Acidity of Alcohols and the Alkoxide Ion

Alkoxide generation is the opening move of an enormous fraction of practical organic synthesis. Sodium ethoxide in ethanol is the classic base for Claisen condensations and malonic ester alkylations in undergraduate and industrial synthesis alike. Sodium methoxide is produced on a multi-ton industrial scale as a catalyst and reagent for transesterification — most famously in biodiesel production, where triglycerides are converted to fatty acid methyl esters via methoxide-catalyzed transesterification with methanol. Potassium tert-butoxide, a bulky, poorly-nucleophilic but strongly basic alkoxide, is a workhorse reagent in E2 eliminations (favoring the less-substituted Hofmann alkene precisely because its bulk disfavors approach to the more hindered proton) and in Wittig-type and other carbanion-generating reactions. In pharmaceutical process chemistry, controlled, quantitative alkoxide formation (using NaH or KOtBu, chosen by exactly the pKa-difference reasoning developed here) underlies large-scale Williamson ether syntheses used to install ether linkages in active pharmaceutical ingredients — for example, the aryl ether linkages found in many kinase inhibitors and antihistamines are installed by generating a phenoxide or alkoxide and displacing a leaving group on an alkyl halide or mesylate. In biochemistry, serine and threonine hydroxyls in enzyme active sites are deprotonated in analogous fashion by active-site bases (often a histidine acting as a general base, or assisted by a metal ion polarizing the O–H bond), a mechanistic parallel to the alkoxide-formation chemistry here that underlies catalytic triads in serine proteases. Even something as mundane as the choice of drying agent or base in a reaction flask — should you use NaH or K₂CO₃ to deprotonate a phenol before an alkylation? — is answered instantly by the same pKa-difference calculation performed throughout this article.

Ask the tutor: How exactly does solvation reverse the gas-phase acidity order of methanol, ethanol, and tert-butanol when they're dissolved in water?

An alcohol R–O–H has an oxygen bearing two lone pairs, a σ-bond to carbon, and a σ-bond to hydrogen. The O–H bond is polarized toward oxygen (electronegativity of O ≈ 3.44 vs. H ≈ 2.20), so oxygen already carries partial negative character in the neutral molecule. Acting as a Brønsted acid means a base's lone pair attacks the O–H hydrogen, and the O–H bonding electron pair collapses entirely onto oxygen. The product is an alkoxide ion, RO⁻, isoelectronic in its oxygen environment with a fluoride-like center: three lone pairs (formally) and a single bond to carbon, net charge −1. The key mechanistic and energetic question is: how does that lone-pair-rich, negatively charged oxygen compare, in stability, to the neutral O–H oxygen it came from? Three effects dominate the comparison, and understanding all three is what lets you predict and rank pKa values rather than memorize them.

Inductive effect. Any electron-withdrawing group near the O–H (a halogen, a carbonyl, another oxygen) pulls electron density away through the σ-framework, stabilizing the developing negative charge in the transition state and in the alkoxide product. This is why 2,2,2-trifluoroethanol (pKa ≈ 12.4) is dramatically more acidic than ethanol (pKa ≈ 16), even though both are simple primary alcohols — the three fluorines are not conjugated to oxygen at all, they act purely through space and through σ-bonds.

Solvation. In protic solvent (water, or the alcohol itself acting as solvent), the alkoxide oxygen is stabilized by hydrogen bonds donated to it by solvent O–H groups. A small, unhindered alkoxide like methoxide can be surrounded efficiently; a bulky, branched alkoxide like tert-butoxide cannot pack solvent molecules around its oxygen as effectively, and it is measurably less well solvated. This solvation effect is large enough that it can dominate over intrinsic (gas-phase) electronic effects, which is why the solution-phase acidity order of simple alcohols in water is

(with pKa rising through that series) — the opposite of the gas-phase order, where alkyl branching, by hyperconjugative electron donation into the developing anion, actually stabilizes the alkoxide (more on this inversion under substituent effects; here it suffices to fix the solution trend, which is what matters for essentially all synthetic chemistry).

Hybridization and orbital character of the lone pairs. Comparing an alcohol's O–H acidity to a completely different X–H bond (say, a terminal alkyne C–H, pKa ~25, or ammonia's N–H, pKa ~38) is a story about which atom holds the negative charge and in what kind of orbital. Oxygen is more electronegative than carbon or nitrogen, so a lone pair sitting in an oxygen-centered orbital is intrinsically lower in energy (better stabilized) than the same electron density on carbon or nitrogen. This is the primary reason alcohols (pKa 16–18) are so much more acidic than amines, despite the superficial similarity of "heteroatom–H bond."

The acidity of an alcohol is defined by the standard acid dissociation equilibrium in water,

with equilibrium constant

where:

• $[\text{RO}^-]$, $[\text{H}_3\text{O}^+]$, $[\text{ROH}]$ are equilibrium molar concentrations,
• water, the bulk solvent, is omitted from the expression (its concentration ≈ 55.5 M is folded into $K_a$ by convention).

The pKa is simply

and the free energy of deprotonation follows from

where:

• $R = 8.314\ \text{J mol}^{-1}\text{K}^{-1}$,
• $T$ is absolute temperature (K),
• $2.303\,RT \approx 5.71\ \text{kJ mol}^{-1}$ (1.36 kcal/mol) per pKa unit at 25 °C.

For any acid–base proton transfer between an acid HA and a base B⁻ (conjugate base of acid HB),

where:

• $\text{p}K_a(\text{HA})$ is the pKa of the acid being deprotonated (the alcohol, here),
• $\text{p}K_a(\text{HB})$ is the pKa of the conjugate acid of the base doing the deprotonating,
• the equilibrium favors products (alkoxide) whenever $\text{p}K_a(\text{HB}) > \text{p}K_a(\text{HA})$, i.e., whenever the base's conjugate acid is a weaker acid than the alcohol.

Reading this equation: it says the position of any proton-transfer equilibrium is controlled by nothing but the difference in pKa values of the two acids involved, and every 1-unit difference shifts the equilibrium constant by a full power of ten. This single relationship is the entire quantitative machinery you need to decide, instantly, whether a given base will deprotonate a given alcohol to a synthetically useful extent.

Example 1 — Foundational: equilibrium constant for ethoxide formation from ethanol and hydroxide. Ethanol has pKa ≈ 16.0; water (the conjugate acid of hydroxide) has pKa = 15.7. The reaction is

Apply the pKa-difference rule with HA = ethanol and HB = water:

Interpretation: only about a third of the ethanol present is converted to ethoxide at equilibrium (since $K_{eq}=[\text{EtO}^-][\text{H}_2\text{O}]/([\text{EtOH}][\text{OH}^-])\approx 0.5$, roughly a 1:2 ratio of products to starting acid-base pair, meaning ethoxide is a minority species). This confirms mechanistically why hydroxide is not the reagent of choice when a clean, high-yield alkoxide solution is required — the equilibrium sits close to 1, not driven to completion. Limiting case check: if the acid and base had identical pKa (ΔpKa = 0), $K_{eq}=10^0=1$, an exactly 50:50-type equilibrium (statistically), consistent with the definition of pKa as the point of half-dissociation — this is the correct limiting behavior of the formula.

Example 2 — Applied: choosing a base to deprotonate cyclohexanol quantitatively before a Williamson ether synthesis. Cyclohexanol (a secondary alcohol, pKa ≈ 18, slightly higher than ethanol due to the extra alkyl branching's effect on solvation, foreshadowed above) must be converted essentially completely to cyclohexoxide before alkylation with an alkyl halide (the classic Williamson Ether Synthesis setup, where the alkoxide performs an SN2 displacement). Compare two candidate bases: (a) Triethylamine (conjugate acid Et₃NH⁺, pKa ≈ 10.7):

Essentially no deprotonation occurs — triethylamine is far too weak a base for this substrate, and using it would leave the alcohol untouched, killing the subsequent SN2 step entirely. (b) Sodium hydride (conjugate acid H₂, pKa ≈ 36):

Overwhelmingly favorable — NaH cleanly and (for practical purposes) completely deprotonates cyclohexanol, evolving H₂ gas as visible confirmation, and delivering a clean solution of sodium cyclohexoxide ready for SN2 alkylation. Limiting case check: as the base gets weaker and weaker (pKa of its conjugate acid → −∞, e.g., a very strong acid's conjugate base), $K_{eq}\to 0$ — no deprotonation, matching physical intuition that a strong acid's conjugate base cannot deprotonate anything appreciably.

Example 3 — Exam/competition level: quantitative comparison of two competing deprotonation sites with a single equivalent of base. A substrate contains both a free alcohol O–H (pKa 17.0, an unusually hindered secondary neopentyl-type alcohol) and an alkynyl C–H (pKa 25.0) in the same molecule (e.g., a propargylic alcohol such as HC≡C–CH(OH)–C(CH₃)₃). One equivalent of NaNH₂ (conjugate acid NH₃, pKa 38) is added. Which proton is removed, and by how much is that site thermodynamically favored over the other? Compute $K_{eq}$ for each possible deprotonation: Alcohol O–H removal:

Alkyne C–H removal:

Both are, in isolation, essentially irreversible with NaNH₂ as base — this reagent is strong enough to deprotonate either site completely. The question of which site reacts first, with a single equivalent, is answered by comparing the relative rates/equilibrium preference directly:

The alcohol O–H is deprotonated preferentially, by a factor of 10⁸ in thermodynamic driving force, because it is the more acidic proton (lower pKa) in the molecule. Mechanistically, this also matches expectation from orbital reasoning: the O–H proton is on a small, sterically accessible, highly electronegative oxygen, while removing the C–H proton requires forming a much higher-energy, less-stabilized alkynyl carbanion (only partially compensated by the sp-hybridized carbanion's extra s-character). Practically, this means a chemist who wants to deprotonate only the alkyne (to do a subsequent alkylation at that carbanion, for instance) must first protect the alcohol, or use exactly one equivalent of a base weak enough to leave the alkyne alone but strong enough to take the O–H (e.g., NaH, pKa(H₂)=36, gives $K_{eq}^{(\text{CH})}=10^{36-25}=10^{11}$ still favorable for the alkyne too — so in practice excess NaH would deprotonate both; a single controlled equivalent is titrated to the more acidic O–H site first, consuming the base there before any alkyne deprotonation becomes kinetically significant).

Quantitative acid–base thermodynamics, and the very definition of pKa used throughout this article, traces back to Søren Sørensen's introduction of the pH scale (1909) at the Carlsberg Laboratory, and to the broader physical-organic framework built by Louis Hammett, whose 1930s–40s work systematized how substituents quantitatively shift acid strength — the direct conceptual ancestor of every "electron-withdrawing group lowers pKa" argument used here and expanded further when treating substituent effects on alcohol acidity.