Two Directing Groups Used for Metal Catalysed Meta‐C−H Functionalisation Only Effect Ortho Electrophilic C−H Borylation

Abstract Two templates used in meta‐directed C−H functionalisation under metal catalysis do not direct meta‐C−H borylation under electrophilic borylation conditions. Using BCl3 only Lewis adduct formation with Lewis basic sites in the template is observed. While combining BBr3 and the template containing an amide linker only led to amide directed ortho C−H borylation, with no pyridyl directed meta borylation. The amide directed borylation is selective for the ortho borylation of the aniline derived unit in the template, with no ortho borylation of the phenylacetyl ring – which would also form a six membered boracycle – observed. In the absence of other aromatics amide directed ortho borylation on to phenylacetyl rings can be achieved. The absence of meta‐borylation using two templates indicates a higher barrier to pyridyl directed meta borylation relative to amide directed ortho borylation and suggests that bespoke templates for enabling meta‐directed electrophilic borylation may be required.


Introduction
Directed CÀ H functionalisation has developed into an extremely powerful methodology to selectively transform arenes. The interaction of the directing group with the catalyst / reagent can overcome the intrinsic reactivity of the arene and enable highly selective CÀ H functionalisation [1] with regiochemistry otherwise hard to achieve. The ortho CÀ H functionalisation of arenes using covalently bound directing groups is relatively straight forward due to the formation of favoured 5/6 membered intermediates/products. In this area CÀ H borylation is one of the most developed transformations due to the synthetic versatility of CÀ B containing units. [2] Indeed, a multitude of transition metal catalysed, directed lithiation, and metal free (electrophilic) directed ortho-CÀ H borylation methodologies have been reported. [1a,d,3,4] In contrast, achieving the regioselective meta (or para) CÀ H borylation of arenes is much more challenging in the absence of substrate control (such as in the C5 selective iridium catalysed borylation of 1,3-disubstituted aromatics, note, the iridium catalysed borylation of monosubstituted aromatics generally leads to mixtures of meta and para products). [5,6] Nevertheless, notable progress in meta and para CÀ H borylation has been reported recently, e.g. utilising non-covalent interactions. [3b,7] However, these methods require iridium catalysis, thus developing a transition metal-free route, such as electrophilic borylation, to achieve directed meta borylation would be highly desirable.
Directed electrophilic CÀ H borylation proceeds via the interaction of a boron Lewis acid (generally BCl 3 or BBr 3 ) with a covalently bound directing group that contains a sufficiently basic heteroatom (Figure 1, top). [1d] While the use of covalently bound directing groups to effect meta and para CÀ H functionalisation under transition metal (TM) catalysis is now wellprecedented, [1e] to the best of our knowledge meta CÀ H borylation has not been reported using the covalently bound directing group approach (with or without TM catalysis). Since the pioneering work of Yu and co-workers, [8] meta and para CÀ H functionalisation reactions have been reported using a range of directing groups. [1e,9] Analysis of the covalently attached directing groups successful in meta functionalisation show them to be relatively complex due to the requirement to selectively form large rings (e. g. 12 membered) during CÀ H functionalisation. Therefore, the directing groups generally contain a flexible unit, then one aromatic moiety (or more) and finally a donor atom that's part of a rigid group (Figure 1, middle). The latter is most commonly an Aryl-C�N unit, however cyano groups are not appropriate as directing groups in electrophilic borylation [3a] due to their low basicity and their tendency to undergo reactions with nucleophiles at C on binding to an electrophile at N (e. g., the Hoesch reaction).
Another class of meta selective covalently attached directing groups use N-heterocycles as the donor. [10] These include heterocycles such as pyridine and pyrimidine which are well documented to enable directed ortho electrophilic borylation. [1d,11] However, it should be noted that the meta directing groups often contain other basic sites (e. g. amides, imines) in the linker that could also interact with boron electrophiles to effect directed borylation, [12] which would result in the undesired ortho borylation. Herein we report our study into how two covalently attached directing groups used in metal catalysed meta-CÀ H functionalisation react under electrophilic borylation conditions.

Results and Discussion
To commence our study, a template that is similar to a directing group pioneered by Yu and co-workers for metal catalysed meta functionalisation was selected, compound 1 (Figure 2, top). The major differences between 1 and the successful templates used by Yu (e. g., compound A, inset Figure 2) are the absence of a fluorine ortho to N in compound 1, and the absence of a (R = alkyl) blocking group ortho to the aniline NH (used in Yu's report to prevent CÀ H functionalisation at this position). It was important to avoid the ortho-fluorine in 1 as ortho halogenated pyridyls have much lower Lewis basicity and are known to bind BX 3 weakly. [13] This is undesirable as it would disfavour BX 3 binding and also make the halide abstraction from the pyridyl!BX 3 derivative more endergonic, this step is essential to form the borenium (three coordinate boron) cation, [pyridyl-BX 2 ][BX 4 ] that effects CÀ H borylation. [14] The ortho-to NH alkyl group also was omitted as based on previous studies borylation conditions were envisaged that would only proceed via the borenium cation formed from activation of the pyridyl! BX 3 moiety. Initially BCl 3 was utilised as it is reported that BCl 3 does not affect amide directed electrophilic borylation, [12a,b,15] but it is known to effect pyridyl directed electrophilic borylation. [1d,14a] Addition of excess BCl 3 to a DCM solution of 1 resulted in a single new species being observed in the in-situ 1 H NMR spectrum (after 45 mins. at room temperature). However, analysis revealed no CÀ H borylation (based on integration of the aromatic resonances). Analysis of the 11 B NMR spectrum revealed three resonances which were consistent with BCl 3 (δ 11B = 42.2), the pyridyl!BCl 3 adduct (δ 11B = 8.4) and the amide(O)!BCl 3 adduct (δ 11B = 6.8). Leaving the reaction at room temperature or heating it in DCM (to 60°C in a sealed tube) led to no change in the NMR spectra. The lack of any CÀ H borylation was confirmed by work-up involving pinacol/NEt 3 addition leading to the formation of no observable CÀ BPin species by in-situ 11 B NMR spectroscopy or after work-up. The absence of borylation is in contrast to the reactivity of 2phenylpyridine and excess BCl 3 , which under identical conditions leads to 50 % CÀ H borylation (with the other 50 % 2phenylpyridine protonated by the acidic by-product from S E Ar). [1d] Thus, these findings indicate that the absence of pyridyl directed borylation with 1 is due to a higher barrier to borylation via a 12 membered transition state relative to a five membered transition state (in the ortho borylation of 2-phenylpyridine). Calculations (see Supporting information, section 12) disfavour substrate electronics precluding pyridyl directed borylation. Specifically, the close in energy HOMO and HOMO-1 of 2 are principally located on the aryl of the PhCMe 2 -unit and have significant character at the ortho and meta carbons. Despite this, neither site undergoes pyridyl directed electrophilic borylation under these conditions. The product from  addition of excess BCl 3 to 1 was confirmed as the bis-BCl 3 adduct 2 by X-ray crystallography (Figure 2, bottom). The solidstate structure of 2 is unremarkable, containing OÀ B and NÀ B bond lengths of 1.485(2) and 1.594(2) Å, respectively, for the Lewis adducts.
The formation of borenium cations from pyridyl!BCl 3 using additional BCl 3 (to form [pyridyl-BCl 2 ][BCl 4 ]) is endergonic, [14a] thus this step will be contributing to the overall barrier to meta CÀ H borylation using 1/BCl 3 . Therefore, the addition of AlCl 3 to 2 was explored to make borenium cation formation exergonic and thus lower the overall barrier to meta borylation, [16] but this combination failed to affect any CÀ H borylation (even on heating). Instead, addition of AlCl 3 appears to displace BCl 3 from the amide, as only the pyridyl!BCl 3 resonance is observed post AlCl 3 addition (at δ 11B = 8.4 ppm). More insight into the relative stability of the pyridyl-BCl 3 and amide(O)!BCl 3 adducts was forthcoming from the exposure of 2 to "wet" (non-purified) solvent/chromatographic work up which produced compound 3 ( Figure 2, right) in which the amide(O)!BCl 3 dative bond had been cleaved, but the pyridyl!BCl 3 bond has persisted (this is indicted by the NÀ H shifting from δ 1H = 9.97 for 2 to δ 1H = 6.62 for 3 and there being only a single 11 B resonance (δ 11B = 8.4) now observed attributable to the pyridyl!BCl 3 , Figure S1). This confirms the expected stronger binding of BCl 3 to pyridyl relative to amide.
To determine if [pyridyl-BBr 2 ][BBr 4 ] boreniums could be accessed selectively (over [amide-BBr 2 ][BBr 4 ]) we explored the controlled addition of BBr 3 . One equivalent of BBr 3 was added to a DCM solution of 1. Analysis of the 11 B NMR spectrum showed four resonances at δ 11B = À 1.3, À 7.4, À 11.5 and À 24.3. These resonances can be assigned as follows: the δ 11B = À 7.4 is in the region expected for pyridyl!BBr 3 adducts, the δ 11B = À 11.5 is closely comparable to benzoyl!BBr 3 adducts [12a] thus can be assigned to the amide(O)!BBr 3 moiety, the δ 11B = À 24.3 is consistent with [BBr 4 ] À , while the broad resonance at δ 11B = À 1.3 is assigned as the product from amide directed CÀ H borylation. This indicates that BBr 3 reacts unselectively with the Lewis basic sites in 1 thus can affect amide directed ortho borylation even with only 1 equiv. of BBr 3 . Indeed, heating the reaction mixture to 60°C in a sealed tube resulted in disappearance of three of the resonances in the 11 B NMR spectrum with only δ 11B = À 7.4 ppm (pyridyl!BBr 3 ) persisting and a new broad resonance appearing at δ 11B = 0.9 ppm, the latter is in the region for acyl-coordinated arylÀ BBr 2 species in 6-membered boracycles. [12a] As two boron atoms are incorporated into the product (based on the 11 B NMR spectra), > 2 equivalents of BBr 3 are required to achieve complete conversion of 1. Therefore, ca. 3 equivalents of BBr 3 were added to 1, which led to a complex mixture of species in the 1 H NMR spectrum at room temperature. The 11 B NMR spectrum exhibited mostly the same resonances as observed when one equivalent of BBr 3 was used, however the resonance at δ 11B = À 24.3 (due to BBr 4 À ) was no longer present and instead a broad resonance at δ 11B = + 22.0 was observed attributed to a halide transfer equilibrium between BBr 3 and BBr 4 À (vide infra). Heating this reaction mixture to 60°C resulted in complete conversion to a single species in the 1 H NMR spectrum that was consistent with compound 4 (Figure 3). Additionally, the in-situ 11 B NMR spectrum showed the expected three resonances at δ 11B = 34.7 (for unreacted BBr 3 ), 2.9 and À 7.8; the δ 11B = 2.9 resonance is assigned as the CÀ H borylated unit in species, 4. Note this species changes chemical shift in the presence of excess BBr 3 due to reversible bromide abstraction (vide infra). The species at δ 11B = À 7.8 is as expected for a pyridyl!BBr 3 moiety. To further confirm this assignment and enable full characterisation crystals suitable for X-ray diffraction analysis were grown by layering a DCM solution of 4 with pentane. The resultant solid-state structure confirmed that ortho borylation had occurred via amide direction producing a 6-membered boracycle, while the pyridyl group forms a Lewis adduct with BBr 3 . Ortho borylation occurred exclusively on the aniline derived phenyl and results in planarization of part of the directing template (displacement maximum of 0.023 Å from the plane of O1À B1À C12À C11À N1À C10). Locking the template in the conformation required for the 6-membered boracycle reduces the flexibility of the template and may help prevent meta borylation from occurring as post ortho borylation, (and using excess BBr 3 ) prolonged heating of 4 does not lead to any further CÀ H borylation. This indicates that preventing amide directed ortho borylation is essential when using amide containing templates and BBr 3 , this is closely related to the findings of Yu with Pd catalysed meta-functionalisation. [10b] With no meta electrophilic borylation observed using substrate 1 due to preferential amide directed ortho CÀ H borylation, the ester analogue, 5, was next targeted (Scheme 1). Compound 5 was selected as ester/BBr 3 combinations do not affect directed ortho borylation [12b] (in contrast to the more basic amide analogues), this will preclude ortho borylation without having to install alkyl blocking groups onto the template. Furthermore, an extremely similar ester linked template has been used successfully in palladium catalysed meta CÀ H deuteration, [10c] and alkenyl-/acetoxylation. [17] However, the combination of excess BCl 3 or BBr 3 with compound 5 led to no CÀ H borylation (ortho or meta) under a range of conditions, with complex mixtures formed from which the only boron containing species that can be assigned with confidence being due to pyridyl!BX 3 adducts (for X = Cl δ 11B = 8.3, for X = Br δ 11B = À 7.4). The absence of CÀ H borylation was supported by work up with pinacol/NEt 3 , which revealed no species containing CÀ Bpin moieties were formed (by 11 B NMR spectroscopy). Therefore, the failure of template 5 in meta borylation under standard electrophilic borylation conditions is not due to preferential ortho borylation but is presumably due to a high energy barrier to electrophilic borylation using BX 3 via a 12 membered transition state.
It is notable that during the amide directed borylation of 1 only one ortho borylation product is formed, compound 4, with no alternative ortho borylation product, compound B, observed (Figure 4), despite the presence of the CMe 2 unit in the linker which may have been expected to favour ring closure onto the phenylacetyl unit. Therefore, we were interested in the feasibility of amide directed CÀ H borylation where a six membered boracycle is still formed but there is one sp 3 unit in the linker (e. g. the CH 2 and CMe 2 groups in 6 and 7). To the best to our knowledge amide directed electrophilic borylation has not been reported to date for these types of substrates.
The reaction of 6 with 2.5 equivalents of BBr 3 resulted in a species containing a resonance in the 11 B NMR spectrum at δ 11B = À 11.0, in the region for an amide(O)!BBr 3 adduct, thus it is tentatively assigned as 6-BBr 3 (Figure 5, top). Heating of the reaction mixture to 60°C is required to result in the formation of a CÀ H borylated species. This contains only four aromatic protons ( Figure S2) and a resonance consistent with HBr formation (δ 1H = À 2.6 ppm observed only pre-vacuum treatment) -the by-product of CÀ H borylation (visible only in these sealed tube conditions). However, two new resonances were observed in the in-situ 11 B NMR spectrum suggesting two boron centers are incorporated into the product. Under these conditions the two new resonances in the in-situ 11 B NMR spectrum are at δ 11B = + 10.2 and + 42.8, which we assign as an equilibrium between 8 A and 8 B, with the BBr 3 associated with either O or N in 8 B as it is not removed in-vacuo. Note the 11 B chemical shifts were dramatically affected by the equivalents of BBr 3 used in this reaction (vide infra).
Crystals of the borylated product suitable for X-ray diffraction studies were grown by slow diffusion of pentane into a DCM solution of 8 A/8 B. The solid-state structure ( Figure 5, bottom) confirmed the presence of two boron molecules in the product, as in 8 A, in the form of a cation and a [BBr 4 ] À counteranion. In the structure of 8 A the closest B···BrÀ BBr 3 contact is at 3.474(6) Å, within the combined van der Waals radii for B and Br (Σ = 3.75 Å) [18] suggesting transfer of bromide between cation and anion is possible in solution (vide infra for more detailed structural discussion). To further examine the proposed equilibrium between 8 A and 8 B, variable temperature NMR spectroscopy studies were conducted. Incremental cooling of a DCM solution of 8 A/8 B (Figure 6, made from 6 and 2.5 equiv. of BBr 3 and heated and then dried in-vacuo) showed a gradual change in the 11 B NMR spectra whereby the boron centre in the cationic component was shifted gradually downfield from δ 11B = 37.5 to δ 11B ca. 42 ppm (in the range expected for an ArylB(OR)Br species), [19] whereas the second resonance was shifted upfield from δ 11B = À 16.6 to δ 11B = À 25 ppm (Figure 6), the latter is as expected for a discrete BBr 4 À anion. Thus, it can be concluded that upon cooling of the sample, the product favours the salt form, 8 A. We also conducted studies involving the addition of increasing amounts of excess BBr 3 to the 8 A/8 B mixture (from 0 to 6 equiv.) to probe the effect on both 11 B resonances ( Figure S5); while the cationic component moves to a limiting δ 11B = + 43.2 (consistent with the three   coordinate boron centre in 8 A), the second resonance shifts closer and closer to that for free BBr 3 (e. g. at 0 equiv. excess BBr 3 δ 11B = À 10.2 and at 6 additional equivalents of BBr 3 δ 11B = 30.1 ppm), as expected for a fast exchange of bromide between BBr 3 /BBr 4 À and an increasing quantity of BBr 3 . With an understanding of the products formed from 6/BBr 3 compound 7 was reacted with BBr 3 . While the Lewis adduct 7-BBr 3 forms rapidly, using 2.5 equivalents of BBr 3 and heating at 60°C (in a sealed vessel) resulted in incomplete conversion of 7-BBr 3 to the borylated species 9 A/9 B, after 16 hours. High levels of conversion (by NMR spectroscopy) to 9 A/9 B required 3 days of heating. It should be noted only 9 A is shown in Figure 7, but an analogous equilibrium occurs for 9 A (to form 9 B) as observed for 8 A/8 B. It is noteworthy that the borylation of 7 is slower than the borylation of 6, this is attributed to steric clash between the À NMe 2 moiety and the À CMe 2 in 7-BBr 3 (and the borenium derived from 7-BBr 3 ). Such interactions will presumably lead to rotation around the Me 2 CÀ C(O)NMe 2 bond to orientate the NMe 2 unit to reduce clash with the CMe 2 unit and thus position the boron centre unfavourably for CÀ H borylation (disfavouring formation of the key transition state -which itself maybe higher in energy due to the unfavourable interactions between NMe 2 /CMe 2 when the carbonyl is positioned appropriately). Regardless, the formation of 8 A/8 B and 9 A/9 B confirms that carbonyl directed electrophilic borylation tolerates CH 2 / CMe 2 groups in the linker. Therefore the preference to form compound 4 over B is attributed to a lower kinetic barrier to borylate the aniline derived unit in carbonyl directed borylation (which has been previously observed to undergo borylation at room temperature). [12a,b,15] As noted earlier, borylation selectivity is not controlled by the location of the HOMO/HOMO-1 in this case, as both these orbitals are principally located on the aryl unit of ArylCMe 2 . Despite this borylation still proceeds on the aniline derived unit to form 4.
The structure of 9 A was confirmed by X-ray diffraction studies (Figure 7 bottom). In the solid state, both compounds 8 A and 9 A show short BBr 4 ···B cation contacts between a bromide and boron of 3.474(6) and 3.487(5) Å for 8 A and 9 A, respectively. Notably, due to the planar N1À C8À O1À B1 unit 9 A has one of the methyls of the À NMe 2 orientated between the two CMe 2 . Thus, the structure of 9 A shows minimal deviation of planarity between the plane of the cyclic boronate ring and the À NMe 2 (max. 0.026 Å), whereas the À NMe 2 moiety in 8 A is deviated by upto 0.423 Å from the plane of the boracycle. Here, the entire À NMe 2 unit in 8 A is bent out of the plane possibly due to packing effects (as short contacts of 2.92-2.98 Å are observed between the À N(CH 3 ) 2 and BBr 4 À anion) of the cyclic boronate framework with the CÀ NMe 2 moiety remaining planar (angles around N Σ = 359.9°for both 8 A and 9 A). Additionally, the angles around C8 in both 8 A and 9 A sum to to 359.9 and 360.0°, respectively. Other notable features include the CÀ O bond in the cations which at 1.336(6) Å is lengthened relative to an uncoordinated amide C=O bond (typically~1.23 Å in length). [20] This is as expected for a carbonyl unit upon Lewis acid coordination. [21] Additionally, the N1À C8 bond length is 1.287(7) and 1.291(7) Å for 8 A and 9 A, respectively, slightly shortened relative to that in an uncoordinated amide (typicallỹ 1.35 Å). [20] Thus, the slightly contracted CÀ N and lengthened C=O suggest delocalisation of the cationic charge across the BÀ OÀ CÀ N unit, and thus the cationic products 8 A and 9 A can be considered to have iminium character as well borocation character (i. e. the positive charge in these cations will be localised predominantly on the least electronegative atoms, in this case C and B, as previously observed in other borenium  cations). [22] Next, the conversion of 8 A/9 A into bench stable products familiar to synthetic chemists was targeted. Attempts to form the pinacol-protected product derived from 8 A was successful using pinacol and NEt 3 , with the product having a δ 11B = 30.7 ppm. However, in our hands we could not isolate this product sufficiently pure due to its instability on silica. Isolation of the pure ortho borylated products as bench stable compounds was achieved by protecting at boron with 1,8diaminonaphthalene (1,8-Dan) to form À BDan protected 10 and 11 in 65 % and 34 % isolated yields, respectively, via a one-pot procedure (Scheme 2). The 11 B NMR spectra of the À BDan protected compounds each showed a single resonance at δ 11B = 30-31, consistent with a 3-coordinate boron centre. The absence of any significant BÀ O dative bond post Dan installation is similar to the 11 B NMR spectra obtained for the pivaloyl-directed borylation of anilines which show minimal coordination to the carbonyl directing group after pinacol installation. [12a] Finally, several additional substrates related to 6 were explored to test the generality of this directed electrophilic borylation process. The successful formation of compounds 12-14 (Scheme 2, bottom) demonstrates that substituents in the o, m and p positions are tolerated. While the CÀ H borylation step in the formation of 12 occurs under the same conditions as that for 6, having a fluoride meta to the CÀ H borylation site significantly retards the electrophilic borylation step. For this substrate borylation required 24 h at 100°C (in chlorobenzene) to produce reasonable yields of 13 post protection. The meta (to the acetylamide unit) bromo derivative leads to two inequivalent ortho CÀ H positions, but borylation is only observed at the less hindered ortho site. Again, the electrophilic borylation step is slower relative to that of 6, this time due to the electron withdrawing bromo group para to the CÀ H borylation position. While requiring longer reaction times/ higher temperatures (than 6), the successful formation of 13 and 14 nevertheless demonstrates that challenging substrates (in terms of substrates that are deactivated towards S E Ar) can be borylated with high selectivity and reasonable yield using this methodology. Finally, variation in the nitrogen substituents was investigated, with bulkier i Pr substituents used in place of methyl. This led to formation of 15 in reasonable yield, with the CÀ H borylation step proceeding to high conversion within 24 h at 60°C (by-in-situ NMR spectroscopy and by isolation of the intermediate before protection with 1,8-Dan).

Conclusion
In summary, two close analogues of directing templates effective in transition metal catalysed meta-CÀ H functionalisation were not able to effect meta directed CÀ H borylation via electrophilic borylation under a range of conditions. Thus bespoke (i. e., not transferred directly from transition metal catalysed approaches) covalently attached directed groups may be required to enable meta-selective electrophilic borylation. Removing any Lewis basic groups in the template that could affect ortho CÀ H borylation (via 5 or 6 membered boracycles) is one obvious next step emerging from this work. Indeed, the template with an amide unit can be used to effect amide directed ortho borylation which proceeds selectively on the aniline derived aryl unit in preference to the phenylacetyl unit. Phenylacetyl units were shown to be amenable to carbonyl directed ortho borylation with BBr 3 on heating. This process tolerated electron withdrawing substituents and groups at all three positions on the aryl unit.