Scanning Kelvin Probe Blister test measurements of adhesive delamination – Bridging the gap between experiment and theory

International Journal of Adhesion & Adhesives 73 (2017) 8–15

Contents lists available at ScienceDirect

International Journal of Adhesion and Adhesives journal homepage: www.elsevier.com/locate/ijadhadh

crossmark

Scanning Kelvin Probe Blister test measurements of adhesive delamination – Bridging the gap between experiment and theory R. Grothea, Chen-Ni Liua, M. Baumertb, O. Hesebeckb, G. Grundmeiera, a b

⁎

University of Paderborn, Department for Technical and Macromolecular Chemistry, Warburger Str. 100, 33098 Paderborn, Germany Fraunhofer IFAM, Wiener Str. 12, 28359 Bremen, Germany

A R T I C L E I N F O

A BS T RAC T

Keywords: Epoxides Metals Non-destructive testing Scanning kelvin probe blister Test Peel Finite element stress analysis Microscopy Delamination

The delamination of an epoxy-adhesive film from a zinc coated steel substrate was studied by means of the electrochemical Height Regulated Scanning Kelvin Probe Blister Test (HR-SKP-BT1) under controlled atmospheric conditions, applied pressure and interfacial electrode potential. The experimental studies focused on the analysis of the critical environmental water activity that leads to a corrosive delamination process under applied mechanical load and the analysis of the corrosion and delamination mechanisms at the front of delamination. The influence of applied pressure and relative humidity on the increase in the maximum blister height and the delamination rate was measured under constant polarization of the defect. 90° peel-tests were performed in order to correlate the water activity with the resulting peel force. The corrosion products that formed across the delamination front were analyzed by Raman microscopy. Through these HR-SKP-BT studies, a critical value was found for relative humidity for the delamination process. A transition zone was detected in which electrochemical degradation precedes mechanical delamination. In addition to the experimental studies, the critical energy release rates of the blister were calculated in finite element (FE2) simulations so as to enable a better understanding of the delamination of adhesives on metal surfaces. The combined experimental and theoretical studies show that the delamination process is controlled by the interfacial electrochemical reactions at the delamination front and that a transition area of few hundred micrometers exists in which the adhesion strength is lowered by the cathodic oxygen reduction process to a value which can be overcome by the mechanical stress in this area.

1. Introduction The durability of structural-adhesive-bonded engineering metals is mainly determined by the ingression of water, oxygen and corrosive ions into the polymer/metal interface [1]. Water ingression leads to wet de-adhesion and, in combination with corrosive ions, to a corrosive attack in the interface region. Early models of wet adhesion and deadhesion were developed by Brockmann and others [2]. The access of water to the interface and interfacial hydrolytic reactions play a major role in both wet adhesion and corrosion processes [3]. However, the interplay between the chemical and corrosive delamination mechanisms of adhesives is still not fully understood. On the one hand, studies have not taken the mechanics of the composites into account and, on the other hand, the mechanical aspects of the delamination of composites and adhesive interfaces have been very much highlighted in recent years [4].

⁎

Corresponding author. E-mail address: [email protected] (G. Grundmeier). HR-SKP-BT: Height Regulated Scanning Kelvin Probe Blister Test. 2 FE: Finite Element. 1

http://dx.doi.org/10.1016/j.ijadhadh.2016.11.006 Accepted 5 October 2016 Available online 12 November 2016 0143-7496/ © 2016 Published by Elsevier Ltd.

The electrochemical fundamentals of the corrosive delamination of organic coatings have been intensively studied over the last twenty years [5–13]. The corresponding mechanisms for adhesives on metals, however, have only recently attracted significant interest [14,15]. While oxygen is always dissolved in adhesives in sufficiently large concentrations to sustain a cathodic undermining process, the ingress of water depends on the changes in the water activity in the environment [14]. Based on the understanding acquired of the electrochemical delamination processes, the focus has recently been placed on the development of new surface technologies for metals prior to organic paint application [16,17]. Galvanized steel and aluminum alloys, however, are adhesively bonded in the automotive industry without any chemical treatment, such as conversion chemistry, prior to the application of the adhesive. Only recently, have advanced techniques been developed for aluminum alloys, such as local laser treatments [18]. Here, the durability of the joint relies on the prevention of water

International Journal of Adhesion & Adhesives 73 (2017) 8–15

R. Grothe et al.

The concept covered in the work presented here is a combination of FE simulation and the experimental study of blister formation under controlled corrosive conditions and mechanical load. This approach involves the analysis of the blister shape and the progress of the interfacial delamination.

and ion ingression into the joint. This is often accomplished by protecting joint parts with ED3-paint in addition to the barrier properties of the adhesive itself [19–21]. The Height-Regulated Scanning Kelvin Probe Blister Test was developed by Grundmeier et al. especially for analyzing the corrosive delamination of adhesive films on metals [22,23]. This development is based on an analysis of interfacial electrode potentials using the Kelvin probe, as published by Stratmann et al. [14,15,22,24,25]. This electrochemical technique permits contact-free, nondestructive, spatially resolved analysis of the interfacial corrosion potential between polymer films and metals based on the evaluation of the contact potential between a vibrating scanning metallic needle and the polymer-covered metal electrode. The scanning Kelvin probe has thus become a wellknown and established technique for implementing nondestructive and, of course, in situ electrochemical measurements with controlled atmospheric conditions during the corrosive degradation of bare and coated metal substrates. The introduction of height regulation, based on the dependence of the AC current between the needle and the substrate on the needle-substrate surface distance, allowed for additional height regulation. This approach permits an analysis of the interfacial electrode potentials and the simultaneous characterization of the shape of a polymeric blister on a corroding metal substrate. Based on the combination of the Scanning Kelvin Probe Blister Test and in-situ FTIR-ATR4, Grundmeier et al. analyzed the delamination processes of adhesives in the presence and absence of an applied mechanical load [14,22]. As for organic coatings, free-standing adhesive films showed a cathodic delamination process, which was, however, accelerated by a simultaneously applied mechanical load [18,26]; the authors were able to correlate the delamination kinetics with the water activity in the environment and the applied mechanical load [14,15]. However, the authors did not quantitatively analyze the resulting mechanical stress in the delamination zone. The fracture energy was extracted from a pressurized blister test in 1983 by Hinkley, who performed an elastic analysis of a spherical cap. An analysis of the blister test, considering blister plasticity as well as the interaction of the blister and the substrate represented by a cohesive zone model, was suggested by Liecht et al. [27]. A closed form solution for the pressurized blister, describing the transition from the bending to the stretching limit case, was developed by Arjun and Wan and compared to earlier models and a numerical solution [28]. Xu et al. [29] proposed an analysis of the peninsula blister test with allowance for the effect of residual stresses. Both finite element analysis and closed form solutions were applied by Nie et al. [30] with blister tests that employed loading by a rigid central block. More recently, a plate model for pressurized blisters was proposed by Cao et al. [31], who also considered residual stresses. Furthermore, Cao et al. deviated from the simple clamped boundary conditions at the delamination front. They did not use a cohesive law like Liecht et al. but clamping with rotational compliance. A modified test using a constant amount of pressurized gas instead of a constant pressure reveals the advantage of stable crack growth and was covered by Boddeti et al. [32]. The FEA5 of Nie et al. used an axisymmetric model consisting of quadrilateral elements. Only the part of the blister located over the defect was modeled, while clamped boundary conditions were applied at the outer edge of the defect area. The load was applied by prescribing the vertical displacement of the nodes in the region of the central block. The energy release rate was calculated by comparing the elastic energy to a model with a slightly increased defect radius. This tallied well with the closed form solution as long as the central deflection was no larger than the blister thickness. Furthermore, the phase angle (mode-mix) was calculated from the stresses at the crack tip.

2. Experimental 2.1. Materials and chemicals 2.1.1. Substrates All the chemicals and solvents were of p.a. quality. Use was made of zinc substrates (type ZE, electrically galvanized) from Salzgitter, which were cleaned with organic solvents (THF6, IPA7, EtOH8), for 10 min each in an ultrasonic bath at RT. Iron sheets, covered with an approximately 7-µm thick electrogalvanized zinc layer, were obtained from Salzgitter AG, Salzgitter/ Germany. These plates were ultrasonically degreased with organic solvents, and then alkaline cleaned, before being subsequently rinsed with ultra-pure water and dried in a nitrogen stream. As the next step, a circular hole with a diameter of 1 mm was drilled in the rear side of the metal sample (4×4 cm) through to the adhesive layer. For removing the alloy residues, the use of concentrated and tempered (60 °C) HNO39 was considered. Employing this procedure, an undamaged adhesive film was obtained above the hole. These samples were used for the HR-SKP-BT investigations. 2.1.2. Preparation of adhesive films The substrates were coated with a model adhesive consisting of three components provided by Dow Chemical (Midland/ USA): Bisphenol-A-diglycidilether (DGEBA, DER 332) and diglycidyl ether of polypropylene glycol (DGEPG, DER 736), plus the amine component Jeffamine (D 230) from Huntsman Corporation (Woodloch Forest Drive/ USA). The polymer coating was hardened for 2 h at 130 °C. To form a uniform layer, two spacers (one layer of magic tape, two layers of Scotch tape) were placed at the end of the sample, and use was made of a polycarbonate block wrapped with aluminum foil with an applied pressure of 50 g/cm2 (sandwich technique). The thickness of the adhesive film (120 ± 30 µm) was confirmed by an electric sliding caliper. The samples were exposed to humid air ( > 90% r.h.) for 1 h at 40 °C before employing the HR-SKP-BT method in order to prevent any electrostatic charging of the polymer. 2.2. Analytical methods 2.2.1. Height regulated scanning Kelvin probe blister test To investigate the epoxy amine/zinc oxide interface stability, the HR-SKP-BT measurements were carried out using a custom-made SKP, which is described in [22]. The vibrating needle of the scanning Kelvin probe was made of CrNi and had a diameter of 50 μm. Based on the simultaneous analysis of the Volta potential difference between the needle and the adhesive surface, and the current flow between the substrate and the needle which form a capacitor, both the interfacial electrode potential and the topography of the blister can be studied [22]. The height regulation is based on the value of the alternating current of the capacitor measured at low frequency while the potential measurements is based on the high frequency analysis of the zero crossing of the current [22]. The blister hole was filled with 0.5 Mol NaCl solution and the sheets were fixed on the sample carrier. The area around the defect was polarized to −1.2 VSHE in a three-electrode arrangement (platinum counter electrode, an Ag/AgCl-reference electrode and a grounded 6

THF: tetrahydrofuran. IPA: isopropyl alcohol. 8 EtOH: ethanol. 9 HNO3: nitric acid.

3

7

ED: electrophoretic deposition. 4 FTIR-ATR: Fourier-Transform-Spectroscopy under attenuated total reflection. 5 FEA: Finite Element Analysis.

9

International Journal of Adhesion & Adhesives 73 (2017) 8–15

R. Grothe et al.

blister-test sample as a working electrode). A Potentiostat/ Galvanostat of type “PGU-10V-200mA-OEM“ (Elektronik Labor Peter Schrems, Münster, Germany) was employed. For pressurization, use was made of a Low ΔP-Flow pressure-meter/ controller (Bronkhorst E-7500-AAA) with a 2-valve-pressure meter (Bronkhorst-Mättig GmbH, Kamen, Germany). To set 70, 80 and > 90 ± 3.0% r.h. within the SKP chamber, a 3-way-mixer and a software controlled moisture-dry-air mixer – the HyGrow RH-Generator (GrowControl, Germany) – was used. To obtain a humid air atmosphere, synthetic air was passed through three water filled fritted gas wash bottles. Several parameters were varied during the measurement of the individual samples. The starting point was at 70% r.h. and 100 mbar pressure. Once no more essential changes were observed in the blister geometry, the pressure was increased up to 500 mbar in 100 mbar steps. The humidity was then increased up to > 90% in 10% steps. Stable conditions were reached within a very short equilibration time (5–10 min). In this work, delamination rates are mean values calculated from several different measurements; a basic cathodic polarization of around 100 mV vs. OCP10, always to the same fixed value of ≈−860 mVSHE11, was applied in order to obtain comparable electrochemical conditions for the corrosive processes within the defect area. Figs. 1 and 2. Based on this custom made set-up, the adhesive/oxide/metal interface potential can be measured with a spatial resolution of about 100 μm. For the simultaneous topographic information, the height resolution is in the range of 1–2 μm, while the x and y directions are the same as for the potential measurement.

Fig. 1. Set-up of the Height Regulated Scanning Kelvin Probe Blister Test with control of the electrode according to [22].

Fig. 2. Finite element model of the blister test (contours display strain in the horizontal direction).

commercial FE software were applied, namely the VCCT12 and the Jintegral. To integrate the VCCT into the model shown in Fig. 3, the clamped boundary condition was replaced by a debonding interaction with a rigid support. The critical energy release rate at the crack tip was evaluated for the opening (mode I) and the sliding (mode II) components separately, so that the mode ratio GII/GI was obtained. For the J-integral evaluation, a modified model was used which considered part of the substrate and applied the clamping to its bottom edge. A special mesh refinement was conducted at the crack tip in order to improve the calculation accuracy of the J-integral (Fig. 3, refinement up to 6 nm edge length). The J-integral only provides a value for the total critical energy release rate, without any information on the mode mix. The load transfer in the cathodic zone was not taken into account in the FEA. While a model using a cohesive zone approach like [27] is certainly able to give a more realistic representation of the experiment, the unique identification of the additional model parameters is difficult. Since we just want to show that the strain energy release rate is very much smaller than the value measured in other tests, the higher accuracy that a cohesive zone model can potentially provide is not necessary, and the simpler fracture mechanical approach employed here is sufficient.

2.2.2. Raman-Spectroscopy Raman spectra were recorded on the sample after blister tests with an "in-Via" disperse spectrometer (Renishaw, Gloucestershire, UK) excited by a RL633 nm HeNe laser with a power of 8.8 or 1.7 mW (only spectrum A), and a 50x objective was used. The recording time was set to 10 s. Use was made of the program provided for data recording “Wire 3.1” (Renishaw, Gloucestershire, UK). 2.2.3. 90° - peel tests Peel tests were carried out at a constant angle perpendicular to the sample surface and a constant velocity of 1 mm/s. The experimental details are described in [22]. Strips with a width of 5 mm were peeled off at different relative atmospheric humidity at room temperature. The peel-off forces were evaluated in the plateau region of the force distance curve. 2.2.4. Simulation tools The strain energy release rate G during defect growth in the blister test can be calculated using the closed form solution for the flexurally rigid, pressurized blister with simple clamped boundary conditions (see e.g. [30]): G=ph/2 with the pressure p and the height of the blister h. A finite element model of the test was set up to check the validity of this approximation and to gain further information on the mode mix at the crack tip. In the same way as for the FEA of Nie et al. [30], use was made of an axisymmetric model, which consisted solely of the adhesive layer, assuming a comparatively rigid substrate. Unlike the work of Nie et al., the clamped boundary condition was not applied to a vertical line at the end of the defect. Instead, a larger part of the adhesive was modelled and the boundary condition was applied at its edge to the substrate. A linear elastic material model was used for the adhesive. Its elastic constants and the size of the defect were obtained by inverse identification, comparing the simulated blister deformation to the measurements. The fracture energy was not calculated using two models with different-sized delamination like by Nie and coworkers; instead, two tools currently available in 10 11

3. Results and discussion 3.1. Mechanical studies of the blister at reduced relative humidity The mechanical properties of the adhesion film above the blister hole were analyzed at a relative humidity of 70%. Under such conditions, no corrosive delamination was observed within the measurement time, with variation of the hydrostatic pressure between 0 and 500 mbar. For each set pressure, the shape of the blister was recorded on the basis of SKP-based topographic analysis, as shown in Fig. 4. A constant shape and height of the blister was typically established within 60 min after the respective pressure had been set. Due to the given slight tilt of the sample surface which is unavoidable based on the need to press the sample against an O-ring, the base line

OCP: open circuit potential. SHE: standard hydrogen electrode potential.

12

10

VCCT: Virtual Crack Closure Technique.

International Journal of Adhesion & Adhesives 73 (2017) 8–15

R. Grothe et al.

Fig. 3. Modified model for J-integral evaluation with local refinement at the crack tip. Fig. 5. Plot of the blister height vs. the applied hydrostatic pressure at varying hydrostatic pressures for a relative humidity of 70%.

h=

pa 4 64 D

With the blister radius a=0.7 mm and the slope dh/dp of Fig. 5, the flexural rigidity of the blister can be estimated as D=0.015 N mm. This is in good agreement with the value D=0.016 N mm, which will be evaluated in Section 3.5 from the height profile of a blister test at 80% humidity and 500 mbar pressure using finite element analysis.

3.2. Progress of delamination at increased relative humidity For relative humidity values of 80 and > 90% a delamination process was observed as indicated by the symmetric broadening of the potential curves following the progress of the blister topography. The diameter of the blister and the interfacial potential profile increased with time. In Fig. 6, the three-dimensional profiles for the topography and the potential are shown for > 90% r.h. and an applied pressure of 500 mbar, as an example which proves the symmetric growth of the blister. In all the studies, the polarization of the defect by the potentiostat was perfectly reflected in the measured interface potentials by means of the SKP. On the basis of the symmetric delamination process, the crosssectional analysis allowed for a kinetic analysis of the corrosive delamination process, as shown in Figs. 7 and 8. Both the topographic profile and the progress of the drop in the transients of the interfacial potential indicate corrosive delamination that correlates linearly with the time. This means that the kinetics of the corrosive reaction at the front of the delamination determines the overall kinetics of delamination [14]. The transport of ions by migration, as reported by Stratmann et al. for lacquers without the simultaneous application of a mechanical load, is not rate-determining [15]. This can be explained by the proximity of the blister electrolyte and the delamination front in the blister experiment. Moreover, the sufficient bond strength of the adhesive to the oxide-covered substrate, even under wet conditions, limits the migration of ions along the interface. As already shown by Posner and Wapner et al., however, the potential front precedes the mechanical delamination front [14]. In the system reported here, the length by which the potential front precedes this latter front is in the range of few hundred micrometers. From the detailed kinetic evaluation shown in Fig. 8, it is obvious that the electrochemical delamination rate and the rate of topographic blister growth is accelerated by the increase in relative humidity. The increase in the electrochemical delamination rate, however, was observed to be more significant than the increase in the mechanical delamination rate.

Fig. 4. a, b. HR-SKP-BT: corrosive delamination study of an epoxy amine/ zinc oxide/ zinc interface at a) 100 mbar electrolyte pressurization and 70% r.h. and b) 500 mbar electrolyte pressurization and 70% r.h.

of the height versus distance cannot be perfectly zero. The plot of the blister height versus the applied pressure is shown in Fig. 5. The blister height analysis is based on the manual subtraction of the baseline. No delamination was observed at this water partial pressure, indicating that the adhesion of the adhesive to the zinc oxide-covered metal was still high enough to resist the peel force applied. An almost linear correlation was observed between height and applied pressure. Due to the preparation method, the height at zero pressure was already 11 µm, which creates the offset. This linear increase of blister height is predicted by plate theory which provides the following closed form solution [31]:

11

International Journal of Adhesion & Adhesives 73 (2017) 8–15

R. Grothe et al.

Fig. 6. a, b. 3-D Plots of the blister height a) and b) interface potential at 500 mbar and 7 days of delamination.

3.3. Peel force studies as a function of relative humidity Fig. 7. a and 7b. Cross-sectional analysis of interface potential and blister topography for a) 80% r.h. and b) > 90% r.h. at an applied pressure of 500 mbar as a function of time.

Complementary peel force studies were performed in order to prove the dependence of adhesion on the bond between the adhesive and the oxide-covered substrate. As the defect does not corrode in this case, no cathodic delamination can occur at the interface. As expected, the peel force at an angle of 90° fell with an increased relative humidity, as shown in Fig. 9. This reduction in adhesion is due to the equilibrium between the chemical potential of water in the surrounding atmosphere and in the adhesive. The increasing water activity in the adhesive leads to an accumulation of water at the adhesive/oxide interface, which lowers the macroscopically measured peel force. The strong reduction in peel force which results when the relative humidity is increased from 80 to > 90% can be explained by the multilayer formation of water adsorbates at the interface. As shown by in-situ FTIR-ATR spectroscopy, an increase in the relative humidity leads to a corresponding increase in the interphasial water absorbance [14]. The HR-SKP-BT results and the peel forces both show that the interface stability decreases with an increasing interfacial water activity. In the case of the peel test and the HR-SKP-BT study, an equilibrium state of water becomes established in the gas phase, the adhesive and at the adhesive/ oxide interface, due to the extended equilibration times (see Eq. (1)) μgas(H2O) = μadhesive(H2O) = μinterface(H2O)

Fig. 8. Illustration of the progress of delamination based on an evaluation of the reversal points of the topographic profiles for the right side of the blister at 80 and > 90% r.h. and 500 mbar mechanical load.

(1) spectra are summarized in Table 1. Microscopically, the delaminated zone shows two different areas. The inner area (red area) is rather strongly corroded, while a ring of about 200–300 µm (blue area) is formed as a transition between the inner blister volume and the nondelaminated area (green area). The superposition of the microscopic image on the final topographic and potential profile in Fig. 10 indicates that the transition area in which the metal surface has not yet been strongly attacked can be assigned to a state in which the adhesive has undergone partial

where μ is the chemical potential. The interfacial activity of water is directly correlated to the water partial pressure in the gas phase above the adhesive.

3.4. Microscopic and spectroscopic analysis of the delaminated area The Raman spectroscopic evaluation of the delaminated area is shown in Fig. 10 and Fig. 11, and related peak assignments of the 12

International Journal of Adhesion & Adhesives 73 (2017) 8–15

R. Grothe et al.

Table 1 Assignment of Raman peaks in the delaminated zone. Observed peak position [cm−1]

Assignment

Peak position in the literature [cm−1]

Reference

224

223/ 224

[33,34]

240.5

[35]

257

[36]

293/ 291 392.2/ 397

[33,34] [35,36]

406 498 543

[34,35] [33] [35]

564

Fe-oxyhydroxides (lepidocrocite) Simonkolleite [4Zn (OH)2*ZnCl2*H2O] Simonkolleite [4Zn (OH)2*ZnCl2*H2O] iron oxides Simonkolleite [4Zn (OH)2*ZnCl2*H2O] Iron oxides Hematite Non-stoichiometric oxide Zn1+xO Native ZnO Layer

[37]

609

Hematite

Broad band 300– 600, max. located around 560 611

243 255 290 394 408 494 547

Fig. 9. Peel-off forces for the adhesive on galvanized steel substrates measured at ambient temperature for different relative humidity values of 70, 80 and > 90%.

[33]

chloride corrosion product, simonkolleite, and non-stoichiometric zinc oxides. For the inner blister volume, two areas can be distinguished. Point C shows simonkolleite while point D shows additional corrosion produced by iron. The Raman data and the optical image visualize a transition zone of a few hundred micrometers between the intact and the delaminated area where almost full access for the sodium chloride solution leads to the formation of hydroxychlorides, even at an applied potential that is more negative than the free corrosion potential of zinc. In the transition zone, ingress of the electrolyte is still hindered by the small gap between the adhesive and the substrate. In the center of the blister, the zinc coating is already dissolved and iron oxides are formed.

Fig. 10. Superposition of the microscopic image of the surface after removal of the adhesive film and the corresponding final potential distribution as measured by the HRSKP-BT. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

3.5. FEM13 simulation of interfacial delamination The energy release rate during defect growth in the blister test performed at 500 mbar and 80% r.h. is calculated using the closed form solution for the flexurally rigid, pressurized blister with simple clamped boundary conditions (see e.g. Cao et al. 2014, [29]): GC=ph/2≈ (500 hPa 24 µm)/2≈ 0.6 Jm−2. The central deflection of the blister h is about five times smaller than its thickness t. According to Nie et al. [30], the closed form solution should be a good approximation. This was validated by means of finite element analysis carried out using the finite element software Abaqus. First, the elastic constants and the defect size were identified in order to achieve an optimum fit of the simulated deflection profile and the test result (Fig. 12). According to the approximate closed form solution, the deformation is only influenced by the elastic modulus E and Poisson's ratio v in the form of the flexural rigidity: D=Et2/(12(1-v2). In agreement with this, the FEA yields the same solution for different pairs of elastic parameters exhibiting the same flexural rigidity (D=0.016 N mm). Next, the energy release rate was obtained using the VCCT and Jintegral. To ensure mesh-independent results, a discretization study with a minimum element size of 1 µm was performed. The different methods yielded a fracture energy of 0.6 ± 0.05 Jm−2, in good agreement with the closed form solution. Additionally, the VCCT made it possible to estimate the mode ratio GII/GI=1.2. This means that the loading of the crack tip is neither mode I nor mode II but a mixed mode state. In tapered double cantilever beam tests (ISO 25217), the critical energy release rate of the same adhesive in a layer of 0.6 mm thickness

Fig. 11. Raman spectra measured in the indicated areas (see Fig. 10). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

electrochemical delamination, based on the oxygen reduction reaction. This zone, however, still adheres to the substrate strongly enough to withstand the applied peel force. In the transition zone, the potential is significantly more positive than in the center of the blister, where the anodic corrosion activity is higher. The inner area (inner volume) is rather strongly corroded, while a ring of about 200–300 µm (transition) is formed as a transition zone between the inner blister volume and the non-delaminated area (intact area). Raman spectra were recorded at five different points. Point A is located on the non-delaminated area and shows no detectable Raman signal. By contrast, Raman spectra of the transition zone with points B and B* both show several peaks at 255 cm−1, 394 cm−1, 733 cm−1, 912 cm−1 and 547 cm−1 indicating the presence of the zinc hydroxide/

13

13

FEM: Finite Element Model.

International Journal of Adhesion & Adhesives 73 (2017) 8–15

R. Grothe et al.

was observed to be in the range of few hundred micrometers for the studied system. A critical relative humidity of between 70 and 80% was revealed which was necessary to start a corrosive delamination. This value is assigned to the formation of an interfacial water activity which is sufficient to allow for an interfacial electrode reaction and the incorporation of corrosive ions. When the interaction forces have been sufficiently lowered by the corrosion process, the applied force is sufficient to mechanically peel off the film. The simulation data prove that the delamination is driven not by the mechanical force but by the interfacial electrochemical reaction. However, an increase in the applied pressure of the blister leads to an accelerated delamination process, since higher peel forces can be overcome by a higher applied force. Acknowledgement Fig. 12. Measured and simulated blister deflection, simulated defect radius 0.78 mm.

The authors acknowledge the financial support of the German Science Foundation (DFG) and the AiF within the “BestKleb” joint research cluster. The authors of the University of Paderborn gratefully acknowledge the financial support from the German Science Foundation (DFG project grant No. GR 1709/14-1). The authors of the IFAM in Bremen gratefully acknowledge the financial support from the AiF (IGF project grants No. 17276 N). IGF project 17276 N "Fracture behaviour of adhesive joints and cohesive-zone-model – influences of manufacturing and ageing" run by research association Forschungsvereinigung Stahlanwendung e.V. (FOSTA), Sohnstraße 65, D-40237 Düsseldorf was funded by the AiF under the programme for the promotion of joint industrial research and development (IGF) by the Federal Ministry of Economics and Energy on the basis of a decision of the German Bundestag.

was measured in mode I loading. Tests were performed on both fresh and aged samples. A hot/wet (60 °C/90% r.h.) environment and immersion in water at 60 °C were used to age samples for durations of up to 14 weeks. The results of tests performed at room temperature and at 40 °C showed critical energy release rates in a range of 100 to 2000 Jm−2. Thus, the energy release rate in the blister test is smaller by a factor of approximately 1000 than the energy release rate in the fracture mechanical test. This large difference is observed regardless of the variation of TDCB test conditions: test on samples stored in laboratory climate showed the difference to the energy release rate of the blister test as well as samples immersed in hot water, and neither TDCB tests performed at 23 °C nor at 40 °C yielded values close to the blister test results. Therefore, the discrepancy between the results of the two test methods cannot be attributed to differences in humidity of the adhesive. This difference between the energy release rates in the two tests is due in part to the difference in adhesive thickness, which is five times larger in the TDCB14 test. Another possible reason for discrepancies is the different mode-mix of the tests. The critical energy release rate in mixed mode is usually higher than in pure mode I, however, so that would explain a higher but not a lower energy release rate in the blister test. Since these differences in the test conditions do not explain a difference of a factor of one thousand in the energy release rate, we can conclude that the energy release rate is not the main driving force behind crack growth in the evaluated blister tests. At the low level of pressure applied in the electrochemical blister test, the crack growth is driven by the kinetics of the electrochemical reactions and the correlated transport of ions at the adhesive/metal interface.

References [1] Kinloch AJ. Adhesion and adhesives: science and technologyDordrecht, s.l. Netherlands: Springer; 1987. [2] Brockmann W, Hennemann O, Kollek H. Failure mechanisms in the boundary layer zone of metal/polymer systems. In: Mittal KL, editor. Adhesive Joints. Boston, MA. Springer US; 1984. p. 469–84. [3] da Silva LFM, Sato C, editors. Design of adhesive joints under humid conditions. Berlin, Heidelberg: Springer Berlin Heidelberg; 2013. [4] Pascoe JA, Alderliesten RC, Benedictus R. Methods for the prediction of fatigue delamination growth in composites and adhesive bonds. A critical review. [5] Williams G, McMurray HN Chromate inhibition of corrosion-driven organic coating delamination studied using a scanning Kelvin probe technique. [6] Edavan RP, Kopinski R. Corrosion resistance of painted zinc alloy coated steels. Corros Sci 2009;51(10):2429–42. [7] Fürbeth W, Stratmann M. The delamination of polymeric coatings from electrogalvanised steel – a mechanistic approach. Corros Sci 2001;43(2):207–27. [8] Fürbeth W, Stratmann M. The delamination of polymeric coatings from electrogalvanized steel – a mechanistic approach. Corros Sci 2001;43(2):243–54. [9] Fürbeth W, Stratmann M. The delamination of polymeric coatings from electrogalvanized steel – a mechanistic approach. Corros Sci 2001;43(2):229–41. [10] Leng A, Streckel H, Hofmann K, Stratmann M. The delamination of polymeric coatings from steel Part 3: effect of the oxygen partial pressure on the delamination reaction and current distribution at the metal/polymer interface. Corros Sci 1998;41(3):599–620. [11] Leng A, Streckel H, Stratmann M. The delamination of polymeric coatings from steel. Part 1: calibration of the Kelvinprobe and basic delamination mechanism. Corros Sci 1998;41(3):547–78. [12] Leng A, Streckel H, Stratmann M. The delamination of polymeric coatings from steel. Part 2: first stage of delamination, effect of type and concentration of cations on delamination, chemical analysis of the interface. Corros Sci 1998;41(3):579–97. [13] Prosek T Evaluation of the tendency of coil-coated materials to blistering: field exposure, accelerated tests and electrochemical measurements. [14] Posner R, Giza G, Vlasak R, Grundmeier G. In situ electrochemical Scanning Kelvin Probe Blister-Test studies of the de-adhesion kinetics at polymer/zinc oxide/zinc interfaces. Electrochim Acta 2009;54(21):4837–43. [15] Wapner K, Stratmann M, Grundmeier G. In situ infrared spectroscopic and scanning Kelvin probe measurements of water and ion transport at polymer/metal interfaces. Electrochim Acta 2006;51(16):3303–15. [16] Ferreira M, Duarte R, Montemor M, Simões A. Silanes and rare earth salts as chromate replacers for pre-treatments on galvanised steel. Electrochim Acta 2004;49(17-18):2927–35. [17] Critchlow GW, Yendall KA, Bahrani D, Quinn A, Andrews F. Strategies for the

4. Conclusions The results show that the combination of the experimental HRSKP-BT analysis and FEM simulation leads to a detailed insight into the corrosive delamination process of adhesive films on zinc-coated steel. The SKP-BT studies as performed under controlled hydrostatic pressure, interfacial electrode potentials and water activity in the adhesive film and showed that the oxygen reduction at the front of the delamination leads to a transition area in which corrosive delamination takes place. This transition area shows a limited ingress of chlorides and thereby the corrosive delamination in this area is dominated by the cathodic reaction. As proven by the Raman spectroscopic data, this process is followed by an anodic corrosion of the zinc coating. In the center of the blister, the anodic reaction completely predominates, leading to a more negative electrode potential and strong dissolution of the zinc coating. The width of the transition area 14

TDCB: Tapered Double-Cantilever Beam.

14

International Journal of Adhesion & Adhesives 73 (2017) 8–15

R. Grothe et al.

[18]

[19]

[20] [21]

[22]

[23]

[24]

[25]

[26]

[27] Liecht KM, Shirani A, Dillingham RG, Boerio FJ, Weaver SM. Cohesive zone models of polyimide/aluminum interphases. J Adhes 2000;73(2-3):259–97. [28] Arjun A, Wan K-T. Derivation of the strain energy release rate G from the first principles for the pressurized blister test. Int J Adhes Adhes 2005;25(1):13–8. [29] Xu D, Liechti KM, Lumley-Woodyear THde. Closed form nonlinear analysis of the peninsula blister test. J Adhes 2006;82(8):831–66. [30] Nie P, Shen Y, Chen Q, Cai X. Effects of residual stresses on interfacial adhesion measurement. Mech Mater 2009;41(5):545–52. [31] Cao Z, Wang P, Gao W, Tao L, Suk JW, Ruoff RS, Akinwande D, Huang R, Liechti KM. A blister test for interfacial adhesion of large-scale transferred graphene. Carbon 2014;69:390–400. [32] Boddeti NG, Koenig SP, Long R, Xiao J, Bunch JS, Dunn ML. Mechanics of adhered, pressurized graphene blisters. J Appl Mech 2013;80(4):040909. [33] Larroumet D, Greenfield D, Akid R, Yarwood J. Raman spectroscopic studies of the corrosion of model iron electrodes in sodium chloride solution. J. Raman Spectrosc. 2007;38(12):1577–85. [34] Bouchard M, Smith DC. Catalogue of 45 reference Raman spectra of minerals concerning research in art history or archaeology, especially on corroded metals and coloured glass. Spectrochim Acta Part A: Mol Biomol Spectrosc 2003;59(10):2247–66. [35] Hammouda N, Chadli H, Guillemot G, Belmokre K. The corrosion protection behaviour of zinc rich epoxy paint in 3% NaCl solution. ACES 2011;01(02):51–60. [36] Le Hugot Goff A, Bernard MC, Phillips N, Takenouti H. contributions of raman spectroscopy and electrochemical impedance to the understanding of the underpaint corrosion process of zinc-coated steel sheets. MSF 1995;192-194:779–88. [37] Bernard MC. In situ raman study of the corrosion of zinc-coated steel in the presence of chloride. J Electrochem Soc 1995;142(7):2167.

replacement of chromic acid anodising for the structural bonding of aluminium alloys. Int J Adhes Adhes 2006;26(6):419–53. Klimow G, Fink N, Grundmeier G. Electrochemical studies of the inhibition of the cathodic delamination of organically coated galvanised steel by thin conversion films. Electrochim Acta 2007;53(3):1290–9. Maege I, Jaehne E, Henke A, Adler HP, Bram C, Jung C, et al. Self-assembling adhesion promoters for corrosion resistant metal polymer interfaces. Progress Org Coat 1998;34(1-4):1–12. Jaehne E, Oberoi S, Adler HP. Ultra thin layers as new concepts for corrosion inhibition and adhesion promotion. Progress Org Coat 2008;61(2-4):211–23. Wapner K, Stratmann M, Grundmeier G. Structure and stability of adhesion promoting aminopropyl phosphonate layers at polymer/aluminium oxide interfaces. Int J Adhes Adhes 2008;28(1-2):59–70. Wapner K, Schoenberger B, Stratmann M, Grundmeier G. Height-Regulating scanning Kelvin probe for simultaneous measurement of surface topology and electrode potentials at buried polymer/metal interfaces. J Electrochem Soc 2005;152(3):E114. Wapner K, Grundmeier G. Scanning Kelvin probe measurements of the stability of adhesive/metal interfaces in corrosive environments. Adv Eng Mater 2004;6(3):163–7. Stratmann M, Leng A, Fürbeth W, Streckel H, Gehmecker H, Groβe-Brinkhaus K. The scanning Kelvin probe; a new technique for the in situ analysis of the delamination of organic coatings. Progress Org Coat 1996;27(1-4):261–7. Fürbeth W, Stratmann M. Scanning Kelvinprobe investigations on the delamination of polymeric coatings from metallic surfaces. Progress Org Coat 2000;39(1):23–9. Leidheiser H, Wang W, Igetoft L. The mechanism for the cathodic delamination of organic coatings from a metal surface. Progress Org Coat 1983;11(1):19–40.

15