Reflectance Studies of the Gold/Electrolyte Interface BY B. D. C m N JEAN HORKANS AND ERNEST YEAGER Chemistry Dept. Case Western Reserve University Cleveland Ohio 44106 U.S.A. Received 4th September 1970 The gold/electrolyte interface has been examined with both single and multiple reflection tech- niques. Gross errors arising from scattering have been identified with the latter using a large number of reflections. An explanation involving transitions of electrons in surface states (5d-+6s) is proposed to account for the sensitivity of the reflectivity of gold electrodes to potential and to adsorbed ionic and neutral species. Previously proposed explanations involving changes in the Fermi level relative to non-surface electronic states appear untenable. A method is also proposed for examining the refractive index within the double layer if it is lower than that of the bulk solution.A number of papers have recently appeared dealing with the phenomenon of the electro-modulation of light at a reflecting electrode/solution interface. These effects have been detected with single reflection using a phase-lock amplifier,'. ellipso- m e t ~ y ~ ' ~ multiple reflection,6 and with attenuated total reflection.'" The potential dependence of the reflectivity was originally interpreted as a change in the optical constants of the double layer,2 but it was later shown by Hansen and Prostak 7-9 that the changes in the double layer should be too small to cause the observed effect. They re-interpreted the data in terms of a change in the electron concentration in the surface layers of the electrode.Their calculations explain the occurrence of a peak in the AR/R curve for gold at the wavelength of the absorption edge of gold associated with the 5d+6s interband transition. This peak is predicted to be much narrower than that obtained experimentally. It is still not fully clear that the pre- dominant optical effect is attributable to changes in the metal rather than in the double layer. Calculations by Stedman O predict some dependence of reflectivity of the electrode on the changes of the optical constants of the double layer although this effect is smaller than that predicted on the basis of the Hansen-Prostak calculation. Takamura et aL6 reported a large increase in sensitivity by using multiple (- 19) reflections. Their (A,R/R,E) curves were highly sensitive to monolayer and sub- monolayer quantities of oxides adsorbed anions and foreign metal adatoms.Ellipsometry also provides adequate sensitivity for the detection of submonolayer quantities of adsorbates. followed anion adsorption on Hg and Au. Buckman studied the electro-modulation of silver in KCl electrolyte. The purpose of the present paper is to elucidate the mechanism responsible for electro-modulation of the reflectivity. Sirohi and Genshaw 4* EXPERIMENTAL The optical system consisted of a conventional prism monochromator with a tungsten light source and a mirror-cell arrangement for single or multiple reflection measurements. The use of a highly stable light source and photodetector system has afforded sufficient stability to permit the determination of the potential dependence with a single reflection after suitable amplification and without a.c.or square-wave modulation techniques. The 36 B . D. CAHAN J . HORKANS A N D E . YEAGER 37 signal-to-noise and hence the sensitivity to changes in reflectivity with potential were superior to that reported for multiple reflection.6 The electrodes used in these studies were evaporated gold films 5 O00-10 000 A thick. These films were deposited on substrates of flame-polished glass with layers of sputtered tantalum followed by sputtered platinum between the gold and the glass to yield good adhesion.'l The gold film electrodes gave voltammetric and reflectivity data similar to that on mechanically polished gold plates with the added advantage that the light scattered from the surface was very small compared to that obtainable from a polished bulk specimen.Most of the measurements were done in 1 N HC104 to minimize specific adsorption of the anion but measurements were also made in H2S04 and NaC104. The effects of small additions of non-polar materials (e.g. benzene) and adsorbable anions (e.g. the halides) were also studied. RESULTS (Reflectivity potential) curves obtained in the present work with single reflection and with seven reflections are generically similar (fig. 1) to those obtained by Takamura et aL6 with a reported 19 reflections. The magnitude of the reflectivity change per reflection however is larger by a factor of three in the present work. The slopes l/R(aR/aE) of these curves at three potentials in the anodic branch of the sweep have been plotted against wavelength in fig.2 for a single reflection. The analogous curves obtained for seven reflections have the same basic shape and are of the same magnitude after division by a scale factor of seven. Curves obtained in H2S04 showed similar behaviour. There is a shift of the position of the maximum by about 35 mp between 0.2 V and 1.0 V. The cwves also broaden at higher poten- tials. The maxima of the curves occur in a wavelength region similar to that observed by Fejnlieb.2 This is in contrast to the curves obtained by Takamura et aL6 (see inset fig. 2) which exhibit a sharp drop-off in the blue and a maximum displaced to ward longer wavelengths. - 0 0.4 0.8 1:2 1.6 0 0'4 0-8 1-2 1.6 potential against NHE (V) FIG. 1.-Relative reflectivity curve for an evaporated gold film electrode in 1 N HC104 at 540 mp normalized to a reflectivity af 1.0 at 0.0 V.The inset is a similar curve from ref. (6) for multiple reflection at a bulk gold electrode in 0.2 N HC104 at the same wavelength. Relative reflectivity curves (normalized to unit reflectivity at 675 mp) have been obtained for the evaporated gold film electrodes potentiostatted in HC104. These data have been processed digitally and fitted to a ninth-order polynomial. A typical curve and the resultant derivative curve (lIR)(iJR/i?il) against il are shown in fig. 3. 38 REFLECTANCE STUDIES OF GOLD/ELECTROLYTE INTERFACE 16 r 01 I I I I I I I I 400 500 600 700 1 (mt.) FIG. 2.-1/R(3R/3E)~ as a function of wavelength - is at 0.2V -- at 0.6V and - - -at 1.0 V. Slopes are taken from the anodic sweep of curves similar to fig. 1.The insert is the data obtained in ref. (6) using multiple reflection A is at 0.52 V ; B is at 1.03 V. 0'4 0-2 0 mt.) FIG. 3.-The reflectivity (normalized to R = 1.0 at 675 mp) and its derivative 1/R ( ~ R / ~ X ) E for a gold film electrode potentiostatted at 0.0V in 1 N HC104. B . D . CAHAN J . HORKANS AND E. YEAGER Fig. 4 shows the effect of adding 5 x 39 All of the curves described earlier show marked sensitivity to even minute quantities of additives or impurities. M benzene to the solution. At this concentration the benzene should have a surface coverage of only 15 % of a monolayer at the maximum of adsorption (0.5-0.6 V) against NHE),12 yet its effect on the reflectivity is clearly observable. [The noise level was sufficiently low to permit curves with considerable additional amplification to be obtained but they were not included in the present paper because of the difficulty of showing the entire curves within practical space limitations.] Although the curves show unmistakably the optical effects of benzene adsorption the data in its present form are not suitable for the determination of the adsorption isotherm for benzene because of the wide range of potentials used in the scan (> 1.6 V) and the possibility of oxidation products.Further work is planned to study more closely the effects of neutral adsorbed species on the reflectivity. I I I I I I t 1 0 0.4 0-8 I. 2 1.6 potential against NHE (V) FIG. 4.4hange in the relative reflectivity curve of gold (7 reflections) in 1 N HClO at 500 mp upon the addition of 5 x M benzene - in absence of benzene ; -- in presence of benzene.The lower curve has been shifted vertically for purposes of display. DISCUSSION COMPARISONS OF SINGLE AND MULTIPLE REFLECTION TECHNIQUES For a given polarization state of the incident light Zo the intensity Zn after n reflections can be expressed as where R is the reflectivity of the interface. Differentiating and dividing by In gives The implication is that the shape and magnitude of (n/R)(dR/aE)A against R should be independent of the number of reflections. The curves obtained in this study using single reflection and seven reflections exhibit this property within experimental tolerances. The earlier multiple reflection work,6 however deviates significantly (see fig. 2). A consideration of the cumulative effects of 19 reflections shows that at a wavelength where the reflectivity of gold is 0.5 only of the incident light is available for detection.Any stray or scattered light (including extraneous wave- lengths produced by scatter in the monochromator itself) would thus cause a back- ground intensity orders of magnitude greater than the desired signal such that In Zn = RnIO (1) (2) (1 1RXaRlaE)A = (1 /nrn)(aLlaE),i* 40 in eqn (2) is no longer the desired value. The mechanically polished surfaces used in the earlier studies should scatter light down the multiple reflection path giving emergent light equivalent to far fewer reflections than the number calculated on the basis of geometry. Indeed a comparison of the n values obtained in the red (where the reflectivity is relatively high) indicates an effective number of reflections between seven and nine.13 Therefore the displacement toward the red of the peak in the [(l/R)(8R/i?E)4 curve in their work is probably an artifact caused by a burying of the true signal I in the scattered light and the consequent misapplication of the l/n factor in eqn (2).Although quantitatively in error most of the qualitative observations of their work are still valid. REFLECTANCE STUDIES OF GOLD/ELECTROLYTE INTERFACE THE PROBLEM OF THE EDGE SHIFT The theory discussed by Hansen 7-9 ascribes the apparent changes in the optical constants of gold upon electro-modulation to a change in the concentration of free electrons in the surface region of the bulk gold. An increase in the number of free electrons n in the metal will then cause a shift in the plasma frequency of the metal cop = (4me2/rn)+ (3) and the optical properties of the metal will be modulated.Hansen and Prostak * state that the Fermi level will be modulated but that the energy of the bound electrons will be essentially unaffected (Le. the top of the 5d valence band). On this basis a shift of the absorption edge associated with transitions from the 5d to the Fermi level in the middle of the 6s conduction band is to be expected. The idea that the concentration of the electrons within the bulk of a metal film can be varied by an appreciable amount electrochemically is difficult to accept since it violates the principle of electroneutrality within a conductive medium. The excess electrons accumulated within the metal during electrochemical charging of the double layer are statistically localized at the surface by electrostatic interaction with the corresponding ion and dipole charge of opposite charge in the double layer.Stoner et ~11.'~ showed that the resistance of a thin platinum film used as an electrode was unchanged over the entire double layer region of potentials in the presence of a non-adsorbing electrolyte (i.e. HC10J. The concept of a shift in the Fermi level relative to the top of the 5d band in the bulk is contrary to accepted solid-state theory. The energy difference between the Fermi level and the top of the surface electron states however is expected to vary with electrochemical potential. In accounting for their ATR studies of gold film electrodes of - 50 a thickness Hansen and Prostak * assume the change in free electron concentration causes the optical properties of the metal including the optical absorption edge (5d-+6s transition) of gold to shift by 0.0076 eV for a potential sweep of 0.70 V.This shift is estimated on the basis that the double-layer charging leads to a change in the number of free electrons in the 50A gold film and that the (optical constant frequency) curve is then shifted in frequency by the change in cop as predicted by eqn (3). If the change in number of electrons is restricted to 10 % of the film then An/n and hence the frequency shift is 10 fold greater (i.e. 0.076 ev). Hansen and Prostak * found the calculated change in reflectivity with potential to be independent of whether the change in free electrons is in the entire film or only 10 % of it and thus that the reflectivity change is not sensitive to the fraction of the film in which the change in free electron concentration occurs.A similar calculation * performed by the authors shows that this is indeed the * A programme was written in Fortran using the general Fresnel-Drude equations for oblique incidence using a three-layer model. B . D. CAHAN J . HORKANS AND E . YEAGER 41 case for a small A (eV) shift for specular reflection and that this shift of the surface optical constants should lead to an effective wavelength shift of the @,A) curve with potential. The magnitude of this shift is estimated to be 2-3 mp/V at the wavelength of the edge and the shift should be constant (in terms of eV) with wavelength. For a simple linear shift the one point of the curve whose position should be invariant with scale factor is the inflection point.Therefore the fitted polynomial for the single reflection data was analyzed for the location of this point. Computer analysis of these data however showed a negligible simple monotonic trend of the position of the inflection point as the potential was changed. A linear regression analysis of the position of the infection points against potential showed a shift of only 0.12 mp/V with a standard deviation of 0.4mp/V compared to the expected 2-3 mp/V shift. The derivatives (l/R)(dl?/~?A)~ (fig. 3) were calculated from the shape of the (reflectivity wavelength) curves and (1 /R)(dR/dE)A was determined from the (reflectivity potential) curves (fig. 2). The shift of the edge in terms of the energy U (in eV) with respect to potential is then and should according to the edge-shift theory be a constant.From the data in table 1 (aU/aE) is far from constant. This is not unexpected since the peak calculated by Hansen and Prostak * is much narrower than that found experimentally by them by Feinleib,2 or ourselves. While the edge-shift theory can befitted to account for the potential dependence of the peak in the reflectivity at the absorption edge it does not explain the appreciable modulation in either the red or the blue. (3 u/aE)R = - (dU/dA)[(l /R)(3R/aE)A/(l /R)(aR/aA)El (4) TABLE 1 .-WAVELENGTH DEPENDENCE AND POTENTIAL DEPENDENCE OF (8 U / ~ E ) R -(a UIaE)R (eV/V> ( x 103) wavelength A (mi4 0.2 v 0.6 V 1.0 v 450 150 118 98 475 11.4 8.8 7.0 500 4.2 3.4 2.6 525 4.3 4.4 3.5 550 6.1 6.6 6.5 575 8.8 9.6 10.7 600 10.8 13.8 15.4 In view of the lack of a simple shift of the edge itself the displacement of the maximum of (1 /R)(aR/dE) curves (fig.2) with increasing voltage cannot be attributed to a simple linear shift but can be interpreted in terms of a gross distortion of the optical constants in a surface layer of the metal. SURFACE STATE THEORY At a temperature of absolute zero for a perfect crystal if the primary mechanism of light absorption in gold is the 5&6s transition there should be an abrupt step in the (reflectivity photon energy) curve. At higher temperatures the step should become rounded as shown by the Fermi distribution function. This rounding should be symmetric around the inflection point whereas the actual reflectivity curve still shows significant excess absorption,* more than 0.5 eV below the actual absorption * The term excess absorption is here used to denote absorption in the red at wavelengths longer than the absorption edge over and above that expected on the basis of kT (thermal) broadening.42 REFLECTANCE STUDIES OF GOLD/ELECTROLYTE INTERFACE edge while only 2-3 kT (i.e. less than 0.075 eV) is attributable to thermal fluctuations in the electron distribution. Theoretical considerations of the electronic surface states of finite lattices show that the effect of truncation of an infinite lattice generates highly localized surface eigenstates with energies in the band gap i.e. for gold the top of the 5d band is raised and the bottom of the 6s band is lowered at the surface. Fig. 5a is a possible diagrammatic representation of the 5d and 6s bands at a metal-vacuum interface.The cross-hatched areas represent the states induced by introducing a boundary in the crystal. The existence of these states in the forbidden band may account for the shape of the reflectivity curve of gold. The existence of these occupied states makes the (5d)surface +Fermi level transitions possible with photon energies lower than the absorption edge. Conversely it may be possible to calculate the shape of the (5d)surface band from the excess absorption in the red. METAL- VACUUM METAL- ELECTROLYTE FIG. 5.-Schematic representation of energy bands and surface states (cross-hatched area) in gold a metal-vacuum interface ; b metal-solution interface. Fig. 5b illustrates the surface immersed in an electrolyte ; the 5d and 6s bands are distorted because of the potential across the interface.Consequently the (energy distance) relationship of the surface bands is also changed and thus the shape of the excess absorption spectrum can be shifted markedly. If the hypothesis that the round- ing in the red of the reflectivity curve is caused by the excess absorption due to the surface states is correct then we have available sufficiently large energy changes to account for electro-modulation over a wide wavelength range. The modulation of the double layer thus affects a very small region in the metal. Very large effects can be produced by the relatively small charge (33 pC/cm2 for the 0.7 V in the work of Hansen and Prostak 8 associated with the modulation of this double layer. The small shifts in photon energy (0.0076 to 0.076 eV) used by Hansen and Prostaks in their calculations of AR/R against A can be considered as approximating differentials and their resultant curve is equivalent to the derivative of the (reflectivity A) curve.B . D . CAHAN J . HORKANS AND E. YEAGER 43 The half-width of the peak in the (ARIR A) curve derived by Hansen and Prostak is invariant with energy shift as long as the shifts are sufficiently small. Shifts of a size that can no longer be treated as linear do in fact produce broadening and a shift of the peak. With surface states of the type represented in fig. 5 even a few tenths of a volt should be sufficient to produce such non-linear effects. Surface hetero- geneity would also contribute to broadening. EFFECTS I N SOLUTION PHASE The question remains whether there are any appreciable electro-modulation effects specifically due to changes in the refractive index and/or thickness of the double layer.Assuming that these effects do exist it is interesting to consider the best method of observing them independent of the effects within the metal. If we assume that the index of refraction of the double layer is lower than that of the solution for perpendicular polarization there exists a particular angle at high angles of incidence at which the phase difference between the light reflected from the solution/double layer and the double layer/metal interfaces approaches A/2. The resultant interference causes a sharp dip in reflectivity. The assumption of a lower index of refraction in the double layer is not unreasonable at least under some cir- cumstances ; such a layer could exist because of the formation of an ice-like structure the repulsion of anions at negative potentials leaving only surface oriented H30+ ions or the adsorption of a low index of refraction organic material.A three-layer calculation was programmed in Fortran to compute the reflectivity for high angles of incidence varying the index of refraction of the double layer. At each angle of incidence the predicted dip in reflectivity is associated with a unique index of refraction. As the index of refraction approaches that of the bulk solution the required angle of incidence approaches 90". For perpendicular polarization the change of reflectivity can be greater than 20 % for a 0.05 change in index of refrac- tion. The following experimental set-up is being implemented to study this effect.A flat electrode is to be mounted at the centre of rotation of a scanning 8-20 gonio- meter. A small modulation of the fixed average electrode potential will be impressed potentiostatically and the goniometer scanned through small angles. The modula- tion will be detected and divided by the average reflectivity. Any electro-modulation of the index of refraction will show up as a blip at the corresponding angle. From this angle the index of refraction of the double layer can be calculated and hence its dielectric constant. In summary an explanation is offered for the sensitivity of the reflectivity of gold to potential which involves surface electronic states. Specular reflection on gold electrodes is very sensitive to adsorbed species including neutral species and provides a means for studying adsorption phenomena.Further multiple reflection studies can be subject to gross errors particularly associated with scattering phenomena ; this indicates the desirability of performing future work with single or at most a few reflections. Finally a new method has been proposed for measuring the refractive index of the double layer in instances where it is lower than that of the bulk solution. The authors acknowledge the support of this research by the U.S. Office of Naval Research and also the Electrochemical Society through the Edward Weston Fellow- ship to one of the authors (J. H.). J. D. McIntyre 135th National Meeting Electrochem. Sac (New York. May 1969) (extended abstr. p. 578). J. Feinlieb Phys. Reu. Letters 1966 16 1200. A. B. Buckman Surface Sci. 1969 16 193. 44 REFLECTANCE STUDIES OF GOLD/ELECTROLYTE INTERFACE R. S. Sirohi and M A. Genshaw J. Electrochem. SOC. 1969,116,910. M. A. Genshaw private communication. T. Takamura K. Takamura W. Nippe and E. Yeager J. Electrochem. SOC. 1970,117,626. A. Prostak and W. N. Hansen Phys. Rev. 1967,160,600. * W. N. Hansen and A. Prostak Phys. Rev. 1968,174 500. W. N. Hansen Surface Sci. 1969 16,205. lo M. Stedman Chem. Phys. Letters 1968 2,457. 11 B. D. Cahan Ph.D. Thesis (University of Pennsylvania 1968). l2 H. Green and M. Dahms J. Electrochem Soc. 1963,110,1075. l3 T. Takamura K. Takamura and E Yeager unpublished data. l4 J. O'M. Bockris B. D. Cahan and G. E. Stoner Chem. Instr. 1969,1,273. l6 P. Mark in Clean Surfaces G. Goldfinger ed. (Marcel Dekker N.Y. 1970) p. 307. R. J. Archer Ellipsometry (Gaertner Scientific Chicago 1968) p. 11.