2 Collaborative Innovation Center of Extreme Optics, Shanxi University, Taiyuan 030006, China
† Corresponding author. E-mail:
The kesterite thin film solar cells based on the quaternary Cu2ZnSnS4 and Cu2ZnSnSe4 and their alloys Cu2ZnSn(S,Se)4 have been considered as environment-friendly and non-toxic alternatives to the currently commercialized CdTe and Cu(In,Ga)Se2 thin film solar cells. From the theoretical point of view, we will review how the group I2–II–IV–VI4 quaternary compound semiconductors are derived from the binary CdTe and the ternary CuInSe2 or CuGaSe2 through the cation mutation, and how the crystal structure and electronic band structure evolve as the component elements change. The increased structural and chemical freedom in these quaternary semiconductors opens up new possibility for the tailoring of material properties and design of new light-absorber semiconductors. However, the increased freedom also makes the development of high-efficiency solar cells more challenging because much more intrinsic point defects, secondary phases, surfaces, and grain-boundaries can exist in the thin films and influence the photovoltaic performance in a way different from that in the conventional CdTe and Cu(In,Ga)Se2 solar cells. The experimental characterization of the properties of defects, secondary phase, and grain-boundaries is currently not very efficient and direct, especially for these quaternary compounds. First-principles calculations have been successfully used in the past decade for studying these properties. Here we will review the theoretical progress in the study of the mixed-cation and mixed-anion alloys of the group I2–II–IV–VI4 semiconductors, defects, alkaline dopants, and grain boundaries, which provided very important information for the optimization of the kesterite solar cell performance.
The kesterite semiconductors Cu2ZnSnS4 and Cu2ZnSnSe4 and their alloys Cu2ZnSn(S,Se)4 have drawn intensive attention as the light-absorber materials in thin film solar cells since 2010.[1–3] The rapid development of the kesterite solar cells in the past years has been reviewed in a series of review papers or books[4] as well as in this special issue.
Both Cu2ZnSnS4 and Cu2ZnSnSe4 belong to the group I2–II–IV–VI4 quaternary compound semiconductors. The study on the I2–II–IV–VI4 quaternary semiconductors dates back to 1950s, and a series of I2–II–IV–VI4 semiconductors have been synthesized by different methods.[5–8] From the view of theoretical material design, the quaternary I2–II–IV–VI4 semiconductors, in which both the anions and cations have tetrahedral coordination, can be designed through a two-step cation mutation (substitution) from the binary II–VI semiconductors in the zincblende or wurtzite structure with tetrahedral coordination,[1,9–13] as shown schematically in Fig.
![]() | Fig. 1. (color online) The schematic plot of the cation-mutation from the group IV elemental semiconductors to the group III–V and II–VI binary compound semiconductors, and then to the group II–IV–V2 and I–III–VI2 ternary compound semiconductors and the group I–III-IV2–V4 and I2–II–IV–VI4 quaternary compound semiconductors.[1,9–13] |
The application of Cu2ZnSnS4 (CZTS) in solar cells dates back to 1988 when Ito and Nakazawa synthesized Cu2CdSnS4 and Cu2ZnSnS4 thin films, fabricated a CZTS solar cell, and reported a 165 meV open-circuit voltage.[16] During the 14 years from 1996 to 2009, Katagiri et al. have pioneered the fabrication of Cu2ZnSnS4 thin film solar cells with a device structure ZnO:Al/CdS/CZTS/Mo/SLG (similar to that of the Cu(In,Ga)Se2 solar cells), and increased the energy-conversion efficiency from 0.66% (an open-circuit voltage of 400 mV) in 1996 to 6.7% in 2009.[3,17] In 1997, Friedlmeieret al. also fabricated a Cu2ZnSnS4 solar cell with 2.3% efficiency and a Cu2ZnSnSe4 cell with 0.6% efficiency.[18]
Despite the steady efficiency increase during the period of 1996–2009, Cu2ZnSnS4, Cu2ZnSnSe4, and also all other I2–II–IV–VI4 semiconductors had not attracted special attention in the photovoltaic field, and many fundamental properties (crystal structure, electronic structure, optical properties) of I2–II–IV–VI4 had not been well characterized and were even wrong.[19] Until 2009, Cu2ZnSnSe4 was even reported experimentally to crystallize in the stannite structure (rather than the kesterite structure), and its band gap was reported to be around 1.5 eV, which is close to that of Cu2ZnSnS4, indicating that it is impossible to tune the band gap through forming Cu2ZnSn(S,Se)4 alloys.[1,9] Using the first-principles calculations, we studied the chemical trends in the crystal structure and electronic band structure of a series of I2–II–IV–VI4 semiconductors, showing that the ground-state structure of Cu2ZnSnSe4 is the kesterite structure and its band gap is 1.0 eV (0.5 eV smaller than that of Cu2ZnSnS4), and thus pointed out the long-standing misunderstanding in the crystal structure[20] and band gaps of Cu2ZnSnSe4 as well as other I2–II–IV–VI4 semiconductors.[1,9]
Since 2010, the Cu2ZnSnS4 and Cu2ZnSnSe4 systems become a hot topic in the photovoltaic field, especially after the publication of the experimental paper by Mitzi et al. who reported a hydrazine-based deposition of the Cu2ZnSn(S,Se)4 (CZTSSe) alloys and a fabrication of a 9.7% efficiency solar cell based on the CZTSSe alloys.[2] The record efficiency was then increased to 11.1% in 2012,[21] and to 12.6% in 2013.[22] In 2010, Ahn et al. also reported their experimental determination of the band gap of Cu2ZnSnSe4 at 1.02 eV.[23] Nowadays the thin film solar cells using Cu2ZnSnS4 (with a band gap at 1.5 eV), Cu2ZnSnSe4 (with a band gap at 1.0 eV), and their alloys Cu2ZnSn(S,Se)4 (with a linearly and continuously tunable band gap from 1.0 eV to 1.5 eV) as the light-absorber layer are called CZTS, CZTSe, and CZTSSe solar cells, respectively, or more generally kesterite solar cells.
The fundamental crystal structure, electronic structure, and defect properties have been reviewed in a series of papers[13,24,25] or book chapters.[4] Here we will introduce the crystal structure and electronic band structure first, and review the theoretical study on the mixed-cation and mixed-anion alloys, alkaline dopants, and grain boundaries in detail, which we believe will be very important for the further optimization of the kesterite solar cell performance.
Since the I2–II–IV–VI4 semiconductors can be derived from the binary II–VI semiconductors which usually crystallize in the zincblende or wurtzite structure with the tetrahedral coordination, their crystal structures can also be derived from the zincblende and wurtzite structures, as shown in Fig.
![]() | Fig. 2. (color online) The crystal structure mutation from (a) zinc-blende ZnS to (b) chalcopyrite CuGaS2 and (c) CuAu-like CuGaS2, then to (d) kesterite Cu2ZnSnS4, (e) stannite Cu2ZnSnS4, and (f) PMCA-Cu2ZnSnS4, and the mutation from (g) wurtzite ZnS to (h) wurtzite-chalcopyrite CuGaS2 and (i) wurtzite-CuAu CuGaS2, then to (j) wurtzite-kesterite Cu2ZnSnS4 and (k) wurtzite-stannite Cu2ZnSnS4. Adapted from Refs. [4], [9], and [10]. |
However, when the cation-size difference is large, both the kesterite and stannite structures may become the lowest-energy structures, e.g., the stannite structure has lower energy than kesterite for Cu2Cd–IV–S4, while Ag2ZnSnS4 and Ag2ZnSnSe4 are still more stable in the kesterite structure. Zhang et al.[28] have studied a series of Cu-based quaternary semiconductors and proposed that the stannite structure is energetically more stable for systems containing Cd. The calculated lowest-energy structures for a series of I2–II–IV–VI4 semiconductors can be found in Ref. [20]. For the lattice constants, no matter the ground state structure is kesterite or stannite, the most stable one always has larger a and smaller
The crystal structure mutation is also possible for the wurtzite structure. As shown in Figs.
Although the ground state structures of Cu2ZnSnS4, Cu2ZnSnSe4, Cu2ZnGeS4, and other I2–II–IV–VI4 semiconductors have been determined, the unique partial disorder makes the real structure of the synthesized thin films more complicated for the standard x-ray diffraction characterization. In kesterite structure, the/Cu+Zn/Cu+Sn/Zn+Cu/Sn+Cu/layers are ordered along the (001) direction, as shown in Fig.
![]() | Fig. 3. (color online) The crystal structures of (a) the kesterite and (b) Cu/Zn partially disordered kesterite Cu2ZnSnS4. Partially disordered means that the Cu/Zn disordered occupation exists only in the (001) CuZn layers, so the symmetry of the crystal lattice is also I-4, equivalent to that of the stannite structure. Adapted from Ref. [4]. |
It is also interesting to see that the partial disorder in the (001) layers has negligible influence on the electronic structure, so its influence on the electrical properties may be small although the partial disorder may increase the carrier scattering and decrease the mobility. In contrast, the structure change from kesterite to stannite can decrease the band gap by 0.2 eV. The exchange of Cu and Zn in the kesterite structure (formation of a CuZn+ZnCu antisite pair in a 128-atom supercell) decreases the band gap by about 0.1 eV and costs a formation energy about 0.2 eV/pair. As we can see, the partial disorder in the (001) layer costs less energy and also induces smaller band gap change than single antisite-pair or the structure change from kesterite to stannite.
Since 2013, the cation mutation has attracted more attention, and Be, Mg, Ca, Sr, and Ba are also considered as the group II cations, whereas Ti, Zr, and Hf are considered as the group IV cations.[32–36] One important reason for the substitution of elements is to increase the size difference between the cations and thus suppress the formation of caton antisites. Zhong et al.[33] showed in 2016 that when Zn is replaced by Mg and Ca, leading to Cu2MgSnS4 and Cu2CaSnS4, respectively, the lowest-energy (ground state) structure changes from kesterite to stannite, which shows the same trend as in the case of Cu2Cd–IV–S4 series. In 2017, Chen et al.[36] calculated the electronic structure of Cu2BeSnS4 and predicted a band gap of 1.76 eV, and also showed that Cu2BeSnS4 exhibits benign defect properties for the large difference of its atomic size, thus evidently lowering the concentration of the antisite defects. The absorption coefficient is as high as 105 cm−1 according to their calculation. However, the toxic element Be limits the development of practical solar cells.
Interestingly, Cu2–II–Sn–VI4 (II = Ba, Sr; VI = S, Se) were proposed as promising light-absorber materials for solar cells and photoelectrochemical water-splitting applications during the last year.[34,35] For these systems with very large group II cations, the crystal structure is very different from the kesterite or stannite structure, and the lowest-energy structure is usually not those with the tetrahedral coordination. Since the commercialized solar cells are most based on Si, CdTe, and Cu(In,Ga)Se2 with tetrahedral coordination, it will be interesting to see the photovoltaic semiconductors with non-tetrahedral coordination, such as the perovskite semiconductors.
The electronic structure analysis is helpful for understanding the electrical conductivity of the I2–II–IV–VI4 semiconductors and why the band gaps and band edge positions change as the component elements change. The chemical trends in the band gaps of these semiconductors have been summarized in Ref. [37]. Figure
![]() | Fig. 4. (color online) The calculated band structure, total and partial density of states, and the schematic plot of the band component for kesterite Cu2ZnSnS4. Adapted from Ref. [24]. |
Compared to ZnS, the Cu-based chalcogenides, such as Cu2ZnGeS4, have much smaller band gaps, because the valence band edge is much higher due to the shallower Cu 3d levels, which can be seen clearly in the calculated band alignment in Fig.
![]() | Fig. 5. (color online) (a) The calculated band alignment of ZnS, CuGaS2, and Cu2ZnGeS4. Adapted from Ref. [9]. (b) The calculated band alignment of binary CdS, ternary CuInSe2 and CuGaSe2, and quaternary Cu2ZnSnS4 and Cu2ZnSnSe4. The red dashed lines near the conduction band show the n-type doping limit. Adapted from Ref. [42]. |
Similarly, the valence band edge of Cu2ZnSnS4 is about 1 eV higher than that of Cu-free CdS, as shown in Fig.
The conduction band edge of Cu2ZnSnSe4 is 0.25 eV lower than that of CdS, giving a type-I band alignment between Cu2ZnSnSe4 and CdS. The type-I and type-II band alignment difference of Cu2ZnSnS4 and Cu2ZnSnSe4 relative to CdS has been discussed in Ref. [42]. The experimental measurement of the valence band offset between Cu2ZnSnSe4 and CdS is around 0.9–1.2 eV.[42–48] The 0.3 eV difference may result from the different interfaces, elemental composition ratios, and measurement methods used by different groups. If the valence band offset is only 0.9 eV, the band alignment between Cu2ZnSnS4 and CdS becomes type-I. The accurate measurement of the band offset and the influence of the different interfaces, elemental composition ratios, and measurement methods on the results are still interesting topics deserving further study.
Different structures (kesterite, stannite, and PMCA) have the valence band edges close to each other, while the conduction band edge shifts down from kesterite to stannite and PMCA, leading to smaller band gaps of stannite and PMCA. Figure
![]() | Fig. 6. (color online) The GGA calculated band gaps of CuZn–IV–S4. Compared to the experimental values, the GGA calculated values have a general underestimation by about 1.4 eV for these semiconductors. Adapted from Ref. [10]. |
Liu et al.[49] have studied the electron and hole effective masses of KS and ST structured Cu2Zn–IV–VI4. They found that the electron effective masses are rather small and almost isotropic while the hole effective masses show strong anisotropies, and the effective masses of the Se-based compounds are much smaller than those of the S-based compounds. Table
![]() | Table 1.
The |
The different hole effective masses for the kesterite and stannite structures result from the opposite crystal filed splitting,[9,50] and can be understood by their partial charge densities shown in Fig.
![]() | Fig. 7. (color online) Partial charge density of the topmost 3 VBs at the Γ point in (a) the Cu–S–Zn plane for the kesterite Cu2ZnGeS4 and (b) the Cu–S plane for the stannite Cu2ZnGeS4. Adapted from Ref. [49]. |
Since the band gaps of Cu2ZnSnS4 and Cu2ZnSnSe4 are 1.5 eV and 1.0 eV, respectively, there is a 0.5 eV space for the band gap tuning through forming the S-Se alloys. It is interesting to see that the solar cells based on the mixed-anion Cu2ZnSn(S,Se)4 (CZTSSe) alloys have higher efficiency than those based on the pure Cu2ZnSnS4 (CZTS) and Cu2ZnSnSe4 (CZTSe) compounds,[2,51] although CZTS has an optimal band gap for single-junction solar cells according to the Shockley–Queisser model.[52] The physical properties of various alloys based on CZTS have been studied by several groups,[53–56] which are very important for understanding the better performance of the CZTS-based alloys. In order to simulate the random occupation of S and Se on the anion sites of the CZTSSe alloys, we used the special quasi-random structure (SQS) method[50,57] to describe the geometrical structure of the CZTSSe alloys. To know the miscibility of the alloys, the alloy formation enthalpy is defined as
At the same time, for most alloys, the alloy formation enthalpy obeys a quadratic function of the composition x
The bandgap dependence of the conventional semiconductor alloys on the composition x follows the function
From Fig.
![]() | Fig. 8. (color online) The calculated enthalpy of formation for Cu2ZnSn(S1 −xSex)4 alloys as a function of composition x. Adapted from Ref. [42]. |
In Fig.
![]() | Fig. 9. (color online) The calculated bandgaps of the Cu2ZnSn(S1−xSex)4 alloys using the (a) GGA functional and (b) HSE functional, respectively. Adapted from Ref. [42]. |
Besides the mix-anion alloys, several theoretical groups also studied the mixed-cation alloys, including (Cu,Li)2ZnSnS4,[61] Cu2(Zn,Fe)SnS4(CZFSS),[62] Cu2(Zn,Cd)SnS4,[63] Cu2Zn(Sn,Si)S4,[64] Cu2Zn(Sn,Ge)Se4,[64] and (Cu,Ag)2ZnSn(S,Se)4.[60] Shu et al. studied the properties of the Cu2Zn(Sn,Si)S4 and Cu2Zn(Sn,Ge)S4 alloys.[64] They found that the Cu2ZnSn1−xGexSe4 alloys are highly miscible and their band gap increases from 1.0 eV to 1.5 eV with composition x increasing from 0 to 1, while the Cu2ZnSn1 −xSixSe4 alloys are much less miscible and span a bandgap range from 1.0 eV to 2.4 eV. The bowing parameter of the Cu2Zn(Sn,Ge)S4 alloy is much smaller than that of Cu2Zn(Sn,Si)S4, because the size and chemical differences between Sn and Si are larger. Shen et al. investigated the electronic and optical properties of Cu2ZnGe(SexS1 −x)4 alloy and found that the gap size may range from 1.52 eV to 2.25 eV, which will provide potential applications as solid state lighting[65] and water splitting photocatalyst. Shibuya et al. found that Fe can be easily mixed into CZTS.[62] It is interesting to see that the structure of Cu2Zn1−xFexSnS4 alloys changes from kesterite to stannite at x = 0.4, which may result in a small discontinuity of the bandgap. The CZFTS alloys with high Fe concentrations may be used in the Si-based tandem solar cells. Other groups[61,63] tuned the bandgap from 1.55 eV to 1.09 eV through Cu2(Zn,Cd)SnS4, and from 1.5 eV to 1.9 eV through (Cu,Li)2ZnSnS4. These alloys are also good candidates for tandem solar cells.
Not only the component-uniform mixed-anion and mixed-cation alloys, but also the component-graded (Cu,Ag)2ZnSn(S,Se)4 have been proposed to improve the photovoltaic performance of kesterite solar cells. The efficiency increase of kesterite solar cells stopped after the latest record efficiency was reported in 2013, and it is believed that there is a bottleneck to the further improvement of the open-circuit voltage (
To combine the advantages of Ag2ZnSnS4 and Cu2ZnSnS4, they proposed that the component-graded (Cu,Ag)2ZnSnS4 and (Cu,Ag)2ZnSnSe4 alloys can be ideal absorber layer materials if the Ag component is low in the interior of the absorber layer (thus less recombination-center defects) but high near the p–n junction interface (thus less
The beneficial effect of alkali-metal dopants on Cu(In,Ga)Se2 thin film solar cells has been investigated for several decades. The alkali-metal can be doped into the lattice of Cu(In,Ga)Se2, compensating the donors or promoting the acceptor carrier concentration in the absorber layer,[68–71] and can also increase the grain size and passivate the grain boundaries (GB) which act as the recombination centers in the thin films.[68]
Inspired by the prominent effect of alkali-doped Cu(In,Ga)Se2 devices, it is expected that the alkali metals will also have beneficial effects on CZTS and CZTSe based solar cells because of the similar crystal and device structures. Using the first-principles calculations, Maeda et al. studied the doping effect of Li, Na, and K in CZTS and CZTSe.[72] Their results show that alkali metal is prone to occupy the Zn site first, then the Cu site, followed by the Sn site and the interstitial site. The substitution defect of alkali metal on the Zn site will be the primary p-type defect in CZTS and CZTSe. However, in the study of Xiao et al.,[73] NaCu has the lowest formation energy, but NaCu is an isovalent substitution which cannot produce carriers and increase the hole concentration. Whereas, the formation energy of NaZn is a little higher than that of NaCu, which will contribute to the electrical conductivity of the CZTS absorber layer. The difference of the two studies results from the different chemical potentials considered in their calculations. In 2016, Yuan et al. proposed a new picture on why the isovalent NaCu substitution can increase the hole carrier concentration based on the Na diffusion in the lattice,[70] which opens a new view on the role of sodium in promoting the hole concentration from the electronic point of view.
Meanwhile, calculations on the transition energy levels of Na dopants on different sites revealed that NaZn is a shallow acceptor. Although NaSn introduces two transition energy levels, one of which is close to the middle of the band gap, the formation energy of NaSn is so high that its detrimental effect as recombination center can be ignored.
Experimentally, Heish et al. investigated the efficiency enhancement via alkali metal doping on CZTSSe. They found that small alkali metals are suitable for increasing the carrier concentration because they are easy to substitute the lattice atom in CZTSSe, while the larger alkali metals are favorable for increasing the grain size because of the low melting point of binary selenides, which will suppress the non-radiative recombination caused by the GBs, since less GBs are produced in the absorber layer.[74]
The benign effect of alkali metal on the increase of the carrier concentration in CZTSSe has also been proved by several previous studies,[75–77] where one order of increment of the carrier concentration was obtained by doping sodium into CZTS.[75] It seems that the alkali metal has the similarly beneficial effect as that in CIS. The enhancement of the mobility of the hole carriers was also observed by Nagaoka et al.[77] Moreover, Zhao et al. calculated the effective mass of holes in the CZTS with Na impurity, and found that effective mass of hole in Na-doped CZTS is lighter, implying that Na benefits the hole mobility in CZTS.
Based on the observation of Prabhakar et al.,[75] Na impurity in CZTS enhances the (112) texture of the film. Also, potassium was shown to enhance the (112) preferred orientation,[78] which indicates that the alkali metal improves the crystallinity of the film. The improvement of the grain size was also observed after introducing Li,[74,79] Na,[74,75,77,79,80] K,[74,80] Rb,[74,79] and Cs.[74] It is worth noting that the grain size reaches its maximum at 1% K doping and decreases as the K doping concentration continues rising as reported in the study of Tong et al.[78] Based on the aforementioned studies, the effect of the alkali metal on the grain size cannot be drew out, but a detailed investigation of the sodium concentration in CZTS grains has pointed out that the electronic passivation of the GB requires less sodium than that required for producing large grains.[81] This means that the size of the grain becomes not critical to the solar cell performance as long as the GB is electronically passivated.
Based on the discussions above, it is clear that alkali metal also plays a beneficial role in CZTS, but it is still not clear whether the influence mechanism in CZTS is similar to that in Cu(In,Ga)Se2. Further investigation on the effect of alkali metal in CZTS and CZTSe is necessary.
Besides the point defects, the grain boundaries of the CZTS and CZTSe thin films are also important for their photovoltaic performance. Usually the polycrystalline semiconductors exhibit a poorer performance in comparison with their monocrystalline counterparts, because there are usually dangling bonds, wrong bonds, and wrong bonding angles on the grain boundaries, which may introduce deep levels in the band gap and act as non-radiative recombination centers in the system. Fig.
![]() | Fig. 10. (color online) Schematic plot of the p–n junction interface between the n-type CdS buffer layer and p-type absorber layer: Cu2ZnSn(S,Se)4 with high concentration of CuZn antisite defects limiting ![]() ![]() |
The lack of experimental observations on the structures of grain boundaries in CZTS caused the difficulty in obtaining the exact atomic structure of the grain boundaries, however, the theoretical model of the grain boundaries in CZTS can be adopted from those in CdTe which had been observed via the high-resolution transmission electron microscopy (HRTEM).[82] Considering the ordered substitution of Cu, Zn, and Sn for the cation ions, CZTS has 4 different grain boundaries as shown in Fig.
![]() | Fig. 11. (color online) The 4 relaxed atomic structures of Cu2ZnSnSe4 GBs. The grain boundary (114) plane is indicated by the dashed line. Adapted from Ref. [83]. |
Several groups have investigated the feature of these GBs in CZTS. Li et al. calculated the density of states of the GB structures in Fig.
Experimentally, Kim et al. found that the highest efficiency samples of CZTS exhibit downward potential bending at GB area and upward bending at the inter-grains, while the poorer efficiency samples exhibit an opposite behavior,[87] which shows that the quality of the GB strongly affects the property of the CZTS devices.
Since the intrinsic GB of CZTS is detrimental, an important issue of the GB in CZTS is whether it can be passivated or not. Using DFT calculation, Yin et al. found the segregation of ZnSn, OSe, and Nai has the beneficial effect of eliminating the deep defect states in the band gap and creating a hole barrier at GB.[84] Liu et al. found that Na not only can passivate the deep level in the band gap, but also prefers to segregate at the GBs of CZTS because it has negative formation energies in the system.[88] These calculation studies pointed out the potential passivation method that can be applied on CZTS to ensure the improvement of GB in CZTS.
The effect of sodium in the GB of CZTS was studied by Gerson experimentally, who demonstrated that the CZTS samples without sodium treatment contain non-radiative recombination center defects, and these deep-level defects can be effectively passivated by the addition of sodium.[81] Sodium prefers to locate on the surface and GBs of CZTS and it can help to produce large grains, and a small amount of sodium is sufficient to passivate GBs effectively.
The number of component elements increased steadily in the 60-year development of photovoltaic semiconductors, i.e., from silicon in 1950s, to GaAs and CdTe in 1960s, CuInSe2 in 1970s, Cu(In,Ga)Se2 in 1980s, Cu2ZnSnS4 in 1990s, and more recently Cu2ZnSn(S,Se)4 and CH3NH3PbI3. The increased number of elements in Cu2ZnSnS4 and Cu2ZnSnSe4 and their alloys Cu2ZnSn(S,Se)4 relative to binary II–VI or III–V semiconductors results in the increased structural and chemical freedom for tuning the material properties. We reviewed how these group I2–II–IV–VI4 semiconductors can be derived from the binary II–VI semiconductors through element-mutation, i.e., from II–VI (ZnS) to I–III–VI2 (CuGaS2) and then to I2–II–IV–VI4 (Cu2ZnSnS4). Following the mutation, we can determine the crystal structures of the I2–II–IV–VI4 semiconductors and reveal the chemical trends in their electronic band structure, which makes the band structure engineering possible. The band gaps of I2–II–IV–VI4 compounds can be tuned from negative (metal or topological insulator[89]) to more than 4 eV (wide-gap semiconductor). Furthermore, the increased structural freedom enhances the miscibility (component-uniformity) of the Cu2ZnSn(S,Se)4 and Cu2Zn(Sn,Ge)Se4 alloys, so their band gaps can be tuned continuously and also linearly as a function of the alloy composition.
The easy formation of CuZn antisite defect and related defect complexes was believed to be one critical factor limiting the open-circuit voltage and efficiency of kesterite solar cells. In order to suppress the formation of these defects, the mixed-cation (Ag,Cu)2ZnSn(S,Se)4 and Cu2(Zn,Cd)Sn(S,Se)4 alloys were proposed as alternative light-absorber layers and are now under intensive study. The recent experimental breakthrough on the component-graded (Ag,Cu)2ZnSn(S,Se)4 alloy solar cells may point out a possible direction or method for breaking the 12.6% efficiency record.
The increased structural and chemical freedom also causes the dramatic increase of possible point defects in the lattice of Cu2ZnSn(S,Se)4 thin films, which can significantly influence the optical and electrical properties and thus the photovoltaic performance. A series of point defects have been predicted since 2010.[24,90–92] After that various defect levels have been experimentally observed using different characterization techniques, and compared with the calculated defect levels. Although the ionization levels of the dominant defects such as CuZn and VCu have been experimentally well confirmed, other defect levels have not well determined. A careful and systematical characterization of the defect levels in Cu2ZnSnS4 and Cu2ZnSnSe4 and their alloys Cu2ZnSn(S,Se)4 with different chemical composition ratios and different growth techniques will be very useful for the knowledge-based defect optimization. Although the defect properties are not reviewed here, the importance of the defect optimization cannot be neglected.
Besides the intrinsic defects, extrinsic elements may be doped into these quaternary semiconductors intentionally or unintentionally. Whether the real thin films can work as ideal light-absorber layers depends also on the behavior of the possible extrinsic dopants. Here we reviewed the study on the alkaline dopants which may exist in both the grain interiors and on the grain boundaries. Many opinions that were proposed for understanding the alkaline dopants and grain boundaries in Cu(In,Ga)Se2 solar cells have been borrowed in the study of the alkaline dopants and grain boundaries in Cu2ZnSnS4, Cu2ZnSnSe4, and Cu2ZnSn(S,Se)4. However, obvious differences in these properties have been found between Cu(In,Ga)Se2 and Cu2ZnSn(S,Se)4. More careful, accurate, and systematical studies on the alkaline dopants and grain boundaries in the kesterite solar cells are still necessary and fundamental to the search for new strategies for the performance optimization.
The interfaces between the kesterite Cu2ZnSnS4, Cu2ZnSnSe4, and Cu2ZnSn(S,Se)4 and CdS are currently not well-studied, because the microstructure near the interfaces can be very complicated and also depend on the specific fabrication methods and conditions. It is not clear whether the interfaces between the kesterite absorber layer and CdS are similar to those between the chalcopyrite absorber layer and CdS, so it is still an open question whether the kesterite solar cells should inherit the device structure from the chalcopyrite thin film solar cells. Systematical simulation and characterization of these interfaces are of fundamental importance to the future development of high-efficiency kesterite solar cells.
[1] | |
[2] | |
[3] | |
[4] | |
[5] | |
[6] | |
[7] | |
[8] | |
[9] | |
[10] | |
[11] | |
[12] | |
[13] | |
[14] | |
[15] | |
[16] | |
[17] | |
[18] | |
[19] | |
[20] | |
[21] | |
[22] | |
[23] | |
[24] | |
[25] | |
[26] | |
[27] | |
[28] | |
[29] | |
[30] | |
[31] | |
[32] | |
[33] | |
[34] | |
[35] | |
[36] | |
[37] | |
[38] | |
[39] | |
[40] | |
[41] | |
[42] | |
[43] | |
[44] | |
[45] | |
[46] | |
[47] | |
[48] | |
[49] | |
[50] | |
[51] | |
[52] | |
[53] | |
[54] | |
[55] | |
[56] | |
[57] | |
[58] | |
[59] | |
[60] | |
[61] | |
[62] | |
[63] | |
[64] | |
[65] | |
[66] | |
[67] | |
[68] | |
[69] | |
[70] | |
[71] | |
[72] | |
[73] | |
[74] | |
[75] | |
[76] | |
[77] | |
[78] | |
[79] | |
[80] | |
[81] | |
[82] | |
[83] | |
[84] | |
[85] | |
[86] | |
[87] | |
[88] | |
[89] | |
[90] | |
[91] | |
[92] |