Marina
Uzelac
,
Kang
Yuan
,
Gary S.
Nichol
and
Michael J.
Ingleson
*
EaStCHEM School of Chemistry, The University of Edinburgh, David Brewster Road, Edinburgh, EH9 3FJ, UK. E-mail: michael.ingleson@edinburgh.ac.uk
First published on 15th September 2021
Amido-zincates containing hydrides are underexplored yet potentially useful complexes. Attempts to access this type of zincate through combining amido-organo zincates and pinacolborane (HBPin) via Zn–C/H–BPin exchange led instead to preferential formation of amide–BPin and/or [amide–BPin(Y)]− (Y = Ph, amide, H), when the amide is hexamethyldisilazide or 2,2,6,6-tetramethylpiperidide and the hydrocarbyl group was phenyl or ethyl. In contrast, the use of a dipyridylamide (dpa) based arylzinc complex led to Zn–C/H–BPin metathesis being the major outcome. Independent synthesis and full characterisation of two LnLi[(dpa)ZnPh2] (L = THF, n = 3; L = PMDETA, n = 1) complexes, 1 and 3, respectively, enabled reactivity studies that demonstrated that these species display zincate type reactivity (by comparison to the lower reactivity of the neutral complex (Me-dpa)ZnPh2, 4, Me-dpa = 2,2′-dipyridyl-N-methylamine). This included 1 performing the rapid deprotonation of 4-ethynyltoluene and also phenyl transfer to α,α,α-trifluoroacetophenone in contrast to neutral complex 4. Complex 1 reacted with one equivalent of HBPin to give predominantly PhBPin (ca. 90%) and a lithium amidophenylzincate containing a hydride unit, complex 7-A, as the major zinc containing product. Complex 7-A transfers hydride to an electrophile preferentially over phenyl, indicating it reacts as a hydridozincate. Attempts to react 1 with >1 equivalent of HBPin or with catecholborane led to more complex outcomes, which included significant borane and dpaZn substituent scrambling, two examples of which were crystallographically characterised. While this work provides proof of principle for Zn–C/H–BPin exchange as a route to form an amido-zincate containing a hydride, amido-organozincates that undergo more selective Zn–C/H–BPin exchange still are required.
One rarely explored class of zincates are those containing a hydride substituent.1b The limited reports in this area have focused principally on determining the solution and solid state structures of the hydrido-zincates formed on combining MH or M with R2Zn (M = group one metal).7 Furthermore, there is only one amido-hydrido zincate8 and only one alkoxy-hydrido zincate9 well characterised to date to the best of our knowledge. The use of hydrido-zincates in organic transformations is also limited, and in all cases the hydrido-zincates are formed and used in situ.1b Nevertheless, notable reports from Uchiyama and co-workers using R2Zn/MH combinations revealed highly selective reductions with no competing Brønsted basic reactivity.10 Since this work hydrido-zincates have remained largely overlooked in organic transformations. Changing the latter would be facilitated by the development of alternative routes to access hydrido-zincates, such as methods that use bench stable hydride sources in place of MH or M (M = group 1 metal).
In contrast to anionic zinc hydrides, the chemistry of neutral and cationic molecular zinc hydrides has experienced significant progress, particularly in the past decade.11 This includes the development of multiple synthetic routes to access these zinc hydrides, some of the most common being: complexation of metastable ZnH2 with ligand(s); substituent exchange reactions between LnZn–X and MHn, or between LnZn–OR (or LnZn–NR2) and R4−xSi–Hx or LnZn–NR2 and R2HNBH3.11 More recently, we reported that a metathesis type reaction between neutral and cationic LnZn–R complexes (R = alkyl, aryl, alkenyl or alkynyl) and pinacolborane, HBPin, provides an alternative route to neutral and cationic molecular zinc hydrides (Scheme 1).12 While this approach is related to the formation of zinc-hydrides by reaction of zinc alkyls with LiAlH4 (reported for both neutral organozinc and alkylzincate complexes),13 HBPin is a bench stable hydride source, and the products from Zn–C/H–B exchange, R–BPin, are ubiquitous in synthesis (in contrast to the by-products using LiAlH4, e.g. [R4−XAlHx]−). The formal metathesis of organozincates with HBPin represents an unexplored route to form hydrido-zincate complexes concomitant with organoboranes that could underpin future catalytic developments mediated by zincates. Herein we report our studies into assessing the reactivity of amido-organozincates with HBPin as a potential route to form hydrido-zincates.
Entry | MNR2 | ZnR2 | Major producta |
---|---|---|---|
a Identity of major product(s) determined by 11B NMR spectroscopy. b 1 eq. of DME added. c THF was added dropwise until a solution was obtained. d 2 eq. HBPin used. For more details see ESI.† | |||
1 | LiHMDS | ZnEt2 | R2NBPin |
2b | LiHMDS | ZnPh2 | R2NBPin |
3c | KHMDS | ZnPh2 | R2NBPin |
4c | LiTMP | ZnPh2 | R2NBPin/PhBpin/[R2NBPin(Y)]− |
5b,d | LiTMP | ZnPh2 | R2NBPin/PhBpin |
6c | KTMP | ZnPh2 | [R2NBPin(Y)]−/PhBPin |
7 | (dpa)Li | ZnPh2 | PhBPin |
The more sterically hindered amide TMP was next explored, with equimolar combinations of LiTMP and ZnPh2 addition of one equivalent of HBPin afforded mixtures containing approximately 1:1:1 ratios of TMPBPin, PhBPin (δ11B ≈ 25 and 31, respectively) and a broad resonance centered at 6 ppm (presumably [(NR2)YBPin]− anion(s) (Y = H, Ph or NR2)) in contrast to outcomes with HMDS/Ph-zincates (where PhBPin is only observed at ca. 10%). These findings indicate that increasing the amide steric bulk is beneficial for improving selectivity towards Zn–C/H-BPin exchange, but even with TMP this still does not lead to high selectivity. Addition of further HBPin (now two equivalents in total relative to LiTMP/ZnPh2) led to complete consumption of HBPin and formation of an approximately 1:1 mix of PhBPin and TMPBPin as the major products (by 11B NMR spectroscopy) with minor quantities of borate salts (e.g. Li[PhxBHy], x + y = 4) also observed. This indicates that both Zn–C and Zn–NR2 units undergo complete exchange with HBPin, indicating that the undesired formation of insoluble homoleptic hydridozincates will occur with these systems. Similar results were also observed replacing LiTMP with KTMP.
We next turned our attention to replacing monofunctional amides (TMP/HMDS) with multifunctional amides: it was hypothesised that by preparing zincates containing less nucleophilic amides (e.g. aniline derived) incorporated into a chelating group could preclude Zn–N/H–BPin exchange. 2,2′-Dipyridylamide (dpa) was selected as it has three Lewis basic N-atoms incorporated within a flexible scaffold, favouring chelation to zinc,19 and a significantly lower nucleophilicity amide (potentially disfavouring Zn–N/H–B exchange). Mixing Lidpa with ZnPh2 and reacting this with one equivalent of HBPin afforded PhBPin as the major product (entry 7 and Fig. S14†) with no PinB-NR2 observed (minor borate resonances at δ11B 6–8 ppm were observed consistent with formation of [R2N(Y)BPin]− anions, but these were only 20% the intensity of the PhBPin resonance in the 11B NMR spectra). Importantly, a new sharp resonance in the region 4.5–5.7 ppm (depending on the solvent) consistent with a Zn–H was observed in the 1H NMR spectra suggesting successful formation of a soluble and stable Zn–H species derived from Zn–C/H–BPin exchange. Encouraged by this observation, we set out to identify the constitution of the relevant dpa-based zincates formed prior to addition of HBPin to enable more controlled studies and more insight into the putative hydrido-dpa-zincate.
Deprotonation of dpaH with ZnPh2 in toluene afforded on crystallisation trimeric [(dpaZnPh)3] (2) in 77% isolated yield, which could be converted into mono-zinc complex 1 by addition of one equivalent of PhLi in THF. Addition of the chelating ligand PMDETA (N,N,N′,N′′,N′′-pentamethyldiethylenetriamine) during zincate synthesis afforded the PMDETA analogue of 1, [(PMDETA)Li(μ-dpa)ZnPh2] (3) in 65% isolated yield. In addition, a neutral monomeric Zn-complex incorporating the dpa scaffold was prepared for reactivity comparison studies (vide infra). Adding an equivalent of 2,2′-dipyridyl-N-methylamine (Me-dpa) to a solution of ZnPh2 afforded adduct [(Me-dpa)·ZnPh2] (4). Complexes 1–4 were all fully characterised in solution (comparing well with the data previously reported for other dpa-zincates (e.g. [(TMEDA)2Na2(μ-dpa)2Zn(tBu)2], A, and [Na(THF)6{Zn(tBu)2(dpa)Zn(tBu)2}], B)19 and L2ZnPh2 systems, e.g. TMEDA·ZnPh2)20 and in the solid state by single crystal X-ray diffraction studies. It should be noted that 1H DOSY NMR studies are consistent with 2 persisting as an oligomer in solution by comparison to parameters derived from 1H DOSY studies on monomeric 4 (see ESI, Fig. S38 and S39†).
In the solid state (Fig. 1) 2 is a trimer in which each dpa assumes an anti/anti conformation to act as a bridge between neighbouring Zn atoms. Each Zn atom is in a distorted tetrahedral environment, bonded to the Namido of one dpa unit and chelated within the dipyridyl pocket of another dpa, with its coordination completed with a terminal Ph ligand. Within the trimer, the three Zn-atoms define a plane, to which all three dpa ligands are effectively perpendicular, while the Ph rings are 7.20°, 22.40° and 34.87° tilted to the Zn3 plane. The Zn–C bond distances are in close agreement with that reported for the dimer [(dpaZntBu)2], while the Zn–N bonds are only slightly shortened in 2 (cf. for [(dpaZntBu)2] Zn–N = 2.116(2) Å; 2.079(2) Å; 2.070(2) Å).19
Fig. 1 ORTEP-representation of [(dpaZnPh)3] (2) with ellipsoids at 50% probability level. The disorder component, hydrogen atoms and crystallisation solvent have been omitted for clarity. |
In both 1 and 3 (Fig. 2a and b), the anionic dpa ligand is also in an anti/anti conformation, but now bridges between a donor-capped Li+ cation coordinated to the Namido with a ZnPh2 unit chelated within the dipyridyl pocket, similar to that observed in zincates A and B.19 Close comparison of 1 and 3 (Table 2) reveal that the tetrahedral environment of Li in 3 is more distorted (τ4 = 0.7) from ideal than in 1 (τ4 = 0.9) but the average angles and Li–Namido bond lengths are identical within error. In both complexes Zn is in a distorted N2C2-tetrahedral environment, displaying essentially identical bond distances, average angles and the extent of distortion from an ideal tetrahedral environment (τ4). However, the C–Zn–C angle is significantly different at 137.17(8)° for 1 and 128.07(4)° for 3, the former angle is even greater than the ZntBu2 species, A (130.31(7)°) and B (131.9(1)°),19 where a larger steric effect from the tBu groups would be expected. Previous studies have shown that the narrowing of the C–Zn–C bond angle in diorganylzinc complexes with C2ZnN2 coordination environments increases the Lewis acidity of Zn-centre.21 However, based on reactivity studies and DFT calculations (vide infra) this C–Zn–C angle disparity does not indicate a significantly different electronic structure for 1 and 3. Finally, the tetrahedrally coordinated Zn centre in 4 (Fig. 2c) exhibits very similar parameters to those found in 3, with a C–Zn–C angle of 126.20(7)°.
1 | 2 | 3 | 4 | 5 | |
---|---|---|---|---|---|
Zn–C (Å) | 2.004(2) | 1.987(2)–1.992(2) | 2.0088(11) | 1.9988(18) | 1.973(4) |
2.010(2) | 2.0018(10) | 1.9964(19) | 1.971(5) | ||
Zn–Npy (Å) | 2.0952(16) | 2.035(2)–2.056(2) | 2.0966(9) | 2.1032(15) | 2.059(4) |
2.0985(17) | 2.1071(9) | 2.1309(15) | 2.064(4) | ||
Zn–Namide (Å) | — | 2.047(2)–2.063(2) | — | — | — |
C–Zn–C (°) | 137.17(8) | — | 128.07(4) | 126.20(7) | 130.60(19) |
Li–Namide (Å) | 2.047(4) | — | 2.028(2) | — | 2.028(9) |
The solid state geometries of complexes 1 and 3 were used for single point calculations at the B3PW91/6-311G(d,p) (H, Li, C, N, O)/lanl2dz (Zn)//PCM(THF) level (PCM = polarisable continuum model). Notably, these calculations show that despite the difference in the C–Zn–C bond angle between 1 and 3 the two structures possess effectively identical NBO charge distributions (e.g. axial ipsoPh charge = −0.499e and −0.506e for 1 and 3, respectively) and frontier orbital energies and character. For example, the HOMO of 1 and 3 (Fig. 3) both contain Zn–C σ bonding character and are effectively identical in energy (−5.02 eV and −5.06 eV), while the charges at Zn are within 0.006e in 1 and 3 (+1.169 and +1.163e, respectively). The most notable difference is in the NBO charge at lithium which is +0.875e for 1 and +0.846e for 3; this small difference is presumably due to a more electrostatic dominated interaction between (THF)O3–Li (in 1) than between (PMDETA)N3–Li in 3.
Fig. 4 Formation of [(THF)3Li(μ-dpa)Zn(CCC6H4Me)2] (5). ORTEP representation of 5 with ellipsoids at 30% probability level and hydrogen atoms omitted for clarity. |
Single crystal X-ray diffraction analysis of 5 revealed it to have an essentially isostructural Li-dpa-Zn core to 1 and 3 (Table 2). The most notable metric is the C–Zn–C angle which is different between 1 and 5, with 5 having a compressed C–Zn–C angle (130.60(19)°) relative to that in 1 (137.17(8)°), despite both having an identical lithium environment. Solution studies on 5 were precluded due to the poor solubility of crystals of 5, even in THF. Furthermore, during reactions to form 5 small amounts of two other alkynyl-containing species also were observed which could correspond to different dpaZn structural motif(s) or to products obtained by alkyne deprotonation partially proceeding through Li–Namido fragment. The same minor species are observed even when 5 was prepared by reacting dialkynylzinc with Lidpa. Finally, compound 3 reacts with two equivalents of 4-ethynyltoluene in a comparable manner to 1, with all alkyne consumed within 2 hours using 3, indicating similar reactivity between 1 and 3 in metalation reactions supporting the comparable electronic structures in the Ph2Zn-dpa units of 1 and 3 found by DFT calculations.
We next turned our attention to phenyl transfer to α,α,α-trifluoroacetophenone (to facilitate reaction monitoring by 19F NMR spectroscopy). While neutral complexes 2 and 4 afforded no transfer of a phenyl group (by 19F NMR spectroscopy) even after days, complex 1 displayed significant reactivity and afforded 2,2,2-trifluoro-1,1-diphenylethanol (6) in 71% yield after only 2 h reaction time at room temperature (Scheme 3). This observation is in line with previous reports on the sluggish reactivity of neutral diorganozinc reagents towards aldehydes and ketones that can be enhanced by formation of zincates.22 This further indicates that 1 displays zincate type reactivity. Monitoring the reaction of 1/α,α,α-trifluoroacetophenone in situ (prior to work up) revealed formation of resonances consistent with 2 as the only observed zinc containing by-product from phenyl transfer. In contrast to fast phenyl transfer with 1, compound 3 led to no phenyl transfer to α,α,α-trifluoroacetophenone after 2 h by 19F NMR spectroscopy (though slow phenyl transfer does occur at longer reaction times). Given the similar calculated electronic structures and comparable reactivity towards terminal alkynes observed for 1 and 3 this disparity was surprising. It is attributed to the lithium centre in 1 being more accessible for coordination of a carbonyl and that this Lewis acid activation of the ketone accelerates phenyl transfer (consistent with observations from reported DFT calculations on lithium-zincates reacting with carbonyls).10 In contrast, PMDETA would bind lithium more strongly and disfavour coordination of the ketone to lithium. A similar observation has been previously reported where TMEDA addition retards nucleophile transfer from a lithium zincate to carbonyl containing electrophiles.23 This hypothesis is supported further by the addition of PMDETA to 1 resulting in formation of 3, confirming the stronger binding affinity that PMDETA has towards lithium in these lithium zincates relative to three molecules of THF.
With confirmation of zincate type reactivity exhibited by 1 in hand we explored next the reactivity of 1 towards HBPin targeting an amido-hydrido-zincate by Zn–C/HBPin exchange.
Mass spectrometry on equimolar 1/HBPin reactions only showed fragmentation products such as [dpaZnPh]˙+. It should be noted that combining HBPin/1 in aromatic solvents also formed some insoluble material, however performing these reactions in more polar solvents, e.g. THF, led to more complex mixtures by NMR spectroscopy, (though PhBPin is still the dominant boron containing species when reactions are performed in THF), thus aromatic solvents were the preferred reaction media.
On standing in benzene 1/HBPin mixtures slowly produced more insoluble material over several days, with the only significant species ultimately left in solution being PhBPin. The amount of PhBPin formed was quantified versus an internal standard which revealed >90% of HBPin had converted into PhBPin indicating that Zn–Ph/H–BPin exchange is the dominant outcome on combining equimolar 1 and HBPin. Finally, attempts to access the putative amido-hydrido-zincate 7-A by other routes were unsuccessful (see ESI† for more details).
Fig. 7 1H DOSY NMR spectrum of reaction mixture 1 with 2 eq. of HBPin in C6D6 at room temperature after 45 min. |
On standing the δ1H 5.10 resonance slowly decreases in intensity, concomitant with the very broad δ11B 5 ppm resonance being replaced with several other boron resonances. The most intense new resonances correspond to [BH4]− species (δ11B = −40 ppm quintet) and a sharper δ11B resonance at +5 ppm (consistent with [diamido–BPin]−).18 This decomposition is consistent with previous work on the reactivity of anionic nucleophiles with HBPin, that form metastable [HBPin(Y)]− anions that undergo substituent scrambling, this indicates that the 1:2 1/HBpin reaction mixture contains significant [(Y)HBPin]−.24,25 Attempts to crystallise these 1:2 reaction mixtures to isolate the Zn–H species 7-B also were unsuccessful, with the only species isolated being a small quantity of the redistribution product, 8, containing both [BH4]− and [(R2N)2BPin]− moieties consistent with the +5 and −40 ppm resonances observed in the 11B NMR spectra. The PMDETA analogue, 3, displayed comparable outcomes to combinations of 1 with HBPin, undergoing Zn–Ph/H–BPin metathesis (albeit more slowly than 1) to give PhBPin and Zn–H 7-B (δ1H 5.1 ppm), with scrambling to form BHx species (x > 1) also observed on standing. Finally, attempts to use catecholborane, HBCat, in place of HBPin led to much more complex outcomes, rapidly forming new boron containing compounds including: [BCat2]−, species containing BH2 and BH3 units and [BH4]−. Single crystals of a substituent scrambled product, [dpa3Zn2][BCat2] (9), were isolated from these mixtures (see ESI, Fig. S132†). These observations indicate that both the borane and dpaZn fragments can undergo exchange processes, and that these occur much more rapidly with BCat derived species.
Scheme 4 Top and bottom, disparate outcomes on reaction of Zn–H species 7-A and 7-B derived from 1/HBPin with α,α,α-trifluoroacetophenone. Middle, alcohols formed by carbonyl reduction using 7-A. |
In contrast, the combination of the reaction mixture derived from 1:2 1/HBPin with one equivalent of α,α,α-trifluoroacetophenone led to formation of a significant quantity of the hydroboration product 14 (by 19F NMR spectroscopy), indicating the presence of significant [(Y)HBPin]− species. Furthermore, the Zn–H resonance at 5.10 ppm is not consumed on addition of this ketone, indicating that ketone reduction is achieved by a B–H species not a Zn–H complex. Note, in a control reaction using HBPin under identical conditions no hydroboration of α,α,α-trifluoroacetophenone is observed. The formation of significant hydroboration product 14, combined with the spectroscopic data, suggests that on using two equivalents of HBPin (with respect to 1) the outcome involves a single Zn–C/H–BPin exchange followed by formation of a [H(Y)BPin]− species that is metastable (towards substituent redistribution).
Footnote |
† Electronic supplementary information (ESI) available: Full experimental procedures, crystallographic data and additional figures. CCDC 2101027–2101033. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/d1dt02580e |
This journal is © The Royal Society of Chemistry 2021 |