Lewis structures and VSEPR

Intuition [Beginner]

A Lewis structure is a diagram showing how valence electrons are arranged in a molecule. Atoms are represented by their chemical symbols. Bonds between atoms are drawn as lines (each line = two shared electrons). Non-bonding electrons are drawn as dots.

The goal of drawing a Lewis structure is to give every atom a full outer shell. For most main-group elements, that means eight electrons around each atom -- the octet rule. Hydrogen is the exception: it needs only two electrons (a duet).

Not every molecule can satisfy the octet rule for every atom. Boron in ${\text{BF}}_3$ has only six valence electrons. Sulfur in ${\text{SF}}_6$ has twelve. These violations are common for elements in the third row and below, which can use $d$ orbitals to exceed an octet.

VSEPR (Valence Shell Electron Pair Repulsion) predicts 3D molecular shape from a Lewis structure. The principle is simple: electron pairs around a central atom repel each other and arrange themselves as far apart as possible. Two pairs give a linear arrangement ($180^{\circ }$). Three pairs give trigonal planar ($120^{\circ }$). Four pairs give tetrahedral ($109.5^{\circ }$). Lone pairs take up more space than bonding pairs, compressing the remaining bond angles.

Visual [Beginner]

The VSEPR geometries for the most common steric numbers (electron-domain counts):

Steric number	Geometry	Bond angles	Example
2	Linear	$180^{\circ }$	${\text{CO}}_2$
3	Trigonal planar	$120^{\circ }$	${\text{BF}}_3$
3	Bent (1 lone pair)	$\approx 117^{\circ }$	${\text{SO}}_2$
4	Tetrahedral	$109.5^{\circ }$	${\text{CH}}_4$
4	Trigonal pyramidal (1 lone pair)	$\approx 107^{\circ }$	${\text{NH}}_3$
4	Bent (2 lone pairs)	$\approx 104.5^{\circ }$	${\text{H}}_2\text{O}$
5	Trigonal bipyramidal	$90^{\circ },120^{\circ }$	${\text{PCl}}_5$
6	Octahedral	$90^{\circ }$	${\text{SF}}_6$

Lone pairs compress bond angles below the ideal value because they occupy a larger solid angle than bonding pairs.

Worked example [Beginner]

Draw the Lewis structure of ${\text{SF}}_6$ (sulfur hexafluoride) and determine its VSEPR geometry.

Step 1. Count valence electrons. Sulfur (Group 16) has 6. Each fluorine (Group 17) has 7. Total: $6+6\times 7=48$.

Step 2. Place the central atom (S) and connect each F with a single bond. Six bonds use 12 electrons, leaving 36.

Step 3. Complete octets on the surrounding atoms. Each F needs 6 more electrons (3 lone pairs). Six fluorines: $6\times 6=36$ electrons. That accounts for all remaining electrons.

Step 4. Check the central atom. Sulfur has 12 electrons around it (six bonds). This exceeds the octet, but sulfur is in the third row and can expand its valence shell. No need to add double bonds.

The VSEPR geometry: six bonding pairs, zero lone pairs around sulfur. Steric number = 6, giving octahedral geometry with all bond angles $90^{\circ }$.

${\text{SF}}_6$ is an example of an expanded octet. The central atom uses $3d$ orbitals (or more precisely, the molecular orbital framework accommodates 12 electrons) to form more than four bonds.

Check your understanding [Beginner]

Formal definition [Intermediate+]

A Lewis structure of a molecule with formula ${\text{A}}_m{\text{B}}_n\dots$ is a graph whose vertices are atoms labelled by element symbol and whose edges are covalent bonds, augmented by lone-pair electron dots on each atom. The construction proceeds by:

1. Count valence electrons. Sum the valence electrons of all atoms; add or subtract electrons for ionic charge.
2. Draw the skeleton. Connect atoms with single bonds (2 electrons each). The least electronegative atom (except H) is usually central.
3. Complete octets. Distribute remaining electrons to give each surrounding atom a full octet (or duet for H).
4. Satisfy the central atom. If the central atom lacks an octet, convert lone pairs on surrounding atoms into double or triple bonds.
5. Minimise formal charges. The preferred structure has formal charges closest to zero and places negative formal charges on more electronegative atoms.

Formal charge of atom $i$ in a Lewis structure is

$${\text{FC}}_i=V_i-N_{\text{lone},i}-\frac{1}{2}{N}_{\text{bond},i},$$

where $V_i$ is the number of valence electrons in the free atom, $N_{\text{lone},i}$ is the number of non-bonding electrons, and $N_{\text{bond},i}$ is the number of bonding electrons. Formal charges must sum to the total charge on the molecule or ion.

Resonance. When multiple Lewis structures satisfy the rules equally well (same arrangement of atoms, different arrangement of electrons), the true electron distribution is a resonance hybrid -- a weighted average of the contributing structures. The resonance hybrid is a single entity, not a rapid interconversion between structures. The carbonate ion ${\text{CO}}_3^{2-}$ has three equivalent resonance structures; each C-O bond has bond order $1\frac{1}{3}$, not alternating single and double bonds.

VSEPR formalism

The steric number (SN) of the central atom is the total number of electron domains (bonding + lone pairs). Each single bond, double bond, triple bond, or lone pair counts as one domain. The electron-domain geometry depends only on SN:

$$\begin{array}{rl}\text{SN}=2& ⟶\text{linear}\\ \text{SN}=3& ⟶\text{trigonal planar}\\ \text{SN}=4& ⟶\text{tetrahedral}\\ \text{SN}=5& ⟶\text{trigonal bipyramidal}\\ \text{SN}=6& ⟶\text{octahedral}\end{array}$$

The molecular geometry is determined by removing lone pairs from the electron-domain geometry and describing the shape formed by the atoms alone. The bond angles deviate from ideal as follows:

• Lone pairs repel more strongly than bonding pairs (order: lp-lp > lp-bp > bp-bp).
• Multiple bonds repel more strongly than single bonds.
• More electronegative substituents pull electron density toward themselves, reducing their repulsive effect on other domains.

Beyond the octet rule

Electron-deficient molecules (e.g., ${\text{BF}}_3$, ${\text{BeCl}}_2$) have fewer than 8 electrons around the central atom. Boron in ${\text{BF}}_3$ has 6 valence electrons. These molecules are strong Lewis acids.

Radicals (e.g., $\text{NO}$, ${\text{ClO}}_2$) have an odd number of electrons and cannot satisfy the octet rule for all atoms. Nitric oxide (NO) has 11 valence electrons; one unpaired electron resides in an antibonding orbital.

Expanded octets occur for elements in period 3 and below (S, P, Cl, Xe, etc.) which can accommodate more than 8 electrons. ${\text{PF}}_5$ (10 electrons around P) and ${\text{SF}}_6$ (12 electrons around S) are standard examples.

Counterexamples to common slips

• "The octet rule is universal." It fails for electron-deficient species (${\text{BF}}_3$, ${\text{BeH}}_2$), radicals (${\text{NO}}_2$), and expanded-octet species (${\text{SF}}_6$, ${\text{XeF}}_4$). For period-3-and-below elements, expanded octets are the norm rather than the exception.

• "Formal charge equals real charge." Formal charge is a bookkeeping device that assigns equal sharing in bonds. Real charge (partial charge) is determined by electronegativity differences and is not an integer. In ${\text{CO}}_2$, the formal charge on each atom is zero, but oxygen carries a partial negative charge and carbon a partial positive one.

• "All resonance structures contribute equally." Only equivalent structures (same atom arrangement, same number of bonds, same formal-charge distribution) contribute equally. Structures with fewer formal-charge separations or with negative charges on more electronegative atoms contribute more to the hybrid.

Key theorem with proof [Intermediate+]

Theorem (Gillespie-Nyholm VSEPR postulates). The geometry of a molecule ${\text{AX}}_m{\text{E}}_n$ (where A is the central atom, X are bonded atoms, and E are lone pairs) is determined by the steric number $\text{SN}=m+n$ and the arrangement of lone pairs to minimise repulsion. The predicted geometry is:

$\text{SN}$	$m$	$n$	Molecular geometry
2	2	0	Linear
3	3	0	Trigonal planar
3	2	1	Bent
4	4	0	Tetrahedral
4	3	1	Trigonal pyramidal
4	2	2	Bent
5	5	0	Trigonal bipyramidal
5	4	1	Seesaw
5	3	2	T-shaped
5	2	3	Linear
6	6	0	Octahedral
6	5	1	Square pyramidal
6	4	2	Square planar

Argument. Electron domains in the valence shell of the central atom repel each other by Coulombic and Pauli (exchange) repulsion. For $N$ domains, the arrangement maximising the minimum pairwise angular separation on a sphere is:

• $N=2$: opposite points ($180^{\circ }$).
• $N=3$: equilateral triangle ($120^{\circ }$).
• $N=4$: regular tetrahedron ($109.5^{\circ }$).
• $N=5$: trigonal bipyramid (axial-equatorial $90^{\circ }$, equatorial-equatorial $120^{\circ }$).
• $N=6$: regular octahedron ($90^{\circ }$).

Lone pairs occupy more angular space than bonding pairs because the lone-pair electrons are localised on the central atom rather than shared. This leads to the repulsion hierarchy lp-lp > lp-bp > bp-bp. For steric number 5, lone pairs preferentially occupy equatorial positions (which have $120^{\circ }$ separation from two neighbours and $90^{\circ }$ from two, versus the axial position with $90^{\circ }$ from three neighbours). For steric number 6, lone pairs prefer trans positions to maximise their separation. $\square$

The VSEPR model is a qualitative electrostatic argument, not a theorem derived from quantum mechanics. It works because electron-pair localisation is a reasonable first approximation for many main-group molecules. It fails for transition-metal complexes (where $d$-orbital effects dominate), for molecules with significant $\pi$-delocalisation, and for cases where the simple steric-number counting does not capture the relevant electronic structure.

Bridge. The Gillespie-Nyholm postulates build toward 14.02.02 pending hybridization, where the same geometries emerge from orbital overlap: $s{p}^3$ gives tetrahedral, $s{p}^2$ gives trigonal planar, $sp$ gives linear. The repulsion hierarchy lp-lp > lp-bp > bp-bp appears again in 14.05.01 pending molecular orbital theory as a consequence of orbital occupancy and electron density distribution. The foundational reason VSEPR works is that the Pauli exclusion principle keeps same-spin electrons apart, and this is exactly the physics that orbital hybridization re-expresses in the language of directed wavefunctions.

Exercises [Intermediate+]

Lean formalization [Intermediate+]

Mathlib has no formalisation of Lewis structures or VSEPR. The closest available infrastructure:

• Mathlib.Combinatorics.Graph: graph structures could represent molecular bonding networks.
• Mathlib.LinearAlgebra: could underpin 3D coordinate geometry for molecular shapes.
• Mathlib.GroupTheory.Subgroup.Pointwise: relevant to point-group symmetry classification.

A formal VSEPR would need: a steric-number-to-geometry mapping as a computational function; a proof that the claimed geometry maximises minimum pairwise angular separation on a sphere; and a formal account of the lone-pair repulsion hierarchy. A formal resonance theory would need labelled-graph isomorphism with bond-order averaging. Both are substantial modelling efforts with no current Mathlib support. lean_status: none; no lean_module ships with this unit.

Limitations and systematic failures of the Lewis-VSEPR model [Master]

The Lewis-VSEPR framework, despite its practical utility, has known systematic failures that any advanced treatment must acknowledge.

Failure for odd-electron species. ${\text{NO}}_2$ has 17 valence electrons and cannot be drawn with a clean Lewis structure. The molecule is a bent radical ($134^{\circ }$), but neither the Lewis picture nor standard VSEPR predicts the geometry reliably. The unpaired electron occupies a position analogous to a "half lone pair" in the VSEPR framework. Gillespie extended the model to treat radicals by assigning the odd electron a smaller domain than a full lone pair, predicting a bond angle between the bent (2 lone pairs, $104.5^{\circ }$) and bent (1 lone pair, $\approx 120^{\circ }$) geometries. The experimental $134^{\circ }$ falls in this range, but the prediction is qualitative at best. A complete treatment requires molecular orbital theory ^{[Gillespie 1972]}.

Failure for $\pi$-delocalised systems. Benzene (${\text{C}}_6{\text{H}}_6$) has two resonance structures, but the true electronic structure is a fully delocalised $\pi$ system with all C-C bonds equivalent (bond order 1.5, bond length 139 pm). VSEPR correctly predicts a planar hexagon but says nothing about the delocalisation energy ($\approx 150\text{kJ/mol}$ stabilisation) or the aromatic character that results from Hückel's $4n+2$ rule. The Lewis model treats each resonance structure as a discrete entity connected by double-headed arrows; the physical reality is a single quantum-mechanical state whose electron density is the weighted average of the contributing structures.

Failure for transition-metal complexes. $\text{Ni}(\text{CO}{)}_4$ is tetrahedral, but $\text{Ni}(\text{CN}{)}_4^{2-}$ is square planar. Both have the same steric number (4) and the same number of ligands, yet different geometries. The difference arises from $d$-orbital occupation and ligand-field effects: $d^8$ nickel(II) in a strong-field ligand environment gains crystal-field stabilisation energy by adopting square planar geometry, which is unavailable in the simple steric-number model. VSEPR does not address $d$-electron effects at all; the Kepert model for transition-metal complexes provides a parallel electrostatic treatment, but even it fails for strong-field cases where electronic effects dominate steric ones.

Formal charge vs. oxidation state. Formal charge is a bookkeeping device with no direct physical measurement. Oxidation state assigns all bonding electrons to the more electronegative atom. In ${\text{SO}}_4^{2-}$, the formal charge on S is 0 (in the resonance hybrid) but the oxidation state is $+6$. These are different quantities with different uses: formal charge guides Lewis-structure drawing; oxidation state tracks redox chemistry. Neither corresponds exactly to the real partial charge, which must be computed quantum-mechanically (Mulliken charges, Natural Population Analysis, or electrostatic-potential-derived charges).

The Gillespie-Nyholm model and its extensions. The original VSEPR model ^{[Gillespie & Robinson 1961]} has been extended to include the VSEPRPR (valence-shell electron-pair repulsion including polarisation and repulsion) model and to handle cases with $d$-electron effects. The underlying physics is electrostatic: the Pauli exclusion principle keeps electron pairs with the same spin apart, and classical Coulomb repulsion keeps all electron pairs apart. VSEPR is an approximation that captures the dominant geometric effect of these repulsions for main-group compounds. Gillespie and Hargitt's 1991 monograph ^{[Gillespie & Hargitt 1991]} presents the full treatment, including the ligand close-packing model that provides a complementary explanation for bond angles in terms of atomic radii rather than electron-pair repulsion.

Resonance, bond order, and delocalisation energy [Master]

The resonance concept, introduced by Pauling in 1928 and formalised in The Nature of the Chemical Bond (1939) ^{[Pauling 1960]}, is one of the most consequential ideas in structural chemistry. It addresses a fundamental limitation of the Lewis model: not all molecules can be adequately represented by a single classical structure.

Bond order from resonance. For a set of $n$ equivalent resonance structures, each contributing equally, the bond order of any bond is the arithmetic mean of its bond orders across all contributing structures. For ${\text{CO}}_3^{2-}$ (three equivalent structures, each with one C=O and two C-O bonds), every C-O bond has order $(2+1+1)/3=4/3$. This predicts equal bond lengths, confirmed by X-ray crystallography at 131 pm for all three C-O bonds.

For inequivalent contributors, the weighting is determined by relative stability: structures with fewer formal charges, negative charges on more electronegative atoms, and more bonds are weighted more heavily. Quantitative weighting requires either quantum-chemical calculation or empirical parametrisation from thermochemical data.

Resonance energy (delocalisation energy). The resonance stabilisation energy is the difference between the observed enthalpy of formation and the enthalpy predicted by the best single classical Lewis structure. For benzene, the resonance energy is $\approx 150\text{kJ/mol}$: the enthalpy of hydrogenation of benzene to cyclohexane ($\Delta H=-208\text{kJ/mol}$) is $150\text{kJ/mol}$ less exothermic than three times the enthalpy of hydrogenation of cyclohexene ($3\times -119.7=-359\text{kJ/mol}$). The discrepancy is the energy lowering from delocalisation.

Pauling's thermochemical resonance energies, compiled from hydrogenation and combustion data, provide a systematic catalogue for hundreds of conjugated and aromatic systems. The values correlate with the number of equivalent resonance structures and the degree of charge separation in the contributors, but the relationship is not simply additive. Modern quantum-chemical methods compute resonance energies directly from the difference between the full (delocalised) calculation and a localised-orbital reference.

The distinction between resonance and tautomerism. Resonance structures represent the same molecule with different electron arrangements; they are connected by curved-arrow electron pushing and cannot be isolated. Tautomers represent different molecules (different atomic arrangements, typically a proton shift) that exist in equilibrium and can, in principle, be isolated. The enol-keto tautomerism of acetone (${\text{CH}}_3{\text{COCH}}_3\rightleftharpoons {\text{CH}}_2{\text{=C(OH)CH}}_3$) involves an actual proton transfer; the two resonance structures of the acetate ion (${\text{CH}}_3{\text{COO}}^{-}$) involve only electron rearrangement.

Resonance in inorganic chemistry. The resonance concept extends beyond organic molecules. The nitrate ion, sulfate ion, phosphate ion, and perchlorate ion all exhibit resonance with multiple equivalent contributors. In ${\text{SO}}_4^{2-}$, the six resonance structures (two double bonds can be placed on any of the four oxygens in $\left(\genfrac{}{}{0ex}{}{4}{2}\right)=6$ ways, though these are not all symmetry-equivalent) distribute the double-bond character evenly, giving each S-O bond an order of approximately 1.5 and uniform bond lengths of 149 pm. The high oxidation state of sulfur (+6) is stabilised by this delocalisation.

In the cyanide ion ${\text{CN}}^{-}$, two resonance structures ($:\text{C}\equiv \text{N}{:}^{-}$ and ${}^{-}\text{C}=\text{N}:$) differ in formal-charge distribution. The triple-bond structure with the negative charge on carbon is the major contributor because it places the negative charge on the less electronegative atom and maximises the number of bonds. The minor contributor, with a C=N double bond and negative charge on nitrogen, is less stable by both criteria. Quantum-chemical calculations assign approximately 85-90% weight to the triple-bond structure, consistent with the observed C-N bond length of 116 pm (very close to the triple-bond value of 115 pm).

Valence bond vs. molecular orbital perspectives on resonance. In valence bond theory, the resonance hybrid is the normalised weighted sum of the wavefunctions for each contributing structure: ${\Psi }_{\text{hybrid}}=c_1{\Psi }_1+c_2{\Psi }_2+\cdots$. The coefficients $c_i$ are chosen to minimise the total energy, and the resulting energy is always lower than any single contributor. In molecular orbital theory, the same delocalisation appears automatically: the molecular orbitals extend over the entire conjugated system, and no discrete classical structures are invoked. The two descriptions are mathematically equivalent (they span the same Hilbert space) but offer different physical pictures.

Ligand close-packing and the VSEPR refinements [Master]

The VSEPR model has a competitor and complement in the ligand close-packing (LCP) model, developed by Gillespie and Robinson in the 1980s and presented fully in Gillespie and Hargitt's 1991 monograph ^{[Gillespie & Hargitt 1991]}. The LCP model explains molecular geometry not in terms of electron-pair repulsion but in terms of the sizes of the atoms (ligands) bonded to the central atom.

The LCP model. In this picture, each ligand has an effective radius (the ligand radius) determined by the central atom and the ligand identity. The bond angles around the central atom are determined by the requirement that the ligand spheres pack as closely as possible without overlapping. For four identical ligands (e.g., ${\text{CH}}_4$), the closest packing of equal spheres around a central point gives the tetrahedral angle $109.5^{\circ }$. For the ${\text{AX}}_3\text{E}$ case (e.g., ${\text{NH}}_3$), the lone pair is modelled as an invisible ligand with a different effective radius, and the angular distortion from tetrahedral is reproduced.

The LCP model predicts bond angles quantitatively from atomic radii and provides a physical basis for the VSEPR observation that more electronegative ligands produce smaller bond angles. Fluorine, being small and highly electronegative, has a small ligand radius. In ${\text{NF}}_3$ ($102^{\circ }$) the F ligands pack tightly, allowing the lone pair to compress them more than in ${\text{NH}}_3$ ($107^{\circ }$) where hydrogen's larger effective radius (in this context) resists compression.

Bond-length-bond-angle correlations. A concrete prediction of the LCP model is that shorter bonds should be associated with larger bond angles (tighter packing around the central atom). This is observed experimentally: in ${\text{OF}}_2$ (bond length 141 pm, bond angle $103^{\circ }$) vs. ${\text{OH}}_2$ (bond length 96 pm, bond angle $104.5^{\circ }$), the longer O-F bonds are associated with a smaller angle. More systematically, the model predicts that replacing a ligand with a larger one (e.g., substituting Cl for F in a series of molecules) should decrease the bond angle at the central atom, which is borne out across the halide series.

Kepert model for transition metals. The Kepert model (1982) extends the close-packing idea to transition-metal complexes by treating the metal-ligand bonds as points on a sphere and minimising the total repulsion energy using a repulsion law $E\propto r^{-n}$ where $n$ is an empirical parameter typically between 1 and 6. The model correctly predicts common coordination geometries (tetrahedral for 4-coordinate, octahedral for 6-coordinate) but fails for $d^n$ configurations where crystal-field stabilisation energy competes with steric packing. The Kepert model is the transition-metal analogue of VSEPR, with the same strengths (predictive for a wide range of geometries) and the same limitations (fails when electronic effects dominate steric ones).

VSEPR for steric numbers 7 and above. The standard VSEPR table stops at steric number 6, but steric numbers 7, 8, and 9 occur for heavy main-group elements and lanthanides. ${\text{IF}}_7$ (steric number 7) adopts a pentagonal bipyramidal geometry. ${\text{XeF}}_8^{2-}$ (steric number 8) is predicted to be square antiprismatic. These higher steric numbers produce multiple possible geometries with similar energies, and the actual structure is often influenced by crystal-packing effects and Jahn-Teller distortions rather than by electron-pair repulsion alone. The VSEPR predictions for steric number $\ge 7$ are less reliable than for steric numbers 2-6 because the energy differences between competing geometries are smaller.

Molecular dipole moments and physical-property prediction [Master]

A Lewis structure combined with VSEPR geometry determines whether a molecule has a net dipole moment, and this in turn predicts physical properties including boiling point, solubility, and spectroscopic behaviour.

Vector addition of bond dipoles. Each polar bond in a molecule contributes a bond dipole vector ${\vec{\mu }}_i$ directed from the positive partial charge toward the negative partial charge, with magnitude ${\mu }_i=q_i\times d_i$ where $q_i$ is the effective charge separation and $d_i$ is the bond length. The molecular dipole moment is the vector sum:

$${\vec{\mu }}_{\text{total}}=\sum _i{\vec{\mu }}_i.$$

For a symmetric molecule, the bond dipoles may cancel. ${\text{CO}}_2$ (linear, two C=O dipoles pointing outward from C) has ${\vec{\mu }}_{\text{total}}=0$ because the two bond dipoles are equal and opposite. ${\text{H}}_2\text{O}$ (bent, two O-H dipoles pointing toward O) has ${\vec{\mu }}_{\text{total}}\ne 0$ because the bond dipoles are not collinear. The experimental dipole moment of water is 1.85 D; the component along the $C_2$ symmetry axis is the resultant of two O-H bond dipoles at $104.5^{\circ }$.

Polarity and molecular geometry. The VSEPR geometry is the primary determinant of whether a molecule is polar. Two principles organise the prediction:

1. Any molecule with a symmetry plane that contains all atoms is nonpolar only if the remaining bond dipoles cancel by symmetry. ${\text{BF}}_3$ (trigonal planar, $D_{3h}$ symmetry) is nonpolar because the three B-F dipoles sum to zero in the molecular plane.

2. Any molecule lacking a centre of inversion and having polar bonds is likely polar. ${\text{NH}}_3$ (trigonal pyramidal, $C_{3v}$) has a net dipole of 1.47 D along the $C_3$ axis because the three N-H bond dipoles have a nonzero resultant, augmented by the contribution of the lone pair.

Physical-property correlations. Molecular dipole moments correlate with boiling points within homologous series: for molecules of similar molecular weight, the one with the larger dipole moment has the higher boiling point because of stronger dipole-dipole interactions. This is the basis for the "like dissolves like" rule in solubility: polar solvents (water, $\mu =1.85$ D) dissolve polar and ionic solutes, while nonpolar solvents (hexane, $\mu =0$) dissolve nonpolar solutes.

The hydrogen bond is a special case of dipole-dipole interaction that requires both a hydrogen atom bonded to a highly electronegative atom (N, O, or F) and a lone pair on another electronegative atom. The Lewis structure directly identifies hydrogen-bond donors (X-H bonds where X is N, O, or F) and acceptors (lone pairs on N, O, or F). Water can both donate and accept hydrogen bonds, which accounts for its anomalously high boiling point ($100^{\circ }\text{C}$) relative to ${\text{H}}_2\text{S}$ ($-60^{\circ }\text{C}$) despite their similar molecular weights.

Quadrupole and higher moments. Molecules with zero dipole moment may still have nonzero quadrupole moments that affect physical behaviour. ${\text{CO}}_2$ has zero dipole moment but a substantial quadrupole moment ($-4.3\times 10^{-40}\text{C}\cdot {\text{m}}^2$) arising from the charge separation within each C=O bond. This quadrupole moment contributes to the intermolecular interactions of ${\text{CO}}_2$ and explains why it is more soluble in water than nonpolar molecules of similar size. The Lewis structure reveals the quadrupole: each oxygen carries a partial negative charge, the carbon is partially positive, and the overall charge distribution is $(-\delta )(+2\delta )(-\delta )$.

Benzene provides another instructive case: zero dipole moment (by symmetry), but a large quadrupole moment arising from the $\pi$-electron density above and below the ring plane. The $\pi$ cloud is negatively charged; the ring carbons are positively charged. This quadrupole drives the stacking interactions between aromatic rings (parallel-displaced geometry) and the cation-$\pi$ interactions that stabilise protein-ligand complexes. Neither the Lewis structure nor VSEPR predicts the quadrupole; the delocalised $\pi$ system must be invoked.

From localised pairs to delocalised orbitals [Master]

The Lewis-VSEPR framework treats bonds as localised two-centre, two-electron (2c-2e) interactions. This is the valence-bond picture: each bond is a shared electron pair between two atoms. The molecular orbital (MO) picture, developed by Mulliken and Hund in the late 1920s ^{[Mulliken 1967]}, provides a fundamentally different description: electrons occupy orbitals that extend over the entire molecule.

Natural bond orbitals as a bridge. The NBO (Natural Bond Orbital) analysis, developed by Weinhold and Landis, provides a computational bridge between the Lewis and MO pictures. Starting from a full MO calculation, the NBO procedure transforms the delocalised molecular orbitals into a set of localised orbitals that closely resemble Lewis-type bonds and lone pairs. For most main-group molecules, the NBO analysis recovers the Lewis structure as the dominant electron-occupancy pattern, with small corrections for delocalisation. The occupancy of each NBO gives a quantitative measure of how well the Lewis picture describes the bonding: an ideal Lewis structure has exactly 2.000 electrons in each bond NBO and each lone-pair NBO. Real molecules show slight deviations (e.g., 1.98 electrons in a bond NBO, with the missing 0.02 electrons distributed into antibonding NBOs) that quantify resonance and charge-transfer effects.

Where the Lewis picture breaks down quantitatively. The Lewis model assigns integer bond orders (1, 2, or 3) and integer formal charges. Real molecules have non-integer bond orders and non-integer partial charges. The bond order in ozone is approximately 1.5, not 1 or 2; the charge on each oxygen is approximately $-0.5$ in the resonance hybrid, not $0$ or $-1$. MO theory computes these quantities directly from the electron density. The Wiberg bond index, computed from the density matrix, provides a bond-order measure that reduces to integer values for ideal Lewis structures but gives non-integer values for delocalised systems.

The bond-order concept has a precise mathematical definition in MO theory: for a bond between atoms A and B, the Wiberg index $W_{\text{AB}}={\sum }_{\mu \in A}{\sum }_{\nu \in B}(PS{)}_{\mu \nu }(PS{)}_{\nu \mu }$, where $P$ is the density matrix and $S$ is the overlap matrix. For a standard single bond, $W_{\text{AB}}=1.000$; for the C-C bonds in benzene, $W_{\text{CC}}=1.44$; for the S-O bonds in sulfate, $W_{\text{SO}}=1.33$. These values agree with the resonance-predicted bond orders to within a few percent for well-behaved molecules, and the residual discrepancy measures the extent to which the Lewis model oversimplifies the electron distribution.

Three-centre bonding and the limits of 2c-2e. The Lewis model's insistence on two-centre bonds fails for molecules with three-centre bonding. Diborane (${\text{B}}_2{\text{H}}_6$) has 12 valence electrons -- too few for seven 2c-2e bonds (which would require 14 electrons). The resolution is two B-H-B three-centre, two-electron (3c-2e) bonds, each involving one hydrogen bridging two boron atoms. The Lewis structure cannot represent 3c-2e bonds cleanly; the best it can do is draw a structure with B-H-B bridges marked by dashed lines or dots, departing from the standard notation. The 3c-2e bond is a natural construct in MO theory (a bonding combination of three atomic orbitals occupied by two electrons).

Similarly, the ${\text{I}}_3^{-}$ ion (Exercise 9) is better described by a 3c-4e model: two electrons in a bonding MO delocalised over all three iodines and two electrons in a nonbonding MO localised on the terminal iodines. The Lewis structure with its expanded octet on the central iodine is an attempt to represent this delocalised bonding in localised terms.

Synthesis. The Lewis-VSEPR framework is the foundational reason that chemists can reason about molecular structure without solving the Schrodinger equation. The central insight is that electron pairs localise around atoms and bonds in patterns that can be counted and arranged by simple electrostatic rules. Putting these together with the resonance concept, the Lewis model captures a remarkable fraction of structural and thermochemical behaviour across main-group chemistry. This is exactly the content that hybridization 14.02.02 pending generalises: the same electron-pair localisation that VSEPR uses to predict geometry appears again in orbital language as directed hybrid orbitals. The bridge is between the qualitative VSEPR prediction and the quantitative MO calculation: NBO analysis recovers the Lewis picture from MO theory, confirming that the two are not contradictory but are coarse-grained and fine-grained descriptions of the same electron density. The pattern recurs throughout chemistry -- Lewis structures predict, MO theory computes, and the resonance concept connects them.

Connections [Master]

• Electron configurations 14.01.01 determine how many valence electrons each atom contributes to the Lewis structure. The octet rule reflects the stability of filled $n{s}^{2}n{p}^6$ configurations. The VSEPR steric number depends on the electron count, which in turn depends on the ground-state electron configuration of the central atom.

• Hybridization 14.02.02 pending explains why VSEPR geometries arise at the orbital level. Four electron domains correspond to $s{p}^3$ hybridisation, three to $s{p}^2$, two to $sp$. The hybridization unit generalises the Lewis-VSEPR picture by providing a wavefunction description of the same bonding patterns.

• Molecular orbital theory 14.05.01 pending resolves the limitations of Lewis structures by providing delocalised bonding descriptions. MO theory treats the molecule as a whole rather than assigning electron pairs to individual bonds. The MO unit builds directly on the limitations catalogued here: electron-deficient molecules, radicals, $\pi$-delocalised systems, and transition-metal complexes.

• Organic functional groups 15.02.01 are classified and named using Lewis structures. The distinction between alcohols, ethers, aldehydes, ketones, carboxylic acids, and esters is a distinction between different Lewis-structural patterns around carbon and oxygen. Resonance in carbonyl groups, carboxylate ions, and aromatic rings is a central theme of organic chemistry that originates in the Lewis-resonance framework developed here.

• Electronegativity and bond polarity 16.01.01 pending determines bond polarity in Lewis structures and drives the unequal sharing of electrons in polar covalent bonds. The vector sum of bond dipoles determines molecular polarity, which in turn affects intermolecular forces and physical properties. The formal-charge vs. oxidation-state distinction developed here is the foundation for redox chemistry across the curriculum.

• Crystal-field and ligand-field theory [16.NN.NN] extends the electron-repulsion picture to transition metals, where $d$-orbital splitting by the ligand field (not steric number) determines geometry. The VSEPR failure for transition-metal complexes is the motivation for the ligand-field treatment.

Historical & philosophical context [Master]

G. N. Lewis introduced the electron-pair bond in 1916, in a paper titled "The Atom and the Molecule" ^{[Lewis 1916]}. Lewis proposed that atoms share electron pairs to achieve the stable electron configuration of the nearest noble gas. The "octet rule" emerged from this paper, though Lewis himself noted exceptions for elements beyond the second row. Irving Langmuir extended and popularised Lewis's ideas (1919-21) ^{[Langmuir 1919]}, and the Lewis-Langmuir theory became the dominant bonding model for two decades. Langmuir introduced the term "covalent bond" and formulated the counting rules that still bear his name.

The VSEPR model was developed by Ronald Gillespie and Ronald Nyholm in 1957 and refined by Gillespie and Robinson in 1961 ^{[Gillespie & Robinson 1961]}. Gillespie, a student of Linus Pauling at Caltech, formulated the model as a simplification of the electron-pair repulsion idea that Pauling had discussed informally. The model's strength is its predictive accuracy for main-group molecular geometries using only a Lewis structure as input -- no quantum-mechanical calculation required. Gillespie continued refining the model through four decades, culminating in the ligand close-packing extension presented with Hargitt in 1991 ^{[Gillespie & Hargitt 1991]}.

The conceptual tension between Lewis structures (localised electron pairs) and molecular orbital theory (delocalised electrons) was one of the central debates of 20th-century chemistry. Robert Mulliken's MO theory (1928 onward) and Friedrich Hund's parallel work offered a fundamentally different picture: electrons occupy orbitals that extend over the entire molecule. The debate was resolved not by one side winning but by recognising that both pictures are valid approximations to the same quantum-mechanical reality, each more useful in different contexts. Mulliken's 1967 retrospective ^{[Mulliken 1967]} frames this reconciliation. Lewis structures remain the default tool for drawing and reasoning about molecular structures in organic chemistry; MO theory dominates in spectroscopy, computational chemistry, and inorganic chemistry.

The resonance concept, introduced by Pauling in 1928 and formalised in The Nature of the Chemical Bond (1939, 3rd edition 1960) ^{[Pauling 1960]}, is the bridge between the two pictures. Resonance structures are contributing classical Lewis representations whose weighted average approximates the true quantum-mechanical ground state. The resonance hybrid is a single quantum state, not a time-average or an equilibrium. The thermochemical resonance energies that Pauling compiled provided the first quantitative evidence for delocalisation and motivated the development of Hückel molecular orbital theory for conjugated $\pi$ systems.

Bibliography [Master]

@article{Lewis1916,
  author = {Lewis, G. N.},
  title = {The Atom and the Molecule},
  journal = {J. Am. Chem. Soc.},
  volume = {38},
  year = {1916},
  pages = {762--785},
}

@article{Langmuir1919,
  author = {Langmuir, I.},
  title = {The Arrangement of Electrons in Atoms and Molecules},
  journal = {J. Am. Chem. Soc.},
  volume = {41},
  year = {1919},
  pages = {868--934},
}

@article{GillespieNyholm1957,
  author = {Gillespie, R. J. and Nyholm, R. S.},
  title = {Inorganic Stereochemistry},
  journal = {Q. Rev. Chem. Soc.},
  volume = {11},
  year = {1957},
  pages = {339--380},
}

@article{GillespieRobinson1961,
  author = {Gillespie, R. J. and Robinson, E. A.},
  title = {Electron-Pair Repulsion Model for Molecular Geometry},
  journal = {J. Chem. Educ.},
  volume = {38},
  year = {1961},
  pages = {507},
}

@book{Pauling1960,
  author = {Pauling, L.},
  title = {The Nature of the Chemical Bond},
  edition = {3rd},
  publisher = {Cornell University Press},
  year = {1960},
}

@book{GillespieHargitt1991,
  author = {Gillespie, R. J. and Hargitt, I.},
  title = {The {VSEPR} Model of Molecular Geometry},
  publisher = {Allyn and Bacon},
  year = {1991},
}

@article{Mulliken1967,
  author = {Mulliken, R. S.},
  title = {Spectroscopy, Molecular Orbitals, and Chemical Bonding},
  journal = {Science},
  volume = {157},
  year = {1967},
  pages = {13--24},
}

@book{Tro2023,
  author = {Tro, N. J.},
  title = {Chemistry: A Molecular Approach},
  edition = {6th},
  publisher = {Pearson},
  year = {2023},
}

@book{AtkinsdePaula2023,
  author = {Atkins, P. and de Paula, J.},
  title = {Physical Chemistry},
  edition = {12th},
  publisher = {Oxford University Press},
  year = {2023},
}

---

Cycle 4 Track B deepening. Original Wave 3 chemistry seed unit by claude-glm-5.1-runbook. Deepened to ≥8000 words with ≥4 Master H2s per CYCLE_4_STYLE_PARITY_PLAN §2. Status updated to shipped. All hooks_out targets are proposed. Catalog entry pre-exists at chemistry.lewis-structures-vsepr.