Evans Enolates: Structures and Mechanisms Underlying the Aldol Addition of Oxazolidinone-Derived Boron Enolates

Abstract

The soft enolization of an acylated oxazolidinone using di-n-butylboron triflate (n-Bu₂BOTf) and trialkylamines and subsequent aldol addition was probed structurally and mechanistically using a combination of IR and NMR spectroscopies. None of the species along the reaction coordinate shows a penchant for aggregating. Complexation of the acylated oxazolidinone by n-Bu₂BOTf was too rapid to monitor, as was the subsequent enolization with Et₃N (triethylamine). The pre-formed n-Bu₂BOTf·Et₃N complex displaying muted Lewis acidity and affiliated tractable rates reveals a rate-limiting complexation via a transition structure with a complicated counterion. n-Bu₂BOTf·i-Bu₃N bearing a hindered amine shifts the rate-limiting step to proton transfer. Rate studies show that the aldol addition with isobutyraldehyde occurs as proffered by others.

TOC Graphic

Introduction

During the development of polyketide total syntheses and the emergence of biomimetic aldol additions, few reagents have been as central as the oxazolidinone enolates, colloquially referred to as Evans enolates.¹ In their seminal 1981 paper, Evans and co-workers² showed that the acylated oxazolidinone scaffold controls additions with high facial selectivity. Although their first attempts likely involved alkali metal enolates, boron enolates derived from di-n-butylboron triflate³ (n-Bu₂BOTf) provide exceptional diastereoselectivities (eq 1).⁴ Since that first publication, Evans enolates have been reported in an astonishing 1600 patents.⁵

(1)

Boron enolates, including several derived from oxazolidinones,⁶ have been examined crystallographically,⁷ spectroscopically,^7,8 and computationally.^9,10 By contrast, few experimental probes of mechanism have been carried out with either simple boron enolates¹¹ or oxazolidinone-based variants.^6,12–15 On the heels of investigations of lithium-based aldol additions of Evans enolates,¹² we undertook structural and mechanistic studies of the boron variant in Scheme 1.¹⁶

Scheme 1.

Mechanism of oxazolidinone-based aldol addition.

Results

Structure Determinations: General Strategies.

The aldol addition was carried out as prescribed in the literature;² however, CHCl₃ was used interchangeably with CH₂Cl₂ owing to the convenience of CDCl₃ for ¹H and ¹³C NMR spectroscopies. ¹¹B NMR spectroscopy was ineffectual owing to broad, poorly resolved resonances. IR spectroscopy, by contrast, proved particularly informative; carbonyl absorbances of key species are summarized in Table 1. Representative IR spectra are contained in Figure 1. ¹H and ¹³C NMR spectroscopies were used to confirm the structures of key species (supporting information).

Table 1.

IR Absorbances of Key Intermediates^a

Compd	C=X (cm−1)
1	1777	1704
2	1777	1704
3	1727b
4	1706	1658
6	1777	1704

^a1777 and 1704 correspond to the oxazolidinone and propionyl group carbonyls, respectively.^14,15

^bSuperimposed single absorbance.

Figure 1.

IR spectra of 0.10 M 1 in CHCl₃ recorded at −60 °C with (a) no additive, (b) 0.11 M n-Bu₂BOTf affording 3, (c) 0.11 M n-Bu₂BOTf and 0.12 M Et₃N affording 4, and (d) 0.11 M n-Bu₂BOTf, 0.12 M Et₃N, and 0.13 M i-PrCHO affording 6.

The method of continuous variations (MCV) was used to ascertain whether key intermediates associate into higher aggregates that would otherwise go undetected with standard spectroscopic methods.¹⁷ If boron enolates are dimeric in solution, binary mixtures of two structurally related enolates would contain homodimers, A₂ and B₂ and a heterodimer, AB (eq 2). AB would appear as a new species or, in the event of rapid exchange, elicit changes in time-averaged ¹H or ¹³C NMR chemical shifts. In the event, numerous binary mixtures of complexes (3), enolates (4), or alkoxides (6) derived from substrates in Chart 1 showed no evidence of AB.

Chart 1.

Enolate precursors

$${\mathbf{A}}_{\mathbf{2}}+{\mathbf{B}}_{\mathbf{2}}\rightleftharpoons 2\mathbf{AB}$$ (2)

Complexation.

The addition of 1.0 equiv of n-Bu₂BOTf to oxazolidinone 1 (0.10 M in CHCl₃) at 20 °C resulted in approximately 50% consumption of 1 and the appearance of a single new absorbance at 1727 cm⁻¹ corresponding to complex 3. We suspect that the electron distribution exemplified by resonance structure 3a (eq 3) accounts for the single C=X absorbance.

The reaction was instantaneous but not quantitative. Complete complexation was observed when the samples were either cooled with 1.1 equiv n-Bu₂BOTf to −60 °C or combined with >3.0 equiv of n-Bu₂BOTf at 20 °C (vide infra).

(3)

¹H NMR spectroscopy showed the two diastereotopic protons of the propionyl methylene fragment in 3 as doublets of quartets with markedly different chemical shifts (δ 3.0 and 1.7 ppm). In partially complexed samples containing only 1.0 equiv n-Bu₂BOTf, the upfield resonance appeared at 1.7 ppm at 20 °C and shifted to 1.4 ppm at −60 °C. It would be tempting to ascribe this result to complexation promoted at low temperature, but the shift occurred independent of the n-Bu₂BOTf concentration, which also promoted complexation. We suspect that the temperature dependence stems from an aromatic π-interaction with the proximate α-proton.^18,19 The analogous complex derived from 9, which bears an isopropyl moiety, showed no anomalous shifting of the proximate α proton. Density functional theory (DFT) computations²⁰ for 3 with geometries optimized at the B3LYP/6–31G(d) level and corrected by single-point calculations at the MP2/6–31G(d)//B3LYP/6–31G(d) level did not support a proton–arene interaction.

Slow exchange on ¹H and ¹³C NMR timescales made the two magnetically inequivalent n-butyl groups of 3 discernible. The treatment of complex 3 with 1-d₂ showed rapid incorporation at −60 °C on ¹H NMR spectroscopy (eq 4), confirming that exchange was fast on laboratory timescales.

(4)

The results of ¹⁹F NMR spectroscopy at −60 °C showed a single resonance (δ −77 ppm) consistent with a free triflate ion that shifted downfield with additional n-Bu₂BOTf. By comparison, CF₃SO₃H and n-Bu₂BOTf displayed ¹⁹F resonances at −79 and −76 ppm, respectively. The triflate counterion story becomes much more complicated.

Enolization

Adding triethylamine (Et₃N; 1.2 equiv) to solutions containing >95% complex 3 (1.1 equiv n-Bu₂BOTf, −60 °C) afforded enolate 4 instantaneously and quantitatively, as shown by the replacement of the absorbance of 3 with two new C=X absorbances (Table 1). Chelation by the oxazolidinone carbonyl was evidenced by the lower energy carbonyl absorbance at 1706 cm⁻¹ (Table 1). The magnetically inequivalent n-butyl moieties of 4, however, were time-averaged in the ¹H and ¹³C NMR spectra at −60 °C, suggesting weak chelation.

Tandem Complexation–Enolization

The complexation of 1 to form 3 was too fast to monitor, as was the enolization of 3 to form enolate 4. However, 1 reacted with a preassociated n-Bu₂BOTf·Et₃N complex at 0 °C at tractable rates with no detectable pre-complex 3 owing to the attenuated (inhibited) Lewis acidity.²¹ n-Bu₂BOTf·Et₃N reacted in slow exchange with free Et₃N, and the ¹H NMR spectra confirmed the 1:1 complex noted previously.^22,23

Rate studies revealed both expected and decidedly unexpected results. The conversion of 1 to enolate 4 at 0 °C under pseudo-first-order conditions (0.0020 M 1) followed a first-order decay as confirmed by pseudo-first-order rate constants (k_obsd) that were independent of the concentration of 1. Comparing 1 and 1-d₂ afforded no isotope effect (k_H/k_D = 1.00 ± 0.01). Post-rate-limiting enolization was confirmed by two additional experiments.²⁴ A mixture of pre-formed complexes 3 and 3-d₂ (1.0 equiv each) was treated with a deficiency (0.80 equiv) of Et₃N, thereby forcing the instantaneous deprotonation to select H over D. The selective loss of 3 monitored with ¹H NMR showed a large competitive isotope effect (k_H/k_D = 10). To confirm that the deprotonation of 3 was faster than decomplexation—a requirement for post-rate-limiting proton transfer²⁴—we added low concentrations of an equimolar mixture of 1-d₂ and Et₃N to a solution of 3. Enolization proceeded to the exclusion of the exchange of 1-d₂ into 3. Conversely, adding low concentrations of an equimolar mixture of 1 and Et₃N to a solution of 3-d₂ again resulted in dominant enolization, with approximately 10% incorporation of 1 into complex 3-d₂. Slowing the isotopically sensitive enolization 10-fold relative to exchange with substrate revealed a 10% competing back reaction, which would have gone undetected in a measured isotope effect.²⁴

A plot of k_obsd versus Et₃N concentration in excess showed an approximate inverse-first-order dependence (n = −0.9 ± 0.1) consistent with the reversible loss of the amine from n-Bu₂BOTf·Et₃N (Figure 2). Quite unexpectedly, a plot of k_obsd versus n-Bu₂BOTf·Et₃N concentration in which the excess amine concentration was held constant showed a third-order dependence (n = 3.3 ± 0.3; Figure 3). We neither anticipated this result nor have a fully satisfactory explanation, as discussed below.

Figure 2.

Plot of k_obsd versus added Et₃N concentration for the enolization of 1 (0.0020 M) by n-Bu₂BOTf·Et₃N (0.040 M) in CHCl₃ at 0 °C. y = ax^b, a = 0.00015 ± 0.00003, b = −0.88 ± 0.05.

Figure 3.

Plot of k_obsd versus n-Bu₂BOTf·Et₃N concentration for the enolization of 1 (0.0020 M) by 0.10 M free (uncomplexed) Et₃N in CHCl₃ at 0 °C. y = ax^b, a = 72 ± 3, b = 3.3 ± 0.3.

$$–\mathrm{d}\left[\mathbf{1}\right]/\mathrm{d}t=k’\left[\mathbf{1}\right]{\left[n\text{-}{\mathrm{Bu}}_2\mathrm{BOTf}·{\mathrm{Et}}_3\mathrm{N}\right]}^3{\left[{\mathrm{Et}}_3\mathrm{N}\right]}^{–1}$$ (5)

Taken together, the reaction orders afforded the rate law in eq 5, which implicated a transition structure with [(1)(n-Bu₂BOTf)₃(Et₃N)₂]‡ stoichiometry.²⁵ The triflate-based counterion denoted as “X⁻” at the transition state would necessarily be complicated. Neither of the two amines in the rate-limiting transition state served as a Brønsted base because the deprotonation is post-rate-limiting. We defer the interpretation of stoichiometry to the discussion; however, we probed the amine dependence and obtained the following relative reaction rates: Me₂NEt (DMEA) < MeNEt₂ < Et₃N < i-Pr₂NEt < i-Bu₃N.

i-Bu₃N shifted the equilibrium to non-limiting behavior that has the effect of shifting the rate-limiting step to proton transfer as described by Scheme 2 and eqs 6–9. A plot of k_obsd versus i-Bu₃N concentration showed saturation kinetics (Figure 4) in which substrate and i-Bu₃N competitively coordinate to boron. K_eq was commensurate with a value measured with ¹H NMR using enolization-resistant isobutyrate 20 (see below) as a surrogate. The enolization of 1 using i-Bu₃N showed rate-limiting proton transfer, as evidenced by a large kinetic isotope effect (k_H/k_D = 10) at low and high amine concentrations. Changing the addition sequence by adding i-Bu₃N to the substrate–boron complex provided the same behavior. The counterintuitive normal (rather than inverse) saturation owing to an unproductive side equilibrium stemmed from the dual role of amine as both inhibiting Lewis base and accelerating Brønsted base.

Scheme 2.

Competitive amine-mediated complexation and enolization.

Figure 4.

Plot of k_obsd versus i-Bu₃N concentration for the enolization of 1 (0.0020 M)by n-Bu₂BOTf·i-Bu₃N (0.040 M) and i-Bu₃N in CHCl₃ at 0 °C. y = ax/(x + b), a = 0.046 ± 0.004, b = 0.14 ± 0.03.

$$\mathrm{d}\left[\mathbf{4}\right]/\mathrm{d}t=k_{\mathrm{obsd}}\left[{\mathbf{1}}_{\mathrm{total}}\right]$$ (6)

$$k_{\mathrm{obsd}}=k_2{K}_{\mathrm{eq}}\left[n\text{-}{\mathrm{Bu}}_2\mathrm{BOTf}·i\text{-}{\mathrm{Bu}}_3\mathrm{N}\right]/\left\{1+K_{\mathrm{eq}}\left[n\text{-}{\mathrm{Bu}}_2\mathrm{BOTf}·i\text{-}{\mathrm{Bu}}_3\mathrm{N}\right]/\left[i\text{-}{\mathrm{Bu}}_3\mathrm{N}\right]\right\}$$ (7)

Low Amine Concentration

$$k_{\mathrm{obsd}}=k_2\left[i\text{-}{\mathrm{Bu}}_3\mathrm{N}\right]$$ (8)

High Amine Concentration

$$k_{\mathrm{obsd}}=k_2{K}_{\mathrm{eq}}\left[n\text{-}{\mathrm{Bu}}_2\mathrm{BOTf}·i\text{-}{\mathrm{Bu}}_3\mathrm{N}\right]$$ (9)

Aldol Addition.

The reaction of enolate 4 (0.10 M) and excess Et₃N (0.020 M) with i-PrCHO (0.13 M) in CH₂Cl₂ at −78 °C afforded adduct 6 at tractable rates. The absence of chelation in alkoxide 6 was shown by carbonyl absorbances that were nearly indistinguishable from those of starting oxazolidinone 1. The absence of aggregation was shown experimentally using MCV (vide supra) and supported by DFT computations predicting the dimerization of alkoxide 6 (eq 10) to be highly endothermic.²⁶

(10)

Reactions under pseudo-first-order conditions (0.0050 M enolate 4 and 0.050 M i-PrCHO) showed first-order decays of enolate and k_obsd that were independent of the initial enolate concentration (Figure 5). The rate was unaffected by excess Et₃N or added tetrahydrofuran. The rate law in eq 11 is consistent with a simple aldol addition mechanistically akin to that proposed by Evans in the original work (5 in Scheme 1). DFT studies mirroring those reported by Kobayashi and co-workers⁶ showed transition structure 5a to be both viable and 4.7 kcal/mol more stable than transition structure 5b leading to the wrong isomer.^1,2

$$–\mathrm{d}\left[\mathbf{4}\right]/\mathrm{d}t=k’\left[\mathbf{4}\right]\left[i\text{-}\mathrm{PrCHO}\right]$$ (11)

Figure 5.

Plot of k_obsd versus i-PrCHO concentration for aldol addition by enolate 4 (0.0050 M) with a slight (0.0010 M) excess of Et₃N in CHCl₃ at −60 °C. y = ax^b, a = 0.067 ± 0.007, b = 1.09 ± 0.10.

Discussion

The structural and mechanistic studies of the Evans boron aldol addition proceeded largely according to conventional wisdom. Treating oxazolidinone 1 at 20 °C with n-Bu₂BOTf causes the instantaneous formation of complex 3. Full complexation requires cooling to −60 °C or the use of excess n-Bu₂BOTf at 20 °C. Treatment of 3 with trialkylamines at −78 °C effects instantaneous enolization to give boron enolate 4. Aldol addition to i-PrCHO at −60 °C affords boron alkoxide 6.

IR spectroscopy provided compelling support for the chelates in 3 and 4, whereas the carbonyl of 6 does not coordinate to boron. Computations were in full accord with the monomeric alkoxide 6.²⁶ MCV also provided no evidence that 3, 4, or 6 aggregate.

The high rates of complexation and enolization were overcome using n-Bu₂BOTf·Et₃N mixtures, which caused a marked attenuation of the complexation rate. The rate studies confirmed a rate-limiting complexation followed by rapid (post-rate-limiting) enolization. The success of the standard boron aldol addition stems from the fact that the incomplete 1–3 equilibrium is driven to enolate 4 by the enolization. Switching to n-Bu₂BOTf·i-Bu₃N displaces the equilibrium toward observable complex 3 and shifts the rate-limiting step to proton transfer (Scheme 2).²³

The details of the complexation are very odd. Although complex 3 represents a simple CF₃SO₃⁻ (“X⁻”), a third-order dependence on n-Bu₂BOTf·Et₃N in conjunction with an inverse-first-order dependence on amine implicates a rate-limiting transition structure with an unexpected [(n-Bu₂BOTf)₃(Et₃N)₂(1)]‡ stoichiometry.^25,27 Chart 2 shows a number of fragments—possible building blocks—that could be in play ranging from highly plausible to merely conceivable. The titration of n-Bu₂BOTf with Et₃N clearly shows a 1:1 complex consistent with 15. A single n-Bu group excludes the ion pair of 16 and 17. Dimer 18 is supported by limited literature precedent,²⁸ but it offers us nothing useful. The 3:2 stoichiometry demands a reactive form such as that composed of Lewis acidic 16 and counterion 19 with five-coordinate borons. The evidence of five-coordinate boron is sound²⁹. We piece this together into to create transition structure 20 for rate-limiting complexation. The five-coordinate, trigonal bipyramidal substitution at boron has strong precedent from some rate studies we did a dozen years ago on imine activation by BF₃-R₃N complexes.³⁰ The complex genenion remains the controversial portion. And, as a referee noted, if the third-order dependence in Figure 3 is actually a second-order dependence that is in error—it happens—one can replace gegenion 19 with the far more conventional 17. It would be an understatement to say that we are uneasy about parts of the model.

Chart 2.

Possible boron-containing fragments.

As an aside, the kinetics using the n-Bu₂BOTf·i-Pr₂NEt bearing a more hindered trialkylamine to form enolate 4 shows a 30-fold acceleration relative to the kinetics with the Et₃N variant, as expected for a mechanism requiring an amine dissociation. Interestingly, the measured reaction order in n-Bu₂BOTf·i-Pr₂NEt approaches unity (n = 1.3). The complex counterion may be unfavorable if the cation fragment (analogous to 16) becomes too congested.

The final step—the aldol addition of 4 to give 6 via transition structure 5 (Scheme 1)—is mechanistically uneventful. Neither Et₃N nor added tetrahydrofuran inhibits the reaction, which shows that the putative four-coordinate boron in 4 is undisturbed by Lewis basic ligands. Kobayashi and co-workers⁶ carried out calculations probing the nuances of 5 and competing orientations. These reactions showed consistency in the open transition structure and approach of the aldehyde anti to the benzyl group.

Conclusion

We described studies of the boron enolate-based Evans aldol addition. The most surprising aspect proves to be the mechanism of complexation, in which we may have uncovered some unusual organoboron coordination chemistry. The potentially most useful part, however, is probably the insights gained about the dual role of trialkylamines as inhibitors—complexants to the Lewis acid—and Bronsted bases. Suspecting that we could use this information to optimize the quaternization of the α-carbon using boron enolates,⁴ we formed complex 22 from isobutyryl derivative 21. The addition of Et₃N pushes the equilibrium to 21 rather than enolate 23, however. More hindered trialkylamines promote complexation but are too unreactive as Brønsted bases. Thus, the quaternization faces challenges posed by such soft enolization methods.

It seems generally useful, even advisable, to understand the structural and mechanistic principles underlying synthetically prevalent reactions such as the Evans aldol addition.

Experimental

Reagents and Solvents.

CH₂Cl₂, CHCl₃, and CDCl₃ were distilled from molecular sieves. Trialkylamines were distilled from sodium benzophenone ketyl. n-Bu₂BOTf was used as a neat oil by evaporating the solvent from a commercial 1.0 M n-Bu₂BOTf solution in CH₂Cl₂. Air- and moisture-sensitive materials were manipulated under argon using standard glovebox, vacuum line, and syringe techniques. Oxazolidinones 1 and 7–11 were either purchased or prepared as described previously.³¹

NMR Spectroscopy.

An NMR tube under vacuum was flame-dried on a Schlenk line and allowed to return to room temperature, backfilled with argon, and placed in a −78 °C dry ice/acetone bath. The appropriate amounts of n-Bu₂BOTf, Et₃N, and oxazolidinone in CDCl₃ were added sequentially via syringe. The tube was flame-sealed under partial vacuum, mixed on a vortex mixer three times for ~10 s with cooling between each vortexing, and stored in a freezer at −80 °C. Standard 1H, 13C, and ¹⁹F NMR spectra were recorded on a 500 MHz spectrometer at 500, 125, and 470 MHz, respectively. The 1H, 13C, and ¹⁹F resonances are referenced to CDCl₃ (CHCl₃ 7.26 and CDCl₃ 77.16 ppm) and fluorobenzene (−113.15 ppm).

IR spectroscopic analyses.

IR spectra were recorded with an in situ IR spectrometer fitted with a 30-bounce, silicon-tipped probe. The spectra were acquired in 16 scans at a gain of 1 and a resolution of 4 cm⁻¹. A representative reaction was carried out as follows: The IR probe was inserted through a nylon adapter and O-ring seal into an oven-dried, cylindrical flask fitted with a magnetic stir bar and a T-joint. The T-joint was capped with a septum for injections and a nitrogen line. After evacuation under full vacuum, heating, and flushing with nitrogen, the flask was charged with CHCl₃ and cooled in a −60 °C bath prepared with fresh acetone. After a background spectrum was recorded, oxazolidinone 1 (23.3 mg, 0.10 mmol) was added as a 1.0 M solution in CHCl₃ with stirring, followed by neat n-Bu₂OTf (30.2 mg, 0.11 mmol), neat Et₃N (12.1 mg, 0.12 mmol), and i-PrCHO (9.4 mg, 0.13 mmol). IR spectra were recorded every 15 s with monitoring of the absorbance at 1777 cm⁻¹ and 1658 cm⁻¹ over the course of the reaction.