Structure and Hydrogen Bonding in Alcohols

Hydrogen bonding rooted in alcohol-type O–H groups is not a museum-piece concept; it is operative constantly in modern chemistry, biology, and industry. Ethanol–water azeotrope behavior (the reason 190-proof, not 200-proof, ethanol comes off a simple distillation column) traces directly to the hydrogen-bonded network reorganization that occurs when ethanol and water mix, altering the effective vapor pressure relationship from Raoult's-law ideality. Formulation chemists choosing solvents for oral drugs routinely use the same donor/acceptor/chain-length reasoning worked out above (via computed logP and hydrogen-bond-donor/acceptor counts, as codified in Lipinski's Rule of Five for drug-likeness) to predict whether a candidate molecule will dissolve well enough in the gut to be absorbed. Cryoprotectants such as glycerol and ethylene glycol exploit exactly the polyol hydrogen-bonding density discussed in Example 3 to disrupt ice-crystal formation in frozen biological tissue and in antifreeze formulations. Carbohydrates — polyols writ large — owe their water solubility, their syrupy viscosity, and their ability to form the hydrogen-bonded networks of cellulose fibers (giving wood and paper their strength) to the identical chemistry of exposed, polarized O–H groups clustered densely on a carbon backbone. And in supramolecular and materials chemistry, chemists deliberately design molecules bearing arrays of alcohol or alcohol-like hydrogen-bond donors and acceptors to self-assemble into predictable solid-state architectures, hydrogels, and liquid crystals — a direct engineering application of the same sp³-oxygen, polarized-O–H picture developed here. This same electron-rich oxygen is also the reactive center that will later be deprotonated to form the nucleophilic alkoxide ion (see Acidity of Alcohols and the Alkoxide Ion) or protonated under acid catalysis to become a superb leaving group (see Basicity and Protonation of Alcohols) — hydrogen bonding, acidity, and basicity are three faces of the identical lone-pair-bearing, polarized oxygen center.

Ask the tutor: How does this same polarized, sp³ oxygen become a strong acid or a strong base depending on what it reacts with?

Start with the oxygen atom itself. Free oxygen has electron configuration $1s^2 2s^2 2p^4$: two paired electrons in one 2p orbital, one electron each in the other two. When oxygen bonds to carbon and hydrogen in an alcohol, the standard valence-bond picture mixes the 2s and three 2p orbitals into four sp³ hybrid orbitals pointing toward the corners of a tetrahedron. Two of those hybrids form σ bonds — one to carbon (the C–O bond), one to hydrogen (the O–H bond) — and the remaining two hybrids each hold a lone pair. This is the same hybridization water uses, and it is why alcohols are often described as "alkylated water": replace one O–H of H₂O with an O–R group and you get an alcohol. The ideal tetrahedral angle is 109.5°, but real H–O–C and lone-pair repulsions compress the H–O–C bond angle to about 108.9° in methanol (close to tetrahedral) — lone pair–bond pair repulsion is slightly stronger than bond pair–bond pair repulsion (the same VSEPR reasoning that pushes water's H–O–H angle down to 104.5°, but less severely here because only one hydrogen sits at the vertex instead of two). The consequence of this geometry is crucial: the two lone pairs are not buried — they project outward from the oxygen into the space around the molecule, fully available to accept hydrogen bonds from other molecules' O–H or N–H groups. Simultaneously, the alcohol's own O–H hydrogen, highly polarized toward the very electronegative oxygen, is exposed and ready to donate a hydrogen bond to some other molecule's lone pair. Every alcohol molecule in the liquid, therefore, is simultaneously a donor (via O–H) and an acceptor (via either lone pair) — up to three hydrogen bonds per molecule in principle, though in liquid alcohols the time-averaged number per molecule is roughly 1.5–2, well below water's ~3.5, because each alcohol oxygen has only one O–H donor (versus water's two) while retaining two acceptor lone pairs. Quantifying the polarity: using the Pauling electronegativity scale, the C–O bond has $\Delta\chi = 3.44 - 2.55 = 0.89$ and the O–H bond has $\Delta\chi = 3.44 - 2.20 = 1.24$. Both are firmly in the polar-covalent regime (0.5–1.7 on the Pauling scale), and the O–H bond is the more polarized of the two. This shows up directly in the measured dipole moment: methanol has $\mu = 1.70$ D, comparable to water's 1.85 D, and far larger than an alkane's essentially zero dipole. The molecular dipole is the vector sum of the C–O and O–H bond dipoles plus lone-pair contributions, and because the geometry is bent-tetrahedral rather than linear, these do not cancel — they reinforce into a net dipole pointing roughly from the alkyl/hydrogen side toward the oxygen lone pairs.

Now widen the lens to the liquid phase. Because the O–H hydrogen bears a large partial positive charge ($\delta^+$, roughly +0.4 to +0.5 of an electron's charge by typical computed atomic charges) and the oxygen lone pairs are compact, high-electron-density regions ($\delta^-$), an O–H group on one alcohol molecule is strongly attracted to a lone pair on a neighboring molecule's oxygen. This is the hydrogen bond: $\text{O–H}\cdots\text{:O}$. It is fundamentally an electrostatic attraction (dipole–dipole, sharpened because the positive end is an essentially "bare" proton with almost no shielding electron cloud of its own) supplemented by a small amount of genuine orbital overlap — donation from the acceptor oxygen's lone pair into the O–H σ antibonding orbital of the donor, an $n \rightarrow \sigma^*$ interaction analogous to (though weaker than) a covalent bond. This partial covalent character is why hydrogen bonds are strongly directional (they want the O–H···O angle near 180°) rather than the isotropic pull of a simple point-charge interaction. In liquid methanol or ethanol, each molecule hydrogen bonds, on average, to roughly two neighbors, forming chains and rings that are constantly breaking and reforming on a picosecond timescale — never a single static structure, but a dynamic network whose average* extent of bonding is what determines the bulk thermodynamics.

The dipole moment of a bond is the product of the partial charge separated and the bond length:

where: $q$ is the magnitude of partial charge (in the polarized bond) and $d$ is the bond length; $\mu$ is conventionally reported in debye (1 D = $3.336\times10^{-30}$ C·m). The overall molecular dipole is the vector sum over all bond and lone-pair contributions:

where: the sum runs over every polarized bond and lone pair in the molecule, each treated as a vector pointing from negative to positive (or vice versa by convention) along its local axis, combined according to molecular geometry (here, near-tetrahedral). Reading: because the C–O–H unit is bent rather than linear, and the lone pairs point away from the bonds, none of these individual dipole vectors cancel; the alcohol dipole is large precisely because oxygen's tetrahedral geometry keeps the negative (lone-pair) character and positive (O–H) character spatially separated rather than symmetric. The strength of a hydrogen bond, treated at the level of a simple electrostatic (Coulombic) approximation between the donor's partial positive hydrogen and the acceptor's partial negative lone-pair center, scales as:

where: $k_e$ is Coulomb's constant, $q_H$ and $q_O$ are the (opposite-sign) partial charges on the donor hydrogen and acceptor oxygen, and $r_{H\cdots O}$ is the hydrogen-bond distance (typically 1.8–2.0 Å for O–H···O in alcohols, giving an O···O distance around 2.7–2.9 Å). Reading: this crude point-charge formula already explains the two central empirical facts about hydrogen bonds: they are short-range and strongly distance-dependent (getting only a little longer weakens them a lot, since $E \propto 1/r$), and they require close approach of a genuinely electron-poor H to a genuinely electron-rich lone pair — a diffuse, non-polar C–H bond does not qualify because $q_H$ there is far too small. The boiling point of a hydrogen-bonded liquid, at the level of a simple thermodynamic argument, is set by the balance between the enthalpy needed to break the network and the entropy gained on vaporization:

where: $\Delta H_{\text{vap}}$ is the molar enthalpy of vaporization (dominated, for alcohols, by the energy needed to break hydrogen bonds plus overcome dispersion forces) and $\Delta S_{\text{vap}}$ is the entropy of vaporization (empirically close to a universal constant, Trouton's rule, $\Delta S_{vap} \approx 85$–95 J mol⁻¹K⁻¹ for non-associated liquids, but often larger for hydrogen-bonded liquids because the liquid is unusually ordered). Reading: since alcohols have both an anomalously large $\Delta H_{vap}$ (because of hydrogen bonding, on top of dispersion) and, per Trouton's rule deviations, an enlarged $\Delta S_{vap}$ (because the hydrogen-bonded liquid is far more ordered than the free gas, so vaporizing it is a bigger entropy jump than for a simple liquid), the ratio nonetheless comes out strongly positive — alcohols boil at dramatically higher temperatures than nonpolar isomers or isoelectronic species of similar mass.

Example 1 (foundational): Ranking three C₄H₁₀O isomers by boiling point using only structure. Consider three isomers of formula $\text{C}_4\text{H}_{10}\text{O}$: 1-butanol ($\text{CH}_3\text{CH}_2\text{CH}_2\text{CH}_2\text{OH}$), diethyl ether ($\text{CH}_3\text{CH}_2\text{OCH}_2\text{CH}_3$), and tert-butanol ($(\text{CH}_3)_3\text{COH}$). Predict the boiling-point order from structure alone. Reasoning. All three have identical molecular formula, so dispersion forces are roughly comparable in magnitude (though shape matters somewhat: more spherical, compact molecules like tert-butanol have slightly reduced surface-area contact and hence slightly weaker dispersion forces than an extended straight chain). The decisive factor is hydrogen-bonding capacity. 1-Butanol has a free O–H: it can both donate and accept hydrogen bonds, building an extended liquid-phase network exactly as ethanol does. Diethyl ether has an oxygen with two lone pairs but no O–H — it can only accept a hydrogen bond (from some other donor), never donate one to another ether molecule, so pure diethyl ether cannot hydrogen-bond to itself at all; its intermolecular forces are dipole–dipole (from the polarized C–O–C, dipole moment about 1.15 D) plus dispersion only. tert-Butanol also has a free O–H and can hydrogen-bond, but its bulky, branched, nearly spherical shape reduces the surface area available for both dispersion contacts between molecules and for close approach needed for optimal hydrogen-bond geometry (steric hindrance around a tertiary, congested oxygen). Prediction and check against data: 1-butanol (extended chain, full H-bonding, largest effective surface area) should boil highest; tert-butanol (full H-bonding but sterically hindered/compact) should boil lower than 1-butanol but still well above the ether; diethyl ether (no H-bond donation at all) should boil lowest. Measured boiling points: 1-butanol 117.7 °C, tert-butanol 82.4 °C, diethyl ether 34.6 °C — confirming the predicted order 1-butanol > tert-butanol > diethyl ether, with a gap of roughly 48 °C between the two alcohols (steric/shape effect) and a further roughly 48 °C gap down to the ether (loss of H-bond donation entirely).

Example 2 (applied): Predicting relative water solubility of 2-butanol vs. 1-hexanol from the crossover model. Using the crossover relation derived above, $n_{crossover} \approx \delta/\gamma \approx 5.1$, estimate whether 2-butanol ($n=4$ carbons past the functional carbon, i.e., $\text{C}_4\text{H}_{10}\text{O}$) and 1-hexanol ($\text{C}_6\text{H}_{14}\text{O}$, $n=6$) should be miscible or only partially soluble in water, and rank their expected solubilities. Reasoning. For 2-butanol, $n = 4 < 5.1$, so the model predicts $\Delta G_{solution} = \gamma(4) - \delta = 3.5(4) - 18 = 14 - 18 = -4$ kJ/mol, mildly favorable — predicting substantial, likely fully miscible or near-miscible behavior. (2-Butanol, additionally, is a secondary alcohol with a somewhat less linear, slightly more compact shape than 1-butanol, which nudges $\gamma_{\text{eff}}$ down a bit further, reinforcing the favorable prediction.) For 1-hexanol, $n = 6 > 5.1$, giving $\Delta G_{solution} = 3.5(6) - 18 = 21 - 18 = +3$ kJ/mol, mildly unfavorable — predicting limited solubility, with the alcohol preferring to phase-separate into an alcohol-rich layer above a mostly-aqueous layer. Check against real data: 2-butanol's water solubility is about 12.5 g per 100 mL at room temperature (substantial, consistent with the small negative/near-zero predicted $\Delta G$, i.e., "borderline miscible" in practice — full loop miscibility is temperature-dependent for 2-butanol, an experimentally well-known partially-miscible-with-an-upper-consolute-behavior case), while 1-hexanol's solubility is only about 0.6 g per 100 mL — dramatically lower, consistent with the predicted sign flip of $\Delta G_{solution}$ between $n=4$ and $n=6$. The direction of the trend (solubility collapsing by more than an order of magnitude between four and six carbons) is exactly what the simple linear cavity-cost model predicts, even though the crude constants $\gamma$ and $\delta$ are only order-of-magnitude estimates.

Example 3 (exam/competition level): Explaining an inversion — why glycerol (propane-1,2,3-triol) is a much more viscous, higher-boiling liquid than 1-propanol despite having a nearly identical carbon skeleton. Glycerol, $\text{HOCH}_2\text{CH(OH)CH}_2\text{OH}$, boils at 290 °C (with decomposition) and has a viscosity roughly 1000 times that of water at room temperature; 1-propanol, $\text{CH}_3\text{CH}_2\text{CH}_2\text{OH}$, boils at 97 °C. Both have a three-carbon backbone. Explain the enormous gap mechanistically, using a quantitative hydrogen-bond count. Reasoning. 1-Propanol has exactly one OH group: one hydrogen-bond donor site, two acceptor lone pairs (on one oxygen), giving each molecule an average participation of perhaps 1.5–2 hydrogen bonds with neighbors at any instant, exactly as in ethanol. Glycerol has three OH groups on the same three-carbon backbone — three independent donor sites and three oxygens' worth of acceptor lone pairs (six lone pairs total). Each glycerol molecule can therefore participate in a hydrogen-bonding network several times denser than 1-propanol's: experimentally, each glycerol molecule engages, on average, in roughly 4–5 hydrogen bonds simultaneously with its neighbors (some donated, some accepted), compared to roughly 1.5–2 for a monool of similar chain length. Quantitatively, using the same per-hydrogen-bond enthalpy increment (~20–25 kJ/mol) used for ethanol above, and counting bonds per molecule while dividing by two to avoid double counting shared bonds between pairs:

compared to roughly 20 kJ/mol estimated for 1-propanol (essentially identical to the ethanol calculation above, since chain length beyond the OH contributes negligibly extra to hydrogen-bond count, only to dispersion). The measured enthalpies of vaporization bear this out: $\Delta H_{vap}(\text{1-propanol}) \approx 47.5$ kJ/mol, while $\Delta H_{vap}(\text{glycerol}) \approx 91$ kJ/mol near its (decomposition-limited) boiling range — a difference of about 44 kJ/mol, in good qualitative agreement with the extra-hydrogen-bond estimate above (the remaining discrepancy reflects the crudeness of assigning a single fixed per-bond energy and of estimating average coordination numbers in a real disordered liquid). The same logic explains glycerol's viscosity: viscosity reflects the kinetic barrier to molecules sliding past one another, which scales with how extensively and strongly each molecule is hydrogen-bonded to its neighbors — with three OH groups per molecule, each glycerol molecule is effectively cross-linked into the surrounding liquid at multiple points simultaneously (much like a small-molecule analog of hydrogen-bonded polymer networks), so breaking enough bonds to let one molecule slip past another requires disrupting several hydrogen bonds at once rather than just one or two, exactly the reasoning that also explains why sugars (polyols like glucose) are viscous syrups and high-melting solids rather than simple liquids.

The modern understanding of hydrogen bonding did not spring from a single discovery but crystallized gradually across the early twentieth century, with the concept named and systematized largely through the work of Linus Pauling.