ARTICLE IN PRESS

Journal of Crystal Growth 284 (2005) 477485 www.elsevier.com/locate/jcrysgro

Properties of Te-rich cadmium telluride thin lms fabricated by closed space sublimation technique
N. Abbas Shah, A. Ali, Z. Ali, A. Maqsood, A.K.S. Aqili1
Thermal Physics Laboratory, Department of Physics, Quaid-i-Azam University, Islamabad 45320, Pakistan Received 4 May 2005; received in revised form 29 July 2005; accepted 1 August 2005 Communicated by P. Rudolph

Abstract Cadmium telluride (CdTe) thin lms were prepared by the closed space sublimation (CSS) technique, using CdTe powder as evaporant onto substrates of waterwhite glass. In the next step, the same procedure was adopted by using tellurium as evaporant and already deposited CdTe lm as substrate. Such compositions were then annealed at 300 1C for 30 min to obtain Te-enriched lms. The samples were characterized by X-ray diffraction (XRD), scanning electron microscopy (SEM), spectrophotometry, DC electrical resistivity, dark conductivity and activation energy analysis as a function of temperature by two-probe method. The electron microprobe analyzer (EMPA) results showed an increase of Te content composition in the samples as the mass of the Te-deposition increased in CdTe. The Hall measurements indicated the increase in mobility and carrier concentrations of CdTe lms by addition of tellurium. A signicant change in the shape and size of the CdTe grains were observed. r 2005 Elsevier B.V. All rights reserved.
PACS: 61.10.Nz; 68.37.d; 78.66.w; 81.05.Dz Keywords: A1. Crystal structure; A1. Optical microscopy; A1. Recrystallization; A1. Substrates; B1. Cadmium compounds; B2. Semiconducting IIVI materials

1. Introduction
Corresponding author. Tel.: +92 51 2828187; fax: +92 51 9210256. E-mail addresses: nabbasqureshi@yahoo.com (N. Abbas Shah), tpl.qau@usa.net (A. Maqsood). 1 Physics Department, The Hashemite University, Zarqa, Jordan.

As a result of recent advances in the efciency of Cadmium telluride (CdTe)-based solar cells [13], interest in this material for thin lm terrestrial photovoltaic applications [46] is increasing. CdTe has a direct band gap of 1.5 eV at room

0022-0248/$ - see front matter r 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.jcrysgro.2005.08.005

ARTICLE IN PRESS
478 N. Abbas Shah et al. / Journal of Crystal Growth 284 (2005) 477485

temperature, which is optimum for single junction solar cell efciency [3]. CdTe can exhibit both types of conductivity, n and p, which makes diode technology [7] and eld effect transistors [8] possible, and it can exhibit a semi-conducting state as well [9]. CdTe-based semi-magnets, like CdMnTe, display extremely exciting properties, which have not so far been completely exploited. Wu et al. [10] have reported on a CdTe/CdS solar cell with an efciency of 16.5%. Because of its low sublimation temperature, CdTe polycrystalline lms can be prepared by several techniques, such as thermal evaporation [11], metal organic chemical vapour deposition (MOCVD) [12], radio frequency (RF) sputtering [13], closed space sublimation (CSS) [14], screen printing [15], hotwall ash evaporation [16], molecular beam deposition [17,18], hot-wall vacuum evaporation [1921], etc. The CSS technique is one of the various techniques that have produced encouraging results. The CSS process offers the advantages of simple deposition apparatus and high transport efciency conducted under low vacuum conditions at moderate temperatures, thus, simplifying scale-up for high-volume continuous processing [22,23]. It has been pointed out that high efcient solar cells possess Te-rich CdTe surfaces with smooth interfaces of p-CdTe/ n-CdS [24]. The successful experimental approach will open a pathway for improving devices based on CdTe, such as solar cells, g and IR detectors and eld effect transistors [8]. The efciency of these devices is strongly determined by the electrical and optical properties of the lms. The systematic research on these properties in correlation with doping/excess of materials and with the lm preparation conditions is still necessary for a more comprehensive understanding of this correlation and for the control of the properties of these lms. In the present work, we have fabricated Te-rich CdTe lms using a home made CSS system and investigated the effects of Te addition on the crystal structure, morphology, electrical and optical properties such as lm thickness, refractive index, absorption coefcient and optical band gap.

2. Experimental procedure Usually, the deposition of CdTe lms is based on the following reversible high-temperature process [25]: 2CdTes32Cdg Te2 g. The CSS technique, however, offers the neardistance deposition without marked dissociation of the components. The CSS system used in our experiment is shown in Fig. 1. Thin lms of CdTe powder (99.99% pure) of Aldrich chemical company were prepared onto microscope slides of waterwhite glass substrates of size 25.4 mm 76.2 mm. For that source material (14 mg) was put in the graphite boat of size 20 mm 45 mm, heated by a halogen lamp (1000 W) connected to the main power through temperature controller with K-type thermocouple (Chromel & Alumel,2701350 1C). The substrate was xed at a distance of about 4 mm from the source material. It was heated by a second halogen lamp (500 W), while the thermocouple was placed over the substrate to measure its surface temperature. Source and substrate temperatures were 400 and 300 1C, respectively. The chamber was evacuated down to $103 mbar with the help of rotary pump. The deposition time for each lm was 5 min at the source temperature of 400 1C. After that the source and substrate lamps were switched off for cooling down to 100 1C before opening the chamber. In the next step, these CdTe lms were covered with varying excess Te mass of 0.5, 1, 2, 3, 4 and 5 mg, respectively, which was evaporated onto them by CSS technique. These lms were numbered 27 as given in Table 1. The temperature of the source and substrate was 200 and 100 1C, respectively. Again, the substrate was xed at a distance of about 4 mm from the source material. The time of Te evaporation was 5 min for each sample. The vacuum of the system was $103 mbar. After that all samples were annealed at 300 1C for 30 min under the same vacuum. Finally they were gold coated for ohmic contacting to perform Hall measurements. Table 1 compiles the samples denoted by Nos. 27 with successively varying Te content obtained after tellurium

ARTICLE IN PRESS
N. Abbas Shah et al. / Journal of Crystal Growth 284 (2005) 477485 479

Fig. 1. Schematic diagram of closed space sublimation system.

deposition and annealing procedure by the above technique. The structure of the lms was studied by X-ray diffraction (XRD) using Cu-Ka (1.5418 A) radiation with operating conditions voltage/current 40 kV/30 mA. The microstructure of the samples was checked using scanning electron microscope (SEM). The SEM accelerating voltage was 10 kV. The compositions of Cd and Te of all the samples were measured with the help of electron microprobe analyzer (EMPA) attached to the SEM. The optical properties such as lm thickness, refractive index, absorption coefcient and optical band gap were calculated from the transmission spectra between 300 and 2500 nm recorded by Lambda 900 spectrophotometer. The most direct and perhaps the simplest method for probing the band structure of semiconductors is to measure the absorption spectrum. A method for the calculation of the lm thickness d and refractive index n from the transmission measurements was developed by Aqili and Maqsood [26] where d can be determined from two

maxima or minima. Thickness should be in nanometer range for good transmission spectra and other optical measurements. The values of n and d, calculated from the above method are used as initial least-squares tting parameters in the theoretical model [27] for the following equation: T Ax , B Cx cosf Dx2 (1)

where A, B, C, D are constants and T is the normal transmittance for the system consisting of a thin lm on transparent substrate surrounded by air (refractive index n 1) and taking all multiple reections at the interface into account. In case of k2 ( n2 , which is true for this kind of semiconductor thin lms, where k is the extinction coefcient of the lm, an empirical formula for the n dependence of l as n a b=l2 can be used with a and b constants obtained from tting the transmission spectra. During the present investigation a two-point probe method was used to measure the sheet resistance Rs of all the lms by the relation Rs

ARTICLE IN PRESS
480 N. Abbas Shah et al. / Journal of Crystal Growth 284 (2005) 477485

Table 1 Te composition, lattice constant, thickness, optical constants, sheet resistance and electrical resistivity of the CSS grown CdTe samples 17

Energy gap Eg (ev)

V =I Ri where Ri is the internal resistance of the multimeter DM 350. The electrical properties such as sheet resistance, DC resistivity, dark conductivity and activation energy as functions of temperature (50200 1C) were measured by the two-probe method at a pressure of 106 mbar in a 306 A Edward vacuum coating unit. Silver paste was used above the gold coatings for electrical contacts. The resistivity, mobility and carrier concentrations were determined by room temperature Hall measurements.

Resistivity r (O cm) at 200 1C

Sheet resistance (O/ sq) at 200 1C

1.50 1.48 1.47 1.467 1.461 1.44 1.43

5.58 106 5.09 106 4.19 106 3.35 106 3.1 106 3.0 106 2.12 106

122.24 73.05 50.23 47.38 44.88 42.37 4.37

3. Results and discussion The XRD patterns (arranged by XRD software) of the samples 1, 3, 5, 6 and 7 are compared in Fig. 2. The main reections of four samples (1, 3, 6 and 7) are the same and can be indexed according to fcc CdTe lattice. The values of the lattice constant a0 are for the sample (1) and for Teenriched samples (3, 5, 6 and 7) 6.49 and 6.44, 6.86(3), 6.57(2) and 6.54(3) A, respectively. These values are calculated and averaged after comparison of respective experimental d values in agreement with the d standard values of the lattice constant according to 1999 JCPDSInternational Centre for Diffraction Data 10-0207, 15-0770, 75-2086 and 75-2083 (PCPDFWIN v. 2.02). The increase in the lattice constant of the Te enriched samples can be explained as follows. During the annealing the diffusing Te atoms may replace the Cd atoms by a kick off mechanism besides its diffusion at interstitial position in the lattice, which results in an increase of the lattice constant. In sample 5 a new peak corresponding to 200 reex appeared as shown in Fig. 2c. It matches with CdO mineral powder diffraction le (5-640) according to data book published by Joint Committee on Powder Diffraction Standards ICDD USA 1980, where-as the (1 1 1) plane disappeared completely. The presence of oxide is due to the heat treatment under low vacuum. It was observed in sample 3 that all the peaks were shifted by (0.41) towards the higher range (right side). This shows that the lattice is in uniform strain due to expansion of the cell. The expansion of the cell can be explained on the basis

Refractive index n a b=l2

Thickness d (nm)

Experimental (A) Standard (A)

Lattice constant ao

Te (at%)

52.55 52.66 54.70 55.80 57.26 59.84 60.43 1 2 3 4 5 6 7

6.481a

6.49 6.49 6.44 6.44 6.86(3) 6.57(2) 6.54(3)

a

732 323 348 416 400 264 663 I CDD 15770.

Sl. no.

2.48 2.23 2.43 2.52 2.55 2.61 2.21

43384 167493 138367 128460 138562 196591 366486

ARTICLE IN PRESS
N. Abbas Shah et al. / Journal of Crystal Growth 284 (2005) 477485 481

Fig. 2. X-ray diffraction patterns of CdTe samples: (a) as deposited and (b, c, d, e) Te- enriched according Table 1 after heat treatment at 300 1C for 30 min, 103 mbar.

of radius of the Te atom, which is larger than that of the Cd atom. Thus, the incorporation of extra Te leads to the expansion of the CdTe lattice. A similar observation has been reported by Li et al. [14]. The change in intensities was also observed. The strongest (1 1 1) reection in the pattern indicates a preferential orientation of [1 1 1] in the lms as already reported [28]. Also the grains of the lm became larger with the increase of Te content in the samples, which indicates that the crystallinity of the lms can be improved by appropriate Te enrichment. The Te content after its deposition and annealing determined by EMPA is shown in Table 1. The results show that the Te content in samples 17 is increasing. The ratio of at% of Cd and Te in asdeposited CdTe sample is 47.45:52.55, i.e. less than 1 which is considered the reason for CdTe lms with slight p-type conduction due to the formation of Cd vacancies in CdTe lattice acting as acceptor centers. The Te content was increased up to 60.43% in sample 7, which is maximum Te-enriched. It is clear from these values that already the samples grown from the pure CdTe source have a deciency in Cd, which can be almost 5 at%. On the other hand, the additional Te in the source leads to an

increase of the Te content from 52.55 to 60.43 at%, as in samples 17. The lower Cd content gives rise to the formation of vacancies of Cd in the CdTe lattice. From Hall measurements samples 17 show p-type conductivity. The amount of excess Te in the lms mainly affects their electrical properties. As can be seen from Table 1 the resistivity of the samples decreases with increasing Te content by two orders of magnitude. SEM images in Fig. 3 show the main feature of the CSS technique, i.e. the Te-enriched lms have larger grains. The crystallographic arrangements are related to the stress in the layer and to the application of annealing at high temperature. After optimization of annealing time, the average grain size of as-deposited lm is under 540 nm while that of Te-enriched lms in Figs. 3(b, c) are about 990 nm and 2 mm, respectively. The creation of larger grains is as a result of coalescence of small grains into bigger ones which can be seen from Fig. 3(c). This is consistent with the above results of XRD. Also a signicant change of the grain shape was observed. It appears that the annealing in the presence of Te facilitates the grain growth in the CdTe lms in such a way that some of the small grains coalesce together and

ARTICLE IN PRESS
482 N. Abbas Shah et al. / Journal of Crystal Growth 284 (2005) 477485

Fig. 3. SEM images of the grown samples: (a) as-deposited and (b, c) Te-enriched CdTe lms.

reorient themselves, which decrease the number of grain boundaries. As a result an improved different microstructure and morphology is created. Grain growth in CdTe is well known and reported for a solid-state recrystallization (SSR) method [29,30]. Usually CdTe lms have a high absorption in the region of visible and near infrared. Indeed, the same was observed in the lms deposited during this study. The transmission spectrum of the lms did not change signicantly after the optimization of heat treatment for 30 min, which ensured the diffusion of Te in the samples. The obtained thickness values are in the range of 264730 nm. The transmittance slightly decreases with increase of Te content as shown in Fig. 4(a). Fig. 4(b) shows the resulting t of Eq. (1) to experimental data. A good tting in the transparent region is obvious. Fig. 5. illustrates the change of refractive index n with Te content. The increase in the refractive index is due to the change in the lattice constant.

For the calculation of absorption coefcient a in the high absorption region the values of n and d from the tted curve are used. A high absorption coefcient is important from both technological and economic considerations. Large absorption coefcient permits the absorber layer to be thin and hence less material is necessary for photovoltaic device fabrication. Since the region of high absorption in our lms matches well with the solar spectrum, these lms are optically well-suited for use as an absorber in solar cells. For the band gap determination the following relation can be used [31]: ahn Ahn E g N=2 . (2)

Here A is a constant, hn the photon energy, Eg the optical band energy gap and N depends on the nature of the transition (N 1 for direct band gap and N 4 for indirect band gap transition). By extrapolating (ahn)2 vs. the incident photon energy (hn) plot the band gap can be obtained. The values

ARTICLE IN PRESS
N. Abbas Shah et al. / Journal of Crystal Growth 284 (2005) 477485 483

1.0 0.8 Transmittence 0.6 0.4 0.2 0.0 500

(a)

As- deposited (sample 1) Te enriched (sample 7)

10

(1) (7)
n

9 8 7 6 5 4 Sample(2) Sample(3) Sample(4) Sample(5) Sample(6) Sample(7)

1000 (nm)

1500

2000

3 2

1.0

500

1000

1500 (nm)

2000

2500

3000

0.8 Transmittence

Fig. 5. Refractive index as a function of wavelength.

0.6

0.4

0.2

0.0 500
(b)

1000

1500 (nm)

2000

2500

Fig. 4. Transmittance vs. wavelength (a) for samples 1 and 7 and (b) transmittance data along the curve tted for sample 2.

of energy gap, thickness and refractive index of all the sample series are listed in Table 1. The decrease in the energy gap is also due to tellurium diffusion into the CdTe lms. The sheet resistance in the present case for the as-deposited sample 1 was 5.58 106 O/sq at 200 1C, while it dropped to 2.12 105 O/sq in Teenriched sample 7 at the same temperature. For the resistivity determination by two-probe method the following relation could be used: r Rs d w , (3)

where Rs is the sheet resistance, d the thickness, w the width and the length of the lm.

In general, as-prepared CdTe lms, fabricated by various techniques, show a high electrical resistivity (107109 O cm) at room temperature [14]. The change in sheet resistance and resistivity of the samples as function of Te content at temperature 200 1C is shown in Table 1. As can be seen these values decrease with the increase of Te content. The change of the electrical property of Teenriched samples can be easily understood according to the above analysis of the sample microstructure. The grains of Te-rich samples become larger and the crystallinity improves which results in the decrease of the number of grain boundary and, hence, of the decrease in the grain barriers. All these changes are favorable for the transport of carriers. It is shown in Figs. 6 and 7 that the dark conductivity of the samples as a function of temperature (50200 1C) is increasing. The dark conductivity activation energy (E a E f 2E v ) was deduced by linear tting ln (s) vs.1/kBT (kBBoltzmann constant). The slope gives the activation energy. The representative activation energy of one of the Te- rich sample 7 is shown in Fig. 7. The resistivity of sample 7 with a van der Pauw geometry was $156 O cm at room temperature, while the mobility and carrier concentration at room temperature were about 129.3 cm2 V1 s1 and 3.09 1014 cm3, respectively. After Te addition the conductivity of the samples was markedly

ARTICLE IN PRESS
484
-0.5 -1.0 -1.5 -2.0 Log () ( cm)-1 -2.5 -3.0 -3.5 -4.0 -4.5 -5.0 -5.5 2.0 sample1 sample2 sample3 sample4 sample5 sample6 sample7 2.2 2.4 2.6 1000 / T (K)-1 2.8 3.0 3.2

N. Abbas Shah et al. / Journal of Crystal Growth 284 (2005) 477485

lattice that might create Cd vacancies, which increase the p-type carriers. Further work on this material is under progress and will be addressed in future.

4. Conclusion In this work, Te-enriched CdTe lms have been fabricated by the CSS technique with subsequent annealing. On the basis of the XRD, SEM and Hall measurements one can conclude that the post annealing strongly affects the morphology as well as electrical properties. The samples show large grains with size $2 mm which can grow larger with increasing Te content. The values of refractive index increases with the increase of Te concentration. The transmission spectra showed a decrease of transmittance with increase of Te. The band gap values are also inuenced by Te incorporation. The sheet resistance, DC electrical resistivity and activation energy of Te-rich samples were decreased but the dark conductivity, mobility and carrier concentration were increased signicantly compared to pure CdTe samples. The increase of the lm mobility could be due to the enlargement of the grains size, while the diffusion of Te in the CdTe lms cause an increase in the number of the charge carriers.

Fig. 6. Dark conductivity of the heated samples (17) plotted against 1/T.

-1 Conductivity of sample (7) Linear fit to the data

-2 ln () ( - cm )-1

-3

-4

-5

-6 24 26 28 30 32 34 36 1 / kBT (ev)-1

Acknowledgements This work was supported by Pakistan Science Foundation project No. PSF/Res /C-QU/ PHY.(121). The authors would like to thank Prof. Dr. M. Riaz Khan of Central Resource Laboratory, Peshawar University, Pakistan, for providing the SEM facilities. Dr. M Anees-ur Rehman and Mrs. Naghma Haider are thanked for their help in the X-ray diffraction analysis. References
[1] G. Fulop, M. Doty, P. Meyers, J. Betz, C.H. Liu, Appl. Phys. Lett. 40 (1982) 327. [2] K. Yamaguchi, H. Matsumoto, N. Nakayama, S. Ikegami, Jpn. J. Appl. Phys. 15 (1976) 1575.

Fig. 7. Fitting of ln (s) against 1/kB T of sample 7.

improved. This improvement of the lm conductivity could be attributed to the increased grain growth taking place in the CdTe lms in presence of Te. As the concentration of Te increased the resistivity was reduced, probably due to the Te doping effect in CdTe. Similar results are reported by Turkevych et al. [32] for similar temperatures. It has been observed by Hall measurements that p-type doping is achieved in Te-enriched CdTe lms prepared by CSS technique. This is caused by the excess of Te atoms replacing a part of the

ARTICLE IN PRESS
N. Abbas Shah et al. / Journal of Crystal Growth 284 (2005) 477485 [3] H. Matsumoto, K. Kuribayashi, Y. Komatsu, A. Nakano, H. Uda, S. Ikegami, Jpn. J. Appl. Phys. 22 (1983) 891. [4] R. Chakrabarti, J. Dutta, A.B. Maity, S. Chaudhuri, A.K. Pal, Thin Solid Films 288 (1996) 32. [5] J. Touskova, D. Kindl, J. Tousek, Thin Solid Films 293 (1997) 272. [6] Y.L. Soo, S. Huang, Y.H. Kao, A.D. Compaan, J. Appl . Phys. 83 (1998) 4173. [7] X.L. Saldana, C. Vasquez-Lopez, A. Zehe, H. Navarro, R. Triboulet, Appl. Phys. Lett. 39 (1981) 433. [8] G. Landwehr, A. Waag, K. Hoffmann, R.N. BicknellTassius, Proc. SPIE 1362 (1991) 382. [9] D. De Nobel, Philips Res. Repts 14 (1959) 361. [10] X. Wu, J.C. Keane, R.G. Dhere, C. Dehart, D.S. Albin, A. Duda, T.A. Gessert, S. Asher, D.H. Levi, P. Sheldon, in: Seventeenth European Photovoltaic Solar Energy Conference, Proceedings of the International Conference, vol. 1, WIP-Munich and ETA- Florence 2001, p. 995. [11] C.S. Ferekides, D. Marinskiy, V. Viswanahan, B. Tetali, V. Palekis, P. Selvaraj, D.L. Morel, Thin Solid Films 361362 (2000) 520. [12] S.J.C. Irvine, V. Barrioz, A. Stafford, K. Durose, Thin Solid Films 480481 (2005) 76. [13] N. EI-Kadry, M.F. Ahmed, K. Abdel Hady, Thin Solid Films 274 (1996) 120. [14] J. Li, Y.F. Zheng, J.B. Xu, K. Dai, J. Semicond. Sci. Technol. 18 (2003) 611. 485

[15] Z.C. Feng, H.C. Chou, A. Rohatgi, G.K. Lim, A.T.S. Wee, K.L. Tan, J. Appl. Phys. 79 (1996) 2151. [16] C.A. Menezes, J. Electrochem. Soc. 127 (1980) 155. [17] T.H. Myers, A.W. Waltner, J.F. Schetzina, J. Appl. Phys. 53 (1982) 5697. [18] T.H. Myers, J.F. Schetzina, J. Vac. Sci. Tech. 20 (1982) 134. [19] W. Huber, A. Lopez-Otero, Thin Solid Films 58 (1979) 21. [20] J. Humenburger, H. Sitter, W. Huber, N.C. Sharma, A. Lopez-Otero, Thin Solid Films 90 (1982) 101. [21] A. Lopez-Otero, Thin Solid Films 49 (1978) 3. [22] J. Britt, C. Ferekides, Appl. Phys. Lett. 62 (1993) 2851. [23] P.D. Paulson, V. Dutta, Thin Solid Films 370 (2000) 299. [24] A. Picos-Vega, M. Becerril, O. Zelaya-Angel, R. RamirezBon, F.J. Espinoza-Beltran, J. Gonzalez-Hemandez, S. Jimenez-Sandoal, B. Chao, J. Appl. Phys. 83 (1998) 760. [25] K. Mitchell, A.L. Fahrenbruch, R.H. Bube, J. Vac. Sci. Tech. 12 (1975) 909. [26] A.K.S. Aqili, A. Maqsood, J. Appl Opt 41 (2002) 218. [27] R. Swanepoel, J. Phys. E: Sci. Instrum. 17 (1984) 89. [28] K.H. Kim, J.S. Chun, Thin Solid Films 141 (1986) 287. [29] R. Triboulet, Cryst. Res. Technol. 38 (2003) 215. [30] S. Hassani, A. Lusson, A. Tromson-Carli, R. Triboulet, J. Crystal Growth 249 (2003) 121. [31] A.K.S. Aqili, A. Maqsood, Z. Ali, J. Appl. Surf. Sci. 180 (2001) 73. [32] I. Turkevych, R. Grill, J. Franc, E. Belas, P. Hoschl, P.J. Moravec, Semicond. Sci. Technol. 17 (2002) 1064.