Experimental and Computational 77Se NMR Spectroscopic Study on Selenaborane Cluster Compounds

Calculated and measured 77Se nuclear magnetic resonance (NMR) chemical shift data on a diverse collection of 13 selenaborane cluster compounds, containing a total of 19 selenium centers, reveals a correlation between chemical shifts and the intracluster coordination of selenium atoms within their borane frameworks. A plot of the measured against calculated 77Se NMR chemical shifts shows an approximately linear relationship that can serve as a predictive tool in assessing the chemical shift range in which a selenium vertex from a particular compound might be expected to be found, thereby reducing expensive experimental time. Furthermore, the relative chemical shifts between selenium vertices in clusters containing more than one selenium atom are consistent across the range, thus allowing the assignment of the selenium resonances with a high degree of confidence even in relatively low-level density functional theory calculations. A new macropolyhedral 20-vertex selenaborane Se2B18H20 (A) is also reported.


■ INTRODUCTION
−3 For example, it has proven useful in enabling bioorganic selenametabolite compounds to be individually identified by their 77 Se chemical shifts, thus providing a 77 Se ′fingerprint′. 4 In inorganic carboranyl boron cluster compounds, the presence of selenium has been used to assess the relative basicity of carboranylphosphines via the P−Se coupling constants. 5We, however, are interested in selenium chemical shifts in selenaborane macropolyhedral cluster compounds and the use of quantum chemical calculations to predict their chemical shifts.
Binary boron hydride compounds (boranes), a group of inorganic polyhedral molecules comprising clusters of boron and hydrogen atoms, display a diversity of structures surpassed only by the hydrocarbons. 6,7−18 In the case of selenium and sulfur, these have also been combined in so-called macropolyhedral clusters, which are defined as "borane compounds that contain two or more cages, with individual cages that are joined or fused to each other with two or more atoms held in common". 19The synthesis and characterization of a number of new anionic macropolyhedral selenaboranes have been described, 20,21 and, more recently, the photophysical properties of their luminescent neutral conjugate acids together with other luminescent macropolyhedral thiaboranes have also been reported. 22In terms of NMR spectroscopy, these selenaborane compounds were characterized by their 11 B and 1 H spectra.The nuclear spin-half property of selenium-77, its 7.6% natural abundance, and gyromagnetic ratio of 19.071523 × 10 −7 rad• s −1 •T −1 allow an additional and easily available metric using a routine single-pulse sequence for the characterization of selenaborane clusters, which can potentially deliver additional structural information.Nevertheless, historically, the measurement of 77 Se chemical shifts during the characterization of selenaborane species has not been routine, 17,23−30 and it has thus hitherto received little attention, although a recent publication has described a similar approach to our work reported herein, with a pairing of measured and calculated 77 Se NMR chemical shifts in a number of perchlorinated monoselenaborane clusters. 31There is also a number of compounds that contain both B−H units, chalcogens and metal units, such as Mo or Fe, in which 77 Se chemical shifts have been measured, but they feature cubane-type or other structural motifs that do not resemble conventional borane clusters based usually on an icosahedron. 15,16,32They thus fall outside the scope of the work reported here.
In selenaborane compounds with only one selenium vertex, the assignment of the selenium resonance is clear.However, the assignments are not so straightforward in the macropolyhedral and other selenaborane species reported here, which contain two or three selenium atoms per molecule.There are several NMR techniques for assigning 11 B and 1 H resonances in borane clusters to their cluster positions, and they typically include different variations of 1D experiments (coupled, broad-band or selective decoupled, etc.) together with 2D methods such as 1 H− 11 B HMQC (heteronuclear multiple-quantum correlation) and 1 H− 11 B HSQC (heteronuclear single-quantum correlation) and 11 B− 11 B-COSY (homonuclear correlation spectroscopy) methods.However, there are no 1D or 2D correlation NMR techniques suitable for assigning the positions of selenium vertices in heteroborane species holding more than one selenium atom.This is because both 1 H and 11 B resonances are too broad to observe selenium-77 satellites or to apply the correlation methods based on long-range 1 H− 77 Se or 11 B− 77 Se couplings, which would enable a direct assignment of the boron resonances adjacent to the selenium vertices.We have thus resorted to quantum chemical calculations to investigate the possibility of achieving a suitably close correlation between calculated and measured values in order to assign the selenium resonances to their positions in selenaboranes.Herein, we endeavor to collate and describe the measured and calculated 77 Se NMR chemical shifts of this range of selenaboranes, limited to those in which selenium acts as a vertex in the borane cluster rather than, for example, an exo-polyhedral substituent replacing an exoterminal hydrogen atom and also where selenium is the only cluster heteroelement in order to eliminate the complications that will ensue from the inclusion of, for example, heavier ligated transition metal moieties mentioned above.We show how these collective data can be used to predict the region in which to expect the 77 Se NMR signals to be found in new selenaboranes.This approach has been used previously for correlations between 11 B calculated and measured values 33 and a range of organic and inorganic selenium compounds 34,35 but has not previously been applied to selenaboranes.Measured 77 Se resonances are very sensitive to the local surroundings such that they can be spread over a long range of ca.3000 ppm, 36−38 and thus they can be time-consuming to locate, especially in weaker samples.Consequently, there is utility in accurately calculating the potential chemical shift range in which a particular selenium atom might be expected to be found.

■ RESULTS AND DISCUSSION
This paper deals primarily with the comparison of the calculated and measured 77 Se chemical shifts of the previously reported neutral and anionic selenaborane cluster compounds of known structure, as shown in Chart 1.Nevertheless, we first describe the characterization of a previously unreported new neutral selenaborane Se 2 B 18 H 20 (Compound A).Data for A were not included in our recent paper on the luminescent properties of chalcogen-containing macropolyhedral species, 22 as the low amounts of the compound available were not sufficient for full photophysical investigation.It is, however, useful to demonstrate how a comparison of the measured and calculated Se-77 chemical shifts can be used to support the proposed structure.
Compound A is produced almost quantitatively from the protonation of a low-yield (3%) side product, the [Se 2 B 18 H 19 ] − anion, formed in the reaction between syn-B 18 H 22 and elemental selenium. 21This is in contrast to the higher yield (48%) equivalent reaction between elemental sulfur and syn-B 18 H 22 . 39Similarly to the reaction for [S 2 B 18 H 19 ] − , the 20 vertex macropolyhedral species Se 2 B 18 H 20 A may be prepared from [Ph 4 P][Se 2 B 18 H 19 ] by the addition of H 2 SO 4 or CF 3 COOH to dichloromethane solutions of the anion (Scheme 1).An almost quantitative formation of protonated compound A was observed by boron-11 NMR spectroscopy on acidification of the anion (Figure S1) with H 2 SO 4 .The thiaborane analogue has been described, and its structure deduced from NMR spectroscopy allied with density functional theory (DFT) calculation. 40We similarly have only been able to characterize the cluster compound by 11 B, 11 B{ 1 H}, 11 B{ 1 H selective}, and 1 H-11 B HMQC spectroscopy, with the addition of 77 Se NMR spectroscopy, and by comparison of these with the thiaborane analogue.Our attempts to obtain single crystals of both A and the sulfur analogue suitable for Xray diffraction analysis were unsuccessful.The 11 B and 1 H NMR data for A are listed in Table 1 together with the boron chemical shift data for S 2 B 18 H 20 , and they show an excellent correspondence between the measured sulfur and selenium species and the calculated boron NMR chemical shifts for A. The 11 B{ 1 H} spectrum for A and the 77 Se spectrum are shown in Figure 1. Figure 2 shows an ORTEP-type drawing of the DFT calculated structure together with the cluster numbering.The 77 Se NMR spectrum in Figure 1 shows two selenium resonances labeled Se(9′) and Se (9).These were assigned by a comparison of the calculated and measured chemical shifts.The means by which we arrived at these assignments are described next.
Calculations.Precisely calculated values for heavy elements such as 77 Se may ideally include relativistic corrections using routines other than DFT methods, but these would require considerable computing power for even simple selenium-containing compounds. 36GIAO-MP2 methods have been used to good effect in the calculation of 77 Se chemical shieldings. 34These, however, are also computationally expensive, and we therefore compare them to less expensive DFT methods that may be more routinely used in the characterization of selenaborane compounds.Here, we parallel lower-level DFT calculated isotropic shielding constants using B3LYP and mPW1PW91 functionals to those of MP2 (2nd order Møller−Plesset perturbation theory).In all cases, use the same 6-31+G(d,p) basis set for the main group elements B, H, and Cl and with the Binning and Curtiss 962 + d polarization basis set for Se, as used previously by

Inorganic Chemistry
Buehl et al. 34 The calculated chemical shift values presented in Table 2 are derived from linear regression analyses of the calculated isotropic shielding relative to the calculated value for the dimethylselenide reference at the appropriate level versus the measured chemical shifts.Plots of the data are shown in Figure 3 together with the linear regression parameters.The three methods can be seen to produce very similar values for R 2 , the coefficient of determination, and the two relatively lowcost DFT methods, which are not significantly different from the more computationally costly MP2 calculations.The mPW1PW91/6-31+G(d,p) method shows a good correlation between measured and calculated values, and these will be used in further discussions.
Although there is no way to definitely assign the calculated values to the measured values in the multiselenium compounds, the close conformity of the calculated and measured values in the compounds containing a single selenium vertex, including the perchlorinated species SeB 11 Cl 11 M and SeB 5 Cl 5 N, allows us a large degree of confidence in the assignments in those macropolyhedral species where the identities of the measured selenium resonances cannot be directly assigned.The differences in the chemical shifts between the sites in the macropolyhedral species are, in most cases, considerably greater than the standard deviations in the calculated values.This is, however, currently a limited set of data.
In the three methods, there is one prominent outlier point (I in Figure 3).This arises from the [(Se 2 B 9 H 10 )(SeB 9 H 11 )] − anion, compound I in Chart 1.Here, the 3-connected Se(3), vertex held on a nido cluster, affords a measured value of δ( 77 Se) + 158 ppm and a calculated value of +239 ppm, making it the selenium atom presenting the largest deviation for all three methods.In contrast, the Se(1) held on an arachno cluster in the same molecule gives δ( 77 Se) −286 ppm, which is very close to the calculated value of −285 ppm, and also a bridging Se(2) center with very close measured and calculated chemical shifts of +97 and +63 ppm, respectively.This is an   S2) shows that the bridging hydrogen resonance at +0.57 ppm is coupled to the boron resonance at −20.4 ppm as would be expected if the boron resonance is due to vertex B(5).Some cross peaks are not found, but these were located by selectively decoupled boron in the proton spectra, 1 H{ 11 B selective }, as shown in Figure S3.Additionally, as we have noted before, 41 it is often possible to differentiate the assignments of close boron resonances from calculation by also looking at the calculated values for their individual directly attached exo-terminal proton resonances.Thus, for B(5)-H, the calculated value is +1.17 ppm compared to that for B(1)-H, where it is +2.93 ppm.These nicely match the measured values from the HMQC spectrum of +1.17 and +2.82 ppm, respectively, thereby supporting the assignments given in the table.This molecular flexibility may introduce effects not well modeled by the calculated static structures.Nevertheless, in nearly all cases, each calculational method produces the chemical shifts in the same order as the measured values.Thus, the ordering of the higher and lower field resonances in the measured macropolyhedral selenaboranes is always mirrored by the calculations.This is also true for compound I above.The only exception is in the 11-vertex nido clusters for the mPW1PW91/6-31+G(d,p) level, where the ordering of the very close measured values for the anion [SeB 10 H 11 ] − J and SeB 10 H 12 K at +105 and +100 ppm is reversed in the calculated chemical shifts (+82 and +93 ppm, respectively).This indicates that the predictions for separate molecules with very close chemical shifts are not reliable.Nevertheless, the predictions are still sufficiently close to the observed chemical shifts to allow the resonances to be located easily during the measurement.
Overall, with this small sample of compounds, the anionic species seem to appear generally at higher field than the neutral species as shown in the A/B, G/F, and K/J conjugate acid/ base pairs.Figure 4 illustrates the distribution of the selenium resonances in polyhedral boron hydride species A to L. The perchlorinated closo species M and N are not considered here, as they constitute a separate subgroup.From this presentation of the data, two trends are revealed.First, the measured 77 Se chemical shifts in the nido subclusters with a 4-connected Se vertex span a smaller range (from +62 to +322 ppm) compared to the subclusters with a 3-connected Se vertex, which span a much larger range of almost a 1000 ppm (from +474 ppm in neutral SeB 17 H 19 , H, to −516 ppm in anionic [Se 2 B 18 H 19 ] − B).Second, the data so far suggest that selenium vertices on the nido clusters and subclusters are to lower field than that for the arachno subclusters.
Also noteworthy is the resonance in the 20-vertex [Se 2 B 18 H 19 ] − anion B at −516 ppm, which is 230 ppm more shielded than the closest next resonance�the arachno 3connected Se(1) vertex in I (−286 ppm).The extent of this difference led us to consider its reason.Thus, in B, one subcluster contains a rare 6-connected boron vertex, 21,43,44 and the subcluster could be regarded as being based geometrically on a closo-14-vertex tetradecahedron 45,46 of either arachno 12vertex constitution with 2 missing vertices or of hypho 11vertex geometry based on 3 missing vertices (Chart 2a) depending on whether or not the boron atom labeled B is included in or excluded from the subcluster vertex count.In both of these, the selenium vertex is in an antipodal position with respect to the 6-connected boron in the tetradecahedron.We consequently hypothesized that the chemical shift might be due to the antipodal effect 47 of the boron vertex, and we therefore calculated the chemical shift of selenium in a hypothetical closo-SeB 13 H 13 cluster using the mPW1PW91 level regression analysis.However, as shown in Chart 2b, the chemical shift of +145 ppm is in the same region as for closo-SeB 11 H 11 (L) and no extreme effect of the unusual boron vertex is evident.Interestingly, the chemical shift of the 4connected selenium vertex directly adjacent to the boron atom in the dicommo linkage that is protonated on the acidification of [Se 2 B 18 H 19 ] − (see Scheme 1) does not undergo a large chemical shift change (+62 to +119 ppm), whereas the more remote 3-connected selenium changes from −79 to −516 ppm.This indicates that the second subcluster has a large influence on the electronic environment of the selenium.Separate calculations of the δ( 77 Se) for the hypothetical arachno and hypho subcluster geometries, Chart 2c,d, respectively, show a much better correlation between the 11-vertex hypho constitution and compound B (see Chart 2), suggesting this to be the better description of its subcluster geometry and may account for its increased shielding.
Finally, as we noted earlier in the description of the characterization of Se 2 B 18 H 20 A, primarily by 11 B and 1 H NMR spectroscopy and their comparison to those of S 2 B 18 H 20 , the prediction of the selenium chemical shifts by all three methods gives very close correlations to the measured values, and they therefore give independent confirmation of the proposed structures of both Se 2 B 18 H 20 and S 2 B 18 H 20 .

■ CONCLUSIONS
This work indicates that fast and computationally inexpensive lower-level quantum chemical calculations are sufficiently accurate to enable the assignment of measured data in macropolyhedral selenaborane species.There are, however, currently only a limited number of examples from which to make this assessment, and we hope to expand this data set with further work.Nevertheless, being able to predict 77 Se NMR chemical shifts in selenaborane cluster compounds offers several benefits.
1 The chemical shift of the selenium nucleus in a selenaborane cluster compound can provide information about its molecular structure.Accurate prediction of these chemical shifts allows researchers to better understand the arrangement of atoms within the cluster, as well as subcluster geometries (nido, arachno, etc.). 2 Accurately predicted chemical shifts enable the position of 77 Se resonances to be more efficiently located in experimental measurements.Indeed, in the final sample we measured, SeB 18 H 20 , for which only a small amount of material was available, requiring a long accumulation time, we looked in the predicted region of zero to +400 ppm and found the measured resonance at +322 ppm (calculated: +303 ppm).3 Reliable prediction of 77 Se NMR chemical shifts aids the interpretation of experimental spectra, permitting the assignment of specific signals to different structural motifs within selenaborane cluster compounds.This may aid in the identification and characterization of these compounds in complex mixtures or reaction intermediates.
■ EXPERIMENTAL SECTION Caution.Although selenium is a biologically useful trace element, it is toxic, 48 and selenium-containing compounds should be handled accordingly.
NMR Spectroscopy.NMR spectra were recorded on a JEOL ECZ 600 R (14.1 T) spectrometer using 77 Se, 11  The 77 Se NMR spectra were measured using a standard singlepulse sequence (available from the spectrometer library) with a 90°p ulse length and relaxation delay of 0.1−0.2s.The line widths of the 77 Se resonances were in the range of 200−300 Hz, and the spectra were recorded with an FID resolution of 3−4 Hz.Due to the absence of the hydrogen atoms attached to the Se atom, the spectra were acquired without proton decoupling.
Computational Details.Calculations were performed using Gaussian16 package. 49For the DFT/B3LYP methodology, the 6-31+G(d,p) basis sets for B, Cl, and H were used, and the Binning and Curtiss 962 + d polarization basis set for Se was taken from Basis Set Exchange. 50The higher level MP2 calculations were carried out using

Inorganic Chemistry
the same basis sets.The polarizable continuum model was implemented with CHCl 3 solvation.Frequency analyses to confirm the true minima were performed at the appropriate level.

Figure 1 .
Figure 1.(Upper) 11 B-{ 1 H} NMR spectrum of Se 2 B 18 H 20 , A. All resonances are doublets in the 11 B spectrum except for the singlet resonance due to the commo-boron atom linking the two subclusters, as denoted by an asterisk.(Lower) 77 Se NMR spectrum of A.

Figure 2 .
Figure 2. ORTEP-type diagram of the DFT-calculated structure of Se 2 B 18 H 20 A, showing the cluster numbering.

Figure 4 .
Figure 4. Schematic illustration of the distribution of 77 Se resonances in selenaborane species.See Chart 1 for compound identity.

Chart 2 .
Structural Considerations of the [Se 2 B 18 H 19 ] − Anion (B).(a) Measured Chemical Shifts and (b−d) Calculated Chemical Shifts at the mPW1PW91 Level.

Table 1 .
Measured Proton and Boron-11 NMR Data for Se 2 B 18 H 20 (A) at 293 K in CDCl 3 Solution with B3LYP/6-31+G(d,p)/GIAO Calculated Chemical Shifts in Square Brackets and Together with Measured δ( 11 B) NMR Comparison Data for S 2 B 18 H 20

Table 2 .
Measured and Calculated 77 Se NMR Chemical Shifts in CDCl 3 Square brackets indicate that the compound is monoanionic.See Chart 1 for schematic structures of the compounds and Figures 1 and S4−S16 for the 77 Se measured spectra.b 4-Connected selenium on an 11-vertex subcluster.c 3-Connected selenium on a 10-vertex subcluster.d Chemical shift values estimated from the linear regression analysis shown below in Figure 3. e The [SeB 10 H 11 ] − anion 25 was obtained as a byproduct in the synthesis of [Se 2 B 17 H 18 ] − , 21 and its conjugate acid, SeB 10 H 12 , was obtained by acidification of the anion with H 2 SO 4 .SeB 11 H 11 was donated by Josef Holub. 42f Se(1), see Chart 1.

Preparation of Se 2 B 18 H 20 (A).
The compound was prepared by adding excess concentrated H 2 SO 4 to a CH 2 Cl 2 solution of [Ph 4 P][Se 2 B 18 H 19 ] in a small sample tube.The mixture was shaken, then allowed to settle, and the upper CH 2 Cl 2 layer was decanted.The solvent was removed under a stream of nitrogen and redissolved in CDCl 3 for NMR spectroscopic measurement.The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.inorgchem.4c01890.Boron and proton NMR spectra for compound A, plots of 77 Se NMR spectra for those compounds measured by the authors, tables of calculated chemical shieldings versus measured chemical shifts, and tables of calculated Cartesian atomic coordinates (PDF) Institute of Inorganic Chemistry of the Czech Academy of Sciences, Husinec-R ̌ež250 68, Czech Republic; orcid.org/0000-0003-3615-1938;Email: bould@iic.cas.cz