Tuan V. Vu*ab,
A. A. Lavrentyevc,
B. V. Gabreliand,
Dat D. Voab,
Pham D. Khange,
L. I. Isaenkofg,
S. I. Lobanovfg,
A. F. Kurus’fg and
O. Y. Khyzhun*h
aDivision of Computational Physics, Institute for Computational Science, Ton Duc Thang University, Ho Chi Minh City, Vietnam. E-mail: vuvantuan@tdtu.edu.vn; voduydat@tdtu.edu.vn
bFaculty of Electrical & Electronics Engineering, Ton Duc Thang University, Ho Chi Minh City, Vietnam
cDepartment of Electrical Engineering and Electronics, Don State Technical University, 1 Gagarin Square, 344010 Rostov-on-Don, Russian Federation
dDepartment of Computational Technique and Automated System Software, Don State Technical University, 1 Gagarin Square, 344010 Rostov-on-Don, Russian Federation
eLaboratory of Applied Physics, Advanced Institute of Materials Science, Ton Duc Thang University, Ho Chi Minh City, Vietnam
fV.S. Sobolev Institute of Geology and Mineralogy SB RAS, 630090, Ac. Koptyug Avenue 3, Novosibirsk, Russia
gNovosibirsk State University, 630090, Pirogova Street 2, Novosibirsk, Russia
hFrantsevych Institute for Problems of Materials Science, National Academy of Sciences of Ukraine, 3 Krzhyzhanivsky Street, 03142 Kyiv, Ukraine. E-mail: khyzhun@ipms.kiev.ua
First published on 17th July 2020
We report the relation between the optical properties and electronic structure of lithium thiogallate (LiGaS2) by performing XPS and XES measurements and theoretical calculations. According to the XPS measurements, the LiGaS2 crystals grown by the Bridgman–Stockbarger method possess promising optical qualities, low hygroscopicity and high stability upon middle-energy Ar+-ion irradiation. The difference in the LiGaS2 band gaps obtained by theoretical calculations and experimental measurements was, for the first time, reduced down to 0.27 eV by applying the Tran–Blaha modified Becke–Johnson (TB-mBJ) potential where the Coulomb repulsion was considered by introducing Hubbard parameter, U. The TB-mBJ+U method also reproduces the XPS spectrum well. The TB-mBJ+U band-structure calculations of LiGaS2 are found to be in good agreement with the XPS and XES experimental data. The accurate electronic structure of LiGaS2 allows further investigation of the optical properties. The relation between the photoluminescence of LiGaS2 and its electronic structure was revealed. Moreover, the theoretical results show the possibility of emissions at higher energy levels in LiGaS2, that has not been measured in experiments yet. Good phase-matching of LiGaS2 was expected to occur at energy levels of 5, 6, 6.2, 7, 7.2, and 8 eV.
It is well-known that the electronic structure of NLO materials plays a crucial role in many properties such as the second-order susceptibilities, the generation of free charge carriers, photoluminescence, absorption edge, and the higher order wave-mixing processes.18–20 Despite the fact that LiGaS2 has been thoroughly studied in both theoretical and experimental work, its electronic structure remains controversial. In 1990, the atomic positions, bond length, and bond angles were determined by Leal-Gonzalez et al. using X-ray techniques.14 In 2003, Isaenko et al. grew LiGaS2 single crystals to study the crystallographic parameters, absorption spectra, and refractive indices.10 In 2009, theoretical calculations were used to describe the electronic structure of LiGaS2 based on which the optical properties of the compound can be revealed. However, the LDA+U calculation was modified according to X-ray photoelectron spectroscopy (XPS) measurements resulting in a band gap of 3.28 eV.21 Many theoretical studies have been performed leading to the same situation where the band gap of the electronic structure is about 3.08–3.29 eV,21–24 which is much lower than the experimental band gap of 4.15 eV.10
In order to achieve more reliable data on the electronic structure, and optical properties of LiGaS2, both experimental and theoretical studies have been performed in the present work. A sample of this compound was grown and measured with XPS and X-ray emission spectroscopy (XES) methods. The band gap was experimentally determined by the cross point between abscissa axis and the line of the absorption spectrum. As the Tran–Blaha modified Becke–Johnson potential with Hubbard parameter U (TB-mBJ+U)25–30 technique has been successfully applied to study the electronic structure and optical properties of different families of sulfides,31–38 it is also applied in the current study. The theoretical result was verified by comparing it with experimental data.
Fig. 1 (a) Photo of as-grown LiGaS2 crystal boule, (b) polished LiGaS2 plate used in the present XPS and XES measurements. |
Fig. 2 (a) Orthorhombic structure of LiGaS2, and distorted (b) LiS4 and (c) GaS4 tetrahedra in the LiGaS2 crystal. |
The electronic structure calculation for LiGaS2 crystals was performed by applying the augmented plane wave plus local-orbital (APW+lo) method43,44 as implemented in the WIEN2k package.45 The Muffin-Tin (MT) model was used to reproduce the system’s full potential without shape approximations, where the MT radius for Li, Ga, and S was 2.09 (a.u), 2.29 (a.u) and 1.97 (a.u), respectively. The expansion of the wavefunction, within a Brillouin zone of 3000 k-points, was limited by the maximum angular momentum lmax = 10, and RMTminkmax = 8. The Fourier expansion of charge density was carried out until Gmax = 14 (a.u.)−1. The exchange–correlation potential was modeled by both traditional GGA-PBE method46 and a TB-mBJ functional.47,48 The Coulomb repulsion of strongly correlated Ga-3d states was treated by the Hubbard technique.49,50 The Hubbard parameter U = 0.515 Ry was applied for 3d states of Ga, because this U value was proven to be suitable for systems containing the Ga element.26 The convergence of a self-consistent process was controlled by setting the charge density difference to be less than 10−4 Ry.
The optical properties of the LiGaS2 crystal were calculated based on the dielectric function ε(ω) = ε1(ω) + ε2(ω)51 which in turn depends on the unit-cell volume V, the crystal momentum k, spin σ, and the eigenvalue energy Ekn of the wave function |knp〉. Applying the Kramers–Kronig relations,52 ε2(ω), and ε1(ω) can be calculated as follows:53
(1) |
(2) |
Fig. 3 Survey XPS spectra of (1) pristine and (2) Ar+-ion irradiated surfaces of the LiGaS2 crystal. |
Detailed XPS measurements of the most important core-level spectra associated with lithium, gallium and sulfur atoms (Fig. 4) indicate that the 3 keV Ar+ ion-treatment does not lead to visible changes in the shapes of the XPS spectra or the binding energy values (Table 1). We do not see the formation of new fine-structure peculiarities in the XPS valence-band spectra in such a case (Fig. 5). Therefore, like in the case of selenides LiGaSe2,26 and LiGa0.5In0.5Se2,28 the LiGaS2 crystal also reveals the high chemical stability. In the LiGaS2 crystal, the XPS S 2p core-level spectrum superimposes the Auger Ga L2M45M45 line, as Fig. 4(c) demonstrates.
Fig. 5 XPS valence-band spectra of (1) pristine and (2) Ar+-ion irradiated surfaces of the LiGaS2 crystal. |
It is worth mentioning that XPS is recognized as an experimental technique that is very sensitive to peculiarities in the chemical bonding in solids.40,54 However, it is also very sensitive to surface charge effects and calibration methods. Therefore, it is beneficial to ensure that XPS measurements are performed with one instrument and the same experimental conditions are used. Previously, we measured XPS spectra of the related selenide, LiGaSe2 (ref. 26), and derived a binding energy of 1117.34 ± 0.05 eV for Ga 2p3/2 core-level spectrum. Since the experiments in ref. 26 were performed under similar experimental conditions as compared to those used in the present work, comparison of the above binding energy value with that derived for the Ga 2p3/2 core-level of LiGaS2 (1117.88 ± 0.05 eV; Table 1) allows us to conclude that the ionic component of the Ga–X bonds (X = S, Se) decreases when in the LiGaX2, the chalcogenide sulfur is substituted by selenium. The comparison of the ionic component of the Li–X bonds (X = S, Se) in LiGaX2 can not be made because the Li 1s spectrum superimposes the Se 3d spectrum in selenide LiGaSe2.26
As can be seen from the data reported in Table 2, the LiGaS2 band gap obtained by PBE-GGA and PBE-GGA+U are 3.259 eV, and 3.202 eV, respectively. These values are just slightly different from previous theoretical studies, and the introduction of the Hubbard parameter U does not lead to a significant improvement of the LiGaS2 band structure, as shown in Fig. 6. The TB-mBJ method results in a remarkable shift of conduction band (CB) to an upper energy level, Fig. 6, giving a LiGaS2 band gap of 4.566 eV, which is about 0.4 eV bigger than the experimental data.10 Meanwhile, the TB-mBJ+U method reduces the difference between theoretical and experimental band gaps down to 0.27 eV. The TB-mBJ+U band structure of LiGaS2 is presented in Fig. 7. As both VB maximum and CB minimum are located at Γ-point, LiGaS2 is also a direct band gap semiconductor like other members in the family LiMX2 (M = In, Ga, X = S, Se).10 This result is consistent with experimental studies.58 The band dispersion near the valence region is mainly flat implying rather high effective mass, and peculiarities of the bonding nature in LiGaS2. Subsequently, the hopping of particles between different regions in real space is expected to be inhibited.
Fig. 6 Band structure of LiGaS2 calculated by (a) GGA-PBE, (b) GGA-PBE+U, and (c) TB-mBJ methods (the Fermi level is set to zero). |
Fig. 7 (a) TB-mBJ+U band structure and DOS of LiGaS2 along the high-symmetric direction in (b) the Brillouin zone. |
In previous work, the TDOS calculated by GAA/LDA fails in describing the d-orbital of Ga which is at very low density.18 The d-orbital of Ga as well as the whole VB is shifted to a higher energy level, and therefore, the calculated band gap is significantly reduced.22 The reliability of the TB-mBJ+U method is confirmed by comparing the TDOS with the XPS spectrum of the VB in the LiGaS2 crystal (Fig. 8). It can be seen that the TB-mBJ+U method is very good with respect to reproducing the XPS spectrum with main peaks in the vicinity of −17.6 eV, −13 to −11 eV, −6 to −4 eV, and −3 to 0 eV. The highest part of the LiGaS2 VB is mainly constructed by the hybridization of s/p-orbitals from the Li, Ga, and S elements. The strong hybridization of Li-2s and S-3p orbitals reveals the covalent nature of the Li–S bonds. Meanwhile, the admixture of S-3p and Li-2p orbitals results in a sharp increase of DOS at the VB maximum. While the d-orbitals of Ga are the only contributors to VB with the lowest part at −17.5 eV, the unoccupied CB is formed by hybridization of Li-2p, Ga-4s, and S-3p orbitals.
Regarding the occupancy of the LiGaS2 VB by S-3p and Ga-4p states, the substantial contributors in the valence-band region, as Fig. 9 presents, show that the above theoretical predictions are reasonably supported by the data of the present XES measurements. Fig. 9 presents the comparison results, on a general energy scale, of the XES S Kβ1,3 and Ga Kβ2 spectra, providing the energy distribution of the S 3p and Se 4p states in LiGaS2, respectively, with the XPS VB spectrum. From Fig. 9, it is clear that the main maximum of the S Kβ1,3 band (S-3p states) is detected at the top of the XPS VB spectrum (peculiarity A), while the main maximum of the XES Ga Kβ2 band (Ga-4p states) is positioned near peculiarity B of the XPS spectrum (cf. Fig. 8 and 9). As can be seen from Fig. 8, the theoretical data also predict substantial input from the Ga-4s states at the bottom of the LiGaS2 VB, while minor contributions of Li-2s, 2p states contribute to the upper portion. However, the available ability of our group does not allow experimental measurements of those states in the LiGaS2 crystal.
Fig. 9 XPS valence-band spectrum of the LiGaS2 crystal matched on a common energy scale with the XES S Kβ1,3 and Ga Kβ2 bands. |
One of the emission maxima in LiGaS2 was observed at 5 eV,58 this is exactly the band-to-band transition between peak A in the VB maximum and the highest peak A* in the CB lowest peak, as shown in Fig. 8. The remaining extrema in the spectra of pink, blue, and violet photoluminescence are at lower intensity and they are observed at 3.4 eV, 3.94 eV, 2.9 eV, 2.74 eV, 2.98 eV, and 3.28 eV.58 The reduction in emission energies may arise from experimental factors including temperature, defects, self-trapped exciton.61,62 In the framework of theoretical calculations, this reduction may also originate from the low intensity of bands near the band gap, as shown in Fig. 8. Since the photoluminescence originates from absorption and recombination processes, the optical properties of LiGaS2, including dielectric function, and absorption rate are calculated in the next paragraph.
The ε1(ω) spectrum of LiGaS2, as shown in Fig. 10(a), is characterized with main extrema A, B, C, D, E, and F in the vicinities of 4 eV, 5 eV, 5.5 eV, 6.5 eV, 7.5 eV, and 9 eV, respectively. The static values of ε1(0) are almost the same (ε1xx(0) = 4.185, ε1yy(0) = 4.150, ε1zz(0) = 4.225). It can be seen in Fig. 10(a), that LiGaS2 possesses high reflectivity for photons with energy of 3–6.5 eV. The polarization of dielectric function is strong in the energy level range 4–11 eV, within this range (ε1xx(0) is slightly higher than (ε1yy(0), and (ε1zz(0). The damping of the electro-magnetic wave occurs at about 9 eV where ε1(ω) becomes negative. At energy levels higher than 13 eV, ε1(ω) is almost isotropic, and slightly increases to a positive value near zero. The imaginary part of the dielectric function (Fig. 10(b)) is only anisotropic in the energy range 6–11 eV. The extrema A, B, C, D, E, and F occur near energy levels of 5.5 eV, 6 eV, 6.5 eV, 7 eV, 8, and 9 eV, respectively. Analyzing Fig. 8 and 10, it is obvious that the extrema in the ε2(ω) spectra correspond to the inter-band transitions from S-3p/Li-2s/Li-2p/Ga-4s states to Ga-4s/Ga-4p/Li-2p/Li-2s states. The inter-band transition happens only from energy levels of −4 to 0 eV in the VB to energy levels of 4–6 eV in the CB. It is also shown that the value of ε2xx(ω) is always lower than those of ε2yy(ω), and ε2zz(ω) which are close to each other.
Fig. 10 Polarization of (a) ε1(ω), and (b) ε2(ω) spectra along x-, y-, and z-directions in the LiGaS2 crystal. |
It is well-known that the imaginary part of the dielectric function is proportional to the joint density of states,66,67 which is relative to the power of emission. As ε2(ω) reaches its extrema at 5.5 eV, 6 eV, 6.5 eV, 7 eV, 8, and 9 eV, it is expected that the emission is from a high intensity of photons in these energy ranges. However, previous measurements58 show only one theoretical maximum at 5 eV. Our theoretical results indicate a need to measure the emission in LiGaS2 at higher energy levels. Meanwhile, in the framework of theoretical calculations, some aspects of the photoluminescence can be revealed via the absorption spectrum α(ω), which was calculated by the TB-mBJ+U method and is presented in Fig. 11.
Fig. 11 Polarization of the absorption spectrum α(ω) along x-, y-, and z-directions in the LiGaS2 crystal. |
For the energy range 0.1–3.9 eV (0.32–11.6 μm), the LiGaS2 compound is reported to be nearly transparent.58 This experimental result is reasonable because the electromagnetic wave absorption of LiGaS2 in this energy level is very low. However, the α(ω) neatly maintains at 104 cm−1 which results in pink, blue, and violet emission with main extrema at 2.74 eV, 2.9 eV, 2.98 eV, 3.28 eV, 3.38 eV, 3.4 eV, and 3.94 eV.63
As shown in Fig. 11, the optical absorption edge is observed near 4.4 eV. The first extremum of α(ω) is near 5 eV, which is also the highest intensity of emission observed in previous work.62 The emission at higher energy levels has not been available, however it can be seen that LiGaS2 single crystal intensively absorbs propagating electro-magnetic waves with photon energies in the range of 7–17 eV, where the α(ω) is in the order of magnitude of 106 cm−1. At energy ranges of 8–11 eV, the absorption αyy(ω) is slightly higher than αxx(ω), and αzz(ω), whose intensities are almost the same. At energy levels higher than 17 eV, there is a sharp decrease of absorption in LiGaS2, and negligible polarization.
Since the polarization of absorption α(ω) is significant at high energy levels, it is expected that the intensity of photoluminescence in LiGaS2 also depends on different directions. It is necessary to note that the emission at high energy level results in a high refractive index.68 Therefore, the photoluminescence is actually observed to be best preserved in the normal direction.69 On the other hand, the reflected emission can cause attenuation of photoluminescence.69 Therefore, the refractive index n(ω) and extinction coefficient k(ω) were calculated and are presented in Fig. 12.
At equilibrium, the refractive indices n(ω) along x-, y-, and z-directions in the LiGaS2 crystal are very close to each other. The difference in the values of n(ω) remains minor in the transparent region of 0–3.9 eV.58 The polarization becomes strong in the energy range of 4.4–13 eV, where nxx(ω) > nzz(ω) > nyy(ω) (Fig. 12(a)). In the energy range of 5–8 eV, n(ω) reaches its extrema in the vicinity of 5, 6, 6.2, 7, 7.2, and 8 eV, where LiGaS2 exhibits its best phase-matching as the birefringence is most remarkable.70 As the conditions for phase-matching are met, the constructive interference is expected to occur leading to a high intensity field. Generally, the intensities of k(ω) along the three crystal directions of LiGaS2 are nearly the same, Fig. 12(b). However, there is a remarkable shift of kyy(ω) to higher energy levels within the energy range of 7–11 eV. At the energy range of 5–17 eV, the attenuation of the propagating wave is at a high rate as the extinction coefficient k(ω) is as high as 1–2. This is also the energy range where the absorption rate α(ω) is the highest. So, the extinction coefficient k(ω) is well proportional to α(ω). Due to the absorption process, the energy-loss L(ω) is partly caused by thermalization during the inter-band transition. As shown in Fig. 12(c), the energy loss becomes significant in the energy range of 5–18 eV, then it sharply increases to the highest values of about 4.5 in the narrow energy range of 18–20 eV. The plasma frequency is expected to be at 19 eV, signifying a plasma frequency of about 4594 THz. This is also the energy level of strong polarization, where Lzz(ω) is obviously higher than Lxx(ω), and Lyy(ω). The reflectivity spectrum R(ω) extends to 22 eV with strong polarization at the extrema in the vicinity of 5, 6, 7, 10, 12, and 16 eV (Fig. 12(d)). The static values of the reflectivity spectrum R(ω) are as follows: (Rxx(ω) = 11.79%, Ryy(ω) = 11.661%, Rzz(ω) = 11.933%).
Despite the fact that the LiGaS2 crystal structure was modeled based on experimental data, both GGA and LDA methods reproduce a band gap that is about 1 eV smaller than the experimental measurements. The improvement of exchange–correlation potential via the TB-mBJ method leads to a band gap of about 0.4 eV higher than the experimental data. The Hubbard parameter in the TB-mBJ+U method plays an important role in reducing the difference between theoretical and experimental results to 0.27 eV.
The nature of the band-to-band transition, which is relative to the photoluminescence in LiGaS2, was studied by analyzing the DOS, TDOS, and XPS spectra. Both the TB-mBJ+U method and XPS spectrum separate the valance band of LiGaS2 into the lowest part at −17.6 eV (mainly contributed to by the d-orbitals of Ga); the middle part at −13 to −11 eV (donated by s/p-orbitals from Li, Ga, and S elements); and the highest part at −6 to 0 eV (donated by s/p-orbitals of Li, and S elements). Meanwhile, the Li-2p, Ga-4s, and S-3p orbitals provide the major contribution to the CB. Regarding the occupancy of the LiGaS2 VB by S-3p and Ga-4p states, the above theoretical predictions are reasonably supported by data of the present XES and XPS measurements when comparing them on the general energy scale of the XES S Kβ1,3 and Ga Kβ2 spectra bringing the energy distribution of S 3p an Se 4p states with the XPS VB spectrum. Furthermore, the XPS data indicate the low hygroscopicity of the LiGaS2 crystal surface. The ionic component of the Ga–X bonds (X = S, Se) decreases when in the LiGaX2 chalcogenide sulfur is substituted by selenium.
LiGaS2 possesses high reflectivity for photons with energies 3–6.5 eV, where the ε1(ω) spectrum reaches its extrema. The polarization of dielectric function is strong in the energy level range 4–11 eV. The theoretical spectra of LiGaS2 dielectric function ε(ω) confirm the band-to-band transition relating to the photoluminescence of pink, blue, and violet spectra, where the emission is observed at energy levels lower than 5 eV. As ε2xx(ω) reaches its extremum beyond 5 eV, it is recommended to measure the emission of LiGaS2 at higher energy levels. The transparency of LiGaS2 in the energy range 0.1–3.9 eV (0.32–11.6 μm) is confirmed by the current theoretical results. The LiGaS2 single crystal is predicted to intensively absorb photons with energies of 7–17 eV, with the absorption rate α(ω) is in the order of magnitude of 106 cm−1, where the polarization is remarkable in such a case.
LiGaS2 can exhibit constructive interference with high intensity fields at 5, 6, 6.2, 7, 7.2, and 8 eV, where the birefringence is most remarkable (nxx(ω) > nzz(ω) > nyy(ω)). The extinction coefficient k(ω) is highest in the energy range of 5–17 eV. The loss energy L(ω) reaches the highest value near 19 eV, signifying a plasma frequency of about 4594 THz. The reflectivity spectrum R(ω) is at high intensity (0.1–0.5) in the energy range of 0–20 eV.
This journal is © The Royal Society of Chemistry 2020 |