Electronic Partition Function

Chlorine atoms are yellow-green as a gas, but that is not why their thermodynamics gets interesting. Deep in the electronic structure of atomic chlorine sits a spin-orbit excited state only 881 cm⁻¹ above the ground term. At room temperature this state is partly populated, and it measurably changes the atom's entropy and heat capacity. The electronic partition function, $q_{elec}$ , is the bookkeeping tool that captures this effect. Most textbook problems quietly set $q_{elec} = 1$ , and for closed-shell molecules like H₂O or CO₂ at 298 K that is an excellent approximation. But for open-shell species, radicals, and many transition-metal atoms, low-lying excited electronic states sit close enough in energy to matter. Getting $q_{elec}$ right is often the difference between a calculated entropy that matches experiment and one that is off by several J/mol·K. This article builds $q_{elec}$ from first principles: ground-state degeneracy, Boltzmann population of excited terms, and the practical truncation of an infinite sum. You will see the general formula. Then you will see when it collapses to a single number. Finally you will work through NO and atomic chlorine as concrete, checkable examples.

Key takeaways

$q_{elec}$ sums Boltzmann factors over electronic energy levels, each weighted by its degeneracy $g_i$ .
Electronic energy gaps are typically thousands of cm⁻¹, so $q_{elec} \approx g_0$ (ground-state degeneracy) for most closed-shell molecules at 298 K.
Open-shell atoms and radicals (O, Cl, NO) often have low-lying spin-orbit states within a few hundred to a thousand cm⁻¹, giving $q_{elec} > g_0$ .
The reference energy is always set at the ground state, so the ground term contributes exactly $g_0$ with no exponential suppression.
$q_{elec}$ feeds directly into electronic contributions to entropy, heat capacity, and free energy, and ultimately into equilibrium constants computed from partition functions.
Key idea: $q_{elec}$ is a finite, temperature-dependent sum over electronic states measured from the ground state, and it is essentially never an infinite or divergent series in practice because upper electronic states lie so high that they never contribute.

The core insight

The electronic partition function is the degeneracy-weighted Boltzmann sum over electronic energy levels measured from the ground state; because electronic spacings are usually huge compared to $k_BT$ it reduces to the ground-state degeneracy, except in the important minority of cases (open-shell atoms, some radicals) where a low-lying excited term contributes a calculable fractional population.

The physical picture

Recall that the total molecular partition function factorizes into translational, rotational, vibrational, and electronic pieces,

q = q_{trans}\, q_{rot}\, q_{vib}\, q_{elec},

because each degree of freedom has its own, independent set of energy levels (see Factorization of the Molecular Partition Function). The electronic partition function handles the population of electronic states — different arrangements of electrons around the nuclei, each with its own energy and degeneracy. Electronic energy differences are large. A typical electronic excitation energy is 20,000–50,000 cm⁻¹, corresponding to visible or ultraviolet light. Compare this to $k_BT$ at 298 K, which is about 207 cm⁻¹. The ratio $\Delta\varepsilon / k_BT$ is on the order of 100–200, so $\exp(-\Delta\varepsilon/k_BT)$ is astronomically small — the excited electronic state is essentially unpopulated. This is why chemists almost always write $q_{elec} \approx g_0$ , the degeneracy of the ground electronic state, and move on. The exception is open-shell species with low-lying states arising from spin-orbit coupling or near-degenerate orbital occupancies. Atomic oxygen's ³P ground term splits into $J = 2, 1, 0$ sublevels spanning about 226 cm⁻¹. Atomic chlorine's ground term ²P splits into $J = 3/2$ (ground) and $J = 1/2$ (excited, 881 cm⁻¹ above). Nitric oxide, NO, has a low-lying spin-orbit doublet split by 121 cm⁻¹. In all these cases $k_BT$ is comparable to or larger than the gap, so the excited state is thermally accessible and must be included explicitly.

Figure

Energy-level ladder and Boltzmann population for a two-level electronic system

Figure

q_elec versus temperature for the chlorine atom two-level model

The core relations

q_{elec} = \sum_i g_i \, e^{-\varepsilon_i / k_B T}

where: $g_i$ is the degeneracy of electronic level $i$ , $\varepsilon_i$ its energy measured from the ground state ( $\varepsilon_0 \equiv 0$ ), $k_B$ Boltzmann's constant, $T$ absolute temperature.

q_{elec} = g_0 + g_1 e^{-\varepsilon_1/k_BT} + g_2 e^{-\varepsilon_2/k_BT} + \cdots

where: the ground term always contributes exactly $g_0$ because $\varepsilon_0 = 0$ by convention.

q_{elec} \xrightarrow{\;\varepsilon_1 \gg k_BT\;} g_0

where: this is the ordinary closed-shell limit used almost universally for stable molecules at room temperature.

P_i = \frac{g_i\, e^{-\varepsilon_i/k_BT}}{q_{elec}}

where: $P_i$ is the fraction of molecules or atoms found in electronic level $i$ at temperature $T$ .

U_{elec} = N k_BT^2 \left(\frac{\partial \ln q_{elec}}{\partial T}\right)_V, \qquad S_{elec} = Nk_B \ln q_{elec} + \frac{U_{elec}}{T}

where: these give the electronic contributions to internal energy and entropy once $q_{elec}(T)$ is known.

Derivation AFrom energy levels to the electronic partition function

Goal. Show how the general partition function definition reduces to a finite sum over electronic states, and why truncation after one or two excited terms is normally exact for chemical purposes.

Step 1 — start from the definition. For any set of energy levels accessible to a system, the partition function is

q = \sum_{\text{all states}} e^{-\varepsilon_{state}/k_BT}.

Electronic states often come in degenerate groups — for example, the $M_J = -3/2, -1/2, +1/2, +3/2$ sublevels of Cl's ²P₃/₂ term are degenerate at zero field. Grouping states of equal energy into levels of degeneracy $g_i$ turns the sum over states into a sum over levels,

q_{elec} = \sum_i g_i\, e^{-\varepsilon_i/k_BT}.

Step 2 — fix the energy zero at the ground state. Partition functions are only meaningful up to a choice of energy origin, and the conventional choice for $q_{elec}$ sets $\varepsilon_0 = 0$ for the ground electronic level. This is different from, say, vibrational partition functions referenced to the bottom of the well versus the zero-point level — the choice must always be stated, and here it means the ground level's Boltzmann factor is exactly 1. Consequently

q_{elec} = g_0 + \sum_{i \geq 1} g_i\, e^{-\varepsilon_i/k_BT}.

Step 3 — estimate the size of the excited-state terms. For a typical stable closed-shell molecule, the first electronic excitation energy $\varepsilon_1$ is of order 40,000 cm⁻¹ (a UV transition near 250 nm). At 298 K, $k_BT \approx 207$ cm⁻¹, so

\frac{\varepsilon_1}{k_BT} \approx \frac{40000}{207} \approx 193, \qquad e^{-193} \approx 10^{-84}.

This term is utterly negligible — no realistic measurement could detect its effect on $q_{elec}$ . Hence for the vast majority of stable molecules,

q_{elec} \approx g_0.

Step 4 — identify when the approximation fails. The approximation breaks down when $\varepsilon_1$ is only a few hundred to a couple of thousand cm⁻¹ — the regime of spin-orbit splittings in open-shell atoms (O, Cl, F, Br, I) and some diatomic radicals (NO, OH in certain states). There $\varepsilon_1/k_BT$ is of order 1–5, and $e^{-\varepsilon_1/k_BT}$ is not negligible. The sum must then be kept to at least two terms:

q_{elec} \approx g_0 + g_1\, e^{-\varepsilon_1/k_BT}.

Step 5 — check convergence and truncation validity. Because degeneracies $g_i$ do not grow without bound (they are set by angular momentum coupling and are typically 1–6), and because energies $\varepsilon_i$ generally increase monotonically and steeply for successive electronic terms, the series converges rapidly. In practice, once the second excited state lies more than about 5,000 cm⁻¹ above the ground state, its Boltzmann factor is below $10^{-10}$ at room temperature and can be dropped. This justifies truncating the infinite sum to one or two terms — never an approximation of convenience alone, but one that can be checked numerically term by term.

\boxed{q_{elec} = g_0 + g_1 e^{-\varepsilon_1/k_BT} + \cdots \;\approx\; g_0 \text{ (closed-shell)} \quad\text{or}\quad g_0 + g_1 e^{-\varepsilon_1/k_BT} \text{ (low-lying excited term)}}

Worked examples

Example 1 — Ground-state degeneracy only: atomic sodium at 298 K. Atomic sodium in its ground state has electron configuration [Ne]3s¹, giving a ²S₁/₂ term. The degeneracy of a term with total angular momentum $J$ is $g = 2J+1$ , so here $g_0 = 2(1/2)+1 = 2$ . The first excited state (the famous sodium D-line transition, 3s → 3p) lies at about 16,978 cm⁻¹ (the D₂ line, 589.0 nm) above the ground state. Compute the Boltzmann factor at 298 K. First, $k_BT$ in cm⁻¹:

k_BT = (0.695\ \text{cm}^{-1}\text{K}^{-1})(298\ \text{K}) = 207.1\ \text{cm}^{-1}.

Then

\frac{\varepsilon_1}{k_BT} = \frac{16978}{207.1} = 82.0, \qquad e^{-82.0} \approx 2 \times 10^{-36}.

The excited-state contribution is

g_1 e^{-\varepsilon_1/k_BT} = 2 \times (2\times10^{-36}) \approx 4\times10^{-36}.

q_{elec} = 2 + 4\times10^{-36} = 2.000\ldots

to any digit precision that matters chemically. Sanity check: the answer is simply $g_0 = 2$ , exactly as the qualitative argument predicts for a species whose first excitation is a genuine electronic (not spin-orbit) transition tens of thousands of cm⁻¹ up. This is the "boring but correct" default outcome for the overwhelming majority of molecules and atoms.

Example 2 — Two-term sum: ground-state atomic chlorine at 298 K and 1000 K. Atomic chlorine's ground configuration is [Ne]3s²3p⁵, giving a ²P term split by spin-orbit coupling into $J = 3/2$ (lower, ground) and $J = 1/2$ (upper). Degeneracies:

g_0 = 2(3/2)+1 = 4, \qquad g_1 = 2(1/2)+1 = 2.

The spin-orbit splitting is $\varepsilon_1 = 881\ \text{cm}^{-1}$ above the ground level. There are no other electronic states low enough to matter (the next excited terms of Cl lie tens of thousands of cm⁻¹ higher), so a two-term sum is appropriate. At $T = 298$ K: $k_BT = 207.1\ \text{cm}^{-1}$ as before.

\frac{\varepsilon_1}{k_BT} = \frac{881}{207.1} = 4.254, \qquad e^{-4.254} = 0.01418.

q_{elec} = 4 + 2(0.01418) = 4 + 0.0284 = 4.028.

The fraction of Cl atoms in the excited $J=1/2$ state is

P_1 = \frac{2(0.01418)}{4.028} = \frac{0.0284}{4.028} = 0.00704,

about 0.7% — small, but not zero, and enough to shift the calculated entropy measurably from the naive $R\ln 4$ value. At $T = 1000$ K (relevant to combustion or high-temperature chlorine chemistry):

k_BT = (0.695)(1000) = 695\ \text{cm}^{-1}, \qquad \frac{881}{695} = 1.268, \qquad e^{-1.268} = 0.2815.

q_{elec} = 4 + 2(0.2815) = 4 + 0.563 = 4.563.

Now $P_1 = 0.563/4.563 = 0.123$ , roughly 12% of atoms in the excited spin-orbit state — a genuinely significant population that must be included in any serious thermodynamic calculation of Cl atom entropy or heat capacity at combustion temperatures.

Sanity check: $q_{elec}$ increases monotonically with $T$ from a low-temperature limit of exactly $g_0 = 4$ (all population in the ground level as $T \to 0$ ) toward a high-temperature limit of $g_0+g_1 = 6$ (both levels equally accessible relative to their degeneracies as $\varepsilon_1/k_BT \to 0$ ). Both computed values, 4.028 and 4.563, sit correctly between these bounds and increase with $T$ , as required by the physics.

Limiting cases & common traps

Situation	Behavior of $q_{elec}$	Practical rule
Closed-shell molecule (H₂O, CO₂, CH₄)	$\varepsilon_1 \sim 10^4$ – $10^5$ cm⁻¹	$q_{elec} = g_0 = 1$ (singlet, nondegenerate)
Open-shell atom with spin-orbit splitting (O, Cl, F)	$\varepsilon_1 \sim 10^2$ – $10^3$ cm⁻¹	Keep two (or more) terms explicitly
$T \to 0$	All population in ground level	$q_{elec} \to g_0$
$T \to \infty$ (or $\varepsilon_1 \ll k_BT$ )	Levels populated by degeneracy ratio	$q_{elec} \to \sum_i g_i$
Radical with near-degenerate ground state (NO₂, some transition-metal atoms)	Several low states possible	Sum may need three or more terms

The chemist behind it

The modern treatment of $q_{elec}$ sits at the intersection of atomic spectroscopy and statistical mechanics, a synthesis most associated with Giuseppe Occhialini's contemporaries in early 20th-century spectroscopy — but the person most directly responsible for connecting electronic term structure to measurable thermodynamic quantities was Robert Mulliken, whose systematic work on electronic states of diatomic molecules gave spectroscopists the term symbols and degeneracies that feed directly into $q_{elec}$ sums used today.

Origins · 1 minRobert S. Mulliken and how this was discovered

Robert S. Mulliken

1896-1986 · American physical chemist who developed molecular orbital theory and systematized electronic term symbols used to assign degeneracies and energies in q_elec

Ask the tutor: How does a nonzero q_elec change the calculated standard entropy of atomic chlorine gas compared to using q_elec = 1?

Why doesn't q_elec ever diverge, given that atoms and molecules have infinitely many electronic states?

In principle an atom has infinitely many bound electronic states approaching an ionization limit, plus a continuum above it. In practice, energies $\varepsilon_i$ increase steeply enough, and $k_BT$ is small enough at chemically relevant temperatures, that terms beyond the first one or two excited states contribute a completely negligible Boltzmann factor. The infinite series converges rapidly in the regime chemists care about, so truncating after one or two terms introduces no meaningful error. Divergence becomes a real concern only near ionization or in exotic plasma conditions, well outside standard thermodynamic calculations.

How is q_elec different from q_vib or q_rot in terms of its energy reference?

Vibrational partition functions are commonly referenced either to the bottom of the potential well or to the zero-point vibrational level, and both conventions appear in textbooks with different resulting formulas. Rotational partition functions naturally start from $J=0$ with zero energy. The electronic partition function follows the same logic as rotation: the ground electronic level is defined as the energy zero, so $\varepsilon_0 = 0$ and its Boltzmann factor is always exactly 1, contributing $g_0$ to the sum with no ambiguity.

Does q_elec affect equilibrium constants?

Yes, directly. Equilibrium constants computed from partition functions involve ratios of $q$ values for reactants and products, each including its $q_{elec}$ factor. For reactions involving open-shell atoms or radicals — combustion intermediates like O, Cl, OH, or NO — the electronic partition function contributes a real, calculable correction to $K$ beyond the dominant translational and rotational terms. This becomes especially important at combustion or atmospheric temperatures where excited spin-orbit states are appreciably populated.

Can q_elec ever be less than the ground-state degeneracy g0?

No. Every term in the sum $q_{elec} = \sum_i g_i \exp(-\varepsilon_i/k_BT)$ is positive, and the ground-state term always contributes exactly $g_0$ with $\varepsilon_0=0$ . Adding any further positive terms can only increase the total. So $q_{elec} \geq g_0$ always, with equality only in the strict limit where excited states are completely inaccessible, either because $T \to 0$ or because $\varepsilon_1$ is enormous compared to $k_BT$ .