Contributions to Heat Capacity from Molecular Modes

Heat a gas of diatomic molecules and watch its heat capacity climb in steps rather than smoothly. At low temperature only translation and rotation absorb energy. At high temperature vibration wakes up too. Then $C_V$ jumps toward a new plateau. This staircase behavior is invisible to classical mechanics alone. The puzzle mattered historically. Nineteenth-century kinetic theory predicted a single universal heat capacity for all diatomic gases. This was contradicted by experiment. Resolving that contradiction required quantizing molecular motion. It also required comparing thermal energy $k_BT$ against the spacing of rotational and vibrational energy levels. This article builds the equipartition theorem from classical statistical mechanics. It shows precisely how and why it fails. It also derives the temperature-dependent heat capacity of an ideal diatomic gas mode by mode. You will see the classical limit first. Then you will see quantum corrections. Finally you will combine both into a working formula you can evaluate for any real gas.

Key takeaways

Equipartition assigns $\tfrac12 k_BT$ to each quadratic term in the classical energy, giving $C_V = \tfrac32 R$ for translation and $\tfrac12 R$ per rotational or vibrational quadratic term.
Equipartition works only when $k_BT$ greatly exceeds the spacing between quantum energy levels of that mode.
Rotational modes "unfreeze" near room temperature because $\theta_{rot} \sim 1$ – $100$ K; vibrational modes unfreeze only above $\theta_{vib} \sim 1000$ – $6000$ K.
Each mode contributes a Schottky-like heat capacity curve that rises from zero and saturates at its classical equipartition value.
Electronic degrees of freedom are almost always frozen out at ordinary temperatures because electronic gaps vastly exceed $k_BT$ .
Key idea: heat capacity per mode grows from zero to its classical equipartition value as temperature crosses that mode's characteristic quantum temperature, $\theta = \epsilon/k_B$ .

The core insight

Each molecular degree of freedom contributes to $C_V$ only once thermal energy $k_BT$ becomes comparable to or larger than its quantum level spacing; equipartition is simply the high-temperature limit of this general statement, and the transition temperature $\theta_{mode}$ is fixed by the mode's characteristic energy scale, $B$ for rotation or $h\nu$ for vibration.

The physical picture

Consider a diatomic gas molecule. It has three ways to store energy: translation of its center of mass, rotation about axes perpendicular to the bond, and vibration along the bond. Classically, each of these looks like a bundle of quadratic terms in the Hamiltonian — momentum-squared terms for translation and rotation, and momentum-squared plus coordinate-squared terms for the harmonic vibration. Classical statistical mechanics treats energy as continuously divisible among all of these terms, giving the equipartition theorem: each quadratic term carries an average energy of $\tfrac12 k_BT$ , regardless of the particle's mass or the force constant involved. Quantum mechanics changes this picture only when the level spacing of a mode becomes comparable to $k_BT$ . Translational levels in a macroscopic container are so closely spaced that they are effectively continuous at any accessible temperature — equipartition for translation is essentially always valid. Rotational levels are spaced by an energy $B \sim 1$ – $60$ cm $^{-1}$ for common diatomics, corresponding to characteristic temperatures $\theta_{rot} = hcB/k_B$ of order 1–100 K. Room temperature easily exceeds this, so rotation is classical for most molecules well above their boiling points. Vibrational levels are spaced by $h\nu$ , typically 500–4000 cm $^{-1}$ , giving $\theta_{vib} = h\nu/k_B$ of order 1000–6000 K — usually far above room temperature. Vibration therefore stays "frozen" in its ground state at 298 K for most diatomics, contributing almost nothing to $C_V$ , and only gradually unfreezes as $T$ climbs toward $\theta_{vib}$ . This picture — a ladder of energy scales, each with its own onset temperature — is the master idea that explains the shape of $C_V(T)$ for any real molecular gas.

Figure

Heat capacity per mole rises in steps as translational, rotational, and vibrational modes unfreeze with increasing temperature

The core relations

\langle \epsilon \rangle = \tfrac12 k_BT \quad \text{per quadratic term (classical equipartition)}

where: $\epsilon$ is the average energy stored in one quadratic term of the classical Hamiltonian, valid when $k_BT \gg$ level spacing.

C_{V,m}^{trans} = \tfrac32 R, \qquad C_{V,m}^{rot,\,linear} = R, \qquad C_{V,m}^{vib} \to R \ \text{(per mode, classical limit)}

where: $R = N_Ak_B$ is the gas constant; a linear rotor has two rotational quadratic terms giving $R$ , and each vibrational mode has two quadratic terms (kinetic + potential) giving $R$ classically.

C_{V,m}^{vib}(T) = R\left(\frac{\theta_{vib}}{T}\right)^2 \frac{e^{\theta_{vib}/T}}{\left(e^{\theta_{vib}/T}-1\right)^2}, \qquad \theta_{vib} = \frac{h\nu}{k_B}

where: $\theta_{vib}$ is the vibrational characteristic temperature and $T$ is the absolute temperature.

C_{V,m}^{rot}(T) \approx R \left[1 + \frac{1}{45}\left(\frac{\theta_{rot}}{T}\right)^2 + \dots \right], \qquad \theta_{rot} = \frac{hcB}{k_B}

where: this high-temperature expansion (valid for $T \gg \theta_{rot}$ ) shows the rotational heat capacity approaching $R$ from slightly above.

C_{V,m}^{total}(T) = \tfrac32R + C_{V,m}^{rot}(T) + \sum_{modes} C_{V,m}^{vib}(T) + C_{V,m}^{elec}(T)

where: the sum runs over all vibrational normal modes; electronic terms are negligible unless low-lying excited states exist.

Derivation AFrom the vibrational partition function to the Einstein-type heat capacity

Goal. Derive $C_{V,m}^{vib}(T)$ for one harmonic vibrational mode directly from its quantum partition function, and show it reduces to $R$ at high $T$ and vanishes at low $T$ .

Step 1. Partition function. For a harmonic oscillator with energy levels $\epsilon_v = (v+\tfrac12)h\nu$ , $v = 0,1,2,\dots$ , measured from the bottom of the well, the molecular partition function is a geometric series:

q_{vib} = \sum_{v=0}^{\infty} e^{-\beta(v+1/2)h\nu} = \frac{e^{-\beta h\nu/2}}{1-e^{-\beta h\nu}}, \qquad \beta = \frac{1}{k_BT}

Step 2. Molar internal energy. Using $U_m - U_m(0) = N_A k_BT^2 \left(\partial \ln q_{vib}/\partial T\right)_V$ (from the internal-energy–partition-function relation), differentiate $\ln q_{vib}$ :

\ln q_{vib} = -\frac{\beta h\nu}{2} - \ln\left(1-e^{-\beta h\nu}\right)

Converting the $\beta$ -derivative to a $T$ -derivative gives

U_m - U_m(0) = \frac{N_A h\nu}{e^{h\nu/k_BT}-1}

This is the mean vibrational energy above the zero-point level, distributed over the oscillator's excited states according to the Boltzmann population $e^{-\beta h\nu}$ .

Step 3. Differentiate to get $C_V$ . By definition $C_{V,m} = (\partial U_m/\partial T)_V$ . Writing $x = h\nu/k_BT = \theta_{vib}/T$ ,

C_{V,m}^{vib} = N_A h\nu \cdot \frac{d}{dT}\left(\frac{1}{e^{x}-1}\right) = N_A h\nu \cdot \left(-\frac{1}{(e^x-1)^2}\right)e^{x}\cdot\left(-\frac{h\nu}{k_BT^2}\right)

Collecting terms and substituting $N_Ak_B = R$ gives

C_{V,m}^{vib}(T) = R\, x^2 \frac{e^{x}}{(e^{x}-1)^2}, \qquad x = \frac{\theta_{vib}}{T}

Step 4. Check the two limits. For $T \gg \theta_{vib}$ ( $x\to0$ ), expand $e^x \approx 1+x$ , so $e^x-1 \approx x$ and $(e^x-1)^2 \approx x^2$ ; the whole expression collapses to $C_{V,m}^{vib}\to R\cdot x^2\cdot 1/x^2 = R$ , recovering the classical equipartition value for one vibrational mode (two quadratic terms, each $\tfrac12 R$ ). For $T \ll \theta_{vib}$ ( $x\to\infty$ ), the factor $e^x$ in the numerator is overwhelmed by $(e^x-1)^2 \approx e^{2x}$ in the denominator, so

C_{V,m}^{vib} \to R\,x^2 e^{-x} \to 0

The mode is frozen: essentially all molecules sit in $v=0$ , and raising $T$ slightly cannot populate $v=1$ because $k_BT \ll h\nu$ .

Boxed result.

\boxed{C_{V,m}^{vib}(T) = R\left(\frac{\theta_{vib}}{T}\right)^{2}\frac{e^{\theta_{vib}/T}}{\left(e^{\theta_{vib}/T}-1\right)^{2}}}

This is the Einstein heat-capacity function, later reused (with a modified level structure) to model lattice vibrations in solids.

Worked examples

Example 1 — Foundational: room-temperature $C_{V,m}$ of HCl. HCl has vibrational wavenumber $\tilde\nu = 2990\ \text{cm}^{-1}$ and rotational constant $B = 10.59\ \text{cm}^{-1}$ . Find $C_{V,m}$ at 298 K. Convert to characteristic temperatures using $\theta = hc\tilde\nu/k_B$ , with $hc/k_B = 1.4388\ \text{cm K}$ :

\theta_{vib} = 1.4388 \times 2990 = 4302\ \text{K}

\theta_{rot} = 1.4388 \times 10.59 = 15.24\ \text{K}

Rotation: since $T/\theta_{rot} = 298/15.24 = 19.6 \gg 1$ , rotation is essentially fully classical. Using the expansion $C_{V,m}^{rot} \approx R[1 + \tfrac{1}{45}(\theta_{rot}/T)^2]$ :

C_{V,m}^{rot} = R\left[1 + \frac{1}{45}(0.0511)^2\right] = R(1.000058) \approx R = 8.314\ \text{J K}^{-1}\text{mol}^{-1}

Vibration: compute $x = \theta_{vib}/T = 4302/298 = 14.44$ . Since $x \gg 1$ ,

C_{V,m}^{vib} \approx R\,x^2 e^{-x} = 8.314 \times (14.44)^2 \times e^{-14.44}

$e^{-14.44} \approx 5.36\times10^{-7}$ , and $(14.44)^2 = 208.5$ , so

C_{V,m}^{vib} \approx 8.314 \times 208.5 \times 5.36\times10^{-7} \approx 9.3\times10^{-4}\ \text{J K}^{-1}\text{mol}^{-1}

Essentially zero — vibration contributes nothing at 298 K. Translation contributes $\tfrac32 R = 12.47\ \text{J K}^{-1}\text{mol}^{-1}$ . Total:

C_{V,m} \approx 12.47 + 8.31 + 0.0009 \approx 20.8\ \text{J K}^{-1}\text{mol}^{-1} = \tfrac52 R

This matches the classical rigid-rotor–harmonic-oscillator prediction of $\tfrac52 R$ for a diatomic at ordinary temperature, confirming rotation is fully classical and vibration is fully frozen at 298 K.

Example 2 — Applied: at what temperature does I₂ vibration contribute half its classical value, and what is $C_{V,m}$ for Cl₂ at 1000 K? I₂ has $\tilde\nu = 214.5\ \text{cm}^{-1}$ , giving

\theta_{vib}(\text{I}_2) = 1.4388 \times 214.5 = 308.7\ \text{K}

We want $T$ such that $C_{V,m}^{vib}/R = 0.5$ . Solve

x^2 e^x/(e^x-1)^2 = 0.5

numerically. Trial $x=2.0$ : $e^2 = 7.389$ , $(e^2-1)^2 = 40.79$ , so value $= 4\times7.389/40.79 = 0.724$ . Trial $x=2.4$ : $e^{2.4}=11.02$ , $(10.02)^2=100.4$ , value $=5.76\times11.02/100.4=0.632$ . Trial $x=3.0$ : $e^3=20.09$ , $(19.09)^2=364.4$ , value $=9\times20.09/364.4=0.496$ — close. Refine $x=2.96$ : $e^{2.96}=19.30$ , $(18.30)^2=334.9$ , value $=8.76\times19.30/334.9=0.505$ . Take $x\approx2.98$ .

T = \theta_{vib}/x = 308.7/2.98 = 103.6\ \text{K}

So I₂'s low vibrational frequency (weak, heavy bond) means its vibration is already half-active around 104 K — far below room temperature, consistent with I₂ vibration being significantly excited even near 298 K (there, $x = 308.7/298 = 1.036$ , giving $C_{V,m}^{vib}/R = (1.036)^2 e^{1.036}/(e^{1.036}-1)^2 = 1.073\times2.818/(1.818)^2 = 3.025/3.305 = 0.915$ — vibration is 91.5% activated at room temperature). Now Cl₂: $\tilde\nu = 559.7\ \text{cm}^{-1}$ , so

\theta_{vib}(\text{Cl}_2) = 1.4388 \times 559.7 = 805.3\ \text{K}

At $T = 1000$ K, $x = 805.3/1000 = 0.8053$ .

C_{V,m}^{vib} = R\,x^2\frac{e^{x}}{(e^{x}-1)^2} = 8.314\times(0.8053)^2 \times \frac{e^{0.8053}}{(e^{0.8053}-1)^2}

$e^{0.8053} = 2.238$ , $(e^{0.8053}-1)^2 = (1.238)^2 = 1.533$ , and $(0.8053)^2 = 0.6485$ :

C_{V,m}^{vib} = 8.314 \times 0.6485 \times \frac{2.238}{1.533} = 8.314 \times 0.6485 \times 1.460 = 7.87\ \text{J K}^{-1}\text{mol}^{-1}

Rotation is fully classical at 1000 K ( $\theta_{rot}$ (Cl₂) $\approx 0.351$ K), contributing $R = 8.314\ \text{J K}^{-1}\text{mol}^{-1}$ . Translation gives $\tfrac32R = 12.47\ \text{J K}^{-1}\text{mol}^{-1}$ . Total:

C_{V,m}(1000\,\text{K}) \approx 12.47 + 8.31 + 7.87 = 28.7\ \text{J K}^{-1}\text{mol}^{-1}

This is close to, but below, the fully classical $\tfrac72 R = 29.1\ \text{J K}^{-1}\text{mol}^{-1}$ — vibration is 90% activated but not saturated at 1000 K, exactly as expected since $T/\theta_{vib} = 1.24$ is only modestly above one.

Limiting cases & common traps

Regime	Behavior	Contribution to $C_{V,m}$
$T \ll \theta_{rot}, \theta_{vib}$	only translation active	$\tfrac32 R$
$\theta_{rot} \ll T \ll \theta_{vib}$	rotation classical, vibration frozen	$\tfrac52 R$ (linear rotor)
$T \gg \theta_{vib}$	all modes classical	$\tfrac72 R$ per vibrational mode active
$T \to \infty$ , polyatomic	every quadratic term equipartitioned	$\tfrac12 R \times$ (number of quadratic terms)

The chemist behind it

The equipartition theorem itself grew out of nineteenth-century kinetic theory, but the decisive step — recognizing that its failure for vibration (and later for solids) demanded quantization of energy — belongs to the generation bridging classical and quantum thermodynamics. Ludwig Boltzmann developed the statistical machinery of energy distribution among degrees of freedom that made equipartition rigorous, while the anomalous, temperature-dependent heat capacities of gases and solids became one of the clearest experimental cracks in classical physics, motivating Einstein's 1907 quantum treatment of solid heat capacities — a direct conceptual cousin of the vibrational formula derived above.

Origins · 1 minLudwig Boltzmann and how this was discovered

Ludwig Boltzmann

1844–1906 · Austrian physicist whose statistical mechanics gave equipartition its rigorous foundation and whose entropy formula underlies the partition-function methods used throughout this derivation

Ask the tutor: How would this heat-capacity ladder change for a nonlinear polyatomic molecule with three rotational axes and multiple distinct vibrational frequencies?

Why does translation always behave classically, even at very low temperatures?

Translational energy levels in a macroscopic box are spaced incredibly finely — proportional to $h^2/(8mL^2)$ for a container of length $L$ . For any laboratory-sized container and any atom or molecule, this spacing is many orders of magnitude smaller than $k_BT$ even at temperatures of a few kelvin. Equipartition for translation therefore holds essentially universally for gases above the point where quantum effects like Bose–Einstein or Fermi–Dirac statistics become relevant, which is far below normal experimental conditions.

Does equipartition apply to electronic degrees of freedom?

Rarely at ordinary temperatures. Electronic energy gaps are typically several electron-volts, corresponding to characteristic temperatures of tens of thousands of kelvin. Since $k_BT$ at 298 K corresponds to roughly 0.026 eV, electronic excitation is utterly negligible for most closed-shell molecules. Exceptions exist: molecules with low-lying excited electronic states (like NO, with a small spin-orbit splitting) or with degenerate ground states can show measurable electronic contributions even near room temperature.

How does this connect to the Dulong-Petit law for solids?

Dulong and Petit observed that molar heat capacities of many solids approach $3R$ at high temperature, matching equipartition applied to three-dimensional vibrations (six quadratic terms per atom, three kinetic and three potential, each worth $\tfrac12R$ ). At low temperature, real solids deviate sharply below $3R$ , just as gas-phase vibrations freeze out. Einstein's 1907 model applied exactly the harmonic-oscillator heat-capacity formula derived here, atom by atom, to explain that deviation.

Why does the rotational heat capacity approach R from above rather than from below?

The high-temperature expansion $C_{V,m}^{rot} \approx R[1 + \theta_{rot}^2/(45T^2) + \dots]$ contains a small positive correction term. This reflects the discreteness of the rotational ladder: even when mostly classical, the quantization very slightly enhances the heat capacity above the strict continuum value at finite $T$ , a correction that vanishes as $T\to\infty$ and is normally negligible except for very light molecules like H₂ near room temperature.

Where does this heat-capacity analysis lead next?

Once $U_m(T)$ and $C_{V,m}(T)$ are known mode by mode from the partition function, the same derivative techniques generate entropy, Helmholtz energy, and Gibbs energy. That broader machinery, covered under Other Thermodynamic Functions from the Partition Function, reuses exactly the vibrational and rotational partition functions introduced here.