Thin Solid Films, 127 (1985) 305-312 PREPARATION AND CHARACTERIZATION
LUMINESCENCE AND PHOTOELECTROCHEMISTRY OF CdS THIN FILM ELECTRODES I. JIMt~NEZ AND P. SALVADOR
lnstituto de Cat(diMs y Petroleoquimica, Consejo Superior de lnvestigaciones Cientificas, Serrano 119, Madrid 6 (Spain) F. DECKER lnstituto de Fisica, UNICAMP, 13100 Campinas, Sao Paulo (Brazil) (Received June 25, 1984; accepted December 24, 1984)
Thin polycrystalline CdS films were prepared for photoelectrochemical applications. The presence of electronic states in the band gap of the material was investigated by means of several optical techniques. The sub-band-gap photocurrent, the photoluminescence and the electrolyte electroluminescence indicated a high density of such states which were attributed mainly to sulphur and cadmium vacancies.
The fabrication and study of thin polycrystalline cadmium chalcogenide films for use in both solid 1 and liquid 2 junction solar cells has received considerable attention in recent years. CdS is a particularly interesting chalcogenide since it can be used in both CdS/Cu2S and CdS/CdTe thin film photovoltaic systems 1. The behaviour of polycrystalline CdS in photovoltaic and photosynthetic photoelectrochemical cells has also been widely studied 3. Intra-band-gap states play an important role in bulk trapping and recombination processes and in charge transfer at the semiconductor-electrolyte interface. Photoluminescence (PL) is a useful non-destructive technique for the characterization of deep-level traps in the semiconductor, while electrolyte electroluminescence (EL) is an interesting tool for the in situ study of band gap states at the interface with the electrolyte. In PL experiments radiative electron-hole recombination is stimulated by light. In EL, in contrast, minority carriers are injected into the semiconductor by means of an electrochemical reaction at the contact with the electrolyte and undergo radiative electron-hole recombination with the majority carriers 4. The PL of CdS single crystals has been widely studied from 1950 onwards s. However, EL studies are a much more recent development4'6. Although from a practical point of view polycrystalline thin films are much more interesting than single crystals, very little attention has been paid to the study of CdS thin films by means of luminescence techniques. 0040-6090/85/$3.30
© ElsevierSequoia/Printedin The Netherlands
I. JIM~NEZ, P. SALVADOR, E. DECKER
We report here the PL and EL behaviour of CdS polycrystalline thin films deposited onto titanium substrates from an aqueous dispersion of CdS powder. By combining the information obtained from the PL and photocurrent action spectra of the films in a photoelectrochemical cell we have been able to establish the presence of bulk band gap states (electron and hole traps) which act as centres for both recombination and sensitization to sub-band-gap light. EL is shown to be useful for the detection of surface states. 2. EXPERIMENTAL DETAILS
2.1. Fabrication of the CdS electrodes Titanium strips of area about 1 c m 2 and thickness 0.5 mm were lapped with 240 grade emery paper, cleaned with distilled water and acetone and etched for 30 s in an aqueous solution of 4 wt.~ H F and 30wt.~ HNO3. 10mg of finely powdered CdS (Fluka; purity, 99.999~o; grain size about 1 ~m) were ultrasonically dispersed in 10 cm 3 of ultrapure water (Milli-Q, Millipore, Bedford, MA). The dispersion was deposited dropwise onto the etched titanium sheet and was dried under a 200 W IR lamp. The electrode was then heated in air at 640°C for 10min and was subsequently cooled to room temperature for about 5 min. The thickness of the CdS layers, estimated from the weight, was a few microns and the electrode resistance was of the order of 10D. 2.2. Techniques The electrochemical and photoelectrochemical measurements were carried out using a Wenking POS 73 Potentioscan connected to a fast YEW 3033/3 x - y recorder. A single-compartment photoelectrochemical cell, with a saturated calomel electrode (SCE) as the reference electrode and a platinum counterelectrode, was used. The light source was a stabilized 150 W xenon lamp coupled to a high intensity Bausch & Lomb 200-800nm monochromator. The electrolyte was an aqueous solution of 0.2 M Na2SO4 and 0.2 M Na2SO 3. The PL spectra were taken at room temperature using a scanning monochromator (McPherson 218) and a cooled photomultiplier (RCA 7102) with useful sensitivity down to 1.1 eV. The light source was the same as that used in photoelectrochemical experiments; finally an H e - N e laser (LT-16, CW Radiation Inc.) with a maximum power of 25 mW was employed. The EL spectra were obtained by pulsing the working electrode with a double-pulse control generator (Wenking D P C 72) (between 0 V (SCE) (3 s) and - 1.0 to - 2.0 V (SCE); pulse duration, 1 s) while the monochromator was scanned at 20 nm min-2. Aqueous solutions of 1 M K O H plus 0.1 M K 2 S 2 0 8 and of 1 M K O H plus 0.1 M K3Fe(CN)6 were used in EL experiments. The electrolyte was magnetically stirred during the experiments. 3.
The photocurrent action spectrum of the CdS photoelectrode in an aqueous solution of 0.2 M Na2SO3 plus 0.2 M Na2SO 4 is shown in Fig. 1 where a clear subband-gap photoresponse can be seen. The spectral response of the Schottky junction
THIN FILM ELECTRODES
Fig. 1. Photocurrent action spectrum of the CdS thin film in an aqueous solution of 0.2 M SO32- plus 0.2 M SO42 - (pH 11) (polarization potential, 0 mV (SCE)). The photocurrents Iph were corrected for the xenon lamp photon flux. Curve A is a fourfold magnification of the first curve for 2 > 550 rim.
at the semiconductor-electrolyte interface can be analysed using G~rtner's model 7. If it is assumed that 1/ct is much greater than W and Lp, where ~t is the optical absorption coefficient, W is the thickness of the depletion layer and Lp is the hole diffusion length, the quantum efficiency 11of the photoelectrochemical cell is given by the following equation 8:
(rlhv) 2/n =
where A is a constant, Eg is the band gap and n is 1 or 4 for direct or indirect allowed transitions respectively. In the case ofCdS (n = 1) a plot of(~/hv) 2 versus hv should be linear and the intercept with the abscissa should give E~. Figure 2 shows a plot of (Iph/2) 2 versus 2 from the data of Fig. 1. At least two linear segments appear in Fig. 2. One of them extrapolates at Ex = 2.3 eV (i.e. very close to the CdS band gapS), and the other at E2 = 2.1 eV. The existence of E2 can be attributed to the existence of filled band gap states (narrow band) 0.2eV above the valence band from which electrons can be excited to the conduction band with sub-band-gap light (Fig. 3(C)). The hole mobility in this narrow band should be sufficiently high for the holes to reach the interface with the electrolyte and to react with the reduced SO32- species according to the reaction
8032- + O H -
+2h+ ~SO42- +H +
Figure 4 shows the PL spectra at room temperature of the CdS thin film electrode in a 1 M Na2SO4 aqueous solution (curve a) and the CdS powder on the titanium substrate prior to sintering in air (curve b). Both spectra are characterized by an intense large peak with maxima at about 800 nm and about 730 nm respectively. The positions of the maxima do not depend sensibly on the excitation intensity. Figure 5 shows the EL spectra of the CdS thin film in 0.1 M K2S20 s. Two broad bands are observed, one at about 800 nm (peak I), i.e. at the same position as the peak of the PL spectrum, and the other at about 925 nm (peak II). The relative intensities of the two peaks depend on the potential of the applied cathodic pulse. At the lowest applied negative potential ( - 1 . 0 5 V (SCE)) peak I is much more intense than peak II. However, as the applied pulse becomes more cathodic the intensity of peak II
I. JIMI~NEZ, P. SALVADOR, F. DECKER
,'~ 2.0 ...,.~
i/ xl0/ /
• I, "r= /
"/ 1.5 E2
I i I 2.0 (1/~)/10 -3 nrn-1
Fig. 2. Plot of (Iph/J.)2 v s . 1/2 for direct transitions from the photocurrent action spectrum of Fig. 1. The intercept E1 represents the CdS band gap (E ~ 2.3 eV). E 2 corresponds to electronic transitions from filled band gap states (about 0.2 eV above the valence band) to the conduction band.
increases with respect to that of peak I; at - 1.7 V (SCE) peak II becomes more intense than peak I. Only peak II is observed in the EL spectrum with Fe(CN)63 - as the acceptor electrolyte species (Fig. 6). 4. DISCUSSION The PL spectrum of a pure CdS single crystal at r o o m temperature shows a rather narrow peak at 524 nm (green emission) and a broad band which peaks at about 720 nm (red emission) 9. The green emission is attributed to direct band-toband recombination, while the red emission (1.7 eV) is attributed to the recombination of an electron trapped at a sulphur vacancy (0.7 eV below the conduction band edge) with a hole at the valence band generated by band gap illumination 9 (Fig. 3(A)). Very intense r o o m temperature photoluminescence has also been observed in polycrystalline spray-pyrolysed CdSI o. In this case two very broad lines at 750 and 1030 nm were present in the PL spectrum. The former line was identified with a complex sulphur interstitial 1t, and the latter with a transition between a sulphur vacancy and a cadmium vacancy VCd acting as a hole trap very near the valence band edge t 2. The absence of green luminescence may indicate a very large
CdS THIN FILM ELECTRODES
~ 0;+OH-+2h'e-SO;+H+) '~~ o;/so;
-2. e_ Ec
EF . . . . . . .
O" Fe(CN)3"14" o +1-
Fig. 3. Diagram showing the CdS and electrolyte energy levels and the EL and PL mechanisms: (A) 1.7eV PL produced under band gap illumination of the non-sintered CdS powder (Vs represents sulphur vacancieswhere electrons are first trapped and then recombine radiativelywith photogenerated holes);(B) 1.5eV PL produced under band gap illumination of the CdS thin filmelectrode (radiative recombination takes place between an electron trapped at Vs and a hole trapped at a cadmium vacancy Vcd; (C) mechanism for the sub-band-gap photoresponse of the CdS thin filmelectrode (holes are photogenerated at the Vcd levelswith photons with hv < Eg and move through the Vca narrow band to reach the interface with the electrolyte where they are injected into the SO32- filled levels);(D) Fe(CN)63-/FetCN)64- and SO~/SO42- redox potentials relative to the CdS energy levels; (E) mechanisms for the 1.5 eV and 1.37 eV EL of the CdS thin filmelectrode in contact with Fe(CN)63-/Fe(CN)64- and SO4~/SO42- redox couples respectively(holes injected into the valence band from overlapping SO4 ~ radicals may either be trapped at a Vca and then recombine with an electron trapped at a Vs (mechanism a) or be trapped at a surface state (mechanism b); mechanism b also takes place when holes are directly injected into the filled surface states from empty levelsof the Fe(CN)63- electrolyte). density of these l u m i n e s c e n t defects which should a c c u m u l a t e in the i n t e r g r a n u l a r regions 13. The large a m o u n t o f stress existing in the region between the grains should explain the m a r k e d line b r o a d e n i n g l°. The P L spectrum of the u n t r e a t e d CdS powder exhibits the same red emission as the pure CdS single crystal. However, a shift towards the I R is clearly observed after the thermal t r e a t m e n t used to prepare the p o w d e r e d CdS thin film electrode. This shift m a y indicate the existence of c a d m i u m vacancies acting as relatively shallow hole traps which are generated, together with s u l p h u r vacancies, d u r i n g the t h e r m a l sintering t r e a t m e n t 14. As s h o w n in Fig. 3(B) the emission at 800 n m ( a b o u t 1.5 eV) can be explained as the radiative r e c o m b i n a t i o n of a n electron t r a p p e d at a s u l p h u r vacancy with a hole trapped at a c a d m i u m vacancy. The c a d m i u m vacancies should lie a b o u t 0.2eV a b o v e the valence b a n d edge (Fig. 3(B)), in good agreement with the value of E 2 in Fig. 2,
I. JIMI~NEZ, P. SALVADOR, F. DECKER
II -I 05V
: I I
600 h (nrn)
Fig. 4. PL spectra at room temperatureof polycrystalline CdS: curve a, thin film electrode sintered in air in contact with a 1 M Na2SO 4 aqueous solution (pH 6); curve b, CdS powder deposited onto the titanium substrate prior to the sintering treatment. The light excitation source was an He Ne laser. Fig. 5. EL spectra of the CdS thin film electrode in 0.1 M K2S208 aqueous solution under various cathodic potential steps. w h i c h means that c a d m i u m vacancies are very probably i n v o l v e d in the observed sub-band-gap p h o t o r e s p o n s e of Fig. 1. The existence of hole traps identified with c a d m i u m vacancies and located at about the same position with respect to the valence band in both single-crystal and polycrystalline p-CdTe has also been reported by several workers t~-16. W i t h respect to the EL of the C d S - e l e c t r o l y t e j u n c t i o n only single-crystal data were found in the literature 4'6. Here, we report the EL spectra of the polycrystalline C d S - l i q u i d j u n c t i o n for the first time. Peak I of the EL spectrum in the presence of $ 2 0 8 z - is very similar to the peak characterizing the PL spectrum, which indicates
600 /~ (nm)
Fig. 6. Comparison of the EL spectra of the CdS thin film in 0.1M K2SzOs (curve a) and 0.1 M K3Fe(CN)6 (curve b). The potential of the cathodic step is - 1.3 V (SCE).
CdS THIN FILM ELECTRODES
an analogous recombination mechanism. SO4- radicals generated by electroreduction with conduction band electrons of $2082 - ions according to S2Os 2- +e--~SO,,
a r e electrolyte acceptor states which overlap the valence band and can inject holes efficiently into the semiconductor (Fig. 3(E)). Injected holes are first trapped at cadmium vacancies in the bulk (Fig. 3(E), mechanism a) and subsequently recombine with electrons trapped at sulphur vacancies in a radiative process. A similar radiative recombination mechanism due to a second acceptor state lying deeper in the band gap must be responsible for peak II. Various explanations can be given for the absence of this peak in PL spectra. First we must consider that the spatial distribution of injected holes in the case of photon excitation is very different from that in the case of elecrolyte excitation. Different bulk states have different radiative recombination rates depending on the energy, the temperature, the local density of electron-hole pairs and other parameters. Therefore we cannot exclude the involvement of a second bulk state in peak II of the EL spectra. A second explanation is the involvement of surface states, again hole acceptors, in the EL mechanism (see Fig. 3(E), mechanism b). This explanation has already been proposed for other cases of electrolyte EL 17'1s. At potentials close to - 1 . 0 V (SCE) the injection rate is low and mechanism a is predominant. For decreasing potentials and increasing injection rates mechanism a saturates and mechanism b becomes predominant. The surface state acceptor model is reinforced by the fact that in the presence of Fe(CN)6 3- only peak II is observed in the EL spectrum (Fig. 6). The F e ( C N ) 6 3 acceptor states do not overlap with the CdS valence band and therefore cannot inject free holes. In contrast, trapped holes can be injected into filled surface states (Fig. 3(E), mechanism b) and then recombine radiatively with electrons from the conduction band. 5. CONCLUSIONS
Thin film polycrystalline CdS electrodes in aqueous electrolytes were studied using luminescence spectroscopy. States lying in the band gap were observed using this technique. A close correspondence to the states detected by means of sub-bandgap photocurrent spectroscopy was found. Further experiments will be needed in order to elucidate the real involvement of surface states in the hole injection mechanism. REFERENCES 1 Proc. 16th 1EEE Photovoltaic Specialists Conf., San Diego, CA, 1982, IEEE, New York, 1982. 2 W.L. Wallance, 1. I. Nozik, S. K. Deb and R. H. Wilson (eds.), Proe. Syrup. on Photoelectrochemical Processes and Measurement Techniques for Photoelectroehemieal Solar Cells, Electrochemical Society, Pennington, N J, 1982. 3 A.B. Ellis, S. W. Kaiser and M. S. Wrighton, J. Am. Chem. Soc., 98 (1976) 6855. A. Heller, K. C. Chang and B. Miller, J. Electroehem. Soc., 124 (1977) 697. H. Gericher and J. Gobrecht, Ber. Bunsenges. Phys. Chem., 82 (1978) 520.
6 7 8 9 10 11 12 13 14 15 16 17 18
I. JIMI~NEZ, P. SALVADOR, F. DECKER
T. Inoue, T. Watanabe, A. Fujishima and K. Honda, Bull. Chem. Soc. Jpn., 52 (1979) 1243. M. Neuman-Spallart and K. Kalyanasundaram, Ber. Bunsenges. Phys. Chem., 85 (1981) 1112. P. Chartier, B. Ba, J. Ebothe, N. A. Vante and H. N. Cong, J. Electroanal. Chem., 138 (1982) 38l. B. Pettinger, H.-R. Sch6ppel and H. Geriseher, Ber. Bunsenges. Phys. Chem., 80 (1976) 849. R.E. Halsted, D. Curie and J. S. Prener, in M. Aven and J. S. Prener (eds.), Physics and Chemistry of I I - I V Compounds, North-Holland, Amsterdam, 1967, Chapters 8, 9. H.H. Streckert, B.R. Karas, D.J. MoranoandA. B. Ellis, J. Phys. Chem.,84(1980) 3232. Y. Nakato, A. Tsumura and H. Tsubornura, Chem. Phys. Lett., 85 (1982) 387. W.W. G~irtner, Phys. Rev., 116 (1959) 84. M.A. Butler, J. Appl. Phys., 48 (1977) 1914. A.A. Vuylsteke and Y. T. Sihvonen, Phys. Rev., 113 (1959) 40. B.J. Feldman and J. A. Duisman, Appl. Phys. Lett., 37 (1980) 1092. C.N. Elsby and J. M. Meese, IEEE Trans. Nuel. Sci., 21 (1974) 14. C.N. ElsbyandJ. M. Meese, J. Appl. Phys.,43(1972)4818. J.C. Knights, J. Non-Cryst. Solids, 35-36 (1980) 159. S.S. Ou, A. Bindal, O. M. Stafsudd, K. L. Wang and B. W. Basol, J. Appl. Phys., 55 (1984) 1020. F.A. Kr6ger, Rev. Phys. Appl., 12 (1977) 205. B.J. Feldman, J. L. Boone and T. Van Doren, Appl. Phys. Lett., 38 (1981) 708. K.H. Beekmann and R. Memming, J. Electrochem. Soe., 116 (1969) 368. Y. Nakato, A. Tsubomura and H. Tsubomura, Chem. Phys. Lett., 85 (1982) 387.