Journal of Alloys and Compounds 362 (2004) 241–247 Rietveld refinement for CuInSe 2 and CuIn 3 Se 5 W. Paszkowicz a,∗ , R. Lewandowska b , R. Bacewicz b a Institute of Physics, P.A.S., Al. Lotników 32/46, 02-668 Warsaw, Poland b Faculty of Physics, Warsaw University of Technology, Ul. Koszykowa 75, 00-662 Warsaw, Poland Received 15 December 2002; received in revised form 19 February 2003; accepted 25 February 2003 Abstract Rietveld refinements for copper indium selenides CuInSe 2 and CuIn 3 Se 5 and their mixture were performed using data collected at a Bragg–Brentano diffractometer. The values of lattice parameters, axial ratio and positional parameters are discussed and compared to available literature data for single crystals and polycrystals. Results for CuInSe 2 give a 0 = 5.78149(1) Å and c 0 = 11.61879(4) Å. The positional parameter x is determined to be equal 0.2271(4). Refinements for CuIn 3 Se 5 give an indication concerning the choice of a structural model for this compound. Namely, the best fit is obtained for a model based on I ¯ 42m space group, yielding a composition close to the experimental one and the lattice parameters a 0 = 5.75812(2) Å and c 0 = 11.53593(7) Å. Parameters of this model are discussed. © 2003 Elsevier B.V. All rights reserved. Keywords: Semiconductors; Crystal growth; Crystal structure and symmetry; X-ray diffraction 1. Introduction Copper indium diselenide, CuInSe 2 , is a subject of inves- tigation in many laboratories because of its application as an efficient absorber in polycrystalline-thin-film solar cells. An advantage of such cell is that the CuInSe 2 absorbing layer is substantially thinner (several micrometers) than that of polysilicon based solar cells (several hundred microm- eters). CuInSe 2 -based laboratory devices attain 21% effi- ciency while the reported value for large-scale devices is 12% [1,2]. The technological importance of this compound attracted attention of the scientific community to Cu–In–Se system [3–6], in particular to CuIn 3 Se 5 compound which exhibits some structural similarities to CuInSe 2 . The structure of copper indium diselenide, CuInSe 2 , is closely related to the zincblende structure type. This com- pound crystallises in a tetrahedral chalcopyrite-type struc- ture: space group I ¯ 42d, with Cu on (0, 0, 0), In on (0, 0, 0.5), Se on (x, 0.25, 0.125), x being close to 0.225. It exhibits a range of off-stoichiometry [3,7], attributed to remarkably low formation enthalpy of defect pairs [8]. For CuIn 3 Se 5 , some structural properties are not fully understood up to now. Its structure involves a consider- ∗ Corresponding author. E-mail address: paszk@ifpan.edu.pl (W. Paszkowicz). able contribution of vacancies and allows for a Cu:In ra- tio varying in the range 0.27 to 0.50. Only recently, the I ¯ 42m space group has been proposed by Hanada et al. based on the stannite-type structure formula ABIn 2 Se 4 ; site occu- pancy factors (SOFs) for sites 2a,2b,4d,8i have been dis- cussed [9]. According to this proposition, the ideal formula is Cu 0.8 In 0.4 In 2 Se 4 with Cu on 2a at (0, 0, 0), In on 2b at (0, 0, 0.5) and on 4d at (0, 0.5, 0.25), Se on 8i at (x, x, z), where x and z are free parameters close to 1/4 and 1/8, respec- tively. This model (termed below as the ‘basic’ model) in- cludes vacancies at 2a and 2b sites allowing for the variable composition. For a quenched sample of a similar composi- tion, the space group P ¯ 42c has been proposed [10]. Merino et al. [11] have concluded that P ¯ 42c space group leads to a slightly better structure refinement results for the sample studied than does the I ¯ 42m group. Knight [12] has pointed out the importance of Rietveld method in studies of CuInSe 2 and related compounds, justi- fied by difficulties in growing their single crystals; the diffi- culties are connected, in particular, with twinning. Due to se- vere peak overlap in powder diffraction patterns (tetragonal unit cells of CuInSe 2 and CuIn 3 Se 5 are of similar size, with axial ratios differing slightly from 2), the Rietveld method is a suitable tool for X-ray characterisation of CuInSe 2 and CuIn 3 Se 5 . Indeed, several research groups used this method successfully for CuInSe 2 (see, e.g., Refs. [13,14]) as well as for CuIn 3 Se 5 [11] and for related bulk crystals such as 0925-8388/$ – see front matter © 2003 Elsevier B.V. All rights reserved. doi:10.1016/S0925-8388(03)00592-9