Contents lists available atScienceDirect
Journal of Alloys and Compounds
j o u r n a l h o m e p a g e :w w w . e l s e v i e r . c o m / l o c a t e / j a l l c o m
Polarization-dependent electrolyte electroreflectance study of Cu 2 ZnSiS 4 and Cu 2 ZnSiSe 4 single crystals
S. Levcenco
a,1, D. Dumcenco
a,1, Y.S. Huang
a,∗, E. Arushanov
b, V. Tezlevan
b, K.K. Tiong
c, C.H. Du
daDepartment of Electronic Engineering, National Taiwan University of Science and Technology, Taipei 106, Taiwan
bInstitute of Applied Physics, Academy of Sciences of Moldova, Chisinau, MD 2028, Republic of Moldova
cDepartment of Electrical Engineering, National Taiwan Ocean University, Keelung 202, Taiwan
dDepartment of Physics, Tamkang University, Tamsui 251, Taiwan
a r t i c l e i n f o
Polarization-dependent electrolyte electroreflectance (EER) measurements were carried out on the ori-ented Cu2ZnSiS4 and Cu2ZnSiSe4 single crystals at room temperature. Thin blade single crystals of Cu2ZnSiS4and Cu2ZnSiSe4were grown by chemical vapor transport technique using iodine as a trans-port agent. Laue pattern normal to the basal plane of the as-grown crystal revealed the formation of orthorhombic structure with the normal along [2 1 0] and the c axis parallel to the long edge of the crystal platelet. The polarized EER spectra in the vicinity of the direct band edge of Cu2ZnSiS4displayed distinct structures associated with transitions from two upper-most valence bands to the conduction band min-imum at point. In the E⊥c configuration, the feature designated as EA∼ 3.345 eV was detected and for Ec, only EB∼ 3.432 eV appeared. For Cu2ZnSiSe4, three features denoted as EA, EB, and ECat around 2.348, 2.406 and 2.605 eV, respectively, were recorded for E⊥c polarization, whereas in the Ec, only EB
and ECwere the allowed transitions. Based on the experimental observations and a recent band-structure calculation by Chen et al. [Phys. Rev. B 82 (2010) 195203], plausible band structures near the direct band edge of Cu2ZnSiS4and Cu2ZnSiSe4have been proposed.
© 2011 Elsevier B.V. All rights reserved.
1. Introduction
Cu2ZnSiS4 and Cu2ZnSiSe4 belong to the family of quaternary chalcogenide crystallizing in the wurtzite–stannite structure. In these compounds each sulfur/selenium atom has four nearest neighbor cation atoms (two copper atoms, a zinc, and a silicon atom) at the corners of the surrounding tetrahedron[1–4]. The qua-ternary chalcogenide semiconductors have drawn wide interest for their nonlinear optical properties[5,6]and potential application as solar-cell absorbers[7–9], photocatalysts for solar water splitting [10,11], and high-temperature thermoelectric materials [12,13].
Despite their interesting optical and thermoelectric properties, and possible applications, up-to-date, only a few studies have been reported concerning the basic properties of Cu2ZnSiS4(Se4), due to the difficulty of preparing suitable size, compositionally homo-geneous and high purity single crystals. Furthermore, the reported results of these studies[14,15]contain some discrepancies.
∗ Corresponding author. Tel.: +886 2 27376385; fax: +886 2 27376424.
E-mail address:[email protected](Y.S. Huang).
1 Permanent address: Institute of Applied Physics, Academy of Sciences of Moldova, Chisinau, MD 2028, Moldova.
In this article, we report a detailed study of the near direct band edge anisotropic optical transition properties of Cu2ZnSiS4 and Cu2ZnSiSe4 single crystals by polarization-dependent electrolyte electroreflectance (EER) at room temperature. High quality sin-gle crystals of Cu2ZnSiS4and Cu2ZnSiSe4were grown by chemical vapor transport using iodine as the transport agent. Hall measure-ments indicated p-type semiconducting behavior for the crystals.
The EER measurements were carried out on the as-grown basal plane with the normal along [2 1 0] and the axis c parallel to the long edge of the crystal platelets. The polarization-dependent near band-edge excitonic transition energies were determined. Base on a recent band-structure calculation by Chen et al.[16,17], a schematic representation of the plausible assignments for the observed near direct band edge optical transitions for Cu2ZnSiS4and Cu2ZnSiSe4
is presented and discussed.
2. Experimental
Single crystals of Cu2ZnSiS4(Cu2ZnSiSe4) were grown by vapor transport of sto-ichiometric amounts of the elements with 5 mg iodine/cm3as the transport agent.
Optimum crystal growth was achieved with the charge zone maintained at 950◦C (850◦C) and the growth zone at 900◦C (800◦C). The transport process was carried out for a period of 14 days. Single crystals Cu2ZnSiS4(Cu2ZnSiSe4) formed thin, greenish (reddish), blade shape up to 10 mm× 1.5 mm (20 mm × 1.0 mm) in area 0925-8388/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.jallcom.2011.04.013
7106 S. Levcenco et al. / Journal of Alloys and Compounds 509 (2011) 7105–7108
Fig. 1. A schematic arrangement of the polarization-dependent EER measurements.
and 300 (100)m in thickness were synthesized. Hall measurements indicated p-type semiconducting behavior for the as-grown crystals. The orientation of the basal plane was determined by comparing back-reflection Laue pattern with computer generated Laue plots. Analyzing the symmetry of Laue pattern confirms the forma-tion of orthorhombic structure and reveal that the normal of the basal plane is [2 1 0]
and the long-edge of the crystal platelet is parallel to c axis[18].
Fig. 1depicts the schematic arrangement of the polarization-dependent EER measurements with polarization configurations of E⊥c and Ec performed on the as-grown basal plane with the normal along [2 1 0] and c parallel to the long edge of the crystal platelet. A 150W xenon arc lamp filtered by a 0.25 m grating monochro-mator provided the source for EER measurements. Model PRH 8020 CASIX Rochon prisms were employed for polarization dependent measurements. A model 3378 Hamamatsu photomultiplier tube was used to detect the reflected signal. For EER measurements an electrolyte of the tartaric acid (3 wt.%) in ethylene alcohol was used. A 200 Hz, 4 V peak-to-peak, square wave with a−0.5 V (vs. Pt electrode) DC bias was used to modulate the electric field. A dual-phase lock-in amplifier was used to measure the detected signals. The entire data acquisition procedure was per-formed under computer control. Multiple scans over a given photon energy range was programmed until a desired signal-to-noise level has been obtained.
3. Results and discussion
Fig. 2(a)–(c) shows the unpolarized, E⊥c and Ec polarization EER spectra of Cu2ZnSiS4in the energy range 3.2–3.6 eV, respec-tively. As shown inFig. 2(a), the unpolarized EER spectrum has two prominent features labeled as EAand EBand located between 3.20 and 3.55 eV. The polarized spectra showed evidence of the exis-tence of a strong polarization effect on the EER spectra related to the optical anisotropy of Cu2ZnSiS4orthorhombic crystal structure.
For E⊥c polarization (Fig. 2(b)), only the feature EAis seen, while the EBfeature is observed for Ec polarization (seeFig. 2(c)). These fea-tures may be related to the interband excitonic transitions at the point of the Brillouin zone[18,19]. The EER spectra were analyzed using the first derivative Lorentzian lineshape function model for excitonic transitions[20–22]. This model is given by
R R = Re
n j=1Ajeij(E − Ej+ ij)−2 (1)
where n is the number of spectral features to be fitted, Aj and
j are the amplitude and phase of the lineshape, Ejandjare, respectively, the energy and broadening parameter of the interband transitions. InFig. 2the best least-squares fits to experimental data are shown by the solid curves. Arrows above the curves inFig. 2 show the positions of the EAand EBinterband excitonic features.
The obtained values of EAand EBare found to be 3.345± 0.005 eV and 3.432± 0.005 eV, respectively, and are listed inTable 1. For comparison purpose, the energy positions determined by PzR[19]
are also listed in Table 1. The energy difference between the
Fig. 2. The (a) unpolarized, (b) E⊥c polarization and (c) Ec polarization EER spectra of Cu2ZnSiS4at 300 K. The solid lines are the least-squares fits of experimental data to Eq.(1). The obtained values of the transition energies denoted as EAand EBare indicated by the arrows.
low- and high-energy transitions represents the crystal-field split-ting between the two upper-most valence bands. Shay et al.[23]
reported the symmetries and splitting of the uppermost valence bands in orthorhombic AgInS2 by using polarized EER measure-ments on oriented crystals. The observed valence band splitting was explained by crystal field splitting alone, neglecting any spin–orbit interaction. The results showed that the direct band edge transition for Ec polarization is higher than that of E⊥c polarization. Our EER results showed that the excitonic transition energy of EBfeature observed at Ec polarization is 87 meV larger than that of EA at E⊥c polarization, similar to the deduced value from PzR measure-ments[19], concurred well qualitatively with the previous report on orthorhombic AgInS2[23].
Fig. 3(a)–(c) shows the unpolarized, E⊥c and Ec polarization EER spectra of Cu2ZnSiSe4in the energy range 2.0–3.0 eV, respec-tively. As shown inFig. 3(a), the three dominant structures located between 2.20 and 2.80 eV are associated with band-edge excitonic transitions from different origins and are assigned as EA, EB, and EC. As can be seen inFig. 3(b), three features are recorded for E⊥c polarization, whereas in the Ec configuration (Fig. 3(c)), only EB
and ECare the allowed transitions. Shown by the solid curves in Fig. 3(a)–(c) are the least-squares fits to Eq.(1). The dashed curves show the theoretical fit of each transition. Arrows above the curves inFig. 3show the positions of the EA, EBand ECinterband exci-tonic features. The obtained values of EA, EBand ECare found to be 2.348± 0.005, 2.406 ± 0.005 and 2.605 ± 0.005 eV, respectively, and are listed inTable 1. The observed polarization dependent EER
Table 1
Values of the direct band-edge excitonic transitions EA, EBand ECobtained by fitting EER data to Eq.(1). The corresponding values for the direct excitonic transition energies of Cu2ZnSiS4obtained by piezoreflectance are also listed for comparison.
Materials Method EA(eV) EB(eV) EC(eV)
Cu2ZnSiS4 EER 3.345± 0.005 3.432± 0.005 – PzRa 3.323± 0.005 3.413± 0.005 –
Cu2ZnSiSe4 EER 2.348± 0.005 2.406± 0.005 2.605± 0.005
aRef.[16].
S. Levcenco et al. / Journal of Alloys and Compounds 509 (2011) 7105–7108 7107
Fig. 3. The (a) unpolarized, (b) E⊥c polarization and (c) Ec polarization EER spectra of Cu2ZnSiSe4at 300 K. The solid lines are the least-squares fits of experimental data to Eq.(1). The dashed curves show the theoretical fits to each transition. The obtained values of the transition energies denoted as EA, EBand ECare indicated by the arrows.
spectra for Cu2ZnSiSe4are found to be similar to that reported on photoreflectance (PR) study of wurtzite-CdS by Imada et al.[24].
The results of their polarization dependent PR spectra indicated that as expected from the optical-transition selection rule, the fea-ture EAcan be recognized only for E⊥c, but not for Ec.
Our experimental findings of the near direct band edge transi-tions for Cu2ZnSiS4and Cu2ZnSiSe4are similar to that for I–III–VI2 semiconductors reported by Shay et al.[25]. They had observed three valence bands in every selenide-containing compound inves-tigated, but never observed three valence bands in a sulphide-based compound. For sulphide contained compounds, only two valence bands were observed, one of which was observed only for E⊥c and the other for Ec. In order to understand the observed interband transitions, a band diagram near the direct band edge is needed.
Recently Chen et al. reported a band-structure calculation using first-principles total energy calculations on a family of I2–II–IV–VI4 wurtzite-derived polytypes of kesterit and stannite quaternary chalcogenide semiconductors[16,17]. From the calculation, the fol-lowing results can be found: (i) I2–II–IV–VI4semiconductors have usually direct band gaps at the point, (ii) the top of the valence band is mainly the antibonding component of the p–d hybridiza-tion between the group-VI anion and group-I cahybridiza-tion, (iii) the bottom conduction band is mainly the antibonding component of the s–s and s–p hybridization between group-IV cation and group-VI anion [17], except for those containing Si, the group-I and group-II cations also have significant contribution to the bottom of the conduction band as well as Si and group-VI anion. Adopting the band-structure calculation by Chen et al.[16]the schematic representation of the plausible assignments for the observed optical transitions near the direct band edge for Cu2ZnSiS4and Cu2ZnSiSe4 are presented in Fig. 4(a) and (b). As shown inFig. 4(b), for Cu2ZnSiSe4three closely spaced valence bands (v1,v2andv3) caused by the p–d hybridiza-tion of Se p-like and Cu d-like valence band states form the top-most valence bands. We attribute the EA, EB, and ECstructures inFig. 3 to transitions from three split valence bands to a single conduc-tion band minimum at the point, where the differences between v1andv2, andv2andv3 are the crystal-field (cf) and spin-orbit
Fig. 4. The schematic representations of the plausible assignments for observed optical transitions near the direct band edge at point for (a) Cu2ZnSiS4and (b) Cu2ZnSiSe4.
(so) splitting parameters. For the case of Cu2ZnSiS4cf so, i.e.
spin–orbit splitting can be neglected. Therefore onlyv1andv2bands form the top-most valence bands (seeFig. 4(a)), and EAand EB fea-tures inFig. 2can be attributed to the transitions fromv1andv2to conduction band minimum at the point. Our results also show that the spin split-off energysoin Cu2ZnSiSe4is much stronger than that in Cu2ZnSiS4and are attributed to a direct consequence of the heavier Se element. The experimental results agreed well with the recent report by Person[26]on the electronic structure study of Cu2ZnSnS4and Cu2ZnSnSe4by a relativistic full-potential linearized augmented plane wave method. The study revealed that the spin split-off energysois stronger in the Se-based compounds compared to the S-containing compounds.
As shown inTable 1the crystal field splitting parameter for Cu2ZnSiS4is larger than that of Cu2ZnSiSe4. The possible reason can be understood as follows. In the simplest approach, the perfect unit cell of orthorhombic Cu2ZnSiS4(Se4) quaternary compounds can be derived by doubling an orthohexagonal wurtzite cell in the a-direction so that the relationship between cells dimensions are aor= 2aw,bor=√
3aw, and cor= cw[2]. In addition, for ideal wurtzite structure,cw= 2
2/3aw. From the polarized magnetoreflectance measurements on oriented crystals Shih et al.[27]reported that the increase of the crystal-field splitting in Cu2Zn1−xMnxGeS4is due to the increasing of the fractional distortion from perfect orthorhom-bic geometry. Using the actual lattice constants of Cu2ZnSiS4and Cu2ZnSiSe4[15], the values of the fractional distortion from per-fect orthorhombic geometry for a and b axis are −0.0106 and
−0.0171 for Cu2ZnSiS4and−0.0085 and −0.0159 for Cu2ZnSiSe4, respectively. Therefore our experimental finding that the energy difference between the EBand EAin Cu2ZnSiS4is higher than that in Cu2ZnSiSe4concurred well qualitatively with relative degree of distortion of the respective crystal structures.
4. Summary
Polarization-dependent EER measurements were carried out on the oriented Cu2ZnSiS4 and Cu2ZnSiSe4 single crystals at room temperature. The near direct band edge anisotropic EAand EB exci-tonic transitions of Cu2ZnSiS4are found to be 3.345 eV for E⊥c and 3.432 eV for Ec configurations. For Cu2ZnSiSe4, three features EA, EB, and ECat around 2.348, 2.406 and 2.605 eV, respectively, were observed for E⊥c polarization, whereas in the Ec, only EB and ECwere recorded. Based on the experimental results and a recent band-structure calculation by Chen et al., plausible band structures near direct band edge of Cu2ZnSiS4and Cu2ZnSiSe4have been pre-sented. The optical transitions are attributed to the transitions from
7108 S. Levcenco et al. / Journal of Alloys and Compounds 509 (2011) 7105–7108
split valence bands, caused by crystal field and spin orbit interac-tions, to a single conduction band minimum at the point. Our results reveal that the spin split-off energyso in Cu2ZnSiSe4 is much stronger than that in Cu2ZnSiS4and are attributed to a direct consequence of the heavier Se element.
Acknowledgments
The authors acknowledge the supports of National Science Council of Taiwan under Project nos. NSC 099-2811-E-011-016, NSC 97-2112-M-011-001-MY3, 98-2221-E-011-015-MY2 and 99-2112-M-032-00-MY3. CHD is thankful to NSRRC for the beam time available on beamlines BL07 and BL12B2.
References
[1] R. Nitsche, D.F. Sargent, P. Wild, J. Cryst. Growth 1 (1967) 52–53.
[2] W. Schafer, R. Nitsche, Mater. Res. Bull. 9 (1974) 645–654.
[3] H. Matsushita, T. Ichikawa, A. Katsui, J. Mater. Sci. 40 (2005) 2003–2005.
[4] O.V. Parasyuk, L.V. Piskach, Y.E. Romanyuk, I.D. Olekseyuk, V.I. Zaremba, V.I.
Pekhnyo, J. Alloys Compd. 397 (2005) 85–94.
[5] J.W. Lekse, M.A. Moreau, K.L. McNerny, J. Yeon, P.S. Halasyamani, J.A. Aitken, Inorg. Chem. 48 (2009) 7516–7518.
[6] J.W. Lekse, B.M. Leverett, C.H. Lake, J.A. Aitken, J. Solid State Chem. 181 (2008) 3217–3222.
[7] H. Katagiri, K. Jimbo, S. Yamada, T. Kamimura, W.S. Maw, T. Fukano, T. Ito, T.
Motohiro, Appl. Phys. Expr. 1 (2008) 041201 (2pp).
[8] R.A. Wibowo, E.S. Lee, B. Munir, K.H. Kim, Phys. Status Solidi A 204 (2007) 3373–3379.
[9] H. Matsushita, T. Ochiai, A. Katsui, J. Cryst. Growth 275 (2005) e995–e999.
[10] I. Tsuji, Y. Shimodaira, H. Kato, H. Kobayashi, A. Kudo, Chem. Mater. 22 (2010) 1402–1409.
[11] D. Yokoyama, T. Minegishi, K. Jimbo, T. Hisatomi, G. Ma, M. Katayama, J. Kubota, H. Katagiri, K. Domen, Appl. Phys. Expr. 3 (2010) 101202 (3pp).
[12] M.L. Liu, F.Q. Huang, L.D. Chen, I.W. Chen, Appl. Phys. Lett. 94 (2009) 202103 (3pp).
[13] C. Sevik, T. Cagin, Phys. Rev. B 82 (2010) 045202 (7pp).
[14] D.M. Schleich, A. Wold, Mater. Res. Bull. 12 (1977) 111–114.
[15] G.Q. Yao, H.S. Shen, E.D. Honig, R. Kershaw, K. Dwight, A. Wold, Solid State Ionics 24 (1987) 249–252.
[16] S. Chen, X.G. Gong, A. Walsh, S.H. Wei, Phys. Rev. B 79 (2009) 165211 (10pp).
[17] S. Chen, A. Walsh, Y. Luo, J.H. Yang, X.G. Gong, S.H. Wei, Phys. Rev. B 82 (2010) 195203 (8pp).
[18] S. Levcenco, D. Dumcenco, Y.S. Huang, E. Arushanov, V. Tezlevan, K.K. Tiong, C.H. Du, J. Appl. Phys. 108 (2010) 073508 (5pp).
[19] S. Levcenco, D. Dumcenco, Y.S. Huang, E. Arushanov, V. Tezlevan, K.K. Tiong, C.H. Du, J. Alloys Compd. 506 (2010) 46–50.
[20] F.H. Pollak, H. Shen, Mater. Sci. Eng. R 10 (1993), xv–xvi.
[21] Y.S. Huang, F.H. Pollak, Phys. Status Solidi A 202 (2005) 1193–1207.
[22] D.E. Aspnes, in: T.S. Moss, M. Balkanski (Eds.), Handbook of Semiconductors, vol. 2, North-Holland, Amsterdam, 1980, pp. 109–154.
[23] J.L. Shay, B. Tell, L.M. Schiavone, H.M. Kasper, F. Thiel, Phys. Rev. B 9 (1974) 1719–1723.
[24] A. Imada, S. Ozaki, S. Adachi, J. Appl. Phys. 92 (2002) 1793–1798.
[25] J.L. Shay, B. Tell, H.M. Kasper, L.M. Schiavone, Phys. Rev. B 5 (1972) 5003–5005.
[26] C. Person, J. Appl. Phys. 107 (2010) 053710 (8pp).
[27] O.W. Shih, R.L. Aggarwal, T.Q. Vu, Y. Shapira, K. Doverspike, R.N. Kershaw, K.
Dwight, A. Wold, Phys. Rev. B 45 (1992) 14025–14035.