Closely related yet different: a borylene and its dimer are non-interconvertible but connected through reactivity

This side-by-side reactivity study of a borylene and a diborene with the same empirical formula demonstrates their non-interconvertibility.


Introduction
The dimerisation of carbenes [R 2 C:] to form alkenes (Fig. 1A) is a fundamental reaction pattern in carbene chemistry 1 and the basis of the so-called Wanzlick equilibrium. 2 A similar dimerisation process has also been observed with heavier analogues of carbenes. 3 However, despite the existence of isoelectronic group 13 analogues of carbenesborylenes of the form [LRB:] (R ¼ anionic substituent, L ¼ neutral Lewis basic substituent)no carbene-analogous dimerisation (Fig. 1B) has been observed for group 13 species.
Singly base-stabilised [LRB:] borylenes belong to the family of monovalent boron species discovered and recently reviewed by Bertrand et al. 4 Perhaps the most denitive and well-dened example of this class of compounds is the linear borylene [(cAAC Cy )BN(SiMe 3 ) 2 ] (cAAC Cy ¼ 2-(2,6-diisopropylphenyl)-3,3dimethyl-2-azaspiro[4.5]decan-1-ylidene 5 ) reported by Stephan and Bertrand. 6 Over the past few years, however, our group has demonstrated that a number of doubly base-stabilized borylene species ([LL 0 RB:]) can act as synthons for [LRB:] borylenes, either through photodecarbonylation 7,8 or the base-mediated deaggregation of tetrameric cyanoborylene [(cAAC)B(CN)] 4 (I, cAAC ¼ 1-(2,6-diisopropylphenyl)-3,3,5,5-tetramethylpyrrolidin-2-ylidene). 9 The generation of [LRB:] borylenes, either stable or transient, has allowed the study of coordination chemistry at monovalent boron centers through binding of new Lewis bases, in addition to intramolecular C-H and C-C bond activations in the absence of a suitable base. [6][7][8][9] The currently unrealised possibility of dimerising [LRB:] species would provide a new synthetic route to doubly base-stabilised diborenes of the form [LRB]BRL], which still suffer from considerable synthetic restrictions. 10 Our recent synthesis of the aforementioned tetraborylene macrocycle [(cAAC)B(CN)] 4 (I; Scheme 1, le) 9   realise the interconversion of I and II through photolysis and/or heating, we were eager to test their respective reactivity with suitable reagents in order to dene any differences or similarities, and hopefully gain denitive proof of their ability (or inability) to interconvert. Herein we present the reactivity of I and II with elemental chalcogens and dichalcogenides, based on our recently-reported reactions of these chalcogen reagents with B-B multiply-bonded species. 12,13 Our results indicate that, although extensive similarities exist between borylene I and its formal dimer II, and they may therefore be viewed as close relatives, subtle differences in reactivity conrm that no Wanzlick-type equilibrium exists between the two.

Synthetic studies and characterisation
The reaction of tetrameric borylene I with four equivalents of diorganyldichalcogenides, E 2 R 2 (E ¼ S, Se, R ¼ Ph; E ¼ Se, R ¼ Me) in benzene proceeded selectively to the corresponding cAAC-supported cyanoboron dichalcogenides, [(cAAC) B(CN)(ER) 2 ] (ER ¼ SPh 1, SeMe 2, SePh 3), which were isolated in moderate to good yields (Scheme 2A). Whereas the reaction with diphenyldisulde required prolonged heating at 80 C to proceed to completion, reactions with diselenides proceeded at room temperature, the reaction with Se 2 Me 2 being signicantly faster than that with Se 2 Ph 2 . Compounds 1-3 showed 11 B NMR resonances typical of sp 3 hybridized boranes, with the bis(selenides) 2 (À18.4 ppm) and 3 (À14.4 and À15.8 ppm) exhibiting shis signicantly downeld from the bis(sulde) 1 (À9.6 ppm). Furthermore, at room temperature the bis(phenylselenide) 3 displayed highly broadened 1 H NMR ligand resonances as well as two distinct, very broad 11 B NMR resonances (À14.4 and À15.8 ppm), which coalesced upon heating to 70 C, indicating hindered rotation, presumably owing to steric interactions between the bulky cAAC substituents and the large phenylselenide ligands.
NMR spectroscopic data of reaction mixtures of diborene II and Ph 2 S 2 or Ph 2 Se 2 showed the monoboron bis(chalcogenides) 1 and 3 to be sole products of these reactions, independent of the stoichiometry used (Scheme 2B). These reactions proceeded much faster and at lower temperatures than the corresponding additions of dichalcogenides to the tetrameric borylene I. Although no intermediates were observed under these reaction conditions, it is likely that the formation of 1 and 3 from II proceeds via the successive 1,2-addition of the E-E s-bond rst across the B]B double bond and, subsequently, across the remaining B-B single bond. Although both compounds I and II reacted with Ph 2 Te 2 at high temperatures, these reactions were rather unselective and did not yield any tractable products.
Compounds 1-3 readily crystallized from THF at room temperature, 1 as a colorless crystalline solid, and 2 and 3 as yellow crystals. ‡ Fig. 2 shows the crystallographically determined structures of the bis(phenylchalcogenides) 1 and 3, which crystallize in isomorphous unit cells. The B-C cAAC bond Scheme 1 Syntheses of two boron(I) compounds of the same empirical formula, cyanoborylene I and dicyanodiborene II.
Scheme 2 Reactivity of I and II with dichalcogenides. Fig. 2 Crystallographically determined solid-state structures of 1 (left) and 3 (right). Atomic displacement ellipsoids depicted at 50% probability level. Hydrogen atoms and atomic displacement ellipsoids of peripheral substituents omitted for clarity. Selected bond lengths (Å) and angles ( ) for 1: B1-C1 1.6297 (19), B1-C21 1.584(2), B1-S1 1.9248(15), B1-S2 1.9578(16); B1-S1-C22 100.21 (6), B1-S2-C28 112.33 (6).  Stirring of a dark red suspension of tetrameric borylene I with elemental sulfur in a 1 : 1 boron-to-sulfur ratio in benzene for 5 d at room temperature resulted in a yellow suspension, which upon ltration and slow evaporation yielded compound 4 as a yellow crystalline solid (Scheme 3). Compound 4 displayed a 11 B NMR singlet at À17.9 ppm and a single set of 1 H NMR resonances for the cAAC ligand, suggesting a highly symmetrical species. Despite repeated recrystallization attempts, crystalline samples of 4 always contained 5-10% of another species displaying a higher eld 11 B NMR resonance at À9 ppm (vide infra). The analogous reaction with elemental selenium at 60 C for 3 d similarly provided compound 5 as an orange crystalline solid in 70% yield. Compound 5 was isolated as a single species with an 11 B NMR singlet at À33.5 ppm, shied ca. 25 ppm upeld from that of 4, and a highly shielded, broad 77 Se{ 1 H} NMR resonance at À143.1 ppm. In solution at room temperature 5 partially isomerized to a second species presenting an 11 B NMR shi at À31.8 ppm and slightly shied 1 H and 13 C NMR resonances. High resolution mass spectrometry experiments performed on 4 and 5 provided molecular masses consistent with dimeric compounds of the formula This was conrmed by X-ray crystallographic analyses, the results of which are displayed in Fig. 3. Compounds 4 and 5 crystallized in near-identical triclinic unit cells as centrosymmetric species presenting planar 1,3-dithia-and 1,3-diselena-2,4-diboretane cores, respectively. The B 2 S 2 ring in 4 is an approximate square with four almost identical B-S bonds (1.939(3) and 1.940(3)Å) and near-perpendicular B-S-B and S-B-S angles (85.21 (12) and 94.80 (12) , respectively), whereas the planar B 2 Se 2 ring in 5 displays two slightly different B-Se bond lengths (2.069(4) and 2.100(4)Å) as well as nearperpendicular B-Se-B and Se-B-Se angles (85.13 (17) and 94.87 (17) , respectively). There are only a couple of structurally characterized 1,3,2,4-dichalcogenadiboretanes in the literature, all displaying sp 2 hybridized boron centers stabilized by pdonating amino substituents. 17,18 Due to the lower coordination number at boron, the B-E bonds in Nöth's 2,2,6,6tetramethylpiperidine-supported B 2 S 2 and B 2 Se 2 heterocycles are ca. 0.08-0.10Å shorter than in 4 and 5, while the B-E-B and E-B-E angles are ca. 3 narrower and wider, respectively. 18 For both compounds the plane consisting of the cyanoboron moiety and the p-framework of the cAAC ligand forms a ca. 80 angle with the B 2 E 2 plane, with the respective ligands lying in trans-conformation with respect to the B 2 E 2 core. While the solid-state structure of 5 shows a single conformational isomer, it remains unclear whether the observation of two isomers in solution is the result of a 180 rotation of one of the cAAC ligands around the B-C cAAC bond or the existence of a cis-Scheme 3 Synthesis of 1,3-dichalcogena-2,4-diboretanes from I and II.  (3), B1-S1 1.939(3), B1-S1 0 1.940(3); B1-S1-B1 0 85.21 (12), S1-B1-S1 0 94.80 (12). conformer, in which both cyano ligands occupy a cis-arrangement with respect to the heterocyclic core. The B-C cAAC bond lengths (4: 1.639(4), 5: 1.606(5)Å) suggest that the cAAC ligand functions as a pure s-donor ligand to the sp 3 borane, unlike in borylene I, where a signicant contribution of p-backbonding from the electron-rich B(I) center shortens the bond (ca. 1.47Å). It is noteworthy that both the B-C cAAC and B-C CN bond lengths are signicantly longer in 4 than in 5 (B-C CN : 4 1.597(4), 5 1.577(6)Å).
Whereas 4 proved indenitely stable in solution at room temperature, the 11 B NMR spectrum of an analytically pure sample of 5 in CD 2 Cl 2 le at room temperature for 3 d showed the partial disappearance (<10%) of 5 concomitant with the appearance of a new boron-containing species at À22.8 ppm (Scheme 4A). A few crystals of this species could be isolated by recrystallization from diethyl ether and analyzed by mass spectroscopy, revealing a compound of the formula [(cAAC) 2 -B 2 (CN) 2 Se] (6). X-ray crystallographic analysis of this compound showed that 6 is indeed a bis(cAAC)-stabilized 2,3-dicyano-2,3,1diboraselenirane (Fig. 4). § The structure of 6 is similar to that of the bis(NHC)-stabilized 2,3-dithienyl-2,3,1-diboraselenirane obtained by our group upon reaction of one equivalent of selenium with the corresponding diborene precursor, 13 and likewise displays a trans-arrangement of the cyano and cAAC ligands with respect to the B 2 Se core. Attempts to generate 6 from I using a 2 : 1 boron-to-selenium ratio failed, resulting instead in 50% conversion to the diselenadiboretane 5, with the remaining 50% of I le unreacted (Scheme 5). It thus appears that 6 is not accessible directly from borylene I but can be generated in solution at room temperature through loss of one selenium atom from 5, concomitant with B-B bond formation. This was also conrmed by closer inspection of the mass spectrum of compound 5 which showed mass patterns corresponding to 6.
À22.6 ppm, is nearly identical to that of the selenium analogue 6, suggesting little electronic inuence of the chalcogen atom in these B 2 E heterocycles. Both 6 and 7 also displayed a nearidentical set of unsymmetrical 1 H NMR cAAC resonances owing to the presence of the two chiral centers at boron.
The reaction of 6 with one molar equivalent of selenium heated for 6 hours at 60 C in benzene resulted in clean conversion to the 1,3-diselena-2,4-diboretane 5, as indicated by the 11 B NMR shis of the two isomers observed at À31.8 and À33.5 ppm (Scheme 4B). Similarly, the addition of one equivalent of elemental sulfur to thiadiborirane 7 cleanly yielded the 1,3-dithia-2,4-diboretane 4 (Scheme 6). But whereas compounds 5 and 6 are in equilibrium in the presence of selenium, with {5 + Se} favoured at room temperature and 6 favoured at elevated temperature, toluene solutions of 5 showed no evidence of ring contraction to 7 even aer prolonged storage at À30 C. The equilibrium between 5 and 6 could not be further quantied due to the poor solubility of compound 5 and elemental selenium at low temperature in toluene. It does, however, constitute a rare example of fully reversible cleavage of a B-B single bond under extremely mild conditions, and is thereby reminiscent of the reversible insertion of molecular CO into the B-B bond of the borylene borane [(cAAC)H 2 B-B(CO)(cAAC)], which we reported recently. 19 Single crystals of 7 showed a structure very similar to that of its selenium analogue 6, displaying the same trans-arrangement of the cyano and cAAC ligands with respect to the B 2 S core (Fig. 4). The only notable difference lies in the slightly longer B-B bond (7 1.777 (6) 20 neither compound I or II showed any reactivity toward elemental tellurium in toluene, even aer prolonged heating in benzene at 80 C.
In repeated X-ray crystallographic experiments on single crystals in 4 the renement of the data generated some residual electron density around the B 2 S 2 ring that hinted at its overlap with a ve-membered 1,2,4-dithia-3,5-diborolane ring representing less than 5% of the structure. This was consistent with the observation that recrystallized samples of 4 always showed some contamination with another species showing an 11 B NMR resonance at À9 ppm, and that closer inspection of the mass spectrum of 4 revealed traces of a compound with the formula [(cAAC) 2 B 2 (CN) 2 S 3 ]. Based upon this observation, attempts were made to obtain this compound by addition of sulfur to 4. However, even in the presence of excess sulfur with prolonged heating at 60 C, only starting material was recovered (Scheme 7A).
In contrast, the room temperature reaction of tetrameric borylene I with elemental sulfur in a 2 : 3 boron-to-sulfur ratio yielded the corresponding yellow 1,2,4-trithia-3,5-diborolane 8 as the major reaction product in 71% isolated yield (Scheme 7B). This suggests that 8 is formed directly from I and independently of 4, rather than by insertion of a sulfur atom into the 1,3-disulfa-2,4-diboretane ring. Isolated samples of 8 showed two distinct 11 B NMR resonances at À8.5 and À9.1 ppm in a 55 : 45 ratio, ca. 9 ppm downeld from that of 4. The 1 H NMR spectrum of 8 displayed two isomers in that same ratio, each presenting two sets of cAAC resonances in a 1 : 1 ratio. Furthermore, for each isomer, the isopropyl resonances of the Dip residue were split into two sets of asymmetric resonances. The ratio of the two isomers was temperature-independent, suggesting they may be non-exchanging diastereomers.
The analogous 1,2,4-triselena-3,5-diborolane 9 was successfully isolated as an orange-colored solid as the major product (83% yield) of the reaction of borylene I with selenium in a 2 : 3 boron-to-selenium ratio in benzene at 60 C (Scheme 7B). Compound 9 presented an 11 B NMR resonance at À12.3 ppm, ca. 20 ppm downeld from that of 5. Similarly to 8, the 1 H NMR resonances of the isoproyl groups of the Dip residue were split into two sets, indicative of an asymmetric structure. Attempts to detect the 77 Se{ 1 H} NMR resonances of 9 failed due to strong broadening of the signal. Furthermore, while the B 2 S 3 analogue 7 proved stable in solution, compound 9 was observed to fully decompose to the B 2 Se 2 heterocycle 5 in benzene solution over a period of 4 days at room temperature (Scheme 8).
Like their B 2 E 2 counterparts, compounds 8 and 9 crystallized in isomorphous unit cells. For both compounds, the X-ray crystallographic structures (Fig. 5) show a ve-membered central 1,2,4-trichalcogena-3,5-diborolane ring bearing one cyano and one cAAC ligand on each boron atom, arranged in a trans-conguration with respect to the B 2 E 3 ring. While the structures thus represent a single (R,R/S,S) diastereomer, the observation of two non-exchanging isomers of 8 in solution suggests that the meso diastereomer of 8 may also be formed. The formation of a single diastereomer of 9 may be attributable to the higher reaction temperature favoring the thermodynamic (R,R/S,S) diastereomer. The structure of 8 is reminiscent of that of the bis(NHC)-stabilized 3,5-dithienyl-1,2,4-trithia-3,5diborolane obtained by the reductive insertion of three sulfur atoms into the B]B double bond of a diborene precursor, with very similar B-S bond lengths and B-S-B and S-B-S angles. 13 The 1,2,4-triselena-3,5-diborolane ring of 9 is reminiscent of that obtained by Tokitoh and co-workers upon irradiation of a boron bis(methylselenide) bound to a very bulky aryl ligand (Tbt ¼ 2,4,6-(C(SiMe 3 ) 2 H) 3 C 6 H 2 ). 21 Due to the sp 3 hybridization of the cAAC-supported boron atoms in 9, however, the B-Se bonds (2.068(4), 2.086(4)Å) are slightly elongated and the Se-B-Se angle (109.4(2) ) is signicantly more acute than in [(Tbt) 2 -B 2 Se 3 ] (B-Se: 1.942(7), 1.926(8)Å; Se-B-Se: 118.8(4) ). 21 NMR spectroscopic analysis of the crystallization ltrate of 1,2,4-trithia-3,5-diborolane 8 revealed the presence of another boron-containing species presenting an 11 B NMR singlet at À11.2 ppm and a single symmetrical cAAC ligand environment in its 1 H NMR spectrum. Surmising that this may be a tetrathiadiborinane resulting from a 1 : 4 boron-to-sulfur reaction, a scaled-up reaction with this stoichiometry was carried out (Scheme 9). The resulting orange suspension was ltered and the ltrate slowly evaporated to give compound 10 as a pale yellow solid in 53% isolated yield based on boron and its formulation conrmed by LC-MS.

Mechanistic considerations
With all these data in hand, it was now possible to reassess the viability of the mechanism proposed by Tokitoh and co-workers for the formation of [(Tbt) 2 B 2 Se 3 ] from [(Tbt)B(SeMe) 2 ]. 21  with Se, which dimerizes to a 1,3-diselena-2,4-diboretane and nally inserts the third selenium atom. Regarding path A, numerous attempts on our part have failed to convert borylene I into its diborene relative, compound II, under thermal and/or photolytic conditions. The fact that the three-membered B 2 E heterocycles 6 and 7 can only be accessed directly from diborene II, but not from borylene I, is further evidence that, despite extensive overlap of reactivity outcomes, I and II do not interconvert.
Our previous report on the deaggregation of tetramer I by Lewis bases showed that only a relatively small and strong Lewis basethe NHC 1,3,4,5-tetramethylimiazol-2-ylideneis able to break up the tetramer and generate the mixed base-stabilised It is noteworthy that, while the selenium-based reactions were highly selective, those based on sulfur always yielded a mixture of products. For example, based on 11 B NMR spectroscopic analysis of the nal reaction mixture, the reaction of I with sulfur in a 2 : 3 boron-to-sulfur ratio yielded 8 in 70-85% selectivity at most, alongside the smaller and larger heterocycles 4 and 10, whereas the analogous reaction with selenium provided 9 in near-quantitative yield. The increased selectivity in the formation of 9 may be ascribed to the possibility of selenium de-insertion, which allows any B 2 Se 2 heterocycle (5) formed in the course of the reaction to lose a Se atom, forming the diboraselenirane 6, which can in turn be converted to 9 (Scheme 10). In contrast, any B 2 S 2 (4) and B 2 S 4 (10) heterocycles formed in the course of the reaction are inert towards sulfur de-Scheme 10 Possible pathways to the boron-chalcogen heterocycles presented herein. insertion and will therefore remain as by-products. Finally, the fact that the four-and ve-membered B 2 E 2 and B 2 E 3 heterocycles are inert towards chalcogen insertion, whereas the three-membered B 2 E heterocycles may be converted to both the larger B 2 E 2 and B 2 E 3 heterocycles, suggests that ring-expansion only proceeds by insertion of a single chalcogen atom or an E 2 unit into any remaining B-B bonds.

Conclusions
This rst comparative study on the reactivity of two boron(I) compounds of the same empirical formula, the tetrameric, selfstabilizing cyanoborylene I and its dicyanodiborene relative II, has demonstrated that, while both compounds provide access to the same products in many cases, this occurs via different pathways as shown in Scheme 11.
In reactions with dichalcogenides, both I and II yielded the mononuclear cyanoboron bis(chalcogenides) as sole products (Scheme 11). In the case of I this presumably occurs by insertion of a monomeric borylene into the chalcogen-chalcogen bond, whereas for II a successive 1,2-addition mechanism across the B]B double bond, resulting in full B-B bond cleavage is most likely at work.
Reactions with elemental sulfur and selenium were found to be highly dependent on the stoichiometry used. Reactions of I or II employing a 1 : 1 or 2 : 3 boron-to-chalcogen ratio yielded the corresponding 1,3-dichalcogena-2,4-diboretanes or 1,2,4trichalcogena-3,5-diborolanes, respectively. A unique 1,2,3thiadiborirane could only be accessed from the reaction of diborene precursor II with sulfur in a 2 : 1 boron-to-sulfur ratio, whereas the corresponding diboraselenirane was accessible both directly from the analogous reaction with selenium, and indirectly by de-insertion of one selenium atom from the fourmembered 1,3-diselena-2,4-diboretane (Scheme 11). In the case of sulfur a rare example of a 1,2,4,5-tetrasulfa-3,6diborinane was isolated from the reaction of borylene I with sulfur in a 1 : 2 boron-to-sulfur ratio.
Careful stepwise addition of chalcogen equivalents to either I or II and stability studies of the resulting heterocycles also gave insight into several mechanistic aspects of these reactions: (i) Borylene-based reactions do not proceed via a diborene intermediate but likely via monomeric borachalcone and dichalcogenaborirane intermediates; (ii) Ring-expansion reactions can only proceed by insertion of chalcogens into existing B-B bonds; (iii) Ring-contraction is possible in the case of boron-selenium heterocycles only by de-insertion of Se atoms.
Although the subtle divergences in reactivity between I and II provide conrmation that no Wanzlick-type equilibrium exists between a putative monomeric form of I and its B-B bonded dimer, diborene II, the question remains as to whether this is a result of the extremely stable, tetrameric constitution of borylene I. To date, I and II represent the only existing borylene/ diborene pair with the same empirical formula, however, recent advances in the synthesis of borylenes will hopefully enable a more denitive answer to the question of a possible interconversion between [LRB:] borylenes and [LRB]BRL] diborenes. Beyond the interest of such an interconversion from a fundamental point of view, its undeniable potential for providing a new, more reliable route towards hitherto inaccessible diborenes should continue to stimulate research into this area.

Conflicts of interest
The authors declare no conicts of interest.