Beruflich Dokumente
Kultur Dokumente
///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
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
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
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.
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
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.
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
^ 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
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
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 groups. Redefined substituent constants Efs and a*' have previously been
reported [9] and are presented in Table 2.
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
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).
This page intentionally left blank
Silanes and Other Coupling Agents, Vol. 3, pp. 11 -20
Ed. K. L. Mittal
© VSP 2004
HELMUT MACK*
Degussa, AG, Untere Kanalstrasse 3, D-79618 Rheinfelden, Germany
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
i— Silanol
Silanediol —i
Trialkoxysilane
mr^
T T T
- 45 - 55 - 65 ppm
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*i^i**^^*mii#**
T
•20 •40 -60 •80 ppm
Figure 2. Oligomeric siloxanes of methacrylate silane DYNASYLAN® MEMO (analytically
monitored in the 29Si-NMR spectrum).
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).
Silane oligomers: a class of their own 17
Figure 4. SEM image and AES 2-dimensional carbon mapping of aminosilane DYNASYLAN®
AMEO.
50 nm 50 ^im
Figure 7. AES depth profiles of plate E-glass treated with monomeric and oligomeric cationic
aminosilanes (N, nitrogen).
Monomer Oligomer
xlO 4 xlO 4
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
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.
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]
1250 ...1000
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
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 ).
Characterization of an ALS and MPS mixture adsorbed onto titanium substrate 27
1250 ...1000
3400 ...3200
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 ).
1250 ...1000
A
1250< , . B
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
-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
CO
O
i (/) Q.
CN CN
CO co
t niMiiiitoil 1
I^^M.*'*'!^
'
•01s
Ti + ALS
Q.
CN
CO
"""v
LHN
^*-^U 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).
Characterization of an ALS and MPS mixture adsorbed onto titanium substrate 31
1.05i
30
0.53 H
20
0.00^
30 ^m
20 10
\iro
10
(Q
/ 0
(D)
Figure 7. (Continued).
32 J. P. Matinlinna et al.
0
(E)
0
(F)
Figure 7. (Continued).
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.
Characterization of an ALS and MPS mixture adsorbed onto titanium substrate 33
489-j
20
(.0
Mm
20 10
10
(G)
(H)
Figure 7. (Continued).
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
34 J. P. Matinlinna et al.
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.
36 J. P. Matinlinna et al.
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).
Characterization of an ALS and MPS mixture adsorbed onto titanium substrate 37
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.
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
*To whom correspondence should be addressed. Tel.: (1-513) 556-5431; Fax: (1-513) 556-3773;
e-mail: dale.schaefer@uc.edu
40 G. Pan et al.
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].
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.
D20 2 5 * C 1 4 h
A Redried
O As-prepared
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°.
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.
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.
Effect of bridging group on the structure ofbis-silane water-barrier films 47
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).
5. E. P. Plueddemann, Silane Coupling Agents, 2nd edn. Plenum Press, New York, NY (1991).
Effect of bridging group on the structure ofbis-silane water-barrier films 49
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
*To whom correspondence should be addressed. Tel: (47-2) 206-7849; Fax: (47-2) 206-7350;
e-mail: Bente.G.Tilset@sintef.no
52 B. G. Tibet et al.
Table 1.
Silanes and adhesive employed for pre-treatment of aluminium AA6060
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(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
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
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%
56 B. G. Tilset et al.
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
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.
58 B. G. Tibet et al.
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.
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
60 B. G. Tilset et al.
Figure 5. Photographs taken after ASST on samples pre-treated with (a) y-APS and (b) BTSE and
coated with Araldite 2014.
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
Amino- and bis-silane pre-treatments for adhesive bonding of Al 61
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
Amino- and bis-silane pre-treatments for adhesive bonding of Al 63
/-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,
64 B. G. Tibet et al.
140
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.
Amino- and bis-silane pre-treatments for adhesive bonding of Al 65
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.
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.
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.
R.E HAMADE*
Department of Mechanical Engineering, American University of Beirut (AUB), P.O. Box 11-0236,
RiadEl-Solh, Beirut 11072020, Lebanon
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
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.
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 + c^a + 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.
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
y-APS as a durability enhancer of elastomer-to-metal adhesive bonds 75
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
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).
y-APS as a durability enhancer of elastomer-to-metal adhesive bonds 11
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.
1&
\ •
(0
v.
\
\
CC
Q
D
0.1
i 3 r
Y- APS concentration (wt%)
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).
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.
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,
y-APS as a durability enhancer of elastomer-to-metal adhesive bonds 81
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.
y-APS as a durability enhancer of elastomer-to-metal adhesive bonds 83
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
-^A 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 '
Y- APS concentration (wt%)
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
y-APS as a durability enhancer of elastomer-to-metal adhesive bonds 85
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).
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.
y-APS as a durability enhancer of elastomer-to-metal adhesive bonds 87
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. 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).
This page intentionally left blank
Silanes and Other Coupling Agents, Vol. 3, pp. 89-103
Ed. K. L. Mittal
© VSP 2004
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].
*To whom correspondence should be addressed. Tel.: (31-53) 489-4621; Fax: (31-53) 489-3823;
e-mail: j.w.m.noordermeer@utwente.nl
90 W. K. Dierkes et al.
OEt
I
EtO —Si —(CH2)3
OEt /
I /
EtO—Si —(CH2)3
OEt
I
EtO — Si — (CH2)3-
OEt
^ O E t ^
(CH a ) 3 -
\
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),
92 W. K. Dierkes et al.
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
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.
Silane coupling agents for silica-filled tire-tread compounds 93
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)
Figure 3. Mixing curves of TESH- and TESPT-containing compounds. Temperatures indicate the
dump temperatures of the compound.
94 W. K. Dierkes et al.
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
Silane coupling agents for silica-filled tire-tread compounds 95
124
% 8
E 6
2-\
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.
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.
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)
Figure 6. Influence of dump temperature on the storage modulus at 0.56% strain (A) and 100%
strain (B): (•) TESPT; (•) TMeSP.
Silane coupling agents for silica-filled tire-tread compounds 97
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.
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
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,
Silane coupling agents for silica-filled tire-tread compounds 99
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)
Figure 9. Influence of dump temperature on the storage modulus at 0.56% strain (A) and 100%
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.
100 W. K. Dierkes et al.
14
12'
I 10
*-#-
J 8
6
E
^ 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.
Silane coupling agents for silica-filled tire-tread compounds 101
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
102 W. K. Dierkes et al.
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.
Silane coupling agents for silica-filled tire-tread compounds 103
REFERENCES
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.
*To whom correspondence should be addressed. Tel.: (1-513) 556-3194; Fax: (1-513) 556-3773;
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
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
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).
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.
-0.500
-0.600
t -0.700
UJ
£ -0.800|
>
15
•*-»
c
S
Immersed
+ 10V
(a)
(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.
Electrodeposition of organofunctional bis-silanes 111
IMMmt
1.00
-2.00 -1.00 0.00 1.00 2.00 3.00 4.00 5.00
Log Freq (Hz)
(a)
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.
112 J. S. Gandhi and W. J. van Ooij
-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).
Electrodeposition of organofunctional bis-silanes 113
(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
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.
114 J. S. Gandhi and W. J. van Ooij
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
Electrodeposition of organofunctional bis-silanes 115
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
116 J. S. Gandhi and W. J. van Ooij
OH
OH "OH
Oxide layer
!> ',1 '.^ '/• 1> 1> •> •.*• '•> "> '.^ 1> *> "•> "•*» '•*• 1% '/• 1>t> :> ' ^ 1^ '/• ".1 1> 'A "•
^'/j^VW_»J'?^WVWV'^Vf^'^^lf'/V^^J.
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)
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.
Electrodeposition of organofunctional bis-silanes 117
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).
This page intentionally left blank
Silanes and Other Coupling Agents, Vol. 3, pp. 119-133
Ed. K. L. Mittal
© VSP 2004
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
*To whom correspondence should be addressed. Tel: (1-513) 556-3194; Fax: (1-513) 556-3773;
e-mail: vanooiwj@email.uc.edu
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.
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
Silane 4 Bis-[trimethoxysilylpropyl]amine/vinlytriacetoxysilane (5% water
based) in a ratio of 5 :1
Silane 5 Bis-[trimethoxysilylpropyl]amine/vinlytriacetoxysilane (5% water
based) in a ratio of 4 : 1 , with 10 ml Sol-Gel
Silane 6 Bis-[trimethoxysilylpropyl]amine/vinlytriacetoxysilane (5% water
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
Silane 10 Bis-[trimethoxysilylpropyl]amine/vinlytriacetoxysilane (5% water
based) in a ratio of 4:1, with 200 ppm Ce(NO)3
Silane 11 Bis-[trimethoxysilylpropyl]amine/vinlytriacetoxysilane (5% water
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— •
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
Silane coupling agents as corrosion fatigue inhibitors 123
^'^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,
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
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
Silane coupling agents as corrosion fatigue inhibitors 127
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
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
Silane coupling agents as corrosion fatigue inhibitors 129
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
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.
Silane coupling agents as corrosion fatigue inhibitors 131
50000
45000
35000 -H
30000
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.
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).
11. W. J. van Ooij and D. Zhu, Corrosion 157, 413 (2001).
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).
This page intentionally left blank
Silanes and Other Coupling Agents, Vol. 3, pp. 135-159
Ed. K. L. Mittal
© VSP 2004
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].
*To whom correspondence should be addressed. Tel: (1-513) 556-3194; Fax: (1-513) 556-3773;
e-mail: vanooiwj@email.uc.edu
136 V. Palanivel and W. J. van Ooij
2 >•
,cW&-
cyclic
siloxane
H-bond
"-&T~0
* >$
ox ^o- L<TV*
«K <& x / / \ \
o
IV I
55
O "A.
X
/
^;s ^ . •••"
/ ?
^o-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
Following the hydrolysis, the silane film, when applied to the metal substrate,
undergoes condensation reaction to form siloxane groups.
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 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.
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.
2.2.3. 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.
-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
Log Current Density (A/cm2)
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.
(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
144 V Palanivel and W. J. van Ooij
(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).
O-Si
Si-OH + ) -S^OH
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
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
146 V Palanivel and W. J. van Ooij
7.00 -i 3
2V^22SSgg%
6.00- 1 " " ^ ^ •Sfe
O 5.00 - 4 **AtotoAa
*^fe w£*
&fr
-o
**«*
o 4.00 -
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.
Modified silane coatings for corrosion protection of Al alloys 147
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.
148 V Palanivel and W. J. van Ooij
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
Figure 12. Modulus of elasticity of bis-sulfur silane films loaded with and without silica nanoparticles
measured by nanoindentation.
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
Modified silane coatings for corrosion protection ofAl alloys 149
H 2 0, 0 2
silane film
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
150 V. Palanivel and W. J. van Ooij
HzO, O z
(b)
Figure 13. (Continued).
pores in the film. This leads to deterioration of the corrosion inhibition properties
of the film, an effect similar to the overpigmentation of paints.
silane film
high
silica
• Si02 particles adsorb silanes
• CPVC exceeded
• CPVC is low because of relative
particle size
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
152 V Palanivel and W. J. van Ooij
-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
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
Modified silane coatings for corrosion protection of Al alloys 153
-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
Log Current Density (A/cm2)
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.
154 V. Palanivel and W. J. van Ooij
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.
(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)
156 V Palanivel and W. J. van Ooij
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.
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
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.
(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.
Modified silane coatings for corrosion protection of Al alloys 159
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).
6. W. J. van Ooij and D. Zhu, Corrosion 157, 413 (2001).
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).
9. E. P. Plueddemann, Silane Coupling Agents, 2nd edn. Plenum Press, New York, NY (1991).
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).
This page intentionally left blank
Silanes and Other Coupling Agents, Vol. 3, pp. 161 -176
Ed. K. L. Mittal
© VSP 2004
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.
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.
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
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
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
Musical instrument strings and corrosion 167
CD
G
CD
>
JO
2 2
> iE CD
£a
CO
E
CO
E
aa a a. aa
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.
Musical instrument strings and corrosion 169
10
Steel/ Phosphor Bronze, 24 hours
c Titanium/ Phosphor Bronze, 21 days
o
§
o
o
O)
2
5
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].
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.
Musical instrument strings and corrosion 171
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
Musical instrument strings and corrosion 173
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.
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.
Musical instrument strings and corrosion 175
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).
20. K. L. Mittal (Ed.), Silanes and Other Coupling Agents, Vol. 2. VSP, Utrecht (2000).
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
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.
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.
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,
Cyclic azasilanes: volatile coupling agents for nanotechnology 183
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.
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.
(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
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
186 B. Arkles et al.
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"
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.
Cyclic azasilanes: volatile coupling agents for nanotechnology 187
Table 1.
Properties of azasilacyclopentanes
nBu
A
MeO OMe
42% 58-60° C/3 mmHg 0.932
tBu
/S's
MeO OMe
X i ^ CH 2 CH=CH 2
MeO OMe
v
\ Q - ^ Me
/ N
MeO OMe
EtO
A OEt
0% Not observed
/ \
MeO OMe
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
H Me OMe
MeO OMe
A ^Si-OH
I
CH0
I "
CH^CH 2 —Si-OH
CH2
CH3
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.
Cyclic azasilanes: volatile coupling agents for nanotechnology 189
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)-
190 B. Arkles et al.
H H
I I
NCH 2 CH 2 CH 2 CH 3 NCH 2 CH 2 CH 2 CH 3
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
NCH 2 CH 2 CH 2 CH 3 NCH 2 CH 2 CH 2 CH 3
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.
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
Cyclic azasilanes: volatile coupling agents for nanotechnology 191
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
1. K. L. Mittal (Ed.), Silanes and Other Coupling Agents. VSP, Utrecht (1992).
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).
This page intentionally left blank
Silanes and Other Coupling Agents, Vol. 3, pp. 193-203
Ed. K. L. Mittal
© VSP 2004
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.
*To whom correspondence should be addressed. Tel.: (49-2151) 843-159; Fax: (49-2151) 843-141;
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.
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
Silane-based inorganic-organic hybrid polymers for coating textiles 197
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.
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.
Silane-based inorganic-organic hybrid polymers for coating textiles 199
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.
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%.
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
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).
Silane-based inorganic-organic hybrid polymers for coating textiles 203
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).
This page intentionally left blank
Silanes and Other Coupling Agents, Vol. 3, pp. 205-223
Ed. K. L. Mittal
© VSP 2004
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].
Table 1.
The tensile properties of the LY1556/GY298 matrix resin
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
Adhesion of glass fibres to epoxy resin using plasma copolymers 209
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.
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
Adhesion of glass fibres to epoxy resin using plasma copolymers 211
Table 2.
Single filament tensile strength of plasma polymer coated and unsized E-glass fibres
allylamine series, the trend is reversed with the 74% 1,7-octadiene plasma polymer
providing the highest improvement with the largest Weibull modulus.
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.
Adhesion of glass fibres to epoxy resin using plasma copolymers 213
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
214 D. J. Marks and F. R. Jones
z
CSTF - " ' - 1 J?=N , (3)
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.
216 D. J. Marks and F. R. Jones
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.
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)
Table 6.
XPS elemental analysis (in at%) of allylamine/octadiene plasma copolymers deposited onto E-glass
fibres
Table 7.
Fraction of components (%) in the Ci s narrow scan for a series of allylamine/octadiene plasma
copolymers deposited onto E-glass fibres
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
Adhesion of glass fibres to epoxy resin using plasma copolymers 219
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.
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.
4.1.4. 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
Adhesion of glass fibres to epoxy resin using plasma copolymers 221
Table 8.
Ranking of plasma polymer functionalised E-glass fibres from STE with interfacial micromechanics
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
Table A2.
Binding energies and assignments for Ci s curve-fitting of spectra from acrylic acid based plasma
copolymers and their fluorinated derivatives
Table A3.
Binding energies and assignments for Ci s curve-fitting of spectra from allyl amine based plasma
copolymers
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.
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
*To whom correspondence should be addressed. Tel: (1-330) 972-4936; Fax: (1-330) 972-6918;
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
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
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.
Zonyl UR
O O O
P i I
RO^l^OR RO^i^OR RO^i^OH
OR OH OH
R = CF3(CF2CF2)z(CH2CH2)
Z=1-7
MSAP
O
II
CH3(CH2)i7Cr l^OH
OH
Figure 3. Structural representations of the phosphates studied in this work.
Organophosphate adsorption on metal oxide surfaces 229
was purchased from Baikowski International (Charlotte, NC, USA) and used as
received for the solid-state NMR and infrared studies.
(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
<
Wavenumber (cm" )
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.
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.
NMjINl llll|llll|lllljllll|llll|llll|llll|llll|IUI|llll|IIN|llll|llll|llll|llll|llll|IIM|mi|llll|
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.
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.
Organophosphate adsorption on metal oxide surfaces 235
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).
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.
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.
236 M. J. Shepard et al.
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.
Organophosphate adsorption on metal oxide surfaces 237
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
238 M. J. Shepard et al.
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
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).
This page intentionally left blank
Silanes and Other Coupling Agents, Vol. 3, pp. 241-256
Ed. K. L. Mittal
© VSP 2004
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
*To whom correspondence should be addressed. Tel.: (358-9) 451-4291; Fax: (358-9) 451-4259;
e-mail: petri.widsten@hut.fi
242 P. Widsten and J. E. Laine
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.
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].
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
Figure 5. Surface lignin coverage of birch and spruce HTMP fibers [16, 17].
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].
100
ro tn
o ro
o .t
a:
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].
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].
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.
252 P. Widsten and J. E. Laine
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.
(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 -
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].
Resin-freefiberboardsfrom wood fibers activated with Fenton's reagent 253
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
Defibration temperature (°C)
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
254 P. Widsten and J. E. Laine
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
Defibration temperature (°C)
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.
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).
256 P. Widsten and J. E. Laine
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).