Introduction

Halide alloying has been widely used to expand the optoelectronic landscape of halide perovskites. [1,2] Indeed, the band gap and band dispersion of 3D lead-halide perovskites (APbX 3 ; A = monovalent cation, X = halide), which are promising photovoltaic absorbers, [3,4] predictably vary with the electronegativity and size of the halide. Thus, mixing halides in APbX 3 perovskites affords band gaps ranging from 1.5-3.0 eV-encompassing near-ideal values for absorbers in single-and multi-junction solar cells. [5][6][7] However, halide mobility and light-induced halide segregation in mixed-halide compositions have impeded the use of many of these compositions with desirable band gaps, particularly for multi-junction cells with silicon. [8,9] Thus, other ways of tuning the band gap of halide perovskites need to be investigated. In order to maintain its exceptional band structure, the halides in the perovskite should ideally be replaced by isolobal, π-donor ligands. We recently reported the new family of 3D organosulfide-halide perovskites: (CYS)PbX 2 (CYS = + NH 3 (CH 2 ) 2 S À , X = Cl, Br), where the zwitterionic organosulfide replaces both the A + site and an X À site in the prototypical APbX 3 perovskite, introducing sulfide into the perovskite framework. [10] The stronger PbÀS bonds and covalent attachment of the A + -site cation to the inorganic framework likely contribute to the materials' improved thermal and moisture stability compared to those of (CH 3 NH 3 )PbX 3 . DFT calculations show direct band gaps with dispersive bands in (CYS)PbX 2 . However, the band gaps of 2.31 eV and 2.16 eV for the X = Cl and Br analogs, respectively, are still higher than the ideal values (1.7-2.0 eV) [11] required for coupling the perovskite with a lower-band gap bottom absorber in a multi-junction solar cell.

Although the valence band maximum (VBM) of APbX 3 perovskites is dominated by halide and Pb states, [12] the VBM of (CYS)PbX 2 has mostly S and Pb states, with a markedly smaller halide contribution. [10] Hence, sulfur plays a crucial role in dictating the band gap magnitudes. The band gaps of APbX 3 decrease monotonically from X = Cl, Br, to I due to the increasing energy of the valence np orbitals in the halide. Therefore, we hypothesized that replacing the S in (CYS)PbX 2 with the less electronegative Se may result in a similar band gap reduction. Indeed, Se alloying in other sulfide-based solar absorbers has reduced the band gaps, including in chalcopyrite Cu(In,Ga)(S,Se) 2 and kesterite Cu 2 ZnSn(S,Se) 4 . [13,14] Selenide perovskites are predicted to exist and are expected to be promising solar absorbers with desirable band gaps and carrier mobilities. [15][16][17] However, to our knowledge, there are very few examples of known selenide perovskites (LaScSe 3 and uranium-based perovskites) [18] and Se-alloying of oxide perovskites [19,20] requires high-temperature O 2 -free synthetic conditions.

Following our strategy of employing zwitterions to introduce chalcogenides into halide perovskites, we describe the synthesis of the first examples of organoselenidehalide perovskites: (SeCYS)PbX 2 (Figure 1, X = Cl, Br) using the zwitterion SeCYS ( + NH 3 (CH 2 ) 2 Se À ). We describe the average and local structures of these materials and show that Se substitution for S indeed reduces the band gaps (E g ), affording E g = 2.07 eV and 1.86 eV for X = Cl and Br, respectively, in the ideal range for a multi-junction solar cell with Si. We are further able to alloy the halide (Cl/Br) site or the chalcogenide site (S/Se) to tune the band gaps continuously from 1.86 to 2.31 eV.

Results and Discussion

Synthesis and Basic Characterization

Replacing CYS with SeCYS, in the synthetic procedure used for (CYS)PbX 2 , does not afford phase-pure (SeCYS)PbX 2 perovskites. Instead, the (SeCYS)PbX 2 perovskites required a new synthetic procedure of mixing the lead precursors (ca. 1 : 1 molar ratio of PbX 2 and Pb-(OAc) 2 • 3H 2 O) and selenocysteamine hydrochloride (Se-CYS•HCl) in concentrated NaX aqueous solutions at 90 °C to afford red and black powders for X = Cl and Br, respectively. A lower precursor concentration was used to yield crystals; only the crystals of (SeCYS)PbCl 2 were suitable for single-crystal X-ray diffraction (SC-XRD; Figure S1, see the Supporting Information for detailed procedures). We were unable to synthesize the iodide analog using similar synthetic conditions. Elemental analysis of the bulk powder (Combustion analysis for C, H, and N and inductively coupled plasma mass spectrometry for Pb and Se) confirmed the formula as (SeCYS)PbX 2 for both perovskites. X-ray photoelectron spectroscopy (XPS) was also used to estimate the elemental compositions and relative ratios (Figure S3). Solution-state 1 H NMR, measured by dissolving the perovskites in deuterated acid, showed that the SeCYS ligand remains unchanged after forming the perovskites.

To probe the ligand environment in the solid (SeCYS)PbX 2 , we then applied solid-state 13 C NMR with magic angle spinning (MAS). The 13 C chemical shifts of the α-C and β-C are 18.7 ppm (19.0 ppm) and 46.8 ppm (46.9 ppm) for (SeCYS)PbCl 2 ((SeCYS)PbBr 2 ), respectively, which are consistent with those in SeCYS•HCl: 17.8 ppm and 44.3 ppm for α-C and β-C, respectively (Figure S16). The thermogravimetric analyses (scan rate 1 °C/minute) under N 2 atmosphere determine the decomposition temperatures (T d , corresponding to 5 % mass loss) to be 234.5 °C and 222.8 °C for (SeCYS)PbCl 2 and (SeCYS)PbBr 2 , respectively (Figure S5). These values are slightly higher than the values for (CYS)PbX 2 (207.1 °C and 199.0 °C for X = Cl and Br, respectively), presumably due to the less volatile ligand.

The Average Structure of (SeCYS)PbX 2

SC-XRD data of (SeCYS)PbCl 2 were solved in the trigonal space group R-3c, with a disordered organic component (ÀCH 2 CH 2 NH 3 + ) and indistinguishable Se/Cl sites, similar to the SC-XRD solution of (CYS)PbCl 2 . [10,21] The freely refined occupancies of Cl (0.65) and Se (0.35) in the disordered anion sites are close to the theoretical values (0.67 and 0.33 for Cl and Se, respectively) based on the formula.

The SC-XRD solution reflects the intrinsic disorder of the mixed-anionic perovskite with an organic component that adopts different configurations. For independent structural verification, high-energy X-ray scattering data, suitable for pair distribution function (PDF) analysis, were collected at room temperature using fine powders of (SeCYS)PbX 2 . The experimental PDF data were both modeled in the R-3c space group, only considering the inorganic framework (i.e., Pb and anion sites) in the initial refinement. The anion site was modeled as 1/3 Se and 2/3 X. The contribution of the disordered organic component (~C 2 N = 19 electrons) in the cuboctahedral cavity was modeled using a virtual atom with an equivalent number of electrons (K atom) with a large, isotropic atomic displacement parameter reflecting the dynamic disorder of this group. The most significant misfit to experimental data is an asymmetry in the first peak that is associated with the positional Se/X disorder and the resulting distribution of the PbÀSe and PbÀX bond lengths, as seen in other mixedanion perovskites, [22] which results in local distortions of the Pb-centered octahedra (Figure 1D). The distribution of PbÀSe/X distances was modeled by allowing separate atomic coordinates for the Se and X sites, while constraining the atomic displacement parameters to be isotropic. To best account for local disorder, the PDF data were refined over a short range (2-12 Å) (Figure S7). The lattice parameters and fitting results from the long-and shortrange PDF refinements are summarized in Tables S2 and S3. The good agreement between the experimental PXRD data and simulated patterns from the PDF models validates the PDF fitting (Figure S2). We then compared the roomtemperature cell parameters of the four organochalcogenide-halide perovskites LPbX 2 (L = CYS, SeCYS; X = Cl, Br) all obtained by PDF fitting. [10] A linear lattice expansion is evident from (CYS)PbCl 2 < (SeCYS)PbCl 2 < (CYS)PbBr 2 < (SeCYS)PbBr 2 since they crystallize in the same space group and due to the similar radii between Cl À (1.81 Å) and S 2À (1.84 Å) and between Br À (1.96 Å) and Se 2À (1.98 Å) (Figure S8). [23] Since the inherent disorder of the perovskite crystal structure does not allow us to see the configuration of the organoammonium tail of the organochalcogenide, we obtained Raman spectra of (CYS)PbCl 2 , (SeCYS)PbCl 2 , and (SeCYS)PbBr 2 to probe the ligand vibrations (Figure 2A, Table S7, Figure S4). The Raman spectra were collected at ambient conditions and repeated scans showed no signs of material decomposition, by oxidation or laser damage, over 10 minutes. The low-energy Raman signals below 200 cm À1 are assigned to lattice phonon modes through comparison to the Raman data of (MA)PbX 3 perovskites. [24,25] We observed a redshift of lattice modes with S to Se and Cl to Br substitution. The higher-energy modes (> 200 cm À1 ) are attributed to CYS/SeCYS ligand modes, which we assigned based on reported experimental/ theoretical studies on CYS (see Table S7 for assignments). The peaks at 380 and 327 cm À1 in the spectra of (CYS)PbCl 2 and (SeCYS)PbCl 2 , respectively, are assigned to the CÀN torsion mode. In (MA)PbX 3 perovskites, this mode is more pronounced and broader due to the free rotation of MA + . For (CYS)PbX 2 and (SeCYS)PbX 2 , the CÀN rotation is more restricted due to sterics and covalent bonding between Pb and S/Se, resulting in a peak with much weaker intensity. Next, we analyzed the highly polarizable CÀCh (Ch = S, Se) stretching vibrations. Importantly, these modes are well-known to be sensitive to the rotational isomerization around the CÀC bond (Figure 2B). [26][27][28][29][30] In (CYS)PbCl 2 , the peaks for the gauche (G) and trans (T) CÀS stretching modes are centered at 650 and 738 cm À1 , respectively. This is consistent with the reported values of 640 and 725 cm À1 , observed for CYS chemisorbed on a Ag surface via the terminal S. [31] We observed two

Angewandte

Chemie closely spaced ν(CÀS) T modes in (CYS)PbCl 2 (at 736, 749 cm À1 by fitting), possibly from two trans conformers with slightly different rotational angles. In (SeCYS)PbX 2 , the ν(CÀSe) G and ν(CÀSe) T peaks center at 550 (554) and 654 (655) cm À1 , respectively, for X = Cl (Br). The relative ratios of the two conformers can be semi-quantitatively calculated from the integrated peak intensity ratios (I T /I G ) for ν(CÀCh) T and ν(CÀCh) G . At ambient conditions, the zwitterionic ligands prefer to adopt a trans conformer inside the cuboctahedral cavity of the perovskite, with I T / I G = 15, 8.6, and 6.3 for (CYS)PbCl 2 , (SeCYS)PbCl 2 , and (SeCYS)PbBr 2 , respectively. The most intense peak of ligand modes (ca. 1230-1260 cm À1 ) is tentatively assigned to the twisting/wagging modes of the methylene groups, which have also been observed as strong peaks (ca. 1220-1230 cm À1 ) in the Raman spectrum of (Az)PbX 3 (Az = aziridinium), although assigned differently. [32]

The Local Structures in LPbX 2 (L = CYS and SeCYS)

Since the organochalcogenide-halide perovskites are heteroanionic materials [33,34] with anion disorder, a closer look at their local structures is warranted. Previous structural studies of (CYS)PbX 2 through X-ray scattering (SC-XRD, PXRD, PDF) indicated disordered sulfide and halide sites in the long-range or average structure. [10] Yet, the local ordering of anions around the Pb center remains unknown. We first applied solid-state 77 Se NMR to probe the chemical environment of the selenide in the solid. The PbÀSe bonding dramatically changes the 77 Se chemical shift from À51.3 ppm (free ligand) to 4.7 ppm and À6.0 ppm for (SeCYS)PbCl 2 and (SeCYS)PbBr 2 , respectively, indicating the formation of metal-selenide bonds. [35,36] The broad linewidths of the 77 Se signals in the perovskites are attributed to the local disorder in the structure (Figure 3C).

Solid-state 207 Pb NMR has served as a powerful tool to directly probe the local environment of Pb and investigate halide order on the local scale in lead-halide perovskites. [37,38] The 207 Pb nuclide, with I = 1/2, at 22 % abundance, and with high receptivity (12 times compared with that of 13 C), enables the NMR measurements to be relatively accessible. [39] The chemical shift, spanning a large window (~20,000 ppm) and linewidth are both highly sensitive to the local coordination. [40] The coordination environments of Pb in mixed-Cl/Br 3D perovskites have Angewandte [40][41][42] The ratios of these seven environments are consistent with the binomial distribution (Figure 3D), which indicates that the two anions randomly occupy the X À site to form solid solutions.

Here, we considered the two extreme local orderings of Pb coordination in the LPbX 2 : (1) fully ordered: each Pb is coordinated by two chalcogenides and four halides; (2) fully disordered: the chalcogenide ligands randomly distribute in the framework. In the first scenario, we would expect only one chemical environment of Pb, corresponding to one 207 Pb NMR signal arising from [PbL 2 X 4 ] 2À (not distinguishing between cis/trans isomers). For the fully disordered distribution, there are 7 possible local structures, [PbL 6-x X x ] n (x = 0, 1, 2, …, 6), which closely resembles the case of mixed-halide perovskites (Figure 3B). The theoretical distributions of L and X in the LPbX 2 perovskites are more complicated than the simple binomial distribution observed in mixed-halide perovskites due to the additional restraint that the zwitterions must connect the A + site and anion site. We designed an algorithm to simulate the distribution of Pb coordination environments in LPbX 2 using a large superlattice (200×200×200) assuming fully disordered ligands (Figure 3A, see Supporting Information for assumptions and code). The calculation results are similar, though not identical, to the theoretical distribution of Pb coordination spheres in mixed-halide perovskites with a halide alloying ratio of 1 : 2, as shown in Figure 3D.

We obtained the solid-state 207 Pb NMR spectra of (CYS)PbCl 2 , (SeCYS)PbCl 2 , and (SeCYS)PbBr 2 as shown in Figure 3E. All three materials exhibit a wide range of 207 Pb signals, from À1000 ppm to 3000 ppm, with a relatively narrow peak at a lower frequency followed by a broad main peak at a higher frequency. The wide NMR spectra were collected using the variable offset cumulative spectra (VOCS) approach. [43] The resulting spectra are full chemical shift powder patterns, yet we can assign the features according to the local bonding environment. The narrow peaks at the lower frequency are assigned as the [PbX 6 ] 4À species ([PbCl 6 ] 4À : À696 ppm and À656 ppm in (CYS)PbCl 2 and (SeCYS)PbCl 2 , respectively, and À646 ppm in (MA)PbCl 3 ; [PbBr 6 ] 4À : 232 ppm in (SeCYS)PbBr 2 , 344 ppm in (MA)PbBr 3 ). The differences in the 207 Pb NMR chemical shift observed in different perovskites even for the same local coordination (e.g., for [PbCl 6 ] 4À in (CYS)PbCl 2 and (SeCYS)PbCl 2 ) show the sensitivity of the NMR signal to the next-next-nearest neighbors, as seen in the NMR spectra of (MA)PbX 3-x Y x (X, Y = halides) as well. [42] The broad signals are therefore attributed to chalcogenide-containing species [PbL 6-x X x ] n (L = CYS, SeCYS; x = 0, 1, 2, …, 5). The existence of [PbX 6 ] 4À species rules out the fully ordered structure. We then integrated the [PbX 6 ] 4À peaks to obtain their ratios in the whole spectra semi-quantitatively. The ratios of [PbX 6 ] 4À were estimated to be 11 %, 16 %, and 11 % in (CYS)PbCl 2 , (SeCYS)PbCl 2 , and (SeCYS)PbBr 2 , respectively, which are consistent, within the precision of the measurement, with the theoretical ratio of [PbX 6 ] 4À calculated for the fully disordered model (7.0 %).

Two-dimensional 207 Pb exchange spectroscopy (EXSY) spectra were also collected at room temperature with a mixing time of 2 ms (Figure S17). Due to the wide range of 207 Pb signals, we were unable to measure 2D EXSY covering the whole range within a feasible timescale. Therefore, we selectively excited at À720 ppm and 240 ppm for (CYS)PbCl 2 and (SeCYS)PbBr 2 , respectively, to obtain signals from [PbX 6 ] 4À and from some [PbL 6-x X x ] n species. In contrast to the mixed-halide perovskites, where halide exchange causes cross-peaks in room-temperature 2D EXSY, [40] no cross-peaks were observed, indicating the absence of halide/chalcogenide ligand exchange in the LPbX 2 perovskites. The immobility of the organochalcogenide can be explained by the stronger bonding interaction between Pb and the chalcogenide and the expected low mobility of the larger zwitterionic ligand. [44,45]

Electronic Structure

We performed density functional theory (DFT) calculations of the electronic band structures of (SeCYS)PbX 2 , including spin-orbit coupling, [46][47][48] using atomic coordinates obtained from the PDF analysis (see Supporting Information for computational details). We model the ligand disorder similarly as in the case of (CYS)PbCl 2 by first building an ordered model, assuming a trans coordination of SeCYS to Pb. The calculations indicate a direct band gap at the Γ point, with E g values of 0.53 eV and 0.41 eV for X = Cl and Br, respectively (Figures 4A and S10). Although the band gaps are underestimated, as expected at this level of theory, the difference in gaps between the Cl and Br analogs (0.12 eV) is close to the experimentally measured value (0.21 eV, described later). The valence band maximum (VBM) consists of 67 % (60 %) Se and 22 % (23 %) Pb states, with only 11 % (17 %) halogen states for (SeCYS)PbCl 2 ((SeCYS)PbBr 2 ). The dominant contributions from the chalcogenide at the VBM are also seen in (CYS)PbX 2 (58 % S for X = Cl and 52 % S for X = Br). [10] The greater contribution of Se states in the VBM of (SeCYS)PbX 2 , compared to S states in the VBM of (CYS)PbX 2 , is consistent with the lesser electronegativity of Se. The conduction band minimum (CBM) at the Γ point mostly consists of Pb (> 90 %) states, with small contributions from halide and chalcogenide states. Therefore, the band gaps in the organochalcogenide-halide perovskites are mostly determined by the chalcogenide: the higherenergy filled np frontier orbitals of Se, compared to those of S, afford a smaller band gap in (SeCYS)PbX 2 . We then compared the band structures of (SeCYS)PbX 2 with those of (CYS)PbX 2 (Figures 4B and S10) by arbitrarily aligning their CBM. The band edges remain almost unchanged in shape, with the VBM appearing higher in (SeCYS)PbX 2 , as expected due to the band gap reduction. The calculated carrier effective masses of (SeCYS)PbX 2 are comparable with those in CsPbX 3 (e.g., < m* h > = 0.131, < m* e > = 0.133 for (SeCYS)PbBr 2 ) as shown in Table S5. Further, we tested the sensitivity of the band structures to the anion arrangements about the central Pb 2 + . The two structures for the hypothetical CsPb(SeH)X 2 , where Cs + and HSe À mimic the zwitterionic + NH 3 (CH 2 ) 2 Se À , feature (1) all trans and (2) both cis and trans arrangements of HSe À in the [Pb(SeH) 2 X 4 ] 4À octahedra. The CsPb(SeH)X 2 models show similar band dispersion and orbital contributions at the band edges compared with the band structure of (SeCYS)PbX 2 , especially at the Γ point (Table S4, Figure S13), indicating that these anion arrangements should have minimal effects on the electronic structure.

The valence-band dispersion of (SeCYS)PbX 2 , and of its model complex CsPb(SeH)X 2 , is reduced compared to that of a hypothetical CsPbX 3 adopting the same lattice constants and space group as for (SeCYS)PbX 2 . To test whether the flatter valence band was due to the Se orbitals, we also calculated the band structure of a hypothetical all-Se perovskite: CsPb(SeH) 3 . We see a sharper dispersion in the valence band of CsPb(SeH) 3 compared to the valence band of CsPbCl 3 (Figure 5), consistent with the more diffuse Se orbitals providing greater overlap with Pb orbitals. Thus, we conclude that the reduced valence band dispersion in our calculations of (SeCYS)PbX 2 is due to the anisotropic arrangement of ligands about the central Pb atoms in our models. This conclusion is consistent with the profile of the squared modulus of the electron wave function, corresponding to the VBM (Γ point) shown in Figure S15, which clearly displays an asymmetric orbital environment between the central Pb cation and the adjacent anions, in the case of CsPb(SeH)X 2 .

The ordered atomistic models used in our DFT calculations cannot capture the distribution of local environments in (SeCYS)PbX 2 and they do not account for structural distortions associated with the inhomogeneous distributions of different Pb environments. However, our calculations hint at the possibility that locally chalcogeniderich Pb centers (e.g., Pb(SeCYS) 6 , Pb(SeCYS) 5 X, etc.), which have a non-zero probability of occurring in the mixed-anion landscape (Figure 3D), may form sub-band gap trap states.

Alloying the Anion Site

The organochalcogenide-halide perovskites allow us to continuously tune the band gap by alloying both the chalcogenide and halide sites, since they both contribute states to the VBM, according to the calculated electronic structures.

Halide alloying for (SeCYS)PbCl 2(1-x) Br 2x was conducted by tuning bromide and chloride concentrations in the precursor solution (solution-state synthesis). We found that ball-milling together different ratios of the solid perovskite endmembers: (SeCYS)PbCl 2 and (SeCYS)PbBr 2 (mechanochemical synthesis), could also form (SeCYS)PbCl 2(1-x) Br 2x , similar to the mechanochemical syntheses of mixed-halide [8,40] and mosaic [49] perovskites. Chalcogenide alloying for (CYS) 1-x (SeCYS) x PbX 2 (X = Cl, Br) was accomplished by tuning the ratio of the two ligands The Pb, Br, and Se orbitals that contribute to the bands are depicted as turquoise, brown, and magenta dots, respectively, where the dot size is proportional to the atomic contribution. The band structure is calculated along the high symmetry path (in crystal coordinates) Γ (0,0,0) -M (0.5,0,0) -K (0.333,0.333,0) -Γ (0,0,0) -A (0,0,0.5) -L (0.5,0,0.5) -H (0.333,0.333,0.5) -A (0,0,0.5). [47] The SeCYS molecules are modeled as ordered (see Supporting Information). (B) Comparison of the band structures of (CYS)PbBr 2 and (SeCYS)PbBr 2 , where the conduction band minima have been arbitrarily aligned. in the solution-state synthesis (see Supporting Information for detailed synthetic procedures). Notably, we were unable to mix the chalcogenides through a mechanochemical synthesis, starting directly from the two endmembers: (CYS)PbCl 2 and (SeCYS)PbCl 2 (Figures S20, S21). Our ability to mechanochemically alloy the halide but not the chalcogenide corroborates the lower mobility of the larger zwitterionic ligand and its stronger bond with Pb compared to those with the halides, which are also supported by the 2D 207 Pb NMR studies.

The alloyed perovskites were characterized by PXRD and Le Bail refinements to obtain unit-cell parameters (Figures S24-S29, Table S6). The compositions of the alloyed perovskites were determined by 1 H NMR (of the dissolved product) to quantify the ratios of SeCYS and CYS ligands, and inductively coupled plasma mass spectrometry (ICP-MS) to quantify the Pb to Br ratios (Figures S18, S19). The dependence of unit-cell volumes on the anion alloying ratio is summarized in Figure Combining both the Cl/Br and S/Se alloying strategies, we realized three series of compositions: i) (SeCYS)PbCl 2(1-x) Br 2x (through solution-and solid-state syntheses), ii) (CYS) 1- x (SeCYS) x PbCl 2 , and iii) (CYS) 1-x (SeCYS) x PbBr 2 .

Optical Properties

We obtained optical gaps of 2.07 eV and 1.86 eV for (SeCYS)PbCl 2 and (SeCYS)PbBr 2 , respectively, using diffuse reflectance spectroscopy and direct-band gap Tauc plots (Figures 7A and S9). In accordance with the calcu- 7B), from 2.31 to 2.07 eV for the 1-x (SeCYS) x PbCl 2 series, (Figures 6, 7B) and from 2.16 to 1.86 eV for the (CYS) 1-x (SeCYS) x PbBr 2 series (Figure S21).

Similar to (CYS)PbX 2 , (SeCYS)PbX 2 and its alloys show no room-temperature photoluminescence (PL) and a broad, Stokes shifted PL at 80 K. The 80-K emission maxima of 1.58 eV and ca. 1.4 eV, for (SeCYS)PbX 2 with X = Cl and Br, respectively, (Figure 7C) are considerably redshifted compared to those of the (CYS)PbX 2 analogs (1.98 and 1.66 eV for X = Cl and Br, respectively). [10] Because much of the emission from (SeCYS)PbBr 2 falls outside the detector range, we used an IR detector to estimate the PL maximum (Figures S22-S23). The broad emission with a large (ca. 0.5 eV) Stokes shift is potentially indicative of trap states originating from selenide-rich local coordination or from other defects.

Powders of (SeCYS)PbCl 2 and (SeCYS)PbBr 2 , exposed to 1-sun illumination in air at 40 °C, maintain their PXRD patterns over the course of 4 days, with some reduction of crystallinity evident by 27 days although we do not see the growth of new crystalline phases (Figure S30). The longerterm stability of the organochalcogenide-halide perovskites should be further assessed.

Conclusions

We present the first examples of organoselenide-halide perovskites: (SeCYS)PbX 2 (X = Cl, Br). The zwitterionic organoselenide ligand occupies both the A + site and X À site in the prototypical ABX 3 perovskite, further expanding the family of organochalcogenide-halide perovskites, in addition to the recently reported organosulfide-halide perovskites: (CYS)PbX 2 . [10] X-ray diffraction and total scattering reveal disordered anions over long length scales in the average structures, and a systematic lattice expansion with (CYS)PbCl 2 < (SeCYS)PbCl 2 < (CYS)PbBr 2 < (SeCYS)PbBr 2 . The ligand dynamics and rotational isomerism were probed by Raman spectroscopy. Further, we apply solid-state 77 Se and 207 Pb NMR to study the local structures of the organochalcogenide-halide perovskites, complemented by theoretical simulations of the distributions of different coordination environments. We find that the heteroanionic perovskites LPbX 2 (L = CYS, SeCYS) can be viewed as solid solutions with the chalcogenide/ halide randomly distributed at the anion site, similar, but not identical, to mixed-halide perovskites with a 2 : 1 ratio between different halides.

The incorporation of Se affords smaller band gaps compared to the S analogs, with gaps of 2.07 eV and 1.86 eV for (SeCYS)PbCl 2 and (SeCYS)PbBr 2 , respectively. The electronic effects of Se substitution were further studied by DFT, revealing reduced band gaps, which we attribute to the higher frontier p orbital energy level in Se compared to that of S. Using both solution-state and solidstate mechanochemical syntheses, we demonstrate halide alloying and chalcogenide alloying to afford a continuous band gap shift from 1.86 to 2.31 eV with the band gap increasing as (SeCYS)PbBr 2 < (SeCYS)PbCl 2(1-x) Br 2x < (SeCYS)PbCl 2 < (CYS) 1-x (SeCYS) x PbCl 2 < (CYS)PbCl 2 . These band gaps, encompassing near-ideal values for the top absorber in tandem solar cells, [11] or for photocatalysis with visible light, [50] motivate further studies to form films of organochalcogenide-halide perovskites and tune their defect chemistry to assess their potential for photocarrier extraction.

Knowledge of both the long-range structure and distribution of local structures, presented here, will be important to understand and exploit the, as yet unoptimized, optoelectronic properties of this nascent family of organochalcogenide-halide perovskites. Further studies should probe whether the local disorder seen in the solidstate NMR studies (Figure 3) leads to chalcogen-rich subband gap trap states, as suggested by the lower band gaps computed through DFT of chalcogen-rich models (Figure 5), and how these potential trap states, or other photoluminescence-quenching defects (e.g., halogen vacancies), can be suppressed. Furthermore, the organochalcogenides exhibit much lower anion mobility than the halides, based on 2D NMR and mechanochemical synthetic attempts, likely due to the ligand size and stronger bonding enthalpy with Pb. The immobility of anions, at least in part of the inorganic framework, may enhance stability, potentially mitigating the well-known halide mobility [44,51,52] and light-induced halide segregation in mixed-halide perovskites. [8] Overall, the introduction and expansion of the family of organochalcogenide-halide perovskites offer new handles for tuning the properties of halide perovskites using insights from solid-state as well as organo-maingroup chemistry.