K L Mittal Silanes and Other Coupling Agents, Volume 3

Silanes and Other Coupling Agents, Volume 3
This page intentionally left blank

Silanes and
Other
CouplingAgents
VOLUME 3
Editor: K.L. Mittal
///VSP///
UTRECHT • BOSTON
2004
CRC Press
Taylor & Francis Group
6000 Broken Sound Parkway NW, Suite 300
Boca Raton, FL 33487-2742
© 2004 by Koninklijke Brill NV
CRC Press is an imprint of Taylor & Francis Group, an Informa business
No claim to original U.S. Government works

Version Date: 20120727
International Standard Book Number-13: 978-9-04-741402-5 (eBook - PDF)
This book contains information obtained from authentic and highly regarded sources. Reasonable
efforts have been made to publish reliable data and information, but the author and publisher cannot
assume responsibility for the validity of all materials or the consequences of their use. The authors and
publishers have attempted to trace the copyright holders of all material reproduced in this publication
and apologize to copyright holders if permission to publish in this form has not been obtained. If any
copyright material has not been acknowledged please write and let us know so we may rectify in any
future reprint.
Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced,
transmitted, or utilized in any form by any electronic, mechanical, or other means, now known or
hereafter invented, including photocopying, microfilming, and recording, or in any information stor-
age or retrieval system, without written permission from the publishers.
For permission to photocopy or use material electronically from this work, please access www.copy-
right.com (http://www.copyright.com/) or contact the Copyright Clearance Center, Inc. (CCC), 222
Rosewood Drive, Danvers, MA 01923, 978-750-8400. CCC is a not-for-profit organization that pro-
vides licenses and registration for a variety of users. For organizations that have been granted a pho-
tocopy license by the CCC, a separate system of payment has been arranged.
Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are
used only for identification and explanation without intent to infringe.
Visit the Taylor & Francis Web site at
http://www.taylorandfrancis.com
and the CRC Press Web site at
http://www.crcpress.com
CONTENTS
Preface vii
Part 1: Silane Coupling Agents
Sterically hindered silanes for waterborne systems: A model study

of silane hydrolysis
E. R. Pohl and A. Chaves 3
Silane oligomers: A class of their own

H.Mack 11
FTIR and XPS surface characterization of allyltrimethoxysilane and

3-methacryloxypropyltrimethoxysilane mixture adsorbed onto titanium
substrate
/. R Matinlinna, K. Laajalehto, L. V 7. Lassila, A. Yli-Urpo and R K. Vallittu 21
Effect of bridging group on the structure of bis-silane water-barrier films

G. Pan, H. Kim, M. S. Kent, J. Majewski and D. W. Schaefer 39
Amino- and bis-silane pre-treatments for adhesive bonding of aluminum

B. G Tilset, R Lapique, A, Bj0rgum and C. 7. Simensen 51
y-APS as a durability enhancer of elastomer-to-metal adhesive bonds

R, R Hamade 69
Silane coupling agents for silica-filled tire-tread compounds:

The link between chemistry and performance
W. K. Dierkes, L. A. E, M. Reuvekamp, A. J. W. ten Brinke
and 7. W. M. Noordermeer 89
Electrodeposition of organofuncational bis-silanes and their effectiveness

in prodiving corrosion resistance of metals
/. S. Gandhi and W. J. van Ooij 105
Silane coupling agents as corrosion fatigue inhibitors

A. Seth and W. J. van Ooij 119
Modified silane coatings as an alternative to chromates for corrosion

protection of aluminum alloys
V Palanivel and W. J. van Ooij 135
VI Contents
Musical instrument strings and corrosion: A comparative study

of aminosilane and benzotriazole surface treatments
A.A.Parker 161
Part 2: Other Coupling Agents / Adhesion Promoters
Cyclic azasilanes: Volatile coupling agents for nanotechnology

B. Arkles, Y Pan, G. L. Larson and D. H. Berry 179
Inorganic-organic hybrid polymers based on silanes for coating textile

substrates
T. Textor, D. Knittel, T. Bahners andE. Schollmeyer 193
Optimising the adhesion of glass fibres to an epoxy resin using plasma

copolymers
D. J. Marks and F. R. Jones 205
Organophosphate adsorption on metal oxide surfaces

M. J. Shepard, J. R. Comer, T. L. Young, J. 5. McNatt, M. P. Espe,
R. D. Ramsier, T. R. Robinson and L. Y. Nelson 225
Manufacture of resin-free fiberboards from wood fibers activated

with Fenton's reagent (H202/Fe2+)
P. Widsten and J. E. Laine 241
Silanes and Other Coupling Agents, Vol. 3, pp. vii-viii
Ed. K. L. Mittal
© VSP 2004
Preface
This book chronicles the proceedings of the Fourth International Symposium on

Silanes and Other Coupling Agents held under the aegis of MST Conferences, LLC
in Orlando, FL, June 11-13, 2003. The premier symposium on this topic was held
in honor of the 75th birthday of Dr. Edwin P. Plueddemann in Midland, MI, April
3-5, 1991 the proceedings of which were properly documented in a hard-bound
book [1]. The second symposium in this series (with a slightly different title) was
held under the auspices of C4 Technologies, Inc. in Newark, NJ, October 21-23,
1998 and the proceedings were also chronicled in a book form [2]. Apropos, it
should be recorded that the third symposium in this vein was organized by MST
Conferences, LLC in Newark, NJ, June 18-20, 2001 but, for a variety of reasons,
the proceedings of this event were not documented in the form of a hard-bound
book.
Silanes have been used for about half a century as coupling agents/adhesion pro-
moters to promote adhesion between dissimilar materials in a variety of situations,
e.g., coating technology, adhesive bonding, and reinforced composites. However,
recently silanes have found other applications, for example, as corrosion inhibitors.
Currently, there has been tremendous R&D activity in unravelling the mechanisms
by which silanes work as well as in devising new and improved silanes.
The technical program for this symposium was comprised of 22 papers reflecting
both overviews and original research contributions. These presentations discussed
a number of topics including new applications of silanes, and both fundamental and
applied aspects were accorded due coverage.
Now turning to this volume, it contains a total of 16 papers which were properly
peer reviewed, revised and edited. So this book is not a mere collection of papers,
rather represents the highest standard of publication. The book is divided into two
parts: 1. Silane Coupling Agents; and 2. Other Coupling Agents/Adhesion Pro-
moters. The topics covered include: sterically hindered silanes; silane hydrolysis;
silane oligomers; adsorption of silanes and their surface characterization; structure
of bis-silane water-barrier films; silanes for improving adhesive bonding of alu-
minum, elastomer-to-metal adhesive bonds, and adhesion in silica-filler tire-tread
compounds; electrodeposition of bis-silanes; silanes to provide corrosion resistance
and as corrosion fatigue inhibitors; silane and other treatments for musical instru-
ment strings; cyclic azasilanes as coupling agents for nanotechnology; hybrid poly-
mers based on silanes for coating textile fabrics; plasma copolymers as adhesion
promoters; organophosphate adsorption; and activation of wood fibers.
Vlll Preface
This volume and its predecessors [1,2] containing bountiful information should
serve as a reference source for the latest R&D activity in the arena of coupling
agents. Anyone interested or involved in promoting adhesion between dissimilar
materials for any application should find this volume of great use and value.
Acknowledgements
First, my sincere thanks are due to my colleague and friend, Dr. Robert H. Lacombe,
for taking care of myriad details entailed in organizing this symposium. Second, all
authors are thanked for their interest, enthusiasm and contribution without which
this book would not have been possible. Special thanks are conveyed to unsung
heroes (reviewers, working behind the scene) for their many valuable comments.
Last, but not least, my appreciation goes to the staff of VSP (publisher) for doing an
excellent job in producing this book.
K. L. Mittal
P.O. Box 1280
Hopewell Jet., NY 12533
REFERENCES
1. K. L. Mittal (Ed.), Silanes and Other Coupling Agents. VSP, Utrecht, The Netherlands (1992).
2. K. L. Mittal (Ed.), Silanes and Other Coupling Agents, Vol. 2. VSP, Utrecht, The Netherlands
(2000).
Parti
Silane Coupling Agents

Silanes and Other Coupling Agents, Vol 3, pp. 3-9
Ed. K. L. Mittal
© VSP 2004
Sterically hindered silanes for waterborne systems:

a model study of silane hydrolysis
E. R. POHL * and A. CHAVES

OSi Specialties, Crompton Corporation, 771 Old Saw Mill River Road, Tarrytown,
NY 10591-6716, USA
Abstract—The reactivity of silanes in waterborne systems was studied using model compounds.
Vinyltrialkoxysilanes were used to investigate the effects of the leaving group on the rates of silane
hydrolysis. The alkoxy groups were methoxy, ethoxy, 2-propoxy and 3-oxabutoxy. The acid-catalyzed
rates of hydrolysis for these silanes were measured under pseudo-first-order conditions in an aqueous
acetone solution at 22°C. Steric and polar characteristics of the alkoxy leaving group were found to
affect the hydrolysis rates.
Keywords: Hydrolysis kinetics; silane; acid-catalyzed; steric; polar; Taft plot.
1. INTRODUCTION
Alkoxysilanes are used commonly as crosslinkers and adhesion promoters for
numerous coating, adhesive and sealant formulations. They are often chosen
because of their chemical reactivity, formulation flexibility and environmental
stability. The alkoxysilyl group hydrolyzes to form silanols that condense to form
crosslinks or react with surfaces to form chemical bonds. The condensation is
initiated by exposure to moisture and catalysts.
The reactivity of the methoxysilyl group with water has limited the use of many
alkoxysilanes in waterborne formulations. The methoxysilyl group prematurely hy-
drolyzes and condenses. These reactions reduce the shelf-life of the formulations to
only days or weeks and thereby make one-component waterborne formulations im-
practical. The current trend in the marketplace is towards water-based formulations
because they comply with regulatory mandates and meet consumers' environmental
and use requirements.
Silanes that contain sterically hindered alkoxysilyl groups have been successfully
used in one-component waterborne formulations [1-4]. The improved shelf-life of
*To whom correspondence should be addressed. Tel.: (1-914) 784-4911; Fax: (1-914) 784-4890;
e-mail: Eric.Pohl@gesm.ge.com
4 E. R. Pohl and A. Chaves
these formulations was achieved by restricting or eliminating the exposure of the

alkoxysilyl group to water. The approach was to isolate the silane in an organic
phase and thereby minimize contact with water. Emulsions or organic polymer
dispersions in which these silanes are partitioned into the organic phase provide
effective barriers to water attack.
The solubility of water in these organic phases is low. The stability of the
formulations containing silanes can be improved further by slowing down the
rate of water reaction with these silanes by using sterically hindered alkoxysilyl
groups. However, a balance must be reached whereby the silanes will be sufficiently
stable to hydrolysis to allow preparation of an emulsion or a dispersion containing
these silanes with a reasonable shelf-life, but the silanes will still hydrolyze upon
application.
An understanding of the hydrolysis kinetics for these silanes would aid in the
preparation of formulations with long shelf-lives and with sufficient reactivity to
condense and react with surfaces upon application. Much is already known about
the kinetics and mechanisms of alkoxysilane hydrolysis [5]. The effects of alkyl
substituents on the hydrolysis of trialkoxysilanes in aqueous solutions have been
reported previously [6]. Hydrolysis studies of alkoxy silanes have also been done in
mixtures of water and organic solvents [5]. In these studies it was found that the
reaction was first-order with respect to the silane. The order with respect to water
was between one and six depending on the solvent and water concentration [7].
The structure of the alkoxy group also had a significant effect on the catalyzed
hydrolysis of tetraalkoxysilanes. Tetrakis~(2,6-dimethylheptoxy)silane was found
to hydrolyze 17-times slower than tetraethoxysilane [8]. Knowing the effects of the
alkoxy groups on the hydrolysis of alkoxysilanes would be useful in designing new
materials for waterborne formulations.
A previous model study for investigating the reactions of alkoxysilanes in wa-
terborne formulations examined the hydrolysis of vinyltrialkoxysilanes in an ace-
tone and water mixture [9]. The base-catalyzed rates of hydrolysis were measured
for trimethoxy-, triethoxy-, tripropoxy-, tri-2-propoxy- and tris-(3-oxabutoxy)vinyl
silanes. Vinylsilanes were chosen because they do not contain other reactive groups
which would complicate the kinetic study. The reactions were carried out under
pseudo-first order conditions where the concentration of water was much greater
than the concentration of the silane. This study examines the acid-catalyzed rates of
hydrolysis of these silanes under similar conditions.
2. EXPERIMENTAL
2.1. Materials
Vinyltrimethoxysilane, vinyltriethoxysilane, vinyltri-2-propoxysilane and vinyltris-
(3-oxabutoxy) silane were obtained from Crompton Corporation under the trade
names Silquest® A-171, Silquest® A-151, CoatOSil® 1706 and Silquest® A-172,
respectively.
Reactivity of silanes in waterborne systems 5
Aqueous stock solutions were freshly prepared using perchloric acid. Potassium
chloride was used to maintain the ionic strength, /x, of the final silane solutions at
fi = 0.1. The pH of the solution was measured with an Accumet 910 pH meter and
a Fisher combination electrode.
2.2. Method for monitoring hydrolysis reactions

The hydrolysis of the silanes was monitored using proton nuclear magnetic reso-
nance (!H-NMR) spectroscopy. The vinyltrialkoxysilane solutions were prepared
by adding the silane, 5.0 g of acetone and 1.8 g of aqueous perchloric acid solu-
tion to a 10 ml volumetric flask. Additional acetone was added to the 10-ml mark.
The vinyltrialkoxysilane concentration was 0.009 M. The samples were placed in
a 5-mm NMR tube with a 3-mm NMR tube insert containing acetone^. The tem-
perature of the solutions was maintained at 295 K (22°C). Proton NMR spectra were
taken on a Briiker ACP-300 or a Varian VXR-300 spectrometer. A 9.2-degree pulse
width was used with an acquisition time of 2 s and a relaxation delay of 5 s. Chem-
ical shifts are reported in parts per million (ppm) downfield from tetramethylsilane
(0.0 ppm). Spectra were taken at several time intervals from the initiation of the
reaction.
3. RESULTS AND DISCUSSION

3.1. Hydrolysis of hindered alkoxysilanes
The rates of hydrolysis of a series of vinyltrialkoxysilanes were determined by
proton NMR spectroscopy [9]. The alkoxy groups studied in the series included
methoxy, ethoxy, 2-propoxy and 3-oxabutoxy. The concentration of water in the
silane/acetone solutions was 18% (w/v).
The hydrolysis of trifunctional silanes proceeds in a stepwise manner, as shown
in equations (l)-(3)
R'Si(OR)3 + H 2 0 % R'Si(OR)2OH + ROH (1)
R'Si(OR)2OH + H 2 0 % R,Si(OR)(OH)2 + ROH (2)
R/Si(OR)(OH)2 + H 2 0 ^ R'Si(OH)3 + ROH (3)
Each step of the hydrolysis reaction can be monitored independently using the
NMR technique. It has been shown previously that the amount of silanol and
silanediol intermediates remains small and that the ratios of the rate constants k\/k2
and k[/k3 are small [9]. The first step of the hydrolysis is, therefore, the slow step.
Pratt and co-workers [10] found a similar result for the hydrolysis of phenyltris-(3-
oxabutoxy)silane in aqueous solution. The presence of the silanol and silanediol
intermediates was not detected for the more sterically hindered alkoxysilanes.
6 E. R. Pohl and A. Chaves
3.2. Acid catalysis

The rate of hydrolysis of silane esters in mixtures of organic solvent and water may
be expressed by equation (4):
^ P = WH 2 Of[Sr + ^H+][H2Of[Sr
+ £ HO [HO-][H 2 Or [S]m + /:B[B][H2Of [S]m, (4)
where [S] is the concentration of the alkoxysilane and [B] is the concentration of
a basic species that can accept a proton [4]. Previous work has also found that
the reaction order with respect to silane in these solvent systems was one [6]. By
using a large excess of water relative to the silane, pseudo-first-order conditions are
achieved.
The hydrolysis of vinyltrimethoxysilane was carried out in the acetone-water
solvent system with perchloric acid as the catalyst. The ionic strength was
maintained at /x = 0.1 with potassium chloride. Because the reaction was
carried out in aqueous acetone, pH* was used to determine the hydronium ion
concentration. pH* is the negative logarithm of the hydronium ion concentration
in aqueous acetone. It is determined by the equation,
P H* = pHobsd-5, (5)
where pH0bSd is the ordinary pH reading from the pH meter using a combina-
tion electrode and 8 is a correction factor. The correction factor for acetone
was-0.21 [11].
The hydrolysis reaction proceeds under specific acid catalysis [5]. The observed
rate of reaction is dependent only upon the hydronium ion concentration. Under
these conditions, equation (4) may be simplified to
"d[S] v r*i <K

—7— =*obsd[S], (6)
at
+
wherefc0bSd= &H[H ], Integrating equation (6) gives
ln[S] = -£ ob sd' + ln[S0], (7)
where t is the reaction time and [So] is the initial concentration of the silane.
The observed rate constants for the hydrolysis of the vinyltrialkoxysilanes were
obtained from the slopes of ln[S] versus time plots. The correlation coefficients
were generally greater than r 2 = 0.98. The second-order rate constants (fcH) were
obtained from £0bSd and the hydronium ion concentrations. The results are reported
in Table 1.
The rates of hydrolysis of the vinyltrialkoxysilanes are affected by the alkoxy
groups that are attached to the silicon atom. Vinyltrimethoxysilane was found to
hydrolyze 50-times faster than vinyltri-2-propoxysilane. The steric and inductive
effects of the alkoxy groups control the overall rate of hydrolysis. However, steric
substituent constants (£ s ) and polar substituent constants (a*) do not exist for
Reactivity ofsilanes in waterborne systems
Table 1.
The acid-catalyzed rate constants for the hydrolysis
of vinylsilanes in aqueous acetone at 22° C
Silane A^CM^s-1)
Vinyltrimethoxysilane 5.59
Vinyltriethoxy silane 2.49
Vinyltri-2-propoxysilane 0.11
Vinyltris-(3-oxabutoxy)silane 2.79
pH* was equal to 3.2.
Table 2.
The redefined steric and polar substituent constants for the alkoxy groups used to correlate the rate
constants for silane hydrolysis with the Taft equation
Alkoxy group, Steric substituent Polar substituent

RlRIIRmC0_ constant (E's) constant {a*')
CH3O- 1.67 1.47
CH3CH2O- 1.38 0.980
CH3OCH2CH2O- 1.34 1.58
(CH 3 ) 2 CHO- 0.81 0.49
alkoxy groups. Redefined substituent constants Efs and a*' have previously been
reported [9] and are presented in Table 2.
3.3. Polar effects

The Taft equation is often used to make an estimate of the magnitude of polar and
steric effects on chemical reactions [12]. The Taft equation is [12]
log k/ko = pJ2a* + s^2E^ (8)

where a* and £ s are the polar and steric substituent constants, respectively, and p
and s are coefficients that measure the contributions of the polar and steric effects,
respectively.
A Taft equation was constructed using the observed acid-catalyzed rate constants
in Table 1, the substituent constants in Table 2 and the Taft equation (8) and solving
for p, s and fcno &HO is the rate constant for the hydronium-ion-catalyzed hydrolysis
of vinyl tris-(l,l-dimethylethoxy)silane.
The Taft equation that correlates the hydrolysis of vinylalkoxysilanes to steric and
polar effects is
log*H/*HO = 0.10 J > * ' + 0.65 £ E'„ (9)
where £Ho is equal to 4.41 x 10~6 M _ 1 s"1 [9]. Figure 1 is a plot of the exper-
imentally determined log/rn/^HO versus the calculated expression O-IO^V*7 +
0.65 ^2 K- The slope of the plot and correlation coefficient were found to be 1.001
E. R. Pohl and A, Chaves
4.0 -1
y=1,0009x+0.0728
0.9907
0.0 1.0 2.0 3.0 4.0

0.10 £o*' + 0.65ZES'
Figure 1. The Taft plot of the experimentally determined log&H/^HO versus the expression
0.10 Yl &*f + 0.65 J2 E's shows a good correlation between experimental and calculated results.
and r2 = 0.9907, respectively. These data indicate a good fit between the ex-
perimental rate constants and the calculated values using the Taft equation (9).
The p value in equation (8) is 0.10. The positive value of p indicates that electron-
withdrawing groups on the alkoxy groups, R I R II R III CO-, increase the rate of reac-
tion. The small positive value of p indicates that there is a small increase in negative
charge on the silicon atom in the transition state. This suggests that the mechanism
for acid catalyzed hydrolysis is protonation of the substrate followed by a bimolec-
ular SN2-type displacement of the leaving group by water as shown below:
1 t
5+,OR
OR
R'- -Si
OR
OH
The influence of steric factors on the rate of hydrolysis was small. The coeffi-
cient s was 0.65. The small effect of steric bulk may be due to the oxygen atom
of the alkoxy group. The oxygen atom moves the R of the alkoxy group one atom
away from the silicon center. Because R is further away from silicon, changes in its
bulkiness would have a smaller effect than if it were bonded directly to the silicon
atom.
4. CONCLUSIONS
Model systems were used to investigate the premature crosslinking of silylated
latexes. The model systems used were a series of vinyltrialkoxysilanes in 18% (w/v)
Reactivity ofsilanes in waterbome systems 9
water in acetone solvent. The rates of silane hydrolysis were determined for these
systems. Aqueous acetone was used to simulate the environment of the two-phase
waterborne systems. The alkoxy substituents attached to the silicon atom were
varied to determine their effects on the reactivity of the silane ester with water.
The acid-catalyzed hydrolysis rates of silane esters were affected by the structure
of the alkoxy groups bonded to the silicon atom. The rates of hydrolysis were
retarded when the steric bulk or electron donating characteristics of the alkoxy
leaving group were increased. Branched alkoxy groups, especially when the
substitution occurred a to the oxygen atom of the alkoxy group, were effective
at retarding the hydrolysis reactions. The effectiveness was due primarily to steric
bulk of the branched alkoxy groups.
REFERENCES
1. R. J. DePasquale and M. E. Wilson, U.S. Patent No. 4,648,904 (1987).
2. M. E. Wilson, U.S. Patent No. 4,877,654 (1989).
3. E D. Osterholtz, E. R. Pohl, M. J. Chen and A. Chaves, U.S. Patent No. 5,714,532 (1998).
4. M. J. Chen, F. D. Osterholtz, A. Chaves, P. E. Ramdatt and B. A. Waldman, Proc. Twenty-Fourth
Waterborne, High-Solids and Powder Coatings Symposium 24, 132 (1997).
5. E D. Osterholtz and E. R. Pohl, J. Adhesion Scl TechnoL 6, 127 (1992).
6. E. R. Pohl, Proc. 38th Annu. Tech. Conf., Reinforced Plastics/Composites Inst Section 4-B
(1983).
7. J. R. Chipperfield and G. E. Gould, /. Chem. Soc, Perkins II, 1324 (1974).
8. R. Aelion, A. Loebel and F. Eirich, Rec. Trav. Chim. 69, 61 (1950).
9. E. R. Pohl, A. Chaves, C. T. Danehey, A. Sussman and V. Bennett, in: Silanes and Other
Coupling Agents, Vol. 2, K. L. Mittal (Ed.), p. 15. VSP, Utrecht (2000).
10. K. J. McNeil, J. A. DiCaprio, D. A. Walsh and R. F. Pratt, J. Am. Chem. Soc. 102, 1859 (1980).
11. G. Douheret, Bull. Soc. Chim. France, 1412 (1967).
12. J. E. Leffler and E. Grunwald, Rates and Equilibria of Organic Reactions. John Wiley, New
York, NY (1963).
Silanes and Other Coupling Agents, Vol. 3, pp. 11 -20
Ed. K. L. Mittal
© VSP 2004
Silane oligomers: a class of their own
HELMUT MACK*
Degussa, AG, Untere Kanalstrasse 3, D-79618 Rheinfelden, Germany
Abstract—The hydrolysis and condensation of organofunctional silanes, Y-(CH2)3-Si(OX)3 (Y =

organofunctional group, OX = alkoxy), was studied using time-dependent 29Si-NMR spectroscopy.
Data for epoxysilane (DYNASYLAN® GLYMO) are presented. It was found that, on average,
1.5 hydroxyl groups per silicon atom proved to be a limit for the activity of the silane solution
in water. In additon, a monomeric aminosilane (DYNASYLAN® AMEO) and monomeric and
oligomeric cationic benzylamino-functional silanes (DYNASYLAN® 1161) were applied to E-glass
surfaces and the resulting aminosilane films were investigated by AES (Auger Electron Spectroscopy).
It was demonstrated that oligomeric cationic aminosilanes exhibited a superior film-formation over
commonly used monomeric aminosilanes thus leading to more homogeneous, defect-free silane layers
on the E-glass surface.
Keywords: Alkoxysilanes; hydrolysis; condensation; NMR analysis; E-glass; characterisation of

silane layer; AES analysis.
1. INTRODUCTION
Silanes are bifunctional molecules that possess dual reactivity and act as surface
modifiers, adhesion promoters, crosslinkers and moisture scavengers in many
different industrial applications. The properties and effects of silanes are defined
by their molecular structure:
Y-(CH2)„-Si(OX)3
Y = organofunctional group, OX = silicon-functional group (alkoxy), n = 0 or 3.

The organofunctional group Y links with the polymer resin. This group must be
chosen carefully to ensure maximum compatibility with the polymer resin of choice.
The silicon-functional groups OX, usually alkoxy groups, must be hydrolyzed to
silanols (Si-OH) first before they can bond to the substrate or crosslink.
*Tel.: (49-7623) 91-8233; Fax: (49-7623) 91-8571; e-mail: helmut.mack@degussa.com

12 H. Mack
2. HYDROLYSIS AND CONDENSATION OF SILANES

Silane adhesion promoters are typically applied in a concentration of 0.2-2.0% by
weight as additives as such or from aqueous or water/alcohol solutions as a primer.
As a result of the reaction with water, the alkoxy groups on the central silicon
atom are replaced by hydroxyl groups. These Si-OH moieties prove to be very
reactive. In addition to a fast hydrolysis reaction of the silicon-functional group,
the availability of reactive hydroxyl groups on the central silicon atom significantly
impacts the reactivity of a silane. The more hydroxyl groups available, the better
the adhesion promotion effect [1]. In this context it has long been postulated that
exclusively monomeric trisilanols (or silanetriols) account for the reactivity of a
silane [2]. A maximum concentration of monomeric trisilanols should, therefore,
be the key for excellent silane performance [31.
Subsequently, to the formation of monomeric silanols, short-chain and ring-
structured oligomeric siloxanes are formed by Si-OH condensation reactions.
These oligomeric siloxane structures are still soluble in water and bear active Si-OH
groups. As the hydrolysis and condensation reactions proceed, silane activity
decreases [4]. Finally, crosslinking occurs, thus leading to a substantial decrease in
the adhesion promoting effect and the formation of insoluble, polymeric siloxanes
(gel structures). Knowledge of the concentration of the reactive Si-OH functions is,
therefore, important for the practical use of a silane.
Detailed information on the silane type (alkoxysilane or silanols) and silane
structures (monomeric, oligomeric, or polymeric) present and their changes as a
function of time can be gained by 29Si-NMR spectroscopy.
3.29Si-NMR SPECTROSCOPY OF SILANE HYDROLYSATES [5,6]

All experimental work was carried out using glass apparatuses. NMR tubes
were pre-treated with hexamethyldisilazane. The 4 wt%, 10 wt% and 50 wt%
concentrations of the silane investigated were stirred with demineralized water in
a glass beaker at 500 rpm (revolutions per min). At fixed time intervals, the stirrer
was switched off and a 2-ml sample was taken from the reaction mixture. The
29
Si-NMR spectroscopic analysis was carried out after addition of 1 ml acetone-
rf6 with 3 wt% chromium(III)-acetylacetonate as internal standard and relaxation
accelerarator. At the same time the homogeneity of the aqueous silane solution was
checked by measuring the turbidity.
In the 29Si-NMR spectrum of 3-glycidyloxypropyltrimethoxysilane (DYNASY-
LAN® GLYMO) all monomeric units are detectable the initial stage of the hydroly-
sis reaction (see Fig. 1).
Silanols are metastable units which undergo condensation reactions to thermo-
dynamically more stable siloxane units by elimination of water. The formation of
silanols and oligomeric structures through silanol condensation reactions is influ-
enced both by the type of organofunctional group (polar or non-polar) and by the
conditions of the hydrolysis reaction (pH, temperature, silane concentration, or cata-
lyst). Monomeric silanols form oligomeric siloxanes. With the aid of 29Si-NMR it is
Silane oligomers: a class of their own 13
i— Silanol
Silanediol —i
Trialkoxysilane
mr^
T T T
- 45 - 55 - 65 ppm
Figure 1. Beginning of the hydrolysis reaction of epoxysilane DYNASYLAN® GLYMO.
OMe OMe
I
i i i
Y-(CH2)3-'/Si\o//
OMe Y-(CH 2 ) 3
M-structures D-structures
Trialkoxysilane and I
silane(x)ols
Y-(CH 2 ) 3
T-structures
\+*fai*im*W*»^»0*iî**^^*mii#**
T
•20 •40 -60 •80 ppm
Figure 2. Oligomeric siloxanes of methacrylate silane DYNASYLAN® MEMO (analytically
monitored in the 29Si-NMR spectrum).
possible to obtain information on the structure and amount of oligomeric siloxanes

formed. The 29Si-NMR spectrum (Fig. 2) shows oligomeric siloxanes of 3-metha-
cryloxypropyltrimethoxysilane (DYNASYLAN® MEMO).
The characterization of the still soluble oligomeric siloxanes can be done us-
ing the terms M-structure (for mono-crosslinked silicon units), D-structure (for di-
crosslinked silicon units) and T-structure (tri-crosslinked silicon units) [7]. The
chemical shift, which characterizes the environment of the silicon atom, i.e., the
number of siloxane links, is determined in terms of a large number of different
resonance lines. In the region of monomeric silane structures a maximum of four
resonance lines are present. These four resonance lines correspond to the unhy-
drolyzed organofunctional silane, the monosilanol, the disilanol and the trisilanol.
Depending on the degree of hydrolysis and condensation of monomeric silane struc-
tures, a large number of resonance lines result in the regions of the M-, D- and
T-structures.
14 H. Mack
4. 29Si-NMR STUDY OF 3-GLYCIDYLOXYPROPYLTRIMETHOXYSILANE
In demineralized water, at pH 6, hydrolysis of 3-glycidyloxypropyltrimethoxysilane

to the corresponding silanols and consecutive oligomerization occurs. The rate
of the hydrolysis and condensation reactions depends very much on the H +
concentration and anion (e.g., chloride or acetate). At the same time, the oxirane
ring opens up, although much more slowly. In addition, a condensation reaction
product of oxirane diol and monomeric silanols or OH-functional siloxanes can
be detected in the 29Si-NMR spectrum after 1 h (branched or X-product). The
concentration of this X-product increases to more than 25% after 10 weeks. The
proportion of monomeric trisilanol decreases from initially 75% (after 15 h) to less
than 10% after 10 weeks.
Important for the practical use of 3-glycidyloxypropyltrimethoxysilane is the
absolute concentration of remaining reactive Si-OH functions. This concentration
is decisive with regard to the silane reactivity with an inorganic surface. It was
found that oligomerized siloxanes having free Si-OH functions were capable
of chemically binding to inorganic surfaces [8]. In addition, a more uniform
silane layer can be expected when an inorganic surface is coated with short-chain
oligomeric siloxanes (D-structures). Short-chain oligomeric siloxanes exhibit a
better wettability than monomeric silanes [9]. It is obvious that a homogeneous
silane coverage of the inorganic surface results in an excellent adhesion promoting
effect [10].
Active Si-OH functions in mol% were determined as (3 mol% trisilanol + 2 mol%
M-structures + 1 mol% D-structures)/3. In a 10 wt% 3-glycidyloxypropyltrimeth-
oxysilane solution, starting from 85% after 15 h, the active Si-OH functions are
still more than 70% after 1 week. This corresponds to about 2.1 hydroxyl groups
per silicon atom. In a 50 wt% 3-glycidyloxypropyltrimethoxysilane solution a
maximum, more than 93%, of active Si-OH functions is reached after 8 h. This
decreases to less than 15% after 3 weeks, thus corresponding to about 0.4 hydroxyl
groups per silicon atom.
In contrast to many postulates [2], initially no reduction in the reactivity of
the aqueous silane solution is associated with the oligomeric siloxane build-up in
aqueous solutions of organofunctional silanes. In contrast, inorganic surfaces to
be modified are offered relatively large oligomeric siloxanes which lead to a more
homogeneous, more complete, and more defect-free silane layer than monomeric
silanes [5]. It was also found that non-amine organofunctional silanes had a
significantly lower tendency to form crosslinked structures upon reaction with
water [11]. Ongoing condensation leads to a loss of hydrophilic functions thus
resulting in poor solubility of the resulting siloxanes. A minimum concentration
of 50% active Si-OH functions (an average of 1.5 hydroxyl groups per silicon atom)
proves to be a limit for the reactivity of a silane solution. Consequently, the optimum
processing period is 5-40 h for a 10 wt% 3-glycidyloxypropyltrimethoxysilane
solution in water at a pH of 3 to 4 [11].
5. SILYLATED SURFACES: AES STUDY

The analysis of silanes on substrates and the elucidation of the exact nature of the
silane surface layers and their relationship to the coating conditions has proven
to be difficult. Various techniques are used in the industry. Qualitative evalua-
tions are based on colorimetric and flotation tests. A properly alkylsilane-treated
mineral filler floats on water, whereas a properly amino- or epoxysilane-treated
mineral filler floats on toluene. Quantitative evaluations include, e.g., pyrolysis-
GC (Gas Chromatography), ESCA (Electron Spectroscopy for Chemical Analysis),
SIMS (Secondary Ion Mass Spectroscopy) and FT-IR (Fourier Transform InfraRed)
spectroscopy [12-14]. To obtain information on the degree of silane crosslinking
29
Si-CP/MAS-solid-state-NMR (Cross Polarization Magic Angle Spinning) is
used [15]. However, open questions related to the distribution of the silane film
on the surface, the orientation of the organofunctional groups on the surface (im-
pacts the adhesion promoting or surface modification effect), the thickness of the
silane film (mono- or multilayer) and the silane film topography and roughness still
remain.
AES [16] has served as a powerful tool for silane film analysis on inorganic sur-
faces. AES identifies the elemental compositions of surfaces by measuring the en-
ergies of Auger electrons. An Auger spectrum plots electron signal intensity versus
electron energy. Secondary electrons have an energy of less than approx. 50 eV.
As a consequence, Auger electrons fail to emerge with their characteristic energies
if they start from deeper than about one to five nm into the surface. Thus, Auger
analysis is surface specific. In its basic form AES provides compositional informa-
tion on a relatively large area (approx. 1 mm2) of surface. Auger spectroscopy is
non-destructive. To obtain information about the variation of the surface composi-
tion with depth, it is necessary to gradually remove material from the surface region
being analyzed, while continuing to monitor and record the Auger spectra. This
controlled surface etching of the analyzed region can be accomplished by simul-
taneously exposing the surface to an ion flux (e.g., Ar + ) which leads to sputtering
(removal) of the surface atoms. Elemental depth profiling of the surface with uni-
form sensitivity can then be realised. Other advantages of AES include monolayer-
sensitive surface analysis with high spatial resolution and elemental mapping across
the surface. Surface sensitivity typically is less than one nm and lateral resolution
typically is lower than 50 nm. AES is sensitive to all elements, except hydrogen and
helium, with detection limits of 0.1 to 1.0 at% [16].
A PHI® model 595 Auger electron spectroscopy unit was used to conduct the
elemental surface analysis. This unit combines Auger analysis with the ability to
also perform X-ray Photoelectron Spectroscopy (XPS) and SIMS. The multiple
analysis techniques provided within this single ultra-high vacuum chamber allow
analysis of most materials. The inorganic substrate employed was plate E-glass.
E-glass is widely used in commercial and industrial products, e.g., glass fibers.
Glass fibers made with E-glass exhibit good strength, a low modulus, are available
in many forms and have the lowest cost. Argon ions were used for AES depth
profiling.
16 H. Mack
6. AES DEPTH PROFILING STUDY

Information on the variation of composition with depth below the surface can be
obtained by collecting the Auger spectra as the sample is simultaneously subjected
to etching by ion bombardment. Figure 3 shows the AES analysis of the plate
E-glass control sample.
The plate E-glass was rinsed with ethanol prior to AES analysis. As a result,
the elemental composition consists only of silicon and oxygen (Si02) after a few
seconds of sputtering (shown by the arrow in Fig. 3).
For the AES depth profiling, the ethanol-rinsed plate E-glass was dipped for 5 min
in a freshly prepared 1 wt% aqueous aminosilane solution and then dried for 1 h at
room temperature. Then AES surface analysis was performed. Figure 4 shows
the 2-dimensional visualization of the monomeric 3-aminopropyltriethoxysilane
(DYNASYLAN® AMEO) areal distribution.
The varying intensity of the AES 2-dimensional carbon mapping proves that
the aminosilane surface coverage is not homogeneous. The inhomogeneity of the
aminosilane surface layer is also shown by the shady SEM (Scanning Electron
Microscopy) image.
The AES depth profiling of a "dark" intensity spot out of the AES 2-dimensional
mapping (Fig. 5) shows that the aminosilane layer can be completely removed
after a few seconds. This translates into an aminosilane film thickness of only
approx. 3 nm for this "dark" intensity spot. In contrast, AES depth profiling of
a "bright" intensity spot shows that a complete removal of the aminosilane layer
takes much longer. The much longer sputtering time translates into an aminosilane
film thickness of approx. 60 nm. The AES data show that the aminosilane surface
coverage consists of areas with silane films of only 3 nm thickness and areas
with silane films of up to 60 nm thickness. These findings make the concept
ioo H
90 H
80 H
70 J
g 60 j 0
IO
50
1
40 1
Si
a
o 30 |
o
20 \
1 10 -f Sn, Ca, etc.
0 T 1 1 I T
0 5 10 15 20 25 30 35 40
Sputtering time;[s]
Figure 3. AES depth profiling of control plate E-glass sample (O, oxygen; Si, silicon; C, carbon;
Sn, tin; Ca, calcium).
Secondary electron image AES 2-dimensional carbon mapping
100 \xm 100 nm Intensity
Figure 4. SEM image and AES 2-dimensional carbon mapping of aminosilane DYNASYLAN®
AMEO.
Film thickness Film thickness

approx. 3 nm approx. 60 nm
0 50 100 150 200 250 300 350 400

Sputtering time [s] Sputtering time [s]
Figure 5. AES depth profiles of plate E-glass treated with aminosilane DYNASYLAN® AMEO.
of homogeneous silane monolayer coverage on substrates based on simple silane

surface reaction of dubious value. Optimum surface coverage is obtained with
coating levels well in excess of a nominal monolayer.
Cationic aminosilanes typically are used on an industrial scale for the produc-
tion of computer circuit boards. Monomeric and oligomeric cationic aminosilanes
18 K Mack
Secondary electron image Secondary electron image

Monomer Oligomer
50 nm 50 îm
Figure 6. SEM images of cationic aminosilanes on E-glass.
are commercially available. The advantages of oligomeric silanes include chem-

ical multifunctionality within a single product, low volatility and viscosity, high
flash point, greatly reduced volatile by-product, improved film-forming properties
on substrates, and being a polymer by OECD (Organisation for Economic Co-
operation and Development) definition.
Figure 6 shows the 2-dimensional visualizations of monomeric and oligomeric
cationic aminosilanes on E-glass. As can be clearly seen, the oligomeric cationic
aminosilane leads to a much more homogeneous surface coverage than the mo-
nomeric cationic aminosilane. Only the SEM image of the monomeric cationic
aminosilane is shady.
AES depth profiling (see Fig. 7) demonstrates that the monomeric cationic
aminosilane leads to an average silane layer thickness of approx. 10 nm, whereas the
oligomeric cationic aminosilane forms much thicker silane layers with an average
thickness of approx. 200 nm.
The AES linescan, as shown in Fig. 8, clearly demonstrates the resulting very
homogeneous silane layer of the oligomeric cationic aminosilane. The treatment
with a monomeric cationic aminosilane leads, according to the AES linescan, to a
more inhomogeneous silane coverage of the E-glass surface.
It can be concluded that oligomeric silanes wet surfaces much better than
monomeric silanes, thus leading to a more homogeneous silane film on the surface.
It was found that AES served as a powerful tool for silane surface analysis.
It can be clearly seen that oligomeric silanes do exhibit significant advantages
Film thickness Film thickness

Monomer Oligomer
approx. 5 to 10 nm approx. 170 nm
10 15 20 25 30 400 600 800
Sputtering time [s] Sputtering time [s]
Figure 7. AES depth profiles of plate E-glass treated with monomeric and oligomeric cationic
aminosilanes (N, nitrogen).
Monomer Oligomer
xlO 4 xlO 4
200 300 200 300 500

Distance [\xm] Distance [nm]
Figure 8. AES linescans of plate E-glass treated with monomeric and oligomeric cationic aminosi-
lanes.
over commonly used monomeric silanes. Oligomeric silanes are characterized

by excellent surface wetting. In the case of oligomeric cationic aminosilanes the
resulting silane layer thickness is approx. 200 nm.
7. SUMMARY
The 29Si-NMR investigations demonstrated that the concept of holding exclusively
silanetriols responsible for silane reactivity is of dubious value. No reduction in
20 H. Mack
silane reactivity of an aqeuous silane solution can be associated with the build-
up of soluble oligomeric species. In contrast to the concept of silane monolayer
coverage of substrates, the use of silanes in substrate surface treatment results in
silane multilayers. The use of oligomeric silanes leads to a more homogeneous
silane film on the substrate surface.
REFERENCES
1. E. P. Plueddemann, Silane Coupling Agents, 2 nd edn. Plenum Press, New York, NY (1991).
2. F. D. Osterholtz and E. R. Pohl, in: Silanes and Other Coupling Agents, K. L. Mittal (Ed.),
pp. 119-141. VSP, Utrecht (1992).
3. H. Ishida and J. L. Koenig, Appl. Spectrosc. 32, 469 (1978).
4. E. P. Plueddemann, SPI24th Ann. Tech. Conf. Reinf. Plasty 19-A (1969).
5. M. Brand, A. Frings, P. Jenkner, R. Lehnert, H. J. Metternich, J. Monkiewicz and J. Schramm,
Z. Naturforsch. 54b, 155 (1999).
6. L. Delatire and F. Babonneau, Mater. Res. Soc. Symp. Proc. 346, 365 (1994).
7. G. Engelhardt and H. Jancke, Polym. Bull. 5, 577 (1981).
8. H. Ishida and J. D. Miller, Macromolecules 17, 1659 (1984).
9. H. Ishida, Polym. Sci. Technol. 27, 25 (1985).
10. K. Itadani, H. Kawasaki and T. Nakatsuka, J. Appl. Polym. ScL 24, 1985 (1979).
11. F. Beari, M. Brand, P. Jenkner, R. Lehnert, H. J. Metternich, J. Monkiewicz and H. W. Siesler,
J. Organomet. Chem. 625, 208 (2001).
12. R. K. Her, in: The Chemistry of Silica, R. K. Her (Ed.), Ch. 6. Wiley, New York, NY (1979).
13. C. G. Armistead and J. A. Hockey, Trans. Faraday Soc. 63, 2549 (1967).
14. W. Herd, J. Phys. Chem. 73, 2372 (1969).
15. T. Gobel, U. Gorl, A. Hunsche, M. Knaack and A. Muller, Kautschuk Gummi Kunststoffe 50,
881 (1997).
16. D. Briggs and J. T. Grant (Eds), Surface Analysis by Auger and X-ray Photoelectron Spec-
troscopy. IM Publications (2003).
Silanes and Other Coupling Agents, Vol. 3, pp. 21-37
Ed. K. L. Mittal
© VSP 2004
FT-IR and XPS surface characterization

of allyltrimethoxysilane and
3-methacryloxypropyltrimethoxysilane mixture
adsorbed onto titanium substrate
J. P. MATINLINNA1 •*, K. LAAJALEHTO2, L. V. J. LASSILA1, A. YLI-URPO1

and P. K. VALLITTU]
1
Institute of Dentistry, Department of Prosthetic Dentistry and Biomaterials Research,
University of Turku, FIN-20520 Turku, Finland
Laboratory of Materials Science, Department of Physics, University of Turku,
FIN-20014 Turku, Finland
Abstract—Organofunctional trialkoxysilane coupling agents are used in restorative dentistry in order

to enhance the bonding between restorative and veneering composites and inorganic materials, such
as E-glass fibres, grit-blasted dental ceramics, base and noble alloys, and also for composite filler
treatment. The most commonly used silane for coupling composite resins to metals, ceramics
and silica-coated composites (for repair) is 3-methacryloxypropyltrimethoxysilane (MPS). The aim
of this study was to determine whether an allyltrimethoxysilane (ALS) and MPS mixture could
form a stable, adhesion-promoting metallo-siloxane layer on a titanium substrate. Two mixtures
(0.25 vol%+0.25 vol% and 0.50 vol%+0.50 vol%) were applied to the titanium substrate and cured at
room temperature or in an oven (at 110°C for 1 h). The siloxane films were characterized using XPS,
RAIR and AFM. The results showed that all samples had siloxane films on them. According to the
XPS analysis it might be suggested that silanization with ALS seemed to form a more Si-rich siloxane
film compared to silanization with MPS. It was concluded also that MPS + ALS silane mixtures would
be worthy of further investigation. The FTIR analysis suggested that Si-O-Ti bonds were formed to
bond the siloxane film to the Ti substrates.
Keywords'. Silane coupling agent; titanium; XPS; FT-IR; AFM; allyltrimethoxysilane; 3-methacryl-
oxypropyltrimethoxysilane.
*To whom correspondence should be addressed. Tel: (358-2) 333-8288; Fax: (358-2) 333-8390;
e-mail: jukka.matinlinna@utu.fi
22 J. P. Matinlinna et al.
1. INTRODUCTION
Trialkoxyorganosilanes are dual-functional organic-inorganic adhesion promot-
ers [1,2]. They are known to enhance the adhesion at the organic/inorganic interface
through a dual reactivity. Vinyltrimethoxysilane and 3-methacryloxypropyltrime-
thoxysilane (MPS) have been successfully applied in glass-fibre reinforced plas-
tics [3-5] and in prosthetic dentistry [6-8]. The commercial silanes used in den-
tistry are predominantly based on pre-hydrolyzed MPS and are usually applied
from 50 to 95 vol% ethanol/water solutions. Ethanol is used because of its low
toxicity while methanol is poisonous. Silanes are molecules which first undergo
hydrolysis to transform their alkoxy groups into labile silanol groups:
R'-Si(OR)3 + 3H 2 0 -> R'-Si(OH)3 + 3R-OH (1)
where R' is the non-hydrolyzable organofunctional group and OR is methoxy or
ethoxy group. Reactive silanol groups, Si-OH, are formed during hydrolysis. These
silanol-containing species are highly reactive intermediates which are responsible
for bond formation with the substrate. They subsequently condense and form
dimers:
R-Si(OH) 3 + R/-Si(OH)3 -> R / -Si(OH) 2 -0-Si(OH) 2 (R / ) + H 2 0 (2)
Silanol dimers then form oligomers that react to form a branched siloxane layer.
Hydrogen bonding can also occur and reactions (1) and (2) can occur simulta-
neously [9-11]. In the presence of an inorganic substrate with -OH groups,
e.g., Ti-OH, the formation of Si-O-Ti interfacial bonds is possible. A very sim-
plified reaction scheme is shown below:
... R'-Si(OH) 2 -0-Si(OH) 2 (R')-... + 20H-Ti
-> -R / -Si(R / )(0-Ti)-0-Si(R / )(0-Ti)-0-... + 2H 2 0 (3)
The organofunctional group can be linked to a silicon atom via the propylene group
(-CH 2 CH 2 CH 2 -). The typical linker length is three carbon atoms, since in that
case the thermal stability is good. Hydrolysis and condensation of alkoxysilanes is
dependent on both pH and catalyst [9]. In the presence of appropriate organofunc-
tional group, polymerization with monomers of the composite resin (e.g., veneering
composite) is possible [12, 13].
Aminosilanes and the siloxane films have been characterized in conjunction with
E-glass fibres [14, 15]. Investigations have been undertaken with certain disiloxanes
on silica [16-18] and on aluminum substrates [17]. Glycidoxysilanes (epoxysi-
lanes) have been studied as primers for aluminum and aluminum alloys [19-21].
Silicones are synthesized starting from certain silanes, and silicones have a wide
range of applications as lubricants, sealants, coatings, biomaterials, etc. [22, 23].
Certain aminosilanes have been shown to be useful in the steel industry as a surface
pre-treatment agent (paint adhesion promoter) substituting harmful chromates [24],
protecting aluminum alloys and silane mixtures in particular have shown better
performance than silanes alone [11, 25]. Non-functional silanes can also be used
Characterization of an ALS and MPS mixture adsorbed onto titanium substrate 23
\
CH3
(A)
/CH3
O
N
CH 3
(B)
Figure 1. Structures of 3-methacryIoxypropyltrimethoxysilane (A) and allyltrimethoxysilane (B).
for corrosion protection [26]. The silane most often used in dental applications is
MPS [27-30], but also vinyltrimethoxysilane is applied [31, 32]. In restorative den-
tistry silanes are used either as a composite filler surface modification agent or for
coupling repair composites to ceramics or metals (Ti, base alloys and noble alloys).
Allyltrimethoxysilane (ALS, systematic name 2-propenyltrimethoxysilane) is well
known as a coupling agent, especially for vinyl-addition silicones, but to the au-
thors' best knowledge, it has not been previously studied for dental applications.
The structures of MPS and ALS are shown in Fig. 1.
An adequate bond strength between materials applied in intra- and extra-oral
restorations and dentures is of crucial importance [33-35]. Titanium instanta-
neously forms a thin oxide layer. It is a biocompatible, non-toxic, relatively cheap,
non-degradable material with the additional benefit of its superior resistance to-
wards corrosion and erosion at body temperature. Titanium is used for implants, for
crowns and in biomedical applications, e.g., for hip joints, bone splints, pacemaker
cases and heart valve components [27, 36-38].
The aim of this study was to characterize the siloxane film formation, derived
from ALS and MPS mixtures on a Ti substrate surface, at two curing temperatures.
The formation of a metallo-siloxane film, from a silane treatment, is a prerequisite
for a composite resin to be bonded to the substrate. Silanizations with ALS alone
and MPS alone were carried out for comparison to silanizations with ALS and MPS
mixtures.
2. MATERIALS AND METHODS

The titanium used was commercially pure (grade 2) and supplied by Permascand
(Ljungaverk, Sweden, lot ASTM B265 89). It was cut into planar coupons
24 7. P. Matinlinna et al.
(20 mm x 40 mm x 1 mm), which were finished and polished by grinding them

with silicon carbide (SiC) paper (1200 grit). Before silanization, the Ti was rinsed
and degreased ultrasonically with ethanol and acetone (Quantrex 90 WT, L&R
Manufacturing, Kearny, NJ, USA).
In this study, 0.25 vol% + 0.25 vol% and 0.50 vol% + 0.50 vol% solutions of
allyltrimethoxysilane (ABCR, Karlsruhe, Germany, 97%, lot 3A-2263-2D23-BS)
and 3-methacryloxypropyltrimethoxysilane (Sigma-Aldrich Chemie, Steinheim,
Germany, Purum 98%, lot S01603-022) were prepared. The silanes were used
without re-distillation. The silanes amounts were measured, and rapidly added and
sealed into 25-ml polyethylene bottles to avoid atmospheric humidity, followed by
the addition of 95 vol% 2-propanol solution (Riedel-de Haen, Seelze, Germany, pro
analyst, lot 11310) in de-ionized water, with an electrical resistivity of 18.2 MQ cm.
The pH was adjusted to 4 with 1 M acetic acid. Next, the sealed silane solutions
were allowed to hydrolyze for 1 h at RT. Silanes, one drop at a time, were brushed
(a new brush used each time) onto titanium coupons and gently air-dried with oil-
free compressed air, following clinical procedure (in dental laboratories and at chair-
side) and using the same amount of silane [33], to form the siloxane films. The
samples were silanized at RT, and cured either at RT (15 min) or at 110°C (1 h).
One hour curing time was selected based on the silanization procedures described
in the literature [9].
FT-IR analysis was performed (mid-IR 3800 cm _1 -600 cm" 1 ) with a Reflectance
Absorbance-Fourier Transform Infrared (RA-FT-IR) spectrometer (Perkin-Elmer
Spectrum One, Perkin-Elmer, Beaconsfield, UK) using a variable angle specular
reflectance monolayer/grazing angle accessory (Specac, Smyrna, GA, USA). The
grazing angle was 80°, the number of scans was 32, the scan speed was 0.50 cm s~l
and the resolution was 2 cm^1.
Water storage was utilized to compare whether a short-term storage in water
would have an effect on the siloxane film. Silanized Ti samples were immersed
in de-ionized water for 24 h at 37°C, in sealed glass bottles that were kept in an
autoclave. The storage effect was then evaluated (samples were first air-dried) by
FT-IR analysis [39].
X-ray photoelectron spectra (XPS) were recorded using the Perkin-Elmer PHI
5400 ESCA System (Perkin-Elmer, Eden Prairie, MN, USA) with a monochromatic
aluminum Ka X-ray source. This source was operated at 300 W and 14.4 kV.
The samples were mounted with a double-sided carbon-tape onto a holder and then
placed in the chamber at a pressure of 10~7 Torr. A take-off angle of 45° was used.
The photoelectrons were generated by X-ray photons of energy 1486 eV (Al Ka).
Both survey and multiple spectra were recorded.
An atomic force microscope (AFM) AutoProbe CP (Park Instruments, Sunnyvale,
CA, USA) was utilized for imaging 30 x 30 fxm2 and 4 x 4 fjum2 surface areas. The
cantilever was gold coated and the tip was used in the AFM contact mode. The
frequency used was 1 lps (line/s) for 4 x 4 /xm2 and 0.3 Ips for 30 x 30 jttm2 surface
areas.
3. RESULTS
3.1. FT-IR spectra
The FT-IR spectra showed some interesting changes (cf., Table 1). ^Si-OH peaks
were seen to diminish after curing: in particular, none was seen after curing at ele-
vated temperature. For uncured samples the signals were s at 3700 cm" 1 -3500 cm - 1
and m at 880 cm - 1 (where s = strong, m = medium). Hydrogen-bonded -OH in
the cured samples was minimal (broad, s 3400 cm _1 -3200 cm - 1 ). Free water was
detected, adsorbed on the siloxane film (s, broad ca. 3280 cm" 1 ) [40,41]. Si-
methoxy signals (m, s 2840 cm - 1 , 820 cm - 1 ) had disappeared after curing. The
carbonyl peak - C = 0 , ca. 1730 cm - 1 (from methacrylate), was, of course, not
seen for ALS-silanized samples in Fig. 2. However, it remained unchanged, as
did the double bond C=C of the allyl group (m, ca. 1620 cm _1 -1640 cm - 1 ) in
Figs 3-5 [42-44]. Oven curing did not appear to decompose any of these groups.
An indication of the siloxane film formation were -Si-O-Si- siloxane peaks that
appeared (s 1130 cm _1 -1000 cm - 1 ). Evidently long ^Si-O-Si-O-Si chains were
present (s ca. 1080 cm -1 -1040 cm" 1 ) [41]. Si-O-Ti peaks (s ca. 925 cm" 1 ) were
detected in all the samples. RAIR spectra are presented in Figs 2-5.
Table 1.
Some significant IR absorption frequencies (wavenumbers) [11, 13, 55]
Wavenumber (cm *) Assignment

3740 Free Si-OH stretching
3740-3500 Bridged Si-OH stretching
3690 Free hydroxyl, -OH
3400-3200 Hydrogen bonded, -OH
3280 Free water, H2O
3385 Bonded silanol, Si-OH
2940, 2840 Asymmetric and symmetric stretchings of Si-0-CH 3
2840 S1-O-CH3
1735 Carbonyl C = 0 stretching
1638 Si0 2 wH 2 0 (H-O-H bending motion)
1620-1640 C=C stretching
1480-1300 CH2, CH3 bending
1260 Si-CH 3
1250-1220 Si-CH 2 CH 2 CH 2 CH 2 - long chain
1250-1020 Si-O-Si asymmetric stretching vibration
1190,1100-1080, 1087,818 Si-0-CH 3
1130-1000 Si-O-Si siloxane bonds, often broad and complex
1080-1040 -Si-O-Si-O-Si- long chains
1050 Inorganic silicates
1000-900 Si-O-M bonds (M = transition metal)
925-950 -Si-O-Ti- metallo-siloxane bonds
880 Si-OH, Si-O stretching
820 Si-0-CH 3
487 Inorganic silicates (also Si-0-CH 2 CH3 symmetric deformation)
26 /. P. Matinlinna et al.
1250 ...1000
3000 2000 1000

Wavenumber/cm" 1
Figure 2. FT-IR spectra of titanium substrate (A) untreated, (C) 0.50% ALS silanized at RT and
(B) 0.50% ALS silanized at 110°C (y-axis, absorbance in arbitrary units; jc-axis, wavenumber
in c m - ' ) .
1250 ...1000
1080 ...1040
3000 2000 1000

Wavenumber/cm" 1
Figure 3. FT-IR spectra of titanium untreated (A), 0.50% ALS -f 0.50% MPS silanized and oven
cured at 110°C for 1 h (B) (y-axis, absorbance in arbitrary units; x-axis, wavenumber in cm - 1 ).
1250 ...1000
3400 ...3200
3000 2000 1000

Wavenumber/cm' 1
Figure 4. FT-IR spectra of titanium untreated (A) and 0.50% MPS + 0.50% ALS silanized, cured in
an oven, and after water storage at 37°C for 24 h (B) (j-axis, absorbance in arbitrary units; x-axis,
wavenumber in cm - 1 ).
3.2. XPS spectra

The formation of the siloxane film could easily be observed from the appearance
of Si2s and Si2P intense peaks in the XPS survey spectra at binding energies of
150 eV and 100 eV, respectively. Survey spectra are presented in Fig. 6. Atomic
concentrations based on spectral intensities, presented in Table 2, were also in close
agreement with the AFM images showing dramatic changes in the morphology of
the surface (see next section).
3.3. AFM images

The AFM images of untreated Ti and Ti with siloxane films confirmed the presence
of film on the surface. AFM images of titanium surfaces, without silanization with
different chemical treatments (MPS, ALS and MPS + ALS) are shown in Fig. 7.
Eight images were selected from a larger series. Half of the images presented had
an image size of 4 /xm x 4 /im and the other half were 30 /zm x 30 /xm. The
height scale of all the images was less than 2 ^m. As can be seen, the surface
morphology changed dramatically as a consequence of silanization. Rough Ti
surfaces (Fig. 7A and 7B) appeared much smoother after 0.25 vol% MPS treatment
at RT (Fig. 7C and 7D) and after 0.25 vol% ALS (Fig. 7E and 7F) and 0.50 vol%
J. P. Matinlinna et al.
1250 ...1000
A
1250< , . B
3000 2000 1000

Waven umber/cm' 1
Figure 5. FT-IR spectra of titanium untreated (A), silanized with 0.50% ALS for 15 min (B) and
cured at 110°C for 1 h followed by 24 h water storage at 37°C (C) (j-axis, absorbance in arbitrary
units; x-axis, wavenumber in cm - 1 ).
Table 2.
XPS atomic concentrations (%) of adsorbed silanes based on spectral intensities
Ti and silane solutions vol% Ti c O Si

+ curing conditions
Ti (control) 13.7 43.0 35.4 0
0.25 MPS, oven 2.4 55.5 35.4 6.7
0.50 MPS, oven 0 54.5 37.4 8.3
0.50 MPS, RT 0 57.1 34.6 8.3
0.25 ALS, oven 0 51.1 32.8 16.1
0.50 ALS, oven 0 56.8 34.7 8.5
0.50 ALS, RT 0 51.0 32.8 16.1
0.50 MPS + 0.50 ALS, oven 0 55.3 34.8 9.9
0.25 MPS + 0.25 ALS, oven 0 56.2 34.4 9.4
0.25 MPS + 0.25 ALS, RT 0 54.6 35.4 10.0
0.50 MPS + 0.50 ALS, RT 0 56.5 34.2 9.3
Abbreviations: ALS = allyltrimethoxysilane, MPS = 3-methacryloxypropyltrimethoxysilane,

oven = cured at elevated temperature (110°C), RT — cured at room temperature.
-01s
Ti + MPS
r
c
_i
O
m Q-
<8 CM
tl m co co
pyt0mr+4* i MM"fp^*A • •
U i mi * »II«"^ 11
1 1 —i 1 j i | i 1
1000 800 600 400 200

Binding energy [eV]
Ti + MPS + ALS -01s
CO
O
i (/) Q.
CN CN
CO co
t niMiiiitoil 1
I^^M.*'*'!^
'
1000 800 600 400 200 C

Binding energy [eV\
•01s
Ti + ALS
Q.
CN
CO
"""v
LHN
^*-Û J
1 • 1— -i 1—
1000 800 600 400 200
Binding energy [eV]
Figure 6. XPS survey spectra of adsorbed silanes on titanium. (Top) 0.25% MPS; (middle) 0.25%
MPS + 0.25% ALS; (bottom) 0.25% ALS. All samples were oven cured for 1 h at 110°C.
30 J. R Matinlinna et al.
30
20
2.06- pm
1,03 -
:
0.00^ 10
30
Sim 10
(A)
(B)
Figure 7. AFM images (A-H): unsilanized Ti (A, 30 /xm x 30 /xm; B, 4 /xm x 4 /xm); silanized
with 0.25% MPS at RT (C, 30 /xm x 30 /xm; D, 4 /xm x 4 /xm); silanized with 0.25% ALS at RT
(E, 30 /xm x 30 /xm; F, 4 /xm x 4 /xm); and silanized with 0.50% ALS + 0.50% MPS, cured at 110°C
(G, 30 /xm x 30 /xm; H, 4 /xm x 4 urn).
1.05i
30
0.53 H
20
0.00^
30 ^m
20 10
\iro
10
(Q
/ 0
(D)
Figure 7. (Continued).
0
(E)
0
(F)
MPS + 0.50 val% ALS treatment (Fig. 7G and 7H). The rough surface texture
originating from mechanical grinding diminished after silanization in all cases.
Therefore, a multilayer siloxane film formation might be suggested based on the
AFM images.
489-j
20
(.0
Mm
20 10
10
(G)
(H)
4. DISCUSSION
The type of metal substrate or metal oxide obviously has a significant influence
on the siloxane film [45], The Si-O-Ti covalent bonds formed are assumed
to be responsible for bonding of the siloxane film to the metal substrate. The
hydrophobic, dense and non-porous silane-derived siloxane film is about 50-100

molecular layers thick when the silane solution is 2 vol% and 200^-00 nm thick
when it is 5 vol% [11]. Its thickness is mainly dependent on the concentration
of silane solution (not the reaction time), but also on the siloxane film drying
method and the film curing [18, 46, 47]. Water does not easily penetrate this
hydrophobic 3D structure, though some water is needed for its formation. Some
early studies claimed that the hydrophobic siloxane film still could allow water
to penetrate it; however, the more hydrophobic the film, the less the water can
penetrate it [4]. Silanes in water/alcohol solutions are known to adsorb onto
metal surfaces immediately, and start reacting with them [27]. Alcohol, usually
ethanol or propanol, is needed for dissolution, since most silanes are not soluble
in water. Silanols from dilute (e.g., less than 5 vol%) hydrolyzed silane solutions
spontaneously adsorb onto the metal surface through hydrogen bonding, when the
metal is dipped for a few seconds [48].
Curing in the oven seemed to yield stronger -Si-O-Si- peaks in the correspond-
ing FT-IR spectra than curing at RT. Oven curing did not destroy the organofunc-
tional groups, viz., methacrylate and allyl which are still visible in the spectra. The
silanized samples were kept in water (for 24 h at 37°C), then gently air-blast dried
and the FT-IR spectra were recorded. Water storage did not seem to diminish the
intense -^Si-O-Si peaks. There were no clear differences obtained between differ-
ent curing temperatures. These results might mean that the siloxane film did not
decompose during water storage, but the effect needs to be studied more carefully.
The XPS analysis was based on the appearance of Si2S and Si2P intense peaks.
For Si2P, the binding energy is between 103 and 104 eV when Si is chemically
bonded in silica, and between 102 and 103 eV when Si is bonded in silicates.
Also, silicones (and silanes) have their Si2P signals at the same binding energy
scale. The XPS signals for titanium, Ti02, nitrogen and oxygen do overlap, and no
other elements were expected to be present in significant amounts [49]. However,
the difference between the binding energies of the O l5 and Si2P in inorganic Si
compounds is practically invariant. In the XPS survey spectra the Ois peaks appear
between 429.6 eV and 429.0 eV, and Si2p peaks appear between 150 eV and 100 eV.
The difference between the binding energies of the 0 ] s and Si2P lines in the
organic silicon compounds is almost invariant, 429.0 to 429.6 eV. The magnitudes
of these binding energies have a high positive correlation which can be shown by
plotting Ois binding energy as a function of Si2P binding energy in a diagram, as
suggested and presented by Wagner et al. [50]. For example, dimethylsilicone
and methylsilicone resins can be separately identified, and also silicates from
silicones [50]. The Si2s peak is usually weaker than the Si2p peak in XPS survey
spectra. Applying this method in the XPS analysis in this study, the results
suggested that all the films on titanium substrates mainly consisted of oxidized
silicon, most probably in the form of Si0 2 . Also the lack of Ti in the spectra
suggested a thick layer of siloxane.
According to the XPS results, the pure ALS based siloxane films seemed to
have the highest silicon concentration, 16.1%, when the silane concentration was
0.25 vol%. All silanizations with pure ALS seemed to yield higher Si concentrations
than silanizations with pure MPS (Table 2). There were no significant differences
between silane blends. When comparing the temperature effect, both curing
temperatures gave an average 9.6% for the silicon concentration. For silanization
with 0.50 vol% ALS and curing at RT, XPS seemed to yield almost double the
Si-% than for curing at elevated temperature. Nevertheless, for 0.50 vol% MPS,
this effect did not seem to exist as the Si concentrations were the same. In this
study, the 0.25 vol% MPS silane (cured at elevated temperature) yielded the lowest
value for silicon concentration in the siloxane film, but oven-cured 0.25 vol% ALS
yielded as high a value as the 0.50 vol% ALS, cured at RT. These results seem to be
problematic and might be explained by an uneven silane spreading onto the titanium
substrate during application with a brush. Also, ALS, in particular, might contain
some silica as impurity. Using photoelectron take-off angles other than 45° might
show differences in Si content for the different siloxane films. The AFM images
supported the siloxane film coverage on all the samples.
Many silane mixture studies are based on the use of bis-functional silanes (also
called dipodal silanes, or 'bis-silanes\ i.e., a silane molecule with two Si atoms) in a
blend with conventional mono-functional (with one Si atom in the molecule) silane
coupling agents [51, 52]. These mixtures produce hydrolytically very resistant
siloxane films, and the reason is thought to be a result of both the increased crosslink
density and the ability of dipodal silanes to form six bonds (conventional silanes
form only three bonds) with the substrate surface [9, 53]. The silane mixtures
studied here were prepared from trialkoxysilanes, so the results might not be
comparable with those derived from the silane mixtures with dipodal silanes.
Next we will study the silane blend performance. A dental composite resin
(dimethacrylate or methacrylate type) would be bonded to pre-treated Ti substrates
with different siloxane films. This would predict the clinical use of these silane
mixtures when Ti is used as prosthetic material. The shear bond strength values
should exceed 5 MPa, and preferably be ca. 20 MPa, according to the ISO
standard for dental materials [54]. The silane film thickness measurements, e.g., by
ellipsometry and electrochemical impedance measurements should be of interest.
5. CONCLUSIONS
In this study, the surface of the titanium coupons appeared to be smoother after
silanization and curing, regardless of curing temperature. Curing temperature did
not produce differences in Si content, according to XPS analysis. Silane mixtures
did not show differences in Si concentration at different curing temperatures and
silane concentrations. All the silanes/silane mixtures seemed to form Si-O-Ti
bonds with the substrate. Nevertheless, these results might suggest that a stable
metallo-siloxane film formation onto the Ti substrate could be achieved using ALS,
or ALS + MPS. As the next step, methacrylate resin samples will be light-cured on
the Ti substrates with adsorbed siloxane films.
Acknowledgements
This study was conducted as a part of the "Bio- and Nanopolymers Research
Group" activity of the Centre of Excellence of The Academy of Finland (Project
77317/53301). It was financially supported by grants from "POTRA Polymer Re-
search Programme", funded by TEKES (National Technology Agency of Finland,
Project 40696/01), and from the Finnish Dental Society APOLLONIA. Mrs Taina
Laiho (University of Turku) is acknowledged for the AFM images, Dr. Mervi Puska
(University of Turku) for the drawings, and Mr John Wright (London, UK) is
thanked for proofreading this text.
REFERENCES
1. K. L. Mittal (Ed.), Silanes and Other Coupling Agents. VSP, Utrecht (1992).
2. K. L. Mittal (Ed.), Silanes and Other Coupling Agents, Vol. 2. VSP, Utrecht (2000).
3. H. A. Clark and E. P. Plueddemann, Modern Plastics 40, 133-196 (1963).
4. E. P. Plueddemann, Modern Plastics 47, 92-98 (1970).
5. E. P. Plueddemann, H. A. Clark, L. E. Nelson and K. R. Hoffman, Modem Plastics 39, 135-187
(1962).
6. J. P. Matinlinna, L. V. J. Lassila, M. Ozcan, A. Yli-Urpo and P. K. Vallittu, Int. J. Prosthodontics
17, 155-164 (2004).
7. P. K. Vallittu, J. Oral Rehabil 24, 560-567 (1997).
8. P. K. Vallittu, J. Oral Rehabil. 24, 124-130 (1997).
9. B. Arkles (Ed.), Silane Coupling Agents: Connecting Across Boundaries. Gelest, Morrisville,
PA (2003).
10. T. F. Child and W. J. van Ooij, Trans. Inst. Metal Finish. 11, 64-70 (1999).
11. W. J. van Ooij, D. Q. Zhu, G. Prasad, S. Jayaseelan, Y. Fu and N. Teredesai, Surface Eng. 16,
386-396 (2000).
12. J. G. Marsden, in: Handbook ofAdhesives, I. Skeist (Ed.). Van Nostrand Reinhold, New York,
NY (1990).
13. E. P. Plueddemann, Silane Coupling Agents, 2nd edn. Plenum Press, New York, NY (1991).
14. D. Wang and F. R. Jones, Surf. Interface Anal 20, 457-467 (1993).
15. D. Wang and F. R. Jones, J. Mater. Set 28, 1396-1408 (1993).
16. M. R. Alexander, R. D. Short, F. R. Jones, M. Stollenwerk, G. Mathar, J. Zabold and W. Michaeli,
Br. Ceram. Proc. 54 (Ceramics and Coatings), 87-99 (1995).
17. M. R. Alexander, R. D. Short, F. R. Jones, M. Stollenwerk, J. Zabold and W. Michaeli, J. Mater.
Set 31, 1879-1885(1996).
18. F. R. Jones, M. S. Vedamuthu and J. P. Blitz, Special Publication — R. Soc. Chem. 235
(Fundamental and Applied Aspects of Chemically Modified Surfaces), 173-182 (1999).
19. K. W. Allen, in: Silanes and Other Coupling Agents, K. L. Mittal (Ed.), pp. 81-90. VSP, Utrecht
(1992).
20. M. W. Daniels and L. F. Francis, J. Colloid Interface Set 205, 191-200 (1998).
21. S. Debnath, S. L. Wunder, J. I. McCool and G. R. Baran, Dental Mater. 19, 441-448 (2003).
22. B. Arkles, Chemtech 13, 542-555 (1983).
23. B. Arkles, Chemtech 29, 7-14 (1999).
24. P. R. Subramanian and W. J. van Ooij, Surface Eng. 15, 168-172 (1999).
25. W. J. van Ooij and A. Sabata, in: Silanes and Other Coupling Agents, K. L. Mittal (Ed.), pp. 323-
344. VSP, Utrecht (1992).
26. W. van Ooij and T. F. Child, Chemtech 28, 26-35 (1998).
27. K. Ekstrand, I. E. Ruyter and H. Oysaed, Dental Mater. 4, 111-115 (1988).
28. M. Kern and V. P. Thompson, J. Dentistry 23, 47-54 (1995).

29. W. G. McDonough, J. M. Antonucci and J. P. Dunkers, Dental Mater. 17, 492-498 (2001).
30. J. F. Roulet, K. J. Soderholm and J. Longmate, J. Dental Res. 74, 381-387 (1995).
31. R. L. Bowen, J. Am. Dental Ass. 66, 57-64 (1963).
32. P. K. Vallittu and K. Ekstrand, J. Oral. Rehabil 26, 666-671 (1999).
33. M. Ozcan, J. Prosthet. Dentistry 87, 469-472 (2002).
34. M. Ozcan and A. Akkaya, J. Prosthet. Dentistry 88, 252-254 (2002).
35. M. Ozcan, P. Pfeiffer and Y. Nergiz, Quintessence Int. 29, 713-724 (1998).
36. C. Ohkubo, I. Watanabe, T. Hosoi and T. Okabe, J. Prosthet. Dentistry 83, 50-57 (2000).
37. K. B. May, J. Fox, M. E. Razzoog and B. R. Lang, J. Prosthet. Dentistry 73, 428-431 (1995).
38. H. Yanagida, H. Matsumura, Y Taira, M. Atsuta and S. Shimoe, J. Oral Rehabil. 29, 121-126
(2002).
39. T. Berry, N. Barghi and K. Chung, J. Oral Rehabil. 26, 459^63 (1999).
40. H. Ishida and J. L. Koenig, /. Colloid Interface ScL 64, 565-576 (1978).
41. P. L. Launer, in: Silicone Compounds Register and Review, B. Arkles (Ed.). Petrarch Systems
(1987).
42. H. Ishida and J. L. Koenig, J. Colloid Interface Sci. 64, 555-564 (1978).
43. S. R. Culler, H. Ishida and J. L. Koenig, Appl Spectrosc. 38, 1-7 (1984).
44. S. Naviroj, S. R. Culler, J. L. Koenig and H. Ishida, /. Colloid Interface Sci. 97, 308-317 (1984).
45. J. P. Bell, R. G. Schmidt, A. Malofsky and D. Mancini, in: Silanes and Other Coupling Agents,
K. L. Mittal (Ed.), pp. 49-66. VSP, Utrecht (1992).
46. A. Franquet, H. Terryn and J. Vereecken, Appl. Surf. Sci. 211, 259-269 (2003).
47. F. D. Osterholz and E. R. Pohl, /. Adhesion Sci. Technol. 6, 127-149 (1992).
48. T. Eklund, J. Backman, P. Idman, A. E. E, Norstrom and J. B. Rosenholm, in: Silanes and Other
Coupling Agents, Vol. 2, K. L. Mittal (Ed.), pp. 55-78. VSP, Utrecht (2000).
49. J. E Moulder, W. F. Stickle, P. E. Sobol and K. D. Bomben (Eds), Handbook of X-ray
Photoelectron Spectroscopy. Perkin-Elmer Corporation, Eden Prairie, MN (1992).
50. C. D. Wagner, D. E. Passoja, H. F. Hillery, T. G. Kinisky, H. A. Six, W. T. Jansen and J. A. Taylor,
/. Vac. Sci. Technol. 21, 933-944 (1982).
51. W. J. van Ooij and G. P. Sundararajan, J. Corrosion Sci. Eng. 2, Paper 14 (1999).
52. V. Palanivel, W. van Ooij and Y Huang, Silicon, in press.
53. P. G. Pape and E. P. Plueddemann, in: Silanes and Other Coupling Agents, K. L. Mittal (Ed.),
pp. 105-116. VSP, Utrecht (1992).
54. ISO 10477, Dentistry: Polymer-Based Crown and Bridge Materials, Amendment 1996.
55. B. Arkles, Chemtechl, 766-778 (1977).
Ed. K. L. Mittal
© VSP 2004
Effect of bridging group on the structure of bis-silane

water-barrier films
GUIRONG PAN l, HYUN YIM 2, MICHAEL S. KENT 2,

JAROSLAW MAJEWSKI3 and DALE W. SCHAEFERl•*
1
Department of Chemical and Materials Engineering, and the Polymer Research Center,
The University of Cincinnati, Cincinnati, OH 45221-0012, USA
2
Department 1811, Sandia National Laboratories, Albuquerque, NM 87185, USA
3
Manuel Lujan Jr. Neutron Scattering Center, Los Alamos National Laboratory, Los Alamos,
NM 87585, USA
Abstract—Silane molecular structure, including functionality and the length and character of organic
modifying group, is the key factor that controls the performance of silane films in corrosion protection
applications. In this paper, we review the use of neutron reflectivity to determine the influence
of the bridging group on the swelling of bis-silane films. Our goal is to clarify the relationship
between silane molecular structure, silane film morphology and water barrier performance while
developing a database for optimizing the performance in anti-corrosion applications. In nitrobenzene,
bis-trimethoxysilylpropylamine (bis-amino silane) swells more than bis-triethoxysilylethane (BTSE),
a behavior that we attribute to differences in crosslink density. From swelling studies using deuterated
nitrobenzene, we conclude that the crosslink density of BTSE is higher than that of bis-amino silane.
High crosslink density leads to improved water-barrier performance and explains why bis-silanes are
more effective than monosilanes as water barrier films. Bis-amino silane is found to swell more in
water than in nitrobenzene, which indicates the importance of the character of the bridging group. The
secondary amino group is the driving force for absorbing water. Finally, we observe exchange of the
amine proton by the deuteron when bis-amino silane is swollen in heavy water.
Keywords: Silane; neutron reflectivity; degradation; swelling; bridging group.
1. INTRODUCTION
Because silanes are credible candidates to replace chromates as corrosion inhibitors,
a number of studies have appeared regarding the performance of silane films [1-10].
One important conclusion from current research is that silane films can act as a
water barrier, but they show no electrochemical corrosion inhibition. The formation
e-mail: dale.schaefer@uc.edu
40 G. Pan et al.
of protective films involves numerous chemical and physical processes, so it is

not surprising that the relationship between film chemistry, film morphology and
corrosion performance is not completely understood. Silane molecular structure,
processing protocol and the nature of the metal substrate all affect the water-barrier
performance. Processing variables, including the solution pH, hydrolysis time [1],
solution concentration, curing time and temperature [81 have all been manipulated
to optimize film performance.
Silanes differ from each other regarding functionality and the character of
the organic modifying group. Silanes can be divided into two major classes
based on functionality: mono-silanes (3 hydrolyzable groups, (XO)3-Si-(CH2)3-R)
and bis-silanes (6 hydrolyzable groups, (XO)3-SHCH2)3-R-(CH2)3-Si-(OX)3).
Films formed by certain bis-silanes have substantially better corrosion resistance
properties than films of mono-silanes [2, 3, 8, 11]. Concerning the character of the
organic group, for mono-silanes the organic functional group typically serves as
a coupling agent to bond with a polymer overcoat; for bis-silanes, the length and
character of the bridging group are the dominating factors controlling water-barrier
properties.
This report focuses on the structure of bis-silane films as determined by neu-
tron reflectivity. We study two silanes, bis-triethoxysilylethane (BTSE) and bis-
trimethoxysilylpropylamine (bis-amino). These silanes show contrasting behavior
traceable to the bridging group. The chemical structures are shown in Fig. 1. BTSE
has a short alkyl bridging group. Bis-amino silane has a longer bridging group
containing a central secondary amine group that imparts both functionality and hy-
drophilicity to the molecule. BTSE is known to yield excellent corrosion resistance
on many metal surfaces under appropriate conditions [2, 12]. However, processing
of BTSE is more difficult due to slow hydrolysis of the precursor solution. The bare
corrosion performance (without topcoating) of bis-amino silane is not as good as
that of BTSE, so bis-amino silane cannot be used alone for corrosion protection.
Bis-amino silane does perform well, however, when used as a paint adhesion pro-
moter [13]. It may be possible to combine silanes such as these to obtain both good
paint adhesion, low moisture penetration and good corrosion resistance.
Neutron reflectivity (NR) was measured using the time-of-flight SPEAR instru-
ment at Los Alamos Neutron Scattering Center. Neutron reflectivity is an excellent
OC 2 H 5 OC 2 H 5
CaHgC/ ^ OC 2 H 5
(a) Bis-triethoxysilylethane (BTSE)
OCH3 H OCH3
H 3 C O - S i ^ / ^ / V-^\s/Si-OCH3
H3CO OCH3
(b) Bis-trimethoxysilylpropylamine (bis-amino silane)
Figure 1. Molecular structures of BTSE and bis-amino silane.
Effect of bridging group on the structure of bis-silane water-barrier films 41
tool to probe both the structure of films and the response of films to solvents. NR
is sensitive to structure on length scales from 10 A to 2000 A, enabling a detailed
study of the structure within a silane film. Because neutrons readily penetrate most
materials, one can easily examine interfaces that are buried within a sample.
NR is sensitive to the scattering length density (SLD) profile perpendicular to
substrate. The SLD is determined by the chemical composition and density of the
film. In neutron reflectivity, contrast between different chemical species arises from
variations in the SLD, which is characteristic of each material. In particular, the
large difference in SLD for protons and deuterons is particularly useful for barrier
film research, since it is easy to track solvent penetration by exposure to deuterated
solvents [14]. In our study, we use deuterated nitrobenzene (d-NB) and heavy
water (D2O) to investigate the structure and swelling response of silane films. A 1%
silane solution was prepared by adding the silane to a mixture of DI water and
ethanol. The ratio of silane/DI water/ethanol was 1:9:90 (w/w/w). The solution of
bis-amino silane was aged at ambient temperature for 17 hours before spin coating.
The pH of the solution was adjusted to 7.8 for bis-amino silane and 4.1 for BTSE
by adding acid.
The silane solution was deposited onto the wafers using a glass pipette, and then
spun off for 1 min at 3000 rpm using a Headway spin-coater. The samples were then
dried in an oven at 90°C for 1 h. Following removal from the oven, each sample was
stored in a desiccator until the reflectivity experiments were performed. For detailed
information about the testing procedure and data analysis, we refer the reader to our
previous papers [15,16].
2. THE EFFECT OF THE LENGTH OF BRIDGING GROUP

The crosslink density of silane films is related to the length of the bridging group.
The crosslink density of the film can be probed using solvent-swelling. This
method, however, is complicated by the fact that swelling is affected by both
crosslink density and the monomer-solvent interaction parameter (x). The theory
of Flory and Rehner [17] can be used to estimate the crosslink density from the
equilibrium volume fraction of a swelling solvent, assuming a perfect network (no
dangling chain ends) and prior knowledge of x • Since in our systems x is not known
precisely and the network is not perfect, information regarding the crosslink density
is only qualitative. Nevertheless, the trends are still significant.
Swelling with d-NB provides a first-order measure of relative crosslink density
for different films. The reason is that d-NB interacts mainly through van der Waals
interactions and is a good solvent for most silanes based on solubility assays. Thus,
X can be assumed to be comparable for the silanes studied. Water, on the other hand,
can interact very differently with various silane films depending on the strength
of hydrogen bonding interactions and chemical reactivity of the film; therefore,
swelling with water cannot be used to infer directly the relative crosslink densities
of different films. Our NR studies have shown that BTSE and bis-amino films show
different swelling behaviors in response to d-NB vapor. BTSE film does not swell
42 G. Pan et aL
at all, whereas bis-amino swells substantially. From this behavior we conclude that
BTSE is more highly crosslinked than bis-amino, as one might expect given the
short bridging group for BTSE.
The d-NB swelling of bis-amino film (17%), however, is substantially less than
that of the mono-silane 3-glycidoxypropyltrimethoxysilane (GPS), which is found
to swell 50-60% [18]. This result, along with the results for BTSE, clarifies
the reason for exploring bis-silanes rather than mono-silanes for water barrier
applications — the crosslink density of bis-silanes is higher than that of mono-
silanes. Bis-silanes have 6-fold functionality, whereas mono-silanes have 3-fold
functionality.
3. THE EFFECT OF THE HYDROPHOBICITY OF THE BRIDGING GROUP

The effect of the nature of the bridging group on water barrier performance
was investigated by observing the effects of hydrothermal conditioning, both at
room temperature and 80°C. The extent of swelling and the reversibility upon
drying were examined. Neutron reflectivity data were obtained for the films as-
prepared, after exposure of the film to saturated D2O vapors, and again after
re-drying. Kent et al. [6, 7] proposed that chemical changes in the film could
be measured by comparing the "re-dried" state following conditioning to the
as-prepared film. Physically-bound water or water that is bound by reversible
interactions would eventually be removed during drying, but water (or individual
deuterium atoms) bound by irreversible chemical bonds would remain within the
film. This difference makes it possible to distinguish between deuterium carried
in by physically absorbed water and deuterium incorporated by the hydrolysis of
siloxane bonds.
The secondary amino group of the bis-amino silane increases the hydrophilicity
and also imparts basic character to the film. By contrast, the ethyl bridging group
of BTSE is hydrophobic and neutral. The bridging group influences the hydrolysis
rate of silane molecules, the crosslink density of the film, the structure of the film,
and the overall performance as a water barrier.
3.1. Water swelling

When a silane film is exposed to D2O, the increase in reflectivity compared to
the as-prepared state can be attributed exclusively to physically absorbed water if
the reflectivity returns to nearly that of the as-prepared state upon drying at room
temperature. When exposed to D 2 0 at room temperature, bis-amino film shows a
large increase in reflectivity (Fig. 2a) that is nearly reversible upon drying at room
temperature. The increase in reflectivity is consistent with a volume fraction (0D 2 O)
of D 2 0 in the film of 41%, which is substantially higher than the value of 17%
obtained for d-NB (Fig. 3). We conclude that the amino group increases the
hydrophilicity of the film by hydrogen bonding with water.
More information can be obtained from the SLD profiles (Fig. 2b). For the
swollen system, there is an SLD peak at the air surface indicating a higher
Effect of bridging group on the structure ofbis-silane water-barrier films 43
D20 2 5 * C 1 4 h
A Redried
O As-prepared
0.00 0.02 0.04 0.06 0.08 0.10 0.12

q(A'1)
(a)
I I I I IT! I I I I 1 I | I I I I
- D 2 0 25°C 14h (rough)|

• Redried
As-prepared
0 200 400 600

Distance from silicon surface (A)
(b)
Figure 2. (a) Neutron reflectivity data from bis-amino silane film as-prepared, after exposure to D2O
at room temperature for 14 h and after re-drying. The curves through the data points correspond to the
best fits using model SLD profiles, (b) Best-fit scattering length density profiles corresponding to the
curves through the data in (a) for the samples as-prepared, after exposure to D2O at room temperature
for 14 h and after re-drying. "Rough" in the legend means roughness was included to fit the data. The
film was found to be rough in the swollen state. The calculated volume fraction of D2O in the swollen
film is <?D2O = 0 . 4 1 .
concentration of deuterium (D). This peak could be due either to a more hydrophilic
layer near the air surface (lower monomer-solvent x parameter) or to a lower
crosslink density. Figure 3b shows that no such peak is observed for bis-amino
silane swollen with d-NB. This fact suggests that the peak at the air surface in the
case of swelling with D2O is related to hydrophilicity rather than crosslink density.
Near the silicon substrate surface and within the native oxide layer, there is also
a D-rich layer, although not nearly as pronounced as at the air surface.
Considering the molecular structure of bis-amino silane, we propose that cluster-
ing of the secondary amino groups occurs at the air surface to form a hydrophilic
region. The amino groups have a higher surface energy than hydrocarbon chains
and, thus, they would not be expected to be in excess at the surface in dry air. An
excess of amino groups, however, may result, upon exposure to water (liquid or
vapor) through surface reconstruction. Regarding the slight excess of D at the sub-
strate surface, this excess may simply reflect the presence of hydrophilic sites on
the substrate surface not occupied by silane molecules. It is also possible that there
44 G. Pan et al.
F" 1 —•I 1 1 —— i ^
+ d-NB 25 °C
O As-prepared
[• \ H
<D .3
110
10" 4
It—._—I— _
10" 5
0.00 0.02 0.04 0.06 0.08 0.10 0.12
q (A"1)
(a)
4M I I I I | I I I I | I I I I | I I I I | I I I I | I I I I | I I i I |
- - - d-NB 25 °C (rough)
As-prepared
= 0
<PNB -17
3 2
CO
Ll
0 200 400 600
Distance from silicon surface (A)
(b)
Figure 3. (a) Neutron reflectivity data from bis-amino silane film as-prepared and after swelling to
equilibrium with d-NB. The curves through the data points correspond to the best fits using model
SLD profiles, (b) Best-fit scattering length density profiles corresponding to the curves through the
data in (a) for the samples as-prepared and after swelling to equilibrium with d-NB. "Rough" in the
legend means roughness was included to fit the data. The film was found to be rough in the swollen
state. The calculated volume fraction of d-NB in the swollen film is V?NB = 0-17.
is a small excess of amino groups in the substrate surface due to the upside-down
orientation of silane molecules [19, 20], in which case the molecule is adsorbed
onto the substrate through hydrogen bonding between an amino group and a surface
hydroxy 1 group.
Little or no physically absorbed water exists within BTSE film (see Section 3.2).
The differing hydrophobic/hydrophilic nature of these silanes can also be demon-
strated by contact angle measurements. BTSE film has a high water contact an-
gle (104°), whereas the bis-amino silane film has a water contact angle of 61°.
3.2. Hydrolysis rate and extent

Further insight into the structure and chemistry of the films can be obtained by
conditioning with water vapor at 80°C. At this temperature, it is more likely that
chemical interactions occur than at room temperature. At 80°C, the increased re-
flectivity of the swollen system must be attributed to both physically and chemically
absorbed water. An important question is whether hydrolysis of siloxane bonds (Si-
O-Si) occurs under these conditions.
Effect of bridging group on the structure of bis-silane water-barrier films 45
10u T 1 1 1
•*\
10" 1
_ A +
A
A
A
A
+
+ D 2 0 80°C 48h
> 10" 2
A
f
+ • As-prepared
+
10'3
A
A
A
10" 4
4--H
10- 5 - 1 1 1 1 1 -|
10 20 30 40 50x10
q (A"1)
Figure 4. Neutron reflectivity data from BTSE film as-prepared and after exposure to D2O at 80° C
for 48 h.
10 u — 1 — ~1 1 ~~l l -r H
10-1
Expected change n
• As-prepared
• >
10- 2
+ H2O80°C48h
0
<D
10- 3
<D
a:
1<T*
1 1 1 1 1 1 i + +++
10 20 30 40 50 60 ^0x10"3
q(A-1)
Figure 5. Neutron reflectivity data from BTSE film as-prepared and after exposure to H2O at 80°C
for 48 h. The solid curve shows the expected reflectivity after H2O conditioning if H2O was physically
absorbed.
For BTSE film, an increase in reflectivity is observed upon conditioning at 80°C

for 48 h in D2O saturated air (Fig. 4). Reversibility upon air-drying at room
temperature has not yet been examined. However, the change in reflectivity was
found to be reversible upon evacuation at room temperature. It is interesting that no
change in reflectivity was observed upon conditioning at 80°C for 48 h in HÔ-sa-
turated air as opposed to D2O (Fig. 5). Therefore, the increased reflectivity observed
with D 2 0 conditioning indicates incorporation of D into the film in the form of
either free D2O or Si-OD. Si-OD could be formed either by hydrolysis of siloxane
bonds or residual ethoxysilyl groups (Si-OCH2CH3), or by exchange of H for D
in existing silanol groups (Si-OH). From the result of H 2 0 conditioning of BTSE
film, both physical absorption of water and hydrolysis of siloxane bonds can be
ruled out since both would lead to substantial changes in the reflectivity. For either
case, the reflectivity of H20-conditioned silane would have reached the solid line
in Fig. 5. The SLD values of -OCH 2 CH 3 and -OH, on the other hand, are very
similar, so, negligible change in reflectivity results upon hydrolysis of ethoxysilyl
groups with H 2 0, consistent with Fig. 5. Therefore, for BTSE film conditioned
46 G. Pan et al.
10°
10-1
£> 10-2
>
35
10*
0.00 0.05 0.10 0.15 0.20
q(A"1)
Figure 6. Neutron reflectivity data from BTSE film as-prepared and after exposure to H2O at 80° C
for 5, 9, 27 and 41 days.
10°
+ D2O80°C14h
10-1 -^r-Redried
Q As-prepared
! > 10"2
% 10"3
a:
10-4
10*
0.00 0.02 0.04 0.06 0.08 0.10 0.12
q(A"1)
Figure 7. Neutron reflectivity data from bis-amino film as-prepared and after exposure to D2O
at 80°C for 14 h and after re-drying.
with water vapor at 80°C, we conclude that the only possible chemical change is the
hydrolysis of residual ethoxysilyl groups, and that the siloxane bonds are robust.
Other experiments involving H 2 0 conditioning indicate that the siloxane bonds of
BTSE film are not hydrolyzed upon conditioning at 80°C for up to 42 days, which
is demonstrated by the negligible change in reflectivity (Fig. 6).
For bis-amino film, an increase in reflectivity occurs upon conditioning at 80°C
in air saturated with D2O. This change is not entirely reversible upon re-drying
(Fig. 7), which contrasts with the results for conditioning at room temperature
(Fig. 2). Some D must be chemically incorporated into the re-dried film. We know
from the SLD of the as-prepared film that bis-amino silane is highly condensed with
very few residual methoxysilyl groups. Therefore, the irreversible component of D
incorporation is either in the form of -ND from amine proton exchange (H -> D)
or hydrolysis of siloxane bonds. Even though hydrolysis of siloxane bonds was
not observed with BTSE under these conditions, it may nevertheless occur at a
measurable rate for bis-amino due to a higher local concentration of water in this
more hydrophilic film. However, we actually observe less swelling at 80°C (Fig. 7)
than at 25°C (Fig. 2a), which is inconsistent with hydrolysis of Si-O-Si at 80°C.
[Dangling chains from

lunhydrolyzed ethoxv group
(a) BTSEfilm:more dangling chains, shorter distance between crosslinks,

higher crosslink density.
(b) Bis-amino silanefilm:fewer dangling chains, longer distance between

crosslinks, lower crosslink density.
Figure 8. Schematics of network structures of BTSE (a) and bis-amino silane (b) films.
The irreversible D incorporation is most likely exchange of the amine proton with a
deuteron from D 2 0.
By comparing the SLD values of the as-prepared dry films with calculated values,
we conclude that the degree of condensation within bis-amino silane film is higher
than that for BTSE film. Therefore, less dangling ends would be expected in the bis-
amino film compared to BTSE. Figure 8 shows difference in the network structures
of BTSE and bis-amino silane schematically. Although there are more dangling
chains in BTSE than in bis-amino silane, the distance between crosslinks is larger in
bis-amino silane, due to the longer bridging group. Therefore, the overall crosslink
density is lower for bis-amino than for BTSE. Moreover, in bis-amino silane film,
secondary amine group attracts water by hydrogen bonding. These two factors lead
to poorer overall water-barrier performance of bis-amino silane film compared to
that of BTSE.
During precursor conditioning and film deposition, not only the extent of hydrol-
ysis, but also the rate of hydrolysis is affected by the secondary amine group. Bis-
48 G. Pan et al.
amino silane hydrolyzes much faster than BTSE because of the basic character of
the secondary amine, which can act as a catalyst for the silane hydrolysis. Another
reason for slow BTSE hydrolysis is that BTSE has a larger alkoxy group (ethoxy
group) than bis-amino silane (methoxy group). Under acid catalyzed conditions, the
hydrolysis of methoxy group is five times faster than that of ethoxy group [21].
4. CONCLUSIONS
Crosslink density and hydrophobicity are the two main factors that determine the
water barrier performance of the silane films. Regarding the relationships among
the crosslink density, the molecular structure, and the performance the following
observations can be made:
• Bis-silanes have greater potential as water barriers than mono-silanes due to the
higher crosslink density,
• BTSE has a higher crosslink density than bis-amino silane due to the shorter
length of the bridging group.
• The character of the bridging group is very important with regard to water-barrier
properties. For bis-amino silane, the driving force for absorbing water is mainly
from the hydrophilic secondary amine group.
• The secondary amine group also brings basicity to the silane film. The amino
group acts as a catalyst to accelerate the hydrolysis of the precursor solution,
leading to a more condensed silane film.
• The amino groups also appear to concentrate water at the air surface of the film.
Acknowledgements
Work at the University of Cincinnati was sponsored by the Strategic Environmental
Research and Development Program (www.serdp.org). We thank Professor Wim
van Ooij for introducing us to this problem. This work benefited from the use
of Surface Profile Analysis Reflectometer (SPEAR) at Lujan Neutron Scattering
Center at Los Alamos National Laboratory and was supported by Los Alamos
National Laboratory under DOE contract W7405-ENG-36, and by the DOE Office
of Basic Energy Sciences. Sandia is a multiprogram laboratory operated by Sandia
Corporation, a Lockheed Martin Company, for the United States Department of
Energy under contract DE-AC04-94AL85000.
REFERENCES
1. D. Q. Zhu and W. J. van Ooij, /. Adhesion ScL Technol 16, 1235-1260 (2002).
2. W. J. van Ooij and T. Child, Chemtech 28, 26-35 (1998).
3. D. Susac, X. Sun and K. A. R. Mitchell, Appl Surface ScL 207, 40-50 (2003).
4. V. Subramanian and W. J. van Ooij, Corrosion 54, 204-215 (1998).
6. M. S. Kent, W. F. McNamara, D. B. Fein, L. A. Domeier and A. P. Y. Wong, /. Adhesion 69,

121-138(1999).
7. M. S. Kent, W. F. McNamara, D. B. Fein, L. A. Domeier and A. P. Y Wong, J. Adhesion 69,
139-163 (1999).
8. A. Franquet, C. Le Pen, H. Terryn and J. Vereecken, Electrochim. Acta 48, 1245-1255 (2003).
9. U. Bexell and M. Olsson, Surface Interface Anal 31, 212-222 (2001).
10. U. Bexell and M. Olsson, Surface Interface Anal 31, 223-231 (2001).
11. V. Subramanian, Ph.D. thesis, University of Cincinnati (2000).
12. S. K. Jayaseelan and W. J. van Ooij, J. Adhesion ScL Technol 15, 967-991 (2001).
13. W. J. van Ooij and D. Zhu, Corrosion 57, 413-427 (2001).
14. T. P. Russell, Physica B 221, 267-283 (1996).
15. M. S. Kent and H. Yim, in: Silanes and Other Coupling Agents, K. L. Mittal (Ed.), Vol. 2,
pp. 115-125. VSP, Utrecht, The Netherlands (2000).
16. G. Pan, H. Yim, M. S. Kent, J. Majewski and D. W. Schaefer, J. Adhesion Sci. Technol 17,
2175-2189(2003).
17. P. J. Flory, Principles of Polymer Chemistry. Cornell University Press, Ithaca, NY (1953).
18. H. Yim, M. S. Kent and J. S. Hall, J. Phys. Chem. B 106, 2474-2481 (2002).
19. A. Franquet, J. De Laet, T. Schram and H. Terryn, Thin Solid Films 384, 3 7 ^ 5 (2001).
20. M. R. Horner, F. J. Boerio and H. M. Clearfield, J. Adhesion ScL Technol 6, 1-22 (1992).
21. E. D. Osterholtz and E. R. Pohl, J. Adhesion Sci. Technol 6, 127-149 (1992).
Ed. K. L. Mittal
© VSP 2004
Amino- and bis-silane pre-treatments for adhesive bonding

of aluminium
BENTE GILBU TILSET*, FABRICE LAPIQUE, ASTRID BJ0RGUM

and CHRISTIAN J. SIMENSEN
SINTEFMaterials Technology, P.O.Box 124 Blindern, N-0314 Oslo, Norway
Abstract—The goal of the present study was to understand the relative importance of chemical
bonding and interfacial stability when thin silane layers were used as pre-treatments for adhesive
bonding of aluminium with epoxy adhesives. The main focus was on hydrolysed y-APS and BTSE,
used either alone or in combination. Samples were subjected to wedge tests, acetic acid salt spray
tests (ASSTs) and surface characterisation. BTSE used alone was found to be the most efficient, its
performance approaches that of a standard pre-treatment developed by the Boeing Company for use
on airplanes. BTSE might reinforce the adhesive bond by interacting with silanes in the adhesive.
Samples were bonded using silane-containing and silane-free adhesives to test this hypothesis. The
results suggested that BTSE improves the durability of adhesive bonds by decreasing the hydrolysis
and/or corrosion rate at the adhesive-metal interface, while /-APS increases the original adhesive
bond strength by interacting chemically with the adhesive. Bis-silanes with and without amino
groups were tested in an attempt to combine chemical bonding and corrosion protection. The
samples with the highest amine concentration initially gave the strongest adhesive bond, while the
best durability was obtained for samples providing good protection against interfacial hydrolysis and
corrosion. It seems that it is not necessary for a silane to react with the adhesive in order to obtain
a durable adhesive bond, and judging from the current results, two non-functional bis-silanes, BTSE
and bis(trimethoxysilyl)hexane, actually show the greatest promise with respect to replacing current,
chromium-containing pre-treatments.
Keywords: Adhesive bond durability; aluminium pre-treatments; y-APS; BTSE; amino-silanes; bis-
silanes.
1. INTRODUCTION
Owing to its high strength and low weight, the use of aluminium in the automotive
industry is increasing. When bonding aluminium, traditional methods such as
riveting and welding are often associated with problems due to weakening and
e-mail: Bente.G.Tilset@sintef.no
52 B. G. Tibet et al.
corrosion of the metal. These problems can be avoided by structural adhesive

bonding. Adhesive bonding is presently used for aluminium space frames for
advanced sports cars, like the Lotus Elise, Opel Speedster and Aston Martin
Vanquish. However, pre-treatments are quite demanding, and thus too costly for
large production series. In addition, most high quality pre-treatments involve
the use of hexavalent chromium (CrVI), which will be banned from use in the
European automotive industry in 2007 due to environmental, health and recycling
requirements [1, 2]. Thus, alternative, chromium-free pre-treatments must be
developed and assessed for industrial applications. Several requirements must be
fulfilled: the pre-treatment must provide a chemically inert and mechanically strong
interface between the adhesive and substrate, in addition to being inexpensive, easy
to apply and environmentally benign.
One type of pre-treatment that may fulfil these requirements is silanisation.
Silanes have the general formula X3Si(CH2)„Y, where X is a hydrolysable group
and Y is an organofunctional group that can react with the adhesive. During the
process, the silane is hydrolysed, giving Si-OH groups that interact with a metal
(oxide) surface via hydrogen bonds or react with surface M-OH groups to form
covalent Si-O-M bonds. In addition, they react with neighbouring silane molecules
to form a Si-O-Si network on the sample surface. The extent of cross-linking
can be increased by employing bis-silanes, with two hydrolysable silane centres.
Their general formula is X3Si(CH2),7 Y(CH2)mSiX3. They may contain one or more
organofunctional groups Y that can react with the adhesive employed. However, if
the bis-silane is used together with an organofunctional silane, this is not necessary,
as demonstrated for coating applications by van Ooij and co-workers [3-6].
The goal of the present study was to understand the relative importance of
chemical bonding and interfacial stability when thin silane layers were used as
pre-treatments for adhesive bonding of aluminium with epoxy adhesives. Both
functional and non-functional mono- and bis-silanes were used. The functional
mono- and bis-silanes have amino groups which can react with the epoxy groups in
the adhesive, while the bis-silanes (both functional and non-functional) can provide
a more cross-linked film with more stable bonding to the substrate material.
Initially, y-aminopropyltriethoxy silane (y-APS) was used as the functional and
bis(triethoxysilyl)ethane (BTSE) as the non-functional bis-silane. Their structures
are shown in Table 1. Different combinations of these two silanes were used as
surface pre-treatments for coating and adhesive bonding of aluminium.
It has been reported that y-APS can be adsorbed "upside-down" so that the amino
group instead of the silanol groups interacts with the substrate [3-5, 7, 8]. The
"upside-down" adsorption is expected to decrease with decreasing pH in the silane
solution [9]. The thermal treatment of BTSE after deposition is important for the
degree of cross-linking and thus, for the density of the obtained layer [10, 11]. This
is assumed to hold true also for y-APS. Furthermore, van Ooij and co-workers [3, 5]
have reported that y-APS is easily removed by rinsing with water, while BTSE is
more stable. Based on the above considerations, the pH of the y-APS solution, heat
treatments and rinsing procedures were systematically varied in the experiments.
Amino- and bis-silane pre-treatments for adhesive bonding ofAl 53
Table 1.
Silanes and adhesive employed for pre-treatment of aluminium AA6060
Silane name, supplier Structure

and abbreviation
y - Aminopropyltriethoxysilane OC 2 H 5
Witco, SilquestA-1100
K-APS
Bis(triethoxysilyl)ethane H5C20
H5C20 \ /OC2H5
ABCR \ ^ \ ^Si
BTSE OC2H5
HcCoO \
OC2H5
y-Glycidoxypropyltrimethoxysilane H3CO
Witco, SilquestA-187
X-GPS HoCcr \
Bis(trimethoxysilylpropyl)-
H3C0
ethylenediamine
ABCR, 62% in methanol
HoCCT \
"diamine" OCHo
Bis(trimethoxysilylpropyl)amine H3CO OCH3

ABCR, purity 95% /
"amine" hUCCT \ / ÔCH 3
OCH3 H3CO
Bis(trimethoxysilyl)hexane H3CO
H.CO
ABCR
"hexane"
H3CO \
OCH3
Bis(triethoxysilyl)nonane H5C2q
ABCR
HcCoO \
OC 2 H 5 H5C20
Diglycidylether of bisphenol A
Ciba: Araldite CY 219, epoxy resin,
MW < 700
DGEBA
It has been reported [9] that hydration of the oxide is the weak link in adhesive-
aluminium system. Improved corrosion resistance is, therefore, vital, and the quality
of the pre-treatments was evaluated from sample behaviour in wedge tests and in
acetic acid salt spray tests (ASSTs).
Most adhesives contain silanes to improve their interactions with the substrate.
It is deemed possible for these silanes to interact with those from the silane pre-
treatments. Both silane-free and silane-containing adhesives were employed to test
this hypothesis.
54 B. G. Tilset et al.
Based on the results with y-APS and BTSE, a second series of functional and non-
functional bis-silanes was chosen, where the size of the alkyl group and number of
amino groups between the silane centres were varied. The names and structures of
the silanes chosen are given in Table 1, along with the abbreviations used throughout
this paper. Attempts were also made to increase the interaction between the silanes
and the epoxy adhesive by coating with the epoxy constituent of the adhesive,
diglycidylether of bisphenol A (DGEBA), prior to applying the adhesive.
Finally, the role of silane pre-treatments in adhesion and failure of bonded
aluminium assemblies is discussed.
2. EXPERIMENTAL
2 J. Materials and chemicals
The substrates used were extruded aluminium AA6060-T6 profiles (2 mm thick and
110 mm wide). The composition of the alloy was Al, 0.51% Mg, 0.43% Si and
0.18% Fe. Two different commercial epoxy adhesives were used: Araldite 2014
and Araldite 2011. Araldite 2014 is opaque and grey and contains both silanes and
fillers. It was chosen because it is stiff and thus transfers mechanical forces to the
interfacial region during wedge tests. Araldite 2011 was chosen because it contains
no silanes. It is transparent and yellow and probably contains no fillers. 1% y-gly-
cidoxypropyltrimethoxysilane (y-GPS) was added to Araldite 2011 for some of the
wedge test and ASST samples to determine the effect of silane addition. )/-GPS
may react with y-APS via the epoxy group, in a fashion similar to the reaction
between the adhesive and y-APS.
The structures, names and presently used abbreviations of the silanes employed
and DGEBA are shown in Table 1, along with supplier information. Absolute
ethanol (>99.5%) and approx. 96% ethanol were obtained from Arcus, while
acetone (p.a. >99.5%), methanol (p.a. >99.8%), NaOH (p.a. >99%), CH3COOH
(p.a. >99.8%) and HNO3 (p.a. 65%) were from Merck. Distilled water was used
for silane hydrolysis and sample rinsing.
2.2. Pre-treatment
The deformed top layer of the AA6060 substrate was removed, together with
contaminants, to reduce the corrosion rate of the aluminium. Substrates were
cleaned in acetone, rinsed in water and etched for 50 s in 10 wt% NaOH(aq) at 60°C
before rinsing in water, desmutting in 65% HNO3 at room temperature and thorough
rinsing in water. Samples were stored in distilled water (pH 5) until silanisation,
which was performed within six hours after etching and desmutting. During the
process, about 7 /xm of the AA6060 top layer was removed. The oxide layer formed
was between 3 and 10 nm thick, thus very little oxidation occurred during sample
storage in water.
For pre-treatments, 1 wt% solutions, with compositions as specified in Table 2,
were used. The pH of the water used for making the BTSE, "hexane" and "nonane"
Amino- and bis-silane pre-treatments for adhesive bonding of Al 55
Table 2.
Mixing ratios for silane solutions
Silane Distilled water (wt%) Alcohol (wt%)
y-APS 9 90
BTSE 33° 66
"diamine" 9 90
"amine" 9 90
"hexane" 33 a 66
"nonane" 33* 66
For all silanes, 1 wt% solutions were prepared. The alcohol was either ethanol
or methanol, depending on the alkoxy group on the silane.
a
pH adjusted to 3.5 with CH3COOH.
Table 3.
Pre-treatment procedures for samples with BTSE and/or / - A P S
Sample name BTSE H20 125°C K-APS H20 125°C
BTSE + + + — — —
y-APS* - - - + - -
BAl f l + + + + - -
BA2^ + + + + + +
BA3* + + — + + —
BA4*> + + + + + -
BTSE refers to 1 min immersion in BTSE solution, H 2 0 to rinsing in water, 125°C to 15 min curing
at 125°C and y-APS to 1 min immersion in )/-APS solution.
Symbols: ( + ) in a column means that the treatment was employed, (—) that it was not. ( - ) in the
125°C column refers to drying at ambient conditions.
fl
pH(y-APS)=11.8.
^pH(y-APS)= 1.8.
solutions was adjusted to 3.5 with acetic acid. The pH values of the y-APS solutions
were 11.8 or 1.8 (adjusted with HNO3). "Diamine" and "amine" solutions were
adjusted to pH 5.0 with acetic acid. All silane solutions were stirred for half an
hour at room temperature after mixing to allow hydrolysis prior to use. Etched
and desmutted samples were rinsed in the appropriate alcohol to remove the excess
water from the surfaces prior to silanisation.
For samples with /-APS and BTSE, the silanes were applied in various combi-
nations as shown in Table 3, with immersion in silane solutions for 1 min, followed
by rinsing in water for 10 s or drying at room temperature. Curing was performed
either by heating at 125°C for 15 min or by drying at room temperature for a min-
imum of 1 h. Samples were always coated or adhesively bonded within 24 h after
preparation.
Samples with "diamine", "amine", "hexane" and "nonane" were immersed in the
relevant silane solutions for 1 min, rinsed in water for 10 s and cured at 125°C for
15 min. Some of the samples with "diamine" and "amine" were coated with a 1 wt%
solution of DGEBA in absolute ethanol by immersion for 1 min, followed by drying

at room temperature.
A standard pre-treatment method developed by Boeing [12] was used as a
benchmark in wedge tests with Araldite 2014. The pre-treatment consists of etching
in a solution of 33 g/1 sodium dichromate and 330 g/1 sulphuric acid (FPL etch) for
15 min at 68°C, followed by phosphoric acid anodising (PAA) at 10 V (DC) for
20 min in 10 wt% phosphoric acid solution at 25°C, using a computer aided pulse
plating (CAPP) unit, as previously described in Ref. [13].
2.3. Wedge test

Wedge test samples (Fig. 1) were 25.4 mm wide and 152.4 mm long, with 2-mm-
thick AA6060 backing plates on both sides to control the stiffness of the samples.
The adhesive layer was 0.10 mm thick, controlled by steel spacers. The length
of the bondline was 133 mm. Assemblies with Araldite 2014 were cured at 64°C
for 4 h [14] and those with Araldite 2011 (with and without y-GPS) at 60°C for
4 h. The wedge was inserted after cooling, at a rate of 10 mm/min. Samples were
allowed to relax at room temperature for 24 h before exposure to an environment
with 100% relative humidity at 40°C. The total length of the crack was measured on
both sides of each assembly. Six samples were prepared with each pre-treatment;
thus, the reported crack growth is an average of 12 measurements. In addition to
silanised samples, AA6060 samples etched and desmutted only, were prepared and
used as references, as well as samples pre-treated with FPL etch followed by PAA.
2.4. Acetic acid salt spray test (ASST)

For acetic acid salt spray testing (ASST), pre-treated AA6060 substrates (size:
110 x 50 x 2 mm3) were bar coated with the adhesive, giving an adhesive film with
thickness varying between 40 and 200 /xm. Both silanised samples and reference
samples (etched and desmutted only) were used. After curing (the same procedure
as for wedge test samples), two scribes were made in each sample, and edges were
covered with an adhesive tape. Samples were then exposed in the salt spray cabinet
at 35°C and pH 3.1-3.3 in accordance with ASTM G85-02el.
2.5. Characterisation techniques

Surfaces of etched and desmutted AA6060 were characterised by Scanning Electron
Microscopy (SEM) (JEOL JSM-5900LV), while Electron Probe Microanalysis
Figure 1. Wedge test assembly after exposure to 100% relative humidity at 40° C.
Amino- and bis-silane pre-treatments for adhesive bonding ofAl 57
(EPMA) (Cameca SXIOO) was used to determine the chemical composition of

adhesive/metal interfaces after wedge tests and ASSTs.
Time-of Flight Secondary Ion Mass Spectroscopy (ToF-SIMS) (Cameca ToF-
SIMS IV) was employed to investigate the chemical composition of silanised
surfaces. Intermittent sputtering with Ar+ was used in an attempt to determine the
silane layer thickness. It was assumed that a layer was traversed when the intensities
of the relevant fragments had decreased to half of their maximum values.

3.L Characterisation of pre-treated surfaces
An SEM image of an etched and desmutted AA6060 surface is shown in Fig. 2.
Intermetallic particles and pits after particle removal are readily seen.
ToF-SIMS was used to study the samples coated with combinations of BTSE
and y-APS. In ToF-SIMS, the detection of the SiOAl fragment is used as an
indication that silanes have reacted with a hydrolysed Al surface and formed
covalent bonds [15, 16]. However, the current alloy contains about 0.4% Si.
Thus, any change in the concentration of the SiOAl fragment was masked by its
presence in the natural oxide. The SiOSi fragment was also masked. It would
otherwise be used as an indication of cross-linking in the silane top layer. These
features are shown in Fig. 3, which are typical ToF-SIMS depth profiles for a
sample consecutively coated with BTSE and y-APS. The intensities of the SiOAl
and SiOSi fragments are more or less constant during sputtering from the silane
Figure 2. SEM image of the AA6060 surface after etching in 10 wt% NaOH at 60° C and desmutting
in 65% HNO3. Small, bright, intermetallic particles, probably Al-Fe-Si phases, are present in the
surface. Rounded features are etching pits.
Silane Oxide
4
10
: CH4N
^ 1H33
AlO '_
10
AlOSi
2
10
\ _
1
•\
SiOSi
^ 10 •\
10°
[ SiOC2H5
Y x
l ft A
\/V'\
V
A/ \
U_/V
/•*
/
IA
\
100 200
Time (s)
Figure 3. Typical ToF-SIMS depth profiles of AA6060 surfaces with BTSE and y-APS. Approximate
sputtering rate: 0.0065 nm/s (measured on T1O2).
top layer into the hydrated aluminium oxide layer, which is represented by the
fragment AlO in Fig. 3. The AlO intensity increased gradually and reached half
its saturation level after 50-60 s, indicating a silane layer thickness of 0.3-0.4 nm.
Prolonged sputtering showed that the hydrated oxide layer was 3 to 10 nm thick.
CH4N and S1OC2H5 are fragments characteristic of y-APS and BTSE, respectively
(data not shown). The intensity of these fragments decreased in a similar way. Thus,
despite the consecutive deposition, our results indicate that the silanes appear as a
single, mixed layer and not as separate layers.
3.2. Wedge tests andASSTs

3.2.1. BTSE and y-APS with Araldite 2014. Wedge test results for BTSE and
y-APS samples bonded with Araldite 2014 are shown in Fig. 4. The AA6060
reference (etched and desmutted only) performed worst, with total failure of all
samples after 4 h. Due to the standard deviations in the measurements, the quality
of 3/-APS, BA3 and BA2 is regarded to be equal. BA4 and BTSE are also equal
and slightly better than BA1. In all cases, the initial crack (due to wedge insertion)
was within the adhesive. Upon exposure to humid environment, crack propagation
changed from cohesive mode to interfacial mode for most samples, as observed
optically. For the best silane pre-treatments, BTSE and BA4, a few samples still
showed failure within the adhesive during exposure. Their durability was close to
that of the benchmark samples with FPL etch + PAA pre-treatment, which all failed
within the adhesive.
Interfacial failure in wedge tests is assumed to be caused by hydration of the
interfacial region. The corrosion protection properties of the pre-treated samples
were, therefore, tested by exposing the coated samples to the ASST environment
for about 4 months (approx. 2900 h). In ASST, samples with y-APS (Fig. 5a)
performed worst, followed by BA3. Large blisters were found on both samples.
Amino- and bis-silane pre-treatments for adhesive bonding of A! 59
1 10 100 1000
Time (h)
Figure 4. Wedge test results using Araldite 2014 for samples with BTSE and/or y-APS, Average
standard deviations are noted on the curves. AA6060 denotes the reference (etched and desmutted
only) sample and FPL + PAA the benchmark sample.
Some small blisters appeared on BA4, while samples BA1, BA2, AA6060 reference
and BTSE (Fig. 5b) were still intact after 4 months of exposure. It was surprising
that the AA6060 reference sample, which performed poorly in the wedge test,
performed so much better than most of the samples pre-treated with y-APS, either
in combination with BTSE or alone. A close inspection showed that blistering only
occurred when corrosion had proceeded to a large degree, and it was actually the
corrosion products that lifted the adhesive off the aluminium surface. This is shown
in the case of y-APS in Fig. 6, which is a cross section of the scribe at the position
marked in Fig. 5a. Since samples with y-APS performed poorly, it was suggested
that the amino groups might facilitate the transport of Cl~ from the scribes along
the adhesive-aluminium interface, and that these ions might increase the rate of
corrosion underneath the adhesive.
EPMA images at the crack tip of the y-APS sample (marked in Fig. 6) are shown
in Fig. 7. The Al and O plots clearly show the presence of an aluminium oxide
or hydroxide phase on the delaminated adhesive. At the magnification available,
corrosion products are not detected in the area with intact adhesion. From the
CI image, it is evident that Cl~ diffuses into the adhesive. The highest Cl~
concentration is found at the Araldite 2014 surface. Preferential diffusion along
the adhesive/aluminium interface would result in an increased CI" concentration
in the adhesive near the interface. This is not detected. Thus, the theory that CI"
transport in the interfacial region is facilitated by amino groups on the silanes is not
supported by the present investigation.
A more probable process is that once the adhesive/aluminium interface is partially
exposed, as it is around the scribe, acid from the salt spray reacts with free amino
groups present at the interface, giving positively charged NH^" groups. These will
attract Cl~, which will increase the dissolution rate of aluminium [17]. Corrosion
Figure 5. Photographs taken after ASST on samples pre-treated with (a) y-APS and (b) BTSE and
coated with Araldite 2014.
200 \xm Area in Fig. 7
Figure 6. Back-scattered electron (BSE) image of the scribe cross section marked in Fig. 5a. The
sample was pre-treated with /-APS and subjected to AS ST.
products (hydrated alumium oxides) then lift the adhesive, exposing more of the
interface to hydrolysis and corrosion. This underfilm corrosion does not occur on
samples without aminosilane (AA6060 reference and BTSE). In samples with both
BTSE and y -APS, BTSE gives partial protection of the aluminium, especially when
cross-linking at 125°C was performed before application of y-APS. This pertains
to BA1, BA2 and BA4, which all perform better in the ASST than y-APS and BA3.
The superior performance of BTSE in wedge tests with Araldite 2014 was a
surprise, since BTSE has no functional groups that can react with the adhesive.
Furthermore, the formation of an interpenetrating network with the adhesive was
considered improbable, since our ToF-SIMS results indicated only near-monolayer
coverage. One possible explanation for the good performance is that BTSE can react
with silanes that are present in Araldite 2014 and form a strong covalent network
which would stop crack propagation.
3.2.2. BTSE and y-APS with Araldite 2011. To check the theory that BTSE
interacts with silanes in the adhesive, a silane-free adhesive (Araldite 2011) was
employed for wedge and ASST samples with BTSE, y-APS and AA6060 reference
20 mn o 2pfim^
*v:; s?e ••***•vif- *•
~'..;v * V ^ V ^ V V : 4" • s J-*. ' • '• ..r*, v ' , • j ^ * * '% % •
• > ; • • • •
M%V*$ . * ( | ^ : ^ . ; ; • k *;#,;^ ^ # J
Figure 7. EPMA images of BSE, Al, CI and O, in the crack tip area marked in Fig. 6. The maps
of Al and O show the presence of alumina on both aluminium and adhesive sides of the crack
tip. The CI map shows a decreasing concentration of Cl~ from the adhesive surface towards the
adhesive/aluminium interface.
Figure 8. Wedge test results for Araldite 2011 with (+ in the legend) and without added y-GPS.
The standard deviation in crack length is about 13 mm. Differences are seen in sample durability.
All BTSE + samples had reached total failure after 48 h, while one BTSE sample had still not failed
completely after 168 h exposure at 40°C and 100% RH. AA6060 denotes the reference (etched and
desmutted only) sample.
(etched and desmutted only). To check the effect of silane addition, the experiments
were repeated with 1% y-GPS mixed into the adhesive. The results from wedge
tests are shown in Fig. 8.
62 B. G. Tibet tied.
Figure 9. Representative samples after ASST for 1000 h (42 days). Samples where y-GPS was
added to the adhesive, Araldite 2011, are marked by a "+" after the name of the pre-treatment.
AA6060 denotes the reference (etched and desmutted only) sample.
Araldite 2011 itself has poor adhesion to aluminium; the initial failure (before
exposure) was interfacial for most samples. Cohesive failure (within the adhesive)
was only found for samples that were pre-treated with y-APS and had no y-GFS
added to the adhesive. All wedge samples failed interfacially upon exposure to
the humid environment. This was confirmed by EPMA for one sample, where
intermetallic particles were found both on the aluminium and adhesive sides of the
failed joint.
The initial crack growth was almost equal for all samples, while differences were
found in sample durability. No systematic variations were found for samples with
and without y-GPS added to the adhesive. With regard to durability, the relative
performance is the same as that found with Araldite 2014; BTSE performs better
than y-APS, which is only slightly better than the AA6060 reference.
Larger differences are found in ASSTs. Representative samples from each series
are shown in Fig. 9. Without y-GPS added to the adhesive, etched and desmutted
AA6060 lost adhesion within one week and BTSE samples after two weeks. The
sample pre-treated with y-APS outperformed both, even though extensive blistering
was observed after six weeks (1000 h) of exposure. For all pre-treatments, the
addition of silane to the adhesive enhanced the performance in the ASST.
The poor adhesion of Araldite 2011 to aluminium is also reflected in the ASST
results. Araldite 2011 absorbs water and expands during the tests, giving rise to
forces that lift the adhesive off the surface. In addition, water diffusing through the
adhesive contributes to hydrolysis and thus weakening of the interface. This is the
reason for blistering in areas far from the sample scribes. For the AA6060 reference,
the interface is sufficiently weak that the forces due to expansion of the adhesive
rapidly lead to loss of adhesion. For samples with stronger initial adhesion, some
hydrolysis/corrosion must take place before the interface is sufficiently weakened
to result in adhesive delamination. The best adhesion in ASST is obtained with
/-APS, probably because the amino group of the silane has reacted with the epoxy
groups in the adhesive. BTSE also increases the adhesion relative to the AA6060
reference, probably because it increases the interfacial hydrophobicity and thereby
prevents hydrolytic weakening.
When comparing wedge and ASST results, it is important to bear in mind that
there are at least two factors determining the durability: initial adhesion and rate
of hydrolysis/corrosion at the interface. Thus, the differences in time scale are
important when evaluating the results of these tests. In wedge tests, the samples with
Araldite 2011 fail rapidly, with little time for water to diffuse through the adhesive,
as it does in the ASST. Most of the water in the adhesive/aluminium interfacial
region must be diffusing along the interface.
For the AA6060 reference, initial adhesion is poor. Combined with rapid
hydrolysis of the interface, this explains the poor performance in wedge tests. When
y-APS is used, the initial adhesion is good. This is seen from both wedge tests,
where y-APS is the only pre-treatment for which initial failure is partly within the
adhesive and from ASSTs, where y-APS has best durability. However, y-APS is
hydrophilic and produces a basic environment upon hydrolysis, which leads to a
weakening of the interfacial region. This combination of bonding and hydrophilicity
gives medium performance in wedge tests. For BTSE, ASST results indicate
an initial adhesion between that of the AA6060 reference and y-APS samples.
Compared to y-APS, BTSE provides a more hydrophobic surface, which reduces
the rate of interfacial hydrolysis and thereby enhances wedge test performance.
For all ASST samples, the addition of y-GPS to Araldite 2011 improves durabil-
ity. This may either be due to an increase in the degree of covalent bonding between
the adhesive and the aluminium substrate or to a decrease in the water permeability
and absorption capacity of the adhesive. In any case, the effect found in ASST is
largely masked by the dominating effect of interfacial hydrolysis in wedge tests.
3.2.3. Bis-silanes. Based on the considerations above, it was expected that the
best performance would be obtained with a silane providing good anchoring to
the aluminium surface, a hydrophobic interface and a chemical interaction with
the epoxy. Mono- and di-amino bis-silanes were expected to give enhanced
performance, since they might form a hydrophobic siloxane network with multiple
bonds to the aluminium surface [3, 5, 6, 17], as well as covalent bonds with the
adhesive. Comparisons were made with non-functional bis-silanes with similar
lengths of the organic chains. It was expected all that these bis-silanes would
outperform BTSE. If Araldite 2014 was used, failure would be cohesive, within
the adhesive, for all the pre-treatments. Thus, the results would be non-conclusive.
To distinguish the effects of the different silanes, the less durable adhesive,
Araldite 2011, was used.
The "nonane" bis-silane did not interact well with the adhesive. During bar
coating of the ASST samples, the adhesive partially retracted from the sample,
forming a quite uneven coating. The initial adhesion was very poor; in wedge
tests, the "nonane" samples failed interfacially upon wedge insertion; and in ASSTs,
140
0 50 100 150 200

Time (h)
Figure 10. Wedge test results for Araldite 2011 with various bis-silane pre-treatments. The standard
deviation in crack length is about 13 mm for the BTSE sample and in the range 5 to 9 mm for the
others. The "nonane" sample is not included in the figure, since it failed totally upon wedge insertion.
some delamination occurred already during scribing of the sample. "Nonane" has
a rather long alkyl chain, without any functional groups. This probably makes
the critical surface tension of the "nonane"-pre-treated aluminium too low for
efficient spreading of the epoxy adhesive, giving rise to very weak interactions,
as observed.
The results of wedge tests are shown in Fig. 10, where the plot for BTSE is
included to facilitate comparisons with the previous pre-treatments. All samples
performed better than BTSE in the early stages of exposure. Due to large standard
deviations, "diamine", "amine", "amine" + DGEBA and "diamine" + DGEBA
samples must be considered equal, with the "hexane" sample performing best.
In ASSTs, "diamine" performed best of the bis-silanes (Fig. 11), followed by
"amine", "diamine" + DGEBA, "amine" + DGEBA, "hexane" and "nonane". When
evaluated after the same exposure time, none of these samples was better than
y-APS. One reason might be that y-APS has a higher concentration of amino
groups per unit surface area than the other silanes and, thus, has a more efficient
interaction with the epoxy groups of the adhesive.
The intermediate DGEBA layer was originally assumed to react efficiently with
the applied aminosilane to form a stable epoxy surface which would interact closely
with Araldite 2011. This was, however, not the case; ASST results show that the
adhesion/hydration resistance is reduced when DGEBA is used together with these
aminosilane pre-treatments.
Despite the previously mentioned poor adhesion, the "nonane" sample performed
better than both the AA6060 reference and BTSE samples during ASST exposure.
The reason might be that the "nonane" pre-treatment gives a very hydrophobic
surface which efficiently repels water from the adhesive-aluminium interface and
thus stabilises the rather weak interface.
Both ASST and wedge test results with Araldite 2011 show that all the bis-silanes
in the current series, except "nonane", provide better initial adhesion than BTSE.
Diamine Nonane
Figure 11. Best (left) and worst (right) samples after ASST for 288 h (12 days). For the "nonane"
sample, some delamination occurred already when the scribes were made.
However, when durability in wedge tests is important, once again the silane which
has a reasonable initial adhesion, combined with the best interface protection
outperforms the others, irrespective of whether or not it can form covalent bonds
with the adhesive.
32 A. The role of silanes in adhesion and failure of bonded assemblies. If a pre-

treatment is going to replace current pre-treatments for structural adhesive bonding,
it must give excellent bond durability during exposure and loading. The present
results show that two conditions must be fulfilled: (1) The initial adhesion must be
sufficient and (2) the interface must be stable, especially with respect to hydrolysis
and corrosion.
When using a silane pre-treatment, strong initial adhesion can be obtained by
formation of covalent bonds or an interpenetrating network (IPN) between the silane
and the adhesive. In either case, it is necessary that the silane and adhesive are
compatible in terms of surface energies and wettability [5, 6]. For the current pre-
treatments, covalent bonding is possible for the silanes containing amino groups,
and ASST results with Araldite 2011 imply that the initial adhesive bond strength
increases ("amine" < "diamine" < y-APS) with increasing surface concentration
of amino groups at the adhesive-aluminium interface.
After pre-treatment, silanes are bonded to the aluminium surface via hydrolysable
Si-O-Al bonds. In addition, the silanes crosslink and form hydrophobic Si-O-Si
bonds, which are assumed to play a major role in stabilising the interface with re-
spect to hydrolysis and corrosion [3, 5, 6, 17]. Increasing the degree of crosslink-
ing improves the corrosion protection properties afforded by the coatings [5, 11].
Thus, all bis-silanes outperform the /-APS monosilane in wedge tests, even though
y-APS has the best initial adhesion.
An additional factor is the acid-base properties of silane functional groups.
Amino groups are hydrophilic and will increase the rate of water diffusion in the
interfacial region. Upon reaction with water, they provide a basic environment
underneath the adhesive coating. If the resulting pH is above 8.5, corrosion of
aluminium proceeds and the interface is weakened. This effect is seen for nearly all
samples where y-APS, "amine" and "diamine" are used.
For samples with both BTSE and y-APS, the balance between crosslinking
and acid-base properties is reflected in both ASST and wedge test results with
Araldite 2014. Of all samples pre-treated with both silanes, the sample BA3
performed worst. This sample was the only one where no curing was carried out.
The BTSE layer is, therefore, quite porous, giving a large degree of intermixing
when y-APS is applied, as indicated in the ToF-SIMS depth profiles in Fig. 3. Upon
exposure to the ASST environment, protonated amino groups provide easy access
for water and Cl~ to the aluminium surface [17]. In wedge tests, water penetrates,
providing a basic, corrosive environment. For the other samples, the BTSE layer
is probably denser and less intermixing of the layers is expected. This gives
better corrosion protection and better initial adhesion, since more amino groups
are available for bonding with the adhesive. This effect seems to be independent of
the pH of the y-APS solution during deposition.
Judging from the good performance of the BTSE samples in wedge tests, it may
not be necessary to incorporate an organofunctional group to react with the adhesive.
This is in line with the observation that 3-mercaptopropyltrimethoxysilane (MPS),
which does not react with the epoxy adhesive, performed better in wedge tests than
y-GPS used either alone or in combination with MPS [9]. However, the silane used
must provide protection against hydrolytic weakening of the interface. In addition,
the critical surface tension of the pre-treated surface must be high enough to allow
spreading of the adhesive. This probably sets an upper limit to the size of the alkoxy
group between the silane centres in bis-silanes.
3.2.5. Silanisation as replacement for pre-treatments involving hexavalent

chromium. Many studies of silane pre-treatments for corrosion protection and
paint adhesion have been conducted by van Ooij and co-workers [3-6]. In gen-
eral, they applied thicker silane layers than those used in the present investigation.
However, their results correspond well with our Araldite 2011 ASST results, in-
cluding the observation that BTSE does not adhere well to epoxy adhesive coat-
ings. They also compared silane pre-treatments to chromate pre-treatment. In many
cases, combinations of BTSE and y-APS, as well as the "amine" bis-silane alone
or in combination with a bis-polysulphur-silane outperformed chromating.
Due to the differences in failure mechanism, the results obtained for coatings
are not directly transferable to structural bonding applications. Our results show
that hydrolytic stability of the interface is of major importance in structural
bonding, whereas chemical bonding seems to have a greater importance in coating
applications. In structural bonding, thin (near-monomolecular) layers are sufficient
for increasing the initial strength, as well as the durability of adhesively bonded
joints. The performance of the best of the current silane pre-treatments approaches
that of the (FPL etch + PAA) process developed by Boeing. Thus, the results
are promising with respect to developing silane-based pre-treatments for structural
adhesive bonding of aluminium.
4. CONCLUSIONS
1. The best bond durability in wedge tests is obtained with silane pre-treatments
which provide reasonable initial adhesion, in addition to providing excellent
hydrolysis and/or corrosion protection.
2. Amino-silanes improve the chemical interaction with the epoxy adhesive, and
thereby the initial adhesion.
3. The bis-amino-silanes show better performance in wedge tests than y-APS,
probably because increased cross-linking gives better corrosion protection of the
aluminium substrate.
4. BTSE and other non-functional bis-silanes decrease the rate of corrosion at the
adhesive-aluminium interface.
5. The sample durability in the wedge test is not increased by adding a DGEBA
layer between the adhesive and amino-silane pre-treatment.
6. Of the current pre-treatments, BTSE and "hexane" are the most promising
with respect to replacing CrIV-containing pre-treatments for adhesive bonding
of aluminium.
Acknowledgements
The present work was performed as part of a Strategic Institute Program (SIP) at
SINTEF Materials Technology, where one goal is to understand the function and
failure of various surface pre-treatments for aluminium. We wish to thank The
Norwegian Research Council for financing this program. We also wish to thank
Jens-Anton Horst (SINTEF) for performing the SEM and EPMA analyses and Peter
Sjovall and Jukka Lausmaa (Swedish National Testing and Research Institute) for
help with ToF-SIMS analyses.
REFERENCES
1. J. Lohse, K. Sander and M. Wirts, Heavy Metals in Vehicles //, Report compiled for the
Directorate General Environment, Nuclear Safety and Civil Protection of the Commission of
68 B. G. TilsetQtdX.
the European Communities, July 2001. Accessible at: http://europa.eu.int/comm/environment/

waste/studies/elv/heavy_metals.pdf
2. 2002/525/EC Commission Decision of 27 June 2002 amending Annex II of Directive 200/53/EC
of the European Parliament and of the Council on end-of-life vehicles, Official J. Eur. Commun.
L170, 81-84 (2002).
3. W. J. van Ooij and T. Child, Chemtech 28, 26-35 (1998).
4. V. Subramanian and W. J. van Ooij, Surface Eng. 15, 168-172 (1999).
5. J. Song and W. J. van Ooij, /. Adhesion ScL TechnoL, in press (2004).
6. W. J. van Ooij, D. Q. Zhu, G. Prasad, S. Jayaseelan, Y. Fu and N. Teredesai, Surface Eng. 16,
386-396 (2000).
7. E. T. Vandenberg, L. Bertilsson, B. Liedberg, K. Uvdal, R. Erlandsson, H. Elwing and
I. Lundstrom, J. Colloid Interface Sci. 147, 103-118 (1991).
8. M. K. Harun, S. B. Lyon and J. Marsh, Prog. Org. Coatings 46 (1), 21-27 (2003).
9. P. R. Underhill and D. L. Duquesnay, in: Silanes and Other Coupling Agents, Vol, 2, K. L. Mittal
(Ed.), pp. 149-158, VSP, Utrecht (2000).
10. A. Franquet, J. D. Laet, T. Schram, H. Terryn, V. Subramanian, W. J. van Ooij and J. Vereecken,
Thin Solid Films 384, 37-45 (2001).
11. A. Franquet, C. Le Pen, H. Terryn and J. Vereecken, Electrochim. Acta 48, 1245-1255 (2003).
12. P. G. Sheasby and R. Pinner (Eds), The Surface Treatment and Finishing of Aluminium and its
Alloys, 6 th edn. ASM International, Materials Park, OH (2001).
13. A. Bj0rgum, F. Lapique, J. Walmsley and K. Redford, Int. J. Adhesion Adhesives 23, 401-412
(2003).
14. F. Lapique and K. Redford, Int. J. Adhesion Adhesives 22, 337-346 (2002).
15. U. Bexell and M. Oisson, Surface Interface Anal 31, 223-231 (2001).
16. M.-L. Abel, R. P. Digby, I. W. Fletcher and J. F. Watts, Surface Interface Anal 29, 115-125
(2000).
17. M. A. Petrunin, A. P. Nazarov and Y. N. Mikhailovski, /. Electrochem. Soc. 143, 251-257 (1996).
Ed. K. L. Mittal
© VSP 2004
y-APS as a durability enhancer of elastomer-to-metal

adhesive bonds
R.E HAMADE*
Department of Mechanical Engineering, American University of Beirut (AUB), P.O. Box 11-0236,
RiadEl-Solh, Beirut 11072020, Lebanon
Abstract—The beneficial effect of y-aminopropyltriethoxysilane (y-APS) coupling agent on the

durability of adhesive bonds under cathodic conditions is investigated. To improve the durability
of commercially available primer/adhesive systems commonly used in rubber-to-metal joints, the
silane is incorporated into the primer component. Mass uptake tests on thin free-standing films in
highly alkaline NaOH solutions show that the primer re-formulation resulted in enhanced hydrolysis
resistance of the primer itself. The mass uptake following exposure to alkali is found to be inversely
proportional to silane concentration. Similarly, chemical degradation rate of the primer film is found
to decrease with increasing y-APS concentration. Also, joints (Strip Blister Specimens, SBS) utilizing
mild steel substrates bonded with the J/-APS silane-modified primer were found to be more resistant
to delamination as compared with the control primer/adhesive system under the same test conditions.
The concentration of the silane is correlated with the reduction of delamination rates in both 1 M
NaOH and artificial seawater.
Keywords: Silane; y-APS; elastomer; mild steel; metal; adhesive bond; primer; durability; cathodic
debonding; free film; delamination.
1. INTRODUCTION
The adhesion literature abounds [1, 2] with research that demonstrates the important
role of silanes in adhesion. Evidence of chemical interfacial bonding and adsorption
of y-APS on metallic surfaces was investigated early by Plueddemann [3] where
the ethoxyl groups in aminosilane were found to hydrolyze, giving rise to silanol
groups which, in turn, covalently bond to the hydroxyl groups available on the
metal surface. Boerio and co-workers [4-6] used XPS to study the adsorption
of aminosilane (y-APS) onto metal substrates (including iron) and concluded that
the silane's amino groups were protonated via interaction with hydroxyl groups
on the metal surfaces. Also, evidence of covalent bonding between y-APS and
*Tel: (961-1) 350-000, ext. 3481; Fax: (961-1) 744-462; e-mail: rhamade@aub.edu.lb
70 R. F. Hamade
steel surfaces as investigated by time-of-flight SIMS (TOFSIMS) and XPS was

presented by van Ooij and Sabata [7]. More recently, Kranias et al. [8] used XPS to
measure the binding energies of chemisorbed y-APS on metallic surfaces. Kallury
et al. [9] demonstrated that not only a covalently-bonded interface was created, but
also that an ordered structure existed in the interphase as it moved from the adsorbed
monolayer region into the bulk region.
In addition to classical applications, such as bonding glass to a multitude of other
materials, the adhesion-promoting capabilities of silanes were demonstrated; for
example, in improving the adhesion of epoxy resin to alloy 42 leadframes [10] and
in bonding difficult-to-bond materials such as polyethylene to metal substrates [11]
and polyimide to alumina [12]. When the silane is properly applied [13] to metal
surfaces, stable and durable bonds are created. The creation of these bonds is
generally acknowledged to be responsible for improving the wet durability of
bonded joints of organic polymers to metals such as in the reported improved wet
durability of epoxy paints [14] and epoxy adhesives [15, 16] to several metals
(including steel). y-APS, in particular, was cited to have a pronounced effect on
the adhesion of epoxy to copper [17], Pyralin to silicon [18] and poly (ethylene
naphthalate) film to several materials [19], as well as on the durability of epoxy to
aluminum bonded joints [20]. Such findings and numerous others demonstrate why
silanes in general, and APS in particular, have proven to be popular in polymer-to-
metal bonding applications.
This paper reports on the use of y-APS to improve the resistance to cathodic
degradation of elastomer-to-metal (mild steel) adhesive bonds. For the zinc-steel
galvanic pair in seawater, it is widely accepted that one of two electrochemical
reduction reactions can occur at exposed locations. At voltages less negative than
— 1020 mV (standard calomel electrode, SCE) the oxygen reduction reaction
l / 2 0 2 + H 2 0 + 2e" -> 2 0 I T (1)
is favored, while the water reduction reaction
2H 2 0 + 2e~ —• H2(g) + 20H" (2)
is dominant at more negative potentials. Both of these reactions increase the

OH~ ion concentration (and, thus, the local pH). In addition to the hydroxyl ions,
reaction (2) also produces bubbles as a result of hydrogen gas evolution. The
presence of hydroxyl ions at the elastomer/metal bondline is detrimental to the
durability of cathodically exposed polymeric coatings as demonstrated by several
pioneering researchers in the area of corrosion [21-24]. Similar undesirable
consequences take place in the case of adhesively bonded elastomer-to-metal joints
as well [25, 26].
The objective of this work was to evaluate the beneficial effects of blending
y-APS with the primer as measured by the primer's resistance to degradation under
harsh environments. This involved two types of studies:
y -APS as a durability enhancer of elastomer-to-metal adhesive bonds 71
1. Net mass gain studies of primer free-standing films (made from both the 'control'
as well as the silane-modified primer) under highly alkaline conditions (1 M
NaOH).
2. Delamination studies of adhesively bonded joints either with the 'control' primer
or with the silane-enhanced primer. These studies utilized a compact, beam-type
adhesively bonded joint called the Strip Blister Specimen (SBS, see Fig. 1),
which allowed for the simultaneous application of peel stress and cathodic
conditioning.
2. EXPERIMENTAL
2.1. Specimen material and fabrication
In this study, the control primer/adhesive system, referred to as A, was a proprietary,
commercially available, two-coat (a primer bottom-coat and an adhesive top-coat)
vulcanizing primer/adhesive system used in bonding elastomers to steel. Boerio
and co-workers [27] found that the adhesive top-coat was based on allylically-
brominated poly-2,3-dichloro-l,4-butadiene. Regarding the primer bottom-coat,
they found that it consisted of phenol formaldehyde and chlorinated isoprene, and
included titanium dioxide (titania) and zinc compounds as additives and carbon
black as filler. For durability improvement, the formulation of this control primer
coat was modified by blending in 6% by weight of y-APS (A-1100 from Union
Carbide, now OSi). The resulting primer/adhesive system is referred to as the
modified system B. This approach of re-formulating the primer component by
blending in the silane was pioneered by Plueddemann [28]. Walker [29] investigated
several approaches to use silanes as adhesion promoters including blending with the
polymer and reported impressive improvements in terms of mechanical properties
as well as bond durability. While mechanical properties were reported to peak for
blended polymer with 5 wt% APS, no more than 0.2 wt% APS was needed to attain
maximum bond strength on mild steel.
The strip blister specimen (SBS) was utilized to study bond delamination. Both
a schematic and a photograph of the SBS are shown in Fig. 1. This configuration
was first introduced by Liechti and co-workers [30]. In fabricating these specimens,
the substrates were made of ANSI 1026 mild steel, while the rubber was specially
formulated for marine applications and is known as 5109S Neoprene. Prior to
bonding, sheets of metal were vapor degreased using trichloroethane before and
after grit blasting with 40-grade steel grit at 275 kPa. The primer was then brush-
coated on the metal sheets and, upon drying, was followed by application of the
adhesive coat also by brushing. Non-vulcanized (green) neoprene rubber strips were
then placed in a mold in contact with the primer/adhesive-coated steel substrates.
The mold was then placed in a heated platen press at a temperature of 157°C and
under a pressure of 3.4 MPa for 50 min. This was followed by a post-cure cycle
at 177°C for 3 h. This resulted in fully vulcanizing the elastomer as well as in
curing the primer/adhesive bondline.
72 R. F. Hamade
L_.j2.7~J
-RUBBER DOWEL*
UNIT'mm
Figure 1. Schematic diagram and a photograph (side view) of the self-loading strip blister
specimen (SBS). The self-loading capability is accomplished by inserting a plastic dowel of a known
diameter, d (total of two dowels per specimen are used one on either side of the specimen), in the
built-in flaw at the center of the specimen. The length of the built-in flaw is 2a where a = 12.7 mm
is the flaw length in either direction as measured from the center of the specimen. Given the
geometric symmetry of the SBS and considering only one half of the specimen, one may presume
that as delamination proceeds the crack grows from the initial length of a = 12.7 mm to length of
a = 12.7 mm + the delaminated distance. On the left-hand side of the photograph, notice the plastic
screw used to hold an electric wire attached to the power supply in order to polarize the SBS.
2.2. Specimen configuration and characteristics

22A. Free-standing film specimen (FFS). In addition to the 'control' silane-
free primer free-standing films, films containing four concentrations of silane were
prepared: 0.5, 2, 5 and 8 wt%. The solvent-based primer material (both with and
without y-APS) was cast into neat free-standing films with an average thickness
of 1.6 mm. Due to the solvent content of the primer, a technique was developed
by which the casting and subsequent drying was done on Teflon-coated cloth made
from fiberglass. Due to the flexibility of the cloth, the film was able to shrink while
drying without developing cracks thus resulting in defect-free films. Upon drying
at room temperature, the FFS samples were cured at a temperature of 157°C and
under a pressure of 275 kPa.
y-APS as a durability enhancer of elastomer-to-metal adhesive bonds 73
Table 1.
Four different combinations of strip blister specimen configurations with increasing tR (rubber
thickness) and d (dowel diameter), resulting in progressively increasing values of initial strain energy
release rates, GJO (which corresponds to the initial built-in flaw length of 12.7 mm)
'R d GTO c\ C2 C3
(1(T 3 m) (1(T 3 m) (kJ/m2)
3.18 9.52 1.04 0.0042 -0.227 3.175
8.13 9.52 2.24 0.0055 -0.341 5.603
8.13 12.7 3.8 0.0103 -0.614 9.793
12.7 12.7 4.6 0.0091 -0.538 10.541
Also tabulated are values for the coefficients c\, C2 and c^ of the second-order polynomial in
equation (3) that relate the total strain energy release rate, G j , at the delaminated tip vs. crack
length, a, which is equal to the length of the initial built-in flaw (12.7 mm) -f length of delamination.
2.2.2. Strip blister specimen (SBS). The strip blister specimen was utilized to
collect delamination data. The SBS was designed to have a self-loading capability,
which was accomplished by inserting a plastic dowel of a known diameter, d, in the
built-in flaw at the center of the specimen. The length of the built-in flaw was 2a
where a = 12.7 mm is the flaw length in either direction as measured from the
center of the specimen. Given the geometric symmetry of the SBS and considering
only one half of the specimen, one may presume that as delamination proceeds the
crack grows from an initial length of a = 12.7 mm to length of a = 12.7 mm + the
delaminated distance. The driving force for delamination in the SBS was described
by Liechti et al. [31] in terms of a fracture energy parameter, the total strain energy
release rate, Gj. The initial value of GT corresponding to the initial built-in flaw
of 12.7 mm, GTO, was determined to be a function of rubber thickness (tR) and
the imposed deflection (d) as shown in Table 1. As delamination proceeds and
the crack length grows, the driving force for delamination, GT, decays from a high
of GTO according to the following second-order polynomial equation
GT = c\a2 + câ + c3. (3)
This variable-GT vs. a behavior is graphically illustrated in Fig. 2 where actual
analysis results are shown as symbols and the fits from the polynomial equation
(equation (3)) are shown as solid lines.
2.3. Environmental conditioning

In order to perform the net mass gain studies, FFS samples (with and without APS)
were conditioned in high pH solution (1 M NaOH) and relatively high temperature
of 42°C FFS samples were occasionally pulled out of the conditioning bath, blotted
to remove any excess water that might had simply clung to the surface but not truly
absorbed.
To measure in situ bond delamination, the SBS samples were conditioned in
high-density polyethylene water tanks and the temperature was regulated with
74 R. E Hamade
6H
tR d
(mm) (mm)
D 3.18 9.53
X 8.13 9.53
A 8.13 12.70
12.70 12.70
o - Second - Order Equation
10 12.7 20 40
Debond length (mm)
Figure 2. A second-order polynomial fit (equation (3)) to the Gj vs. delaminated length for SBS
specimens. Actual analysis results are shown as symbols and the fits from the polynomial equation
are shown as solid lines. Note that the initial built-in debond length is 12.7 mm, as shown in the plot.
Also note that the total strain energy release rate is functions of both rubber thickness, tR, and dowel
diameter, d [30,31].
submergible heaters. All the delamination data reported in this work correspond
to SBS samples loaded initially to a GT0 value of 2.24 kJ/m2 using rubber thickness
of 8.13 mm and dowel diameter of 9.52 mm. The specimens were placed in
the electrolyte (artificial seawater, ASW or 1 M NaOH solution), becoming the
working electrode of an electrochemical cell, with a graphite counter electrode
and a reference standard calomel electrode (SCE) also immersed in the solution.
An HP model 6214B power supply was used to provide the desired voltage, which
was monitored and adjusted if needed. To provide the needed oxygen, room
air was pumped into the bath after passing through a C0 2 trap. The cathodic
electrochemical setup was such that a number of specimens were assembled on
a common pin and exposed to the same voltage.
In order to assess the adhesive bond susceptibility to delamination under the com-
bined action of peel stress and 'galvanic environment', a suitable test matrix was
needed. Table 2 contains five target current densities (and their corresponding ca-
thodic voltages) and four temperature settings in two electrolytes: 1 M NaOH and
artificial seawater (ASW). Going vertically along any one column in the table yields
the same current density regardless of the environment's electrolyte and tempera-
ture. Table 2 is based on electrochemical potentiometery data partially reported in
Table 2.
Target cathodic potentials for the test matrix for the SBS specimens bonded with both control A and
y-APS-modified B adhesive systems
Solution Temp. # Current density (x 10- 4 mA/m 2 ) (level)

<°Q oW~ 0.59 (1) 1(2) 1.475(3) 14.75 (4) 73.3 (5)
ASW 25 A A-l A-2* A-3* A-4* A-5*
-830 -1000 -1100 -1420 -1825
30 B B-1 B-2* B-4
-800 -990 -1380
40 C C-l C-2 C-3 C-4* C-5
-770 -970 -1020 -1310 -1620
55 D D-l D-2 D-3 D-4* D-5
-770 -930 -1000 -1250 -1520
1 M NaOH 25 E E-0* E-l* E-2* E-3* E-4* E-5*
0 -960 -1150 -1250 -1420 -1550
30 F F-0* F-1*
0 -900
40 G G-l G-2 G-3 G-4* G-5
-920 -1000 -1200 -1360 -1500
55 H H-l H-2 H-3 H-4* H-5
-890 -980 -1060 -1310 -1470
The environments include two solutions: ASW and 1 M NaOH, five current density levels
(denoted levels 1-5) ranging from low of 0.59 x 10~4 to high of 73.3 x 10 - 4 mA/m2, and four
temperatures: 25, 30,40 and 55°C (some specimens were tested in 1 M NaOH without being polarized
and the current density level is referred to as level 0). The cells with (*) represent those conditions
for which delamination tests were conducted. For instance, the applied voltage for case A-4 is
— 1420 mV (SCE). This case represents conditioning in ASW, 25°C, and under cathodic current
density level 4 (14.75 mA/m2).
Ref. [32]. In ASW, case A-2 is the 'reference' case, since the natural potential of
zinc-steel galvanic pair is about -1000 mV (SCE). At this voltage in ASW and
at 25°C, the corresponding cathodic current density is 1 mA/m2 (or level 2).
Delamination testing of large number of SBS samples was performed under
a wide variety of environmental conditions as shown in the table. After being
conditioned for a desired duration, the bonded sample was pulled out and the
four delaminated fronts B, C, D and E (Fig. 1) measured using calipers in a
nondestructive manner after which the specimen was returned to the bath and the
test continued.
3. DEGRADATION BEHAVIOR
3,1. Bulk primer degradation
Under cathodic conditions, bond degradation is commonly associated with OH~
accumulation that results in high pH value at the interface. In order to determine the
resistance of the primer to the alkali, FFS samples (with and without y-APS) were
76 R. F. Hamade
60
- 0 - - - 0 wt%Silane
I & — 0.5 wt% Silane
5 0 4J—-Q
A 2.0 wt% Silane
5.0 wt% Silane >--^" -o—o-,- o—o
•% 8.0 wt% Silane ,o^'
£ 40
as<*>
O
t 0
Q~—&—Q
30
/> ..-••-a
.-•ET"
o
/> P--
J2--
.--A
20 H .A-
-A-
/> -A"
— - • & '
10H , 0 'ET
O A
^ ' ^ '
r
50 1 0 0 1 5 0 25o 250
Time (hours)
Figure 3. Net mass gain for the control free-standing primer film specimens (0 wt% }/-APS), as well
as the y-APS-modified primer (0.5, 2, 5 and 8 wt% y-APS). Environment: 1 M NaOH, 42°C.
conditioned in 1 M NaOH. Since the free-standing films are relatively thin, chemical
reaction is expected to dominate the degradation process with mass transfer kinetics
playing only a secondary role.
The net mass gain for FFS samples upon conditioning in 1 M NaOH at 42° C is
shown in Fig. 3. The FFS samples used were made from the control primer and from
y- APS -modified primer at 0.5, 2, 5 and 8 wt% y-APS concentration. The net mass
gain measured is the net balance of the mass gain (i.e., absorbed water, ions, and
retained byproducts of the degradation reaction) and the mass loss due to leaching
of degraded primer constituents. The following observations are made from Fig. 3:
1. The net mass gain at saturation is inversely proportional to the y-APS concen-
tration in the primer, i.e., the higher the silane wt% the lower is the saturation
level. The net mass gain for the primer films with 5-8 wt% APS is less than 5%
which is one order of magnitude lower than that of the control silane-free primer
(over 50% mass gain).
2. The same observation as above applies to the rate of net mass gain. The
near-linear nature of the mass gain plots vs. time suggests a degradation
process dominated by chemical reaction. For the control case and for each
of the four silane concentrations, a linear chemical reaction rate (mm/day)
was calculated. These degradation rates are plotted in Fig. 4 that shows the
calculated rate for each concentration (plotted on a logarithmic scale) vs. y-APS
concentration (wt%). Figure 4 shows the stability of the primer in 1 M NaOH to
be optimum at high concentration levels (5-8 wt%) at y-APS. Compared with
the degradation rate of the control primer of 0.2 mm/day, the rate at 5 and 8 wt%
y-APS reduces to about 0.04 mm/day (or only 20% of that of the control).
c
o
D
s i 0.1 "D--
D
- - ^m
0.01
1 1
Y- APS concentration (wt%)
Figure 4. Degradation rate of free-standing primer film specimens (plotted on a logarithmic scale)
vs. y-APS concentration. Environment: 1 M NaOH, 42°C. A line fit is shown through the
experimentally collected data points.
32. In situ bond delamination

Preliminary delamination testing was done on SBS samples bonded using primer
at the same four y-APS concentrations as used in the free-standing film study (as
well as the control primer). These specimens were conditioned in 1 M NaOH at
— 1200 mV (SCE) and 30°C. In all the samples tested, a dowel diameter, d, of
9.51 mm and rubber thickness, fR, of 8.13 mm were used, resulting in an initial
value of GJO of 2.2 kJ/m2. When plotted against time, delamination distances yield
plots that are fairly linear. The initial delamination rates DRQ (mm delamination per
day) for the SBS samples tested are plotted in Fig. 5. Similar to the results obtained
in Fig. 4 for the free-standing primer film samples. Figure 5 shows the stability of
the bonded joints under peel stress and in 1 M NaOH to be optimum at the high
concentration of 5-8 wt% )/-APS. An order of magnitude reduction in the initial
delamination rate was accomplished under these harsh conditions. Compare, for
example, the delamination rate of the control (1 mm/day) to that of 8 wt% y-APS
in the primer (0.1 mm/day). These y-APS concentrations necessary to make such an
impact are quite high as compared to what is reported in the literature. For example,
Walker [29] reports the need for only a fraction of a percent for optimum adhesion
of polymers to metals. Such low concentrations of silanes are also reported by Bell
et al. [16] in evaluating the enhancement of lap shear strength of acrylic and epoxy
to steel joints. However, since adsorption and film building on surfaces of metals
and metal oxides including that of titania is well documented (see, e.g., Ref. [33])
and since fillers make up a considerable percentage of the total composition of the
primer, chemisorption on the hydroxy 1-rich filler surfaces in the primer appears to
be responsible for this high wt% requirement.
78 R. R Hamade
1&
\ •
(0
v.
\
\
CC
Q
D
0.1
i 3 r
Figure 5. Initial delamination rate, DRo (plotted on a logarithmic scale) of the strip blister specimens
vs. /-APS concentration in the primer; Environment: 1 M NaOH, -1200 mV (SCE), 30°C. A line fit
is shown through the experimentally collected data points.
Based on the above, 6 wt% was chosen as the target y-APS concentration in
the enhanced-primer formulation. SBS samples were then fabricated using the two
extremes: the control containing no silane and the modified primer using y-APS
concentration of 6 wt%. Next, the samples were conditioned as per Table 2.
Delamination results shown in Figs 6-11 represent the arithmetic averages of
the delamination measurements made along the four edges B, C, D and E (see
Fig. 1).
3.2.1. Effect of temperature.

3.2.1.1. Control A primer/adhesive system. Figure 6 shows that temperature
plays a major role in accelerating bond delamination. The delamination histories
presented here correspond to four different SBS samples made with the control
primer/adhesive system 4A and conditioned in ASW (held at 25 and 40°C in
separate tanks) and maintained at current density level 4 of 14.75 x 10~4 mA/m2
(cases A-4 and C-4 in Table 2). Two other cases were run in 1 M NaOH solution
at 25 and 40°C, and at current density level 4 of 14.75 x 10~4 mA/m2 (cases E-4
and G-4). Although initially linear, delamination rates appear to slowly decrease.
This is likely because as the delaminated crack grows it causes the instantaneous GT
to decrease from an initial high GT0 = 2.24 kJ/m2. From the plot, it is obvious
that delamination proceeds at higher rates in 1 M NaOH than in ASW for the
temperatures tested. The initial delamination rates, DR0, were evaluated at the
onset of the test at GTo = 2.24 kJ/m2. For example, the initial delamination rates for
cases A-4 and C-4 are 0.66 and 5.6 mm/day, respectively. Although some tests were
conducted at 55°C, the results were simply excluded from further consideration
34. -]
28-
D
E D
E, 24- +
0)
a—
o
c 20- \ ^\ +
(0
%
'•5 16 -J a
c +
+
$ • >
o 40°C , 1M NaOH
"£
C
E
12
8-
/ +
V 25°C,1MNaOH
/I /^ >^P \
0)
Q
4 Jf/ jZ
r •'
It-TV- 40°C , ASW
25°C , ASW
u ~1
ib A 3'0
Time (days)
Figure 6. Temperature effect on bond delamination rate (DR); SBS data using control A adhesive
system at current density level 4 (14.75 x 1(T4 mA/m2) in ASW (cases A-4 and C-4) and 1 M NaOH
(cases E-4 and G-4).
because of buildup of calcareous deposits on the samples thus masking the true
accelerating contribution of temperature.
3.2.1.2. Silane-modified B primer/adhesive system. Figure 7 shows that temper-
ature, although to a lesser extent, also plays a strong accelerating effect on the de-
lamination rate of specimens bonded using the y-APS-modified primer/adhesive
system B. In ASW, delamination results for two cases at 25°C and 55°C (A-4
and D-4) are shown, while for 1 M NaOH, three cases (E-4, G-4, and H-4) at three
levels of temperature, 25, 40 and 55°C, are shown. Again and because of concerns
with calcareous deposits, it was decided to exclude the delamination data corre-
sponding to 55°C (313 K), leaving only data corresponding to two temperatures,
25 and 40°C. Examining the results reveals that in 1 M NaOH (bulk pH = 14) the
initial delamination rates are significantly higher than the rates in ASW. Contrasting
with the results for the control primer/adhesive system A in Fig. 6, it appears that the
resistance of the silane-modified system B to the combined degrading effect of peel
stress and environment is quite impressive (especially under extremely harsh envi-
ronments). For example, after 10 days in 1 M NaOH at current density level 4, the
bonds delaminated only about 4 and 10 mm at 25 and 40°C, respectively, which is
less than one third of the delamination length encountered by the control system 'A
under identical conditions as shown in Fig. 6.
3.2.2, Effect of voltage.

3.2.2.1. Control A primer/adhesive system. The accelerating effect of voltage
on the delamination rate of the control primer/adhesive A system is illustrated
80 /?. F. Hamade
20
X
x O
/x
? 16-
A &
E / A\y
o
o
§ 12- V* >X A
» / x /^ + \
c X
x i^/\A O / ^ g — " - 55°C, 1M NaOH
o /A/ iM
00
aminati
^ / A \
^r/ + *_.±- s<£\ 25°C, IMNaOH
40°C,1MNaOH
<D 4-
55°C, ASW
25°C, ASW
—
u
(
~ ' Vo 2'o 3'0 4b '
Time (days)
Figure 7. Temperature effect on the bond delamination rate (DR); SBS data using modified B
adhesive system at current density level 4 (14.75 x 10~4 mA/m2) in ASW (cases A-4 and D-4) and
1 M NaOH (cases E-4, G-4 and H-4).
160
Time (days)
Figure 8. Voltage effect on the bond delamination rate (DR); SBS data using control A adhesive
system in ASW at 25°C. Cases: A-2, A-3, A-4 and A-5.
in Fig. 8. The test was run for extended periods of time on four SBS samples
bonded with the control system and tested in different ASW tanks set at 25°C and
subjected to increasing current density levels 2-5 (i.e., cases A-2, A-3, A-4 and A-5,
corresponding to -1000, -1100, -1420 and -1825 mV (SCE), respectively). The
corresponding values for the initial delamination rate were found to be 0.03, 0.21,
Figure 9. Voltage effect on the bond delamination rate (DR); SBS data using control A adhesive
system in 1 M NaOH at 25°C. Cases: E-l, E-2, E-3, E-4 and E-5.
0.66 and 4.0 mm/day, respectively. When compared to the reference case A-2, these
DR values result in acceleration factors of 1, 6.6, 21 and 128 times, respectively.
Figure 9 shows the influence of voltage on the delamination behavior of the
control primer/adhesive system A in alkaline 1 M NaOH environment (cases: E-l
through E-5). The initial delamination rates were estimated for the cases E-l, E-2,
E-3, E-4 and E-5 at 0.33, 1.6, 1.85, 1.88 and 2.94 mm/day, respectively. With
the exception of current density level 5, these initial DRs are considerably larger
than those in ASW. When compared to the reference case A-2, acceleration factors,
respectively, of 0.2, 1, 1.15, 1.17 and 1.8 times are estimated for these cases.
32.22. Silane-modified B primer/adhesive system. Figure 10 shows the effect
of the voltage on the initial bond delamination rate of the silane-modified B
primer/adhesive system in ASW at 25°C (cases A-2, A-3, A-4 and A-5). The
durability of the silane-modified B primer/adhesive system is reflected by the low
values of the initial DRs for these four cases of 0.03, 0.13, 0.14 and 0.13 mm/day,
respectively. Under the same conditions, the samples bonded with the control
system A delaminate 1.14, 1.6, 4.7 and 30.7 times faster than their silane-modified
siblings. Interestingly, the DRs for the silane-enhanced system for cases A-3
through A-5 are identical, testifying to the resistance to delamination of the silane.
Delamination data for the modified B system in 1 M NaOH environment at 25°C are
presented in Fig. 11 (cases E-l through E-5). Compared with the delamination rates
of system A in polarized 1 M NaOH (shown in Fig. 9), delamination acceleration
of the B system appears more modest. For example, after 10 days of exposure,
delamination distances of only 0.9, 2, 2.3, 3.8 and 4.8 mm were recorded after
conditioning at -900, -1150, -1250, -1420 and -1550 mV (SCE), respectively.
Comparing these results to those in Fig. 9 reveals that the resistance to delamination
82 R. F. Hamade
160
Time (days)
Figure 10. Voltage effect on the bond delamination rate (DR); SBS data using silane-modified B
adhesive system in ASW at 25°C. Cases: A-2, A-3, A-4 and A-5.
1550 mV
1420 mV
o
o
c
CO
CO
T5
c
o
'•J3
CO
c
£
J2
o
a
40
Time (days)
Figure 11. Voltage effect on the bond delamination rate (DR); SBS data using modified B adhesive
system in 1 M NaOH at 25°C. Cases: E-l, E-2, E-3, E-4 and E-5.
of the B system in polarized 1 M NaOH is 1.3-4.8 times better than that of the
control A system. This is in line with the resistance of the silane to delamination
in ASW as shown above. The initial DRs corresponding to cases E-l through E-5
are 0.25, 0.33, 0.43, 0.42 and 0.63 mm/day, respectively. This again reflects the
delamination resistance of the silane under harsh conditions.
3.2.3. Summary of the beneficial effect of silane-modified primers in 1 M NaOH

and in ASW.
3.2.3.1. Quantifying the beneficial effect ofsilane modified primer in 1 M NaOH.
Data for the free-standing film degradation rate in 1 M NaOH at 42°C (Fig. 4) and
initial delamination rates in 1 M NaOH, -1200 mV, 30°C (Fig. 5) are re-plotted
in one plot (Fig. 12). Although the net mass gain of the primer film may appear
at the first glance to follow a different degradation mechanism than that of SBS
delamination, it is found that in both cases degradation appears to be governed by
chemical reaction. This is especially true at the onset of SBS delamination where
stress at the delamination front is high enough to constantly cause debonding of
the degraded (weakened) interface and thus constantly exposing fresh material for
degradation. In both cases, mass transfer (diffusion) plays a secondary role. This
justifies in our opinion plotting both sets of data in the same figure. The bottom two
lines in the figure show that the degradation rate becomes smaller as y-APS wt%
increases. The lines also show that this improvement is of the same magnitude for
both free-standing film and bonded joints. This is further substantiated where the
initial DRs for three more representative cases in 1 M NaOH (E-4, E-5 and G-4)
are co-plotted in the same figure. Although the delamination data available are
only for the two extreme cases (0 wt% and 6 wt% y-APS), the slopes for the
delamination rate vs. y-APS concentration appear to equal those of the other cases.
Furthermore, and for comparison's sake, a lone data point corresponding to SBS
delamination under no voltage (E-0) is also added to this plot. When compared
Figure 12. Data for degradation rates (plotted on a logarithmic scale) vs, y-APS concentration in the
primer: free film degradation rate in 1 M NaOH, 42°C and initial delamination rate (DRo) of SBS
samples in 1 M NaOH at 30°C. Also plotted are initial delamination rates (DRQ) of SBS samples for
four cases in 1 M NaOH: E-0, E-4, E-5 and G-4.
84 R. F. Hamade
10
-Â A2(-1000mV,25°C)
- X- - A3 (-1100 mV, 25°C)
•-y- — A4(-1420mV,25°C)
...g..- A5(-1825mV,25°C)
- ^ - - C4(-1310mV,40°C)
1J
E •-o
E
o ^L. __ "—"••—-—.. ^^
DC
Q — -T$
0-1 J
0.01
1 ' 1 '
Figure 13. Plot of initial delamination rate, DRQ (plotted on a logarithmic scale) vs. y-APS
concentration in the primer. Five cases of SBS samples tested in ASW: cases A-2 through A-5, as
well as case C-4.
against the degradation rate for the free-standing film, the delamination rate for the
stress-assisted delamination is lower but apparently only because of the difference
in bath temperature (42 vs. 25 °C). Examining the figure one may conclude that the
6-8 wt% y-APS blended in with the primer would result in roughly an order of
magnitude delamination rate reduction for practically all cases tested in 1 M NaOH.
3.2.3.2. Quantifying the beneficial effect of the silane-modified primer in ASW.
Figure 13 presents the initial delamination rates (again plotted on log scale) in ASW
for cases A-2 through A-5 as well as case C-4. The data are available only for the
two extreme cases: control at 0 wt% y-APS concentration and that at 6 wt%. When
connected, the lines are not as parallel as in the case of Fig. 12 showing results
of conditioning in 1 M NaOH solution. Nevertheless, the benefit of silane is quite
obvious in reducing delamination rates across all levels of temperature and voltage
in ASW with the highest improvement being at the more extreme conditions.
4. DISCUSSION
Significant improvement in the resistance of the bonded specimens to delamination
is accomplished through modifying the primer component of the two-component
primer/adhesive system A with y-APS. Figure 14 is a plot of the initial DRs
(plotted on a log scale) vs. the inverse of temperature for cases A-4 and C-4
(in ASW) and E-4 and G-4 (in 1 M NaOH). Having excluded the data corresponding
to 55°C because of calcareous deposits, the figure presents data corresponding
to the two remaining temperatures, 25 and 40°C. Nevertheless, the plot shows
10q * ^
O-
8"
Q—.
^
fc
Q *-Q
0.1 d
. - B — A in ASW
- - ^ _ _ - A in INNaOH
— g - . . - Bin ASW
- - $ - - - Bin INNaOH
0.01
0.00315 0.0632 0.0CJ325 0.0b33 0.0CJ335 0.0034
i/r(K 1 )
Figure 14. Plot of the initial deiamination rate, DRo (plotted on a logarithmic scale) vs. the inverse of
temperature for both the control primer/adhesive system A and the y-APS-modified primer/adhesive
system B: cases A-4 and C-4 (in ASW) and E-4 and G-4 (in 1 M NaOH).
interesting trends. Under harsh environments, the silane-modified system B

outperforms the control system A where the improvement in the rate approaches an
order of magnitude. This indicates that significant improvement in the resistance
to deiamination is accomplished by blending y-APS with the primer of the
two-component control primer/adhesive system A. Notice that for both adhesive
systems, deiamination proceeds faster in NaOH than in ASW. Although the
available data are limited only to two temperatures, the slopes of the deiamination
rate vs. the inverse of temperature lines appear to be equal. Since the slope
is proportional to the activation energy of the process responsible for the initial
deiamination, it may be argued that the same process (presumably chemical
reaction) controls bond deiamination. Because the slopes appear equal for both
adhesive systems in ASW and in 1 M NaOH, the same chemical reaction appears to
control deiamination (for the cases plotted in Fig. 14).
The effect of cathodic voltage on the deiamination rates for both the control and
the silane-modified system is shown in Fig. 15. The figure illustrates the relation
between the initial deiamination rates (plotted on a log scale) and cathodic potential
(going horizontally in Table 2). In ASW, testing covers cases A-2 through A-5,
while in 1 M NaOH testing cases E-l through E-5 were conducted. Note that the
silane-modified system B significantly outperforms system A having consistently
lower DRo values than those of the control in both environments. Most sensitive to
voltage (as indicated by the slope of the trendlines) is the control A in ASW with
the y -APS silane-modified system B is only slightly sensitive to the applied voltage
in both ASW and 1 M NaOH.
86 R. F. Hamade
iu:
~ •""'"
- x*"
- .' •
- „
"
"*"• S
• •
^+.+
• .x'
—
,
>> 1- ,x* |
co : * +~ s'
5£ ---** *"
..-o-—
o
t o-°;v-^"
tr
a us' n
0.1: . ^ ' _ _ . . — • - - — - *
*£.*'• ""
i — * — AinASW
---+--- A in 1N NaOH
—Q—- BinASW
,.^9>... BinlNNaOH
0 01 -
U.U 1
8( 10 1000 12^0 1400 1600 1800 20
Voltage (-mV)
Figure 15. Plot of the initial delamination rate, DRQ (plotted on a logarithmic scale) vs. applied ca-
thodic voltage for both the control primer/adhesive system A and the y-APS-modified primer/adhesive
system B. Environment: both ASW and 1 M NaOH (all data at 25°C).
5. CONCLUSIONS
1. As measured by the net mass uptake at saturation levels and by the degradation
rates of the free-standing primer film samples, the resistance of the y-APS
modified bulk primer to the alkali is enhanced by about an order of magnitude
as compared to the control primer.
2. As measured on actual bonded joints, the y-APS modified primer/adhesive
system B outperforms the control A system practically under all combinations
of peel stress, cathodic voltage, and temperature in both ASW and in 1M NaOH
solutions. This is evidenced by up to an order of magnitude reduction in the
initial bond delamination rates in highly alkaline 1 M NaOH.
Acknowledgements
The author wishes to thank the Office of Naval Research, the Naval Research
Laboratory-Underwater Sound Reference Detachment, the Texas Research Institute
and the Virginia Center for Innovative Technology for their support of this research.
Dr. J. S. Thornton, the president of the Texas Research Institute, and his staff are
acknowledged especially for their help in fabricating the strip blister specimens.
Thanks are also due to Drs. Wim van Ooij and F. J. Boerio at the University of
Cincinnati for their help and advice. The mass uptake work done by Zhiqiang
Wang and Yeou Chang is greatly appreciated. The author is especially grateful
to the Engineering Science & Mechanics Department at Virginia Tech for providing
facilities and to the Center for Adhesive & Sealant Science (especially the Center's
director Dr. David A. Dillard) for fostering the interdisciplinary environment for
this research.
REFERENCES
3. E. P. Plueddemann, Silane Coupling Agents. Plenum Press, New York (1982).
4. D. J. Ondrus, F. J. Boerio and K. J. Grannen, J. Adhesion 29, 27 (1989).
5. D. J. Ondrus and F. J. Boerio, J. Colloid. Interface Sci. 124, 349 (1988).
6. M. R. Horner, F. J. Boerio and H. M. Clearfield, in: Silanes and Other Coupling Agents,
K. L. Mittal (Ed.), p. 241. VSP, Utrecht (1992).
7. W. J. van Ooij and A. Sabata, in: Silanes and Other Coupling Agents, K. L. Mittal (Ed.), p. 323.
VSP, Utrecht (1992).
8. S. Kranias, C. Bureau, D. P. Chong, V. Brenner, I. George, R Viel and G. Lecayon, J. Phys.
Chem. B 101, 10254-10261 (1997).
9. K. M. Kaliury, U. J. Krull and M. Thompson, Anal. Chem. 60, 169 (1988).
10. S. M. Song, C. E. Park, H. K. Yun, S. Y Oh and J. M. Park, J. Adhesion Sci. Technol. 11,797-809
(1997).
11. H. Fujimatsu, K. Iyo, H. Usami, S. Ogasawara and K. Kajiwara, Composite Interfaces 9, 259-
272 (2002).
12. M. Anschel and P. D. Murphy, J. Adhesion Sci. Technol. 8, 787-806 (1994).
13. R. A. Gledhill, S. J. Shaw and A. Tod, Int. J. Adhesion Adhesives 10, 192-198 (1990).
14. P. Walker, /. Coatings Technol. 52, 49-61 (November 1980).
15. F. J. Boerio, F. J. Armogan and S. Y Cheng, J. Colloid Interface Sci. 73, 416 (1980).
16. J. P. Bell, R. G. Schmidt, A. Malofsky and D. Mancini, in: Silanes and Other Coupling Agents,
K. L. Mittal (Ed.), p. 49. VSP, Utrecht (1992).
17. X. H. Gu, G. Xue and B. C. Jiang, Appl. Surf. Sci. 115, 66-73 (1997).
18. I. George, P. Viel, C Bureau, J. Suski and G. Lecayon, Surf. Interface Anal. 24, 774-780 (1996).
19. B. Hu, R. M. Ottenbrite and J. A. Siddiqui, Polym. Prepr. 41, 252-253 (2000).
20. W. Theidman, F. C. Tolan, P. J. Pearce and C. E. M. Morris, J. Adhesion 22, 197 (1987).
21. H. Leidheiser, Jr., Ind. Eng. Chem. Prod. Res. Dev. 20, 547-551 (1981).
22. H. Leidheiser, Jr., /. Adhesion Sci. Technol. 1, 79-98 (1987).
23. J. S. Hammond, J. W. Holubka, J. E. DeVries and R. A. Dickie, Corrosion Sci. 21, 239-253
(1981).
24. E. I. Koehler, Corrosion 40, 5-8 (1984).
25. A. Stevenson, Int. J. Adhesion Adhesives 5, 81-91 (1985).
26. A. Stevenson, J. Adhesion 21, 313-327 (1987).
27. D. A. Dillard, J. S. Thornton and F. J. Boerio, Polym. Mater. Sci. Eng. 56, 214^215 (1987).
28. E. P. Plueddemann, Prog. Org. Coatings 11, 297 (1984).
29. P. Walker, in: Silanes and Other Coupling Agents, K. L. Mittal (Ed.), pp. 21-^7. VSP, Utrecht
(1992).
30. D. A. Dillard, K. M. Liechti, D. R. LeFebvre, C. Lin, J. S. Thornton and H. F. Brinson,
in: Adhesively Bonded Joints: Testing, Analysis, and Design, STP 981, pp. 83-97. ASTM,
Philadelphia (1988).
31. K. M. Liechti, E. B. Becker, C. Lin and T. H. Miller, Int. J. Fracture 39, 217-234 (1989).
32. R. F. Hamade and D. A. Dillard, J. Adhesion Sci. Technol. 17, 1235-1264 (2003).
33. S. Naviroj, J. L. Koenig and H. Ishida, J. Adhesion 18, 93 (1985).
Ed. K. L. Mittal
© VSP 2004
Silane coupling agents for silica-filled tire-tread

compounds: the link between chemistry and performance
WILMA K. DIERKES, LOUIS A. E. M. REUVEKAMP,

ANNEMIEKE J. W. TEN BRINKE and JACQUES W. M. NOORDERMEER*
Twente University, Faculty of Chemical Technology, P.O. Box 217, 7500 AE Enschede,
The Netherlands
Abstract—Silanes are commonly used as coupling agents for silica-filled compounds in the rubber
industry, mainly for tire-tread compounds. The replacement of carbon black by silica results in an
improvement of tire performance in terms of wet grip, abrasion resistance and rolling resistance. The
most widely used silanes are bis-(triethoxysilylpropyl)tetrasulfide (TESPT) and the corresponding
disulfide (TESPD). The coupling agent reacts first with the silanol groups of the silica filler, forming a
hydrophobic shell around the filler particle and improving the compatibility between the filler and the
rubber. Secondly, the sulfur moiety reacts with the rubber with formation of a stable network between
the filler and the rubber polymer. The different steps of the formation of the filler-rubber network
and their influence on the in-rubber properties have been investigated separately. The silanisation
efficiency of different types and numbers of alkoxygroups, as well as the effect of the coupling of the
polysulfide groups to the rubber polymer were analyzed. It was found that all functionalities have to
be combined in a single molecule to provide the well-balanced property profile that is characteristic
for silica compounds.
Keywords'. Coupling agent; hydrophobation; rolling resistance; silane; silanisation; silica filler.
1. INTRODUCTION
Silica is constantly gaining importance as a reinforcing filler for rubber compounds.
It offers several advantages compared to carbon black. In tire treads, silica yields a
lower rolling resistance with equal wear resistance and higher wet grip than carbon
black [1]. When polar silica is mixed with generally non-polar dienic hydrocarbon
rubbers, there will be a greater occurrence of hydrogen-bond interactions between
surface silanol groups to form silica agglomerates than interactions between the
polar siloxane or silanol groups of the silica and the rubber. So mixing silica with
rubber poses major problems [2-6].
e-mail: j.w.m.noordermeer@utwente.nl
90 W. K. Dierkes et al.
Therefore, it is necessary to reduce the filler-filler interaction and to enhance the

compatibility of precipitated silica with hydrocarbon rubbers by modifying the sur-
face of the silica. Bifunctional organo-silanes, such as bis(triethoxysilylpropyl)te-
trasulfide (TESPT), are commonly used as coupling agents [7, 8]. The same type
of compounds also enhance the rubber-metal bonding of sulphur-cured rubber poly-
mers [9, 10].
Silica, coupling agent and rubber are usually simultaneously introduced into a
mixer. During mixing, the coupling agent potentially shows two main chemical
reactions as illustrated in Figs 1 and 2 [11, 12]:
• bonding of the organosilane to the silica surface, the primary reaction; and
• reaction between the organosilane and the rubber polymer, based on sulphur
donation or an interaction of the sulphur moiety of the silane with the rubber.
It is assumed that the organosilane does not directly react with the alkoxy groups of
the coupling agent, but with an intermediate of hydrolyzed ethoxy groups. These
intermediate silanol groups of the coupling agent, in turn, react with the silanol
groups of the filler in a condensation reaction. A precondition for the hydrolysis
reaction is the presence of water in the compound. Commercially available silica
types contain approx. 5-6% of water, reported to be the optimal level of moisture for
the silanisation reaction. Investigations have shown that moisture levels up to 7%
OEt
I
EtO — Si—<CH2).
OEt
OEt
I
EtO —Si —(CH2)3
OEt /
I /
EtO—Si —(CH2)3
OEt
I
EtO — Si — (CH2)3-
OEt
^ O E t ^
°\ *\ / Further hydrolysis and

Si—O—Si-^.,/,. , condensation of adjacent
OEt
TESPT molecules
OEt •
°-fScH2)3-
OEt
OEt
(CH a ) 3 -
\
Figure 1. Bonding of TESPT to the silica surface.

Silane coupling agents for silica-filled tire-tread compounds 91
Figure 2. Bonding of TESPT-attached silica to unsaturated hydrocarbon rubber.
result in an acceleration of the silanisation reaction; for higher moisture levels no

further increase of the reaction rate is found [13-16].
The reaction between the organosilane and the rubber can cause premature scorch,
a reaction that should be avoided at this stage. The primary reaction of the coupling
agent during mixing takes place at a relatively slow rate at moderate temperatures.
At the commonly used dump temperature of approx. 120°C for mixing of rubber
compounds the reaction rate is rather low [17]. Therefore, higher batch temperatures
are needed to achieve shorter reaction times. At elevated temperatures, the second
reaction between the coupling agent and the rubber is provoked, resulting in the
aforementioned problem of scorch.
The objective of the present study was to understand the mechanistic aspects of the
reactions between the coupling agent and the silica surface, as well as between the
coupling agent and the rubber polymer, based on the dynamic mechanical properties
of silica-filled tire-tread compounds.
2. EXPERIMENTAL
In order to investigate the effect of different functional groups of TESPT, a series of
different coupling agents were synthesized as shown in Table 1.
TMeSPT shares the sulfur moiety with TESPT, but cannot react with the silica.
TESH can couple to silica, lacks the sulfur group needed to react with the rubber
polymer. TESD can couple to silica and has the sulfur moiety of TESPT replaced
by four methylene groups; it cannot react with the rubber polymer either. TESH
and TESD should react in the same manner, but their hydrophobation efficiencies
are different due to the different numbers of alkyl groups.
All experiments were carried out on a reference tire tread composition shown
in Table 2, and represents a "green tire" recipe in accordance with the Michelin
patent [18] (as the molar masses of the various coupling agents differ from
that of TESPT, the amounts of the alternative coupling agents were adjusted to
represent equimolar quantities). In the case of compounds containing both bis(tri-
ethoxysilyl)decane (TESD) and bis-(trimethylsilylpropyl)tetrasulfide (TMeSPT),
Table 1.
Chemical structures of silane coupling agents investigated
Bis(triethoxysilylpropyl)tetrasulfide (TESPT)
(C2H 5 -0)3-SKCH2)3-S3.83-(CH2)3-Si-(0-C 2 H 5 )3
Bis(trimethylsilylpropyl)tetrasulfide (TMeSPT)
(CH3)3-SKCH2) 3 -S3.83-(CH 2 ) 3 -Si-(CH 3 ) 3
Bis(triethoxysilyl)hexane (TESH)
(C 2 H5-0)3-SHCH 2 ) 6 -Si-(0-C 2 H5)3
Bis(triethoxysilyl)decane (TESD)
(C2H5-O)3-SKCH 2 ) 10 -SKO-C 2 H 5 )3
Table 2.
Tire-tread recipes (parts per hundred parts rubber, phr) with the various coupling agents
Component Coupling agent

TESH TMeSPT TESD + TESPT
TMeSPT
Solution-SBR (Buna® VSL 5025-1 HM, Bayer, 75 75 75 75
Germany)0
Butadiene rubber (Kosyn® KBR 01, Korea Cumho 25 25 25 25
Petrochemical)
Silica (Zeosil® 1165 MP, Rhodia Silica Systems, 80 80 80 80
France)
Coupling agent 5.4 4.7 6.1+4.7 7
Aromatic oil (Enerflex® 75, BP Oil Europe, 32.5 32.5 32.5 32.5
The Netherlands)
ZnO (Merck, Germany) 2.5 2.5 2.5 2.5
Stearic acid (Merck, Germany) 2.5 2.5 2.5 2.5
Sulphur (J.T. Baker, USA) 1.4 1.4 1.4 1.4
CBS (N-cyclohexyl-2-benzothiazylsulphenamide, 1.7 1.7 1.7 1.7
Santocure®, Flexsys, The Netherlands)
DPG (diphenyl guanidine, Perkacit®, Flexsys,
The Netherlands)
Total 228.0 227.3 233.4 229.6
a
Vinyl content: 50%; styrene content: 25%; oil content: 37.5 phr.
the TESD was adjusted to an equimolar quantity of ethoxy groups. The TMeSPT
was adjusted on an equimolar basis to the sulfur present in TESPT. As filler a highly
dispersible silica with a BET surface area of 140-180 m2/g was chosen. This silica
was used as such without any pre-treatment of the surface or drying.
The various ingredients were mixed in three steps. The first two steps were done
in a Brabender Plasticorder lab station internal mixer with a volume of 390 ml.
The mixing chamber was heated up to a temperature of 50°C, and the temperature
control unit was set to 50° C. The fill factor used in this investigation was 66%. After
every mixing step the compound was sheeted out on a Schwabenthan two-roll mill.
The third mixing step was performed on the same two-roll mill, and the accelerators
and sulphur were added during this step.
To study the reactions between the coupling agents and silica as well as with the
rubber, measurements were performed with an RPA 2000 dynamic mechanical rhe-
ological tester from Alpha Technologies (USA), by subjecting uncured compounds
to different strain amplitudes at 100°C and a frequency of 0.5 Hz. When a rubber is
reinforced with a filler, filler-filler interactions take place above a critical filler con-
centration as a result of the decrease of the distance between the filler aggregates in
the rubber compound. At low strain amplitudes, such as the 0.56% minimum strain
specified for the RPA 2000, such filler-filler interactions can be measured particu-
larly well. This effect of decreasing modulus with increasing strain is commonly
known as the Payne effect [19]. The remaining filler-rubber and rubber-rubber
interactions are broken at higher strains of approx. 100% elongation.
3. RESULTS
3.LTESHvs. TESPT
TESH is able to couple to the silica particle via the ethoxy groups in the same
manner as TESPT, but does not possess the sulfur moiety to react with the rubber
polymer. The compound containing TESH was mixed in the Brabender plasticorder
at different rotor speeds to achieve different dump temperatures. A fingerprint of the
mixing curves is shown in Fig. 3. The TESPT-containing compound is also given in
this figure as a reference. All mixing curves for TESH-containing compounds are
lower in end torque compared to TESPT, indicating a lower compound viscosity at
the end of the mixing cycle.
Figure 4A shows the Payne effect of the compounds after mixing. The Payne ef-
fect is directly related to the degree of hydrophobation of the silica by reaction with
400
TESPT (150°C)
TESH(176°C)
TESH(160°C)
TESH(168°C)
TESH(123°C)
200 300 500

Time (sec.)
Figure 3. Mixing curves of TESH- and TESPT-containing compounds. Temperatures indicate the
dump temperatures of the compound.
1,6n
1,4
£1,2
•- 1 0
0,8
SB
d 0,6
@
ID 0,4J
0,2
100 120 140 160 180 200
Dump temp. (°C)
(A)
0,12-,
0,10-
ff 0,081
£ 0.06J
I
P 0,04^
CD 0,02 J
0,00-
100 120 140 160 180
Dump temp. (°C)
(B)
Figure 4. Influence of dump temperature on the storage modulus at 0.56% strain (A) and 100%
strain (B): ( • ) TESPT; (•) TESH.
the silane. The storage modulus G' at 0.56% strain, a measure of the silica-silica in-
teraction, decreases with increasing dump temperature for both coupling agents and
reaches the same level for high dump temperatures. TESH improves the dispersion
of the silica filler particles with increasing dump temperature in a similar manner
as TESPT does [20]. The efficiency of hydrophobation is similar for both coupling
agents, as both molecules have the same number of alkyl groups. These non-polar
groups are responsible for the shielding effect of the coupling agent [21, 22]. The
degree of filler-filler interaction is strongly reduced as a result of the formation of
the hydrophobic shell around the silica particles. The consequence of the reduction
of the filler-filler interaction is the suppression of filler flocculation [23].
On the high strain side (100% strain), the filler-filler interactions are broken and
a chemically cross-linked network prevails (Fig. 4B). The compound with TESPT
124
% 8
E 6
2-\
80 100 120 140 160 180 200

Dump temp. (°C)
(A)
20
10H
-=
to
$ 5
0
80 100 120 140 160 180 "200
Dump temp. °C)
(B)
Figure 5. Tensile properties as a function of dump temperature. (A) 300% modulus, (B) tensile
strength: ( • ) TESPT; (•) TESH.
shows a strong increase in G at dump temperatures above 150°C. This effect is

commonly interpreted as a premature scorch of the compound due either to donation
of reactive, radical form of sulfur to the compound [24] or the coupling reaction of
the sulfur moiety of the already coupled TESPT to the rubber molecules. Both lead
to crosslinking, resulting in an increase in G'. It shows that the sulfur group in
TESPT is the reactive species towards the rubber matrix, because for TESH the G
at 100% strain remains almost constant at a low level.
The effect of the sulfur moiety is also visible in the plot of the 300% modulus
versus dump temperature (Fig. 5A). The trend is similar to the G trend at 100%
strain (Fig. 4B). A significant difference between the two coupling agents is found
for the 300% modulus: In the case of TESPT the modulus shows a sharp increase
at temperatures higher than approx. 160°C, indicating that a pre-scorch reaction
occurs caused by the sulfur contained in the molecule. As TESH does not contain
any sulfur, it cannot form crosslinks with the rubber polymer chains, and no scorch
reaction is observed. In Fig. 5B the tensile strength of the vulcanized compounds Is
plotted against dump temperature. The main difference between the two coupling
agents is a higher tensile strength for the TESPT-containing material compared to
the TESH-containing material.
3.2. TMeSPT vs. TESPT

TMeSPT cannot react with the silica surface, but contains a sulfur moiety to react
with the rubber matrix. TMeSPT was mixed into the tread compound to study the
effect of this sulfur group. Figure 6A shows G' at 0.56% strain as a function of dump
5- •-
•s
\
1$ 4" \
\
\
1 3-
%
58 2 -
o
® 1-
'O •
•
^^•"•"•#" l^H B • •_ •
0- , r
100 120 140 160 180
Dump temp. (°C)
(A)
0,12-
0,10-
S. 0,08-
1 0,06-
0,04-
C9 0,02-
0,00-
80 100 120 140 160 180 200
Dump temp. (°C)
(B)
strain (B): (•) TESPT; (•) TMeSP.
0,30-1
0,25^
0.05H
0,00-1 . 1 . , . 1 . . . , .
100 120 140 160 180 200
Temp. (°C)
Figure 7. Temperature sweep measurements for TMeSPT and TESPT containing compounds
prepared using various dump temperatures. TMeSPT: ( • ) 117°C; (A) 141°C; ( • ) 157°C; TESPT:
( • ) 150°C.
temperature. The storage modulus Gf at low strains gives an indication of the filler-
filler interaction. This shows that TMeSPT does not react with the silica surface due
to the lack of alkoxy groups: the values of Gf at the various dump temperatures are
all significantly higher than those obtained for TESPT. The silica filler particles are
poorly dispersed in the rubber matrix because their surface is not hydrophobized
and, therefore, not compatible with the hydrophobic rubber polymer. The strong
filler-filler interactions contribute to a stronger Payne effect.
In the G' values at 100% strain (Fig. 6B), the effect of the sulfur group on
the coupling agent is visible for TESPT as an increase of the storage modulus
above 150°C. In the case of TMeSPT this effect is not clearly visible, and TMeSPT
shows only a small increase. A strong filler-filler network in presence of TMeSPT
apparently prevails over the sulfur donation or scorch effect of TMeSPT.
Another method for measuring the reactivity of the sulfur group in TMeSPT, as
well as in TESPT, is the temperature sweep, measured with the RPA (Fig. 7). The
increase in G' starts for TMeSPT and for TESPT at the same temperature level
(approximately 160°C). In the case of TMeSPT G1 reaches a higher level caused by
the poor dispersion of the silica, as already mentioned.
The 300% modulus and tensile strength of the TMeSPT-containing compounds
are plotted in Fig. 8. The 300% modulus and tensile strength values for the TMeSPT
compounds are in general lower than for TESPT, but similar to TESH. The sulfur
donating effect of the TMeSPT is also visible as at higher dump temperatures only
a slight increase in the 300% modulus is measured.
33. Effect of combination of TMeSPT and TESD compared to TESPT

Using the combination of TMeSPT and TESD allows a separation of the reaction
with the silica surface on the one hand and with the rubber polymer on the other
15-,
14
13
12
^ 11
& 10
^ 9 -*-#-
_2 8
7
E 6
E
4
8 3
2
1
0
80 100 120 140 160 180 200
Dump temp. (°C)
(A)
• .
16- • •
<<—v
CD • • ^^^"^ •
Q- • ^-~~""^"
_ •
^ • {
£ 14-
B •
c(D
i—
"So •
^ '
1 12- /
|2
• ^ **
• #*
10- 1 1 -* i • 1— • 1 • 1
100 120 140 160 180 200

Dump temp. (°C)
(B)
Figure 8. Tensile properties as a function of dump temperature. (A) 300% modulus, (B) tensile
strength: ( • ) TESPT; (•) TMeSPT.
hand. Each of the two coupling agents is capable of performing only one of
the two functions: TMeSPT is able to couple to the rubber polymer and TESH
is able to hydrophobize the silica surface. In this test series TESD was used,
because it is known from the literature that this coupling agent provides even better
hydrophobation of the silica surface than TESH [25]. In the final product the silica
will be very well dispersed, but no chemical coupling between the silica and the
rubber can take place. An equimolar amount of TESD with regard to the amount of
ethoxy groups in TESPT was used. The same was done for TMeSPT relative to the
amount of sulfur present in the coupling agent.
Figure 9A shows the storage modulus G at 0.56% strain. There is a correlation
between the storage modulus and the dispersion of the filler: the lower the Gf value,
1,2
1,0-J
£
0.8H
c
ft 0,6H
S 0.4H
o 0,2 * •
0,0
80 100 120 140 160 180 200
Dump temp. (°C)
(A)
0,12-
0,10-
| 0,08-
£ 0,06-
I 0,04-
i) 0.02J
ID
0,00-
80 100 120 140 160 180 200
Dump temp. (°C)
(B)
strain (B): ( • ) TESPT; (•) TESD + TMeSPT-containing compounds.
the better the dispersion of the filler. Very low values are obtained compared to,
e.g., TMeSPT (Fig. 6A), indicating a good dispersion of the silica. The effect of
sulfur donation from the coupling agent TMeSPT is visible in Gf at 100% strain
(Fig. 9B). A problem in this context is that the degree of hydrophobation by TESD
is high and, as a consequence, the resulting viscosities of the compounds are low,
making it impossible to reach dump temperatures well above 160°C. Therefore, the
effect of scorch is not very distinct. In this figure the G' at 100% strain increases at
dump temperatures above 150°C, similar to the curve for TESPT (Figs 4B and 6B).
The TMeSPT cannot react with the silica and, consequently, the increase in G at
100% strain must be the result of sulfur donation or coupling to the rubber polymer
matrix.
14
12'
I 10
*-#-
J 8
6
E
80 100 120 140 160 180 200

Dump temp. (°C)
(A)
18
16
14^
J?1 2 .
I
£ 10H
I 8-I
CO
^ 6
£ 4
2-|
080 100 120 140 160 180 200
Dump temp. °C)
(B)
Figure 10. Influence of dump temperature on the modulus at 300% elongation (A) and the tensile
strength (B): ( • ) TESPT; ( • ) TESD + TMeSPT-containing compounds.
Figure 10A shows the 300% modulus versus dump temperature. The values
for the combination of TESD + TMeSPT are significantly lower compared to the
values of the compounds containing TESPT only, which show only a slight increase
with dump temperature. The tensile strength of the TESD + TMeSPT containing
compounds shows the same trend as for TESPT (Fig. 10B), but the absolute values
for the material containing TESD + TMeSPT are at a lower level.
4. DISCUSSION
This series of experiments shows that a proper balance in functionalities is required
for a coupling agent to enhance the properties of silica-reinforced tread compounds.
The utilisation of TESPT gave the best overall properties, because it allows for the
coupling reaction with the silica via its tri-ethoxysilyl groups, and for a sulphur do-
nation, as well as a coupling reaction with the rubber polymer via its sulphur moiety.
Alternative coupling agents possessing only one of these functionalities lead to
an inferior balance of properties. TESH with only a tri-ethoxysilyl group and
no sulphur does hydrophobize the silica and provides a good silica dispersion,
like TESPT. However, the properties of the vulcanised products are considerably
poorer than those obtained with TESPT. On the other hand, TMeSPT, having only
the active sulphur group for coupling to the rubber polymer, fails in its ability
to properly disperse the silica and shows premature scorch at high mixer dump
temperatures. Additionally, the properties of the vulcanised products are inferior to
those obtained using TESPT.
Adding two coupling agents, TESD and TMeSPT, each of them carrying one of
the functionalities, does result in a silica dispersion comparable to the dispersion
obtained with TESPT. Scorch occurs at high mixer dump temperatures, like
for TESPT. However, the properties of vulcanised compounds are still considerably
inferior to those obtained with TESPT.
Only in the case of the TESPT containing compound the full coupling mechanism,
bonding to the silica via the primary and secondary reactions and to the rubber
polymer via the sulphur moiety, is developed. Good dispersion of the silica is
reached and the sulphur reaction with the rubber polymer takes place. Both
reactions are indeed necessary in order to achieve good mechanical properties, as
reflected in the 300% modulus and tensile strength values.
The experiments with TMeSPT and the combination with TESD show that at
elevated dump temperatures the M30o increases when sulphur is present in the
coupling agent. Consequently, the ratio M300/A/100 increases as well. As this
property is related to the rolling resistance, the increase in the M300/M100 ratio
should result in a decrease of the rolling resistance. At high dump temperatures,
i.e., above 150°C, the coupling agent TESPT donates sulphur to the rubber matrix
causing a surplus of sulfur in the neighbourhood of the silica particle. This creates
extra cross-links in the rubber matrix around the silica particle. Thus the silica filler
particle is trapped within the rubber matrix. During vulcanisation at a later stage,
the silica particles are then coupled via the coupling agent to the rubber. This results
in a very tightly cross-linked shell around the silica particle, formed during mixing
in the first silanisation step when the silanol groups of the filler are reacting with the
silane, and during the second step of the silanisation reaction when the remaining
hydrolysed alkoxy-groups are reacting. A second network is formed during the
vulcanisation step when the sulphur-moiety of the coupling agent reacts with the
rubber polymer. A schematic depiction of this model is given in Fig. 11.
5. CONCLUSIONS
For an optimal enhancement of the properties of a silica-reinforced tire-tread
compound a proper balance of the functionalities is required. TESPT gives the
Figure 11. Silica particle embedded in cross-linked rubber polymer shell and coupled to the rubber
polymer via the coupling agent (CA).
best results, as it contains functionalities for the primary and secondary reactions of
the silanisation process, as well as the sulfur moiety for the coupling to the rubber
polymer.
Coupling agents with only one functionality give inferior results in terms of filler
dispersion and performance of the vulcanized material. Coupling agents without a
sulfur moiety such as TESH are able to hydrophobize the silica particles resulting
in a good dispersion, but are unable to chemically react with the rubber polymer.
Therefore, the mechanical properties of the vulcanized material are inferior to the
properties of rubber containing TESPT. On the other hand, a coupling agent able
to connect to the rubber polymer but without functionality for bonding to the filler
does not have the ability to properly disperse the silica.
The addition of two coupling agents, each of them providing one of the function-
alities, does result in a good dispersion of the filler. However, the properties after
vulcanization are inferior compared to TESPT due to the lack of the filler-rubber
polymer network.
Acknowledgements
This research was supported by the Technology Foundation STW, the Applied Sci-
ence Division of the Dutch Scientific Foundation NWO and the technology program
of the Dutch Ministry of Economic Affairs in the Materials Priority Program (MPP).
We gratefully acknowledge the additional financial support provided by an indus-
trial consortium consisting of DSM Elastomers B.V., Flexsys B.V., The Dutch Rub-
ber Foundation, Schill & Seilacher, Vredestein Banden B.V. and OSi Specialties,
Crompton Corporation.
REFERENCES
1. S. Wolff, Tire Sci. TechnoL 15, 276 (1987).

2. D. Berkemeier, W. Haeder and M. Rinker, Rubber World, 34 (July 2001).
3. L. A. E. M. Reuvekamp, W. Dierkes and J. W. M. Noordermeer, Tire TechnoL Int. 44
(March 2003).
4. W. Dierkes, J. W. M. Noordermeer, M. Rinker, K.-U. Kelting and C. van de Pol, Kautschuk
Gummi Kunststoffe 56, 338 (2003).
5. W. Dierkes, Paper presented at the International Rubber Conference 2003, Nurnberg, June 30-
July 3 (2003).
6. L. A. E. M. Reuvekamp, J. W. ten Brinke, P. J. van Swaaij and J. W. M. Noordermeer, Rubber
Chem. TechnoL 75, 187 (2002).
7. B. T. Poh and C. C. Ng, Eur. Polym. J. 24, 975 (1998).
8. A. Scurati, D. L. Feke and I. Manas-Zloczower, Paper presented at the ACS Rubber Division
Meeting, Cleveland, OH, October 16-19 (2001).
9. S. K. Jayaseelan and W. J. van Ooij, Gummi Asbest Kunststoffe 56, 497 (2003).
10. M. J. Moore, Paper presented at the ACS Rubber Division Meeting, Cleveland, OH, Octo-
ber 16-18 (2001).
U . S . Wolff, Kautschuk Gummi Kunststoffe 34, 280 (1981).
12. A. Hunsche, U. Gorl, A. Mtiller, M. Knaack and Th. Gobel, Kautschuk Gummi Kunststoffe 50,
881 (1997).
13. H.-D. Luginsland and A. Hasse, Paper presented at the ACS Rubber Division Meeting, Dallas,
TX, April 4-6 (2000).
14. U. Gorl and A. Hunsche, Paper presented at the ACS Rubber Division Meeting, Louisville, KY,
October 8-11(1996).
15. H.-D. Luginsland, Paper presented at the Slovak Rubber Conference Meeting, Puchov,
May 25-26 (1999).
16. U. Gorl and A. Parkhouse, Kautschuk Gummi Kunststoffe 2,493 (1999).
17. A. Hunsche, U. Gorl, H.G. Koban and T. Lehmann, Kautschuk Gummi Kunststoffe 51, 525
(1998).
18. R. Rauline, European Patent 0501227 A1 (to Compagnie Generale des Etablissements
Michelin-Michelin & Cie), (1992).
19. A. R. Payne, Rubber Chem. TechnoL 33, 365 (1966).
20. L. A. E. M. Reuvekamp, J. W. ten Brinke, P. J. van Swaaij and J. W. M. Noordermeer, Kautschuk
21. I. Ladouce, Y. Bomal, I. Flandin and D. Labarre, Paper presented at the ACS Rubber Division
Meeting, Dallas, TX, April 4-6 (2000).
22. J. W. ten Brinke, S. C. Debnath, L. A. E. M. Reuvekamp and J. W. M. Noordermeer, Composit.
Sci. TechnoL 63, 1165 (2003).
23. C.-C. Lin, W. L. Hergenrother, E. Alexanian and G. G. A. Bohm, Paper presented at the ACS
Rubber Division Meeting, Cleveland, OH, October 16-19 (2001).
24. S. Debnath, Paper presented at the ACS Rubber Division Meeting, Savannah, GA, April 29-
May 1 (2002).
25. J. W. ten Brinke, P. J. van Swaaij, L. A. E. M. Reuvekamp and J. W. M. Noordermeer, Kautschuk
Ed. K. L. Mittal
© VSP 2004
Electrodeposition of organofunctional bis-silanes and their

effectiveness in providing corrosion resistance of metals
JASPREET S. GANDHI and WIM J. VAN OOIJ *

Department of Chemical and Materials Engineering, University of Cincinnati, Cincinnati,
OH 45221-0012, USA
Abstract—Hydrolyzed silanes of the type (HO)3Si(CH2)3Y(CH2)3Si(OH)3, in which Y is a func-

tional group, are water-soluble, ionized molecules, so it should be possible to deposit them on metals
by electrodeposition. Various combinations of silane mixtures were tested at different voltages, pH
values, bath concentrations and exposure times on panels of aluminum alloys. The corrosion re-
sistance of silane-coated aluminum panels was investigated by DC polarization and electrochemical
impedance spectroscopy techniques. The thickness of the silane films was measured by ellipsometry.
Electrodeposition results in a more organized and uniform silane film with less porosity as compared
to immersed films. The layer formed at the interface between the silane film and the metal oxide
has a high ohmic resistance and low permeability to electrolyte. Especially the resistance to pitting
corrosion is improved by the electrodeposition process as compared to the dipping process.
Keywords: Electrodeposition; organofunctional silanes; corrosion; aluminum; coatings.
1. INTRODUCTION
Organofunctional silanes have recently emerged as an environmentally-friendly
anti-corrosion treatment for metal substrates and have shown the potential to
replace conventional chromate treatments [1, 2]. The use of chromates is being
discouraged owing to the carcinogenicity and environmental hazards related to the
toxic hexavalent chromium ions.
Organofunctional silanes are hybrid organic-inorganic compounds having the
structure X3Si(CH2)n Y, where X represents a hydrolyzable group such as methoxy
or ethoxy and Y is an organofunctional group. When the silane is symmetrical
about the functional group Y, i.e., if there are two trialkoxy (X3) groups in the
molecule, then these molecules are bis-functional silanes having the structure
X3Si(CH2)rtY(CH2)nSiX3.
e-mail: vanooiwj@email.uc.edu
106 J. S. Gandhi and W. J. van Ooij
Various techniques have been used to coat metal surfaces with these silanes,
e.g., by immersion, brush or spray coating [1, 2], However, these techniques
produce a non-uniform, sometimes porous and poorly uncontrolled thickness of
the film. This paper demonstrates a new technique for coating silanes on aluminum
surfaces, i.e., by electrodeposition. Although Woo et al. [3] used this technique for
depositing silanes on metal surfaces, their emphasis was on adhesion properties of
the films formed. They performed lap-shear tests with the silane-treated surfaces
to evaluate bond strength and durability. According to these authors, as the voltage
is applied, hydroxyl anions are generated at the cathode and the cationic silane
containing a protonated amine group is deposited on the cathode by the alkali-
catalyzed condensation reaction. The results obtained were similar, and in some
cases better, than compared with the immersion techniques.
The data presented in this paper are concerned with the metal corrosion-resistance
properties of the films. Our work is based on the use of bis-silanes. Since
hydrolyzed silanes are water-soluble, ionized molecules, they dissociate so it should
be possible to electrodeposit them. At high pH, R-SiO~ anions are present so they
will be attracted towards the anode. Similarly, at low pH, cations of the type R-
NH3" or R-SiOH2 are attracted towards the cathode. Various combinations of silane
mixtures were tested at different voltages, pHs, bath concentrations and exposure
times on panels of aluminum 5005 alloys. Good results were obtained with a typical
bis-silane, bis-[trimethoxysilylpropyl]amine, whose structure is shown below.
0 C 3
f OCH 3
Si CH2-CH2-CH2-NH-CH2-CH2-CH2
OCH3 OCH3 ' OCH3
This silane has three methoxy groups on each Si atom which, on hydrolysis, form
up to six silanol Si-OH groups. These silanols are assumed to dissociate as follows:
(HO)3Si-(CH2)3-NH-(CH2)3-Si(OH)3 neutral
(HO) 3 Si-(CH 2 )3-NH-(CH 2 ) 3 -Si(OH) 2 0- at high pH
(HO)3Si-(CH2)3-NH+-(CH2)3-Si(OH)3 at low pH
or
(HO) 3 SHCH 2 ) 3 -NH 2 -(CH 2 ) 3 -Si(OH) 2 OH+ at low pH
The performance of the films formed at different pH values was investigated by

DC polarization and electrochemical impedance spectroscopy (EIS). Based on the
electrochemical data, a mechanism is proposed.
Electrodeposition of organofunctional bis-silanes 107
2. EXPERIMENTAL
2.1. Materials
The silane bis-[trimethoxysilylpropyl]amine, also known as A-1170 Silquest®,
was obtained from OSi Specialities (Greenwich, CT, USA). This silane was used
without further purification. Two solvent-based silane solutions were prepared, one,
5:75 :20 (v/v/v), by mixing 5 vol% bis-amino silane with 75 vol% methanol and
20 vol% DI water and the other, 5:90:5 (v/v/v), with 5 vol% bis-amino silane
with 90 vol% methanol and 5 vol% DI water. The natural pH of the resulting
solutions was 10 and was adjusted to 8-8.5 by the addition of acetic acid. The
solution was continuously stirred for 24 h for hydrolysis. Substrates for deposition
were Al-5005 panels (15 cm x 10 cm x 0.06 cm) obtained from ACT laboratories,
Hillsdale, Michigan. Prior to deposition, these panels were cut into smaller size
(5 cm x 5 cm x 0.06 cm) and were degreased ultrasonically in hexane and ethanol
for 5-7 min and ?>-A min, respectively. The substrates were then dipped into a dilute
alkaline cleaner (AC1055®, provided by Brent, Lake Bluff, IL, USA) at 60-70°C
for 5-7 min, after which they were rinsed with DI water and air dried.
The 0-30 V voltage supply required for electrodeposition was built by Sanction
Electronics (Trenton, OH, USA).
2.2. Silane deposition

2.2.1. Immersion. The cleaned aluminum substrates were immersed into the
silane solution for 10 min. After immersion the substrates were rinsed by the same
solution and dried at 100°C for 10 min in an oven. Curing in the oven was done to
crosslink the film.
2.2.2. Electrodeposition. The set-up comprised of an electrolytic cell with

the aluminum substrate and graphite as the opposite electrodes. Two series of
experiments were performed:
1. Substrate as anode and graphite electrode as cathode.
2. Substrate as cathode and graphite electrode as anode.
A series of positive and negative voltages were applied for 10 min. After
electrodeposition the substrates were dried at 100°C for 10 min in an oven.
2.3. Electrochemical techniques

2.3.1. DC polarization test. One sample of each type (uncoated, immersed,
electrodeposited) was kept immersed in neutral 0.6 M NaCl solution for 2 h and
a DC polarization test was performed. The reason for pre-immersion of the panels in
the electrolyte before testing was to achieve a steady state. On average, 3 replicates
were tested for each condition. The data were recorded at potentials EC0TT ± 250 mV
(ECOTY = corrosion potential or open circuit potential, OCP), with a scan rate of
1 mV/s.
A saturated calomel electrode SCE and a platinum mesh were used as the
reference and counter electrodes, respectively. The exposed area was 0.78 cm2.
2.3.2. Electrochemical impedance spectroscopy (EIS). EIS measurements on

Al-5005 panels coated with bis-amino silane was carried out in 3.5 vol% NaCl
aqueous solution, using an SR 810 frequency response analyzer and a Gamry
CMS 100 potentiostat. Impedance data were recorded at frequencies ranging
from 10~2 to 105 Hz, with alternating current (AC) voltage amplitude of ±10 mV.
A commercial saturated calomel electrode (SCE) served as the reference electrode,
coupled with a graphite counter electrode. An area of 3.14 cm2 of the specimen was
exposed to the electrolyte during the measurement. All EIS spectra were recorded
after immersing one sample of each kind (uncoated, immersed, electrodeposited) in
3.5 vol% NaCl for different intervals of time.
2.4. Thickness measurement

A Variable Angle Spectroscopic Ellipsometer manufactured by J. A. Woollam was
used to measure the thickness of the silane film formed on the surface. The
instrument was used in a rotating analyzer set-up. For the analysis of the spectra, the
WVASE32 software was used. The measurements were done at different points on
the surface at four angles of incidence, 60°, 65°, 70° and 75°, with wavelengths
in the range of 300-800 nm. In order to ensure an identical position of the
measurement spot on the sample at all angles of incidence, the instrument was
equipped with a two-step alignment procedure.

3.1. DC polarization results
The polarization curves for blank, immersed, and electrodeposited samples are
shown in Fig. 1. The corrosion potential of + 10 V sample is lower by —0.250 V as
compared to the immersed sample. Similar results are obtained for +2 V and +5 V
samples which showed a trend of decreasing corrosion potential with increasing
electrodeposition voltage. However, the corrosion current icorT remains the same
for immersed and electrodeposited samples. These results suggest that, although
the overall corrosion rate is not reduced, the electrodeposition leads to cathodic
inhibition which may ultimately decrease the rate of anodic reaction also. In the
case of negative voltage (Fig. 2), the same trend is observed with —2 V and —5 V
curves, i.e., the corrosion potential is lowered. The test curve for —10 V is not
shown here, because when —10 V was applied to the substrate, the silane film fully
condensed on the surface. It could, however, be scratched off easily.
3.2. EIS results

The results of DC polarization were confirmed by the EIS curves obtained for the
positive voltage samples. The impedance curves (Fig. 3a) and phase angle curves
-0.500
-0.600
t -0.700
UJ
£ -0.800|
>
15
•*-»
c
S
-6.0 -5.0 -4.0 1.0

2
Log Current Density (A/cm )
Figure 1. DC polarization curves showing a comparison of blank, immersed, +2 V, +5 V and +10 V

electrodeposited samples. The samples were tested after 2 h immersion in 3.5 vol% NaCl solution.
The silane used was bis-amino 5 :75 :20 (vol% silane/ethanol/water).
-6.0 -5.0 -4.0
Log Current Density (A/cm2)
Figure 2. DC polarization curves showing a comparison of blank, immersed, - 2 V and —5 V

electrodeposited samples. The samples were tested after 2 h immersion in 3.5 vol% NaCl solution.
The silane used was bis-amino 5 :75 :20 (v/v/v).
(Fig. 3b) showed a trend of increasing impedance values for immersed, +2 V, +5 V

and +10 V samples even after 9 days of exposure to 3.5 vol% NaCl salt solution. On
the other hand, the curves for the samples obtained with - 2 V and —5 V coincided
with the curve of immersed sample (Fig. 4a and 4b). These results indicated
that, although negative voltage showed cathodic inhibition in DC polarization, the
impedance value remained the same for all of them. Secondly, with higher voltages,
6.00
Immersed
+ 10V
-2.00 -1.00 0.00 1.00 2.00 3.00 5.00
Log Freq (Hz)
(a)
-1.00 0.00 1.00 2.00 3.00 4.00 5.00
Log Freq (Hz)
(b)
Figure 3. (a) Impedance curves showing a comparison of immersed, +2 V, +5 V and +10 V
electrodeposited samples after 24 h of immersion in 3.5 vol% solution, (b) Phase-angle plot showing
a comparison of immersed, +2 V, +5 V and +10 V electrodeposited samples after 24 h of immersion
in 3.5 vol% solution. The silane used was bis-amino 5 :75 :20 (v/v/v).
— 10 V and over, the film started to condense on the substrate and could be scraped
off easily. Moreover, the 5 :75 :20 (v/v/v) silane solution also lost its stability and
gelled after 5 days. The reason was the high pH of the solution. Since dissociation
of a silane is highly dependent on the pH of the solution, another solution 5 :90:5
(v/v/v) was tried and the procedure was repeated. In the 5:75:20 (v/v/v) silane
solution, +10 V showed the best performance, so electrodeposition for the 5:90:5
(v/v/v) silane solution was also done at this voltage only. Figure 5 shows the
DC polarization curves for blank, immersed and +10 V samples. The corrosion
current iC0IT, obtained by the intersection of extrapolated anodic and cathodic curves
for +10 V electrodeposited film, shows a shift of half decade towards the left as
compared to the immersed film. At the same time the ECOTT for both films remains
the same clearly indicating about equal anodic and cathodic inhibition in the case
of electrodeposited films.
When a positive voltage is applied to the substrate in the case of 5:75:20 (v/v/v)
silane solution, the high water content of the solution results in a high flux of
dissolved aluminum ions, As a result, the R-SiO~ ions are not deposited on the
anodic sites and the substrate shows a lower Econ due to cathodic inhibition only.
IMMmt
1.00
-2.00 -1.00 0.00 1.00 2.00 3.00 4.00 5.00
Log Freq (Hz)
(a)
1.00 2.00 3.00 4.00 5.00
Log Freq (Hz)

(b)
Figure 4. (a) EIS curves showing a comparison of immersed, —2 V and —5 V electrodeposited
samples. The samples were tested after 24 h of immersion in 3.5 vol% NaCl solution, (b) Phase-angle
plot showing a comparison of immersed, —2 V and —5 V electrodeposited samples. The samples were
tested after 24 h of immersion in 3.5 vol% NaCl solution. The silane used was bis-amino 5 :75 : 20
(v/v/v).
However, when the same voltage is applied in the case of the 5 :90:5 (v/v/v) silane
solution then, due to lower water content, the flux of dissolved aluminum ions
decreases. As a result, the silane deposits on both the anodic and cathodic sites
resulting in a lower /corr.
On the other hand, when a negative voltage is applied to the substrate in the
case of the 5:75:20 (v/v/v) silane solution, water decomposition occurs at the
substrate resulting in alkali-catalyzed condensation of the silane at high voltage.
Moreover, due to a high aluminum flux R-SiO - is deposited only on the cathodic
sites which results in a lower £ corr . The aluminum flux consists of AlO^" ions in this
case.
The EIS curves, Fig. 6a and 6b, also show a higher value of impedance for the
+ 10 V electrodeposited sample as compared to the immersed sample. The EIS
test was conducted continuously for 9 days and the same trend was observed.
-0.500
-1.200
-5.5 -4.5 -2.5
Log Current Density (A/cnri2)
Figure 5. DC polarization curves showing a comparison of blank, immersed and +10 V electrode-
posited samples. The samples were tested after 2 h immersion in 3.5 vol% NaCl solution. The silane
used was bis-amino 5 :90:5 (v/v/v).
Based on the trend, the film pore resistance (/?p0) was calculated for both im-
mersed and electrodeposited films by modeling the EIS curves [4-10]. The curve
fitting was done using the equivalent circuit model shown in Fig. 7. The EIS
curves in Fig. 6a and 6b showed only the capacitive behavior of the film, but in
the later stages of exposure the phase angle plot showed emergence of film pore
resistance which also became clear in the impedance plot with time. Figure 8
shows the trend of pore resistance with time for both electrodeposited and im-
mersed films, clearly indicating higher pore resistance for the electrodeposited film
initially.
The samples were removed from the salt exposure after 9 days and were visually
inspected (Fig. 9). The untreated substrate showed uniform corrosion all over
the exposed area. The immersed sample showed localized corrosion in the form
of pitting at various points, but the electrodeposited +10 V substrate remained
clean, i.e., showed no corrosion. These results confirmed the argument that
electrodeposited sample led to a slower corrosion rate as compared to the immersed
sample.
A plausible mechanism for silane-metal interaction by electrodeposition could
be that at high pH the functional group R-NH^ (cation) is not active, so when
a negative voltage is applied to the substrate no special effect occurs and the
electrodeposited curves coincide with the immersed curves in the EIS test. On the
contrary, the R-SiO~ anions are highly active at this basic pH so when a positive
voltage is applied to the substrate, the R-SiO~ ions are attracted towards the positive
substrate and are deposited (Fig. 10).
0.00 1.00 2.00 3.00 4.00 5.00

L o g F r e q (Hz)
(a)
20.00 -
Immersed
*5T -D-+10V
2 0.00 -
Ol
O)
G- -20.00-
03
Ol
= -40.00 -
N / y
« \ X •
ca -60.00 - A. y a
^ ^ S s ^ _ ^ ' ' r j F
-80.00 -
-100.00 - 1 1 1 1 , 1
-2.00 -1.00 0.00 1.00 2.00 3.00 4.00 5.00
Log Freq (Hz)

(b)
Figure 6. (a) Impedance curves showing comparison of immersed and +10 V electrodeposited films.
The samples were tested after 24 h of immersion in 3.5 vol% NaCl solution, (b) Phase-angle plot
showing comparison of immersed and +10 V electrodeposited films. The samples were tested after
24 h of immersion in 3.5 vol% NaCl solution. The silane used was bis-amino 5:90:5 (v/v/v).
CPE (Y0,n)
o-
Figure 7. Equivalent circuit model for EIS data fitting. Rso is the uncompensated solution resistance,
CPE is the constant phase element, YQ and n are the CPE parameters, /? po is the film pore resistance.
45 i
Immersed
m
, 40 \
35 + 10V
X 30
V)
| 25
£ 20
g 15 }
g 5 i
£ 0
o- -5 -6 6
Time (Days)
Figure 8. Film pore resistance vs. time (days) for immersed and electrodeposited bis-amino silane
films. The silane used was bis-amino 5 :90 :5 (v/v/v).
3.3. Ellipsometry
The silane film thickness was measured at three different points on a single substrate
for both immersed and electrodeposited films. A two-layer optical model (Fig. 11)
was used for curve fitting and determining the silane layer thickness and non-
uniformity. Since the surface was manually polished, a Bruggeman effective mean
approximation (EMA) layer was considered at the film/substrate interface to take
into account the effect of roughness or any diffused oxide layer. Also, a non-
ideal model was used to incorporate the non-uniformity of the film. A detailed
discussion about the model can be found elsewhere [11, 12]. The immersed sample
had a thickness of 512 ± 15 nm with 18% non-uniformity. The EMA layer for the
immersed film comprised of 6% aluminum and 94% bulk silane film. In the case of
the +10 V electrodeposited film, the thickness observed was 350±24 nm with 15%
non-uniformity and 9% aluminum in the EMA layer.
The thickness data were used in the following equations to calculate the dielectric
constant of the films for both immersed and electrodeposited samples [13].
Z(CPE) = (Mr
Yo
where Z is the impedance of constant phase element (CPE) in ohm, co is the angular
frequency of the input signal (rad/s), n and Fo are the CPE parameters. Using these
values the film capacitance C can be obtained from the following expression
Y0co>7 1 - 1
C =
sin (ft j )
Figure 6a shows that at low frequency the films possess a fully capacitive behavior,
and the values of capacitance for immersed and electrodeposited films were 3.16 x
10"4 and 1 x 10~4 F, respectively, at a frequency of 0.16 Hz. The film capacitance
always tends to increase with immersion time as a result of water uptake [14]. The
Al-5005 immersed in 5/90/5 bis-amino

silane solution at pH 9
Al-5005 with silane electrodeposited at +10 V

in 5/90/5 bis-amino silane solution at pH 9
Figure 9. Blank, immersed and +10 V electrodeposited silane-treated samples after 9 days of
3.5 vol% salt solution exposure. The silane used was bis-amino 5:90:5 (v/v/v).
reason for such increase is due to a significant increase of the dielectric constant (s)
of the coating, which is influenced strongly by water penetration into the coating.
Assuming the film thickness remains constant during water penetration, the film
OH
Si,- CH 2 -CH 2 -CH 2 -NH-CH 2 -CH 2 -CH 2 Si'
OH "OH
Oxide layer
!> ',1 '.^ '/• 1> 1> •> •.*• '•> "> '.^ 1> *> "•> "•*» '•*• 1% '/• 1>t> :> ' ^ 1^ '/• ".1 1> 'A "•
^'/j^VW_»J'?^WVWV'^Vf^'^^lf'/V^^J.
Positively charged aluminum substrate

Figure 10. Silane-substrate interaction mechanism when positive voltage is applied to the substrate.
coating relation Thickness fmm}\

Sont niformity (%)
— — - - ^ ^ ^ w ^ —
EM A layer .. m
C^v ^"""^rs j'*wwS jr^*-"'"" '"^ j p r ^ y ^ j^****? ' fCracky + %% Al)
Thkknesff fnmj
in
T?j™" >ig[iij "^HET"'"1' pjp|
A I (pmido-constants
\ At \ naiulk)
Presumed structure Optical model

Figure 11. Two-layer optical model used to calculate the thickness and non-uniformity of the silane
film. EMA layer is the Bruggeman effective mean approximation layer.
capacitance can be defined as [15, 16]

A
C = e0s — ,
a
where C is the film capacitance in F, A is the exposed sample area in cm2, d is
the thickness in cm, s is the dielectric constant of the electrolyte-free silane film
and £o is the permittivity of vacuum = 8.87 x 10 - 1 4 F/cm. Incorporating the values
of C in the above equation yields simm = 3.16 Electro- This suggests that when both
films are exposed to the test electrolyte, diffusion of water into the film occurs with
time, but due to lower pore resistance of the immersed film the dielectric constant
increases more than the electrodeposited film.
4. CONCLUSIONS
This study shows that electrodeposition of organofunctional silanes on metals
has the potential to result in a more organized, void-free and uniform film with
less porosity at the metal surface. The resistance to pitting corrosion of metal
is improved by this technique, as compared with the more common immersion
process.
5. FUTURE WORK
A further in-depth study of the structure of the film is required which includes
characterization of the surface morphology by various analytical techniques such
as FT-IR, XPS, TOFSIMS and AFM. Since it has become clear from the above tests
that the functional group of the silane does not play any significant role during
electrodeposition, other non-functional silanes such as bis[triethoxysilyl]ethane
(BTSE) at high pH can also be used. The performance of paint coatings on
electrodeposited silane films will also be studied.
Acknowledgements
The authors gratefully acknowledge the American Electroplaters and Surface Fin-
ishers Society (AESF) for financial support. Mr. Terry Craycraft is acknowledged
for building the power supply, Matt Stacy for project guidance and Akshay Ponda
for his help with the ellipsometry.
REFERENCES
1. W. J. van Ooij, D. Zhu, G. Prasad, S. Jayaseelan, Y. Fu and N. Teredesai, Surface Eng. 16, 386
(2000).
2. W. J. van Ooij and T. Child, Chemtech 28, 26 (1998).
3. H. Woo, P. J. Reucroft and R. J. Jacob, /. Adhesion. Sci. Technol. 7, 681 (1993).
4. D. Zhu and W. J. van Ooij, J. Adhesion ScL Technol. 16, 1235 (2002).
5. W. S. Tait, An Introduction to Electrochemical Corrosion Testing for Practicing Engineers and
Scientists. PairODocs Publications, Racine, WI (1994).
6. J. N. Murray, Prog. Org. Coatings 31, 375 (1997).
7. V. Subramanian and W. J. van Ooij, Corrosion 54, 204 (1998).
8. M. Kendig and J. Scully, Corrosion 46, 23 (1990).
9. P. L. Bonora, F. Deflorian and L. Fedrizzi, Electrochim. Acta 41, 1073 (1996).
10. M. D. G. Destreri, J. Vogelsang, L. Fedrizzi and F. Deflorian, Prog. Org. Coatings 37, 69 (1999).
11. W. J. van Ooij and D. Zhu, Corrosion 57, 413 (2001).
12. A. Franquet, C. Le Pen, H. Terrynn and J. Vereecken, Electrochim. Acta 48, 1245 (2003).
13. A. Franquet, J. De Laet, T. Schram, H. Terryn, V. Subramanian, W. J. van Ooij and J. Vereecken,
Thin Solid Films 384, 37 (2001).
14. E. P. M. van Westing, G. M. Ferrari and J. H. W. de Wit, Corrosion Sci. 36, 979 (1994).
15. J. Ross MacDonald, Impedance Spectroscopy, Wiley, New York, NY (1987).
16. E. P. M. van Westing, G. M. Ferrari and J. H. W. de Wit, Corrosion Sci. 34, 1511 (1993).
Ed. K. L. Mittal
© VSP 2004
Silane coupling agents as corrosion fatigue inhibitors
ANUJ SETH and WIM J. VAN OOIJ *

OH 45221-0012, USA
Abstract—The prevention of corrosion of aerospace alloys is accomplished using chromate com-

pounds as corrosion inhibitors. Chromate processes have also been shown to prevent the alloys from
corrosion fatigue cracking and to enhance their life under conditions of combined corrosive envi-
ronment and cyclic stress applications. However, the presence of Cr(VI) ions in chromates with the
consequent health hazards has led to the restrictions imposed on the use of these systems. Since
they are economically competitive, provide excellent corrosion inhibition and possess environmen-
tal compatibility, organo-functional silane coatings can be considered as prospective replacements for
chromating technology. In this project we have investigated the efficiency with which various bis-type
silanes, such as bis-amino, bis-sulfur and bis-ureido silanes, can be used to prevent corrosion fatigue
cracking of AA2024-T3 and AA7075-T6 alloys along with prevention of bare corrosion, as already
shown in previous research. The testing encompasses loading a dog-bone specimen at 65% of its
ultimate tensile strength for a 7075-T6 alloy and at 70% of its ultimate tensile strength for a 2024-T3
alloy. The dog-bone specimens are loaded and alternated with salt spray exposure. This cycle of
fatigue loading and salt spray continues until the sample fails. As control specimens we are compar-
ing the silane-coated with chromated samples and uncoated samples. The results indicate that some
silanes may increase the corrosion fatigue life for one alloy and not for the other when compared with
the chromated counterpart, but when compared to the blank, uncoated ones a positive effect is always
observed.
Keywords: Silanes; corrosion fatigue cracking; corrosion inhibition; chromate conversion coatings.
1. INTRODUCTION
Aluminum AA 2024-T3 and aluminum AA 7075-T6 alloys are widely used in
the aerospace industry. The AA 2024-T3 alloy containing copper has superior
mechanical properties compared to the base metal due to the different phases present
in the Al-Cu system. The formation of the precipitates in these alloys due to the
presence of alloying elements results in improved mechanical properties of both
120 A. Seth and W. J, Van Ooij
alloys. The Cu in the AA 2024-T3 alloy results in the CuAl2 precipitates [1]. The
AA 7075-T6 alloy contains zinc and magnesium as the main alloying elements to
aluminum. These form a precipitate of MgZn2 which imparts higher strength to this
alloys [2].
The presence of intermediate precipitates results in a drastic decrease in corrosion
resistance of the alloy. The corrosion resistance in bare aluminum and its alloys is
attributed to the thin, well-adherent, uniform, coherent and tough oxide layer present
on the surface of the alloy. For the AA 2024-T3 and 7075-T6 alloys, the oxide is
non-uniform due to the intermetallic particles, which results in the formation of
numerous corrosion cells. Using a layer of pure aluminum on top can rectify this
problem. However, the alloy still remains prone to stress corrosion cracking under
extended service conditions [3].
Various models have been proposed for describing the phenomenon of corrosion
fatigue cracking. Goswami and Hoeppner [4] proposed a 7-stage model for
corrosion fatigue crack propagation. The seven stages that contribute to the failure
due to corrosion fatigue are as follows: (1) pit nucleation, (2) pit growth, (3) pit
transition to short crack, (4) short crack growth, (5) short crack transition to long
crack, (6) long crack growth and (7) failure.
The rate of growth of a short crack is observed to be higher than the long crack
growth [5]. The life for corrosion fatigue cracking has been proposed as the sum of
four time periods. The time for pit nucleation, the time for pit propagation, the time
for short crack growth and the time for long crack growth.
The chromating process is used in the industry for corrosion protection of the
aerospace alloys. This involves dipping the cleaned alloy into an acid bath
containing chromic acid in which the alloy develops a conversion coating of
chromate on its surface. This conversion coating slowly leaches out chromate
ions out from the coating when damaged, which imparts a "self-healing" effect
to the coating. It also provides excellent adhesion to topcoats. However, this
process employs Cr6+ ions, which have been shown to be carcinogenic and are
defined as hazardous by the US Environmental Protection Agency (EPA). The use
of chromates has been proven to be harmful to both the people working in the
chromating industry and the end users. Also, chromate conversion coatings have
been known to reduce the fatigue life of the alloys [6]. The development of an
alternate technology has been slow, and silane-based techniques have shown a good
promise in this regard [1].
In this study it is investigated if the results of the efficiency with which the
bis-silanes [1, 7-12], besides acting as coupling agents [13], can be used for the
enhancement of the corrosion fatigue life. The effect of the coating of a silane film
on the fatigue life of the sample without any corrosive environment involved in the
testing was also studied and is reported. The results presented in this paper give
trends and suggest that silane coupling agents can be used for corrosion fatigue
inhibition. The research activity is still in progress.
Silane coupling agents as corrosion fatigue inhibitors 121
2. EXPERIMENTAL
2.1. Materials and Sample Specimens
The silanes used in this research work are listed in Table 1. They were obtained
from OSi Specialties (Greenwich, CT, USA).
Sol-Gel AL 9201, an aqueous aluminum coating solution, was also used in the
study. The Sol-Gel was acquired from Chemat (Northwood, CA, USA). This was
mixed in various ratios with the hydrolyzed silane solutions and used for coating
the samples.
The concentration of the silane solution was maintained at 5% by volume for the
bare corrosion protection on bare metal substrates without further painting.
Dog-bone specimens of the dimensions shown in Fig. 1 were machined from
coupons of AA 2024-T3 and AA 7075-T6 of thickness 2 mm according to
ASTM E466 specifications. The AA 7075-T6 coupons were obtained from Amer-
ican Metals Supply (Cincinnati, OH, USA) and AA 2024-T3 coupons were from
Copper and Brass Sales (Dayton, OH, USA). The samples were machined for load-
ing in the direction of rolling.
2.2. Test Procedures

2.2.1. Silane solution preparation. Several silanes, individually and in combi-
nations with other silanes, were tested for their effect on the number of cycles to
Table 1.
Silanes investigated for corrosion fatigue crack resistance
Silane formulation
Silane 1 Bis-[triethoxysilylpropyl]tetrasulfide (5% ethanol-based solution)
Silane 2 Bis-[trimethoxysilylpropyl]urea (5% water based, pH 6)
Silane 3 Bis-[trimethoxysilylpropyl]amine/vinlytriacetoxysilane (5% water
based) in a ratio of 1.5 : 1
based) in a ratio of 5 :1
based) in a ratio of 4 : 1 , with 10 ml Sol-Gel
based) in a ratio of 4 : 1 , with 20 ml Sol-Gel
Silane 7 Bis-[trimethoxysilylpropyl]amine and bis-[triethoxysilylpropyl]tetrasulfide
(5% ethanol) in a ratio of 1 :3
Silane 8 Tris-[3-trimethoxysilylpropyl]isocyanurate (5% ethanol-based
solution)
Silane 9 Tris-[3-trimethoxysilylpropyl]isocyanurate/bis-[triethoxysilylpropyl]tetrasulfide
(5% ethanol based solution) in a ratio of 3 :1
based) in a ratio of 4:1, with 200 ppm Ce(NO)3
based) in a ratio of 4 : 1 , with 1000 ppm potassium dichromate
122 A. Seth and W. J. Van Ooij
A k
_^y
1.30 cm
2.54 cm
^ ^ >
< 3.18cm—•
^ 3.81cm w
t
-15.19 cm— •
Figure 1. Dog-bone specimen dimensions.
failure in the presence and absence of a corrosive environment. Before coating the
samples the silanes were hydrolyzed. The hydrolysis of the silanes was carried out
by addition of water or/and alcohol to the silanes. Hydrolysis of the silane does not
need to be complete for the silane to form a corrosion-resistant coating. However,
sufficient silanol groups must be formed during the hydrolysis for the formation of
Si-O-Al bonds.
Silane 1 was a 5% by volume bis-[triethoxysilylpropyl]tetrasulfide silane. The
bis-silane was hydrolyzed by mixing the silane, DI water and ethanol in the ratio
of 5 :5 :90. This results in a 5-vol% of the above silane solution. The pH value of
the solution was maintained at 7-8. The solution was left to stand for 2 days for
hydrolysis before the solution was used for coating the samples.
Silane 2 was 5% by volume bis-[trimethoxysilylpropyl]urea. The 5% by volume
solution was prepared by mixing the silane and DI water in the ratio of 5 :95. The
pH was maintained at 6 for coating the alloys with this silane. A day was allowed
for hydrolysis before the samples were coated.
Silanes 3, 4, 5 and 6 were combinations of 5% by volume bis-[trimethoxysilylpro-
pyl] amine and vinyltriacetoxysilane in various ratios. Silanes 3 and 4 were
mixtures of bis-[trimethoxysilylpropyl] amine and vinyltriacetoxysilane in ratios
of 1.5:1 and 5 : 1 , respectively. For preparing the silane solutions in which the
mixtures of vinyltriacetoxysilane and bis-[trimethoxysilylpropyl] amine were used,
undiluted silanes were mixed in the desired ratios prior to diluting them in DI water.
The ratio of the undiluted silane mixture to DI water was 5:95. The pH was
maintained at 7-8. The solution was left to stand for about an hour for hydrolysis
before coatings were made from it. The effect of addition of Sol-Gel was also
investigated. Silanes 5 and 6 were these silanes where 10 ml and 20 ml of Sol-
Gel were added to a hydrolyzed mixture of bis-[trimethoxysilylpropyl]amine and
vinyltriacetoxysilane. In these silanes the ratio of bis-[trimethoxysilylpropyl]amine
and vinyltriacetoxysilane was 4 : 1 .
Silane 7 was a 5% by volume mixture of bis-[trimethoxysilylpropyl]amine and
bis-[triethoxysilylpropyl]tetrasulfide in a ratio of 1: 3. The bis-sulfur and the bis-
amino silanes were separately hydrolyzed in ethanol and the hydrolyzed silanes
were mixed in the desired ratio.
The natural pH of 5% by volume of water/ethanol bis-amino silane solution
was 10.8. The solution at such a high pH would gel in a few hours. To avoid this,
acetic acid was added. For optimum solution stability and anticorrosion efficiency
of the resultant silane coating a pH of 7 to 8.5 is desirable. The hydrolysis of this

silane is very fast.
A tris-silane, tris-[3-trimethoxysilylpropyl]cyanurate, was also studied. Two
systems were investigated. Silane 8 was made up of the tris-silane alone and Silane 9
was a combination of tris-silane with bis-[triethoxysilylpropyl]tetrasulfide in a ratio
of 3 : 1 . The tris-silane was hydrolyzed in ethanol. A 5% by volume silane solution
was prepared by mixing the silane, DI water and ethanol in the ratio of 5 :5 :90. The
pH of the solution was maintained at 6.5-7 for coating the samples. For coatings
which incorporated mixtures of tris-silane and bis-[triethoxysilylpropyl]tetrasulfide,
the silanes were hydrolyzed separately and then mixed in the desired ratio prior to
coating.
Silanes 10 and 11 were 5% by volume (water based) bis-[trimethoxysilylpropyl]-
amine and vinyltriacetoxysilane in combinations of 4: 1 with cerium nitrate and
potassium chromate in amounts of 200 ppm and 1000 ppm added, respectively. The
effect of the presence of a corrosion inhibitor in the silane coating in trace amounts
was investigated. Addition of inhibitors was done by adding the desired amount of
salt to the hydrolyzed silane solution and stirring the solution for 30 minutes until
the salt had completely dissolved in the silane solution.
22.2. Coating procedure. The samples to be coated were cleaned with an

alkaline cleaner AC 1055 (Brent International, Lake Bluff, IL, USA) at 65-75°C
for 3-5 min, after they had been cleaned with hexane, acetone and methanol in that
sequence in an ultrasonic bath for 5 min each. The samples were then rinsed with
water until they were completely wetted by it, and then blow dried with air. The
samples were then dipped into the silane solutions for 30 s, followed by drip drying
for 10 s. This was followed by curing the samples at 110°C for 45 min, which
was completed by further curing for 24 h at room temperature. This ensures further
cross-linking in the coating and avoids embrittlement of the film coating.
22.3. Corrosion fatigue life determination. A probabilistic model developed by

Perez was used [14]. The sample was loaded at a maximum stress of 65% of its
ultimate tensile strength for AA 7075-T6 and 70% of its ultimate tensile strength
for AA 2024-T3 alloy. The stress ratio for both alloys was 0.02 and the frequency
of loading was 10 Hz. The cyclic loading of the alloys was stopped after loading the
samples to one-third of the determined fatigue life of the alloy. This was followed by
a week of standard ASTM B117 salt spray exposure in absence of any stress. The
samples were then again loaded for another one-third of their fatigue life, which
was followed by another round of salt spray exposure for a week. The test was
concluded by cyclically loading the sample until failure.
This method of loading and alternating with stress cycles simulates the actual
environmental conditions to which the alloys are exposed in real life. The aircraft
where these alloys are predominantly used are cyclically loaded in operation
conditions at high altitudes where low temperatures prevailing lower the rate of
corrosion occurring. The enhanced rates of corrosion are encountered when the
^'^JHJ
' ,",]
/
A
'V"' J
^ ;--"
:"v: :-.';•
& £
<D <L>
3 3
W) fv^-' - OD
iK> ,
<S &%4 .1
a
<E
jjffeii'.. -a *M|
<D Vl
"£S !?" S 1
C/5
<D
^<U
(73
.»J
«*-
O
<+-
o
> --5>d
> -H
T3 -o
S -V.
j -
s-
• dii J3
"S1 •+-*
•mi <L>
C C
O O
c '•'*fM %
c
^
- J
3 3
j~^0.
1 -a
ctf
O
j
- ,»s
3
p J
Figure 2. Schematic diagram of fatigue loading of the samples for the probabilistic model of corrosion
fatigue life determination. Samples were loaded to 70% of ultimate tensile strength for A A 2024-T3
and to 65% of ultimate tensile strength for AA 7075-T6 alloy at R = 0.02 (R = stress ratio) and at
10 Hz. Duration of loading in the first two loadings was one-third the fatigue life of the sample. For
the last loading it was done until failure.
aircraft is grounded and in the hangars during non-service conditions i.e., in absence
of any cyclic loading.
For each silane investigated, three samples were loaded to fatigue failure. Figure 2
is a schematic diagram to represent the probabilistic corrosion fatigue model used
for testing.
2.2A. Effect of silane film on fatigue life. Tests were also performed for each
silane system investigated to determine the effect of the presence of a silane film on
the sample's fatigue life in the absence of any corrosive environment. For each of
the silanes, three samples were loaded at the same specifications of fatigue loading,
as specified above, until failure without any salt spray exposure,
2.2.5. Controls. Alkaline-cleaned blank samples and chromated samples were

used for determination of the effect of silane film on both fatigue life and also the
corrosion fatigue life. Chromating of the dog-bone specimens was done at Boeing
(Renton, WA, USA) using Alodine 1200 S.

Silanes have already been demonstrated to be excellent coupling agents that inhibit
the corrosion process in AA 2024-T3 and AA 7075-T6 alloys in the absence of
cyclic loading [1, 7-12].
3.1. The effect of the silane film on the number of cycles to failure in fatigue
forAA2024-T3
The effect of the presence of a silane film on the fatigue properties of AA 2024-T3
is shown in Fig. 3. The number of cycles to failure in fatigue of a blank control
sample of A A 2024-T3 was found to be 5.23 x 104 cycles. The chromated control
70000
80000
50000
3
40000 -H
30000
20000 -H
10000
# ^ # <fr* ^ & # # 4>* 4? <&* #* ** s** <f

J? J* ^ oP d? & <P x® & & >P xfc ,© J& ,
^ £& J& £& £& & & & ** <& £& •&* <£ fc* &
<*• J* ^ J> S v v £ v *> v £ <? &* >y
* cP o° ^ ^ ^ ^ ^ ^ ^ ^ <sr <fcr
Figure 3. Effect of silanes on fatigue life for AA 2024-T3 alloy.

samples resulted in a number of cycles to failure of 4.62 x 104 cycles. For reference
purposes, the numbers of cycles to failure in corrosion fatigue for the blank and
chromated samples of AA 2024-T3 are also shown in Fig. 3. Here, a reduction of
0.61 x 104 cycles due to chromating resulted in the number of cycles to failure in
fatigue for the AA 2024-T3 alloy.
Silanes 1 and 2 resulted in an increase in the number of cycles to failure in fatigue
of the alloy with respect to the blank control samples. Silane 1-coated samples had
a life of 0.80 x 104 cycles greater than that of a blank control and Silane 2-coated
samples had a life 0.16 x 104 cycles greater than that of a blank control sample.
The increase observed in the case of Silane 1, which is a bis-sulfur silane, is
attributed to the fact that the resulting silane coating is uniform, flexible and well
adherent to the entire surface of the alloy. In an AA 2024-T3 alloy 2.7% of the
total surface area is comprised of the intermetallic particles which disrupt the oxide
film which is formed on the surface of the alloy [15]. These particles are cathodic
with respect to the surrounding metal matrix they are imbedded in. The hydrolyzed
silanes have silanol groups which react with the oxide film to form Si-O-Me bonds
(Me = metal) which are overcoated by the hydrophobic siloxane network above it.
Since the oxide film is not continuous, a uniform silane film is not produced at
the interface in most of the silanes. It is also known that the common laboratory
atmosphere is a corrosive environment and reduces the number of cycles to failure
in fatigue loading of the alloys [16]. This means that if the fatigue loading of a
sample is conducted in vacuum the number of cycles to failure would be higher.
The failure of a silane-coated sample is due to the failure of the silane film which
exposes the surface of the alloy to the atmosphere which initiates corrosion of the
alloy and the nucleation of pits which are the classical sites for crack initiation
and propagation. The failure of the silane film could start at locations where the
adhesion to the metal surface is the weakest or where the silane film is the thinnest.
However, in the case of a bis-sulfur silane the sulfur in the silane reacts with the
Cu-rich intermetallic precipitates to form Cu^S, which results in a continuous film
at the interface [1]. Thus, the failure of the silane film is delayed and the observed
number of cycles to failure is increased.
The bis-sulfur silane film is relatively flexible. A silane-coated sample of
A A 2024-T3, which was deep-drawn in the Erickson test, survived 500 h of standard
ASTM B 117 salt spray test in previous investigations in our research group [10],
This means that the hydrophobic siloxane network formed above the interfacial
Si-O-Me bonds is flexible and can stretch to accommodate strains which develop
during drawing or cyclic loading of the substrate without failing. As such, the
samples coated with Silane 2 perform better under conditions of cyclic loading and
increase the number of cycles to failure when compared with the blank uncoated
samples.
Silanes 4, 7 and 9 showed reductions of 0.98 x 104, 0.84 x 104 and 0.93 x 104
cycles, respectively, to failure in fatigue as compared with number of cycles to
failure in corrosion fatigue of the blank control sample. However, these reductions
in the numbers of cycles to failure in fatigue for Silanes 4, 7 and 9 are less when
compared with the reduction observed due to chromating, which is 1.01 x 104 cycles.
Silanes 3, 5, 6, 8, 10 and 11 also showed decreases of 2.13 x 104, 1.54 x 10\
1.84 x 104, 1.21 x 104, 1.43 x 104 and 1.41 x 104 cycles, respectively, with respect
to the number of cycles to failure in fatigue of a blank control sample.
All the silanes investigated affected the number of cycles to failure in fatigue
for the AA 2024-T3 alloy. Silanes 1 and 2 were found to increase the number of
cycles to failure in fatigue of the alloy. These are the silanes which also had the
best performance in corrosion fatigue. Silanes 4 and 7 which had a lower degree
of reduction in the number of cycles to failure in corrosion fatigue for the alloy
were also found to have a lesser effect in reducing the number of cycles to failure in
fatigue loading. It is also seen that the presence of inhibitors in Silanes 10 and 11 has
a different effect in the presence of a corrosive environment. The reductions in the
number of cycles to failure in fatigue for these silanes, with respect to the number of
cycles to failure for a blank control sample, are found to be 1.43 x 104 and 1.41 x 104
cycles, respectively. However, when a corrosive environment is introduced, the
corresponding reductions in the number of cycles to failure are 0.79 x 104 and
0.66 x 104 cycles, respectively. This clearly shows that the inhibitors added to
the silanes leach out to the damaged sites of the silane coating during loading and
protect the alloy by retarding pit nucleation and growth during corrosion.
3.2. The effect of the silane film on the number of cycles to failure in corrosion
fatigue for AA 2024-T3
The effect of the presence of silane films on the corrosion fatigue resistance
properties of the specimens is shown in Fig. 4. The number of cycles to failure
in corrosion fatigue of a blank control sample of AA 2024-T3 was found to be
3.69 x 104 cycles. The chromated control samples resulted in a number of cycles
to failure of 2.68 x 104 cycles. Thus, a reduction of 1.01 x 104 cycles due to
chromating resulted in the number of cycles to failure in corrosion fatigue. For
reference purposes, the numbers of cycles to failure in fatigue for the blank and
chromated samples of AA 2024-T3 are also shown in Fig. 4.
Silanes 1 and 2 resulted in the highest number of cycles to failure in corrosion
fatigue of the alloy with respect to the blank control sample. Silane-1-coated sample
had a life of 1.30 x 104 cycles greater than that of a blank control, and the Silane-
2-coated sample had a life 0.77 x 104 cycles greater than that of a blank control
sample. These increases can be explained with the model suggested above. The
increases in the number of cycles to failure as compared with the blank uncoated
sample are greater because a corrosive environment is present which further reduces
the life of a blank uncoated sample.
Silanes 4 and 7 showed, respectively, reductions of 0.39 x 104 cycles and
0.25 x 104 cycles to failure in corrosion fatigue as compared to the number of cycles
to failure in corrosion fatigue of blank control sample. However, these reductions
in the number of cycles to failure in corrosion fatigue for Silanes 4 and 7 when
128 A. Seth and W J. Van Ooij
Figure 4. Effect of silanes on corrosion fatigue life for AA 2024-T3 alloy.
compared with the reduction observed due to chromating, which is 0.61 x 104 cycles,
are less.
Silanes 3, 5, 6, 8, 9, 10 and 11 also showed decreases of 1.03 x 104, 0.79 x 104,
1.12 x 104, 1.09 x 10\ 1.20 x 104, 0.79 x 104 and 0.66 x 104 cycles, respectively,
with respect to the number of cycles to failure in corrosion fatigue of a blank control
sample.
It is clearly seen that all the silanes investigated have an effect on the number of
cycles to failure in corrosion fatigue for the AA 2024-T3 alloy, which is greater than
or comparable with the reduction of 1.01 x 104 cycles, i.e., the number of cycles to
failure in corrosion fatigue when the AA 2024-T3 alloy is chromated.
3.3. The effect of the silane film on the number of cycles to failure in fatigue
forAA7075-T6
The effect of the presence of silane films on the fatigue properties of A A 7075-T6 is
shown in Fig. 5. The number of cycles to failure in fatigue of a blank control sample
of AA 7075-T6 was found to be 3.80 x 104 cycles. The chromated control samples
Figure 5. Effect of silanes on fatigue life for AA 7075-T6 alloy.

resulted in the number of cycles to failure of 3.00 x 104 cycles. Thus, a reduction
of 0.80 x 104 cycles due to chromating resulted in the number of cycles to failure in
fatigue. For reference purposes, the numbers of cycles to failure in corrosion fatigue
for the blank and chromated samples of A A 7075-T6 are also shown in Fig. 5.
All the silanes resulted in a decrease in the number of cycles to failure in fatigue
for the AA 7075-T6 alloy. However, for Silane 2, a reduction of only 0.52 x 104
cycles with respect to the number of cycles to failure in fatigue of a blank control
was observed. This is less than the reduction of 0.80 x 104 cycles which resulted
due to chromating. All other silanes resulted in a larger reduction in the number
of cycles to failure in fatigue with respect to the reduction of 0.80 x 104 cycles
observed due to chromating.
The effect of leaching out of the corrosion inhibitors to mimic the "self healing"
effect was observed in the case of AA 7075-T6 alloy. Silanes 10 and 11 show,
respectively, reductions of 2.33 x 104 and 1.80 x 104 cycles with respect to the
number of cycles to failure in fatigue in the absence of a corrosive environment for
the blank control sample. However, the corresponding reductions are only 0.13 x 104
and 0.08 x 104 cycles for Silanes 10 and 11, respectively, when corrosion fatigue
is considered. This corroborates the fact that the "self healing" property of the
chromate conversion coatings can be incorporated in silane coatings and this results
in enhanced corrosion fatigue resistance of the silane coatings.
3.4. The effect of the silane film on the number of cycles to failure in corrosion
fatigue for AA 7075-T6
The effect of the presence of silane films on the corrosion fatigue resistance
properties of AA 7075-T6 is shown in Fig. 6. The number of cycles to failure
in corrosion fatigue of a blank control sample of AA 7075-T6 was found to be
2.18 x 104 cycles. The chromated control sample resulted in the number of cycles
to failure of 1.81 x 104 cycles. Thus, a reduction of 0.36 x 104 cycles due to
chromating was observed for the number of cycles to failure in corrosion fatigue.
For reference purposes, the number of cycles to failure in fatigue for the blank and
chromated samples of AA 7075-T6 are also shown in Fig. 6.
Silanes 5 and 7 resulted in an increase in the number of cycles to failure in cor-
rosion fatigue of the alloy with respect to the blank control sample. Silane-5-coated
sample had a life of 0.05 x 104 cycles greater than that of a blank control and the
Silane-7-coated sample had a life 0.38 x 104 cycles greater than that of the blank
control sample.
Silanes 1, 2, 3, 6, 10 and 11 showed, respectively, reductions of 0.01 x 104,
0.10 x 104, 0.26 x 104, 0.20 x 104, 0.13 x 104 and 0.08 x 104 cycles to failure
in corrosion fatigue as compared with the number of cycles to failure in corrosion
fatigue of the blank control sample. However, these reductions in the numbers of
cycles to failure in corrosion fatigue for Silanes 1, 2, 3, 6 and 10, when compared
with reduction observed due to chromating, which is 0.36 x 104 cycles, are less.
50000
45000
35000 -H
30000
Figure 6. Effect of silanes on corrosion fatigue life for AA 7075-T6 alloy.

Silanes 4, 8 and 9 also showed decreases of 0.26 x 104, 0.39 x 104 and 0.50 x 104
cycles, respectively, with respect to the number of cycles to failure in corrosion
fatigue of a blank control sample.
Again, it is clearly seen that all the silanes investigated result in a change of the
number of cycles to failure in corrosion fatigue for the AA 7075-T6 alloy, which is
greater than or comparable to the reduction of 0.36x 104 cycles, which is the number
of cycles to failure in corrosion fatigue when the AA 7075-T6 alloy is chromated.
4. SUMMARY AND CONCLUSIONS

1. Chromating was found to reduce the number of cycles to failure in fatigue and
corrosion fatigue for both AA 2024-T3 and AA 7075-T6 alloys.
2. Silanes 1 and 2 resulted in coatings for the AA 2024-T3 alloy that had better
performance as compared to both the blank and chromated controls for AA 2024-
T3 alloy in corrosion fatigue. Silanes 4 and 7 resulted in coatings for the AA
2024-T3 alloy that had better performance as compared to the chromated control
for the AA 2024-T3 alloy in corrosion fatigue. Silanes 3, 5, 6, 8, 9, 10 and 11
resulted in coatings for the AA 2024-T3 alloy that had comparable performance
as compared to corrosion fatigue of the chromated control alloy.
3. Silanes 5 and 7 resulted in coatings for the A A 7075-T6 alloy that had better
performance as compared to both the blank and chromated control for AA 7075-
T6 alloy in corrosion fatigue. Silanes 1, 2, 3, 6, 10 and 11 resulted in coatings for
the AA 7075-T6 alloy that had better performance as compared to the chromated
control for AA 7075-T6 alloy in corrosion fatigue. Silanes 4, 8 and 9 resulted
in coatings for the AA 7075-T6 alloy that had comparable performance as
compared to the chromated control specimen in corrosion fatigue.
4. Silanes 1 and 2 resulted in coatings for the AA 2024-T3 alloy that increased
the number of cycles to failure in fatigue as compared to both the blank and
chromated controls for AA 2024-T3 alloy in fatigue. Silanes 3, 4, 5, 6, 7, 8, 9,
10 and 11 resulted in coatings for A A 2024-T3 alloy that reduced the number of
cycles to failure in fatigue as compared to chromated control alloy in fatigue.
5. All the silanes reduced the number of cycles to failure in fatigue for the A A
7075-T6 alloy as compared to the blank control of AA 7075-T6 alloy in fatigue.
However, Silane 2 resulted in a coating that had a greater number of cycles to
failure as compared to the chromated control of the AA 7075-T6 alloy.
6. The failure of a silane-coated sample is believed to occur due to failure of the film
which exposes the surface of the metal and facilitates the formation of corrosion
cells around the intermetallic precipitates. The silane which results in a uniform
and flexible film with a specific affinity for copper-bearing intermetallics, such
as the Silane 1, results in a coating for the AA 2024-T3 alloy which increases
the resistance of the alloy to corrosion fatigue and fatigue.
The above results indicate that, depending on the silane chosen, a better or com-
parable resistance to corrosion fatigue as compared with the chromate conversion
coatings can be obtained. However, because of the statistical nature of these tests,
it is essential to treat the results as trends. Further tests will be performed for vali-
dating the veracity of the observations. The effect of silane films under primers and
paints also needs to be investigated.
Acknowledgements
The authors are indebted to Dr. Joseph Osborne, The Boeing Company (Renton,
WA, USA), for the assistance rendered in chromating the dog-bone specimens,
and Dr. Rigberto Perez, The Boeing Company (St. Louis, MO, USA), for the
guidance regarding the probabilistic corrosion fatigue testing procedure, Chemat
Technologies (Northridge, CA, USA) is acknowledged for the use of their facilities
for salt spray.
REFERENCES
1. D. Zhu, Ph.D. Dissertation, University of Cincinnati, Cincinnati, OH (2002).
2. F. Viana, A. M. P. Pinto, H. M. C. Santos and A. B. Lopes, J. Mater. Process. Technol 92, 54
(1999).
3. V. Subramanian and W. J. van Ooij, in: Silanes and Other Coupling Agents, Vol. 2, K. L. Mittal
(Ed.), p. 159. VSP, Utrecht (2000).
4. T. K. Goswami and D. W. Hoeppner, in: Structural Integrity in Aging Aircraft, AD-Vol. 47,
p. 129. ASME, New York, NY (1995).
5. S. Mahadevan and P. Shi, Progress in Structural Engineering and Materials 3 (2), 188 (2001).
6. P. G. Sheasby and R. Pinner, The Surface Treatment and Finishing of Aluminum and its Alloy,
Vol. 1, p. 229. ASM International, Materials Park, OH (2001).
7. W. J. van Ooij and T. F. Child, CHEMTECH 28, 26 (1998).
8. V Subramanian, Ph.D. Dissertation, University of Cincinnati, Cincinnati, OH (1999).
9. W. J. van Ooij, D. Zhu, G. P. Sundararajan, S. K. Jayaseelan, Y. Fu and N. Teredesai, Surf. Eng.
16, 86 (2000).
10. G. P. Sundararajan, M.S. Thesis. University of Cincinnati, Department of Materials Science and
Engineering, Cincinnati, OH (2000).
12. W. J. van Ooij, D. Zhu, V. Palanivel, J. A. Lamar and M. Stacey, Silicon, submitted.
13. E. P. Plueddemann, Silane Coupling agents, 2nd edn. Plenum Press, New York, NY (1991).
14. R. Perez, Presented at 3rd Joint FAA/DoD/NASA Conference on Aging Aircraft, Albuquerque,
MN(1999).
15. R. G. Buchheit, J. Electrochem. Soc. 142, 3994 (1995).
16. D. W. Hoeppner, ASTM Special Technical Publication 675, 841 (1979).
Ed. K. L. Mittal
© VSP 2004
Modified silane coatings as an alternative to chromates

for corrosion protection of aluminum alloys
VIGNESH PALANIVEL and WIM J. VAN OOIJ *

OH 45211-0012, USA
Abstract—Silane coatings on aluminum aircraft alloy AA2024-T3 have been proposed as an

alternative to toxic chromate conversion coatings for surface pretreatment. The corrosion protection
by these silanes is comparable to chromates and in certain cases silanes out-perform them. However,
the silane coatings are not visible and do not have the "self-healing" effect that chromates are known
for. In this paper we show that silane films can be colored and that a self-healing effect can be
achieved by adding optimum amounts of leachable corrosion inhibitors to the silane film which can
be released at a controlled rate. We have studied the corrosion protection properties of the colored
silane films and have shown that the corrosion inhibition property of the silane film is not hindered
by the colorant. We also show that loading silanes with sheared nanoparticles enhances the corrosion
protection performance and the mechanical properties of the film. Two different silane systems are
described in this paper, viz., a solvent-based system and a completely water-based system.
Keywords: Silane; corrosion inhibitors; self-healing; chromate replacement; adhesion.
1. INTRODUCTION
Chromates are among the most common substances used as corrosion inhibitors in
surface pretreatment of aircraft aluminum alloys. However, these chromates are
considered toxic and carcinogenic causing serious environmental hazards [1]. This
paper discusses the potential of modified silanes as an alternative to the conventional
chromating process.
Silanes are organofunctional hybrid organic/inorganic compounds used primarily
as coupling agents for adhesion between organic and inorganic materials. These
functional silanes have hydrolyzable groups such as methoxy or ethoxy and an
organofunctional group. The silane structure is typically X3Si(CH2)„Y, where X
is the hydrolyzable group and Y the organofunctional group [1-8].
136 V. Palanivel and W. J. van Ooij
Silanes were first used as efficient coupling agents on glass reinforcements

for polymeric bulk [9]. For decades, silanes were considered to be primarily
coupling agents and their potential to function as corrosion inhibitors had not been
explored in any detail. Van Ooij and co-workers [2-5] have studied the corrosion
inhibition performance of silanes extensively in recent years. Their initial research
was devoted to mono-silanes with one silicon atom in each molecule. Later it
was found that bis-type silanes, such as the bis-(3-triethoxysilylpropyl)tetrasulfide,
which have two silicon atoms in their molecule, provided better corrosion inhibition
performance on a wide range of metals and alloys as compared with mono-silanes.
The improved corrosion inhibition performance was attributed to the fact that the
bis-type silanes contained an extra silicon atom which could help build up the
silane film by forming siloxane groups (Si-O-Si) at increased thickness, whereas
the mono-silanes are unable to further cross-link and build up the silane film and
at the same time form a dense Si-O-Me interface [10, 11]. The bis-type silanes,
especially the bis-sulfur silane, provide excellent adhesion of coatings to metal
substrates and better corrosion protection under severe corrosive conditions such
as seawater immersion tests. The corrosion behavior of these bis-silanes was found
to be comparable to chromates [2-5].
Silane treatment involves two important stages: (1) hydrolysis of the silane in a
water/alcohol mixture in order to form active silanol groups (Si-OH) and (2) cross-
linking of the silane film after surface treatment by curing. When dipping a metal
into a dilute silane solution (e.g., 2-5 vol%) for a few seconds, silanols (Si-OH) in
the silane solution adsorb spontaneously onto the metal surface, as shown in Fig. 1.
Upon drying, the condensation reactions occur in the silane/metal interfacial region:
between Si-OH groups from the silane solution and the metal hydroxyls (Al-
OH) from the metal surface hydroxide, forming covalent metallo-siloxane bonds
(Si-O-Al) [9, 10, 12, 13].
2 >•
,cW&-
cyclic
siloxane
H-bond
"-&T~0
* >$
ox ô- L<TV*
«K <& x / / \ \
o
IV I
55
O "A.
X
/
^;s ^ . •••"
/ ?
ô-sz-cf
Si
§f \ ~ ' I / Si'
H
/ \
<X>V
+4-
<
+ -O,—u
o ..P—o
1-
linear
siloxane
Me Me Me M i r f - O J t o ^ , , ^ , 0 |i d e Me Me
Figure 1. Schematic of bonding mechanism between silane molecules and metal surface.
Modified silane coatings for corrosion protection of Al alloys 137
The hydrolysis reaction of a bis-type silane in a water/alcohol mixture takes place

in sequential steps as
-Si-(OC 2 H 5 ) 3 + 3H 2 0 ^ 3C2H5OH + -Si-(OH) 3
Following the hydrolysis, the silane film, when applied to the metal substrate,
undergoes condensation reaction to form siloxane groups.
Si-OH + Si-OH — • Si-O-Si + H 2 0

Si-OH + Al-OH —-> Si-O-Al + H 2 0
The corrosion inhibition efficiency is assumed to be due to the hydrophobicity of the

silane film [1-5, 7]. Water does not easily penetrate hydrophobic films. The cross-
linking process can be accelerated by increasing the curing time or temperature.
The alcohol-based silane-surface treatments that were developed in our laboratory
were the first-generation silanes [3, 5]. Second-generation water-based silanes have
now been developed. These silanes are completely solvent-free and are made up of
a mixture of bis-amino silane and a vinylacetoxysilane. The silane solution is made
up of 95% DI water and 5% silane. The corrosion inhibition performance of this
novel water-based silane mixture is comparable to some alcohol-based hydrophobic
bis-type silanes.
After having developed silane treatments to effectively replace chromate coatings,
the need arose to modify these silane films further because chromates have certain
distinct advantages over silanes. Chromate films are easily recognizable, i.e., they
can be detected by their color as an indication that the coating exists on the metal,
e.g., on coil paint lines. Chromates also possess a "self-healing" capability, i.e., the
ability to slowly leach chromate ions into an aqueous medium to unprotected areas
of the metal and protect them from further corrosion. Silanes are clear liquids and
when applied to a metal are not visible to the naked eye. Also, silane films do not
possess the self-healing ability when films are locally scratched, as they are barrier
coatings only and have no electrochemical activity. In order to replace chromates
these factors should be considered. We have attempted to give the silane film a
color by adding suitable colorants, dyes and nanoparticles. The effect was tested
by DC polarization measurements. Further, organic and inorganic inhibitors were
added to the silane film to (1) study whether or not the addition further increased
the corrosion inhibition performance of the silane film and (2) to determine whether
these inhibitors had the capability to leach out and self-heal.
In addition to the anticorrosive efficiency, another major concern in the use of
silane films is the mechanical properties of such films. In service, these silane films
on metals should be capable of resisting mechanical damages by impact, scratch and
wear. Therefore, one of our current research subjects is to improve the mechanical
properties of silane films by loading them with nanoparticles.
2. EXPERIMENTAL
2.1. Materials and methods
2.LI. Silanes. The silanes bis-(3-triethoxysilylpropyl)tetrasulfide, bis-(trime-
thoxysilylpropyl)amine were obtained from OSi Specialities, and vinlytriacetoxysl-
lane (VTAS) was obtained from Gelest in pure form. The silanes were in unhy-
drolyzed form and in order to hydrolyze them DI water and/or alcohol was added.
The solvent-based bis-sulfur silane solution was prepared by mixing 5 vol% silane
with 90 vol% ethanol and 5 vol% DI water. The solution was hydrolyzed for 2 days
at the natural pH of 6.5. The water-based silane solution was prepared by mixing
the bis-amino silane and VTAS in two different ratios of 1: 1 and 4 : 1 , i.e., 4 parts
of bis-amino and one part of VTAS by volume. About 5 vol% of this mixture was
hydrolyzed with 95 vol% of DI water. The solution was hydrolyzed for one day at
the natural pH of 4.
2.1.2. Metals/alloys. The substrate samples for the present study were AA2024-
T3 alloy (10 cm x 15 cm x 0.8 cm) obtained from ACT (Hillsdale, MI, USA).
The rolled sheets obtained were cleaned with an alkaline cleaner (AC 1055®,
Brent America, Lake Bluff, IL, USA). The metal substrates were consecutively
ultrasonically cleaned in hexane and ethanol for 10 min each. The cleaned panels
were further alkaline cleaned with 7.5 vol% of the AC 1055 aqueous solution
at 60-70° C for 5-7 min. The cleaned panels were immediately rinsed with DI water
until a water break-free surface was obtained. The panels were then blow-dried with
compressed air.
2.1.3. Silane treatment of alloys. The metals/alloys were cleaned as described

above. The silane films were applied by immersing the cleaned metal panel into
the prepared silane solution. The immersion time was between 15 and 30 s. After
immersion the panels were baked at 100°C for 10 min for the silane with the colorant
and for 20 h for the silane with the inhibitors. This was done to crosslink the silane
film more extensively, so as to lower the leaching rate.
2.1 A. Incorporating color in silane films. Organic dyes were obtained from
BASF Canada. Two different organic dyes were chosen: a basonyl yellow NB 122
dye and a basonyl red 482 (xanthene) powder dye. The bis-sulfur silane solution
was hydrolyzed as mentioned above and the dyes were added in low quantities. The
silane solution was stirred for 10 minutes for the dye to dissolve completely in the
solution.
2.1.5. Incorporating inhibitors in silane films. Organic and inorganic inhibitors

were selected based on their inhibiting properties on aluminum alloys [14, 15].
Cerium nitrate, tolyltriazole, benzotriazole were obtained from Sigma Aldrich
(St. Louis, MO, USA). The solvent-based bis-sulfur silane solution with an inhibitor
was prepared by mixing 0.1 g of the inhibitor with 90 vol% ethanol and the mixture
Modified silane coatings for corrosion protection ofAl alloys 139
was stirred well for 15 min. The ethanol with the inhibitor was added to 5 vol%
of the bis-sulfur silane and stirred for 5 min. The hydrolysis was then carried out
by adding 5 vol% of DI water to the silane with ethanol. The silane solution was
allowed to stand for 2 days for hydrolysis to occur. The water-based silane with
an inhibitor was prepared by adding 1000 ppm of inhibitor to 100 vol% water, the
solution was stirred and optionally heated for complete solubility of the inhibitor in
DI water. The DI water with the inhibitor (95 vol%) was added to 5 vol% silane
mixture of bis-amino/VTAS and stirred for 5 minutes. Hydrolysis was carried out
for one day before application.
2.1.6. Incorporating silica nanoparticles. Fumed silica nanoparticles of l-/xm

size (Cabot, Tuscola, IL, USA) were added into the bis-sulfur silane solution
according to the following two-step procedure. First, different amounts of silica
nanoparticles, i.e., 0.01%, 0.03%, 0.04% and 0.1% (by weight) were mixed with DI
water using a high-speed blender until uniform silica colloidal suspensions were
obtained. The mixing time was about 20 min. 5 parts of the silica colloidal
suspensions were then added into the silane/ethanol mixture at a mixing ratio of
5 :90 (v/v). The concentrations of silica nanoparticles in the silane solutions then
became 5 ppm, 15 ppm, 20 ppm and 50 ppm.
2.2. Test procedures

2.2.1. Electrochemical polarization tests. These were carried out in aerated
0.6 M NaCl solution at pH 6.5. The silane-treated panels were pre-immersed in
the electrolyte for 1-2 h before data acquisition, in order to achieve a steady state.
The same conditions were used for all samples tested. The bare AA2024-T3 panels
were tested immediately after exposure to the electrolyte. A commercial Saturated
Calomel Electrode (SCE) and a platinum mesh were used as the reference and
counter electrodes, respectively. The exposed area was 0.78 cm2. On average, three
replicate samples were tested for each condition. The data were recorded over the
range of Econ ± 0.25 V/SCE (where Ecorr is the equilibrium corrosion potential of
the tested sample). The scan rate was 1 mV/s.
2.2.2. Electrochemical Impedance Spectroscopy (EIS). EIS was employed to

evaluate the performance of the silane-treated AA2024-T3 systems in a 0.6 M
NaCl solution (pH 6.5). The EIS measurements were carried out using an SR810
frequency response analyzer connected to a Gamry CMS 100 potentiostat. The
measured frequency range was from 10~2 to 105 Hz, with AC excitation amplitude
of 10 mV. SCE was used as the reference electrode and was coupled to a graphite
counter electrode. The surface area exposed to the electrolyte was 5.16 cm2.
2.2.3. Crosscut adhesion tape test (ASTM D3359). This test was used to assess
the adhesion of paint coatings on metallic substrates by applying and removing
pressure. A cutting tool device with a cutting edge angle between 15 and 30°
140 V Palanivel and W. J. van Ooij
was chosen, which made several cuts at once. After making two such cuts at 90°
the grid area was brushed and a 2.5-cm wide semi-transparent pressure sensitive
tape was placed over the grid. The tape was rubbed with an eraser. After 30 s of
application, the tape was removed rapidly and the grid inspected according to the
ASTM standards.
2.2.4. Mechanical tests. Fifteen indentation experiments were performed on bis-

sulfur silane-treated samples using an MTS Nanoindenter XP and the Continu-
ous Stiffness Measurement (CSM) technique, the details of which are given in
Ref. [16]. The instrument was enclosed in an environmental chamber maintained
at 25°C located in the laboratory where the ambient temperature was controlled to
within zbl°C. The samples were placed in the chamber for approximately 12 h prior
to testing.
2.2.5. Ellipsometry thickness measurements. Ellipsometry experiments were

performed over a wide range of wavelengths of 300-800 nm with angles of
incidence at 60, 65, 70 and 75 degrees. The ^ and the A values were calculated and
a computer was used for data acquisition. Further information about ellipsometry
can be found in Ref. [17].
2.2.6. ICP-MS. It is a versatile, rapid and precise analytical technique that

provides a high quality multi-element and isotopic analysis package. This technique
was used to detect leaching of inhibitors from the silane film when immersed in a
DI water and/or salt solution. The detection limit for most elements is in the sub-
parts per billion (ppb) range. The instrument used was a VG plasma system which
is equipped with a number of different sample introduction techniques. The sample
is introduced into a radio-frequency (RF) induced plasma in the form of a solution,
vapor or solid. A quadrupole mass spectrometer permits the detection of ions at
each mass in rapid sequence, allowing signals of individual isotopes of an element
to be scanned.

3.1. Colored si lanes
Silanes modified with dyes when applied on the metal substrate were visible to the
naked eye. Figure 2 shows the pictures of colored silane and silane without the
colorant on AA2024-T3. It can be seen that the silanes with the yellow and the red
dyes are clearly visible to the naked eye indicating the presence and the uniformity
of the silane film. The idea behind the dye is to make the water-soluble dye insoluble
by heating the silane with the dye to higher temperatures after deposition. Figure 3
shows the DC polarization curves for AA2024-T3 panels treated with silane with
and without the yellow colorant. A blank panel was also tested for comparison
-0.400 -p
-0.500 I
-0.600 i
Potential
(V)
-0.700 4-
-0.800 I
-0.900 |
-1.000 4 i 1 1 1 1 1 1
-8.0 -7.0 -6.0 -5.0 -4.0 -3.0 -2.0 -1.0
Figure 3. DC Polarization curves of AA2024-T3 treated with and without yellow dye in bis-sulfur
silane films in 0.6 M NaCl. (1) Blank AA2024-T3; (2) bis-sulfur treated; (3) bis-sulfur with 10 ppm
yellow dye; (4) bis-sulfur with 20 ppm yellow dye.
to determine the corrosion inhibition properties of the silane film. All the silane-
treated panels were immersed in the 0.6 M NaCl electrolyte for 2 h before testing
to achieve a steady state. It can be clearly seen that the silane treatment reduces
both the anodic and cathodic current densities by at least one decade. Curves 3
and 4 show that addition of organic colorants does not affect the performance of the
silane film. Curves 3 and 4 overlap with curve 2, which is the silane-treated sample
without the colorant. Addition of these colorants does not affect the electrochemical
properties of the silane film.
Figure 4 shows the immersion test results of AA2024-T3 treated with bis-sulfur
silane and with silane loaded with colorants. The immersion time in the 3.5% NaCl
solution was 1 week at room temperature. The results show that the corrosion
inhibition performance of the silane with the colorant is equal to that without it.
Also, the silane with the colorant does not lose its color, i.e., it does not leach out,
indicating that the soluble colorant becomes insoluble upon curing the silane film
on the metal substrate. Further, it also indicates that the performance of the film is
not lost when the colorant is added.
5.2. Paint adhesion due to colored silanes

Surface pretreatment of aluminum aircraft alloys is performed to provide corrosion
protection, as well as to promote a strong bond between the topcoat and the alloy.
So the adhesion of paints to the silane-pretreated alloy is of great importance. As we
have already mentioned, silanes are effective coupling agents and so are conducive
(a)
Figure 4. Immersion test results (7 days 3.5% NaCl) with bis-sulfur silane with (10 ppm) (b) and
without dye (a) on AA2024-T3 alloy.
to paint adhesion. Colored silanes solve the problem of detecting the silane film but
adhesion due to the colored silane was lost completely when the polyester-painted
colored silane (yellow) coated alloy was placed in a humidity chamber for one
week at 90% relative humidity. The paint along with the colored silane came off
when the crosscut adhesion test was carried out as shown in Fig. 5b and 5c, both
exposed to humidity. The paint virtually came off from the grid when the tape was
removed. Figure 5a shows the good adhesion due to the colored silane before the
humidity test, the paint in the grid remains intact. This loss of adhesion may be
attributed to the fact that the polyester paint applied is a solvent based paint and the
yellow dye is leached out from the silane film by the solvent. This was confirmed
by immersing a yellow colored silane sample in an alcohol solvent, where the color
leached out completely. This led us to explore new dyes and colorants, which would
be water-soluble but become insoluble on curing. A red xanthene dye was added
to the bis-sulfur silane and adhesion of polyester paint to it was studied in the same
way as for the yellow organic dye. Figure 6 shows the pictures of painted colored
silanes. Figure 6a shows the result for the painted colored silane without exposure to
humidity and Fig. 6b indicates the painted colored silane after exposure to humidity
(b) (c)
Figure 5. Polyester paint adhesion test results on colored (yellow) silanes before (a) and after (b, c)
exposure to humidity for one week on AA2024-T3 alloy.
(a) (b)
Figure 6. Paint adhesion test results on colored (red) silanes before (a) and after (b) exposure to
humidity for one week on AA2024-T3 alloy.
for 1 week. The paint does not peel off as in the case of the yellow dye in the grid,
when the tape is removed, indicating a considerably more stable system. The reason
for this may be the structure of the xanthene dye which has two hydroxyl groups.
These groups may react with the silanol in the silane to form a complex which when
cured can become part of the hydrophobic silane film (Fig. 7).
33. Effect of addition of silica nanoparticles: electrochemical tests on AA2024-T3

in a 0.6MNaCl solution (pH 6.5)
Figure 8 shows the DC polarization curves of AA2024-T3 treated with and without
bis-sulfur silane. All the silane-treated panels were immersed in the electrolyte for
O-Si
Si-OH + ) -SÔH
Curing
Figure 7. Possible reaction mechanism between xanthene dye and silanol groups.
Potential
(V)
1. Untreated AA2024-T3
2. Bis-sulfur silane treated
3. Bis-sulfur silane (5 ppm of silica) treated
4. Bis-sulfur silane (15 ppm of silica) treated
Log Current Density (A/cm 2 )
Figure 8. DC polarization curves of AA 2024-T3 treated with and without nanoparticle-fllled bis-
sulfur silane films in 0.6 M NaCl.
24 h before testing to reach a steady state, while the untreated panel was tested
without delay. Curve 1 represents the untreated AA2024-T3; while curves 2 to 4
are for the bis-sulfur silane-treated AA2024-T3 panels but loaded with different
amounts of silica nanoparticles in the silane films. It is seen that after being treated
with the bis-sulfur silane (curve 2), both anodic and cathodic current densities have
been reduced by at least 1 decade. The Ecorr corresponding to the film from 5 ppm
silica-containing silane solution shifts significantly in the cathodic direction, from
the original value of -0.6 V/SCE to -1.0 V/SCE (curve 3). This shift indicates that
the incorporation of a small amount of silica particles alters the cathodic kinetics
on the alloy surface. When increasing the amount of silica in the silane solution
to 15 ppm, such cathodic shift disappears for the corresponding silane film, with
the £ corr shifting back to around -0.6 V/SCE (curve 4). Another pronounced
feature in Fig. 8 is that two cathodic regions (reactions) are shown in curves 3
and 4, but not in curves 1 and 2. This further suggests that a small amount of silica
particles in the silane film changes the mechanism of the cathodic reactions on the
alloy surface. Instead of one cathodic reaction, two cathodic reactions reflected by
7.00 -i 3
2V^22SSgg%
6.00- 1 " " ^ ^ •Sfe
O 5.00 - 4 **AtotoAa
*^fe w£*
&fr
-o
**«*
o 4.00 -
° 3.00- 1. without silica

2. Sppmsilica
3.15 ppmsilica
2.00- 4. 50 ppmsilica
1.00 - .... t ___^

-2.50 -1.50 -0.50 0.50 1.50 2.50 3.50 4.50
Log Freq (Hz)
Figure 9. Impedance plots of bis-sulfur si lane treated AA2024-T3 systems loaded with different
amounts of silica nanoparticles.
the two cathodic regions dominate over the applied cathodic voltages. The proposed
mechanism is explained in Section 3.6.
Figure 9 compares the EIS behavior of silica-loaded bis-sulfur silane film de-
posited on AA2024-T3. The results were obtained in a 0.6 M NaCl solution at
pH 6.5. All panels were immersed in the electrolyte for 24 h before data acquisi-
tion. A similar trend is observed here: the low-frequency impedance values (Zif) of
the systems increase with the increase of the silica content until 15 ppm (curves 2
and 3 in Fig. 9) compared with the unloaded panel (curve 1 in Fig. 9). The Z\f value,
however, drops sharply for the film obtained from 50 ppm silica-containing silane
solution (curve 4 in Fig. 9). This, again, confirms that a large amount of silica
particles is not required from the point of view of a good corrosion inhibition per-
formance of the bis-sulfur silane film. In addition, a two-time constant behavior is
clearly shown in curve 4, corresponding to 50 ppm silica. The time constant at high
frequencies is for the nano-structured silane film, while the one at low frequencies
may be attributed to a double layer formed at the silane/metal interface. The forma-
tion of the double layer in the 50 ppm silica-loaded silane system indicates that an
excess of silica particles in the film has a negative effect on the interfacial adhesion,
leading to a premature film delamination from the substrate.
On the basis of the above results, it can be generally concluded that a small
amount of silica nanoparticles (e.g., films obtained from the silane solution with
silica ^15 ppm) does improve the corrosion inhibition performance of the bis-sulfur
silane film on AA2024-T3. However, such improvement diminishes on further
increasing the silica amount in the film (e.g., silica in the solution >15 ppm).
Moreover, a large amount of silica particles in the film even degrades the corrosion
inhibition performance of the silane film (e.g., >50 ppm), as confirmed in both DC
and EIS tests.
3.4. Thickness dependence of corrosion inhibition performance of

nanoparticle-filled bis-sulfur silane films on AA2024-T3
Figure 10 displays the bis-sulfur silane film thickness and the corresponding icon
values (corrosion current measured by Tafel plots) as a function of silica content
in the silane solution. The film thicknesses were obtained from ellipsometry
measurements using mirror-like stainless steel substrates. It is clear in Fig. 10
that the film thickness increases with increasing silica amount in the corresponding
silane solution until 15 ppm, and remains constant afterwards. This indicates that
the film is indeed thickened by incorporating silica particles. However, an additional
amount of silica does not thicken the silane film further.
3.5. Mechanical properties of nanopartilce-filled bis-sulfur silane films

onAA2024-T3
The values of hardness and modulus of elasticity of the silane films loaded with and
without silica nanoparticles are shown in Figs 11 and 12, respectively. The same
trend is seen for both hardness and modulus values. The silane films obtained from
5 ppm, 15 ppm and 50 ppm silica-containing silane solutions show somewhat higher
hardness as compared to the film without silica particles. This confirms that the
silane film can be hardened by loading with silica nanoparticles as expected. As the
silica content is increased from 5 to 15 ppm (Fig. 11) the hardness value increases,
however for the 50 ppm silica in the silane solution the hardness value does not
increase much indicating that the film has become porous and further addition of
silica will not enhance the hardness of the film.
700 1.00E+00
600 1.00E-01
Icorr ( A / C m 2 )
Thickness
1.00E-02
( nm ) 500
1.00E-03
400 H
^ 1.00E-04
Icorr of untreated AA 2024-T3: 2.86E-05 A/cm2
300
1.00E-05
200
1.00E-06
100 H 1.00E-07
1.00E-08
10 20 30 40 50
Silica content in the silane solution (ppm)
Figure 10. Film thicknesses and / corr values of the bis-sulfur silane film as a function of silica content
in the bis-sulfur silane solution.
1.4
-—Oppm
LO o SOppm
1.2 + A 15ppm
o 5ppm
-J >,„-! | -f-
100 200 300 400 500 600
Displacement Into Surface (nm)
Figure 11. Hardness of bis-sulfur silane films loaded with and without silica nano-particles measured
by nanoindentation.
40 r
-Without silica
-SOppm silica
J—i—i—i—i—i—I—u
100 200 300 400 500

Displacement Into Surface (nm)
Figure 12. Modulus of elasticity of bis-sulfur silane films loaded with and without silica nanoparticles
measured by nanoindentation.
3.6. Proposed corrosion inhibition mechanism
The bis-sulfur silane film on loading with silica nanoparticles tends to improve the
corrosion inhibition and mechanical properties up to a certain extent, beyond which
further loading of silica nanoparticles tends to degrade the film. This observed effect
is seen in the paint industry where this critical point is referred to as the critical
H 2 0, 0 2
silane film
A = anode, Al -> Al3+ + 3e

C = cathode, 0 2 + 2H20 + 4e -> 40H~
(a)
Figure 13. Model for effect of silica loading in silane films on corrosion inhibition, (a) No silica,
(b) low silica, (c) high silica.
pigment volume concentration (CPVC) and, thus, coatings are typically formulated
below this critical concentration.
The proposed corrosion inhibition mechanism is based on the above principle.
Figure 13 shows the model for the silica effect in the silane film. Figure 13a
shows the silane film with no silica, A refers to the anodic site (metal dissolution)
and C to the cathodic site (oxygen reduction). When the silane film is loaded with
small amounts of silica, as shown in Fig. 13b the silica suppresses the cathodic
reaction (oxygen reduction) by forming silicates, in a reaction with the cathodically
generated OH - ions which react with aluminum to form a passive silicate film.
Cathodic inhibition occurs as confirmed by the DC Tafel curves with a shift in
potential to lower values.
Si0 2 + 20H" -> S i O ^ + H 2 0
Figure 13c shows the effect of high amounts of silica in the silane film. The silica
particles adsorb the silane and the film becomes more porous. The silica particles
start to protrude out of the surface thus making water penetration easier through the
HzO, O z
• Si0 2 + 20IT - • SiQ32" + H 2 0 low

• Al + Si032~ -> passive silica
• ICorr lower because of cathodic
inhibition
f ) = silica particle
(b)
pores in the film. This leads to deterioration of the corrosion inhibition properties
of the film, an effect similar to the overpigmentation of paints.
5.7. Effect of organic and inorganic corrosion inhibitors on silane film

performance
It has been observed that solvent-based silane systems give better corrosion inhi-
bition performance than water-based silane systems [4, 5]. In attempts to increase
the performance of these water-based systems, it was investigated whether organic
or inorganic inhibitors could be added to the silane film. About 1000 ppm of the
inhibitors tolyltriazole, benzotriazole, or cerium nitrate were added to the silane sys-
tem and DC polarization tests were conducted to check if there were any changes
in the corrosion inhibition performance of the silane system due to the addition of
these inhibitors.
H 2 0, Q2
silane film
high
silica
• Si02 particles adsorb silanes
• CPVC exceeded
• CPVC is low because of relative
particle size
- silica particle with

silane
• Film becomes porous
(c)
Figure 14 shows the DC polarization curves for the water-based silane system
(bis-amino/vinyltriacetoxy) loaded with 1000 ppm (8 x 10~4 g/m2 in the silane
film) of tolyltriazole inhibitor. The AA2024-T3 panels were immersed in a 0.6 M
NaCl solution for 2 h before the test. It can be seen that as the silane is loaded
with the tolyltriazole inhibitor, the cathodic current density drops by half a decade
as compared to the silane system without the inhibitor, indicating better corrosion
protection. The curves show that the net reduction in the corrosion rate is achieved
by limiting the cathodic reaction rate. The decrease in Econ to lower values indicates
that tolyltriazole acts as a cathodic inhibitor. Figure 15 shows the effect of adding
benzotriazole inhibitor to the silane system. The same test procedures as above
-0.400
-0.500
-0.600
S
~ -0.700 |
eo
| -0.800 |
Q.
-0.900 .
-0.1000.
-9.0 -8.0 -7.0 -6.0 -5.0 -4.0 -3.0 -2.0
Figure 14. DC polarization curves for silane (bis-amino:VTAS) (3) with and (2) without tolyltriazole
inhibitor on AA2024-T3 alloy. (1) Blank.
-0.200
Potential
(V)
-0.700
-7.0 -6.0 -5.0 -4.0

Figure 15. DC polarization curves for silane (bis-amino:VTAS) (3) with (1000 ppm in solution) and
(2) without (benzotriazole inhibitor on AA2024-T3 alloy. (1) Blank.
were followed. The DC curves indicate a drop in cathodic current density with the
addition of the benzotriazole inhibitor implying a reduction in the cathodic reaction
rate and better corrosion inhibition performance as compared to the silane without
the inhibitor.
The inorganic inhibitor cerium nitrate (1000 ppm) was added to the water-based
(bis-amino/vinyltriacetoxy) silane system and DC polarization tests conducted on
AA2024-T3 alloy after immersion in the 0.6 M NaCl electrolyte for 2 h to achieve
a steady state. Figure 16 shows the DC polarization curves for the silane with
-0.500 j -
-0.600 I
-0.700 J-
Potential
(V)
-0.800 +
-0.900 4-
-1.000 I
-1.100 I
-1.200 -I 1 1 1 1 1 1 1 1
-9.0 -8.0 -7.0 -6.0 -5.0 -4.0 -3.0 -2.0 -1.0
Figure 16. DC polarization curves for silane (bis-amino:VTAS) (3) with (1000 ppm in solution) and
(2) without cerium nitrate inhibitor on AA2024-T3 alloy. (1) Blank.
and without the inhibitor and the blank as control. Cerium salts act as cathodic
inhibitors [15], they suppress the oxygen reduction reaction by forming an oxide
or hydroxide film. Ce 3+ ions precipitate on cathodic sites (reduction reaction
sites) [15] and protect the metal by forming precipitates. The silane containing the
cerium nitrate inhibitor shows a lowering of the Ecorr value, i.e., a shift of Econ in the
cathodic direction and a reduction in the cathodic current density and zcorr values.
These results indicate that the silane solution loaded with water-soluble inhibitors
can result in enhanced corrosion inhibition performance of the silane film.
The drop in /corr for the organic inhibitor loaded silanes indicates a blocking
effect of active reaction sites. Comparing the corrosion current (/COrr) and the
potential (EcorT) of the organic inhibitor loaded silanes, the following is observed.
Figures 14 and 15 show that for the organic-inhibitor-loaded silane the corrosion
current icorr decreases but there is no effect on the corrosion potential Ecorr. This
shows that the water is not able to penetrate the silane network and reach the active
metal reaction sites (anode and cathode). The triazole organic inhibitor in the silane
film can react with the copper containing phases and form Cu-triazole complexes
thus making the copper unreactive to the corrosive medium. The inhibitor thus
gets immobilized at the reactive metal surface and gives a reduction in the cathodic
current density. Whereas the inorganic cerium nitrate shows a decrease in current
density (cathodic) and Ecorr, indicating that the inorganic inhibitor does not interact
with the silane network and can be soluble when the electrolyte penetrates the silane
film thereby giving the silane film an electrochemical effect.
We have, thus, seen that addition of cathodic inhibitors (organic and inorganic)
gives the silane film a secondary form of protection and improves the corrosion
resistance of the silane film further. It also gives the silane film an electrochemical
effect. The corrosion inhibition efficiency for the silane with the inhibitors shows
an increase in the protection ability of the film compared to the silane without them.
3.8. Self-healing effect in silane films

The leachability of corrosion inhibitors can be used to achieve the self-healing effect
that is observed with chromates. The idea of adding inhibitors is to improve the
performance of the silane and also to achieve the self-healing capability. The silane
with cerium nitrate inhibitor coated onto AA2024-T3 panels was scribed by a sharp
edged knife. The width of the scribe was about 100-120 /xm. The scribed panel was
immersed in 0.6 M NaCl electrolyte for 7 days. Figure 17 shows pictures of panels
treated with silane (bis-sulfur) and silane with cerium nitrate inhibitor (1000 ppm).
The scribed area is subjected to corrosion as it is unprotected by the silane film. The
scribe in the silane film with the cerium nitrate inhibitor (Fig. 17a) shows less or
no corrosion along the scribe, whereas the silane film (Fig. 17b) shows corrosion
along the scribe, indicating that the silane film cannot protect the damaged area
and, thus, lacks the self-healing effect. The white spots are the corrosion products,
which cannot be observed in the silane with the cerium nitrate inhibitor as compared
to the scribe in the silane-treated only panel. Figure 18 shows the SEM images
(a) (b)
Figure 17. 7-day immersion test results in 3.5% NaCl for AA2024-T3 panels coated with bis-sulfur
silane with (a) and without (b) cerium nitrate inhibitor.
Modified silane coatings for corrosion protection ofAl alloyi. 155
Figure 18. (a) Scanning electron image of AA2024-T3 coated with silane containing cerium nitrate
(1000 ppm in solution) shows little corrosion product after immersion for 7 days in 0.5 M NaCl.
(b) Scanning electron image of AA2024-T3 coated with silane and scribed, showing distribution of
corrosion product in the scribe immersed for 7 days in 0.5 M NaCl.
of the scribes in the silane film with (Fig. 18a) and without inhibitors (Fig. 18b).
The scribe in Fig. 18a shows little or no corrosion products whereas the scribe in
Fig. 18b shows corrosion products along the scribe. This indicates that the scribe
with little or no corrosion must have cerium nitrate deposits. EDX analysis (Fig. 19)
Figure 19. EDX analysis of the scribe in AA2024-T3 coated with silane (bis-amino:VTAS)
containing cerium nitrate (1000 ppm in solution), immersed in 0.5 M NaCl for 7 days.
confirmed the presence of cerium in the scribe. The cerium ions must have leached
out of the silane film and precipitated on the scribed area, where they reduce the rate
of the cathodic reaction.
These results indicate that the silane film can be a reservoir for water-soluble and
leachable inhibitor, which can be released so that it can interact with anodes or
cathodes in the unprotected regions and prevent further corrosion. The silane film,
while being hydrophobic, allows enough water penetration to leach out the soluble
cerium ions without causing delamination of the film. This is of great importance
in the aircraft and building industries as the chromates possess this self-healing
capability and its replacement must be able to reproduce this property. These results
show promise and further research is focused on this area.
3.9. Inhibitor release "on demand"

The above tests confirm that cerium nitrate gives a self-healing effect to the silane
film. But the water-soluble inhibitor must be released only on demand, i.e., the
inhibitor should not be washed away when exposed to water. Ideally when the
electrolyte penetrates the silane film with cerium nitrate inhibitor, local anodic and
cathodic reactions should start to occur at the metal surface. The areas where the
cathodic reactions occur (at high pH) should trigger the release of cerium nitrate
inhibitor (cathodic inhibitors).
Based on the above assumption a simple test was done. AA2024-T3 panels
were coated with silane doped with 100 mg of cerium nitrate as described in the
Experimental section. The panels of known area were immersed in deionized water
Table 1.
Amount of cerium (in ppm) leached out at different pH levels in 100 ml of water from silane film
loaded with cerium nitrate (1000 ppm) inhibitor
Time (h) NaCl (3.5%) pH 2 (ppm) pH 7 (ppm) pH 11 (ppm)

1 0.0096 0.0085 0.174
48 1.091 0.0139 0.0124 1.336
96 0.0127 0.0141 1.414
at different pH levels of 2, 7 and 11. After exposing the panels at different pH levels
for 1, 48 and 96 h, the deionized water at various pH levels was tested using ICP/MS
for presence of cerium. Table 1 shows the levels of cerium detected (in ppm) at
different pH levels. At pH 11 the amount of cerium detected is higher by more than
two orders as compared to pH 7 and pH 2. This indicates that at high pH cathodic
reaction occurs and the cerium nitrate is released into the solution. The generation
of hydroxyl ions due to the cathodic reaction swells up the silane film leading to the
release of the inhibitor. This shows that the cerium nitrate is not washed away when
exposed to water and is released only on demand to achieve self-healing.
The same test was conducted by exposing the panel to 3.5% NaCl solution instead
of the deionized water. The panel was exposed for 48 h and the solution was tested
for cerium using ICP/MS. Table 1 shows the ppm level of cerium in the electrolyte.
The amount released is comparable to that released at high pH. This shows that
when subjected to a corrosive seawater environment, where more than one cathodic
reaction can take place the cerium nitrate inhibitor is released. The sodium ions can
possibly replace the cerium ions in the silane film triggering the displacement of
cerium ions from its positions in the film. The displaced cerium ions come out of
the film and are detected in the solution. These tests confirm that the cerium nitrate
inhibitor is released only when the cathodic reaction takes place. The storage and
the life expectancy of the cerium nitrate inhibitor in the silane film will, therefore,
depend on the severity of the corrosion environment.
3.10. Paint adhesion due to cerium-doped silane film

Figure 20 shows the crosscut tape adhesion test results on silane-treated
AA2024-T3 panels with and without cerium nitrate inhibitor. Figure 20a shows
the bis-amino/VTAS (2: l)-treated AA2024-T3 panel coated with sprayed powder
polyester paint cured at 200°C; the part of the panel above the black line shows
results for crosscut adhesion test done immediately. The part below the line was
exposed to deionized water for 2 days before the test. There was no delamination
of the paint in the grid on both parts of the panel. The same procedure was fol-
lowed for bis-amino/VTAS with 1000 ppm of cerium nitrate in the silane solution.
Figure 20b shows the adhesion test results before and after exposure to deionized
water. The results show that there is no loss of adhesion after doping the silane film
with cerium nitrate inhibitor.
(a) (b)
Figure 20. Crosscut adhesion tape test result on AA2024-T3 coated with 5% bis-amino/VTAS (2:1)
and polyester powder painted (a); adhesion test result on bis-amino/VTAS with 1000 ppm cerium
nitrate inhibitor and polyester powder painted (b). Top: before exposure to DI water. Bottom: after
exposure to DI water.
4. CONCLUSIONS
• Silane films can be colored by adding organic colorants. The corrosion perfor-
mance does not decrease by the addition of these organic dyes, even after paint-
ing. The colored films can be easily detected when applied to metal substrates.
Paint adhesion after storage in humidity can be achieved by properly choosing
dyes and colorants that become insoluble after curing.
• Incorporation of a small amount of silica improves both the corrosion inhibition
and mechanical properties of the film. An excess amount of silica nanoparticles
into the bis-sulfur silane film seemed to degrade the corrosion inhibition perfor-
mance of the film. The EIS results indicated that the bis-sulfur silane film loaded
with 50 ppm of silica nanoparticles tended to prematurely delaminate from the
substrate due to the weakened interfacial adhesion and, therefore, was no longer
able to protect the substrate underneath. Such delamination may be caused by an
increase of the porosity and/or stresses in the silane film due to the large amount
of particles, whereas optimum addition of silica nanoparticles shows an increase
in corrosion inhibition performance.
• Silanes loaded with water-soluble corrosion inhibitors can act as a reservoir

of such inhibitors and release them slowly to the defect sites and achieve the
self-healing effect exhibited by chromate coatings. While cerium nitrate shows
promise of release, the organic inhibitors do not heal defects in the coating.
Acknowledgements
The authors are grateful for the financial support provided by Chemat Technologies,
Inc. (Northridge, CA, USA).
REFERENCES
1. W. McGovern, P. Schmutz, R. G. Buchheit and R. L. McCreery, / Electrochem. Soc. 147, 4494
(2000).
2. W. J. van Ooij and T. F. Child, CHEMTECH 28, 26 (1998).
3. V. Subramanian, Ph.D. Dissertation, University of Cincinnati, Department of Materials Science
and Engineering, Cincinnati, OH (1999).
4. G. P. Sundararajan, M.S. Thesis, University of Cincinnati, Department of Materials Science and
Engineering, Cincinnati, OH (2000).
5. W. J. van Ooij, D. Zhu, G. P. Sundararajan, S. K. Jayaseelan, Y. Fu and N. Teredesai, Surface
Eng. 16, 386 (2000).
7. M. A, Petrunin, A. P. Nazarov and Yu. N. Mikhailovski, J. Electrochem. Soc. 143, 251 (1996).
8. P. R. Underhill and D. L. Duquesnay, in: Silanes and Other Coupling Agents, K. L. Mittal (Ed.),
Vol. 2, p. 149. VSP, Utrecht (2000).
10. D. Zhu and W. J. van Ooij, J. Adhesion ScL Technol. 16, 1235 (2002).
11. F. D. Osterholtz and E. R. Pohl, J. Adhesion ScL Technol. 6, 127 (1992).
12. V. Palanivel, D. Zhu and W. J. van Ooij, paper presented at the Workshop on Nanoscale
Approaches to Multifunctional Coatings, Keystone, CO (2002).
13. D. Zhu and W. J. van Ooij, Corrosion Set 45, 2177 (2003).
14. A. N. Onal and A. A. Aksut, Anti-Corrosion Methods Mater. 12, 339 (2000).
15. K. Aramaki, Corrosion ScL 43, 2201 (2001).
16. R. L. Parkhill, E. T. Knobbe and M. S. Donley, Prog. Org. Coatings 41, 261 (2001).
17. R. M. A. Azzam and N. M. Bashara, Ellipsometry and Polarized Light. North Holland,
Amsterdam (1977).
Silanes and Other Coupling Agents, Vol. 3, pp. 161 -176
Ed. K. L. Mittal
© VSP 2004
Musical instrument strings and corrosion: a comparative

study of aminosilane and benzotriazole surface treatments
ANTHONY A. PARKER*
A. A. Parker Consulting <£ Product Development, 10 Columbine Circle, Newtown, PA 18940, USA
Abstract—The corrosion resistance of musical instrument strings is improved through surface modi-
fication with polymeric film-forming materials including benzotriazole (BTA) and N-2-aminoethyl-3-
aminopropyltrimethoxysilane (AAPS). The improvements are most noteworthy when the core of the
string is cathodic in character (i.e., titanium as opposed to steel). The corrosion inhibiting function
of benzotriazole and other azole compounds is known to result from a combination of factors, includ-
ing the ability of the azole functionality to chelate with the metal oxide surface to form a polymeric
film. AAPS possesses analogous molecular-level capabilities, and also provides improved corrosion
protection when applied from dilute solutions. FT-IR analyses of AAPS-treated strings reveal that
improved corrosion protection is accompanied by a frequency shift in both the N-H deformation and
N-H stretching bands. Dynamic Mechanical Analysis (DMA) reveals that the resonant characteristics
are influenced by the weight of the polymer coating. Collectively, these results provide unique insight
into the common molecular level attributes that are required to simultaneously optimize the corrosion
resistance and vibrational characteristics of musical instrument strings.
Keywords: Aminosilane; coatings; FT-IR; corrosion; musical instrument strings.
1. INTRODUCTION
1.1. Chemistry's role in the evolution of musical instrument strings
Musical instrument strings have benefited from a 100 year progression of improve-
ments ranging from the use of synthetic polymer fibers and protective coatings to
the use of chemical surface treatments and galvanically matched corrosion-proof
materials — all for the purpose of increasing longevity.
Prior to the 1890s, guitar strings were predominantly made from natural biopoly-
mer known as "catgut", or, more specifically, fibrous polymers taken from the in-
testines of various animals (a few specialized products are still made from such
materials). These strings were not only difficult to make, they were also difficult to
keep in tune due to their sensitivity to moisture and humidity.
*TeL: (1-215)497-3065; e-mail: starmacpar@aol.com

162 A. A. Parker
The catgut core with its deficiencies was eventually replaced by other materials,
including ferrous-based alloys (in the 1890s), synthetic polymers like nylon (after
the 1930s), polymer-coated metallic strings (in the 1990s) and, most recently,
a titanium alloy (in 2001). Today's popular steel guitar strings are still made
from geometric-shaped steel core wires that are wound like springs, typically with
copper or nickel alloy wires to control mass (a similar version was first patented
in 1878 [1]). When originally introduced, these types of strings had the immediate
advantage of staying in tune better than catgut strings. They were also louder in
volume as a result of their higher mass and tension. Indeed, the tension was so
high that instruments had to be redesigned with more structural reinforcements to
combat warping and bowing. Unfortunately, the major disadvantage of these strings
was and still continues to be corrosion.
The battle against corrosion was at first fought with a tried and true 15th century
European chemical innovation [2] that is still being used to protect metals in a
variety of applications. The steel core wire was coated with a thin layer of molten tin
prior to being wound with the other copper and nickel alloy wires. The tin coating
served the same purposes then that it serves now: it provided a soft base so that the
winding wire could be firmly embedded, and it passivated the relatively anodic steel
from the more cathodic winding wires. Although this vintage chemistry is still used
to slow down the corrosion process, it by no means has solved the problem.
The advent of a second chemical innovation in the 1930s (thanks to Carothers)
led to synthetic nylon strings, which are still in use today. More apt to stay in
tune and less moisture sensitive than their catgut counterparts, nylon strings also do
not corrode, and their soft tones are aesthetically pleasing to many musicians and
listeners alike. Several other synthetic fibers have come into use as well [3]; and,
like steel core strings, many are wound with both metal alloys and other synthetic
fibers to control their masses and resultant tensions. These offshoots of polymer
chemistry continue to fill a market niche, but there is still a large audience for the
sound and playing characteristics of metallic strings. For this reason, inventors have
been quietly occupied through most of the 20th century in attempts to improve the
performance of metallic strings.
As early as the 1930s, inventors were attempting to arrest corrosion by coating
metallic strings with natural and synthetic polymer lacquers [4, 5], Unfortunatefy,
such coatings also reduced the brightness of the strings during use. The perceived
brightness of a string arises from its ability to excite the resonance vibrations of a
musical instrument. Anything that interferes with these vibrations will deteriorate
sound quality. Thus, corrosion byproducts, contamination from finger contact and
even coatings that are designed to help prevent corrosion can all contribute to the
dampening of string vibrations.
When an ideal string under tension is plucked, struck, or bowed, it freely vibrates
at a fundamental frequency / , which is controlled, in part, by the tension on the
string T (higher tension produces higher frequencies), the mass per unit length m
(heavier strings vibrate at lower frequencies) and the distance between its end
points L, otherwise known as the string's speaking length. Analytically, the
Musical instrument strings and corrosion 163
vibration of an ideal string can be expressed as follows [6]:
f(n)=n/2L(T/m)l/\
where n = 1 for the fundamental tone. Note that in addition to the fundamental
tone, a geometric series of overtone vibrations are produced at integer values of
n > 1. These overtones excite a complimentary ensemble of instrument resonance
frequencies whose amplitudes are very dependent on both the type of instrument and
the properties of its component materials. In fact, the overtones and the resultant
resonance vibrations that they excite are responsible for each instrument's unique
tone, or timbre. They are at least, in part, responsible for the audible difference
between a note plucked on a guitar, and the same note struck on a piano or
harpsichord.
A string's overtone vibrations occur at higher frequencies than the fundamental
tone, but their amplitudes are significantly lower. Hence the overtones are the
first vibrations to be perceptibly dampened by frictional losses from corrosion
byproducts, or by mechanical losses from polymers that are otherwise designed
to slow down corrosion. This problem was, in part, overcome in the 1990s by
a poly(tetrafluoroethylene) (PTFE) coated string innovation [7] known as Elixir™
(W. L. Gore & Associates). The PTFE coating of this product is a thin film
with very specific machine direction and cross-machine direction mechanical
properties. The film is spirally wound around the traditional metallic string in
such a way so as to minimize its dampening effect on the motions that produce the
overtone vibrations. Although the dampening effect is not entirely eliminated, the
resultant corrosion protection and longevity have enabled this product to become an
important example of chemistry's influence on the evolution of musical instrument
strings.
Perhaps the most recent application of chemistry has come from Rohrbacher
Technologies' introduction of corrosion-proof metallic strings with titanium alloy
cores [8]. As opposed to addressing the corrosion problem with protective
coatings, Rohrbacher Technologies has devised a string that eliminates the galvanic
couple between the core and winding wires. Conventional steel core strings are
comprised of materials that are galvanically mismatched, and hence the propensity
for corrosion is always present. An electrochemical couple is established between
the traditional materials when salt and moisture (from human hand contact) create
a type of salt-bridge that completes the contact. Unlike traditional strings, the
titanium core string is comprised of a relatively cathodic metal core (cathodic
because titanium metal is passivated with a thin oxide layer) and a second metal
winding wire, where the difference in galvanic potential between the two metals
(as measured by the difference in galvanic potential with respect to a saturated
calomel electrode in seawater) is as close to zero as possible. A special nickel
wound titanium alloy core satisfies both the mechanical property requirement
(to withstand the tension of tuning), and the galvanic requirement (to prevent
corrosion).
164 A. A. Parker
1.2. Surface treatments and corrosion

Still another area of chemical innovation deals with the use of surface treatments
to inhibit corrosion. For example, one patent discloses a method for treating steel-
core musical instrument strings with dry lubricant particles, moisture displacement
agents and a corrosion inhibitor [9]. In another case, surface treatments were used
to increase the longevity of copper-alloy wound titanium-core strings [8].
The efficiency of such surface treatments can be influenced by the type of
metal alloy and by the type of treatment. For example, in this study, both
benzotriazole (BTA) and N-2-aminoethyl-3-aminopropyltrimethoxysilane (AAPS)
surface treatments are shown to provide the greatest protection when the core is
comprised of titanium (as opposed to steel).
Triazole compounds (including benzotriazole and benzimidazole) have long been
reported to provide corrosion protection for copper and zinc alloys [10]. Although
the mechanism is still not completely understood, it is believed that certain
azole compounds interact with metal oxide surfaces to form polymeric protective
complexes which are resistant both to oxidation under anodic conditions, and to
dissolution under cathodic conditions [10].
Several spectroscopic techniques have been used to study the chemical interaction
between azole compounds and metals [11-14]. For example, FT-IR studies have
shown that the coordination between benzimidazole and various metals leads to
an increase in the C-N stretching frequency near 1500 cm - 1 [15]. Similarly,
XPS studies have revealed that the coordination between azoles and metals leads
to an increase in the N l s electron binding energies for specific azole nitrogen
atoms [11,13]. Solid state NMR has also been used to study the interaction between
13
C-labeled benzimidazole and ZnO, where the C-2 carbon of the surface-adsorbed
azole ring experiences a 9 ppm downfield shift, consistent with an electron density
withdrawal from the imidazole carbon center [16].
Studies on chromatographic columns modified with various imidazole compounds
have revealed that the efficiency for metal adsorption depends on azole molecular-
level structure [15], where the most efficient adsorbers have the ability to form
bidentate metal-chelate complexes comprised of 5-6-membered rings [17]. Sol-
ubility studies of model compounds have also revealed that the best corrosion in-
hibitors form relatively insoluble metal-chelate complexes [10]. Analogously, the
best corrosion inhibitors for cuprous metal surfaces are triazoles that not only bind
with the oxide surface through nitrogen lone pairs, but also have the ability to
form [cuprous(I)triazole]„ polymeric complexes that reinforce the protective oxide
layer [10, 12].
1.3. N-2-aminoethyl-3-aminopropyltrimethoxysilane (AAPS) as a potential

corrosion inhibitor
In light of the extensive literature on corrosion inhibition by azoles, it is apparent
that a good corrosion inhibitor should have the ability to stabilize the metal oxide
surface layer against both cathodic and anodic reactions. Although there may be
many mechanisms through which this can be accomplished, benzotriazole (BTA)

provides protection through a unique combination of chelation and polymer network
formation. Hence, by using BTA as a "template", it becomes possible to imagine
other types of molecules that might provide similar protection. For example,
when viewed in this way, N-2-aminoethyl-3-aminopropyltrimethoxysilane (AAPS)
appears to possess similar potential for both chelation and polymer network
formation.
AAPS is known to be a polymeric film-forming adhesion promoter, and has
been extensively studied in many applications as reported in the scientific litera-
ture [18-34]. Like the triazoles and imidazoles, AAPS possesses nitrogen atoms
with lone electron pairs that are available for chelation [18]. Furthermore, the
nitrogen atoms of AAPS are separated by two methylene carbons, making them
ideally suited for the formation of five-membered metal-chelate complexes. Such
structures were found to be favorable for facilitating metal adsorption onto silica
gels functionalized with &>-amino-alkylbenzimidazoles [17]. In addition, AAPS has
the capacity to polymerize through alkoxide hydrolysis and silanol condensation to
form a three-dimensional polymeric network. Thus, AAPS, like benzotriazole, has
the potential both to chelate and polymerize on the surface of copper alloys.
Hence, it follows that AAPS, like BTA, may also have the ability to impart
corrosion protection to copper alloys. In the context of the present study, this
possibility is particularly interesting; especially since copper alloys are often used
as winding metals for musical instrument strings. In fact, one of the most popular
winding metals is a phosphor bronze copper alloy [35], which is used to produce
both titanium [8] and steel-core strings [36].
Certain organosilanes have recently been used to provide corrosion protection to
various non-copper alloys [37, 38]. In one study, the corrosion resistance of 2024
and 7075 aluminum alloys was improved with a mercapto-functional silane surface
treatment, whereas an aminosilane provided little to no improvement [37]. The poor
performance of aminosilane was attributed to the probable orientation of protonated
amines toward the negatively charged aluminum oxide surface. The implication
was that the resultant aminosilane bonds would be more susceptible to moisture
than those formed with mercaptosilane.
In another study, the best protection was reported to occur from a two-step
treatment process, where the substrate was first coated with a non-functional silane
and then with a functional silane, so that the functional group of the second silane
would not be allowed to interact with the metal surface [38].
By contrast, one of the objectives of the present study has been to investigate the
use of a simple, one-step coating process to protect phosphor bronze-wound strings.
In light of the aforementioned similarities between AAPS and BTA, and given that
BTA is known to be a protectant for phosphor bronze [8], AAPS was chosen for its
potential ability to perform a similar function. In addition, the protection imparted
by poly(AAPS) was compared to the protection imparted by benzotriazole (BTA)
under both anodic and cathodic conditions (as uniquely enabled through the use of
steel and titanium cores, respectively).
166 A, A. Parker
2. EXPERIMENTAL
2.1. Materials and procedures for corrosion testing
The musical instrument strings for this study included three types of commer-
cially available constructions that were designed for acoustic guitar: phosphor
bronze-wound titanium-core strings from a set of Rohrbacher low-tension Titanium
Acoustic Guitar Strings™ (available from Rohrbacher Technologies, Bordentown,
NJ, USA); and two types of phosphor bronze wound steel-core strings — one from
a set of John Pearse™ medium-tension acoustic guitar strings, and the second from
a set of DR-Rare-Bronze™ light-tension acoustic guitar strings. All of the strings
were used as received.
In each corrosion experiment, 30 ml glass vials with lids were filled with 5 g of a
saturated aqueous NaCl stock solution (prepared with deionized water and reagent
grade NaCl from Aldrich). The string samples were cut into 3.17 cm strips, and
were placed into the glass vials with a portion of each submerged below the water
line and with a portion of each above the air/water interface. Two visual criteria
were used to qualitatively rank corrosion: the degree of winding discoloration above
the water line and the degree of turbidity below the water line. Differences above
the water line were generally discernible within the first 24 h (if corrosion was to
occur), whereas turbidity below the water line was discernible either within a few
hours (steel-core samples), or within several days (titanium-core samples).
In the first series of surface treatment experiments, the corrosion resistance of
phosphor bronze wound strings was determined as a function of the relative coating
weight of benzotriazole (BTA, Aldrich). The strings for this comparison included
a John Pearse™ medium-tension "E" string (phosphor bronze-wound steel core);
and a Rohrbacher Technologies low-tension "E" string (C521 phosphor bronze
wound Ti alloy core). Both types of strings were cut into 3.17 cm strips and were
then surface treated by dipping into solutions of benzotriazole (30, 300, 600 and
1000 ppm concentrations by weight) dissolved in a 95/5 weight percent mixture
of denatured ethanol (denatured reagent grade from Aldrich) and deionized water.
Each wound section was separately dipped into a solution for 1 min, and then was
removed to air dry for 15 h prior to corrosion testing. Samples of each string were
also tested as received (with no treatment), and after solvent washing with ethanol.
The samples were qualitatively evaluated as described above, and were ranked (from
low to high) according to the relative degree of turbidity after 30 h of exposure.
In the second series of surface treatment experiments, phosphor bronze wound
strings were treated to determine the effect of an organosilane surface treatment
on the corrosion resistance of both steel-core and Ti alloy-core strings. The
strings for this comparison included a DR-Rare-Bronze™ light-tension "A" string
(C521 phosphor bronze-wound steel core); and a Rohrbacher Technologies low-
tension "A" string (C521 phosphor bronze-wound Ti alloy core). The composition
of C521 phosphor bronze is given by ASTM B159 [35] as Pb (0.05% max), Fe
(0.10% max), Sn (7.0-9.0%), Zn (0.20% max), P (0.03-0.35%) and Cu (remainder).
Both types of strings were cut into 3.17 cm strips as described above and were
then surface treated by dipping into solutions of pre-hydrolyzed N-2-aminoethyl-

3-aminopropyltrimethoxysilane (AAPS), otherwise known as Z6020 from Dow
Corning. The AAPS was prehydrolyzed by mixing 50 parts silane with 50 parts
ethanol and 5 parts distilled water at 25°C. After 24 h, the prehydrolyzed concentrate
was diluted into 70:30 (v/v) isopropanol/water (Aldrich) to yield solutions with
concentrations of 0.05, 0.1, 0.2, 0.4, 0.8, 1.6, 3.2 and 6.4% active silane by weight.
Each wound section was separately dipped into a solution for one minute, and then
was flash dried with a hot-air gun with a 2-s dwell time. The samples were then
allowed to equilibrate under atmospheric conditions for 48 h prior to corrosion
testing.
Qualitative rankings were established based on the relative degrees of visual
corrosion (as judged by turbidity), where the titanium-core sample with the lowest
degree of corrosion (after 21 days of exposure) was ranked as 1, and the remainder
of the group was ranked in sequence from low to high according to increasing
degrees of turbidity. The steel-core samples (after 24 h of exposure) were then
ranked in comparison to this baseline, where the visual turbidity of the untreated
steel-core control (after 24 h of exposure) was noted to be equivalent in rank to the
untreated titanium-core control sample (after 21 days of exposure). The other steel-
core samples were then ranked in relative comparison to one another (from low to
high) with respect to the titanium-core baseline.
2.2. FT-IR studies
Attenuated total reflectance FTIR analyses were performed on two DR-Rare-

BronzerM light-tension "A" strings treated with 0.2 and 6.4 wt% AAPS solutions,
respectively. In addition, an FTIR spectrum was also collected for a neat AAPS
film that was cast onto a TeflonrM-coated pan under equivalent conditions.
Spectra were collected with a Magna-IR™ 560 instrument from Thermo-Nicolet
that was equipped with a DTGS KBr detector and a MIRacle™ ZnSe single
reflection HATR accessory from Pike Technologies. The HATR accessory was
equipped with a clamp to maintain adequate contact between the sample and the
ZnSe crystal. A total of 64 scans were signal averaged at a resolution of 4 cm - 1 .
2.3. DMA studies
Dynamic Mechanical Analysis (DMA) was also performed on DR-Rare Bronze™

light-tension "A" strings (treated and untreated). The treated samples included
constructions that were surface coated with 0.2 and 6.4 wt% AAPS solutions using
the methods as described above. The analyses were performed on a TA Instruments
model 983 DMA equipped with vertical clamps. The data were collected in resonant
mode under ambient conditions using a sample length of 21 mm and an oscillation
amplitude of 0.8 mm.
168 A. A. Parker

3.1. The corrosion behavior of phosphor bronze wound musical instrument strings:
the effect of BTA concentration
The qualitative corrosion results for BTA-treated steel-core and titanium-core
strings are presented in Fig. 1. The graph depicts the combined relative ranking
(based on visual turbidity) of all of the solutions after 30 h of exposure. Although the
overall degree of corrosion among the titanium-core samples was significantly less
than that of the steel-core samples, the protection afforded by BTA was observed to
increase monotonically in both sets. The initial differences were apparent above the
water line (from partial oxidation and discoloration of the phosphor bronze winding)
and below the water line (from the turbidity due to the build-up of corrosion
byproducts and precipitates). After 3 days, the differences in steel-core samples
were no longer discernible (due to extreme corrosion), whereas the differences
among titanium-core samples remained visible for more than 1 week.
These results are consistent with prior corrosion studies of BTA treated copper
alloys [10-14], Previous studies have shown that BTA provides protection against
cathodically induced oxidation of copper alloy surfaces [10]. This is consistent with
the finding that BTA lessens the oxidation of phosphor bronze when it is wound
around a more cathodic titanium alloy core. Similarly, BTA also provides anodic
protection in cases where copper alloys are in contact with more anodic metals [10].
This is consistent with the finding that BTA lessens the oxidation of the more anodic
steel core when it is wound with phosphor bronze.
12
Steel/ Phosphor Bronze
Titanium/ Phosphor Bronze
S
8
8
CD
G
CD
>
JO
2 2
> iE CD
£a
CO
E
CO
E
aa a a. aa
BTA Concentration (ppm by weight)
Figure 1. The effect of BTA concentration (parts per million, ppm by weight) on the relative degree
of corrosion of phosphor bronze wound steel-core and titanium-core strings. Comparisons include
strings tested as-received, and strings washed with ethanol prior to corrosion testing. Results represent
the combined relative rankings (based on visual turbidity) of all the NaCl solutions after 30 h of
exposure, where 1 = low and 11 = high.
3.2. The effect of AAPS organosilane on the corrosion behavior of phosphor

bronze-wound musical instrument strings
The results of the present study show that AAPS improves the corrosion resistance
of phosphor bronze wound metallic strings. Figure 2 provides a qualitative visual
ranking of the relative degree of corrosion (as judged by the turbidity of NaCl
solutions) for both steel-core and titanium-core strings as a function of the silane
solution concentration. For the case of steel-core strings, the NaCl solutions were
orange in color (due to the likely formation of Fe203 precipitates), whereas the
solutions for the titanium-core strings were blue in color (due to the likely formation
of copper(II) precipitates). Although the trends in both sets were similar, the
differences among the steel-core strings were discernible within 24 h of exposure,
whereas the differences among titanium-core samples could not be discerned for
several days.
These results show that the corrosion resistance improves with increasing con-
centration up to a level of 0.2% AAPS by weight. Beyond this level, the degree of
corrosion appears to become higher. Thus, up to a certain threshold level of surface
treatment, AAPS appears to stabilize the surface of phosphor bronze against both
10
Steel/ Phosphor Bronze, 24 hours
c Titanium/ Phosphor Bronze, 21 days
o
§
o
o
O)
2
5
AAPS Concentration (weight %)
Figure 2. The effect of AAPS concentration (% by weight) on the relative degree of corrosion
of phosphor bronze wound steel-core and titanium-core strings. Results represent the qualitative
combined relative rankings (based on visual turbidity) of all the NaCl solutions. The baseline for
this relative comparison was established by assigning a rank of 1 to the titanium-core sample with the
lowest degree of visual corrosion (the sample treated with the 0.2 wt% AAPS solution by weight).
The titanium-core strings were then ranked from low to high in accordance with increasing levels
of turbidity with respect to this baseline. Given that the turbidity of the untreated steel-core sample
after 24 h of exposure was qualitatively similar to the turbidity of the untreated titanium-core sample
after 21 days of exposure, the untreated steel-core sample was assigned the same qualitative ranking
as the untreated titanium-core string (rank = 6). The remaining steel-core strings were then ranked
with either higher or lower values, depending on the relative degree of turbidity with respect to this
baseline.
170 A. A. Parker
cathodic and anodic reactions. Beyond this threshold, little to no visible corrosion
protection is provided.
Given that AAPS improves the corrosion resistance of phosphor bronze wound
strings, it appears that AAPS (like BTA) has the ability to stabilize the metal oxide
surface layer. Given that BTA provides similar protection, and given that Cu and
its oxides are known to bind with BTA to form polymeric complexes [10, 12], it
follows that Cu may also bind with the amine groups of polymeric AAPS. In fact,
AAPS has been observed to complex with Cu 2+ ions in other applications [18, 39].
3.3. DMA studies of AAPS coated phosphor bronze-wound strings
DMA studies indicate that the mechanical properties of the strings are strongly
influenced by the presence of the poly(AAPS) coating. Figure 3 shows that the
coated strings resonate at higher frequencies, and that the relative string stiffness
increases in proportion with the weight of the coating. From the design standpoint,
it is important to minimize these mechanical effects, especially since any change in
mechanical properties can dramatically influence the acoustic characteristics of the
string [6-8]. Fortunately, the results of this study show that the mechanical effects
can be minimized, and the corrosion protection can be simultaneously maximized
through the use of lower coating weights.
2.78- \
6.4 wt. % AAPS Solution Treatment "
2.76- "
o
a
2.74-
= 0.2 wt. % AAPS Solution Treatment
a4
U
2.72-
No Treatment
2.70-
Si
2.68- L t 1 ' > ' 1 < T- * 1 ' > < 1 <
__, ,
Time (minutes)
Figure 3. Resonant mode DMA (frequency vs. time) for a 6.4 wt% AAPS-treated steel-core string,
a 0.2 wt% AAPS-treated steel-core string and an untreated steel-core string.
3.4. FT-IR studies of AAPS-coated phosphor bronze-wound strings

Regardless of the specific nature of the chemical interactions between AAPS and
phosphor bronze, it is apparent that, below a certain coating weight, AAPS has the
ability to stabilize the phosphor bronze surface. In order to gain further insight
into this phenomenon, attenuated total reflectance FT-IR was used to analyze
the surfaces of two AAPS-coated strings, including one that displayed improved
corrosion resistance (the string coated with the 0.2% silane solution), and one
that exhibited worse corrosion resistance (the string coated with the 6.4% silane
solution).
Figure 4 shows attenuated total reflectance FTIR spectra (scaled for relative com-
parison) for the 6.4 wt% AAPS coated string, the 0.2 wt% AAPS-coated string
and a neat AAPS film (from top to bottom). The presence of a polymeric siloxane
is confirmed by the strong Si-O-Si asymmetric stretching band near 1100 cm - 1
(peak A) in all three spectra [18, 40]. In addition, both the neat film and the
6.4%-coated string show many of the spectral characteristics that have been pre-
viously associated with the presence of protonated and strongly hydrogen bonded
aminosilanes [18, 41-43], These characteristics include an absorption at approx-
3500 3000 2500 2000 1500 1000
Wavenumber (cm 1 )
Figure 4. Attenuated total reflectance FTIR spectra for a 6.4 wt% AAPS-treated steel-core string,
a 0.2 wt% AAPS-treated steel-core string and a neat AAPS film (from top to bottom; scaled for
relative comparison). Assignments include Si-O-Si asymmetric stretching band near 1100 c m - 1
(peak A); overlap of N-H stretching and NH3" stretching modes near 3250 c m - 1 (peak B);
N-H deformation band for protonated and/or hydrogen bonded primary amine near 1580 cm - 1
(peak C); N-H deformation mode for free, or associated, primary amine at 1620 cm" 1 (peak D*);
N-H stretching for free, or associated, primary amine at 3400 cm - 1 (peak E*).
172 A A. Parker
imately 3250 cm"1 (peak B) which is consistent with the hydrogen bonded N-H
stretching band and the NH^j" stretching mode [40], a broad set of weak absorptions
between 2100 cm"1 and 2800 cm - 1 which is consistent with a previous assign-
ment to combination bands for amine bicarbonate salts [41, 42], and an absorp-
tion at approximately 1580 cm - 1 (peak C) that is consistent with previous assign-
ments to the N-H deformation band for protonated and/or hydrogen bonded primary
amines [40-42].
The spectrum for the 0.2% AAPS-treated string appears quite different from
the other two. Although it shares some of the same spectral characteristics, two
additional absorption bands appear at 3400 cm - 1 (peak E) and at 1620 cm - 1
(peak D). These bands are consistent with previous assignments to the N-H
stretching and N-H deformation modes, respectively, for free, or associated,
primary amines [18, 40-42].
The appearance of the higher frequency bands on the surface of the 0.2%-treated
string (at 3400 cm - 1 and 1620 cm" 1 ), and the simultaneous diminution of these
bands on the 6.4% AAPS-treated string, suggest that the chemical state of the
amine plays a major role in corrosion inhibition. Specifically, the coating with the
highest fraction of protonated amines provides no corrosion protection, whereas the
coating with the higher fraction of either free or associated amines provides greatly
improved corrosion protection. Thus, like BTA, it appears that AAPS provides
improved corrosion protection when its amine moieties are free either to associate
with the metal oxide surface, or to chelate with metal ions that may otherwise
dissociate and migrate from the surface during the corrosion process. Conversely,
when the amine moieties are predominantly protonated (as in the case of the thicker
coatings), corrosion protection is not observed.
Figure 5 provides a description of the chemical environment as it might exist
near the interface between a poly(AAPS) coating and an inorganic substrate. Prior
studies have shown that poly(AAPS) films may contain a combination of free
and protonated amines [33, 44], as well as hydrogen bonded and chelated amines
(depending on the nature of the inorganic substrate) [18, 39]. In addition, the
protonated amines can associate with a number of possible counterions, including
protonated bicarbonates [41, 42], "bridging" bicarbonates (linked with metal oxide
Bronsted base sites) [34] and negatively charged metal oxide sites [37].
In the present study, the relative concentration of the protonated amines was
found to depend on the solution concentration employed during the coating process.
Specifically, deposition from a dilute solution leads to high fractions of free or
associated amines, whereas deposition from a concentrated solution leads to high
fractions of protonated amines.
Previous studies have suggested that protonated amines can influence both the
density and the moisture sensitivity of polymeric aminosilane films [37]. Thus, it is
conceivable that a high fraction of protonated amines could lead to an increase in
the water permeability of the poly(AAPS) coating. In addition, an increase in the
permeability of electrolytes could also accelerate the corrosion process. This could
Polymeric AAPS
S i ^ ^
/°^/V/°'
"^0 ,OH 0 0"H'"*0-H
/ >
-M M- -M.
Figure 5. The chemical environment as it might exist near the interface between a poly(AAPS)
coating and an inorganic substrate.
be one explanation for the worse corrosion behavior that is observed when films are
deposited from higher AAPS solution concentrations.
3.5. The analogies between BTA and AAPS

Taken collectively, the analogies between BTA and AAPS (from the literature
and from this study) can provide unique insight into the common molecular level
attributes of good corrosion inhibitors. The present study suggests that AAPS
(like BTA) has the ability to provide corrosion protection to copper alloys, but
only below a certain critical concentration, where it appears that the AAPS amines
are free to associate with the metal oxide surface. Specifically, it appears that
like BTA, AAPS has the ability to stabilize the protective oxide layer on the
surface of a phosphor bronze alloy under both anodic and cathodic conditions (as
simulated by steel-core and titanium-core musical instrument strings, respectively).
Further extrapolation of these analogies leads to the suggestion that both compounds
provide corrosion protection through analogous mechanisms (i.e., through chelation
and polymerization).
In order to gain more insight into these mechanisms, the progression of future
AAPS research could be modeled after the historical progression that was once
taken with BTA [10-15]. For example, it would be beneficial to use XPS to
study the surface compositions of AAPS treated copper alloys (i.e., as a function of
coating weight, deposition process, and exposure time under corrosive conditions).
Homologous compounds (i.e., monofunctional amines and diamines with and
without silane functionality) could also be tested for the purpose of isolating the
effects of chelation and polymerization on corrosion protection. It would also be
174 A. A. Parker
helpful to determine the relative solution concentrations of metal ions throughout

the course of each corrosion experiment. In addition, potentiometric measurements
could be used to better quantify the effects of passivation by poly(AAPS) under both
anodic and cathodic conditions. Also, it would be important to determine whether
AAPS, like BTA, can protect other types of copper alloys.
Clearly, a great deal of research remains to be done, but from a practical
perspective, it nevertheless appears that AAPS (like BTA) can successfully inhibit
the corrosion of phosphor bronze wound musical instrument strings — especially
when the strings are constructed with titanium alloy cores. In addition, AAPS
can be analogously applied through a relatively simple, one-step coating process to
produce a thin film that has a minimal effect on resonant characteristics. Thus, short
of creating a "corrosion-proof string (i.e., a string constructed with galvanically
matched materials like nickel and titanium [8]), an AAPS or BTA surface treated
string could be the next best alternative.
4. CONCLUSIONS
The corrosion resistance of phosphor bronze wound musical instrument strings can
be improved through surface modification with two analogous, polymeric film-
forming compounds: benzotriazole (BTA), and N-2-aminoethyl-3-aminopropyltri-
methoxysilane (AAPS). The improvements are most noteworthy when the core of
the string is cathodic in character (i.e., titanium as opposed to steel). In addition, the
degree of corrosion protection depends on the concentration of the coating solution.
For the case of AAPS, deposition from a dilute solution leads to the formation of
a thin, protective coating that contains a high fraction of free or associated amines,
whereas deposition from a concentrated solution leads to a thicker, non-protective
coating that contains a high fraction of protonated amines. Thus, like BTA, it
appears that poly (A APS) provides improved corrosion protection when its amine
moieties are free to associate with the metal oxide surface.
Stiffness and resonant characteristics are also influenced by the presence of
poly(AAPS) coatings, but the effects are minimized at low coating weights, where
corrosion inhibition is simultaneously maximized. Thus, taken collectively, these
results provide unique insight into the common molecular level attributes thait
are required to simultaneously optimize the corrosion resistance and vibrational
characteristics of musical instrument strings.
Acknowledgements
The musical instrument strings for this study were kindly provided by Mr. Peter
Rohrbacher of Rohrbacher Technologies. In addition, FT-IR analyses were per-
formed with the help of Ms. Nadata Green, corrosion tests were performed with the
assistance of Ms. Abby Parker, and acoustic tests were performed in concert with
Starlite-MacPark Music Productions.
REFERENCES
1. E. J. Watson and P. Bauer, US Patent No. 210,172 (1878).
2. R. J. McKay and R. Worthington, Corrosion Resistance of Metals and Alloys, American
Chemical Society Monograph Series. Reinhold, New York, NY (1936).
3. P. Infield, US Patent No. 4,854,213 (1989).
4. C. B. Gray, US Patent No. 2,049,769 (1936).
5. H. C. Ralls, US Patent No. 2,892,374 (1959).
6. J. Jeans, Science and Music. Dover Publications, New York, NY (1968).
7. C. G. Hebestreit and D. J. Myers, A. Huppenthal and G. T. Bethke, US Patent No. 5,801,319
(1998).
8. A. A. Parker and P. J. Rohrbacher, US Patent No. 6,348,646 (2002).
9. A. Lazarus, US Patent No. 4,539,228 (1985).
10. J. B. Cotton and I. R. Scholes, Br. Corrosion J. 2, 1-5 (1967).
11. D. Chadwick and T. Hashemi, Corrosion Sci. 18,39-51 (1978).
12. J. C. Rubim, Chem. Phys. Lett. 167, 209-214 (1990).
13. L. J. Jha, G. Singh and G. Kaur, Trans. SAEST 26, 182-188 (1991).
14. J. F. Walsh, H. S. Dhariwal, A. Gutierrez-Sosa, R. Lindsay, G. Thornton and R. J. Oldman, Nucl.
Instrum. Methods Phys, Res. B 97, 392-396 (1995).
15. N. L. Dias Filho, Y. Gushikem, W. L. Polito, J. C. Moreira and E. O. Ehirim, Talanta 42, 1625-
1630(1995).
16. A. A. Parker and T. Janinni, to be published.
17. H. J. Hoorn, P. deJoode, W. L. Driessen and J. Reedijk, React. Funct. Polym. 27, 223-235 (1995).
18. E. P. Plueddemann, Silane Coupling Agents. Plenum Press, New York, NY (1982).
19. K. L. Mittal (Ed.), Silanes and Other Coupling Agents, VSP, Utrecht (1992).
21. E. P. Plueddemann, J. Adhesion Sci. Technol. 5, 261-277 (1991).
22. H. Ishida, Polym. Composites 5 (2), 101-123 (1984).
23. E D. Osterholtz and E. R. Pohl, J. Adhesion Sci. Technol 6, 127-149 (1992).
24. E. R. Pohl and F. D. Osterholtz, in: Molecular Characterization of Composite Interfaces,
H. Ishida and G. Kumar (Eds), pp. 157-170. Plenum Press, New York, NY (1985).
25. H. Kang, W. Meesiri and F. D. Blum, Mater. Sci. Eng. A126, 265-270 (1990).
26. L. G. Britcher, D. C. Kehoe and J. G. Matisons, Langmuir 9, 1609-1613 (1993).
27. F. Garbassi, E. Occhiello, C. Bastioli and G. Romano, J. Colloid Interface ScL 117, 258-270
(1987).
28. F. D. Blum, Ann. Rep. NMR Spectrosc. 28, 277-321 (1994).
29. C. W. Chu, D. P. Kirby and P. D. Murphy, J. Adhesion ScL Technol 1, 417-433 (1993).
30. H. Kang and F. D. Blum, J. Phys. Chem. 95, 9391-9396 (1991).
31. T. P. Huijgen, H. Angad Gaur, T. L. Weeding, L. W Jenneskens, H. E. C. Schuurs, W. G. B. Huys-
mans and W. S. Veeman, Macromolecules 23, 3063-3068 (1990).
32. T. L. Weeding, W. S. Veeman, L. W. Jenneskens, H. Angad Gaur, H. E. C. Schuurs and
W G. B. Huysmans, Macromolecules 22, 706-714 (1989).
33. L. W. Jenneskens, A. Venema, N. Van Veenendaal and W. G. B. Huysmans, J. Polym. ScL Part A:
Polym. Chem. 30, 133(1992).
34. A. A. Parker and P. M. Kolek, J. Adhesion 73, 197-214 (2000).
35. J. R. Davis (Ed.), Metals Handbook Desk Edition, 2nd edn. ASM International, Materials Park,
OH (1998).
36. N. H. Stone and A. S. Falcone, US Patent No. 4,063,674 (1977).
37. P. R. Underhill and D. L. Duquesnay in: Silanes and Other Coupling Agents, K. L. Mittal (Ed.),
Vol. 2, pp. 149-158. VSP, Utrecht (2000).
38. V. Subramanian and W. J. van Ooij, in: Silanes and Other Coupling Agents, K. L. Mittal (Ed.),
Vol. 2, pp. 159-174. VSP, Utrecht (2000).
176 A. A. Parker
39. D. H. Park, S. S. Park and S. J. Choe, Bull. Kor. Chem. Soc. 20, 291-296 (1999).
40. N. B. Colthup, L. H. Daly and S. E. Wiberley, Introduction to Infrared and Raman Spectroscopy,
3rd edn. Academic Press, New York, NY (1990).
41. J. R. Culler, S. Naviroj, H. Ishida and J. L. Koenig, /. Colloid Interface ScL 96, 69 (1983).
42. J. R. Culler, H. Ishida and J. L. Koenig, Polym. Composites 7, 231 (1986).
43. C. H. Chiang, H. Ishida and J. L. Koenig, J. Colloid Interface ScL 74, 396 (1980).
44. F. J. Boerio, C. A. Gosselin, R. G. Dillingham and J. M. Burkstrand, Proc. 15th Natl SAMPE
Technical Conference, 212 (1983).
Part 2
Other Coupling Agents /

Adhesion Promoters
Ed. K. L. Mittal
© VSP 2004
Cyclic azasilanes: volatile coupling agents

for nanotechnology
BARRY ARKLES 1*, YOULIN PAN ] , GERALD L. LARSONl

and DONALD H. BERRY 2
1
Gelest, Inc., 11 East Steel Road, Morrisville, PA 19067, USA
2
Department of Chemistry, University of Pennsylvania, Philadelphia, PA 19104, USA
Abstract—New cyclic azasilanes have been synthesized with the purpose of developing coupling
agents appropriate for a variety of nanotechnologies including surface modification of nanoparticles
and chemical vapor deposition (CVD) consistent with nanoscale features. A facile method for the
formation of Si-N bonds which comprises heating aminoalkoxysilanes in the presence of a neutral
ammonium salt catalyst has been developed. Compounds ranging from simple azasilacyclopentanes
to pentacoordinate azasilacycloctanes, as shown below, were prepared.
H
I
"1ST
R
Ni^ H' O °
,x M
R R R R
Preliminary data for the treatment of inorganic surfaces, including nanoparticles and oxidized silicon
wafers, with cyclic azasilanes suggest high density monolayer deposition by a ring-opening reaction.
Keywords: Silane; azasilacyclopentane; coupling; nanoparticles; silica; silanol.
1. INTRODUCTION
Surface modification of hydroxyl-containing surfaces, particularly inorganic sur-
faces such as nanoparticles, microelectronic and optoelectronic devices with fea-
tures less than 10 nm, poses challenges for conventional alkoxysilane coupling
agents. Monolayer deposition with a high density of functional groups is critical.
In the modification of surfaces with small or nano-scale features, it is desirable to
*To whom correspondence should be addressed. Tel.: (1-215) 547-1015, Fax: (1-215) 547-2484;
e-mail: Info@gelest.com
180 B. Arkles et al.
effect the functionalization of surface hydroxyl groups in high yield and at low tem-
peratures. Byproducts of substrate reactions with alkoxysilanes and chlorosilanes
can remain strongly adsorbed to the surface interfering with the desired functional
or chemical behavior of the modified surface. Most significantly, nano-scale feature
modification is generally preferred in an environment free of water.
Organofunctional silanes remain the most versatile chemical "platform" for the
modification of inorganic surfaces with organofunctional groups [1]. The most com-
mon organofunctional silanes have alkoxy groups. The reaction of surface hydroxyl
groups with alkoxysilanes normally requires hydrogen bonding of hydroxyl groups
with either neighboring hydroxyl groups or the addition of hydrogen-bonding addi-
tives such as amines. Alternatively, alkoxysilanes can be prehydrolyzed, but these
silanol-containing species tend to self-react and polymerize, generating their own
nano-scale domains, often large enough to bridge across features. Prehydrolyzed
silanes are not suitable for vapor phase deposition, the method of deposition for
most nano-scale applications, since they are not volatile.
The challenge of depositing monolayers with high functional group density on
nano-scale features is further exacerbated when both the absolute number and the
scarcity of reactive sites within the geometrical confines of nano-scale features is
considered. Elaborating on the earlier discussions of Iler [2] on the theoretical
concentration of surface hydroxyl groups, Figure 1 depicts a 20-nm diameter
particle (a typical dimension for pyrogenic silica) which has a surface area of
approx. 12.6 x 102 nm2. The step and repeat area (Fig. 2) for S1O2 on the surface
is approx. 0.13 nm2, giving a total population of about 10000 silicon atoms on
the surface (or about 7.8 silicon atoms/nm2). The structure of a fumed silica
nanoparticle is thought to approach the ordered structure of tridymite, with one
silicon atom extending above the mean surface plane and one below it (Fig. 3).
Assuming that only silicon atoms above the plane can be hydroxylated, the potential
Figure 1.
ai3nm2
Figure 2.
Cyclic azasilanes: volatile coupling agents for nanotechnology 181
OH OH OH
Surface
>Bulk
Figure 3.
OH
HO^T
OH
Figure 4.
population of available hydroxyls then is 5000. This calculation correlates with

reported values for rehydrated pyrogenic silica of 4.4-4.6 OH/nm2. However,
the hydroxyls are not equivalent. They include free (isolated) surface hydroxyl
groups, internal hydroxyl groups, hydroxyl groups bound to molecular water,
geminal hydroxyl groups (silanediols) and mutually hydrogen bonded hydroxyl
pairs (Fig. 4). Mutually-bonded hydroxyl pairs, which comprise 50-70% of
the silanol population, are the principal reactive species and only one member
of the pair is thought to react under uncatalyzed conditions with conventional
organosilanes. The final calculation is that most silane surface treatments result in
only 1250-2500 modified sites on the the surface of a single particle of pyrogenic
silica. Interestingly, if the radius of most organofunctional groups is considered,
only about 10% of the surface area of a modified particle actually possesses
functionality.
The intrinsic limitations in reactivity of conventional silanes stem from a number
of factors, ranging from inadequate volatility of most silanes for vapor phase
techniques, the steric restraints of bulk surfaces and, most critically, the absence
of a significant thermodynamic driver for formation of oxane bonds with substrates.
Exothermic formation of oxane bonds was considered the primary requirement for
new coupling agents in this study. Secondarily, candidates with lower molecular
weight (more volatile) and less sterically demanding structures than conventional
silanes were preferred. Cyclic azasilanes with the general structure shown below
emerged as theoretical candidates with appropriate structures to test this hypothesis
since both the formation of a silicon-oxygen bond in place of a silicon-nitrogen

bond and relief of ring strain are exothermic processes.
R
> ,
R f/ R"
The silicon-nitrogen bond energy is approx. 100 kcal/mol, compared to the silicon-
oxygen bond energy of approx. 110 kcal/mol [3]. Further, the cyclic azasilanes
could react with hydroxyl groups by a ring-opening reaction, presumably with the
loss of ring strain energy, and would not require water as a catalyst.
Cyclic azasilanes were prepared much earlier by Speier [4, 5] in a relatively
inefficient synthesis according to the following equation:
H*C
CH3
I
ClCH2CHCH2Si(CH3)2Cl + 3CH3NH2 ' N \ _ T + 2 CH3NH2+C1-
CH
7sr 3
CH3 CH3
There were few intermittent reports of their synthesis and no reports of their use for
the treatment of inorganic surfaces. One of the reasons for the lack of interest may
be that no practical, high yield syntheses have been developed for any member of
this class of compounds and no method has been reported for the synthesis of the
most volatile members of the series. While Speier demonstrated the formation of
azasilacyclopentanes, he was not able to prepare the most volatile member of the
series, namely alkoxy substituted cyclic azasilanes in which there were no methyl
substituents on the hydrocarbon portion of the ring structure. Later, Pepe [6],
using a different process, explicitly failed to form the 2,2-dimethoxy-l-aza-2-si-
lacyclopentane. Again, he noted success in those cases in which the hydrocarbon
portion of the ring structure had methyl substituents. It must be mentioned that
evaluation of Speier's work by Ziche et al [7] showed that the structure assigned to
the reaction product of chloropropyltrimethoxysilane with 1,2-diaminoethane was
incorrect and that it was in fact a pentacoordinate diazasilaoctane.
2. EXPERIMENTAL
2.7. Synthesis of cyclic azasilanes
2.LI. 2,2-dimethoxy-l,6-diaza-2-silacyclooctane. A 1-1, 3-neck flask equipped
with a magnetic stirrer, pot thermometer and short column with distillation head was
charged with 679.08 g (3 mol) of N-(2-aminoethyl)-3-aminopropyltrimethoxysilane
and 6.80 g (1 wt%) of ammonium chloride. After heating to 120-140°C for 30 min,
vacuum was gradually applied and adjusted to 10 mmHg. The head temperature
rose slowly to 85 °C. The product mixture was collected within a temperature range
of 85-105°C at 10 mmHg. At the same time, byproduct methanol that formed was
removed continuously and condensed separately in a dry-ice trap. A 580 g mixture
was generated in 12 h. White solids formed in the distillate and were separated. The
liquid portion was predominantly the unreacted starting material. The solids were
then washed with pentane and dried under vacuum for 4 h: 248 g (yield: 42.5%);
mp: 61-62°C, bp 71-73°C/2.5 mmHg. The recrystallized solids were analyzed
by NMR and X-ray diffraction. Data consistent with the proposed structure were
obtained. 'H-NMR (C6D6): 0.77 (m, 2H), 1.36 (m, 2H), 1.85 (m, 2H), 2.06 (m,
2H), 2.58 (m, 2H), 3.64 (s, 6H). The X-ray structure is provided (see Section 3).
The experiment was repeated with ammonium sulfate, ammonium trifluorometha-
nesulfonate and ammonium bromide. In all cases, identical products in similar
yields were generated.
2.7.2. 2,2-dimethoxy-N-n-butyl-l-aza-2-silacyclopentane. A 1-1, 3-neck flask

equipped with a magnetic stirrer, pot thermometer and short column with distillation
head was charged with 478.80 g (2 mol) of N-(n-butyl)-aminopropyltrimethoxysi-
lane and 4.80 g (1 wt%) of ammonium sulfate. After heating to 120-140°C for
30 min, vacuum was gradually applied and adjusted to 10 mmHg. The head
temperature rose slowly to approx. 85°C. The distillation receiver was maintained at
ambient temperature (20-24°C). Distillate was collected within a temperature range
of 85-105°C at 10 mmHg. The more volatile methanol was allowed to bypass the
receiver and was collected in a separate dry-ice trap. A total of 420 g of mixture,
which GC analysis indicated to consist of two principal components, was generated
in 12 h. The mixture was then distilled under vacuum. 2,2-dimethoxy-N-n-butyl-
l-aza-2-silacyclopentane, 183 g (44.6%), was obtained with bp: 69-71°C/3 mmHg;
!
H-NMR (C6D6): 0.56 (t, 2H), 0.90 (t, 3H), 1.28 (m, 2H), 1.75 (m, 2H), 2.85 (t,
H), 3.43 (s, 6H) (higher boiling distillate was identified as starting material). The
effective yield was greater than 70%.
2.1.3. 2-methyl-2-methoxy-l,6-diaza-2-silacyclooctane. Conditions were simi-

lar to 2,2-dimethoxy-l,6-diaza-2-silacyclooctane. A 1-1, 3-neck flask equipped with
a magnetic stirrer, pot thermometer and short column with distillation head was
charged with 412.72 g (2 mol) of N-(2-aminoethyl)-3-aminopropylmethyldimetho-
xysilane and 4.13 g (1 wt%) of ammonium chloride. After heating to 120-140°C
for 30 min, vacuum was gradually applied and adjusted to 10 mmHg. The head
temperature rose slowly to approx. 85°C. The product mixture was collected within
a temperature range of 85-105°C at 10 mmHg. A 370 g mixture was generated
in 12 h. The mixture was then subjected to fractional vacuum distillation. 122 g
(yield: 29%) of the title product, a semi-solid at room temperature, was obtained;
bp: 70-72°C/3 mmHg. ^ - N M R (C6D6): 0.21 (s, 3H), 0.72 (m, 2H), 1.76-2.55 (m,
6H), 2.67 (m, 2H), 3.54 (s, 3H).
2.1.4. 2,2-diethoxy-l-aza-2-silacyclopentane. A 1-1 3-neck flask equipped with

magnetic stirrer, pot thermometer and short column with distillation head was
charged with 442.74 g (2 mol) of aminopropyltriethoxysilane and 4.80 g (1 wt%) of
ammonium chloride. After heating to 120-140°C for 30 min, vacuum was gradually
applied and adjusted to 10 mmHg. The head temperature rose above 75°C. Product
mixture was collected within a temperature range of 85-105°C at 10 mmHg; 380 g
of mixture was generated in 12 h. The mixture was then subjected to vacuum
distillation: 14.2 g (yield: 8%) of liquids were obtained; bp: 69-71°C/2.5 mmHg.
The product was less than 90% pure. Addition of ethanol gave an exothermic
reaction and a single product, aminopropyltriethoxysilane, the starting material, was
formed.
2.1.5. 2>2-dimethoxy-N-t-butyl-l-aza-2-silacyclopentane. A 1-1, 3-neck flask

equipped with a magnetic stirrer, pot thermometer and short column with distillation
head was charged with 239.40 g (2 mol) of N-(t-butyl)aminopropyltrimethoxysilane
and 4.80 g (1 wt%) of ammonium sulfate. After heating to 100-120°C, vac-
uum was applied and gradually adjusted to 10 mmHg. The head temperature
rose slowly to 80-100°C. The distillation receiver was maintained at ambient
temperature (20-24°C). The distillate was collected within a temperature range
of 85-105°C at 10 mmHg. The more volatile methanol was allowed to bypass the
receiver and was collected in a separate dry-ice trap. The distillate was then redis-
tilled under vacuum. 205.0 g of 2,2-dimethoxy-N-t-butyl-l-aza-2-silacyclopentane
was obtained; bp: 58-60°C/3 mmHg. 'H-NMR (C6D6): 0.55 (t, 2H), 1.21 (s, 9H),
1,63 (m, 2H), 2.73 (t, 2H), 3.45 (s, 6H).
2.1.6. 2,2-dimethoxy-N-methyl-l-aza-2-silacyclopentane. Under conditions si-

milar to example (Section 2.1.2), 2,2-dimethoxy-N-methyl-l-aza-2-silacyclopenta-
ne was prepared, but in lower (14.2%) yield; bp: 48-49°C/3 mmHg; ^ - N M R
(C6D6): 0.51 (t, 2H), 1.72 (m, 2H), 2.51 (s, 3H), 2.68 (t, 2H), 3.43 (s, 6H).
2.1.7. 2,2-dimethoxy-N-allyl-l~aza-2-silacyclopentane. Under conditions simi-

lar to example (Section 2.1.2), 2,2-dimethoxy-N-allyl-l-aza-2-silacyclopentane was
prepared in 31.1% yield with bp 46^8°C/3 mmHg; ^ - N M R (C6D6): 0.54 (t, 2H),
1.70 (m, 2H), 2.72 (t, 2H), 3.45 (s, 6H), 5.06 (d, 3H), 5.84 (m, 2H).
2.1.8. N-aminoethyl-aza-2,2,4-trimethylsilacyclopentane. Under conditions si-

milar to those used by Speier [4, 5], 3-chloroisobutyl-dimethylchlorosilane was re-
acted with ethylenediamine to produce N-aminoethyl-aza-2,2,4-trimethylsilacyclo-
pentane in 54% yield, bp 54-56°C/2 mmHg; density: 0.905 g/m3.
22. Preliminary deposition efficiency screening

2.2.1. Nanoparticle method — mixed liquid vapor deposition. A tared 100-ml
flask was charged with approx. 2 g of carefuly weighed pyrogenic fumed silica with
a nominal surface area of 200 m2/g (Aerosil 200). The flask was slowly evacuated to
<0.1 mmHg and then heated to 200°C for 2 h (or to constant weight). The flask was
returned to room temperature. The vacuum was broken with approx. 1 g of silane.
The mixture was shaken vigorously by hand for 100 s and then re-evacuated and
heated to 180°C for 1 h, and the weight increase was recorded. Silanes evaluated
were n-butyltrimethoxysilane, N-n-butylaminopropyltrimethoxysilane, and N-n-
butyl-aza-dimethoxysilacyclopentane which gave weight gains of 6.4%, 22.5%
and 38.0%, respectively.
2.2.2. Substrate reactivity by spin-on deposition. Oxidized silicon wafers were

treated with a 1:1 mixture of 50% aqueous sulfuric acid and 30% hydrogen
peroxide for 30 min and then rinsed with deionized water and dried at 110°C.
Silanes were prepared in 5 wt% concentrations in diglyme and then applied to
the wafers at 2000 rpm. Ellipsometric thickness was then calculated. For N-n-
butylaminopropyl-trimethoxysilane and N-n-butyl-aza-dimethoxysilacyclopentane
the thicknesses were 110 nm and 130 nm, respectively.

In the course of these investigations it was discovered that volatile cyclic azasilanes
with or without alkoxy substituents on the silicon could be produced from alkoxysi-
lanes by treating them with simple ammonium salts, preferably ammonium sulfate,
removing the alcohol as it formed. A mixture containing the cyclic azasilane, poly-
meric azasilanes and alkoxysilane starting materials initially forms. The pure cyclic
azasilane can be distilled in high yield from the equilibrating mixture, indicated
below for the example of N-n-butyl-aza-2,2-dimethoxysilacyclopentane.
(NH 4 ) 2 S0 4
CH 3CH 2CH 2CH 2 NHCH 2CH 2CH 2Si(OCH 3 ) 3 N + CH3OH
120-140°C ^ s i ^ CH 2 CH 2 CH 2 CH 3
C H 3 o ' S OCH 3
H oC "~0 CH oCH ^CH 9CH o

I I
CH3 CH 2 CH 2 CH 2NHCH 2 CH2 CH2 Si -NCH 2CH 2CH 2Si(OCH 3 )3
H3C-0
dimeric and polymerized products
The ring-opening reaction of the cyclic azasilanes with the appropriate alcohol in
the absence of a catalyst proceeds rapidly and quantitatively to form the starting
material.
Cyclic azasilanes of several different general structures were synthesized by this
method. Azasilacyclopentanes without substituents on the hydrocarbon portion
of the ring can be prepared with or without alkoxy substitution. They have the
following representative structures:
,N N,
*R \ ; /^R
:si X
:sic:
/ ^R
R'O' OR" R'O / R" R' / R"
Specific examples of compounds produced include 2,2-dimethoxy-N-n-butyl-l-aza-

2-silacyclopentane and 2,2-diethoxy-N-(2-aminoethyl)-l-aza-2-silacyclopentane.
The method yields silicon compounds with the following typical structure from
the ring closure of 3-(2-aminoethyl)aminopropylsilanes when there are two alkoxy
groups bound to silicon:
H
I
"1ST
NT/ \
H O °x
/
R R
A specific example of a compound produced is 2,2-dimethoxy-l,6-diaza-2-silacy-
clooctane. This particular compound shows strong coordination of one nitrogen
with silicon and may be regarded as a bicyclic compound with a pentacoordinate
silicon. The pure pentacoordinate compounds are generally low melting crystalline
solids. The X-ray structure of 2,2-dimethoxy-l,6-diaza-2-silacyclooctane is de-
picted in Fig. 5.
C14
Figure 5.
Table 1.
Properties of azasilacyclopentanes
Compound Yield Bp Density

(g/cm 3 )
47% 6 9 - 7 l ° C / 3 mmHg 0.941
nBu
A
MeO OMe
42% 58-60° C/3 mmHg 0.932
tBu
/S's
MeO OMe
40% 52-54°C/3 mmHg 0.938
X i ^ CH 2 CH=CH 2
MeO OMe
19% 48-49° C/3 mmHg 1.008
v
\ Q - ^ Me
/ N
MeO OMe
5-7% Not isolated
EtO
A OEt
0% Not observed
/ \
MeO OMe
CH^ 54-56° C/3 mmHg 0.905
A
Me Me
New compound prepared by method of Speier in 54% yield; all others by ring closure with loss of
alcohol.
When there are less than two alkoxy substituents, the azasilacyclopentane is
preferred. At room temperature these compounds are usually liquids. The cyclic
azasilanes prepared in this study are summarized in Tables 1 and 2.
Prior to studying the comparative reactions of the azasilacyclopentane and alko-
xysilanes with silica, a model compound for isolated hydroxyl groups was exam-
B. Arkles et al.
Table 2.
Properties of diazasilacyclooctanes
Compound Yield Bp Mp
29% 70-72° C/3 mmHg Semi-solid
H Me OMe
H 42% 71-73°C/2.5mmHg 61-62°C

I
MeO OMe
ined. The hydroxyl group of triethylsilanol can be likened to an isolated hydroxyl

group on a surface. The model reaction, shown below, can be performed and fol-
lowed in homogeneous liquid phase.
CH,
A ^Si-OH
I
CH0
I "
CH^CH 2 —Si-OH
CH2
CH3
The reactions of N-n-butylaminopropyltrimethoxysilane and its cyclic analog,

2,2-dimethoxy-N-n-butyl-l-aza-2-silacyclopentane, were compared in room tem-
perature reactions with molar equivalents of triethylsilanol. The alkoxysilane (be-
low) showed less than 1 % reaction over 24 h.
C2H5
C2H5—Si-OH + CH3CH2CH2CH2NHCH2CH2CH2Si(OCH3)3 -7^- no reaction
C2H5
The reaction of the cyclic azasilane was quantitative in less than 5 min and
demonstrated a strong exotherm according to the equation shown below.
.c
C,HS
r C 9 H^ OCH,
QHc—Si-OH
i
r
C 2 H S — S i • 0-Si-CH 2 CH 2 CH 2 NCH 2 CH 2 CH 2 CH 3
C H O ' *OCH^
C2H5 OCH3
The cyclic l-aza-2-silanes rapidly react with a variety of hydroxy lie substrates,
particularly siliceous and inorganic structures in vapor, liquid or solution state
without the formation of byproducts. Depicted below is the ring-opening deposition
of an N-alkyl-2,2-dimethylaza-2-silacyclopentane.
CH 2
CH^
\ /
N—srCH3
N
CH 3 CH 2 CH 2 CH/ CH3
OH OH
Substrate
H H
I I
NCH 2 CH 2 CH 2 CH 3 NCH 2 CH 2 CH 2 CH 3
I
CH 9 CH 2
I " I
CH 2 CH 2
I I
CH 2 CH 2
I
CH 3 —Si- CH, CH 3 —Si- -CH,
I
O
L Substrate
When dried pyrogenic (fumed) silica is treated with cyclic azasilanes, a strong
exotherm is observed while no exotherm is observed for the alkoxysilane.
When alkyl groups (typically methyl) are substituted on the silicon, a crosslinked
film will not form. In many applications these monolayers are sufficiently robust. If
the substitutions on the silicon are alkoxy groups (typically methoxy), the deposition
still leads to monolayers, but subsequent hydrolytic condensation of the monolayer
after the excess (unreacted) silane is removed from the substrate results in formation
of a more durable monolayer as shown below (see Scheme 1).
The extent of reaction of silanes with hydroxylic substrates, i.e. the effectiveness
in reacting with the different types of hydroxyl groups, can be measured by a
number of different techniques. The earlier discussion on silica nanoparticles can
be applied directly to commercial pyrogenic silicas. A typical commercial grade
has a surface area of 325 m2/g or 3.25 x 1020 nm2/g. If the hypothetical number of
4.5 hydroxyls/nm2 (i.e., 1 m2 of silica contains 7.5 /xM of hydroxyl) is accepted,
then one gram of silica contains 1.5 x 1021 hydroxyls or 1.5 x 1021/6 x 1023
or 2.4 x 10~3 mol. Monolayer bonding of a silane with a molecular weight
of 200 would deposit 0.5 g silane per gram of silica. In fact, most monolayer
depositions of silanes result in less than 0.05 g per gram of silica.
An interesting practical experiment is to measure the weight increase of dried
fumed silica after treatment with silanes followed by vacuum devolatization af-
ter a 100-s silane exposure. While more accurate and representational data can
be obtained by IR [8] or solid state 29Si-NMR [9] studies, simple weight gain
can provide a quick screening method for effectiveness of silane surface modifi-
cation. A series of n-butyl functional silanes were selected for the study. The
n-butyltrimethoxysilane is comparable to typical coupling agents. N-(n-butyl)-
H H
I I
CH 2 CH 2
I
CH, CH 2
I *"
CH 2 CH 2
I
CH30-Si-OCH3 CH 3 0—Si—OCH 3
I I
O O
Substrate
- CH3OH
H H
I
CHo CH 2
CH, CH 2
I - I
CH 2 CH 2
I I
)—Si — -Si — -o-
I
O O
I I
Substrate
Scheme 1.
aminopropyltrimethoxysilane contains an amine which is expected to have a cat-

alytic effect in promoting surface reaction [10]. Finally, 2,2-dimethoxy-N-n-butyl-
l-aza-2-silacyclopentane, the cyclic analog, was evaluated. The ratio of weight gain
was 1:3.5 :5.9. This clearly indicates that, although there is a significant catalytic
effect by the amine, the surface reaction is driven even further by the ring-opening
reaction.
Modification of fillers with cyclic azasilanes without hydrolyzable groups has
recently been reported [11]. However, solution deposition conditions utilized
were such that adsorbed silanes probably reacted after formal contact times.
Nevertheless, the one example of a cyclic azasilane demonstrated higher bonding
efficiency than a non-analogous linear silane at ratios similar to the screening
experiment reported here.
Spin-on deposition of 2,2-dimethoxy-N-n-butyl-l-aza-2-silacyclopentane on oxi-
dized silicon wafers provided about 20% greater deposition than the linear analog
as measured by ellipsometry, a result comparable to the vapor phase deposition.
4. CONCLUSIONS
New cyclic azasilanes have been synthesized for the purpose of developing coupling
agents appropriate for a variety of nanotechnologies including surface modification
of nanoparticles and chemical vapor deposition (CVD) consistent with nanoscale
features. Cyclic azasilanes appear to be ideal candidates for these applications since
they undergo a ring-opening reaction with hydroxyl groups driven thermodynami-
cally by the formation of an oxane bond with silicon without byproduct formation.
Preliminary data for the treatment of inorganic surfaces, including silica nanopar-
ticles and oxidized silicon wafers, with cyclic azasilanes suggest that high density
monolayer deposition is achieved.
Acknowledgements
D.H.B. thanks the National Science Foundation for partial support of this work
under the MRES program at the University of Pennsylvania.
REFERENCES
2. R. Her, The Chemistry of Silica. Wiley, New York, NY (1979).
3. R. Walsh, in: Silicon, Germanium and Tin Compounds, B, Arkles (Ed.), p. 96. Gelest,
Morrisville,PA(1998).
4. J. Speier, US Patent 3,146,250 (1964).
5. J. Speier, J. Org. Chem. 36, 3120 (1971).
6. H. Pepe, US Patent 5,354,880 (1992).
7. W. Ziche, B. Ziemer, P. Hohn, J. Weiss and N. Auner, J. Organomet Chem. 521, 29 (1996).
8. C. Armistead, A. Tyler, F. Hambleton, S. Mitchell and J. Hockey, /. Phys. Chem. 73, 3947
(1969).
9. G. Maciel and D. Sindorf, J. Am. Chem. Soc. 102, 7607 (1980).
10. S. Kanan, W. Tze and C. Tripp, Langmuir 18, 6623 (2002).
11. M. Vendamuthu, S. Painter, J. Ancheta and J. Blitz, J. Undergrad. Chem. Res. 1, 5 (2002).
Ed. K. L. Mittal
© VSP 2004
Inorganic-organic hybrid polymers based on silanes

for coating textile substrates
TORSTEN TEXTOR*, DIERK KNITTEL, THOMAS BAHNERS

and ECKHARD SCHOLLMEYER
Deutsches Textilforschungszentrum Nord-West eM (DTNW), Adlerstr 1, D-47798 Krefeld, Germany
Abstract—Approaches for coating textile materials with inorganic-organic hybrid polymers were
investigated to introduce specific properties to the substrates. The present investigations deal with
the finishing of technical textiles, which might, e.g., be used for architectural, automotive or medical
applications. The coatings reported aim at improving of the wear-resistance of glass fiber material
and creating thin barrier coatings to protect different fiber materials in aggressive environments.
The results show that UV-absorbing coatings that prevent photochemical decomposition, coatings
changing color when irradiated with UV-light or coatings for bullet-proof vests, that provide stab-
resistance, are possible.
Keywords: Wear resistance; water repellence; stab resistance; barrier coating; UV protection.
1. INTRODUCTION
Inorganic-organic hybrid polymers have been an active research area in the last
decade. New or improved products have been developed in different fields of ap-
plication. Investigations were carried out dealing with scratch resistant coatings
for polymer surfaces based on hybrid polymers filled with nano-particles, protect-
ing, e.g., glasses or compact discs [1-3]. Barrier films, e.g., made of polypropy-
lene or poly(ethylene terephthalate), used as packaging material coated with certain
hybrids, have been reported that seal up flavors or hydrocarbons [4]. Other in-
vestigations present coatings that protect against UV radiation [5]. Coatings with
hybrid polymers which are modified with minor amounts of perfluorinated com-
pounds were introduced to create hard and hydrophobic and oleophobic surfaces
yielding excellent water and oil repellencies [6, 7]. Promising studies have been
carried out at the DTNW dealing with the modification of fiber materials by coat-
ing the textile with inorganic-organic hybrid polymers using the sol-gel technique.
e-mail: textor@dtnw.de
194 T. Textor et al.
The focus of these ongoing investigations is on technical textiles, i.e., textiles used
for architectural applications, as filter materials, for ballistic protection, for medical
applications or in the automotive sector. The purpose of coating textiles with the
hybrids mentioned was to provide specific surface properties especially hydropho-
bic, oleophobic and hydrophilic properties. By coating fabrics made of polyester,
polyamides, glass fiber material and also natural fibers with sols based on organi-
cally modified trialkoxysilanes, excellent water- and oil-repellency can be achieved.
Compared to common textile finishing strategies, finishing with inorganic-organic
hybrid polymers allows a combination of a variety of properties in one coating ma-
terial, which means a number of properties can be improved in one finishing step.
By an appropriate composition of the sol, coatings can be achieved that improve
both water and oil repellency, offer excellent barrier properties to protect the fiber
material in aggressive basic or acidic atmospheres and simultaneously improve the
wear-resistance. Incorporation of nano-particles broadens the range of properties or
improvements which can be obtained without losing the transparency of the coating
material, which is very important for many textile applications. Modification with
particles that show high absorption in the UV-region can be employed, e.g., to pre-
vent decomposition due to photochemical reactions induced by sunlight. Coatings
containing aluminium oxide particles lead to improved wear resistance for sensitive
materials such as glass fiber fabrics, which are very susceptible to abrasive stresses.
A large number of new applications for hybrids filled with nano-particles are con-
ceivable [8-16].
The fact that coating by the sol-gel-technique can be carried out yielding very
thin layers guarantees only a slight increase in weight which is important, e.g., for
architectural applications where light weight is much desired. The actual coating
process can be carried out with a comparatively low technical effort, i.e., by simple
dipping or padding (dipping followed by squeezing between two rollers) processes
which are common techniques in this branch of industry. The only drawback is
the fact that sols mostly contain alcohol, so the textile machinery needs exhaust
systems.
2. EXPERIMENTAL
The textile materials (fabrics made of polyamide 6, poly(ethylene terephthalate),
p-aramide, wool and glass fiber) used for the present investigations were technical
materials and were used as received. Solvents and chemicals were of reagent grade
and were used as received.
2.1. Preparation of sols

All sols described here were based on 3-glycidoxypropyltrimethoxysilane
(GPTMS). 10 ml GPTMS were prehydrolyzed with 1.222 ml of 0.01 M hydrochlo-
ric acid and stirred for at least 1 h, and the resulting sol was used for further prepa-
rations. An amount of 2.34 g Bisphenol A was added as a cross-linking agent. The
sol was diluted with about 30 ml ethanol.
Silane-based inorganic-organic hybrid polymers for coating textiles 195
2.2. Preparation of coatings

Barrier coatings for protection against aggressive environments were prepared by
additionally dissolving 1 vol% of Zonyl FSA® (as a hydrophobic component) into
the sol. Wear-resistant coatings were prepared by dispersing 10 wt% of aluminium
oxide particles (Degussa C, average primary particle size 13 nm) into the GPTMS
sol by intense stirring and ultrasonic treatment. UV-absorbing coatings were pre-
pared either by dispersing 10 wt% ZnO-particles (received from Sachtleben, av-
erage particle size 5-10 nm) into the sol with a bead mill or by adding a freshly
prepared Ti02-sol yielding an amount of also 10 wt% Ti02 in the resulting coat-
ing material. The sol for the UV-spectrum was prepared without bisphenol A.
The Ti02-sol was prepared by hydrolysis of 2.6 ml tetraethyl orthotitanate diluted
in 20 ml isopropanol with 0.9 ml hydrochloric acid (0.01 M). The highly repel-
lent coatings were based on the GPTMS sol modified by dispersing aluminium
oxide as described before and mixing with 1 vol% of tridecafluoro-l,l,2,2-tetra-
hydrooctyltriethoxysilane. Photochromic coatings were prepared with an additional
amount of 0.76 g bisphenol A and 0.19 g of the photochromic dyestuff 1,3-dihydro-
l,3,3-trimethylspiro[2//-indole-2,3/-[^//]naphth[2,l-Z?][l,4]oxazine] (receivedfrom
Aldrich). Coatings for stab-resistant p-aramide fabrics were prepared by adding an
Al 2 0 3 -sol yielding 10 wt% A1203 (relative to GPTMS). The aluminium oxide sol
was prepared by mixing 4.08 g aluminium triisopropylate dissolved in 10 ml iso-
propanol with 4.00 g acetylacetone also dissolved in 10 ml isopropanol. The mix-
ture was slowly hydrolyzed with 0.24 ml of water before the pH was adjusted to 4.5
with nitric acid.
Before coating the fabrics, all sols were diluted to a volume of 30 ml ethanol
and 5 mol% 1-methylimidazole as a cross-linking catalyst was added. Coating was
carried out with a lab-padder (dipping followed by squeezing between two rollers)
and then the fabrics were fixed on a needle frame and dried in an oven at 130°C for
about 1 h (photochromic coatings were dried at 100°C because of the temperature
sensitive dyestuff).
The tensile strength of the textile materials was tested by tearing 5 cm broad fabric
strips following DIN 53857-T1 until they broke. Wear-resistance tests followed the
Martindale test (DIN 53863-T4), in which a standard material is scrubbed over the
sample surface with a load of 12 kPa. The UV treatment of textiles to investigate
the UV stability was carried out in a UV-reactor (Rayonet Photochemical Chamber
Reactor, RPR-100). Investigation methods for bullet-proof vests (stab-resistance
and ballistic test) followed the technical instructions of the German and British
police.
3. TEXTILE APPLICATIONS
3.1. Barrier properties
In the building industry, concrete is usually reinforced by embedding steel compo-
nents, but to prevent corrosion, the steel has to be covered by a certain thickness
196 T.Textor etal
Figure 1. SEM micrographs of uncoated (top) and coated (bottom) polyamide fabrics with different
magnifications (left and right).
of, at least a few centimeters, concrete, which limits the application and shaping
of components. Recently, concrete has been more and more reinforced with glass
fiber fabrics, which allows to produce lighter components. During the setting —
which takes several weeks to be completed — of the concrete these fabrics have
to be protected against the hydrolytic decomposition by the basic concrete (pH of
about 14!).
Very thin coatings with hybrid polymers can act as barriers to guarantee an
effective protection of the fiber material in aggressive environments. In the
following, results are shown for technical polyamide (PA) and glass fiber fabrics
that were coated with different coatings based on the epoxy-modified alkoxysilane,
GPTMS. The basic sols were modified with a hydrophobic component (Zonyl
FSA®), nano-sized aluminium oxide particles (Degussa C) and organic network
modifiers, e.g., bisphenol A. Figure 1 shows scanning electron micrographs of an
uncoated and a coated PA fabric. The coating mainly covers single fibers and has a
thickness of only a few micrometers.
Such coatings showed excellent results in tests used to investigate the barrier
properties. For these tests, textile samples were stored in aggressive environments
for a certain time. Due to the decomposition of the fiber material in these
ammonia concrete sulfurous acid untreated
Figure 2. Tensile strength of uncoated (front) and coated poly amide fabrics (back) after storage in
aggressive environments.
environments the tensile strength of the uncoated fabrics decreases. The barrier
coatings protect the fibers against the decomposition; therefore, the remaining
tensile strength was taken as a measure of the barrier quality. Three environments
were chosen: an acidic one, storing the samples for 6 days at 60°C above a
concentrated sulfurous acid (with 5-6% free SO2); a basic one, storing over
concentrated ammonia solution, for 24 h at 60°C. Additionally, the tests were
carried out at 60° C for 6 days in a concentrated concrete solution to simulate the
conditions in the building sector.
The results of these tests are shown in Figs 2 and 3. It can be clearly seen that the
coated samples do not show even a slight decrease of the tensile strength, due to the
storage in aggressive environments, whereas the uncoated glass fiber shows a strong
decrease of its strength. Further investigations showed that the tensile strength of
the tested glass fiber fabric showed a fast decay in the first 72 h and a decrease of
about 80% within two weeks, while the coated fabric did not show any decrease
during the same time period.
5.2. Wear resistance

Technical textiles, especially fabrics made of glass fiber, show an extremely poor
wear resistance due to the fragile fiber material. Coatings based on filled hybrids,
e.g., with aluminum oxide nano-particles are known to improve the resistance
against abrasive forces [5]. Comparable coatings were used to improve the wear
resistance of technical glass fiber fabrics. The left sample in Fig. 4 shows a
fabric that was completely destroyed in an abrasion test. This test is the so-called
ammonia concrete sulfurous acid untreated
Figure 3. Tensile strength of uncoated (front) and coated glass fiber fabrics (back) after storage in
aggressive environments.
Figure 4. Photographs taken after the abrasion test from an uncoated (left) and a coated (right) glass
fiber fabrics.
Martindale Abrasion Test, where samples are scrubbed over another fabric for a
certain number of cycles. The uncoated sample was totally destroyed after about
100 cycles of the abrasion test.
The same test was performed with the sample, on the right in Fig. 4, which was
coated with a hybrid polymer filled with nano-particles. Even after 10000 cycles
the sample was not destroyed.
33. UV protection
By introducing other nano-sized particles such as ZnO or TiC>2 into the network the
absorption of UV radiation can be increased to protect the fiber material itself or
to create UV-protecting textiles. The incorporation of particles of less than 50 nm
size retains the transparency of the coatings, because no refraction of visible light
occurs. An advantage of the UV protection with the above-mentioned inorganic
oxide particles is that they are non-toxic (ZnO, e.g., is used for suncream) and
more stable compared to many organic compounds that are used as UV-absorbers.
Figure 5 shows, for example, the UV-spectra of a polyethylene film coated with a
ZnO-filled and an unfilled sol. The coating with ZnO particles strongly absorbs in
the UV-A, -B and -C-regions.
Coatings filled with different nano-sized particles were applied to p-aramid fabrics
which are know to be sensitive to UV radiation. Such materials are used because
of their high tensile strength, which makes them useful for ballistic applications as
bullet-proof vests. Figure 6 shows, however, that storage in a UV-reactor for about
6 h leads to a strong decrease of the tensile strength of more than 80%. The results
for different samples coated with different thin hybrid layers are also depicted in
PE film
PE film coated with GPTMS-Sof
PE film coated with ZnOfilledGPTMS-Sol
"—I 1 1 1 i 1 1 1 1 r
250 300 350 400 450 500 550 600 650 700
Wavelength [nm]
Figure 5. UV-spectra of PE film uncoated and coated with an unfilled sol and a sol filled with ZnO
particles.
200 T, Textor et al.
uncoated ZnO TiC^
Figure 6. Tensile strength of differently coated p-aramid fabrics before and after UV exposure for
6h.
the same figure, and it can be clearly seen that the decomposition is significantly
slowed down. The decrease in tensile strength for coated fabrics is merely between 2
and 9%.
3.4. Oil- and water-repellency

Many textiles have a hydrophobic finish in order to achieve water- and/or oil-
repellency. A coating based on one of the hybrid polymers made by the sol-gel
technique can be modified with certain hydrophobic alkoxysilanes that contain
hydrocarbon chains or with highly fluorinated groups yielding surfaces with low
surface energies, giving the fabrics excellent repellency. The basic wool fiber, for
instance, is very hydrophobic due to wool waxes. By dry cleaning, these waxes
are removed with a corresponding loss in repellency. By coating afterwards with
a perfluoro-modified sol, however, the water-repellency can be achieved to an even
higher level than before, as demonstrated in Fig. 7 (the amount of the perfluoro
component was about 1 vol%). Polyester- or polyacrylic fabrics coated with such
sols showed no water penetration and achieved the highest oil repellency in the
AATCC-118-1972 test.
3.5. Host matrices

The hybrid polymer matrices may act as hosts for a variety of materials. The
networks can, e.g., be filled with a dye to dye the coating, as well as the coated
textile. Properly incorporated substances have an extremely high washing fastness
and are as permanent as the coating itself. If a photochromic dye is incorporated in
the network and has sufficient free volume in the cavity to change its molecular
conformation due to UV-irradiation, coatings can be achieved that show the
Figure 7. Water-repellency of a coated (left) and an uncoated (right) wool sample.
Figure 8. PET fabric finished with a photochromic coating.
photochromic effect [17-19]. Figure 8 shows an example of such a coated polyester

fabric which was irradiated with UV light shortly before. During the irradiation it
was covered with the letters "DTNW". A few minutes after the picture was taken
the sample was colorless again.
3.6. Ballistic protection bodywear

Bullet-proof vests can stop bullets from different portable firearms, but unfortu-
nately the protection does not include attacks with sharp weapons. An ongoing
project in our Institute is concerned with the improvement of the stab-resistance
of p-aramide-based ballistic protection materials by coating with Al203-particle-
filled hybrid polymers. The results of this work are very promising, since bundles
of the coated fabrics passed a certain stab-resistance test (Technische Richtlinie
Schutzweste mit integriertem Stichschutz, German police instruction), without los-
ing the ballistic properties of the material. The stab-resistance test is performed
with a knife missie falling down on a bundle of fabrics, fixed above a big block of
plastilina (here the energy before hitting the fabrics was 25 N • m). The plastilina
is used to simulate the human body: the advantage of this setup is that the plas-
tilina is inelastic so the penetration into the body by deformation can be measured
after the test. Passing the stab-resistance test means that the blade penetrates less
than 20 mm into the plastilina after piercing the fabrics. In addition, the dent in the
platilina caused by the stab, simulating the deformation of the human body during
the stab, has to be less than 20 mm. Exceeding one of these limits might cause
lethal injuries. The results shown in Table 1 are from a typical bundle of 30 lay-
ers of p-aramide fabric as is typically used for commercial bullet-proof vests (20 of
these layers were coated with a hybrid polymer). This bundle achieved the required
values, whereas an unmodified bundle failed the test.
The same bundle was additionally tested in a ballistic test to show the retention of
ballistic protection after coating the material with the hybrid polymer. The pictures
in Fig. 9 show the bundle of modified fabrics after passing the ballistic test.
Table 1.
Results achieved in a stab-resistance test
Bundle Penetration of the blade Dent

(mm) (mm)
20 layers of coated p-aramide, 0.5 8.8
10 layers of uncoated p-aramide
Figure 9. Bundle of p-aramide fabric after shooting with a 9 mm Parabellum, the bullet was stopped
between layers 9 and 10 (the picture on the left shows the complete bundle after the test, the right one
shows layer number 10 after removal of the first 9 layers).
4. CONCLUSIONS
Coatings based on inorganic-organic hybrid polymers have been intensively inves-
tigated in the last years and a number of commercial applications have already
been found. However, the modification of textile material has been excepted so far.
Promising approaches for far-reaching possibilities leading to creative surface de-
sign, not only with regard to an improved water- and oil-repellency or a better wear
resistance, were found in our investigations. A number of interesting properties are
possible by suitable modifications of the coating materials and several of them can
be combined in a single coating material. Examples presented were sun protection,
self-adapting colorations, resistance to extreme environments, or improved stab re-
sistance for ballistic body wear. Further applications are possible for medical textiles
as transdermal therapeutical systems, or textiles with magnetic properties.
Acknowledgements
The authors wish to thank the Forschungskuratorium Textil e.V. for their financial
support for these projects (AiF-No. 10954N, 12000N, 12882N). This support
is granted from resources of the Bundesministerium fiir Wirtschaft und Arbeit
(BMWA) via a supplementary contribution by the Arbeitsgemeinschaft Industrieller
Forschungsvereinigungen "Otto-von-Guericke" e.V. (AiF).
REFERENCES
1. F. Bauer, V. Sauerland, H.-J. Glasel, H. Ernst, M. Findeisen, E. Hartmann, H. Langguth,
B. Marquardt and R. Mehnert, Macromol. Mater. Eng. 287, 546-552 (2002).
2. H. Schmidt, H. Scholze and G. Tiinker, J. Non-Cryst. Solids 89, 557 (1989).
3. H. Schmidt and H. Wolter, J. Non-Cryst. Solids 121, 428-435 (1990).
4. S. Amberg-Schwab, M. Hoffmann and H. Bader, Kunststoffe 86, 660-664 (1996).
5. M. Saito,/. Coated Fabrics 23, 150-184 (1993).
6. H. Schmidt, J. Non-Cryst Solids 178, 302-312 (1994).
7. C. Roscher and M. Popall, Mater. Res. Soc. Symp. Proc. 435, 547-552 (1996).
8. H. K. Schmidt, Makromol. Symp. 101, 333-342 (1996).
9. J. M. Yang, H. S. Chen, Y. G. Hsu and W. Wang, Angew. Makromol. Chemie 251, 49-72 (1997).
10. R. M. Laine (Ed.), Inorganic and Organometallic Polymers with Special Properties, proceedings
of the NATO Advanced Research Workshop on Inorganic and Organometallic Polymers with
Special Properties, Cap d'Agde, France, pp. 297-317 (1990).
11. T. Iwamoto and J. D. Mackenzie, J. Mater. ScL 30, 2566-2570 (1995).
12. D. Knittel, T. Textor, Th. Bahners and E. Schollmeyer, Proc. UMIST-Conference on Textiles
Engineered for Performance, pp. 1-13. Manchester (1998).
13. T. Textor, T. Bahners and E. Schollmeyer, Melliand Textilber. 80, 847-848 (1999).
14. T. Textor, T. Bahners and E. Schollmeyer, Technische Textilien 44, 304-306 (2001).
15. T. Textor, T. Bahners and E. Schollmeyer, Progr. Colloid Polym. ScL 117, 76-79 (2001).
16. T. Textor, T. Bahners and E. Schollmeyer, Technische Textilien 45, 169-172 (2002).
17. B. Hoffmann, M. Mennig and H. Schmidt, Proceedings of XVIIInternational Congress on Glass,
Peking, Vol. 4, pp. 399-404 (1995).
18. M. Mennig, K. Fries and H. Schmidt, Mater. Res. Soc. Symp. Proc. 576, 409-414 (1999).
19. L. Hou, M. Mennig and H. Schmidt, Proc. SPIE 2255, 26-37 (1994).
Ed. K. L. Mittal
© VSP 2004
Optimising the adhesion of glass fibres to an epoxy resin

using plasma copolymers
D. J. MARKS and R R. JONES *

Department of Engineering Materials, University of Sheffield, Sir Robert Hadfield Building,
Mappin Street, Sheffield SI 3JD, UK
Abstract—Plasma copolymers from acrylic acid and allylamine with 1,7-octadiene have been used
to deposit conformal functional coatings of approximately 0.01 /xm thickness onto E-glass fibre tows
in a semi-continuous process. The adhesion of the fibres to an epoxy resin was studied using the
fragmentation test on randomly selected fibres. The optimum concentration of the functional groups
(amine or carboxylic acid) required for adhesion has been identified. The stress transfer efficiency
has been calculated from the fragment length and debond distributions. This methodology provides
a method for determining the ineffective length for the individual combinations of fibre and resin.
This new method of quantifying fibre-matrix adhesion would appear to be more discriminating than
previous models. The plasma polymers are also effective in protecting the fibres from damage. The
Weibull modulus for the distribution of fibre strengths was found to increase after plasma coating.
Furthermore, the chemistry of the coating had a major impact with the amine functionality exhibiting
the highest Weibull modulus.
Keywords: Glass fibres; functional plasma polymers; adhesion; stress transfer efficiency; fragmenta-
tion test; fragmentation analysis.
1. INTRODUCTION
The durability of a fibre composite is strongly dependent on an optimum degree
of adhesion between the matrix resin and the reinforcing fibre. Traditionally for
carbon fibres, surface oxidation is carried out to provide a degree of adhesion
between the resin and the fibre which provides for a small degree of debonding
when a fibre fractures. In the case of glass fibres, it is typical to add a silane
coupling agent to the film-forming resin in order to promote adhesion of the resin
to the fibre. Therefore, the probability of interphase formation is high and a yield
front, rather than debonding, is often observed. The sizing or finish applied to
*To whom correspondence should be addressed. Tel: (44-114) 222-5477; e-mail: f.r.jones@
sheffield.ac.uk
206 D. J. Marks and F. R. Jones
fibres is also required to protect them from damage during processing. For aramid
fibres, a more complex coating technology is often used. This has the additional
aims of improving the mechanical integrity of the fibre and modifying the moisture
absorption characteristics.
There is a requirement for a future sizing technology for reinforcing fibres such
as carbon, aramid, glass or other high performance fibres. This technology needs
to be environmentally clean but at the same time needs to be able to provide
a functionalised conformal coating with chemistry that can be matched to the
chemistry of the resin. The degree of adhesion must be controllable to provide
the requisite interfacial micromechanics for an optimum composite. In order to
molecularly engineer the fibre surface, an important aspect must be to conceal
any inherent surface chemistry and microstructure and at the same time provide a
functional group for reaction with the selected matrix. It is also important to be able
to vary the chemistry of the coating in such a way that adhesion to different matrices
can be readily switched. A simple gaseous coating technique which has the potential
for varying the chemistry of the deposit is an important way forward. One such
technique for applying conformal polymeric coatings with retained functionality is
plasma polymerisation. Plasma polymers do not have a regular repeat unit but can
retain the functionality of the monomer when low power or pulsed radio frequency
plasmas are utilised. The deposit will be a pin-hole free film which is crosslinked
and, therefore, mechanically stable with the retention of the chemistry associated
with the chosen monomer. The concentration of the functionality on the surface
of the reinforcement can be tuned simply by incorporating an inert comonomer.
The comonomer can also provide a mechanism for additional crosslinking and
hence film stability. In this way, the chemistry of the surface can be readily tuned
to meet the needs of the matrix into which the fibre will be embedded. In a
previous paper [1], we have demonstrated that the fibres can be uniformly coated
continuously. We have also shown in earlier publications that the chemistry of the
plasma polymer can be used to vary the degree of adhesion [2-4]. In this paper, we
have used allylamine and acrylic acid plasma copolymers to modify the adhesion
of E-glass fibres to an anhydride cured epoxy resin. The degree of adhesion has
been assessed using the single filament fragmentation test and the data have been
analysed using the stress transfer efficiency methodology which has been described
elsewhere [2],
2. EXPERIMENTAL
E-glass fibres without coupling agent or sizing were used in this work. They had
a diameter of 15.46 ± 1.74 /xm. The fibres were received in tow form and were
spread out using a blown air source and wound onto glass spools. These spools
with wound fibres were placed in the plasma reactor. This procedure avoids the
non-uniform deposition of the plasma polymer throughout the fibre bundle. The
effect of separating the tows and passing them through the reactor was assessed by
measuring the strength of single filaments removed at random from the bundle. The
Adhesion of glass fibres to epoxy resin using plasma copolymers 207
reactor has electric motors for the transfer of fibres through the plasma. These are
included in the end-pieces which were designed to prevent parasitic deposition in
this region. The reactor, which has been described in detail elsewhere [1], used a
matching network to provide a plasma at a radiofrequency of 13.56 MHz and at a
power of 1 W. The monomers were metered through needle valves into the reactor
at an overall flow rate of 2 seem (cm3 (at STP) per min). The power and flow rate
were chosen to give the required power density for the retention of the monomer
functionality in the plasma polymer. The reactor was pumped by a two-stage rotary
pump with a base pressure of 2 x 10~3 mbar. The pressure during the plasma
polymerisation was generally of the order of (2-3) x 10~2 mbar. The fibres were
resident in the reactor barrel for 15 min. This gave a conformal coating on the fibres
which was found by X-ray photoelectron spectroscopy (XPS) to be of the order of
0.01 /xm.
XPS spectra were obtained with Al Ka X-rays either with a VG Clam 2 spectrom-
eter (at Sheffield) or with the high-resolution Scienta 300 spectrometer (at Dares-
bury Laboratory, UK). A take-off angle of 45° relative to the analyser was used.
The elements in the glass surface were masked by the plasma polymer, so that the
coating thickness was greater than the core electron escape depth. The Cj s peak in
the spectrum of the plasma polymers was used to quantify the retained chemistry on
the surface of the coated glass fibres. The binding energies for the assignments are
given in the Appendix. Details are given in an earlier publication [1]. The fraction
of carboxylic acid groups on the fibre surface was quantified using trifluoroethanol
labelling. The methodology is given by Alexander et al. [5]. Detailed discussion
of the results for acrylic acid plasma polymers on glass fibres has been given else-
where [1].
2.7. Single-fibre strength measurements

Single fibres were selected at random from fibre tows for strength determination
at a gauge length of 6.25 mm. The details of the test methodology are given
elsewhere [3]. These fibres were tested in tension at a loading rate of 0.52 mm/min.
The result was discarded if there was any indication of slippage in the grips or
mounting adhesive. The diameter of the fibres was measured in a Camscan Series II
scanning electron microscope. At least 50 samples of each batch of fibres were
tested and the data analysed using Weibull statistics. The Weibull modulus was
obtained using a linear regression algorithm. The control fibres represent those
that were separated, as described above, and passed through the plasma reactor
in the absence of monomers. These were compared to the as-received unsized or
unseparated fibres. The coated fibres refer to the separated fibres with a plasma
deposit.
2.2. Assessment of interfacial adhesion from single-fibre fragmentation

The adhesion of the various glass fibres to an epoxy resin was determined using
the single fibre fragmentation test. Single filaments were extracted from the fibre
Table 1.
The tensile properties of the LY1556/GY298 matrix resin
Yield strength (MPa) 45.9 ± 3.1

Cold draw strength (MPa) 32.3 ± 2.2
Shear yield strength (MPa) 26.5 ± 1 . 8
Shear cold draw strength (MPa) 18.7 ± 1.3
Yield strain (%) 4.5 ± 0.2
Failure strain (%) 13.7 ± 1.2
Modulus (GPa) 2.4 ± 0.2
tows and embedded into the resin matrix and cured. The resin was a blend of
Araldite LY1556 which is a diglycidylether of Bisphenol A and a flexible aliphatic
epoxy resin Araldite GY298 (Vantico, UK). A 60:40 ratio of these two components
was found to have an appropriate failure strain to achieve saturation in the lengths
of the fragments. The resin was cured with 70 phr (parts per hundred of resin
by weight) of nadicmethylenetetrahydrophthalic anhydride (NMA, Stag Polymers
and Sealants, UK) and 23.91 phr Capcure 3-800 (Henkel Nopco, UK), a mecaptan
terminated polymeric hardener. The matrix was cured at 80° C for 4 h followed by a
post-cure for 3 h at 130°C. The samples were allowed to cool down overnight within
the oven. A subsequent annealing stage was utilised to ensure the fibres remained
straight for the fragmentation tests [1]. The cured resin had an elastic modulus
of 2.4 GPa and a tensile yield strength of 46 MPa. Assuming a von Mises yield
criterion, a shear yield strength of 26.5 MPa has been calculated. This resin has
a failure strain of >10% so that the fragmentation of E-glass fibre can proceed to
saturation. The properties of the resin given in Table 1 are an average of 5 samples.
The fragmentation test was carried out with dumbbell shaped specimens with a
gauge length of 33 mm on a custom designed testing machine (Micromaterials,
UK) using a test speed of 0.13 mm/min. At intervals of 1% strain, the test was
interrupted and the number of fragments recorded digitally using image grabbing
software. The test was considered to be complete when the fibre length reached a
saturation value. Both the fragment lengths and the debond lengths were measured
digitally from the images taken during the test as described elsewhere [4, 6].
3. RESULTS
3.1. Uniformity of coating
Figure 1 shows the surface elemental composition (from XPS) of E-glass fi-
bres coated with plasma polymers of acrylic acid and 1,7-octadiene of various
comonomer compositions. The Si2P peak falls to an almost undetectable level for
all of the plasma-polymer-coated fibres. The fibres were extracted at random from
the coated tows. From the escape depth of the Si2P core electrons at a take-off an-
gle of 45° an estimate of the coating thickness can be made. This was found to be
r-
Unsized 20 40 60 80 100
Acrylic acid (mol%)
Figure 1. XPS elemental composition of the surface of E-glass fibres with plasma polymer deposits
as a function of the concentration (mol%) of acrylic acid in the monomer feed.
0.01 jinn. The Ois peak follows the expected trend falling rapidly from the com-
position of the uncoated fibre to approximately 10% for the plasma-polymer-coated
fibre. Thus, some oxygen is incorporated into the plasma polymer films as a result
of residual water in the reactor, principally arising from the high surface area which
the E-glass fibres present to the plasma gases. This has been discussed in detail
elsewhere [1]. The Cis peak was charge corrected to 285 eV so that the components
in the narrow scan spectrum could be correctly assigned. The carboxylic acid func-
tionality (COOH) cannot be precisely quantified because it has the same binding
energy as an ester (COOR). The former was esterified with trifluoroethanol (TFE)
to give a CF3 component of higher binding energy. Therefore, comparison of the
relative intensities of these peaks gives an estimate of the retained COOH in the
plasma polymer coating [1, 5]. The degree of retained carboxylic functionality in
these plasma polymers was found by this TFE labelling method to be less than that
obtained for deposits on other substrates, such as carbon fibres. A full discussion
of these data has been reported in reference [1]. To confirm the uniformity of the
coating single filaments were taken at random from the tows and analysed on the
Scienta ESCA 300 high resolution spectrometer. XPS spectra were obtained from
these individual filaments and the intensities and components of the Ci s peaks were
found to be similar [1]. The allylamine plasma copolymers have been similarly
analysed. In this case, the Ni s peak can be used to assess the retention of the amino
functionality.
3.2. Strength ofplasma-polymer-coated fibres

To achieve a uniform conformal plasma polymer deposition onto all of the fibres in
the tows, the as-received unsized tows were separated with an air brush and wound
onto a glass spool which was inserted into an end-piece of the plasma reactor.
-llo Separated unsized

.||n Unsized
i—
i—i—i
i -o i—i—i,
ko
D
1 i !
J
-0.5 0.5
In (strength) (GPa)
Figure 2. Weibull plots of the cumulative failure probability (P) for the strength distributions of the
separated unsized control fibres and as-received unsized E-glass fibres.
The fibres could be transported through the plasma reactor to another glass spool
enclosed in the end-piece at the other end of the reactor. One pass through the
reactor, which took 15 min, was sufficient to provide a coating which masked the
glass substrate elements.
The degradation in the strength of the fibres after air separation and transport
through the reactor was assessed using Weibull statistics. Figure 2 compares the
strength statistics of the unsized fibre before and after spreading and transpon:
through the reactor. The slopes of the lines represent the breadths of the distrib-
utions.
The distribution of strengths for the fibres which have been spread and transported
through the reactor is clearly broader indicating that some degradation of strength
has occurred. This has been quantified by calculating the Weibull moduli, m, from
the slopes of these logarithmic probabilities. The Weibull moduli are given in
Table 2. The reduction in m from 3.1 to 1.65 indicates that some weaker fibres
have been introduced into the population. However, the average strength may not
have been reduced beyond the standard deviation. After deposition of a plasma
polymer onto the fibres the value of m has increased to a higher value than that for
the as-received unsized fibres.
This indicates that the fibre strength distribution has narrowed. Since the
average strength has returned to a value equal to or greater than that of the as-
received unsized fibres, we conclude that the damage introduced during handling
has been reduced in impact, by flaw filling, which reduces the associated stress
concentration.
What is particularly interesting is that the plasma polymer coatings containing
the differing functional groups appear to behave differently. Table 2 shows that for
the acrylic acid plasma polymer series, the highest concentration of hydrocarbon
(1,7-octadiene) appears to give the least improvement. On the other hand, with the
Table 2.
Single filament tensile strength of plasma polymer coated and unsized E-glass fibres
Coating parameters Fibre parameters

Monomer 1,7-Octadiene No. Diameter Average strength Weibull
composition (mol fraction) of samples (/xm) (GPa) modulus m
None 56 15.46 ± 1.74 1.46 ± 0 . 8 3.10
(as-received,
unsized)
None - 52 15.46 ±1.74 1.35 ± 0 . 7 1.65
(separated
unsized)
Acrylic acid/ 0 51 15.23 ±1.23 1.53 ±0.58 4.16
1,7-octadiene 0.11 51 14.94 ± 1.68 1.33 ±0.56 3.94
0.51 56 15.12 ±1.48 1.38 ±0.57 4.04
0.71 55 14.99 ± 1.82 1.51 ±0.71 3.40
Allylamine/ 0 52 15.34 ± 1.29 1.42 ±0.61 3.48
1,7-octadiene 0.18 52 14.83 ±1.37 1.64 ±0.59 4.55
0.40 54 15.39 ± 1.52 1.47 ±0.57 4.26
0.74 52 15.57 ± 1.80 1.57 ±0.6 4.88
1,7-Octadiene 1 64 14.87 ± 1.24 1.58 ±0.64 3.58
allylamine series, the trend is reversed with the 74% 1,7-octadiene plasma polymer
providing the highest improvement with the largest Weibull modulus.
3.3. Adhesion offunctionalised glass fibre to epoxy resin

The single embedded fragmentation test has been used to examine the differing
degrees of adhesion between the plasma polymer coated fibres and epoxy resin.
The test was interrupted at 1% intervals of applied strain for the measurement of
fragment and debond lengths. The test was continued until saturation was reached.
This occurs when the fragments are too short to be loaded, through shear at the
interface, to their fracture strength so that the length remains constant. The data are
summarised in Table 3.
As a first approximation, the fibre with the shortest average fragment length can be
considered to have the strongest interfacial bond. In this way, the 49% acrylic acid
plasma copolymer and the 60% allylamine plasma copolymer functional coatings
appear to have promoted the highest degree of adhesion to the epoxy resin matrix.
As can been seen in Table 2, the strength of the differently coated fibres varies.
Since this will also contribute to the fragmentation process it is necessary to use a
model that provides a true interfacial adhesion parameter.
3.3.1. Constant shear stress model. The conventional methodology assumes

that the interfacial shear stress becomes constant at saturation where complete
debonding has occurred. It is then assumed that the value obtained can be equated
to the interfacial shear (or adhesion bond) strength (ra) between that fibre and resin
combination.
Table 3.
Fragmentation data at saturation for E-glass fibres coated with plasma copolymers from acrylic acid or allylamine and 1,7 octadiene, in an anhydride cured
epoxy
Fibre/plasma Uncoated" Acrylic acid (%) Allylamine (%)

polymer 89 49 29 0 Too 82 60 26
Number of fragments 64 71 64 50 36 35 42 55 53
(±9) (±13) (±5) (±2) (±6) (±2) (±10) (±7) (±7) s
Average fragment length (mm) 0.42 0.4 0.38 0.46 0.60 0.65 0.56 0.39 0.41
(±0.03) (±0.02) (±0.12) (±0.12) (±0.19) (±0.17) (±0.17) (±0.10) (±0.14) Co
3
Debonding (%) 0 2.5 0 0 44 50 49 0 0
IT!
(±5) 50
Critical fibre length /c (mm) 0.56 0.54 0.51 0.62 0.8 0.87 0.75 0.52 0.55 as
(±0.07) (±0.03) (±0.05) (±0.02) (±0.14) (±0.05) (±0.16) (±0.08) (±0.04) Co
Fibre strength at /c (GPa) 5.89 2.91 2.65 3.03 2.81 2.66 2.61 2.83 2.85
(±0.42) (±0.13) (±0.06) (±0.03) (±0.14) (± 0.05) (±0.13) (±0.11) (±0.05)
r a (MPa) 82 42 39 37 26 23 26 41 39
(±17) (±11) (±4) (±4) (±7) (±2) (±8) (±9) (±4)
a
These control fibres are unsized and have been separated and transferred through the reactor.
The analysis is based on the Kelly-Tyson model [7] which gives equation (1)
where rf is the radius of the fibre, <7fu is the tensile strength of the fibre at its critical
length /c. lc is the length of the shortest fibre which can be loaded through stress
transfer to fracture. Thus a fragment slightly longer than lc will fracture but one
equal or shorter than lc cannot be fractured. Therefore, lc has to be calculated from
equation (2)
4/
k- j , (2)
where / is the average measured fragment length at saturation.
The apparent interfacial shear strengths (r a ) calculated from equation (1) are
given in Table 3. The uncoated control fibres have an apparent interfacial shear
strength of 82 MPa, which is approximately three times the shear yield strength of
the matrix (26.5 MPa). Furthermore, debonding of the interface was not observed,
so that yielding in the interfacial region might be expected. Thus the interfacial
bond strength must be larger than the matrix shear yield strength. The Kelly-Tyson
analysis is clearly not valid because of the assumptions employed in the constant
shear model [7]. Since the matrix is elastoplastic and not perfectly plastic, a constant
interfacial stress can only be achieved with 100% debonded interface. Thus, an
alternative analytical procedure is required. This methodology must include the
elastoplastic properties of the matrix and the degree of debonding. Furthermore,
additional micromechanical features such as transverse matrix cracking will also
need to be considered.
3.3.2. The cumulative stress transfer model [8]. Tripathi et al. [9] proposed the
plasticity effect model which enables the shear stress profile of a fragment of any
length to be simply calculated. In this model, the variational approach of Nairn [10]
for a perfectly-bonded elastic fibre in an elastic matrix is used to calculate the shear
stress profile of a fragment. When the shear stress reaches the shear yield strength of
the matrix, assuming the von Mises criterion, the profile is truncated. Cold draw in
the matrix (at the high strains required for fragmentation) can also be incorporated
in a similar manner. In addition, the shear stress operating in debonded regions
can be included by employing Coulomb's law to calculate the degree of frictional
adhesion operating as a result of the radial stress obtained from the variational
model. Therefore, a full shear-stress profile of a fragment of any length can be
calculated from the measured debond length and the properties of the fibre and
matrix. This can be converted into a tensile fibre stress profile Of (x) using a balance
of forces approach.
Within the fragmentation test at each strain interval, there will be a distribution
of fragments of differing lengths. At low strains, the statistics of fibre strength
will dominate fragmentation. However, as the test progresses, the tensile stress
profile within each fragment becomes a function of the interfacial bond and the
stress transfer characteristics. To obtain an interfacial parameter it is necessary to

average the stress profiles for every fragment. For the cumulative stress transfer
function (CSTF) [8] this is done by integrating each Gf{x) profile, summing and
normalising to the total length of the fragments, as shown in equation (3)
z
CSTF - " ' - 1 J?=N , (3)
where N is the number of fragments, Lt is the fragment length for / = 1, 2, 3 . . . N

and Of (JC) is the tensile stress in the fibre at a distance x from its end, calculated from
the Plasticity Effect Model [9]. A full discussion of the CSTF equation is given in
Ref. [8].
The CSTF approach provides a good analysis of interfacial properties. However,
since CSTF is a function of fragment length [2] we have normalised the CSTF to an
ideal value of CSTF, calculated for a perfectly bonded fibre of the same length in an
elastic matrix (CSTFo). This function is referred to as the stress transfer efficiency
(STE) and represents the efficiency of that interface as a function of the ideal stress
transfer capability.
CSTF
STE= , (4)
CSTFo
where CSTF is given by equation (3). C S T F Q is calculated using a similar procedure
but from ideal elastic properties, at the same length.
This analysis requires complete fragmentation data at a range of strain intervals.
Only the average data are given in Table 3 but full details can be obtained from
Ref. [11].
The input parameters for the calculation of CSTF and STE are the matrix
properties in Table 1 and the fibre and other properties in Table 4. The variational
model [10] requires a knowledge of the volume of matrix perturbed by the presence
of the fibre. Tripathi et al. [12] have shown by FE modelling that R/r? should be
Table 4.
Fibre and other properties used in the cumulative stress transfer function (CSTF)
and stress transfer efficiency (STE) calculations
Mechanical property Magnitude
Elastic modulus of E-glass fibre (GPa) 76

Poisson's ratio for E-glass fibre (GPa) 0.22
Shear modulus for E-glass fibre (GPa) 31.2
Coefficient of friction between fibre and matrix 0.2
Fibre volume fraction 0.000625
Thermal expansion coefficient of E-glass fibre (K _ 1 ) 4.9 x 10~6
Thermal expansion coefficient of matrix (K _ 1 ) 40 x 10 - 6
R/rf 40
a
R is the radius of the resin matrix perturbed by the presence of the fibre of
radius rf.
given a value of 40, where R is the radius of the matrix under the influence of the
fibre.
The shear transfer efficiency (STE) was calculated according to equations (3)
and (4) from the fragmentation data at each interval of applied strain from 3 to 13%
for each coated fibre. Since the STE of a fibre is a function of its length, it is more
informative to compare the values of STE at a constant length. The most efficient
interface is given by the fibre with the largest value of STE at the shortest fragment
length.
The STE has been plotted against average fragment length in Figs 3 and 4.
The most efficient interface is defined by the curve associated with the shortest
fragments. Thus, the 49% acrylic acid (Fig. 3) and the 60% allylamine plasma
copolymers (Fig. 4) provide the highest adhesion to this epoxy resin.
To examine the correlation between the surface chemistry of the fibres and their
adhesion to epoxy resin, an appropriate ranking parameter is needed. Examination
of Figs 3 and 4 shows that an ineffective length can be estimated by extrapolating
STE to zero. This gives the shortest length of an embedded fibre which can be
reloaded, through the matrix, after fracture. The higher the interfacial bond strength
the shorter the ineffective length.
If the shear stress generated at the interface causes debonding, then the stress
transfer efficiency will decrease with a consequent increase in the ineffective length
of the fibre. The friction coefficient and the radial residual stress will determine
the magnitude of the reduction in STE on debonding. With a strong interface,
shear yielding in the matrix near the interface will also reduce the stress transfer
efficiency. This is because the STE methodology for assessing interfacial adhesion
50
-29% Acrylic acid
45
-49% Acrylic acid
40 89% Acrylic acid
35 Unsized control
-100%Octadiene
30
w 20
15
10
5
0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Average fragment length (mm)
Figure 3. Variation in stress transfer efficiency (STE) with average fragment length for the acrylic
acid/l,7-octadiene plasma polymer functionalised E-glass fibres. The control fibres were unsized tows
which had been separated and transported through the reactor in the absence of monomers.
60
- 26% Allyl amine
- 60% Allyl amine
50
-82% Allyl amine
100% Allyl amine
40 -Unsized control
-100%Octadiene
?30
*2(H
10
0
0.2 0.4 0.6 0.8
Average fragment length (mm)
Figure 4. Variation in stress transfer efficiency (STE) with average fragment length for the ailyl-
amine/l,7-octadiene plasma polymer functionalised E-glass fibres. The control fibres were unsized
tows which had been separated and transported through the reactor in the absence of monomers.
includes the mechanical properties of the matrix and fibre and the radial stresses
which are responsible for physical adhesion where no chemical bonding exists.
Thus, the contribution of a debonded or partially debonded fibre can be included
in the analysis. As such it is a more precise technique for quantifying fibre-matrix
bond strength.
3.4. Surface chemistry of plasma-polymer-coated fibres

3.4.L Acrylic acid/lf7-octadiene plasma polymers. Figure 1 gives the concen-
tration of elements in the surface of E-glass fibres coated with acrylic acid/l,7-octa-
diene plasma polymers. The major elements present in the surface of E-glass fibres
are Si and O. Carbon is also present as a contaminant. The other elements (8.5% by
weight) in the E-glass surface (Al, Ca, Na) were not detectable on the surface of
the coated fibres showing that the fibres were uniformly coated with the plasma
polymer. We have already reported high resolution XPS spectra which showed
complete coverage of the fibres in the tow [1]. Of importance is the retention of
the carboxylic acid functionality on the glass surface. This has been quantified by
esterification with trifluoroethanol (TFE) which provides a CF3 label representa-
tive of the proportion of the carbon functional groups present as COOH. Table 5
shows that the retained concentration of COOH is lower than that observed pre-
viously with other substrates [2, 5]. This has been attributed to residual water in
the plasma reactor, as a result of the high surface area of the substrate. Despite
the non-quantitative labelling (85%) for conventional poly(acrylic acid) (PAA), the
concentration of COOH groups decreases approximately linearly with 1,7-octadiene
concentration in the feedstock.
Table 5.
XPS analysis results for trifluoroethanol (TFE) derivatised acrylic acid/l,7,-octadiene plasma copoly-
mers on E-glass fibres. PAA is a conventional poly(acrylic acid)
Deposit C-R (%) yS-Shift C-OR C=0 C-CF 3 COOR CF 3 (%)

(%) (%) (%) (%) (%)
PAA on 29.07 ± 19.10± - - 16.45 ± 19.10± 16.38 ±
glass 1.21 0.80 0.66 0.80 0.65
coverslip
100% acrylic 43.92 ± 13.28 ± 7.51 ± 1.31 ± 10.34 ± 13.28 ± 10.37 ±
acid on 1.83 0.55 0.31 0.06 0.41 0.55 0.41
aluminium
foil
100% acrylic 44.45 ± 9.78 ± 13.88 ± 10.22 ± 5.94 ± 9.78 ± 5.94 ±
acid on E- 1.85 0.41 0.58 0.43 0.25 0.41 0.25
glass fibres
89% acrylic 53.88 ± 12.79 ± 4.01 ± 6.89 ± 4.83 ± 12.79 ± 4.82 ±
acid on E- 2.25 0.53 0.17 0.29 0.20 0.53 0.20
glass fibres
57% acrylic 45.54 ± 11.23± 10.23 ± 15.95 ± 2.91 ± 11.23± 2.90 ±
acid on E- 1.90 0.47 0.43 0.67 0.12 0.47 0.12
glass fibres
49% acrylic 68.87 ± 3.93 ± 12.91 ± 5.08 ± 2.64 ± 3.93 ± 2.64 ±
acid on E- 2.87 0.16 0.54 0.21 0.11 0.16 0.11
glass fibres
29% acrylic 71.92± 4.21 ± 11.30± 4.63 ± 1.88± 4.22 ± 1.85 ±
acid on E- 3.00 0.18 0.47 0.19 0.08 0.18 0.08
glass fibres
0% acrylic 69.05 ± 3.50 ± 15.21 ± 7.26 ± 0.74 ± 3.50± 0.74 ±
acid on E- 2.88 0.15 0.63 0.30 0.03 0.15 0.03
glass fibres
Table 6.
XPS elemental analysis (in at%) of allylamine/octadiene plasma copolymers deposited onto E-glass
fibres
Deposit 100% 82% 60% 26%

allylamine allylamine allylamine allylamine
Cis 76.79 ±3.33 84.75 ±3.53 84.60 ±3.53 91.69 ±3.82

Nis 18.41 ±0.77 10.85 ±0.45 4.94 ± 0.21 3.07 ±0.13
Ols 4.80 ± 0.20 4.40 ±0.18 8.67 ± 0.36 4.14 ±0.17
Si2P - - 1.79 ±0.08 1.10 ±0.05
3.4.2. Allylamine/1} 7-octadiene plasma polymers. The elemental analysis of the

surface of the coated fibres is given in Table 6. The binding energies for the different
components in the Nis peak are insufficiently separated to provide a quantification of
the nitrogen functionality. Therefore, the results of the Q s curve fit are also shown
in Table 7. C-N-R component is assumed to be mostly assignable to retained C-NH2
218 D. J. Marks and E R. Jones
Table 7.
Fraction of components (%) in the Ci s narrow scan for a series of allylamine/octadiene plasma
copolymers deposited onto E-glass fibres
Deposit 100% 82% 60% 26%

allylamine allylamine allylamine allylamine
C-R 55.46 ±2.31 74.58 ±3.11 71.33 ±2.97 75.41 ±3.14
C-N-R, C=N-R, 32.84 ±1.37 21.06 ±0.88 19.88 ±0.83 17.84 ±0.74
C-O-R
C=0, N-C=0 9.27 ± 0.39 3.66 ±0.15 6.02 ± 0.25 5.59 ±0.23
C(=0)-0-R, 2.43 ±0.10 0.71 ±0.03 2.76 ±0.12 1.17 ±0.05
(N 2 )-C=0
groups and these decline with the concentration of 1,7-octadiene in the monomer
feed.
The Cis component peak at a binding energy of 286.4 eV can be assigned
to amine, imine or ether. It has an intensity equivalent to the theoretical for
100% amine retention in the 100% allylamine plasma polymer. With a 26% al-
lylamine in the monomer feed, the intensity is higher than theoretical. Oxygen
incorporation occurs during deposition, as with acrylic acid, some of which can be
assigned to C-OR. It is shown in Table 5 that the highest concentration of C-OR is
associated with the 100% 1,7-octadiene plasma polymer. As a result, we conclude
that there is a high retention of the amine functionality in these plasma copolymers
and that the total concentration of nitrogen (Nis) is a good approximation of the
reactive functional groups.
4. DISCUSSION
4.L The effect of surface functionality on adhesion to epoxy resin
In plasma polymerisation at low RF power, substrate activation is followed by the
deposition of a strongly adhering conformal coating with high retained functional-
ity. In this study the concentration of functional groups has been varied by incor-
porating a reactive diluent, 1,7-octadiene, which also adds to the mechanical sta-
bility of the coating through increased crosslinking. Therefore, it can be assumed
that the yield strength of this coating is significantly higher than that of the matrix,
so that the strength of the interfacial bond between the functional coating and the
elastoplastic matrix dominates the stress transfer mechanism. The STE methodol-
ogy utilises these properties to quantify the changes in the degree of adhesion with
surface functionality. The most appropriate adhesion parameter has been shown to
be the ineffective length, l\.
4.1.1. The role of carboxylic acid functional groups in adhesion. The plasma
copolymers of acrylic acid and 1,7-octadiene with varying composition have
been deposited onto E-glass fibres. The relative concentration of carboxylic acid
0.62 -I
0.58 \
0.54 \
^ 0.46 \ \
\ CONTROL
0-42 f \
v
\ ^ •
0.38 \ V— "
0.34 -I . 1 . 1 . 1
0 1 2 3 4 5 6
[ C O O H ] r e t a i n e d (mol%)
Figure 5. Variation of ineffective length (l\) (obtained by extrapolation of STE) with relative retained
carboxylic acid concentration obtained by TFE labelling and XPS. The control refers to the unsized,
separated fibres without a coating.
groups on the functionalised E-glass fibres was estimated using trifluoroethanol

labelling [1, 5]. This has been plotted against the adhesion parameter, lu which
is the ineffective length of that fibre in the epoxy resin. lx is inversely proportional
to the degree of adhesion. Figure 5 shows that the degree of adhesion appears to
reach an optimum value at a carboxylic acid concentration of 3 mol% of the surface
carbon functionality.
The 100% octadiene plasma polymer coating is seen to have provided a lower
degree of adhesion than the control fibres. Therefore, we conclude that deadhesion
occurs at the interface between the functional coating and the resin.
The carboxylic acid groups can react with the epoxy groups in the resin and
couple the fibres to the resin. Since the non-functional 1,7-octadiene coating makes
the adhesion worse than the control, the involvement of functional groups in the
chemistry of adhesion is clearly demonstrated.
4.1.2. The role of amino functionality in adhesion. In Fig. 6 the ineffective

length, /i, is plotted against the nitrogen concentration, fNis] retained in the
allylamine plasma copolymers. As discussed above, the latter is the best indication
of the reactivity of the fibre surface through retained amino groups. Figure 6 shows
that the degree of adhesion, as shown by the shortest ineffective length, is optimised
at a nitrogen concentration of ^ 5%. The 1,7-octadiene plasma polymer does not
promote adhesion to the fibre confirming that interface being probed is that between
the plasma polymer and the resin. Furthermore, the amine groups at the coated fibre
surface are clearly involved in a chemical reaction with the anhydride-cured epoxy
resin. Of interest is the 100% allylamine plasma polymer coating, which has the
highest functionality but gives a weak interface and exhibits debonding. Further
research is required to understand why a surface with the highest functionality is
10
[N]retained (mol%)
Figure 6. Variation of ineffective length (/j) (obtained by extrapolation of STE) against the relative
retained surface nitrogen concentration. The control refers to the unsized, separated fibres without
coating.
apparently less reactive. However, it is highly likely that at this high concentration
the resin-cure mechanism in the interfacial region has been perturbed by the fibre
surface.
4.1.3. Micromechanics of interfacial failure in the fragmentation test. A simple

embedded fibre fragmentation test was used to examine the differing degrees of ad-
hesion. The individual fragments were examined for debonding at 1% intervals of
applied strain to give a database for interfacial quantification. Transverse microc-
racking (TMC) emanating from a fibre break is also observed in the fragmentation
test. The micromechanics associated with a fibre break, is summarised in Table 8
in the order of adhesion ranking (1 = highest, 9 = lowest). Transverse matrix
cracking arises from the release of stored energy on fibre fracture and is indicative
of a strong interfacial bond. In a recent study [13] phase-stepping photoelasticity
has shown that debonding can be initiated at the interface near a transverse matrix
crack illustrating that both phenomena are operative in the fragmentation test. AH
of the fibres which are partially debonded have, as expected, the lowest degree of
adhesion according to the STE ranking in Table 8. It should be noted that the STE
methodology is able to differentiate between the effectiveness of relatively strong
bonded interfaces.
4.1.4. The unsized control E-glass fibres. The unsized E-glass fibres without
plasma polymer coating have a relatively high stress transfer efficiency (Fig. 3)
and low ineffective length, l\ (Figs 5 and 6). Thus, a degree of adhesion is observed
despite the recognised need for silane coupling agents as adhesion promoters for
E-glass. Introduction of a hydrocarbon surface with only a very small degree of
Table 8.
Ranking of plasma polymer functionalised E-glass fibres from STE with interfacial micromechanics
Rank Acrylic acid/ Allylamine/ Failure mode Debonding at

1,7-octadiene 1,7-octadiene saturation
(%)
1 49% acrylic acid TMC 0
2 60% allylamine TMC 0
3 89% acrylic acid MM 2
4 26% allylamine TMC 0
5 <— Non-unsized —> TMC 0
6 29% acrylic acid TMC 0
7 82% allylamine MM 49
8 <— 100% 1,7-octadiene —> MM 44
9 100% allylamine MM 50
TMC = transverse matrix cracking, MM = mixed mode of transverse matrix cracking and debon-
ding.
functionality (from oxygen incorporation into the coating) decreases the adhesion
but complete debonding was not observed. This is in contrast to 1,7-octadiene
plasma polymer coated carbon fibres which were completely debonded [2, 4].
The major difference between these two fibre systems (carbon and E-glass) is in
transverse modulus and fibre diameter. E-glass has approximately 3 times the
transverse modulus of Type-A carbon fibre and twice the diameter. Therefore, a
higher radial compressive stress is expected for the glass fibres in a hot cured epoxy
resin. Thus, a strong frictional adhesion bond can explain the apparent adhesion in
the absence of a functional coating.
5. CONCLUSIONS
Strong adhering plasma copolymers have been deposited onto E-glass fibres. The
plasma polymers were deposited onto tows as they were transferred through
the reactor. A uniform conformal coating was deposited as shown by XPS
analysis. Acrylic acid and allylamine have been used as the functional comonomers
for introducing either carboxylic acid or amino groups onto the fibre surface.
1,7-Octadiene is a non-functional comonomer which has been used to provide
mechanical stability to the coating and optimise the concentration of functional
groups. By varying the composition of the monomer feed, the adhesion to the epoxy
resin could be optimised. After handling and passage through the plasma reactor
(without the presence of monomers and a plasma), the fibre strength distribution
is broadened. However, after coating the strength-distribution returns to a similar
level as the as-received fibres. This demonstrates that the coating has a protective
function at a thickness of approximately 0.01 /xm.
222 D. J. Marks and R R. Jones
Acknowledgements
We thank EPSRC and Advanced Composites Group Ltd, UK for financial support.
We acknowledge A.C. Johnson for help with the interpretation of the fragmentation
data.
REFERENCES
1. D. J. Marks and F. R. Jones, Composites Part A 33, 1293-1302 (2002).
2. N. Lopattananon, S. A. Hayes and F. R. Jones, /. Adhesion 78, 313-350 (2002).
3. A. P. Kettle, A. J. Beck, L. O'Toole, F. R. Jones and R. D. Short, Composites ScL Technol. 57,
1023-1032 (1997).
4. N. Lopattananon, A. P. Kettle, D. Tripathi, A. J. Beck, E. Duval, R. M. France, R. D. Short and
F. R. Jones, Composites Part A 30A, 49-57 (2000).
5. M. R. Alexander, P. V. Wright and B. D. Ratner, Surf. Interface Anal. 24, 217-220 (1996).
6. D. Tripathi, N. Lopattananon and F. R. Jones, Composites Part A 29A, 1099-1109 (1998).
7. A. Kelly and W. R. Tyson, J. Mech. Phys. Solids 13, 329-350 (1965).
8. D. Tripathi and F. R. Jones, Composites Sci. Technol. 57, 925-935 (1997).
9. D. Tripathi, F. P. Chen and F. R. Jones, J. Composite Mater. 30, 1514-1538 (1996).
10. J. A. Nairn, Mech. Mater. 13, 131-154 (1992).
11. D. J. Marks, PhD Thesis, University of Sheffield, Sheffield, UK (2002).
12. D. Tripathi, F. Chen and F. R. Jones, Proc. Roy. Soc. London A 452, 621-653 (1996).
13. F. M. Zhao, R. D. S. Martin, S. A. Hayes, E. A. Patterson, R. J. Young and F. R. Jones,
Composites: Part A, in press (2004).
APPENDIX
Assignment of binding energies in XPS
The binding energies in the XPS spectra were charge compensated to hydrocar-
bon Cis at 285 eV. The following assignments were used:
Table Al.
Binding energies of elements in XPS spectra
Element Peak Binding

energy (eV)
Carbon Cls 285
Oxygen Ols 329
Nitrogen Nls 400
Fluorine Fls 68
Silicon Si 2p3/2 100
Aluminium Al 2p3/2 72
Calcium Ca 2p3/2 347
Table A2.
Binding energies and assignments for Ci s curve-fitting of spectra from acrylic acid based plasma
copolymers and their fluorinated derivatives
Functional group Notation Energy Chemical shift

(eV) (eV)
Hydrocarbon C-R 285.0 —
Hydroxyl/ether C-OR 286.5 1.5
Carbonyl C=0 288.0 3.0
Carboxyl/ester C(=0)-OR 289.5 4.5
P shift from C-C(=0)-OR 285.7 0.7
carboxyl/ester
TFE derivatised O-C-CF3 287.99 2.99
TFE derivatised CF3 293.35 8.35
Table A3.
Binding energies and assignments for Ci s curve-fitting of spectra from allyl amine based plasma
copolymers
Peak Notation Energy Chemical shift

(eV) (eV)
Hydrocarbon C-R 285 0
Amine, hydroxyl C - N - R , C=N-R, 286.4 1.4
or ether, imine C-O-R
Carbonyl, amide, C=0, N - C = 0 288 3.0
imide
Ester, carboxylic C(=0)-0-R, 289.3 4.3
acid, urea (N2)-C=0
Ed. K. L. Mittal
© VSP 2004
Organophosphate adsorption on metal oxide surfaces
M. J. SHEPARD', J. R. COMER ', T. L. YOUNG2, J. S. McNATT ',

M. P. ESPE 2 , R. D. RAMSIER'•"•*, T. R. ROBINSON4 and L. Y. NELSON4
1
Department of Physics, The University of Akron, Akron, OH 44325-4001, USA
Department of Chemistry, The University of Akron, Akron, OH 44325-4001, USA
3
Department ofChemical Engineering, The University of Akron, Akron, OH 44325-4001, USA
4
Korry Electronics Co., 901 Dexter Avenue North, Seattle, WA 98109, USA
Abstract—Aluminum, titanium and zirconium play a major role in industrial applications where
environmental stability and surface functionalization are important. We are studying the adsorption of
hydrogenated and fluorinated phosphates on these materials to determine their potential as corrosion
inhibiting coatings, adhesion promoters, or anti-stiction modifiers. In this paper we compare and
contrast methods of phosphate deposition as well as different adsorbates and substrates primarily
using infrared spectroscopy and solid state nuclear magnetic resonance. We show that sonication can
be used to enhance the homogeneity of organophosphate films adsorbed on aluminum oxide surfaces.
For titanium, zirconium and Zircaloy-2 surfaces, the amount of adsorbed phosphate increases due to
sonication for one species studied but decreases for the other. This indicates that the type of phosphate,
the type of substrate and the deposition method all play a role in determining the adhesion of these
films to industrially relevant surfaces.
Keywords: Adsorption; phosphates; aluminum; zirconium; titanium.
1. INTRODUCTION
Functionalized metal surfaces are an important part of many industrial processes
and may involve coatings which enhance adhesion to other materials, minimize
corrosion of the base metal, or exhibit non-stick behavior. Understanding the
formation and performance of such coatings requires that we approach the problem
from a variety of directions and disciplines, and search for common trends and
concepts that transfer easily across material systems. We have been interested
for some time in the adsorption of organo-phosphorus compounds on aluminum
surfaces [1, 2], and have recently begun investigating titanium and zirconium
surfaces to extend the range of potential applications in the aerospace and other
e-mail: rex@uakron.edu
226 M. J. Shepard et al.
0 •- " 0 /^ R i
\ +- \, ~P/ \
M-OH 2
/ 0 ' ^ —R2
o
°M/
0
\ +
M-OH 2 --.
""V
--R1
0 * ^ —R2
Where M is a Metal that forms a Metal Oxide

with an IEP greater than the pH of the Solution
Figure 1. Schemes depicting long-range ionic interactions (left) and covalent bonding modes (right)
for phosphates near hydroxylated metal surfaces.
communities. This report is the first from our laboratory to compare the adsorption
of phosphate species on these surfaces by two different deposition methods: mild
agitation (stirring) and sonication. We identify and discuss transferable trends that
fit well within our conceptual models of phosphate film formation on surfaces.
Most metal surfaces exposed to ambient conditions become covered with native
oxide layers, and many form surface hydroxyl groups as well. In acidic solutions
these surfaces can become positively charged [3, 4] and spontaneously attract
the negatively charged (partially de-protonated) conjugate base of the acid. This
long range ionic interaction [5] can account for the growth of phosphate films on
metal surfaces by simply soaking substrates in solution. However, the tenacious
adhesion of some of these phosphate films would also suggest that strong covalent
bonds form via condensation reactions with surface hydroxyl groups. It would
seem likely that these phosphates are spontaneously attracted to hydroxylated metal
oxide surfaces by ionic interactions and upon reaching appropriate adsorption sites
undergo condensation reactions to form durable thin films. Figure 1 conceptually
outlines the combined process, where the ligands Ri and R2 are considered in the
discussion here to be non-reactive alkyl and fluoro-alkyl ester chains or reactive
hydroxyl groups.
Adsorbates of this type have been studied for a variety of applications on many
different metal surfaces, including those of aluminum alloys [6,7], carbon steels [8],
aluminum [1, 2, 9, 10], titanium [11] and others [12]. Many of these studies use hy-
drophobic end-terminated functional groups to form self-assembled thin films with
high film density which can lead to enhanced environmental resistance and mechan-
Organophosphate adsorption on metal oxide surfaces 227
Bubble Transport Implosion

Time •
Figure 2. Schematic showing role of bubbles and cavitation in delivering phosphates to surfaces for
enhancing film density and growth.
ical durability. However, as the films condense, the resulting hydrophobic properties
can inhibit wetting of the remaining regions of exposed surface. Thus, while the film
growth may be spontaneous and the adhesion tenacious, autophobicity can lead to
slow adsorption rates and patchy film structures.
In an effort to circumvent this problem, we have been investigating energetic
methods of film growth. Instead of only immersion in the phosphate solution
of interest, we subject our substrates to mild agitation by magnetically stirring
or aggressive agitation by sonicating the solution during exposure. The latter is
based on the fact that many commercial processes use sonication for preparing
components immersed in cleaning solutions. The sound waves create small
energetic bubbles which implosively collapse at the surface. This cavitation can
generate intense microjets, shock waves and locally high temperatures that remove
contaminants from the surfaces of the component. As depicted in Fig. 2, bubbles
can also transport reactive materials from solution to the substrate surface [13]. In
the case of phosphate-based reactants, the combination of strong ionic and covalent
adsorption mechanisms with energetic delivery can produce a durable and dense
surface coating.
It is our goal in this paper to demonstrate the ultrasonic assisted deposition
(USAD) of phosphate films on aluminum, titanium and zirconium surfaces, and
compare and contrast these with films adsorbed from solutions subjected to mild
mechanical stirring. We use industrially relevant deposition processes, metal oxide
surfaces and organophosphates. In doing so, we will be comparing and contrasting
different systems and methods of surface preparation. The data presented here, in
combination with our previous work [2], give a clear indication that phosphates
bond most strongly to alumina in the tridentate manner. Better surface coverage is
obtained with phosphates having fluorinated rather than hydrogenated alkyl chains,
and sonication can be used to improve the environmental stability of the fluorinated
coatings. In this paper we also extend these efforts to the titanium and zirconium
materials systems. Although we do not have a complete understanding of phosphate
adsorption on these surfaces, we do identify trends that can be explained in a self-
consistent manner.
228 M. /. Shepard et al.
2. EXPERIMENTAL
2.1. Chemicals
Zonyl® UR (DuPont, Wilmington, DE, USA) is a mixture of anionic fluorosurfac-
tants with the chemical formula CF3(CF2CF2)z(CH2CH20)xPO(OH)3; where x = 1,
2 o r 3 , x + ;y = 3 and z ranges from 1 to 7, as indicated in Fig. 3. Monostearyl acid
phosphate (MSAP), also depicted in Fig. 3, was obtained from Calgene Chemicals
(Skokie, IL, USA) and has the chemical formula CH3(CH2)17OPO(OH)2. Spec-
troscopy grade 2-propanol purchased from Aldrich (Milwankee, WI, USA) was
used as a solvent. All materials were used as received without further purification.
A clear solution comprised of 5 g Zonyl UR, 300 ml 2-propanol and 200 ml wa-
ter was obtained with magnetic stirring and mild heating (50°C) for 2 h. A cloudy
suspension comprised of 2.5 g MSAP, 300 ml 2-propanel and 200 ml water was pro-
duced in a similar manner. These solutions were used to treat the metal and metal
oxide surfaces using procedures described below.
2.2. Substrate preparation

Bulk substrates of aluminum (ESPI, Ashland, OR, USA) and Zircaloy-2 (Wah
Chang, Albany, OR, USA) were cut into rectangular coupons (nominally 20 mm x
48 mm) from sheet stock material and polished using successively finer diamond
pastes in stages to achieve a mirror finish. The coupons were then degreased, rinsed
in de-ionized water and dried in air. Thin film substrates were deposited in vacuum
onto clean glass slides by sputtering, in the case of Al, and by ion-beam-assisted
deposition for Zr and Ti. A thin native oxide forms on these substrates when they are
exposed to air, so we refer to them as oxidized metal surfaces. We used the planar
substrates for the infrared, contact angle, ellipsometry and atomic force microscopy
investigations discussed in this paper. Aluminum oxide powder (0.6 /xm (X-AI2O3)
Zonyl UR
O O O
P i I
RO^lÔR ROîÔR ROîÔH
OR OH OH
R = CF3(CF2CF2)z(CH2CH2)
Z=1-7
MSAP
O
II
CH3(CH2)i7Cr lÔH
OH
Figure 3. Structural representations of the phosphates studied in this work.
was purchased from Baikowski International (Charlotte, NC, USA) and used as
received for the solid-state NMR and infrared studies.
2.3. Phosphate deposition

Phosphate adsorption was performed either by stirring or sonicating in solution. We
refer to the latter method as ultrasonic assisted deposition (USAD) for which we
used a liquid processor sonicator (Misonix, Farmingdale, NY, USA) equipped with a
sapphire tip. The stationary sonicator horn was immersed in a shallow water-alcohol
bath containing the MSAP or Zonyl UR species. Planar substrates lay submerged
directly beneath the horn, and in some cases the water/alcohol bath container was
mounted on a computer controlled x-y translation system for controlled rastering
of the substrate beneath the horn. We found no significant differences in films
that resulted from stationary USAD vs. those formed by rastered USAD. Following
USAD, the substrates were removed from the solution, rinsed with water, and placed
in an oven or blown dry with nitrogen gas to remove traces of water. Substrates
coated by stirring were handled similarly except that the USAD procedure was
replaced by magnetic stirring.
For USAD on the alumina powder, a beaker (held stationary in this case) was
filled with 2 g alumina powder and 80 ml of the desired solution. The sonicator
horn was lowered into the liquid to a depth of approx. 13 mm and the mixture was
sonicated for 5 min. Other powder samples were coated by magnetically stirring the
powder in the solution for 15 min. Following adsorption, the powder was collected
via filtration, washed twice, and dried under vacuum at 80° C for 12 h.
2.4. Environmental stability

For these studies we inserted phosphate covered substrates into a reflux distillation
column containing 100°C water and steam. The coupons were removed from the
column periodically and IR scans collected, and then the samples were placed back
into the same column for continued exposure. Fresh water was used when a new
sample was inserted to avoid cross-contamination.
2.5. Infrared spectroscopy

Infrared spectroscopy was performed using a dry air purged Mattson 7020 spec-
trometer in the range 700-4000 cm - 1 at 4 cm"1 resolution. Transmission IR spec-
tra of neat reagents and coated powders were acquired using a micro-cell sampling
fixture with AgCl windows. Specular reflection spectroscopy was performed on the
planar substrates with a fixed-angle sampling attachment containing a polarizer and
gold mirrors. A liquid-nitrogen-cooled HgCdTe detector was used for all measure-
ments.
2.6. Solid state nuclear magnetic resonance

A Varian Unityplus-200 (4.7 T) spectrometer with a Doty Scientific Magic-Angle-
Spinning (VTMAS) probe was used for solid-state nuclear magnetic resonance
230 M, J. Shepard et al.
(NMR) measurements. Samples were packed into 7 mm silicon nitride rotors with
Kel-F end-caps, and the spectra were collected using cross-polarization (CP) and
magic-angle spinning. The 31P spectra were acquired with a CP time of 5 ms and ait
a sample-spinning speed of 3 kHz, with proton decoupling. All 31P chemical shifts
were referenced to 85% H3PO4 at 0 ppm. The 13C spectra were acquired with a
cross-polarization (CP) time of 1 ms, sample-spinning speed of 5 kHz and proton
decoupling, with chemical shifts referenced to TMS at 0 ppm.
3. RESULTS
3.1. Zonyl UR and MSAP on planar aluminum substrates
Figure 4 presents a set of IR absorption spectra from thin films (estimated by
spectroscopic ellipsometry to be less than 50 nm thick) of Zonyl UR deposited
on aluminum substrates by rastered USAD or stirring. These absorption spectra
were calculated from the ratio of single-beam spectra from coated substrates to
Zonyl UR on Planar Aluminum Substrates
USAD, Refluxing Time 5.0 Hours
USAD, Refluxing Time 0.0 Hours
<
Stirred, Refluxing Time 5.0 Hours

Stirred, Refluxing Time 0.0 Hours
1000 1500 2000 2500 3000
Wavenumber (cm" )
Figure 4. Infrared spectra indicating significantly enhanced coverage of Zonyl UR on aluminum

surfaces using rastered USAD as compared to magnetic stirring. Note that almost no changes occur
in the US AD-coated surfaces during refluxing, whereas evidence of corrosive attack of the aluminum
substrate exists for the films deposited by stirring. The relative vertical scale of this figure is 140.
those from uncoated substrates. No baseline corrections were performed for any
absorption spectra presented in this paper, but we do uniformly shift the baselines
for clarity of presentation. It is inferred from Fig. 4 that sonicating deposits much
more Zonyl UR material on aluminum surfaces than stirring. The presence of the
same spectral features at 1095 cm - 1 , 1155 cm - 1 , 1220 cm" 1 and 1255 cm - 1 before
refluxing indicates that the same phosphate species adsorbed in both cases.
These results are consistent with our atomic force microscopy (AFM) studies [2],
which indicate dense Zonyl UR films when USAD is used, but patchy films
when deposition is by stirring. The IR spectral features observed in this study
are consistent with the literature relevant to the adsorption of phosphate species
on surfaces [1, 6, 9, 12, 14-16]. In Fig. 4 the main Zonyl UR features result
from vibrations of the fluorocarbon chains and P - 0 moieties, although definite
assignments are not possible due to overlapping spectral regions. The band
associated with the P = 0 stretch expected near 1125 cm - 1 is not present [1, 9],
consistent with the condensation reactions illustrated in Fig. 1.
The environmental stability of the phosphate films on aluminum surfaces was also
investigated by exposure to refluxing water. Figure 4 shows the IR spectra from the
films before and after refluxing for 5 h. For the dense films formed by USAD,
the peak intensities decreased by about 30% after 5 h, implying that some Zonyl
UR was removed by refluxing. In the case of the patchy films formed by stirring,
broad hydroxyl bands at 1600 cm - 1 and 3000-3600 cm - 1 appear. In addition, a
sharp feature at 1080 cm"1 and a large band near 800 cm - 1 dominate the spectra.
These features do not appear during refluxing of the dense Zonyl UR films grown
by USAD and are indicative of hydroxylation and oxidation of the substrate through
gaps in the patchy films.
Figure 5 shows that MSAP film deposition and environmental stability differ
substantially from that of Zonyl UR seen in Fig. 4. In the case of MSAP, rastered
USAD results in a slightly higher coverage than stirring. Prior to refluxing, the same
P-O features are evident in the spectra of both USAD and stirred MSAP films, with
the main peak at 1091 cm - 1 assigned to the PO3 group stretching vibrations. Note
that the P = 0 stretching mode is not present [1,9], again consistent with the reaction
scheme depicted in Fig. 1. Also in both cases before refluxing, the asymmetric
methylene stretching mode appears at 2918 cm - 1 , indicative of a fair amount of
molecular ordering within the films [6-8, 11, 12, 16].
Refluxing of MSAP coated aluminum for 5 h results in a shift of the 1091 cm - 1
band to 1087 cm" 1 , and the appearance of a new band near 1056 cm - 1 . These
changes occur for films deposited by stirring as well as by rastered USAD. This
indicates that some of the covalent tridentate bonds with the substrate found in
the as-formed films are lost, and that the refluxed films have terminal or bridging
bidentate ionic interactions with the substrate as well [11, 17]. In addition,
the methylene stretching modes shift to 2920 cm - 1 , also indicative of a less
homogeneous environment induced by refluxing. Note that the C-H stretching
bands are more intense after refluxing, which is attributed to a reorganization of
the ligands that permits a stronger coupling to the polarized IR beam.
232 M. /. Shepard et al.
MSAP on Planar Aluminum Substrates
T 1 1 1 r
1000 1500 2000 2500 3000
Wavenumber (cm" )
Figure 5. Infrared spectra indicating only a slight coverage enhancement of MSAP on aluminum
surfaces using rastered USAD as compared to magnetic stirring. Note that in both cases refluxing
causes the asymmetric methylene stretching mode to increase in intensity and to shift from 2918 cm" 1
to 2920 cm" 1 . Refluxing also induces the appearance of a new mode at 1056 cm - 1 in the P-O region.
The relative vertical scale of this figure is 28.
3.2. Zonyl UR and MSAP on aluminum oxide powder

In order to further investigate phosphate-surface reaction mechanisms, Zonyl UR
and MSAP films were also deposited on alumina powders and studied by IR
and NMR. Although not shown here, IR spectra of USAD formed Zonyl UR films
on alumina powder contain the same features as those formed on planar aluminum
substrates and indicate that the same type of species are dominant on these surfaces
following USAD [2]. However, the IR spectra from alumina powders stirred in
the Zonyl UR solution exhibit broad features. These broad spectral features result
from an inhomogeneous mixture of adsorbed species which is not present after
sonication. This is in contrast to MSAP treated powder, which yields essentially the
same IR spectra independent of deposition method (USAD vs. stirring). Our powder
IR data are, therefore, consistent with our work on planar aluminum substrates and
demonstrate that USAD dramatically enhances the growth of homogeneous films of
Zonyl UR on aluminum oxide but does not significantly improve the deposition of
MSAP coatings under the same conditions.
NMjINl llll|llll|lllljllll|llll|llll|llll|llll|IUI|llll|IIN|llll|llll|llll|llll|llll|IIM|mi|llll|
80 60 40 20 0 -20 -40 -60 -80 -100 ppm
6(3IP)
Figure 6. CP/MAS 3iP-NMR spectra from MSAP/alumina powder. The MSAP coatings were
generated either by (A) sonication or (B) stirring. * — spinning side bands.
Studies of alkylphosphates/alumina by solid-state NMR provide a detailed mole-

cular level characterization of these systems. Insights into the effect of sample
processing on surface coverage, the interactions between the phosphate groups and
the alumina, and the alignment of the alkyl chains can be obtained with this tech-
nique. The NMR data were collected using cross-polarization to enhance the signal
amplitude and magic-angle spinning to increase resolution. Spinning side bands,
separated from the isotropic peak by the sample-spinning frequency, appear in all
spectra.
The 31P-NMR spectra from the MSAP/alumina powder generated by US AD or
stirring are shown in Fig. 6. In contrast to the use of Zonyl UR as the surface
modifier, where the surface coverage is enhanced with sonication [21, the NMR
data show that the surface coverage obtained is nearly the same upon application
of MSAP by USAD or stirring. The impact of the deposition method on surface
coverage, as identified by NMR, is consistent with the IR results of Fig. 5.
Sonication does, however, affect the specific interactions between the phosphate
group and the alumina. Our studies on Zonyl UR indicate that the bonding between
the phosphate group and the alumina is heterogeneous — having mono-, bi- and
tridentate ligation (1, 2, or 3 P-O-Al bonds) — when the alkylphosphate films
are formed with stirring. When USAD is used, providing additional energy to the
system, the tridentate interaction is substantially favored and a more homogeneous
film is formed. The 31P-NMR spectrum from the stirred MSAP/alumina material
(Fig. 6B) consists of a broad peak with a linewidth of approx. 16 ppm. The
linewidth indicates that there is structural heterogeneity in the phosphate-alumina
234 M. 7. Shepard et al.
interactions, giving rise to a series of different chemical species with different

chemical shifts, yielding a broad line. The different interactions may include the
formation of one, two, or three bonds between the aluminum and the oxygens of the
phosphate group of MSAP, the presence or absence of hydrogen bonding between
phosphate groups, and interactions at different types of adsorption sites.
The spectrum from the USAD sample (Fig. 6A) has a set of narrow peaks at - 4
and —20 ppm. In the work of Reven and co-workers [14], these peaks were
shown to arise from the dissolution of alumina and the formation of a bulk form
of alumina-phosphate. The data of Fig. 6 show that in the sonicated material the
amount of dissolved alumina is very small and that dissolution does not occur
in the stirring procedure. The spectrum from the sonicated sample also has
resolved peaks at —7 and —14 ppm, whereas the chemical shift of the unreacted
MSAP occurs at 0 ppm. Comparison with previously published NMR data from
alkylphosphate films on metal oxide surfaces [2, 14,18-21] indicates that the peaks
at —7 and —14 ppm correspond to bidentate and tridentate interactions, respectively,
between the phosphate and alumina. Similar to Zonyl UR, the favored interaction
between MSAP and alumina, when the film is formed with USAD, is the tridentate
interaction, as the intensity of the peak at —14 ppm is substantially larger than the
peak at —7 ppm.
Solid-state NMR can also be used to characterize the structure of the alkyl chains
of the alkylphosphates upon attachment to the metal oxide surface. When the chains
adopt the d\\-trans conformation, this allows for the highest packing density and is
the most ordered configuration. In this case the methylene carbons will have a
chemical shift of 33-34 ppm in the 13C-NMR spectrum. Incorporation of disorder
in the chains and the formation of methylene groups in the gauche conformation
results in peak shifts to 30 ppm [22]. The 13C-NMR spectra from the sonicated or
stirred MSAP/alumina samples are shown in Fig. 7. The spectrum from the stirred
all trans—• 1
B / \ *— trans/gauche
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 i i
70 60 50 40 30 20 10 ppm
6(,3C)
Figure 7. CP/MAS 13C-NMR spectra from MSAP/alumina powder. The MSAP coatings were
generated either by (A) sonication or (B) stirring.
sample (Fig. 7B) has a peak at 34 ppm from the ordered (aM-trans) regions but it
also has a sizeable peak at 31 ppm from the disordered regions. In contrast, the 13C-
NMR spectrum from the sonicated material only has a peak at 34 ppm, indicating
that the chains have adopted the all-trans conformation (Fig. 7A).
3.3. Zonyl UR and MSAP on other planar substrates
The work presented above and in Ref. [2] provides a clear picture of the adsorption
behavior of these two phosphates on alumina. We now focus attention on extending
what we have learned to other industrially relevant substrates. In particular, we have
deposited Zonyl UR and MSAP on titanium, zirconium, and Zircaloy-2 surfaces by
magnetic stirring and stationary USAD methods. As an example of what we observe
in these studies, Figure 8 presents IR spectra of the P - 0 and C-F regions following
Zonyl UR deposition and Fig. 9 shows the C-H stretching region following MSAP
adsorption.
Zonyl UR on Planar Metal Substrates
T 1 1 1 r
1000 1100 1200 1300 1400
Wavenumber (cm" )
Figure 8. Infrared spectra of the C-F and P - 0 regions, indicating only a slight coverage enhancement
of Zonyl UR on titanium and zirconium surfaces using stationary USAD for 8 min as compared to
magnetic stirring for 2 h. Note that more enhancement is seen for Zircaloy-2, which contains tin as a
minor alloying element. The relative vertical scale of this figure is 1.
MSAP on Planar Metal Substrates
2700 2800 2900 3000 3100

1
Wavenumber (cm )
Figure 9. Infrared spectra of the C-H stretching regions indicating coverage enhancement of MSAP
on titanium, zirconium, and Zircaloy-2 surfaces using magnetic stirring for 2 h as compared to
stationary USAD for 8 min. The relative vertical scale of this figure is 2.
It can be seen in Fig. 8 that USAD barely enhances the adsorption of Zonyl UR on
zirconium, offers some improvement for titanium and makes the most difference in
the case of Zircaloy-2. It is interesting to note that the features are very broad for Zr
and Ti surfaces, but much better resolved for Zircaloy-2. In addition, comparison
with Fig. 4 shows that the relative intensities of the Zonyl UR bands differ on
Zircaloy-2 from those on aluminum. We suggest that these differences are due to
the small amount of tin that is part of the Zircaloy-2 composition. Note that for thin
films of organophosphates on surfaces, broad IR spectra with poor signal-to-noise
are not uncommon [6, 7, 9].
Equally interesting is Fig. 9, which indicates that, in the case of MSAP deposition
on Zr, Ti and Zircaloy-2 surfaces, stirring yields higher adsorbate coverages than
stationary USAD. This is in sharp contrast to the Zonyl UR case on these surfaces,
and also differs from the case of MSAP adsorption on aluminum. It is reasonable
that more MSAP is adsorbed in the stirring cases in Fig. 9 because of the time
difference between the stationary USAD (8 min) and the stirring (2 h). In all cases
the asymmetric methylene stretching mode is at 2913 cm - 1 , indicating a high degree
of ordering in the MSAP films.
4. DISCUSSION
USAD has the advantages that coatings can be applied in a relatively short period
of time and without the need for high temperature treatments. These features are
especially important in the deposition of phosphonates/phosphates on metal oxide
surfaces. For example, efforts to adsorb octadecylphosphonic acid onto y - A l 2 0 3 by
heating to 100°C resulted in dissolution of some of the alumina and the formation
of bulk aluminooctadecylphosphonate [14]. In our NMR studies of phosphate
adsorption on alumina there is little evidence for the dissolution of alumina induced
by USAD, and compared to stirring the surface coverage of Zonyl UR is higher
with the use of sonication [2]. While the autophobic behavior of a partially Zonyl
UR-covered alumina surface limits the extent of surface coverage during stirring,
USAD overcomes this barrier and yields surfaces with higher adsorbate coverages,
better film homogeneity and improved corrosion resistance.
For Zonyl UR on aluminum surfaces, the film heterogeneity is reduced by USAD
and the principal phosphate species present on the surface attach through three P -
O-Al bonds. Recalling from Fig. 3 that Zonyl UR is a mixture of species, the
energy supplied by sonication presumably desorbs the more weakly-interacting
phosphates and increases surface migration. The energy and momentum transferred
during bubble cavitation at the surface is schematically represented in Fig. 2 by
the last set of arrows in the time sequence. USAD leads to more of the tridentate
bonding structures by keeping the coverage of monodentate and bidentate species
low. In the case of stirring deposition of Zonyl UR on alumina powder, the more
weakly bound mono- and bidentate compounds are stabilized by the heterogeneity
of the surface and are not removed by the washing and drying steps following
adsorption. Such heterogeneity does not exist on the planar aluminum substrates,
leading to the removal of weakly-bound species by rinsing and drying following
stirring deposition.
NMR data show that the use of USAD results in more chemical/structural
homogeneity than stirring for MSAP films deposited on alumina. Both USAD and
stirring yield nearly the same surface coverage of MSAP, and the environmental
resistance is good in both cases. MSAP is a single component compound capable
of tridentate bonding and has alkyl chains long enough for self-assembly, whereas
Zonyl UR is a mixture of species (recall the structures from Fig. 3). We see that both
the deposition method and the structure of the phosphate play a role in achieving
surface coatings with good environmental stability. Prolonged exposure to hot water
and steam reverses the MSAP chemisorption process no matter how the films are
deposited by replacing covalent bonding with ionic interactions as evidenced by
the data of Fig. 5. This intrusion of water back into the interfacial region and
modification of the binding modes of phosphates to metal oxides can be represented
by Fig. 1 with the direction of the arrow reversed.
Comparing our data sets from the aluminum system with those from titanium,
zirconium, and Zircaloy-2 surfaces reveals the following trend. The isoelectronic
points of the relevant metal oxides follow the sequence Z r 0 2 ^ T i 0 2 ^ S n 0 2 <
Al 2 03 [23], which is the same ordering we find for IR intensities of Zonyl UR
and MSAP on these surfaces deposited in the same way. This is consistent with
the scheme depicted in Fig. 1 involving ionic interactions and demonstrates that
the role of the substrate cannot be neglected. It should be noted that contact angle
measurements were not useful for distinguishing these surfaces. For example, all
of the uncoated substrates had contact angles between 65 and 70°, whereas all of
the Zonyl UR coated surfaces yielded 108-115°, independent of whether stirring or
USAD was used.
For Zonyl UR deposition on alumina, USAD is an effective means of introducing
energy into the system and overcoming the autophobic barrier established by the
growing film. Since USAD is less effective on titanium and zirconium, it is
reasonable to suggest that autophobicity is not as significant in these systems. The
fact that the behavior of Zircaloy-2 (composition 69 ppm Hf, 570 ppm Ni, 1.4% Sn,
balance Zr) is distinguishable from that of nominally pure Zr indicates that even
a small change in surface composition might alter the adsorption of phosphate
films. It is also possible that Zircaloy-2 surfaces are preferentially enriched in Sn
by outward diffusion, and we are currently investigating this possibility with X-ray
photoelectron spectroscopy (XPS).
Our data presented here and elsewhere [2] for alumina indicate that phosphates
bond most strongly to alumina in the tridentate manner and demonstrate the
influence of the alkyl chains. In this paper we have also extended our efforts to the
titanium and zirconium materials systems and see different behavior. Thus, although
we do not yet have a complete understanding of phosphate adsorption on Ti and Zr
surfaces, we have discovered interesting trends and identified directions for future
research.
5. CONCLUSIONS
We have shown that local sonication can be used to enhance the coverage, homo-
geneity and environmental stability of Zonyl UR films formed on alumina surfaces.
In the case of MSAP on aluminum oxide surfaces, IR and NMR data indicate
that little coverage enhancement occurs during sonication but additional structural
homogeneity is introduced. For titanium, zirconium and Zircaloy-2 surfaces, the
coverage of Zonyl UR films slightly increases by USAD with the most influence
seen in the case of the alloy. The opposite effect is observed for MSAP deposi-
tion, where stirring yields higher surface coverages than USAD on all three sur-
faces.
Acknowledgements
Acknowledgement is made to the Donors of the American Chemical Society

Petroleum Research Fund for partial support of this research. We are also grateful
to Wah Chang for supplying the Zircaloy-2 material.
REFERENCES
1. R. D. Ramsier, P. N. Henriksen and A. N. Gent, Surf. Sci. 203, 72 (1988).
2. J. S. McNatt, J. M. Morgan, N. Farkas, R. D. Ramsier, T. L. Young, J. Rapp-Cross, M. P. Espe,
T. R. Robinson and L. Y. Nelson, Langmuir 19, 1148 (2003).
3. J. C Berg, in: Wettability, J. C. Berg (Ed.), Ch. 2. Marcel Dekker, New York, NY (1993).
4. K. Hass, W. Schneider, A. Curioni and W. Andreoni, Science 282, 265 (1998).
5. J. C Bolger and A. S. Michaels, in: Interface Conversion, P. Weiss and D. Cheevers (Eds), Ch. 1.
Elsevier, New York, NY (1969).
6. I. Maege, E. Jaehne, A. Henke, H.-J. P. Adler, C. Bram, C. Jung and M. Stratmann, Prog. Org.
Coatings 34, 1 (1998).
7. I. Maege, E. Jaehne, A. Henke, H.-J. P. Adler, C. Bram, C. Jung and M. Stratmann, MacromoL
Symp. 126, 7 (1997).
8. X. H. To, N. Pebere, N. Pelaprat, B. Boutevin and Y. Hervaud, Corrosion Sci. 39, 1925 (1997).
9. G. A. Nitowski, Ph.D. dissertation, Virginia Polytechnic Institute and State University, Blacks-
burg, VA (1998).
10. L. B. Goetting, T. Deng and G. M. Whitesides, Langmuir 15, 1182 (1999).
11. E. S. Gawalt, G. Lu, S. L. Bernasek and J. Schwartz, Langmuir 15, 8929 (1999).
12. J. G. Van Alsten, Langmuir 15, 7605 (1999).
13. R. L. Stefan and A. J. Szeri, J. Colloid Interface Sci. 212, 1 (1999).
14. W. Gao, L. Dickinson, C. Grozinger, F. G. Morin and L. Reven, Langmuir 12, 6429 (1996).
15. M. Textor, L. Ruiz, R. Hofer, A. Rossi, K. Feldman, G. Hahner and N. D. Spencer, Langmuir
16, 3257 (2000).
16. C. Yee, G. Kataby, A. Ulman, T. Prozorov, H. White, A. King, M. Rafailovich, J. Sokolov and
A. Gedanken, Langmuir 15, 7111 (1999).
17. B. L. Frey, D. G. Hanken and R. M. Corn, Langmuir 9, 1815 (1993).
18. G. A. Neff, C. J. Page, E. Meintjes, T Tsuda, W.-C. Pilgrim, N. Roberts and W. W. Warren,
Langmuir 12, 238(1996).
19. W. Gao and L. Reven, Langmuir 11, 1860 (1995).
20. D. Akporiaye and M. Stocker, Zeolites 12, 351 (1992).
21. I. Lukes, M. Borbaruah and L. D. Quin, J. Am. Chem. Soc. 116, 1737 (1994).
22. A. E. Tonelli and F. C. Schilling, Ace. Chem. Res. 14, 223 (1981).
23. G. A. Parks, Chem. Rev. 65, 177 (1965).
Ed. K. L. Mittal
© VSP 2004
Manufacture of resin-freefiberboardsfrom wood fibers

activated with Fenton's reagent (HÔi/Fe2*)
PETRI WIDSTEN * and JAAKKO E. LAINE

Laboratory of Paper Technology, Department of Forest Products Technology,
Helsinki University of Technology PO. Box 6300, FIN-02015 Hut, Finland
Abstract—Production of wood fibers at high defibration temperatures (171-202°C) results in partial

depolymerization of fiber lignin. The extent of lignin depolymerization increases with an increase
in defibration temperature and is greater for hardwood than for softwood fibers. The resulting low-
molecular-weight lignin fragments are believed to play a key role in radical formation in the fibers
during their treatment with Fenton's reagent (H2C>2/Fe2+) in water suspension or when the reagent
is sprayed onto the fibers in the defibrator blowline. Fiberboards of high internal bond (IB) strength
can be made by treatment of wood fibers with a low dose of Fenton's reagent (0.25% H2C>2/0.01%
FeS04 • 7H2O) before the fibers are fabricated into fiberboards. The IB strength and static bending
properties of the boards improve as the defibration temperature is increased, which is mainly attributed
to the concomitant increase in the amount of low-molecular-weight lignin, allowing more radicals to
be formed in the fibers. The adhesion effect observed when fiberboard is made from fibers treated
with Fenton's reagent may be largely due to covalent interfiber bonds formed by coupling of phenoxy
radicals in the lignin-rich fiber surfaces during hot-pressing. The thickness swell of fiberboards
depends on the amount of sizing agent added to the fibers and shows a low correlation with the
mechanical properties. The boards made with this process are best suited for indoor applications for
which high water resistance is usually not a critical factor.
Keywords: Wood fibers; fiberboard; hydrogen peroxide; Fenton's reagent; oxidative activation;
phenoxy radicals.
1. INTRODUCTION
Wood composite boards [1], such as fiberboard and particleboard, are manu-
factured industrially by gluing particles of mechanically disintegrated wood or
other lignocellulosic material together with thermosetting resins such as urea-
formaldehyde (UF) and phenol-formaldehyde (PF). Boards glued with these syn-
thetic resins show high mechanical strength. Many of them also exhibit good water
e-mail: petri.widsten@hut.fi
242 P. Widsten and J. E. Laine
resistance, necessary for outdoor applications. However, as the synthetic resins

are produced from unrenewable petrochemicals and the worldwide consumption
of wood composites is steadily increasing, resin-free bonding methods would be
an attractive alternative to conventional board manufacture. Another advantage of
resin-free bonding is that the used products need not be disposed of as toxic waste
but can be recycled or treated in the same manner as ordinary household waste. The
formaldehyde emissions during board manufacture and from the boards also pose
a threat to employees' and consumers' safety. In addition, synthetic resins, which
are more expensive than the wood particles, need to be used in large amounts for
most composite boards. For example, medium-density fiberboard (MDF) contains
approx. 10%ofUF.
The manufacturing processes for dry-process MDF and hardboard (high-density
fiberboard, HDF) are quite similar [1], Wood chips or other lignocellulosic raw
material is first preheated at high temperature (e.g., 170°C for MDF and 230°C
for HDF) to plasticize the lignin. During the following defibration stage, fibers
are separated along the lignin-rich middle lamella region. After this, adhesive and
other chemicals, such as sizing agents, are added to the fibers which are formed
into sheets and the sheets are pressed into boards at elevated temperatures. The
pressing temperature depends on the adhesive and product type and is typically in
the range 170-200° C. Fiberboards are called either dry-process or wet-process,
depending on whether the medium distributing and conveying the fibers is air or
water. The densities of MDF boards lie typically in the range 0.5-0.8 g/cm;,
while those of hardboards exceed 0.9 g/cm3. Hardboards are often called Masonite
boards after their inventor W. H. Mason whose company was the only hardboard
manufacturer for a long time after he invented hardboard in 1924.
An important application of thin (2-7 mm) HDF and MDF boards is doorskin, a
substitute for solid wood glued to both sides of wood frame to make a door. Thick
MDF boards (thickness approx. 12 mm) are manufactured for different furniture,
fixture, cabinet and display parts. The mechanical properties of fiberboards tend
to improve with an increase in board density, particularly when little or no added
binder is used as is often the case with hardboard. In such a case, hydrogen and
other secondary bonds are largely responsible for imparting mechanical strength to
the boards but also some covalent bonding, called auto-crosslinking [2], takes place
during the hot-pressing. Auto-crosslinking is thought to involve radicals and other
reactive groups formed from lignin, lipophilic extractives and sugars. In MDF and
other boards glued with synthetic resins, interfiber bonding is created mainly by the
resin [1], Sizing agents are routinely added to fiberboard products. Dry-process
HDF, for example, typically shows poor water resistance (large thickness swell on
water soaking) because the hydrophilic hemicelluloses are retained in the furnish.
Therefore, it is suitable for interior use only.
The so-called nonconventional bonding methods for various wood composites
have been reviewed by several authors [3-6]. Some of the applications were con-
cerned with thermosetting binder applications involving the use of polymerizable
chemicals or pulping residues such as lignosulfonates or spent sulfite liquor (SSL).
Resin-free fiberboards from wood fibers activated with Fenton's reagent 243
In other studies inorganic oxidants were used in conjunction with thermosetting

crosslinking agents. Direct bonding of wood surfaces using inorganic chemical
oxidants has been studied less frequently than applications involving added copoly-
merizing agents. A closely related field of study for direct bonding, which is another
interesting resin-free method for board fabrication but shall not be discussed here in
more detail, involves activation of wood fibers by laccase treatment prior to manu-
facture of fiberboards or particleboards [7-91.
During hot-pressing of wood particles into boards, interaction between radicals
or other reactive groups formed by an oxidative treatment in the fiber surfaces
generates covalent bonds between the particles. Applications of such approach
have been reported by Stofko [10] and Stofko and Zavarin [11], who used different
oxidants such as ferric chloride or hydrogen peroxide with ferric chloride as
peroxide decomposition catalyst for bonding of plywood. The products obtained
with the two oxidants mentioned were mechanically strong and resistant to boiling
water. Stofko [10] also used hydrogen peroxide together with ferrous sulfate and
hydrochloric acid as catalyst to make particleboard, but the internal bond strength
of the product was not very high. Johns and Nguyen [12] made particleboards
using peracetic acid as surface activator. Due to the lack of gap-filling capability
of the bonding process, only boards with high densities were resistant to boiling
water. Direct coupling of active sites on wood surfaces required that the surfaces be
brought close together. Because of the roughness of wood surfaces, this was only
achieved by pressing at high pressures, which resulted in boards with high densities.
Very recently, promising results involving the manufacture of particleboard from
softwood particles activated by treatment with Fenton's reagent (H202/Fe2+) have
been reported [13].
The above-cited and other published studies involving the use of inorganic
oxidants for the manufacture of wood composites have focused mainly on the
manufacture of particleboards and plywood. An application that had received
little attention until recently is the activation of high-temperature thermomechanical
pulp (HTMP) fibers used for the production of fiberboards [14, 15] with inorganic
oxidants. The bonding methods involving oxidative activation of wood particle
surfaces so far developed have not been put to industrial use despite the fact that
many of them have been a technical success. A reason for this is the board
manufacturers' reticence in using strong oxidants such as hydrogen peroxide at their
production plants. To help overcome this problem and also to reduce production
costs, the amount of oxidant should be kept to a minimum. Moreover, thick
(approx. 12 mm) board densities should preferably be in the MDF rather than in
the hardboard range to save raw material. In addition, to keep investment costs
low, resin-free methods for board manufacture should also be implemented without
major alterations to the existing production lines.
In this paper, a method for the production of 12-mm-thick dry-process fiberboards
from HTMP fibers activated by treatment with Fenton's reagent (H 2 0 2 /Fe 2+ ) is
presented. The aim was to generate phenoxy radicals in the fiber lignin so that
radicals on the fiber surfaces would form covalent interfiber bonds during pressing
of fibers into boards and a product with sufficient mechanical strength for indoor
applications could be obtained using hardwood or softwood as raw material. As
industrially manufactured HDF boards are much thinner than 12 mm, typically
around 3-7 mm, the properties of the boards are compared to the minimum
requirements for mechanical strength specified for 12 mm MDF boards, although
the densities of the softwood boards are mostly in the HDF range.
2. CHEMICAL CHARACTERIZATION OF HTMP FIBERS USED

FOR FIBERBOARDS
2.1. Changes in the structure of fiber lignin during defibration
Lignin is the polyphenolic substance found in all woody plants. It acts as a

cementing material, binding the fibers together and giving the plant its rigidity.
A structural scheme of native softwood lignin is shown in Fig. 1.
When wood is defibrated at temperatures above 170°C, the fiber lignin is partly de-
polymerized by cleavage of ether linkages connecting the lignin phenylpropane (C9)
units whereby new phenolic hydroxyl groups are formed [16, 17]. The extent of
depolymerization depends on the wood type (softwood or hardwood) and defibra-
tion temperature used. The change in /3-aryl ether bonds (the most common lignin
Figure 1. Structure of softwood lignin.

Resin-freefiberboardsfrom wood fibers activated with Fentons reagent 245
interunit ether linkage) as determined by solid-state 13C-CP/MAS-NMR and con-

comitant formation of phenolic hydroxyls as determined by periodate oxidation is
shown in Fig. 2 for samples of the original wood raw materials and HTMP fibers
produced at different preheating temperatures for softwood (spruce) and in Fig. 3
for hardwood (birch) [16, 17]. The results show that a gradual decline in the degree
of polymerization of fiber lignin takes place with an increase in defibration temper-
ature. The magnitude of depolymerization is clearly larger for birch than for spruce
fibers.
Figure 2. Depolymerization of softwood (spruce) lignin at different defibration temperatures in terms

of cleavage of /3-aryl ether bonds and phenolic hydroxyl content [16].
Figure 3. Depolymerization of hardwood (birch) lignin at different defibration temperatures in terms

of cleavage of /2-aryl ether bonds and phenolic hydroxyl content [17].
4.5-~c I s—i
D Birch
4.0-] / j • Spruce
o 0 j f
C _Q 3.5 j
Q. O
3.0
® #
2.5-
2 o 2.0J
Y
II
CO «
1.5-
1.0-
0.5-
o.o-
171 188 196 202
Defibration temperature (°C)
Figure 4. Depolymerization of spruce and birch lignins at different defibration temperatures in terms
of the content of water-extractable low-molecular-weight lignin in the HTMP fibers [16, 17].
Native lignin is hydrophobic and has a very high molecular weight. Because of
this, only small amounts of phenolic substances, mainly lignans, can be extracted
from wood with neutral water. However, the proportion of lignin that can be
extracted from wood with neutral water also increases as the lignin becomes more
hydrophilic (richer in phenolic hydroxyls) and its molecular weight is lowered by
cleavage of interunit ether linkages. There is also a much larger increase in total
water extractives, which mostly consist of hydrophilic hemicelluloses [16-18].
As shown in Fig. 4, the higher extent of depolymerization observed for birch lignin
relative to spruce lignin is also reflected in the amount of water-extractable phenolic
material present in the HTMP fibers. The effect of defibration temperature on the
water extractive content is also clear. Similar results have been obtained with other
softwood (pine) and hardwood (aspen and eucalyptus) species f 16, 17].
2.2. Surface lignin content of HTMP fibers
The spruce raw material and HTMP fibers produced by defibration between 171
and 202°C have a bulk lignin content of 25-28% by weight [16], while the birch raw
material and HTMP fibers obtained by defibration between 171 and 196°C contain
less lignin (20-22 wt%) [17]. As mentioned earlier, during high-temperature
defibration fiber separation occurs along the middle lamella region, which contains
over 40% lignin. As a result, the lignin coverage of HTMP fiber surfaces, as
determined by ESCA analysis, varies from 45% to 60%, which is much higher than
the bulk lignin contents of the fibers (Fig. 5). However, neither the differences in
the bulk lignin contents between the wood species nor the defibration temperature
affect the surface lignin coverage of the fibers.
Re sin-freefiberboards from wood fibers activated with Fenton 's reagent 247
171 188 196 202

Figure 5. Surface lignin coverage of birch and spruce HTMP fibers [16, 17].
3. RADICAL FORMATION IN HTMP FIBERS DURING TREATMENT WITH

FENTON'S REAGENT (H202/FE2+)
3.1. Formation of oxygen-based radicals by catalytic decomposition of hydrogen
peroxide
The decomposition of hydrogen peroxide is strongly catalyzed by transition metal
ions such as ferrous ions (Fe 2+ ). The combination of hydrogen peroxide and its de-
composition catalyst is called Fenton's reagent. Peroxide decomposition gives rise
to several types of oxygen radicals such as hydroxyl (HO*) and superoxide (RO°2IO'2)
radicals, which show high reactivity toward lignin [19].
3.2. Formation ofphenoxy radicals in lignin

Hydroxyl and superoxide radicals are highly reactive toward all fiber constituents,
particularly lignin. The reactions of superoxide radicals with lignin include ring-
opening of aromatic lignin units to conjugated muconic acid derivatives while
hydroxyl radicals bring about hydroxylation of aromatic lignin units (formation of
new phenolic hydroxyls) and hydrogen atom abstraction from phenolic hydroxyl
groups, which gives rise to phenoxy radicals, as illustrated in equation (1):
PhOH + HO' -> PhO* + H 2 0 (1)
The wood chips used as fiberboard raw material have a very low radical content.
However, during defibration, homolytic cleavage of lignin interunit ether bonds
generates carbon and phenoxy radicals (mechanoradicals). Most of the radicals
formed in lignin and carbohydrates are only short-lived but some of the phenoxy
radicals are stabilized in the rigid lignin matrix and are relatively stable.
The treatment of HTMP fibers with Fenton's reagent results in a large increase
in their phenoxy radical content, as shown by electron spin resonance (ESR)
spectroscopic analyses [14, 15]. Figure 6 shows the relative radical contents of
Birch Birch Birch Spruce Spruce Spruce Spruce

171°C 188°C 196°C 171°C 188°C 196°C 202°C
Figure 6. Phenoxy radicals in birch and spruce HTMP fibers, produced between 171 and 202° C,
formed by treatment with Fenton's reagent (3% H2O2/0.3% FeSC>4 • 7H2O) in water suspension at 5%
consistency for 5 min [15].
HTMP fibers after treatment with Fenton's reagent (3% H2O2/0.3% FeS0 4 • 7H 2 0)
in water suspension at 5% consistency for 5 min. The increase in phenoxy
radical content obtained by the treatment ranged from 250% to 620%. The
radicals present in the fibers before the treatment have been deducted from the
values. The absolute numbers of radicals formed, measured against cu,a'-diphenyl-
,6-picrylhydrazyl (DPPH, 1.5 x 1021 spins/g), ranged from 0.5 x 1017 to 1.9 x
1017 spins/g. Oniki and Takahama [20] report that various wood and nonwood
dioxane lignins treated with Fenton's reagent (H202/K3[Fe(CN)6]) or K3[Fe(CN)6]
contained (0.3-21) x 1017 spins/g. For the birch fibers the defibration temperature
has no clear effect on the number of radicals formed, whereas an increase in
defibration temperature up to approx. 200°C increases radical formation in spruce
fibers. These results agree with those from studies with other hardwood (aspen) and
softwood (pine) fibers [14, 15].
3.3. Effect of peroxide dose on radical formation

As shown in Fig. 7, a large increase in peroxide dose is needed to obtain only a
modest increase in the radical concentration of HTMP fibers when the peroxide
dose is increased gradually from 0.3% to 4.8%. In fact, radical formation is roughly
proportional to the square root of the peroxide dose.
3.4. Heat stability of phenoxy radicals

The decay of phenoxy radicals (formed by laccase-treatment [21, 22]) in birch and
spruce HTMP fibers was found to proceed slowly at 50° C and much faster at 100°C
(Fig. 8). After 40 min of heat treatment at 100°C, the surviving radicals in the
spruce fibers were stable within the experimental time frame. The radicals in the
birch fibers decayed at a steady rate for 60 min. As seen from the linearity of the
plot of reciprocal radical concentration of birch and spruce fibers vs. time of heat
treatment (Fig. 9), radical decay followed second-order kinetics, indicating that the
Resin-freefiberboardsfrom wood fibers activated with Fenton's reagent 249
100
ro tn
o ro
o .t
a:
1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0

Peroxide dose (% of fiber)
Figure 7. Effect of peroxide dose on radical formation on treatment of birch HTMP fibers, produced
at 188°C, with Fenton's reagent [15].
100
•coS
* 8
-B-196, 100°C
20 A -B-196, 50°C
-S-202,100°C
-S-202,50°C
10 20 30 40 50 60
Time of heat treatment (min)
Figure 8. Effect of heat treatment at 50° C and 100°C on phenoxy radical concentration of birch (B)
and spruce (S) HTMP fibers produced at 196°C and 202° C, respectively. The radicals were generated
by laccase-catalyzed oxidation [21, 22].
radicals probably decayed by coupling reactions. Based on these findings, it is

advisable that after radicals have been generated in the fibers, the fibers should be
maintained at low temperature to preserve the radicals until the fibers are fabricated
into boards in order to maximize the extent of interfiber bonding.
0 10 20 30 40 50 60
Time of heat treatment (min)
Figure 9. Reciprocal phenoxy radical concentration of birch and spruce HTMP fibers vs. time of heat
treatment at 100°C [21, 22].
4. FIBERBOARDS MADE FROM HTMP FIBERS TREATED WITH FENTON'S

REAGENT
4.1. Production of HTMP fibers and fiberboards
The wood raw material used for the manufacture of fiberboards consisted of
debarked chips of fresh spruce (Picea abies) and beech (Fagus sylvatica) [14]. The
HTMP fibers were produced using a pressurized pilot-scale Asplund double-disc
refiner and sprayed with Fenton's reagent solution at the wax addition point located
at the beginning of the blowline when the fibers are in a fluffed state. The preheating
temperature ranged from 171°C to 202°C. The H 2 0 2 / F e S 0 4 • 7 H 2 0 charge was
0.25%/0.01% based on dry fiber weight. After the oxidative treatment, the fibers
were formed into mats and the mats hot-pressed into 12-mm-thick boards in a
daylight press. Board densities were in the range 0.73-0.95 g/cm 3 .
4.2. Property standards of fiberboards

Composite panels have to conform to standards regarding their mechanical strength
and water resistance properties. The standards depend on the type of composite
product and also vary from country to country. MDF boards are generally tested
for internal bond strength (IB), which is the tensile strength perpendicular to the
board surface, and thickness swell (TS) on 24-h soak in cold water. Also the
static bending strength is often tested, involving the measurement of modulus of
rupture (MOR) and modulus of elasticity (MOE). Typical European MDF standards
(Europe-Norm No. CIN DIN 622-5) specify that for 12 mm MDF boards the IB
should be >0.6-0.65 MPa, MOR >30-35 MPa and MOE >2500 MPa, while TS
should not exceed 10-15%. The IB in particular is good for evaluation of the
frequency of interfiber bonds formed in MDF boards provided that the boards to
be compared have approximately equal densities. Board strength is governed by
bonding area, which increases as board density increases [2]. As MOR and MOE
depend largely on factors such as fiber length distribution and board density profile,
they are less reliable measures of the extent of interfiber bonding than IB.
4,3. Bonding mechanisms

The basic idea in the work discussed here was to achieve bonding of HTMP fibers
by coupling of phenoxy radicals generated on the fiber surfaces during pressing
of the fibers into boards. However, the exact nature of the adhesion occurring
when fibers activated by oxidative treatment are fabricated into fiberboards is not
known. This is due to the complexity of the chemical structure of oxidized fiber
surfaces. When fibers treated with Fenton's reagent are compressed under heat,
different types of interfiber covalent and secondary bonds are formed via interaction
of reactive groups on the fiber surfaces. Some of these potential interfiber bonding
mechanisms are illustrated in Fig. 10.
FIBER SURFACE
u
CH 3 o .CH30
FIBER SURFACE
B D
FIBER SURFACE
\"
CH3O
oil .CH3O
FIBER SURFACE
Figure 10. Adhesion processes possibly occurring by interaction of oxidized HTMP fiber sur-
faces (top) as they are brought together by hot-pressing (below). (A) Coupling of phenoxy radicals,
(B) esterification, (C) hydrogen bonding, (D) condensation of lignin and furfural.
Phenoxy radicals and other functional groups on oxidized fiber surfaces may
contribute to adhesion in various ways. It has been suggested [7] that phenoxy
radicals on fiber surfaces interact with radicals or other reactive groups on the
surfaces of other fibers. An increase in interfiber bonding area due either to bonding
of low-molecular-weight lignin (see Fig. 4) or lignans to the fiber surfaces or to
a loosening of lignin structure by the action of laccase resulting in a more even
surface topography have also been considered as possible explanations for enhanced
adhesion [7]. Bonding area could be increased also by an increase in polymer
mobility, due to breakdown of lignin and carbohydrate bonds during refining or
as a result of softening and fractionation of fiber lignin by oxidant treatment [23].
An increased bonding area would result in a higher frequency of hydrogen bonds
and other secondary forces binding fibers together [2].
The treatment of wood fibers with Fenton's reagent generates not only radicals but
also carboxyl and other carbonyl groups in fibers [2, 4, 24], Carboxyl groups may
esterify with hydroxyl groups on other fibers while carbonyl groups may crosslink
by forming acetal, hemi-acetal and ether bonds. Also, self-condensation of furfural
or other furan-type carbohydrate degradation products may take place, or these
compounds may condense with lignin or other phenolic compounds.
4.4. Effect of density on the mechanical properties of fiberboards

The effect of board density on the mechanical properties of spruce fiberboards is
shown in Figs 11 and 12 [14]. It is seen that the IB standard for 12 mm MDF
boards (0.6 MPa) is met at a board density in the MDF range whereas the static
bending properties reach the standards at considerably higher densities, lying in the
hardboard range.
4.5. Effect of defibration temperature on the mechanical properties of fiberboards

An increase in defibration temperature causes a large improvement in the IB strength
of spruce fiberboards (Fig. 13) and also improves their static bending properties
(Fig. 14) [14]. A large part of the adhesion effect obtained by treatment of
(0
1.4
Q_ 1.2
-D J ,
1.0 -I
— D) 0.8 -I R2 = 0.974
TO C
£ s> 0.6
— CO 0.4 -I
0.2
0.0 T 1 1 -
0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95

3\
Density (g/cm)
Figure 11. Effect of density on the IB strength of spruce fiberboards made from HTMP fibers
produced at 202°C and treated with 0.25% H2O2/0.01 FeS0 4 • 7H 2 0 [14].
03
O
LU
O
T5
C
03
Is
Q.
a: 0.95
o
Density (g/cm )
Figure 12. Effect of density on the static bending strength properties (MOE and MOR) of
spruce fiberboards made from HTMP fibers produced at 202°C and treated with 0.25% H2O2/0.01
FeS0 4 -7H 2 0[14].
1.2
1.0
0.8
03
0.6
CO
- 0.4
0.2
0.0
171 188 202
Figure 13. Effect of defibration temperature on IB strength of spruce fiberboards made from HTMP
fibers treated with Fenton's reagent [14].
fibers with Fenton's reagent before hot-pressing the fibers into boards may be due
to formation of interfiber bonds via coupling of radicals on adjacent fibers, as
suggested by the rate and mechanism of radical decay on heat treatment (Figs 8
and 9).
A control board made from spruce fibers produced at 202° C without treatment
with Fenton's reagent showed a 50% lower IB value (0.41 MPa) than a correspond-
ing board of equal density (0.91 g/cm3) bonded with Fenton's reagent [14]. More
control boards were not available for testing because they delaminated when the
press was opened after hot-pressing (a sign of insufficient interfiber bonding). The
delamination of boards consisting of weakly bonded fibers is caused by the build-up
of vapor pressure in the panel core during pressing.
Fiberboards made from hardwood (beech) fibers produced at 176°C and treated
with Fenton's reagent (0.25% H2O2/0.01 FeS0 4 • 7H 2 0) meet the IB, MOR and
5U -|
^ 45 -
CO
40 -
a.
o" 3b -
LU
30 -
o^
~o 2b -
cCD 20 -
^—V,
CD 15 -
g^ 10 -
01 5 -
o
^ 0 -
171 188 202
Figure 14. Effect of defibration temperature on MOR and MOE of spruce fiberboards (density
0.96 g/cm3) made from HTMP fibers treated with Fenton's reagent [14].
MOE standards for 12 mm MDF boards [14]. The average IB (1.2 MPa) of four
beech fiberboards is much higher than that of spruce boards made with fibers
produced at a higher temperature, 188°C (Fig. 13). A possible reason for this is
that at a given defibration temperature, more reactive (depolymerized and with a
high phenolic hydroxyl content) lignin is present in hardwood than in softwood
fibers (Figs 2-4). The MOR (32 MPa) and MOE (34 GPa) of the beech boards also
meet the requirements for the 12 mm MDF boards.
4.6. Thickness swell of fiberboards
The water resistance of fiberboards is generally determined by 24 h soak in cold

water. For 12 mm MDF boards, the maximum thickness swell (TS) allowed in this
test is about 12% by European board standards. However, hardboards do not have
such strict TS standards.
The TS of spruce boards is clearly reduced as the defibration temperature
increases. On the other hand, the amount of paraffin wax added as sizing agent to the
fibers has a much larger effect on TS than the defibration temperature. Boards made
with the lowest wax dose (0.5%) show extremely high TS. Doubling the amount of
wax from 0.5% to 1% causes a strong reduction in TS, and as the amount of wax is
increased to 4%, the TS declines further to <20%. The standard TS value for MDF
boards is thus not met. Also the addition of 2% CaCl2 to fibers causes a reduction in
TS comparable to the addition of 4% wax. The beech boards, for which the amount
of wax added was only 0.5%, all have very high TS despite their high IB strength.
There is, thus, no clear relationship between the mechanical properties and TS of
the boards. Unless a large amount of sizing agent is added, the boards are only
suitable for interior use under such conditions that they are not exposed to water.
5. CONCLUSIONS
Activation of wood fibers by treatment with a low dose of Fenton's reagent (0.25%
H2O2/0.01% FeS04 * 7H2O) before their use for fiberboard manufacture results
in a large increase in the board IB strength. IB strength also improves as the
defibration temperature increases, which can be attributed to an increase in the
amount of reactive material such as low-molecular-weight lignin in the fibers.
The adhesion resulting from treatment with Fenton's reagent may be largely due
to covalent interfiber bonds formed by coupling of phenoxy radicals in the fiber
surface lignin during pressing but the relationship between the board IB strength
and extent of radical formation is not straightforward. Other bonding mechanisms,
such as hydrogen bonding, contribute to adhesion to an unknown extent. The
thickness swell of fiberboards does not correlate with the mechanical properties, and
is strongly dependent on the amount of wax added to the fibers. Because of the high
thickness swell of most boards, which is typical for hardboards, they are suitable
for interior use only. The results suggest that resin-free bonding processes involving
the activation of fibers with inorganic oxidants may be used for the manufacture of
MDF and HDF boards for indoor applications.
REFERENCES
1. T. M. Maloney, Forest. Prod. J. 46, 19-26 (1996).
2. E. L. Back, Holzforschung 41, 247-258 (1987).
3. W. E. Johns, in: Wood Adhesive s — Chemistry and Technology, A. Pizzi (Ed.), pp. 75-96. Marcel
Dekker, New York, NY (1983).
4. E. Zavarin, in: The Chemistry of Solid Wood, Advances in Chemistry Series No. 207, R. Rowell
(Ed.), pp. 354-355. American Chemical Society, Washington, DC (1984).
5. J.-F. Matte and J. Doucet, Cellulose Chem. Technol 22, 71-78 (1988).
6. E. Roffael and B. Dix, Holz Roh-Werkst. 49, 199-205 (1991).
7. C. Felby, L. S. Pedersen and B. R. Nielsen, Holzforschung 51, 281-286 (1997).
8. A. Kharazipour, A. Huttermann and H. D. Ludemann, J. Adhesion Sci. Technol 11, 419^4-27
(1997).
9. P. Widsten, J. E. Laine, P. Qvintus-Leino and S. Tuominen, J. Adhesion Sci. Technol. 17, 67-78
(2003).
10. J. Stofko, 'The Autoadhesion of Wood", PhD Thesis, Univ. of California, Berkeley, CA (1974).
11. J. Stofko and E. Zavarin, US Patent 4,007,312 (1977).
12. W. E. Johns and T. Nguyen, Forest Prod. J. 27, 17-23 (1977).
13. U. Westermark and O. Karlsson, in: Proc. 12th Int. Symp. Wood Pulping Chem., Vol. 1, pp. 365-
368, Madison, WI (2003).
14. P. Widsten, J. E. Laine, P. Qvintus-Leino and S. Tuominen, Holzforschung 57, A41-452 (2003).
15. P. Widsten, J. E. Laine, P. Qvintus-Leino and S. Tuominen, Paperija Puu 85, 96-99 (2003).
16. P. Widsten, J. E. Laine, P. Qvintus-Leino and S. Tuominen, J. Wood Chem. Technol. 21, 227-245
(2001).
17. P. Widsten, J. E. Laine, P. Qvintus-Leino and S. Tuominen, Holzforschung 56, 51-59 (2002).
18. R.-C. Sun, G. L. Jones, R. M. Elias and J. Tomkinson, Cellulose Chem. Technol 33, 239-250
(1999).
19. J. Gierer, Holzforschung 51, 34-46 (1997).
20. T. Oniki and U. Takahama, Mokuzai Gakkaishi 43, 493^98 (1997).
21. P. Widsten, J. E. Laine and S. Tuominen, Nordic Pulp Paper Res. J. 17, 139-146 (2002).
22. P. Widsten, J. E. Laine and S. Tuominen, Cellulose Chem. TechnoL 36, 161-172 (2002).
23. H. Yamaguchi, M. Higuchi and I. Sakata, Mokuzai Gakkaishi 40, 185-190 (1994).
24. H. H. Nimz, in: Wood Adhesives — Chemistry and Technology, A. Pizzi (Ed.), pp. 247-288.
Marcel Dekker, New York, NY (1983).

K L Mittal Silanes and Other Coupling Agents, Volume 3

Hochgeladen von

Dokumentinformationen

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

K L Mittal Silanes and Other Coupling Agents, Volume 3

Hochgeladen von

Copyright:

Verfügbare Formate

Silanes and Other Coupling Agents, Volume 3

This page intentionally left blank

Editor: K.L. Mittal

No claim to original U.S. Government works

International Standard Book Number-13: 978-9-04-741402-5 (eBook - PDF)

Part 1: Silane Coupling Agents

Sterically hindered silanes for waterborne systems: A model study

Silane oligomers: A class of their own

FTIR and XPS surface characterization of allyltrimethoxysilane and

Effect of bridging group on the structure of bis-silane water-barrier films

Amino- and bis-silane pre-treatments for adhesive bonding of aluminum

y-APS as a durability enhancer of elastomer-to-metal adhesive bonds

Silane coupling agents for silica-filled tire-tread compounds:

Electrodeposition of organofuncational bis-silanes and their effectiveness

Silane coupling agents as corrosion fatigue inhibitors

Modified silane coatings as an alternative to chromates for corrosion

Musical instrument strings and corrosion: A comparative study

Part 2: Other Coupling Agents / Adhesion Promoters

Cyclic azasilanes: Volatile coupling agents for nanotechnology

Inorganic-organic hybrid polymers based on silanes for coating textile

Optimising the adhesion of glass fibres to an epoxy resin using plasma

Organophosphate adsorption on metal oxide surfaces

Manufacture of resin-free fiberboards from wood fibers activated

This book chronicles the proceedings of the Fourth International Symposium on

Silane Coupling Agents

Sterically hindered silanes for waterborne systems:

E. R. POHL * and A. CHAVES

Keywords: Hydrolysis kinetics; silane; acid-catalyzed; steric; polar; Taft plot.

these formulations was achieved by restricting or eliminating the exposure of the

2.2. Method for monitoring hydrolysis reactions

3. RESULTS AND DISCUSSION

R'Si(OR)3 + H 2 0 % R'Si(OR)2OH + ROH (1)

R'Si(OR)2OH + H 2 0 % R,Si(OR)(OH)2 + ROH (2)

R/Si(OR)(OH)2 + H 2 0 ^ R'Si(OH)3 + ROH (3)

3.2. Acid catalysis

"d[S] v r*i <K

pH* was equal to 3.2.

Alkoxy group, Steric substituent Polar substituent

3.3. Polar effects

log k/ko = pJ2a* + s^2E^ (8)

log*H/*HO = 0.10 J > * ' + 0.65 £ E'„ (9)

0.0 1.0 2.0 3.0 4.0

Silane oligomers: a class of their own

Abstract—The hydrolysis and condensation of organofunctional silanes, Y-(CH2)3-Si(OX)3 (Y =

Keywords: Alkoxysilanes; hydrolysis; condensation; NMR analysis; E-glass; characterisation of

Y = organofunctional group, OX = silicon-functional group (alkoxy), n = 0 or 3.

*Tel.: (49-7623) 91-8233; Fax: (49-7623) 91-8571; e-mail: helmut.mack@degussa.com

2. HYDROLYSIS AND CONDENSATION OF SILANES

3.29Si-NMR SPECTROSCOPY OF SILANE HYDROLYSATES [5,6]

Figure 1. Beginning of the hydrolysis reaction of epoxysilane DYNASYLAN® GLYMO.

possible to obtain information on the structure and amount of oligomeric siloxanes

4. 29Si-NMR STUDY OF 3-GLYCIDYLOXYPROPYLTRIMETHOXYSILANE

In demineralized water, at pH 6, hydrolysis of 3-glycidyloxypropyltrimethoxysilane

5. SILYLATED SURFACES: AES STUDY

6. AES DEPTH PROFILING STUDY

Secondary electron image AES 2-dimensional carbon mapping

100 \xm 100 nm Intensity

Film thickness Film thickness

0 50 100 150 200 250 300 350 400

of homogeneous silane monolayer coverage on substrates based on simple silane

Secondary electron image Secondary electron image

Figure 6. SEM images of cationic aminosilanes on E-glass.

are commercially available. The advantages of oligomeric silanes include chem-

Film thickness Film thickness

logH/HO = 0.10 J > * ' + 0.65 £ E'„ (9)