Add methylmagnesium bromide to acetone and you obtain tert-butyl alcohol. Add the same reagent to ethyl acetate and the reaction does not stop at a ketone. Two equivalents add, and the product is again tert-butyl alcohol. Meanwhile, dissolve acetone in water and almost nothing happens, yet formaldehyde in water exists almost entirely as its hydrate, methanediol.

These outcomes look unrelated, but they share one event. A nucleophile approaches a carbonyl carbon, the π bond breaks, and a trigonal sp2 carbon becomes tetrahedral. Almost every reaction in carbonyl chemistry passes through this step or its reverse, including reductions, Grignard additions, acetal formation, ester hydrolysis, and amide coupling.

What differs is what happens next. The new tetrahedral carbon can keep its nucleophile, expel a different group, or fall back to the starting material. Which one it does depends on three questions: how electrophilic the carbonyl is, what groups it carries, and how readily the nucleophile can leave again. Answer those, and most carbonyl chemistry becomes predictable rather than memorised.

The carbon–oxygen double bond is polarised before any reagent arrives. Oxygen has an electronegativity of 3.44 against carbon's 2.55, so it pulls electron density toward itself in both the σ and π bonds. The π bond is more diffuse and more polarisable, so it shifts furthest. Acetone carries a dipole moment of about 2.9 debye. A useful way to picture this is a resonance contributor in which carbon bears a positive charge and oxygen a negative one. It is never the whole truth, but it is a meaningful share of it.

The orbital picture sharpens the point. The lowest unoccupied molecular orbital of a carbonyl is the π* orbital. Because oxygen's atomic orbitals sit lower in energy, π* carries its larger coefficient on carbon. A nucleophile donates electrons from its highest occupied orbital into that π*. Overlap is best not perpendicular to the plane but at roughly 107 degrees to the C=O axis. This approach, the Bürgi–Dunitz trajectory, maximises contact with π* while avoiding repulsion from the electron density on oxygen.

Substituents tune that orbital. Aldehydes react faster than ketones for two reasons. A second alkyl group donates electron density through hyperconjugation, and it crowds the tetrahedral product, where bond angles compress from 120 to about 109.5 degrees. Electron-withdrawing groups such as trifluoromethyl do the opposite. Heteroatoms with lone pairs, such as the oxygen of an ester or the nitrogen of an amide, donate into π* and raise its energy. That donation explains the familiar reactivity order: acid chlorides, then anhydrides, then esters, then amides, the least electrophilic of all.

When the nucleophile is weak, the carbonyl can be activated. Protonating the oxygen, or coordinating it to a Lewis acid such as a metal cation, makes the carbocation-like resonance contributor dominant and lowers π* further. This is why water and alcohols, which are poor nucleophiles, need acid catalysis to add at useful rates. It is also why Lewis acids like boron trifluoride and titanium tetrachloride appear throughout synthetic routes to pharmaceutical intermediates.

Takeaway

A carbonyl's reactivity comes down to the energy of its π* orbital and how much of that orbital sits on carbon. Every substituent, catalyst, or solvent effect is ultimately a statement about that orbital.

Attack converts the planar carbonyl into a tetrahedral intermediate: an alkoxide on a carbon that now carries four groups. This species is high in energy relative to most of its possible fates, partly because it has given up a C=O π bond. A carbon–oxygen double bond is worth roughly 740 kJ/mol, compared with about 360 kJ/mol for a single bond. The alkoxide's lone pair is therefore strongly inclined to push back down and rebuild the double bond. The question is what gets pushed out.

Leaving-group ability follows the acidity of the departing group's conjugate acid. On an aldehyde or ketone, the candidates are hydride and carbanions, whose conjugate acids have pKa values near 35 and 50. Neither leaves. The intermediate persists until protonation gives an alcohol, so the net result is addition. On an acid derivative, the candidates are chloride (HCl, pKa about −7), carboxylate (about 5), or alkoxide (about 16). Here the intermediate collapses and expels the group, so the net result is substitution by addition–elimination.

This logic explains the Grignard reaction with an ester. The first addition expels ethoxide and generates a ketone. That ketone is more electrophilic than the ester it came from, so it captures a second equivalent of Grignard reagent faster than the remaining ester captures the first. Stopping at the ketone requires preventing collapse. The Weinreb amide does exactly this. Magnesium chelates between the alkoxide and the N-methoxy oxygen, holding the tetrahedral intermediate intact until aqueous workup destroys the remaining organometallic.

Amides sit at the other extreme. Nitrogen donation stabilises their carbonyl, and the amide anion, with a conjugate acid pKa near 38, is a very poor leaving group. That is why peptide bonds survive for years in neutral water. Serine proteases overcome both problems at once. An oxyanion hole stabilises the tetrahedral intermediate through hydrogen bonds, and a histidine protonates the departing nitrogen, turning it into a neutral amine that leaves readily.

Takeaway

The tetrahedral intermediate is a junction, not a destination. Chemists choose between addition and substitution by controlling how long that intermediate lives and how readily its substituents can leave.

Many carbonyl additions are equilibria, and the clearest data come from hydration. Formaldehyde in water is more than 99.9 percent methanediol, with a hydration constant around 2000. Acetaldehyde sits close to 1, roughly half hydrated. Acetone's constant is about 0.001, so its hydrate is barely detectable. Hexafluoroacetone, with two strongly withdrawing groups, hydrates almost completely. Chloral forms a hydrate stable enough to be isolated as a crystalline solid.

The span from formaldehyde to acetone is a factor of about a million in equilibrium constant. That corresponds to roughly 35 kJ/mol in free energy at room temperature. The same factors that control rate also control equilibrium position. Alkyl groups stabilise the carbonyl through hyperconjugation and destabilise the crowded tetrahedral adduct. Withdrawing groups destabilise the carbonyl, which leaves the adduct comparatively favoured. Electrophilicity, in other words, is both a kinetic and a thermodynamic property.

The identity of the nucleophile decides whether reversal is possible at all. The rule mirrors the previous section: a nucleophile that is also a competent leaving group adds reversibly. Hydride and carbanions add irreversibly because expelling them is prohibitively uphill. Cyanide, whose conjugate acid HCN has a pKa of 9.2, adds reversibly, which is why cyanohydrins revert under basic conditions. Water, alcohols, and amines also add reversibly. Entropy matters too: glucose exists overwhelmingly as a cyclic hemiacetal because its alcohol and aldehyde groups belong to the same molecule.

Reversible additions can be driven to completion by coupling them to an irreversible step. Acetal and imine formation are pushed forward by removing water with a Dean–Stark trap or molecular sieves. When a protecting group must come off, aqueous acid reverses the reaction. Reductive amination, one of the most widely used bond-forming reactions in pharmaceutical synthesis, follows the same pattern. It forms an imine reversibly, then traps the iminium ion with sodium cyanoborohydride, which reduces the iminium far faster than it reduces the parent ketone.

Takeaway

Ask whether the nucleophile could leave again. If it could, the reaction is an equilibrium you must manage; if it cannot, the first addition is the final word.

Carbonyl addition reduces to three linked questions. How low in energy is the π* orbital, and how much of it sits on carbon? Which group on the tetrahedral intermediate can leave? Can the nucleophile itself depart again? Each answer narrows the possible outcomes.

These questions translate directly into design decisions. A process chemist choosing an acylating agent weighs electrophilicity against selectivity. A medicinal chemist who needs a ketone reaches for a Weinreb amide because it holds the intermediate in place. A plant engineer drives an acetal forward by controlling water rather than by adding more reagent.

There is a quiet economy in this. From one polarised double bond and one change in geometry, from trigonal to tetrahedral, come reductions, esterifications, protein hydrolysis, and sugar chemistry. Follow the electrons into that tetrahedral carbon, and the outcome is rarely a surprise.