The Corey-Bakshi-Shibata (CBS) Reduction

What the CBS reduction does

A ketone carbonyl is flat and symmetric: the two faces of the C=O plane are mirror images. Add a hydride to one face and you get one enantiomer of alcohol; add it to the other face and you get its mirror image. An achiral reductant like sodium borohydride cannot tell the two faces apart, so it delivers a 50:50 racemate. The CBS reduction installs a small chiral catalyst that can tell them apart, funnelling the reduction through a single well-ordered transition state so that hydride lands on one face over and over.

The net transformation is simple to state and hard to do well: prochiral ketone → single-enantiomer secondary alcohol, with an enantiomeric excess that is typically 90-97% and can exceed 99% on well-behaved substrates. The chirality comes entirely from the catalyst — a chiral oxazaborolidine — so by swapping to its mirror-image catalyst you get the opposite alcohol from the same ketone.

        O                              OH
        ‖         (S)-CBS cat.          |
   R_L—C—R_S   ──────────────────→   R_L—C—R_S     (single enantiomer)
                  BH₃ (0.6 eq)           H
   prochiral ketone               chiral 2° alcohol

The mechanism, arrow by arrow

The catalyst is a five-membered ring containing both a nitrogen and a boron — the oxazaborolidine. Those two atoms play two completely different roles, and understanding the mechanism is really just tracking what each of them does.

1. The nitrogen activates the borane. The ring nitrogen is Lewis-basic. Its lone pair attacks the empty p-orbital of BH₃, forming an N→B dative bond. This does two things at once: it turns the once-sluggish BH₃ into a much more reactive, "loaded" hydride donor, and — by pulling electron density through the ring — it makes the endocyclic boron (the one already in the ring) markedly more Lewis acidic.
2. The endocyclic boron binds the ketone. That now-hungry ring boron coordinates to the lone pair of the ketone oxygen. Crucially, the oxygen coordinates through the lone pair that is anti to the bulky substituents of the catalyst, so the ketone can only dock in one orientation: its large group (R_L) swings away from the ring, its small group (R_S) tucks in toward it.
3. A six-membered transition state assembles. Now everything is pinned in one rigid piece: catalyst boron–O–C of the carbonyl on one arc, and catalyst nitrogen–B(of BH₃)–H on the other. The migrating hydride and the carbonyl carbon are held face-to-face in a chair-like six-membered ring.
4. Hydride transfers. A hydride slides from the N-bound BH₃ across to the carbonyl carbon in a concerted step — the C=O π bond breaks, a C–H bond forms, and the oxygen picks up the boron as a new B–O bond. Because the ketone was locked with R_L pointing one way, the hydride necessarily hits one specific face. That is the enantioselective step.
5. The catalyst turns over. The product is a boron alkoxide still tethered to the catalyst. It cleaves off (a retro-coordination / B–O metathesis with more BH₃), releasing the catalyst to bind another BH₃ and another ketone. During aqueous workup the boron esters hydrolyse to give the free alcohol.

   cat-N: ──→ BH₃           (1) N activates borane, ring-B gets Lewis acidic
       \                          |
        B(ring) ←: O=C            (2) ring-B binds ketone O, ketone locked R_L out
        |          / \
   six-membered TS:  R_L R_S      (3) chair-like 6-ring pins H and C face-to-face
        H···C ──────────►         (4) hydride crosses to ONE face → chiral C
        |
    boron alkoxide → workup → R_L-CH(OH)-R_S   (5) catalyst released, alcohol freed

The reason this beats a simple achiral hydride is geometric: the two boron/nitrogen roles force the ketone and the incoming hydride into one mutual arrangement. There is no low-energy way for the ketone to flip and present its other face, so one enantiomer dominates.

Predicting the product enantiomer

You do not need to solve the whole transition state to call the answer. The working rule:

• Identify the larger group R_L and smaller group R_S flanking the carbonyl.
• In the transition state R_L points away from the catalyst; hydride is delivered from the face set by the catalyst's chirality.
• The (S)-CBS catalyst (from natural (S)-proline) delivers hydride so that acetophenone becomes (R)-1-phenylethanol. The (R)-CBS catalyst gives (S)-1-phenylethanol.

Because the selectivity is read off the size difference between the two groups, the rule is remarkably robust across substrates as long as one group is clearly bigger than the other. Aryl versus methyl is a big gap and gives excellent ee; ethyl versus methyl is a small gap and gives poor ee. Note the R/S label of the product can flip depending on CIP priorities even when the geometric face is the same — always assign priorities on the actual product rather than memorising "(S)-cat gives R".

Reagents, catalyst and conditions

• The catalyst. A chiral oxazaborolidine made from α,α-diphenylprolinol (two steps from proline). The bench-stable, commercially sold version has a methyl group on the ring boron and is called (S)- or (R)-2-methyl-CBS-oxazaborolidine ("MeCBS"), usually supplied as a ~1 M solution in toluene. Loading is typically 5-10 mol%, though as little as 2 mol% works on easy substrates.
• The stoichiometric reductant. Borane, as BH₃·THF or the more stable BH₃·SMe₂ (BMS). Each BH₃ carries three hydrides, so you only need about 0.6 equivalents per ketone. For sensitive substrates (enones, 1,2-diketones), catecholborane at low temperature is milder and more chemoselective.
• Solvent and temperature. Toluene, THF, or CH₂Cl₂ under a dry, inert atmosphere. Reactions run anywhere from −20 °C to room temperature; catecholborane variants often go to −78 °C.
• Order of addition. This matters more than almost anything else. Pre-mix catalyst and a little borane, then add the ketone (and the rest of the borane) slowly — often by syringe pump. Keeping free BH₃ low starves the racemising uncatalysed background reduction.
• Workup. Quench excess borane with methanol (watch for H₂ evolution), then aqueous acid to hydrolyse the boron esters; extract and purify.

Scope, selectivity and stereochemistry

The CBS reduction is at its best exactly where an achiral reductant is useless: creating a new stereocentre at a carbon flanked by two groups that differ mainly in size.

• Aryl alkyl ketones — acetophenone and relatives give 90-97% ee. The aryl/alkyl size gap is ideal.
• α-Halo ketones — reduced to chiral halohydrins, which cyclise to enantiopure epoxides. A classic route to single-enantiomer terminal epoxides.
• α,β-Ynones and enones — propargylic and allylic alcohols in high ee, especially with catecholborane at low temperature.
• Diaryl ketones with subtle differences — even ortho-substitution or a single isotope-like size difference can be read, though ee falls as the two sides converge in size.
• Dialkyl ketones with similar groups — the weak spot. 2-butanone (methyl vs ethyl) gives low ee because the catalyst cannot resolve which side is "large."

The reaction is highly chemoselective for ketones over esters, nitriles, nitro groups, and halides, because the catalyst-templated pathway is fast only for a ketone that can dock on boron; most other reducible groups simply ride through untouched.

CBS vs. other ways to make a chiral alcohol

	CBS reduction	Noyori asymmetric hydrogenation	NaBH₄ (achiral)
Chirality source	Chiral oxazaborolidine catalyst	Chiral Ru/BINAP or Ru/diamine	None — gives racemate
Reductant	BH₃ or catecholborane	H₂, or HCO₂H / iPrOH (transfer)	NaBH₄
Catalyst loading	5-10 mol%	0.01-1 mol% (very high TON)	n/a
Typical ee	90-97%	95-99%	0% (racemic)
Conditions	−78 to 25 °C, inert atmosphere, no pressure vessel	Often 1-50 bar H₂, heated	Room temperature, protic
Metal-free?	Yes	No (Ru, Rh, Ir)	Yes
Reads which property?	Steric size gap R_L vs R_S	Coordination to metal + directing groups	Nothing
Best scale	Grams — bench / med-chem	Kilos to tonnes — process	Any (no selectivity)
Pick when	Fast asymmetric reduction, no H₂ rig	Cheap reductant matters at scale	You don't need a single enantiomer

Worked example: acetophenone → (R)-1-phenylethanol

The textbook demonstration. Acetophenone has a big phenyl (R_L) and a small methyl (R_S) — a large size gap, so ee is excellent.

    Ph-C(=O)-CH₃  ──(S)-MeCBS 10 mol%, BH₃·SMe₂ 0.6 eq, toluene, −20→25 °C──→  (R)-Ph-CH(OH)-CH₃

• Setup. Charge (S)-2-methyl-CBS-oxazaborolidine (0.10 equiv, 1 M in toluene) and a small portion of BH₃·SMe₂ into dry toluene under nitrogen; cool.
• Addition. Add a solution of acetophenone and the remaining BH₃·SMe₂ (0.6 equiv total borane) slowly by syringe pump over 30-60 min — this keeps free borane low and suppresses the racemic background reduction.
• Workup. Quench carefully with methanol (H₂ evolves), then dilute HCl to break the boron esters; extract, dry, concentrate.
• Result. (R)-1-phenylethanol in ~95% yield and 96-97% ee. Switching to (R)-MeCBS under identical conditions gives (S)-1-phenylethanol in the same ee.

A named-application highlight: the asthma drug montelukast (Singulair) needs exactly this kind of transformation — setting a benzylic alcohol stereocentre on a large aryl ketone intermediate. Merck's commercial route actually locked that centre with a related stoichiometric chiral borane, (−)-DIP-Cl, but the same intermediate is a textbook oxazaborolidine target and (R)-MeCBS reductions of it appear throughout the montelukast patent literature — a reminder that CBS and its chiral-borane cousins run cleanly on multi-kilogram process scale.

Limitations and side reactions

• Small size gap = poor ee. When R_L and R_S are similar (many dialkyl ketones), the catalyst can't decide which side is large and selectivity collapses. Choose a different method or a substrate handle.
• The uncatalysed background reduction. Free BH₃ reduces ketones on its own with no facial control. Any protocol that lets borane accumulate — fast addition, too much borane, warm start — bleeds ee. Slow, controlled addition is the antidote.
• Moisture and air. Both the oxazaborolidine and borane are moisture-sensitive; wet solvent hydrolyses the catalyst's B–N/B–O framework and kills activity. Dry everything.
• Over-reduction / competing reduction. Aldehydes are reduced far faster than ketones and with little selectivity; α,β-unsaturated ketones can suffer some 1,4-reduction with BH₃ unless you switch to catecholborane at low temperature.
• Cost of borane on scale. BH₃ is more expensive (and more hazardous to handle in bulk) than H₂, which is why large-scale routes sometimes prefer catalytic hydrogenation despite CBS's convenience on the bench.

Discovery: who, when and where it came from

The reaction is named for Elias James (E. J.) Corey, R. K. Bakshi, and S. Shibata, who reported it in 1987 in the Journal of the American Chemical Society (J. Am. Chem. Soc. 1987, 109, 5551 and 7925). Corey — already a Nobel laureate-to-be (1990, for the logic of organic synthesis and retrosynthetic analysis) — and his co-workers rationalised and dramatically improved an earlier observation.

The seed came from Shinichi Itsuno and co-workers, who around 1981 found that an amino-alcohol–borane combination could reduce ketones enantioselectively (the reduction is sometimes called the Corey–Itsuno reduction in recognition). Corey's group recognised that a preformed, well-defined oxazaborolidine — a single molecule doing both the borane activation and the ketone binding — was the active catalyst, worked out the six-membered transition-state model that predicts the product, and engineered the bench-stable B-methyl "MeCBS" version that made the reaction a routine tool. Within a few years it had become one of the most-used asymmetric reductions in both academic and industrial synthesis.

Practical and industrial notes

• Both enantiomers are cheap. Proline is available in natural (S) and unnatural (R) forms, so both (S)- and (R)-MeCBS catalysts are sold off the shelf — you simply pick the one that gives the alcohol you want.
• Catalyst is recoverable in principle but on lab scale the small loading is usually just consumed; process routes optimise borane addition and temperature rather than catalyst recycling.
• Safety. Borane reagents are flammable, react violently with water and protic solvents (H₂ release), and BH₃·SMe₂ has a strong odour; handle under inert atmosphere with proper venting. Quench slowly and cold.
• Process pedigree. Beyond montelukast, oxazaborolidine reductions appear throughout pharmaceutical process chemistry precisely because they run without a hydrogenation rig, tolerate many functional groups, and give predictable stereochemistry — a rare combination of speed, selectivity, and operational simplicity.