Alcohols from Grignard and Organometallic Addition
A flask of magnesium turnings, an alkyl halide, and dry ether — left alone, nothing happens. Add a single crystal of iodine or a trace of 1,2-dibromoethane, though, and the mixture changes. Within minutes it begins to boil gently under its own reaction exotherm. This is the birth of a Grignard reagent. It is one of the most powerful carbon nucleophiles in the chemist's toolkit. The moment it is poured onto a carbonyl compound, a new C–C bond forms where none existed before. This is the essence of building carbon skeletons. It is not just decorating a chain with new functional groups. It is literally lengthening it. The addition of organometallic reagents (RMgX, RLi, and their cousins RLi·LiBr, R₂CuLi) to carbonyl electrophiles is arguably the single most important C–C bond-forming reaction. It is taught at the undergraduate level. Reduction of a carbonyl with Alcohols from Reduction of Carbonyls uses hydride donors like NaBH₄ or LiAlH₄. These deliver H⁻ and preserve the carbon count. Grignard and organolithium addition delivers an entire carbon fragment R⁻ instead. This effectively stitches two independent molecules into one larger one. This single conceptual shift — nucleophile as carbanion rather than hydride — is important. It is why the reaction anchors nearly every retrosynthetic analysis a student will ever draw. Any tertiary alcohol, on paper, disconnects into a ketone plus an organometallic. Or it disconnects into an ester plus two equivalents of the same. The mechanistic story is deceptively simple to sketch. A curved arrow moves from a filled sp³ (or, for vinyl/aryl systems, sp²) carbon orbital into the carbonyl π*. But the real subtlety lies elsewhere. It lies in the aggregated, Lewis-acidic nature of the organometallic itself. Grignard reagents are not simple "R⁻ MgX⁺" ion pairs. They exist as solvated monomers, dimers, and Schlenk-equilibrium mixtures. The magnesium cation plays an active role in activating the electrophile. It does this even as the carbon delivers its lone pair. Understanding this dual role is essential. Carbanion nucleophile and Lewis acid template work in concert. This separates rote memorization of "Grignard + aldehyde → secondary alcohol" from real mastery. It gives a mechanistic command of the reaction. That command predicts chemoselectivity, over-addition, and failure modes. This article builds the mechanism from the ground up. It covers formation of the organometallic via oxidative addition at magnesium. It covers the four-centered (or six-centered, solvent-bridged) transition state of carbonyl addition. It covers the stoichiometric consequences of using esters versus aldehydes versus ketones. It also covers the practical incompatibilities. These include acidic protons, other electrophilic sites, and functional groups. Such groups must be masked before a Grignard reagent is ever introduced. It closes with reasoning that leads directly into the next topic. That reasoning explains why chemists reach for Alcohol Protecting Groups: Silyl Ethers once these alcohols are made.
- RMgX and RLi are strong carbanion nucleophiles; the metal is both the source of electron density (via the polarized C–M bond) and a Lewis-acidic carbonyl activator.
- Addition to formaldehyde gives 1° alcohols; to aldehydes gives 2° alcohols; to ketones gives 3° alcohols; to esters/acid chlorides gives 3° alcohols after two additions.
- The mechanism is a single nucleophilic addition step: filled C–Mg σ or carbanion lone pair attacks the carbonyl π, C=O collapses to C–O⁻, aqueous acidic workup protonates the alkoxide.
- Grignard/organolithium reagents are destroyed by any acidic proton (–OH, –NH, –COOH, terminal alkyne C–H) and add irreversibly to esters/nitriles/epoxides — all must be considered before choosing a route.
- Chemoselectivity (mono- vs bis-addition to esters, 1,2- vs 1,4-addition to enones) is controlled by reagent identity: RLi/RMgX favor 1,2-addition; R₂CuLi (Gilman) favors conjugate 1,4-addition.
- Stereochemistry: addition to a prochiral carbonyl with an adjacent stereocenter is governed by Felkin–Anh (or chelation) models, giving predictable diastereomer ratios.
- Key idea: a Grignard/organolithium reagent is a carbanion equivalent whose metal both delivers and (via Lewis-acid coordination to oxygen) accelerates addition into a carbonyl π* system, forging a new C–C bond and, after protonation, a new alcohol.
Where this lives today
Grignard and organolithium addition remains the industrial-scale workhorse for constructing tertiary alcohols and complex carbinol centers found throughout pharmaceuticals. Tamoxifen, a triarylethylene selective estrogen receptor modulator used in breast cancer treatment, is synthesized via a key Grignard addition step building the triaryl-substituted alkene backbone from a ketone precursor. Many antihistamines and steroid-modifying syntheses similarly rely on organometallic addition to install quaternary or tertiary carbinol centers that are otherwise difficult to construct with comparable atom economy. In materials chemistry, organolithium reagents (particularly n-BuLi) are the standard tool for generating aryllithiums via halogen–metal exchange, feeding directly into carbonyl additions that build the extended, sterically defined cores of liquid-crystal and organic-semiconductor molecules. In an academic research or industrial process-chemistry laboratory, the practical challenge these methods raise most often — protecting an alcohol or amine elsewhere in a molecule so that a Grignard addition can be run selectively at a distal carbonyl without self-destructing the reagent — is precisely the motivation for the protecting-group strategies (silyl ethers, acetals, carbamates) developed to shield sensitive functionality during multi-step synthesis.
Ask the tutor: How does the choice between Felkin–Anh and chelation control change if the α-substituent is a bulky silyl ether rather than a free hydroxyl or methoxy group?
The physical picture — thorough prose
Picture the carbonyl carbon of acetone. It is sp² hybridized, and the C=O π bond is built from the overlap of a carbon 2p orbital with an oxygen 2p orbital, polarized toward oxygen because oxygen is more electronegative. The π antibonding orbital — empty, and lower in energy on carbon than on oxygen because of that same polarization — is the target. Any nucleophile that can donate electron density into that π orbital, approaching along the Bürgi–Dunitz trajectory (roughly 105–107° off the C=O axis, not a perpendicular 90° approach), will convert the sp² carbonyl carbon into an sp³ alkoxide carbon. A Grignard reagent, R–MgX, presents exactly such a nucleophile. The C–Mg bond is highly polarized — carbon is far more electronegative than magnesium (Pauling electronegativity of C ≈ 2.5, Mg ≈ 1.3) — so the bonding electron pair sits close to carbon, giving R essentially carbanion-like reactivity even though no free R⁻ ever exists in solution. Crucially, Grignard reagents in ethereal solvents are not monomeric "RMgX" species floating independently; they exist in a dynamic Schlenk equilibrium,
and both sides of this equilibrium are further solvated and often dimeric or oligomeric, with ether oxygens coordinating to the electron-deficient, four-coordinate magnesium. This aggregation matters mechanistically: the same magnesium that carries the nucleophilic R group is also Lewis-acidic and coordinates the carbonyl oxygen lone pair of the electrophile before C–C bond formation, pre-organizing the transition state and making the addition intramolecular in character — a six-membered (or four-membered, in simplified textbook drawings) cyclic transition state rather than a naked bimolecular collision. Organolithium reagents behave analogously but are typically more aggregated (commonly tetrameric or hexameric in hydrocarbon solvents, breaking down to dimers or monomers in coordinating solvents like THF), and lithium's smaller ionic radius and higher charge density make RLi both more basic and more nucleophilic than the corresponding RMgX — useful when a sluggish ketone needs a harder-hitting reagent, but also more prone to side reactions like enolization or reduction via β-hydride transfer with hindered substrates.
Once the C–C bond forms, the tetrahedral alkoxide is bound tightly to magnesium (or lithium) as a covalent alkoxide salt — not yet an alcohol. Aqueous acidic workup (dilute HCl or, for base-sensitive products, saturated NH₄Cl) protonates the alkoxide oxygen and dissolves the magnesium salts, releasing the free alcohol into the organic layer.
The core ideas — the governing relationships
The reaction is best organized around a formal oxidation-state/electron-count bookkeeping rather than a single "equation," because the chemistry is a mechanism, not an algebraic law. Still, several relationships must hold exactly and are worth stating as governing rules.
Rule 1 — carbonyl carbon oxidation state. Nucleophilic addition converts an sp² carbonyl carbon (bonded to one O by a double bond) into an sp³ alkoxide carbon (bonded to O by a single bond, and now also to the new R group):
where: [M] = MgX or Li; the carbonyl carbon gains one new C–C σ bond and the π bond is fully consumed, converting C from sp² (3 σ + 1 π) to sp³ (4 σ).
Rule 2 — stoichiometry with esters and acid chlorides. Because the tetrahedral intermediate formed from the first addition to an ester carbon can expel the alkoxide leaving group (R″O⁻) to regenerate a new, more reactive ketone in situ, the reaction does not stop after one equivalent:
where: the intermediate ketone R′C(=O)R is always more electrophilic than the starting ester (esters are stabilized by resonance donation from the second oxygen), so the second addition outcompetes any attempt to isolate the ketone at room temperature with excess Grignard reagent. Two equivalents of the same R group therefore end up on the final carbinol carbon.
Rule 3 — count of new C–C bonds and required equivalents.
where: carboxylic acids are normally not good substrates for direct dialkylation because the first equivalent of Grignard reagent is consumed unproductively deprotonating the O–H (Brønsted acid–base reaction, faster than any addition step by many orders of magnitude), forming an unreactive magnesium carboxylate that requires a second equivalent just to begin nucleophilic addition, and even then the product commonly stops at the ketone stage under carefully controlled low-temperature/limited-equivalent protocols, though under standard excess-reagent conditions full addition to the tertiary alcohol is observed.
where: this 45+ unit gap in acidity is the single most important number in the entire chapter — it is the thermodynamic reason that any O–H, N–H, S–H, or terminal alkyne C–H (pKa ≈ 25) in a substrate will destroy a Grignard/organolithium reagent by simple proton transfer, far faster than any carbonyl addition can compete.
Two mechanisms in full curved-arrow detail
Goal: show, step by step, how 2-butanone (methyl ethyl ketone) plus phenylmagnesium bromide gives, after aqueous workup, 2-methyl-2-phenylbutan... more precisely 2-phenyl-2-butanol, a tertiary alcohol, tracking every electron pair and the exact role magnesium plays as a Lewis acid template.
Step 1 (pre-complexation). The carbonyl oxygen of 2-butanone, bearing two lone pairs, uses one lone pair to coordinate to the electrophilic magnesium center of PhMgBr (itself solvated by two THF or Et₂O molecules in the resting state — omitted here for clarity). This is a simple dative Lewis acid–base interaction: O lone pair → empty Mg orbital. No bonds break yet; this step only polarizes the C=O further, lowering the LUMO (π*) energy and raising the electrophilicity of the carbonyl carbon. Experimentally this pre-complexation is why Grignard additions proceed even in noncoordinating hydrocarbon solvents once the Mg is otherwise ligated, and why added Lewis acids (e.g., CeCl₃ in the Luche-type variants for organometallics) can accelerate sluggish additions.
Step 2 (nucleophilic C–C bond formation). With the carbonyl activated, the Ph–Mg σ bond electron pair swings from carbon into the carbonyl carbon along the Bürgi–Dunitz trajectory (~105° from the C=O axis, avoiding eclipsing interactions with the departing π-system and allowing best overlap with the developing sp³ lobe). Simultaneously, the π electrons of C=O flow onto oxygen. This is drawn as two curved arrows: one from the C–Mg bond to the carbonyl carbon, one from the C=O π bond to O. The result is a four- (or six-, counting the coordinated ether) membered cyclic transition state in which Mg is already tethered to the incipient alkoxide oxygen — this is why the reaction, despite forming two new formal charges' worth of separation in a naive Lewis-structure sense, has a modest activation barrier: charge is never fully separated because Mg–O bonding character grows in step with C–O π-bond loss.
Step 3 (collapse to tetrahedral alkoxide). The former sp² carbonyl carbon is now sp³, bonded to: the original methyl group, the original ethyl group, the new phenyl group, and the alkoxide oxygen (now bearing a negative formal charge, tightly ion-paired/covalently bonded to Mg²⁺Br⁻). This magnesium alkoxide, R₃C–O–MgBr, is configurationally a discrete, isolable (in principle) species — the reaction is complete on the carbon framework at this point; only the countercation remains to be exchanged.
Step 4 (aqueous acidic workup). Addition of dilute aqueous HCl (or, for acid-sensitive substrates, saturated aqueous NH₄Cl, a mild proton source that avoids side reactions like elimination) protonates the alkoxide oxygen: a lone pair on O attacks H of H₃O⁺, O–H bond forms, H–OH₂⁺ bond breaks, releasing water. Simultaneously, Mg²⁺ and Br⁻ dissolve into the aqueous phase as MgBrCl or MgCl₂ salts. The organic layer, separated and dried, contains the neutral tertiary alcohol, 2-phenyl-2-butanol, C₆H₅–C(CH₃)(C₂H₅)–OH.
Stereochemical note. Because the ketone carbon was planar (sp²) and attack can occur from either face with equal probability in this achiral substrate, the product is racemic — a 1:1 mixture of (R)- and (S)-2-phenyl-2-butanol, since the new stereocenter has four different groups (CH₃, C₂H₅, C₆H₅, OH) and no chiral influence was present anywhere in the reaction to bias face selection.
Go deeper · Derivation B · 3 minMechanism 2 — Double addition to an ester: methyl benzoate + 2 eq CH₃MgBr
Goal: derive explicitly why an ester delivers a tertiary alcohol with two identical new alkyl groups, by tracking the tetrahedral intermediate's collapse and the fate of the leaving group.
Step 1 (first addition, formation of tetrahedral alkoxide). Methylmagnesium bromide's C–Mg bond attacks the carbonyl carbon of methyl benzoate (PhCO₂CH₃) exactly as in Mechanism 1: one arrow from C–Mg to the carbonyl carbon, one arrow from the C=O π bond to oxygen. This generates a tetrahedral intermediate carbon bearing four different substituents: the phenyl ring, the newly added methyl group, the original methoxy group (–OCH₃), and the alkoxide oxygen (–O–MgBr).
Step 2 (collapse of the tetrahedral intermediate — the crucial branch point). Unlike the ketone case, this tetrahedral intermediate has a viable leaving group attached directly to the same carbon: the methoxide, –OCH₃, still tethered through oxygen. The alkoxide oxygen's lone pair reforms the π bond (arrow from O lone pair to C), pushing out the methoxide as a leaving group (arrow from the C–OCH₃ bond onto the departing oxygen, which takes both electrons as CH₃O⁻). This regenerates a carbonyl — but now it is a ketone, acetophenone (PhCOCH₃), not the original ester. This step is exactly analogous to the tetrahedral-intermediate collapse in ester nucleophilic acyl substitution chemistry (saponification, transesterification), except here the incoming nucleophile is a carbanion rather than hydroxide or alkoxide.
Step 3 (why the reaction does not stop at the ketone). The methoxide leaving group, once expelled, is itself sufficiently basic (pKa of CH₃OH ≈ 15.5) that under the reaction conditions it is simply protonated on workup or remains as an inert magnesium alkoxide spectator — it does not compete further. Meanwhile, the acetophenone generated in situ is a better electrophile than the original ester: esters are stabilized relative to ketones by roughly 15–20 kcal/mol of resonance donation from the second (ether) oxygen's lone pair into the carbonyl π system, an effect entirely absent once that oxygen has left as methoxide. Because excess CH₃MgBr (the second equivalent) is present in the same pot, this newly formed, more electrophilic ketone is attacked essentially as fast as it forms — faster, in fact, than unreacted ester is attacked in competing side pathways — so under standard conditions (excess RMgX, no attempt at low-temperature control) the ketone intermediate is never observed; it is consumed as rapidly as it is generated.
Step 4 (second addition). A second equivalent of CH₃–MgBr delivers its methyl group to the newly formed ketone carbon exactly as in Mechanism 1: C–Mg bond electrons attack the carbonyl carbon, π electrons flow to oxygen, giving a new tetrahedral alkoxide — now bearing phenyl, methyl, methyl (the two methyls indistinguishable), and O–MgBr.
Step 5 (workup). Aqueous acid protonates the alkoxide, releasing 2-phenyl-2-propanol (also called 2-phenylpropan-2-ol or, informally, "cumyl alcohol" without the extra methyl — precisely, PhC(CH₃)₂OH), a tertiary alcohol bearing two identical methyl groups installed by the two equivalents of the same Grignard reagent. This is the mechanistic reason ester + 2 RMgX always gives a tertiary alcohol with a pair of identical R groups flanking the original acyl carbon — a disconnection every retrosynthesis student must recognize instantly.
Limiting case check. If instead only 1.0 equivalent (rigorously controlled, low temperature, slow addition) of RMgX is used with an ester, and the more reactive intermediate ketone is trapped before further addition — a demanding but real experimental regime, more reliably achieved using Weinreb amides (R′C(=O)N(OMe)Me) which chelate the metal and block over-addition entirely — the ketone can be isolated cleanly. This is why Weinreb amides, not esters, are the standard tool when a chemist wants to stop cleanly at the ketone stage using an organometallic nucleophile.
Worked examples
Example 1 (foundational): PhMgBr + acetaldehyde → 1-phenylethanol. Acetaldehyde, CH₃CHO, is treated with one equivalent of phenylmagnesium bromide in dry diethyl ether at 0 °C, followed by aqueous NH₄Cl workup. Mechanistic walk-through. The carbonyl carbon of acetaldehyde bears one H and one CH₃ — it is less hindered and more electrophilic than a ketone carbonyl (only one alkyl group donating electron density to raise the LUMO, versus two in a ketone), so this addition is fast even at 0 °C. Phenylmagnesium bromide's ipso carbon (sp² aryl carbanion character, C–Mg bond) attacks the carbonyl carbon along the Bürgi–Dunitz trajectory; the π electrons flow to oxygen, giving the magnesium alkoxide CH₃CH(OMgBr)(C₆H₅). Aqueous acidic workup protonates this alkoxide. Product identification. The product is 1-phenylethanol, C₆H₅–CH(OH)–CH₃ — a secondary alcohol, exactly as predicted: aldehyde + RMgX → 2° alcohol. The new stereocenter (C bonded to Ph, CH₃, OH, H — four different groups) is formed with no facial bias (achiral aldehyde, achiral Grignard reagent), so the product is racemic, a 50:50 mixture of (R)- and (S)-1-phenylethanol. Sanity/limiting check. If the aldehyde is instead replaced by formaldehyde (H₂C=O, the parent case with two H's on the carbonyl carbon rather than one H and one alkyl), the same mechanism gives C₆H₅–CH₂–OH, benzyl alcohol — a primary alcohol, confirming Rule: formaldehyde is the unique carbonyl that yields a 1° alcohol from RMgX addition, because it is the only carbonyl compound with zero pre-existing carbon substituents on the carbonyl carbon.
Example 2 (applied): synthesis of 3-ethyl-3-pentanol from an ester with asymmetric, non-round stoichiometry check. A synthetic chemist wants 3-ethyl-3-pentanol, (C₂H₅)₃C–OH, a tertiary alcohol with three identical ethyl groups on the carbinol carbon. Working backward: a tertiary alcohol with three identical R groups (not two) cannot come from a ketone (which gives only two new identical groups plus whatever was on the ketone) — it must come from an ester or acid chloride where the acyl R group is itself identical to the Grignard's R group, so that all three substituents on the final carbon end up the same. Retrosynthesis. Disconnect 3-ethyl-3-pentanol as (C₂H₅)–C(OH)(C₂H₅)₂ → the acyl precursor must be ethyl propanoate's acyl fragment or, more directly, ethyl formate would give only two ethyls plus an H — insufficient. The correct precursor is ethyl propanoate, CH₃CH₂CO₂CH₂CH₃ (an ester whose acyl R group, ethyl, matches the Grignard's ethyl group), combined with 2 equivalents of ethylmagnesium bromide. Forward mechanism, briefly re-applied. First equivalent of EtMgBr adds to the ester carbonyl carbon, tetrahedral intermediate collapses, ejecting ethoxide (CH₃CH₂O⁻) to regenerate a ketone — specifically 3-pentanone, CH₃CH₂C(=O)CH₂CH₃ (diethyl ketone), formed transiently in situ. Second equivalent of EtMgBr adds to this ketone, giving the tetrahedral alkoxide (C₂H₅)₃C–OMgBr. Acidic workup delivers 3-ethyl-3-pentanol. Stoichiometric accounting (explicit, non-round numbers to verify the bookkeeping). Suppose a chemist runs this on 4.35 mmol of ethyl propanoate. Rule 2 demands exactly 2 equivalents of EtMgBr per equivalent of ester for complete, clean double addition — so the required amount is
If the EtMgBr solution is 0.87 M in THF, the required volume is
In practice a modest excess (commonly 2.2–2.5 equivalents) is used to compensate for adventitious moisture destroying some reagent by protonation, so a real-world charge might be 9.6–10.9 mmol (≈ 11.0–12.5 mL of the same solution) — illustrating why Grignard reactions are always run with rigorously anhydrous glassware and solvent, since even trace water competes stoichiometrically with the intended substrate.
Example 3 (exam/competition level): diastereoselective addition — Felkin–Anh prediction for a chiral aldehyde. Consider (S)-2-phenylpropanal, CH₃–CH(C₆H₅)–CHO, treated with methylmagnesium iodide. This substrate has a stereocenter adjacent to the carbonyl, so the two faces of the C=O are diastereotopic, not enantiotopic — the reagent will add preferentially to one face, giving unequal amounts of the two possible diastereomeric alcohols (a mixture, but not 50:50). Setting up the Felkin–Anh model. Around the C(carbonyl)–C(stereocenter) bond, we consider the Newman projection looking from the incoming nucleophile toward the carbonyl carbon (front) and the stereocenter (back). The three substituents on the back (stereocenter) carbon are ranked by size in this specific stereoelectronic model: the largest group (here, phenyl, since it is bulkier than methyl and much bulkier than H) is placed perpendicular to the carbonyl, anti to the trajectory of nucleophilic attack — this minimizes steric clash along the Bürgi–Dunitz approach angle (~105–107°) while simultaneously allowing the C–C(large) σ orbital to align antiperiplanar to the forming nucleophile–carbon bond, providing stabilizing σ(C–C)→π(C=O)-like hyperconjugative assistance (the electronic refinement of the original Felkin model, due to Anh and Eisenstein). Applying the trajectory. With phenyl held perpendicular and anti to the approaching nucleophile, the remaining two back-carbon substituents (CH₃ and H) occupy the two gauche positions relative to the carbonyl oxygen. The nucleophile (methyl from MeMgI) approaches past the smaller of these two gauche substituents — past H rather than past CH₃ — along the Bürgi–Dunitz trajectory on the face where steric interference is least. Predicted major product. This trajectory delivers the incoming methyl group to the face of the carbonyl that places the new OH and the pre-existing phenyl in a specific relative (anti, "Felkin") arrangement in the extended (zig-zag) chain — for this substrate the Felkin–Anh model predicts the anti (sometimes labeled 3,4-anti or "Felkin") diastereomer as major, typically in ratios around 3:1 to 5:1 for a modestly differentiated system like methyl/phenyl/H (exact ratios depend on solvent, temperature, and the specific reagent — organolithiums and cerium-modified Grignard reagents often sharpen this selectivity because tighter chelation-free ion pairing better enforces the open, non-chelated Felkin transition state). Limiting case check. If instead the adjacent stereocenter bears an α-alkoxy group capable of chelating the incoming Mg²⁺ or Li⁺ (e.g., an α-methoxy aldehyde), the mechanism changes qualitatively: the metal bridges both the carbonyl oxygen and the α-alkoxy oxygen in a five-membered chelate, locking the conformation and reversing the facial preference relative to the non-chelated Felkin–Anh case — this is the Cram chelation model, and recognizing which model applies (open Felkin–Anh vs. chelation-controlled) is a classic exam discriminator: the determining question is simply "does this substrate have a Lewis-basic group positioned to chelate the metal in a 5-membered ring?"
Limiting cases, scope & common traps
| Situation | What happens | Why (mechanistic reason) |
|---|---|---|
| Substrate has –OH, –NH, –CO₂H, terminal alkyne C–H | RMgX/RLi destroyed by acid–base reaction before any addition | pKa gap (∼25–50 vs. R–H) makes proton transfer overwhelmingly faster than addition |
| Ester or acid chloride + excess RMgX | Tertiary alcohol, two identical new R groups | Tetrahedral intermediate collapses, ejects leaving group, regenerates more electrophilic ketone in situ, consumed immediately |
| Ketone/aldehyde + 1 eq RMgX, no leaving group at carbonyl C | Clean 1 : 1 addition, stops at alcohol stage | No group on carbonyl carbon can leave (H or alkyl are not leaving groups) |
| Weinreb amide (R′CON(OMe)Me) + RMgX | Clean ketone, no over-addition | Chelated tetrahedral intermediate is stable until aqueous workup; collapses only then |
| α,β-unsaturated ketone + RLi or RMgX | Mostly 1,2-addition (direct carbonyl attack) | Hard nucleophile, kinetic preference for the more electrophilic, less hindered carbonyl carbon |
| α,β-unsaturated ketone + R₂CuLi (Gilman) | Mostly 1,4-(conjugate) addition | Softer, more polarizable copper-stabilized nucleophile favors the softer β-carbon electrophilic site |
| Epoxide + RMgX | Ring-opened alcohol, two carbons longer | SN2-like backside attack at less hindered epoxide carbon, C–O bond breaks as new C–C bond forms |
| Water/moist glassware present | Reagent partly or wholly quenched (RH formed) | Grignard/organolithium is simply a very strong base toward H₂O; competes stoichiometrically |
The chemist behind it
Origins · 1 minFrançois Auguste Victor Grignard and how this was discovered
Victor Grignard was a French chemist working under Philippe Barbier at the University of Lyon in the final years of the nineteenth century. Barbier had been attempting magnesium-mediated additions of alkyl halides to carbonyl compounds but found the reaction unreliable and difficult to reproduce; he handed the problem to his young assistant Grignard, who in 1900 discovered that pre-forming the organomagnesium halide in anhydrous diethyl ether — rather than attempting a one-pot combination of magnesium, halide, and carbonyl simultaneously — gave a clean, reproducible, and remarkably general reagent. This procedural refinement, seemingly small, is what transformed an erratic curiosity into the most reliable carbon-nucleophile method available to organic chemists for the next century. Grignard published his results in 1900 and defended his doctoral thesis on the subject in 1901; the reaction bears his name rather than Barbier's despite Barbier's earlier priority, a naming choice contemporaries attributed to the sheer scope and reliability Grignard demonstrated across dozens of substrates.
Why can't you just use NaOH or another simple base instead of forming a Grignard reagent to add carbon to a carbonyl?
Hydroxide and alkoxide bases are oxygen or nitrogen nucleophiles (or, at best, can generate stabilized carbanions like enolates from acidic C–H bonds), but they cannot deliver an unactivated alkyl or aryl carbon as a nucleophile because a simple alkane or arene C–H bond has a pKa near 45–50 — far too high for any conventional base to deprotonate under normal laboratory conditions. Grignard and organolithium reagents solve this by using a metal–carbon bond, formed directly from an alkyl or aryl halide via oxidative addition at the metal, to deliver carbanion-like reactivity without ever needing to deprotonate a C–H bond. This is precisely why organometallics, not simple bases, are required whenever the synthetic goal is to attach a genuinely unactivated alkyl or aryl group to a carbonyl carbon.
What happens if you try to make a Grignard reagent from a bromide that also has a ketone in the same molecule?
As magnesium metal begins inserting into a few of the C–Br bonds, the small amount of organomagnesium halide that forms will immediately find and attack the ketone carbonyl present on other molecules of the same starting material (or, in especially unfavorable cases, intramolecularly), consuming the reagent as fast as it forms and typically producing an oligomeric or polymeric tar rather than a clean, discrete organometallic solution. This self-condensation is why any substrate destined for Grignard-reagent formation must have all other electrophilic carbonyls (and all acidic protons) removed or protected first — the halide-bearing fragment and the electrophilic partner are normally kept as two entirely separate molecules until the moment of the intended addition.
Why does adding excess Grignard reagent to an ester give a tertiary alcohol with two identical groups, rather than stopping at the ketone?
The tetrahedral intermediate formed after the first addition to an ester has a leaving group (the alkoxy portion of the original ester) still attached to the same carbon, so it collapses by ejecting that group and regenerating a carbonyl — but the newly formed carbonyl is a ketone, which lacks the extra resonance stabilization from a second oxygen lone pair that made the ester less electrophilic in the first place. This freshly generated ketone is therefore more reactive toward the remaining excess Grignard reagent than the original ester was, so under standard conditions it is consumed by a second addition essentially as quickly as it is produced, and the reaction proceeds cleanly through to the tertiary alcohol without any ketone ever accumulating in solution.
How is a Weinreb amide different, and why does it stop at the ketone?
A Weinreb amide, R–C(=O)–N(OCH₃)(CH�3), forms a tetrahedral intermediate upon addition of one equivalent of RMgX or RLi in which the metal (Mg²⁺ or Li⁺) is chelated by both the newly formed alkoxide oxygen and the adjacent N–OCH₃ oxygen in a stable five-membered ring. This chelate keeps the tetrahedral intermediate from collapsing (expelling the amide nitrogen as a leaving group) until the chelate is broken during aqueous acidic workup, at which point the ketone is released cleanly, after the excess organometallic reagent has already been quenched. Because the collapse only happens after the nucleophile is gone, no second addition can occur, giving controlled, single-addition access to ketones from an acyl-level precursor — something ordinary esters cannot achieve with organometallic nucleophiles.
Can a Grignard reagent add to a nitrile?
Yes: the C≡N triple bond is attacked by the organometallic nucleophile in a manner mechanistically parallel to carbonyl addition — one arrow from the C–Mg bond to the nitrile carbon, one arrow from a C≡N π bond to nitrogen — giving a metalated imine (an iminate salt, R′C(=NMgX)R). This nitrogen-stabilized intermediate is generally more resistant to a second addition than an ester-derived ketone intermediate would be, so nitriles typically stop cleanly at the imine stage under the reaction conditions and, crucially, only proceed to a ketone (not a tertiary amine or further addition) upon subsequent aqueous acidic hydrolysis of the resulting imine — this is a standard and reliable method for ketone synthesis from a nitrile precursor.
Why is diethyl ether or THF used as solvent instead of, say, dichloromethane or acetone?
Ethereal solvents provide two oxygen lone pairs per molecule that coordinate to the electron-deficient magnesium or lithium center, both stabilizing the organometallic against decomposition and, crucially, solubilizing it in solution (bare "RMgX" without solvation is a poorly soluble, highly aggregated species). Dichloromethane and other halogenated solvents are unusable because their C–halogen bonds are themselves electrophilic and would simply react with the very organometallic reagent you are trying to use, while a solvent like acetone contains its own reactive carbonyl group and would be consumed instantly as an unwanted electrophile competing with the intended substrate.
Does the stereochemistry at the new carbinol carbon matter if the starting material is achiral?
When both the organometallic reagent and the carbonyl substrate are achiral (or, more precisely, when neither introduces any pre-existing chirality near the reacting carbon), the new sp³ stereocenter formed at the carbinol carbon is generated with exactly equal probability of either face being attacked, because the transition states leading to the two enantiomeric products are mirror images of each other and therefore isoenergetic. The result is necessarily a racemic mixture, 50:50, regardless of reaction time, temperature, or reagent excess — only the presence of a pre-existing stereocenter (Felkin–Anh/chelation control) or an external chiral catalyst/ligand can break this symmetry and produce an enantioenriched alcohol.