Boltzmann Factor

Why is the air thinner on a mountaintop? Why do chemical reactions run faster when you heat them? Why does a hot gas glow while a cold one does not? A single expression answers all three: the Boltzmann factor $e^{-E/k_BT}$ , the exponential weight that tells you how likely a system is to be found in a state of energy $E$ at temperature $T$ . It is arguably the most-used formula in all of physics — the exponential that quietly sets populations, rates, densities, and spectral intensities across chemistry, biology, astrophysics, and materials science. The factor's message is stark and simple: energy costs exponentially. To occupy a state that lies energy $E$ above the reference costs a probability suppression of $e^{-E/k_BT}$ , and the natural unit of "expensive" is the thermal energy $k_BT$ . If $E\ll k_BT$ the state is freely accessible; if $E\gg k_BT$ it is exponentially rare. At room temperature $k_BT\approx 25\ \mathrm{meV}=1/40\ \mathrm{eV}$ — the yardstick against which every energy scale in a room-temperature system is measured. An electronic gap of a few eV is $\sim100\,k_BT$ and utterly frozen; a rotational spacing of a few meV is $\lesssim k_BT$ and thermally populated. Because the reference and the normalization cancel when you take a ratio, the cleanest statement is the population ratio $p_1/p_2 = e^{-(E_1-E_2)/k_BT}$ : the relative chance of two states depends only on their energy gap in thermal units. From this one relation flow the barometric formula (states = heights, $E=mgh$ ), the Arrhenius law of reaction rates (states = over-the-barrier configurations, $E=E_a$ ), and the population inversions that make lasers work. This article isolates the Boltzmann factor as a tool — its origin, its ratio form, its role as a thermometer — and computes real numbers with it. The full canonical-ensemble machinery that produces it is developed in the companion articles; here the factor stands on its own.

Key takeaways

The Boltzmann factor $e^{-E/k_BT}$ is the exponential weight of a state of energy $E$ at temperature $T$ ; energy costs exponentially, in units of $k_BT$ .
The population ratio $p_1/p_2 = e^{-(E_1-E_2)/k_BT}$ depends only on the energy gap, not on the reference or the normalization.
Thermal energy $k_BT\approx 25\ \mathrm{meV}$ at $300\ \mathrm{K}$ is the yardstick: $E\ll k_BT$ freely populated, $E\gg k_BT$ exponentially rare.
With $E=mgh$ it gives the barometric formula $n(h)=n_0 e^{-mgh/k_BT}$ ; with $E=E_a$ it gives the Arrhenius law for reaction rates.
A measured population ratio is a thermometer: invert $e^{-\Delta E/k_BT}$ to read the temperature (how astronomers use spectral lines).
Key idea: the probability of a state falls off exponentially with its energy, $\propto e^{-E/k_BT}$ , so a system's occupancy of high-energy states is set entirely by the energy gap measured in units of the thermal energy $k_BT$ .

The core insight

The Boltzmann factor $e^{-E/k_BT}$ gives the exponential suppression of a state with energy $E$ at temperature $T$ . In ratio form, $p_1/p_2 = e^{-(E_1-E_2)/k_BT}$ , the relative population of two states depends only on their energy gap in units of $k_BT$ — the single fact behind the barometric formula, Arrhenius rates, and population statistics everywhere.

Where this lives today

The Boltzmann factor is everywhere the equilibrium occupancy of states matters. In chemistry it is the Arrhenius law, setting reaction rates and the temperature dependence of everything from combustion to enzyme kinetics; in semiconductor physics it sets the density of thermally excited charge carriers ( $\propto e^{-E_g/2k_BT}$ ), which is why a transistor's leakage current rises with temperature and why cooling electronics improves performance. In astrophysics, the ratios of spectral-line intensities — governed by the Boltzmann factor for atomic level populations, refined by the Saha equation for ionization — are the primary thermometers of stars and nebulae; the surface temperature of the Sun is read directly off Boltzmann-weighted line strengths. The same exponential runs through biology and technology. Protein folding, ligand binding, and ion-channel gating are all Boltzmann-weighted equilibria over conformational energies; the melting of DNA is a Boltzmann competition between base-pairing energy and configurational entropy. In machine learning, "Boltzmann machines" and the softmax function used in every neural-network classifier are the Boltzmann distribution in disguise, with a "temperature" parameter controlling how sharply the model commits to its most probable output. And in simulation, the Metropolis Monte Carlo acceptance rule $\min(1, e^{-\Delta E/k_BT})$ — the algorithm behind much of computational science — is nothing but the Boltzmann factor deciding which moves to accept. Few formulas of nineteenth-century physics reach so deeply into twenty-first-century life.

The physical picture

Think of energy as a cost and $k_BT$ as your budget. Occupying a state of energy $E$ "costs" $E$ , and thermal fluctuations give you a budget of about $k_BT$ to spend. If the cost is small compared to the budget ( $E\ll k_BT$ ), the state is easily afforded and freely occupied. If the cost is large ( $E\gg k_BT$ ), you can rarely scrape together enough energy, and the occupancy is exponentially suppressed by $e^{-E/k_BT}$ . The exponential is not a gentle roll-off — it is a cliff. Each additional $k_BT$ of energy divides the population by $e\approx2.7$ ; ten $k_BT$ 's divide it by $\sim22000$ .

Figure

The Boltzmann factor e^{-E/kT}: population falls exponentially with energy. States within ~k_BT are freely occupied; states many k_BT up are exponentially rare.

Because a fixed reference energy and the normalization cancel in a ratio, the comparison between two states is even cleaner than the absolute occupancy. The chance of being in state 1 versus state 2 is $e^{-(E_1-E_2)/k_BT}$ — you never need to know the zero of energy, only the gap. This is why the Boltzmann factor is so portable: it does not care what "states" mean. Let the states be heights in a gravitational field, $E=mgh$ , and you get the exponential atmosphere; let them be configurations at the top of a reaction barrier, $E=E_a$ , and you get the temperature dependence of reaction rates. The punchline is the atmosphere itself. Air molecules at height $h$ have gravitational energy $mgh$ , so their number density obeys $n(h)=n_0\,e^{-mgh/k_BT}$ — an exponential falloff whose scale height $H=k_BT/mg$ is about $8\ \mathrm{km}$ for air. The plot below shows this exponential decay of density with altitude, the Boltzmann factor writ across the sky: the same math that depletes an excited atomic level thins the air on Everest.

Figure

The barometric formula n(h)=n₀ e^{-mgh/kT}: air density falls off exponentially with altitude, scale height H=k_BT/mg ≈ 8 km — the Boltzmann factor with E=mgh.

The core equations

The Boltzmann factor is the exponential weight of a state, and the probability of that state is the factor divided by the sum over all states:

p(E) \propto e^{-E/k_BT} = e^{-\beta E},\qquad \beta=\frac1{k_BT} \qquad (1)

where: $p(E)$ — relative probability of a state of energy $E$ ; $\beta$ — inverse temperature; $k_B=1.381\times10^{-23}\ \mathrm{J/K}=8.617\times10^{-5}\ \mathrm{eV/K}$ . The factor is dimensionless because $E$ and $k_BT$ share units. The most useful form is the ratio, where the reference and normalization drop out:

\frac{p_1}{p_2} = \frac{g_1}{g_2}\,e^{-(E_1-E_2)/k_BT} \qquad (2)

where: $g_1,g_2$ — degeneracies (numbers of states) at each energy; the exponential depends only on the gap $\Delta E=E_1-E_2$ . Without degeneracies ( $g_1=g_2$ ) it is the bare $e^{-\Delta E/k_BT}$ . Applied to a gravitational field, states are heights and $E=mgh$ , giving the barometric (isothermal-atmosphere) formula:

n(h) = n_0\,e^{-mgh/k_BT},\qquad H=\frac{k_BT}{mg} \qquad (3)

where: $n(h)$ — number density at height $h$ ; $n_0$ — density at $h=0$ ; $H$ — scale height, the altitude over which the density falls by a factor $e$ . For air ( $m\approx4.8\times10^{-26}\ \mathrm{kg}$ , $T\approx288\ \mathrm{K}$ ), $H\approx8.4\ \mathrm{km}$ . Applied to a reaction barrier, states are over-the-barrier configurations and $E=E_a$ , giving the Arrhenius temperature dependence of rates:

k_{\text{rate}} = A\,e^{-E_a/k_BT} \qquad (4)

where: $E_a$ — activation energy (the barrier height); $A$ — pre-exponential (attempt) factor. The exponential Boltzmann factor is the fraction of molecules with enough thermal energy to surmount the barrier, and it is why modest heating dramatically accelerates reactions.

Derivation AThe population ratio and its use as a thermometer

The goal is to establish the ratio form of the Boltzmann factor — the statement that the relative population of two states depends only on their energy gap — and to invert it into a thermometer, showing how a measured population ratio yields the temperature. The strategy is to write the two probabilities, take their ratio so the normalization cancels, and solve for $T$ .

Step 1. The probability of a state of energy $E$ is $p(E)=g(E)e^{-\beta E}/Z$ , where $g(E)$ is the degeneracy and $Z$ the normalizing sum. Form the ratio of two states: $\dfrac{p_1}{p_2}=\dfrac{g_1 e^{-\beta E_1}}{g_2 e^{-\beta E_2}}=\dfrac{g_1}{g_2}e^{-\beta(E_1-E_2)}$ . Why: the common denominator $Z$ cancels, and only the ratio of Boltzmann factors and degeneracies survives — no reference energy is needed.

Step 2. For non-degenerate states ( $g_1=g_2$ ) this is the bare $e^{-\Delta E/k_BT}$ with $\Delta E=E_1-E_2$ . If state 1 is higher, $\Delta E>0$ and the ratio is $<1$ : the upper state is less populated, exponentially so. Why: the exponential penalizes the higher-energy state by exactly the gap in thermal units, independent of everything else.

Step 3. Invert to get temperature. Solve $p_1/p_2=(g_1/g_2)e^{-\Delta E/k_BT}$ for $T$ : $\dfrac1{T}=\dfrac{k_B}{\Delta E}\ln\!\left(\dfrac{g_1 p_2}{g_2 p_1}\right)$ . A single measured ratio of populations, plus the known gap, fixes the temperature. Why: the population ratio is a monotonic function of $T$ , so measuring it and knowing $\Delta E$ pins down $T$ uniquely — the physical basis of spectroscopic thermometry.

Step 4. Note the sign structure: normal (positive- $T$ ) systems always have the lower state more populated, $p_1<p_2$ for $E_1>E_2$ . A population inversion ( $p_1>p_2$ for $E_1>E_2$ ) cannot occur at any positive temperature — it corresponds formally to $T<0$ and is the essential ingredient of a laser. Why: at any positive temperature the exponential forces the upper level to be less populated, so achieving inversion requires driving the system out of thermal equilibrium. *Verified: $p_1/p_2=e^{-\beta(E_1-E_2)}$ symbolically; the population-ratio thermometer and the $g$ -weighted ratio are checked numerically, in verify_aq-sm-boltzmann-factor.py.*

Go deeper · Derivation B · 1 minThe barometric formula and the meaning of the thermal energy scale — the owned result

Derivation BThe barometric formula and the meaning of the thermal energy scale — the owned result

This is the article's owned structural result: the Boltzmann factor applied to gravitational potential energy produces the exponential atmosphere, and the scale height $H=k_BT/mg$ makes concrete the meaning of " $k_BT$ is the thermal energy yardstick" — it is the height at which gravitational energy equals thermal energy. The goal is to derive $n(h)=n_0 e^{-mgh/k_BT}$ and interpret the scale height. The strategy is to apply the Boltzmann factor with $E=mgh$ and then cross-check against mechanical (hydrostatic) equilibrium.

Step 1. A gas molecule at height $h$ has gravitational potential energy $E=mgh$ . By the Boltzmann factor, the probability of finding a molecule at height $h$ is $\propto e^{-mgh/k_BT}$ , so the number density is $n(h)=n_0 e^{-mgh/k_BT}$ . Why: height is just another state label, and the Boltzmann weight $e^{-E/k_BT}$ with $E=mgh$ gives the relative occupancy of different altitudes.

Step 2. The scale height $H=k_BT/mg$ is the altitude at which the density falls by a factor $e$ : $n(H)=n_0/e$ . It is the height where the gravitational energy $mgH$ equals the thermal energy $k_BT$ . Why: setting $mgh=k_BT$ gives $h=k_BT/mg=H$ , so the scale height is literally where the energy cost of climbing equals the thermal budget — the geometric meaning of $k_BT$ .

Step 3. Cross-check with hydrostatics. Mechanical equilibrium of a gas column gives $dp/dh=-\rho g=-(mn)g$ and the ideal-gas law $p=n k_BT$ , so $k_BT\,dn/dh=-mgn$ , i.e. $dn/n=-(mg/k_BT)dh$ , integrating to $n(h)=n_0 e^{-mgh/k_BT}$ . Why: the same exponential emerges from balancing pressure against gravity — a completely independent derivation, confirming the Boltzmann-factor result and showing thermodynamics and mechanics agree.

Step 4. Numbers: for air, $m\approx4.8\times10^{-26}\ \mathrm{kg}$ , $g=9.81\ \mathrm{m/s^2}$ , $T\approx288\ \mathrm{K}$ give $H=k_BT/mg\approx8.4\ \mathrm{km}$ — consistent with the observed thinning of the atmosphere (density roughly halves every $\sim5.5$ km, since $H\ln2\approx5.8$ km). Why: the derived scale height matches the real atmosphere, demonstrating the Boltzmann factor governs a system you can climb. *Verified: $H=k_BT/mg\approx8.4$ km for air; $n(8\,\mathrm{km})/n_0=e^{-mgh/k_BT}\approx0.33$ – $0.40$ ; and the thermal energy $k_BT=25.85$ meV at 300 K, all in the verify script.*

Limiting cases

The Boltzmann factor across regimes:

Case	Result	Meaning
High temperature	$\beta\to0$ : $e^{-\beta E}\to1$	All states equally weighted; energy differences irrelevant.
Low temperature	$\beta\to\infty$ : $e^{-\beta E}\to0$ (for $E>0$ )	Only the lowest state survives; upper states frozen out.
$E = k_BT$	factor $= e^{-1} = 0.37$	The natural crossover: states at one $k_BT$ are suppressed by $1/e$ .
$E \gg k_BT$	exponentially tiny	Electronic gaps ( $\sim$ eV $\approx 40\,k_BT$ ) are essentially unoccupied at room $T$ .
Formal $T<0$	upper state MORE populated	Population inversion — impossible at equilibrium, the basis of lasers.

Common trap“The Boltzmann factor gives the probability of an energy level”

Careful — it gives the relative probability of a single microstate (or state), not of an energy level. If a level of energy $E$ is $g$ -fold degenerate (there are $g$ distinct states with that energy), the probability of finding the system at that energy is $g\,e^{-E/k_BT}/Z$ , not $e^{-E/k_BT}/Z$ . Forgetting the degeneracy $g$ is the single most common error: it makes highly-degenerate excited levels look far less populated than they are, and it wrecks quantitative population ratios and spectroscopic thermometry. A related trap is the entropy versus energy competition: a level can be high in energy (small Boltzmann factor) yet highly populated if it is enormously degenerate (large $g$ ) — this entropy-energy tug-of-war is exactly what drives phase transitions and the temperature dependence of chemical equilibria. Finally, remember $E$ must be measured from a consistent reference, and only differences matter for ratios.

Worked examples

1 · Why atoms sit in their ground state at room temperature (a real number)

An atom has an electronic excited state $\varepsilon = 1.0\ \mathrm{eV}$ above the ground state (both non-degenerate). At $T=300\ \mathrm{K}$ , what fraction of atoms are electronically excited? Contrast this with a rotational level $\varepsilon' = 2\ \mathrm{meV}$ above the ground state.

Step 1. Compute the thermal energy: $k_BT = (8.617\times10^{-5}\ \mathrm{eV/K})(300\ \mathrm{K}) = 0.02585\ \mathrm{eV} = 25.85\ \mathrm{meV}$ .

Step 2. Electronic level: $\beta\varepsilon = \varepsilon/k_BT = 1.0/0.02585 = 38.7$ . The Boltzmann factor is $e^{-38.7} = \mathbf{1.6\times10^{-17}}$ . Essentially no atoms are electronically excited — the gap is nearly $40\,k_BT$ , an exponential wall.

Step 3. Rotational level: $\beta\varepsilon' = 0.002/0.02585 = 0.077$ . The Boltzmann factor is $e^{-0.077} = \mathbf{0.926}$ . The excited rotational state is populated almost as much as the ground state — the gap is a small fraction of $k_BT$ .

Step 4. The contrast is the whole point. Electronic transitions ( $\sim$ eV) are frozen at room temperature, which is why matter is chemically stable and does not glow; rotational and vibrational levels ( $\sim$ meV) are thermally active, which is why molecules have temperature-dependent heat capacities. The single knob is $E/k_BT$ : the same temperature freezes one and activates the other.

2 · Speeding up a reaction by heating it (exam level)

A chemical reaction has activation energy $E_a = 0.75\ \mathrm{eV}$ . By what factor does its rate increase when the temperature rises from $300\ \mathrm{K}$ to $350\ \mathrm{K}$ , according to the Arrhenius law?

Step 1. The rate is $k=A e^{-E_a/k_BT}$ , so the ratio of rates at the two temperatures is $\dfrac{k(350)}{k(300)} = \exp\!\left[-\dfrac{E_a}{k_B}\left(\dfrac1{350}-\dfrac1{300}\right)\right]$ — the pre-factor $A$ cancels.

Step 2. Compute $E_a/k_B = 0.75/(8.617\times10^{-5}) = 8703\ \mathrm{K}$ . And $\dfrac1{300}-\dfrac1{350} = 0.003333-0.002857 = 4.76\times10^{-4}\ \mathrm{K^{-1}}$ .

Step 3. Exponent: $-E_a/k_B\times(1/350-1/300) = +8703\times4.76\times10^{-4} = +4.14$ . (Careful with the sign: the rate increases on heating, so the exponent is positive.)

Step 4. Rate ratio: $e^{4.14} = \mathbf{63}$ . A $50\,\mathrm{K}$ temperature rise speeds the reaction up more than sixtyfold — the familiar "rule of thumb" that reaction rates roughly double every $10\,\mathrm{K}$ (here $63^{1/5}\approx2.3$ per $10\,\mathrm K$ ). This exponential sensitivity is why cooking, corrosion, and biology are all so temperature-dependent, and it is the Boltzmann factor at work.

Origins · 1 minboltzmann distribution and how this was discovered

boltzmann distribution

statistical-mechanics law · Boltzmann 1868–77

The Boltzmann distribution — the exponential law $\propto e^{-E/k_BT}$ that the Boltzmann factor embodies — is Ludwig Boltzmann's central contribution to statistical physics, developed through the 1860s and 1870s. Boltzmann showed that a system in contact with a heat bath populates its states with probability falling off exponentially in the energy, a result that unified the kinetic theory of gases (Maxwell's velocity distribution is a special case) with the emerging science of thermodynamics. The constant $k_B$ , relating energy to temperature, was named in his honour by Planck, who first wrote it explicitly; it is engraved (as $S=k\log W$ ) on Boltzmann's tombstone in Vienna. The distribution is the beating heart of equilibrium statistical mechanics: it governs everything from the color of a flame to the rate of a reaction to the occupancy of a semiconductor's conduction band. Boltzmann fought bitter battles defending the atomic and statistical view of matter against the energeticists; the ubiquity of "his" factor is a vindication written across the whole of modern physics.

History

1860

— Maxwell derives the velocity distribution, the kinetic-energy special case of the Boltzmann factor.

1868

— Boltzmann generalizes to the exponential distribution over arbitrary energies for a system in a heat bath.

1877

— Boltzmann's relation S = k log W connects entropy to microstate counting, underpinning the factor.

1889

— Arrhenius formulates the rate law k = A e^{−E_a/RT}, the Boltzmann factor applied to reaction barriers.

1900

— Planck introduces k_B explicitly (and h) in the blackbody law, giving the constant its name and value.

Going further

References. Reif, Fundamentals of Statistical and Thermal Physics, Ch. 6 · Pathria & Beale, Statistical Mechanics, 3rd ed., Ch. 3 · Kardar, Statistical Physics of Particles, Ch. 4 · Schroeder, An Introduction to Thermal Physics, Ch. 6 · Blundell & Blundell, Concepts in Thermal Physics, Ch. 4 · Atkins, Physical Chemistry (for the Arrhenius and spectroscopy applications).

Provenance. The Boltzmann factor, its ratio form, the barometric formula, and the Arrhenius law are standard (Reif Ch. 6; Schroeder Ch. 6). A verifier should reproduce: $p_1/p_2=e^{-\beta(E_1-E_2)}$ ; the barometric $n(h)=n_0 e^{-mgh/k_BT}$ with scale height $H=k_BT/mg\approx8.4$ km for air; the Arrhenius rate ratio; and $k_BT=25.85$ meV at 300 K. Numerical instances (electronic $e^{-38.7}\approx1.6\times10^{-17}$ ; rotational $e^{-0.077}\approx0.93$ ; reaction rate ratio $e^{4.14}\approx63$ for $E_a=0.75$ eV over 300→350 K; population-ratio thermometer) are recomputed in verify_aq-sm-boltzmann-factor.py. $k_B=8.617\times10^{-5}\ \mathrm{eV/K}$ (CODATA).

Builds on: The Microcanonical Ensemble and the Postulate of Equal a Priori Probabilities · Temperature, Pressure, and Chemical Potential as Entropy Derivatives. Leads to: Canonical Partition Function · Canonical Ensemble · Derivation of the Canonical Distribution from the Microcanonical Ensemble · Helmholtz Free Energy from the Partition Function · Gibbs Factor and Chemical Potential · Factorization of the Partition Function.

Where does the k_B in the Boltzmann factor come from, and why temperature in the denominator?

The Boltzmann constant $k_B$ is the conversion factor between temperature and energy: multiply a temperature by $k_B$ and you get an energy, the "thermal energy" $k_BT$ . It appears because temperature and energy are physically different quantities measured in different units (kelvin versus joules), and $k_B\approx1.38\times10^{-23}\ \mathrm{J/K}$ bridges them. Temperature sits in the denominator of the exponent because it sets the scale of accessible energy: a higher temperature means a larger thermal budget $k_BT$ , so a given energy $E$ becomes a smaller fraction $E/k_BT$ and the state is less suppressed. As $T\to\infty$ the ratio $E/k_BT\to0$ and every state is equally weighted; as $T\to0$ the ratio blows up and all but the lowest states are frozen out. The combination $E/k_BT$ is dimensionless — it is energy measured in units of the thermal energy — which is why "how many $k_BT$ 's" is the natural way to judge whether a state is accessible.

Why does the Boltzmann factor make reaction rates so sensitive to temperature?

Because the fraction of molecules with enough energy to cross the reaction barrier is a Boltzmann factor, $e^{-E_a/k_BT}$ , and an exponential of a large number is ferociously sensitive to its argument. Activation energies are typically many $k_BT$ (often $20$ – $40\,k_BT$ at room temperature), so the exponent is large and negative, and a small change in $T$ produces a large fractional change in the rate. Concretely, raising the temperature slightly shrinks $E_a/k_BT$ a little, but because it sits in an exponential, the rate can double or triple — the familiar rule that reaction rates roughly double for every $10\,\mathrm K$ . This is why food spoils faster in summer, why cold slows biology to a crawl (hence refrigeration and cryopreservation), and why industrial chemistry runs hot. The sensitivity is entirely the exponential character of the Boltzmann factor: linear changes in temperature produce exponential changes in the number of molecules that clear the energy barrier.

Does the Boltzmann factor apply to individual particles or to the whole system?

Both, and keeping the two straight avoids confusion. Fundamentally, $e^{-E/k_BT}$ is the weight of a microstate of the whole system of total energy $E$ — that is the rigorous statement from the canonical ensemble. But when the system is made of independent, non-interacting parts (an ideal gas, independent spins, separate molecules), the total energy is a sum and the total Boltzmann factor factorizes into a product of single-particle factors. Then you can, and usually do, apply the Boltzmann factor to each particle's energy separately: the probability that a given molecule is in a single-particle state of energy $\epsilon$ is $\propto e^{-\epsilon/k_BT}$ . This is why we speak of "the fraction of molecules in the excited state" or "the density of air at height $h$ " — each particle independently obeys the Boltzmann law. The factorization fails when particles interact strongly (then only the whole-system statement holds) or when quantum indistinguishability matters at low temperature (then Bose–Einstein or Fermi–Dirac statistics replace the simple factor).

Can the population of a higher-energy state ever exceed that of a lower one?

Not at thermal equilibrium with a positive temperature. The ratio $p_{\text{upper}}/p_{\text{lower}}=(g_u/g_l)e^{-\Delta E/k_BT}$ is always less than $g_u/g_l$ for $\Delta E>0$ , so absent large degeneracy the upper level is less populated — that is what "positive temperature" means. However, you can get a population inversion (upper more populated than lower) by driving the system out of equilibrium with an external energy source — optical or electrical pumping. Such an inversion corresponds formally to a negative temperature (plug $p_u>p_l$ into the ratio and solve: you need $T<0$ ), which is not "colder than absolute zero" but "hotter than infinite temperature" in a system with a bounded spectrum. Population inversion is the essential ingredient of a laser: you pump atoms into an excited state faster than they decay, creating more upper-state atoms than lower-state ones, so that stimulated emission amplifies light rather than being absorbed. Degeneracy can also boost an upper level's total population if $g_u\gg g_l$ , but that is the level being favored by entropy, not by the Boltzmann factor.

How do astronomers use the Boltzmann factor to measure the temperature of a star?

By reading the ratio of spectral-line intensities, which reflects the ratio of atomic level populations, which the Boltzmann factor ties to temperature. When starlight passes through the star's atmosphere, atoms absorb at wavelengths corresponding to transitions between energy levels, producing dark absorption lines whose strength depends on how many atoms are in the relevant lower level. Two lines originating from levels of different energy have an intensity ratio proportional to $(g_1/g_2)e^{-(E_1-E_2)/k_BT}$ — the Boltzmann factor. Measuring the ratio and knowing the energies and degeneracies (from atomic physics), astronomers invert the relation to solve for $T$ . This is refined by the Saha equation, which handles the temperature dependence of ionization (how many atoms are ionized versus neutral), but the core thermometer is the Boltzmann factor: the strength of the hydrogen Balmer lines, for instance, peaks around $10000\ \mathrm K$ precisely because that is where the Boltzmann and Saha factors conspire to put the most hydrogen atoms in the second energy level. Stellar spectral classification (O, B, A, F, G, K, M) is essentially a temperature sequence read through Boltzmann-weighted line strengths.

Is the Boltzmann factor the same thing as the partition function?

No — they are related but distinct. The Boltzmann factor is the un-normalized weight of a single state, $e^{-\beta E_i}$ . The partition function $Z=\sum_i e^{-\beta E_i}$ is the sum of the Boltzmann factors over all states, and it serves as the normalization that turns the factors into genuine probabilities, $p_i=e^{-\beta E_i}/Z$ . Think of the Boltzmann factors as un-normalized "votes" for each state and $Z$ as the total number of votes, so that dividing gives the fraction. The partition function is much more than a normalizer, though: because it packages the entire spectrum, its logarithm and derivatives generate all of thermodynamics ( $U=-\partial_\beta\ln Z$ , $F=-k_BT\ln Z$ , and so on). So the Boltzmann factor is the atom, and $Z$ is the molecule built from summing them — you need the factor to write down a probability at all, and you need $Z$ to normalize it and to extract thermodynamic quantities. The next articles develop $Z$ and its consequences in full.