Pretreatment of aluminium: topography, surface chemistry and adhesive bond durability.
R.P. Digby & D.E. Packham
School of Materials Science,
University of Bath, BA2 7AY.
Abstract
This paper reports on work in which bonded joints between a clad aluminium alloy [L165 (Cu 4.4%, Mg 0.5%, Si 0.8%, Mn 0.8%)], with different pretreatments, and an epoxy resin (Ciba Geigy's Redux 312/5) have been examined. The datum for comparison of the pretreatments is the Boeing phosphoric acid anodisation (BAC 5555). Other treatments include sulphuric acid anodising in combination with a phosphoric acid dip and a sulphuric acid/ferric sulphate etch..
The topographical structure of the surface layers formed have been examined using scanning and transmission electron microscopy in conjunction with ultramicrotome sectioning. The chemical composition of pretreated surfaces has been characterised by X-ray photoelectron spectroscopy and depth-profiling using argon ion etch performed.
Adhesive bonds have been prepared and their durability assessed using wedge test specimens. The data collected has been used to calculate crack growth and strain energy release rate as functions of time for the bonds produced. Surfaces of failed specimens have been examined to establish the locus of failure.
(Keywords: adhesively-bonded joints; aluminium; pretreatment; surface topography; surface chemistry; durability)
Introduction
For many years the aerospace industry has used adhesive bonding for a variety applications. During the last 30 years there has been increased usage of adhesives in structural applications . Whilst it is relatively easy to achieve an adhesive bond with high initial strength, it is often much more difficult to produce a bond which is durable. One of the most hostile environments for many adhesive bonds is one which is warm and humid. Unfortunately, these are precisely the conditions frequently encountered in aerospace applications.
The strength and durability of an adhesively-bonded repair depend partly on the adhesive and partly on the metal and its surface pretreatment. Many pretreatments have been developed to increase the initial strength and durability of bonds to aluminium and its alloys. The most effective of these are complex, multi-stage processes involving a combination of degreasing, etching and anodising. In some cases further controlled etching is carried out after anodising. Other methods for increasing bond strength and durability include the use of primers and coupling agents (e.g. silanes).
This paper reports results of a detailed examination of the effectiveness of nine different pretreatments for aluminium. Their effectiveness for producing strong, durable adhesive bonds is assessed by use of the Boeing wedge test. The nature of the surface layers involved is examined by X-ray photoelectron spectroscopy (XPS). Conventional anodising treatments are used in combination with post-anodising dips, and a non-chromate containing substitute for the FPL etch is investigated.
Experimental Methods
The materials used in this project are those of particular relevance to the aircraft industry, as this is a major user of structurally bonded aluminium alloys.
As a point of reference to which other treatments are compared, the classic Boeing specification (BAC 5555) has been used. This specification involved the chromate-containing FPL etch as a preliminary stage. A ferric sulphate/sulphuric acid treatment, designated P2, is claimed to have a similar effect,, so in some experiments this has been used as a replacement for the FPL etch.
A post-anodising etch in phosphoric acid has been shown to increase durability for sulphuric acid anodised specimens; this procedure has been investigated here as a final dip for both sulphuric and phosphoric acid anodised specimens.
The wedge test has been used to evaluate durability. Strips of aluminium alloy were bonded after the appropriate pretreatment, a standard wedge was inserted and they were exposed in an environment of 50C/96% relative humidity. The crack growth was monitored and graphs prepared showing total crack growth as a function of time. For a number of the tests the strain energy release rate was calculated using the equation:
Where:
GI = mode I strain energy release rate
E = Young's Modulus
d = displacement of the load point (thickness of the
wedge)
h = specimen thickness
a = crack length
0.6 = geometric correction factor for rotation about
the crack tip
The above equation is one of the simpler solutions for strain energy release rate and has the Griffiths energy balance criteria as its basis. More complex equations, such as those proposed by Cognard and Ripling et al., give different values of strain energy release rate. All of these equations contain both empirical and theoretical correction factors, some of which are not as yet universally accepted; therefore this simple form is used for the purely comparative analysis presented here.
The surfaces of the alloy strips has been characterised after each stage of the different pretreatment schedules using a combination of scanning electron microscopy, extended resolution scanning electron microscopy (SEM/XSEM), transmission electron microscopy (TEM) and XPS. The fracture surfaces after environmental failure have been examined in a similar way to establish the mode of failure.
Experimental Details
The L165 clad aluminium alloy (Cu 4.4%, Mg 0.5%, Si 0.8%, Mn 0.8%) and the Redux 312/5 (Ciba-Geigy) epoxy resin, a bisphenol-A epoxy resin film adhesive supported on a woven nylon carrier, were selected as they are both materials of particular interest to the aircraft industry. Coupons with dimensions 150 x 25 mm were cut from 3 mm thick L165 (clad) sheet.
Degreasing
Gross surface contamination by residues of the manufacturing process was removed from the coupons by wiping with a cloth impregnated with 1,1,1-trichloroethane. The coupons were then degreased in both vapour and hot liquid 1,1,1-trichloroethane (5 min immersion in each) using a Dawe Miniclene.
Alkaline cleaning
Alkaline cleaning was carried out in a commercial cleaning solution (Minco N24205) at 60-70C for five min. After cleaning, solution residues were removed by rinsing in cold running tap water for 5 min.
Etching
Initial etching was in either the FPL etch at 68°C (±3°C ) or the P2 etch at 65°C (±3°C ) for 10 min. The FPL etch was made up with 1.98 l sulphuric acid (specific gravity = 1.84), 363g sodium dichromate, 16.5g aluminium, 11g copper pre-dissolved as sulphate in distilled water and 11 l distilled water. The P2 etch consisted of 3330g sulphuric acid (specific gravity = 1.84), 1350g ferric sulphate and distilled water to make 9 l. Heating and agitation of the etchants was provided by a Gallenkamp Thermostirrer 100. After etching the coupons were rinsed in cold running tap water (5 min).
Anodising
In addition to etching, some coupons were also anodised in 10 vol.% phosphoric acid at 23°C (±3°C) for 20 min at 10 V (PAA), or in 10 vol.% sulphuric acid for 12 min with a current density of 4 Adm-2 at -5C (SAA). After anodising the coupons were rinsed in cold running tap water for 5 min.
Post anodising dip (PAD)
A post-anodising dip in 10 vol.% phosphoric acid at 60°C for immersion times up to 1 min was performed in some cases. The coupons were once again rinsed in cold running tap water for 5 min on completion of this stage.
Final rinsing and drying
Once all the required pretreatments had been applied to the coupons, and after their last rinse in cold running tap water, they were given a final rinse in distilled water for 3 min. Next the coupons were dried in warm air (<50°C) for approximately 20 min to ensure that the surfaces to be bonded were completely dry.
Adhesive bonding
Wedge test specimens were assembled by bonding two of the pretreated coupons together using Redux 312/5 adhesive. Temperature (120°C for 30 min) and pressure (170-250 kPa ) during the cure cycle were provided by a heated platen press. The bonded specimens were kept under pressure until they had cooled sufficiently to be handled comfortably.
Environmental testing
Environmental performance of the bonded joints was assessed by means of the Boeing wedge-test (ASTM D3762). The specimens were placed in air at 50°C and 96% relative humidity. Crack tip positions were marked on the specimen edges at various times and measured with a travelling microscope. The marked crack tip positions on both sides of each specimen were used to provide a more accurate indication of the crack length.
Electron microscopy
Electron-optical examinations were carried out using JEOL T330 and 35C scanning electron microscopes, a JEOL 2000FX transmission electron microscope and a JEOL 1200EX transmission electron microscope fitted with a scanning image device. A number of specimens were bent sharply through 180, thereby fracturing the oxide layer, to allow examination of both the surface and the cross-section of the anodic layer. Ultramicrotomy was also employed to prepare specimens for TEM examination. Some microscopy specimens were coated with a very thin layer of gold/palladium alloy to prevent charging.
Electron spectroscopy
XPS analysis of the pretreated surfaces has been performed and depth-profiling, by controlled argon ion beam etching at 10kV and 50A for various times, has provided information about chemical variations within the oxide layer. The spectrometer was a VG Scientific ESCALAB MKII using an aluminium anode operated at 12.5kV, 40mA.
Results and Discussion
Characterisation of pretreated surfaces
The nine main pretreatment schedules investigated are summarised in Table 1. Electron microscopy (Figure 1a) shows that the FPL treatment produced the scalloped surface with a network of shallow pores and protrusions on top of a thin barrier layer described by Venables. The P2 treatment produces similar topographical features (Figure 1b).
The anodising treatments produced the classic open pore structure of the PAA surfaces (Figure 2) and the much finer pore structure produced by SAA (Figures 3 and 4). Electron-optical examinations have shown that the anodic layer formed by PAA after P2 etching is somewhat different to that formed after FPL. The P2+PAA oxide layer (Figure 2b) appears to be more dense without the continuous pores and very 'open' surface of the FPL+PAA oxide (Figure 2a). The persistence of the scallops from the first stage of pretreatment is particularly clear from Figure 3.
The post-anodising dip in phosphoric acid (PAD) attacks the anodic oxide, opening up the surface. In Figure 5 the extended surface achieved by controlled etching in phosphoric acid after sulphuric acid anodising can be seen. The previous fine pore structure, see Figure 3, has been opened to leave the brush-like topography with features of of 1 m in scale. A similar structure has been described by Arrowsmith et al.. Application of this dip treatment for various lengths of time to a PAA surface produced the topographies shown in Figure 6. Here it can be seen that the surface has initially been opened and then progressively etched away. After 30 s etching the original structure produced by the PAA has been completely transformed to one with pores or etch pits 0.1 m in size, Figure 6c.
X-ray photoelectron spectroscopy
The elemental compositions of the surfaces as indicated by XPS are given in Table 2. No dramatic differences in chemical composition between the surfaces are apparent. In most cases the ratio of AlIII to oxygen is consistent with Al2O3, within reasonable error limits, and allowing for the presence of other oxide species. The carbon 1s peaks of the XPS spectra show that the carbon present is bonded predominantly to carbon or to hydrogen. Some of these peaks display very slight broadening which suggests that a very small amount of carbon is bonding to other elements such as oxygen. All of the carbon detected is due to contamination of the surfaces from the atmosphere during preparation. This analysis is broadly consistent with that Davis for FPL treated surfaces11. A small percentage of phosphorus is detected when phosphoric acid is used at some stage of the pretreatment. The presence of chromium was detected on the surface of the FPL-etched specimens. However, there was no evidence for its presence after further treatments had been applied (FPL+PAA, FPL+PAA+PAD). In these further treatments the surface of the oxide layer is continuously dissolved. The disappearance of chromium after these treatments implies that it is present only on the surface of the oxide formed in the FPL etch. The P2 etch introduces a small percentage of iron into the surface. This, like the chromium from the FPL etch, disappears when further treatments are applied once again showing that the inclusion is only present on the surface of the original , P2-formed oxide.
Controlled argon ion etching allows XPS analysis at successive depths into the oxide layer, so that changes in the chemical nature of the oxide throughout its depth can be observed. This technique is particularly useful for establishing the whereabouts of any inclusions introduced by the various pretreatments. Depth calibration was performed by taking an oxide layer of known thickness and establishing the time taken for complete removal. This method, using a 400nm oxide layer produced by PAA (measured using the ultramicrotome sectioning technique described above), gives an etch rate of approximately 5nm min-1.
The aluminium spectra obtained for the two schedules culminating in a phosphoric acid dip after PAA show a minor peak indicating Al0 (aluminium metal) as well as the AlIII (oxide) peak. This confirms that the PAD employed is removing the anodic oxide sufficiently for the underlying metal to be detected. Depth profiling by argon ion etching reveals that the remaining surface layer is extremely thin; possibly completely removed in some places. The fact that the signal from AlIII disappears from the XPS spectrum after only 3 min of argon-ion etching shows that the remaining oxide is less than 15 nm thick .
Depth profiling of the surfaces reveals that the iron introduced by the P2 etch occurs throughout the P2-formed oxide layer (Table 3), only disappearing completely when argon-ion etching has removed the entire oxide layer (indicated by the eventual disappearance of the AlIII peak and its replacement by Al0 ). The Cu 2p3 peak could be due to the underlying alloy or, in the case of pretreatments involving the FPL etch, the copper used to optimise the etchant. It is known that copper tends to diffuse to the surface of copper-containing aluminium alloys and could, therefore, find its way to the metal-oxide interface. This would explain the presence of the Cu 2p3 peak with PAD-treated surfaces where the oxide layer has been destroyed and its appearance as argon-ion etching removes the oxide layer during depth profiling. With the FPL-etched surface copper does not appear to be present on the outer surface but is evident only after argon-ion etching (Table 3). It is unclear whether the copper here is from the underlying alloy or from the solution optimisation; either way it seems to occur at or near the metal-oxide interface. There is little sulphur or chromium in the FPL-formed surface layer.
The argon-ion etch time required to remove the AlIII peak from the data of Table 3 indicates that oxide layer produced by the P2 etch is somewhat thinner than that formed by the FPL process. The P2-formed oxide is completely removed by 7.5 min of argon-ion etching (Table 3) giving an oxide thickness of <37.5 nm. After the same etch time the FPL-treated surface still has a clear AlIII signal, indicating the presence of residual oxide.
The nitrogen detected during depth profiling of the FPL, P2 and P2+PAA surfaces is an artefact caused by nitrogen gas leaking into the spectrometer during argon-ion etching; this was due to a fault in the equipment.
The P2+PAA oxide layer has a small percentage of phosphorus present throughout its depth (Table 3). Whether this is incorporated into the oxide itself, or just on the pore walls, is unclear. There is also a trace reading for sulphur both on the oxide surface and after depth profiling. Again, the precise location of this inclusion is not known.
Analysis of the surfaces produced by P2+SAA (Table 3) shows a small inclusion of sulphur throughout the anodic layer. Indeed, once the surface contamination has been removed by a short argon-ion etch, the chemical analysis of the anodic layer remains constant throughout the thickness analysed (Table 3). XPS analysis of the P2+SAA+PAD treated surfaces reveals that a small percentage of phosphorus is incorporated into the anodic layer (Table 3). The amount of phosphorus present decreases as depth increases. This implies that the phosphoric acid has penetrated the pores of the anodic layer but is unable to penetrate to the base of the anodic layer. This could be due to a number of factors including pore narrowing, pore branching and termination and the limits of capillary action.
For many of the pretreatment schedules the ratio of aluminium to oxygen determined by XPS after depth profiling does not appear to be consistent with Al2O3. This can be accounted for by the presence of other oxygen-containing species (in cases of excess oxygen) and effects of the argon-ion beam on the stoichiometry of the oxide under ultra high vacuum (in cases of oxygen deficit). It is likely that the argon ion beam is removing some of the oxygen and maintaining a high positive charge on the surface, thereby allowing the aluminium atoms formerly bound to the absent oxygen to remain in the 3+ state (Al3+).
Environmental Exposure
Interpretation of wedge test data
There are two frequently used methods for displaying and analysing the results obtained from wedge tests. The first of these presents the cumulative crack growth beyond the initial crack, induced by insertion of the wedge, as a function time in the hostile environment. The second method uses information about the test specimen geometry in conjunction with the crack-growth data to calculate the strain energy release rate (GI) as a function of time as the test progresses. This work shows the importance of considering both approaches to avoid misleading interpretation. For example, if crack growth alone is considered, the effect of variations in the initial crack length are not taken into account. A bond may apparently display good durability, but the initial crack length may be so large as to be unacceptable (indicating a weak initial bond). It must be remembered that, as initial crack length increases, the 'driving force' for further crack growth decreases.
Calculation of strain energy release rate (GI) can provide a more complete picture; by its very nature it includes information about the strength of the initial bond. However, care must still be taken when interpreting the results. A bond which has high initial strength may maintain that strength and thus be classified as durable. A bond of similarly high initial strength would have poor durability if its strength reduced considerably during environmental exposure. Bonds of low initial strength can be considered in the same way. It is quite possible for bonds with quite different strength/durability characteristics to have very similar values of strain energy release rate once the crack has reached equilibrium. This raises the question: which is the best pretreatment? The simple answer is the pretreatment which produces the bond with both highest initial strength and the best durability. However, this may not be the most appropriate treatment to use for a particular application; the answer then is dependent on the design requirement. The situation is further complicated by the fact that many of the specimens displaying high initial bond strength also have the most rapid reduction in strength on exposure to the hostile environment. This may of course be due to the synergistic effect of stress and environment. Nevertheless, this must be borne in mind. It should also be noted that the mean GI which are often used when discussing the relative merits of various pretreatments can be dramatically altered by any variation in initial crack length. By the very nature of the wedge test, the initial crack length has considerable inherent variability; once again this must be borne in mind when interpreting the results.
Figures 7 and 8 are given to illustrate some of the above points; it can be seen that during the first 24 h of exposure the crack growth for both the FPL+PAA and the P2+PAA+PAD(10 s) are similar, with the FPL+PAA perhaps showing marginal superiority (Figure 7). However, when one looks at GI for the same period of time the FPL+PAA treatment produces substantially lower values thereby implying that the FPL+PAA pretreatment is producing a weaker bond than the P2+PAA+PAD(10 s). Figure 8 also clearly demonstrates how GI values reduce rapidly in the initial stages for some treatments (P2+PAA+PAD(10 s)) whilst others fall of much more slowly. This behaviour is an indication of the susceptibility of the bond to early environmental degradation.
Comparison of P2 and FPL
These results are of particular interest as the P2 treatment does not involve use of chromium, and may be considered as a replacement for the long-established FPL sulphochromate treatment. Both of these can be used as single-stage treatments or as a preliminary etch in a multi-stage pretreatment schedule. Figures 9 and 10 show crack-growth data and GI values respectively for both single-stage etching and etching followed by phosphoric acid anodising. It can be seen from Figure 9 that both of the two-stage treatments offer considerable reduction in crack-growth over the single-stage treatments. Indeed, when one considers the crack-growth data alone, both P2+PAA and FPL+PAA appear to produce almost identical results. However, when one looks at the variation in GI with exposure time (Figure 10) a very different story emerges. FPL+PAA has a much lower initial GI value than the other four treatments; indicating a greater initial crack length. As time progresses the GI values for FPL and P2 reduce rapidly whilst those for FPL+PAA and P2+PAA reduce more slowly until, after 100 h, FPL and FPL+PAA have an almost identical GI value with that for P2 being somewhat lower and for P2+PAA substantially higher. The reduction in GI for P2+PAA during the early stages of environmental exposure is greater than that for FPL+PAA but, despite this, the equilibrium value reached is much higher. Overall, it would appear that P2 used alone is not as effective a pretreatment as the FPL etch as it leads to higher crack growth and lower GI. Conversely, when P2 is used as a preliminary etch to PAA it outperforms FPL both in terms of crack growth and strain energy release rate.
During the course of this study the scatter in the raw data obtained from wedge tests has been noticeably lower for the P2 etch than for the FPL etch when both are used on their own. Thus, although the P2 durability may be lower than the FPL it is more consistent in its effectiveness. The scatter in the data for the two- and three-stage processes is considerable lower than for either of the single-stage processes.
The implication is that the classic FPL+PAA treatment is not producing as good a bond as the single-stage FPL etch. The data presented in the summary graph (Figure 10) are based on mean values. The FPL+PAA treatment is let down by considerable variation in the initial crack lengths prior to environmental exposure. A few specimens with high initial crack lengths lead to a reduction in the mean initial GI value, a lowering of the 'driving force' for crack growth, and hence misleadingly low mean values for GI and mean crack growth. The so-called 'good' pretreatments do not show such variation in initial crack length.
Visual examination of the fracture surfaces shows that for the two-stage treatments (FPL+PAA and P2+PAA) the failure is cohesive within the adhesive, well away from the treated alloy surface. Failure for the single stage processes (FPL and P2) specimens, by contrast, appeared largely interfacial with the fracture path changing from time to time from one adherend surface to the other. XPS results for all these surfaces, however, gave no evidence of aluminium on the surface of the epoxy; indeed all four spectra were very similar and resembled the spectrum of the epoxy resin, Table 4.
Sulphuric acid anodising
The results for sulphuric acid anodising are shown in Figures 11 and 12. The crack growth is much more extensive than for the treatments shown in Figures 9. GI values for both treatments reach an extremely low level after only a very short exposure to the high humidity environment (Figure 12). Although it may appear from Figure 12 that the P2+SAA produces better GI values than P2+SAA+PAD in fact, the scatter in the raw data is such that the difference between the two curves is not statistically valid. A modest improvement in durability does result from the post-anodising dip, P2+SAA+PAD. However, this is still very different from the substantial improvements reported by Arrowsmith et al.12 . It seems likely that this lack of improvement is due to differences in the adhesive being used. The best results reported by Arrowsmith et al. were achieved using a low viscosity, two-part, rubber-toughened acrylic adhesive. They found it was difficult to get full wetting of the adherend surface using two higher viscosity epoxy adhesive unless a primer was used. This is consistent with observations made during the course of this investigation where many voids have been identified at the bondline for SAA-treated surfaces. There also appears to be little penetration of the surface by the adhesive. This reduces the area available for chemical bonding and reduces the degree of mechanical interlocking. This is also consistent with the work of Arrowsmith et al. 12. Electron-optical and XPS examinations support these observations: Figures 3 and 4 show the narrow, closed nature of the pores; XPS detects relatively small amounts of the organic groups associated with epoxy resins on the adherend side of a P2+SAA+PAD treated specimen after failure (Table 4). This latter result indicates that little penetration of the surface has taken place.
These results demonstrate how important it is to consider both the adherend pretreatment and the intended adhesive when designing an adhesively bonded joint. Pretreatments which produce excellent durability with one adhesive may produce mediocre, or even poor results with another. There is no benefit to be gained from developing a porous oxide layer or extending the surface area if the adhesive being used is unable to penetrate those pores or fully wet the surface.
Visual inspection suggested that the P2+SAA specimens had failed interfacially. XPS detected a small amount of aluminium on the epoxy after failure and quantities of the carbon groups associated with the epoxy resin (indicated by peak shifts within the C 1s peak of the XPS spectra) found on the aluminium (Table 4). This implies that the failure is close to interfacial, but with the crack passing through the extremities of the oxide layer where the oxide and adhesive form a 'micro-composite' region.
The specimens with the additional post-anodising dip (P2+SAA+PAD) tended to fail in a slightly less interfacial manner. This is reflected in the increase in the amount of epoxy resin detected on the failure surface by optical examination and XPS analysis (Table 4).
PAA+PAD
The 60 s PAD treatment which was partially successful for the SAA surfaces is clearly disastrous for the PAA treatments (Figures 11 and 12, 13 and 14). XPS results from the pretreated surface, Table 2 discussed above, suggest that the 60 s treatment removes most of the anodic layer, as indicated by the presence of a substantial Al0 peak in the spectrum.. Table 4 presents XPS data from which it can be seen that a very small amount of AlIII is present on the adhesive side of the failed bond, and a relatively small amount of the carbon groups associated with epoxy is present on the adherend side. This supports evidence from optical examinations which had indicated that the crack in the wedge test specimens propagates close to the oxide-epoxy interface.
In view of the poor results obtained from a 60 s dip, it was considered that a shorter post-anodising dip might be effective in opening up the porous structure and improving durability. The results in Figures 13 and 14 show that durability is much improved. The crack growth for the 10 s dip is less than that for the P2+PAA treatment (Figure 9), but the difference is probably not significant. The GI results for this 10 s dip (Figure 10) show a high initial value which falls rapidly during the first few hours (similar to P2+PAA), but then falls much more slowly to attain a near-equilibrium state at a higher value than for any other treatment in this programme.
Visual examination shows that the failure mode for all the shorter dip times is a mixture of interfacial and cohesive, with the more durable tending towards cohesive within the adhesive. So, whilst this treatment is quite susceptible to environmental attack during the first few hours or so of exposure to the high humidity environment, it would appear, on balance, to be the most effective pretreatment for this alloy/adhesive system
Conclusions
This work has provided detailed results on the composition of various surface layers produced as pretreatments for aluminium, and has related them to the durability of bonds with a particular epoxy resin. The utility of the Boeing wedge test has been confirmed, but the importance of considering changes in strain energy release rate with time as well as simple the extent of crack growth has been brought out in a number of examples.
The newer P2 treatment, using ferric sulphate and sulphuric acid, has been shown to be broadly as effective as the well-established chromate-containing FPL etch.
Penetration of the adhesive into the pores of the surface film is an important feature of a durable bond. Penetration depends not only on the pore dimensions, but also on the contact angle between the adhesive and substrate, the adhesive viscosity and the viscosity-time characteristics at the temperature of application. Thus care must be exercised in specifying a 'good' pretreatment for durability without consideration of the specific adhesive to be used. In this work the finer pores of the SAA films did not allow adequate adhesive penetration for good durability, even with a phosphoric acid post-anodising dip, which had been shown to be efficacious with a lower viscosity adhesive. A short phosphoric acid dip after PAA treatment opened the pores further and gave modest improvement in durability.
Acknowledgements
The authors would like to acknowledge the financial support of DRA Farnborough.
Click here to return to David Packham's home page.
Revised 12.ix.96