Interpreting q as a Count of Accessible States

Imagine a single molecule bathed in a heat bath at temperature $T$ . It can sit in its ground state, or hop up into excited states, and the warmer the bath, the more states it can plausibly reach. The molecular partition function $q$ turns this vague idea into a precise number. It counts how many states are "in play" thermally. This matters because $q$ is the bridge between a molecule's energy-level spectrum and every bulk thermodynamic property you care about: entropy, heat capacity, equilibrium constants, and more. Without a clear physical reading of $q$ , it becomes just a scary sum. With one, it becomes intuitive. This article builds that intuition. You will see why $q$ approaches 1 at low temperature. You will see why it grows without bound at high temperature. You will connect both limits to a simple counting picture, and you will work two full numerical examples.

Key takeaways

$q$ is defined as a Boltzmann-weighted sum over all states, $q = \sum_i g_i e^{-\varepsilon_i/k_BT}$ .
At $T \to 0$ , only the ground state contributes meaningfully, so $q \to g_0$ , the ground-state degeneracy.
At $T \to \infty$ , or whenever level spacing is small compared to $k_BT$ , many terms contribute comparably and $q$ grows large.
$q$ is best read as the effective number of thermally accessible states, not literally a state count.
The crossover temperature, where $k_BT$ matches the level spacing, sets the boundary between the two regimes.
Key idea: $q$ measures how many quantum states a molecule can meaningfully explore at a given temperature, smoothly interpolating between "just the ground state" and "a huge accessible manifold."

The core insight

The molecular partition function $q$ is a temperature-dependent headcount of quantum states weighted by their Boltzmann probability, so it shrinks toward the ground-state degeneracy when thermal energy is scarce and grows toward the total number of low-lying states when thermal energy is abundant.

The physical picture

Every molecule has a ladder of quantum energy levels, set by its electronic, vibrational, rotational, and translational structure (the details of which live in the article on the molecular partition function). At any instant, a molecule sampled from a huge ensemble at temperature $T$ has some probability of occupying each level, given by the Boltzmann distribution:

p_i = \frac{g_i e^{-\varepsilon_i / k_BT}}{q}

The partition function $q$ is simply the normalizing sum in the denominator, but that undersells its importance. Because $q$ appears as the sum of all the unnormalized Boltzmann factors, it directly measures how "spread out" the population is across the ladder. Picture the extremes. If the bath is extremely cold, $k_BT$ is tiny compared to the gap between the ground state and the first excited state. Every excited-state Boltzmann factor $e^{-\varepsilon_i/k_BT}$ is vanishingly small, so the sum collapses to just the ground-state term. The molecule is, for all practical purposes, frozen into its lowest energy state (or states, if degenerate). In this limit $q$ equals the ground-state degeneracy $g_0$ — literally the number of quantum states tied for lowest energy. Now picture the opposite extreme: a hot bath, or a ladder with very closely spaced rungs (translational and rotational levels are typically like this at room temperature). Here $k_BT$ dwarfs the spacing between many levels. Boltzmann factors for level after level are close to 1, so the sum accumulates many comparable contributions. The molecule is no longer confined; it can be found anywhere across a wide swath of the ladder. In this limit $q$ becomes large, and it behaves like a genuine count of the number of levels for which $k_BT$ exceeds the excitation energy. This is why the phrase "effective number of accessible states" is the right mental image. It is not that $q$ literally counts discrete occupied states one-by-one at high $T$ — it is a weighted count, where states contribute in proportion to their Boltzmann probability relative to the ground state. A state with $\varepsilon_i \gg k_BT$ contributes almost nothing; a state with $\varepsilon_i \lesssim k_BT$ contributes close to a full unit.

Figure

Population distribution across an energy ladder at low versus high temperature

The core relations

q = \sum_i g_i\, e^{-\varepsilon_i / k_BT}

where: $g_i$ is the degeneracy of level $i$ , $\varepsilon_i$ its energy above the ground state, $k_B$ Boltzmann's constant, $T$ the temperature.

\lim_{T \to 0} q = g_0

where: $g_0$ is the degeneracy of the ground state alone; all excited-state terms vanish.

\lim_{k_BT \gg \Delta\varepsilon} q \approx \sum_{i=0}^{N} g_i

where: $\Delta\varepsilon$ is the characteristic level spacing, $N$ the number of levels lying within roughly $k_BT$ of the ground state.

q \approx g_0\left(1 + \frac{g_1}{g_0}e^{-\Delta\varepsilon/k_BT} + \cdots\right)

where: this is the two-level (or low-lying-level) expansion useful near the crossover between the limits.

Derivation AThe two limiting forms of q from a simple two-level ladder

Goal. Show explicitly how $q$ reduces to $g_0$ at low $T$ and grows toward the full state count at high $T$ , using the simplest nontrivial system: two nondegenerate levels separated by energy $\Delta\varepsilon$ .

Step 1. Write the exact sum. Take $g_0 = g_1 = 1$ , ground energy $\varepsilon_0 = 0$ , excited energy $\varepsilon_1 = \Delta\varepsilon$ . Then

q = e^{-0/k_BT} + e^{-\Delta\varepsilon/k_BT} = 1 + e^{-\Delta\varepsilon/k_BT}

This is exact for any temperature — no approximation yet.

Step 2. Take the low-temperature limit. As $T \to 0$ , the exponent $-\Delta\varepsilon/k_BT \to -\infty$ , so

e^{-\Delta\varepsilon/k_BT} \to 0

Substituting back,

q \to 1 + 0 = 1 = g_0

Physically: the thermal energy $k_BT$ is too small to promote any appreciable population to $\varepsilon_1$ , so the molecule is essentially locked in the ground state, and $q$ literally equals the ground-state degeneracy.

Step 3. Take the high-temperature limit. When $k_BT \gg \Delta\varepsilon$ , expand the exponential:

e^{-\Delta\varepsilon/k_BT} \approx 1 - \frac{\Delta\varepsilon}{k_BT} + \cdots \approx 1

Substituting,

q \to 1 + 1 = 2

Both levels are now equally likely — $q$ equals the total number of accessible levels, exactly matching the naive counting picture.

Step 4. Generalize to a ladder of $N+1$ closely spaced levels. If all levels lie within $k_BT$ of the ground state (spacing $\Delta\varepsilon \ll k_BT$ ), each Boltzmann factor is close to 1, so

q \approx \sum_{i=0}^{N} g_i e^{-\varepsilon_i/k_BT} \approx \sum_{i=0}^N g_i

which is just the total degeneracy of all thermally reachable levels.

Step 5. Identify the crossover. The two regimes meet where $k_BT \sim \Delta\varepsilon$ , i.e.

T^{*} \sim \frac{\Delta\varepsilon}{k_B}

Below $T^*$ , $q$ is pinned near $g_0$ . Above $T^*$ , $q$ climbs roughly linearly (for evenly spaced ladders) or exponentially (for continuum-like spacing) with $T$ .

Boxed result.

\boxed{q \xrightarrow{T \to 0} g_0 \, ; \qquad q \xrightarrow{k_BT \gg \Delta\varepsilon} \sum_{i} g_i \; (\text{effective total state count})}

Worked examples

Example 1 — A two-level system at two temperatures (foundational). Consider a molecular degree of freedom modeled as two nondegenerate levels separated by $\Delta\varepsilon = 300\ \text{cm}^{-1}$ , a typical low-frequency vibrational spacing. Convert to joules:

\Delta\varepsilon = 300\ \text{cm}^{-1} \times (1.986 \times 10^{-23}\ \text{J per cm}^{-1}) = 5.96 \times 10^{-21}\ \text{J}

(a) At $T = 50$ K. Compute $k_BT$ :

k_BT = (1.381\times10^{-23}\ \text{J K}^{-1})(50\ \text{K}) = 6.90\times10^{-22}\ \text{J}

\frac{\Delta\varepsilon}{k_BT} = \frac{5.96\times10^{-21}}{6.90\times10^{-22}} = 8.64

q = 1 + e^{-8.64} = 1 + 1.77\times10^{-4} = 1.0002

Nearly the whole population sits in the ground state; $q \approx g_0 = 1$ , consistent with the cold limit. (b) At $T = 500$ K. Now

k_BT = (1.381\times10^{-23})(500) = 6.905\times10^{-21}\ \text{J}

\frac{\Delta\varepsilon}{k_BT} = \frac{5.96\times10^{-21}}{6.905\times10^{-21}} = 0.863

q = 1 + e^{-0.863} = 1 + 0.422 = 1.422

Sanity check: $q$ must lie strictly between 1 (all ground state) and 2 (both levels equally populated); 1.422 sits comfortably in between, appropriately closer to 2 since $k_BT$ now nearly matches $\Delta\varepsilon$ . The excited-state population fraction is $p_1 = 0.422/1.422 = 0.297$ , meaning roughly 30% of molecules occupy the upper level — a substantial, but not dominant, excited population, exactly the qualitative picture expected near the crossover $T^* = \Delta\varepsilon/k_B = 5.96\times10^{-21}/1.381\times10^{-23} = 431\ \text{K}$ .

Example 2 — Rotational-like ladder approaching the high-temperature count (applied). Consider a set of five evenly spaced, nondegenerate levels at $\varepsilon_i = i \times \delta$ for $i = 0,1,2,3,4$ , with $\delta = 50\ \text{cm}^{-1} = 9.93\times10^{-22}\ \text{J}$ (a rough stand-in for closely spaced rotational sublevels). We ask: at what temperature does $q$ reach 80% of its maximum possible value of 5 (i.e., $q = 4.0$ ), and what does that say about accessible states? Exact sum:

q(T) = \sum_{i=0}^{4} e^{-i\delta/k_BT}

Let $x = \delta/k_BT$ . We need

q = 1 + e^{-x} + e^{-2x} + e^{-3x} + e^{-4x} = 4.0

Try $x = 0.30$ : terms are $1, e^{-0.30}=0.741, e^{-0.60}=0.549, e^{-0.90}=0.407, e^{-1.20}=0.301$ , summing to

q = 1 + 0.741 + 0.549 + 0.407 + 0.301 = 2.998

Too small — need smaller $x$ (higher $T$ ). Try $x = 0.15$ : terms are $1, e^{-0.15}=0.861, e^{-0.30}=0.741, e^{-0.45}=0.638, e^{-0.60}=0.549$ , giving

q = 1 + 0.861 + 0.741 + 0.638 + 0.549 = 3.789

Still slightly low; try $x = 0.12$ : terms $1, 0.887, 0.787, 0.698, 0.619$ , sum

q = 1 + 0.887 + 0.787 + 0.698 + 0.619 = 3.991 \approx 4.0

Good enough. So $x \approx 0.12$ , meaning

\frac{\delta}{k_BT} = 0.12 \implies T = \frac{\delta}{0.12\,k_B} = \frac{9.93\times10^{-22}}{0.12\times1.381\times10^{-23}} = 599\ \text{K}

At $T \approx 599$ K, $k_BT = 8.27\times10^{-21}\ \text{J}$ , which is over eight times the level spacing $\delta$ — consistent with $q$ approaching its ceiling of 5, since when $k_BT \gg \delta$ nearly all five levels are comparably populated. This confirms the high-temperature reading of $q$ : it approaches the literal count of levels once thermal energy comfortably exceeds the spacing between them, exactly as predicted by Step 4 of the derivation.

Limiting cases & common traps

Regime	Condition	Behavior of $q$	Physical reading
Very cold	$k_BT \ll \Delta\varepsilon$	$q \to g_0$	Only ground state populated
Crossover	$k_BT \sim \Delta\varepsilon$	$q$ rises smoothly	Partial excited population
Very hot / dense ladder	$k_BT \gg \Delta\varepsilon$	$q \to \sum g_i$ (large)	Many levels comparably populated
Continuum-like (translation)	spacing $\to 0$	$q$ grows roughly as $T^{3/2}$	Effectively unbounded state count

The chemist behind it

The molecular partition function's modern form owes much to the statistical mechanics built by Ludwig Boltzmann, but its systematic application to molecular energy levels — translational, rotational, vibrational, electronic — and its use to compute real thermodynamic quantities from spectroscopic data was substantially developed and popularized in physical chemistry by Richard C. Tolman and later codified pedagogically by figures like Donald McQuarrie, whose textbook treatment of $q$ as an "accessible states" counter is the direct ancestor of the picture used here.

Origins · 1 minLudwig Boltzmann and how this was discovered

Ludwig Boltzmann

1844–1906 · Established the statistical link between microscopic states and macroscopic entropy, the conceptual root of the partition function

Ask the tutor: How does this state-counting picture of q lead directly to the translational partition function's characteristic $T^{3/2}$ scaling?

Why does q equal the ground-state degeneracy at low temperature rather than just 1

Degeneracy means multiple distinct quantum states share the same lowest energy. Each of those states contributes an equal Boltzmann factor of 1 (since $\varepsilon_0$ is the energy reference), so the sum over the ground manifold alone gives $g_0$ , not 1. Only when the ground state is nondegenerate ( $g_0 = 1$ ), as in most vibrational problems, does $q \to 1$ exactly. For a degenerate ground term, such as some electronic states with orbital or spin degeneracy, $q \to g_0 > 1$ even at absolute zero.

Can q ever be less than 1

No. Every term in the sum $q = \sum_i g_i e^{-\varepsilon_i/k_BT}$ is positive, and the ground-state term alone contributes at least $g_0 \geq 1$ (assuming energies are measured relative to the ground state, so its exponential factor is exactly 1). Therefore $q \geq g_0 \geq 1$ always, with equality only in the strict zero-temperature limit for a nondegenerate ground state.

How does this counting picture connect to entropy

A larger $q$ signals that population is spread across more states, which is precisely what higher entropy means microscopically. The statistical entropy formula $S = k_B \ln W$ , where $W$ counts accessible microstates for the full assembly, is intimately linked to $q$ : for an ideal gas, $S$ can be written in terms of $\ln q$ (per the Sackur-Tetrode-type expressions built from $q$ ). Roughly, $\ln q$ behaves like a measure of the "spread" of occupied states, so growing $q$ directly tracks growing entropy.

Does every degree of freedom reach the high-temperature limit at the same temperature

No, and this is a crucial point. Translational and rotational level spacings are typically far smaller than $k_BT$ at room temperature, so their partition functions are already deep in the high-temperature, large- $q$ regime. Vibrational spacings are often comparable to or larger than $k_BT$ at 298 K, so vibrational $q$ can still sit close to 1. Electronic spacings are usually so large that electronic $q$ stays at $g_0$ (often 1) even at very high temperatures. Each degree of freedom has its own effective crossover temperature.

Is q ever exactly equal to an integer at finite temperature

Only in special, engineered cases — for example, if you truncate a ladder and drive $T$ so high that every retained level's Boltzmann factor rounds to numerically 1 within your precision, $q$ approaches (but mathematically never exactly reaches) the integer sum of degeneracies. In real, infinite ladders (like true rotational or translational spectra) $q$ is essentially never exactly an integer; it is a smooth, continuously varying function of $T$ .