NMR Coalescence: How Two Peaks Merge as Rotation Speeds Up

What Coalescence Is and Where It Shows Up

Coalescence is the spectral signature of a molecule interconverting between two (or more) sites that have different chemical shifts. When the interconversion is slow on the NMR timescale, each site produces its own resonance. When it is fast, the nucleus samples both environments many times during a single measurement and reports a single, population-weighted average shift. Coalescence is the crossover point between these limits — the moment the two lines merge into one.

The phenomenon appears wherever an exchange scrambles magnetic environments:

• Hindered rotation about a partial double bond — the classic amide C–N bond in DMF, peptides, and carbamates.
• Ring inversion — cyclohexane's axial/equatorial proton swap (chair-flip).
• Conformational exchange and atropisomerism about biaryl bonds.
• Ligand fluxionality — Berry pseudorotation in PF5, Cope rearrangements in bullvalene.
• Protonation / tautomeric and host–guest binding exchange.

Because temperature tunes the rate, a variable-temperature (VT) experiment lets you walk the same molecule from the slow-exchange limit up through coalescence and into fast exchange.

The Mechanism: From Two Lorentzians to One

The physics is captured by the Bloch equations modified for chemical exchange, derived by H. M. McConnell in 1958 building on Gutowsky, Saika, and McCall (1953). Consider a nucleus hopping between site A and site B with equal populations, shift difference Δν (in Hz), and first-order rate constant k for the exchange.

• In slow exchange each line is a Lorentzian of natural width; the exchange adds broadening of Δν1/2(exchange) = k/π to each peak.
• As k rises, the two Lorentzians broaden, their inner edges fill in, and the valley between them shrinks.
• At coalescence the second derivative of the line shape at the midpoint goes to zero — the dip just vanishes, giving a single flat-topped signal.
• Beyond coalescence, the merged line narrows toward the average shift; its residual width Δν1/2 = π·Δν²/(2k) reports the (now fast) rate.

Solving the modified Bloch equation for where the central minimum disappears gives the coalescence condition k_c = π·Δν/√2. Everything downstream — the barrier, the stopwatch reading — flows from this one relation.

Key Quantities and a Worked DMF Example

Two equations do the heavy lifting. The coalescence rate for an equally-populated two-site singlet exchange is:

k_c = π·Δν/√2 = 2.22·Δν, where Δν is the peak separation in Hz measured in the slow-exchange limit.

Feeding k_c and the coalescence temperature Tc into the Eyring equation (with transmission coefficient κ = 1) gives the activation free energy:

ΔG‡ = 4.575·Tc·[10.32 + log₁₀(Tc/k_c)] cal/mol (equivalently ΔG‡ = RTc·ln(k_B·Tc/(h·k_c))).

Worked example — DMF C–N rotation: the two N-methyl singlets are separated by Δν ≈ 10 Hz on a 60 MHz spectrometer, and they coalesce near Tc ≈ 393 K (120 °C).

• k_c = 2.22 × 10 Hz ≈ 22.2 s⁻¹.
• ΔG‡ = 4.575 × 393 × [10.32 + log₁₀(393/22.2)] cal/mol.
• log₁₀(17.7) ≈ 1.248, so the bracket ≈ 11.57.
• ΔG‡ ≈ 4.575 × 393 × 11.57 ≈ 20 800 cal/mol ≈ 21 kcal/mol ≈ 88 kJ/mol.

Note Δν is field-dependent (in Hz), so k_c and Tc shift with spectrometer frequency, but ΔG‡ does not.

How It's Measured in Practice

The standard tool is a variable-temperature (VT) NMR experiment. A stream of heated or cooled dry nitrogen flows over the sample; the probe thermocouple is calibrated against the known chemical-shift separations of a neat methanol (low-T) or ethylene glycol (high-T) standard, typically to ±0.5 K.

• Acquire spectra in ~5–10 K steps across the exchange range.
• Identify Tc as the temperature at which the two peaks just merge into one flat-topped signal.
• Measure Δν at the slow-exchange limit (extrapolate — the separation can drift slightly with T).
• Apply k_c = 2.22·Δν and then the Eyring/Gutowsky expression for ΔG‡.

For more rigor, chemists perform full line-shape analysis: simulate the whole spectrum at each temperature with software (WINDNMR, DNMR3, gNMR, MEXICO) to extract k(T) over the entire range, then build an Eyring plot of ln(k/T) vs 1/T. Its slope gives ΔH‡ and its intercept ΔS‡, separating enthalpic and entropic contributions rather than reporting only ΔG‡ at the single point Tc. Uncertainty in ΔG‡ from the coalescence method alone is typically ±0.2–0.4 kcal/mol.

How Coalescence Differs From Its Cousins

Coalescence is often confused with unrelated line-broadening phenomena. Keep them distinct:

• vs. spin–spin coupling (J-splitting): Coupling multiplets are temperature-independent and their spacing (J, in Hz) is field-independent. Exchange broadening is strongly temperature-dependent and its raw separation (Δν, in Hz) scales with field.
• vs. T2 / natural line width: Instrumental broadening from field inhomogeneity or short T2 does not merge two peaks into one at higher temperature — only chemical exchange does.
• vs. saturation-transfer / EXSY: These 2D and selective-inversion methods measure the same exchange rates but in the slow regime, where coalescence has not yet occurred; they complement DNMR at slower rates.
• vs. Le Chatelier equilibrium shift: Coalescence probes the kinetics (rate of interconversion), not the position of an equilibrium. Populations set peak areas; the rate sets the line shape.

Coalescence lives specifically in the intermediate-exchange regime, roughly k on the order of 1–10³ s⁻¹ for typical Δν of a few to hundreds of Hz.

Range, Limits, and Famous Cases

The accessible barrier window is set by what rates coalescence can reach in a practical temperature range. For Δν of ~1–500 Hz, coalescence probes ΔG‡ roughly 5–25 kcal/mol (20–100 kJ/mol) — fast enough not to freeze out below −100 °C, slow enough not to average out below the solvent's boiling limit.

• Amides (DMF, DMA): the textbook case; the partial C=N double-bond character (~40%) from n→π* donation raises the rotation barrier to ~18–21 kcal/mol.
• Cyclohexane: ring inversion, ΔG‡ ≈ 10.3 kcal/mol; the axial/equatorial protons coalesce near −60 °C in the ¹H spectrum.
• Bullvalene: Doering–Schröder's fluxional molecule interconverting among more than 1.2 million degenerate structures via rapid Cope rearrangements (rate ≈ 4000 s⁻¹ at room temperature) — its 10 protons average to a single line.
• Biphenyls / BINOL: atropisomer interconversion barriers of 20–25 kcal/mol define whether a chiral axis is configurationally stable.

Limits: unequal populations, overlapping multiplets, and strong coupling complicate the simple k_c = 2.22·Δν formula, and require full line-shape simulation. Still, coalescence remains the fastest way to read a conformational barrier straight off a spectrum.

Exchange regimes in dynamic NMR for a two-site system with separation Δν and rate constant k

Regime	Rate vs. Δν	Appearance in spectrum	What you extract
Slow exchange	k ≪ πΔν/√2	Two sharp separate peaks; slight broadening as T rises	Δν, populations, individual site lifetimes
Approaching coalescence	k < πΔν/√2	Peaks broaden and lean toward each other	Line widths → k via Δν1/2
Coalescence	k_c = πΔν/√2	One flat-topped / just-merged broad signal	k_c directly, then ΔG‡ at Tc
Fast exchange	k ≫ πΔν/√2	One sharp line at population-weighted average shift	Averaged δ, residual broadening → k