J . Chem. SOC., Faraday Trans. I , 1982, 78, 3433-3445 Photoelectrochemical Study of the Amorphous-W0,-semiconductor-Electrolyte Junction BY FRANCESCO DI QUARTO,* GIUSEPPE Russo, CARMELO SUNSERI AND AGATINO DI PAOLA Istituto di Ingegneria Chimica, Universita di Palermo, Wale delle Scienze, 90 134 Palermo, Italy Received 2nd November, 198 1 The photoelectrochemical behaviour of amorphous anodic films grown on tungsten has been studied. The wavelength of the incident light is shown to influence the photoresponse of the amorphous films. The experimental results are interpreted on the basis of the semiconducting properties of the film and by taking into account the various mechanisms of transport occurring in amorphous materials. At longer wavelengths a Poole-Frenkel mechanism of electrical conduction in the non-extended states of the amorphous semiconductor is invoked in explaining the transport of photoinjected carriers.At the shortest wavelengths a 'free-carrier-like' mechanism of transport of the photogenerated carriers is suggested. In each case a different electrode-potential dependence of the photocurrent is obtained experimentally. We have previously made a preliminary study of the photoelectrochemical behaviour of amorphous WO, anodic films.' Its main aim was to improve our knowledge of the physico-chemical properties of the anodic films grown on tungsten by anodic polarization in different acid solutions. The experimental results could be explained reasonably well using the existing theory of the electrochemical behaviour of single-crystal semiconducting electrodes.Some discrepancies in the flat-band potential values obtained using different experimental methods remained unresolved because of various experimental inaccuracies, such as the use of polychromatic light, as well as through the lack of an adequate model of the amorphous-semiconductor-electrolyte interface. On the other hand, the solid-state properties of films grown on an electrode surface are ultimately the factors which control the electrochemical behaviour of electrodes during their use in several practical applications (electrocatalysis, corrosion, solar cells). Within this framework a more detailed study of the photoelectrochemical behaviour of the amorphous semiconducting films grown on tungsten by anodic polarization has been undertaken, in order to obtain more information on the influence of the amorphous nature of these films on their photoelectrochemical behaviour.We propose that different mechanisms of transport of the photogenerated carriers can operate in an amorphous electrode, depending on the energy of the incident photons, and thus different relationships between measured photocurrent and electrode potential have been obtained for the amorphous-semiconductor-electrolyte interface. EXPERIMENTAL Spectrographically pure tungsten foils or rods were anodized in 0.1 normal H,PO, solutions at a constant current density of 8 mA cm-2, until various final voltages were reached. Before anodization the electrode surface was prepared as previously described.'? The corresponding 34333434 PHOTOELECTROCHEMISTRY OF AMORPHOUS wo, ELECTRODES thicknesses of the anodic films were estimated by assuming an anodizing ratio of 17.0 8, V-1.3 After anodization the electrodes were inserted into a quartz electrochemical cell, where both photoelectrochemical and differential capacitance measurements were performed. The differential capacitance of the semiconducting films was measured by a lock-in technique by using a P.A.R.124 A/ 1 16 lock-in amplifier in connection with a P.A.R. 173/ 179 potentiostat equipped with a P.A.R. 175 universal programmer. The modulating a.c. voltage was a 10 mV peak-to-peak sinewave at 160 Hz, and the scanning rate was 10 mV s-l. The same apparatus was used in the linear-potential-sweep experiments. Monochromatic light was obtained by filtering light from a Bausch-Lomb 150 W Xenon lamp using a hgh-intensity Bausch-Lomb no.5 u.v.-visible monochromator equipped with focusing lenses. The exit slit-width was 3 mm, but a variable entrance slit-width was used in order to change the light intensity. A P.A.R. 125A mechanical light chopper was used in the chopped-light experiments. All capacitance and photoelectrochemical measurements were performed in 0.5 mol dm-3 H,SO, solution; the counter-electrode was a 10 cm2 Pt foil and the reference electrode was Hg I Hg2S0, 10.5 mol dmP3 H2S0, (mercurous sulphate electrode, MSE). Solutions were pre- pared from distilled water and analytical-grade reagents. All the experiments were performed at room temperature (25 & 1 "C). RESULTS The photoelectrochemical behaviour of the amorphous WO, electrodes was studied by investigating in linear-potential-sweep experiments the photoresponse of the oxidized electrodes illuminated with light of different wavelengths.The use of focusing lenses as well as the very low dark current (idark < 0.1 pA cmd2) under anodic polarization allowed steady-state photocurrent measurements without chopping the light. Because of the slow dissolution of the films, after each experiment with illumination the values of the dark current were verified. No significative change in these values were measured, provided that a relatively thick film of oxide (ca. 500 %.) was still on the electrode surface. The good reproducibility of the films as well as of the experimental data for the same experiments performed with different films was carefully checked.Unless explicitly stated, experiments reported in the various figures were performed with different films. It is significant that no influence on the shape of the photocurrent against potential curves was seen if films grown to different thicknesses were employed in the same experiments. Thus most of the investigations were performed with films grown at two different voltages, 70 and 100 V, in order to avoid experimental complications. All potentiodynamic experiments were performed by starting from + 3.0 V us. MSE and decreasing the electrode potential until the dark-current value was attained. For the various samples the value of the onset photocurrent potential was Uon = - 0.1 & 0.05 V.The same value was measured by performing the experiments with chopped light (chopping frequency 89.5 Hz) at a very low light intensity and by measuring the photocurrent using the lock-in technique. In the latter experiments the same functional dependence in the Iph against UE curves was seen. Slight hysteresis was observed in the reverse scan, especially in the experiments performed at 230 nm, where saturation effects were recorded at higher potentials ( U , 2 2 V). Following this finding and in order to reduce the dissolution of the films under illumination, a standard procedure was followed in carrying out the potentiodynamic experiments. Before each experiment under illumination the electrodes were kept at 3 V for 10 min in the dark and then swept under light to -0.2 V.In this way, for the thicker electrodes anodized to 100 V it was possible to perform five different sweeps without any appreciable change in the dark-current values.F. D I QUARTO, G. RUSSO, C. SUNSERI AND A. DI PAOLA 3435 In fig. 1 two curves are reported for different wavelengths, 340 and 230 nm, respectively. The curves, obtained with two different electrodes, show the general behaviour of the photocurrent against electrode-potential curves at the shortest and longest wavelengths. The shapes of the curves were not influenced by changing the intensity of the incident light (see also fig. 4). The influence of film thickness on the values of the photocurrent will be discussed in detail later. From the two curves, different relationships between photocurrent and electrode potential were obtained. In particular, from 380 to 270 nm a Iph against u& law was observed, whilst at 230 nm a square-root dependence was found.I ’ 1 I I 1 /I 5t Y’/l I I I I 0.0 0.5 i.0 1.5 2.0 U,lV us. MSE FIG. 1.-Photoresponse of amorphous WO, anodic films at two different wavelengths in 0.5 mol dm-3 H,SO,. Initial film thickness, 1700 A; scan rate, 10 mV s-l. Solid curve: 340 nm (electrode area, 0.13 cm2); hatched curve: 230 nm (electrode area, 0.16 cm2). In fig. 2 are reported I$h against UE: plots for a film anodized to 100 V and swept potentiodynamically at three different wavelengths, starting with A = 270 nm. A straight line was observed for the I$! against UE plots. Furthermore, the electrode potential at zero photocurrent was in fair agreement with the onset photocurrent potential and the previously reported flat-band potential obtained for Mott-Schottky plots.1 In fig. 3 we report & against UE plots for three consecutive sweeps at 300 nm. The thicknesses of the films were different because of the continuous dissolution process, which occurred evenly provided that the illumination of the electrode surface was uniform. Although a different slope is obtained for each sweep, all the straight lines have a common intersection voltage. The different slopes of the plots seem to be related to the change in the distribution of donors occurring in the films during the experiment, rather than to the thinning of the oxide films by dissolution. This aspect will be discussed below on the basis of the proposed model.In fig. 4 and 5 are reported IEh against UE plots at II = 230 nm. As shown for the experiments performed at 300 nm, both the light intensity and the film thickness influence only the slope of the lines. All the previous results did not change appreciably if the experiments were performed by starting with electrodes anodized to 70 V, or if different current densities or solutions were employed during the anodic formation of the films. Measurements of differential capacitance in the dark or under illumination were performed in order to gain a better knowledge of the changes which occurred in the films during both the dissolution process and illumination.7 I I 0 1 2 3 FIG. 2 UElV VS. MSE I I UE/V VS. MSE FIG. 3 FIG. 2.-Plots of I$, against U$: obtained at different wavelengths.Initial film thickness, 1700 A; scan rate, 10 mV s-l; electrode area, 0.16 cm2. Slit width: entrance 6 mm, exit 3 mm. Wavelengths: 0, 270 nm (1st sweep); 0, 340 nm (2nd sweep); A, 380 nm (3rd sweep). FIG. 3.-Influence of dissolution process on plots of & against U , at constant wavelength and light intensity. 2 = 300BAA nm; initial film thickness, 1700 A; scan rate, 10 mV s-l; electrode area, 0.16 cm2. Sweeps: 0, 1st; n, 2nd; A, 3rd.I ' I I 40 30 c1 \ 9 @ 20 10 0 UEIV US. MSE FIG. 4 UE{V us. MSE FIG. 5 FIG. 4.-Influence of the light intensity on plots I& against U,. 1 = 230 nm; initial film thickness, 1700 A; scan rate, 10 mV s-'; electrode area, 0.16 cm2. Slit width: 0, entrance 6 mm, exit 3 mm; n, entrance 2 mm, exit 3 mm.Both expenments are relative to the first sweep on two virgin samples. FIG. 5.-Influence of the dissolution process on plots of Gh against U , at constant wavelength and light intensity. 1 = 230 nm; initial film thickness, 1700 A; scan rate, 10 mV s-l; electrode area, 0.16 cm2. Sweeps: 0, 1st; R, 3rd; A, 5th.3438 PHOTOELECTROCHEMISTRY OF AMORPHOUS WO, ELECTRODES Fig. 6 shows the Mott-Schottky plots obtained from the differential capacitance measurements performed with the electrode used in the experiments of fig. 3, before and after the sweeps under illumination. There is a change in the semiconducting properties of the electrode which leads to an increase in the density of donors and a change in their spatial distribution.These results are in agreement with the ageing effects reported by Butler4 for TiO, electrodes. Fig. 7 and 8 show the influence of illumination and the dissolution process on both the capacitance measurements and the Mott-Schottky plots, for two electrodes anodized at 70 and 100 V, respectively. Although some doubts have been raised very recently as to the application of Mott-Schottky theory to amorphous semiconductor^,^ both these figures show unequivocally that under illumination the semiconductor electrodes exhibit behaviour which is in accordance with the existence of a depletion layer much thinner than the thickness of the anodic films. Moreover these figures agree in showing that during an experiment the distribution of donors in the films is changing as a result of both the dissolution process and ageing under electrical polarization.By comparing fig. 6 and 8 it is possible to see that under illumination the space-charge region of the electrode decreases substantially with respect to the dark conditions. Analogous effects on amorphous silicon have been reported by Wronsky.6 By assuming a direct proportionality between the absorption coefficient a and the photocurrent, we have plotted in fig. 9 the (Iph hv)i values as a function of the photon energy hv, in order to obtain a measure of the optical band-gap E g p t of the amorphous WO, films. The measured Egpt values for the various films usually ranged between 3.0 and 3.1 eV, and this scattering of values does not seem to be related to the thickness of the film or to anodizing parameters, such as current density or nature of the anion, provided that the anodization process is stopped before the onset of film breakdown.For amorphous semiconductors7 other relationships between the absorption coefficient and the energy of the incident photons have been proposed. In our case, however, the best fitting of the experimental data collected for several specimens of different thicknesses and at various light intensities was obtained by using the square-root interpolation. Although the same square-root dependence has been reported for both a m o r p h o u ~ ~ - ~ and crystalline materials having indirect optical transitions, the nature of the transitions is different in the two cases. In fact, in the case of amorphous materials no intervention of phonons is required in order that the optically induced transitions take place, whilst phonon-assisted transitions are involved in crystalline materials.For this reason transitions in amorphous materials have been defined as n~n-direct.~ With these differences in mind, fig. 9 shows a plot relative to a thermally crystallized anodic film. The crystallization process was performed under an argon atmosphere at 350 OC. The details of this process and the properties of the crystallized electrodes will be discussed in a separate paper. Owing to the crystallization, a decrease in the optical band-gap of ca. 0.3 eV is observed. The good agreement between our Egpt values and those reported in the 1iterature1O-l2 for both polycrystalline and amorphous WO, oxides is evidence of the direct propor- tionality which exists at each wavelength between the measured photocurrent and the absorption coefficient.This aspect will be discussed further below. DISCUSSION The experimental results previously outlined cannot be explained, in our opin on the basis of Gartner’s modeP3 formerly employed by various authors crystalline-semiconductor-electrolyte junctions under illurnination.l2-l8 In fact, on, for theU,lV us. MSE FIG. 6 0 1 2 3 U,/V US. MSE FIG. 7 FIG. 6.-Plots of l/Cz against UE obtained in the dark. The experiments were performed on the same electrode as fig. 3, before (0) and after (A) the three consecutive sweeps under illumination. Scan rate, 10 mV s-l; a.c. frequency, 160 Hz. FIG. 7.-Influence of the dissolution process on plots of differential capacitance against UE at constant wavelength and light intensity.,I = 300 nm; initial film thickness, 1260 A; scan rate, 10 mV s-l; electrode area, 0.16 cm2; ax. frequency, 160 Hz. Sweeps: 0, 1st; 0, 3rd.3440 PHOTOELECTROCHEMISTRY OF AMORPHOUS wo, ELECTRODES UE/V us. MSE FIG. 8.-Influence of the dissolution process on plots of l/cZ against UE obtained under constant illumination at /1 = 300 nm. Initial film thickness, 1700 A; scan rate, 10 mV s-l; a.c. frequency, 160 Hz. Sweeps: 0, 1st; A, 3rd. sub-linear Iph against UB law in the range 270-380 nm, coupled with the change to a square-root dependence at shorter wavelengths (A = 230 nm), cannot be justified by the above models. This disagreement is not surprising if we take into account that transport of the photogenerated carriers can be noticeably different in amorphous materials when compared with the crystalline case.In the latter, the photoinjected carriers can move under the action of an electric field in the delocalized states of the valence or conduction band of the semiconductor, regardless of the existence of possible mechanisms of recombination in the various regions of the semiconductor (bulk, depletion layer, surface). For amorphous semiconductors different mechanisms of transport have been invoked depending on whether the carriers move in the extended-state or in the non-extended-state 19-25 In the first case an activationless ' free-carrier-like ' transport mechanism will be operating; in the second case a field-assisted hopping mechanism with a related activation energy has been suggested.The most usual hopping mechanism proposed by various authors in explaining the electrical behaviour of amorphous semiconductors in the dark is a generalized Poole-Frenkel me~hanism.l~-*~ Moreover, a dependence of the photocurrent on the electric field following the Poole-Frenkel mechanism has been reported by different authors for344 1 a number of amorphous films,26-28 and very recently also for single-crystal semi- conductors illuminated with photons of sub-band-gap energy.16* 29 According to Hill’s theoretical analysis of the Poole-Frenkel mechanism of electrical conduction in amorphous materials, the following low-field expressions, depending on the assump- tions made, can be obtained for the dark current: iPpF = A F ~ (1) iPpF = BF (2) iPwF = CF: (3) where P is the electric field inside the electrode, and A , B and C can be considered as constants at a fixed temperature and for a given material.According to Hill, eqn (1)-(3) can be obtained as a result of the substitution of sinh /?Fi by exp /?FB in the expression for the probability of the carriers being emitted from the coulombic trapping centres.20 When this substitution is performed, a field dependence for crystalline materials analogous to that reported in eqn (3) for semicrystalline materials is obtained. On the basis of the previous considerations an interpretative model of the experimental results can be proposed according to the following assumptions. (i) The existence under constant illumination of a steady-state distribution of the photoexcited carriers among the localized states of the conduction (electrons) or valence (holes) band.These states will be assumed to have, respectively, donor-like or acceptor-like characteristics so that a field-lowering of the activation energy barriers, in accordance with the Poole-Frenkel effect, takes place under electric polarization.201 28 (ii) The absence of any contributions to the photocurrent from the carriers generated outside3442 PHOTOELECTROCHEMISTRY OF AMORPHOUS wo, ELECTRODES the space-charge region. (iii) The existence under illumination of a depletion layer whose thickness is less than the film thickness over the whole range of electrode potentials used. The first assumption is strictly related to the use of eqn (1) according to Hill's theoretical analysis of the Poole-Frenkel mechanism in amorphous semiconductors.This assumption will be relaxed in the discussion of the experiments performed at A = 230 nm, where the hypothesis of. transport in the extended states will be made for the photogenerated carriers. The second assumption is explained by considering the amorphous nature of WO, as well as the experimental finding of a negative value for the diffusion length of the holes, Lh, as obtained by extrapolation to zero photocurrent of the iph against l/C plots.30 A negative value of &,, which is physically meaningless, can be taken as evidence of a strong recombination process in the field-free region of the semiconductor, so that a negligible contribution to the photocurrent will come from this region.The third assumption is in agreement with the results described in fig. 6-8. The use of the same formulae for crystalline semiconductors in the depletion region will be considered as a simplifying good approximation. According to previous assumptions, the final expression for the photocurrent in the absence of kinetic control by the electrode-electrolyte interface can be written as iph = ip-F Gl(x,,) = 74, El - exp (- 2a x,,)] (4) where ip-F is given by eqn (1) and GA(xsc) is the total density of photogenerated carriers in the space-charge region x,,. In eqn (4) the existence of a possible reflection at the metal-oxide interface has been taken into account, and q50 represents the photon flux incident on the surface of the anodic films after reflection losses.z is the lifetime of the excess carriers photogenerated in the space-charge region and F,, is the average electric field in this region. The use of eqn (1) rests only on the experimental fitting of the photocurrent curves for WO, anodic films, but the use of eqn (2) or ( 3 ) could be more correct in other systems. According to the a values reported in the literaturelo? le for amorphous evaporated films or single-crystal WO, semiconductors, and on the basis of the experimental results reported in fig. 6-8, it seemed a good approximation to expand the exponential term in eqn (4) by retaining the first two terms, In such a way the final expression for the photocurrent becomes iph = 2Az4, Fhv a xsc.By taking into account the semiconducting nature of our films and by substitution of the mathematical expressions for Cv and x,, usually employed with crystalline semiconductors in the depletion region 32 eqn ( 5 ) becomes where E,, is the dielectric constant of the film, E , the permittivity of free space, Nd the density of donors in the film, and #sc the potential drop in the semiconductor; the remaining symbols have their usual meaning. By identifying the parameters which play the most important role in determining the photocurrent behaviour, it is possible to write e iph = constant x aN;, (7)F. DI QUARTO, G. RUSSO, C. SUNSERI A N D A. DI PAOLA 3443 where bSc has been expressed as a function of the electrode potential UE and the flat-band potential UFB.317 32 In agreement with eqn (7), a plot of i$ against U, should give a straight line, whose intercept with the voltage axis should be a measure of the flat-band potential.In fig. 2 and 3 the experimental data seem to fit both provisions. In fact good straight- line behaviour is observed at the various wavelengths in the range 270-380 nm. The intercepts with the voltage axis (0.0+0.1 V us. MSE) are not dependent on the wavelengths used and are in good agreement with the flat-band potentials previously reported.' This shows that the assumption 2axs, 6 1 made in the expansion of the exponential term in eqn ( 5 ) is valid over the range of wavelengtbs used. In eqn (7) note that, at a fixed wavelength, the slopes of the ibh against UE plots are inversely proportional to the cube root of the density of donors in the film.This seems qualitatively in accordance with the experimental results reported in fig. 3, 7 and 8 showing the influence of thickness and ageing under polarization on the photocurrent behaviour as well as on the distribution of donors in the oxide films. In fact, both the photocurrent behaviour and the capacitance data could be interpreted on the basis of a donor density which grows during the experiment. The dependence of iph on a shown in eqn (7) indicates that in fig. 9 the use of iph instead of a at every wavelength is a good approximation, provided that the experiments are performed at constant potential, and the eventual change of donor distribution is negligible over the time-scale of the experiment, It is useful at this point to observe that the use of eqn (2) or (3) rather than eqn (1) would give, respectively, the following final expressions after the mathematical treatment outlined above : e i,, = constant x a or iph = constant x aNa e (9) Preliminary experiments performed with amorphous Nb,O, anodic oxide films show a photocurrent against potential behaviour which seems in aqeement with eqn (8).Further investigations are, however, necessary before reaching a final conclusion. Eqn (9) shows a dependence of the photocurrent on the electrode potential analogous to that reported for single-crystal semiconductors, in which a Poole-Frenkel mechanism has been invoked to explain the photocurrent characteristic^.^^^ 29 This is due to the use in Hill's analysis of a relationship between velocity and mobility of carriers in semicrystalline materials which is analogous to that existing between mobility and drift velocity in crystalline materials.If this is true, the result of eqn (9) would be more general than the equation i,, = CV1-28 reported by Lemasson et al. for the zinc-selenide-electrode-electrolyte junction. 16t 29 The experimental results obtained at 3, = 230 nm can be interpreted in a straight- forward manner by relaxing assumption (iii) and by assuming that the electron-hole pairs generated by photons absorbed in the depletion layer are swept away by the electric field existing in this region in such a short time that the thermalization process in the non-extended states does not occur. The appearance of this conduction process in the extended states of higher energy agrees with the results reported in fig.4 and 5. In fact the square-root dependence of the photocurrent on the electrode potential can be explained by taking into account the dependence of the space-charge region of the3444 PHOTOELECTROCHEMISTRY OF AMORPHOUS WO, ELECTRODES semiconductor on the electrode potential and by assuming no recombination in the depletion layer, in accordance with Butler’s mode1.12 In this hypothesis the expression for the photocurrent becomes e iph = constant x aN;: In this case the slopes of the straight lines reported in fig. 5 should be in qualitative agreement with the hypothesis previously made concerning the change in donor distribution in the films during the experiment.CONCLUSIONS On the basis of the experimental results and the existing theories concerning the Poole-Frenkel mechanism of electrical conduction in amorphous materials, we have suggested an interpretative model of the photocurrent characteristics observed at the amorphous-WO,-electrolyte interface. It has also been shown that a change in the photocurrent behaviour with wavelength could be ascribed to different mechanisms of electrical conduction in the amorphous materials : hopping in localized states or ‘ free-carrier-like ’ transport in extended states at higher energies. According to the experimental data, a mobility gap ca. 0.3 eV larger than the band-gap energy of crystalline WO, has been attributed to the WO, amorphous anodic films. The existence of non-extended states below the mobility gap, where the hopping mechanism of electrical conduction follows a Poole-Frenkel law, suggests an over- lapping of these states by a distribution of defects having a donor-like or acceptor-like behaviour in the conduction or valence bands, re~pectively.~ The n-type semiconducting behaviour, as well as the non-stoichiometry usually reported for polycrystalline or single-crystal WO, electr~des,~~ suggest that the oxygen vacancies produced during the anodization process could behave as donor-like centres. Finally, we stress that a photoelectrochemical study of the amorphous- semiconductor-electrolyte interface can be a valuable tool in obtaining further knowledge of the electronic properties of these materials.F. Di Quarto, A. Di Paola and C.Sunseri, Electrochim. Acta, 1981, 26, 1177. S. Roy Morrison, Electrochemistry at Semiconductors and Oxidized Electrodes (Plenum Press, New York, 1980). F. Di Quarto, A. Di Paola and C. Sunseri, J . Electrochem. Soc., 1980, 127, 1016. M. A. Butler, J . Electrochem. Soc., 1979, 126, 338. W. E. Spear, P. G. Le Comber and A. J. Snell, Philos. Mag., 1978, 38, 303. C. R. Wronski, IEEE Trans. Electron Devices, 1977, ED-24, 351. N. F. Mott and E. A. Davies, Electronic Processes in Non-crystalline Materials (Clarendon Press, Oxford, 1971), p. 237 ff. * S. Ikonopisov, E. Klein, A. Stanchev and T. S. Nikolov, Thin Solid Films, 1975, 26, 99. S. Kapusta and N. Hackermann, Electrochim. Acta, 1980, 25, 1001. l o S. K. Deb, Philos. Mag., 1973, 27, 801. l1 Z. M. Hanafi and M. A. Khilla, 2. Phys. Chem. (N.F.), 1974, 89, 230. l2 M. A. Butler, J . Appl. Phys., 1977, 48, 1914. l 3 W. W. Gartner, Phys. Rev., 1959, 116, 84. l4 R. H. Wilson, J . Appl. Phys., 1977, 48, 4292. l 5 H. Reiss, J . Electrochem. Soc., 1978, 125, 937. l6 P. Lemasson and J. Gautron, Phys. Status Solidi (A), 1979, 53, 303. l7 J. Reichmann, Appl. Phys. Lett., 1980, 36, 574.F. DI QUARTO, G. RUSSO, C. SUNSERI AND A. DI PAOLA 3445 W. J. Albery, P. N. Bartlett, A. Hamnett and M. P. Dare-Edwards, J. Electrochem. SOC., 1981, 128, 1492. Iy A. K. Jonscher and R. M. Hill, in Physics of Thin Films, ed. G. Hass. M. H. Francombe and R. W. Hoffman (Academic Press, New York, 1975), vol. 8, p. 170 ff. 2o R. M. Hill, Philos. Mag., 1971, 23, 59. 21 R. M. Hill, Philos. Mug., 1971, 24, 1307. 22 N. F. Mott, Philos. Mag., 1971, 24, 911. 23 E. A. Owen, J. Non-Cryst. Solids, 1977, 25, 372. 24 P. G. Le Comber, W. E. Spear and D. Allan, J. Nun-Cryst. Solids, 1979, 32, 1. 25 R. M. Hill and A. K. Jonscher, J. Non-Cryst. Solidy, 1979, 32, 53. 26 D. M. Pai and S. W. Ing Jr, Phys. Rev., 1968, 173, 729. 27 M. D. Tabak and P. J. Warter Jr, Phys. Rev., 1968, 173, 899. 28 A. K. Jonscher and A. A. Ansari, Philos. Mag., 1971, 23, 205. 29 J. Gautron, P. Lemasson, F. Rabago and R. Triboulet, J. Electrochem. SOC., 1979, 126, 186. 30 S. Roy Morrison, J. Vm. Sci. Technol., 1978, 15, 1417. 31 H. Gerischer, in Advances in Electrochemistry and Electrochemical Engineering, ed. P. Delahay and 32 V. A. Miamlin and Yu. V. Pleskov, Electrochemistry of Semiconductors (Plenum Press, New York, 33 M. J. Sienko and J. M. Berak, in The Chemistry of Extended Defects in Non-metallic Solids, ed. C. W. Tobias (Interscience, New York. 1961), vol. 1, p. 139. 1967). L. Eyring and M. O’Keeffe (North-Holland, Amsterdam, 1970), p. 541. (PAPER 1 / 1706)