Curved Arrows for Resonance

Correct curved-arrow resonance bookkeeping underlies drug design at the most basic level: medicinal chemists routinely reason about which nitrogen in a heterocyclic drug scaffold is basic (protonatable) versus which is amide-like and non-basic, purely from whether that nitrogen's lone pair is delocalized into an adjacent $\pi$ system — get this wrong and you mispredict a compound's ionization state, membrane permeability, and plasma protein binding at physiological pH. The same reasoning explains why the guanidinium group of arginine side chains is such an effective, always-protonated cationic center in proteins: three resonance structures delocalize the positive charge symmetrically over three nitrogens, just as the acetate mechanism above delocalizes negative charge over two oxygens, making the cation both highly stable and a poor base to deprotonate. In materials science, curved-arrow resonance reasoning is the qualitative entry point to understanding conjugated polymers used in organic light-emitting diodes and organic photovoltaics (polyacetylene, polythiophene, and their derivatives): the continuous chain of aligned p orbitals that legalizes arrow-pushing across many repeat units is the same continuous overlap responsible for the delocalized band-like electronic states that give these polymers useful conductivity and optical absorption, a connection explored quantitatively once you move from resonance structures to true Conjugation: Overlapping p Orbitals and molecular orbital theory. In synthetic methodology, the ambident reactivity of enolates (Example 1 above) is exploited constantly — directing alkylation to carbon versus silylation to oxygen (forming a silyl enol ether) is a matter of controlling which resonance contributor's reactive site is exposed to which electrophile, under kinetic versus thermodynamic control. Nitro groups, carbamates, ureas, and carbonate esters are all rationalized, and their relative reactivities predicted, by the same two-arrow, formal-charge-audited procedure developed here, making this small piece of notation one of the most heavily reused pieces of technical vocabulary in the whole of organic chemistry, useful again the moment you need to rank competing resonance contributors by importance in Evaluating Resonance Structures.

Ask the tutor: How do I decide which of several legitimate resonance structures contributes most to the real hybrid?

A Lewis structure is a cartoon of a wavefunction: it pretends that bonding and lone-pair electrons sit in fixed, localized two-center or one-center orbitals. For most molecules that cartoon is a good approximation. But whenever a $\pi$ system is flanked by an adjacent empty orbital, an adjacent lone pair, or another $\pi$ bond, the true molecular orbitals are delocalized over three or more atoms, and a single localized picture cannot capture the electron distribution correctly. Resonance theory patches the cartoon by drawing several localized structures — the resonance structures or contributing structures — and asserting that the real molecule is a weighted quantum superposition of them, the resonance hybrid. The curved arrow is simply the algorithm for generating the next contributing structure from the current one. Mechanically: choose a source of electron density (a $\pi$ bond, or a lone pair sitting in a p orbital aligned with a $\pi$ system), draw an arrow whose tail sits on that pair and whose head points to the destination, and redraw the structure with the electrons in their new position. The head of the arrow must land on an atom, and it must represent either (a) formation of a new $\pi$ bond to that atom, or (b) formation of a new lone pair on that atom. Nothing else is permitted. In particular, you never draw an arrowhead that terminates in empty space, and you never move a single dot (one electron) with the standard double-barbed arrow — that symbol is reserved for full electron pairs. Orbital language makes the rule precise. Delocalization is possible only when a p orbital (holding a $\pi$ bond, a lone pair, or empty) is parallel to an adjacent p orbital, so that the two can overlap sideways and mix into delocalized $\pi$ molecular orbitals. A curved arrow is a shorthand for "let this p-orbital's electron density mix into the neighboring, aligned p orbital system." If the neighboring orbital is not a p orbital aligned for $\pi$ overlap — for instance, if it is a $\sigma^*$ orbital of a saturated C–H bond pointing the wrong way, or if free rotation has twisted the p orbital out of alignment — no resonance arrow can be drawn there, because there is no physical overlap to represent.

Resonance structures are not different molecules; they are different partitionings of the same total electron count among the same fixed set of atomic positions. Three bookkeeping identities must hold across every valid pair of resonance structures $\text{A}$ and $\text{B}$ connected by curved arrows:

where: $N_{\text{atoms}}$ is the count and connectivity of atoms — the skeleton is frozen, so this identity is trivial but is the first thing to check when you suspect an illegal arrow (if atoms moved, it's a different compound, not resonance).

where: $q_{\text{total}}$ is the overall (net) charge of the species — resonance redistributes charge among atoms but the sum is conserved, since arrows only relocate existing electron pairs, never create or destroy electrons.

where: $FC_i$ is the formal charge on atom $i$, $Z_i$ its number of valence electrons as a free atom, $L_i$ the number of nonbonding (lone-pair) electrons it carries in that particular resonance structure, and $B_i$ the number of electrons it shares in bonds (so $\tfrac12 B_i$ is "its half" of every bonding pair). This is the formula you re-evaluate at every atom, every time an arrow moves a pair, to confirm the new structure is charge-balanced and to locate which atom becomes more electron-rich or electron-poor. Reading the formal-charge equation: each curved arrow head that lands a new lone pair on an atom increases $L_i$ by 2 (formal charge drops by 1 relative to before); each arrow head that forms a new bond to an atom (without loss of an existing lone pair pair) increases $B_i$ by 2 while decreasing $L_i$ by 2 at the atom that donated the pair, shifting formal charge by $\mp 1$ at the two ends respectively. This is exactly how you track, arrow by arrow, that acetate's two oxygens swap between $FC=-1,\;L=3,\;B=2$ (single-bonded O⁻) and $FC=0,\;L=2,\;B=4$ (double-bonded O), while carbon and the total charge never change. A fourth, purely topological rule governs where arrows are legal at all: an arrow's tail must sit on a $\pi$ bond or a lone pair that occupies (or can rotate into) a p orbital aligned with an adjacent, continuous system of parallel p orbitals; its head must land on an atom belonging to that same contiguous p-orbital system. This is the "adjacency and alignment" test, and it is what rules out pushing electron density from an ordinary alkane C–H $\sigma$ bond, or across a $sp^3$ (tetrahedral, non-conjugated) atom that has no p orbital available to relay the electrons onward.

Example 1 (foundational): the enolate of acetone. Deprotonation of acetone, $(CH_3)_2C=O$, at the $\alpha$-carbon by a base gives the enolate anion $(CH_3)(CH_2^-)C=O \leftrightarrow (CH_3)(CH_2=)C-O^-$. Start from the carbanion form: $C_\alpha$ carries a lone pair (from having lost $H^+$) and is bonded to the carbonyl carbon $C_1$, which bears a $C_1{=}O$ $\pi$ bond. Check alignment: is $C_\alpha$'s lone pair in a p orbital parallel to the $C_1{=}O$ $\pi$ system? Yes — deprotonation occurs preferentially from a conformation where the $C_\alpha$–H (now lone pair) axis is perpendicular to the carbonyl plane, i.e., the resulting lone pair sits in a p orbital aligned with $C_1=O$. Draw the arrow: tail on the $C_\alpha$ lone pair, head forming a new $\pi$ bond $C_\alpha{=}C_1$. Because $C_1$ cannot hold five bonds, the pre-existing $C_1{=}O$ $\pi$ bond must simultaneously curl onto oxygen as a new lone pair (second arrow, tail on $C_1=O$, head on $O$). Audit: before, $FC(C_\alpha)=-1$ ($Z=4, L=2, B=6 \Rightarrow 4-2-3=-1$), $FC(O)=0$ ($Z=6,L=4,B=4\Rightarrow 6-4-2=0$); after, $FC(C_\alpha)=0$ ($4-0-4=0$), $FC(O)=-1$ ($6-6-1=-1$). Total charge $-1$ before and after — conserved. This pair of structures is the reason enolates react as ambident nucleophiles, attacking electrophiles through either carbon (giving the $\alpha$-alkylated ketone) or oxygen (giving the enol ether), a duality you will meet again when ranking these two resonance contributors for relative importance.

Example 2 (applied): the amide bond and restricted rotation. Formamide, $H_2N{-}CHO$, is drawn conventionally with a full $C{=}O$ double bond and a lone pair localized on nitrogen. But nitrogen's lone pair sits in a p orbital, and that p orbital is parallel to the carbonyl $\pi$ system, because the nitrogen is (nearly) planar $sp^2$, not pyramidal $sp^3$, in the ground state. Draw the arrow: tail on the nitrogen lone pair, head forming a new $\pi$ bond $N{=}C$; simultaneously, arrow tail on the pre-existing $C{=}O$ $\pi$ bond, head landing as a new lone pair on $O$. Audit: nitrogen goes from $FC=0$ ($Z=5,L=2,B=6\Rightarrow5-2-3=0$) to $FC=+1$ ($5-0-4=+1$); oxygen goes from $FC=0$ ($6-4-2=0$) to $FC=-1$ ($6-6-1=-1$); total charge $0$ throughout, conserved. This second structure, $H_2\overset{+}{N}=CH-O^-$, is a real, significant contributor (not the major one, but far from negligible) — it is precisely why the C–N bond in an amide has substantial double-bond character (rotation barrier around $75$–$85\ \text{kJ mol}^{-1}$, versus a negligible barrier for a genuine C–N single bond in an amine), why amide nitrogen is a poor base and poor nucleophile compared with amine nitrogen (its lone pair is partly tied up in the $\pi$ system, delocalized as this arrow shows), and why the amide oxygen, not nitrogen, is the site protonated under strongly acidic conditions. As a limiting-case check: if you artificially twist the N–C(=O) dihedral to $90°$ (as strain forces in some bridged bicyclic amides), the nitrogen lone pair's p orbital rotates out of alignment with the carbonyl $\pi$ system, the arrow becomes illegal by the adjacency test, resonance stabilization vanishes, and such twisted amides indeed show sharply increased reactivity (faster hydrolysis, IR carbonyl stretch shifted toward normal ketone values) — exactly consistent with loss of the delocalization the arrow represents.

Example 3 (exam/competition level): furan versus cyclopentadienyl anion — counting arrows around a ring, non-round formal charges. Consider the cyclopentadienide anion, $C_5H_5^-$, an aromatic ring with six $\pi$ electrons delocalized over five carbons. Starting from one localized Lewis structure (say, negative charge and a lone pair localized on $C_1$, with alternating double bonds $C_2{=}C_3$ and $C_4{=}C_5$), generate all symmetry-related resonance structures by chasing the lone pair around the ring one bond at a time. Arrow set A: tail on the $C_1$ lone pair, head forming $\pi$ bond $C_1{=}C_2$; simultaneously tail on $C_2{=}C_3$, head landing as new lone pair on $C_3$. Check alignment at each step: since all five ring carbons are $sp^2$ and coplanar (this is the aromatic ring geometry), every lone pair and every $\pi$ bond around the ring sits in mutually parallel p orbitals, so the adjacency test is satisfied at every single bond in the ring — this is precisely what "fully conjugated cyclic system" means mechanically. Repeating the two-arrow shift four more times walks the negative charge and lone pair sequentially onto $C_3$, then (by a different two-arrow path) onto $C_4$, then $C_5$, then $C_2$, generating five degenerate resonance structures, one for each ring position. Formal charge audit at each stage is identical by the symmetry of the shift ($FC=-1$ at the lone-pair carbon, $FC=0$ elsewhere, total charge $-1$ throughout every structure) — but the true hybrid divides that single unit of negative charge over five equivalent carbons, so each ring carbon carries an average formal charge of exactly

per carbon — a genuinely non-round, asymmetric-looking number that is nonetheless required by the fivefold symmetry, and it is confirmed by NMR: all five ring protons in cyclopentadienide are equivalent by symmetry and appear as a single upfield-shifted singlet, exactly as five equal partial charges (rather than one full charge fixed at one position) would predict. Contrast this with furan, where one ring position is oxygen: oxygen's lone pair (in the p orbital aligned with the ring) can be pushed around by the identical arrow-chasing procedure, generating structures with formal charge $+1$ on oxygen and $-1$ distributed onto the four ring carbons — but here the four carbon positions are not all equivalent (two are adjacent to O, two are one bond further), so the resonance structures are not all degenerate, and the smaller, unequal coefficients that result are the mechanistic reason furan is far less aromatic-stabilized and far more reactive (e.g., toward Diels–Alder cycloaddition as a diene) than benzene.

The curved-arrow formalism for resonance is inseparable from the broader development of electronic structural theory in organic chemistry during the 1920s–1930s, and its two central intellectual fathers are usually named as Robert Robinson, who introduced curved-arrow notation for depicting electron movement in organic reactions and structures in Britain, and Christopher Ingold, who — working in close (often contentious) parallel — built the rigorous physical-organic vocabulary of mesomerism, inductive and mesomeric effects, and reaction mechanism classification that turned Robinson's diagrams into a predictive, quantitative discipline.