Lithium Hexamethyldisilazide-Mediated Enolizations: Influence of Triethylamine on E/Z Selectivities and Enolate Reactivities

Abstract

Lithium hexamethyldisilazide (LiHMDS) in triethylamine (Et₃N)/toluene is shown to enolize acyclic ketones and esters rapidly and with high E/Z selectivity. Mechanistic studies reveal a dimer-based mechanism consistent with previous studies of LiHMDS/Et₃N. E/Z equilibration occurs when <2.0 equiv of LiHMDS are used. Studies of the aldol condensation and Ireland-Claisen rearrangement of the resulting Et₃N-solvated enolates show higher and often complementary diastereoselectivities when compared with analogous reactions in THF. The Et₃N-solvated enolates also display a marked (20-fold) acceleration of the Ireland-Claisen rearrangement with evidence of autocatalysis. A possible importance of amine-solvated enolates is discussed.

Introduction

On the heels of investigations of the solvent-dependent structures of lithium hexamethyldisilazide (LiHMDS),^1,2 we began examining structure-reactivity relationships in LiHMDS-mediated ketone enolizations (eq 1).³ We found that enolization of ketone 1 in Et₃N/toluene^3a is >100 times faster than the analogous enolization in neat THF.^3b,c. This was surprising given that simple trialkylamines are particularly feeble ligands^2,3a,b^,4 that have rarely been used in organolithium chemistry.^5,6 The high rates imparted by trialkylamines were traced to a dimer-based pathway exemplified by transition structure 3. The exceptional steric demands of the trialkylamines were shown to destabilize dimeric LiHMDS reactants more than they destabilize transition structure 3. By contrast, the LiHMDS/THF-mediated enolizations proceed via a more conventional disolvated monomer-based transition structure 4.^3b.

(1)

We describe herein LiHMDS/Et₃N-mediated enolizations of acyclic ketones. The enolizations are fast, dimer-based, and, most important, highly E selective (eq 2). We also examine the reactivity of the Et₃N-solvated lithium enolates toward aldol condensations and Ireland-Claisen rearrangements and find some distinct advantages when compared with THF-solvated enolates.⁷

(2)

Results

Stereochemistry of Enolization

The E/Z selectivities for LiHMDS-mediated enolizations were monitored by quenching the resulting enolate solutions with Me₃SiCl/Et₃N⁸ and analyzing the products with gas chromatography (GC) using well-established protocols (eq 3).⁹ The proportion of LiHMDS to ketone proves to be the key determinant of stereochemistry (Figure 1). At low base loadings the enolization shows a modest Z selectivity akin to that observed for LiHMDS/THF mixtures.^10,11,12 The E selectivity rises markedly with increasing quantities of LiHMDS, approaching 110:1 at ≥2.0 equiv of LiHMDS per ketone.

Figure 1.

E/Z selectivity (E-6:Z-6, eq 3) versus molarity (M) of LiHMDS in samples containing 0.10 M ketone 5 and 1.2 M excess Et₃N in toluene at −78 °C.¹⁶

(3)

The dependencies of the E/Z selectivities on the equivalents of LiHMDS could be construed as evidence of intervening lithium enolate/LiHMDS mixed aggregates in a kinetically controlled enolization.^{9, 13,14,15} It is mathematically impossible, however, for a 100:1 selectivity at 50% conversion (2.0 equiv of base) to be reversed to a 1:2 selectivity at full conversion (1.0 equiv of base) without an intervening equilibration. Indeed, when a reaction mixture resulting from a highly E-selective enolization of ketone 5 is subsequently treated with cyclohexyl ethyl ketone as a surrogate to consume the remaining LiHMDS (eq 4), the E selectivity in the enolization of 5 is lost. Under no circumstance, however, is a regiochemical equilibration observed.

(4)

A limited survey of hindered trialkylamines and dialkyl ethers in toluene as the cosolvent afforded the E/Z selectivities listed in Table 1. LiHMDS/Et₃N affords the highest selectivities.¹¹ The selectivity reversal caused by decreasing the THF concentrations was noted by Xie and Wielgosh.¹²

Table 1.

Solvent-dependent E/Z selectivities resulting from the enolization of ketone 5 (eq 3)^a

solvent, S	E-6:Z-6
Et3N	110:1
Me2NEt	90:1
(i-Pr)2NEt	40:1
i-Bu3N	60:1
THF (0.5 M)	13:1
THF (10 M)	1:13
Me4THFb	40:1
TMEDA	30:1

^a[5] = 0.050 M; [LiHMDS] = 0.15 M; [S] = 1.5 M in toluene; −78 °C.

^bMe₄THF = 2,2,5,5-tetramethyltetrahydrofuran (1.5 M in toluene).

The generality of the E-selective enolization by LiHMDS/Et₃N was established by surveying the enolizations summarized in Table 2. Results derived from LiHMDS/THF are included for comparison. Although the selectivities were not optimized for each case, we suspect there is only limited room for improvement.

Table 2.

Solvent-dependent E/Z selectivities¹¹ for enolization by LiHMDS (eq 5).^a

	E/Z selectivity
R	Et3N/tolueneb	THFc
Et	140:1	1:411a
i-pr (3)	100:1	1:1411a
Cy	80:1	1:3010
Ph	3.5:1	1:>10011a
CH3O	22:1	1:1211b,k
	50:1	1:811c,d

^a[ketone] = 0.05 M; [LiHMDS] = 0.15 M.

^b[Et₃N] = 1.5 M in toluene; −78 °C.

^cNeat THF.

(5)

Aggregate Structures

Adding ketone 5 to LiHMDS in either toluene or Et₃N/toluene mixtures at −78 °C and monitoring with in situ IR spectroscopy reveals that 5 (1714 cm⁻¹) is quantitatively converted to LiHMDS-ketone complexes (1698 cm⁻¹).^17,18 (The less reactive 2,4,4-trideuterated ketone, 5-d₃,¹⁹ was used to monitor structures and rates in some instances.) Dimers bearing one or two complexed ketones (7 and 8) are readily characterized with ⁶Li and ¹⁵N NMR spectoscopy^1,20,21 using [⁶Li,¹⁵N]LiHMDS²² (Table 3). Putative LiHMDS-ketone complex 9 was too reactive to characterize, but is almost certainly isostructural to the closely related complex characterized previously.^3b Enolization of 5 with 3.0 equiv of LiHMDS/Et₃N at −78 °C affords mixed dimer 10 (Table 3). If the enolization is carried out under poorly selective conditions at 0 °C using 3.0 equiv of [⁶Li,¹⁵N]LiHMDS, both mixed dimer 10 and the corresponding Z-enolate-derived mixed dimer 11 are observed.^15b

Table 3.

NMR spectroscopic data^a

compd	δ 6Li (mult, JLiN)	δ 15N (mult, JLiN)
7	1.00 (t, 4.0)	41.4 (q,3.8)
	1.48 (t, 3.1)
8	1.85 (t, 3.3)	38.7 (q, 3.4)
10	0.48 (d, 3.4)	40.34 (q, 3.4)
11	0.51 (d, 3.5)	40.31 (m, --)

^aSpectra were recorded in 0.10 M [⁶Li,¹⁵N]LiHMDS in toluene containing Et₃N. Coupling constants were measured after resolution enhancement and reported in hertz. Multiplicities are denoted as follows: d = doublet, t = triplet, and q = quintet. The chemical shifts are reported relative to 0.30 M ⁶LiCl/MeOH at −90 °C (0.0 ppm) and neat Me₂NEt at −90 °C (25.7 ppm).

Mechanism of Enolization

The structural similarities of ketones 1 and 5 certainly suggest a common mechanism. Indeed, rate studies using well-documented protocols^3,21 confirm the mechanistic parallels as follows. Pseudo-first-order conditions were established by maintaining low concentrations of ketone 5 (0.004–0.01 M) and high, yet adjustable, concentrations of recrystallized²² LiHMDS (0.05–0.40 M) and Et₃N (0.15–2.40 M), with toluene as the cosolvent. Monitoring the loss of LiHMDS-ketone complex 9 with in situ IR spectroscopy (1698 cm⁻¹) shows clean first-order decays to ≥5 half-lives. The resulting pseudo-first-order rate constants (k_obsd) are independent of ketone concentration (0.004 – 0.040 M). Re-establishing the IR baseline and monitoring a second aliquot of ketone reveals no significant change in the rate constant, showing that conversion-dependent autocatalysis or autoinhibition are unimportant under these conditions.^23,24 Comparison of 5 and 5-d₃ provided a large kinetic isotope effect (k_H/k_D > 10),²⁵ confirming rate-limiting proton transfer.^26,27

A plot of k_obsd versus [Et₃N] shows saturation kinetics (Figure 2).^3b,24,28 At low Et₃N concentrations, unsolvated complex 7 is the dominant form (see above), affording a first-order [Et₃N] dependence. At high concentrations of Et₃N, fully solvated complex 9 becomes the dominant form, and a zeroth-order Et₃N dependence results. A plot of k_obsd versus [LiHMDS] at elevated Et₃N concentration reveals a zeroth-order dependence (Figure 3). (The slight upward drift results from residual saturation kinetics affording complex 9.)^3a The idealized rate law (eq 6) is consistent with the generic mechanism described by eqs 7 and 8 and monosolvated dimer-based transition structure 12.²⁹

$$-d{[\text{complex}]}_{\text{total}}/\mathit{\text{dt}}=(k\prime K\text{eq}[{\text{Et}}_3\mathrm{N}]{[\text{complex}]}_{\text{total}})/(1+K_{\text{eq}}[{\text{Et}}_3\mathrm{N}]){\text{such}\text{that}[\text{complex}]}_{\text{total}}=[7]+[9]$$ (6)

$$\begin{array}{l}[\underset{(\mathbf{7})}{{\{{({\text{Me}}_3\text{Si})}_2\text{NLi}\}}_2(\text{ketone})]}+{\text{Et}}_3\mathrm{N}\stackrel{K_{\text{eq}}}{\rightleftharpoons }[\underset{(\mathbf{9})}{{\{{({\text{Me}}_3\text{Si})}_2\text{NLi}\}}_2}({\text{Et}}_3\mathrm{N})(\text{ketone})]\\ \end{array}$$ (7)

$$[\underset{(\mathbf{9})}{{\{{({\text{Me}}_3\text{Si})}_2\text{NLi}\}}_2({\text{Et}}_3\mathrm{N})}(\text{ketone})]\stackrel{k\prime }{\to }{[\underset{(\mathbf{12})}{{\{{({\text{Me}}_3\text{Si})}_2\text{NLi}\}}_2}({\text{Et}}_3\mathrm{N})(\text{ketone})]}^{‡}$$ (8)

Figure 2.

Plot of k_obsd versus [Et₃N] in toluene cosolvent for the enolization of 5-d₃ (0.005 M) by LiHMDS (0.10 M) at −78 °C. The curve depicts an unweighted least-squares fit to k_obsd = a[Et₃N]/(1 + b[Et₃N]), where a = 4.0 ± 0.1, b = 8.2 ± 0.6 × 10¹.

Figure 3.

Plot of k_obsd versus [LiHMDS] in 4.0 M Et₃N/toluene solution for the enolization of 5-d₃ (0.005 M) by LiHMDS at −78 °C. The curve depicts an unweighted least-squares fit to k_obsd = a[LiHMDS] + b, where a = 5 ± 1, b = 3.0 ± 0.1.

Aldol Condensation

We examined the reactivities and selectivities of enolates bearing poorly coordinating and sterically demanding Et₃N ligands and compared them with those observed in neat THF. For example, eq 9 illustrates a sequential enolization with LiHMDS (3.0 equiv) in Et₃N/toluene followed by an aldol condensation with isobutyraldehyde affording primarily anti-13 (anti/syn = 25:1).³⁰ By contrast, the analogous reaction in neat THF shows decidedly inferior stereocontrol with opposite selectivity. Several additional aldol condensations are included as supporting information.

(9)

By carrying out the enolizations in different solvents, we are comparing the chemistry of enolates that differ in both stereochemistry and coordinating ligand. To isolate the influence of solvent on only the aldol condensation, we enolized 5 with LiHMDS/Et₃N and subsequently added THF prior to the addition of i-PrCHO; the anti/syn selectivity drops to 18:1. Thus, there is no apparent advantage offered by the THF-solvated enolate. Nonetheless, we believe that such strategies can be important (vide infra).

Ireland-Claisen Rearrangement

We examined a lithium enolate variant of the Ireland-Claisen rearrangement^7,11b,31 to compare selectivities and reactivities of THF- and Et₃N-solvated enolates. Enolization of 14 using LiHMDS (3.0 equiv) and Et₃N (10 equiv per lithium) in toluene at −78 °C affords an estimated 20:1 preference for enolate 15 (see Table 2, R = OMe). Claisen rearrangement of 15 at −20 °C affords a 35:1 ratio of 16³¹ and 17³¹ in 79% isolated yield. Addition of 10 equiv of THF (per lithium) subsequent to enolization in Et₃N/toluene affords a comparable 33:1 selectivity in 70% yield (albeit much more slowly; vide infra). Curiously, the standard protocol--enolization and rearrangement in neat THF--proceeds in 11% yield. The low yields using LiHMDS/THF could stem from either a poor enolization or a poor rearrangement from the Z isomer as noted by Ireland in a similar silyl Claisen.¹¹ⁱ

Qualitative rate studies of the rearrangement are quite interesting. In situ IR spectroscopy revealed the loss of the Et₃N-solvated ester enolate 15 (1675 and 1659 cm⁻¹) and formation of carboxylate absorbances (1610, 1600, 1582, and 1424 cm⁻¹).³² A sigmoidal behavior (Figure 4, curve A) is indicative of autocatalysis^24,33 occurring as the enolate is replaced with carboxylates, presumably within aggregates. By contrast, addition of 10 equiv of THF (per lithium) subsequent to enolization affords a THF-solvated enolate, as evidenced by the replacement of absorbances at 1675 and 1659 cm⁻¹ with an absorbance at 1657 cm⁻¹. Of special importance, rearrangement in Et₃N/toluene (Figure 4, curve B) proceeds approximately 20 times faster than with added THF. The Et₃N solvated enolate is substantially more reactive.

Figure 4.

IR absorbance of enolate 15 versus time in (A) 1.5 M Et₃N/toluene, and (B) 1.5 M Et₃N/1.5 M THF/toluene.

Discussion

We previously showed that LiHMDS/Et₃N-mediated enolizations are very fast compared to LiHMDS/THF and traced the accelerations to a dimer-based mechanism.^3b,29 We hasten to add that the high rates are only observed using ≥2.0 equiv of LiHMDS. (Equimolar LiHMDS and ketone afford an inert bis-ketone solvated LiHMDS dimer.) Of course, such accelerations are notable, but the procedure may also be cost effective. LiHMDS is now used routinely on large scales for complex drug synthesis.^35,36 Using Et₃N/hydrocarbon mixtures instead of THF could offer considerable savings.^10,37

E-Selective Enolizations

Acting on a hunch, we found LiHMDS/Et₃N-mediated enolizations of acyclic ketones are very fast and afford high E/Z selectivities (Tables 1 and 2). Excess (≥2.0 equiv) LiHMDS is required to maintain high rates and selectivities. Enolizations of ketone 5 with only 1.0 equiv of LiHMDS/Et₃N suffer from E/Z equilibration. Overall, the procedure is more selective and more convenient than that using LiTMP/LiBr mixtures, which has afforded lithium enolates with the highest E-selectivity (up to 50:1) reported to date.⁹

Mechanism of Enolization

NMR spectroscopic studies of the enolization of ketone 5 using 3.0 equiv of LiHMDS revealed lithium enolate-LiHMDS mixed dimer 10. The loss in stereoselectivity when <2.0 equiv LiHMDS/Et₃N was used was traced to a facile equilibration and consequent formation of mixed dimers 10 and 11. There is no affiliated regiochemical equilibration.

Rate studies of the enolization under pseudo-first-order conditions (a large excess of LiHMDS) confirmed that a dimer-based enolization--transition structure 12--is the source of the high selectivities. Previous semiempirical computational studies of enolizations by amine-solvated lithium amides failed to predict the observed high selectivities.³⁸

Aldol Condensation

The facile enolizations in Et₃N/toluene allowed us to examine the reactivity of Et₃N-solvated enolates and to compare them with their THF-solvated counterparts. We first examined the aldol condensation--the most prevalent application of E-selective enolizations--to ascertain whether the high enolization selectivities could be translated into high syn/anti selectivities. The aldol condensations are generally anti selective using the LiHMDS/Et₃N-based protocol (eq 9 and supporting information). We examined whether the selectivity might improve if THF was added before the addition of the aldehyde. This sequence is tantamount to doing the enolization in Et₃N and the aldol condensation in THF. The selectivities did not improve.

Claisen Rearrangement

We investigated a lithium enolate variant of the Ireland-Claisen rearrangement shown in eq 10.⁷ The low yields using LiHMDS/THF were previously observed by Ireland on a closely related case and were attributed to the formation of an unreactive stereoisomer. We concur with this notion. The improved selectivity toward enolate 15 expected for LDA/THF could, in principle, solve the problem, but we obtained complex mixtures that probably derived from allylic metalation of or 1,2-addition to the cinnamyl ether moiety.³⁹ By contrast, LiHMDS/Et₃N-mediated enolization affords 15 rapidly and selectively. Subsequent rearrangement proceeded in good yield and exceptional selectivity.

A distinct benefit of the Ireland-Claisen rearrangement is that it is slow enough to allow us to readily monitor reaction rates. We found that Et₃N-solvated enolates rearrange approximately 20 times faster than the analogous THF solvates, and there is evidence of autocatalysis.³³ We are likely to examine Ireland-Claisen rearrangements in more detail later.

On the Role of Solvent Mixtures

There are very few ligands--probably far fewer than most organic chemists realize--that can displace THF from lithium.^1,20 When an enolate is generated in THF, the subsequent reaction is necessarily THF dependent in most instances. By contrast, highly hindered trialkylamines are substitutionally labile, potentially allowing many ligands to coordinate and influence subsequent reactions. Although substituting THF for Et₃N in a few instances described herein did not offer obvious benefits, the principle is an important one. Organolithium intermediates generated in poorly coordinating solvents are especially well suited for subsequent ligand substitutions. If Et₃N and the electrophile are incompatible (alkyl halides, for example), highly hindered dialkylethers may suffice.^3d

Conclusion

LiHMDS/Et₃N offers a particularly convenient, cost-effective protocol for generating lithium enolates with the highest E selectivity reported to date. The Et₃N-solvated enolates are also provocative species. As expected, the high E selectivities can be translated into relatively highly stereoselective aldol condensations and Ireland-Claisen rearrangements. Moreover, the potential importance of amine-solvated lithium enolates is underscored by a marked (20-fold) acceleration of the Ireland-Claisen rearrangement when compared with the rates observed using THF-solvated variant. One might ask why, if Et₃N elicits such favorable reactivities, have simple (monodentate) trialkylamines played such a minor role in organolithium chemistry?^5,6 The answer may be simple. Et₃N is a poorly coordinating ligand that cannot compete with ethereal solvents for coordination to lithium: Et₃N requires hydrocarbons as cosolvents to be effective. We suggest that trialkylamine/hydrocarbon mixtures are worthy of more careful consideration in other applications within organolithium chemistry.

Experimental Section

Reagents and Solvents

Coordinating ligands and toluene were distilled by vacuum transfer from blue or purple solutions containing sodium benzophenone ketyl. The toluene still contained 1% tetraglyme to dissolve the ketyl. ⁶Li metal (95.5% enriched) was obtained from Oak Ridge National Laboratory (Oak Ridge, TN). LiHMDS, [⁶Li]LiHMDS, and [⁶Li,¹⁵N]LiHMDS were prepared and purified as described previously.²² Ketone 5-d₃ was prepared as described previously.¹⁹ Air- and moisture-sensitive materials were manipulated under argon or nitrogen using standard glove box, vacuum line, and syringe techniques.

NMR Spectroscopic Analyses

Samples were prepared, and the ⁶Li and ¹⁵N NMR spectra were recorded as described elsewhere.²

IR Spectroscopic Analyses

Spectra were recorded with an in situ IR spectrometer fitted with a 30-bounce, silicon-tipped probe optimized for sensitivity. The spectra were acquired in 16 scans (30-second intervals) at a gain of 1 and a resolution of 4 or 8. The probe was inserted through a nylon adapter and O-ring seal into an oven-dried, cylindrical flask fitted with a magnetic stir bar and T-joint. The T-joint was capped with a septum for injections and an argon line. After evacuation under full vacuum and flushing with argon, the flask was charged with the reagents via syringe.

Representative enolization

A 5-mL serum vial was charged with a solution of LiHMDS (50 mg, 0.30 mmol) in Et₃N (420 µL, 3.0 mmol) and toluene (1.53 mL) and cooled to −78 °C. Ketone 5 (12 µL, 0.10 mmol) in toluene (38 µL) was added with stirring. After 20 min, the reaction was quenched with 180 µL of a 4:1 solution of Me₃SiCl/Et₃N (centrifuged free of solid Et₃N-HCl) in toluene (3.0 mL). After warming to RT, dilution with pentane, and quenching with cold aqueous NaHCO₃, the solution was subjected to a standard aqueous workup. The crude enol silyl ethers E-6 and Z-6 were analyzed by GC as described previously.⁹

Representative Aldol Condensation

The enolate solution prepared as described above was quenched with isobutyraldehyde (27 µL, 0.30 mmol). After an additional 10 min at −78 °C and quenching with aqueous NH₄Cl, a standard aqueous workup afforded a crude mixture of adducts anti-13 and syn-13 in 25:1 ratio as shown by GC. Flash chromatography afforded pure anti-13 in 72% yield as shown by comparison with previously reported spectroscopic data.³⁰

Ireland-Claisen Rearrangement

A solution of enolate in Et₃N (1.5 M)/toluene was prepared at −78 °C as described above. The sample was slowly warmed to RT to allow the rearrangement to proceed. After 10 min at RT, the reaction mixture was poured into 10 mL of 5% aqueous NaOH and washed with ether. The aqueous layer was cooled to 0 °C and acidified with concentrated HCl, and the carboxylic acid was extracted with ether. The carboxylic acid was converted to its methyl ester by adding distilled CH₂N₂⁴² at 0 °C until a yellow color persisted. The resulting methyl esters were analyzed using GC to show a 35:1 mixture of 16³¹ and 17.³¹ Flash chromatography afforded a 79% combined yield.

Kinetics

Rate constants were determined using a standard protocol exemplified as follows: The IR cell described above was charged with a solution of LiHMDS (167 mg, 1.00 mmol) in Et₃N (2.09 mL, 1.5 M) and toluene (7.74 mL) and cooled to −78 °C. After recording a background spectrum, ketone 5 (50 µL, 0.050 mmol, 0.005 M final concentration) was added neat with stirring. IR spectra were recorded over 5 half-lives. To account for mixing and temperature equilibration, spectra recorded in the first 1.0 min were discarded. The rate constant was determined using an unweighted, non-linear least squares fit to the function f(x)=ae^bx.