versão impressa ISSN 0366-1644
Bol. Soc. Chil. Quím. v.47 n.4 Concepción dez. 2002
ELECTRODEPOSITION AND CHARACTERIZATION OF ZnX
(X=Se, Te) SEMICONDUCTOR THIN FILMS.
G. RIVEROS, H GÓMEZ, R. HENRÍQUEZ, R. SSHREBLER, R.CÓRDOVA
Instituto de Química, Universidad Católica de Valparaíso,
Av. Brasil 2950, Valparaíso, Chile.
R. E. MAROTTI, E.A. DALCHIELE
Instituto de Física, Facultad de Ingeniería, Herrera y Reissig 565,
C.C. 30, 11000 Montevideo, Uruguay.
El presente trabajo describe la síntesis y la caracterización de películas delgadas de ZnX (X = Se y Te) obtenidas por electrodeposición a potencial constante en medio ácido. Previamente, se realizó un estudio voltamétrico y fotovoltamétrico sobre diferentes substratos conductores los que permitieron determinar las mejores condiciones para la electro obtención de estos compuestos. Las películas delgadas de ZnSe y de ZnTe fueron analizadas por diferentes técnicas (SEM, EDS, XRD y medidas ópticas). Las películas de ZnTe presentaron una composición muy cercana a la estequiometrica, en tanto que las de ZnSe presentaron un exceso de Se el cual puede ser eliminado por un adecuado tratamiento térmico. La caracterización óptica de ambos semiconductores depositados sobre titanio arrojó valores de ancho de banda prohibida de transición directa de 2,64 eV para ZnSe y 2,27 eV para ZnTe, muy cercanos a los aceptados en bibliografía.
PALABRAS CLAVES: ZnSe, ZnTe, semiconductores, electrodisposición
In present work we report the one step electrodeposition of ZnX (X = Se y Te) thin films in acid solution. In order to establish the appropriate conditions for the electrodeposition, a voltammetric and photovoltammetric study on different substrates was previously performed. The films were analyzed by different techniques (SEM, EDS, XRD and optical reflectance). The composition of the ZnTe films was very close to the stoichiometric one, instead, ZnSe films presented a selenium excess that can be eliminated with a proper annealing. Optical reflectance characterization of ZnSe and ZnTe samples grown on titanium gave direct band gaps values of 2.64 eV and 2.27 eV, respectively, in agreement with those reported in the bibliography.
KEY WORDS: ZnSe, ZnTe, semiconductors, electrodeposition.
Thin films heterostructures involving wide band gap semiconductors are widely studied for optoelectronic applications, such as, light emitting diodes or laser diodes. A number of methodologies are employed in the formation of high quality thin films , including: chemical vapor deposition, molecular beam epitaxy, pulsed laser, evaporation, and sputtering. There is, however, an interest to investigate other approaches which could open new or supplementary possibilities in terms of device properties, structure or engineering. Deposition from solutions belongs to these alternative methods that are particularly adapted for the deposition of chalcogenide materials either by electrodeposition or chemical deposition. Chemical bath deposition of sulfides and selenides is already used for producing interfacial buffer layers in high efficiency thin film solar cells based in copper indium gallium diselenide. Additionally, highly efficient devices can be obtained by electrodeposition such as for electrodeposited cadmium telluride solar cells or electrodeposited wide band gap sulfide or oxides buffer/window layers.
Within direct wide-band semiconductor materials, the zinc chalcogenides compounds have been the object of numerous studies concerning thin film electrodeposition from aqueous solutions. Most of the studies deal, mainly, with ZnSe1-15 and, in a lesser extent, with ZnTe16-20 . However, most of the experimental conditions employed for the electrodeposition usually do not lead to electrodeposits of enough quality for device application. The poor crystalline character of the as grown films and a chalcogenide excess in the deposits are often the main problems associated with this technique and an annealing is required in order to improve film quality.
To characterize the electrochemical behaviour of the Zn/X system in acid solution a detailed voltammetric study of the precursors and their mixtures was firstly accomplished. Then, from this study we selected the experimental conditions for the one step electrodeposition of ZnSe and ZnTe thin films at controlled potential. The films were characterized as grown and also after annealing by XRD, SEM and EDS. Furthermore, considering the lack of references related with appropriate procedures for the optical characterization of ZnX electrodeposits, we developed a methodology in order to determine reliable values of the energy gap in our films. As the samples were grown onto opaque substrates, we have studied the optical properties through the study of their reflectance. The clear coloration of the samples let us have well determined sharp edges in these spectra associated with the band edge properties. We have determined the energy gap from these spectra using two different methods. Excellent agreement is obtained when comparing the results between them and with the bibliographic reported values.
All solutions were prepared from analytical grade reagents and 18 MW.cm Millipore water. ZnSe and ZnTe were cathodically deposited in a conventional three-electrode single compartment electrochemical cell under potentiostatic conditions, from unstirred solutions. In the case of ZnSe film growth, the electrolyte bath was an acidic aqueous solution (pH=2.0-2.5) containing 0.5 M ZnSO4 and 1mM H2SeO3, held at 60 oC. In the case of ZnTe deposition, electrolytic bath was an aqueous solution of 0.5 M ZnSO4 and 0.1 mM TeO2 at a pH of 4.3, thermostated at a temperature of 95 oC. To avoid the possible oxygen electroreduction, the electrolyte solutions were purged with nitrogen for 30 min prior to each experimental series and then kept under flowing nitrogen during the experiments. The substrates were titanium ( Aldrich ) plates (about 1 cm2 of geometric area), that before each experiment were polished with emery paper, rinsed with deionized water, immersed during 10s in 3% HF solution and finally thoroughly rinsed with water. The electrochemical measurements (cyclic voltammetry and photovoltammetry) were performed in a conventional three-electrode glass cell using polycrystalline gold and glassy carbon as working electrodes. The gold (Johnson Matthey) and glassy carbon(Aldrich) electrodes were polished with a 0.3 m m alumina suspension and ultrasonically cleaned in deionized water. A platinum wire was used as a counterelectrode and a saturated Ag/AgCl (Ag/AgCl(sat)) served as reference electrode in both, electrochemical growth and electrochemical measurements. All potential values are reported versus Ag/AgCl(sat).
The cyclic photovoltammetry technique consisted in examining the current response of the electrodeposit/electrolyte interface under conditions of voltammetric scanning and modulated light excitation8, 21. The electrode (gold, glassy carbon)/electrolyte interface was illuminated with white light which was manually chopped during voltammetric scanning. Thus, the photoresponse of the forming semiconductor serves as an in situ diagnostic probe of the deposition mechanism and product formation.
The electrochemical measurements were done using model IM6e BAS-ZAHNER potentiostat whereas electrodeposits were obtained using model 263A EG&G Princeton Applied Research potentiostat.
Powder X-ray diffraction spectra of films were recorded with a Philips PW3710 diffractometer using CuKa 1 radiation and a nickel filter in order to filter the Kß radiation. The accelerating voltage was set at 40 KV with a 25 mA flux. Scatter and diffraction slits of 1o and 0.1 mm collection slit were used.
Quantitative standard less microanalyses were obtained using an energy disperse X-ray analysis (EDS) with a JEOL JSM-5410 apparatus.
As the films were deposited onto opaque substrates, the optical characterization of the samples was made from optical reflectance measurements. We used a 1000 W Xe lamp (ORIEL 6271) light source, and a S2000 Ocean Optics fiber optic spectrometer (with a Sony ILX511 Linear CCD Array and a 100 m m fiber acting as input slit). The effective optical range of this equipment is 350 - 1000 nm. Both specular and diffused light were measured with no considerable differences between the results. We have determined the energy gap by two different methods: the maximum of the reflectance first derivative against wavelength and from the optical absorption estimated from reflectance.
RESULTS AND DISCUSSION
3.1. Voltammetric studies
3.1.1. Study of Se(IV) and Te(IV) in aqueous acid solutions
The Zn2+ ion in an acidic solution (pH 2 - 2.5) is cathodically deposited on inert electrodes (Ti, glassy carbon (GC)) at potentials lower than -0.950 V vs. Ag/AgCl. This is in agreement with what is expected according to its standard reduction potential (-0.940 V vs. Ag/AgCl (sat)). Cyclic voltammograms show that the Zn deposited during the cathodic hemicycle is stripped in the anodic hemicycle, without any other electrochemical process, according to :
However, both the H2SeO3 (SeO2 in an acidic medium) and the HTeO2+ (TeO2 in an acid medium) show more complex voltammetric responses. Figure 1 shows the voltammetric scan recorded at a gold electrode in a 0.002 M H2SeO3 (pH = 1.5) solution. In order to facilitates the discussion of the voltammetric data, the following notation is used: individual voltammetric waves are labelled with a combination of capital letters, either "A" or "C", denoting anodic or cathodic processes, respectively; they are followed by consecutive arabic numerals and the element symbol of the species involved. Three cathodic waves, labelled C1Se, C2Se , and C3Se, respectively are observed. In spite of its very low wave amplitude, the C1Se process can be assigned to the reduction of H2SeO3 to Se0 on the surface of the gold electrode22. Additionally, when Se electrodeposits are made in the potential interval where C1Se process occurs, all the deposited Se0 is stripped to H2SeO3 in A1Se process. In spite of the fact that the latter presents two maximums (see below), they correspond to one single overall oxidation process. This is summed up in the following equation:
| ||Fig. 1. Cyclic voltammogram for a gold electrode in 0.2 M Na2SO4 at a pH=1.5 containing 2 mM H2SeO3. Potential scan rate 10 mV.s-1.|
The resolution of the stripping wave A1Se into two peaks is consistent with the fact that the former (more cathodic) comes from a Se0 surface, whereas the latter is related to Se0 catalytically deposited on the gold surface22.
Carrying out cyclic voltammetry tests with a rotating disk electrode, it was observed that process C2Se presents a limiting current(Fig.2) which obeys Levichs equation, i.e. it is proportional to the square root of disc rotation rate23. This indicates that the process is controlled by the mass transport from the bulk of the solution towards the electrode surface. Considering that apart from the H+ ion the only electroactive species in solution is H2SeO3, C2Se can be related to the transport of this species to the electrode surface. As this process is developed, the electrode surface appears covered with the characteristic red colour of amorphous selenium. This fact supports the assumption that C2Se is associated to the electroreduction of H2SeO3 to Se0 onto the selenium previously deposited on C1Se, through a process similar to that represented by Eq. (2). However, it is observed that the charge under C1Se and C2Se is much greater than that corresponding to A1Se , indicating that not all the deposited selenium is stripped. This partial oxidation is due to the amorphous character of red selenium, which makes it very resistive to charge transfer. In this way, the stripped selenium corresponds to the grey crystalline phase which is jointly deposited in a lesser amount with the amorphous red selenium24. This would explain the red colour of the electrode after the cyclic voltammogram is carried out.
| ||Fig. 2. a) Voltammograms of the cathodic hemicycle for a rotating disk gold electrode in 0.2 M Na2SO4 at a pH = 1.5 containing 2 mM H2SeO3. Potential scan rate 10 mV s 1 ;
b) jlim v/s w 1/2 graph (Levichs graph). jlim data were selected at - 0.500 V.
According to Eq. (2), the C2Se process should present a of -59 mV variation per pH unit. In fact, measurements of the C2Se peak potential between pH 1 and 2 gives a variation of -64 mV pH-1, which is close to the expected value for this reaction. Lastly, the C3Se process is attributed to hydrogen discharge reaction.
A similar approach was used to study the behaviour of a HTeO2+ solution in acidic medium. Due to the low solubility of this species, which is pH dependent (log CHTeO2+ = -2.07-pH)25, work was done at lower pH in order to maintain an appreciable amount of HTeO2+ in solution. Figure 3 shows the voltammetric response at a glassy carbon electrode in 4.1x10-4 M HTeO2+, 0.5 M Na2SO4 , pH 1 solution. Unlike selenium, four cathodic processes (C1Te, C2Te, C3Te and C4Te) and one single anodic process (A1Te) ) are observed.
The latter is proportional to C1Te and are directly related. In this way, C1Te is attributed to Te0 deposition and A1Te to its subsequent stripping according to the following reaction:
| ||Fig. 3. Cyclic voltammogram of a glassy carbon electrode in 0.5 M Na2SO4 at a pH=1.04 containing 0.4 mM HTeO2+. Potential scan rate 10 mV.s-1.|
Experiments with the rotating disk electrode show that the C1Te limiting current obeys the Levich equation (Fig4). Assuming that HTeO2+ is the only electroactive species present (same as in the case of selenium), this result confirms the participation of the process presented in equation (3). Accordingly, for C1Te a variation of -45 mV pH-1 is predicted. In fact, a variation of -50 mV pH-1 between pH 1 and 3 was observed, which is very close to what is expected.
The C2Te process corresponds to the reduction of the deposited Te to H2Te according to the following reaction26:
| ||Fig. 4. Cathodic hemicycle for a rotating disk glassy carbon in 0.5 M Na2SO4 at a pH=1.04 containing 0.4 mM HTeO2+. Potential scan rate 10 mV.s-1. ; inset: jlim v/s w 1/2 graph (Levichs graph) for processes CTe1,and CTe3. jlim data were selected at - 0.500 V and - 0.900 V respectively.|
According to this equation, a variation of 59 mV pH-1 should be observed in this process. Measurements of the variation of C2Te peak potential between pH 1 and 2 give a value of 56 mV pH-1, which agrees with what is expected.
As H2Te cannot go on reducing, the C3Te process is initially attributed either to the progressive reduction of Te0 , that may continue being deposited, or to the direct reduction, via six electrons, of HTeO2+ to H2Te. When experiments with a rotating disk electrode are carried out (Fig.4), it is observed that the C3Te processes is under mass-transport control (obeying Levichs equation), presenting a limiting current density that is equivalent to that found in the C1Te process. These findings demonstrate that C3Te depends on the Te(IV) in solution (the concentration of H+ ion is very high to be able to control the reaction by mass transport) and that it is a reaction occurring with the same number of electrons as in C1Te. Assuming that 4e- are involved in C1Te we can rule out the idea that C3Te might be the direct reduction of HTeO2+ to H2Te via 6e-. Nevertheless, it is also possible that H2Te reacts chemically with HTeO2+ present in solution to form Te0 according to:
In order for reaction (5) to occur, it is necessary that reaction (4) occurs twice. In this way, combining equations (4) and (5), we have:
The combination of the two reactions results in equation (3), which depends on the HTeO2+ present in the solution, and in which 4e- are involved. It should be borne in mind that the way in which reaction (6) occurs is different to that of reaction (3), although the reagents and products are the same. Finally, the C4Te process corresponds to the hydrogen discharge.
3.1.2. Study of the Zn(II)-Se(IV) and Zn(II)-Te(IV) systems
No considerable changes are observed in the voltammetric response by the addition of Zn(II) to a H2SeO3 solution, excepting for processes related to A1Se as the A1Zn-Se oxidation is much more intense than A1Se. The cyclic photovoltammetry technique, unlike its "dark" cyclic voltammetry counterpart, has the potential of identifying the formation of photoactive species (e.g., semiconductors) on an electrode surface8,21. When a photovoltammogram is done, see inset (b) of Fig. 5, it is observed that the appearance of a cathodic photocurrent after the C2Zn-Se process has reached its maximum development. This photocurrent is not observable when only the H2SeO3 is present in the solution (result not shown) and from is direction a p-type conductivity response could be inferred.
| ||Fig. 5. Cyclic voltammogram (a) and cyclic photovoltammogram (inset figure (b)) for a gold electrode in 0.2 M Na2SO4 electrolyte (pH=1.5) containing 2 mM H2SeO3 and 0.05 M ZnSO4. In the inset, the vertical arrow denotes the direction of the photoresponse, which was always cathodic in this case. Potential scan rate 10 mV.s-1.|
The appearance of a photocurrent indicates that in presence of Zn(II) a semiconducting phase is forming on the gold electrode, probably ZnSe. However, in agreement with Kröger theory27 , the formation of this semiconducting phase occurs after Se0 deposition. In this way, the formation of the binary compound ZnSe would occur in accordance with the following sequence: first, Seo formed via reaction (2) undergoes a subsequent electrochemical reduction that proceeds via reaction
which is assisted by the free energy of formation of the ZnSe (D Gº = -137 KJ mol-1). Then, the deposition of the binary compound is displaced to a more positive potential than that for elemental Zn alone.
The stripping is attributed to both, Se0 (as happened in the previous case, see reversal of Equation (2)), and ZnSe oxidation, explaining thus the charge increasing of A1Zn-Se as compared to the oxidation in the absence of Zn(II):
In fact, as is shown in Figure 6, the deconvolution of peak A1Zn-Se supports this assumption.
| ||Fig. 6. Deconvolution of anodic peak A1ZnSe showing the contributions of Se and ZnSe stripping to the overall anodic charge.|
As Zn(II) concentration is increased, the C2Zn-Se process diminishes in intensity. This is possibly due to the ZnSe increases in the electrodeposits, which in turn increases the resistance for the charge transfer at the electrode/solution interface.
A similar general behaviour regarding the addition of Zn(II) to Se(IV) solution is observed when adding Zn(II) to HTeO2+ solution (see Fig. 7). However, a new reduction process (C3Zn-Te) and two new oxidation processes (A2Zn-Te and A3Zn-Te) are observed.
| ||Fig. 7. Cyclic voltammogram (a) and cyclic photovoltammogram (inset figure (b)) for a glassy carbon electrode in 0.5 M Na2SO4 electrolyte (pH=1.04) containing 0.4 mM HTeO2+ and 0.05 M ZnSO4. In the inset, the vertical arrow denotes the direction of the photoresponse, which was always cathodic in this case. Potential scan rate 10 mV.s-1.|
C3Zn-Te corresponds to Zn2+ electroreduction and A3Zn-Te to zinc stripping. It is observed that the charge of A3Zn-Te is lower than that of C3Zn-Te, due to the chemical reaction between the electrodeposited metallic Zn and the strong acid medium. The A2Zn-Te process, however, does not have a clear origin, although it is believed that could be associated with the ZnTe oxidation.
The inset (b) of Fig. 7 shows the photovoltammogram for a glassy carbon electrode in the Zn(II)+Te(IV) electrodeposition solution. As in the case of selenium , a photocurrent is also observed after the Te0 has been deposited, something that is not observed when the experiment is carried out in the absence of Zn(II). The semiconducting phase responsible for the photocurrent is believed to be ZnTe formed in accordance with the following equation:
where Teo is previously formed via reaction (3).
Following Krögers theory reaction (9) is assisted by the ZnTe formation free energy (ZnTe = -141.6 KJ mol-1). From the direction of the photocurrent a p-type response could be inferred. When the concentration of Zn(II) in solution is increased, the C2Zn-Te process gradually disappears, due to the greater quantity of ZnTe formed, and to the lesser quantity of Te0 to be reduced.
3.2. Zn-Se and Zn-Te film growth.
Several set of zinc chalcogenide thin films were prepared for various cathodic potentials, ranging between -0.8 V to -1.0 V vs. Ag/AgCl and -0.6 to -0.8 V vs. Ag/AgCl for the ZnSe and ZnTe films respectively. All the films were typically ca. 0.4 µm thick, estimated from the total charge passed and applying the Faraday law to reactions (2)+(7) and (3)+(9).
Bath temperature was found to affect growth significantly. Increasing bath temperature resulted in higher growth rates, enhancing smoothness and adhesion of the films. Crystallinity of the films also has been improved by higher bath temperatures16, 17. Figure 8 presents the morphology of the deposits, the mean size of the crystallites is about 10µm.
| ||Fig. 8 SEM image of an annealed ZnSe film electrodeposited at -0.900 V vs Ag/AgCl(sat).|
The Zn/chalcogenide electrodeposits ratio was determined by EDS analysis. Practically for the whole potential interval studied, the occurrence of large amounts of elemental selenium was observed in the as-grown Zn+Se samples (Table 1). This elemental selenium probably was in an amorphous phase, as it was not possible to detect it by XRD diffraction (see below). Furthermore, the as-grown films present a reddish color when observed with the naked eye, changing to lemon-yellow after the heat treatment. In fact, for example, Zn/Se stoichiometric ratios of 20/80 and 48/52 for the as-grown and annealed samples respectively, were observed for films deposited at 0.950 V(Table 2).
|E Dep. / V||% Zn||% Se|
|E Dep 0.950 V||% Zn||% Se|
The compositions of the Zn+Te electrodeposits analyzed by EDS were studied in the -0.600 to -0.800 V vs. Ag/AgCl potential range. A Zn/Te stoichiometric ratio of 40/60 relatively independent of the applied potential in the potential region between -0.600 and -0.700 V, has been observed Table (3). Because both Zn and Te are not soluble into the other, the only compound formed should be ZnTe, with some tellurium in excess. At potentials more negative than -0.700 V the zinc content in the deposits increases; i.e. at -0.800 V Zn/Te composition ratios of 55/45 were obtained. In general, the ZnTe layers were brick-red to greyish brown, and well adding to the substrate. Figure 9 shows the morphology of the deposits whose size is very similar to that found for ZnSe films.
E Dep. / V
|% Zn||% Se|
| ||Fig 9. SEM image of an annealed ZnTe film electrodeposited at -0.800 Vvs Ag/AgCl(sat)|
3.3. Structural characterization.
X-ray diffraction studies were done in order to identify the crystallinity and phases of the thin films. Figure 10 a shows the XRD pattern obtained for a typical Zn+Se as-grown film deposited at 0.900 V. The JCPDS pattern of ZnSe stilleite28 is also reported for comparison in Fig. 10 b. The diffraction peaks at 2q = 27.4o, 45.4o and 53.9o are attributed to the (111), (220) and (311) cubic ZnSe planes, respectively. As can be seen, in comparison with the JCPDS pattern, apart from the titanium substrate, no other compound is detected. Raising the deposition potential further to -1.050 V, however, resulted in the disappearance of the ZnSe diffraction peaks. Only those corresponding to the substrate can be seen (results not shown).
| ||Fig. 10. (a) X-ray diffraction spectra for an as-grown ZnSe thin film deposited at -0.900 V vs. Ag/AgCl(sat) and (b) JCPDS pattern for the cubic ZnSe stilleite structure. The film thickness is about 0.4 _. (__) Peaks due to ZnSe; the others are due to titanium|
In order to improve the crystallinity of the films and to eliminate the eventual excess of elemental selenium, some of the samples were annealed at 300 oC during 30 minutes, under argon atmosphere7,11. However, XRD patterns of annealed films are very similar to the as-grown ones, indicating that a good morphology was obtained in the as-grown ZnSe films.
An estimation of the mean size of the crystallites in the as-grown ZnSe polycrystalline films was obtained from the broadening of the (111) X-ray diffraction peak according to the Scherrer formula29. The thinnest titanium peak was used as the measure of instrumental broadening. The mean crystallite size was found to be on the order of 9 µm similar to values obtained from SEM micrographs.
The X-ray diffraction pattern of a typical as-grown Zn+Te film deposited at -0.600 V on a titanium substrate is shown in Fig. 11 a. Its thickness is approximately 0.4 µm. The sharp (111) peak at 2q =25.2º establishes the formation of a highly oriented cubic phase ZnTe film, however diffractions peaks at 41.6o and 49.5o corresponding to the (220) and (311) planes were also seen. The JCPDS pattern for cubic ZnTe30 is also shown in Fig. 11 b, for comparison. ZnTe is known to possess either a cubic or hexagonal structure like ZnS and ZnSe. However, in our case no diffraction peaks corresponding to a hexagonal phase (e.g. (103) diffraction plane peak31) were observed , as was the case in electrodeposited ZnTe from a nonaqueous bath32. Furthermore, no diffraction peaks corresponding to Te can be observed. The crystallite sizes estimated from the X-ray diffraction peak width (see above) appear to be on the order of 400 Å.
| ||Fig. 11. (a) X-ray diffractogram of as-electrodeposited ZnTe layer at -0.600 V vs. Ag/AgCl(sat) and (b) JCPDS pattern for the cubic ZnTe structure. The film thickness is about 0.4 µm. ()Peaks due to ZnTe; the others are due to titanium.|
3.4 Optical characterization.
We have studied the electronic band-edge of these samples from the reflectance spectra, as their opaque substrate did not allowed us to determine the optical transmission. We have tried to transfer the film to glass substrates using a transparent glue33, but the electrodeposited films were so well adhered to the titanium substrates that the gluing method did not work: the films remained firmly adhered to the substrate instead of being transferred to the glass. We have also tried to grow the films onto ITO coated glass, but selenium excess appeared on such samples34. Thus, we have determined the band-edge properties just from the reflectance.
The upper dotted curves of Fig. 12 are the reflectance spectra R(l) for ZnSe (a) and ZnTe (b). For both cases, the reflectance in the low wavelength (high photon energy) region of the spectrum is smaller than in the high wavelength (low photon energy) region, although these reflectance spectra decrease also for wavelengths well into the infrared region35. In addition, both spectra show a sharp edge in the middle of the visible region (close to 500 nm). These edges were always present for different specular or diffused configurations, between the incident and detected light beams. For ZnSe the edge is between 460 to 520 nm, and for ZnTe is between 520 to 580 nm; i. e. for ZnSe is in lower wavelengths (higher energies) than in ZnTe. These edges in the reflectance give the samples their characteristic colorations: yellow for ZnSe15, 35 and brown for ZnTe16. In some cases, for very thin films, interference fringes appear altering these spectra and the sample color; but such fringes are not remarkably present in the spectra of Fig.12
| ||Fig. 12 Optical Reflectance Spectra R(l) (upper dotted curves) for ZnSe (a) and ZnTe (b) films deposited onto titanium. Both curves have sharp edges near 500 nm. First Derivatives of the Reflectance dR(l)/dl (lower full curves) have peaks at lg = 470 nm (for ZnSe) and 550 nm (for ZnTe). These values correspond with energy gaps at Eg = 2.64 and 2.27 eV, respectively.|
The sharp edges indicate that they are originated in direct gap transition. For determining the energy gaps associated with them we have studied the first derivative dR(l)/dl of the reflectance spectra against the wavelength l. These derivatives are shown in Fig. 12 (full lines) and have prominent peaks in the onset of the sharp edges of the reflectance: the peak positions are at 470 nm (for ZnSe) and 547 nm (for ZnTe). As we show in the Appendix, the energies corresponding to these peak positions (2.64 eV and 2.27 eV for ZnSe and ZnTe, respectively) give us a direct band-gap energy measurement. Both values are very close with the most accepted room temperature ones: 2.68 eV for ZnSe36, 37 , and 2.26 eV36, 38 for ZnTe.
We have also obtained these energy gaps using a different method: we have deduced them from the absorption coefficient estimated from the same reflectance spectra. In this kind of samples, the diffused reflectance may be assumed essentially proportional to the transmittance through the thin film. This is because the main contribution to the diffused light comes from the backscattering at the opaque substrate. The backscattered light goes twice through the semiconducting thin film. Assuming the substrate is optically neutral (and highly reflective) in the region of interest, the reflectance spectra will be proportional to the transmittance and we can essentially write, in this approximation39:
where is the absorption coefficient spectrum against photon energy hn , l is the film thickness, is the diffuse reflectance spectrum in any arbitrary direction, and Rmax (Rmin) is the maximum (minimum) value of the reflectance in the vicinity of the sharp edges under study. The presence of Rmax and Rmin in the previous equations is for normalization reasons.
Fig. 13a shows the absorption coefficient spectra calculated using Eq. 10. The absorption increases abruptly for both spectra close to the expected values given by the first derivative method. Fig. 13 b shows plots of (a (hv)* hv)2 vs. hv, and linear fittings of them in the region corresponding to the sharp edges. Extrapolating from these linear plots to their zero crossing, we obtain two new estimations to the energy gaps: 2.58 eV for ZnSe and 2.24 for ZnTe. These values, although rather smaller than the previous ones are still quite close to the accepted ones.
Fig. 13 . Optical Absorption Spectra for ZnSe and ZnTe films deposited onto titanium. (a) Absorption Coefficient a against photon energy (hv) obtained from the reflectance spectra using Eq. (10). (b) Direct Energy Gap determination extrapolating from linear fitting of (a * hv)2 against h?. Note that the zero reference line is up shifted for ZnTe spectra in both figures.
The difference between the energy gap values found by one method and the other are inherent of each method. Effectively, it is usually found that the determination of the absorption coefficient from reflectance measurements give rise to underestimations of energy gap values with uncertainties on the order of 0.1 eV40, 41. Then the values given by this last method are quite reasonable, while better results are expected by the first derivative method.
The bigger discrepancy is found for ZnSe, for which the value found by the (a (hv)* hv)2 vs. hv method is 0.06 eV smaller than the value found from the first derivative method. We see from Fig. 13b that the linear region of the data in this plot, although sharp, spreads over a narrow region of the spectrum. This is in agreement with the wide peak of the first derivative method (Fig. 12a), although the peak agrees quite well with the accepted value for ZnSe. Fig. 13a also shows that, for energies smaller than the 2.58 eV energy gap value, the ZnSe absorption coefficient increases smoothly. This dependence follow closely a linear dependence in a (a (hv)* hv)1/2 vs. hv Tauc plot42. The zero crossing of a linear fitting of this plot gives evidence of an indirect gap in 1.83 eV. This may correspond with the presence of trigonal selenium, which has an indirect gap in 1.85 eV43. The presence of excess selenium was already reported by the authors for ZnSe grown onto ITO/glass films34.
For ZnTe, instead, both methods give a lesser discrepancy of just 0.03 eV. Moreover, the linear plot of (a (hv)* hv)2 vs. hv. for ZnTe on Fig. 13 b, agrees in a far wider region than for ZnSe. This is in agreement with the narrower peak of the first derivative in Fig.12 b, and it is important because it supports the first derivative method for this case. Some oscillations can be seen in both the reflectance spectrum and its first derivative, which may be due to interference effects. The presence of interferences peaks would invalidate the first derivative method, but the good agreement between both methods denies any uncertainties due to this effect.
The electrochemical behaviour of the Zn/X (X= Se,Te) system and their corresponding precursors has been characterized through a detailed voltammetric and photovoltametry study. From this study we got useful information intended to select experimental conditions in order to prepare ZnX thin films by one step electrodeposition at constant potential.
Regarding films composition, as has been previously reported by other authors, EDS films characterization showed that the one step electrodeposition of ZnSe and ZnTe in acid media gave as grown films that always contained an excess of elemental chalcogenide that can be eliminated by annealing. However, XRD analysis revealed a good crystallinity for the as grown films, indicating that the experimental conditions we employed were appropriates for achieving good quality films.
The quality of the films was also evidenced after the results obtained in their optical characterization. In fact, optical reflectance characterization of ZnSe and ZnTe samples grown on titanium gave us direct band gaps values of 2.64 eV and 2.27 eV, respectively, that were obtained from the first derivative method. While the same parameters obtained from the absorption coefficient deduced from reflectance gave us smaller values, 2.58 eV and 2.24 eV respectively, the first method gave closer results with the accepted room temperature values of 2.68 (for ZnSe) and 2.26 eV (for ZnTe). This is in agreement with the statement that the first derivative method has less uncertainty than that one arising from absorption obtained from reflectance, that is on the order of 0.1 eV40, 41.
This work was supported by FONDECYT (Grant Nº 8000022) and DI-UCV (Grant 125-721). R. E. M. and E. A. D. thank CSIC Universidad de la República and PEDECIBA-Física (Uruguay) for financial support.
The dependence of the reflectance R(l), with the wavelength l, in this kind of samples, comes from its dependence f with the wavelength dependent refraction index n(l) and extinction coefficient k(l). Thus:
where the extinction coefficient can be expressed in terms of the absorption coefficient a(l):
The mathematical details of f are due to the specific experimental conditions. For example, for normal incidence of light over a flat interface we will have44:
The first derivative of the reflectance as a function of the wavelength l has two distinguishable terms, one depending on the variation of the index of refraction, and the other on the extinction coefficient:
The most relevant of these dependences is the one in the extinction coefficient:
Near a direct or indirect energy band gap, the absorption coefficient follows a potential dependence with the photon energy , measured from the energy gap Eg we want to determine:
where a = ½ (2) for a direct (indirect) energy band gap, and A is just a proportionality constant. As:
then (for hv > Eg):
We expect, for the wavelength that corresponds with the energy gap :
for both direct and indirect energy band gaps, while:
This would indicate the presence of a sharp peak in the derivative of (unless ) for a direct band-gap, while this would not be so for an indirect band-gap.
Moreover, the Kramer-Krönig relation let us write:
where the P-integration means Cauchy principal value; i. e. the integration should be carefully done between the hv value:
Taking in consideration the absorption coefficient given by Eq. (A-6) we will have that:
For hg < Eg the integration is well defined for a direct band gap (a = ½). For an indirect band gap (a = 2) problems may arise with the limit for E®¥. Nevertheless, in this case, we must remember the dependence of given by (A-6) is verified just near the gap, while the function must be convenient smooth for . Then we can calculate the derivative for both cases of the refraction index, at least in this region hg< Eg:
In the limit for hg smaller than (but close to) Eg we will have:
Then for a direct band-gap (a = ½):
while for an indirect band-gap (a = 2)
Finally, the index of refraction also verifies Eq. (A-8) as the extinction index, so its derivative, and the first term in Eq. (A-4), diverges for an direct band-gap, while it is not necessary the case for an indirect band-gap.
In practice, other contributions to the reflectance (impurity states, crystal disorder, etc.) appear that will prevent a divergence on the reflectance. However, unless some very particular conditions are fulfilled, the reflectance will have a peak near a direct energy gap, being the position of the peak a measurement of the energy gap.
The derivations presented in this appendix, to justify the validity of the first derivative method against the absorption coefficient derived from reflectance follows similar arguments that are used in modulation reflectance measurements45, 46.
1. K. Singh and J. P. Rai. J. Mater. Sci. 22, (1987) 132. [ Links ]
2. K. Singh and J. P. Rai. Indian J. Chem. 28 A, (1989) 557. [ Links ]
3. J. P. Rai. Solar Energy Mater. Solar Cells. 30, (1993) 119. [ Links ]
4. J. P. Rai and K. Singh. Indian J. Chem. 32 A, (1993) 376. [ Links ]
5. K. Singh and R. K. Pathak. Electrochim. Acta. 36, (1994) 2693. [ Links ]
6. K. Singh, A. K. Shukla and M. R. Tanveer. Indian J. Chem. Technol. 3, (1996) 112. [ Links ]
7. C. Natarajan, M. Sharon, C. Lévy-Clément and M. Neumann-Spallart. Thin Solid Films. 237, (1994) 118. [ Links ]
8. K. K. Mishra and K. Rajeshwar. J. Electroanal. Chem. 273, (1989) 169. [ Links ]
9. C. Wei and K. Rajeshwar. J. Electrochem. Soc. 139, (1992) L40. [ Links ]
10. R. Chandramohan, C. Sanjeeviraja and T. Mahalingam. Phys. Stat. Sol. 163, (1997) R11. [ Links ]
11. A. P. Samantilleke, M. H. Boyle, J. Young and I. M. Dharmadasa. J. Mater. Sci.: Mater. Electron. 9, (1998) 231. [ Links ]
12. I. M. Dharmadasa, A. P. Samantilleke, J. Young, M. H. Boyle, R. Bacewicz and Wolska. J. Mater. Sci.: Mater. Electron. 10, (1999) 441. [ Links ]
13. G. Pezzatini, S. Caporali, M. Innocenti and M. L. Foresti. J. Electroanal. Chem. 475, (1999) 164. [ Links ]
14. C, Natarajan, H. Matsumoto and G. Nogami. Bull. Electrochem. 13, (1997) 123. [ Links ]
15. D. Gal and G. Hodes. J. Electrochem. Soc. 147, (2000) 1825. [ Links ]
16. M. Neumann-Spallart and C. Königstein. Thin Solid Films. 265, (1995) 33. [ Links ]
17. C. Königstein and M. Neumann-Spallart. J. Electrochem. Soc. 145, (1998) 337. [ Links ]
18. B. Bozzini, M. A. Baker, P. L. Cavallotti, E. Cerri and Lenardi. Thin Solid Films. 361 362, (2000) 388. [ Links ]
19. A. B. Kashyout, A. S. Aricò, P. L. Antonucci, F. A. Mohamed and V. Antonucci. Mater. Chem. Phys. 51, (1997) 130. [ Links ]
20. N. B. Chaudre, R. Jayakrishnan, J. P. Nair and R. K. Pandey. Semicond. Sci. Technol. 12, (1997) 1171. [ Links ]
21. K.K. Mishra, K. Rajeshwar, J. Electroanal. Chem. 271, (1989) 279. [ Links ]
22. G. Riveros, R. Henriquez, R. Córdova, E. A. Dalchiele, R. Schrebler, H. Gómez, J. Electroanal. Chem. 504, (2001) 160. [ Links ]
23. A. J. Bard, L. R. Faulkner, Electrochemical Methods: Fundamentals and Applications, John Wiley and Sons Inc., New York NY, 1980, p. 288. [ Links ]
24. R. Henriquez, License Thesis , Universidad Católica de Valparaíso (1999). [ Links ]
25. M. Pourbaix, Atlas D'Equilibres Electrochimiques, Gauthier-Villars, Paris, 1963. [ Links ]
26. E. Mori, C. K. Baker, R. Reynolds, K. Rajeshwar, J. Electroanal. Chem. 252, (1988) 441. [ Links ]
27. F. A. Kröger, J. Electrochem. Soc. 125, (1978) 2028. [ Links ]
28. JCPDS ZnSe, File 37-1463. [ Links ]
29. H. P. Klug, L. E. Alexander, X-ray Diffraction Procedures: For Polycrystalline and Amorphous Materials, John Wiley and Sons Inc., New York NY, 1954, Chap. 9. [ Links ]
30. JCPDS ZnTe, File 15-746. [ Links ]
31. JCPDS ZnTe, File 19-1482. [ Links ]
34. G. Riveros, H. Gómez, R. Henriquez, R. Schrebler, R. E. Marotti, E. A. Dalchiele, Solar Energy Mater. Solar Cells. 70, (2001) 255. [ Links ]
35. J. M. Doña, J. Herrero, J. Electrochem. Soc. 142, (1995) 764. [ Links ]
36. R. N. Bhargava (Ed.), Properties of Wide Bandgap II-VI Semiconductors, EMIS Datareviews Series No. 17, INSPEC, London, 1997. [ Links ]
37. S. Adachi, T. Taguchi, Phys. Rev. B 43 ,(1991) 9569. [ Links ]
38. K. Sato, S. Adachi, J. Appl. Phys. 73, (1993) 926. [ Links ]
39. M. Husain, S. Kumar, Solar Energy Mater. Solar Cells (to be published). [ Links ]
40. M. Sotelo-Lerma, R. A. Zingaro, S. J. Castillo, J. Organomet. Chem. 623, (2001) 81. [ Links ]
41. D. Gal and G. Hodes, J. Electrochem. Soc. 147, (2000) 1825. [ Links ]
42. J. Tauc, Mat. Res. Bull. 5, (1970) 721. [ Links ]
43. R. A. Zingaro, W. C. Cooper (Eds.), Selenium, Van Nostrand Reinhold, New York, 1974. [ Links ]
44. J. I. Pankove, Optical Processes in Semiconductors, Dover Publications, Inc., New York, 1975. [ Links ]
45. M. Cardona, Modulation Spectroscopy, Solid State Physics, Supplement 11, Academic Press, Inc., New York, 1969. [ Links ]
46. F. H. Pollak, H. Shen, Mater. Sci. Eng. R10, (1993), 275 [ Links ]