Controlled Reduction of Sn4+ in the Complex Iodide Cs2SnI6 with Metallic Gallium

Metal gallium as a low-melting solid was applied in a mixture with elemental iodine to substitute tin(IV) in a promising light-harvesting phase of Cs2SnI6 by a reactive sintering method. The reducing power of gallium was applied to influence the optoelectronic properties of the Cs2SnI6 phase via partial reduction of tin(IV) and, very likely, substitute partially Sn4+ by Ga3+. The reduction of Sn4+ to Sn2+ in the Cs2SnI6 phase contributes to the switching from p-type conductivity to n-type, thereby improving the total concentration and mobility of negative-charge carriers. The phase composition of the samples obtained was studied by X-ray diffraction (XRD) and 119Sn Mössbauer spectroscopy (MS). It is shown that the excess of metal gallium in a reaction melt leads to the two-phase product containing Cs2SnI6 with Sn4+ and β-CsSnI3 with Sn2+. UV–visible absorption spectroscopy shows a high absorption coefficient of the composite material.


Introduction
Complex tin-based halides [1] with a general formula A I M II X 3 (where A is an inorganic or organic cation such as Cs + , K + , Rb + , or CH 3 NH 3 + [2]; M IV = Sn; and X = a halide anion of F − , Cl − , Br − , or I − ) are most promising as light-harvesting components in modern photovoltaic solar cells (PSCs) as alternatives to lead halides. Recently, the powerconversion efficiency of Pb-based inorganic-organic perovskite PSCs has overrun 25% in single-junction architecture [3]. Such progress of perovskite-solar-cell efficiency is a result of both the chemical design of new light-harvesting compounds and also the evolution of the overall structure of the photovoltaic cell that originated from the architecture of the dye-sensitized solar cell [4] and transformed later to a thin sandwich-like structure with nano-sized functional layers of solid materials [5]. Research of complex halides as new light-harvesting materials gives a chance to minimize significantly the size and make flexible perovskite photovoltaic cells.
It is remarkable that all of the most efficient perovskite solar cells are based on leadcontaining light absorbers; however, this raises concerns owing to their high toxicity [6,7]. Despite the progressive increase in the efficiency of lead PSCs, the toxicity of lead requires development of new lead-free analogues with appropriate optical and electrical characteristics. Lately, γ-CsSnI 3 with the p-type conductivity was used in PSCs as a light-harvesting compound; however, the phase has demonstrated rather poor stability to oxidation and hydrolysis [8][9][10][11][12]. Ichiba and Kanatzidis described four polymorphs for CsSnI 3 , including black B-α-CsSnI 3 , black B-β-CsSnI 3 , black B-γ-CsSnI 3 , and yellow (or green) Y-γ-CsSnI 3 [13,14]. Optical methods have shown that, among these polymorphs, the The reactions were carried out in sealed quartz ampoules and evacuated to a pressure of 1.6·10 −2 Torr. The reduction samples (RS compounds) (where x = 0-0.2) were prepared by grinding cesium iodide (CsI) (SigmaAldrich, St. Louis, MO, USA, 99.9%), tin iodide (SnI 4 ), and metallic gallium (Ga) (SigmaAldrich, 99.9999%) with the stoichiometric mass ratios given in Table S1 in Supplementary Materials in an agate mortar. The solid-solution samples of Cs 2 Sn 1−x Ga x I 6−x (SS compounds) were prepared in the same way but with the addition of elemental iodine in small excess amounts proportional to the cesium content. Mass ratios are given in Table S2 in Supplementary Materials. The pristine Cs 2 SnI 6 phase Nanomaterials 2023, 13, 427 3 of 12 was obtained using the stoichiometric molar ratio of CsI to SnI 4 (2:1). The mixtures were sealed in preliminary dried-quartz ampoules and heated with a rate of~0.2 • C/min to 300 • C and then annealed at this temperature for 96 h. The specimens were cooled to room temperature slowly and opened inside a nitrogen-filled glove box. All products were black in color, and irregularly shaped crystals were observed under a microscope. All samples were kept in closed black Ziploc bags in nitrogen and then studied at room temperature in air.

Characterization Methods
X-ray diffraction measurements (XRD) were performed on a Rigaku D/MAX 2500 diffractometer (Rigaku, Tokyo, Japan) equipped with a rotating copper anode (Cu-Kα 1,2 radiation) and operated at 45 kV and 250 mA from 5 to 80 • in 2θ, and the scanning rate was 3 • min −1 at a step of 0.02 • . The experimental data were analyzed using WinXPow (database PDF2) (STOE & Cie, Version 1.07, Darmstadt, Germany) to define the phase composition, whereas for the lattice parameter calculations, Jana 2006 software (ECA-SIG#3/Institute of Physics, Prague, Czech Republic) was applied. The crystal structure of the materials was designed using the VESTA (JP-Minerals, Tsukuba, Japan) and Diamond (Crystal Impact, Bonn, Germany) programs.
Mössbauer spectroscopy experiments were carried out in closed polycarbonate ampoules during 2 days using an original setup equipped with a Ba 119m SnO 3 source. Mössbauer spectra at 119 Sn nuclei were recorded on an MS-1104Em electrodynamic spectrometer operating in a constant-acceleration mode (CJSC Cordon, Rostov-on-Don, Russia). The analysis and model approximation of the spectra were performed with the SpectrRelax software application [37,38].
UV-visible diffuse reflectance spectra were collected using the UV-visible spectrometer Lambda 950 (PerkinElmer, Waltham, MA, USA). Measurements were performed in a spectral range of 200-2000 nm with a step scan of 1 nm at 298 K with a scanning rate of 1 nm/s using quartz glass as a reference. The optical energy band gap (E g ) was acquired using a Tauc plot as a dependence of (αhν) 2 on energy (hν).
The microstructure of the samples was studied using a scanning electron microscope with the field emission source LEO SUPRA 50VP (LEO Carl Zeiss SMT Ltd., Oberkochen, Germany) with 250 X-2.5k X magnification. The samples were analyzed by X-ray emission microanalysis with an X/MAX X-ray energy dispersive detector (EDX) (Oxford Instruments, High Wycombe, UK).

Results and Discussion
The SS samples of the estimated composition of Cs 2 Sn 4+ 1−x Ga 3+ x I 6−x and RS samples (assuming a composition of Cs 2 Sn 4+ 1−5x Sn 2+ 3x Ga 3+ 2x I 6−8x ) with x = 0-0.2 were synthesized by a reactive sintering method in vacuum (in closed ampoules), while almost in all quoted publications various wet-chemistry methods were applied for synthesizing Cs 2 SnI 6 and other complex iodide materials. The compounds taken in the stoichiometric mass ratios for the SS series are given in Table S1 in Supplementary Materials. Compositions of the RS samples had a lack of elementary iodine and an excess of metal gallium (Table S1, in Supplementary Materials). The weight of each sample was 0.01 g. We observed no traces of the possible reactions between iodine, tin iodide, or metal gallium with quartz ampoules because any reactions were negligible and did not change the stoichiometry of elements during the synthesis.
Samples of the SS series were synthesized in small excess amounts of elemental iodine to produce the I-reach materials with the n-type conductivity and low deficiencies of the anionic sublattice. We supposed the Sn 4+ octahedral positions in the Cs 2 SnI 6 structure to be the most preferable for heterovalent substitution by Ga 3+ (0.620Å in the octahedral environment), while the reduced tin (Sn 2+ ) cation (1.16 Å) is larger significantly than the 0.690 Å of the Sn 4+ ionic radii, which works against the formation of the substitutional solid Nanomaterials 2023, 13, 427 4 of 12 solution. We proposed the possible products in the RS series to be Cs 2 SnI 6 -like compounds with Sn 4+ cations substituted partially by Ga 3+ or Sn 2+ cations (Figure 1a).
Samples of the SS series were synthesized in small excess amounts of elemental iodine to produce the I-reach materials with the n-type conductivity and low deficiencies of the anionic sublattice. We supposed the Sn 4+ octahedral positions in the Cs2SnI6 structure to be the most preferable for heterovalent substitution by Ga 3+ (0.620Å in the octahedral environment), while the reduced tin (Sn 2+ ) cation (1.16 Å) is larger significantly than the 0.690 Å of the Sn 4+ ionic radii, which works against the formation of the substitutional solid solution. We proposed the possible products in the RS series to be Cs2SnI6-like compounds with Sn 4+ cations substituted partially by Ga 3+ or Sn 2+ cations (Figure 1a).
It is shown experimentally that all characteristic XRD reflections for the SS series ( Figure 1) match well to the peaks generated from the crystal data and are consistent with the experimental data for the Cs2SnI6 phases reported elsewhere [23,39]. For all samples of the SS series, XRD reflections are consistent with the cubic vacancy-ordered perovskite-like structure ( Figure 1b). The noticeable shift in the reflection positions to higher or lower 2Ɵ values with increasing Ga/Sn ratio x is not observed.  It is shown experimentally that all characteristic XRD reflections for the SS series ( Figure 1) match well to the peaks generated from the crystal data and are consistent with the experimental data for the Cs 2 SnI 6 phases reported elsewhere [23,39]. For all samples of the SS series, XRD reflections are consistent with the cubic vacancy-ordered perovskite-like structure (Figure 1b). The noticeable shift in the reflection positions to higher or lower 2θ values with increasing Ga/Sn ratio x is not observed.
According to Figure 1b, the XRD patterns of two samples related to the SS series (namely, x = 0.01 and 0.03) and one related to the pristine Cs 2 SnI 6 (where x = 0) can be regarded as single phase. Contrastingly, the XRD patterns of the RS series (Figure 1c) distinctly show the presence of the β-CsSnI 3 secondary phase. The most intensive diffraction peaks are associated with reflections of Cs 2 SnI 6 (PDF2 file #73-330), while the weaker group of peaks is associated with reflections of γ-CsSnI 3 (PDF2 file #71-1898). The intensities and number of CsSnI 3 -characteristic reflections increase with the percentage of gallium in the samples before the heat treatment (0→0.15). Meanwhile, diffractograms do not show reflections of any iodogallates such as CsGaI 4 and CsGa 2 I 7 (Figure 1c).
It is remarkable that for the samples with the substitution rate x of 0.09-0.15, the presence of B-β-CsSnI 3 (PDF2 file #80-2138) was observed. This β-phase is stable thermodynamically at higher temperatures of 78-152 • C but is present in the RS samples together with low-temperature B-γ-CsSnI 3 . Table 1 shows the unit-cell parameters for the perovskite-related phase of Cs 2 SnI 6 , which were determined using the Le Bail method. The estimated unit-cell parameter a for the undoped Cs 2 SnI 6 perovskite phase was found to be 11.6416 (8) Å, which is close to recently reported data (11.630 (10) Å, according to the PDF2 file 73-330). The evolution of the unit-cell parameter with increasing dopant content for the SS samples demonstrated a slight increase of the cell parameter a to 11.6426 (5) Å (x = 0.05). The unit-cell volume V changes only slightly. It changes the same way with an increase of gallium percentage in the charge (Table 1). It is important that the radius of Ga 3+ is smaller than that of Sn 4+ (0.62 Å vs. 0.69 Å), and only an insignificant decrease of the unit-cell parameter is expected if gallium(III) partially substitutes for tin(IV). The deficiency in the iodine sublattice could also influence cubic unit-cell parameters in the same manner. In contrast, substitution of Sn 4+ ions with the larger cations of Sn 2+ could act in a different way by increasing the unit-cell parameter and volume. The unit-cell parameter a of the "double perovskite" Cs 2 SnI 6 phase was estimated for the RS series to demonstrate its evolution with the substitution rate. The results are presented in Table 2. As soon as the samples include more than one phase of Ga-doped Cs 2 SnI 6 , these values are not characteristic for the substituted phases.  (10) To investigate the effect of gallium doping on the crystallization process of the Cs 2 SnI 6 phase, the morphological analysis of the as-prepared SS samples (before grinding) was carried out with scanning electron microscopy (SEM). In order to correctly estimate the distribution of gallium over the volume of the sample, the annealed materials were divided into several parts so that the surface and the fracture (volume) could be explored ( Figure S1). In the SEM images ( Figure 2 and Figure S2), we can see that, on average, 100~200-micron spherical grains consist of crystallites of different sizes and shapes. Comparatively, Figure  S2a4-d4 (x = 0, x = 0.01, x = 0.03, and x = 0.05) provides crystal size spreading based on the fracture top-view images. It is noticeable also that the characteristic grain size increased up to 50 µm with a gallium percentage in a melt, probably as a result of an excess of iodine in the ampoules. A similar effect was found recently for Cs 2 SnI 6 and many other complex iodides [40].  The EDX images and tables of the elemental composition of the samples sh gallium is not evenly distributed across the samples. A large amount of gallium served at the grain boundaries, which indicates the possibility of the formatio CsGaI4 phase. Since CsGaI4 is an ionic liquid and remains in a liquid state at a t ture of 300 °C, it becomes an ion exchanger (or solvent) between precursors/Cs2S this is the reason for the increase in the crystal size in doped samples. Gallium can seen in Cs2SnI6 crystallites ( Figures S3 and S7), albeit in small quantities. We assu The EDX images and tables of the elemental composition of the samples show that gallium is not evenly distributed across the samples. A large amount of gallium is observed at the grain boundaries, which indicates the possibility of the formation of the CsGaI 4 phase. Since CsGaI 4 is an ionic liquid and remains in a liquid state at a temperature of 300 • C, it becomes an ion exchanger (or solvent) between precursors/Cs 2 SnI 6 , and this is the reason Nanomaterials 2023, 13, 427 7 of 12 for the increase in the crystal size in doped samples. Gallium can also be seen in Cs 2 SnI 6 crystallites ( Figures S3 and S7), albeit in small quantities. We assume that gallium in the Cs 2 SnI 6 structure does not replace tin but is present in the form of interstitial defects in octahedral vacancies (as shown in Figure 1a), so it is not distributed uniformly. Additionally, it should be noted that in the micrographs of backscattered electrons, a chemical contrast that would show the presence of two phases is not observed.
The 119 Sn Mössbauer spectroscopy was used to analyze the oxidation state of tin atoms in both series of samples. Figure 3a-c shows characteristic 119 Sn Mössbauer spectra for RS samples (x = 0; 0.05; 0.09) after 2 days of spectra collecting (Table 3). In the Mössbauer spectra of the RS series (x = 0.05 and 0.09), the presence of a major sensibly unresolved resonant subspectrum characteristic of Sn 4+ (isomer shift δ ≈ 1.36 mm/s, quadrupole splitting ∆ ≈ 0.15 mm/s) and one minor doublet that corresponds to Sn 2+ (isomer shift δ = 3.84 mm/s and δ = 3.72 mm/s for x = 0.05 and x = 0.09, respectively) are observed. was not found. For the RS series, Mössbauer spectroscopy data do not contradict the results of the XRD phase composition analysis. Additionally, the experimental values of the hyperfine parameters correspond to previously reported data for the three presented individual phases [39,41]. According to Yamada et al. [42], minor quadrupole doublets correspond to the β-CsSnI3 phase. The doublet related to the B-γ-CsSnI3 phase, which has higher quadrupole splitting, is not found, and we assume that this is due to the small amount of this phase in the sample. An excess of the β-CsSnI3 phase in the RS sample with x = 0.05 could be a result of the concentration gradient of gallium in the sample. To investigate the optical properties of the black-color materials, diffuse reflectance spectra were analyzed in the wavelength range of 200-2000 nm. The estimated optical band gap values for the samples are in the range of 1.23-1.35 eV and diminish with increasing the gallium percentage in the samples. The profiles of the collected absorption spectra correspond to direct-gap semiconductors [43], while the estimated optical Eg values for the samples demonstrate larger Urbach energies for both RS and SS samples with the smaller concentration of gallium dopant. We assume that the decrease in the absorption edge intensity and its slight shift are due to a suppression of vacancies/intrinsic defects with a low formation energy and formation of in-gap states. If the pristine Cs2SnI6 phase is the p-type semiconductor, this phenomenon would contradict the theory. If the SS samples of Cs2Sn1-xGaxI6-x are n-conductive, as reported by most authors, the Ga 3+ atoms will serve as electron traps or increase the e'-h˙ recombination probability.
For the RS series (Figure 4a,b), we observed a wider absorption edge than that of the SS series. Most likely, most of the spectra correspond to the superposition of subspectra of the SS phase of Cs2Sn1-xGaxI6-x (Eg observed at 1.31-1.35 eV) and the black-color phase of The major component can be attributed to the cubic Cs 2 SnI 6 phase [39]. The isomer shift and profile of the spectrum for the double perovskite is further confirmed by the data for the SS sample (x = 0.05) given in Figure 3d. The hyperfine parameters of the phases are given in Table 3. The presence of Sn 2+ ions within the sample of the SS series was not found. For the RS series, Mössbauer spectroscopy data do not contradict the results of the XRD phase composition analysis. Additionally, the experimental values of the hyperfine parameters correspond to previously reported data for the three presented individual phases [39,41]. According to Yamada et al. [42], minor quadrupole doublets correspond to the β-CsSnI 3 phase. The doublet related to the B-γ-CsSnI 3 phase, which has higher quadrupole splitting, is not found, and we assume that this is due to the small amount of this phase in the sample. An excess of the β-CsSnI 3 phase in the RS sample with x = 0.05 could be a result of the concentration gradient of gallium in the sample. To investigate the optical properties of the black-color materials, diffuse reflectance spectra were analyzed in the wavelength range of 200-2000 nm. The estimated optical band gap values for the samples are in the range of 1.23-1.35 eV and diminish with increasing the gallium percentage in the samples. The profiles of the collected absorption spectra correspond to direct-gap semiconductors [43], while the estimated optical E g values for the samples demonstrate larger Urbach energies for both RS and SS samples with the smaller concentration of gallium dopant. We assume that the decrease in the absorption edge intensity and its slight shift are due to a suppression of vacancies/intrinsic defects with a low formation energy and formation of in-gap states. If the pristine Cs 2 SnI 6 phase is the p-type semiconductor, this phenomenon would contradict the theory. If the SS samples of Cs 2 Sn 1-x Ga x I 6-x are n-conductive, as reported by most authors, the Ga 3+ atoms will serve as electron traps or increase the e'-h˙recombination probability.
For the RS series (Figure 4a,b), we observed a wider absorption edge than that of the SS series. Most likely, most of the spectra correspond to the superposition of subspectra of the SS phase of Cs 2 Sn 1-x Ga x I 6-x (E g observed at 1.31-1.35 eV) and the black-color phase of γ-CsSnI 3 (E g = 1.23 eV [8]). This correlates partly to the absorption spectra of these materials, revealing local maxima at~630 nm for all samples (Figure 4a,c), with small discrepancies within the SS and RS series.
13, x FOR PEER REVIEW 9 of 13 γ-CsSnI3 (Eg = 1.23 eV [8]). This correlates partly to the absorption spectra of these materials, revealing local maxima at ~630 nm for all samples (Figure 4a,c), with small discrepancies within the SS and RS series.   Thus, the reduction of Sn 4+ to Sn 2+ by gallium melt leads to the formation of perovskite phases, namely, the black γ-CsSnI3 genuine perovskite and β-CsSnI3 phase as an admixture. The latter is promising for light harvesting. It is remarkable also that Cs2SnI6 is a p-type semiconductor [22] and γ-CsSnI3 is an n-type semiconductor [7]. This fact makes Thus, the reduction of Sn 4+ to Sn 2+ by gallium melt leads to the formation of perovskite phases, namely, the black γ-CsSnI 3 genuine perovskite and β-CsSnI 3 phase as an admixture. The latter is promising for light harvesting. It is remarkable also that Cs 2 SnI 6 is a p-type semiconductor [22] and γ-CsSnI 3 is an n-type semiconductor [7]. This fact makes the growth of the sandwich-like planar structures of Cs 2 SnI 6 and γ-CsSnI 3 attractive, aiming at the p-n heterojunction materials for the solid-state solar cells [44]. Thus, the composite made of Ga-doped Cs 2 SnI 6 with γ-CsSnI 3 and β-CsSnI 3 perovskites seems to be a model for further investigation of corresponding cesium iodostannates (II, IV) composite films.

Conclusions
Our study improves the understanding of the double-perovskite Cs 2 SnI 6 perspective as a perovskite solar-cell compound. According to the XRD data, no iodogallates were found in the samples for the x < 0.05 compositions, and so the formation of the substitutional solid solution of the double-perovskite Cs 2 Sn 1-x Ga x I 6-x up to 5 at.% of Ga is still under discussion. On the other hand, the interaction of the tin-based double-perovskite Cs 2 SnX 6 with metal gallium leads to the formation of a new light-harvesting composite compound and a hole-or electron-transport material with improved grain boundaries and appropriate conductivities of charge carriers. It is remarkable also that the presence of Sn 2+ cations in the double-perovskite structure is still under discussion due to the easy segregation of tin(II) in a form of individual phases of β-CsSnI 3 and γ-CsSnI 3 . These results suggest that it is possible to optimize the crystal quality and optoelectronic properties by doping the perovskite structures (ABX 3 , A 2 BX 6 , A 3 B 2×9 ) with relevant heterovalent cations. We expect that this research will be a significant step for obtaining "less toxic" light-absorber materials with improved characteristics.
Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/nano13030427/s1. Table S1: Composition of Cs 2 Sn 1-x Ga x I 6-x solidsolution samples (SS series); Table S2: Composition of samples for the reduction of Cs 2 SnI 6 by metallic gallium (RS); Figure S1: Optical photographs of a piece of compound after synthesis (before grinding) for SEM and EDX measurements (an example of the SS series); Figure