Research Update: Emerging chalcostibite absorbers for thin-film solar cells

The most studied materials for the thin-film photovoltaic (PV) technologies are Cu(In,Ga)Se₂ (CIGS, 22.6% one-sun energy conversion efficiency) and CdTe (22.1% efficiency).¹ Nevertheless, the future environmental benefits and potential scale of deployment of these solar cells may be limited by their elemental scarcity (In, Ga, Te) and toxicity (Cd).² These considerations have led to increased interest in emerging light-absorber materials with less toxic and more abundant elements, including Cu₂SnS₃ (CTS),³ Cu₂O,⁴ Sb₂Se₃,⁵ Cu₃N,^6,7 SnS,^8,9 ZnSnN₂,¹⁰ and Cu₂ZnSn(S,Se)₄ (CZTSSe).¹¹ Among these, the highest PV efficiency (certified 12.6%^1,12) has been reported for the CZTSSe absorber. However, its further improvements may be hindered by chemical and crystallographic complexity, pointing researchers toward less complex materials.

Most of the emerging absorbers, like CZTSSe, have a three-dimensional (3D) crystal structure, where dangling bonds in grain boundaries (GBs) can act as recombination centers, causing efficiency loss.^13,14 In contrast, the low-dimensional crystal structure of some potential absorbers, such as 1D-Sb₂Se_3,^5,15 1D-SbSeI,¹⁶ 2D-CuSbS₂,¹⁷ 2D-CuBiS₂,¹⁸ and 2D-CuSbSe₂,¹⁹ can be oriented hypothetically in such a way to decrease the number of dangling bonds, even at the GBs. In turn, this can minimize photoexcited charge-carrier recombination losses, which is one of the major limiting factors for high-efficiency thin-film solar cells.

One class of such non-toxic Earth-abundant less-complex and low-dimensional absorbers is ternary copper chalcogenides, such as CuBiS₂¹⁸ and CuSbCh₂ (where Ch=S, Se) with a chalcostibite crystal structure.^17,19 Despite a relatively small amount of research, these layered semiconductors have shown promising PV device efficiencies approaching 5%.^19–21 The PV-relevant properties of these materials include slightly indirect bandgaps of 1.1–1.6 eV (that is, well-suited for single-junction terrestrial solar cell applications) and strong optical absorption coefficient >10⁵ cm⁻¹ at 0.5 eV above the bandgap (slightly better than CIGSe, CdTe, and CZTSSe, for example).^19,21–23 Many other characteristics that make these chalcostibite materials worth exploring will be discussed in this research update.

A good review of the PV applications of CuSbS₂ (CAS) was published three years ago,²⁴ but much CuSbS₂ research activity has occurred since that time, and there are no review papers about CuSbSe₂ (CASe) as a PV absorber. Thus, in this article, we focus on the most recent research efforts (past three years) on CuSbCh₂ (Ch=S, Se) as a promising family of absorber materials for thin-film solar cell applications. This paper summarizes both the currently used methods to prepare these materials and their most important PV-relevant properties, as compared to other thin-film PV absorbers. We also discuss the factors that have limited current CuSbCh₂-based solar cell performance, as well as some strategies that have been used to address them.

As mentioned earlier, most of the absorbers used in solar cells have a tetrahedrally bonded lattice, where different atoms are linked to one another in 3D diamond-like structures (e.g., Si, CdTe, CuInS₂, and Cu₂ZnSnS₄), as can be seen in Fig. 1(a). However, the CuSbCh₂ chalcostibites show a very different 2D-like layered crystal structure, despite the same 1:1:2 stoichiometry as in CuInS₂ chalcopyrites, and similar Sb/In ionic radii.²⁵ In contrast to the four-fold coordinated chalcopyrite, in the chalcostibite structure the Cu atoms are four-fold coordinated, whereas the Sb atoms are in a distorted three-fold coordination [Fig. 1(b)]. This difference in the metal coordination number makes chalcostibites “line compounds”: namely, the small deviations from the nominal CuSbCh₂ stoichiometry (Cu/Sb = 1) lead to phase impurities, indirectly suggesting a smaller propensity to cation disorder for the stoichiometric material. This is in contrast to the tetrahedrally coordinated CIGS, CTZS, and CTS, which are known to show cation disorder and to tolerate high levels of off-stoichiometry (up to ∼20%).²⁵ The crystallographic differences between the 3D (e.g., CuInS₂), 2D (e.g., CuSbS₂), and 1D (e.g., Sb₂Se₃) absorbers can be seen in Fig. 1.

As can be seen in Fig. 1(b), the CuSbCh₂ crystal structure is formed by stacking layers composed of SbSe₂ and CuSe₃ motifs along the “c”-axis, with interlayer distances of ∼7.10 Å (for Ch=S) and ∼7.29 Å (for Ch=Se).²⁶ Theoretical calculations reveal that these layers (along with other characteristics) are responsible for the increase in the density of states (DOS) in CuSbCh₂ and that each trivalent Sb atom has a non-bonding electron pair from its 5s orbital. These factors are responsible for the larger optical extinction coefficient of CuSbCh₂ than in CIGS, for example, but they also lead to higher effective masses and less efficient collection of the photogenerated charge carriers.^21,26–28

Another interesting property that results from the low-dimensional structure is the possibility of crystallographically orienting the layers in different ways with respect to the substrate. Such preferential orientation may create chemically inert surfaces in 2D structures and benign grain boundaries (GBs) in 1D structures, both with fewer dangling bonds compared to the 3D structures. The benign surfaces or GBs can reduce the carrier recombination loss, improving the efficiency of thin-film solar cells. The preferential orientation is also important to the charge transport inside CuSbCh₂, which is easier within the sheets than between them. Within the CuSbCh₂ sheets, the carriers are transported along chemical bonds, whereas in the other case, they are required to hop between the layers, as in Sb₂Se₃.^19,21,22 Thus, the crystallographic orientation control is vital for reduced carrier recombination and improved carrier transport in CuSbCh₂ PV devices.^19,21,22

A wide variety of techniques has been used to grow CuSbCh₂ and they can be grouped into physical and chemical methods. Physical methods include thermal evaporation²⁹ and co-sputtering,^19,30,31 whereas chemical methods include chemical bath deposition,^32,33 spray pyrolysis,³⁴ spin coating,^20,35 solution processing,³⁶ electrodeposition,^37,38 and solvo-/hydro-thermal synthesis.^39,40 Below, we focus mostly on the physical vapor deposition methods for CuSbCh₂ growth.

The control of the CuSbCh₂ composition during growth is very important because small deviations from the nominal stoichiometry more easily lead to phase impurities than in CIGS and CZTSSe.²⁵ For this reason, self-regulated growth (a.k.a. adsorption-controlled growth) is the most promising approach to achieve reproducible synthesis of this material, regardless of the deposition method. This kind of CuSbCh₂ growth control has been demonstrated for CuSbCh₂ by radio-frequency (RF) magnetron co-sputtering of Cu₂Ch and Sb₂Ch₃ targets in excess of Sb₂Ch₃ vapor.^19,21,22,41 For CAS, it was found that it is possible to grow highly stoichiometric phase-pure thin films at the substrate temperature of 350 °C in a wide range of Sb₂S₃ over-flux.²² The CAS films showed some Sb₂S₃ impurities at lower temperatures or decomposed to Cu₁₂Sb₄S₁₃ at higher temperatures.

To determine the stability region for the deposition of stoichiometric phase-pure CAS films, the phase diagram has been calculated as a function of temperature and Sb₂S₃ partial pressure [Fig. 2(a)].²² It was concluded that for a given substrate temperature, phase-pure CuSbS₂ can be grown in a ∼10² Torr dynamic range of Sb₂S₃ partial pressure. In turn, for a given Sb₂S₃ partial pressure, the region of stability extends ∼70 °C of substrate temperature in agreement with experimental data. Overall, these results suggest that no fine control of the Sb₂S₃ precursor flux or substrate temperature is required within this region of CuSbS₂ self-regulated growth. Similar conclusions hold for CuSbSe₂ adsorption-controlled growth in an Sb₂Se₃-rich atmosphere, but at slightly higher temperatures (∼380 °C), and with somewhat different competing phases [Sb₂Se₃, Cu₃SbSe₃, and Cu₂Se in Fig. 2(b)].

It is worth mentioning that the CuSbCh₂ films grown by different methods often show small grains and/or low crystallinity, encouraging the use of post-deposition thermal treatments (TTs) to improve film quality. Such a treatment has been demonstrated under atmospheric conditions,^30,35,42 H₂S gas and S vapor,²⁰ or Sb₂S₃ vapor.^30,31,43,44 For example, taking into account the calculated stability of the CuSbCh₂ phase under an Sb₂Ch₃-rich environment [Fig. 2(a)], it has been shown that the TT of the sputtered CAS thin films under Sb₂S₃ vapor increases their grain size without affecting the composition or phase purity. This TT also improved the structural quality and optoelectronic properties of the films, leading to more reproducible and efficient TT-CuSbS₂ PV devices.⁴⁴ In this context, it is also important to note that CuSbCh₂ has a lower melting point of 480 °C (for Ch=Se) and 535 °C (for Ch=S), compared to CIGS, CZTSSe, and CdTe (∼1000 °C^21,45). This should lead to grain growth at a lower temperature, making CuSbCh₂ a suitable absorber for the fabrication of flexible thin-film solar cells on polymer substrates.

Based on the experimental and calculated data shown in Fig. 2, the most likely impurities found in CuSbCh₂ (Ch=S or Se) are the following phases: Sb₂Ch₃, Cu_1.8-2Ch, Cu₁₂Sb₄S₁₃ (for the Cu-Sb-S system), and Cu₃SbSe₃ (for the Cu—Sb—Se system). As shown by the simulated X-ray diffraction (XRD) reference patterns in Fig. 3, there are several regions of superposition between the diffraction peaks of the CuSbCh₂ and impurity phases. For CAS [Fig. 3(a)], the most difficult impurity to identify by XRD is Cu₁₂Sb₄S₁₃ (12-4-13 phase) because its main peak is at the same position as the CAS (200)/(013) peaks. Thus, a common mistake made in the literature is attributing an increase in the (200)/(013) peaks compared to the (111)/(104) peaks to preferential orientation of CAS, instead of the actual presence of the very stable 12-4-13 secondary phase.

The Raman spectroscopy can be useful for determining the presence of this impurity because its main peak is ∼20 cm⁻¹ higher compared to the CuSbS₂ main peak at 332 cm⁻¹.⁴⁷ Also, the CuSbS₂ thin film with the Cu₁₂Sb₄S₁₃ phase impurity often has higher conductivity and higher hole density (10⁻¹ to 10² S/cm and 10¹⁸–10²¹ cm⁻³) than phase-pure CuSbS₂ (10⁻³ to 10⁻² S/cm and 10¹⁶–10¹⁷ cm⁻³). Thus, it should be possible to combine electric measurements with XRD and Raman spectroscopy characterizations to rule out the existence of the 12-4-13 phase.^22,41 The phase purity is equally important but less difficult to determine in CuSbSe₂ [Fig. 3(b)], where less stable Cu₃SbS₃ and Cu_1.8Se secondary phases can be more easily identified by XRD, despite some overlap with the weaker CuSbSe₂ (113)/(105)/(203) peaks.

The knowledge of the band structure and DOS helps us to understand the optoelectronic properties of CuSbCh₂ compared with the more studied absorber materials such as CIS, as shown in Fig. 4. It has been predicted that chalcostibite CuSbS₂ and CuSbSe₂ have slightly indirect bandgaps [E_g^dir–E_g^ind ≈ 0.1 eV, Figs. 4(a) and 4(b)] in contrast with the well-defined direct bandgap of chalcopyrite CuInSe₂ [Fig. 4(c)]. Chalcostibites also present higher DOS compared to chalcopyrites. This band-structure feature not only results in a larger optical extinction coefficient (more effective absorption of photons) but also in higher effective masses (less efficient charge-carrier collection for the PV device).²⁸ 

Theoretical calculations in combination with experimentally measured X-ray photoemission spectra show that the top of the valence band (VB) of CAS is mainly formed by strongly antibonding Cu 3d and S 3p states. In contrast, Sb 5p and S 3p states are main contributions for the bottom of the conduction band (CB).^28,48 It is also interesting to note that Sb 5s electrons in CuSbCh₂ are not fully inert and localized lone pairs. Instead, these Sb 5s states contribute somewhat to the formation of bonding states in the VB and antibonding states in the CB. These states are also partially associated with the rising of the energy levels of the bands,⁴⁸ which has important implications for the selection of contacts to CuSbCh₂ absorbers.

In agreement with the theoretical observations, the CuSbCh₂ materials feature strong experimental optical absorption coefficients of >10⁵ cm⁻¹ for photon energies 0.5 eV higher than their bandgaps,^19,21–23 as shown in Fig. 5(a), and have slightly indirect bandgaps of 1.1–1.2 eV^19,21 (for Ch=Se) and 1.4–1.6 eV (for Ch=S)⁴⁴ [Figs. 5(b) and 5(c)]. In Fig. 5(b), the presence of an electronic defect 0.22 eV below the bandgap energy can also be observed (i.e., 0.22 eV close to VB or CB); the different defects likely present in CuSbCh₂ samples will be discussed in Sec. II D.

CuSbCh₂ possesses p-type conductivity and presents a tunable hole concentration in a range of 10¹⁵–10¹⁸ cm⁻³ suitable for PV application.²² The hole mobility (μ_h) has been measured by the Hall effect, showing values of 0.1–49 cm² V⁻¹ s⁻¹ (for Ch=S)³⁵ and 12 cm² V⁻¹ s⁻¹ (for Ch=Se).²¹ The large variation in the values of μ_h reported in the literature may be related to differences in both the concentration of defects and anisotropic crystal structures in combination with different crystallographic textures of the films. Based on optical-pump THz probe (OPTP) transient reflection spectroscopy, it has been observed that the hole and electron mobilities in 2D-like CuSbS₂ are similar to each other, in contrast to the significantly different hole and electron mobilities in 3D-bonded materials such as CIGS, CZTS, and CTS.⁴⁴ 

Some of the most important characteristics of the PV absorbers that should be considered for their device performance are defect properties and photoexcited charge-carrier dynamics. Thus, despite a relatively small number of CuSbCh₂ publications on these topics, such studies are crucial for understanding CuSbCh₂ thin-film solar cell performance and for enhancing their efficiencies. There are many methods that can be used to study defects, including first-principles calculations, photoluminescence (PL) spectroscopy, admittance spectroscopy (AS), and deep-level transient spectroscopy (DLTS). The photoexcited charge-carrier dynamics can be studied using transient OPTP reflection spectroscopy and time-resolved photoluminescence (TRPL). Some of these techniques have been previously reported for studying defects and recombination in CuSbCh₂ materials.

First-principles calculations of native point defects in CuSbS₂⁴⁹ have been performed for different chemical potentials, such as the experimentally used Sb₂S₃-rich conditions [Fig. 6(a)]. The results show that the most plentiful acceptor is the copper vacancy (V_Cu), with a very shallow transition level (0/1−) at ∼0.03 eV above the valence band maximum (VBM). Another active acceptor defect is the copper on antimony antisite (Cu_Sb), showing deeper transition levels of 0.1 and 0.2 eV. The likely dominant donor defects are Cu interstitials (within the layer—Cu_i-in and between the layers—Cu_i-out), which are very shallow (transition levels close to the conduction band minimum, CBM). In addition, the sulfur vacancies (V_S-in and V_S-out) also have relatively low formation enthalpy (ΔH_f). The (0/1+) transition level is around 0.15 eV and 0.35 eV for V_S-out and V_S-in, respectively, and the (1+/2+) transition level for V_S-out is around 0.10 eV. Hence, these vacancies are amphoteric defects that may act as donors or as acceptors, depending on the position of the Fermi level (E_F). These data are in agreement with the previously reported properties of defects in CuSbS₂.³⁵ Similar conclusions for CuSbSe₂^19,26 can be made, as can be seen in Fig. 6(b).

From an experimental point of view, defects in different CuSbS₂ samples (as-deposited, TT, and micro-crystals) have been studied by AS, PL, and DLTS.⁴⁹ Three acceptor defects were found at about 0.08, 0.17, and 0.24 eV above the VBM by the capacitance-based methods, in agreement with free-to-bound radiative photoluminescent transitions observed in these thin films. Comparing these results to the theoretical calculation [Fig. 6(a)], the shallower defect was associated with V_Cu, whereas the two deeper ones were attributed to Cu_Sb and/or V_S, as can be seen in the diagram of Fig. 7(a). No investigation of the experimental behavior of the CuSbSe₂ defects can be found in the literature.

To understand the effects of point defects on CuSbCh₂ photoexcited charge carriers, transient OPTP reflection spectroscopy [a.k.a time-resolved terahertz spectroscopy (TRTS)] has been performed. An experimental study of the carrier dynamics in the solution-processed CuSbS₂ nanoplates³³ indicates a minority lifetime (τ) of up to ∼1 ns and a mobility (μ) of ∼1 cm² V⁻¹ s⁻¹, which result in a diffusion length (L_d) estimate of ∼60 nm [L_d = (kTμτ/e)^1/2 Ref. 21]. Similar results have been shown using the OPTP for the thermally treated CuSbS₂ films (τ of 0.7 ns, μ of ∼4 cm² V⁻¹ s⁻¹, and estimated L_d of ∼90 nm).⁴⁴ The photoexcited charge-carrier dynamics in CuSbSe₂ thin films studied using the OPTP indicate higher μ (∼12 cm² V⁻¹ s⁻¹), but smaller τ (0.2 ns), and hence similar L_d (∼80 nm). Figure 7 shows the time-domain OPTP spectra for CuSbS₂ and CuSbSe₂ thin films.

Finally, a summary of the important optoelectronic properties of the chalcostibite CuSbCh₂ is shown in Table I, in comparison with the chalcopyrite Cu(In,Ga)Se₂ and kesterite Cu₂ZnSn(S,Se)₄. Overall, both chalcostibites show similar properties to one another but somewhat different if compared to chalcopyrites and kesterites. Despite the higher absorption coefficient shown by the 2D chalcostibite materials, they show larger effective masses of charge carriers and lower photoexcited charge-carrier lifetime. However, the carrier concentrations and the bandgaps of chalcostibites and chalcopyrites/kesterites are quite similar.

Advancing from intrinsic properties of the CuSbCh₂ material to CuSbCh₂ device integration, the development of suitable front contacts (also known as “buffers”) and back contacts is an important direction of research. This is related to the anisotropic properties of the layered CuSbCh₂ structure, which makes the surface energies, densities of surface states, and energy band positions [ionization potentials (IP) and electron affinities (χ)] different than in CIGS and related absorbers. First-principles calculations indicate that CuSbS₂ has two lower-energy surfaces: the (001)-plane with a surface energy of 12.4 meV Å⁻¹ [(a) and (b) planes in Fig. 1(c)] and the (010)-plane with 14.6 meV Å⁻¹. As shown in Fig. 8(a), the (001) surface has a lower density of surface states than the (010) surface, which is consistent with the weak bonding of the (001) planes. Figure 8(b) shows the calculated energy band positions for [001]- and [010]-oriented CuSbS₂ absorber surfaces compared to the commonly used CdS contact layer. These results suggest a cliff-type CuSbS₂/CdS band offset of 0.85–1.43 eV, depending on the surface orientation.⁴¹ 

Experimental measurements of IP and work function (Φ) on CAS thin films have been performed by X-ray photoemission spectroscopy (XPS)⁴⁸ and ultraviolet photoemission spectroscopy (UPS).³⁵ XPS for the as-deposited and thermally treated CuSbS₂ is shown in Fig. 9(a). The IP of 4.98/5.25 eV (XPS/UPS) and the Φ of 4.73/4.86 eV (XPS/UPS) were found. Combining these results with the experimentally measured direct/indirect bandgap energy (1.58 eV/1.48 eV⁴¹) of the CuSbS₂ and the estimated band position of the CdS buffer (IP of 7.1 eV vs. vacuum level and bandgap of 2.6 eV⁵⁵), the CBM of this absorber is expected to be 0.7–1.1 eV higher than that of CdS. These experimental results (Fig. 9) are consistent with the theoretical results (Fig. 8) and suggest the presence of interfacial recombination centers.⁵⁶ Temperature-dependent J–V measurements (JV–T) performed on CuSbS₂/CdS devices [Fig. 9(b)] showed an open-circuit voltage (V_OC) extrapolated to 0 K of ∼0.7 V, also indicative of a 0.7–0.9 eV cliff-type band offset at the CuSbS₂/CdS interface.⁴⁴ In contrast, the Fermi level and VBM of CuSbSe₂ measured by UPS are −4.63 eV and −4.88 eV with respect to vacuum level (0 eV), respectively.²⁶ Thus, CuSbSe₂ may have a less detrimental CB offset with CdS, compared to CuSbS₂.

Motivated by the high CB offset between the CuSbCh₂ absorber and the commonly used CdS contact, some research has been directed at alternative buffer layers, such as undoped and Ga-doped Cd_1−xZn_xS deposited by atomic layer deposition.⁵⁷ The CuSbS₂ devices made with Zn-rich undoped buffers (Cd_0.61Zn_0.39S and Cd_0.14Zn_0.86S) showed high series resistance and very low photoresponse, but 2.4%–5.5% Ga doping in Cd_0.6Zn_0.4S partially addressed these problems. Comparing the performance of the standard chemical-bath-deposited CdS, V_OC improved from 211 mV to 449 mV, the short-circuit current density (J_SC) improved from 3.82 to 6.24 mA cm⁻², and the efficiency increased from ∼0.3% to 1%. Addressing the issue of Cd toxicity, the CdS/ZnO layers were replaced with GaN/In_0.15Ga_0.85N in CAS devices, achieving the efficiency of ∼3% and the highest value of J_SC (∼34 mA cm⁻²) reported in the literature,³¹ which is above the theoretical limit for this 1.5 eV bandgap material.

Different back-contact materials have been also evaluated for CAS PV devices, such as Au, W, Ni, Pd, Pt, FTO, and Mo.⁴¹ Many of these back contacts did not result in functioning PV devices due to delamination, pinholes, interface reaction, or other reasons. Among these back contacts, Mo provided the best current collection, even though its work function (4.35–4.90 eV⁵⁸) was not favorable for charge extraction from the CuSbS₂ absorber. Thus, for improving the back contact, the effect of the addition of charge-selective layers (CSLs) on Mo was also studied.⁴¹ The MoO_x CSL-based device showed a significant increase in efficiency (from 0.49% to 0.86%) because of the improvement in J_SC (from 3.53 to 8.91 mA cm⁻²). This effect on J_SC was explained by the deeper (6.6 eV⁵⁹) work function of the MoO_x layer, which promotes an upward band bending in the absorber, reflecting the photogenerated electrons.⁴¹ A similar device with a sulfurized hybrid ink/spin-coated CuSbS₂ thin-film absorber has shown a record efficiency of 3.22% and J_SC of 15 mA cm⁻² (see Table II), albeit without statistical histograms or certification results reported in the paper.²⁰ This high performance may be related to the carbon-containing layer at the Mo/CuSbS₂ interface measured by Auger electron spectroscopy (AES) depth profiling, which may also aid charge selection.

For the most part, CuSbCh₂ research has so far focused on the intrinsic optoelectronic properties of the materials (e.g., absorption coefficient, bandgap, carrier concentration). However, significant challenges arise on integrating absorber thin films into PV device prototypes,¹⁹ such as the choice of adequate back/front contacts, growth of pinhole-free absorber layers, optimization of absorber thickness, and possible chemical reactions or inter-diffusion of the absorber with contacts. All of these parameters are very time-consuming to evaluate using the traditional experimental approach. To address this challenge, high-throughput experimental (HTE) combinatorial research methods^60,61 can be extended from materials studies^62,63 to accelerate PV device research.^64,65

The first attempt to apply the HTE approach to CuSbCh₂ devices has focused on studying the effects of the crystallographic orientation, phase purity, composition, morphology, and thickness of the CuSbS₂ absorber on the PV performance of devices with the SiO₂/Mo/CuSbS₂/CdS/i-ZnO/Al-ZnO/Al architecture and on evaluating different back contacts.⁴¹ In this particular research, a <1%-efficient device was obtained with the optimal absorber thickness of 0.8 μm and MoO_X CSL (device #3 in Table II).⁴¹ Further, an improvement in CuSbS₂ PV device performance (1%) and reproducibility was achieved after absorber thermochemical treatment (TT) under Sb₂S₃ vapor followed by selective KOH chemical etch of a likely Sb₂S₃ surface layer, prior to the deposition of front contacts (device #4 in Table II).⁴⁴ 

A similar HTE approach was also applied in the accelerated development of 3%–5%-efficient CuSbSe₂ devices (device #12 in Table II) and to understand the tradeoffs of light absorption and charge transport in these layered materials.^19,21 In an effort to suppress the tradeoff between J_SC and V_OC observed in this drift PV device, the Na incorporation in the absorber was investigated. The Na incorporation has enhanced V_OC by ∼50 mV (comparing device #13 with #12 in Table II), but it also caused a decrease in J_SC and efficiency.²¹ Several other adjustments in the device fabrication process were attempted, compared to the baseline device (#14 in Table II).¹⁹ Some improvement of the efficiency resulted from using an MgF anti-reflection coating and from better design of the front-contact collector grids.²¹ 

Comparing the performances of the CuSbS₂ and CuSbSe₂ devices with similar structure (e.g., devices #4 and #12 in Fig. 10 and Table II), it is seen that V_OC for both materials is 300–400 mV [Fig. 10(a)], which is very low compared to their bandgaps (1.1–1.5 eV). This effect may be directly related to the low quasi Fermi-level splitting due to insufficient absorber quality, and to large CB offset at the interface between these absorbers and the commonly used CdS front contacts. However, the CuSbSe₂ devices have shown higher efficiencies, mainly caused by higher J_SC values. The higher J_SC may be related to the better overlap of the CuSbSe₂ absorption with the solar spectrum indicated by better external quantum efficiency (EQE) [Fig. 10(b)] and by lower recombination of the photogenerated charge carriers at the front interface. To represent this comparison, Fig. 10 shows comparative graphs of current density-voltage (J-V) and EQE for these CuSbCh₂ PV devices. The poor carrier collection for photon energies near the band edge can be observed in both CuSbS₂ and CuSbSe₂ EQE graphs [Fig. 10(b)], which is mainly indicative of the short transport lengths of the photogenerated charge carriers.

In addition to the traditional substrate architecture (CIGS-like) of the PV devices described earlier, several other devices in substrate architectures and one in superstrate architectures (CdTe-like) have been attempted. These diverse CuSbCh₂ PV device configurations and performances are summarized in Table II. Most of the CuSbCh₂ PV device prototypes fabricated in the substrate configuration have the glass/back-contact/CuSbCh₂/buffer/TCO/front-contact architecture, where TCO is a transparent conductive oxide such as Al:ZnO or Sn:In₂O₃. Only one of the PV devices has a superstrate configuration with the glass/TCO/CuSbCh₂/buffer/back-contact architecture (i.e., device #6 in Table II), where TCO is F:SnO₂. So far, substrate devices had higher efficiencies than this superstrate device, further supporting the analogy of CuSbCh₂ materials with CIGS. However, in both cases, more research and development toward higher-quality absorbers and suitable contacts is needed to improve the efficiencies of CuSbCh₂ PV devices.

This article provides a research update on the emerging chalcostibite family of absorber materials for thin-film photovoltaic solar cells, with particular focus on the most recent research efforts (past 3 years). The CuSbCh₂ (Ch=S, Se) features layered the 2D-like chalcostibite crystallographic structure, in contrast to the 3D-like chalcopyrite structure of CuInSe₂ and CuGaSe₂. Because of the narrow composition phase width of CuSbCh₂, these absorber materials require a self-regulated adsorption-limited synthesis method to avoid detrimental Cu-rich competing phases such as Cu₁₂Sb₄S₃ or Cu₂Se. The band structure and optoelectronic properties of CuSbCh₂ feature higher density of states and optical absorption, but heavier effective masses and worse charge transport, compared to CIGS. The moderate hole density is set by compensation between copper vacancies and interstitials, whereas the charge recombination and carrier dynamics are determined by chalcogen vacancies and Cu-on-Sb antisites. The PV-relevant optoelectronic properties of CuSbS₂ and CuSbSe₂ are summarized in Table I, where they are also compared with the more studied CIGS and CZTS absorbers.

Both CuSbS₂ and CuSbSe₂ have large cliff-like interface band offsets with CdS, calling for further development of the chalcostibite contacts. The chalcostibite photovoltaic device efficiencies are currently limited to 1%–3% for CuSbS₂ and 3%–5% for CuSbSe₂ by short transport lengths of the photogenerated charge carriers and by detrimental voltage and current effects of the contact band offset. Hence, more research is needed on the chalcostibite absorber quality and contact selection. Table II summarizes device architectures and the resulting device performances of different CuSbCh₂ solar cells reported in the literature. In summary, chalcostibite CuSbCh₂ materials are a promising non-toxic and Earth-abundant family of absorbers for applications in thin-film photovoltaic solar cell devices.

The CuSbCh₂ work at the National Renewable Energy Laboratory (NREL) managed and operated by Alliance for Sustainable Energy, LLC, was supported by the U.S. Department of Energy, Office of Energy Efficiency and Renewable Energy, Solar Energy Technologies Office, under Contract No. DE-AC36-08GO28308 to NREL. The views expressed in the article do not necessarily represent the views of the DOE or the U.S. Government. The U.S. Government retains a nonexclusive, paid-up, irrevocable, worldwide license to publish or reproduce the published form of this work, or allow others to do so, for U.S. Government purposes. F.W.S.L. was funded by the São Paulo Research Foundation (FAPESP), Grant Nos. 2016/10513-3 and 2014/12166-3.