Jay J.
Dunsford
,
Ewan R.
Clark
and
Michael J.
Ingleson
*
School of Chemistry, University of Manchester, M13 9PL, UK. E-mail: Michael.ingleson@manchester.ac.uk
First published on 29th October 2015
New aryl- and heteroarylboronate esters chelated by dipropanolamine are synthesised directly from boronic acids. The corresponding anionic borates are readily accessible by deprotonation and demonstrate an increase in hydrocarbyl nucleophilicity in comparison to other common borates. The new borates proved competent for magnesium or zinc additive-free, direct boron-to-iron hydrocarbyl transmetallations with well-defined iron(II) (pre)catalysts. The application of the new borate reagents in representative Csp2–Csp3 cross-coupling led to almost exclusive homocoupling unless coupling is performed in the presence of a zinc additive.
Fig. 1 Proposal to favour hydrocarbyl transmetallation over alkoxide transfer to iron from boronate reagents. |
Borate reagents incorporating tethered, trianionic ligands are known,19 with the triolborates particularly noteworthy,20 however these are non-productive in Fe catalysed SM couplings.12 Borates incorporating a trianionic 6,6-bicyclic chelate are rare,21 but are attractive as they will afford a reduction in ring strain in the conjugate 3-coordinate boronate ester formed post hydrocarbyl transfer relative to 5,5-bicyclic and 6,6,6-tricyclic (e.g. triolboronate) systems. This would facilitate the planarisation of the neutral 3-coordinate boronate ester by-product and thus its stabilisation through increased π-orbital overlap, an outcome enhanced further by the greater multiple bond character present in B-NR2 relative to B-OR; both factors were envisaged to enhance the nucleophilicity of the anionic hydrocarbyl-borate.
Combining 1.1 equiv. of dipropanolamine (which can be easily prepared on multi-gram scales from cheap, commercially available reagents)22 with phenylboronic acid (1 equiv.) in THF afforded the desired neutral boronate ester 1a within 10 minutes at ambient temperature. The poor solubility of 1a in THF allows for its isolation by filtration and subsequent recrystallisation from hot acetone affords analytically pure 1a in 89% yield (9.74 g) (Fig. 2). The rigid nature of the 6,6-bicyclic chelate is evident in the 1H NMR spectra of 1a through the diastereotopic nature of the CH2 protons within the chelate backbone, a feature also observed in related 5,5-bicyclic systems.23 It is noteworthy that 5,5-bicyclic analogues demonstrate fluxional behaviour at temperatures above 40 °C, whereas 1a is considerably more conformationally rigid, demonstrating no fluxional behaviour up to 110 °C (in d6-DMSO), analogous to the extremely robust MIDA boronate esters.24 This simple esterification protocol can be extended to a number of aryl- or heteroarylboronic acids, affording neutral boronate esters 1a–1q in good to excellent isolated yields (97–72%) (Fig. 2).
Generation of the anionic borates is simple as exemplified by the formation of K[2a] which can be achieved in quantitative yield via deprotonation of 1a with a variety of bases (KH, tBuOK, KOMe) in anhydrous THF.25 The employment of KOH leads only to the recovery of neutral boronate ester, 1a. This observation correlates with the fact that K[2a] slowly hydrolyses to 1a (and presumably KOH) with prolonged storage in the solid or solution state under air. With methods for the generation of the anionic borates in hand, we next focused upon calculating relative phenyl ion affinities (PhIAs) to assess the relative nucleophilicity of K[2a] and other common borate anions. PhIAs were calculated from the isodesmic reactions between tetraphenylborate and the appropriate neutral boron Lewis acid (Table 1). This methodology is analogous to previous approaches for calculating hydride and chloride ion affinities (HIAs and CIAs) of boron Lewis acids, where calculations of HIA correlated well with hydride transfer reactivity.26,27 Analysis of the calculated PhIA values revealed neutral borane 3 to have a substantially lower PhIA (hence K[2a] will provide a better source of Ph−) than other common borate reagents tested (Table 1). The alleviation of ring strain in 3 relative to the 5,5-bicyclic analogue derived from diethanolamine (5,5-ONO, entry 4) and to a greater extent the 6,6,6-tricyclic containing triolborane (entry 1) does indeed have a considerable effect on PhIA and thus the energetics of aryl transfer.
Entry | Lewis acid | Abbreviation | PhIAa (kcal mol−1) |
---|---|---|---|
a PhIAs calculated at the M06-2X/6-311G+(d,p) level of theory with incorporation of a dichloromethane PCM solvent model. b In parenthesis the PhIA is calculated using atomic coordinates from the solid state structure of K[2a]. | |||
1 | Triolborane | −22.8 | |
2 | BF3 | — | −16.3 |
3 | BPh3 | — | 0.0 |
4 | 5,5-ONO | +17.0 | |
5 | B(OMe)3 | — | +22.7 |
6 | B(Pin)(tBu) | +23.2 | |
7 | B(Pin)(nBu) | +23.9 | |
8 | 3 (This work) | +36.3 (+31.3)b |
The trends observed through calculated PhIAs were probed experimentally by phenyl ion transfer reactions. Initial studies commenced with K[2a]/3 and boron reagents possessing considerably different calculated PhIAs. The addition of K[2a] to triolborane in THF resulted in slow dissolution of triolborane (which is poorly soluble in THF), with initial 11B NMR spectra (after ca. 10 minutes at 20 °C) showing a mixture of K[2a] (δ11B = 2.5) and a new species (δ11B = 4.4) which is tentatively attributed to reversible Lewis adduct formation between K[2a] and triolborane, presumably by coordination of the N or O nucleophilic sites in K[2a] to the Lewis acidic boron centre of the triolborane. Nevertheless, heating this mixture at 60 °C for 18 h resulted in complete disappearance of both of these signals, full dissolution of the triolborane and formation of 3 (δ11B = 20.0) and [Ph-triolborate]− (δ11B = 3.6) as the major species. The new boron species at δ11B = 20 was confirmed as the expected neutral 3-coordinate boron by-product of hydrocarbyl transmetallation via the independent synthesis and characterization of 3 from dipropanolamine and Me2S·BH3.25 The reaction of K[2a] with Et2O-BF3 proceeded more rapidly with no starting materials observed after <10 min at 20 °C (by 11B NMR spectroscopy), instead 3 and [PhBF3]− are the major species observed, along with an unidentified intermediate (δ11B = 0.1) observed at short times which is fully consumed on heating to 60 °C.
It is more significant in the context of this investigation to experimentally benchmark the PhIA of 3 against boron Lewis acids whose Ph-borate anions transmetallate to Fe (with or without additives). However, attempts to experimentally confirm the relative PhIA of BPh3 and 3 were frustrated by solubility issues (in THF and DCM) along with the formation of an intermediate (δ11B = 0.8) that persisted even on heating to 60 °C for 18 h, this species frustrated crystallisation attempts and is tentatively assigned to a Lewis adduct between K[2a] and BPh3. In contrast, combination of K[2a] and B(OMe)3 resulted in rapid phenyl transfer at 20 °C (with a short lived intermediate again observed, δ11B = 3.1), to ultimately form 3 and [PhB(OMe)3]−, confirming the relative PhIA of 3 and B(OMe)3 (Scheme 1). The combination of K[2a] and B(Pin)(nBu) resulted in only a minor amount of 3 formed after 16 h at 60 °C. We attribute the limited phenyl transfer to a kinetically slow process as opposed to an equilibrium favouring B(Pin)(nBu) and K[2a] as the reverse reaction of 3 with [PhB(Pin)(nBu)]− resulted in minimal formation of [2a]− after 16 h at 60 °C.
Encouraged by the calculated PhIA value of 3 and the stoichiometric phenyl ion transfer experiments, we moved on to probe whether high hydrocarbyl nucleophilicity in [2a]− would translate to facile boron-to-iron hydrocarbyl transmetallation with common iron(II) (pre)catalysts. We began this investigation with the (pre)catalyst, [Fe(dppe)Cl2]n4 (dppe = 1,2-bis(diphenylphosphino)ethane) (4 was found by X-ray diffraction studies to exist as a 1D coordination polymer in the solid state on recrystallisation from THF).28In situ generation of K[2a] with KH (1 equiv.) in anhydrous THF followed by the addition of 4 (0.5 equiv.) and dppe (0.5 equiv.) then agitation at ambient temperature led to no observable reaction (by 11B{1H} NMR spectroscopy, with K[2a] the only signal observed) after 1 h. However, heating the reaction mixture for 10 minutes at 60 °C in THF (or in toluene) led to the pale green reaction mixture becoming deep red in colour. Analysis of the reaction mixture by 1H NMR spectroscopy was uninformative, however the 11B{1H} NMR spectra revealed the complete disappearance of the 4-coordinate signal of K[2a] and the presence of a single new boron containing species in the three-coordinate region at δ11B 23.9 ppm attributed to 3.29 Isolated samples of K[2a] generated from tBuOK or KOMe produced analogous outcomes, provided all protic by-products from boronate generation (tBuOH/MeOH) were removed prior to use due to their detrimental effect on these specific iron complexes. The requirement for raised temperatures for conversion of K[2a] to 3 on addition of 4/dppe is in contrast to the reaction of Li[(Ph)B(pin(tBu)] (5) with 4 (0.5 equiv. + 0.5 equiv of dppe) which in our hands was found to react within minutes at ambient temperature in the absence of MgBr2 to produce tBuBPin (by 11B NMR spectroscopy). Thus whilst [2a]− is calculated to be thermodynamically a more nucleophilic source of Ph− than 5 it is kinetically slower to react by phenyl transfer with the Fe species present in solution under these conditions.
With the in situ11B{1H} NMR data being consistent with hydrocarbyl loss from [2a]− we sought evidence for the formation of arylated iron species. Most notably the generation of significant quantities of biphenyl (ca. 0.5 equiv.) post aryl transfer was confirmed by analysis of the reaction mixture by GC-MS against a calibrated internal standard,25 indicative of Fe-Ph species.14b Consumption of all K[2a] on addition of 0.5 equiv. of 4 (by 11B{1H} NMR spectroscopy) suggests the transfer of two aryl equivalents to iron, although the transfer of just one aryl equivalent to Fe and formation of an adduct between an Fe species and a second equivalent of [2a] (which may not be observable in the 11B NMR spectra) is also feasible. We disfavour a reaction stoichiometry other than 2:1 as increasing the ratio of K[2a]:Fe from 2:1 to 3:1 and to 4:1 resulted in an increasing amount of K[2a] persisting in solution (by 11B NMR spectroscopy). Analogous outcomes were observed on combining K[2c] (or K[2d]) and 4 in a 2:1 ratio, with significant biaryl formed in each case (by GC-MS versus an internal standard). Repeated efforts to isolate any Fe species from transmetallation to iron complexes with K[2a] were unsuccessful in our hands.
[Fe(dpbz)2Cl2], 6 (0.5 equiv.) (dpbz = 1,2-bis(diphenylphosphino)benzene) reacted analogously with K[2a] or K[2b], with two equivalents of borate consumed per Fe centre (by 11B NMR spectroscopy) and significant biphenyl or bi-tolyl (ca. 0.5 equiv. by GC-MS) formed again indicating significant transmetallation from boron to iron. The previously reported iron(I) complex, [Fe(dpbz)2(p-Tol)] 730 (Fig. 3) was observed from this reaction mixture by X-band continuous-wave EPR measurements at 120 K further confirming successful boron to Fe transmetallation. The spectrum of 7 is in excellent agreement with previous X-band CW-EPR measurements obtained for 7 by Bedford and co-workers30 and with an independently prepared sample of 7 obtained from the reaction of 6 with p-TolMgBr (2 equiv.) (Fig. 3). Whilst the presence of EPR and NMR spectroscopy silent Fe complexes cannot be precluded from these reactions the formation of significant quantities of the homocoupled biaryl compounds and the observation of 7 by EPR spectroscopy combined do confirm significant boron to iron hydrocarbyl transmetallation using [2x].
With boron-to-iron hydrocarbyl transmetallation confirmed we moved on to evaluating the efficiency of K[2a] in iron catalysed Csp2–Csp3 SM cross-coupling. Employing K[2a] and 3-methoxybenzyl bromide in the presence of 4 + dppe (both at 10 mol% loading) led to the complete consumption of 3-methoxybenzyl bromide and almost exclusive formation of 1,2-bis(3-methoxyphenyl)ethane with only a very minor amount of heterocoupled product (by GC-MS). This observation is in contrast to the outcomes reported employing the tBuLi activated arylpinacol boronate esters, such as 5, or Grignard activated triorganoboranes, which under similar conditions, in the presence of magnesium or zinc additives led predominantly to heterocoupling products.14,15 Analogous homocoupling outcomes to that observed through the use of K[2a] were found in the employment of the lithium salt Li[2a], hence the difference in reactivity between M[2a] and 5 is not cation derived. Furthermore, repeating the cross coupling reactions using Li[2a] with MgBr2 additive, led to homocoupling again being the dominant pathway.
Diarylzinc co-catalysts are used widely in Fe SM catalysis, but it is important to note that species formulated as aryl2Zn are actually often ionic zincates, and this can significantly impact reactivity in cross-coupling.31 For example, the previously reported attempted synthesis of (p-tol)2Zn led instead to the zincate [Mg(THF)4Br2(Zn(p-tol)2)2]n, 8 (on recrystallisation, Scheme 2, inset).32 Using 10 mol% of a stock solution of 8 (equating to a total of 0.08 mmol of tolyl nucleophile) and 5 mol% 4/dppe resulted in the heterocoupling product from K[2a] and cyclo-heptylbromide being the major product (Scheme 2). However, the product distribution from this reaction is significantly less selective towards heterocoupling than cross coupling cycloheptylbromide using [tBuB(Pin)Ar]− (or [Ar(iPr)BBN]−) with magnesium or zinc additives.14,15 Repeating the coupling of cyclo-heptylbromide with K[2a] in the absence of 8, with or without MgBr2 as an additive, led to no heterocoupling product formation under a range of conditions. The omission of 4 also resulted in no heterocoupling confirming the importance of Fe under these conditions.
The disparity between magnesium and zinc additives led us to investigate the direct reaction between MX2 (M = Zn or Mg, X = Br or Cl) and K[2a]. Combination of MgBr2 and K[2a] in THF resulted in no observable reaction at 20 °C and 60 °C (after 24 h by 11B NMR spectroscopy). In contrast, K[2a] reacts rapidly with ZnCl2 with complete consumption of K[2a] and formation of 3 as the only new observable boron containing product after 5 h in THF at 20 °C or after 30 min in toluene at 60 °C (by 11B NMR spectroscopy). This is suggestive of transmetallation from boron in [2a]− to ZnCl2 but not to MgBr2, consistent with the disparate additive performance in Fe catalysed cross coupling reactions using K[2a]. Related transmetallation from boron to ZnCl2 has been previously observed, e.g. using [sBu3B(p-tolyl)][Mg(solv)Cl],15b which is a boron based aryl nucleophile also effective in Fe catalysed SM reactions. It is notable that less nucleophilic borates such as [PhBF3]− and [PhBtriolborane]− give no heterocoupling under Fe catalysed SM conditions even in the presence of zinc additives.12 In control reactions 2 equivalents of [PhBF3]− (or [PhBtriolborane]−) were heated in THF at 60 °C for 16 h with ZnCl2. Subsequent analysis of these reaction mixtures in d6-DMSO (in which all boron species, e.g., [PhBY3]− and the conjugate neutral borane, are soluble) revealed no evidence for transmetallation to zinc with [PhBF3]− (or [PhBtriolborane]−) being the only major boron product observed (by 11B NMR spectroscopy).
10-phenyloctahydro-[1,3,2]oxazaborinino[2,3-b][1,3,2]oxazaborinin-5-ium-10-uide21 (1a): Prepared according to the general procedure. Phenylboronic acid (6.10 g, 50.0 mmol) and 3,3′-azanediylbis(propan-1-ol) (7.33 g, 55.0 mmol) in tetrahydrofuran (200 mL) afforded 1a as a free-flowing white solid (9.74 g, 89%). 1H NMR (400 MHz, DMSO-d6, 298 K): 7.39 (2H, d, J = 6.7 Hz, o-CH); 7.19 (2H, t, J = 7.5 Hz, m-CH); 7.09 (1H, t, J = 7.5 Hz, p-CH); 5.90 (1H, bs, NH); 3.76–3.70 (2H, m, CH2); 3.50–3.44 (2H, m, CH2); 3.09–3.02 (2H, m, CH2); 2.86–2.80 (2H, m, CH2); 1.81–1.72 (2H, m, CH2); 1.51–1.42 ppm (2H, m, CH2). 13C{1H} NMR (100 MHz, DMSO-d6, 298 K): 131.8; 126.9; 125.5; 59.7; 46.0; 24.3 ppm. 11B{1H} NMR (128.4 MHz, DMSO-d6, 298 K): 3.2 ppm. HRMS (ESI) m/z: calculated for [M + H]+, C12H19BNO2+, 220.1509, found: 220.1516. Crystals suitable for X-ray diffraction were grown by the slow evapouration of a concentrated acetone solution of 1a at ambient temperature.
See ESI† for further details for 1b–1q synthesised via the general procedure.
K[2a]: Prepared according to the general procedure. 1a (750.0 mg, 3.42 mmol) and KH (136.8 mg, 3.42 mmol) in tetrahydrofuran (10 mL) afforded K[2a] as a free-flowing white solid (862 mg, 98%).1H NMR (400 MHz, THF-d8, 298 K): 7.54 (2H, d, J = 6.9 Hz, o-CH); 7.09 (2H, t, J = 7.2 Hz, m-CH); 6.96 (1H, t, J = 7.5 Hz, p-CH); 3.61–3.56 (2H, m, CH2); 3.52–3.43 (2H, m, CH2); 3.34–3.25 (2H, m, CH2); 2.74–2.66 (2H, m, CH2); 1.88–1.75 (2H, m, CH2); 1.36–1.26 (2H, m, CH2) ppm. 13C{1H} NMR (100 MHz, THF-d8, 298 K): 135.1; 127.1; 125.0; 62.6; 51.8; 31.1 ppm. 11B{1H} NMR (128.4 MHz, THF-d8, 298 K): 2.5 ppm. Anal Calcd for C12H18BKNO2: C, 56.04; H, 6.66; N, 5.45. Found: C, 55.84; H, 6.49; N, 5.39. Crystals suitable for X-ray diffraction were grown by the slow evaporation of a concentrated THF solution of K[2a] at ambient temperature.
Footnote |
† Electronic supplementary information (ESI) available: Full experimental, computational and X-ray crystallographic details and all relevant NMR spectra. CCDC 1010524, 1010526 and 1058782. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/c5dt03835a |
This journal is © The Royal Society of Chemistry 2015 |