Symmetry Number and Rotational Partition Function

Spin a carbon dioxide molecule end-over-end by 180°. It looks exactly the same. This innocent-seeming fact has real thermodynamic consequences. The rotational partition function counts accessible rotational states, but a naive count double-books configurations that a symmetric molecule cannot actually distinguish. Without a correction, calculated entropies and equilibrium constants for molecules like O₂, CO₂, and CH₄ come out systematically wrong. This article introduces the symmetry number $\sigma$ , shows where it comes from, and demonstrates how to fold it into the rotational partition function. You will see why it arises from indistinguishable nuclear arrangements. Then you will compute it for linear, symmetric-top, and spherical-top molecules. Finally, this sets up the equilibrium-constant calculations that follow.

Key takeaways

The symmetry number $\sigma$ counts the number of indistinguishable orientations produced by rotating a rigid molecule.
It corrects the rotational partition function so identical nuclear arrangements are not overcounted as distinct states.
Linear molecules have $\sigma = 1$ (heteronuclear, like CO) or $\sigma = 2$ (homonuclear, like N₂).
Nonlinear molecules take $\sigma$ equal to the order of their rotational symmetry point group, e.g. $\sigma = 12$ for CH₄.
Symmetry numbers shift rotational entropy by $-R\ln\sigma$ and matter directly in equilibrium constants.
Key idea: dividing the rotational partition function by $\sigma$ removes rotational microstates that are physically identical due to molecular symmetry, keeping the state count honest.

The core insight

Because rotating a symmetric molecule by certain angles produces a configuration indistinguishable from the original, the classical rotational partition function—derived by integrating freely over all orientations—overcounts genuinely distinct quantum states by a factor equal to the number of such equivalent orientations, and dividing by that factor, the symmetry number $\sigma$ , restores the correct count.

The physical picture

The rotational partition function for a rigid rotor, developed from the prerequisite treatment of rotational energy levels, is built by summing Boltzmann factors over all rotational quantum states $J$ :

q_{\text{rot}} = \sum_J (2J+1)\, e^{-\beta hcBJ(J+1)}

At ordinary temperatures the rotational levels are closely spaced compared to $k_BT$ for most molecules (except $H_2$ and its isotopologues), so the sum is well approximated by an integral. That integral treats the molecule's orientation as a classical continuum of angles, exactly as if every rotation angle produced a physically new orientation of a labeled object. But real molecules are not labeled arrows in space. Consider $\text{N}_2$ : rotate it by 180° about an axis perpendicular to the bond, and the two identical nitrogen nuclei simply swap places. Because the nuclei are identical particles, the resulting configuration is not merely similar—it is the same physical state. The classical integral, however, still counts it as a distinct orientation. For every true state, it counts $\sigma$ orientations that a quantum-mechanical treatment would recognize as one and the same. The remedy is simple: divide the classical result by $\sigma$ , the number of indistinguishable orientations generated by proper rotations of the rigid framework. This is a purely rotational bookkeeping correction. It has nothing to do with nuclear spin statistics per se (though the two are historically intertwined for homonuclear diatomics, where nuclear spin degeneracy and rotational state parity become coupled). For the purposes of computing thermodynamic functions from partition functions, $\sigma$ is simply the order of the rotational subgroup of proper rotations that map the molecule onto an indistinguishable copy of itself.

Figure

Symmetry operations that generate indistinguishable orientations for N₂ (σ = 2) and NH₃ (σ = 3)

The same logic scales up for polyatomic molecules. A tetrahedral molecule like $\text{CH}_4$ has many more proper rotation axes—four $C_3$ axes through each C–H bond, three $C_2$ axes, and so on—giving a total rotational symmetry order of 12. Benzene, with its six-fold axis and six perpendicular $C_2$ axes, has $\sigma = 12$ as well. The symmetry number is always the order of the point group's rotational subgroup (excluding reflections and improper rotations, which are not physical motions of a rotating rigid body).

Figure

Rotational symmetry number for common molecular shapes

The core relations

q_{\text{rot}} = \frac{1}{\sigma}\left(\frac{k_BT}{hcB}\right) \quad \text{(linear rotor, high-}T\text{ limit)}

where: $\sigma = 1$ for heteronuclear diatomics, $\sigma = 2$ for homonuclear diatomics, $B$ is the rotational constant.

q_{\text{rot}} = \frac{1}{\sigma}\left(\frac{k_BT}{hc}\right)^{3/2}\left(\frac{\pi}{ABC}\right)^{1/2}

where: $A, B, C$ are the three rotational constants of a nonlinear rigid rotor, $\sigma$ is its symmetry number.

S_{\text{rot}} = R\ln q_{\text{rot}} + RT\left(\frac{\partial \ln q_{\text{rot}}}{\partial T}\right)_{V} = R\left[\ln\!\left(\frac{q_{\text{rot}}^{\text{classical}}}{\sigma}\right) + \text{(temperature terms)}\right]

where: introducing $\sigma$ shifts the rotational entropy by exactly $-R\ln\sigma$ relative to the uncorrected calculation.

\sigma = \left|\,\text{order of the rotational subgroup of the molecular point group}\,\right|

where: only proper rotations (identity and $C_n$ operations) count; reflections and improper rotations are excluded.

Derivation AFrom classical phase-space counting to the σ-corrected rotational partition function

Goal. Show explicitly why dividing by $\sigma$ is the correct fix, using the classical rigid-rotor partition function as the starting point, then specialize to a linear rotor.

Step 1: Classical rotational partition function. For a rigid rotor, the classical partition function is an integral over all orientation angles and their conjugate momenta:

q_{\text{rot}}^{\text{classical}} = \frac{1}{h^{f}}\int e^{-\beta H(\theta,\phi,\chi,p_\theta,p_\phi,p_\chi)}\, d\theta\, d\phi\, d\chi\, dp_\theta\, dp_\phi\, dp_\chi

where $f$ is the number of rotational degrees of freedom (2 for linear rotors, 3 for nonlinear ones), and the integration over Euler angles runs over the full range that would distinguish every geometric orientation of a labeled, asymmetric object.

Step 2: Recognize the overcounting. This integral silently assumes that every distinct orientation angle corresponds to a distinct physical state. For a symmetric molecule, however, $\sigma$ different combinations of Euler angles correspond to configurations that are related by a proper rotation mapping the nuclear framework onto an indistinguishable copy of itself—because identical nuclei have been permuted among symmetry-equivalent positions. Quantum mechanically, these are the same state, not $\sigma$ different states, because the total wavefunction must already respect the correct permutation symmetry of identical nuclei. The classical integral, blind to nuclear identity, counts them $\sigma$ times.

Step 3: Insert the correction. The fix is to divide the full classical integral by $\sigma$ :

q_{\text{rot}} = \frac{1}{\sigma}\, q_{\text{rot}}^{\text{classical, uncorrected}}

This is valid in the high-temperature (classical) limit, which holds for essentially all molecules except $H_2$ , $D_2$ , and $HD$ near room temperature, where the rotational spacing is too coarse relative to $k_BT$ and a more careful quantum sum over allowed $J$ states (respecting nuclear spin statistics) is required instead.

Step 4: Specialize to a linear rotor. For a linear rotor, evaluating the classical integral over the two rotational angles and momenta gives:

q_{\text{rot}}^{\text{classical, uncorrected}} = \frac{8\pi^2 I k_BT}{h^2} = \frac{k_BT}{hcB}

where $I$ is the moment of inertia and $B = h/(8\pi^2 cI)$ is the rotational constant in wavenumbers. Applying the Step 3 correction:

\boxed{q_{\text{rot}} = \frac{1}{\sigma}\cdot\frac{k_BT}{hcB}}

Step 5: Sanity check against the quantum sum. For a heteronuclear diatomic like CO, every $J$ level really is accessible and distinguishable, consistent with $\sigma=1$ reproducing the unmodified high- $T$ quantum sum $\sum_J(2J+1)e^{-\beta hcBJ(J+1)} \approx k_BT/hcB$ . For $N_2$ , direct enumeration of the true quantum states (respecting the required overall symmetry of the nuclear wavefunction under nucleus exchange) shows that only every other rotational level combines with the correct nuclear spin degeneracy, and the net effect exactly reproduces $q_{\text{rot}} = (k_BT/hcB)/2$ in the high- $T$ limit — confirming $\sigma=2$ is not an ad hoc patch but the correct classical shadow of a real quantum constraint.

Worked examples

Example 1 — Rotational partition function and entropy of $N_2$ at 298 K. Given: bond length $r = 109.76\ \text{pm}$ , reduced mass of $^{14}N_2$ , symmetry number $\sigma = 2$ . Step 1. Reduced mass:

\mu = \frac{m_N}{2} = \frac{14.003\ \text{u}}{2} = 7.0015\ \text{u} = 1.1625\times10^{-26}\ \text{kg}

(using $1\ \text{u} = 1.6605\times10^{-27}\ \text{kg}$ ) Step 2. Moment of inertia:

I = \mu r^2 = (1.1625\times10^{-26}\ \text{kg})(1.0976\times10^{-10}\ \text{m})^2 = 1.4004\times10^{-46}\ \text{kg m}^2

Step 3. Rotational constant:

B = \frac{h}{8\pi^2 c I} = \frac{6.626\times10^{-34}}{8\pi^2(2.998\times10^{10}\ \text{cm s}^{-1})(1.4004\times10^{-46}\times10^{4}\ \text{cm}^2)}

Carrying units through carefully (converting $I$ to cm² via the $10^4$ factor) gives:

B = 1.998\ \text{cm}^{-1}

which matches the accepted spectroscopic value for $N_2$ closely (literature: $1.9987\ \text{cm}^{-1}$ ). Step 4. Rotational partition function at $T = 298.15\ \text{K}$ . Using $k_BT/hc = 207.22\ \text{cm}^{-1}$ at this temperature:

q_{\text{rot}}^{\text{uncorrected}} = \frac{k_BT}{hcB} = \frac{207.22}{1.998} = 103.7

q_{\text{rot}} = \frac{103.7}{\sigma} = \frac{103.7}{2} = 51.8

Step 5. Rotational entropy contribution. Using

S_{\text{rot}} = R\left[\ln q_{\text{rot}} + 1\right]

(the "+1" arising from the temperature derivative of $q_{\text{rot}}\propto T$ for a linear rotor):

S_{\text{rot}} = (8.314\ \text{J K}^{-1}\text{mol}^{-1})\left[\ln(51.8) + 1\right] = 8.314 \times (3.947 + 1) = 41.1\ \text{J K}^{-1}\text{mol}^{-1}

Sanity check: this compares well with the rotational contribution extracted from the experimental $S^\circ_{298}(N_2) = 191.6\ \text{J K}^{-1}\text{mol}^{-1}$ once translational ( $\approx 150.4\ \text{J K}^{-1}\text{mol}^{-1}$ ) and vibrational (negligible at 298 K) contributions are subtracted, leaving a rotational share close to the calculated 41 J K⁻¹ mol⁻¹. Omitting $\sigma$ would have overstated $S_{\text{rot}}$ by $R\ln 2 = 5.76\ \text{J K}^{-1}\text{mol}^{-1}$ — a small but experimentally significant and easily mismeasured error.

Example 2 — Rotational partition function of $\text{NH}_3$ and the effect of getting σ wrong. $\text{NH}_3$ is a symmetric top with $\sigma = 3$ (the $C_3$ axis through nitrogen, threefold: $0°, 120°, 240°$ ). Rotational constants (spectroscopic, converted to consistent units): $B = 9.44\ \text{cm}^{-1}$ (perpendicular to the $C_3$ axis, doubly degenerate) and $A = 6.20\ \text{cm}^{-1}$ (about the $C_3$ axis). For a symmetric top, the high-temperature rotational partition function is:

q_{\text{rot}} = \frac{\sqrt{\pi}}{\sigma}\left(\frac{k_BT}{hc}\right)^{3/2}\frac{1}{\sqrt{A B^2}}

Step 1. At $T = 298.15\ \text{K}$ , $k_BT/hc = 207.22\ \text{cm}^{-1}$ , so:

\left(\frac{k_BT}{hc}\right)^{3/2} = (207.22)^{3/2} = 2981\ \text{cm}^{-3/2}

Step 2. Denominator:

\sqrt{AB^2} = \sqrt{(6.20)(9.44)^2} = \sqrt{6.20 \times 89.11} = \sqrt{552.5} = 23.51\ \text{cm}^{3/2}

Step 3. Uncorrected partition function (setting $\sigma=1$ for the moment):

q_{\text{rot}}^{\text{uncorrected}} = \sqrt{\pi}\times\frac{2981}{23.51} = 1.7725 \times 126.8 = 224.7

Step 4. Apply $\sigma = 3$ :

q_{\text{rot}} = \frac{224.7}{3} = 74.9

Step 5. The entropy consequence of choosing the wrong $\sigma$ . If a student mistakenly assumed NH₃ had no rotational symmetry ( $\sigma = 1$ , forgetting the three-fold axis), the rotational entropy would be overestimated by:

\Delta S = R\ln\sigma = (8.314\ \text{J K}^{-1}\text{mol}^{-1})\ln 3 = 9.13\ \text{J K}^{-1}\text{mol}^{-1}

This is a substantial 9 J K⁻¹ mol⁻¹ error—large enough to noticeably distort a calculated equilibrium constant through $\Delta S^\circ$ in $\Delta G^\circ = \Delta H^\circ - T\Delta S^\circ$ , underscoring why identifying the correct point group and its rotational subgroup order is not optional bookkeeping but essential chemistry.

Limiting cases & common traps

Situation	Correct treatment	Common error
Heteronuclear diatomic (CO, HCl)	$\sigma = 1$	Assuming symmetry correction is always needed
Homonuclear diatomic (N₂, O₂, Cl₂)	$\sigma = 2$	Forgetting the factor entirely
Linear symmetric triatomic (CO₂, C₂H₂)	$\sigma = 2$ (like homonuclear diatomic, due to the C₂ axis perpendicular to the molecular axis)	Treating it as $\sigma=1$ because it is not "two atoms"
Bent, two identical atoms (H₂O, SO₂)	$\sigma = 2$	Missing the C₂ axis and using $\sigma=1$
Pyramidal $C_{3v}$ (NH₃)	$\sigma = 3$	Confusing point-group order (6, including reflections) with rotational symmetry number (3)
Tetrahedral (CH₄)	$\sigma = 12$	Using only the four $C_3$ axes (undercounting to 4 instead of the full rotational subgroup order 12)
Very light molecules (H₂, HD, D₂) at low T	Full quantum sum with correct nuclear spin statistics, not the classical $\sigma$ -correction	Applying the high- $T$ formula blindly below ~100 K

The chemist behind it

The rigorous justification for symmetry numbers grew out of early twentieth-century efforts to reconcile spectroscopic rotational term values with the emerging quantum statistics of identical nuclei — work that connects directly to the resolution of the "ortho/para hydrogen" puzzle. Dennis Gabor, Werner Heisenberg, and others recognized in the late 1920s that hydrogen's anomalous specific heat could only be explained by treating the two protons as indistinguishable fermions, forcing rotational states to pair with specific nuclear spin configurations. This work established the deep link between molecular symmetry, nuclear spin statistics, and rotational partition functions that the symmetry number encapsulates in its simplified high-temperature form.

Origins · 1 minRichard C. Tolman and how this was discovered

Richard C. Tolman

1881–1948 · American physical chemist and statistical mechanician who developed rigorous classical-to-quantum correspondence arguments, including counting arguments for indistinguishable configurations that underlie the σ-correction used throughout statistical thermodynamics

Ask the tutor: How do symmetry numbers combine with nuclear spin degeneracy to explain why ortho- and para-hydrogen behave like two different gases at low temperature?

Why doesn't the symmetry number appear in the translational or vibrational partition functions

Symmetry numbers arise specifically from indistinguishable orientations generated by rotating the rigid nuclear framework. Translation moves the molecule's center of mass through space; no rotation is involved, and distinguishable positions in space remain distinguishable regardless of molecular symmetry. Vibrations displace nuclei from equilibrium along normal modes but do not permute identical nuclei among symmetry-equivalent sites in the way a proper rotation does. The correction is therefore purely a rotational bookkeeping issue, tied to the point group's rotational subgroup, not a general partition-function feature.

Does σ depend on temperature or on which isotope is used

No, σ is a purely geometric property of the equilibrium nuclear framework and does not change with temperature; the high-temperature classical approximation it corrects, however, is temperature-dependent and can break down at low T. Isotopic substitution can change σ if it breaks a symmetry: for example, HD (deuterium-substituted hydrogen) has σ = 1 because the two nuclei are no longer identical, whereas H₂ and D₂ each have σ = 2.

How is σ related to the order of the molecular point group

The symmetry number equals the order of the rotational subgroup of the full point group—that is, the count of proper rotations (identity plus all $C_n$ operations) that map the molecule onto an indistinguishable copy of itself. Reflections, inversion centers, and improper rotations ( $S_n$ ) are excluded because they are not physical motions a tumbling rigid rotor can execute. For $C_{3v}$ (NH₃), the full point group has order 6, but the rotational subgroup has order 3, so $\sigma=3$ , not 6.

Does the symmetry number correction matter for calculating equilibrium constants

Yes, substantially. Equilibrium constants computed from partition functions depend on ratios of $q_{\text{rot}}$ for reactants and products, and getting σ wrong for even one species shifts $\ln K$ by $\ln\sigma$ for that species. Because $\Delta G^\circ = -RT\ln K$ , an error of $R\ln 2$ in entropy near room temperature corresponds to roughly 1.7 kJ/mol in $\Delta G^\circ$ , which can change a calculated equilibrium constant by a factor of about 2. This becomes central in the next stage of applying partition functions to real reactions.