Microsolvation and sp2-stereoinversion of monomeric α-(2,6-di-tert-butylphenyl)vinyllithium as measured by NMR

Summary The β-unsubstituted title compound dissolves in THF as a uniformly trisolvated monomer, whereas it forms exclusively disolvated monomers in tert-butyl methyl ether, Et2O, TMEDA, or toluene with TMEDA (1.4 equiv). This was established at low temperatures through the observation of separated NMR signals for free and lithium-coordinated ligands and/or through the patterns and magnitudes of 13C,6Li NMR coupling constants. An aggregated form was observed only with Et2O (2 equiv) in toluene as the solvent. The olefinic geminal interproton coupling constants of the H2C= part can be used as a secondary criterion to differentiate between these differently solvated ground-states (3, 2, or <2 coordinated ligands per Li). Due to a kinetic trisolvation privilege of THF, the cis/trans sp2-stereoinversion rates could be measured through analyses of 1H NMR line broadening and coalescence only in THF as the solvent: The pseudomonomolecular (because THF-catalyzed), ionic mechanism is initialized by a C–Li bond heterolysis with the transient immobilization of one additional THF ligand, followed by stereoinversion of the quasi-sp2-hybridized carbanionic center in cooperation with a “conducted tour” migration of Li+(THF)4 along the α-aryl group within the solvent-separated ion pair.


Introduction
Organolithium compounds tend to aggregate in solution unless the structure of their carbanionic part favors the nonaggregated (monomeric) species [2]. For example, tetrameric and dimeric n-butyllithium (n-BuLi) can coexist [3] in a highly mobile equilibrium, so that their reliable kinetic differentiation [4] required rapid-injection NMR techniques at very low temperatures. Without such techniques, however, a reactivity study may furnish dubious kinetic evidence if two or more equilibrium components in unknown proportions contribute to a global reaction rate. Therefore, it may be preferable to investigate (at least in opening studies) a purely monomeric species, in particular because this may be the most reactive form [2] that can dominate the reactivity profile of its mixture with aggregates. On the other hand, the analytic characterization of such a reactive species may be problematic due to a diminished kinetic stability against certain solvents, especially tetrahydrofuran (THF). For instance, barely avoidable decomposition products may interfere with traditional methods of molecular mass determinations that measure the averaged colligative properties of a solution. Fortunately, the 13 C NMR techniques [5] exemplified further below can identify the ground-state structures of organolithiums even in partially decayed or contaminated solutions. However, the important problem of microsolvation in the coordination shell of lithium can normally not be addressed by NMR spectroscopy, because scrambling of coordinated and free ligands is usually so fast that only averaged resonances can be recorded even at low temperatures. Among the rare opposite cases are the very strong donor ligand (Me 2 N) 3 PO (HMPA) [6], or THF at the endocyclic Li of dimeric Me 2 CuLi&LiCN [7], or intramolecular (chelating) donor functions [8], whereas nonchelating monodentate ethereal donor ligands such as THF, Et 2 O, and tert-butyl methyl ether (t-BuOMe) normally show only the averaged signals. However, our sterically congested model system [9] 1 exhibited separate resonances of the latter three donor ligands ("Don" in Scheme 1) above the melting points of the solvent mixtures, so that the microsolvation numbers d could be measured by simple NMR integrations. On this basis, dimerization equilibria of 1a and 1b were analyzed [10] with a proper allowance for the differing microsolvation of the monomeric and dimeric components. The alkenyllithiums 1a-e and 2 are exclusively trisolvated (d = 3) monomers [9,11] in THF as the solvent, whereas 1a forms purely disolvated (d = 2) monomers in Et 2 O, t-BuOMe, and 1,2-bis(dimethylamino)ethane (TMEDA). All dimeric species of 1a and 1b are disolvated (d = 1 at each Li) by the above three donor ligands in toluene solution and in the solid state [9]. These d values were found [9] to change inversely with the magnitude of the scalar one-bond 13 C, 6 Li NMR coupling constants 1 J C,Li as quantified by the empirical Equation 1, where n is the number of Li cations coordinated to the carbanionic center 13 C-α under consideration, while a is the number of C-α centers in direct contact with a certain Li cation; obviously, n = a = 1 for monomeric species such as 1 or 2. Synthesized with the experimentally most convenient (because least line-broadening [12]) isotope 6 Li, almost [9] all species of 1a and 2 were found to have a common sensitivity factor of L = 42. 8 Hz in Equation 1, whereas L decreased slightly in a nonlinear fashion with increasing Hammett parameters σ p for R in 1a-e [13]. Numerically different L values apply to alkyl-, phenyl-, and alkynyllithiums [9]. (1) The 13 C, 6 Li coupling constants 1 J C,Li can usually be detected only at sufficiently low temperatures where inter-and intramolecular scrambling of the 6 Li cations becomes slow on the NMR time scales. Due to its spin quantum number I = 1, the single 6 Li nuclear spin in a monomer (n = 1) will split the 13 C-α resonance into 2In + 1 = 3 equally intense components. This 1:1:1 splitting pattern establishes a single 13 C,Li contact, which provides evidence for the monomeric constitution of an organolithium compound that does not carry heteroatoms X as possible donor ligands for an intermolecular coordination (such as C-Li-X). The frequency intervals of this 1:1:1 triplet are numerically equal to the magnitude of 1 J C,Li which can reveal the degree of microsolvation via Equation 1: Decreasing (but nonzero) magnitudes of 1 J C,Li = L × [n (a + d)] -1 can result from increasing values of d (higher microsolvation). Using these two analytic tools of 13 C-α splitting patterns and 1 J C,Li magnitudes, we will now establish the monomeric nature and the microsolvation numbers of the title compound which carries the small H 2 C= group in place of the bulky tetramethylindan-2ylidene part of 1 and 2. The latter constitutional difference was planned to become a touchstone for the cis/trans stereoinversion mechanism that was deduced earlier [1,11] for 1 and 2.

Results and Discussion
Preparation and ground-state properties of 4 The Br/Li interchange reaction (Scheme 2) between n-BuLi and the bromoalkene [14] 3 that generates the title compound α-(2,6-di-tert-butylphenyl)vinyllithium (4) in cyclopentane as the solvent was almost finished after 70 min at room temperature (rt), affording 1-bromobutane (n-BuBr), the olefin [14] 8a, the known [14] lithium arylacetylide 6, and residual n-BuLi (by 1 H NMR in situ analysis). Obviously, 4 had not reacted with its coproduct n-BuBr to produce 7; instead, it had reacted faster than the concomitant n-BuLi to eliminate HBr from 3 and/or subsequently to deprotonate the generated arylalkyne [14] with formation of 6 and 8a. After carboxylation and aqueous workup, the acidic product fraction contained only the arylpropiolic acid (δ H = 1.56 and 7.33 ppm in CDCl 3 ) derived from 6 but not the α-arylacrylic acid to be expected from 4. In t-BuOMe as the solvent, bromoalkene 3 was consumed by n-BuLi in less than 20 min at rt, yielding roughly equivalent amounts of 4, 6, 8a, and residual n-BuLi (the expected n-BuBr signal was perhaps overlaid). While n-BuLi vanished within 75 hours from this solution, the much slower decay of 4 at rt extended over more than 145 hours. Et 2 O as the solvent accelerated the Br/Li interchange reaction so much that a mixing problem emerged: While bromoalkene 3 was consumed by n-BuLi in less than five minutes at −70 °C and furnished mainly 4 along with a little of arylacetylide 6, the slow introduction at or above −20 °C and faster local consumption of n-BuLi by a portion of 3 gave rise to a local accumulation of 4 in the presence of residual 3, so that the elimination of HBr from 3 by 4 could produce comparable amounts of arylacetylide 6 and olefin 8a. The finally surviving portion of 4 reacted very slowly (during >23 hours) at rt with its coproduct n-BuBr to yield hydrocarbon 7. In THF as the solvent, such a mixing problem was encountered already at −70 °C: During the cautious addition of n-BuLi to bromoalkene 3, 4 was generated and rapidly converted into 6, 7, and 8a despite the presence of residual n-BuLi. In consideration of these various aspects of the reactivity profile of 4, it became clear that sufficiently stable solutions of 4 could not be prepared from 3 directly.
The successful detour preparation of more stable solutions of 4 made use of the trimethylstannyl precursor 5 that had been obtained [14] from bromoalkene 3 with LiSnMe 3 . A cyclopentane solution of 5 and n-BuLi did not react during several days at rt; on addition of TMEDA (ca. 3-5 equiv), however, merely one quickly recorded 1 H NMR spectrum could be taken before crystals of 4&TMEDA began to precipitate. Surplus n-BuLi converted the coproduct n-BuSnMe 3 into n-Bu 2 SnMe 2 (and sometimes into n-Bu 3 SnMe) along with a freely floating powder of MeLi. All such contaminations could be removed from the bunchy needles of 4&TMEDA through washings with (cyclo)pentane, whereupon these purified crystals could be dissolved in an anhydrous solvent and stored under argon gas cover at −70 °C until NMR spectra could be measured. In such a colorless solution of 4&TMEDA in THF/cyclopentane (ca. 83:17 by volume), THF had displaced TMEDA from its coordination to 4; at below −50 °C, the olefinic proton resonances of 4 (Table S2, [15]) displayed a well resolved AB spectral system (two doublets) with the two-bond coupling constant 2 J H,H = 8.5 Hz (entry 2, Table 1). Conclusive evidence of the monomeric nature of [ 6 Li]4 arose from the 13 C-α NMR triplet splitting (1:1:1) that became resolved also at and below −50 °C (Table S10, [15]). The magnitude of 1 J C,Li = 10.8 Hz indicated solvation by d = 3 THF ligands at Li, assuming that L = 42 (± 1) Hz in Equation 1 applies also to 4 (as will be confirmed further below). Practically equal 1 H (Table S3, [15]) and 13 C (Table S11, [15]) NMR data with identical magnitudes 1 J C,Li and 2 J H,H (entry 3, Table 1) were found for unpurified 4&3THF that arose immediately from 5 with n-Bu 6 Li in the absence of TMEDA in a less clean (pale violet) THF/hydrocarbons mixture (47:53% by volume) . These coincidences established that even considerable amounts of residual n-Bu 6 Li and the emerging tetrameric Me 6 Li did not form mixed aggregates with 4 in THF. The further numerical similarity with 1 J C,Li = 10.7 Hz of 1a&3THF [9] and 2&3THF [9] (entry 1, Table 1; Don = THF and d = 3 in Scheme 1) is understood to indicate practically equal electronic properties (equal L parameters in Equation 1) of the C-α centers in all of these trisolvated monomers. As a topological requirement for that, the differing α-aryl groups in monomeric 1a and 4 maintain almost orthogonal orientations with respect to the olefinic C=C double bond (as was also reported [14] for 3 and 5). This orientation provides for an almost optimum delocalization of the σ-type (quasi-sp 2 ) carbanionic electric charge at C-α of 4 (Scheme 2) into the α-aryl π-orbital system. The resulting negative lithiation NMR shifts Δδ = δ(4) − δ(8a) for 13 C-4 and 4-H (entries 2 and 3, Table 1) point to a significant charge transfer [9,11] from C-α to C-4 of 4&3THF. The full set of Δδ values of 4 in the two THF solutions is shown in formulae 9a and 9b of Figure 1, which provide a crude impression of the quasi-benzyl anion character of the α-aryl substituent with negatively charged ortho and para positions [16], although it must be admitted that signs and magnitudes of Δδ are not necessarily dominated by local electric charges in positions that are nearer than C-4 to the C-Li bond. Almost all lithiation shifts Δδ shown in 9c ( Figure 1) for α-(2,4,6-tri-tert-butylphenyl)vinyllithium (10) in THF [15] are closely similar to those of 4 in 9a and 9b, even though the chemical shifts δ(C-4) of 4 and 10 (Tables S11 and S12, [15]) differ by 20 ppm. Accordingly, 10 was established to be also a trisolvated monomer ( 13 C-α triplet splitting) with characteristic values (entry 4, Table 1) of 1 J C,Li = 10.9 Hz, 2 J H,H = 8. 6 Hz, and Δδ(C-α) = +73.7 ppm.
Both the clean (with TMEDA) and the contaminated (no TMEDA) THF solutions of 4&3THF decayed at or above +3 °C through proton transfer from the solvent that was cleaved into ethylene and the enolate LiO-CH=CH 2 [the latter identified through δ H = 6.93 ppm (dd) and confirmed by δ C = 81.9 and 158.9 ppm]. In t-BuOMe as the solvent, a purified sample of [ 6 Li]4&TMEDA (entry 5, Table 1) displayed the triplet (1:1:1) of the 13 C-α resonance already on cooling to −24 °C. The monomeric species identified by this triplet remained the only component of 4 between +25 and −88 °C, as documented by the practically temperature-independent 1 H and 13 C NMR chemical shifts (Tables S5 and S13, [15]). The increased magnitude of 1 J C,Li = 13.9 Hz (compared to 10.8 Hz in entries 2 and 3, Table 1) cannot be caused by the changed bulk solvent´s (t-BuOMe) macroscopic properties [9]; instead, it agrees with disolvation (d = 2 only) of a monomer for which L = 42 (± 1) Hz in Equation 1 predicts 1 J C,Li = 14 (± 0.3) Hz. This provided the above postponed confirmation that Equation 1 is valid in the 4 family. The NMR signals of TMEDA shifted and broadened on cooling and became split below −67 °C into separate absorptions of free and coordinated TMEDA (ca. 4:1). NMR integrations at −88 °C revealed the coordination of one molar equivalent of TMEDA. Although the severely broadened NCH 2 and NCH 3 signals did not enable us to differentiate between a symmetric (chelating) and an unsymmetrical (nonchelating) ligand binding of TMEDA at lithium, we are sure to have met the chelating mode which must be favored over the immoblization of additional ligands by its less negative entropy contribution. If so, this disolvation of monomeric 4 implies that TMEDA did not admit the solvent t-BuOMe to participate in direct microsolvation. After final quenching with DOCH 3 (6 equiv), the in situ 1 H and 13 C NMR spectra showed the deuteriated olefin 8b and displayed the expected signal shifting of TMEDA to the resonances of the free ligand.
Returning to the above aberration of unpurified 4 in Et 2 O (without TMEDA, but contaminated by n-BuLi and MeLi), we searched for more convincing evidence of aggregating 4 as follows. After evaporation of the solvent under a reduced pressure of dry argon gas, the remaining oil (ca. 0.02 mL) dissolved readily in [D 8 ]toluene, except for an insoluble powder of (CH 3 Li) 4 . This solution contained [ 6 Li]4 along with Et 2 O (ca. 2 equiv), n-Bu 6 Li, olefin 8a, and a portion of the starting material 5 with partially modified α-trialkylstannyl (n-Bu x Me 3-x Sn) groups. Although a 1 J C,Li coupling could not be resolved in this experiment (Table S17, [15]), the significantly diminished quasi-benzyl anion character shown in 9h ( Figure 1) indicated that this form of [ 6 Li]4 was no longer the monomer but must have become aggregated. In particular, the magnitudes of Δδ(C-α) and Δδ(C-4) in entry 9 of Table 1  The pseudomonomolecular, ionic cis/trans sp 2 -stereoinversion of α-(2,6-di-tertbutylphenyl)vinyllithium (4) In a stereochemically "frozen" ground-state of 4, the olefinic protons H-cis and H-trans give rise to two NMR doublets ("AB" spectral system) with the above-mentioned geminal coupling constant 2 J H,H . These two "diastereotopic" protons (which differ stereochemically but are equivalent by connectivity) can interchange their environments (cis and trans) through a "diastereotopomerization" [23] that interconverts the ground-state contact-ion pairs (CIP) 11 and 11' in Scheme 3.
Increasing rates of this interchange will cause at first an increasing line broadening of the AB system and then a "coalescence" of all four lines into a singlet at the averaged resonance position of (δ cis + δ trans )/2 [24]. The temperature-dependent pseudo-first-order rate constants k ψ of this interconversion were determined with a computer program [25][26][27][28] that can simulate and plot AB line shapes for visual comparison with the experimental spectra. This technique demands extrapolating the values of δ cis -δ trans and 2 J H,H from a series of low-temperature (less broadened) AB spectra into the coalescence domain.
The temperature dependence of the resultant rate constants k ψ (Table S1, [15]) is quantified in Figure 2 by a single straight line, which established that k ψ did not depend on the concentrations of 4 in THF (47% by volume). Therefore, the diastereotopomerization mechanism can be neither associative (for example, bimolecular) nor dissociative (generating the free ions from 4); instead, the rates depend on the concentration of the substrate 4 in a first order as defined in Equation 2. This important kinetic property raises the problem of devising a monomolecular (yet nondissociative) manner of cleaving the C-Li bond in preparation of the stereoinversion step. An obvious possibility consists in the transient heterolysis with formation of an NMR-invisible, solvent-separated ion pair (SSIP) 12 in Scheme 3. Such a heterolysis must be induced by the transitory immobilization of a THF ligand, which will be released on formation of the inverted product 11' via the diastereotopomer 12' of 12. Thus, the sp 2 -stereoinverion process is catalyzed by THF, so that its pseudo-first-order rate constant k ψ in Equation 2 is the product of a second-order rate constant k 0 and the constant concentration of free THF. One might be tempted to call on Equation 2 for interpreting the twofold increase of k ψ that was observed for 4 in a more concentrated (85% in place of 47%) THF solution (−8 °C in Table S2, [15]); however, this weak acceleration might also be due to the increased solvent polarity. Table 2: Pseudoactivation parameters ΔG ψ ‡ (kcal mol −1 at 0 °C), ΔH ψ ‡ (kcal mol −1 ), and ΔS ψ ‡ (cal mol −1 K −1 ) of cis/trans diastereotopomerization rates of three monomeric 1-aryl-1-alkenyllithiums (1a, 2, and 4)   The pseudomonomolecular, ionic mechanism in Scheme 3 had previously [11] been deduced through finding a Hammett reaction constant of ρ = +5.2 for the sp 2 -stereoinversions of the paradigm family of 1a-e, which established that charge separation increased substantially on the way to the transition state (corresponding to 13). In a similar process, C-α in 13 should become nearly sp-hybridized (linear) and more distant from Li + than it is in 11. Consequently, the negative charge at C-α in 13 will be delocalized into the aromatic π-electron system more efficiently than in 11 and 12, so that the quasi-benzyl anion character and the concomitant charge separation increase on the way from 11 to 13. Li + (THF) 4 can then utilize the increasing charge gradients to perform the required migration shown in 13 during stereoinversion of the carbanion, as had been postulated [11] for 1. This proposal is now supported in Table 2 through the similar pseudoactivation parameters of 4&3THF (entry 3, Table 2), 2&3THF (entry 2, Table 2), and 1a&3THF (entry 1, Table 2). Although migration across the β-unsubstituted H 2 C= region in 13 might be sterically much more comfortable than across the bulky β,β-di-tert-alkyl substituents in the transition states deriving from 1 or 2, this opportunity is apparently disdained in 13: The stereoinversion is not facilitated for 4, since the ΔG ψ ‡ barrier of 4 is slightly higher than those of 1a and 2, which may be caused by a structure-dependent decrease of the ground-state free enthalpies (G) that is somewhat stronger for 4. We conclude that the migration of Li + (THF) 4 in 13 occurs likewise away from the C-β region, namely, toward the negative charge center C-4 of the orthogonally oriented α-aryl group without being seriously impeded by the two t-Bu groups. With a 4-t-Bu substituent in 10 [15], however, the corresponding diastereotopomerization was found to be roughly threefold retarded (albeit not quantified because of decomposition); this may have been caused by the decelerating (because electron-donating) effect of 4-t-Bu (σ p -= −0.13 [29,30]) in combination with some steric hindrance of the Li + (THF) 4 migration past 4-t-Bu, even though a concomitant accelerating effect might be expected from a somewhat stronger negative charge at C-4 in 10 (compare the ground-state 9c with 9a in Figure 1). Altogether, these considerations argue for the possibility that Li + (THF) 4 circumnavigates 3-H (or 5-H) in 13, conducted by the negative charges in α-aryl.
As outlined previously [11], the trisolvation privilege of THF facilitates the sp 2 -stereoinversion through limiting the negative entropy contribution of ligand immoblization: Since 4&3THF needs only one further THF ligand to generate the SSIPs 12 and 13 from 11, this immobilization costs only a one-particle entropy contribution of ca. −11 cal mol −1 K −1 [10], as also known [11] for the diastereotopomerization of 1a&3THF and 2&3THF whose ΔS ψ ‡ values (entries 1 and 2, Table 2) are similar to those of 4&3THF (entry 3, Table 2), which suggests a similar course of the stereoinversion processes. In contrast, the disolvated monomers 4&2t-BuOMe and 4&2Et 2 O would have to immobilize two additional ligands at the expense of doubled entropy contributions (ca. −22 cal mol −1 K −1 ) if generating SSIPs (like 12) with tetrasolvated Li + , which explains why their diastereotopomerization rates remained below our NMR time scales even though the doubled THF fixation would provide an increased negative contribution to ΔH ψ ‡ . For a critical evaluation of these pseudoactivation parameters, it may be recalled [10,11] that the "true" activation enthalpies ΔH ‡ will be almost equal to ΔH ψ ‡ , whereas true ΔS ‡ will be more negative than ΔS ψ ‡ by a mathematical correction of roughly R ln[free THF] = up to 5 cal mol −1 K −1 .
Conclusion α-(2,6-Di-tert-butylphenyl)vinyllithium (4) and its 4-t-Bu congener (10) are the first β-unsubstituted vinyllithiums whose microsolvation numbers in solution could be established directly by NMR integrations and/or through 1 J C,Li values via the empirical Equation 1. This became possible through twosided shielding of the C-Li part of 4 by two tert-butyl substituents: these gave rise to unusually low rates of donor ligand exchange at lithium and of some reactions with 4, and they caused also a very weak inclination of monomeric 4 to aggregate in solution, in contrast to the majority of known H 2 C=C(Li)-R species. Such conditions enabled us to establish also the magnitude of the olefinic 2 J H,H coupling constant as a secondary criterion for the degree of microsolvation: Independent of the kind of ligands and solvents (Table 1), the values were 8.5 Hz for trisolvated and 7.4 Hz for disolvated monomers.
The rapid cis/trans stereoinversion of monomeric 4 in THF provides a possible explanation of the stereorandom formation [14] of [β-D]5 through Br/Sn interchange and is partially due to trisolvation of the ground-state: The necessary C-Li bond heterolysis is achieved through immobilization of only one further THF ligand at lithium with a correspondingly low entropic penalty (ca. 44% of ΔS ψ ‡ ). The energetic expenditure also remains low (ca. 50% of ΔG ψ ‡ at 0 °C) because the carbanion and Li + (THF) 4 do not dissociate from the solventseparated ion pair and because the α-aryl substituent stabilizes the carbanionic charge through an increasingly efficient quasibenzyl anion resonance in the transition state 13. The close agreement of these pseudoactivation parameters of 4 with those of 2 and 1a suggests that the migration of Li + (THF) 4 [25] spectra. Natural logarithms (ln) refer to the dimensionless magnitudes of the employed quantities.

[ 6 Li]-α-(2,6-Di-tert-butylphenyl)vinyllithium ([ 6 Li]4).
A dried NMR tube (5 mm) was charged with the trimethylstannyl alkene [14] 5 (51 mg, 0.13 mmol), dry cyclopentane (0.60 mL), and TMEDA (0.068 mL, 0.45 mmol), then cooled under argon gas cover in an ice-bath during the addition of n-Bu 6 Li (0.15 mmol) in cyclopentane (0.060 mL). The tightly closed tube was stored in a large Schlenk tube filled with argon gas at −30 °C until the following washing procedure: The supernatant was withdrawn from the crystals of 4&TMEDA by syringe under argon gas; fresh dry cyclopentane (0.3 mL) was added, the tube was gently shaken, and the crystals were allowed to settle in the tube. After one or more repetitions of such washings and final withdrawal, the crystals were blown dry in a soft stream of dried argon gas for ca. five seconds, then dissolved in the appropriate anhydrous solvent (0.5 mL) together with TMS for NMR measurements [15]. 1 -1-ene (7). A solution of bromoalkene 3 (54 mg, 0.18 mmol) [14] in anhydrous THF (0.50 mL) was placed in an NMR tube (5 mm) and cooled at −70 °C under argon gas cover during the addition of n-BuLi (0.20 mmol) in hexane (0.10 mL). After 120 min at −70 °C, the dark red solution was poured onto solid CO 2 , warmed up, and diluted with Et 2 O (10 mL) and aqueous NaOH (20 mmol,