Molecular Shape and VSEPR Theory

VSEPR is the first filter every chemist applies before reaching for computation. Medicinal chemists use it to predict whether a drug candidate's reactive or hydrogen-bonding groups point the right way to fit a binding pocket — the trigonal pyramidal geometry of an amine nitrogen, for instance, determines whether its lone pair is presented for a key hydrogen bond or protonation event in a protease active site. Materials chemists rely on it to rationalize the geometry of hypervalent main-group compounds used in flame retardants (phosphorus SN=5 centers) and semiconductor precursors. In atmospheric and environmental chemistry, the bent geometry of ozone (SO₂-like, SN=3 with one lone pair) is essential to understanding its dipole-driven reactivity and absorption spectrum. Structural biology courses use VSEPR reasoning as the entry point before teaching that the resulting geometries correlate with orbital hybridization schemes, which themselves set up later topics like Bond Polarity and Electronegativity (geometry plus bond polarity together determine whether a molecule has a net dipole moment) — a bent, polar-bonded molecule like water has a large net dipole, while a linear, polar-bonded molecule like CO₂ has canceling dipoles and none at all, a distinction that is invisible without first getting the VSEPR shape right.

Ask the tutor: Why does bond angle theory (VSEPR) sometimes fail for transition-metal complexes, and what replaces it there?

Start from the Lewis Structures: Rules and Construction picture: a central atom surrounded by some number of bonding pairs (shared with substituents) and possibly some lone pairs (localized on the central atom only). Each of these pairs occupies, on average, a roughly conical region of space pointing away from the nucleus — an electron domain. The valence electrons of the central atom are far more mobile than the tightly held core electrons, and they repel each other via ordinary Coulombic repulsion, modulated by the Pauli exclusion principle (two electrons of the same spin cannot occupy the same region). The result: domains push apart until they reach a stable geometric arrangement that maximizes the smallest domain–domain angle. This is precisely the same electrostatic logic as arranging point charges on a sphere — for 4 domains, the tetrahedron; for 5, the trigonal bipyramid; for 6, the octahedron — because these are the arrangements that maximize minimum pairwise angular separation. Not all domains are equally repulsive. A lone pair is held by only one nucleus, so its electron density is less confined and spreads into a larger, "fatter" angular region than a bonding pair, whose density is pulled toward the region between two nuclei and thus compressed. This asymmetry produces the empirical repulsion ranking:

This ranking is the single correction needed to go from "ideal polyhedron" predictions to real, experimentally measured bond angles — it is why water's HOH angle is 104.5° rather than the tetrahedral 109.5°, and why ammonia's HNH angle is 107.2° rather than 109.5°. A multiple bond (double or triple) is treated as one electron domain for counting purposes — all the electron density of a π system is still confined to the same internuclear axis as the σ bond, so it does not create an independent direction in space. However, a double or triple bond's domain is broader and more electron-rich than a single bond's domain (more electrons crowded into the same angular region), so it repels adjacent domains slightly more than a single bond does, pushing other bond angles open — this is why the O=C=O like angles in formaldehyde or ketones sit slightly above 120°, e.g. the H–C–H angle in formaldehyde is 116°, not 120°, because the C=O domain widens the other two angles at its expense... more carefully, the H–C–H angle narrows and the H–C=O angles widen, since the fat C=O domain pushes the two C–H bonds together.

The entire predictive scheme reduces to one counting rule and one geometric lookup.

where: σ bonds are counted once per bonded neighbor regardless of bond order (a double bond is one σ + one π, but still one domain); lone pairs come directly from the completed Lewis structure.

where: this mapping is the electron-domain geometry — the arrangement of all domains (bonding + lone) in space, derived purely from maximizing angular separation for that number of repelling points.

where: the tetrahedral angle $109.47^\circ = \arccos(-1/3)$ is the exact solution for four equal point charges on a sphere repelling via inverse-square law — not a memorized number but a geometric consequence.

where: this is a qualitative ordering (not a closed-form formula) that predicts the direction of angle distortion from the ideal SN-based value; each additional lone pair typically compresses adjacent bond angles by roughly 1.5–2.5°.

Reading these relationships together: first assign SN from the Lewis structure, second look up the idealized electron-domain geometry, third remove lone pairs from the verbal description of shape (but not from space — they still occupy a vertex and still push other domains around), and fourth apply the repulsion-ranking correction to predict whether real angles will be compressed relative to ideal.

Example 1 (foundational): Ammonia, NH₃ — the canonical lone-pair case. Step 1 — Lewis structure: nitrogen has 5 valence electrons, forms 3 N–H bonds (using 3 electrons in bonds contributed by N, plus 3 H electrons) and retains 1 lone pair, satisfying the octet: 3 bonding pairs + 1 lone pair = 4 electron domains around N. Step 2 — Steric number: $\text{SN} = 3\ (\sigma\text{ bonds}) + 1\ (\text{lone pair}) = 4$. Step 3 — Electron-domain geometry: SN = 4 → tetrahedral arrangement of all four domains (three N–H bonds and the lone pair each occupy a tetrahedral vertex). Step 4 — Molecular geometry (what an X-ray or electron-diffraction experiment "sees", ignoring the invisible lone pair): with one vertex occupied by a lone pair rather than an atom, the visible atomic arrangement is trigonal pyramidal. Step 5 — Angle correction: lone pair repels the three N–H bonds more strongly than they repel each other (lp–bp > bp–bp), compressing H–N–H from the ideal 109.5° down to the experimental 107.2°. Step 6 — Hybridization correlate: to accommodate four roughly tetrahedral domains, nitrogen's one 2s and three 2p orbitals mix into four sp³ hybrids; three overlap with hydrogen 1s orbitals (σ bonds), and the fourth houses the lone pair. This is the same Hybridization: sp Carbon and Triple Bonds-style orbital-mixing logic applied at SN=4 instead of SN=2.

Example 2 (applied): Sulfur tetrafluoride, SF₄ — the "seesaw" from an expanded octet. Step 1 — Lewis structure: sulfur (6 valence electrons) forms four S–F single bonds (using 4 electrons) and retains one lone pair (2 electrons), for a total valence-electron count of $4(2) + 2 = 10$ electrons around sulfur — an expanded octet, permissible because sulfur is a third-row element with accessible 3d-orbital character (or, in modern bonding language, three-center four-electron bonding) to stabilize hypervalency. Step 2 — Steric number: $\text{SN} = 4\ (\sigma) + 1\ (\text{lp}) = 5$. Step 3 — Electron-domain geometry: SN = 5 → trigonal bipyramidal, which has two inequivalent site types: two axial positions (180° apart from each other, 90° from equatorial) and three equatorial positions (120° apart from each other, 90° from axial). Step 4 — Lone-pair placement rule: because axial positions have three 90°-close neighbors (compressed, high-repulsion sites) while equatorial positions have only two 90°-close neighbors, the bulky, more-repulsive lone pair always occupies an equatorial site to minimize its number of strong 90° interactions. This is a general rule for all SN=5 species with lone pairs. Step 5 — Molecular geometry: with the lone pair sitting in one equatorial vertex, four fluorines occupy the remaining vertices (2 axial + 2 equatorial) — this arrangement, viewed atom-only, is called seesaw (or disphenoidal). Step 6 — Angle corrections: the equatorial lone pair repels the two axial F atoms (which are 90° away from it) more than it repels the two remaining equatorial F atoms (120° away), bending the axial F–S–F angle down from 180° to an experimental ≈173°, and bending the equatorial F–S–F angle down from 120° to ≈101–102°, both compressions being consistent with lp–bp > bp–bp repulsion acting most strongly on the closest (90°) neighbors.

Example 3 (exam/competition level): Chlorine trifluoride, ClF₃ — multi-step lone-pair placement with asymmetric distortion. This is a classic "trap" question because two different lone-pair placements are geometrically plausible at SN=5, and only careful minimization of total repulsion picks the right one. Step 1 — Lewis structure: chlorine (7 valence electrons) forms three Cl–F bonds (3 electrons used) and retains two lone pairs (4 electrons), for a total of $3(2)+4 = 10$ electrons — again an expanded octet on a third-row halogen. Step 2 — Steric number: $\text{SN} = 3\ (\sigma) + 2\ (\text{lp}) = 5$ → trigonal bipyramidal electron-domain geometry. Step 3 — Candidate placements. With 2 lone pairs and 3 bonding pairs to distribute among 2 axial + 3 equatorial sites, there are two chemically distinct possibilities to compare: (a) both lone pairs equatorial (leaving one F equatorial, two F axial), or (b) one lone pair axial and one equatorial. Step 4 — Count 90° (strong) interactions for each option, since 90° pairs dominate total repulsion (they are the closest, highest-repulsion contacts): For option (a): the two equatorial lone pairs are 120° apart from each other (a weak lp–lp interaction) but each equatorial lone pair has two 90° axial-F neighbors, giving $2\times 2 = 4$ lp–bp 90° contacts, and zero lp–lp 90° contacts (since both lone pairs are equatorial, they see each other only at 120°). For option (b): one axial lone pair has three 90° equatorial-F neighbors (3 lp–bp 90° contacts), and the equatorial lone pair has two 90° axial neighbors — but one of those axial slots is now the other lone pair, giving 1 lp–lp 90° contact (the worst possible interaction) plus 1 lp–bp 90° contact, totaling 3 lp–bp-equivalent + 1 very costly lp–lp contact. Step 5 — Compare. Option (a) has zero 90° lp–lp contacts (the most repulsive type), only 90° lp–bp contacts. Option (b) introduces one 90° lp–lp contact, which is disproportionately costly. Therefore option (a) — both lone pairs equatorial — is the lower-energy, experimentally correct structure. Step 6 — Molecular geometry: with 2 equatorial lone pairs and 3 F atoms occupying the remaining 2 axial + 1 equatorial sites, the observed atom-only shape is T-shaped. Step 7 — Angle correction: the axial F–Cl–F angle bends from 180° down to the experimental 175°, and the axial-to-equatorial F–Cl–F angles bend from 90° down to about 87.5°, both compressions driven by the two equatorial lone pairs squeezing the axial fluorines together — a small but measurable distortion, confirming the qualitative lp–bp > bp–bp prediction even in this more crowded five-domain system.

VSEPR as a systematic theory is credited primarily to Ronald Gillespie, building on earlier suggestions by Nevil Sidgwick and Herbert Powell in the 1940s that electron pairs around a central atom should be treated as mutually repelling. Gillespie, working at University College London and later McMaster University, spent the 1950s–1960s turning this qualitative intuition into a systematic, teachable algorithm — explicitly formalizing the lone-pair-vs-bonding-pair repulsion hierarchy, the treatment of multiple bonds as single domains, and the rules for lone-pair placement at trigonal-bipyramidal centers (the equatorial preference used in the SF₄ and ClF₃ examples above). He deliberately built VSEPR as a teaching and prediction tool that required no quantum mechanical computation — remarkable given that, at the time, the alternative was full molecular-orbital calculation, computationally infeasible for routine classroom or bench use.