Oxidation of Secondary Alcohols to Ketones

Secondary alcohol oxidation to ketones is one of the most frequently executed steps in pharmaceutical process chemistry and total synthesis, precisely because of its guaranteed chemoselectivity ceiling — chemists can oxidize a secondary alcohol in a complex polyfunctional molecule with reasonable confidence that the reaction stops exactly at the ketone. Steroid chemistry depends on it constantly: converting the 3β-hydroxyl of a steroid precursor to the corresponding 3-ketone (as in the industrial synthesis of progesterone and testosterone derivatives) is textbook Jones- or PCC-type chemistry, and it is precisely this ketone that later serves as an electrophile for further C–C bond-forming steps (Grignard addition, enolate alkylation) elsewhere in steroid total synthesis. Dess–Martin periodinane and Swern oxidation are default choices in modern academic and industrial total synthesis of natural products and drug candidates whenever a substrate carries acid- or base-sensitive functionality (epimerizable stereocenters, acetals, silyl ethers, alkenes prone to Wacker-type side reactions) that would not survive Jones' strongly acidic aqueous conditions. The paired strategy of protecting a primary alcohol as a robust ether — most commonly a silyl ether such as TBS or TBDPS — while oxidizing an adjacent free secondary alcohol, is one of the most common two-step sequences in synthetic planning, because silyl ethers are completely inert to DMP, Swern, and PCC conditions and can be removed later under mild fluoride-based conditions without disturbing the newly installed ketone. TEMPO (2,2,6,6-tetramethylpiperidine-1-oxyl) with a stoichiometric co-oxidant such as bleach (NaOCl) or BAIB provides a catalytic, environmentally friendlier alternative used at scale in industrial and green-chemistry contexts; it operates through a different initial mechanism (a nitroxyl radical/oxoammonium cycle rather than direct O-activation by a metal or hypervalent center) but converges on the same overall bond-forming logic — a species with an O-based electrophilic center reacts with the alcohol oxygen, and a subsequent step removes the carbinol C–H to deliver the carbonyl, again halting cleanly at the ketone for secondary substrates and failing (as expected) for tertiary ones. The oxidation-state ceiling logic taught here also underlies biochemical alcohol dehydrogenase chemistry: NAD⁺-dependent alcohol dehydrogenases oxidize secondary alcohols (e.g., in steroid and terpenoid metabolism) to ketones via a hydride-transfer mechanism mechanistically distinct in detail (hydride goes directly to NAD⁺ rather than through an O-activated leaving-group intermediate) but identical in outcome and in the same fundamental constraint — the enzyme active site removes exactly the carbinol C–H as a hydride, and a tertiary substrate lacking that hydride is simply not a substrate for these enzymes at all.

Ask the tutor: How does the mechanism change when TEMPO/bleach is used instead of a chromium or sulfonium-based oxidant, and why is it considered "greener"?

Look at the carbinol carbon before and after oxidation. In the alcohol, that carbon is sp³, tetrahedral, bonded to O, H, and two carbon substituents (for a secondary alcohol) or three carbon substituents (for a tertiary alcohol). Oxidation converts the C–O single bond into a C=O double bond. Because carbon can only ever have four bonds, making a second bond to oxygen requires giving up one of the other three bonds — and the bond that is given up is always the one to hydrogen, never a bond to carbon, because C–C bond homolysis/heterolysis is enormously higher in energy than deprotonation at a properly activated center. This is the entire reason tertiary alcohols resist oxidation by every standard alcohol-oxidizing reagent: a tertiary carbinol carbon has three C–C bonds and zero C–H bonds. To make the second C–O bond you would have to break a C–C bond, which none of Cr(VI), DMP, Swern-activated DMSO, or TEMPO/NaOCl are capable of doing under the conditions used for alcohol oxidation. The alcohol may still be activated at oxygen (chromate esters and alkoxysulfonium salts form perfectly well from tertiary alcohols), but the elimination step that actually forms the carbonyl has no C–H to remove, so the activated intermediate simply falls apart back to starting material (or, under forcing acidic/oxidative conditions, undergoes a completely different pathway — acid-catalyzed dehydration to an alkene, or oxidative C–C cleavage under very strong conditions like hot concentrated permanganate, which is a different chapter entirely). Contrast this with a primary alcohol, whose product — an aldehyde — still has one C–H left on the carbonyl carbon, so a second identical oxidation event (hydration to the gem-diol, then a second E2-like elimination) can carry it onward to the carboxylic acid. Secondary alcohols occupy the mechanistic middle ground: one oxidation event is possible and it terminates automatically at the ketone, because the ketone carbon has no C–H bond left to remove. This is why chemists reach for secondary alcohol oxidation as one of the most reliable, "no-over-oxidation-to-worry-about" steps in a synthesis.

Now the actual mechanism. The dominant strategy across chromium(VI), hypervalent-iodine, and activated-sulfoxide oxidants is identical in logic even though the atoms differ: (1) the alcohol oxygen attacks an electrophilic center on the oxidant (Cr, I, or S) to form an O-linked ester/intermediate with a good leaving group built onto that oxidant center; (2) a base — which can be another molecule of the oxidant, a chloride, an amine, or simply the departing leaving group's own conjugate base — removes the α C–H in a six-membered, cyclic, syn-periplanar transition state while the C–O bond of the activated ester breaks and the leaving group departs, generating the C=O π bond in the same concerted step.

Formally, the oxidation of a secondary alcohol is a two-electron, two-proton-equivalent process:

where: R₂CH–OH is the secondary carbinol, R₂C=O the ketone product, and the two electrons/two H⁺ are absorbed by whatever the terminal oxidant is (Cr(VI)→Cr(IV) or Cr(III), the DMSO-derived sulfonium ylide's proton transfers, or I(V)→I(III) for hypervalent iodine reagents). Oxidation state bookkeeping at the carbinol carbon is the single most load-bearing relationship in the whole topic:

where: each bond to a more electronegative atom (O here) counts +1, each bond to H counts −1, and bonds to carbon count 0. Applying this to the three relevant species:

Reading this: a secondary alcohol carbon sits at oxidation level 0; the ketone product sits at +2; the difference of +2 is exactly the two-electron oxidation captured in the equation above. Because a ketone carbon's O.S. is already +2 with zero C–H bonds contributing to lower it, there is no accessible pathway (short of C–C bond cleavage, which is a distinct oxidative-cleavage regime, e.g. ozonolysis or periodate cleavage of diols) to push it any higher while it remains a four-bonded sp² carbon with two carbon substituents. For the tertiary alcohol, the same accounting gives:

identical starting oxidation state to the secondary case — but the mechanism below shows why the pathway to raise it is blocked at the very last step, not by thermodynamics but by the complete absence of a proton to remove.

MECHANISM 1 — Chromic acid / PCC oxidation of a secondary alcohol

Goal: convert a secondary alcohol R₂CH–OH into ketone R₂C=O using Cr(VI) (chromic acid, Jones reagent, or pyridinium chlorochromate, PCC), and show explicitly why the mechanism cannot proceed for a tertiary alcohol.

Step 1 — Formation of the chromate ester. The alcohol oxygen's lone pair attacks the electrophilic Cr(VI) center of HCrO₄⁻ (or CrO₃·pyridine for PCC), displacing a hydroxide (or chloride, in PCC) from chromium. This is a nucleophilic substitution at a metal center, mechanistically analogous to phosphorylation or sulfonation of an alcohol: the O–H bond survives this step (it is lost in the next step by acid–base chemistry), and a new O–Cr bond is formed, releasing water or HCl. The product is a chromate(VI) ester, R₂CH–O–CrO₃H (or R₂CH–O–CrO₂Cl for PCC).

Step 2 — Deprotonation and concerted elimination (the redox step). A base (a second equivalent of chromate, water, or in PCC's non-aqueous conditions, the acetate/chloride counterions or another chromate oxygen) removes the α C–H proton on the carbinol carbon, in a cyclic six-membered transition state exactly as drawn above. Simultaneously, the O–Cr bond breaks heterolytically, its electron pair flowing to become the new C=O π bond, while the C–H bond's electron pair flows to fill in the same π system as the proton leaves. All three bond reorganizations — C–H breaking, C=O forming, O–Cr breaking — happen in one concerted step, which is why the reaction shows a primary kinetic isotope effect (k_H/k_D typically 6–7) when the α C–H is replaced by α C–D: the C–H bond is being broken in the rate-determining step itself, not in a pre-equilibrium.

Step 3 — Chromium fate. Cr(VI) is reduced to Cr(IV) in this step; Cr(IV) is unstable and disproportionates with another Cr(VI) to give two Cr(V), which then each oxidize a second alcohol molecule via the same two-step sequence, ultimately delivering Cr(III) as the final reduced species — the color change from orange (Cr VI) to green (Cr III) that is the classic visual endpoint of a Jones oxidation.

Step 4 — Why this cannot happen for a tertiary alcohol. Step 1 proceeds identically: a tertiary alkoxide oxygen is perfectly nucleophilic and forms the tertiary chromate ester without difficulty. But Step 2 requires a proton on the carbinol carbon itself — the same carbon bonded to the O–Cr group — for the base to remove in that six-membered cyclic transition state. A tertiary carbinol carbon is bonded to O, C, C, C: there is no such proton. The chromate ester can only decompose by breaking the O–Cr bond heterolytically with no concomitant C–H removal, which simply regenerates the starting tertiary alcohol (or, under acidic aqueous Jones conditions, the alcohol may instead ionize via protonation and loss of water to a tertiary carbocation, which is then oxidized/eliminated by a completely different, non-carbonyl-forming pathway — typically E1 dehydration to an alkene, or slow further oxidative degradation that fragments the carbon skeleton). Either way, the clean R₃C–OH → ketone-type transformation is mechanistically forbidden, not merely slow.

MECHANISM 2 — Swern-type oxidation (DMSO activation) of a secondary alcohol

Goal: show the alternative, metal-free activation strategy — turning DMSO into an electrophilic sulfonium species, then using the identical six-membered elimination logic — and confirm it fails for the same structural reason on tertiary substrates.

Step 1 — Activation of DMSO. Dimethyl sulfoxide's oxygen lone pair attacks an electrophile — oxalyl chloride, (COCl)₂, in the classic Swern protocol. The DMSO oxygen displaces chloride from one carbonyl carbon of oxalyl chloride, generating a chlorosulfonium ion, Me₂S⁺–O–C(=O)–Cl, with loss of Cl⁻; this intermediate rapidly loses CO and CO₂ (a retro-ene-like fragmentation driven by relief of strain and formation of two very stable small gas molecules), leaving the reactive chlorodimethylsulfonium ion, Me₂S⁺–Cl, in solution at −78 °C.

Step 2 — Alcohol attacks the activated sulfur. The alcohol oxygen's lone pair displaces chloride from sulfur, forming an alkoxysulfonium ion: R₂CH–O–S⁺Me₂, with Cl⁻ now free in solution as the base for the next step. This alkoxysulfonium ion is the "activated ester" analog of the chromate ester in Mechanism 1 — oxygen now bears an excellent leaving group (a neutral dimethyl sulfide) built onto it.

Step 3 — Base-mediated ylide formation and concerted elimination. A base — in the classic Swern protocol, triethylamine, added at this point — removes a proton not from the S-methyl groups (that would give a different, non-productive ylide pathway that is actually the real mechanism!) — let's be precise: the accepted Swern mechanism has the base deprotonate one of the S-methyl C–H bonds to generate a sulfonium ylide, Me(CH₂⁻)–S⁺(–OCHR₂), which then undergoes an intramolecular five-membered cyclic proton transfer/elimination: the ylide carbon's lone pair is not what attacks the carbinol; rather, the mechanism proceeds through the ylide acting as the internal base that removes the carbinol α C–H in a five-membered cyclic transition state, while the C–O(alkoxysulfonium) bond's departure delivers dimethyl sulfide and forms the new C=O π bond, all concertedly. The electron flow: the carbanion-like ylide carbon's lone pair pulls the α C–H proton toward itself (forming a new C–H bond on what becomes a methyl group of the departing Me₂S), the C–H bond on the carbinol carbon breaks heterolytically with its electrons flowing into the forming C=O π bond, and the O–S bond breaks with its electrons leaving as the lone pair on the departing neutral sulfide sulfur.

Step 4 — Products. The ketone R₂C=O is released, along with dimethyl sulfide, Me₂S, and Et₃NH⁺Cl⁻. The characteristic unpleasant sulfurous odor of a Swern oxidation is exactly this dimethyl sulfide byproduct.

Step 5 — Why tertiary alcohols fail here too. Steps 1–2 are unaffected by substitution at the carbinol carbon: a tertiary alkoxide is a perfectly good nucleophile and forms the tertiary alkoxysulfonium ion without any problem. But Step 3's cyclic elimination requires a proton on the carbinol carbon (R₂CH–O–S⁺Me₂ has that proton; R₃C–O–S⁺Me₂ does not). With no α C–H to remove, the tertiary alkoxysulfonium ion cannot collapse to a carbonyl; at best it hydrolyzes back to the tertiary alcohol on aqueous workup, or persists as an unreactive intermediate. This is the general pattern: every oxidant that works by "activate oxygen, then do a concerted intramolecular E2-type elimination using the carbinol C–H" is structurally blind to tertiary alcohols — not because the mechanism is slow for them, but because a required atom (the carbinol C–H) simply is not there.

Example 1 — Foundational: Jones oxidation of cyclohexanol. Cyclohexanol, treated with Jones reagent (CrO₃ in aqueous H₂SO₄/acetone), gives cyclohexanone. Mechanistically: the ring oxygen's lone pair attacks Cr(VI) of HCrO₄⁻ (formed in situ from CrO₃ + H₂O in the acidic medium), displacing water and forming the cyclohexyl chromate ester. A base — here, a water molecule or a second HCrO₄⁻ — then removes the axial or pseudo-axial C1–H proton in the six-membered cyclic transition state described in Mechanism 1, while the O–Cr bond breaks and the new C=O forms. Stereochemical note worth making explicit: cyclohexanol's carbinol carbon (C1) is a stereocenter only in substituted derivatives (unsubstituted cyclohexanol has a plane of symmetry through C1 and C4, so it is not chiral), but the mechanistic point generalizes — the carbinol carbon rehybridizes from sp³ to sp², so any stereochemical information at that specific carbon is permanently erased upon oxidation. This becomes important the moment you consider a chiral secondary alcohol (see Example 3). Product accounting: the oxidation-state change at C1 is exactly the $0 \to +2$ shift derived above,

matching the two-electron reduction of Cr(VI). Because cyclohexanone's carbonyl carbon (C1 of the product) is bonded to O, O, C2, C6 — no hydrogen — the reaction terminates cleanly at the ketone; there is no over-oxidation pathway available to a ring-opened diacid under these mild aqueous-acidic conditions (ring C–C cleavage requires much more forcing oxidants, e.g. hot KMnO₄ or HNO₃, which is a separate industrial process for making adipic acid).

Limiting-case check: if instead 1-methylcyclohexanol (a tertiary alcohol, methyl and OH both on C1) is subjected to the identical Jones conditions, the chromate ester still forms at C1–O, but C1 bears no hydrogen (its four substituents are O, CH₃, C2, C6) — Step 2 of Mechanism 1 has no proton to remove, so no ketone forms. Under the strongly acidic aqueous conditions of Jones reagent, the actual observed outcome is acid-catalyzed E1 dehydration to 1-methylcyclohexene (protonation of OH, loss of water to the tertiary carbocation, then loss of a β-H to give the alkene) — a completely different reaction pathway, confirming that "resistance to oxidation" for 3° alcohols does not mean "inert to these reagents," it means "diverted to an entirely different mechanism that does not involve the α-C–H elimination step at all."

Example 2 — Applied: chemoselective oxidation of a diol in a synthesis-style substrate. Consider a substrate bearing both a primary and a secondary hydroxyl — for concreteness, take (S)-4-(hydroxymethyl)cyclohexan-1-ol's secondary alcohol position, i.e., a molecule with a primary CH₂OH and a ring secondary CH–OH, and suppose the synthetic goal is to oxidize only the secondary alcohol to the ketone while leaving the primary alcohol untouched (a common requirement when the primary alcohol is destined to become, e.g., a leaving group or is protected in a later step). Naively, Jones reagent would over-oxidize the primary alcohol all the way to the carboxylic acid (via the aldehyde hydrate mechanism) while also oxidizing the secondary alcohol to the ketone — poor chemoselectivity by oxidation level, since primary alcohols are, if anything, more reactive toward Cr(VI) oxidation, not less. The correct strategic choice is not really "differentiate 1° vs 2° directly" (both oxidize readily) but rather to protect the primary alcohol first, most commonly as a bulky silyl ether — this is exactly the strategy foreshadowed by silyl-ether protecting group chemistry — oxidize the free secondary alcohol with PCC or DMP, then deprotect. Running the oxidation itself with Dess–Martin periodinane (DMP) on the free secondary alcohol: the alcohol oxygen attacks the electrophilic hypervalent I(V) center of the periodinane, displacing one of its acetate ligands (an associative ligand exchange at iodine, mechanistically parallel to Step 1 of Mechanism 1 — a nucleophilic substitution at a hypervalent center). A second acetate, now acting as an intramolecular base, removes the carbinol α C–H in a five-membered cyclic transition state (the smaller ring size compared to Cr and S oxidants reflects iodine's different preferred geometry/bond lengths), with the C–H electrons flowing into the forming C=O π bond and the I–O bond breaking to release I(III) (the reduced iodinane, which becomes the byproduct 1-hydroxy-1,2-benziodoxol-3(1H)-one after loss of the last acetate and reduction). The reaction is mild, buffered (typically run with solid NaHCO₃ present to mop up any acetic acid byproduct), and famously tolerant of acid-sensitive groups. Numbers worth internalizing for practical lab work: DMP oxidations are routinely run at 0 °C to room temperature and go to completion in 15 minutes to 2 hours; the reagent's mildness compared to Jones (strongly acidic, aqueous) or even PCC (mildly acidic, can epimerize sensitive stereocenters or promote elimination in acid-sensitive substrates) is precisely why DMP has become a default choice in complex-molecule total synthesis since its popularization in the 1980s–90s.

Example 3 — Exam/competition level: stereochemical consequence of oxidation–reduction sequences. A classic mechanistic-stereochemistry problem: (R)-butan-2-ol is oxidized with PCC to butan-2-one, and the ketone is then reduced with NaBH₄. What is the stereochemical outcome, and what does this tell you about using oxidation as a stereochemical "reset" step?

Step 1 (oxidation). (R)-Butan-2-ol has its stereocenter at C2, bonded to OH, H, CH₃, and CH₂CH₃. PCC forms the chromate ester at C2–O (Step 1 of Mechanism 1); the base then removes the C2–H in the concerted six-membered elimination (Step 2), and because that same carbon's C–H bond is the bond being broken, C2 rehybridizes from sp³ to sp². The moment C2 becomes the trigonal-planar carbonyl carbon of butan-2-one, all stereochemical information at C2 is destroyed — butan-2-one is achiral (planar at C2, and its two faces, Re and Si, are diastereotopic only in the sense of face selectivity for subsequent nucleophilic addition, not stereocenters in the ground state).

Step 2 (reduction). NaBH₄ delivers hydride to the carbonyl carbon of butan-2-one. Because butan-2-one has no adjacent chiral centers and no chiral catalyst/reagent is present, hydride adds with equal probability to the Re face and the Si face of the planar sp² carbon — the two transition states leading to (R)- and (S)-butan-2-ol are enantiomeric (mirror images of each other) and therefore isoenergetic.

Conclusion: the product is racemic, $(\pm)$-butan-2-ol — a 1:1 mixture of (R) and (S) — even though the starting material was enantiopure (R). This is a mechanistic teaching point disguised as a stereochemistry problem: oxidation of a stereocenter that resides at the carbinol carbon itself is not merely "a reaction," it is an act of erasing configurational memory, because the rate-determining, stereo-defining step converts an sp³ center into an sp² center. Any synthesis that needs to retain configuration at a secondary alcohol's carbon must avoid oxidizing that specific carbon, or must reintroduce the stereocenter afterward with a stereoselective reduction (chiral catalysts, enzymatic reduction, or substrate-controlled diastereoselective delivery from an adjacent existing stereocenter) rather than with achiral NaBH₄.

Limiting-case check: if instead the alcohol carried a second, remote stereocenter unaffected by the oxidation (e.g., oxidizing the C2 secondary alcohol of a molecule that also has a fixed stereocenter at C5), that remote center is completely untouched by the mechanism — the six-membered cyclic TS only engages the C2–H and C2–O bonds — so the product retains full enantiomeric/diastereomeric purity at C5 while C2 becomes achiral sp². This confirms the mechanism is strictly local to the carbinol carbon and its immediate substituents, exactly as the curved arrows in Mechanism 1 depict.

Georg Bredig, Hans Fischer, and a long line of 19th- and 20th-century chromium chemists established the empirical chromate-oxidation toolkit, but the mechanistic understanding that unifies all these oxidants — a concerted, cyclic, E2-like elimination from an O-activated intermediate — is best credited to the physical-organic tradition of rate-law and isotope-effect studies applied to chromic acid oxidation in the mid-20th century, most notably by Frank H. Westheimer, whose kinetic and deuterium-isotope-effect experiments on chromic acid oxidation of isopropanol in the 1940s–1950s provided the decisive evidence that C–H bond cleavage occurs in the rate-determining step itself — precisely the concerted mechanism taught today. Westheimer measured a substantial primary kinetic isotope effect (k_H/k_D on the order of 6–7) for the oxidation of isopropanol versus its α-deuterated analog by chromic acid, a result that rules out mechanisms in which the chromate ester decomposes via a slow ionization step not involving the C–H bond (which would predict little to no isotope effect) and instead demands that the C–H bond is broken concertedly with C=O formation and Cr–O cleavage — exactly the picture built into Mechanism 1 above.