Anhydrides: Formation and Acylating Power

Acetic anhydride remains one of the highest-volume acylating reagents in the chemical industry, used in the manufacture of cellulose acetate (photographic film base, cigarette filter tow, textile fibers) and in the industrial synthesis of aspirin (acetylsalicylic acid) by O-acetylation of salicylic acid — a textbook nucleophilic acyl substitution on an anhydride run at enormous scale. Maleic anhydride, a cyclic unsaturated anhydride, is a monomer workhorse in polyester resin and copolymer manufacture, its ring-opening acylation chemistry exploited both for polymer cross-linking and for Diels–Alder dienophile chemistry in materials science. In peptide and medicinal chemistry, the mixed carbonic-anhydride method (isobutyl chloroformate/N-methylmorpholine) is still taught and occasionally used for solution-phase fragment coupling, valued for its operational simplicity, low epimerization risk when activation and coupling are done quickly at low temperature, and the mild, volatile byproducts (CO₂, isobutanol) that simplify purification — a direct descendant of the chemoselectivity logic worked through in Example 3. Mixed anhydrides more generally underpin the classical Yamaguchi esterification (using 2,4,6-trichlorobenzoyl chloride to form a bulky mixed anhydride with a carboxylic acid, which is then attacked selectively by an alcohol with DMAP catalysis) — a mainstay of complex natural-product macrolactonization, precisely because the trichlorobenzoate "carrier" group is both extremely hindered and forms an extremely stable, easily expelled carboxylate, guaranteeing that only the desired acyl group is transferred to the alcohol nucleophile even in highly functionalized, sterically congested substrates. Cyclic anhydrides (succinic, phthalic, glutaric) remain standard reagents for reversible protein/amine modification (e.g., citraconic anhydride for reversible lysine masking in protein chemistry) and for introducing a carboxylic acid handle plus an amide/ester in a single ring-opening step, a strategy used throughout bioconjugation and polymer functionalization chemistry.

Ask the tutor: In the Yamaguchi esterification, why does using 2,4,6-trichlorobenzoyl chloride as the "carrier" acyl group give even higher chemoselectivity for alcohol acylation than pivaloyl chloride does?

An anhydride's central architecture is two carbonyls flanking a single bridging oxygen: R–C(=O)–O–C(=O)–R′. That bridging oxygen has two lone pairs, and each is anti-periplanar-ish to (better: can conjugate into) both adjacent carbonyl π systems, but geometrically it can only donate substantially into one carbonyl at a time because the two acyl planes twist relative to each other to minimize dipole–dipole and steric clash between the two carbonyl oxygens. This partial, shared conjugation is weaker per carbonyl than the analogous donation of an ether or amide nitrogen lone pair into a single carbonyl, which is precisely why each carbonyl of an anhydride remains a good electrophile — better than an ester carbonyl, worse than an acid chloride carbonyl. Compare the three we already know from the ladder. In an acid chloride, chlorine's lone pairs conjugate weakly (poor orbital overlap, electronegative Cl holds its lone pairs tightly) — the carbonyl carbon is left highly electrophilic, and chloride, once expelled, is an outstanding, resonance-uninvolved leaving group. In an ester, the alkoxy oxygen donates its lone pair well (good energy match, no competition), pushing electron density onto the carbonyl oxygen and deactivating* the carbonyl carbon substantially; alkoxide is also a poor leaving group (high pKaH of the conjugate alcohol, ~16–18) — so esters are unreactive both ways. An anhydride splits the difference: the bridging oxygen donates into one carbonyl only partially (it must "share" its conjugation, and geometry forces the two acyl groups out of full coplanarity), so the carbon remains moderately electrophilic; the leaving carboxylate, once expelled, is resonance-stabilized (pKaH of a carboxylic acid ~4–5, far more acidic/stabilized than an alcohol) and hence a much better leaving group than alkoxide, though still not as unconditionally good as chloride. The consequence: anhydrides react with nucleophiles fast enough to be synthetically useful acylating agents, but slow enough, and with mild enough byproducts (a carboxylic acid, not HCl), that they are compatible with more sensitive substrates and can be handled/stored without violent hydrolysis in ambient moisture (though they are certainly moisture-sensitive over time).

The governing relationship is again the reactivity-ladder logic applied to a specific leaving-group comparison. The rate of nucleophilic acyl substitution correlates with leaving-group ability, which correlates inversely with the basicity (pKaH) of the leaving group once expelled:

where:

• LG⁻ is the leaving group expelled from the tetrahedral intermediate (Cl⁻, RCO₂⁻, or RO⁻),
• pKaH(LG–H) is the pKa of the conjugate acid of that leaving group (HCl ≈ −7, RCO₂H ≈ 4–5, ROH ≈ 16–18),
• lower pKaH ⇒ more stable, less basic anion ⇒ faster expulsion from the tetrahedral intermediate.

For anhydride formation itself, the reaction is an equilibrium condensation whose position is controlled by removing water (or by using a large excess of dehydrating agent), captured by:

where:

• the forward reaction is disfavored thermodynamically at equilibrium (anhydride bond formation from two acids is only mildly exothermic/often endergonic under standard aqueous conditions),
• practical anhydride synthesis therefore avoids this equilibrium entirely, going instead through an acid chloride or a mixed anhydride with an irreversible driving force (e.g., loss of CO₂ and HCl from a chloroformate, or formation of a very stable byproduct like dicyclohexylurea).

For mixed anhydrides, the chemoselectivity of nucleophilic attack — which of the two nonequivalent carbonyls reacts — is governed by a competition of electrophilicity (rate of nucleophilic attack) versus the stability of the two possible tetrahedral-intermediate collapse pathways:

where:

• C₁, C₂ are the two distinct carbonyl carbons of the mixed anhydride (e.g., the acyl carbon vs. the alkoxycarbonyl carbon of a mixed carbonic anhydride),
• ΔΔG‡ is the difference in activation energy for nucleophilic attack at each site, dominated by sterics (bulkier R disfavors attack there) and by how well each acyl group's leaving carboxylate is stabilized (the worse a potential leaving group is, the more that carbonyl is disfavored as the site of attack, because collapse would have to expel it),
• in practice, for mixed carboxylic–carbonic anhydrides (R–CO–O–CO–OR′, from chloroformates), nucleophiles attack the carboxylic carbonyl almost exclusively, because collapse there expels the excellent, resonance-delocalized, easily lost carbonate/alkyl carbonate leaving group (which further fragments to CO₂ + RO⁻), an essentially irreversible sink.

MECHANISM 1 — Synthesis of a symmetrical anhydride from an acid chloride and a carboxylate salt

Goal: convert an acid chloride and the sodium (or pyridinium) salt of the same carboxylic acid into the symmetrical anhydride, with chloride as the thermodynamic sink.

Step 1 — Deprotonation/salt formation. The carboxylic acid RCO₂H is first converted to its carboxylate RCO₂⁻ (using aqueous NaOH, or simply using excess pyridine as an in-situ base during the reaction). The carboxylate oxygen's lone pairs are now fully delocalized over both oxygens, making it a competent, reasonably soft nucleophile at oxygen.

Step 2 — Nucleophilic attack of carboxylate oxygen on the acid chloride carbonyl. One oxygen lone pair of RCO₂⁻ attacks the electrophilic carbonyl carbon of R–C(=O)–Cl. The carbonyl π bond breaks, pushing electron density onto the carbonyl oxygen, generating a tetrahedral alkoxide intermediate bearing: the original R group, the newly bonded O–C(=O)R (from the carboxylate), Cl, and O⁻.

Step 3 — Collapse of the tetrahedral intermediate, chloride expelled. The alkoxide lone pair re-forms the C=O π bond, and in a concerted curved-arrow step the C–Cl bond breaks heterolytically, chloride departing as the leaving group (pKaH of HCl ≈ −7, an outstanding leaving group, far better than the carboxylate that just attacked). This regenerates a full carbonyl and produces the anhydride R–C(=O)–O–C(=O)–R along with Cl⁻.

Step 4 — Irreversibility. Because chloride is such a poor nucleophile (delocalized negative charge over a large, polarizable ion, weak conjugate base) and such a stable, low-energy anion, the reverse reaction (chloride attacking the anhydride to regenerate acid chloride) is negligible under the reaction conditions. The equilibrium lies overwhelmingly toward the anhydride, unlike the direct acid + acid condensation, which is why this route (via the acid chloride) is preferred over trying to dehydrate two carboxylic acids directly. Net transformation:

MECHANISM 2 — Nucleophilic acyl substitution of an anhydride by an amine (acylation), with mechanistic account of chemoselectivity in a mixed anhydride

Goal: show, arrow by arrow, how a primary amine acylates using a mixed anhydride formed from acetic acid and pivalic acid — commonly used to selectively deliver the acetyl, not the pivaloyl, group — and explain why.

Step 0 — Setting up the mixed anhydride. Acetic acid is converted to its mixed anhydride with pivalic acid (2,2-dimethylpropanoic acid) via pivaloyl chloride + sodium acetate (mechanism identical to MECHANISM 1, but with two different acids): CH₃CO–O–CO–C(CH₃)₃.

Step 1 — Amine lone pair attacks the less hindered, more electrophilic carbonyl. The amine's nitrogen lone pair (sp³, HOMO) attacks the carbon of the acetyl carbonyl rather than the pivaloyl carbonyl. Two reinforcing reasons converge: (a) sterics — the pivaloyl carbon is flanked by a bulky tert-butyl group that sterically blocks nucleophilic approach along the Bürgi–Dunitz trajectory, while the acetyl carbon (flanked only by a methyl) is comparatively open; (b) electronics — the tetrahedral intermediate formed by attack at the acetyl carbon, upon collapse, must expel pivalate, which is a perfectly good, resonance-stabilized carboxylate leaving group, so this pathway is both kinetically (sterically) and thermodynamically reasonable.

Step 2 — Tetrahedral intermediate formation. As the N lone pair attacks, the acetyl C=O π bond breaks, generating an alkoxide oxygen on the now-tetrahedral carbon, which bears: CH₃, the incoming NR′H₂⁺ (ammonium, positively charged nitrogen with N–H available for proton transfer), O⁻, and the O–CO–C(CH₃)₃ (pivalate ester oxygen) leaving-group arm.

Step 3 — Proton transfer and collapse. A proton is transferred (often intermolecularly, mediated by a second equivalent of amine or by solvent) from the ammonium nitrogen to a basic site, restoring neutral nitrogen while the alkoxide oxygen's lone pair re-forms the π bond. In the same electronic event (or immediately following), the C–O bond to the pivalate oxygen breaks heterolytically; the electrons go fully onto that oxygen, generating free pivalate ion, (CH₃)₃C–CO₂⁻, and regenerating the carbonyl of the new amide product, CH₃–C(=O)–NHR′.

Step 4 — Why not the other regiochemistry? Had the amine instead attacked the pivaloyl carbonyl, the tetrahedral intermediate there would be severely destabilized by A^1,3-like steric strain from the tert-butyl group crowding the incoming nucleophile and the departing acetate arm simultaneously; in addition, collapse would expel acetate, which is a fine leaving group too, so there is no strong electronic bias favoring that channel — the steric argument alone dominates, and acetyl transfer wins overwhelmingly. This is exactly the design principle exploited industrially and in the lab: pair a small, "sacrificial" acyl group you want delivered with a bulky, non-transferable "carrier" acyl group (pivaloyl, or in the Vaughan mixed-anhydride peptide method, an isobutyl carbonate) so that only the desired group is transferred, with high selectivity, and mild carboxylate/carbonate byproducts rather than harsh acids. Net transformation:

Example 1 (foundational): predicting the product and byproduct of acetic anhydride + ethanol. Acetic anhydride, (CH₃CO)₂O, reacts with ethanol under mild acid or base catalysis (or even slowly uncatalyzed) to give ethyl acetate and acetic acid. Mechanism. Ethanol's oxygen lone pair attacks one of the two (electronically and sterically equivalent) carbonyl carbons of the anhydride. The carbonyl π bond breaks onto the carbonyl oxygen, forming a tetrahedral alkoxide intermediate bearing: CH₃ (from that acyl group), O⁻, OEt (protonated, or deprotonated depending on catalysis), and the O–COCH₃ leaving arm. A proton transfer from the incipient oxocarbenium-adjacent OEt–H⁺ to a base (or the leaving carboxylate itself acting as an internal base, in the uncatalyzed pathway) precedes or accompanies collapse: the alkoxide reforms the π bond, and the C–O bond to the acetate arm breaks, releasing acetate/acetic acid and generating ethyl acetate, CH₃CO–OEt. Because the two acyl groups in acetic anhydride are identical, there is no chemoselectivity question — either carbonyl gives the same product. This example fixes the basic four-step choreography (attack, tetrahedral intermediate, proton transfer, collapse) before we add the complication of distinguishing two different acyl groups.

Limiting case check: if ethanol is replaced by water (the "nucleophile" most commonly encountered adventitiously), the identical mechanism gives back two equivalents of acetic acid — i.e., simple hydrolysis, confirming that the anhydride's inherent electrophilicity, not some special property of alcohols, drives the reaction; hydrolysis is simply slower than with acetyl chloride because acetate is a poorer leaving group than chloride but faster than ester solvolysis because acetate is still a much better leaving group than alkoxide.

Example 2 (applied): selective mono-acylation using a mixed anhydride in a real synthetic step — acetylation of a hindered secondary alcohol in the presence of an unprotected primary amine, using acetic–formic mixed anhydride. Acetic–formic anhydride, HCO–O–COCH�3, is a classic mild formylating agent (it is unstable to isolate as pure acetic anhydride equivalent, and is generated in situ from acetic anhydride and formic acid). Suppose a synthetic intermediate contains both a primary amine and a secondary alcohol, and you want to formylate the amine selectively (a common protecting/masking strategy) using this mixed anhydride, leaving the alcohol untouched. Mechanistic reasoning for chemoselectivity between the two carbonyls. The mixed anhydride has two distinct carbonyls: the formyl carbon (H–C=O, essentially unhindered, and whose leaving-group arm upon attack there would be acetate) and the acetyl carbon (CH₃–C=O, whose leaving-group arm upon attack there would be formate). Nucleophiles overwhelmingly attack the formyl carbon: it is far less sterically hindered (H versus CH₃ flanking the carbonyl) and formate, once expelled, is only a marginally different leaving group in basicity from acetate (pKa of formic acid 3.75 vs. acetic acid 4.76, so formate is if anything the better leaving group, reinforcing the same regiochemical outcome). Both steric and electronic factors point the same way: amine attacks the formyl carbon. Mechanistic reasoning for chemoselectivity between the amine and the alcohol. Given attack occurs at the formyl carbon, we then ask which nucleophile in the substrate reacts: the primary amine's nitrogen lone pair is both more nucleophilic (nitrogen is less electronegative than oxygen, holds its lone pair more loosely, and is a poorer match for the hard, small proton grip that keeps an alcohol's lone pair more tightly held — in Pearson's soft/hard language, nitrogen is a better match for a moderately soft acyl electrophile) and present in higher effective local concentration if it is unhindered, so under carefully controlled substoichiometric anhydride and low temperature, the amine reacts selectively, transferring the formyl group to give a formamide, while the secondary alcohol remains an alcohol.

This is exactly the logic exploited in real total synthesis: mixed anhydrides let you exploit small differences in steric bulk and leaving-group pKa to achieve chemoselective acylation without protecting every other functional group first.

Example 3 (exam/competition level): quantitative estimate of relative acylation rates and full mechanism for the Vaughan (isobutyl chloroformate) mixed-anhydride peptide coupling. In classical mixed-anhydride peptide synthesis, an N-protected amino acid (Boc-Ala-OH, say) is first converted to its mixed carbonic anhydride using isobutyl chloroformate (iBuO–CO–Cl) and a tertiary amine base (N-methylmorpholine, NMM):

Mechanism of anhydride formation. NMM deprotonates the carboxylic acid (or acts as a nucleophilic catalyst first attacking the chloroformate — either pathway converges); the carboxylate oxygen then attacks the electrophilic carbonyl of the chloroformate, displacing chloride exactly as in MECHANISM 1, giving the mixed carbonic–carboxylic anhydride Boc-Ala-CO–O–CO–OiBu. Mechanism and regiochemistry of aminolysis by the incoming amino acid ester (H-Gly-OMe). The mixed anhydride has two very different carbonyls: the acyl carbonyl (from the amino acid, flanked by the Boc-Ala side chain) and the alkoxycarbonyl (carbonate-type) carbonyl (flanked by OiBu). Nucleophilic attack occurs almost exclusively at the acyl carbonyl, for a decisive electronic reason: collapse of the tetrahedral intermediate formed by attack at the acyl carbon expels the carbonate-type leaving group, iBuO–CO–O⁻, an alkyl carbonate anion, which is exceptionally well stabilized (it is essentially a "super-carboxylate," delocalized over three oxygens through the carbonate framework) and which, once formed, decomposes irreversibly to CO₂ and isobutoxide/isobutanol — an irreversible sink that pulls the equilibrium of the entire collapse step strongly forward. Attack at the alkoxycarbonyl carbon, by contrast, would have to expel the amino-acid carboxylate, a normal (not specially stabilized) leaving group, and is additionally disfavored because the alkoxycarbonyl carbon is a poorer electrophile (the extra alkoxy oxygen donates into that carbonyl more effectively than a simple acyl carbon's substituent, partially deactivating it, similar to how esters are less electrophilic than acyl carbons flanked by carbon substituents). Quantitatively, the selectivity for reaction at the desired (amino acid) carbonyl in these mixed carbonic anhydrides is reported in the synthetic literature to exceed 100:1 in typical peptide couplings — consistent with an estimated ΔΔG‡ on the order of

favoring attack at the amino-acid carbonyl (a modest but decisive few kJ/mol difference, entirely reasonable for a combination of a mild steric preference and a leaving-group stabilization difference of this size). Full product. Aminolysis at the correct carbonyl gives the dipeptide Boc-Ala-Gly-OMe plus isobutyl carbonic acid monoester, which spontaneously decarboxylates:

Limiting-case check: if the "R′" carrier group on the chloroformate were replaced by something with negligible steric bulk and a leaving group of comparable stabilization to the amino acid carboxylate (an artificial, non-physical limit), the chemoselectivity would collapse toward statistical (∼50:50) attack at either carbonyl — confirming that it is precisely the combination of superior carbonate leaving-group stabilization and deliberate design of the chloroformate partner that gives the observed high selectivity, not some intrinsic magic of "anhydrides."

Anhydride chemistry as a deliberate synthetic tool is inseparable from the industrial and academic history of acetic anhydride itself, one of the highest-tonnage organic intermediates ever made (feeding into cellulose acetate, aspirin manufacture, and countless acylations). But the figure most worth profiling for mechanistic insight into acyl transfer chemistry, and whose kinetic framework underlies how we now reason about leaving-group ability in anhydrides, esters, and amides alike, is Christopher Kelk Ingold — the physical organic chemist who, together with contemporaries, established the electronic (inductive/resonance) vocabulary and the systematic kinetic study of nucleophilic acyl substitution that lets us rank leaving groups quantitatively rather than qualitatively.