Choosing Best Lewis Structures

Choosing the correct dominant Lewis structure is not a museum-piece skill confined to introductory courses — it underlies quantitative predictions used across chemistry. In atmospheric chemistry, correctly identifying that nitrate's three N–O bonds are equivalent (via the resonance/symmetry argument above) is essential to modeling its role in acid rain and aerosol nucleation, where bond strength and reactivity toward radicals depend on the true, averaged bond order, not a single arbitrarily drawn structure. In biochemistry, the carboxylate group (–COO⁻) in amino acid side chains and in citrate, acetate, and countless drug molecules is correctly described only by two degenerate, formal-charge-and-electronegativity-favored resonance structures, both placing negative charge on oxygen — this delocalization is why carboxylic acids are markedly more acidic than simple alcohols, since the conjugate base is stabilized by genuine charge delocalization onto electronegative oxygens rather than sitting on a single carbon. In materials and coordination chemistry, correctly ranking Lewis structures for oxo-anions and ligands (thiocyanate binding through S or N to a metal center, for example) depends on knowing which resonance structure — and hence which terminal atom — carries more negative character and is thus the better donor site; this directly feeds into the "ambidentate ligand" problem covered when studying coordination compounds. In computational and structural chemistry, the classical formal-charge/electronegativity ranking is now cross-checked against natural bond orbital (NBO) analysis and computed electrostatic potential maps, and the two methods agree remarkably well for simple main-group anions — a reassuring validation that a purely pencil-and-paper 1916–1930s-era heuristic still tracks real quantum-mechanical charge distributions computed with modern density functional theory. Finally, this entire selection logic is the essential stepping stone to two further topics: recognizing when the octet rule itself must be relaxed (treated under Hypervalency and Expanded Octets), and using the finally-selected best Lewis structure as the correct starting point for predicting molecular geometry via electron-pair repulsion (the subject of VSEPR Theory Fundamentals).

Ask the tutor: Why does thiocyanate (SCN⁻) genuinely have two significant resonance contributors while cyanate (OCN⁻) is dominated by essentially one — even though both ties involve similarly modest electronegativity differences?

Formal charge is computed by pretending every bonding pair is split exactly in half between the two atoms, regardless of which atom is more electronegative:

where $V$ is the number of valence electrons the free atom has, $N$ is the number of non-bonding (lone pair) electrons on that atom in the structure, and $B$ is the number of electrons in bonds to that atom. This is a purely electron-counting exercise — it does not know that oxygen pulls harder on electrons than carbon does. That is precisely its power and its limitation: formal charge tells you how far a given Lewis structure has strayed from an idealized equal-sharing picture, atom by atom, and a structure that requires large or numerous deviations from equal sharing is one that requires a large, artificial redistribution of electron density relative to what covalent bonding "wants" by default. But real atoms are not equally electronegative, so nature does not actually enforce equal sharing. This is where the second criterion enters. If a Lewis structure is forced to carry formal charges (because no all-neutral structure exists, or because minimizing formal charge already leaves residual charges), then those residual formal charges should be assigned to the atoms that are electronically comfortable holding them. An electronegative atom such as oxygen or fluorine has a low-lying, contracted valence shell and stabilizes extra electron density (negative formal charge) well. An electropositive atom such as carbon, boron, or a metal destabilizes extra negative charge but can better tolerate being formally short of electrons (positive formal charge), because its valence orbitals are higher in energy and less electron-greedy to begin with. The two rules are not independent — they are ordered. You always minimize the number and size of formal charges first, because fewer/smaller formal-charge deviations mean a structure closer to the "natural," undistorted covalent-bonding pattern implied by the atoms' valences. Only once you have a shortlist of equally formal-charge-minimal structures do you break the tie using electronegativity, because among structures that are equally far from ideal covalent sharing, the one whose charge separation matches the atoms' real electron affinities will be lower in energy and closer to the true distribution.

The tie-breaking step deserves a physical picture of its own. Think of formal charge as an accounting artifact and electronegativity as the thermodynamic reality correcting that artifact. A carbon–oxygen double bond with an extra lone pair pushed onto carbon (giving carbon a formal negative charge) is a poor structure not because the electron count is wrong, but because carbon's low electronegativity means it does not want that extra density — the real molecule minimizes energy by keeping the electron density where it is energetically happiest, which is on oxygen. Rule 2 is a shortcut for "put the charge where the actual electrostatic/orbital energy is lowest," verified time and again by measured dipole moments, bond lengths, and computed atomic charges (e.g., NBO or Mulliken charges) that consistently correlate with the electronegativity-favored resonance structure.

The formal charge formula, restated for clarity as the operational tool:

where: $V_{\text{atom}}$ is the group-number valence electron count of the free neutral atom, $L_{\text{atom}}$ is the number of electrons the atom holds as lone pairs in that structure, and $S_{\text{atom}}$ is the number of electrons the atom shares in all its bonds (i.e., twice the number of bonds, counting each bond in a double bond separately).

Reading: formal charge answers "if I split every bond exactly 50/50, does this atom end up with more or fewer electrons than it started with as a neutral atom?" A structure is preferred when this hypothetical redistribution is as small as possible everywhere — a signal that the structure is close to the "natural" bonding pattern implied by each atom's own valence. The ranking criterion combines this with electronegativity difference. Define, for a bond between atoms X and Y,

where: $\chi_X, \chi_Y$ are Pauling (or Allred–Rochow) electronegativities of the two bonded atoms.

Reading: $\Delta\chi$ measures how far real electron sharing deviates from the formal-charge assumption of equal sharing; large $\Delta\chi$ means the electronegative partner really does hold more electron density than the 50/50 formal-charge bookkeeping assumes, which is exactly why residual formal negative charge belongs there. Putting the two together, the composite "energy penalty" heuristic for ranking a candidate structure $i$ can be written qualitatively as

where: the first sum penalizes any nonzero formal charge (weighted quadratically, so ±2 is penalized much more than two separate ±1's), $\lambda$ is a positive weighting constant, and the second sum rewards structures that place negative formal charge on atoms with large electronegativity relative to their bonded neighbors, lowering the overall penalty $P_i$. The best structure is the one minimizing $P_i$.

Reading: this is not a formula you calculate numerically in practice (there's no universally agreed $\lambda$) — it's the compact statement of the two-tier logic: quadratic formal-charge penalty dominates first (so it is checked first and structures failing it are discarded outright), and electronegativity match is the secondary correction used only to discriminate among formal-charge-degenerate survivors.

Example 1 — Foundational: the thiocyanate ion, SCN⁻ Thiocyanate has 5 (N) + 4 (C) + 6 (S) + 1 (extra electron for the negative charge) = 16 valence electrons. Three candidate skeletons (all with the atoms connected S–C–N, the correct connectivity, confirmed experimentally): Structure I: S=C=N⁻ (cumulated double bonds, lone pair count: S has 2 lone pairs, C has none, N has 2 lone pairs plus the negative charge lone pair — total valence electrons check: S(2 lp × 2 = 4) + 2 bonds(4 shared) + C(2 bonds ×2 each =4 shared, 4 shared) + N(2 bonds shared =4, 2 lone pairs =4) — let's do formal charges directly. For S=C=N⁻:

Sum of |FC| = 1. Structure II: ⁻S–C≡N (single bond S–C, triple bond C≡N):

Sum of |FC| = 1. Structure III: S≡C–N²⁻ (triple bond S≡C, single bond C–N, with N doubly negative and S positive — deliberately a poor structure):

Sum of |FC| = 3 — immediately rejected by Rule 1; it is not even considered further. Structures I and II tie at Σ|FC| = 1. Now apply Rule 2: Pauling electronegativities are $\chi_S = 2.58$, $\chi_N = 3.04$. Structure I places the −1 charge on N (χ = 3.04); Structure II places the −1 charge on S (χ = 2.58). Since N is more electronegative than S, Structure I (S=C=N⁻) is favored by electronegativity — but the comparison is close (Δχ only 0.46), so both contribute significantly to the real resonance hybrid, with experiment showing SCN⁻ bond lengths (C–S ≈ 1.65 Å, C–N ≈ 1.15 Å) closer to the S=C=N⁻ picture, confirming it as the dominant, though not exclusive, contributor. This example is a genuine, well-documented case where two structures are both important — the criteria correctly identify the dominant one without claiming the other contributes nothing.

Limiting case check: if S and N had equal electronegativity (hypothetically Δχ → 0), Rule 2 would give no preference at all, and both structures should contribute equally to the hybrid — which is exactly the qualitative situation seen here, since Δχ = 0.46 is small and both structures are known experimentally to matter (unlike OCN⁻, where Δχ = 0.40 between O and N might seem similar, but the connectivity constraint, O terminal vs N terminal, breaks the near-tie more decisively in that ion because the favored structure also satisfies better bond-length matching).

Example 2 — Applied: nitrate ion, NO₃⁻, relevant to atmospheric and soil chemistry Nitrate is central to the nitrogen cycle and atmospheric aerosol chemistry. Total valence electrons: N(5) + 3×O(6) + 1(charge) = 24. Consider the structure with one N=O double bond and two N–O single bonds (each terminal O with the single bond bearing 3 lone pairs, the double-bonded O bearing 2 lone pairs):

Sum of |FC| = 1 + 0 + 1 + 1 = 3, but note this is the minimum possible for this ion — you cannot write an all-neutral Lewis structure for NO₃⁻ that also satisfies the octet rule on every atom, because nitrogen has only 5 valence electrons and must form 3 bonds plus accommodate the extra negative charge somewhere; the arithmetic of a 24-electron, 4-atom, 3-bond-to-N structure forces at least a +1/−1/−1 pattern (or equivalent). Three equivalent resonance structures exist (one for each choice of which oxygen bears the double bond), and by symmetry all three are exactly degenerate — a textbook case of true resonance rather than one dominant structure. This matters practically: it explains why all three N–O bonds in NO₃⁻ are experimentally identical, each 1.24 Å, intermediate between a single (~1.36 Å) and double (~1.20 Å) N–O bond, and why the ion is perfectly trigonal planar (D₃ₕ) rather than showing one short and two long bonds. Applying Rule 2 here: nitrogen (χ = 3.04) bears the +1, and the electronegative oxygens (χ = 3.44) bear the −1 charges — exactly the electronegativity-favored assignment, reinforcing that this is about as good as a Lewis structure can get for this ion, and the true structure is the equally-weighted average (a genuine "resonance hybrid," not one dominant contributor with minor others, unlike SCN⁻).

Sanity/limiting check: if you tried instead to draw NO₃⁻ with two double bonds to nitrogen and one single bond (forcing nitrogen to exceed an octet, or alternatively leaving oxygen electron-deficient), you would either violate the octet rule without justification (nitrogen, a second-period element, cannot expand its octet — no accessible low-lying d orbitals, and even by modern MO arguments no valence expansion is physically available) or produce even larger formal charges. Both alternatives are rejected outright: the octet-rule constraint on nitrogen combines with Rule 1 to uniquely fix the correct family of three resonance structures.

Example 3 — Exam/competition level: the azide ion, N₃⁻, and a common trap structure Azide, N₃⁻, has 3×5 + 1 = 16 valence electrons, linear, D∞h symmetry in its symmetric resonance structures. The textbook-correct structure is the symmetric cumulated-double-bond form:

Formal charges: central N,

each terminal N (2 lone pairs, one double bond):

This gives a symmetric −1, +1, −1 pattern, Σ|FC| = 3, and by the molecule's symmetry this exact structure and its mirror image (swap which terminal N you call "left") are indistinguishable — a single resonance structure up to relabeling, contributing dominantly, consistent with the experimentally observed N–N bond lengths in azide salts, both close to 1.16 Å (essentially equal, consistent with two equivalent N=N-like bonds each of order roughly 2).

The trap: students are often tempted to draw the "linear with one triple bond and one single bond" structure, N≡N–N²⁻ or its mirror ⁻²N–N≡N, reasoning that a triple bond "uses more electrons productively." Check the formal charges:

Sum of |FC| = 0 + 1 + 2 = 3 — numerically tied with the symmetric structure on Σ|FC|! This is exactly the kind of case where students stop too early. The correct next check is magnitude concentration: Rule 1, properly stated, penalizes concentrating charge on one atom (a −2) more than spreading equal magnitude across two atoms (two separate −1's), because of the quadratic-in-magnitude nature of electrostatic self-energy noted in the core-ideas section:

so even though the linear, sum, Σ|FC|, is the same (3, since |+1|+|0|+|−2| = 3 in both structures if you're only summing absolute values naively), the quadratic penalty used to discriminate correctly identifies the symmetric −1/+1/−1 structure (quadratic penalty $1+1+1=3$) as superior to the asymmetric 0/+1/−2 structure (quadratic penalty $0+1+4=5$). This refinement — never concentrate formal charge magnitude greater than 1 on a single atom if a spread-out alternative with the same total Σ|FC| exists — is the correct resolution, and it is confirmed by the experimental bond-length equivalence in azide salts (both N–N distances equal at ~1.16 Å), which rules out the asymmetric triple/single-bond structure as dominant (that structure predicts markedly unequal bond lengths, ~1.10 Å for the triple bond and ~1.45 Å for the "single" bond, which is not observed).

Limiting case check: if instead you consider a hypothetical isoelectronic species with a genuinely different central atom electronegativity — e.g., replacing the central N with a hypothetically much more electronegative atom — the −2 charge landing on the electronegative terminal atom would become progressively more favorable, illustrating that Rule 2 and the quadratic-concentration refinement can in principle trade off; for azide, all three atoms are identical (N), so no electronegativity tie-break is even possible, and the pure charge-distribution (quadratic minimization) argument must carry the full weight — exactly what the symmetric resonance picture reflects.