General Introduction BY P. C. S. HAYFIELD Imperial Metal Industries Limited P.O. Box 216 Kynoch Works Witton Birmingham B6 7BA Received 7th May 1971 In order to review the present state of optical studies for the detection of and quantitative measurement on adsorbed layers at interfaces it seems appropriate to consider four main topics (i) circumstances leading to the acceptance of optical methods for studying interfaces ; (ii) present-day interests in applying optical methods ; (iii) relative merits of the various optical methods ; (iv) interpretation of optical data. CIRCUMSTANCES LEADING TO THE ACCEPTANCE OF OPTICAL METHODS FOR STUDYING INTERFACES During the 19th century much attention was given by physicists to the charac- teristics of reflection of light from surfaces and during this period and after much debate it became firmly established that interfacial films could exert considerable effects on reflection.There were numerous contributors to this finding,l including such names as Malus Fresnel Brewster and Rayleigh but Drude is perhaps most frequently remembered because in connection with optical studies he derived equations relating to the optical properties and thickness of a surface film on a solid to the characteristics of reflection of polarized light and these equations still carry his name tan $ exp (iA) = FP/FS; 6 = -27& cos ;62dlL where Fi2 FS,, F g 3 and Fi3 are the so-called Fresnel reflection coefficients and d is the thickness of film. These form the basis for interpretation of most present-day ellipsometry and reflectivity data. At this stage optical techniques for evaluating surface films had been formulated but it was after another 40 years to the beginning of the 1930’s before such techniques found appreciable application.While the basic derivation of the Drude equations which involve multiple reflection within a surface film does not seem to have been in much doubt Constable’s explana- tion for the interference colours produced on tarnished copper-his spectrophoto- metric data being published in 19274id not find ready acceptance. However any sceptism existing on the validity of the Drude equations was removed by the experi- mental work reported in the early 1930’s. Using the ellipsometer technique Tronstad and collaborators 4-7 investigated several surface phenomena. Concerning the passivation of stainless steel in sulphuric acid it was shown that growth of film occurred of average thickness 50A and average film refractive index 3.0 and this was presumed responsible for the improved corrosion resistance imparted.In relation to adsorption studies Feachem and Tronstad made ellipsometer studies on films having well-defined properties choosing long-chain fatty acids and alcohols on 7 8 GENERAL INTRODUCTION mercury and their results interpreted on the basis of change in phase A between rp and rs components of reflected elliptically polarized light being linearly proportional to film thickness gave strikingly good agreement between known figures and those calculated as a result of the optical studies (table 1). TABLE 1 .-ELLIPSOMETRIC DATA FOR THE THICKNESS OF ADSORBED LONG CHAIN MOLECULES ON MERCURY.* G(J/ - $0) nj = Cl+ Simplification for thin uniform transparent (A - Ao) sin 2+ (A - Ao) (1 - 1 In%) isotropic films d = C j - film thickness nM in nM ellipsometry data chain length calc.from pelargonic acid 1.2 1.5 lauric acid 1.6 1.7 myristic acid 1.9 2.2 * C. G. P. Feachem and L. Tronstad,Proc. Roy. Sot. A 1934,23,127. Various shortcomings of the method were appreciated such as the fact that interpretation was based upon a perfectly sharp interface between the different media whereas with mercury for example the excursion of surface molecules due to thermal agitation could become a factor. The possibility of error from the roughness of solid surfaces optical anisotropy in films and also changes in absorption due to the increase of free electron concentration induced by applied electric fields were mentioned and it was argued justly by reference to recent work that the magnitude of the errors was smaller than the accuracy of their techniques for measurement of elliptically polarized light.On the theoretical side there was not much incentive for progress because the Drude equations fairly adequately covered the situation. These become difficult to handle in treating optically anisotropic media and Darwin * proposed an alternate theory based upon the film consisting of a three-dimensional array of scattering centres. These equations themselves are complicated and have not been much used but for the special case of an optically isotropic film do simplify to the form of the Drude equations. At the time Tronstad and collaborators were experimenting with the detection of monolayer films Strachan was motivated to put forward a theoretical treatment of reflection from a surface film consisting of a two-dimensional array of scattering centres in the plane of the surface and this like Drude’s equation is of particular interest in connection with present-day considerations of partial monolayer coverage of adsorbed species.PRESENT-DAY INTERESTS I N THE APPLICATION OF OPTICAL STUDIES From the foregoing section it may be seen that the basis for optical studies on interfaces had been both theoretically formulated and experimentally justified in the early 1930’s and all. that remained was for the techniques to be applied and further refined. Development in ellipsometry has mainly come in the precision and speed of recording reflected elliptically polarized light modern equipment incorporating the advances of light detectors and electronic equipment generally but an important consideration also has been the use of computers which largely remove the considerable tedium at one time attached to the computation side, P .C. S. HAYFIELD 9 Archer’s work lo in the early 1960’s on adsorbed films on silicon in the 0-1 nM range is a fine example of the considerable sensitivity and scope for study that can be achieved by ellipsometric methods and it is worth pausing to consider his paper to the 1st International Conference of Ellipsometry as it embodies many of the considerations that will no doubt come up for discussion in the present Symposium. Working at a fixed wavelength and with films of thickness down to one mono- molecular layer he considered that one might reasonably apply the simplified Drude equations where A and $ have the usual connotations used in ellipsometry d is the layer thickness and a and are constants.It seemed to be scarcely credible to suppose that a theory based upon a homogeneous parallel-sided solid film could be extrapolated to predict the change in reflectivity from a surface covered with a partial monolayer. How for example would one define the refractive index of the layer? Perhaps by defining an effective refractive index proportional to coverage e.g. A = Ao-ad; $ = $o-Bd (3) n = 8n2+(1-8) effective film refractive index (4) for a film of refractive index n2 for a complete monolayer and 8 representing fractional coverage. There would perhaps be better justification for using a refractive index based upon molecular refractions nZ,-1= ( IZ’ - I) ( n i - 1) n2,+2 n2,+2 n,+2 4 a + 2 4 b where qa and ab are the fractional volumes of components a and b and proportional to 0.Rearranging and putting qb = 1 -qa this expression simplifies to the perhaps more widely-known Maxwell-Garnet relationship Archer considered it more realistic to think in terms of Strachan’s theory of a two- dimensional array of scattering centres in the plane of the surface and this he showed led to equations of the form a and p are constants oi is the oscillator strength for unit area and assumed pro- portional to coverage by writing oi = Oaf. Hall later commented that the oscillator strengths were more likely to be proportional to the intensity of the scattered wave rather than its amplitude and thus proposed This situation thus set rather a challenge to the experimentalists which has been answered for a number of situations by comparison of the results of ellipsometer investigations with those of other techniques table 2.Archer lo found good corre- spondence for a linear relationship between analysis of isotherms for the thickness of adsorbed molecules-water carbon tetrachloride and acetone on silicon surfaces having relatively smooth surfaces and the results of elIipsometer estimations- 10 GENERAL INTRODUCTION but not for rougher substrates. Tennyson Smith l 2 worked on adsorption on mercury measuring surface tension contact potential and ellipsometry data for coverages from 0.05 to 1.0 and also found that ellipsometer changes were linearly proportional to coverage rather than to the square root of coverage.However it cannot be assumed that a linear relationship exists for all configurations of coverage and it is most valuable to have further evidence such as e.g. reported in the first paper at this Symposium; it involves not only ellipsometry but Auger electron spectroscopy and gas-volumetric analysis. The relationship might be expected to depart from linearity where the sites of adsorption show a preferred regularity such as particular crystallographic sites on a single crystal material. TABLE 2.-uSE OF MULTIPLE TECHNIQUES INCLUDING ELLIPSOMETRY FOR STUDYING ADSORP- adsorption process water vapour etc. on silicon organic molecules on mercury oxygen on tunsten polar organic compounds on various metals various compounds on silicon krypton oxygen etc.on silver oxygen on platinum TION PROCESSES techniques investigator($ ellipsometry B.E.T. Archer ellipsometry surface Tennyson Smith l tension contact potential ellipsometry L.E.E.D. Melmed Layer and field electron emission spectroscopy and Melmed l4 Kruger l 3 and Carrol ellipsometry Miller and Berger l5 radio t racer. ellipsometry B.E.T. Meyer and Bootsma Auger electron spectroscopy Meyer and Spamaay l7 ellipsometry L.E.E.D. Muller S teiger mass spectrometry Somorjai and Morabito ellipsometry Barrett and Parsons l7 reflectivity With the availability of such powerful optical techniques has come a diversity of interests which can arbitrarily be subdivided into three groups. (a) Those interested not so much in mechanism of adsorption as identification of adsorbed ~ p e c i e s l ~ - ~ ~ but information on the geometry of adsorption nature of the attachment forces and strength of the adsorption are also important related features.The work stems largely from trying to apply infra-red absorption spectroscopy to thinner and thinner layers and eventually adsorbed Much use has been made of internal multiple reflection between parallel plate technique^,^ and they have usually been applied with non-polarized light although after a few multiple reflections the beam has become polarized by absorption of the component in the plane of incidence. A new facet to optical studies using reflectivity methods was given in the papers by Koch 27* 28 on the change of reflectivity from surfaces covered with film the film being formed electrochemically.The greater sensitivity to film changes was found in the ultra-violet which is perhaps to be expected since light penetration is limited to a few atom layers only and it would be logical to expect an adsorbed layer to P . C . S . HAYFIELD 11 introduce a proportionately larger effect on the reflection process compared with a situation where the light penetrates more deeply. This result seems to have provided the stimulus for considerable further work on the application of optical methods and in particular modulated reflectivity measurements to electrochemical studies. (b) With the refinement of optical techniques particularly modulation processes giving sensitivity for studying small changes in reflectivity (AR/R N 1 0 3 induced by various perturbing forces has come renewed interest in optical methods of studying the electron structure in the surface layer of semiconductors and metals.S e r a ~ h i n ~ ~ and later Cardonna et have employed this technique with con- siderable success to the fine structure of semiconductors achieving resolutions of peaks and/or dips which may be as narrow as 0.04 eV. Subsequently the technique has been applied to metals the objective being to perturb the surface layer in a manner which can be predicted to affect the free electron concentration. From measurements of change in reflectivity over a range of wavelengths the results are interpreted in terms of dielectric changes with frequency by means of the Kramers-Kronig trans- formations. In turn the dielectric changes are interpreted in terms of free electron concentration.On the basis of the original Drude theory this can be written as E = E - (i4m/m) and E = 5 ; = 1 - [4nNe2/m(w2 + y 2 ) ] CT = Ne2y/m(02 +y2) (9) where E is the complex dielectric constant and CJ is the conductivity; e and m are the charge and mass of the electron N is the concentration of free electrons m the angular frequency and y is a constant describing the viscous damping of the electrons and equals 1 /z where z is a relaxation time. This relationship holds for frequencies less than coo which corresponds to the internal photoelectric effect but from these formulae if the effective number of free electrons changes then accordingly there will be a change in dielectric and hence optical properties. (c) The third branch of optical studies is the extension of optical methods to define aspects of corrosion and electrochemical processes.For example there has been continuing use of ellipsometry to study adsorption processes,16 and others will be discussed in the present Symposium. In connection with basic electrochemical reactions of which adsorption of species forms an integral part there have been both ellipsometric and reflectivity investigations. Particular success has accompanied the modulated reflectivity process which is well exemplified by the papers of McIntyre and Kolb and Bewick and Tuxf0rd.l’ COMPARISON OF REFLECTIVITY AND ELLIPSOMETRIC METHODS The various basic optical parameters that may be measured are itemized below. q number of single reflection multiple reflections R = 3 ( ( r P ) z + r y ( s ) 2 ) Rq reflectivity ( r p ) ( r p ) 2 q ellipsometry tan $ exp (iA) (tan $)q exp (iAq) These include overall reflectivity 3 (rP2 + rs2) rp2 rs2 or ellipsometry A = dp- ds and t,b = tan-l ~ r p / / ~ r s ~ all quantities that can be calculated directly from the Drude equations (1).Some comparisons between reflectivity and ellipsometry changes for the same situation are shown in fig. 1 and other data relating to reflectivity changes alone in fig. 2 and 3. Since the total reflectivity is the sum of rp and rs components and the sign of the reflectivity change with increasing thickness may differ for the two components it is (rSY (rS)2q 12 GENERAL INTRODUCTION better in some respects to measure one or other component individually rather than the resultant of the two. The relative change in AR/R for unfilmed and filmed situations can be improved by multiple reflection between parallel plates.This seemingly is a direct way of improving the sensitivity of both reflectivity and ellipso- metry methods but there are some experimental disadvantages. For example the overall reflectivity level decreases for both methods. With ellipsometry (tan $)* becomes very small so that effectively the component in the plane of incidence vanishes and there remains no basis for a method. Unless the collirnation of the beam is exceptionally good the beam size spreads with consequent loss in both intensity and discrimination. Where however the precision of a single reflection P. C. S . HAYFIELD 13 film thickness in nM FIG. 2.-Comparison of theoretically computed reflectivity curves corresponding to filmed platinum.Single reflection A = 546.1 nM = 75". A. n = 1.39; n2 = 2.0; fi3 = 1.85-3.83'; B. nl = 1.39; W 2 = 2.0-03; j i 3 = 1.85-3.83'; k2 = 0.5; H = 1 . 1 5 ~ lo5 cm-'. technique is adequate for detection of adsorbed layers t h s is clearly the preferred technique and many of the modulated reflectivity methods recently described are of this type. Comparative ellipsometric and reflectivity changes resulting from a particular model of a filmed surface are illustrated in fig. 4. More detailed informa- tion is forthcoming from ellipsometric evaluation but there are instrumentation problems in applying the technique over a wide spectral range. This has now largely been overcome and although not a great deal of ellipsometry has been I I I I 5 0 100 150 2 0 0 film thickness in nM FIG. 3.-Comparison of theoretically computed reflectivity curves for a parallel plate system consisting of filmed platinum surfaces.Three major reflections. A. nl = 1.39; n2 = 2.0; fig = 1.85-3.8i; A = 546.1 nM; = 75"; M.R. = 3. B. nl = 1.39; fi2 = 2.0-0.5j; fi3 = 1.85-3.8i; 1 = 546.1 nM; = 75"; M.R. = 3. 14 GENERAL INTRODUCTION carried out for determining the spectra of molecules adsorbed on metal surface,31 work is in progress in several laboratories. In connection with the studies on metal surfaces per se electro-modulated ellipsometry provides a method of establishing dielectric changes without recourse to the Krainers-Kronig relationships which Buchman and Bashara maintain is advantageou~.~~ film thickness in nM phase change A-A. (degrees) R-Ro RO where RO is the reflectivity at E = +.02V R1- R 0 -3 .O x lo2 '*O 0 t + 1.0 - 1.0 -2 .o I 0 t0.2 I .o 2 .o asterisks corres- pond to phase changes reported in fig.4 p. 39. 0-a-.-a /-@-a-e J I I I I I I I I I I 1 compare with fig. 2 p. 73. compare with fig. 2 p. 103. electrode potential FIG. 4.-Theoretically computed optical changes produced by growth of film on a platinum substrate illuminated with light of wavelength 546.1 nM. nl = 1.39; n2 = 2.0; fi3 = 1.85-3.83; = 75"; M.R. = 1. A further feature of ellipsometry which is worth comment concerns the change in sign of A with change in film thickness and with change in optical properties of the substrate. Thus change due to the substrate can be of either sign dependent upon the direction of the perturbing field or force but for growth of film for oblique angle of incidence at least the sign is always negative see fig.5. INTERPRETATION OF OPTICAL DATA Having determined the optical properties of a surface in terms of an ellipsometry or reflectivity response and changes with change in applied potential temperature or whatever the variant that may be employed the next stage is that of establishing the equivalent optical circuit so that correct interpretation may be applied. In comparison with interpretation applied to growth of thick films and about which most experience is to hand some problems diminish and others become more import- P . C. S. HAYFIELD 15 ant. With thick films the exact determination of film refractive index is very necessary to obtain a film thickness of any accuracy. With thin films for example changing the refractive index from 1.2 to 3.4 was shown by Archer not to lead to more than a maximum error of +34 %.Indeed in some studies on adsorbed layers it has been deemed sufficient to calibrate the ellipsometer (and presumably this would also be applicable to reflectivity methods) by measuring change in response with build-up of barium stearate films these being deposited using the Langmuir trough method. 0 9 0 P - O0 FiG. 5.-Diagram to indicate the change in ellipsometry readings with (a) change in substrate optical constants fij3 and (b) growth of film. However in dealing with ellipsometry changes appreciably smaller than lo or the equivalent AR/R changes such as occur in modulated ellipsometry and reflectivity studies it is necessary to consider changes in reflected optical properties arising from other causes namely in the metal and in the double 1ayer.l'.33 At the present time the response from most modulated optical studies have been interpreted either as arising from changes in the optical properties of the substrate or the film and correction factors for the double-layer contribution have not been applied. Methods of distinguishing between the contribution to optical reflectivity changes introduced by metal film and environment are obviously crucial to correct interpreta- tion and design of experiment to clarify particular circumstances is essential. The paper by Bockris Genshaw and Brusic to the present Symposium for example highlights the effects of surface roughness of the metal/film interface and in a similar manner irregularities at the filmlenvironment interface should also be considered.Some success has attended the treatment of the etched surface of iron as a film layer having an effective refractive index based upon combining molecular refractions in 16 GENERAL INTRODUCTION proportion to the volume of metal and second phase in the interphase region,34 see eqn (5). The gross optical changes which it seems may be associated with non- uniform interfaces between phases and which had been predicted by Fenstennaker and McCra~kin,~~ are likely to affect ellipsometric and reflectivity changes alike. The evidence is that extreme caution should be observed over the interpretation of optical data for systems where roughening or smoothing of an interface may occur such as in the study of active/passive transitions for metals exposed to acid condi- t i o n ~ ~ ~ electropolishing reactions and the lj ke.Similar care seems necessary with those modulated reflectivity and ellipsometry studies where electrode potential changes cover a sufficiently wide range for roughening effects to occur such as those reported by Biegler 37 for platinum. Further clarification over the effects of non- uniform interfacial layers is awaited. Notwithstanding the problems associated with interpretation which may affect the application of optical methods in particular circumstances ellipsometric and reflectivity methods of examination are nevertheless established and very sensitive techniques which can be uniquely applied to in situ measurements of adsorbed layers at interfaces. A. C. Hall Recent Developments in Ellipsometry Symp. Proc.(University of Nebraska) N. M. Bashara A. B. Buckman and A. C. Hall ed. (North Holland Pub. Co.) chap. 1. ' P. Drude Ann. Phys. 1887,32,584 ; 1888,34,489 ; 1889,36,532,865 ; 1890,39,481. F. H. Constable Proc. Roy. SOC. A 1927,115 570; 1928,117 376. L. Tronstad Trans. Faruduy SOC. 1933 29 502. L. Tronstad and T. Hoverstad Trans. Faraday SOC. 1934,30,362. L. Tronstad and T. Hoverstad Tram. Faraduy Soc. 1934,30 11 14. C. G. P. Feachem and L. Tronstad Proc. Roy. SOC. A 1934,23 127. C. G. Darwin Trans. Cambr. Phil. Soc. 1924,23 137. C. Strachan Proc. Cambr. Phil. SOC. 1933,29,116. lo R. J. Archer Ellipsometry in the Measurement of Surfaces aizd Thin Films Symp. Proc. (Nat. Bur. Stand. Misc. Pub. 256) E. Passaglia R. R. Stromberg and J. Kruger ed. ( U S Govern- ment Printing Office Washington D.C.1964) p. 255. l 1 A. C. Hall J. Phys. Chem. 1966,70 1702. '' T. Smith J. Opt. SOC. Amer. 1968 58 1069. l3 A. J. Melmed H. P. Layer and J. Kruger Surface Sci. 1968,9,476. l4 J. J. Carroll and A. J. Melmed Surface Sci. 1969 16,251. l5 J. R. Miller and J. E. Berger J. Phys. Chem. 1966 70 3070. l6 F. Meyer and G. A. Bootsma Surface Sci. 1969 16 221. l7 this Symposium. l 8 R. H. Muller R. E. Steiger G. A. Somorjai and J. M. Morabito Surface Sci. 1969,16,234. l9 H. B. Mark and B. S. Pons Anal. Chem. 1966,38 119. 2o W. N. Hansen T. Kuwana and R. A. Osteryoung Anal. Chern. 1966,38,1811. 21 G. W. Poling J. Electrochem. SOC. 1967 114 1209. 22 G. W. Poling J. Electrochem. SOC. 1969,116 958. 23R. G. Greenler J. Chem. Phys. 1969,50,1963. 24 G. W. Poling Corrosion Sci. 1970,10 359. 25 N.J. Harrick Internul Reflection Spectroscopy (Interscience Publishers 1967). 26 P. T. Kissinger and C. N. Reilley Anal. Chem. 1970,42 12. 27 D. F. A. Koch Nature. 202 3871. 28 D. F. A. Koch and D. E. Wife J. Electrochem. SOC. 1966,113 1302. 29 B. 0. Seraphin Phys. Reu. A 1965,140,1716. 30 M. Cardoma K. L. Shaklee and F. H. Pollak Phys. Rev. 1967,154,696. 31 C. L. McBee and J. Kruger ref. (l) p. 340. 32 A. B. Buckman and N. M. Bashara J. Opt. SOC. Amer. 1968,58,700. 33 M. Stedman Chern. Phys. Letters 1968,2,457. 34 C . J. F. Bottcher Theory of Electric Polarization (Elsevier Publishing Co. Amsterdam 1952). 35 C. A. Fenstermaker and F. L. McCrackin ref. (lo) p. 85. 36 J. O'M. Bockris A. K. N. Reddy and B. Rao J. Electrochem. SOC. 1966,113 1133. 37 T. Biegler J. Electrochem. Soc. 1969 116 1131.