ANALYTICAL PROCEEDINGS, JANUARY 1993, VOL 30 Surface Analysis 1 13 ATR The following are five of the papers presented at the Industrial Division symposium of the Annual Chemical Congress of the Royal Society of Chemistry, held on April 13-16, 1992, in the University of Manchester Institute of Science and Technology. ~~ Reflection-absorptlon(RAIRS) Emission (EMS) Fourier Transform Infrared Reflection Spectroscopy for Surface Analysis Monolayer Thin films on metals. Electrode surfaces ~ 100A Metal (wire, rough surfaces, etc.,) catalysts, flames J. Yarwood Department of Chemistry, University of Durham Science Laboratories, South Road, Durham DH 1 3LE Introduction Infrared spectroscopy is already extensively used in analysis, both qualitatively (for compound identification)l and quanti- tatively2 (in order to determine concentration).It is less commonly used in ‘surface’ analysis. Indeed it may be thought by some that, because one is dealing with photons of relatively long wavelength (typically 5-25 pm), true ‘surface’ analysis is not possible. It is worth recapping, therefore, on the attributes 1 Substrate-molecule and molecule-molecule interactions (from band multiplicity, bandwidths and frequency shifts) 2 Molecular orientation and packing; degree of ‘chain‘ ordering and ‘tilting’ from linear dichroism measurements 3 Chemical changes (‘finger printing’) e.g., proton transfer, surface degradation, chemical reactions) 4 Electronic changes (conductivity, pyroelectricity, etc.) 5 Molecular arrangement and re-arrangements (crystallinity, fluidity, phase changes, efc.) Major Advantages of FflR ( a ) Non-destructive and ‘universally’ available ( 6 ) ng sensitivity (monolayer and upwards) ( C ) ’in situ’ measurements possible (in devices) (d Relatively cheap (= f20K) Fig. 1 Information available using Fourier transform infrared spectroscopy 1000 8, 500 A Surface ~1 pm XPS SERS 4 Depth profile 4 SIMS Fig.2 Comparison of sampling depths of various ‘surface’ sensitive techniques. ATR (attenuated total reflection); XPS (X-ray photoelec- tron s ectroscopy); SERS (surface-enhanced Raman scattering); SIMS gecondary ion mass spectrometry) of the infrared technique (Fig. 1) in a form which emphasizes that surface sensitivity may be achieved down to S 1 mono- layer (typically 10-20 A of surfactant or polymer or a few tens of nanograms).Surface selectivity is more problematic because, as indicated above, penetration depths are of the order of the wavelength (up to a few micrometres). This is, of course, very high compared with some other techniques (Fig. 2) but infrared reflection techniques do offer opportuni- ties for depth profiling if the sample is suitably prepared and if the correct spectral parameters are measured (see below). Infrared spectra can at present be collected in a wide variety of ways. Fig. 3 gives a summary of the sampling procedures which have been employed. Some of the techniques have been devised specifically to cope with particular sample types [e.g., diffuse reflectance infrared Fourier transform spectroscopy (DRIFT) for powders, photoacoustic spectroscopy (PAS) for ‘black’ materials such as coal, surface electromagnetic wave (SEWS) for organic thin films on metals].This paper deals only with reflection techniques [attenuated total reflectance (ATR) and reflection-absorption infrared spectroscopy (RAIRS)] which may be employed to examine a wide variety of samples but focusing here on the ways in which surface sensitivity and selectivity may be achieved in a (well-founded) analytical laboratory. Fig. 4 compares the conventional transmission experiment with RAIRS and ATR techniques for an organic film with a thickness that may vary from <20 A to several micrometres. The key to obtaining information about molecular orientation and/or ordering lies in the ability to choose a particular electric field polarization (see later). First, the method of obtaining structural and quantitative information is outlined using these techniques and then some examples are presented from our recent work (and that of others) to illustrate the principles involved.paint, mbbers, cloth, etc. Diffuse reflectance G S T i c ( P A S ) I -1 ktm I Carbon, coal, ceramics 1 Surface electromagnetic wave (SEWS) = 1oA Polymer films, organic thin films on metal Fig. 3 FTIR spectroscopy Summary of the most common sampling techniques used in14 ANALYTICAL PROCEEDINGS, JANUARY 1993, VOL 30 (a) Transmission or absorption Samples dipoles parallel to substrate surface ( b ) Reflection-absorption (RAIRS) dipoles perDendicular to substrate surface Uses reflection from a metallic surface.Samples EL I s Thin film sample (c) Attenuated total reflection (ATR) interface to increase sensitivity. Samples dipoles both parallel and perpendicular to substrate surtace Uses many reflections at solidiair or solid/liquid Silicon ATR crystal Fig. 4 Three infrared sampling techniques compared Theoretical Background The infrared intensity of the band (corresponding normal mode Qi) which arises from the transition v = k t m depends on the interaction between the incoming photons and the vibrating molecule .3-4 This interaction is computed via the transition moment <rnlp.Elk> where 1.1. is the total dipole moment and E is the external electric field. The transition probability (the probability per unit time for the k c m transition to occur) is given by Pmk 0~ [<mi P.EI k >I2 (1) Expanding the dipole moment in a Taylor series about the equilibrium atom positions for a given normal mode gives 1.1.= PO + (31.1.IaQi) Qz (2) in the harmonic approximation.5 Thus the transition probabil- ity Pmk becomes? ( 3 ) (4) There is a population factor ( N , - N,) involved also but for fundamental transitions (v = 1 t 0) in absorption this is virtually constant. Eqn. (4) indicates that the infrared intensity depends on the square of the cosine of the angle between the transition dipole Mi = ap/aQj and the incident electric field E (see Fig. 5 ) . As Mi directions depend on the t The integral < r n / ~ l k > is zero as Vrm and Yk are orthogonal wave tz (a) RAIRS Sh bstrate t' (b) ATR \ Substrate Fig. 5 Orientation of electric field components Ex, E , and E, relative to the substrate and a typical transition dipole (acl/aQi) of the molecule ( a ) for a RAIRS experiment, ( b ) for an ATR experiment ( a ) Incident beam Reflected beam Thin organic -n b Ere,.film einc 4 45 90 I I 68 Fig. 6 ( a ) Incident and reflected infrared beam polarization com- ponents (p, polarized light is in the plane of incidence). ( 6 ) Corresponding phase shifts as a function of Oinc on a metal surface way in which molecules are oriented (or arranged) at a surface or interface (Fig. 5 ) , and as the Cartesian components of E; El., Ey and EZ can be varied using various optical techniques (including polarization), eqn. (4) forms the basis for surface- functions. sensitive analysis. Clearly, for a given molar absorptivity, theANALYTICAL PROCEEDINGS, JANUARY 1993, VOL 30 15 observed intensity depends on the size of E* and the value of cos $.Results and Discussion The easiest technique to understand in this context is RAIRS3.6.7 in which infrared radiation is incident at 'grazing' incidence on a metal surface [Fig. 6(a)]. It may be seen [Fig. 6(6)] that s polarized light suffers a 180' phase shift (8) on reflection from a metal surface and so E,(l) and E,(r) cancel at the metal/sample interface. On the other hand E,(l) and E,Cr) differ by -90" at large 8 and so vectorially add together in such a way that maximum field (and maximum intensity) occurs near grazing incidence Oinc = 85-88'. This is the origin of the surface selection rule6-8 which results in the very useful ability to distinguish vibrations which have a transition dipole ( M i ) with a large component perpendicular to the substrate. Vibrations parallel to the substrate have zero intensity.Furthermore, there should be no spectrum for s polarized light (see Fig. 7). Fig. 8 shows that for methylene chains oriented perpendicular to the substrate the va(CH2) and vs(CH2) transition dipoles ( M , ) are parallel to the substrate. Fig. 9 shows16 then that such bonds are extremely weak in RAIRS 0.04 0.03 Q, C m I) v) I) 0.02 a 0.01 0 2918, / mp 285 1 1639 1012 ! 1470 I I 1 I I 3500 3000 2500 2000 1500 1000 Waven u m be rlcm - Fig. 7 Demonstration that the s polarized RAIRS spectrum (a,) has zero intensity. The sample is 10 Langmuir-Blodgett monolayers of a dye JT-11 which has a CZ2 chain attached to an acetylenic head group M - SH 0) t m I) 0.40 0.30 0.20 0.10 0 L 0, a n 1 .oo 0.90 0.80 0.70 0.60 3000 2800 3000 2800 0.20 0.16 0.1 2 0.08 0.04 1700 1500 1300 1.60 1.40 1.20 1 .oo 0.80 0.60 1700 ,1500 1300 Wave n u m be r/cm - 1 Fig.9 Spectra of 21 Langmuir-Blodgett layers of cadmium docosan- oate ( ~ 4 0 0 8, thick) on a silicon ATR crystal; (a) and ( b ) show the spectra in ATR mode. The v,(CH2) and v,(CH2) bands are intense; (c) and (d) show the spectra in RAIRS mode (with an overcoating of metal). Now the v,(CH2) and v,(CHZ) bands are weak, relative to those due to CH3 vibrations. (For details see ref. 16) Substrate Fig. 8 Orientation of the va(CHZ) and v,(CH2) transition dipoles parallel to the substrate for an alkyl chain oriented perpendicular to the substrate.The v,(CH3) and v,(CH3) transition dipoles have components perpendicular to the substrate 1 .o 0 El €0 Fig. 10 Schematic representation of the basic ATR experiment. The evanescent wave penetrates the sample and decays exponentially with distance in the z direction16 ANALYTICAL PROCEEDINGS, JANUARY 1993, VOL 30 (compared with ATR, where two components of the field are parallel to the substrate) [Fig. 5(b)]. Such comparisons enable the degree of orientation of molecular monolayers to be assessed semi-quantitatively.6 One disadvantage of RAIRS is that a single reflection leads to relatively poor sensitivity, although monolayer spectra have been reported .9JO The ATR experiment3.12-14 is more complex (Fig. 10) because the incident electric field will have non-zero com- ponents in all three directions (for unpolarized light) [Fig.5(b)] and the use of linear dichroism is more difficult although not impossible.3,10-14 The principal advantages of ATR are (a) sensitivity and ( b ) the ability to carry out depth profiling,l5 (see below). Sensitivity enhancement is achieved by multiple internal reflection (up to 50 reflections are possible).l6 Depth Profiling With ATR Spectroscopy Depth profiling possibilities arise because the evanescent wave falls off exponentially into the rarer medium (usually the sample). Thus the field (E/Eo)2 varies rapidly as a function of distance z away from the substrate surface (Fig. 10). Varia- tions in concentration and crystallinity, for example, might then be detected by a variation of the penetration depth, d,.with E = Eo exp(- zld,) (6) where h is the wavelength of the band of interest, and Eo is the electric field at the surface of the ATR crystal. Clearly d, can be varied by changing A, Oinc or n2 (and therefore nz1). Figs. 11-13 illustrate that this does work. In principle, at least, one can adjust d, so that the evanescent wave samples material near an interface between (say) two polymers. By spectral subtraction of the data at two or more d, makes it is possible for interfacial interactions (or structure) to be examined. Convincing results have yet to be reported. Very recently, however, there has been a report17 of depth profiling using the method originally advocated by Hirschfeld. 18 This involves the recognition that absorbance measurements as a function of €lint may be converted to a concentration profile C(z) via the Laplace transform where R(0) is the reflectivity at incident angle 8 and a(z) is the absorption coefficient as a function of z away from the crystal surface.The sample transmission is Z/Zo = exp(- ad,), where d, is the thickness in transmission which is equivalent to the amount of material probed in the ATR experiment. Fina and Chen17 have shown how eqn. (7) may be solved to yield depth-dependent information about the crystallinity of PET (polyethylene terephthalate) (Fig. 14). Bands attributed to trans-conformers (of the PET chains) decrease in intensity as a function of z. This is because a high level of all-trans chains is found (as expected) for highly crystalline material formed by quenching from the melt at the ATR crystal surface. In a corresponding way, the bands attributed to gauche-conform- ers increase with z reflecting a higher proportion of amor- phous polymer further from the surface.This work promises to provide a basis for important future work in this area. Determination of Adsorption Isotherms Using ATR Spectroscopy For in situ adsorption studies on to an ATR crystal (or a surface layer on an ATR crystal) it is usual to bring the crystal into contact with a solution containing the adsorbate. The resulting spectrum will, of course, reflect both adsorbed and bulk species (after solvent spectrum removal). The procedure for separation of such components in order to measure the surface excess concentrationlg has been described by Sperline and co-workers.20921 The surface excess is defined as Ti = $/o 0 C m e 8 a B I I I I I I 4000 3500 3000 2500 2000 1500 1000 Wavenum bedcm - 1 Fig.11 45". ( b ) Difference spectrum after subtraction of the underlying PMMA spectrum FT infrared spectra of ( a ) a poly(methy1 methacrylate)-poly(viny1 alcohol) polymer laminate on a ZnSe ATR crystal with @inc =ANALYTICAL PROCEEDINGS, JANUARY 1993, VOL 30 17 Fig. 12 Difference spectra of PVA [poly(vinyl alcohol)] at different 'barrier film' thicknesses of PMMA [poly(methyl methacrylate)] in the laminate (€Iinc = 45") showing smaller effective penetration depths into the PVA as the PMMA layer is made thicker; d = A, 0.345; B, 0.396; C, 0.646; D, 1.065; and E, 1.900 pm 0 1 .o N u7 s 0 I I I I I 1 0.4 0.8 1.2 1.6 0.4 0.8 1.2 1.6 Distance z/pm Fig.13 Comparison of (LIE# fields and PVA (1100 cm-l band) infrared intensities as a function of distance away from the ATR crystal surface. (a) A, €Iinc = 39"; B, €Iinc = 45"; and ( b ) A, Oinc = 60" where qi' is the number of moles adsorbed at the interface and <T is the surface area of the interface. In order to measure T$' separately (from the bulk concentration) one needs to integrate the equation for ATR intensity. The absorbance per reflection, AIN of a particular ATR band of interest, is, then where E is the molar absorpthity of the band and 8 the incident angle. [The other parameters are as defined in eqn. (5).] The n g l i 0 0 2.0 4.0 6.0 8.0 10.0 5 0.025 .- + sl s: $ 0.022 0.019 0.016 0.01 3 0.010 0 2.0 4.0 6.0 8.0 10.0 Depth from interface/pm Fig.14 Concentration profiles of trans (T) and gauche (G) conform- ers in a film of quenched PET (or an ATR crystal) as a function of distance from the interface. (a): A, The 1340 cm-1 band (T) and B, the 1505 cm-1 band (T and G). ( b ) : A, The 1370 cm-1 band (G) and B, the 1455 cm-1 band (G). (Reproduced, with permission, from ref. 17) integral has been performed for 'bulk' and surface phase to give AIN = E Cb d, + E (2 d,Id,) (tit) (9)18 ANALYTICAL PROCEEDINGS, JANUARY 1993, VOL 30 € 3 I I I I 1 I t 0 0.01 0.02 0.03 0.04 0.05 B u I k co nce nt rat ion/mol d m - 3 Fig. 15 Adsorption isotherm of sorbitan monopalmitate on an Si-SiOH surface from carbon tetrachloride at 298 K 0 II CH2-0-C-C H2 ( CH2 ) 1 jC H3 I HO-CH OH Fig. 16 Sorbitan monopalmitate, SPAN 40 w 4.4 4.6 4.8 5.0 5.5 6.0 6.5 7.0 I I I I I I I I 2200 2000 1800 1600 1400 Wave n u m be rlc m - Fig.17 Detection of diffusion across a polymer-polymer interface: A, the spectrum of an epoxy resin (4 ym); B, the spectrum of an epoxy-urethane laminate; and C, the difference spectrum A - B showing the two bands (marked) due to the isocyanate-urethane product diffusion into the epoxy layer where cb and ci are the bulk and surface concentrations, t is the adsorbed layer thickness and d, is the effective depth into the bulk sample.l3%14 Note that the molar absorptivity is assumed to be unchanged between ‘bulk’ and ‘surface’ species. This is expected to be a good approximation for v(CH2) bands of surfactant chains but it is probably a very poor approximation for ‘head group’ vibrations.Fig. 15 shows the adsorption isotherm of sorbitan monopalmitate (SPAN-40) (Fig. 16) or Si-SiOH from carbon tetrachloride. Note that a calibration of the ATR cell is required to determine N and Oinc accurately and that some ATR cell types22 are unsuitable for such quantitative measurements. Studies of Molecular Diffusion Using ATR Spectroscopy From Fig. 10 it can be seen that diffusion of one material into another across an interface or from the external surface into an organic thin film can be detected by the evanescent field. Fig. 17 shows an example of such detection. In this case a 4 pm polymeric film, a cross-linked epoxy resin, has been ‘over- coated’ onto a ZnSe ATR crystal with a polyurethane containing both isocyanate and amide chemical groupings.It may be immediately observed that bands at 2260 and 1680 cm-1 corresponding to v(N=C=O) and v(NHC=O) , respec- tively, arise (and increase with time) in the subtracted spectrum. As the epoxy film thickness is much higher than 3 x 4 1 3 such bands must arise from diffusion across the polymer/ polymer interface. From this, and the other examples, given in this paper it is clear that there is much scope for development of the infrared surface sensitive techniques into areas impor- tant to, although rarely tackled by, analysts and other industrial scientists. Thus, areas such as depth profiling and surface excess measurements, although known to be feasible for some time, have not been used extensively in our industrial context.It is hoped that this review will have whetted a few appetites. The author acknowledges and thanks the collaborators who performed the innovative spectroscopy described here: M. Pereira, Y. Song, J. Tait, G. Davies and S. Nunn. Thanks are also due to Dr. M. Petty for access to Langmuir-Blodgett dipping facilities and to the SERC, Shell (UK) and Courtaulds Coatings for financial support. 1 2 3 4 5 6 7 .8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 References Bellamy, L. J., Infrared Spectra of Complex Molecules, Chap- man and Hall, London, 1973, vols I and 11. Haaland, D . H . , in Practical Fourier Transform Infrared Spectroscopy, eds. Ferraro, J . R., and Krishnan, K., Academic Press, New York, 1990, ch. 8 (and references cited therein).Debe, M. K., Prog. Surface Sci., 1987, 24, 1. Steele, D . , Theory of Vibrational Spectroscopy, W. B. Saund- ers, London, 1971. Atkins, P. N., Molecular Quantum Mechanics, Oxford Univer- sity Press, Oxford, 1984, p. 50. Golden, W. G. in Fourier Transform Infrared Spectroscopy, eds. Ferraro, J. R., and Basile, L. J., Academic Press, New York, 1985, ch. 8. Hayden, B. E., in Methods of Surface Characterisation, eds. Yates, J. T., and Madey, T. E., Plenum Press, New York, 1987, vol. 1, p. 287. Greenler, R. G . , J. Chem. Phys., 1966, 44, 310. Rabolt, J. R., Jurich, M.. and Swalen J. D . , Appl. Spectrosc. 1985, 39, 281. Chabal, Y. J., Surface Sci. Reprints, 1988, 8, 211. Wendlandt, W. W., and Hecht, H. G., Reflectance Spectro- scopy, Interscience, New York, 1966. Ishida, H . , Rubber Chem. Technol., 1987, 60, 497. Harrick, N. J., Internal Reflection Spectroscopy, Harrick Science Corporation, New York, 2nd edn 1979. Mirabella, F. M., Appl. Spectrosc. Rev., 1985, 21, 45. Urban, M. W., and Koenig, J. L., Vib. Spectra Struct., 1990,18, 127. Davies, G. H., and Yarwood, J., Langmuir, 1989, 5 , 229. Fina, L. J., and Chen, G . , Vib. Spectrosc., 1991. 1, 353. Hirschfeld, T., Appl. Spectrosc., 1977. 31, 289. Hunter, R. J., Foundations of Colloid Science, Oxford Univer- sity Press, Oxford, 1989, vol. 1, ch. 8. Sperline, R . P.. Muralidharan. S . , and Freiser, H., Langmuir, 1987, 3, 198. Sperline, R. P., Muralidharan, S., and Freiser, H . , Appl. Spectrosc., 1986, 40, 1019. Afran, A . , SPECAC Applications Note No. 17.