You are on page 1of 483

Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008

Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
STP 1277
Electrochemical Noise
Measurement for Corrosion
Applications
Jeffery R. Kearns, John R. Scully, Pierre R. Roberge,
David L. Reichert, and John L. Dawson, Eds.
ASTM Publication Code Number (PCN)
04-012770-27
ASTM
100 Barr Harbor Drive
West Conshohocken, PA 19428-2959
Printed in the U.S.A.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
L i b r a r y o f C o n g r e s s C a t a l o g i n g - i n - P u b l i c a t i o n D a t a
El ect r ochemi cal noi se measur ement f or cor r osi on appl i cat i ons/ Jef f er y
R. K e a r n s . . . [et al.], edso
p. cm. - - (STP 1277)
" ASTM Publ i cat i on Code Number (PCN) 04- 012770- 27. "
Pr oceedi ngs of t he First I nt ernat i onal Symposi um on
El ect r ochemi cal Noi se Measur ement f or Cor r osi on Appl i cat i ons, hel d in
Mont real , Quebec, 15- 16 1994; sponsor ed by ASTM Commi t t ee G-1
on Cor r osi on of Metal s.
I ncl udes bi bl i ogr aphi cal ref erences and i ndex.
ISBN 0- 8031- 2032- X
1. Cor r osi on and ant i - cor r osi ves- - Test i ng- - Congr esses.
2. El ect rol yt i c cor r osi on- - Test i ng- - Congr esses. 3. Pot ent i omet r y- -
- - Congr esses. 4. El ect roni c measur ement s- - Congr esses. I. Kearns,
Jef f ery R. II. ASTM Commi t t ee G-1 on Cor r osi on of Metal s.
II1. I nt ernat i onal Symposi um on El ect r ochemi cal Noi se Measur ement
f or Cor r osi on Appl i cat i ons (1st: 1994: Mont real , Quebec)
IV. Seri es: ASTM speci al t echni cal publ i cat i on: 1277.
TA462. E535 1996
620. 1' 1223- - dc 20 96- 23890
CIP
Copyright 9 1996 AMERICAN SOCIETY FOR TESTING AND MATERIALS, West Conshohocken,
PA. All rights reserved. This material may not be reproduced or copied, in whole or in part, in any
printed, mechanical, electronic, film, or other distribution and storage media, without the written
consent of the publisher.
P h o t o c o p y R i g h t s
A u t h o r i z a t i o n t o p h o t o c o p y i t e m s f o r i n t e r n a l , p e r s o n a l , o r e d u c a t i o n a l c l a s s r o o m u s e , o r
t h e i n t e r n a l , p e r s o n a l , o r e d u c a t i o n a l c l a s s r o o m u s e o f s p e c i f i c c l i e n t s , i s g r a n t e d b y t h e
A m e r i c a n S o c i e t y f o r T e s t i n g a n d M a t e r i a l s ( A S T M ) p r o v i d e d t h a t t h e a p p r o p r i a t e f e e i s p a i d
t o t h e C o p y r i g h t C l e a r a n c e C e n t e r , 2 2 2 R o s e w o o d D r i v e , D a n v e r s , M A 0 1 9 2 3 , Tel : 5 0 8 - 7 5 0 -
8 4 0 0 ; o n l i n e : h t t p : / / w w w . c o p y r i g h t . c o m / .
P e e r R e v i e w P o l i c y
Each paper published in this volume was evaluated by three peer reviewers. The authors
addressed all of the reviewers' comments to the satisfaction of both the technical editor(s) and the
ASTM Committee on Publications.
The quality of the papers in this publication reflects not only the obvious efforts of the authors
and the technical editors, but also the work of these peer reviewers. The ASTM Committee on
Publications acknowledges with appreciation their dedication and their contribution of time and ef-
fort on behalf of ASTM.
Printed in Philadelphia, PA
September 1996
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
Foreword
This First International Symposium on Electrochemical Noise Measurement for Corrosion
Applications was held in Montreal, Quebec, Canada on 15-16 May 1994. ASTM Committee
G-1 on Corrosion of Metals sponsored the symposium. Jeffery R. Kearns, Alcoa Technical
Center, Alcoa Center, PAL, served as chairman of the symposium and was principal editor of
this publication. John R. Scully, University of Virginia, Charlottesville, VA; Pierre R. Rob-
erge, Royal Military College of Canada, Kingston, Ontario; David L. Reichert, DuPont Ex-
perimental Station, Wilmington, DE; and John L. Dawson, CAPCIS Ltd., Manchester, United
Kingdom, served as symposium vice-chairmen and co-editors.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
Contents
Overview ix
KEYNOTE ADDRESS
Electrochemical Noise Measurement: The Definitive In-Situ Technique for
Corrosion Applications?--JOHN L. DAWSON
CORRELATIONS TO OTHER ELECTROCHEMICAL TECHNIQUES
A Comparison of Electrochemical Noise and Impedance Spectroscopy for the
Detection of Corrosion in Reinforced Concr et e- - uco BERTOCCI
Electrochemical Noise and Impedance Analysis of Iron in Chloride Medi a- -
FLORIAN MANSFELD AND HONG XIAO
Electrochemical Noise Measurement for Determining Corrosion Rat es- -
DAVID L. REICHERT
39
59
79
DATA ANALYSIS
The Effects of Solution Resistance on Electrochemical Noise Resistance
Measurements: A Theoretical Analysis--R. A. cOTrlS, S. TUR~OOSE, AND
J. MENDOZA-FLORES
Analysis of Electrochemical Noise for Type 410 Stainless Steel in Chloride
Solutions--ROBERT G. KELLY, MARIA E. INMAN, AND JOHN L. HUDSON
Comparison of Spectral Analysis with Fast Fourier Transform and Maximum
Entropy Method. Application to the Role of Molybdenum Implantation
on Localized Corrosion of Type 304 Stainless Steel--LUC BEAUNIER,
JOEL FRYDMAN, CLAUDE GABRIELL[, FRAN(~OIS HUET, AND MICHEL KEDDAM
93
101
114
A Comparison of Spectral and Chaotic Analysis of Electrochemical Noise---
ANDRAZ LEGAT AND EDVARD GOVEKAR 129
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
Qua nt i f yi ng t he St ochas t i c Be ha vi or of El e c t r oc he mi c a l Noi se Me a s u r e me n t
Du r i n g t he Cor r os i on of Al umi num- - PI ERRE R. ROBERGE
The I dent i f i cat i on of Pi t t i ng a n d Cr evi ce Cor r os i on Sp e c t r a i n El e c t r oc he mi c a l
Noi se Usi ng an Ar t i f i ci al Ne ur a l Net work--THOMAS E. BARTON,
DAVID L. TUCK, AND D. BRE3~F WELLS
142
157
INDUSTRIAL APPLICATIONS
A Revi ew of EPRI Pr oj e c t s Si nce 1984 t ha t Used El e c t r oc he mi c a l Noi se
Me a s u r e me n t I ns t r ume nt a t i on- - BARRY c. SYREqq" AND WILLIAM M. COX
A Pr ogr e s s Re p o r t on t he Use of El e c t r oc he mi c a l Noi se t o I nves t i gat e t he
Ef f ect s of Ze b r a Mus s el At t a c h me n t on t he Co r r o s i o n Res i s t ance of AI SI
Type 304 St ai nl ess St eel a n d Ca r b o n St eel i n La k e Wa t e r - -
A. M. BRENNENSTUHL, BLAIR SIM, AND RENATA CLAUDI
The Use of El e c t r oc he mi c a l Noi se Me a s u r e me n t s wi t h Nuc l e a r Wa s t e T a n k s - -
JOHN I. MICKALONIS, RICHARD J. JACKO, GRAHAM P. QUIRK AND
DAVID A. EDEN
On- Li ne Moni t or i ng Usi ng El e c t r oc he mi c a l Noi se Me a s u r e me n t i n CO-CO2-
a20 Sys t e mSmHENNI E J. DEBRUYN, KEVIN LAWSON, AND EDWARD E. HEAVER
Si mul t a ne ous Cor r os i on a nd Foul i ng Moni t or i ng Un d e r He a t Tr a ns f e r i n
Cool i ng Wa t e r Sys t EmSmMI CHAEL A. WINTERS, PATRICK S. N. STOKES, AND
HENRY F. NICHOLS
El e c t r oc he mi c a l Noi se Me a s ur e me nt s on Ca r b o n a n d St ai nl ess St eel s i n Hi gh
Subc r i t i c a l a nd Supe r c r i t i c a l Aque ous En v i r o n me n t s - -
DIGBY D. MACDONALD, CHUN LIU, AND MICHAEL P. MANAHAN, SR.
The Use of El e c t r oc he mi c a l Noi se t o I nves t i gat e t he Cor r os i on Res i s t ance of
UNS Al l oy N04400 Nuc l e a r He a t Ex c h a n g e r 1~ubes--A. M. BRENNENSTUHL,
G. PALUMBO, F. S. GONZALEZ, AND G. P. QUIRK
The Use of El e c t r oc he mi c a l Noi se Me a s u r e me n t i n t he Eva l ua t i on of Ma t e r i a l s
f or St e a m Ge n e r a t o r s - - P. E. DOHERTY, M. J. PSAILA-DOMI3ROWSKI,
S. L. HARPER, AND w. G. SCHNEIDER
173
186
201
214
230
247
266
288
PITTING
Ana l ys i s of El e c t r oc he mi c a l Noi se f r om Me t a s t a bl e Pi t t i ng i n Al u mi n u m, Age d
AI - 2% Cu, a n d AA 2024- T3- - SHELDON T. PRIDE, JOHN R. SCULLY, AND
JACK L. HUDSON 307
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
Ef f ect o f Met al l urgi cal Fac t or s o n t he El e c t r oc he mi c al Noi s e Me a s ur e d o n
AI SI Type 430 St ai nl es s St eel s i n Chl or i de - Cont ai ni ng Me d i a - -
DOMINIQUE GORSE, CLAIRE BOULLERET, AND BERNARD BAROLIX 332
The Ef f ect o f S o me Funda me nt a l As pe c t s o f t he Pi t t i ng Co r r o s i o n o f St ai nl es s
St eel o n El e c t r oc he mi c al Noi s e Me a s ur e me nt s - - P. CHRIS PISTORIUS 343
METHODS
Coupl i ng o f Ac ous t i c Emi s s i o n and El e c t r oc he mi c al Noi s e Me a s ur e me nt
Techni ques i n Sl urry Er o s i o n- Co r r o s i o n St udi es--ROLAND OLTRA,
BRIGrVrE CHAPEY, FRAN(~OIS HUET, AND LIONEL RENAUD 361
I nf l uence o f Hy dr o g e n Abs o r pt i o n o n t he El e c t r oc he mi c al Pot ent i al Noi s e of
an I r on El e c t r ode Unde r Co r r o s i o n wi t h Gas Evol uti on--e~ANCOIS HUET,
MICHEL J~ROME, PANAGIOTIS MANOLATOS, AND FRAN(~OIS WENGER 375
El e c t r oc he mi c al Noi s e Re s pons e o f St eel Unde r Hy dr o dy na mi c Co n d i t i o n s - -
JOSI~ M. MALO AND OCTAVIO VELAZCO 387
El e c t r oc he mi c al Noi s e Me a s ur e me nt s Dur i ng Exf ol i at i on o f Al u mi n u m
Al i oys ~JUAN DE DAMBORENEA AND BERNARDO FERNANDEZ 398
Cha r a c t e r i z a t i o n o f Pr e r us t e d St eel s i n S o me I be r o - Ame r i c a n At mo s phe r e s by
El e c t r oc he mi c al Pot ent i al Noi s e Me a s u r e me n t - -
ELISABETE ALMEIDA, LIBORIA MAR1ACA, ABELARDO RODRIGUEZ,
JORGE URUCHURTU CHAVAR[N, AND MARIA A. VELOZ 411
STANDARDIZATION
Reproduci bi l i t y o f El e c t r oc he mi c al Noi s e Dat a f r o m Co a t e d Me t al S y s t e ms m
GORDON P. BIERWAGEN, DOUGLAS J. MILLS, DENNIS E. TALLMAN, AND
BRIAN S. SKERRY 427
AS TM St a nda r di z a t i o n o f El e c t r oc he mi c al Noi s e Me a s u r e me n t m
JEFFERY R. KEARNS, DAVID A. EDEN, MAX R. YAFFE, JEFFERSON V. FAHEY,
DAVID L. REICHERT, AND DAVID C. SILVERMAN 446
I nde xe s 471
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
Overview
El ect r ochemi cal noi se measur ement (ENM) is a cont roversi al subject. There are no est ab-
l i shed t est met hods, and t here is no consensus on a t heoret i cal f r amewor k for i nt erpret i ng
data. It is unusual for ASTM Commi t t ee G-1 on Corrosi on, Det eri orat i on, and Degradat i on
of Mat er i al s to begi n st andards devel opment under such ci rcumst ances. Nevert hel ess, a t ask
group was f or med by member s that were both concer ned about t he r api dl y growi ng com-
merci al mar ket for t he t echnol ogy and i nt erest ed in pursui ng fundament al wor k on a " ne w"
el ect rochemi cal technique. The basi c quest i ons that ASTM Commi t t ee G-1 asks about any
el ect rochemi cal t echni que that is pr oposed for st andardi zat i on are " What does t he pr oposed
t echni que measur e?, " " Have i nst rument s and met hods been est abl i shed to make rel i abl e
measurement s?, " and " Does t he pr oposed t echni que pr ovi de useful i nformat i on that is not
possi bl e to obt ai n by exi st i ng st andardi zed t echni ques?" To answer t hese quest i ons, the
ASTM Commi t t ee G-1 Task Gr oup on ENM and the symposi um authors represent ed by this
STP were char ged wi t h t he t ask of devel opi ng consensus on t hree basi c issues:
(1) how shoul d a measur ement be made so that it can be compar ed wi t h confi dence to
others,
(2) what el ect roni c measur ement capabi l i t i es and cal i brat i on procedures are necessary to
make a val i d measurement , and
(3) how can the dat a be most effi ci ent l y anal yzed and r el i abl y i nt erpret ed.
It is hoped that progress wi t h t hese i ssues wi l l l ead to the creat i on of new corrosi on t est
st andards and the wi der accept ance of t he t echni que in research and industry.
Gi ven t he cont roversy surroundi ng the subject, it is appropri at e to pr ovi de wor ki ng defi-
nitions for t he t erms used in t he title of this STP.
Electrochemical noise--Fluctuations of pot ent i al or current, t ypi cal l y of low frequency
( <10 Hz) and l ow ampl i t ude. El ect r ochemi cal noi se ori gi nat es, in part, from nat ural varia-
tions in el ect r ochemi cal rate ki net i cs duri ng a corrosi on process. El ect r ochemi cal noi se is
oft en regarded as a r andom (st ochast i c) phenomena coupl ed to det ermi ni st i c ki net i cs.
Electrochemical noise measurement--A t echni que i nvol vi ng the acqui si t i on processi ng and
anal ysi s of el ect r ochemi cal noi se data. Dat a are t ypi cal l y acqui red by moni t ori ng the evo-
lution of a corrosi on process on t wo or mor e coupl ed el ect rodes wi t hout the appl i cat i on of
an ext ernal signal. Al t ernat i vel y, pot ent i ost at i c or gal vanost at i c met hods can be used to mea-
sure el ect rochemi cal current or pot ent i al noise, respect i vel y. Various el ect roni c filtering and
mat hemat i cal met hods can be used to anal yze el ect rochemi cal noi se signals. Wi t h appr opr i at e
corroborat i ng measurement s and observat i ons, ENM dat a can pr ovi de i nformat i on concern-
i ng t he nature of t he corrosi on process and the magni t ude of t he corrosi on rate for a gi ven
system.
Historical Perspective
Fl uct uat i ons of the pot ent i al or current from a cor r odi ng el ect r ode are a wel l - known and
easi l y observabl e phenomenon. Before t he 1970s, el ect r ochemi cal noi se was regarded as a
source of bi as and error that compr omi sed el ect rochemi cal measurement s rat her than as a
ix
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
X ELECTROCHEMICAL NOISE MEASUREMENT
rich source of i nformat i on. (The r eader is referred to the keynot e address for a revi ew of
much of the earl y work on what is now consi dered to be ENM. ) The use of el ect rochemi cal
pot ent i al noi se measur ement (EPNM) as a corrosi on research t ool has i ncreased st eadi l y
si nce I ver son' s paper in 1968. The measur ement of el ect rochemi cal current noi se (ECN)
bet ween coupl ed el ect rodes combi ned wi t h the measur ement of EPNM was est abl i shed as a
vi abl e i ndust ri al moni t ori ng met hod by workers at the Uni versi t y of Manchest er in the earl y
1980s. The use of the t echnol ogy has grown in l aborat ori es and i ndust ri al pl ant s wor l dwi de,
such that by the end of t he decade it was bei ng cal l ed the "Techni que of the Ni net i es. "
ASTM Commi t t ee G-1 began to consi der t he devel opment of st andards on ENM in 1991
and f or med a t ask group on the t opi c (J. R. Kearns, chair). The scope of t he t ask group is
to "devel op st andards that descri be i nst rument s and met hods for maki ng and anal yzi ng
el ect rochemi cal noi se measurement s. " One of the first act i vi t i es of t he t ask group was to
i ni t i at e the organi zat i on of an i nt ernat i onal conference on ENM to assess this t echnol ogy
and i dent i fy new areas for research and st andardi zat i on.
Summary of the STP
Thi s ASTM Speci al Techni cal Publ i cat i on (STP) resul t ed from the first i nt ernat i onal sym-
posi um on El ect r ochemi cal Noi se Measur ement for Corrosi on Appl i cat i ons hel d in Mont real ,
Quebec, Canada on 16 and 17 May 1994. Of the 36 present at i ons made at the symposi um,
28 papers were approved (3 peers revi ewed each paper) for this STP. The pri nci pal authors
were from ei ght countries, i ncl udi ng Canada (4), Fr ance (4), Mexi co (3), New Zeal and (1),
Sl oveni a (1), Sout h Af r i ca (2), Uni t ed Ki ngdom (3), and the Uni t ed States (10) and repre-
sented a vari et y of i ndust ri al and academi c perspect i ves on ENM. The papers have been
grouped into six sections on Correl at i ons to ot her El ect rochemi cal Techniques, Dat a Anal -
ysis, Indust ri al Appl i cat i ons, Pitting, Met hods, and St andardi zat i on. Thi s STP provi des a
snapshot port rai t of the state of the art for ENM in the earl y 1990s.
The fact that many of the papers in the STP, part i cul arl y in the Indust ri al Appl i cat i ons
section, have one or mor e coaut hors that are associ at ed wi t h the Uni versi t y of Manchest er
reflects the hi st ory of the commer ci al devel opment of the t echnol ogy at this institution. The
r eader shoul d al so note that an even great er number of papers in this STP are coaut hored
by i ndi vi dual s i ndependent of the Manchest er group. Thi s fact not onl y serves to val i dat e
aspect s of the early, i ndust ri al l y ori ent ed wor k done in Manchest er, but it also expl ai ns the
rapi d i ncrease in t he use of t echnol ogy around the worl d duri ng the past five years.
Overview of Papers
The keynot e address (Dawson) is i nt ended pr i mar i l y to pr ovi de backgr ound for the non-
speci al i st . Shoul d the r eader want mor e dept h, an ext ensi ve list of references is i ncl uded
coveri ng the scientific research l eadi ng to ENM, i nvest i gat i ons of vari ous forms of corrosi on
usi ng ENM, and the appl i cat i on of ENM to i ndust ri al cor r osi on- moni t or i ng situations. The
title of the keynot e address poses the quest i on as to whet her or not ENM is the definitive
el ect rochemi cal technique. The answer to the quest i on posed in the keynot e title is No! ENM
is not a panacea and is best used as one of several t echni ques in what is often referred to
as a mul t i t echni que approach. For this reason, the first sect i on of this STP on Correl at i ons
to Ot her El ect rochemi cal Techni ques is very useful in est abl i shi ng the val i di t y of ENM and
in demonst rat i ng how it can be effect i vel y combi ned wi t h mor e est abl i shed techniques.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
OVERVIEW x i
Correlations to Other Electrochemical Techniques
Linear polarization resistance (LPR) and electrochemical impedance spectroscopy (EIS)
are well-established corrosion measurement tools and are logical standards for qualifying
ENM as a method for measuring general corrosion. In the first paper in this section (Bertocci)
note that one must eliminate data from unwanted noise sources, and that trend removal may
also be warranted. Whet her using analog or digital (mathematical) filtering, caution must be
exercised to prevent the elimination of valid data.
The next two papers (Mansfeld and Xi ao and Reichert) found that the relatively new
parameter unique to ENM called the noise resistance, R, , correlated well with polarization
resistance, Rp, as measured by ENM and LPR techniques, respectively. The manner in which
R, should be calculated was debated between the t wo papers ( R , , = V , , I I , , where 17, and 1,
were taken as the root mean square [RMS] values of electrochemical potential noise and
electrochemical current noise, respectively, in the first paper and as standard deviation values
in the second paper). Selection of RMS or standard deviation values depends upon how the
data were collected and filtered.
The second paper of the session (Mansfeld and Xiao) confirmed that frequency plots of
ENM data are similar to EIS data and good agreement can be obtained between ENM and
EIS data at the zero frequency limit. The authors prefer the EIS technique because it provides
more information than ENM, especially for evaluating coated systems. It was observed in
the Reichert paper that for very active systems (1.3 r am/ y) the correlation of corrosion rate
calculated from ENM did not correlate with that from Rp or mass loss data. The interpretation
of the practical studies of uniform corrosion reported in this paper is regarded from a the-
oretical perspective in the first paper of the next session.
Data Analysis
The first paper of the section (Cottis et al.) addresses, in part, the need expressed in
Reichert' s paper for a theoretical basis relating ENM to corrosion rate. The authors concl ude
that the noise resistance parameter is a measure of essentially the same resistance as that
obtained by conventional LPR methods. ENM is superior to LPR in measuring resistance in
high-resistance systems.
The second paper (Kelly et al.) critically evaluates the correlation of two of the more
common noise parameters, noise resistance and the pitting index, to conventional electro-
chemical and physical measures of degradation. The results, for AI SI Type 410 stainless
steel exposed to several test conditions that induce different modes and rates of corrosion,
cast doubts on the credibility of these parameters for all systems. More work needs t o be
done to explain the apparent lack of correlation in the results of ENM and other methods
for the alloy and environments in question.
A common way to analyze noise data in many science and engineering disciplines has
been to transform time records to the frequency domai n t o obtain power spectra. Electro-
chemical noise signals often consist of a compl ex combination of deterministic and stochastic
processes, so a common approach to analysis has been to correlate domi nant frequencies
and deconvolute unwanted signals in an iterative manner using established mathematical
functions. For instance, power spectral density (PSD) plots are comput ed using algorithms
such as the fast Fourier transforms (FFT) or the maxi mum entropy method (MEM). A com-
parison of these two methods is made in the third paper (Beaunier et al.) in which it was
revealed that the validity of the low-frequency plateau obtained by the analysis of noise
results with MEM can yield questionable results. The authors apply the technique to elucidate
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
x i i ELECTROCHEMICAL NOISE MEASUREMENT
a practical corrosion problem, the influence of mol ybdenum additions (in the case by im-
plantation) to iron-chromium-nickel (Fe-Cr-Ni) alloys on pitting behavior.
The paradi gm of nonlinear dynami cs (also called chaos theory or factals) is impacting
everything from stock price forecasting to corrosion science. Analytical tools have been
developed to detect and characterize deterministic or chaotic features in seemingly random
data. The fourth and fifth papers of the session (Legat and Govekar and Roberge) review
the j argon and some of the mathematics as would seem to be relevant to the analysis of
ENM data. The techniques presented in both papers are illustrated by their use to reveal
important features of compl ex ENM data. Legat and Govekar provide a general introduction
to chaos and then demonstrate the utility of the techniques by identifying in experimental
ENM data records changes in the mode of corrosion for a carbon and a stainless steel exposed
to a changi ng test environment. Roberge describes and then applies two chaos techniques,
the stochastic process detector and rescaled range analysis. The critical pH of a 5000 series
aluminum alloy exposed to a series of salt solution of various degrees of acidity was iden-
tified by these techniques. Extensive lists of references for more in-depth knowl edge in this
important area are provided in both papers.
A quite different computer-based approach concludes the session (Barton et al.). Back-
ground to the t echnol ogy is presented along with a case history that demonstrates the power
of artificial neural networks to classify ENM data. In a case history, an artificial neural
network is trained with many ENM time series records for AI SI type 304 stainless steel in
dilute chloride solution t o identify the onset of localized corrosion.
Industrial Applications
The eight papers contained in this section survey the use of ENM in modern corrosion
monitoring and in industrial service-relevant testing of materials. The goal of industrial cor-
rosion monitoring is to provide a clearer picture of the conditions in an operating plant that
lead to corrosion damage to avoid unscheduled downtime. The operating conditions resulting
in corrosion are often transient and are frequently masked or by other events. Syrett and
Cox present a modern approach to corrosion monitoring as realized through a series of
programs sponsored by the Electric Power Research Institute (EPRI) over the past decade.
This significant body of work demonstrates that corrosion monitoring by multiple techniques,
including ENM, can be used as a sensitive indicator of the conditions within a system for
purposes of corrosion prevention and process control.
In the second paper (Brennensthul et al.) of the section, ENM was successfully used to
detect the effect that zebra mussel settlement has on the corrosion performance for stainless
steel in comparison to steel. The exposure and monitoring was done at a field test site near
the water inlet to Ontario Hydr o' s Nanticoke Thermal Generating Station. A number of
parameters were monitored simultaneously, including coupling current, electrochemical cur-
rent and potential noise, degree of localization, and noise resistance. ENM proved to be
sensitive to changes associated with mollusk settlement, and collaborating evidence revealed
that it was not detrimental to corrosion performance; in fact, the binding bi opol ymers were
actually proposed to inhibit corrosion.
Nuclear wastes at the Savannah River site are partially processed and stored in plain carbon
steel tanks. As described in the third paper of this session (Mickalonis et al.) based on the
encouraging results of a preliminary laboratory study, Westinghouse decided to use an ENM-
based monitoring system to assure that the corrosion inhibitors used to protect these tanks
remained effective. Under inhibited storage conditions, the noise data was fairly stable. ENM
was sensitive to the steel passive-active transition associated with greater solution nitrate
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
OVERVIEW x i i i
concentrations. The authors proposed that with appropriate calibration, the technique may
also be useful for the detection of stress corrosion cracking (SCC).
Switching from the power industry, the next t wo papers in this section (DeBruyn et al.
and Winters et al.) deal with the use of ENM for on-line monitoring in the oil and gas
industries. At a Mossgas (Pty.) Ltd. (Republic of South Africa) gas plant in a pipeline
downstream of reformer units, ENM was used to provide a continuous indication of the
capacity of the service environment to support SCC and other types of localized corrosion.
Conventional corrosion monitoring was ineffective for the condensing thin-film conditions
in a wet CO- CO 2 gas service environment. A preliminary laboratory study was conduct ed
to establish opt i mum probe and instrument configurations as well as measurement capabil-
ities. Stressed and unstressed sensor probes were inserted directly into the process line. The
ENM-based moni t or was used to identify times when process conditions are such that the
carbon steel pipeline may be susceptible to SCC. By instituting effective countermeasures
at such times, there is the opportunity to prevent SCC of the pipeline by process control.
The second paper about monitoring in the oil and gas industries (Winters et al.) describes
the development of a side-stream syst em that was designed t o overcome many of the short-
comi ngs of fouling monitors currently used in the treatment of cool i ng water. I n addition to
monitoring indicators of fouling, the system simultaneously monitors corrosion with ENM
and t wo other supporting electrochemical techniques. A sensor array is integrated into the
heat exchanger tubing bundle so that critical operating conditions can be monitored. In a
field trial, the system was successfully used to verify the operating limits and scale deposit
control capability of the water treatment packages.
The last three papers of the section (Macdonald et al., Brennenstuhl et al., and Doherty
et al.) offer examples of how ENM can be used to evaluate materials of construction under
test conditions relevant to the actual plant operating environment. The goal of the first ser-
vice-relevant testing paper (Macdonald et al.) is to provide a clearer picture of the devel-
opment of damage in a plant by measuring relevant parameters under actual operating con-
ditions. The simplicity of the coupl ed electrode for monitoring electrochemical current noise
provides for a sensor capable of monitoring changes in key parameters at elevated temper-
atures and pressures. The lack of such a sensor has impeded the introduction of monitoring
systems for the thermal transport circuits of power plants. In quite a different test apparatus,
ENM was explored as a means of detecting and monitoring stress corrosion cracking in
sensitized Type 304 stainless steel in high-temperature water (up to 288~ The magnitude
and pattern of the data were used to detect changes in the chemical and physical properties
of the fluid as well as the mode of corrosion.
The last t wo papers of the session (Brennensthul et al. and Dohert y et al.), describe the
use ENM in a multitechnique approach (ENM, open-circuit potential measurement, and zero
resistance ammetry). From the data collected by the multitechnique approach, several ENM
parameters were t racked versus time (for example, ECN standard deviation, mean coupling
current, noise resistance, degree of localization, and pitting function) and related to subtle
differences in the composition and microstructure of nickel base alloys as well as the sim-
ulated process environment. Workers at Ontario Hydro Technologies (Brennensthul et al.),
suspected that differences in corrosion performance, as observed in-service, were masked in
conventional testing by two factors: the aggressiveness of the test environment and the limited
sensitivity of the corrosion measurement method. In the last paper of the session (Doherty
et al.), workers from McDermott, Babcock, and Wilcox used ENM parameters to compare
differences in simulated cleaning solution (NaOH) corrosivity, crevicing conditions, and com-
position on steam generator tubing performance. The noise resistance, degree of
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
xi v ELECTROCHEMICAL NOISE MEASUREMENT
localization, and standard deviation of ENM were found to be the most reliable parameters
for making these three types of comparisons, respectively.
Pi t t i ng
The goal of the papers in this session is elucidate the mechani sm of pitting through the
sensitivity of the ENM technique. In the first paper of the session, Pride et al. measure EPN
and ECN by galvanostatic and potentiostatic techniques, respectively, to exami ned the tran-
sition from metastable to stable pitting. High-purity aluminum, A1-Cu binary alloy, and AA
2024-T3 were examined under conditions of increasing applied potentials or chloride-ion
concentration. ECN spikes were directly linked to individual metastable pitting events. Ratios
of pit currents to pit radii at various growth times were analyzed. The formation of a stable
pit was correlated to pit stabilization conditions producing ECN events that continuously
exceeded the threshold ratio of 10 -2 A/ cm. This approach is promising since this ratio is
fundamentally linked to the pit solution concentration required to maintain active pits.
In the second paper, Gorse et al. attempted to examine the effects of sulfur and titanium
on the metastable pitting behavi or of AI SI 430 stainless steel by investigating the shapes of
current transients. Individual ECN events associated with metastable pitting were described
by current spikes which obeyed power law relationships. Attempts were made to relate the
power law behavior of the current spike to the identity of the pit site, such as titanium
carbide and manganese sulfide inclusions.
In the last paper of this session, Pistorius considers the merits of using either ECN or
EPN to predict the transition from metastable to stable pitting in AI SI 304 stainless steel.
By coupling the stainless steel to a platinum cathode, it was shown that a large portion of
the pit charge was supplied by capacitive currents from the adjacent platinum cathode. The
size of potential fluctuations during metastable pit growth was argued to be related mainly
to the pit charge (size) instead of pit current or current density. From prior work it was
known that increasing pit currents for similarly sized pits provide a good indication of the
transition to stable pitting. Therefore, EPN was viewed to have less merit as an indicator of
pit stabilization than ECN. It was theorized that the standard deviation of metastable pit
currents would provide a better indicator of the transition to stable pitting. However, the
presence of heat tint oxides formed during heat treatment of samples complicated the estab-
lishment of straightforward relationships between ECN and conditions for pit stabilization
for a variety of reasons.
Me t ho ds
In this section, ENM methods were applied to a diverse number of corrosion problems.
In each paper, alternative corrosion monitoring tools augmented ENM studies. The first paper
by Oltra et al. examined ECN and acoustic emission (AE) associated with the abrasive action
of a slurry on a metallic target. The fluctuating AE and ECN signals were processed in two
ways (statistical analysis of signals in the time domain and in terms of a spectral density
plot) with the overall goal of quantifying the contributions to metal loss by mechanical wear
and corrosion. A linear relationship was established between abrasive metal loss and the
statistical analysis of acoustic signals. Interestingly, total corrosion currents could not be
considered as the linear superposition of single current transients suggesting that the observed
depassivation-repassivation events were not completely independent.
In the second paper, Huet et al. examined EPN associated with the proton discharge
reaction both at open circuit and under applied cathodic polarization on iron using Devan-
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
OVERVIEW XV
athan permeation foils exposed to sulfuric acid. Simultaneously with EPN, associated fluc-
tuations in electrolyte resistance on the charging surface and the hydrogen permeation flux
on the exit surface of the foil were characterized. The hydrogen permeation flux is coupled
to the behavior on charging surfaces through solid state hydrogen diffusion and fluctuations
in charging surface lattice hydrogen concentrations could be estimated. These fluctuations
were linked with hydrogen bubble formation and detachment from the charging surface under
cathodic polarization. However, the origins of fluctuations for open circuit hydrogen pro-
duction and ingress remained undetermined. Insight into factors affecting hydrogen absorp-
tion was gained.
In the third paper, Mal o and Velazco sought to distinguish artificial noise signals associated
with a rotating disk electrode device and other sources from noise of fundamental electro-
chemical origins. Artificial sources produced noise signals that did not exceed 10 v~V and
0.1 ixA. Corrosion processes for mild steel in Na2SOa produced much larger noise levels in
this system. In the next to last paper in the session, de Damborenea and Fernfindez examined
ENM during the exfoliation corrosion of AA 8090 in the ASTM Test Met hod for Exfoliation
Corrosion Susceptibility in 2XXX and 7XXX Series Al umi num Alloys (EXCO Test) (G 34).
Power spectral density (PSD) and statistical methods, including noise resistances, were com-
pared to charge transfer resistances obtained from EIS. The early stages of localized corrosion
were detected by ENM, but strong correlations between ENM data and corrosion rates during
severe exfoliation were generally lacking. EPN was used in the last paper of the session
(Almeida et al.) to examine the protective nature of the corrosion products developed on
pretreated and untreated carbon steels exposed at various Ibero-Ameri can atmospheric test
sites. These samples were subsequently immersed in Na2SO4 solutions for linear polarization
resistance (LPR) and EPN measurements at different times. Samples with higher corrosion
rates, because of less protective oxide layers or more aggressive atmospheric conditions or
both, correlated well with increased magnitudes of EPN particularly in the cases in which
localized corrosion was promot ed by chloride ion incorporation into oxide layers during
exposure to marine atmospheres.
St andar di z at i on
The first paper by Bierwagen et al. covers several topics and is the onl y one in the STP
with the goal of evaluating the reproducibility of data obtained for any given system. The
authors present a significant body of data that should serve as a benchmark for future work
in this area. By rigourously controlling test variables and statistically validating test results,
the authors justify the use of ENM for the evaluated coated metal systems. The final paper
(Kearns et al.) of the session and the symposi um reviews the activities of the ASTM Com-
mittee G- i task group on ENM since its beginning in 1991. The paper presents the group' s
first attempt to develop consensus on: (1) terminology, (2) specifications and configurations
for laboratory instrumentation, (3) laboratory apparatus, and (4) data analysis methods. The
pr oof of these deliberations is a test method and instrument specification that when followed
yield a set of data for a given system that can be compared with confidence to others
following the same. To this end, instrument specifications, test procedures, and test results
are presented. It is hoped that this paper encourages participation in future task group activ-
ities and the round robin test program that is planned. An extensive list of terms related t o
ENM technology is defined in the Appendix to the paper, which the reader may find useful
when reading this and other papers in this STP.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
XVI ELECTROCHEMICAL NOISE MEASUREMENT
Cur r e nt St at e and Fut ur e Di r e c t i ons
The reliable interpretation of ENM data is the major challenge to the advance of the
t echnol ogy in corrosion research, materials testing, and industrial monitoring. At this time,
it has been established that ENM systems are capable of detecting subtle changes in the
material or environment or both under a wide variety of plant processes and laboratory test
conditions. With strong corroborating information from other sources, ENM data can be
related with confidence to changes in corrosion behavior or environmental conditions or
both. However, until a theoretical framework for interpreting corrosion events from ENM
data are established and standardized testing practices are widely accepted, the reader is
encouraged to be skeptical about detailed physical interpretations of ENM data in the absence
of strong corroboration: As was stated at the symposium, "correlation alone is not sufficient
to establish causality."
In addition to the reliable interpretation of data, other exciting developments are also
anticipated. No doubt there will continue t o be more unique applications for ENM in cor-
rosion monitoring and even process control. The challenges encountered in the field will set
the pace for much laboratory experimentation and research. In addition to successful indus-
trial applications, the next major step toward validating ENM is anticipated to be in regard
its use in elucidating various forms of localized corrosion, such as SCC and pitting. In
addition to comparisons of parameters common to ENM and more established electrochem-
ical techniques, such as LPR and EIS, one can also look forward to the analysis of ENMs
by computer-intensive tools associated with artificial neural networks and nonlinear
dynamics.
Measurement is the final area for future work that is anticipated based on a reading of
this STP and discussions at the symposium. There is a need, in both the lab and plant, for
techniques/instruments that can handle the large amounts of data that are acquired during
real-time monitoring without losing valuable information. Specifically, more work is needed
to define the best methods for electronic filtering, trend subtraction, and localized event
recognition of ENM signals. However, in an era of incredibly powerful, inexpensive, user-
friendly electronics and software, it is very important t o consider just how electrochemical
signals are processed into ENM data to avoid mis- and over-interpretation of data. With this
caution in mind, it is essential that calibration procedures for instrumentation and reference-
corroding systems be established along with diagnostic tests for the validity of data so that
measurements can be compared with confidence. The ASTM Committee G-1 task group on
ENM is an excellent forum to address these latter tasks.
S u mma r y
The combi ned offerings of this STP provide the reader with a review of the state of the
art of ENM in the early 1990s. The integration of ENM parameters with other electrochem-
ical measures of system status demonstrate the opt i mum use of ENM technology. Many
diverse industrial applications for ENM are presented along with the latest laboratory meth-
ods and data analysis techniques. The approaches of the individual groups varied accordi ng
to the anticipated nature of the corrosive environment and the anticipated response of the
material to it. In addition to material-environment considerations, the individual goals of the
monitoring or testing operation also influenced decisions about what is t o be measured and
how data should be acquired and analyzed. Strategies to monitor and control corrosion in
heat exchangers, pipelines, storage tanks, and steam generators using ENM t echnol ogy are
discussed from an industrial perspective. Workers from laboratories around the world are
using ENM to glean new insights into traditional corrosion problems. From this plethora of
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
OVERVIEW xvi i
knowledge and experiences, a consensus is emerging on how to make ENMs, but the inter-
pretation of ENM data and the optimum use of ENM technology for corrosion prevention
and process control in industrial systems are issues that still present many challenges. It is
hoped that some of the proposed test methods and data analysis techniques presented in this
STP will gain wider acceptance and eventually lead to the development of new ASTM
standards.
Jeffery R. Kearns
Alcoa Technical Center
Alcoa Center, PA; symposium
chair and principal editor.
John R. Scully
University of Virginia
Charlottesville, VA; symposium
vice-chair and editor.
Pierre R. Roberge
Royal Military College of Canada
Kingston, Ontario; symposium
vice-chair and editor.
David L. Reichert
DuPont Experimental Station
Wilmington, DE; symposium
vice-chair and editor.
John L. Dawson
CAPCIS Ltd.
Manchester, UK; symposium
vice-chair and editor.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
Keynote Address
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
J o h n L. Da ws o n t
Electrochemical Noise Measurement:
The Definitive in-Situ Technique for
Corrosion Applications?
REF ERENCE: Dawson, J. L., "El e c t r oc he mi c al Noi s e Me as ur e me nt : The Def i ni t i ve l n -
Si t u Te c hni que f or Corros i on Appl i c at i ons ?, " El ect r ochemi cal Noi s e Me as ur e me nt f o r Cor-
rosi on Appl i cat i ons, AS T M STP 1277, Jeffery R. Kearns, John R. Scully, Pierre R. Roberge,
David L. Reichert, and John L. Dawson, Eds., American Society for Testing and Materials,
1996, pp. 3-35.
ABSTRACT: A review is presented of electrochemical noise (EN), the generic term given to
fluctuations of current and potential seen in high-temperature corrosion, molten salt corrosion,
and aqueous corrosion. EN levels in corrosion and particularly localized corrosion are signif-
icantly greater than EN observed in redox systems. EN associated with corrosion is the result
of stochastic pulses of current generated by, for example, sudden film rupture, crack propa-
gation, discrete events involving metal dissolution at etch pits, grain boundaries and kink sites,
and hydrogen discharge with gas bubble formation and detachment. EN in corrosion includes
low-frequency, nonstationary, and weakly stationary processes; transients; and cyclic or oscil-
latory phenomena.
The use of EN, obtained either by potentiostatic/galvanostatic measurements or at freely
corroding potentials, has been shown to offer advantages over conventional DC and AC tech-
niques in research studies, testing, and corrosion monitoring. In many cases, reaction mecha-
nisms can be elucidated and corrosion rate information can be obtained.
Under freely corroding conditions, current noise impulses give rise to changes in potential
through the interracial impedance, (1,) - - (E, , )/IZI. Correlation of current and potential noise
from uniform corrosion may be used to obtain the electrochemical impedance of the corroding
interface; these noise signals are produced by weakly stationary or deterministic processes.
Corrosion rates may therefore be evaluated from this low-frequency impedance or from the
associated but simplified noise resistance, obtained from the standard deviations of the current
and potential, trE/tr , = R , .
Localized corrosion events produced by Poisson processes can be easily identified from time
records and spectral density plots. Localized corrosion involves nonstationary systems. The
anodic and cathodic processes can be separated in both time (the anodic event occurring before
the cathodic event or vice versa) and space (at different parts of the electrode surface). The
electrons produced or consumed during these corrosion events are stored temporarily in the
electrochemical double layer and corrosion product film capacitors.
Assessment of individual transients, use of signal analysis techniques, modeling of ensem-
bles of transients as developed for electrocrystallization studies, and use of the chaos theory
have all been used in EN evaluations.
KEYWORDS : electrochemical noise, corrosion monitoring, testing, electrochemical noise im-
pedance, uniform corrosion, localized corrosion, pitting, stress corrosion cracking, coatings
The obj ect i ves o f t hi s paper ar e to pr ovi de a hi st or i cal per s pect i ve and ove r vi e w o f el ec-
t r ochemi cal noi s e ( EN) t hat c ompl i me nt s pr evi ous sci ent i f i c r evi ews by Gabr i el l i , Huet , and
Senior Consultant, CAPCIS Ltd., Bainbridge House, Granby Row, Manchester, M1 2PW, UK.
Copyrighl 9 1996 by ASTM lntcrnational
3
www.astm.org
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
4 EL ECT ROCHEMI CAL NOI SE ME A S URE ME NT
Keddam [1-3]. The emphasis is on the wider use of electrochemical noise measurement
(ENM) not onl y in corrosion research but also in corrosion testing and on-site monitoring.
Included are a basic introduction for those scientists and engineers who are not familiar with
the subject, an outline of scientific and practical developments to date with appropriate
reference to the literature, and finally, an indication of some areas of possible future study.
EN from many corrosion systems, but particularly uniform corrosion, is observed as spon-
taneous low-frequency fluctuations of current and potential (Fig. 1). In the case of localized
corrosion, distinct anodic or cathodic transients are observed (Fig. 2). EN data are obtained
potentiostatically (Fig. 3a), or galvanostatically (Fig. 4a), or at the natural corrosion potential
(Figs. 1 and 2). In all cases, stochastic processes are involved; that is, naturally occurri ng
events take place in both a random manner across the corrodi ng surface and a random manner
throughout the measurement period.
Sources of electrochemical noise depend on the particular corrosion circumstances. These
may be: (1) the result of mechanical effects, as during the propagation of stress corrosion
cracks (Fig. 5) and abrasion of metallic surfaces by particles (Fig. 6a); (2) the result of
chemical effects, such as adsorption of chlorides on passive films leading to film rupture and
repassivation during pit initiation (Fig. 7); or (3) the growth and detachment of hydrogen
bubbles during active dissolution in acidic solutions (Figs. 3a and 4a). EN can be described
in terms of Poisson processes; the result of a series of random (stochastic) events or bursts
of activity associated with either the anodic or cathodic kinetics.
The growi ng international interest in EN over the past 15 years can be seen in the increase
in papers and publications describing scientific investigations and practical applications.
There is also a growing consensus that EN can be used quantitatively, and it is therefore
appropriate that the First International Symposi um on Electrochemical Noise Measurements
for Corrosion Applications in 1994 was sponsored by ASTM.
Background
EN in corrosion [11,12] and electrochemical studies [13-31] were first reported in the
early 1960s and 1970s. The research on electrochemical processes has been evaluated and
presented in a comprehensive theoretical treatment by Rangarajan and Serclathan [32,33],
but they did not consider corrosion processes. It is onl y since the corrosion publications of
the late 1970s and early 1980s [34-50] that there has been an increasing interest in the
+0.73 +0.024
#_
-0.645
r ' y
/
t l ii
8 1 .
4O96
4096 0
lime (seconds) "rime ( s e c o n d s )
a ) P o t e n t i a l - t i m e r e c o r d b ) C u r r e n t - t i m e r e c o r d
FIG. l - - Si mul t aneous changes of coupling current and corrosion pot ent i al f rom a mi l d steel and a
sacrificial zinc anode [ 4 ] .
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
KEYNOTE ADDRESS 5
Potential (V)
-20.00E-04
- 40. 00E- 04
- 80. 00E- 04
Potential Noise
I
L~.,~, , l ~ ~ k J ~ , . ,-, ,. , I,
F" ' t i ' q J ~ 1 ~ " ' ' " 1 ' ~
r t . , f l h , l ' "
h 4 I I I I I I I I
I 1 ( 1 1 1 H I ! !
! 1 1 1 1 1 1 I l U l l l
I ~ 1 1 i 1 1 1 I I I I I !
i d l l l l l U l l l !
~ 1 1 t l l I I I I I I
t
O.OOE+O0 10.00E+02
I :-' L t t
~ ~ / l l C d ~ 1
s I" ' rl
I' !
f
i
20.00E+02
l i me (s)
Current (A) Current Noise
- 10. 00E- 09 J i
o.oo~.oo l i b I., lJ~ i
I III ill ILlt I ~
' ~ 1 7 6 1 7 6 , , I , , I H , l l l l l l i l l , u
~oo.o ~ ~ v l ~ O
30.00E-09 I
O.OOE+O0 10.00E+02 20.00E+02
"rime (s)
FIG. 2- - Si mul t aneous current and pot ent i al noise transients showi ng pi t initiation and met ast abl e
pi t t i ng [5].
subject. This is partially because the availability of digital equipment makes the measurement
and recording of noise data a relatively easy task. Equally important is the recognition that
EN interpretation provides scientific answers in corrosion research and solves practical prob-
lems in corrosion engineering.
Most workers who have made DC polarization or electrochemical impedance measure-
ments have noticed the apparent "chat t er" on chart recordings or "scat t er" in low-frequency
i mpedance data. Such random signals, often ascribed to instrumentation noise, are now rec-
ogni zed as resulting from distinctive phenomena that are of significant importance in
corrosion. Not e that low-frequency electrochemical corrosion data that do not conform to
the Krammers-Kroni g analysis are stochastic and probably a result of localized corrosion.
There are also sufficient reviews [ 1 - 3 , 4 9 , 5 1 ] , scientific publications [ 9, 52, 53] and informa-
tion on practical applications [ 5 , 5 1 , 5 4 - 5 6 ] available in the literature on EN, including pres-
entations at this symposia, to provide a formalization that will aid future scientific debate
and development of the subject.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
0
3

0
.
5
-

0
,
4
"

0
.
3
-

>

a

E

0
.
2
-

0
.
1
-

!
o
.

-
0
.
1
-

~

.
o
.
3

-
0
.
4

-
0
.
5

0

2
0
0

4
0
0

6
0
0

8
0
0

1
0
0
0

1
2
0
0

1
4
0
0

1
6
0
0

1
8
0
0

2
0
0
0

-
1
0

N

~

-
1
1

-
1
2

X

i

-
1
3

-
1
4

-
1
5

-
1
6

o

Q

-
1
7

-
1
8

A

S
q
u
a
r
e
/

H
z

1
0
~

1
0
"
=

I

%
,
I

1
0
"
~

"
t
i
m
e
,

s

F
r
e
q
u
e
n
c
y
,

H
z

b

U
l

1
0

F
I
G
.

3
-
-
(
a
)

C
u
r
r
e
n
t

n
o
i
s
e

r
e
c
o
r
d

f
r
o
m

p
o
t
e
n
t
i
o
s
t
a
t
i
c

m
e
a
s
u
r
e
m
e
n
t
s

o
f

i
r
o
n

i
n

I
M

s
u
l
p
h
u
r
i
c

a
c
i
d
;

(
b
)

P
o
w
e
r

s
p
e
c
t
r
a
l

d
e
n
s
i
t
y

p
l
o
t

o
f

c
u
r
r
e
n
t

n
o
i
s
e

[
6
]
.

m

r
-

m

0

o

-
1
-

m

>

r
'
-

z

o
_

m

C

m

m

Z

Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
2
.
0

1
.
6
t

a

1
.
2

~

0
.
8

~
;

V
V
:

v

g
V
~
'

,
v
v
'
C
Y
,

~

'
:
:
:

~

-
0
.
8
-

~

-
1
.
2
-

-
1
.
6
-

-
2
.
0

0

2
0
0

4
0
0

6
0
0

8
0
0

I

O
(
X
)

1
2
(
)
0

1
4
0
0

~

1
8
0
0

2
0
0
0

"
n
m
e
,

s

,
i
t

o

X

Z
-

Z
-

-
1

-
2

-
3

-
4

1
0
4

1
0
"
s

1
0
"
=

1
0
-
1

b

f

/
.
,

I

1
0

C
a
l
c
u
l
a
t
e
d
p
o
w
e
r
s
p
e
c
t
r
a
l
d
e
n
s
i
t
y

F
I
G
.

4
-
-
(
a
)

V
o
l
t
a
g
e

t
i
m
e

r
e
c
o
r
d
s

f
r
o
m

g
a
l
v
a
n
o
s
t
a
t
i
c

m
e
a
s
u
r
e
m
e
n
t
s

o
f

i
r
o
n

i
n

1
M

s
u
l
p
h
u
r
i
c

a
c
i
d
;

(
b
)

P
o
w
e
r

s
p
e
c
t
r
a
l

d
e
n
s
i
t
y

p
l
o
t

o
f

t
h
e

v
o
l
t
a
g
e

n
o
i
s
e

[
6
]
.

m

-
<

z

m

"
'
4

Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
8 ELECTROCHEMICAL NOISE MEASUREMENT
-18.0
-18.4
-18.8
I L l
-19.2
0
,
' i l : i '
I I t t , ~ i
I ' ' ~ : I ~ , , , i . l I P
i
f l I ~ I , ~ ! ; I ~1 ~ I I i : " q I , I 1 1
" , , . i , ~ i I ~ , ~ . ~ i ~ 1 ! ; I % 1 1 I I I I I
! I I ; I \ ~ I ;
I I I
20 40 60 80 100
"time, s
"~" 0
n
LLI
-1
I I ' l J l r [
t ~ 9 t }
i f [ j
1
! ; i
I " Ii
i [
I ~ i i 9
! : . , ! '
T i [ i i i
0 20 40 60 80 100
~me, $
FIG. 5 - - ECN and EPN transients produced by stress corrosion cracking at the f ree corrosion
potential.
R a n d o m P r o c e s s e s
Al l physi cal processes are di vi ded into det ermi ni st i c or nondet ermi ni st i c (nondet ermi ni st i c
bei ng r andom or st ochast i c) cat egori es. Det ermi ni st i c processes may be peri odi c or nonpe-
ri odi c (t ransi ent ); they are descr i bed by t i me- var yi ng functions, such as t hose pr oduced by
si nusoi dal and pseudor andom pert urbat i ons used in el ect rochemi cal i mpedance studies, or
sawt oot h and st ep functions, as used in pol ari zat i on and el ect r ocr yst al l i zat i on studies. In
convent i onal el ect r ochemi cal studies, the det ermi ni st i c i nput is t ypi cal l y cont r ol l ed by an
ext ernal pot ent i ost at i c or gal vanost at i c pert urbat i on.
EN phenomena bel ong to the general cat egory of r andom l ow-frequency processes. These
stochastic processes are descr i bed ei t her by pr obabi l i t y densi t y function equat i ons or in
st at i st i cal terms. Random processes can be st at i onary or nonstationary. The first moment s
(mean values) of a st at i onary process are invariate wi t h t i me. Some r andom corrosi on noi se
dat a can be descr i bed by a st at i onary r andom process mul t i pl i ed by a det ermi ni st i c factor.
In pract i ce, si ngl e- t i me records may be consi dered st at i onary i f the first moment s comput ed
over short -t i me intervals do not vary more than expect ed from normal st at i st i cal sampl i ng
variations. Hence, it may be possi bl e to treat some corrosi on syst ems, such as t hose with
l ow rates of uni form corrosi on, as weakl y st at i onary r andom processes.
Nonst at i onar y r andom processes have t i me- var yi ng functions that are det er mi ned by ob-
t ai ni ng i nst ant aneous averages over the set (ensembl e) of t i me records formi ng t he process.
One of the most compr ehensi ve st udi es of such ensembl es was by the Har wel l group on
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
KEYNOTE ADDRE S S 9
.O
m
I
$
0 4
i
i I F
t
- i l l
t I
ZH / ~V',O I. ' NON 1o O S d
V'd ~,ueuno
O N
ILL
o
"7
O
E
"E
2
O "~
' ~"C
C~
L h ' ~
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
10 E L E CT ROCHE MI CA L NOI SE ME A S URE ME NT
1.1
0.9
0.7
0.5
~ 0.3
0.1
-0.1
0 30 60 90 120 150
" t i me, s
FIG. 7--Current noise transients at 0.2 V versus SCE, A/ S/ 304 stainless steel in (a) 0.075M nCl O 4
+ 0.025M HCI, (b) in O.1M HCIO 4, and (c) specimen passivated in perchloric acid before chloride
addition [9,10].
current noi se t ransi ent s pr oduced duri ng pi t t i ng (Fig. 8). The dat a sets were generat ed by a
met hod in whi ch a pot ent i ost at i c sweep was t ermi nat ed at a fixed pot ent i al [58,59].
St andard met hods of anal ysi s are t ypi cal l y used to show noi se dat a as spectra; t hese are
represent at i ons of the t i me r ecor d as " aver ages " in t he frequency domai n di spl ay. Two
common mat hemat i cal pr ocedur es used to convert t i me domai n records i nt o frequency do-
mai n spect ra are t he fast Four i er t ransform ( FFF) and the maxi mum ent ropy met hod (MEM).
SO I
j ~A , ' i
25 . ~ p : " J ~ : t
- - p s . ~ ~ I f - " -
( - - ' # : o
r , /~&: ; . V ; e
" ~ 9 . k " ~ s
r . ;q, v J ' - ~ . . . , ~ . . .
0 1200 2400
" l i me, s
FIG. 8--Ensemble of current time transients on 18Cr:13Ni:lNb at 50 mV versus SCE in 0.028M
NaCI [57].
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
KEYNOTE ADDRESS 11
The PTT method [60] is widely used in many branches of engineering and science; it pro-
duces somewhat "noi s y" spectra and is appropriate for repetitive signals and data sets with
a reasonable number of sample points (>1040, 2080, and so forth). MEM was developed
by Burg [61] to analyze a limited number of sample points in geophysical studies and uses
filter coefficients to describe the data. Details of the computational procedures for MEM are
available [62], and MEM is considered by some to produce superior spectral representations
of data [63]. MEM produces smoot her spectral curves from which roll-off slopes and fre-
quency break points can he easily evaluated. Results t o date indicate that FFT and MEM
produce identical spectra from electrochemical noise [64,65]. Selection of the transformation
method (FFT or MEM) and noise data display (PSD or amplitude) would therefore appear
to be a personal choice.
Noise spectra are presented as either power spectral density (PSD) or amplitude (dB) plots.
In both cases, the log of frequency (Hz) is plotted either against l og spectral power density
( V/ Hz or A/ Hz ) or against the amplitude (dB) as log power spectral amplitude ( V/ ~/ Hz or
A/ ~/ Hz) . The noise amplitude is ~v / noise power, given by dB = 20 log (voltage ratio),
correlated according to the measured bandwidth. Not e that spectral analysis provides valid
information below 1/(the time record length) and above 1/ (2 x the sample period).
EN observed in corrosion can be classified under the general heading of low frequency
or 1 I f noise (sometimes referred to as 11f~' noise). This is true of potential noise measure-
ments made at the freely corroding corrosion potential. With some uniform corrosion pro-
cesses, the current noise amplitude spectra appear "whi t e"; with the current fluctuations in
the time domain, records produce a fiat spectral plot in the frequency domain, and the freely
corroding process is slow enough to appear as a weakl y stationary random process. The
major feature of 1/ f noise is that the noise amplitude in each measured bandwidth increases
with decreasing frequency. 1/ f noise is due to the randomness inherent in the system under
investigation and is associated with discrete Poisson stochastic-type processes. In some spec-
tral density plots, distinct peaks are observed in the 1/ f s l ope indicating a repetitive transient
of a specific frequency. This is analogous to the single "spi ke" seen in frequency domain
spectra when a constant frequency sinusoidal signal is used t o perturb a system t o obtain an
impedance value at a specific frequency as part of the overall spectra.
Noise Sources
The maj or sources of noise observed in electrochemistry and corrosion can be ascribed to
macroscopi c random (stochastic) phenomena. They include partial faradaic currents,
adsorpt i on/ desorpt i on processes, surface coverage, and, particularly in the case of localized
corrosion, the initiation of pits and mechanical effects resulting from cracking and erosion
processes. Al t hough similar 1/ f noise spectra are observed in studies from other scientific
fields, including biochemistry and electronics, this does not necessarily mean that the same
processes are involved. Nevertheless, the same basic principles of data interpretation apply
in all fields, even t hough converting this data into process information will be different
[66,67]. A common feature of 1/ f Poisson noise is that improvements in accuracy by in-
creasing the measurement time is not possible because it is a random signal. This is in
contrast to "whi t e" Gaussian noise in which the accuracy increases as the square root of the
measurement time.
The lowest magnitude or intensity of electrical noise generated in systems is the thermal
or Johnson noise as a result of the random mot i on of the electrons and charge carriers in
thermal equilibrium with their surroundings. This vibration occurs at temperatures above
absolute zero and is given by the Nyqui st formula [68,69]. Thermal noise is "whi t e, " with
the noise power directly proportional t o the measured bandwidth. Johnson noise produced
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
12 ELECTROCHEMICAL NOISE MF_.ASUREMENT
by t he measur ement i nst rument at i on noi se must be mi ni mi zed. In some studies i nvol vi ng
low levels of EN, one wi del y used approach is to use comput at i onal means to r emove the
i nst rument noise. Most pract i cal corrosi on studies pr oduce sufficient EN for the signal l evel
to be well in excess of the i nst rument at i on noise.
Shot noi se in el ect ri cal ci rcui t s is due to t he quant i t i zed nature of the el ect roni c charge
and is t ypi cal l y associ at ed with low levels of current flow in el ect roni c devi ces. Such fluc-
t uat i ons are great er than t hose pr oduced by t hermal noise, but convent i onal el ect ri cal shot
noi se is unl i kel y to be a maj or source of el ect r ochemi cal noi se in corrosi on.
Bet ween 1967 and 1972, Hooge [24-31] i nvest i gat ed noi se in met al s and semi conduct ors
and concl uded that the vari ance of the noi se was const ant with a Gaussi an di st ri but i on.
St udi es of syst ems in aqueous sol ut i ons [29] showed 1/ f and 1/f2 spectra, whi ch l ed to the
concl usi on that 1/ f noi se was due to the fluctuations in t he mobi l i t y of free charge carri ers,
not the change in the act ual number of carri ers. Thi s approach was not ext ended to corrosi on
studies.
Initial studies of noi se in el ect r ochemi cal syst ems by Tyagai [13-19], Baker [21-23], and
Fl ei schmann and Ol dham [31] have been revi ewed by Seral at han and Rangaraj an [32,33].
They have present ed det ai l ed mat hemat i cal model s based on the Langevi n met hod for as-
sessi ng fundament al processes. The Rangaraj an model s consi der el ect rochemi cal noi se as a
process i nvol vi ng faradai c shot noise; t hei r met hodol ogy provi des a fundament al t reat ment
of the previ ous studies of basi c el ect rochemi cal react i ons and part i cul arl y t hose by Tyagai
(but not corrosi on). It involves, first, consi derat i on of pr i mar y noi se sources, whi ch are
fundament al , but i nvari abl y uncorrel at ed (the Langevi n approach), and second, assessment
of macr oscopi c phenomena that provi de the secondar y source of noise. The t reat ment in-
cl udes equat i ons for conversi on of the current into power spectra. The macr ovar i abl es i ncl ude
part i al faradai c currents, el ect ron transfer, and chemi cal ki net i cs, but these are descr i bed as
di scret e Poi sson processes. The f or mal i zat i on was not ext ended to corrosi on studies such as
pit nucl eat i on and growth; it is nevert hel ess an appr oach that coul d be used to f or mal i ze
corrosi on processes.
These earl i er studies hi ghl i ght two maj or drawbacks to the use of EN in studies of redox
syst ems: (1) the noi se signal is rel at i vel y smal l and (2) as poi nt ed out by Barker [21], t here
is no part i cul ar advant age of noi se measurement s compar ed to det ermi ni st i c met hods (i m-
pedance) for t hese syst ems.
The i mport ance of noi se measurement s in the st udy of passi vi t y and its br eakdown was
r ecogni zed by the Japanese researchers Okamot o et al. in the mi d 1970s [35-37]. Thi s work,
t oget her with Ber t occi ' s earl y studies [43-45], were revi ewed in 1981 by Tachi bana and
Okamot o [49].
The devel opment of equat i ons to model pr obabi l i st i c (st ochast i c) processes was used
for bot h el ect r ocr yst al l i zat i on studies [46,72] and corrosi on [38,71,73,74]. A si mi l ar ap-
proach was adopt ed in bot h cases. The publ i cat i ons of Gabri el l i , Keddam, and Huet in Pari s
are part i cul arl y not ewort hy devel opment s [7,8,38,39,71,73,74,75-77], see al so Refs 1-3 for
t hei r reviews. For exampl e, Refs 2 and 3 show how model i ng the current t ransi ent s may be
consi der ed as a nucl eat i on or poi nt process that can be t ransformed to provi de a power
spectral densi t y of det ermi ni st i c events. Thi s part i cul ar corrosi on model uses a Poi sson pro-
cess to show how the frequency may vary wi t h the physi cal or chemi cal process; here a
PSD pl ot of 1/ f2 i ndi cat es a "sudden bi r t h" or "sudden deat h" process, whi l e a "sl ow bi r t h"
or "deat h pr ocess" gi ves 1 ]f4 plot.
The shapes of t ransi ent s pl ay a crucial rol e in det ermi ni ng the spect ral pl ot s [1,3]. In the
case of si mul t aneous measur ement s of current and pot ent i al at the corrosi on pot ent i al , current
t ransi ent s wi t h a slow ri se and a r api d decay will pr oduce spectral plots that are di fferent to
t hose wi t h a rapi d rise and a sl ow decay. Transi ent s wi t h a rapi d rise and fall or st at i onary
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
KEYNOTE ADDRESS 1 3
noi se sour ces can pr oduce whi t e noi se spect ra. For exampl e, anal ysi s o f powe r spect r al
dens i t y pl ot s f r om pot ent i al noi se on f r eel y cor r odi ng el ect r odes [78-80] gave 1/ f ' - f or pi t
i ni t i at i on and 1//f4 f or unst abl e pi t s (Fig. 9). However , in this cont ext , not e that, in t he f r ee
cor r os i on si t uat i on, it is the t r ans f or mat i on o f t he cur r ent noi se t hr ough t he el ect r ode i nt er-
f aci al i mpe da nc e t hat pr oduces t he pot ent i al noi se. Thus, it woul d a ppe a r t hat a whi t e noi se
cur r ent si gnal is r es pons i bl e f or t he pot ent i al noi se powe r spect r al dens i t y o f 1/ f 2 ( or 1 / f in
t he ampl i t ude pl ot ) whi l e a 1 / f cur r ent noi se s our ce gave ri se t o a p o we r spect r al dens i t y
pl ot o f 1/ f 4 f or t he pot ent i al noi se ( equi val ent t o a 1/ f 2 s l ope in t he ampl i t ude pl ot ).
I t is al so i nst r uct i ve when cons i der i ng noi se sour ces and anal ys i s o f noi se si gnal s, t o
eval uat e t he wor k by Wi l l i ams et al. o f the Har wel l Gr oup [57-59]. The i r me t hodol ogy
i nvol ved a r i gor ous exper i ment al and dat a col l ect i on pr oc e dur e t hat used set s o f dat a r at her
t han st udi es o f s i ngl e- t i me r ecor ds, but the anal ysi s is, in ma n y ways , anal ogous in appr oach
t o t hat adopt ed by t he Pari s school . The Har wel l st udi es l ed t o a gr eat er under st andi ng o f
pi t t i ng pr oces s es by def i ni ng unst abl e and st abl e pi t t i ng pr oces s es ( see be l ow f or f ur t her
det ai l s on pi t t i ng).
Noi s e sour ces in cor r os i on have been s hown t o be ma ny and vari ed. The s e i ncl ude:
(1) hydr oge n evol ut i on i nvol vi ng bubbl e nucl eat i on, gr owt h, and de t a c hme nt
[21,41,42,44]; changes in sol ut i on r esi st ance and di f f usi on i nt o pr opa ga t i ng mi cr o-
cr acks [10]; and di schar ge on f r eshl y e xpos e d met al at pr opagat i ng cr acks [51];
(2) pr opagat i ng st r ess cor r os i on cr acks i nvol vi ng met al di ssol ut i on or wat er di s char ge
[51, 81-86];
6 + X
t
~>
E
P,
CO
9
5 + X
4 + X
3 + X
2 + X
1 + X
O+ X
2 + X
- I + X
- 2 + X
- 3+) (
- 4+X
-4
R I I ' E : N ~ - . . .
[ X = O l ' I ~
- ~
C A S E 3
[ X = 3 !
-3 -2 -1 0
L OG F r e q u e n c y ( H z )
FIG. 9--Corrosion potential noise, pit initiation, and metastable pitting; experimental data and sim-
ulation showing 1/ f 2 and 1/ f 4 [79].
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
14 ELECTROCHEMICAL NOISE MEASUREMENT
(3) pit i ni t i at i on, met ast abl e pitting, and stable pi t growt h [9-11, 87-92];
(4) crevi ce corrosi on [47,93,94];
(5) mechani cal i mpi ngement and abrasi on [95];
(6) underfi l m cor r osi on [ 95- 97] ;
(7) passi ve syst ems [49,54,98];
(8) hi gh-t emperat ure processes [99-101];
(9) mi cr obi al - i nduced corrosi on [102-104]; and
(10) uni form corrosi on [55,105].
In some cases, mechani cal l y i nduced st ochast i c events occur (2, 3, and 5), in ot her syst ems,
st at i onary processes are i nvol ved (1, 6, and 10), whi l e mi grat i on of charge carri ers and sol i d-
state di ffusi on pr ocesses pr obabl y occur in (7) and (8). In all cases, noi se dat a from corrosi on
studies onl y i ndi cat e possi bl e t ypes of st ochast i c processes. The dat a have then to be inter-
pret ed to pr ovi de corrosi on i nformat i on or an underst andi ng in t erms of model s and corrosi on
pri nci pl es.
ENM Techniques
El ect r ochemi cal noi se dat a can be obt ai ned in a convent i onal manner usi ng pot ent i ost at i c
and gai vanost at i c t echni ques (Figs. 3, 4, 7, and 8). The current t i me records or pot ent i al t i me
records are then anal yzed to provi de i nformat i on on the corrosi on processes. Such measure-
ment s can be useful in research i nvest i gat i ons si nce onl y one measured par amet er varies
wi t h time. However, it must be r ecogni zed that i mposi t i on of a cont r ol l ed pot ent i al or con-
t rol l ed current, al t hough a val i d research met hod, may not al ways provi de a true represen-
tation of an act ual corrosi on situation. For exampl e, spect ra with a posi t i ve 1 I f sl ope coul d
be suspect; ei t her the i nst rument at i on or the t echni que used coul d i nt roduce an artifact.
An al t ernat i ve appr oach to pot ent i ost at i c and gal vani c measur ement s devel oped at UMI ST
[48,50,51,54,106-108] is to al l ow the natural corrosi on process to occur on the el ect r ode
and then to record the pot ent i al changes of the freel y cor r odi ng met al on a sensi t i ve di gi t al
voltmeter. A preferred vari at i on is to combi ne the corrosi on pot ent i al measur ement wi t h a
current measurement by coupl i ng t wo freel y cor r odi ng el ect rodes t oget her t hrough a sensi t i ve
zero-resi st ance ammeter. The t wo met ers provi de si mul t aneous t i me records of t he spont a-
neous changes in coupl i ng current and corrosi on pot ent i al . Anal ysi s of such dat a can be used
in ei t her research studies or pract i cal cor r osi on- moni t or i ng appl i cat i ons. Not e that it is the
nat ural l y occurri ng changes in corrosi on current flow, or the occurrence of current t ransi ent s
as a result of the st ochast i c nature of the basi c processes, that result in the obser ved changes
in corrosi on pot ent i al . In all corrosi on cases, it is the l ocal current flow (current noi se)
t hrough the i nt erfaci al i mpedance that gi ves ri se to pot ent i al changes (pot ent i al noise).
Potentiostatic and Galvanostatic Measurements
The use of pot ent i ost at i c measurement s has been a preferred met hod in many research
studies. The t echni que is part i cul arl y useful for i nvest i gat i ons of passi vi t y and its br eakdown
l eadi ng to pitting. Of maj or concern in earl y i nvest i gat i ons of el ect rochemi cal noi se in cor-
rosi on was the devel opment and use of low noi se pot ent i ost at s and gal vanost at s [38,40,42].
Ini t i al exper i ment s al so favored the use of a t wo-channel cross-correl at i on met hod and t wo
reference el ect rodes, all cont ai ned wi t hi n a faradai c cage. Thi s appr oach was abandoned
when it was r ecogni zed that in corrosi on studies most noi se si gnal s are si gni fi cant l y in excess
of the i nst rument at i on noise. Obvi ousl y, in all studies, it is necessary to use wel l - desi gned
i nst rument at i on wi t h a l ow i nst rument noi se out put to achi eve mi ni mum i nt erference f r om
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
KEYNOTE ADDRESS 15
the measurement equipment. Examples of the use of potentiostatic and galvanostatic mea-
surements are given below.
One of the most interesting studies of pitting corrosion on stainless steel was through the
evaluation of the statistics of ensembles of current-time transients by the Harwell group
[57-59]. In this approach, the potential is slowly increased, as in a conventional polarization
sweep, and the sweep is terminated when the current starts to rise. Any increasing current
is then logged as a current-time record containing a series of transients. A series of ensembles
of such current-time transients are thus obtained at a fixed applied "pi t t i ng" potential. These
data are then analyzed statistically in terms of the mean, standard deviation, and variance.
The pitting process is modeled as stochastic events randoml y distributed in time and space.
The events giving rise to the transients have a nucleation frequency, a probability of death,
a survival time probability, and an induction time. The pitting potential is shown to be a
statistical mean. The pitting process can be divided into nucleation of unstable (metastable)
and stable pits. The concept is that all pits have a finite lifetime but will terminate at different
pitting stages as either pit initiation events, unstable pits, metastable pits, or stable pits.
The relationships between the electrochemical behavior of the passive film on stainless
steel under potentiodynamic, nonsteady state conditions and the current fluctuations in the
prepitting range have been evaluated further using the probabilistic approach by Keddam et
al. [109]. Pit nucleation is described by a Poisson process. The amplitude and frequency of
the current noise decreased drastically with aging of the passive film. The capacitance mea-
surements correlated with the passive current and suggested an increasing charge storage
with completion of the full passive state.
In contrast to the large number of papers using potentiostatic measurements, there are few
reports of the use of noise measurements obtained galvanostatically. Of particular interest
has been the work by Gabrielli and Keddam [2] on noise generated by iron dissolution and
simultaneous hydrogen generation. The data are analyzed to demonstrate that the fluctuations
are due to the formation and release of hydrogen bubbles (Figs. 3 and 4). This analysis is
an extension of the original postulates by Baker [21] and Bertocci [41] that electrochemical
noise may be due to hydrogen bubble formation. The model has been further extended by
the Paris school to consider the diffusion of hydrogen into the metal [110].
Free Corrosion Measurements
Electrochemical Potential Noise Measurement
Many of the early observations of electrochemical noise were made by monitoring the
corrosion potential of freely corrodi ng electrodes [11,12,47]. Work sponsored jointly at
UMI ST by the United Ki ngdom Ministry of Defence and ICI between 1979 and 1985 was
directed at using this approach as a means of detecting and monitoring localized corrosion
in practical situations [48]. When used in combination with more conventional techniques,
such as linear polarization and electrochemical impedance, EPN monitoring of the corrosion
potential was able to discriminate between periods of uniform corrosion and localized cor-
rosion on-site and in a range of operating plants and test apparatus [111-114], for example,
Figs. 10. Other researchers have also suggested subsequently that potential noise measure-
ments may be used to quantify corrosion rates of steel in concrete [115,116] and active
dissolution of iron [2,117]. A significant number of workers have now reported on the use
of EPN in corrosion studies.
In the case of film breakdown and pitting corrosion, distinct transients with a characteristic
shape are observed for a range of passive metals. These comprise a rapid potential decrease
followed by an exponential type of recovery. Hagyard and Williams [11] in studies on alu-
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
16 ELECTROCHEMICAL NOISE MEASUREMENT
1 0 O . . . . . . . ~ . . . . . . . .
( m (~)
10O
1 0 2
10 ~
1 0 o
v .
o T i m e ( h ) 1 2
( a ) H i g h I n h i b i t o r / L o w C 1 ~ ( P a s s i v e )
1 0 ~
1 0 0
0 Tirne (h) 1 2
( b ) H i g h I n h i b i t o r / M e d . C 1 ~ ( P i t i n i t i a t i o n )
104
103
102
.ct~ (0),
Ec~(M)
.
~ m ~ m ~ . . ~ , . m ~ . . ~ . m
~.~ (~)
101 ~_. ( ~1
1 0 0
0 ~ r n e (h) 1 2
c ) L o w I n h i b i t o r / h i g h C 1 ~ ( G e n e r a l C o r r o s i o n )
1 0 4
10O ~
L a , = ( = v )
1 0 o i ~
0 Time (h) 1 2
( d ) M o d e r a t e I n h i b i t o r / M e d . C 1 ~ ( P i t t i n g )
FIG. lO--Corrosion monitoring in a cooling water system showing the effect of change of chloride
and inhibitor concentrations [ 121 ].
minum in 1961 ascribed these to a sudden rupture of the oxide film followed by repassi-
vation; the decrease in potential was between 4 to 40 V/ s. Williams [106] suggests that the
exponential type of recovery on stainless steel is due to recharging of the double layer. This
has been subsequently confirmed in other studies, for example, Fig. 11. The magnitude of
the transients and particularly the recovery time is dependent on the metal environment
circumstances. Steel in concrete samples has given rise to transients greater than 300 mV
with recoveries of up to 14 h [53] (Fig. 12). This reflects the slow diffusion of oxygen to
the repassivated electrode. On the other hand, when pitting transients in highly turbulent
fluids, such as stainless steel under an impinging jet, the recovery (Fig. 13) [92] may be a
few seconds or less as a result of the high diffusion rate of the cathodic species. The film
capacitance is also part of the interfacial impedance, hence the transient recovery will rise
quickly for aluminum compared to steels, while valve metals will produce a slower rise.
The analysis of corrosion potential transients on iron has been extended by Hashi mot o et
al. in a recent series of papers [78-80]. They have used spectral analysis to study the film
breakdown and, as discussed above, showed that the power spectral density decreased with
the frequency first to 1/f2 and then to 1/f4. The 1/ f2 Lorentz dependence shows that the
pit initiation follows a Poisson process. The transition to 1/f4 that occurred at about 10
Hz was ascribed to pit growth (Fig. 9). The model assumes that pit growth also occurs during
the rapid decrease in potential. (Simultaneous current transient measurements can provide
further information on this corrosion process, see Fig. 13.)
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
E
~ 2
Q
O
9
KEYNOTE ADDRESS 17
0 100 200 300
"time (seconds)
FI G. 1 l - - A n a l y s i s of the potential recovery f rom a pi t initiation event indicating a recharge mecha-
nism [91].
- 500
- 550
~ 0 0
~ 5 0
-700
-750
- 800
~ 5 0
~00
- 950
o1000
potential (mV SCE)
0 2 4 6 8 10 12 14 16 18 20
"time (hours)
FI G. 1 2 - - E x t e n d e d pot ent i al time transient f rom passi vat ed steel in chloride-contaminated concrete
[53].
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
18 ELECTROCHEMICAL NOISE MEASUREMENT
+2.3E-01
E (mV, SCE)
-5,0E-02
POTENTIAL
0 Seconds 204
+1.77E-01
E (mV, SCE)
+0.77E-02
b
0 Seconds 2040
+3.OE+Oo
l ( x l O~c m -z)
CURRENT
I . ,
+2.oE+O0 I CURRENT
1
-1.0E +00 " -1.38E+00
0 Seconds 204( 0 Seconds 2040
U n s t a b l e p i t t i n g S t a b l e pitting
FIG. 13--Current and potential noise transients showing: (a) pit initiation and unstable pitting, and
(b) stable pitting [92].
El e c t r oc he mi c al Cur r e nt Noi s e Me a s ur e me nt s f r o m Coupl e d El ect rodes
The use of coupl ed el ect rodes where t he coupl i ng current flow is moni t ored usi ng a zer o
resi st ance ammet er provi des a met hod of moni t ori ng current noi se on freel y corrodi ng spec-
i mens [107,108[. In most cases, nomi nal l y i dent i cal el ect rodes are used. In pri nci pl e, t here
shoul d be no net current flow bet ween the el ect rodes; however, in pract i ce, the st at i st i cal
vari at i on in corrosi on rates across surfaces means that current flow and el ect r ochemi cal noi se
are observed. Thi s process can be consi der ed as the experi ment al veri fi cat i on of the ori gi nal
Evans' concept of mobi l e anodi c and cat hodi c areas on a uni f or ml y cor r odi ng surface.
Changes in corrosi on pot ent i al of the coupl ed el ect r ode assembl y are measured usi ng a
t hi rd el ect rode, ei t her a reference el ect rode in the case of l aborat ory studies or in the case
of i n-pl ant moni t or i ng a pseudoreference and even a pi ece of met al of si mi l ar composi t i on
as the el ect r ode assembl y. Si nce in this l at t er case there are t wo uncorrel at ed pot ent i al
sources, then V,( . . . . . . . d) = X/ (V~ 2 + V22), where Vj and V2 are t he noi se si gnal s f r om the
coupl e and reference el ect rode, respect i vel y, and it is onl y requi red to di vi de the measur ed
pot ent i al noi se by X/ 2 to correct the signal. The t hree-el ect rode arrangement enabl es changes
in current noi se t hat pr oduce changes in corrosi on pot ent i al to be f ol l owed as t he envi ron-
ment al condi t i ons change.
In the case of l ocal i zed corrosi on, one of t he el ect rodes may generat e a pit, and hence,
the net current t ransi ent is observed. It is al so possi bl e to use speci al l y desi gned el ect rode
assembl i es in whi ch one el ect r ode is stressed; this may be used to eval uat e stress corrosi on
cracki ng [51,86]. Crevi ce corrosi on can be assessed usi ng an el ect rode covered by a cr evi ce
f or mer wi t h t he second el ect r ode act i ng as t he ext ernal cat hodi c ar ea [93,94,118].
Thi s UMI ST appr oach has been furt her devel oped to correl at e the current noi se and po-
t ent i al noi se si gnal s in bot h magni t ude and phase t o pr ovi de the el ect r ochemi cal i mpedance
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
KEYNOTE ADDRESS 19
of the corrosion system [77] (Fig. 14). In this context, Bierwagen has recently derived an
analysis of stationary noise for corrosion [119] using a probability density function to show
that low-frequency current noise, i,, and potential noise, E,,, are related through the interfacial
impedance, (E,,) ~ [Z] (i,,). This provides a more satisfactory pr oof for the use of simple
statistics [108] and use of the standard deviation, ~, of the low-frequency potential noise,
~r,., and current noise, %, signals to provide a noise resistance, ~r. = ~r i 9 R,,. At sufficiently
low frequencies, IZ] = R,,. This resistance is analogous to the linear polarization resistance
normally obtained by DC polarization techniques. The met hodol ogy is valid provided that
the system under investigation can be considered stationary during the measurement period.
A basic model of the combi ned use of current and potential noise to obtain uniform corrosion
rates is presented by Cottis and Turgoose at this symposi um [120].
When a uniform corrosion system becomes nonstationary (as localized corrosion starts),
then the ratios of the mean coupl i ng current to the standard deviation or the root-mean-
square (RMS) of the current noise can be used to identify the presence of localized corrosion.
For example, the onset of pit initiation increases the potential noise without a corresponding
increase in current noise (the current noise remains virtually white), but the amplitude of the
potential noise changes from pure 1/ f to l / f plus some l / f 2. Thus, the change in the coef-
ficient of variation indicates an increase of localized pitting corrosion compared to the uni-
form corrosion rate,
The simultaneous monitoring of current and potential noise signals was first used in 1984
by Eden [108,121] using simple analog instrumentation (Fig. 10). In 1985, the met hodol ogy
was then extended using digital equipment to identify galvanic corrosion (Fig. 1), uniform
corrosion, and localized corrosion [4,54,118]. This combi ned monitoring of corrosion poten-
tial and coupling current has been particularly useful in research studies of localized cor-
rosion, materials testing, and on-site for troubleshooting and process control. For example,
in the case of pitting, it can be shown that metal dissolution (electron flow from the pit site
l s x l O , , t - t -
I I I
~- - - %- - - - -
E
5 x 1 0 s
" ~ ' I I
- I I
I I
o f 1 "
I I
0 10 x 10 e 2 x 1 0 e
Real I mpedance, ohms
FI G. 14--1mpedance spectra obtained by correlation of current and potential noise, steel corroding
in chloride-contaminated concrete [ 1 2 9 ] .
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
20 ELECTROCHEMICAL NOISE MEASUREMENT
to externally connect ed metal surfaces) occurs during the initiation and propagation stages,
but there is no metal dissolution during the exponential-type recovery [51,91,92]. This sug-
gests that the anodic and cathodic reactions are separated in both space and time, and that
the electrons produced by metal dissolution are essentially held within the passive film and
double-layer capacitors. The potential recovery is then due to consumption of the increased
electron charge by the subsequent and continuing cathodic reaction that occurs across the
total surface.
ENM Characteristics for Various Modes of Corrosion
Uniform Corrosion
Active dissolution of iron and steel in acidic solutions is controlled by the hydrogen
evolution process. The noise signals obtained under galvanostatic and potentiostatic control
have been interpreted in terms of a stochastic process determined by bubble formation,
growth, and detachment [2].
Monitoring measurements made on mild steel in dilute sulfuric acid at the corrosion po-
tential indicate that resistance noise, charge transfer resistance from impedance measure-
ments, and weight-loss data provide similar results [51,91] (Fig. 15). Here the rate of reaction
increases with time as the carbon remaining from the alloy dissolution builds up to increase
the cathode area. In other systems, for example, in carbon-dioxide-containing brines and
aerated sodium chloride solutions, the corrosion rates fall with time as corrosion product
films develop (Fig. 16). Some organizations now recommend ENM for determining corrosion
rates [55]. It not only gives the same corrosion rate data as linear polarization resistance,
but it also provides information on localized corrosion.
The combi ned corrosion potential and current noise technique has also been successfully
used in a number of studies of coated steel [95-97, 122-128]. ENM and electrochemical
impedance provide similar indications of deterioration trends. However, EN~VI appears more
sensitive in the detection of corrosion resulting from through-film penetration of water during
t = 1
3o0 R e s i s t a n c e noi s e
~ 200 ,
0
100
0
0 20 40 60 80 100 120 140 160 180 2 0 0 2 2 0 2 4 0 2 6 0 2 8 0 3 0 0
Time (hours)
FIG. 15--Corrosion of mild steel in 0.5M sulphuric acid: a comparison of resistance noise, impedance
(charge transfer resistance), and weight loss data [91].
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
KEYNOTE ADDRESS 21
~ 2
i~,1 EN5 in 3 % NaCI ~ l m p e d a n c e
1 9 Noise Resistance
, . O Z R A Current
4 0
30
2 0 --"
e -
t.)
l o r
0 10 2 0 30 40 50 60 7 0 8 0 9 0 100
Time, h
FIG. 16--Corrosion of carbon steel in aerated 3% sodium chloride [70].
immersion in the test environment and the monitored resistance changes are quantified more
easily (Fig. 24).
Pitting Corrosion
As indicated previously, EN has been used by a number of investigators in studies of
pitting. Pitting is a stochastic process, and noise analysis has provided greater understanding
of the initiation and propagation stages. In particular, analysis of individual noise transients
is a useful research method for the study of fundamental corrosion mechanisms. Similar
information can be obtained from potentiostatic and free corrosion data. In studies at the
free corrosion potential, a "chemi cal potentiostat" is usually empl oyed in which small
amounts of ferric chloride are added to the sodium chloride test solution to raise the potential.
Initiation events are observed as short-lived transients. In some cases, transients may con-
tinue and give rise to unstable or metastable pits of limited life (Figs. 2, 7, 13a, and 17).
Onl y a small number of initiation events continue to form stable pits (Fig. 13b).
Most unstable and stable pits on stainless steel grow from an initiation event, which starts
to repassivate, but then appears to continue by a further reactivation process; the partial
recovery in the potential ceases as the anodic metal dissolution recommences. Metastable
pits have only a limited life and repassivate within a few seconds; stable pits continue to
grow for up to a few minutes (Figs. 13b and 18). Even stable pits have a finite life. There
are, therefore, probabilities associated with birth and death of the initiating transients and
the metastable and stable pits. The shape of the initial transients and other transients typical
of unstable and stable pits is of continuing research interest, as seen in the studies presented
by Pistorius at this symposium.
The noise trace from a stable growi ng pit is typically that of an actively corrodi ng metal
surface (Fig. 13b). Analysis of the current and potential relationships [92] show that in the
free corrosion situation, the pit is not driven by an IR effect, as with potentiostatic experi-
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
22 ELECTROCHEMICAL NOISE MEASUREMENT
Ct.
e*
o
2 0
10
0
0, 0 1.0 2.0 3.0 4.0 5.0
"rime I s
FIG. 1 7 - - Pi t initiation t ransi ent obtained potentiostatically; repassivation occurs wi t hout pi t growth
[10].
ments, but by the retained local environment. In this context, it is significant to note the
observations of Cavalcanti [130], who demonstrated the presence of a preexisting passive
film across a pit mouth, and also the more recent work by Burstein et al. [9, 10, 90] in which
initiation and metastable transients are consistent with the absence or presence of the original
passive film.
The partial coveri ng of a pit mouth would appear to play an important part in controlling
the diffusion processes in the incipient pit environment. This local environment in turn de-
termines the pit life and the type of noise transient observed. This t ype of retained film
model is confirmed by pitting studies at UMI ST [92] conducted under turbulent flow con-
ditions that showed that some stable (propagating) pits can be prematurely terminated by
eddy penetration into the pit from the liquid flow.
The aut hor' s opinion is that initiation transients are due to local film rupture events [98];
electrochemical pitting noise is mechanically triggered. This concept is based on the elec-
trostrictive model suggested by Sato [131] in which internal forces generated within the film
0. 4 1. 3
1.1
0. 9
0.5
0, 3
0.1
J , - - - -
0 ~ -0.1
5 10 15 0. 0 2. 0 4. 0 6, 0 8. 0
" ni ne I s 1 1 r n e I s
FIG. 18- - Met as t abl e pi t t i ng initiated by nucleation transients t hat do not immediately repassivate
[10].
0.3
~E 0. 2
8
0.1
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
KEYNOTE ADDRESS 23
are relieved by film rupture. The overall mechani sm involves adsorption of negatively
charged chlorine ions, and this adsorption increases as the electrode potential rises towards
the pitting potential. The change in adsorption affects the film surface tension which in turn
changes the pressure within the film. A film rupture event that occurs at an area of high
chloride adsorption (or salt island) has a greater probability of continuing to a metastable or
stable pit. Other authors have considered diffusion processes associated with the formation
of salt islands [132].
Measurements on-site used to moni t or plant corrosion also exhibit stochastic noise pro-
cesses that change in response to the environmental conditions. In general, there are often
gradual changes from uniform corrosion to pitting and back again in many industrial situa-
tions. The presence of active corrosion in combination with pit initiation can be seen in
many noise-monitoring data records. Nevertheless, in-plant data frequently show evidence
of typical pit initiation and metastable pitting processes described above (Fig. 2), but onl y
infrequently is stable pitting detected.
Electrochemical noise measurements on pitting, either by analysis of individual transients
or a series of transients, is able to provide detailed information on the fundamental processes
involved. The actual stochastic processes are influenced by factors such as the age and
thickness of the passive film, the composition of the environment, the electrode potential,
diffusion restrictions as a result of pit geometry, and flow enhancement in turbulent
environments.
Crevice Corrosion
Crevice corrosion on stainless steel is typically observed as pitting, but with the anodic
dissolution assisted by restricted diffusion produced by the crevice geometry. Initial film
rupture or breakdown occurs within the crevice, and then, provided the corrosion potential
is sufficiently anodic, the metal dissolution will continue as a large current transient
[93,94,118,133]. This noise transient is similar to that produced by a stable pit but with an
extended life that continues over a number of hours.
In the case of iron or carbon steel, which has a less protective film, regular potential
oscillations may be observed (Fig. 22). These appear to be characteristic of diffusion control
but are more likely to be due t o the formation and dissolution of unstable films in the
restricted crevice environment [47]. Similar oscillations have been observed with stress-
assisted intergranular corrosion [51] (Fig. 23) and unstable film formation and dissolution
[135].
Stress Corrosion Cracking
Electrochemical noise measurements have been reported for bot h potentiostatic controlled
experiments [81,84,85] (Fig. 21), and corrosion monitoring at the free corrosion potential
[51,82,83,86] (Figs. 19 and 20).
Anodi c transients are the result of metal dissolution at the crack wall and are therefore
similar to pit initiation. In many cases, there could be difficulty in the ability to discriminate
between anodic SCC, pitting, and crevice corrosion. Extensive active corrosion outside the
crack can mask the cracki ng transients [82,118,136].
Cathodic transients can be observed in some SCC systems (Figs. 19 and 20). These result
from the hydrogen generation reaction that occurs on the newly fractured metal surface of
a propagating crack. This freshly exposed surface is highly catalytic, and hence, conversion
of adsorbed water to atomic hydrogen occurs rapidly at the base" of the crack. The fast
cathodic reaction consumes electrons that are removed from the film and the electrochemical
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
24 ELECTROCHEMICAL NOISE MEASUREMENT
7. 6
> 7.2
E
i
o.
uJ 6.8
6. 4
0 1 O0 2 0 0
"rime, s
-12
-10
-8
z"
o
LLI
-6
-4
. l I I ,
I I I I I L
I , I I I I d i l l , ,
I , I , i1 1 ,1 - I , H I I I d ,
J I L I I I H l i k l , l , H H , I 1 1 1 4
" , ~ ~
0
I
I . I I J .
. 1 1 II I . J l H L
II1,11 I I I 1 ~ 1
I l J l | I d d i l | l
l oo
"lime, S
i , I
J l t i
i i . I I I I I
200
FIG. 19--Rapid stress corrosion cracking; hydrogen discharge within the crack [51].
double layer present on the metal surface, both external to the crack and on the crack walls.
The electrochemical interface is then recharged by subsequent metal dissolution. The ob-
served initial potential and the current transients therefore both increase in a positive direction
during the propagation steps. Rapid branched cracking (Fig. 19) and slowly propagating
single cracks (Fig. 23) have been observed with electrochemical noise. Figure 23 also shows
an anodic metal dissolution transient following the initial cathodic transient.
Laboratory Testing and Product Testing
EN has been used in a number of inhibitor studies and evaluations [88,134,137-140].
Corrosion rate data and indications of pitting are typical examples of these applications.
One of the most significant ongoi ng developments over the past ten years has been the
use of EN in coatings evaluations [141]. Particularly noteworthy is the work by Skerry et
al. [56, 95-97, 122, 124, 125, 128] that has demonstrated that noise can now be considered as
a standard test procedure. Its main advantages are ease of application and production of
quantifiable data which, unlike impedance measurements, are also easy to interpret.
Typical data can be presented as a change of noise resistance with time of exposure. In
Fig. 24, the coatings were assessed as maintenance painting systems in a salt spray test; the
acrylic had a good initial performance but quickly deteriorated to give a value similar to the
mild steel control, and the preferred option was the urethane alkyd. In general, through-
coating corrosion is a relatively slow process with good systems, and measured parameters
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
KEYNOTE ADDRESS 25
- ~ . o : i ~ I I ' J I ]
I ~ ~ J i t
I I ! ~ i 1
-60.0
> , . . !..k ! . , , ! ' ~ t . . . .
. . ,
- 62. 0
' " I ! i ~ i I I !
- 6 4 . 0 i , i , I ! i i I 1
-66.0 --
0 2 0 40 6 0 8 0 100
Time, s
o 0
tu %
t
i i
b
f l f i l i
"0.8
0 2 0 40 60 8 0 100
~mo, s
FIG. 20--Slow stress corrosion cracking propagation: (a) hydrogen discharge; (b) hydrogen discharge
followed by metal dissolution [51].
do not change rapidly during the test exposure period. EN is, however, very sensitive to any
change, and hence, trends of resistance noise with exposure time are ideal for coating eval-
uations. The data are more amenable to simple analysis involving water uptake and increased
corrosion of the substrate than that provided by electrochemical impedance.
Corrosi on Moni tori ng of Industrial Plants
The advantages of electrochemical noise for corrosion-monitoring applications have
been demonstrated in numerous publications over the past ten years [3,5,51,54,113,114,121,
144-147].
Evaluations have been made of acid dew point corrosion in flue gas exhaust equipment
and duct work [113] (Fig. 25), flue gas desulphurizing systems [142,143] (Fig. 26), boiler
plant [144], cooling water systems [5,121,145], acid cleaning, oil and gas production, process
plant [121], and reinforced concrete [112,115,147].
The mai n emphasis has been troubleshooting, assessment of process control as it affects
corrosion control, effect of environmental changes, and inhibitor evaluation. Assessment of
corrosion rates, identification of the onset of localized corrosion, and trend monitoring are
typical of most studies. Studies of microbially induced corrosion have also been reported
[148].
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
26 E L E C T R O C H E M I C A L N O I S E M E A S U R E M E N T
500
r
o
U
2
( 1 )
i
O n s e t o f
s t a b l e
c r a c k i n g
P o t e n t i a l ~ ~ . t
J
P o t e n t i a l ,
! ! 1 1 1 g o ~ s m o r e
I C u r r , - t , , - -
~
(to approx, s ~ )
" t i m e f r o m y i e l d I h o u r s
I
1 5 5 . 5
- I O0
- 0 >
E
- l O o
- 2 0 0
F IG . 21- - Cur r ent noise transients obtained in potentiostatically iron in tO-ppm thiosufphate [8 4 ].
a b
E
- 6 L , , . J _ . k L L I . ~ . . . . t . . , I k . ' _ ) ~ , , I
(mV) -8 V'L411hll.adA,t/l~| ~ '
I I I I W I I I I M I I I I I
))IIllliIllll)I I 11
I I I I I m l l l l l l l l l I
0. 1 I i ~ - 12
o_ I I I I l m l l l l l l l l l I I
w I I I l i l l l i i l l l l i l I I
-14 III11111111 I
o I
- 0. 1 100 2 0 0
" r i me , s
- 30 i ' L
i
10 2 0 t ( r ni n)
.2o i i
z l i H i l l i l l ) ,?, :,
0 100 2 0 0
' ~m0, $
F IG . 2 2 - - ( a ) Potential noise f rom crevice corrosion [ 4 7 ] and (b ) pot ent i al and current noise f rom
pl ant monitoring.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
KEYNOTE ADDRESS 27
- 4 3 4 i,m
i
- 4 3 6
~ -438
uJ -440
S
t I
0 20
x.=
4 0 6 0 8 0 1 0 0
~me, s
60
40
'~. 20
O 0
iii
-20
-40
0 2 0 4 0 6 0 8 0 1 0 0
"time, s
FIG. 23--Oscillating current noise from stress-assisted intergranular corrosion typical of a micro-
crevice [51 ].
Concl usi ons
The above review has outlined the major parameters and factors to be considered in
undertaking EN measurements and their evaluations. EN has been shown to be useful in
fundamental research studies; for testing of materials, coatings, and inhibitors; on-site for
online corrosion monitoring both for process control and surveillance; and for troubleshoot-
ing or investigation of plant corrosion problems. An appreciation of the likely sources of
noise and their measurement is essential to recogni ze the advantages and limitations of the
techniques.
Noise analysis of time records for research purposes is not necessarily a trivial task. It
may require detailed analysis of individual transients to provide information on the funda-
mental processes that underlie the corrosion mechanisms. In other cases in which a series
of transients are evaluated, a modeling exercise is often used to convert a Poisson process
into a power spectral density plot (the frequency domain representation) by means of the
Fourier transform. This modeled data, which describes the noise transients in fundamental
terms, is then compared with experimental data to propose viable mechanisms.
In practical situations, the shapes of the random noise and the transients can be used to
discriminate between uniform corrosion and various forms of localized corrosion. A simple
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
28 ELECTROCHEMI CAL NOI SE MEASUREMENT
10 a
9
E 10 s
r
0
8"
9 ~ 1 0 4
O
N
' = 10 3
n o
1 0 2
x - E po xy
9 - U r e l h a n e A l k y d
+ - A c r y l i c
9 9 - M i l d S t e e l
I
. . . / I \ < - , ^ .
5 9 13 17 21 25 29 33 37
T i m e ( d a y s )
F I G . 24- - Changes in coating resistance obtained by monitoring the corrosion current and noise [ 146] .
st at i st i cal i nt erpret at i on of the EN dat a based on the first and second moment s of the random
signal (the st andard devi at i on and the vari ance, respect i vel y) can al so pr ovi de i nformat i on
on uni form corrosi on rates (where ~,./~i -- R,,, ico~ = B/R,,, and B is a St er n- Gear y const ant )
and the degr ee of l ocal i zed act i vi t y (from ~ili . . . . . ). Passi ve syst ems have low levels of noi se
200 . . . . . 9 . . . . . . . . . . . . . . . . . . . . . . . . .
f
i
, ElM
i
i ' ~ = Fl ue gas temperature
_ . ~ . E 0 N . . ~
1 2 3
t i me (hours)
F I G . 25- - Onl i ne monitoring during a boiler startup showing dilute sulfuric aci d condensation [ 1 1 1 ] .
150
G"
~" 100
5O
0
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
KEYNOTE ADDRESS 29
20K
10k
0
i
Impedance
v m
I m A
lOl.tA
Coupling current
I I _ l t
lOmV
loo~
t ~
1/JA ~Current n o i s e
1 o n A
0 . 1 n A ~ c "i" - -
(~
100
0
Pmbetemp.
, t
m
i
1, / . , , , I
(~ ~Gas temp.
0 6
t
12
L ~
I
Absorber Reheater
o n - l i n e s t a r t - u p
FIG. 26--On-line monitoring in a flue gas desulphurization plant.
with potential fluctuations of about 1 mV and current fluctuations of less than a few
i~A/cm ~ of electrode area. Uniform corrosion has an increased noise level, typically a few
10s mV and up to 10 ~A/ cm 2, while localized corrosion gives rise to distinct transients up
to 100s mV and 100s i~A/cm 2.
As to whether "electrochemical noise is the definitive electrochemical corrosion tech-
nique" will only be fully answered in the long term as further scientific information becomes
available. However, the literature presented demonstrates that ENM provides some of the
most fundamental information on the basic (stochastic) processes involved in electrochemical
corrosion reactions. Some current and potential noise is also sufficiently stationary to be
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
30 ELECTROCHEMICAL NOISE MEASUREMENT
transformed into electrochemical impedance spectra, and this will provide an alternative way
to present EN data. It is also likely in the future that EN data will be analyzed using chaos
theories.
At a more practical level, EN information has to date been used in the evaluation of
materials, coatings, and inhibitors. In particular, ENM could be adopted as a standard test
method for coatings and used to determine corrosion rates. Technical developments have
also enabled EN to be used on-site for troubleshooting and corrosion control.
Acknowl edgment s
The author wishes to express his appreciation to the many students, research workers,
colleagues, and friends at UMIST, CAPCIS, and RTCML and to researchers overseas who
contributed in many ways to his understanding of electrochemical noise. In particular, the
research support from UK MoD (John Rowtands) and ICI (the late Mervyn Turner) for Les
Callow and Karl Hladky is gratefully acknowledged, as this was responsible for initiating
the electochemical noise development. The research of Pete Searson, Bill Cox, Wai Him
Mok, Ihasan A1-Zanki, Jorge Uruchurtu, and Dave Moore is also appreciated, as were the
stimulating discussions with Dave Eden over the past 15 years. Acknowledgments are also
given to CAPCIS for support in preparing this review and RTCML for the contribution to
expenses in attending the electrochemical noise symposium.
Re f e r e nc e s
[1] Gabrielli, C., Huet, E, Keddam, M., and Oltra, R., Corrosion, Vol. 46, 1990, p. 266.
[2] Gabrielli, C. and Keddam, M., Corrosion, Vol. 48, 1992, p. 794.
[3] Gabrielli, C., Huet, E, and Keddam, M., "Investigation of Metallic Corrosion by Electrochemical
Noise Techniques," in Electrochemical and Optical Techniques f or the Study and Monitoring of
Corrosion, M. G. S. Ferriera and C. A. Melendres, Eds., 1993, p. 135. Electrochemical Society,
Inc., Princeton, NJ.
[4] Warris, A., "The Monitoring of Sacrificial Anode Galvanic Couples," M.Sc. dissertation, Univer-
sity of Manchester, 1985.
[5] Winters, M. A., Stokes, E S. N., Zuniga, E O., and Schlottenmeier, D. J., Corrosion Science,
1993.
[6] Gabrielli, C., Huet, E, and Keddam, M., Journal of Applied Electrochemistry, Vol. 15, 1985,
p. 503.
[7] Gabrielli, C., Huet, E, Keddam, M., Oltra, R., and Colson, J. C., Materials Science Forum, Vol.
8, 1986, p. 491.
[8] Gabrielli, C., Huet, E, and Keddam, M., Electrochimica Acta, Vol. 31, 1986, p. 1025.
[9] Burnstein, G. T., Pistorius, E C., and Martin, S. E, Corrosion Science, Vol. 35, 1993, p. 57.
[10] Mattin, S. P. and Burstein, G. T., in Progress in the Understanding and Prevention of Corrosion,
Vol. 2, Proceedings of the tOth European Corrosion Congress, Institute of Materials, London,
1993, p. 1109.
[11] Hagyard, T. and Williams, J. R., Transactions of the Faraday Society, Vol. 57, 1961, p. 2288.
[12] Iverson, W. P., Electrochemical Society, Vol. 115, 1968, p. 617.
[13] Tyagai, V. A., Elektrokhimiya, Vol. 3, 1967, p. 1331.
[14] Tyagai, V. A., Elektrokhimiya, Vol. 7, 1971, p. 6.
[15] Tyagai, V. A., Elektrokhimiya, Vol. 10, 1974, p. 1.
[16] Tyagai, V. A., Electrochimica Acta, Vol. 16, 1971, p. 1674.
[17] Tyagai, V. A. and Ya. Kolbasov, G., Zeitschrift Fiir Physikalische Chemic (Leipzig), Vol. 251,
1972, p. 122.
[18] Tyagai, V. A. and Lukjanchikova, N. B., Surface Science, Vol. 12, 1968, p. 331.
[19] Tyagai, V. A. and Ya. Kolbasov, G., Elektrokhimiya, Vol. 7, 1971, p. 299.
[20] Tyagai, V. A., Electrical Phenomena at the Biological Membrane Level, E. Roux, Ed., Elsevier
Science Publishing Co., London, 1977, p. 325.
[21] Barker, G. C., Journal of Electroanalytical Chemistry, Vol. 21, 1969, p. 127.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
KEYNOTE ADDRESS 31
[22] Barker, G. C., Journal ofElectroanalytical Chemisto; Vol. 39, 1972, p. 484.
[23] Barker, G. C., Journal of Electroanalytical Chemistry, Vol. 82, 1977, p. 145.
[24] Hooge, F. N. and Hoppenbrouwers, A. M. H., Physica, Vol. 42, 1969, p. 331.
[25] Hooge, E N. and Hoppenbrouwers, A. M. H., Physica, Vol. 45, 1969, p. 386.
[26] Hooge, E N., Physics Letters, Vol. 29A, 1969, p. 139.
[27] Hooge, E N., Physics Letters, Vol. 29A, 1969, p. 642.
[28] Hooge, E N., Physics Letters, Vol. 33A, 1970, p. 169.
[29] Hooge, E N. and Gaal, J. L. M., Phillips Research Reports, Vol. 26, The Netherlands, 1971,
p. 77.
[30] Hooge, E N., Physica, Vol. 60, 1972, p. 130.
[31] Fleischmann, M. and Oldfield, J. W., Journal of Electroanalytical Chemistry, Vol. 27, 1970,
p. 207.
[32] Seralathan, M. and Rangarajan, S. K., Journal of Electroanalytical Chemistry, Vol. 208, 1986,
p. 13.
[33] Seralathan, M. and Rangarajan, S. K., Journal of Electroanalytical Chemistry, Vol. 208, 1986,
p. 29.
[34] Blanc, G., Gabrielli, C., and Keddam, M., Electrochimica Acta, Vol. 20, 1975, p. 687.
[35] Okamoto, G., Sugita, T., Nishiyama, S., and Tachibana, K., Boshoki Gijutsu, Vol. 23, 1974,
pp. 439 and 445.
[36] Okamoto, G., Tachibiana, K., Nishiyama, S., and Sugita, T., Passivity and Its Breakdown on Iron
and Iron Base Alloys, R. Staehle and H. Wand Okada, Eds., National Association of Corrosion
Engineers, Houston, 1976, p. 106.
[37] Okamoto, G., Sugita, T., Nisbiyama, S., Tacbibana, K., and Takaishi, T, Proceedings of the 6th
International Congress on Metal Corrosion, Australia, 1975.
[38] Blanc, G., Gabrielli, C., Ksouri, M., and Wiart, R., Electrochimica Acta, Vol. 23, 1978, p. 337.
[39] Gabrielli, C., Ksouri, M., and Wiart, R., Journal of Electroanalytical Chemistry, Vol. 86, 1978,
p. 233.
[40] Epelboin, I., Gabrielli, C., Keddam, M., and Raillon, L., Journal of Electroanalytical Chemistry,
Vol. 105, 1979, p. 389.
[41] Bertocci, U., Proceedings of the 7th International Congress Metallic Corrosion, Associacao Bras-
ilereira de Corrosao, Rio de Janeiro, 1979.
[42] Shideler, R. W. and Bertocci, U., Journal of Research of the National Bureau of Standards, Vol.
85, 1980, p. 211.
[43] Bertocci, U., Journal of the Electrochemical Society, Vol. 127, 1980, p. 931.
[44] Bertocci, U. and Krugger, J., Surface Science, Vol. 101, 1980, p. 608.
[45] Bertocci, U., Journal of the Electrochemical Society, Vol. 128, 1981, p. 520.
[46] Fleischmann, M., Labram, M., Gahrielli, C., and Sattar, A., Surface Science, Vol. 101, 1980,
p. 583.
[47] Hladky, K. and Dawson, J. L, Corrosion Science, Vol. 21, 1981, p. 317.
[48] Hladky, K., European Patent 084404A3, USA Patent 455709, Canadian Patent 418938.
[49] Tachibana, K. and Okamoto, G., Reviews on Coatings and Corrosion, Vol. 3, 1981, p. 229.
[50] Hladky, K. and Dawson, J. L., Corrosion Science, Vol. 22, 1982, p. 231.
[51] Eden, D. A., Rothwell, A. N., and Dawson, J. L., in Corrosion/91, Paper 444, National Association
of Corrosion Engineers, Houston, 1991.
[52] Skerry, B. S., Chen, C-T., and Ray, C. J., Journal of Coatings Technology, Vol. 64, 1992, p. 77.
[53] Searson, E C., "Electrochemical Noise and Corrosion of Steel in Concrete," Ph.D. thesis, Uni-
versity of Manchester, 1982.
[54] Dawson, J. L., Farrell, D. M., Aylott, P. J., and Hladky, K., CORROSION~89, Paper 31, National
Association of Corrosion Engineers, Houston, 1989.
[55] Webster, S., Nathanson, L., Green, A. G., and Johnson, B. V., in Proceedings of the 12th Inter-
national Corrosion Congress, National Association of Corrosion Engineers, Houston, 1993.
[56] Moon, M. and Skerry, B. S., Polymer Preprints, Vol. 35, 1994, p. 303.
[57/ Williams, D. E., Westcott, C., and Fleischmann, M., Journal of the Electrochemical Society, Vol.
132, 1985, p. 1796.
[58] Williams, D. E., Westcott, C., and Fleischmann, M., Journal of Electroanalytical Chemistry, Vol.
180, 1984, p. 549.
[59] Williams, D. E., Westcott, C., and Fleischmann, M., Journal of the Electrochemical Society, Vol.
132, 1985, p. 2804.
[60] Cooly, J. W. and Tukey, J. W., Mathematics of Computation, Vol. 19, 1965, p. 297.
[61] Burg, J. P., "' Maximum Entropy Spectral Analysis," in Proceedings of the 37th International SEG
Meeting, 1967.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
32 ELECTROCHEMICAL NOISE MEASUREMENT
[62] Ables, J. G., Astronomy and Astrophysics, A European Journal Supplement Series, Vol. 15, 1974,
p. 383.
[63] Andersen, N., Geophysica, Vol. 39, 1974, p. 69.
[64] Searson, P. C. and Dawson, J. L., Journal of the Electrochemical Socie~., Vol. 135, 1988, p. 1908.
[65] Cottis, R. A. and Loto, C. A., Corrosion, Vol. 46, 1990, p. 2.
[66] van der Ziel, A., Fluctuation Phenomena in Semi-Conductors, Butterworth, London, 1959.
[67] van der Ziel, A., Noise--Sources. Characterization, Measurement, Prentice Hall, Englewood
Cliffs, NJ, 1970.
[68] Nyquist, H., Physics Revue, Vol. 129, 1927, p. 614.
[69] Nyquist, H., Physics Revue, Vol. 32, 1928, p. 110.
[70] Rothwell, A. N., private communication, CAPCIS-March, Ltd., Manchester, UK, 1992.
[71] Cachet, C., Gabrielli, C., Huet, E, Keddam, M., and Wiart, R., Electrochimica Acta, Vol. 28, 1983,
p. 899.
[72] Budevski, E., Comprehensive Treatise ofElectrochemistr3,, Vol. 7, Plenum Press, New York, 1983,
p. 399.
[73] Gabrielli, C., Huet, E, Keddam, M., Colson, J. C., and Oltra, R., Proceedings of the International
Congress of Metallic Corrosion, Ottawa, Vol. 3, 1984, p. 427.
[74] Gabrielli, C., Huet, E, Keddam, M., and Macias, A., Proceedings of the Electrochemical Socie~.
Conference, Vol. 86-7, 1986, p. 507.
[75] Gabrielli, C., Huet, E, Keddam, M., and Oltra, R., Proceedings of the Advances in Localized
Corrosion. National Association of Corrosion Engineers, Houston, 1990, p. 93.
[76] Gabrielli, C., Huet, F., and Keddam, M., "Investigation of Metallic Corrosion by Electrochemical
Noise Techniques," in Proceedings of the Conference on Electrochemical and Optical Techniques
for the Study and Monitoring of Metallic Corrosion, Kluer Academic Publishers, 1991, p. 135.
[77] Gabrielli, C., Huet, E, and Keddam, M., Journal of Chemist~ and Physics, Vol. 99, 1993,
p. 7240.
[78] Hashimoto, M., Miyajima, S., and Murata, T., Corrosion Science, Vol. 33, 1992, p. 885.
[79] Hashimoto, M., Miyajima, S., and Murata, T., Corrosion Science, Vol. 33, 1992, p. 905.
[80] Hashimoto, M., Miyajima, S., and Murata, T., Corrosion Science, Vol. 33, 1992, p. 917.
[81] Isaacs, H. S, "Initiation of Stress Corrosion Cracking on Sensitized Type 304 Stainless Steel,"
Final Report, EPRI Research Project RP 1167-8, 1985.
[82] Cottis, R. A. and Loto, C. A., Materials Science Forum, Vol. 8, 1986, p. 201.
[83] Loto, C. A. and Cottis, R. A., Corrosion, Vol. 45, 1989, p. 136.
[84] Stewart, J., Scott, P. M., Williams, D. E., and Wells, D. B., Corrosion Science, Vol. 33, 1992,
p. 73.
[85] Wells, D. B., Stewart, J., Davidson, R., Scott, P. M., and Williams, D. E., Corrosion Science, Vol.
33, 1992, p. 13.
[86] Heaver, E. E., Lawson, K., and Eden, D. A., in UK CORROSION~93, Institute of Corrosion,
London, 1993.
[87] Bertocci, U., Mullen, J. L., and Ye, Y. X., in Proceedings of the 5th International Symposium on
Passivio' and Metallic Semiconductors, M. Froment, Ed., Elsevier, New York, 1983, p. 229.
[88] Gollner, J., Garz, I., and Meyer, K., Korrosion (Germany), Vol. 17, 1986, p. 244.
[89] Nachstedt, K. and Heusler, K. E., Electrochimica Acta, Vol. 33, 1988, p. 311.
[90] Burstein, G. T. and Mattin, S. P., Phil. Mag. Letters, Vol. 66, 1992, p. 127.
[91] Brown, S. A., "Electrochemical Noise in Corrosion," M.Sc. dissertation, University of Manchester,
1990.
[92] Shih, C. C., "Fl ow Affected Corrosion--Laboratory Studies and Models," Ph.D. thesis, University
of Manchester, 1991.
[93] Simoes, A. M. P. and Ferreira, M. G. S., British Corrosion Journal Vol. 27, 1987, p. 21.
[94] Simoes, A. M. P. and Ferreira, M. G. S., in Proceedings of the lOth International Congress on
Metal Corrosion, Vol. IV, Key Engineering Material Vols. 20-28, 1988, p. 3125.
[95] Eden, D. A., Hoffman, M., and Skerry, B. S., in Polymeric Materials for Corrosion Control, ACS
Symposium Series 322, R. A. Dicke and E L. Floyd, Eds., American Chemical Society, 1986.
[96] Eden, D. A. and Skerry, B. S, in Advances in Corrosion Protection By Organic Coatings, Cam-
bridge, 1989.
[97] Chen, C. T. and Skerry, B. S., Corrosion, Vol. 47, 1991, p. 598.
[98] Dawson, J. L. and Ferreira, M. G. S., Corrosion Science, Vol. 26, 1986, p. 1009.
[99] Bottomley, P. D. W., Gill, J. S., and Dawson, J. L., Materials Science Forum, Vol. 8, 1986,
p. 509.
[100] Gao, G., Stott, F. H., Dawson, J. L., and Farrell, D. M., Oxidized Metal Vol. 33, 1990, p. 79.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
KEYNOTE ADDRESS 33
[101] Whitlow, G. A., Mok, W. Y., Gallagher, E, Cox, W. M., Elliott, E, and Lee, S. Y.,
CORROSION~91, National Association of Corrosion Engineers, Houston, 1991.
[102] Iverson, W. E and Heverly, L. E, in Corrosion Monitoring in Industrial Plants Using Nondestruc-
tive Testing and Electrochemical Methods, ASTM STP 908, American Society for Testing and
Materials, West Conshohocken, PA, 1986, p. 459.
[103] Sequeira, C. A. C., in Proceedings of the Conference on Microbiological Corrosion, Elsevier
Applied Science, London, 1988, p. 99.
[104] Dexter, S. C., Duquette, D. J., Siebert, O. W., and Videla, H. A., Corrosion, Vol. 47, 1991,
p, 303.
[105] AI-Zanki, I., Gill, J. S., and Dawson, J. L., Materials Science Forum, Vol. 8, 1986, p. 463.
[106] Williams, D. E., in Electrochemical Corrosion Testing, Dechema. Monograph, Vol. 101, 1986,
p. 253.
[107] Eden, D. A., Dawson, J. L., and John, D. G., UK Patent App. 861158, May 1986, US Patent
5139627.
[108] Eden, D. E., Hladky, K., John, D. G., and Dawson, J. L., in CORROSION~86, Paper 274, National
Association of Corrosion Engineers, Houston, 1986.
[109] Keddam, M., Krarti, M., and Pallotta, C., Corrosion, Vol. 43, 1987, p. 454.
[I10] Benzaid, E., Huet, E, Gabdelli, C., Jerome, M., Wenger, E, and Galland, G., in Progress in the
Understanding and Prevention of Corrosion, Vol. 2, Proceedings of the lOth European Corrosion
Congress, Institute of Materials, London, 1993, p. 1304.
[111] Dawson, J. L., Gearey, D., and Cox, W. M., "Recent Experience of Monitoring Condensed Acid
Corrosion in Boiler Flue Gas Ducts," U.K. CORROSION~82, I.Corr.S.T., London, 1982, p. 99.
[112] Dawson, J. L., in Proceedings of the International Conference on Corrosion of Reinforcement in
Concrete Construction, Ellis Hoiwood, London, 1983, p. 175.
[113] Cox, W. M., Gearey, D., and Dawson, J. L., "Corrosi on--l ndust ri al Problems Treatment and
Control Techniques," in Proceedings of the Kuwait Conference, Pergamon Press, Oxford, 1987,
p. 83.
[114] Hladky, K., John, D. G., Worthington, S. E., and Herbert, D., in UK CORROSION~85, l.Corr.S.T.,
London, 1985, p. 295.
[115] Hardon, R. G., Lambert, P., and Page, C. L., British Corrosion Journal, Vol. 23, 1988, p. 225.
[116] Glass, G. K., Page, C. L., and Short, N. R., UK CORROSION~88, Vol. II, I.Corr.S.T., London,
1988, p. 243.
[117] Carassiti, F., Cigna, R., Gusmano, G. K., and Goolamallee, R., Material Science Forum, Vols.
44-45, 1989, p. 271.
[118] AI-Zanki, I. A., "The Application of Electrochemical Noise to the Monitoring of Corrosion,"
Ph.D. thesis, University of Manchester, 1989.
[I19] Bierwagen, G. P., submitted to Journal of the Electrochemical Society, 1994.
[120] Cottis, R. A. and Turgoose, S., in this volume.
[121] Dawson, J. L., Cox, W. M., Eden, D. A., Hladky, K., and John, D. G., Giornale delle Prove non
Destructive (Italy), Vol. 2, 1986, p. 49.
[122] Eden, D. A., Hoffman, M., and Skerry, B. S., Polymer Material Science Engineering, Vol. 53,
1985, p. 388.
[123] Lomas, J. P., Callow, L. M., Sussex, G. A. M., and Scantleberry, J. D., Materials Science Forum,
Vol. 8, 1986, p. 317.
[124] Skerry, B. S., Progress in Organic Coatings, Vol. 15, 1987, p. 269.
[125] Skerry, B. S., Alavi, A., and Lingren, K. I., Journal of Coatings Technology, Vol. 60, 1988, p. 97.
[126] Bastidas, J. M., Feliu, S., Jr., Morcillo, M., and Feliu, S., Progress in Organic Coatings, Vol. 18,
1990, p. 265.
[127] Bierwagen, G. P., Mills, D. J., and Tallman, D. E., in Proceedings of the 12th International
Corrosion Congress, Paper 576, National Association of Corrosion Engineers, Houston, 1993,
p. 4208.
[128] Mills, D. J., Bierwagen, G. P., Tallman, D. E., and Skerry, B. S., Proceedings of the 12th Inter-
national Corrosion Congress, Paper 486, National Association of Corrosion Engineers, Houston,
Vol. l, 1993, p. 182.
[129] Carr, R., Eden, D. E., and Dawson, J. L., patent application, 1993.
[130] Cavalcanti, E. H. de S., "' Electron Optical and Electrochemical Observations of Stainless Steels,"
Ph.D. thesis, University of Manchester, 1984.
[131] Sato, N., Journal of the Electrochemical Society, Vol. 123, 1976, p. 1197.
[132] Okada, T., Journal of the Electrochemical Society, Vol. 105, 1993, p. 1261.
[133] Ferreira, M. G. S. and Simoes, A. M. E, Materials Science Forum, Vol. 8, 1986, p. 373.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
34 ELECTROCHEMICAL NOISE MEASUREMENT
[134] Dawson, J. L., Rothwell, A. N., Walsh, T. G., Lawson, K., and Palmer, J. W., in CORROSION/
93, National Association of Corrosion Engineers, Houston, 1993.
[135] Li, W., Nobe, K., and Pearlstein, J., Corrosion Science, Vol. 31, 1990, p. 615.
[136] Loto, C. A. and Cottis, R. A., Bulletin of the Electrochemical Society, Vol. 4, 1988, p. 1001.
[137] Bastidas, J. M. and Malo, J. M., Revista Del Metal (Madrid), Vol. 21, No. 6, 1985, p. 337.
[138] Roberge, P. R., Sastri, V. S., and Revie, W., in Proceedings of the 7th European Symposium on
Corrosion lnhibitors, Vol. 1, University of Ferrara, Italy, 1990, p. 1003.
[139] Monticelli, C., Brunoro, G., Frignani, A., and Trabanelli, G., Journal of the Electrochemical So-
ciety, Vol. 139, 1992, p. 706.
[140] Hirozawa, S. T. and Tnrcotte, D. E., Electrochemical Impedance: Analysis and Interpretation,
ASTM STP 1188, American Society for Testing and Materials, West Conshohocken, PA, 1994,
p. 205.
[141] Xiao, H. and Mansfeld, F., Journal of the Electrochemical Society, Vol. 141, 1994, p. 2332.
[142] Farrell, D. M., Cox, W. M., Wrobel, B. A., and Syrett, B. C., in Proceedings of the Symposium
on Corrosion in Power Utilities, National Association of Corrosion Engineers, Houston, 1987.
[143] Farrell, D. M., Cox, W. M., Syrett, B. C., and Keeth, R. J., in Proceedings of the 1st Combined
Flue Gas Desutfurization and Dry SO z Control Symposium, EPRI GS-6307, April 1989.
[144] Roarty, D. H., Bogard, W. T., Cox, W. M., Moore, D. C. A., and Quirk, G. P., Paper 192, in
CORROSION/93, National Association of Corrosion Engineers, Houston, 1993.
[145] Winters, M. A., Stokes, P. S. N., Zuniga, P. O., and Schlottenmeier, D. J., Paper 392, in
CORROSION/93, National Association of Corrosion Engineers, Houston, 1993.
[146] Hew, Y. O., "Assessment of Paint Coatings," M.Sc. dissertation, University of Manchester, 1988.
[147] Dawson, J. L., John, D. G., Jafar, M. I., Hladky, K., and Sherwood, L., in Corrosion of Reinforce-
ment in Concrete, Elsevier Applied Science, London, 1990, p. 358.
[148] Brennenstuhl, A. M. and Gendron, T. S., in Microbiologically Influenced Corrosion Testing, ASTM
STP 1232, American Society for Testing and Materials, West Conshohocken, PA, 1994.
Bibliography
Bataillon, C. and Fiaud, C., Materials Science Forum, Vol. 8, 1986, p. 141.
Bertocci, U., Journal of the Electrochemical Society, Vol. 133, 1986, p. 1782.
Bertocci, U., in Proceedings of the Advances in Localized Corrosion, National Association of Corrosion
Engineers, Houston, 1990, p. 127.
Budevski, E., Obretenov, W., 13ostanov, W., Staikov, G., Doneit, J., et aI., Etectrochimica Acta, Vol. 34,
1989, p. 1023.
Cao, C-N., Shi, Q., and Lin, H., Bulletin of Electrochemical Society, Vol. 6, 1990, p. 710.
Doherty, P. E., Moore, D. C. A., Cox, W. M., and Dawson, J. L., in Proceedings of the 4th International
Symposium on Environmental Degradation of Materials in Nuclear Power Stations, 1989.
Eden, D. A. and Rothwell, A. N., in CORROSION/92, National Association of Corrosion Engineers,
Houston, 1992.
Edgemon, G. L., Wilson, D. E, and Bell, G. E. C., Paper 179, in CORROSION/93, National Association
of Corrosion Engineers, Houston, 1993.
Ferreira, M. G. S. and Simoes, A. M. P., Rev. Corros. Prot. Mat., Vol. 3, No. 2, 1984, p. 40.
Flis, J., Dawson, J. L., Gill, J., and Wood, G. C., Corrosion Science, Vol. 32, 1991, p. 877.
Heusler, K. E., Corrosion Science, Vol. 29, 1989, p. 131.
Heusler, K. E., Corrosion Science, Vol. 31, 1990, p. 597.
Isaacs, H., Corrosion Science, Vol. 29, 1989, p. 313.
Magaino, S., Corrosion Engineering (Boshoku Gizatsu), Vol. 37, 1988, p. 692.
Magaino, S., Yamazaki, R., and Kobayaski, N., Journal of Electroanalytical Chemistry and lnterfacial
Electrochemistry, Vol. 305, 1991, p. 141.
Mansfeld, F. and Ziao, H., Journal of the Electrochemical Society, Vol. 140, 1993, p. 2205.
Metikos-Hukovic, M., Loncor, M., and Zevnik, C., Werkstoffe Und Korrosion, Vol. 40, 1989, p. 494.
Miya, K. and Tachibana, K., Corrosion Engineering (Boshoku Gizatsu), Vol. 37, 1988, p. 97.
Okada, T., Journal of Electroanalytical Chemistry and Interfacial Chemistry, Vol. 297, 1991, p. 349.
Okada, T., Journal of Electroanalytical Chemistry and Interfacial Chemistry, Vol. 297, 1991, p. 361.
Okada, T., Journal of the Electrochemical Society, Vol. 140, 1993, p. 1261.
Okada, T., in Proceedings of the Symposium on Critical Factors in Localized Corrosion, Electrochemical
Society, 1992.
Pistorius, P. C., in this volume.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
KEYNOTE ADDRESS 35
Riley, A. M., et al., Corrosion Science, Vol. 32, t991, p. 1307.
Roarty, D. H., Lawson, K., Ryder, J. C., Eden, D. A., and Quirk, G. P., Paper 191, in CORROSION/93,
National Association of Corrosion Engineers, Houston, 1993.
Roberge, P. R., Beaudoin, R., and Sastri, V. S., Corrosion Science, Vol. 29, 1989, p. 1231.
Rothwell, A. N. and Eden, D. A., in CORROSION/92, National Association of Corrosion Engineers,
Houston, 1992.
Shibata, T., Corrosion Science, Vol. 31, 1990, p. 413.
Shibata, T., Horikawa, J., and Fujimoto, S., Corrosion Engineering, Vol. 38, 1989, p. 155.
Skerry, B. S. and Eden, D. A., Progress in Organic Coatings, Vol. 19, 1991, p. 379.
Smith, S. and Francis, R., British Corrosion Journal, Vol. 25, 1990, p. 285.
Smith, S. and Francis, R., in COPPER/90, Institute of Metals, London, 1990, p. 81.
Stewart, J. and Williams, D. E., Corrosion Science, Vol. 33, 1992, p. 457.
Uruchurtu, C., Corrosion, Vol. 47, 1991, p. 471.
Uruchurta, J'. C. and Dawson, J. L., Corrosion, Vol. 43, 1987, p. 19.
Uruchurtu, J. C., "Corrosion of Aluminium in Seawater," Ph.D. thesis, University of Manchester, 1985.
Uruchurta, J. C., Proceedings of the Advances in Localized Corrosion, National Association of Corrosion
Engineers, Houston, 1990, p. 141.
Wolsiffer, S., Moore, D. C. A., and Radcliffe, P., in Proceedings of the 2nd Combined Flue Gas and
Dry SO 2 Control Symposium, EPA/ EPRI, 1990.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
Correlations to Other Electrochemical
Techniques
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
Ugo Be r t oc c i 1
A Comparison of Electrochemical Noise
and Impedance Spectroscopy for the
Detection of Corrosion in Reinforced
Concrete
REFERENCE: Bertocci, U., "A Comparison of Electrochemical Noise and Impedance
Spectroscopy for the Detection of Corrosion in Reinforced Concrete," Electrochemical
Noise Measurement for Corrosion Applications, ASTM STP 1277, Jeffery R. Kearns, John R.
Scully, Pierre R. Roberge, David L. Reichert, and John L. Dawson, Eds., American Society
for Testing and Materials, 1996, pp. 39-58.
ABSTRACT" Electrochemical impedance spectroscopy (EIS) and electrochemical noise mea-
surements (ENM) were carried out on small concrete blocks with embedded steel rods, im-
mersed in solutions of highway deicers, mainly sodium or magnesium chlorides with corrosion
inhibitors added. Voltage and current fluctuations between identical electrodes were recorded,
and different signal-processing schemes were tested. Various problems experienced in the
course of the work are described. Corrosion polarization resistances obtained by EIS are com-
pared with values of the ratio of the standard deviations of the voltage and current fluctuations.
The results suggest that the type of ENM used in this work is subject to considerable uncer-
tainties, and further work would be necessary before the method can be applied with sufficient
confidence to corrosion monitoring.
KEYWORDS: corrosion, reinforced concrete, electrochemical noise, impedance spectroscopy
This paper presents part of the research carried out at the Turner-Fairbank Research Center
of the Federal Highway Administration on highway deicers. The main purpose of the work
was to examine if the addition of corrosion inhibitors to commercial deicing mixtures, which
contain mainly chloride salts, affects the corrosion behavior of steel in concrete. The solu-
tions used in the testing are listed in Table 1, together with the numerical designation used
throughout this paper.
The principal measurement method used was electrochemical impedance spectroscopy
(EIS), but, at the same time, it was decided to carry out electrochemical noise measurements
(ENM) on the same set of test specimens to learn about the utility of ENM in monitoring
corrosion in steel-reinforced concrete.
Problems encountered in implementing the specific ENM technique selected for this work
are described. The ENM results obtained on electrochemical ceils, which were also being
studied by EIS, are presented and discussed, to compare their electrochemical behavior, as
shown by the EIS measurements, with that obtained from EN data.
Senior research associate, Turner-Fairbank Highway Research Center, Federal Highway Administra-
tion, 6300 Georgetown Pike, McLean, VA 22101.
Copyrighl 9 1996 by ASTM lntcrnational
39
www.astm.org
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
.
I
X

0

T
A
B
L
E

1
-
-
E
N
M

a
n
d

E
I
S

r
e
s
u
l
t
s
.

m

r
"

m

o

-
1
1

O

(
3

"
1
"

m

a
r

)
,

S
o
l
u
t
i
o
n

#

S
o
l
u
t
i
o
n

E
N
M

D
a
t
a

P
r
o
c
e
s
s
i
n
g

M
e
t
h
o
d

P
o
l
y
n
o
m
i
a
l

F
i
l
t
e
r

P
o
l
y
n
o
m
i
a
l

F
i
l
t
e
r

E
I
S

V

C
h
a
n
n
e
l

I

C
h
a
n
n
e
l

V

C
h
a
n
n
e
l

I

C
h
a
n
n
e
l

X
,

X
,

R
o

R
c
,

p
,
V

n
A

t
x
V

n
A

k
D

x

c
m
2

k
D

x

c
m
z

k
D


c
m
:

k
D

x

c
m
2

r
-

z

C
)

o
)

m

c

0
.
2
M

N
a
C
I

5
5
8

1
0
4

4
7
0

1
4
7

4
8
.
4

1
2
.
3

8
8
.
5

1
4
.
6

0
.
2
M

N
a
C
1

+

1
7
4

1
2
.
5

1
5
7

1
0
.
4

p
h
o
s
p
h
a
t
e

i
n
h
i
b
i
t
o
r

0
.
1
M

M
g
C
I
E

+

3
3
7

1
2
.
7

2
7
8

1
1
.
6

c
i
t
r
a
t
e

i
n
h
i
b
i
t
o
r

0
.
1
M

M
g
C
l
z

+

6
0
.
3

1
.
9
5

6
0
.
6

2
.
0
3

l
i
g
n
o
s
u
l
f
a
t
e

i
n
h
i
b
i
t
o
r

1
0
0
-
g
/
L

C
a

a
n
d

M
g

2
3
.
5

1
.
9
0

1
4
3

1
.
9
7

a
c
e
t
a
t
e

8
3
.
4

4
9
.
1

6
1
.
2

9
4
.
4

2
1
9

2
4
0

4
1
1

3
7
0

4
7
8

4
6
2

1
9
2

1
1
2
0

1
7
0

4
0
0

2
0
0

2

2
0
0

1
1

5
0
0

4
0

0
0
0

4
9
.
6

4
1

1
2
2

6
2
0

~
0

m

m

z

-
-
4

Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
BERTOCCI ON DETECTION OF CORROSION IN REINFORCED CONCRETE 41
ENM Method
Various met hods have been used for ENM studies requi ri ng mor e or l ess compl i cat ed
i nst rument at i on. To si mpl i f y the ci rcui t ry and avoi d t he necessi t y for t ransformat i ons in the
frequency domai n, a number of workers have car ded out ENM bet ween t wo i dent i cal el ec-
t rodes at open ci rcui t [ 1 - 3 ] . Thi s met hod, in its most el abor at e form, uses t wo pai rs of
el ect rodes, al l made of t he mat eri al under st udy and of equal surface area. Two of t he
el ect rodes, separat ed by a short di st ance in the el ect rol yt e, are connect ed by a zero-resi st ance
ammet er ( ZRA) , whi ch records t he fluctuating current s flowing bet ween the two. The ot her
t wo el ect rodes are connect ed t hrough a hi gh i mpedance vol t met er, whi ch records t he vol t age
fluctuations bet ween them. The proponent s of this arrangement [ 4 , 5 ] poi nt out that it is
si mpl er than t hose usi ng a pot ent i ost at , and t hat t he measurement s are per f or med at t he
corrosi on pot ent i al , whi ch makes t hem mor e rel evant to the act ual servi ce condi t i ons.
The anal ysi s of t he dat a is carri ed out by cal cul at i ng the st andard devi at i on of t he current
and vol t age fluctuations over a chosen t i me i nt erval and then consi der i ng t he rat i o of t he
t wo as a measure of the corrosi on resi st ance of t he mat eri al [ 4 , 5 ] .
The si gni fi cance of this t ype of measur ement has been di scussed in a previ ous paper in
cooper at i on wi t h Huet [6]. In it, a model is pr oposed to account for t he vol t age fluctuations
in an i sol at ed el ect rode, whi ch assumes that t he surface can be di vi ded in a l arge number
of pat ches. On each of them, smal l changes at the i nt erface l ead to short -l i ved changes in
the i nt erfaci al resi st ance. These, in turn, cause fluctuations in the anodi c and cat hodi c part i al
currents. Inst ant aneous devi at i ons f r om a zero net current in each patch, summed over the
whol e ar ea A, wi l l dri ve the el ect r ode pot ent i al , posi t i vel y or negatively, so as to reach
compensat i on bet ween anodi c and cat hodi c current. It is al so assumed that t he resi st ance
changes of the vari ous surface pat ches are st at i st i cal l y uncorrel at ed; therefore, the size of
the pat ch is det er mi ned by t he spat i al ext ent of the correl at i on.
Al t hough the descr i pt i on of the model is easi er starting from t he current fluctuations as
the " caus e" of t he vol t age fluctuations, bot h are the consequence of the physi cal changes
on the surface, and it is i dl e to argue over which, current or vol t age, causes the fluctuations
of the other.
The i nst ant aneous devi at i on of the current f r om zero is the sum of the l ocal fluctuations
over al l M pat ches, each fluctuation bei ng the pr oduct of a scal i ng factor, i o , wi t h t he di -
mensi ons of a current density, and a r andom number, Pk, wi t h an average val ue of zero and
st andard devi at i on, or, so t hat the t ot al current, I ( t ) , is
k - M
z ( t ) = io ~ p~ (1)
k- I
Assumi ng that the fluctuations are nor mal l y di st ri but ed, the vari ance of the current I ( t ) is
then proport i onal to the number of pat ches, that is, t o t he surface area of the el ect rode, and
the st andard devi at i on of the current fluctuations, s~, wi l l be pr opor t i onal to t he square r oot
of the area A
sz = KV~ (2)
where K is a proport i onal i t y constant. Because the fluctuations are smal l , one can assume
that t here is a l i near rel at i onshi p bet ween t he faradai c part of t he current densi t y of the
fluctuations and t he el ect r ode pot ent i al . The resul t i ng rel at i onshi p for the vol t age is
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
42 ELECTROCHEMICAL NOISE MEASUREMENT
V(t) = R i(t) - C dt ] = R k:,~ Ok-- C- - ~
(3)
where i is the corrosion (or exchange) current density and R and C are the unit values of
the electrode resistance and double layer capacitance, respectively. From this formula it can
be seen that high-frequency current fluctuations are shunted by the capacitance and cause a
smaller effect on the voltage than low-frequency fluctuations of the same amplitude. Whereas
the standard deviation of the total current defined above varies with surface area as shown
in Eq 2, that of the current density is inversely proportional to the square root of the area
and, because of Eq 3, so will the standard deviation of the voltage fluctuations, sv
(4)
FIG. 1--Shape and size of the cast concrete specimens.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
BERTOCCI ON DETECTION OF CORROSION IN REINFORCED CONCRETE 43
where B is a proportionality constant. Therefore, for the measurement of the voltage fluc-
tuations, it is better to use as small a surface area as possible t o increase the amplitude of
the signal.
The current fluctuations are driven by the voltage fluctuations that occur independent of
each other on the t wo electrodes. I f the current flowing between the two electrodes is neg-
ligible compared to the partial currents within each one, the electrical connection will not
affect significantly the random current fluctuations within each electrode. With the same
reasoning as that used t o describe the influence of the surface area on the voltage fluctuations,
it can be shown that the standard deviation of the current density i e x t is inversely proportional
to the square root of the area. Therefore, the standard deviation of the total external current
s~r t, whi ch is the quantity being measured, increases as the square root of the area. As a
consequence, although small electrodes may be more suited for voltage measurements, the
current measurements favor the use of large electrodes.
The current density generated by the voltage fluctuations of the t wo electrodes, i e x t , de-
pends on the admittance of the circuit, whi ch is formed by a faradaic impedance in parallel
to a double layer capacitance C, followed by the resistance of the solution, R,, between the
electrodes. Here this circuit is simplified by assuming that the faradaic branch is a simple
resistance, R. From Eq 3 we have
v R c ( d r R
i"x' = R + R------~ + R + R'-----~, - ~ - " d t ] ( 5 )
It is easy t o see that, for R s = 0, substitution of Eq 3 into Eq 5 gives Eq 1, the external
current is identical to the fluctuations within an electrode. Because the current to be measured
decreases as R, increases, it is convenient to keep the solution resistance small. This justifies
the practice of using t wo electrodes facing each other at a short distance.
Accordi ng t o this model, because of the existence of a capacitative branch in the electrode,
the amplitude of the current iext should be larger for high-frequency fluctuations. It would,
therefore, be more reasonable to presume that the ratio, X, of the standard deviations woul d
be proportional to the modul e of the impedance at some frequency larger than zero, so that
X woul d not necessarily be a good measure of the corrosion rate, that is, to the zero frequency
limit of the electrode impedance.
The model describes above presupposes uniformity of conditions on all the electrodes, a
thing that may not be easily realized i f the mai n form of corrosi on is, for instance, pitting.
Therefore, experimental testing is of great interest to build confidence in the usefulness of
this ENM technique.
Experimental Procedures
The measurements were car ded out on steel rods embedded in concrete, exposed to various
solutions by immersion, over about seven months. The shape and size of the specimens are
shown in Fig. 1. To promot e uniform corrosion conditions for the four carbon steel rods in
each block, they were kept short-circuited unless measurements were being performed.
To maintain an oxygen-ri ch environment inside the concrete, the specimens were subjected
to a regular regime of i mmersi on in the solution from Monday t o Friday and let to dry over
the weekend. This regime was followed, with a few exceptions, for several months.
Immersi on of the specimens was done in 600- mL glass beakers, in which about 150 mL
of solution was poured. The level of the liquid reached about 1 cm below the top of the
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
44 ELECTROCHEMICAL NOISE MEASUREMENT
COMPUTER
FIG. 2--Schematic view of the EN measurement system.
concret e bl ock, and car e was t aken to prevent the sol ut i on from wet t i ng bet ween the pr o-
t rudi ng rods.
Measurement Scheme for EN
Data Acqu&ition
The el ect ri c ci rcui t for ENM used t hree of the f our r ods in each concret e bl ock: one of
t hem was common to the current and vol t age measur ement circuits. The current bet ween it
and anot her rod was measur ed by means of a ZRA, whereas t he vol t age fluctuations bet ween
the common rod and t he t hi rd one were r ecor ded by a hi gh-i nput i mpedance voltmeter. The
use of t hree el ect rodes rat her than four does not al t er t he pri nci pl e of t he measur ement
scheme [7], but si mpl i fi es the groundi ng of the circuitry. The onl y observabl e di fference is
that, wi t h t hree el ect rodes, some correl at i on bet ween current and vol t age fluctuations has to
be expect ed.
No hi gh-pass filters were i nst al l ed to el i mi nat e DC and very l ow-frequency si gnal s, what
we may cal l " DC" drift, but it was deci ded to try to mi ni mi ze t hei r effect by di gi t al proc-
essing. Di fferent schemes were tried, whi ch wi l l be di scussed later.
A t wel ve-bi t , mul t i channel , i nput / out put ( I / O) card i nst al l ed in a personal comput er al-
l owed t he si mul t aneous r ecor di ng of current and vol t age from four concret e bl ocks. On each
of t he ei ght si gnal channel s, an ant i al i asi ng t ow-pass t hree-pol e filter was i nst al l ed, wi t h a
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
BE RT OCCI ON DE T E CT I ON OF C O R R O S I O N I N R E I N F O R C E D C O N C R E T E 45
~ >
E
J ~
" O
d
a .
o ~
<
J ~
" O
c~
a .
u ~
~ >
E
J O
0 . .
U ~
- 6 0 - ' , , , , , j
- 7 0
eo 4 so B l oo 12o
Fr e que nc y , Hz
- 4 0 , , , , ,
- 5 0
-60
- 7 0 ' ~ 6 1 0 ~ , I
2 0 4 8 1 0 0 1 2 0
Frequency, Hz
-40 , , , , ,
-50 ~
-60
-70
- 8 0 ' ~ , ~ I ,
0 2 0 4 6 0 8 1 0 0 1 2 0
Fr e que nc y , Hz
-40
: >
E -5o
"~ d
~ -60
-70
0 2 4 6 8 1 0 1 2 1 4 1 6
Frequency, Hz
FIG. 3- - Spectr a of the i nstrumental noi se: (a) Voltage channel O. Ampl i fi cati on -~500. Acqui si ti on
rate = 256 pts/s. (b) Current channel 1. Ampl i fi cati on = 100. ZI A ampl i fi cati on = 105 V/A. Acqui si ti on
rate = 256 pts/s. (c) Voltage channel 6. Ampl i fi cati on = 500. Acqui si ti on rate = 256 pts/s. (d) Voltage
channel 6. Ampl i fi cati on = 500. Acqui si ti on rate = 32 pts/s.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
46 E L E C T R O C H E MI C A L N OI S E M E A S U R E M E N T
cut of f frequency o f 15 Hz. A schemati c circuit is gi ven in Fig. 2. For the acqui si ti on and
processi ng o f the signals, commerci al software produci ng virtual instruments was used.
Efforts were made to decrease the instrumental noi s e o f the s ys t em by electrical shi el di ng
o f the cel l s and the instruments and usi ng battery-powered component s. The instrumental
noi se was much affected by the amplification, whi ch coul d be applied t o the input signals.
For the vol t ages, the I / O card provi ded four levels: 1, 10, 100, and 500. Each l evel corre-
sponded to different maxi mum signals, respecti vel y, 10 V, 1 V, 100 mV, and 20 mV. For the
current si gnal s, besi de the same amplification choi ces in the I / O card, the ZRAs coul d be
set for conversi on factors f rom 104 to 107 VI A. Because the instrumental noi se was l owes t
when the amplification o f the I / O card was the largest, a major factor limiting the sensi ti vi ty
was the presence o f a DC si gnal that compel l ed a reduction in amplification t o stay wi thi n
79
O
C
P
O
Z
UJ
78
77
76
75
74
73
72
71
- 18
5 0 0 1 0 0 0 15 0 2 0 0 0
Ti me, s
- 19
- 20
-21
- 22
>
E
=;
O
>
Z
UJ
' ' ' 0 5 0 0 1 0 0 0 15 0 0 0 0
Ti me, s
F I G. 4- - - E N recordi ngs bef ore pr ocessi ng (Sol uti on 3 ) : ( a ) Vol tage. Acqui si ti on rate = 0.5 pts/ s. (b)
Current. Acqui si ti on rate = 0.5 pts/s.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
BERTOCCI ON DETECTION OF CORROSION IN REINFORCED CONCRETE 47
the allowed input limits. It was also found that the instrumental noise varied from channel
to channel. Cross-talk between adjacent channels was found to be significant. This problem
was solved by using onl y every other channel of a 16-channel I / O card.
For the voltage-measuring channels, the standard deviation of the instrumental noise (mea-
sured with inputs short-circuited) was between 8 and 16 ixV at the maxi mum amplification
of 500, but it was about four times larger if the amplification had to be reduced to 100. For
the current-measuring channels, the instrumental noise (measured with inputs open-circuited)
ranged from 4 to 0.04 nA, but the settings most often used had an instrumental noise from
0.1 to 0.4 nA. Despite the antialiasing filters, some structure was observed in the spectra of
the instrumental noise. Figure 3 shows the examples of instrumental noise spectra. Al t hough
the spectrum of Channel 0 (Fig. 3a) did not show any peaks, that of the noisiest channel
(Channel 6, Fig. 3c), had several peaks, some ten times larger than the background and up
to - 4 5 db mV, which is equal to 6 ~V. When measuring at low acquisition rates (31 ms/pt,
in the example of Fig. 3d), these peaks were aliased and appeared as l ow-frequency peaks,
as shown in Fig. 3d.
Data Processing
Al t hough the acquisition of the time series could be observed in real time, the data proc-
essing for the calculation of the standard deviations and their ratios was done subsequently.
For the purpose of testing different processing schemes, it is clearly advantageous t o apply
t hem to the same data sets. However, i f EN measurements have t o be used for monitoring
corrosion, processing in real time, with immediate display of the results, would be necessary.
Methods for doing so are being developed, and onl y preliminary results are available.
The time series of N voltage and N current readings were taken mostly at a rate of a point
every 2 s (v = 0.5 Hz), generally with N = 1000 or 1024. The acquired data immediately
indicated that the assumption of a zero average difference in voltage and current between
the supposedly identical electrodes was not fulfilled. Substantial voltage differences between
rods embedded in the same concrete bl ock had already been observed in open circuit poten-
tial (Eo~) measurements. As soon as the short circuit was broken, in many instances the Eo~
would drift apart rapidly. Figure 4 shows voltage and current recordings before any proc-
essing: the drift in voltage is evident, but the current data (Fig. 4b) also show a large DC
component. Standard deviations calculated from the raw data would be quite large because
of the low-frequency components. Therefore, it was necessary to eliminate both the DC
component and slow drift. For this, the value of the first point (n = 0), was subtracted from
all following values, and then the time series was passed t hrough a fifth order, high-pass,
Butterworth digital filter, with a cut off frequency vc = v/N, or approximately 0.5 mHz. The
result of such a treatment is illustrated in Fig. 5, which shows the same curves of Fig. 4
after filtering. Here it is possible to see that some of the spikes that appear in the current
record coincide in time with transients in the voltage, which were almost impossible to detect
in the original recording.
Anot her scheme for the elimination of DC and slow drift was tried. This consisted of
fitting the raw data to a second- or third-order pol ynomi al and then using the residuals as a
measure of the noise. The method gave results similar to high-pass filtering, except when
the fluctuations appeared to have extremely low frequencies, close t o the cut off limit of the
digital filter.
Because sporadic transient spikes generated by other electrical equipment in the laboratory
could not be completely avoided, and because they considerably affected the standard de-
viations, the high-pass-filtered time series were additionally processed, when necessary,
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
48 E L E C T R O C H E MI C A L NOI S E M E A S U R E M E N T
E
. ~
0
= >
Z
u J
c
e .
P
i .
0
Z
UJ
F I G. 5- - Same
0 , 6
0 , 4
0 , 2
0 , 0
- 0 , 2
- 0 , 4
- 0 , 6
0
I
5 0 0 1 o l o o 1 5 1 0 o 2 0 0 0
Ti me, s
-1
- 3
o
, i
I
5 0 0 1 0 1 0 0 1 5 1 0 0 2 0 0 0
Ti me, s
recordi ng as i n F i g . 4 a f t e r f i l t e r i n g : ( a ) vol tage and (b) current.
through a digital filter to remove the unwanted spikes. The occurrence of these spikes coul d
be detected easi l y because they were much larger than normal and appeared at the same
ti me on all measuring channels.
Once the ti me series were adequately filtered, running standard devi ati ons (Sv). and (st).
values coul d be computed. Thi s was done usi ng either all points from 0 to n for the value
of index n or a shorter record length L (L < N) for n > L. In this work, the first alternative
was used. From these standard deviations, their ratios X. = ( sv) ./ ( st) . were obtained, a typical
exampl e of whi ch is shown in Fig. 6. In general, the running ratio tended to stabilize after
a few hundred points. It is therefore reasonable to try to summari ze the curves by a si ngl e
number. To do so, running means and standard deviations of (Sv)., (st)., and X . were com-
puted for record lengths L = N / 2, one hal f of the original time record, N. The last value of
the mean can be considered as an indication of the average val ue of these quantities, whereas
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
B E R T OC C I O N DE T E CT I ON OF C O R R O S I O N I N R E I N F O R C E D C O N C R E T E 49
8 0
7 0
6 0
E 50
t -
O
4 0
, , =
3 0
2O
10
0
0
I I I
5 0 0 1 0 0 0 1 5 0 0
T i me , s
2 0 0 0
FI G. 6--Typical standard deviation ratio X. . Acquisition rate = 0 . 5 p t s / s .
the standard devi at i on i s an i ndi cat i on o f thei r st abi l i t y wi t h t i me. The val ues report ed i n
Tabl e 1 we r e c omput e d thi s way.
EI S Re s u l t s
To c ompar e the ENM data wi t h the e l e c t r oc he mi c al behavi or as s hown by EI S, the resul ts
o f the i mpe danc e me as ur e me nt s wi l l be bri efl y s ummari zed. Al l me as ur e me nt s we r e carri ed
7 , o
_ ' I ' - - - * - L o g q z l l
6 - - 2 0
0
E ~
t -
O 5 - -4o
N
n r . .
O1 0 .
O
- - I 4 - - 6 0
3 t - 8 0
- 4 - 3 - 2 - 1 0 1 2
L o g ( F r e q ) , Hz
FI G. 7--Typical Bode plot showing high Ro and only one time constant (Solution 5).
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
50 ELECTROCHEMICAL NOISE MEASUREMENT
5,0 0
i i i i i i
9 -10
4, 5
% 2~
t~ "o
x
E - - 30 &
0 4,0
c
~,, - -4o
t~
O - -50 ' -
--I 3,5 O.
I
,--.o--- Log IZl -60
3,0 -70
-4 -3 -2 -1 0 1 2
L o g ( F r e q ) , Hz
FIG. 8--Typical Bode plot showing lower Ro and two time constants (Solution 1).
out at the Eoc. The ci rcui t s that best si mul at ed t he exper i ment al dat a were ei t her a si mpl e
RC paral l el fol l owed by a resi st or or t he one wi t h an addi t i onal const ant phase el ement
(CPE) of Sl ope 1 (45 ~ in seri es to the resistor. The CPE pr obabl y is t he effect of di ffusi onal
t ransport in the concret e, as a Warburg i mpedance over a finite-length di ffusi on layer. The
fitting rout i ne was abl e to pr ovi de an ext rapol at i on to zero frequency, gi vi ng a val ue for the
-100 m
[] a m ~ B , ~ m - - - m n
- o ~ q~.~_ _ ~ =
t,l,,I -200 m ~ ~ 1 ~
O
(t'}
-300
S -400 [] m B" " ~ - - I~
% l ,
-500 m m
y = - 2 9 5 . 4 2 - I 0 3 . 0 1 X R ^ 2 = 0 . 8 2 9
-600 I 11 I I
3 -2 - O 1
l o g ( l / R) , l l S/ cm 2
FI G. 9--Cumulative plot of measurement potential E m v e r s us electrode admittance.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
BERTOCCI ON DETECTION OF CORROSION IN REINFORCED CONCRETE 51
E
O~
O
Z
LU
5 0 0 1 0 0 1 5 0 0 2 0 0 0
Ti me, s
i i i i
1
0
1-
r
= b
P
i . - 1
( J
Z
I L l
- 2
- 3 I
5 0 0 1 0 l o 0 1 5 1 0 0 2 0 0 0
Ti me, s
FIG. l O- - Fi l t er ed EN recordings. Solution 1: (a) Voltage. Acqui si t i on rate = 0. 5 pt s / s . (b) Current.
Acqui si t i on rate = 0. 5 pt s / s .
overall electrode resistance. This value, however, is, in general, not very accurate because
the measurements had to be extended to very low frequencies where the scattering of the
experimental points was large.
The simpler RC circuit was associated with very large values of the electrode resistance,
R o, 10 MI I cm 2 or more, and with double-layer capacitance values, Cdt, between 30 and
60 p~F/cm 2, indicating very low corrosion rates. A typical Bode plot is shown in Fig. 7,
obtained in Solution 5. In the range of frequencies sampled by the ENM, the slope of the
absolute impedance I~ with respect to the logarithm of the frequency, v, is about - 1 , and
its value is of the order of 1 MI-I x cm 2 at 1 mHz. The equivalent circuit with t wo time
constants fits better the experimental results when the steel is corrodi ng more rapidly. The
resistance associated with the first time constant may vary from a few kl'~ cm 2 t o almost
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
- 2 0
I I I I I
- 4 0
" 0
{E
I ' l - 6 0
( / )
- 8 0
52 E L E C T R O C H E M I C A L N O I S E M E A S U R E M E N T
-I00 I I I I I
- 3 , 5 - 3 , 0 - 2 , 5 - 2 . 0 - 1 , 5 - 1 , 0 - 0 , 5
L o g ( F r e q ) , H z
4O
20
,<
C
J ~
"o o b
O.
i f )
- 2 0
-40 I I I I I
- 3 , 5 - 3 , 0 - 2 , 5 - 2 , 0 - 1 , 5 - 1 , 0 - 0 , 5
L o g ( F r e q ) , H z
FIG. 11- - Power spectra of the EN recordings shown in Fig. 10: (a) vol tage and (b) current.
1 mI~ x c m 2, whereas Cd~ i s larger than in the previ ous case, but in general l e s s than
100 ~ F / c m 2. The Bo de pl ot ma y be as in Fi g. 8 ( Sol ut i on l ) , wi t h a s l o pe o f about - 0 . 5 at
1 mHz , whi c h changes t o about - 1 be t we e n l 0 and 100 mHz , dependi ng on the val ue o f
the resi st ance. The val ues o f I ~ in the f requency range s ampl e d by EN can be qui te di fferent,
f r om 1 Ml ) x c m 2 t o l e s s than 1 kI~ x c m 2. The res i s t ance in s eri es Rs o f the e l e c t r ol yt e
permeat i ng the concret e wa s o f the order o f 5 0 t o 100 12 The resul ts o f the EIS me as ur e me nt s
s ho we d a g o o d correl at i on be t we e n R o and the pot ent i al at whi c h the me as ur e me nt was
made, as s hown in Fi g. 9. Corros i on rates coul d be cal cul at ed f r om EIS data s uppl e me nt e d
by pol ari zat i on me as ur e me nt s t o e s t i mat e the coef f i ci ent s as s oci at ed wi t h the Tafel s l ope s .
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
BERTOCCI ON DETECTI ON OF CORROSI ON IN REI NFORCED CONCRETE 53
0 4
E -20
O
X
E
. C
o
- 4 0
m
- 6 0 I ! i BI 0
0 2 0 4 0 6 0 1 0 0
5 , 0
Z', k Oh mx c m2
i i i i
- 1 0
4 , 5 - 2 0
o Q
X
E -30
,,..
O 4 , 0 O' ;
C
. ~ - 4 0 t ~
u
0 - 5 0 ~ "
. . . I 3 , 5 I 1 .
I
- - - - o - - - L o g I Zl - 6 0
3 , 0 I 7 0
- 4 - 3 - 2 - 1 0 1 2
Log ( Fr eq) , Hz
FIG. 12--Nyquist and Bode plots obtained from the same galvanic cell from which the EN recordings
of Fig. 10 were taken. E~ = -393 mV versus SCE.
E N M Re s u l t s
We wi l l gi ve s o me e x a mpl e s o f the t ypes o f resul ts obt ai ned by ENM. Nume r i c al val ues
are c o l l e c t e d in Tabl e 1, t oget her wi t h Ro (the z e r o f requency l i mi t o f 1~), and, whe n avai l -
abl e, Rc, ( t he charge transfer res i s t ance as s oci at ed wi t h the shorter o f the t i me cons t ant s ) ,
obt ai ned f r om EIS on the s ame concret e bl o c ks and wi t hi n t wo or three days f r om the date
o f the ENM.
The first three r ows o f Tabl e 1 concern c as e s in whi c h the corros i on res i s t ance i s not very
large: the corros i on rate, f r om EIS, wa s es t i mat ed t o be about 0. 5 ~ A / c m 2 for the t wo
s ampl e s i mme r s e d in 0. 2M s o di um chl ori de ( NaCI) ( Sol ut i on 1, first and s e c ond row) , and
about 0. 2 ~ A / c m 2 for the s pe c i me n in Sol ut i on 2, third row. Bot h vol t age and current no i s e
l e ve l s we r e we l l above the i nst rument al l i mi t s (as gi ve n in the s e c t i on ent i t l ed "Dat a Ac -
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
E
c-
o
Z
UJ
0,0
0,2
-0,2
-0,4
-0,6
0
0 , 3
I
500 10100 I 5L00
Ti me, s
1 I I
0,2
E
O 0,1
e -
~ o,o
0 -OJ
-0,2
0,4
000
54 EL ECT ROCHEMI CAL NOI SE ME A S U R E ME N T
-0,3 I / i
500 1000 1500 000
Ti me, s
F I G. 13- - F i l ter ed E N recordi ngs. Sol uti on 2: (a) Vol tage. Acqui si ti on rate = 0.5 pts/ s. (b) Current.
Acqui si ti on rate = 0.5 pts/ s.
qui si ti on"). Al s o, the choi ce o f the digital filtering procedure was in general o f little i m-
portance, increasing conf i dence in the results.
Figure 10 s hows filtered ti me records o f vol t age and current fluctuations. In this case, the
correlation bet ween the t wo i s good, i ndi cati ng that the el ectrode in c o mmo n to the t wo
gal vani c cel l s was generating mos t o f the noi se. The correspondi ng power spectra are s hown
in Fig. 11. The vol t age spectrum flattens above 50 mHz because it approaches the instru-
mental noi se ( compare Fig. 3a) , whereas the spectrum o f the current fluctuations, whi ch is
still more than one order o f magni tude larger than the instrumental noi s e at the hi ghest
frequency o f 0. 25 Hz ( compare Fig. 3b), has throughout a sl ope o f about - 1 . EIS spectra
on the same speci men are s hown in Fig. 12. Al l these data were obtai ned in Sol ut i on 1.
The ENM recordi ngs obtai ned in Sol ut i on 2 are s hown in Fig. 13 in whi ch the correlation
i s onl y partial. Here, al though the current fluctuations are o f the s ame order as in t hos e in
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
BERTOCCI ON DETECTI ON OF CORROSI ON IN REI NFORCED CONCRETE 5 5
E
O~
O
Z
t/J
1,0
0,8 t
0,6
0,4
0,2
0,0~
-0,2
-0,4
-0,6
-0,8
-1,0
i
I
500
10100 15100 2000
Ti me, s
i i i
a
u
4 . ,
e-
= =
I .
O
Z
i l l
0
-1
-2
-3
f
' o o 500 10 0 15 0
Ti me, s
1;
200
FIG. 14- - Fi l t er ed EN recordings. Solution 3: (a) Voltage. Acqui si t i on rate = 0. 5 pt s / s . (b) Current.
Acqui si t i on rate = 0. 5 pt s / s .
pure NaC1, the standard devi ati on o f the vol t age fluctuations is larger, l eadi ng to a larger
val ue o f X, , in agreement wi th EIS results.
On the speci men e xpos e d to a magnes i um chl ori de (MgCi2) sol uti on wi th a citrate inhibitor
( Sol ut i on 3), fairly regularly spaced transients were observed, as s hown in Fig. 14. Such
behavi or suggest s l ocal i zed corrosi on, but both ENM and EIS results gi ve a fairly hi gh
resistance, at about 5 0 0 kl ~ c m 2, whi ch shoul d correspond to a l ow corrosi on rate o f the
order o f 50 nA/ c m 2. Becaus e the transients occurred in both vol t age and current at the same
time, the curve o f the standard devi ati on ratio X, is not affected, as s hown in Fig. 15.
Therefore, si mpl e moni tori ng o f X, woul d not have reveal ed this behavior.
The last t wo rows in Table 1 refer to data taken on a dei ci ng salt wi th an organic l i gno-
sulfate inhibitor ( Sol ut i on 4) and on an alternative deicer, cal ci um magnes i um acetate ( CMA)
( Sol ut i on 5), whi ch both exhibit hi gh Ro wi th EIS, i ndi cati ve o f a very l ow corrosi on rate.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
56 ELECTROCHEMICAL NOISE MEASUREMENT
E
c-
O
v
X
40
30
20
10
i 1 i i j
o J 9 ,
0 500 100 0 1500 2000
Ti me, s
FIG. 15--Standard deviation ratio X, obtained from the recordings of Fig. 14.
Here the EN, particularly in the voltage channels, is barely above the instrumental limit, and
the time records seem to be devoid of meaningful content, as shown in Fig. 16. In these
cases, the instrumentation appears to have insufficient sensitivity.
Di s cus s i on
One of the major problems that affect the ENM presented here stems from the lack of
corrosion uniformity in the electrodes, which leads to large DC differences and slow drifts.
This undermines one of the assumptions made in the model used in the analysis of the
fluctuations. It may be argued that a zero average in the current and voltage would be
achieved if the measurements were extended for arbitrarily long times, but this does not
solve the problem in most practical cases. It can be conceded that differences in behavior
are, in general, more pronounced in slowly corrodi ng systems, of which steel in concrete is
obvi ousl y an example, but experience suggests that such differences are common. I f vanish-
ing differences in Eoc were a necessary requirement, the applicability of this method would
be severely restricted.
A second problem is that of the most appropriate way to eliminate the noise below a
certain frequency, including the DC component. Results presented here show that onl y when
the choice of the filtering procedure has little influence on the value of the standard deviation
can there be confidence that significant fluctuations in the frequency range under examination
have been detected. Many authors have preferred to use analog high-pass filters in noise
measurements. This may turn out to be a better approach, but it should be remembered that
analog filters affect the signals just as digital filters, so that the issues about the validity of
the experimental data are present in this case too, although they may be less obvious.
In some of the data obtained in this work, the values of X, were smaller than expected.
This was the case when the corrosion rate was low, the noise in the voltage measuring
channel was onl y marginally greater than that produced by the instrumentation, and signif-
icant fluctuations could be detected in the current channel. This is in contradiction to our
understanding of the phenomenon: i f the electrode impedance is large, whatever causes
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
BERTOCCI ON DETECTI ON OF CORROSI ON IN REI NFORCED CONCRETE 5 7
0,2 i i i
0,1
ca o, o a
..=.
o
>
z
tJJ
-o,1
-0,2
0 5 0 0 1 0 0 0 1 5 0 0 2 0 0 0
Time, s
0, 3 j , ,
0 , 2
0, 1
c
o,o b
e=
,-.t
(=) -0, I
Z
tkl
-0, 2
- 0 , 3 i * I 20100
5 0 0 1 0 0 0 1 5 0 0
Time, s
FI G. 16- - Fi l t er ed EN recordings. Solution 4: (a) Voltage. Acquisition rate = 0. 5 pt s / s . (b) Current.
Acqui si t i on rate = 0. 5 pt s / s .
vol t age fl uctuati ons s houl d i nduce s mal l currents. Such a puz z l i ng resul t cannot be di s mi s s e d
as an ef f ect o f l i mi t e d s ens i t i vi t y o f the i nst rument at i on, whe n the current fl uctuati ons are
cl earl y above the i nstrumental l i mi t .
In c onc l us i on, the data present ed here i ndi cat e that EN moni t ori ng has cons i derabl e l i m-
i tati ons whe n appl i ed t o corros i on i n rei nf orced concret e, and that further st udi es on ot her
e l e c t r oc he mi c al s y s t e ms , as we l l as c ompar i s on wi t h di fferent experi ment al t echni ques are
neces s ary bef ore the me t hod can be us ed wi t h conf i dence.
Re f e r e n c e s
[1] Searson, P. C., Dawson, J. L., and John, D. G., UMIST Conference on Electrochemical Methods in
Corrosion Testing and Research, UMIST, Manchester, UK, 1982, p. 1.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
58 ELECTROCHEMICAL NOISE MEASUREMENT
[2] Zevnik, C., "Electrochemical Noise Corrosion Monitoring Method in Laboratory Use," Proceedings
of the International Conference on Metallic Corrosion, Madras, Oxford & IBH Publ. Co., New
Delhi, 1987, p. 4061.
[3] Kendig, M., Jeanjaquet, S., and Mahoney, M., "Electrochemical Noise Analysis of the Corrosion of
Aluminum Alloys and Composites," Paper 383, CORROSION 88, NACE, Houston, TX, 1988.
[4] Rothwell, A. N. and Eden, D. A., "Electrochemical Noise Techniques for Determining Corrosion
Rates and Mechanisms," Paper 223, CORROSION 92, NACE, Houston, TX, 1992.
[5] Eden, D. A. and Rothwell, A. N., "Electrochemical Noise Data: Analysis, Interpretation and Pres-
entation," Paper 292, CORROSION 92, NACE, Houston, TX, 1992.
[6] Bertocci, U. and Huet, E, Corrosion, Vol. 51, 1995, p. 131.
[7] Legat, A. and Zevnik, C., Corrosion Science, Vol. 35, 1993, p. 1661.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
F l ori an Mansfel d 1 and Hong Xi ao 1
Electrochemical Noise and I mpedance
Analysis of I ron in Chloride Medi a
REFERENCE: Mansfeld, E and Xiao, H., "Electrochemical Noise and Impedance Analysis
of Iron in Chloride Media," Electrochemical Noise Measurement f or Corrosion Applications,
ASTM STP 1277, Jeffery R. Keams, John R. Scully, Pierre R. Roberge, David L. Reichert, and
John L. Dawson, Eds., American Society for Testing and Materials, 1996, pp. 59-78.
ABSTRACT: Potential and current noise data have been collected for pure iron foils exposed
to 0.5N sodium chloride (NaC1) that was aerated, deaerated, or aerated with sodium nitrite
(NaNO2) added as inhibitor. Potential and current noise data were obtained at the beginning
of each hour over a 24-h period either sequentially or simultaneously. Statistical and spectral
analyses were performed resulting in noise resistance values. For the noise resistance R, =
~rVIo-l, where ~rV and crl are the standard deviation values of potential and current, respectively,
similar values as for the polarization resistance Rp determined with electrochemical impedance
spectroscopy (EIS) at the end of the tests were obtained. The spectral noise resistance R~
obtained from spectral analysis of simultaneously collected current and potential data agreed
satisfactorily with Rp and R, for aerated and inhibited solutions. Electrochemical noise and
impedance data have also been obtained for two polymer coatings on cold-rolled steel during
immersion in 0.5N NaC1 for five months. The two types of measurements clearly distinguished
between the relatively poor performance of the alkyd system and the excellent performance of
the epoxy polyamide system. Similar trends with exposure time were observed for the pore
resistance Rpo and R e, on the one hand, and R. and R~ on the other hand. However, the
numerical values of the parameters obtained with EIS were different from those obtained with
electrochemical noise analysis (ENA).
KEYWORDS: noise resistance, spectral noise resistance, polarization resistance, iron, inhib-
itor, polymer coating, electrochemical noise analysis, electrochemical impedance spectroscopy
El ect r ochemi cal noi se anal ysi s ( ENA) is uni que among al l el ect r ochemi cal t echni ques used
i n cor r osi on research because no ext er nal si gnal needs to be appl i ed for t he col l ect i on of
exper i ment al data. Ther ef or e, ENA is of great i nt er est for cor r osi on st udi es especi al l y i f one
consi der s t he addi t i onal benef i t t hat t he cost s of i ns t r ument at i on are great l y r educed i n c om-
par i son wi t h ot her popul ar t echni ques. I n general , a zero r esi st ance ammet er ( ZRA) and one
or t wo vol t met er s under comput er cont r ol are suffi ci ent . It is obvi ous t hat ENA has great
pr omi s e not onl y for basi c l abor at or y st udi es, but perhaps even mor e so for cor r osi on mon-
i t ori ng. Ther ef or e, it is not t oo sur pr i si ng t hat a n u mb e r of r esear ch gr oups have r epor t ed
resul t s obt ai ned ma i nl y wi t h t he me a s ur e me nt of pot ent i al noi se. Ther e has been a t endency
to rel at e al l el ect r ochemi cal noi se dat a to l ocal i zed cor r osi on; however, recent l y, t he pr esent
aut hors have f ound [1-5] t hat pot ent i al noi se can be mi s l eadi ng i n cases i n whi ch s mal l
f l uct uat i ons i n mass t r anspor t cont r ol can pr oduce l arge changes i n t he cor r osi on pot ent i al
Ecorr. Such is t he case for met al s and al l oys t hat are i mmu n e to cor r osi on such as pl at i num
Professor and Ph.D. student, respectively, Corrosion and Environmental Effects Laboratory (CEEL),
Department of Materials Science and Engineering, University of Southern California, Los Angeles, CA
90089-0241.
Copyrighl 9 1996 by ASTM lntcrnational
59
www.astm.org
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
60 ELECTROCHEMICAL NOISE MEASUREMENT
(Pt) or are passive such as stainless steels in neutral, aerated environments. Mansfel d and
Xi ao [1-3] have shown that potential noise as analyzed with the use of power spectral density
(PSD) plots was similar for Pt and for an AI / Si C metal matrix composite, which pits severely
in aerated 0. 5N sodium chloride (NaCl). However, the current noise PSD plots were very
different for the two materials. To obtain more information about experimental and theoretical
details of ENA, the well-known syst em Fe/ NaCI has been investigated in three solutions of
different corrosivity. Dat a analysis has been carried out using both spectral and statistical
analysis. The system Fe/NaC1 was also of interest because Lumsden et al. [6] found poor
agreement between the noise resistance R, obtained with ENA and the polarization resistance
Rp obtained with electrochemical impedance spectroscopy (EIS).
As discussed in a recent review [7], one of the most successful applications of EIS has
been in the evaluation of the protective properties of pol ymer coatings and their degradation
with exposure time. It was of interest in this study to evaluate whether similar information
concerning the coating degradation can be obtained with ENA. Mills et al. recently deter-
mined R, and the mean current I,~ for a number of different pol ymer coatings and concl uded
in tests lasting 185 h that ENA distinguished between paints that were blistering and those
that were not [8].
Experimental Approach
Materials and Methods
Materials--Pure iron foils (99.999%, 0. 1-mm thickness) were exposed to 0. 5N NaC1 that
was open to air or deaerated. Samples were polished to 1200-grit finish, then degreased in
50~ hexane for 20 min. Pol ymer-coat ed cold-rolled steel with an alkyd coating system (CR-
2) or an epoxy pol yami de coating (CR-9) was exposed to the 0. 5N NaCl (open t o air). These
two coatings have been studied earlier in detail with EIS [9-11].
Methods--The experimental arrangement for the measurement of potential and current
noise is shown schematically in Fig. 1. Two pieces of the same iron foil or pol ymer-coat ed
steel with 5-cm 2 exposed area each were used as test electrodes. These two samples were
placed vertically and parallel to each other in a specially designed test cell and connected
to a Schlumberger model 1286 potentiostat, which was used as a ZRA to apply 0 mV
potential between the two electrodes. A saturated calomel electrode (SCE) reference elec-
trode was placed between the t wo electrodes. The potential of the coupl ed electrodes versus
the reference electrode was measured using a Hewlett Packard (HP) 3457A multimeter. The
current output of the potentiostat was connected to the rear channel of a computer-controlled
HP 3478A multimeter providing the current noise record (see Fig. 1).
For the iron foil couple, the electrochemical potential and current noise were initially
recorded sequentially during a 24-h period. The t wo test electrodes were kept at 0- mV
potential difference from the start of immersion until the end of the test. At the beginning
of every hour of immersion time, the potential noise was recorded first, immediately followed
by the coupling current noise measurement. For both the potential and current measurement,
a 2-poi nt sl s sampling rate was used for 500 s in each measurement for the tests with iron
as in the tests by Lumsden et al. [6]. The timing and measuring mode switching were
controlled by an IBM-AT compatible computer. A comput er program was developed to read
and store the noise data as well as to remotely control the experiment through the general-
purpose interface bus (GPIB) (IEEE-488) interface [1-4]. The simultaneous sampling of
current and potential noise data makes it possible to reveal the relationships between potential
and current fluctuations in the frequency domain [4,5].
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
MANSFELD AND XlAO ON ANALYSIS OF IRON IN CHLORIDE MEDIA 61
".~
I
~5
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
62 ELECTROCHEMICAL NOISE MEASUREMENT
To determine the noise behavior of iron in solutions of different corrosivity and at different
potentials, three solutions were used: (1) 0. 5N NaC1 open to air, (2) 0. 5N NaC1 deaerated
with N2, and (3) 0. 5N NaC1 + 0. 01M sodium nitrite (NaNO2) ad inhibitor, open to air. After
recording the electrochemical noise data for 24 h, the impedances of the samples were
measured using a two-electrode arrangement. The experimental noise data were further pro-
cessed using bot h statistical analysis and simple spectral analysis.
In addition to the sequentially collected data, potential and current noise were recorded
simultaneously for iron in the aerated and inhibited solutions. For the coated steel, mea-
surements were made using a slightly different two-voltmeter arrangement and a more ad-
vanced control program [4,5]. The sampling rate was 2 poi nt s/ s, and each test was carded
out for 1024 s.
Dat a Anal ysi s
For all tests, the mean value and the standard deviation of the potential and current fluc-
tuations were determined. From these data the noise resistance R. defined as
R . = ~ V / ~ I (1)
was calculated, where crV is the standard deviation of potential fluctuation and crl is the
standard deviation of current fluctuations [12]. Experimental values of R. for carbon steel
in NaC1 have been reported by Lumsden et al. [6], who tried to correlate R. with R. deter-
mined with EIS.
In addition to the statistical analysis, the noise data obtained by simultaneous collection
of potential and current noise were anal yzed in the frequency domain using fast Fourier
transform (FFT). To evaluate the frequency dependence of the electrochemical noise data,
the spectral noise response Rs . ( f ) was calculated at each frequency f as
R ( f ) = V ( f ) / I ( f ) (2)
and
R s . ( f ) = j R( f ) [ = [Rre 2 ql_ R i , n 2 ] l / 2 (3)
where V( f ) and I( f ) are compl ex numbers obtained from FFT [4,5]. A new term called
the spectral noise resistance R~ introduced by the authors is defined as the value of R s . ( f )
at f = 0:
R~ = lim {R, , (f)} (4)
f ~0
Experimental Results and Discussion
The experimental results obtained for iron in NaCI with sequential collection of potential
and current noise data and the results of the data analysis are displayed in Figs. 2 through
7. Figure 2 shows experimental potential and current noise data after exposure for 1 h to
the deaerated NaC1 solution. For the example in Fig. 2, Eco, decreased by about 10 mV in
500 s. For the current, fairly rapid fluctuations of low amplitude ( <0. 3 ~LA) were observed.
The mean values of Ecor~ for iron exposed to NaC1, which was open to air, deaerated, or
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
MANSFELD AND Xl AO ON ANALYSIS OF IRON IN CHLORIDE MEDIA 6 3
0 . 5 " c u r r e n t - - pot ent i al - 7 5 0
i - 7 5 5 ~ -
0 . 2 5 " ~ ' ~ i i : ., : , : , ! i : , : ,
~ , , . , ~ ' ' , " " , ~ . , , , , ' ~ ; t ~ . ' . . . . . . , , ~ ' ~ ~ . ' ; ? , ' " , ' , . " ~ , ; . ' , ' ~ , ' , ' ' * . , , t " , , " , , , , '
o ~ : , : ~ : ~ : , , , . , , , . , , , . , , : ' . . , , ~ . : , . , , , . . . . , . . . . , , , , . . , ,
' ' ' " " " ' :" - 7 6 5 uJ
- 0 . 2 5 ~ ~ , , , - 7 7 0
0 1 2 0 2 4 0 3 6 0 4 8 0
S a m p l i n g t i m e ( s e c . )
FIG. 2--Potential and current noise recorded f or iron in 0.5N NaCI (deaerated) after exposure f or
l h .
open to air with addition of 0. 01M NaNO2, are shown in Fig. 3. Figure 4 presents the mean
values of the coupling current I,,. Eco . in the aerated solution decreased slowly with time as
the rust layer formed. In the deaerated solution Eco ~ was more negative and fairly constant
after a few hours of immersion. In the aerated solution containing NaNO2, positive potentials
typical for this type of inhibitor were observed. I,, was very small for the deaerated and the
inhibited solution. Bot h positive and negative current peaks occurred in the aerated solution.
The data shown in Fig. 4 are different from those reported by Lumsden et al. [6], in which
a very large DC current of 700 to 750 I~A occurred, indicating large differences in the
corrosion kinetics of the two steel samples used by these authors.
Figures 5 and 6 show the standard deviation values of potential and current, respectively.
The largest ~rV values in the range between 1 and 10 mV are observed for the inhibited
solution in which iron is in the passive state and the fluctuations of E~o,~ are larger, whi ch
is similar to the results reported previously for platinum [1-3]. For the deaerated solution,
the potential fluctuations are very small and fairly constant (about 0.1 mV). These obser-
vations are consistent with earlier findings [1-3] that mass transport of oxygen can play a
large role in potential noise. The (rl values are the largest and show the highest variations
with time for the aerated solution in which significant corrosion occurs (see Fig. 6). For the
deaerated and the inhibited solutions, much smaller values of (rl are observed with larger
fluctuations for the inhibited solution. In these solutions the sample surfaces remained shiny
during the 24-h test.
The noise resistance R, , which is derived from the data in Figs. 3 t hrough 6 accordi ng
to Eq 1, is shown in Fig. 7 for the three test solutions. The highest values of R, (10 to 100
kO) are observed for the inhibited solution. For the deaerated solution, a very stable time
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
64 ELECTROCHEMICAL NOISE MEASUREMENT
-100,
-200
ua -300
(..)
r
> -400
>
E
"-" -500
E
O
-600
--~ NaCI / ai r
-800
0
-700
+ " ~ I I - ~ [ I I I t I I t I I I I I I I I i i
6 12 18 24
Exposure Ti me (hours)
FIG. 3- - Ti me dependence of the mean Ecorr f or iron in three solutions.
- - NaCI / N2
NaCI/NaNO,~
f l -
y
E
3 o
20
10
0
-10
-20
-30
0
l J-
6 12 18
Exposure Ti me (hours)
FIG. 4- - Ti me dependence of the mean coupling c u r r e n t I m f or iron in three solutions.
24
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
MANSFELD AND Xl AO ON ANALYSIS OF IRON IN CHLORIDE MEDIA 65
E
v
>
1 O 0 1 4 _
NaCI / ai r
/
t - F NaCI / N2
' ,. ~ NaCI/NaNO21
I
0.1
0 . 0 1 J I I
0 6 12 18 24
Exposur e Ti me (hours)
FIG. 5--o' V values f or iron in three solutions.
v
b
100
10
0,1
0.01
0
-=- NaCI / ai r
- r - NaCI / N2
->~ NaCI / NaNO2
I I I
6 12 18 24
Exposur e Ti me (hours)
FIG. 6--o' 1 values f or iron in three solutions.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
66 ELECTROCHEMICAL NOISE MEASUREMENT
1E+07
1E+06 t
1E+05
E
c-
O 1E+04
v
C
r r
1E+03
I E+02
1E+01
0
4- NaCI / ai r
-}- NaCI / N2
NaCI / NaN02
6 12 18 24
Exposure Time (hours)
FIG. 7--Noise resistance R. for iron in three solutions.
behavior is observed with R, slowly increasing from about 1 to about 10 kl'l after 24 h. The
sample exposed to NaC1/air has the lowest R, values with the largest fluctuations between
10 fl and 5 kll
Figure 8 shows the impedance spectra for the iron samples after exposure in the three
solutions for 24 h. The spectra have been fit to a one-time-constant model to obtain Rp,
which is listed in Table l for comparison with R, determined for 24 h of exposure. Fairly
good agreement between these two parameters is observed. Lumsden et al. reported much
larger values of R, than Rp for carbon steel in NaCI [6]. Some of these differences could be
due to the dissimilar nature of the steel samples used by these authors.
The results of the analysis of simultaneously recorded potential and current noise data are
presented in Figs. 9 through 11 for the aerated NaCI solution with and without NaNO2.
Similar R, values were determined (Fig. 9) as in the experiments with sequential recording
of potential and current noise (see Fig. 7). This result demonstrates that for relatively short
sampling times sequential and simultaneous recording of noise data provides similar results.
However, for spectral analysis, in which potential and current fluctuations are correlated in
the frequency domain, simultaneous recording of potential and current noise data is neces-
sary. Figure 10 shows the spectral noise response R,, calculated from Eqs 2 and 3 as a
function of frequency for the first (see Fig. 10a) and last hour (see Fig. 10b) of exposure.
The frequency independent low values of Rs, (see Fig. 10) for the aerated solution indicate
uniform corrosion at high rates in contrast to the curves with higher and frequency-dependent
Rsn values for the inhibited solution, which show larger scatter (see Fig. 10). From curves
obtained for each hour of exposure, such as those shown in Fig. 10a and b, the spectral
noise resistance R~ was determined by extrapolation to f = 0 according to Eq 4. These
R~ values are plotted in Fig. 11 as a function of exposure time for both solutions. A
comparison of the results in Fig. 9 and Fig. 11 shows that R, as defined in Eq 1 (see Fig.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
MANSFELD AND XlAO ON ANALYSIS OF IRON IN CHLORIDE MEDIA 67
1E+05
90
E
t -
O 1E+03
N
-=- NaCI / ai r
-><- NaCI / NaNO2
- - NaCI / N2
1E+04
) _
1 E + 0 2
1 E + 0 1 I I I I L . . . . 0
0.01 0.1 1 10 1 O0 1000 10000
Frequency ( Hz)
FIG. 8 Bode-plots for iron after immersion in three different solutions f or 24 h.
60 (B
" 0
fl)
O3
c "
<
Q ~
30 ~
I -
9) and R~ as defined in Eq 4 (see Fig. 11) have similar values at each exposure time. A
compari son of the 24-h values of R, , R~ and Rp is given in Table 2 for the t wo test
solutions. Good agreement between R, and R~ obtained with t wo different experimental
and analysis techniques and Rp determined with EIS is observed.
The potential fluctuations for the alkyd coating CR-2 after 28 and 130 days of exposure
are shown in Fig. 12a; the corresponding current fluctuations are displayed in Fig. 12b.
Considering that the E28 scale in Fig. 12a is ten times larger than the E13o scale, it is clear
that the potential fluctuations in the early stages of exposure are much larger than after
longer times. On the other hand, in Fig. 12b, the current fluctuations at the 28th day were
much smaller than at the 130th day of exposure (note the ten times difference between the
12g and lj3 o scales). This qualitative evaluation of the experimental data suggests that, initially,
potential noise levels are high and current noise levels are low. As the coating degrades,
potential noise levels decrease and current noise levels increase, For the more protective
epoxy pol yami de coating CR-9, no systematic changes of noise levels with time could be
detected [13].
The time records for potential and current noise were first analyzed statistically by cal-
culating the mean and the standard deviation for each data set to obtain R, (Eq 1). The
TABLE 1--Comparison of R, and Rp f or three different test
environments,
Environment Open to Air Deaerated (N2) Inhibited (NaNO2)
R,(kfl) 0.8 8.3 54.7
Rp(kD) 1.6 15.9 22.1
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
6 8 ELECTROCHEMI CAL NOISE MEASUREMENT
A
E
r
O
r
r l -
10 6
1 0 5
10 4
10 3
1 0 2
, N a C l / a i r
~< N a C I / N a N O 2
I I I
0 6 12 18 2 4
Exposure Ti me (hours)
FIG. 9- - Noi s e resistance R n determined f rom simultaneously recorded pot ent i al and current noise
data f or iron in aerat ed and in aerated, inhibited 0. 5N NaCI.
experimental data were also transformed from the time domain into the frequency domain
by FFF to determine R,.( f ) (Eqs 2 and 3). The R,.( f ) functions for CR-2 and CR-9 after
28 and 130 days of exposure in Fig. 13 show that the noise spectra for CR-2 and CR-9 after
28 and 130 days of exposure are independent of frequency. At the 28th day of exposure,
the R s. values (Eq 3) for CR-2 are about one order of magnitude lower than for CR-9,
indicating the CR-2 had already started to degrade. At the 130th day of exposure, the dif-
ference in Rs. for CR-2 and CR-9 was about four orders of magnitude over the entire
frequency region.
Figure 14 shows R. and R~ as a function of exposure time for CR-2 and CR-9. For
CR-2, the same trend of R. and R~ to decrease with time can be found as was observed
for Rp and Rpo determined from EIS. For CR-9, R. and R~ remained at much higher values
and did not change significantly with time, indicating that degradation had not occurred,
which is consistent with previous conclusions [9-11]. Similar to the Fe/NaC1 system dis-
cussed above, R. and R~ had very similar values despite the fact that they are mathemati-
cally unrelated.
Figure 15 shows that the time dependence of Rp, Rpo, R . , and R~ for CR-2 is very
similar. However, the numerical values of the parameters obtained from ENA are much lower
than those obtained from EIS, which raises questions concerning the relation of R. and R~
to the corrosion/coating degradation processes for the systems under investigation.
Another important part of spectral analysis is the calculation of the power spectral density
(PSD) functions of the noise data. Being another frequency domain representation of a given
signal, the PSD function is defined as the power of a signal at a given frequency and can
be obtained by FFF or the maximum entropy method (MEM) [14]. A previously developed
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
MANSFELD AND XIAO ON ANALYSIS OF IRON IN CHLORIDE MEDIA 6 9
1 0 7
1 0 6
. . . , 1 0 5
E
e -
o
v
= 1 0 4
M )
1 0 3
1 0 2
. =
iT"
: e )
w
N A N 0 2
i I I I i i i ]
. 0 0 1 , 0 1 .1 1
1 0 7
1 0 6
1 0 5
1 0 4
1 0 3
1 0 2
. 0 0 1
F r e q u e n c y ( H z )
b )
N A N 0 2
a i r
A - 1 1 - ~ " W - ' a ' ' ~ t ' ~ ' A 1 l i i i i i i i
i i i I I I I t I J i i i i t l I I i I i i I I
. 0 1 . I I
F r e q u e n c y ( H z )
FIG. lO--Frequency dependence of spectral noise resistance Rs, for iron in aerated and for aerated,
inhibited 0.5N NaCl after exposure for 1 h ( a ) and 24 h ( b ) .
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
70 ELECTROCHEMI CAL NOISE MEASUREMENT
A
E
t , -
o
v
c
r162
O
nr-
106
10 5
10 4
103
10 2
, NaCI / ai r
~ NaCI / NaNO2
I _ _ I [
0 6 12 18 24
Exposure Ti me (hours)
FIG. 1 l--R~ values as a function of exposure time f or iron in aerated and aerated, inhibited
0.5N NaCl.
MEM comput er program was applied in these experiments to calculate the PSD functions
for potential and current fluctuation. Figure 16 presents potential (Fig. 16a) and current PSD
curves (Fig. 16b) for CR-2 at different exposure times, whereas Fig. 17 shows the potential
(Fig. 17a) and current (Fig. 17b) PSD plots for CR-9. As the exposure time increased, the
potential PSD curves for CR-2 decreased in magnitude with similar slopes except for the
130th day (see Fig. 16a). This is another presentation of the results in Fig. 12a indicating
that potential PSD decreased with increasing degradation of the protective properties of the
coating. In Fig. 17a, which shows the potential PSD for CR-9, no obvious pattern of changes
with exposure time can be seen. Potential fluctuation levels were higher for CR-9 than for
CR-2 at all exposure times. The current PSD for CR-2, which presents the frequency de-
pendence of the experimental results in Fig. 12b obtained in the time domain, increased with
exposure time maintaining a constant slope of about - 2 0 dB (Fig. 16b). On the other hand,
TABLE 2--Comparison of R., R~ and Rp for simultaneously
recorded data in two solution environments, a
Environment Open to Air Inhibited (NaNO2)
n,(kl2) 0.4 195.8
R~ 0.4 79.1
R ~ ( k f l ) 1 . 4 220.0
a The exposed area of each sample is 5 cm2; the values of R,
and R~ were recorded at 24-h exposure time.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
M A N S F E L D A N D X I A O O N A N A L Y S I S O F I R O N I N C H L O R I D E M E D I A 71
u. I
r
W
>
==
uJ
-490
-500
-510
a )
0
>
J
J
<
' I ' I ' I
300 600 900
Sampling Time (sec.)
-0.07
-0.09 -
g
-0.11
-0.13
b )
2 8 t h day
<
- 4 4 8
o
-449
>
E
- 4 5 0
0.0
- - 0. 2
- !
J
-0.4
-0.6
' I ' I ' I
0 300 600 900
Exposure Time (sec.)
FIG. 12- - l i me records of potential ( a ) and current fluctuations ( b ) f o r CR-2 after 28 and 130 days
of exposure.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
7 2 E L E C T R O C H E M I C A L N OI S E MEASUREMENT
A
E
t -
O
r
c / )
n--
1010
109
108
107
106
105
104
103
102
.001
--c--CR-2 ( 2 8 d a y )
o c a - 9 ( 2 8 d a y ) , . ~ T ! - , I N , a N
CR-9 ( 1 3 0 d a y ) " . ,. ~ , ' I S . ! : " '
~ ~ ~176 L o ' ~ t = k t k r L l ~ l l t l ~ i t l l ~
!
' d
[ I
.01 .1
Frequency ( Hz )
FIG. 13--Noise spectra for CR-2 and CR-9 after 28 and 130 days of exposure.
A
E
r
0
C
clJ
t , , -
r r -
1 0 9
1 0 8
107
1 0 6
105
104
103
- - t ~- Rn
- - ~ - - R * s n
C R - 9
' I t I '
50 100
Exposure T i me (day)
FIG. 14--Time dependence of R. and R~ CR-2 and CR-9.
1 5 0
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
MANSFELD AND XIAO ON ANALYSIS OF IRON IN CHLORIDE MEDIA 7 3
E
t -
O
t --
r
r r
0
rr-
Q.
r r
10 9
10 8
10 7
10 6
10 5
104
10 3
0 150
' I ' I
5 0 100
E x p o s u r e T i m e ( d a y )
FIG. 15--Time dependence of R., R%, R~, and Rpofor CR-2.
the current PSD for CR-9 (see Fig. 17b) had much lower levels than for CR-2 without an
obvious relation with exposure time. The four PSD plots (see Figs. 16a and 17b) indicate
that the PSD level is correlated qualitatively with loss of corrosion protection provided by
the coating system. From Figs. 16 and 17, it is clear that the frequency bandwidth in the
present study was t oo narrow to show the DC limit po, which apparently occurs at extremely
low frequencies ( < 1 mHz). Therefore, the PSD values at 1 mHz for potential (P%) and
current (P~ PSD plots were used for comparative purposes. The P% and p o values obtained
in this way are shown in Fig. 18, whereas the values of the slope Sp in the linear region of
Figs. 16 and 17 are shown in Fig. 19 as function of exposure time. For CR-2, P% decreased
and P~ i increased with exposure time, whereas for CR-9, P% remained high and p o was very
low (see Fig. 18). The slopes for potential (Sew) and current (Spi) PSD remained constant at
about - 2 0 dB for CR-2 and CR-9, except for Sp~ of CR-2 at the 130th day of exposure.
This constant value of Sp is consistent with previous suggestions [15] that the slope of PSD
plots is a characteristic parameter for a given corrosion mechani sm independent of extent of
corrosion activity.
Summary and Conclusions
The main goal of this study was t o determine whether parameters obtained by statistical
and spectral analysis of potential and current fluctuations agree with those obtained from
EIS, which served as a reference technique. For pure iron in 0. 5N NaC1, which was either
aerated, deaerated, or aerated containing NaNO2 as inhibitor, the mean and the standard
deviation values of Ecor~ and the coupling current between two electrodes of the same material
have been determined on an hourly basis for a 24-h period. Noise data have been collected
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
74 EL ECT ROCHEMI CAL NOI SE ME A S URE ME NT
A
E
r
Cz
r
. m
Q
r
6 0
3 0
-30
- 6 0
. 0 0 1
--c-- 28 day
32 day
- c - 83 day
, 80 day
a ) ~ - ~ . ~ %. ~ . . ~ , __~
I L
. 0 1 .1 1
Frequency (Hz)
~ L
" O
r
Q .
t . -
P
3
O
1 0
--c- 28 day
-s3 day
| . ~ _ ~ - + - 8 0 day
-20 ~ ~ - ~- 1 3 0 day
- 5 0 - ~
b )
- 8 0 I I
. 0 0 1 . 0 1 .1 1
F r e q u e n c y (Hz)
FIG. 16--Potential ( a ) and current ( b ) PSD curves of CR-2 for different exposure times.
sequentially and simultaneously. The latter approach allows construction of spectral noise
plots from which the spectral noise resistance R~ can be determined as the DC limit. R.
and R~ have been determined as a function of exposure time. Satisfactory agreement be-
tween the polarization resistance Rp obtained from EIS and R~ and R. has been observed.
The most relevant parameters obtained from analysis of electrochemical noise data for the
polymer-coated steel exposed to NaC1 are R. . R~ and the PSD parameters revealing in-
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
MANSF EL D AND XI AO ON ANAL YSI S OF I RON I N CHL ORI DE MEDI A 75
A
E
" o
r',,
cn
13.
E
o .
A
. <
=I .
n~
" 1 0
O'3
O,1.
"E
z, -
C.)
6 0
3 0
-30
- 6 0
- 32 day
- 0 - 6 3 day
, 80 day
- ~ 1 3 0 day
I I
. 0 0 1 . 01 .1 1
Frequency (Hz)
10
i 1)) ---o-- 28 day
x 32 day
--=- 83 day
t 80 day
-20 ~ 130 day
- 5 0
- 8 0 l I
. 0 0 1 . 01 .1 1
F r e q u e n c y ( Hz)
FIG. 17--Potential ( a ) and current ( b) PSD curves of CR-g f or different exposure times.
formation concerning coating degradation and corrosion of the substrate from different as-
pects. Very similar values have been obtained for R. and R~ despite the fact that these
parameters have been determined using different methods of data analysis. The decrease of
R. and R~ with time for the coating with relatively poor performance is similar to that of
EIS parameters related to coating disbonding (Rpo) and corrosion at the metal/coating in-
terface (Rp); however, the numerical values are quite different. As far as the PSD parameters
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
7 6 E L E C T R O C H E M I C A L N O I S E M E A S U R E M E N T
, - , z
a:z
o_
=r
>
E
" O
o
a .
6 0
30
- 3 0
- 6 0
-90
O
0 ....
..... " . . O . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
~
....,.--'" "
- 0 - P~ ( CR- 2)
--~--P~ ( CR- 2)
..o,-P~ ( CR- 9)
- - + - P~ ( CR- 9)
I
0 1 0 0 1 5 0
T i me ( d a y )
I
5 0
E x p o s u r e
F I G. 18- - T i me dependence of PSI:)parameters po for potential (P~ and current ( P ~ noise for CR-
2 and CR-9.
N
. q r -
0
" O
- 1 0
- 2 0
O
| - 3 0
>
E
t o
" D
r - 4 0
- o - Spv ( CR- 2)
--~--Spi (CR 2)
. . . . Spv ( CR- 9)
- ( CR- 9)
' I ' I '
0 5 0 1 0 0 1 5 0
E x p o s u r e Ti me ( d a y )
F I G. 19--Time dependence of PSD parameters Spfor potential ( S p y ) and current ( S v i ) noise for CR-2
and CR-9.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
MANSFELD AND XIAO ON ANALYSIS OF IRON IN CHLORIDE MEDIA 77
are concerned, the increase of P~ i obtained from current PSD data and the decrease of P~ v
derived from potential PSD data seem to be useful indicators of coating degradation or
substrate corrosion or both. The slopes Spy and Sei with values of about - 2 0 dB independent
of exposure time and coating performance seem to be characteristic values of the coat i ng/
steel/NaC1 system and the prevailing corrosion mechanism.
Recommendations f or Future Work
Several points require further theoretical and experimental examination. In the PSD plots
for the pol ymer-coat ed steel, a DC limit was not observed even at 1 mHz. On the other
hand, the spectral noise plots were independent of frequency in the examined frequency
range (0.001 < f < 1 Hz). Analysis of EIS data obtained over a very wide frequency range
results in a number of parameters that allow assessment of the changes of coating properties
such as water uptake and decrease of coating conductivity, determination of the extent of
the disbonded area, and estimation of the average corrosion rate of this area [7]. Statistical
analysis of noise data results in onl y one par amet er - - t he noise resistance R, , whereas spec-
tral analysis of the same data produces another par amet er - - t he spectral noise resistance
R~ In future work, it needs to be established to which of the parameters characterizing a
coating system exposed to a corrosive environment R, and R~ are related. A similar ap-
proach is needed for correlation of parameters obtained from PSD and spectral noise plots
with coating properties. At the present state of our understanding of ENA, it has to be
concl uded that this technique can produce qualitative information concerning coating per-
formance and is, therefore, suited for corrosion-monitoring purposes. Important advantages
of ENA are low equipment cost and relative ease of data collection using appropriate soft-
ware, which allows collection of data from multiple corrosion sensors and data transmission
t o a central control facility. 2 This approach is used by the authors in a study of the effects
of microorganisms on the protective properties of pol ymer coatings applied to steel exposed
to natural seawater [16].
Acknowledgment
This project has been funded in part by Mont ana State University under Subcontract
291435-01 and in part by the Office of Naval Research (Dr. A. J. Sedriks) under Contract
N00014-94-1-0026. The simultaneously collected noise data for the Fe/NaC1 system have
been displayed and analyzed by L. T. Han, who is a graduate student at CEEL.
Re f e r e nc e s
[1] Mansfeld, E and Xiao, H., "Evaluation of Electrochemical Techniques for the Study of Localized
Corrosion Phenomena in MIC," NSF-CONICET Workshop on Biocorrosion and Biofouling, Mar
del Plata, Argentina, Oct. 1992, Buckman Laboratory, 1993.
[2] Mansfeld, E and Xiao, H., "Electrochemical Techniques for the Detection of Localized Corrosion
Phenomena in MIC," in Microbiologicatty Influenced Corrosion (MIC) Testing, STP 1232, Amer-
ican Society for Testing and Materials, West Conshohocken, PA, 1994, p. 12.
[3] Mansfeld, E and Xiao, H., "Development of Electrochemical Test Methods for the Study of Lo-
calized Corrosion Phenomena in Biocorrosion," in Proceedings of Biofouling and Biocorrosion in
Natural Waters, ACS, Aug. 1992, Lewis Publishing, 1994, p. 265.
[4] Mansfeld, F. and Xiao, H., Journal of the Electrochemical Society, Vol. 140, 1993, p. 2205.
2 H. Xiao and E Mansfeld, unpublished data, 1994.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
78 ELECTROCHEMICAL NOISE MEASUREMENT
[5] Mansfeld, E and Xiao, H., in Proceedings, 12th International Corrosion Congress, Houston, TX,
Sept. 1993, NACE, p. 1388.
[6] Lumsden, J. B., Kendig, M., and Jeanjaquet, S., presented at CORROSION/92, Paper 224, NACE,
1992.
[7] Mansfeld, E, "The Use of EIS for the Study of Corrosion Protection by Polymer Coatings," Journal
of Applied Electrochemistry, Vol. 25, 1995, p. 187.
[8] Mills, D., Bierwagen, G., Tallman, D., and Skerry, B., in Proceedings of the 12th International
Corrosion Congress, Houston, TX, Sept. 1993, NACE, p. 182.
[9] Mansfeld, F. and Tsai, C. H., Corrosion, Vol. 47, 1991, p. 952.
[10] Tsai, C. H. and Mansfeld, F., Corrosion, Vol. 49, 1993, p. 726.
[11] Mansfeld, E and Tsai, C. H., in Proceedings, 12th International Corrosion Congress, Houston, TX,
Sept. 1993, NACE, p. 128.
[12] Rothwell, A. N. and Eden, D. A., presented at CORROSION/92, Paper 223, NACE, 1992.
[13] Xiao, H. and Mansfeld, F., Journal of the Electrochemical Society, Vol. 141, 1994, p. 2332.
[14] Fougere, P. F., Maximum Entropy and Bayesion Methods, Kluwer Acad. Publ., Boston, 1990.
[15] Searson, P. and Dawson, J. L., Journal of the Electrochemical Society, Vol. 135, 1988, p. 1908.
[16] Mansfeld, F., Xiao, H., Han, L. T., Lee, C. C., Little, B., Ray, R., and Jones-Meehan, J., "Evaluation
of the Effects of Microorganisms on Corrosion Protection of Steel by Polymer Coatings," presented
at 1995 International Conference on Microbially Influenced Corrosion, NACE and AWS, New
Orleans, May 1995.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
Davi d L. Rei chert ~
Electrochemical Noise Measurement for
Determining Corrosion Rates
REF ERENCE: Reichert, D. L. , "El e c t r oc he mi c al Noi s e Me a s ur e me nt f or De t e r mi n i n g
Corros i on Rates," Electrochemical Noise Measurement for Corrosion Applications, ASTM
STP 1277, Jeffery R. Kearns, John R. Scully, Pierre R. Roberge, David L. Reichert, and John
L. Dawson, Eds., American Society for Testing and Materials, 1996, pp. 79-89.
ABSTRACT: Electrochemical noise measurements (ENM), linear polarization tests and mass
loss measurements were performed in sulfuric acid, acetic acid and other solutions. The ENM
data were converted to corrosion rates by calculating the noise resistance, R, = crV/trI where
trV and crl are the standard deviations of the potential and current. Good correlation among
the three methods was obtained for low to moderate corrosion rates, but poor correlation was
observed for high rates. ENM has proven valuable for determining corrosion rates in low-
conductivity solutions, which are not suitable for linear polarization resistance (LPR) testing,
and for measuring very low corrosion rates in which mass loss tests would have required at
least 30 days exposure to provide meaningful results.
KEYWORDS : electrochemical noise, current noise, potential noise, corrosion rate, noise
resistance
Electrochemical noise measurements (ENM) record the fluctuations in current and poten-
tial generated by corroding metals. These phenomena were first reported by Iverson in 1968
[1]. Since then, a number of researchers have been involved in evaluating ENM [2-5].
Roberge et al. collected current noise by measuring voltage drop across a resistor rather than
using a zero resistance ammeter [6]. Scully et al. examined a novel way of analyzing the
data by applying chaos theory [7]. This paper presents a number of systems in which ENM
data were analyzed to determine corrosion rates and correlated with independent measure-
ments such as mass loss or linear polarization resistance (LPR). Systems generating good
and poor correlation are reported. Explanations for the observations are suggested in the
discussion of the results.
Experi ment al
Because ENM was being evaluated as a possible tool for field monitoring, all the ENM
tests were run using three identical electrodes. This avoids the issue of possible noise con-
tribution from a saturated calomel (SCE) reference electrode [8]. The LPR tests used two
identical electrodes. The electrodes were cylindrical and polished to a 600-grit finish. All
electrodes used in a given environment were made from a single heat of material to minimize
differences between the coupons. For sulfuric acid tests, where sample area was a variable,
a long rod was cut into different lengths to obtain 1-, 3-, 5-, and 15-cm 2 samples. In the
n Semorresearch enginee~ DuPont Experimental Station, Rou~ 141, Box 80323, Wilmington, DE
19880-0323.
Copyright 9 1996 by ASTM lntcrnational
7 9
www.astm.org
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
80 ELECTROCHEMICAL NOISE MEASUREMENT
acetic and sulfuric aci d ENM tests, el ect rodes were mount ed in a t ri angul ar pi t ch spaced
about 2 cm apart. The cyl i ndr i cal el ect rodes were t hreaded on a gl ass- encl osed met al rod
and seal ed wi t h a Teflon | gasket. The gamma- but yr ol act one (gBL) tests used a field probe
wi t h a st acked ri ng desi gn. The el ect rodes were each 4.5 cm 2 in area and separat ed by 0.03-
cm- t hi ck Teflon gaskets. The hi gh-puri t y wat er tests used 0. 3- cm- di amet er rods spaced about
2.5 cm apart i nsert ed into an aut ocl ave t hrough Tefl on-l i ned hi gh-pressure fittings. The t ot al
i mmer sed ar ea for the rods was 12 cm 2.
Test sol ut i ons were made in l arge bat ches so t here woul d be no vari at i ons in sol ut i on
composi t i on from cel l to cel l . LPR and ENM tests were run concurrent l y usi ng separat e
el ect rodes and cells. Thi s was to prevent any resi dual effects of pol ar i zi ng the el ect rodes
from affect i ng the ENM data.
Cor r osi on rates were est i mat ed bef or e the t est and el ect rode area was fact ored in to mi n-
i mi ze sol ut i on depl et i on effects. Thi s was especi al l y i mport ant for the l arger el ect rodes in
the hi ghl y corrosi ve sulfuric aci d tests. The test cel l s were pl aced in a wat er bat h to mai nt ai n
a st eady t emperat ure. The aut ocl aves were heat ed wi t h heat t ape power ed by a pr opor t i onal
t emperat ure cont rol l er and wr apped wi t h i nsul at i on to mi ni mi ze t emperat ure drift.
ENM dat a were obt ai ned usi ng CML di gi t al noi se moni t ori ng equi pment and DENI S
software. Thi s comput er - cont r ol l ed syst em has a l ower sensi t i vi t y l i mi t of 10 ~2 A and 1 0 - 7
V. These val ues were obt ai ned by col l ect i ng dat a wi t h t he pot ent i al measur ement l eads
short ed and t he current measur ement l eads l eft as an open circuit. A si ngl e dat a set consi st ed
of 600 dat a poi nt s col l ect ed at a rate of 1 poi nt per second. Mul t i pl e dat a sets were gat hered
for each envi ronment . The LPR tests were run wi t h an EG&G Model 273 pot ent i ost at and
352 soft ware package. The el ect rodes from all but the hi gh-puri t y wat er exper i ment s were
al so used as mass loss coupons.
Data Analysis Method
The ENM dat a were convert ed to cor r osi on rates by cal cul at i ng t he noi se resi st ance, R,
and i nsert i ng it into the St er n- Gear y equat i on in pl ace of R e. R, , was cal cul at ed accordi ng
to the f ol l owi ng equat i on [9, 10]:
R n = ( t rV/ ~rl )
where
r = st andard devi at i on of the pot ent i al noi se and
t rl = st andard devi at i on of the current noise.
( 1 )
Not e that this equat i on is for unfiltered di gi t i zed data. Anal og equi pment t ypi cal l y uses filters
and, in some cases, act ual l y col l ect s a r oot mean square ( RMS) val ue of the pot ent i al and
current noise. In the anal og case, R, is the rat i o of t he mean RMS pot ent i al noi se to the
mean RMS current noi se [11].
For bot h di gi t al and anal og noi se measurement s, when the reference el ect r ode used to
obt ai n t he pot ent i al noi se is a t hi rd i dent i cal el ect rode, t he pot ent i al noi se must be nor mal i zed
to account for t he noi se generat ed by the reference el ect r ode [12]. ~ Because t hey are i dent i cal
el ect rodes in the same envi ronment , the st andard devi at i on of the pot ent i al noi se from each
2 D. A. Eden, personal interviews, Sept. 1990-Dec. 1995.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
REICHERT ON DETERMINING CORROSION RATES 81
electrode will be the same. The measured potential noise will be the combined noise from
both electrodes, but the potential noise from one electrode is the quantity used to calculate
R, . Therefore, ~rVmeas is divided by X/2 to obtain ~V. The %/2 factor comes from the fol-
lowing relationships:
~V. , 0. , = (,~V, ~ + o v ~ ) ~
(2)
given:
crV1 = trV: = oV (3)
then:
, r v = ~ V m o a s / X / 2
where
crVmeas = standard deviation of measured potential noise,
~V~ = standard deviation of potential noise from first electrode, and
trV2 = standard deviation of potential noise from second electrode.
(4)
The LPR and ENM corrosion rates reported are averages of the multiple scans run peri-
odically during the immersion time. The mass loss corrosion rate data are the average rate
of all electrodes. In the sulfuric acid tests, there was considerable variability in the mass loss
rates from test to test despite identical solutions and nominally identical electrodes. There-
fore, separate averages were calculated for the mass loss rates from LPR electrodes and
ENM electrodes. In the other tests, mass loss results were highly reproducible.
Test Results
Ac e t i c A c i d
The first set of tests used 304L stainless steel in 50% acetic acid at 30 and 60~ The
LPR data were corrected for solution conductivity and correlated well with the ENM cor-
rosion rate (Table 1). An inspection of the coupons by optical microscopy did not reveal
evidence of any attack. The low corrosion rate and short test duration resulted in very small
mass loss values. Coupled with the sensitivity level of the balance, the uncertainty in the
mass loss corrosion rate exceeded the calculated rates. Because the LPR technique is an
established method for obtaining general corrosion rates and the LPR and ENM data
matched, this test was not repeated for a longer duration.
TABLE 1- - 316L stainless steel in 50% acetic acid.
Temperature, ~ ENM, mm/y LPR, mm/y Weight Loss, mm/y
30 0.0005 0.0005 0.023"
60 0.0008 0.0008 0.005 ~
a Error as a result of small weight loss and accuracy of balance = +0.025 mm/y.
b Error as a result of small weight loss and accuracy of balance = +0.018 mm/y.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
82 ELECTROCHEMICAL NOISE MEASUREMENT
Hi gh- Pur i t y Wat er
The low corrosi on rate of AI SI 1018 carbon steel in dei oni zed wat er at 230~ was hi gh
enough to generat e sufficient cor r osi on product s causi ng cont ami nat i on pr obl ems. The ex-
t ended t i me requi red for meani ngful mass l oss tests was unaccept abl e to t he pl ant request i ng
the data, so el ect rochemi cal t echni ques were investigated. A stable corrosi on pot ent i al was
not obt ai ned wi t h an EG&G Model 273 pot ent i ost at . Thi s may have been caused by ground
l oops bet ween the grounded pot ent i ost at and the el ect ri cal l y heat ed aut ocl ave. The very l ow
sol ut i on conduct i vi t y woul d have posed a pr obl em for t he LPR t echni que even i f a st abl e
pot ent i al had been obt ai ned. An at t empt was made to col l ect i mpedance spectra, but dat a
were t oo noi sy even at frequenci es above 1000 Hz.
The ENM t echni que successful l y di fferent i at ed bet ween t he i nhi bi t ors and the noni nhi bi t ed
wat er (Table 2). The di spar i t y in the cor r osi on rates for the t wo cal ci um hydr oxi de Ca( OH) 2
i nhi bi t ed sol ut i ons was expl ai ned by t he di fference in t he pH levels of t he sol ut i ons. Ca(OH)2
sol ut i ons are hi ghl y suscept i bl e to a l oweri ng of the pH from absorpt i on of carbon di oxi de
from the air, and this was the suspect ed source of the pH di screpancy. Fi el d trials usi ng
Ca(OH)2 were al so inconsistent.
Sul f uri c Ac i d
Sul furi c aci d (H2SO4) at 40~ offered the abi l i t y to obt ai n a wi de range of general cor-
r osi on rat es on 316L st ai nl ess steel whi l e changi ng onl y one vari abl e. The concent rat i ons
s el ect ed- - 10, 20, 30, and 4 0 %- - we r e expect ed to pr ovi de a range of corrosi on rat es from
hundredt hs t o tens of mi l l i met er s per year [13]. Al t hough this range was observed, t here was
a t remendous amount of scat t er in the data. In 30 and 40% H2SO4, t he initial passi ve film
somet i mes prot ect ed t he 316L from at t ack for mor e than 10 h even t hough i so- cor r osi on
di agr ams put t he corrosi on rat e at l ess than 10 mm/ y for 30% H2SO4 and great er t han 10
mm/ y for 40% H2SO4 [13]. In one LPR test, one el ect r ode was vi gor ousl y gi vi ng off hy-
dr ogen gas, whereas t he ot her coupon was inert. A sl i ght scrat ch wi t h a steel screwdri ver
caused r api d corrosi on at t he scrat ch mark, and wi t hi n art hour the ent i re coupon was r api dl y
corrodi ng.
The l arge amount of var i abi l i t y in the cor r osi on behavi or caused l arge vari at i ons in the
i ndi vi dual corrosi on rat es cal cul at ed from LPR. For one 30% H2SO 4 test, the i ni t i al LPR
rat e was 0.06 mm/ y and t he rate i ncreased over t he next 50 h until it was 43 mm/ y at the
end of t he test. In most of t he 30 and 40% concent rat i on tests, corrosi on rat es di d not fol l ow
a definite trend, j umpi ng up and down and di fferi ng by as much as a fact or of 5.
The 10 and 20% H2SO 4 tests showed r epr oduci bl e behavior. A represent at i ve pl ot of dat a
f r om t he 10% H2SO 4 1-cm 2 dat a set is shown in Fi g. 1. These ENM dat a are t ypi cal of
TABLE 2--Carbon steel in deionized water + inhibitors
at 230~
Inhibitor pH ENM, mm/ y
None 5 0.0075
l0 mg/ L Ca(OH) 2 l 0 0.0004
10 mg/ L Ca(OH)z 8 0.0032
25 mg/ L Na2MoO4 + 25 mg/ L Na2SiO 3 9 0.0007
75 mg/ L Na2SiO 3 9 0.0008
100 mg/ L NaNO 2 9 0.0495
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
E
E
c
? ,
-0. 4400
-0. 4405
-0. 4410
-0. 4415
-0. 4420
-0. 4425
-0.4430
Pot ent i al Noi se f or 316L in 10% Sul f uri c Aci d at 40"C
! i i I :
100 200 300 400 500 600
Ti me [S]
Cur r ent Noi se f or 316L in 10% Sul f uri c Aci d at 40"C
6, 50E-05
i i i I
0 100 200 300 400 500 600
6. 48E-05
6~46E-05
6. 44E- 05
6. 42E-05
6. 40E-05
6. 38E-05
6. 36E-05
REICHERT ON DETERMINING CORROSION RATES 83
Ti me [ s]
FIG. I - - ENM data f or 316L in 10% H2SO 4 at 40~
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
84 ELECTROCHEMICAL NOISE MEASUREMENT
TABLE 3- - 316L i n 10% HzS04 at 40~
Weight Loss of LPR
Area, cm 2 LPR, mm/y Electrodes, mm/y ENM, mm/y
Weight Loss of ENM
Electrodes, mm/y
1 0.4318 0.3048 0.3556 0.3302
3 0.4318 0.2540 0.0051 0.0051
5 0.0025 0.0000 0.0178 0.0025
15 0.1178 0.2540 0.0762 0.0279
samples exhibiting general corrosion rates below 1.3 mm/y. A common potential noise trait
was an overall trend toward a higher or lower potential with superimposed random fluctu-
ations. The current noise did not tend to form patterns during general corrosion. An exception
to this was found when the current noise had regular oscillations that were found to correlate
with a heater cycling on and off. This effect was avoided by using water baths when possible
or proportional heating controllers to minimize temperature cycling.
Although there was some variation from cell to cell in the 10 and 20% HzSO 4 tests, the
mass loss rate correlated well with the values calculated from the electrochemical technique
used (Tables 3 and 4), In the cases in which LPR- and ENM-calculated corrosion rates
differed, the mass loss data for those coupons confirmed the different electrochemically
determined rates.
The 30 and 40% H2SO 4 results are listed in Tables 5 and 6. Significant end grain attack
was observed on the electrodes used in the tests, so the change in diameter was used as a
means of calculating the corrosion rate while discounting any grain dropping that may have
been caused by the end grain attack. It was found that the difference between corrosion rates
from mass loss and diameter change was less than 5% for the all 30% H2SO4 tests and for
the 15-cm 2 electrodes in 40% H2SO4. In 40% H2SO 4, the diameter change determined a
lower corrosion rate by 30, 15, and 10% for the 1-, 3-, and 5-cm 2 samples, respectively. This
correlation between electrode size and the relative impact of the end grain attack was not
surprising. Because all electrodes were of the same diameter, the larger electrodes were
longer and had a smaller relative amount of end grain area. Because electrochemical tech-
niques do not measure metal lost by grain dropping, the rates predicted by ENM and LPR
would have been expected to show the same level of disparity as the diameter change mea-
surements. In practice, end grain attack was insignificant compared to the overall scatter in
the data.
Tri-lso-Propanolamine.Hydrochloric Acid (TIPA.HCI)
Above about 160~ TIPA.HC1 has a corrosivity similar to 35% HC1. Below 160~ HC1
is bound to TIPA, and TIPA.HCI is noncorrosive to carbon steel. Plant conditions indicated
TABLE 4--316L in 20% H2S04 at 40~
Weight Loss of LPR Weight Loss of ENM
Area, crn 2 LPR, mm/y Electrodes, mm/y ENM, mm/y Electrodes, mm/y
1 1.0668 1.1684 1.3970 0.9906
3 0.8636 0.9906 0.7874 0.7894
5 0.5080 0.6096 0.0610 0.2794
15 0.5537 0.5588 0.5458 0.4318
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
REICHERT ON DETERMINING CORROSI ON RATES
TABLE 5 - - 3 1 6 L i n 30% H2SO 4 at 40~
85
Weight Loss of LPR Weight Loss of ENM
Area, cm 2 LPR, mm/y Electrodes, mm/y ENM, mm/y Electrodes, mm/y
1 2.1 5.6 0.2 5.9
3 6.5 5.2 0.6 5.8
5 18.0 5.0 0.0 1.8 a
15 18.0 4.3 0.1 4.8
"Average of three electrodes with rates of: 0.0051, 0.0102, 5.3340 mm/y.
that there was a possibility of TIPA.HC1 being present under the elevated, corrosive condi-
tions. Plant inspections, however, showed no corrosion attack. Corrosion monitoring was
performed in the lab at 190~ to determine if the steel formed a protective layer of corrosion
products.
Mass loss coupons provided corrosion rates of 190 mm/ y for a 30-min exposure and 250
mm/ y for a 60-min exposure. A coupon left overnight (16 h) was consumed, indicating a
rate on the order of 250 mm/ y.
The ENM tests were run as nine consecutive 10-min scans. The initial data set showed a
corrosion rate of 15 mm/ y. After 20 min, the measured corrosion rate decreased to 5 to 8
mm/y. The ENM data followed a predictable decline in corrosion rate as a layer of loosely
adhered corrosion product grew on the electrode, slowing mass transport. If the ENM data
after the first 20 rain are considered representative of long-term corrosion behavior, then the
ENM corrosion rate calculated was about 30 times smaller than that determined by mass
loss.
Gamma-Butyrolactone (gBL) + Trace Contaminants
gBL was tested by ENM simply to evaluate the ability of ENM to measure corrosion rate
in a low-conductivity organic system. ENM data were collected on carbon steel and 316L
stainless steel at 130~ in gBL with unidentified trace impurities. The conductivity of the
gBL used was less than 10 ~mho/cm. Mass loss rates were 0.042 and 0.0038 mm/ y, whereas
noise resistance calculated rates were 0.043 and 0.0023 mm/ y for carbon steel and 316L,
respectively. The electrodes were threaded onto glass-coated stainless steel rods. The rods
were held in a plastic spacer so the electrodes were spaced approximately 0.03 cm apart to
minimize the effects of solution resistance.
D i s c u s s i o n
Potential noise data for the ENM tests performed ranged from 10 -5 to 10 z V, whereas
the current noise data ranged from 10 -~~ to 1 0 - 4 m. The baseline potential and current noise
TABLE 6--316L in 40% H2 S O 4 a t 40~
Weight Loss of LPR Weight Loss of ENM
Area, cm 2 LPR, mm/y Electrodes, mm/y ENM, mm/y Electrodes, mm/y
1 8.9 32 1.2 31
3 15 21 1.3 21
5 12 19 0.51 18
15 67 16 0.051 9.1
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
86 ELECTROCHEMICAL NOISE MEASUREMENT
levels from t he moni t ori ng equi pment were bot h t wo orders of magni t ude smal l er than the
noi se generat ed by the corrosi on processes studied. Therefore, t he i nst rument at i on was suit-
abl e for col l ect i ng EN data. The basel i ne noi se dat a were col l ect ed by l eavi ng the pot ent i al
l eads as an open ci rcui t and short ci rcui t i ng the current leads.
The corrosi on rates measured by R, fluctuated over t he durat i on of the i mmer si on t i me
for all t e s t s - - whe t he r the correl at i on to R e or mass l oss was good or poor. The magni t ude
of t hese vari at i ons t ended to mat ch t he magni t ude of the corrosi on rate. For exampl e, i f the
average corrosi on rate was 0.1 mm/ y, then i ndi vi dual ENM dat a sets woul d measure cor-
rosi on rat es from 0.01 to 0.2 mm/ y. Despi t e t hese vari at i ons, trends were still e vi de nt - - s uc h
as an i ncrease in the corrosi on rate measur ed i f the t emperat ure i ncreased. The i ncreased
cor r osi on rate was observed as an i ncrease in the mean val ue of the current noise. An i ncrease
in sol ut i on corrosi vi t y duri ng a scan (such as by addi ng an aci d) resul t ed in an i mmedi at e
j ump in the current noise. The magni t ude of the i ncrease in current noi se refl ect ed the degree
of change in the aggressi veness of the solution.
Variations in corrosi on rates were al so obser ved in the LPR tests, al t hough the magni t ude
of the vari at i ons was about hal f as large. Because general corrosi on is a st ochast i c process,
i t fol l ows that the corrosi on rat e wi l l vary wi t h t i me and that the average of t hose vari ed
rat es wi l l correl at e wi t h rates obt ai ned by a technique, such as mass l oss, that measures an
average corrosi on rate.
The sulfuric aci d tests were desi gned to l ook at the effect of el ect r ode area in addi t i on to
the val i di t y of usi ng R, to cal cul at e corrosi on rat es from ENM. For the 10 and 20% H2SO 4
tests, corrosi on rates det er mi ned by the mass loss met hod and by the R, t echni que mat ched
regardl ess of el ect r ode size. Thi s meant that the magni t ude of the current noi se correl at ed
wi t h el ect r ode area for t he 1- to 15-cm 2 areas tested.
In t he cases in whi ch cor r osi on rates were high, the ENM dat a di d not correl at e wel l wi t h
mass loss data. In the 30 and 40% H2SO 4 experi ment s, the st andard devi at i on of t he pot ent i al
noi se was very hi gh and t ended to domi nat e (crV > > ~/ ) t he R, value. The ~V val ue for
t hese tests gener al l y r anged bet ween 1 and 10 mV. Much of the vari at i on was due to sub-
stantial dri ft i ng of the corrosi on pot ent i al (Fi g. 2). Some of this dri ft may have been due to
the hi gh corrosi on rates obser ved in many tests. Anot her source for t he hi gh pot ent i al noi se
was the l arge amount of hydr ogen bubbl i ng occurri ng on the surface, Mansf el d has report ed
that fluctuations in mass t ransport cont rol can pr oduce l arge changes in the corrosi on pot en-
t i al [14].
A t hi rd aspect of the poor correl at i on may be sat urat i on of the noi se data. The TIPA.HC1
t est al so showed poor correl at i on bet ween R, and mass l oss corrosi on rates. In this case, crV
was much smal l er than in t he 30 and 40% H2SO 4 experi ment s and di d not domi nat e the
rat i o R, = ~rV/~l. The ENM pat t ern for the TIPA.HC1 tests (Fig. 3) appears very si mi l ar to
that shown in Fi g. 1 for t he 10% HzSO4 tests, yet the magni t ude of the current noi se does
not reflect the hi gh cor r osi on rate.
Based on current noi se val ues from t he 10 and 20% HzSO 4 tests, current noi se roughl y
cor r el at ed wi t h corrosi on rate. The current noi se ranges for vari ous corrosi on rat es are l i st ed
in Tabl e 7. Ignori ng the apparent t rend that t he i ncrease in current noi se is faster than the
i ncrease in corrosi on rate, a l i near ext rapol at i on of Table 7 woul d ant i ci pat e a current noi se
on the or der of 10 -2 A/ c m z for the cor r osi on rat e of 250 mm/ y obser ved in TIPA. HCI. Such
a l arge current noi se magni t ude woul d requi re a subst ant i al segregat i on of the anodi c and
cat hodi c sites bet ween t he current measuri ng el ect rodes. As t he corrosi on rat e i ncreases and
segregat i on becomes mor e difficult to achi eve, t he current noi se wi l l saturate, resul t i ng in
poor correl at i on bet ween ENM and mass l oss data.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
9.00E-03
B
o
Pot ent i al No i s e f o r 3 1 6 L i n 3 0 % Sul f ur i c Ac i d at 4 0 ~
(J
8.50E-03
8.00E-03
7.50E-03
7,00E-03
6,50E-03
6.00E-03
5.50E~
5.00Eo03
4.50E-03
4.00E-03
REICHERT ON DETERMINING CORROSION RATES 87
I I r I I
100 200 300 400 500 600
I Ti me Is]
Cu r r e n t No i s e f o r 3 1 6 L in 3 0 % Sul f ur i c Ac i d a t 4 0 ~
4.00E-06
I I I I I
0 100 200 300 400 500 600
Ti me [s]
3.50E-06
3,00E-06
2,50E-06
2.00E-06
1.50E-06
1,00E-06
FIG. 2- - ENM data f or 316L in 30% H2SO 4 at 40~C.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
-0.0006
= :
~ o
a.
Pot ent i al Noi se f or Car bon Steel i n TI PA. HCl
i i b i i
100 200 300 400 500 600
Time Is]
-0.0008
-0.001
-0.0012
-0.0014
-0.0016
-0.0018
88 ELECTROCHEMI CAL NOI SE MEASUREMENT
Cur r ent Noi se f or Car bon Steel i n TI PA. HCl
O.OOE+O0
c
-2.00E-05 -
-4.00E-05 -
-6.00E-05
-8.00E-05
-1.00E-04
-1.20E-04
-1.40E-04 i i b i i
1 O0 200 300 400 500 600
Time [s]
FIG. 3- - ENM data f or carbon steel in TIPA.HCt at 190~
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
REICHERT ON DETERMINING CORROSION RATES
TABLE 7--Corrosion rate approximations f or current
noise ranges.
Current Noise Range, A/ cm 2 Corrosion Rate, mm/ y
10- 9~ 10 8 0.003
10 -8 ~ 10 -7 0.03
10 -6 ~ 10 -5 0.3
1 0 - 5 ~ 10 4 1,3
89
Conclusions
The R, anal ysi s met hod pr oved hi ghl y effect i ve for det er mi ni ng corrosi on rates for syst ems
exhi bi t i ng general cor r osi on rat es bel ow 1.3 mm/ y. Agr eement bet ween R, , Rp and mass
loss values was very good for bot h carbon steel and st ai nl ess steel in di fferent envi ronment s.
Correl at i ons bet ween R, and mass loss measurement s broke down at hi gh corrosi on rates.
Hi gh pot ent i al noi se coupl ed with l ower - t han- pr edi ct ed current noi se resul t ed in l ow cal cu-
l at ed corrosi on rates usi ng R, . Hydr ogen bubbl i ng cont ri but ed to the pot ent i al noise, t hereby
i ncreasi ng the cal cul at ed val ue of R, . Anot her pr oposed source for l ow corrosi on rat es
cal cul at ed by R, was sat urat i on of t he current noi se at hi gh corrosi on rates. Ext r apol at i ng
t he current noi se ranges obser ved for l ow to moder at e corrosi on rat es to hi gh rat es i mpl i es
a very hi gh coupl i ng c ur r e nt - - on the order of tens of mA/ c m2 - - f o r hi gh corrosi on rates.
The l arge segregat i on of the anodi c and cat hodi c sites requi red to generat e such currents
woul d be difficult to achieve, and current noi se of this magni t ude was not obser ved even for
corrosi on rat es of 250 mm/ y.
References
[1] Iverson, W. P., Journal of the Electrochemical Society, Vol. 115, No. 6, 1968, pp. 617-618.
[2] Hladky, K. and Dawson, J. L., Corrosion Science, Vol. 22, No. 3, 1982, pp. 231-237.
[3] Bertocci, U., Journal of the Electrochemical Society, Vol. 127, No. 9, 1980, pp. 1931-1934.
[4] Gabrielli, C., Huet, E, Keddam, M., and Oltra, R., Corrosion, Vot. 46, No. 4, 1990, pp. 266-278.
[5] Mansfeld, E, Research in progress, CORROSION/94, NACE, Baltimore, MD, 1994.
[6] Roberge, E R., Beaudoin, R., and Sastri, V. S., Corrosion Science, Vol. 29, No. 10, 1989,
pp. 1231-1233.
[7] Scully, J. R. and Hudson, J. L., Oral presentation, T-3L-15, ENM Task Group, CORROSION/93,
NACE, New Orleans, LA, 1993.
[8] Hladky, K. and Dawson, J. L., Corrosion Science, Vol. 21, No. 4, 1981, pp. 317-322.
[9] Eden, D. A., Hladky, K., and John, D. G., NACE CORROSION/86 Conference, NACE, Houston,
TX, 1986, Paper 274.
[10] Dawson, J. L., Farrell, D. M., Aylott, P. J., and Hladky, K., NACE CORROSION/89 Conference,
NACE, New Orleans, LA, 1989, Paper 31.
[11] Rothwell, A. N. and Eden, D. A., NACE CORROSION/92 Conference, NACE, Nashville, TN,
1992, Paper 223.
[12] CML Technology Transfer, "Mathematical Analysis of Electrochemical Noise Signals," Capcis
March Limited, Manchester, UK, circa 1993 (not dated).
[13] Davis, J. R., Ed. Metals Handbook, 9th ed., Vol. 13 Corrosion, ASM International, Metals Park,
OH, 1987, p. 1150.
[14] Mansfeld, E and Xioa, H., Journal of the Electrochemical Society, Vol. 140, No. 8, 1993,
pp. 2205-2209.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
Data Analysis
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
R. A. Cottis, l S. T u r g o o s e , 1 and J. Me ndoz a- Fl or e s ~
The Effects of Solution Resistance on
Electrochemical Noise Resistance
Measurements: A Theoretical Analysis
RE F E RE NCE : Cottis, R. A., Turgoose, S., and Mendoza-Flores, J., " Th e Ef f e c t s o f So l ut i o n
Re s i s t a nc e o n El e c t r o c h e mi c a l No i s e Re s i s t a n c e Me a s u r e me n t s : A The o r e t i c a l Ana l y s i s , "
Electrochemical Noise Measurement f or Corrosion Applications, ASTM STP 1277, Jeffery R.
Keams, John R. Scully, Pierre R. Roberge, David L. Reichert, and John L. Dawson, Eds.,
American Society for Testing and Materials, 1996, pp. 93-100.
ABSTRACT: The theoretical basis of electrochemical noise resistance measurements in the
presence of significant solution resistance is examined with a simple linear circuit model. A
shot noise model of the noise generation process is assumed to develop the dependence of
electrochemical potential and current noise on corrosion rate, although the conclusions in re-
spect to electrochemical noise resistance do not depend on this. It is concluded that the elec-
trochemical noise resistance method measures essentially the same resistance as is measured
by a conventional linear polarization resistance measurement, although it is found to be capable
of making measurements in higher resistance systems.
KEYWORDS: electrochemical noise, noise resistance, polarization resistance, corrosion rate
measurement
It has been shown by several workers (see, for example, the other papers in this volume)
that electrochemical noise resistance measurements compare well with other electrochemical
techniques, such as linear polarization resistance and electrochemical impedance spectros-
copy, for the estimation of corrosion rate (although not all investigations have given a good
correlation). However, recent work has shown that noise resistance measurements are subject
to errors as a result of the resistance of the solution between the measuring electrodes. This
paper is concerned with the analysis of the effects of solution resistance, although this com-
mences with the development of a simple theoretical basis for noise resistance measurements.
Note that more detailed mathematical analyses of a range of possible electrochemical noise
generation processes have been performed, and Gabrielli provides a good summary of these
[11.
A s s u m p t i o n s
For the purpose of our analysis, several simplifying assumptions are made. The physical
system considered is the conventional three-electrode noise resistance measurement with the
electrochemical current noise being measured between two electrodes, and the electrochem-
ical potential noise of the pair of electrodes being measured with respect to a third electrode,
Senior lecturer, lecturer, and research student, respectively, Corrosion and Protection Centre, UMIST,
PO Box 88, Manchester M60 IQD, UK.
Copyrighl 9 1996 by ASTM lntcrnational
9 3
www.astm.org
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
94 ELECTROCHEMICAL NOISE MEASUREMENT
whi ch may ei t her be a l ow-noi se reference el ect r ode or a t hi rd el ect rode of the same mat eri al
as the ot her two. For the purposes of this paper, t he reference el ect r ode is assumed to be
noi se-free, but it is a si mpl e mat t er to add the noi se generat ed at a " noi s y" reference el ec-
trode. Bot h t he anodi c and cat hodi c react i ons are consi der ed to be processes that generat e
current. Hence, current noi se is a fundament al out put of the corrosi on react i on, whereas
pot ent i al noi se is an i ndi rect output that results from the act i on of t he current noi se on the
component s of t he el ect r ochemi cal syst em.
Bot h anodi c and cat hodi c react i ons are assumed to result in pul ses of charge of rel at i vel y
short durat i on. For the cat hodi c react i on of oxygen reduct i on, it is expect ed that the current
pul ses wi l l be smal l , pr obabl y consi st i ng of j ust the four el ect rons pr oduced by t he reduct i on
of one oxygen mol ecul e (mass t ransport fluctuations may pr ovi de a mor e si gni fi cant source
of noi se in pract i ce, but t hese are not consi der ed here). This wi l l pr obabl y al so appl y to
hydr ogen evol ut i on at l ow current densi t i es, such that the hydrogen gas mol ecul es diffuse
into the solution, but at hi gher current densi t i es the maj or source of noi se wi l l be the fluc-
t uat i on of exposed el ect r ode area and mass t ransport processes as a result of hydr ogen bubbl e
evol ut i on, whi ch wi l l cl ear l y not fit our assumpt i on. For the anodi c process, i t has been
argued [2] that act i vat i on-cont rol l ed corrosi on wi l l generat e bursts of charge of t he or der of
10 3 el ect rons, whereas it is wel l est abl i shed that noi se generat ed by met ast abl e pi t t i ng wi l l
consi st of pul ses of charge wi t h durat i ons of a few seconds and wi t h a charge of t he order
of 1 p.C. Ot her forms of noi se generat i on process for the anodi c react i on may not produce
pul ses of charge, wi t h chaot i c [3] and peri odi c [4] behavi or havi ng been report ed. It is
assumed that the pul ses of charge pr oduced by t he anodi c and cat hodi c react i ons are i nde-
pendent. Thi s seems a r easonabl e assumpt i on, al t hough it is cl ear that t he average DC cur-
rents must be equal to mai nt ai n charge neut ral i t y (short -t erm di fferences can be accommo-
dat ed by chargi ng and di schargi ng of the doubl e- l ayer capaci t ance).
To use st andard st at i st i cal results for t he anal ysi s of the noi se charact eri st i cs, it is assumed
that the charge pul ses are i ndependent of one anot her and that t he anodi c and cat hodi c
react i ons are uncorrel at ed and can t herefore be t reat ed as i ndependent sources of current
noise. At first sight, the l at t er assumpt i on is invalid, at l east at l ow frequenci es, because the
average anodi c and cat hodi c current s must be the same, ot herwi se the speci men pot ent i al
wi l l change as a resul t of the accumul at i on of charge. However, the l i nk bet ween the t wo
react i ons is bot h compl ex and indirect. The det ai l s of this l i nkage are not yet compl et el y
clear, and at present, t he assumpt i on of i ndependence is t he onl y vi abl e approach.
The anal ysi s present ed here wi l l be rest ri ct ed to the l ow-frequency l i mi t of the noi se
behavior, so that current pul ses may be t reat ed as del t a functions (that is, i nst ant aneous pul ses
of charge), al t hough it is rel at i vel y st rai ght forward to ext end the anal ysi s to the full frequency
spectrum.
Current Noise
Gi ven the assumpt i ons made, it is cl ear that each el ect rochemi cal react i on can be t reat ed
wi t h a shot noi se anal ysi s. Essent i al l y, this assumes that the noi se arises as a resul t of the
r andom vari at i on in the number of pul ses ari si ng in a gi ven t i me interval. Through a si mpl e
st at i st i cal anal ysi s [5], this resul t s in a current noi se power gi ven by
= 2qI b (1)
where
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
CO'I-I'IS ET AL. ON EFFECTS OF SOLUTION RESISTANCE ON ENM 95
m
I , ~ = mean squared noi se current or current noi se power,
q = charge in each pulse,
I = mean current, and
b = bandwi dt h of measurement .
Not e that, as is common in si gnal processi ng, the t er m p o w e r is used t o descri be bot h
currenF and voltage, 2 on t he basi s that the power di ssi pat ed in a resi st i ve l oad is gi ven by
V 2 R or 12R.
I f we consi der t he react i ons occurri ng at the met al - s ol ut i on i nt erface, we can see t hat
t here are act ual l y t wo currents fl owi ng on each el ect r ode as a resul t of t he anodi c and
cat hodi c react i ons. These are assumed to be i ndependent and may t herefore be anal yzed
separat el y, and the noi se powers are summed to obt ai n t he t ot al noi se power. Thi s l eads t o
a total current noi se power gi ven by
1,2 = 2(qa + qc)Icor~b (2)
where
qa = charge in anodi c react i on pul se,
q,. = charge in cat hodi c react i on pulse,
Icor~ = cor r osi on current on t he el ect rode, and
b -- bandwi dt h of measurement .
I f t wo such el ect rodes are coupl ed t oget her t hrough a ci rcui t of zero resi st ance, t he noi se
current power det ect ed bet ween the t wo el ect rodes wi l l be t he same as t hat of t he i ndi vi dual
el ect rodes (because hal f of the power wi l l be di ssi pat ed on t he same el ect r ode as t he current
source and, hence, wi l l not be det ect ed). However, when a sol ut i on resi st ance is i nt erposed
bet ween the t wo el ect rodes, this reduces t he noi se current; effectively, t he sol ut i on resi st ance
forces a great er part of t he noi se current to r emai n l ocal i zed on t he i ndi vi dual el ect rodes. To
anal yze this situation, we need to devel op a si mpl e met hod of t reat i ng the pot ent i al noi se
pr oduced by the passage of the noi se current.
Electrochemical Potential Noise
I f we consi der a noi se generat i on process such as act i vat i on-cont rol l ed corrosi on, i t can
be seen that an exact descr i pt i on of t he pot ent i al noi se generat i on process wi l l consi der t hat
t he current noi se consi st s of current pul ses that have a pot ent i al dependent rat e of generat i on,
and that the pot ent i al noi se arises by t he chargi ng and di schargi ng of t he i nt erfaci al capac-
i t ance by the anodi c and cat hodi c current pulses. Thi s syst em is bei ng model ed numer i cal l y
by the authors but is rat her compl ex for use in a si mpl e anal ysi s. However, we can obt ai n
a si mpl er met hod of t reat i ng t he pot ent i al dependence by not i ng that the pol ari zat i on resi s-
t ance (Rp) wi l l be pr oduced by t he act i on of ext ernal l y i mposed pot ent i al changes on exact l y
the same pul se generat i on processes. Thus, it seems r easonabl e t o model t he pot ent i al noi se
generat i on process as t he response of t he pol ar i zat i on resi st ance t o the noi se current. The
exact val i di t y of this approach r emai ns to be tested, because it may be not ed t hat t he charge
pul ses and t he apparent pol ari zat i on resi st ance that results are pr oduced by t he same process.
Hence, t he i mpl i ed assumpt i on of st at i st i cal i ndependence is not st ri ct l y correct. Conversel y,
when t wo pol ar i zat i on resi st ances are in paral l el , as when measur i ng el ect r ochemi cal current
noise, it is not cl ear t hat the behavi or wi l l be model ed correct l y by the val ue cal cul at ed f r om
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
96 ELECTROCHEMICAL NOISE MEASUREMENT
conventional circuit theory because the currents through the two resistances will be uncor-
related. For the purposes of this paper and for consistency, we have assumed that the use of
Rp is valid; any errors resulting from this assumption are likely to be a factor of ~ or less
(because this is the difference between adding correlated and uncorrelated noise currents),
and errors in potential and current noise may cancel in calculating noise resistance. Al so
note that we are working in terms of currents and polarization resistances for the electrode
area, not with normalized current densities and polarization resistances.
I f we adopt this simplified approach, the electrochemical potential noise power is obtained
by the application of Ohm' s law
u
Not e that this relationship should not be taken as a theoretical justification of the noise
resistance method because, to a certain extent, we are assuming the validity of the method.
Rather, the experimental applicability of the noise resistance method lends support to this
model. Not e also that the potential and current noise power terms in Eq 3 are generalized,
rather than the specific values for this analysis (that is, a potential noise power of E~, will
be produced i f a current noise of ~ is applied to the polarization resistance).
Because of the relationship between Rp and Icorr that is implied by the Stern-Geary
equation,
B
Ic~ : pR-- (4)
where B = the Stern-Geary coefficient, we can derive two useful relationships:
B 2 - 2BZ(qa + q c ) b
VE-] I ] (6)
( q " + q~) = 2 B b
Equation 5 suggests that potential noise power will be i n v e r s e l y proportional to corrosion
rate. This result is inconsistent with the results of early work that suggested that potential
was proportional to corrosion rate [6]. This apparent contradiction may possibly be ration-
alized by consideration of changes in the value of qa or qc- Equation 6 provides a met hod
for determining the value of (q, + qc), which may provide useful information on localized
corrosion processes. These questions will be treated in more detail elsewhere.
Effect of Solution Resi stance
To model the influence of solution resistance, we shall consider an equivalent circuit model
of the process (Fig. 1). Not e that both the noise current sources and the polarization resis-
tances ( R p ) associated with the working electrodes in Fig. 1 are at the met al -sol ut i on
interface.
This model considers that the reference electrode is equidistant from the two working
electrodes, such that it samples the potential halfway between the two electrodes. We also
assume that a noise-free reference electrode is used; i f a reference electrode of the same
material as the working electrode is used, this will add to the potential noise power.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
COI-IIS ET AL. ON EFFECTS OF SOLUTION RESISTANCE ON ENM 97
I R e f e r e n c e ]
E le c tr o d e I
S o l u t i o n
I . . . . . . . .
Current ~ ~ C u r r e n t saole tu tailo'n
T S ~ i n t e r f a c e
s~ Y z.,~, zm . . . . . . . .
J W o r k i n g
E le c t r o d e I ] J W o r k i n g
E le c t r o d e 2 ] M e t a l
I Q i
FIG. 1--Equivalent circuit for three-electrode noise resistance measurement.
In the absence of solution resistance, the currents resulting from the t wo noise current
sources pass through the polarization resistance of the two electrodes to give a potential
noise as a result of each current of
m
I 2 R~ (7)
~ n n ( d u e t o o n e c u r r e n t ) - - 4
The potential noise powers caused by the two independent noise sources add to give a
total potential noise
~ . = ~ R~ ( 8 )
2
Not e that, from Eq 2, ft, is proportional to Icon, and hence to specimen area (because Icor~ is
the corrosion current density times the specimen area). When combi ned with Eq 8, and
remembering that Rp is inversely proportional to specimen area, this leads to the somewhat
counterintuitive relationship between potential noise and area,
E - ~ 1
Area (9)
I f the solution resistance is non-zero, we may apply a simple circuit analysis to determine
the effect of the noise current on the potential detected at the midpoint of the solution
resistance and the corresponding current through the solution resistance.
Considering one current source, the current will flow through the polarization resistance
associated with the same working electrode as the current source, Rp, in parallel with the
solution resistance, Rs, plus the polarization resistance of the other working electrode.
The noise potential across this resistor network can be calculated, and the potential noise
at the reference electrode can be determined by considering the potential divider consisting
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
98 ELECTROCHEMICAL NOISE MEASUREMENT
of the t wo part s of the sol ut i on resi st ance pl us the r emot e pol ari zat i on resi st ance. The po-
t ent i al noi se power as a resul t of the t wo i ndependent current sources is then t wi ce that
resul t i ng from the one current source, gi vi ng the same resul t as Eq 8. Thus, t he measured
pot ent i al noi se is unaffect ed by the sol ut i on resi st ance.
Electrochemical Current Noise
In the absence of sol ut i on resi st ance, t he measured noi se current caused by each source
wi l l be hal f the total current from the source (because the ot her hal f wi l l flow t hrough t he
pol ari zat i on resi st ance on t he same el ect r ode and wi l l not t herefore be det ect ed; as t he
current s are correl at ed, because t hey come from the same current noi se source, the current s
wi l l be hal ved, not the current powers). The current noi se powers as a resul t of t he t wo
i ndependent sources wi l l add, gi vi ng a measur ed noi se current power
. . . . red - - T ( 10)
Not e that this anal ysi s is correct f o r the equivalent circuit, in whi ch t he separat i on of the
noi se current s bet ween the t wo el ect rodes results in the di vi si on of the current power by 4
(22 ) because the t wo current s are not i ndependent (bei ng due to t he same vol t age appl i ed
across t wo resistors). However, in t he real el ect rochemi cal interface, it seems pr obabl e that
the t wo current s wi l l be i ndependent , in whi ch case t he separat i on of currents wi l l resul t in
di vi si on of the current power s by onl y 2. Hence, the measured noi se current wi l l be equal
to t he noi se current on one el ect rode. Thi s quest i on wi l l be consi der ed further el sewhere,
and for the present anal ysi s i t is assumed t hat the equi val ent ci rcui t resul t is val i d (because
the result for the sol ut i on resi st ance case necessar i l y makes this assumpt i on).
When the sol ut i on resi st ance is non-zero, the di st ri but i on of current bet ween t he t wo
el ect rodes is modi fi ed. Consi der i ng the current source on one el ect rode, i t can be seen that
the proport i on of current flowing to the ot her el ect r ode (and hence bei ng measured) is Re/
(Rs + 2Rp). Therefore, t he measur ed current power wi l l be gi ven by
h
( 2 , 2
12 . . . . . d = ( R s + 2 e p ) ]
(11)
Compar i son of Eqs 11 and 12 shows that the measured current wi l l be in error by a fact or
2Rf f ( R, + 2Rp), as mi ght perhaps be expect ed intuitively, and as woul d al so be obt ai ned for
a pol ari zat i on resi st ance measurement . Hence, the measured resi st ance wi l l si mpl y be in-
creased by t he sol ut i on resi st ance, and, i f the sol ut i on resi st ance is known, t he resul t can, in
pri nci pl e, be correct ed (al t hough errors are l i kel y to l i mi t the possi bi l i t y for correct i on i f the
sol ut i on resi st ance is si gni fi cant l y great er t han the pol ari zat i on resi st ance).
Experimental Results
To test the effects of sol ut i on resi st ance on noi se resi st ance measurement s and compar e
the results of el ect rochemi cal and wei ght l oss measurement s, a series of exper i ment s have
been per f or med wi t h mi l d steel cor r odi ng in sodi um sulfate (Na2SO4) sol ut i ons wi t h a range
of concent rat i ons from 0.001 to 1% and at a t emperat ure of 60~ Ful l det ai l s of t hese
exper i ment s are present ed el sewher e [7], but t he results are summar i zed here in Fi g. 2. The
corrosi on rate, as det er mi ned wi t h wei ght l oss measurement s over a per i od of 15 days, was
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
CO'I-i-IS ET AL. ON EFFECTS OF SOLUTION RESISTANCE ON ENM 99
FIG. 2--Comparison of corrosion rate measurements (without compensation f or solution resistance).
essential l y independent of Na2SO4 concentration. The corrosion rate is compatibl e with ox-
ygen mass-transport control . Hence, the Na2SO4 is thought simpl y to provide sufficient so-
l ution conductivity to coupl e the anodic and cathodic reactions. With high N~2SO 4 concen-
t rat i ons, such that the sol ution resistance is much smal l er than the pol arization resistance,
all three el ectrochemical techniques give reasonabl y good resul ts in comparison with weight
l oss measurements. ( Note that the weight l oss measurements are necessaril y made over a
reasonabl y l ong period, whereas the el ectrochemical measurements reported here were made
af ter 3-h exposure, thus it is not expected that a perf ect correl ation wil l be obtained.) I n
more dil ute sol utions, the el ectrochemical techniques become more difficult. El ectrochemical
impedance al l ows the separate measurement of Rp and R,, but when Rs becomes l arge rel ative
to Rp, noise renders the discrimination of R e impossibl e. Linear pol arization resistance and
noise resistance measurements appear to measure the same thing, but the noise resistance
method is abl e to make measurements in more dil ute sol utions ( al though it is, of course,
measuring R, + Rp rather than Rp al one). I f the sol ution resistance is known, both pol arization
resistance and noise resistance measurements can be corrected to determine Rp, al though the
error increases rapidl y as Rs becomes l arge comp ared to Rp. This is il l ustrated in Fig. 3.
Concl usi ons
For corrosion processes that resul t in bursts of charge, the noise current p ower is expected
to be proportional to the corrosion rate, whereas the potential noise p ower is expected to be
inversel y proportional to corrosion rate. Both current and potential noise powers are expected
to be proportional to the quantity of charge in each pul se.
I n l ow-resistivity sol utions, noise resistance measurements appear to be comp arabl e to
el ectrochemical impedance spectroscopy and l inear pol arization resistance measurements in
their abil ity to estimate corrosion rate.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
100 ELECTROCHEMICAL NOISE MEASUREMENT
FIG. 3--Comparison of corrosion rate measurements (with compensation f or solution resistance).
Linear pol arization resistance and noise resistance measurements both measure the sol ution
resistance in addition to the pol arization resistance.
The noise resistance technique can make measurements in sol utions of much higher re-
sistivity than either pol arization resistance or impedance (al though the resul ts are, of course,
l argel y measuring sol ution resistance).
Ref erences
[1] Gabrielli, C., Huet, E, and Keddam, M., "Investigation of Metallic Corrosion by Electrochemical
Noise Techniques," in Electrochemical and Optical Techniques f or the Study and Monitoring of
Metallic Corrosion, M. G. S. Ferreira and C. A. Melendres, Eds., NATO AS1203, Kluwer, Dordrecht,
1991, pp. 135-190.
[2] Cottis, R. A., "Modelling of Electrochemical Noise Due to the Activation-Controlled Dissolution of
Metals," in Modelling Aqueous Corrosion, K. R. Threthewey and P. R. Roberge, Eds., NATO ASI
Series E, Vol. 266, Kluwer, Dordrecht, 1994, pp. 369-379.
[3] Li, W., Nobe, K., and Pearlstein, A. J., "Electrodissolution Kinetics of Iron in Chloride Solutions
VIII: Chaos in Potential/Current Oscillations," Journal of the Electrochemical Society, Vol. 140,
1993, pp. 721-728.
[4] Russell, P. and Newman, P., "Current Oscillations Observed Within the Limiting Current Plateau
for Iron in Sulfuric Acid," Journal of the Electrochemical Society, Vol. 133, 1986, pp. 2093-2097.
[5] Schottky, W., Annals of Physics, Vol. 57, 1918, p. 541. (See King, R. A., ElectricalNoise, Chapman
and Hall, London, 1966, for a more recent description.)
[6] Iverson, W. P., "Transient Voltage Changes Produced in Corroding Metals and Alloys," Journal of
the Electrochemical Society, Vol. 115, 1968.
[7] Mendoza-Flores, J., "Electrochemical Techniques Applied to the Study of Mild Steel Corrosion in
1%, 0.1%, 0.01% and 0.001% Na2SO4 Solutions," MSc dissertation, UMIST, 1993.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
Rober t G. Kel l y, 1 M ar i a E. I nman, 1 and J ohn L . H udson 1
Analysis of Electrochemical Noise for Type
410 Stainless Steel in Chloride Solutions
REFERENCE: Kelly, R. G., Inman, M. E., and Hudson, J. L., "Analysis of Electrochemical
Noise for qPype 410 Stainless Steel in Chloride Solutions," Electrochemical Noise Measure-
mentfor Corrosion Applications, ASTM STP 1277, Jeffery R. Kearns, John R. Scully, Pierre
R. Roberge, David L. Reichert, and John L. Dawson, Eds., American Society for Testing and
Materials, 1996, pp. 101-113.
ABSTRACT: The electrochemical noise resulting from the corrosion of Type 410 stainless
steel under open circuit conditions in solutions of widely different corrosivity has been ex-
amined. Parameters derived from electrochemical noise measurements and conventional elec-
trochemical measurements were correlated with one another and with post-test examination of
the specimens. Both the noise resistance and the polarization resistance increased with de-
creases in solution corrosivity. In all of the solutions studied, the noise resistance was found
to be consistently higher than the polarization resistance. In solutions in which pitting occurred,
the open circuit potential of the electrode versus a true reference electrode was more sensitive
to the transition to stable pitting than was the pitting index.
KEYWORDS: stainless steel, pitting, uniform corrosion, polarization resistance, electrochem-
ical noise
The corrosion of 13Cr stainless steels is a concern in the oil and gas industry because of
the use of this material for downhole tubulars. Although chemical inhibition is used to
combat degradation, its effectiveness is difficult to monitor. The emergence of electrochem-
ical noise measurement (ENM) as a monitoring technique is of particular interest because
of its noninvasive character. This paper describes the preliminary investigation of a variety
of analysis methods to determine what correlations exist among electrochemical noise mea-
surements, conventional electrochemical measurements, and the nature and extent of the
observed corrosion attack of Type 410 stainless steel (SS410) in room temperature solutions.
Electrochemical noise measurement has been investigated as a monitoring tool for the last
20 years [1]. One way to categorize the types of electrochemical noise is by the electro-
chemical signal that is collected. Potential noise refers to measurements of the open circuit
potential of an electrode versus either a reference electrode or versus a nominally identical
electrode. Although measurements with a conventional reference electrode have the advan-
tage of being relatable to thermodynamic conditions, these reference electrodes have their
own noise associated with them, which could complicate analysis. In addition, the application
of ENM to field conditions is much more attractive if maintenance of the reference electrode
is not required. Because the fluctuations of the signal are typically considered to be the most
important aspect of electrochemical potential noise (EPN), rather than the absolute value of
Assistant professor, postdoctoral research associate, Department of Materials Science & Engineering,
and professor, Department of Chemical Engineering, respectively, University of Virginia, Thornton Hall,
Charlottesville, VA 22903.
Copyrighl 9 1996 by ASTM lntcrnational
101
www.astm.org
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
102 ELECTROCHEMICAL NOISE MEASUREMENT
the pot ent i al , the use of an el ect r ode nomi nal l y i dent i cal to the wor ki ng el ect rode has become
popular. Al t hough the average pot ent i al di fference bet ween t rul y i dent i cal el ect rodes woul d
be zero, under condi t i ons for whi ch l ocal i zed corrosi on is possi bl e, large DC di fferences can
devel op. For exampl e, i f pi t t i ng were to i ni t i at e on onl y one el ect rode, its pot ent i al wi l l fall
wi t h respect to the other. Electrochemical current noise (ECN) is oft en measured as the
fluctuation in the current that passes bet ween t wo nomi nal l y i dent i cal el ect rodes that are
exposed to the same envi ronment and connect ed t hrough a zero resi st ance ammet er (ZRA).
The fluctuations in this coupl i ng current are due to each el ect rode becomi ng mor e or less
anodi c with respect to the ot her as a funct i on of time. The initiation, short -t erm pr opagat i on
and repassi vat i on of a pi t on one el ect r ode whi l e the ot her remai ns passi ve is an exampl e
of a physi cal ori gi n of such a phenomenon. Al t hough the el ect rodes are nomi nal l y i dent i cal ,
mi nor di fferences in el ect rochemi cal behavi or do exi st (or somet i mes devel op), whi ch l ead
to di fferences in pot ent i al and t herefore the coupl i ng current. These t ransi ent di fferences are
the basi s for the ENM. As with the pot ent i al , duri ng l ocal i zed corrosi on, subst ant i al DC
currents can pass bet ween el ect rodes short ed through a ZRA, i f onl y one i ni t i at es a pit.
Most field appl i cat i ons of ENM have i nvol ved a t hree-el ect rode pr obe system. Two nom-
i nal l y i dent i cal el ect rodes are connect ed vi a a ZRA, and the ECN is measured. The pot ent i al
of this coupl e is then measured agai nst the t hi rd nomi nal l y i dent i cal el ect r ode to moni t or
the EPN associ at ed wi t h t he same processes. This approach has been found useful in that it
al l ows an addi t i onal correl at i on (bet ween the t wo t ypes of noi se measurement s) to be made.
Successful correl at i ons bet ween noi se measurement s and corrosi on processes have been
demonst r at ed for uni form corrosi on as wel l as pi t t i ng in i ndi vi dual syst ems. Dawson et al.
[2] found that the rat i o of the st andard devi at i ons of the pot ent i al and current noi se (whi ch
t hey t er med the noi se resi st ance, R, ) correl at ed wel l with bot h the pol ari zat i on resi st ance
det er mi ned by convent i onal met hods and wei ght l oss of carbon steel in acid. By usi ng
di fferent sol ut i on composi t i ons, di fferent t ypes of at t ack were created, and t hey found qual -
itative di fferences bet ween the el ect rochemi cal noi se si gnat ures for uniform, pi t t i ng, and
crevi ce corrosi on. A number of workers have correl at ed noi se measurement s to the occur-
rence of l ocal i zed corrosi on. Pitting has been the most studied, wi t h appl i cat i ons wi t h steels
[3], st ai nl ess steels in C1 [4-6], and al umi num al l oys [7,8]. The pi t t i ng i ndex (PI) is cal -
cul at ed as shown accordi ng to Eden et al. [9]:
P I ~-- ~
Irm ~
where a~ is the st andard devi at i on of the ECN and Irm~ is the r oot mean square of t he coupl i ng
current. Because the ECN may vary with t i me, t hese quantities shoul d be cal cul at ed over
the same t i me peri od.
The pi t t i ng i ndex can have a val ue bet ween 0 and 1. Eden et al. have report ed the cl as-
sification for the pi t t i ng i ndex shown in Tabl e 1. The reasoni ng behi nd the formul at i on of
TABLE 1--Correlation between pitting index (PI) and type of
corrosion expected [9].
Pitting Index Value Range Type of Corrosion Expected
0.001 < PI < 0.0l uniform corrosion
0.0l < PI < 0.1 mixed corrosion
0.1 < PI < 1.0 localized corrosion
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
KELLY ET AL. ON ANALYSIS OF ELECTROCHEMICAL NOISE 103
the PI is that pi t t i ng occurs on i ni t i al l y passi ve el ect rodes, whi ch exhi bi t smal l coupl i ng
currents, and the i ni t i at i on of l ocal i zed attack can be det ect ed by the rel at i vel y l arge current
transients, whi ch l ead to a large st andard devi at i on in t he current. The r oot mean square of
the current signal is used i nst ead of the mean current to account for all current passi ng
bet ween t he t wo el ect rodes, i ndependent of its di rect i on.
The present wor k report s the results of exper i ment s whose ai m was to eval uat e cri t i cal l y
the correl at i on among ENM, convent i onal el ect r ochemi cal measurement s used in corrosi on,
and physi cal measurement s of attack.
Experimental Procedure
Resul t s are report ed for the exper i ment al arrangement shown in Fi g. 1. Three nomi nal l y
i dent i cal SS410 el ect rodes (24.5 by 19 by 1.6 mm thick) were i mmer sed 15 mm into the
sol ut i on so t hei r spot - wel ded el ect ri cal connect i ons were not in cont act wi t h solution. Thi s
Stabilizing holders
_ I
L_
L
SCE
J
$410 identical electrodes
1 L beaker - quiescent solution
~ A u x i l i a r y
ctrode (Pt)
Reference ~ ~ " ---~ sldentical
electrode ( S C E ) ~ electrodes
FIG. l--Experimental arrangement of electrodes in solution. The electrical connections were made
above the waterline to avoid crevice corrosion. No preferential attack at the waterline was observed.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
104 ELECTROCHEMICAL NOISE MEASUREMENT
arrangement was used to avoi d crevi ces and resul t ed in an exposed area of 13.5 cmL One
of the nomi nal l y i dent i cal el ect rodes served as a reference el ect r ode (referred to herei naft er
as the E-el ect rode), whereas t he ot her t wo (referred t o as the 1-electrodes) were connect ed
t hrough a Kei t hl ey Model 485 pi coammet er. A sat urat ed cal omel el ect r ode ( SCE) was al so
i ncl uded in t he cel l for measur ement of t he corrosi on pot ent i al (Eco~r)- A pi ece of pl at i num
mesh was used as an auxi l i ary el ect rode.
The t hree i dent i cal el ect rodes wer e connect ed to t wo ampl i fi er ci rcui t s based on Anal og
Devi ces Model AD524BD operat i onal ampl i fi ers shown in Fi g. 2. Thi s al l owed smal l de-
vi at i ons in pot ent i al bet ween the E- el ect r ode and I-el ect rodes to be moni t or ed at a hi gh
resol ut i on. For the 12-bit anal og- t o- di gi t al ( A/ D) convert er used, the 100 ampl i fi er resul t ed
in an i nst rument at i on resol ut i on of 2.4 I~V. The negat i ve i nput channel s on t he ampl i fi ers
were connect ed to t he el ect r ode that hol ds t he ground i nput for t he pi coammet er. The posi t i ve
i nput channel was connect ed to the E-el ect rode. The common out put channel was used as
the gr ound signal for t he dat a acqui si t i on board. It was found that i t was i mport ant t hat this
signal was connect ed to the ground i nput on the Kei t hl ey Met r abyt e Model DAS- 16 dat a
Terminal
box
Ampl i f
Data acquisition s ys t em l
aal
3
ithley 485
oammeter
Identical electrodes
FIG. 2- - El ect roni cs o f data acquisition with details of connections. Amplifiers were f or measurement
of EPN whereas the pi coammet er was used f or measurement of ECN.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
KELLY ET AL. ON ANALYSIS OF ELECTROCHEMICAL NOISE 105
acqui si t i on syst em (DAS) board, not t he l ow input. The Vou , channel was connect ed to the
hi gh-i nput channel on the DAS board.
The pi coammet er ground i nput was connect ed to t he sampl e that hel d the ground i nput
for the ampl i fi er as shown in Fi g. 2. The hi gh-i nput channel was connect ed t o t he sampl e
used onl y for current measurement . The pi coammet er was connect ed to the DAS boar d wi t h
the ground output from the pi coammet er connect ed to gr ound i nput on the DAS board. The
hi gh out put from t he ammet er was connect ed to the hi gh i nput on t he DAS board.
Measur ement s were per f or med t hree t i mes a day at 6-h i nt erval s for seven days. A single
measur ement consi st ed of moni t ori ng t he current bet ween t wo of t he el ect rodes (the I-
el ect rodes) and the pot ent i al di fference bet ween this coupl e and t he t hi rd SS410 el ect rode
(the E-el ect rode) at 20 Hz for 30 min. The col l ect i on frequency of 20 Hz was chosen to
det ect fast er transients t han are nor mal l y consi der ed in ENM. No frequency dependence was
found for t he el ect r ochemi cal noi se par amet er s report ed here over a frequency range of 1 to
1000 Hz. Resul t s from ot her anal yses of t hese dat a wi l l be report ed later.
Af t er the measur ement of the EPN and ECN (for use in the det er mi nat i on of R, and PI),
four ot her paramet ers were measured. The measur ed par amet er s i ncl ude Eco ~, whi ch we
define as the pot ent i al of t he I-el ect rodes versus a sat urat ed cal omel reference (SCE); E,vg,
whi ch we define as the average pot ent i al di fference bet ween the E- el ect r ode and the
1-electrodes for a gi ven t i me peri od; and the average coupl i ng current, lavg, bet ween the I-
el ect rodes over the same t i me peri od. Not e that EPN refers to the fluctuations in the pot ent i al
about some average val ue bet ween t wo nomi nal l y i dent i cal el ect rodes, whereas E, vg is the
average value. Wher eas for t wo t rul y i dent i cal el ect rodes Eavg woul d be zero, for passi ve
mat eri al s undergoi ng l ocal i zed cor r osi on l arge di fferences in pot ent i al can devel op bet ween
t wo i sol at ed, nomi nal l y i dent i cal el ect rodes. The average current has the same magni t ude as
the r oot mean square of the current, but can be of ei t her pol ari t y. Thi s was used to al l ow us
to moni t or whi ch of t he I-el ect rodes was the net anode at any poi nt duri ng an experi ment .
In addi t i on, in some tests, the pol ar i zat i on resi st ance of the E- el ect r ode was measured ac-
cordi ng to t he ASTM Pract i ce for Conduct i ng Pot ent i odynami c Pol ari zat i on Resi st ance Mea-
surement s (G 59) usi ng an SCE as a reference el ect r ode and a pl at i num count er el ect rode.
Three test sol ut i ons were used: 0. 1M hydr ogen chl or i de (HC1), phosphat e buffer at pH 7
wi t h addi t i on of 0. 5M sodi um chl ori de (NaCI) and 0. 01M hydrogen per oxi de (H20/ ) aft er
four days of i mmersi on, and 1M NaC1 at pH 3. Al l chemi cal s were reagent grade. Al l
measurement s were made at r oom t emperat ure.
Res ul t s
Results in Acid
The use of 0. 1M HCI sol ut i on al l owed the st udy of uni form at t ack of the SS410. Because
of t he l ow chr omi um cont ent of this mat eri al , as shown in Table 2, it cannot spont aneousl y
passi vat e in this solution. Instead, the mat eri al undergoes active di ssol ut i on at a hi gh rate
(about 600 ixA/cm2). Al t hough the l arge surfaces of the sheet speci men underwent uni form
corrosi on, some pi t t i ng was obser ved on the exposed edges, most l i kel y a result of a pref-
erent i al attack of i ncl usi on stringers. In the seven-day tests, mass l osses of appr oxi mat el y
TABLE 2--Composition of SS410 stainless steel (all in wt %).
Fe Cr Ni Mo Cu C AI Mn P S Si
Balance 11.73 0.09 0.0l 0.05 0.12 0.007 0. 41 0.015 0.005 0.34
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
106 ELECTROCHEMICAL NOISE MEASUREMENT
100 mg / c m 2 were observed, whi ch cor r esponded to t hi ckness l osses on the or der of 230
la.m.
The corrosi on pot ent i al (Ecor~) of the SS410 duri ng exposure to 0. 1M HC1 was - 0 . 4 9 V
(SCE) and fai rl y const ant ( + 2 mV) aft er the first 40 h. The Ear g was al so st eady and less
than 1 mV. The pol ari zat i on resi st ance was st eady at appr oxi mat el y 15 f~ (195 f~-cm2). The
noi se resi st ance was not qui t e as st abl e as Rp, as shown in Fi g. 3, and was gener al l y hi gher
than the pol ari zat i on resi st ance by appr oxi mat el y a fact or of 3, al t hough in rare i nst ances
R n was less than Rp. The lavg reached a st eady state of appr oxi mat el y 5 ~A. The pi t t i ng
i ndex vari ed bet ween 0.069 and 0.87 as shown in Fi g. 4.
The addi t i on of 0. 2M sodi um hydr oxi de (NaOH) (vi a t wo separat e addi t i ons, each of
whi ch i ncreased the concent rat i on by 0. 1M) was used to passi vat e the el ect r ode and to
det ermi ne i f the vari ous el ect rochemi cal paramet ers coul d fol l ow such a change. Af t er the
addi t i on of the NaOH, t he open ci rcui t pot ent i al first fel l and then i ncreased dramat i cal l y.
Bot h the pol ari zat i on resi st ance and the noi se resi st ance i ncreased dr amat i cal l y as wel l , from
values on the order of 15 11 to values on the order of 10 to 12 kD, The pi t t i ng i ndex fell
aft er the addi t i on of NaOH t o val ues of bet ween 0.1 and 0.3.
The wei ght loss of the E- el ect r ode was found to be 0. 6806 g, whereas the t wo I-el ect rodes
l ost 0.4803 and 0. 5510 g.
Results in pH 7 Phosphate Buffer with Added Chloride and Peroxide
SS410 spont aneousl y passi vat es in neutral phosphat e buffer solutions. In t hese solutions,
a prot ect i ve oxi de forms, whi ch subst ant i al l y di mi ni shes uni form corrosi on. The addi t i on of
chl ori de i ons and per oxi de (to rai se the pot ent i al ) can di srupt the fi l m and l ead to pitting.
Thi s envi ronment was chosen to st udy the el ect rochemi cal noi se from a passi ve el ect r ode as
wel l to serve as a test vehi cl e for a st udy of t he t ransi t i on t o pitting. The speci mens exposed
2 0 0
Rp
R.
150 --
~ 1 0 0 -
9 - - _ - _ - _ - _ - = = -
O - l i F - I I i
0 2 0 4 0 60 80 100 120 140
Ti me / h
FIG. 3- - Rp and R.f or SS410 in O.1M HCI as function of exposure time. Note that effect of addition
of NaOH at 135 h is not shown.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
KELLY ET AL. ON ANALYSIS OF ELECTROCHEMICAL NOISE 107
.8 -
~
0 . 6
e ~
~ 0 . 4 4
0 . 2 -
0
0
I I i I I 1
25 50 125 150
1
75 100 175
T i me / h
FIG. 4- - Pi t t i ng index f or SS410 in O. 1M HCl as f unct i on o f time. Arrows indicate addition o f O. 1M
NaOH.
--]~l. 0.5 M NaC1 and - I - E~o
0 ' 0 . 0 1 MH2 0 2 a d d e d I I ~ E , ,
- 50
- 100 t
i ; i ~ L
- - 2 0
- - 2 2
- - 24
- - 2 6
- - 2 8
- - 3 0
0 25 50 75 100 125 150
T i me / h
FIG. 5 - - E . , r ~ a n d E.,,g versus time f or SS410 in borate buffer. E~orr was measured v e r s u s a saturated
cal omel electrode, whereas E , ~ , , g was the average pot ent i al difference between the E- and 1-electrodes.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
108 ELECTROCHEMICAL NOISE MEASUREMENT
to this envi ronment showed no measur abl e uni form corrosi on, but each el ect rode devel oped
pi t s aft er the sol ut i on addi t i ons.
The t i me dependenci es of t he t wo pot ent i al s measured (Eco, and E a v g ) a r e shown in Fig.
5, al ong wi t h an i ndi cat i on of the t i me at whi ch the sol ut i on addi t i ons were made. The
corrosi on pot ent i al of the speci mens r ose sl owl y before t he sol ut i on addi t i ons. The Eavg
fl uct uat ed bet ween - 2 2 and - 2 9 mV. Af t er addi t i on of the chl ori de and per oxi de, t he Eco ~
fel l dramat i cal l y, i ndi cat i ng the i ni t i at i on of a stable pit on one of the I-el ect rodes. The
pol ari t y of Eavg was such that a posi t i ve val ue of the open ci rcui t pot ent i al cor r esponded to
the E- el ect r ode bei ng at a hi gher pot ent i al than the I-el ect rodes.
The coupl i ng current was ext r emel y smal l (on the or der of l 0 hA) before the addi t i on of
chl ori de. Af t er addi t i on of t he chl ori de and peroxi de, a large, l ong- l i ved t ransi ent was mea-
sured, wi t h a peak val ue of over I l 0 txA as shown in Fi g. 6, i ndi cat i ng that one of t he I-
el ect rodes had i ni t i at ed a pi t and a subst ant i al amount of the cat hodi c react i on support i ng
pi t growt h was occurri ng on t he ot her I-el ect rode. Thi s current decr eased over 24 h to 60
ixA at the t ermi nat i on of the experi ment . Smal l pits, confined most l y to the edges of the
coupons, were obser ved on bot h I-el ect rodes. A few large pits were found on the flat surface
of the c oupons - - one was a hemi spher i cal pi t roughl y 300 ixm in di amet er, the others were
el ongat ed pi t s rangi ng from 0.7 to 2 mm in length and 0.2 to 0.7 mm in width.
The noi se resi st ance and the PI decr eased dr amat i cal l y aft er the i ni t i at i on of pi t t i ng as
shown in Fi g. 7. The PI decr eased dr amat i cal l y and fell from a val ue of 0.5 to 0.015 aft er
the i ni t i at i on of pitting, al t hough it l at er i ncreased to 0.15 before t he t ermi nat i on of the
experi ment . The noi se resi st ance decreased from appr oxi mat el y 1.1 Ml-I to appr oxi mat el y 2
kD.
Results from 1M NaCl at pH 3
In some instances, subst ant i al pi t t i ng can occur in the presence of si gni fi cant uni form
corrosi on. Thi s is the case for SS410 in 1M NaC1 aci di fi ed to pH 3. Af t er the test, sampl e
<
J
120
100
8 0 -
60
4 0 -
20
o r
-20 -
0
0.5 M NaCI and I
Y 0.01 M H:O2 added /
. J
! t
75 100 125
T i me / h
25 50 150
FIG. 6--Average current as a function of time for SS410 in borate buffer. Arrow indicates addition
of O.5M CI- and O.01M H202 before measurement.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
KELLY ET AL. ON ANALYSIS OF ELECTROCHEMICAL NOISE 109
o ~
10000 -
1000 J
1 o o - '
1 0 -
1 -
4
0.5 M NaC1 and 1
0 . 0 1 M H2 0 2 a d d e d
0.1 I ~ I
0 25 50 75
Time / h
i
- 0.1
P I
i
~ - 0.01
F j - - 0.001
100 125 150
FIG. 7- - R, and pitting index as a function of time for SS410 in borate buffer. Arrow indicates addition
of O.5M CI- and O.01M HeO 2 before measurement.
surfaces showed some uniform corrosion, although the weight loss was t oo small to be
measured ( <0. 2 mg) after onl y seven days of exposure. A large number of small pits were
found on all exposed surfaces.
The corrosion potential of the I-electrodes rose over the first 30 h from - 0. 55 V (SCE)
to - 0. 325 V (SCE) before falling slowly to - 0 . 4 4 V (SCE) over the next 100 h as shown
in Fig. 8. The magnitude of E, vg slowly increased from less than 5 mV at the start of the
test to approximately 300 mV. The polarization resistance increased from 150 to 2800 f l as
Ecor~ increased as shown in Fig. 9. As the corrosion potential fell slowly, so did the polari-
zation resistance. The noise resistance generally tracked the polarization resistance. However,
the R,, values were again consistently higher than the R, values by approximately a factor
of two, as shown in Fig. 10 for a portion of the test.
The mean current varied sporadically during most of the test period, although its magni-
tude increased significantly after approximately 100 h as shown in Fig. 11. The pitting index
was high (close to 1) early, but then fell to approximately 0.1 after 25 h before oscillating
between 0.1 and 0.8 as shown in Fig. 11.
D i s c u s s i o n
Type 410 stainless steel has been used to investigate the correlation between a variety of
electrochemical parameters and the nature of the physical attack that occurs when it is
exposed to a variety of solutions. Reasonable correlation was found between the degree of
the uniform corrosion and the values of the polarization and noise resistances over a broad
range of solutions in which R, varied from 15 I I to 1 MD. However, the noise resistance
was consistently higher than the polarization resistance. Al t hough generally R, was approx-
imately three times greater than Rp, R, was sometimes more than an order of magnitude
larger, and occasionally slightly less than Rp. The noise resistance also appeared to be more
volatile in situations in which general corrosion was accompani ed by pitting, as in the 1M
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
110 ELECTROCHEMICAL NOISE MEASUREMENT
-300 [- - - I - F_ , ! 0
- 5 5 0 , , , x /
-600 r i t ~ [
0 25 50 75 100 125 150
Ti me / h
FIG. 8- - E .... and Ea,,g venus time f or 1M NaCl at pH 3.
-50
I
~ - 1 5 0 ~
J ~
-250
r
~ - 3 0 0
NaC1 solution at pH 3 (see Fig. 10). Nonetheless, both the noise resistance and the polari-
zation resistance were able to detect the effects of a large change in the corrosivity of the
environment. For example, when the pH of the acid solutions was raised to 10, both the
polarization and noise resistances increased by three orders of magnitude.
The weight loss measured for the E-electrode exposed to acid (on which the polarization
resistance measurements were performed) corresponds to a corrosion current of 4.57 mA
3000
2500 -
200O
1500
1000
50O
1 I I 1 I I
0 20 40 60 80 100 120 140
Time / h
FIG. 9--Rp versus time f or 1M NaCl at pH 3.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
KELLY ET AL. ON ANALYSIS OF ELECTROCHEMICAL NOISE 111
10000
--I I -- R,
8000 -
6000 -
4000
2000
I 7 - - 7 - k ~ - I
70 80 90 100 110 120 130
Time / h
FIG. 10--Rp and R, versus time f or 1M NaCl at pH 3.
140
(340 ixA/cm2). This woul d agree well with the polarization resistance values i f both the
anodic and cathodic Tafel slopes were 120 mV/ decade, that is, for a Stern-Geary B coeffi-
cient of 0. 026 V. Good agreement with the noise resistance woul d occur if the B coefficient
were 0. 052 V. This woul d require an extremely high anodic Tafel slope (implying passivity),
because the expected Tafel slope for hydrogen evolution is 120 mV/ decade. Polarization
- 2 - ~
<
:::L
- 6 - -
- 8 ~ ~
0 25 50 75
T i me / h
P , 1
i
I- 0.8
i 0.6
- 0 . 4
- 0 . 2
I
100 125 150
FIG. 11--Average current and pitting index f or I M NaCl at pH 3.
~f
fro
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
112 ELECTROCHEMICAL NOISE MEASUREMENT
curves generated for SS410 in 0. 1M HCI exhibited an anodic Tafel slope of 109 mV/ decade
and a cathodic Tafel slope of 105 mV/ decade. The importance of these results lies in the
use of ENM for corrosion rate monitoring. Al t hough R, can be used for monitoring changes
in corrosivity, these results indicate that it will generally overestimate the polarization resis-
tance, leading to an underestimation of the corrosion rate.
An electrochemically measurable parameter that characterizes the degree of localized cor-
rosion would be of great practical use. The PI has been proposed [9] to be that parameter,
with higher values indicating localized corrosion and lower values indicating uniform cor-
rosion. This classification has been developed based on measurements in complicated sys-
tems in which post-test verification of the type of attack was difficult, and the determination
of when pitting initiated was impossible. By choosi ng a simple system, a critical test of the
PI has been performed. SS410 is spontaneously passive in phosphate buffer solutions. In
quiescent solutions, as the passive film grows, the anodic reaction slows, leading to a rise
in the corrosion potential. In the absence of an aggressive species, the material will remain
passive. However, for the tests reported here, the pitting index ranged in value from 0.2 to
0.9, whereas the SS410 was exposed onl y to phosphate buffer. This would normal l y be
interpreted as indicative of localized corrosion. Clearly, this would be an erroneous
interpretation.
After the addition of 0. 5M chloride ion and 0. 01M peroxide, SS410 will undergo pitting
in phosphate buffer. When a material begins to pit, the corrosion potential falls dramatically.
This can be rationalized based on mixed potential theory. When a small, high-rate dissolution
site such as a pit is produced, the remainder of the surface acts as the cathode. Because
oxygen reduction is sluggish on the oxide-covered surface and is diffusion limited, the po-
tential of the pit and the passive surface falls, Figure 5 shows this effect. Not e that the Eavg
was not affected by the solution addition (see Fig. 5), but there was a large increase in the
Iavg (see Fig. 6). The most likely explanation of this is that pitting occurred on one of the I-
electrodes and the E-electrode.
The PI fell to very low levels during the pitting (0.015), which would indicate predomi-
nantly uniform corrosion. However, this was not the case. On each electrode, one or two
large pits and a number of smaller pits were observed in the otherwise passive surface.
These results have implications for corrosion monitoring via ENM. Clearly, the Ecor~ of
test specimens appears to be more sensitive to changes in the corrosivity than was either the
Eavg or PI. This is unfortunate, because the maintenance of reliable reference electrodes is
difficult in many industrial applications that require corrosion monitoring. Of particular con-
cern is the possibility of misinterpretation of the ENM. The pitting index was found to be
unreliable as presently constituted. During the period when the SS410 was clearly passive,
the PI indicated that localized corrosion was occurring. In addition, when localized corrosion
was occurring, the PI indicated that the predominantly uniform corrosion was taking place.
The development of pitting on the E-electrode prevented it from following the course of the
pitting process on one of the I-electrodes. This would complicate its use in the monitoring
of localized corrosion as well.
Although the reasons behind the lack of correlation between the PI and the observed attack
are still unclear, these results cast doubt on its use for the detection of the onset, propagation,
and cessation of localized corrosion. Mathematically, the PI is a relative standard deviation,
indicating the size of the current fluctuations relative to the average coupling current. In the
case of SS410 in phosphate, the very small values of I , ~ when the system is passive drive
the index to values indicative of localized corrosion, as the measured fluctuations (th) are of
the same magnitude. Because both I~n s and tr~ are an order of magnitude above the electrical
noise of the instruments used, it appears that the values reported reflect the electrochemical
processes under study. During the propagation of localized corrosion, the I, , s rises more than
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
KELLY ET AL. ON ANALYSIS OF ELECTROCHEMICAL NOISE 113
does the Orl, which causes the PI t o fall to low values indicative of uniform corrosion.
Unfortunately, the PI, as presently constituted, is not able to differentiate localized and uni-
form corrosion for this system.
The system studied was specifically chosen to provide an unambi guous test of the various
parameters derived from ENM. The noise resistance shows promise for monitoring changes
in corrosivity. While the fundamental basis for noise resistance remains a topic of debate
[10-12], this is an important practical application of electrochemical techniques to corrosion
engineering. The pitting index, as presently constituted, shows much less promise as a tool
to differentiate localized and uniform corrosion.
Co nc l us i o ns
1. The polarization resistance of SS410 in solutions ranging from pH 1 to 7 increases by
five orders of magnitude. Over this range of solution composition, the noise resistance
generally tracks the polarization resistance. Hence, R, can be used t o moni t or changes
in solution corrosivity.
2. The noise resistance was found to be consistently higher than the polarization resis-
tance. The use of R, in the Stern-Geary relation woul d lead to an underestimation of
the corrosion rate. In general, the ratio of R, to Rp was usually t wo t o three for the
results reported here, although values for this ratio ranged from less than one to
fourteen.
3. The pitting index did not correlate to the occurrence of pitting. For SS410 in a phos-
phate buffer solution, the PI indicated that localized corrosion was occurring. After the
addition of 0. 5M CI - and 0. 01M H202, the P! indicated that the SS410 was undergoing
predominantly uniform corrosion when post-test examination showed that pitting was
the sole mode of attack.
Acknowl edgment s
The technical assistance of N. M. Brush in the performance of the experiments is gratefully
acknowledged. The authors would like to acknowledge the financial support of Mobil Ex-
ploration and Production Technol ogy Company (Dr. C. Kroen).
Re f e r e nc e s
[1] Iverson, W. E, Journal of the Electrochemical Society, Vol. 115, 1968, pp. 617-618.
[2] Dawson, J. L., Farrell, D. M., Aylott, E J., and Hladky, K., Paper 31, Corrosion/89 Conference,
NACE International, Houston, TX, 1989.
[3] Hladky, K. and Dawson, J. L., Corrosion Science, Vol. 23, 1982, p. 231.
[4] Magaino, S., Kawaguchi, A., Hirata, A., and Osaka, T., Journal of the Electrochemical Society,
Vol. 134, 1987, p. 2993.
[5] Stolica, N., Corrosion Science, Vol. 9, 1969, p. 205.
[6] Schwenk, W, Corrosion, Vol. 20, 1964, p. 129t.
[7] Magaino, S., Boshoku Gijutsu, Vol. 32, 1983, p. 712.
[8] Magaino, S., Yasuda, M., and Yamada, H., Boshoku Gijutsu, Vol. 34, 1985, p. 157.
[9] Eden, D. A., John, D. G., and Dawson, J. L., International Patent WO 87107022, World Intellectual
Property Organization, 1987.
[10] Mansfeld, E and Xiao, H., Journal of the Electrochemical Society, Vol. 141, 1994, pp. 1403-1404.
[11] Eden, D. A., Journal of the Electrochemical Society, Vol. 141, 1994, pp. 1402-1403.
[12] Bierwagen, G. P., Journal of the Electrochemical Society, Vol. 141, 1994, pp. L155-L157.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
L u c Be a u n i e r , t Jo#' l F r y d ma n , 1 C l a u d e Ga b r i e l l i , ~ F r a n g o i s Hu e t , J
a n d Mi c h e l K e d d a m ~
Comparison of Spectral Analysis with Fast
Fourier Transform and Maximum Entropy
Method. Application to the Role of
Molybdenum Implantation on Localized
Corrosion of Type 304 Stainless Steel
REFERENCE: Beaunier, L., Frydman, J . , Gabrielli, C., Huet, E, and Keddam, M., "Com-
parison of Spectral Analysis with Fast Fourier Transform and Maximum Entropy
Method. Application to the Role of Molybdenum Implantation on Localized Corrosion of
Type 304 Stainless St eel , " Electrochemical Noise Measurement f or Corrosion Applications,
ASTM STP 1277, Jeffery R. Kearns, John R. Scully, Pierre R. Roberge, David L. Reichert, and
John L. Dawson, Eds., American Society for Testing and Materials, 1996, pp. 114-128.
ABSTRACT: A comparison of a spectral analysis using the fast Fourier transform (FFT) and
the maximum entropy method (MEM) was carded out in the case in which both methods can
be performed, that is, when several time acquisitions can be recorded. A summary of the
principles of the MEM is given. Then the main properties of this method are investigated, that
is, influence of the MEM order on the spectrum accuracy, validity of the low-frequency plateau
usually given by this technique, overlapping of spectra measured for different frequency band-
widths, and influence of a slow evolution of the amplitude of the signal fluctuations.
The susceptibility to pitting corrosion of type 304 stainless steel and type 304 modified by
molybdenum (Mo) by means of ion implantation was studied. The power spectral densities
(PSD) measured with the FFT and MEM techniques are in reasonable agreement, except for
low electrochemical current noises (ECN) buried in the parasitic noise generated by the power
supply. In that case, the FFT technique is more appropriate than the MEM, which gave qual-
itative results only. The type 304 stainless steel showed a large metastable pitting leading to
only a few macroscopic pits, whereas the type 304 Mo-implanted specimen showed a very low
metastable pitting leading to many hemispheric pits covered by the Mo-implanted layer, under
which localized corrosion occurred.
KEYWORDS: potentiostatic electrochemical noise measurement, fast Fourier transform, max-
imum entropy method, localized corrosion, molybdenum implantation, stainless steel, data
analysis techniques
The i nvest i gat i on o f l ocal i zed cor r os i on of t en l eads t o t he anal ysi s o f r a ndom cur r ent
f l uct uat i ons r el at ed t o t he r a ndom bi rt h, gr owt h, and r epassi vat i on o f pi t s at an el ect r ode
surface. It is gener al l y t hought that t he anal ysi s o f t he t i me seri es o f el ement ar y event s
occur r i ng at pot ent i al s bel ow t he pi t t i ng pot ent i al al l ows t he suscept i bi l i t y o f an al l oy or
met al t o l ocal i zed cor r os i on t o be as s es s ed [1,2].
i Physique des Liquides et Electrochimie, UPR15 CNRS, Universit6 Pierre et Marie Curie, Tour 22,
4 place Jussieu, 75252 Paris Codex 05 France.
Copyrighl 9 1996 by ASTM International
114
www.astm.org
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
BEAUNIER ET AL. ON COMPARISON OF SPECTRAL ANALYSIS 115
During the prepitting stage, the usual approach for characterizing such a random signal is
to estimate its power spectral density (PSD) or spectrum, that is, the distribution of the power
of the signal in the frequency domain.
Usually the PSD of a random signal is obtained through a fast Fourier transform (FFT)
algorithm. However, especially when onl y one experiment is available, or when the signal
drifts t oo much and therefore limits the measurement time, the maxi mum entropy method
(MEM) was proposed to obtain the PSD [ 3 , 4 ] . This method was first applied in geophysics,
more generally in earth sciences, and from 1967 many papers have been published on this
topic (see Ref 5 and references t h e r e i n ) . The proponents of the MEM claimed that it had
some advantages:
1. It needed only one time acquisition.
2. It gave smooth spectra.
3. It led to the values of the spectrum at frequencies lower than the inverse of the acqui-
sition time.
This method has already been used in corrosion investigation for studying pitting of alu-
mi num [6], mild steel [7], and pure iron [8], to obtain data at frequencies much lower than
with the FFT technique while keeping reasonable acquisition times.
In this paper, a compari son of a spectral analysis with the FFT and the MEM was at-
tempted on systems where the t wo methods can be carried out, that is, when several time
acquisitions can be recorded, to assess their relative efficiencies. This compari son was con-
ducted by processing signals delivered, first, by electronic devices, then by type 304 stainless
steel electrodes. I n that case, the effect of mol ybdenum (Mo) implantation on the localized
corrosion susceptibility in a chloride-containing acidic medi um was studied by using Mo-
implanted and unimplanted specimens.
Principles of the Analysis Methods
The PSD can be obtained through the peri odogram method based on the Cool ey Tuckey
FFT algorithm [9]. Here, the spectrum is estimated by:
~b(f) = T. t*j~-"7~ IlJ(f)P (1)
where
f = the frequency,
T = the acquisition time,
l j ( f ) = the Fourier transform of the elementary time recording l j ( t ) ,
9 I = the number of time recordings.
This method has the advantages of being simple, easily implementable on a computer, and,
above all, applicable to a large class of random signals.
However, despite these advantages, the FFT needs, to obtain accurate spectra, t o average
ove r . l" squares of the modul us of the Fourier transforms of I j ( t ) (see Eq l); therefore, the
peri odogram method will be designated as FFT(. I ). This leads to the following drawbacks:
1. The time to acquire the data is multiplied by 9 ~ compared to a method that would
need only one acquisition.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
1 1 6 ELECTROCHEMICAL NOISE MEASUREMENT
2. It is appropri at e onl y i f the t i me si gnal s are reproduci bl e, that is, i f several acqui si t i ons
can be recorded duri ng the same experi ment .
Therefore, this met hod is i ncompat i bl e wi t h experi ment s that can be carri ed out onl y once
or that do not l ast a suffi ci ent l y l ong t i me. For this reason, t he MEM was devel oped.
Si mi l ar to the FFT anal ysi s, there are t wo ways for est i mat i ng t he PSD by t he MEM,
ei t her a di r ect met hod, in whi ch the spect ral est i mat i on is made on t he signal sampl es, or
an i ndi rect met hod, in whi ch t he PSD is est i mat ed t hrough the aut ocorrel at i on funct i on [5].
We have tested bot h met hods and, for our condi t i ons, t hey gave si mi l ar results. Therefore,
we use, in this work, t he second met hod (Levi nson al gori t hm), whi ch l eads to faster
cal cul at i ons.
The sampl i ng duri ng a t ot al t i me T of the t i me series x(t) of N dat apoi nt s separat ed by
At = T/ N l eads to a di scret e set of dat a x(n) t aken at t i me nAt (in our case N = 2048).
The MEM is an est i mat i on of the PSD suggest ed by the i nformat i on theory. It was first
assumed that the sampl ed data, x(k), can be descr i bed by an aut oregressi ve model :
y(k) =x ( k ) - a , y ( k - 1 ) - a 2 y ( k - 2 ) - . . . - a My ( k - M) (2)
where y(k) is the est i mat ed val ue and M is the or der of the model. Then it was assumed that
the est i mat i on of the PSD + ( f ) was:
1. Coherent wi t h the N known val ues of the aut ocorrel at i on function R~(m) defi ned at
t i me mAt by
1 u m-I
R~(m) = ~ ~ x(n) x(n + m)
,~=o
(3)
Hence the Wi ener - Khi nt chi ne t heorem gi ves
f [
m.~ e z' ~a' d~(f ) df = R~(k)
Fmax
O<- k <- N - 1 (4)
where Fm,x is the Nyqui st frequency, that is, the maxi mum anal yzed frequency Fmax =
1 ~2At.
2. Independent of any hypot heses on t he unknown values of Rx~(m) for m -> N.
This last poi nt is equi val ent to i mposi ng maxi mum ent ropy to the PSD versus the values
R~(m), m >- N, hence the name of the method. Not e that t he per i odogr am met hod assumes
that R,~(m) = 0 for m -> N.
An al gori t hm was deri ved to est i mat e the PSD that can be i mpl ement ed on a comput er
[5]. Cont rary to the FFI" technique, whi ch gi ves a spect rum in the range [ 1/ T = 2Fm,x/N,
Fmax], the PSD is est i mat ed by the MEM at any frequency f - < Fm,x, wi t h
2 tr 2 At
(5)
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
BEAUNIER ET AL. ON COMPARISON OF SPECTRAL ANALYSIS 1 17
where o ~, a I . . . . , a M are M + 1 coefficients dependent on M, which are calculated from
the (M + 1) (M + 1) matrix equation
[
R~(O) R=( I )
, %. 0) , %(0)
R=iM) Rxx(M- 1)
R x ~ ( M - l ) a I [ o l
a2
9 - . R~(O)
The maj or parameter of this algorithm is the order M o f the model. Therefore, the MEM
will be called MEM(M).
Properti es o f t he MEM
The MEM properties of interest to electrochemists will be reported below by processing
either a white noise signal generated by a Solartron JM 1861 or a signal generated by a
corrodi ng electrode. The signals were digitized with a real-time comput er (Concurrent Com-
puter RTU 6600). No drift was observed on the analyzed signals; therefore, the onl y oper-
ation on the time recordings was the elimination of their mean value before comput i ng the
PSD.
Influence of the Order of the MEM on the Accuracy of the Spectra
To examine the influence o f the order of the MEM on the accuracy of the estimated
spectra, the spectrum of a white noise was calculated for various orders 10, 30, 100, and
300 [MEM(10), MEM(30), MEM(100), MEM(300)] and compared with a F F r processing
averaged over 20 elementary spectra [FFT(20)]. The MEM spectra were calculated with the
same time record, and, for a better reading of Fig. 1, the MEM(10), MEM(30), MEM(100),
and MEM(300) spectra were divided by 10, 102, 10 3, and 104, respectively. The comparisons
plotted in Fig. 1 show that the levels of all spectra are the same, and that the opt i mum order
of the MEM is a trade-off: an order that is too low leads to poor frequency resolution (smooth
spectrum) with a few wiggles in the high-frequency range, whereas an order that is too high
leads to a finely resolved spectrum with many spurious oscillations. In Figs. 1 and 4, the
decrease of the spectra at frequencies higher than 0.67 Fro, x is due to the anti-aliasing low-
pass filter.
Performance of the MEM for Signals Without a Low-Frequency Plateau in the Analyzed
Frequency Bandwidth
The signals whose spectra are plotted in Fig. 2 were generated by filtering a white noise
signal by an electronic circuit in which the characteristic frequency of the RC circuit is lower
than the inverse of the time record duration 1/T, whi ch is the mi ni mum frequency 2Fm~x/N
available with the FFr . In this experiment, the FFI" was chosen as a standard, although it
needed measurement times 20 times longer than the MEM to be estimated. Here, the
MEM(10) spectrum was calculated with a logarithmic frequency distribution with 20 fre-
quencies per decade between 1 Hz and 50 kHz to save calculation time. The curves in Fig.
2 show that the MEM leads to spectra whose low-frequency parts [ f -< 41T] are not in
good agreement with those obtained through FFr . Therefore, the validity of the low-
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
118 ELECTROCHEMICAL NOISE MEASUREMENT
t
N
r
>
I0--~
i O - s
I 0 -6
I 0 -~ '
1 0 - 8 .
i 0 - 9
lo_iO i
i O - u
iO - m
i i
il
b
c
d
e
i i i i t i
i i ! i i
lO 2 l O s
F REQUENCY
i g ! 1
i t ! i
/ Hz
i i t i , ~ 1
i i i !
5
FIG. 1- - PSD of white noise analyzed by means of an FFT algorithm (a : FFT(20)) and an MEM of
different orders (b : MEM(IO); c : MEM(30); d : MEM(IO0); e : MEM(300)). The curves b, c, d, and
e were divided by 10, 103, 103, and 104, respectively.
frequency plateau obtained by the MEM is often questionable. Not e that Eq 5 always leads
to a finite value +(0).
Overl appi ng o f t he Meas ur ed Spect ra
The MEM measurement can be carried out accurately over three decades (because the
number of samples in one time recording is usually N = 2048). For broadband random
signals, as the spectrum changes over more than three decades, it is necessary to perform
measurements on several overlapping bandwidths to analyze the whole useful frequency
range. In Fig. 3, these measurements were carried out for filtered white noise (Curve a) and
unfiltered white noise (Curve b). For the high-frequency measurement with Fma x = 50 kHz,
the same probl em as ment i oned previously arises for the filtered signal: curve a shows that,
in the considered bandwidth (2Fmax/N, Fmax), there is no low-frequency plateau, so the low-
frequency part of the calculated spectrum is wrong. However, by carrying out a low-fre-
quency measurement (Fm,~ = 500 Hz), the whole spectrum is accurately obtained. For white
noise, onl y one high-frequency measurement is sufficient because, in this case, data can be
accurately obtained below 1/T. For both cases, the overlap of the two measured spectra is
reasonably good.
PS D o f Poor St at i onari t y Si gnal s
When the analyzed signal changes slowly with time, the spectrum estimated by the MEM
depends on the time record chosen. As an example, the random current generated by a type
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
BEAUNIER ET AL. ON COMPARISON OF SPECTRAL ANALYSIS 1 19
"7,
N
r
r
9,
10L
10 -s_
l0 -t~
t0 -~_
i0 -13_
i o -
i0 -~
!
a
b
i i i i i i i i i i i
, \
[ W h it e n ~ t . ~ V ~ , i
- - g e n e r a t o r ~ )
_ _
A,
I I
i i i i i i i i i i i i
\
I ] I [ I I I I i i i i
FREQUENCY / Hz
FIG. 2 - - PS D o f f i l t ered white noise by an electronic circuit given in the inset. (a) FFT(20) measured
in two overlapping bandwidths [0.5 to 500 Hz] and [50 Hz to 50 kHz]. (b) MEM(IO) measured f rom a
time record obtainecl f or F ~ = 50 kHz (logarithmic f requency distribution).
304 stainless steel corroding electrode was recorded over a time 204.8 s, giving then ten
recordings, each lasting T = 20.48 s. From the ten time T-recordings, the FFT(10) spectrum
was estimated, and each time T-recording was processed by an MEM of order 10, which
produced ten estimated spectra MEM(10). In Fig. 4a, the estimated spectrum FFI~(10) is
plotted with two estimated spectra MEM(10). It can be observed that the level of the low-
frequency plateau of the spectra MEM(10) (Curves a and b) changes widely and depends
on the time record chosen.
In Fig. 4b, the estimated spectra MEM(10) given in Fig. 4a are compared to the estimated
spectra FFT(1) obtained by processing the considered time record by FFI ~ without averaging
(that is, the acquisition time of the processed signal is the same for FFT(1) and MEM(10)
in each case). It can be seen that MEM(10) gave a smooth average of FFT(1) in each case.
In particular, the mean low-frequency plateau given by FFF(1) depended, as for MEM(10),
on the considered time recording, that is, depended on the pitting activity. Of course, the
average of the ten FFT(1) led to the spectrum FVF(10), which gave the averaged low-
frequency plateau measured over a time 10T.
Role of Mol ybdenum on Localized Corrosion of Type 304 Stainless Steel
To investigate the influence of molybdenum ion implantation on the susceptibility of a
type 304L stainless steel to localized corrosion, the ECN induced by the birth, growth, and
repassivation of pits at potentials in the prepitting range was analyzed for two different
specimens of type 304 steel (75% Fe, 17% Cr, 8% Ni in approximate weight percentages);
the surface of one of them (called type 304 Mo steel in the following) was implanted with
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
120 ELECTROCHEMICAL NOISE MEASUREMENT
=~
i
N
: =
>
t 0 - ' ~
1 0 - s _ :
1 0 - 6 _
t 0 -7 -
1 0 4 -
I 0 -s_
1 0 -l~
1 0 - u _
1 0 - ~ _
io-~_:
io-._!
i i ! I
b
a ~
I I I I I I l I
! I I [
I I I I
i
T I
i i I i ~ i
' ' " 1
|
io-' I b ~ i o ~ I b " i o 3 I b ' I b 5
F R E Q U E N C Y / H z
FIG. 3- - Overl appi ng of MEM(IO) spectra measured f rom a 20.48 ms ( F . . . . = 50 kHz) and a
2.048 s ( F , , ~ = 500 Hz) time record. ( a ) Filtered white noise. ( b ) White noise.
|
r
r
<
I E a
i 0 - s _
io-,O_
1 o - " _
I o - ~ _
Io-,3_
Io-"_
!
i 0 - ~ -
i !
a
l I
c ' - ~
b - - I
I I
! i i i
. T7
I I I [ I I I I I I I
l o - ' t b ~ l o ' l o ~
F RE QUE NCY / Hz
FIG. 4a- - Sl owl y time-varying random signals. Comparison of FFT(10) (Curve c) and two MEM(10)
(Curves a and b) obtained f rom t wo different elementary time records used to determine the FFT(10).
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
BEAUNIER ET AL. ON COMPARISON OF SPECTRAL ANALYSIS 121
i
N
r
<
~
9 ,
10 4
1 0 _ 9 '
10
| ! | !
a
b _ _
. . . . . . . . . . . . I I , ~,
' l l . . . . I
, , ,
, 4 mL -
t l f ' , l -
i , I ' F I -
,, J '11 t
F RE QUE NCY / Hz
FIG. 4b- - Sl owl y time-varying random signals. Comparison of MEM(IO) (Curves a and b, same as
in Fig. 4a) and FFT(1) (Curves c and d) obtained f rom the same time records.
mol ybdenum ions (Mo+). Mol ybdenum was chosen for its beneficial role against corrosion,
especially pitting.
Ma t e r i a l s
The electrode surface was a disk of 0. 5-cm diameter (area 0.2 cm2). The lateral surface
was covered with an insulating electrophoretic coating (cataphoretic paint used in car in-
dustries, Corona EW 2258 and EW 2269). The implantation process was performed at room
temperature with a beam of 100 keV and a fluence of 2.5 x l016 ions/cmL The structure
was characterized by transmission electron mi croscopy (TEM) at 100 keV. The mi crochem-
ical analysis was performed on a 200-keV TEM- STEM equipped with an X-ray energy
dispersive spectrometer (EDS analysis). The implantation profile was measured by Ruther-
ford backscattering analysis (RBS).
For interpretation of electrochemical noise results, the structural description of the im-
planted film can be summarized as follows: the RBS profile of the thin implanted film was
pseudo-gaussian with a maxi mum at a 25-rim depth and a maxi mum implantation depth of
60 nm shown in Fig. 5. By TEM, we could observe an austenitic-ferritic transformation
induced by the implantation process. Correlatively, the grain size was decreased from 5 to
12 i~m in the initial austenitic phase to 0.1 to 0.3 ixm in the ferritic phase. No growi ng of
martensite was observed by TEM at the surface region. X-ray analysis revealed a large
enrichment up to a mean value of 15% in weight in mol ybdenum in the transformed film.
However, the local composition measured with a 4-nm beam showed important fluctuations
in the distribution of the elemental concentrations. The structural description of the implanted
film is further detailed by Beaunier et ai. [10].
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
122 ELECTROCHEMICAL NOISE MEASUREMENT
FIG. 5--Gaussian distribution of the Mo implantation and associated microstructural changes versus
depth below the surface.
Experimental Procedure
The electrochemical investigations were carried out at room temperature in an aqueous
solution (volume 300 cm 3) of 0.5M sulfuric acid with 0.05M sodium chloride (NaCI) for the
type 304 stainless specimen at potential 0.3 V versus a saturated sulfate electrode (SSE).
For the type 304 Mo steel specimen, the conditions were more aggressive (0.3M NaCI,
potential 0.55 V/SSE) for testing protection against pitting corrosion. The counter-electrode
was a large surface platinum grid. The experimental setup is based on homemade low-noise
potentiostat and amplifiers; the ECN was recorded with an audiotape recorder (Biologic DTR
1200) for more convenient signal processing after the experiment. Later, the signal was
digitized and the PSD was calculated with the real-time computer.
The intrinsic noise of the instrumentation was modeled in [11]. The PSD of the parasitic
current noise superimposed to the ECN is the ratio of the PSD of the voltage noise measured
between the reference electrode and the working electrode, and the square of the modulus
of the working electrode impedance. Therefore,
1. It depends not only on the instrumentation used, but also on the system under study.
2. In the high-frequency range, it increases with frequency because the impedance mod-
ulus decreases with frequency (Figs. 6a and b).
3. It can be neglected in the low-frequency range where the impedance modulus is
important.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
BEAUNIER ET AL. ON COMPARISON OF SPECTRAL ANALYSIS 1 2 3
I N
104 -
10-9-
10_1o_."
10-"_
10 -m_
l 0 -'+_
1 0 - :
10 -m_
1 0 - ` 7 -
to-dO_
i O - ' s _
1 i 1
a ~
hi . ,
i i i i i i i i
1
I
b
I I I I I I I I I I I I i i i i
I
i i i i
l.J
t
i i i i
! i i i -
i i i
l 0 - 2 1 0 " l 1 3 ~ 1 0 1 1 0 2 l 0 a 1 0 + i 0
F REQUENCY / Hz
FIG. 6 a - - PS D of the ECN measured with the FFT technique in prepitting conditions. (a) Type 304
stainless steel electrode, 0. 5M H2S04 + 0. 05M NaCl at pot ent i al 0.3 V/ SSE. (b) Molybdenum-implanted
type 304 steel electrode, 0. 5M H2SO 4 + 0. 3M NaCl at potential 0.55 V/ SSE.
Resul t s and Di s cus s i on
Fi gur e 6 shows t he PSD o f t he ECN f or the i mpl a nt e d and uni mpl ant ed s peci mens , me a -
sur ed wi t h t he ~ t echni que (see Fi g. 6a) , and wi t h t he MEM ( see Fi g. 6b) f or compar i s on.
I n each case, t he me a s ur e me nt was pe r f or me d on t hr ee over l appi ng bandwi dt hs (Fmax, = 50
Hz, Fmax2 = 500 Hz, Fmax3 = 20 k Hz f or t he t ype 304 s pe c i me n and 5 k Hz f or t he ot her
one f or l i mi t i ng t he hi gh- f r equency r i si ng par t o f t he s pect r um as a r esul t o f t he pot ent i ost at
noi se) . I n Fi g. 6a, t he over l appi ng o f t he me a s ur e d s pect r a is f ai r l y good; f or t he t ype 304
Mo s pe c i me n (Fig. 6a, Cur ve b), t he l evel o f t he ECN wa s so l ow t hat we had t o use
Ha nni ng wi ndowi ng t o r educe t he wi dt h o f t he spect r al l i nes i nduced by t he p o we r s uppl y
( 50 Hz and har moni cs ) [9]. Wi t h t he MEM, t he over l appi ng o f t he me a s ur e d s pect r a ( see
Fi g. 6b) is accur at e f or a hi gh l evel o f ECN si gnal (t hat is, t ype 304 s peci men) but is
i naccur at e f or a l ow l evel o f ECN, wi t h as s oci at ed hi gh- ampl i t ude par asi t i c l i nes (t hat is,
t ype 304 Mo s peci men) , be c a us e o f t he i nabi l i t y o f t hi s t echni que t o r educe t he wi dt h o f
spect r al l i nes. Once mor e, Fig. 6b shows t hat t he ext r apol at i on o f t he s pect r um at f r equenci es
l owe r t han t he i nver se o f t he me a s u r e me n t t i me can be danger ous. The l ow- f r equency pl a-
t eaus obt ai ned wi t h t he t i me r ecor di ngs cor r es pondi ng t o Fma~ = 500 Hz and Fm,x = 5 k Hz
ar e dr amat i cal l y di f f er ent .
Whe n t he r a n d o m t i mes o f pi t gener at i on are a s s ume d t o be i ndependent ( Poi sson pr ocess) ,
t he PSD l evel is pr opor t i onal t o t he pi t gener at i on r at e [1]. Hence, t he l arge di f f er ence o f
t he PSD l evel , seen wi t h bot h FFF and MEM anal ysi s, f or t he t wo s pe c i me ns ( seven decades
at l ow f r equenci es ) r eveal s an i nt ense pr epi t t i ng act i vi t y f or the t ype 304 el ect r ode and a
ver y l ow act i vi t y f or t he t ype 304 Mo el ect r ode. However , it is not i ceabl e t hat t he t i me
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
124 ELECTROCHEMICAL NOISE MEASUREMENT
m
( ' ~
<
1 0 - a
! i ! t
1 0 - 9 _ : . - - _ _
1 0 - t 0 _ ! a
10
1 0 -t3 _ :.
10-t~_:
i o - :
i o - " l
1 o - t ! b
1 0 - t _ :
1 0 - t !
1 I 1 1
! ! ! !
1 1 1 1 l 1 I I
' ' " l
"q
1
J
I l I I
I 17
) ) i i t ~ ) T
1 0 - z t O - t i 3 0 1 0 t i 0 2 1 0 3 1 0 4 t 0 5
F RE QUE NCY / Hz
FIG. 6b- - PSD of the ECN measured with the MEM in the same conditions as in Fig. 6a; three
overlapping bandwidths were successively analyzed (0.05 Hz, 50 Hz), (0.5 Hz, 500 Hz), and (20 Hz, 20
kHz) (Curve a) or (0.05 Hz, 50 Hz), (0.5 Hz, 500 Hz), and (5 Hz, 5 kHz) (Curve b).
constants of the repassivation process are very similar (about 0.5 s). The typical time re-
cordings given in Fig. 7 confirm the presence of
1. Numerous depassivation-repassivation current transients (about a hundred per second)
for the unimplanted specimen with an average electrical charge of 1 I~C (see Fig. 7a).
2. Very few transients for the Mo-implanted specimen (about 0.02 per second) with an
average charge of 0.1 ixC (see Fig. 7b).
After the experiment, we observed considerable differences in the pitting state of the
unimplanted and implanted specimens: the type 304 stainless steel exhibited three or four
pits per square centimeter. A single pit with classic features is represented in Fig. 8, with
20 to 100 v.m in width. By comparison, the mol ybdenum implantation induced the formation
of numerous pits in the implanted layer (up to 30 times more than for the type 304 steel)
(Fig. 9). These effects were attributed to Mo because of its well-known role in the passive
film during pitting corrosion. Moreover, a classical pitting process was observed for speci-
mens implanted with nonchemical sensitive species such as argon, krypton, and neon, which
were used for testing the physical effect of implantation.
Second, pits in the Mo-implanted ferritic film had a dramatically smaller area compared
to the unimplanted specimen. We observed a very special feature. After initiation, the pits
developed as holes about 50 nm in width in the ferritic layer at points corresponding to
physical or chemical defects introduced by the implantation process [12]. After this initial
pitting process, the dissolution occurred under the implanted ferritic film in the austenitic
matrix, which gave rise to hemispheric pits of about 50 v.m in diameter. The cap on the pit
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
BEAUNIER ET AL. ON COMPARISON OF SPECTRAL ANALYSIS 125
<
:=J..
600
500
400
300
200
100
%
' I ' I ' I , I , I , I , I , i , i , i , i , I , i , [ , i , i , i , i , i ,
0 20 40 60 80 iO0 120 i40 180 i80 200
t i me / ms
FIG. 7a--Typical time recording of ECN corresponding to experimental conditions given in Fig. 6a
for type 304 stainless steel electrode.
1 , Z r m
1,2
1,0
.8 _
<
=I. .6
<1
. 2
0,0 ..~ . . . . . . A. , . ~ . . . . . . . ~ . . . . . . . . . . . . . . . . . . . . . . . . . . . . t, . . . . ~ . . . . . . .
/
0 1 2 3 4 5 6 7 8 9 lO
t i m e / s
FIG. 7b--Typical time recording of ECN corresponding to experimental conditions given in Fig. 6a
for Molybdenum-implanted type 304 steel electrode.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
126 ELECTROCHEMICAL NOISE MEASUREMENT
FIG. 8--Pi t observed on the type 304 stainless steel specimen.
was identified as the ferritic-implanted layer by TEM and EDS analysis. Figure 10 shows
that a confined corrosion process occurred in the pit covered by the implanted film, and the
corrosion products, issued from the occluded cell, which have crystallized over the primary
50-nm hole.
This study shows that the pit initiation process is very different after surface modification
by molybdenum ion implantation:
FIG. 9--Pits distributed on the Mo-implanted type 304 stainless steel specimen.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
BEAUNIER ET AL. ON COMPARISON OF SPECTRAL ANALYSIS 127
FIG. lO--Photograph of surface (top) and schematic illustration of cross section (bottom) of the pit
observed on the Mo-implanted type 304 stainless steel specimen.
1. For type 304 stainless steel with an austenitic matrix, the pit initiation occurs at chem-
ical defects, such as sulfate inclusions, and, immediately after, an important anodic
dissolution event takes place. This process leads to an open pit.
2. For Mo-implanted type 304 stainless steel, the pit initiation occurs at physical and
chemical defects. This process leads to a covered hemispheric pit, which gives rise to
confined corrosion with low current density followed by fracture of the implanted layer
and normal anodic dissolution.
Conclusion
1. The MEM and the FPT methods give similar results for calculating power spectral
densities, with a shorter measurement time for the first method. However, the advan-
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
128 ELECTROCHEMICAL NOISE MEASUREMENT
t ages cl ai med by the proponent s of the MEM are not al ways cl ear in corrosi on
i nvest i gat i ons.
2. Smoot h MEM spect ra are often rel at ed to a poor frequency resol ut i on. In addition, i f
a spectral line exi st s in the PSD, the frequency resol ut i on has to be i mproved by
i ncreasi ng the order of the MEM, whi ch det eri orat es the smoot hness.
3. Measurement s in t he very l ow-frequency range shoul d be made prudent l y because t he
MEM al ways l eads to a l ow-frequency pl at eau for suffi ci ent l y l ow frequencies.
4. The over l appi ng of MEM spect ra for several frequency bandwi dt hs can be difficult
wi t h sharp, st ruct ured spectra.
5. In some cases, the MEM is the onl y appl i cabl e met hod when onl y one t i me record is
avai l abl e, but, the result has to be careful l y anal yzed, especi al l y in the case of an
unst abl e corrosi on process. The resul t can depend on the specific t i me record used to
cal cul at e the PSD.
6. Concerni ng the corrosi on of mol ybdenum i mpl ant ed stainless steel:
a. a prot ect i ve surface l ayer was f or med as a result of ferritic t ransformat i on and mo-
l ybdenum enri chment , whi ch is di fferent from the mechani sm for prot ect i on obt ai ned
wi t h bul k al l oyed mol ybdenum- bear i ng stainless steels.
b. The i mpl ant at i on decr eased the prepi t t i ng activity, but, nevert hel ess, a number of
pi t s devel oped that perforat ed the prot ect i ve layer, whi ch was in cont rast wi t h the
t ype 304 st ai nl ess steel.
c. It seems that a syst emat i c rel at i onshi p bet ween the l evel of the prepi t t i ng act i vi t y
and the pi t densi t y observed aft er some t i me per i od is quest i onabl e. A l i mi t ed num-
ber of severe pi t s may somet i mes coexi st with an al most noi sel ess passi ve sampl e.
Re f e r e nc e s
[1] Gabrielli, C., Huet, F., Keddam, M., and Oltra, R., Corrosion, Vol. 46, 1990, p. 266.
[2] Gabrielli, C., Huet, E, Keddam, M., Oltra, R., and Pallotta, C., "Stochastic Aspects of Mechanical
and Chemical Breakdown of Passivity," in Passivity of Metals and Semiconductors, M. Froment,
Ed., Elsevier Science Publication, Amsterdam, 1983, p. 293.
[3] Burg, J. P., "Maximum Entropy Spectral Analysis," in Proceedings of the 37th Meeting of the
Society of Exploration Geophysicists, Oklahoma City, Oct. 1967.
[4] Andersen, N. O., Geophysics, Vol. 39, 1974, p. 69.
[5] Kay, S. M. and Marple, S. L., "Spectrum Analysis, a Modem Perspective," Proceedings of the
IEEE, Vol. 69, 1981, p. 1380.
[6] Uruchurtu, J. C. and Dawson, J. L., Corrosion, Vol. 43, 1987, p. 19.
[7] Searson, P. C. and Dawson, J. L, Journal of the Electrochemical Society, Vol. 135, 1988, p. 1908.
[8] Mansfeld, E and Xiao, H., Journal of the Electrochemical Society, Vol. 140, 1993, p. 2205.
[9] Bendat, J. S. and Piersol, A. G., Engineering Applications of Correlation and Spectral Analysis,
John Wiley and Sons, New York, 1980.
[10] Beaunier, L., Pillier, E, and Beaunier, R, "Phase Transformation in Molybdenum Surface Implanted
Stainless Steel: Microstructural Corrosion Characteristics," in Proceedings of the 6th International
Congress on Intergranular and Interphase Boundaries in Materials, Vol. 126-128, Ph. Komninou
and A. Rocher, Eds., Thessaloniki (Greece), Materials Science Forum, Trans Tech Publications,
Switzerland, 1993, p. 507.
[ll] Gabrielli, C., Huet, E, and Keddam, M., Electrochimica Acta, Vol. 31, 1986, p. 1025.
[12] Beaunier, L., "Surface Modification by Molybdenum Ions for Corrosion Protection," in Proceedings
of the 3rd Euromat European Conference on Advanced Materials and Processes, Paris (France),
1993, p. 356.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
Andr a~ L egat ~ and E dvar d Govekar ~
A Comparison of Spectral and Chaotic
Analysis of Electrochemical Noise
REFERENCE: Legat, A. and Govekar, E., "A Comparison of Spectral and Chaotic Anal-
ysis of Electrochemical Noise," Electrochemical Noise Measurement f or Corrosion Applica-
tions, ASTMSTP 1277, Jeffery R. Kearns, John R. Scutly, Pierre R. Roberge, David L. Reichert,
and John L. Dawson, Eds., American Society for Testing and Materials, 1996, pp. 129-141.
ABSTRACT: Potential and current fluctuations spontaneously generated by corrosion reactions
are known as electrochemical noise. In certain cases, good correlation can be obtained between
the results of the spectral analysis of electrochemical noise and corrosion rate and type. How-
ever, because of the chaotic nature of corrosion processes, a special mathematical treatment
may be needed. In the present study, the electrochemical noise measured on various metals
was treated by methods known from the theory of chaos, and the results were compared with
the results of spectral analysis. It has been shown that the chaotic characteristics of electro-
chemical noise are related to corrosion type, whereas the rate of corrosion has no influence on
the fractal dimensions of the noise.
KEYWORDS: electrochemical noise measurement, power-spectrum density, chaos, fractal
analysis
Of all t he met hods of cor r osi on measurement , onl y a few are abl e to di st i ngui sh di fferent
t ypes of corrosi on. Most of t hem (pol ari zat i on curves, el ect r ochemi cal i mpedance spect ros-
copy) use an ext ernal vol t age or current source; for this reason, corrosi on processes can be
di st urbed, and the resul t s may not be perfect l y rel i abl e. On t he ot her hand, el ect rochemi cal
noi se consi st s of spont aneousl y generat ed pot ent i al and current fluctuations, whi ch can be
measur ed in freel y cor r odi ng syst ems.
The phenomenon of el ect r ochemi cal noise, part i cul arl y its measur ement as a possi bl e
par amet er for a cor r osi on- moni t or i ng met hod, has recei ved consi der abl e at t ent i on over t he
l ast 20 years. In earl y exami nat i ons of el ect r ochemi cal noi se measur ement ( ENM) at open
ci rcui t pot ent i al , onl y one el ect rochemi cal quant i t y was measur ed and anal yzed [1- 3],
whereas most l at er st udi es st ri ct l y di st i ngui shed bet ween el ect r ochemi cal pot ent i al and cur-
rent noi se [4,5]. Anal yses of ENM wer e per f or med not onl y in t he i nvest i gat i ons of vari ous
t ypes of corrosi on (uni form [41, pi t t i ng [61, and crevi ce corrosi on [7]), but al so in the eval -
uat i on of corrosi on i nhi bi t ors [8] and organi c coat i ngs [9]. In several studies, t he ENM resul t s
were compar ed t o t he resul t s of pol ari zat i on curves, l i near pol ari zat i on resi st ance (LPRM),
and el ect r ochemi cal i mpedance spect r oscopy (ELM). It has been confi rmed that ENM enabl es
t he det er mi nat i on of t he rat e of cor r osi on [10-12] and t he r ecogni t i on of the t ype of corrosi on
[5,13]. Two mai n met hods of anal ysi s of measured el ect r ochemi cal noi se were i mpl ement ed:
Researcher, Institute for Testing and Research in Materials and Structures, Dimi~eva 12, 61000
Ljubljana, Slovenia.
2 Researcher, Faculty of Mechanical Engineering University of Ljubljana, Agker~eva 6, 61000 Lju-
bljana, Slovenia.
Copyrighl 9 1996 by ASTM lntcrnational
129
www.astm.org
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
130 ELECTROCHEMICAL NOISE MEASUREMENT
statistical (root mean square) and spectral (power-spectrum density). Over the last five years,
ENM has been used for corrosion monitoring in various industrial systems, especially in
electric power plants, both in situ [14-16] and under simulated conditions [17-20]. Although
several studies have been carried out, as yet, no unified theory exists for the sources of ENM.
Over the last 25 years, in nonlinear dynami c theory, some new mathematical parameters
have been developed to characterize nonperiodic and chaotic systems [21], such as: the
reconstruction of attractors [22,23], fractal dimensions [24-26], and Lyapunov exponents
[27,28]. The theory of chaos has been also used in chemistry [29], particularly to analyze
electrochemical processes [30-35]. Various chaotic parameters of current oscillations have
been estimated to describe the electrodissolution of copper [30,31] and iron [32], and the
initiation of localized corrosion [33], as well as to simulate the evolution of pit on silver
[34]. An excellent review of the applications of deterministic chaos theory to corrosion was
presented by Stringer and Markworth [35].
It was established by Mandelbrot that the power law ( l / f " ) of power spectrum may in-
dicate the chaotic nature of a time series, where the exponent c~ is considered to be an index
of its irregularity [36]. The results of the spectral analysis of measured electrochemical noise
usually indicate the continuous broadband nature of fluctuations. Whereas this t ype of spec-
t rum is characteristic for random and chaotic signals, rescaled range analysis has been used
to extend the statistical approach of electrochemical voltage noise [37,38]. Fractal dimensions
of voltage and current noise measured on stainless steel have been calculated to estimate the
dimensionality of the phase space of different types of corrosion [39]. The aim of the present
study has been to apply the theory of chaos to ENM for various metals to improve corrosion-
monitoring techniques based on the spectral analysis.
Experimental Procedure
Electrochemical potential and current noise were measured in a freely corrodi ng system.
The probe contained three identical metal rods encapsulated in an epoxy tube so that onl y
the transverse surface of each was exposed t o the electrolyte. The first pair of electrodes
formed a voltage source, and the second pair a current source, with one electrode in common.
The measuri ng system consisted of two multimeters (HP 3457A) connect ed via an I EEE 488
bus to a personal computer. The input impedances were 100 ~ for current input and 109 1~
for voltage input. Resolution was 1 nA for current and 1 i~V for voltage measurements. The
sampling rate was 500 mHz. In 1 test record, 4096 voltage and current values were recorded.
Measurements of electrochemical noise were made on various metals (carbon steel AI SI
C1008, stainless steel AI SI 302, and pure aluminum 99.99%) in aqueous solutions of various
pH and conductivity values (deionized water, 0.1% sodium hydroxide [NaOH], 0.1% sodium
chloride [NaC1], 0.5% NaCI, 3.5% NaCI, 0.1% HCI, 5% HC1, 0.1% sulfuric acid [H2SO4],
5% H2SO~) to generate a wide spectrum of corrosion rates and types. All tests were per-
formed at constant temperature (25~ and mixing of the electrolyte, so that the level of
dissolved oxygen remained nearly unchanged. Electrode surface areas varied from 7.1 to
50.2 mm 2. To determine the corrosion type and rate, various reference methods were used:
visual inspection, scanning electron mi croscope (SEM) analysis (JEOL 840A), and polari-
zation curves (EG&G PARC 273).
Results and Discussion
It has been shown in previous studies [12,13] that the spectral analysis of electrochemical
noise may indicate the rate and type of corrosion. The power-spectrum density of electro-
chemical current noise (PSD(1) in the frequency interval from 10 to 500 mHz) can be related
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
LEGAT AND GOVEKAR ON SPECTRAL AND CHAOTIC ANALYSIS 131
to the general corrosion rate, whereas the slope of the power-spectrum density of electro-
chemical voltage noise on a log versus l og scale (the parameter et being obtained from the
1/ ff function) has been recognized as a significant parameter for distinguishing among dif-
ferent types of corrosion. To improve the pattern recognition met hod for distinguishing dif-
ferent types of corrosion, the attractors of potential and current noise were reconstructed and
their fractal dimensions were estimated.
Three typical forms of electrochemical voltage and current fluctuations were generated by
the various types of metal corrosion in the chosen electrolytes. The passivation process was
observed on carbon steel and stainless steel in 0.1% NaOH and on aluminum in deionized
water (Fig. 1). With the exception of the high initial values of voltage and current as a result
of immersion, the values of measured noise were very low (Fig. 2.). In some cases, the
current noise was below measurement limits, which was defined by instrumental noise (see
Fig. 2).
Uni form corrosion was detected on the stainless steel in 5% H2SO4 (Figs. 3 and 4),
whereas the aluminum corroded uni forml y and intensively in 0.1% NaOH (see Fig. 4). The
electrochemical noise generated by uniform corrosion consisted of very frequent voltage and
current transients appearing in the frequency domai n much like white noise (Fig. 5). The
mean values of measured voltage and current were close to zero, but the amplitude of
fluctuations was much higher than during the passivation process.
Pitting corrosion was observed on the stainless steel and aluminum in 3.5% NaC1 (Figs.
6 and 7). The fluctuations of voltage and current were less frequent than during uni form
corrosion. As can also be seen in Fig. 8, the rise time of voltage fluctuations was, in general,
shorter than the descent time; however, for current transients the situation was the opposite.
The power-spectrum densities of potential and current noise could be interpolated by a 1/ ff
function on a log versus l og scale. The voltage and current transients often occurred
simultaneously.
The electrochemical noise records measured during the different types of corrosion were
split into four shorter time series, transformed to the frequency domain, and averaged (see
Figs. 2, 5, and 8). The parameter (x was obtained by linear interpolation of the power-
spectrum densities on a l og versus log scale (Table 1). Additionally, the fractal dimensions
of the attractors of electrochemical noise were estimated and compared with the results of
the spectral analysis.
The behavior of a dynami c system is best described in its phase space, a coordinate system
whose coordinates are the variables necessary to describe completely the state of the system
0 , 6 J i I I ) I H I i h l l l r l l J ~ ~ ' ~ ] H I H J J ' ~ - - T i ~ ] T F 0 ' I ] I T H 1 ] T " ~ ~ I H I T 1 F T T T T i l ~ r r J u m 1 T F r T n ~
|
0. 4 Q ) ~ J - ~ . . . . . . q -0.2 I | b )
- 0 . 4 /
-0. 2
~0~6
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . i
10-2 10-1 101 10 2 )-3 10-2 10-1 10 o 101 0 2
i / laA c m -2 i / p A c m - 2
FIG. l--Polarization curves measured on: (a) stainless steel AISI 302 (area = 19.6 mm 2) in 0.1%
NaOH and (b) pure aluminum 99.99% (area = 50.1 mm 2) in deionized water.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
132 ELECTROCHEMICAL NOISE MEASUREMENT
-20
-60
2 ' o 4 0
. . . . . L . ~ j
80 I{D 120
0 . . . . r - - i , i J , i i i ~ r , l , . ~ T - l ~ 1 1 q T i I ' T
-20" a
' ~
-8 0 i
1 [ a J _ LJ- t \
20 d
0 ~ -80
-100
<= -zo - ~' -azo
-40 b b
-140
-60 d ta
-80 -150 c ~ .
-1000 20 40 60 80 100 120 - 1% _z , , 10 ~ 101 102 , ~rl03
TIME/min log FREQUENCY/mHz
FIG. 2--Electrochemical noise generated by an alloy in a passive condition: (a) stainless steel AISI
302 (area = 19.6 mm 2) in 0. 1% NaOH, (b) pure al umi num 99. 99% (area = 50.1 mm 2) in dei oni zed
water, (c) pure al umi num 99. 99% (area = 7.1 mm 2) in dei oni zed water, and (d) i nst rument al noise.
at any moment . To each possi bl e state of t he syst em a poi nt corresponds in the phase syst em,
whereas the evol ut i on of states is r epr esent ed by a trajectory. A phase port rai t is a graph
depi ct i ng the evol ut i on of the syst em from di fferent i ni t i al condi t i ons. The final state or the
equi l i bri um state of the syst em is model ed by l i mi t sets al so cal l ed attractors.
It has been shown by Packar d et al. [22] and Takens [23] that the phase space can be
appr oxi mat ed by usi ng a r ecor d of a si ngl e observabl e. Thi s procedure, known as the t i me
del ay met hod, has been used by the authors to reconst ruct the attractors of t he measured
el ect rochemi cal noise. The poi nt s of at t ract or of a scal ar t i me series X ( t 3 [in our case E( t )
and l(t)] in D E di mensi onal phase space were deri ved by Eq 1. 9 is the t i me del ay mul t i pl e
of the sampl i ng t i me, and DE is t he embeddi ng di mensi on of the reconst ruct ed phase space,
whi ch shoul d be pr oper l y chosen. The values of x used in reconst ruct i ng the at t ract ors were
det er mi ned by the i nt egral l ocal def or mat i on met hod based on the condi t i on that t wo di st i nct
t raj ect ori es shoul d not i nt ersect [40].
X( t i ) = { [X(t), X(t~ + -r)] . . . X[t~ + ( D E - 1) 9 "r]} ( l )
For a graphi c present at i on of attractors, D E is usual l y chosen to be 2. In this case, Eq 1
is si mpl i fi ed, and the phase- space pl ot s are reconst ruct ed as X( t i ) versus X( t i - "0. The t wo-
di mensi onal phase space pl ot s of the el ect rochemi cal noi se generat ed by di fferent t ypes of
corrosi on can be seen in Fi g. 9. The at t ract or of pot ent i al noi se measured duri ng the passi -
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
LEGAT AND GOVEKAR ON SPECTRAL AND CHAOTIC ANALYSIS 133
FIG. 3- - SEM micrograph of stainless steel AISI 302 after exposure to 5% H2SO 4.
vation process is nearly a straight line (see Fig. 9a-I), whereas the phase-space plots of
current are dispersed as a result of instrumental noise (see Fig. 9a-II). The attractors of
electrochemical noise generated by uniform corrosion formed disordered sets of points (see
Fig. 9b). The two-dimensional phase space plots of electrochemical noise measured during
pitting corrosion consisted of multiple loops (see Fig. 9c).
From Fig. 9 it is evident that different types of electrochemical noise form different types
of attractors. However, for the characterization of the structure of these attractors, the fractal
dimensions have to be estimated. The dimensionality of an attractor also has an important
physical meani ng- - i t is equal t o the minimum number of variables present in the evolution
of a corresponding dynamic system (degree of freedom). Various types of fractal dimensions
o
> -0.2
-0.4
-0.6
0. 4 ~ 1 1 i i i i i , J ~ ~ 1 i i i i i i i i r ~ rrerT - ~r Tl ~l l 1
a) / ~\ \
0.2 f
-0.8
10-1 10 o
-0.6
b)
101 102 103 104 105 10-1 10 o 101 102 103 104 105
i / FA cm "2 i / pA cm -2
FIG. 4--Polarization curves measured on: (a) stainless steel AISI 302 (area = I9. 6 mm 2) in 5%
1-12804 and (b) pure aluminum 99.99% (area = 50.1 mm 2) in 0.1% NaOH.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
1 3 4 ELECTROCHEMICAL NOISE MEASUREMENT
>
20
10
-10
0 20 40 60 80 100 120
0 ~ - i i i i , l l l l i - ~ r T x z r ~ i l l l l l ~ - - , r E I p l I ,
J
-20 i o
I
~ 8 f
-100
- I - - i i i i u 2 z J _ ~ _ _ l i i i l l ' J J ~ _ ~ I i I i i [ i i _ l i i l i l
100 l0 t 102 103
3 ~ I
2
1
<
- 1
-2
-80
-100
<
-120
t2
-140
-3 -160
-4u 0 " 1~ ~' ~u- I 100 ' ~ J - ~ J ' ~
20 40 60 80 100 120 101 102 103
T I M E / m i n FREQUENCY/mHz
FIG. 5--Electrochemical noise generated by uniform corrosion: (a) stainless steel AISI 302 (area =
19.6 mm 2) in 5% H2SO 4 and (b) pure aluminum 99.99% (area = 50.1 mm 2) in 0.1% NaOH.
are known, but for application to real data, particularly for hi gher-di mensi onal phase spaces,
the so- cal l ed correlation di mensi on is wi del y used. The procedure for cal cul ati ng the cor-
relation di mensi on (descri bed in detail in Re f 24) was used. Gi ven the vector ti me series
X(t3, the correlation functi on Co~ is defined as:
1 ~ 0{ r - IlX(ti) - X(t)H} (2)
C~ = M 2 - M , , /
where 0 is the Heavi si de step function, M is the number o f poi nts in the vector ti me series
X(tl ), and the vertical bars indicate the norm o f the vector. The summat i on counts the number
o f pairs o f points in the attractor for whi ch the di stance IPf(ti) - X(t)lP is l ess than r. The
correlation di mensi on D2 at various embeddi ng di mensi ons DE can be obtai ned from the
expressi on:
D 2 = l i m l n C~ (3)
,.~o l n r
The correlation di mensi on D 2 is actual l y a measure o f the extent to whi ch the presence
o f a data poi nt affects the posi t i on o f other poi nts l yi ng on the attractor ("spatial correla-
ti on"). It can be estimated onl y i f Eq 3 has a val ue that i s limited by i ncreasi ng the embed-
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
LEGAT AND GOVEKAR ON SPECTRAL AND CHAOTIC ANALYSIS 135
FIG. 6- - SEM micrograph of stainless steel AISI 302 after exposure to 3.5% NaCI.
ding di mensi on D E of the reconst ruct ed phase space. The correl at i on funct i ons Co~ and the
correlation di mensi ons D2, cal cul at ed for attractors of el ect rochemi cal noise at di fferent em-
beddi ng di mensi ons D e, can be seen in Figs. 10, 11, and 12.
It is evi dent that the correl at i on di mensi ons D 2 of el ect rochemi cal noise measured during
pitting corrosi on converge by i ncreasi ng De (see Fig. 11), whereas for uni form corrosi on no
saturation is observed in D 2 (see Fig. 12). It can be concl uded that the el ect rochemi cal noise
generat ed by uni form corrosi on is random noise. The plot lnCD~ versus l nr for el ect rochem-
ical noi se in the passivation process has two scaling regions, whi ch are also reflected in the
cal cul at i on of D2 (see Fig. 10). Such characteristics of CD~ are typical for l ow-di mensi onal
>
0.6
0.4
0.2
0
-0.2
-0.4
10-1
i i i i i i i i J ~ i i i i l f l I
10 ~ 10 ] I(P 103 104 l0 s
i / pA cm -2
0 ~ ~ i i i , t l ,
-o.2 b )
-0.4
;>
-0.6
-0.8
-1
E , r ~ r r r ~ l l . . , , , , i Ll . ~ , , H, I . , l i Hi n
. ~ [ i l l l l i
1~0.1 10 ~ t~.L.~_~.,,,,,.lO 1 10 a ,, ,,,,,,,10 3 ,, ,,J,,.lO 4 , , ,,~0 s
i / pA cm -2
FIG. 7--Polarization curves measured on: (a) stainless steel AISI 302 (area = 19.6 mm 2) in 3.5%
NaCl and (b) pure aluminum 99.99% (area = 50.1 mm 2) in 3.5% NaCL
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
136 ELECTROCHEMICAL NOISE MEASUREMENT
100
50
> -50
E
-aso L I
-250 20 40 60 80 10O 120
~ a
- 4 0 ~ - - ~
-60 ] ~ ~ ' ~
-80
l
.lOO ~
I
-t 10 o 101 10 2 10 3
10
200
100
0
<
-100
-20O
-30O 0
i L
20
r . . . . . . i i - - 7
!
J
4'0 f~0 8'0 100 1 89
TIME/min
t
-80 }
-100 ~-
-120 1
FREQUENCY/mHz
FIG. 8--El ect rochemi cal noise generated by pitting corrosion: ( a) stainless steel AI SI 302 (area =
19.6 mm 2) in 3.5% NaCI and ( b ) pure aluminum 99.99% (area = 50.1 mm 2) in 3.5% NaCl.
syst ems wi th an external noi s e that di ffuses the structure o f the attractor [21]. It can be
recogni zed, in general, that D2 o f current noi se s l owl y reaches a l i mi ted val ue than D2 o f
potential noi se (see Fig. 11). However, it is not known whether this i s i nfl uenced by the
different nature o f potential and current noi se, or by instrumental noi s e (a l ower s i gnal / noi s e
ratio for current ( see Fig. 2).
The correlation and embeddi ng di mensi ons o f el ect rochemi cal vol t age and current noi se
generated by different t ypes and rates o f corrosi on is summari zed in Table 2. For each
combi nat i on o f metal and sol uti on, ten pol ari zati on curves were recorded. The val ue o f icor~
was obtai ned as the average o f current densi ti es calculated by Tafel extrapolation. It i s evi dent
that the general corrosi on rate did not affect the fractal di mensi ons.
T A B L E l - - Sl ope s of power-spectrum densities and capacity f ract al dimensions of electrochemical
noise in different types of corrosion (minimum and maximum values).
e t E/ d BV % / d B A
Ty p e of Corros i on decade -~ decade J Do( E) Do( / )
Passi ve - 2 0 - 15 - 1 1 1. 09 1. 22 l . 10 1. 30
Pi tti ng - 3 0 - 2 0 - 15 - 10 1. 35 1. 45 1. 24 1. 38
Uni f o r m - 7 2 - 6 1 1. 68 1. 83 1. 69 1. 79
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
L E G A T A N D G O V E K A R O N S P E C T R A L A N D C H A O T I C A N A L Y S I S 137
9 L ~ i l
" , . , . . , ' . < , ~
. , . , , . , , + : ~
\ ;? + : j
L L , , ~ L L
0 0 0 0
t I
( ~ - - + 1 3
9 , . " 7 ; ~ : ~ "
, " .'":".C ~" ' ~i , ~" ~' -"
t I [ J
0 ~ 0
I
( 1 - ~ ) 3
- - T ~ I I - -
I _ _ i i I
0 cO ~ ~'
( : ~ - ~ ) 3
o
I
0
C~
I
0
o . I
o ~,
C~
o
0
I . ' C )
9
I
0 4
0
C~
I 1 ~ I ]
. ' j . + .
: , . . : : . ; . ~ . ~ : L . r . , . . . . . . . . . . . .
l
" i + 9
I ,
2
0 0 0 0
0 0 0
I I
(3. -), ) I
I I
r , , ,
. / . . , . ' + + 9 .
L 1 I
0 0 0 0
I I
( i - ~) i
I
"G
I
0 0
8
I
A
t ~
Oi
03
I
0 ~
o I
I
, + ~
~, tr
v N . , , ~
II
" e ~
b
2 ~
- 5 '
} , , .
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
138 ELECTROCHEMICAL NOISE MEASUREMENT
0
- 5
- q - 1 0
- 1 5
- 1 0
1 0 - -
8
6
S
a)
i i i I
- 8 - 6 - 4
I n r
0 . . . .
- 2
- 4
m - 6
-8 1
- 1 0 . . . . ~ . . . . 9 . . . . ~ - L __ _
- - 4 - 3 - 2 - 1 0
I n r
c}
"Z"
cY
J
- 8 - 6 - 4 - 2 0 5
I n r
d)
~3
- 4 - 3 - 2 - 1
I n r
FIG. lO---Correlation functions and correlation dimensions of electrochemical noise (measured on
alloy in a passive condition--pure aluminum 99.99% in deionized water) at various embedding dimen-
sions (DE from 2 to 10, step 1): (a) CI~ (r) of potential noise, (b) CD~ (r) of current noise, (c) D 2 (r) of
potential noise, and (d) D 2 (r) of current noise.
The met hod of estimating the correlation dimension of measured electrochemical noise
can be a reliable basis for the determination of corrosion type. Unfortunately, the procedure
is very t i me-consumi ng [41], so it is unsuitable for corrosion-monitoring systems. To reduce
the complexity of the calculations, a simpler met hod has been implemented. The box-count-
ing procedure was applied to the measured signals, presented as two-dimensional geometrical
objects, t o estimate the capacity fractal dimension [42].
Let N(~) be the mi ni mum number of 8-mesh squares needed to cover the graph [in this
ease E(t) or I(0]. The capacity fractal dimension, denoted by Do, can be obtained by taking
the limit as 8 approaches zero:
Do = lim l nN(8) (4)
~. o I n( l ~8)
A comparison between the slopes of the power-spectrum densities on a l og versus log
scale, and the estimated capacity fractal dimensions of the measured electrochemical noise,
can be seen in Table 1. The parameters a and D O were calculated for different types of
corrosion of the metals (carbon steel AI SI C1008, stainless steel AI SI 302, and pure alu-
mi num 99. 99%) and onl y mi ni mum and maxi mum values of estimate values are presented.
It is evident that capacity fractal di mensi on Do is related t o the parameter a obtained by the
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
LEGAT AND GOVEKAR ON SPECTRAL AND CHAOTIC ANALYSIS 139
0
-5
U
dO
-15
o) T
D E = 2 ~ /
-10 -8 -6 -4 -2
In r
U
f
b) T O ~
In r
5 1 i i T
c)
4 \ ~
1 DE= 2 ~ ~ _ z
d)
DE = 2
~ 0 . . . . . . . . . . . . . .
-8 -6 -4 -2 0
In r In r
FI G. 1 1 - - C o r r e l a t i o n functions and correlation dimensions of electrochemical noise (measured during
uniform corrosion--pure aluminum 99.99% in 0.1% NaOH) at various embedding dimensions (DE from
6 to 24, step 2): ( a) C ~ (r) of potential noise, ( b) CDE (r) of current noise, ( c) D2 (r) of potential noise,
and ( d) D 2 ( r ) of current noisel
spectral analysis; moreover, the fractal dimension was recognized as an even more accurate
parameter distinguishing among different types of corrosion. It can also be seen from Table
1 that Do, when estimated from potential noise, enables the classification of the different
types of corrosion without the overlapping of intervals, as opposed to current noise in which
some overlapping occurs.
Conc l us i ons
The analysis of electrochemical noise by mathematical tools known from the theory of
chaos enables the distinction of different types of corrosion. The correlation dimension in-
dicates that the dimensionality of the phase space of electrochemical noise increases with
the mode of corrosion from passivation, pitting, to uniform corrosion. From the chaotic
analysis of electrochemical noise, it can be assumed that potential noise contains more in-
formation about the form of corrosion than current noise.
The capacity dimension of measured electrochemical noise was also estimated and has
been recognized as a reliable parameter for the determination of the type of corrosion. The
corrosion rate can be estimated by the spectral analysis, but it has no influence on the fractal
dimension of electrochemical noise. Thus, spectral and chaotic treatment can be used as
compl ement ary methods in the analysis of electrochemical noise for corrosion monitoring.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
140 ELECTROCHEMICAL NOISE MEASUREMENT
S
n
u [ O ) . . . . . . . l - - -
!
-5 j DE:
i
-10)
r
) // / / / /
V , / ,
/ / / /
5 / /
L _ _ t 9
-10 -8 -6 -4 -2 0
- ' ~ ":)
- l o - 8 - 6 4 - i
In r In r
] 5 c) ' /\ 24 15i d) '
!
10 10
5 - ~ 0 5
DE: 6
0-10 -6 -4 -2 -01O -~8 & -4 -2 0
In r In r
FIG. 12--Correlation function and correlation dimension of electrochemical noise (measured during
pitting corrosion--pure aluminum 99.99% in 3.5% NaCl) at various embedding dimensions (DE from 2
to 10, step 1): (a) CD~ (r) of potential noise, (b) Ct~ (r) of current noise, (c) D 2 (r) of potential noise,
and (d) D2 (r) of current noise.
TABLE 2--Correlation and embedding dimensions of electrochemical noise in different types
of corrosion.
Type of Corrosion Metal ico~/~A cm -: D2(E) De(E) D2(/) De(/)
Passive stainl ess steel AI SI 302 0.4 1.35 8 >10 >26
Passive al uminum ( 99.99%) 0.4 1.15 4 >10 >10
Pitting stainl ess steel AI SI 302 1 1.8 7 2.5 10
Pitting al uminum (99.99%) 2 1.8 5 4.2 9
Unif orm stainl ess steel AI SI 302 40 >12 >26 >9 >26
Unif orm stainl ess steel AI SI 302 80 >12 >26 >10 >26
Unif orm al uminum (99.99%) 100 >12 >24 >14 >24
Unif orm carbon steel AI SI C1008 >1000 >11 >26 >12 >24
Acknowledgment
The au t hors t hank al l r evi ewer s f or t hei r hel p f u l comment s, w hi ch have enabl ed t he au t hors
to l ook at t he ex p er i ment al resu l t s wi t h a mor e cri t i cal ey e and to cl ari f y t he concl u si ons.
Ref erences
[1] Barker, G. C., Journal of Electroanalytical Chemistry, Vol. 21, 1969, p. 127.
[2] Hladky, K. and Dawson, J. L., Corrosion Science, Vol. 21, 1981, p. 317.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
LEGAT AND GOVEKAR ON SPECTRAL AND CHAOTIC ANALYSIS 141
[3] Hladky, K. and Dawson, J. L., Corrosion Science, Vol. 22, 1982, p. 231.
[4] A1-Zanki, I. A., Gill, J. S., and Dawson, J. L., Electrochemical Methods in Corrosion Research,
Materials Science Forum, Vol. 8, 1986, p. 463.
[5] Eden, D. A. and Rothwell, A. N., in Proceedings Corrosion/92, No. 292, NACE International,
Houston, 1992.
[6] Uruchurtu, J. C. and Dawson, J. L., Corrosion, Vol. 43, 1987, p. 19.
[7] Simoes, A. M. P. and Ferreira, M. G. S., British Corrosion Journal, Vol. 22, 1987, p. 21.
[8] Monticelli, C., Brunoro, G., Frignani, A., and Trabanelli, G., Journal of the ElectrochemicalSociety,
Vol. 139, 1992, p. 706.
[9] Xiao, H. and Mansfeld, E, Journal of the Electrochemical Society, Vol. 141, 1994, p. 2332.
[10] Lumsden, J. B., Kendig, M. W., and Jeanjaquet, S., in Proceedings Corrosion~92, No. 294, NACE
International, Houston, 1992.
[11] Mansfeld, E and Xiao, H., Journal of the Electrochemical Society, Vol. 140, 1993, p. 2205.
[12] Legat, A. and Zevnik, C., Corrosion Science, Vol. 35, 1993, p. 1661.
[13] Legat, A. and Dole~ek, V., in Proceedings Corrosion~94, No. 319, NACE International, Baltimore,
1994.
[14] Doherty, P. E., Moore, D. C. A., Cox, W. M., and Dawson, J. L., in Proceedings of Environmental
Degradation of Materials in Nuclear Power Systems-Water Reactors, Jekyll Island, 1989, pp. 13-65.
[15] Syrett, B. C., Corrosion Science, Vol. 35, 1993, p. 1189.
[16] Farrell, D., in Proceedings of the 12 'h International Corrosion Congress, Vol. 6, NACE International,
Houston, 1993, p. 4131.
[17] Winters, M. A., Stokes, P. S. N., Zuniga, P. O., and Schlottenmeier, D. J., Corrosion Science, Vol.
35, 1993, p. 1667.
[18] Brennestuhl, A. M., in Proceedings of the 12 th International Corrosion Congress, Vol. 6, NACE
International, Houston, 1993, p. 4102.
[19] Legat, A., in Proceedings of the 12 th International Corrosion Congress, Vol. 3, NACE International,
Houston, 1993, p. 1410.
[20] Liu, C., Macdonald, D. D., Medina, E., Villa, J. J., and Bueno, J. M., Corrosion, Vol. 50, 1994, p.
687.
[21] Tsonis, A. A., Chaos--From Theory to Applications, Plenum Press, New York, 1992.
[22] Packard, N. H., Crutchfield, J. P., Farmer, J. D., and Shaw, R. S., Physical Review Letters, Vol. 45,
1980, p. 712.
[23] Takens, F., in Dynamical Systems and Turbulence, Warwick 1980, D. A. Rand and L. S. Young,
Eds., Lecture Notes in Mathematics, Vol. 898, Springer-Verlag, New York, 1981, p. 366.
[24] Grassberger, P. and Procaccia, I., Physica D, Vol. 9, 1983, p. 189.
[25] Higuchi, T., Physica D, Vol. 31, 1988, p. 277.
[26] Theiler, J., Journal of the Optical Society of America, Vol. 7, 1990, p. 1055.
[27] Eckmann J.-P. and Ruelle, D., Reviews of Modern Physics, Vol. 57, 1985, p. 617.
[28] Broomhead, D. S. and King, G. P., Physica D, Vol. 20, 1986, p. 217.
[29] D. Avnir, Ed., The Fractal Approach to Heterogeneous Chemistry: Surfaces, Colloids, Polymers,
John Wiley & Sons, Chichester, Great Britain, 1989.
[30] Basset, M. R. and Hudson, J. L., Journal of Physical Chemistry, Vol. 93, 1989, p. 1989.
[31] Dewald, H. D,, Parmananda, P., and Rollins, R. W., Journal of the Electrochemical Society, Vol.
140, 1993, p. 1969.
[32] Li, W., Nobe, K., and Pearlstein, A. J., Journal of the Electrochemical Society, Vol. 140, 1993, p.
721.
[33] Sharland, S. M., Bishop, C. M., Balkwill, P. H., and Stewart, J., in Advances in Localized Corrosion,
H. S. Isaacs, U. Bertocci, J. Kruger, and S. Smialovska, Eds., NACE International, Orlando, 1987,
p. 109.
[34] Corcoran, S. and Sieradzki K., Journal of the Electrochemical Society, Vol. 139, 1992, p. 1568.
[35] Stringer, J. and Markworth, A. J., Corrosion Science, Vol. 35, 1993, p. 751.
[36] Mandelbrot, B. and van Ness, J. W., SlAM Review, Vol. 10, 1968, p. 422.
[37] Roberge, P. R., Journal of Applied Electrochemistry, Vol. 23, 1993, p. 1223.
[38] Roberge, P. R., Corrosion, Vol. 50, 1994, p. 502.
[39] Legat, A. and Govekar, E., Fractals, Vol. 2, 1994, p. 241.
[40] Buzug, T. and Pfister, G., Physical Review A, Vol. 45, 1992, p. 7073.
[41] Parker, T. S. and Chua, L. O., Practical Numerical Alghorithms f or Chaotic Systems, Springer-
Verlag, New York, 1989.
[42] Falconer, K., Fractal Geometry: Mathematical Foundations and Applications, John Wiley & Sons,
Chichester, Great Britain, 1990.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
Pi erre R. Rober ge I
Quantifying the Stochastic Behavior of
Electrochemical Noise Measurement During
the Corrosion of Aluminum
REFERENCE: Roberge, R R., "Quantifying the Stochastic Behavior of Electrochemical
Noise Measurement During the Corrosion of Aluminum," Electrochemical Noise Measure-
ment f or Corrosion Applications, ASTM STP 1277, Jeffery R. Kearns, John R. Scully, Pierre
R. Roberge, David L. Reichert, and John L. Dawson, Eds., American Society for Testing and
Materials, 1996, pp. 142-156.
ABSTRACT: This paper describes a study of electrochemical measurements that included the
analysis of potential noise fluctuations recorded during the free corrosion of a strain-hardened
aluminum alloy exposed to an aerated saline solution. The analysis of spontaneous electro-
chemical noise (EN) generated during these experiments was made by various analysis tech-
niques specifically designed to reveal the stochastic and fractal nature of signals and shapes.
The paper demonstrates how rescaled range analysis and stochastic process detector techniques
can serve to highlight features present in the frequency domain of transformed data records.
It is also shown that the corrosion cell made with aluminum exposed to an aqueous aerated
saline solution is an excellent source of noise, which can be easily varied by changing the pH
of the solution to force depassivation phenomena. The analysis of spontaneous EN by com-
plementary methods is demonstrated to be a practical approach for the identification of the
physical phenomena causing the EN. It is also demonstrated that monitoring the electrochem-
ical impedance spectroscopy behavior in parallel to EN can be valuable because it helps to
establish some fundamental correlations in real time and define the sensitivity of each technique
under similar conditions. In the present study, the drastic changes recorded by EN measure-
ments when the solution pH was slightly reduced are indicative of the tremendous sensitivity
of EN monitoring to changes in the corroding environment.
KEYWORDS: electrochemical noise, stochastic, chaos, depassivation, aluminum corrosion,
analysis methods
Many biological, chemical, and physical systems have been found to exhibit chaotic be-
havior [1-5]. The extensive development of techniques for studying electrochemical reactions
has similarly stimulated the study of fluctuating electrochemical processes that translate into
electrochemical noise (EN) [6]. In this context, the study of spontaneous current or potential
fluctuations for the characterization of corrosion processes has also received considerable
attention in recent years. The study of corrosion potential fluctuations was applied, for ex-
ample, to monitor the onset of events characterizing pitting [7-10] and stress corrosion
cracking corrosion [11-13]. During localized corrosion, EN seems to be generated by both
stochastic processes typically present during passivation breakdown and repassivation events
and deterministic processes such as film formation or pit propagation. In this paper, fluctu-
ations of the corrosion potential observed on freely corroding aluminum material exposed
Professor, Department of Chemistry and Chemical Engineering, Royal Military College, Kingston,
Ontario, Canada, K7K 5L0.
Copyrighl 9 1996 by ASTM lntcrnational
142
www.astm.org
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
ROBERGE ON QUANTIFYING STOCHASTIC BEHAVIOR OF ENM 143
to aerated saline solutions will be analyzed with techniques specifically designed to reveal
the stochastic and chaotic features that are present in EN signatures.
Noise and Chaos Analysis
The most common way to analyze EN data has been to transform time records in the
frequency domain to obtain power spectral density (PSD) plots. PSD plots can typically be
computed using fast Fourier transforms (FFT) [14] or other algorithms such as the maximum
entropy method (MEM) [15,16]. But quite a few other analytical tools have been developed
to detect and characterize hidden deterministic or chaotic features in seemingly random data
and reveal the properties of the equations that could represent fluctuating behavior. These
tools include calculations of the probability distribution, Lyapunov exponent, and various
measures of the fractal dimension. Many of these tools have recently become commercially
available [17]. Two other techniques, rescaled range scale (R/ S) analysis and stochastic
process detector (SPD), which are not yet available in software form, are included in the
present study because they have already been applied by the author in different corrosion
studies [18-21].
Chaos Dat a An a l y z e r
Although the concepts and models necessary to characterize chaotic behavior can be com-
plex, the association of low-dimensional chaotic dynamics with the presence of strange at-
tractors has permitted the quantification of chaos. Various methods provide the estimation
of properties of strange attractors. These attractors have been characterized, for example, for
their fractal presence in a spatial representation or for the degree of divergence of repeated
patterns using Lyapunov exponents. Owing to the ingenuity of many experimenters, the study
of deterministic chaos has undergone a transition from the derivation of purely mathematical
and theoretical results to the quantitative identification of chaotic features in experimental
data.
Re s c a l e d Range An a l y s i s
This approach, which is relatively new in electrochemistry, was derived from one of the
most useful mathematical models for analyzing time-series data that was proposed by Man-
delbrot and van Ness [22]. Although the fractional Brownian motion (fBm) model can help
to describe very complex geometries or time series, its analysis can be made relatively simply
by using the rescaled range analysis technique originally proposed by Hurst [23] and applied
by Mandelbrot and Wallis to the determination of the fractal characteristics of a time series
[24]. A detailed description of the R/ S technique (where R or R(t , s) stands for the sequential
range of the data points increments for a given lag s and time t, and S or S(t , s) for the square
root of the sample sequential variance) can be found in Fan et al. [25]. Hurst [23] and later
Mandelbrot and Wallis [24] proposed that the ratio R( t , s ) l ( t , s ) , also called the rescaled range,
is itself a random function with a scaling property described by Eq 1 in which the scaling
behavior of a signal is characterized by the Hurst exponent (H), which can vary between
0 < H< I .
R~,.,) ~ S H ( 1)
S ( t , s )
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
144 ELECTROCHEMICAL NOISE MEASUREMENT
It has additionally been shown [26] that the local fractal dimension dFi of a fBm noise
trace is related to H through Eq 2, which makes it possible to characterize the fractal di-
mension of given time series by simply calculating the slope of a R/ S plot.
d F ~ = 2 - H 0 < H< 1 (2)
Stochastic Process Detector
The SPD technique was designed to quantify the stochasticity of EN records as a function
of frequency. In the present study, this frequency was the sampling frequency, that is, 2 Hz.
The main idea behind the SPD technique is that specific corrosion modes, such as pitting or
other forms of localized corrosion, will potentially generate EN containing deterministic
features. It was hoped that by focusing on the stochasticity of EN records the technique
would be highly sensitive to any feature falling out of such category and reveal some infor-
mation not visible by any other analysis technique. The SPD technique, which has been
explained in details elsewhere [19], consists in two levels of EN transformation. First, EN
records are transformed into a series of singular events, that is, each point of a time series
is examined for its appurtenance to either positive or negative noise peaks. In the second
level of transformation, the distribution of peak lengths f ( t ) is examined and compared to
the theoretical exponential decay distribution, represented by Eq 3 where ),, is the mean
value and t is the peak length (time), that would characterize a series of stochastic events
[27]. The goodness-of-fit (GF) of real data by the exponential function is then calculated to
serve as a measure of stochasticity of the time records.
f ( t ) = X," e -a' (3)
The slope of the basic events (AV/ At ) is also comput ed by the SPD technique because
this parameter is thought to be an important characteristic of the electrochemical systems
being studied.
Experimental Procedure
The specimens were cut from a block of strain-hardened 5083-H131 aluminum alloy armor
material (thickness 5.0 cm; analyzed composition chromi um: 0.19, copper: 0.06, iron: 0.53,
magnesium: 4.1, manganese: 1.3, silicon: 0.50, zinc: 0.21) t o appropriate sizes for mount i ng
in epoxy according t o metallographic techniques. A pair of identical aluminum specimens
exposing their short transverse face was immersed in a 2-L beaker containing a solution of
3% sodium chloride. Before mounting, provisions were made for electrical connection to the
unexposed back of the samples and the unexposed edges were coated with an aluminum-
vinyl anticorrosive paint to prevent crevice corrosion between the epoxy mount and the
aluminum electrodes. After mounting, the specimens were polished up to a 600-grit finish
and cleaned with acetone and dichloromethane. The cell was equipped with an air purge and
a saturated calomel electrode (SCE) brought into close proximity with one electrode by a
Luggi n probe. The mount ed specimens were separated by 2.5 mm and kept in a stable
parallel position with plastic holders. The samples were exposed to the neutral aerated saline
solution for three days before the solution was acidified to a pH of ~4. 8 by adding small
increments of a 0.1 N hydrochloric (HC1) solution.
The reference electrode (SCE) served to measure the corrosion potential and its fluctua-
tions. The fluctuations were monitored through a high-pass filter (resistance (R) = 1 m~ in
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
ROBERGE ON QUANTIFYING STOCHASTIC BEHAVIOR OF ENM 145
series with a capacitor (C) -- 50 p,F and a lower frequency cutoff (f~) = 1/(2~rRC) that is,
3.2 mHz) to increase the sensitivity of the measurements. The high-sensitivity multimeter
used in these studies (Hewlett Packard Model 3457A) has a resolution of l 0 nV on its most
sensitive scale (30 mV). A sampling interval (At) of 0.5 s and a number of consecutive data
points (N) of 1800 were chosen because this gives a factual measuring frequency window
in which corrosion processes are dominant [28]. The frequency domain corresponding to
these sampling conditions can be evaluated to be between 1 Hz (fmax) and 1 mHz (fmi,) with
the help of Eqs 4 and 5 [29].
1
f"~' = 2A---t (4)
1
fmin = NAt (5)
The electrochemical impedance spectroscopy (EIS) measurements were performed with a
commercial generator/analyzer (Solartron Model 1255) at the corrosion potential (Eco~r), and
each measurement was repeated five times to ascertain the stability of the system being
evaluated. A frequency range of 100 to 0.1 Hz was selected for its sensitivity to corrosion
resistance. A potentiostat was not used in these measurements. The alternating current was
applied directly between the two aluminum electrodes and kept at a value that would not
FIG. 1--pH and parameters calculated from E1S data (corrosion rates, V, and Cdl , [~) during the
exposure of alloy 5083 specimens to an aerated 3% NaCl solution.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
146 ELECTROCHEMICAL NOISE MEASUREMENT
cause more than lO-mV difference (peak to peak) across the cell. The polarization resistance
(Rp) values were determined with the projection of centers analysis technique described
elsewhere [30-32].
Results and Di scussi on
The corrosion rates (CR) and the double-layer capacitance (Cdl) values calculated from
the EIS results obtained during the twelve-day experiment are presented in Fig. 1. The
transformation of EIS Rp values into corrosion rates was possible by using Eq 6, after cor-
recting for the two-electrode configuration (each electrode = Rp/2). In Eq 6, EW is the
equivalent weight of the corroding metal (8.99 g eq -t for aluminum alloys), d is the metal
density (2.7 g cm-3), A is the exposed surface area (cm2), and K is a constant that has a
value of 3.3 if Rp is in kft and the polarization resistance constant (B) is in mV. A B value
of 24 mV has been reported as an average value for aluminum alloys exposed to chloride
solutions [33].
K.EW.B
Corr. rate (p.m/y) = d- Rp 9 A (6)
The initial low corrosion rates of the aluminum specimens immersed in a neutral 3%
sodium chloride (NaCI) solution increased markedly when the pH was lowered during the
fourth day of the experiment. Lowering the pH is known to accelerate the breakdown of the
FIG. 2--E-pH diagram of the chemical species involved in the system O-H-AI with concentration of
soluble species set at 10 -5 M.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
ROBERGE ON QUANTIFYING STOCHASTIC BEHAVIOR OF ENM 147
protective film on aluminum and prevent self-repairing before finally leading to an acceler-
ation of localized corrosion [34]. Figure 2 illustrates the passive/corroding frontier of alu-
minum in the traditional form of a E-pH diagram computed from available thermodynamic
data [35] and with the concentrations of soluble species set at 10 -5 M. Note that the presence
of chloride has no effect on the predominance diagram of Fig. 2, but that the presence of
chlorides greatly accelerates the kinetics of the reactions involved [35].
When the solution pH was adjusted to 4.8, the corrosion potential temporarily moved to
slightly more anodic values (Fig. 3). Although the corrosion rates and double-layer capaci-
tance values calculated from the EIS results increased notably when the environment was
acidified, three times for CR and two times for Cdt, note that the average Eco~ values, expect
for the temporary anodic excursion after acidification, remained stable. But lowering the pH
dramatically increased the EN recorded at Er and even favored the production of potential
oscillations that could last for many hours. The oscillations encountered when the pH was
lowered were difficult to reproduce and could hardly be used for monitoring. But their
presence, even temporary, is indicative of the complexity of the competing corrosion pro-
cesses occurring at such metallic/environment interface.
Examples
Three EN records have been chosen to illustrate the different noise characteristics observed
during that experiment. The first file (Fig. 4) was sampled a few hours after immersion of
FIG. 3--Eco,-, and pH measured during the exposure of 5083 specimens to an aerated 3%
NaCl solution.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
1 4 8 ELECTROCHEMICAL NOISE MEASUREMENT
0.4
>
E
5n
O
r
II)
O)
O
0. 3
0. 2
0.1
0
-0.1
-0. 2
-0. 3 [ ' ~
-0.4 =
-0.5
0 100 200 300 460 500 6 ( ~ 0 700 860 900
Ti me/ s
FIG. 4 - - EN recorded during 15 min of the first day of the experiment with 5083 exposed to an aerated
3% NaCI solution.
the freshly polished specimens into the aerated saline solution. The level of noise activity at
that point was relatively small, the fluctuations slow or sluggish, and without specific fea-
tures. The PSD plot (Fig. 5) corresponding to this noise record shows a rapid decline of the
power density in the low-frequency portion of the PSD plot (between 10 3 and 10 -2 Hz)
and very small energy attributable to higher frequency values. The second example chosen
(Fig. 6) was sampled during the day that followed the pH adjustment. The tremendous level
of potential activity, which is visible by considering the scale of the potential fluctuations
recorded (10 mV peak to peak), translated into very high values on the corresponding PSD
plot (Fig. 5). The level of electrochemical activity was, in this second case, prevalent right
to the frequency domain corresponding to fmi, (10 2.5 HZ). The final example (Fig. 7) was
sampled on Day 9 of the experiment, at which point the EN was apparently dominated by
chaotic features. The PSD plot (Fig. 5) of the noise file indicated that a dominant frequency
component existed at approximately 0.016 Hz (10 -18 Hz) with a period around 1 min or
63 s.
The results obtained by analyzing the EN of the first example (see Fig. 4) with the SPD
method (Fig. 8) indicated a very poor fit (82%) of the exponential decay model at the
sampling frequency (2 Hz), whereas the R/ S analysis revealed a fractal dimension (D =
1.37) indicative of a small degree of persistence (or regularity). The fit (Fig. 9) made by the
SPD method was excellent (96.7%) for the example presented in Fig. 6, and the R/ S analysis
showed a fractal dimension (D = 1.65) indicative of antipersistence [25].
Although the results obtained with the SPD method (Fig. 10) applied to the third example
(see Fig. 7) appeared to fit the distribution of peaks tightly (GF of 95.7%), the distribution
of positive potential slopes presented an important anisotropy that was not paralleled on the
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
ROBERGE ON QUANTIFYING STOCHASTIC BEHAVIOR OF ENM 149
FIG. 5--Results of analysis by FFT of EN recorded during: Day 1 of experiment (Fig. 4); Day 4
(Fig. 6); and Day 9 (Fig. 7).
15-
~ I . . . . . . . . . . . . . . . . . . . . . . ~ I
> -10
-15 . . . . . . . . . . . . . . . . . . . . . . . . . . .
-20 j ~- - - . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
-250 100 260 3( ) 0 460 560 600 700 860 900
Time/s
FIG. 6--EN recorded during 15 min of Day 4 of the experiment with 5083 exposed to an aerated 3%
NaCI solution.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
150 ELECTROCHEMICAL NOISE MEASUREMENT
>
E
.oo
o
m
0
2
0.5
0-
-0.5
-1-
-1.5-
-2
0 160 200 300 400 5( ) 0 600 760 860 900
Time/s
FIG. 7- - EN recorded during 15 min of Day 9 of the experiment with 5083 exposed to an aerated 3%
NaCl solution.
FIG. 8--Resul t s of analysis by SPD of EN recorded during Day 1 of experiment (Fig. 4); histogram =
real distribution, line = calculated distribution, and hatched surface = potential slopes.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
ROBERGE ON QUANTIFYING STOCHASTIC BEHAVIOR OF ENM 151
350
300
O.
50
250
t -
O
"5 200
0
Q..
~ 150
|
Q..
100
0
.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5
3, 500
3,000
2, 500 m
>
2, 000
O.
O
1,500 a~
1,000 :~
5O0
Peak durat i on/ s
[ I I F i t t e d model [~Observed peaks [~Observed slopes 1
FIG. 9--Resul t s of analysis by SPD of EN recorded during Day 4 of experiment (Fig. 6): histogram =
real distribution, line = calculated distribution, and hatched surface = potential slopes.
negative side. This can be explained by the sharpness of the positive or anodic potential
excursions, visible in Fig. 7, symptomatic of a fast repassivation process following the much
slower depassivation of the partially protective surface. The R/S analysis performed on these
data clearly indicated the presence of persistence (D = 1.33), which is characteristic of the
presence of deterministic features.
Correl at i ons
The analysis of spontaneous EN by different and complementary methods, illustrated by
the previous examples, can serve to quantify the visible differences present in noise records.
In the present study, the drastic changes recorded by EN measurements, when the solution
pH was slightly reduced, are indicative of the tremendous sensitivity of noise monitoring to
changes in the corroding environment.
The data generated by the various techniques used in this experiment were analyzed for
the presence of correlations between the numerous parameters involved. For example, the
results obtained by monitoring the system with EIS seemed to correlate fairly well with the
pH measurements. The correlation coefficient calculated between CR and pH values was, in
fact, excellent (-0.96), indicating a simple inverse relation existed between pH values and
the corrosion rate. The Cat values also followed an inverse relation with the pH, but the fit
was slightly less, that is, correlation coefficient = -0.84. But, not surprisingly, the Eco =
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
250
200
o3
o~
Q .
~_ 100
50
.5 1 .5 2 2.5 3 3.5 4 4.5 5 5.5
300
152 ELECTROCHEMICAL NOISE MEASUREMENT
300
! 2 0
200 >
"-i
150 o
c r )
1 0 0 o
>
50
Peak durat i on/ s
I IIFi tted model [ ] Observed peaks [ ~ Observed slopes I
FIG. lO--Results of analysis by SPD of EN recorded during Day 9 of experiment (Fig. 7):
histogram = real distribution, line = calculated distribution, and hatched surface = potential slopes.
me a s ur e me nt s di d not cor r el at e wel l wi t h any of t hese par amet er s ( - 0 . 1 9 wi t h CR, - 0 . 3 4
wi t h Cdt, and 0. 18 wi t h t he pH).
The cor r el at i on coef f i ci ent s cal cul at ed be t we e n pH, CR, Ddt, and Eco,~ and s ome o f t he
par amet er s obt ai ned by anal yzi ng EN r ecor ds wi t h var i ous t echni ques ar e pr es ent ed i n Tabl e
1. Amo n g t he par amet er s eval uat ed by t he SPD met hod (GF, l / h , and pot ent i al sl opes) ,
none s e e me d t o cor r el at e ver y wel l wi t h any o f t he f our st andard el ect r ochemi cal par amet er s.
Thi s was not a bi g sur pr i se becaus e t he SPD was known t o be mos t l y sensi t i ve t o t r ansi t i ons
TABLE 1--Correlation coefficients calculated between standard electrochemical measurements and
parameters estimated from EN records.
Correlation Coefficient
Parameter GF a 1 / h a Slop@ dFi b CapD c Lyap d
pH - 0.17 0.03 0.45 - 0. 84 - 0. 44 - 0. 65
CR - 0. 20 - 0. 03 0.55 0.85 0.49 0.71
C~ - 0. 20 0.45 0.25 0.82 0.57 0.82
Eco ~ 0.17 - 0. 09 0.40 - 0. 21 - 0. 09 - 0. 17
Goodness-of-fit, mean, and potential slopes measured by SPD.
b Fractal dimension calculated by R/ S analysis (Eq 2).
c Capacity dimension calculated, with the Chaos Data Analyzer [17].
d Lyapunov exponent calculated with the Chaos Data Analyzer.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
ROBERGE ON QUANTIFYING STOCHASTIC BEHAVIOR OF ENM 153
between stochastic and deterministic processes that are themselves related to localized cor-
rosion problems [18,19]. Figure 11 illustrates the sensitivity of the potential slopes, for
example, to the modification of the pH done on Day 4.
Another feature revealed by the SPD method is the presence of anisotropy between the
anodic and cathodic potential excursions and between short and long duration peaks. This
phenomenon appears as differences in the calculated values presented in Fig. 11. It can be
seen that there existed some differences between the positive and negative potential slopes
before the pH was modified and that these differences disappeared, from that point onward,
from the negative excursions and became variable for the positive excursions. These subtle-
ties are also more characteristic of localized corrosion phenomena.
But the analysis for correlations revealed that an interesting relation seemed to exist be-
tween the fractal dimension that can be evaluated by the R/ S technique and three of the
four electrochemical parameters, that is, CR, Cat, and pH. This is illustrated in Fig. 12, in
which the Hurst exponent that leads to the fractal dimension through Eq 2 is seen to change
from values indicating persistence (~0.6) to values indicative of antipersistence (~0.4) after
the pH was modified. Both the capacity dimension and the Lyapunov exponents calculated
with the Chaos Data Analyzer also seemed to follow these three electrochemical parameters,
although with much more scatter than dn, indicating that the chaotic nature of the EN records
is also linked, at least in this experiment, to the corrosion kinetics (CR) measured indepen-
dently by EIS and the corrosivity of the solution (pH). Once sound correlations are estab-
lished, the advantages of any of the transforms examined in this study become obvious: the
FIG. 11--Positive (black symbols) and negative (white symbols) potential slopes corresponding to
short ([[] and U) and long (A and ~r) duration peaks measured during the experiment with 5083
aluminum.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
154 ELECTROCHEMICAL NOISE MEASUREMENT
FIG. 12--Lyapunov (A) and Hurst (~) exponents calculated during the experiment with 5083
aluminum.
patterns or absence of patterns could give immediate indications concerning the state of a
corrosion situation. By looking at various aspects of the EN signatures, it may become
possible to relate specific combination of characteristic features to the corrosion processes
themselves.
Summary
This paper has described a study of electrochemical measurements that included the anal-
ysis of potential noise fluctuations recorded during the free corrosion of a strain-hardened
aluminum alloy used for armor applications exposed to an aerated saline solution. The anal-
ysis of spontaneous electrochemical noise generated during this experiment was made by
various analysis techniques specifically designed to reveal the stochastic and fractal nature
of signals and shapes. It was found that the corrosion cell made with aluminum exposed to
an aqueous aerated saline solution is an excellent source of noise which can be easily varied
by lowering the pH of the solution to force depassivation phenomena. If this pH is adjusted
to 4.8, it is even possible to initiate the production of oscillations that can last for many
hours and be a vivid demonstration of the complexity of the processes that can occur at a
corroding interface.
But to make sense of these features and reveal their essential components, it is important
to consider appropriate signal-processing techniques. In the present paper, the fractal dimen-
sion estimated by R/S analysis seemed to fit the trends observed independently by measuring
the corrosion rates with EIS and the corrosivity as it relates to the pH of the solution. These
findings, if they are confirmed for other systems, would indicate that the potential fluctuations
can be used to monitor the corrosion rates.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
ROBERGE ON QUANTIFYING STOCHASTIC BEHAVIOR OF ENM 155
Acknowl edgment
The aut hor is grat eful to t he Def ence Resear ch Es t abl i s hment Paci fi c, and par t i cul ar l y t o
Dr. Der ek Lenar d, wi t hout whose suppor t this wor k woul d not have been done.
Re f e r e nc e s
[1] Schuster, H. G., Deterministic Chaos, Physik-Verlag, Weinheim, Germany, 1984.
[2] Berge, P., Pomeau, Y., and Vidal, C., Order Within Chaos, John Wiley & Sons, New York, 1984.
[3] Thompson, J. M. T. and Stewart, H. B., Nonlinear Dynamics and Chaos, John Wiley & Sons, New
York, 1986.
[4] Grassberger, P. and Procaccia, I., "Characterization of Strange Attractors," Physica D, Vol. 13, No.
5, 1984, pp. 34-54.
[5] Moon, F. C., Chaotic Vibrations, John Wiley & Sons, New York, 1987.
[6] Tyagai, V. A., "Noise in Electrochemical Systems," Elektrokhimiya, Vol. 10, No. 1, 1974, pp. 3-24.
[7] Monticelli, C., Brunoro, G., Frignani, A., and Trabanelli, G., "Surface Active Substances as Inhib-
itors of Localized Corrosion of the Aluminum Alloy AA 6351," Journal of the Electrochemical
Society, Vol. 139, No. 7, 1992, pp. 693-705.
[8] Roberge, P. R., Halliop, E., and Sastri, V. S., "A Study of Passivation/Depassivation of Carbon
Steel: Electrochemical Impedance Spectroscopy vs Potential Noise Fluctuations," Materials Science
Forum, Vols. 111-112, 1992, pp. 147-154.
[9] Gabrielli, C., Huet, F., Keddam, M., and Oltra, R., "A Review of the Probabilistic Aspects of
Localized Corrosion," Corrosion, Vol. 46, No. 4, 1990, pp. 266-278.
[10] Gabrielli, C., Huet, E, and Keddam, M., "Investigation of Electrochemical Processes by an Elec-
trochemical Noise Analysis," Electrochimica Acta, Vol. 31, No. 8, 1986, pp. 1025-1039.
[11] Cottis, R. A. and Loto, C. A., "Electrochemical Noise Generation during SCC of a High-Strength
Carbon Steel," Corrosion, Vol. 46, No. 1, 1990, pp. 12-19.
[12] Loto, C. A. and Cottis, R. A., "Electrochemical Noise Generation During Corrosion of Stainless
Steel-Type 316 in Acid Chloride Environment," British Electrochemistry, Vol. 4, No. 12, 1988, pp.
1001-1005.
[13] Benzaid, A., Gabrielli, C., Huet, F., Jerome, M., and Wenger, F., "Investigation of the Electro-
chemical Noise Generated during the Stress Corrosion Cracking of a 42CD4 Steel Electrode,"
Materials Science Forum, Vols. 111-112, 1992, pp. 167-176.
[14] Bendat, J. S. and Piersol, A. G., Random Data: Analysis and Measurement Procedures, 2nd ed.,
John Wiley & Sons, New York, 1986.
[15] Gull, S. F. and Skilling, J., "Maximum Entropy Method in Image Processing," lEE Proceedings,
Vol. 131, Pt. F, No. 6, 1984, pp. 646-659.
[16] Andersen, N., "On the Calculation of Filter Coefficients for Maximum Entropy Spectral Analysis,"
Geophysics, Vol. 39, 1974, pp. 69-72.
[17] Sprott, J. C. and Rowlands, G., "Chaos Data Analyzer," Physics Academic Software, IBM v. 1.0,
New York, 1992.
[18] Roberge, P. R., "The Analysis of Spontaneous Electrochemical Noise for Corrosion Studies," Jour-
nal of Applied Electrochemistry, Vol. 23, No. 12, 1993, pp. 1223-1231.
[19] Roberge, P. R., "Analysis of Electrochemical Noise by the Stochastic Process Detector Method,"
Corrosion, Vol. 50, No. 7, 1994, pp. 502-512.
[20] Roberge, P. R., Lenard, D., Moores, J. G., and Halliop, E. "Marine Corrosion Behavior of A1-Li
Alloy Sheet," in Advancements in Synthesis and Processes, Society for Advancements of Materials
and Process Engineering, Covina, CA, 1992, pp. M51-62.
[21] Roberge, P. R., Halliop, E., Lenard, D. R., and Moores, J. G., "Electrochemical Characterization
of the Corrosion Resistance of Aluminum-Lithium Alloys," Corrosion Science, Vol. 35, Nos. 1--4,
1993, pp. 213-221.
[22] Mandelbrot, B. B. and van Ness, J. W., "Fractional Brownian Motion, Fractional Noises and Ap-
plications," SlAM Review, Vol. 10, No. 4, 1968, pp. 422-437.
[23] Hurst, E. H., "Methods of Using Long-term Storage in Reservoirs," Proceedings, Institute of Civil
Engineering, Vol. 5, No. 5, Part I, 1956, pp. 519-591.
[24] Mandelbrot, B. B. and Wallis, J. R., "Some Long Term Properties of Geophysical Records," Water
Resources Research, Vol. 5, 1969, pp. 321-340.
[25] Fan, L. T., Neogi, D., and Yashima, M., Elementary Introduction to Spatial and Temporal Fractals,
Springer-Vedag, Bedim 1991.
[26] Feder, J., Fractals, Plenum, New York, 1988.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
156 ELECTROCHEMICAL NOISE MEASUREMENT
[27] Walpole, R. E. and Myers, R. E., Probability and Statistics f or Engineers and Scientists, 4th ed.,
Macmillan, New York, 1987.
[28] Searson, P. C. and Dawson, J. L., Journal of the Electrochemical Society, Vol. 135, 1988, pp.
1903-1918.
[29] Uruchurtu, J. C. and Dawson, J. L., Corrosion, Vol. 4, 1987, pp. 19-26.
[30] Roberge, P. R. and Beaudoin, R., Journal of Applied Electrochemistry, Vol. 18, 1988, pp. 601-607.
[31] Roberge, P. R., Halliop, E., Asplund, M., and Sastri, V. S., Journal of Applied Electrochemistry,
Vol. 20, 1990, pp. 1004-1008.
[32] Roberge, P. R., Halliop, E., and Sastri, V. S., Corrosion, Vol. 48, 1992, pp. 447-454.
[33] Grauer, R., Moreland, P. J., and Pini, G., "A Literature Review of Polarisation Resistance Constant
(B) Values for the Measurement of Corrosion Rate," National Association of Corrosion Engineers,
Houston, TX, 1982.
[34] Nguyen, T. H. and Foley, R. T., Journal of the Electrochemical Society, Vol. 127, No. 12, 1980,
pp. 2563-2566.
[35] Foley, R. T. and Nguyen, T. H., Journal of the Electrochemical Society, Vol. 129, No. 3, 1982, pp.
2563-2566.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
Thomas E Barton, Davi d L. Tuck, and D. Brett Wells ~
The Identification of Pitting and Crevi ce
Corrosion Spect r a in Electrochemical Noi se
Using an Artificial Neural Net work
REFERENCE: Barton, T. E, Tuck, D. L., and Wells, D. B., "The Identification of Pitting
and Crevice Corrosion Spectra in Electrochemical Noise Using an Artificial Neural Net-
work," Electrochemical Noise Measurement f or Corrosion Applications, ASTM STP 1277,
Jeffery R. Kearns, John R. Scully, Pierre R. Roberge, David L. Reichert, and John L. Dawson,
Eds., American Society for Testing and Materials, 1996, pp. 157-169.
ABSTRACT: An artificial neural network has been developed to identify the onset and classify
the type of localized corrosion from electrochemical noise spectra. The multilayer feedforward
(MLF) network was trained by classical back-propagation to identify corrosion from the char-
acteristics of the initial current ramp. Using 50 training files and 39 test files taken from
measurements on Type 304 stainless steel in a dilute chloride electrolyte, the network accurately
detected and classified 96% of the data and reported no misclassifications. Experiments with
high levels of adventitious noise superimposed on the original data have been carried out to
examine the noise tolerance of the network.
KEYWORDS: artificial neural networks, pitting corrosion, crevice corrosion, electrochemical
noise, stainless steel
El ect rochemi cal Noi s e and Loc al i z e d Corros i on
Electrochemical noise associated with corrosion processes is readily detected on passive
or coated surfaces as low-frequency current or potential fluctuations. Previous analysis of
electrochemical noise has concentrated on two broad treatments: under steady-state condi-
tions, small fluctuations in potential and current have been correlated to produce measure-
ments of low-frequency impedance allowing estimation of corrosion rate [1,2]. Second, under
stationary conditions, the stochastic nature of localized corrosion processes has been inves-
tigated by spectral analysis in the frequency domain or statistical counting in the time domain
[3-10]. In the latter treatment, stochastic models have been explored yielding parameters
such as intensity or rates of pit generation [6,7], and criteria for sustained localized disso-
lution based on a critical pit age have been developed [8,9]. Within the framework of these
localized corrosion models and with the recognition of electrochemical noise signatures as-
sociated with specific corrosion processes, particularly metastable or incipient localized cor-
rosion processes [It)I, an opportunity exists to develop monitors for localized corrosion. The
objective was to develop and test an artificial neural network to identify the onset of localized
corrosion from el ect rochemi cal noise spectra.
The practical interest in neural networks as a detection and classification tool lies in the
analysis of field-generated electrochemical noise spectra. Fluctuations in the environmental
conditions found in industrial plants frequently represent a larger source of uncertainty than
Research scientists, Industrial Research Ltd., EO. Box 2225, Auckland, New Zealand.
Copyrighl 9 1996 by ASTM International
157
www.astm.org
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
158 ELECTROCHEMICAL NOISE MEASUREMENT
the random variation associated with the initiation of localized corrosion in laboratory ex-
periments. The assumption of a stationary state is often invalid. The successful identification
of precursors to localized corrosion using neural networks may reduce the risk of corrosion
failure for plant conditions that are not adequately defined.
A fundamental requirement for direct monitoring of corrosion activity is the efficient
interpretation of spectral characteristics. Criteria for identifying localized corrosion have been
proposed based on changes in the noise resistance R, = V / I (where V and 1 are root mean
square [rms] values of potential and current) [2] or the characteristic shape of the power
spectral density curve [3, 4]. The averaging procedures associated with these techniques are
expected to limit the sensitivity to infrequent corrosion activity, and this study concentrates
on the direct analysis of data in the time domain only. A neural network has been designed
to operate as a decision-making filter, to distinguish the onset of pitting or crevice corrosion
based on the characteristics of the current spectra. Onl y the initial current ramp has been
used to identify between corrosion processes. The advantage of using a neural net work for
identification is that the network establishes its own "noi se-t ol erant " criteria for classification
and offers the potential for on-line operation and actual field training.
Electrochemical data have been collected for the initiation of pitting and crevice corrosion
on AISI Type 304 stainless steel in a dilute chloride environment (1000 mg 9 L 1 C1-) to
enable training and testing of the network.
Neural Networks and Learning
The architecture of an artificial neural network consists of highly interconnected processor
elements with groups of elements operating in parallel. Because of the distributed nature and
interconnectivity of the network, large amounts of input activity converge on each of these
processor elements, which operate as simple summers. Weights exist on the processor inter-
connections, which proportion the flow of data through the network. Nonlinear transfer func-
tions operate on the summed converging data at each processor element. The outputs of the
network therefore depend on the
9 Input data,
9 Net work architecture,
9 Weights of the network interconnections, and
9 Form of the transfer functions.
The choice of the size and type of architecture and the form of the transfer functions
depend on the type of input data and the problem to be solved [11,12]. The type of training
algorithm is usually determined by the architecture chosen. A multilayered feedfoward
(MLF) network [13,14] was chosen for this pattern detection and classification problem.
Weights on the processor interconnections of the MLF network were developed by a back-
propagation training algorithm [15].
Multilayered Feedforward Network
The MLF network is the most widely applied artificial neural network. This network has
input and output layers of processors and one or more hidden layers of processors. (Hidden
refers to those that are not directly accessible by, or to, external manipulation.) Each layer,
beginning at the input layer, connects to the next layer ahead of it in a feedforward manner
(Fig. 1). Using the classical back-propagation training algorithm, this type of net work has
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
BARTON ET AL. ON ARTIFICIAL NEURAL NETWORK 159
D ~ W O R K
I N ~ PUTS
HIDDEN HIDDEN
LAYER LAYER
FIG. 1--A schematic of a simple multilayer feedforward network with two hidden layers.
been shown to be able to solve most neural network applicable problems [12,13,16]. Some
solutions take considerable computing time and do not always provide the best possible
solutions. However, successful application areas include modeling or mapping data and sys-
tems, pattern recognition, classification, optimization and time series prediction, a special
case of modeling.
Net work Size~Generalization
The optimal size of the MLF neural network can be roughly defined by the size of the
problem set, but significant controversy remains regarding the minimum number of layers
and processors per layer [17,18]. In general, the larger the network, the longer computational
time required for training. A practical solution to most problems can be found using MLF
networks with ideally one [12] or, generally, two "hidden layers" of processors; two hidden
layers were used in this study. The size of the network following training must be able to
generalize, that is, map appropriate input values to the correct output pattern result. The
number of weights should be kept to less than the number of training data samples to prohibit
the network from learning that data explicitly (assuming each weight has the potential to
store one piece of training data). The validity of this rule depends on the complexity of the
underlying information content in the data and does not allow for redundancy as a result of
repetition in the data set. Too few weights can provide insufficient freedom for the network
to produce a good model of the information implicit in the training data set or result in no
training at all. Generalization of a network can be a difficult goal to accomplish, and ex-
perimentation with alternate network architectures is one way to verify that a satisfactory
solution has been determined.
Transfer Functions
Within any single processor, the summation of the incoming connection activity is per-
formed by calculating the sum of the products of the respective weight and input activity
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
160 ELECTROCHEMICAL NOISE MEASUREMENT
Xj : ~ i Wj i * Y i
( 1 )
where
j = the i ndex number for the pr ocessor in the l ayer under consi derat i on,
y, = the output level of the ith pr ocessor in the previ ous layer, and
wjl = the wei ght for the connect i on bet ween the ith and j t h processors.
The resul t of this wei ght ed summed process is the act i vat i on level, xj, whi ch is then operat ed
on by the t ransfer function, f i x ) , to pr oduce the new output for that processor, z~ = f ( x ~ ) . The
t ransfer funct i on is usual l y chosen to be nonl i near to assi st in the f or mat i on of a compl ex
represent at i ve space. As the out put s must be const rai ned in magni t ude and be di fferent i abl e
for the i mpl ement at i on of back- pr opagat i on t rai ni ng, t he most commonl y used t ransfer func-
tion t ypes are si gmoi dal or hyper bol i c functions. These l i mi t the out put s to bet ween 0 and
+ 1 for l ogi st i c functions, and - 1 and + 1 for hyperbol i c functions. The fol l owi ng l ogi st i c
t ransfer function was chosen for this st udy:
f(x) = (1 - e x)-, (2)
L e a r n i n g / T r a i n i n g
The MLF net work is t rai ned by back- pr opagat i on, whi ch is a t wo-pass supervi sed pr o-
cedure. Fi rst , successi ve i nput sets of the t rai ni ng dat a are operat ed on by the net work in a
feedforward manner to pr oduce cor r espondi ng out put dat a sets. The net work out put dur i ng
t rai ni ng is compar ed with target out put sets, known t ransformat i on resul t s bet ween i nput and
out put dat a sets. The initial wei ght s are usual l y preset to smal l r andom values. Errors bet ween
the actual and target out put dat a sets are cal cul at ed and accumul at ed for the whol e t rai ni ng
set. In the second pass, the i nt erconnect i ve wei ght s are syst emat i cal l y changed by appl yi ng
t rai ni ng rules that first reduce the output l ayer errors and then subsequent l ayer errors, movi ng
backwar d ( back- pr opagat ed) t hrough the net wor k to the input. An i t erat i ve pr ocedur e is
fol l owed to r educe the out put error until
9 the mean (output) squared error (MSE) is suffi ci ent l y low,
9 the t i me deri vat i ve of the MSE is suffi ci ent l y low, or
9 a sufficient percent age of the l earni ng and test set are correct l y cl assi fi ed.
where
MSE = 0. 5N Ec EjN (zj.,. - d j . , . ) :
MSE = the mean out put l ayer squared error,
c = the case bei ng anal yzed,
Zj,c = the actual out put val ue of t he j t h output l ayer pr ocessor for case c,
dzc = its desi red output, and
N = the total number of cases used to train or test the net work.
( 3 )
Thi s mi ni mi zat i on rul e gi ves a maxi mum est i mat e for the wei ght adj ust ment s necessary
to cancel the out put errors. General l y, smal l i ncrement al adj ust ment s proport i onal to a learn-
ing rate coeffi ci ent (usual l y < < 1.0) are made. Often ot her t echni ques are used to assi st wi t h
the convergence to near opt i mal sol ut i ons; one common met hod al so uses the second deri v-
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
BARTON ET AL. ON ARTIFICIAL NEURAL NETWORK 161
ative of the training rate and the product of an exponential decay or moment um coefficient
( < 1.0) to contribute to the weight adjustments.
The correctness of the solution is evaluated and confirmed by testing with an independent
data set. Once a suitable solution is determined, the weights can then be fixed and new data
processed in a feedforward manner so the network output provides the appropriate answers.
The information stored by the network in its internal representational space during training
is clearly related to the weight values. However, the complexity of the weight values and
network structure mask the underlying logic rules used to solve the problem. Onl y with
extensive testing can confidence be gained in this "bl ack box" model i ng approach. It is usual
practice to use t wo independent sets of data: one for training the network and the second
for testing the resultant solution determined.
Experimental Procedure
Data Collection
Pitting and crevice experiments were performed on AI SI Type 304 stainless steel in a
deaerated, 1000-mg 9 L ~ chloride electrolyte. Electrodes were prepared from solution-
annealed Type 304 stainless steel disks with a 1.5-cm 2 surface area. The electrodes were
mount ed flush in epoxy. Electrochemical data were gathered using a multispecimen cell and
an eight-channel potentiostat with data collected simultaneously on all eight channels with
a comput er system. The counter-electrode was a large platinum foil. The cell was placed
inside a Faraday cage to minimize external noise, although adventitious noise was super-
imposed on the data during testing of the network.
To initiate localized corrosion, the potential of each electrode was slowly ramped from
- 3 0 0 to +700 mV (versus a saturated calomel reference electrode) at 0.1 mV/ s , and the
current was recorded at one data point per second. The level of the baseline current was
typically 0.1 txA at the onset of localized corrosion. The full-scale current limit was set at
30 ~A at which point the working electrode was isolated from the potentiostat t o prevent
excessive corrosion damage.
Because of the mount i ng geomet ry of the electrodes, localized corrosion could occur via
two mechanisms: pit formation on the face of the electrode or crevice formation at the edge
of the electrode between the stainless steel and the epoxy mount. The potential ramp and
electrolyte conditions were such that there was an equal likelihood of pitting corrosion or
crevice corrosion.
After each experiment, the electrodes were examined by optical mi croscopy to identify
the t ype o f corrosion that had occurred. Files containing up to I 0 000 data points were
examined and then shortened so that they contained the pit or crevice corrosion initiation
data, plus some additional leading data points to establish the baseline level of current. When
training the network, the start of the corrosion event was arbitrarily assigned as the first of
at least three consecutive data points above the mean baseline plus two standard deviations
calculated from local baseline noise. Fifty training files were produced, 25 pitting and 25
crevice corrosion data files. An additional 39 files were prepared for testing, 21 pitting and
18 crevice corrosion data files. Typical current / t i me data files showing the onset of corrosion
can be seen graphically in Fig. 2.
The Artificial Neural Network
The network initially had two outputs for classifying corrosion events, that is, pits or
crevices. Using this network, the number of processors in the hidden layers was optimized
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
162 ELECTROCHEMICAL NOISE MEASUREMENT
' . D
rr
u
t~ . . ' ' 2
~ g J - - -
Z
r
i
0 4 8 12 16
Ti me i n Se c onds
pi t s
cr evi ces . . . .
FIG. 2--Typical current/time spectra of the initiation of pitting and crevice corrosion.
wi t h the first hi dden l ayer requi ri ng seven processors and the second hi dden l ayer requi ri ng
five. Af t er this pr el i mi nar y study, a t hi rd out put was added in the t rai ni ng phase to rei nforce
the basel i ne- dat a- onl y si t uat i on before corrosi on was i ni t i at ed to avoi d fal se det ect i on of
corrosi on events. A sl i di ng wi ndow el even dat a poi nt s wi de (Fig. 3) was used to i nput the
cur r ent / t i me dat a (that is, a net work i nput of el even consecut i ve dat a poi nt s). Duri ng t rai ni ng
and testing, the wi ndow was scanned al ong the dat a poi nt s for ei ght or ten shifts to si mul at e
dat a i nput from a corrosi on monitor, each shift pr oduci ng a mat hemat i cal l y i ndependent dat a
set of el even points. Therefore, the t ot al number of i ndependent dat a sets was the number
of dat a files t i mes the dat a sets per file, or a mi ni mum of 400 t rai ni ng dat a sets and 312
testing sets. The t rai ni ng pr ocess was desi gned to rei nforce the net work out put cl assi fi cat i on
aft er t he present at i on of the i nput dat a set, such that
9 i f t he i nput val ue in t he mi ddl e of the sl i di ng wi ndow i ncreased consi st ent l y above the
mean basel i ne pl us t wo st andard devi at i ons for at l east three consecut i ve shifts of the
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
BARTON E'F AL. ON ARTIFICIAL NEURAL NETWORK 163
a) Detection t l I - " "
Level =0. 1 l / 9 9 I 9 9
9 . . . . . ' - . . . . . . . I" . . . . . . . . . . . . . . . . . . . . Baseline Mean
2 shifts f ur t her on:
b) Detecti~ i . . . . . . ' I 9
Level = 0.5[ ~ 9 9 "
. . . . . . . . . . . . . . . . . . . . . . . I ] [ 9 9 9 ~
Mean
' " 9 1 7 6 1 4 9 . . . . 1 " 1 . . . . . . . ..i . . . . . . 1 . 7 . . - . - . . 7 7 . 2 a s e l i n e M e a n
d) LDeetelc ti Oo7 [ I I 9 ~- 9 9 9
FIG. 3--Schematic operation of the eleven-poim sliding window.
Baseline Mean
input window along the data points, then the output value should increase from 0.1 to
0.5 for the correct corrosion mechani sm (either pitting or crevice corrosion) and
9 Each additional shift of the sliding window, which was above the threshold level (mean
plus two standard deviations), should provide a further increase in the correct output
value by 0.1 to a maxi mum of 0.90.
Figure 3 illustrates the sliding window process, and Table 1 shows how the pitting and
crevice corrosion output values develop after training as the sliding window is shifted over
a typical pitting corrosion data file. I f the data set did not exhibit evidence of either type of
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
164 ELECTROCHEMICAL NOISE MEASUREMENT
TABLE l--Output levels during operation of the sliding window.
Sliding Window Pitting Corrosion Crevice Corrosion
Shift Number Detection Level Detection Level Baseline Detection Level
1 0.10 0.10 0.90
2 0.10 0.10 0.90
3 0.12 0.08 0.88
4 0.50 0.15 0.45
5 0.62 0.10 0.31
6 0.71 0.13 0.23
7 0.73 0.07 0.24
8 0.81 0.10 0.18
9 0.85 0.09 0.13
10 0.89 0.11 0.10
corrosion initiation, then the corrosion type outputs were trained to give a detection level of
0.1 and the output for the baseline presence trained to 0.90.
Each of the 50 training files was processed through the network multiple times during the
training period with each full set of training data considered as one cycle. The net error was
monitored during the training cycles, and, when the rate of change of the error value was
sufficiently low, the test data files were run through the network (typically after 15 000 and
30 000 training cycles) to confirm a valid solution.
Resul t s
Initially, only a two-output network was used with outputs for indicating pitting corrosion
and crevice corrosion, but no output to reinforce the presence of baseline data. The criteria
for acceptable identification were three consecutive output values above 0.45 of the correct
corrosion process and increasing confidence in the output values such that the sum of the
three consecutive values was above 1.65. These criteria were arbitrary but offered a relatively
stringent measure of acceptable classification. After 15 000 cycles of the training data, using
eight shifts of the sliding window, the network was able to identify successfully 100% of
the pitting corrosion training files and 76% of the crevice corrosion training files (Table 2).
When the test files were run through the network, 95% of the pitting corrosion files and
TABLE 2--Percentage of accurate classification by the neural network.
2 Output, a 3 Output, b 3 Output, b
15 000 Cycles, 30 000 Cycles, 30 000 Cycles,
8 Shifts, % 8 Shifts, % 10 Shifts, %
Training: Total 88 96 96
Pitting 100 96 100
Crevicing 76 96 92
Testing: Total 85 90 95
Pitting 95 95 100
Crevicing 72 83 90
Baseline false detections <5 0 0
a Two outputs: pitting, crevicing.
b Three outputs: pitting, crevicing, baseline.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
BARTON ET AL. ON ARTIFICIAL NEURAL NETWORK 165
72% of the crevi ce corrosi on files, were identified. There were 5% fal se det ect i ons in whi ch
t he net wor k recorded a crevi ce corrosi on i ni t i at i on in whi ch t here was onl y basel i ne data.
A t hi rd out put to the net work to di st i ngui sh sect i ons of the dat a files that cont ai ned onl y
basel i ne l evel current was added to mi ni mi ze t he occurrence of fal se det ect i ons. By t rai ni ng
this new net wor k to 30 000 cycl es wi t h t he t hi rd output, the correct det ect i ons i ncreased to
96% for bot h the pi t t i ng and crevi ce corrosi on t rai ni ng files, wi t h no fal se det ect i ons. The
t hree-out put net work had sl i ght l y mor e di ffi cul t y i dent i fyi ng the t est files correct l y wi t h 95%
of t he pi t t i ng corrosi on t est files and 83% of the crevi ce corrosi on t est files correct l y i den-
tified. There were, however, no fal se det ect i ons in the basel i ne data.
By i ncreasi ng t he number of shifts of the sl i di ng wi ndow f r om ei ght to t en in t rai ni ng of
anot her t hree-out put net work (for 14 500 cycl es), t he percent age of correct det ect i ons in-
creased to 100% for pi t t i ng corrosi on t rai ni ng files and 92% for the crevi ce corrosi on t rai ni ng
files. Resul t s for the test files al so i mpr oved with 100% of the pi t t i ng corrosi on t est files and
90% of t he crevi ce corrosi on test files correct l y i dent i fi ed and no fal se det ect i ons.
As t he ori gi nal dat a files wer e col l ect ed wi t h a l ow level of advent i t i ous noise, furt her
exper i ment s were carri ed out to t est the net wor k' s noi se t ol erance by art i fi ci al l y superi m-
posi ng advent i t i ous noi se on the signal. The levels of noi se in the ori gi nal dat a were anal yzed,
and files cont ai ni ng an addi t i onal 2.0, 4.0, and 5.0 st andard devi at i ons of basel i ne noi se were
produced. The results are l i st ed in Tabl e 3 for a net work t rai ned wi t hi n the ori gi nal dat a files
but t est ed wi t h addi t i onal advent i t i ous noi se added. Two st andard devi at i ons of noi se r educed
the correct i dent i fi cat i on rates onl y sl i ght l y to 96% of t he pi t t i ng corrosi on t rai ni ng files and
90% of t he crevi ce corrosi on t rai ni ng files, wi t h no fal se det ect i ons. Of the test data, 100%
of t he pi t t i ng corrosi on files and 88% of the crevi ce corrosi on files were correct l y identified.
The addi t i on of 4. 0 and 5.0 st andard devi at i ons of noi se r educed t he correct i dent i fi cat i on
of crevi ce corrosi on onset but di d not great l y affect the accuracy of the net work for pi t t i ng
corrosi on identification. Exampl es of files wi t h advent i t i ous noi se super i mposed on the ex-
peri ment al dat a are present ed in Fi g. 4.
Di s c us s i on
The results i ndi cat e t he neural net work coul d successful l y i dent i fy the onset and t ype of
l ocal i zed corrosi on usi ng a narrow wi ndow and few dat a poi nt s. Accept abl e i dent i fi cat i on
levels of >95% are encour agi ng wi t hi n the stringent defi ni t i on of success, that is, gi ven t hat
a posi t i ve cl assi fi cat i on is expect ed at or soon aft er the fourth shi ft and requi res t hree con-
secut i ve val ues of >0. 45 to be obt ai ned within the four or six shifts used. Vari abi l i t y is
observed in the number of shifts requi red to first i dent i fy the onset of pi t t i ng corrosi on and
mor e so for crevi ce corrosi on. The scat t er appears to cor r espond to the arbi t rary defi ni t i on
of i ni t i at i on used wi t h the t rai ni ng and test data. The l ower per cent age of correct classifi-
TABLE 3--Percentage of accurate classification by the neural network with noise-added spectra.
No Noise, %" 2~ Noise, %~ 4tr Noise, %~ 5cr Noise, %~
Training: Pitting 100 96 88 92
Crevicing 96 90 64 36
Testing: Pitting 100 100 100 100
Crevicing 90 88 61 55
Baseline false detections 0 0 0 5
Q Refers to noise added to spectra.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
166 ELECTROCHEMICAL NOISE MEASUREMENT
o ' 3
~
.- = / ,
c ' - I
L)
t ~
C' 4
i
0 4 8 12 16
Ti me in Seconds
pits
crevices . . . .
FIG. 4--Current~time spectra with adventitious noise added.
cations found for the eight-shift procedure is believed to result from this constraint because
three consecutive correct values must be established in only five shifts.
The onset of pitting corrosion was identified under all test conditions with considerable
certainty despite the overlap in ramp rate of some crevice and pit data. Detection or classi-
fication errors were usually associated with crevice corrosion measurements, and this was
not substantially improved by the addition of a third output to reinforce the baseline data,
although false detections were eliminated by the presence of the third output during training.
As the adventitious noise was increased to five standard deviations above the baseline level,
the output values for crevice corrosion were subject to greater variability and errors arose
from the failure to measure three consecutive values above 0.45 in seven window shifts.
Only a small percentage of the test results indicated no increase in crevice output values as
the data were shifted through the window. The number of shifts required to classify both
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
BARTON ET AL. ON ARTIFICIAL NEURAL NETWORK 167
pitting and crevice corrosion systematically increased with increasing adventitious noise.
Refinement of the classification criteria and an increase in the number of shifts of data
through the network are expected to assist in increasing noise tolerance, although it is not
possible to quantitatively predict the level of improvement. Preliminary results for networks
that have been trained with adventitious noise have not revealed substantial improvements
in the noise tolerance. Other networks may offer better discrimination for very noisy data
sets (for example, adaptive resonance theory network [19]) or more efficient transformation
mapping processes (for example, Kohonen networks [20] or recurrent networks [16]). No
attempt was made to evaluate the efficiency of our network architecture against other
possibilities.
Although it may be cited as a strength that the MLF neural network requires no author-
imposed classification model, the inability to identify the criteria used for classification leads
to lengthy development and training times to optimize the technique. It is speculated that
the network classifies pitting and crevice corrosion primarily by the variation in current ramp
rate above the baseline value. A larger window size should provide greater confidence in
classifying crevice corrosion because the value of the current over time progressively rises
above the baseline. It is impractical to increase the input data set substantially as the number
of weights, and therefore the complexity of the network, rises rapidly.
Although fast Fourier transform (FFT) and maximum entropy techniques are normally
valid as steady-state operations, they can act like this neural network as a window analysis
on a snapshot of real-time data. For comparison, the results of FFT and noise resistance (rms
V/ 1) techniques using pitting and crevice test data with the original level of baseline noise
are presented in Fig. 5. The FFF analysis was based on a 16-point window; the resistance
noise method used an 11-point running mean. Figure 5 shows that, although the onset of
pitting corrosion can be recognized by both techniques, only FFF can easily distinguish
crevice corrosion under these condi t i ons. More work is required to provide an adequate
comparison between the performance of these networks simply trained to operate in the time
domain, and a neural network that uses maximum entropy methods or FF1 r techniques to
preprocess the data before interpretation.
The next step in the development of the neural network system is to make alterations to
allow continuous signal analysis for on-line monitoring. This development would allow field
study of localized corrosion. Additional developments will include producing a hybrid system
incorporating an expert system to allow rule-based judgments from neural network outputs.
This approach will allow other more logical criteria to be added to classification analysis
including prepitting or repassivation behavior. Multiple neural networks could also be linked,
in which each network is trained for an individual classification, and the combined system
produces multiple output for decision making.
Although it has been demonstrated in this work that neural networks are successful clas-
sifiers of electrochemical noise spectra, there are other applications in the area of corrosion
and electrochemical noise where neural networks might be valuable. Smets and Bogaerts
used a neural network to forecast chloride-induced stress corrosion cracking based on tem-
perature, chloride, and oxygen concentrations [21]. Rosen and Silverman successfully used
a neural network to link potentiodynamic polarization scans with prediction of pitting, crev-
ice, and general corrosion probabilities [22], although Silverman generally used experimental
values as inputs rather than spectral data [23,24]. Ben-Haim and MacDonald did similar
predictive work by modeling chemical characteristics of salt brines using neural networks
[25]. Other areas that show potential include modeling AC impedance spectra and classifi-
cation of multivariable solution conditions with the characteristics of corrosion initiation.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
168 ELECTROCHEMICAL NOISE MEASUREMENT
2 0 0 0 0
1 5 0 0 0 ~
C
D \
i 0 0 0 0 ~
"~ 5 0 0 0 ~
< " ._.
0
0 1 2
Pi t t i ng
Cr e v i c i ng . . . . . . . .
Bas e l i ne
Fas t Fo ur i e r Tr a n s f o r m
3 4
n
5 6 7
Fr e q u e n c y Ha r mo n i c
>
N o i s e Re s i s t a n c e
o
E
"6
>
2800
1400
\
\
4 6 8
n
! 0
Wi n d o w s hi f t s
Pi t t i ng
Cr e v i c i n g . . . . . . . .
FI G. 5--F ast Fourier transform and resistance noise analysis of data files.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
BARTON ET AL. ON ARTIFICIAL NEURAL NETWORK 169
Summary
A multilayer feedfoward neural network has been used to identify the onset of localized
corrosion by the analysis of electrochemical noise data. The advantage of using a neural
network for this identification is that the network appears to be able to generalize and choose
its own optimal criteria for classification. The network was able to detect and classify cor-
rectly 96% of the data within seven to ten data points (recorded at one sample per second).
It is concluded that neural networks in conjunction with electrochemical noise could offer a
robust, nondestructive monitoring technique for the detection of localized corrosion.
Acknowl edgment
This research was funded by the New Zealand Foundation for Research, Science, and
Technology under Contract CO8216.
References
[1] Gabrielli, C. and Keddam, M., Corrosion, Vol. 48, 1992, p. 794.
[2] Rothwell, A. N. and Eden, D. A., Corrosion~92, No. 223, NACE, Houston, TX, 1992.
[3] Gabrielli, C., Huet, E, Keddam, M., and Oltra, R., in Advances in Localised Corrosion, NACE 9,
H. Isaacs, et al, Eds., NACE, Houston, TX, 1990, p. 93.
[4] Umchurtu, J. C. and Dawson, J. L., Corrosion, Vol. 43, 1987, p. 19.
[51 Bertocci, U. and Yang-Xiang, Y., Journal of the Electrochemical Society, Vol. 131, 1984, p. 1011.
[6] Shibata, T. and Takeyama, T., Corrosion, Vol. 33, 1977, p. 243.
[7] Baroux, B., Corrosion Science, Vol. 28, 1988, p, 969.
[8] Williams, D. E., Westcott, C., and Fleischmann, M., Journal of the Electrochemical Society, Vol.
132, 1985, p. 1796.
[9] Williams, D. E., Westcon, C., and Fleischmann, M., Journal of the Electrochemical Society, Vol.
132, 1985, p. 1804.
[10] Stewart, J., "Pit Initiation on Austenitic Stainless Steels," Ph.D. thesis, University of Southampton,
UK, 1990.
[11] Barton, T. E, Tuck, D. L., and Wells, D. B., in Proceedings of the 93 ANNES Conference, IEEE
Computer Society Press, Los Alamitos, CA, 1993, p. 325.
[12] Lippmann, R., IEEE ASSP Magazine, Vol. 4, 1987, p. 4.
[13] Rumelhart, D. E., McClelland, J. L., et al., Parallel Distributed Processing, Vols. I and II, MIT
Press, Cambridge, MA, 1986.
[14] McClelland, J. L. and Rumelhart, D. E., Explorations in Parallel Distributed Processing, MIT Press,
Cambridge, MA, 1988.
[15] Hinton, G. E., Scientific American, Vol. 267, No. 3, 1992, p. 105.
[16] Bressloff, P. C. and Weir, D. J., GECJournal of Research, Vol. 8, No. 3, 1991, p. 151.
[17] Albertson, E C. and Sontag, E. D., Neural Networks, Vol. 6, 1993, p. 975.
[18] Bulsari, A., Neural Networks, Vol. 6, 1993, p. 991.
[19] Carpenter, G., Grossberg, S., and Reynolds, J., Neural Networks, Vol. 4, 1991, p. 565.
[20] Kohonen, T., Self-Organization & Associative Memory, Springer-Verlag, Berlin, 1987.
[21] Smets, H. M. and Bogaerts, W. F., Materials Performance, Vol. 31, No. 9, 1992, p. 64.
[22] Rosen, E. M. and Silverman,D. C., Corrosion, Vol. 48, 1992, p. 734.
[23] Silverman, D. C., Corrosion, Vol. 50, 1994, p. 411.
[24] Silverman, D. C., in Application of Accelerated Corrosion Tests to Service Life Prediction of Ma-
terials, ASTM STP 1194, G. Cragnolino and N. Sridhar, Eds., American Society for Testing and
Materials, West Conshohocken, PA, 1994, pp. 43-64.
[25] Ben-Haim, M. and MacDonald, D. D., Corrosion Science, Vol. 36, 1994, p. 385.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
Industrial Applications
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
Bar r y C. Syrei~,t ) and Wi l l i am M. Cox 2
A Review of EPRI Projects Since 1984 that
Used Electrochemical Noise Measurement
Instrumentation
RE F E RE NCE : Syrett, B. C. and Cox, W. M., "A Review of EPRI Projects Since 1984 t hat
Used Electrochemical Noise Instrumentation," Electrochemical Noise Measurement f or Cor-
rosion Applications, STP 1277, Jeffery R. Kearns, John R. Scully, Pierre R. Roberge, David
L. Reichert, and John L. Dawson, Eds., American Society for Testing and Materials, 1996, pp.
173-185.
ABS TRACT: This review paper documents the development of a modem approach to online
corrosion evaluation: the origins of the approach, configuring multiple techniques to maximize
data retrieval, and taking the instrumentation out of the laboratory and into the field. Several
Electric Power Research Institute (EPRI) projects conducted over the past decade have used
this approach. This paper describes how with the use of modem corrosion surveillance instru-
mentation the objectives of each of these EPRI projects were achieved and an appropriate
corrosion management strategy was developed. The final section includes a discussion of the
future of corrosion management systems, and their relation to, and dependence on, electro-
chemical noise (EN) instrumentation.
KEYWORI)S: technology demonstrations, electrochemical noise, real time, Electric Power
Research Institute (EPRI), online, surveillance, process control, instrumentation
In the late 1970s, it was recognized that considerable fuel cost savings were possible if
the outlet gas temperature on large power generation boilers could be reduced. It was cal-
culated that a small (approximately 17~ or 31~ reduction in outlet flue gas temperature
would yield annual savings of approximately $3 million in a single 2000-MW power station
[1]. However, a reduced outlet gas temperature increases the risk of condensation of aggres-
sive acids on the walls of the outlet duct and stack, and the onset of so called dewpoint
corrosion. An investigation was launched in 1978 at the University of Manchester Institute
of Science and Technology (UMIST), along with two parallel studies, to examine the mech-
anism of acid dewpoint corrosion in combustion systems. In one study, the electrochemical
impedance spectroscopy technique of corrosion rate evaluation was used [2]. In one of the
other studies, Hladky, working on electrochemical noise evaluations, established that a non-
perturbative approach to the examination of electrochemical response from corroding systems
had considerable advantages over the more widely accepted linear polarization techniques
[3]. The onset and propagation of localized corrosion, such as pitting, were clearly identified
using the electrochemical noise approach.
Subsequent work for the nuclear industry investigated localized corrosion in condensing
nitric acid systems. A multitechnique approach was used in which linear polarization, elec-
Manager, Corrosion Science and Technology, Strategic Research and Development, Electric Power
Research Institute, Palo Alto, CA 94303.
2 Director, Corrosion Management Ltd., Hyde, Cheshire, UK, SK14 5LT.
Copyrighl 9 1996 by ASTM lntcrnational
173
www.astm.org
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
174 ELECTROCHEMICAL NOISE MEASUREMENT
trical resistance, electrochemical impedance, zero resistance ammetry, and electrochemical
noise techniques were used simultaneously to maximize data retrieval. 3 The advantages of
combi ni ng several methods were recognized, and sensor configurations that had been de-
veloped for impedance studies in condensing conditions were adapted to permit measure-
ments, using several electrochemical techniques simultaneously, under essentially identical
environmental conditions. These compact multielement sensors allowed electrochemical
measurements to be conduct ed conveniently in the laboratory and operating plants.
In late 1979, impedance equipment was transported to the Drax Power Station in York-
shire, England, and tests were conducted in the flue gas system of a 660- MW coal-fired
power generation boiler. This and similar work undertaken in the outlet duct of oil-fired
boilers at the Esso Oil Terminal at Trafford Park, Manchester, England, marked the entry of
modem corrosion instrumentation into online corrosion monitoring and opened the way for
a new approach to corrosion control.
Commenci ng in 1981, electrochemical noise instrumentation was used on site in the sec-
ond phase of the dewpoint investigation. In this investigation, the instrumentation demon-
strated that few materials could withstand operation at temperatures just below the dewpoint
temperature. Aggl omerat i ons of ash, bound together by condensed sulfuric acid, could form
causing fouling and bl ockage of the air heaters. Heat recovery could be optimized by mon-
itoring the corrosion rate, establishing an acceptable maxi mum corrosion rate, and reducing
the exhaust gas temperature as much as possible without exceeding the corrosion rate
maximum.
Electrochemical Techniques
Electrochemical Impedance Spectroscopy
This technique is an i mprovement over the conventional linear polarization method of
corrosion rate determination. A small alternating signal, typically of 20 or 40 mV peak to
peak, is applied to the corrosion cell. The frequency of the signal is varied, usually between
10 kHz and 100 mHz for metallic systems, and impedance measurements are plotted in a
Nyqui st or Bode format. From these data, a value of the charge transfer resistance can be
estimated, and a corrosion rate can be deduced via the Stern Geary relationship and Faraday' s
law [4,5].
The technique is especially useful to evaluate changes in solution resistance (or conduc-
tivity), specimen charge transfer resistance, and diffusion behavior. It works well in activa-
tion-controlled corrosion systems but is less effective where systems are entirely diffusion
controlled. When used in combination with the electrochemical noise analysis, it can verify
general corrosion rate estimates obtained from the noise analysis.
Electrochemical Noise Measurement
The electrochemical noise (EN) analysis techniques are based on the evaluation of spon-
taneous current and voltage transients and low level signals. These fluctuations and transients
can be used to characterize the electrochemical condition at a corroding interface.
In its simplest application, the data may be analyzed using an approach analogous to
Ohm' s law ( V/ I = R). For linear polarization, Av/ Ai = Rp would be used (where Av is the
3 D. A. Eden, "Pitting of Stainless Steels in Iodide Solution," CAPC1S Internal Report (private com-
mercial communication), 1981.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
SYRETT AND COX ON ELECTROCHEMICAL NOISE INSTRUMENTATION 175
i ncr ement al c ha nge in vol t age, Ai is t he as s oci at ed i ncr ement al c ha nge in cur r ent , and Rp i s
t he cons equent pol ar i zat i on r esi st ance) , and f or noi se gener at ed by uni f or m cor r os i on, t he
st andar d devi at i on o f t he si gnal s i s ~rv/cr i = R, ( wher e tro is t he ma gni t ude o f t he vol t age
noi se si gnal , tr, is t he ma gni t ude o f t he cur r ent noi se si gnal , and R, i s t he cons equent " n o i s e
r e s i s t a nc e " val ue, whi ch i s anal ogous t o t he pol ar i zat i on r esi st ance) . Be f or e t he UMI S T
pat ent s wer e i ssued [6,7], nume r ous st udi es ma de r ef er ence t o el ect r ochemi cal noi se, but t he
dat a wer e not conver t i bl e i nt o cor r os i on r at es unt i l i t wa s r ecogni zed t hat (i n es s ence) it wa s
ne c e s s a r y t o c ombi ne t he pot ent i al noi se and cur r ent noi se c o mp o n e n t t o obt ai n t he r at i o
val ue or " r e s i s t a nc e noi s e " ( al t hough in par t i cul ar c i r c ums t a nc e s ei t her c o mp o n e n t c a n be
a s s ume d t o be essent i al l y cons t ant be t we e n def i ned l i mi t s). I t i s t her ef or e ar guabl e t hat
el ect r ochemi cal noi se t heor y can be cons i der ed t o be t he gener al case, obt ai ned wi t hout
ext er nal pol ar i zat i on, wher eas convent i onal l i near pol ar i zat i on t heor y is a speci al case, whi ch
e na bl e d pol ar i zat i on t heor y t o b e f or mul at ed, and whi c h ha ppe ne d t o b e di s c ove r e d first.
Noi s e anal ysi s is es peci al l y usef ul f or t hr ee r easons:
1. Cur r ent and vol t age t r ansi ent s r esul t i ng f r om el ect r ochemi cal pr oces s es as s oci at ed wi t h
t he ons et o f cor r os i on act i vi t y ar e det ect abl e l ong be f or e any vi s i bl e d a ma g e i s evi dent
at a me t a l / e n v i r o n me n t i nt er f ace, a l l owi ng t he r i sk o f at t ack t o be i dent i fi ed, and en-
abl i ng t i mel y and ef f ect i ve r e me di a l act i on t o mi ni mi z e cor r os i on d a ma g e t o t he pl ant .
2. The cur r ent and vol t age noi se t r ansi ent s can be eval uat ed t o char act er i ze and quant i f y
l ocal i zed cor r os i on be ha vi or (pi t t i ng cor r osi on, cr evi ce at t ack, st r ess cor r os i on cr acki ng,
cavi t at i on, and under - depos i t at t ack) .
3. El e c t r oc he mi c a l vol t age and cur r ent noi se si gnal s are obt ai ned i n r eal t i me.
The i mpor t ance o f r eal - t i me i nf or mat i on has been ove r l ooke d i n convent i onal me t hods o f
cor r os i on cont r ol , whi ch i nst ead e mpha s i z e d a ma g e l i mi t at i on. The avai l abi l i t y o f accur at e
and r epr esent at i ve r eal - t i me cor r os i on i nf or mat i on f aci l i t at es pr eci s e onl i ne pr oc e s s cont r ol ,
whi c h was di ffi cul t or i mpos s i bl e t o achi eve by i nf er r i ng cor r os i on st at us b y eval uat i on o f
cumul at i ve cor r os i on r at es; or moni t or i ng o f i ndi r ect pa r a me t e r s as t emper at ur e, pr essur e,
pH, or cor r os i on pot ent i al moni t or i ng; or even onl i ne c he mi c a l anal yses. I mp r o v e d cont r ol
is t he mos t cos t - ef f ect i ve a ppr oa c h b y whi c h t o r educe oper at i ng and ma i nt e na nc e cost s,
mi ni mi z e downt i me, i ncr eas e pr oduct i vi t y, i mp r o v e pr oduct qual i t y, and i ncr eas e safety.
EPRI Project Revi ew
I n 1984, t he El ect r i c Powe r Re s e a r c h I nst i t ut e ( EPRI ) b e c a me i nt er est ed in t he c o mb i n e d
el ect r ochemi cal i mpe da nc e and el ect r ochemi cal noi se i nst r ument at i on. Ini t i al l y, EPRI us ed
t he i ns t r ument at i on t o i dent i f y whe n and why condens at e cor r os i on t ook pl ace in fl ue ga s
des ul f ur i zat i on ( FGD) s ys t ems . I t s successf ul us e al l owed oper at i ng cost s t o be r educed b y
s peedi ng t he de ve l opme nt o f ef f ect i ve r emedi al st r at egi es and al l owi ng t hei r ef f ect i venes s t o
be ver i f i ed whi l e t he pl ant wa s onl i ne. Lat er, EPRI us ed s i mi l ar i ns t r ument at i on i n ot her
p o we r pl ant appl i cat i ons. The s e pr oj ect s ar e des cr i bed bel ow.
LaQue Center for Corrosion Technology/R. M. Schahfer Generating Station
Be f or e a f ul l - scal e pl ant trial, t he f easi bi l i t y o f onl i ne cor r os i on moni t or i ng in condens i ng
fl ue ga s condi t i ons wa s de mons t r a t e d at t he La Que Ce nt e r f or Cor r os i on Technol ogy, i n
Wr i ght s vi l l e Beach, Nor t h Car ol i na, usi ng a pi l ot - s cal e we t l i mes t one F GD s ys t e m t hat had
been bui l t f or mat er i al s t est i ng.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
176 ELECTROCHEMICAL NOISE MEASUREMENT
The instrumentation was commissioned in February 1985. The sensor, manufactured from
Type 316L stainless steel, was located immediately downstream of the absorber column. It
operated in a continuously wetted condition and was occasionally partially covered with
deposits, comprising ash (introduced deliberately to simulate plant slurries) and limestone
dust particles. Electrochemical noise transients typical of the onset of pitting corrosion were
observed, and the extent of corrosion was seen to depend on operating variables, such as
flue gas temperature.
After a three-month exposure, the sensor was removed for inspection by project staff who,
on the basis of electrochemical noise data, had been confidently predicting the onset of pitting
corrosion, but no pitting was evident! The sensor was disconnected from the rig and taken
to the laboratory for more detailed inspection. Examination on a stereo optical microscope
at 100 magnification revealed that pitting had, in fact, occurred. This was the first success
on the EPRI online monitoring program, and a second stage of work for plant investigation
was authorized [8,9].
The instrumentation was relocated on a dual alkali FGD unit at the R. M. Schahfer Gen-
erating Station, operated by the Northern Indiana Public Service Company (NIPSCO). Con-
siderable corrosion damage was being sustained by the outlet flue gas ducting on the FGD
system. The ducting was constructed from Type 317LM stainless steel, and substantial pitting
attack had been observed after nine months of operation. Station engineers wished to un-
derstand the underlying cause of attack and formulate effective remedial measures.
The location of the damage was unexpected. It was most marked just downstream of the
reheater, where the gas stream should have been dry. It was important to establish when
these duct surfaces became wetted with condensate, so a duct wall sensor was installed to
provide a continuous indication of the corrosion status. It was located immediately down-
stream of the reheater approximately 4 ft (1 m) above the duct floor (Fig. 1). It was soon
FIG. 1--External view of corrosion-monitoring probes located in duct at R. M. Schahfer Generating
Station.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
SYRETT AND COX ON ELECTROCHEMICAL NOISE INSTRUMENTATION 177
confirmed that the exhaust flue gas, although reheated to the target outlet temperature, was
not dry during much of the operating time. An example of the traces provided by the in-
strumentation at this installation is shown in Fig. 2. The outlet duct, on this occasion, sus-
tained damage for almost a complete 8-h night shift, even t hough the outlet gas temperature
was considerably higher than should have been necessary to dry it [10]. Calculations revealed
that, i f the flue gas could be discharged in the unreheated state, corrosion rates were actually
much lower and cost savings approaching $8 million could be achieved over a ten-year
operating period by the use of lower cost construction materials and improved thermal
efficiency.
S c h o l z Ge n e r a t i n g C e n t e r
In October 1986, EPRI adopted the online monitoring concept to assist in the investigation
of corrosion phenomena in cyclic reheater systems for FGD plants. Using two inline, tube-
type heat exchangers, a cyclic reheat system extracts heat from hot flue gas before the gas
enters a wet sulfur dioxide (SO2) scrubber and returns the heat to the cooler clean gas after
the FGD system. This process increases gas buoyancy and prevents condensation. The service
environment of the first-stage heat exchanger (heat extractor) is harsh, and a materials in-
vestigation was in progress in a pilot cyclic reheat unit, installed at Gul f Power Company' s
Scholz Generating Center, to identify candidate alloys and coatings for this application.
Over the course of the trials, it has been reported that many alloys sustained "localized
corrosion. " This finding was unexpected on Alloy 276, which in other trials had exhibited
r
i g ~ D ~
r
l ~ _ !
i , r ~l v
_ I M _ u
I - - M - - ~ M
!
I~ % , i | ~11* t , # ~ . "
- ~ 1 - - t l l
_ u _ |
. . . . . . . . . . . . . . . ' . . ' : - " " . t . . . . . . . . . . . . . . . .
m . . q m . . I
* i t
I
I
& t - - I
I
FIG. 2 - - S u b s t a n t i a l i ncrease in corrosi on act i vi t y c aus e d by a decr eas e in t he reheat t empera-
t ure ri se f r o m 25 to 14~ i dent i f i ed by al l corrosi on moni t ors. ECN = el ect rochemi cal cur r ent noi se,
EPN = el ect rochemi cal pot ent i al noi se, ZRA = zero resi st ance ammet ry, and El M = el ect rochemi cal
i mpedance measurement .
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
178 ELECTROCHEMI CAL NOISE MEASUREMENT
sat i sfact ory performance. It was deci ded that a real -t i me i ndi cat i on of corrosi on act i vi t y
woul d be helpful, and t wo of the 1-in. (25. 4-mm) di amet er test heat exchanger t ubes were
repl aced by sensor assembl i es of si mi l ar di mensi ons. The t ubes were modi fi ed to enabl e t he
sensi ng el ement s to be reposi t i oned at di fferent l ocat i ons across the gas st ream (Fig. 3).
It was qui ckl y evi dent that the unexpect ed damage resul t ed from low gas-fl ow r egi mes
adj acent to the t ube sheets, whi ch al l owed ash deposi t s to accumul at e. The deposi t s i nsul at ed
the tube sheets, pr omot i ng sulfuric aci d condensat i on. The sulfuric aci d caused r api d general
corrosi on attack beneat h the deposi t . The at t ack was uni form in nature but was onl y sust ai ned
over a l i mi t ed ar ea [11]. The di fferences in corrosi on performance bet ween seam- wel ded and
unwel ded Al l oy 625 mat eri al were then eval uat ed [12].
The Hi gh Sul f ur Test Cent er
In a cont i nuat i on of t he R. M. Schahfer program, a series of compar at i ve trials commenced
in 1988 at the EPRI Hi gh Sul fur Test Cent er near Buffalo, New York. The pur pose of t hese
trials was to eval uat e candi dat e al l oy mat eri al s for cost -effect i ve retrofit in FGD out l et flue
gas duct i ng. A t wel ve- sensor compound exposur e assembl y was i nst al l ed i mmedi at el y down-
st ream of the mi st el i mi nat or in the outlet sect i on of the 4- MW pi l ot FGD syst em (Fi g. 4).
The scrubber was not fitted wi t h a gas-reheat i ng facility, so the sensors were t emper at ur e
cont rol l ed to al l ow cool i ng and heat i ng of the sensor el ement s. The t i me for t he onset of
at t ack on each of the exposed al l oys and t he degree to whi ch t hey repassi vat ed aft er a change
in the servi ce envi ronment were investigated. In a second phase of the work, smal l quant i t i es
of chemi cal s were added to the scrubber l i quor to i ncrease scrubbi ng efficiency or cont rol
scal i ng charact eri st i cs. Because car r yover of the scrubber l i quor into the outlet duct was
possi bl e, t he effects of the addi t i ves on t he rat e and mor phol ogy of at t ack were al so inves-
t i gat ed. These results were report ed in 1993 [13,14].
An i mport ant finding of the EPRI tests di scussed so far was that corrosi on damage oft en
occurs in operat i ng pl ant s because actual servi ce condi t i ons depar t si gni fi cant l y f r om desi gn-
Watar
J J
I rm'rf'rrm'rrrrrrm'rMl i I
/ / - - "
adi~.m~)
d P i l o t S c a l e \
I C y c l i c R e h e a t
U n i t
I = H o, .I
, I I , ~
t k~. ~
FIG. 3--General layout of the corrosion-monitoring installation in the pilot-scale cyclic reheat unit
at Scholz Generating Center.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
SYRETF AND COX ON ELECTROCHEMICAL NOISE INSTRUMENTATION 179
FIG. 4--Corrosion probes installed in the FRP duct in the 4-MW pilot scrubber at EPRI's High Sulfur
Test Center near Buffalo, New York.
operating conditions. Ducts can be operated satisfactorily with low-cost materials, such as
plain carbon steel, if adequate reheat is provided at all times and the outlet ductwork and
stack always operate hot and dry. Systems constructed from the lower-grade stainless steels
can operate satisfactorily if the system is operated wet, without reheat, provided they are
adequately drained. Ducts operated under alternating wet and dry conditions accumulate acid,
which, in combination with dissolved salts, is highly corrosive. Highly corrosion-resistant
alloys can be used to minimize unplanned down time, but the cost for this insurance is high.
Indianapolis Power and Light Company
Indianapolis Power and Light Company' s (IPL's) Petersburg Generating Station Unit 3
participated in a trial. The station uses fossil fuels, and the unit is fitted with a limestone
slurry FGD system. In 1988, there was concern that an organic lining on the chimney stack
had sustained damage. Budget provision had been made to reline it with a corrosion-resistant
nickel base alloy or titanium alloy. A monitoring investigation was funded by EPRI to
establish if the relining was necessary. Instrumentation from the cyclic reheater program at
Gulf Power in Florida was transferred to IPL.
Preliminary trials demonstrated that corrosion could be avoided provided the gas reheater
systems were online and mist eliminators operated satisfactorily. A permanent surveillance
system was installed in 1990. The system was comprised of six sensors (see Fig. 5 to review
sensor layout). An online control room display informed plant operations personnel of cor-
rosion risk in stack, thereby prompting remedial action. The monitoring system was the
subject of an EPRI "First Use" publication, which described how knowledge of the corrosion
activity allowed IPL to avoid relining the stacks and duct [15]. The resulting investment
saving was $3.2 million and savings in operating and maintenance costs of $0.5 million
would be achieved every five years.
Compressed Ai r Energy Storage
Continuous electrochemical monitoring was used in connection with an EPRI-sponsored
program in which materials were being considered for use in a recuperator to be fitted in a
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
180 ELECTROCHEMICAL NOISE MEASUREMENT
FIG. 5--Schematic diagram of the Petersburg Unit 3 exhaust train showing location of corrosion-
monitoring points.
compressed energy storage system (CAES) at Al abama Electric Cooperative, Inc. A CAES
plant uses off-peak power to compress air into an underground reservoir. During peak de-
mand, the air is heated and released into an expansion turbine to produce electricity. Use of
a recuperator to preheat the compressed air with turbine exhaust heat can reduce operating
costs by 27%. However, the intermittent operation can put the recuperator at risk from
dewpoint corrosion. The program, which commenced in 1986, was managed by Encot ech
Inc. of Schenectady, New York, and the initial stages of work were conducted in a pilot-
scale recuperator at the Creare R&D Company in Hanover, New Hampshire.
Carbon steel was the onl y cost-effective material for the installation, but the normal tem-
perature excursions during operation allowed sulfuric acid condensation and corrosion. It
was decided, therefore, that instrumentation used on the pilot rig for three years would be
reconfigured to provide continuous surveillance on the full-scale recuperator. The Maclntire
CAES installation at Al abama Electric Cooperative, Inc. and its corrosion surveillance in-
strumentation were commi ssi oned in 1991 [16,17].
High-Temperature Monitoring
Field trials of an online high-temperature monitoring system commenced at Rougel ey
Power Station in England in 1986 [18]. The first such installation in the United States was
on behalf of Westinghouse in 1990 at a Waste-to-Energy plant in Panama City, Florida [19].
In 1992, EPRI financed a long-term, high-temperature corrosion trial on a 660- MW boiler
at the Drax Power Station operated by National Power in Yorkshire, England.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
SYRETT AND COX ON ELECTROCHEMICAL NOISE INSTRUMENTATION 181
The Drax project was an investigator of the comparative corrosion performance of super-
heater and rebeater tube materials. An air-cooled coupon exposure lance was adapted to
incorporate electrochemical sensors for t wo candidate materials, and a twin-channel moni-
toring system was used to interrogate the samples. The results were evaluated along with
data obtained from the mass-loss coupon tests [20,21]. An interesting difference in corrosion
behavior was exhibited by the t wo candidate materials after a high-temperature excursion.
The corrosion rate of Type 316 stainless steel increased immediately after the excursion,
whereas on the Type 310 material, the rate decreased (Fig. 6). This information was not
evident from the conventional methods of cumulative metal loss assessment. Electrochemical
data also revealed that the predominant corrosion mechanism was sulfidation/oxidafion dam-
age and not the molten salt attack predicted. Inspection of the corrosion product films at the
completion of the trial confirmed this result.
Stress Corrosion Cracking
The most recent EPRI monitoring feasibility study was conduct ed in collaboration with
PowerGen at the Ratcliffe Technol ogy Centre and Cot t am Power Station. Work in other
fields had indicated that it was possible to detect the electrochemical transients associated
with the onset and propagation of stress corrosion cracking (SCC). Slow crack growth is a
significant concern in certain steam turbine designs. The ability to detect the onset of crack-
ing in real time potentially offers considerable i mprovement in service life prediction and
safety.
The EPRI / Power Gen program had two primary goals: to characterize the onset and prop-
agation of SCC and to transmit the signals across a brushless contact interface as woul d be
necessary i f the data were to be obtained from a power turbine rotor while in service.
The first part of the program was conduct ed at the Ratcliffe Technol ogy Centre laboratories
of PowerGen. Characteristic SCC transients were obtained from a compact tension specimen
made from a rotor steel and exposed to caustic soda solution and then similar transients were
monitored in condensing steam. The electrochemical noise transients were then transmitted
across a brushless contact transducer without degradation of the signal. Finally, the instru-
mentation was moved to Cottam Power Station, and high-speed SCC transients were mon-
itored on a 550- MW power generation turbine without loss of signal integrity [22].
The Future
Each of the EPRI projects described above was successful. At each stage, greater confi-
dence has been developed in the technology, and the power industry is able to take advantage
of the cost savings available [23]. Practical experience is assisting the development of noise
theory [24,25], which has assisted in the understanding of empirical data obtained during
the early EPRI trials.
The feature that distinguishes electrochemical noise instrumentation from earlier corrosion
monitoring t echnol ogy is its capability to provide an online real-time indication of conditions
at a corrodi ng surface. Traditional corrosion monitoring methods do not offer "cont rol , " but
instead offer "accountability. " The resulting damage limitation strategy is consequently ex-
pensive. Real-time feedback allows control of corrosion to be redefined in terms of detection,
prediction, and prevention. For process control purposes, it is more useful t o detect the
threshold onset of corrosion than to quantify the rate and morphol ogy of attack. Plant inci-
dents can then be correlated with the occurrence of corrosion. The approach allows imprecise
data to be processed to produce an effective and robust corrosion control strategy, enabling
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
182 ELECTROCHEMI CAL NOI SE MEASUREMENT
0. 4
0. 35
0.3-
, ~ o. 2s-
15
-~ o. 2-
: t
> .
= 0.15-
o
u 0.1
0.05
0
23/ I I
Alloy .316
, 2 2 o , 0 , ; ' 0 3 2 2 ) ' 0 , , , ; o , 5 , ) ' o ,
Oate
I I I I
I I I
T ~ t r o k e ' e ~ I ITch | l mpl r ol v r l ( ~ u r ~ Tl e-eem~rl ~ 2
t s/ os
0. 018
0. 016-
O.014-
E
E
' - - ' 0. 012-
o
~ 0.01
o
2 0. 008
!
~ 0.006-
~ 0. 0O4-
0. 002-
0
25/11
Alloy 310
12/01 0 ~ o 5 2 2 ~ 0 , , , ~ 0 6 5 , ; o 7
Oote
I I I 1
I I I
Tm~perdvre Scan I ~ Tenw4~kare Cxcurm Temeerakare S~n 2
FIG. 6--Cumulative metal loss profiles of Alloys 316 and 310.
t s/ os
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
SYRETI" AND COX ON ELECTROCHEMICAL NOISE INSTRUMENTATION 183
prompt and effective remedial measures to be initiated before measurable damage has been
sustained.
Open Systems
It is appropriate to consider the developing architecture of noise-based corrosion surveil-
lance instrumentation. The main components of a plant control and information display
system are shown in Fig. 7 [26]. The system consists of:
9 An information-gathering and plant control front-end that is distributed and connected
together using local area network (LAN) technology. The operator interfaces consist of
video display unit (VDU) screens with an interactive pointing device (mouse, tracker
ball, and touch screen), connected to the input/output ( I / 0) subsystems via LAN tech-
FIG. 7--Process plant data-gathering and information technology installation.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
184 ELECTROCHEMICAL NOISE MEASUREMENT
nology. This system forms the primary instrumentation layer and could be a proprietary
distributed control system (DCS).
9 The DCS system. This system can be supported by engineering configuration worksta-
tions, and, on may plants, a process data-archiving comput er is also provided. Such
facilities provide l ong-t erm storage of all measured plant variables.
9 Networked personal computers. These provide plant-based engineers and managers with
productivity tools, such as word processors, spreadsheets, databases, and electronic mail.
9 Remot e systems. These are used by head office personnel to interrogate plant systems,
company-wi de efficiency engineers, or senior company management who require access
to the day-t o-day status of operating systems to determine optimal company strategies.
New systems must be smoothly integrated with the existing infrastructure. "Open syst em"
control and instrumentation (C&I) architecture promotes system integration, and easy use
speeds introduction of new t echnol ogy solutions. System consistency is of particular impor-
tance, and compliance with the emerging system standards in these areas helps to avoid
premature obsolescence of the monitoring system.
For anticipated benefits to be realized, a corrosion management system must impact op-
erating and maintenance practice. I f this is not done, instrumentation and associated main-
tenance and spares support increase the operating cost of the plant. This process is understood
in the vibration-monitoring field. Here, the installation of a monitoring system is normally
preceded by a structured evaluation and audit program that defines potential cost savings
and compares t hem with the actual costs of the proposed condition-monitoring installation.
This procedure is not well established in the corrosion field. Corrosion control has become
the province of control instrumentation groups and operations engineers. Modem electro-
chemical corrosion surveillance methods interfaced effectively to standard C&I systems are
able to achieve continuing real-time savings.
Conclusions
During the past decade, demonstrations of electrochemical noise technology were under-
taken, in which EPRI played a significant role, on behalf of the U.S. power generation
industry. During these investigations, the initial uncertainties were explored, evaluated, and
clarified. Each case history discussed in this paper gave considerable insight into corrosion
problems previously considered to be complicated or not amenable to control. Modem elec-
trochemical surveillance systems are robust and can be used to provide a retrospective cu-
mulative corrosion rate estimate, but the benefits of real-time feedback of corrosion status
are considerably more valuable. Earlier hardware and software that were not based on stan-
dard instrumentation architecture are being replaced by equipment that incorporates "open
syst em" modular principles, which will assist successful technology uptake.
Acknowledgments
The authors wish to thank colleagues in CML, EPRI, and the numerous client organiza-
tions for support, advice, and encouragement during the projects mentioned in this paper.
References
[1] Moore, W., Willet, M. J . , Mok, W. Y., and Cox, W. M., "An Approach to Reducing Stack Losses
in Coal Fired Power Boilers," paper presented at the EPRI Symposium on Improved Technology
for Fossil Plants, Washington, March 1993.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
SYRE'I-[ AND COX ON ELECTROCHEMICAL NOISE INSTRUMENTATION 185
[2] Cox, W. M., M.Sc. dissertation, University of Manchester Institute of Science and Technology,
Manchester, UK, 1979.
[3] Hladky, K. and Dawson, J. L., Corrosion Science, Vol. 22, 1981, p. 317.
[4] Hladky, K., Callow, L. M., and Dawson, J. L., British Corrosion Journal, Vol. 15, 1980, p. 20.
[5] Dawson, J. L. and John, D. G., Journal of the Electrochemical Society, Vol. 110, 1980, p. 37.
[6] Hladky, K., UK Patent GB0084404; USA Patent 4575678; Canadian Patent 1210066.
[7] Eden, D. A., Dawson, J. L., and John, D. G., UK Patent 5139627: European Patent 0302073.
[8] EPRI Report CS-5605, Multitechnique Corrosion Monitoring in Flue Gas Desulfurization Systems:
Phase I, Jan. 1988.
[9] Cox, W. M., Phull, B. S., Wrobel, B. A., and Syrett, B. C., "On-line Monitoring of Desulfurization
Scrubber Corrosion with Electrochemical Techniques," Materials Performance, Vol. 25, No. 11,
Nov. 1986.
[10] EPRI Report CS-5734, Multisystem Corrosion Monitoring in FGD Systems, Phase 2, April 1988.
[11] "'Multisystem Corrosion Monitoring in a Cyclic Reheat Test Facility: Phase 1," Report CS-5776,
Electric Power Research Institute, April 1988.
[12] "Multisystem Corrosion Monitoring in a Condensing Flue Gas Heat Exchanger: Phase 2," Report
GS-7540, Electric Power Research Institute, Oct. 1991.
[13] Mok, W. Y., Moore, D. C. A., Cox, W. M., and Syrett, B. C., "On-line Electrochemical Corrosion
Assessment of Materials for FGD Outlet Ducts," Proceedings of the NACE Air Pollution Seminar,
Orlando, FL, 1992.
[14] "On-Line Corrosion Monitoring at the High Sulfur Test Center," Report TR-102538, Electric Power
Research Institute, June 1993.
[15] EPRI "First Use" publication, Electric Power Research Institute, Nov. 1990.
[16] "Corrosion Testing of Materials in a Simulated CAES Plant Exhaust Atmosphere," Final Report
for EPRI Project RP2488-04, Jan. 1993.
[17] Cook, E. T., Lucas, H., Mok, W. Y., Elmore, D., and Mehta, B., "The Development of an On-Line
Electrochemical Corrosion Monitoring System," Paper 93-JPC-GT-4, Joint Power Conference, Kan-
sas City, KS, Oct. 1993.
[18] Farrell, D. M., Mok, W. Y., and Pinder, L. W., "On-line Monitoring of Furnace Wall Corrosion in
a 125 MW P.F. Boiler," paper presented at the 2rid International Symposium of High Temperature
Corrosion of Advanced Materials and Coatings, Les Embiez, France, May 1989.
[19] Whitlow, G. A., Mok, W. Y., Gallagher, P., Cox, W. M., Elliott, P., and Lee, S. Y., "On-Line
Materials Surveillance for Improved Reliability in Power Generation Systems," NACE Corrosion/
91, Cincinnati, OH, March 1991.
[20] "Hi gh Temperature Fireside Corrosion Monitoring in the Superheater Section of a Pulverized-Coal-
Fired Boiler," Report TR-101799, Electric Power Research Institute, Dec. 1992.
[21] Mok, W. Y., Cox, W. M., Stokes, P. S. N., Chamberlain, J. R., and Bakker, W. T., "On-Line
Corrosion Surveillance in the Superheater Section of a Coal-Fired Power Boiler," EPRI Symposium
on Improved Technology for Fossil Plants, Washington, DC, March 1993.
[22] "Feasibility of On-Line Monitoring of Stress Corrosion Cracking in Rotating Components," Report
TR-102537, Electric Power Research Institute, June 1993.
[23] Syrett, B. C., "Corrosion Control in Electric Power Plants," Materials Performance, Vol. 31, No.
8, Aug. 1992, pp. 52-61.
[24] Bierwagen, G., "Theoretical Derivation of Resistance Noise," Journal of the Electrochemical So-
ciety, 1994, submitted for publication.
[25] Eden, D. A., "A Probabilistic Model of Electrochemical Noise Generation," presentation to ASTM
Committee GI , Atlanta, GA, June 1993.
[26] Collier, D. and Cox, W. M., "Modem Plant Corrosion Surveillance Systems," Institute of Electrical
Engineers, London, March 1994.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
A. M. Br ennenstuhl , 1 Bl ai r Si m, ~ and Renata Cl audi 2
A Progress Report on the Use of
Electrochemical Noise to Investigate the
Effects of Zebra Mussel Attachment on the
Corrosion Resistance of AISI Type 304
Stainless Steel and Carbon Steel in
Lake Water
RE F E RE NCE : Brennenstuhl, A. M., Sim, B., and Claudi, R., " A Pr o g r e s s Re p o r t o n t he
Us e o f El e c t r o c h e mi c a l No i s e t o I nv e s t i g a t e t he Ef f e c t s o f Ze br a Mu s s e l At t a c h me n t o n
t he Co r r o s i o n Re s i s t a n c e o f AI S I Ty pe 3 0 4 St a i nl e s s St e e l a n d Ca r b o n St e e l i n La k e
Wat er , " Electrochemical Noise Measurement for Corrosion Applications, ASTM STP 1277,
Jeffery R. }(earns, John R. Scully, Pierre R. Roberge, David L. Reichert, and John L. Dawson,
Eds., American Society for Testing and Materials, 1996, pp. 186-200.
ABS TRACT: The electrochemical noise technique was used to determine the effect of zebra
mussel settlement on the corrosion performance of AISI Type 304 stainless steel and carbon
steel (ASTM A53 Grade B). These materials represent alloys commonl y used for handl i ng
untreated Great Lakes water at Ontario Hydro' s power generating plants. Thi s work was mo-
tivated by a concern that zebra mussel settlement will lead to accelerated attack of these
materials as a result of the establishment of stable crevice conditions and the growth of cor-
rosion influencing anaerobic sulfate-reducing bacteria (SRB). Corrosion monitoring was carried
out in a field test facility that uses the same untreated Lake Erie water as Ontario Hydro' s
Nanticoke Thermal Generating Station. The test program extended from May through Decem-
ber 1993. During this period, a number of electrochemical parameters were monitored simul-
taneously, including coupling current, electrochemical potential noise (EPN), electrochemical
current noise (ECN), degree of localization (DOLL and resistance noise (R,). Differences were
observed in the performance of the control samples and the samples to which mussels were
attached. The results for the AISI Type 304 stainless steel suggested that over the period
monitored, mussel at t achment reduced corrosion activity. Similarly, signals from carbon steel,
samples exposed to mussels, although initially displaying relatively hi gh corrosion rates, ex-
hibited less corrosion damage than did control samples over the longer term. The reason for
this difference in performance is not known but is considered to have resulted from a change
in the surface envi ronment as a result of mussel attachment, which appeared to di mi ni sh
corrosion. One possible explanation may be the generation of inhibitive species by the mussels.
KE Y WOR D S : coupling current, potential noise, current noise, degree of localization, resis-
tance noise, freshwater, zebra mussels, sulfate-reducing bacteria, corrosion monitoring
Research scientist, Materials Technology Unit, and technologist, Sustainable Development, respec-
tively, Ontario Hydro Technologies, 800 Kipling Ave., Toronto, Ontario, M8Z 5S4 Canada.
2 Senior scientist, Environmental Department, Ontario Hydro, 700 University Ave., Toronto, Ontario,
M5G l X6 Canada.
Copyrighl 9 1996 by ASTM lntcrnational
186
WW'vV.aStIll.OFg
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
BRENNENSTUHL ET AL. ON EFFECTS OF MUSSELS 187
The accidental introduction of zebra mussels into the Great Lakes has led to problems in
conveying water to process systems that use this important resource. Settlement of zebra
mussels on the inside surface of pipes that extract and discharge water from the lakes has
resulted in equipment outages as a result of reduced flow. In addition to the severe fouling,
there are also concerns that the crevice conditions created beneath the mussels could also
lead to accelerated corrosion of metallic substrates. Zebra mussel attachment may result in
the formation of differential aeration conditions. It was feared that mussels could cause the
establishment and growth of anaerobic sulfate-reducing bacteria (SRB) beneath the attach-
ment site. To ensure the continued reliable operation of process equipment and formulate
effective remedial measures, should these have been found necessary, it would be helpful if
the effect of zebra mussel settlement on the corrosion performance of metallic materials
exposed to untreated lake water was better understood.
Corrosion tests were conducted at Ontario Hydro' s (OH' s) Nanticoke Thermal Power Gen-
erating Station. This power station is located on the shores of Lake Erie and has experienced
problems associated with heavy zebra mussel fouling. Electrochemical instrumentation was
used to investigate the effects of zebra mussel settlement on materials commonly exposed
to untreated lake water. Carbon steel (ASTM A53 Grade B) and AISI Type 304 stainless
steel samples were monitored.
This paper presents the experimental methods used and the results obtained to date during
the field exposure. Consideration is also given to the possible causes of electrochemical
activity observed during the program.
Experimental Procedure
Test Facility and Test Conditions
The exposure trials were conducted in a specially built facility located at the Nanticoke
Power Generating Station on the shores of Lake Erie. Figure 1 is a schematic of the test
setup. The test samples were held in aquaria. One aquarium contained zebra mussels. The
FIG. 1--Diagram of the Nanticoke zebra mussel corrosion test side stream facility.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
188 ELECTROCHEMICAL NOISE MEASUREMENT
TABLE l ~Appr oxi mat e analysis o f lake wat er (Lake Erie)
chemistry parameters at Nanticoke Thermal Power Generating
Station.
pH 8.1
CI, mg/ L 20.5
SO4, mg/ L 25.0
Alkalinity, mg CaCO3/L 99.0
Total iron, mg/ L 0.08
Specific conductivity, mS / cm 293.1
bot t om of this was carpet ed wi t h adul t mussels. A second aquari um, the " cont r ol " aquari um,
was mai nt ai ned essent i al l y mussel free, al t hough mussel vel i gers di d ent er the cont rol aquar-
i um duri ng the exposure per i od and became est abl i shed on the surface of the coupons.
However, the ar ea of coverage was very low ( <5%) and the mussel s were very smal l , less
than 2 mm in length. Unt reat ed wat er t aken from the station i nt ake was passed over the
sampl es at a nomi nal rate of 0. 6 L/ mi n. Duri ng the l at t er part of the exposure (at 3700 h),
the flow wat er t hrough the aquari a was st opped for 14 days to det ermi ne t he effect of no
flow condi t i ons on t he el ect rochemi cal response. Because the t ops of bot h aquari a were open,
the wat er had an i nt erface wi t h ai r and thus a suppl y of oxygen. However, the devel opment
of a bi ofi l m over the ent i re sampl e surface shortly aft er the flow was st opped woul d have
severel y l i mi t ed the access of oxygen to the met al surface. The composi t i on of the wat er is
gi ven in Tabl e 1. Wat er t emperat ure was moni t or ed cont i nuousl y and ranged from 8 to 26~
(the t emperat ure of the wat er duri ng the trial per i od is shown in Fi g. 2). Wat er sampl es also
were t aken rout i nel y for pH det ermi nat i on; no si gni fi cant changes were observed.
The test sampl es were i nspect ed on a weekl y basi s to i dent i fy when mussel s at t ached to
the surface and to i nvest i gat e how the mussel densi t y changed wi t h time. Compl et e coverage
30
2 5
s0
' j r
E
1o
5 J
o 1
l i i
2 3 4
Th o u s a n d s
Time/Hours
FIG. 2--Temperat ure variation o f the wat er f l owi ng into the test aquaria during the corrosion-mon-
itoring period.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
BRENNENSTUHL ET AL. ON EFFECTS OF MUSSELS 189
of the surface of a sample with mussels was considered undesirable because this could have
resulted in the development of completely uniform anaerobic conditions and thus an envi-
ronment that is not conducive to significant corrosion [1]. It was suspected that the worst-
case condition would involve the presence of isolated anaerobic areas beneath crustatia and
adjacent mussel-free regions in which oxygen reduction could occur. During these trials,
zebra mussel coverage did not exceed approximately 25% coverage on the stainless steel,
and 50% coverage on the carbon steel.
Sample Design
Figure 3 illustrates the sample configuration used during the experiment. The electrode
assembly was comprised of three tube-coupons, each of which had a surface area of 27 cmL
Electrical isolation between each coupon was achieved with polymethyl methacrylate
(PMMA) insulators. Epoxy resin was used to join the coupons to the insulators. Electrical
leads were spot-welded to each coupon. The welds and the outer surface of the coupons
were then coated with a high-build epoxy paint.
Electrochemical Monitoring
Monitoring was performed using CML Ltd. electrochemical instrumentation. The instru-
ment collected, stored, and performed the following summary statistics:
I. The differential potential between two of each test electrode group.
2. Electrochemical potential noise (EPN)
3. The value of the overall coupling current between the third electrode of each group
and one common electrode.
4. The fluctuation in the coupling current or electrochemical current noise (ECN).
FIG. 3--Three-electrode tube sample used during this study. Samples were fabricated from AISI Type
304 stainless steel and ASTM A53 Grade B carbon steel.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
190 ELECTROCHEMICAL NOISE MEASUREMENT
5. Degree of localization (DoL) derived by dividing the current noise value by the root
mean square (RMS) of the coupl i ng current [2]. This parameter is used to classify
broadly the corrosion process occurring on the sample surface. Values between 0.1 and
1.0 were considered to be an indication of a Poisson-type process, that is, an unstable
system such as pitting, crevice corrosion, or stress corrosion cracking. Values less than
0.1 were assumed to be associated with uniform corrosion.
6. The value of the resistance noise (R,), derived from the ratio of the potential noise and
considered to be analogous to polarization resistance (Rp) in charge transfer resistance
(R, ) obtained by linear polarization resistance measurements and from electrochemical
impedance spectroscopy [2-4].
The instrumentation was multiplexed to interrogate twelve samples semicontinuously (six
control samples and six mussel samples). Each interrogation period consisted of collecting
1024 data points with a separation interval of 4 s. The time domain data were stored on the
hard drive of the logging computer. This cycle continued for the duration of the tests, which
started in May 1993 and continued until December 1993.
Results
A1SI Type 304 Stainless Steel
The results presented here were typical of the observations made for the "cont r ol " and
"mussel " sample covered test electrodes monitored during this study. Good reproducibility
was observed, and the trends described were consistent among samples of similar type.
A DoL plot for a typical control sample is given in Fig. 4. The data indicate relatively
high and erratic fluctuations during the first 500 to 700 h of exposure. Some values ap-
proached the maxi mum of 1. These values subsided to values on the order of 0.1 before
gradually increasing to around 0.3 after 2600 h of exposure, when a sudden large drop in
intensity occurred. This low-amplitude condition persisted for approximately 1200 h, sub-
sequent to which sharp short-duration increases became apparent, a condition that continued
for about 400 h. DoL results from the corresponding mussel sample are also given in Fig.
4. Initially high signal values, typically greater than those observed for the "cont r ol " samples,
were observed, a condition that continued for the first 200 to 400 h. This initial period of
high activity was followed by a period characterized by intermittent, relatively short, tran-
sients with no consistent pattern. After approximately 3500 h of exposure, an increase in the
frequency of the noise fluctuations became evident, and this lasted for 200 to 300 h. This
coincided with the increase in electrochemical activity observed for the control samples. This
increased activity corresponded to a 14-day period during which water was not flowing
through the aquaria.
A typical RMS coupling current plot for one of the "cont r ol " and "mussel " samples is
shown in Fig. 5. Initially, coupling current values for the control samples were relatively low
for all samples. A significant change in RMS coupling current occurred approximately 1000
h into the test, at which time a relatively sharp increase in the coupling current was observed.
For a period of approximately 1500 h after this increase was initially evident, the coupl i ng
current remained fairly steady at around 0.25 I~A. After approximately 2600 h of exposure,
the amplitude and frequency of the outputs increased dramatically, and coupling currents of
up to 5 I~A were reached. This condition persisted for the remainder of the test, despite a
large reduction in the water temperature. Occasional transients were detected, and one of the
largest decreases occurred during the period of no flow, during which the coupl i ng current
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
BRENNENSTUHL ET AL. ON EFFECTS OF MUSSELS 191
0.5 r
0. 4
0201 i '
o _ _ I _ _
0 1 2 3 4 5
Thousands
Time/Hrs
_ _ Control + Zebra
FIG. 4--DoL plot for AISI Type 304 stainless steel showing the outputs for "control" and "mussel"
samples.
values decreased on average to approximately 1 I~A for 300 to 400 h. However, the general
trend was toward increasing coupl i ng current with l onger exposure time. These results con-
trasted with those f r om the "mussel " samples, for which the coupling current values started
higher relative t o the "cont rol s, " peaked at around 2000 h into the test, and then gradually
fell. The peak current for the "mussel " samples was approximately 2 ixA. The no flow
condition at the end of the test did not appear to affect significantly coupling current values.
ECN data for the "cont r ol " sample were similar to the DoL data except that the ECN
values appeared to display the changes in activity more clearly (that is, the sensitivity seemed
to be greater for ECN dat a) dur i ng the last quarter of the exposure period. There was no
sign of the high-amplitude events seen at the start of the tests (Fig. 6a), but high-amplitude,
short-duration transients were evident, during the later stages of the exposure during the
period of no water flow through the aquaria. Figure 6a also shows the EPN data for the
same "cont r ol " sample. There was reasonable correlation between the EPN and ECN results.
A typical ECN trace for the "mussel " samples is shown in Fig. 6b. The plots were similar
to the DoL data for this sample, and good correlation between the ECN and EPN outputs
was evident.
None of the data from both the "cont r ol " and the "mussel " samples appears to display
any dependence on temperature during the exposure period. The increase in electrochemical
noise activity associated with the 14-day no-flow condition appeared greater with the "con-
trol" samples. This increase coincided with a decrease in the coupling current of the control
samples.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
192 ELECTROCHEMICAL NOISE MEASUREMENT
0.02
0.015
0.01
0.005
0
0.02
1 2 3
Thousands
Time/Hrs
__ EPN _r ECN
a)
2.5
2
1.5
1
0.5
0
5
0.015
Z 0.01
0.01)5
2.5
2
1
0.5
o o
0 l 2 3 4 5
Thousands
Time/Hrs
__ EPN + ECN
b)
FIG. 5--Typical ECN and EPN data f or (a) control AISI Type 304 sample, and (b) a "mussel" AISI
304 sample. Note the good correspondence between these two parameters.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
BRENNENSTUHL ET AL. ON EFFECTS OF MUSSELS 193
3
.~ .~ 2
1
.
i
1 2
Thousands
Time/Hrs
3 4
Control + Zebra
FIG. 6--RMS coupling current plot for representative AISI Type 304 "control" and "mussel" samples.
Carbon Steel (ASTM A53 Grade B)
Bot h R, and DoL are presented here because it was believed a compari son of these data
might characterize better the corrosion behavior of the carbon steel samples, although the
latter parameter is normally associated with the evolution of localized corrosion.
Resistance noise outputs for the "cont r ol " samples are shown in Fig. 7. Significant data
scatter was evident for both the "cont r ol " and "mussel " samples. However, the general
tendency was toward an increase in R, (that is, a decrease in corrosion rate) with time for
both sets of samples. It was not possible to distinguish whether the mussel attachment made
any difference in the corrosion rate during the first 1000 h of exposure. However, as exposure
time was increased, activity decreased to a greater extent with the mussel samples.
A typical DoL output for the "cont r ol " and "mussel " samples is presented in Fig. 8. DoL
data for the "cont r ol " samples were generally of a very low amplitude throughout the ex-
periment, with occasional large transients which decayed quickly. These results were not
unexpected because general corrosion was prevalent. The mussel results appeared to differ
slightly from those for the "cont r ol " samples in that relatively hi gh short-duration, high-
value signals were initially observed, some being as high as DoL = 0.8. These high DoL
values were evident during the first 1000 h of the exposure, but subsequently values less
than 0.05 became normal. A very slight increase in activity was detected during the 14-day
no-flow period.
Typical EPN and ECN plots obtained during the study are shown in Fig. 9. A good
correlation between EPN and ECN was evident for the "cont r ol " samples. Bot h the EPN
and ECN signals were relatively low except at the middle and end of the exposure. EPN
outputs up to 1.8 mV and ECN values of up to 2.0 ixA were evident, but average values on
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
194 ELECTROCHEMICAL NOISE MEASUREMENT
20
20
15
10
t
I
1 2 3
Thousands
Time/Hrs
m Sample #1 + Sample #2 __ Sample #3
0 4
a~
r ~
r~
15
10
g~
0 1 2 3 4
Thousands
Time/Hrs
Sample #1 + Sample #2 __ Sample #3
b)
FIG. 7- - R. data f or the carbon steel (a) "'control" and (b) "mussel " samples.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
BRENNENSTUHL ET AL. ON EFFECTS OF MUSSELS 195
0.2
0.15
o 0.1
0.05
0
0 1 2 3 4
Thousands
Time/Hrs
_ _ Control + Zebra
FIG. 8--DoL data for the carbon steel samples showing "control" and "mussel" samples.
the order of 0.15 mV and 0.1 ~A were recorded for EPN and ECN, respect i vel y. Hi gher
values at the end of the exposure coi nci ded wi t h the 14-day no-flow condi t i on, as was t he
case for the st ai nl ess steel sampl es. Noi se dat a from " mus s el " sampl es appear ed to have a
di fferent pattern from t he "cont r ol s. " " Mus s el " sampl es exhi bi t ed si gni fi cant l y hi gher EPN
and ECN than di d t he " cont r ol " sampl es duri ng the first 3000 h of the exposure. The t rend
di d not persi st , and the noi se levels gr adual l y r educed wi t h l onger exposure t i me until it was
l ower than that of cont rol sampl es and st ayed for the r emai nder of the test. No si gni fi cant
i ncrease in act i vi t y was obser ved at the end of the experi ment , in cont rast to behavi or ex-
hi bi t ed by "cont r ol " sampl es.
RMS coupl i ng current pl ot s for t hese carbon steel " cont r ol " and " mus s e l " sampl es are
shown in Fi g. 10 The RMS coupl i ng currents for the " cont r ol " sampl es di spl ayed an erratic
behavior, with no cl ear trends with t i me, and out put ranged from 18 p~A to less than 1 I~A,
wi t h t he mean bei ng around 7 ixA. A slight i ncrease in coupl i ng current was evi dent duri ng
the per i od when the flow to the aquari a was st opped. Resul t s for the " mus s el " sampl es,
unl i ke t hose for the " cont r ol " sampl es, exhi bi t ed a pr onounced t rend in that the coupl i ng
currents started at rel at i vel y low values, t ypi cal l y 10 to 15 ixA, and aft er 700 to 1000 h of
exposure i ncreased sharpl y to a maxi mum of 30 I~A. A fluctuating hi gh- coupl i ng current
prevai l ed until about 1800 h of exposure, and then gr adual l y decr eased to values much l ower
than t hose obser ved for the cont rol sampl es.
The di fference in corrosi on act i vi t y det er mi ned by t he presence or absence of zebra mus-
sels on the surface was mor e pr onounced for carbon steel sampl es than was the case for
st ai nl ess steel, but agai n little evi dence of t emperat ure dependence was seen in ei t her case.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
196 ELECTROCHEMICAL NOISE MEASUREMENT
0. 5
1.5
0. 4
.~ 0. 3
z
t~
m ~ 0 . 2
0.1
0. 5
0 3
Th o u s a n d s
Ti me/ Fi r s
_ _ EPN + ECN
a)
o
0. 5
0
5
1.5
0 . 4
a~ 0. 3
t ~ 4 0 . 2
0. 1
0 1 2 3 4 5
T h o u s a n d s
T i me / Hr s
E P N + E C N
b)
1
0. 5
FI G. 9 - - ECN and EPN data f or carbon steel samples showing ( a) "control" signals, and ( b) zebra
mussel signals. Good correspondence between these parameters is evident.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
BRENNENSTUHL ET AL. ON EFFECTS OF MUSSELS 197
0.035
0.03
< 0. 025
-~ 0. 02
~ ~ 0.015
~A
0.01
0.005
0
o 1 2 3 4
Thousands
Time/Hrs
_ _ Control + Zebra
FIG. IO--RMS coupling current plots for "control" and "'mussel" carbon steel samples.
Di scussi on
Electrochemical noise measurement was used to assist in the investigation into the effects
of zebra mussel attachment on AI SI Type 304 stainless steel and carbon steel ( ASTM A53
Grade B). Bot h materials were exposed to untreated lake water at Ontario Hydr o' s Nanticoke
power generation plant. Carbon steel normally sustains general corrosion, and, as such, its
service life in lake water should be predictable provided the chemistry of the environment
to which the steel is exposed is known. The stainless steel examined during this investigation
is the least corrosion-resistant of the stainless steels used at power stations and has a history
of failing by localized corrosion in untreated natural water. AISI Type 304 is susceptible to
microbiologically influenced corrosion (MIC) when exposed to Lake Erie water [5]. Micro-
environments beneath deposits may support SRB, which can induce severe damage on both
the carbon steel and stainless steel. There was concern, therefore, that the corrosion per-
formance of these alloys might be affected adversely by changes in the environment at metal
surfaces as a result of the attachment of zebra mussels. Crevice formation, and the estab-
lishment of differential aeration cells, might further accelerate corrosion by the establishment
and growth of SRB beneath the mussels. This study investigated the contribution of zebra
mussels to the corrosion behavior of these alloys. Not e that the trial is still in progress, and
that samples have not yet been removed for metallurgical examination.
AISI Type 304 Stainless Steel
The results for the AI SI Type 304L stainless steel show differences between samples that
had mussels attached to the surface and those that were mussel-free. Initially, very little
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
198 ELECTROCHEMICAL NOISE MEASUREMENT
act i vi t y was observed on the cont rol sampl es. Thi s was in sharp cont rast to the mussel
sampl es, whi ch exhi bi t ed al most i mmedi at e activity. The di fference in per f or mance is ra-
t i onal i zed in t erms of the devel opment of crevi ce condi t i ons at the surface beneat h the
mussels, whi ch at t ached to the surface of t he sampl es in less than a week. The est abl i shment
of di fferences in the el ect rochemi cal pot ent i al bet ween t he crevi ce sites and the free met al
surface results in the format i on of di scret e anodi c and cat hodi c regi ons. Not e that the t erm
crevice does not refer to the cl assi cal geomet ri cal condi t i on. Zebr a mussel s at t ach fai rl y
l oosel y to the surface and di spl ay some mobi l i t y. Therefore, the condi t i on is more aki n to
underdeposi t corrosi on. The coupl i ng current for the "cont r ol " sampl es di d not show any
mar ked sust ai ned i ncrease duri ng this period. It can be i nferred that any di ssol ut i on was
qui ckl y fol l owed by repassi vat i on. The per i od of i ni t i al hi gh act i vi t y on the " mus s e l " sampl e
was short -l i ved, and it was suspect ed that the noi se det ect ed at the start of the experi ment
was possi bl y as a result of the di ssol ut i on and repassi vat i on at sites of weakness in the
passi ve fi l m at crevi ce location. These sites may have been caused duri ng the t ube fabri cat i on
and sampl e preparat i on. The act i vi t y di mi ni shes once t he l ocat i ons in whi ch t he film is
defect i ve have been repai red and pr ovi ded that the di ssol ut i on process has not reduced t he
pH in the crevi ce regi on to a level in whi ch pi t i ni t i at i on in " s ound" oxi de occurs. It was
concl uded that act i vi t y evi dent at the start of the exper i ment cor r esponded to t he r epai r of
such defects. Previ ous wor k [6] on the same test faci l i t y to i nvest i gat e the effects of chl o-
rination added credence to this hypot hesi s because chl ori nat i on tests on "as - r ecei ved" AI SI
Type 304 and on i dent i cal sampl es that were et ched before testing exhi bi t ed very di fferent
behavior. The et ched sampl es di spl ayed very little el ect rochemi cal noi se at the start of the
experi ment , whereas the "as - r ecei ved" sampl es exhi bi t ed many met ast abl e pi t t i ng events.
In t he case of the cont rol sampl es, very little noi se was det ect ed dur i ng the first 1000 h.
Previ ous studies i ndi cat ed that it may t ake as l ong as six months to est abl i sh deposi t s that
woul d pr omot e pi t i ni t i at i on [7]. Bet ween 1000 and appr oxi mat el y 2500 h of exposur e an
i ncrease in el ect rochemi cal act i vi t y was di scerni bl e, and the coupl i ng current, ECN, and
EPN were al l hi gher than t hose col l ect ed duri ng the earl y stages of the exposure. These
changes suggest a l arger number of di ssol ut i on repassi vat i on events on the met al surface. A
substantial i ncrease in the coupl i ng current was al so evi dent aft er 2500 h of exposure. Thi s
corresponded to hi gher EPN and ECN outputs and suggest ed that a rel at i ve i ncrease in the
rate of degradat i on was in progress. DoL dat a i ndi cat e that what was t aki ng pl ace duri ng
this per i od was not consi st ent with dat a from the " cont r ol " sampl es at the start of the test.
The DoL outputs were of a l ower ampl i t ude whi ch is i ndi cat i ve of less l ocal i zed attack.
Previ ous moni t ori ng wor k on AI SI Type 304 sampl es in Lake Eri e wat er i ndi cat ed that
ECN and EPN i ncreased with exposure t i me until the l ake wat er t emperat ure decreased
subst ant i al l y [6]. It was al so report ed that SRB l i kel y accel erat e at t ack [5]. An overal l in-
crease in the corrosi on rate of t he " cont r ol " sampl es wi t h t i me was t herefore expect ed, and
this was observed. Thi s di d not appear to be the case for sampl es exposed to mussel s. Af t er
the initial per i od of high noi se intensity, little further act i vi t y was observed, whi ch suggest s
that in the short t erm zebra mussel at t achment di d not have a del et eri ous effect on corrosi on
per f or mance and seemed to reduce corrosi on activity. The reason is unclear, al t hough it
woul d seem from the evi dence obt ai ned to dat e that the presence of zebra mussel s may
condi t i on the envi ronment in such a way that corrosi on is reduced. A possi bl e answer may
be the product i on of i nhi bi t i ve mat eri al s by the mussels. Pol yani oni c pept i des have been
report ed to i nhi bi t corrosi on. Hydr ophi l i c and hydr ophobi c prot ei n fractions are commonl y
found in the cal ci um carbonat e skel et al structures of mari ne organi sms. These can be rich
in pol yani oni c prot ei ns [8]. Cl ues to the cause for this r educed corrosi on may come from
the surface anal ysi s of the sampl es when the t est i ng is compl et ed.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
BRENNENSTUHL ET AL. ON EFFECTS OF MUSSELS 199
The effect of zero waterflow was particularly evident from the control samples. ECN and
coupling current appeared to have decreased, and DoL and EPN increased. The decreases
can be explained by reduced oxygen available for the cathode reaction. A DoL high suggests
high localized activity, which is not consistent with the other outputs. This result highlights
the problems with this monitoring parameter. DoL is frequently at its highest during periods
when the noise resistance increases, that is, with very low rates of attack the noise data are
almost certain to indicate a relatively high degree of localization even if the rate of attack
is low. This problem with inconsistent DoL results is one reason why several noise pa-
rameters were considered during this study. The samples exposed to mussels exhibited almost
no discernible change in electrochemical behavior during the flow stoppage.
Carbon Steel (ASTM A53 Grade B)
The resistance noise data obtained from carbon steel samples were not consistent with the
differences in EPN, ECN, and coupling current data from the control- and mussel-covered
samples. A decrease in the corrosion rate with time was inferred from the R, outputs for
both test conditions. Results from the mussel-exposed samples suggested that the rate of
corrosion decreased at a greater rate with the mussel samples. The high starting electro-
chemical noise for the mussel samples was certainly not evident in the R, plots. The large
scatter in the R, data masked this detail and highlighted the importance of considering EPN
and ECN data outputs in addition to resistance noise values. DoL data, considered to be
only of use for evaluating localized corrosion, yielded information that may be of signifi-
cance. The results suggested that localized activity of mussel samples decreased slightly
during the first 1000 h of exposure. These data indicated that flow stoppage at the end of
the exposure period had less effect than was the case with stainless steel.
On the basis of electrochemical noise data obtained to date during this program, it would
be concluded that zebra mussel attachment may affect corrosion performance of the carbon
steel. Their presence on stainless steel may reduce corrosion. This is suspected to be the
case at least while the mussels are alive, although the behavior may be different with dead
or decaying mussels. The low EPN, ECN, and coupling current, after extended exposure and
relative to the control samples, may be indicative of either the formation of a protective film
or the conditioning of the chemistry of the "lake water" in the immediate vicinity of the
mussels or both, rendering it less corrosive. Again, a possible explanation may be the release
of inhibitive material from the mussels. Surface analysis may confirm if there is a difference
in the composition of the corrosion products.
Summary
Electrochemical noise instrumentation can be used to investigate the differences in cor-
rosion of ASTM 53A Grade B carbon steel and AISI Type 304 stainless steel samples
exposed to zebra mussels in lake water. The electrochemical outputs obtained from these
tests were compared with results from samples that were essentially mussel-free.
Data for the stainless steel revealed differences between the mussel-free and the mussel
environments. The mussel-free stainless steel samples exhibited the expected performance
for this material in untreated lakewater, that is, gradually increasing activity with time. A
different behavior was found with the samples exposed to mussels. After high initial activity,
infrequent short-duration events were detected with no real pattern to the signals. Overall,
the results suggested that zebra mussel attachment does not have a deleterious effect on the
corrosion performance in the short term, and this presence may even reduce corrosion. How-
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
200 ELECTROCHEMICAL NOISE MEASUREMENT
ever, the effect of dead or decaying mussels or both on corrosion performance of the materials
is unknown.
Data for carbon steel revealed erratic outputs for the control samples. No clear trends were
observed. This contrasted with the data obtained from the mussel samples. Relatively high
initial activity was noticed. This peaked after 1800 h of exposure and then gradually de-
creased to very low values. The presence of zebra mussels appeared to inhibit corrosion, at
least during the period of the trials. The difference between the two test conditions was not
obvious from the R, parameter as a result of a very high degree of scatter. RMS coupling
current, EPN, and ECN outputs revealed greater differences between the "control" and
"mussel" samples than was the case with R, assessment. Hence, electrochemical noise data
were observed to identify differences in the behavior of samples that were free of zebra
mussels and those that had mussels attached, but it was concluded that single-output pa-
rameters such as the DoL or R, were insufficient to make a clear discrimination.
Ac k nowl e dgme nt s
The authors would like to acknowledge the helpful discussions and technical support from
Dr. E. G. Leitch and Dr. E Spencer.
Re f e r e nc e s
[1] Newman, R. C., Webster, B. J., and Kelly, R. G., ISIJ Int., Vol. 31, 1990, pp. 201.
[2] Rothwell, A. N. and Eden, D. A., Paper 223, Corrosion' 92, NACE, Nashville, 1992.
[3] Lumsden, J. B., Kendig, M., and Jeanjaquet, S., Paper 224, Corrosion' 92, NACE, Nashville, 1992.
[4] Mansfeld, E and Xiao, H., Journal of the Electrochemical Society, Vol. 140, No. 8, 1993.
[5] Brennenstuhl, A. M. and Gendron, T. S., "The Use of Field Tests and Electrochemical Noise to
Define Conditions for Accelerated MIC Testing," paper presented at the Symposium on Microbio-
logically Influenced Corrosion (MIC) Testing, Miami, FL, Nov. 1992.
[6] Brennenstuhl, A. M., in Proceedings of the 12 'h International Corrosion Congress, Vol. 6, p. 4102.
[7] Brennenstuhl, A. M., Doherty, P. E., Dunstall, T., and Evans, D., in Proceedings of the Fourth
International Symposium on Environmental Degradation of Materials in Nuclear Power Sys-
t ems - Wat er Reactors, 1989.
[8] Mueller, E., Sikes, C. S., and Little, B. J., Corrosion, Vol. 49, No. 10, 1993.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
J ohn I . Mickalonis,' Ri chard J . J acko, 2 Graham P Qui rk, 3 and
Davi d A. Eden 3
The Use of Electrochemical Noise
Measurements with Nuclear Waste Tanks
RE F E RE NCE : Mickalonis, J. I., Jacko, R. J., Quirk, G. R, and Eden, D. A., " T h e Us e o f
El e c t r o c h e mi c a l No i s e Me a s u r e me n t s wi t h Nu c l e a r Wa s t e Tanks, " Electrochemical Noise
Measurement for Corrosion Applications, ASTM STP 1277, Jeffery R. Kearns, John R. Scully,
Pierre R. Roberge, David L. Reichert, and John L. Dawson, Eds., Ameri can Society for Testing
and Materials, 1996, pp. 201-213.
ABSTRACT: Processing of nuclear materials for ei t her military or civilian purposes generates
waste products. The nuclear wastes are generally stored in plain carbon or stainless steel
underground tanks. Corrosive wastes are inhibited to minimize the corrosion of these tanks
and mai nt ai n tank structural integrity under normal and off-normal conditions. Storage tanks
have fairly static conditions, although processing of the waste in these tanks to more stable
forms will produce dynami c conditions, which may lead to localized corrosion. Corrosion
monitoring under either condition, therefore, would ensure t hat inhibitive conditions are
maintained.
A preliminary investigation was conducted to evaluate electrochemical noise measurement s
for corrosion monitoring of waste tanks. The initial results indicated that current and potential
noise correlate to the solution corrosivity. Further development is necessary to differentiate
among the various forms of corrosion.
Electrochemical noise measurements were made in the laboratory wi t h plain carbon steel
electrodes under simulated conditions for storage and processing of nuclear waste. Coupons
were also exposed in the laboratory test vessels. The pri mary aggressive species of the waste
was the nitrate anion, whi ch induces pitting. The simulated waste was initially i nhi bi t ed wi t h
sodium nitrite and sodium hydroxide. The nitrate concentration was increased over the test
period to increase the corrosivity of the solution. The temperature was held constant at 40~
The current and potential noise showed an increasing trend with nitrate concentrations. Under
i nhi bi t ed storage conditions, the noise data were fairly stable. Inspection of the electrodes
revealed little general corrosion with mi ni mal pitting. The coupons, however, had a significant
amount of pitting. This difference was attributed to the microstructure of the electrodes and
coupons.
KEYWORDS: electrochemical noise measurements, pitting, carbon steel, microstructure, nu-
clear waste tanks, nitrate, inhibitors
Nu c l e a r wa s t e r es ul t s f r o m t he ni t r i c aci d di s s ol ut i on of f uel a nd t ar get e l e me nt s dur i ng
t he p r o d u c t i o n of n u c l e a r ma t e r i a l s f or na t i ona l d e f e n s e a nd s pa c e pr ogr a ms . At t he Sa v a n n a h
Ri ve r Si t e ( SRS) , mo s t o f t hi s r a di oa c t i ve wa s t e i s s t or ed i n u n d e r g r o u n d t a nks ma d e o f
va r i ous gr a de s of c a r b o n st eel . Th e t a nks r a nge i n c a pa c i t y f r o m 0. 75 t o 1.3 mi l l i on gal
( 2. 84 t o 4. 9 mi l l i on L). Th e was t e, whi c h i s ne ut r a l i z e d be f or e p l a c e me n t i n t he t anks , i s
s t or ed i n t hr e e f or ms : a s l udge c o n t a i n i n g me t a l oxi des , me t a l c a r bona t e s , a nd mo s t o f t he
' Senior engineer, Westinghouse Savannah River Company, RO. Box 616, Aiken, SC 29802.
2 Westinghouse Electric Company STC, 1310 Beul ah Rd., Pittsburgh, PA 15235.
3 CAPCIS March Ltd., Bainbridge House, Granhy Row, Manchest er M1 2PW, Uni t ed Kingdom.
Copyrighl 9 1996 by ASTM lntcrnational
201
www.astm.org
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
202 ELECTROCHEMICAL NOISE MEASUREMENT
radi oact i ve component s; a sal t cake pr i mar i l y cont ai ni ng sodi um nitrate; and a supernat ant
l i qui d cont ai ni ng sodi um nitrate, hydroxi de, and nitrite al ong wi t h the wat er - sol ubl e radi o-
active cesi um and strontium. Al l wast e forms cont ai n mi nor amount s of ot her chemi cal
const i t uent s such as sulfates and chl ori des.
Ni t rat e was found to be the most aggressi ve ion to the carbon steel t ank [1]. The pr i mar y
corrosi on mechani sms of concern for t he t anks are nitrate stress corrosi on cracki ng and
pitting. The wast e is i nhi bi t ed wi t h sodi um hydr oxi de and nitrite to prevent t hese forms of
corrosi on. General corrosi on of the carbon steel is bel i eved to be i nsi gni fi cant because of
the hi gh pH of the wast e [2]. At SRS, this corrosi on is moni t ored t hrough a pr ogr am con-
sisting of vi deo observat i on of t he t ank i nt eri ors and the surroundi ng annul us spaces, chem-
i cal anal yses of the wast e by grab sampl es, sporadi c use of short -t erm coupon tests, and
l i mi t ed ul t rasoni c t hi ckness measurement s of the t ank wal l s [2]. Trials are al so under way
wi t h a corrosi on coupon ri g for schedul ed, l ong-t erm corrosi on rate measurement s.
The nucl ear wast e wi l l be processed t o st abi l i ze the radi oact i ve component s for l ong-t erm
storage. The hi gh-l evel radi oact i ve component s wi l l be vitrified i nt o a bor osi l i cat e glass.
Low- l evel liquid wast e wi l l cont i nue to be mi xed wi t h grout to form a concret e. Several
tanks, whi ch were desi gned as st orage tanks, wi l l be used to prepare the hi gh-l evel radi o-
active wast e for vitrification. The envi ronment wi t hi n t hese tanks wi l l change f r om static to
dynami c condi t i ons. The dynami c condi t i ons coul d pr oduce operat i ng condi t i ons, such as
concent rat i on fluctuations, t hat coul d pr omot e l ocal i zed corrosi on. Therefore, a cont i nual ,
on-l i ne met hod of moni t or i ng corrosi on duri ng wast e processi ng is necessary t o det ect and
mi ni mi ze any del et eri ous l ocal i zed corrosi on.
An assort ment of cor r osi on- moni t or i ng t echni ques are avai l abl e commer ci al l y or under
devel opment [3]. For wast e t ank appl i cat i ons, a t echni que must be abl e to det ect various
forms of l ocal i zed corrosi on di rect l y on a cont i nual basis. These requi rement s l i mi t ed t he
number of candi dat e t echni ques. El ect r ochemi cal noi se measurement s (ENM) met t he re-
qui rement s. Al t hough avai l abl e commer ci al l y, this t echnol ogy needs further devel opment for
appl i cat i on to the wast e t ank envi ronment .
As part of a cor r osi on- moni t or i ng devel opment pr ogr am at SRS, a pr el i mi nar y l aborat ory
st udy was conduct ed to eval uat e ENM. The resul t s from this st udy show that ENM offers
pr omi se in det ect i ng pi t i ni t i at i on in a si mul at ed wast e solution. Dat a anal ysi s and i nt erpre-
tation t hrough the cal cul at i on of a pi t t i ng funct i on were requi red to det er mi ne the changes
in pi t t i ng suscept i bi l i t y. Duri ng the study, sodi um nitrate was var i ed t hrough a range of
concent rat i ons to i ncrease the corrosi vi t y of the si mul at ed wast e sol ut i on and to i nduce
pitting. The experi ment al results al so showed that t he mi crost ruct ure of carbon steel had a
si gni fi cant effect on t he pi t t i ng suscept i bi l i t y of a sampl e. Furt her devel opment al wor k is
necessary to det ermi ne i f ENM can accurat el y moni t or pi t propagat i on.
Ex p e r i me n t a l Pr o c e d u r e
The pur pose of the l aborat ory st udy was to det er mi ne i f ENM coul d det ect pi t t i ng in a
si mul at ed wast e sol ut i on when the ni t rat e concent rat i on was above t he l i mi t for i nhi bi t i on.
The experi ment al setup, dat a anal ysi s, and mi crost ruct ural anal ysi s are descri bed.
Experimental Setup
The t hree-mont h test was conduct ed wi t h t wo probe assembl i es. One probe assembl y was
pl aced in a covered vessel in whi ch the wast e composi t i on was hel d const ant and wi l l be
referred to as the cont rol t hroughout this paper. The second probe assembl y was pl aced in
a separat e cover ed vessel in whi ch the ni t rat e concent rat i on was i ncreased. Ni t r ogen was
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
MI CKALONI S ET AL. ON ENM WI TH NUCLEAR WASTE TANKS
TABLE 1--Composition of simulated waste solution.
Constituents Concentration, M
Aluminum nitrate 2.05E-2
Iron nitrate 5.50E-3
Sodium hydroxide 9.37E-2
Sodium nitrate 5.18E-3
Sodium nitrite 2.57E-2
Sodium oxalate 8.66E-3
Sodium sulfate 5.94E-3
Sodium carbonate 5.80E-2
Sodium bicarbonate 1.20E-2
Sodium hypophosphate 3.90E-4
Calcium carbonate 6.60E-4
Sodium chloride 1.03E-3
Sodium fluoride 7.30E-4
Sodium chromate 1.40E-4
Chromium chloride 6.00E-5
Sodium molybdate 1.85E-5
Sodium metasilicate 1.60E-4
Uranyl nitrate 3.30E-4
Cobalt nitrate 1.50E-6
Copper nitrate 2.00E-5
Manganese dioxide 1.14E-3
Nickel nitrate 5.20E-4
Mercury nitrate 1.10E-4
Zinc nitrate 4.00E-5
Lead nitrate 2.00E-5
Phenol 4.30E-6
203
slowly bubbled in each vessel to simulate the waste tank operation of maintaining a nitrogen
flow over the waste.
The initial solution, which was used in each vessel, had the simulated waste composition
shown in Table 1. The composition represents an average chemistry expected during SRS
waste processing and is based on the present waste inventory. For this study, the initial nitrate
concentration was set at the highest level for the specified inhibitor concentrations (that is,
the sodium hydroxide and nitrite concentrations) [1]. The nitrate concentration was adjusted
by adding variable quantities (8.8 to 300 mL) of a 5M sodium nitrate solution. The nitrate
additions and estimated resulting nitrate concentrations are given in Table 2.
The pH of the simulated waste was monitored during the test. The pH of the control
solution slowly decreased over the course of the test from 10 to 9.5. This pH drop may have
TABLE 2--Nitrate additions and estimated
nitrate concentrations.
0 0.022
1 "8~8 0.037
2 26.4 0.0797
3 66.0 0.184
4 75.0 0.298
5 300.0 0.704
Final Nitrate
Addition Quantity, mL Concentration, M
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
204 ELECTROCHEMICAL NOISE MEASUREMENT
resul t ed from t he pi ckup of carbon di oxi de when the sol ut i on was exposed to the ai r weekl y.
The pH of the ni t rat e-adj ust ed sol ut i on decr eased f r om 10 to 9.0, at t ri but ed bot h to carbon
di oxi de pi ckup and t he addi t i ons of the 5M sodi um nitrate solution.
Each ENM probe assembl y consi st ed of t hree nomi nal l y i dent i cal cyl i ndr i cal el ect rodes;
t wo el ect rodes were coupl ed to moni t or the current noise, and the third moni t or ed the po-
tential noise. The el ect rodes were made of ASTM A537, Cl ass 1, carbon steel (A537) wi t h
the di mensi ons of 3.1 cm in length and 0.5 cm in diameter. This grade of steel is t he mat eri al
of const ruct i on for t he wast e t anks that are to be used as processi ng tanks. The nomi nal
composi t i on for A537 is 0. 24% carbon, 0.7 to 1.6% manganese, 0. 04% sulfur, 0. 035% phos-
phorus, and 0.15 to 0. 5% silicon. The el ect rodes were prepared wi t h a 600-gri t surface finish
by the manufacturer. The pot ent i al and current noi ses were measured on a CAPCI S Mar ch
Lt d. Deni s V4. 40 system. The noi se si gnal s were measured for 40 mi n every 3.5 h wi t hout
appl yi ng a pot ent i al . Duri ng each measur ement peri od, the soft ware cal cul at ed t he fol l owi ng
st at i st i cal val ues for t he current and pot ent i al noises: mean, st andard devi at i on, and r oot
mean square (rms).
Fl at corrosi on coupons were al so exposed in each vessel. The coupons had a cent ral hol e
for at t achment to the support. The coupons were al so made of A537 carbon steel wi t h the
di mensi ons of 5.1 by 2.6 cm. Several coupons were r emoved from t he test aft er approxi -
mat el y one and a hal f months. The removal was j ust bef or e i ncreasi ng the nitrate concen-
t rat i on to 0.298M.
D a t a A n a l y s i s
Several corrosi on paramet ers were cal cul at ed from t he ENM data: noi se resi st ance, whi ch
is a measure of t he general corrosi on rate; the l ocal i zat i on index, whi ch i ndi cat es the pre-
domi nant corrosi on mechani sm; and a pi t t i ng function, whi ch is an i ndi cat i on of t he si gni f-
i cance of l ocal i zed corrosi on.
The noi se resi st ance, R. , is defi ned as
R, = (rv/(r / (1)
where cry is t he st andard devi at i on of the pot ent i al noi se and ~ is the st andard devi at i on of
t he current noi se [4].
The l ocal i zat i on i ndex, L, is defi ned as
L = t r l / l ~ m ~ (2)
where l ~s is the rms val ue of the current noise. The l ocal i zat i on i ndex fal l s in a range of
0. 0 to 1.0. Values on the order of 0.001 i ndi cat e that uni form general corrosi on is the
pr edomi nant mechani sm, whereas values appr oachi ng 1.0 i ndi cat e the pr edomi nance of a
l ocal i zed mechani sm [5].
Dur i ng t he i ni t i al anal ysi s of the data, the l ocal i zat i on i ndex was found not to di fferent i at e
bet ween passi ve behavi or and l ocal i zed corrosi on in this envi ronment . However, further anal -
ysi s of the ENM dat a t hrough the appl i cat i on of a pi t t i ng funct i on was abl e to di fferent i at e
t hese behavi ors. The pi t t i ng funct i on was empi r i cal l y deri ved as descr i bed below. For br oader
appl i cat i on, the pi t t i ng funct i on must be refi ned and veri fi ed experi ment al l y.
The pi t t i ng funct i on quantifies l ocal i zed behavi or by combi ni ng the l ocal i zat i on i ndex wi t h
the average magni t ude of the current t ransi ent [6]. Therefore, passi vat i on and l ocal i zed cor-
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
MICKALONIS ET AL. ON ENM WITH NUCLEAR WASTE TANKS 2 0 5
rosi on can be di fferent i at ed because passi ve behavi or with smal l er current t ransi ent s wi l l
have smal l er pi t t i ng funct i ons than occur duri ng pi t t i ng.
The pi t t i ng function, P, is defi ned as
P = L {[di(1) + "" + di ( n) ]/ n}
where
L = the l ocal i zat i on i ndex gi ven by Eq 2,
di(n) = the current t ransi ent magni t ude or current i ncrease over a 1-s interval, and
n = the number of intervals.
(3)
I f the current t ransi ent cont i nues over mor e than 1 s, de(n) is the summat i on of al l the current
changes in the transient. For exampl e, i f the current i ncreased by t he quant i t i es of 1, 2, and
2 ~A over t hree successi ve seconds and the pot ent i al condi t i on ( descr i bed bel ow) was sat-
isfied in al l cases, di(n) = 5 I~A. The dat a were anal yzed in sect i ons of 4- mi n durat i on, s o
the pi t t i ng function was cal cul at ed for 1024 seconds or intervals, (that is, n -< 1024). Si m-
ilarly, tr I and I~n~, whi ch are used to cal cul at e L, are al so cal cul at ed for each section.
As ment i oned above, a pot ent i al condi t i on exists for accept i ng a current t ransi ent i nt o t he
summat i on of the mean current t ransi ent magni t ude t o sat i sfy pi t t i ng behavior. In the case
of i dent i cal el ect rodes, the current t ransi ent must have an opposi t e si gn to that of t he pot ent i al
transient. I f the current t ransi ent i ncreases, the pot ent i al t ransi ent decreases. The absol ut e
val ue of the current dat a is t aken bef or e det er mi ni ng the sl opes of t he t ransi ent s because
pi t t i ng may i ni t i at e on ei t her el ect rode. The pot ent i al on the t hi rd el ect rode, t herefore, must
be negative. Onl y dat a that sat i sfy t he pot ent i al t ransi ent condi t i on are accept ed.
The sl ope condi t i ons i mposed for t he summat i on of the mean current t ransi ent s came f r om
a consi derat i on of the Tafel sl opes duri ng pi t t i ng or l ocal i zed anodi c di ssol ut i on. Fi gur e 1
i l l ust rat es this behavior. As pi t t i ng occurs, the anodi c Tafel l i ne shifts down al ong the ca-
t hodi c Tafel curve as a resul t of an i ncrease in anodi c di ssol ut i on. As shown in Fi g. 1, this
shift resul t s in the pot ent i al havi ng a negat i ve change (that is, a shi ft t o mor e el ect ronegat i ve
val ues) and t he current havi ng a posi t i ve change.
Microstructural Anal ysi s
Af t er t hey were r emoved from the sol ut i ons, t he coupons and el ect rodes were r i nsed wi t h
wat er and bl own dry. The sampl es were vi sual l y exami ned bot h bef or e and aft er cl eani ng
wi t h l i ght - opt i cal ( LOM) and scanni ng el ect ron mi cr oscopes (SEM). Surface chemi st ri es
were eval uat ed usi ng an energy di spersi ve X- r ay syst em (EDS) on t he SEM. Measurement s
of pi t si ze were al so made on t he flat coupons wi t h a confocal l aser mi croscope. Af t er an
i ni t i al exami nat i on of t he sampl es, t he cor r osi on product s were cl eaned f r om t he surface
usi ng a Cl ar ke' s sol ut i on per t he ASTM Pract i ce for Prepari ng, Cl eani ng, and Eval uat i ng
Cor r osi on Test Speci mens ( GI ) .
Re s u l t s
The pr i mar y findings of this st udy were that t he ENM dat a, t hrough t he anal ysi s of a
pi t t i ng function, i ndi cat ed that t he cor r osi on act i vi t y of the t wo vessel s was different. The
ENM corrosi on paramet ers of noi se resi st ance and l ocal i zat i on i ndex showed no si gni fi cant
di fference bet ween the t wo vessel s because of the general passi vi t y of the syst em. Mi cr os-
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
206 ELECTROCHEMI CAL NOI SE MEASUREMENT
r"
0
13_
- ve
r
(1)
* 6
I:L
E o.-
z~E
-Me
m/m+
Erev
r e ' ~ " ~ An o d i c
L ~ Active dissolution
9 Mi xed 7 1 1 t he p,t
Potential" ~
II Cathodic
" * Cu r r e n t De n s i t y , L o g i
At
+ve
An o d i c
~ Coupl ed Potential
ooo,c o 2 , ~
within the pit
AT Cu r r e n t , L o g I
+ve
FI G. 1--Schematic anodic and cathodic Tafel c ur v e s dur i ng pi t t i ng.
tructural anal ysi s o f the e l e c t r ode s f r om each v e s s e l s ho we d a di f f erence i n the degree o f
corros i on. The s t eel mi crost ruct ure wa s al s o f ound t o af f ect corros i on.
Noi se Resi stance
The no i s e resi st ance i s anal ogous t o the c o mmo n l y accept ed charge transfer res i s t ance
used f or es t i mat i ng general corros i on rates. In the St ern- Geary rel at i onshi p, the corros i on
current i s equal t o the St ern- Geary const ant di vi ded by the charge transfer resi st ance. Si m-
ilarly, the no i s e res i s t ance i s i nvers el y proporti onal t o the corros i on current. The corros i on
current i s convert ed t o a corros i on rate us i ng Faraday' s law. A noi s e res i s t ance o f 100 k f l
c m 2 corres ponds t o a general corros i on rate o f 0. 0035 mm ye ar - l i f a Stearn-Geary const ant
o f 30 mV i s us ed ( t ypi cal val ues range f r om 2 0 t o 5 0 mV) . Thi s corros i on rate i s i nsi gni f i cant ,
whe r e as a no i s e res i s t ance o f 100 f~ c m z corres ponds t o a cons i derabl e corros i on rate o f 3. 5
mm y e a r - k
The no i s e res i s t ances we r e rel at i vel y hi gh f or bot h v e s s e l s in t hi s st udy and di d not s ho w
any si gni f i cant di f f erence f r om the addi t i on o f nitrate. The no i s e res i s t ance val ues for each
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
MICKALONIS ET AL. ON ENM WITH NUCLEAR WASTE TANKS 2 0 7
vessel are pl ot t ed versus t he t i me in test in Fi g. 2. In the figure, changes in t he measuri ng
resi st or of t he zero resi st ance ammet er ( ZRA) are i ndi cat ed. These changes affect ed t he val ue
of the noi se resi st ance.
The average noi se resi st ance was of the or der of 100 kl~ cm 2, whi ch i ndi cat es that t here
was i nsi gni fi cant general corrosi on. Thi s resul t was expect ed because st eel gener al l y is pas-
sive in al kal i ne solutions. These val ues are in agr eement with pol ari zat i on resi st ance values
det er mi ned with DC t echni ques on a r od el ect r ode in si mi l ar sol ut i ons [7].
Localization Index
The l ocal i zat i on i ndex showed some response to the nitrate addi t i ons; however, the i ndex
values pr ovi ded l i mi t ed, i f any, i nformat i on because of the l ow rates of attack. The response
of the i ndex over the t est per i od is shown in Fi g. 3 for each vessel. The l ow corrosi on was
at t ri but ed to the passi ve behavi or of t he steel. The i ndex reached t he hi ghest val ues when
the noi se resi st ance i ncreased, i ndi cat i ng that at l ow rates of at t ack the noi se dat a showed a
hi gh degr ee of l ocal i zat i on even i f the l ocal i zed at t ack was mi ni mal . Therefore, in this
al kal i ne syst em, the l ocal i zat i on i ndex i ndi cat ed t hat passi ve film mai nt enance and l ocal i zed
corrosi on are si mi l ar processes, in that t hey are bot h random. Because current magni t udes
are not i ncl uded in t he i ndex, l ocal i zed cor r osi on and passi ve behavi or cannot be di fferen-
tiated. The i ndex, however, can di fferent i at e bet ween ei t her of t hese processes and uni f or m
corrosi on.
Pitting Function
The pi t t i ng funct i on combi nes the cor r osi on current measur ement wi t h the l ocal i zat i on
i ndex to quant i fy t he severi t y of pitting. The nitrate addi t i ons i ncr eased the corrosi on current
of t he el ect rode. Thi s i ncrease was at t ri but ed to pi t i ni t i at i on. The mi crost ruct ural anal ysi s,
as di scussed below, support ed this i nt erpret at i on of t he ENM data. For t he cont rol vessel,
I~ms was on the order of 10 -8 A. l~ms for t he ni t rat e-adj ust ed sol ut i on i ncreased by an order
of magni t ude to l 0 7 A aft er t he final nitrate addi t i on.
Fi gur e 4 shows t he changes in the pi t t i ng funct i on over the course of the test. The pi t t i ng
funct i on for t he cont rol vessel was flat t hroughout t he experi ment . The initial nitrate addi t i ons
of 8 and 26 mL had little effect. The l arger addi t i ons resul t ed in current transients, whi ch
t ended t owar d stable, hi gher values. The t ransi ent s were al so more i mmedi at e wi t h t he l arger
nitrate addi t i ons. The pi t t i ng funct i on di d not cont i nue to i ncrease aft er the i ni t i al response,
whi ch i ndi cat ed that t he el ect rodes were not forced i nt o pi t propagat i on. Pi t pr opagat i on may
have occurred i f t he exper i ment durat i on was longer. The pi t t i ng function, however, dem-
onst rat ed that the ENM dat a were di fferent for the t wo vessels.
Microstructural Analysis
Pi t t i ng of t he st eel in t hese si mul at ed wast e sol ut i ons was i ni t i at ed on t he exami ned cou-
pons and el ect rodes. The degree of pi t t i ng, however, was di fferent for the coupons and
el ect rodes. Thi s di fference was at t ri but ed t o t hei r mi crost ruct ures.
Electrodes--The macr oscopi c exami nat i on of t he el ect rodes from bot h vessel s reveal ed
superfi ci al r eddi sh brown cor r osi on product s l ocat ed bet ween t he surface gri ndi ng marks.
These product s were bel i eved to have resul t ed from ai r oxi dat i on aft er r emoval from the
vessel. The el ect rodes f r om t he ni t rat e-adj ust ed vessel al so had some areas of heavi er cor-
rosi on product s or great er degradat i on.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
208 ELECTROCHEMICAL NOISE MEASUREMENT
10000
= o o E 1 0 0 o
r r ' ~
wE
10
Control Solution
[ ]
i i ,
0 10 20 30
40 50 60 70 80 90
Ti me in Test (days)
100
Nitrate-Adjusted Solution
10000
r 1000
lO
0 10 20 30 40 50 60 70 80 90 100
Ti me in Test ( days)
FIG. 2 - - No i s e resi st ance val ues f o r car bon st eel in a s i mul at ed was t e sol ut i on.
At hi gher magni fi cat i on ( x500) in the SEM, the el ect rodes were found to have pi t t ed, as
shown in Fi g. 5 for an el ect r ode from the ni t rat e-adj ust ed test. The severi t y of pi t t i ng ap-
pear ed great er on the el ect rodes from the ni t rat e-adj ust ed vessel. The severi t y was not quan-
tified as a resul t of the l i mi t ed number and smal l si zes of t he pi t s. The pi t s were pr i mar i l y
hemi spher i cal and measur ed 10 ~Lm across. The pi t bot t oms were seen, whi ch i ndi cat ed that
the pi t s had not grown ext ensi vel y into the coupons. These resul t s were expect ed based on
the pi t t i ng funct i on anal ysi s of the ENM data. Al so, on one el ect rode, whi ch was cross-
sect i oned and vi ewed, pi t s coul d not be found, al t hough the cr oss- sect i on cut s may have
been in areas wi t hout pits. Most of the obser ved pi t s were associ at ed wi t h i ncl usi ons, al-
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
MICKALONIS ET AL. ON ENM WITH NUCLEAR WASTE TANKS 209
X
I1)
t . -
r
o
N
0
0
. . J
1
0.8
0.6
0.4
0.2
0
0 10 20
Cont r ol Sol ut i on
30
i i i i i i i
40 50 60 70 80
Ti me in Test ( days)
i
90 100
Ni t r at e- Adj ust ed Sol ut i on
1
x 0.8
~ 0.6
~ 0.4
~ 0.2
0 .
0 10 20 30 40 50 60 70 80 90
Ti me in Test ( days)
FIG. 3--Localization index values for carbon steel in a simulated waste solution.
100
t hough some i ni t i at ed near surface scratches. The i ncl usi ons, as anal yzed by EDS, were
composed pr i mar i l y of manganese and sulfur or of al umi num and silicon.
These mi crost ruct ural observat i ons, t herefore, support the i nt erpret at i on of the ENM data.
Pi t growt h on the coupons was si gni fi cant to that of t he el ect rodes where pi t s had initiated.
Pi t t i ng was al so mor e severe at hi gher ni t rat e concent rat i ons. ENM may al so be useful for
ot her mat eri al s, such as aust eni t i c st ai nl ess steels, whi ch al so have a pr opensi t y for pi t
i ni t i at i on at i ncl usi ons.
Coupons- - The coupons t hat were r emoved bef or e the 75- mL addi t i on ( 0. 298M nitrate
concent rat i on) were found to have numerous pi t s, as shown in Fi g. 6. Some l i ght staining
was al so not ed beneat h the washers that were used to fasten the coupons. The severi t y of
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
210 ELECTROCHEMICAL NOISE MEASUREMENT
1E-06
Co n t r o l S o l u t i o n
O
O
t "
- i
t l
e "
CX.
1E-07
1E-08
1E-09
1E-10
1E-11
[ ] [ ]
I I I I I I I I I I I I I I
0 10 20 30 40 50 60 70 80 90
Ti me in Test (days)
100
t -
o
. _
O
t - -
I t
t ~
._c
. ~
1E-06
1E-07
1E-08
1E-09
1E-10
Ni t r a t e - A d j u s t e d S o l u t i o n
0 10 20
i ' i i i
30 40
1E-11 ' ' ' ' ' '
50 60 70 80
Ti me i n Test (days)
FIG. 4--Pitting function values f or carbon steel in a simulated waste solution.
i i
90 100
pi t t i ng on the coupons was si mi l ar for t he t wo solutions. In this case, the mi crost ruct ure of
the coupon was a mor e si gni fi cant vari abl e for i ni t i at i ng pi t t i ng than the sol ut i on corrosi vi t y.
Recal l that the cont rol sol ut i on was at the l i mi t for i nhi bi t i on. As can be seen in Fi g. 6, the
pi t s were in l i near arrays. Thi s f or mat i on resul t ed from t he steel mi crost ruct ure. The pi t s
were associ at ed with i ncl usi ons like the electrodes. These inclusions were along the ferri t e/
pearl i t e boundari es. These boundar i es were st rai ght because of the mat eri al processi ng, whi ch
resul t ed in a l ami nar mi crost ruct ure of ferri t e (l i ght grains) and pearl i t e (dark grai ns) as
shown in Fi g. 7.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
MICKALONIS ET AL. ON ENM WITH NUCLEAR WASTE TANKS 211
FIG. 5- - SEM micrograph of pits on a carbon steel electrode (
The pit morphology for the coupons was similar to that for the electrodes--hemispherical
and approximately l0 p,m across, The severity of pitting was much greater on the coupons.
This difference in pitting susceptibility is attributed to the steel chemistry. More inclusions
were found on the surface of an untested coupon than of an electrode. Figure 8 shows an
SEM micrograph of a coupon surface. The small dark spots were identified as inclusions. A
probable explanation for this difference is that the steel used to make the electrodes had
lower impurities than that used for the coupons.
The difference in the severity of corrosion between the electrodes and the coupons high-
lights an important consideration for corrosion-monitoring techniques that use samples (that
FIG. 6--Photograph of pits on a carbon steel coupon.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
212 ELECTROCHEMICAL NOISE MEASUREMENT
FIG. 7--Microstructure of carbon steel coupons (Nital Etchant).
is, either electrodes or coupons). The microstructure of the sample must be representative of
the material of construction for the structure or equipment that is to be monitored. If not,
erroneous data may be obtained.
Conclusions
The results of this study show that:
1. Analysis of ENM data, through pitting function values, aided in distinguishing between
passivity and pitting.
FIG. 8--SEM micrograph of carbon steel coupon surface.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
MIGKALONIS ET AL. ON ENM WITH NUCLEAR WASTE TANKS 213
2. The pi t t i ng funct i on i dent i fi ed the initiation of smal l pi t s (that is, 10 g,m or less).
3. The mi crost ruct ure of the steel si gni fi cant l y affect ed the corrosi on t hrough the presence
of i ncl usi ons and must be consi der ed when usi ng cor r osi on- moni t or i ng techniques.
The el ect rochemi cal noi se t echni que may meet the requi rement s for corrosi on moni t or i ng
of nucl ear wast e tanks. Furt her anal ysi s and devel opment , however, are necessary to confi rm
pi t pr opagat i on and to opt i mi ze the pi t t i ng funct i on anal ysi s. The t echni que may al so be
useful for the det ect i on of stress corrosi on cracki ng wi t h appropri at e sel ect i on cri t eri a and
anal yt i cal limits.
References
[1] Congdon, J. W., "Inhibition of Nuclear Waste Solutions Containing Multiple Aggressive Anions,"
Materials Performance, Vol. 27, No. 5, May 1988, pp. 34-38.
[2] Donovan, J. A., "Materials Aspects of SRP Waste Storage--Corrosion and Mechanical Failure,"
USDOE Report DP-1476, Savannah River Site, Aiken, SC, Nov. 1977.
[3] Moran, G. C. and Labine, P., Eds., Corrosion Monitoring in Industrial Plants Using Nondestructive
Testing and Electrochemical Methods, ASTM STP 908, American Society for Testing and Materials,
West Conshohocken, PA, 1984.
[4] Rothwell, A. N. and Eden, D. A., "Electrochemical Noise Techniques for Determining Corrosion
Rates and Mechanisms," Paper 223, Corrosion92, NACE International, Houston, TX, 1992.
[5] Webster, S., Nathanson, L., Green, A. G., and Johnson, B. V., "The Use of Electrochemical Noise
to Assess Inhibitor Film Stability," in UKCorrosion 92, 1992.
[6] Hearer, E. E., Lawson, K., and Eden, D. A., "Stress Corrosion Cracking In Refinery Systems," in
UKCorrosion'93, Institute of Corrosion, Surrey, United Kingdom, 1993.
[7] Mickalonis, J. I., "Corrosion of Steel in Simulated Waste Solutions," Paper 117, Corrosion94, NACE
International, Houston, TX, 1994.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
He nni e J. De Br uy n, l Ke v i n Laws on, 2 a n d Ed wa r d E. He a v e r 2
On-Line Monitoring Using
Electrochemical Noise Measurement in
C0- C02- H20 Systems
REFERENCE: DeBruyn, H. J., Lawson, K., and Heaver, E. E., "On- Li ne Monitoring Using
Electrochemical Noise Measurement in CO-CO2-H20 Systems," Electrochemical Noise
Measurement for Corrosion Applications, ASTM STP 1277, Jeffery R. Kearns, John R. Scully,
Pierre R. Roberge, David L. Reichert, and John L. Dawson, Eds., American Society for Testing
and Materials, 1996, pp. 214-229.
ABSTRACT: The stress corrosion cracking (SCC) susceptibility of carbon steel extends over
a narrow range of potentials in environments that contain a mixture of wet carbon monoxide-
carbon dioxide (CO-CO2) gas. Normal fluctuations in plant operation may cause movement in
and out of this range of potentials. Stress corrosion cracking in this case is discontinuous, and
an opportunity exists to prevent it by process control.
Electrochemical noise measurement, because of its capability for operation in thin film con-
densate conditions, is an attractive technique for monitoring corrosion in wet CO-CO2 systems.
Laboratory tests and plant trials were carded out to determine if electrochemical noise mea-
surement could be used to characterize periods of SCC.
KEYWORDS: carbon steel, stress corrosion cracking, process control, carbon monoxide-
carbon dioxide, on-line monitoring, oil industry
St ress cor r osi on cr acki ng ( SCC) is a pr obl em t hat affects a r ange of engi neer i ng mat er i al s
used i n a wi de vari et y of i ndust r i al appl i cat i ons. The pr obl em is t r adi t i onal l y appr oached by
sel ect i ng nons us cept i bl e mat er i al s or a stress r el i ef heat t reat ment . Bot h met hods of pr even-
t i on can ent ai l cons i der abl e expense, and t he l at t er ma y not al ways be effect i ve, par t i cul ar l y
i f t he heat t r eat ment is done i n t he field or aft er a l ocal i zed repair. Wh e n t he suscept i bi l i t y
to SCC for a steel ext ends over a nar r ow r ange of pot ent i al s, t here is a possi bi l i t y t hat nor mal
pl ant f l uct uat i ons wi l l cause move me nt i n and out of t hi s r ange, and cr acki ng wi l l be di s-
cont i nuous . Thi s pr ovi des an oppor t uni t y to pr event SCC by pr ocess cont r ol i f per i ods of
SCC act i vi t y can be i dent i f i ed and effect i ve count er meas ur es defi ned. Such a syst em is f ound
downs t r eam of r ef or mi ng uni t s i n whi ch pr essur i zed mi xt ur es of wet car bon monoxi de and
car bon di oxi de pr omot e stress cor r osi on cr acki ng of car bon steel. Thi s paper descr i bes t he
appl i cat i on of el ect r ochemi cal noi s e me a s ur e me nt ( ENM) to moni t or stress cor r osi on crack-
i ng i n such an envi r onment .
' Mossgas (Pty) Ltd., Duin Zicht Ave., Private Bag X14, Mossel Bay 6500, Republic of South Africa.
2 Real Time Corrosion Management Ltd., Rutherford House, Manchester Science Park, Manchester,
MI5 6SZ UK.
Copyrighl 9 1996 by ASTM International
214
www.astm.org
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
DEBRUYN ET AL. ON ON-LINE MONITORING 215
Stress Corrosion Cracking in Wet Mixtures of Carbon Monoxide and
Carbon Dioxide
Transgranular SCC has been observed in as-welded carbon steel pipelines and equipment
exposed to wet mixtures of carbon monoxide (CO) and carbon dioxide (CO2) gas [1-4].
Cracking is most prevalent in the temperature range from 20 to 60~ and occurs in many
mixtures of CO/CO2 gas. Published test results [1-5] and service experience at Mossgas
and similar process plants are summarized in Fig. 1, which indicates a likelihood of cracking
in almost any wet mixture of CO and CO2. Potentiostatically controlled slow strain rate tests
carried out by Hannah [4] showed that in pressurized systems SCC is promoted by both
anodic and cathodic polarization, indicating that the system is capable of supporting both
SCC and hydrogen embfittlement. At the free corrosion potential, the mechanism is SCC
reportedly [3] as a result of film rupture. In CO/CO2 gas systems, an aqueous solution of
COz is the principal corroding agent, and surface adsorption of CO provides passivation
[1-3]. SCC occurs in a narrow band of potentials within which the steel exhibits active-
passive transition behavior. Dunlop and Olson [5] report cracking in the range from -475
to -575 mV (SCE). This is supported by results obtained by Harrison [3] who reports
cracking at a similar range of potentials with maximum crack growth at - 475 mV (SCE).
Berry and Payer [1] found SCC at potentials between - 510 and -625 mV (SCE).
The schematic in Fig. 2 reflects the corrosion behavior of carbon steel in aqueous solutions
of CO and CO2. The system behaves as a chemical potentiostat, depending upon CO2 con-
centration, CO adsorption, and the presence of oxygen [2,3]. There is a possibility that
normal fluctuations in the operation of an industrial plant will cause an excursion into and
out of the region of SCC susceptibility. Straightforward measurement of corrosion potential
would seem to offer a means of monitoring when an environment is capable of supporting
SCC in this system, however, the service media are principally gaseous, and cracking occurs
beneath a thin film of condensate. Meaningful measurement of potential is virtually impos-
FIG. 1--Summary of published test results and service experience showing range of SCC susceptibility
[1].
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
216 E L E C T R O C H E M I C A L NOI SE M E A S U R E M E N T
. ~ : ~ M a t e r i a l ::' :
: : 1 , n e e i~ :
B e h a v i o r
D e s c d p t i o n
E n v i r o n m e n t a l
C o n d i t i o n s
Act i ve
A p a s s i v e
film i s not
f o r me d
':::= a n d : ~: So~ce '
Tr ans )assive
A passi ve film i s f o r me d b u t i s ruptured by t he
appl i cat i on o f stress. Ma t e r i a l s e x p o s e d at ~}m
rupture s is sl ow t o repassivate a n d Iocalised metal
loss b y anodi c di ssol uti on resul ts in S C C
I ncreasi ng con' osi vi t y p r o mo t e d by:
Hi gher PCO 2
Hi gher PO 2
H i g h e r total acids
P a s s i v e
A pass! ve surf ace
i s f o r me d a n d full
passi vi t y i s
mai nt ai ned
Increasing passi vi t y p r o mo t e d b y :
Hi gher PCO
a n d
s o m e o r g a n i c c o m p o u n d s
FIG. 2--Schematic showing behavior of carbon steel in CO-CO 2 environments [ 1].
sible, and measurement of the corrosion potential in the bulk environment may not reflect
conditions at the crack tip. Electrochemical noise instrumentation, because of its capability
to moni t or under thin film conditions, is attractive for use in C0- CO2- H20 systems.
El ect rochemi cal Noi se Meas urement for SCC Appl i cat i ons
Electrochemical noise has been characterized from laboratory and plant observations. It
has been noted that the amplitude of low frequency potential or current electrochemical noise
is related to the rate of the corrosion process, and that the frequency range, as measured by
the shape of the power spectrum, is influenced by the nature of the process [6].
The ability of ENM to distinguish corrosion processes has led to an interest in its use to
detect SCC. Cottis and Lotto report that the noise technique can be used to moni t or SCC
processes in alpha-brass [7] and high-strength aluminum alloys [8]. In most cases, SCC gave
the highest EN amplitude and the highest standard deviation peaks. The source of the elec-
trochemical current noise (ECN) was believed to be repassivation transients resulting from
exposure of fresh metal following rupture of the passive film by cracking. Similar tests
carried out on high strength steel in conditions in which environmentally assisted cracking
was almost certainly because of hydrogen embrittlement resulted in characteristic noise only
when the specimen broke [9]. There is an opportunity to prevent SCC by process control i f
periods of SCC activity can be identified and effective countermeasures defined. Control
requires a signal that reflects the susceptibility of the material to SCC. Laborat ory tests and
plant trials were carded out to assess the use of electrochemical noise measurement in
CO-CO2-H~O systems.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
DEBRUYN ET AL. ON ON-LINE MONITORING 217
Experimental Procedure and Results
Laborat ory Met hod
St ai nl ess st eel aut ocl aves were used to si mul at e t wo servi ce envi ronment s by pressuri zi ng
an aqueous condensat e to 2 MPa wi t h t he gas mi xt ures shown in Tabl e 1. The syst em was
i nvest i gat ed usi ng a prest ressed C-ri ng as one el ect r ode (Worki ng 1) coupl ed to a second
unst ressed count er-el ect rode (Worki ng 2) fabri cat ed f r om i dent i cal mat eri al . The aut ocl ave
body was used as the reference el ect rode. The C-ri ngs t ypi cal l y had an out si de di amet er of
appr oxi mat el y 48 mm and a wal l t hi ckness of 3 mm. They were f abr i cat ed f r om l ow-st rengt h
carbon st eel Type SA 106 Gr ade B, in accor dance wi t h t he requi rement s of t he ASTM
Pract i ces for Maki ng and Usi ng C- Ri ng Stress Cor r osi on Cr acki ng Test Speci men (G 38).
Several stress levels were exami ned: 50, 75, and 100% of yi el d stress. The el ect r ol yt e was
degassed by purgi ng wi t h oxygen- f r ee ni t rogen. The el ect r ode arrangement was then pl aced
in t he aut ocl ave whi ch was cl osed and subsequent l y repurged wi t h ni t rogen. A gas envi ron-
ment was used to br i ng t he aut ocl ave up to pressure, and the gas mi xt ure was cont i nuousl y
bubbl ed t hrough the el ect r ol yt e for the durat i on of the test. A pr eci si on l ow-noi se dat a
acqui si t i on syst em was used to col l ect and anal yze t he el ect r ochemi cal noi se data. The
i nst rument vol t age range was 0 to l 0 V, and t he current range was l 0 pA to 10 mA at a
resol ut i on of 2 mV and l 0 pA. The maxi mum r ecommended dat a acqui si t i on rat e was one
sampl e per second. A schemat i c of t he i nst rument at i on is shown in Fi g. 3.
At t he end of t he t est peri od, the C-ri ngs were opt i cal l y exami ned usi ng a st ereomi cro-
scope. Ar eas of i nt erest were marked, and aft er sect i oni ng, t he severi t y of SCC was assessed
usi ng met al l ography. The EN t races from each speci men were exami ned to det er mi ne i f t here
were charact eri st i cs that coul d be mat ched to the obser ved vi sual SCC.
Laboratory Results
Met al l ography
Met al l ogr aphy was used to est abl i sh the presence of SCC in the C-ri ngs. Speci mens ex-
posed to Gas Mi xt ur e 1 showed the hi ghest i nci dence of SCC. The maxi mum dept h of
penet rat i on and hi ghest frequency of cracks was obser ved in the speci mens st ressed to 100%
of yi el d stress. Cracks coul d be di st i ngui shed at about 60 ~ f r om t he center, whi ch corre-
sponded to appr oxi mat el y 45% of yi el d, as det er mi ned f r om t he t heoret i cal stress di st ri but i on.
The maxi mum crack dept h measured at t he most hi ghl y st ressed regi on at t he cent er of t he
speci mens was 0. 4 mm. The maxi mum crack dept h measur ed in t he C-ri ngs st ressed to 75
and 50% of yi el d was 0.17 and 0.05 ram, respectively. As before, no cracks were obser ved
at areas of t he speci mens in whi ch the t heoret i cal stress l evel fel l bel ow 45% of yi el d stress.
In Gas Mi xt ur e 2 onl y ver y shal l ow cracks, l ess than 0.015 mm, were observed. It was not
possi bl e to correl at e the i nt ensi t y of SCC to the magni t ude of the appl i ed stress.
TABLE l --Anal ysi s of the gas mixtures used during laboratory tests.
Carbon Monoxide, % Carbon Dioxide, % Hydrogen, % Methane, %
Gas Mixture 1 5 15 40 40
Gas Mixture 2 15 15 65 5
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
218 ELECTROCHEMICAL NOISE MEASUREMENT
FIG. 3--Schematic of electrochemical noise measurement instrumentation.
El ect rochemi cal Noise Measurement
Figure 4 shows a typical set of electrochemical current and potential noise signals collected
from the samples of carbon steel stressed to 50, 75, and 100% of yield and exposed to Gas
Mixture 1 (5% CO). At high stress levels, positive-going potential transients and unidirec-
tional negative-going current transients were identified and appeared to be characteristic of
periods of stress corrosion cracking. The time traces showed an increase in transient ampli-
tude and frequency at increasing levels of stress. The electrode stressed to 100% of yield
showed potential transients approaching 0.5 mV, with several current transients in excess of
6 ~A. For the 50 and 75% specimens, the amplitude and frequency of the potential and
current transients was significantly less (Fig. 4). Figure 5 shows an example of SCC observed
during metallographic examination of a C-ring stressed to yield.
Samples monitored in Gas Mixture 2 containing a higher CO concentration exhibited less
transient activity, and the mean coupling current was substantially lower. This was indicative
of lower corrosion as a result of better inhibition by the CO. Positive-going potential tran-
sients and corresponding unidirectional negative-going current transients were present in the
electrochemical noise measurement traces of specimens stressed at the higher levels (Fig.
6).
Plant Method
Purpose-designed SCC sensors were fabricated for use with 50.8-mm (2-in.) high-pressure
access fittings (Fig. 7). The sensors comprised three elements which included stressed and
unstressed elements fabricated from carbon steel similar to that used for construction of the
pipeline being monitored. In this case, carbon steel Type SA 106 Grade B was used. The
reference electrode was fabricated from Type 316 stainless steel. The stressed electrode was
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
DEBRUYN ET AL. ON ON- LI NE MONI TORI NG 219
+8.47
A
+8.29
-4.30
g
-4.60
Time (seconds) 2048
50% yield-stress

+11.33
-5.00
g
-5.50
0 Time (seconds) 2048
75% yield-stress
+9.50
g

-6.40
g
-7.30
r .
0 lqme (seconds) 2048
100% yield-stress
FIG. 4- - Typi cal pot ent i al and current time traces obtained during laboratory tests using Gas Mixture
1 and working electrodes stressed to 50, 75, and 100% of yi el d stress.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
220 ELECTROCHEMICAL NOISE MEASUREMENT
I +18.882 [ A~
[ +18.615 L
0 Time (seconds) 2o48
50% yi el d-st ress
-378.45
g
-380.71
" i
0 "[]me (seconds) 2048
75% yi el d-st ress
I
- 3 " A 2 ' I
-369.70
-746.90
Time (seconds) 2o48
100% yi el d-st ress
FIG. 5- - SCC cracks found after microsectioning a specimen stressed to yield.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
DEBRUYN ET AL. ON ON-LINE MONITORING 221
FIG. 6--Typi cal potential and current time traces obtained during laboratory tests using Gas Mixture
2 and working electrodes stressed to 50, 75, and 100% of yieM stress.
stressed to yield point, confirmed by strain gage measurements. The working electrodes were
electrically coupled so that the mean coupling current and electrochemical current noise
could be monitored. For the plant trials, the sensors were wired such that Working 2 was
the stressed electrode. This reversed the sign of the current transients. Electrochemical po-
tential noise between the couple and the reference electrode was measured simultaneously
against a stainless steel reference electrode included in the probe (Fig. 3). Three sensors
were placed at key locations in the plant, as shown by the simplified process diagram shown
in Fig. 8. Table 2 summarizes the operating conditions at each location. A high resolution
electrochemical noise system was used to interrogate sensors placed in the plant and log
data for a representative exposure period.
Plant Results
The salient features of electrochemical noise time traces obtained from Monitoring Probes
1 and 2 were similar. The data files exhibited positive-going potential transients and positive-
going current transients that were associated with active SCC conditions. The average mag-
nitude of current transients was 48 nA, but occasionally approached 200 nA, and the average
size of the potential transients was 1.0 mV. Periods of pitting were associated with a change
in the direction of the transients to the negative direction for both current and potential. An
absence of significant transients indicated periods when SCC or pitting was not occurring.
Typical examples of data traces are presented in Figs. 9 through 11. Unstable pitting is
characterized by sharp negative-going current transients as a result of film breakdown. The
signal returns to "baseline" once the passive film has reformed.
The corresponding potential transients showed an exponential-type recovery over a short
time period. Probe 3 was found to display consistently crevice-type behavior (Fig. 12) typ-
ified by a series of repetitive transients indicative of restricted diffusion. The time of recovery
to baseline levels was longer than for pitting corrosion.
Statistical analysis techniques normally used in conjunction with electrochemical noise
were not found to be particularly useful for identifying periods of high and low SCC activity.
Neither degree of localization, which is the ratio of the standard deviation (current) to rms
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
222 ELECTROCHEMICAL NOISE MEASUREMENT
(current), nor MEM spectral analysis, was able to differentiate between SCC, pitting corro-
sion, and crevice corrosion.
The observation that positive-going transients are recorded during periods of SCC gave
an opportunity to use a simple algorithm to identify periods when the collected data signify
that SCC is occurring. The algorithm assesses SCC activity in terms of a function (F), which
is a measure of the frequency and magnitude of positive-going transients:
f(frequency, SCC/Intensity) = l/n EfT_ 7 Ai "Tt~ ~E . . . .
3t ' r
where
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
DEBRUYN ET AL. ON ON-LINE MONITORING 223
FIG. 8--Simplified process diagrams showing location of sensors.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
224 ELECTROCHEMICAL NOISE MEASUREMENT
TABLE 2---Summary of operating conditions at the probe locations.
Nominal Gas Composition
Temperature, Press, CO, CO2, CH4, H2, N2, PCO, PCO2,
~ kPa tool% mol% mol% mol% tool% kPa kPa
Probe ] 40 2660 1.7 12 36 36 Bal 45 319
Probe 2 40 1810 1.7 12 36 36 Bal 30 217
Probe 3 18 3600 1.7 12 36 36 Bal 61 432
n = number of t ransi ent s measured in the sampl er per i od (256 s),
Ai = peak hei ght of the t ransi ent s, ~A,
Gr i = st andard devi at i on of the current, and
i . , s = r oot mean square of the current.
The peak hei ght and number of t ransi ent s are onl y count ed when Oi/Ot and OE/Ot are bot h
positive.
Fi gur e 13 shows t ypi cal traces of SCC i nt ensi t y (F) in real t i me for Probes 1 and 2. Al so
i ncl uded in t he figure is the CO cont ent of t he process gas measured at the t i me of el ect ro-
chemi cal noi se moni t ori ng. Act i vi t y was hi ghest on Probe 2, and per i ods of i ncreased SCC
act i vi t y ( F) coul d be coor di nat ed wi t h i ncreases in the CO cont ent in the gas.
Di scussi on
Characterization of ECN During Periods of SCC
Met al l ogr aphy of C-ri ngs aft er exposure to the l aborat ory test envi ronment confi rmed the
presence of stress corrosi on cracks. The severity of SCC was est i mat ed from t he number
and dept h of cracks found in the speci mens. Those exposed to Gas Mi xt ur e 1 cont ai ned
more cracks t han speci mens exposed to Gas Mi xt ur e 2. Cr acks were al so mor e frequent and
deeper in the speci mens stressed to hi gher levels.
Scrut i ny of t he cor r espondi ng EN traces from the speci mens reveal ed an i ncrease in tran-
sient behavi or at hi gher levels of stress. At stress levels near to the yi el d stress of t he mat eri al ,
posi t i ve- goi ng pot ent i al and uni di rect i onal negat i ve- goi ng current t ransi ent s were i dent i fi ed
duri ng l aborat ory moni t ori ng. It was concl uded that t ransi ent behavi or of this t ype was
sympt omat i c of SCC.
The si gn of the current t ransi ent s was dependent upon whet her Wor ki ng 1 or Worki ng 2
made the st ressed el ect rode, but, in all cases, el ect rons flowed to the st ressed el ement .
Mechanism
The si gn of the current t ransi ent s recorded duri ng per i ods of SCC suggest ed that SCC
events were associ at ed wi t h an el ect r on- consumi ng react i on: cat hodi c di scharge on the
st ressed el ect rode. A cat hodi c- cont r ol l ed process is consi st ent wi t h a model for hydrogen
embr i t t l ement and is in agr eement with fract ographi c observat i ons in t hese syst ems (Fi g.
14). The ori gi n of the embr i t t l i ng hydr ogen is as fol l ows:
C02 + H20 =- FI2CO 3
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
DEBRUYN ET AL. ON ON-LI NE MONI TORI NG 225
a . , . , , , . , . .
-2.834
+I.320
8
- ~ A 4 0
e Ti me (seconds) 2048
Fig. 9: Probe 1. Pa s s i ve - no SCC
-17.91 -
g -
- 1 9 . 3 3 - -
- 1 3 0 . 0
-155.9
Fig. 10: Probe 1. SCC
, , I . . . . L, , . ,
0 Ti me (seconds) 102,
+16.15
"36.20
"1(/0.2
8
- 5 1 1 1 . 2
i i i
0 Ti me (seconds) 1024
Fig. 11: Probe 1.
Fig. 12: Probe 3.
Pitting Corrosion
+~a3s
A
: >
o
o.
+0.910
-7.990
=
- 2 2 . 4 2
.ti me (seconds) 1024
Crevicing
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
226 ELECTROCHEMI CAL NOI SE MEASUREMENT
o
(J
g
0.00001
0.000001
1E-07
1E-08
1 E - 0 9
1E-10
1E-11
1E-12
01/10/92
O.0QO~l
0.000001
09/10/92
i
17110/92 25/10/92
Pr obe I
02/11/92
1F-07
1E-08
IE-09
I E-I 0
1E-11
IE-12 I
01 I10/92 09110/92
9
8
7
6
5
4
3
2
1
[
17110/92
Pr obe 2
25/10/92
0 n .
01/10/92 17/10/92 09110/92
Co mmo n t o
Pr obe s l & 2
OP-J11192
FI G. 13--Stress corrosion cracking activity and CO content monitored in the plant.
I
!
I
1
25/10/92 ~1t 92
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
DEBRUYN ET AL. ON ON-LINE MONITORING 227
FIG. 14--Fracture appearance of a carbon steel specimen stressed in a C0-C02-H20 environment.
H2CO 3 + e- ---, H'.a ~ + HCO 3
H2CO 3 ~ H + + HCO 3-
HCO3- =- H + + CO3 2-
Once a crack has been initiated by hydrogen embrittlement, the freshly exposed iron is
able to catalyze water reduction to hydroxyl ions and hydrogen atoms. The electrons con-
sumed in this process appear to be supplied by a discharge of the electrochemical double
layer at metal exposed to the bulk of the solution. This results in a spontaneous polarization
of the bulk metal surface to a more noble value measured as transients in the electrochemical
potential noise signal.
Electrochemical Noise Measurement to Identify Periods of SCC
During the period of on-line monitoring, intermittent episodes of stress corrosion cracking
were identifiable as were periods of pitting and crevice corrosion. To understand the reasons
for the variance, consider briefly the CO-C02-H20 system. Carbon monoxide and carbon
dioxide act as a chemical potentiostat with carbon monoxide the passivating agent and carbon
dioxide the principal corrodant. Additionally, SCC does not take place if water is absent.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
228 ELECTROCHEMICAL NOISE MEASUREMENT
Consequent l y, any process change that i nt erferes wi t h the absor bed CO passi vat i ng l ayer
influences the nature of the corrosi on observed.
SCC was obser ved to coi nci de wi t h i ncrement s in CO concent rat i on (Fig. 13), however,
little di fference was obser ved as a resul t of rai si ng the part i al pressure of CO f r om 30 t o 45
kPa. Lean gas mi xt ures l ow in carbon monoxi de favor pi t t i ng and crevi ci ng in exi st i ng
cracks, as evi denced by fluctuations bet ween per i ods of pi t t i ng and SCC on bot h probes
(Figs. 10 and 11).
Ot her process vari abl es that are capabl e of i nfl uenci ng the corrosi on behavi or are: carry-
over of wat er - sol ubl e aci d compounds (acet i c and pr opi oni c aci ds) that favors pi t t i ng and
hi gh t urbul ence and t emperat ure ( > 150~ t hat dest roy the CO passi vat i ng l ayer and pr omot e
corrosi on as a resul t of wet carbon di oxi de. The resul t i ng mesa corrosi on t akes its name
from the geogr aphi c feat ure it resembl es, consi st i ng of st eep-si ded pl at eaus. Operat i ng the
pl ant above the dew poi nt prevent s at t ack altogether.
Cr ack i ni t i at i on and pr opagat i on in t he CO-CO2-H20 syst em is si gni fi cant l y i nfl uenced by
process fluctuations t hat make the syst em amenabl e to i nvest i gat i on by on- l i ne moni t ori ng
t echni ques. El ect rochemi cal noi se has pr oved capabl e of di fferent i at i ng bet ween peri ods of
active SCC and others of no SCC act i vi t y in CO-CO2-H20 envi ronment s.
Conclusions
El ect rochemi cal current and pot ent i al noi se moni t ori ng of st ressed el ect rodes exposed to
a CO-CO2-H20 envi ronment gi ve charact eri st i cs noi se traces that can be det ect ed bot h in
the l aborat ory and on-l i ne.
The si gn of the current and pot ent i al t ransi ent s suggest ed that cracki ng in this syst em was
due to hydr ogen embr i t t l ement rather than anodi c di ssol ut i on.
The pl ant moni t ori ng was abl e to i dent i fy peri ods when the servi ce condi t i ons favored
SCC, pi t t i ng, and cr evi ce corrosi on.
Lean gas mi xt ures low in carbon monoxi de favor pitting. Hi gher concent rat i ons of CO
pr omot e SCC.
There is an opport uni t y to use el ect rochemi cal noi se dat a gat hered on-l i ne to opt i mi ze the
operat i on of t he pl ant to mi ni mi ze the ri sk of SCC.
In pract i ce, it was found that process changes to operat e the pl ant out si de t he i dent i fi ed
SCC scenari o were l i mi t ed. The i nvest i gat i on, however, has proven to Mossgas t hat peri ods
when the envi ronment wi l l support SCC can be identified. Thi s assi st ed the pl anni ng of
r egul ar i nspect i on wor k for t he earl y det ect i on of cracki ng in cri t i cal lines.
References
[1] Berry, W. E. and Payer, J. H., Proceedings of the 6th Symposium on Line Pipe Research, American
Gas Association, Texas, 1979, pp. 251-268.
[2] Kowaka, M. and Nagata, S., Corrosion, Vol. 32, No. 10, 1978, pp. 396-401.
[3] Brown, A., Harrison, J. T., and Wilkins, R., paper presented at the Conference on Stress Corrosion
Cracking and Hydrogen Embrittlement of Iron Base Alloys, 12-16 June 1973.
[4] Hannah, I. M., Ph.D. thesis, UMIST, Manchester, 1991.
[5] Dunlop, A. K. and Olson, E. E., "Inhibition of Stress Corrosion Cracking of Steel in Aqueous
CO/CO 2 Environments," in Proceedings of the Sixth International Congress on Metallic Corrosion,
1975, pp. 332-342.
[6] Zanki, A. L., Gill, J. S., and Dawson, J. L., "Electrochemical Noise Measurements on Mild Steel
in 0.5 M Sulfuric Acid," in Electrochemical Methods in Corrosion Research, M. Duprat, Ed.,
Materials Science Forum, Vol. 8, 1986, pp. 463-476.
[7] Lotto, C. A. and Cottis, R. A., "Electrochemical Noise Generation During Stress Corrosion Crack-
ing of Alpha-Brass," Corrosion, Vol. 43, No. 8, Aug. 1987, pp. 499-505.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
DEBRUYN ET AL. ON ON-LI NE MONI TORI NG 229
[8] Lotto, C. A. and Cottis, R. A., "Electrochemical Noise Generation During Stress Corrosion Crack-
ing of High Strength Aluminum AA 7075-T6 Alloys," Corrosion, Vol. 45, No. 2, Feb. 1989,
p. 136.
[9] Lotto, C. A. and Cottis, R. A., "Electrochemical Noise Generation During SCC of a High Strength
Carbon Steel," Corrosion, Jan. 1990, pp. 12-19.
[10] Schmitt, G. and Rothman, B., "Corrosion of Unalloyed and Low Alloyed Steels in Carbonic Acid
Solutions," Werkstoffe and Korrosion, Vol. 29, 1978, p. 816.
[11] Schmitt, G., "Fundamental Aspects of CO2 Corrosion," Paper 53, Corrosion ' 83, NACE, 1983.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
Mi c hae l A. Winters, l Pat ri ck S. N. St okes, 2 and He nr y F. Ni chol s ~
Simultaneous Corrosion and Fouling
Monitoring Under Heat Transfer in Cooling
Water Systems
REFERENCE: Winters, M. A., Stokes, E S. N., and Nichols, H. E, "Simultaneous Corro-
sion and Fouling Monitoring Under Heat Transfer in Cooling Water Systems," Electro-
chemical Noise Measurement f or Corrosion Applications, ASTM STP 1277, Jeffery R. Keams,
John R. Scully, Pierre R. Roberge, David L. Reichert, and John L. Dawson, Eds., American
Society for Testing and Materials, 1996, pp. 230-246.
ABSTRACT: Corrosion and fouling in cooling water systems can potentially reduce heat-
transfer capability, increase maintenance costs, reduce plant availability, and contaminate pro
cess lines. Conventional monitoring systems have measured corrosion and fouling separately.
Fouling monitors alone are not able to indicate real-time corrosion activity beneath surface
deposits. Traditionally, corrosion information has been derived from destructive evaluation of
heat exchanger tubes, weight loss coupons, and probes under nonheat flux conditions. To
overcome these shortcomings, a monitoring system has been developed that measures corrosion
and fouling simultaneously. By using electrochemical noise measurement technology, the sys-
tem is particularly sensitive to detecting localized corrosion beneath a fouled surface, a major
and frequent mode of corrosion failure in heat exchangers. Development of the system has
progressed from field trials with a prototype unit to a commercially available system. This
paper reports and discusses the most recent evaluation of the system at Amoco's Corporate
Research facility in Naperville, Illinois, focusing on the sensitivity of electrochemical noise
measurement in detecting localized corrosion (pitting and so forth) on a heat transfer surface
in a cooling water environment.
KEYWORDS: fouling, under-deposit corrosion, electrochemical noise measurement, heat
exchangers
Inadequate control of cooling water treatment programs results in high operating and
maintenance costs in many situations. Pitting corrosion, scaling, and microbial fouling cause
damage to the installation and result in efficiency losses. Heat transfer surfaces are especially
prone to fouling and associated corrosion damage. A major drawback of field monitors has
been the inability to indicate real-time corrosion activity beneath fouled deposits. In the past,
it has been necessary to destructively evaluate heat exchanger tubes or rely on the results of
coupons and probes exposed under nonrepresentative conditions, for example, without heat
flux conditions, or both. Higher operating temperatures promote scaling and microbial growth
(up to =50~ which subsequently impedes access of a corrosion inhibitor to the heat
transfer surfaces. This can result in high rates of localized corrosion (pitting attack), which
is not detected by conventional control instrumentation. A combined corrosion and fouling
monitoring instrument (CFM500 Unit) has been recently developed that incorporates state-
Water technology specialist, Amoco Corp., WEC, 3700 Bay Area Blvd., Houston, TX 77058.
2 Business development manager, CML, Manchester M15 6SZ, UK.
3 Corrosion specialist, Amoco Corp., P.O. Box 3011, Naperville, IL 60566.
Copyrighl 9 1996 by ASTM lntcrnational
2 3 0
Www.astIII.OFg
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
WINTERS ET AL. ON SIMULTANEOUS CORROSION 231
of-the-art electrochemical techniques for corrosion assessment under in-plant service con-
ditions. This paper reports and discusses recent results from a "shakedown" evaluation of
the system at the Amoco Corporate Research Center before commercial use at an Amoco
chemical plant. This paper summarizes the continuing evolution of the monitoring technology
being used to evaluate corrosion behavior on a heat transfer surface.
Instrumentation
The system is comprised of a miniature heat exchanger (incorporating an integrated cor-
rosion sensor arrangement), an electronic flow and heat control unit, and a data collection
device that is linked to an electronic storage device (Fig. 1). The heat exchanger uses a
segmented tube (typically carbon steel). Tube segments are externally coated with heat trans-
fer paste and mechanically clamped in an electrically heated block. A side stream of cooling
water is fed through the exchanger and a paddlewheel flow meter enables the flow to be
FIG. 1--Schematic representation of the corrosion and fouling monitoring system (CFM500 unit).
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
232 ELECTROCHEMICAL NOISE MEASUREMENT
mai nt ai ned bet ween 0.15 and 4.6 ms -~, whi ch equat es to Reynol ds numbers bet ween
100 000 and 100 000. The flow is kept wi t hi n +_0.02 ms -I of t he set value. The ot her
vari abl e that may be set is ei t her heat t ransfer rate or wal l t emperat ure. Pl at i num resi st ance
t emper at ur e det ect ors (RTD) measure the t emperat ure of t he fluid and the t emperat ure of the
bl ock out si de the tube. The i nsi de wal l t emperat ure is cal cul at ed from t hese measurement s.
I f t he heat t ransfer rat e is set, then as t he surface of t he t ube fouls, t he wal l t emper at ur e wi l l
ri se to mai nt ai n the set heat rate. I f the wal l t emperat ure is set, the heat i ng rat e woul d
decrease to mai nt ai n the set wal l t emper at ur e as the resi st ance to heat t ransfer i ncreases.
Maki ng al l owances for convect i ve and conduct i ve resi st ance, changes in heat t ransfer resi s-
t ance (HTR) from that of the cl ean surface condi t i on are found anal yt i cal l y. Measur ed val ues
of vari abl es, scanned every minute, are fed t o a personal comput er for storage. Heat t ransfer
rat e can be set bet ween 50 and 1000 W. Thi s equates to a t ypi cal ski n t emperat ure bet ween
30 and 90~ The heat t ransfer rate is set t ypi cal l y at 200 W.
In conj unct i on wi t h the foul i ng tendency, t he corrosi on i nformat i on is obt ai ned (by means
of an i nt egrat ed sensor array) usi ng four el ect r ochemi cal t echni ques that make measurement s
si mul t aneousl y under heat flux: zero resi st ance ammet r y ( ZRA) , el ect r ochemi cal current
noi se (ECN), el ect r ochemi cal pot ent i al noi se (EPN), and l i near pol ar i zat i on resi st ance (LPR).
Zer o resi st ance ammet r y has been used convent i onal l y to det er mi ne the gal vani c current
bet ween t wo di ssi mi l ar el ect rodes, but may be al so used to det er mi ne the current bet ween
t wo nomi nal l y i dent i cal el ect rodes. It has been found t hat "i dent i cal " el ect rodes nor mal l y
have di fferent bal ances of anodi c and cat hodi c areas and hence t ake up sl i ght l y di fferent
pot ent i al s. When the el ect r odes are coupl ed vi a the ZRA, a measur abl e current wi l l flow.
The ZRA out put is t he root mean square ( RMS) of t he ZRA current. The DC val ue of the
current dur i ng act i ve cor r osi on is proport i onal t o t he corrosi on act i vi t y in progress on the
el ect rodes [1].
The EPN t echni que measures l ow-l evel r andom fluctuations of the corrosi on pot ent i al
bet ween t wo nomi nal l y i dent i cal el ect rodes. The fluctuations are usual l y of l ow ampl i t ude,
l ess than a mi l l i vol t , and of l ow frequency, in the range of 1 Hz and bel ow [2]. The EPN
measur ement out put is a l ow frequency bandpass filtered RMS value. The EPN measur ement
provi des dat a that can be correl at ed wi t h t he mode of corrosi on attack, because the r andom
fluctuations recorded are rel at ed to surface corrosi on event s/ act i vi t i es. The si gnat ures of
pi t t i ng corrosi on and crevi ce at t ack are especi al l y clear. For a fi xed el ect r ode area, it has
been shown t hat dat a obt ai ned at t he free cor r osi on pot ent i al can be rel at ed to t he degr ee of
l ocal i zed at t ack occurri ng at t he t i me of measur ement [3]. The el ect rochemi cal noi se t ech-
niques are sensi t i ve to the onset of uni form at t ack (general corrosi on) and can charact eri ze
pr opagat i on of pi t t i ng or cr evi ce corrosi on. The ECN measur ement t echni que is si mi l ar to
EPN measurement , except t hat fl uct uat i ons in t he coupl i ng current bet ween t he t wo si mi l ar
el ect rodes, t ypi cal l y less than 1 CA, are r ecor ded and anal yzed. The out put is a l ow-frequency
bandpass filtered RMS val ue (DC and hi gh-frequency AC component s removed). It corre-
sponds to the l ow-l evel frequency noi se (the di fferent i al of the ZRA signal), whi ch is part
of the ZRA signal, but has a much l ower ampl i t ude when general corrosi on is occurri ng.
ZRA t i mes records compl ement the i nformat i on obt ai ned by EPN and ECN measurement s.
Passi ve fi l m breakdown, for exampl e, is char act er i zed by sporadi c t ransi ent s in t he current
flow [3]. These current s are drawn from any avai l abl e cat hodi c site, i ncl udi ng t hose of the
second el ect rode, and the current t ransi ent s are obser ved vi a the ZRA.
It is possi bl e to rel at e el ect r ochemi cal current and pot ent i al noi se to l ocal i zed at t ack pr o-
pensi t y [4]. The l ocal at t ack fact or is det er mi ned from the rat i o of t he ECN measur ement
val ue and t he ZRA out put value. Thi s rat i o gi ves an i ndi cat i on of the st ochast i c nat ure of
t he processes occurri ng on t he el ect r ode surface. When this val ue is low, that is, 0.001, the
at t ack is l i kel y to be general in nature because this rat i o i ndi cat es a rel at i vel y gaussi an
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
WINTERS ET AL. ON SIMULTANEOUS CORROSION 233
distribution of the electrochemical events. Pit initiation events, whi ch are identifiable as low-
level transients in the EPN (<1 mV) and ECN (<0. 1 I~A) traces, indicate when the system
is becomi ng unstable. At this point, the local attack factor will begin to increase (as the ECN
measurement increases against a relatively stable ZRA value) depending on the amplitude
and frequency of the events. The stochastic nature of the installation process produces a
more Poisson-type behavior in which the limit is indicated by the levels of the ECN and
ZRA being very similar, that is, the local attack factor approaches a value of 1. This reflects
film instability, which invariably relates to chemistry or hydrodynami c changes or both in
the corrosion process. The value of the local attack factor will increase to perhaps 0.01 for
low-level activity. When pits begin to propagate (active pitting), several features will become
noticeable. The electrochemical potential noise (EPN) level will increase dramatically t o
values >1 mV. The electrochemical current noise (ECN) activity will increase t o >0. 1 I~A.
The ZRA will begin to show signs of transients, which correspond with the ECN activity.
The local attack factor will rise to values >0. 1 and may reach a maxi mum value of 1. Pit
death is associated with general cessation of local activity as indicated by a decrease in the
frequency of events and signal amplitudes. In the extreme case in which very active pit
propagation occurs, the overall amplitude of the signals will stay high, and the local attack
factor may drop slightly.
ECN and EPN are also used to obtain the resistance noise (R, ) value [2]. This value is
used to estimate corrosion rates from the Stern-Geary approximation [5,6]. Unlike the linear
polarization resistance (LPR) technique, the electrochemical noise techniques are nonpertur-
bative: No extemat voltage or current is applied. The potential and current measurements
that were made are due to naturally occurring, spontaneous fluctuations associated with the
corrosion processes taking place on the surface of the material.
Linear polarization resistance is a DC met hod used for measurement of uniform corrosion
[ASTM Practice for Conduct i ng Pot ent i odynami c Polarization Resistance Measurements
(G 59)]. The technique assesses the electrode potential versus current relationship close to
the corrosion potential. This is achieved by monitoring the external current flow between
two nominally identical electrodes held at a known potential difference. The difference is
kept small, typically 10 mV, and the change of potential relative to the change of current
gives the polarization resistance R p . The corrosion current is then given by the simple rela-
tionship (again from Stern-Geary):
ioo~r = B / R e
where B is a polarization constant dependent on the type of metal or alloy used for the
electrode probe and on the environment. The LPR met hod is used extensively for prediction
of general or uniform corrosion rates but has limited value in the presence of scales or
deposits, which can introduce an increased resistance as can low-conductivity electrolytes.
The technique is also insensitive to localized corrosion.
Applications
The CFM500 unit gives real-time corrosion information on fouled heat transfer surfaces.
The unit gives a continuous online indication of the corrosion rate, local attack factor (pitting
tendency), and the heat transfer resistance (HTR fouling tendency) on a single graphical
output. The data are also stored for more detailed retrospective evaluation. In the past, the
heat flux and flow parameters of fouling monitors [ 7 - 9 ] have been set to encourage contin-
uous fouling, that is, high heat flux and low flow. This mode of operation is known to be
sensitive t o small changes in water treatment chemistry, pH, and so forth, in terms of im-
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
234 ELECTROCHEMICAL NOISE MEASUREMENT
mediate changes in the fouling tendency that may occur. Conditions of operation can also
be set t o mimic a critical heat exchanger, thus tracking fouling at a key location in the
cooling system. By using the CFM500, this will allow the corrosion reading to be used as
an alarm for both corrosion and fouling changes because the corrosion readings are sensitive
enough to be a good indication of minute changes in fouling behavior. I f operated under
nonrepresentative (heavy) fouling conditions, the CFM500 system is likely to indicate unduly
high rates of corrosion that are not representative of the heat exchanger unit. However, the
"posi t i ve fouling mode" operation may be acceptable i f the water treatment inhibits under-
deposit corrosion. Otherwise, the information can be misleading.
In the past, there has been a tendency to consider that foul ed/ scal ed heat transfer surfaces
are protected by the presence of the scale. This assumption has been shown to be misplaced
by the recent experiences/results from the field studies. Fouling has an immediate and sig-
nificant impact on the surface corrosion condition. Operation of the unit such that it mimics
critical heat exchanger conditions is therefore imperative to obtain a representative indication
of corrosion activity in that location.
Res ul t s
Introduction
Laboratory and field trials on the prot ot ype CFM500 unit have been carried out previously
and are reported elsewhere [10,11]. This paper reports on the most recent evaluation on the
fully commercialized unit at the Amoco Corporate Research facility in Naperville, Illinois.
The CFM500 unit was installed in a pilot plant cooling water system within the research
and development facility. Throughout this evaluation period, a number of operational para-
meters were changed to investigate the relationship between flow, wall temperature, and
localized and general corrosion attack. The effect of operational upsets, such as chlorine
overfeed or inhibitor underfeed on the corrosion and fouling behavior, was also investigated.
The CFM500 unit was installed in the hot return line of a recirculating cool i ng water
system fed with municipal makeup water. Typical recirculating cooling water quality was
pH 8.5 t o 9.0, alkalinity 300 to 350 ppm as cal ci um carbonate (CaCO3), calcium 280 to 320
ppm as CaCO3, and total dissolved solids (TDS) of 800 to 1200 ppm equating to a Langelier
Scale Index (LSI--CaCO 3 scaling tendency) between + 1.9 and +2. 4 at the estimated max-
i mum wall temperature of 43~ (110~ corresponding to a relatively "l i ght -dut y" service.
The cooling water was continuously treated with a zinc-molybdate-based program incorpo-
rating zinc mol ybdat e as primary carbon steel corrosion inhibitor and tolytriazole (TTA) as
a "copper al l oy" corrosion inhibitor coupled with hydroxyet hyl enedi phosphoni c acid
(HEDP) and a compl ex pol ymer for scale deposit control and particulate dispersancy. Chlo-
rine was also fed on a periodic basis to control microbiological activity.
Baseline Tests
The CFM500 unit was initially set up with a constant heat input of 300 W and a flow
rate of 0.76 ms- t to generate a skin temperature of ~43~ and reflect the typical extreme
hydrodynami c operating conditions in the cooling water system. A typical set of raw data
collected during this phase of testing is illustrated in Fig. 2. These data show (1) no fouling
tendency as the cooling water quality falls well within the chemical treatment scale control
limit of +2. 5 LSI, (2) a mean general corrosion rate of 1.5 mpy, and (3) a local attack factor
of 0.0002, suggesting more generalized corrosion over this period.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
W I N T E R S E T A L . O N S I M U L T A N E O U S C O R R O S I O N 235
U n i t # I
C o r r o s i o n D a t a
+I .OOE+Q1
L P K
M a x : + ! . 4 9 E + 0 3
M i n : + ! . 4 0 E + 0 3
M e a n : + ! . 4 6 E + 0 3
S D e v : + 2 . 2 2 E + 0 1
C V a r :
U n i t s : O h m s
~ 1 . 0 0 E - O 2
, ' 1 . p O E t 0 9 t ,
I r I I i r ~ r i r J r I i i ~ i r
Z R A
M a x : + 0 . 4 5 E - 0 5
M i n : + 0 . 3 9 E - 0 5
M e a n : + 0 . 4 2 E - 0 5
S D e v : + 0 . 2 3 E - 0 6
C V a r : + 0 . 1 3 E - 0 7
U n i t s : A m p s
t-1.00E-Q2
. I . 9 0 E . - ~ , ~ l ~ i r T 1 , , = r , l , , l ,
E P N
M a x : + 0 . 5 2 E - 0 5
M i n : + 0 . 2 0 E - 0 5
M e a n : + 0 . 3 0 E - 0 5
S D e v : + 0 . 4 5 E - 0 6
C V a r : + 0 . 6 9 E - 0 7
U n i t s : V o l t s
*'1 .nOE-O6
+ I J O O E r 1 O , r i I I r i f i I r I r r I I t = t x ,
ECN
M a x : + 0 . 1 2 E - 0 8
M i n : + 0 . 6 8 E - 0 9
M e a n : + 0 . 8 3 E - 0 9
S D e v : + 0 . 6 6 E - ! 0
C V a r : + 0 . 5 3 E - I I
U n i t s : A m p s
I . . , 1 , 0 0 ~ , , , , - 0 5 , i ' 9 f r r ' : ! ' ~ [ , T I ; . , , T
0 0 : 0 0 T i m e O f D a y 2 4 : 0 0
FIG. 2--24-h time record of raw corrosion and summary data recorded during baseline tests.
E R N
M a x : + 5 . 5 9 E + 0 3
M i n : + 2 . 2 0 E + 0 3
M e a n : + 3 . 6 5 E + Q 3
S D e v : + 5 . 6 8 E + 0 2
C V a r : + 8 . 8 3 E + 0 1
U n i t s : O h m s
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
236 ELECTROCHEMICAL NOISE MEASUREMENT
U n i t # i
C F M 5 0 0 S u m m a _ ~ y
1 0
0 r I I ] J l I t I f I I I I i ] I I I 1 T
; I . OOE.,.O0
+ 1 . D O E : - 0 3 ,
9 ( - 0 .&OE-02
; , r I r I I , I t ' I I , * r I i
00:00 Time Of Day ,?.&:O0
FIG. 2--Continued.
G e n e r a l R a t e
M a x : + l . 5 0 E + 0 0
M i n : ~ 1 . 4 2 E + 0 0
M e a n : + 1 . 4 5 E + 0 0
S D e v : + 0 . 2 2 E - 0 !
C V a r : + 0 . 3 5 E - 0 3
U n i t s : m p y
L o c a l A t t a c k F a c t o r
M a x : + 0 . 2 7 E - 0 3
M i n : + 0 . 1 5 E - 0 3
M e a n : + 0 . 2 0 E - 0 3
S D e v : + 0 . 1 9 E - 0 4
C V a r : + 0 . 1 8 E - 0 5
U n i t s :
H T R
M a x : + 0 . 3 0 E - 0 4
M i n : - 0 . 7 0 E - 0 4
M e a / l : - 0 . l i E - 0 4
S D e v : + 0 . i 2 E - 0 4
C V a r : - 0 . 1 4 E - 0 4
i t s : h r . f t 2 . ~
t
Biocide Overfeed
The unit has demonstrated the sensitivity of corrosion behavior to biocide overfeed in a
previous field trial [10]. A similar response to chlorine overfeed was observed during the
evaluation at Naperville and is illustrated in Fig. 3. This time record shows the increase in
local attack factor corresponding to a period of chlorine overfeed in the cooling water system.
The general rate of corrosion (derived from the LPR measurement) did not detect the in-
creased localized activity. Transients in the local attack factor trend corresponded to transients
in the ECN, whi ch is symptomatic of pitting behavior. Not e that the chlorine overfeed was
detected by an operator who was able to identify quickly the process excursion by the online
visual histogram ( VI SI COR TM software) display of the localized corrosion activity.
Several local corrosion transients related to chlorine overfeed occurred during the early
stages of testing. Once recognized, the operating remedy was to reduce chlorine feed and
blowdown, as indicated by the reduction in local attack factor in the last 2 h of the time
record (see Fig. 3). Ultimately, the incidence of upsets lessened as a result of tighter chlorine
feed control implemented t o minimize chemical wast age/ overfeed and potential corrosion.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
WINTERS ET AL. ON SIMULTANEOUS CORROSION 237
U n i t # I
C F M 5 0 0 S u m m a r y
1 0
J
r
I I ~ I , , , I I ~ 7 I I I I
G e n e r a l R a t e
M ~ x : + ! . 4 3 E + 0 0
M i n : + 0 . 2 1 E - 0 i
M e a n : + 1 . 3 7 E + 0 0
S D e v : ~ 0 . 4 6 E - 0 !
C V a r : + 0 . 1 5 E - 0 2
U n i t s : m p y
" 1 . O O E + O 0
R e d u c t i o n
C h l o r i n e o v e r f l o w in c h l o r i n e
f e e d ",
9 ~ ~ . . . . . .~.~.,,.~,~,~,.~.,. ~ . . . . . .
* 1 . ~ 0 E r 0 3 , r i I J , , T T I f f r ~ ~ I , ~ ~ , ~
L o c a l A t t a c k F a c t o r
M a x : + 0 . 1 4 E - 0 1
M i n : + 0 . 9 7 E - 0 3
M e a n : + 0 . 5 5 E - 0 2
S D e v : + 0 . ! 5 E - 0 2
C V a r : + 0 . 4 3 E - 0 3
U n i t s :
~.4oE-o2 h ~ f R
- - M a x : + 0 . 2 0 E - 0 4
M i n : - 0 . 6 0 E - 0 4
- - M e a n : - 0 . 1 5 E - 0 4
S D e v : + 0 . 1 1 E - 0 4
C V a r : - 0 . 8 0 E - 0 5
Z n i t s : h r . f t = . ~
i
- 1 .,OOE:'O 3 , r , [ , , r l , I , I J [ , ! a , , r T J
0 0 : 0 0 T i m e O f D a y 2 4 : 0 0
FIG. 3- - 24- h time record of corrosion and f oul i ng summary data showing the increase in the local
attack f act or as a result o f chlorine overfeed and the remedial effect o f reducing chlorine feed.
St at i c Test
Before i sol at i ng t he unit wi t h ent r apped cool i ng water, zero flow, and zer o appl i ed heat
flux, the CFM500 was i nt ent i onal l y f oul ed/ s cal ed by i ncreasi ng the heat i nput ( ~400 W)
and reduci ng flow vel oci t y ( - 0 . 3 0 ms-1) to generat e a ski n t emperat ure of - 8 2 ~ ( - 180~
LSI ~ +3. 0) and cl ear l y to exceed the scal e cont rol capabi l i t y of the cool i ng wat er treatment.
Ul t i mat el y, the "s hut - i n" si mul at i on was desi gned to reflect a t ypi cal operat i ng si t uat i on in
whi ch a " f oul ed" unit is shut in for some t i me, such as duri ng a mai nt enance shutdown. To
si mul at e st agnant condi t i ons in a heat exchanger, t he heat flux and flow t hrough the unit was
shut off compl et el y for seven days. The cool i ng wat er in t he model heat exchanger was
i sol at ed, and corrosi on moni t or i ng in the st agnant condi t i ons cont i nued. Af t er appr oxi mat el y
t hree days of passi ve behavi or, t ransi ent s in the current and pot ent i al noi se (i ndi cat i ve of
l ocal i zed corrosi on) were r ecor ded (Fi g. 4). The operat i ng dat a showed a cl ear combi nat i on
of general and l ocal i zed corrosi on: a st ep reduct i on of general corrosi on as rel at ed to LPR
measur ement and a cor r espondi ng i ncrease in l ocal i zed cor r osi on t endency as r el at ed to EPN
and ECN measurement . Af t er the shut -i n test, t he modes of corrosi on were confi rmed to be
i ni t i at i on of l ocal i zed cor r osi on beneat h deposi t s based on t he observat i on of the presence
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
238 ELECTROCHEMICAL NOISE MEASUREMENT
U n i t # I
C o r r o s i o n D a t a
+ 1 ~ 0 0 E + 0 1
E n O E * 0 5 , r I l I r
~
O O E - 0 2
O e ~ 0 9 , , , I r
r I I I I I I r ! r f I I
r e f I r i I I
" I . O O E - O 2

D
~ , ~ 9 ,--.,7-..r M-' : ' ~ l ~
t " p , i i i , l I i
+1.00E~I
L P R
M a x : + 1 . 7 8 E + 0 4
M i n : + I . 1 9 E + 0 3
M e a n : + 2 . 8 8 E + 0 3
S D e v : + 4 . 4 0 E + 0 3
C V a r : + 6 . 7 3 E + 0 3
U n i t s : O h m s
Z R A
M a x : + 0 . 3 5 E - 0 6
M i n : + 0 . 2 7 E - 0 6
M e a n : + 0 . 3 ! E - 0 6
S D e v : + 0 . 2 0 E - 0 7
C V a r : + 0 . 1 3 E - 0 8
U n i t s : A m p s
E P N
M a x : + 0 . 4 6 E - 0 5
M i n : + l . 0 0 E - 0 6
M e a n : + 0 . 2 0 E - 0 5
S D e v : + 0 . 6 2 E - 0 6
C V a r : + 0 . 2 0 E - 0 6
U n i t s : V o l t s
E C N
M a x : + 0 . 4 2 E - 0 8
M i n : + I . 0 0 E - 1 0
M e a n : + 0 . 2 0 E - 0 9
S D e v :
C V a r : + 0 . 2 5 E - 0 9
U n i t s : A m p s
M a x : + 2 . 0 8 E + 0 4
M i n : + 8 . 3 7 E + 0 2
M e a n : + 9 . 6 2 E + 0 3
S D e v : + 4 . 7 7 E + 0 3
C V a r : + 2 . 3 7 E + 0 3
U n i t s : O h m s
~ - 1 . , 0 0 E + 0 5 , v ~ I r r f ~ i i t i 1 . , , i ,
0 0 : 0 0 T i m e O f O a y 2 4 : 0 0
FIG. 4--24-h time record of raw corrosion data collected during the static test illustrating current
and potential noise transients associated with localized corrosion.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
WINTERS ET AL. ON SIMULTANEOUS CORROSION 2 3 9
of active tubercles on the internal tube surface (activity was verified by boroscopic exami-
nation) and an inordinately high iron content of precipitated solids in the "ent rapped" cooling
water. It is t hought that the interaction of general and localized corrosion reflects suppression
of generalized corrosion by cathodic protection of locally corroding regions.
Acidification/Repassivation
After the shut-in test, the unit was run with a heat input of 400 W and a flow rate of 0.46
ms -~ to induce a positive fouling tendency. Increasing the fouling rate in the unit, which
resulted in an increased heat transfer resistance value, was intended to assess online repas-
sivation of under-deposit corrosion under nonopt i mum operating conditions as well as to
enhance fouling on the CFM500 unit in preparation for online acidification. Acidification of
the cool i ng water with hydrochloric acid was designed to simulate an on-stream descaling
treatment or an extreme acid pH excursion. Summary fouling and corrosion data before and
after acidizing are seen in Fig. 6. Before acidizing, the CFM500 unit flow rate was progres-
sively reduced to 0. 20 ms-~ to induce the heaviest fouling tendency and assess the impact
of running at very low flow rates. Immediately before acidification, the CFM500 unit was
internally inspected by a boroscopic unit to establish the baseline "f oul ed" condition. This
inspection found:
1. that the CFM500 unit was heavily scaled with a whi t e/ t an deposit, subsequently found
to be mainly cal ci um carbonate;
2. that the previously active tubercles formed during the shut-in test had been repassivated
in-service; and
3. that deposits had been mechanically dislodged by the boroscope, which woul d explain
the reduced level of HTR after internal inspection.
After inspection, the CFM500 unit was returned to service and the cool i ng water pH
depressed to < 1 with hydrochloric acid. Figure 6 illustrates the increased corrosion activity
on all the electrochemical techniques during the acidification. This corresponded to a max-
i mum general corrosion rate of 2.25 mm/ year . The HTR time record (see Fig. 5) also clearly
shows the effect of mechanical removal of deposit during inspection and the effect of acid-
ification. Mechanical cleaning of the unit reduced HTR to - 3 0 percentile of t he range
between the baseline HTR and full-scale HTR value. For the first 11 h of this time record
(Fig. 5), the HTR value was full scale. After the addition of acid, the HTR trace returned
to a baseline level once all the deposit had been removed.
After cessation of acidification, the CFM500 unit showed a repassivation trend, as illus-
trated by the downward trend in corrosion rate through the following day (Fig. 7). However,
an increased general rate of corrosion and local attack factor was seen the next day (Fig. 8).
Based on the unusually high level of general corrosion, the unit operator investigated and
found the main bl owdown valve was j ammed open, resulting in increased discharge to waste
and low inhibitor level in the cool i ng water. Remedyi ng this problem corrected the inhibitor
dosage and promot ed full repassivation of the CFM500 unit over the next 24 h.
Localized Corrosion Versus General Corrosion
On several occasions, transients in the electrochemical current and potential noise (with
associated transients in the local attack factor) correlated with instability in the LPR trace
(with associated increase in the polarization resistance and reduction in general corrosion
rate), as illustrated in Figs. 9 and 10. The initiation and growth of pits on the sensor surface
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
240 E L E C T R O C H E M I C A L N O I S E M E A S U R E M E N T
U n i ~ #1
C F M 5 0 0 S u m m a r y
I Q
- - A c i d i f i c a t i o n
0 , , r , J I I ; , , , ~ I
9 ' 1 . O Q E , ~ 3 Q
l i a I L t T r '
9 . - ~ . ~ . ~ .
- - 9 ' - ~ , i : ' , ~ : . ;
9 . - ~ . . . . - i . - " - ~ " ~ ; ~
1 ; P Q E F O 3 1 , ' l f , I , - . " ; " i k . . [ ,~ , t I "~ "
4 ~ . t ~ E - O 2
*~JO(]E = 0 3 r z :
00:00
[ r f r
L
I J 1 I r I I T : l I . T I I r i I i I
T i m e O f D a y 2 4 : 0 0
G e n e r a l R a t e
M ~ x : + 8 . 9 9 E + 0 1
M i n : + 0 . 2 1 E - 0 i
M e a n : + 4 . 0 7 E + 0 0
S D e v : + 9 . 6 8 E + 0 0
C V a r : + 2 . 3 0 E + 0 1
U n i t s : m p y
L o c a l A t t a c k F a c t c r
M a x : + 0 . 4 3 E + 0 0
M i n : + 0 . 1 3 E - 0 4
M e a n : + 0 . 3 8 E - 0 2
S D e v : + 0 . 1 7 E - 0 1
C V a r : + 0 . 7 3 E - 0 1
U n i t s :
H T R
M a x : + 0 . 1 0 E + 0 0
M i n : - 0 . 3 6 E - 0 1
M e a n : + 0 . 2 3 E - 0 2
S D e v : + 0 . 5 0 E - 0 2
C V a r : + 0 . ! I E - 0 !
h i t s : h r . f t 2 . ~
J
Z32
r i I r I i i I ! r 1 I I r
B l o c k T e m p
M a x : + 2 . 2 2 E + 0 2
M i n : + 5 . 8 3 E + 0 !
M e a n : + 1 . 6 6 E + 0 2
S D e v : + 4 . 0 9 E + 0 1
C V a r : + I . 0 1 E + 0 1
U n i t s : ~
z3Z W a l l T e m p
M a x : + 2 . 1 9 E + 0 2
M i n : + 5 . 8 3 E + 0 1
_ _ .:~ M e a n : + I . 6 3 E + 0 2
-; ~ S D e v : + 4 . 0 4 E + 0 !
\ '
! ; ~ ' ~ - - ~ " - ~ " ~ C V a r : + ! . 0 0 E + 0 1
!. . U n i t s : ~
1 2 , , , , , , L r L 9 I , L I i ~ i
FIG. 5- - 24- h time record of corrosion and foaling summary data showing the increased general rate
of corrosion during acidification, the reduction in HTR, and the effect of acidification on the wall
temperature.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
WINTERS ET AL. ON SIMULTANEOUS CORROSI ON 2 4 1
U n i t # !
C o r r o s i o n D a t a
r 1 . O O E ' , " O 1
- - A c i d i f i c a t i o n
~ I . , Q O E * 0 5 ,
+ 1 . O O E - O 2
+ 1 . , O O E ; ' 0 9 t I I I I I 1 I I I
+ 1 . 0 0 E - Q 2 " ' . : . ; :
~ v
i ~ . '
t i t ~ I I ~ i 1 i
- - ~=.~ . ~.. - ~ 3 ~ : ~ s t h , ~
- ' I O E ; - 0 6 t I l I r l I r i I = ] t = I I I I 1 l [
. ~ O ( ] E - 0 6 - . - . . r
_ : - : : - . . . t . ~
L P R
M a x : + 5 . 4 4 E + 0 3
M i n : + 2 . 3 5 E + G I
M e a n : + 1 . 2 0 E + 0 3
S D e v : . 7 . 3 0 E + 0 2
C V a r : + 4 . 4 5 E + 0 2
Units : O h m s
Z R A
M a x : + 0 . 4 9 E - 0 3
M i n : + 0 . 3 8 E - 0 6
M e a / l : + 0 . 1 6 E - 0 4
$ D e v : + 0 . 3 0 E - 0 4
C V a r : + 0 . 5 5 E - 0 4
U n i t s : A m p s
E P N
M a x : + 0 . 7 0 E - 0 2
M i n : + 0 . 2 6 E - 0 5
M e a n : + 0 . 9 0 E - 0 4
S D e v : + 0 . 5 9 E - 0 3
C V a r : + 0 . 3 9 E - 0 2
U n i t s : V o l t s
ECN
M a x : + 0 , 9 6 E - 0 6
M i n : + 0 . 1 9 E - 0 9
M e a n : + 0 . 3 3 E - 0 7
S D e v : + 0 . 1 3 E - 0 6
C V a r : + 0 . 5 2 E - 0 6
U n i t s : A m p s
~ 1 . 0 0 E + O l ~ i ~ l : ~ ,
_ _ ~ M a x : + I . 5 1 E + 0 7
~ ~ ; . _ ~ j : ~ , ~ : ~ , : : , : , ~ , M i n : + 1 . 6 3 E + 0 1
.-:~,---~,:.-:~,,~.~< ~.~:..tT~'~:-:;~', M e a ~ : + I . 3 6 E + 0 4
" ~ 9 .... " "" S D e v : + 4 . 0 1 E + 0 5
" r . : . ~ - . : ~ ~ . . . . ~. :.. ~-~ :, : .:./.:..~- C V a r : + I . 1 9 E + 0 7
- - . ' h A t : " " ' ~! 9 U n i t s : O h m s
' , I . B O E ~ . 0 5 i ' ' 1 I ~ I , :
9 ' = r ] ~ , t
O 0 : O O T i m e O f D a y Z & : O 0
F I G . 6- - 24- h time record of raw corrosion data collected during acidification illustrating the in-
creased activity on al l the electrochemical techniques.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
242 ELECTROCHEMI CAL NOI SE MEASUREMENT
U n i ~ # 1
C F M 5 0 0 S u m m a r y
1 0
r I I [ I I I [ I
G e n e r a l R a t e
M a x : + ~ . 4 6 E + 0 0
M i n : + I . 2 0 E + 0 0
M e a n : + 2 . 7 2 E - 0 0
S D e v : + I . 0 9 E + 0 0
C V a r : + 0 . 4 4 E + O O
U n i t s : m p v
+I .OOE~O
I 1 I ~ I I ' I l I [ ' l " I ; ' I " [ I I - - " ' "
L o c a l A t t a c k F a c t o r
M a x : + 0 . 2 0 E - 0 1
M i n : + 0 . 2 3 E - 0 3
M e a n : + 0 . 5 7 E - 0 3
S D e v : + 0 . 8 6 E - 0 3
C V a r : + 0 . 1 3 E - 0 2
U n i t s :
+0 . 4 0 E- 0 2
- :
" l . n 0 E , ' 0 3 1 , r I t r , , , I r , L , I I , , , , ,
00:00 Time Of Day Z4:00
M a x
M i n
M e a n
S D e v
C V a r
~ n i t s :
H T R
: + 0 . 5 5 E - 0 1
: - 0 . 1 6 E - 0 2
: + 0 . 2 7 E - 0 3
: + 0 . 1 5 E - 0 2
: + 0 . 8 0 E - 0 2
h r . f t =. Q F / B T ~
FIG. 7- - 24- h time record of corrosion and fouling summary data showing the downward trend in the
general corrosion rate during repassivation.
appear to protect cathodi cal l y the surrounding surface area and, consequentl y, i ncrease the
measured pol ari zati on resistance suggest i ng s ome associ at i on bet ween l ocal i zed and gener-
al i zed attack. Thi s type o f behavi or was observed onl i ne as wel l as during the shut-in test.
Moreover, the occurrence o f this behavi or on both the noi se and LPR measurements suggest s
that similar corrosi on processes occurred over all the corrosi on sensors.
Di s c us s i on
The results obtai ned during this eval uati on cl earl y demonstrate the capabi l i ty o f the c om-
bi ned corrosi on and f oul i ng moni tori ng syst em to moni t or corrosi on and differentiate type
o f attack under deposi ts, whi l e expos i ng the shortcomi ngs o f convent i onal moni t ori ng ap-
proaches. The i nformati on obtai ned f rom the el ect rochemi cal met hods o f corrosi on appraisal
has s hown that a cont i nuous i ndi cati on o f corrosi on status beneath a scal e or f oul i ng deposi t
can be obtained. The unit and the techni ques used therein have proven sensi ti ve in changes
in cool i ng water process and chemistry. Changes in l ocal attack factor and the general rate
o f corrosi on have gi ven an i nsi ght i nto the passi vat i on/ repassi vat i on behavi or o f the heat
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
WI NTERS ET AL. ON SI MULTANEOUS CORROSI ON 243
U n i t # I
C F M 5 0 0 S u m m a r y
1 0
O I : , , I 1 I [ J I I , 1 I T I I r
G e n e r a l R a t e
M a x : + 2 . 7 5 E + 0 0
M i n : + ! . 3 2 E + 0 0
M e a n : + 2 . 1 7 E + 0 0
S D e v : + 0 . 2 7 E + 0 0
C V a r : + 0 . 3 4 E - 0 !
U n i t s : m p y
+ 1 . OOE+Orl
m
. . . . . . . . . ~. ~-~ ;~;: : ~ : ~ . : ~ : ~ 2 ~ :~ " " " ~ ; : : ! ~ ~ ' : ~ : - U ' ~
9 , " I . D O E : : . O ] I ~ , I ' " ~, " r , ~ ' , , ~ , : - u ; ~ - 4 ~ ' ; ; ' ; " ~ , ' i ~ a ' ' ~ ' ~ ' ~ ' : ' ' ' x ' = - 5 ; ; ' v ; ' - i " ; " I l I I " l ~
+ O . , ; O E - O 2
L o c a l A t t a c k F a c t o r
M a x : + 0 . 5 4 E - 0 2
M i n : + 0 . 3 9 E - 0 3
M e a n : + 0 . 1 6 E - 0 2
S D e v :
C V a r : + 0 . 5 3 E - 0 3
U n i t s :
H T R
M a x : + 0 . 6 7 E - 0 3
M i n : - 0 . 1 8 E - 0 3
M e a n : + 0 . 1 8 E - 0 3
S D e v
C V a r
: + 0 . 5 6 E - 0 4 1
- - : + 0 . 1 7 E - 0 4
I J - I J ] O E r 0 3 , , ~ 1 , , , [ r r ~ ~ r ~ I r r , ~ L T
0 0 : 0 0 T i ~ of Day 2 4 . : 0 0
FIG. 8--24-h time record of corrosion and fouling summary data showing an increased general
corrosion rate and increased local attack factor as a result of low inhibitor level.
transfer surface during normal operati on and after process upsets such as bi oci de overf eed,
acidification, and so forth.
Severe l ocal i zed attack was not detected during this eval uati on al though s ome transient
bursts o f pitting behavi or were observed in the el ect rochemi cal noi s e measurement data.
Not e that the l ocal attack factor does not indicate rate but, in associ ati on wi th the rate
esti mates from LPR, can be used to provi de an i ndi cati on o f worst case pit penetration rates.
It i s difficult to esti mate actual pit penetration rates because the geomet ry and number o f
the pits were not determined, and, because pits have a l i fe c yc l e (birth, growth, and death),
the active l i fe o f a si ngl e pit is difficult to determine.
From the vi ewpoi nt o f corrosi on control , it is much more important to be aware o f s ys t em
stability and whether there is general corrosi on, pit initiation, or pit propagation. At the pit
initiation stage, there is no real criticality, but it i s an early warni ng sign. More important
operati onal l y is to kno w when initiation occurs and to track subsequent change. The key
steps are: ( l ) initiation (alarm), (2) repassivation ( no action), and (3) propagat i on- - s us t ai ned
or intensified l ocal i zed activity (action). The individual pit penetration rate esti mates need
interpretation because they may be very hi gh but the l i feti me may be so short as to be
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
244 E L E C T R OC H E MI C A L N OI S E ME A S U R E ME N T
U n i t # i
C o r r o s i o n D a t a
_ ~ O O E ' ~ ' O 1
[ , , , * ~ I , , , , , I , , , ,
I + 1 . 0 0 E - 0 2
- 2
!*1.p0Er09. = , I . . , , , , , I , ,
I I r I
L ~ R
M a x : + 1 . 5 2 E + 0 4
M i n : + 2 . 5 6 E + 0 3
M e a n : + 4 . 6 B E + 0 3
S D e v : + 2 . 7 1 E + 0 3
C V a r : + ! . 5 6 E + 0 3
U n i t s : O h m s
Z R A
M a x : + 0 . 2 0 E - 0 5
M i n : + 0 . ! 2 E - 0 5
M e a n : + 0 . 1 6 E - 0 5
S D e v : + 0 . 2 0 E - 0 6
C V a r : + 0 . 2 6 E - 0 7
U n i t s : A m p s
~ ' 1 . 0 0 6 - 0 2
E P N
M a x : + 0 . 9 5 E - 0 5
M i n : + 0 . 1 1 E - 0 ~
M e a n : + 0 . 1 ~ E - 0 5
S D e v : + 0 . 6 5 E - 0 6
C V a r : + 0 . 2 6 E - 0 6
U n i t s : V o l t s
I ' 1 . 0 0 E - 0 6
e 1 - ~ O E r l O l I , t f , * z 9 " ' J ~ ' ,~ I ~ I t ~ 1 I
E C N
M a x : + 0 . 3 1 E - 0 7
M i n : + 0 . 1 9 E - 0 8
M e a n : + 0 . 3 6 E - 0 8
S D e v : + 0 . 1 7 E - 0 8
C V a r : + 0 . 8 1 E - 0 9
U n i t s : A m p s
~' I . ~ 1
9 . . . . . , , . . . : : 9 . . . . " - . ~ -
+ I . ~ ] O E . * C ] 5 , I t I I L ~ I 9 ' E ~ :
( ] O : O 0 T i m e O f O a y 2 4 . : 0 0
E R N
M a x : + 1 . 4 8 E + 0 3
M i n : + 1 . 5 2 E
M e a n : + 4 . 8 0 E + 0 2
S D e v : + 1 . 4 1 E + 0 2
C V a r : + 4 . 1 5 E + 0 1
U n i t s : O h m s
FIG. 9- - 24- h time record o f raw corrosion data illustrating the current and pot ent i al noise transients
and LPR instability caused by localized corrosion.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
WI NT E RS ET A L ON SI MUL T ANEOUS C O R R O S I O N 245
U n i t # l
C F M 5 0 0 S u m m a r y
10
, 0 , , , T I I I I I r I I ~ ' ~ ' ~ I - - ~ ' !
+I.00E+O0
- - " . - _
9 : . . . . . . . - : . - . . . . . : . , , i ? " " : . 3 : . ,
" J ] g E r 0 3 1 I I [ I I r I t / I r l I " 1 ~ [ r r " i " ' ~ [ " I "
*O.~E-OZ
I-1.DOErO 3r , , I , , , l , I r , , , [ I i i
00:00 Time Of O a y Z&:O0
G e n e r a l R a t e
M a x : + 0 . 8 2 E
M i n : + 0 . 1 4 E + 0 0
M e a n : + 0 . 5 9 E
S D e v : + 0 . 2 5 E + 0 0
C V a r : + 0 . 1 i E + 0 0
U n i t s : m p y
L o c a l A t t a c k F a c t o r
M a x : + 0 . 1 8 E - 0 Z
M i n : + 0 . 1 2 E - 0 2
M e a n : + 0 . 2 3 E - 0 2
S D e v : + 0 . 1 0 E - G 2
C V a r : + 0 . 4 4 E - 0 3
U n i t s :
Ma x
M i n
M e a n
S D e v
C V a r
[ n i t s :
blTR
: + 0 . 1 9 E - 0 2
: + 0 . 1 6 E - 0 2
: + 0 . 1 7 E - 0 2
: + 0 . 3 5 E - 0 4
: + 0 . 7 5 E - 0 6
h r . f ~ Z . O F / B ~ " 7 ~
I
i
FIG. 10- - 24- h t i me record o f corrosion and f oul i ng summary data showing the reduction in the
general corrosion rate and transient increases in the local attack factor.
insignificant; on the other hand, the pit penetration rate may be very significant i f attack i s
sustained. It is therefore consi dered to be dangerous to attempt to linearize pit penetration
rates, rather to use i ndi cati ons o f pitting initiation as a means o f process control to hel p
operati ons in a safe, nonpitting regi me.
So, what was learned about this particular cool i ng water syst em? Overall, it was f ound
that the normal c ool i ng water servi ce was not particularly aggressi ve from the standpoint o f
f oul i ng and related corrosi on activity. Indeed, under normal operating ( hydrodynami c) con-
ditions, the cool i ng water s howed no f oul i ng tendency and was not corrosi ve to carbon steel:
this was attributed to l ow chl ori de content and hi gh al kal i ni t y/ pH coupl ed wi th a hi ghl y
ef f ect i ve anodi c/ cat hodi c corrosi on inhibitor. Thi s was reflected by an average general cor-
rosi on rate o f - 0 . 0 1 2 5 mm/ y e a r ( f rom LPR data) and mi ni mal l ocal i zed attack seen on the
CFM500 unit during the trial. Thes e results compared favorabl y wi th corrosi on coupon re-
sults and post exposure assessment o f the CFM500 sensors: average general corrosi on rate
o f 0. 0025 mm/ y e a r wi th s ome l ocal i zed attack correspondi ng to a corrosi on rate o f 0. 125
mm/ ye ar .
Not wi t hst andi ng the general l y unaggressi ve nature of the c ool i ng water, it was possi bl e
t o sti mul ate f oul i ng on the CFM500 unit by appl yi ng higher heat duty or l ower fl ow rates
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
246 ELECTROCHEMICAL NOISE MEASUREMENT
than that ordi nari l y exper i enced in the cool i ng wat er syst em or both. In doi ng this, it was
possi bl e to demonst rat e the sensi t i vi t y of the CFM500 to hydr odynami c changes of heat flux
and flow rat e as wel l as to verify the operat i ng l i mi t s and scal e cont rol capaci t y of the
chemi cal treatment.
In t erms of l ocal i zed corrosi on tendency, the cool i ng wat er showed some defi ni t e insta-
bi l i t y as a resul t of chl ori ne overfeed or i nhi bi t or underfeed or both. Accent uat ed in the
presence of deposi t s, the l ocal i zed corrosi on act i vi t y faci l i t at ed opt i mi zat i on of bot h the
chl ori ne and i nhi bi t or t reat ment , and rei nforced t he need to mai nt ai n effect i ve scal e deposi t
control. Cri t i cal l y, this l ocal i zed corrosi on sensi t i vi t y offers a means of i dent i fyi ng an unsta-
bl e operat i ng situation before any si gni fi cant at t ack has mani fest ed. Of concern was t he fact
that LPR r emai ned unaffect ed by corrosi vi t y changes ot her than duri ng the aci di fi cat i on
t reat ment . Al so, the evi dent reduct i on of LPR measur ement coi nci dent wi t h l ocal i zed cor-
rosi on act i vi t y quest i ons the convent i onal i nt erpret at i on of LPR measur ement data.
Conclusions
1. Moni t or i ng corrosi on and foul i ng si mul t aneousl y in a cool i ng wat er syst em enhances
the capabi l i t y of measuri ng represent at i ve corrosi on that occurs beneat h scal e or foul i ng
deposi t s when compar ed wi t h convent i onal techniques.
2. The combi ned moni t or i ng approach has been used to st udy t he oper at i ng l i mi t s of
chl ori ne (as a bi oci de) and some corrosi on i nhi bi t ors used to cont rol cor r osi on and
foul i ng in a pi l ot pl ant cool i ng wat er system.
3. The CP-M500 unit has proven sensi t i ve to t he onset of l ocal i zed corrosi on and has
i dent i fi ed unst abl e operat i ng in the pi l ot pl ant cool i ng wat er syst ems.
4, The CFM500 unit has faci l i t at ed the opt i mi zat i on of bi oci de and i nhi bi t or t reat ment in
a pi l ot pl ant envi ronment and has been used effect i vel y to cont rol scal e deposi t i on in
this part i cul ar appl i cat i on.
5. Duri ng this study, suppressi on in general corrosi on rat e (as deri ved from LPR mea-
surement ) has oft en cor r esponded to i ncreased l ocal i zed act i vi t y as det ect ed by el ec-
t rochemi cal noi se measurement s. Thi s may reflect t he cat hodi c prot ect i on afforded by
l ocal l y corrodi ng regi ons and is si gni fi cant in t erms of whi ch t echni ques are used to
assess bot h general and l ocal i zed corrosi on damage.
References
[1] Hladky, K. and Dawson, J. L., Corrosion Science, Vol. 23, 1982, p. 231.
[2] Dawson, J. L., Eden, D. A., and Hladky, K., "Electrochemical Noi se--Some New Developments
in Corrosion Monitoring," paper presented at UK Corrosion 83, Birmingham, UK, CCEJV/ICorrSt,
Nov. 1983.
[3] Dawson, J. L., Farrell, D. M., Aylott, P. J., and Hladky, K., Paper 31, in Corrosion 89, NACE, New
Orleans, LA, 1989.
[4] Hladky, K., UK Patent Application 8,200,196, Jan. 1982, US Patent 4,575,678, March 1986.
[5] Eden, D. A., Dawson, J. L., and John, D. G., UK Patent Application 8,611,518, May 1986.
[6] Stem, M. and Geary, A. J., Journal of the Electrochemical Society, Vol. 56, 1957, p. 104.
[7] DATS Fouling Monitoring System, Bridger Scientific, Inc., P.O. Box 1923, Sandwich, MA 02563.
[8] Zuniga, P. O., Klee, H., Miller, K., and Winters, M. A., Corrosion 90, NACE, Houston, TX, 1990.
[9] Zuniga, P. O., Miller, K., and Winters, M. A., "Cooling Water Fouling Monitor Senses Upsets,
Evaluates Changes." Chemical Processing, April 1990.
[10] Winters, M. A., Stokes, P. S. N., Zuniga, P. O., and Schlottenmier, D. J., Corrosion Science, Vol.
35, No. 5-8, 1993, pp. 1667-1675.
[11] Winters, M. A., Stokes, P. S. N., Zuniga, P. O., and Schlottenmier, D. J., Paper 392, Corrosion 93,
NACE, New Orleans, 1993.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
Di gby D. Macdonal d, l Chun Liu, 2 and Mi chael P. Manahan, S i r : . 3
Electrochemical Noise Measurements
on Carbon and Stainless Steels in
High Subcritical and Supercritical
Aqueous Environments
REFERENCE: Macdonald, D. D., Liu, C., and Manahan, M. E, St., " El e c t r o c h e mi c a l No i s e
Me a s u r e me n t s o n Ca r b o n a n d St a i nl e s s St e e l s i n Hi g h Subc r i t i c a l a n d Supe r c r i t i c a l Aq u e -
o u s Env i r o nme nt s , " Electrochemical Noise Measurement for Corrosion Applications, ASTM
STP 1277, Jeffery R. Kearns, John R. Scully, Pierre R. Roberge, David L. Reichert, and John
L. Dawson, Eds., American Society for Testing and Materials, 1996, pp. 247-265.
ABSTRACT: The electrochemical noise induced in a pair of corroding carbon steel probes in
simulated heat transport circuit fluids of a supercritical fossil-fueled power plant at temperatures
up to 550~ and at pressures as high as 276 bar (4000 psi) was measured as a function of
oxygen concentration, flow rate, temperature, and pressure. Additionally, electrochemical noise
measurement (ENM) was explored as a means of detecting and monitoring stress corrosion
cracking in sensitized Type 304 SS in high-temperature water (up to 288~ Experimental
results from these studies indicate that the electrochemical noise induced from corrosion of
steels in high-temperature aqueous systems can be measured in a reproducible manner. The
magnitude of the electrochemical noise responds sensibly to changes in the chemical and
physical properties of the fluid, and the pattern of the electrochemical noise contains infor-
mation on the nature of the corrosion processes that occur, including general corrosion, pitting
attack, and stress corrosion cracking.
KE Y WOR D S : electrochemical noise, supercritical aqueous system, corrosion, pitting attack,
stress corrosion cracking
The effect i ve cont r ol of cor r osi on and of mass t r anspor t and deposi t i on i n the heat t r anspor t
ci rcui t s of t her mal and nucl ear power pl ant s requi res t he cont i nuous moni t or i ng of key wat er
chemi st r y and syst em paramet ers, not abl y t he pH, concent r at i ons of oxygen and hydr ogen,
cor r osi on pot ent i al , conduct i vi t y, and cor r osi on activity. Nor mal l y, onl y a few of t hese pa-
r amet er s are moni t or ed, and meas ur ement s are gener al l y rest ri ct ed t o ambi ent t emper at ur e
and pressure. Consequent l y, onl y a part i al pi ct ur e of t he devel opment of damagi ng condi t i ons
i n an oper at i ng pl ant can be obt ai ned, becaus e t he r el evant par amet er s are not det er mi ned
under t he act ual oper at i ng condi t i ons. Thi s pr obl em has been r ecogni zed for ma ny years,
but t he l ack of sensor s for moni t or i ng key par amet er s at el evat ed t emper at ur es and pressures
has i mpeded t he i nt r oduct i on of i n si t u moni t or i ng syst ems.
1 Professor of materials science and engineering and director, Center for Advanced Materials, The
Pennsylvania State University, University Park, PA 16802.
2 Research associate, Center for Advanced Materials, The Pennsylvania State University, University
Park, PA 16802.
3 President, MPM Research and Consulting, PO Box 840, Lemont, PA 16851.
Copyrighl 9 1996 by ASTM lntcrnational
247
Www.astIII.OFg
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
248 ELECTROCHEMICAL NOISE MEASUREMENT
All general and localized corrosion processes, such as general corrosion, pitting attack,
crevice corrosion, and stress corrosion cracking, result in fluctuations in the free corrosion
potential and in the anodic and cathodic partial currents. These fluctuations are observed as
electrochemical noise that is thought to contain significant information about both the activity
and mechani sm of attack. Electrochemical noise measurement (ENM) has been used exten-
sively for monitoring corrosion activity in a variety of systems [1-17], although the sources
of the electrochemical noise are still not clearly defined [2 ]. Some believe that the electro-
chemical noise is due principally to the formation of hydrogen bubbles [3], at least in the
case of general corrosion, or the absorption of hydrogen by the first layers of metal beneath
the surface of passive film [4 ]. In the case of localized corrosion, such as pitting attack,
crevice corrosion [5-9], and stress corrosion cracki ng [10-12], the noise frequently arises
from discrete events that can be related to the development of damage. In some instances,
it has been suggested that ENM can distinguish between different corrosion mechanisms
[14]. However, in most cases, the experiments are preliminary, and viable theories for the
origin of the noise have yet to be developed [3].
A program has been established in this laboratory to develop in situ monitors for moni-
toring various physical, chemical, and electrochemical parameters in the heat transport cir-
cuits of thermal power plants. The monitors include reference electrodes, pH sensors, probes
for measuring the concentrations of oxygen and hydrogen in water, probes for measuring
electrochemical noise, and a fracture unit to detect the onset of stress corrosion cracking
(SCC) and to moni t or its propagation rate, all of which operate at elevated temperatures and
pressures [18,19]. In this paper, we describe the development of an electrochemical noise
analyzer (ENA) for monitoring corrosion activity at temperatures above the critical point of
water. The ENA monitors fluctuations in the galvanic coupling current between identical
electrodes, which, in this case, were fabricated from carbon steel. In parallel, we have also
explored ENM as a means of detecting and monitoring stress corrosion cracking in sensitized
Type 304 stainless steel (SS) (UNS $30400) in water at temperatures up to 288~ using an
experimental setup that isolates the external surfaces of a precracked compact tension (CT)
specimen from the environment, but couples the specimen to cathodes located on the spec-
imen sides.
Experimental Apparatus
The ENA developed in this study is shown schematically in Fig. 1. The analyzer operates
on an analog basis for simplicity and to maintain a very low cost. The circuit consists of a
] H ea ter [
FIG. 1--Schematic of the electrochemical noise analyzer (ZRA = zero resistance ammeter, RM = root
mean square).
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
MACDONALD ET AL. ON CARBON AND STAINLESS STEELS 249
zero resistance ammeter (ZRA), the output of whi ch passes through a band-pass filter with
independently adjustable upper and lower roll-off frequencies. The lower roll-off frequency
can be set at 1 Hz, 10 Hz, 100 Hz, and 1 KHz, whereas the upper roll-off frequency can be
set at 10 Hz, 100 Hz, 1 KHz, and 10 KHz. Because there is virtually no electrochemical
noise at very high frequencies, we have selected 10 KHz as the high limit of upper-roll-off
frequency. Therefore, the bandwidth of the signals that pass this filter can be selected between
10 Hz and 10 KHz. Amplification of the signals can be regulated through a gain from 0.1
to 2000. In addition, the sensitivity of the analyzer can be regulated by adjusting the response
time over the range from 250 ms to 10 s. All of these regulation functions are accessible
from the front panel of the analyzer. After amplification by the ZRA (AMP-05) and filtering
by the band pass filter, the signal is processed by a root -mean square module (RMS, AD636)
and is then presented on a digital display and on a strip-chart recorder. The electronic circuits
are carefully shielded to exclude electronic noise from the environment and from sources
other than the probes. The probes consisted of two identical SAE- AI SI 1013 (UNS GI 0130)
carbon steel rods (+ = 1.3 mm), with 12 mm being exposed to the environment through a
high-pressure fitting in the autoclave. The probe assembly was located in a pressure vessel
that was part of a once-through, recirculating, high-temperature/high-pressure loop capable
of operating at temperatures to 550~ and at pressures to 310 bar ( - 4 5 0 0 psi or 31.0 MPa).
A schematic of the loop is presented in Fig. 2. The operating parameters of the loop are
such that the density of water in the supercritical region at 550~ could be varied from zero
to less than 0.1 g / c m 3 [20].
For stress corrosion cracking studies on sensitized Type 304 SS in high-temperature water,
a precracked compact tension (CT) specimen was loaded in a high-temperature, high-pressure
cell and was isolated (except for the crack) from the environment using baked-on polytet-
rafluoroethylene (PTFE). The specimen was coupled to side-mounted cathodes via a ZRA
(Fig. 3). Because the cathodic reaction occurs on the external surface (in this case, the
cathode) whereas the anodic reaction is restricted to the crack, a current is detected by the
ZRA. It is the electrochemical noise in this coupling current that is of prime interest in this
work [21, 22].
Resul t s and Di s cus s i on
Electrochemical noise measurements on corrodi ng carbon steel were carried out at high
subcritical (T < 374.12~ and at supercritical (T > 374.12~ aqueous environment under
N2
inlet
Valve
to drain
_
r~e~s~e l ~ ~ . ~ l --H-ZZ-I
reducing ~ ~ u 2 Pm~
cheek valve[I ...... II II - I
Test Cell
i
Heater
_ / Hi P fitting
[ I ~ Carbon steel probes
' ~ x (to ENA)
Modified
Cooling Conax
coil fitting
FIG. 2--Experimental loop (HPLC = high pressure liquid chromatography, HiP = High Pressure
Equipment Co.).
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
2 5 0 ELECTROCHEMICAL NOISE MEASUREMENT
Cathode over
specimen side
surface
J,
To ZRA
T
Machi ned Notch
Beneath Cat hode
I
Side Back
FIG. 3--Schematic of CT specimen: (1) Specimen, (2) Clevis, (3) Cathode, and (4) PTFE washer.
conditions of temperature, pressure, and oxygen concentration that simulate those in the heat
transport circuits of supercritical power plants. However, because of the limited capacity of
the high pressure liquid chromat ography (HPLC) pump used to pressurize the system, it was
not possible to simulate the flow conditions fully. Nevertheless, our studies demonstrate that
ENM is a powerful means of monitoring corrosion activity, as will become evident from the
material presented below.
In the present design of the monitoring system, we used an RMS module for signal
conditioning. The purpose of this module was to convert any sinusoidal component in the
electrochemical noise into a DC output that could be easily displayed on a digital meter.
However, the output was found to be "noi sy, " thereby indicating that the electrochemical
noise generated on the carbon steel is stochastic in nature and is not simply sinusoidal (Fig.
4). Other possible sources that may be responsible for the measured noise have been dis-
cussed, and it was demonstrated that the measured noise is indeed from the corrodi ng probes
in high-temperature aqueous environment [20]. As shown below, the electrochemical noise
consists of principally of low-frequency component s (1 to l 0 Hz), as determined by the
bandpass filter.
An important finding of this study is that electrochemical noise is generated on carbon
steel corrodi ng in low density (p < 0.1 g / c m 3) supercritical water, and that the noise can be
used to moni t or corrosion activity. To our knowledge, the present work represents the first
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
MACDONALD ET AL. ON CARBON AND STAINLESS STEELS 251
0
e-
r
10'
J ~ a n n i ~ I 1 6 o " C I C o o t i n a
I
I
1 "
E
0 6 0 0 0 1 2 1 ~ 1 8 0 0 0
" l ' b n e i s )
FIG. 4- - Evol ut i on of electrochemical current noise induced by corrosion o f A/ S / 1 0 1 3 carbon steel
during an experimental heat i ng-cool i ng cycle (oxygen pressure = 40 psi (0.28 MPa), in-vessel
pressure = 1200 psi (8.28 MPa), gain = 1000, l ower roll-off f requency = 1 Hz, upper roll-off
f requency = 10 Hz).
attempt to perform ENM in supercritical aqueous systems and, hence, to demonstrate the
potential application of this technique in supercritical power plant technology.
Analysis of the electrochemical current noise from the compact tension test indicates that
the electrochemical current noise consists of domi nant low-frequency components of - l to
2 Hz, depending on the load. This finding is consistent with crack growth being due to
discrete events (jumps), perhaps resulting from an embrittlement phenomenon at the crack
tip. Details of this work are given below.
The Ef f e c t o f Temper at ur e
The general response of the electrochemical noise as a function of temperature in sub-
critical water (liquid phase) is shown in Fig. 4. The amplitude of the electrochemical noise
is observed to increase with increasing temperature. This is an expected response, because
elevation in temperature increases the activity of the corrosion process. The observed elec-
trochemical noise is found to be reversible with respect to temperature, in that lowering
temperature decreases the amplitude of the noise in a predictable and reversible manner.
In the supercritical region, however, the amplitude of the electrochemical noise versus
temperature is found to be complicated (Fig. 5). It is observed that the amplitude of the
electrochemical noise initially increases with increasing temperature to a temperature of
- 434~ but then decreases as the temperature increases to 550~ The origin of the maxi-
mum in the electrochemical noise is most likely due to the compet i ng effects of temperature
and water density on the corrosion processes. Thus, we speculate that, for T < 434~ the
change in the noise amplitude is due principally to the effect of increasing temperature on
the kinetics of the corrosion process, but at higher temperatures, decreasing water density
exerts a dominant effect. Not e that in this figure and subsequent figures the error bars rep-
resent the range of fluctuation in the noise signal as observed over a 1-min period for the
indicated gain and for the stated rol l -off frequencies.
The Ef f e c t o f Pr e s s ur e
Pressure is found to have strong impact on the amplitude of the electrochemical noise
from corroding carbon steel measured in supercritical water (Fig. 6). Our experiments have
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
252 ELECTROCHEMICAL NOISE MEASUREMENT
2 . 0 1. 00
. . . . . . . . . I . . . . . . . . . I . . . . . . . . . I . . . . . . . . . I . . . . . . . . .
I [Oe] = 0 . 0 0 6 ppm
g"
E / P = 3 5 5 0 p s i
T [ Flow r a t e 172.5 ml / mi n
v 1.5 ~ 0. 75
gmt . ude
,-. 1 . 0 0 . 5 0 o
t.O
r t'*3
0 II
2 ,',
~ '
" 0. 5 - 0. 25 "N
. , . o e - *
0 . 0 . . . . . . . . i . . . . . . . . . ; . . . . . . . . . I . . . . . . . . . r . . . . . . . . . O.00
3 . 5 0 4( 30 4 5 0 5 0 0 5 5 0 6 0 0
T e mp e r a t u r e (~
FIG. 5--Dependence of electrochemical current noise amplitude for AISI 1013 carbon steel in su-
percritical water as a function of temperature (data compiled from three heating-cooling cycles), and
water density (water density data from Ref 15 ).
shown that t he ampl i t ude of the el ect r ochemi cal noi se i ncreases sharpl y wi t h i ncreasi ng
pressure of t he supercri t i cal fluid to 241 bar (3500 psi or 24.1 MPa), cor r espondi ng to a
densi t y i ncrease from 0.062 to 0.091 g / c m 3. The i ncrease in the noi se ampl i t ude wi t h in-
creasi ng densi t y is pr esumabl y due to ei t her an i ncrease in conduct i vi t y of the fluid bet ween
t he carbon steel probes or to an i ncrease in the effect i ve concent rat i on of oxygen at the
met al surface, or bot h (see Fi g. 6).
The Effect of Stirring and Flow Rate
The sol ut i on in t he aut ocl ave coul d be agi t at ed t hrough a st i rrer act i vat ed by magnet i c
drive. Exper i ment s carri ed out at subcri t i cal t emperat ures demonst r at e that st i rri ng has a
strong qual i t at i ve influence on the ampl i t ude of the el ect rochemi cal noi se (Fi g. 7). Thus,
activation of the stirrer, in this part i cul ar case, results in a great er than fivefold i ncrease in
the noi se ampl i t ude, pr esumabl y a result of enhanced mass transport. By varyi ng the pump
speed, it was al so possi bl e to regul at e the flow rate. The i nfl uence of mass flow rat e t hrough
the test cel l i nner channel di amet er 3.18 mm on the el ect rochemi cal noi se has been st udi ed
over the range of 0 to 20 mL/ mi n, and the ampl i t ude of the noi se was found to i ncrease
modest l y as t he flow rate was i ncreased over the accessi bl e range (Fi g. 8). Part of the
observed i ncrease in the noi se ampl i t ude may be at t ri but ed to a modi fi cat i on of the hydro-
dynami cs of t he syst em, but it is al so l i kel y that t he hi gh flow rat es count eract the con-
sumpt i on of oxygen by corrosi on of the l oop wal l s (especi al l y in the hot zone), so that a
great er effect i ve oxygen concent rat i on is present in the vi ci ni t y of the pr obes at hi gher
flow rates. The dat a shown in Fi gs. 7 and 8 cl earl y demonst r at e the i mport ance of
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
MACDONALD ET AL ON CARBON AND STAINLESS STEELS 253
1.5
De n s i t y ( g / c m 3)
0. 061 0 . 0 7 4 0 . 0 9 1
. . . . . . . . . I . . . . . . . . . 1 . . . . . . . . . 1 . . . . . . . . .
~" [ 0o] = 0 . 0 0 6 p p m
E
- - - T = 4 8 1 o c
Fl ow r a t . e 12. 5 r n l / mi n
-~ 1. 0
1 1 ,
.1~ / /
( , 1
0 . 5
. =
0
0 9
e ~
__.n . . . . . . . . . ' . . . . . . . . . r . . . . . . . . . I . . . . . . . . .
2000 2500 5000 3500 4000
P r e s s u r e o f t h e s y s t e m ( p s i )
FIG. 6--Electrochemical current noise amplitude for AISI 1013 carbon steel in supercritical water
as a function of pressure (water density data from Ref 15).
hydr odynami c/ mas s t ransport effects in det er mi ni ng the ampl i t ude of the el ect r ochemi cal
current noi se i nduced by the corrosi on of carbon steel in the subcri t i cal and supercri t i cal
water.
The Effect of Oxygen Concentration
The effect of oxygen concent rat i on was st udi ed under bot h supercri t i cal and subcri t i cal
condi t i ons. The sol ut i on was pr epar ed to be sat urat ed wi t h pure oxygen (wi t h an oxygen
concent rat i on in t he sol ut i on [02] = - 3 2 ppm). The reservoi r was then sparged wi t h ni t rogen
to deaerat e t he wat er (see Fi g. 2). The pressure in t he sol ut i on reservoi r was mai nt ai ned
sl i ght l y hi gher than one at mosphere to prevent ai r f r om ent eri ng i nt o t he l ow- pr essur e sect i on
of the syst em. By varyi ng the ni t rogen bubbl i ng rate, the oxygen cont ent in t he sol ut i on
coul d be decr eased at a desi r ed rate. A t ypi cal deoxygenat i on curve is shown in Fi g. 9. The
oxygen concent rat i on was measured at ambi ent t emperat ure usi ng a commer ci al oxygen
probe ( I ngol d Model 170) at the out l et of an aut ocl ave aft er cool i ng the process stream.
Because a sl ow deaerat i on rate was chosen, and because of t he smal l i nner channel di amet er
and vol ume of the t est cel l (4) = 3.18 mm, effect i ve heat ed l engt h L = 101.6 mm), t he t i me
for the sol ut i on to flow from t he aut ocl ave to the oxygen probe was an or der of magni t ude
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
2 5 4 ELECTROCHEMICAL NOISE MEASUREMENT
>
E
0
._.9,
0
10 i
No st m' m|
" r l mr
F I G. 7--E f f ect of sti rri ng on the el ectrochemi cal current noi se i nduced by corrosi on of A1Sl 1013
carbon steel in water at 244~ (oxygen pressure = 40 psi (0.28 MPa), i n-vessel pressure = 1200 psi
(8.28 MPa), gai n = 1000, l ower rol l -off frequency = 1 Hz, upper rol l -off frequency = 10 Hz).
s ma l l e r t han t he t i me r e qui r e d f or a n o t i c e a b l e c h a n g e i n t he o x y g e n c o n c e n t r a t i o n ( s e e Fi g.
9 f or a t y p i c a l de a e r a t i ng p r o c e s s ) . Th u s , w e c o n c l u d e t hat t he o x y g e n c o n c e n t r a t i o n me a -
s ur e d b y t he o x y g e n pr o be r e pr e s e nt s t he o x y g e n c o n t e n t i n t he s o l u t i o n wi t h i n t he t e s t c e l l ,
t o a r e a s o n a b l e a p p r o x i ma t i o n .
2. 0 . . . . , . . . . . . . . . . . . . . . . . . , . . . .
E
0
t.-
o
E
@
c-
O
0
-d
e ~
1.5
].0
T = 4 7 0 " C
P = 3 5 0 0 p s i
[ 0 2 ] = 0 . 4 p p m
t J - l J i
0. 5 , , . , I . . . . . . , , , , I , , , , , , , , , f ' ' ' '
4 5 G
Fl ow r a c e ( m l / m i n )
FIG. 8--Electrochemical current noise from AI SI 1013 carbon steel in supercritical water as a f unc-
tion of flow rate (lower roll-off frequency = 1 Hz, upper roll-off frequency = 10 Hz).
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
MACDONALD ET AL, ON CARBON AND STAINLESS STEELS 255
E
e~
3
O
O
= 2
t _
r "
o
t -
O 1
o
x
0
0
T=410~ ml / mi n
I P = 3 5 0 0 pm
LFlow r a t e 8. 4
.~ A O
U 2UU 4 0 0 ~UU 8(.)(.-) 100U
Ti me (rai n),
FI G. 9- - Typi cal deoxygenation curve as determined by sparging the reservoir tank with nitrogen
([02 ] measured at the outlet of the autoclave at ambient temperature).
o
e -
" ~ . 0 1
. 00,
. 0 0 0 1 9 - ~ . . . . . , . . . . . . . . J . . . . . . . ~ . . . . . . . ", . . . . . .
. 0 0 1 , 0 1 . 1 1 1 0 1 0 0
Oxygen concentr=ion (ppm)
FIG. l O- - Evol ut i on o f electrochemical current noise induced by corrosion of AI Sl 1013 carbon steel
as a f unct i on o f oxygen concentration in the solution (T = 250~ in-vessel pressure = 1400 psi (9.66
MPa), gain = I, lower roll-off f requency = I Hz, upper roll-off f requency = 10 Hz).
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
256 ELECTROCHEMICAL NOISE MEASUREMENT
In subcritical systems, the magnitude of the electrochemical noise is found to decrease
monotonically with decreasing oxygen concentration in the solution (Fig. 10). Although the
data are of only moderate precision, particularly at high oxygen concentrations, the noise
amplitude is found to be represented by
RMS = 0.01 [02] 0.5 (1)
where RMS is the root-mean square of the noise amplitude, in units of millivolts and [02]
is the oxygen concentration in ppm. The square root dependence of RMS on [02] appears
to indicate that the amplitude of the noise is not controlled by the diffusion of oxygen to
the surface, because, in that case, a linear dependence might be expected. However, the
stirring and flow-rate experiments discussed above clearly demonstrate a mass transport-
controlled component in the observed electrochemical noise. At the present time and until
the mechanism of noise generation is better understood, the observed relationship between
the electrochemical noise amplitude and the oxygen concentration should be considered
purely empirical in nature.
In supercritical systems (Fig. 11), the amplitude of the noise was again found to decrease
monotonically with decreasing oxygen concentration for [02] > 0.1 ppm, with approximately
the same exponent (0.5) as observed at lower temperatures. However, for [02] < 0.1 ppm,
the amplitude was found to be independent of [O2], although the magnitude of the fluctua-
tions in the noise (as expressed by the error bars) renders this observation only semiquan-
titative in nature. We note that in this experiment the density of the water is less than 0.1
g/cm 3, compared with -0. 75 g/cm 3 for the subcritical system shown in Fig. 10. Thus, the
data shown in Fig. 11 may well indicate that the amplitude of the noise at low oxygen level
o
' ' . . . . . . . . '
[hieh b a n d pass I K H z I
/ T = 4 1 0 * C / T
I P = a s o o I T , I ,
i i i . . . . . I
0.01
FIG. 11--Electrochemical current noise amplitude f or AISI 1013 carbon steel in supercritical water
as a function of oxygen concentration.
0 . 0 1 , . . . . . . a . . . . . . . . r . . . . . . . . i . . . . . .
0 . 0 0 1 0 . 1 I 1 0 1 0 0
[02] in the solution (ppm)
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
MACDONALD ET AL. ON CARBON AND STAINLESS STEELS 257
is controlled by the density of the fluid (and presumably, by the resistance of the fluid
between the probes), and that the effect of oxygen become apparent only for [02] > 0.1
ppm. This interpretation is consistent with a model that supposes that an electrical analog
for the probes used in this study consists of a series of resistance (Rs), representing the
resistance between the probes that is inversely related to the density of the fluid, and two
(identical) interfacial independences that describe the corrosion processes that occur at the
surface of the two probes. For [O2] < 0.1 ppm, the impedance of the cell is dominated by
R s 9
The Fr equency Composi t i on o f the El ect r ochemi cal Noi se
It has been demonstrated in previous studies [18, 19] that at high oxygen concentrations
and at subcritical temperatures the electrochemical noise generated by the corrosion of carbon
steel is composed mainly of low-frequency components (1 to 10 Hz). A similar distribution
in frequency of the components of the electrochemical noise from corroding carbon steel is
observed at supercritical temperatures, as shown in Fig. 12. The spectra show that the noise
is composed principally of low-frequency components (1 to 10 Hz). Although no literature
data are available for the corrosion of carbon steel in supercritical water, data from the
literature for systems at ambient temperature show that the electrochemical noise consists
principally of low-frequency components [2,14,17].
Observat i on o f Pi t t i ng Corrosi on
During the experiments carried out in this work, certain patterns in the electrochemical
noise could be associated with specific experimental conditions. Figure 13 shows two dif-
g
2
o
-g
0. 15
0. 10
0.05
,=.
I T=4IO~ P=3500 psi
[02]= 7 ppb
Flow r at e 8.4 ml / mi n
0.00
1-10 I0- lO0 IO0-1K IK-IO K
Fr equency r ange (Hz)
FIG. 12--The distribution in frequency of the electrochemical noise from AISI 1013 carbon steel in
supercritical water (lower roll-off frequency = 1 ttz, upper roll-off frequency = 10 KHz, gain = 1 ).
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
258 ELECTROCHEMICAL NOISE MEASUREMENT
D
E
2
50
4O
30
2O
lO
_,/
0
P-- 1 ~0 )si
:!
i
1
I
a. ,.# ,~.
1 2
P=2750 psi
5 1 2 3 4 5
Tune (min)
50
40
30
20
I0
0
FIG. 13--The electrochemical current noise from AISI 1013 carbon steel as a function of time in
supercritical water (T = 405~ [02] = ppb, flow rate = 4.6 mL/min, gain = 20, lower roll-off
frequency = 1 Hz, upper roll off frequency = 10 KHz, Pressure = 2150 and 2750 psi (14.84 and 18.98
MPa ), as indicated).
ferent patterns in the magnitude of the electrochemical noise that were observed under two
different set of conditions. Shown on the right-hand side of Fig. 13 is a continuous, steady
level of the electrochemical noise, whereas on the left-hand side, the noise pattern consists
of a series of spikes. The only difference between the two cases is the pressure in the test
cell. In searching for a relationship between the noise pattern and the mechanism of corro-
sion, the surface morphology of the steel specimens before and after exposing to the high-
temperature water was examined. (Scanning electron microscopy (SEM) micrographs shown
in Figs. 14 and 15 clearly demonstrate the occurrence of pitting corrosion during these
FIG. 14--Surface morphology of the AISI 1013 carbon steel specimen before the experiment.
9 Copyright 1994 by NACE International. All Rights Reserved by NACE; reprinted with permission.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
MACDONALD ET AL. ON CARBON AND STAINLESS STEELS 259
FIG. 15--Surface morphology of the AISI 1013 carbon steel specimen after the experiment. 9 Copy-
right 1994 by NACE International. All Rights Reserved by NACE; reprinted with permission.
experiments. Although the electrochemical noise pattern shown on the left-hand side in Fig.
13 cannot be directly linked to the occurrence of pitting corrosion because SEM examination
was performed at ambient temperature after disassembling the system, the experiments
clearly associate the occurrence of discrete events in the noise pattern with localized cor-
rosion, as has been found for ambient temperature systems [8].
Stress Corrosion Cracking
As noted in the Experimental Apparatus section, in these studies we monitored the cou-
pling current that flows from the crack to the external cathodes where it is consumed by
oxygen reduction. Thus, the observed current and the noise contained therein are expected
to reflect the properties of the source (the crack) as well as the properties of the sink (the
external cathodes). We have previously shown [18,19] that corrosion-resistant alloys, such
as stainless steel (and presumably nickel and nickel alloys), exhibit low-amplitude noise in
oxygenated pure water that is random in character. Therefore, any structure in the coupling
current noise is presumed to arise from processes occurring within the crack. A typical
coupling current versus time trace on loading and subsequently unloading the specimen is
shown in Fig. 16. Thus, on increasing the load to a stress intensity (K0 of 27.5 MPa.m '/2,
the ZRA current initially increases to a plateau at - 50 I~A, corresponding to the growth of
the crack through the highly damaged plastic zone ahead of the crack (as a result of fatigue
precracking). Subsequently, the ZRA current increases to - 150 to 200 I~A as the crack
propagates through the virgin steel. On unloading, the current immediately decreases to the
previously observed unloaded value of - - 50 I~A (this is not zero because the nickel cath-
odes were lightly platinized, so that a galvanic couple exists in the system; Fig. 17).
The data acquisition rate used for recording the trace shown in Fig. 16 was low (0.07 Hz)
so that the noise observed on the trace does not accurately reflect the actual noise structure.
On increasing the acquisition rate to 100 Hz (in this case, for a specimen with Type 304 SS
cathodes), a remarkable pattern is discerned (see Fig. 17). Thus, the noise consists of a
dominant components of ~2 Hz with periodic bursts of higher frequency noise at about
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
260 ELECTROCHEMICAL NOISE MEASUREMENT
Water Temperature = 252 Degrees C
Pressure = 82.7 bar
7 0 0 2 5 0
6 0 0 K~ = 27.5 ~a ~m 2 0 0
( , . )
<E
5 0 0 1 5 0 : 1 .
C
4oo lOO
300 5 0
2 0 0 ~ 0
100 ~ - 5 0
0 ~ o t u t n p - 1 0 0
0 . 0 0 0 . 5 0 1 . 0 0 1 . 5 0 2 . 0 0 2 . 5 0 3 . 0 0 3 . 5 0 4 . 0 0
Time (hours)
FIG. 16- - Load and ZRA coupling current versus time f or the initial l oadi ng/ unl oadi ng of a pre-
cracked specimen (platinized nickel cathode, test conducted at a temperature o f 252~ pressure 1200
psi [82. 7 MPa ]).
every 5 s. Because of t he ext r aor di nar y nature of this noi se pattern, we went to consi der abl e
l engt hs to ensure that the noi se was real. Thus, the noi se was no l onger obser ved when the
speci men was unl oaded to K l < 11 MPA-m 1/2, whi ch apparent l y rul es out i nduct i on from
ext ernal sources. Al so, the noi se (at K~ > 11 MPA. m 1~) persi st ed when vari ous el ect ri cal
equi pment was swi t ched off, i ncl udi ng the servohydraul i c l oad frame (not e t hat t he noi se in
the l oad is due to the acqui si t i on syst em and does not si gni fy fluctuations in t he act ual load).
Fi nal l y, t he frequency of t he noi se in t he coupl i ng current (i ncl udi ng bot h the backgr ound
noi se and t he frequency of the bursts) was found to i ncrease wi t h i ncr easi ng/ ( 1 [22], whi ch
cl ear l y shows that the noi se ori gi nat es from the crack.
The effect of l oadi ng to successi vel y hi gher stress i nt ensi t i es is shown in Fi g. 18. Loadi ng
to K[ = 11 MPa. m ~/2 di d not cause a response in t he ZRA current, but l oadi ng to 22 MPa-m 1/z
did, t hereby i ndi cat i ng that the val ue of the cri t i cal stress i nt ensi t y for stress corrosi on crack-
i ng (K~scc) is bet ween 10 and 22 MPa. m 1/2. Furt hermore, stress i nt ensi t i es above 33 MPa. m 1~2
di d not cause a si gni fi cant i ncrease in t he ZRA current, whi ch is consi st ent wi t h the current
ari si ng from crack growt h in the St age II regi on for this al l oy/ envi r onment combi nat i on
[23]. These observat i ons further support our cont ent i on that t he ZRA coupl i ng current (in-
cl udi ng the noi se) arises di rect l y from crack advance, and hence, that the structure of the
noi se pr ovi des val uabl e i nformat i on on the mi cr oscopi c processes that l ead t o cr ack pr opa-
gation. Fi nal l y, the magni t ude of the coupl i ng current is much great er than that expect ed
t heoret i cal l y, i f it is assumed that al l of the current ori gi nat es from the crack tip. However,
as we show el sewhere [22], the current can be account ed for i f it is assumed t hat crack
advance occurs by mi cr oscopi c brittle fracture events, wi t h the current ori gi nat i ng from the
crack t i p and the near-t i p crack flanks. In this model , the mean current is apparent l y domi -
nat ed by the current from the cr ack flanks, whereas the current noi se reflects, pri nci pal l y,
processes that occur at the crack front,
As not ed above, a model that is consi st ent wi t h the noi se dat a for crack pr opagat i on is
one that proposes that cr ack advance occurs by di scret e events, the frequency of whi ch
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
MACDONALD ET AL. ON CARBON AND STAINLESS STEELS 261
( V ~ i ) ; u ~ u n o v ~ z
8
m
0
. . J
0 0 0 0 0 C ) 0 ~ ' ~ 0 0 P ~ 0
( : 0 0 0 ~ 0 0 C : ) 0 0
3
I
r3.~
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
r
~

"
o

o

.
J

1
2
0
0

1
0
0
0

8
0
0

6
0
0

W
a
t
e
r

T
e
m
p
e
r
a
t
u
r
e

=

2
8
0

D
e
g
r
e
e
s

C

P
r
e
s
s
u
r
e

=

8
2
.
7

b
a
r

4
0
0

2
0
0

t
o
a
d


"
~

K
=

=

2
2

M
P
a
~
'
m

9

K
I
=

I
I

M
P
a
~
m

I

I

I

~
,

=

3
3

M
P
a
~
'
.

K
I

=

4
4

M
P
a
~
'
m

7
0
O

6
O
O

5
o
0

'
<

:
:
L

4
0
0

'
~

3
0
0

d

2
0
0

N

1
0
0

0

I

~

-
1
0
0

0
.
0
0

0
.
5
0

1
.
0
0

1
.
5
0

2
.
0
0

2
.
5
0

3
.
0
0

3
.
5
0

T
i
m
e

(
h
o
u
r
s
)

F
I
G
.

1
8
-
-
L
o
a
d

a
n
d

Z
R
A

c
u
r
r
e
n
t

v
e
r
s
u
s

t
i
m
e

f
o
r

m
u
l
t
i
p
l
e

l
o
a
d
i
n
g
~
u
n
l
o
a
d
i
n
g

c
y
c
l
e
s

t
o

v
a
r
i
o
u
s

K
/

l
e
v
e
l
s

(
p
l
a
t
i
n
i
z
e
d

n
i
c
k
e
l

c
a
t
h
o
d
e
)
.

m

r
-

m

c
)

O

c
)

n
-

m

E

>

r
-

Z

6
O

m

E

C

m

E

m

z

Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
MACDONALD ET AL. ON CARBON AND STAINLESS STEELS 263
depends on the stress intensity. By assuming that each spike in the noise corresponds to a
single semicircular event of dimension, g, we can write the crack growth rate as
da _ 2g 2
dt L f (2)
where
a = the crack length,
L = the specimen width, and
f = the event frequency (1 - 2 Hz).
From the measured crack growth rate and the specimen dimension [22], it was found that
the average dimension is of the order of 3 ~m, which is considerably smaller than the grain
size of the steel (30 to 50 ixm) [21, 22].
We can test the internal consistency of this model for crack advance by calculating the
current density at the interfaces of the discrete events from the amplitude of the noise and
the dimensions of the events. Thus, taking the previous estimate of g = 3 I~m, and assuming
that the event is semicircular, we calculate the total area exposed per event as 2.8 1 0 - 7
cm ~. Now, the amplitude of the noise (see Fig. 17), on average, is about 15 IxA, which yields
an instantaneous current density at the moment of fracture of - 5 3 ~LA/cm 2. This value is
consistent with the results from fast-straining experiments on austenic stainless steel [24],
allowing for the fact that these alloys do not accurately simulate the sensitized grain bound-
aries of the materials used in this study and recogni zi ng the approximate nature of the
calculation.
To our knowledge, the experiment described above is unique, and it provides a powerful
illustration of how ENM may be used to monitor the rate of crack growth and explore the
mechani sm by which growth occurs. The method is particularly convenient for use in high-
temperature (supercritical and subcritical) systems, because of its great precision and sensi-
tivity (afforded by our ability to measure very small currents) and its simplicity. Al t hough
experimental challenges do exist (for example, in coating the specimen with plasma-sprayed
ZrO2 rather than with PTFE), the technique has already been proven at 288~ and we see
little impediment to extending the temperature to T > T c.
Summary and Conclusions
An electrochemical noise analyzer has been developed that is capable of detecting and
measuring the electrochemical noise induced in the coupling current between two identical
metal or alloy probes when exposed to conditions that simulate the heat transport circuits of
supercritical fossil-fuelled power plants. This study shows that the electrochemical noise
induced on carbon steel in high subcritical and supercritical water depends on temperature,
pressure, flow rate, and oxygen concentration in an understandable manner. These experi-
ments demonstrate thai: the electrochemical noise analysis is a sensitive technique for mon-
itoring corrosion activity in high subcritical and supercritical aqueous systems.
The compact tension testing experiment on sensitized Type 304 SS in high-temperature
water demonstrates that noise in the coupling current between the crack and the external
cathodes reveals important information on the mechani sm of crack propagation. The exper-
imental results are co~asistent with crack growth being due t o a discrete event cleavage
mechanism, in which tile events are possibly induced by embrittlement of the grain boundary
matrix.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
264 ELECTROCHEMICAL NOISE MEASUREMENT
Acknowledgment
The aut hors gr at ef ul l y acknowl edge t he suppor t of t hi s wor k by UNI ON FENOSA, Ma-
dri d, Spai n, and by t he Ni agar a Mohawk Power Cor por at i on, Syr acuse, New York.
References
[1 ] Mills, D. J . , Bierwagen, G. B., Skerry, B., and Tallman, D., "Characterization of Corrosion Under
Coatings by Electrochemical Noise Methods," in Proceedings of the 12th International Corrosion
Congress, Vol. 1, NACE, Houston, 1994, p. 182.
[2 ] Gabrielli, C., Huet, F., and Keddam, M., "Introduction of Metallic Corrosion by Electrochemical
Noise Techniques," in Electrochemical and Optical Techniques for the Study and Monitoring of
Metallic Corrosion, M. G. S. Ferreira and C. A. Melendres, Eds., Kluwer Academic Publishers,
The Netherlands, 1991, p. 135.
[3] Gabrielli, C., Huet, E, Keddam, M., and Macias, M., "Electrochemical Noise Generated by a
Corroding Electrode, Effect of Metallurgy and Inhibitors," in Surface, Inhibition, and Passivation,
E. McCafferty and R. J. Brodd, Eds., Proceedings, Vol. 86-7, The Electrochemical Society, Inc.,
Pennington, NJ, 1986, p. 507.
[4] Benzaid, E., Huet, F., Gabrielli, C., Jerome, M., Wenger, E, and Galland, J., "Investigation of
Electrochemical Noise in the Study of Hydrogen Embrittlement of a 42CD4 Carbon Steel Elec-
trode," in Progress in the Understanding and Prevention of Corrosion (Proceedings, lOth European
Corrosion Congress) J. M. Costa and A. D. Mercer, Eds., The Institute of Materials, London, 1993,
Vol. 2, p. 1304.
[5] Hladky, K. and Dawson, J. L., Corrosion Science, Vol. 21, No. 4, 1981, p. 317.
[6] Dawson, J. L., Farell, D. M., Aylott, P. J., and Hladky, K., "Corrosion Monitoring Using Electro-
chemical Noise Measurements," Paper 31, NACE Corrosion~89, April 1989, New Orleans, LA,
NACE, Houston, TX.
[7] Kending, M. and Anderson, D., Corrosion, Vol. 48, No. 3, 1992, p. 178.
[8] Bertocci, U., Mullen, J. L., and Ye, Y. X., "Electrochemical Noise Measurements for the Study of
Localized Corrosion and Passivity Breakdown," in Passivity of Metal and Semiconductors, M.
Froment, Ed., Elsevier Science Publishers B. V., New York, 1983, p. 229.
[9] Flis, J., Dawson, J. L., Gill, J., and Wood, G. C., Corrosion Science, Vol. 32, No. 8, 1991, p. 877.
[10] Eden, D. A. and Rothwell, A. N., "Electrochemical Noise Data: Analysis, Interpretation and Pres-
entation," Paper 292, Corrosion 92, the NACE Annual Conference and Corrosion Show, May 1992,
Nashville, TN, NACE, Houston, TX.
[11 ] Eden, D. A., Rothwell, A. N., and Dawson, J. L., "Electrochemical Noise for Detection of Sus-
ceptibility to Stress Corrosion Cracking," Corrosion 91, The NACE Annual Conference and Cor-
rosion Show, Paper 444, March 1992, Cincinnati, OH, NACE, Houston, TX.
[12] Stewart, J., Wells, D. B., Scott, P. M., and Williams, D. E., Corrosion Science, Vol. 33, 1991,
p. 73.
[13] Monticelli, C., Brunoro, G., Frignani, A., and Trabanelli, G., Journal of the Electrochemical Society,
Vol. 139, No. 3, 1992, p. 706.
[14] Mansfeld, E and Xiao, H., "Electrochemical Noise Analysis of Iron Exposed to NaCI Solutions
of Different Corrosivity," in Proceedings, 12th International Corrosion Congress, Vol. 3A, NACE,
Houston, TX, p. 1388.
[15] Bain, R. W., Steam Tables, Her Majesty' s Stationary Office, Edinburgh, UK, 1964.
[16] Roberge, P. R., "Analysis of Spontaneous Electrochemical Noise with Stochastic Pattern Detector
Method," in Progress in the Understanding and Prevention of Corrosion (Proceedings, lOth Eur-
opean Corrosion Congress, Vol. 2), J. M. Costa and A. D. Mercer, Eds., The Institute of Materials,
London, 1993, p. 1345.
[17] Dowling, N. J. E., Thual-Duret, C., Auclair, G., Audouard, J. P., and Combrade, P., "Assessment
of the Pitting Susceptibility of AISI 303 and 304 Stainless Steels as a Function of Sulphur Content
Using Electrochemical Noise," in Progress in the Understanding and Prevention of Corrosion
(Proceedings, lOth European Corrosion Congress, Vol. 2), J. M. Costa and A. D. Mercer, Eds.,
The Institute of Materials, London, 1993, p. 1328.
[18] Macdonald, D. D., Pang, J., Liu, C., Medina, E., Villa, J., and Bueno, J., "Development of Sensors
for In-Situ Monitoring of Corrosion and Water Chemistry Parameters For Electric Power Utility
Industry," in Proceedings, 12th International Corrosion Congress, Vol. 6, NACE, Houston, TX,
p. 4727.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
MACDONALD ET AL. ON CARBON AND STAINLESS STEELS 265
[19] Macdonald, D. D., Pang, J., Liu, C., Kriksunov, L. B., Medina, E., Villa, J., and Bueno, J., "Mon-
itoring Corrosion and Chemistry Phenomena in Supercritical Aqueous Systems," Paper 627, Cor-
rosion 94, The NACE Annual Conference and Corrosion Show, Feb. 1994, Baltimore, MD, NACE
International, Houston, TX.
[20] Liu, C., Macdonald, D. D., Medina, E., Villa, J., and Bueno, J., Corrosion, Vol. 50, No. 9, 1994,
p. 687.
[21] Manahah, M P., Sr., Newman, K. E., Macdonald, D. D., and Peterson, A. J., Jr., "Experimental
Validation of the Basis for the Coupled Environment Fracture Model," paper presented at EPRI
Workshop on Secondary Side-Initiated IGA/ SCC, Minneapolis, MN, Oct. 1993.
[22] Manahah, M. P., Sr., Peterson, A. J., Jr., and Macdonald, D. D., Corrosion Science, Vol. 37, No.
1, 1995, p. 189.
[23] Gerberich, W. W. and Chen, S., "Environment-Induced Cracking of Metals Fundamental Processes:
Micromechanics,'" in "Environment-Induced Cracking of Metals" in Proceedings of the 1st Inter-
national Conference on Environmentally-lnduced Cracking of Metals, R. P. Gangloff and M. B.
Ives, Eds., NACE, Houston, TX, 1990, p. 167.
[24] Steward, J., Wells, D. B, Scott, P. M., and Williams, D. E., Corrosion Science, Vol. 33, No. 73,
1992.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
A. M. Brennenstuhl, t G. Palumbo, l F. S. Gonzalez, l and G. P. Qui re
The Use of Electrochemical Noise to
Investigate the Corrosion Resistance
of UNS Alloy N04400 Nuclear Heat
Exchanger Tubes
REFERENCE: Brennenstuhi, A. M., Palumbo, G., Gonzalez, E S., and Quirk, G. E, " Th e
Use of El ect r ochemi cal Noise t o I nvest i gat e t he Cor r os i on Resi st ance of UNS Alloy N04400
Nucl ear Heat Exchanger Tubes, " Electrochemical Noise Measurement f or Corrosion Appli-
cations, ASTM STP 1277, Jeffery R. Kearns, John R. Scully, Pierre R. Roberge, David L
Reichert, and John L. Dawson, Eds., Ameri can Society for Testing and Materials, 1996, pp.
266-287.
ABSTRACT: Underdeposit corrosion of freshwater-cooled shutdown heat exchangers (SHX)
and steam generators (SG) resulted in unscheduled unit shutdowns very early in the life of the
Ontario Hydro Pickering B Nuclear Generat i ng Station, whereas the same SHX and SG i n the
identical A station had not experienced any problems in over 19 years of operation. UNS alloy
N04400 tubes were used for SHX and SG at bot h stations. Because detailed metallurgical
analysis of tubes removed from the A and B stations suggested that alloy differences may be
a maj or factor to the early failures at the B station, an investigation to characterize the corrosion
resistance of the two alloys was carried out. From standard accelerated corrosion t est methods
it was inferred that the A and B alloys had similar corrosion resistance, suggesting t hat subtle
but important material differences may have been masked by the aggressiveness of the test
methods. Therefore, the electrochemical noise measurement (ENM) technique in conjunction
with less aggressive test environments was used to characterize better the corrosion resistance
of the A and B alloys. Variations in the corrosion resistance of the material removed from
Pickering A and B heat exchangers (HX) were evident. During an initial exposure to deionized
(DI) water at a pH of 5, the B alloy generally displayed greater passivity t han the A material.
However, the B alloy exhibited far less corrosion resistance t han the A material when the pH
of the water was reduced to 1 by the introduction of a nitric aci d/ acet i c acid mixture. The
predomi nant mode of degradation in the B material was intergranular attack with some local-
ized corrosion in grain interiors. The A alloy, on the other hand, exhibited general corrosion.
The study also showed that the corrosion resistance of the A alloy was markedly reduced
by heat treatment. A solution heat treatment at 900~ for 1 h followed by a cold-water quench
or furnace cool from this temperature led to a corrosion performance similar to the nonheat -
treated B alloy. The same heat treatment di d not significantly alter the corrosion performance
of the B alloy. Changes in the mode of attack of the A and the B alloys were also observed
when samples had been thermally treated. Bot h exhibited crystallographic attack, whi ch was
very localized in some areas. Transmission electron microscopy and secondary ion mass spec-
trometry revealed changes in the microstructure, whi ch may account for the differences in t he
mode of attack. The predomi nant precipitate chemistry in the A alloy was Cr23C6, whereas in
the B material it was titanium carbide. The B alloy also displayed boron segregation at t he
grain boundaries. Thi s seemed to influence the extent of intergranular attack i n this alloy.
Research scientists, Ontario Hydro Technologies, 800 Kipling Ave., Toronto, Ontario, Canada, M8Z
5S4.
2 CML Ltd., Manchester, Uni t ed Kingdom.
Copyrighl 9 1996 by ASTM lntcrnational
2 6 6
Www.astIII.OFg
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
BRENNENSTUHL ET AL. ON NUCLEAR HEAT EXCHANGER TUBES 2 6 7
KEYWORDS: UNS N04400, alloy heat exchangers, steam generators, resistance noise, pitting
function, degree of localization, electrochemical potential noise, microstructure, intergranular
corrosion, pitting corrosion, nuclear power
The exceptionally good performance of UNS N04400 alloy heat exchanger (HX) tubes at
the Picketing A Nuclear Generating Station (PNGS A station) resulted in the use of this
alloy in the Picketing B Nuclear Generating Station (PNGS B station). This alloy is used
for shutdown heat exchangers (SHX) and steam generators (SG), which have essentially the
same design and same water sources at both stations. No problems as a result of corrosion
had occurred at the A station during its ten years of operation before the commi ssi oni ng of
the first unit at the B station. However, after onl y six months of service, the first B unit
developed leaks in the SHX tubes, whi ch led to an unscheduled shutdown. Failure analysis
of the leaking tubes suggested that perforation was the result of underdeposit corrosion. Pits
had developed on the outside surface of the tubes. Leaks also developed in the SG at the B
station; these occurred after approximately six years of operation. Again, the leaks were
confined to locations where deposits had developed on the outer diameter (OD) surface. To
establish why the N04400 alloy performed so poorly at the B station, tubes were removed
from both stations for a comparative study.
Detailed metallurgical analysis of the A and B tube samples revealed some differences,
specifically bulk chemistry, inclusion size, and distribution, Analysis of deposits found on
the surface of the A and B SHX tubes indicated that they were similar and typical of what
was expected on SHX tubes exposed to untreated Lake Ontario water. This result seemed
to indicate that metallurgical variations might be an important factor influencing the corrosion
performance of the SHX. The deposits found on the A and B SG tubes were of a compl ex
nature and exhibited both compositional and physical differences. The consequence of this
was that it was impossible to separate alloy and deposit contributions to the degradation
process without carrying out difficult SG simulations involving the two alloys and represen-
tative SG deposits. Therefore, the initial approach used to understand the difference in per-
formance focused on characterizing the corrosion resistance of the two alloys in a simple
test environment.
A corrosion assessment using anodic polarization and open circuit accelerated laboratory
tests failed t o highlight any large variations between the t wo alloys [1]. Because it is possible
that these tests were t oo severe to emphasize subtle but important differences in corrosion
resistance of the A and B alloys, electrochemical noise measurement (ENM) was used to
study corrosion response of the A and B alloys. ENM was selected because its high sensi-
tivity would facilitate the use of a less aggressive test medium, thus potentially providing
more detailed information on the corrosion processes. This technique also has the attraction
of being electrochemically noninvasive and could provide information on the mechanisms
of corrosion [2-4].
This paper outlines the testing carried out on UNS N04400 alloy tubes obtained from the
two stations and describes the results. The effect of heat treatment, specifically quench rate,
was also investigated because the corrosion resistance of this alloy depends on its thermal
treatment history. The thermal history is known to be different for the A and the B alloys
[5].
Experimental Procedure
Bulk Chemical Analysis
Table 1 gives the results of the bulk chemical analysis of the A and B material used in
this investigation. The composition of the t wo materials both conformed to the nominal for
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
268 ELECTROCHEMICAL NOISE MEASUREMEN'f
TABLE 1--Chemical composition of the A and B UNS Alloy
N04400 investigated during this evaluation.
Element PNGS A PNGS B
Ni 65.30 64.5
Cu 31.3 33.5
Fe 1.93 1.16
Mn 1.01 1.0
Si 0.23 0.15
Mg 0.104 0.079
C 0.097 0.083
Ti 0.006 0.040
B 0.0013 0.0021
Cr 0.066 0.007
S 0.011 0.011
Ca 0.0019 0.001
UNS N04400 alloy; however, some differences between the two batches were evident. This
was particularly the case with iron and copper.
Heat Treat ment s
Two treatments were used. Both included a solution treatment at 900~ for 1 h. One set
of samples was quenched in cold water from this temperature; the other series was allowed
to cool in the furnace. The heat treatments were conducted in an argon atmosphere to elim-
inate oxide scale formation.
The effect of heat treatment was investigated because the A and B HX were subjected to
different thermal conditions during the HX fabrication. No accurate records of the thermal
treatments were available.
Mat eri al s Charact eri zat i on
The microstructure of the as-received (AR) and heat-treated samples were characterized
using transmission electron microscopic/scanning transmission electron microscopic (TEM/
STEM), scanning electron microscopic/energy dispersive spectroscopic (SEM/EDS), and
secondary ion mass spectrometry (SIMS) techniques. Thin foils were extracted from tube
samples and examined with a Philips CMI2 TEM/ STEM with a LaB 6 electron source,
operating at 120 keV. The location density and composition of precipitates found were care-
fully recorded. Information on the distribution of boron in the two alloys was obtained using
a Cameca IMS 3f ion microscope using a positive oxygen primary beam current of 1000
nA and an acceleration voltage of 12.5 kV. The beam was rastered over an area of 250 ~m 2,
and transfer optics of 150 and 400 ~Lm 2 were used. The boron distribution was studied
because a previous examination of the A and B materials revealed large differences in the
distribution of this element in the two alloys [1]. Grain boundary misorientation, inclusion
composition data, and the mode of corrosion were obtained using SEM/EDS. A JEOL 840
SEM, operated at 20 kV, was used for this study. The grain boundary microstructure data
were extracted from electron channeling patterns; misorientations are expressed in terms of
the proportion of special boundaries [6,7].
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
BRENNENSTUHL ET AL. ON NUCLEAR HEAT EXCHANGER TUBES 269
Corros i on Testi ng
The corrosion resistance of the A and B material was assessed using a CML Ltd. digital
electrochemical noise integrated system (DENIS) instrument and a three-electrode arrange-
ment. Two "i dent i cal " N04400 alloy electrodes and a saturated silver chloride electrode
(Ag/ AgC1 4M) were used in each cell. The coupling current between the pair of N04400
electrodes was monitored, and the corrosion potential between one of them and the standard
reference electrode was obtained simultaneously. The ENM equipment was programmed to
acquire signals from the electrodes every 4 s. Each file included 1024 data points. Onl y two
cells were monitored at any one time, one cell containing the A material and the other the
B alloy. The monitoring was carried out in series, that is, 1024 data points were obtained
from the A cell before the instrument switched to the B cell. The DENIS system collected,
stored, and performed summary statistics on the raw time domai n data. The summary sta-
tistics included corrosion potential, electrochemical current noise (ECN), coupl i ng current,
electrochemical potential noise (EPN), degree of localization (DoL), and resistance noise
(R,,).
ENM Meas urement s
The ECN and EPN have their usual significance. DoL is the result of the ECN divided
by the rms coupling current and broadly classifies the corrosion process occurring on the
sample surface. DoL shows the variation of localized attack during an experiment but does
not reflect the severity of pitting because the absolute current level from the electrochemical
current noise is not factored in. In fact, the DoL is at its highest during periods when the
noise resistance increases, that is, with very low rates of attack the ENM data are almost
certain to indicate a reasonably high degree of localization even i f localization of attack is
actually low. Therefore, pitting function, a new parameter recently derived by CML Ltd. to
overcome this problem, was used in addition to DoL during this study. The pitting function
is an indication of the significance of localized corrosion and is derived from the DoL, but
is weighted by the average magnitude of the current transients. The calculation of the pitting
function is as follows:
[Ai(1) + . . . + Ai(n)] cri
Pitting Function = " - -
n / ~s
where Ai (I) is the current increase over a 4-s interval (that is, current transient magnitude)
and n is the number of readings taken during Ai. For the data to be accepted into the
summation of the mean transient magnitude for anodic dissolution, the current transient must
have a positive slope and the potential transient must have a negative slope.
Conditions:
- - : +r e
~t
and
~e
m ~ - ve
~t
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
270 ELECTROCHEMICAL NOISE MEASUREMENT
~ilgt is the current transient over 4 s, and ~c/~t is the potential transient over 4 s. I f the
transient continues over more than 1 s, the summation of current for Ai(n) is continuous, tri
is the standard deviation over the 1024 readings, and Irms is equal to the root mean square
of the current.
Resistance noise is equal to the EPN divided by the corresponding ECN value and is
analogous to polarization resistance (Rp) and the charge transfer resistance (R, ) obtained by
linear polarization resistance measurements and impedance spectroscopy, respectively [8].
Test Samples and Environments
Test samples were made from tubes extracted from PNGS A and PNGS B SG tubes. The
SG tubes were cut into 2. 5-cm lengths, and the samples requiring thermal processing were
given the heat treatment already outlined. This sample length gave an outside surface area
of 7.5 cm 2. The test samples were connect ed to the monitoring equipment by a stainless
steel wire attached to the inside surface of the tube. A pol ymer coating was shrunk onto the
stainless steel wire to prevent contact with the test environment. The ends of the tube samples
were sealed with epoxy resin. All samples were abraded with 320-grit silicon carbide paper
and swabbed with isopropyl alcohol j ust before testing. Testing was carded out in triplicate.
Figure 1 shows the electrode arrangement in the test cell. The test cells, which contained
1 L of DI water, were immersed in a water bath, which was held at a constant temperature
of 30~ ( + / - 2 ~ The DI water had a pH of 5. Exposure to DI water lasted 120 h. The
purpose of this initial exposure was to determine the response of the two alloys in a benign
environment. After 120 h, a 10-mL aliquot containing concentrated nitric acid (70%) and
acetic acid (99.8%) (HNO3/ HAC) in equal amounts was injected into the DI water. Two
additions to the DI water were made 24 h apart. The pH had decreased to one after the
second acid addition.
The HNO3/HAC mixture was used because it is commonl y used as a etchant for revealing
microstructural details in copper-containing alloys including UNS N04400 alloy and was,
therefore, believed to be an appropriate starting medium. Data acquisition was continued for
an additional 34 h after the second etchant injection.
Results
The results presented here are typical outputs for the three sample conditions. Onl y the
pitting function, resistance noise, DoL, and EPN outputs are presented because they best
show the behavior of the alloys examined during this study. The arrows in the pitting function
plots indicate when the HNO3/ HAC additions were made.
Precipitate and inclusion compositions are given in Table 2. Special boundary fractions
and information on boron segregation are also presented.
As-Received (AR) Samples
Figure 2 shows pitting function output for the A material in the AR condition. The pitting
function was initially quite high for both A and B samples with values for the latter being
much greater than the former. Hi gh values are a common observation at the start of moni-
toring and are attributed to increased corrosion associated with defects in the surface film
resulting from the sample preparation process. Bot h passivation charge and passivation cur-
rent density are observed to be significantly affected by residual surface deformation intro-
duced during sample preparation [I]. The initially high pitting function outputs diminished
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
BRENNENSTUHL ET AL. ON NUCLEAR HEAT EXCHANGER TUBES 271
FIG. l --Det ai l s of the test cell and sample arrangement used in this study.
TABLE 2- - Met al l urgi cal differences between A and B as received and heat-treated samples.
PNGS A PNGS B
Solution Treated Solution Treated
As Received CWQ FC As Received CWQ FC
Precipitate composition Cr23C 6 none none TIC" none none
Inclusion composition MgS MgS MgS CaSb CaS CaS
Boron at grain boundaries none none none extensive none some
% Special grain boundaries 47% . . . . . . . . . . . . . . .
Predominant precipitate composition Cr23C 6 w as also present.
b CaS inclusions al so contained t to 5% magnesium.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
272 ELECTROCHEMICAL NOISE MEASUREMENT
1E-04
1E-05
IE-06
O
1E-07
1E-08
IE-09
~, 11~ ~ ~ o 4~"
1E-10
0 50 100 150 200
Time/Hours
' A' ' B'
- - - , o, - -
FIG. 2- - Pi t t i ng f unct i on versus time f or AR A and B alloy N04400 tubes.
wi t h t i me. The B al l oy showed a cont i nuous decrease duri ng t he DI wat er exposur e until
the echant addi t i on was made. A reduct i on of appr oxi mat el y t wo orders of magni t ude was
seen. Fi nal pi t t i ng funct i on values ranged from 4 to 7 l 0 - l ~ A. A di fferent pat t ern was
apparent wi t h the A mat eri al . A decrease was first observed, f ol l owed by a gradual i ncrease
from 1 to 7 l 0 9 A. The i nt roduct i on of the et chant i nt o the test cel l resul t ed in an i ncrease
in the pi t t i ng funct i on of t he B mat eri al of mor e than t hree orders of magni t ude. Dur i ng t he
final hours of the HNOa / HAC phase of the exposure, t he pi t t i ng funct i on r anged from 1 to
4 10 -6 A. An i ncrease in the pi t t i ng funct i on of the A mat eri al was al so evi dent . However,
the i ncrease was not as l arge as was seen for the B mat eri al , and the pi t t i ng funct i on recov-
ered to previ ous levels short l y aft er et chant had been added. The second HNO3/ HAC addi t i on
appears as a sharp i ncrease in t he pi t t i ng funct i on for a br i ef per i od of t i me f ol l owed by an
equal l y r api d decay. Bot h the A and B al l oys r esponded in the same manner.
The DoL outputs (Fig. 3) exhi bi t some di fferences when compar ed wi t h t he pi t t i ng func-
tion data. At the start of t he exposure, the DoL was l owest for t he B mat eri al . The converse
was true wi t h the pi t t i ng funct i on results, wi t h the except i on of t wo l arge transients. The A
mat eri al di spl ayed a rel at i vel y hi gh and fluctuating out put duri ng the DI wat er part of t he
exposure. Etchant i nj ect i on into the t est cel l resul t ed in an i ncrease in the DoL for the B
mat eri al , then an i mmedi at e decrease, fol l owed by a further t ransi ent when the second ad-
di t i on was made. The general t rend was t oward an i ncrease in the DoL, suggest i ng i ncreasi ng
l ocal i zed attack. In the case of the A mat eri al , the t endency was al so t oward a decr ease in
t he degr ee of l ocal i zat i on aft er t he HNO3/ HAC addi t i ons, i ndi cat i ng general corrosi on.
Fi gure 4 shows the R, data. These plots, l i ke the pi t t i ng funct i on results, al so show a l arge
di fference bet ween t he A and B al l oys. The A mat eri al di spl ayed bet t er corrosi on resi st ance
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
BRENNENSTUHL ETAL. ON NUCLEAR HEAT EXCHANGER TUBES 273
IE+O0
1E-01
1E-02
1E-03 , I , ,
/
0 50 100 150 200
Time/Hours
I
- - ' A ' - 4 - ' B '
FIG. 3 - - Do L versus t i me f o r A and B al l oy N04400 AR tubes.
at the very start of the exposure. However, this qui ckl y decr eased to val ues si gni fi cant l y
bel ow t hose for t he B mat eri al . Further, the A mat eri al exhi bi t ed a rel at i vel y smoot h output
during this period, whereas the B signal was variable. R, values ranged from 200 to 500 kf~
for the A al l oy wi t h the pr eponder ance of dat a poi nt s bei ng in the 200- to 300-kl~ range.
This cont rast s wi t h t he B alloy, whi ch exhi bi t ed values rangi ng from 300 to 6000 kl-I wi t h
the average bei ng about 1000 kI~. When t he et chant was i nj ect ed into the t est cel l , a dramat i c
reduct i on in R, of appr oxi mat el y four decades was observed for the B alloy, and t he values
r emai ned at l ow levels for the rest of t he exposure. Resi st ance noi se val ues rangi ng f r om 60
to 300 f~ were t ypi cal . The A al l oy al so di spl ayed a reduct i on in R, but onl y by a factor of
t wo to five (values were in the 30 to 200- kf l range), and the out put seemed to be more
vari abl e than before the HNO3/ HAC i nt roduct i on. These dat a suggest t hat the A al l oy t ends
t oward general corrosi on whereas that for the B al l oy is l ocal i zed. Thi s conflicts wi t h the
pi t t i ng funct i on and DoL results.
The EPN outputs are shown in Fi g. 5. Before the addi t i on of the et chant in the DI water,
the EPN for the B mat eri al was gener al l y sl i ght l y hi gher than for t he A alloy, and appear ed
to fluctuate more. The aci d addi t i ons are cl ear l y evi dent in t hese plots. Af t er the aci d addi t i on,
bot h mat eri al s seem to behave in a si mi l ar manner. The ampl i t ude of the fluctuation was a
fact or of appr oxi mat el y five less t han bef or e the aci d addi t i on. These dat a al so seem to
conflict wi t h the pi t t i ng funct i on and DoL outputs for t he B alloy, whi ch show an i ncrease
in l ocal i zed act i vi t y aft er the et chant addi t i on was made. A concomi t ant i ncrease in EPN
was expect ed from our experi ence of passi vat i ng al l oys l i ke st ai nl ess steels.
Scanni ng el ect ron mi cr ogr aphs for the t wo bat ches of al l oy N04400 aft er the exposure
are shown in Fi g. 6. The A mat eri al di spl ays uni form corrosi on; grai n boundari es are evident,
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
274 ELECTROCHEMICAL NOISE MEASUREMENT
1E+07
1E+06
1E+05
9
Z 1E+04
o
~ 1E+03
1E+02
IE+01
I
0 50 100 150 200
Time/Hours
_ _ ' A ' + ' B '
FIG. 4--Resistance noise versus time f or A and B alloy N04400 AR tubes.
but there was no appreciable ditching. This contrasted with the surface of the B samples in
which attack was mostly at grain boundaries with some localized corrosion in the grain
interiors. The localized corrosion took the form of numerous small pits all at the same
location and often extending across several grain boundaries.
The TEM/ STEM examination revealed precipitates at grain boundaries, and within the
grain interiors, predominantly at the dislocations of both materials. Cr 23C 6 precipitates were
found in the A alloy, and a combination of TiC and Cr 23C 6 wa s in the B alloy. The SIMS
imaging of boron indicated segregation of this element in many of the grain boundaries of
the B material and no segregation in the A alloy. No other elements were found with boron,
which seems to imply that it is likely in solid solution. Results from the grain boundary
misorientation measurements suggested a similar proportion of special boundaries 47 and
49% for the A and B alloys, respectively. Nonmetallic inclusion elemental analysis showed
that inclusions in the A alloy contained mostly magnesi um and sulphur; those in the B
material were essentially calcium and sulphur with a small amount of magnesi um (<5%).
In addition, the inclusions in the A alloy were small and uniformally distributed over the
entire surface, whereas these in the B alloy were larger and at isolated locations.
Col d- Wat er - Quenched ( CWQ) Sampl es
Pitting function results for the cold-water-quenched samples are shown in Fig. 7. Bot h the
A and B material start at similar levels and decrease together during the first 25 h. The
decrease is relatively smooth, showing little or no large fluctuations. The initial decrease for
the A material was followed by a relatively variable signal about 65 h into the test when a
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
BRENNENSTUHLETAL ON NUCLEAR HEAT EXCHANGER TUBES 275
1E+00
1E-OI
IE-02
1E-03
1E-04
1E-05
L
50 100 150 200
Time/Hours
_ _ ' A ' ~ ' B '
FIG. 5- - EPN versus time for A and B alloy N04400 AR tubes.
steady increase became evident, the increase ranging from 3 10 -9 to 4 x 10 -8 A. A
variable signal was also observed for the B material, but, unlike the A alloy, no major
increase was evident during the DI water exposure. The average pitting function value during
this phase of the test was 1 x 10 8 A. When the HNO3/ HAC addition was made to the cell,
both the A and B Al l oy N04400 responded in an identical manner - - t he pitting function
increased dramatically. The final pitting function levels were similar for both alloys. Average
pitting functions were 1.8 x 10 -6 and 3 x 10 -6 A for the A and B alloys, respectively. The
extent of pitting of these samples was like that for the AR B material.
Figure 8 gives the DoL data. The A material displayed numerous high amplitude fluctu-
ations during the DI water exposure; the trend was toward a gradual increase in DoL. A
similar tendency was observed with the B material except that the amplitude of the transients
was lower and of higher frequency. When the etchant was added to the test cells, the B
material exhibited a sharp increase whereas the A material showed very little change. After
the acid addition, the DoL for the A alloy was about twice that of the B sample. The pitting
function and DoL outputs suggest localized corrosion for both alloys.
Resistance noise results are given in Fig. 9. Bot h alloys show high R, values during the
DI water exposure. The A material exhibits decreasing R, with time, whereas the B material,
although fluctuating considerably, displays an increasing trend. R, values for the B material
range from 150 to 3500 kf~ and 200 to 900 kf~ for the A alloy. Once the acid was added
to the cells, the R, for both of the alloys decreased to very low values. The B material (60
f~) was consistently lower than the A (80 f~). The second etchant addition had little lasting
effect. These data contradict the pitting function and DoL outputs and suggest general
corrosion.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
276 ELECTROCHEMICAL NOISE MEASUREMENT
FIG. 6--(A) AR A sample after testing showing general corrosion (x2000). (B) AR B sample after
testing showing intergranular attack and corrosion in the grain interiors (x2000).
EPN results are given in Fig. 10. Both materials showed similar noise before and after
the acid addition. Large potential noise spikes were evident when the acid additions were
made. The amplitude of the outputs was greatly reduced after the acid addition and displayed
much less noise than was the case before the etchant was introduced into the cells. This
kind of response was also seen with the AR samples.
Scanning electron micrographs for the A and B materials after the exposure are shown in
Fig. 11. The attack in the B was much less concentrated at the grain boundaries than was
the case with the AR samples, and distinct pits were clearly evident. Etching of crystallo-
graphic features in the grain interiors was a major mode of corrosion. The corrosion mor-
phology of the A material is similar to that for the B alloy. Corrosion was localized and pits
were evident. This was in sharp contrast to the general corrosion observed for the AR
samples.
The TEM/ STEM examination indicated that the solution treatment had dissolved almost
all of the precipitates found in the AR samples. Only a small number of isolated large TiC
precipitates was detected in the B material. Results from the SIMS analysis showed no boron
segregation in the A or B alloys. No change was found to have occurred in the distribution
or morphology of the sulphur-beating intrusions as a result of the water quench.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
BRENNENSTUHL ET AL. ON NUCLEAR HEAT EXCHANGER TUBES 277
1E-04
1E-05
1E-06
O
"gl
o
~. 1E-07
2
1E-08
IE-09
1E-10
0 50 100 150 200
Time/Hours
_ _ ' A ' ' B '
FIG. 7--Pi t t i ng function versus time f or A and B alloy N04400 after solution treatment f or 1 h at
900~C followed by a cold water quench.
Fur nace- Cool ed ( FC) Sampl es
Figure 12 shows pitting function plots for the furnace-cooled samples. The outputs for the
A and B samples were almost identical. The pitting function values for both materials started
at relatively high values and after 15 h of exposure had dropped by one to two orders of
magnitude to approximately 1 x 10 9 A showing little further change until the acid mixture
was added. Not e that during this "st abl e period" bot h alloys exhibited fluctuations of rela-
tively short duration; the B material was somewhat worse in this regard. The introduction
of the acid mixture resulted in a three-decade increase in the pitting function, Both alloys
showed a similar response, Values were typically in the 3 to 5 x 10 6 A range.
DoL results are shown in Fig. 13. The plots for the t wo materials were similar, except
that the fluctuations in the outputs for the B alloy were of a larger amplitude. After the acid
addition, some separation was evident, the A material showing significantly higher values
than the B alloy.
Figure 14 shows the R, results for the t wo alloys. The plots were similar, with the B alloy
displaying slightly higher values than the A sample during the DI water phase of the test.
Data obtained during the low pH exposure revealed almost no difference. Overall, the outputs
were generally slightly higher than those obtained from the cold-water-quenched samples,
EPN results are shown in Fig. 15. Again, during the higher pH phase of the test the noise
for the B alloy was greater than that for the case with the A, but after the acid addition the
outputs outs were similar.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
278 ELECTROCHEMICAL NOISE MEASUREMENT
1E+O0
1E-01
O
1E-02
i t 7 t
T
1E,-03 ' ' '
0 50 100 150 200
Time/Hours
' A' . . 4 - 'B'
FIG. 8--DoL versus time for A and B alloy N04400 after solution treatment at 900~ f or 1 h followed
by a cold water quench.
Scanni ng el ect ron mi cr ogr aphs for the A and B sampl es are shown in Fi g, 16. The A
al l oy exhi bi t s general at t ack wi t h some cr yst al l ogr aphi c et chi ng in the grai n interiors. Some
of the grai n boundari es were al so di t ched. Si mi l ar features were obser ved for the B mat eri al ;
however, a l arger number of the grai n boundari es were heavi l y di t ched. The grai n boundar y
at t ack obser ved on the B sampl e was not as l ocal i zed as was the case wi t h the AR B alloy.
The TEM/ STEM exami nat i on suggest ed that, even t hough t he cool i ng rat e had been
r educed by furnace cool i ng, t he rate was still t oo fast for preci pi t at i on to occur. The mi -
crost ruct ure appear ed to be si mi l ar to that seen aft er wat er quenchi ng. Resul t s from the SI MS
anal ysi s i ndi cat ed that boron had ret urned to some of the grai n boundari es. Agai n, no change
had occurred in the mor phol ogy and di st ri but i on of t he i ncl usi ons as a resul t of the heat
treatment.
Di s c u s s i o n
The reason for the poor per f or mance of UNS Al l oy N04400 in PNGS B SHX and SG
must be det er mi ned to i nt roduce r emedi al measures to ensure the cont i nued r el i abl e operat i on
of t hese syst ems for the r emai ni ng l i fe of the nucl ear units. Because t he HX t hat exhi bi t the
most pr onounced degradat i on ( SG and SHX) are subj ect ed to vast l y di fferent operat i ng en-
vi ronment s, attention was focused on the i nherent cor r osi on resi st ance of t he A and B al l oys.
Previ ous chemi cal and met al l urgi cal anal ysi s of the A and B al l oys reveal ed di fferences that
mi ght cause vari at i ons in corrosi on resi st ance. St andard accel er at ed cor r osi on tests were used
t o compar e the al l oys, but t hese fai l ed t o reveal any l arge di fferences bet ween t he t wo
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
BRENNENSTUHL E-I- AL. ON NUCLEAR HEAT EXCHANGER TUBES 279
IE+07
1E+06
1E+05
Z IE+04
~ 1E+03
1E+02
1E+01 r
0 50 100
Time/Hours
[
150 200
_ _ ' A ' + ' B '
FIG. 9- - Resi st ance noise versus time f or A and B alloy N04400 af t er solution treatment at 900~ f or
1 h and col d wat er quench.
materials [1]. The possibility that the accelerated tests selected were so severe that they
masked subtle but important differences in the corrosion resistance led to the ENM approach
being selected as a method for better defining the corrosion performance of the two alloys.
ENM was attractive because it is very sensitive, electrochemically nonintrusive, and can
provide mechanistic information [2,4].
As - Re c e i v e d A a n d B Tube Ma t e r i a l
The ENM results for the AR tube samples clearly revealed a difference in the corrosion
response of the A and B alloys. This was particularly noticeable after the etchant addition.
A much larger pitting function and lower R, were recorded for the B material, indicating
less corrosion resistance than the A alloy in the pH 1 environment. This result is consistent
with the field performance. Interestingly, the variation was clear with both the R, and pitting
function outputs; this is not commonl y the case. Generally, resistance noise is onl y useful
when corrosion is of an active nature (that is, general corrosion), and the pitting function is
only a practical parameter when attack is localized. Variations in t opography of the test
samples help to explain this unusual behavior. Local i zed corrosion was evidently occurring
in the B alloy but was very widespread; almost all the grain boundaries exhibited significant
corrosion together with some attack in the grain interiors. This increase in pitting function
is therefore likely the consequence of propagation at the grain boundaries, a process that is
relatively unstable because of significant crystallographic heterogeneities at these locations.
Because the process is widespread, the mean potential of the system and the EPN, as a
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
280 ELECTROCHEMICAL NOISE MEASUREMENT
IE-01
1E-02
~ 1E-03
1E-04
IE-05 i J
0 50 100
Time/Hours
_ _ ' A ' _ . _ ' B '
L J
150 200
FIG. I O- - EPN outputs f or A and B alloy N04400 af t er solution treatment f or 1 h and coM water
quench.
consequence, is dr amat i cal l y lowered. In cont rast to t he B al l oy t opography, t he A surface
di spl ayed general corrosi on; grai n boundari es coul d be seen, but t hey were not heavi l y at-
t acked. The el ect rochemi cal outputs aft er the etchant addition, speci fi cal l y t he resi st ance
noise, refl ect ed a general corrosi on condi t i on at a low rate.
A compar i son of the grai n boundar y mi crost ruct ure of the t wo al l oys reveal ed di fferences
in preci pi t at e chemi st ry, Cr23C 6 and TiC for the A and B al l oys, respectively, wi t h the l at t er
al l oy al so cont ai ni ng some Cr23C 6. These preci pi t at es were al so seen at grai n i nt eri ors pre-
domi nant l y at di sl ocat i ons. Boron was al so found to be segregat ed at the grai n boundari es
of the B alloy. Because the degree of mi sori ent at i on of the grai n boundar i es of the t wo al l oys
was similar, it appears that t hese mi cr ochemi cal di fferences are a maj or fact or in t erms of
the i nt ergranul ar at t ack of the B alloy. The cause of the l ocal i zed corrosi on in the grai n
i nt eri ors is not as clear. The mode of at t ack suggest s that it may be segregat i on related.
Because t here was no evi dence of t her modynami cal l y possi bl e spi nodal decomposi t i on
[9- 11], whi ch mi ght account for l ocal changes in al l oy mi crost ruct ure, it is not possi bl e to
ci t e this as the cause. A possi bl e expl anat i on is the presence of sul phur-beari ng i ncl usi ons.
In addi t i on to chemi st ry di fferences, the si ze and di st ri but i on of t hese i ncl usi ons are mar kedl y
different. A previ ous SI MS exami nat i on of t he A and B al l oys suggest s that t hey were evenl y
di st ri but ed in the A al l oy and hi ghl y segregat ed in the B al l oy [12]. Thi s si t uat i on coul d l ead
to di fferences in the anode/ cat hode areas, wi t h the B mat eri al suffering the hi gher l ocal i zed
corrosi on because of its smal l er anode- t o- cat hode ratio.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
BRENNENSTUHL ET AL. ON NUCLEAR HEAT EXCHANGER TUBES 281
FIG. 1 I--(A) Solution-treated and CWQ A sample after testing showing crystallographic attack
(x2000). (B) Solution treated and CWQ B sample after testing showing similar corrosion to the A
material (
Some interesting observations were also made during the D! water phase of the test. The
pitting function, R, and DoL outputs seem to indicate that the B material was more passive
in this relatively benign environment than the A alloy. This is also possibly due to the lower
concentration of sulfur-bearing inclusions. These inclusions act as sites for local protective
film breakdown.
Evidently, the noise parameters selected for assessing the corrosion resistance of the two
alloys examined in this investigation clearly highlighted the differences between them with
the exception of the DoL. Overall, the DoL for these samples followed the pitting function
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
282 ELECTROCHEMICAL NOISE MEASUREMENT
-04
1E-05
IE-06
O
~ w 0 7
1E-08
IE-10 ' J '
0 50 100 150 ~.00
Time/Hours
_ _ ' A ' + ' B '
FIG. 12--Pitting function versus time f or A and B alloy N04400 after solution heat treatment at 900~
f or 1 h followed by a furnace cool.
outputs. However, the t ransi t i on bet ween t he DI wat er exposure and l ow pH condi t i on was
oft en less t han clear.
Sol ut i on- Treat ed and Col d- Wat er - Quenched A and B Tube
A sol ut i on t reat ment at 900~ fol l owed by a wat er quench l ed to a dramat i c change in
t he corrosi on response of t he A mat eri al . The A mat eri al agai n gener al l y showed sl i ght l y
hi gher act i vi t y than t he B mat eri al bef or e t he aci d mi xt ure addi t i on. Duri ng t he DI wat er
part of the exposure, the B sampl e out put s sl i ght l y depart ed from what was seen in the AR
exami nat i on. It was evi dent t hat the out put was now noisier. The reason for this is not cl ear
but does seem to suggest t hat the heat t reat ment has al t ered the corrosi on resi st ance of the
passi ve film. A very l arge i ncrease in the pi t t i ng function was evi dent when the HNO3/ HAC
was added to the t est cel l for bot h A and B al l oys. Results for the A al l oy i ndi cat e that the
ext ent of corrosi on was appr oachi ng that found for the B alloy. Al t hough the act i vi t y of the
B was not si gni fi cant l y great er than was obser ved for the AR B mat eri al , the mode of at t ack
was different. Corrosi on was pr edomi nant l y cryst al l ographi c. Grai n boundar y at t ack was
great l y reduced. Thi s absence of grai n boundar y corrosi on coi nci ded wi t h an absence of
preci pi t at es and boron at the grai n boundari es. Thi s support s the suggest i on that t here is a
rel at i onshi p bet ween grai n boundary chemi st r y and i nt ergranul ar corrosi on. Unfortunately,
however, t he pri ce for r educed grai n boundary corrosi on is great er corrosi on of the grai n
interiors. Cor r osi on was al so cr yst al l ogr aphi c in t he A al l oy but di d not appear to be as
pr onounced as was t he case wi t h the B mat eri al . The resi st ance noi se out put s were al so
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
BRENNENSTUHL ET AL. ON NUCLEAR HEAT EXCHANGER TUBES 283
1E+00
1E-01
O
1E-02
1E-03 ~ I 1
0 50 100 150 200
Time/Hours
_ _ ' A ' _ - r ' B '
FIG. 13- - DoL versus time f or A and B alloy N04400 af t er solution heat treatment at 900~ f or 1 h
f ol l owed by a f urnace cool.
si mi l ar for bot h A and B sampl es (that is, the resi st ance noi se dr opped in the same manner
as it di d wi t h the AR B al l oy when t he et chant was added. Agai n, the R, and EPN dat a
were not consi st ent wi t h t hose for the pi t t i ng function. Thi s confl i ct cannot be account ed for
by grai n boundar y at t ack because this was great l y r educed by t he heat t reat ment but can be
r at i onal i zed in t erms of the l ocal i zed cr yst al l ogr aphi c et chi ng and pitting.
The di fference bet ween the AR A sampl es and t he sol ut i on-t reat ed and col d- wat er -
quenched A is evi dent in the DoL pl ot s but is not as cl ear as was the case wi t h t he pi t t i ng
funct i on and R n plots. Further, t he t ransi t i on bet ween the DI wat er exposure and the et chant
addi t i on was not obvi ous. Al so, the results suggest that the degr ee of l ocal i zat i on of the A
mat eri al was great er than that for the B alloy.
Sol ut i on- He at - Tr e at e d a n d Fu r n a c e - Co o l e d A a n d B Tube
It seems from the results of this eval uat i on t hat reduci ng the cool i ng rate f r om 900~ by
furnace cool i ng does not al t er the cor r osi on resi st ance from t hat found with the col d wat er
quench. Agai n, the R, resul t s confl i ct wi t h t he pi t t i ng function dat a duri ng the HNO3/ HAC
phase of t he test. Before the et chant addi t i on, t he B al l oy was noi si er than the A alloy, but
the average pi t t i ng funct i on was appr oxi mat el y the same i ndi cat i ng an i mpr ovement in t he
passi vi t y of the A alloy. The DoL appeared, aft er an initial decrease, to i ncrease gr adual l y
wi t h exposure time. Before the aci d addi t i on, t he A sampl e seemed to be less active than
was the case wi t h the A sol ut i on-t reat ed and col d- wat er - quenched sampl es. Af t er the
HNOf l HAC i nt roduct i on, the A mat eri al di spl ayed the great est degree of l ocal i zat i on. These
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
284 ELECTROCHEMICAL NOISE MEASUREMENT
1E+07
1E+06
~ 1E+05
O
Z 1E+04
8
1E+03
1E+02
1E+01 i L
0 50 100 150 200
Time/HourS
_ _ ' A ' ~ ' B'
FIG. 14--Resi st ance noise versus time f or A and B alloy N04400 af t er solution heat treatment at
900~ f or 1 h f ol l owed by a fi4rnace cool.
dat a appear to i ndi cat e that gener al l y the B al l oy has a great er t endency to passi vi t y than
the A mat eri al .
The reappearance of si gni fi cant grai n boundary at t ack especi al l y in the B mat eri al appears
to correl at e wi t h the segregat i on of boron at some grai n boundari es suggest i ng that this
el ement may be the reason for the sensi t i vi t y of grai n boundar i es to attack. Preferent i al grai n
boundary corrosi on was al so obser ved in the A alloy. However, it was not as preval ent as
was the case with the B sampl es, and it was not possi bl e to correl at e t he cor r osi on wi t h a
change in grai n boundary chemistry.
I m p l i c a t i o n s o f T e s t R e s u l t s
The results cl earl y show that in the envi ronment s used duri ng this study, pH 5 DI wat er
and pH I HNO3/ HAC, the as-recei ved (nonbeat -t reat ed) A and B UNS N04400 behave
differently. It appears that the B al l oy gener al l y has great er passi vi t y in the pH 5 DI wat er
than the A material. Thi s di fference can l i kel y be at t ri but ed to the di st ri but i on and compo-
sition of sul fur-beari ng i ncl usi ons. The i ncl usi ons in the A al l oy were uni f or ml y di st ri but ed,
whereas in the B al l oy they t ended to be l ocal i zed [12]. Sulfide i ncl usi ons are a maj or cause
of loss of passi vi t y in ni ckel al l oys and st ai nl ess steels [13,14]. Thi s great er passi vi t y of the
B al l oy duri ng the DI wat er phase of t he exposure was sl i ght l y affect ed by heat treatment.
Unfortunately, B al l oy does not retain the advant age associ at ed wi t h superi or passi vi t y in
the low pH envi ronment . Int ergranul ar at t ack occurs. The results appear to i mpl y t hat grai n
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
1E+00
1E-01
1E-02
1E-03
IE-04
BRENNENSTUHL ET AL. ON NUCLEAR HEAT EXCHANGER TUBES 285
1E-05
I I I
0 5O 100 15O 20O
Time/Hours
__ ' A' ~ ' B'
FIG. 15- - EPN outputs f or the A and B alloy N04400 af t er solution heat treatment at 900~ f or 1 h
f ol l owed by a f urnace cool.
boundary chemistry of the B is less corrosion resistant than the A in the pH 1 environment.
Based on the solution-treated and furnace-cooled results, boron appears have a significant
influence on the corrosion resistance of the grain boundaries. The intensity of the attack at
the grain boundaries was likely promot ed by the relatively good corrosion resistance of the
grain interiors of the B alloy. This resulted in a high potential existing between the grain
boundaries and grain interiors.
Some localized attack was observed in the grain interiors; the reason for this is not clear
but may be connected to the presence of isolated inclusions in the B alloy. The relatively
good corrosion resistance of the material surrounding the inclusions would lead to the de-
velopment of a relatively high potential difference between it and the sulphide inclusions in
the interface region between the alloy and the inclusion, thus promoting localized attack.
Therefore, although the B alloy generally shows greater corrosion resistance than the A,
which suffers from general attack, this superior corrosion resistance does not translate into
better localized corrosion performance. To the contrary, the performance of the B alloy is
diminished because of the microstructural weaknesses.
The solution heat treatment used during this study shows that the A material can be made
to behave like the B alloy at least in the pH 1 environment. This highlights the heat treatment
sensitivity of the A alloy. Whet her the corrosion resistance of the B alloy can be i mproved
by heat treatment has yet to be determined. Certainly, the limited number of thermal treat-
ments examined in this study did not appear to lead to a significant change.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
286 ELECTROCHEMICAL NOISE MEASUREMENT
FIG. 16--(A) Solution-treated and FC A sample after testing showing crystallographic attack and
some grain boundary ditching ( x2000). (B) Solution-treated and FC B sample after testing showing
similar corrosion to the A sample but greater grain boundary attack (X2000).
Electrochemical Noise Measurement Considerations
The results of this investigation suggested that the system under examination was one in
which the pitting function and resistance noise were both broadly applicable. This infers a
system that ties between general corrosion and localized attack, thereby making rate deter-
minations difficult. No attempt was made to obtain rate information during this investigation,
but an analysis of the test samples and the ENM outputs with the aim of determining whether
reliable rate information can be arrived at using the pitting function and R, is planned for
the future.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
BRENNENSTUHL ET AL. ON NUCLEAR HEAT EXCHANGER TUBES 287
Conclusions
1. The ENM resul t s of this st udy cl earl y i ndi cat ed that the AR B al l oy has a l ower
corrosi on resi st ance than t he AR A t ubes when exposed to the hi ghl y oxi di zi ng l ow
pH ( appr oxi mat el y pH l ) condi t i ons used is this study.
2. Duri ng the DI wat er exposure phase ( appr oxi mat el y pH 5), t he B mat eri al gener al l y
di spl ayed great er passi vi t y than the A sampl e.
3. The corrosi on resi st ance of the A t ube mat eri al can be si gni fi cant l y r educed by a 900~
sol ut i on heat t reat ment to the ext ent that it had a si mi l ar corrosi on resi st ance to the B
alloy. UNS Al l oy N04400 is evi dent l y sensi t i ve to heat treatment.
4. The heat t reat ment used duri ng this eval uat i on l ed to a change in the mode of degra-
dat i on for bot h al l oys. Thi s seems to correl at e wi t h grai n boundar y mi cr ochemi st r y in
t he B alloy, wi t h boron bei ng the maj or factor. The A al l oy was not as sensi t i ve to
grai n boundary degr adat i on as the B mat eri al .
5. The hi gh pi t t i ng funct i on and unexpect ed r educed EPN and R, response at l ow pH are
expl ai ned in t erms of t he nature of attack, which, al t hough l ocal i zed, was ext ensi ve.
Acknowl edgment s
The authors woul d l i ke t o t hank M. Pri kryl and G. C. Weat herl y at McMast er Uni versi t y
for the TEM/ STEM studies and G. Mount at The Uni versi t y of West ern Ont ari o for t he
grai n boundar y SI MS i magi ng.
References
[1] Palumbo, G., Brennenstuhl, A. M., and Gonzalez, F. S., in Application of Accelerated Corrosion
Tests to Service Life Prediction of Materials, ASTM STP 1194, American Society for Testing and
Materials, West Conshohocken, 1992, p. 252.
[2] Hladky, K. and Dawson, J. L., Corrosion Science, Vol. 21, No. 4, 1981, p. 317.
[3] Eden, D. A., Rothwell, A. N., and Dawson, J. L., Paper 444, Corrosion 91, NACE, Cincinnati,
1992.
[4] Searson, E C. and Dawson, J. L., Journal of the Electrochemical Society, Vol. 135, No. 8, 1988,
p. 1908.
[5] De, P. K. and Bose, A., Corrosion, Vol. 49, No. 2, 1993.
[6] Brandon, D. G., Acta Metallurgica, Vol. 14, 1966, p. 1479.
[7] Palumbo, G., Aust, K. T., Erb, U., King, E J., Brennenstuhl, A. M., and Lichtenberger, E C.,
Physica Status Solidi (a), Vol. 131, 1992, p. 425.
[8] Lumsden, J. B., Kendig, M., and Jeanjaquet, S., Paper 224, Corrosion 92, NACE, Nashville, 1992.
[9] Srikanth, S. and Jacob, K. T., Materials Science and Technology, Vol. 5, 1988, p. 427.
[10] Vrijen, J. and Radelaar, S., Physical Review, Series B, Vol. 17, 1978, p. 409.
[11] Sachtler, W. M. H. and Dorgelo, G. J. H., Journal of Catalysis, Vol. 4, 1965, p. 654.
[12] Brennenstuhl, A. M., Doherty, E E., Bellamy, G., King, E J., and Dunstall, T. G., "The Corrosion
of Freshwater Cooled Nuclear Heat Exchangers at Ontario Hydro," in Canadian Nuclear Society
Proceedings of a Conference on Steam Generators and Heat Exchangers, Toronto, 1990, pp. 1-69.
[13] Marcus, E and Talah, H., Corrosion Science, Vol. 29, 1989, p. 455.
[14] Kesten, M., Corrosion, Vol. 32, No. 3, 1976, p. 94.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
P. E. Dohert y, ~ M. J. Ps ai l a- Dombr ows ki , 2 S. L. Harper, 2
and W. G. Schnei der ~
The Use of Electrochemical Noise
Measurement in the Evaluation of
Materials for Steam Generators
REFERENCE: Doherty, E E., Psaila-Dombrowski, M. J., Harper, S. L., and Schneider, W.
G. , " Th e Us e o f El e c t r o c h e mi c a l No i s e Me a s u r e me n t i n t he Ev a l u a t i o n o f Ma t e r i a l s f o r
S t e a m Ge ne r a t o r s , " Electrochemical Noise Measurement for Corrosion Applications, ASTM
STP 1277, Jeffery S. Kearns, John R. Scully, Pierre R. Roberge, David L. Reichert, and John
L. Dawson, Eds., American Society for Testing and Materials, 1996, pp. 288-304.
ABS TRACT: In the past ten years, there has been a progressive development in corrosion
monitoring technology [1-12]. Initial approaches to evaluate the performance of steam gen-
erator (SG) alloys were based on mass-loss coupons and polarization resistance measurements
of probes in SG environments or aggressive water chemistries. Corrosion rates and mechanisms
were characterized and quantified with data obtained from metallurgical examination of test
specimens. Several of these earlier autoclave tests were conducted over a number of years with
inconclusive results. Recent interest has focused on the application of modern electrochemical
techniques [6,7], such as current and potential electrochemical noise, for greater sensitivity in
the evaluation of localized and general corrosion behavior.
This paper reviews the progress to date of a multidisciplinary study to investigate the feas-
ibility of electrochemical techniques to evaluate steam generator materials. The primary tech-
niques investigated were electrochemical current noise measurement (ECNM), zero resistance
ammetry (ZRA), and electrochemical potential noise measurement (EPNM). Several applica-
tions, including monitoring localized and general corrosion of steam generator tubes and other
material combinations, with and without crevices, are discussed. Future developments and their
potential impact on plant reliability, integrity, and safety programs are reviewed.
KE Y WOR D S : Alloy 690, Alloy 600, steam generator, resistance noise, current noise, potential
noise, degree of localization, nuclear power, crevice corrosion
A coor di nat ed r esear ch pr ogr a m was i ni t i at ed t o expand our under s t andi ng o f st eam gen-
er at or ( SG) mat er i al s ' el ect r ochemi cal behavi or . The first phase o f t hi s pr ogr a m f ocus ed on
demons t r at i ng t he usef ul ness o f el ect r ochemi cal noi se me a s ur e me nt t echni ques in eval uat i ng
t he cor r os i on behavi or o f SG t ubi ng al l oys. In addi t i on, envi r onment s t hat woul d accel er at e
t he SG t ube cor r os i on pr oces s es whi l e al l owi ng t est i ng to be conduct ed in a r easonabl e
per i od o f t i me wer e r evi ewed.
Cor r os i on moni t or i ng o f Al l oy 690 and Al l oy 600 t est s peci mens was conduct ed usi ng a
mul t i t echni que appr oach t hat uses a c ombi na t i on of e l e c t r oc he mi c a l met hods. A mul t i t ech-
Supervisor of Chemistry and Metallurgy and manager of Nuclear Technology, respectively, Babcock
& Wilcox International, 581 Coronation Blvd., Cambridge, ON, Canada N1R 5V3.
z Research engineer and research chemist, respectively, Babcock & Wilcox Research and Development
Division, 1562 Beeson St., Alliance, OH 44601.
Copyrighl 9 1996 by ASTM International
2 8 8
www.astm.org
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
DOHERTY ET AL. ON EVALUATION OF MATERIALS 289
nique corrosion monitoring system was connect ed directly to stressed and unstressed Alloy
690 and 600 SG tube C-ring specimens in various concentrations of deaerated sodium hy-
droxide (NaOH) at 300~ (0.01M NaOH, 0.5M NaOH, and 5M NaOH). Creviced Alloy 690
SG tube samples were also evaluated in 5M NaOH solution.
Small, spontaneously generated current and voltage signals originating from electrochem-
ical corrosion processes were recorded on a multiplexed multichannel data logger. The ab-
sence of external polarization eliminates the possibility of perturbing the natural corrosion
process. The data can provide information on the mechani sm and rate of corrosion occurring
on the Al l oy 690 and Al l oy 600 C-ring specimens. The nonperturbative electrochemical
potential and current noise monitoring techniques therefore can be appropriate for the eval-
uation of in-service corrosion, for example, during operation or chemical cleaning processes.
Muititechnique Corrosion Monitoring System
The multitechnique corrosion moni t or used a combination of four monitoring techniques
to evaluate the corrosion behavior of the Alloy 690 and Al l oy 600 SG tubes. These tech-
niques are described below:
1. Potential Measurement--Platinum wire was used as a reference electrode in this study
to give indications of overall potential change. A high-temperature reference electrode
was not used during the test program because the test temperature was too high (300~
for the electrodes currently used. Other electrodes for higher temperatures are being
evaluated. An Alloy 690 coupon is used as the worki ng electrode in this case.
2. Zero Resistance Ammetry ( ZRA) - - ZRA is normally used to determine the galvanic
current between two dissimilar electrodes, but may be effectively used to determine
the coupling current between two nominally identical electrodes.
3. Electrochemical Potential Noise Measurement ( EPNM) - - The term electrochemical po-
tential noise (EPN) is used to describe fluctuations in potential measured during a
corrosion. In this case, EPN was measured by monitoring the potential between two
nearly identical specimens. It may be possible to use EPN data (as well as ECN data)
to identify the mechani sm of corrosion attack [8,9]. For instance, the signatures of
pitting corrosion and stress corrosion cracking have been identified.
4. Electrochemical Current Noise Measurement (ECNM)--Electrochemical current noise
(ECN) can be described as fluctuations of current during the corrosion process. In this
case, the current flow between two nominally identical SG tube electrodes was recorded
and analyzed. The overall corrosion rate may be obtained from the ECN moni t or output
using a calibrated system and controlled environments.
Experimental Procedure
Tests were performed on Alloy 600 SG tube and Al l oy 690 SG tube specimens from two
different tube vendors (A and B). C-ring specimens were machined from lengths of SG tubes
for testing. The chemical compositions of the three alloys used are provided in Table 1.
Three specimens were machined from each alloy to enable the ECN and EPN measurements
to be taken from a three-specimen probe configuration (Fig. 1). One C-ring specimen from
each set of specimens was fully loaded (until the legs touched). No quantitative measurement
of strain was performed, since it was not a test variable and the strain level was above yield.
These specimens were electrically isolated from the loading nuts and bolts, as shown in Fig.
2. No replicates were used during this initial test program. The bolt threads were removed
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
290 ELECTROCHEMI CAL NOI SE MEASUREMENT
TABLE 1--Chemi cal composition of test alloys.
Species Concentration
Alloy C Mn S P Si Cr Ni AI Ti Fe N 2
A 690 0.017 0.21 0.000 3 0.007 0.15 30.26 59.36 0.41 0.28 9.26 0.0086
B 690 0.021 0.23 0.000 3 0.007 0.10 26.81 59.81 0.06 0.22 9.56 0.0070
600 0.048 0.31 0.005 0.009 0.10 15.5 73.0 . . . 0.24 8.2 0.018
in t he regi on in whi ch the speci men woul d nor mal l y cont act the speci men and a TFE-
fl uorocarbon sl eeve was pl aced in this regi on to i sol at e t he speci men f r om the bol t . Cer ami c
washers were pl aced bet ween the speci men and the l oadi ng nut.
A cr evi ced C- r i ng speci men was added to t he t est mat ri x in the 5M NaOH t est solution.
The speci men was const ruct ed of t wo C- r i ng speci mens nest ed together. The mat eri al used
for t he cr evi ced C- r i ng speci mens was Vendor B Al l oy 690 tubing. No at t empt was made
to i sol at e t he nest ed C-ri ngs from one another. The C-ri ngs were i sol at ed from t he l oadi ng
bol t and nuts as descr i bed above. The entire assembl y was used as a t est speci men in this
experi ment .
Tests were per f or med in a 7. 57-L (2-gal ) Hast el l oy | al l oy C-276 aut ocl ave. Before i ni t i al
testing, t he aut ocl ave was mechani cal l y cl eaned and flushed wi t h dei oni zed water. The t est
speci mens were mount ed in the aut ocl ave shown in Fig. 3. The speci mens in t he t hree
speci men pr obe were l ocat ed at the same l evel in the aut ocl ave. In addi t i on, a pl at i num wi r e
and an Al l oy 690 coupon were mount ed in the aut ocl ave head to gi ve an i ndi cat i on of
pot ent i al . Three test sol ut i ons were used duri ng t he t est program: 0. 01M NaOH, 0. 5M NaOH,
and 5M NaOH. Af t er the t est sol ut i on was pl aced in t he aut ocl ave and t he head seal ed in
pl ace, t he sol ut i on was deaerat ed overni ght and t he aut ocl ave was heat ed to 300~
Computer
Current Potential
Anal og t o Digital Anal og t o Digital
Converter Converter
ZRA
I
F
J
FIG. 1--Three-specimen probe configuration.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
DOHERTY ET AL. ON EVALUATION OF MATERIALS 291
~ A
,p
- E
J _
Oo,a~ L
Washer ~ ~
i
D
Inconel
Wa s he r
/o
5
Al l oy- 690 Al l oy- 600
A 0 . 6 8 7 5 " A 0. 625"
B 0 . 3 4 3 8 " B 0 . 3 1 2 5 "
C 0 . 6 8 7 5 " C 0 . 6 2 5 "
D 0 . 2 5 0 " D 0 . 2 5 0 "
E 1. 750" E 1. 750"
HG. 2--Drawing of C-ring specimen used showing ceramic isolation washers.
Dat a were obt ai ned usi ng a CML DENNI S 9 (Version 4. 5) moni t or i ng syst em and a com-
put er was used to l og data. ECN and EPN dat a were t aken on each speci men pr obe once
every 8 h.
R e s u l t s
The results for the t hree tests are summar i zed below. In addi t i on, the st andard devi at i ons
of the ECN dat a are al so given. An i ncrease in the st andard devi at i on of t he ECN dat a (cri)
i ndi cat es l ocal i zed events. The resi st ance noi se (Ro) is cal cul at ed by tri/tr v (where ~r v is the
st andard devi at i on of the EPN data). R, provi des i nformat i on on the general corrosi on be-
havi or of the mat eri al s t est ed and is i nversel y pr opor t i onal to t he corrosi on rate. The degree
of l ocal i zat i on for the t est mat eri al s in each test envi ronment is al so provi ded. Thi s is cal -
cul at ed by ~i/IRMs, where IRMS is t he r oot mean square of t he current data. The degree of
l ocal i zat i on provi des some i ndi cat i on of t he occurrence of l ocal i zed events, al t hough this
may be mi sl eadi ng. However, when a l ocal i zed event occurs, t he val ue drifts t owards 1 [4].
Test 1 was conduct ed in 0. 01M NaOH sol ut i on usi ng Vendor A Al l oy 690, Vendor B
Al l oy 690, and Al l oy 600 C-ri ng speci mens. Tests were carri ed out for 320 h. ECN or EPN
dat a i ndi cat ed that no cracki ng had occurred. Typi cal ECN and EPN results are shown in
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
292 ELECTROCHEMICAL NOISE MEASUREMENT
FIG. 3--Test specimens mounted in autoclave head.
Fig. 4. The measured coupling currents and potentials were low as shown in Fig. 5. The
standard deviations for ECN (~ri) were the smallest for Vendor A Alloy 690 and largest for
Alloy 600 (Fig. 6). There was no increase in ~r i, indicating that no significant activity had
occurred for any of the sample alloys. These alloys demonstrated relatively low levels of
general corrosion, as indicated by the relatively high resistance noise (R, ) as shown in Fig.
7. Vendor B Alloy 690 tubing had the highest noise resistance followed by Vendor A Alloy
690 tubing and then Alloy 600. The degree of localization was approximately one for all
materials, indicating a localized process [4], perhaps some grain boundary dissolution or
pitting.
Test 2 was conducted in 0.5M NaOH solution using the same specimens previously de-
scribed. These tests were terminated after approximately 100 h without indication of crack-
ing. The test was stopped to apply a more aggressive environment. Typical ECN and EPN
scans for this environment were not significantly different in appearance than those for the
0.01M NaOH, as shown in Fig. 3. Corrosion activity was general in nature, and there was
no indication of cracking in the electrochemical noise data taken. The measured coupling
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
DOHERTY ET AL. ON EVALUATION OF MATERIALS 2 9 3
2. 0E- 06
1. 0E- 06
Z O.OE+O0
O
uJ
- 1. 0E- 06
-2. OE-06
- 4. 0E- 04
--L---- J_
500 1000 1500
Time (secondS)
2000
- 5. 0E- 04
Z - 6. 0E- 04
O.
tr
- 7. 0E- 04
~. OE- 04 ~ r j
0 500 1000 1500
~me (seconds)
2000
FIG. 4a--Typical ECN and EPN data for specimens in O.O1M NaOH (Specimen A).
current ( Fi g. 8) wa s hi gher than t hos e meas ured in 0 . 0 1 M Na OH sol ut i on, cr i data f or al l
mat eri al s we r e al s o hi gher than t hos e meas ured i n the 0 . 0 1 M Na OH sol ut i on ( Fi g. 9). Al l
the al l oys had a l o we r R , and, f r om this, it was i nferred that t hey al s o had a hi gher corros i on
rate in thi s t est envi ronment as demons t rat ed by the l o we r R , (Fi g. 10). The res i s t ance noi s e
on Vendor B Al l o y 6 9 0 wa s hi gher than the Vendor A Al l o y 690. Al l o y 6 0 0 had the l o we s t
res i s t ance i ndi cat i ng the hi ghes t corros i on rate. The degree o f l oc al i z at i on decreas ed i n al l
cas es to approxi mat el y 0. 4, i ndi cat i ng that general corros i on wa s occurri ng. There was no
cl ear i ndi cat i on o f l o c a l i z e d attack. The t est wa s st opped and rei ni ti ated wi t h a mor e ag-
gr e s s i ve t est envi ronment .
Test 3 wa s carded out in a 5M Na OH s ol ut i on for approxi mat el y 500 h. In thi s test, a
crevi ced Al l o y 690 C- ri ng s pe c i me n was added t o the t est mat ri x. Typi cal ECN and EPN
scans for this envi ronment we r e not si gni f i cant l y di fferent in appearance than t hos e f or the
0 . 0 1 M Na OH as s hown in Fi g. 4. Duri ng t est i ng (after approxi mat el y 260 h), there was a
sudden change in the e l e c t r oc he mi c al outputs. Af t er the t est was c o mpl e t e , the s pe c i me ns
we r e e xami ne d, and it wa s f ound that areas o f the TFE- f l uorocarbon coat i ng on the ni c ke l
wi re l eads had cracked. Thi s i s the l i ke l y reason for the sudden c hange i n e l e c t r oc he mi c al
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
2 9 4 ELECTROCHEMICAL NOISE MEASUREMENT
2 . 0 E - 0 6 i ~ I
1 . 0 E - 0 6 ~
~ f q " " " i ' 1 " r l ' ' r r l F " I " ' " r , " w t r I T 1
- 2 . 0 E - 0 6 ~ ' *
0
- 8 . 0 E - 0 4
5 0 0 1 0 0 0 1 5 0 0
" l i m e ( s e c o n d s )
2 0 0 0
> ~ - 9 . 0 E - 0 4
Z - 1 . 0 E - 0 3
O .
" ' t
- 1 . 1 E - 0 3
- 1 . 2 E - 0 3 J '
0 5 0 0 1 5 0 0
i
1000
T i m e ( s e c o n d s )
2000
FIG. 4b--Typical ECN and EPN data for specimens in O.O1M NaOH (Specimen B).
readings. The data obtained during the first 250 h of exposure compared with data from
Tests 1 and 2.
The mean coupling current and potential standard deviation for the uncreviced Vendor A
Al l oy 690 test material are shown in Fig. 11. The values did not significantly change between
the 0. 5M NaOH test solution and the 5M test solution. The measured coupling potential and
potential standard deviation for the other test materials was higher (the Alloy 600 specimens
coupling potential changed sign). The coupling current and the current standard deviation
(Fig. 12) was higher in all cases. The resistance was lower in all test materials, indicating a
higher corrosion rate (Fig. 13). The degree of localization, also shown in Fig. 13, was also
lower. This indicated that general corrosion was occurring [4].
The creviced Vendor B Alloy 690 specimen demonstrated significant differences in the
measured coupling potential and potential standard deviation from the uncreviced Vendor A
Al l oy 690 specimen. The creviced Alloy 690 specimen current standard deviation was sig-
nificantly higher than the uncreviced specimens. There was no significant difference between
the coupl i ng current measured in the creviced and uncreviced Al l oy 690 specimens (Fig.
l l ) .
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
DOHERTY El" AL. ON EVALUATION OF MATERIALS 295
2 . o E 6 l t l i
i I I a
A
<
- 1 . 0 E- 0 6
- 2 . 0 E* 0 6 i i i
0 5 0 0 1 0 0 0 1 5 0 0 2 0 0 0
T i me ( s e c o n d s )
O,OE+O0
- 1 . 0 E- 0 4
A
>
Z - 2 . 0 E- 0 4
EL
UJ
- 3 . 0 E - O 4
- 4 . 0 E - 0 4 ~ r
0 5 0 0 1 0 0 0 1500 2 0 0 0
T i me ( s e c o n d s )
FIG. 4c--Typical ECN and EPN data for specimens in O.O1M NaOH (Alloy 600).
This degree of localization, shown on Fig. 13, was also significantly higher for the creviced
specimens (0.88 versus 0.28), indicating localized attack in the creviced specimen [4].
After testing was completed, the specimens were examined. There was no significant
evidence of localized attack in the creviced specimen. It would appear that the localized
corrosion events were masked by high levels of general corrosion. Figure 14 shows the
micrographs of each specimen.
Di s c us s i on
The main objective of this test program was to demonstrate that electrochemical noise
monitoring (ENM) techniques could be used to monitor the increased corrosiveness of an
environment and the different corrosion processes occurring on test materials in a nonper-
turbative manner. These goals were achieved. Differences in solution aggressiveness can be
clearly seen by compari ng R, values, as shown in Figs. 15 t hrough 17. The resistance noise
of all specimens decreased with increasing NaOH concentration, indicating an increase in
corrosion rates. The resistance noise of the alloys in the 0. 5M NaOH solution decreased with
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
O.OE+O _~
I.-
Z
LLI
0
,m
2. 0E- 7 - - 4. 0E- 3
2. 0E- 3
~ - 2. 0E- 7
0 B (cu~ent)
~2,OE~ 3
~7 600 (current)
/ --~-- A (potential)
B (potential)
600 (potent=al)
- 4. 0E- 7 : ~ i ' - 4. 0E- 3
0. 00 200. 00 400. 00
T I ME ( h o u r s )
FIG. 5 - - Me a n coupling current and pot ent i al f or test specimens in O.OIM NaOH.
5. 00E- 7 - -
Vendor A Alloy 690
( ~ Ve ndor B Alloy 690
~z Alloy 600
"~" 4. 00E- 7
Q.
E
Z
o_
I. -
~ 3. 00E- 7
I.U
121
121
n-
<
Z
CO 2 . 0 0 E- 7
1. 00E- 7
296 E L E C T R OC H E MI C A L N OI S E ME A S U R E ME N T
r 1 1 f r 1 T [ T - - - -
0. 00 100o00 200. 00 300. 00 4 0 0 . 0 0 500. 00
T I ME ( h o u r s )
FIG. 6- - St andar d deviation of current noise f or test specimens in O.O1M NaOH.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
DOHERTY ET AL. ON EVALUATI ON OF MATERI ALS 297
FIG. 7--Resistance noise and degree of localization f or test specimens in O.01M NaOH.
time approaching the 5M NaOH R, values, indicating that the corrosion rates in 0.5M and
5M NaOH may not be that different, that is, high general corrosion rates.
Differences in alloy composition could also be seen in all environments. This could be
most clearly seen by examining the standard deviation of the ECN data for all environments.
In all cases, the ECN standard deviation was significantly higher for Alloy 600 than for
Alloy 690. This was expected, since Alloy 690 has a higher chromium content than Alloy
600.
The creviced Vendor B Alloy 690 tube specimen demonstrated significantly different be-
havior to the uncreviced Alloy 690 specimens. The degree of localization was higher, indi-
E
I.-
Z
ILl
r',,"
n,"
C3
r
6.0E-5
4.0E-5
2.0E-5
0.0E+0
8.0E-4
I
~ 4.0E-4
O. 0E+O --"~
I --
I z
i w
L
-4.01::-4
i
1
!
-2.0E-5 I I -8.0E-4
I I i '
0.00 40.00 80.00
TI ME (hours)
FIG. B--Mean coupling current and potential f or test specimens in 0.5M NaOH.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
298 ELECTROCHEMICAL NOISE MEASUREMENT
2.50E-5
O.
E
z
0
<
iii
C~
ra
r
<
t:)
z
<
Vendor A Alloy 690
C) Vendor IB Alloy 690
2.00E-5
1.50E-5
1.00E-5
5.00E-6
O.OOE+O ' ) ' I i I ' ) '
0.00 20.00 40.00 60.00 80.00 100.00
TIME (hours)
FIG. 9mSt andard deviation of current noise f or test specimens in 0.5M NaOH.
FIG. lO--Resistance noise and degree of localization f or test specimens in 0. 5M NaOH
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
DOHERTY ET AL. ON EVALUATION OF MATERIALS 299
E
j-.
z
uJ
n~
r~
0
3. 0E- 3
5 M NaOH
A (r
e (cu,ent~
600 (current)
A A ~ ~
t r e y ( p a t )
4.0E-4
0. 0E+0
v
-4. 0E-4 I j i ] r [ -1. 0E-3
0.00 100.00 200.00 300.00
T I M E ( h o u r s )
FIG. 11--Mean coupling current and potential f or test specimens in 5M NaOH.
2.0E-3
I .OE-3
O.OE+O
.,..,
O
v
. . I
<
F-
Z
u.I
I--
O
O_
cating that, as expected, there was localized corrosion in the creviced region. However,
further work is required to identify localized corrosion activity or to demonstrate that general
attack was greater than the local activity. The calculated resistance of the creviced Vendor
B Al l oy 690 specimen was not significantly lower than that of the uncreviced Speci men A.
The lack of evidence of localized attack is probabl y a function of relatively short test times.
Anot her objective of this test program was t o identify a test environment that would
produce cracking of Al l oy 690 within a reasonable period of time and to identify the initi-
ation of cracking using ENM. This has not been achieved, as yet, but work will continue t o
identify an opt i mum chemistry. It appears that general attack was high in Tests 2 and 3.
Future work will focus on: (1) using 0. 01M NaOH, (2) incorporating a SCC cracki ng agent
(for example, a lead compound) into the test solution, and (3) applying potential chemically
or electrochemically to a loaded specimen. Once this environment has been identified, testing
with mockups of specific steam generator component s will be initiated.
Co n c l u s i o n s
1. The corrosion behavior of Al l oy 690 steam generator tube material exposed to various
sodium hydroxi de environments was successfully monitored using online electrochem-
ical techniques. Activity was identified as being general and not severe f or the moni-
toring period.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
300 ELECTROCHEMI CAL NOI SE MEASUREMENT
1.00E.-4
0.
E
z
O
I -
<
UJ
C3
C3
0E
<
o
z
<
I -
o3
8.00E-5
6.00E-5
4.00E-5
2.00E-5
<~ Vendor A Alloy 690
C) Vendor B Alloy 690
[ ] Alloy 600
~)~ Creviced Vendor B Alloy 690
O.OOE+O I ' I ' I ' I '
0.00 100.00 200.00 300.00 400.00
TIME (hours)
FIG. 12--Standard deviation of current noise f or test specimens in 5M NaOIt.
FIG. 13--Resistance noise and degree of localization f or test specimens in 5M NaOH.
Copyright by ASTM Int'l (all rights reserved); Fri Nov 14 15:05:54 EST 2008
Downloaded/printed by
Sergio Serna () pursuant to License Agreement. No further reproductions authorized.
DOHERTY ET AL. ON EVALUATION OF MATERIALS 301
FIG. 14--Specimen micrographs taken at the apex of the C-ring specimen at a magnification of x lO0.
Upper left is Vendor A, upper right is Vendor B, lower left is Alloy 600, and lower