Sodiated Oppolzer Enolates: Solution Structures, Mechanism of Alkylation, and Origin of Stereoselectivity

Abstract

NMR spectroscopic studies reveal camphorsultam-derived sodium enolates known as Oppolzer enolates reside as monomers in neat THF and THF/HMPA solutions and as dimers in toluene when solvated by N,N,N’,N’-tetramethylethylenediamine (TMEDA) and N,N,N’,N’’,N’’-pentamethyldiethylenediamine (PMDTA). Density functional theory (DFT) computations attest to the solvation numbers. Rate studies show analogy with previously studied lithiated Oppolzer enolates in which alkylation occurs through non-chelated solvent-separated ion pairs. The origins of the selectivity trace to transition structures in which the alkylating agent is guided to the exo face of the camphor owing to stereoelectronic preferences imparted by the sultam sulfonyl moiety. Marked secondary-shell solvation effects are gleaned from the rate studies.

Graphical Abstract

Sodium camphorsultam enolates are structurally characterized uncovering monomers and dimers; rate studies reveal ionic transition structures and curious solvation effects.

Introduction

Alkylations of camphorsultam-derived enolates referred to colloquially as Oppolzer enolates have occupied a central position in asymmetric synthesis (eq 1).¹ In addition to the relatively low-cost auxiliary, there is an unmistakable appeal to the widely held logic that a chelate of type 1 imparts rigidity while the camphor methyl protruding above the enolate directs the electrophile to the endo face of the bicyclo[2.2.1]heptane ring system. The Oppolzer enolates merged our interest in understanding the structure-reactivity-selectivity relationships of alkali metal enolates² with efforts to promote the long-overlooked potential of organosodium chemistry.³

(1)

In the first of a two-part study, we explored the structures of alkyl- and aryl-substituted lithiated Oppolzer enolates in a variety of solvents and their reactivities in THF/HMPA.⁴ Focusing on the aryl cases was justified by a notable lack of their development for generating medicinally important aryl propionate derivates,⁵ and they were more tractable in what proved to be a complex study under the best of circumstances. There are, of course, always surprises when one probes alkali metal chemistry, the most poignant in that study was that the alkylation proceeds via a mechanism based on a solvent-separated ion pair (1). The enolate showed a marked preference to rotate nearly 180 degrees relative to chelate 2 with the preferred facial attack coming from the exo face proximate to the protruding camphor methyl. Needless to say, the irrelevance of chelation, the counterion, and even van der Waals interactions with the camphor methyl moiety flew in the face of conventional wisdom. We attributed the high selectivity to a stereoelectronic preference of the electrophile to approach anti to the endo-oriented sulfonyl oxygen.

Our motivations to examine the sodium enolates focused on confirming or refuting the stereochemical model in which the alkali metal cation was simply a counterion of little structural importance. We immediately confronted the first challenge: unbeknownst to us and probably the practitioners who routinely use NaHMDS/THF to enolize the acylated Oppolzer sultam,^6,7 the alkyl-substituted substrates (3a–f below) are only partially (K_eq < 1) metalated by NaHMDS (eq 2).⁸ Although the transiently formed sodium enolate appears to be adequate for the applications, the lack of full enolization makes enolate structure-reactivity relationships elusive.

We considered several approaches to overcoming this issue. First, rate studies with NaHMDS and alkyl-substituted sultams as the resting state would provide detailed insights into the mechanism of the enolate alkylation and were appealing given the detailed understanding of the aggregation and solvation of NaHMDS in solution.^3a Second, we could retreat to the aryl-substituted cases that are quantitatively metalated, which also allows direct comparisons with the aryl-substituted lithium enolates. Third, a stronger base would readily generate the alkyl- and aryl-substituted enolates quantitatively. Although sodium diisopropylamide (NaDA) offers more than adequate basicity,^3b the excellent stability and solubility profiles of sodium isopropyl(trimethylsilyl) amide (NaPTA) with an additional 4–5 pK_b units above NaHMDS make it a far superior choice (eq 2).^3c

(2)

We describe herein studies showing a high penchant for sodiated Oppolzer enolates to reside as monomers in THF solution and dimers when solvated by N,N,N’,N’-tetramethylethylenediamine (TMEDA) and N,N,N’,N’’,N’’-pentamethylethylene triamine (PMDTA) in toluene. Rate studies show that the ion-pair-based model represented by 2 held firm. The results also shed light on the role of HMPA and how two research groups managed to omit the HMPA through careful control of reaction conditions.⁷

Results and Discussion

As foreshadowed in the introduction, the alkyl-substituted substrates 3a–f (Scheme 1) require the more basic NaPTA.^3c Substrates 3g–j bearing aryl substituents were enolized using solutions of NaHMDS or NaPTA in a variety of solvents including toluene, MTBE, and THF. Enolates solvated by diamine TMEDA and triamine PMDTA were generated in toluene. Aging studies revealed no evidence of temperature-dependent aggregate changes and that decomposition commenced above 0 °C.

Scheme 1.

The N-acyl camphorsultam derivatives and their enolates that are studied in this work.

We routinely determine enolate aggregation states using the Method of Continuous Variations (MCV)^3a,4,9 in which binary mixtures of structurally similar enolates or even enolate enantiomers afford heteroaggregates whose number, symmetries, and mole-fraction dependencies monitored using ¹³C, ¹H, or ¹⁹F NMR spectroscopies attest to the aggregation states of the homoaggregates. This study, however, is dominated by monomeric enolates in THF wherein binary mixtures of either aryl- or alkyl-substituted enolates afford no heteroaggregates. Density functional theory (DFT) computations provided best-guess solvation numbers of the monomers,^10,11,12 which we use to interpret the rate data.

THF

Two enolates showing lack of heteroaggregation are illustrated in Figure 1. The region of the ¹³C NMR spectrum shown affords the most well-resolved hetero- and homoaggregate resonances when heteroaggregates appear (vide infra). Analogous pairings showing no additional resonances expected for heteroaggregates are in the Supporting Information. Both aryl- and alkyl-substituted enolates are monomeric in neat THF. The rate studies described are also consistent with the monomer assignment. DFT computations on enolate 4g suggest di-, tri-, and tetrasolvated exo-chelates 5–7 are nearly isoenergetic. A more definitive distinction would have been more satisfying.

Figure 1.

Partial ¹³C NMR spectra of 0.20 M of 4i, a 1:1 mixture of enolates 4i and 4j, and 4j in THF at –20 °C. No new peaks appear indicating the absence of heteroaggregation consistent with a monomeric enolate.

At low THF concentrations or in weaker donor solvents such as MTBE or N,N-dimethylethylamine (DMEA) resonances of higher aggregates appear. Similarly, binary mixtures of enolates in these weak donor ligands afford additional resonances attributed to heteroaggregates. However, the results from lithium Oppolzer enolates revealed extraordinary stereochemical complexity within tetramers examined in excruciating detail.⁴ Such potential complexity, lower data quality for the sodium enolates when compared to their lithium counterparts, and occasional solubility problems left us with no appetite for trying to repeat such a study.

THF/HMPA

The predominant protocol for alkylation of sodiated Oppolzer enolates uses THF/HMPA mixtures.⁶ MCV analysis using binary enolate mixtures confirm the absence of heteroaggregation. Probing serial solvation by ³¹P NMR spectroscopy under optimal conditions can provide explicit solvation numbers, but resolution was insufficient to distinguish differential solvation states in this case. DFT computations suggest disolvate 8 and trisolvate 9 to be viable with a 2.5 kcal/mol preference for 9. This is not a definitive assignment but provides adequate context for the rate studies. The lithium enolates were shown spectroscopically and computationally to be disolvates analogous to 8.

TMEDA

N,N,N’,N’-tetramethylethylenediamine (TMEDA) has a longstanding reputation for promoting deaggregation despite mounting evidence over three decades of a more nuanced story.^13,14 The impetus to examine TMEDA comes from previous studies of both simple and complex sodium enolates which suggest that TMEDA, while not necessarily as strongly binding or as prone toward deaggregating sodium salts, is often superior to THF for controlling structure.¹⁴ Binary mixtures of enolates 4g and 4i in TMEDA/toluene show resonance duplications consistent with a heterodimer (Figure 2). DFT computations show exothermic serial solvation to form doubly chelated dimer. The dimer motif has three possible stereoisomers, 10–12, owing to chelation with the endo or exo sulfonyl oxygens (Scheme 2). DFT computations support doubly exo-chelated dimer 12 as the preferred form by 10.6 kcal/mol.

Figure 2.

Job plot showing relative integrations of homodimers of 4g (A₂, blue) and 4i (B₂, red) and the heterodimer (A₁B₁, orange) plotted against the measured mole fraction¹⁵ of 4g (Χ_A) for 0.20 M mixtures of sodium enolates 4g and 4i in 0.88 M TMEDA/toluene at –80 °C monitored by ¹³C{¹H} NMR spectroscopy. The curves result from a parametric fit to a dimer model.

Scheme 2.

Stereoisomers of κ²,κ²- TMEDA-solvated camphorsultam-enolate dimers. DFT computations support 12 as the preferred form.

PMDTA

N,N,N’,N’’,N’’-pentamethyldiethylenetriamine (PMDTA) has a penchant for forming κ³-solvated monomers,¹⁶ yet binary mixtures of 4g and 4h in 3.0 equiv PMDTA/toluene show marginally resolved but unambiguous heterodimers akin to those observed with TMEDA solvates (Supporting Information). Computational studies support κ²,κ²-solvated dimer 13 akin to TMEDA solvate 12. (Formally, there could be as many as 10 stereoisomers owing to the lower symmetry of the chelate and the exo and endo isomerism of the sultam sulfonyl moieties.)

Kinetics of alkylation

The mechanism of alkylation was studied using enolate 4g and allyl bromide (eq 3). The enolate was generated in situ using recrystallized NaHMDS.^3a Rates were monitored by in situ IR spectroscopy¹⁷ following the loss of enolate 9 (1616 cm⁻¹) and formation of product 14 (1701 cm⁻¹). A typical decay is illustrated in Figure 3. Alkylations under non-pseudo-first-order conditions (0.10 M enolate and 3.0 equiv of allyl bromide) displayed no unusual curvatures emblematic of intervening autocatalysis or autoinhibition.¹⁸

(3)

Figure 3.

Alkylation of 4g (0.025 M; 1616 cm⁻¹) with 0.275 M allyl bromide in 9.24 M THF/toluene at –78 °C to form 14 (1701 cm⁻¹). The red curve depicts an unweighted least-squares fit to eq 4, such that n = 1.00 ± 0.01, k_obsd = 0.0022 ± 0.0009, and [enolate]₀ = 0.0430 ± 0.0001.

Following protocols developed to study the lithium enolates with enolate 4g as the limiting reagent, the enolate order, n, is determined by best-fit to the non-linear van’t Hoff equation (eq 4).¹⁹ The curve in Figure 3 stems from such a fit. Considerable variation of n from run to run is compensated by replication, affording n =1.04 ± 0.25 from the 48 independent decays used to obtain values for k_obsd. The first order in excess allyl bromide was confirmed by a three-point control experiment showing a direct relationship of k_obsd to concentration. Curiously, comparing the relative rates for the lithium and sodium enolates under otherwise identical conditions afforded k_Li/Na = 2.7: the lithium enolate is almost threefold more reactive than the analogous sodium enolate.

$$\left[\text{enolate}\right]=\left\{\left(n-1\right)k_{\text{obsd}}\mathrm{t}+\text{[enolate]}{0}^{\left(1-n\right)}{\text{}}}^{1/\left(1-n\right)}\right.$$ (4)

As a preface to the rate studies described below it is constructive to consider the backdrop provided by the alkylations of the lithium enolates. In particular, first-order dependencies of the alkylation rates on alkyl halide and enolate monomer implicated a monomer-based alkylation. Second-order dependencies on both THF and HMPA suggested a hexasolvated ion pair. However, the THF dependence proved to be 100% secondary shell (medium) effects rather than primary solvation in which an inverse-first-order-dependence on the toluene cosolvent owing to stabilization of the HMPA in the reactant and a first-order dependence on THF owing to stabilization of the transition structure combined to create an apparent second-order THF dependence. Using 2,5-dimethyltetrahydrofuran (2,5-Me₂THF) removed both the influence of toluene and the drifting dielectric constant of the medium,²⁰ revealing a true zeroth-order dependence on primary-shell solvation by THF. We suspected this primary and secondary-shell solvation narrative would reappear with sodium.

In this event, the first-order dependencies on enolate and allyl bromide followed for the sodium enolates. A second-order HMPA dependence (Figure 4) implicated a ⁺Na(HMPA)₅ gegenion based on the computed trisolvated monomer 9. (If disolvate 8 is the dominant form, then we have a ⁺Na(HMPA)₄ gegenion.) The second-order THF dependence in toluene (Figure 5) and approximate zeroth-order THF dependence using 2,5-Me₂THF with a downward drift (see insert) mirrored the lithium enolates. Once again, the second order in THF is attributed to a first-order medium effect on THF and an inverse-first-order medium effect on toluene (Figure 6), both of which were negated using 2,5-Me₂THF as cosolvent.

Figure 4.

Plot of k_obsd against [HMPA]_free for the alkylation of enolate 4g (0.025 M) with 0.275 M allyl bromide in 9.24 M THF/2,5-Me₂THF at –78 °C.²¹ The blue curve depicts an error-weighted least-squares fit to f(x) = ax^b such that a = 0.061 ± 0.001, b = 2.56 ± 0.01.

Figure 5.

Plot of k_obsd versus THF concentration for alkylation of 0.025 M 4g with allyl bromide (0.275 M) in 0.250 M free HMPA²¹ at –78 °C in toluene (■) or 2,5-Me₂THF (●) co-solvent. The blue curve depicts an error-weighted least-squares fit to f(x) = y₀ + ax^b such that y₀ = 1.71 × 10⁻⁴ ± 0.57 × 10⁻⁴, a = 8.16 × 10⁻⁶ ± 5.10 × 10⁻⁶, b = 2.2 ± 0.2. The red curve depicts an error-weighted least-squares fit to f(x) = y₀ + ax such that y₀ = 3.62 × 10⁻³ ± 0.14 × 10⁻³, a = –1.5 × 10⁻⁴ ± 0.17 × 10⁻⁴.

Figure 6.

Plot of k_obsd versus toluene concentration for the alkylation of 0.025 M 4g at –78 °C with 0.275 M allyl bromide in 0.250 M free HMPA in 0.275 M THF using cyclopentane cosolvent.²¹ The curve depicts a least-squares fit to f(x) = y₀ + ax^b such that y₀ = 2.13 × 10⁻⁴ ± 0.58 × 10⁻⁴, a = 3.30 × 10⁻⁴ ± 0.79 × 10⁻⁴, b = –0.98 ± 0.24.

Omitting HMPA

Hoping to eliminate HMPA from the protocol⁷ we tried alkylations using TMEDA and PMDTA without much luck. In neat THF the alkylation (eq 5) is approximately 100-fold slower than in THF with 10 equiv of HMPA. However, alkylation with the highly reactive methyl iodide is sufficiently fast to allow the methylation to occur at −40 °C over 2 hours, which is sufficiently below the −20 °C onset of decomposition.

(5)

The highly abbreviated mechanistic story is that alkylation follows the rate law in eq 6 (see Supporting Information), which, given the assignment of enolate 4g as tetrasolvated monomer, is consistent with a hexa-solvated ion-pair-based mechanism. This is the mechanism found for the HMPA-solvated lithium and sodium enolates with the only difference being the ⁺Na(THF)₆ counterion. Supporting data suggests to us that ⁺Na(THF)₆ is the most probable.²² As a reminder, the assignment of enolate 4g in neat THF-solvated monomer as a tetrasolvate was not unassailable, though the ⁺Na(THF)₆ is computationally quite credible.

$$-d\left[\text{enolate}\right]/d\text{t}=k^{\prime }\left[\text{enolate}{]}^1\right[\text{THF}{]}^2{[\text{Mel}]}^1$$ (6)

Conclusions

The structural and rate studies of sodiated Oppolzer enolates in THF/HMPA solution are strongly aligned with the results from the lithium enolates in THF/HMPA, differing only in the solvation number of the monomeric reactant and the counterion in the solvent-separated ion-pair.²³ Consequently, the stereochemical model involving an open transition structure crudely depicted as 1 in the introduction and as the ball-and-stick structure in Figure 7 differs only in the remote ⁺MS_n counterion—⁺Li(HMPA)₄ or ⁺Na(HMPA)₅. The secondary shell influences of toluene on the reactant HMPA and THF on the transition structure are some of the most dramatic we have seen from literally hundreds of rate studies. We will not reiterate the extensive computations pointing to the stereoelectronically preferential attack of the electropile anti to the endo sulfonyl oxygen.⁴ Curiously, several recent reactions that do not involve potentially chelate-forming at all can be explained by the same stereoelectronic control.²⁴ One might surmise that the only advantage of sodium over lithium is the much higher reactivity but that would be wrong. The lithium and sodium enolates studied show a threefold greater reactivity of the lithium enolate.

Figure 7.

DFT computed transition structure for the alkylation of enolate 4g with allylbromide. Si-facial approach of the electrophile is favored (2.6 kcal mol⁻¹) over the re face due to stereoelectronic effects emanating from the sultam sulfonyl moiety.⁴

We can take some solace in locating a narrow window of reactivity using the highly reactive methyl iodide in which the undesirable HMPA can be excluded as reported by two groups.⁷ The greater efficacy of NaPTA when compared with NaHMDS for generating recalcitrant sodium enolates is also notable.

Experimental

Reagents and solvents

NaHMDS and NaPTA were prepared as white crystalline solids.^3a,c toluene, hexanes, THF, MTBE, and cyclopentane were distilled from blue or purple solutions containing sodium benzophenone ketyl. Methyl iodide and allyl bromide were distilled from 4Å molecular sieves. Substrates 3a–j were prepared according to literature procedures.^1,4

NMR spectroscopic analysis

An NMR tube fitted with a double-septum under vacuum was flame-dried on a Schlenk line and allowed to cool to room temperature, backfilled with argon, and placed in a dry ice/acetone cooling bath. Individual stock solutions of the N-acyl sultams and NaHMDS were prepared at room temperature and 0 °C, respectively. The appropriate amounts of the N-acyl sultams, NaHMDS, solvent, and (when applicable) co-solvent were added sequentially via a gas-tight syringe. The tube was flame-sealed under a partial vacuum while cold to minimize evaporation. The tubes were mixed on a vortex mixer and stored at −80 °C.

Unless otherwise stated all tubes were sealed with a total enolate concentration of 0.10 M. Standard ¹H, ¹⁹F, ¹³C, and ³¹P direct detection spectra were recorded on a 11.8 T spectrometer at 500.1, 470.6, 125.8, and 202.5 MHz, respectively. ¹H, ¹³C, and ³¹P resonances are referenced to their respective standards (Me₄Si and H₃PO₄ at 0.0 ppm). ¹⁹F spectra are referenced to C₆H₅F (–113.15 ppm).

For quantitated ¹³C spectra, the spin-lattice relaxation (T1) was determined by standard inversion-recovery experiments on several samples. The relaxation delay (d1) was set to seven times the average relaxation lifetime. Integration of the NMR signals was determined using the line-fitting method included in MNova (Mestrelab research S.L.).

Rate studies

IR spectra were recorded with an in situ IR spectrometer (ReactIR iC 10, Mettler Toledo AutoChem) fitted with a 30-bounce, silicon-tipped probe. The spectra were acquired at a gain of 1 and a resolution of 4 cm⁻¹. All tracked reactions were conducted under a positive pressure of argon from a Schlenk line. A representative reaction was carried out as follows:

The IR probe was inserted through a Teflon 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 fitted with a septum for injections and an argon line. After evacuation under full vacuum, heating, and flushing with argon, the flask was charged with the THF/cosolvent mixture of choice (toluene, 2,5-Me₂THF, toluene/cyclopentane) and cooled to −78 °C in a dry ice−acetone bath. A set of 256 baseline scans were collected and IR spectra were recorded every 15 seconds from 30 scans. The reaction vessel was charged to 0.025 M 4g (1704 cm⁻¹). A 1.00 M stock solution of NaHMDS was injected (0.030 M, 1.2 equiv) through the septum, and enolization was tracked to completion (1616 cm⁻¹), typically ~10 min. Following full disappearance of 4g, HMPA was added to the reaction as a 4.31 M (75 v/v%) stock solution in toluene to avoid freezing. For IR studies of the HMPA-free methylation this step was omitted. The reaction was left to stir for another 10 min. At this point, the spectral collection was halted, and an additional 256 baseline scans were collected. The spectrometer was configured to collect spectra every 5 seconds from 16 scans. 1 set of scans was collected before the addition of neat alkylating agent (methyl iodide or allyl bromide) through the septum. The reaction was tracked over 5 half-lives monitoring the disappearance of the enolate (1616 cm⁻¹) and the appearance of the adduct (1701 cm⁻¹).

Density functional theory (DFT) computations

All DFT calculations were carried out using Gaussian 16.⁹ Prompted by a recent benchmarking of modern density functionals, all calculations were conducted at the M06–2X level of theory using Grimme’s zero-dampened DFT-D3 dispersion corrections.^10a–d A pruned (99, 590) integration grid (equivalent to Gaussian’s “UltraFine” option) was used for all calculations. Where appropriate solvation effects were accounted for by the Self Consistent Reaction Field method using the SMD model of Truhlar and coworkers.^10e Jensen’s polarization-consistent segment-contracted basis set, pcseg-1, was used for geometry optimizations and the expanded pcseg-2 basis set for single-point energy calculations.^10f Basis set files were obtained from the Basis Set Exchange.^10g Ball-and-stick models were rendered using CYLview 1.0b.^10h A large number of DFT-computed energies are archived in the Supporting Information.¹¹ A frequency calculation was conducted at all stationary points to ensure the existence of real minima. All reported geometries have exactly zero imaginary (negative) vibrational frequencies.