P. Rochon, E. Fortin, J. P. Zielinger
Sep 1, 1975
Citations
0
Citations
Journal
Le Journal De Physique Colloques
Abstract
The spectral distribution of the photoconductivity and, principally, of the photovoltaTc and wavelength modulated photovoltai'c effect for a Au-CuGaS2 Schottky barrier, has been studied at 300 K and 90 K. Several structures are observed in the spectra at energies above and below the gap, which can be related to absorption reflectivity or luminescence structures in the material. The presence of deep electron traps is also revealed. The transfer of electrons under optical excitation. from compensated acceptors to these traps enhances the dark conductivity at low temperatures and has a sensitizing effect on the photoconductivity. 1 . Introduction. Thc new ternary materials CuGaS, and CuInS, have recently been the subject of interest for their semiconducting and optical properties in view of thcir possible applications in non linear optics [l], [2], [3] and electroluminescent diodes [4], [S], [6]. Optical absorption [5], [7], [8], [9], [IO], reflectivity [5], [7], [g], [9], [l01 and luminescence [l l ] have been investigated for CuGaS,. Rccently Tell and al. [l21 have studied the structure of the valcncc band of the compounds by alloying them in the form of CuGa,In,-,S,. Recent efforts to investigate the low temperaturc photoconductivity of CuGaS, havc met with difficulties related to photomemory effects and to the nonreproducibility of results. PEM effect studies are also likely to be difficult because of the very low carrier mobility in the material. We present here some preliminary results on the photoconductivity a t 300 K and on the photovoltaic and wavelength modulated photovoltaic effects a t a Au-CuGaS, Schottky barrier a t 300 K and 90 K. Some structures seen in absorption or reflectivity are also observed in the PV spectra and mostly in their derivatives. This method of observation is relativcly simple ; the signal is strong and the results are reproducible and less sensitive to many of the extrinsic efrects seen in photoconductivity and even in luminescence. 2. Experimental. The samples of a few mm2 X 1 mm were of thc orange variety (undoped) with natural (1 12) faces. Contacts were made in the sandwich configuration, thc ohmic back contact consisting of aquadag (colloi'dal graphite) and the front rectifying contact concisting of a semi-transparent evaporated gold layer. The material was scmiinsulating with a resistivity of about 106 R.cm at room temperature. The samples were mounted stressfree with General Electric DC-Z9 compound in a cryostat equipped with an independent experimental chamber and good thermal contact was provided by a He exchange gaz a t a pressure of about 1 torr. The optical system consisted of a 3 meter spectrometer equipped with a vibrating rcfractor plate (12 Hz) located near the entrance slit and producing a modulation amplitude of about 10 A. For A. C. measurements (light intensity modulation) light was chopped a t 85 Hz and synchronous detection used with the electrometer a t unity gain serving as a high to low impeArticle published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphyscol:1975313 C3-68 P. ROCHON, E. FORTIN, J. P. ZIELINGER AND C. SCHWAB dance transformer. The D. C . measurements (constant illumination) were made with a high impedance electrometer. The P. C. and P. V. spectra were obtained under unpolarized illumination. All P. V. spectra were recorded at zero bias since during a test run, external bias of up to one volt showed no significant influence on the results. 3. Preliminary remarkson P. C. and P.V. detectors. 3.1 P. C. DETECTOR. The photoconductiyity in a semi-conductor in the intrinsic range may be a combined majority and minority carrier effect. Moreover the lifetimes of the photocarriers depend strongly on the actual population of the recombination centers or traps. Thus if during illumination a charge exchange occurs between two types of centers these parameters may be modified. Another difficulty may arise from electrode effects if the contacts are not strictly ohmic. 3.2 P. V. DETECTOR. The P. V. erect is in principle essentially a minority carrler (electrons in the case of CuGaS,) phenomenon. T h ~ s is known to be true for low resistivity materials, but can also be easily de~ilonstrated for high resistivity materials [13]. On the other hand owing to the strong electric field in the barrier, electrons tend to remain in it for a period substantially shorter than their incan lifetime. Consequently no recombination occurs in the effective photocell space and all electrons generated in the barrier region will contribute to the P. V. efllect. Thus the P. V. effect appears to be less sensitive to extrinsic effects than the photoconductivity. For low values of the optical absorption coeflicient a with respect to the width W (generally of the order of 10-"10-4 cm) of the barrier (a-' > W) only a fraction proportional to a of the incident photons is absorbed. For high values of a (a-' < W) the P. V. effect is governed only by the total amount of light energy penetrating into the sample. Thus if the incident photon flux is independent of wavelength, the spectral response will be proportional to a in the low energy region (provided that free carriers only are created) and show a saturation in the high energy region which begins at a wavelength which satisfies approximately the condition a' 2: W. However reflectivity anomalics resulting in a change in the eKective photon flux penetrating into the sample may well be detected in this region especially in the wavelength modulated spectrum. 4. Results and discussion. 4.1 P 1 3 o r o c o ~ ~ ~ c TIVITY AT 300 K. Photoconductivity spectra at 300 K present no particular problems. Such a spectrum is shown in figure 1 in both direct and wavelength modulated forms. The high energy maximum at 0.501 pm (2.48 eV) can be attributed to band to band transitions. This value found for the band gap is in FIG. l . Intensity modulated and wavelength modulated photoconductivity spectra at 300 K. good agreement with the one deduced from optical measurements. The low energy peak at about 0.516 pm seems to correspond to : a) the electroluminescence band at 0.518 pm observed at room temperature by Wagner et alii [l41 ; b) the luminescence peak in the interval 0.5165 pm0.519 pm, observed by J. von Bardeleben et alii [ I l l at low temperature ; c) the absorption line at 0.513 pm (2.416 eV) observed at 4.2 K by Ringeissen et alii [9], Regolini [l01 and Regolini et alii [g] in green crystals. The absorption band has been attributed to impurity centers. However, it has not been observed in the spectrum of orange crystals, probably because the concentration of impurity absorbers is too low in this type of material. J. von Bardeleben et alii attribute the corresponding luminescence band (which is also present in the spectrum of undoped, i. e. orange crystals) to donor-acceptor recombinations. The absorption at 0.513 pm may thus correspond to valexe band-donor transitions. This correlation allovts to estimate the donor ionization energy at about 80 meV. Taking into account the value 160 meV obtained by J. von Bardeleben et alii for E, + E, (E,,, E, = acceptor and donor ionization energy respectively), a weaker absorption band corresponding to acceptordonor transitions may be expected to lie at about 0.525 pm (2.36 CV). So far, no absorption as well as no structure in the photoconductivity spectrum could be detected in this wavelength region. The importance of the extrinsic peak compared to the intrinsic photoconductivity leads to the eonclusion that the lifetime of the holes created in the bulk of the material by extrinsic excltation is very high with respect to the lifetime of the photocarriers generated near the surface by the strongly absorbed intrinsic excitation. Optical excitation may also be coupled with a shift of the Fermi level towards the valence band. Such a shift gives rise to a strong increase in conducDIRECT AND WAVELENGTH MODULATED PHOTOCONDUCTIVITY C3-69 tivity which may remain for a more or less long time after the light has been removed ; however it needs an appreciable charge transfer from ionized acceptors to deep electron traps and consequently it is more likely to occur at low temperatures. Indeed as the temperature is lowered the P. C. signal in the extrinsic region becomes very high and the spectra are very difficult to reproduce because of photomemory effects : the light induces a strong enhancement of the dark conductivity which depends on the light intensity and on the exposure time and simultaneously an increase of the photosensitivity is observed. The decrease of the conductivity after interruption of the illumination is a very slow process. These preliminary observations indicate the presence of dccp electron traps in CuGaS,. As a result of illumination, clcctrons are transferred from compensated acceptor centers to these traps. The charge exchange has a sensitizing effect on photoconductivity. If the temperature is sufficiently low, the non equilibrium state may exist for an indefinite time after illumination. Thus the simple but classical photoconductivity measuring technique could not give meaningful results. For this reason no further attempt was made to record low temperature P. C. spectra. 4 . 2 PHOTOVOLTA~C EFFECT AT 300 K . In principle, for the reasons invoked above the P. V. spectra should be less sensitive to extrinsic effects. A typical direct (intensity modulated) spectrum is shown in figure 2. This spectrum exhibits, in general, a single F~G. 2. Photovolta'ic excitation spectra at 300 K and 90 K. structure. In all cases it drops off more sharply than the P. C . spectrum, in the low energy range. The reason is that the signal measured by the P. V. detector depends mainly on the absorption coefficient a, thus extrjnsic absorption, which is weak compared to intrinsic absorption in orange CuGaS,, can hardly be detected. For the same reason the P. V. response decreases more slowly i