Bromine Pentafluoride BrF5, the Formation of [BrF6]− Salts, and the Stereochemical (In)activity of the Bromine Lone Pairs

Abstract BrF5 can be prepared by treating BrF3 with fluorine under UV light in the region of 300 to 400 nm at room temperature. It was analyzed by UV‐Vis, NMR, IR and Raman spectroscopy. Its crystal structure was redetermined by X‐ray diffraction, and its space group was corrected to Pnma. Quantum‐chemical calculations were performed for the band assignment of the vibrational spectra. A monoclinic polymorph of BrF5 was quantum chemically predicted and then observed as its low‐temperature modification in space group P21/c by single crystal X‐ray diffraction. BrF5 reacts with the alkali metal fluorides AF (A=K, Rb) to form alkali metal hexafluoridobromates(V), A[BrF6] the crystal structures of which have been determined. Both compounds crystallize in the K[AsF6] structure type (R 3‾ , no. 148, hR24). For the species [BrF6]+, BrF5, [BrF6]−, and [IF6]−, the chemical bonds and lone pairs on the heavy atoms were investigated by means of intrinsic bond orbital analysis.


Introduction
Bromine pentafluoride is one of the most reactive halogen fluorides, probably outperformed in its reactivity only by ClF 3 . [1][2][3][4][5] BrF 5 was discovered by Ruff and Menzel in 1931 and described as a colorless liquid that freezes at 211.85 K (À 61.30°C) and boils at 313.65 K (+ 40.50°C). [6] BrF 5 is the highest known binary fluoride of bromine. The existence of the hypothetical BrF 7 molecule has not yet been proven as attempts of a further oxidation of BrF 5 F 11 ]. [7] The conventional synthesis of BrF 5 [6] takes place by reacting fluorine gas with gaseous BrF 3 at 473 K in a platinum or copper vessel [Eq. (1)].
The pale yellow (red if more heavily contaminated by mixtures of other bromine fluorides or Br 2 ) crude BrF 5 is then fractionally distilled for purification. [6] Another synthetic route for the preparation of BrF 5 is the fluorination of KBr, as described by Hyde and Boudakian. [8] Contact with moisture or the use of unsuitable or insufficiently passivated vessel materials will also cause BrF 5 to take on a pale yellow to deep red color. [6] In this case, fluorine can be passed through the liquid at room temperature until all the Br 2 and BrF 3 has reacted. [7,9] After degassing, BrF 5 can be distilled onto dry NaF for storage, that reacts with traces of HF and BrF 3 . [9] Pure BrF 5 is colorless. [6] Compounds such as BrF 5 are hard to come by or simply unobtainable for academic research purposes, therefore we looked for a simple method for their synthesis at room temperature on a laboratory scale. While the photosynthesis of ClF 5 had already been reported in various sources, [10][11][12] to the best of our knowledge no methods are yet known for the photochemical preparation of BrF 5 . However, the photochemical fluorination of bromine and BrF 3 , respectively, to BrF 5 should be straightforward, since the ionization energies of bromine are even lower than those of chlorine.
Due to the lower temperature of the synthesis presented here compared to the traditional routes, no or significantly less wall reactions with the vessel materials take place. Also, the decomposition of BrF 5 at higher temperatures into BrF 3 and F 2 is suppressed. [1] Therefore, BrF 5 is obtained in nearly quantitative yield and high purity.
The crystal structure of solid BrF 5 was determined in 1957 by Burbank and Bensey, [13] but due to the data quality a redetermination of the crystal structure was performed here. We present a modified structure model for HT-BrF 5 , a novel lowtemperature modification LT-BrF 5 , reactions of BrF 5 with KF and RbF, and discuss the effect of the lone pair on the Br atom within the BrF 5 molecule and the [BrF 6 ] À anion in comparison to other species.

Results and Discussion
Scheme 1 provides an overview of the works reported in this manuscript, the photochemical synthesis of BrF 5 , its hightemperature and low-temperature crystal structures, its reactions with some alkali metal fluorides and a discussion of the free valence electron pairs on the Br atoms of BrF 5 molecules and [BrF 6 ] À anions by using intrinsic bond orbitals (IBOs).

Photochemical preparation of bromine pentafluoride
BrF 5 was obtained by irradiation of BrF 3 and F 2 with UV light at room temperature according to Equation (2).
The reaction can be performed either with pure fluorine gas or gas diluted with argon. The use of diluted fluorine does not entail any major disadvantages, apart from the smaller amount of F 2 being present for BrF 5 synthesis at the same overall pressure. Pure BrF 5 could be obtained at irradiation wavelengths between 300 and 400 nm in both cases. The product was characterized by X-ray diffraction, NMR, UV-Vis, Raman and IR spectroscopy and by reaction with alkali metal fluorides.

Selection of UV lamps of suitable wavelength
Various UV lamps of different wavelengths were tested for the photochemical synthesis of BrF 5 . Their emission spectra are shown in Figure 1. BrF 5 is formed independent of the used UV lamp, however, it was not possible to achieve complete conversion when the low-pressure mercury vapor lamp with a peak wavelength of 254 nm was used. Instead, yellowish samples were obtained which additionally contained BrF 3 , as was shown by Raman spectroscopy. The use of the other lamps with emission maxima at longer wavelengths resulted in all cases in pure, colorless BrF 5 . This leads to the conclusion that BrF 5 is photodissociated due to its self-absorption at low wavelength and an equilibrium between BrF 5 and BrF 3 is established. However, as demonstrated by the transmission spectra of BrF 5 and F 2 in Figure 1, their absorption maxima are separated well enough, so that at higher wavelengths fluorine molecules can be photo-dissociated selectively.
NMR spectroscopy 19 F NMR spectra of the as-obtained BrF 5 were recorded at room temperature and at 213 K and are shown in Figure 2. Two Scheme 1. A summary of the works reported here. Preparation of BrF 5 , solid-state structures of its high-temperature (HT) and low-temperature (LT) polymorph, its reactions with the alkali metal fluorides KF and RbF. Photographs of BrF 3 and BrF 5 , as well as the Lewis structures of the BrF 5 molecule and the [BrF 6 ] À anion, sections of the crystal structures, and the lone pairs on the Br atoms using IBOs are shown. Figure 1. UV-Vis transmittance spectra of BrF 5 (red) and F 2 (black) as well as emission spectra of the UV lamps used: Osram Puritec HNS UVÀ C, [14] 254 nm (green), ExoTerra Reptile UVB200 [15] (blue), Nichia NVSU233B, [16] 365 nm (cyan), Osram LuxiGen, 395 nm (magenta). [17] An artifact due to the lamp change of the UV-Vis spectrometer at 320 nm is marked by an asterisk. Data for the emission spectra were extracted from the literature [14][15][16][17] by using the Engauge Digitizer software. [18] signals with an integral ratio of 1 : 4 are present and assigned to the one apical, F ap , and the four equatorial, F eq , fluorine atoms of BrF 5 . Spectra recorded at 273 K or below show a splitting of the two signals into a quintet and a doublet due to 2 J( 19 F, 19 F) coupling. Thus, assignments of the signals with a chemical shift of around 276 ppm to the F ap atom and the one at around 139 ppm to the F eq atoms are evident. The obtained spectra are in agreement with the literature. [19] NMR spectra and data obtained from variable temperature measurements from 213 to 300 K are summarized in Figure S1 and Table S1 in the Supporting Information. NMR chemical shifts calculated at the CCSD(T)/cc-pwCVTZ level of theory agree with the experimental findings (see the Experimental Section for computational details). The chemical shift for the F ap atom was calculated to be at 277 ppm and the one for the F eq atoms at 138 ppm.

Vibrational spectroscopy
IR spectra of gaseous BrF 5 were recorded at room temperature at various pressures ( Figures S6 and S7) in order to resolve strong fundamental vibrations as well as weaker combination modes. A spectrum together with the quantum chemically calculated modes is shown in Figure 3. The quantum-chemical calculation for band assignment of the fundamental and combination modes was performed at the CCSD(T)/cc-pVTZ level of theory.
The Raman spectrum shown in Figure 4 was recorded for liquid BrF 5 . For practical reasons, no IR spectrum could be acquired on liquid BrF 5 . Both the Raman as well as the gasphase IR spectrum are in agreement with those given in the literature. [20,21] The tetragonal-pyramidal BrF 5 molecule (C 4v ) shows twelve normal modes of vibration. Three modes are doubly degenerate, six are non-degenerate. These are assigned into the irreducible representations A 1 , B 1 , B 2 , and E of the point group. All these vibrations are Raman-active, whereas only the A 1 and E vibrations are IR-active. The band assignment for the fundamental vibrations is given in Table 1. Two of the active Raman frequencies, ν 5 and ν 7 , were not observed in the recorded spectrum. Based on the calculated bands, it can be assumed that ν 5 overlaps with ν 9 , and ν 7 most likely coincides with ν 1 , leading to its slightly asymmetric shape. The numbering of the vibrations follows the literature. [22] A more comprehensive band assignment for the IR spectrum including the combination bands is given in Table S4. Raman spectra of liquid and solid BrF 5 at various temperatures are shown in Figures S2-S5, their band assignments in Tables S2  and S3. Calculated Raman spectra of the two crystalline modifications are also available in Figures S8 and S9.  19 F NMR spectra of neat BrF 5 at room temperature (top) and at 213 K (bottom), slightly above the melting point. The assignment of the observed signals to the F atoms is indicated in yellow (F ap ) and orange (F eq ). . IR spectrum of gaseous BrF 5 at ca. 125 mbar (black line). The corresponding calculated and intensity-scaled IR bands are shown as blue lines, of which the strongest were cut at the same arbitrary value. In the region above 1500 cm À 1 and up to 4000 cm À 1 , no additional bands were observed or calculated.

Handling and reactivity
In several cases, especially during our attempts to record IR spectra on BrF 5 , we visually observed a color change of BrF 5 gas from colorless to yellowish-brown, which indicated the decomposition of some traces of BrF 5 . The nature of the yellowishbrown species is still under investigation. Even before the color change is visible to the eye, additional IR bands appear in the region above 3690 cm À 1 corresponding to the rotation-vibration bands of HF. This decomposition of BrF 5 does not occur when all surfaces that come in contact with it are well passivated, thoroughly baked out and as free of moisture as possible. In order to achieve this, steel apparatuses are usually baked out in a hot air bath at about 873 K in vacuo. Because it is technically not possible to bake out our IR gas cell completely, which is due to the limited thermal stress resistance of the optical windows, we suspect traces of water being the reason for the decomposition of BrF 5 . To overcome this issue, we first filled the gas cell with BrF 5 or ClF 3 so that they react with "all" the moisture, then pumped of the volatiles using a vacuum of circa 10 À 3 mbar and finally filled the cell with fresh BrF 5 for the measurement.
When BrF 5 is stored in vessels made out of fluoropolymers like PFA or FEP, HF impurities are also present after some days. Both BrF 5 and moisture dissolve and diffuse through the walls of the fluoropolymer vessel so that the contamination increases over time. That is why vessels made out of passivated stainless steel, nickel or Monel should be preferred for storage and handling. When fluoropolymers are required because of their translucency, the vessels can be baked out in vacuum and then flushed with fluorine several times. This "saturates" the polymer with F 2 and temporarily displaces other contaminants such as H 2 O dissolved in the fluoroplastic.

Crystal structure of HT-BrF 5
The crystal structure of BrF 5 was first published in 1957 by Burbank and Bensey. [13] According to them, BrF 5 crystallizes in the orthorhombic space group Cmc2 1 (36) at 153 K. However, our diffraction data show reflections that clearly violate the extinction condition of the C-centering. These violations were observed not only in a powder diffraction pattern of BrF 5 recorded at 180 K, see below, but also in the single-crystal X-ray diffraction data acquired at temperatures of 200, 180, 150 and 100 K.
Below its melting point of 211.85 K (À 61.3°C), BrF 5 crystallizes in space group Pnma (62) but undergoes a phase transition at about 142 K as shown by variable temperature powder X-ray diffraction, see the Supporting Information. At 130 K we observed the formation of a monoclinic low-temperature (LT) modification, crystallizing in space group P2 1 /c (14). The first frames collected during the X-ray diffraction experiment at this temperature showed still only the orthorhombic polymorph, then both modifications were observed, while the last frames contained only reflections of LT-BrF 5 . A dataset containing only the reflections of LT-BrF 5 was collected at 100 K. Selected crystallographic data and details of the structure determinations are given in Table 2.
We solved and refined the crystal structure of the hightemperature (HT) polymorph in space group Pnma (62). Attempts to refine the crystal structure in the non-centrosymmetric subgroup Pna2 1 also lead to the essentially same structure model, however the Flack parameter of 0.46 (7), the ADDSYM [23,24] algorithm implemented in the PLATON [25] software, as well as strong correlation of atomic coordinates clearly suggested the centrosymmetric space group Pnma to be the superior choice.
The following description of the crystal structure refers to the diffraction data collected at 150 K, allowing for a direct comparison with the previous structure model. In the solid state, BrF 5 adopts a square pyramidal molecular structure ( Figure 5), as already indicated by NMR and vibrational spectroscopy for the liquid and gas phase. The bromine atom resides on Wyckoff position 4c (.m.) and is surrounded by the F ap (1) atom (4c, .m.), the F eq atoms F(2) and F(3) on positions 4c (.m.), and the F eq (4) atom on the 8d (1) position. The Br atom is not located within the base of the pyramid, but lies with 0.1726(18) Å outside a least-squares plane spanned by the four F eq atoms, showing the putative space requirement of the lone pair of the Br atom. The bond angles between the apical and equatorial fluorine atoms F ap À BrÀ F eq are therefore smaller than 90°and are 83.75(11)°, 2 × 84.16(10)°, and 85.36 (14)°. The same phenomenon is visible in the previously reported structure model, where the distance between the Br atom and the base of the pyramid was reported with 0.1734 Å (with no esds given in literature). [13] The respective F ap À BrÀ F eq angles diverge much more with 80.4°, 2 × 85.4°, and 86.5°. The BrÀ F eq bonds range from 1.741(2) to 1.777(2) Å and are therefore slightly longer than the BrÀ F ap bond with 1.678(2) Å. The BrÀ F eq bond lengths of the previously reported structure show a larger deviation from the mean value with 1.7460 to 1.8206 Å, which can be attributed to the choice of the space group. However, the reported BrÀ F ap distance agrees with 1.6784 Å with the value redetermined here.
An intermolecular interaction is present between the Br atoms and F eq atoms of neighboring BrF 5 molecules with a Br···F eq distance of 2.880(2) Å. As expected, the F eq atom involved in the intermolecular interaction is the one with the longest intramolecular BrÀ F eq distance of 1.777(2) Å. The planes spanned by the four equatorial fluorine atoms of each of two adjacent BrF 5 molecules intersect at an angle of 106.42(6)°. As a result of this intermolecular interaction, flat, zigzag-like chains are formed running parallel to the a-axis. The crystal structure of BrF 5 can therefore be described with the Niggli formula To the best of our knowledge, the crystal structure of BrF 5 represents a new structure type. The formation of oligomeric molecular structures, as known for other pentafluorides such as MoF 5 (mS48), that is rings of Mo 4 F 20 , RuF 5 (mP48), differently shaped Ru 4 F 20 rings, or AuF 5 (oP48), Au 2 F 10 dimers, does not occur. [26][27][28] Similarity exists between the structures of BrF 5 (oP24) and BrF 3 (oS16) . [29] As BrF 5 , BrF 3 also forms one-dimensional infinite strands with a flat zigzag-like shape in its solid state. Thus, the bromine atoms are coordinated in a kite shape by F atoms and have a coordination number of 3 + 1. The intramolecular BrÀ F distances in the BrF 3 structure are in between 1.71(1) and 1.888(9) Å and therefore are in average longer than in the BrF 5 molecule, as expected for bromine in oxidation state + III. The F atom with the longest

Chemistry-A European Journal
Research Article doi.org/10.1002/chem.202202466 intramolecular BrÀ F bond of 1.888(9) Å is the one that makes intermolecular contact with the neighboring BrF 3 molecule. The intermolecular Br···F distance is 2.451(12) Å, and thus, more than 0.4 Å shorter than in the structure of BrF 5 . The shorter intermolecular Br···F distance is likely attributed to steric reasons.
Looking at the arrangement of the Br atoms within the crystal structure of BrF 5 , a structural relation to the Cu type becomes obvious and a Bärnighausen-tree is given in Figure S10. A Br atom is surrounded by 12 other Br atoms in the shape of a distorted cuboctahedron ( Figure 6). The distortion that occurs is due to the formation of the chain motif adopted by the BrF 5 molecules. This leads to a deviation from the pseudo-face centering of the bromine atoms compared to the atom positions in the Cu type. In the setting chosen by Burbank and Bensey, this centering corresponds to the C-centering in space group Cmc2 1 .
The structure model proposed here in space group Pnma is reasonable, only the comparatively large displacement ellipsoids, especially of the F eq atoms, may be unexpected. In order to examine whether these are physically meaningful and to be able to exclude that artificially large displacement parameters are obtained from a flawed structure model or an incorrect absorption correction, we recorded single-crystal diffraction data at different temperatures. The obtained equivalent displacement parameters of the atoms were then plotted against the temperature (Figure 7). We observe that the extrapolated displacement parameters intersect the coordinate origin at 0 K within their standard uncertainties, implying that the strong atomic displacement is due to thermal effects indicating a "soft" molecular packing with weak intermolecular interactions.
However, the powder X-ray diffraction pattern of BrF 5 recorded at 180 K ( Figure 8) shows that the compound is diffracting rather well as sharp reflections are observed up to approximately 70°2θ which indicates a proper three-dimensional long-range order of the BrF 5 molecules within the crystal structure and that the thermal vibrations of the atoms are not severe at all.
All reflections could be assigned to the high-temperature modification of BrF 5 that crystallizes in space group Pnma. Space group Cmc2 1 , reported for the previous structure model, [13] can be completely ruled out as several distinct  The displacement parameters observed at 100 K correspond to the structure of LT-BrF 5 and are therefore not directly comparable to those of HT-BrF 5 . However, as the phase transition from HT-to LT-BrF 5 has no significant effect on the displacement parameters, we decided to use the data obtained at 100 K for extrapolation anyway.  Table S6.

LT-BrF 5
The phase transition from HT-to LT-BrF 5 occurs at about 142 K, as shown by variable temperature powder X-ray diffraction ( Figure S11). The single-crystal structure was recorded at 100 K. The shape of the BrF 5 molecules as well as their interconnection to one-dimensional infinite strands in the low-temperature modification is similar compared to the high-temperature polymorph. The BrÀ F eq bond lengths range from 1.744(3) to 1.779(3) Å and are therefore identical within the standard uncertainties in comparison to the high-temperature modification. The same applies to the BrÀ F ap bond length measuring 1.686(2) Å and to the distances of the intermolecular Br···F eq contacts measuring 2.881(3) Å. The Br atom is located 0.1707(16) Å below the least-squares plane defined by the F eq atoms. Thus, within the standard uncertainty this distance is also identical compared to the HT structure. The major structural difference between the two polymorphs is that in LT-BrF 5 , together with the deviation of the monoclinic angle from 90°, the orientation of the BrF 5 molecules within the zigzagstrands also changes ( Figure 9). The intrachain Br···Br distance shortens from 4.6555(3) Å in HT-BrF 5 to 4.6447(4) Å in LT-BrF 5 . The angle by which the planes spanned by the F eq atoms of two neighboring BrF 5 molecules intersect decreases from 106.42(6)°i n HT-BrF 5 to 103.17 (7)°in LT-BrF 5 . The shortest distance of two Br atoms in adjacent strands decreases from 4.8257(1) Å in HT-BrF 5 to 4.6489(4) Å in LT-BrF 5 . The group-subgroup relation of HT-and LT-BrF 5 is described in Figure 9 by means of a Bärnighausen tree. [30] Solid-state quantum-chemical calculations for BrF 5 Quantum-chemical calculations on BrF 5 with dispersion-corrected hybrid density functional method (DFT-PBE0-D3/TZVP), accounting for the weak intermolecular interactions, aided in the discovery of the low-temperature modification. Calculations on the high-temperature polymorph of BrF 5 in space group Pnma showed a number of imaginary frequencies at 0 K, the largest being 33i cm À 1 . Following the largest imaginary mode allowed to predict the correct space group P2 1 /c and Wyckoff sequence of the low-temperature modification of BrF 5 even before the single-crystal structure determination. A geometry optimization starting from the single-crystal X-ray structure of LT-BrF 5 (P2 1 /c) and a harmonic frequency calculation resulted in a small imaginary frequency of 5.3i cm À 1 , which could also arise due to accuracy limitations in the numerical integration of the DFT exchange-correlation functional. Following this mode would lead to space group Pc, but the energy is lowered only by 0.1 kJ mol À 1 per formula unit compared to P2 1 /c. The intermolecular interactions in the crystal structure of LT-BrF 5 were underestimated without the empirical dispersion correction, resulting in poor prediction of the BrÀ F bond lengths, FÀ BrÀ F angles, the crystal structure and the lattice parameters. Without the dispersion correction, the Br···F eq contacts, responsible for the formation of one-dimensional strands, are with 3.106 Å significantly longer than in the experimentally determined crystal structure with 2.881(3) Å, and differ only slightly from the BrÀ F eq distances between the different strands, so that the motive of isolated strands was no longer discernible. The errors in lattice parameters with and without dispersion correction are shown in Table 3. Dispersion correction reduced the error in the lattice parameters to 1-2 %, and the strand motive was correctly described with the shortest intermolecular Br···F eq distance being 2.946 Å compared to 2.881(3) Å in LT-BrF 5 .
The energetics of the two polymorphs of BrF 5 were further compared with the previously reported Cmc2 1 structure [13] and a hypothetical modification in the dimeric structure of AuF 5 . [28] The previously reported Cmc2 1 structure showed two imaginary frequencies of 33i and 32i cm À 1 , and was 2.9 kJ mol À 1 per formula unit higher in energy compared to LT-BrF 5 . For a hypothetical dimeric AuF 5 structure (Au 2 F 10 ), imaginary frequencies of over 300i cm À 1 were present and the structure was 76 kJ mol À 1 per formula unit higher in energy compared to LT-BrF 5 , indicating that BrF 5 is energetically very unlikely to adopt a

Chemistry-A European Journal
Research Article doi.org/10.1002/chem.202202466 dimeric molecular structure like AuF 5 . All optimized crystal structures are given in CIF format in the Supporting Information.
The optimized crystal structures and their energetics obtained from the quantum-chemical calculations agree with the experimental results when a dispersion-corrected DFT method is used. DFT calculations show that the LT-as well as the HT-BrF 5 crystal structures are energetically favored over the previously reported structural model and the phase transition from HT-to LT-BrF 5 could be followed by studying the imaginary vibrational modes obtained for HT-BrF 5 .

Reactions of BrF 5 with AF (A = K, Rb)
By the reaction of BrF 5 with the alkali metal fluorides AF (A = K, Rb), the corresponding alkali metal hexafluoridobromates(V), A[BrF 6 ], are formed [Eq. (3)]. [31,32] BrF 5 þ AF ! A½BrF 6 � ðA ¼ K, RbÞ While the crystal structure of Cs[BrF 6 ] is already known, [33] the crystal structures of K[BrF 6 ] and Rb[BrF 6 ] have not been reported. Suitable crystals of these compounds were obtained by slow cooling of a saturated solution of the respective alkali metal fluoride in BrF 5 warmed to about 373 K in an air bath.  6 ]. The compounds have already been studied by Bougon and co-workers using powder X-ray diffraction methods, [32] and two possible space groups R � 3 and R � 3m have been proposed. Accordingly, two possible structure types were considered, the K[AsF 6 ] type in R � 3 and the Ba[SiF 6 ] type in R � 3m. We can clearly exclude space group R � 3m as a refinement of the crystal structures in that space group leads to a striking elongation of the displacement ellipsoids of the fluorine atoms perpendicular to the m mirror plane. The crystal structure of the compounds in space group R � 3 is shown in Figure 10 using K[BrF 6 ] as an example. Details of the structure solutions and refinements are given in Table 4 and the atomic coordinates and isotropic displacement parameters are given in Table 5.

Crystal structures of the compounds A[BrF
The bromine atoms reside on position 3a ( � 3:) and are coordinated octahedron-like by six symmetry-equivalent fluorine atoms (18 f, 1) forming the [BrF 6 ] À anion with point group symmetry S 6 . As already discussed by Seppelt and co-workers for the crystal structure of Cs[BrF 6 ], the lone pair of the Br atom within the [BrF 6 ] À anion seems to show no stereochemical activity. [33] For a discussion on the lone pair, see below. The alkali metal cations (3b, � 3.) are 12-fold coordinated by fluorine atoms in a cuboctahedron-like environment.
Seppelt and co-workers observed a BrÀ F bond length within the [BrF 6 ] À anions of Cs[BrF 6 ] of 1.854(1) Å by X-ray diffraction on single crystals and 1.847(1) Å by powder neutron diffraction. [33] We found 1.8637 (12)  While the observed BrÀ F bond lengths agree within the tripled standard uncertainties for the Rb and Cs compounds, the BrÀ F bond length is slightly longer in the determined crystal structure of the K compound. The observed bond lengths agree with those quantum chemically calculated for the solids and the BrÀ F bonds are longest in the gas-phase [BrF 6 ] À anion, as may be expected. [a] For HT-BrF 5 , quantum-chemical calculations with DFT-PBE0-D3 were also performed and LT-BrF 5 was determined to be 1.7 kJ mol À 1 per formula unit lower in energy compared to HT-BrF 5 (calculations at 0 K). The structural motive is also well described by the calculation with errors between 0.9 and 1.4 % for the BrÀ F bond lengths and an error of 2.50 % for the intermolecular Br···F distance.   6 ] are 91.77°and 91.30°, respectively. Therefore, observed and quantum chemically calculated values agree. Based on these findings we conclude that the influence of the K + cations on the molecular structure of the [BrF 6 ] À anions is seemingly the strongest and weakest for the Cs + cations, as might have been expected on the basis of Pearson's concept of hard and soft acids and bases. [34]

Raman spectroscopic investigation of the compounds A[BrF 6 ] (A = K, Rb)
The bulk phases of the samples were analyzed by Raman spectroscopy. For both, K[BrF 6 ] and Rb[BrF 6 ], three bands were observed corresponding to the Raman active vibrations of the octahedron-like [BrF 6 ] À anion. The Raman spectra are shown in Figure 11 and the band assignments are given in Table 6.
The observed Raman bands agree with the ones reported in the literature [32] where the assignment of bands had been carried out using O h symmetry, while we used the crystallographic site symmetry of S 6 . We also see an agreement with the calculated Raman spectra (Figure 11). The calculated bands at 246 cm À 1 for K[BrF 6 ] and at 245 cm À 1 for Rb[BrF 6 ] are likely not  Figure 11. Raman spectra of K[BrF 6 ] (left) and Rb[BrF 6 ] (right) at room temperature recorded with an excitation laser wavelength of 532 nm, calculated spectra shown in red. No additional bands were present in the region above 600 cm À 1 and up to 4000 cm À 1 .

Chemistry-A European Journal
Research Article doi.org/10.1002/chem.202202466 observed due to their small intensities in comparison to the other bands or due to band overlap.

Lone pair effects
Canonical molecular orbitals from quantum-chemical calculations are typically rather difficult to interpret when chemical bonds between two or a few atoms are considered. Many chemists like to think of localized chemical bonds, draw molecular structures with Lewis formulas, and predict the shapes of molecules with simple models such as the VSEPR theory. [35][36][37] These however may bias judgment: For example, the action of BrF 5 as a Lewis acid might be unexpected as one may think that the free valence electron pair, the lone pair, on the Br atom of the BrF 5 molecule should repel an incoming F À anion and therefore no [BrF 6 ] À anion would form based on this over-simplified view. However, BrF 5 acts as a Lewis acid under formation of [BrF 6 ] À anions according to Scheme 2, for examples see above and the literature. [31,33,38] Previous gas-phase DFTÀ B3LYP calculations suggest the reaction in Scheme 2 to be exoenergetic by over 300 kJ mol À 1 . [39] We obtained a gas-phase reaction energy of À 299 kJ mol À 1 at the CCSD(T)/cc-pVTZ level of theory. Furthermore, we estimated the Fluoride Ion Affinity [40][41][42][43][44] (FIA) of BrF 5 to be 276 kJ mol À 1 at the same level of theory (COF 2 as reference system). With that it is comparable to PCl 3 and PBr 3 , it is stronger than AsF 3 and SF 4 , weaker than SbF 3 , SiF 4 , BF 3 and much weaker than AsF 5 , SbF 5 or Sb(OTf) 5 . [43] In order to describe the chemical bonds of the BrF 5 and [BrF 6 ] À molecules, population analyses were carried out at the DFT-PBE0/TZVP level of theory by using intrinsic atomic orbitals (IAOs) and the bonding was analyzed with the aid of IBOs. [45,46] For the gas-phase BrF 5 molecule, a local minimum is obtained in point group C 4v , as may be expected. The Br atom shows a positive partial charge of + 2.36 e À , while the four F eq atoms show negative partial charges of À 0.49 e À and the apical fluorine atom F ap a smaller charge of À 0.40 e À . The high positive partial charge on the Br atom is surely also a reason why BrF 5 reacts with a fluoride anion. Figure 12 shows selected IBOs for the BrF 5 molecule. As expected, the BrÀ F bonds are strongly polarized covalent. The BrÀ F eq bonds are essentially plike with a small 4 s contribution of the Br atom of 3 %. The BrÀ F ap bond is slightly less polarized in comparison and the Br 4 s orbital contributes with 19 % to the bond. The shape of the lone pair on the Br atom is essentially s orbital-like with only 18 % contribution from the 4p z orbital.
Quantum-chemical calculations lead to a true local minimum for the [BrF 6 ] À anion in O h symmetry, that is, an "ideal" octahedron is obtained with no signs of stereochemical activity of the Br lone pair. The Br atom shows a positive partial charge of + 2.32 e À which is by 0.04 e À lower compared to the Br atom within the BrF 5 molecule. The F atoms show negative partial charges of À 0.55 e À . Figure 13 shows selected IBOs of the [BrF 6 ] À anion.
The BrÀ F bonds in the [BrF 6 ] À anion become even more polarized compared to the BrF 5 molecule and the Br atom no longer contributes with its 4 s orbital to the BrÀ F bonds but only with its p orbitals. The lone pair of the Br atom is completely 4 s-like and therefore shows the absence of any stereochemical activity.
We calculated the [BrF 6 ] + cation, where there is no Br lone pair, for comparison and found a true local minimum for O h symmetry, as expected from the literature. [7,47] The calculated BrÀ F bond length is 1.67 Å which agrees with estimated  6 ] À anion because of the presumed orientation and shape of the lone pair.

Chemistry-A European Journal
Research Article doi.org/10.1002/chem.202202466 1.69 Å [33] based on its reported force constants. [47] Its IBOs of the BrÀ F bonds are shown in Figure 14.
In The stereochemical inactivity of the lone pair of the Br atom within the [BrF 6 ] À anion had been discussed previously by different researchers. Christe and co-workers reasoned wellfounded that "in a rigid molecule the space requirement of a sterically active free valence electron pair slightly exceeds that of a fluorine ligand", [48] and that therefore anions such as [IF 6 ] À are not octahedral as there is enough space around the I atom for the lone pair to become active because I atoms can adopt coordination numbers up to eight with F ligands, for example, in [IF 8 ] À anions. "In [BrF 6 ] À there is little or no room left for a seventh ligand and, therefore, the free valence electron pair should be sterically inactive" they concluded. [38] Seppelt and coworkers had the opinion that the 4 s electrons were particularly strongly bound to the Br atomic nucleus as the fully occupied 3d shell shields the nuclear charge incompletely. Therefore, the BrÀ F bonds would mainly show p character. [33] Gillespie and coworkers reasoned that the six F atoms around the small Br atom are close packed. Therefore, no room for the lone pair is left and it belongs to the Br core. [36,49] A quantum-chemical calculation of the [IF 6 ] À anion shows that its molecular structure is not a local minimum in point group O h . Following the imaginary mode of 39i cm À 1 leads to molecular structures in point groups C 2v and C s that are also no local minima. Finally, a true local minimum in point group C 1 is reached, and overall the energy lowered by 14 kJ mol À 1 compared to the ideal O h symmetry. The IBOs of the I lone pair and the IÀ F bonds of the [IF 6 ] À anion in its energy minimum are shown in Figure 15.
The I lone pair is essentially 5 s-orbital-like with 93 % 5 s and 7 % 5p x orbital contribution. It therefore shows some stereochemical activity in contrast to the Br lone pair of the [BrF 6 ] À anion. Both reasons for the stereochemical inactivity of the Br lone pair of the [BrF 6 ] À anion given by Seppelt or Christe are valid and in agreement with our quantum chemical findings.    6 ] À anion in its energy minimum in point group C 1 . Percentages next to the IBOs indicate the contribution of each atom to that IBO. The larger a percentage, the more polarized the covalent bond. The percentages in the labels show the contributions of atomic orbitals to each IBO. F atoms are in yellow, I atoms in violet. IBO isosurfaces are drawn so that 80 % of the density is enclosed within them.

Conclusions
BrF 5 was synthesized at room temperature by the photochemical reaction of BrF 3 with F 2 . UV-Vis spectra of BrF 5 and F 2 were recorded to identify the region of suitable wavelength, 300 to 400 nm, to photodissociate F 2 but not the product BrF 5 . BrF 5 can be obtained in yields above 90 % with respect to the starting material BrF 3 and pure on the basis of NMR, IR and Raman spectroscopy. NMR spectra of BrF 5 were taken at various temperatures. Quantum-chemical calculations aided in the determination of the new low-temperature modification and allowed the band assignment of the recorded IR and Raman spectra. By treating BrF 5 with KF and RbF, crystals of the compounds K[BrF 6 ] and Rb[BrF 6 ], respectively, were obtained and their crystal structures determined. The chemical bonds and the lone pairs on the Br atoms within the molecules [BrF 6 ] + , BrF 5 , and [BrF 6 ] À were investigated by using intrinsic bond orbitals. These show that the contribution of the 4 s orbitals of the Br atoms to the BrÀ F bonds decreases to zero in the octahedral [BrF 6 ] À anion where the Br lone pair is purely 4 sorbital-like. In the [IF 6 ] À molecule, which is not octahedral, the 5p orbitals contribute to the I lone pair, making it stereochemically active.

Experimental Section
General: All operations were performed on a Monel metal Schlenk line, which was passivated with fluorine and ClF 3 at various temperatures and pressures before use. The alkali metal fluorides were purchased from Merck, purified according to literature procedures, [50] and were stored in PTFE vials in an Ar-filled glove box (MBraun).
Reaction vessels were made out of fluoropolymer (perfluoroalkoxy alkanes, PFA or perfluorinated ethylene propylene copolymer, FEP) and sealed with a bellows valve made out of Monel or stainless steel. The vessels were baked out in vacuum (~10 À 3 mbar) at circa 393 K for several times and then filled up to a pressure of 4 bar with diluted F 2 (F 2 /Ar 20 : 80, v/v) for 16 h in order to saturate the polymer with fluorine.
CAUTION! Fluorine, the halogen fluorides and fluoridobromates(V) must be handled with appropriate protective gear with ready access to proper emergency treatment procedures in the event of contact. The aforementioned are potent oxidative fluorinators that are only stable under the rigorously anhydrous conditions employed in the experimental procedures outlined in the Experimental Section. They react vigorously to explosively upon hydrolysis or contact with organic materials. The utmost precautions must be taken when disposing of these materials and their derivatives.

Preparation of BrF 5 in PFA vessels:
When only small amounts of BrF 5 around 1 g are required, the reaction can be performed in a reaction vessel with a volume of 50 to 80 mL made out of PFA or FEP and equipped with a stainless-steel valve.
For a typical synthesis 692 mg (5.08 mmol) BrF 3 was loaded into the reaction vessel (V = 74 mL) and diluted fluorine (F 2 /Ar 20 : 80, v/v, 3 bar, 1.79 mmol) was added. After irradiating the reaction mixture with a UV LED (Osram, LuxiGen-UV395, 395 nm, 1380 mW) for 6 h, the reaction vessel was cooled to 77 K and all volatiles were pumped off. The vessel was then allowed to warm to room temperature and diluted fluorine (3 bar, 1.79 mmol) was added again. This process was repeated for three times. After the third addition of F 2 a slight excess (5.37 mmol) with respect to Br was reached. The liquid completely decolorized upon irradiation, which can be attributed to the absence of BrF 3 . Bromine pentafluoride (792 mg, 4.55 mmol, 90 % with respect to the starting material BrF 3 ) was obtained as a colorless liquid. As BrF 5 , like ClF 5 , [5] tends to dissolve in the PFA/FEP vessel wall over time, the yield decreases with longer reaction times. Therefore, for larger batches requiring longer reaction times, vessels made of PFA/FEP should be avoided if possible.
Preparation of BrF 5 in a stainless-steel reactor: For the preparation of larger amounts of circa 60 g BrF 5 per batch, a stainless-steel (1.4571) reactor ( Figure 16) with sapphire windows was used. For a typical synthesis 48.46 g (354 mmol) BrF 3 was loaded into the reactor (V = 924 mL) and undiluted fluorine (1 bar, 42.7 mmol) was added. The reaction mixture was irradiated with a UV LED (Nichia, NVSU233B, 365 nm, 1450 mW) for 12 h during which the pressure dropped to circa 400 mbar. The F 2 pressure was then again increased to 1 bar and the process was repeated until the reaction mixture completely decolorized after 9.5 d of irradiation. Bromine pentafluoride (59.11 g, 338 mmol, 95 % with respect to BrF 3 ) was obtained as a colorless liquid.

Preparation of A[BrF 6 ] (A = K, Rb):
A PFA reaction vessel treated as described above was charged in a glove box with 40 mg of the respective dry alkali metal fluoride. The reaction vessel was then cooled to 77 K and an excess of BrF 5 (0.1 mL, 246 mg, 1.4 mmol) was condensed. The reaction mixture was then allowed to warm to room temperature and diluted fluorine was added to a pressure of 3 bar to counteract diffusion of external moisture and air. The reaction mixture was heated to circa 373 K in a hot air bath for 1 h. By slowly cooling the mixture down to room temperature, the product was obtained in form of colorless, needle shaped crystals. After a few crystals were transferred from the BrF 5 solution directly into the perfluorinated oil for the single-crystal preparation, BrF 5 was removed from the reaction vessel under reduced pressure and the remaining solid was heated to 373 K in vacuum for 1 h in order to release any adsorbed BrF 5 . We were not able to obtain the compounds phase-pure this way, there seems to be still too much diffusion through the PFA vessels because the HF 2 À salts were formed in varying amounts during the syntheses. NMR spectroscopy: 19 F NMR spectra were recorded using a Bruker Avance III HD 300 NMR spectrometer. CFCl 3 was used as an external standard. A sample of neat BrF 5 was distilled into a thoroughly baked out 3 mm diameter FEP tube and sealed under vacuum. The sample was then placed into a regular glass NMR tube (5 mm) and stored under argon at 237 K until assayed.

Raman spectroscopy:
The Raman spectra were measured with a Monovista CRS + confocal Raman microscope (Spectroscopy & Imaging GmbH) using a 532 nm solid-state laser and either a 300 grooves/mm (low-resolution mode, FWHM: < 4.62 cm À 1 ) or an 1800 grooves/mm (high-resolution mode, FWHM: < 0.368 cm À 1 ) grating. Sample preparation of BrF 5 : A silica capillary was baked out under vacuum and flushed with diluted fluorine for several times. BrF 5 was distilled at liquid nitrogen temperature into the capillary which was then flame-sealed under vacuum. The capillary was allowed to warm to room temperature and was placed under the Raman microscope for data acquisition. Sample preparation of A[BrF 6 ] (A = K, Rb): Samples were filled and sealed in quartz capillaries inside the glovebox and were then placed under the Raman microscope for data acquisition.
Infrared spectroscopy: The gas-phase IR spectrum of BrF 5 was recorded on a Bruker Tensor 37 FTIR with a resolution of 4 cm À 1 using a passivated measuring cell manufactured from 316 L stainless steel, equipped with diamond or BaF 2 windows. IR spectra of solids were recorded on a Bruker alpha FTIR spectrometer using the ATR Diamond module with a resolution of 4 cm À 1 . The spectrometer was located inside a glovebox (MBraun) under argon atmosphere. The spectra were processed with the OPUS software package. [51] Single-crystal X-ray diffraction: A crystal of K[BrF 6 ] or Rb[BrF 6 ], respectively, was selected under pre-dried perfluorinated oil (Fomblin YR-1800) and mounted using a MiTeGen loop. Intensity data of a suitable crystal were recorded with an IPDS 2 diffractometer (Stoe & Cie). The diffractometer was operated with Mo Kα radiation (0.71073 Å, graphite monochromator) and equipped with an image plate detector. In case of K[BrF 6 ], split reflections were observed for several crystals, which indicated non-merohedral twinning. Evaluation, integration and reduction of the diffraction data was carried out using the X-Area software suite. [52] A numerical absorption correction was applied with the modules X-Shape and X-Red32 of the X-Area software suite. The structures were solved with dual-space methods (SHELXT-2014/5) and refined against F 2 (SHELXL-2018/3). [53,54] For the structure solution of K[BrF 6 ] only the non-overlapping reflections of the major twin component were used. The data were initially refined with the HKLF5 format option in SHELXL-2018/3 with all reflections (overlapping reflections and non-overlapping reflections of three twin components). [54] The data were then processed with the HKLF5Tools [55] program: the non-overlapping reflections of the weaker diffracting twin components were removed, the non-overlapping reflections of the major twin component were merged in point group � 3 and the overlapping reflections were merged in point group 1. The final refinement was carried out against the detwinned dataset (created by SHELXL with the LIST 8 option as an FCF file and converted to an HKL file with HKLF5Tools).
For the determination of the crystal structure of BrF 5 , a sample of the latter was prepared in a 0.3 mm quartz capillary as described for its Raman spectroscopy and mounted on a D8 Quest diffractometer (Bruker). The diffractometer was operated with monochromatized MoÀ Kα radiation (0.71073 Å, multi layered optics) and equipped with a PHOTON 100 CMOS detector. By cooling the sample to 200 K on the diffractometer, a polycrystalline material was obtained, as shown by the diffraction pattern in which Debye-Scherrer rings are indicated. By repeated zone melting of the sample in the capillary with the help of a copper wire, a single crystal suitable for X-ray structure determination was obtained. Full datasets of the crystal were measured at 200, 180, 150 and 100 K respectively. Evaluation, integration, and reduction of the diffraction data was carried out within the APEX3 software suite. [56] The data was corrected for absorption utilizing the multi-scan method of SADABS [57] within the APEX3 software suite, the structure was solved with dual-space methods (SHELXT-2014/5) and refined against F 2 (SHELXL-2018/3). [53,54] When the crystal was cooled below about 130 K, split reflections were observed in the diffraction pattern, indicating that the crystal was shattered by a phase transition. In consequence, the diffraction data recorded at 100 K is not from a single crystal but contains reflections from multiple crystallites. Using the CELL_NOW [58] indexing algorithm more than five domains with identical lattice parameters were found, of which only the two strongest were used for further processing. Only the non-overlapping reflections of the major twin component were used for the structure solution. The structure was solved with dualspace methods (SHELXT). [53] The data were initially refined with the HKLF5 format option in SHELXL with all reflections (overlapping reflections and non-overlapping reflections of the three twin components). [54] The data were then processed with the HKLF5Tools [55] program: the non-overlapping reflections of the weaker diffracting twin components were removed, the nonoverlapping reflections of the major twin component were merged in point group 2/m and the overlapping reflections were merged in point group 1. The final refinement was carried out against the detwinned dataset (created by SHELXL with the LIST 8 option as a FCF file and converted to a HKL file with HKLF5Tools).
In the datasets recorded at 150, 180 and 200 K, several additional weak reflections occur at half the diffraction angle of particularly strong reflections. This can be attributed to the minor lambda-half fraction of the primary X-ray beam that is not filtered out by the monochromator and gets diffracted due to the big crystal size with significant intensity. As no subsequent correction was made, the goodness of fit values are somewhat higher than might be expected.
All atoms were refined with anisotropic displacement parameters. Representations of the crystal structures were created with the Diamond software. [59] Deposition Numbers 2193268 (for K[BrF 6 ]), 2193269 (for HT-BrF 5 at 200 K), 2193270 (for HT-BrF 5 at 150 K), 2193271 (for LT-BrF 5 ), 2193272 (for HT-BrF 5 at 180 K), 2193273 (for Rb[BrF 6 ]) contain the supplementary crystallographic data for this paper. These data are provided free of charge by the joint Cambridge Crystallographic Data Centre and Fachinformationszentrum Karlsruhe Access Structures service.
Powder X-ray diffraction: A sample of BrF 5 was prepared in a 0.3 mm quartz capillary as described for Raman spectroscopy and single-crystal X-ray diffraction. The powder X-ray pattern was recorded with a StadiMP diffractometer (Stoe & Cie) in Debye-Scherrer geometry. The diffractometer was operated with Cu Ka 1 radiation (1.5406 Å, germanium monochromator) and equipped with a MYTHEN 1 K detector. The sample was cooled to 180 K using a capillary cooling system (Osford Cryosystems). Diffraction data of the resulting polycrystalline material were collected in eight subsequent runs in a range of 15 to 71°2θ with an irradiation time of 80 s per 1°. Due to ice formation on the surface of the capillary during the measurement only the first tree runs were used for the refinement.
Rietveld refinements [60] were performed using the TOPAS-Academic software (version 7). [61] The structure model derived from the X-ray structure analysis of the single crystal was used as the starting point for the refinement. A shifted Chebyshev polynomial with 12 terms was used to describe the background, the peak profiles were fitted with a modified Thompson-Cox-Hastings pseudo-Voigt ("TCHZ") function as implemented in TOPAS, and the zero offset was refined. To account for absorption, an intensity correction for cylindrical samples was applied as implemented in TOPAS. An eighth-order spherical harmonics function was used to account for the preferred orientation of the crystallites. The strong preferred orientation of the sample can be explained by the fact that the liquid BrF 5 was frozen directly inside of the capillary, making it impossible to grind the sample before measurement. Therefore, various different directions of crystal growth are to be expected. The same phenomenon was observed when trying to obtain single crystals of the compound. In this case, the preferred orientation was also clearly visible in the diffraction image ( Figure 17).
The approach of Le Bail and Jouanneaux [62] was used to accurately describe the peak half-width and shape anisotropy effects, and the corresponding parameters could be freely refined. The final refinement cycles converged with free refinement of all background, profile, and lattice parameters, including the coordinates and isotropic displacement parameters of all atoms.
Computational details: We carried out anharmonic force field and 19 F NMR chemical shift calculations on the gas-phase BrF 5 molecule with CFOUR v2.1 [63,64] program suite. CCSD(T) coupled cluster method was used in combination with correlation-consistent polarized triple-zeta basis sets. For geometry optimizations and anharmonic frequency calculations, we used triple-valence correlation consistent basis set (cc-pVTZ) [65,66] The structure was fully optimized within the C 4v point group (wavefunction was obtained for the highest Abelian subgroup C 2v ). The geometry of CFCl 3 was also optimized and it was used as the NMR reference species for 19 F NMR chemical shifts. The 19 F NMR chemical shift calculations were carried out at the optimized geometries with weighted corevalence basis sets (cc-pwCVTZ). [67,68] Anharmonic vibrational spectra were calculated within the second-order vibrational perturbation theory (VPT2). [69] For evaluating the Fluoride Ion Affinity (FIA) of BrF 5 , we optimized [BrF 6 ] À , COF 2 , and COF 3 À at the CCSD(T)/cc-pVTZ level of theory and used the COF 2 reference system with an experimental FIA of 208.08 kJ mol À 1 to estimate the FIA at an absolute scale. [43] The Z-matrices of the optimized molecular structures are included in the Supporting Information.
All solid-state calculations were performed with the CRYSTAL17 [70] program suite. HT-and LT-BrF 5 , Rb[BrF 6 ], and K[BrF 6 ] were investigated with hybrid density functional methods (DFT-PBE0) [71,72] combined with triple-valence basis sets TZVP for Br [73] and F, [74] and a split-valence basis set SVP for K [75] and Rb [76] (derived from molecular Karlsruhe basis sets [77] ). In case of the molecular crystals HT-and LT-BrF 5 , empirical D3 dispersion correction with zero damping [78,79] was applied for the geometry optimizations and frequency calculations to take into account weak intermolecular interactions. Atomic positions and lattice parameters were fully optimized within the space group symmetry of each system. The reciprocal space was sampled with Monkhorst-Packtype k-point grids. [80] The employed k-point meshes are reported in Table S5. Tight truncation criteria (TOLINTEG 8,8,8,8,16) were applied for the evaluation of the bielectronic Coulomb and exchange series in all calculations. Default DFT integration grids and optimization convergence thresholds were used in all calculations. Harmonic vibrational frequencies and Raman intensities were calculated with the schemes implemented in CRYSTAL. [81][82][83] Raman intensities were calculated for a polycrystalline powder sample. For Rb[BrF 6 ] and K[BrF 6 ] the experimental setup of T = 298.15 K and λ = 532 nm was considered. For the simulation of the spectra a pseudo-Voigt band profile (50 : 50 Lorentzian/Gaussian) with a FWHM of 8 cm À 1 was used. Assignment of the vibrational bands was done with the visualization tool CRYSPLOT. [84] The atomic positions and lattice parameters of the optimized structures are given in the Supporting Information.