Adhesion aspects of polymeric coatings. Vol. 2

Adhesion Aspects of Polymeric Coatings, Volume 2 K.L. Mittal, Editor VSP Adhesion Aspects of Polymeric Coatings, Vol...

Author: K. L. Mittal

61 downloads 851 Views 14MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!

Report copyright / DMCA form

DOWNLOAD PDF

Adhesion Aspects of Polymeric Coatings, Volume 2

K.L. Mittal, Editor

VSP

Adhesion Aspects of Polymeric Coatings, Volume 2

This page intentionally left blank

ADHESION ASPECTS OF POLYMERIC COATINGS VOLUME 2

Editor: K.L. Mittal

UTRECHT BOSTON, 2003

VSP BV P.O. Box 346 3700 AH Zeist The Netherlands

Tel: +31 30 692 5790 Fax: +31 30 693 2081 [email protected] www.vsppub.com

© VSP BV 2003 First published in 2003 ISBN 90-6764-377-7

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without the prior permission of the copyright owner.

Printed in The Netherlands by Ridderprint bv, Ridderkerk

Contents

Preface Interphase: Formation, characterization and relevance to practical adhesion A.A. Roche, J. Bouchet and S. Bentadjine

vii

1

Depletion, a key factor in polymer adhesion G. Frens

21

Attaining adhesion/cohesion within painted plastics R.A. Ryntz

29

Scanning electric potential microscopy (SEPM) and electric force microscopy (EFM) imaging of polymer surfaces E.F. de Souza, M.M. Rippel, A.J. Keslarek, A. Galembeck, C.A.R. Costa, É.T. Neto and F. Galembeck

45

The residue (smut) formed on aluminum alloys during hydrofluoric acid etching and its effect on a coating process A.P.S. Tihaiya, J.P. Bell and G.D. Davis

65

Surface modification of metals by silanes D.Zhu and W.J. van Ooij

81

Application of X-ray photoelectron spectroscopy in assessing the adsorption of siloxane polymers onto E-glass fibers L.G. Britcher, D. Kehoe and J.G. Matisons

101

Surface modification of polyphenylene sulfide plastics to improve their adhesion to a dielectric adhesive Y. Wang and S. Rak

121

Metal surface conditioning concepts for resin bonding in dentistry P. Pfeiffer, I. Nergiz and M. Özcan

137

vi

Contents

Measurement of internal stresses in polymeric coatings using time resolved fluorescence T. Ikawa and T. Shiga

151

Adhesion of an alkyd paint to cold rolled steel sheets: Effect of steel surface composition S. Maeda

165

Analysis of the wet adhesion of coatings on wood M. de Meijer

183

Modified tape test: Measurement of adhesion of insulator films to low dielectric constant organic polymers L.L.N. Goh, S.L. Toh, S.Y.M. Chooi, Y. Xu and T.E. Tay

203

Adhesion Aspects of Polymeric Coatings, Vol. 2, pp. vii–viii Ed. K.L. Mittal © VSP 2003

Preface This volume documents the proceedings of the Second International Symposium on Adhesion Aspects of Polymeric Coatings held under the auspices of MST Conferences in Newark, New Jersey, May 25–26, 2000. The premier symposium on this topic was held under the aegis of the Electrochemical Society in Minneapolis, Minnesota in 1981, the proceedings of which were properly documented in a hard-bound book [1]. As almost 20 years had passed since the first symposium was held so we decided to organize the second event on this topic. In the interim, there had been a great deal of research activity relative to the adhesion aspects of polymeric coatings, so this symposium was both timely (rather overdue) and needed. Polymeric coatings are used for a variety of purposes, e.g., decorative, protective, functional (as dielectrics or insulators) and a special application of polymeric coatings is their use as lithographic materials for making integrated circuit elements. Irrespective of the intended purpose of the coating, it must adequately adhere to the underlying substrate, otherwise delamination and other undesirable phenomena can occur. So the need to understand the factors which influence adhesion of polymeric coatings and to control it to a desirable level is quite patent. In the last 20 years there have been new theoretical developments and advancements in instrumentation which have helped immensely in the arena of polymeric coatings. The acid-base theory of adhesion has found particular application in controlling the adhesion behavior of coatings. The technical program for this symposium consisted of 23 papers covering many subtopics dealing with adhesion aspects of polymeric coatings. There were lively and illuminating – not exothermic – discussions, both formally and informally, throughout the symposium. The presenters hailed from many corners of the globe and represented varied disciplines and research interests. Now coming to this volume (called Volume 2) it contains a total of 13 papers (others are not included for a variety of reasons) addressing many different issues. It must be recorded that all manuscripts were rigorously peer reviewed and suitably revised (some twice or thrice) before inclusion in this volume. So this book is not a mere collection of unreviewed papers – which is commonly the case with many symposia proceedings – rather it represents information which has passed peer scrutiny. Also it should be pointed out that, for a combination of reasons, the publication of this book got delayed but the authors were asked and given the opportunity to update their manuscripts. So the information contained in this book should be current and fresh.

viii

Preface

The topics covered in this volume include: factors influencing adhesion of polymeric coatings; ways to improve adhesion; formation and relevance of interphase in practical adhesion; adhesion/cohesion in painted plastics; imaging of polymer surfaces; effect of substrate residue (smut) on coating process; surface treatment of metals and glass by silanes; surface modification of polyphenylene sulfide plastics; resin bonding in dentistry; measurement of internal stresses in polymeric coatings; effect of steel surface composition on adhesion of paint; wet adhesion of coatings on wood; and modified tape test to measure adhesion of coatings. Yours truly sincerely hopes that this book will be of interest to everyone interested or involved in the arena of polymeric coatings. Also it should provide some new ideas as to how to control adhesion durability of coatings in different environments. Acknowledgements First, my sincere thanks are extended to my colleague and friend, Dr. Robert H. Lacombe, for taking care of the organizational aspects of this symposium. The comments from the peers are a sine qua non to maintain the highest standard of a publication, so I am most appreciative of the time and efforts of the unsung heroes (reviewers) in providing many valuable comments. I am deeply thankful to the authors for their interest, enthusiasm, patience and contribution without which this book would not have seen the light of day. In closing, my thanks go to the staff of VSP (publisher) for giving this book a body form. K.L. Mittal P.O. Box 1280 Hopewell Jct., NY 12533 1. K.L. Mittal (Ed.), Adhesion Aspects of Polymeric Coatings, Plenum Press, New York (1983).

Adhesion Aspects of Polymeric Coatings, Vol. 2, pp. 1–20 Ed. K.L. Mittal © VSP 2003

Interphase: Formation, characterization and relevance to practical adhesion A.A. ROCHE,∗ J. BOUCHET and S. BENTADJINE INSA de Lyon, Ingénierie des Matériaux Polymères / Laboratoire des Matériaux Macromoléculaires (CNRS, UMR 5627), 20 Ave. Albert Einstein, F-69621 Villeurbanne Cedex, France

Abstract—Epoxy-diamine networks are extensively used as adhesives or paints in many industrial applications. When the precursors are applied onto metallic substrates and cured, an interphase, having chemical, physical and mechanical properties quite different from that of bulk polymer, is created between the substrate and the polymer. Moreover, chemical reactions between diamine and metallic surfaces induce an increase in the practical adhesion (adherence). When the same epoxydiamine mixtures are applied onto gold coated or polyethylene substrates, the interphase properties are the same as bulk ones. When epoxy-diamine mixtures are applied onto aluminum or titanium alloy surfaces, the glass transition temperature, amine and epoxy reaction extent, the interphase thickness, residual stresses within the interphase, Young’s modulus of the interphase all depend on the amine nature (aromatic, aliphatic or cycloaliphatic), the stoichiometric ratio, the processing conditions (time and temperature), the organic layer thickness and the metallic surface treatment. Coating analyses (FTIR, FTNIR, DSC, DMTA, H+ and C13 NMR, SEC, ICP and POM) suggest that diamine monomers chemically react with and dissolve in the metallic hydrated oxide layer. Then, metallic ions diffuse through the organic layer to form a complex by coordination with diamine monomers (chelate or ligand). Metal-diamine complexes are insoluble, at room temperature, both in diamine as well as in DGEBA monomers and they induce a phase separation during the curing cycle of the epoxy-diamine precursors. Furthermore, the chemical bonding of diamine monomers to the metallic surfaces and the diamine-metal crystal orientation parallel to the metallic surface within the interphase lead to chemical, physical and mechanical properties to the epoxy-diamine network which are different from those of the bulk. Keywords: Epoxy-diamine networks; practical adhesion; interphase; diamine-metal chemical reactions.

1. INTRODUCTION

Epoxy-diamine mixtures are extensively used as adhesives or paints in many industrial applications. When they are applied onto metallic substrates and cured, epoxy-amine liquid monomers react with the metallic oxide and/or hydroxide to form chemical bonds [1, 2] increasing practical adhesion (or adherence) between ∗

To whom all correspondence should be addressed. Phone: (33) 4 72 43 82 78, Fax: (33) 4 72 43 85 27, E-mail: [email protected]

2

A.A. Roche et al.

the epoxy polymer and the substrate surface [3, 4]. Different studies concern the influence of the metallic substrate nature on the prepolymer cross-linking. Some authors have shown a catalytic effect of the substrate on the cross-linking mechanisms and the prepolymer adsorption onto metallic surfaces [5]. Others have attempted to determine the existence of monomer/substrate specific reactions. As an example, Dillingham and Boerio [6] have studied the polymerization of a diglycidyl ether of bisphenol A (DGEBA) and a triethylene tetramine (DGEBATETA) system applied onto aluminum by using both FTIR and XPS. Close to the polymer/metal interface, the hardener is partially protonated by aluminum hydroxides. Moreover, the polymerization is catalyzed by the presence of hydroxide acid groups leading to an interphase formation. Nigro and Ishida [7] have studied, by using FTIR, an epoxy/BF3-monoethylamine system applied onto polished steel. The epoxy conversion rate is more important in the polymer/metal interfacial region suggesting the existence of chemical reactions with the steel surface. The epoxy prepolymer was applied onto the steel surface (without hardener), and the same phenomenon was still observed suggesting that the chemical species formed at the steel surface were able to catalyze the homopolymerization of the epoxy monomer. Gaillard and coworkers [8-9] have shown that the DGEBADDA polymerization rate was faster on the zinc surface of galvanized or electrogalvanized steel than on polished steel. They concluded that this was due to a catalytic effect of zinc metallic ions leading to a higher extent of cross-linking. Some researchers have studied the adsorption of various monomers, used as adhesives, at the substrate surface [10-12]. Kollek [13] studied, using FTIR, the polymerization of a diglycidyl ether of bisphenol A and a dicyandiamide (DGEBADDA) system applied onto aluminum and observed that both the DDA and the DGEBA monomers were adsorbed onto the aluminum oxide surface. For the DDA monomer, the adsorption was due to the acidic proton of the aluminum oxide. For the epoxy monomer, the adsorption was achieved by the oxirane (or epoxy) ring opening. Unfortunately, only a few papers [14-17] have dealt with molecular structures formed within the interphase region. Moreover, when epoxy resins are applied onto metallic substrates and cured, intrinsic and thermal residual stresses develop within the entire organic layer [18]. Intrinsic stresses are produced as a result of the mismatch between the active sites of the metallic substrate and the organic network and/or the formation of the polymer network. Thermal stresses are mostly developed during cooling [19] and are the result of thermal expansion mismatch between the metallic substrate and the polymer or cure-induced shrinkage of the organic layer [20]. Whatever their source, these residual stresses reduce the practical adhesion and may induce cracks in coating materials [21-23] resulting in a drop of the overall performance of adhesives or paints. To gain a better understanding of epoxy/metal adhesion requires a full knowledge of chemical and physical reactions which take place in the epoxy/metal interphase [24, 25]. Thus, the polymer/substrate interphase is a complex region containing gradients of residual stresses, as well as allowing structural rearrangement, intermolecular and inter-atomic interactions and diffu-

Interphase: Formation, characterization and relevance to practical adhesion

3

sion phenomenon [24]. When the adhesion of epoxy/metal systems failed, it was possible not only to correlate the residual stresses at the interphase/metal interface to practical adhesion but also to correlate the theoretical adhesion and durability to the presence or not of some chemical species [26]. However, in such systems, chemical and physical mechanisms leading to the interphase formation were not well understood. The aim of this paper was to develop an understanding of this interphase formation and to characterize it using model epoxy-amine systems. 2. EXPERIMENTAL

2.1. Materials 2.1.1. Substrates The metallic substrates used were commercial 0.516 ± 0.005 mm thick 5754 aluminum alloy from Péchiney and 0.600 ± 0.005 mm thick Ti6Al4V titanium alloy from Aérospatiale. Titanium and aluminum sheets were made into squares of dimensions 100 x 100 mm2. Before any monomer application, aluminum and titanium substrate surfaces were ultrasonically degreased in acetone for 10 min. and wiped dry. However, some titanium and aluminum panels were also degreased and chemically treated. Titanium sample were immersed in a solution of 10 g/l of ammonium difluoride for 2 min. at room temperature, rinsed in running tap water for 1 min., immersed in deionized water for 5 min. and wiped dry. Aluminum panels were degreased and submerged in a solution of 250 g/l of sulfuric acid, 50 g/l chromium acid and 44 g/l aluminum sulfate at 60°C for 20 min., rinsed and dried as titanium. After surface treatment, all substrates were kept in an airconditioned room (22 ± 2°C and 55 ± 5% R.H) for 2 h. Some aluminum sheets were coated with gold (≈ 100 nm) using a SCD005 Sputter Coater from Bal-Tec. 2.1.2. Monomers and polymers The bifunctional epoxy prepolymer used was a liquid diglycidyl ether of bisphenol A (DGEBA, M = 348 g/mole, DER 332) from Dow Chemical. The cycloaliphatic diamine curing agent used was isophorone-diamine (IPDA or 3aminomethyl-3,5,5-trimethylcyclohexylamine) from Fluka. An aromatic diamine, (4,4'-methylenebis(3-chloro-2,6-diethylaniline) or MCDEA) from Aldrich and an aliphatic diamine, (polyoxypropylene diamine or D400) from Texaco were also used. Epoxy prepolymer and curing agents were used without further purification. Assuming a functionality of 4 for the diamines and 2 for the epoxy monomer, a stoichiometric ratio (a/e = aminohydrogen/epoxy) equal to 1 was used throughout the work. Homogeneous mixtures of DGEBA and diamine were achieved by stirring under vacuum (≈ 1 Pa) at room temperature for 1 h (Rotavapor RE211 from Büchi, Switzerland) to avoid air bubble formation. The epoxy-amine adhesive cure cycle [27-29] was adapted to obtain both the maximum of the cure conversion, i.e. the highest glass transition temperature denoted Tg∞, and the thickest interphase formation. For IPDA, the curing cycle was: 3 h at 20°C, 20 → 60°C

A.A. Roche et al.

4

(2°C/min.), 2 h at 60°C, 60 → 140°C (2°C/min.), 1 h at 140°C, 140 → 190°C (2°C/min.), 6 h at 190°C, cooling (8 h) in the oven to 20°C; for MCDEA: 3 h at 20°C, 20 → 60°C (2°C/min.), 2 h at 60°C, 60 → 160°C (2°C/min.), 4 h at 160°C, 160 → 190°C (2°C/min.), 9 h at 190°C, cooling (8 h) in the oven to 20°C; and for D400: 3 h at 20°C, 20 → 60°C (2°C/min.), 2 h at 60°C, 60 → 100°C (2°C/min.), 3 h at 100°C, cooling (8 h) in the oven to 20°C. 2.1.3. Sample preparation Several layers of an adhesive tape (5413 from 3M), approximately 50 µm thick, were applied all around the periphery of 100 x 100 mm2 treated metallic sheets to obtain the desired liquid coating thickness. The epoxy-diamine mixtures were poured onto the metallic surfaces and spread with a cylindrical glass rod. For bulk materials, 10 x 10 x 100 mm3 bars were prepared using a silicone (RTV 501 from Rhône-Poulenc) or PTFE mould. Only the central parts of the bars were used for analysis. After curing and cooling down, the coating thicknesses (from 40 to 1500 µm) were determined using an EG-100 Digital Linear Gauge (from Ono Sokki Co, Japan) having a ± 2 µm sensitivity. 2.1.4. Monomer analysis To characterize the changes in monomers, both the liquid monomers DGEBA and IPDA were applied between two chemically treated metallic substrates (100 x 50 mm2) to form a 110-150 µm thick liquid film and kept at room temperature for 3 h in a desiccator under continuous nitrogen flow to prevent any monomer carbonization or oxidation. Then, the liquid monomers were scraped from the metallic surfaces (and will be called “modified monomers” in the following) with a PTFE spatula and stored in polyethylene vials under a nitrogen atmosphere. 2.2. Experimental techniques 2.2.1. Differential scanning calorimetry (DSC) DSC experiments were carried out in a Mettler DSC 30 apparatus to determine the glass transition temperature (Tg) of epoxy resins. Sealed aluminum pans containing 15-20 mg of resin were heated from –50°C to 250°C at a rate of 10°C/min under a continuous flow of U-grade argon. Samples were weighed using a Mettler balance having a ± 1 µg sensitivity. The calorimeter was calibrated with both indium and zinc. The glass transition temperature (onset) was determined with a ± 0.5°C sensitivity. To evaluate the variation in Tg with the coating thickness, the (Tg)eq value at thickness (i) was calculated using: (Tg ) eq =

hiTgi − hi−1Tgi−1 hi − hi−1

(1)

where Tgi corresponds to the Tg of a hi thick coating and Tgi–1 value corresponds to the glass transition temperature of a hi–1, thick coating.

Interphase: Formation, characterization and relevance to practical adhesion

5

2.2.2. Dynamic mechanical thermal analysis (DMTA) Dynamic viscoelastic measurements on coatings, after removal from the metallic substrate, were performed in a Rheometrics Solids Analyzer RSA II apparatus, using the tensile mode to determine the storage (E'), and loss (E") moduli as well as the ratio E"/E' = Tan δ as a function of temperature. Samples (4x40 mm2) were heated from –150°C to 250°C using a heating rate of 2°C/min by a forced convection oven using a nitrogen stream. The sample was deformed sinusoidally to a controlled strain amplitude of 0.05% at a fixed frequency of 1 Hz. Measurements were made at 30 s intervals. 2.2.3. Fourier transform infrared and near-infrared spectroscopy (FTIR, FTNIR) An infrared spectrometer (FTIR Magna-IR 550 from Nicolet) was used with Omnic FTIR software. An Ever-GloTM source was used along with a KBr beamsplitter and DTGS-KBr or MCT/A detector. The mid-infrared spectra were recorded in the 400-4000 cm–1 range and in the 4000-7000 cm–1 range for the nearinfrared. A transmission accessory was used for bulk or free standing film characterization. For the bulk material, KBr was mixed in a 100:1 ratio with the various epoxy resins which had been ground cryogenically. This mixture was then pressed under vacuum to obtain disks. For analysis, pure KBr disks were used as background. For each analysis, 64 or 96 scans were collected at 4 cm–1 resolution. The cure conversions for epoxy and amine groups were calculated by using, respectively, the 4530 cm–1 epoxy combination band and the 6500 cm–1 amine band [30]. The 4623 cm–1 aromatic C-H ring stretch combination band was considered as reference. Thus, the amine (Xa) and the epoxy (Xe) conversions were determined using the ratio of the respective peak areas (A) by: Xa =1 −

( A6500 / A4623 )t ( A6500 / A4623 )t=0

and

Xe =1 −

( A4530 / A4623 )t ( A4530 / A4623 )t=0

(2)

Once again, to evaluate the Xa and Xe variation versus the coating thickness we calculated the (Xa)eq and (Xe)eq values at thickness (i) using:

( X a,e )eq =

(

hi X a,e

)i − hi−1 ( X a,e )i−1 hi − hi−1

(3)

According to Bentadjine [31], it is possible to correlate the stoichiometric ratio (a/e) to the Xa/Xe ratio as shown in Fig. 1. Then, the variation of the equivalent stoichiometric ratio (a/e)eq versus the coating thickness can be determined. 2.2.4. Nuclear magnetic resonance (NMR) spectroscopy For proton and carbon Nuclear Magnetic Resonance (1H and 13C NMR) a Bruker AC 400 spectrometer was used with the sodium salt of deuterated trimethyl silyl propionic acid (TSPd4) as the internal standard. Analyses were carried out at 293 K using deuterated water in which pure IPDA and modified IPDA were com-

6

A.A. Roche et al.

Figure 1. Variation of the (Xa)eq/(Xe)eq ratio as a function of the stoichiometric ratio (a/e).

pletely soluble. By comparing pure IPDA and modified IPDA spectra it was possible to determine the differences in chemical shifts for the equivalent carbon nuclei. 2.2.5. Inductively coupled plasma (ICP) spectroscopy An ICP spectrometer (Modula by Spectro Analytical Instruments) was used with a 2.5 kW plasma generator at 27 MHz. It was fitted with various detectors (an UV (0.75 m, 2400 grooves mm–1; 160-480 nm) monochromator, UV (0.75 m, 3600 grooves mm–1) polychromator, and visible (0.75 m, 1200 grooves mm–1) polychromator). A cross-flow nebulizer was used to introduce the liquid sample. Distilled water was used for dilution. 2.2.6. Size exclusion chromatography (SEC) A Waters apparatus was used with double detection (UV Waters 484 Tunable Absorbance Detector at λ = 254 nm and Waters 410 differential Refractometer) and a Waters 510 HPLC pump. The elution solvent used was tetrahydrofuran (THF) and the separation was carried out on two styrene-divinyl benzene columns (Nucleogel 100-5 and 500-5 from Macherey-Nagel) with a flow rate of 1 ml/min. The average molar mass calibration curve was constructed from monodispersed polystyrene standards and used to correlate the retention time to the mass average molar mass (M) .

Interphase: Formation, characterization and relevance to practical adhesion

7

2.2.7. Polarized optical microscopy (POM), scanning electron microscopy (SEM) and electron microprobe analysis (EMPA) Drops of diamine or epoxy-diamine mixture were confined between two glass plates and mounted on a hot plate under the optical microscope (Laborlux 12POLS from Leica equipped with a hot plate FP82 from Mettler and a CCD-IRIS color video camera from Sony). Samples were heated from 30°C to 190°C at 10°C/min. A scanning electron microscope (Philips XL20) fitted with an electron microprobe analysis accessory (Edax-Econ4) was also used. Samples were neither coated with gold nor with carbon. The accelerating electron voltage was 5 kV, the electron spot diameter for microanalyses was about 200 nm and the tilt angle used was 15°. 2.2.8. Free corrosion potential and pH measurements Different aqueous solutions with isophorone diamine (IPDA) were prepared. The pH’s of the solutions were measured at room temperature (22°C) using a PHN 850 pH-meter from Tacussel equipped with a combined electrode (Pt-Ag/AgCl). The pH variation versus the diamine concentration in aqueous solution is shown in Fig. 2. The basic behavior of IPDA is clearly shown. For DGEBA/IPDA prepolymer with a/e = 1, the IPDA concentration is equal to 20 wt% which corresponds to a pH value of 12. The same basic behavior was obtained for aqueous D400 (pH = 12.1) and MCDEA (pH = 10.3) monomer solutions. Free corrosion potential (or open circuit potential) measurements were carried out by using a Tacussel Potentiostat (PRT 40-1X) with a combined Pt electrode (Ag/AgCl).

Figure 2. pH variation of aqueous solution as a function of IPDA concentration.

8

A.A. Roche et al.

3. RESULTS AND DISCUSSION

3.1. Bulk and coating characteristics According to the curing cycle mentioned above, the maximum glass transition temperatures (Tg∞) of DGEBA/IPDA, DGEBA/MCDEA and DGEBA/D400 bulk epoxies were 163°C, 188°C and 34°C respectively. For these bulk materials, the amine conversion (Xa) and the epoxy conversion (Xe) were equal to 1. These values are in good agreement with the previous works [27-29]. Variations in the glass transition temperatures (Tg)eq versus DGEBA/IPD coating thickness applied onto both degreased titanium and aluminum are shown in Fig. 3. Variations in the equivalent stoichiometric ratio ((a/e)eq) versus the thickness of DGEBA/IPDA coatings applied onto degreased titanium and aluminum substrates are shown in Fig. 4. Regardless of coating thickness, for the coatings applied onto a gold coated substrate, the (Tg)eq’s were equal to the (Tg∞) of the cured bulk DGEBA/IPDA material and the equivalent stoichiometric ratio ((a/e)eq) was equal to 1. Similarly, for coatings applied either onto degreased titanium or aluminum that were thicker than 400 µm or 200 µm respectively, (Tg)eq and (a/e)eq were all equal to those of the cured bulk DGEBA/IPDA system. For the thinnest coatings (≤ 50 µm) applied onto titanium or aluminum, (Tg)eq values were quite different (≈ 129°C and 108°C respectively) from the bulk ones (163°C) and the equivalent stoichiometric ratios (a/e)eq were 1.19 and 1.14, respectively. Thus it can be assumed that for these thin coatings, a different epoxy network was formed. For coating thicknesses between 50 ≤ h ≤ 400 µm for Ti and 200 µm for Al, a gradient region was observed. In all the following work, the interphase thickness (or width) will be defined as the region where the properties are different from those of the bulk. In the case of the DGEBA/IPDA coatings, the interphase is thicker for the titanium substrate (400 µm) than for the aluminum one (200 µm). So, we can assume that the dissolution rate of TiO2 amorphous oxide is higher than that of the Al2O3 amorphous oxide. According to Ellingham [32], the standard free energies at room temperature of formation of Al2O3 and TiO2 crystalline oxides are equal to 250 and 210 kcal/mol O2, respectively. In a first approximation, we can assume that the amorphous oxide layer formed on aluminum alloy was more stable than the one obtained on titanium alloy. Thus, for the same diamine monomer, the dissolution rate of TiO2 should be higher than that of Al2O3 leading to a wider interphase for TiO2, and this is observed experimentally. The large interphase thickness also means that for a 1 mm thick coating on degreased titanium, the first 400 µm from the substrate surface corresponds to the interphase and that only the remaining 600 µm will have the same properties as the DGEBA/IPDA bulk epoxy. Also, we observe that the interphase thickness depends on the substrate nature (Fig. 4) as well as on the basic behavior of the amine as shown in Fig. 5. The formation of such a thick interphase has to be explained.

Interphase: Formation, characterization and relevance to practical adhesion

9

Figure 3. Variation of the equivalent glass transition temperature versus the coating thickness for degreased titanium and aluminum substrates.

Figure 4. Variation of the equivalent stoichiometric ratio (a/e)eq as a function of the coating thickness for both degreased titanium and aluminum substrates.

10

A.A. Roche et al.

Figure 5. Variation of the interphase thickness versus the pH of monomer aqueous solution.

3.2. Monomers characterization To characterize the chemical changes occurring, DGEBA and IPDA liquid monomers were placed between two chemically treated metallic substrates (Ti or Al) and kept at room temperature for 3 h. Then, Ti or Al modified monomers were scraped from their metallic substrates with a PTFE spatula and visually compared to the pure monomers. Both pure monomers (DGEBA and IPDA) were transparent. No color change of the DGEBA monomers after application onto titanium or aluminum was observed. The IPDA monomers from both aluminum and titanium became milky white (Fig. 6). After two hours within test tubes, a separation was observed. The floating part remained transparent while the precipitate was milky white. One drop each of the precipitate from both Ti and Al modified IPDA were placed between two glass plates and observed using a polarized optical microscope (Fig. 7). Crystals in the modified IPDA liquid were observed corresponding to a new chemical product formed when the IPDA monomer reacted with the metallic oxide and/or hydroxide surface of both titanium and aluminum. Taking into account the basic nature of the pure diamine curing agent, a dissolution of the metallic oxide or hydroxide by the amine is assumed when the monomer is applied onto the metallic substrate. The free potential variation versus the immersion time for chemically etched titanium immersed in a 20 wt% aque-

Interphase: Formation, characterization and relevance to practical adhesion

11

Figure 6. Photographs of pure IPDA and IPDA modified either on titanium or aluminum.

Figure 7. POM micrographs of the modified IPDA precipitates obtained after application onto titanium (a) and aluminum (b) substrates.

ous IPDA solution is reported in Fig. 8 and it suggests a dissolution of the metallic oxide surface. A comparison of the SEM micrographs (Fig. 9) of the degreased titanium and aluminum surfaces to the ones obtained after an application of IPDA for 3 h and rinsing with acetone shows indisputably a chemical attack of the metallic surfaces by the liquid IPDA monomer. The dissolution of metallic oxide or hydroxide surface layers may induce a metallic ion diffusion within the IPDA liquid monomer. Experimentally, this was observed by ICP analysis of pure IPDA and the precipitate portion of Ti and Al modified IPDA as reported in Fig. 10. The presence of titanium or aluminum is shown clearly in the modified IPDA monomer spectrum. Thus the hypothesis of the dissolution of the titanium or aluminum oxide and/or hydroxide by the basic diamine monomer is strengthened. FTIR spectra of pure IPDA and modified IPDA after application onto aluminum,

12

A.A. Roche et al.

Figure 8. Variation of the free potential versus time for a chemically etched titanium substrate immersed in an aqueous solution of IPDA (20 wt%).

Figure 9. SEM photomicrographs of the initial surfaces of degreased titanium and aluminum substrates and after application for 3 h of liquid IPDA monomer and removal using acetone rinse.

Interphase: Formation, characterization and relevance to practical adhesion

13

Figure 10. ICP spectra of pure IPDA and titanium and aluminum modified IPDA.

Figure 11. FTIR spectra of pure IPDA and IPDA modified on titanium, aluminum or gold coated substrates.

titanium or gold coated substrate are shown in Fig. 11. The spectra of pure IPDA and modified IPDA after application onto a gold coated substrate or the floating portion of the modified IPDA on aluminum and titanium are identical. However, new bands (between 1100-1350 cm–1 and 600-700 cm–1) are observed only when IPDA was applied onto titanium and aluminum substrates. FTIR results confirmed that no chemical reaction occurred when coatings were applied onto the gold coated substrates. The diamine monomer (IPDA) clearly reacts with both titanium and aluminum substrates. Fig. 12 shows the reflectance index (RI) SEC spectra obtained from pure diamine monomer (MCDEA) and modified MCDEA with both titanium and aluminum substrates. New small peaks are observed which correspond to M w = 1014 g/mol for titanium and M w = 1011 g/mol for aluminum. These values are twice higher than the pure MCDEA monomer mass average molar mass ( M w = 489 g/mol).

14

A.A. Roche et al.

Figure 12. SEC spectra of pure MCDEA and MCDEA modified on titanium and aluminum substrates.

To characterize the chemical bindings that may appear between the IPDA monomer and metallic ions, nuclear magnetic resonance (NMR) spectroscopy was used. Pure IPDA monomer was actually a mixture of 25% “cis” and 75% “trans” isomers and the corresponding conformations and the carbon atom labeling are shown in Fig. 13. Peak assignment of pure IPDA was achieved using 1H, 13 C and 2D NMR spectra as reported elsewhere [33, 34]. Spectra of pure IPDA monomer and modified IPDA applied onto titanium and aluminum were also obtained. After a complete peak assignment, it can be noted that some of the characteristic nuclear resonance frequencies vary (magnetic shielding) when IPDA was applied onto titanium and aluminum. From the differences between characteristic resonance frequencies, chemical shifts can be calculated [35]. Fig. 13 shows also the NMR chemical shifts for various carbon atoms when IPDA was applied onto titanium and aluminum. Only carbon atoms associated with nitrogen atoms or neighboring carbon atoms were affected when the monomer was applied onto metallic substrates and the shift was found to be the same regardless of the isomer. The chemical shifts obtained for aluminum were about two to three times higher than the ones observed on titanium. From these results, it can be assumed that the bond between amine groups and metallic ions results from the electron-rich nitrogen atom donating its lone pair to the electron-deficient metallic center. Thus, organo-metallic complexes were created by coordination bindings. When the com-

Interphase: Formation, characterization and relevance to practical adhesion

15

Figure 13. Cis and trans isomers configurations and labeling of the two IPDA isomers and chemical shifts observed for modified IPDA on titanium and aluminum.

plex concentration within the liquid monomer reaches the solubility limit (Ks), complexes crystallize as sharp needles as shown in Fig. 7. About forty years ago, the complex formation was reported for organic pigments where metallic ions were introduced to stabilize monomers [36]. More recently other authors have described the complex formation between amines and metals [37-38]. Using such modified monomers, it was interesting to add them to the DGEBA epoxy prepolymer and to characterize the properties of such new materials. 3.3. Pure DGEBA/modified IPDA characterization To fully understand the properties of interphases when DGEBA-diamine mixtures are applied onto metallic surfaces and to verify our various hypotheses, we prepared some epoxy mixtures with pure DGEBA and both Ti and Al modified IPDA monomers. The stoichiometric ratio was kept equal to 1 and the curing conditions were the same as for the pure DGEBA/IPDA system. Figure 14 contains POM photographs obtained before and after the cure cycle for both pure DGEBA/titanium or aluminum modified IPDA mixtures. Comparing Fig. 7 to Fig. 14, a dilution effect was observed before curing, but crystals were still present in the pure DGEBA/Ti or Al modified IPDA mixtures. After curing, crystals remain observable and all around these crystals, a phase separation (black parts in

16

A.A. Roche et al.

Figure 14. POM photographs of mixtures (a/e = 1) of pure DGEBA/Ti and Al modified IPDA.

the photograph) can be observed. It is quite evident that the presence of crystals and the phase separation lead to the formation of a new network. The properties of such modified polymers are reported in Table 1 and are compared to thin coatings on titanium and aluminum and to the pure bulk polymer. The properties of the pure DGEBA/Ti or Al modified IPDA are the same as those of thin films formed onto titanium or aluminum substrates. This means that we have been able to reproduce within a modified bulk system the various phenomena observed for the thin coatings. Moreover, an increase of Young’s modulus of DGEBA/modified IPDA systems was obtained as shown in Fig. 15. Randomly dispersed organo-metallic crystals act as short fibers in a polymer matrix [39-43]. However, this 50% increase in Young’s modulus does not explain the huge increase (300%) in the longitudinal Young’s modulus of thin films as shown in Table 1. Lastly, one drop of IPDA monomer was applied close to an aluminum surface between two glasses. The POM photographs were taken versus time. After

Interphase: Formation, characterization and relevance to practical adhesion

17

Table 1. Physical, chemical and mechanical properties of bulk coating and of thin films applied onto degreased titanium and aluminum substrates

DGEBA/IPDA bulk 40 µm thick film on Al DGEBA/Al modified IPDA 250 µm thick film on Ti DGEBA/Ti modified IPDA

Tg [°C]

a/e

E [GPa]

163 110 108 129 132

1.0 1.14 1.15 1.20 1.22

3.2 10 5.0 11 4.8

E: is the Young modulus a/e: is the stoichiometric ratio

Figure 15. Tensile stress versus strain curves for pure DGEBA/IPDA system and pure DGEBA/Al modified IPDA system.

30 min, the first crystals appeared and after 3 h it can be clearly observed that most of crystals are oriented parallel to the metallic surface as shown in Fig. 16. Thus, the huge increase in the longitudinal Young’s modulus as determined by the flexure test can be easily explained.

18

A.A. Roche et al.

Figure 16. POM photographs of liquid IPDA monomer in the vicinity of degreased aluminum surface after 30 min (left) and 3 h (right).

4. SUMMARY

When epoxy-diamine prepolymers are applied onto metallic substrates, an interphase between the coating part having the bulk properties and the metallic surface was created. Chemical, physical and mechanical properties of the interphase formed depend on the substrate nature and its surface treatment, the nature of the diamine hardener, and the curing cycle. Use of an appropriate curing cycle for the prepolymer allowed us to reach the maximum cure conversion of the polymer and the full interphase formation. Mechanisms describing the formation of the interphase were deduced from comparison of behaviors when epoxy and diamine monomers were applied onto aluminum and titanium. When either pure DGEBA or diamine monomers were applied onto gold coated surfaces, no chemical reaction was observed. In the same way, when the pure DGEBA monomer was applied onto the metallic surfaces, no chemical reaction was observed. On the contrary, when pure diamine monomers were applied onto metallic surfaces, chemical reactions occurred. According to the basic behavior of diamine monomers, a partial dissolution of the surface metal oxide and/or hydroxide was observed. Metallic ions diffuse within the liquid monomer mixture (epoxy-diamine) and react with the amine groups of the diamine monomers to form organometallic complexes by coordination. When the complex concentration is higher than the solubility product, these complexes crystallize as sharp needles. They align themselves parallel to the metallic surfaces leading to an oriented crystalline layer in the vicinity of the substrate surface. Crystals act as short fibers in an organic matrix leading to an increase of the mechanical properties. During the curing cycle, crystals were not fully dissolved and a phase separation was observed to occur all around these crystals inducing the formation of a new epoxy network. The same phenomena would be expected for all metallic substrates as soon as they are covered with an oxide or hydroxide layer. Also, since dissolution and diffusion phenomena are expected, the extent of interphase formation should be related to the contact duration between the liquid diamine and metallic substrates.

Interphase: Formation, characterization and relevance to practical adhesion

19

This enabled us to propose interphase formation mechanisms leading to a better understanding of its chemical, physical and mechanical properties. Acknowledgements We are very grateful to R. Petiaud and V. Massardier for performing the NMR analyses and to R. Diemiaszonek for the ICP analyses. REFERENCES 1. 2. 3. 4. 5. 6. 7. 8.

E. Peillex, Ph.D. thesis, Université de Lyon 1 (1996). C. Fauquet, Ph.D. thesis, Université Paris 6 (1992). P. Walker, J. Coating. Technol., 52, 49 (1980). R.D. Guminski and F.M.P. Meredith, J. Oil. Colour Chem. Assoc., 44, 93 (1961). B. De’neve, Ph.D.thesis, Ecole Nationale Supérieure des Mines de Paris (1993). R.G. Dillingham and F.J. Boerio, J. Adhesion, 24, 315 (1987). J. Nigro and H. Ishida, J. Appl. Polym. Sci., 38, 2191 (1989). F. Gaillard, H. Hocquaux, M. Romand and D. Verchère, Proceedings of Euradh’92, pp. 122127, Karlsruhe (Germany) (1992). 9. D. Verchère, H. Hocquaux, T. Marquais and F. Gaillard, Proceedings of Euradh’92, pp. 488495, Karlsruhe (Germany) (1992). 10. J.E. De Vries, L.P. Haack, J.W. Holubka and R.A. Dickie, J. Adhesion Sci. Technol., 3, 203 (1989). 11. S.G. Hong, N.G. Cave and F.J. Boerio, J. Adhesion, 36, 265 (1992). 12. J.W. Holubka and J.C. Ball, Ind. Eng. Chem. Res., 28, 48 (1989). 13. H. Kollek, Int. J. Adhesion Adhesives, 5, 75 (1985). 14. J. March, L. Minel, M.G. Barthes-Labrousse and D. Gorse, Appl. Surface Sci., 133, 270 (1998). 15. G.S. Crompton, J. Mater. Sci., 24, 1575 (1989). 16. Y.H. Kim, G.F. Walker, J. Kim and J. Park, J. Adhesion Sci. Technol., 1, 331 (1987). 17. D.L. Allara and C.W. White, Amer. Chem. Soc., 273 (1978). 18. A. Entenberg, V. Lindberg, L. Fendrock, Sang-Ki Hong, T.S. Chen and R.S Horwath, in: Metallized Plastics 1: Fundamental and Applied Aspects, K.L. Mittal and J.R. Susko (Eds.), pp. 103113, Plenum Press, New York (1989). 19. F.J. Von Preissing, J. Appl. Phys., 66, 4262 (1989). 20. P. Scafidi and M. Ignat, J. Adhesion Sci. Technol., 12, 1219 (1998). 21. H. Orsini and F. Schmit, J. Adhesion, 43, 55 (1993). 22. M.D. Thouless and H.M. Jensen, J. Adhesion Sci. Technol. 8, 579 (1994). 23. D.R. Mulville and R.N. Vaishnav, J. Adhesion, 7, 215 (1975). 24. K.L. Mittal, in: Adhesion Measurement of Thin Films, Thick Films and Bulk Coatings, K.L. Mittal (Ed.), pp. 5-17, ASTM, STP 640, Philadelphia (1978). 25. L.H. Sharpe, J. Adhesion, 4, 51 (1972). 26. M.F. Finlayson and B.A. Shah, J. Adhesion Sci. Technol., 4, 431 (1990). 27. O. Sindt, J. Perez and J.F. Gerard, Polymer, 37, 2989 (1995). 28. L. Bonnaud, Ph.D. thesis, INSA de Lyon (1999). 29. H. Masood Siddiqi, Ph.D. thesis, INSA de Lyon (1997). 30. N. Poisson, G. Lachenal and H. Sautereau, Vibrational Spectros., 12, 237 (1996). 31. S. Bentadjine, Ph.D. thesis, INSA de Lyon (2000). 32. H.J.T. Ellingham, J. Soc. Chem. Ind., 63, 125 (1944). 33. S. Bentadjine, A.A. Roche and J. Bouchet, in: Adhesion Aspects of Thin Films, K.L. Mittal (Ed.), Vol. 1, pp. 239-260, VSP, Utrecht (2001). 34. S. Bentadjine, R. Petiaud, V. Massardier and A.A. Roche, Polymer, 42, 6271-6282 (2001).

20

A.A. Roche et al.

35. F.K. Bovey, High Resolution NMR of Macromolecules, Academic Press, London (1972). 36. S. Wernick and R. Pinner, Les traitements de surface et la finition de l’aluminium et de ses alliages, Eyrolles, Paris (1962). 37. M. Kinzler, M. Grunge, N. Blank, H. Shenkel and I. Scheffler, J. Vac. Sci. Technol., 10, 26912697 (1992). 38. J.S. Kuo and J.W. Rogers, Surface Sci., 453, 119-129 (2000). 39. J. Bouchet, A.A. Roche, E. Jacquelin and G.W. Scherer, in: Adhesion Aspects of Thin Films, K.L. Mittal (Ed.), Vol. 1, pp. 217-237, VSP, Utrecht (2001). 40. M.G. Bader and W.H. Bowyer, J. Phys. D: Appl. Phys., 5, 2215-2226 (1972). 41. A.R. Saadi and M.R. Piggott, J. Mater. Sci., 20, 431-437 (1985). 42. M. Miwa and N. Horiba, J. Mater. Sci., 29, 973-977 (1994). 43. M.L. Mehan and L.S. Schadler, Composites Sci. Technol., 60, 1013-1026 (2000).

Adhesion Aspects of Polymeric Coatings, Vol. 2, pp. 21–27 Ed. K.L. Mittal © VSP 2003

Depletion, a key factor in polymer adhesion G. FRENS∗ Laboratory of Physical Chemistry, Delft University of Technology, Julianalaan 136, 2628 BL Delft, The Netherlands

Abstract—The chemical properties of polymers are essential for the strength of an adhesion bond. Generally, a random coil of a polymer is repelled, entropically, from an impenetrable surface by the depletion effect. To bond the polymer molecules it must be forced actively towards the interface. Suppression of depletion is brought about by adsorption of specific groups, interacting with polar sites on the substrate. Altering the balance between adsorption and depletion is the most important effect of pretreatments on the strength of an adhesion bond. Adsorption diminishes the effective distance between the surface and the polymer. This, rather than the effect of polar groups on the work of adhesion, leads to order of magnitude increases of the adhesion force between a polymer and a surface. Keywords: Depletion; adhesion; polymers; polar groups; adsorption.

1. INTRODUCTION

A starting point for quantitative adhesion theories is the work of adhesion WA, expressed as: WA = γS + γL – γSL

(1)

The energy WA, of the adhesion between a solid substrate S and a wetting liquid L, follows from the surface energies (in mJ/m2) γS and γL and from their interfacial energy γSL. WA can be determined from contact angle measurements using groups of standard liquids [1-2]. The idea behind the work of adhesion is that when the solid and the liquid are separated, two free surfaces (of the solid S and of the liquid L) are created out of an equal area of contact which had, originally, the interfacial energy γSL. Polymer molecules which are attached to the substrate surface are instrumental in the strength of an adhesion bond. When forces are applied to separate the bond, there is a concentration of stresses at the interface. The stored mechanical energy is released when a crack develops. Propagation of the crack can be stopped, and the bond is saved, when the stresses are redistributed away from the tip of the ∗

Phone: +31 15 278 5180, Fax: +31 15 278 4135, E-mail: [email protected]

22

G. Frens

crack. This is what the relatively large and interwoven polymer coils can do for the strength of an adhesion bond. The polymer molecules can only have such an effect if they are fixed strongly onto the surface. This fixation is the chemical secret in the formulation of adhesives for different materials. Pre-treatment procedures are designed for the same purpose: to optimize the strength of the bond, by attaching the polymers to the bonding substrate. But the mechanism which determines the strength of the bond is not very well understood, as yet, from a theoretical point of view. New observations about the adhesion between polymers and solid surfaces [3] indicate why theories of adhesion which are based on the value of WA [4] have always had serious trouble to adequately predict the strength of the bond between a polymer and the surface of a solid material. This is because the effective distance H between the polymer and the substrate has been overlooked as a variable. It can vary by an order of magnitude, depending on experimental conditions which affect polymer adsorption. 2. ADHESION STRENGTH

Surface energies are the macroscopic expression of the intermolecular (van der Waals) attractions between the molecules near the interface. From the Hamaker equation [5] for the long range dispersion forces across the interface it follows that: VA = WA (δ/H)2,

(2)

VA is the attraction energy between two slabs of materials - such as a solid and an adhesive - separated by a gap of width H. The molecular parameter δ can be understood as the distance of closest approach between the two molecules which are in contact across the interface. The adhesion force, FA is the derivative; i.e., FA = dVA/dH

(3)

FA is the force (per m2) which resists the separation of the materials by outside forces. It represents the strength of the adhesion bond and decreases as the cube of the distance H. The bond strength is, therefore, not only dependent on WA, but also on the effective distance H between the bulk polymer and the surface of the substrate. This may seem trivial when the two materials are in close contact, and, therefore, its consequences are often overlooked in theories of adhesion. These theories then proceed and concentrate on methods to maximize WA for making strong adhesion bonds, e.g. through pre-treatments of the substrate surface which lead to an extra (“polar”) attraction between molecules in close contact at H = δ. Indeed, for a strong bond WA must be large. But it is more important that the effective distance H has the minimum possible value. It is in this aspect of adhesion that the physical chemistry of polymer molecules and of the solvent has a special role to play. Polymer molecules in a solvent or in a melt will keep their distance to the interface at H δ.

Depletion, a key factor in polymer adhesion

23

3. POLYMER SOLUTIONS

Adhesives are fairly concentrated polymer solutions. In cured and dried layers of adhesives, like in a hot melt of a polymer, the per volume concentration Cs of polymer segments approaches that of the pure polymer. In a dilute polymer solution the volume of the random coil is characterised by the extension vector R. The value of R is related to the size of the segment A and the number of segments N in the coil by áR2ñ= N6/5 A2. When the polymer concentration increases, coils of this size will fill the complete volume of the solution at a segment concentration Cs*. At higher concentrations than Cs* the polymer coils interpenetrate. Such a system, with a concentration above Cs* is called semi-dilute. The polymer coils form a network with a characteristic length scale ξ, which is called the correlation length. At Cs*, the correlation length is approximately equal to R. In a semi-dilute solution, where Cs > Cs*, the relation between ξ and Cs can be found from a scaling argument. This relation can always be written as a power law, such as ξ = R (Cs* / Cs)x with an unknown power x. The correlation length ξ is a local parameter, which describes the interpenetrating network. As such, it cannot depend on the lengths of the individual coils. From the theory of a dilute polymer solution in a good solvent it follows that R is proportional to N3/5 and that Cs* is proportional to N–4/5. The power x in the relation between ξ and Cs must, therefore, be 3/4. With Cs = nN/V, in which n is the number of polymer coils in the solution, and β as the excluded volume of one polymer segment of length A, it follows for a semi-dilute system (Cs Cs*). ξ = β–1/4 · A–1/2 · Cs–3/4

(4)

The correlation length determines, among others, the modulus of elasticity in the viscoelastic adhesive material [6], and therefore the rupture dynamics in thin viscoelastic films [7]. In Eq. 4 A is the length of the Kuhn segment in the polymer chain. The excluded volume β of the segment describes its interaction with the solvent. At high volume fractions of the polymer, like in a melt, the correlation length ξ decreases. It approaches the segment length A as ξ = A · φ–3/4 in the limit of a pure polymer without solvent. The Kuhn segment in a polymer molecule is always larger than just one single monomer. In flexible polymers the length A is typically of the order of 1 nm. It may, however, sometimes be much larger – with stiff macromolecules or in special solvents. The segment length A, however, is always the lower limit for the correlation length ξ in a polymer solution when its concentration increases. 4. DEPLETION AND ADSORPTION

All polymer molecules tend to avoid contact with an impenetrable surface. This effect is known as “depletion”. The entropy of the polymer coils would decrease, if they should approach a surface to distances (H) smaller than the correlation

G. Frens

24

length (ξ). This would raise the free energy of the system, and, therefore, polymer molecules leave a gap of width H = ξ at the interface. The free energy involved in this depletion effect is of the order T∆S = kT/ξ 2

(5)

2

per m of contact surface. The effective distance H for the most intimate contact between a polymer coil and a completely inert, but impenetrable, surface is then at least of the order of ξ ≈ A ≈ 1 nm and depends on the stiffness of the chain of segments. This implies, indeed, that for a concentrated solution in a good solvent or a polymer melt the mininum thickness of the depletion layer H δ. The contribution of polar interactions to the adhesion will remain very small as long as H > δ. The adhesion is then mostly determined by the long range dispersion forces across the interface. For strong adhesion this depletion effect must be suppressed. To bring polymer molecules nearer to the interface, the repulsion force due to depletion has to be compensated by attraction between the polymer segments and the surface material. In adhesive systems this attraction is through the specific energies of interaction between polar adsorption sites on the surface and segments of the adjacent polymer chains. Given the right circumstances, the polar groups may pull the polymer chains nearer to the surface. For this, the adsorption energy of the polar groups in segments which attach themselves to the surface must exceed the repulsive depletion energy of kT/A2 (per m2 of interface). The adsorption which determines the quality of the adhesion bond can be measured in the same way as WA, through the determination of contact angles. The method has been described by Pennings [8-11], who showed that for many polymers: ∆γ/γp = α[γs /γp]1.85

(6)

Eq. 6 is called the Pennings Equation. It is used in polymer technology to establish optimum process conditions for precision injection moulding. In the Pennings equation ∆γ is the difference between the surface tension immediately after the contact with the substrate was disjoined and the equilibrium surface tension γp of the polymer layer in air. It registers the presence of the extra groups which have become attached to the surface by adsorption. For optimum bond strength ∆γ can be influenced by pre-treatment of the substrate surface. Pretreatments create more polar groups at the surface and add an extra “polar” contribution to the surface energy γs. The parameter α is the coil compliance of the specific polymer [3], which is a measure of its entropic resistance to the conformation change when it becomes adsorbed. The coil compliance can be influenced by the choice of the solvent, by the temperature and by other process conditions. These may be adjusted and define a technology window for reaching an optimum value of ∆γ and of the bond strength which depends on the adsorption of polymer segments directly onto the bonded surface.

Depletion, a key factor in polymer adhesion

25

5. THE ATTRACTION RANGE, H

If the depletion is overcome by the adsorption of the polymer molecules, the effective distance decreases from H = ξ to H = δ, i.e. from about 1 nm to atomic dimensions. The contact then is as if the substrate were wetted with a liquid instead of covered with polymer segments. At first sight, the difference in H may seem a subtle change indeed, but the strength of the adhesion varies as H 3. Therefore, suppression of the depletion and filling up the gap H between the polymer and the substrate strengthens the adhesion bond by at least an order of magnitude. This is purely an effect of the van der Waals dispersion forces between the bulk of the two materials. These forces become stronger at close range. As a consequence of the interaction of polar groups in the interface the gap width H is reduced, and this has a much larger effect for the bond strength than whatever small contributions the polar groups themselves may have to WA. It is not the work of adhesion which is an indication for the strength of an adhesion bond, but the distance H (of the order of nm) between the bonded materials being reduced by adsorption. An example of the increased adhesion which is created by a pretreatement which generates polar groups on the surface of a substrate is given [12] in Figure 1. Polypropylene samples were treated at different UV-wavelengths in air. This radiation forms ozone, which is adsorbed and reacts at the surface of the polymer so that polar groups are introduced. The surface energy γS of the material increased from 18 mJ/m2 to 52 mJ/m2. Afterwards the samples were glued together with MS-polymer adhesive (a silylated thermosetting adhesive with a polypropylene-ether backbone). The pretreatment led, compared to cured but untreated sam-

Figure 1. Effect of UV treatment of polypropylene on its adhesion strength.

26

G. Frens

ples, to a more than tenfold increase in the adhesion strength when glued together parts were tested for the tensile strength of the adhesive bond. This effect seems too large to be explained from the increase in WA alone. The strongest effect must be due to the reduction of the distance H which is brought about by the adsorption of the MS-polymer on the pre-treated polypropylene substrate. 6. DELAMINATION

For a strong adhesion bond it is essential that the depletion effect is suppressed. The competition between adsorption and depletion will, however, always lead to a preferential adsorption of polar molecules, like water or monomers, at the interface. These small molecules have comparable adsorption energies as the segments, but there is no depletion energy to compete with it. Replacing the adsorbed polar groups of the polymer at the interface with water can undo the adsorption of the polymer segments and activate depletion. Sometimes this is a useful effect, e.g. when release agents separate injection moulded plastic products from their metal mould. But for polymer adhesion bonds this sort of depletion is catastrophic. It causes delamination along the interface between the two materials. Therefore, in order to prevent delamination, the adsorption energy of the polar groups on the substrate surface must always exceed that of water by the (depletion) energy kT/A2. The tendency towards depletion is always lurking, hidden in every polymer adhesion bond which has, of necessity, polymer molecules in the interface. This is the reason for interfacial failure in the presence of small polar molecules like monomers or H2O. Such molecules compete with the polymer segments for the polar sites at the surface. Adsorption of water molecules instead of polymer segments on these sites releases the entropic forces of depletion and causes delamination of the polymer material from the surface. 7. CONCLUSION

Depletion from an impenetrable surface is a universal property of polymeric substances, which is potentially detrimental for the strength of adhesion bonds. Depletion is suppressed when the adsorption energy of polar groups in the polymer is large enough to compensate for the free energy of depletion. Adsorption reduces the effective distance H between the substrate and the polymer from the length A of a Kuhn segment to the size of an adsorbed monomer δ. The change ∆γ is the effect of adsorption too, but it is too small to explain why the presence of polar groups in the interface is so important for the strength of an adhesion bond. The reason why adsorption does increase the strength by more than an order of magnitude is that it suppresses the depletion effect and creates a closer contact of the polymer molecules with the substrate surface.

Depletion, a key factor in polymer adhesion

27

REFERENCES 1. W.A. Zisman, in Contact Angle, Wettability and Adhesion, Adv. Chem. Series No. 43, p.1, Amer. Chem. Soc., Washington DC (1964). 2. F.M. Fowkes, in Contact Angle, Wettability and Adhesion, Adv. Chem. Series No. 43, p.99, Amer. Chem. Soc., Washington DC (1964). 3. G. Frens and A. van der Put, J. Adhesion. Sci. Technol. 12, 1355 (1998). 4. K.L. Mittal (Ed.), Physicochemical Aspects of Polymer Surfaces, Vols. 1&2, Plenum Press, New York (1982). 5. H.C. Hamaker, Physica, 4, 1058 (1937). 6. L.J. Evers, S.Yu. Shulepov and G. Frens, Faraday Discussions, 104 (1997). 7. L.J. Evers, S.Yu. Shulepov and G. Frens, Phys. Rev. Lett., 79, 4850 (1997). 8. J.F.M. Pennings, Colloid. Polym. Sci. 256, 1155 (1978). 9. J.F.M. Pennings and B. Bosman, Colloid. Polym. Sci. 257, 720 (1979). 10. J.F.M. Pennings, Colloid. Polym. Sci. 258, 1099 (1980). 11. J.F.M. Pennings, “Adsorption of Vinyl Copolymers from Melts and Solutions”, Ph.D. Thesis, TU Eindhoven (1982). 12. H.P. Poulis, in: “UV/Ozone Surface Treatment of Polymers for High-Quality Adhesion in the Footwear and Coating Industry”, Synthesis report (for publication), Brite Euram III-project no. BE-S2-2032, p. 9 (1998).

This page intentionally left blank

Adhesion Aspects of Polymeric Coatings, Vol. 2, pp. 29–43 Ed. K.L. Mittal © VSP 2003

Attaining adhesion/cohesion within painted plastics ROSE A. RYNTZ∗ Visteon Corporation, 401 Southfield Road, Dearborn, MI 48121

Abstract—Adhesion to thermoplastic olefin (TPO) substrates, blends of polypropylene (PP) and an elastomer, is often difficult to attain due to the low surface energy, nonpolar nature of the surface. One often specifies, therefore, that the surface of the substrate be pretreated. Pretreatment methods can vary from oxidation of the surface, e.g., plasma or flame, to diffusion and mechanical interlocking of an applied polymer with the surface. Mechanical interlocking with the nonpolar TPO surface can be achieved through the use of an adhesion promoter, namely a chlorinated polyolefin (CPO). The type of CPO utilized, in addition to the types of solvents and heat effects utilized, can substantially influence the degree of adhesion/cohesion obtained within the CPO/TPO system. This paper reviews the factors influencing the adhesion/cohesion of painted TPO substrates. Heat histories, TPO molding variations, CPO types, including solvent and resin variations, and topcoat (basecoat/clearcoat) chemistries are all found to influence the adhesion/cohesion of the painted TPO assembly. Keywords: Adhesion; cohesion; plastic; chlorinated polyolefin; thermoplastic olefin.

1. INTRODUCTION AND DISCUSSION

The lack of adhesion to low surface energy substrates (surface energy ~ 30 mJ/m2 for PP) has been attributed to their inertness, including poor wettability, lack of polar functionality, and good solvent resistance. The introduction of an elastomer, producing a TPO, generally provides better adhesion than PP alone. Attaining adhesion to TPO, however, is still often problematic. The adhesion of organic coatings to a TPO is strongly dependent upon the type of surface treatment the thermoplastic material receives prior to painting. One such treatment involves the application of a thin layer (7 to 14 µm) of an adhesion promoting primer that is comprised of a CPO. Although materials of this type are known to establish good adhesion between the TPO and subsequent paint layers, the mechanism of adhesion promotion is not unequivocally known. The mechanism by which CPO coatings promote adhesion to TPO has been speculative over the last several years [1-5]. Waddington and Briggs [5] have reported experiments that suggest CPO promotes adhesion by diffusing into the ∗

Phone: (313)755-6164, Fax: (313)755-0601, E-mail: [email protected]

30

R.A. Ryntz

substrate and forming a chain entangled interphase between substrate and adhesion promoter. They used secondary ion mass spectrometry (SIMS) and x-ray photoelectron spectroscopy (XPS) to study the peel failure between a CPO and two grades of TPO: one that exhibited strong adhesion and one that exhibited weak adhesion. Chlorine was detected on the surface of the TPO that exhibited strong adhesion, but not on the surface of the TPO that exhibited weak adhesion. In a separate experiment, they coated the two grades of TPO with a CPO and measured the amount of chlorine at the air interface of the CPO. Lower levels of chlorine were found over the TPO grade that exhibited better adhesion. This depletion of chlorine supported a mechanism based on the diffusion of CPO into the TPO. Clemens et al. [1] utilized surface characterization techniques, including electron spectroscopy for chemical analysis (ESCA), energy dispersive X-ray analysis (EDXA), time-of-flight secondary mass spectrometry (TOFSIMS), and transmission electron microscopy (TEM), in attempts to elucidate the mechanism by which CPO primer coatings promoted adhesion of paints to polypropylene and TPO. The coatings, their interphasial failures, and taper-cross sections were studied, using both waterborne and solventborne CPO primers. It was determined that CPO primers did not penetrate deeply into the polyolefin substrates, but were quite mobile following application of the topcoat. The mechanism by which CPO primer coatings promoted adhesion was discussed in terms of better surface wetting and thus contact, allowing for sufficient dispersion interactions with the substrate, and intermixing/bonding with the more polar topcoat. Solventborne CPOs generally showed failure at the CPO/polyolefin interface when dried at ambient temperatures. Waterborne CPOs showed equivalent adhesion promotion, although cohesive failure within the primer layer was more common and appeared to be a function of formulation parameters. Work done by Ryntz and coworkers [6, 7] demonstrated that adhesion to TPO was strongly influenced by the type and amount of solvent contained within the adhesion promoter (CPO containing primer) as well as the surface morphology of the TPO substrate. The morphological changes at and near the surface of TPO affect not only paint adhesion but also the cohesive integrity of the painted plastic assembly. Morphological changes in the TPO substrate are accomplished in the presence of solvent from the adhesion promoter (as well as from subsequent topcoats) and vary depending upon bake time and temperature. If it is assumed that the morphology of TPO is indicated by a closely packed, crystalline surface layer of PP, under which lies a rubber-rich boundary layer, then adhesion attainment is dependent upon the diffusion of CPO adhesion promoter through the crystalline PP surface layer and its swelling and entanglement with the rubber-rich layer beneath it (Figure 1). In general, a greater degree of rubber swelling is indicative of a higher amount of solvent penetration. Initial adhesion to TPO appears to be a function of polypropylene crystallinity at the surface only as it relates to the ability of solvents to penetrate through and into the rubber-rich layer beneath it.

Attaining adhesion/cohesion within painted plastics

31

Figure 1. Painting TPO.

One must be aware, however, of the rubber swelling effect on the ability of the TPO to withstand stress (compressive or shear) within its boundary layers in a swollen state. For example, when the rubber is swollen (elongated) to near its tensile break point, it would require little imparted stress to delaminate the rubbery boundary layer from the bulk morphology of the TPO beneath it. Peel strengths (given in terms of cohesive integrity values) performed on TPO plaques exposed to a compatible solvent (xylene, solubility parameter near that of TPO) indicated that those plaques exposed to solvent (30 minutes at 25°C) afforded lower values than those not exposed to solvent (Figure 2) [7]. This variance can be accounted for by increased swelling or plasticization of the top TPO layers. When the same plaques were swollen by slow evaporating (Aromatic 100) versus fast evaporating (tolu-

32

R.A. Ryntz

Figure 2. TPO cohesive integrity as a function of solvent exposure.

ene) solvents and then flashed-off through the use of an infrared oven, very little difference in peel strength values was seen. In order to attain maximum adhesion to TPO, therefore, one tries to maximize solvent diffusion into the substrate while minimizing the amount of solvent remaining in the substrate after bake. Solvent diffusion, however, is only indicative of the ability of the CPO, the actual adhesion promoting moiety, to be carried into the substrate and not of the actual CPO penetration. In order to understand the amount of diffusion accomplished by both the CPO and solvent a fluorescent “tagging” experiment was performed to monitor the depth of interpenetration of both substances via fluorescence microscopy. The use of an unbound fluorescent dye, 7,7'-dimethyl-4aminocoumarin, in the solvent was utilized to follow the depth of solvent penetration. A covalently bound dye, attached directly to the maleic anhydride component of the CPO, was utilized to follow the depth of CPO penetration. This technique is not unfounded, being studied by Li et al. [8] to determine the surface morphology of a polymer blend through laser confocal fluorescence microscopy. The ability of the solvent to swell the TPO substrate (Figure 3) was studied in relation to the depth of solvent penetration into the substrate versus the bake temperature (Figures 4 and 5). This was accomplished through the use of fluorescence microscopy as described above. The depth of penetration of the “free” dye, 7,7'-dimethyl-4-aminocoumarin, was measured in a 70/30 PP/Rubber w/w compounded TPO as a function of bake temperature while maintaining equivalent solvent composition (80/20 xylene/Aromatic 100 based on weight). The dye penetration, as viewed from a cryogenically “cracked” cross section of the treated TPO substrate, was found to increase as a function of bake temperature (Figure 5 vs. Figure 4). The depth of penetration was followed by monitoring the

Attaining adhesion/cohesion within painted plastics

Figure 3. Swellability of TPO.

Figure 4. Free dye penetration into TPO at 25°C.

Figure 5. Free dye penetration into TPO at 121°C.

33

34

R.A. Ryntz

Figure 6. Topographical fluorescence of “tagged CPO” on TPO.

increase in the optical intensity of the dye fluorescence (right hand side of both Figures). Dye penetration values were taken at 50% of the intial intensity of the fluorescence trace shown in the right side of the Figures. At room temperature, the dye is shown to migrate roughly 150 µm into the substrate, whereas at 121°C, the depth of penetration increases to roughly 600 µm. This depth of penetration merely mimics the amount of free dye carried into the substrate by the solvent and not the actual diffusion of the much higher molecular weight CPO molecule. In a separate fluorescence experiment the fluorescent tag was covalently bound directly to the maleic anhydride functionality of the CPO molecule. The penetration of the CPO molecule was then determined by monitoring the fluorescence with a laser confocal microscope. In Figure 6 one can view the CPO/TPO topographical fluorescence as a function of temperature. A surface profile of the orientation of fluorescence in the same section of the substrate as a function of bake temperature is displayed. It can be seen that as the fluorescently tagged CPO adhesion promoter is baked the fluorescent (light sections of the diagrams) regions appear to coalesce. One can postulate that the CPO diffuses through the TPO surface and entwines with the rubberrich areas of the TPO directly beneath the surface. As the swelling continues, through the interaction of the solvent-laden CPO with the rubber, the CPO/rubber regions begin to break through the surface and coalesce.

Attaining adhesion/cohesion within painted plastics

35

Figure 7. Chloride ion diffusion into TPO via TOFSIMS analysis.

Laser confocal fluorescence microscopy also confirmed depths of penetration of the fluorescently tagged CPO. The depth of penetration was determined by rastering the laser across the selected section of the TPO at various focal planes. The depth of CPO penetration was then re-created by addition of the various focal plane depths where fluorescence was seen. The fluorescence penetration after application of the fluorescently tagged CPO and a 30-minute dwell time at 25°C was approximately 4 µm. After a 121°C bake, the depth of fluorescence increased to ca. 10 µm. Time-of-flight secondary-ion-mass spectrometry was utilized to evaluate the depth of penetration of the adhesion promoter based on imaging of the chloride ion contained within the CPO. The chloride images and line scans of crosssectioned samples of the chlorinated polyolefin based adhesion promoter coated TPO system are shown in Figure 7. The pixel intensity of the chloride ion (as determined with the gallium laser) can be viewed pictorially on the left side of the Figure while the absolute intensity is shown graphically on the right side of the Figure. A number of important features are immediately apparent in the images. A well-defined line, representing the adhesion promoter layer, is readily apparent in the pictorial side of the Figure. The thickness of this layer is about 10-15 µm. Localized areas of high intensity chloride observed along this sharp line are suggestive of topographical anomalies introduced into the sample surface by the microtoming technique used to prepare the sample. This turns out to be a critical factor in determining the degree of useful information that can be gained from the analysis. Most importantly, chloride diffusion into the TPO is noted graphically on the right side of the Figure. The diffusion into the TPO substrate appears to be on the order of 3 to 9 µm. The diffusion of a chlorinated polyolefin polymer having a molecular weight of about 15,000 could reasonably be expected to result in several orders of magnitude less diffusion into the TPO substrate. However, non-Fickian behavior resulting from solvent induced swelling, crazing and porosity induced by phase separation of the elastomer from the polypropylene would not be unexpected.

36

R.A. Ryntz

Figure 8. Effect of CPO resin type on its adhesion (peel strength) to TPO.

These results correlate quite well with those determined by Komatsu et al. [9] as well as the TOF-SIMS studies. The slightly higher degree of interpenetration of the CPO into TPO, as suggested by the fluorescent studies relative to the SIMS data, can be explained in the context of molecular weight. A gel permeation chromatography (GPC) examination of the tagged CPO showed small amounts of low molecular weight moieties present within the tagged CPO. The low molecular weight components in the CPO compound are believed to contribute to the deeper depths of penetration. The effects of structural differences within the chlorinated polyolefins and the effect they have on coating performance is described in the work of Fujimoto [10] and Ryntz [11]. In effect, the work by the above-mentioned authors shows the effect of chlorination of the polyolefin on adhesion and crystallinity of the polymer. In summary, the higher the degree of chlorination the less crystalline is the chlorinated polyolefin, thus making it more compatible with other typical filmforming compounds used in coatings. However, the lower the level of chlorine the better the adhesion to olefinic substrates. These two opposing parameters must be balanced with one another to provide adhesion promoters with the best balance of properties. In the work performed by Ryntz [11], the degree of maleation of the chlorinated polyolefin, in conjunction with the chlorinated polyolefin molecular weight and molecular weight distribution, was also shown to affect the degree of adhesion between the chlorinated polyolefin and the TPO substrate (Figure 8). As can be seen in the figure, the resins with the highest 90 degree peel strength (triangles are indicative of relative peel strength only) were those with highest chlorine levels, some level of maleic anhydride, low number average molecular weight, low polydispersities, and low viscosities (CPO resins labeled “A” and “B”).

Attaining adhesion/cohesion within painted plastics

37

Figure 9. Effect of molding condition on TPO swellability.

Although the work described above identified the importance of CPO structural attributes in adhesion of CPO to TPO, another anomaly was found to occur with respect to adhesion and location along the TPO substrate. As shown in Figure 9, when swelling experiments were performed near or away from the proximity to the “gate” (the location in a injection molded substrate where the hot plastic melt is introduced into the part), variations in swellability occurred [6]. As would be expected, those materials with more rubber should swell more. However, it was noted that the areas “near” the gate in the TPO part exhibited lower amounts of swelling. In a separate experiment, the adhesion of the topcoat “near” and “away” from the gate in TPO injection molded plaques was determined via a gas soak test [12]. It was found that when white topcoated, black pigmented TPO substrates were cross-hatched and immersed in ethanol laden gasoline, the adhesion began to become differentiated at the different locations. As shown in Figure 10, the white topcoated TPO substrates, when exhibiting adhesion loss, became black (indicative of the black substrate). In traversing from the top of the Figure to the bottom, the TPO flexural modulus increases and the fluorescent dye penetration decreases. With each set of TPO materials, there are portions of the substrate representing the “near” gate and “away” from the gate locations (there are four TPO materials, and each set of 2 represents the gate location). As is evidenced in the Figure, the lower modulus (higher rubber content) TPO materials exhibit better adhesion than those with higher modulus. As one traverses the Figure from top to bottom, it is also evident that the area “near” the gate in higher modulus substrates begins to fail before the area “away” from the gate. Several researchers have examined the importance of the variation in TPO surface and its affect on adhesion [7, 13, 14]. The surface morphology of semi-

38

R.A. Ryntz

Figure 10. Gasoline resistance of various modulus topcoated TPO substrates as a function of position from gate.

Attaining adhesion/cohesion within painted plastics

39

Figure 11. Birefringence vs. gate location in injection molded TPO.

crystalline polymers has been shown to be similar to or substantially different from that present in the bulk as a consequence of processing conditions [15]. The relatively minor changes in morphology lead to specific changes in the chemical, physical, and mechanical properties of the surface and of the interior or bulk. Since chemical reactivity, gas permeability, adhesion, friction, and abrasion resistance are dependent on surface morphology, it is of interest to delineate those factors that influence the formation of surface structure. It has been found that cross-polarized light can be utilized as an effective means with which to measure the skin/shear layer in injection molded TPO [7]. The shear zone, under cross-polarized light, is birefringent. As shown in Figure 11, however, the shear layer thickness can vary as a result of the way a part is filled. Near the injection molding gate, the shear layer appears thinner (57 µm) versus away from the gate (119 µm). It was also determined that the skin layer near the gate exhibited a greater “microhardness” than that away from the gate, possibly as a result of increased packing density or variations in surface crystallinities. The surface hardness variation may account for the adhesion variations, since it was also determined that areas “near” the gate exhibited lower dye penetration values. Fujiama and Kimura [13] found that a relationship existed between the thickness of the skin layer in polypropylene and its yield strength. The skin-layer is composed of a “shish-kebab” like structure where crystalline lamellae align perpendicular to the flow direction (referred to as transcrystallinity). Generally thicker shear zone skin layers afforded higher tensile strengths (Figure 12).

40

R.A. Ryntz

Figure 12. Yield strength of TPO as a function of skin layer thickness [14].

The types of materials that contribute to the “skin layer” in injection molded TPO can vary [7]. If the TPO contains low molecular weight additives, e.g., processing stabilizers, viscosity adjusters, or even process degraded or low molecular weight “tails” of the polymers, one can experience additive migration to the TPO surface (Figure 13). The species migrating to the surface can amount to formation of a “weak boundary layer (WBL)” that exhibits poor cohesion to the bulk TPO. The relationship of “skin layer” thickness to cohesive integrity of the TPO substrate became an important consideration when a topcoated substrate was exposed to a compressive/shear stress [16]. Thermoplastic polyolefin substrates, under compressive tensile shear stress, often delaminate cohesively within the TPO substrate, where cohesive delamination was found to be dependent on load, temperature, and shear velocity parameters of the tensile shearing event. The failure was found to occur in the weak boundary layers of the TPO substrate, most often between 25 and 100 µm into the substrate (Figure 14). It is interesting to note that if one exposes unpainted TPO to the compressive/tensile shearing, no failure occurs. It appears that the coefficient of friction of the surface, being very low for polypropylene, allows the shear load to be transmitted across the monomolecular low friction layer. When painted, however, the friction coefficient of the paint, being higher than that of the TPO surface, allows the force to be transmitted into the painted TPO surface.

Attaining adhesion/cohesion within painted plastics

41

Figure 13. Formation of weak boundary layers within TPO as a function of heat.

Figure 14. Gouge in topcoated TPO.

It was determined that the most significant parameters in the paint which affected “gouge” damage resistance were glass transition temperature, secant modulus at break, and the static coefficient of friction [17]. In order to better understand, however, the attributes of the coating type utilized in painting TPO in relationship to the gouge resistance of the system a model coating system was de-

42

R.A. Ryntz

Figure 15. Gouge performance vs. clearcoat type.

signed [16]. Waterbased hydroxyl functional styrenated acrylic clearcoats, containing non-aggressive solvents so as not to swell the TPO, were formulated. The gouge resistance of the formulated clearcoats was measured on a compressiveshear loading device. Figure 15 depicts the gouge resistance of the clearcoats. The gouge area/energy rating for the most severe test conditions, labeled DOE Run #8 (68.3°C, sliding velocity of 3.4 cm/sec, and a load of 272.4 kg) versus the less severe conditions (labeled DOE Runs #2, #4, and #6, where the loads, sliding velocities, and temperatures are lower) are shown. The smaller the value (a rating of 0 depicts no damage) the better the performance. The coating formulation nomenclature is: #77A, D, G, and J are hydroxyl functional styrenated acrylics with a Tg of 45°C; where 77D and J contain surface free energy modifier where 77 A and D are not crosslinked where 77 G and J are crosslinked with a blocked isocyanurate #77C, F, I, and L are hydroxyl functional styrenated acrylics with a Tg of 65°C; where 77 F and L contain surface free energy modifier where 77 C and F are not crosslinked where 77 I and L are crosslinked with a blocked isocyanurate As viewed in Figure 15, all of the coatings, regardless of the Tg or crosslinker level, when formulated with added surface free energy modifier (mar agent), showed no gouge (77-D, 77-J, 77-F, and 77-L). It appears generally that as the Tg increased (77-C vs. 77-A and 77-I vs. 77-G) the gouge damage increased. It is be-

Attaining adhesion/cohesion within painted plastics

43

lieved that the brittleness of the higher Tg resin accounts for brittle fracture of the coating during applied compressive load resulting in asperity formation and subsequent gouging. Added crosslinker, 77-G vs. 77-A and 77-I vs. 77-C, appeared to improve gouge resistance. This is believed to be due to an increase in tensile strength of the coating when crosslinked. 2. SUMMARY

Mechanical interlocking of topcoat with the nonpolar TPO surface can be achieved through the use of an adhesion promoter, namely a chlorinated polyolefin (CPO). The type of CPO utilized, in addition to the types of solvents and heat effects utilized, can substantially influence the degree of adhesion/cohesion obtained within the CPO/TPO system. Heat histories, TPO molding variations, CPO types, including solvent and resin variations, and topcoat (basecoat/clearcoat) chemistries were all found to influence the adhesion/cohesion of the painted TPO assembly. REFERENCES 1. R.J. Clemens, G.N. Batts, J.E. Lawniczak, K.P. Middleton and C. Saas, Proc. XIXth Conf. on Organic Coatings Science and Technology, Athens, Greece, p. 105 (1993). 2. K. Grundke and H.J. Jacobasch, Farbe und Lacke, 98, 934 (1992). 3. Y. Aoki, J. Polym. Sci.: Part C, 23, 855 (1968). 4. T.J. Prater, S.L. Kaberline, J.W. Holubka and R.A. Ryntz, J. Coatings Technol., 68 (857), 83 (1996). 5. S. Waddington and D. Briggs, Polymer Commun., 32 (16), 506 (1991). 6. R.A. Ryntz, Q. Xie and A.C. Ramamurthy, J. Coatings Technol., 67 (843), 45 (1995). 7. R.A. Ryntz, Prog. Org. Coatings, 27, 241 (1996). 8. L. Li, S. Sosnowski, C. Chaffey, S. Balke and M. Winnik, Langmuir, 10, 2495 (1994). 9. Y. Komatsu, J. Kamimura, O. Aoki, D. Chung and M. Wiseman, SAE Technical Paper #940187 (1994). 10. F. Fujimoto, Paint and Resin, 36 (February 1986). 11. R.A. Ryntz, Advanced Coating Technology Conference Proceedings, Engineering Society of Detroit, Detroit, MI (April 1997). 12. R.A. Ryntz and B. Buzdon, Prog. Organic Coatings, 32, 167 (1997). 13. M. Fujiyama and S. Kimura, Kobunshi Kobunshu (English Edition), 4 (10), 777 (1975). 14. M.R. Kantz, H.D. Newman and F.H. Stigale, J. App. Polym. Sci., 16, 1249 (1972). 15. D.R. Fitchmun and S.J. Newman, J. Polym. Sci.: Part A2, 8, 1545 (1970). 16. R.A. Ryntz, M. Everson and G. Pollano, Prog. Organic Coatings, 31, 281 (1997). 17. B. Buzdon and R.A. Ryntz, SAE Paper, presented at SAE Conference, Detroit, MI (1996).

This page intentionally left blank

Adhesion Aspects of Polymeric Coatings, Vol. 2, pp. 45–64 Ed. K.L. Mittal © VSP 2003

Scanning electric potential microscopy (SEPM) and electric force microscopy (EFM) imaging of polymer surfaces ELIZABETH F. DE SOUZA,1 MÁRCIA MARIA RIPPEL,2 AMAURI JOSÉ KESLAREK,2 ANDRÉ GALEMBECK,2 CARLOS ALBERTO RODRIGUES COSTA,2 ÉRICO TEIXEIRA NETO2 and FERNANDO GALEMBECK2,∗ 1

Instituto de Ciências Biológicas e Química, Pontifícia Universidade Católica de Campinas PUC-Campinas, PO Box 1111, CEP 13020-904, Campinas - SP, Brazil 2 Instituto de Química, Universidade Estadual de Campinas - UNICAMP, PO Box 6154, CEP 13083-970, Campinas - SP, Brazil

Abstract—Scanning electric potential microscopy (SEPM) and electric force microscopy (EFM) images of polymer surfaces are presented and compared to standard non-contact AFM images, acquired concurrently for the same areas. Samples used were synthetic latex films with well-known distribution of chemical constituents and thus of ionic electrical charges, as well as a natural latex film and two finished industrial samples. Pairs of AFM and SEPM or EFM images show a variable degree of correlation, thus evidencing the independence of topographical and electrical features of the samples, on the micro- and nanoscopic scales. In every case, contrasting domains with non-zero electric potentials are observed, extending for a few tenths of a micrometer and creating an electric mosaic in the otherwise neutral polymers. Large electric potential gradients across the film surfaces are observed which are probably relevant for polymer surface adhesion, coating and contamination. Keywords: Electric charges in polymers; scanning electric potential microscopy (SEPM); electric force microscopy (EFM); electric maps.

1. INTRODUCTION

Polymer surface properties are closely related to the performance of these materials in areas of adhesion and coating, biocompatibility, and contamination [1]. Polyolefins and related thermoplastics are often used as dielectrics and they have a strong ability to acquire electrical static charges especially at their surfaces. Many other important problems are related to the presence of electric charges in

∗

To whom all correspondence should be addressed. Phone: 055 51 788 3080, Fax: 055 51 788 3023, E-mail: [email protected]

46

E.F. de Souza et al.

dielectrics, such as: electret formation, space and residual charges, double-layer formation at interfaces, and interfacial polarization [2]. There is a relevant conceptual constraint in dealing with these problems, which is the idea of charge neutrality as a normal state for polymer dielectrics at every relevant length scale, from the macromolecules up to the macroscopic plastic parts (films, coatings, tubes) and devices [3]. A direct measurement of electric potentials in dielectrics is a difficult task, although suitable techniques have been available for the past century. Specific techniques have been devised for the case of monolayers formed by organic substances (e.g. surfactants, polymers) on top of aqueous solutions [4]. For instance, the oscillating electrode technique has revealed a great deal of information on the distribution and orientation of different substances on liquid surfaces [5]. More recently, the advent of the scanning probe microscopes has made available techniques for producing maps showing local changes of charge density on insulator surfaces [6]. Other properties measured with spatial resolution by using these instruments are: dielectric constant [7], film thickness of insulating layers [8], surface potential [9] and ferroelectric susceptibility [10]. A scanning force microscope measures forces between the probing tip and the sample surface [11]. It is comprised of a force sensor mounted on a cantilever (tip), an optical sensor, a positioning feedback system, an x-y-z piezoelectric scanning system, and a microcomputer for the acquisition of data and image processing [12]. A scheme of the basic setup of an atomic force microscope (AFM) is presented in Figure 1. Many modes of operation for AFM have been developed, and non-contact mode AFM is the most common. In the dynamic non-contact operation, the cantilever is mechanically oscillated to vibrate on or near its resonance frequency by a piezoelectric vibrator. The gradient of the force between the tip and sample modifies the compliance of the cantilever, hence inducing a change in vibration amplitude as a function of the tip-sample spacing, which is in the 10-100 nm range. The associated circuitry monitors the resonant frequency or vibrational amplitude of the cantilever and keeps it constant, with the aid of a feedback system that moves the scanner up and down. By keeping the resonant frequency or amplitude constant, the system allows the tip-to-sample spacing to be held constant for profiling applications [12, 13]. Several forces contribute to the deflection of an AFM cantilever, but long-range van der Waals forces predominate in the non-contact mode [14]. Electrostatic forces can also deflect the cantilever when it scans over locally charged domains of the sample surface. This is caused by the induction of charge in the tip, which results in changes in the tip-to-surface force gradient [15]. For instance, over a negatively charged region of the sample surface, positive charges are induced in the tip, resulting in an additional attractive electric force. The same attraction occurs between a positively charged region of the sample and the negative charges induced in the tip. In these two cases, the force between the tip and

SEPM and EFM imaging of polymer surfaces

47

Figure 1. Schematic diagram of the setup of an atomic force microscope operating in the noncontact mode. The inset shows a magnified representation of the tip-sample contact region.

the sample is increased and it is impossible to identify the electric charge signal in the area under examination. However, if a positive potential is applied to the sample-holder, the negative charges of the sample are compensated, resulting in a decrease of the attractive electric force between the sample and the tip due to a decrease in the induced charge. In the neutral regions of the sample, the positive potential applied to the sample holder induces negative charges in the tip, which is attracted to the surface. Finally, in the positive domains of the sample the tip is under the added electrostatic attraction of the sample and the sample holder. Therefore, the use of an external potential applied to the sample-holder allows us to identify the charge signal in the sample surface. Figure 2 presents a schematic representation of charges induced on the scanning tip, and the corresponding profiles of the attractive electrostatic force acting on the tip. In this work, we used non-contact atomic force microscopy (AFM) to identify the morphological features of various polymeric surfaces and electric force microscopy (EFM) as well as scanning electric potential microscopy (SEPM) to investigate the microelectric characteristics of the examined surfaces. These two techniques for electric sensing can be practiced using available scanning force microscopes, but their use is still very limited, as can be gathered from the literature and relevant databases.

48

E.F. de Souza et al.

Figure 2. Schematic representation of charges induced on the scanning tip, and the corresponding profiles of the attractive electrostatic force acting on the tip. (a) The sample holder is electrically neutral: the electric force between the tip with induced charges and the sample surface charged domains are always attractive. (b) The sample holder is positively charged by the application of an external electric potential: the attractive electric force between the tip and the sample surface is reduced over the negative domains of the sample surface, which are compensated by the positive charges of the sample holder. (c) The sample holder is negatively charged by the application of an external electric potential. (d) The negative charge of the sample holder is increased by the application of a more negative external electric potential.

2. EXPERIMENTAL

2.1. Preparation of samples Titanium dioxide: This sample was used as a control, to verify the separation of van der Waals and electrostatic contributions to imaging. A drop of a very dilute aqueous dispersion of titanium dioxide nanometric particles (Riedel-de Haën AG, Seelze-Hannover, Germany) was placed on freshly cleaved mica and allowed to dry in air. Poly(styrene-co-2-hydroxyethylmethacrylate) (PS-HEMA) latex: The PSHEMA latex was prepared by batch surfactant-free emulsion copolymerization of styrene and 2-hydroxyethylmethacrylate. Details on the latex preparation and characterization are given in previous works from this laboratory [16, 17]. The

SEPM and EFM imaging of polymer surfaces

49

PS-HEMA latex film was prepared by depositing of one droplet of the latex dispersion (1% solids) directly on top of a freshly cleaved mica surface, and allowing to dry at room temperature for 48 h. Poly(styrene-co-acrylamide) (PS-AAM) latex: The PS-AAM latex was prepared in this laboratory by batch surfactant-free emulsion copolymerization of styrene and acrylamide [18]. The PS-AAM latex was dialyzed and further fractionated by zonal centrifugation in a linear aqueous sucrose density gradient (100µL latex, 1% solids, 19 krpm). Each latex fraction was then dialyzed to remove sucrose by centrifugal ultrafiltration [19], and the denser latex particles were deposited on mica, as in the case of PS-HEMA. Natural latex (NL): Raw latex from Hevea brasiliensis (from a Rubber Research Institute of Malaysia 600 clone) without any stabilizing agent was collected at the Instituto Agronômico de Campinas and centrifuged at 10 krpm, for 2 h at 5°C. One droplet of the latex dispersion (2.6% solids) was dried on top of a freshly cleaved mica surface at room temperature for 48 h, thus yielding a film. Poly(styrene-co-n-butylacrylate) (PS-BA) latex: Droplets of the non-dialyzed latex dispersion (1% solids) were deposited directly on top of a glass slide cover and dried at (20 ± 2)°C, under 50% relative humidity, for 48 h. Commercial RetroplanTM polyester film (Polyester): Small fractions of the printable and of the back sides of a polyester film were fixed with carbon-filled adhesive onto a freshly cleaved mica surface. Commercial high-density polyethylene film (HDPE): The HDPE film was washed with water, ethanol and air-dried. Then a small fraction of the film was also fixed with a conductive adhesive onto a freshly cleaved mica surface. 2.2. Atomic force microscopy (AFM) The dry latex and commercial films obtained as described above were examined by AFM in order to observe their morphological features. AFM measurements were performed in a Topometrix DiscovererTM instrument operating in noncontact mode, under air and concurrently with EFM or SEPM imaging. Topography changes were sensed by monitoring the detector signal amplitude at 300 x 300-pixel resolution, using platinum-coated silicon nitride tips with 20 nm nominal radius. This is one-third or less of the pixel size, in the images presented in this paper, and we avoided considering image features smaller than 60 nm. The scan rate was 2.5 µm/s. 2.3. Electric force microscopy (EFM) To perform EFM measurements, each sample line was scanned at two heights above the surface: 10 nm and then 60 nm. In the first scan (10 nm) the tip response is dominated by short-distance van der Waals forces. In the second scan (60 nm) the electrostatic interactions predominate, due to their slower dependence on distance. As the tip crosses over an electrically charged region, charge is induced in the tip resulting in changes in the tip-to-surface force gradient, causing a

50

E.F. de Souza et al.

change in the effective cantilever spring constant and in the resonance frequency of the tip. This change in the resonance frequency modifies the signal phase detected by a lock-in amplifier. Therefore, the EFM signal is measured by monitoring the changes in the signal phase [20]. Figure 3 shows a scheme of the basic setup of an atomic force microscope used to perform electric force microscopic measurements. Both EFM and AFM measurements were made simultaneously, by monitoring the detector signal amplitude at 300 x 300-pixel resolution, using silicon nitride

Figure 3. (a) Schematic diagram of the TopometrixTM setup for simultaneous non-contact topographic and EFM measurements. (b) The force versus distance plot indicates the different forces acting on the tip and their relative magnitudes at different tip-sample distances. Note that when the distance between the tip and the sample surface is higher than d0, the electrostatic attractive force predominates. (c) The schematic diagram shows the tip-sample distances to obtain topographic (d < d0) or electric force data (d > d0).

SEPM and EFM imaging of polymer surfaces

51

tips coated with platinum with 20 nm nominal radius (resonance frequency = 7099 kHz, stiffness constant = 1.8-5.2 N/m). 2.4. Scanning electric potential microscopy (SEPM) A common method to measure the contact potential difference between two materials is the vibrating capacitor or Kelvin method. The principle of the SEPM is similar to the Kelvin method except that forces are measured instead of currents [21, 22]. Scanning electric potential microscopy (SEPM) is then performed by applying a DC voltage between the tip and the sample while the cantilever hovers above the surface, without touching it. The cantilever deflects when it scans over the locally charged domains of the sample surface, and the magnitude of the deflection detected by the displacement of the sensor is proportional to the charge density [14, 15, 23]. In the case of SEPM, contrast is obtained when there are local electric potential gradients, across the sample surface. The images are obtained using an experimental arrangement in which the tip of the scanning force microscope is electrically biased as if it were a classical vibrating electrode in a monolayer surface potential measuring instrument. During the measurement, the resonant frequency of the cantilever is tracked by the four-quadrant photodetector and analyzed by two feedback loops. The first loop is used in the conventional way to move the scanner up and down, keeping a constant distance between the tip and the sample surface. The second loop is used to minimize the electric field between the tip and the sample. The second lock-in amplifier measures the tip vibration at the AC frequency oscillation while scanning, and adds a DC bias to the tip, to recover the undisturbed AC oscillation. The image is built using the DC voltage fed to the tip at every pixel [22]. The scheme of the basic setup of an atomic force microscope used to perform SEPM measurements is shown in Figure 4. SEPM imaging was done using the same tips and cantilevers as in AFM/EFM, and a DC voltage in the ±13 V range was applied between the sample and the tip of the microscope. 3. RESULTS AND DISCUSSION

The topographic image of titanium dioxide nanometric particles on mica (Figure 5 - top) shows that some of the particles are deposited adjacent to a step in the mica surface, the height of which is ca. 0.8 µm. In the SEPM image (Figure 5 - bottom) it is not possible to observe any difference in the electric force due to the mica step, but the sides of the oxide particles are perfectly visible. The outer layers of the oxide particle are positive, relative to the particle centers. This result shows an excellent separation of the contributions of the van der Waals forces and attractive electric forces acting on the tip. The electric map is thus completely independent of sample roughness.

52

E.F. de Souza et al.

Figure 4. Schematic diagram of the TopometrixTM setup for SEPM measurements.

The PS-HEMA latex forms ordered arrays in both the liquid and the dry states, very easily [24]. Figure 6 shows the AFM and EFM images of a self-assembled macrocrystalline film formed by PS-HEMA latex. In the EFM image, the particle shells appear positive, relative to their cores. While considering the EFM and SEPM images, we should recall that these are not strictly surface imaging techniques, because the electrostatic interactions are long-range. Consequently, the sampled depth is estimated as some tens of nanometers. The elemental distribution maps of these particles were previously obtained using energy-loss spectroscopy imaging in the transmission electron microscope (ELSI-TEM) [16, 17]. The dry particles contain sulfur-rich (from sulfate) cores and potassium-rich shells. Consequently, the EFM and ELSI-TEM results are in excellent agreement concerning the electrical characteristics of particle shells and cores. The presence of ionic charges in a latex is well acknowledged, but their location in the dry films has not been examined in detail, up to now [25]. The present results show that ions are clustered in dry latex, forming domains with excess charge, extending for tens of nanometers.

SEPM and EFM imaging of polymer surfaces

53

Figure 5. AFM (top) and SEPM (bottom) images of titanium dioxide nanometric particles on mica. The AFM image was rotated 90° counterclockwise and redrawn in perspective, to facilitate the observation of the TiO2 particles adjacent to the mica step.

54

E.F. de Souza et al.

Figure 6. AFM (top) and EFM (bottom) images of the self-assembled macrocrystalline film formed by the poly(styrene-co-2-hydroxyethylmethacrylate) (PS-HEMA) latex.

SEPM and EFM imaging of polymer surfaces

55

The AFM and SEPM images of the self-assembled film formed by poly(styrene-co-acrylamide) (PS-AAM) latex are presented in Figure 7. Particle cores are negative in relation to particle shells, as in the previous case of PSHEMA. This agrees with the known elemental distribution in these particles, as determined by ELSI-TEM [18]. In the natural latex film the situation is completely different: the topography image (Figure 8 - top) shows a continuous smooth film (maximum height = 12 nm) and without marked morphological features that could be identified as the original latex particles. On the other hand, the SEPM image (Figure 8 - bottom) shows a large contrast between neighboring domains, which can be associated with individual (darker, negative) latex particle cores, immersed in an electrically positive (bright) continuum. The SEPM image also shows some small brighter spots within the darker domains, which are probably related to latex non-rubber components [26]. Figures 9 and 10 present AFM and SEPM images of the surfaces of two regions of a film formed by a low Tg poly(styrene-co-n-butylacrylate) (PS-BA) latex. The topography images (Figure 9 and 10 - top) show continuous smooth films (maximum height = 34 nm), but with distinct morphological features. In both areas examined, there is correlation between the SEPM image and the topography observed, but in Fig. 9 it is possible to observe a sharp electric contrast between neighboring areas of the same morphological domain while the background is almost homogeneous. On the other hand, the topographic image in Fig. 10 presents roughly elliptical, negative domains, and an undulated background. Figure 11 presents the shadowed-AFM and EFM images of a high-density polyethylene film. The EFM contrast is very high, and both the upper and lower limits of the bias voltage scale were reached in a large number of pixels, which means that some parts of the image were saturated. Generally speaking, the surface protuberances appear as more negative than the depressions, but there are some areas in which this type of correlation is not observed. This is in agreement with observations for other polyolefins [27]. Both sides of a commercial polyester desk-jet printing film were imaged, as presented in Figures 12 and 13. At the printing side of this film there are many fuse-shaped elevations, pointing towards the right side of the picture. In the EFM image (Figure 12 - bottom), these elevations appear as if they were made of two parts, one of which to the left is more negative than the other. The printing side of this film is coated with a hydrophilic material, and the observed elevations are thus part of this material. The shape of the elevations suggests that the coating droplets spread during application, and the observed separation into domains of different electrical nature may be understood assuming that the coating is, e.g., made of a polyelectrolyte solution, in which the solvent liquid (or the serum) has a charge opposite to the polyelectrolyte. The backside of the polyester film is very smooth, both topographically (Figure 13 - top) and electrically (Figure 13 - bottom). Tiny shallow depressions appear as positive domains, but there is no correlation between the background topography and electrical contrast features.

56

E.F. de Souza et al.

Figure 7. AFM (top) and SEPM (bottom) images of the self-assembled film formed by poly(styrene-co-acrylamide) (PS-AAM) latex.

SEPM and EFM imaging of polymer surfaces

57

Figure 8. AFM (top) and SEPM (bottom) images of the film formed by the natural latex from Hevea brasiliensis (collected from a Rubber Research Institute of Malaysia 600 clone, film thickness = 250 nm).

58

E.F. de Souza et al.

Figure 9. AFM (top) and SEPM (bottom) images of the film formed by the non-dialyzed poly(styrene-co-n-butylacrylate) latex (film thickness = 185 µm).

SEPM and EFM imaging of polymer surfaces

59

Figure 10. AFM (top) and SEPM (bottom) images of the film formed by the non-dialyzed poly(styrene-co-n-butylacrylate) latex (film thickness = 185 µm).

60

E.F. de Souza et al.

Figure 11. AFM (top) and EFM (bottom) images of a high-density polyethylene film (film thickness = 93 µm). The AFM image was pseudo-illuminated at 90°, to facilitate the observation of the sample relief.

SEPM and EFM imaging of polymer surfaces

61

Figure 12. AFM (top) and EFM (bottom) images of the printable side of a commercial RetroplanTM polyester film (film thickness = 115 µm).

62

E.F. de Souza et al.

Figure 13. AFM (top) and EFM (bottom) images of the backside of a commercial RetroplanTM polyester film (film thickness = 115 µm).

SEPM and EFM imaging of polymer surfaces

63

4. CONCLUSIONS

The SEPM and EFM images of the various polymer samples examined reveal strongly contrasting image domains, with variable degrees of correlation with the respective topography images. This variability in the correlation between the imaging modes (topographic and electric) shows that there is a variable correlation between topographical and electrical features. The absence of correlation observed in some cases shows that these images are formed by independent factors. On the other hand, there is sufficient correlation in many image pairs, to allow for the identification of electrical features with corresponding topographical features. Generally speaking, the EFM images show more narrow features than SEPM, which may be an indication for the need to improve the feedback loop of the electric signal in SEPM. Of course, the detailed interpretation of all the systems examined in this work cannot be done without further additional characterization data. However, we believe that presenting these results at this stage may help motivate others to use the imaging capabilities of EFM and SEPM and other scanning probe techniques, and to develop the required interpretation tools. Progress is also required to obtain local electric charge densities from the electric force and electric potential maps, analogous to results already achieved in other force microscopy techniques, e.g. in topography-adhesion-stiffness imaging [28]. Finally, the present results show that neutral dielectrics are indeed made out of domains of different electrical nature, extending from a few nanometers to micrometers. Consequently, polymer electroneutrality is a result of charge balance but on a supramolecular or colloidal scale, not necessarily on the size scale of ion pairs or ion clusters. Acknowledgements The authors thank FAPESP, Pronex/Finep/MCT and CNPq. REFERENCES 1. S.-B. Lee, in: Polymer Interfaces and Emulsions, K. Esumi (Ed.), pp. 33–54. Marcel Dekker, New York (1999). 2. P. Robert, Electrical and Magnetic Properties of Materials. Artech House, Norwood, MA (1988). 3. K. Wu, M.J. Iedema and J.P. Cowin, Science, 286, 2482–2485 (1999). 4. G.L. Gaines, Jr., Insoluble Monolayers at Liquid-Gas Interfaces. Wiley-Interscience, New York (1966). 5. D.J. Shaw, Introduction to Colloid and Surface Chemistry. Butterworths, London (1966). 6. T. Hidaka, T. Maruyama, M. Saitoh, N. Mikoshiba, M. Shimizu, T. Shiosaki, L.A. Wills, R. Hiskes, S.A. Dicarolis and J. Amano, Appl. Phys. Lett. A 68, 2358–2359 (1996). 7. Y. Martin, D.W. Abraham and H.K. Wickramasinghe, Appl. Phys. Lett. 52, 1103–1105 (1988). 8. M. Nonnenmacher, O. Volter, J. Greschner and R. Kassing, J. Vac. Sci. Technol. B 9, 1358– 1362 (1991).

64

E.F. de Souza et al.

9. J.W. Hong, S.-I. Park and Z.G. Khim, Rev. Sci. Instrum. 70, 1735–1739 (1999). 10. F. Saurenbach and B.D. Terris, Appl. Phys. Lett. 56, 1703–1705 (1990). 11. G. Binning, C.F. Quate and C. Gerber, Phys. Rev. Lett. 56, 930–933 (1986). 12. R. Lüthi, H. Haefke, K.-P. Meyer, E. Meyer, L. Howald and H.-J. Güntherodt, J. Appl. Phys. 74, 7461–7471 (1993). 13. Y. Martin, C.C. Williams and H.K. Wickramasinghe, J. Appl. Phys. 61, 4723–4729 (1987). 14. R. Howland and L. Benatar, A Practical Guide to Scanning Probe Microscopy, Park Scientific Instruments, Sunnyvale, CA (1996). 15. S. Belaidi, P. Girard and G. Leveque, J. Appl. Phys. 81, 1023–1030 (1997). 16. A.H. Cardoso, C.A.P. Leite, and F. Galembeck, Langmuir 14, 4447–4453 (1998). 17. A.H. Cardoso, C.A.P. Leite, M.E.D. Zaniquelli and F. Galembeck, Colloids Surfaces A 144, 207–217 (1998). 18. E. Teixeira-Neto, C.A.P. Leite, A.H. Cardoso, M.C.V.M. Silva, M. Braga and F. Galembeck, J. Colloid Interface Sci. 231, 182-189 (2000). 19. F. Galembeck and E.F. Souza, in: Polymer Interfaces and Emulsions, K. Esumi (Ed.), pp. 119– 164, Marcel Dekker, New York (1999). 20. H. Bluhm, A. Wadas and R. Wiesendanger, Phys. Rev. B 55, 4–7 (1997). 21. G. Ertl and J. Küppers, Low Energy Electrons and Surface Chemistry, Verlag Chemie, Weinheim (1974). 22. M. Nonnenmacher, M.P. O’Boyle and H.K. Wickramasinghe, Appl. Phys. Lett. 58, 2921–2923 (1991). 23. R.M. Nyffenegger, R.M. Penner and R. Schierle, Appl. Phys. Lett. 71, 1878–1880 (1997). 24. A.H. Cardoso, C.A.P. Leite and F. Galembeck, J. Brazilian Chem. Soc. 10, 497–504 (1999). 25. M. Braga, C.A.R. Costa, C.A.P. Leite, M.C.V.M. da Silva and F. Galembeck, J. Phys. Chem. B 105, 3005–3001 (2001). 26. M.R. Sethuraj and N.M. Mathew, Natural Rubber: Biology, Cultivation and Technology, Elsevier Science, Amsterdam (1992). 27. A. Galembeck, C.A.R. Costa, M.C.V.M. da Silva, E.F. de Souza and F. Galembeck, Polymer, 42, 4845–4851 (2001). 28. Y. Zhang, A. Pungor, G. Jogikalmath and V. Hlady, J. Adhesion Sci. Technol., 14, 1751–1765 (2000).

Adhesion Aspects of Polymeric Coatings, Vol. 2, pp. 65–80 Ed. K.L. Mittal © VSP 2003

The residue (smut) formed on aluminum alloys during hydrofluoric acid etching and its effect on a coating process ANURAG P.S. TIHAIYA,1 JAMES P. BELL∗,1 and GUY D. DAVIS2 1

Institute of Materials Science, University of Connecticut, 97 N. Eagleville Road, Storrs, CT 06269-3136, USA 2 DACCO SCI Inc., 10260 Old Columbia Rd, Columbia, MD 21046, USA

Abstract—Hydrofluoric acid (HF) etching was studied as a pretreatment process for aluminum alloys. A coating process involving spontaneous polymerization on aluminum surfaces was used as a basis for the study. Five different aluminum alloys (2024, 3105, 5000, 6061 and 7075) were examined. The HF-treated aluminum surfaces were analyzed by means of XPS, AES and ESEM. HF etching resulted in differing behaviors among the alloys, in terms of extent of etching, surface morphology obtained after etching, and the quantity of residual black etching product on the surface. These factors led to different coating rates among the alloys. The residual etching product (“smut”) was observed to affect the polymerization rate on the surface and was necessary in some cases for coating formation. The smut consisted of hexagonal crystals and was identified as being primarily AlF3 · 3H2O. The presence of smut also affected the adhesion strength of the coating formed on the surface. Keywords: Hydrofluoric acid etching; aluminum pretreatment; residual product of etching; coating process.

1. INTRODUCTION

There is an abundant literature related to the surface pretreatment of aluminum. The subject has been the focus of numerous books and review articles [1-3]. However, studies of hydorofluoric (HF) etching of aluminum have not been as extensive. While some work has been done on the use of HF, alone or in solution with other acids, in aluminum pretreatment [4-7], the understanding of the effect of HF etching on various aluminum alloys is still limited. HF etching has been observed to result in formation of a residual product on the metal surface; some authors have reported formation of some kind of smut with alkaline [1, 8, 9] or acidic [2, 9] solutions. Little has been done to identify the smut material and un∗

To whom all correspondence should be addressed. Phone: (860) 486 4629, Fax: (860) 486 4745, E-mail: [email protected]

66

A.P.S. Tihaiya et al.

derstand its role in subsequent processing. The present work involves a study of etching by HF of different aluminum alloys with a focus on the smut. A novel coating process (S-Poly©) involving Spontaneous Polymerization [10, 11] on treated aluminum surfaces was used as a basis for the study. 2. EXPERIMENTAL

2.1. Materials/chemicals Five different aluminum alloys (AAs) were used for this study. The alloys were 2024-T3, 3105, 5000, 6061-T6 and 7075-T6, supplied by United Technologies’ Sikorsky Aircraft division. AAs 2024 and 7075 were Al-clad. AAs 7075, 6061 and 2024 samples were 1.0 mm thick. The thicknesses of AA 3105 and 5000 were 0.6 mm and 0.3 mm, respectively. Aluminum coupons used in the coating process were 7.5 cm x 2.5 cm in size. They were cut from the source metal using a shear cutter in the machine shop at the University of Connecticut. BlueGold®, an industrial alkaline cleaner, manufactured by Modern Chemical Inc. was used for cleaning aluminum substrates. The cleaner contained diethylene glycol n-butyl ether, and had a pH of 13 in its concentrated state. A diluted solution of the cleaner in tap water (10% by vol.) was used for cleaning. The pH of the dilute solution was ~ 12. The aluminum samples were cleaned with a Kimwipes® tissue which had been dipped in the dilute solution of the cleaner, and then rinsed under flowing tap water. The reagent grade HF acid (48-50%) used for preparing the etch bath was purchased from Fisher Scientific Company. N-phenyl maleimide (NPMI) (TCI America), bismaleimide (BMI) (TCI America), styrene (Aldrich Chemical Company) and 2 (methacryloyloxy) ethyl acetoacetate (MEA) (Aldrich Chemical Company) were the chemicals used as monomers in the S-Poly© coatings. N-methyl pyrrolidinone (NMP) (Lancaster Synthesis, Inc., Pelham, NH, USA) and water (distilled using a Barnstead Model 210 Distilled Water Center) constituted the solvent system for the coating. The chemicals were used as received without any further purification. 2.2. Pretreatment The HF etch bath was prepared by diluting 48-50% HF with distilled water to the desired concentration. The concentration levels used were: 2.5%, 5.0% and 7.5% (which were obtained by diluting the original 48-50% HF by 20, 10 and 6.67 times, respectively). The volume of a typical etch bath (prepared and used in a polypropylene container) was 350-400 ml. The dip time in the etch bath varied from 20 seconds to 5 minutes. The following sequence of steps was followed for pretreatment of substrate coupons with HF solution. (1) Degreasing with a Kimwipes® tissue dipped in 10% solution of an industrial cleaner, BlueGold®.

The residue (smut) formed on aluminum alloys

67

(2) Washing thoroughly (by wiping with a wet tissue) under running cold tap water with moderate flow for 1-2 minutes. (3) Wiping dry with Kimwipes® tissue. (4) Immersing in the HF solution (using stainless steel tweezers). (5) Rinsing under running tap water with or without wiping the surface with wet Kimwipes® for 15-20 seconds. (6) Rinsing with distilled water (using a wash bottle). (7) Drying in an air flow, using rubber tubing connected to a filtered and compressed air supply, for 45 sec-1 min. The HF etching resulted in the formation of smut on the treated surface, with the amount varying as a function of alloy and the etching conditions. The smut was loosely adherent and could visually be removed from the surface by wet wiping (step 5). To study the effect of smut, some of the samples were left with the smut intact while the others were wiped; thus some of the samples were only rinsed after the etch, while the others were wiped with wet Kimwipes® tissue during rinsing. 2.3. Metal surface study Treated aluminum surfaces were analyzed by means of X-ray Photoelectron Spectroscopy (XPS), Environmental Scanning Electron Microscopy (ESEM) and Auger Electron Spectroscopy (AES). The XPS unit (Surface Science Instruments, Model SSX-100-03) used a monochromatic Al Kα source. The X-ray spot was 600 µm; an electron flood gun was used for charge neutralization. The survey spectra pass energy was 150 eV. The AES analysis (Physical Electronic Industries, Model PHI 25-110 spectrometer) was conducted using a LaB6 source. The ESEM unit featured a LaB6 source gun and a Gaseous Secondary Electron Detector (GSED). Some of the AA 6061 samples with smut were dried in a laboratory hood at ambient conditions after being removed from the etching bath. The smut was scraped off the surfaces with a thin polycarbonate blade and was analyzed by means of wide angle X-ray diffraction (XRD) and Inductively Coupled PlasmaAtomic Emission Spectroscopy (ICP-AES). The X-ray diffractometer (Bruker, D5005) used a copper tube (λ = 1.5418 Å). The smut powder was dissolved in an acidic solution (2 parts nitric acid (70%), 1 part hydrochloric acid (35%) and 1 part distilled water) at ambient conditions. The solution was further diluted with distilled water (1:75) for the ICP-AES analysis. 2.4. Studies on coatings The pretreated samples were coated by polymerization on the samples (coupons) surface via the S-Poly© method [10, 11]. The S-Poly© process is a novel coating method in which Spontaneous Polymerization forms a coating on a metal substrate immersed into a solution of monomers. The reaction initiates without the help of any external initiator, and the polymer is formed on the metal surface but

68

A.P.S. Tihaiya et al.

not in the solution. The metal surface is required to initiate the polymerization. The polymer forms on the metal surface with the polymer chains growing out from the surface instead of being pre-formed and deposited on it. This facilitates conformal and uniform coatings on substrates of various shapes, with excellent adhesion and minimal defects. The process has been studied in detail with respect to its application to aluminum [10], steel [11] and copper [12]. It is known to be able to apply to zinc, tin, silver and other metals as well. The pretreated samples were weighed and immediately placed in the polymerization bath. There was little time between the removal from the etch bath and coating of the samples (3-4 min). The samples were handled with stainless steel tweezers and placed on a Kimwipes® sheet for weighing. The conditions used for polymerization, extraction of residual monomers from wet coating (rinsing) and drying the coating were based on the results of prior work [10]. The volume ratio of NMP and water in NMP/water solvent system was kept at 57/43. Concentrations of the monomers in the S-Poly© solution were NPMI (0.2 M)/Styrene (0.2 M)/MEA (0.1)/BMI (0.005 M). The pH of the polymerization bath was adjusted to 3 by adding 0.5 M sulfuric acid solution. The polymerization was carried out at ambient conditions. Coated samples were rinsed in an aqueous solution of NMP (10% vol.) for 10 min to remove the residual monomers and NMP from the coating. Rinsed samples were placed in a laboratory oven with air circulation for drying. The oven was preheated to 110ºC and coatings were dried at this temperature for around 1-1.5 hour. Then the temperature was increased to 220ºC and maintained at this level for 0.75-1 hour. The samples were allowed to cool to room temperature in laboratory conditions. The coating rates on the samples were the basis for most of the comparative studies carried out in this work. The measured coating rate was actually based on a standard polymerization time of 10 minutes. Based on prior work on the kinetics of the polymerization [10], the coating rate for a shorter polymerization time (less than 10 min) is expected to be linear. At longer times the process becomes diffusion controlled and varies with the square root of time. First, the sample coupons were weighed before coating (after air drying) and, later, they were weighed after the coating was dried. The coated area was measured and the coating rate, which was based on the dry weight, was calculated in g/cm2/sec. The weight loss during etching was obtained by the difference of the weights before and after etching (air drying). Since the coating is formed as a result of the growth of polymer chains on the metal surface, the coating rate depends upon the number of the active sites (initiation sites) on the metal surface. This is discussed in detail in prior work [10]. The practical adhesion of the polymer coatings to the aluminum substrate was measured using the torsional testing method of Bell and Lin [13]. In the test, aluminum joints were used. The coating-metal bond was made to fail in shear and the bond shear strength was calculated. By visual examination, the failure occurred in the coating-metal interphase.

The residue (smut) formed on aluminum alloys

69

The substrates were etched and coated using the standard procedure for SPoly© coatings [10]. The joint members were then adhered together by applying 0.045 g of an equimolar mixture of epoxy, Epon 828®, with methylene dianiline (MDA) curing agent, uniformly on the area of contact and cured at 120°C for 1 hour and at 150°C for 2 hours. The preparation of joints for the torsional tests has been described in detail by McCarvill and Bell [14]. For water immersion tests, the coated substrates were soaked in 60°C water for 3 days before testing. The testing was conducted on an Instron Universal Testing Machine (Model TM-S) at room temperature and humidity (lab conditions). A 4448 N load cell was used to measure the force at break. The crosshead displacement speed was 0.5 cm/min for all the tests. 3. RESULTS AND DISCUSSION

3.1. Effect of HF etching on various alloys The HF treatment was used with five alloys: 2024, 3105, 5000, 6061 and 7075. The compositions of the aluminum alloys are given in Table 1. AA 2024 and 7075 alloys were Al-clad, and that possibly was a reason that 2.5% HF was not an effective pretreatment for these two for the coating process. While the other alloys could be coated after etching in 2.5% HF, AA 2024 and 7075 alloy samples failed to coat after the treatment. Etching with 5% HF solution was found to be effective for all the alloys. Since fluoride ions are well known to induce an increase in the oxide dissolution rate in aluminum [15], a higher concentration HF (also lower pH) aids the process. The alloys, however, exhibited different coating rates; a comparison of the rates is shown in Figure 1. The coating rates presented in Figure 1 are derived from samples which were not wiped after HF etch. As shown, the order of alloys with respect to the coating rate is: 7075, 3105, 5000 > 6061 > 2024. Figure 2 presents the weight loss experienced by the alloys during HF etch. The weight loss for alloys 7075, 5000 and 3105 was higher than for 2024 and 6061. The difference among the etching rates of alloys was likely to have resulted from Table 1. Composition of the Al alloys used for the study Aluminum alloy 2024 3105 5000 6061 7075

Alloy elements, % (aluminum and impurities constitute remainder) Si

0.6

Cu

Mn

Mg

Cr

Zn

4.4

0.6 0.6 0–0.7

1.5 0.5 2.5–4.5 1.0 2.5

0–0.25 0.20 0.25

5.6

0.25 1.6

70

A.P.S. Tihaiya et al.

Figure 1. Coating rates for various Al alloys (non-wiped) after 1 min etch in 5% HF solution.

Figure 2. Weight loss of the alloys during 1 min etch in 5% HF solution.

the differences in the oxides present on various alloys. Different oxides would be expected to dissolve at different rates in the HF bath. In general, it appeared that a higher weight loss during etching (more etching) led to a higher coating rate. However, the variation in the coating rates could not be explained completely by the weight loss during etching. An interesting observation was that though both

The residue (smut) formed on aluminum alloys

71

Figure 3. ESEM pictures of alloys surfaces after 2 min of etch in 5% HF.

AA 7075 and 2024 were Al-clad, the two still exhibited different etching and coating rates. There was no apparent correlation between the bulk composition and the coating rate on the alloys. It is likely that other factors such as surface composition and morphology resulting from the HF etch also affected the rate of coating. Since a detailed surface analysis was beyond the scope of the project, the work to relate the surface composition of the alloys to their coating rate was not pursued. As mentioned above, the alloy samples used in this comparison were not wiped after etch. Some of the alloys, 6061 and 3105 in particular, had a layer of black

72

A.P.S. Tihaiya et al.

Figure 4. ESEM pictures of alloys surfaces after 2 min etch in 5% HF.

smut formed on their surfaces. It could be largely removed (visually) by wiping the surface with a wet tissue. The surfaces of alloys were observed under the Environmental Scanning Electron Microscope (ESEM) before and after etch to study the effects of the HF etch process. The effect of wiping the surface after etch was also studied. The smut was observed to be a collection of crystalline solid particles dispersed randomly on the surface. The ESEM pictures of surfaces of various alloys (non-wiped) after 2 min etch in 5% HF, are presented in Figures 3 and 4. As the pictures show, there were significant quantities of smut particles present on 6061 and 3105 alloys. Alloys 7075 and 5000 had a moderate number of these particles, while 2024 showed the smallest buildup. The surface of AA 5000 alloy was affected most by the HF etch; it showed a severely etched surface. The wiped surface of AA 7075 alloy, at a higher magnification, is shown in Figure 3. It had a granular appearance, and thus would be expected to have a high surface area. Observations of such granular appearances with F incorporation on the surface and their effect on practical adhesion have previously been reported [16-19]. 3.2. Effect of the smut The coating rates on wiped samples were compared with the rates on non-wiped samples (wiping usually removed the smut visually, but complete removal cannot be assumed). Figure 5 shows the coating rates on AA 2024. A summary of the coating rates on wiped and non-wiped samples is presented in Table 2. The results comparing the weight loss on wiped and non-wiped samples during their pretreatment are shown in Figures 6 and 7.

The residue (smut) formed on aluminum alloys

73

Figure 5. Coating rates on AA 2024 alloy for different etch times in 5% HF. Table 2. Coating rates (10-6 g/cm2/s) on wiped (W) and non-wiped (NW) samples Etch

2024 a

time

NW

W

NW

W

NW

W

NW

W

NW

W

20 sec 2 min 5 min

2.3 3.6 4

1.8 2.2 1.3

3.6 4.6 5.1

2.8 0 1.9

3.6 4.9 5.0

2.8 3.1 1.2

3.5 4.3 4.7

2.1 0 0

3.4 4.1 5.9

2.5 2.1 2.1

a b

6061 a

7075 a

3105 b

5000 b

- 5% HF etch - 2.5% HF etch

It was observed that the samples which had smut (non-wiped) exhibited higher coating rates than those without the smut (wiped). The data in Table 2 indicate that this difference was observed with all the alloys. The higher coating rates on the non-wiped samples suggest that presence of smut somehow helps the polymerization on the metal surface. The rate of coating on the non-wiped samples increased with etching time, likely because of an increasing amount of smut. The weight loss results on AA 7075 and 6061 alloys are presented in Figures 6 and 7, respectively. The weight loss for the samples with smut was lower than for samples without it. The difference in the weight losses is a measure of the amount of smut which was wiped away. The smut measurably contributed to the weight of the non-wiped samples. Clearly much more smut was present on AA 6061 than on AA 7075 alloy. More smut was formed at longer etch times and this led to higher coating rates. Although alloys 7075, 2024 and 5000 did not form as much smut as

74

A.P.S. Tihaiya et al.

Figure 6. Weight loss for wiped and non-wiped AA 7075 alloy samples for different etch times in 5% HF.

Figure 7. Weight loss for wiped and non-wiped AA 6061 alloy samples for different etch times in 5% HF.

The residue (smut) formed on aluminum alloys

75

Figure 8. ESEM picture (x2500) of AA 6061 alloy, wiped after 5 min of etch in 5% HF.

6061 and 3105, the difference between coating rates of the wiped and non-wiped samples was still significant, suggesting that even a small amount of smut can make a marked difference. No trend was observed among the coating rates on wiped samples. One possible reason could be the inherent inconsistency in the sample wiping technique. Wiping is not likely to remove the smut completely from the surface; the variation in the quantity of smut left on the surface after the wiping can explain the results. Some unexpected results were observed with wiped samples when different etching times were used. For most of the wiped samples, the coating rate after a 5 min etch was less than the rate after a 2 min or a 20 sec etch. This was not expected; a better etch resulting from a longer etch time was expected to lead to a higher coating rate. However, that was not observed. One infers that in the absence of smut, higher etching (more surface area) may not necessarily result in a higher coating rate. The AA 6061 samples wiped after 2 min of etch coated poorly while the samples wiped after 20 sec or 5 min of etch were coated easily. The ESEM studies showed that AA 6061 samples etched for 2 min had little or no smut left on them after wiping. On the other hand, AA 6061 samples wiped after 5 min of etch gave a different result. A layer of crystals, which could not be removed by wiping, was observed embedded into the surface (Figure 8). The presence of smut particles on the surface assisted the polymerization on the metal surface.

76

A.P.S. Tihaiya et al.

The AA 3105 samples wiped after 2 min and 5 min of etch did not coat. The ESEM studies on 3105 samples showed that wiping after etch effectively removed the smut. Here again, a lack of smut particles seemed to have caused the failure to coat. Interestingly, the critical role of smut in facilitating polymerization was observed more with AAs 3105 and 6061, which generated more smut than the other alloys. Smut, as is discussed below, was found to contain metals such as Fe and Cu which favor the coating (S-Poly©) process [10, 11]. During etching, these metals tend to concentrate in the smut. As a result of a longer etch, the surfaces of the alloys are likely to become depleted of these trace metals as they go into the smut. This is one possible explanation why wiped AA 3105 samples could be coated after a 20 sec etch but could not be coated after a longer etch. Wiped samples of AAs 2024, 7075 and 5000 could be coated, but their coating rate was significantly lower than that of non-wiped samples. The difference in the coating rates on wiped and non-wiped samples underlines the importance of the smut in the polymerization process. As mentioned before, wiping may not remove smut completely. Also the tendency of smut to adhere to the surface becomes an important factor as well. In case of a sticky smut, the samples may have some smut left on them even after wiping and give variable results. Similarly, during rinsing and drying steps of the non-wiped samples, some of the smut may shed off the surface (caused by water and air-flow), affecting the rate of coating. The practical adhesion of coatings obtained on AA 6061 substrates after HF etch was measured. A comparison of practical adhesion of coatings on wiped and non-wiped samples is shown in Figure 9. The coating on a non-wiped etched sample had a lower adhesion strength, likely due to the presence of smut on the surface. Smut particles are loosely adhered to the substrate and their presence in the coating-substrate interphase might have adversely affected the bond strength. The failure appeared to have occurred in the coating-metal interphase. To study the effect of water on the coatings, practical adhesion strengths of dry coatings were compared to those for the samples immersed in 60°C distilled water for 3 days. A comparison of bond strengths of an etched (non-wiped) and alumina blasted sample is presented in Figure 10. The effect of water was more severe on the non-wiped etched samples than that on the blasted sample, most likely due to the presence of smut particles on the metal surface. The presence of smut particles likely enhances diffusion of water within the interphase. The water within the interphase forms a weakened layer, resulting in loss of practical adhesion. The effect of water on coatings with smut could also be observed in the form of some dark spots which formed under the coating. 3.3. Study of the smut character The ESEM studies on AA 6061 and 3105 samples revealed that the smut crystals were hexagonal in shape (Figure 11). The size of the crystals (in their lateral di-

The residue (smut) formed on aluminum alloys

77

Figure 9. Shear strength of S-Poly coatings on HF etched AA 6061 alloy joints, with and without smut.

Figure 10. Comparison of the bond strength of coatings on AA 6061 alloys: dry vs. those immersed in water at 60°C for 3 days.

A.P.S. Tihaiya et al.

78

Figure 11. ESEM picture (x3000) of smut particles on AA 6061 alloy, after 5 min etch in 5% HF.

mension), as observed on etched AA 3105 alloy, increased with etch time. The smut collected from AA 6061 alloy was analyzed by wide angle XRD. The smut was identified as primarily tri-hydrated aluminum fluoride (AlF3 · 3H2O). The etched surface of AA 6061 alloy was analyzed by means of both XPS and AES. The analyses detected trace quantities of Fe and Cu along with high concentration of Al, F and O on the etched (wiped) surface. The XPS and AES results are summarized in Tables 3 and 4, respectively. Table 3. XPS analyses of etched Al 6061 samples Sample (Al 6061) As-received 5 min etch, non-wiped 5 min etch, wiped

% of the main elements detected on the surfaces of the samples Al

O

C

F

5.9 17.5

48.1 16.6

28.5 10.2

55.1

24.8

41.1

21.1

10

Mg

Si

12.9

2.8

Cu

Fe

Zn 0.5

1.0

0.5

0.3

The residue (smut) formed on aluminum alloys

79

Table 4. The elements detected by AES on the HF etched surface of Al 6061 alloy Sample (Al 6061)

Elements detected on the surface

As-received Non-wiped, 5 min etch Wiped, 5 min etch

Al, O, C, Mg, Ca, Cl Al, O, F, C Al, O, C, Mg, F, Cu, Fe, Zn, Ca

Table 5. ICP-AES results on Al 6061 alloy and the smut collected from the alloy Sample

Al 6061 Smut

% individual metal based on total metal content as 100% Al

Cr

Cu

Mg

Fe

Mn

98.2 95.55

0.20

0.17 0.95

0.98 0.6

0.32 0.95

0.12 1.95

The ICP-AES analysis of the smut from AA 6061 alloy also showed primarily Al with traces of Cu, Fe and Mn (Table 5). The concentrations of these trace metals in the smut were 3-5 times higher than those detected in the AA 6061 bulk. 4. CONCLUSIONS

The results of HF etching were observed to vary among aluminum alloys. The differences were in terms of the weight loss, the extent of coverage with the residual product (smut) and the surface morphology obtained after etching. This resulted, in turn, in the differences in coating rates obtained on the alloys. The smut was observed to have a significant effect on polymerization rate on the metal surface. In some alloys the presence of smut was found to be more critical for the coating formation than in others. The presence of the smut affected the adhesion strength of the coating, and resulted in lowering the adhesion strength further in the presence of water. The smut collected from AA 6061 alloy was found to consist of hexagonal crystals of tri-hydrated aluminum fluoride with traces of other metals such as Fe and Cu. Acknowledgements The authors appreciate the advice of Dr. T.R. Hanlon of United Technologies Corporation in this work.

80

A.P.S. Tihaiya et al.

REFERENCES 1. S. Wernick, R. Pinner and P.G. Sheasby, The Surface Treatment and Finishing of Aluminum and its Alloys; 5th ed., Vols. I & II., ASM International (1987). 2. J.D. Minford, Handbook of Aluminum Bonding Technology and Data, Marcel Dekker, New York (1993). 3. G.W. Critchlow and D.M. Brewis, Int. J. Adhesion Adhesives, 16, 255-275 (1996). 4. A.D. Yfantis and D.K. Yfantis, ATB Metall., 37 (2-4), 25-30 (1997). 5. F.J. Monteiro and M.A. Barbosa, Surface Coatings Technol. 35, 321-331 (1988). 6. J.W. Golby and J.K. Dennis, Surface Technol. 12, 141-155 (1981). 7. N.T. McDevitt, W.L. Baun and J.S. Solomon, WPAFB technical report, AFML-TR-75-122 (1975). 8. J.S. Solomon and W.L. Baun, in: Surface Contamination-: Genesis, Detection, and Control, K.L. Mittal (Ed.), Vol. 2, pp. 609-634, Plenum Press, New York (1979). 9. W.L. Baun, N.T. McDevitt and J.S. Solomon, paper presented at the Symposium on Surface Analysis of ASTM Committee E-2 on Emission Spectroscopy, Cleveland, Ohio (1975). 10. R. Agarwal and J.P. Bell, Polym.Eng. Sci, 38, 299 (1998). 11. X. Zhang and J.P. Bell, J. Appl. Polym. Sci., 66, 1667 (1997). 12. K. Nainani, Thesis, University of Connecticut, Storrs CT (1998). 13. J.P. Bell and C.J. Lin, J. Appl. Polym. Sci. 16, 1721-1733 (1972). 14. W.T. McCarvill and J.P. Bell J. Adhesion, 6, 185 (1974). 15. A.D. Yfantis and D.K. Yfantis, ATB Metall., 37, 25-30 (1997). 16. W.L Baun, WPAFB technical report, AFML Report TR-79-4172 (1979). 17. W.L Baun, WPAFB technical report, AFML Report TR-79-4178 (1979). 18. W.L Baun, WPAFB technical report, AFML Report TR-79-4155 (1980). 19. W.L. Baun and J.S. Solomon, WPAFB technical report, AFML Report TR-79-4165 (1979).

Adhesion Aspects of Polymeric Coatings, Vol. 2, pp. 81–99 Ed. K.L. Mittal © VSP 2003

Surface modification of metals by silanes DANQING ZHU and WIM J. VAN OOIJ∗ Department of Materials Science and Engineering, University of Cincinnati, Cincinnati, OH 45221-0012

Abstract—Various types of silane films on metals were evaluated in a series of ASTM tests for the purpose of corrosion protection. The silane films deposited on AA 2024-T3 were characterized using FTIR-RA. The results showed that: (1) some bis-type silane films, such as bis-[3triethoxysilylpropyl]tetrasulfide (bis-sulfur silane) provided both excellent bare corrosion protection and paint adhesion to different metals; (2) a novel water-borne silane mixture also offered outstanding corrosion protection and paint adhesion to metals, the performance of which was equal to the bis-type silane films; and (3) the structure of the bis-sulfur silane film was affected by factors such as the type of solvent in the silane solution and the nature of the metal substrate on which the silane film was deposited. An interfacial layer at the silane/metal interface was investigated using an FT-IR technique. This interfacial layer is believed to play a key role in the corrosion inhibitory performance of silane films on metals. Keywords: Silanes; Fourier-Transform Infrared Spectroscopy; AA 2024-T3; silane; coatings; corrosion.

1. INTRODUCTION

Silane treatment as a potential replacement for conventional chromating of metals has been studied extensively in our laboratory for several years [1–4]. Recently, bis-type silanes, such as bis-[triethoxysilyl]ethane (BTSE), bis-[trimethoxysilylpropyl]amine (bis-amino silane) and bis-[3-triethoxysilylpropyl]tetrasulfide (bissulfur silane), have become a new focus of our study due to their even better corrosion protection on a wider range of metals or alloys, as compared with monotype silanes. Bis-type silanes had been used primarily as crosslinkers for silane coupling agents until their anticorrosion behavior on a variety of metals or alloys was uncovered [1–6]. Bis-type silanes are usually applied onto metal surfaces from their water/alcohol solutions by a dipping process. The silane-treated metals are then dried in air. The crosslinking of the silane films on the metals in air is rapid, and can be accelerated at elevated temperatures (~ 100 °C). In a silane/metal system it is ∗

To whom all correspondence should be addressed. Phone: 513-556-3194, Fax: 513-556-3773, E-mail: [email protected]

82

D. Zhu and W.J. van Ooij

thought [7] that two types of chemical bonds are formed at the interface: (1) siloxane bonds (SiOSi) in the crosslinked silane film; and (2) metallo-siloxane bonds (MeOSi) at the silane/metal interface. Compared with mono-type silanes, bis-type silanes contain more silanol groups if the hydrolysis is complete. Both crosslinking density of the silane film and the number of metallo-siloxane bonds per unit area at the silane/metal interface could be higher. Consequently, an improvement in water resistance and adhesion of the metal to the silane film would be expected, assuming that other functional groups in bis-type silanes do not have detrimental effects. It was found in most cases [2, 3, 6] that bis-type silanes indeed gave a much improved paint adhesion to metal substrates, and, correspondingly, offered better corrosion protection of metals with and without painting. It should be pointed out that although bis-amino silane performs very well when used under paints, it cannot be used alone for the purpose of bare corrosion protection (i.e., without further painting). In a seawater immersion test (i.e., a 3.5% NaCl neutral aqueous solution), for example, AA 2024-T3 alloy coated with the bis-amino silane film applied from a 5% bis-amino water/methanol solution showed some local film delamination in the early stage of the immersion (within 1 day). A complete delamination occurred within 2 hrs of immersion under the harsher condition by adjusting the pH of the NaCl solution to 2, and severe corrosion in the delaminated area was evident soon after. This indicates that the hydrophilic secondary amino groups (-NH-) in the bisamino silane have a detrimental effect on the bare corrosion protection of metals. Subramanian and Van Ooij [2] observed a similar phenomenon with γ-aminopropyltriethoxysilane (γ-APS). They reported that amino groups in γ-APS preferred to bond to metal surfaces, which made the interface between γ-APS silane and metals more hydrophilic, leading to the delamination of the silane film from the substrates more easily by the attack of water. Such delamination, however, has never been observed for the other bis-type silanes. Besides the above bis-type silanes, a novel water-borne silane, a mixture of bisamino and vinyltriacetoxysilane (VTAS), has been developed recently in our laboratory. The major advantages of this silane mixture are listed as follows. (1) The hydrolysis of the mixture is instantaneous. The hydrolysis of VTAS is expressed as: H2C=CH-Si(Ac)3 + 3H2O ⇔ H2C=CH-Si(OH)3 +3HAc

(1)

The acetic acid (HAc) produced from reaction (1) makes the pH of the VTAS aqueous solution very low, e.g., pH 2-3 at a concentration of 5%. The solution at such low pH is unstable, and becomes hazy rapidly due to condensation of the silanol groups. The basic bis-amino silane is therefore mixed with VTAS in certain ratios, functioning as a pH adjuster. (2) The silane solution is almost alcohol-free, i.e., only de-ionized water (DI) water is used as the solvent. Only a small amount of methanol is formed from the rapid hydrolysis of the methoxy ester groups in the bis-amino silane.

Surface modification of metals by silanes

83

(3) The shelf lives of these aqueous solutions at certain mixing ratios are fairly long, more than 4 months. (4) The corrosion performance of this mixed silane film on several alloys equaled that of other bis-type silane films in a series of ASTM industrial tests. In order to obtain a better understanding of the anticorrosion behavior of these silane films, characterization studies were carried out. Fourier-Transform infrared spectroscopy in a reflection absorption mode (FTIR-RA) was employed extensively in this work. The effects of solvent type and metal substrates on the resulting silane film structures were investigated, as well as the interfacial layer suggested in silane/Al system. 2. EXPERIMENTAL

2.1. Materials a) Silanes. Silanes and their structures used in this work are listed in Table 1. Table 1. Silanes and their structures Silane

Structure

Bis-[triethoxysilyl]ethane (BTSE) Bis[3-(triethoxysilyl)propyl]tetrasulfide (Bissulfur silane)

(H5C2O)3Si(CH2)2Si(OC2H5)3 (H5C2O)3Si(CH2)3S4(CH2)3Si(OC2H5)3 (subscript 4 = the average number of sulfur atoms per molecule) (H3CO)3Si(CH2)3NH(CH2)3Si(OCH3)3

Bis(trimethoxysilylpropyl)amine (Bis-amino silane) Vinyltriacetoxysilane (VTAS)

(CH3COO)3SiCH=CH2

The bis-amino and bis-sulfur silanes (Silquest® A1170 and Silquest® A1289, respectively), were provided by Witco Co. (Tarrytown, NY), and VTAS was purchased from Gelest Co. (Tullytown, PA). They were used without further purification. To hydrolyze the bis-sulfur and bis-amino silanes, DI water and/or alcohol were added to the unhydrolyzed silanes. In a 5% bis-sulfur water/ethanol silane solution, for example, the volume ratio between the bis-sulfur silane, DI water and ethanol was 5:5:90. The natural pH of a 5% water/methanol bis-amino silane solution is around 10.8. The solution at such high pH would gel within a few minutes. Acetic acid, therefore, was added to lower the pH of the bis-amino silane solution to 7–8. Preparation of water-soluble silane mixture of bis-amino and VTAS was much simpler than the above silanes. For a 2% solution, 2 parts of the silane mixture of bis-amino and VTAS were simply added into 98 parts of DI water, followed by stirring for a certain period until the solution became clear. Ta-

D. Zhu and W.J. van Ooij

84

ble 2 shows the solvents commonly used and the pH values of the silane solutions. Table 2. Silanes, solvents and the pHs of the silane solutions Silane

Bis-polysulfur silane Bis-amino silane Mix (Bis-amino silane+VTAS)

Ratio of Solvents

pH

(v/v)

2%

5%

Ethanol/DI water: ~ 90/5 Methanol/DI water: ~ 90/5 DI water only

Natural pH5-6

Natural pH=5-6

pH=7-8 (adjusted by acetic acid) Natural pH=4-5 (at the mixing volume ratio of bis-amino to VTAS of 3:2)

pH=7-8 (adjusted by acetic acid) Natural pH=4 (at the mixing volume ratio of bis-amino to VTAS of 3:2)

Two solution concentrations were used in this study. 2% silane solution was used mainly for improving the paint adhesion to metal substrates and 5% for bare corrosion protection of metal substrates without further painting. 5% solution was also employed in the IR characterization work. b) Metals/alloys. Metals and alloys (15 cm ×10 cm × 0.07 cm) used in this work were as follows: ● 2024 Al alloy in T3 condition (AA 2024-T3, purchased from ACT Laboratories, Hillsdale, MI) ● Cold-Rolled Steel (CRS, purchased from ACT Laboratories, Hillsdale, MI) ● Hot-Dip Galvanized steel (HDG, provided by Bethlehem Steel Co., Bethlehem, PA); the Zn coating thickness was 20 µm. ● Galvalume ® (GVL provided by Bethlehem Steel Co., Bethlehem, PA); the 55% Al-Zn coating thickness was 7 µm. ● Mg alloy (AZ 91B, Norsk Hydro Research Centre, Porsgrunn, Norway) c) Paints. Polyester powder paint from Ferro Corp. and polyurethane powder paint from O'Brien Corp. were used in this study. The paint thicknesses were between 30 to 45 µm and were cured for approximately 15 min at 200 °C. d) Controls. Untreated and chromated panels were used as the controls in ASTM tests. The untreated panels were cleaned with an AC 1055® alkaline cleaner only; the chromated panels were treated with a chromate product with a trade name CHEM-COTE 3509® (Brent International, Lake Bluff, IL).

Surface modification of metals by silanes

85

e) Silane treatment of metals. The metals/alloys to be treated were cleaned with a diluted AC1055® (Brent International, Lake Bluff, IL) alkaline cleaner at 65– 75 °C for 3–5 min, followed by rinsing with tap water and drying with compressed air. The cleaned panels were dipped into the silane solutions for 30 s, and then blow-dried with air at room temperature. For the purpose of bare corrosion protection (without further painting), the silane-treated panels were cured at 100 °C for 10 min to obtain an extensively cross-linked structure in the silane films; otherwise the silane-treated panels were not heat-treated. 2.2. Test procedures a) ASTM tests. Three types of ASTM tests [8] were used to evaluate the corrosion performance of silane-treated metals and alloys in this work. ● ASTM B117 (Salt Spray Test): to evaluate the bare corrosion performance of silane-treated metals and alloys. ● ASTM E643-84 (Ball Punch Deformation Test): to evaluate the formability of silane films on metal and alloy sheets. ● ASTM D 1654-92 (Corrosion Test for Painted or Coated Metals and Alloys): to evaluate the corrosion performance of painted alloys with silane pretreatments. b) Characterization of silane films. The structures of silane/metal systems were investigated using FTIR-RA techniques. Experiments were carried out to study: (1) the effect of solvent type on the structure of the bis-sulfur silane film on AA 2024-T3; (2) the effect of metal substrates on the structure of the bis-sulfur silane film; and (3) observation of the interfacial structure in the bis-sulfur silane coated AA 2024-T3 system. FTIR-RA measurements were mainly conducted in the midIR range (e.g., 4000–400 cm–1) on a BIO-RAD infrared spectrometer. The resolution was 4 cm–1, and the number of scans was 256. An FTIR study was also done on a Nicolet® IR spectrometer in the far-IR range from 1800 to 150 cm–1, in order to investigate the interfacial layer in silane-treated metal systems. 3. RESULTS AND DISCUSSION

3.1. ASTM test results 3.1.1. Bare corrosion protection of silane-treated alloys Some bis-type silane films protect a variety of metal substrates against corrosion without topcoats. Empirically, it was found that a thickness of silane films above 200 nm was required for bare corrosion protection. The thickness values of silane films applied onto metals can be controlled by the silane solution concentration. This is because a linear relationship was consistently observed between the film thickness as measured by ellipsometry and the corresponding silane solution con-

86

D. Zhu and W.J. van Ooij

centrations [1, 6]. The optimum concentration for bare corrosion protection is about 5% for all silanes reported [1]. Figure 1 shows AA 2024-T3 panels after 7 days (168 hrs) of salt spray testing. It can be seen that the silane-treated AA 2024-T3 panels retained their original shiny surface with only slight edge corrosion after the test, while the untreated AA 2024-T3 corroded severely and the chromated alloy surface also showed slightly more corrosion than the silane-treated one. Figures 2 and 3, respectively, present bis-sulfur silane-treated GVL and HDG panels after the deformation test followed by the salt spray test according to ASTM E643-84 and ASTM B117. The aim of this work was to test the bare corrosion behavior of the deformed bis-sulfur silane films on alloys. Apparently, the deformed areas on both GVL® and HDG panels (i.e., the domes in the center of the panels) were protected by the silane films in comparison to the untreated panels. The silane-treated deformed area on the GVL® shows no corrosion; the silane-treated deformed dome on HDG has a few small pits. 3.1.2. Corrosion protection of silane-treated alloys under paints When silanes are used in the pretreatment before painting, the optimum concentration of silane solutions in this application normally is 2% [1]. Figures 4 and 5, respectively, present polyurethane (PU) and polyester (PET) powder-painted AA 2024-T3 panels after 1008 hours of salt spray test (SST). The panels were subjected to different pretreatments (i.e., silane treatment and chromating) before painting and corrosion testing. The silane used here was a water-soluble mixture of bis-amino silane and VTAS. It is evident in Figures 4 and 5 that both silane treatment and chromating provide equally good paint adhesion and corrosion protection for the PET and PU painted AA 2024-T3 panels. The untreated AA 2024-

Figure 1. AA 2024-T3 panels after 7 days (168 hrs) of Salt Spray Test (SST); (a) untreated, alkaline cleaned only; (b) chromated (CHEM-COTE 3509); and (c) mix (bis-amino silane + VTAS) pretreated, (Bis-amino/VTAS=1.5/1(v/v), 5%, natural pH 3.8).

Surface modification of metals by silanes

87

T3, on the contrary, shows a large degree of delamination with the PU paint and a complete delamination from the PET paint. PET powder-painted GVL panels were also subjected to salt spray test (SST), as shown in Figure 6. The silane treated GVL panel presents a slight local paint delamination at one end of the scribe line on the surface after 1008 hours of SST exposure, whereas the blank panel (i.e., untreated) totally failed in the test, showing a complete paint delamination and severe corrosion of the substrate nearby the scribe line.

Figure 2. GVL panels after deformation test followed by SST for 48 hrs; (a) untreated, alkaline cleaned only; (b) bis-sulfur silane treated, (5%, natural pH 5-6).

Figure 3. HDG panels after deformation test followed by SST for 48 hrs; (a) untreated, alkaline cleaned only; (b) bis-sulfur silane treated, (5%, natural pH 5-6).

88

D. Zhu and W.J. van Ooij

Figure 4. Polyurethane powder painted AA 2024-T3 panels after 1008 hrs of SST; (a) untreated, alkaline cleaned only; (b) chromated (CHEM-COTE 3509); and (c) mix (bis-amino silane + VTAS) pretreated, (bis-amino silane/VTAS=1.5/1(v/v), 2%, natural pH 4-5).

Figure 5. Polyester powder painted AA 2024-T3 panels after 1008 hrs of SST; (a) untreated, alkaline cleaned only; (b) chromated (CHEM-COTE 3509); and (c) mix (bis-amino silane + VTAS) pretreated, (bis-amino silane/VTAS=1.5/1(v/v), 2%, natural pH 4-5).

3.2. Characterization of silane films on metals 3.2.1. The effect of organic solvents on the film structure of bis-sulfur silane deposited on AA 2024-T3 It has been reported [9] that the hydrolysis of silanes in protic solvents, such as water and ethanol, approaches equilibrium rather than completion, due to the competition between silane hydrolysis and alcoholysis. The hydrolysis process in non-protic solvents such as dioxane, on the contrary, tends to completion. In the case of the bis-sulfur silane, we also observed a similar phenomenon [5, 10]. In

Surface modification of metals by silanes

89

Figure 6. Polyester powder painted GVL panels after 1008 hrs of SST; (a) untreated, alkaline cleaned only; and (b) bis-sulfur silane pretreated, (2%, pH 5-6).

this work, one critical issue that needed to be clarified is whether or not the solvent type would further affect the corrosion performance of the resulting bissulfur silane films through influencing the hydrolysis process in the silane solution. To clarify this, FTIR-RA and salt immersion test were employed here. Two types of solvents studied here were ethanol and dioxane. The former is a commonly-used solvent for the bis-sulfur silane solution. Two kinds of bis-sulfur silane films were deposited on AA 2024-T3 from 5% ethanol-based and dioxanebased bis-sulfur silane solutions, followed by curing at 100 °C for 10 min, to make the films sufficiently crosslinked. The FTIR-RA measurements were then conducted on these two silane-coated AA 2024-T3 specimens, and the results are shown in Figure 7. The IR peak assignments are given in Table 3. In Figure 7, a broad band at around 3300 cm–1 appears for both silane films, indicating that there still exists a considerable amount of hydrogen-bonded SiOH groups in the films after curing at 100 °C for 10 min [10]. Two peaks, 2974 cm–1 and 953 cm–1, correspond to the undydrolyzed ester groups (-SiOC2H5). The one at 2974 cm–1 is due to the CH stretching vibration in the ester groups, and the other at 953 cm–1 is attributed to the SiO deformation bending mode in the ester groups. It is noted that for the ethanol-based bis-sulfur silane film these two peaks are pronounced. In the case of the dioxane-based bis-sulfur silane film, however, the peak at 2974 cm–1 has totally disappeared and that at 953 cm–1 has diminished. This difference indicates that the ester groups have been converted completely to silanols in the dioxane-based solution but not in the ethanol-based solution. In other words, a complete hydrolysis of the bis-sulfur silane can be obtained using dioxane as a solvent instead of ethanol. As also shown in Figure 7, only one broad and strong band is seen for the ethanol-based silane film at around 1130 cm–1, which corre-

90

D. Zhu and W.J. van Ooij

Figure 7. FTIR-RA spectra of ethanol- and dioxane-based bis-sulfur silane films on AA 2024-T3. The silane-treated specimens were cured at 100 °C for 10 min. before IR measurements; the concentration of bis-sulfur silane solutions was 5% with a natural pH 5-6; IR measurements were conducted on a BIO-RAD Win-IR spectrometer. Incident angle: 75°, resolution: 4 cm-1, and number of scans: 100. Table 3. Assignments of absorption bands for bis-sulfur silane films on AA 2024-T3 [11, 12] Absorption band (cm-1)

Assignment

3336, 3367 2974, 2924, 2886 1130, 1030 953 886

Si-OH (H-bonded) Si-O-C2H5 (C-H str.) Si-O-Si (Si-O asym.) Si-O-C (Si-O asym.) Si-OH (H-bonded)

sponds to SiOSi bonds. In the case the dioxane-based film, however, two intense peaks are observed at 1130 cm–1 and 1030 cm–1 for SiOSi bonds. The appearance of these double peaks for SiOSi bonds has been reported to be associated with long SiOSi chains in silane films [11, 12], which apparently can be formed in completely hydrolyzed films and not in partially hydrolyzed films. These results allow a conclusion that the type of organic solvent does affect the structures of the resulting bis-sulfur silane films. Figures 8a and b illustrate schematically the molecular structures of these two types of bis-sulfur silane films on the Al alloy, on the basis of the structural information extracted from the IR spectra discussed above. In the ethanol-based bis-

Surface modification of metals by silanes

91

sulfur silane film (Figure 8a), there still exists a large number of unhydrolyzed ester groups (OR) attached to Si and hydrogen-bonded silanols groups (SiOH). These unhydrolyzed ester groups would further hydrolyze to silanol groups in the existing film when in contact with water or moisture [10]. The silanol groups generated in the interfacial region might react with the oxide layer on the Al alloy surface to form an interfacial layer enriched with AlOSi bonds and those in the outer film might condense with themselves to form an extensively crosslinked three-dimensional siloxane network [10]. The structure of the dioxane-based bissulfur silane film on the Al alloy is given in Figure 8b. No unhydrolyzed ester groups are shown in the film, as the hydrolysis in the dioxane-based silane solution tends to completion. The silanol groups in the film can condense readily and completely provided that the curing time is long enough. It seems that the SiOH groups condense in the film to form long siloxane chains, as was supported by the appearance of the double peaks at 1130 cm-1 and at 1030 cm–1. The corrosion performance of these types of bis-sulfur silane films on AA 2024-T3 was evaluated by a salt immersion test (i.e., 3.5%NaCl aqueous solution, pH 7). After 7 days of immersion, no significant difference was seen between these two types of silane films (Figure 9). This indicates that different organic solvents do not have a noticeable effect on the corrosion performance, although they do affect the film structures by influencing the hydrolysis of the bis-sulfur silane in different ways. Based on this result, another interesting conclusion can be drawn here that a completely hydrolyzed silane solution is not necessary for optimum corrosion protection of AA 2024-T3. 3.2.2. The effect of metal substrates on the film structure of bis-sulfur silane In practice, it is often observed that the same silane film performs differently when deposited on different metal substrates. So in order to understand this phenomenon we characterized the bis-sulfur silane films deposited on four different metal substrates. The following common metals/alloys were selected here: AZ 91B (Mg alloy), CRS (Fe alloy), AA 2024-T3 (Al alloy) and HDG (Zn-coated steel). All the metals were treated with the same 5% bis-sulfur silane water/ethanol solution, followed by curing at 100 °C for 10 min before FTIR-RA measurements. The results are shown in Figure 10, and the corresponding IR peak assignments are similar to those given in Table 3. In the SiOH region (3500–3000 cm–1 in Figure 10), a broad band is seen around 3300 cm–1, indicating that H-bonded SiOH groups still exist in the bis-sulfur silane films formed on all the alloys. The intensity of the band, however, depends strongly on the metal substrate. The intensity in the spectrum of the film formed on AZ 91B (Mg alloy) is the smallest; while the intensity in the spectrum obtained from HDG (Zn-coated steel) is the greatest. Obviously, the film on the AZ 91B has been almost fully crosslinked after cured at 100 °C for 10 min, as the very broad band at 3392 cm–1 has diminished compared with the others. The film on HDG, on the contrary, is the least crosslinked because a considerably intense broad band is seen at 3350 cm–1. Further, in the siloxane range (1200–1000 cm–1

92

D. Zhu and W.J. van Ooij

Figure 8. Schematic structure of the two types of bis-sulfur silane films on Al alloy; (a) ethanolbased bis-sulfur silane film; and (b) dioxane-based silane film.

in Figure 10), only one broad and intense band around 1130 cm–1 due to SiOSi bonds is observed for the silane films on CRS, AA 2024-T3 and HDG; whereas double peaks at 1030 cm–1 and at 1130 cm–1 are seen for AZ 91B. As discussed earlier, the appearance of these double peaks for SiOSi is associated with long chains of SiOSi bonds [11, 12]. Apparently, the structure of the bis-sulfur silane film deposited on the Mg alloy (AZ 91B) is different from the others. The one on the Mg alloy is characteristic of long chains of SiOSi bonds. This phenomenon may be explained by taking into account the highly basic nature of magnesium oxide (MgO).

Surface modification of metals by silanes

93

Figure 9. 7-day seawater immersion test on different types of bis-sulfur silane films on AA 2024T3; (a) Ethanol-based bis-sulfur silane film on AA 2024-T3; (b) Dioxane-based bis-sulfur silane film on AA 2024-T3.

Figure 10. FTIR-RA spectra of ethanol-based bis-sulfur silane films on different metals; (a) HDG; (b) AA 2024-T3; (c) CRS; and (d) AZ 91B. The silane-treated specimens were cured at 100 °C for 10 min. before IR measurements; the concentration of bis-sulfur silane solutions was 5% with a natural pH 5-6.

94

D. Zhu and W.J. van Ooij

It is known that the equilibrium pH of MgO is ~ 11, i.e., very basic in nature [13]. At pH < 11, MgO undergoes acidic dissolution according to the reaction, MgO + H2O ⇔ Mg2+ +2OH–

(2) 2+

It is noted that dissolution of MgO not only generates magnesium ions (Mg ), but also produces hydroxide ions (OH–). In this work, the pH of the 5% water/ethanol bis-sulfur silane solution was 5–6, thus some of the MgO layer was very likely to dissolve during the process of silane treatment, resulting in a high pH region close to the metal surface, as shown schematically in Figure 11a. Both hydrolysis and condensation would be accelerated by a base-catalyzed mechanism in this high pH region. As a result, long SiOSi chains were formed in the resulting silane films, as evidenced by the appearance of the double SiOSi peaks for AZ 91B in Figure 10. Grubb observed a similar effect caused by alkali metal hydroxides [14]. He stated that the condensation of trimethylsilanol in methanol was catalyzed by alkali metal hydroxides. The rate of condensation was dependent on the concentration of hydroxide anion, but not on the alkali metal cation. In contrast with MgO, the other oxides, such as Al2O3, ZnO, and FeO, are much more stable at neutral pH. Therefore, oxide dissolution is not expected on the other alloy surfaces. In the case of AA 2024-T3, for example, a hydroxide layer containing Al hydroxyls (AlOH) is formed (Figure 11b). The as-generated AlOH groups would react with SiOH groups in the silane solution, forming Al-siloxane (AlOSi) later on. Figures 12a and b show schematically the interfacial structure of silane/metal systems, on the basis of the structural information extracted from the FTIR-RA spectra in Figure 10. Figure 12a illustrates that the interfacial structure of the bissulfur sialne treated Mg system is affected by the local alkalization discussed above. The new SiOSi peak at 1033 cm–1 in Figure 10 clearly reflects this pH effect on the structure. As a comparison, Figure 12b depicts the interfacial structure of the silane/Al system. AlOSi covalent bonds form at the silane/Al interface as a result of the condensation between SiOH groups and AlOH groups at the alloy surface. A small amount of H-bonded SiOH and AlOH may also exist at the interface as a consequence of the equilibrium [7]. 3.2.3. Observation of the interfacial layer in silane/metal systems It was found in our recent electrochemical impedance spectroscopy (EIS) work with a non-corrosive electrolyte (0.5M K2SO4) [10] that there existed three time constants in the EIS spectra for a bis-sulfur silane coated AA 2024-T3 system. These three time constants were then assigned to, from outside to inside, the outer silane film, a new interfacial layer between the silane film and the oxide layer, and the inner oxide layer. FTIR-RA measurements also showed some evidence which supported this suggestion [10]. It was further speculated that the interfacial layer was very likely the product of the reaction (1) between AlOH groups on the AA 2024-T3 surface and the SiOH groups in the silane film, and (2) among the SiOH groups in the interfacial region. This layer was, therefore, assumed to be composed of aluminum silicates ([Al2O3•XSiO2]) [10, 15].

Surface modification of metals by silanes

95

To further clarify this inorganic interfacial layer, the FTIR-RA spectra of two types of silane films on AA 2024-T3 were obtained in the far-IR range (2000 to 200 cm–1), as shown in Figure 13. The silane films were deposited on AA 2024-

Figure 11. Schematic of the interfaces between metals and bis-sulfur silane water/ethanol solution; (a) MgO/silane-solution; (b) Al2O3/silane-solution.

96

D. Zhu and W.J. van Ooij

Figure 12. Schematic of the interfacial regions between metals and bis-sulfur siloxane network; (a) Mg /siloxane-network; (b) Al/siloxane-network.

T3 from the 5% silane solutions of bis-amino and bis-sulfur silanes. The silane films were cured at 100 °C for 10 min before the IR measurements. In Figure 13 two broad but weak bands in the region from 1200 cm–1 to 800 cm–1 are observed for the untreated AA 2024-T3, which are associated with the Al oxide layer on the alloy surface. The AA 2024-T3 substrate coated with the bis-amino silane film shows a strong and broad band in the region from 1200 cm–1 to 1000 cm–1, known as the SiOSi region, and a medium one around 500 cm–1. The AA 2024-T3 panel coated with the bis-sulfur silane film exhibits similar results: one intense peak at 1100 cm–1 indicating SiOSi bonds and one medium peak around 500 cm–1. As no peaks were observed for the untreated AA 2024-T3 at 500 cm–1, the peak around 500 cm–1 for both silane films on AA 2024-T3 is thus attributed to the silane films.

Surface modification of metals by silanes

97

Figure 13. FTIR-RA spectra of untreated and silane-treated AA 2024-T3; (a) AA 2024-T3 untreated; (b) treated with bis-sulfur silane (5%, natural pH 5-6); and (c) treated with bis-amino silane (5%, pH adjusted to 7 with acetic acid). The silane-treated specimens were cured at 100 °C for 10 min. before IR measurements; IR measurements were conducted on a Nicolet spectrometer.

Since few peaks are normally observed for polymeric materials below 600 cm–1, the peaks at around 500 cm–1 are probably due to an inorganic region formed in the silane film and/or at the silane/metal interface [16]. The authors in Ref. [15] assumed that some bands corresponding to metallosiloxane (MeOSi) bonds also appeared in the region from 1200 cm–1 to 1000 cm–1. It is, however, difficult to extract such information from Figure 13, as the 1200– 1000 cm–1 region is dominated by the bands corresponding to SiOSi bonds. The bis-sulfur silane solution was further diluted from 5% to lower concentrations, in order to deposit thinner films on AA 2024-T3. More specific information on the interfacial region is thus possible to be extracted from the spectra of the thinner films. Figure 14 compares the FTIR-RA spectra of bis-sulfur silane films on AA 2024-T3 alloy applied from the concentrations of 5%, 1%, 0.5%, and 0.1%. Some weak peaks in the case of 5% can be distinguished clearly at low concentrations (1–0.1%) (Figure 14). In the region from 550 cm–1 to 400 cm–1, three medium peaks have become distinct with decreasing concentrations. These peaks may again be associated with the inorganic layer at the silane/metal interface. The peaks could be attributed to the Si-O deformation bending mode for SiOSi in inorganic silica or silicates [16]. We thus tentatively postulate an interfacial structure of the organic/inorganic silicates of the type [Al2O3•XSiO2], in which X is an organic group, such as –(CH2)3, still attached to the Si atom.

D. Zhu and W.J. van Ooij

98

Figure 14. FTIR-RA spectra of ethanol-based bis-sulfur silane films on AA 2024-T3 applied from silane solutions with different concentrations; (a) 0.1%; (b) 0.5%; (c) 1%; and (d) 5%. All specimens were polished and finished with 1 µm alumina paste; the silane-treated specimens were cured at 100 °C for 21 hrs. before the measurements.

4. CONCLUSIONS ●

●

●

Bis-type silanes and a novel waterborne silane mixture, bis-amino and VTAS, provide excellent bare corrosion protection of and paint adhesion to a variety of alloys, as shown in a series of industrial ASTM tests. The type of organic solvent in the bis-sulfur silane solution has a remarkable effect on the hydrolysis of the silane solution. The FTIR-RA results indicated that the hydrolysis of bis-sulfur silane can be accelerated using dioxane as a solvent instead of ethanol. The structure of the bis-sulfur silane films and their corresponding corrosion performance are also influenced by the solvent type in the bis-sulfur silane solution. The ethanol-based bis-sulfur silane film shows only a partly crosslinked structure due to the incomplete hydrolysis in the ethanol-based solution, while that of dioxane-based film has a fully crosslinked structure as a result of the complete hydrolysis in the dioxane-base solution. The difference in the corrosion performance of these two types bis-sulfur silane films on AA 2024-T3, however, was insignificant. This indicates that a complete silane hydrolysis in the silane solution is not necessary for bare corrosion protection of metals.

Surface modification of metals by silanes

●

●

99

Metal substrates, on which the bis-sulfur silane film is deposited, have a significant effect on the film structure. In the case of the Mg alloy (AZ 91B), the basic MgO dissolves partly in the bis-sulfur silane water/alcohol solution. The as-generated OH– ions increase the local pH close to the alloy surface; as a result, both hydrolysis and condensation of the bis-sulfur silane are accelerated by a base-catalyzed mechanism. The less basic oxide layers on the other alloys, i.e., AA 2024-T3, HDG, and CRS, only form hydroxide layers in the presence of water or moisture, which do not vary the pH nearby the alloy surfaces. Therefore, the bis-sulfur silane films formed on these alloys are similar in structure. The existence of an inorganic interfacial layer at the silane/metal interface was concluded from FTIR studies in both mid-and far-IR ranges.

Acknowledgment The authors are grateful for the financial support provided by Brent International PLC in Bletchley, UK, the use of salt spray facilities at Brent International in Lake Bluff, IL, and the assistance of Thomas Rose for the far-IR measurements at Chemetall/GmbH in Frankfurt, Germany. REFERENCES 1. W.J. van Ooij and T.F. Child, CHEMTECH, 28, 26 (February 1998). 2. V. Subramanian and W.J. van Ooij, Corrosion 54, 204 (1998). 3. V. Subramanian, Ph.D. Thesis, University of Cincinnati, Department of Materials Science and Engineering (1999). 4. W.J. van Ooij, D. Zhu, Guru Prasad, Senthil K. Jayaseelan, Yuan Fu and Niranjan Teredesai, Surf. Engg. 16, 386 (2000). 5. Th. Van Schaftinghen, M.S. Thesis, Vrije Universiteit Brussel, Department of Materials Science and Electrochemistry, Brussels, Belgium (1999). 6. Guru Prasad Sundararajan, M.S. Thesis, University of Cincinnati, Department of Materials Science and Engineering (2000). 7. E.P. Plueddemann, Silane Coupling Agents, 2nd Edition, Plenum Press, New York (1991). 8. R.F. Allen and N.C. Baldini (Eds), 1998 Annual Book of ASTM Standards, ASTM, West Conshohocken, PA (1998). 9. F.D. Osterholtz and E.R. Pohl, in: Silanes and Other Coupling Agents, K.L. Mittal (Ed.), p. 119, VSP, Utrecht (1992). 10. W.J. van Ooij and D. Zhu, Corrosion, 57, 413 (2001). 11. G. Socrates, Infrared Characteristic Group Frequencies, John Wiley & Sons, New York (1994). 12. M.W. Urban, Vibrational Spectroscopy of Molecules and Macromolecules on Surfaces, John Wiley & Sons, New York (1993). 13. E. Ghali, in: R.W. Revie (Ed.), Uhlig’s Corrosion Handbook, 2nd ed, John Wiley & Sons, New York, p. 793 (2000). 14. W.T. Grubb, J. Am. Chem. Soc. 76, 3408 (1954). 15. A.M. Beccaria and L. Chiaruttini, Corrosion Sci. 41, 885 (1999). 16. Infrared Spectroscopy Committee of the Chicago Society for Paint Technology. Infrared Spectroscopy: Its Use in the Coatings Industry. Federation of Societies for Coatings Technologies, Philadelphia, PA (1969).

This page intentionally left blank

Adhesion Aspects of Polymeric Coatings, Vol. 2, pp. 101–119 Ed. K.L. Mittal © VSP 2003

Application of X-ray photoelectron spectroscopy in assessing the adsorption of siloxane polymers onto E-glass fibers LEANNE G. BRITCHER, DAVID KEHOE and JANIS G. MATISONS∗ Polymer Science Group, Ian Wark Research Institute Special Research Centre for Particle and Material Interfaces, University of South Australia, Mawson Lakes 5095, South Australia, Australia

Abstract—This paper focuses on the use of X-ray Photoelectron Spectroscopy (XPS) in establishing the adsorption of siloxanes and silanes onto E-glass fibers. When a siloxane or silane is adsorbed onto E-glass fibers, the O(1s) and Si(2p) binding energies in the XPS spectra do not shift or broaden significantly to allow identification of the silane/siloxane coating. We have found that the difference between the binding energies of the O(1s) and Si(2p) photoelectron lines, however, can be used to characterize the surfaces of siloxane coated E-glass fibers, as either resembling a siloxane (good coating) or the E-glass fiber (poor coating). The O(1s)-Si(2p) binding energy difference was also used to assess the adsorption of various functional siloxanes to E-glass fibers as a stringent consecutive washing procedure was applied to the surfaces using a variety of selected solvents. Siloxane coupling agent analogs were also prepared and they appeared to interact with the E-glass fiber surfaces in the same way as the analogous alkylsilane coupling agents. DRIFTS and SEM studies have also been used to support the XPS data. Keywords: Siloxane polymers; XPS; surface modification; E-glass fiber; siloxane adsorption.

1. INTRODUCTION

The XPS is a surface sensitive technique, which is capable of providing chemical information to a depth of only a few nanometers, and has been applied to a wide range of modified surfaces [1]. Although the XPS allows quantification of elements at the surface, the interpretation of such data in terms of the ‘degree of coverage’ or coating ‘thickness’ is the subject of much discussion in the literature [2-6]. There have been a number of studies, which have applied XPS analysis to modified E-glass fibers. In most cases, a semi-quantitative approach has been taken. Fagerholm et al. [7] have used XPS (together with streaming zeta potential and dynamic contact angles) to analyze the aqueous application of a surfactant, ∗

To whom all correspondence should be addressed. Phone: 61 8 8302 3162, Fax: 61 8 8302 3755, E-mail: [email protected]

102

L.G. Britcher et al.

alkylphenylpoly(oxyethylene)alcohol onto E-glass fibers. The oxygen elemental concentration, almost 50% in the clean fiber, was reduced to 21-24% by application of 0.2-2% solution of the surfactant. If the entire glass surface was being progressively covered by the surfactant, then oxygen percentage should decrease correspondingly. Fagerholm et al. [7] suggested that preferential adsorption onto some of the glass active sites occurred, followed by progressive multilayer adsorption onto the surface patches. Both streaming zeta potential and dynamic contact angle data support this hypothesis. Wang et al. [8] used XPS and Time-of-Flight Secondary Ion Mass Spectrometry (TOF-SIMS) to examine the interaction of hydrolyzed γ-aminopropyltriethoxysilane onto E-glass surfaces. In particular, they were interested in the effects of varying the pH and of treating the finished fibers with hot water. They determined that a pH of 9 resulted in maximum silane adsorption, and as the temperature of the water extraction increased, more adsorbed silane was removed from the surface. Interestingly, the observed aluminium concentration remained the same, as the amount of adsorbed silane increased, suggesting that either aluminium from the glass must have migrated into the silane layer, or the aluminium sites on glass were not covered. The former proposal was supported by the water extraction results, where the elemental concentration of Al was greater after extraction (~ 14%) than that for untreated glass (~ 8%). The Ca level was negligible, after warm water extraction, yet returned to almost 5% after hot water extraction, suggesting that some of the Ca was incorporated into the layer in a water soluble form, such as Ca(OH)2. Wesson et al. [9] used XPS and Electrolytic Thermodesorption Analysis of Water, Inverse Gas Chromatography & Programmed Thermal Desorption to examine the acid-base characteristics of silane treated E-glass fiber surfaces. Aqueous solutions of γ-aminopropyltriethoxysilane (APTES), γ-chloropropyltrimethoxysilane (CPTMS) and methyl-trimethoxysilane (MTMS) were applied to E-glass fibers, and angle resolved XPS was used to determine the orientation of the adsorbed silanes. For APTES, a small percentage of N in the form of NH3+, was found but was observed only at the largest penetration depth (take-off angle 90°). This indicates that only a small percentage of the amino groups are oriented towards the surface, and do in fact react with it. Similarly, the C elemental concentration decreased with increasing penetration depth, indicating that the propyl groups were also oriented towards the surface. The results for CPTMS were surprising, as both the Cl and C elemental concentrations increased with penetration depth. A significant amount of the silane must be oriented with the chlorine and propyl groups towards the glass surface. The aluminium elemental concentration suggested that the silanes did deposit preferentially on aluminols. Pantano and Wittberg similarly used XPS to examine the aqueous application of a number of silane coupling agents onto E-glass fibers [10] (upon silane treatment Ca and Al surface elemental concentrations were reduced). Pantano and Wittberg [10] suggested that the silanes uniformly covered the glass fiber surface, so the thickness, t, of the adsorbed silane film could be determined by:

Application of X-ray photoelectron spectroscopy

t = - λC ln [1 - IC / IC*] . 2/π

103

(1)

where IC is the measured carbon concentration on a particular silane-treated sample, IC* is the carbon concentration that would be measured in an infinitely thick film of that silane and λC is the mean free path of the C 1s photoelectron. Since calcium is present only in the E-glass fiber, the thickness of the overlayer was also determined by: t = - λCa ln [ICa / ICa*] . 2/π

(2)

where ICa is the measured Ca 2p concentration on a particular treated sample, ICa* is the Ca 2p concentration for a clean E-glass surface and λCa is the mean free path of the Ca 2p photoelectron in the silane overlayer. The factor 2/π in these expressions accounts for the curvature of the fiber, which can increase the effective path length of photoelectrons in the surface layer. Pantano and Wittberg [10] then calculated the silane film thickness using equation (2) for each silane, and found that the calculated result corresponded closely to the molecular length of each silane. Such methods make a number of assumptions in order to simplify the calculations. The uniformity of the coating is very much system dependent, and thus may present a real problem in applying these techniques to certain systems. An alternate way of assessing the “degree of coating” takes advantage of the accurately determined binding energies in XPS measurements. Wagner et al. [11] applied XPS to the analysis of 31 silicon-oxygen compounds, and found that the difference between the O(1s) and Si(2p) photoelectron line energies in inorganic silicon compounds ranged between 429.0-429.6 eV. However, for siloxane polymers, the line energy differences between 429.8-430.1 eV, were observed with an accuracy of 0.1 eV reported for the samples run on a non-monochromated source. This difference between the O(1s) and Si(2p) photoelectron line energies may, therefore, be useful to characterise how a silane or siloxane coating covers silicate surfaces. This method has a number of advantages. It is less affected by variations in adventitious carbon levels; it does not rely upon assumptions regarding the uniformity of the coating; and it can be accurately determined from the narrow scan multiplex spectrum of the sample. Previously we have presented the use of Diffuse Reflectance Infrared Fourier Transform Spectroscopy (DRIFTS) to monitor the adsorption of siloxanes onto Eglass fibers [12]. The objectives of this work were to determine if the O(1s)Si(2p) binding energy difference could be used to distinguish siloxanes on E-glass fiber surfaces. This paper discusses the analysis of functionalised siloxane polymers and silanes on E-glass fibers by using the O(1s)-Si(2p) binding energy difference in the XPS spectra.

104

L.G. Britcher et al.

2. EXPERIMENTAL SECTION

2.1. Materials Untreated E-glass fibers (water washed only) were obtained from ACI (Australia) (composition: 55% SiO2; 21.5% CaO; 14.5% Al2O3; 6.0% B2O3; 0.6% MgO; 0.9% Na2O and K2O; 0.4% Fe2O3 and 0.6% F2) and used without any further modification. Methylene chloride was dried with anhydrous CaCl2 for 24 hours and then distilled (Ace Chemicals, South Australia). Toluene (Ace Chemicals, South Australia) was distilled over sodium wire with benzophenone until a violet colour developed, indicating the formation of a ketyl radical. Hexane (Shell Australia), was distilled and stored over molecular sieves. Acetone (Chemplex), and AR grade ethanol (Merck) were used as supplied. The following functionalized siloxane polymers were prepared by Duplock et al. [13], from the direct hydrosilation of various poly(dimethylsiloxane-comethylhydrogensiloxane) fluids with the appropriate unsaturated reagents: poly[methyl(3-methacrylatepropyl)siloxane] fluids (degree of polymerization, D.P. = 22 and 33); poly[dimethylsiloxane-co-methyl(ketenimine)siloxane] (D.P. = 20; m = 16, n = 4); α,ω-[(3-amino-n-propyl)terminated-dimethylsilylpoly(dimethylsiloxane)] fluid (D.P. = 19 or 836) and poly[dimethylsiloxane-co-methyl(3aminopropyl)siloxane] (D.P. = 20; m = 16, n = 4). The synthesis of the siloxane couplings agents via hydrosilation of vinyltrimethoxysilane and vinyltris(2methoxyethoxysilane) onto poly(methylhydrogensiloxane) and poly(dimethylsiloxane-co-methylhydrogensiloxane) fluids has been outlined previously [14]. 2.2. Application procedures The application procedure consisted of dissolving the silanes or functionalized siloxane polymers in an appropriate organic solvent (see Tables 1 and 2), and placing this solution (~ 150 cm3) into a 250 cm3 glass vessel together with unsized Eglass fibers (~ 1 g). The glass fibers were left in this solution overnight (16 hours). The fibers were then removed from the solution, and washed with a series of organic solvents (see Table 2) and dried in an oven (3 hours at 100ºC in air). All treated fibers were analyzed by XPS, DRIFTS and SEM. 2.3. Diffuse reflectance infrared Fourier transform (DRIFT) spectroscopy DRIFT spectra were measured at room temperature with a single beam BIO-RAD Model FTS 65 spectrometer in the wavenumber region from 4000 to 400 cm–1, at a resolution of 4 cm–1, using an MCTA detector. The interferogram was apodized using the boxcar method installed in the FTIR software (BIO-RAD, 3200 Data Stations Users Manual, 0910260E, 1989). Signal-to-noise ratio was generally better than 100 to 1, using 256 scans.

Application of X-ray photoelectron spectroscopy

105

Table 1. Functionalized siloxanes applied to E-glass fibers Functionalized siloxane

Designation CAA1

CAA2

CAA3

CAA4

106

L.G. Britcher et al.

Table 1. (Continued) Functionalized siloxane

Designation methacryl 1 (n = 22) methacryl 3 (n = 33)

amino 1 (n = 19) amino 2 (n = 836)

amino 3

imino

Application of X-ray photoelectron spectroscopy

107

Table 2. Application procedures for the silanes and functionalized siloxanes Sample name

Siloxane or silane

Conc. (%)

Application procedure

Post-treatment (organic solvent wash)

VTMES1

VTMES

4.98

ethanol

VTMES2

VTMES

4.98

ethanol

VTMS

VTMS

3.40

ethanol

CAA1

CAA1

3.06

toluene

CAA2

CAA2

3.06

toluene

CAA3

CAA3

3.06

toluene

CAA4

CAA4

3.06

toluene

ethanol, methylene chloride (2x75 cm3 of each). methylene chloride, toluene, hexane, acetone, methylene chloride (100 cm3 of each). methylene chloride, toluene, hexane, acetone, methylene chloride (100 cm3 of each). methylene chloride, toluene, hexane, acetone, methylene chloride (100 cm3 of each). methylene chloride, toluene, hexane, acetone, methylene chloride (100 cm3 of each). methylene chloride, toluene, hexane, acetone, methylene chloride (100 cm3 of each). methylene chloride, toluene, hexane, acetone, methylene chloride (100 cm3 of each).

methacryl 1

methacryl 1

3.73

toluene

toluene (2 × 75 cm3)

methacryl 2

methacryl 1

3.73

toluene (2 × 75 cm3)

methacryl 3

methacryl 3

1.41

methacryl 1 + toluene + 0.7% ethylene diamine toluene

methacryl 4

methacryl 3

1.41

toluene (2 × 75 cm3)

amino 1

amino 1

1.70

methacryl 3 + toluene + 0.8% ethylene diamine toluene

amino 2

amino 2

5.20

toluene

toluene (2 × 75 cm3), methylene chloride (3 × 75 cm3)

amino 3

amino 3

4.10

toluene

toluene (2 × 75 cm3), methylene chloride (3 × 75 cm3)

imino

imino

0.94

toluene

toluene (2 × 75 cm3), methylene chloride (3 × 75 cm3)

toluene (2 × 75 cm3)

toluene (2 × 75 cm3), methylene chloride (3 × 75 cm3)

The fibers were mounted parallel to each other in a specially constructed sample holder [12], which was placed in a SpectraTech diffuse reflectance apparatus. Each glass fiber sample was scanned 256 times, with the fibers at an angle of 90°, with respect to the direction of the infrared beam, for maximum signal-to-noise

108

L.G. Britcher et al.

Figure 1. DRIFT spectrum of a clean E-glass fiber a) and DRIFT spectra of coated E-glass fiber samples b) VTMES 1 and c) VTMES 2.

ratio. The background spectrum was taken from high purity (IR) grade KBr powder (Merck), placed in a sample cup, and levelled to the top of the cup using a spatula. No pressure was applied to the KBr powder in the cup. DRIFT spectra, such as those in Figure 1, were obtained by subtracting the B-O overtone at 2671 cm–1 of the E-glass fibers from the same overtone on the treated glass fibers, using the BIO-RAD software. 2.4. X-ray photoelectron spectroscopy (XPS) XPS spectra were obtained using a Perkin-Elmer PHI 5600 XPS system with a concentric hemispherical analyzer and Mg Kα non-monchromated X-ray source operating at 300 W. Pass energies of 89 and 18 eV were used for survey and elemental spectra, respectively. Atomic concentrations were obtained using standard sensitivity factors [15]. The angle between the sample surface and the analyzer was fixed at 45°. The pressure during analysis ranged from 10–8 to 10–9 Torr. A five minute Argon ion etch was applied after the initial XPS analysis; the sputter rate was calibrated from the complete removal of a Ta2O5 film on tantalum (5 nm thick). All binding energies (B.E.s) were referenced to the C1s neutral carbon peak at 284.6 eV, to compensate for the effects of surface charging.

Application of X-ray photoelectron spectroscopy

109

2.5. Scanning electron microscopy (SEM) A Cambridge stereoscan 100 SEM was employed in this study to routinely examine the modified E-glass fiber surfaces. The treated fibers were coated with a thin (~ 20 nm) evaporated carbon layer to reduce the effects of charging. 3. RESULTS AND DISCUSSION

3.1. Adsorption of siloxane and silane coupling agents When assessing the adsorption of silane and siloxane coupling agents onto Eglass fibers it is difficult to determine the extent of coverage by the binding energies and elemental concentrations in the XPS spectra. Even if there are present other elements besides oxygen, silicon and carbon, it is not possible to determine if the coating is ‘patchy’ or extensively covers the surface. As discussed earlier, Wagner et al. [11] found that the difference between the O(1s) and Si(2p) binding energies could be used to distinguish between silicate and siloxane compounds. This difference maybe useful in characterising the surfaces of silane/siloxane coated E-glass fibers as either resembling that of a siloxane (indicating extensive coverage), or an inorganic silicate (indicating ‘patchy’ coverage). We have been investigating the usefulness of the O(1s)-Si(2p) binding energy difference on a number of different silane/siloxane treated E-glass fibers. When the binding energy difference is ≥429.8 eV, the glass is thought to have mainly a silicone coating; a difference of 429.7 eV is interpreted as being in between a silicone and silicate, while a difference of ≤429.6 eV is interpreted as a ‘patchy’ coating or not detected (based on SEM results). In order to determine the error involved in these measurements we have done triplicate measurements on treated Eglass fibers. Three XPS spectra were obtained on three different days with a fresh sample of the fiber taken for each measurement. The binding energy values did not vary more than 0.2 eV while the O(1s)-Si(2p) binding energy difference error value was <0.1 eV [16]. In assessing the adsorption characteristics of siloxanes upon E-glass fibers, it is useful to compare them with the adsorption of a commercial silane coupling agent. Therefore, vinyltris(2-methoxyethoxy)silane (VTMES) was applied to Eglass fibers in ethanol as outlined in Table 2. Sample VTMES 1 was washed with ethanol and methylene chloride, while sample VTMES 2 was sequentially washed with a series of organic solvents: methylene chloride, toluene, hexane, acetone and methylene chloride. The solvents were chosen in order to ensure that the most physisorbed silane/siloxane would be removed. The methylene chloride, toluene and hexane have an increasing order of ability to remove siloxanes (formed during condensation of silane with the glass surface), based on the difference in Hildebrand solubility parameter (Table 3). The smaller the difference in the solubility parameter with that of the polymer, the more soluble that polymer will be in the solvent [17]. Wet acetone (i.e. not dried) is reported also to remove any re-

110

L.G. Britcher et al.

Table 3. Hildebrand solubility parameter Polymer or solvent

Solubility parameter (MPa)1/2

Poly(dimethylsiloxane) methylene chloride toluene n-hexane acetone ethanol

14.9 19.8 18.2 14.9 20.5 26.0

maining physisorbed siloxane from the surface [18] and was also used in the washing sequence. The physisorbed oligomers can chemisorb (by formation of siloxane bonds to the glass surface) upon oven drying at 100°C. The O(1s)-Si(2p) binding energy difference obtained from the XPS spectrum of E-glass fibers is 429.2 eV, which clearly indicates a silicate compound. When the E-glass fibers are coated with a silane (VTMES 1), the O(1s)-Si(2p) binding energy difference is 429.9 eV, after washing (VTMES 2) the difference is 429.3 eV. The XPS results indicate that the surface of the treated glass fiber sample VTMES 1 resembles that of a siloxane, but the surface of the other sample resembles clean E-glass fibers (429.2 eV). Similar results were also obtained for E-glass fibers coated with vinyltrimethoxysilane (VTMS, see Table 2). The DRIFT spectra of VTMES 1 and 2 coated E-glass fiber surfaces are shown in Figure 1 spectra a and b, respectively. A full discussion of the DRIFTS technique has been presented previously [12]. Such spectra support the XPS results, since there are substantial C-H infrared vibrations present in sample VTMES 1, but almost none is observable for sample VTMES 2. The XPS and DRIFT spectra suggest that the silane coupling agent, in sample VTMES 2, is substantially removed after the extensive organic solvent washing procedure. Siloxane coupling agents were prepared by hydrosilating the silane coupling agents onto a siloxane backbone. The siloxane coupling agents were applied to Eglass fibers as indicated in Table 2. Figure 2 shows the O(1s)-Si(2p) binding energy differences for the coated E-glass fiber samples. It is clear from these XPS results that when the coupling agent analogs bearing trimethoxysilane functional groups are applied from toluene and subjected to the rigorous washing procedure (CAA1 and CAA2), the siloxane is removed from the glass surface or remains in patches. The O(1s)-Si(2p) binding energy differences for CAA1 is in the range for a silicate rather than a siloxane surface, confirming that there is no significant siloxane coating remaining on the glass fibers. While the O(1s)-Si(2p) binding energy difference for CAA2 is higher than for CAA1 but it still is not high enough to indicate complete siloxane coverage. The O(1s)-Si(2p) binding energy difference for coupling agent analog CAA3 bearing trimethoxyethoxysilane functional groups was in the range for a silicate

Application of X-ray photoelectron spectroscopy

111

Figure 2. O(1s)-Si(2p) binding energy differences for E-glass fibers coated with the siloxane coupling agents. Where = binding energy (B.E.) difference before a 5 minute Argon etch and = binding energy difference after the samples were etched for 5 minutes by Argon.

surface, indicating it had been substantially removed after washing. In contrast, CAA4 was not removed from the surface even after washing, reflected in the O(1s)-Si(2p) binding energy difference (429.8 eV). The polymer contained Me2SiO groups, which are randomly dispersed thoughout the polymer. These groups will minimise the number of adjacent trialkoxy groups enabling better interaction with the surface, as there will be less self-association of alkoxy groups within the polymer. The XPS spectra of the coupling agent analogs have given us information on the extent of coating; DRIFTS and SEM data can also be used to confirm such results. Figure 3 spectra a and b show the DRIFT spectra obtained for E-glass fibers coated with the siloxane coupling agents CAA1 and CAA2, respectively. The DRIFT spectrum of CAA1 (spectrum a) adsorbed onto glass fibers shows that no siloxane is present on the glass surface, as the siloxane C-H stretching vibrations are not visible. A subtraction technique to remove the background glass fiber peaks was used to examine in detail the spectral region between 3100 cm–1 and 2700 cm–1 for CAA2. The technique involves subtraction of the clean glass fiber DRIFT spectrum from the DRIFT spectrum of the coated glass fibers by matching the area and intensity of the B-O overtone. Subtraction of the B-O overtone for the spectrum of CAA1 did not reveal any further information. The subtraction spectrum of CAA2 (spectrum b) displays three C-H stretching vibrations between 2842 cm–1 and 2963 cm–1. At 2842 cm–1 the symmetric stretch from unhydrolyzed OCH3 groups is observed, while at 2963 cm–1 the asymmetric stretch for the CH3 groups on the siloxane backbone occurs; both of these stretches are present in the neat siloxane. The stretch at 2924 cm–1 is assigned to the C-H asymmetric stretch

112

L.G. Britcher et al.

for the CH2 groups in CAA2; this stretch is not detected in the neat siloxane spectrum. The DRIFT spectrum for CAA2 adsorbed onto glass fibers has, therefore, shown some siloxane is present on the surface, however not in significant amounts, thereby confirming the O(1s)-Si(2p) binding energy differences observed. Figure 4 shows the DRIFT spectra (a and b) obtained for E-glass fibers coated with the siloxane coupling agents CAA3 and CAA4, respectively. The DRIFT spectra show that the copolymer with Me2SiO groups, which are dispersed in the

Figure 3. DRIFT spectra of coated E-glass fiber samples a) CAA1 and b) CAA2.

Figure 4. DRIFT spectra of coated E-glass fiber samples a) CAA3 and b) CAA4.

Application of X-ray photoelectron spectroscopy

113

copolymer (CAA4), has interacted with the surface better than CAA3, as the spectrum of CAA4 shows strong absorbances in the C-H region whereas these stretches are hardly visible in the spectrum of CAA3. The C-H stretches in spectrum b are assigned as follows: at 2964 cm–1 and 2869 cm–1 are the asymmetric and symmetric CH3 stretches respectively associated with the siloxane backbone; at 2925 cm–1 is the asymmetric CH2 stretch, and at 2817 cm–1 is the C-H stretch associated with the silicon alkoxy group. The SEM micrographs of the coupling agent analogs on the E-glass fiber are shown in Figures 5a-c. There is little surface morphology, with only a few ‘bubble or patch-like’ formations on the surface (less than 0.5 µm in diameter) and some occasional siloxane bridges (some >5 µm long) between adjacent fibers on the otherwise generally smooth surfaces of CAA1, CAA2 and CAA3. These SEM micrographs confirm the DRIFTS data suggesting that some siloxane is on the surface, but only rare patches are present, and therefore the O(1s)-Si(2p) binding energy difference shows a predominantly silicate rather than siloxane surface. The SEM micrographs of the E-glass fibers treated with CAA4 are displayed in Figures 6a and 6b. The surface morphology consists of both small (<0.3 µm) and large bubbles (~ 5 to 6 µm), with more frequent bridges (some >5 µm long) also between adjacent fibers. These micrographs indicate not only that substantial siloxane is on the fiber surfaces (confirming DRIFTS and XPS data), but the coat-

Figure 5. SEM micrographs of coated E-glass fiber samples a) CAA1 b) CAA2 and c) CAA3.

114

Figure 5. (Continued).

L.G. Britcher et al.

Application of X-ray photoelectron spectroscopy

115

Figure 6. SEM micrographs of coated E-glass fiber sample CAA4; the top (a) micrograph shows bridges formed between the fibers, while the bottom (b) micrograph shows the siloxane encapsulating the fibers.

116

L.G. Britcher et al.

ing is not uniform (hence the bubbles on the surface). CAA4 not only encapsulates the fiber surfaces, but also can link adjacent fibers through siloxane bridges. XPS, DRIFTS and SEM have shown that the nature of the siloxane backbone plays a significant role in the adsorption of coupling agent analogs onto E-glass fibers. 3.2. Adsorption of methacryl functionalised siloxane polymers Two different methacryl functionalized siloxane polymers were prepared and applied to E-glass fibers from toluene (methacryl 1, D.P. = 22 and methacryl 3, D.P. = 33). Ethylene diamine was also added to toluene solutions of methacryl 1 and methacryl 3, generating two further samples methacryl 2 and methacryl 4 respectively (see also Table 2). Blitz and coworkers [19] investigated the interactions of silanes and silica in the presence of various amines. They concluded that all amines catalyzed the interaction between silanes and the surface hydroxyl groups on silica. The ethylene diamine was added to determine the effect of increasing the pH on the adsorption of the methacryl siloxane polymers. Figure 7 displays the the O(1s)-Si(2p) binding energy differences for methacryl functionalized siloxanes on E-glass fibers. Samples methacryl 1 and methacryl 3 have binding energy differences typical of a siloxane polymer surface. The binding energy differences for methacryl 2 and methacryl 4 indicate that the presence of the ethylene diamine (see Table 2) has not enhanced but rather reduced the ad-

Figure 7. O(1s)-Si(2p) binding energy differences for methacryl functionalized siloxane polymers on E-glass fibers. Where = binding energy (B.E.) difference before a 5 minute Argon etch and = binding energy difference after the samples were etched for 5 minutes by Argon.

Application of X-ray photoelectron spectroscopy

117

sorption of the methacryl functionalized siloxanes. DRIFTS and SEM results, presented previously [12], did not indicate clearly whether the amine competed with the methacryl functionalized siloxane for adsorption sites. XPS has now shown that the presence of amine does interfere with adsorption as the O(1s)Si(2p) binding energy differences are lower for samples containing the amine (methacryl 2 and methacryl 4). 3.3. Adsorption of imino and amino functionalized siloxanes We have observed that amino acid functionalized siloxanes groups can adsorb effectively onto E-glass fibers [20] and there has been much discussion in the literature regarding the interaction of such functional groups attached to trialkoxysilanes with glass surfaces [21-22]. DRIFTS and SEM results, presented previously [12], did show adsorption of the siloxane polymers, but they were not able to determine if the siloxane polymers entirely covered the surface or were only present as patches which were observed in the SEM micrographs. Several siloxane polymers bearing amino or imino functional groups were prepared and applied to unsized E-glass fibers, in order to investigate how such functional groups interacted with glass surfaces. An imino functionalized siloxane based on a poly(dimethylsiloxane-co-methylhydrogensiloxane) polymer, with an average of 4 imino groups per molecule was prepared. Both amino 1 and amino 2 samples are aminopropyl-terminated poly(dimethylsiloxane) polymers of different molecular weights (amino 1 D.P. = 19; amino 2 D.P. = 836). Sample amino 3 is a pendant aminopropyl functionalized siloxane, with an average of 4 pendant amino groups per molecule. For the imino functionalized siloxane was prepared by reaction of amino 3 with a ketone. The O(1s)-Si(2p) binding energy differences for E-glass fibers coated with the imino and amino functionalized siloxanes are displayed in Figure 8. All of the amino siloxane samples gave O(1s)-Si(2p) binding energy differences typical for a siloxane surface. For the imino siloxane coated E-glass binding energy difference is 429.7 eV. Clearly the amino groups have adsorbed more effectively onto the glass fiber surface than the imino siloxane and are not easily removed after washing with toluene. 4. CONCLUSIONS

The O(1s)-Si(2p) binding energy difference obtained from the XPS spectra is a useful way to assess the surface coverage of a siloxane on E-glass fibers quickly and effectively. When the XPS data are combined with DRIFTS and SEM results, we have a powerful analytical technique for characterising the adsorption processes of siloxane polymers, bearing diverse functional groups, onto E-glass fibers.

118

L.G. Britcher et al.

Figure 8. The O(1s)-Si(2p) binding energy differences for imino and amino functionalized siloxane polymers on E-glass fibers. Where = binding energy (B.E.) difference before a 5 minute Argon etch and = binding energy difference after the samples were etched for 5 minutes by Argon.

Vinylsilanes do absorb onto E-glass fibers from solution but may be effectively removed if the fibers are subsequently washed with a series of appropriate solvents. When the vinylsilanes are attached to a siloxane polymer via hydrosilation, the siloxane backbone determines the effective adsorption onto E-glass fibers. Siloxane polymers bearing amino and methacryl groups remain on the E-glass fiber surface even after washing with toluene, indicating good adsorption onto the Eglass fiber surface. REFERENCES 1. D. Briggs and M.P. Seah (Eds), Practical Surface Analysis by Auger and X-ray Photoelectron Spectroscopy, Wiley, New York (1983). 2. T.L. Barr, Critical Rev. Anal. Chem. 22, 569 (1991). 3. O.A. Baschenko, J. Electron Spect. Rel. Phenom. 57, 297 (1991). 4. L.S. Johansson, Surf. Interf. Anal. 17, 663 (1991). 5. J.E. Fulghum, Surf. Interf. Anal. 20, 161 (1993). 6. K.I. Nishimori and K. Tanaka, J. Vac. Sci. Technol. A8, 3300 (1990). 7. H. Fagerholm, C. Lindsjö, J. Rosenholm and K. Rökman, Colloids Surfaces 69, 79 (1992). 8. D. Wang, F.R. Jones and P. Denison, J. Adhesion Sci. Technol. 6, 79 (1992). 9. S.P. Wesson, J.S. Jen and G.M. Nishioka, J. Adhesion Sci. Technol. 6, 151 (1992). 10. C.G. Pantano and T.N. Wittberg, Surf. Interface Anal. 15, 498 (1990). 11. C.D. Wagner, D.E. Passoja, H.F. Hillery, T.G. Kinisky, H.A. Six, W.T. Jansen and J.A. Taylor, J. Vac. Sci. Technol., 21, 933 (1982).

Application of X-ray photoelectron spectroscopy

119

12. L.G. Britcher, D.C. Kehoe and J.G. Matisons, Silanes and Other Coupling Agents, Vol 2, K.L. Mittal (Ed.), pp. 99-114, VSP, Utrecht (2000). 13. S.K. Duplock, J.G. Matisons, A.G. Swincer and R.F.O. Warren, J. Inorg. Organomet. Polym. 1, 361 (1991). 14. L.G. Britcher, D.C. Kehoe, J.G. Matisons and A.G. Swincer, Macromolecules, 28, 3110 (1995). 15. S. Hofmann, in Practical Surface Analysis, Second Edition, Vol 1 Auger and X-ray Photoelectron Spectroscopy, D. Briggs and M.P. Seah (Eds), John Wiley, New York (1990). 16. L.G. Britcher, J.G. Matisons and D.C. Kehoe, Unpublished results. 17. J.A. Brydson, Rubber Chemistry, Applied Science Publishers, London (1978). 18. E.P. Plueddemann, Silane Coupling Agents, 2nd ed, Plenum Press, New York (1991). 19. J. Blitz, R. Murthy and D. Leyden, J. Colloid Interface Sci. 126, 387 (1988). 20. A. Provatas and J.G. Matisons, Langmuir, 14, 1656 (1997). 21. D.W. Dwight, F.M. Fowkes and T. Huang, J. Adhesion Sci. Technol. 4, 619 (1990). 22. H. Ishida, in: Adhesion Aspects of Polymeric Coatings, K.L. Mittal (Ed.), pp 45-106, Plenum Press, New York (1983).

This page intentionally left blank

Adhesion Aspects of Polymeric Coatings, Vol. 2, pp. 121–136 Ed. K.L. Mittal © VSP 2003

Surface modification of polyphenylene sulfide plastics to improve their adhesion to a dielectric adhesive YING WANG∗ and STANTON RAK Physical Technology Center, Automotive and Industrial Electronics Group, 4000 Commercial Avenue, Motorola, Inc, Northbrook, IL 60062

Abstract—The adhesion strength of a perfluorinated dielectric adhesive to polyphenylene sulfide (PPS) was investigated. The effect of different fillers in the PPS as a function of plasma treatment conditions was evaluated. The change in adhesion as a result of thermal baking was also addressed. The surface composition and surface energy were monitored and systematically quantified by X-ray Photoelectron Spectroscopy (XPS) and contact angle measurements, respectively. The correlations between the presence of certain functional groups, change in surface energy and polarity, and variation in adhesion properties indicate that the adhesion mechanism is mainly due to van der Waals forces. Enhanced wetting at the adhesive/substrate interface and a deeper interfacial diffusion zone are found to be necessary conditions to achieve the optimal adhesion. Keywords: Perfluorinated adhesive; surface modification; RF plasma; polyphenylene sulfide plastics.

1. INTRODUCTION

PPS is a widely used plastic in commercial and automotive industries. It has excellent mechanical strength and resistance to different media due to its structural stiffness provided by aromatic rings. This stiffness, however, sometimes prevents good adhesion of PPS to other materials. In order to improve the adhesion of PPS and other polymer materials, several surface modification techniques have been developed including chemical treatment, corona, arc discharge and plasma etching. Plasma surface treatment is an effective and clean way of improving the adhesion of many plastics. Several mechanisms have been proposed to interpret this improvement. They are: removal of organic contamination from the surfaces; material removal by ablation to increase the surface area or to remove a weak boundary layer; crosslinking or branching to strengthen the surface cohesively; and surface chemistry modification to improve chemical and physical interactions at the ∗

To whom all correspondence should be addressed. Phone: (847) 480-4718, Fax: (847) 480-3064, E-mail: [email protected]

Y. Wang and S. Rak

122

bonding interface [1–3]. To determine the mechanisms behind the modification of the plastic, several surface sensitive techniques have been used including X-ray photoelectron spectroscopy (XPS) [4–6], infrared reflection absorption spectroscopy (IRRAS) [4], and contact angle measurements [2, 7]. It is generally accepted that the first step in the O2 plasma treatment consists of the formation of radicals on the surface layer by the elimination of hydrogen atoms, [8]. The radicals then react with oxygen and water molecules in air to produce functional groups such as alcohols, ketones and carboxylic acids, which are capable of interacting with an adhesive [7, 9]. It has also been reported that these functional groups normally have a short life-time because of the high mobility of plasma-induced, fragmented polymer chains on the surface. Therefore, to maintain good adhesion after plasma treatment, the effect of aging in different environmental conditions must be carefully monitored and controlled. It should be noted that depending on the nature of the bonding between the adhesive and substrate, the adhesion strength as well as the stability in different media, could be very different. In industrial applications, certain fillers including glass fibers and minerals are added into the pure polymer matrix to improve its thermal and mechanical properties. At the same time, the introduction of fillers may result in heterogeneity of surface morphology and chemistry and influence adhesion behavior and mechanisms compared to single phase plastics. The adhesive suppliers add adhesion promoters to promote adhesion with fillers such as SiO2. The mechanism for adhesion to PPS itself is less well understood and is of practical interest. This work describes the use of plasma treatment for improving the adhesion of a perfluorinated adhesive to PPS. The mechanism of this improvement was investigated using XPS and surface energy characterizations. PPS plastics with and without glass fillers and minerals were used. The effects of etching conditions and thermal annealing are also addressed. 2. EXPERIMENTAL DETAILS

2.1. Materials and sample preparation The chemical structure of PPS is shown below,

S n

polyphenylene sulfide (PPS) Three types of natural colored PPS samples from Ticona (Summit, NJ) were used to investigate the effects of fillers on adhesion. They were PPS without glass mineral additive (Fortronâ 0214 P1), PPS with glass fibers (Fortronâ 1140 L4),

Surface modification of polyphenylene sulfide plastics

123

and PPS with glass fibers plus mineral fillers (Fortronâ 4665 B6). The weight percentages of SiO2 fibers and calcium-containing minerals in each sample, obtained by thermogravimetric analysis at 800 °C, are listed in Table 1. As shown in the table, the major components of the three materials are very different. These differences also imply that the materials are formulated under different conditions, and likely with other minor additives. As a result, material properties that are directly and indirectly related to adhesion properties, such as density, glass transition temperature, molecular weight, and molding temperature, can also be different. Due to this complexity, it is very difficult to unambiguously identify the differences in bonding characteristics between materials. Therefore, the emphasis in this work was placed on the mechanism of surface modification and its effects on adhesion within the same material group. The PPS samples in the form of tensile bars were first rinsed in methanol in an ultrasonic cleaner for 5 minutes to remove possible surface grease and low molecular weight contaminants. The thermal treatments after plasma exposure were conducted at 85 °C. Table 1. Compositions of PPS materials (wt%) Brand name

Plastic

Filler

Others additives

Fortronâ 0214 P1 (natural color)

>98

<2 (no filler)

/

Fortronâ 1140 L4 (natural color)

~60

~40 (fibers)

Lubricants

Fortronâ 4665 B6 (natural color)

~35

~65 (20 fibers + 45 minerals)

/

A perfluorinated polymer adhesive from Shin Etsu America (Newark, CA) was cured at the 150 °C for 1 hr. The adhesion between the adhesive and PPS substrates, which was initially poor, could be improved by plasma treatment. Therefore, it provides an appropriate window for investigating the modification mechanisms of the plasma treated PPS surface. 2.2. Plasma treatment Plasma treatment was performed on Tegal Corporation’s Plasmaline 211 plasma etcher with 13.56 MHz RF frequency and an all-aluminum reactor. The samples were treated under different power and time conditions, in oxygen environment with the gas purity greater than 99.9%. The operating pressure was maintained at 2.66×102 Pa. The time lapse between the plasma treatment and applying the adhesive or performing surface analyses was controlled within an hour to minimize aging effects.

124

Y. Wang and S. Rak

2.3. X-ray photoelectron spectroscopy XPS analysis was performed using a Kratos Analytical XSAM 800 spectrometer employing an Al Kα (1486.6 eV) achromatic X-ray source operated at 10 kV with a total power of 150 W. The operating pressure was lower than 1.33×10-6 Pa. The pass energies for wide survey scans and single element scans were 320 eV and 20 eV, respectively. The spectrometer was calibrated to the 3d5/2 and 3p1/2 photopeaks of Ag at 368.3 eV and 573.2 eV, respectively, using a 99.99% Ag foil of 0.125 mm thickness. All binding energies were referenced to the main C-H and C-C peaks at 285.0 eV. The linear background subtraction and Gaussian line fitting were used in the elemental quantification process. Different samples treated under the same conditions were used in the XPS investigation. The angle between the sample surface and the analyzer (take-off angle) was 90° unless noted otherwise. 2.4. Adhesion measurements The adhesion was tested using pre-molded tensile bars of PPS plastics provided by Ticona. The adhesive was manually dispensed and cured at 150 °C/1 hour onto a 12.7 mm × 12.7 mm bond area with bondline thickness of 0.5 mm. The adhesion results, averaged from five test samples, were obtained from the maximum shear strength recorded by an Instron (Model 5500) with a crosshead rate of 5 mm/min. 2.5. Contact angle measurements The contact angle measurements were performed using diiodomethane (DIM) and freshly made de-ionized water on plastic surfaces after acetone was used to remove the possible contaminants such as grease and oil. The contact angles were recorded for 20 µl water drops on the test surface. The time between plsma treatment and contact angle measurements was controlled at approximately three minutes. 2.6. Environmental treatments The environmental treatments of PPS materials after plasma treatments included temperature (85 °C) and humidity (0 %RH and 85 %RH). The environmental exposure was performed before the adhesive was applied. 3. RESULTS

3.1. Adhesion The adhesion results after plasma treatment under different conditions are reported in this section. The same adhesive and test conditions were used for all

Surface modification of polyphenylene sulfide plastics

125

PPS-based materials. Under these conditions, the adhesive fails cohesively at 1.7– 1.9×106 Pa or higher. Therefore, any adhesion data below this range indicate a failure at the interface between the adhesive and PPS substrate. 3.1.1. Treatment parameters The lap shear strength of PPS 0124 P1 under different plasma conditions is plotted in Figure 1. It is observed that as the treatment parameters (time and power) increase from 200 W/5 min. to 300 W/10 min., there is little change in adhesion. All the failures visually appeared to be interfacial between the adhesive and the substrate. As the treatment time is increased to 300 W/20 min., improved adhesion is observed, and the failure mode changes to cohesive in the adhesive. Figure 1 also gives the adhesion results of PPS 1140 and PPS 4665 after similar treatments. Compared to PPS 0214, the adhesion for both PPS 1140 and PPS 4665 is poor before the plasma treatment. However, the adhesion was significantly improved after the treatment, even though the corresponding results are slightly lower than that of PPS 0214 treated under the same conditions. The plasma treatment improves the adhesion of PPS 1140 and PPS 4665 to a greater extent than that of PPS 0214 because of the low initial adhesion value in the PPS 1140 and PPS 4665. Because of its low energy level, plasma has stronger effects on plastic materials than on inorganic fillers, and may produce more radicals on the surface of PPS 0214 than on the surfaces of PPS1140 and PPS 4665. On the

Figure 1. Adhesion of the three PPS materials to perfluorinated adhesive under different plasma treatment conditions.

126

Y. Wang and S. Rak

other hand, the ablation effect of plasma would introduce a more dramatic change in the morphology on the surface of the composite materials (PPS 1140 and PPS 4665) than the single-phased material (PPS 0214). These effects are discussed further in the surface analysis results in the next section. 3.1.2. Time-temperature-humidity effects After plasma treatment (300 W/10 min.), the samples were baked under different humidity conditions (85 °C/85 %RH and 85 °C/0 %RH) for one day before the adhesive was applied and cured. The adhesion results are shown in Figure 2 together with the data from the control (untreated) and 300 W/10 min. plasma treated samples. A similar trend is observed for all three materials. The adhesion was greatly improved as a result of thermal treatment. The failure mode for all three materials changed from interfacial to cohesive after 85 °C baking. There is no significant difference in adhesion between PPS samples treated under 85 °C/85 %RH or 85 °C/0 %RH. The similarities of adhesion after thermal treatment in different humidity conditions exclude the critical contribution to adhesion from surface polar groups introduced by plasma, since humidity would have significant effect on altering the concentration of surface polar groups. Such an effect has been reported in the study of poly(dimethylsiloxane) (PDMS) material [10]. When the oxygen-plasma-treated PDMS surface was aged in air, the surface returned to a low energy state by reorienting its non-polar groups toward the interface. When

Figure 2. Temperature and humidity effects on adhesion of different PPS samples to perfluorinated adhesive.

Surface modification of polyphenylene sulfide plastics

127

aging was performed in an aqueous phase, the plasma treated surface, which had a high concentration of polar groups, maintained its polar groups at the surface. We attempted to induce a similar effect to PPS using 85 °C/85 %RH treatment but the adhesion did not improve. Therefore, we believe that polar groups on the surface may not have a significant effect on the adhesion of the PPS system. 3.2. X-ray photoelectron spectroscopy 3.2.1. Before plasma treatment The spectra from survey scans of the three PPS samples before any treatment are shown in Figure 3. For both PPS 0214 and PPS 1140, the spectra consist of X-ray photoelectron and Auger peaks from C, O and S. No Si was detected on the surfaces of PPS 1140 and PPS 4665, in which SiO2 was used as glass fibers. This could be attributed to the existence of a thin layer of plastic resin on the surface. The thickness of this layer is mainly a function of molding temperature, which may not be the same for different types of PPS plastics. The PPS 1140 resin contains a mold lubricant according to the supplier. The lubricant could be responsible for the poor initial adhesion of the PPS 1140 in comparison to the PPS 0214. As seen in Figure 3c, Ca-2p photoelectron peak was present on PPS 4665 before treatment. The presence of Ca on the surface could also influence the adhesion of PPS 4665 with the adhesive.

Figure 3. a) XPS survey scan of PPS 0124, b) XPS survey scan of PPS 1140, c) XPS survey scan of PPS 4665.

128

Y. Wang and S. Rak

The photoelectron peaks of C-1s, O-1s, S-2p for PPS 0214 are shown in Figure 4. After curve fittings and binding energy calibrations, the peak positions and relative atomic percentages for all three materials are listed in Table 2. Under the current energy resolution, both C-1s and O-1s photoelectrons show single peaks, while the doublet of S comes from S-2p1/2 and S-2p3/2. Also listed in the table are results from PPS 1140 and PPS 4665. As shown in the table, the major components for all three materials are very similar. In addition, Ca, added as a mineral, was present in PPS 4665. 3.2.2. After plasma treatment After oxygen plasma treatment for 5 minutes at 200 W, no new elements were detected. The photoelectron peaks for all elements from PPS 0214 are shown in Figure 5. The results of the detailed scan indicate the presence of a broad shoulder on the higher binding energy side of the C-1s peak, as well as the third peak at the higher energy side of the S-2p doublet. Table 3 documents the results of curve fittings and quantification for each element. Based on the energy shift of each peak with respect to the C-1s peak (285.0 eV), the assignments of these peaks are also included in the table. The C-1s peaks at around 288.0 eV and 289.5 eV are assigned to carbonyl bond, C=O, and carbon bonds in ester, O-C=O, respectively [11–13]. The sulfur peaks at higher binding energy side of the original S-2p doublet is attributed to the S−O bond in the form of sulfate [14]. Figure 5 and Table 3

Figure 4. XPS high-resolution spectra (take-off angle θ=90°) of PPS 0214 before plasma treatment.

Figure 5. XPS high-resolution spectra (take-off angle θ=90°) of PPS 0214 after 200 W/5 min. O2 plasma treatment.

Surface modification of polyphenylene sulfide plastics

129

Table 2. Surface composition of PPS materials before plasma treatment PPS 0214 P1 C O S1 (2p1/2) S2 (2p2/3) Si Ca

eV 285.0 532.6 163.7 164.9 / /

PPS 1140 L4 at% 83.9 3.1 8.3 4.7 / /

eV 285.0 532.9 163.7 164.9 / /

PPS 4665 B6 at% 83.9 3.9 7.6 4.6 / /

eV 285.0 532.7 163.7 164.9 / 348.0

at% 81.8 5.6 7.8 4.7 / 0.3

Table 3. Surface composition of PPS 0214 P1 after O2 plasma treatment (200 W/5 min.)

C1 C2 C3 O1 O2 S1 S2 S3 S4 S5

PPS 0214 P1 (θ=90°)

PPS 0214 P1 (θ=30°)

eV

at%

eV

at%

285.0 288.0 289.8 532.9 534.3 163.6 164.7 166.2 169.2 170.6

35.5 2.5 12.0 24.3 17.7 2.3 1.5 0.4 2.9 1.0

285.0 288.3 289.8 532.9 534.2 163.7 164.7 166.2 169.1 170.3

27.9 6.2 13.4 27.9 19.2 0.8 0.2 0.4 2.9 1.1

Peak assignment

C−C, C−H C=O O-C=O / / sulfide sulfide unknown sulfate/sulfone sulfate/sulfone

Figure 6. XPS high-resolution spectra (take-off angle θ=30°) of PPS 0214 after 200 W/5 min. O2 plasma treatment.

130

Y. Wang and S. Rak

indicate that oxygen plasma treatment increases the amount of oxygen while decreasing the amount of carbon on the surface. The changes in oxygen can be partially attributed to the presence of C=O, O-C=O, and S-O bonds as listed in Table 3. Figure 6 shows the XPS spectra of PPS 0214 after 200 W/5 min. treatment, with a take-off angle θ = 30°. More intense peaks are observed from oxygen related bonds (C=O, O-C=O and S−O). This indicates that the modification by plasma treatment primarily occurs at the surface. If the mean free path (λ) for the C-1s electron is taken as 1.4 nm [4], the sampling depth can be estimated as approximately 3λsinθ, where θ is the take-off angle. Therefore, the sampling depths for θ of 90° and 30° are 4.2 nm and 2.1 nm, respectively. This is consistent with the observation that plasma treatment affects only the top layer with a thickness of the order of 10 nm [2]. 3.2.3. Effect of treatment conditions After the samples were treated under different plasma conditions, the surface compositions for PPS 0214, PPS 1140 and PPS 4665 were determined and shown in Figure 7. The oxygen and sulfur peaks are categorized by their binding energies based on their curve fitting results. Quantitative calculations were performed on the elements in parentheses. It is seen that after plasma treatment, the total amount of oxygen increases considerably while the amount of carbon decreases. At the same time, XPS peak intensity contributed by sulfide decreases while the signals from sulfate (S−O or S=O) increase after the treatment. 3.2.4. Thermal treatment Figure 8 shows the surface compositions of the three samples after a 300 W/10 min. plasma treatment followed by thermal treatment (85 °C/0 %RH) for 24 hours. After 85 °C baking, the C=O (or O-C=O) peak in all three materials almost vanishes in the background while the S−O signal remains fairly strong. If the changes in adhesion shown in Figure 2 are compared, the improvement in adhesion after thermal annealing is not correlated with the presence of C=O or O-C=O functional groups on the PPS surface. 3.3. Contact angle analysis The results of the contact angle measurements averaged over five test spots on each sample are listed in Table 4 for both DI-water and diiodomethane (DIM), before and after oxygen plasma treatment. As shown in the table, before the treatment, the difference in contact angles among various materials is negligible, and plasma treatment greatly reduces the contact angles. Thermal baking at 85 °C for one day increases the contact angles for all three materials. Referring to the XPS data shown earlier, a direct correlation is observed between contact angle and surface radicals created by plasma treatment.

Surface modification of polyphenylene sulfide plastics

131

Table 4. Contact angles (°) on surface treated PPS materials PPS 0214 P1

PPS 1140 L4

PPS 4665 B6

H2O

DIM

H2O

DIM

H2O

DIM

As-received O2 plasma, 200 W/5 min.

80 20

21 18

82 35

38 26

79 8

30 17

Annealing treatment at 85 °C/1 day after 200 W/5 min. plasma treatment

57

27

55

41

31

29

Figure 7. Surface compositions of the three PPS samples after plasma treatments at different conditions.

132

Y. Wang and S. Rak

Figure 8. Surface compositions of the three PPS samples after plasma treatments and thermal annealing at 85 °C for one day.

4. DISCUSSION

In order to form an adhesive bond between an adhesive and a substrate, the initial establishment of interfacial molecular contact is necessary. The molecules on the surface undergo motions toward preferred configurations to achieve the adsorptive equilibrium. Next, the molecules diffuse across the interface to form a smeared interfacial zone, and/or react chemically to form primary chemical bonds across the interface. It has been well accepted that adhesion is a manifestation of the attractive forces that exist between all atoms and which fall into three broad categories: chemical forces, hydrogen bond forces and van der Waals forces.

Surface modification of polyphenylene sulfide plastics

133

In many plasma treated plastics where adhesion is due mainly to chemical forces, it has been reported that a plasma treatment time of 1–2 minutes or less under similar power as employed in this investigation is sufficient to create active bonds [3]. This is because chemical bonding can be initiated at a very sharp interface between the adhesive and the substrate. It has also been experimentally verified that a longer treatment time, or too much power, deteriorates the adhesion behavior of most plastics such as poly(ethylene terephthalate) (PET) [15], poly(etheretherketone) (PEEK) [16]. On the other hand, the optimized plasma treatment time for PPS has been reported to be much longer, typically 10–15 minutes [2–5]. In this work, no significant improvement in adhesion was recorded (Figure 1a) for the PPS without filler until a treatment time of 10 min. at a high power (300 W). XPS analysis shows that this optimized adhesion is not directly correlated to the amount of free radicals induced by plasma treatment. Actually, plasma treatment under lower power and shorter time (200 W/5 min) introduced the maximum amount of oxidized species while the adhesion hardly changed. In addition, thermal baking experiments indicate that the adhesion is improved while the amount of surface oxidation decreases. The fact that long treatment time and baking at high temperature improve adhesion suggests that the secondary forces, not the chemical ones, play a critical role in determining the adhesion mechanism in this system. It is also possible that the surface morphology changes significantly after long treatment time, but we do not have any definite evidence that surface morphological effects influence the adhesion of perfluorinated polymer adhesives to PPS. The formation of interfacial diffusion zones is an important parameter for developing a good adhesive bond. A high power plasma treatment produces a deeper, defect-rich layer than is required by chemical bonding. Thermal baking under the temperature close to the glass transition temperature of PPS (Tg ~ 90 °C) further thickens this layer which greatly facilitates the inter-diffusion between the adhesive and the PPS substrate. Another critical parameter in determining the adhesion by secondary forces is that the adhesive should wet the surface of the substrate since the van der Waals forces diminish rapidly with distance. Good wetting is favored by the mating surfaces with matching polarities. This has been theoretically and experimentally verified by several researchers [3]. To calculate the surface energy and polarity, the expression proposed by Owens and Wendt was used [17],

æ γd l 1 + cos θ = 2 γ sd ç ç γ lv è

p ö æ ÷+ 2 γ p ç γl s ÷ ç γ lv ø è

ö ÷ ÷ ø

where: θ is the contact angle between the liquid and the solid; γ sp and γ sd are the polar and dispersion force components of the free energy of solid, respectively;

Y. Wang and S. Rak

134

γ lp and γ ld are the polar and the dispersion force components of the free energy of liquid, respectively; γ lv is the surface free energy of liquid against its saturated vapor. The total surface energy of the substrate γ sv , i.e. ( γ sp + γ sd ), and its polarity P,

γ sp (γ sd + γ sp ) , can be derived since both γ lp and γ ld are already known for selected standard liquids. Accordingly, the surface energy and polarity before and after plasma treatment were calculated using the data in Table 4. The results are listed in Table 5. A similar calculation for the cured perfluorinated polymer adhesive gives the surface energy and polarity of 25 mJ/m2 and 0.32, respectively. Table 5. Surface energy ( γ sv , mJ/m2) and polarity (P) of surface treated PPS materials

PPS 0214

PPS 1140

PPS 4665

Calculated parameters

As-received

O2 plasma 200 W/5 min.

85 °C/1 day after 200 W/5 min. plasma

γ sv

49.6

76.7

57.5

γ sp

2.2

29.1

12.2

γ sd

47.5

47.6

45.3

P (Polarity)

0.04 43.2

0.38 69.4

0.21 54.3

γ sp

2.7

23.7

15.8

γ sd

40.5

45.7

38.5

P (Polarity)

0.06 47.7

0.34 80.0

0.29 66.0

γ sp

3.6

31.5

21.9

γ sd

44.1

48.5

44.1

P (Polarity)

0.08

0.39

0.33

γ sv

γ sv

The polar and dispersion components of the surface energy of water are 51 mJ/m2 and 21 mJ/m2, respectively; whereas for diiodomethane these components are 0 and 50.8 mJ/m2, respectively.

It can be seen from Table 5 that before the treatment, surface energy of all the PPS samples is of the order of 40–50 mJ/m2 with very low polarity (<0.1). After plasma treatment, both the surface energy and polarity are significantly increased. This is mainly attributed to the increase of the polar component of the surface energy ( γ sp ). Compared with surface analysis results obtained by XPS, we conclude that the changes in surface energy and polarity for PPS 0214 are directly related

Surface modification of polyphenylene sulfide plastics

135

to the amount of free radical products such as C=O and S-O. In the case of PPS 1140 and 4665, in addition to the contribution from the free radicals, exposure of the filler after plasma treatment may also account for the increase in surface energy and polarity at the material surface. The process of thermal baking can diffuse the low molecular weight free radical products deep into the top surface, and therefore decrease the surface energy and polarity. Since the polarity of the adhesive used is only of the order of 0.32, the decrease of polarity for all PPS samples after thermal baking actually better matches the polarity between the adhesive and the adherend. Secondary forces, although being much weaker than chemical forces, contribute to a major part toward the material strength of many polymers such as polyethylene. It is also generally agreed that attraction due only to these forces is sufficient to produce an adhesive joint between polymers with joint strength equal to that of polymers themselves without the need for chemical bonds. However, for the systems that are used in harsh environments, it is still advisable that stronger bonding mechanisms (such as ionic and/or covalent bonding) be developed, by the modification of both plastic surfaces and adhesive, in order to achieve the best adhesion and, therefore, superior performance. There are other theoretical models available that describe the issue of surface energy. We decided to use polar-dispersion approach to address the response of surface changes, because it provides a direct method of comparing the surface changes after different treatments. 5. CONCLUSIONS

The adhesion of RF oxygen plasma treated PPS is noticeably improved. XPS results indicate that the initial improvement in adhesion is attributed to the creation of oxidized species on PPS surface. PPS materials with fillers have poor initial adhesion compared to PPS without glass and mineral additives. With proper plasma treatment, the adhesion of PPS with fillers can be significantly improved. The results show that thermal baking after plasma treatment greatly enhances the adhesion of PPS. Experimental data exclude the critical contribution to adhesion by the surface roughness variation from plasma treatment. Contact angle measurements show that optimal adhesion is achieved by matching the polarity between the PPS substrate and the adhesive. For the perfluorinated polymer adhesive/PPS system, the dominant bonding mechanism is due to van der Waals forces, which require an interfacial diffusion zone and polarity matching between the adhesive and adherend used. In this case, a plasma treatment at longer time and higher power, together with thermal baking, helps to modify the plastic surface providing the appropriate polarity and a deeper inter-diffusion zone.

136

Y. Wang and S. Rak

REFERENCES 1. M. Strobel, C.S. Lyons and K.L. Mittal (Eds.), Plasma Surface Modification of Polymers: Relevance to Adhesion, VSP, Utrecht (1994). 2. E.M. Liston, J. Adhesion, 30, 199 (1989). 3. S. Wu, Polymer Interface and Adhesion, Marcel Dekker, New York (1982). 4. H.F. Webster and J.R. Wightman, J. Adhesion Sci. Technol., 5, 93 (1991). 5. J. Piglowski, M.C. Cohen, R. Schafer, J. Kressler and R. Mulhaupt, Angew. Makromol. Chem., 246 (4293), 97 (1997). 6. H. Yasuda and H.C. Marsh, J. Polym. Sci.: Polym. Chem. Ed., 15, 991 (1977). 7. M.K. Shi, J. Christoud, Y. Holl and F. Clouet, J. Macromol. Sci.: Pure Appl. Chem., A30(2-3), 219 (1993). 8. E. Occhiello, M. Morra, G. Morini, F. Garbassi and O. Johnson, J. Appl. Polym. Sci., 42, 2045 (1991). 9. A. Kaul and K. Udipi, Macromolecules, 22, 12012 (1989). 10. C.-M. Chan, T.-M. Ko and H. Hiraoka, Surface Sci. Reports, 24, 1 (1996). 11. D. Briggs and G. Beamson, Anal. Chem., 64, 1729 (1992). 12. D. Briggs and G. Beamson, Anal. Chem., 65, 1517 (1993). 13. J.F. Moulder, W.F. Stickle, P.E. Sobol and K.D. Bomben, in: Handbook of X-ray Photoelectron Spectroscopy, G.E. Muilenburg (Ed.), Perkin-Elmer Corporation, Eden Prairie, MN (1995). 14. D. Briggs and M.P. Seah (Eds.), Practical Surface Analysis, Wiley, New York (1993). 15. L.J. Gerenser, J. Vac. Sci. Technol., A8, 3682 (1990). 16. W.J. Brennan, W.J. Feast, H.S. Munro and S.A. Walker, Polymer, 32, 1527 (1991). 17. D.K. Owens and R.C. Wendt, J. Appl. Polym. Sci., 13, 1741 (1969).

Adhesion Aspects of Polymeric Coatings, Vol. 2, pp. 137–149 Ed. K.L. Mittal © VSP 2003

Metal surface conditioning concepts for resin bonding in dentistry PETER PFEIFFER,1,∗ IBRAHIM NERGIZ2 and MUTLU ÖZCAN3 1 Department of Prosthodontics, School of Oral and Dental Medicine, University of Cologne, Kerpener Str. 32, 50931 Cologne, Germany 2 Department of Operative Dentistry and Periodontology, School of Dentistry, University of Hamburg, Martinistr. 52, 20246 Hamburg, Germany 3 Department of Prosthodontics, Marmara University, Dentistry Faculty, Büyükçiftlik Sok. No: 6, 80200 Nisantasi, Istanbul, Turkey

Abstract—The retention of resin veneers to metal substructures is still a problem in dentistry. The advent of adhesive dentistry has resulted in the recent introduction of new surface treatment methods. The aim of this in vitro study was to compare the shear and tensile strengths obtained by chemical surface treatment methods with those of mechanical retentions. Samples of a noble alloy were surface treated using Silicoater® MD, Kevloc® AC, Siloc®, Rocatec® and Targis® systems. Shear and tensile bond strengths with Dentacolor®, Artglass®, Signum®, Compo Plus®, Visio-Gem®, Sinfony® and Targis® veneer composites were determined after 5000 thermocycles between 5 to 55 oC with a dwell time of 30 sec followed by 7 days of water storage. The chemical surface treatment systems were found to exhibit mean shear and tensile bond strengths greater than 6 MPa (6–15 MPa for the control group with mechanical retention). Bond strengths for all chemical surface treatment systems exceeded the recommended value of at least 5 MPa for polymer-based crown and bridge veneering materials without macromechanical retention (ISO 10477, Amendment 1, 04-98), thus recommending them as an alternative to mechanical retention systems. Keywords: Silica coating; tribochemical coating; chemical bonding; shear and tensile bond strengths.

1. INTRODUCTION

A variety of methods are available for bonding resin materials to dental alloys. These may generally be categorised as mechanical, chemical or a combination of the two [1]. Macromechanical bonding, one of the oldest methods used to retain metal restorations, involves providing metal undercuts for locking the cement. This simple, old technique of using mechanical beads, loops or mesh has the main disadvantage of adding considerable thickness to the restoration [1, 2]. Recently as an alternative to conventional mechanical retention systems, chemical adhesive sys∗

To whom all correspondence should be addressed. Phone: (49) (221) 478-4717, Fax: (49) (221) 478-6722, E-mail: [email protected]

138

P. Pfeiffer et al.

tems have become popular for bonding resin to metal [3]. Chemical bonding eliminates the need for these bulkier macroretentive features and the pooling of opaquing material around retention beads [4]. One such system is Silicoater® MD (metal dotted, Heraeus-Kulzer, Wehrheim, Germany). This technique involves using a special oven to provide a chromium containing silica layer onto the surface. The principle of the system is that chromium, at temperatures higher than 250 oC, by a complex mechanism, forms waterproof bonds with silicates [5–7]. Kevloc® AC system (Heraeus-Kulzer, Wehrheim, Germany) is relatively a new system, introduced in 1995, that offers a combination of chemical and mechanical bonding. In this system, the surfaces of the alloys are sandblasted with fresh 110 µm grains of aluminum oxide. The temperature needed to activate the bonding layers of the resin is generated by contact heat transfer and heat radiation by an activation chamber [8]. In Siloc® technique (Heraeus-Kulzer, Wehrheim, Germany), the surfaces are sandblasted with 250 µm grains of Al2O3. This activated surface is then coated with silane and a special bonding agent and the specimen then placed in a special heat curing apparatus. The opaquer layer is then applied and light-cured (Dentacolor XS®, Heraeus-Kulzer, Wehrheim, Germany). The Rocatec® system (ESPE, Seefeld, Germany) introduced in 1989, is an acrylic resin-metal bonding system [9]. The principle is a tribochemical application of a silica layer by means of sandblasting. First the surface is conditioned by sandblasting with 110 µm aluminum oxide grains at a pressure of 0.25 MPa using a Rocatector® Delta device (ESPE, Seefeld, Germany with Rocatec Pre). Then the samples are blasted with Rocatec Plus, which is 110 µm Al2O3 modified with silicic acid, using Rocatector Delta at 0.25 MPa at a distance of 1 cm from the metal surface for 13 sec/cm2. The theoretical velocity at which the Rocatec Plus particles hit the alloy surface is 200 m/s, producing spot heating of up to 1000 oC. The tribochemical effect [9] of sandblasting results in deposition of a molecular coating of alumina and silica on the metal surface. The surface is then coated with silane to render the surface even more chemically reactive to the resin. Targis® Link (Ivoclar, Ellwangen, Germany) is a bonding agent based on phosphoric acid ester with a methacrylate functional group. The phosphoric acid ester group of the molecule is a strong acid that reacts with the metal or the metal oxides and forms a phosphate passivation layer on the metal surface. The methacrylate group in the phosphoric acid reacts with the monomer contained in the Targis opaquer and forms a copolymer assuring a bond with the veneering material. Hydrolytic stability is achieved since Targis® Link contains a monomer with an aliphatic hydrocarbon, which is highly water-repellent, which, of course, is very important in clinical applications. In an effort to determine ideal adhesives, promoters and techniques, this in vitro study measured the shear and tensile strengths of bonds between polymer glass composites and noble alloy samples surface treated using these systems, comparing these with mechanical bonding, after subjecting all samples to thermocycling and storage in water.

Bond strength of resin veneer to alloy

139

2. MATERIALS AND METHODS

The samples for the shear tests were fabricated using a cylinder ring and plunger set-up. A copper ring with an outer diameter of 9 mm and a plunger of outer diameter 8 mm were used. A total of 198 resin samples of 9 mm diameter and 0.5 mm thickness were produced as follows: A sheet of casting wax was lightly pressed between two clean glass slabs to produce a sheet 0.5 mm thick. A copper ring of outer diameter of 9 mm was used to punch holes in it. Methyl methacrylate monomer and polymer (Palavit G®, Heraeus-Kulzer, Wehrheim, Germany) were mixed and poured around the periphery of the punched holes to prevent air bubble entrapment. A clean dry glass slab was lightly pressed onto the samples until the resin was hard in order to produce flat and smooth sample surfaces. After the resin had set completely, the disks were carefully removed and cleaned of wax or excess resin around the periphery. A dial caliper was used to check the thickness. A total of 198 samples with 9 mm diameter and 0.5 mm thickness were cast out of a noble alloy (Au74Pt9Ag9) (Degunorm®, Degussa, Hanau, Germany). The samples for the tensile tests (2 x 198, Degunorm) were fabricated according to the suggestions of Laufer et al. [6] and consisted of a metal disk 10 mm thick and 6 mm in diameter with a stem 10 mm long and 3 mm in diameter. The top surfaces of the samples were polished by hand with SiC-P 1200 C extra-fine emery paper. The control group consisted of samples, which had 0.6 mm diameter spherical beads on the surface to provide mechanical retention. They received no chemical coating. The materials used for the substrate samples and in each experimental group are listed in Table 1. There were 9 disk samples in each group. All samples were conditioned according to the manufacturers‘ instructions, veneered with a layer of opaquer approximately 0.2 mm thick, and light cured as detailed below. Dentacolor® Opaquer: The powder and liquid were mixed (1:1) producing a thin consistency, stirring for at least 30 sec to achieve a homogeneous distribution of the pigments. A thin layer was applied with a brush and light cured for 90 sec in the Dentacolor XS unit (Heraeus-Kulzer, Germany). Second and third thin layers were separately applied and cured until complete masking of the metal was achieved. Artglass® Opaquer: The components of opaquer were mixed on a mixing pad. A thin layer of opaquer was dabbed on to the alloy surface with a short bristle brush. As the first layer did not mask the metal sufficiently, further layering was performed. Each layer of opaquer, which did not exceed 75 µm in thickness, was light-cured for 90 sec in the Dentacolor XS. Signum® Opaquer: The opaquer was squeezed onto a mixing pad and stirred. A brush with hard and short bristles was used to apply individual layers cross-wise until the metal framework was covered. Each layer was polymerized separately.

140

P. Pfeiffer et al.

Bond strength of resin veneer to alloy

141

Visio-Gem® and Sinfony® Opaquers: The opaquers for each system were dispersed at a ratio of 1:1 in a glass-mixing cup for 45 sec and painted onto the silanized surfaces, and the opaquer liquid bottle closed immediately after use. The opaquers were polymerised in the VISIO ALFA light unit (Program 1, ESPE, Seefeld, Germany) twice for 5 sec. CompoPlus® Opaquer: The Compo Plus opaquer was applied in two steps. In the first step, it was applied with the special brushes supplied, in a thin even layer, less than 0.1 mm thick to just cover the metal surface below and then intermediate polymerisation was carried out. In the second step, the opaque layer was again applied in an evenly thin layer to just cover the remaining surface. Targis® Opaquer: After being dried, the metal was coated with two individually polymerised layers of Targis opaquer. Opaquer, which did not polymerise because of oxygen inhibition, was removed with an acrylic sponge. All of the veneer composites were cured onto the metal samples conditioned by each of these systems or onto the control metal samples with mechanical retention. The veneer composites were 9 mm in diameter and 2 mm long for the shear tests and 6 mm in diameter and 2 mm long for the tensile tests. The bonded samples were kept in a humidifier (Memmert, Schwalbach, Germany) for 24 h in distilled water at 37 oC, then subjected to 5000 thermal cycles in water (30 sec at 5 oC and 30 sec at 55 oC, in accordance with ISO 10477), before being stored in the humidifier for a further 7 days. Prior to the tensile tests the top surface of the veneering composites was glued (Sekundenkleber, Orbis Dental Offenbach, Germany) to a corresponding metal disk with a stem to fix with the cardan joint of the testing machine (Figure 2). A Zwick® 1445 universal testing machine (Zwick, Ulm, Germany) was used at a crosshead speed of 1 mm/min for the shear (Figure 1, [25]) and tensile tests (Figure 2, [6]) which were performed without drying the samples. The data obtained were statistically analysed using analysis of variance, ANOVA, α = 0.05. Significant differences were calculated utilizing the post hoc test of Bonferroni/Dunn (StatView 5.0, SAS Institute Inc., Cary, NC, USA). 3. RESULTS

Silicoater MD: Mean shear bond strengths varied between 8 to 9 MPa for the veneering composites (Dentacolor, Artglass, Compo Plus or Signum, Table 2). Differences between veneering composites were not significant (Table 3). For the tensile tests, however, the strength for Signum (9 MPa) was significantly higher than that of Artglass (5 MPa) and of Compo Plus (6 MPa) (p<0.05, Tables 2, 4). Kevloc AC: Mean shear and tensile bond strengths varied between 10 to13 MPa for Dentacolor, Artglass or Compo Plus (Table 2). There were no significant differences between them for either shear or tensile strength (p>0.05, Tables 3, 4).

142

P. Pfeiffer et al.

Figure 1. Shear test device.

Siloc: The shear bond strength of Dentacolor (14 MPa) was significantly higher than that of Artglass (9 MPa), of Signum (8 MPa) and of Compo Plus (9 MPa) (p<0.05, Tables 2, 3). However, the tensile strength of Dentacolor was not significantly higher than any of the others (Table 4). Rocatec: Mean shear bond strengths varied between 7 and 12 MPa for VisioGem, Sinfony and Compo Plus (Table 2) but the differences were not significant (Table 3). However the tensile strength of Compo Plus (11 MPa) was significantly higher than that of Visio-Gem (7 MPa) and of Sinfony (8 MPa) (p<0.05, Tables 2, 4). Mechanical retention (control): The shear strengths of Dentacolor (15 MPa) and Sinfony (15 MPa) were significantly higher than that of Artglass (10 MPa) (p<0.05, Tables 2, 3). The tensile strengths of Dentacolor (9 MPa) and Visio-Gem (9 MPa) were significantly higher than that of Signum (6 MPa) (Tables 2, 4). It was observed that the samples of the shear tests treated with the Rocatec system and bonded with Sinfony or coated with the Targis Link system and bonded with Targis showed cohesive failure in the composite or the opaquer layer. The same mode of failure was observed with mechanical retentions in combination with the tested veener composites. Samples subjected to the Silicoater MD system and bonded with Artglass, Compo Plus or Signum, Kevloc system with Artglass and Siloc system in combination with Signum showed interfacial failure. A mixed form of failure was observed for the other combinations.

Bond strength of resin veneer to alloy

143

Figure 2. Tensile test device.

Mechanical retentions in combination with Visio-Gem showed interfacial failure in the tensile test. Samples of the tensile tests subjected to the Silicoater MD system and bonded with Artglass showed a mixed form of failure. Cohesive failure was observed for the other combinations. 4. DISCUSSION

The technology for bonding a resin to dental alloys has significantly improved over the last decade. Now various commercial bonding techniques are available which claim to provide a durable bond. All these new systems involve the conditioning of the substrate to produce bifunctional molecules that adhere to the metal surface with a polymerisable double bond. These molecules react with the methacylate groups contained in the monomers of the applied resin/veneering composite in a radical polymerisation process [10–13].

P. Pfeiffer et al.

144

Table 2. Shear and tensile bond strengths (MPa) for combinations of bonding systems and veneering composites (control group: mechanical retention) Shear Strength Mean Silicoater MD Dentacolor Artglass Signum Compo Plus

Tensile Strength SD

Mean

SD

8.5 8.7 7.8 7.9

2.0 3.8 2.6 2.6

7.7 5.1 9.2 5.6

0.9 1.4 2.2 1.3

Kevloc Dentacolor Artglass Compo Plus

11.3 11.7 12.5

2.5 2.2 3.5

11.8 12.6 10.2

1.9 1.0 2.4

Siloc Dentacolor Artglass Signum Compo Plus

14.4 8.5 7.9 9.0

4.7 2.1 1.8 2.5

6.8 6.0 6.1 7.1

1.0 1.3 1.1 0.8

Rocatec Visio-Gem Sinfony Compo Plus

7.3 10.2 12.2

2.0 3.4 4.5

6.5 8.1 10.9

1.9 2.3 1.2

Targis Link Targis

13.9

2.9

6.6

1.2

Retentions (Control Group) Dentacolor Artglass Signum Visio-Gem Sinfony Compo Plus Targis

15.1 10.2 9.1 13.1 15.3 10.7 13.2

2.2 1.2 2.0 0.9 2.1 1.7 3.4

8.6 7.4 5.8 9.1 8.4 7.5 6.7

1.4 1.3 1.3 1.9 2.1 0.5 1.1

In order to find ideal adhesives, promoters and techniques, this study examined the shear and tensile bond strengths after thermocycling and water storage of veneer composites bonded to noble alloy samples using different opaquer systems. This study found that most of the chemically treated surfaces resulted in bond strengths that amounted to those produced by mechanical retention. Mechanical retention can result in microleakage and requires a high thickness of material necessitating overcontouring of the restoration, which is inconsistent with the conservative approach to clinical applications [14].

Bond strength of resin veneer to alloy 145

146

P. Pfeiffer et al.

Bond strength of resin veneer to alloy

147

The manufacturers of most of the new surface treatment systems require sandblasting of the metal prior to bonding to achieve a high bond strength. This has the potential to remove significant amounts of material. The material loss resulting from these procedures depends on the alloy type. The noble alloy used in the present study is known to be one of the most biocompatible alloys and has high corrosion resistance [15]. Shear bond strengths of non-precious alloys after 5000 thermocycles have been reported to be higher than those of precious ones [16]. Other studies have found that bonds formed via silica coating or silanization are stronger on base alloys than on noble alloys. A good deal needs to be learned about the surface characteristics of dental alloys after being treated in these ways in order improve our understanding of the bonding mechanism [4, 15, 18]. It follows then that in clinical applications involving fixed partial prostheses, where high bond strengths are required, it is important to know the type of underlying alloy used when bonding silica coated and silanized surfaces and composite resin veneers. Thermocycling, though a necessary test for oral applications, is detrimental because it relaxes stresses within the composite resins produced by polymerisation shrinkage [18]. It also results in a rapid increase in the amount of water absorbed into the composite resin at higher temperatures causing hydrolysis of the silane bond. Furthermore, water absorption which causes hydrolysis of the silane bond increases rapidly with temperature [19]. Thermal cycling both ages the resin as well as creates a stress at the bondline. The stress results from the large coefficient of thermal expansion mismatch between the resin and the metal. Although the surface treatment studies conducted have used different thermocycling times [20, 21], the common consensus is that the thermocycling decreases the bond strength as it weakens the resin structure. In this study, shear and tensile tests were employed after 5000 thermocycles between 5–55 °C with a dwell time of 30 sec in accordance with ISO 10477 [22] and stored in water for 7 days. Some authors suggested that long water storage times of at least 180 days along with thermal cycles were needed to test the aging of chemical bonding systems [9, 16, 23]. In 1998 the International Organization for Standardization published thermocycling parameters for testing polymer-based, crown and bridge materials in order to standardize study design [22]. However, in order to estimate the actual clinical performance, due to different consumption temperatures of beverage and food, the temperature effect cannot be standardized [24]. Although differences in experimental parameters make a comparison difficult, the results of the present study for the Silicoater MD-Dentacolor combination were similar to those of Kolodrey et al. [4] in which they used non-precious alloys [4]. The present study also found bond strengths similar to those reported by Kurtz et al. [25] for Siloc and noble alloy combinations. Our study found that with mechanical retention, Dentacolor was significantly stronger than Artglass and also found significant differences in bond strengths between opaquers. It is likely that bond strengths produced by opaquers besides being dependent on chemical composition, may also depend on consistency and thus han-

148

P. Pfeiffer et al.

dling characteristics. When they were mixed in the proportions specified by the manufacturers, the consistency of the opaquers turned out to be slightly different from one another which may have contributed to differences in thickness between the resin and the conditioned metal surface, thereby affecting their bond strength. After bond strengths testing, most metal samples showed an even layer of veneer composite on the metal surface. However, no correlation between bond strength and cohesive or interfacial failure could be detected. In order to withstand functional loads, the bond between the resin and the restoration must be strong and durable. Recently, the advantages of extraoral silica coating have been demonstrated in a number of studies, all of which report results in accordance with those presented in this study [26–28]. The bond strengths obtained for the Kevloc group were significantly higher than those obtained for the Silicoater MD group. In a study by Vojvodic et al. [29], this difference was attributed to differences in bonding layer thickness, opaquer viscosity or opaquer liquid proportion. Kevloc treated surfaces consist of a fused layer of rigid, tough acrylonitrile and a layer of water-resistant urethane elastic, tough highly cross-linked resin, both reinforced by a direct acrylizing network. Thus, the Kevloc system provides a low modulus layer between the alloy surface and the veneer resin. This would allow a degree of self-alignment in the tensile apparatus and hence may explain the higher tensile bond strengths for all composite resins on this surface compared to other bonding systems. A common finding of several studies was that the bond strength decreased after water storage and thermocycling [30–33]. 5. CONCLUSIONS

The following conclusions could be drawn: (1) Dentacolor with Silicoater MD as well as Visio-Gem and Sinfony with Rocatec system exhibited significantly lower shear bond strengths to a noble alloy than they did in combination with macromechanical retention. However, in the tensile tests, the Signum/Silicoater MD, Artglass/Kevloc, Dentacolor/Kevloc and Compo Plus/Rocatec combinations exhibited significantly higher bond strengths than did these veneer composites with macromechanical retention. (2) None of the other resins tested showed significant differences in shear or tensile strength for any of the surface treatment systems compared to macromechanical retention. (3) The encouraging results obtained for the Rocatec tribochemical coating system and the Siloc, Kevloc and Targis chemical coating systems enable them to be recommended as a substitute for mechanical retentions.

Bond strength of resin veneer to alloy

149

Acknowledgement The authors wish to thank the companies Heraeus-Kulzer, ESPE and Ivoclar for supplying the coating systems. REFERENCES 1. H. F. Alberts, Adept Report 2, 29-36 (1991). 2. G. Shen, J. Dental Res. 2, 221 (Abst. 464) (1984). 3. K. F. Leinfelder, J. Am. Dental Assoc. 125, 292-294 (1994). 4. H. Kolodrey, A. Puckett and K. Brown, J. Prosthet. Dentistry 67, 419-422 (1992). 5. H. Lüthy, C. P. Marinello and P. Schaefer, J. Dental Res. 59, 223 (Abst. 884) (1988). 6. B. Z. Laufer, J. I. Nicholls and J. D. Townsend, J. Prosthet. Dentistry 3, 320-327 (1988). 7. H. J. Tiller, R. Göbel, B. Magnus, R. Musil and R. Bimberg, Dental Lab. 38, 78-82 (1990). 8. I. Nergiz, P. Pfeiffer and W. Niedermeier, J. Dental Res. 76, 312 (Abst. 2390) (1997). 9. R. Guggenberger, Dtsch. Zahnaerztl. Z. 44, 874-876 (1989). 10. T. Anagnostopoulos, G. Eliades and G. Palghias, Dental Mater. 9, 182-190 (1993). 11. E. Asmussen, J. P. Attal and M. Degrange, Acta. Odontol. Scand. 51, 235-240 (1993). 12. J. Ishijima, A. A. Caputo and R. Mito, J. Prosthet. Dentistry 67, 445-449 (1992). 13. M. Kern and V. P. Thompson, Dental Mater. 9, 155-161 (1993). 14. I. F. Tulunoglu and M. Öktemer, Quintessence Inter. 28, 447-451 (1997). 15. Council on Dental Materials, Instruments, and Equipment, J. Am. Dental Assoc. 104, 501-505 (1982). 16. M. Kern and V. P. Thompson, J. Dentistry 23, 47-54 (1995). 17. J. H. Van der Veen, A. E. Bronsdijk, A. P. Slaghter, A. C. M. Poel and J. Arends, Dental Mater. 4, 271-277 (1988). 18. J. F. Roulet, Degradation of Dental Polymers, p. 60, Karger, Basel (1987). 19. K. J. M. Söderholm and M. J. Roberts, J. Dental Res. 69, 1812-1816 (1984). 20. J. H. Bailey, J. Prosthet. Dentistry 61, 174-177 (1989). 21. C. H. Pameijer, N. P. Louw and D. Fischer, J. Am. Dental Assoc. 127, 203-209 (1996). 22. International Organization for Standardization, “Polymer-based, Crown and Bridge Materials”, ISO 10477, Amendment 1 (1998). 23. R. M. Smith, M. G. Barrett, W. A. Gardner, T. Marshall, M. J. McLean, D. W. McMichael, P. J. Yerbury and H. R. Rawls, Am. J. Dentistry 6, 111-115 (1993). 24. S. Y. Lee, E. H. Greener and D. L. Menis, Dental Mater. 11, 348-353 (1995). 25. K. Kurtz, C. Ohkuba, A. Lavaca and T. Okabe, J. Dental Res. 78, 121 (Abst. 122) (1999). 26. P. Pfeiffer, P. Proano, I. Nergiz and W. Niedermeier, J. Dental Res. 75, 24 (Abst. 53) (1996). 27. P. Pfeiffer, Dtsch. Zahnaerztl. Z. 48, 692-695 (1993). 28. M. Özcan, A. Schulz and W. Niedermeier, J. Dental Res. 78, 302 (Abst. 1573) (1999). 29. D. Vojvodic, V. Jerolimov, A. Celebic and A. Catovic, J. Prosthet. Dentistry 81, 1-6 (1999). 30. A. Peutzfeld and E. Asmussen, Scan. J. Dentistry 96, 171-176 (1988). 31. P. Pfeiffer, Dtsch. Zahnaerztl. Z. 44, 936-937 (1989). 32. J. Chang, J. M. Powers and D. Hart, J. Prosthodont. 2, 110-114 (1993). 33. O. Hansson and L. Moberg, Scan. J. Dental Res. 101, 243-251 (1993).

This page intentionally left blank

Adhesion Aspects of Polymeric Coatings, Vol. 2, pp. 151–163 Ed. K.L. Mittal © VSP 2003

Measurement of internal stresses in polymeric coatings using time resolved fluorescence TAIJI IKAWA and TOHRU SHIGA∗ Toyota Central Research & Development Laboratories Inc., Nagakute-cho, Aichi-gun, Aichi-ken 480-1192, Japan

Abstract—This paper describes a new nondestructive technique for detecting internal stresses in polymeric coatings using time resolved fluorescence. The measurement principle is based on an experimental result that the lifetime of fluorescence from poly(3-octylthiophene) dispersed in a uniaxially-stretched polymer film decreases with increasing tensile stress acting on the film. The validity of the time resolved fluorescence technique was investigated by comparing it with a traditional bimetallic method. The time resolved fluorescence technique was employed to estimate the internal stresses in polymeric coatings of a multi-layered system. The internal stress distribution in a clear coating of four-layer system was monitored directly. The internal stress in the undercoat of the system was also measured The effects of temperature, humidity, and light on the internal stresses were analyzed. Keywords: Internal stress; polymeric coating; nondestructive technique; time resolved fluorescence; lifetime of fluorescence; poly(3-octylthiophene).

1. INTRODUCTION

Internal or residual stresses strongly affect the fracture and reliability of polymeric coatings, which are used both to protect and enhance the appearance of automotive assemblies. The internal stresses in polymeric coatings are generally determined by the traditional bimetallic method [1]. The measurement principle of the bimetallic method is based on the fact that when a coating is applied and cured on a thin metal substrate, the coated substrate will deflect slightly. The internal stress has been calculated from the generated deflection. However, the internal stresses in coatings of multi-layered systems in practical applications cannot directly be estimated. It is, therefore, important to develop a new nondestructive method for directly measuring internal stresses in multi-layered polymeric coatings.

∗

To whom all correspondence should be addressed. Phone: +81-561-63-5221, Fax: +81-561-63-6137, E-mail: [email protected]

152

T. Ikawa and T. Shiga

The purpose of this paper is to present a new method for detecting internal stresses in multi-layered polymeric coatings using time resolved fluorescence. Internal stresses in topcoat and undercoat of multi-layered systems have first been analyzed by the time resolved fluorescence technique. The effects of temperature, humidity, and light have been investigated. Stress monitoring in an outdoor weathering test has also been reported. 2. TIME RESOLVED FLUORESCENCE TECHNIQUE FOR DETECTING INTERNAL STRESSES

2.1. Fluorescence from poly(3-alkylthiophene) in uniaxially stretched polymer films Poly(3-alkylthiophene), P3AT, is a conjugated polymer with rigid thiophene backbone and flexible sidechains. The P3AT having long sidechains (the number of carbons in the alkyl group > 4) is soluble in organic solvents [2]. P3AT in a polymer matrix emits fluorescence because of its π conjugated structure. It is well known that mechanical stretching influences the photoluminescence [3]. When stretched uniaxially, a P3AT macromolecule in the matrix shows spectral changes in absorption and steady-state emission. We, recently, studied time resolved fluorescence from P3AT dispersed in a poly(methyl methacrylate) PMMA film under a tensile loading [4]. Figure 1 shows a relationship between the lifetime of fluorescence from poly(3octylthiophene), P3OT and the applied strain. The lifetime τ has been calculated using Eq. (1): I= å A i ⋅ exp ( – t / τ i )

(1)

where I is the intensity of fluorescence, Ai is constant for i component, t is time, and τ is lifetime. The stress–strain curve of the PMMA film is also shown in Figure 1. The lifetime of fluorescence from P3OT in an elastically deformed PMMA matrix (strain ~ 0.05%) diminishes with the applied stress or strain. The lifetime of the fluorescence reverses to the starting value after the removal of the applied stress or strain. These results suggest the possibility of using P3OT as a molecular sensor for nondestructive detection of internal stresses in polymer materials. The P3OT can detect internal stresses of the order 1 MPa. Based on our analytical studies [5], the unique phenomenon of time resolved fluorescence under tensile loadings can be induced by distortion or deformation of the π conjugated structure of large P3OT molecules with molecular weight more than 2800. 2.2. Procedure for the time resolved fluorescence technique Consider a polymeric coating containing a small amount of P3OT. When it is applied and cured on a metal substrate, the coating tries to shrink as the crosslinking

Measurement of internal stresses in polymeric coatings

153

Figure 1. Strain dependence of lifetime of fluorescence from P3OT in a uniaxially-stretched poly(methyl methacrylate), PMMA, film. The solid line represents the stress-strain curve of the PMMA film.

reaction proceeds. Since the cured coating cannot slide on the substrate, an internal stress, two-dimensional (2D, in x and y directions in Figure 2) tensile stress, will be generated in the coating. Compressive stress will occur in the z direction. Here we assume that the 2D internal stress is equal to a one-dimensional (1D, y axis) tensile stress produced when a free-standing polymer coating is stretched unidirectionally. We assume that the internal stress profile is uniform within the entire coating. The compressive stress in the z-axis is neglected. The second assumption in our technique is that the fluorescence from P3OT in the coating on the substrate is similar to fluorescence from the label in the free- standing coating under a uniaxial stretching (Figure 2). The procedure for the time resolved fluorescence study consists of the following four steps: Step 1: A coating containing a small amount of fluorescence label is prepared. Then the coating is sprayed and cured on a metal substrate. Step 2: The decay time of fluorescence from the label in the coating is measured.

154

T. Ikawa and T. Shiga

Figure 2. Schematic illustration of internal stresses in a coating cured on a metal substrate and in a uniaxially stretched free-standing coating.

Step 3: A fluorescence decay time-internal stress calibration curve is obtained by measuring the time resolved fluorescence of a free-standing coating having the label under a tensile loading. Step 4: The decay time in Step 2 is finally reduced to an internal stress using the calibration curve in Step 3. 3. EXPERIMENTAL

3.1. Materials 9-methylanthracence, 9-MeAn (Tokyo Kasei Ltd.) and poly(3-octylthiophene), P3OT were used as fluorescence labels for measuring internal stresses. 9-MeAn was used as received. P3OT was synthesized by oxidative coupling of 3octylthiophene with FeCl3 in chloroform [5]. The clear coatings used were an acrylic copolymer crosslinked with melamine formaldehyde resin and an acrylic copolymer having glycidyl group and carboxylic acid groups. The base coatings used were acrylic copolymers containing melamine formaldehyde as a crosslinker and ca.10 wt% aluminum flakes. The clear coats and base coats were obtained from Kansai Paint and Nippon Paint, respectively. 3.2. Specimens 3.2.1. Single-layer coating on a substrate The clear coating containing P3OT or 9-MeAn was sprayed using a spray gun with an air pressure of 3.5 kg/cm2 and cured on a phosphor bronze plate (150 mm length, 50 mm width, 50 µm thickness) at 140 ºC for 30 min. The fluorescence labels were soluble in the clear coating at room temperature. The fractions of the fluorescence labels were 0.05 wt% (P3OT) and 0.001 wt% (9-MeAn).

Measurement of internal stresses in polymeric coatings

155

3.2.2. Multi-layer coating on a substrate The base coating free of the fluorescence label and the clear coating having the label were sprayed separately on a steel flat plate on which an electro-deposited coat and a surfacer coat had already been applied. Both coatings were cured at the same time at 140 °C for 30 min. The thicknesses of the obtained coats were 22 µm (base coat) and 40 µm (clear coat). When the internal stress in the base coat was measured, the base coating having the label and the clear coating free of the label were used. 3.2.3. Free-standing coatings for the calibration curve Both clear coating and base coating containing the label were sprayed separately and cured on a glass plate at 140 ºC for 30 min. After curing, the clear or base coat was removed from the glass plate in a warm water of 40 ºC and cut in size (50 mm length, 20 mm width). 3.3. Fluorescence spectroscopy Time resolved fluorescence measurements were performed at room temperature on a picosecond time resolved photoluminescence spectrometer with a streak camera (Hamamatsu Photonics, C4780). A diagram of the spectrometer is shown in Figure 3. For excitation we used a dye laser with an average power of 150 µJ and a frequency of 15 Hz. When the label was P3OT, the excitation wavelength was 420 nm with a pulse duration of 300 ps. The dye laser output was focused onto a spot with a diameter of 1 mm. The excitation of the specimen by the laser was carried out using a measurement probe. The probe also detected the fluorescence from P3OT or 9-MeAn in the specimen. 3.4. Weathering test An accelerated weathering test was carried out using a Sunshine Weather Meter (Suga Test Instruments, WEL-SUN-DCE). The light source used was a Sunshine

Figure 3. Diagram of picosecond time resolved photoluminescence spectrometer.

T. Ikawa and T. Shiga

156

carbon arc lamp from Suga. The accelerated weathering test was carried out at 50 ºC and 95% RH (relative humidity). Outdoor weathering tests on the specimens were also carried out. 4. ANALYSIS OF INTERNAL STRESSES

4.1. Single-layer coating on a metal substrate The internal stresses in a single-layer coating on a thin bronze phosphor plate were measured by time resolved fluorescence technique as well as by the traditional bimetallic method to understand the validity of the time resolved fluorescence. In the bimetallic method, the deflection of a single-layer coating on a substrate was measured by a laser displacement detector. The internal stress σ was calculated using Eq. (2) [6].

σ=

(

Eh3 2 + am − bm 2 12 ρ ( h + H )

)

(2)

Here E is elastic modulus of the phosphor bronze plate, m is the ratio of elastic modulus of the coating to E. h is thickness of coating, H is thickness of the phosphor bronze plate, and ρ is radius of curvature of the deflected substrate. a and b are constants that are determined by thicknesses of the coating and of the substrate. Figure 4 shows the relationship between internal stresses measured by the two methods. The label used was P3OT (0.5 wt% content). The results from the two techniques are relatively in good agreement. Stress relaxation for a short period after curing is shown in Figure 5. The measurement was carried out using 9MeAn as the label. The results obtained from both methods show that the stress relaxation occurs mainly in one or two days after curing. From the results in Figures 4 and 5, it is concluded that the time resolved fluorescence technique gives reliable values of internal stresses in coatings [7]. 4.2. Internal stresses in a multi-layer coating system The time resolved fluorescence technique was employed to analyze internal stresses in multi-layer coating systems [7]. When the internal stress in the undercoating was measured, the fluorescence label was mixed only with the undercoating. Figure 6a shows a typical four-layer coating system of clear coat, base coat, surfacer coat, and electrodeposited (ED) coat. The system was prepared by 2 coatings, 1 baking (2C1B) method. The 2C1B method represents a coating method used widely in automotive engineering. The surfacer coat is one of undercoats on the ED coat which contains a small amount of metal flake.

Measurement of internal stresses in polymeric coatings

157

Figure 4. Comparison between internal stresses by time resolved fluorescence technique and by the bimetallic method (Specimen: Clear coating A cured on a phosphor bronze plate, Label: P3OT).

Figure 5. Stress relaxation after curing (Specimen: Clear coating B cured on a phosphor bronze plate, Fluorescence label: 9-methylanthracence).

The internal stresses in clear coating B and in base coating B were 0.51 MPa and 0.81 MPa, respectively. Comparing the internal stresses in the clear coating B of single-layer coating (Figure 5) and of the multi-layer coating (Figure 6a), it is

T. Ikawa and T. Shiga

158

Clear coating B 0.51 MPa Base coating B 0.81 MPa Surfacer ED Steel

Upper clear coating D 0.75MPa Lower clear coating E 1.81 MPa Base coating C 2.23 MPa Surfacer ED Steel

(a) Four-layer coating system

(b) Double clear coating systems

Figure 6. Internal stresses in multi-layer coating systems. Four-layer coating system was prepared by 2 coatings, 1 baking (2C1B) method. Double clear coating system was obtained using 3 coatings, 2 baking (3C2B) method. The 2C1B and 3B2C methods are typical coating methods used in automotive engineering. The undercoating on the ED coat is called surfacer coating which contains a small amount of metal flake.

found that the value for the latter was smaller than the former. Figure 6b illustrates stress distribution in a double clear coating system prepared by 3 coatings, 2 baking (3C2B) method. The internal stress in the upper clear coating is smaller than that in the lower coating. The internal stress in the base coating (2.23 MPa) is three times as large as that in the upper coating (0.75 MPa). The internal stress diminishes as the coatings are separated from the steel plate. As shown in Figures 6a and 6b, the order of the internal stress in the topcoat is 1 MPa. It is much smaller than the tensile stress at break of the free-standing topcoat (~ 40 MPa). The time resolved fluorescence technique has given information about internal stresses in multi-layer coating systems. The attractive point of this technique is that it utilizes a pen-shaped measurement probe having optical fibers to measure the lifetime of fluorescence. The probe enables us to analyze stresses in coatings applied in automotive assemblies. We have investigated stress distribution in the clear coating of a four-layer system on the door of an automotive body [8]. Internal stresses in the clear coating at points A to E in Figure 7a are listed in Table 1. The internal stresses at points A to D are the same. Figure 7b shows stress distribution in three clear coatings of A to C on an uneven area of the door. The internal stress reaches a maximum at the edges of the area. Table 1. Internal stress (in MPa) in a clear coat on the door of an automobile (Figure 7a) Point A

Point B

Point C

Point D

Point E

0.68

0.64

0.64

0.68

0.71

4.3. Internal stress monitoring in accelerated weathering tests It is known that the internal stress is influenced by temperature, moisture, and light [9]. The internal stresses in the clear coating and base coating of a four-layer coating system were first investigated when the specimen was exposed to an atmosphere of 50 ºC and 95% RH. The exposure test was not carried out under the

Measurement of internal stresses in polymeric coatings

159

Figure 7a. Internal stress distribution in clear coating B cured on a door plate of an automobile (Specimen: four-layer coating system of clear, base, surfacer, and ED coatings).

influence of light. The fluorescence behavior of P3OT in the coatings was unchanged by the exposure test. Figure 8 shows the relationship between the internal stress and exposure time at a short period of time. After the exposure test was finished, the specimens were cooled to 25 ºC and then the internal stresses were measured at 25 ºC. The internal stress in the clear coating after exposure for 2 h became twice as large as the starting value at 0 h. The increase in the internal stress at 2 h is very interesting. The same phenomenon has been observed in the case of the free-standing clear coating [9]. The internal stress in the base coating decreases with time gradually. The temperature and moisture effect on the clear coating is stronger than on the base coating.

160

T. Ikawa and T. Shiga

Figure 7b. Internal stress distribution in a clear coat cured on the door of an automobile (Specimen: four-layer coating system of clear, base, surfacer, and ED coatings).

The accelerated weathering test in Figure 9 was carried out under the influence of light irradiation [10]. As shown in Figure 9, the internal stress in the clear coating B decreases with exposure time when irradiated with light. When the exposure time exceeded 6 h, the stress was not monitored as the P3OT emitted no fluorescence because it degraded under the influence of light. We have finally monitored internal stress in the clear coatings of two coating systems during outdoor weathering test. The fluorescence label used was 9MeAn. The sunlight influences considerably the internal stress in the outdoor weathering test (Figure 10). The capability of 9-MeAn as a stress sensor decreases in the outdoor test, and therefore, 9-MeAn cannot be employed for the stress monitoring for a long term, e.g., on the order of one year [10]. 5. CONCLUSION

We have developed a new nondestructive technique for detecting internal stresses in polymeric coatings using time resolved fluorescence. The time resolved fluorescence technique has provided us the information about internal stresses both in clear coating and in base coating of a multi-layer system. The stress monitoring during an outdoor weathering test has been carried out for periods at least one

Measurement of internal stresses in polymeric coatings

161

Figure 8. Stress monitoring in an accelerated weathering test. (● ) Clear coat, (□ ) Base coat, Specimen: four-layer coating system prepared by the 2C1B method. Test condition: 50 ºC, 95% RH.

Figure 9. Effect of light irradiation on the internal stress. (○ ) irradiated, (● ) not irradiated. Specimen: clear coat B of four-layer coating system.

162

T. Ikawa and T. Shiga

Figure 10. Stress monitoring during outdoor weathering test. (○ ) Clear coat in single-layer coating system, (□ ) Clear coat in four-layer coating system prepared by the 2C1B method. Fluorescence Label: 9-methyl anthracence.

month. The next target of our research is to find organic dyes for detecting internal stresses for periods of the order of one year. The improvement in the photoluminescence spectrometer with optical fibers is a key factor in the successful application of this technique. It enables us to analyze stress distribution in many cases. Not only internal stresses in coatings but residual stresses at the surfaces of injection-molded polymeric materials have already been investigated [11]. REFERENCES 1. A. Saarnak, E. Nilsson and L.O. Kornum, J. Oil Colour Chem. Assoc., 59, 427-432 (1976). D.Y. Perera and D. Vanden Eynde, J. Coatings Technol. 53, No. 677, 39-44 (1981). K. Ochi, K. Ohnishi, K. Yamashita and M. Zinbo, Koubunshi Ronbunshu, 47, 559-567 (1990). O. Negele and W. Funke, Prog. Organic Coatings, 28, 285-289 (1996). 2. S.D.D.V. Rughooputh, M. Nowak, S. Hotta, A.J. Heeger and F. Wudl, Synthetic Metals, 21, 4150 (1987). K. Yoshino, S. Nakajima, M. Onoda and R. Sugimoto, Synthetic Metals, 28, C349C357 (1989).

Measurement of internal stresses in polymeric coatings

163

3. K. Kaneto and K. Yoshino, Synthetic Metals, 28, C287-C292 (1989). L.S. Swanson, J. Shinar and K. Yoshino, Phys. Rev. Lett., 65, 1140-1143 (1990). T.W. Hagler, K. Pakbaz, K.F. Voss and A.J. Heeger, Phys. Rev. B, 44, 8652-8666 (1991). P.F. van Hutten, R.E. Gill, J.K. Herrema and G. Hadziioannou, J. Phys. Chem., 99, 3218-3224 (1995). 4. T. Shiga, A. Okada, H. Takahashi and T. Kurauchi, J. Mater. Sci. Lett., 14, 1754-1755 (1995). 5. T. Shiga, T. Ikawa and A. Okada, J. Appl. Polym. Sci., 67, 259-266 (1998). 6. Y. Inoue and Y. Kobatake, Kogyo Kagaku Zasshi, 61, 1108-1116 (1958). 7. T. Shiga, T. Narita, K. Tachi, A. Okada, H. Takahashi and T. Kurauchi, Polym. Eng. Sci., 37, 24-30 (1997). 8. T. Shiga, T. Narita, K. Tachi, M. Muramatsu and Y. Yazawa, Proc. 1997 Spring Meeting of Jpn Automotive Engineering Soc., No. 973, 17-20 (1997). 9. D.Y.Perera and D.Vanden Eynde, J. Coating Technol., 59, No. 748, 55-63 (1987). 10. T. Shiga, T. Narita, T. Ikawa and A. Okada, Polym. Eng. Sci., 38, 693-698 (1998). 11. T. Ikawa and T. Shiga, J. Appl. Polym.Sci., 83, 2600-2605 (2002).

This page intentionally left blank

Adhesion Aspects of Polymeric Coatings, Vol. 2, pp. 165–181 Ed. K.L. Mittal © VSP 2003

Adhesion of an alkyd paint to cold rolled steel sheets: Effect of steel surface composition S. MAEDA∗ Nihon Parkerizing Co. Ltd., 1-15-1, Nihonbashi, Chuo-ku, Tokyo 103-0027, Japan

Abstract—The cause of adhesion loss of an amino alkyd paint applied to bare cold-rolled steel sheets manufactured by continuous annealing has been studied using Auger electron spectroscopy(AES), X-ray photoelectron spectroscopy(XPS), Glow discharge spectroscopy(GDS) and zeta potential measurements. For comparison, the surface constituents of a batch-annealed sheet having a good paint adhesion were analyzed. The surface analyses showed the presence of significant amounts of calcium silicate on the surface of continuously annealed sheet. On the other hand, the outermost surface of batch annealed sheet was composed primarily of FeOOH besides carbonaceous contamination. The metal (Na and Ca) silicates were also observed on a paint-delaminated substrate surface; therefore, the deposited metal silicate on the steel surface seems to cause the adhesion failure of paint. The deposition of the silicate was due to poor water rinse after electrodegreasing by an alkaline solution prior to annealing. The zeta potential measurements showed that the Ca-silicate had basic properties, and Na-silicate is known to be basic. A possible explanation for the poor adhesion would appear to involve the loss of the acid-base interaction between the amino alkyd resin and the silicate-contaminated surface also with basic properties. Keywords: Paint adhesion; steel sheet; surface composition; silicate; acid-base interaction.

1. INTRODUCTION

Cold rolled steel sheets are basic materials used in the manufacture of everyday products such as electrical appliances, automobiles, office equipment, and containers. Most of them are usually used after an organic coating with a phosphate pretreatment. Many studies have been carried out to characterize the surface and microstructural features of low carbon cold rolled steels which influence paint adhesion and corrosion resistance [1–10]. These studies were mostly concerned with zinc phosphate pretreatment which has a major effect on paint performance, and have shown that both the surface carbon and elemental segregation such as Mn, Si, P, etc., occurred during heat treatment (annealing) in sheet processing line affected the phosphatability of the steel sheets. For example, the carbonaceous residues in∗

Phone: (81) 03-3278-4361, Fax: (81) 03-3278-4422, E-mail: [email protected]

S. Maeda

166

terfere with the deposition of an effective zinc phosphate conversion coating, resulting in an adverse effect on the corrosion protection of painted sheets [1–7]. On the other hand, surface manganese is known to enhance the phosphate reaction (formation of finer phosphate crystals), thereby providing a superior paint performance [8, 9]. Although in some uses, e.g., steel containers, paints are directly applied to the substrates without chemical pretreatment, very few studies have been made concerning this direct painting process. The aim of this study was to clarify the effect of steel surface composition on paintability, when paint was directly applied, by elucidating the cause of poor paint adhesion observed on a continuously annealed steel sheet. In this study, Glow discharge spectroscopy (GDS), Auger electron spectroscopy (AES) and X-ray photoelectron spectroscopy (XPS) were used to determine the detailed surface composition of the steel sheets. 2. EXPERIMENTAL

2.1. Materials Cold rolled steel sheets were taken from coils (before shipment) produced by continuous annealing line and conventional batch annealing. The chemical compositions of the steel sheets determined by atomic emission spectrometry are shown in Table 1. Table 1. Chemical compositions of steel sheets (wt%)

Sample 1∗ Sample 2∗∗

C

Si

Mn

P

S

Al

Fe

0.042

<0.003

0.24

0.006

0.012

0.016

Balance

0.22

0.013

0.015

0.030

Balance

0.056 <0.003 ∗ : Batch annealed steel sheet ∗∗ : Continuously annealed steel sheet

Figure 1 shows schematic views of cold processing lines for the steel coil production [10]. After pickling, the steel strip is rolled on a continuous cold rolling mill to the desired thickness (0.3–1.2 mm) while being sprayed with a rolling oil emulsion (coolant). Prior to annealing, the steel coils are subjected to electrodegreasing in an alkaline solution (sodium orthosilicate), followed by a hot water rinse to obtain a clean surface. Annealing for giving ductility to the steel is carried out in two different ways: batch (box) annealing in which a stack of coils is annealed for several hours at a soaking temperature of about 700 ºC, and continuous annealing in which a coil is continuously annealed at a soaking temperature of 800~900 ºC as it is unrolled. In

Adhesion of an alkyd paint to cold rolled steel sheets

Figure 1. Schematic views of cold processing lines for steel coils (a) advanced continuous annealing line, and (b) conventional batch (box) annealing.

167

168

S. Maeda

either case, a protective gas (a mixture of H2+N2) is used to prevent the steel from oxidizing during annealing. In the batch annealing process, the gas is allowed to pass through the furnace while the coils cool to about 130 ºC. After this, the steel coils are exposed to the air. Therefore, this process takes about 7 days to complete the annealing. On the other hand, in the continuous annealing process which has been developed by Japanese steel makers, the heated coil is force-cooled using a protective gas jet to around 100 ºC in the cooling zone, followed by water quenching. Therefore, the latter is far more advantageous since the annealing time takes 10 minute or less (including cooling and overaging times). In both processes, the annealed coils are then lightly cold rolled (ca.1% reduction) to adjust the mechanical properties of the steel (temper rolling), and then finally coated with a protective oil. 2.2. Paint adhesion After removing the protective oil in an organic solvent (methyl ethyl ketone, MEK), the sheets were spray coated with an amino alkyd paint (with green pigment) to a dry film thickness of 15 µm, followed by baking for 20 min at 160 ºC in an oven. The ATR-FTIR spectrum of the coated paint is shown in Figure 2, which indicates a typical oil-modified polyester resin, i.e., an alkyd (orthophthalic acid type)/urea-formaldehyde resin. The coated sheets were scribed by a knife, followed by deep drawing using the Ericksen cupping machine, and then subjected to a tape-pull test. 2.3. Surface analysis In order to investigate the cause of adhesion failure of the painted sheets, three kinds of surface analysis techniques (GDS, AES and XPS) were applied to the asproduced sheet surface and the paint-delaminated substrate surface. Prior to the surface analysis, the samples were degreased by an ultrasonic cleaning in MEK, but the paint failure substrate surface was measured in the as-delaminated state (without MEK cleaning). 2.3.1. Glow discharge spectroscopy Glow discharge spectroscopy (GDS) is a rapid depth profiling technique for a surface [11]. In this method, a Grimm-style “hollow anode” lamp is used. The sample is placed against the O-ring seal of the lamp where it forms the cathode. After a rotary pump-evacuation, a plasma gas (Ar+ ion) is bled through the anode at a pressure of a few Torr, then a DC voltage is applied between the sample and anode. The generated Ar+ ions cause a rapid sputtering of the sample surface. The sputtered materials are electronically highly excited by the plasma and the excited atoms emit light radiation consisting of the characteristic lines of the elements. The light emitted by the plasma is dispersed by a grating and the monochromatic rays of the elements to be analyzed are detected by electron photomultipliers. The measurement was performed using an RSV-ANALYMAT 2504 spectrometer under the following conditions:

Adhesion of an alkyd paint to cold rolled steel sheets

169

Figure 2. FTIR spectrum of the paint.

Anode tube diameter 4 mm (analysis area: 0.12 cm2), Ar gas pressure 8.2 Torr, and applied DC voltage 600 V (constant). 2.3.2. Auger electron spectroscopy (AES) The AES analysis was performed using a PHI 600 Micro-Auger Spectrometer. The primary electron beam was set at 3 kV with a beam current of 200 nA, and the spectra were measured with a CMA (Cylindrical Mirror Analyzer) having a resolution of ∆E/E of 0.6%. The operating pressure was 2x10-8 Torr and the measuring area was 10 µm x 20 µm rectangle. Ar+ ion sputtering was carried out under an emission current of 25 mA. The quantitative analysis was made using the peak-to-peak height measurements. The atomic concentrations were calculated from the signals based on known sensitivity factors relative to a Ag standard which are given by PHI charts, and the concentration of each element was represented as the atomic fractions (%) for the total elements detected [12]. 2.3.3. X-ray photoelectron spectroscopy XPS spectra were obtained using a Shimazu ESCA 850 spectrometer employing a Mg-Kα (1253.6 eV) X-ray source. The typical operating pressure was approximately 10-7 Torr and the area analyzed was 0.28 cm2. The spectrometer was calibrated to the 4f7/2 peak (83.8 eV) of Au and all binding energies were referenced to the hydrocarbon line at 285.0 eV. Calcium silicate powder (CaSiO3) was prepared as the standard sample to determine the binding energies of Ca and Si detected on the steel surface.

170

S. Maeda

2.4. Zeta potential measurements In order to investigate the acid-base properties of the surface components present on the sheet surfaces, fine dispersions (under 300 mesh with the Tyler sieve) of CaSiO3 and γ-FeOOH were prepared in 0.01 M KCl solutions for determination of their zeta potentials. The electrophoretic mobility was measured with a Coulter-Delsa 440 instrument at a solution conductivity of 1.3–1.5 mS/cm. 3. RESULTS AND DISCUSSION

3.1. Paint adhesion Figure 3 shows the paint adhesion results for two differently produced sheets (batch and continuously annealed sheets). An excellent paint adhesion was obtained on the batch-annealed steel sheet, while the adhesion to the continuously annealed sheet was very poor. The failure appeared visually to be interfacial. Since there is no difference in the bulk steel composition between the two sheets (Table 1), the remarkable difference in adhesion is attributed to their surface composition. 3.2. Surface analysis Figure 4 shows the elemental depth profiles of the surface obtained by GDS for the sheets with good and poor adhesion. The data for the interfacial substrate surface of the delaminated sample are also shown. Although the surface manganese at the good adhesion sample was greater than that at poor adhesion sample, a re-

Figure 3. Paint adhesion to steel sheets produced by different manufacturing processes.

Adhesion of an alkyd paint to cold rolled steel sheets

171

Figure 4. GDS in-depth profiles of as-received cold-rolled steel sheets with good (a) and poor (b) paint adhesion and the paint-delaminated substrate surface (c).

markable difference in the surface concentrations of Si and Ca was observed between the two sheet surfaces. It can be seen that the surface concentrations of Si and Ca are extraordinarily higher on the sheets with poor adhesion (continuous annealing) and the paint delaminated substrate surface. As stated later, the surface Si is in the form of silica and/or silicate originating from drag-out of the orthosilicate solution in the degreasing bath and Ca pick-up originates from the hardness of the water used for degreasing bath and postrinsing. It is well known that the surface silica is important to avoid the sticking between the turns during batch annealing, but it is not necessary for continuous annealing process. Figure 5 shows the AES spectra of the surfaces of both samples (except the delaminated surface), and the atomic concentrations of the detected elements for three samples (including the delaminated surface) are summarized in Figure 6. Besides carbon contamination and the main components O and Fe, only S was detected on the good adhesion sample; on the other hand, S, Ca and Si were detected on the poor adhesion samples (both the sheet surface as well as substrate failure surface). The presence of 6–7 at.% Si on the outermost surfaces that had poor adhesion as well as on the paint delaminated substrate surface is surprising. Calcium is also high on both sheet surfaces and the paint delaminated substrate surface. These results are in good agreement with the GDS measurement results.

172

S. Maeda

Figure 5. Auger spectra of as-received steel sheets with good and poor paint adhesion.

Sodium detected on the delaminated substrate surface originates from the cleaning solution (sodium orthosilicate) due to poor water rinse. Augustsson et al., who investigated the deposits on the sheet surface after post-rinse in a similar electrodegreasing line showed that the residual sodium was in the form of Na2SiO2(OH)2 [13]. A small amount of N detected on the delaminated substrate surface in Figure 6 seems to be due to the residue of the paint components.

Adhesion of an alkyd paint to cold rolled steel sheets

173

Figure 6. Atomic percentage of the surface elements determined by AES for the two sheets with good and poor paint adhesion and the paint-delaminated substrate surface.

The XPS spectra of C, O, Fe, Si and Ca acquired on the sheet surfaces are shown in Figures 7a and 7b. The C1s spectra of both sheets (good and poor adhesion) contain aliphatic (285.0 eV), ether (256.6 eV) and carboxyl (288.9 eV) spe-

174

S. Maeda

Figure 7a. C1s, O1s and Fe2p3/2 XPS spectra for the steel sheets with good and poor paint adhesion.

cies. The oxygenated C species are attributed to residual oil contamination. The O1s spectrum seems to contain 4 species: (1) O2– in the metal oxide such as Fe2O3 and MnO, (2) OH in FeOOH, (3) carbonyl (C=O) and (4) H2O and/or silicate.

Adhesion of an alkyd paint to cold rolled steel sheets

175

Figure 7b. Si2p and Ca2p XPS spectra for the steel sheets with good and poor paint adhesion.

It is well known that the cold rolled steel sheets(after annealing) are covered with γ-Fe2O3 films with Fe3O4 underlayer (total thickness: 6–12 nm) and that γFe2O3 is covered with a very thin FeOOH film (0.4–0.6 nm thickness) in air atmosphere [7, 14]. The decrease in the OH– peak after Ar+ ion sputtering in the O1s spectrum on the good adhesion sample (batch annealing) supports this fact. The structure of FeOOH was not clarified but the hydrated form of γ-Fe2O3 was supposed to be γ-FeOOH. The Ca2p and Si2p spectra acquired from the poor adhesion sample are compared with the spectrum of the CaSiO3 powder (control) in Figure 8. Judging from the binding energies of Ca and Si, the calcium and silicon species were identified as CaSiO3 in the outermost layer and the underlying layer seemed to be SiO2. The disagreement between Si and Ca profiles in GDS (Figure 4) is probably due to a preferential sputtering of Si and Ca because the sputtering rate of each element in GDS is known to differ considerably depending on the individual element due to very high sputtering speed by Ar+ ions [11, 16].

176

S. Maeda

Figure 8. Ca2p and Si2p XPS spectra of the CaSiO3 powder.

The calcium is closely associated with hardness of water used (tap water in bath and rinsing) in electrodegreasing process. It is said that the calcium ions are easily trapped in the gelatinous silicate films deposited on the surface [15], and, therefore, it could reasonably be assumed that the calcium reacted with the surface silica (SiO2) during the annealing, leading to the formation of calcium silicate (CaSiO3).

Adhesion of an alkyd paint to cold rolled steel sheets

177

As said before, sodium present on the fracture surface (see Figure 6) is due to poor water rinse in electrodegreasing process. In order to confirm this fact, the chemical analyses of both the degreasing bath and rinse water in the electrodegreasing process prior to annealing were carried out. The results are summarized in Figure 9. It was found that the final rinse tank was contaminated with ca.700 ppm of Na-silicate, which was brought in from the previous scrubber tank. Unfortunately, we have no data on the sodium and calcium contamination in the rinse tank of electrodegreasing for batch annealing process, but it was reported that de-ionized water was used for the final hot rinsing in electrodegreasing. In case of batch annealing, there is a possibility of reduction in deposited silicates by diffusion into steel substrate during long time annealing, but the diffusion of calcium silicate is unlikely to occur because of very low solubility limits of Ca (< 0.01%) in a steel matrix. Based on the results given above, it is possible to conclude that the surfacedeposited silicate inhibits the adhesion of paint, resulting in poor adhesion to the continuously annealed sheet. This conclusion was confirmed by the fact that the poor adhesion of the continuously annealed sheet considerably improved by water rinsing after electrodegreasing and by the final purified water quenching at the end of annealing (use of deionized water in place of tap water). Therefore, it should be noted that the poor paint adhesion is not inherent to continuous annealing process, but due to poor water rinsing in electrodegreasing prior to annealing. 3.3. Zeta potential measurements Figure 10 shows the measurement results of the zeta potential for CaSiO3 and γFeOOH powders dispersed in 0.01 M KCl solutions with different pHs. On an alkaline side, the γ-FeOOH is negatively charged whereas the CaSiO3 is positively charged below pH10. The acidity and basicity of solid surfaces can be determined qualitatively from the pH which corresponds to a zero zeta potential and this pH is usually called the isoelectric point, IEP. From the pH dependence of the zeta potential of each powder in Figure 10, the IEP of CaSiO3 is about 10 in pH units

Figure 9. Electrodegreasing process in the continuous annealing line.

S. Maeda

178

and that of γ-FeOOH is about 5.0. The IEP value near 5 for γ-FeOOH agrees with observations by others. For example, Parks has reported values of 5.3–7.4 for γFeOOH and 3.2–6.8 for α-FeOOH powders [17]. 3.4. Acid-base viewpoint [18] According to Fowkes [19, 20], the work of adhesion (W12) of a polymer to a metal (oxide) is determined by two factors, i.e., the London dispersion component of the surface energy and acid-base interaction as shown below:

d

W12 = 2(γ1dγ2d)1/2 + W12ab

(1)

W12ab= – fnab∆Hab

(2)

d

where γ1 and γ2 are the London dispersion components of the surface energies of materials 1 and 2, respectively. W12ab is the work of acid-base interactions. ∆Hab is the enthalpy of the acid–base interaction between two materials in contact, nab is number of acid-base bonds (mol/m2), and f is a constant for converting enthalpy into free energy, close to unity for an acid-base interaction [19]. On the other hand, van Oss, Chaudhury and Good have proposed the following interesting relation by using the contact angle data only [21, 22]: W12 = 2{(γ1LWγ2LW)1/2 + (γ1+γ2–)1/2 + (γ1–γ2+)1/2}

(3)

Figure 10. Zeta potential measurements as a function of pH for CaSiO3 and FeOOH suspensions.

Adhesion of an alkyd paint to cold rolled steel sheets

179

where γLW represents the Lifshitz–van der Waals (LW) component of surface energy which consists of three components: the London dispersion, Keesom dipoledipole and Debye dipole-induced dipole. γ+ is the acid (electron-acceptor) surface energy component, and γ– is the base (electron-donor) surface energy component. It should be noted that in a condensed phase the contributions of dipole-dipole attraction and dipole-induced dipole attraction are much smaller compared to the dispersion force [23]. According to Bolger, the oxyhydroxides of metals MOOH (M: metal) formed on a metal surface have both acidic and basic surface sites, and act both as an acid or a base depending on the ambient conditions [24]. Fowkes also showed by flow microcalorimetry, FTIR and zeta potential measurements that Fe-OH surface sites of the γ-ferric oxide could react with either acidic or basic molecules [25, 26]. Bolger’s concept can be extended to the acid-base interactions in the sense as the Lewis theory according to which bases have electron pair donor properties and acidic materials are electron pair acceptors [25]. The interfacial acid-base interactions require the acidic sites of one phase to interact with the basic sites of the other, so that if one of the phases is neutral or if both phases have only basic or only acidic sites, there can be no acid-base interaction. As is well known, the acidity or basicity of solid surfaces can be determined from the pH which corresponds to the zero zeta potential (isoelectric point, IEP). According to Fowkes, FeOOH has an IEP value of about 7 [26], and as stated above, it could react with either acid (proton-donor or electron-acceptor) or base (proton-acceptor or electron-donor). Our results showed that the IEP value for FeOOH was about 5 (weak acid). On the other hand, the IEP value of CaSiO3 was about 10, which appears to have a basic property. Sodium silicate is also a basic material. Furthermore, Liao and Senna [27] showed using the benzoic acid titration method that calcium silicate produced by the mechanochemical reaction (grinding) of the Ca(OH)2–SiO2 mixture had strong surface basicity. Interestingly, the calcium silicate is known to readily react with poly(vinyl chloride) (acidic polymer), leading to the formation of a cross-linked product. On the other hand, an amino alkyd paint is considered to have basic properties because it has a lone pair of electrons (electron-donor). Therefore, the acid-base interactions are unable to occur between the amino alkyd paint and metal silicateenriched steel surface, thus resulting in very poor adhesion on the continuously annealed sheet. Since the surface composition of the delaminated paint side was not clarified in this experiment, the failure mode including in a weak boundary layer caused by metal silicates cannot be excluded. However, according to GDS measurements, the thickness of Si and Ca layers on the delaminated substrate surface and the asreceived sheet surface was almost the same, and the thickness for both elements was very thin (ca.10–20 nm), which was comparable to the surface oxide films.

180

S. Maeda

Therefore, we presumed that the paint failure mainly occurred at the interface between the paint and silicate-contaminated surface. 4. CONCLUSIONS

The cause of paint adhesion loss on cold rolled steel sheets produced by continuous annealing was investigated using various surface analysis techniques (GDS, AES and XPS) and zeta potential measurements, and the following conclusions were drawn; (1) Adhesion of amino-alkyd paint applied to the bare surface of steel sheets was very dependent on the nature of steel substrate. Poor adhesion on the continuously annealed sheet was attributed to the presence of Ca and Na silicates. (2) The silicate on the sheet surface was caused by the deposition of ortho-silicate from the electrodegreasing bath due to poor hot water rinsing in the sheets production line. Accordingly, the deterioration in paint adhesion in this case has been significantly improved by replacing the tap water with deionized water for the post-rinse. (3) According to zeta potential measurements, the metal silicate-enriched sheet surface was considered to be very basic. Therefore, the adhesion failure on this surface could be attributed to the lack of the acid-base interactions with the basic amino-alkyd resin. REFERENCES 1. V. Hospadruk, J. Haff, R.W. Zurilla and H.T. Greenwood, Soc. Automotive Engineers, Paper No. 780186 (1978). 2. J.J. Wojtkowiak and H.S. Bender, J. Coating Technol., 50 (642), 86 (1978). 3. T.W. Fisher, R.A. Iezzi and J.M. Madritch, Soc. Automotive Engineers, Paper No. 800149 (1980). 4. P.L. Coduti, Metal Finish., 78 (5), 5 (1980). 5. R.A. Iezzi and H. Leidheiser, Jr., Corrosion, 37, 26 (1981). 6. J.A. Kargel and D.L. Jordan, Corrosion, 38, 201 (1982). 7. S. Maeda, J. Coatings Technol., 55(707), 43 (1983). 8. P.-E. Augustsson, T. Olefjord and Y. Olefjord, Werkst. Korros., 34, 563 (1983). 9. L. Schon, W. Leijon and P.-E. Augustsson, Scad. J. Metallurgy, 13, 220 (1984). 10. From Nippon Steel Catalogue. 11. R. Berneron and J.C. Chabonnier, Surf. Interf. Anal. 3, 134 (1981). 12. L.E. Davis, N.C. MacDonald, P.W. Palmberg, G.E. Riach and R.E. Weber, Handbook of Auger Electron Spectroscopy, 2nd. ed., Physical Electronics Division, Perkin-Elmer Corp., Eden Prairie, MN (1976). 13. P.E. Augustsson, T. Olefjord and Y. Olefjord, Surf. Eng., 3, 276 (1986). 14. S. Maeda, Prog. Organic Coatings, 11, 1 (1983). 15. V. Leroy, Mater. Sci. Eng., 42, 289 (1980). 16. R. Payling and D.G. Jones, Surf. Interf. Anal., 20, 783 (1993). 17. G.A. Parks, Chem. Rev., 65, 177 (1965).

Adhesion of an alkyd paint to cold rolled steel sheets

181

18. K.L. Mittal and H.R. Anderson, Jr. (Eds), Acid-Base Interactions: Relevance to Adhesion Science and Technology, VSP, Utrecht (1991). 19. F.M. Fowkes and M.A. Mostafa, Ind. Eng. Chem. Res. Dev., 17, 3 (1978). 20. F.M. Fowkes, J. Adhesion Sci. Technol., 4, 669 (1990). 21. C.J. van Oss, M.K. Chaudhury and R.J. Good, Chem. Rev., 88, 927 (1988). 22. R.J. Good and A.K. Hawa, J. Adhesion, 63, 5 (1997). 23. R.J. Good, in: Contact Angle, Wettability and Adhesion, K.L. Mittal (Ed.), p. 3, VSP, Utrecht (1993). 24. J.C. Bolger, in: Adhesion Aspects of Polymeric Coatings, K.L. Mittal (Ed.), p. 3, Plenum Press, New York (1983). 25. F.M. Fowkes, in: Physicochemical Aspects of Polymer Surfaces, K.L. Mittal (Ed.), Vol. 2, p. 583, Plenum Press, New York (1983). 26. F.M. Fowkes, Y.C. Huang, B.A. Shah, M.J. Kulp and T.B. Lloyd, Colloids Surfaces, 29, 243 (1988). 27. J. Liao and M. Senna, Solid State Ionics, 66, 313 (1993).

This page intentionally left blank

Adhesion Aspects of Polymeric Coatings, Vol. 2, pp. 183–201 Ed. K.L. Mittal © VSP 2003

Analysis of the wet adhesion of coatings on wood MARI DE MEIJER∗ SHR Timber Research, P.O. Box 497, 6700 AL Wageningen, The Netherlands

Abstract—Maintaining good adhesion of coatings under wet conditions is of great importance for moisture sensitive substrates like wood. This paper describes a newly developed method to measure the adhesion of coatings on wood based on a peel test method. By analysing the peel force as a function of the surface structure of the wood, the influence of mechanical anchoring on the adhesion was demonstrated. With this method several model coating formulations based on waterborne acrylic dispersions, alkyd emulsions and solventborne alkyd solutions were compared. The measured adhesion is further analysed with respect to contributions of interfacial work of adhesion and the reduction in adhesion due to internal stress. The work of adhesion was calculated both according to the Lifshitz-van der Waals acid-base theory and from polar and dispersive components. The calculated work of adhesion does not correlate with the observed differences in practical adhesion. The wettability of the wood by the coating is, however, a good predictor for the actual adhesion. For coatings that show a strong swelling at high moisture content the resulting hygroscopic strain causes high level of stored internal stress which has a negative impact on adhesion. The adhesion is related to mechanical anchoring of the coating to the wood. Therefore, the adhesion is controlled by the wetting of the surface and the penetration into the substrate. The viscosity of the coating, especially as a function of solids content, has a major influence on the degree of substrate penetration. Finally, it is briefly addressed how the wettability of wood can be improved by flameionisation treatment. Keywords: Peel test; wood; coating; mechanical anchoring; wettability; stress; flame-ionisation treatment.

1. INTRODUCTION

For good protection of wood by a coating, the adhesion of the coating to the wood is of great importance. Because both wood and coating are hygroscopic materials, the adhesion should be maintained at high moisture content. This so-called wet adhesion is often lower for waterborne coatings compared to traditional solventborne types [1-5]. Several reasons may cause the lower adhesion such as less substrate penetration [6-8], higher moisture uptake in the coating and the surface of

∗

Correspondence should be sent at: Drywood Coatings, P.O. Box 3954, 7500 DZ Enschede, The Netherlands. Phone: +31 53 433 44 22, Fax: +31 53 430 68 32, E-mail: [email protected]

M. de Meijer

184

the wood [9-11], and the presence of surfactants in waterborne coatings that might accumulate at the interface and reduce adhesion [12-14]. The aspects particularly relevant to the adhesion to wood are reviewed elsewhere [15-17] and will only be discussed briefly here. Wood has a rough surface [18] and a porous structure [19] with a typical porosity of 55-70% depending on specific gravity and moisture content [20, 21]. Therefore, good surface wetting and capillary penetration are generally considered a prerequisite for good adhesion of a coating. They increase the contact area between the substrate and the adhering material and reduce the concentration of stresses on specific points during adhesion measurement [15]. The spreading of a liquid on wood is not only influenced by surface energetics but also by capillary penetration into the substrate [22, 23]. The viscosity of the liquid has a significant influence on this as well [24, 25]. The penetration of a liquid into a pore with the capillary pressure as driving force can be described by the Washburn equation [25, 26]. dh 2 γ L cosθr −ρgπhr 2 = dt 8ηh

if cosθ> 0 (θ< 90°)

(1)

The rate of capillary rise (dh/dt) is proportional to the capillary radius r, the liquid surface tension γL and the contact angle between the liquid and the wood θ and is inversely proportional to the height (or distance) of capillary rise h and the viscosity of the liquid η. The term ρgπhr2 is a minor gravity correction factor, only applicable for penetration into vertical direction. The relationship between liquid surface tension and contact angle is given by the Young equation: γ L cosθ = γ S − γ SL

if cosθ<1 ( θ> 0°)

(2)

The term γL cosθ is determined by the difference between the surface tension of the solid (in equilibrium with air) (γS) and the interfacial tension (γSL) between the wood (solid) and the coating (liquid). Most of the above-mentioned factors should not be considered as constant due to fluctuations in time and position within the wood. Firstly, the radius of the wood capillaries will vary due to differences in wood anatomy which has been shown to influence coating penetration [6-8, 43]. Secondly, the surface energy or the contact angle might change due to fluctuations in wood moisture content [24, 44] and dissolution of wood subtances into the coating or selective adsorption of coating components [45]. Thirdly, the viscosity will change due to selective uptake of water or organic solvent from the coating inside the capillaries into the wood cell wall which is impermeable for the larger binder or pigment particles of the coating [25]. The measurement of surface energies for wood was initially based on measuring critical surface tensions (γc) and was found to be around 50 mJ/m2 [24, 27]; this was later followed by measurements of polar (γP) and dispersive (γD ) or nonpolar energy components of the surface energy according to either the geometric or harmonic mean method [24, 28]. The surface free energy of softwoods ranged between 48 and 61 mJ/m2 with a polar fraction of 25 to 40%. More recently, the

Analysis of the wet adhesion of coatings on wood

185

Lifshitz-van der Waals (γLW) and (Lewis) acid-base components (γAB) were used to measure the surface energy [29, 30]. On softwood, the Lifshitz-van der Waals forces are dominant (γLW approx. 40-50 mJ/m2) in comparison to acid-base interactions (γAB approx. 5-8 mJ/m2). This approach also allows the calculation of the work of adhesion between two solids [31]. It is known that the so-called practical adhesion [46-48] is not only determined by the interfacial adhesion between the coating and the substrate. In a peel test experiment, used here, the measured work of peeling (WP) is a function of: interfacial work of adhesion (WaCW; work of adhesion between coating (C) and wood (W)), work expended in plastic deformation during peeling (Wd) and elastic energy stored in the coating because of internal strain (β). On wood the internal strain is a function of the difference in hygroscopic expansion α (swelling) between the coating and the wood for a given change in relative humidity. The relationship between these factors can be described by the following equation [32, 33]: a W p = WCW +Wd − β

(3)

Most of the existing quantitative methods to measure adhesion are based on measuring the force to remove a dolly, glued to a coating in either direct pull-off (ISO 4624) or torque mode [2]. These methods are often difficult to interpret because cohesive failure can occur in the coating, the glue or the wood [34]. Furthermore, the measured force is influenced by the stress distribution under the dolly and the way of cutting the coating around the stamp [35, 36]. The widely used cross-hatch (ISO 2409) or X-cut test (ASTM D3359) is very useful to distinguish between good and poor adhesion but does not give a true quantitative adhesion strength. Furthermore, the outcome of this test is very much dependent on the type of tape used and the speed and peel angle during removal of the tape. The objective of this work was to develop a method to measure quantitatively the adhesion between a coating and wood, in such a way that the measured adhesion strength could be further analysed. Special attention has been given to the relation between adhesion strength, surface free energy or work of adhesion and the microscopic structure of the wood. This gives an opportunity to understand whether the adhesion of the coating is improved by penetration into the substrate. 2. MATERIALS AND METHODS

Pine sapwood (Pinus sylvestris) with an average density of 470 kg m–3 was used in this study. The wood was cut and planed into boards with the convex side of the annual rings parallel to the test surface (tangential section). Growth-ring width was between 2 and 6 mm and the wood was conditioned to an initial moisture content of around 14%. Six model coatings based on commercially available polymeric binders were used in this study. All coatings were pigmented with titanium dioxide to a pigment volume concentration of about 17%. The emulsion and

M. de Meijer

186

dispersion paints were waterbased and the high solid or solventborne alkyd paint were dissolved in white spirit. Table 1 provides a brief overview of their characteristics. From previous studies [6], it is known that there is a wide variation in penetration into the substrate. Coatings were applied with a blade applicator at a wet film thickness of approximately 300 µm. This was done at least two weeks after planing of the wood and adhesion tests were made at least two weeks after application of the coating. The adhesion measurement is based on peeling the coating from the wood by a pressure sensitive tape with a width of 20 mm as shown in Fig. 1. The tape was Table 1. Description of coatings used Coating

Binder type

Wood penetration

E-modulus Surface (a) tension N/mm2 mN m–1

Viscosity Pa.s at 1 s–1

Ac1 Ac2 Ac3 WBA HSA SBA

acrylic dispersion acrylic emulsion acrylic emulsion alkyd emulsion high solid alkyd solventborne alkyd

very limited very limited very limited moderate very deep deep

77 106 209 66 202 260

15.20 4.17 19.10 2.90 6.97 1.61

(a): of liquid coatings measured according to DIN 53419

Figure 1. Peel-test sample configuration.

39.3 32.4 35.8 39.2 30.3 28.2

Analysis of the wet adhesion of coatings on wood

187

removed by a tensile testing machine, which allows measuring the force against the peel-length. The tape was peeled at a rate of 30 mm/min at an angle of 180° which is the easiest angle to maintain constant during the experiment. From a large number of commercial products, the 3M tape no. 396 was found to be most suitable. On aluminium, this tape has an adhesion strength of 600 N/m. The tape was applied by hand followed by pressing it onto the coating with a wheel with a fixed weight. In order to avoid peeling beyond the area of the tape, incisions through the coating were made directly adjacent to the coating. A V-shaped incision was made in the coating under the tape, to have a fixed starting point for delamination between the coating and the wood. Before application of the tape, the coating was cleaned with a 10% ammonia solution to improve adhesion of the tape to the coating. This did not influence the adhesion of the coating to the wood. For wet adhesion measurements, the interface between the coating and the wood was wetted via grooves on both sides of the test area by exposure to water for one day or air with 98% relative humidity for approximately one week. After peeling the coating from the wood, the detached surfaces were examined visually and by light- and electron microscopy. From the measured peel force, the work of peeling was calculated as follows [37]: æFö ç ÷ F w æ ö W p = ç ÷ (1− cos θ) + è ø w 2tE è ø

2

(4)

The work of peeling is a function of the measured force per unit area of tapewidth (F/w), the peel angle (θ), the elastic modulus of the coating (E) and the average coating thickness (t). The elastic energy (β) due to stored hygroscopic strain after exposure to water was calculated as follows [38]:

(∆αcoating −∆αwood ) β = t ⋅ E⋅ 1−ν

2

(5)

The hygroscopic expansion of the coating (∆α) was measured directly by volumetric swelling from equilibrium at 65% RH to immersion in water. The volume of a free coating film was calculated from the upward force of the submersed freehanging film in water in analogy to the determination of the volume percentage solid in a coating (ISO-standard 3233). Poisson’s ratio (ν) was assumed to be around 0.5 [39], modulus of elasticity (E) is given in Table 1 and the average film thickness was calculated from the solids content based on volume. The hygroscopic strain calculated from equation (5) is given as the strain parallel to the direction of peeling. This is due to the anisotropic expansion of the wood, which was around 7% in tangential direction as determined in previous experiments and essentially zero in the longidutional direction of the wood (perpendicular to the direction of peeling). The swelling of the coating was found to be isotropic.

M. de Meijer

188

Advancing contact angles (sessile drops) of the liquid coating on the wood were measured by a video-technique [40] to assess the wetting of substrates. Advancing contact angles of water, diiodomethane and formamide (all reagent grade) were measured on both wood and coatings. The Lifshitz-van der Waals (γLW), acid (γ+) and base (γ–) solid surface (γS) free energy parameters were calculated according to the method described by Good [31]. Polar (γP) and non-polar or dispersive (γD) components of surface energies were calculated according to Owens and Wendt [41]. From the surface free energies of both coating (γC) and wood (γW), the work of adhesion (WaCW) between the two phases was calculated according to the following equation [31] using Lifshitz-van der Waals acid-base parameters: a WCW = γ C + γ W −γ CW a WCW =2

(

LW γ CLW γ W + γ C+ γ −W + γ C− γ +W

(6)

)

(7)

The interfacial work of adhesion after exposure to water (WaWET) was obtained from the interfacial energies between coating-water (γCL), wood-water (γWL) and coating-wood (γCW) as follows: a WWET = γ CL + γ WL −γ CW

(8)

If the surface energy is expressed in terms of Lifshitz-van der Waals and acida base interactions, WWET can be obtained from the following equation [12, 42] in which γL refers to water: a WWET = −2

(

LW γ LLW − γ W

− 2 é γ +L ëê

(

)(

) )+ γ (

γ CLW − γ LLW

γ −W + γ C− − γ −L

− L

)

(9)

− +ù + − γ +W + γ C+ − γ +L − γ W γC − γ W γC ûú

(

)

a DP in the presThe work of adhesion with polar and dispersive components WWET ence of water was calculated according to equation (8). The geometric mean method was used to calculate the polar and dispersive interactions between the coating, the wood and water as shown in equation (10).

γ CL = γ C + γ L − 2 γ CD γ LD − 2 γ CP γ LP ;

D D P P γ WL = γ W +γ L − 2 γ W γL −2 γW γL ;

D P γ CW = γ C + γ W − 2 γ CD γ W − 2 γ CP γ W

(10)

Inserting equation (10) into (8) gives:

(

a DP P P D D P P D D P P WWET = 2 γ L − γ CD γ D L − γCγ L − γ W γ L − γ W γ L + γC γ W + γCγ W

)

(11)

Analysis of the wet adhesion of coatings on wood

189

Flame-ionisation treatments of wood were carried out with a laboratory flameionisation unit (Sherman UK) at a stoichiometric ratio of natural gas to air of 1:10 with four passes for each treatment. 3. RESULTS AND DISCUSSION

3.1. Peel tests and fracture mode Initially, the tape peel test was applied to both dry and wet samples but only the latter appeared to be successful because of the much lower adhesion of the coating to the wood. On dry wood, the adhesion of the coating to the wood was too strong, resulting in failure between the coating and the tape. On wet wood, the adhesion between the tape and the coating was strong enough to measure the adhesion between the coating and the wood. A typical plot of peel force against peeled distance with the corresponding structure of the wood under the peeled area is shown in Fig. 2. The adhesion force (F) measured is expressed in Newton (N), divided by the width (w) of the tape in metres (m). This is equal to energy per unit area (J m–2) because J = N.m. By using this unit, the measured adhesion can be expressed in the same units as the interfacial work of adhesion. The work of peeling is given in Table 2.

Figure 2. Adhesion as a function of peeled distance on pine sapwood; inset photograph of wood structure shows the corresponding early- and latewood zones.

M. de Meijer

190

Table 2. Work of peel adhesion (WP), interfacial work of adhesion (Wa) and internal strain energy (β), all expressed in J m–2 Wp

Coating Ac1 Ac2 Ac3 WBA HSA SBA

EW

LW

152 142 156 238 682 580

107 115 110 126 200 255

β 28 12341 7415 25 74 96

a WCW dry

0.087 0.096 0.094 0.098 0.093 0.093

a LW-AB Wwet

0.020 0.010 0.050 0.019 0.014 0.030

a DP Wwet

0.084 0.082 0.079 0.083 0.084 0.086

The following data for water were used (mJ m–2): γLW 21.8; γ+ 25.5; γ– 25.5; γD 21.8; γP 51.0 EW: earlywood LW: latewood CW: coating-wood LW/AB: Lifshitz-van der Waal/acid-base DP: dispersive-polar

With all measurements, typical increasing and decreasing peel forces were observed. If the peeled distances were compared with the structure of the wood, it was evident that the highest peel forces corresponded with the earlywood zones with the deepest substrate penetration and that the lowest peel forces with the latewood zones with the lowest substrate penetration [6, 7]. Fig. 3 shows an SEMmicrophotograph of the wood-coating interface, illustrating the larger contact area in the earlywood zones. Surface roughness in early- and latewood was also different as can be seen from Fig. 4. Several experiments, in which the orientation and width of growth rings was varied, showed corresponding changes in peel forces. If the same test was performed on a flat metal surface, no fluctuation in peel forces was observed. The measured peel forces for all six coatings tested on pine sapwood and exposed to liquid water are given in Fig. 5. All three acrylic (Acl-Ac3) coatings showed similar adhesion values and differences between early and latewood were also comparable. The adhesion to the earlywood was in most cases 50 percent higher than to the latewood. Adhesion of the waterborne alkyd coating (WBA) was somewhat higher in comparison to Ac1-Ac3, in particular to the earlywood. Adhesion of the SBA and HSA paint was the highest; adhesion to the earlywood was 2.5 to 3.4 times higher than to the latewood. All these differences were in very good agreement with differences in penetration between coatings, and between early- and latewood. The differences between pine sapwood samples exposed to liquid water and water vapour are shown in Fig. 6. Only results for the acrylic coatings are presented, because with the other coatings the failure occurred between the coating and the tape. This clearly indicated a better adhesion of the WBA, HSA and SBA

Analysis of the wet adhesion of coatings on wood

191

Figure 3. SEM image showing in a transverse section the influence of the wood structure on the penetration of the coating.

Figure 4. SEM micrograph of a tangential section of the wood surface showing differences in surface roughness between earlywood (right) and latewood (left).

192

M. de Meijer

Figure 5. Peel adhesion strength of various coatings on pine sapwood after exposure to liquid water.

Figure 6. Difference in peel adhesion strength between pine sapwood samples exposed to liquid water and water vapour (98% RH).

Analysis of the wet adhesion of coatings on wood

193

coatings after exposure to water vapour. The adhesion of the acrylic coatings was very much higher if the samples were exposed to water vapour. Apparently, the presence of liquid water at the wood-coating interface reduced the adhesion strength further, although the wood moisture content was only slightly lower (2528% instead of 30-32%). After exposure to water vapour, larger differences between acrylic coatings and between early- and latewood were observed. After removal of the coating from the wood, the detached surfaces of the wood and the coating were examined by scanning electron microscopy. On the wood face of the coating clear imprints of the wood structure were visible (see Fig. 7). Parts of the coatings were clearly torn out of the wood cells. Fractured pieces of coating material were observed in the wood, especially with the better adhering HSA and SBA coatings. Cohesive failure in the coating was observed in two ways. Firstly as a true cohesive failure with the coating being still in its original position in the wood (Fig. 7 ), and secondly as a cohesive failure after the coating had been partly torn out of the wood (Fig. 8). The measured peel work of adhesion was also compared with the results of the more practical ASTM D 3359 crosshatch and X-cut tests. Both results correlate rather well as can be seen in Fig. 9. The x-cut test resulted in a lower adhesion compared to the crosshatch test.

Figure 7. SEM image of the alkyd emulsion coating detached from the wood, note that coating material is torn out of the wood.

194

M. de Meijer

Figure 8. SEM image of cohesive failure with the coating (HSA) still in its original position in the wood.

Figure 9. Correlation between quantitative data obtained with peel-test and results obtained according to ASTM D3359 adhesion tests on wood with moisture content between 25 and 30%.

Analysis of the wet adhesion of coatings on wood

195

3.2. Work of adhesion and stored strain energy The contact angles for the three test liquids on both coating and wood are given in Table 3. Due to surface roughness contact angles differed depending on whether they were taken along or across the grain orientation. The data obtained along the grain direction were taken as these might be considered most reliable [30]. From the contact angles, interfacial work of adhesion between the coating and the wood was calculated for both dry and wet conditions (Lifshitz-van der Waals acid-base approach) or wet conditions only (dispersive-polar approach) and are given in Table 2. For all coatings, regardless of the method of calculation, the work of adhesion was always positive. No correlation between the work of adhesion under any condition and the measured difference in adhesion could be observed. The maximum volumetric swelling of pine sapwood between 65% RH and immersion in liquid water was around 7%. The alkyd based coatings clearly swelled less than the wood (2.6-2.7%) whereas the acrylic coatings swelled more than the wood. Coating Ac1 swelled 11%, Ac2 swelled 78% and Ac3 swelled 46%. The stored elastic energy due to internal strain (β) from differential swelling between the coating and the wood was calculated according to equation (5) and is given in Table 2. The stored strain energy due to internal strain in the coatings Ac1, WBA, SBA and HSA was lower than the measured adhesion values. This was, however, not the case for the coatings Ac2 and Ac3, which had a very high stored elastic strain energy because the swelling of the coating was much higher than that of the wood. Theoretically, if βWP the coating should detach from the wood spontaneously. As this did not happen it might be due to three reasons. Firstly the hygroscopic stress might have relaxed very rapidly before the adhesion measurements were made. Secondly, it should be noted that the hygroscopic

Table 3. Contact angles and surface energies and their components for coatings and wood Coating

Ac1 Ac2 Ac3 WBA HSA SBA

Contact angles (°)

Surface energy (mJ m–2)

Diiodomethane

Distilled Formawater mide

γSLW

γS+

γS−

γS a

γSD

γSP

γS b

48 47 49 41 53 48

73 68 55 64 80 77

61 64 53 42 77 46

35 36 35 39 32 36

0.01 0.29 0.02 0.86 0.95 1.99

13.5 24.5 33.6 13.6 17.6 3.8

36 41 36 46 41 41

28 25 23 33 23 34

8.6 12.2 21.8 12.5 6.8 6.2

37 37 45 46 30 40

41

50

22

39

2.57

19.7

53

32

21.4

53

Wood Pine sapwood a

γS = γ LW + 2 γS+ γS−

b

γ S = γ SD + γ SP

196

M. de Meijer

stress is calculated from the maximum swelling of the coating that could be considerably higher than the swelling occuring on wet wood because the wood may constrain the swelling of the coating. Thirdly, a spontaneous detachment of the coating might be prevented by mechanical anchoring effects, due to some, but limited, penetration into the substrate. It should also be noted that the differences between coatings Ac1 to Ac3 exposed to water vapour were fairly consistent with the differences in the hygroscopic stress. The coating with the lowest stress (Ac1) showed a much better adhesion than the ones with the higher internal hygroscopic stress (Ac2 and Ac3). The work expended in plastic deformation (Wd) can be neglected in this study, since the forces used to peel the coating from the wood are clearly below the range where these coatings start to deform plastically. The plastic deformation of the coatings starts at forces of 2.6 N/mm2 or higher whereas the force applied to the cross section of the coating is approximately 0.4 N/mm2 (acrylic coatings) to 2.1 N/mm2 (solventborne alkyd). 3.3. Substrate wetting and penetration The contact angles for all the coatings on early and latewood areas are given in Fig. 10 as a function of time. All coatings reached equilibrium angles approximately 50 s after depositing the droplet on the surface, except for the HSA coating which reached equilibrium somewhat quicker. Clear differences in contact angles existed between the coatings. The acrylics (Ac1-Ac3) had the poorest wetting and spreading followed by coatings WBA, SBA and HSA. If contact angles were measured separately on early- and latewood zones (pine sapwood, tangential surface) the former showed a somewhat better wetting. The differences in wetting were consistent with previous findings [6] on the penetration of coatings in the outer cell layers of the wood. The measured adhesion strength shows a good correlation with the equilibrium contact angle of the coating as shown in Fig. 11. Because the adhesion increased with better wetting and substrate penetration, the correlation between liquid surface tension or viscosity and adhesion was evaluated. A good correlation between these values was not found. This can, however, be explained that for most coatings the penetration is not limited by the initial viscosity but by the increase in viscosity with solid matter content [25]. During the capillary penetration of the coating into the wood cells, water or solvent is selectively taken up into the cell wall, hence increasing the solids content of the coating. An example of the differences in viscosity – solids content relationship between a waterborne acrylic dispersion, a waterborne alkyd emulsion and a solventborne alkyd resin is shown in Fig. 12. For those wood species where the wettability is the limiting factor in achieving good wetting and subsequent substrate penetration (γs<γL) the surface energy of the wood might be increased by flame-ionisation of the surface. Some preliminary results from ongoing experiments showing an improved spreading of water on oak after flame-ionisation of a several weeks old surface are given in Fig. 13.

Analysis of the wet adhesion of coatings on wood

197

Figure 10. Contact angles of coatings on early- (EW) and latewood (LW) areas of tangential surface of pine sapwood.

Figure 11. Relationship between peel energy and contact angles of coatings on wood.

198

M. de Meijer

Figure 12. Relative viscosity at a shear rate of 0.01 s–1of an acrylic dispersion, alkyd emulsion and a solventborne alkyd binder as a function of binder content.

Figure 13. Contact angle of water as a function of time on oak heartwood with and without a flame ionisation treatment with a natural gas-air ratio of 1:10 and four passes.

Analysis of the wet adhesion of coatings on wood

199

4. CONCLUSIONS

With the developed tape peel method, the adhesion of coatings to wood can be measured up to adhesion energies of 300 to 400 J m–2. By comparing differences in peel-strength with the wood microstructure, it is evident that penetration of the coating into the pores of the wood improves adhesion. This means that mechanical anchoring and enlargement of the wood-coating contact area contribute to achieving a good adhesion. The results of this method correlate well with more practical adhesion tests as described in ASTM D3359. For the coatings studied, the method was limited to the measurement of wet adhesion, which was very much lower compared to the dry adhesion. The six coatings studied showed very clear differences in wet adhesion strength. Water vapour reduced the adhesion strength, and the reduction was the strongest with liquid water. The critical moisture content in the interface lies around 25% for the acrylic coating and around 30% for the other coatings. An evaluation of the work of adhesion after exposure to water showed no reasons for reduced adhesion. The lower wet adhesion can partially be explained by the stored strain energy due to internal stresses. The poorest adhesion was observed if the swelling of the coating was very much higher than the swelling of the wood. Weak boundary layers might also contribute to the decreased adhesion since the waterborne alkyd emulsion showed a decrease in wet adhesion strength that could not be explained by stored strain energy. For a more detailled quantitative understanding of the adhesion two other factors might also be considered in future work. The first one is the possible viscoelastic energy dissipation within the coating. This might be of particular interest for non-crosslinked coatings, like the ones based on physically drying acrylic dispersions. The second one is to distinguish between the energy dissipated in the rupture of the mechanical links between the coating and the wood and the cohesive energy of cleaving the parts of the coating that might remain in the wood. To obtain good adhesion of a coating to wood there are two critical factors. Firstly, the coating should wet the wood cells properly and to a certain extent penetrate into the open pores in the earlywood. Secondly, the hygroscopic expansion of the coating should be equal to that of the wood to minimise internal stress in the coating. A low viscosity of the coating at increased solids content will improve capillary penetration and hence adhesion. For those wood species that have a poor wettability due to low surface energy, increasing the surface energy by flame-ionisation should improve wetting of the coating. Acknowledgements This research was financed by the Dutch Innovative Research Program on Coatings (IOP-verf). Furthermore, technical and financial support was given by Akzo Nobel Coatings, Sigma Coatings, DSM Resins and Johnson Polymer. The author would like to thank Mark Scheltes, Barend van de Velde, Wiro Cobben and

200

M. de Meijer

Katharina Thurich for the experimental work and Dr. Ronald van der Nesse for his valuable discussions about the initial set-up of this test method. Gratitude is expressed towards the Wageningen Agricultural University Department of Plantcytology and Plantmorphology for the use of the SEM. REFERENCES M.L. Jansen, J. Oil Colour Chem. Assoc. 5, 117-128 (1986). S.L. Bardage and J. Bjurman, J. Coatings Technol. 70 (878), 39-47 (1998). P. Ahola, Materials and Structures 28, 350-356 (1995). G. Hora, A. Belz, D. Kruse and A. Avarlik, Proc. Conference on Advances in Exterior Wood Coatings and CEN-standardisation, paper 2, Paint Res. Assoc., Teddington, UK (1998). 5. K. Bosschaart-Thurich, U. Pathak and L. Berkhout, Proc. Conference on Advances in Exterior Wood Coatings and CEN-standardisation, paper 26, Paint Res. Assoc., Teddington, UK (1998). 6. M. de Meijer, K. Thurich and H. Militz, Wood Sci. Technol. 32, 347-365 (1998). 7. R.M. Nussbaum, Holz als Roh- und Werkstoff 52, 389-393 (1994). 8. R.M. Nussbaum, E.J. Sutcliffe and A.C. Hellgren, J. Coatings Technol. 70 (878), 49-57 (1998). 9. M. de Meijer and H. Militz, Holzforschung 53, 553-560 (1999). 10. P. Ahola, Holz als Roh- und Werkstoff 49, 428-432 (1991). 11. H. Derbyshire and E.R. Miller, J. Inst. Wood Sci. 14 (4), 162-168 (1997). 12. C.J. van Oss, M.K. Chaudhury and R.J. Good, Chem. Rev. 88, 927-941 (1988). 13. E. Kientz, J.Y. Charmeau, Y. Holl and G. Nanse, J. Adhesion Sci. Technol. 10, 745-759 (1996). 14. J.Y. Charmeau, E. Kientz and Y. Holl, Progr. Organic Coatings 27, 87-93 (1996). 15. B.M. Collett, Wood Sci. Technol. 6, 1-42 (1972). 16. A. Pizzi, Holzforschung Holzverwertung 44, 62-97 (1992). 17. M. de Meijer and H. Militz, Progr. Organic Coatings 38 (3-4), 223-240 (2000). 18. K. Richter, W.C. Feist and M.T. Knaebe, Forest Prod. J. 45 (7/8), 91-97 (1995). 19. W.A. Coté, J. Coatings Technol. 55 (699), 25-35 (1983). 20. F.F.P. Kollmann and W.A. Coté, Principles of Wood Science and Technology Vol. I, p. 162, Springer, Berlin (1984). 21. J.F. Siau, Transport Processes in Wood, p. 142, Springer, Berlin (1984). 22. E. Liptáková and J. Kúdela, Holzforschung 48, 139-144 (1994). 23. M. Scheikl and M. Dunky, Holz als Roh- und Werkstoff 54, 113-117 (1996). 24. M. Scheikl and M. Dunky, Holzforschung 52, 89-94 (1998). 25. M. de Meijer, B. van de Velde and H. Militz, J. Coatings Technol. 73 (914), 39-50 (2001). 26. E.W. Washburn, Phys. Rev. 17, 273-283 (1921). 27. V.R. Gray, Forest Prod. J. 12 (9), 452-461 (1962). 28. T. Nguyen and W.E. Johns, Wood Sci. Technol. 12, 63-74 (1978). 29. D.J. Gardner, Wood Fiber Sci. 28 (4), 422-428 (1996). 30. Q. Shen, J. Nylund and J.B. Rosenholm, Holzforschung 52, 521-529 (1998). 31. R.J. Good, in: Contact Angle, Wettability and Adhesion, K.L. Mittal (Ed.), pp. 3-36, VSP, Utrecht (1993). 32. D.Y. Perera, Progr. Organic Coatings 28, 21-23 (1996). 33. S.G. Croll, J. Coatings Technol. 52 (665), 35-43 (1980). 34. P. Ahola, J. Oil Colour Chem. Assoc. 5, 173-176 (1991). 35. P.C. Hopma, J. Oil Colour Chem. Assoc. 7, 179-184 (1984). 36. J. Sickfeld, J. Oil Colour Chem. Assoc. 8, 292-298 (1978). 37. J.L. Prosser, Adhesion of Coatings: Theory and Practice, Paint Res. Assoc. (PRA), Teddington, UK (1993). 38. S.G. Croll, J. Coatings Technol. 51 (648), 64-68 (1979). 1. 2. 3. 4.

Analysis of the wet adhesion of coatings on wood

201

39. D.Y. Perera, in: Paint and Coating Testing Manual, J.V. Koleske (Ed.), pp. 585-599, ASTM Manual Series 17 (1995). 40. M.A. Kalnins, C. Katzenberger, S.A. Schmieding and J.K. Brooks, J. Colloid Interface Sci., 125, 344-346 (1988). 41. C.K. Owens and K.C. Wendt, J. Appl. Polym. Sci. 12, 1741-1747 (1969). 42. C.J. van Oss, Colloids Surfaces A 78, 1-49 (1993). 43. M. de Meijer, K. Thurich and H. Militz, Holz als Roh- und Werkstoff 59, 35-45 (2001). 44. M. de Meijer, S. Haemers, W. Cobben and H. Militz, Langmuir 16, 9352-9359 (2000). 45. M. Wålinder, Wetting Phenomena on Wood, Doctoral thesis, Royal Inst. Technol., Stockholm (2000). 46. K.L. Mittal, in: Adhesion Measurement of Thin Films, Thick Films and Bulk Coatings, K.L. Mittal (Ed.), STP No. 640, pp. 5-17, ASTM, Philadelphia (1978). 47. K.L. Mittal, Electrocomponent Sci. Technol. 3, 21-42 (1976). 48. K.L. Mittal, in: Adhesion Measurement of Films and Coatings, K.L. Mittal (Ed.), pp. 1-13, VSP, Utrecht (1995).

This page intentionally left blank

Adhesion Aspects of Polymeric Coatings, Vol. 2, pp. 203–213 Ed. K.L. Mittal © VSP 2003

Modified tape test: Measurement of adhesion of insulator films to low dielectric constant organic polymers LUONA L.N. GOH,1 S.L. TOH,1,∗ SIMON Y.M. CHOOI,2 YI XU2 and T.E. TAY1 1

Department of Mechanical Engineering, National University of Singapore, 10 Kent Ridge Crescent, Singapore 119260, Singapore 2 Department of Technology Development, Chartered Semiconductor Manufacturing, 60 Woodlands Industrial Park D Street 2, Singapore 738406, Singapore

Abstract—In the microelectronics industry, a simple tape test (ASTM 3359) is typically performed to qualitatively study the adhesion of dielectric films. In this work, a novel approach to conduct the tape test is proposed. This new method is able to eliminate the inconsistency in the traditional tape test, and at the same time, it provides both qualitative and quantitative results. This approach employs a spring-loaded mechanism and eliminates the subjective interpretation of the results of the traditional tape test. It also provides quantitative measures of the peel speed and force, factors that are relevant in the characterization of the interfacial strength of thin films. We illustrate the use of this new modified tape test (MTT) in a study of the adhesion of multi-layered thin film structures. The structures consist of low-k organic polymer with silicon nitride and silicon oxide films. Since the latter two films are often used as hard masks to pattern the low-k dielectric layer, their adhesion to the low-k material is thus very important. Keywords: Tape test; peel force; peel rate; adhesion; low-k material.

1. INTRODUCTION

The continuing trend toward lower Resistance-Capacitance (RC) time constants in advanced integrated circuits has led to intensive studies and development of low dielectric constant (low-k) materials [1, 2]. However, the integration of low-k dielectric materials into a circuit is challenging. One vital aspect of integration relates to the adhesion of the low-k dielectrics to the films that lie above and beneath it. The tape test [3, 4] is a commonly used method by the microelectronics industry to determine the adhesion of thin films. The operator simply needs to apply the tape across a freshly crosshatched surface and then to peel the tape quickly ∗

To whom all correspondence should be addressed. Phone: (65) 874 2578, Fax: (65) 779 1459, E-mail: [email protected]

204

L.L.N. Goh et al.

at an angle of approximately 180 degrees. Tape test is an inexpensive and fast method, without a need for tedious sample preparation. However, the tape test provides only qualitative results and the accuracy of the tape test depends highly on the skill of the operator. When different operators perform the tape test, the peel angle and peel rate may vary significantly. As a result, the amount of film peeled off will be different. This is a major disadvantage of the tape test. However, despite its drawbacks, the tape test is still commonly used because of its simplicity. In this paper, we present an improvised and more consistent method for performing the tape test, using a spring-loaded mechanism, thus reducing the dependency of the test results on the operator. The Modified Tape Test (MTT) also provides force-time graph and average peel rate for analysis and quantification of practical adhesion strength [5, 6]. 2. EXPERIMENTAL SETUP

Figure 1 shows a schematic diagram of the Modified Tape Test (MTT) setup. The setup includes a 50 N load cell for measuring the peel force, a laser beam emission system for measuring the peel rate, a sample holder for securing the sample, a spring-loaded hollow cylinder and an adhesive tape. 2.1. Operation of the modified tape test (MTT) In our study, the MTT was performed under an ambient temperature of 21°C with a percentage relative humidity of 70%. The application of the tape onto the sample followed the recommendation in ASTM 3359. One of the limitations of tape test is that the test results are dependent on the type of tape selected. However, since the particular type of tape mentioned in the ASTM 3359 is no longer commercially produced by 3M, the replacement tape recommended by the company is 3M Scotch tape #600 [7]. Hence, a roll of this commercial tape was used in this work, and the variation in adhesive quality within the roll was assumed to be negligible. The sample size for each test was five. A long tape was carefully removed from the roll and cut into 200 mm strips. One end of the tape was applied onto the sample surface, over a distance of 70 mm (as recommended in ASTM 3359). To ensure good contact with the film, a wooden applicator with a rubber tip was used to secure the tape firmly onto the substrate. The sample was then placed inside a holder (Figure 1), which is attached to a 50 N load cell. The other end of the tape was then firmly attached to a knob, which, in turn, is attached to a spring housed in a metallic cylinder. The spring, which was made of music wire with a spring constant of 0.48 N/mm, is loaded inside the hollow cylinder and held in position by engaging the knob onto a side groove in the cylinder wall. A longitudinal slot was machined in the wall of the cylinder to allow movement of the knob when the spring was released. The spring was released by disengaging the knob from the side groove. Under the

Modified tape test: Measurement of adhesion of insulator films

205

force of the compressed spring, the knob pulls the tape away from the sample at a high speed. The time counter captures the time taken for the knob to intercept the path of the laser beam sensors. At the same time, the load cell detects the peel force exerted. 2.2. Types of samples In the first part of the experiment, the MTT was used to assess the adhesion of a capping layer namely, undoped silicate glass (USG) oxide, on a low-k dielectric film. In sample A, the USG-oxide was deposited onto dielectric A, which is a proprietary organic low-k dielectric material. In sample B, the USG-oxide was deposited onto dielectric B which is a commercially available fluorinated

Figure 1. Experimental Set-up of the Modified Tape Test (MTT).

206

L.L.N. Goh et al.

poly(arylether) known as FLARE [8] manufactured by Honeywell. Both dielectric films were 800 nm thick and were spun onto the wafer using a spin-coater. The USG-oxide layers were 400 nm thick and they were deposited onto the dielectric film using Applied Materials P5000 Chemical Vapor Deposition (CVD) system. In the second part of the experiment, the MTT was used to assess the adhesion of different types of capping layers on dielectric A. Three types of capping layers were silicon nitride (SiN), undoped silicate glass (USG)-based oxide and tetraethylorthosilicate (TEOS) oxide. All the capping layers were 400 nm thick and were deposited by CVD. 3. RESULTS AND DISCUSSION

From the results of the Modified Tape Test (MTT), it was found that there were three different types of waveform force-time characteristics during the peeling process. 3.1. Characteristics of the MTT waveforms The first type of waveform can be seen in Figure 2. The figure shows that the MTT load cell experiences a large re-bound force after the tape has left the specimen. The re-bound force is due to the sudden delamination of the capping layer (e.g. USG-oxide film) from the dielectric film. Initially when the MTT was performed, the adhesive tape was observed to fail interfacially from the surface of the capping layer. However, after the tape had traveled 10 to 20 mm away from the initial starting position, the capping layer began to delaminate from the dielectric film. With the delamination of the capping layer, the spring force transmitted to the load cell starts to drop. The MTT peak peel force for this sample as ob-

Figure 2. Waveform with re-bound force (Type 1).

Modified tape test: Measurement of adhesion of insulator films

207

served in Figure 2 is 16 N with an average peel rate of 1.7 m/s. However, this value does not represent the adhesion force between the capping layer and the dielectric film. It is the adhesion force between the adhesive of the tape and the film surface. It was observed during the tests that there was always a small distance X at the initial part of the peeling process where the capping layer remained intact or attached to the low-k dielectric material. As the peeling proceeds, the capping layer becomes detached or delaminated for distances greater than X. This phenomenon introduces variations in the average peel rate as measured by the laser sensors (Figure 3). In order to reduce these variations, an incision into the test film was deliberately performed across the starting position of the adhesive tape. This has been found to be a very effective way to ensure consistent peeling of the capping layer beginning at the position of the incision. The resulting Force-Time (Type 2) plot is shown in Figure 4. As seen in Figure 4, the delamination of the capping layer takes place right from the start of the test. Hence, the initial unpeeled distance X, as in the case of Type 1 waveform, was not revealed in Type 2 waveform. The force builds up to a peak peel force of 7.1 N and then drops rapidly. Comparing the peak peel force obtained from Type 1 and Type 2 waveforms, the peak peel force for Type 2 waveform is significantly lower than that for Type 1. The MTT shows that the capping layer now peels at a lower force but at a higher average peel rate of 5.1 m/s. By performing an incision into the test film, both the MTT peak peel force and the average peel rate are found to be more consistent and repeatable. Figure 5 shows the third type of waveform, which reflects the interfacial failure of the tape from the surface of the capping layer. No delamination was observed

Figure 3. Correlation of Peel Rate Versus Initial Distance, X.

208

L.L.N. Goh et al.

in the sample. The force gradually decreases at a rather slow and steady rate. This further shows that a large resistance is encountered by the spring in separating the tape from the capping layer. The peak peel force of 15 N is similar to that for Type 1 and is higher than that for Type 2.

Figure 4. Waveform without re-bound force (Type 2).

Figure 5. Waveform of a sample without delamination (Type 3).

Modified tape test: Measurement of adhesion of insulator films

209

3.2. MTT peak peel force for different film structures The MTT test was performed on two types of samples, namely Sample A (USGoxide/Dielectric A/Si) and Sample B (USG-oxide/Dielectric B/Si). The results are shown in Table 1 and Figure 6. Five specimens each of Sample A and Sample B were tested. In all cases reported in this paper, the same 3M Scotch tape #600, was used in the tests. The experiment shows that Sample A has a MTT peak peel force of 10.87 N with an average peel rate of 4.98 m/s while Sample B has a higher peak peel force of 16.03 N with an average peel rate of 3.21 m/s. Visual inspection showed that the capping layers of all five specimens of Sample A had delaminated while the capping layers of all five specimens of Sample B still remained intact to the dielectric film. All five specimens of Sample A therefore exhibit Type 2 force-time characteristics. The waveform of Sample B is identical to the Type 3 waveform as seen in Figure 5, which depicts the interfacial failure of the tape from the sample. This explains the reason why Sample B has a higher peak peel force than Sample A. An SEM picture of the delaminated Sample A is shown in Figure 7. Table 1. MTT Results for the Two Different Film Structures

Average peak peel force (N) Std Dev (N) Peel Rate (m/s) Std Dev (m/s)

Sample A

Sample B

10.87 3.75 4.98 0.02

16.0 2.09 3.21 0.07

Figure 6. MTT Peak Peel Force for Two Different Film Structures.

210

L.L.N. Goh et al.

Although the SEM picture showed an interfacial failure between the capping layer and the dielectric film, Attenuated Total Reflectance Fourier Transform Infrared Spectroscopy (ATR-FTIR) was also used to identify the exact locus of failure. The ATR-FTIR spectroscopy identifies the molecular chain orientation on the delaminated surface and compares it to the dielectric and the USG-oxide spectra (see Figure 8). The ATR-FTIR spectra as shown in Figure 8 reveal that the delaminated sample has a spectrum that more closely resembles the dielectric film spectrum. This

Figure 7. Interfacial failure of Sample A.

Figure 8. ATR-FTIR Spectra for Dielectric A (top), Delaminated Sample (middle) and USG-oxide (bottom).

Modified tape test: Measurement of adhesion of insulator films

211

indicates that the locus of failure [9] for Sample A is either within the dielectric film or at the interface between the capping layer and the dielectric film. 3.3. Effect of different capping layers on dielectric A Different types of capping layers [10], namely silicon nitride (SiN), undoped silicate glass (USG)-oxide and tetraethylorthosilicate (TEOS)-oxide, were deposited onto dielectric A to assess its interface using MTT. All the capping layers were 400 nm thick while the dielectric A was 800 nm thick. The sample size was five for each type. The results are given in Table 2. Table 2. MTT Peak Peel force and Peel Rate for Different Capping Layers on Dielectric A Sample

Ave. Peak Peel Force (N)

Ave. Peel Rate (m/s)

Locus of failure

SiN/Dielectric A/Si

10.57 (σ = 3.06)

5.17 (σ = 0.03)

Mainly either in the dielectric layer or at the interface

USG-oxide/Dielectric A/Si

10.87 (σ = 3.75) 6.20 (σ = 2.07)

4.98 (σ = 0.02) 5.16 (σ = 0.04)

Mainly either in the dielectric layer or at the interface

TEOS-oxide/Dielectric A/Si

Mainly in the oxide layer

Note: σ denotes the standard deviation

Based on visual inspection, all three types of samples were found to have complete interface delamination. Sample with SiN as the capping layer had a MTT peak peel force of 10.57 N with a peel rate of 5.17 m/s, sample with USG-oxide as capping layer had a slightly higher peak peel force of 10.87 N with a peel rate of 4.98 N and sample with TEOS-oxide as the capping layer had the lowest peak peel force of 6.2 N with a peel rate of 5.16 m/s (see Figure 9). Although TEOS-oxide and USG-oxide were both silicon oxide, the differences in MTT peel force could be due to the different methods of deposition, which gave different adhesion properties. ATR-FTIR measurements on the TEOSoxide/Dielectric A/Si specimen indicated that failure occurred in this case mainly in the oxide layer (Figure 10). This is in contrast to the SiN/Dielectric A/Si and USG-oxide/Dielectric A/Si cases where failure occurred mainly either at the interface or within the dielectric layer. 4. CONCLUSIONS

Here we have presented a new method called the modified tape test (MTT) to measure the adhesion of insulator films to low dielectric organic polymers. This

212

L.L.N. Goh et al.

method is an improvement over the original tape test, where its peel angle and its peel rate are highly dependent on the skill of the operator. In MTT, all these important parameters are fixed and secured by the MTT fixture, reducing the uncertainties introduced by manual operation. Furthermore the MTT can accommodate a force transducer, which provides the force-time characteristic plot of the de-

Figure 9. Average Peak Peel Force (N) for Different Capping Layers.

Figure 10. ATR-FTIR Spectra for Delaminated Sample (top), TEOS-oxide (middle) and Dielectric A (bottom).

Modified tape test: Measurement of adhesion of insulator films

213

lamination process. It is shown that there are differences in the force-time curves for samples that do not delaminate and samples that delaminate. From the experiments, it was shown that adhesion of insulator films to dielectric A (proprietary product) was poorer than to dielectric B (FLARE). This is reflected from the MTT peak peel force as 10.87 N and 16.03 N, respectively. Different types of materials deposited on dielectric A give different MTT peak peel forces. From the experiments, it was found that both SiN/Dielectric A/Si and USG-oxide/Dielectric A/Si systems had a higher peak peel force of 10.57 N and 10.87 N respectively, than the TEOS-oxide/Dielectric A/Si system, which yielded a peak peel force of 6.2 N. All the samples showed the locus of failure mainly either in the dielectric layer or at the interface, except the TEOS/Dielectric A/Si system, which had locus of failure mainly within the oxide film. Acknowledgement The authors thank Dr. Jian-Hui Ye of IMRE for providing assistance in the ATRFTIR measurements, and Dr. Mei-Sheng Zhou for technical discussions. REFERENCES 1. W.W. Lee and P.S. Ho, Mater. Res. Soc. Bull., 22 (10), 19 (Oct. 1997). 2. N.H. Hendricks, Mater. Res. Soc. Symp. Proc. 443, 3-14 (1996). 3. S. Lee, L.C. Jensen, S.C. Langford and J.T. Dickinson, J. Adhesion Sci. Technol., 9, 1-26 (1995). 4. L. Scudiero and J.T. Dickinson, J. Adhesion Sci. Technol., 9, 27-45 (1995). 5. K.L. Mittal, in: Adhesion Measurement of Thin Films, Thick Films and Bulk Coatings, K.L Mittal (Ed.), STP 640, pp. 5-17, ASTM, Philadelphia (1978). 6. K.L. Mittal, in: Adhesion Measurement of Films and Coatings, K.L. Mittal (Ed.), pp. 1-13, VSP, Utrecht (1995). 7. S.C. Sun, Y.C. Chiang, C.T. Rosenmayer, J. Teguh and H. Wu, Mater. Res. Soc. Symp. Proc. 443, 85-90 (1996). 8. M. Du, R.L. Opila and C. Case, J. Vacuum Sci. Technol. A, 16, 155-162 (1998). 9. R.J. Good, in: Adhesion Measurement of Thin Films, Thick Films and Bulk Coatings, K.L. Mittal (Ed.), STP 640, pp. 18-29, ASTM, Philadelphia (1978). 10. M. Morgen, E.T. Ryan, J.-H. Zhao, C. Hu, T. Cho and P.S. Ho, J. Mater., 37-40 (Sept. 1999).