You are on page 1of 7

Alexandria Engineering Journal (2014) 53, 977–983

H O S T E D BY
Alexandria University

Alexandria Engineering Journal


www.elsevier.com/locate/aej
www.sciencedirect.com

SHORT COMMUNICATION

Modeling of polarization curves not exhibiting


a Tafel region using Excel spreadsheets
Marco Alfaro

Departamento de Quı´mica, Campus Andre´s Bello, Facultad de Ciencias, Universidad de La Serena, Colina el Pino S/N, La
Serena, Chile

Received 6 December 2013; revised 29 June 2014; accepted 14 July 2014


Available online 4 August 2014

KEYWORDS Abstract This paper describes the modeling of polarization curves not exhibiting a Tafel region
Lineal voltammetry; using Excel spreadsheets. Carbon steel in aerated NaCl unbuffered solutions was studied applying
Steel corrosion rate; a linear potential sweep technique. Current–potential curves were obtained from linear potential
Sodium chloride solution; sweep at a rate of 10 mV s1 in un-buffered 0.1, 0.4 and 0.9 M NaCl solutions, at a constant
Excel spreadsheet temperature of 25 C and potential sweeps that were conducted in a potential range 1.0 to
0.4 V vs. Ag/AgCl.
ª 2014 Production and hosting by Elsevier B.V. on behalf of Faculty of Engineering, Alexandria
University.

1. Introduction allow the determination of corrosion rate from nonlinear


kinetic models have been reported, but the fit of these models
The determination of the corrosion rate of steel is generally to the experimental data requires the development of computa-
conducted by using the extrapolation Tafel method from a tional programs [9,10]. An alternative method is by using the
potentiodynamic polarization curve of the metal in aqueous solver function in Excel for the fitting of nonlinear functions
solution [1–4], but this method requires the preference that along with the nonlinear least squares method, which
the anodic and cathodic branches present Tafelian behavior minimizes the sum of the squared differences between the
[5]. Another method used to determine the corrosion rate from experimental values and the predicted by the model, and it
polarization curves with exclusive activation control for both has already been successfully used in electrochemical
anodic and cathodic branches is the adjustment of the Wagner kinetics [11]. The objective of this work was modeling of polar-
and Traud equation to the curve using a nonlinear least ization curves not exhibiting a Tafel region using Excel
squares method [6,7]. Much modern equipment give the corro- spreadsheets.
sion rate by using this method but do not provide any informa-
2. Materials and methods
tion about the goodness of fitting the experimental data to the
polarization curve in case where clearly the curves do not
present Tafel region [8]. Moreover, several investigations that 2.1. Experimental

E-mail address: malfarog@userena.cl All experiments were conducted in a conventional three


Peer review under responsibility of Faculty of Engineering, Alexandria electrode cell, where the working electrode was made of carbon
University. steel with a 4 mm diameter and an exposed area of 0.5027 cm2,
http://dx.doi.org/10.1016/j.aej.2014.07.005
1110-0168 ª 2014 Production and hosting by Elsevier B.V. on behalf of Faculty of Engineering, Alexandria University.
978 M. Alfaro

a platinum wire as a counter electrode and Ag/AgCl in 1 M the corrosion potential, obtained directly from the experimen-
KCl cell as a reference electrode. The composition of carbon tal polarization curve and electrochemical parameters of the
steel was SAE 1008 with chemical content wt% 98.5 Fe, 0.02 cathodic reaction for the reduction of oxygen under zero total
C, 0.15 Mn and traces of P, S, Si and Cr. The cell was current (it = 0) condition. The corrosion rate, it can be
immersed in a thermostatic Haake D1 bath operating at calculated using the expression given by Khaled and Amin
25 C. Measurements were conducted in water before air was [8] Eq. (4).
bubbled for about 15 min; air was bubbled into the water by CR ¼ 3; 28 icorr ðM=ndÞ ð4Þ
using a small fish-tank pump, this water was used to prepare
the test solutions. A PG 580 (Princeton Instrument) potentio- where CR is the corrosion rate (mm y1), M is the atomic
stat was employed with the software of the same company. weight of iron (55.85), n is the number of electrons transferred
Polarization curves were obtained from linear potential sweep in the corrosion reaction (2 in this case) and d is the density of
at a rate of 10 mV s1. Potential sweeps were conducted in the iron (7,88 g cm3).
potential range 700 to 100 mV/SHE in 0.1, 0.4 and 0.9 M
NaCl un-buffered solution. Unbuffered solutions were used 2.3. Adjustment of the kinetic model to the experimental data
to prevent adsorption of the buffer compounds and their com-
plexes on the surface of metal [12], because the Caceres model In this case in order to adjust Caceres model [13] to the exper-
does not evaluate compound adsorption effect. imental polarization curve a nonlinear least squares method
was used. This is an iterative method that estimates the mini-
2.2. Kinetic model mum of the function (SS), Eq. (5), via Excel solver function
[15].
The kinetic model chosen for this study was developed by and X N     !2
i i
Caceres et al. [13], which uses the superposition model SS ¼ wi  ð5Þ
proposed by Wagner and Traud in 1938 [14], method that con- i¼1
A exp A pred

siders that the total current density (it) is the algebraic sum of
where (i/A)exp is the experimental current density, (i/A)fit is
the anodic (ia) and cathodic (ic) current, which is:
the current density predicted by the model and wi is a weight
h  i
io;O2 exp 2:3
EEeq;O2 function, designed to improve the fit, which in this case is
tc
ic ¼   h  i ð1Þ defined by Eq. (6), as [16].
i 2 EEeq;O2
1 þ io;Ol;O2
exp 2:3 tc 1
   wi ¼ ð6Þ
E  Eeq;Fe s2
ia ¼ io;Fe exp 2:3 ð2Þ
ta where s is the standard deviation of the data predicted, defined
h i
EE
io;O2 exp 2:3ð tceq;O2 Þ    by the Eq. (7) as:
E  Eeq;Fe
it ¼   h  i þ io;Fe exp 2:3 ð3Þ vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
i 2 EEeq;O2 ta uX
1 þ io;O exp 2:3 tc u N  i
i
2
l;O
2
u 
t A Average;fit A i;fit
where iL,O2 (A cm2) is the limiting current density for oxygen s¼ i¼1
ð7Þ
reduction, io,O2 (A cm2) is the exchanged current density for N1
oxygen, Eeq,O2 (V) is the potential balance for oxygen reduc- where (i/A)Average, fit is the average of the predicted current
tion, tc, is the pseudo Tafel slope for oxygen reduction,. io,Fe, density, (i/A)I, fit is the current density of each point and N is
is the exchange current density for iron, Eeq,Fe is the potential the total number of points. The goodness of fit of the data to
balance for iron and ta is the pseudo anodic Tafel slope. Cor- the model will be evaluated by using the regression coefficient
rosion intensity (icorr) can be calculated from Eq. (3) by using (R2) [15]. The sum of squared errors (SSE), the root mean

Table 1 Excel cell, formula and Excel programming.


Excel cell Formula Excel programming
B2 SHE = Ag/AgCl + 0,237 =A2 + 0,237
D2 i = I/A =C2/0.5027
E2 |i| =ABS(D2)
F2 Log i   =LOG(E2)
EEeqO2
G2 ¼ i0O2 exp 2:3 tc =$M$1 * (EXP(2,3 * (B2  0,762)/$M$2))
   
i 2 EEeqO2
H2 ¼ 1 þ i0O lO2
exp 2:3 tc =1 + (($M$1/$M$3) * (EXP(2,3 * (B2  0,762)/$M$2)))
EEeqO

i0O exp 2:3 2


I2 ¼ 2  t
c
EEeqO
=G2/H2
i0 O2 2
1þ exp 2:3
ilO
2  tc 
EEeq;Fe
J2 ¼ io;Fe exp 2:3 ta =$M$4 * (EXP(2,3 * (B2+0,614)/$M$5))
EEeq;O
 
io;O2 exp 2:3 2
EEeq;Fe
K2 ¼   c
EE
t

þ io;Fe exp 2:3 ta =G2 + H2
io;O eq;O
1þ 2 exp 2:3 2
il;O tc
2
P2 =((K2  D2)^2)/($K$94^2)
K94 =STDEVA(K2:K93)
Modeling of polarization curves not exhibiting a Tafel region 979
2 3
square error (RMSE) and chi-square (v2) [17], which compares XN     !2 !=2
1 i i
the values of experimental current density (i/A)exp with those RMSE ¼ 4  5 ð10Þ
predicted by the model (i/A)pred and R2 that include the N i¼1
A exp A fit
experimental average current density (i/A)ave,exp.
N 
X i
i
2
N 
X i
i
2 A exp
 A fit
 i¼1
A exp A fit v2 ¼ ð11Þ
R2 ¼ 1  i¼1
ð8Þ Nz
N 
X 2
i
i

A exp
 A ave; exp
i¼1 2.4. Construction of the Excel spreadsheet

    !2
1X N
i i To construct the Excel spreadsheet, the indications given in
SSE ¼  ð9Þ Table 1 and Figs. 1–3 must be followed. Once the spreadsheet
N i¼1 A exp A fit
is built, the solver function of Excel has to be installed by

Figure 1 Excel spreadsheet showing the cells with their contents A, B, C, D, E y F.

Figure 2 Excel spreadsheet showing the cells with their contents G, H, I, J y K.


980 M. Alfaro

Figure 3 Excel spreadsheet showing the cells with their contents L, M, N, O, P y K.

following the procedure described by Brow [15]. Once 2.5. Sensitivity analysis of corrosion rate prediction Caceres
installed, its mode of execution is as follows: model

1. The option Data has to be clicked and then a dialog Parameters considered in the sensibility analysis for the corro-
box of the solver function of Excel should appear as sion rate prediction Caceres model are listed in Table 3. The
follows. sensibility parameter, S is given by:

2. The set target cell has to be marked, in this case the cell ½ðO2  O1 Þ=O12 
S¼ ð12Þ
marked is $M$7. ½ðI2  I1 Þ=I12 
3. Then Min has to be marked.
where I1 and I2 are the least and greatest values of input used,
4. After marking Min the cells that are marked are
respectively. I12 is the average of I1 and I2. O1 and O2 are the
$M$1:$M$5, these are the cells of the kinetic constants of
output for the two input values, and O12 is the average of the
Caceres model (Eq. (3)) and the ones that solver will change
two outputs [18,19].
to minimize the function SS (Eq. (5)).
5. Finally with this data, the spreadsheet shown in Fig. 4 is
3. Results and discussions
created, and replacing the experimental corrosion potential
value therein, icorr is obtained for the condition where the
total current is approximately zero, i.e. ifit = 0.001 and with Figs. 5–7 show the correlation of polarization curves of carbon
this the corrosion rate. steel in un-buffered 0.1, 0.4 and 0.9 M NaCl solutions, at a
Modeling of polarization curves not exhibiting a Tafel region 981

Figure 4 Excel spreadsheet showing the calculation of the corrosion rate from experimental corrosion potential, the electrochemical
parameters and the condition that the total current density is approximately zero (it = 0.001).

-0,1
-0,1
Experimental
-0,2 Caceres Model Experimental
-0,2
Caceres Model
-0,3
-0,3
E (V vs. SHE)

-0,4
E (V vs. SHE)

-0,4

-0,5 -0,5

-0,6 -0,6

-0,7 -0,7

-0,8 -0,8

-5,0 -4,5 -4,0 -3,5 -3,0 -2,5 -5,0 -4,5 -4,0 -3,5 -3,0 -2,5
-2 -2
Log (i/A) A cm Log (i/A) A cm

Figure 5 Experimental polarization curve (X) for carbon steel in Figure 6 Experimental polarization curve (X) for carbon steel in
aerated 0.1 M NaCl solution un-buffered and fitted curve (con- aerated 0.4 M NaCl solution un-buffered and fitted curve (con-
tinuous line) calculated from the Caceres model using Excel tinuous line) calculated from the Caceres model using Excel
spreadsheet. spreadsheet.

constant temperature of 25 C, with the kinetic model; by


using excel spreadsheets to predict the total current density. 4. Conclusions
It can be appreciated that both, anodic and cathodic branches
of both curves do not present Tafel region and that the hydro- i. The experimental polarization curve was obtained from
gen evolution reaction is almost negligible. The Table 2 shows a stationary electrode of carbon steel in un-buffered 0.1,
the values of the electrochemical parameters, the corrosion 0.4 and 0.9 M NaCl solutions, at a constant temperature
rate and the statisticians to evaluate the quality of fit of the of 25 C.
kinetic model of Caceres. The Fig. 8 shows that a higher molar ii. The superposition Wagner and Traud model was used to
concentration of NaCl to a higher corrosion rate. The results obtain the total current as an algebraic sum of the ano-
from sensitivity index analysis are summarized in Table 3. dic and cathodic currents.
Limiting current density for oxygen reduction, with the aver- iii. The nonlinear least squares method and the solver func-
age index value of 1.65, is the critical input parameter in Các- tion of Excel were used to calculate the electrochemical
eres model. parameters of a kinetic model.
982 M. Alfaro

-0,1
Table 3 Caceres model parameter values used for sensibility
Experimental analysis and parameter S.
-0,2
Caceres Model Parameter Base value Range of test S
-0,3 0.1 M NaCl
io,O2 8.84 * 104 1.0 * 104–1.77 * 103 1.03
E (V vs. SHE)

-0,4 tc,O2 2.04 0.1–4.08 1.10


iL,O2 0.02 0.04–0.1 1.63

-0,5 0.4 M NaCl


io,O2 1.48 * 103 1.0 * 104–2.96 * 103 0.94
-0,6
tc,O2 4.32 0.9–8.64 0.94
iL,O2 0.6 1.0 * 103–1.2 1.73

-0,7 0.9 M NaCl


io,O2 9.53 * 104 1.0 * 104–1.91 * 103 1.00
tc,O2 2.00 0.1–4.00 0.96
-0,8
iL,O2 0.3 1.0 * 103–0.6 1.58
-4,5 -4,0 -3,5 -3,0 -2,5
-2
Log (i/A) A cm iv. An excel spreadsheet was developed that allowed the
determination of the corrosion rate from polarization
Figure 7 Experimental polarization curve (X) for carbon steel in curves that do not present Tafel region.
aerated 0.9 M NaCl solution un-buffered and fitted curve (con- v. The limiting current density for oxygen reduction is the
tinuous line) calculated from the Caceres model using Excel critical input parameter in Cáceres model.
spreadsheet.

Acknowledgments
Table 2 Electrochemical parameters, corrosion rate and
statistical parameters of fit of polarization curves. The generous support of the FAU (Universidad Friedrich-
Electrochemical parameters 0.1 M NaCl 0.4 M NaCl 0.9 M NaCl Alexander of Erlangen-Nuremberg, in Germany, Friedrich-
io,O2 (A cm2) 8.84 * 104 1.48 * 103 9.53 * 104 Alexander-Universität Erlangen-Nürnberg, shortened FAU)
tc,O2 (V dec1) 2.04 4.32 2.00 Germany. Project of Cooperation Universidad de Erlanger
iL,O2 (A cm2) 0.02 0.6 0.3 Universidad de La Serena, Chile is gratefully appreciated.
io,Fe (A cm2) 1.10 * 103 1.48 * 103 2.09 * 103 Thanks to Prof. Dr. Ulrich Nickel for all those encouraging
ta,Fe (V dec1) 0.96 1.23 1.14 discussions.
Ecorr (V) 0.283 0.403 0.359
icorr (A cm2) 2.41 * 103 2.74 * 103 3.41 * 103
CR (mm y1) 2.80 * 102 3.19 * 102 3.97 * 102
References

Statistical parameters [1] P. Raja, A. Qureshi, A. Rahim, H. Osman, K. Awang,


R2 (–) 0.9979 0.9972 0.9926 Neolamarckia cadamba alkaloids as eco-friendly corrosion
SSE (A cm2) 6.93 * 106 1.75 * 105 2.10 * 105 inhibitors for mild steel in 1 M HCl media, Corros. Sci. 69
RMSE (A cm2) 1.12 * 102 1.77 * 102 1.95 * 102 (2013) 292–301.
v2 (A cm2) 1.88 * 102 1.21 * 102 2.09 * 102 [2] M. Moradi, Z. Song, L. Yang, J. Jiang, J. He, Effect of marine
Pseudoalteromonas sp. on the microstructure and corrosion
behaviour of 2205 duplex stainless steel, Corros. Sci. 84 (2014)
103–112.
0,040 [3] K.R. Ansari, M.A. Quraishi, Singh Schiff’s base of pyridyl
substituted triazoles as new and effective corrosion inhibitors for
mild steel in hydrochloric acid solution, Corros. Sci. 79 (2014) 5–15.
0,038
[4] H. Ashassi-Sorkhabi, R. Bagheri, Sonoelectrochemical
synthesis, optimized by Taguchi method, and corrosion
0,036 behavior of polypyrrole-silicon nitride nanocomposite on St-12
steel, Synthetic Met. 195 (2014) 1–8.
CR (mm y )
-1

0,034 [5] E. McCafferty, Validation of corrosion rates measured by the


Tafel extrapolation method, Corros. Sci. 47 (2005) 3202–3215.
[6] A. Zhou, B. Xie, N. Xie, Comparison of polarization curve-and
0,032
electrochemical quartz crystal microbalance methods for
determination of copper corrosion rate, Corros. Sci. 42 (2000)
0,030 469–480.
[7] H. Ardelean, A. Seyeux, S. Zanna, F. Prima, I. Frateur, P.
0,028 Marcus, Corrosion processes of Mg–Y–Nd–Zr alloys in Na2SO4
electrolyte, Corros. Sci. 73 (2013) 196–207.
0,0 0,2 0,4 0,6 0,8 1,0 [8] K.F. Khaled, M. Amin, Corrosion monitoring of mild steel in
sulphuric acid solutions in presence of some thiazole derivatives
NaCl (M)
– molecular dynamics, chemical and electrochemical studies,
Figure 8 Effect of the NaCl concentration on the corrosion rate. Corros. Sci. 51 (2009) 1964–1975.
Modeling of polarization curves not exhibiting a Tafel region 983

[9] H. Flitt, D. Schweinsberg, Evaluation of corrosion rate from [15] M. Brown, A step-by-step guide to non-linear regression
polarisation curves not exhibiting a Tafel region, Corros. Sci. 47 analysis of experimental data using a Microsoft Excel
(2005) 3034–3052. spreadsheet, Comput. Meth. Prog. Biol. 65 (2001) 191–200.
[10] L. Cáceres, T. Vargas, L. Herrera, Determination of [16] D. Harris, Nonlinear least-squares curve fitting with Microsoft
electrochemical parameters and corrosion rate for carbon steel Excel solver, J. Chem. Educ. 75 (1998) 119–121.
in un-buffered sodium chloride solutions using a superposition [17] A. Vega-Galvez, E. Notte-Cuello, R. Lemus-Mondaca, L. Sura,
model, Corros. Sci. 49 (2007) 3168–3184. M. Miranda, Mathematical modelling of mass transfer during
[11] E. Howard, J. Cassidy, Analysis with microelectrodes using rehydration process of Aloe vera (Aloe barbadensis Miller),
Microsoft Excel solver, J. Chem. Educ. 77 (2000) 409–413. Food Bioprod. Process. 87 (2009) 254–260.
[12] M.A. Veloz, I. González, Electrochemical study of carbon steel [18] M.A. Nearing, L. Deer-Ascough, J.M. Laflen, Sensitivity
corrosion in buffered acetic acid solutions with chlorides and analysis of the WEPP hillslope profile erosion model, J. T.
H2S, Electrochim. Acta 48 (2002) 135–144. ASAE 33 (3) (1990) 839–849.
[13] L. Cáceres, T. Vargas, L. Herrera, Influence of pitting and iron [19] K. Siamphukdee, F. Collins, R. Zou, Sensitivity analysis of
oxide formation during corrosion of carbon steel in unbuffered corrosion rate prediction models utilized for reinforced concrete
NaCl solutions, Corros. Sci. 51 (2009) 971–978. affected by chloride, J. Mater. Eng. Perform. 22 (2013) 1530–
[14] C. Wagner, W. Traud, On the interpretation of corrosion 1540.
processes through the superposition of electrochemical partial
processes and on the potential of mixed electrodes, Corrosion 62
(2006) 844–855.