Catalysis of an S_N2 pathway by geometric preorganization

Abstract

Bimolecular nucleophilic substitution (S_N2) mechanisms occupy a central place in the historical development and teaching of the field of organic chemistry¹. Despite the importance of S_N2 pathways in synthesis, catalytic control of ionic S_N2 pathways is rare and notably uncommon even in biocatalysis^2,3, reflecting the fact that any electrostatic interaction between a catalyst and the reacting ion pair necessarily stabilizes its charge and, by extension, reduces polar reactivity. Nucleophilic halogenase enzymes navigate this tradeoff by desolvating and positioning the halide nucleophile precisely on the S_N2 trajectory, using geometric preorganization to compensate for the attenuation of nucleophilicity⁴. Here we show that a small-molecule (646 Da) hydrogen-bond-donor (HBD) catalyst accelerates the S_N2 step of an enantioselective Michaelis–Arbuzov reaction by recapitulating the geometric preorganization principle employed by enzymes. Mechanistic and computational investigations reveal that the HBD diminishes the reactivity of the chloride nucleophile yet accelerates the rate-determining dealkylation step by reorganizing the phosphonium chloride ion pair resting state into a geometry that is primed to enter the S_N2 transition state. This new enantioselective Arbuzov reaction affords highly enantioselective access to an array of H-phosphinates, which are in turn versatile P-stereogenic building blocks amenable to myriad derivatizations. This work constitutes, to our knowledge, the first demonstration of catalytic enantiocontrol of the phosphonium dealkylation step, establishing a new platform for the synthesis of P-stereogenic compounds.

Main Text:

Bimolecular S_N2 pathways between charged species generally proceed more rapidly in polar aprotic solvents than in protic solvents. For example, the reaction of azide with trimethylsulfonium is decelerated by ~10⁵ in methanol relative to the same reaction in acetone (Fig. 1A).⁷ The classical interpretation is that H-bonding interactions from protic solvent molecules stabilize the ionic reactants more strongly than the partially charged S_N2 transition state, leading to a higher activation barrier.^1,7 Selective catalysis of reactions involving ionic S_N2 mechanisms must overcome the fact that any explicit electrostatic interactions with an ion pair are likely to stabilize the reactants relative to the transition state and result in rate attenuation rather than acceleration.

Fig. 1. Applying enzymatic strategies to the catalysis of ionic S_N2 mechanisms.

(A) Qualitative potential energy surfaces for an S_N2 step with (orange) and without (grey) hydrogen bonding from protic solvent. (B) FDAS: an enzymatic exemplar of nucleophile desolvation and preorganization for the catalysis of S_N2. The X-ray crystal structure of the active site bound to the product 5-fluorodeoxyadenosine is shown (PDB 2V7V)^5,6. (C) Qualitative potential energy surfaces for an S_N2 step showing the energetic consequences of substrate desolvation and preorganization, leading to an overall lower reaction barrier. (D) Design concept for accelerating S_N2 reaction with a dual hydrogen-bond donor catalyst, whereby a desymmetrizing nucleophilic substitution in the chiral active site affords a P-stereogenic product.

Of the few enzyme classes that promote S_N2 mechanisms, the majority employ general acid–base catalysis to activate uncharged nucleophiles or electrophiles.² An alternative mode of catalysis has been documented for the nucleophilic halogenase 5’-fluoro-5’-deoxyadenosine synthase (FDAS), which promotes fluoride (or chloride) displacement on a cationic S-adenosyl methionine (SAM) (Fig. 1B).⁸ Solid-state X-ray structural characterization (Fig. 1B), along with stereochemical, theoretical, and kinetic studies, have established that the enzyme effects rate acceleration of ~10⁶ by the halide-binding active site (i) precisely positioning the halide in a collinear relationship to the leaving group^5,6,8 and (ii) providing a “halide hole” to offset the energetic penalty for desolvation of the nucleophile from water⁴. In this enzymatic mechanism, electrostatic stabilization of the transition state relative to the ground state is neither necessary nor sufficient for rate acceleration⁹; rather, the enzyme active site is preorganized to stabilize a “near attack” ground-state conformation that resembles the transition state geometrically¹⁰. The reliance on ground-state preorganization is necessary because electrostatic interactions are more thermodynamically favorable with the fully charged ground state than with the activated complex.

We envisioned a biomimetic approach to designing small-molecule catalysts for ionic S_N2 mechanisms, guided by the FDAS structure–mechanism relationship. Enzyme-mimetic small-molecule organocatalytic systems have been documented to achieve rate acceleration by charge stabilization of the transition state through non-covalent interactions,^11,12 or to recapitulate enzyme-like levels of stereodifferentiation^13–15, yet none to our knowledge has explicitly leveraged the enzymatic principle of geometric ground state preorganization. We posited that successful catalysis is practicable if, like FDAS, a catalyst could (i) precisely preorganize an ion pair into a geometric configuration primed for S_N2 nucleophilic substitution and (ii) provide the immediate solvation shell for the nucleophile (Fig. 1C).

Based on prior work in which dual-hydrogen-bond donor (HBD) organocatalysts were shown to control the reactions of pre-generated ion pairs,^16,17 we considered that an appropriate adaptation within the tert-leucine–arylpyrrolidine HBD family might fulfill both criteria, in addition to providing a chiral environment for enantioinduction^18–22. As a reaction platform to probe our catalytic hypothesis, we selected the Michaelis–Arbuzov reaction^23–25. The key step in the Arbuzov reaction is the S_N2 dealkylation of a phosphonium intermediate, an elementary step that also underpins a broad manifold of other synthetically important reactions of organophosphorus compounds. Thus, achieving catalytic enantiocontrol over this step could unlock catalytic P-asymmetric variants of reactions such as the Appel, Staudinger, and Pudovik reactions^26–28. The envisioned transformation starts from an achiral P(III) species with two identical alkoxy substituents that would present enantiotopic sites for S_N2 dealkylation upon formation of the tetrahedral phosphonium cation (Fig. 1D). If the S_N2 step is turnover-limiting, achieving high selectivity would further require that the catalyst accelerates the S_N2 step over the racemic uncatalyzed reaction. Here, we demonstrate that a precisely designed chiral HBD accelerates S_N2 dealkylation of phosphonium halide ion pair intermediates in highly enantioselective Michaelis–Arbuzov reactions (Fig. 1C). The resulting H-phosphinate products are versatile intermediates for the preparation of stereogenic-at-phosphorus(V) compounds^29–31.

Reaction development

An empirical survey of variants of the Arbuzov reaction with chiral HBD catalysts revealed promising enantioselectivity for the dealkylation of dibenzyl phenylphosphonite 2a with HCl (Fig. 2A). Thiourea 1c promoted the model reaction with higher enantioselectivities than its squaramide (1a) or urea (1b) counterparts (Fig. 2A). Furthermore, the size of the (poly)aromatic substituent as well as α-quarternary substitution at the pyrrolidine correlated positively with enantioselectivity. Ultimately, we found that thiourea 1g bearing a 2-phenanthryl substituent promoted dealkylation in 90% ee and quantitative yield. Although catalyst optimization studies were performed using two equivalents of HCl in methyl tert-butyl ether (MTBE), measurable enhancements in enantioselectivity were achieved by changing the reaction solvent to toluene (Fig. 2B, entry 1), employing a single equivalent of HCl (entry 2), and diluting the reaction mixture to 50 mM (entry 3) to furnish 3a in 95% ee.

Fig. 2. Reaction discovery and isolation of catalyst components.

Yields were quantified by ¹H NMR relative to an internal standard. MTBE, methyl tert-butyl ether. (A) HBD catalyst survey for enantioselective dealkylation of phosphonite 2a and optimization of reaction conditions. *Reaction was quenched with TMSCHN₂ in Et₂O (1.5 equiv) and the yield was quantified by quantitative ³¹P NMR. (B) Kinetic studies probing the role of the hydrogen-bond-donor and t-Leu–arylpyrrolidine domains in rate acceleration. Reaction progress was monitored by following product 3a formation using in situ IR spectroscopy.

Mechanistic investigations

Catalyst 1g was found to accelerate the reaction of 2a by a factor of ca. 30 relative to the uncatalyzed pathway (Fig. S12). As a first step toward elucidating the mechanism of catalysis, and intrigued by the possibility that promotion of the S_N2 step in a manner similar to FDAS might underlie the observed acceleration, we made simple synthetic modifications to 1g to isolate its anion-binding (“halide hole”) domain from the putative cation-binding t-Leu–arylpyrrolidine domain (Fig. 2B). The simple anion-binding variant 4 was constructed by replacing the t-Leu–arylpyrrolidine moiety with a n-octyl group and, when used in superstoichiometic amounts to mitigate uncatalyzed background reactivity, was found to inhibit the dealkylation reaction below the rate of the background reaction (Fig. 2B, yellow vs. grey traces). Thus, simple thiourea analogs of catalyst 1g mimic protic solvents in their ability to inhibit ion-pair collapse. To probe the role of the isolated t-Leu–arylpyrrolidine domain in catalysis, variant 5 was synthesized by S-methylation of the thiourea, thereby removing the dual-HBD properties of catalyst 1g. Compound 5 induced no rate acceleration compared to the background reaction (Fig. 2B, blue). Combining 10 mol % each of 4 and 5 also did not result in rate acceleration compared to background. Taken together, these observations provide compelling evidence that the presence of both domains as well as their precise relative spatial orientation as in 1g are necessary for catalysis.

We undertook a systematic mechanistic investigation to probe the identity and molecular composition of the turnover-limiting and enantiodetermining step(s), guided by the protonation–dealkylation Arbuzov mechanistic sequence. Evidence for the intermediacy of a protiophosphonium ion in the catalytic system was obtained by monitoring the reaction of 2a with HCl catalyzed by thiourea 1g by ³¹P NMR (Fig. 3A). At the earliest timepoints the signal due to 2a at δ 159 ppm is not observed, replaced instead by a doublet at δ 52 ppm that diminishes gradually with concomitant appearance of product 3a at δ 24 ppm (Fig. S2). The transient signal at δ 52 is assigned to the phosphonium chloride intermediate 6a based on the 710 Hz (J¹_P–H) coupling constant³² and the spectrum of an independently prepared phosphonium tetrafluoroborate (Fig. S1). These observations indicate that phosphonium chloride 6a is the resting state of the substrate under the catalytic conditions.

Fig. 3. Mechanistic studies.

(A) Reaction monitoring by ³¹P NMR. Reaction was conducted at 50 mM concentration in toluene-d₈ at −70 °C. (B) Kinetic studies to determine the order of catalyst 1g by varying catalyst loading. Reaction progress was monitored by following product 3a formation using in situ IR spectroscopy. (C) Change in the concentration of reaction components over the reaction timecourse, monitored by in situ IR spectroscopy. (D) Proposed catalytic cycle.

The resting-state form of catalyst 1g was investigated by employing diffusion-ordered NMR spectroscopy (DOSY) to measure its diffusion constant under catalytic reaction conditions as a means of estimating its molecular weight (see Supplementary Information, Section 8.2). The trifluoromethyl groups on 1g provided a sensitive handle for ¹⁹F NMR measurements at concentrations of 1g appropriate to the catalytic conditions (10 mM, 5% Et₂O/toluene-d₈, −70 °C). After calibration studies³³, we established that a solution of catalyst 1g alone affords a measured molecular weight that is consistent with a monomeric state (MW_det = 640, MW(1g) = 646, ΔMW = 1%). Under reaction conditions that reliably afford high yield and enantioselectivity, the measured molecular weight of the catalyst resting state is consistent with a 1:1 1g·6a complex (MW_det = 1043, MW(1g·6a) = 1005, ΔMW = −4%) (Fig. 3B). Upon completion of the reaction, the measured molecular weight is intermediate between that of monomeric 1g and a 1:1 1g·3a complex (MW_det = 739, MW (1g·3a) = 878, MW(1g) = 646, ΔMW = 16%/−14%), consistent with an equilibrium between free and product-bound catalyst, and inconsistent with strong catalyst–product binding. These observations establish that catalyst 1g rests as a 1:1 complex with phosphonium 6a and that rate-limiting dealkylation proceeds from this complex.

With the molecular composition of the resting-state complex established, we endeavored to determine the kinetic dependence on [1g]_T and [6a] to elucidate the stoichiometry of the rate-determining transition-state complex (Fig. 3B). The distinct infrared absorbance of the P=O bond in 3a (1240 cm⁻¹) and the symmetric P–O stretch in 6a (1039 cm⁻¹) provided excellent handles to monitor reaction progress by in situ IR spectroscopy. By systematically varying the concentration of 1g and applying Bures’ normalized time scale treatment to the reaction profiles³⁴, we obtained excellent graphical overlay only when dividing the reaction profiles by [1g]ⁿ for n = 1, indicating that the reaction rate exhibits first-order dependence on catalyst [1g]_T. Furthermore, the consumption of phosphonium species 6a and the formation of product 3a both follow a zeroth-order kinetic rate behavior for the first ~80% of the reaction (Fig. 3C), consistent with a turnover-limiting and enantiodetermining dealkylation transition state proceeding from a 1:1 1g·6a resting-state complex.

A catalytic cycle consistent with all available mechanistic data is depicted in Fig. 3D. This cycle features (i) a protonation equilibrium that favors the phosphonium chloride 6a, (ii) binding of 6a to monomeric catalyst 1g forming the resting state, and (iii) turnover-limiting and enantiodetermining dealkylation to form product 3a and benzyl chloride, which dissociate from the catalyst to turn over the catalytic cycle.

Computational analysis

The relatively small size of catalyst 1g rendered the S_N2 step amenable to explicit modeling by density-functional theory (DFT). Modeling of the catalyzed and uncatalyzed dealkylation reactions of phosphonium chloride 6a was performed with continuum solvation in a low dielectric (PCM, toluene, $ϵ=2.38$) to mimic the experimental reaction conditions (Fig. 4A). The interactions between the catalyst and the phosphonium chloride were further analyzed using the independent gradient model based on Hirshfeld partition (IGMH) (see Supplementary Information, Section 10.4). In the absence of catalyst, the phosphonium chloride 6a is found to rest as a tight ion pair with a cage-like structure in which the chloride anion engages in multiple stabilizing interactions with the cation. All of the stabilizing H-bonding interactions and a significant portion of the Coulombic attraction present in the ground state must be sacrificed, in a manner loosely analogous to desolvation from the aqueous medium in the halogenase reaction, to attain the linear geometry mandated by the S_N2 mechanism as in TS_uncat³⁵. In considering this geometric reorganization, the concerted pathway for dealkylation can be partitioned conceptually into an ion-pair reorganization phase followed by an ion-pair collapse phase. These two phases can be demarcated by a non-stationary state 6a’ located on the computed intrinsic reaction coordinate, wherein the chloride ion is positioned along the S_N2 trajectory but formation of the C–O bond has yet to commence. By this analysis, >75% of the overall electronic activation barrier results from reorganization of the chloride anion in the first phase (6a → 6a’), whereas the ion-pair collapse phase corresponding to the covalent bond-breaking and forming events contributes <25% (ca. 4 kcal/mol) to the overall barrier.

Fig. 4. Origins of catalytic rate acceleration.

All calculations were performed at the B3LYP-D3/6–311++G(d,p)/PCM(Toluene)//B3LYP-D3/def2-SVP/PCM(toluene) level of theory at 195.15 K and 1 atm. Most hydrogens are hidden for clarity. (A) Calculated potential energy surfaces for uncatalyzed background (grey) and catalyzed (red) pathways. Species 6’ is a non-stationary state structure corresponding to maximum P–Cl separation on the intrinsic reaction coordinate of the background reaction. Reported values are electronic energies without zero-point vibrational corrections. (B) Graphical representation of key reaction species and geometrical changes associated with the reorganization and collapse phases of the catalyzed and uncatalyzed pathways.

In the computational model of the catalyzed pathway, the nucleophilic substitution found to proceed in two discrete steps (Fig. 4A). Three critical points along the reaction coordinate may be distinguished: binding of 1g to the ion pair 6a, the reorganized ion pair 6a’, and the transition state of the dealkylation step. In all three structures, the catalyst engages in a network of stabilizing noncovalent interactions with both cationic and anionic components of the phosphonium chloride ion pair (Fig. 5B, bottom), in agreement with previous studies on thiourea–arylpyrrolidine scaffolds²². The chloride binds to the thiourea and the phosphonium associates with the arylpyrrolidine domain. These interactions appear to stabilize both charged components of the ion pair. From the resting-state complex 1g·6a, a structure 1g·6a’ could be located as a stationary state only 1.9 kcal/mol uphill, which is on the intrinsic reaction coordinate to the lowest-energy transition state TS_cat,R (vide infra).

Fig. 5. Origins of enantioselectivity.

All calculations were performed at the B3LYP-D3/6–311++G (d,p)/PCM(toluene)//B3LYP-D3/def2-SVP/PCM(toluene) level of theory at 195.15 K and 1 atm. Most hydrogens are hidden for clarity. (A) Analysis of non-covalent interactions (B) Density-functional-theory-modeled diastereomeric transition states for dealkylation with highlighted differential stabilizing interactions hypothesized to be the origin of enantioinduction.

Remarkably, the geometrical features of the phosphonium chloride resting-state complex 1g·6a’ are very similar to those in 6a’, i.e. the chloride is positioned by hydrogen bonding in a nearly optimal pre-transition-state geometry (Fig. 4B), primed to enter TS_cat,R. Thus, catalyst 1g can be seen as engaging a network of attractive noncovalent interactions to access a relatively stable ground-state complex 1g·6a’ that is primed for the dealkylation reaction. The computational analysis enables a quantitative assessment of the effect of catalyst 1g on both the ion-pair reorganization and the ion-pair collapse. Catalyst association raises the barrier to ion-pair collapse relative to the uncatalyzed pathway (7.5 vs. 4.0 kcal/mol, Fig. 4A), consistent with the expected attenuating effect of H-bonding on the nucleophilicity of chloride. However, inhibition of ion-pair collapse is more than offset by the catalyst mitigating the energetic cost of the requisite geometric preorganization of the phosphonium chloride ion pair (1.9 vs. 13 kcal/mol, Fig. 4A), resulting in overall acceleration relative to the background reaction.

Origin of Enantioinduction

A systematic conformer search led to the identification of low-energy diastereomeric structures leading to the major (R) and minor (S) enantiomers of 3a, with relative energies in excellent agreement with the experimentally observed enantioselectivities. Close analysis of the computational models with the IGMH approach³⁶ (Fig. S14) reveals specific stabilizing interactions that might be responsible for enantioinduction. The diastereomeric transition state structures leading to the minor (S) and major (R) product enantiomers display almost identical catalyst geometries but are related by a 120-degree rotation of the phosphonium ions within the catalyst active site. TS_cat,S positions the phosphonium P-Ph group below the 2-phenanthryl group of the catalyst, while splaying its benzyl groups (highlighted green). The lower energy TS_cat,R, in contrast, positions the phosphonium P–Ph group into solvent and stacks the two benzyloxy groups, with one residing below the 2-phenanthryl group of the catalyst. While TS_cat,S and TS_cat,R both possess several noncovalent attractive interactions in common, TS_cat,S possesses one additional H-bonding interaction between a benzylic C–H and the amide oxygen (Fig. 5B. orange box) while TS_cat,R incorporates two benzylic C–H–π interactions in (Fig. 5B, blue box). The net energetic benefit of losing one benzylic C–H hydrogen bonding interaction and gaining two benzylic C–H–π interactions is thus proposed to play a key role in dictating the sense and magnitude of enantioinduction³⁷.

Substrate scope development and product derivatization

Finally, we examined the substrate scope of our discovered protocol, recognizing that the stereocontrolled synthesis of stereogenic-at-phosphorus compounds is a topic of considerable current interest^28–30. A wide array of dibenzyl phosphonites proved compatible as substrates, furnishing air- and moisture-stable chiral H-phosphinate products that can be purified by silica gel column chromatography (Fig. 2B). Various para-substituted arylphosphonites underwent dealkylation with good yield and enantioselectivity (3a–3g), with lower enantioselectivity observed for substrates bearing highly electron-withdrawing substituents (3h–3j). Meta-substituted phenylphosphonites underwent dealkylation with comparable enantioselectivities to their para-substituted regioisomers (3k–3m). In contrast, sterically demanding ortho substituents proved deleterious for enantioselectivity: o-fluoro substitution decreased enantioselectivity from 90% (3f) to 73% ee (3n), whereas o-phenyl substitution ablated enantioselectivity (3o). Substrates bearing ortho-fused polyaromatic substituents, however, underwent dealkylation with high enantioselectivity (3p and 3q). The method is also compatible with a variety of hetero- and polyaromatic substituents (3r–3t and 3v), including acid-labile functional groups (N-Boc-protected indole, 3u). Phosphonites with non-aryl substituents such as isopropenyl (3w) and adamantyloxy (3x) underwent dealkylation with moderate levels of enantioselectivity while alkyl substituents such as cyclopropyl and methyl afforded low levels of enantioselectivity at 54% and 30% ee respectively (see Supplementary Information, Section 3.1). The enantioselective dealkylation of 2a was performed successfully on a gram scale employing only 3 mol % 1g with high enantioselectivity (93% ee), yield (98%), and efficient catalyst recovery (95%).

Most notably, this protio-Michaelis–Arbuzov reaction yields enantioenriched H-phosphinates, P-stereogenic building blocks with broad synthetic utility but that are currently only accessible by stoichiometric approaches or kinetic resolution,^29,38,39 differentiating our approach from previous catalytic approaches that desymmetrize a P(V) compound by nucleophilic substitution.^40,41 The substituents on the H-phosphinate products obtained through this enantioselective dealkylation are known to exhibit complementary reactivity as the proton and alkoxide groups on phosphorus are prone to substitution by electrophiles and nucleophiles, respectively.²⁹ To highlight the synthetic utility of enantioenriched H-phosphinates, we explored the reactivity of (R)-benzyl phenylphosphinate 3a as an orthogonally bifunctionalizable P-stereogenic building block. We found it amenable to an array of stereospecific synthetic elaborations of the P–H moiety, followed by secondary derivatization of the P–OBn moiety (Fig. 3D). Building on established phospha-Mannich reactivity of phosphinates,⁴² we found that deprotonation with lithium phenoxide enabled 3a to participate in the Pudovik addition to Eschenmoser’s salt,⁴³ affording α-amino phosphinate 7a. The benzyloxy group of 7a could subsequently be substituted in the presence of methyllithium to afford 8a, preserving enantioenrichment at phosphorus.^44–46 We also explored a polarity reversal strategy in the context of the Atherton–Todd reaction,⁴⁷ in which the nucleophilic P–H moiety was converted to electrophilic P–Cl then trapped with heteroatom-based nucleophiles. In this context, nucleophiles such as benzylamine and tyrosine could be employed to access phosphonate esters and phosphonamidates 9a/b with high yields and enantiospecificities. To investigate phosphoryl-radical-mediated reactivity, we adapted the lithium phenoxide conditions to a previously reported 1,6-coupling between H-phosphinates and benzoquinones⁴⁸, finding that O-phosphoryl hydroquinone derivative 10a was obtained with excellent stereospecificity. This result indicates that the 3a-derived phosphoryl radical possesses sufficient configurational stability in the absence of chiral control elements to engage in stereospecific reactions⁴⁹. Finally, 3a was subjected to a sulfuration–methylation sequence to afford phosphonothioate 11a with excellent enantiospecificity⁵⁰.

Outlook

Precise preorganization of reactants in a favorable pre-transition-state geometry, a mechanistic principle fundamental in enzyme catalysis but hitherto unexploited in small-molecule systems, has now been developed and characterized in the context of a HBD catalyst for enantioselective Michaelis–Arbuzov reactions that generate valuable P-stereogenic products and building blocks. The mechanistic scenario characterized here recapitulates the energetic origin for FDAS catalysis in a small-molecule model. We anticipate that this demonstration of the geometric preorganization principle in a small-molecule system will be broadly relevant to the catalysis of ionic pathways where ion-pair reorganization is a significant component of the turnover-limiting reaction barrier.

Fig. 6. Enantioselective phosphonium dealkylation of phosphonites and stereospecific elaborations of H-phosphinate products.

All reported yields are of chromatographically purified and isolated material. (A) Substrate scope. Reactions were conducted with 0.1 mmol of phosphonite. (B) Stereospecific elaborations. Enantiospecificity (es) is defined as ee_product/ee_reactant. *Reaction was conducted with 5 mmol 2a. No isobutylene oxide quench was employed, and 95% of catalyst 1g was recovered from column chromatography. †Eschenmoser’s salt (1.0 equiv), PhOLi (1 equiv.), THF, –40 °C, 17 h. ‡MeLi (1.5 equiv, 1.6 M in Et₂O), THF, −78 °C, 30 min, then 0 °C, 90 min. §Benzophenone (1.0 equiv), PhOLi (20 mol %), THF, −40 °C, 16 h. **Nucleophile (2.0 equiv), Et₃N (2.0 equiv), CCl₄ (2.0 equiv), Et₂O, 0 °C to rt, 7 h. ††2,6-di-tert-butyl-1,4-benzoquinone (1.0 equiv), PhOLi (20 mol %), THF, −40 °C, 16 h. §§S₈ (1.5 equiv), Et₃N (1.05 equiv), Et₂O, 0 °C to rt, 24 h, then MeI (2.0 equiv), Et₂O, rt, 2 h.

Data availability

The data that support the findings in this work are available within the paper and Supplementary Information.