Electrical properties of CdTe films prepared by close-spaced sublimation with screen-printed source layers

Transcription

1 Electrical properties of CdTe films prepared by close-spaced sublimation with screen-printed source layers Gil Yong Chung, Sung Chan Park, Kurn Cho, and Byung Tae Ahna) Department of Materials Science and Engineering, Korea Advanced nstitute of Science and Technology, 373-l Koosung-dong, Yusung-gu, Taejon , Korea (Received 20 December 1994; accepted for publication 12 July 1995) CdTe films have been deposited by a close-spaced sublimation process with screen-printed CdTe layers as a new source. The source-layers were fabricated by screen printing and sintering slurries consisting of propylene glycol, CdC12 and either CdTe powder or (Cd+Te) powder. When CdTe powder was used as a starting material for the source-layer, the electrical resistivity of the CdTe film deposited in O2 was about one-tenth lower than that of the film deposited in He. AES, EDS and PXE analysis showed that the Cd content in the CdTe films deposited in O2 was smaller than that in the CdTe films deposited in He. Especially, no oxygen element was detected in the CdTe films deposited in Oz. t turned out that the sublimated Cd(g) and CdTe(g) from the source-layer formed cadmium oxides in O2 and as a result the overall composition of vapor source became more Cd-deficient. The CdTe film with less Cd content increased cadmium vacancy defects (V&Cl+ and V&) and hole concentrations. As the Cme ratio in the starting (Cd+Te) powder decreased, the tesistivity of the CdTe films decreased and reached at a constant value of about 3~10~ a. cm, regardless of deposition atmosphere. n addition, the resistivity decrease by O2 treatment was diminished when the Cd/Te ratio was below 0.7, where the composition of CdTe film might be limited by solid solubility (Cdc,2Te1,0). The above results indicated that the resistivity decrease of CdTe films deposited in O2 was not due to the effect of previously-known oxygen doping but due to the effect of Cd/Te compositional change American nstitute of Physics. 1. NTRODUCTON Cadmium telluride is currently one of the most promising materials for high efficiency, low-cost thin film solar cells. Polycrystalline thin films of CdTe have been deposited by several techniques such as vacuum evaporation, electrolytic deposition, spray pyrolysis, atomic layer epitaxy, and close-spaced sublimation (CSS). - Especially, Chu et al. reported an efficiency of 15.8% from the CdTe solar cells prepared by CSS process.6 CSS process is a well known method with high deposition rate and low fabrication cost. However, preparing source materials for CSS process is difficult and cumbersome. There are two ways to achieve source materials for CSS. Fit one is forming a source block by using a mixture of Cd and Te powder or CdTe powder7 and second one is growing single crystal CdTe.* Forming a source block requires repetition of high temperature (1000 C), which cause problems in reproducibility and scale-up potential of the process. CdTe source block has to be treated chemically after each deposition process because of the compositional change at the surface. Growing single crystal is a difficult process and is expensive. Therefore, developing a convenient and reproducible source for CSS process is a task to be solved. n this study, we propose a screen-printed CdTe layer on glass substrate as a new source, which can lower preparation cost for CSS source and provides a reproducible composition in CdTe films. Screen-printed CdTe layers have been previously applied to sintered CdSKdTe solar cells. Screenprinted layer can be easily prepared from the mixture of Cd )Elcctronic mail: btahn@cais.kaist.ac.kr and Te powder instead of expensive CdTe powder. * We report the electrical properties of CdTe films prepared by CSS process using screen-printed CdTe layer as a source. The effects of oxygen atmosphere during CSS process and Cd/Te ratio in the starting (Cd+Te) powder on the electrical property of the CdTe films are focused.. EXPERMENTAL PROCEDURE The experimental set up for the CSS process is shown in Figure 1. The substrate temperature and source temperature ranged from 500 to 600 C and from 600 to 700 C, respectively. CdTe films were deposited either in He or in O2 with a pressure ranging from 1 to 200 Tom Substrates included Coming 7059 glass, sintered CdS films, CdS films prepared by chemical bath deposition (CBD), and alumina plate. All the electrical measurement and compositional analysis were carried out with the CdTe films deposited on alumina substrate. This study explored two types of screen-printed sourcelayers: type A source from CdTe powder and type B source from the mixture of Cd and Te powder. The preparation procedure of type A source-layer is as follows. A slurry consisting of CdTe powder with 5 wt% of CdC& and an appropriate amount propylene glycol (PG) was prepared by mixing and dispersing with a mortar and pestle. The slurry was coated on 2.5 cm x 4.0 cm borosilicate glass substrates using a screen printer with a 150-mesh silk screen to obtain CdTe source layer with a thickness of 15 pm. After the CdTe layer dried, the CdTe layer was sintered at 625 C for 1 h in nitrogen. The preparation procedure of type B source-layer is slightly different from type A. The starting slurry consisting of (Cd J. Appt. Phys. 78 (9), 1 November /95/78(9)/5493/8/$6.00 Q 1995 American nstitute of Physics 5493

2 1. Halogen Lamp 3. Graphite Plate 5. Source 7. Gas Output 9. Gas nput 11. Thermocouple 2. Quartz Holder 4. Spacer 6. Substrate 8. Vacuum Gauge 10. Flange Holder FG. 1. Schematic diagram of the appamtus for the deposition of CdTe films by &se spxrd sublima.tion process. +Te) mixture powder with an appropriate Cd/Te ratio was prepared and coated. Then the dried CdTe layers were pushed slowly into a tube furnace at the rate of 14 C/min up to 500 C in nitrogen. A 1 h sintering procedure conducted for the type A source was omitted. XRD analysis showed that the only CdTe phase existed in the prepared type B sourcelayer even though the st*zrting Cd/Te ratio in the slurry varied from 1 to The crystal structure and microstructure of the CdTe films were investigated by s-ray diffraction (XRD) and scanning electron microscopy (SEM. The resistivity of CdTe films was measured by a four point probe method using a Keithlcy current source and nanovoltmeter. A carbon paste electrde was used for electrical ohmic constant. The thickness of the CdTe films was measured by u-step instrument. Chemical analysis was investigated by particle induced x-ray emission U? SE), auger electron spectroscopy (AES) and energy dispersive spectroscopy (EDS) with a reference of CdTe.. RESULTS AND DSCUSSON Figure 2 shows the SEM micrographs of CdTe films deposit& in 20 Torr He atmosphere with type A source on various sutistrates. The source temperature and substrate temperature were 700 C and 600 C, respectively. Figure 2aj shows that the CdTe islands form on glass substrate be&use nucleation sites are rarely available on the smooth surface of amorphous glass. Figure 2(b) shows a continuous 10 pm thick CdTe film with a grain size of about 20,LN~ on a wintered CdS substrate which has a grain size of about 10 p.m. On the other hand, the grain size of CdTe film on a CBD CdS substrate is about a few pm as shown in Figure 2(c), where the CUD CdS substrate has a grain size of about 100 nm. The grain size of the CdTe. film deposited on a CBD CdS substrate is smaller than that of the film deposited on a sintered CdS substrate because the larger grain boundary area in CBD CdS films serves as a nucleation site. Figure 3 shows the SEM micrographs and XRD patterns of the CdTe films deposited on alumina substrate with type A FG. 2. SEM micrographs of CdTe films deposited at substrate tenqxrature of 600 C and in 20 Ton He with type A source on (a) coming 7059 glass, (b) sintered CdS substrate, (c) CRD CdS substrate. source at the substrate temperature of 600 C in 1 Torr 02 and 20 Torr He. We have chosen an alumina plate as a substrate to measure the electrical rrsistivity of the CdTe films itself. The micrographs show that the CdTe films are continu- i s f ~;-, ~-: --.-:,. -. &L-$--& & z: i. & - -io &..J N FG. 3, SEM micrographs (a,b) and XRD patterns ic,d) of CdTe filmr deposited on alumina substrate at substrate temperature of 600 C with type A source in He (a,c) and O2 (b,d) J. Appl. Phys., Vol. 78, No. 9, 1 November 1995 Chung et al.

3 -o- -o- He (a) 0, (b) -j.,, *.. t Substrate temperature ( C) PG. 4. Electrical resistivity of CdTe films deposited with type A source in He (a) and in 0s (b) as a function of substrate temperature OOO/T( 1000/K) PG. 5. Temperature dependence of resistivity of CdTe films deposited on alumina at substrate temperature of 600 C with type A source in He (a) and in 0, (b). ous and the morphologies are similar to each other. The grain size of the films is about l-3 pm. The XRD patterns of the CdTe films show a very strong intensity of the (111) plane, which has the lowest surface energy in zinc blende structure. From the SEM micrographs and the XRD patterns, the films deposited in O2 is slightly less (111) textured along the thickness direction. Figure 4 shows the electrical resistivity of CdTe films deposited on alumina substrate as a function of substrate temperature. The resistivity does not change significantly as the substrate temperature varies. On the other hand, the resistivity significantly changes as the deposition atmosphere varies. The resistivity values of CdTe films deposited in O2 and in He are of the order of lo6 fl.crn and lo5 O-cm, respectively. Note that the resistivity of CdTe in O2 is about an order smaller than that of CdTe in He. From the previous micrographs, it can be seen that the large drop in the resistivity of CdTe films with various gas atmosphere is not due to the difference of microstructure in the CdTe films. Figure 5 shows the temperature dependence of resistivity of the CdTe films. The CdTe films show thermally activated resistivity, i.e., where p. is the pre-exponential factor and EA the resistivity activation energy. The activation energy of CdTe films is about 0.15 ev below 280 K and about 0.63 ev above 280 K regardless of gas atmosphere. n a single crystal CdTe, the values of 0.15 ev and 0.63 ev are ionization energies of cadmium vacancy-donor complex defect (Vt, Cl+) * and cadmium vacancy (V&),13 respectively. Even though the CdTe film in this study is polycrystalline, the measured activation energies are very close to those of the single crystal. Both (Vt,Cl ) and V& are hole generators in CdTe crystal. Therefore, it can be said that the (V&Cl+) defect is an effective hole generator below 280 K and the V& defect as well as (V&Cl ) act as effective hole generators above 280 K. Since the resistivities of CdTe films prepared in O2 and in He are almost parallel with an order difference below 280 K. t can be said that the CdTe film deposited in O2 has about 10 times higher concentrations of (V&Cl ) and holes compared to that deposited in He. Note that (V&Cl+) defects are fully ionized above 280 K. To understand the resistivity variation with various gas atmospheres, we have analyzed the composition of the CdTe films. Figure 6 shows the AES, EDS and PXE spectra of the CdTe films prepared in O2 and He. The calculated relative Cd content in the CdTe films is shown in the upper-left side T 55.E? c s 0 3 &.a al t E P >i.., Kinetic energy (ev) Electron energy (kev) Wavelength Kinetic energy (ev) Electron energy (kev) Wavelength FG. 6. AES (a,b), EDS (c, d) and PME (e, f) spectra of CdTe films deposited at substrate temperature of 600 C with type A source in He (a,c,e) and in 0, (b,d,f). The relative Cd content is shown in the upper left comer of each figure. J. Appl. Phys., Vol. 78, No. 9, 1 November 1995 Chung et al. 5495

4 3 s El E P 2 cl P 0 CdTe 0 Cd0 l CdTeO, (b) 0.81 b A... (c) 0.65 b A... (d) FG. 7. XRD patterns of powders formed on CdTe source layer during CSS process in O2 in a closed system. O FG. 8. XRD patterns of CdTe films as a function of CdTe ratio in starting (Cd+Te) powder. of each figure. The Cd content of the CdTe films varies depending on the analytical method due to the difficulty in absolute quantification. However, it can be clearly seen that regardless of the analytical methods the Cd content in the films deposited in O2 is always lower than that in the films deposited in He. Hsu et a1.l4 reported that CdTe films deposited by the CSS process in O2 atmosphere have a higher oxygen concentration of about 102 /cm3 by R spectroscopy measurement and suggested that these oxygen ions act as a carrier source in CdTe films. However, in our analysis no oxygen ions have been detected in the CdTe films, even in the PXE method which has a very low detection limit about ppm order. nstead, the Cd concentration in the films changed as the gas atmosphere varied. Since the experimental procedure of Hsu et al4 is similar to that of ours, it is likely that the combination of PXE, EDS and AES may provide more accurate composition analysis. But, we do not know the exact experiment conducted by Hsu et al. at this point. There might be a subtle difference in addition to analytical tools. This compositional change might be explained from the thermodynamical point of view. Thermodynamically, the Cd atom tends to react easily with oxygen to form a cadmium oxide at the deposition condition in O2 atmosphere, while the Te atom is less reactive with oxygen compared to the Cd atom. Therefore we could assume that Cd or CdTe vapor sublimated from source materials reacts with oxygen and forms a kind of cadmium oxide. Thus, the vapor source is more Cd-deficient in O2 atmosphere and as a result the Cd content within the deposited film becomes smaller compared to that in the source layer. To verify the above assumption of cadmium oxide formation, we made a closed system of CSS chamber during the deposition process in O2 atmosphere and then cooled the chamber down to room temperature to convert gas phase materials formed during deposition in 02 to solid phase. We found powders on source layer after cooling down the closed system. Figure 7 shows the XRD pattern of these particles which turn out to be CdTe, Cd0 and CdTe03. Therefore, this XRD analysis verified the formation of cadmium oxides such as Cd0 and CdTe03 in an 02 atmosphere. As a result, the CdTe film prepared in O2 shows more Cd-deficient than that prepared in He. From the above composition analysis and thermodynamic analysis, it can be said that the resistivity decrease in an O2 atmosphere is not due to the oxygen doping but due to the decrease in Cd contents. t is not known why cadmium oxide cannot be detected in the deposited CdTe films. The sticking coefficient of cadmium oxide might be too small or both Cd0 and CdTe03 may be volatile and swept away by gas flow during the CSS process. Since the stoichiometric change in CdTe film is the main factor for the resistivity variation, we intentionally prepared the CdTe films with various stoichiometries by controlling the Cd/Te ratio in the starting slurry and analyzed their microstructure, structure, and electrical resistivity. Figure 8 shows the XRD patterns of the CdTe films as a function of CWTe ratio in the starting slurry. The films were deposited on a 600 C alumina substrate in 20 Ton He. Note that no peak of pure Te was detected even though the Cd/Te ratio in source layer decreased down to t is likely that the pure Te is evaporated away during film deposition. All the films show a predominantly (111) oriented texture with no clear difference in the shape of the XRD patterns. The grain size of the films taken by SEM micrographs was about 2 pm regardless of the CdKe ratio and the size distribution of the films prepared by type B source was apparently more uniform than that of the films prepared by type A source. The CdTe films deposited in O2 had the same results. t can be said that there was little difference in the crystal structure and microstructure of CdTe films deposited with type B source even though the Cme ratio in the starting slurry widely varied. Figure 9 shows the electrical resistivity of the CdTe films as a function of Cd/Te ratio in the starting slurry. The substrate temperature and source temperature were 600 C and 650 C, respectively. The Cd/Te ratio in the starting slurry varies from 0.53 to 1.0. The resistivity of CdTe films decrease as the CdiTe ratio decreases down to 0.7 and then it maintains a constant value of 3~10~ fi -cm with further de J. Appl. Phys., Vol. 78, No. 9, 1 November 1995 Chung et al.

5 -a- He -o- 0, 1 s i m 8.48 iii.- a= /.,-: 9 O------O -w- He -o- 0, Cd/Te ratio in (Cd+Te) powder FG. 9. Electrical resistivity of CdTe films deposited as a function of Cd/Te ratio in starting (Cd+Te) powder. crease in the Cd/Te ratio. Above the CdR e molar ratio of 0.7, the resistivity of CdTe films deposited in O2 is lower than that of CdTe films deposited in He. But the difference between the resistivity of CdTe films deposited in O2 and that of CdTe films deposited in He becomes smaller as the Cd/Te ratio decreases. Note that the resistivity values of the CdTe films prepared in He and O2 are essentially the same when the CdRe ratio is below 0.7. t is likely that the stoichiometry of the CdTe films does not change if the CdRe ratio is below 0.7. n other words, the composition of CdTe films is fixed by the limit of solid solubility. Figure 10 shows the calculated relative Cd content in the deposited CdTe films, analyzed by EDS as a function of Cd/Te ratio in the starting slurry. When the Cd/Te molar ratio is above 0.7, the Cd content in CdTe films deposited in O2 is lower than that in the CdTe films deposited in He. t is in agreement with Cd content in Figure 6. As the Cd/Te ratio decreases below 0.7, the stoichiometry of CdTe films may be fixed by Te solid solubility, resulting in the same composi- h 44c -m- He -/ E z Cd/Te molar ratio in (Cd+Te) powder FG. 10. Calculated relative Cd content of CdTe films as a function of Cd/Te ratio in starting (Cd+Te) powder by EDS Cd/Te ratio in (Cd+Te) powder FG. l. Calculated c-lattice parameter of CdTe films as a function of CdfTe ratio in starting (Cd+Te) powder. tion and resistivity, regardless of gas environment. t can be seen that the solubility limited stoichiometry of CdTe is CdeszTe,~a from Figure 10. The Cd content shows a similar variation with the logarithmic resistivity of CdTe films. The compositional change with the Cd/Te ratio might be further confirmed by investigating the c-lattice parameter of the CdTe films. Figure 11 shows the c-lattice parameter of the CdTe films as a function of CdRe ratio in the CdTe starting slurry. The lattice parameter change is in very small quantity and it needs a careful analysis. Even though it is less accurate, it can reflect the compositional change of CdTe films within a single phase. V. CONCLUSONS Based on the results of structure, electrical resistivity, and compositional analysis of the CdTe films prepared by a CSS process with screen-printed CdTe layers as a new source materials, the following conclusions can be drawn. (1) The resistivity of the CdTe films deposited with pure CdTe source in O2 gas was lower than that of the CdTe films deposited in He gas. Compositional analysis showed that the resistivity drop resulted from the decrease of Cd/Te ratio that formed cadmium vacancies and holes. Especially, no oxygen element was detected in the CdTe film deposited in O2 by CSS process. (2) The CWTe compositional change in O2 was originated from the Cd-deficiency of vapor source by the oxidation of Cd(g) to Cd0 during CSS process. (3) As the Cd/Te molar ratio in the starting slurry decreased, the resistivity of the CdTe films decreased and then reached at a constant value of 3~10~ 0. cm. The stoichiometry of Cd-deficient composition was limited to CdcszTel,c when the Cd/Te ratio was 0.7, regardless of gas environment. Therefore, we suggested that the CdKe stoichiometry change is the main reason of resistivity drop in O2 atmosphere, instead of the oxygen incorporation as a carrier source. J. Appl. Phys., Vol. 78, No. 9, 1 November 1995 Chung et a/. 5497

6 W. N. Skafarman, R. W. Birkrnire, D. A. Fardig, B. E. McCandless, A. Mondal, J. E. Philips, and R. D. Vanin, Solar Cells 30, 61 (1991). *J. M. Woodcock, A. K. Turner, M. E. Ozsan. and J. G. Summers. Proceedings of the 22nd EEE Photoioltaic Specialists Conference (EEE, New York, 1991), p S. Albright, J. F. Jordan, B. Ackerman, and R. B. Chamberlin, Solar Cells 27, 77 (1989). 4J. Skarp, Y. Koskinen, S. Liidfors, A. Rautiainen, and T. Suntola, Proceedings of the 10th European Photovoltaic Solar Energy Conference (Lisbon, Portugal, 1991), p D. Bonnet, B. Henricks, and H. Richter, Proceedings of the 22nd EEE Photovoltaic Specialists Conference (EEE, New York, 1991), p T. L. Chu, S. S. Chu, C. Ferekides, C. Q. Wu, J. Britt, and C. Wang, J. Appl. Phys. 70, 7608 (1991). T. L. Chu, Solar Cells. 23, 31 (1988). st. C. Anthony, A. L. Fahrenbnrch, M. G. Peters, and R. H. Bube, J. Appl. Phys. 57, 400 (1985). 9Y. K. Jun and H. B. m, J. Electrochem. Sot. 135, 1658 (1988). D. S. Kim 7 S. Y. Kim, B. T. Ahn, and H. B. m, J. Mater. Sci.:Mater. Electron. 5, 17 (1994). J. S. Roh and H. B. m, J. Mater. Sci. 23, 2267 (1988). r*u. S. vanov, V. B. Stopachinskii, and V. A. Chopin, Sov. Phys. Semicond. 5, 83 (1971). 13M R. Lorenz, B. Segall, and H. H. Woodbury, Phys. Rev. 134, 751 (1963). 14T. M. Hsu, R. J. Jih, P. C. Lin, H. Y. Ueng, Y. J. Hsu, and H. L. H Wang, J. Appl. Phys. 59, 3607 (1986) J. Appl. Phys., Vol. 78, No. 9, 1 November 1995 Chung et a/.