B-N as a C-C substitute in aromatic systems.(Report)

Introduction

The isoelectronic relationship between "B-N" and "C-C" has been appreciated and exploited for several decades. The substitution of one or more B-N fragments for C-C fragments in classic aromatic hydrocarbons was first explored in the seminal work of Dewar and co-workers in the 1950s and 1960s. More recent investigations have revealed the potential for boron-doped aromatic species in new generations of materials with desirable photophysical and electrochemical properties (1, 2) or biological activity (3). First hinted at via computational investigations, these improved properties have begun to be observed experimentally and with recent development of new methods for synthesizing BN substituted aromatic species, the potential for access to new compounds has reinvigorated research into elemental isomers (4) of classic aromatic species. The recent resurgence of interest in this field of research and the fact that 2008 is the 50th anniversary of Dewar's initial report (5) on a BN phenanthrene derivative has prompted this review. While the focus of this account is recent chemistry of analogues of aromatic hydrocarbons incorporating a BN moiety, some of the early work summarized in the last review (6) will be briefly discussed to give context.

Perhaps, the quintessential BN hydrocarbon analog is borazine ([B.sub.3][N.sub.3][H.sub.6]), commonly referred to as "inorganic benzene", an extreme in the continuum of BN for CC substitution in a six-membered aromatic ring. The chemical and physical properties of borazine are well-established, though the debate as to whether or not it can be referred to as "aromatic" continues (7). The sequential replacement of CC moieties for BN in six-membered and more complex aromatic systems has been less thoroughly investigated, and it is therefore interesting to consider the compounds that exhibit such sequential replacements and to elucidate the effects on their structural, electronic, and photophysical properties. As such, this review deals exclusively with compounds obtained by incompletely replacing CC for BN in conjugated systems. That is, fully heteroatomic analogues of cyclopentadienyl and benzene (i.e., borazine) are beyond the scope of this article and have not been included. The intent here is to focus on the intermediate range from organic to inorganic benzene and polycyclic species derived from these. Indeed, there have been numerous accounts examining borazine (8), fully heteroatomic BN-cyclopentadienyl analogues (9-11), linear BN-acenes (12-14), as well as the analogues of graphite, such as hexagonal boron nitride (15), but these will not be specifically addressed here. To further limit the scope of this review, the omission of graphitic structures composed of boron, carbon, and nitrogen commonly referred to as BCN materials, is justified on the basis that the materials so formed have an ill-defined arrangement of BN and CC moieties; that is, the heteroatom substitutions are not at prescribed positions (2, 16, 17). Investigations of BN substitution with "defined" locations in such structures will be addressed.

From the onset of BN aromatic chemistry, several groups have investigated the relative stabilities of the three possible [C.sub.4]BN benzene isomers, with much the same findings (6, 18- 20). The BN-benzene formed by adjacent BN substitution, the 1,2-isomer (Chart 1a), is by far the most stable owing to the complimentary [pi]- and [sigma]-donations from nitrogen and boron, respectively. The "disconnected" 1,4-isomer b has been predicted to be significantly less stable than a, though polycyclic derivatives are known. The least stable substitution pattern is the 1,3-isomer c, for which no classical uncharged bonding depiction can be drawn. Indeed, no examples of this isomer have been reported.

Based on atomic radii, the expected bond length for a B-N single bond is 1.58 [Angstrom], while that for a B=N double bond is 1.40 [Angstrom] (21). Just as the CC bond lengths in benzene are intermediate in length between double and single bonds, the participation of the BN fragment in aromatic systems can also be anticipated to result in a bond length between that of a purely single or double bond. Clearly, this bonding situation is indicative of [pi]-character, and as such an appropriate depiction of this bond is shown in Chart 2b, which also highlights the dipolar nature of this moiety. However, it has been shown that the B and N atoms do not become significantly polarized due to opposing [sigma]- and [pi]-electron polarizations (Chart 2c), and the net formal charges are actually quite similar, even resulting in a net negative charge at nitrogen (6, 22). Given the suppressed polarization, the resonance form shown in Chart 2a is also valid (and commonly depicted), and though it does not highlight the multiple bond character, this depiction will be used throughout this review for the sake of clarity.

[FORMULA NOT REPRODUCIBLE IN ASCII.]

In terms of organization, we classify and discuss BN heterocycles on the basis of the parent azaborinine ring only, and appended rings are considered as substituents on the core heteroaromatic ring of interest. That said, polycyclic systems are regularly referred to in such a way as to highlight their similarity to common aromatic compounds: BN-naphthalene, [B.sub.2][N.sub.2]-triphenylene, and so forth. The abbreviated nomenclature for azaborinines proposed by Thakkar (20) has been adopted in this review, particularly as it pertains to [C.sub.2][B.sub.2][N.sub.2] rings. Thus, for example, in describing [C.sub.2][B.sub.2][N.sub.2] heterocycles, the first two numbers indicate the nitrogen atom positions, and the latter two indicate the boron atoms.

Six-membered rings

Heterocycles with one BN moiety: 1,2-azaborinines

[FORMULA NOT REPRODUCIBLE IN ASCII.]

Benzene analogues

The simplest ring to consider is 1,2-azaborinine, in which one CC pair of benzene has been substituted with a BN unit. As yet, the parent system has eluded isolation, though Dewar and co-workers were able to detect this species via mass spectroscopy (23). Their attempts at forming this parent ring led to rapid polymerization although the products of decomposition were ill-defined. The first substituted examples of this system were prepared by Dewar via desulfurization of 1 with Raney nickel to give 2 (Scheme 1) (5) and by White via dehydrogenation of azaboracyclohexane 3 giving phenyl-substituted 4 (Scheme 2) (24).

In 1980, new derivatives of BN-benzene were prepared by Gronowitz (25, 26), following the analogous desulfurization protocol to that originally utilized by Dewar (Scheme 3). Incorporating a non-fused thiophene moiety at the position ? to boron, monodesulfurization yielded only the ethyl derivative 6, which could then be reacted further to produce the target compound 7, with an aliphatic, rather than aromatic, group at the ??position. Note that utilizing the regioisomer 5, Gronowitz was able to obtain the isomer 7 having an aliphatic substituent at position 4 as opposed to position 3 (26). Derivatives were assessed by UV-vis and NMR spectroscopies.

[FORMULA NOT REPRODUCIBLE IN ASCII.]

In 2000, a communication by Ashe reported the synthesis of 8a via condensation of in situ generated allylboron dichloride with ethylallylamine, followed by installation of the B-phenyl group, ring-closing metathesis, and oxidation with DDQ (Scheme 4) (27). In the subsequent year, this same group reported the synthesis of compounds 8 via a ring-expansion reaction of 9 with C[H.sub.2][Cl.sub.2] and an appropriate lithium base (Scheme 5) (28); overall yields in this latter methodology showed a modest improvement (~35% vs. ~25%). Complexation with Cr[(CO).sub.3] and Mo[(CO).sub.3] showed compounds 8 to be viable ?-ligands, and allowed the first crystallographic characterization of the BN-benzene ring. Here, the BN bond was found to be 1.466(6) [Angstrom], suggesting delocalized [pi]-bonding.

The reaction of 10 (the conjugate base of 4) with [[Cp * RuCl].sub.4] gave rise to mixed sandwich Ru complex 11 (Scheme 6) (29). The N atom of this complex could be protonated by acetic acid, and the acid was determined to have a p[K.sub.a] of 9.21 [+ or -] 0.10 in 85% MeOH/[H.sub.2]O. This complex was also examined for utility as a nucleophilic catalyst in the acylation of benzyl alcohol with phenylethylketene, but was found to have significantly lower activity than conventional catalysts.

Crystallographic analysis of a derivative of 4 without a directly coordinated metal was first accomplished by Ashe and co-workers in 2006 on the 2-(PhCr[(CO).sub.3]) substituted compound 12 (Fig. 1, Scheme 7) (30). The B-N bond length of 1.430(5) [Angstrom] was in the appropriate range for a BN double bond, while the bond lengths throughout this ring provided evidence for significant [pi] delocalization. Complex 12 was formed as the thermodynamic product of 4 with [(MeCN).sub.3]Cr[(CO).sub.3] in THF (Scheme 7), whereas the kinetic product was found to be 13, suggesting that the phenyl ring serves as the better [pi]-ligand to Cr[(CO).sub.3]. Conversely, the deprotonated analogues 14 and 15 exhibited the reverse haptotropic tendency, with 1,2-azaboratabenzene serving as the superior [pi]-ligand in the anionic complexes. These migrations were determined to be largely intramolecular via deuterium-labeling studies.

Complimentary crystallographic data as well as reactivity towards electrophiles were communicated in 2007 by Ashe (31). This first investigation of classic aromatic substitution chemistry on a BN-benzene derivative (other than borazine) (32) revealed that the positions [alpha] (ortho) and [gamma] (para) to boron were susceptible to electrophilic substitution (Scheme 8), in agreement with early investigations by Dewar on larger ring systems (vide infra) (33) as well as previous computations and predicted resonance stabilization (19). Compounds 16a-16c and 17d-17e were formed in poor-to-good yields (10%-91%).

[FORMULA NOT REPRODUCIBLE IN ASCII.]

[FIGURE 1 OMITTED]

Derivatization of the BN-benzene via nucleophilic substitution at boron was recently reported by Liu and co-workers (34). Thus, the reaction of chloro boron species 18 with several anionic nucleophiles, including [H.sup.-], proceeded in good yields (Scheme 9). A demonstration of the synthetic utility came with the formation of the BN analog of the hypolipidemic agent methyl 2-ethylphenoxyacetate 19i, which was achieved through reaction of 18 with methyl glycolate.

[FORMULA NOT REPRODUCIBLE IN ASCII.]

Bicyclic systems

Bicyclic aromatics incorporating the isoelectronic BN unit were first synthesized by Dewar, utilizing condensation reactions of B[Cl.sub.3] with appropriate olefinic amines to give the framework 20 (Chart 3) (35). Dewar also introduced the internalized BN-naphthalene isomer 21, via hydroboration/ cyclization of di-3-butenylamine with subsequent dehydrogenation (33). Spectroscopic and reactivity studies on 21 substantiated the claims that aromaticity was retained upon heteroatom substitution in the rings. The stability of BNnaphthalene 21 to ambient atmosphere is noteworthy, as hydrolysis of 20 had been observed to be appreciable under such conditions; this highlights the chemical stability imparted by "internalizing" the heteroatoms at ring junctions.

The ring-closing metathesis methodology introduced by Ashe (vide supra) was extended to give access to the 3a,7a-azaborindenyl framework, isoelectronic to the indenyl framework (Scheme 10) (36). A zirconium complex of this ligand was structurally analogous to that of the all-carbon indenyl-supported zirconocene in that the metal associated most strongly with the three non-bridgehead atoms of the five-membered ring. Complexation with Cr[(CO).sub.3] showed the formally anionic five-membered ring to be coordinating in the thermodynamic product, just as in BN-benzene complex 14 (30). Competition studies with the zirconium complex in the presence of indene showed the all-carbon system to be a superior [pi]-ligand with free 22 recovered.

A new synthetic route to the BN-naphthalene isomer 21 was introduced by Ashe and co-workers in 2006, utilizing their ring-expansion methodology (vide supra) from 22 (Scheme 10). Overall yields of 5.6%, while still far from desirable, represented an improvement on the 0.2% optimized yield of the original synthesis (33). Importantly, the fully unsubstituted BN-naphthalene 21 was assessed by X-ray diffraction, and found to be isostructural with naphthalene, having an analogous herringbone morphology dominated by CH-[pi] interactions and no [pi]-stacking. Furthermore, the short BN bond (1.461(1) [Angstrom]) was again in the appropriate range for this moiety within an aromatic ring.

The BN-naphthalene isomer 20 was re-visited by Paetzold and co-workers in 2004 (37). Following the procedure originally outlined by Dewar, they were able to access both halogenated and alkylated (at B) derivatives via condensation of 2-aminostyrene with appropriate haloboranes (Scheme 11). Introduction of water to the chloro derivative 23a resulted in an equilibrium mixture of the oxide 24 with the 2-hydroxy derivative 23d. Such behaviour, in which the B-OH group serves as a protic acid, has been noted in several boron aromatics and has been taken as evidence for their aromatic stability, as boron more commonly serves as a Lewis acid under similar conditions (6).

[FORMULA NOT REPRODUCIBLE IN ASCII.]

Using a different approach, derivatives were also obtained in an electrophilic addition of N-methyl aminoborane 25 to phenylacetylene, giving 26 (Scheme 12). Reaction of this derivative with [(MeCN).sub.3] Cr[(CO).sub.3] afforded an [[eta].sup.6] complex with the transition-metal fragment associated with the annulated homoaromatic C6 ring only.

Kar and co-workers computed the stabilities of mono-BN-naphthalenes and determined the isomer 27 to be most stable (Chart 4) (12). Interestingly, this derivative is the "inverse" of isomer 20 (i.e., formed by interchanging B and N …