Physical Properties from Structure

Predicting physical properties from structure is not an academic parlor trick — it is load-bearing infrastructure across chemistry, pharmacology, and materials science. In drug design, Lipinski's "Rule of Five" is a direct, quantitative descendant of the ideas in this article: it flags drug candidates likely to have poor oral bioavailability if they have too many hydrogen-bond donors (>5), too many acceptors (>10), too high molecular weight (>500), or too high lipophilicity (LogP > 5) — each threshold is a proxy for exactly the donor/acceptor/surface-area balance worked out above for acetaminophen versus ibuprofen. Medicinal chemists routinely add a fluorine, remove an N–H, or swap a methyl for a hydroxyl specifically to retune a candidate's melting point (affecting crystallinity and formulation), boiling/subliming behavior, and aqueous solubility. In materials science, polymer chemists exploit exactly the packing-symmetry lesson from para- versus ortho-dichlorobenzene: engineering regularity into a polymer backbone (isotactic versus atactic polypropylene, for instance) changes how well chains can crystallize, directly tuning the melting point, density, and mechanical strength of the resulting plastic — a stereochemical and packing question, not merely an IMF-strength one, closely related to how Conformational Analysis governs solid-state behavior. In green chemistry and solvent selection, the "like dissolves like" heuristic developed rigorously above (donor/acceptor matching, dispersion-based miscibility of hydrocarbons) underlies solvent substitution strategies — replacing chlorinated solvents with more benign ethers, esters, or alcohols requires predicting solubility behavior for new solute–solvent pairs entirely from structure, exactly the exercise performed in the worked examples here. And in atmospheric and environmental chemistry, the vapor pressure/boiling point relationship via Clausius–Clapeyron directly predicts which volatile organic compounds (VOCs) partition into the gas phase (contributing to smog and ozone chemistry) versus remaining condensed on aerosol particles — a real-world, high-stakes application of the exact $\ln P_{vap}$ versus $1/T$ relationship derived above, and a topic that connects naturally to Acid-Base Chemistry when considering how atmospheric humidity and pH affect aerosol partitioning of ionizable organics.

Ask the tutor: How would you predict the relative boiling points and water solubilities of a primary, secondary, and tertiary amine of similar molecular weight?

Think of a liquid as a swarm of molecules held together by a shifting web of noncovalent attractions. To boil the liquid, you must supply enough thermal energy to break essentially all of these attractions and separate molecules to the gas phase, where they interact negligibly. Boiling point is therefore a direct readout of the strength and abundance of intermolecular contacts in the liquid. Every pair of molecules, no matter how nonpolar, experiences London dispersion: instantaneous fluctuations in electron density create a transient dipole in molecule A, which polarizes molecule B, inducing a complementary dipole that attracts back. The interaction energy from this induced-dipole/induced-dipole effect falls off steeply, as $1/r^6$, and its magnitude scales with polarizability $\alpha$ — which in turn scales with the number of electrons and, crucially, with surface area available for contact. This is why an elongated pentane molecule, which can lie alongside its neighbors over its whole length, has stronger net dispersion than a compact, near-spherical neopentane molecule with the same molecular formula but far less contact area. If a molecule carries polar bonds whose dipoles do not cancel, it acquires a permanent dipole moment $\mu$. Two such dipoles orient to attract (positive end near negative end), adding an energy term that also falls off as $1/r^6$ (after orientational averaging) but with a prefactor involving $\mu^4$ for freely rotating dipoles at a given temperature (Keesom interaction) — the essential point for this article is simpler: dipole–dipole forces add a real, extra attraction on top of dispersion, so at similar molecular weight, polar molecules boil higher than nonpolar ones. Hydrogen bonding is dipole–dipole taken to an extreme: because H is so small and O–H/N–H/F–H bonds are so polarized, the departing hydrogen exposes a nearly bare, highly concentrated positive charge that can approach an acceptor lone pair extremely closely, with significant orbital overlap between the lone pair and the σ* of the X–H bond. This partial covalency (an $n \rightarrow \sigma^*$ donor–acceptor interaction) is why hydrogen bonds (typically 15–40 kJ/mol) are far stronger than ordinary dipole–dipole contacts (2–5 kJ/mol) though weaker than covalent bonds (300–400 kJ/mol).

Now overlay hydrogen bonding onto solubility. Water is an extraordinarily cohesive liquid: every molecule can donate two H-bonds (from its two O–H's) and accept two more (via its two lone pairs), building a dynamic 3-D network worth roughly 20 kJ/mol per bond. To dissolve a solute in water, the solute must pay the cost of inserting itself into that network — breaking some water–water H-bonds — and it can only be repaid if the solute itself can donate/accept H-bonds to water in replacement. A nonpolar hydrocarbon can offer nothing in return: water must rebuild an ordered "cage" (a clathrate-like shell) around it, costing entropy, which is the molecular origin of the hydrophobic effect and the low solubility of oils in water.

The unifying quantitative statement is that boiling point (at 1 atm) is the temperature at which the vapor pressure equals atmospheric pressure, and vapor pressure is itself governed by how much enthalpy is needed to remove a molecule from the liquid's web of IMFs:

where:

• $P_{vap}(T)$ is the equilibrium vapor pressure at temperature $T$,
• $\Delta H_{vap}$ is the molar enthalpy of vaporization — essentially, the total IMF energy per mole that must be overcome,
• $R$ is the gas constant (8.314 J mol⁻¹ K⁻¹),
• $C$ is a substance-specific integration constant (related to $\Delta S_{vap}$).

Reading this equation: stronger IMFs → larger $\Delta H_{vap}$ → steeper decline of $\ln P$ with $1/T$ → lower vapor pressure at any given temperature → higher temperature required to reach $P = 1\text{ atm}$ → higher boiling point. This is the rigorous backbone beneath the qualitative rule "more/stronger IMFs → higher bp": it is literally a statement about the enthalpy needed to convert liquid to gas. A second core relationship, Trouton's rule, connects $\Delta H_{vap}$ to $T_b$ directly for "normal" liquids (no strong association):

where:

• $T_b$ is the normal boiling point in kelvin,
• the constant $\approx 88$ J/(mol·K) is the entropy of vaporization for simple liquids (Trouton's constant).

Reading: liquids with the same entropy change on vaporizing (roughly, similar disorder gained) will have $\Delta H_{vap}$ proportional to $T_b$ — but hydrogen-bonded liquids systematically violate this rule (their entropy of vaporization is anomalously large, since the liquid is more ordered than a "normal" liquid due to the H-bond network), which is itself a diagnostic: measuring $\Delta S_{vap}$ and finding it well above 88 J/(mol·K) is direct thermodynamic evidence of hydrogen bonding or other structuring in the liquid.

Example 1 (foundational): Rank the boiling points of propane, dimethyl ether, and ethanol — all with different formulas but comparable "size." Structures: CH₃CH₂CH₃ (propane, MW 44, nonpolar, dispersion only); CH₃OCH₃ (dimethyl ether, MW 46, net dipole ~1.3 D, dipole–dipole + dispersion, no O–H donor); CH₃CH₂OH (ethanol, MW 46, dipole + full H-bond donor/acceptor capability). Mechanistic ranking logic: propane has only dispersion — weakest IMF category — predict lowest bp. Dimethyl ether adds a permanent dipole on top of similar-strength dispersion (similar MW to ethanol) — predict intermediate bp. Ethanol adds hydrogen bonding on top of both — predict highest bp, and by a wide margin since H-bonds (~20 kJ/mol) dwarf dipole–dipole contributions (~2–5 kJ/mol). Predicted order: propane < dimethyl ether < ethanol. Experimental values: propane −42 °C, dimethyl ether −24 °C, ethanol 78.4 °C. Check the gaps: propane→ether gap is 18 °C (consistent with adding one modest dipole–dipole term); ether→ethanol gap is 102 °C (consistent with adding a full hydrogen-bond network, an order of magnitude bigger effect) — the asymmetric size of the two gaps is itself strong mechanistic confirmation that H-bonding, not merely "polarity," is the dominant term once it becomes available.

Example 2 (applied, drug-design flavor): Predict the relative water solubility of acetaminophen vs. ibuprofen, from structure alone. Acetaminophen: 4-hydroxyacetanilide, HO–C₆H₄–NH–CO–CH₃. Count H-bond donors: the phenolic OH (1 donor) and the amide N–H (1 donor). Count H-bond acceptors: the phenolic O (accepts too, though also donates) and the amide carbonyl O (strong acceptor) — roughly 2 strong donors + 2 strong acceptors, plus a moderately sized aromatic ring contributing dispersion-driven hydrophobicity. Ibuprofen: 2-(4-isobutylphenyl)propanoic acid, with a carboxylic acid (–COOH, 1 donor + 2 acceptor-type oxygens) but a large, branched isobutylphenyl hydrocarbon domain (isobutyl group + benzene ring) contributing substantial nonpolar surface area with essentially no polar functionality. Mechanistic prediction: acetaminophen has a favorable donor/acceptor-to-hydrocarbon-surface ratio — its polar groups can effectively "pay for" solvating its modest aromatic ring via multiple strong H-bonds to water, while ibuprofen's single –COOH group must "carry" a much larger hydrophobic isobutylphenyl domain, which resists solvation via the hydrophobic effect (water must cage the aliphatic/aromatic bulk, costing entropy that the single carboxylic acid group cannot fully repay energetically). Predicted: acetaminophen more water-soluble than ibuprofen. Experimental solubility (25 °C, water): acetaminophen ≈ 14 mg/mL; ibuprofen ≈ 0.021 mg/mL — a difference of roughly 700-fold, consistent with (and far larger than) the qualitative donor/acceptor-versus-hydrophobic-surface argument, and this is exactly why ibuprofen is formulated as a more soluble salt (e.g., the lysine or sodium salt) for faster-acting oral formulations — ionizing the carboxylic acid to carboxylate massively increases the ion–dipole interaction with water, overwhelming the hydrophobic penalty from the isobutylphenyl group.

Example 3 (exam/competition level): Explain, with full mechanistic and packing reasoning, why para-dichlorobenzene melts at 53 °C while ortho-dichlorobenzene melts at −17 °C, even though both are constitutional isomers with essentially identical dispersion and dipole character per molecule. This is a genuine trap for students who reflexively reach only for "IMF strength" arguments — melting point is not simply a readout of the strongest IMF; it is a readout of how well the solid lattice packs, since melting requires disrupting a specific 3-D crystal arrangement, not merely overcoming a generic pairwise attraction. Structures: p-dichlorobenzene is a rectangular, highly symmetric molecule (point group $D_{2h}$ for the ring skeleton with substituents at 1,4-positions) — it is essentially a flat "brick" shape. o-Dichlorobenzene has the two chlorines adjacent (1,2-positions), giving a lower-symmetry, slightly wedge-shaped molecule with an appreciable net dipole moment (~2.5 D, versus essentially zero net dipole for the symmetric para isomer, since the two C–Cl bond dipoles cancel by symmetry in the para case). Mechanistic reasoning, step by step: 1. In the liquid or solution phase, dipole moment would suggest ortho should have stronger dipole–dipole attractions (nonzero μ) than para (μ ≈ 0, dispersion-dominated) — if boiling point were the question, you might expect ortho to boil somewhat higher due to its dipole, and indeed ortho does boil slightly higher (180.5 °C) than para (174 °C), consistent with the added dipole–dipole term. 2. But melting requires the solid to break its crystal lattice, and lattice stability is governed by packing efficiency: how densely and symmetrically molecules can tile 3-D space. The para isomer's rectangular, centrosymmetric shape allows it to pack into a highly ordered, dense crystal lattice with excellent point-to-point contact (analogous to how bricks stack more efficiently than irregularly shaped stones) — every molecule sits in an equivalent, symmetric environment, maximizing the number of near-optimal dispersion contacts even though each individual contact is unremarkable. 3. The ortho isomer's bent, lower-symmetry shape cannot tile space as efficiently — its solid lattice necessarily has more "wasted" void space and less-optimal molecule-to-molecule contact geometry, despite the individual molecules attracting each other via dispersion and dipole–dipole forces more strongly than para molecules do pairwise. 4. Net result: melting point (a lattice-packing-sensitive property) favors para enormously (53 °C vs. −17 °C, a 70 °C gap), while boiling point (a property of the disordered liquid, less sensitive to packing symmetry and more sensitive to raw pairwise IMF strength) mildly favors ortho (180.5 °C vs. 174 °C, only a 6.5 °C gap) — the sign of the effect flips between the two properties because they are governed by mechanistically different things: bp by pairwise interaction strength in a disordered liquid, mp by lattice packing symmetry in an ordered solid. This example is the single most important "trap" for students to internalize: never explain melting point using only the same IMF-strength logic you use for boiling point. Symmetry and packing efficiency are an independent, often dominant, factor for solids.

Johannes Diderik van der Waals is the figure whose name is permanently attached to the very forces this article is built on, though the mechanistic, quantum-mechanical explanation of why dispersion forces exist (transient dipole/induced dipole from correlated electron fluctuations) came decades later from Fritz London, working within the newly developed framework of quantum mechanics in the early 1930s. Van der Waals was a Dutch physicist who, in his 1873 doctoral dissertation, proposed a modified equation of state for real gases,

introducing two empirical correction terms: $b$ (excluded volume, accounting for the finite size of molecules) and $a$ (an attractive-pressure correction, accounting for intermolecular attraction reducing the effective pressure a real gas exerts). This was decades before anyone had a mechanistic, electronic explanation for why neutral, nonpolar molecules attract each other at all — van der Waals inferred the attraction purely from the deviation of real-gas PVT behavior from the ideal gas law, an extraordinary piece of physical reasoning from macroscopic data alone. It was Fritz London who, using early quantum mechanical perturbation theory, showed in 1930 that even a molecule with a time-averaged dipole moment of exactly zero has a fluctuating, instantaneous charge distribution whose correlated fluctuations with a neighboring molecule produce a net attractive force scaling as $1/r^6$ — finally giving van der Waals's empirical "$a$" term a first-principles quantum mechanical derivation, and giving organic chemists the "London dispersion force" vocabulary used throughout this article.