Palladium catalyzed cross-coupling of alcohols with olefins by positional tuning of a counteranion

Abstract

Transition-metal catalyzed cross-couplings have great potential to furnish complex ethers; however, challenges in the C(sp³)-O functionalization step have precluded general methods. Here we describe computationally guided transition-metal ligand design that positions a hydrogen-bond acceptor (HBA) anion at the reactive site to promote functionalization. A general cross-coupling of primary, secondary, and tertiary aliphatic alcohols with terminal olefins to furnish complex ethers (>130) is achieved. The mild conditions tolerate functionality prone to substitution, elimination, and epimerization and achieve site-selectivity in polyol settings. Mechanistic studies support that the ligand geometry and electronics directs positioning of the phosphate anion at the π-allyl-Pd terminus and the phosphate’s HBA role to the alcohol. Ligand directed counteranion positioning in cationic transition metal catalysis has the potential to be a general strategy for promoting challenging bimolecular reactivity.

One-Sentence Summary:

Ligand directed counteranion positioning in cationic transition metal catalysis promotes challenging bimolecular reactivity.

Main Text:

A general method by which to accomplish cross-coupling of alcohols and olefins, the most abundant native functionalities in natural products, pharmaceuticals, and feedstock chemicals, would allow unparalleled access to complex ethers, one of the most prevalent features in bioactive molecules (Fig. 1) (1, 2). The poor nucleophilicity of alcohols requires their activation to alkali metal alkoxides as exemplified in the Williamson ether synthesis, the most frequently used method for generating simple aliphatic ethers (3-5). However, this reaction is poorly suited to furnish ethers in sterically hindered and functionally complex settings, in part because the basicity of alkali metal alkoxides promotes undesirable E2 eliminations, racemization, and affords poor control of site-selectivity in polyol settings (6-8). Transition-metal mediated cross-coupling etherification is attractive due to the possibility of activating alcohols mildly at the metal center (9-11). However, C(sp³)-O bond formation proceeding within the sphere of the metal is challenging, and using aliphatic alcohols can afford β-hydride elimination products (10, 11). Allylic C—H functionalization reactions proceeding via π-allyl-Pd metal intermediates represents an attractive alternative approach where functionalization may proceed outside the sphere of the metal (Fig. 1B). However, functionalization proceeding via Pd, Rh, and Ir catalysis with alkali metal alkoxide nucleophiles is also problematic due to competing elimination pathways and a mismatch with the relatively soft π-allyl transition metal electrophile (12-14). In order to effect functionalization, copper, tin, boron, and zinc alkoxides have been used with limited preparative utility (12, 13, 15, 16).

Fig. 1. Reaction design and development.

A. A general cross-coupling of alcohols and olefins for a general and selective etherification reaction. B. Previously observed counterion positioning in π-allylPd complexes (27). Chiral phosphoric acid (R*) engages in dual ion-pairing/H-bonding (30, 31). C. Evaluating the effects of DFT guided SOX ligand design and phosphorus oxoacids on the yield and site-selectivities (L = linear, B = branched) in fragment coupling of an alcohol with an unactivated terminal olefin. D. Crystal structures of (π-allyl)Pd BF₄- transSOX and -DiMeSOX complexes 1 and 2 and DFT calculated structures of cisSOX complex 3 and 4 with accompanying electrostatic potential surfaces of the DFT optimized complexes 1-4. The cisSOX complexes 3 and 4 uniquely have a sterically open coordination site and localization of positive charge proximal to the reactive π-allylPd terminus. E. ¹⁹F – ¹H HOESY NMR of π-allyl/SOX BF₄ in toluene-d⁸ complexes supporting ligand controlled counterion positioning. Yields are reported as an average of 2 runs. 2,5 DMBQ = 2,5 Dimethyl Benzoquinone; DPPA = Diphenyl phosphoric acid; DBP = Dibutyl phosphoric acid; DEHPA = Di(2-ethylhexyl) phosphoric acid. *Linear:Branched (L:B) ratios determined by crude GC. †see SM for procedure for DFT calculations. ‡The electrostatic potential is indicated on a surface of constant electron density of 0.0004 e/ų.

Using unactivated alcohols directly could mitigate many of these problems. Mechanistically distinct etherifications have emerged using unactivated alcohols with activated cationic electrophiles such as S_N1 substitution of stabilized carbocations (17-20), reductive etherification via oxonium intermediates (21), and intermolecular Pd and Rh catalyzed allylic C—H etherifications proceeding via cationic π-allyl-metal intermediates (22, 23). Despite proceeding via highly activated cationic intermediates, superstoichiometric quantities of the alcohol are needed to furnish ethers. For example, in intermolecular allylic C—H etherifications, 5 to 11 equivalents of aliphatic alcohols are coupled with activated aromatic olefins to furnish simple styrenyl products (22, 23). The requirement for excess alcohol and specialized olefins renders these methods unsuitable for cross-coupling etherification where using fragment coupling stoichiometries of complex, diverse substrates is needed.

We hypothesized that increasing the rate of the bimolecular C(sp³)-O bond formation step would be the fundamental challenge in developing a general cross-coupling etherification reaction with alcohols. Enzymes often use proximity and orientation effects to achieve rate accelerations in analogous bimolecular reactions: drawing two reactants out of a dilute solution and holding them close to each other in an orientation suitable for reactivity (24). Substrate enforced proximity has been demonstrated in intramolecular cases; however, this strategy is limited to furnishing cyclic ethers (Fig. 1B) (25). We questioned if tuning the ligand environment of a cationic transition metal catalyst to directly position a H-bond acceptor (HBA) counteranion at the reactive site could accelerate the biomolecular functionalization rate with unactivated alcohols. Although not previously shown to impact catalysis, cationic π-allyl-Pd complexes have demonstrated that ligands can affect the specific positioning of non-coordinating anions (for example BF_4, PF_6,CF₃SO₃), (Fig. 1B, 26-28). Moreover, asymmetric phosphoric acid-catalyzed reactions are well-precented to engage in dual ion-pairing/H-bonding activation modes, primarily to effect ordered transition states (Fig. 1B, 29-34). Modular sulfoxide-oxazoline ligand (SOX)/Pd(OAc)₂ catalysis is an ideal platform to explore these orientation and proximity effects in catalysis due to the demonstrated ability to tune allylic C—H functionalizations through electronic and steric ligand modifications and achiral phosphorus oxoacids (25, 35-37). Aided by x-ray crystallographic and computational analysis, as well as ¹⁹F-¹H HOESY experiments, we developed a cis-SOX ligand L4 that creates a steric and electronic environment that directs ion-pairing of the phosphate next to the reactive terminus of the cationic π-allylPd electrophile (Fig. 1B). We additionally show through spectroscopic and kinetic analysis that changes made to the phosphate counterion increase its H-bond acceptor properties and accelerate bimolecular reactivity. Using this strategy to increase proximity and orientation effects in transition metal catalysis, a general cross-coupling etherification reaction of terminal olefins with 1°, 2°, and 3° native alcohols emerges. The etherification reaction shows excellent scope and compatibility with sterically hindered alcohols, functionality prone to substitution, elimination, and epimerization, as well as site-selectivity in polyol settings.

Catalyst development.

We evaluated the reactivity of 3-phenyl-1-propanol with unactivated olefin allyl cyclohexane under previously reported Pd(OAc)₂/trans-SOX L1 catalytic conditions for intramolecular etherification and found no reactivity in the absence of acid (Fig. 1C, entry 1) (25). Introduction of diphenylphosphinic acid (DPPA), a phosphorous oxoacid that has been previously used for Coulombic activation of cationic π-allylPd(SOX) intermediates, afforded modest reactivity with good site-selectivity (20:1 L:B) indicating that functionalization was occurring predominantly at the terminus (linear, L) of the π-allyl (entry 2). Switching to DiMe-SOX L2 afforded a small increase in yield and slightly diminished site-selectivity (entry 3).

Palladium acetate [Pd(OAc)₂] complexes with trans-SOX, cis-SOX, and DiMe-SOX ligands have been shown to promote allylic C—H oxidations, alkylations, and aminations with high levels of selectivity including asymmetric induction (25, 36, 37). Within this platform, we aimed to develop a SOX ligand that would direct the phosphate counterion to associate near the LUMO at the terminus of the π-allyl-Pd intermediate. To aid our design, we synthesized π-allyl-Pd SOX complexes that represent the three known SOX ligand geometries: (π-allyl)Pd/trans-SOX-L1 complex 1, (π-allyl)Pd/DiMe-SOX L5 complex 2, and (π-allyl)Pd/cis-SOX L3 complex 3 (Fig. 1D). Crystals suitable for x-ray diffraction were obtained for complex 1 and 2 and enabled the first x-ray crystallographic analysis of (π-allyl)Pd/SOX intermediates. Study of (π-allyl)Pd/trans-SOX L1 complex 1 and (π-allyl)Pd/DiMe-SOX L5 complex 2 shows positioning the site of functionalization (allylic terminus) next to the phenyl or the gem-dimethyl groups of the oxazoline, sterically blocking phosphate coordination. Alternatively, DFT evaluation of the (π-allyl)Pd/cis-SOX L3 complex 3 and (π-allyl)Pd/cis-SOX L4 complex 4 based on a crystal structure of Pd(OAc)₂/cis-SOX L3 and PdCl₂/cis-SOX L4 (Fig. S56-S61), uniquely shows an open quadrant III near the allylic terminus that may accommodate the phosphate.

Electrostatic potential maps of the (π-allyl)Pd/SOX complexes 3 and 4 additionally reveal localization of positive charge (blue) at the sterically open quadrant III of the cis-SOX complexes 3 and 4 adjacent to the reactive LUMO of the π-allyl terminus (Fig. 1D) (38). In contrast, in (π-allyl)Pd/trans-SOX L1 complex 1 and (π-allyl)Pd/DiMe-SOX L5 complex 2, concentration of positive charge occurs at quadrants I and II that are orthogonal to the LUMO of the allyl terminus. A ¹⁹F-¹H HOESY study in toluene-d⁸ of both cis- and trans-SOX π-allylPd(II) BF₄ complexes (Fig. 1E, complex 1, complex 4) supports this electrostatic analysis by showing a differential preference for positioning of the anion with respect to the geometry of the SOX ligand. The cis-SOX ligand positions the anion close to localization of positive charge in quadrant III, whereas the trans-SOX ligand complex exhibits contacts with the anion in quadrants I and II. Consistent with this analysis, evaluation of known Pd(II)/cis-SOX L3, furnished allylic ether 1 with an improvement in yield (Fig. 1C, entries 2-3 versus entry 4). Incorporation of electron-donating groups on the aryl backbone and the oxazoline L4 further disfavor positioning of the anion in remote quadrants I and II by introducing concentration of negative charge (red) in those areas and provided a significant increase in the yield from 22% to 64% (entries 2, 5). Notably, the switch in ligand geometry to cis-SOX also leads to an increase in site-selectivity more strongly favoring functionalization at the linear position of the π-allyl terminus where the anion is positioned (entry 2, 20:1 and entry 5, 50:1 L:B). Analogous trends are observed with the linear hydrocarbon decene where trans-SOX L1 affords a modest yield and no site-selectivity (28%, 1.1:1 L:B, SI) whereas cis-SOX L4 affords ether products in 61% yield with an improved 2.4:1 L:B selectivity (Table S18).

Phosphorus oxoacids have been invoked to increase reactivity in Pd(OAc)₂/SOX allylic C—H functionalizations via ionization of π-allylPd(SOX)(OAc) electrophiles (25, 35, 37). Intermolecular etherification precedent using cationic intermediates suggested to us that Coulombic activation alone would not be sufficient to promote reactivity of unactivated alcohols under the desired cross-coupling conditions (17-20). We therefore explored electronic and steric modifications of the phosphorus oxoacid to render it both less coordinating and a better HBA. Switching to more acidic, bulky phosphoric acid would increase both its capacity for Coulombic activation of the π-allyl-Pd intermediate and ability to act as a HBA to the alcohol by increasing its electronegativity. Whereas dibutylphosphoric acid (DBP, pKa ~ 1.5) afforded a modest increase in yield, replacing the butyl groups with more sterically demanding 2-ethylhexyl groups [di-(2-ethylhexyl) phosphoric acid (DEHPA)], effected an increase in reactivity (Fig. 1C, entry 6, 7). Evaluation of trans-SOX L1 or DiMe-SOX L2 with DEHPA did not afford substantial improvements in yields or selectivities (entries 8, 9). Trifluoroacetic acid (TFA), a strong ionizing acid that is a poor HBA, is effective in promoting reactivity for intramolecular etherifications where H-bonding is not needed for increasing the proximity of the alcohol (61% yield, Fig. S63, vide infra). Consistent with the important role of phosphate acting as a HBA to increase the alcohol’s effective concentration, use of TFA affords trace yields of ether 1 with both cis-SOX L4 and trans-SOX L1 catalysts (entry 10, 11). In support of differential counterion positioning impacting site-selectivity between the cis-SOX and trans-SOX ligand frameworks, when moving from DEHPA to TFA cis-SOX L4 shows more significant diminishment in L:B ratios (43:1 to 19:1, entries 7, 10) than trans-SOX L1 (15:1 to 11:1, entries 8, 11). Under optimized reaction conditions with cis-SOX L4, omission of acid additives affords no reactivity (entry 12). Optimal catalyst cis-SOX L4/Pd(OAc)₂ loadings may be reduced to 5 mol% with no diminishment in yield or selectivity (entry 13).

Scope exploration.

We evaluated the generality of this cross-coupling allylic C—H etherification across a range of 1°, 2°, and 3° alcohols (Figs. 2, 3). Electron-poor, -neutral and -rich benzylic 1° alcohols furnished allylic ethers in preparative yields, suggesting small changes in nucleophilicity of aliphatic alcohols does not significantly impact reactivity (2-5, Fig. 2). Benzylic alcohols bearing both electrophilic (e.g. epoxides) and latent nucleophilic functionality (boronic esters, indoles) are effective nucleophiles affording allylic ether products (6-10). This reactivity contrasts with that of previous reports in which Pd(II) catalysis with benzylic alcohols under basic conditions affords oxidized aldehyde products (39). The ability to transform readily accessible chiral allylated compounds, like common mannose intermediate 11, to stereochemically defined benzyl-protected E-allylic alcohols provides a streamlined route to polyol precursors (5 steps versus 1 step, 12,13) (40). Mono-protected 1,2 and 1,3 amino alcohols and diols undergo etherification with broad olefin scope (14-18). Mono-silylated ethylene glycol was coupled with a chiral olefin bearing differentiated 1,2-diols as well as an O-alkylated hydroxy-dihydroquinolinone fragment found in antipsychotic drug aripiprazole to furnish 16 and 17 in preparative yields. Primary alkyl halides, typical electrophiles for the Williamson ether synthesis, are unreactive in the alcohol nucleophile and olefin electrophile under these acidic, oxidative C—H etherification conditions (19-21, 23,24) (6). Remarkably, even primary alkyl bromides that are highly prone to substitution and elimination pathways in etherifications, are well-tolerated under these conditions (19, 21). Installation of such lynchpins enables rapid access to drugs and their derivatives: allylic C—H etherification using bromopropanol furnishes 21 in 59% yield that can be readily converted to anti-narcoleptic drug pitolisant 22, or further diversified with other nucleophiles. Olefins bearing aryl aldehydes, reported to act as electrophiles under reductive etherification conditions (21), can be coupled to halogenated aliphatic alcohol nucleophiles to furnish bi-functional building blocks like 24. Remarkably, phenols that react preferentially to aliphatic alcohols in traditional etherification and allylic substitutions may be present in an unprotected form under these non-basic conditions (25, 26) (41).

Fig. 2. Cross-coupling of primary (1°) alcohol and olefins.

Standard conditions: 1.5 equiv. alcohol, 1.0 equiv. olefin, 1.1 equiv. 2,5 DMBQ, 0.1 equiv. Pd(OAc)₂, 0.1 equiv. cis-SOX L4, 0.2 equiv. DEHPA, 1.0 M toluene, 45 °C, ambient atmosphere. *1.2 equiv. alcohol. †1.0 M dioxane to improve substrate solubility. ‡24 hrs. Yields are reported as an average of 3 runs.

Fig. 3. Cross-coupling of secondary (2°) and tertiary (3°) alcohols and olefins.

Acyclic and cyclic 2° alcohols display analogous scope to 1° alcohols with respect to functional group and steric tolerance. Standard conditions used unless otherwise noted. *1.2 equiv. alcohol. Yields are reported as an average of 3 runs.

The allylic C–H etherification shows noteworthy site-selectivity for allylic C—H bonds of ubiquitous terminal olefins (Fig. 2). E/Z- disubstituted, tri-substituted, and styrenyl internal olefins can be present on the alcohol nucleophile in selective couplings with terminal olefins to furnish allylic ethers (27-30). Allylic alcohols, which have been used as electrophiles in Ir- and Pd-catalyzed allylic substitutions under acidic conditions (34, 42), react as nucleophiles under these allylic C—H etherification conditions (30-32). Notably, the allylic alcohol of myrtenol can be coupled to a 1,7-diene to afford triene (32) in useful yields. Interestingly, no significant reactivity difference is noted between allylic, homoallylic, and saturated alcohols within the same bicyclic monoterpene core (31, 33, 34), supporting that this reaction is not strongly impacted by the electronics of the aliphatic alcohol (vide supra, 2-5).

There is a paucity of neopentyl substitution in both classical and modern methods of ether formation, due to the steric hinderance of the quaternary center that renders nucleophilic substitution sluggish and promotes elimination byproducts (6). 3-phenylpropanol and the 2,2-dimethylated analogue proceed with comparable reactivity in allylic etherification (1 versus 35, Fig. 1C, Fig. 2). Primary (1°) alcohols with β-quaternary centers such as methylcyclohexyl- and adamantanyl-methanol are effective nucleophiles in the allylic C—H etherification with unactivated olefins (36, 37) and an olefin bearing a fragment of the broad spectrum antibiotic tedizolid (38). Cross-coupling etherifications can be effected between such alcohols with terminal olefins bearing homoallylic quaternary and tetrasubstituted centers to afford highly congested allylic ethers (39, 40). Primary alcohols with β-tertiary substitution, abundant and often bearing stereogenic centers like those found in naproxen derivatives, afford allylic ethers in preparative yields (41-52). In contrast to highly acidic or basic etherifications, labile α-methyl carbonyl stereocenters are preserved and afford chiral products (46-48) (7). Underscoring the high functional group compatibility of this method, alcohols with β-nitrogen and oxygen stereogenic centers, such as N-Boc amines, unprotected γ-lactams, a Garner’s aldehyde derivative and a triol derivative, afford preparative yields of chiral ethers (49-52). Evaluation of heterocycle scope diversity showed that both the alcohol and olefin coupling partners tolerate many of the top nitrogen and oxygen heterocycles in FDA approved drugs: oxazolidinone (53), oxetane (54), azetidine (55), pyran (56), piperidine (57, 63, 64), morpholine (58), piperazine (59, 60), pyrrolidine (61), indole (62) and β-lactam (65) (43, 44).

The increased steric hindrance in ethers derived from secondary alcohols (2°) makes them challenging targets. Acyclic benzylic 2° alcohols show the same tolerance for electronic aryl substituents as their 1° counterparts (Fig. 3, 66-70) and tolerate substantial steric bulk in one substituent (71-76). Chiral secondary alcohols, building blocks readily accessible as benzylic alcohols, β-ketoesters, 1,2 amino-alcohols and 1,2 diols, undergo cross-coupling etherification with high enantiospecificity (es) affording chiral products (67, 68, 77-80). Cyclic 2° alcohols, prevalent in natural product settings, undergo oxidative etherification in preparative yields, even with complex olefin coupling partners (81-88). Cyclohexenol as well as medicinally relevant pyrrolidinol, piperidinol and oxetanol are competent nucleophiles despite their attenuated nucleophilicity due to proximal electronegative sp² carbons and heteroatoms (89-96). The capacity to use tetrahydropyranols as nucleophiles affords opportunities to derivatize key core structures in sugars (96).

The counterion positioning strategy directs functionalization from an unhindered quadrant of the catalyst and results in remarkable tolerance for steric hindrance in this etherification. Conformationally rigid t-butyl substituted cyclohexanols show no preference in yield between the more sterically accessible equatorial hydroxyl versus the axial hydroxyl (Fig. 3, 97 vs 98). Additionally trans-1,2-cyclohexanols bearing adjacent methyl, N-Boc protected amines and isopropyl groups are cross-coupled to unactivated olefins in preparative yields (99-101). Sterically congested [3.2.1]-bridged bicyclic alkaloid nortropine and bicyclic terpene alcohols (−)-borneol and its isomer (+)-fenchol, bearing a 2° alcohol flanked with quaternary centers, were all effective nucleophiles for the oxidative etherification furnishing products 102, 103, and 104.

Tertiary (3°) ethers are generally formed by coupling of highly substituted electrophiles capable of stabilizing a positive charge (for example 3° carbocations or oxoniums) with less substituted nucleophiles (i.e. 1° or 2° alcohols or organometallic reagents) and use of sterically hindered tertiary alcohol is rare (17, 45). A range of cyclic and acyclic tertiary alcohols can be cross-coupled with allyl phenyl sulfone in fragment coupling stoichiometries to furnish vinyl sulfone products that could undergo further diversification via cross-couplings or find medicinal use as electrophilic warheads (Fig. 3, 105-110) (46, 47).

Alcohols are the most frequently occurring functional group in natural products (1). We therefore set out to investigate the capacity of natural product derived alcohols to act as nucleophiles in this reaction. Terpenoids are a large and structurally diverse class of natural products that are rich in alcohols whose position and derivatization can influence their biological function. Oxidative etherification can utilize topologically complex 1° and 3° alcohols derived from sesquiterpene natural products (+)-longifolene and (+)-cedrol to furnish ethers in preparative yields (Fig. 4A, 111, 112). The 2° and 3° alcohols of triterpenoids cholesterol and a 3β-allopregnanolone derivative can be cross-coupled to terminal olefins with nitrogen, epoxide, and sulfone functionality to afford allylic ether derivatives (113-115). The sterically congested 2^° alcohol of an enoxolone derivative, which additionally contains an α,β-unsaturated olefin, was cross-coupled with diethyl allylphosphonate to afford allylic ether 116. The 1° and 2° alcohols on pyranose and furanose derivatives, the top two oxygen heterocycles found in U.S. F.D.A. approved drugs, are readily allylated via oxidative etherification (117, 118) (44). Underscoring the high steric tolerance of this reaction, an allylated glucose derivative can be effectively coupled to the 2° alcohol of a tropane derivative to furnish allylic ether 119.

Fig. 4. Cross-coupling of natural product alcohols and polyols.

A. Cross-coupling etherification of complex 1°, 2°, and 3° alcohols found in terpene, steroid, sugar, and alkaloid natural product derivatives. B. Site-selectivity for polyols containing 1° and 2° alcohols, 1° and 3° alcohols and among 2° alcohols in different steric environments. In all cases, preparative yields of mono-alkylated ethers were observed at the least sterically hindered alcohol. Mass balance of alcohol is shown in []= recovered starting alcohol + ether product / total starting alcohol x 100. C. Site-selectivity for phenol natural product derivatives containing 1° and 2° alcohol nucleophiles as well as terminal olefin electrophiles. In all cases no phenol functionalization was observed. Reactions run under standard conditions unless otherwise specified. Yields are reported as the average of 3 runs. *1.2 equiv. alcohol. †0.5 M Toluene. ‡1.0 equiv. alcohol, 1.2 equiv. olefin. §1.0 M dioxane used for substrate solubility. ¶under 1 atm O₂ balloon. #0.33M dioxane. **1.0 M CHCl₃ †† six steps from cholic acid methyl ester: global protection, selective deprotection, allylation, oxidation, reduction, deprotection (8). ‡‡ three steps from estradiol: phenol protection, alkylation, deprotection (52).

Derivatization of natural products and other biologically active compounds with multiple hydroxyl groups is generally dealt with by masking (or protecting) most of the hydroxyls followed by derivatization at the one desired (48). Given that this reaction demonstrates high reactivity with 1°, 2° and 3° alcohols, we questioned if site-selective etherification is possible in polyol settings (Fig. 4B). Evaluation of chloramphenicol bearing a primary alcohol and a more sterically hindered and electron deficient 2° benzylic alcohol showed high selectivity for functionalization at the 1° alcohol: ether 120 was formed in useful yields with significant amounts of recovered alcohol (93% = alcohol mass balance = (ether + rsm)/(starting alcohol)). A sugar derivative with a 1° and an allylic 2° alcohol afforded the 1° alcohol derived ether in preparative yields and excellent mass balance supporting high site-selectivity (121). Evaluation of ambroxdiol and a labdane type diol with 3° alcohol motifs showed selectivity for 1° aliphatic and allylic alcohol etherifications across a range of terminal olefins (122-124). Crystallographic analysis of a steviol derived diol (125) illuminated the relatively exposed adamantane-like 3° alcohol and hindered 1° alcohol adjacent to a quaternary center (Fig. S2). Notably, etherification was only observed at the 1° alcohol.

We next evaluated the cholic acid class of steroids bearing multiple 2° hydroxyl groups in distinct steric environments (Fig. 4B). In both chenodeoxycholic acid ester and deoxycholic acid ester bearing C3 and C7 or C12 (respectively) 2° alcohols, oxidative etherification was only observed at the more sterically exposed C3 alcohol to afford 126 and 127 respectively (Fig. 4B). Whereas site-selective acylation and phosphorylation of alcohols are well-precedented (48-50), such alkylations are rare (51). In cholic acid methyl ester bearing C3, C7 and C12 2° alcohols, high site selectivity and preparative yields afford a streamlined route to allylated C3 derivatives bearing nitrogen and dense oxygen functionality (1 step versus 6 steps, 128, 129). Previous routes using 3-allylbromide required C7- and C12-hydroxyl protection to achieve site-selective allylation, followed by derivatization and deprotection to afford oxygen functionalized ether products (8).

In base mediated etherifications (e.g. Tsuji-Trost allylation or Williamson etherification) of aliphatic alcohols with phenol functionality, the more acidic phenol functionality is protected prior to etherification. Under the conditions of this allylic C—H etherification, more nucleophilic aliphatic alcohols react preferentially to phenols, even in complex natural product settings: vitamin E derivative trolox bearing a bis-ortho-substituted phenol and neopentyl 1° alcohol selectively underwent cross-coupling through the 1° alcohol to afford allylic ethers 130 and 131 (Fig. 4C). Consistent with this chemoselectivity being heavily influenced primarily by electronics, etherification of 17β-estradiol occurs selectively at the hindered 2° alcohol versus the unhindered A-ring phenol (132). This is in contrast with reported routes for alcohol etherification on estradiol which require phenol protection/deprotection sequences (52). Phenols are additionally well-tolerated in unprotected form on the olefin: an allylated 17β-estradiol derivative is cross-coupled with the 2° alcohol of trimethylcyclohexanol to afford ether 133 in preparative yield.

Mechanism:

We hypothesized that the barrier to alcohol functionalization in palladium-catalyzed etherification can be overcome via SOX ligand directed orientation of a phosphate anion that engages in H-bonding with the alcohol to approximate and orient it at the reactive π-allyl terminus.

We first investigated if alcohol functionalization of the π-allylPd intermediate plays a major role in the rate of allylic C—H etherification. Initial rate studies on parallel reactions resulted in a primary kinetic isotope effect (KIE) of 1.65 ± 0.04 (Fig. 5A, 134) that is significantly lower than the intramolecular competition KIE (4.10 ± 0.08) (135). These studies indicate that while this reaction is proceeding via allylic C—H cleavage, it is likely not the rate-determining step (RDS).

Fig. 5. Mechanistic Studies.

A. Intra- and intermolecular kinetic isotope effect (KIE) measurements indicating that etherification proceeds via C—H cleavage but that is not the rate-determining step (RDS). B. The impact of ligand geometry on rate of functionalization, with faster rates for the cis-SOX ligand geometry. The correlation between k_rel for stoichiometric reaction (top) and that of the catalytic reaction (bottom) supports functionalization is the RDS. C. The effect of counteranion on rate of functionalization, suggesting that a weakly coordinating counteranion that is capable of H-bond activation promotes alcohol functionalization. D. Differences (Δ) in proton-coupled 13C NMR spectra (¹J_C1-Ha) of 3-phenyl propanol with π-allylPd(cis-SOX) cationic complexes having different counteranions 137b-141b, supporting H-bonding between the DEHPA counteranion and the alcohol nucleophile. It is estimated that a ~0.2 Hz decrease in ¹J_C-H correlates to 1 kJ of H-bond strength (56). Δ ¹J_C1-Ha and the rate of functionalization as a function of solvent dielectric constant (ε) and H-bond accepting ability (β) support that both phosphate tight ion-pairing with cationic Pd and H-bond activation of the alcohol promote functionalization. E. The inverse rate trends between the intra- and intermolecular reactions when evaluating ligand geometry and counterion identity: in the intramolecular case, selective HBA counterion positioning interferes with functionalization. *The functionalization rate using π-allylPd(trans-SOX-L1)(OAc) was too slow to measure at the same concentration. †¹J_C1—Ha data obtained by using toluene-d⁸. Values using CDCl₃ as a solvent shown in parentheses. ‡ε and β-parameter values taken from the literature (59, 60). §Δ¹J_C1—Ha values obtained in corresponding solvent (complex 140a + alcohol)_solvent – (free alcohol)_solvent.

To interrogate if ligand geometry and phosphoric acid additives promote etherification by increasing the rate of alcohol functionalization, stoichiometric π-allylPd(SOX)(DEHPA) complexes with varying SOX ligand geometries ligands L1-L4 were synthesized and reacted with 3-phenyl-1-propanol under mock catalytic conditions (Fig. 5B). Consistent with ligand geometry influencing the rate of functionalization (Fig. 1D, complex 1-4), π-allylPd(II) with cis-SOX L3 and cis-SOX L4 furnished 1 with an approximately 2.5- and 4.7- fold increase in functionalization rate (respectively) relative to trans-SOX L1. These rates closely mirror the initial rates of the analogous catalytic reactions further supporting that functionalization is rate-determining.

Stoichiometric π-allylPd(cis-SOX-L4) and (trans-SOX-L1) complexes with counterions of varying coordination and H-bond acceptor (HBA) abilities were also evaluated kinetically (Fig. 5C). ¹H NMR analysis of a π-allylPd(cis-SOX-L4)(OAc) complex shows rapid formation of the π-allylPd(SOX-L4)(DEHPA) complex upon addition of one equivalent of the phosphoric acid (Fig. S4). The decreased Lewis basicity and weaker coordinating ability of bulky phosphate counteranion relative to phosphinate and acetate would lead to a less electrostatically stabilized ion-pair with the π-allylPd(SOX) cations, resulting in more rapid functionalization (53, 54). Consistent with alcohol functionalization being promoted with less-coordinating counteranions, π-allylPd(cis-SOX-L4)(DEHPA) 140 underwent functionalization 2 times faster than the corresponding phosphinate complex 139 and 34-fold times faster than the corresponding acetate complex 137. In addition to its electrostatic role, we hypothesized that the phosphate counteranion serves as a H-bond acceptor to the alcohol nucleophile, increasing its effective molarity at the π-allyl-Pd complex. Supporting this, in π-allylPd(cis-SOX-L4)(TFA) 138 and π-allylPd(trans-SOX-L1)(TFA) 138a, with trifluoroacetate as a non-coordinating counterion with poor hydrogen bond acceptor capacity, a 3.1 and 2.7 fold lower rate was observed relative to the corresponding DEHPA complexes 140 and 140a (Fig. 5C) (54, 55). In moving from the weaker HBA counterion DPPA to TFA, the rates of functionalization are 1.5 fold slower in the π-allylPd(cis-SOX-L4) complex (138, 139) but unchanged in the π-allylPd(trans-SOX-L4) complex (138a, 139a).

To provide direct evidence for the proposed hydrogen bonding interaction we performed NMR studies at room temperature, where the rate of functionalization is slow (Table S12), and evaluated the proton-coupled ¹³C NMR spectra of 3-phenylpropanol in the presence of π-allylPd(II)/cis-SOX L4 with varying counterions (Fig. 5D). Hydrogen bonding increases the electron density in the O—H bonding orbital, thereby increasing its hyperconjugation into the σ* orbitals of C—H bonds α to the hydroxyl group. The ensuing lengthening of the C—H bond can be correlated to a decrease in the ¹³C-¹H spin-spin coupling (¹J_C-H) (56, 57). Consistent with a hydrogen bonding interaction between π-allylPd(II) cis-SOX L4/DEHPA 140b and alcohol nucleophile, addition of an equimolar amount of 3-phenyl-1-propanol to complex 140 in toluene d⁸ afforded a 0.99 Hz decrease in ¹J_C-H (140.19 Hz to 139.20 Hz). The decrease in ¹J_C-H is greater for DEHPA relative to the DPPA complex 139b or the acetate complex 137b (0.99 Hz versus 0.59 Hz and 0.48 Hz, respectively) signifying greater hydrogen bond acceptor ability of DEHPA relative to these more coordinating counterions. In the TFA and BF₄ complexes (138b and 141b, respectively) having poor H-bond acceptor abilities, no change in ¹J_C-H was observed (140.19 Hz versus 140.18 Hz and 140.24 Hz). Under catalytic conditions, the benzoquinone oxidant and its dihydroquinone reduction product may act as competitive H-bond acceptors (HBA) and donors, respectively, with the phosphate counterion and alcohol nucleophile. Mixing 2,5-dimethylbenzoquinone (2,5-DMBQ) with 3-phenyl-1-propanol resulted in no change in ¹J_C-H, indicating that the quinone carbonyls are not competitive HBAs in this system (Fig. S21-26). Addition of an equimolar amount of 2,5-dimethyldihydroquinone and 3-phenyl-1-propanol to allylPd(II) cis-SOX L4/DEHPA 140b resulted in a diminishment in the magnitude of the decrease in ¹J_C-H (0.99 Hz versus 0.43 Hz, Fig. S27-29). While phenol can compete with alcohol for H-bonding with phosphate, it does not fully interrupt this interaction.

The ability of the DEPHA counteranion to act as a HBA and orient the alcohol towards functionalization is dependent upon a tight-ion pairing interaction with the cationic π-allylPd cis-SOX complex (28, 32, 58). We interrogated the effect of solvent dielectric constant (ε) on the rate of alcohol functionalization with π-allylPd(II) cis-SOX L4/DEHPA 140 (Fig. 5D). An inverse correlation was observed between solvent dielectric constant and rate of functionalization, with toluene showing a ca. 4-fold increase in functionalization rate relative to dichloromethane (DCM). Dioxane, a low dielectric solvent possessing a high Kamlett-Taft polarity β parameter (β = hydrogen bond acceptor ability) afforded ether 1 with slower rates than the other solvents (59, 60). Consistent with the need for solvent to support phosphate/alcohol H-bonding, the change in ¹J_C-H, was strongest when alcohol was added to complex 140b in solvents with low β parameter values, and smallest in dioxane having the largest β parameter value. These results are consistent with the ability to tolerate free phenols and carbonyls in both coupling partners, but not strong H-bond acceptor functionality like basic amines (See SM, Fig. S70). Collectively, the data support that the phosphate counteranion engages in tight ion-pairing interactions with the cationic π-allylPd(SOX) intermediate and H-bonding interactions with the alcohol nucleophile.

In addition to enhanced steric access, our DFT and ¹⁹F-¹H HOESY experiments suggest that the observed ca. 5 to 6 fold rate accelerations in π-allylPd(cis-SOX-L4) relative to π-allylPd(trans-SOX-L1) phosphorus oxoacids are due to counteranion positioning that orients the alcohol for the reaction at the allyl terminus (Figure 5C). To further investigate this, we tethered the alcohol to the allyl moiety to eliminate the dependence of reactivity on catalyst/counterion-controlled alcohol proximity and orientation effects (Figure 5E). Interrogation of the effect of ligand geometry and counteranion identity on intramolecular reaction rates showed relatively small but statistically significant effects (see SM Table S17) with inverse trends to those observed for intermolecular etherification. Consistent with previous reports of lower yields in intramolecular etherification using cis-SOX versus trans-SOX/Pd(OAc)₂/DPPA catalysis (25), the rates for intramolecular etherification are ca. 1.3 to 1.5-fold slower with cis-SOX L4 than trans-SOX L1 using DPPA and DEHPA, respectively (entries 3, 6, p < 0.05; entries 4, 7, p < 0.01). With trans-SOX L1 where DFT/HOESY experiments show the counteranion is positioned orthogonal to the allyl, the DEHPA counterion shows a 1.3 fold increase in rate relative to TFA (entries 2, 4, p < 0.05). In contrast, with cis-SOX L4 an inverse trend was observed with a ca. 1.4-fold decrease in reaction rate in going from TFA to DEHPA (entries 5, 7, p < 0.05). These results can be rationalized by the cis-SOX ligand positioning the phosphate next to the allyl terminus. H-bonding to the tethered alcohol may now interfere with the geometrically favored 6-membered transition state. Collectively, these results are inconsistent with the cis-SOX ligand only generating a sterically more accessible allyl and the counterion acting as an unselectively positioned hydrogen bond acceptor. Rather, in SOX/Pd catalysis, the ligand geometry and electronics can position counterions in specific locations relative to π-allyl-Pd intermediates and this can influence reactivity.

Ligand directed positioning of HBA counteranion in cationic transition metal catalysis can promote challenging bimolecular reactivity. The etherification method achieved via this strategy is notable with respect to selectivity, generality and scope: fragment coupling amounts of complex alcohols and olefins react under uniform, robust conditions that use the same reagents, and are open to air and moisture to furnish linear E-allylic ethers.