ANALYTICAL PROCEEDINGS, JANUARY 1993, VOL 30 27 Surface Analysis and Adhesive Bonding J. Comyn Centre for Surface and Materials Analysis, Armstrong House, Oxford Road, Manchester M 7 7ED For work on adhesive bonding there are four experimental techniques that are used on a routine basis and which provide much useful information. The purpose of this paper is to review briefly these techniques. There are other techniques that are not yet routine, but which have been applied to adhesive bonding; they include inelastic electron tunnelling spectroscopy (IETS)1-3 and surface enhanced Raman spectro- scopy (SERS) .4 Contact Angle Measurement A useful and simple way of assessing solid surfaces is to measure the contact angle made by small liquid drops. At the centre of the method is an equation due to Fowkess which is Here 0 is the contact angle and yL is the liquid surface tension.The dispersive and polar components of the surface energy of the liquid are yLd and yLP, and ysd and ysP are those of the solid surface. If contact angles are measured for a number of liquids and yL (1 + c0s6)/2(y~d)~/2 is plotted against ( ~ ~ P / y ~ d ) l / ~ , the graph should be linear with intercept (ysd)l/2 and slope (ysP)l/*. A list of available test liquids, basically arranged in decreasing order of the x-parameter of the plot, is shown in Table 1. For a liquid the surface tension and surface free energy are numerically the same, but have units of mN m-1 and mJ m-2, respectively. This information is of value in the selection of adhesives in that the optimum condition is to match the dispersive and polar components of the adhesive and substrate.Surfaces that are difficult to bond such as the polyalkenes have low surface energies which are almost entirely dispersive in character. Surface treatments which render such materials bondable cause increases in the polar component of the surface energy. The thermodynamic work of adhesion6 can be calculated from the polar and dispersive components of the two phases. More important, however, is that the thermodynamic work of adhesion in the presence of water can also be calculated; this will indicate whether a bond is stable or not in wet conditions. Water is a polar liquid which has a tendency to displace adhesives and paints from metal surfaces. Water displacement is not generally a problem for paints or adhesives on plastics because the surfaces are much less polar than metals.However, if one surface has polar groups such as carboxylates, then in the presence of water there may be a significant reduction in the work of adhesion and hence in the stability of the bond. Surface Infrared Spectroscopy In multiple internal reflectance (MIR) the sample is placed in contact with a prism of either germanium or thallium bromide iodide, which can be placed in either a dispersive or Fourier transform spectrometer. The spectra obtained closely resemble conventional transmission infrared (IR) spectra. The sampling depth is about the wavelength of the radiation used. As this is many thousands of molecular layers, it is thus much less surface sensitive than X-ray photoelectron spectro- scopy and static secondary ion mass spectrometry (SSIMS).It is important that the sample makes good optical contact with the prism, and for this reason it is easier to work with rubbery polymers. Multiple internal reflectance has been applied to the oxidative surface treatment of polyalkenes, where it shows the formation of carbonyl groups. However, because of the relatively large sampling depth, the samples were probably treated to a greater depth than is actually required for the purposes of adhesive bonding or printing. Diffuse reflectance in Fourier transform (DRIFT) is parti- cularly useful in that powdered samples can be examined without further preparation; the method has been used extensively by Koenig to study silane coupling agents on powdered silica.With reflection-absorption infrared spec- trometry (RAIR) a beam that is polarized parallel to the plane of incidence is reflected from a polished metal mirror at a high angle of incidence (about SOo). It has been employed by Boerio in studying silane coupling agents on a number of metals. The application of these two techniques to silane coupling agents has been reviewed by the author? Great advantages are conferred on the Fourier transform infrared (FTIR) spectrometer by the incorporation of a computer. These include multiple scanning to improve signal- to-noise ratio, and the ability to produce difference spectra. The latter has been used by Naviroj et a1.8 to study the adsorption of 3-aminopropyldimethylethoxysilane on titan- ium and aluminium oxides.The difference spectra (metal oxide treated with silane-metal oxide) revealed weak bands at 950 and 963 cm-1, which are respectively assigned to Si-0-Ti and Si-0-A1 groups. X-ray Photoelectron Spectroscopy (XPS) In this technique a monochromatic beam of soft X-rays from an A1 or Mg target is directed at a surface in an ultra-high vacuum chamber. Electrons are ejected from the core levels of atoms in the surface, and their binding energies which are characteristic, are determined. An XPS spectrum can have a number of peaks arising from electrons emitted from different core levels, and when the relative atomic sensitivities are taken into account, this can give an atomic analysis of the surface. Hydrogen is not detected by XPS because it has no core electrons.Although the X-rays penetrate deep into the sample, only electrons from near the surface are able to escape and be detected; the sampling depth is about 5-10 nm. This technique was used by Brewis and Briggs9 to demon- strate which chemical groups are introduced on to polyalkene28 ANALYTICAL PROCEEDINGS, JANUARY 1993, VOL 30 Table 1 Liquids for contact angle measurements Liquid Water Glycerol Ethane-1,2-diol Formamide Ethanol Dimethyl sulfoxide 2-Ethoxyethanol Dimethylformamide Dime thylsiloxanes Methylene diiodide Tricresyl phosphate Trichlorobiphen yl Pyridine Hexadecane 1-Bromonaphthalene yLd/mJ m-2 21.8 f 0.7 37.0 k 4 29.3 39.5 _+ 7 17.0 34.86 23.6 32.42 16.9 k 0.5 48.5 k 9 50.76 39.2 f 4 44 k 6 37.16 27.6 47 _+ 7 yLP/mJ m- 51.0 26.4 19.0 18.7 5.4 8.68 5.0 4.88 2.1 2.3 1.7 1.3 0.84 -2.4 -2 IL/mJm-2 72.8 63.4 48.3 58.2 22.4 43.54 28.6 37.30 19.0 50.8 50.76 40.9 45.3 38.00 27.6 44.6 (YL~/Y")"' 1.529 2 0.035 0.845 -t 0.11 0.805 f 0.14 0.688 2 0.19 0.563 0.499 0.460 0.388 0.352 0.218 k 0.446 0.0 0.208 0.172 0.150 0.0 0.2261 _+ 0.14 surfaces by flame and corona treatment.It is particularly useful at detecting contamination involving a heteroatom, such as might be the case with silicone and fluorocarbon mould release agents. Static Secondary Ion Mass Spectrometry This technique places the sample in an ultra-high vacuum chamber and bombards it with ions ( A r f , Xe+ or G a t s 4 keV). The surface then emits secondary ions which are analysed in a mass spectrometer, and in separate experiments both the positive and negative ion spectra can be obtained.In static, as opposed to dynamic, SIMS the primary ion current is so low (=1 nA cm-2) that the surface is not significantly ablated. Data are obtained from the top few molecular layers. A major problem with insulators is the build up of surface charge. This can be avoided by flooding the surface with low energy ( ~ 7 0 0 eV) electrons. The SSIMS spectra are complicated and it is difficult to assign all the mass peaks. However, there have been a number of instances where it has been very successful in identifying surface contamination. One is the detection of ethylene bis(stearamide) on the surface of a polyurethane,lO from the presence of peaks at 282 and 310 u, which are due to the following cleavage reaction: C 17H35-CO-NH-CH2-CH2-NH-CO-C 17H35 + C17H35-CO-NH+ + C17H35-CO-NH-CH2-CH2f 282 u 310 u Silicone contamination is a common problem in adhesive bonding, and polysiloxane contaminants can be readily identified by their fragmentation patterns.1 2 3 4 5 6 7 8 9 10 References Comyn, J., Oxley, D. P., Pritchard, R. G., Werrett, C. R., and Kinloch, A. J., Int. J. Adhes. Adhes., 1989 9, 201. Comyn, J . , Horley, C. C., Pritchard, R. G., and Mallik, R. R., Adhesion, 1987, 11, 38. Comyn, J., Horley, C. C., Oxley, D. P., Pritchard, R. G., Werrett, C. R., and Tegg, J. L., Br. Polym. J . , 1983, 15, 50. Creighton, J. A., in Advances in Spectroscopy, eds. Clark, R. J. H., and Hester, R. E., Wiley. Chichester, 1988, vol. 16, Fowkes, F. M., Ind. Eng. Chem., 1964, 56,40.Comyn, J., in Durability of Structural Adhesives, ed. Kinloch, A. J., Applied Science Publishers, Barking, 1983, ch. 3, pp. Comyn, J.. in Structural Adhesives: Developments in Resins and Primers, ed. Kinloch, A. J., Elsevier, Barking, 1986, ch. 8, pp. 269-3 12. Naviroj, S . , Koenig, J. L., and Ishida, H.. J. Adhesion, 1985,18, 93. Brewis, D. M., and Briggs, D., Polymer, 1981, 22, 7. Briggs, D., Surface Interface Anal., 1986, 9, 391. ch. 2, pp. 37-89. 85-131. Surface Enhanced Raman Spectroscopy J. Alan Creighton Chemical Laboratory, University of Kent, Canterbury CT2 7NH Surface-enhanced Raman spectroscopy (SERS) is becoming recognized as a powerful method for obtaining detailed information on the first monolayer of molecules at certain surfaces. It combines the high structural information content of vibrational spectroscopy with the very high sensitivity and surface specificity which come from the special enhancement mechanism of SERS.Moreover, it is a non-UHV (ultra-high voltage) technique. It is thus very well suited to in situ analyses of surfaces in a wide range of ambient media, and is beginning to find use in fields as diverse as heterogeneous catalysis, electrochemistry and corrosion, adhesion, surfactants, or the interaction of bio-molecules with metal surfaces. Unfortu- nately, it has the limitation that it is only applicable to certain surfaces, namely those of free-electron metals, particularly copper, silver and gold, although some progress has recently been made towards partially overcoming this limitation, as is described below.There have been several reviews describing applications of SERS,1-3 and a number of recent reviews give detailed accounts of its mechanism.4 A more general account of SERS is included among the papers presented at the previous SATA symposium.-5 In its experimental set-up SERS does not differ greatly from normal Raman spectroscopy. The surface is irradiated with monochromatic light from a laser and the scattered light is collected and analysed. The inelastically scattered component in this scattered light gives the vibra- tional spectrum of the molecules adsorbed at the surface. For most surfaces the intensity of Raman bands from the surface monolayer is very low, and indeed, is normally undetectable. If the surface is a suitably roughened or finely divided surfaceANALYTICAL PROCEEDINGS, JANUARY 1993, VOL 30 29 of copper, silver or gold, however, the SERS effect then amplifies the first monolayer Raman scattering by a factor of up to 105, giving easily detectable Raman signals specifically from the first monolayer.The objective of this paper is to highlight some recent applications of SERS which relate to problems of applied interest. Before doing this, however, it is helpful to examine, briefly, the mechanism of the enhancement in order to understand some of the special features of SERS, such as its surface specificity and its restriction only to certain surfaces. There appear to be two enhancement effects, each of which contributes 2-3 orders of magnitude to the over-all Raman intensity enhancement. One of these, often called the electromagnetic SERS effect, depends on the response of the conduction electrons of the metal surface to the electromag- netic field of the incident and scattered light.It is necessary for SERS that the metal surface is very finely divided, and the most widely used surfaces have thus been metal electrodes that have been roughened by electrochemical oxidation- reduction cycling, evaporated metal films deposited at very low temperatures in high vacuum, or colloidal metal disper- sions. The essential feature of this fine state of division is that the conduction electrons of the metal then have localized ‘plasma’ oscillations at optical frequencies, and these oscilla- tions radiate a secondary electromagnetic field close to the surface which augments the incident and scattered electro- magnetic fields.In the electromagnetic SERS effect this augmenting of the surface electromagnetic field can become fairly large because the frequency of the incident light is chosen so that there is a resonant excitation of the plasma oscillations at the surface. Because the large surface electro- magnetic fields depend on this resonant excitation it follows that there is a laser frequency- and surface morphology- dependence in the magnitude of the electromagnetic SERS effect, and that only metals which give undamped plasma resonances in the visible or near-infrared range, namely the essentially free-electron metals, can show the effect. The other enhancement mechanism, often referred to as the chemical SERS effect, arises if the incident radiation is in resonance with a charge transfer transition between the surface and the adsorbed molecules.The effect of such a resonance is to enhance the polarizability of the first layer of molecules at the surface at that frequency, and thus to give rise to enhanced light scattering by the well established resonance Raman mechanism. It is this ‘chemical’ effect, rather than the electromagnetic contribution, which gives the SERS effect its high degree of specificity to the first monolayer. The requirement that there is a surface-adsorbate charge transfer transition in resonance with the incident frequency restricts the adsorbates to some extent, and also the surface sites, which exhibit the chemical SERS effect at a given excitation wavelength, although this restriction is not as limiting as might at first be expected (presumably because of the breadth of the metal bands which are the donor or acceptor levels in such surface-adsorbate charge transfer transitions).Thus it is found that molecules such as alkanes which do not have suitable orbitals to participate in surface-adsorbate charge transfer transitions give very much weaker SERS spectra than alkenes or aromatics with filled low-energy n and empty n* orbitals. It has also become increasingly apparent that there may be sites of widely differing SERS activity on a given surface, and there is currently considerably interest in elucidating the nature of these SERS-active sites (see below).4 A straightforward application of SERS which demonstrates its unique capability for in situ analysis is in competitive adsorption studies at surfaces under normal ambient condi- tions.An example of such a study is a recent measurement by Musiani et a1.6 of the competitive adsorption of corrosion inhibitors at copper surfaces under aqueous conditions. An interesting feature of this work is that it was carried out using a Fourier transform (FT)-Raman spectrometer with near- infrared excitation at a wavelength of 1.064 ym. Most SERS measurements to date have been made with visible-range excitation. It has been known for some time, however, that with copper and gold surfaces it is necessary to use excitation wavelengths no shorter than about 600 nm, and that with these metals the SERS intensity continues to increase into the far red.Recent measurements by Chase and Parkinson7 have shown that the maximum SERS enhancement for copper and gold surfaces, at least for the electrochemically roughened surfaces which they investigated, is obtained at excitation wavelengths even longer than 1.064 pm in the near-infrared, and the excellent quality of SERS spectra which may be obtained for copper surfaces with an FT-Raman spectrometer and 1.064 pm excitation is illustrated by recent work by Sockalingum et al.8 The high degree of surface specificity of SERS has recently been exploited to provide a new method for measuring thin-film diffusion. The thin film is deposited on an SERS- active silver or gold surface and the film is exposed to the diffusate, the spectrum for which appears when the diffusate reaches the SERS-active surface.This method has been applied by Hong et al.9 to the measurement of the self- diffusion of polystyrene at 170°C (the diffusion of per- deuteriopolystyrene through polystyrene), and by Blue et al.,10 who developed a more complete analysis of the experimental data to obtain accurate values for the diffusion coefficient of ethylene through pyrazine over a range of very low temperatures. An earlier use of the surface specificity of SERS to give an analytical measurement relating only to the metal interfacial side of a surface film is that of Hutchinson et al.,” who compared the extent of an electrochemical reaction (the interconversion of thionine and leucothionine) at the metal-film interface measured by SERS with the extent of the reaction in the film as a whole, measured by coulometry.From the hysteresis between the two measure- ments as the electrode potential was cycled, it was deduced that the electrochemical oxidation or reduction began at the solution side rather than at the metal side of the film, and is thus essentially a solution- rather than a metal-interfacial phenomenon. An application of SERS to perhaps a more conventional thin film analytical problem is that of Knight et al. ,12 in which diamond films deposited on silicon by cold vapour deposition (CVD) were characterized by Raman spectroscopy. For very thin films the detection of the characteristic Raman band of diamond was greatly improved by sputtering on to the films a silver overlayer of thickness about 5 nm.This is one of several papers which illustrates a recent important development in SERS, namely the use of the enhanced electromagnetic fields which extend out by 10 A or more from an SERS-active silver or gold surface by virtue of the electromagnetic SERS effect, to obtain enhanced Raman scattering from the adjacent non-SERS-active medium. This method of depositing a particulate silver overlayer has been used, for example, by Alsmeyer and McCreeryl3 to sharpen up the sampling depth in the Raman measurement of the surface defect density induced in graphite electrodes by laser irradiation or electro- chemical anodization. The electron transfer rate between these electrodes and solution species is found to correlate with this surface defect density, but only the first few atomic layers control the electrochemical behaviour.This sharpening of the sampling depth comes about because of the short range of the enhanced electromagnetic fields induced by the silver over- layer, and the correlation with the graphite electrode electron transfer rate was thus better in the presence of the silver overlayer (sampling depth about 20 A) than with the normal Raman sampling depth for graphite of about 135 A. There have also been SERS studies of indium phosphide14 and tin oxide15 surfaces where the SERS is similarly induced by the deposition of a discontinuous thin silver overlayer. An important group of such studies is where the non-SERS-30 ANALYTICAL PROCEEDINGS, JANUARY 1993, VOL 30 active surface is that of another metal.For these surfaces the use of a roughened silver or gold substrate, with the non-SERS-active metal deposited as a thin continuous film, has been found to be the most effective sample arrangement. Several studies of this type have been directed to iron films in an aqueous electrolyte.16J7 In one such study by Mengoli et al. , I 6 approximately 30 atomic layers of iron were deposited electrochemically on to a pre-roughened silver electrode. On exposure to the adsorbate pyridine, this electrode showed an SERS spectrum attributed to pyridine adsorbed on iron, which was clearly different to that of pyridine adsorbed on the underlying silver, and there was also a band attributed to chloride ions adsorbed on iron. Another group of recent papers concerns SERS studies by Weaver et al.18-20 of various adsorbates and electrochemical reactions at the surface of thin continuous films of platinum-group metals deposited on roughened gold substrates. An example is an investigation by SERS of the electrochemical behaviour of CO on platinum. 18 Approximately two atomic layers of platinum were deposited electrochemically on the gold substrate, and the completeness of the film was checked by cyclic voltammetry, which showed predominantly oxidation or reduction waves at potentials corresponding to the formation or removal of platinum oxide. On exposure of the electrode to CO, the SERS spectrum showed well-resolved bands due to CO adsorbed at sites both on platinum and on gold, and the loss of these bands due to electrochemical oxidation at positive potentials was also observed.A parallel study of CO adsorption on a palladium thin-film electrode gave similar results, but showed the greater tendency for adsorption of CO on palladium which has long been known from infrared studies on palladium in UHV. A very interesting feature of this work is the observation that the stretching frequency of CO adsorbed on the platinum electrode is about 10 cm-1 lower in the SERS spectrum than in the surface infrared spectrum, and the potential at which the SERS band disappears due to electro-oxidation of adsorbed CO is about 0.3 V more positive than the potential at which the surface infrared band disappears or at which electro- oxidation is observed by cyclic voltammetry. This suggests that SERS is detecting a sub-set of the totality of adsorbed CO molecules and that these SERS-active sites are particularly tightly binding (and therefore particularly reactive).There are other pieces of evidence which lead to similar conclusions. Among these are some recent results of Gu et a1.21.22 relating to the epoxidation of ethylene in which data from SERS and electron energy loss spectroscopy (EELS) are compared. In one of these studies21 the decomposition of ethylene oxide at a silver surface at 160 K was observed to give rise to a strong SERS band of CO, but in the EELS spectrum no CO was detected at this temperature. In the other study22 a silver surface was exposed to 1,2-dichloroethane at 55 K. From the SERS spectrum it appeared that the adsorbed dichloroethane was completely converted into ethylene at this temperature, whereas the EELS spectrum of the same surface showed only unchanged dichloroethane.This seems clear evidence that the SERS-active sites are particularly reactive, and it was sugges- ted that they may indeed be the sites on the silver surface which are catalytically the most effective. Other evidence relating to the existence of particular SERS-active sites has been reviewed by Otto$ and includes the quenching of SERS from silver surfaces at low temperatures by small amounts of adsorbed oxygen, which adsorbs only at high index surface sites and which thus further indicates that high index sites are the sites of the SERS activity. The fact that there is now strong evidence that SERS gives information that relates only to minority sites and is not representative of a surface as a whole is an intriguing result.There are already several other analytical techniques which fulfil the requirement of giving information which is represen- tative of surfaces as a whole, and it could turn out to be an advantage that SERS is able to probe only certain of the surface sites.4 It will be necessary to identify more clearly the nature of these sites, but the possibility that SERS is not only highly surface specific, but also on some catalytically active surfaces may be specific to the actual catalytic sites them- selves,22 is an exciting idea. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 References Chang, R. K., and Laube. B. L., CRC Crit. Rev. Solid State Muter.Sci., 1984, 12, 1. Pockrand, I.. Springer Tracts in Modern Physics, 1984, 104, 1. Cotton, T. M., Kim, J.-H., and Chumanov, G. D., J . Raman Spectrosc., 1991, 22, 729. Otto, A., Mrozek, I., Grabhorn, H., and Akemann, W., J. Phys.: Condens. Matter, 1992, 4, 1143, and references cited therein. Creighton, J. A , , in Surface Analysis Techniques and Applica- tions, eds. Randell, D. R., and Neagle, W., Royal Society of Chemistry, Cambridge, 1990, p. 13. Musiani, M. M., Mengoli, G . , Fleischmann, M., and Lowry, R. B., J . Electroanal. Chem., 1987, 217, 187. Chase, B., and Parkinson, B., J. Phys. Chem., 1991, 95,7810. Sockalingum, D., Fleischmann, M., and Musiani, M. M., Spectrochim. Acta, Part A , 1991, 47, 1475. Hong. P. P., Boerio, F. J., Clarson. S . J., and Smith, S .D., Macromolecules, 1991, 24, 4770. Blue, D., Helwig, K., and Moskovits, M., J . Phys. Chem., 1989, 93, 8080. Hutchinson, K., Hester, R . E., Albery, W. J., and Hillman, A. R., J . Chem. SOC., Faraday Trans. I , 1984, 80, 2053. Knight, D. S . . Weimer. R., Pilione L., and White, W. B., Appl. Phys. Lett.. 1990, 65, 1320. Alsmeyer, Y. W., and McCreery, R. L., Langmuir, 1991, 7, 2370. Feng, Q., and Cotton, T. M., J . Phys. Chem., 1986, 90, 983. Kaul, B. B.. Holt, R. E., Schlegel, V. L., and Cotton, T. M., Anal. Chem., 1988, 60, 1580. Mengoli, G., Musiani, M. M., Fleischmann, M., Mao, B., and Tian, Z. Q., Electrochim. Acta, 1987, 32, 1239. Uehara. J., Nishihara, H., and Aramaki, K., J . Electrochem. SOC.. 1990, 137, 2677. Leung, L.-W. H., and Weaver, M. J ., J . Am. Chem. Soc., 1987, 109. 5113. Wilke, T., Gao, X., Takoudis, C. G., and Weaver, M. J . , Langmuir, 1991, 7, 714. Feilchenfeld, H.. and Weaver, M. J . , J . Phys. Chem., 1991,95, 7771. Gu, X. J.. Akers, K. L., and Moskovits, M., J . Phys. Chem., 1992, 96, 383. Gu, X. J., Akers, K. L., and Moskovits, M., J . Phys. Chem., 1991, 95, 3696.ANALYTICAL PROCEEDINGS, JANUARY 1993, VOL 30 31 Surface Reactivity of Anti-wear Additives D. Landolt and H. J. Mathieu Labomtoire de Metallurgie Chimique/Departement des Materiaux, Ecole Polytechnique Federale de Lausanne (EPFL), MX-C Ecublens, 1015 Lausanne, Switzerland R. Schumacher Ciba-Geig y Research Laboratories, Marly, Switzerland Introduction Three lubrication regimes can be distinguished in lubricated contacts. 1 In the hydrodynamic lubrication regime, the sliding surfaces forming the contact are separated by a liquid lubricant film, leading to low friction and wear.In the mixed lubrication regime, the surface asperities touch intermittently. Elastic and plastic deformation of the asperities and the formation of microscopic junctions lead to increased friction and wear. In the boundary lubrication regime the surfaces are in direct contact, leading usually to seizure and failure of the contact. Anti-weadextreme pressure (AW/EP) additives improve friction and wear in the mixed regime, mostly, by chemical reaction with the rubbing surfaces. For the formulation of improved lubricants it is important to understand the reaction mechanisms of AW/EP additives. For this the chemical composition of surface layers formed under different conditions must be characterized and correlated with observed friction and wear behaviour.The application of surface analysis methods such as Auger electron spectroscopy (AES), photoelectron spectroscopy (XPS) and secondary ion mass spectrometry (SIMS) to tribological problems is des- cribed in two reviews.2.3 These methods are particularly well suited for studying the reactions of AW/EP additives with rubbing metal surfaces.4 The purpose of the present paper is to present experimental results obtained in these laboratories with model additives on steel surfaces under sliding wear conditions and to propose a mechanistic interpretation of the observations. The data illustrate the usefulness and limits of surface analytical methods in the area of tribology and lubrication.Experimental The most commonly used AW/EP additives contain phospho- rus and sulfur, zinc dialkyldithiophosphate (ZDTP) being a well known example. In our laboratory, zinc-free phosphorus esters based on the molecular structure shown below were used as model compounds for the systematic investigation of the relation between the molecular structure of the additive and its effect on friction and wear.5 The groups X , Y , R1 and R2 were varied systematically. The solvent was a sulfur-free synthetic hydrocarbon. Prl x = s , o C H 2-CO-N-( C3 H7 12 Y = S , O Wear tests were carried out with a ball-on-plate test device (SRV) shown in Fig. 1. This device, described in more detail in ref. 5 , allows one to perform tests with small amounts of lubricant (typically two drops) and the sample size is well adapted for surface analytical investigations.Furthermore, the mechanical quantities, normal force and tangential force, which determine the friction coefficient, can be measured easily. The apparatus allowed the variation of oil temperature in a controlled manner. Thus the friction coefficient and the extent of wear could be determined as a function of temperature for different load conditions. The latter was calculated from the wear scar sectional area, measured by a Talysurf instrument. Typical experimental conditions used are indicated in Fig. 1. In order to obtain reproducible results a running-in period of 10 min under standard load (100 N) was used in all wear tests before starting an experiment under any particular conditions.After the running-in period the friction coefficient varied relatively little with time. Normally the value measured after 2 h was taken. After a wear test the samples were carefully cleaned in an ultrasonic bath containing a 1 + 1 mixture of chemically pure acetone and white spirit. They were then stored over silica gel under argon gas before being introduced into the ultra-high voltage (UHV) system for surface analysis. The described cleaning procedure removed all soluble organic matter, such as residual oil films from the surface, leaving only adherent insoluble matter. Auger depth profiling was carried out in the PHI 550/590 AEYXPS apparatus equipped with a cylindrical mirror analyser and two differentially pumped ion guns.The electron beam of approx 1 pm diameter (2 keV, 0.5 WA) was scanned typically over 200 x 200 vm, the ion beam (2 keV) over 2 x 2 mm. After an initial survey, a depth profile was measured. The sputtered depth was approximately 20 nm. To estimate the concentrations, elemental sensitivity factors were employed throughout. The concentration values at a depth of 1.8 and 4.3 nm were computed (in some cases also at 17 nm). These values were found to be more representative of the reaction layer composition than the surface concentration values measured without sputtering which are strongly af- fected by adsorbed contaminants. The reproducibility of the AES concentration values so determined was good.4 FN 0.5 mm - H 1.2 mm Fig. 1 Schematic diagram of the ball-on-plate wear test apparatus.Operating conditions: load, 100 or 200 N; frequency; 50 Hz; amplitude, 0.5 mm; temperature, 40-120°C. Plate and Ball: Steel 100 Cr 6 (1% C, 1.5% Cr) HRC = 6232 A ANALYTICAL PROCEEDINGS, JANUARY 1993. VOL 30 I I3 1 0.14 4- S a, 0 .- E 0.12 s C .o 0.10 + .- L L L 0.08 S 0 i C D d, t 40 80 120 40 80 120 40 80 120 40 80 120 40 80 120 TemperaturePC Fig. 2 the reaction layer at a depth of 4.3 nm, measured by AES after completion of the respective wear tests Friction coefficient as a function of temperature in the presence of additives A-D. Also indicated is the phosphorus concentration in Auger electron spectroscopy yields atomic concentration, but gives no information on the chemical state of the species present.For this reason, a few complementary experiments were made by XPS, using a Mg Ka source. In addition, SIMS measurements were performed, using a combined AESlSIMS instrument (PHI 54YATOMICA 3000) equipped with a quadrupole mass analyser. This equipment could be used in the static or the dynamic mode.6 Friction and Wear Behaviour of Model Additives The effectiveness of model additives was compared by performing friction and wear experiments at different temper- atures. Increasing temperature decreases the viscosity of the solvent and thus leads to more severe wear conditions. The friction coefficient was plotted as a function of temperature. From this the critical temperature was determined at which lubrication breakdown occurred and the friction coefficient increased markedly.The transition from low to high friction was found to depend on the chemical nature of the additive present. Fig. 2 shows experimentally measured friction coefficients in the presence of different iso-geometrical phosphorus ester additives, present at a concentration of 2%. Below the transition temperature the presence of the additives leads to a lower value of the friction coefficient compared to the pure solvent (PAO). The best additives are those exhibiting the highest transition temperature. The wear rate as a function of temperature (not shown in Fig. 2) exhibits similar behaviour to the friction coefficient. The data clearly demon- strate the effectiveness of AW/EP additives in reducing friction and wear under the present experimental conditions.The worn surfaces were analysed by AES depth profiling. Fig. 2 shows the measured phosphorus concentration at a depth of 1.8 nm. The data correlate well with the friction coefficient. Under conditions of low friction, phosphorus is present at the surface, but at the transition temperature the phosphorus concentration decreases sharply. On the other hand, no such correlation was observed for sulfur. Phosphorus rather than sulfur, is therefore, the active component of the AW/EP additives tested. Chemical Nature of Reaction Layers Information on the chemical nature of the reaction layers was obtained from three independent sources, the oxygen : phos- phorus ratio measured by AES, and XPS binding energy of phosphorus and SIMS fingerprint sprectra. The atomic ratios 0 : P in the reaction layer formed under wear test conditions were found to differ from those observed for steel samples having been simply immersed in the additive containing solvent (Fig.3). The data shown were measured at a depth of 1.8 nm after immersion at 25 “C for 2 min and after a wear test at 200 N and 40°C, respectively. The surface layers formed during the wear test obviously contain much more phosphorus and sulfur than those formed during immersion. Also they are much thicker, an observation not shown in Fig. 3. The atomic ratio P : 0 in the reaction layers formed during wear testing in the low friction regime was approximately 4 : 1. This indicates that the phosphorus in the reaction layer is present mainly as phosphate. The measured XPS binding energy of phosphorus was found to be in agreement with this hypothesis.6 Additional information on the chemical state of the reaction layers can be obtained by SIMS.Negative SIMS spectra from surfaces of iron immersed in phosphoric acid (formation of iron phosphate) and from reaction layers resulting from wear testing were compared and found to contain similar phospho- rus containing fragments (Fig. 4). The data provide further support that the reaction layers formed under friction condi- tions contain iron phosphate. However, a detailed interpreta- tion of the chemical information contained in SIMS spectra was not possible, because too many factors may influence the production of fragments during the analysis.6 100 - s 80 I Er 60 4- m C 0 .- +- 40 4- C a, C s 20 0 Immersion Wear Fig.3 Elemental composition of the reaction layer at 1.8 nm derived from AES measurements performed on samples having been sub- jected to a wear test in the presence of additives A, B and C (see Fig. 2 for formulae) o r simply immersed in the base oil in the presence of 2% of these additivesANALYTICAL PROCEEDINGS, JANUARY 1993, VOL 30 33 Synergistic Effects of Antioxidants Commercial lubricants contain, in addition to the base oil and AW/EP additives, several other compounds, such as viscosity improvers, emulsifiers, detergents and antioxidants. Some of these components might possibly affect the effectiveness of AW/EP additives. Of particular interest in this respect are antioxidants, because of their chemical reactivity. The effect of selected antioxidants was investigated with additive E the structure of which is indicated in Fig.2. The transition from low to high friction and wear occurred at approximately 120 "C in the presence of this additive. At the transition temperature, the friction coefficient and the wear scar area increase and the phosphorus content of the reaction layer decreases. At the same time the oxygen content of the surface layers increases sharply, indicating that severe surface oxidation takes place.7 The presence of an antioxidant shifts the transition tempera- ture characteristic for this additive to a higher value, as shown by the data in Fig. 5 . The antioxidant in this case was a blend of a sulfur containing sterically hindered phenol and an aromatic amine added at a concentration of 0.75%.The additive concentra- 104 PO2 103 c 102 I v) v) 4- 103 102 0 50 m/z 100 Fig. 4 Negative SIMS spectra of a steel surface having been (a) immersed in phosphoric acid or ( b ) having been subjected to a wear test in a synthetic hydrocarbon containing 2% of additive A 100 I I J E *c)/ 0 B 74 40 50 60 70 80 90 100 110 120 130 140 150 Tern peratu re/"C Fig. 5 Wear scar area as a function of temperature measured in an additive containing synthetic hydrocarbon solvent in A, the presence (1.94% additive E +0.75% A03) and B, the absence (1.94% additive E) of an antioxidant 60 50 N E 40 1 2 3 Antioxidant Fig. 6 Influence of additive E and antioxidants A01, A 0 2 and A 0 3 on the measured wear scar area formed during a wear test at 120°C tion was 2%.The data given in Fig. 5 suggest that the antioxidant improves the efficiency of the additive by inhibit- ing surface oxidation. To verify this hypothesis the wear behaviour at 120 "C (just above the transition from low to high wear in the presence of additive E) was investigated for four different lubricant formulations involving a common base oil. Formulation (a) corresponded to the pure base oil, formula- tion ( b ) to the base oil + additive E , formulation ( c ) to the base oil + antioxidant and formulation (d) to the base oil + additive E + antioxidant. Three different antioxidants were used, a sulfur containing phenol (AOl), an aromatic amine (A02) and a mixture of both (A03). The antioxidant concentration was 0.75% in all cases, while the additive concentration was 1.92%. Results of the different experi- ments are summarized in Fig.6. They show that under the experimental conditions chosen (just above the transition temperature) the base oil with and without additive leads to a similar wear rate. Adding only antioxidant to the base oil does not improve the wear behaviour and in one case even leads to significant deterioration. On the other hand, for all three antioxidants tested, a mixture containing both additive and antioxidant gives a much lower wear rate. The results demonstrate the existence of a synergistic effect between the AW/EP additive and the antioxidants. Reaction Mechanism The most widely accepted wear mechanisms are: adhesive wear, abrasive wear, fatigue (delamination) wear and oxida- tion wear.8 Adhesive wear involves the formation and destruction of welded junctions at asperities touching each other.Material removal is due to tearing off of metal from these junctions. By forming a reaction layer at the rubbing surfaces AW/EP additives prevent welding at the metal-metal contacts. The present data and others are in agreement with such a hypothesis. Abrasive wear involves removal of material by ploughing, which results in plastic deformation and fracture. The wear particles thus formed (oxides, metal) act as abrasives. An influence of the additives on this mechanism has not been demonstrated to date. A possible action on the size and mechanical properties of the wear particles formed cannot be excluded a priori, however. Delamination wear, important in brittle materials, involves formation of fatigue cracks below the surface and periodic rupture.It is unlikely that the additives influence this mechanism in a significant way. Oxidation wear involves periodic formation and removal of34 ANALYTICAL PROCEEDINGS, JANUARY 1993, VOL 30 thin oxide layers.3 The AW/EP additive can reduce this type of wear by reacting with the surface in competition with oxygen. The synergistic effect of antioxidants is in favour of such a hypothesis. From the chemical surface and other data presented, three steps can be distinguished in the reaction mechanism of an AW/EP additive with steel. Firstly, the additive molecule dissolved in the oil must be able to adsorb at the metal surface. The molecular structure of the additive, notably its steric effects, may play an important role at this stage.The molecular structure also determines the solubility of the additive in the base oil and its intrinsic thermal stability. These factors are of importance because according to the adsorption isotherm, the surface coverage depends on the bulk concen- tration of the additive. Insufficient thermal stability leads to a decrease in the bulk concentration with time. In a second step, the adsorbed additive molecule is torn apart, by the prevailing shear forces, and highly reactive molecular fragments are formed. This process, called tribo- fragmentation, depends on the prevailing mechanical parameters and the temperature. The occurrence of tribo- fragmentation is supported by the observation that at equal temperature, immersion experiments produce less phos- phorus at the surface than wear experiments.9 In a third step, the molecular fragments react chemically with the metal surface, the end product of the reaction sequence being essentially iron phosphate.In the mixed lubrication regime, under relatively mild conditions, these act as a solid lubricant and prevent adhesive wear. Under more severe conditions they reduce surface oxidation by reacting with surface sites in competition with dissolved oxygen, thus reducing oxidative wear. The thickness of the phosphate reaction layer probably depends on the test conditions. Under the present experimen- tal conditions is was found to be of the order of several tens of nanometres in the low wear regime. Above the transition temperature the phosphorus reaction layer no longer forms, and the reaction between the metal and oxygen becomes dominant. Several factors, not yet well understood, could contribute to the observed change in mechanism at the critical temperature. On one hand, the adsorption equilibrium of the additive is less favourable at higher temperature, decreasing the amount of adsorbed additive and hence the rate of tribo-fragmentation. Another possibility is that the activation energy of the reaction of oxygen with the metal is higher than that of the reaction of additive fragments. In conclusion, the present data show that surface analytical techniques, such as AES, XPS and SIMS, used in combination with well designed wear expqriments, can yield interesting information on surface reactions of additives under wear test conditions, information not accessible by other methods. Many aspects of the reaction mechanism of AW/EP additives need further study. Surface analysis methods should prove very useful for further progress in this field. References Czichos, H., Tribology, Elsevier, Amsterdam, 1978. Buckley, D. H., Surface Effects in Adhesion, Wear and Lubrication, Elsevier, Amsterdam, 1981. Quinn, T. F. J . , Physical Analysis for Tribology, Cambridge University Press, Cambridge, UK, 1991. Mathieu, H. J . , Landolt, D., and Schumacher, R., Wear, 1981, 66, 87. Schumacher, R., Landolt, D., Mathieu, H. J . , and Zinke, H., ASLE Trans., 1983, 26, 94. Mathieu, H. J., Schumacher, R., and Landolt, D.. Wear, 1989, 132, 99. Schumacher, R., Zinke, H., Landolt, D., and Mathieu, H. J., Wear, 1991, 146, 25. Rabinowicz, E., Friction and Wear of Materials, Wiley, New York, 1965. Schumacher, R., Zinke, H., Landolt, D., and Mathieu, H. J . , 5th International Kolloquium Additive fur Schmierstoffe und Arbeitsfliissigkeiten, Technische Akademie, Esslingen, 1986, V O ~ . I, p. 3.9.1-19.