Preparation of 4.8% Efficiency Cu2ZnSnSe4 Based Solar Cell by a Two Step Process

Preparation of 4.8% Efficiency Cu2ZnSnSe4 Based Solar Cell by a Two Step Process A. Fairbrother1, E. Saucedo1, X. Fontané1, V. Izquierdo-Roca1, D. Sylla1, M. Espindola-Rodriguez1,2, F.A. Pulgarín-Agudelo3, O. Vigil-Galán2, A. Pérez-Rodríguez1,4 1

IREC, Catalonia Institute for Energy Research, Sant Adrià del Besòs, Barcelona, Spain 2

Escuela Superior de Física y Matemáticas (ESFM), IPN, México D.F., México 3

4

Centro de Investigación en Energía-UNAM, Temixco, Morelos, México

IN2UB, Departament d’Electrònica, Universitat de Barcelona, Barcelona, Spain

Abstract — Zn-rich Cu2ZnSnSe4 (CZTSe) films were prepared by a two step process. We have varied the Zn/Sn ratio from 1.24 to 1.73 and analyzed the effects of precursor composition and annealing temperature (between 425o C and 550o C) on the morphological, structural, and optoelectronic properties of the films. Raman scattering measurements show the presence of ZnSe as the main secondary phase in the films, as well as SnSe at the back region of the films processed with lower Zn-excess values and annealed at lower temperatures. The effect of the different secondary phases on the optoelectronic properties is discussed. In a first optimization we obtain as a preliminary result a 4.8% efficiency solar cell. Index Terms — Cu2ZnSnSe4, kesterite, composition, secondary phases, ZnSe, SnSe, Raman spectroscopy.

I. INTRODUCTION Cu2ZnSn(S,Se)4 (CZTS), commonly referred as kesterite, its crystal structure, has received increasing interest as a photovoltaic absorber material due to the promising results presented in the last two years in terms of efficiency of the devices [see for example 1-6]. The key advantage of these materials with respect to the more mature CdTe and Cu(In,Ga)Se2 technologies is that kesterites are composed of earth abundant and low toxicity elements. Nevertheless, these materials are still at an incipient state of development, requiring further investigation and optimization. Due to the complexity of these systems, compositional parameters are expected to have a strong influence on the properties of CZTS. In fact, there is a general agreement that Zn-rich conditions are beneficial for the properties of kesterites [1-6], because they avoid the formation of the ternary Cu2Sn(S,Se)3 (CT(S,Se)) compound, known to reduce the Voc, even with the risk of the formation of a Zn(S,Se) binary phase [6]. Thus it is clear that the Zn concentration needs to be high enough to prevent the formation of CT(S,Se), but as low as possible to minimize the occurrence of Zn(S,Se). The majority of the research presented to date is devoted to the study of pure sulfide or sulfo-selenide systems [1-8]. Recent advances on the pure selenide compound (Cu2ZnSnSe4, CZTSe) have permitted an increase of the modest record efficiencies reported for CZTSe from 3.2% [9]

978-1-4673-0066-7/12/$26.00 ©2011 IEEE

to 9.15% [8], showing the high potential of this material. To obtain such a high efficiency, Repins et al. have prepared globally Zn-rich and Cu-poor CZTSe (Cu/(Zn+Sn) = 0.86, Zn/Sn = 1.15) [8] by vacuum co-evaporation. In spite of this substantial increment on the conversion efficiency of CZTSe based devices, and although analyzed in some extent for the sulfide case [3,6,7], reports on the understanding of secondary phase formation in the case of Zn-rich pure selenide compounds are not as well studied. A deeper understanding of the characteristics of the selenide system is an important issue to improve the potential of these technologies and to have better knowledge of the mechanisms limiting efficiency in selenium-rich record devices. Thus, a study of the formation of secondary phases in Zn-rich conditions is a key feature and challenge. In this work we present a systematic study of Cu2ZnSnSe4 films prepared by a two stage process, as a function of the Znexcess concentration and reactive annealing conditions. We have used sufficient Zn-excess to provoke the formation of ZnSe with the aim of studying its distribution and influence on the optoelectronic parameters. The results show the importance of the accurate control on the precursor composition and annealing parameters which are determining factors for the optoelectronic properties of CZTSe devices. In these preliminary results we have obtained a maximum efficiency of 4.8%, describing the possible impact on the optoelectronic characteristics of the cells of two important secondary phases, namely ZnSe and SnSe. II. EXPERIMENTAL Sn/Cu/Zn metallic stack precursors have been deposited by DC-magnetron sputtering deposition (AC450 Alliance Concepts) using 99.99% purity targets, onto 10x10 cm2 Mo coated soda-lime glass (500 nm thickness, 0.25 Ω/ sheet resistance) in Zn-excess and Cu-deficient conditions. The stack order was selected to minimize Sn evaporation [3] and to promote the segregation of excess Zn-rich phases towards the surface. The thickness and composition of the precursor stacks

002679

were measured by XRF (Fisherscope XVD) and are summarized in Table I. TABLE I TOTAL THICKNESS AND COMPOSITIONAL RATIOS OF PRECURSOR

Abet Technology (uniform illumination area of 15x15 cm2). Selected absorbers were also characterized by Raman Spectroscopy with a LabRam HR800-UV Horiba-Jobin Yvon spectrometer and SEM with a FEI Nova™ NanoSEM 230 microscope.

STACKS

Precursor SnCuZn-1 SnCuZn-2 SnCuZn-3 SnCuZn-4

Total thickness (nm) 813.8 834.7 797.6 688.1

Zn/Sn 1.24 1.40 1.49 1.73

Cu/(Zn+Sn) 0.77 0.84 0.87 0.67

Samples of the metallic precursors (2.5x2.5 cm2 in area) were reactively annealed in a three zone tubular furnace capable of working in vacuum (10-4 mBar) and inert gas atmosphere (Ar). A graphite box (23.5 cm3 in volume) was used for the reactive annealing under a Se and Sn atmosphere, varying the temperature between 425 ºC to 550 ºC. Devices were fabricated by depositing CdS (60 nm) by chemical bath deposition, followed by pulsed DC-magnetron sputtering deposition of i-ZnO (50 nm) and ZnO:Al (450 nm thickness, 19 Ω/ sheet resistance) (CT100 Alliance Concepts). For the optoelectronic characterization 3x3 mm2 cells were scribed using a micro-diamond scriber (MR200 OEG), thus avoiding the necessity of metallic grid deposition onto the ZnO:Al surface. Measurement of the optoelectronic properties was carried out using a Sun 3000 class AAA solar simulator from

III. RESULTS AND DISCUSSION The evolutions with annealing temperature of the compositional parameters (Zn/Sn, Cu/(Zn+Sn), Se/(Cu+Zn+Sn)) and the thickness, for the different precursors and absorber films are presented in Fig. 1. SnCuZn-1 – For the precursor SnCuZn-1, both Zn/Sn and Cu/(Zn+Sn) tend to increase with annealing temperature. This is clear evidence that in these samples Sn loss occurs with the increasing temperature, suggesting that the imposed Sn+Se atmosphere during annealing does not entirely prevent Sn evaporation from the precursor. The Se/(Cu+Zn+Sn) ratio is close to the stoichiometric value and almost constant with annealing temperature, and the thickness is around 2 µm as expected. This indicates that the incorporation of Se proceeded in an expected fashion. SnCuZn-2 and SnCuZn-3 – Both of these precursors behave in a very similar way, being more Zn- and Cu-rich than the SnCuZn-1 precursor. In contrast with the previous case, the Zn/Sn ratio seems to be nearly constant with temperature when considering the measurement error. The Cu/(Zn+Sn) ratio tends to decrease suggesting some enhancement in the

Fig. 1. Compositional parameters as a function of the annealing temperature for the four precursors, (A) Zn/Sn ratio; (B) Cu/(Zn+Sn); (C) Se/(Cu+Zn+Sn); and (D) absorber thickness; curves have been added as a visual guide

978-1-4673-0066-7/12/$26.00 ©2011 IEEE

002680

incorporation of Sn. A similar composition adjustment effect has been observed in the sulfide case, where Sn is incorporated from the atmosphere [3]. From these data we propose that a minimum Cu content threshold is necessary for the incorporation of Sn from the annealing atmosphere, probably because Sn is incorporated via reaction with Cu and Se to form ternary Cu-Sn-Se phases which further react with ZnSe binary to form CZTSe [10]. The resulting absorbers are more Se-rich than in the previous case, probably because of the formation of additional ZnSe secondary phases due to the higher starting Zn concentration. Thickness is slightly higher, in agreement with the higher thickness of the corresponding precursors. SnCuZn-4 – In the case of the precursor with highest Znexcess and Cu-deficiency, significant compositional deviations are obtained with respect to stoichiometry. In particular, the Zn/Sn ratio is as high as about 2.4 in the temperature range between 425-450 ºC, suggesting a high degree of formation of the ZnSe secondary phase accompanied by an unexpected Sn loss. This is related to the low Cu/(Zn+Sn) content ratio, which limits the availability of Cu to form Cu-Sn-Se ternary compounds in the chemical reaction pathway. From these results it is evident that by limiting Cu, the probability for Sn loss is enhanced. The thickness of these samples is unusually low suggesting even an incomplete reaction of the metallic stack precursor. At higher annealing temperatures, apparently the Sn-Se pressure on the graphite box is enough to correct to some extent the stoichiometry of the films. The high Se amount obtained by means XRF measurements (approximately 40% in excess from the stoichiometric value), suggest high formation of secondary phases, mostly ZnSe as expected by the precursor composition. The thickness approaches 2 µm with higher annealing temperatures. The morphology and grain size of the CZTSe absorber films as seen by SEM depends on both the annealing temperature and precursor composition. Figs. 2A, B and C show the SEM cross section images of the complete cells fabricated with the

first precursor stack (Zn/Sn = 1.24) and annealed at different temperatures. The grain size increases with annealing temperatures from about 500 nm at 450ºC up to 2 µm at 550ºC. However, increasing the temperatures lead also to a deterioration of the morphology of the Mo/CZTSe back contact region, as voids are formed which are potentially detrimental to the optoelectronic properties and mechanical robustness of the cell. The size of these voids matches closely with the larger features seen in the as-deposited Sn precursor films as seen by AFM (not shown here), thus we tentatively attribute the formation of these voids to the diffusion and evaporation of Sn and SnSe, which is favoured at higher temperatures, even with the use of the Sn containing atmosphere. Increasing the Zn-excess in the precursor leads to the formation of a bi-layer structure, with an upper layer with larger grains and a sub-layer with smaller grains as is observed in Figs. 2D, E and F for films from precursors with varying Zn-excess annealed at 450ºC. The relative thickness of both layers changes with composition, reducing the size of the upper layer, and increasing the thickness of the sub-layer with the Zn/Sn precursor content ratio, and absorbers obtained from precursor SnCuZn-4 exhibit a completely nanocrystalline grain structure. In principle, Cu content in the precursors is known to affect the recrystallisation of the layers [11,12]. However, in our case, and for the range of compositions used in these experiments, the Zn/Sn precursor content ratio appears to be more of a determining factor for the grain size. These data suggest that the grain growth is likely inhibited under high Zn-excess conditions. This might be related to the formation of a more uniform distribution of ZnSe from the back region of the layers, as deduced from the Raman scattering measurements discussed later. To determine the impact of both precursor composition and annealing temperature parameters on the structural properties of the absorbers, Raman spectroscopy measurements have been performed. Fig. 3A shows the Raman spectra measured with an excitation wavelength of 514 nm on the surface of the

Fig. 2. Cross sectional SEM images of different devices and films: (A) Zn/Sn = 1.24, Tanneal = 450ºC; (B) Tanneal = 500ºC; (C) Tanneal = 550ºC; and (D) Tanneal = 450ºC Zn/Sn = 1.40; (E) Zn/Sn = 1.49; (F) Zn/Sn = 1.73 978-1-4673-0066-7/12/$26.00 ©2011 IEEE

002681

absorbers with the same precursor composition (Zn/Sn = 1.24) and annealed at different temperatures. The spectra are characterized by the presence of the main peaks from modes of Cu2ZnSnSe4. The spectra also show a peak at about 250 cm1 that has been identified with the main Raman peak of ZnSe [13]. This shows the presence of this secondary phase at the surface region of the absorbers. Increasing the annealing temperature to 550ºC leads to an increase of the FWHM together with a red shift of the main A1 kesterite peak, as well as a decrease in the relative intensity of the A1 weaker mode. These features suggests a worsening of the crystalline quality of the surface region of this layer. This agrees with the recent observation of a disorder induced spectral contribution at the low frequency side of the main kesterite peak in kesterite Cu2ZnSnS4 [15]. An alternative origin for these spectral changes could be related to the presence of a ternary Cu2SnSe3 secondary phase, which has an A mode at 178 cm-1 [14]. However, as was discussed in the compositional analysis of these samples, Sn-loss is observed with the increasing temperature, making the formation of this phase less likely. More likely, the observed broadening of the CZTSe mode can

be attributed to disorder introduced in the kesterite structure by the high annealing temperature with the consequent Snloss, leading to the formation of the characteristic voids at the back region of the films observed from the cross sectional SEM images. Changes with composition are less evident, and surface Raman spectra from samples with different precursor composition are very similar, as can be observed in Fig. 3B. Nonetheless, Raman measurements performed directly at the back region of the layers – from samples that were mechanically removed from the Mo coated substrate – with a 457.9 nm excitation wavelength have permitted the detection of the ZnSe peaks independent on the Zn-excess conditions. This excitation wavelength is particularly suited to detect ZnSe, because of the existence of a pre-resonant excitation of the ZnSe Raman peak [13] (Fig. 3C). The analysis of the intensity of the ZnSe peaks measured at different regions in the back surface indicate the formation of a more uniform distribution of this secondary phase when the Zn-excess and/or the annealing temperature is increased. This correlates with the observation of a thicker nanocrystalline layer at the

Fig. 3. Raman spectra taken for the different samples: (A) Zn/Sn = 1.24 with different annealing temperatures; (B) Tanneal = 450 ºC with different Zn/Sn ratios; (C) comparison of Raman spectra measured with 514 nm and 457.9 nm excitation wavelengths at the back region (Zn/Sn = 1.24, Tanneal = 450 ºC); and (D) Raman spectra from two different points at the back of the film (514 nm excitation, Zn/Sn = 1.24, Tanneal = 450 ºC) 978-1-4673-0066-7/12/$26.00 ©2011 IEEE

002682

back contact region of the absorbers, as shown in Figs. 2D, 2E and 2F. The other secondary phase detected by Raman spectroscopy is SnSe as is shown in Fig. 3D. SnSe is only detected at certain points in samples with lower Zn-excess (precursors with Zn/Sn = 1.24 and 1.40) and annealed at lower temperatures (Tanneal ≤ 450 ºC), whereas it is not observed in the other samples. This could be related to the inhomogeneity of the Sn precursor layer, which is considerably less uniform than the Cu and Zn precursor films. However, higher Znexcess values likely lead to an inhibition of the formation of this phase, being favored in this case the formation of ZnSe. J-V curves of films from precursor SnCuZn-1 annealed at different temperatures are shown in Fig. 4A, as well as for precursor films with different compositions and annealed at 450ºC (Fig. 4B). For the 425ºC anneal there is a moderate JSC of about 20.8 mA/cm2 and VOC of 230 mV, and at 450ºC these parameters reach maximum values of 22.5 mA/cm2 and 380 mV, respectively. For increasing annealing temperature there are slight successive decreases in JSC and VOC, down to 15.5 mA/cm2 and 175 mV respectively for Tanneal. = 550ºC. These results suggest that the efficiency is mostly controlled by VOC when annealing temperature is increased. This feature could be related to the deterioration observed in the back contact region in Figs. 2B and 2C, as well as the formation of a more uniform ZnSe distribution at this region suggested from the Raman measurements. Apparently in this case, the

presence of ZnSe at the front region of the films and the formation of SnSe at the back region are less detrimental for the efficiency of the devices. For varying composition absorbers there is a general decrease in JSC, to 15.2 mA/cm2 for the highest Zn-excess precursor (Zn/Sn = 1.73), and VOC is also significantly affected, dropping to 270 mV for the same sample. In this case deterioration of the optoelectronic parameters correlates with the formation of the nanocrystalline layer observed in Figs. 2D, 2E and 2F due to the grain growth inhibition, likely related to a more uniform distribution of binary ZnSe secondary phase. VI. CONCLUSION CZTSe based solar cells were prepared by a two step process and the impact of the composition of precursor and temperature of the annealing was studied. Both parameters have a significant effect on the VOC of the devices and thus on the conversion efficiency. We show that ZnSe and SnSe are the main secondary phases in this system, together with the appearance of a mode slightly red shifted with respect to the main A1 CZTSe mode, which we attribute to disorder induced effects at high annealing temperatures. Both the increasing Zn/Sn ratio and the annealing temperature deteriorate the optoelectronic parameters, most notably the VOC. These effects are likely related to the deterioration of the grain structure with the formation of voids at the back absorber region, as well as the favored formation of ZnSe secondary phase. As a preliminary result, and with a first optimization of the processes, we obtain a solar cell with 4.8% conversion efficiency for the pure selenide compound.

ACKNOWLEDGEMENT The research leading to these results has received funding from the People Programme (Marie Curie Actions) of the European Union's Seventh Framework Programme FP7/20072013/ under REA grant agreement n°269167). The research was also partially supported by MINECO, project KEST-PV (ref. ENE2010-121541-C03-1/2). Authors from IREC and the University of Barcelona belong to the M-2E (Electronic Materials for Energy) Consolidated Research Group and the XaRMAE Network of Excellence on Materials for Energy of the “Generalitat de Catalunya”. E.S. thanks the MINECO, Subprogram Ramón y Cajal (ref. RYC 2011-09212) and V.I Subprogram Juan de la Cierva (ref. JCI-2011-10782). REFERENCES Fig. 4. Illuminated AM1.5 J-V curves with (A) varying annealing temperature for precursor Zn/Sn = 1.24; and (B) with varying Zn/Sn ratio annealed at 450 ºC

978-1-4673-0066-7/12/$26.00 ©2011 IEEE

[1] D.A.R. Barkhouse, O. Gunawan, T. Gokmen, T.K. Todorov, and D.B. Mitzi, “Device characteristics of a 10.1% hydrazine-

002683

processed Cu2ZnSn(Se,S)4 solar cell,” Prog. Photovolt: Res. Appl. vol. 20, pp. 6-11, 2011. [2] T. Todorov, K. Reuter, and D. Mitzi, “High-Efficiency Solar Cell with Earth-Abundant Liquid-Processed Absorber,” Adv. Mater. vol. 22, pp. 1–4, 2010. [3] A. Redinger, D.M. Berg, P.J. Dale, and S. Siebentritt, “The Consequences of Kesterite Equilibria for Efficient Solar Cells,” J. Am. Chem. Soc. vol. 133, pp. 3320–3323, 2011. [4] H. Katagiri, K. Jimbo, S. Yamada, T. Kamimura, W.S. Maw, T. Fukano, T. Ito, and T. Motohiro, “Enhanced Conversion Efficiencies of Cu2ZnSnS4-Based Thin Film Solar Cells by Using Preferential Etching Technique,” Appl. Phys. Express vol. 1, pp. 041201-1-2, 2008. [5] Q. Guo, G.M. Ford, W-Ch. Yang, B. C. Walker, E. A. Stach, H. W. Hillhouse, and R. Agrawal, “Fabrication of 7.2% Efficient CZTSSe Solar Cells Using CZTS Nanocrystals,” J. Am. Chem. Soc. vol. 132, pp. 17384–17386, 2010. [6] A. Fairbrother, E. García-Hemme, V. Izquierdo-Roca, X. Fontané, F. A. Pulgarín-Agudelo, O. Vigil-Galán, A. PérezRodríguez, and E. Saucedo, “Development of a selective chemical etch to improve the conversion efficiency of Zn-rich Cu2ZnSnS4 solar cells,” J. Am. Chem. Soc. vol. 134, pp. 8018– 8021, 2012. [7] C. Platzer-Björkman, J. Scragg, H. Flammersberger, T. Kubart, and M. Edoff, “Influence of precursor sulfur content on film formation and compositional changes in Cu2ZnSnS4 films and solar cells,” Sol. Energy Mater. Sol. Cells vol. 98, pp. 110-117, 2012. [8] I. Repins, C. Beall, N. Vora, C. DeHart, D. Kuciauskas, P. Dippo, B. To, J. Mann, W- Ch. Hsu, A. Goodrich, and R. Noufi, “Coevaporated Cu2ZnSnSe4 films and devices,” Solar Ener. Mat. Solar Cells vol. 101, pp. 154-159, 2012.

978-1-4673-0066-7/12/$26.00 ©2011 IEEE

[9] G. Zoppi, I. Forbes, R.W. Miles, P.J. Dale, J.J. Scragg, and L.M. Peter, “Cu2ZnSnSe4 thin film solar cells produced by selenisation of magnetron sputtered precursors,” Prog. Photovolt: Res. Appl. vol. 19, pp. 93–96, 2011. [10] This intermediate phase was proposed for the sulfide compound in, for example: S. Schorr, A. Weber, V. Honkimäki, and H.-W. Schock, “In-situ investigation of the kesterite formation from binary and ternary sulphides,” Thin Solid Films vol. 517, pp. 2461–2464, 2009. [11] G. Suresh Babu, Y.B. Kishore Kumar, P. Uday Bhaskar, and S. Raja Vanjari, “Effect of Cu/(Zn+Sn) ratio on the properties of coevaporated Cu2ZnSnSe4 thin films,” Solar Ener. Mat. Solar Cells vol. 94, pp. 221-226, 2010. [12] K. Tanaka, Y. Fukui, N. Moritake, and H. Uchiki, “Chemical composition dependence of morphological and optical properties of Cu2ZnSnS4 thin films deposited by sol-gel sulfurization and Cu2ZnSnS4 thin film solar cell efficiency,” Solar Ener. Mat. Solar Cells vol. 95, pp. 838-842, 2011. [13] A. Redinger, K. Hönes, X. Fontané, V. Izquierdo-Roca, E. Saucedo, N. Valle, A. Pérez-Rodriguez, and S. Siebentritt, “Detection of a ZnSe secondary phase in coevaporated Cu2ZnSnSe4 thin films,” Appl. Phys. Lett. vol. 98, pp. 101907, 2011. [14] G. Marcano, C. Rincón, S. A. López, G. Sánchez-Pérez, J. L. Herrera-Pérez, J. G. Mendoza-Álvarez, and P. Rodríguez, “Raman spectrum of monoclinic semiconductor Cu2SnSe3,” Solid State Comm. vol. 151, pp. 84-86, 2011. [15] X. Fontané, V. Izquierdo-Roca, E. Saucedo, S. Schorr, V.O. Yukhymchuk, M. Ya. Valakh, A. Pérez-Rodríguez, and J. R. Morante, “Vibrational properties of stannite and kesterite type compounds: Raman scattering analysis of Cu2(Fe,Zn)SnS4,” J. Alloys and Compd., accepted DOI: 10.1016/j.jallcom.2012.06.042.

002684