failure mechanism of phosphated adhesively bonded hot-dipped galvanized steel: a small-area xps...

SURFACE AND INTERFACE ANALYSISSurf. Interface Anal. 29, 131–138 (2000)

Failure mechanism of phosphated adhesivelybonded hot-dipped galvanized steel: a small-areaXPS study

M. F. Fitzpatrick, 1 J. S. G. Ling2 and J. F. Watts1*1 School of Mechanical and Materials Engineering, University of Surrey, Guildford, Surrey GU2 5XH, UK2 Corus plc, Port Talbot, West Glamorgan, SA13 2NG, UK

The failure mechanism of an aged, phosphated, hot-dipped galvanized steel (HDGS) sealant adhesive lap jointsystem has been investigated using XPS. In a previous study, imaging time-of-flight (ToF) SIMS identifiedelectrochemical activity in the form of local cathodic cells set up within an initiation zone: thin strips ofvisually interfacial failure at the ends of the overlaps. Complementary polymer maps also identified areas ofcohesive failure, suggesting a dual effect of electrochemical behaviour and ingress of water being responsiblefor the formation of the initiation zone. The role of electrochemical activity in the initiation zone and itseffect on the substrate are considered in this paper.

Small-area XPS (at a spatial resolution of 20µm) has been used to characterize areas within the initiationzone that are deficient in pretreatment and of similar size to the local cathodes observed in the ToF-SIMSstudy. Quantified XPS linescan data identified areas of high carbon content, once again inferring that wateringress plays a part in the failure mechanism. Scanning electron microscopy also identified denuded areasof the phosphate pretreatment that are a result of exposure to a hostile environment, because a uniformpretreatment exists prior to bonding. Dissolution of phosphate crystals may have occurred within the alkalineenvironment produced by the cathodic half-reaction. Copyright 2000 John Wiley & Sons, Ltd.

KEYWORDS: small area XPS; adhesive bonding; hot dipped galvanized steel; cathodic delamination; failure mechanisms; lap shear joints;phosphating

INTRODUCTION

Adhesive bonding offers many advantages over other join-ing techniques, such as the ability to join dissimilar mate-rials, to improve stress distribution through a joint, etc.,but it is widely accepted that the durability of a bondexposed to a hostile atmosphere is the most significantproblem in the development of adhesive bonding for struc-tural applications.1 The response of adhesively bondedhot-dipped galvanized steel (HDGS) to load and environ-ment was studied by Dickieet al.,2,3 who reported electro-chemical activity as an important precursor to interfacialfailure. Davis and Watts4 reached similar conclusions onmild steel, which confirmed that true interfacial failurewas a result of cathodic delamination. In a previous paperfrom this laboratory, small-area XPS established that elec-trochemistry played an important role in the initial bonddegradation of a phosphated HDGS lap joint.5

Previous work from this laboratory has presented animaging time-of-flight (ToF) SIMS investigation of adhe-sively bonded HDGS, which identified areas of apparentinterfacial failure limited to thin strips at the end of the lapshear joint overlap, termed ‘initiation’ zones.6 The ToF-SIMS images from these regions identified definite localcathodic cells (¾100 µm in size) and also complemen-tary polymer-rich areas, suggesting a cathodic ‘weakening’

* Correspondence to: J. F. Watts, School of Mechanical and Materi-als Engineering, University of Surrey, Guildford, Surrey GU2 5XH, UK.E-mail: [email protected]

effect due to water ingress and cathodic activity, rather thanclassical cathodic delamination. The establishment of elec-trochemical activity within the initiation zone was criticalin defining the mechanism of failure for the joint system.The role of electrochemical activity and its effect on thesubstrate is one of the aspects considered in this paper.

It is generally accepted that the performance of anadhesive system can be improved through the use of apretreatment.7 By changing the nature of the substratesurface (e.g. chemistry or morphology) it is possibleto reduce the rate of corrosion and bring a metallicsurface to a condition that will favour the acceptanceof a continuous and adherent protective organic coating;this is often achieved in practice by the application of aconversion coating. Phosphating is the most widely usedform of metal pretreatment; it has been used for over100 years and if deposited under proper conditions givesthe substrate better corrosion resistance and adhesion ofcoatings.8,9 Phosphating has traditionally been used inthe paint industry for consumer goods and automotiveapplication to provide improved corrosion resistance anda larger interfacial area for bonding, it can also be usedas an effective pretreatment for the adhesive bondingof HDGS. This paper reports the changes that occur inthe pretreatment layer within an initiation zone of anadhesively bonded HDGS lap shear joint on exposure toa humid environment. These observations are interpretedin terms of the degradation of the pretreatment layer andare considered in the light of currently accepted failuremechanisms when the joint is exposed to an aggressiveaqueous environment.

Copyright 2000 John Wiley & Sons, Ltd. Received 13 September 1999Revised 26 November 1999; Accepted 26 November 1999

132 M. F. FITZPATRICKET AL.

EXPERIMENTAL

The substrate used in the study was 1.2 mm thick HDGSthat had been pretreated with a phosphate conversion coat-ing followed by a chrome rinse. Single lap shear speci-mens were fabricated from 110 mmð 20 mm couponswith an overlap of 10 mm. A polybutadiene sealant adhe-sive (Evode M33) was used and the glue line thicknesswas set at 250µm using glass ballotini during fabrication.The joints were aged for 12 months at 35°C in 95% rela-tive humidity and then mechanically tested using a 1195Instron machine with a 5 kN load cell at a cross-headspeed of 5 mm min�1.

X-ray photoelectron spectroscopy analysis was per-formed with a VG Scientific Sigma Probe using amicrofocused monochromatic Al K̨x-ray source. Thisspectrometer has an ultimate spatial resolution of betterthan 15µm and uses a novel form of transfer lens, termeda Radian lens, which, as a result of its large acceptanceangle, improves sensitivity and reduces acquisition time.

For insulating samples a flood gun supplying electronsof 1 V energy was used for the management of sur-face charge. The spectrometer was controlled by a VGScientific Eclipse data system, which was also used forsubsequent data processing. Scanning electron microscopyanalysis was carried out using a Hitachi S-3200N in thefull vacuum mode; samples were gold coated and a pri-mary beam potential of 20 kV was used.

RESULTS AND DISCUSSION

Visual assessment of the failure surfaces

The visual characteristics of the failure surfaces and areasanalysed are shown as a schematic in Fig. 1. The adhesiveexhibits cohesive failure throughout most of the overlap,with large amounts of polymer on both failure surfaces.There are, however, strips of interfacial failure at theends of the joint overlaps akin to the initiation zonesobserved on joint systems fabricated with structural epoxy

Figure 1. A schematic of the visual failure characteristics of a typical joint discussed in the study: (a) a side-on view of the fracturepath; (b) a plan view of failed overlap interfaces, identifying the metal initiation zone, ballotini ‘pits’ and areas from which the XPSspectra shown in Figs 5, 8 and 10 were acquired.

Surf. Interface Anal. 29, 131–138 (2000) Copyright 2000 John Wiley & Sons, Ltd.

FAILURE MECHANISM OF PHOSPHATED HDGS 133

adhesive.6 The load to failure (0.65 kN) was uncharac-teristically low for an adhesively bonded joint, resultingin no plastic deformation of the substrate; with strongerstructural adhesives, plastically deformed hinges are oftenobserved adjacent to the overlap area.

Characterization of the phosphated substrate

One of the principal aims of this study was to investi-gate the effect of any possible electrochemical activity

Figure 2. Scanning electron micrograph of as-received phos-phate crystals on HDGS, showing uniform coverage of pretreat-ment.

Figure 3. Scanning electron micrograph of as-received phos-phate crystals on HDGS, showing ‘micro’ cracks through indi-vidual crystals.

on the substrate within the initiation zone. In order todo this the individual components of the lap joint systemmust be characterized fully. A SEM investigation of theas-received phosphated HDGS found coverage to be uni-form. Figure 2, although only showing a small area, is atypical representation of the surface and sufficiently char-acterizes the phosphate coverage, showing no denudedareas. Closer inspection of the crystals themselves (Fig. 3)indicates that there are defects in the individual crystals.Cracks of the order of micrometres seem to have been

Figure 4. Large-area XPS (400 µm diameter spot size) from unbonded phosphate HDGS substrate: (a) standard survey spectrum0 1200 eV; (b) enlarged section of survey spectrum 0 200 eV.

Copyright 2000 John Wiley & Sons, Ltd. Surf. Interface Anal. 29, 131–138 (2000)


introduced into the crystals, more than likely at some pointduring the deposition process or subsequent processes ofthe treated strip. A large-area XPS survey spectrum takenfrom a section of the unbonded substrate material allowscharacteristic peaks to be identified (Fig. 4). Well-definedP 2p and P 2s transitions at 133 and 190 eV, respec-tively, can be seen and are indicative of the phosphatepretreatment; silicon peaks are present and may well haveoriginated from the phosphating bath, because silicatesmay be present as surfactant additives. Evidence of thechrome rinse is apparent with the Cr 2p doublet dis-cernible, despite the fact that it is coincident with someof the zinc Auger peaks, which effectively means that thedetection limit for Cr in this system is poorer than nor-mally encountered in XPS. Zinc peaks are clearly apparentfrom the zinc phosphate pretreatment. The quantitativesurface chemical analysis by XPS is presented in Table 1.

Analysis of the failure interfaces using XPS and SEM

Large-area XPS analysis (400µm diameter spot size) car-ried out on the middle of joint overlaps confirms thecohesive nature of failure for the majority of the over-lap, Fig. 5 shows a typical spectrum from the middle ofthe overlap and is characteristic of the adhesive only. Thesulphur is present as a cross-linking agent and calcium isadded as a filler in the form of calcium carbonate. Quan-tification of the spectrum shows both sulphur and calcium

Table 1. Quantitative surface analysis of the spectra shownin Figs 4, 5, 8 and 10

Element (at.%)Spectra C 1s O 1s Zn 2p3/2 P 2p S 2p Ca 2p Cr 2p3/2 Si 2p

As-received phos-phated HDGS 38.4 44.4 2.2 3.0 5.2 6.8Cohesively failedpolymer 70.9 19.1 1.3 4.3 4.3Ballotini ‘pit’ 88.8 7.1 4.1Initiation zone,bright area 17.7 49.9 30.2 2.2Initiation zone,dark area 44.8 35 8.4 2.0 5.3 4.5

to exist at¾4 at.%. The zinc peaks (1.2 at.%) are notindicative of the substrate but are actually characteristicof the adhesive, which was included in the formulationas zinc oxide, acting as an inhibitor. A SEM image ofthe overlap (Fig. 6) clearly shows the visually interfacialinitiation zone discussed above but also distinct holes inthe adjacent polymer that have been left by the ballotiniafter mechanical testing of the joints. Scanning electronmicroscopy of these ballotini ‘pits’ appears to identifyphosphate crystals from the pretreatment, this calls intoquestion the integrity of the bond between the polymerand the substrate (Fig. 7). Because the characteristic phos-phate morphology is seen at the base of the ‘pit’, this

Figure 5. Large-area XPS (400 µm diameter spot size) taken from an area of cohesive failure of bulk M33 adhesive: (a) standard surveyspectrum 0 1200 eV; (b) enlarged section of survey spectrum 0 200 eV.



Figure 6. Scanning electron micrograph of an aged joint overlap showing an initiation zone and holes left by ballotini that wereremoved on mechanical testing.

Figure 7. Scanning electron micrograph from the base of aballotini pit, showing the apparent presence of phosphatecrystals.

would seem to indicate that there was no adhesive betweensubstrate and ballotini, thus reducing the effective load-bearing capacity of the adhesive joint. An XPS surveyspectrum at 50µm diameter spot size taken from the bot-tom of a ‘pit’ (Fig. 8) did not reveal the presence of anyphosphorus (characteristic of the phosphate pretreatment),

only carbon, oxygen and sulphur peaks characteristic ofthe polymer, suggesting that the presence of ballotini doesnot compromise the bond between substrate and adhesive.It should be noted, however, that there are differencesbetween the two characteristic polymer spectra shown inFigs 5 and 8, with a noticeable reduction of oxygen andlack of inorganic calcium present in the spectrum acquiredfrom the ballotini pit area. The difference in the spectraof Figs 5 and 8 is not unexpected and indeed typifies pre-vious results in which the polymer adjacent to a solidsurface is devoid of all inorganic material as a resultof the effective dispersion of pigments and extenders inthe resin. This results from a specific interaction betweenpolymer and substrate and is highlighted from the baseof the ballotini pits. The cohesive failure of the adhe-sive, on the other hand, exposes inorganic additions tothe adhesive, such as calcium carbonate and other minorconstituents.

The requirement for small-area analysis in this studywas essential, this is readily achieved with the VG Sci-entific Sigma Probe with an ultimate spatial resolutionof 15 µm. An optical microscope with a 400µm fieldof view, arranged in tandem with a light probe alignedthrough the x-ray monochromator, is another feature thatallows samples to be inspected in detail, as well as ensur-ing the exact location of analysis, which is essential when

Figure 8. Small-area XPS (50 µm diameter spot size) spectrum from the base of a ballotini pit, characteristic of polymer with no signof substrate peaks.



using the smallest spot size. The improved spatial resolu-tion was used to good effect by an initial investigation ofthe initiation zone. The optical image indicated the exis-tence of dark and bright patches¾100 µm in size withinthe initiation zone, as shown in Fig. 9. Small-area XPSat 20 µm resolution was used in these areas to ensurethat analysis was confined to the bright and dark areas,thus allowing the difference in composition between therespective areas to be interpreted. The bright area givesa spectrum indicating a very clean (polymer-free) zincsurface with a high zinc (30.2 at.%) and low carbon(17.7 at.%) content [Fig. 10(a)]. The high amount of zincpresent would suggest that the HDGS substrate was beingdetected, a because 30 at.% is far higher than was foundon the pretreated substrate material. The spectrum fromthe dark areas of the initiation zone [Fig. 10(b)] yieldsa slightly higher C 1s intensity (44.8 at%). This is prob-ably low enough to still indicate an adhesive-free area;however, inorganic components of the adhesive (sulphurcompounds are used as cross-linking agents), togetherwith the slightly increased C 1s line, indicate that a traceof polymer may be left behind at the interface on fail-ure. The presence of phosphorus peaks (P 2p and P 2s) in

the spectrum from the dark region is the most significantdifference between the two spectra shown in Fig. 10. Asalready mentioned, phosphorus peaks are indicative of thephosphate pretreatment and their absence from the brightarea of the spectrum may well be the result of dissolu-tion of the pretreatment layer, resulting in a very cleanspectrum of the galvanized steel substrate, which wouldaccount for the low carbon assay reported above.

Analysis of the bright and dark areas was followed bya series of point analyses (with a 50µm spot) spacedequidistant across the initiation zone and identified aspoints 1–6 in Fig. 11. The subsequent XPS data werequantified and plotted against relative position across theinitiation zone, allowing the intensities of the relevantelements to be determined, as shown in the line scanof Fig. 12. On inspection of the carbon concentrationline it appears that only points 5 and 6 are at a levelthat is low enough to be representative of a clean metalinterfacial surface. Although points 3 and 4 are visuallyinterfacial metal, they possess a level of carbon between70 and 80 at.%. This is high and indicates the presenceof a thin polymer overlayer, which is consistent with thework carried out in the previous SIMS study, where water

Figure 9. Optical microscope image taken by the Sigma Probe of an initiation zone within a joint, exposing the light and dark areas thatexist. During mechanical testing the load was applied in a direction that is consistent with a horizontal orientation in this micrograph.

Figure 10. Spectra from (a) bright area in initiation zone showing clean zinc surface and (b) dark areas with P 2s and P 2p peaksindicative of pretreatment; both of these zones are within the initiation zone.



Figure 11. Photograph of metal initiation zone showing posi-tions of small-area point analysis (50 µm diameter spot size)used to construct a line scan.

Figure 12. Quantified data from point analysis across theinitiation zone used to construct a line scan.

ingress was determined to play a role in the formation ofthe initiation zone, leading to cohesive failure.6 The low-carbon areas are matched with increases in oxygen andzinc, inferring that an interfacial failure initiation zoneof ¾100 µm exists. The 100µm clean zinc region alsocoincides with a significant increase in the amount ofcalcium present, which could be used as a marker forcathodic activity and is once again in agreement withprevious work.6 In point 3 of the line scan the only traceof phosphate pretreatment is observed; this reinforcesthe idea of patchiness of the phosphate layer within theinitiation zone. A SEM inspection of the initiation zone(Fig. 13) confirms that there are areas void of phosphatecrystals within an aged joint’s initiation zone. Bare patchesof phosphate pretreatment could be present as a result ofdamage after mechanical testing, with crystals fracturingat the basal plane; however, if this were the case the

Figure 13. Scanning electron micrograph showing denudedphosphate regions with a metal initiation zone.

phosphate basal plane would still be detected by XPS eventhough the acicular crystals would have been removed.

The denuded areas of phosphate pretreatment may wellbe the result of attack from aggressive hydroxyl ions:the product of cathodic activity. This has been discussedpreviously by Wiggleet al.,7 who, on investigating cor-rosion at scratches in paint films, reported that hydroxideproduced by the cathodic corrosion reaction chemicallydissolves the zinc phosphate, producing adhesion failure.Robertset al.10 reported that there is a very strong depen-dence of the rate of removal of the phosphate coating withpH and has shown direct evidence that zinc phosphatedegradation occurs by alkali attack. Further evidence thatmay support the argument for cathodic activity not onlybeing responsible for the formation of initiation zones butalso for the subsequent dissolution of phosphate conver-sion coating in adhesive bonds is that all interfacial failurehas occurred within a narrow region around the edges ofthe failed interfaces. Corrosion that is dominated by acathodic reaction is limited to areas where there is anabundant supply of oxygen. This is more likely to occurat the edge of the overlap where the diffusion path foroxygen and water is easiest.

CONCLUSIONS

The failure mechanism of an aged phosphated HDGSsealant adhesive lap joint system has been investigatedusing XPS. As expected, owing to the nature of thesealant adhesive type, the majority of the joint overlaphas failed cohesively. Previously reported5 initiation zoneswere observed and studied in detail using small-area XPS.A previous imaging SIMS study had identified corre-sponding cation- and polymer-rich zones showing localand cathodic cells¾100 µm across, the present studyhas also identified traces of polymer within the initiationzone accompanied by denuded phosphate patches. Usingthe optical microscope it proved possible to select typi-cal areas within the 400µm field of view to characterizeareas within the initiation zone that are deficient in pre-treatment and are similar in size to the local cathodesobserved in the ToF-SIMS study. Such areas were anal-ysed by XPS using a 20µm diameter spot. The XPS work



offers complementary data to the ToF-SIMS images andshows the effect of cathodic activity on the phosphatepretreatment. The production of aggressive hydroxyl ionsvia the cathodic corrosion reaction has resulted in thedissolution of phosphate crystals within the pretreatment.Small-area XPS identified areas with and without phos-phate pretreatment and the quantified XPS line-scan datashow areas of high carbon content supporting the ToF-SIMS data, implying the presence of polymer adhesivewithin the initiation zone. This would suggest that fail-ure was not only a result of cathodic behaviour but wateringress also, resulting in a weakening rather than a sep-aration prior to mechanical testing as observed by Davisand Watts in their model of cathodic delamination.3 A

SEM investigation of the phosphated substrate showedthat denuded phosphate areas were a result of exposure torelative humidity because uniform phosphate crystal cov-erage exists prior to bonding, although micro cracks wereobserved at higher magnification.

Acknowledgements

M.F.F. thanks Corus plc. for financial support and Dr Tim English(Swinden Technology Centre) and Alan Seeds and Anthony Cronin(Automotive Engineering Group) for their advice and guidance. Theauthors wish to thank Dr B.J. Hewitt, Director Technical, Corus plcand Dr R. Alderdice, Manager Product Technology Group, WelshTechnology Centre, for permission to publish this paper. They also thankDrs Kevin Robinson and Richard White at VG Scientific who carriedout the preliminary small-area XPS.

REFERENCES

1. Cayless RA. Surf. Interface Anal. 1991; 17: 430.2. Dickie RA, Holubka J, DeVries J. J. Adhes. Sci. Technol.

1990; 4: 57.3. Dickie RA, Debolt M, Haak L. J. Adhes. Sci. Technol. 1994; 8:

1413.4. Davis SJ, Watts JF. J. Mater. Chem. 1996; 6: 479.5. Watts JF, Fitzpatrick MF. ECASIA’97, 7th European Con-

ference on Applications of Surface and Interface Analysis,Olefjord I, Nyborg L, Briggs D (eds). Wiley: Chichester, 1997;113 116.

6. Fitzpatrick MF, Watts JF. Surf. Interface Anal. 1999; 27: 705.7. Wiggle RR, Smith AG, Petrocelli JV. J. Paint Technol. 1968;

40: 174.8. Lorin G. Phosphating of Metals. Clare O’ Molsey: Surrey,

1974.9. Freeman DB. Phosphating and Metal Pre-treatments.

Woodhead-Faulkener: Bucks, 1986.10. Roberts TR, Kolts J, Steele Jr JH. SAE Tech. Pap. 1980;

800443.


failure mechanism of phosphated adhesively bonded hot-dipped galvanized steel: a small-area xps...

Documents