You’ve located the transition state, confirmed it with IRC, computed the Gibbs free energy of activation. Now what? Translating ΔG‡ into a rate constant — and understanding what that rate constant means for experimental predictions — requires a more careful treatment of transition state theory than most DFT workflows document explicitly. This post covers the Arrhenius and Eyring formulations, how DFT outputs map onto each, and the approximations that most often produce disagreements between computed and experimental rates.
Transition State Theory: The Eyring Framework
Classical transition state theory (TST) assumes that reactants are in quasi-equilibrium with a transition state species at the top of the reaction barrier, and that crossing the barrier is irreversible on the timescale of molecular vibration. The Eyring (Evans-Polanyi-Eyring) expression for the rate constant is:
k_TST = (κ × k_B × T / h) × exp(−ΔG‡ / RT)where k_B is the Boltzmann constant (1.380649 × 10⁻²³ J/K), h is Planck’s constant (6.62607 × 10⁻³⁴ J·s), T is temperature in Kelvin, R is the gas constant (1.987 cal/mol·K or 8.314 J/mol·K), ΔG‡ is the Gibbs free energy of activation, and κ is the transmission coefficient (often set to 1 for simple reactions, discussed below).
The pre-exponential factor k_B·T/h evaluates to approximately 6.25 × 10¹² s⁻¹ at 298 K (or equivalently, about 6.25 × 10¹² M⁻¹s⁻¹ for bimolecular reactions when combined with the partition functions). The full Eyring equation is then a function of ΔG‡ alone, given T.
Numerically: at 298 K, each 1.364 kcal/mol increase in ΔG‡ reduces the rate constant by a factor of 10. This is the "10× per 1.4 kcal/mol" rule that gives you quick mental estimates of barrier height changes needed to see rate changes in experimental assays.
What DFT Provides: ΔG‡ Components
A DFT frequency calculation at the TS geometry provides the following thermodynamic quantities (at the harmonic oscillator/rigid rotor approximation):
- Zero-point energy (ZPE): E_ZPE = ½ × Σ_i ν_i × h, summed over all real vibrational frequencies. The TS has one imaginary frequency excluded from this sum.
- Thermal enthalpy correction H(T): Includes translational (3/2 RT), rotational (3/2 RT for nonlinear molecules), and vibrational contributions. ZPE + thermal corrections give the total enthalpy correction ΔH_corr.
- Entropy S(T): Vibrational, rotational, and translational entropy. The vibrational entropy is the most sensitive to low-frequency modes (below ~100 cm⁻¹).
The Gibbs free energy of activation is then:
ΔG‡ = ΔE_electronic + ΔZPE + ΔH_thermal(T) − T × ΔS(T)The key point: the entropy term T·ΔS(T) can be 2–5 kcal/mol for bimolecular reactions at 298 K, because two species combine into one TS (loss of translational and rotational degrees of freedom). Never substitute ΔE_electronic for ΔG‡ in the Eyring equation — the systematic error for bimolecular reactions is 2–5 kcal/mol, corresponding to a rate error of 30–3000×.
The Transmission Coefficient κ
The transmission coefficient κ captures two physical effects that classical TST ignores:
- Quantum mechanical tunneling: For reactions involving hydrogen atom or proton transfer, tunneling through the barrier can substantially increase the rate relative to the classical prediction, especially at low temperatures. The Wigner correction (κ ≈ 1 + u²/24, where u = hν_im/k_BT and ν_im is the magnitude of the imaginary frequency) provides a simple first-order estimate. For H-transfer reactions at 298 K with imaginary frequencies of 1000–2000 cm⁻¹, the Wigner correction gives κ ≈ 1.2–2.0. More rigorous tunneling corrections (Bell correction, instanton methods) are needed when tunneling dominates.
- Recrossing: Some trajectories that reach the TS cross back to reactants without forming products. κ = 1 assumes no recrossing. Dynamical corrections from trajectory calculations suggest recrossing factors of 0.8–1.0 for most simple reactions in solution, making this a secondary correction compared to tunneling.
For heavy-atom reactions (C–C bond forming/breaking, oxidative addition to Pd), κ ≈ 1 is a reasonable approximation. For proton transfer steps, using κ = 1 can underestimate rates by a factor of 2–10 at room temperature.
Arrhenius Parameters from DFT: Activation Enthalpy and Entropy
The Arrhenius form of the rate expression is: k = A × exp(−Ea/RT), where Ea is the activation energy and A is the pre-exponential factor. Comparing to the Eyring equation:
Ea ≈ ΔH‡ + RT (for the activation enthalpy)
A = κ × (k_B T / h) × exp(ΔS‡ / R) × eThe activation entropy ΔS‡ directly controls the pre-exponential factor. Large negative ΔS‡ (highly ordered TS) gives small A; less negative ΔS‡ (less organized TS) gives larger A.
Typical ranges for organic and organometallic reactions:
- Unimolecular reactions (intramolecular cyclization, conformational change): ΔS‡ = −10 to +10 cal/mol·K
- Bimolecular reactions without tight TS (associative ligand exchange): ΔS‡ = −20 to −30 cal/mol·K
- Bimolecular reactions with very ordered TS (Diels-Alder): ΔS‡ = −35 to −45 cal/mol·K
A discrepancy of 10 cal/mol·K in ΔS‡ changes log A by ~2.2 units, which is a factor of ~150 in the rate constant. DFT-computed activation entropies carry systematic errors from the harmonic oscillator approximation, particularly for low-frequency torsional modes near the TS. If your computed k at 298 K differs from experiment by more than 100×, the first diagnostic is the activation entropy, not the activation enthalpy.
Low-Frequency Mode Problem in Entropy Calculation
The harmonic oscillator model assigns entropy contributions from each vibrational mode as S_vib = R × (u/(exp(u) − 1) − ln(1 − exp(−u))), where u = hν/k_BT. For low-frequency modes (ν < 100 cm⁻¹), this expression becomes sensitive to the exact frequency — a 10 cm⁻¹ error in a 50 cm⁻¹ mode can shift S_vib by ~1–2 cal/mol·K. In large flexible systems (organometallic catalysts with rotating phosphine ligands, for example), there can be 10–20 such modes near the TS, making the computed ΔS‡ uncertain by ±10–20 cal/mol·K from this source alone.
Practical mitigation strategies:
- Quasi-RRHO (rigid rotor-harmonic oscillator) correction: Replace the harmonic oscillator vibrational entropy for modes below a threshold frequency (typically 100 cm⁻¹) with a free rotor model interpolated with a damping function (Head-Gordon and coworkers; the qRRHO model in Grimme’s thermo script). This significantly reduces the sensitivity to low-frequency mode errors.
- Report ΔH‡ separately: For comparing to experiment, if activation enthalpy data (from Eyring plots) are available, compare ΔH‡ computed and experimental directly. ΔH‡ is much less sensitive to low-frequency modes than ΔS‡.
Converting DFT Barriers to Rate Constants: A Worked Example
For oxidative addition of PhBr to Pd(0)(SPhos) at 298 K, using ΔG‡ = 18.3 kcal/mol, κ = 1 (no tunneling correction for heavy-atom OA):
k_TST = (1 × 1.38 × 10⁻²³ × 298 / 6.63 × 10⁻³⁴) × exp(−18,300 / (1.987 × 298))
= 6.25 × 10¹² s⁻¹ × exp(−30.9)
= 6.25 × 10¹² × 2.8 × 10⁻¹⁴ s⁻¹
= 0.175 s⁻¹ (first order in Pd(0)(SPhos), pseudo-first order in excess ArBr)This gives t_½ ≈ 4 seconds — consistent with experimental observations that Pd(0)(SPhos)-catalyzed reactions are fast at room temperature for aryl bromides. For aryl chlorides, the computed ΔG‡ is typically 4–6 kcal/mol higher, giving t_½ ~2,000–100,000 seconds (explaining the need for elevated temperatures).
These numerical estimates are approximate — the factor of 10 per 1.4 kcal/mol sensitivity means a ±1.5 kcal/mol DFT error translates to roughly a 10× rate uncertainty. Reported computed rate constants should carry this uncertainty explicitly, particularly when the goal is quantitative matching to experimental kinetics rather than qualitative mechanistic assignment.