A crystalline radical cation derived from Thiele’s hydrocarbon with redox range beyond 1 V

Thiele’s hydrocarbon occupies a central role as an open-shell platform for new organic materials, however little is known about its redox behaviour. While recent synthetic approaches involving symmetrical carbene substitution of the CPh2 termini yield isolable neutral/dicationic analogues, the intervening radical cations are much more difficult to isolate, due to narrow compatible redox ranges (typically < 0.25 V). Here we show that a hybrid BN/carbene approach allows access to an unsymmetrical analogue of Thiele’s hydrocarbon 1, and that this strategy confers markedly enhanced stability on the radical cation. 1•+ is stable across an exceptionally wide redox range (> 1 V), permitting its isolation in crystalline form. Further single-electron oxidation affords borenium dication 12+, thereby establishing an organoboron redox system fully characterized in all three redox states. We perceive that this strategy can be extended to other transient organic radicals to widen their redox stability window and facilitate their isolation.

1) As pointed out above, the reviewer strongly suggests to get crystal structures of radical cation and dication prepared from 2, instead of those from 1. The detailed analysis of bond lengths and angles should provide further understanding the bonding situation. Structural parameters, which would be obtained from non-disordered crystal structure, should be related to the appropriateness of two resonance structures in Schemes 3 and 4. On the assumption that the authors would provide these data, the reviewer guesses the third resonance structure (on the right) in each Scheme would be acceptable only in the case the C-N bond lengths in NHB moiety would be shortened in comparison with those of the neutral precursor. Also, these resonance structures have no contribution of B=N double bond, although the B-N bond lengths seem to be relatively short. 2) In Figure 2, the boron nucleus has a negative distribution of the spin density. Please explain this negative value.
Reviewer #3 (Remarks to the Author): The structures are chemically reliable, but data could probably be improved significantly. The data processing software version is outdated by three years. I strongly recommend to repeat the data reduction employing the latest version of CrysAlisPro. 3D profile fitting and smart background treatment in this version should help to improve the data of 2, but also of the other two structures. An analytical absorption correction from crystal faces would also be very advisable, especially for the [SbF6]-containing compounds.
Reviewer #4 (Remarks to the Author): In this contribution, Aldridge et al. report on very exciting unsymmetrical BN-embedded Thiele's hydrocarbon derivatives where the CPh2 substituents are an NHC and an NHB. This substitution pattern has never been as convincingly demonstrated as in this manuscript. While bis-NHC-and bisboryl-substituted Thiele's hydrocarbon systems have been recently disclosed, their stable redox states are limited to one or two. Such a hybrid NHC/NHB strategy provides access to the systems with three stable redox states. Of note, this approach gives rise to a significantly enhanced redox range (> 1.0 V), thereby preventing disproportionation of the intervening radical cation. The literature is properly cited although a relevant paper involving a bis-BN analogue has been left out (Yamashita et al. Chem. Sci. 2018). Given that many organic mixed-valence systems with three stable redox states have fueled a variety of fields, the manuscript will be of high interest to the organic, inorganic and material communities. In my opinion, this study is suited for an article in Nat. Comm. provided that the following points are addressed: 1. The experimental work is well done, but the theoretical support is modest. The radical cation will be resistant towards disproportionation if the oxidative formation of the radical cation from the neutral state is more favorable than further oxidation to the dication. The separation of the two redox waves in experimental CVs is related to chemical potential change (delta µ), which is easily calculated by frontier molecular orbital energies. How does the BN/CC isosterism effect the FMO energies and thus widen the redox range? Can the authors provide rationale? 2. As shown in Scheme 2, species 1 and 2 have biradical character (open shell singlet or triplet). Can the authors calculate it? How does it compare with their bis-NHC analogs? 3. Are there any experimental evidences for the formation of a FLP consisting of {(HCDippN)2B}OTf and IDipp? As NHCs and CAACs are known to react with pyridinium salts, {(HCDippN)2B}OTf may form a salt with pyridine followed by para-attack by NHC. 4. 13C NMR data are not given for (HCDippN)2BOTf. 5. No EA are given for (HCDippN)2BOTf and 1. 6. Please indicate the contribution of each transition from TD-DFT calculations.

Reviewer #1 (Remarks to the Author):
This manuscript describes the preparation of an unsymmetrical organoboron derivatives of Tiele's hydrocarbon. N-heterocyclic carbene (NHC) and its boron analogue (NHB) have been utilized as synthons in synthesis. Cyclic voltammetry revealed two reversible single-electron oxidations. In fact, upon oxidation by AgSbF6 (1 eq and 2 eq), corresponding radical cation and dication were isolated and fully characterized. EPR spectrum of radical cation fits well with simulated one. Obtained hyperfine coupling constant for 11B, 14N, 1H together with calculated spin density strongly suggest the delocalization of the radical across the entire pi-system. A beautiful gradual red-shifting was shown in the UV-vis spectra of neutral, radical cation and dication, and TD-DFT was also provided/discussed. In summary, this is a great contribution by Aldridge group and the manuscript including a significant breakthrough in the main group/organic chemistry is very well discussed. Thus, I recommend for the acceptance in Nature Comm. Following minor points can be considered for the revised version: 1) NMR: 11B NMR for neutral compounds (1 and 2) and dication 2++ should be discussed/mentioned (whilst they showed more or less similar values). This can be related to the resonance structure(s). Response: we have added in the chemical shifts for compound 1, 2, and [1] 2+ into the manuscript proper. While in theory these data might be used to inform the discussion of electronic structure -in reality (as this reviewer points out) the chemical shifts for these three species are essentially identical.
2) DFT: considering the resonance structures for radical cation and dication, NRT calculations can be used. Also, the aromaticity of these species (they have three rings in the molecules) would be worth to discuss. For example, NICS values can be used (if possible). Response: We have carried out both NRT and NICS (NICS(0) and NICS (1)) calculations as set out below (details of these are also included in the revised supporting information). To summarize: the NICS calculations are certainly informative as to the aromatic/non-aromatic/anti-aromatic character of the three systems, and have now been included in the revised manuscript. Using the NRT approach to determine the weight attached to individual resonance structures proves to be more difficult to apply in these highly delocalized systems (even when the model is simplified to remove the N-bound Dipp groups), with >100 resonance structures being obtained for all three systems (neutral, radical cation, dication), and the most heavily weighted contributing structure being <6% in all cases. However, where the NRT approach does prove to be very useful is in providing natural bond orders for each of the bonds (B1-N6, C9-C12 etc. -as below) in the neutral, radical cationic and dicationic species. In reality this provides a much more useful approach for discussing the bond orders within the three molecules, summed over the various resonance forms. Indeed, this approach provides an informative basis for a more extended discussion of resonance structures for 1, 1 + and 1 2+ , of the type suggested by reviewer 2.  Resonance structures: Over 125 resonance structures found in total, which is in agreement with highly delocalized system, parent structure has a weight of only 5.92%. The 20 highest contributions total to 44.46%, of which ca. 54% have C9-C12 double bond and either C7-C8 or C10-C11 double bond and ca. 39% feature structures that exhibit 3 C-C/C-N double bonds in ring 2.  Resonance structures: (α-spin) Over 180 resonance structures found in total, which is in agreement with highly delocalized system, parent structure has a weight of only 1.73%. The 20 highest contributions total 24.84% of the calculated resonance structures, of which: ca. 4% exhibit a C9-C12 single bond and C7-C8 and C10-C11 double bonds (resembling structure 1 ·+ a). Ca. 66% of structures have a C9-C12 double bond and two C-C double bonds in ring 2 (resembling structure 1 ·+ b). Ca. 30% of the structures feature 3 double bonds in the ring 2 with cationic charge localized on N6 (resembling structure 1 ·+ c) (β-spin) Over 162 resonance structures found in total, which is in agreement with highly delocalized system, parent structure has a weight of only 2.52%. The 20 highest contributions total 43.41% of the calculated resonance structures, of which all feature 3 double bonds in the ring 2 and C9-C12 single bond (resembling structure 1 ·+ c).  Resonance structures: Over 128 resonance structures in total. Parent structure has a weight of only 2.48%. 24 biggest contributions total 45.02% of the overall structures, of which: ca 71% exhibit 3 C-C/C-N double bonds in the ring 2 (resembling structure 1 2+ a). Ca. 23% resemble structure 1 2+ b and ca. 6% of resonance structures have C9-C12 double bond and either C7-C8 or C10-C11 double bond (resembling structure 1 2+ c).  This paper reports synthesis of an unsymmetrical BN analogue of Thiele's hydrocarbon and its one-and twoelectron oxidation reactions. The present molecules are very interesting and attractive considering their stepwise oxidation. This characteristic property is very strong point of this paper. In contrast, the unavoidable disorder due to the same substituents on NHC and NHB moieties in crystallographic study for radical cation and dication would be the weakest point. If the authors would provide crystallographic analysis of all the products without disorder due to the same substituents on NHC and NHB moieties, the reviewer would like to suggest to publish this paper in Nat. Commun. with minor revision. Please consider the following points.

NICS(0) and NICS(1) Calculations (For full systems with N-bound Dipp substituents)
1) As pointed out above, the reviewer strongly suggests to get crystal structures of radical cation and dication prepared from 2, instead of those from 1. The detailed analysis of bond lengths and angles should provide further understanding the bonding situation. Structural parameters, which would be obtained from non-disordered crystal structure, should be related to the appropriateness of two resonance structures in Schemes 3 and 4. On the assumption that the authors would provide these data, the reviewer guesses the third resonance structure (on the right) in each Scheme would be acceptable only in the case the C-N bond lengths in NHB moiety would be shortened in comparison with those of the neutral precursor. Also, these resonance structures have no contribution of B=N double bond, although the B-N bond lengths seem to be relatively short.
Response: The idea of obtaining structural data on compounds 2 + and 2 2+ in order to get round the issues related to crystallographic disorder in 1 + and 1 2+ is indeed a good one -and is similar to the approach we took in synthesizing/characterizing the neutral compound 2 as a surrogate for 1. Indeed, with this in mind we spent a good deal of time both prior to submission of the original manuscript, and in the three-month revision period trying to obtain samples of these two species (i.e. 2 + and 2 2+ ) suitable for single crystal X-ray diffraction. Unfortunately -and in much the same way that we saw for the neutral compound 1 -crystals of 2 + and 2 2+ gave (at best) very weak diffraction, and we have not been able to obtain molecular structures of these systems in the solid state. In response to the reviewer's point regarding structural data on the mono-and dications -of course the solidstate structures of the two disordered system do yield some metrical data that are relevant to the discussion of resonance structures for 1 .+ and 1 2+ , in that it is still possible to gauge the extent of aromaticity in the pyridyl ring through consideration of bond-length alternation. In addition, the mean exocyclic CC/BN distances for the three systems 1, 1 + and 1 2+ also allow for discussion of the trend in mean exocyclic double/single bond character (as do the inter-plane torsion angles). That is not to say that non-disordered structures would not offer a more complete basis for structural discussions -the data pertaining to the individual exocyclic BN and CC bonds would of course be beneficial. In the absence of these data, however, the NRT calculations suggested by reviewer 1 (and the natural bond orders derived therefrom) offer a powerful alternative method to use for the discussion of the weighting attached to different possible resonance structures. This approach can be augmented by consideration of the DFT-calculated exocyclic BN and CC bond lengths for for 1 + and 1 2+ (which we have added in to Table 1). Indeed, it could be argued that for systems such as these (containing only lighter elements from the first period) NRT offers a more definitive approach for the discussion of contributing resonance forms than does an analysis based solely on crystallographic data. As such, we believe that the (revised) discussions of the possible contributing resonance forms for compounds 1 + and 1 2+ (and the associated revised resonance schemes) are now on a rigorous footing. In particular, resonance descriptions of 1 + and 1 2+ as containing B-N single bonds fit well with the calculated natural bond orders for this linkage (in both systems) which are close to one. In addition, the very minor contribution of the left-hand resonance structure 1 2+ c (Scheme 4) -is (as the reviewer suggests) borne out by the NRT analysis.
2) In Figure 2, the boron nucleus has a negative distribution of the spin density. Please explain this negative value. Response: The negative spin densities, rho_alpha -rho_beta < 0, at the boron atom are the result of spin polarisation of underlying orbitals by the SOMO. This effect from the SOMO to s-character orbitals results in large Fermi-contact hyperfine interactions. Some literature calls this a Coulomb and others an exchange interaction, regardless it amounts to a spatial difference in alpha and beta orbitals and can amount a change in sign of the overall spin density. Tabulated below are the Mulliken spin densities of (one) imidazole nitrogen and the C2 carbon, and the corresponding nitrogen and boron of the opposing diazaboroline.