You are on page 1of 14

ARTICLE IN PRESS

International Journal of Adhesion & Adhesives 26 (2006) 215 www.elsevier.com/locate/ijadhadh

Understanding the relationship between silane application conditions, bond durability and locus of failure
M.-L. Abela, R.D. Allingtonb, R.P. Digbyc, N. Porrittd, S.J. Shawb,, J.F. Wattsa
School of Engineering, University of Surrey, Guildford, UK Defence Science and Technology Laboratory, Porton Down, Wiltshire, UK c Airbus, Filton, Bristol, UK d Future Systems Technology Division, QinetiQ, Farnborough, Hampshire, UK
b a

Accepted 10 March 2005 Available online 17 May 2005

Abstract The extent to which an organosilane surface treatment regime can promote durability enhancement of an adhesively bonded aluminium alloy system has been determined. Results have revealed the range of application and lm-conditioning parameters which contribute to joint durability in a simple Boeing wedge joint. Organosilane solution parameters relating to solvent type, solution concentration, pH and hydrolysis time have all been shown to inuence resultant durability. Interestingly, parameters such as lm drying temperature and in-process time delay (time interval between application of the organosilane to the alloy surface and subsequent bonding) have little inuence on joint performance. The factors responsible for the durability variations observed have been considered using various surface analytical techniques. Supercially, failure surfaces indicative of interfacial failure between substrate and adhesive have been observed. More detailed characterisation using both XPS and SIMS has indicated failure processes associated with a diffusion zone comprising aluminium oxide and the organosilane. Crown Copyright r 2005 Published by Elsevier Ltd. All rights reserved.
Keywords: Organosilanes; Surface treatments; Moisture resistance; Durability

1. Introduction The use of adhesive bonding in the manufacture of load bearing structures can provide numerous benets in comparison to more traditional joining techniques such as mechanical fastening. Such advantages include improved performance characteristics, resulting largely from the weight reductions adhesive bonding can provide, together with signicant reductions in both procurement and life-cycle maintainability costs. Unfortunately a lack of condence in the ability of bonded joints to withstand the range of environmental and loading conditions typically encountered in many
Corresponding author. Tel.: +441980614989; fax: +441980613611. E-mail address: sjshaw@dstl.gov.uk (S.J. Shaw).

applications has prevented their widespread use. One issue of concern has been associated with atmospheric moisture. Many studies and much experience with bonded structures have demonstrated the adverse effect moisture can have on a bonded joint [1]. In particular, in many of these investigations, the precise nature of environmentally driven failure has been shown to be associated with the substrateadhesive interphase, thus showing this region to be the primary zone of weakness. Thus, for bonded aluminium structures, in order to promote long-term durability, surface treatment of the alloy prior to bonding has generally been regarded as a vital component of the manufacturing process. Currently much evidence exists indicating that the strongest and most durable adhesive bonds to many types of metallic substrate are provided by pre-treating alloys with a range of aggressive and toxic chemicals.

0143-7496/$ - see front matter Crown Copyright r 2005 Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.ijadhadh.2005.03.009

ARTICLE IN PRESS
M.-L. Abel et al. / International Journal of Adhesion & Adhesives 26 (2006) 215 3

This has been particularly true of aluminium alloys where extensive use has been made of hexavalent chromium compounds and strong acids in surface treatment specications. However, in recent years there has been growing pressure from government and environmental bodies to remove such substances from manufacturing processes. To this end considerable effort has been directed toward the application of organosilanes as a potential pre-treatment process [27]. Indeed, organosilanes have been employed as the basis for adhesion promoting surface treatments for many years. In particular, their use in promoting moisture stability in glass reinforced plastics applications are well known [3]. In addition, they have also been considered and used in various adhesive bonding applications, both in terms of initial manufacture and repair. Thus, although considerable experience has been gained in the use of organosilanes as adhesion promoting substances, most if not all current applications are based on empirically determined silane formulations and application procedures. Thus, the most efcient methods of application, together with the mechanisms by which these materials enhance (and on some occasions do not enhance) the strength and durability of bonded joints, has not been fully understood. In an attempt to enhance our knowledge of organosilanes as the basis of surface treatments for aluminium alloys, our laboratories have led and participated in an extensive collaborative research project. Known as the International Collaborative Programme on Organosilane Adhesion Promoters (ICOSAP) [8], this has had, as its primary aim, the development of a detailed understanding of the relationships between silane application variables, solution and surface chemistry, joint durability and mechanisms of failure. The results obtained from much of this investigation are considered and discussed in this paper.

3762-79. The test arrangement employed is shown in Fig. 1. Prior to surface treatment and bonding, each 25 mm 150 mm adherend was chamfered at one end to aid the eventual insertion of the wedge for durability assessment. The substrates were then degreased using ne grade Scotch Brite abrasive pads with commercial grade liquid detergent, followed by rinsing in running tap water whilst subjecting the adherends to further abrasion with a fresh piece of Scotch Brite. The adherends were then dried with unpigmented tissue paper followed by grit blasting with 50 mm alumina grit. Prior to use, g-GPS was removed from cold storage and allowed sufcient time to reach ambient temperature. The main silane solution and application variables considered in this programme were: a) b) c) d) e) f) solution pH silane concentration the nature of the solvent hydrolysis time drying temperature time lag between silane application and bonding.

2. Experimental 2.1. Materials The organosilane employed throughout this investigation has been a proprietary system based on gglycidoxypropyltrimethoxysilane (g-GPS). The substrate material employed was 2024 T3 unclad aluminium alloy. The adhesive used was a 120 1C curing proprietary lm adhesive based on a toughened epoxy. This system was selected due to the absence of silane coupling agent in the formulation. 2.2. Experimental Procedures All joint durability experiments were conducted using the Boeing wedge joint in accordance with ASTM D

To provide a starting point for the investigation, the silane pretreatment process initially selected was based on a 1% silane solution (aqueous) in which pH was controlled at a value of 5. Prior to application of the solution to the alloy adherends, a pre-hydrolysis time of 1 h was allowed and, following application, the resultant silane lm was dried at a temperature of 93 1C. Each process variable was evaluated in turn, optimised and then xed so that the next variable could be examined and optimised. The pH was controlled by acid (acetic) or alkali (sodium hydroxide) adjustment, with a pH range of from 3 to 11 studied. g-GPS concentrations of from 0.1 to 12% vol/vol were examined, with solution hydrolysis times ranging from 10 min to 48 h. In addition to pure aqueous solutions, water/methanol solvent combinations were also examined with effort focusing on methanol concentrations of 10, 50 and 90%. Following preparation of the silane solution, and the required period of hydrolysis, the silane solution was applied by brush onto the grit-blasted adherend surface, with brushing continuing for a 10-min period. The adherends were then placed on edge and tapped on dry tissue paper so as to remove excess silane solution. The adherends were then placed horizontally, pre-treated side up, on aluminium foil and inserted into a fanassisted oven for accelerated drying at 93 1C for 1 h. Following drying, the adherends were removed from the oven and placed on clean aluminium foil and allowed to cool to below 30 1C at ambient temperature. Although most of the experiments were conducted with a drying temperature of 93 1C, a set of experiments were carried

ARTICLE IN PRESS
4 M.-L. Abel et al. / International Journal of Adhesion & Adhesives 26 (2006) 215
20 Unbonded

Adhesive 150

20

25

3.2

25

Fig. 1. Boeing wedge joint.

out with drying at 23 1C so as to assess the importance of this variable on joint durability. Following surface pre-treatment, adhesive lm previously cut to the required dimensions was sandwiched between two prepared adherend coupons. The resultant wedge test specimens were then sandwiched between two stainless steel plates, with PTFE coated non-porous glass cloth used as a release ply. All test specimens were shimmed using aluminium spacers to give a bond line thickness of 0.2 mm and subsequently cured at 120 1C for 60 min under a pressure of 0.21 MPa. Throughout this work ve replicates per experimental condition were employed. Following cure, the long edges of the joints were subjected to a polishing procedure so as to aid inspection of the bondline during ageing. A wedge was then carefully driven into the unbonded end of the joint, with the entire specimen then being placed in a desiccator at 23 1C for 1 h. The length of the initial crack introduced into the bondline by insertion of the wedge was then determined using a travelling microscope. Any further advance of the crack was then assessed as a function of time during exposure of the joint to 96% RH at 50 1C for 7 days (achieved using a potassium sulphate saturated salt solution). Following removal of the joints from the ageing environment, the specimens were prized apart at the bondline so as to provide evidence as to the nature of the failure process responsible for any advancement of the initial crack. In addition to visual observation of the failure surfaces, X-ray photoelectron spectroscopy (XPS) and time of ight secondary ion mass spectrometry (ToF-SIMS) were employed with selected joints

to provide a detailed indication of the nature of the surfaces exposed after the environmentally induced failure. In the case of XPS, surface analysis was carried out using a VG Scientic ESCALAB Mk II system operated in the constant analyser mode, at a pass energy of 50 eV. An MgKa X-ray source was used and take off angle was set at 451. ToF-SIMS analysis was carried out on a VG Scientic type 23 system equipped with a double stage time of ight analyser and a MIG 300 PB pulsed liquid metal gallium ion source. Specimens were etched in the pulsed analyser mode with spectra collected after 0, 1.5, 2.5, 3.5 and 4.5 h pulsed etching.

3. Results and discussion As previously indicated, the starting point for the silane solution variable study was an aqueous solution with 1% silane and pH 5, subjected to a pre-hydrolysis time of 1 h. Following application of the solution to the aluminium alloy surface, the resultant silane lm was dried at 93 1C for 1 h. Boeing wedge test results obtained from joints subjected to this pre-treatment procedure are shown in Fig. 2. Also shown is data relating to both a simple gritblast and degrease surface treatment, together with results obtained from joints previously subjected to acid anodising surface treatment techniques. In this gure, and throughout this discussion, fracture energy values obtained from the wedge tests are indicated as functions of time in the environment. The fracture energy values were calculated from knowledge

ARTICLE IN PRESS
M.-L. Abel et al. / International Journal of Adhesion & Adhesives 26 (2006) 215
2.5

Silane pre-treatment Grit-blasted only

Fracture energy (kJm-2)

2.0

CAA pre-treatment PAA pre-treatment

1.5

1.0

0.5

0.0 0 20 40 60 80 100 120 140 160 180

Exposure time (hrs)

Fig. 2. Wedge results for standard silane plus comparisons.

of various joint dimensions via use of the following equation [9]: GI 3d 2 Eh3 , 16a 0:6h4 (1)

where G I is fracture energy, a is crack length, d is the width of the joint and h is adherend thickness. The results clearly indicate the poor durability characteristics which result from a very simple degrease/grit blast treatment. As expected, both of the currently employed aerospace aluminium alloy treatments, chromic acid anodising (CAA) and phosphoric acid anodising (PAA), promote substantial improvements in durability. Interestingly, the use of the organosilane applied under the conditions highlighted above, produces a durability enhancing effect virtually identical to that provided by the CAA technique. Other workers have considered comparisons between silane and other traditional treatments. For example, Pearce and Tolan, working with a range of proprietary silane coupling agents, compared durability results with the chromic acid FPL etch [10]. They found that the most effective coupling agent was based on g-GPS when employed with an epoxy-based structural adhesive. Lapshear strengths at both ambient and elevated temperatures were found broadly equivalent to those provided by the chromic acid etch treatment. Hot/wet durability testing, as determined by the Boeing wedge test, showed the g-GPS treatment to be superior to the chromic acid etch. Similarly, work by Theidman et al, who investigated the inuence of epoxy-functional silane based treatments on aluminium/epoxy wedge joints, found hot/wet durability characteristics superior to the FPL chromic acid etch [11]. Of particular relevance to the work discussed here, Baker and Chester considered silane-based treatments for potential aircraft repair applications [12]. They

concluded that g-GPS treatment, when combined with an epoxy-based primer, resulted in durability characteristics approaching those of acid anodising techniques. Similarly, Kuhbander and Mazza, from an extensive study of silane treatments on 2024-T3 aluminium alloy, revealed that treatments based on g-GPS, together with a chromated epoxy primer, could provide durability results similar to phosphoric acid anodising [13]. More recently, Arnott and Kindermann have observed circumstances where a silane-based treatment could promote durability characteristics superior to a phosphoric acid anodise treatment [14]. However, in their work, the anodising process was not preceded by the traditional chromic acid etch and thus cannot be compared directly with our work. Thus, considerable evidence from previous studies would appear to support the observations from our work, namely that organosilanes can, under certain conditions, promote durability characteristics in aluminium alloy joints at least equivalent to anodising techniques. Although the results highlighted in Fig. 2 were of major interest, further work was carried out to assess the extent to which variations in the solution and application conditions employed would modify the silane treatment results. The experiments conducted and the results obtained are discussed in the following sections.

3.1. Inuence of solution pH Several investigations have indicated and discussed the extent to which the pH of an organosilane solution can inuence the nature of the various reactions which occur in that solution [1517]. Thus in an attempt to assess the extent to which such chemistry variations could inuence joint durability, pH variation was considered a vitally important variable in this study. As indicated previously, pH variations of from 3 to 11 were investigated by the use of acetic acid/sodium hydroxide alkali adjustments as appropriate. The results obtained from this exercise are shown in Fig. 3.
2.5

Grit-blasted only pH 3.0 pH 5.0 pH 7.0 pH 9.0 pH 11.0

Fracture energy (kJm )

2.0

-2

1.5

1.0

0.5

0.0 0 20 40 60 80 100 120 140 160 180

Exposure time (hrs)

Fig. 3. Boeing wedge test crack growth for various solution pH values.

ARTICLE IN PRESS
6 M.-L. Abel et al. / International Journal of Adhesion & Adhesives 26 (2006) 215

These reveal the substantial extent to which solution pH inuences the durability enhancing characteristics of a 1% g-GPS aqueous solution. Perhaps the rst point to note is that all the silane solutions, irrespective of solution pH, promoted durability characteristics superior to the base line, grit-blast only specimens. However, the best performing system was the pH 5.0 solution with pH solutions of 3, 9, and 11 exhibiting similar but slightly lower G Ic values after 72 h joint exposure. Interestingly, a pH of 7 was shown to promote environmental resistance signicantly inferior to all other solution pH conditions. To assess the extent to which solution pH exhibits similar effects at other silane solution concentrations, further Boeing wedge experiments were conducted with silane concentrations ranging from 0.1 to 12%. Fracture energy, G Ic , data obtained from the longest exposure times employed (approximately 168 h) are shown in Table 1 for some of the solution concentration/pH combinations examined. The results reveal extremely complex trends with, at each of the solution concentrations studied, there existing an optimum solution pH value, these existing at pH values of 5 for silane concentrations of 0.1 and 1% and pH 7 for a silane concentration of 12%, suggesting an increase in optimum pH with increasing silane concentration. In attempting to discuss these effects it is necessary to consider the reactions which are believed to exist within the aqueous silane solution prior to application to the alloy surface (Fig. 4). As indicated, two main reactions occur within the silane solution. First, in the presence of water the alkoxy groups are sequentially hydrolysed resulting in the gradual build up of silanol species. Second, the silanols condense to form oligomers, which gradually increase in size i.e. converting to dimers/trimers and eventually larger molecules. In considering these reactions, Erickson and Plueddemann have proposed that solution pH effectively controls both the hydrolysis and condensation reactions [15]. For example, for organofunctional trialkoxy silanes in aqueous media, alkoxy silanes hydrolyse rapidly under mildly acidic conditions to form monomeric silanols, but condense slowly to oligomeric variants. However condensation reactions
Table 1 The effect of pH on 168 h fracture energy for 0.1, 1 and 12% silane concentrations Silane pH 168 h fracture energy (kJ m2) 0.1% silane 3 5 7 9 11 0.142 0.427 0.208 0.210 0.164 1% silane 0.655 0.708 0.290 0.509 0.541 12% silane 0.183 0.130 0.569 0.468 0.387

tend to proceed rapidly at pH values in excess of 7. These general observations have been conrmed by Xue et al who observed substantial increases in hydrolysis rates with aqueous g-GPS solutions on reducing pH from 6 to 3.5 [16]. Tesoro and Wu have reviewed various aspects of silane solution chemistry and have commented on the inuence of solution pH in relation to both hydrolysis and condensation reactions [17]. In highlighting the importance of these reactions they also suggested that, for practical adhesion promoting applications, the effectiveness of the silane was inuenced by the extent of condensation; high levels of condensation having detrimental effects on adhesion promoting capability. This would seem logical since the condensation reactions indicated in Fig. 4 would clearly lead to the elimination of silanol groups which, if retained, would promote reaction with surface oxide hydroxyls to produce a covalent bond. The greater the number of covalent bonds generated across the interface, the greater the likelihood of a durable bond. Hence the apparent necessity for enhancing silane hydrolysis whilst minimising resultant silanol condensation. Results obtained from the relatively low silane solution concentrations in this study do appear to support this view with, at silane concentrations of 1%, solution pH values of 3 and 5 providing the highest GIc values and hence better durability behaviours. This appears not to be the case at higher solution concentrations however and the factors which could be responsible, are discussed in the next section. 3.2. Inuence of silane concentration Results obtained from Boeing wedge experiments conducted on joints prepared from silane solution concentrations ranging from 0.1 to 12%, at a solution pH of 5, are indicated in Fig. 5. At this particular pH the results reveal the important inuence of silane concentration on eventual joint durability performance. Clearly, at this pH, concentrations of 0.5% and 1% provide the highest G Ic values after 168 h exposure and thus the best durability. Although silane concentrations above and below these values promote lesser degrees of durability enhancement, the results in Fig. 5 continue to indicate superior durability enhancement in comparison to base-line data i.e. joints simply pre-treated via grit-blast/degrease processes prior to joint formation. Having said this, the differences between grit-blast/degrease joints and those prepared from 12% silane solution concentrations are clearly marginal (at this pH), therefore, indicating clearly the importance of solution concentration on eventual joint durability performance. The extent to which the durability enhancement/silane concentration relationship varies as a function of solution pH is indicated in Fig. 6.

ARTICLE IN PRESS
M.-L. Abel et al. / International Journal of Adhesion & Adhesives 26 (2006) 215 7

R - Si(OR)x

Organic group reacts with adhesive

Inorganic group reacts with metal substrates

(1 ) HYDROLYSIS

R-Si(OR)3 3H2O

3ROH

(2) CONDENSATION

R-Si(OH)3 2Si(OH)3

2H2O

R HO Si OH O

R Si OH O

R Si OH OH

+
OH OH SUBSTRATE
Fig. 4. Organosilane hydrolysis and condensation reaction mechanisms.

OH

As highlighted above, a pH 5 solution results in an optimum silane concentration of 1% with a 168-h exposure G Ic value of 0.71 kJm2. Interestingly, two of the other pH solutions examined, namely 3 and 11, show similar trends with peaks in durability performance at 1% silane concentrations. The exceptions to this trend are for pH 7 and 9 solutions. With the former a silane concentration of 5% appears to promote optimum durability characteristics although it is important to recognise that problems relating to experimental variability was encountered with this specic silane concentration. Indeed at all the solution concentration/pH variations examined, this particular solution (5%, pH 7) together with pH 5, 1% silane solution concentrations provided the highest level of durability enhancement observed. At pH 9, silane concentration appears to have little inuence on extent of durability enhancement at concentrations of from 1 to 12% (after an initial steep

increase in G Ic from 0.1 to 1%). The factors responsible for these effects will clearly be related to the complex nature of both solution and interface chemistries and in particular the manner in which they are inuenced by both silane concentration and solution pH. In discussing these complex trends it is pertinent to note that several researchers have considered the issue of silane solution concentration effects. Sung et al, working with Al2O3/polyethylene joints modied with g-aminopropyltriethoxysilane (g-APS), found an interesting correlation between 1801 peel strength and the concentration of aqueous g-APS solutions [18]. In their work optimum peel strength was observed at an g-APS concentration of approximately 2 vol%. Further increases in silane concentration beyond 2% did not change the peel strength signicantly. Ishida et al have proposed, from FTIR investigations on g-APS/glass interfaces, a silanetriol-oligomer balance

ARTICLE IN PRESS
8
3.500
Base line - no silane pH 5.0 - no silane 0.5% silane 5.0% silane 12.0% silane 2.0% silane

M.-L. Abel et al. / International Journal of Adhesion & Adhesives 26 (2006) 215

3.000

0.1% silane 1.0% silane

2.500

Fracture energy (kJm-2)

8.0% silane

2.000

1.5% silane 3.0% silane

1.500

1.000

0.500

0.000

10

20

30

40

50

60

70

80

90

100

110

120

130

140

150

160

170

180

Exposure time (h) to 96 % rh at 50 C

Fig. 5. Fracture energy against exposure time for pH 5.0 hydrolysing solution at various silane concentrations.

1.2
pH3 pH5 pH7 pH9 pH11 GB only

0.8 GIC (kJm-2 )

0.6

0.4

0.2

0 0 2 4 6 Silane conc. (%) 8 10 12

Fig. 6. Wedge fracture energy as a function of solution concentration at various pH values.

heavily dependent on silane solution concentration [19,20]. Indeed, with g-APS, a very low concentration of 0.15% was observed with concentrations at or below this value comprising virtually complete monomeric triol. Increasing oligomer content was observed for silane concentrations above 0.15%. This would be expected to have perhaps a signicant inuence on the nature of the silane lm deposited; low solution concentrations promoting a more uniform lm with high concentrations promoting defect formation within the silane interphase. Ishida suggested, as a result, that it would, therefore, be likely that the mechanical properties of a resultant composite with a silane above the transition concentration may not be optimum [20]. Ishida further proposed the possibility that this phenomenon, a so-called onset of association, may not simply hold true for g-APS and that it may be a general

trend in surface modication with organosilane adhesion promoters. We are not aware of any similar ndings with the organosilane employed in this study (g-GPS), but it seems reasonable to assume that a transitionary effect of this type could have inuenced the results in our work. Kaul et al examined the inuence of silane solution concentration on the durability of Al2O3/polyethylene joints primed with g-APS [21]. Solution concentration was shown to exhibit a substantial affect on durability, with a concentration range of 0.31% providing optimum performance. Interestingly, concentrations beyond 2% were shown to exhibit particularly poor results, with durabilities roughly comparable to nonsilane treated joints. Similar silane concentration effects were observed with Al2O3/g-APS/nylon-6 joints with, in this case, an optimum silane concentration of 0.3% observed [21]. Indeed, at this specic concentration the integrity of the silane-modied interfacial region was enhanced sufciently to drive failure away from the normally relatively weaker interfacial zone into the nylon 6 substrate. Osterholtz and Pohl have commented on the importance of minimising silanol condensation in aqueous solutions by maintaining silane concentrations below 1% by weight for what they describe as typical organofunctional silanes [22]. Kuhbander and Mazza studied silane solution concentration effects with aqueous solutions of g-GPS and found, via Boeing wedge tests, that optimum durability performance was provided by solution concentrations of between 1% and 2% [13]. As previously mentioned, however, in their work Kuhbander and Mazza primed all silane treated surfaces with a chromated-based primer and thus direct comparison with our work is difcult. However, it is of interest to note that their preferred solution concentrations were

ARTICLE IN PRESS
M.-L. Abel et al. / International Journal of Adhesion & Adhesives 26 (2006) 215 9

similar to the optimum concentrations obtained from our investigation. Work by Quinton and Dastoor used SIMS analysis to build upon earlier work with XPS analysis [23]. In this study, they were able to demonstrate not only the variation in silicon containing species on a silane modied crystalline aluminium oxide surface, but also variation in the concentration of Al-O-Si species associated with the silane layer. SIMS, as with XPS, indicated a maximum in the concentration of silicon containing species at a relatively low silane concentration of 0.75%. Following this low concentration peak, the concentration of silicon was found to gradually increase to something approaching an asymptotic maximum at approximately 12% PTMS concentration. However the concentration of Al-O-Si species, as demonstrated by the presence of the mass 71 peak, although a maximum in concentration was observed at approximately the same silane concentration, the peak was followed by a gradual decline in mass 71 intensity with increasing PTMS concentration. In considering the factors responsible for this effect Quinton and Dastoor hypothesised the existence of two distinct surface species involved in the silane adsorption process i.e. monomeric silanols and condensed forms of PTMS. At low concentrations, where the presence of monomeric silanols would be high, it is reasonable to assume that a relatively large number of available silanols would react with surface hydroxyls to produce interfacial covalent bonds i.e. Al-O-Si bonds. Conversely, at high PTMS concentrations, for a given combination of experimental conditions, a signicant degree of silanol condensation would have almost certainly occurred. Thus the number of silanols available for reaction with the oxide surface would have been less per unit silicon concentration. Thus, the results obtained from our work would appear to conrm the importance of solution concentration in promoting signicant adhesion and durability enhancement on bonded systems. Indeed, as with our results, a considerable body of evidence suggests that silane solution concentrations should be maintained at fairly low levels; low concentrations being necessary to minimise the extent to which neighbouring silanetriol molecules would wish to make contact and undergo condensation. Although we have no conclusive evidence to conrm the existence of the onset of association mechanism, it appears likely that this effect was, partly at least, responsible for the solution concentration trends observed in this work. In theory, the GIc v silane concentration trends highlighted in Fig. 6 would perhaps be expected to provide some conrmation of this mechanism. For example, if an onset of association mechanism is, partly at least, responsible for the silane concentration effects, variation in solution pH should have a predictable effect. Specically, since pH exerts an

inuence on the balance between initial silane hydrolysis and further condensation, then we would expect that lower pH values would allow the use of higher silane concentrations since, although silanols would be in close proximity, the low pH value would retard their desire to condense. In other words, the peak values of G Ic in the G Ic silane concentration trends indicated in Fig. 6 should move to higher silane concentrations with decreasing pH. Unfortunately the trends in Fig. 6 are unable to conrm this hypothesis, an insufcient amount of data almost certainly contributing to this fact. Bearing in mind the optimum solution concentration observed, for all further activities e.g. studies of solution age, solvent type, lm drying temperature etc, a solution based upon 1% g-GPS at pH 5 was used so as to maintain further experimental studies at a manageable level.

3.3. Nature of the solvent Although it is widely accepted that organosilanes are best used in dilute solution, a review of the literature reveals some ambiguity as to the nature of the solvents most appropriate for lm formation [16,22,2429]. Although aqueous solutions have been studied the most, many workers have employed solvent combinations, with ethanol or methanol additions to aqueous solutions being particularly noteworthy. It was, therefore, considered useful to examine the nature of the solvent employed in the silane solutions to assess the effects on durability performance as measured from Boeing wedge joints. In this particular case, methanol/ water solvent combinations were investigated ranging from full aqueous solutions (the primary solvent employed throughout most of this study) to a 90/10 methanol/water system. Silane concentration was maintained at 1% with a pH of 5, with a hydrolysis time of 1 h prior to application of the solution to the alloy surface. The results obtained are shown in Fig. 7 as fracture energy, G Ic , (as obtained from Boeing wedge joints after 168 h exposure) plotted against the concentration of methanol in the hydrolysing solution. As clearly indicated for the solution conditions employed, the incorporation of methanol into the hydrolysing solution has a negative effect on G Ic and hence joint durability. Indeed, with a 90/10 methanol/water binary solvent system, the joint durability characteristics which result are only marginally superior to those provided by a simple grit-blast/degrease surface treatment. Several authors have considered the inuence of solvent modication on solution chemistry and/or eventual bond performance when employed as an adhesion promoter. Ishida has proposed that one of the primary roles of alcohol in aqueous silane solutions

ARTICLE IN PRESS
10
1.0 0.9
GB only

M.-L. Abel et al. / International Journal of Adhesion & Adhesives 26 (2006) 215

0.8 0.7 GIC (kJm-2 ) 0.6 0.5 0.4 0.3 0.2 0.1 0.0 0 20 40 60 80 100 Methanol conc. (%)

Fig. 7. Plot of GIc v methanol conc for 168 h ageing.

is to prevent and/or delay the formation of highly oligomerised siloxane aggregates [30]. Gledhill et al., working with g-GPS on mild steel substrates, compared an aqueous silane solution with a 95/5 ethanol/water system [25]. With g-GPS maintained constant at 1% with pH maintained natural, joint durability was found to be highly dependent on the nature of the solvent. In this case, the aqueous solutions were shown to be substantially superior to their ethanol/ water-based counterparts. Again the detailed factors responsible for this observation were not discussed, although it was recognised that silane solution chemistry was almost certainly responsible for the substantial differences in performance observed. Thus the results from this earlier work would seem to support the results obtained from the current study i.e. that the incorporation of alcohols into aqueous solutions generally has a detrimental effect on bond performance, particularly resistance to hot/wet environments. However, since the introduction of methanol into the solution will almost certainly modify the kinetics of both the hydrolysis and condensation reactions, it is feasible that modications to other solution parameters e.g. silane concentration and hydrolysis time, would have inuenced the nature of the silane adsorbed on to the surface and hence durability. Further work to investigate this effect would be of interest. 3.4. Solution hydrolysis time Several studies have been conducted to consider the nature of the reactions which occur in an aqueous silane solution as a function of solution age [20,31]. As mentioned previously, most of these studies have

demonstrated that two main reactions occur i.e. initial sequential hydrolysis of the alkoxy groups, followed by condensation of the resultant hydroxyls. Some effort has been devoted to assessing the extent to which these two primary reactions are inuenced by various solution variables [20,31]. Since the nature and extent of these reactions will be dependent upon time, it was recognised that the age of the silane solution prior to application to the aluminium substrate could be important. Previous work conducted by Gledhill et al, in which the inuence of solution age with mild steel substrates was studied, revealed signicant effects [25]. Hence solution age prior to application was considered an important variable in this study. To investigate this effect, experiments were focused on g-GPS aqueous silane solutions having a silane concentration of 1% at a pH of 5. Solutions were hydrolysed for times between 10 min and 48 h prior to application. The results obtained from these experiments are shown in Fig. 8. As indicated, two hydrolysis times, 10 min and 1 h exhibited the best durability enhancing effects, with all other hydrolysis times exhibiting lower and similar G Ic values. Although a 10 min hydrolysis time had approximately similar G Ic values to those obtained after 1 h, the amount of experimental scatter obtained from the former was high, leaving considerable doubt as to the reproducibility of results at this short hydrolysis time. Thus an optimum hydrolysis time of 1 h was considered appropriate under the solution conditions employed. This observation is in agreement with the work of Gledhill et al who observed optimum hydrolysis times of between 30 and 90 min with g-GPS aqueous solutions applied to mild steel substrates [25]. Similarly, Khubander and Mazza observed an optimum hydrolysis time of 1 h with an aqueous 1% g-GPS solution applied

1 0.9 0.8 0.7


Baseline Data - no silane treatment

GIC (kJm-2 )

0.6 0.5 0.4 0.3 0.2 0.1 0 0 10 20 30 40 50 Hydrolysis Time (hrs)

Fig. 8. Wedge fracture energy as a function of solution hydrolysis time.

ARTICLE IN PRESS
M.-L. Abel et al. / International Journal of Adhesion & Adhesives 26 (2006) 215 11

to a primed aluminium alloy substrate [13]. Abel et al employed NMR techniques to assess the change in the molecular structure of g-GPS (aqueous solutions) as a function of time and noted that complete hydrolysis under the precise solution conditions employed in this work occurred after 1 h, with only slight condensation of the silanols produced [32]. A detailed study by Bertelson and Boerio essentially conrmed these observations with various concentrations of g-GPS in both aqueous and solvent/water combinations indicating silanol formation in less than an hour [31]. Furthermore, using Si-29 NMR spectroscopy, they also observed the onset of condensation reactions leading to the formation of dimers and, eventually, network type species. In the case of a 10% g-GPS aqueous solution at a pH of 5, they were able to observe the gradual reduction in silanol concentration after 4 h hydrolysis with a corresponding increase in dimer and network species concentration after this time. Indeed after 4 h reaction the dimer concentration was reported to be approximately 22% with signicant increases in this concentration observed with increasing time up to 8 h. Of particular interest, Bertelson and Boerio [31] observed a clear correlation between solution chemistry and the wedge test results obtained from the work of Khubander and Mazza [13] with relatively poor wedge test results correlating well with the formation of dimer and network species in the silane solutions. Thus the results from our work, and the various studies highlighted above, would suggest that optimum durability enhancement is provided by solutions in which virtually complete hydrolysis of alkoxy groups has occurred with minimal silanol condensation. Interestingly, all of the hydrolysis times studied provided durability characteristics superior to the base-line non-silane treated joints. 3.5. Inuence of lm drying temperature A key parameter likely to inuence both the chemical nature of the silane layer deposited on an alloy surface, and its subsequent durability enhancing effect, is drying temperature. To consider this issue in this study, two silane lm drying temperatures were employed i.e. 23 and 93 1C. All other aspects of the silane application process were maintained constant at 1% concentration, pH 5 with a hydrolysis time of 1 h. The results obtained from the Boeing wedge experiments conducted are shown in Fig. 9. As indicated, virtually identical G Ic exposure time trends were obtained, suggesting a virtually insignicant inuence of drying temperature under the experimental conditions employed. This result is of particular interest in that recent work, conducted by Abel et al, has shown that variations in the surface chemical structure of silane lms deposited onto aluminium alloy do indeed occur with variations in

3.500 3.000
93C dried wedges 23C dried wedges

Fracture energy (kJm-2 )

2.500 2.000 1.500 1.000 0.500 0.000 0

10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180

Exposure time (h)

Fig. 9. Boeing wedge crack growth for silane lm drying temperatures of 23 1C and 93 1C.

drying temperature [33]. The factors responsible for this apparent anomaly are unclear. However, one possibility is that the elevated temperatures employed during the adhesive cure cycle, 120 1C, resulted in similar surface chemical characteristics to the extent that differences in interfacial chemistry were negligible following the bonding process. Further work, particularly considering the inuence of lm drying temperature with cold-cure epoxy adhesives, would help to resolve this intriguing issue. Interestingly, several other research groups have considered the effects of this important experimental parameter and a brief overview of some of the research conducted would be instructive. A substantial body of work conducted by Baker et al concluded that optimum durability characteristics with aluminium alloy substrates pretreated with g-GPS solutions was provided by a subsequent lm drying temperature of 93 1C [12]. Similar work conducted by Gledhill et al. on a mild steel alloy strongly suggested that room temperature (20 1C) drying conditions promoted greater durability enhancing effects than elevated temperature drying [25]. Sung et al observed, from DSC experiments, increasing Tg of a bulk silane (g-APS) lm deposited on an Al2O3 substrate with increasing drying time at 110 1C [18]. They attributed this effect to increasing lm molecular weight and/or crosslinking with extensive drying. This proposed mechanism was supported further from observations from SEM-EDX across an interface between a silane lm and a polyethylene substrate. Results from this analysis revealed silicon proles across the interfacial zone with increasing sharpness the greater the drying time. This observation was attributed to decreasing interdiffusion of g-APS into the polyethylene as the silane is polymerised/crosslinked further on continued drying. Of particular interest, these authors noted that elevated temperature drying of the silane lm resulted in a reduction in joint strength together with a change in locus of failure from cohesive within the polyethylene substrate to mixed mode close to the interface. These

ARTICLE IN PRESS
12 M.-L. Abel et al. / International Journal of Adhesion & Adhesives 26 (2006) 215

detrimental effects were attributed to a lack of interdiffusion between the polyethylene and silane due to increased molecular weight/crosslinking in the latter caused by the increased drying temperature. Later work by the same authors considered the inuence of silane lm drying on the durability of polyethyleneAl2O3silane joints in a hot/wet environment [21]. Interestingly, drying of the lm at elevated temperature (110 1C) was, at least for moderate drying times, found to enhance durability. An enhanced inability of water to diffuse through the silane layer, due to increased crosslinking, was proposed as the cause of this effect. Culler et al, exploring reactions at an interface between an epoxy matrix resin and a g-APS silane lm, observed a direct correlation between the extent of reaction and the degree of condensation of the g-APS [34]. In conclusion, they proposed that maximum reactivity at the interface would be achieved when the degree of condensation is smallest, clearly implying that, for optimum performance, the siloxane layer should not be highly condensed. Work by Phillips and Hercules has indicated that heating an adsorbed silane lm increases the water resistance of that lm [35]. Cave and Kinloch suggested that durability enhancements found from the use of silanes deposited onto metallic surfaces without subsequent elevated temperature drying, could have been associated with the inuence of heat cure of the adhesive on the structure of the silane layer [26]. Nishiyama and co-workers studied the adsorption of g-MPS on a colloidal silica and observed, via GPC, that the molecular nature of the adsorbed silane varied with drying time [36]. In particular both an increased lm molecular weight together with a decrease in the amount of removable physisorbed silane was observed with increased drying time. Kuhbander and Mazza have noted silane manufacturers information stating that elevated temperature drying of silane lms is benecial since heating removes hydrolysis products and solvents and promotes more effective silanol condensation [13]. In addition, from Boeing wedge experiments conducted with silane lm drying temperatures of 60 1C to 107 1C, they observed optimum durability performance at 93 1C. Interestingly, the adhesive employed by Kuhbander and Mazza was identical to that used in this study, with, in all probability, the same cure conditions. Bertelsen and Boerio have proposed that the oligomerisation required for good adhesion with silane lms (after deposition) can be obtained through drying of the lm at elevated temperatures [31]. Clearly, the substantial body of research conducted on this particular experimental variant has not proved conclusive in determining whether elevated temperature drying is desirable across a range of silane substrate combinations.

3.6. Inuence of in-process time delay The time interval between drying of the deposited silane lm and the formation of the bonded joint will be of considerable practical importance in both initial manufacture and repair operations. In such circumstances it will, of course, be important to know how long the treated substrate will remain active prior to any deterioration in its durability enhancing capacity. In many industrial bonding operations conventional wisdom has been that a structure should undergo continuous processing from surface treatment to bonding and adhesive cure within approximately 24 h. To ascertain the degree to which applied silane coatings remain active following application, Boeing wedge experiments were conducted in which the time interval between silane application and subsequent bonding were varied between 0 and 17 days. In each case, during the time delay period, silane treated specimens were stored in a dust free cabinet at 23 1C, 50% relative humidity. The results obtained are indicated in Fig. 10. The trends reveal that, under the specimen storage conditions employed, little or no reduction in durability occurs up to approximately 7 days. Although an increase in the in-process time interval to 17 days resulted in a notable reduction in durability, the deterioration in silane surface activity was still not sufcient to reduce G Ic down to values typical of baseline non-silane treated joints. The reduction in durability enhancing characteristics on increasing the in-process time interval from 7 to 17 days may be due to a reduction in the number of active sites on the surface. This could be due to adsorption of atmospheric contamination or some continuing reactions (e.g. silanol condensation) within or at the surface of the silane layer, which could reduce the number of sites available for reaction/interaction with the adhesive layer. In spite of the uncertainty relating to the mechanisms responsible for the above mentioned trends, it is clear that, provided care is taken to avoid gross
1.0 0.9 0.8 0.7
GB only

GIC (kJm )

-2

0.6 0.5 0.4 0.3 0.2 0.1 0.0 0 5 10 15 20

Process time lag (days)

Fig. 10. Wedge fracture energy as a function of inprocess time delay.

ARTICLE IN PRESS
M.-L. Abel et al. / International Journal of Adhesion & Adhesives 26 (2006) 215 13

contamination of the silane treated surface, the time interval between silane application and subsequent bonding is not critical up to approximately 7 days. This observation suggests considerable exibility in both manufacture and repair operations. However it is of course likely that the degree of surface deterioration will depend on storage conditions. Thus it would be of interest to explore the extent to which property variations occur with respect to this experimental variable. 3.7. Fracture surface characterisation On completion of each Boeing wedge experiment, the joints were separated to allow examination of the failure surfaces. Throughout the study, under all silane treatment conditions employed, failure within the regions of crack propagation following ageing in the aggressive environment, was apparent interfacial. In other words, within this region of the joint, one failure surface had the visual appearance of aluminium alloy, whereas its counterpart was ostensibly adhesive. A detailed analysis using XPS was employed to characterise the failure process occurring within joints previously subjected to the optimised silane treatment (aqueous 1% concentration, pH5, 1 hr hydrolysis time, 93 1C drying temperature) and which had been subjected to an ageing period of one week. The photograph in Fig. 11 shows the failure surfaces of a wedge joint separated after just one week exposure to the 96% rh, 50 1C environment. As indicated, four regions of the failure surfaces are highlighted. Region 1 corresponds to the zone which resulted from initial insertion of the wedge. Region 2 was formed as a result of crack growth through the adhesive layer in the rst hour following insertion of the wedge (prior to ageing in the hostile environment). Region 3, the region of primary interest in

this study, resulted from crack growth following insertion of the joint into the 96% rh, 50 1C environment. Following the prescribed ageing time in the hostile environment (in this case one week) joints were removed and split-open to reveal the failure surfaces. Region 4 corresponds to crack propagation within the bondline which occurred during this nal separation process. Of primary signicance in Fig. 11 are the differences in locus of failure between region 3 on the one hand, and regions 1, 2 and 4 on the other. Region 3 indicates a zone of virtually complete apparent interfacial failure, with all other regions demonstrating failure through the adhesive layer. Region 3 was subjected to XPS analysis to determine the precise nature of the failure process. The survey spectrum given in Fig. 12 was obtained from adhesive

Fig. 12. XPS survey spectrum from the adhesive failure surface of one-week aged joint.

Fig. 11. Boeing wedge fracture surface following exposure to test environment for one week.

ARTICLE IN PRESS
14 M.-L. Abel et al. / International Journal of Adhesion & Adhesives 26 (2006) 215

failure surface area. Fig. 13 shows the survey spectrum for the opposing aluminium failure surface. Quantication of both surfaces was undertaken and the resultant data is summarised in Table 2.

Fig. 13. XPS survey spectrum from the aluminium failure surface of one-week aged joint.

Table 2 XPS quantication for one-week aged wedge joints Sample ID Elemental concentration, (at%) O Standard wedge metal interface 39 Standard wedge polymer interface 28 N 2 1 C 44 67 Si 2 4 Al 8 2

The nitrogen and carbon signals recorded from XPS suggest the presence of adhesive on both sides of the failure zone, thus indicating a mixed mode of failure at this exposure time. The presence of the aluminium 2p signal from the adhesive surface conrms the contribution of the oxide layer to the overall failure process at the short exposure condition. Furthermore, silicon was clearly observed on both adhesive and aluminium failure surfaces, this clearly indicating the involvement of the organosilane in the degradation and failure processes. Based upon this information, a schematic representation of the degradation and failure events occurring within the rst week of ageing is depicted in Fig. 14. As indicated, under the ageing conditions employed, crack growth is seen to proceed through a composite zone comprising oxide, silane and adhesive. Interestingly, Arnott et al, working with g-GPS treated 2024-T3 aluminium alloy, observed similar Boeing wedge failure surfaces to those observed in the current study [37]. In their case they also observed, from XPS analysis, the existence of Al, Si, C and O on both sides of the failed joint i.e. on the apparent adhesive and metallic surfaces. This would, as with our studies, imply failure located within an oxide/silane/adhesive diffusion zone. Such observations clearly provide interesting clues as to how further improvements in durability could perhaps result from modications to both the silane and oxide layers.

4. Concluding remarks The application and solution conditions employed contribute to the overall durability enhancing character-

Aluminium oxide

Wedge fracture pat 96%rh, 50Ch

Diffusion layer containing silane and aluminium oxide Silane layer

Epoxy

Fig. 14. Schematic of a sectioned wedge joint showing the apparent failure path for joint aged for one week.

ARTICLE IN PRESS
M.-L. Abel et al. / International Journal of Adhesion & Adhesives 26 (2006) 215 15

istics of the organosilane treatment. Specically, solution variables such as the nature of the solvent (methanol versus water), solution pH, silane concentration and hydrolysis time prior to application, have all been shown to be important. With the aluminium alloy employed, an aqueous solution with a silane concentration of 1% at pH 5 and hydrolysed for 1 h prior to application has been shown to produce best results. Indeed under these conditions durability characteristics comparable to a chromic acid anodise treatment has been demonstrated. Silane lm drying temperature has been shown to have an insignicant effect on resultant joint durability. The factors responsible for this effect, in particular conicting observations from the literature relating to changes in silane lm chemical structure, are unclear. Surface characterisation techniques based upon microscopy, XPS and SIMS have indicated degradation and failure processes associated with both the aluminium oxide and silane layers via a so-called oxide-silane diffusion zone. Such observations, particularly the extent to which the oxide layer contributes to the failure process with the optimised silane treatment, indicates the regions of the interphase from which further improvements in joint durability could emerge. In this study no attempt has been made to investigate the effects of the various solution parameters on the stability of the epoxide group on the silane molecule. Further studies should be carried out to investigate this potentially important issue.

Acknowledgements The authors would like to acknowledge the nancial support of the MoD Corporate Research Programme.

References
[1] Kinloch AJ. Adhesion and adhesives: science and technology. London: Chapman and Hall; 1987.

[2] Plueddemann EP. Silane coupling agents. New York: Plenum Press; 1982. [3] Mittal KL (Ed). Silanes and other coupling agents, VSP: The Netherlands; 1992. [4] Shaw SJ. In: Ellis B, editor. Chemistry and technology of epoxy resins. London: Blackie; 1993. [5] Olsson-Jacques CL, Wilson AR, Rider AN, Arnott DR. Surf Int Anal 1996;24:569. [6] Rider AN, Arnott DR. Surf Int Anal 1996;24:583. [7] Rider AN, Arnott DR. Int J Adhes Adhes 2000;20:209. [8] Digby RP, Shaw SJ. Int J Adhes Adhes 1998;18:261. [9] Mostovoy S, Crossley PB, Ripling EK. J Mats Sci 1967;2:661. [10] Pearce PJ, Tolan FC. International conference on aircraft damage and repair. Melbourne, August 1991, p. 188. [11] Thiedman W, Tolan FC, Pearce PJ, Morris CEM. J Adhes 1987;22:197. [12] Baker AA, Chester RJ. Int J Adhes Adhes 1992;12:73. [13] Kuhbander RJ, Mazza JJ. SAMPE Symp 1993;38:1225. [14] Arnott DR, Kindermann MR. J Adhes 1995;48:101. [15] Erickson PW, Plueddemann EP. In: Plueddemann E P, editor. Composite materials, vol. 6. New York: Academic Press; 1974. [16] Xue G, Koenig JL, Ishida H, Wheeler DD. Rubb Chem Technol 1991:162. [17] Tesoro G, Wu YJ. Adhes Sci Technol 1991;5:771. [18] Sung NH, Kaul A, Chin I, Sung CSP. Polym Eng Sci 1982;22:637. [19] Ishida H, Chiang CH, Koenig JL. Polymer 1982;23:1982. [20] Ishida H. Polym Compos 1984;5:101. [21] Kaul A, Sung N H, Chin I, Sung C S P. Polym Eng Sci 1984;24:493. [22] Osterholtz FD, Pohl E R. J Adhes Sci Technol 1992;6:127. [23] Quinton JS, Dastoor PC. Appl Surf Sci 1999;152:1317. [24] Patsis AV, Cheng S. J Adhes 1988;25:145. [25] Gledhill RA, Shaw SJ, Tod DA. Int J Adhes Adhes 1990;10:192. [26] Cave NG, Kinloch AJ. J Adhes 1991;34:175. [27] Leyden DE, Atwater JB. J. Adhes Sci Technol 1991;5:815. [28] Leung YL, Zhou MY, Wong PC, Mitchell AR, Foster T. Appl Surf Sci 1992;59:23. [29] Woo H, Reucroft PJ, Jacob RJ. J Adhes Sci Technol 1993;7:681. [30] Ishida H. In: Mittal KL, editor. Adhesion aspects of polymeric coatings. New York: Plenum Press; 1983. p. 45. [31] Bertelsen CM, Boerio F J. J Adhes 1999;70:259. [32] Abel M-L, Joannic R, Fayos M, Lafontaine E, Shaw S J, Watts J F. Int J Adhes Adhes, this edition. [33] Abel M- L, Rattana A, Watts JF. J Adhes 2000;73:313. [34] Culler SR, Ishida H, Koenig JL. J Col Interf Sci 1986;109:1. [35] Phillips LV, Hercules DM. Silanes, Surfaces and Interfaces Symposium. Colorado, USA, June 1983. [36] Nishiyama N, Shick R, Ishida H. J Col Interf Sci 1991;143:146. [37] Arnott DR, Wilson AR, Rider AN, Lambriandis LT, Farr NG. Appl Surf Sci 1993;70/71:109.

You might also like