Theoretical insight into the optoelectronic properties of lead-free perovskite derivatives of Cs₃Sb₂X₉ (X = Cl, Br, I)

Abstract

The lead-free perovskites derivatives of Cs₃Sb₂X₉ (X = Cl, Br, I) have been synthesized, but their photocatalytic properties are not explored. To evaluate the feasibility for the visible light catalytic performance, we calculate the structural, electronic, optical and charge transfer properties of Cs₃Sb₂X₉, based on the hybrid density functional theory of HSE06 with the projector augmented wave potential. The results show the decrease of band energy gaps and the redshift of absorption edges from X = Cl to I. The absolute potential of the valence band maximum and conduction band minimum is determined to justify the feasibility of the photocatalytic water splitting or CO₂ reduction. The calculated carrier mobilities reveal that the high electron mobilities of Cs₃Sb₂I₉ are beneficial to the reducing powers for hydrogen generation and CO₂ reduction. The present results indicate that Cs₃Sb₂I₉ is appropriate for the photocatalytic water splitting to produce hydrogen or the CO₂ reduction driven by the visible light.

Introduction

As one of the most abundant energy sources, solar energy is inexhaustible. Photocatalytic water splitting to generate hydrogen and photocatalytic reduction of CO₂ into fuels driven by sunlight are a kind of artificial photosynthesis processes that only require sunlight, water and a photocatalyst for realizing the conversion of solar energy into chemical energy. It is a sustainable strategy to apply solar energy and to address current energy crisis as well as to alleviate the environmental issues caused by CO₂ emissions, such as “the greenhouse effect.” Therefore, it is indispensable to explore and develop the efficient photocatalysts with excellent optoelectronic properties such as high solar-light absorption and superior carrier mobilities to enhance the solar energy photochemical conversion efficiency.

Over the last few years, hybrid organic–inorganic lead halide perovskites have been extensively employed in optoelectronic applications, including photodetectors [1, 2], light-emitting diodes [3, 4], which can be ascribed to their excellent optoelectronic properties. Recently, some studies about the lead halide perovskites for photocatalytic applications such as the degradation of organic dyes, CO₂ reduction and hydrogen production have been reported [5,6,7,8,9,10]. However, the instability of lead-based perovskite and the toxicity of heavy metal lead restrict the maximization of commercial applications of lead halide perovskites. The first-principle calculations for CsPbX₃ (X = Cl, Br, I) [11,12,13,14] confirm that the prominent optoelectronic properties of the Pb-based perovskites are associated with the outermost ns² electronic configuration of Pb²⁺. The hybridization of s orbitals with p orbitals between lead and halogen is responsible for the antibonding character at the valence band maximum (VBM) as shown in Fig. 1. Zakutayev et al. [15] have reported that the presence of antibonding orbitals usually makes the defect levels induced by any kind of intrinsic defects lie within valence band as resonances and does not bring about a deep trap state in the forbidden band. Like Pb²⁺, the outermost electronic configuration of Sb³⁺ is also ns², which is a stable valence electronic configuration. Thus, perovskites based on Sb³⁺ are expected to be promising materials which not only possess the excellent optoelectronic properties like those of lead halide perovskites but are less toxic than Pb [16]. Recently, some investigations about antimony halide complexes [17,18,19] including Sb halide complexes applicable to photodevices have been reported [20,21,22]. As a member of the Sb-based perovskites family, the cesium antimony halide perovskites with the formula of Cs₃Sb₂X₉ (X = Cl, Br, I) have attracted much interest due to their good stability. In 2000, Zemnukhova et al. [23] have reported that the Cs₃Sb₂X₉ (X = Cl, Br, I) perovskites can be prepared from CsX and SbX₃ aqueous solutions. Subsequently, more detailed studies for Cs₃Sb₂X₉ (X = Cl, Br, I) appeared in recent years. Remarkably, Saparov et al. [24] have reported that the layered structure of Cs₃Sb₂I₉ perovskite can be prepared by a two-step deposition approach and the density functional theory (DFT) calculations show that the optical absorption of Cs₃Sb₂I₉ is comparable to that of CH₃NH₃PbI₃. Moreover, the single crystal and quantum dots (QDs) for Cs₃Sb₂Br₉ have been synthesized by Song et al. [25]. Recently, Kuang et al. [26] have reported that their synthesized novel lead-free all-inorganic Cs₂AgBiBr₆ double perovskite nanocrystal shows an excellent performance for CO₂ reduction, implying the possibility of lead-free halide perovskites for photocatalytic applications. With these findings in mind, we explore the photocatalytic performance of Cs₃Sb₂X₉.

Figure 1

Schematic diagram of bonding (π_pp, σ_pp) and antibonding (π_pp*, σ_pp*) orbitals of CsPbI₃ showing the formation of the valence and conduction band. The π ^nb_pp represents the non-bonding orbital

In the present work, we systematically investigate the photocatalytic properties of Cs₃Sb₂X₉ via calculating their electronic and optical properties by the first-principles hybrid DFT. The carrier mobility is calculated to examine the migration and separation of photogenerated electrons and holes. Our results indicate that Cs₃Sb₂X₉ has obvious absorptions in a wide range from near ultraviolet to visible light. Especially, Cs₃Sb₂I₉ exhibits a large absorption coefficient in the high-energy region of the visible light. The band edges alignments have been calculated to demonstrate that the positions of the band edges for Cs₃Sb₂X₉ straddle the corresponding absolute potentials of water splitting or CO₂ reduction. The present findings are expected to provide insights into the photocatalytic properties of Cs₃Sb₂X₉ and a helpful reference to develop the efficient photocatalytic materials.

Computational details

The cesium antimony halide perovskites derivatives with a formula of A₃B₂X₉ are derived from the traditional ABX₃-type perovskites with two-thirds of occupancies of B site. As shown in Fig. 2a, b, Song et al. [25] have reported that the crystal structure of Cs₃Sb₂Br₃ (P$\overline{\text{3}}$m1, no. 164) can be constructed by removing every third Sb layer along <111> in CsSbBr₃ perovskites to achieve correct charge balance. The crystal lattices of Cs₃Sb₂X₉ are demonstrated in Fig. 2c. All the structures have been fully optimized by using a conjugate gradient algorithm with the exchange and correlation potentials of Perdew–Burke–Ernzerhof (PBE) [27] parameterization under the generalized gradient approximation (GGA). The convergence criteria of the total energy of the system and the forces on each atom are set to 10⁻⁶ eV and 0.01 eV/Å, respectively. The 5 × 5 × 5 Monkhorst–Pack k-point meshes are used for sampling the Brillouin zone, and the kinetic energy cutoff is set to 500 eV after a careful convergent test (Table 1). Since the PBE functional usually underestimates the band energy gap and the improvement effect of the Heyd–Scuseria–Ernzerhof (HSE06) hybrid functional [28] has been confirmed by Le Bahers and Sautet et al. [29,30,31], we employ the HSE06 hybrid functional to calculate the band energy structures and optical absorptions of Cs₃Sb₂X₉. All the calculations are carried out with the Vienna ab initio simulation package (VASP 5.4.4) [32, 33]. The core-valence interactions are described by the projected augmented wave (PAW) pseudopotentials method [34]. The choice of valence electrons for Cl, Br, I, Sb, Cs and Pb atoms is Cl (3s²3p⁵), Br (4s²4p⁵), I (5s²5p⁵), Sb (5s²5p³), Cs (5s²5p⁶6s¹) and Pb (6s²6p²), respectively.

Figure 2

Removal of every third Sb layer along the <111> direction of a the perovskite structure results in b the 2D layered modification of Cs₃Sb₂Br₉. The purple, gray and red spheres represent Cs, Sb or Br elements, respectively. c The optimized structure of Cs₃Sb₂Br₉. d The selected high-symmetry lines in the Brillouin zone are Γ–M–K–Γ–A–L–H–A

Table 1 Calculated lattice parameter a, b, c (Å), bond length of X-Sb (X = Cl, Br, I) (Å), band energy E_g (eV) of, Cs₃Sb₂X₉ (Cl, Br, I)

Results and discussion

Geometrical and band energy structures

The calculated lattice parameters, bond lengths of Cs₃Sb₂X₉, including the available experimental and the previous theoretical values, are presented in Table 1. From the table, we can see that the errors between the present lattice parameters of Cs₃Sb₂X₉ with their experimental values [24, 25, 35] are less than 3% and the results for Cs₃Sb₂I₉ are also consistent with the previous theoretical values [24], indicating that the present theoretical level is credible. The lattice parameters and the length of Sb–X bonds increase from Cs₃Sb₂Cl₉ to Cs₃Sb₂I₉, which can be understood from the fact that the effective radius of the X atom increases from Cl to I.

The calculated energy band structures of Cs₃Sb₂X₉ with PBE and HSE06 functional along with the high-symmetry k points in the Brillouin zone are shown in Fig. 3. One can observe from the figure that Cs₃Sb₂Br₉ and Cs₃Sb₂I₉ exhibit a direct band energy gap with the VBM and the conduction band minimum (CBM) located at Г point. However, Cs₃Sb₂Cl₉ has an indirect band energy gap, because the VBM is located at Г point but the CBM at A point (Fig. 3c). Although the PBE functional may underestimate the band energy gap of the semiconductor compared with the HSE06 hybrid functional, the contours of the energy levels remain a similar characteristic for the two calculational methods [36, 37]. The calculated band energy gaps for Cs₃Sb₂X₉ by HSE06 functional are listed in Table 1. The present 3.11, 2.60 and 2.04 eV for the band energy gaps of Cs₃Sb₂X₉ are in good agreement with the available experimental data [24, 25, 38] and the other theoretical values in the literature [24]. The total density of states (TDOS) and the partial density of states (PDOS) of Cs₃Sb₂X₉ can be used to understand the orbital characteristics of VBM and CBM. For Cs₃Sb₂Cl₉, as shown in Fig. 4a, the first interaction is the strong p–p interactions resulting from the overlap of Cl(3p)–Sb(5p) throughout most of the conduction band and the low energy part of the valence band, which leads to the antibonding characteristic of the CBM and the bonding one at the low energy zone of the valence band. The other one is the typical s–p interaction from the hybridization of Sb(5s) atomic orbitals and Cl(3p) atomic orbitals at the valence band, which is responsible for the presence of antibonding orbitals at VBM. Moreover, similar interactions can be found for Cs₃Sb₂Br₉ and Cs₃Sb₂I₉ but not are described here for the sake of brevity.

Figure 3

Band structures of a Cs₃Sb₂Cl₉, b Cs₃Sb₂Br₉, c Cs₃Sb₂I₉ calculated by PBE and HSE06 functional. The red solid lines and blue solid lines represent the PBE and the HSE06 results, respectively. The energy zero represents the Fermi level

Figure 4

Total and partial density of states are calculated by HSE06 functional. a Cs₃Sb₂Cl₉; b Cs₃Sb₂Br₉; c Cs₃Sb₂I₉

Absolute energy-level positions and feasibility for photocatalytic water splitting and CO₂ reduction

In a photocatalytic process, the semiconductor absorbs the light, transfers the electron at valence band to the conduction band and produces the photogenerated electron–hole pairs. Then, the photogenerated electrons and holes transfer to the corresponding reaction sites for the oxidation or reduction reactions. Accordingly, the absolute energy-level positions of CBM and VBM must satisfy the requirement of the photocatalytic reactions which can be described by two half reactions. Specifically, for the hydrogen evolution or CO₂ photoreduction to CH₄, the potential is determined according to the following half-reactions, respectively:

$$ 2{\text{H}}^{ + } + 2 {\text{e}}^{ - } \to {\text{H}}_{ 2} $$

(1)

$$ {\text{CO}}_{ 2} + 8{\text{e}}^{ - } + 8{\text{H}}^{ + } \to {\text{CH}}_{ 4} + 2{\text{H}}_{ 2} {\text{O}} $$

(2)

while the half-reaction steps for the water oxidation of both reactions can be summarized as follows:

$$ 2{\text{H}}_{ 2} {\text{O}} + 4 {\text{h}}^{ + } \to {\text{O}}_{ 2} + 4{\text{H}}^{ + } $$

(3)

Therefore, the CBM for the water-splitting reaction should be higher than the reduction potential of H⁺/H₂ ($E_{{{\text{H}}^{ + } /{\text{H}}_{2} }}$ = − 4.44 eV at pH = 0.0). Similarly, CBM for the photoreduction of CO₂ should be more negative than the potentials for CO₂/CH₄ ($E_{{{\text{CO}}_{2} /{\text{CH}}_{4} }}$ = − 4.62 eV at pH = 0.0). On the other hand, the VBM should be lower than the oxidation potential of O₂/H₂O ($E_{{{\text{O}}_{2} /{\text{H}}_{2} {\text{O}}}}$ = − 5.67 eV at pH = 0.0). Obviously, it is crucial to calculate precisely the edge position of materials to evaluate the feasibility of photocatalytic water splitting and CO₂ reduction; especially, the surface chemistry and interfacial effects are considered in the process. Galli et al. [39] have summarized recent progress and open theoretical challenges present in simulations of PEC interfaces. They provide a good policy to treat the effect of the interface on the band position of the surface structure, in which the aqueous solution model has been considered. However, it is still a challenging task to establish a theoretical approach for predicting the absolute energy positions of CBM and VBM for the bulk structure. Here, we calculate the absolute band edges of Cs₃Sb₂X₉ based on a reliable approximate method supposed by Xu et al. [40] for a bulk semiconductor by the following equations:

$$ E_{{\text{VBM}}} = \chi - \frac{1}{2}E_{\text{g}} $$

(4)

$$ E_{\text{CBM}} = E_{\text{VBM}} + E_{\text{g}} $$

(5)

where E_VBM and E_CBM represent the absolute potentials of VBM and CBM. The $\chi$ is the electronegativity of the semiconductor which can be determined by the absolute electronegativities of the constituent atoms as:

$$ \chi ({\text{compound}}) = \chi_{1}^{a} \chi_{2}^{b} \ldots \chi_{n}^{c} $$

(6)

where $\chi_{1}$,$\chi_{2}$ and $\chi_{\text{n}}$ represent the electronegativities of the atoms in the compound, while a, b and c are the molar fractions of the atoms. Here, the electronegativity of atoms are evaluated by the Mulliken electronegativity as follows [41]:

$$ \chi = {{(I + A)} \mathord{\left/ {\vphantom {{(I + A)} 2}} \right. \kern-0pt} 2} $$

(7)

where I is the ionization energy and A is the electron affinity of the atom. The electronegativities of Cs, Sb, Cl, Br and I can be obtained using Eq. (7). Finally, the calculated values of E_VBM and E_CBM for Cs₃Sb₂X₉ at pH = 0 based on Formulas (4) and (5) are given in Table 2. The results show that the values of E_VBM for Cs₃Sb₂Cl₉, Cs₃Sb₂Br₉ and Cs₃Sb₂I₉ are − 7.15 eV, − 6.58 eV and − 5.92 eV, which are 1.48 eV, 0.91 eV and 0.25 eV lower than the water oxidation potential, respectively, while E_CBMs are about 0.40 eV, 0.46 eV and 0.56 eV higher than the hydrogen reduction potential, respectively. Therefore, Cs₃Sb₂X₉ satisfies the requirement of the absolute potentials of VBM and CBM for water-splitting reaction. Figure 5 visually expresses the absolute energy positions of band edges and the redox potentials of the water-splitting reaction and CO₂ reduction. The figure demonstrates that the absolute potential of CBM for Cs₃Sb₂I₉ is the most negative one among the three structures, which suggests that it is more efficient for hydrogen reduction. Moreover, all the positions of CBM for Cs₃Sb₂X₉ are higher than the reduction potential for CO₂ to CH₄ and their VBMs are located at a more positive position than the water oxidation potential, implying that the E_VBM and E_CBM satisfy the requirement of the CO₂ reduction reaction.

Table 2 Calculated E_VBM (eV) and E_CBM (eV) for Cs₃Sb₂X₉(Cl, Br, I)

Figure 5

VBM and CBM positions of Cs₃Sb₂X₉(Cl, Br, I) with respect to water and CO₂ redox potential

Optical absorption properties

The optical absorption coefficient α(ω) can be used to evaluate the response ability of photocatalyst. A large absorption coefficient means more photons can be captured to impact on the photocatalytic performance. The α(ω) can be obtained from the real parts ε₁(ω) and imaginary parts ε₂(ω) of the dielectric function via the following equation [42]:

$$ \alpha \left( \omega \right) = \sqrt 2 \omega \sqrt {\sqrt {\varepsilon_{1}^{2} \left( \omega \right) + \varepsilon_{2}^{2} \left( \omega \right)} - \varepsilon_{1}^{{}} \left( \omega \right)} $$

(8)

Therefore, we first calculated the dielectric function of Cs₃Sb₂X₉ by employing HSE06 functional. The details of the calculation of the dielectric function can be found from the previous literature [43] and omitted here for brevity. The calculated ε₂(ω) of the dielectric function for Cs₃Sb₂X₉ is shown in Fig. 6a. The characteristics of ε₁(ω) and ε₂(ω) can be understood from the electronic properties. The contours of ε₂(ω) are well consistent with those of the total DOS in Fig. 4a. For example, the peak of ε₂(ω) for Cs₃Sb₂I₉ appears at around 2.8 eV corresponding to the electronic transition from VBM to the sharper peak of DOS at around 2.8 eV.

Figure 6

a Imaginary part of the dielectric function and b absorption coefficient of Cs₃Sb₂X₉(Cl, Br, I)

The calculated optical absorption coefficients for the three structures are presented in Fig. 6b. The optical absorption edge of Cs₃Sb₂Cl₉ occurred near the UV light region corresponding to its band energy gaps (E_gs) of 3.1 eV, while that of the other two structures shows a continuous redshift. For Cs₃Sb₂Br₉, a large absorption coefficient of about 40000 cm⁻¹ has been identified in the high-energy region of the visible light. Notably, Cs₃Sb₂I₉ exhibits a more obvious and wider absorption around the peak at about 3.0 eV in the visible light range, which mainly comes from the transition from I 5p states to Sb 5p states. The larger absorption coefficient suggests that more photons could be captured in the corresponding light range, which is a benefit to generate electron–hole pairs. In the sense, Cs₃Sb₂I₉ is the most efficient one of the three structures to respond to the visible light in the photocatalysis.

Migration and separation of photogenerated electron–hole pairs

The charge spatial separation and the migration rate of the photogenerated charge pair play a key role in the practical photocatalysis. The carrier mobility can give a quantitative description of the transferring ability of the photoinduced charges. The higher carrier mobility implies that charge transfer is more efficient. Moreover, the larger mobility differences between the electrons and the holes are more significant for separation of the photogenerated electron–hole pairs.

The carrier mobility can be calculated via the deformation potential (DP) theory by the following formula [44,45,46]:

$$ \mu = \frac{{2\sqrt {2\pi } e\hbar^{4} {B}}}{{3m_{I}^{*} (m_{b}^{*} k_{B} T)^{{\frac{3}{2}}} {E}^{2} }} $$

(9)

where e, ℏ, k_B and T represent the electron charge, reduced Planck constant, Boltzmann constant and temperature, respectively. B is the bulk modulus defined as ${B} = V_{0} \left( {{{\partial^{2} {E}_{\text{tot}} } \mathord{\left/ {\vphantom {{\partial^{2} {E}_{\text{tot}} } {\partial V^{2} }}} \right. \kern-0pt} {\partial V^{2} }}} \right)$, where E_tot is the total energies of the systems and V₀ is the equilibrium volume of the system. E is the deformation potential constant determined by $E = {{\left( {\Delta E_{\text{edge}} } \right)} \mathord{\left/ {\vphantom {{\left( {\Delta E_{\text{edge}} } \right)} {({{\Delta V} \mathord{\left/ {\vphantom {{\Delta V} {V_{0} }}} \right. \kern-0pt} {V_{0} }})}}} \right. \kern-0pt} {({{\Delta V} \mathord{\left/ {\vphantom {{\Delta V} {V_{0} }}} \right. \kern-0pt} {V_{0} }})}}$, where $\Delta V$ is the variation of volume caused by changing lattice parameters and E_edge is the corresponding change of the energy for CBM or VBM. The conductivity effective mass $m_{I}^{ * } = 3\text{(}{1 \mathord{\left/ {\vphantom {1 {m_{x}^{ * } }}} \right. \kern-0pt} {m_{x}^{ * } }} + {{{1 \mathord{\left/ {\vphantom {1 {m_{y}^{ * } + 1}}} \right. \kern-0pt} {m_{y}^{ * } + 1}}} \mathord{\left/ {\vphantom {{{1 \mathord{\left/ {\vphantom {1 {m_{y}^{ * } + 1}}} \right. \kern-0pt} {m_{y}^{ * } + 1}}} {m_{z}^{ * } }}} \right. \kern-0pt} {m_{z}^{ * } }}\text{)}^{ - 1}$ and the DOS effective mass $m_{b}^{ * } = (m_{x}^{ * } m_{y}^{ * } m_{z}^{ * } )^{{{1 \mathord{\left/ {\vphantom {1 3}} \right. \kern-0pt} 3}}}$ are evaluated by calculating the effective mass along the x, y and z directions, respectively.

By employing the similar treats of Lv et al. [36] and Chin et al. [46], we estimated the effective masses of electrons (m_e*) and holes (m_h*) in different directions by PBE functional with the equation: $m^{*} = \pm \hbar^{2} \left( {\frac{{{\text{d}}^{2} E_{k} }}{{{\text{d}}k^{2} }}} \right)^{ - 1}$. The calculated m_x*, m_y*, m_z*, m_I* and m_b* of electrons and holes for all structures are summarized in Table 3. Considering the high carrier mobility is the obvious characteristic of the lead halide perovskites, we also calculated the carrier mobility of CsPbBr₃ as a contrast for that of Cs₃Sb₂X₉. The spin–orbit coupling (SOC) effects have been employed to overcome the degeneracy of energy level of CsPbBr₃ in the electronic properties calculation due to the heavy element lead (Fig. S1). The calculated results of CsPbBr₃ are in agreement with those of Yettapu et al. [47]. Table 3 demonstrates that m_x*, m_y* of electron or hole for Cs₃Sb₂X₉ are equivalent but different from their m_z*, which can be ascribed to the same arrangement in x and y directions. It implies that the transport properties of Cs₃Sb₂X₉ are anisotropic for both electron and hole. The bulk modulus B and the deformation potential constant E were calculated by fitting parabolic functions of the total energy and band edge positions toward unit strain, respectively. The results are shown in Fig. S2. Finally, the calculated values of m_I*, m_b*, B, E, μ are summarized in Table 4.

Table 3 Calculated effective masses (m*s) of electrons and holes in the unit of free electron mass and other works for pristine CsPbBr₃, Cs₃Sb₂X₉(Cl, Br, I)

Table 4 Calculated conductivity effective mass m_I*(m₀), density-of-states effective mass m_b*(m₀) and elastic constant B (eV/ų), deformation potential constant E (eV), carrier mobility μ (cm² V⁻¹ s⁻¹) of the CsPbBr₃, Cs₃Sb₂X₉(Cl, Br, I), respectively

As we can see from the table that the calculated mobility of electron and hole for CsPbBr₃ is 75.4 cm² V⁻¹ s⁻¹ and 38.5 cm² V⁻¹ s⁻¹, which is consistent with the experimental 52.0 cm² V⁻¹ s⁻¹ and 11.0 cm² V⁻¹ s⁻¹ of Bakr et al. [48], respectively. It is worth mentioning that the electron mobility of Cs₃Sb₂I₉ reaches 75% of that of CsPbBr₃, although that of Cs₃Sb₂Cl₉ and Cs₃Sb₂Br₉ is obviously small. The high electron mobility of Cs₃Sb₂I₉ implies that the photogenerated electrons can rapidly transfer from the inside to the surface of the semiconductor for the proton reduction reaction. In addition, the differences between mobilities of the electron and the hole for Cs₃Sb₂I₉ are more obvious in comparison with those of CsPbBr₃, which will significantly reduce the recombination rate of the photogenerated carriers and enhance the photocatalytic activity of Cs₃Sb₂I₉. In this sense, Cs₃Sb₂I₉ is more suitable for efficient photocatalysis than the other Cs₃Sb₂X₉.

Conclusions

In conclusion, we have investigated the feasibility of Cs₃Sb₂X₉ in the photocatalytic water splitting to produce hydrogen or CO₂ reduction by using the first-principles DFT with the HSE06 hybrid functional. The results reveal that all the three structures have appropriate band edges for the water splitting and CO₂ reduction reactions. It is worth noting that one most suitable band energy gap of 2.04 eV is identified for Cs₃Sb₂I₉, which corresponds to an obvious absorption in the visible light range. Unlike the other two structures, the mobility of the electrons for Cs₃Sb₂I₉ is close to that of the lead perovskite CsPbBr₃, which implies that the rapid migration of photogenerated charge pairs is also possible for Cs₃Sb₂I₉. Moreover, the large differences between the mobilities of the electron and hole for Cs₃Sb₂I₉ could decrease the recombination rate of electron and hole and enhance the photocatalytic performance. To sum up, Cs₃Sb₂I₉ is a promising candidate for the photocatalytic CO₂ reduction or water splitting to produce hydrogen driven by the visible light on the basis of the optimal band energy gap with the appropriate band edges, the strong visible light absorption and the large electron mobility.