- Email: [email protected]
Structural and optical properties of CuGaSe2 layers grown by the hot wall epitaxy method
ARTICLE IN PRESS
Journal of Crystal Growth 310 (2008) 2717–2723 www.elsevier.com/locate/jcrysgro
Structural and optical properties of CuGaSe2 layers grown by the hot wall epitaxy method S.H. Youa, K.J. Honga,, T.S. Jeongb, C.J. Younb a
Department of Physics, Chosun University, Gwangju 501-759, Republic of Korea School of Semiconductor and Chemical Engineering, Chonbuk National University, Jeonju 561-756, Republic of Korea
b
Received 5 December 2007; received in revised form 6 February 2008; accepted 8 February 2008 Communicated by M. Schieber Available online 19 February 2008
Abstract Copper gallium selenide (CuGaSe2, CGS) layers were grown by the hot wall epitaxy method. The optimum temperatures of the substrate and source for growth turned out to be 450 and 610 1C, respectively. The CGS layers were epitaxially grown along the /1 1 0S direction and consisted of Ga-rich components indicating the slight stoichiometric deviations. Based on the absorption measurement, the band-gap variation of CGS was well interpreted by the Varshni’s equation. The band-gap energies at low temperatures, however, had a higher value than those of other CGS. It suggests that the band-gap increase is influenced by the slightly Ga-rich composition. From the low-temperature photoluminescence experiment, sharp and intensive free- and bound-exciton peaks were observed. By analyzing these emissions, a band diagram of the observed optical transitions was obtained. From the solar cell measurement, an 11.17% efficiency on the n-CdS/p-CGS junction was achieved. r 2008 Elsevier B.V. All rights reserved. PACS: 68.60.p; 78.20.e; 78.55.m; 81.15.z Keywords: A1. Characterization; A3. Hot wall epitaxy; B2. Semiconducting ternary compounds; B3. Solar cells
1. Introduction The I–III–VI2 compounds have been known to be ternary analogues of the II–VI compounds. These compounds are crystallized with the chalcopyrite structure, which is closely related to the zinc structure [1]. These materials have received considerable attention in recent years because of their applications in photovoltaic devices [2,3]. They can be used as an environmental-friendly material for the Cd-free buffer layer as an absorbance of the solar cell [4,5]. Among these compounds, copper gallium selenide (CuGaSe2, CGS) has an energy gap of 1.68 eV at room temperature and it is expected to be a promising material for red-light-emitting devices. Currently, solar cells based on CGS have achieved efficiencies of 9.5% and 9.7%, as absorbers of thin film and single Corresponding author. Tel.: +82 62 230 6637; fax: +82 62 234 4326.
E-mail address: [email protected] (K.J. Hong). 0022-0248/$ - see front matter r 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.jcrysgro.2008.02.011
crystal shapes, respectively [6,7]. Despite this technological progress, many of the fundamental properties of CGS are still not fully understood. The physical properties, however, strongly relate to native defects due to deviations of stoichiometry. Therefore, reduction of stoichiometry deviations during crystal growth is required. As such, CGS layers have been grown by many different techniques, such as metalorganic chemical vapor deposition (MOCVD), metalorganic vapor-phase epitaxy (MOVPE), physical vapor deposition (PVD), RF sputtering, and flash evaporation [8–12]. In fact, hot wall epitaxy (HWE), which has been used to grow high-purity ZnSe epitaxial layers at low temperatures [13], is one of the better methods for CGS growth. Thus, HWE has been especially designed to grow epilayers under conditions of near thermodynamic equilibrium [14]. CGS growth by the HWE method, however, was rarely reported before now. In this paper, the CGS layers, which could be used for absorbance in solar cells, were grown on a GaAs substrate
ARTICLE IN PRESS S.H. You et al. / Journal of Crystal Growth 310 (2008) 2717–2723
2718
by the HWE method. The grown CGS layers were investigated for their structural and optical characterization by means of X-ray diffraction (XRD), photoluminescence (PL), and absorption spectroscopy. From these results, we discuss the structural and optical properties of the CGS layers. Furthermore, CGS-based solar cells were investigated. 2. Experimental procedure Prior to the layer growth, polycrystalline CGS was formed as follows. A quartz tube was sequentially cleaned with trichlorethylene (TCE), acetone, methanol, and deionized water. The starting materials were 6 N purity shot-types of Cu, Ga, and Se. After the materials were weighed in the mole fraction of each element, they were sealed in the quartz tube while maintaining a vacuum atmosphere. The sealed ampoule was placed in a synthesis furnace and was continually rotated at a rate of one revolution per minute. In order to avoid an explosion of the ampoule due to the Se vapor pressure, the temperature of the ampoule was gradually increased to 1180 1C, and was then maintained for 48 h. To grow the CGS layer, a polycrystalline CGS ingot was used for the HWE source. The CGS layers were grown on semi-insulating GaAs (1 0 0) by the HWE method using the grown CGS ingot as a source material. Fig. 1 presents a diagram of the HWE apparatus used for the CGS growth and solar-cell sample. In this experiment, the CGS layer was grown by using furnace (a) as shown in Fig. 1. Prior to growing the CGS layers, the GaAs substrate was cleaned ultrasonically for 1 min in successive baths of TCE, acetone, methanol and 2-propanol, and then etched for 1 min in a solution of H2SO4:H2O2:H2O (5:1:1). The substrate was degreased in organic solvents and rinsed with deionized water (18.2 MO). After the substrate was dried off, it was immediately loaded onto the substrate holder as shown in
Substrate Substrate
GaAs & ITO Glass Heater
Source 2 CdS Powder Source 1 Poly CuGaSe2
Reservoir
Furnace (b)
Furnace (a)
Fig. 1. Diagram of the HWE apparatus used for the CGS growth and solar-cell sample.
Fig. 1 and was annealed at 580 1C for 20 min to remove the residual oxide on the surface of the substrate. To obtain optimum growth conditions, the grown CGS layers were estimated through PL and XRD measurements. The PL measurement of the sample was performed at 10 K in a low-temperature cryostat during the excitement of the He–Ne laser (632.8 nm, 50 mW). The thickness, growth orientation, and stoichiometric composition of the CGS layers were measured by using an a-step profilometer, XRD, and an energy-dispersive X-ray spectrometer (EDS), respectively. Thus, the Hall effect experiment was carried out using the van der Pauw method. Optical absorption was used to measure the band-gap energy using an UV–VIS–NIR spectrophotometer for a range of 670– 750 nm with the temperature varying from 10 to 293 K. Also, the CdS/CGS/ITO layer for the solar-cell specimen was grown using HWE. The substrate of the pyrex glass, which was sprayed with indium tin oxide (ITO), was moved successively to the top from furnace (a) to (b), as shown in Fig. 1. The source materials 1 and 2, as shown in Fig. 1, were polycrystalline CGS ingot and CdS powder (rare metallic, 6 N), respectively. After the CGS layer on the ITO glass was grown in furnace, (a), the CGS/ITO layer was immediately moved to furnace (b). Then, the CdS/CGS/ ITO layer was grown by using heating source 2. Also, the solar-cell measurement was conducted by illuminating the tungsten halogen lamp at 80 mW/cm2. 3. Results and discussion 3.1. Growth and structural properties After the heat treatment of the substrate surface, the CGS layers were grown by varying the substrate temperature from 410 to 470 1C, while the source temperature was fixed at 610 1C. At this time, the crystal quality of the grown CGS layers was evaluated by measuring the PL spectra at 10 K. Thus, in order to evaluate the crystal quality, the FWHM value of XRD assisted. Exciton emissions were used to predict the crystal-quality criterion of the grown layer because the exciton could only be observed in a less defective crystal at a low temperature. Here, the emissions of free excitons (EFX) and bound excitons (EBX) used the optimum growth conditions. The grown layers are considered to have good crystal quality when the measured PL spectra show that the intensity of EFX and EBX tend to increase. As shown in Table 1, the EFX and the EBX peaks in the layer, which were grown while the substrate temperature was kept at 450 1C, had the highest intensity of all other samples. On the contrary, the narrowest FWHM, which indicated high-crystal quality, obtained from the XRD experiment was observed in the CGS layer grown at 450 1C as well. This observation indicates that the CGS layer grown at 450 1C has a very high quality because the exciton emission can be observed only under the condition of a long-range Coulomb coupling between the electron and the hole. Therefore,
ARTICLE IN PRESS S.H. You et al. / Journal of Crystal Growth 310 (2008) 2717–2723
2719
Table 1 Comparison of the PL intensities and the XRD as a function of substrate temperature (Arbitrary units)
Table 2 Composition ratios of each element on the synthesized polycrystalline and the CGS layer analyzed by the EDS measurement (Units: %)
Substrate temperature (1C)
Elements
EFX
EBX
FWHM (arcsec)
410 430 450 470
5.75 10.75 26.19 21.35
7.16 16.84 37.16 30.90
1827 1163 457 651
Intensities of PL at 10 K
XRD (1y)
Grown layer
Before synthesis
After synthesis
Before growth
After growth
25 25 50
24.507 25.618 49.875
24.507 25.618 49.875
24.237 26.281 49.482
beneficial for a CGS layer to be fabricated for solar energy conversion. Also, the CGS (1 1 0) peak is located at 2y of 40.4891 and its FWHM value is 0.2051. This FWHM value is similar to that of the CGS grown from MOCVD by Orsal et al. [16]. Therefore, it should be noted that the layers grown by the HWE method are high-quality crystalline. Thus, the mean crystallite size of the layer can be extracted from
GaAs
(400)
3x104
Intensity (arb.units)
Cu Ga Se
Synthesized polycrystalline
2x104
D ¼ 0:94l=ðB cos yÞ; CuGaSe2
(110)
1x104
0 30
40
50
60
70
Diffraction angle (2θ) Fig. 2. XRD o2y scans of the CGS layer grown under optimized conditions.
the optimum temperature of the substrate was found to be 450 1C, while the source temperature was 610 1C. At this time, the thickness and the growth rate of the layer were 2.5 mm and 0.83 mm/h, respectively. Fig. 2 presents the XRD o2y scans of the CGS layers grown at the source and substrate temperatures of 610 and 450 1C, respectively. Thus, the obtained curve corresponds to the diffraction peaks of the CGS (1 1 0) and GaAs (0 0 4). This means that the orientation of the layer grown on the GaAs (1 0 0) substrate was converted to the (1 1 0) plane. This phenomenon was also observed in the CdTe epilayer grown on the GaAs (1 0 0). Faurie et al. [15] reported that the orientation of the CdTe epilayer was related to the pre-annealing process used to remove the residual oxide on the surface of the substrate. They concluded that the growth of the CdTe (1 0 0) or (1 1 1) planes on the GaAs (1 0 0) was possible by controlling the different annealing temperatures and times of the substrate. On the other hand, the observation of only one peak of the CGS (1 1 0) indicates that the CGS layer was grown epitaxially along the /1 1 0S direction onto the GaAs (1 0 0) substrate. This orientation is suggested to be
(1)
where l, y, and B are the X-ray wavelength (0.15405 nm), the Bragg diffraction angle, and the FWHM value on (1 1 0) peak in radians, respectively [17]. From Eq. (1), the mean crystallite size was estimated to be about 81 nm. Table 2 presents the composition ratios of each element during the growth process of the CGS layer, analyzed by the EDS measurement. As shown in Table 2, the grown layer consisted of Cu-poor, Ga-rich, and slightly Se-poor components. From the Hall effect measurement, the grown CGS layer having Ga-rich component was p-type presumably due to slight stoichiometric deviations originating from an excess of copper vacancies (VCu). From the Hall effect measurement, the carrier density and Hall mobility of the layer at 293 K were determined to be 4.87 1017 cm3 and 129 cm2/V s, respectively. 3.2. Optical properties Fig. 3 shows the optical absorption spectra obtained in the temperature range of 10–293 K. To identify the bandgap energy for the CGS layer, we carefully examined the relationship between the optical absorption coefficient (a) and the incident photon energy (hn) from the optical absorption measurements in Fig. 3. The relationship for a direct band-gap between hn and a is given by ðahvÞ2 ðhv E g Þ:
(2) 2
According to Eq. (2), (ahn) linearly depends upon the photon energy. From the plots of (ahn)2 versus photon energy for different temperatures, the band gaps were determined by extrapolating the linear portions of the respective curves to (ahn)2 ¼ 0. Fig. 4 displays the bandgap variation of the CGS layer as a function of temperature. The temperature dependence of the band-gap energy
ARTICLE IN PRESS S.H. You et al. / Journal of Crystal Growth 310 (2008) 2717–2723
2720
1.8
Table 3 Comparison of band gap between our data and published values
76
Optical density (arb.units)
8
5
4
3
2
Sample
1
Band gap
Crystal growth
Temperature (K)
Energy (eV)
Type
Method
1 2 3 4 5 6 7
77 77 10 4.2 4.2 0 77
1.746 1.745 1.73 1.7347 1.75 1.7310 1.7872
Bulk Bulk Layer Bulk Layer Layer Layer
CVT CVT MOVPE VGF MOCVD MBE HWE
7 7 7
10 4.2 0
1.7995 1.7997 1.7998
Reference
1.2 1. 293 K 2. 250 K 3. 200 K 4. 150 K 5. 100 K 6. 77 K 7. 50 K 8. 10 K
0.6
0.0 660
690
720 Wavelength (nm)
750
Fig. 3. Optical absorption spectra according to the variation of the measure temperature.
1.9 Experimental value Fitted energy
Energy (eV)
1.8
1.7 Eg(0) : 1.7998 eV a : 8.7489x10-4 eV/K b : 335 K
1.6 0
100 200 Temperature (K)
300
in our experiment is well fitted numerically by the following formula [18,19]: E g ðTÞ ¼ E g ð0Þ aT 2 =ðT þ bÞ,
with the reported value of 394 K [20]. Thus, the band-gap energy of the CGS obtained at 293 K is estimated to be 1.6802 eV. Even though the band-gap variation of CGS is well satisfied by Eq. (3), our values are slightly larger than that of reported values as shown in Table 3. These differences in the low-temperature region are about 40–60 meV. We suggest that this cause is responsible for a composition ratio of Ga. As the Ga concentration in CGS increases, it is known that the electronic transitions shift towards higher energies [27]. As shown in Table 2, the CuXGa ratio in CGS is about 0.922. Therefore, it suggests that the band-gap increase is influenced by the slightly Garich composition. The absorption-experiment itself can be a problem, however. This method is not accurate due to the difficulty in defining the position of the absorption edge. Fig. 5 shows a typical PL spectrum of the CGS layer measured at 10 K. As shown in Fig. 5, an intense and sharp peak was observed at 747.2 nm (1.6593 eV). This peak is associated with the neutral donor-bound exciton (D1, X), caused by the recombination from free exciton to neutral donor. Also, the EFX at 741.6 nm (1.6718 eV) appears on the shoulder toward the short-wavelength region of the (D1, X) emission. This peak shows a low excitonic emission indicating a redshift due to biaxial tensile stress in the epitaxial film. From the EFX emission, the binding energy, E FXb , is obtained by E FXb ¼ E g ð10Þ EðE FX Þ;
Fig. 4. Band-gap variation of the CGS layer as a function of temperature.
(3)
where a is a constant and b is approximately the Debye temperature. Also, Eg(0) is the band-gap energy at 0 K, which is estimated to be 1.7998 eV. When a and b are taken to be 8.7489 104 eV/K and 335 K, respectively, the curve plotted by Eq. (3) closely fits the experimental values. Our result on the Debye temperature is also in close agreement
[21] [22] [23] [24] [25] [26] This work
(4)
where Eg(10) is the band-gap energy at 10 K. The value of Eg(10) was calculated to be 1.7995 eV using Eq. (3) and the E FXb is obtained to be 0.1277 eV. The E FXb value is about 10 times larger than the reported one [23]. As stated above, it suggests that the large value of E FXb is caused by the increase of Eg due to the Ga-rich component and the inaccuracy of the absorption measurement. Also, the shoulder peak toward the long-wavelength region of the (D1, X) emission appears at 755.5 nm (1.6411 eV). This peak is suggested to be due to the neutral acceptor-bound exciton, (A1, X). This bound exciton has been known to be related to the recombination of the free exciton to the
ARTICLE IN PRESS S.H. You et al. / Journal of Crystal Growth 310 (2008) 2717–2723
2721
tion process on the CGS layer. Fig. 6 plots the band diagram of the observed optical transitions including the band-gap energy of 10 K. In this band diagram, the Eg(10)
100 (Do, X)
EFX FX
PL intensity (arb.units)
Conduction band (Ao, X)
6
Do
7
(Do, Ao)
50
1 (Do, Ao) replicas
2
3
5
4
SA SA Eg (10)
EFX
(Do, X)
(Ao, X)
(Do, Ao) 0 700
800
900 Wavelength (nm)
1000
Fig. 5. PL spectrum of the CGS layer measured at 10 K.
Valence band
(5)
Therefore, the E FXb values of the (D1, X) and the (A1, X) emissions are calculated to be 12.5 and 30.7 meV, respectively. Thus, the neutral donor level is estimated to be 0.1402 eV ( ¼ 0.1277+0.0125). This result is close to the donor level for Ga-annealed CGS [28]. Therefore, the neutral donor level is ascribed to the Se vacancies (VSe) and/or Ga in antisite (GaCu) which is the most probable defect in CGS due to the Ga-rich component [28]. It also suggests that the neutral acceptor level acting as an acceptor is caused by the VCu due to a slight stoichiometric deviation [25]. In addition, the neutral donor–acceptor pair, (D1, A1), emission was observed at 763.9 nm (1.6231 eV). The (D1, A1) emission occurred due to the interaction between the neutral donor and the neutral acceptor states in the energy band gap. Thus, replica peaks in the (D1, A1) emission are observed at 770.2 (1.6098 eV), 776.6 (1.5965 eV), 783.1 (1.5833 eV), and 789.5 nm (1.5704 eV). These peaks are considered to be associated with the longitudinal optical (LO) phonon replicas of the (D1, A1) emission, because the energy difference between these peaks is equal to a multiple of the LO phonon energy of B2-modes such as 13.3 (1LO), 26.6 (2LO), 39.8 (3LO), and 52.7 meV (4LO) [29]. The relatively low and broad peak caused by the self-activated (SA) emission was observed at 919.6 nm (1.3483 eV) in the longer-wavelength region. This emission strongly relates to the native defects formed in the CGS layer. Therefore, based on these PL results, we schemed out a band diagram of the recombina-
Fig. 6. Band diagram of the optical transitions on the CGS layer measured at 10 K. 1. Band-gap energy of CGS at 10 K: 1.7995 eV. 2. Free exciton energy: 1.6718 eV. 3. Neutral donor-bound exciton energy: 1.6593 eV. 4. Neutral acceptor-bound exciton energy: 1.6411 eV. 5. Neutral donor–acceptor pair energy: 1.6231 eV. 6. Binding energy of free exciton: 0.1277 eV. 7. Neutral donor levels: 0.1402 eV. 8. Neutral acceptor levels: 0.0307 eV.
30 Light 20
Current density (mA/cm2)
neutral acceptor. The binding energy (E FXb ) of the EBX emission is obtained by E BXb ¼ EðE FX Þ EðD ; XÞ or EðA ; XÞ:
Ao
8
In n-CdS p-CuGaSe22 ITO Glass ITO
In
10 Dark Illuminated
Vmp 0 0.0
0.4
0.2
0.6
0.8
Voltage (V) -10
Vmp= 0.41 V Jmp= 21.8 mA/cm2
-20
Jmp
Pmax
-30 Fig. 7. Dark and illuminated current density–voltage characteristics of solar cells. (Here, the subfigure shows a diagram of the n-CdS/p-CGS junction. Also, Jmp and Vmp are the current density and voltage values at the maximum power point Pmax, respectively.)
ARTICLE IN PRESS 2722
S.H. You et al. / Journal of Crystal Growth 310 (2008) 2717–2723
Table 4 J–V parameters of the n-CdS/p-CGS solar-cells Cell
Voc (V)
Jsc (mA/cm2)
FF (%)
Efficiency (%)
Reference
CdS/CGS layer CdS/CGS polycrystalline CdS/CGS layer
0.95 0.59 0.51
14.88 8.14 22.90
70.79 44.0 76.53
9.53 2.0 11.17
[6] [31] This work
is the band-to-band transition corresponding to the bandgap energy. Also, the (D1, A1) emission is due to the recombination transition between the neutral donor and the neutral acceptor states in the band-gap energy. 3.3. Solar cells CGS-based solar cells were investigated. Fig. 7 shows the dark and illuminated current density–voltage (J–V) characteristics of solar cells. Also, the sub figure in Fig. 7 shows a diagram of the n-CdS/p-CGS junction. As shown in the subfigure, it is fabricated to a p–n.junction between the p-CGS layer as the absorber material and the n-CdS layer [30]. The solar-cell parameters obtained from the J–V characteristic are given in Table 4. From these results, the open-circuit voltage (Voc) and short-circuit current–density (Jsc) were observed to be 0.51 V and 22.9 mA/cm2, respectively. As Table 4 shows, this means that the achievement of 11.17% efficiency is an improvement in comparison to the previous devices [6–31]. Also, it is superior to the best single-crystal CGS device to date (9.7% active area efficiency) [7]. 4. Conclusions The CGS layers, which are now charming as solar-cell materials, were grown by the HWE method. From the results of the PL and XRD measurements, the optimum growth temperatures of the substrate and the source turned out to be 450 and 610 1C, respectively. Also, the CGS layers were confirmed to be the epitaxially grown layer along the /1 1 0S direction onto the GaAs (1 0 0) substrate. The carrier density and Hall mobility of the layer at 293 K were determined to be 4.87 1017 cm3 and 129 cm2/V s, respectively. Thus, as is evident in the EDS measurement, the grown CGS layers consisted of Ga-rich components indicating slight stoichiometric deviations. The band-gap variation of CGS extracted from the absorption measurement was well fitted by Eg(T) ¼ 1.79988.7489 104T2/ (T+335). The band-gap energy at a low temperature, however, was slightly larger than that of other CGS. It suggests that the band-gap increase is influenced by slightly Ga-rich compositions. From the low-temperature PL measurement, the EFX, the (D1, X), and the (A1, X) emissions, which can be observed at the high-quality crystalline, appeared at 1.6718, 1.6593, and 1.6411 eV, respectively. Also, the (D1, A1) and its LO phonon replicas emissions, which are due to the interaction between the
neutral donor and the neutral acceptor states, were observed at the wavelength region between 770.2 and 789.5 nm. Based on these PL results, a band diagram of the observed optical transitions was obtained. From the solar cell of the n-CdS/p-CGS junction, an efficiency of 11.17% has been reached. Acknowledgment This study was supported by research funds from Chosun University, 2007. References [1] J.L. Shay, J.H. Wernick, Ternary Chalcopyrite Semiconductors: Growth, Electronic Properties, and Applications, Pergamon, Oxford, 1975 (Chapter 4). [2] W.N. Shafarman, J. Zhu, Thin Solid Films 473 (2000) 361. [3] M.A. Contreras, B. Egaas, K. Ramanatan, J. Hiltner, A. Swartzlander, F. Hasoon, R. Nuofi, Prog. Photovoltaics 7 (1999) 311. [4] D. Hariskos, R. Herbeholz, M. Ruckh, U. Ruhle, R. Schaffler, H.W. Schock, in: W. Freielsleben, W. Palz, H.A. Ossenbrink, P. Helm (Eds.), Proceedings of the 13th European Photovoltaic Solar Energy Conference and Exhibition Acropolis, Convention Center, Nice, France, October 23–27, 1995, p. 1995. [5] K. Kushiya, S. Kuriyagawa, T. Nii, I. Sugiyama, T. Kase, M. Sato, H. Takeshita, in: W. Freielsleben, W. Palz, H.A. Ossenbrink, P. Helm (Eds.), Proceedings of the 13th European Photovoltaic Solar Energy Conference and Exhibition Acropolis, Convention Center, Nice, France, October 23–27, 1995, p. 2016. [6] D.L. Young, J. Keane, A. Duda, J.A.M. AbuShama, C.L. Perkins, M. Romero, R. Noufi, Prog. Photovoltaics 11 (2003) 535. [7] M. Saad, H. Riazi, E. Bucher, M. Ch. Lux-Steiner, Appl. Phys. A: Mater. Sci. Process 62 (1996) 181. [8] I.E. Beckers, U. Fiedeler, S. Siebentritt, M. Ch. Lux-Steiner, Thin Solid Films 431 (2003) 205. [9] S. Siebentritt, I. Beckers, T. Riemann, J. Christen, A. Hoffimann, M. Dworzak, Appl. Phys. Lett. 86 (2005) 091909. [10] S. Schuler, S. Siebentritt, S. Nishiwaki, N. Rega, J. Beckmann, S. Brrehme, M. Ch. Lux-Steiner, Phys. Rev. B 69 (2004) 045210. [11] J.H. Yun, R.B.V. Chalapathy, S.J. Ahn, J.C. Lee, K.H. Yoon, J. Korean Phys. Soc. 50 (2007) 1794. [12] M. Rusu, P. Gashin, A. Simashkevich, Sol. Energy Mater. Sol. Cells. 70 (2001) 175. [13] T.S. Jeong, P.Y. Yu, K.J. Hong, T.S. Kim, C.J. Youn, Y.D. Choi, K.S. Lee, B. O, M.Y. Yoon, J. Crystal Growth 249 (2003) 9. [14] A. Lopez-Otero, Thin Solid Films 49 (1987) 3. [15] J.P. Faurie, C. Hsu, S. Sivananthan, X. Chu, Surf. Sci. 168 (1986) 473. [16] G. Orsal, F. Mailly, N. Romain, M.C. Artaud, S. Rushworth, S. Duchemin, Thin Solid Films 361 (2000) 135. [17] B.D. Cullity, Elements of X-ray Diffractions, Addition-Wesley, Reading, MA, 1978, p. 102. [18] Y.P. Varshni, Physica 34 (1967) 149.
ARTICLE IN PRESS S.H. You et al. / Journal of Crystal Growth 310 (2008) 2717–2723 [19] K.J. Hong, T.S. Jeong, C.J. Yoon, Y.J. Shin, J. Crystal Growth 218 (2000) 19. [20] S. Siebentritt, Thin Solid Films 480–481 (2005) 312. [21] I.A. Ponomareva, A. Yu, I.V. Bodnar, Phys. Solid State 49 (2007) 23. [22] B. Tell, P.M. Bridenbaugh, Phys. Rev. B 12 (1978) 3330. [23] A. Bauknecht, S. Siebentritt, J. Albert, M. Ch. Lux-Steiner, J. Appl. Phys. 89 (2001) 4391. [24] K. Yoshino, M. Sugiyama, D. Maruoka, S.F. Chichibu, H. Komaki, K. Umeda, T. Ikari, Physica B 302–303 (2001) 357. [25] G. Orsal, F. Mailly, M.C. Artaud-Gillet, S. Duchemin, Thin Solid Films 387 (2001) 198.
2723
[26] A. Yamada, Y. Makita, S. Niki, A. Obara, P. Fons, H. Shibata, M. Kawai, S. Chichibu, H. Nakanishi, J. Appl. Phys. 79 (1996) 4318. [27] P.D. Paulson, R.W. Birkmire, W.N. Shafarman, J. Appl. Phys. 94 (2003) 879. [28] J.H. Scho¨n, O. Schenker, H. Riazi-Nejad, K. Friemelt, Ch. Kloc, E. Bucher, Phys. Stat. Sol. (a) 161 (1997) 301. [29] D. Papadimitriou, N. Esser, C. Xue, Phys. Stat. Sol. (b) 242 (2005) 2633. [30] N. Romeo, G. Sberveglieri, L. Tarricone, Appl. Phys. Lett. 30 (1977) 108. [31] S. Siebentritt, A. Bauknecht, A. Gerghard, U. Fiedeler, T. Kampschulte, S. Schuler, W. Harneit, S. Brehme, J. Albert, M. Ch. Lux-Steiner, Sol. Energy Mater. Sol. Cells 67 (2001) 129.
Journal of Crystal Growth 310 (2008) 2717–2723 www.elsevier.com/locate/jcrysgro
Structural and optical properties of CuGaSe2 layers grown by the hot wall epitaxy method S.H. Youa, K.J. Honga,, T.S. Jeongb, C.J. Younb a
Department of Physics, Chosun University, Gwangju 501-759, Republic of Korea School of Semiconductor and Chemical Engineering, Chonbuk National University, Jeonju 561-756, Republic of Korea
b
Received 5 December 2007; received in revised form 6 February 2008; accepted 8 February 2008 Communicated by M. Schieber Available online 19 February 2008
Abstract Copper gallium selenide (CuGaSe2, CGS) layers were grown by the hot wall epitaxy method. The optimum temperatures of the substrate and source for growth turned out to be 450 and 610 1C, respectively. The CGS layers were epitaxially grown along the /1 1 0S direction and consisted of Ga-rich components indicating the slight stoichiometric deviations. Based on the absorption measurement, the band-gap variation of CGS was well interpreted by the Varshni’s equation. The band-gap energies at low temperatures, however, had a higher value than those of other CGS. It suggests that the band-gap increase is influenced by the slightly Ga-rich composition. From the low-temperature photoluminescence experiment, sharp and intensive free- and bound-exciton peaks were observed. By analyzing these emissions, a band diagram of the observed optical transitions was obtained. From the solar cell measurement, an 11.17% efficiency on the n-CdS/p-CGS junction was achieved. r 2008 Elsevier B.V. All rights reserved. PACS: 68.60.p; 78.20.e; 78.55.m; 81.15.z Keywords: A1. Characterization; A3. Hot wall epitaxy; B2. Semiconducting ternary compounds; B3. Solar cells
1. Introduction The I–III–VI2 compounds have been known to be ternary analogues of the II–VI compounds. These compounds are crystallized with the chalcopyrite structure, which is closely related to the zinc structure [1]. These materials have received considerable attention in recent years because of their applications in photovoltaic devices [2,3]. They can be used as an environmental-friendly material for the Cd-free buffer layer as an absorbance of the solar cell [4,5]. Among these compounds, copper gallium selenide (CuGaSe2, CGS) has an energy gap of 1.68 eV at room temperature and it is expected to be a promising material for red-light-emitting devices. Currently, solar cells based on CGS have achieved efficiencies of 9.5% and 9.7%, as absorbers of thin film and single Corresponding author. Tel.: +82 62 230 6637; fax: +82 62 234 4326.
E-mail address: [email protected] (K.J. Hong). 0022-0248/$ - see front matter r 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.jcrysgro.2008.02.011
crystal shapes, respectively [6,7]. Despite this technological progress, many of the fundamental properties of CGS are still not fully understood. The physical properties, however, strongly relate to native defects due to deviations of stoichiometry. Therefore, reduction of stoichiometry deviations during crystal growth is required. As such, CGS layers have been grown by many different techniques, such as metalorganic chemical vapor deposition (MOCVD), metalorganic vapor-phase epitaxy (MOVPE), physical vapor deposition (PVD), RF sputtering, and flash evaporation [8–12]. In fact, hot wall epitaxy (HWE), which has been used to grow high-purity ZnSe epitaxial layers at low temperatures [13], is one of the better methods for CGS growth. Thus, HWE has been especially designed to grow epilayers under conditions of near thermodynamic equilibrium [14]. CGS growth by the HWE method, however, was rarely reported before now. In this paper, the CGS layers, which could be used for absorbance in solar cells, were grown on a GaAs substrate
ARTICLE IN PRESS S.H. You et al. / Journal of Crystal Growth 310 (2008) 2717–2723
2718
by the HWE method. The grown CGS layers were investigated for their structural and optical characterization by means of X-ray diffraction (XRD), photoluminescence (PL), and absorption spectroscopy. From these results, we discuss the structural and optical properties of the CGS layers. Furthermore, CGS-based solar cells were investigated. 2. Experimental procedure Prior to the layer growth, polycrystalline CGS was formed as follows. A quartz tube was sequentially cleaned with trichlorethylene (TCE), acetone, methanol, and deionized water. The starting materials were 6 N purity shot-types of Cu, Ga, and Se. After the materials were weighed in the mole fraction of each element, they were sealed in the quartz tube while maintaining a vacuum atmosphere. The sealed ampoule was placed in a synthesis furnace and was continually rotated at a rate of one revolution per minute. In order to avoid an explosion of the ampoule due to the Se vapor pressure, the temperature of the ampoule was gradually increased to 1180 1C, and was then maintained for 48 h. To grow the CGS layer, a polycrystalline CGS ingot was used for the HWE source. The CGS layers were grown on semi-insulating GaAs (1 0 0) by the HWE method using the grown CGS ingot as a source material. Fig. 1 presents a diagram of the HWE apparatus used for the CGS growth and solar-cell sample. In this experiment, the CGS layer was grown by using furnace (a) as shown in Fig. 1. Prior to growing the CGS layers, the GaAs substrate was cleaned ultrasonically for 1 min in successive baths of TCE, acetone, methanol and 2-propanol, and then etched for 1 min in a solution of H2SO4:H2O2:H2O (5:1:1). The substrate was degreased in organic solvents and rinsed with deionized water (18.2 MO). After the substrate was dried off, it was immediately loaded onto the substrate holder as shown in
Substrate Substrate
GaAs & ITO Glass Heater
Source 2 CdS Powder Source 1 Poly CuGaSe2
Reservoir
Furnace (b)
Furnace (a)
Fig. 1. Diagram of the HWE apparatus used for the CGS growth and solar-cell sample.
Fig. 1 and was annealed at 580 1C for 20 min to remove the residual oxide on the surface of the substrate. To obtain optimum growth conditions, the grown CGS layers were estimated through PL and XRD measurements. The PL measurement of the sample was performed at 10 K in a low-temperature cryostat during the excitement of the He–Ne laser (632.8 nm, 50 mW). The thickness, growth orientation, and stoichiometric composition of the CGS layers were measured by using an a-step profilometer, XRD, and an energy-dispersive X-ray spectrometer (EDS), respectively. Thus, the Hall effect experiment was carried out using the van der Pauw method. Optical absorption was used to measure the band-gap energy using an UV–VIS–NIR spectrophotometer for a range of 670– 750 nm with the temperature varying from 10 to 293 K. Also, the CdS/CGS/ITO layer for the solar-cell specimen was grown using HWE. The substrate of the pyrex glass, which was sprayed with indium tin oxide (ITO), was moved successively to the top from furnace (a) to (b), as shown in Fig. 1. The source materials 1 and 2, as shown in Fig. 1, were polycrystalline CGS ingot and CdS powder (rare metallic, 6 N), respectively. After the CGS layer on the ITO glass was grown in furnace, (a), the CGS/ITO layer was immediately moved to furnace (b). Then, the CdS/CGS/ ITO layer was grown by using heating source 2. Also, the solar-cell measurement was conducted by illuminating the tungsten halogen lamp at 80 mW/cm2. 3. Results and discussion 3.1. Growth and structural properties After the heat treatment of the substrate surface, the CGS layers were grown by varying the substrate temperature from 410 to 470 1C, while the source temperature was fixed at 610 1C. At this time, the crystal quality of the grown CGS layers was evaluated by measuring the PL spectra at 10 K. Thus, in order to evaluate the crystal quality, the FWHM value of XRD assisted. Exciton emissions were used to predict the crystal-quality criterion of the grown layer because the exciton could only be observed in a less defective crystal at a low temperature. Here, the emissions of free excitons (EFX) and bound excitons (EBX) used the optimum growth conditions. The grown layers are considered to have good crystal quality when the measured PL spectra show that the intensity of EFX and EBX tend to increase. As shown in Table 1, the EFX and the EBX peaks in the layer, which were grown while the substrate temperature was kept at 450 1C, had the highest intensity of all other samples. On the contrary, the narrowest FWHM, which indicated high-crystal quality, obtained from the XRD experiment was observed in the CGS layer grown at 450 1C as well. This observation indicates that the CGS layer grown at 450 1C has a very high quality because the exciton emission can be observed only under the condition of a long-range Coulomb coupling between the electron and the hole. Therefore,
ARTICLE IN PRESS S.H. You et al. / Journal of Crystal Growth 310 (2008) 2717–2723
2719
Table 1 Comparison of the PL intensities and the XRD as a function of substrate temperature (Arbitrary units)
Table 2 Composition ratios of each element on the synthesized polycrystalline and the CGS layer analyzed by the EDS measurement (Units: %)
Substrate temperature (1C)
Elements
EFX
EBX
FWHM (arcsec)
410 430 450 470
5.75 10.75 26.19 21.35
7.16 16.84 37.16 30.90
1827 1163 457 651
Intensities of PL at 10 K
XRD (1y)
Grown layer
Before synthesis
After synthesis
Before growth
After growth
25 25 50
24.507 25.618 49.875
24.507 25.618 49.875
24.237 26.281 49.482
beneficial for a CGS layer to be fabricated for solar energy conversion. Also, the CGS (1 1 0) peak is located at 2y of 40.4891 and its FWHM value is 0.2051. This FWHM value is similar to that of the CGS grown from MOCVD by Orsal et al. [16]. Therefore, it should be noted that the layers grown by the HWE method are high-quality crystalline. Thus, the mean crystallite size of the layer can be extracted from
GaAs
(400)
3x104
Intensity (arb.units)
Cu Ga Se
Synthesized polycrystalline
2x104
D ¼ 0:94l=ðB cos yÞ; CuGaSe2
(110)
1x104
0 30
40
50
60
70
Diffraction angle (2θ) Fig. 2. XRD o2y scans of the CGS layer grown under optimized conditions.
the optimum temperature of the substrate was found to be 450 1C, while the source temperature was 610 1C. At this time, the thickness and the growth rate of the layer were 2.5 mm and 0.83 mm/h, respectively. Fig. 2 presents the XRD o2y scans of the CGS layers grown at the source and substrate temperatures of 610 and 450 1C, respectively. Thus, the obtained curve corresponds to the diffraction peaks of the CGS (1 1 0) and GaAs (0 0 4). This means that the orientation of the layer grown on the GaAs (1 0 0) substrate was converted to the (1 1 0) plane. This phenomenon was also observed in the CdTe epilayer grown on the GaAs (1 0 0). Faurie et al. [15] reported that the orientation of the CdTe epilayer was related to the pre-annealing process used to remove the residual oxide on the surface of the substrate. They concluded that the growth of the CdTe (1 0 0) or (1 1 1) planes on the GaAs (1 0 0) was possible by controlling the different annealing temperatures and times of the substrate. On the other hand, the observation of only one peak of the CGS (1 1 0) indicates that the CGS layer was grown epitaxially along the /1 1 0S direction onto the GaAs (1 0 0) substrate. This orientation is suggested to be
(1)
where l, y, and B are the X-ray wavelength (0.15405 nm), the Bragg diffraction angle, and the FWHM value on (1 1 0) peak in radians, respectively [17]. From Eq. (1), the mean crystallite size was estimated to be about 81 nm. Table 2 presents the composition ratios of each element during the growth process of the CGS layer, analyzed by the EDS measurement. As shown in Table 2, the grown layer consisted of Cu-poor, Ga-rich, and slightly Se-poor components. From the Hall effect measurement, the grown CGS layer having Ga-rich component was p-type presumably due to slight stoichiometric deviations originating from an excess of copper vacancies (VCu). From the Hall effect measurement, the carrier density and Hall mobility of the layer at 293 K were determined to be 4.87 1017 cm3 and 129 cm2/V s, respectively. 3.2. Optical properties Fig. 3 shows the optical absorption spectra obtained in the temperature range of 10–293 K. To identify the bandgap energy for the CGS layer, we carefully examined the relationship between the optical absorption coefficient (a) and the incident photon energy (hn) from the optical absorption measurements in Fig. 3. The relationship for a direct band-gap between hn and a is given by ðahvÞ2 ðhv E g Þ:
(2) 2
According to Eq. (2), (ahn) linearly depends upon the photon energy. From the plots of (ahn)2 versus photon energy for different temperatures, the band gaps were determined by extrapolating the linear portions of the respective curves to (ahn)2 ¼ 0. Fig. 4 displays the bandgap variation of the CGS layer as a function of temperature. The temperature dependence of the band-gap energy
ARTICLE IN PRESS S.H. You et al. / Journal of Crystal Growth 310 (2008) 2717–2723
2720
1.8
Table 3 Comparison of band gap between our data and published values
76
Optical density (arb.units)
8
5
4
3
2
Sample
1
Band gap
Crystal growth
Temperature (K)
Energy (eV)
Type
Method
1 2 3 4 5 6 7
77 77 10 4.2 4.2 0 77
1.746 1.745 1.73 1.7347 1.75 1.7310 1.7872
Bulk Bulk Layer Bulk Layer Layer Layer
CVT CVT MOVPE VGF MOCVD MBE HWE
7 7 7
10 4.2 0
1.7995 1.7997 1.7998
Reference
1.2 1. 293 K 2. 250 K 3. 200 K 4. 150 K 5. 100 K 6. 77 K 7. 50 K 8. 10 K
0.6
0.0 660
690
720 Wavelength (nm)
750
Fig. 3. Optical absorption spectra according to the variation of the measure temperature.
1.9 Experimental value Fitted energy
Energy (eV)
1.8
1.7 Eg(0) : 1.7998 eV a : 8.7489x10-4 eV/K b : 335 K
1.6 0
100 200 Temperature (K)
300
in our experiment is well fitted numerically by the following formula [18,19]: E g ðTÞ ¼ E g ð0Þ aT 2 =ðT þ bÞ,
with the reported value of 394 K [20]. Thus, the band-gap energy of the CGS obtained at 293 K is estimated to be 1.6802 eV. Even though the band-gap variation of CGS is well satisfied by Eq. (3), our values are slightly larger than that of reported values as shown in Table 3. These differences in the low-temperature region are about 40–60 meV. We suggest that this cause is responsible for a composition ratio of Ga. As the Ga concentration in CGS increases, it is known that the electronic transitions shift towards higher energies [27]. As shown in Table 2, the CuXGa ratio in CGS is about 0.922. Therefore, it suggests that the band-gap increase is influenced by the slightly Garich composition. The absorption-experiment itself can be a problem, however. This method is not accurate due to the difficulty in defining the position of the absorption edge. Fig. 5 shows a typical PL spectrum of the CGS layer measured at 10 K. As shown in Fig. 5, an intense and sharp peak was observed at 747.2 nm (1.6593 eV). This peak is associated with the neutral donor-bound exciton (D1, X), caused by the recombination from free exciton to neutral donor. Also, the EFX at 741.6 nm (1.6718 eV) appears on the shoulder toward the short-wavelength region of the (D1, X) emission. This peak shows a low excitonic emission indicating a redshift due to biaxial tensile stress in the epitaxial film. From the EFX emission, the binding energy, E FXb , is obtained by E FXb ¼ E g ð10Þ EðE FX Þ;
Fig. 4. Band-gap variation of the CGS layer as a function of temperature.
(3)
where a is a constant and b is approximately the Debye temperature. Also, Eg(0) is the band-gap energy at 0 K, which is estimated to be 1.7998 eV. When a and b are taken to be 8.7489 104 eV/K and 335 K, respectively, the curve plotted by Eq. (3) closely fits the experimental values. Our result on the Debye temperature is also in close agreement
[21] [22] [23] [24] [25] [26] This work
(4)
where Eg(10) is the band-gap energy at 10 K. The value of Eg(10) was calculated to be 1.7995 eV using Eq. (3) and the E FXb is obtained to be 0.1277 eV. The E FXb value is about 10 times larger than the reported one [23]. As stated above, it suggests that the large value of E FXb is caused by the increase of Eg due to the Ga-rich component and the inaccuracy of the absorption measurement. Also, the shoulder peak toward the long-wavelength region of the (D1, X) emission appears at 755.5 nm (1.6411 eV). This peak is suggested to be due to the neutral acceptor-bound exciton, (A1, X). This bound exciton has been known to be related to the recombination of the free exciton to the
ARTICLE IN PRESS S.H. You et al. / Journal of Crystal Growth 310 (2008) 2717–2723
2721
tion process on the CGS layer. Fig. 6 plots the band diagram of the observed optical transitions including the band-gap energy of 10 K. In this band diagram, the Eg(10)
100 (Do, X)
EFX FX
PL intensity (arb.units)
Conduction band (Ao, X)
6
Do
7
(Do, Ao)
50
1 (Do, Ao) replicas
2
3
5
4
SA SA Eg (10)
EFX
(Do, X)
(Ao, X)
(Do, Ao) 0 700
800
900 Wavelength (nm)
1000
Fig. 5. PL spectrum of the CGS layer measured at 10 K.
Valence band
(5)
Therefore, the E FXb values of the (D1, X) and the (A1, X) emissions are calculated to be 12.5 and 30.7 meV, respectively. Thus, the neutral donor level is estimated to be 0.1402 eV ( ¼ 0.1277+0.0125). This result is close to the donor level for Ga-annealed CGS [28]. Therefore, the neutral donor level is ascribed to the Se vacancies (VSe) and/or Ga in antisite (GaCu) which is the most probable defect in CGS due to the Ga-rich component [28]. It also suggests that the neutral acceptor level acting as an acceptor is caused by the VCu due to a slight stoichiometric deviation [25]. In addition, the neutral donor–acceptor pair, (D1, A1), emission was observed at 763.9 nm (1.6231 eV). The (D1, A1) emission occurred due to the interaction between the neutral donor and the neutral acceptor states in the energy band gap. Thus, replica peaks in the (D1, A1) emission are observed at 770.2 (1.6098 eV), 776.6 (1.5965 eV), 783.1 (1.5833 eV), and 789.5 nm (1.5704 eV). These peaks are considered to be associated with the longitudinal optical (LO) phonon replicas of the (D1, A1) emission, because the energy difference between these peaks is equal to a multiple of the LO phonon energy of B2-modes such as 13.3 (1LO), 26.6 (2LO), 39.8 (3LO), and 52.7 meV (4LO) [29]. The relatively low and broad peak caused by the self-activated (SA) emission was observed at 919.6 nm (1.3483 eV) in the longer-wavelength region. This emission strongly relates to the native defects formed in the CGS layer. Therefore, based on these PL results, we schemed out a band diagram of the recombina-
Fig. 6. Band diagram of the optical transitions on the CGS layer measured at 10 K. 1. Band-gap energy of CGS at 10 K: 1.7995 eV. 2. Free exciton energy: 1.6718 eV. 3. Neutral donor-bound exciton energy: 1.6593 eV. 4. Neutral acceptor-bound exciton energy: 1.6411 eV. 5. Neutral donor–acceptor pair energy: 1.6231 eV. 6. Binding energy of free exciton: 0.1277 eV. 7. Neutral donor levels: 0.1402 eV. 8. Neutral acceptor levels: 0.0307 eV.
30 Light 20
Current density (mA/cm2)
neutral acceptor. The binding energy (E FXb ) of the EBX emission is obtained by E BXb ¼ EðE FX Þ EðD ; XÞ or EðA ; XÞ:
Ao
8
In n-CdS p-CuGaSe22 ITO Glass ITO
In
10 Dark Illuminated
Vmp 0 0.0
0.4
0.2
0.6
0.8
Voltage (V) -10
Vmp= 0.41 V Jmp= 21.8 mA/cm2
-20
Jmp
Pmax
-30 Fig. 7. Dark and illuminated current density–voltage characteristics of solar cells. (Here, the subfigure shows a diagram of the n-CdS/p-CGS junction. Also, Jmp and Vmp are the current density and voltage values at the maximum power point Pmax, respectively.)
ARTICLE IN PRESS 2722
S.H. You et al. / Journal of Crystal Growth 310 (2008) 2717–2723
Table 4 J–V parameters of the n-CdS/p-CGS solar-cells Cell
Voc (V)
Jsc (mA/cm2)
FF (%)
Efficiency (%)
Reference
CdS/CGS layer CdS/CGS polycrystalline CdS/CGS layer
0.95 0.59 0.51
14.88 8.14 22.90
70.79 44.0 76.53
9.53 2.0 11.17
[6] [31] This work
is the band-to-band transition corresponding to the bandgap energy. Also, the (D1, A1) emission is due to the recombination transition between the neutral donor and the neutral acceptor states in the band-gap energy. 3.3. Solar cells CGS-based solar cells were investigated. Fig. 7 shows the dark and illuminated current density–voltage (J–V) characteristics of solar cells. Also, the sub figure in Fig. 7 shows a diagram of the n-CdS/p-CGS junction. As shown in the subfigure, it is fabricated to a p–n.junction between the p-CGS layer as the absorber material and the n-CdS layer [30]. The solar-cell parameters obtained from the J–V characteristic are given in Table 4. From these results, the open-circuit voltage (Voc) and short-circuit current–density (Jsc) were observed to be 0.51 V and 22.9 mA/cm2, respectively. As Table 4 shows, this means that the achievement of 11.17% efficiency is an improvement in comparison to the previous devices [6–31]. Also, it is superior to the best single-crystal CGS device to date (9.7% active area efficiency) [7]. 4. Conclusions The CGS layers, which are now charming as solar-cell materials, were grown by the HWE method. From the results of the PL and XRD measurements, the optimum growth temperatures of the substrate and the source turned out to be 450 and 610 1C, respectively. Also, the CGS layers were confirmed to be the epitaxially grown layer along the /1 1 0S direction onto the GaAs (1 0 0) substrate. The carrier density and Hall mobility of the layer at 293 K were determined to be 4.87 1017 cm3 and 129 cm2/V s, respectively. Thus, as is evident in the EDS measurement, the grown CGS layers consisted of Ga-rich components indicating slight stoichiometric deviations. The band-gap variation of CGS extracted from the absorption measurement was well fitted by Eg(T) ¼ 1.79988.7489 104T2/ (T+335). The band-gap energy at a low temperature, however, was slightly larger than that of other CGS. It suggests that the band-gap increase is influenced by slightly Ga-rich compositions. From the low-temperature PL measurement, the EFX, the (D1, X), and the (A1, X) emissions, which can be observed at the high-quality crystalline, appeared at 1.6718, 1.6593, and 1.6411 eV, respectively. Also, the (D1, A1) and its LO phonon replicas emissions, which are due to the interaction between the
neutral donor and the neutral acceptor states, were observed at the wavelength region between 770.2 and 789.5 nm. Based on these PL results, a band diagram of the observed optical transitions was obtained. From the solar cell of the n-CdS/p-CGS junction, an efficiency of 11.17% has been reached. Acknowledgment This study was supported by research funds from Chosun University, 2007. References [1] J.L. Shay, J.H. Wernick, Ternary Chalcopyrite Semiconductors: Growth, Electronic Properties, and Applications, Pergamon, Oxford, 1975 (Chapter 4). [2] W.N. Shafarman, J. Zhu, Thin Solid Films 473 (2000) 361. [3] M.A. Contreras, B. Egaas, K. Ramanatan, J. Hiltner, A. Swartzlander, F. Hasoon, R. Nuofi, Prog. Photovoltaics 7 (1999) 311. [4] D. Hariskos, R. Herbeholz, M. Ruckh, U. Ruhle, R. Schaffler, H.W. Schock, in: W. Freielsleben, W. Palz, H.A. Ossenbrink, P. Helm (Eds.), Proceedings of the 13th European Photovoltaic Solar Energy Conference and Exhibition Acropolis, Convention Center, Nice, France, October 23–27, 1995, p. 1995. [5] K. Kushiya, S. Kuriyagawa, T. Nii, I. Sugiyama, T. Kase, M. Sato, H. Takeshita, in: W. Freielsleben, W. Palz, H.A. Ossenbrink, P. Helm (Eds.), Proceedings of the 13th European Photovoltaic Solar Energy Conference and Exhibition Acropolis, Convention Center, Nice, France, October 23–27, 1995, p. 2016. [6] D.L. Young, J. Keane, A. Duda, J.A.M. AbuShama, C.L. Perkins, M. Romero, R. Noufi, Prog. Photovoltaics 11 (2003) 535. [7] M. Saad, H. Riazi, E. Bucher, M. Ch. Lux-Steiner, Appl. Phys. A: Mater. Sci. Process 62 (1996) 181. [8] I.E. Beckers, U. Fiedeler, S. Siebentritt, M. Ch. Lux-Steiner, Thin Solid Films 431 (2003) 205. [9] S. Siebentritt, I. Beckers, T. Riemann, J. Christen, A. Hoffimann, M. Dworzak, Appl. Phys. Lett. 86 (2005) 091909. [10] S. Schuler, S. Siebentritt, S. Nishiwaki, N. Rega, J. Beckmann, S. Brrehme, M. Ch. Lux-Steiner, Phys. Rev. B 69 (2004) 045210. [11] J.H. Yun, R.B.V. Chalapathy, S.J. Ahn, J.C. Lee, K.H. Yoon, J. Korean Phys. Soc. 50 (2007) 1794. [12] M. Rusu, P. Gashin, A. Simashkevich, Sol. Energy Mater. Sol. Cells. 70 (2001) 175. [13] T.S. Jeong, P.Y. Yu, K.J. Hong, T.S. Kim, C.J. Youn, Y.D. Choi, K.S. Lee, B. O, M.Y. Yoon, J. Crystal Growth 249 (2003) 9. [14] A. Lopez-Otero, Thin Solid Films 49 (1987) 3. [15] J.P. Faurie, C. Hsu, S. Sivananthan, X. Chu, Surf. Sci. 168 (1986) 473. [16] G. Orsal, F. Mailly, N. Romain, M.C. Artaud, S. Rushworth, S. Duchemin, Thin Solid Films 361 (2000) 135. [17] B.D. Cullity, Elements of X-ray Diffractions, Addition-Wesley, Reading, MA, 1978, p. 102. [18] Y.P. Varshni, Physica 34 (1967) 149.
ARTICLE IN PRESS S.H. You et al. / Journal of Crystal Growth 310 (2008) 2717–2723 [19] K.J. Hong, T.S. Jeong, C.J. Yoon, Y.J. Shin, J. Crystal Growth 218 (2000) 19. [20] S. Siebentritt, Thin Solid Films 480–481 (2005) 312. [21] I.A. Ponomareva, A. Yu, I.V. Bodnar, Phys. Solid State 49 (2007) 23. [22] B. Tell, P.M. Bridenbaugh, Phys. Rev. B 12 (1978) 3330. [23] A. Bauknecht, S. Siebentritt, J. Albert, M. Ch. Lux-Steiner, J. Appl. Phys. 89 (2001) 4391. [24] K. Yoshino, M. Sugiyama, D. Maruoka, S.F. Chichibu, H. Komaki, K. Umeda, T. Ikari, Physica B 302–303 (2001) 357. [25] G. Orsal, F. Mailly, M.C. Artaud-Gillet, S. Duchemin, Thin Solid Films 387 (2001) 198.
2723
[26] A. Yamada, Y. Makita, S. Niki, A. Obara, P. Fons, H. Shibata, M. Kawai, S. Chichibu, H. Nakanishi, J. Appl. Phys. 79 (1996) 4318. [27] P.D. Paulson, R.W. Birkmire, W.N. Shafarman, J. Appl. Phys. 94 (2003) 879. [28] J.H. Scho¨n, O. Schenker, H. Riazi-Nejad, K. Friemelt, Ch. Kloc, E. Bucher, Phys. Stat. Sol. (a) 161 (1997) 301. [29] D. Papadimitriou, N. Esser, C. Xue, Phys. Stat. Sol. (b) 242 (2005) 2633. [30] N. Romeo, G. Sberveglieri, L. Tarricone, Appl. Phys. Lett. 30 (1977) 108. [31] S. Siebentritt, A. Bauknecht, A. Gerghard, U. Fiedeler, T. Kampschulte, S. Schuler, W. Harneit, S. Brehme, J. Albert, M. Ch. Lux-Steiner, Sol. Energy Mater. Sol. Cells 67 (2001) 129.