SURFACE AND INTERFACE ANALYSIS

Surf. Interface Anal. 27, 705–715 (1999)

Adhesive Bonding of Hot-dipped Galvanized

Steel: Use of ToF-SIMS for Forensic Analysis
of Failed Joints

M. F. Fitzpatrick and J. F. Watts*

School of Mechanical and Materials Engineering, University of Surrey, Guildford, Surrey GU2 5XH, UK

The interfacial chemistry of environmental failure of adhesively bonded hot-dipped galvanized steel joints,
fabricated using a structural epoxy in a lap configuration, has been investigated by time-of-flight (ToF) SIMS.
The failed lap shear joints show areas of apparent interfacial failure but these regions are limited to thin
strips at the ends of the overlap, termed initiation zones. An initial study highlighted the importance of
small area surface analysis, using XPS, to demonstrate that electrochemical activity was responsible for the
initial bond degradation and the formation of the initiation zones at the ends of the overlap. Despite having
been employed successfully in a number of adhesion studies, XPS is unable to give the molecular level of
specificity that is required for a full understanding of the mechanism of such an adhesively bonded system.
A method has been developed, using ToF-SIMS, for mapping the initiation zone of the adhesive joint.
The images obtained support earlier evidence of electrochemical activity at the initiation zone showing the
presence of cations (Mg2Y ), indicating that cathodic behaviour played a role in the formation of the initiation
zone. The ToF-SIMS line scans indicate residual polymer in the initiation zone, which supports the hypothesis
of a dual effect of electrochemical behaviour and ingress of water being responsible for the formation of
the initiation zone. This suggests weakening rather than the clear separation, prior to mechanical testing,
observed in the case of classical cathodic delamination. The ToF-SIMS images extend this model by showing
corresponding cation-rich and adhesive-rich areas within the initiation zone, possibly demonstrating that the
different processes dominate in different regions (a result of localized electrochemical activity), and enables
cathode size to be estimated. Copyright  1999 John Wiley & Sons, Ltd.
KEYWORDS: ToF-SIMS; adhesive bonding; adhesion; environmental failure; hot dipped galvanized steel; epoxy structural adhesive

INTRODUCTION co-workers have studied in detail the interfacial chemistry

of corrosion-induced and mechanical-load-induced bond
degradation of a hot-dipped galvanized steel (HDGS)
One of the most important requirements of an adhe- system.5,6 Dickie discovered that electrochemical activ-
sive joint is the retention of strength during its ser- ity was responsible for interfacial failure regions, with
vice life. Durability is a key issue in the widening use anodic sites identified on the unloaded samples and both
of adhesive bonding technology and is a problem that anodic and cathodic sites identified on the loaded sam-
has been discussed by many authors.1,2 In the automo- ples. Electrochemical activity at an interfacial site in an
tive industry, adhesives are used successfully for non- iron substrate system has also been recognized by Davis
structural parts (bonnet, boot, etc.), but adhesive bonding and Watts, who reported that cathodic delamination was
has yet to be used for the primary structure of a steel responsible for a true interfacial failure.7
high-volume-production automobile. At present, the steel This paper reports a surface analysis investigation,
industry is concentrating on reducing weight with no sac- predominantly by time-of-flight (ToF) SIMS but instigated
rifice to performance, affordability, safety, durability and using preliminary work carried out by small-area XPS, of
cost; another benefit would be reduced fuel consumption.3 the failure mechanism of adhesively bonded hot-dipped
These aims have led to design changes, allowing a reduc- galvanized steel with a commercial epoxy. Lap shear
tion in thickness and the steady growth of materials such joints were exposed to a damp atmosphere for extended
as galvanized steels. These factors, however, have high- periods of time prior to mechanical testing. The failed
lighted potential problems, including in-service vibrations joints show areas of apparent interfacial failure but these
and reduced stiffness when using discontinuous joining regions are limited to thin strips at the ends of the overlap;
techniques such as spot welding. Adhesively bonded gal- for the purposes of this study, these thin strips have been
vanized steel flanges have the potential to overcome termed initiation zones. These initiation zones seem to
such shortcomings with no weight penalty.4 Dickie and be a result of environmental exposure and appear to act
as initiation sites for crack propagation on mechanical
testing, acting as ‘notch-like’ features. It is the study
* Correspondence to: J. F. Watts, School of Mechanical and Materi- of these areas of the failed surfaces that is reported
als Engineering, University of Surrey, Guildford, Surrey GU2
5XH, UK. in this paper, with a view to establishing the role of
E-mail: j.watts@surrey.ac.uk electrochemical activity at the crevice tip and its role in
Contract/grant sponsor: British Steel plc subsequent joint failure.
CCC 0142–2421/99/080705–11 $17.50 Received 27 October 1998
Copyright  1999 John Wiley & Sons, Ltd. Revised 8 February 1999; Accepted 12 February 1999
706 M. F. FITZPATRICK AND J. F. WATTS

The definition of the exact locus of failure and hence

Table 1. Composition of the 3M commercial epoxy
full understanding of failure interfaces is vital when
7823 as given by the safety data sheet
considering the mechanism of bond failure, because it is
the chemistry at the interface that determines the adhe- Composition
Component of adhesive (%)
sion of the system, which in turn has an effect on the
performance of the system in an aggressive environment. Epichlorohydrin bisphenol A epoxy 40 45
Surface analysis techniques such as XPS and SIMS offer Resin fillers 20 25
the potential of interfacial study, and have been widely Methacrylate butadiene styrene Resin 9 11
Aliphatic glycidal ether 5 10
used in previous work from this laboratory in the study Epoxy resin ester adduct 1 10
of both adhesion and subsequent failure mechanisms.7 – 9 Amorphous silica 2 5
Imaging ToF-SIMS, if used in the correct manner, could Strontium chromate(VI) 2 3
aid the study further by creating a two-dimensional visu- Phenolic resin 1 2
alization of the failure surface in a much shorter period 3-(4-Chlorophenyl)-1,1-dimethylurea 1 2
of time than could be achieved using line scan analysis
involving sample positioning and tilting for each analysis. Surface analysis
The current study highlights the ability of ToF-SIMS to
provide not only mass-selected images of diagnostic elec- Small-area XPS (250 µm) and ToF-SIMS were used
trochemical ions but also to map characteristic molecules to study the failure surfaces of the lap shear joints.
and mass fragments from the adhesive, and sets out to Analysis by XPS was performed using a VG Scien-
develop a rapid imaging technique for the analysis of tific EsCALAB 220iXL instrument. The excitation source
adhesively bonded lap joints of an HDGS/epoxy system. used was monochromatic Al K˛ radiation. ToF-SIMS was
carried out using a VG Scientific Type 23 instrument
equipped with a pulsed metal ion source, a two-stage
reflectron mass spectrometer and a 30 kV gallium liquid-
EXPERIMENTAL metal ion gun. The data system used was VGX 7000
software that runs on a DEC PDP11/73 computer under a
micro-RSX operating system. ToF-SIMS line scans were
Materials
constructed from spectra acquired across the initiation
zone of the respective failed surfaces. All the SIMS data
In this study, single lap shear specimens were fabricated were normalized by dividing the intensity of the ion of
from 110 mmð20 mm coupons with an overlap of 10 mm. interest by that of the total ion yield in the mass range
The substrate used in the study was 1.2 mm thick HDGS m/z D 5–400. After carefully establishing the static limit
that had been pretreated with a phosphate conversion coat- for the adhesive system, mass-selected images were col-
ing followed by a chrome rinse. A commercial epoxy lected to determine if certain ions were concentrated in
adhesive (3M 7823) was used and a glue line thickness particular areas, allowing their role in the mode of failure
was set at 250 µm using glass ballotini. Whilst the exact to be determined.
formulation of the adhesive is not revealed by the manu-
facturer, the safety data sheet indicates that the adhesive is
an epichlorohydrin–bisphenol A epoxy with fillers, amor- RESULTS AND DISCUSSION
phous silica and strontium chromate(VI), presumably as a
corrosion inhibitor; more detail is given in Table 1. The Visual assessment of failed surfaces
joints were aged for 12 months at 35 ° C in a 95% relative
humidity chamber and then tested mechanically using a The visual characteristics of the failure surfaces are shown
1195 Instron machine with a 5 kN load cell at a cross-head in Plate 1 and a schematic indicating the areas analysed is
speed of 5 mm min 1 . given in Fig. 1. It can be seen that the fracture path passes

Figure 1. A schematic representation of the failure of a lap shear joint, showing the position of the ToF-SIMS line scan.

Surf. Interface Anal. 27, 705–715 (1999) Copyright  1999 John Wiley & Sons, Ltd.
 Plate 1. Photograph of a typical joint used in the XPS
and ToF-SIMS.

Plate 2. ToF-SIMS image highlighting cation Plate 3. ToF-SIMS image showing that
rich areas, possibly indicating that local charactersitic polymer ions, also, form in
cathodic cells are set up within the initiation concentrated regions within the initiation
zone. (Map shows total counts, m/z = 40, 52 zone. This may give evidence to the ingress
and 24). of water playing a role in initial bond
degradation as well as electrochemical
activity. (Map shows total counts, m/z = 88,
77 + 91, 60 + 43).

Plate 4. Showing that cationic rich areas within the initiation zone do not coincide with polymer adhesive
rich areas. (a) m/z = 88 on 24; (b) m/z = 77 + 91 on 24.

Copyright © 1999 John Wiley & Sons, Ltd Surf. Interface Anal. 27 (1999)
 ADHESIVE BONDING OF HOT-DIPPED GALVANIZED STEEL 707

apparently at random through the overlap, with different for the metal side and 66.2% for the polymer side. More
routes being taken by the crack. With the joint shown, it notably there were cations (Mg2C and Ca2C ) and anions
can be seen that fracture has passed from close proximity (Cl ) present in the initiation zones but not in those adja-
of one substrate to the other, approximately half-way cent to them. The presence of these cations was always
along the overlap. Joints of this particular adhesive system associated with the low assay of carbon, which is indica-
all display this characteristic strip of interfacial failure at tive of the ‘metal’ failure surface. This indicates that they
the ends of the overlaps. By studying the formation of this are not markers for the inorganic components of the adhe-
so called ‘initiation’ zone it is hoped that an understanding sive, although this is probably where they have originated
of initial bond degradation may be acquired, so surface from. It is clear that such ions have leached from the
analysis was concentrated in this region during the study. adhesive during the exposure to high levels of humidity
An HDGS/adhesive system may have the potential for and are subsequently deposited on the area of cathodic
extensive electrochemical activity to occur, with exposed activity. If they were merely indicative of a failure within
iron acting as a cathode and protective zinc acting as the adhesive, such ions would always be observed when
the anode; the penetration of interfacial failure is not as a bulk of the adhesive is examined; this is not the case
great as with Davis and Watts’ true cathodic delamination for either XPS or ToF-SIMS analysis. Zinc peaks were
model experienced using pure iron substrates.7 observed on both sides of the failed joint, suggesting that
degradation of the substrate may have occurred.
The XPS analysis indicates that electrochemical activ-
Small area XPS ity has occurred within the initiation zone, with evi-
dent cathodic and anodic sites existing at the interface.
The XPS survey spectra were recorded from the loca- The carbon level, however, is higher than one would
tions identified in Fig. 1 and are shown in Fig. 2. Table 2 expect for pure cathodic delamination. This may sug-
indicates the quantified results of the elemental surface gest that the initiation was the product of a conjoint
analysis from areas of interest on the failed interfaces. The effect of aggressive hydroxyl ions and water ingress. This
spectra for the two initiation zones show carbon at 47.4% may explain the high carbon concentration and possibly

Figure 2. Small area XPS analysis taken from areas defined in Table 2.

Table 2. Quantification of small-area XPS analysis taken from the areas shown in Fig. 1
Area analysed
Overlap 1 Overlap 2
Metal initiation Adjacent Polymer Adjacent metal
zone polymer zone initiation zone zone
Elements (A) (B) (C) (D)

C ls 47.5 69.9 66.2 55.6

O ls 33.3 23.2 22.3 33.0
Zn 2p3/2 5.1 1.9 1.5 4.1
N ls 4.5 3.0 8.0 4.9
Si 2p 2.1 1.0 2.4
P 2p 6.2 2.03
Cl 2p 1.2 1.1

Copyright  1999 John Wiley & Sons, Ltd. Surf. Interface Anal. 27, 705–715 (1999)
708 M. F. FITZPATRICK AND J. F. WATTS

leads only to weakening of the interfacial bond and mass-selected data. Once these objectives had been met,
not actual separation. Small-area XPS analysis of the ToF-SIMS images could be acquired from the initiation
initiation zones at the ends of the overlap was able to zone of an aged lap shear joint.
demonstrate that electrochemical activity was responsible
for initial bond degradation. X-ray photoelectron spec-
Characterization of the adhesive. This was achieved by
troscopy offers elemental and chemical state informa-
first studying the uncured adhesive. A controlled volume
tion, together with the ability to yield quantitative surface
of a 1%(w/w) solution of the adhesive in acetone was
chemical analyses and indications of overlayer thickness;
dropped onto aluminium foil and the solvent was allowed
however, it inherently lacks the specificity to provide the
to evaporate. This yielded a thin polymer film that was not
molecular information essential for a full understanding
susceptible to electrostatic charging during analysis. The
of the adhesive and interfacial chemistry of a polymer.
ToF-SIMS, in contrast, demonstrates the potential to fulfil resultant positive SIMS spectra are presented in Fig. 3 and
this requirement with high spatial resolution via the use the expected ions from epoxy groups (at m/z D 91, 252
of a focused ion beam and a high degree of molecular and 269) and bisphenol groups (m/z D 135 and 231) are
sensitivity, which hopefully is capable of determining the seen; the relevant fragment patterns are shown in Table 3.
role of the polymer in failure. These observations are in agreement with previous work
on commercial materials of this type.10
Whilst these ions characterize the uncured adhesive,
Time-of-flight SIMS analysis of adhesive and failure they may not be characteristic of the cured adhesive that
surfaces will be present at the locus of failure of an adhesive
joint. A comparison of the adhesive in its uncured and
In order to achieve the intended aim of this work—the use cured state was made by preparing specimens in which
of mass-selected ToF-SIMS images as a diagnostic tool for thin layers of polymer were smeared onto aluminium
the analysis of adhesive joint failure surfaces—a series of substrates; one specimen was then heated at 180 ° C for
preliminary experiments were undertaken to ensure that all 30 min in air to simulate cure. Spectra from the uncured
the potential limitations encountered in static SIMS of a sample showed only traces of siloxane present, with peaks
complex polymer system were considered and, if possible, at m/z D 73 and 147, whereas for the cured material these
circumvented. A series of thin-film analyses were carried peaks were dominant in the spectrum, as shown in Fig. 4.
out to identify the characteristic peaks of the virgin adhe- This observation prompted the question of whether the
sive. This was followed by a damage study to determine new peaks at m/z D 73 and 147 were present as a
the static limit for the polymer system under consid- result of polymerization of siloxanes in the formulation
eration, which would ensure that mass-selected images as adhesion promoters or due to poly(dimethyl siloxane)
could be obtained within the static limit, thus ensuring (PDMS) to which SIMS is particularly sensitive and often
that the analysis was truly representative of the polymer appears in spectra of contaminated samples. In an effort
surface. Point analyses were used to construct line scans to rule out contamination during cure, the experiment was
across the initiation zone and to establish a protocol for repeated using an exceptionally clean oven. Similar spec-
the removal of surface topography in the presentation of tra were found for the two different curing environments,

Figure 3. Positive ToF-SIMS spectrum of a 1% (w/w) solution of Adhesive in acetone, characterizing the adhesive and showing typical
epoxy peaks at m/z 269, 252 and 191.

Surf. Interface Anal. 27, 705–715 (1999) Copyright  1999 John Wiley & Sons, Ltd.
 ADHESIVE BONDING OF HOT-DIPPED GALVANIZED STEEL 709

Table 3. Typical characteristic positive fragment ions of the epoxy-

type adhesive
m/z Fragment/ion

57 (epoxy)

191 (epoxy)

252 (epoxy)

269 (epoxy)

135 (bisphenol A)

213 (bisphenol A)

Figure 4. Positive ToF-SIMS spectrum of the cured adhesive.

indicating that the source of the siloxane was the adhesive the system. If this is the case with our adhesive system,
itself. we may not actually encounter the m/z D 73 and 147
Poly(dimethyl siloxane) is a known mould-release agent peaks in the final analysis of the fractured lap shear joint.
that, on curing, would have a propensity to segregate to the It was decided, therefore, to produce a simulated fracture
surface in an attempt to reduce the surface free energy of surface to analyse the bulk polymer/adhesive. This was
Copyright  1999 John Wiley & Sons, Ltd. Surf. Interface Anal. 27, 705–715 (1999)
710 M. F. FITZPATRICK AND J. F. WATTS

done by applying a thick smear of the adhesive onto a for each analysis. This procedure was repeated in different
metallic sample backing plate, usually used for mount- areas at different magnifications, enabling a static limit to
ing specimens. A circular cover plate was placed over the be obtained via plots of relative counts versus selected
adhesive, forcing a bulb of adhesive to rise; the adhesive masses (at a constant magnification) or relative counts
was cured and a simulated fracture was created by slicing versus magnification (for individual selected masses).
through the base of the adhesive bulb (Fig. 5). Analy- An accurate calculation of the primary ion dose is
sis of the simulated fracture surface showed a dramatic essential so that the limitations of static conditions can be
change in the relative intensities of the suspected PDMS understood clearly. With the initial analysis complete, the
peaks, bisphenol and other characteristic adhesive peaks total primary ion dose rates for the possible experimental
(m/z D 91, 77 and inorganic 88 SrC ), as shown in Fig. 6. variations were calculated using
Comparison of these data with those from a dry joint that
had failed in a cohesive fracture surface showed good Nf PIcont tp ð 6.25
Total primary ion dose D .1/
agreement, confirming that peaks observed at m/z D 73 A
and 147 did not play a major role in the characterization where A D analysis area (cm2 ), Nf D number of frames,
of an adhesive fracture surface. P D pulses per pixel (256 ð 256 pixels per frame),
With the adhesive fully characterized by fingerprint Icont D continuous primary beam current (nA) and tp D
spectra, the next step was to determine the static limit primary ion pulse length D 50 ns (1 amp D 6.25 ð
for the adhesive. 1018 ions s 1 ). In the case of the analytical conditions
used in the current work, the following values apply,
Establishment of the static limit for ToF-SIMS analysis of directly selectable from acquisition software: A and Icont
the adhesive. For any ToF-SIMS study of a polymer, a are determined as described below; N D 50; P D 32767;
static limit for the system must be established in order and tp D 50 ns.
to ensure analysis of the unmodified polymer. Having The primary ion current was measured using a Faraday
determined the static limit, it is important to discover if cup positioned in the path of the beam and a Keithley
useful images are capable of being achieved under the said Model 485 autoranging picoammeter; the analysis area
static limit. Consecutive analyses were carried out at one was interpolated from unpublished work from this labo-
magnification, thus incrementally increasing the ion dose ratory, in which the rastered analysis areas for different

Figure 5. Schematic representation showing the steps involved in fabricating a simulated fracture surface: (a) smear excess of adhesive
onto the metal backing plate; (b) place cover plate over the top, forcing a bulb of adhesive to rise; (c) after curing.

Figure 6. Positive ToF-SIMS spectrum of simulated fracture surface, showing the removal of siloxane peaks.

Surf. Interface Anal. 27, 705–715 (1999) Copyright  1999 John Wiley & Sons, Ltd.
 ADHESIVE BONDING OF HOT-DIPPED GALVANIZED STEEL 711

beam energies were determined by carrying out imaging

SIMS on a precision-ruled Si wafer that was etched at dif- Table 4. Characteristic positive fragments/ions
used in the damage study
ferent beam energies and four selected magnifications.11
With all the data required for total primary ion dose per m/z Fragment/ion
analysis collated, graphs showing relative peak intensity 67
versus total ion doses were plotted (Figs 7 and 8). All peak
intensities were normalized to total ions .m/z D 5–400/.
An alternative route to normalization was carried out using
91
the m/z D 27 ion, representative of C2 H3 C . This pro-
duced good results; indeed, the profiles were the same
shape as those presented below but, because of the pos-
sible presence of 27 AlC in the spectrum (perhaps from 115
an alumanosilicate filler in the adhesive, or residual alu-
minium on the HDGS surface), the normalization scheme
using the total ion intensity was preferred. Normalizing
data was considered necessary to compensate for the vary-
ing sensitivity of different masses, due to their different 128
cross-sections. Characteristic masses selected for moni-
toring damage were as follows: aromatics (m/z D 67, 91,
115, 128); other fingerprinting organic masses (m/z D 105,
135 and 141); and inorganic SrC (m/z D 88/, which was 105
chosen for reasons that will become clear later in the dis-
cussion (Table 4).
Figure 7 shows data from the damage study carried out
at ð500 magnification and, apart from an initial increase 141
in all masses, which may be considered to be a result
of removal of general hydrocarbon contaminant from the
surface, no decrease in the intensity of any of the peaks is
noted. This suggests that the ‘static’ limit for the polymer
system has not been reached by at least 2 ð 1013 ions appears that there are two ‘static’ limits for the system:
cm 2 . With magnification increased to ð1000, Fig. 8(a) one for the lower mass organics (2 ð 1013 ionic cm 2 ) and
and 8(b) show a continuous decrease in the lower mass one for the higher mass organics (4 ð 1013 ions cm 2 ).
aromatics (m/z D 67 and 91) after the first exposure; Despite the apparent ambiguity of the definition of static
this is accompanied by a complementary increase of conditions for the adhesive system, both values given are
inorganic mass 88. Such a result is surprising because within the general static limit of 1013 ions cm 2 as defined
it is normally the larger ions that would have been more by Briggs,12 and imaging should be achieved successfully
sensitive to damage. In this instance the emergence of the within the set experimental constraints to allow imaging
SrC is critically associated with the loss in intensity of of a damage-free surface
m/z D 67 and 91. Often, inorganic components such as
fillers or pigments are covered by a thin layer of organics Line scan across the initiation zone. With the adhesive
in adhesives. These organic molecules may, therefore, characterized and the experimental conditions verified, a
be more susceptible to damage than the other organic ToF-SIMS line scan was carried out across the initiation
molecules. Also, because these organic ions reduce in zone of an aged joint; Fig. 1 shows the areas analysed.
intensity, a matching rise in intensity would be expected Peaks of interest were chosen to conduct a profile of peak
from the attenuated inorganic. Taking these considerations intensity versus position. For example, the cationic peak
into account and examining the data shown in Fig. 8, it of Mg2C (m/z D 24) increased dramatically in intensity at

Figure 7. Graph showing relative peak intensities (normalized to total ions) versus total ion dose at ð500 magnification.

Copyright  1999 John Wiley & Sons, Ltd. Surf. Interface Anal. 27, 705–715 (1999)
712 M. F. FITZPATRICK AND J. F. WATTS

Figure 8. (a) Graph showing relative peak intensity (normalized to total ions) versus total ion dose at ð1000 magnification. Note the
rapid decrease of intensity of medium mass peaks at m/z D 67 and 91, matched with the increase at m/z D 88. (b) Graph showing
relative peak intensity (normalized to total ions) versus total ion dose at ð1000 magnification (medium to higher mass).

the crack tip of the initiation zone. The original profile may be a concentration of cations at the crack tip. This
of element intensity versus relative position (Fig. 9a) is confirmed when looking at normalized line scans of
displays how the line scan was dominated by topography; m/z D 24 and the characteristic epoxy mass m/z D 135.
by once again normalizing intensities to total ions, it The line scan also reinforces the importance of viewing
has been possible to resolve topographical issues and see the initiation zone in the form of a two-dimensional image,
how relative intensities of selected masses vary across because the images shown in the results display cation-
the initiation zone. Here, the sensitivity of SIMS was rich areas not just confined to the initiation crack tip.
highlighted, with traces of polymer being found for the It also reinforces the view that the Mg2C ion can be
first time in the initiation zone. used confidently as a marker for cathodic activity. The
The importance of the line scan becomes apparent when inclusion of the 88 SrC profile in Fig. 9 (SrC is indicative
analysing the normalized intensities (Fig. 9b) and it is of the strontium chromate pigment) is shown to vary in
easy to see distinctive surface characteristics associated concentration with the epoxy fragment at m/z D 135 rather
with different areas across the initiation zone and the than the 24 MgC ion.
thick polymer section. The line scan, therefore, highlights Once all the preliminary steps for the development of a
ions and fragments that would be relevant in subsequent rapid imaging technique for the forensic analysis of failed
imaging. The line scan also offers a one-dimensional joints have been completed and relevant and character-
view of the electrochemical activity in the initiation zone, istic peaks have been identified, it becomes possible for
and visual assessment of the spectra indicates that there suitable images of the initiation zone to be acquired and
Surf. Interface Anal. 27, 705–715 (1999) Copyright  1999 John Wiley & Sons, Ltd.
 ADHESIVE BONDING OF HOT-DIPPED GALVANIZED STEEL 713

Figure 9. Positive ToF-SIMS line scan across the initiation zone as described in Fig. 1: (a) selected masses demonstrating dominance
C
of topography; (b) normalized line scans of m/z 24 (MgC ) and m/z 135 (C9 H11 O ) markers for cation and adhesive, respectively both
normalized to total ion counts.

so contribute to the understanding of the formation of the The images obtained display well-defined cationic-rich
initiation zone. zones (Plate 2), possibly suggesting that local cathodic
cells are set up in within the initiation zone. The images
Time-of-flight SIMS images of a single lap joint initiation characteristic of adhesive peaks (m/z D 77 and 91 and
zone. Peaks of interest for mapping of the initiation zone m/z D 88) also show traces of polymer within the initi-
were considered to be as follows: ation zone to be ‘patchy’ (Plate 3), indicating that water
ingress may weaken areas within the adhesive and so lead
40
Ca2C and 24 Mg2C —cationic markers that seem to to very thin cohesive failure through the polymer within
play a key role in failure; the initiation zone. The obvious question of whether the
Sr—inorganic marker; cationic Mg2C patches coincided with the adhesive patches
m/z D 77 and 91, 105 and 115—organic markers; was simply answered by overlaying the two sets of part-
m/z D 60 and 43—products of amine curing agent; and ner images. Plate 4 confirms that there are areas within the
52
Cr2C and 64 Zn2C —substrate markers. initiation zone that are cation rich and adhesive deficient
Copyright  1999 John Wiley & Sons, Ltd. Surf. Interface Anal. 27, 705–715 (1999)
714 M. F. FITZPATRICK AND J. F. WATTS

and, conversely, adhesively rich areas that are deficient in crack tip. A major advantage of the imaging technique is
cations. that it descibes a two-dimensional profile and so allows an
The data acquired in this manner could be improved estimation of the size of the cathodic cells, which in the
clearly by the use of an appropriate algorithm to correct case shown appears of the order of 100 µm across. The
for sample topography. The data of Fig. 9 show that this sensitivity of SIMS is demonstrated by the clear poly-
can be achieved readily by the use of the total ion yield mer images obtained. These patches of polymer within
or, indeed, a suitable non-specific ion such as 27 C2 H3 C . the initiation zone, along with the corresponding images,
This is not possible using the VG Scientific data system, suggest that a conjoint effect seems to be responsible for
although the problem has been circumvented recently on the formation of the initiation zone. The traces of polymer
our Poshenreider ToF-SIMS by the installation of a Kore seem to indicate that water ingress plays a role, along with
Scientific TDC and data system that is PC based. cathodic behaviour, creating the cathodic weakening effect
proposed earlier rather than the pure cathodic delamina-
tion model proposed by Davis and Watts.7 This theory was
FAILURE MECHANISM: THE ROLE OF supported further by the earlier small-area XPS data that
ELECTROCHEMICAL ACTIVITY showed 47% C 1s at the initiation zone: a level not high
enough for a cohesive polymer surface but too high for a
pure interfacial cathodic delamination failure surface.
As mentioned in the introduction, cathodic behaviour has The mapping technique was able to take the work a
been found to be responsible for initial degradation in step further by analysing a large area relatively quickly,
other lap shear systems. Cathodic activity was defined demonstrating that there seems to be corresponding cation-
by the presence of cations at the interface. The idea of and polymer-rich areas within the initiation zone. This is
cathodic markers identifying electrochemical activity was the principal advantage of the imaging technique; pro-
first introduced by Castle and Epler in the mid-1970s.13 vided that vital preliminary work has been carried out to
Cathodic delamination can be defined as the detachment of understand the spectra of the polymer, a sample with a
strongly adhering organic coatings under the influence of known zone of interest can be analysed quickly to yield
a cathodic potential and has been recognized as a problem spatial information on a two-dimensional plane. This rapid
for electrochemical active surfaces since 1936. imaging technique confirmed the earlier conjoint effect
Consider a metal discontinuously covered with an theories of the formation of an initiation zone and pro-
organic coating. The resulting system is vulnerable to gressed the research by observing that electrochemical
cathodic delamination when subsequently exposed to a behaviour and water ingress dominate in different regions.
hostile atmosphere. The coated metal may behave as a
cathode (electron accepting) and the exposed metal as an
anode (electron donating). The ensuing electrochemical CONCLUSIONS
reaction produces cathodic products (specifically hydroxyl
products) under the coating that may have deleterious
effects on the bonding of the metal to the coating and A thin strip of interfacial failure has been identified at
so cause the coating to ‘delaminate’. Cathodic reactions the ends of overlaps on a Hots/epoxy lap shear sys-
in the presence of oxygen involve an increase in pH, either tem: such strips are called initiation zones. Small-area
by the consumption of hydrogen ions or by the reduction XPS established electrochemical activity in the initiation
of oxygen to hydroxyl ions zone. Following this preliminary result a method has been
developed for mapping the initiation zone of an adhe-
O2 C 2H2 O C 4c ) 4OH .2/ sive joint successfully, with images obtained supporting
earlier evidence of cathodic behaviour playing a role in
The above explanation is widely accepted and has the formation of the initiation zone highlighted by the
been reported extensively by Watts in previous presence of cations .Mg2C /. Line scan analysis showing
communications,14,15 but is particularly applicable to polymer in the initiation zone also supports the claim of
an HDGS substrate. A freshly cut edge of HDGS the previous studies of a possible dual effect of electro-
is a natural galvanic cell, with the zinc coating chemical behaviour and ingress of water being responsible
corroding preferentially to the protected iron under anodic for the formation of the initiation zone, suggesting weak-
dissolution ening rather than separation prior to mechanical testing,
Zn ) Zn2C C 2e .3/ as observed by Davis and Watts’ cathodic delamination
model.7 The ToF-SIMS imaging approach is able to take
As a result, the cathodic reaction will take place, gen- the work a step further by showing corresponding cation-
erating hydroxyl ions due to the reduction of water and rich and adhesive-rich areas within the initiation zone,
oxygen via the consumption of electrons. These aggressive showing local cathodic cells of ¾100 µm across, demon-
hydroxyl ions are able to penetrate between the polymer strating that the different processes dominate in different
adhesive and the metal substrate, thus weakening the inter- regions.
facial bond and allowing anodic dissolution to continue,
as proposed by Dickie et al.6 Acknowledgements
The images obtained support earlier evidence of elec-
trochemical activity at the initiation zone showing the This work is part of a project funded by British Steel plc. The authors
presence of cations .Mg2C /, indicating the role of cathodic thank Dr John Ling (Welsh Technology Centre) and Drs Alan Seeds
and Anthony Cronin (Automotive Engineering Group) for their advice,
behaviour in the formation of the initiation zone. It should guidance and encouragement. The authors also wish to thank Dr Tim
be noted that this concentration of cations is ‘patchy’ and Carney (VG Scientific), who carried out the small area XPS, and
not, as suggested by the line-scan data, concentrated at the Mr Andy Brown, who assisted with the ToF-SIMS data acquisition.

Surf. Interface Anal. 27, 705 715 (1999) Copyright  1999 John Wiley & Sons, Ltd.
 ADHESIVE BONDING OF HOT-DIPPED GALVANIZED STEEL 715

REFERENCES

1. R. A. Cayless, Surf. Interface Anal. 17, 430 (1991). 9. J. F. Watts, Handbook of Surface and Interface Analysis, ed.
2. R. A. Dickie and S. M. Ward, in First International Congress by J. C. Rivière and S. Myhra, Marcel Dekker, New York,
on Adhesion Science and Technology, Amsterdam, 16 20 USA. PP. 781 831 (1998)
October (1995). 10. J. A. Treverton, A. J. Paul and J. C. Vickerman, Surf. Interface
3. J. Kennedy, Mater. World 1, 659 (1993). Anal. 20, 449 (1993).
4. P. L. Couduti and Paratore, International Congress and Expo- 11. A. Rodrigez, MSc Thesis, University of Surrey, Guildford
sition SAE Technical Papers, Detroit, MI, USA. pp. 13 19 (1994).
(1997). 12. D. Briggs and A. B. Wooton, Surf. Interface Anal. 4, 109
5. R. A. Dickie, J. Holubka and J. DeVries, J. Adhes. Sci. (1982).
Technol. 4, 57 (1990). 13. J. E. Castle and D. C. Epler, Surf. Sci. 53, 286 (1975).
6. R. A. Dickie, M. Debolt, L. Haak and J. DeVries, J. Adhes. Sci. 14. J. F. Watts, J. Adhes. 31, 73 (1989).
Technol. 8, 1413 (1994). 15. J. E. Castle and J. F. Watts, Corrosion Control by Organic
7. S. J. Davis and J. F. Watts, J. Mater. Chem. 6, 479 (1996). Coatings, pp. 78 86. NACE, Houston, TX, USA. (1981).
8. J. F. Watts and A. M. Taylor, J. Adhes. 55, 99 (1995).

Copyright  1999 John Wiley & Sons, Ltd. Surf. Interface Anal. 27, 705–715 (1999)