Epoxidation with Peroxyacids

Peroxyacid epoxidation is a workhorse step in pharmaceutical process chemistry: many steroid, terpenoid, and macrolide syntheses use mCPBA (or the cheaper, industrially preferred peracetic acid on large scale) to install an epoxide that is subsequently opened regio- and stereoselectively to set up multiple contiguous stereocenters in a single downstream step — precisely the strategic logic that motivates Anti Dihydroxylation via Epoxide Opening as a synthetic disconnection. Epoxidized natural rubber and epoxidized soybean/vegetable oils, produced industrially by peracid or peracetic-acid-mediated epoxidation of the alkene-rich polymer/triglyceride backbone, are major commodity materials used as plasticizers, PVC stabilizers, and precursors to polyols for polyurethane foams. In fragrance and flavor chemistry, directed epoxidation of terpene alcohols like geraniol (Example 2) is a key step toward rose-oxide and related aroma compounds. In biology, endogenous epoxidation (catalyzed enzymatically rather than by free peracids, via cytochrome P450 monooxygenases using a very differently-mechanism iron-oxo species) converts arachidonic acid into epoxyeicosatrienoic acids (EETs), signaling lipids with vasodilatory and anti-inflammatory roles — a striking case where nature achieves a related net transformation (alkene to epoxide) via an entirely different, metal-mediated radical rebound mechanism, offering a useful mechanistic contrast to the purely organic, metal-free peracid pathway detailed here.

Ask the tutor: How does the mechanism and stereochemical outcome change when the epoxide is opened under acidic versus basic conditions to give diols?

Peroxyacids look deceptively like carboxylic acids with an extra oxygen wedged into the O–H bond: R–C(=O)–O–O–H instead of R–C(=O)–O–H. That extra oxygen changes everything. In a normal carboxylic acid the carbonyl oxygen and the hydroxyl oxygen are both attached to the same electron-poor carbonyl carbon, and the O–H is simply acidic. In a peroxyacid, the terminal oxygen (the one not attached to carbon) sits at the end of a weak, polarizable O–O σ bond (bond dissociation energy of an O–O bond is only about 140–180 kJ/mol, much weaker than a typical C–O or C–C bond). This terminal oxygen is simultaneously electrophilic — pulled on by the electron-withdrawing carbonyl two atoms away, and set up so that if it were to grab two electrons, the developing negative charge could be relayed intramolecularly onto the adjacent carbonyl oxygen — and superbly positioned to accept a pair of π electrons from an alkene approaching from below or above. The classic geometric picture, first proposed mechanistically by Nikolai Prilezhaev in 1909 and refined mechanistically over the following decades, is the "butterfly" or "spiro" transition state. The alkene approaches the terminal peroxyacid oxygen with its π system aligned face-on (not edge-on) to that oxygen's empty σ orbital character, while the internal O–H hydrogen simultaneously swings over to hydrogen-bond to, and then fully transfer onto, the carbonyl oxygen. Six atoms participate in this concerted array: the two alkene carbons, the two peroxyacid oxygens, the carbonyl carbon, and the transferring hydrogen — although the electron count that actually moves is best tracked as a five-membered, quasi-aromatic closed loop* of arrows (a Hückel-like six-electron pericyclic transition state, isoelectronic in arrow-pushing style with an ene reaction).

Because bond formation to both alkene carbons and O–O bond cleavage happen within one transition state, the transition state geometry directly dictates stereochemical outcome. The alkene must approach with its π face parallel to the O–O axis; whichever face it approaches from is the face that receives the oxygen. There is no rotation about the (former) C=C axis possible mid-mechanism, because the two new C–O bonds and the residual C–C σ bond are all formed/retained simultaneously — this is the electronic root of strict syn addition and full stereospecificity, in sharp contrast to the stepwise cationic mechanisms discussed under Stereochemistry of Electrophilic Addition, where a discrete bromonium or carbocation intermediate can (in principle) be attacked from either face. Substituent effects on rate follow directly from frontier orbital matching. The alkene HOMO (π) donates into the peroxyacid's LUMO, which has large coefficient on the electrophilic, O–O-σ-flavored terminal oxygen. Raising the alkene HOMO energy (more alkyl substitution, more hyperconjugative donation, conjugation with electron-donating groups) speeds the reaction; lowering the peroxyacid LUMO energy (electron-withdrawing substituents on the aroyl ring, as in mCPBA's meta*-chlorine) also speeds the reaction, because it is the alkene HOMO — peroxyacid LUMO gap that controls the transition-state energy.

The rate law for peroxyacid epoxidation is second order overall, first order in each reactant, consistent with a bimolecular, single-transition-state mechanism:

where: $k$ is the second-order rate constant (units M⁻¹s⁻¹), sensitive to both alkene substitution and peroxyacid electrophilicity; $[\text{alkene}]$ and $[\text{peroxyacid}]$ are the instantaneous concentrations of each reagent. The Hammett-type substituent dependence on the peroxyacid's aryl ring follows a linear free-energy relationship,

where: $k_X$ is the rate constant for the peroxyacid bearing aryl substituent $X$, $k_H$ is the rate constant for the parent (unsubstituted) peroxybenzoic acid, $\sigma_X$ is the Hammett substituent constant (positive for electron-withdrawing groups like meta-Cl, negative for electron-donating groups), and $\rho$ is the reaction constant; for peroxyacid epoxidation $\rho$ is small and positive (electron-withdrawing aryl substituents accelerate the reaction by stabilizing developing negative charge on the leaving carboxylate oxygen in the transition state). The relative rate of epoxidation across differently substituted alkenes tracks alkene nucleophilicity in an additive, per-alkyl-group fashion; a useful mnemonic ordering (all relative to ethylene = 1) is

where each additional alkyl group raises the alkene HOMO energy through hyperconjugative and inductive donation, narrowing the HOMO–LUMO gap with the peroxyacid and lowering the transition-state energy.

Example 1 (foundational): epoxidation of (E)-4-methylpent-2-ene with mCPBA — predicting product stereochemistry from first principles. The substrate is CH₃–CH=CH–CH(CH₃)₂ with E (trans) geometry: the methyl on C2 and the isopropyl-bearing C4 substituent point to opposite sides of the double bond. Apply Mechanism 2's logic directly. The alkene has one π face attacked (arbitrarily, since mCPBA is achiral and the substrate, while chiral-capable, here has no other stereocenter, so both faces are attacked with statistically equal likelihood, generating a racemic mixture of epoxide enantiomers). Within any single event, both new C–O bonds form on the same face simultaneously. Because the alkene was trans (methyl and the isopropyl-substituted carbon's H/alkyl groups on opposite sides), the resulting epoxide is the trans-configured epoxide: the methyl group at one epoxide carbon and the isopropyl group at the other epoxide carbon end up on opposite faces of the oxirane ring. The product is racemic — a 1:1 mixture of (2R,3R) and (2S,3S) 2-methyl-3-isopropyloxirane-type numbering (exact R/S labels depend on CIP priority assignment at each center, but the relative configuration, "trans", is fixed by the mechanism regardless of which enantiomer forms). As a check on scope: if the starting alkene had instead been the Z-isomer, the product would be the racemic cis (here non-meso, since the two ends of the alkene bear different substituents — methyl vs. isopropyl — so cis here is still a chiral diastereomer, not a meso compound; meso only arises when the substituent pattern itself has an internal mirror symmetry, as in 2,3-dimethyloxirane from 2-butene).

Example 2 (applied): chemoselectivity in the epoxidation of geraniol (a monoterpene with two distinct trisubstituted-type alkene environments) — using the HOMO-energy/steric argument to predict which double bond reacts first. Geraniol, (2E)-3,7-dimethylocta-2,6-dien-1-ol, contains two alkenes: the C2–C3 alkene, conjugated to (adjacent to) the primary allylic alcohol and flanked by one methyl substituent (trisubstituted, allylic-alcohol-proximal), and the C6–C7 alkene, a simple trisubstituted alkene bearing two methyl groups (the "isopropenyl-like" terminus, part of the isopropylidene unit) with no nearby polar group. Both alkenes are trisubstituted, so a naive nucleophilicity argument alone (more alkyl substitution = faster) does not immediately discriminate them — both have comparable intrinsic HOMO energies from alkyl substitution. The discriminating factor is a directing effect: the free allylic/homoallylic hydroxyl group can hydrogen-bond to the peroxyacid's acidic O–H and carbonyl oxygen, pre-organizing the peroxyacid close to the nearer alkene (the C6–C7 alkene, which is closer in space to the OH through the flexible chain, sitting roughly 4–5 atoms away allowing a favorable hydrogen-bonded macrocyclic-like transition state) and delivering the oxygen intramolecularly-assisted to that face. Experimentally, treatment of geraniol with one equivalent of mCPBA gives predominant epoxidation at the C6–C7 alkene (the alkene distal from the OH in terms of conjugation but proximal in terms of achievable hydrogen-bonded transition-state geometry), giving 6,7-epoxygeraniol selectively, leaving the C2–C3 (allylic, conjugated) alkene untouched under carefully controlled conditions. The lesson: intrinsic nucleophilicity sets a baseline rate, but hydrogen-bonding/directing effects from nearby polar groups can dominate site-selectivity — this is the same principle later exploited deliberately and catalytically in Sharpless asymmetric epoxidation, where the allylic alcohol's oxygen coordinates the titanium–tartrate complex to enforce both site- and enantioselectivity.

Example 3 (exam/competition level): quantitative Hammett analysis — determining ρ for a series of substituted peroxybenzoic acids reacting with cyclohexene. Suppose kinetic data (idealized, illustrative values consistent with the known small-positive ρ for this reaction) are collected for the epoxidation of cyclohexene by three meta-substituted peroxybenzoic acids, all in acetic acid solvent at 25 °C:

Task. Extract $\rho$ for this reaction series and use it to predict $k$ for the meta-methoxy peroxyacid ($\sigma = 0.115$).

Step 1 — linearize. Compute $\log(k_X/k_H)$ for each substituent:

Step 2 — fit ρ (slope of $\log(k_X/k_H)$ vs. $\sigma$). Using the Cl and NO₂ points to estimate the slope directly (a full least-squares fit through all three points, including the origin at $\sigma=0$, $\log(k_H/k_H)=0$, gives essentially the same answer since the data are near-linear):

Cross-checking against the Cl point alone:

Averaging the two slope estimates gives $\rho \approx 0.90$, consistent with literature values for peroxyacid epoxidation (small positive $\rho$, typically reported in the range 0.7–1.0 depending on solvent), confirming that electron-withdrawing aryl substituents accelerate the reaction, but only modestly — the transition state has only partial negative charge buildup on the departing carboxylate oxygen, consistent with the concerted (not fully ionic, stepwise) mechanism.

Step 3 — predict $k$ for meta-OMe ($\sigma = 0.115$, mildly electron-donating relative to H but still positive by convention for meta-OMe since resonance donation is attenuated at the meta position, leaving a net small positive inductive $\sigma$):

Limiting-case check. As $\sigma \to 0$ (an ideal, purely inductively-neutral, non-resonance substituent), $k_X \to k_H$ exactly, recovering the reference rate — consistent with the defining property of the Hammett equation, and confirming the fit is self-consistent at the anchor point.

The reaction bears the name of Nikolai Alexandrovich Prilezhaev, a Russian organic chemist who in 1909 first reported that benzoyl peroxide-derived peracids convert alkenes cleanly into epoxides ("Prilezhaev reaction" or "Prilezhaev epoxidation" in older European literature). His original work used perbenzoic acid; the far more practically convenient and shelf-stable meta-chloroperoxybenzoic acid (mCPBA) — solid, easily weighed, commercially available, and less prone to explosive decomposition than many other peracids — became the reagent of choice only decades later, once industrial and academic laboratories needed a peracid robust enough for routine use on the bench. Mechanistic clarification of the concerted, "butterfly" transition-state picture (as opposed to a stepwise ionic mechanism) emerged through kinetic isotope studies, Hammett substituent analyses (of the type worked through above), and stereochemical retention experiments over the mid-20th century, cementing the modern textbook mechanism.