Grignard Addition: Reactivity Order and Stereochemical Outcome

Electrophilicity of a carbonyl carbon tracks how much electron density is pushed into it and how good a leaving group sits next to it. Aldehydes (-CHO) have only one alkyl/H group donating electron density and no resonance-stabilising leaving group, so they are the most reactive. Ketones (-COR) are next — one more alkyl group donates density but there is still no leaving group. Esters (-COOR) are slower still, because the lone pair on the ester oxygen conjugates into the carbonyl, and once RMgX adds, -OR can leave to regenerate a ketone (which then reacts again with a second equivalent — this is exactly why excess Grignard converts esters to tertiary alcohols). Amides (-CONR₂) are the least electrophilic of the four: nitrogen is a far better π-donor than oxygen, so the carbonyl carbon is the most deactivated and amides are essentially inert to Grignard reagents under normal conditions. Choose the reactivity order -CHO > -COR > -COOR > -CONR₂ whenever a question asks which group reacts first, or which survives untouched, in a molecule carrying more than one of these groups.

When MeMgI adds to a ketone that already has an optically pure stereocentre elsewhere in the molecule (not at the carbonyl carbon itself), the Grignard attacks the flat, sp₂ carbonyl carbon and creates a brand-new stereocentre there. The pre-existing stereocentre is untouched by the reaction — it is not involved in any bond-breaking step — so it retains its configuration throughout. The addition can occur from either face of the planar carbonyl, but because the molecule already has a fixed chiral environment nearby, the two faces are not equivalent: the reagent approaches the less hindered face preferentially, giving one diastereomer as the major product rather than a 50:50 mixture. On a molecule that has no other stereocentre at all, the same addition to a planar carbonyl would instead give a racemic (50:50) mixture, since both faces would then be truly equivalent.

Two failure modes to watch for. First, do not assume any Grignard addition automatically racemises the molecule — only bond-breaking at an existing stereocentre can do that, and simple addition to a carbonyl does not touch other stereocentres. Second, when a molecule carries -CHO, -COR, -COOR and -CONR₂ together, resist the instinct to rank them by how oxidised the carbon looks — rank them by π-donation from the attached heteroatom into the carbonyl instead, which is what actually controls how electrophilic the carbon is.