Quantum statistics — distinguishable vs indistinguishable particles: Visual walkthrough

Step 1 — What is "one energy level" and why we stare at just one

WHAT. Picture a shelf of allowed energies for a single particle. Each rung is a single-particle state — a slot a particle can be in — and each rung sits at a height $ε$ (its energy). We zoom in on one rung and ask: how many particles are standing on it right now, on average?

WHY. In a big system there are trillions of rungs. That sounds hopeless — until we notice the rungs are independent: what happens on one rung does not physically depend on another (the particles only talk to the shared reservoir, not to each other, in an ideal gas). So if we solve one rung completely, we can just repeat it for every rung. This is the whole trick: reduce a huge problem to one shelf.

PICTURE. Look at the amber rung — that is the level we fix our eyes on for the entire page.

Figure — Quantum statistics — distinguishable vs indistinguishable particles

Step 2 — Letting particles come and go: the reservoir

WHAT. We connect our one rung to a huge reservoir — a giant bath of particles and heat. Particles hop onto our rung and hop off. The reservoir fixes two knobs:

Temperature $T$ — how violently things jiggle. We package it as $β = \frac{1}{k _{B} T}$ , where $k_{B}$ is Boltzmann's constant (the tiny number that turns temperature into energy). Big $β$ = cold; small $β$ = hot.
Chemical potential $μ$ (mu) — the reservoir's eagerness to hand out particles. High $μ$ = "take more particles"; low $μ$ = "keep them."

WHY. We can't easily count with a fixed particle number — the arithmetic tangles. Instead we let $n$ float and let the reservoir decide the average. This setup is the Grand Canonical Ensemble, and it is the natural language for identical particles because it counts by occupation of levels, not by tracking who-is-who.

PICTURE. The reservoir is the big cyan box; the amber rung trades particles with it across the dashed boundary.

Step 3 — The weight of a configuration (the Boltzmann–Gibbs factor)

WHAT. Statistical mechanics says: a whole-system configuration is not equally likely — its weight depends on two totals for that configuration:

$E$ ::: the total energy of the whole system in that configuration (all particles added up).
$N$ ::: the total number of particles in the whole system in that configuration.

The Gibbs weight is $w = e^{- β (E - μ N)} .$ For our single rung holding $n$ particles, the "whole system" is just that rung, so its total energy is $E = ε n$ (each of the $n$ particles carries energy $ε$ ) and its total particle count is $N = n$ . Substituting, $w (n) = e^{- β (ε - μ) n} .$

WHY. Read the exponent term by term:

$w (n) = exp [coldness - β (energy cost per particle ε - reward per particle μ) how many n] .$

Adding a particle costs energy $ε$ but the reservoir pays back $μ$ . The net price is $(ε - μ)$ .
Multiply by $n$ because $n$ particles each pay that price.
The minus and the $β$ turn "expensive" into "unlikely": a high price and a cold bath ( $β$ large) both crush the weight toward zero.

We use the exponential here (not, say, a straight line) because independent costs must multiply: if two systems are independent, the weight of the pair is the product of the weights, while their energies simply add. So the weight function $f$ must satisfy $f (E_{1}) f (E_{2}) = f (E_{1} + E_{2})$ . This is a famous functional equation, and (for a continuous, non-trivial $f$ ) its only solutions are exponentials $f (E) = e^{c E}$ — that is exactly why $e$ , and nothing else, appears here.

PICTURE. The weight collapses as the net price $(ε - μ)$ or the number $n$ grows.

Step 4 — Fermions: the shelf holds 0 or 1, and that's it

WHAT. The Pauli Exclusion Principle says at most one fermion per state. So $n$ can only be $0$ or $1$ . Add up the weights of the only two allowed configurations: $Z = = 1 w (0) + = e^{- x} w (1) = 1 + e^{- x} .$ This total $Z$ is the grand partition function — the sum of weights, our normaliser.

WHY. Average = (value × its weight, summed) ÷ (total weight): $⟨ n ⟩ = \frac{0 \cdot 1 + 1 \cdot e ^{- x}}{1 + e ^{- x}} = \frac{e ^{- x}}{1 + e ^{- x}} .$ Multiply top and bottom by $e^{x}$ (a legal trick: multiplying by $1 = e^{x} / e^{x}$ ) to make it clean: $⟨ n ⟩_{F D} = \frac{1}{e ^{x} + 1} = \frac{1}{e ^{β (ε - μ)} + 1} .$

The $+ 1$ in the denominator is the fingerprint of "the sum stopped at $n = 1$ ." Because the numerator is $1$ and the denominator is bigger than $1$ , we get $⟨ n ⟩ \leq 1$ always — you literally cannot average more than one fermion onto one rung.

PICTURE. Two configurations, two weights, one weighted average — drawn as a see-saw.

Step 5 — Bosons: the shelf holds any number, so we sum forever

WHAT. Bosons have no cap: $n = 0, 1, 2, 3, \dots$ The total weight is an infinite geometric series: $Z = \sum_{n = 0}^{\infty} e^{- x n} = 1 + e^{- x} + e^{- 2 x} + \dots = \frac{1}{1 - e ^{- x}} .$

WHY. Why does an infinite sum give a tidy fraction? Because each term is the previous one times the same shrink-factor $e^{- x}$ (this ratio is $< 1$ only when $e^{- x} < 1$ , i.e. $x > 0$ , i.e. $ε > μ$ — hold that thought for Step 6). A geometric series with ratio $r$ sums to $\frac{1}{1 - r}$ . We use this tool because the boson tower of occupations is exactly a constant-ratio ladder.

For the average, there is a slick shortcut. Notice $⟨ n ⟩ = \frac{\sum n e ^{- x n}}{\sum e ^{- x n}} = - \frac{\partial}{\partial x} ln Z .$

Why a derivative? Because $\frac{\partial}{\partial x} e^{- x n} = - n e^{- x n}$ — differentiating pulls a factor of $- n$ out of every term, which is exactly the $n$ we need to form the average. The derivative is the tool that "counts" for us. Applying it: $⟨ n ⟩ = - \frac{\partial}{\partial x} ln \frac{1}{1 - e ^{- x}} = - \frac{\partial}{\partial x} [- ln (1 - e^{- x})] = \frac{e ^{- x}}{1 - e ^{- x}} = \frac{1}{e ^{x} - 1} .$

The $- 1$ is the fingerprint of "the sum ran to infinity."

PICTURE. The boson ladder of occupations and the shrinking weights that sum to a finite total.

Step 6 — The edge cases: where the formulas break or blow up

WHAT. A formula you don't stress-test is a formula you don't understand. Set $x = β (ε - μ)$ and push it to the extremes.

Case A — Big $x$ (cold, or $ε ≫ μ$ ): the $\pm 1$ vanishes. When $e^{x} ≫ 1$ , the $+ 1$ or $- 1$ is a rounding error: $⟨ n ⟩ \to \frac{1}{e ^{x}} = e^{- β (ε - μ)} .$ This is the classical Maxwell–Boltzmann Distribution result. Fermi, Bose, and classical all agree — the rung is so rarely occupied that "one per slot vs many per slot" never comes up.

Case B — Fermion at $x = 0$ (level exactly at $μ$ ). $⟨ n ⟩_{F D} = \frac{1}{1 + 1} = \frac{1}{2}$ . Exactly half-filled. This defines the Fermi level: the energy the average occupation crosses one-half.

Case C — Boson as $x \to 0^{+}$ (level approaches $μ$ ): explosion. $⟨ n ⟩_{B E} = \frac{1}{e ^{x} - 1} \to \infty$ . The denominator vanishes, occupation runs away — macroscopically many bosons dump into one state. That is Bose–Einstein Condensation.

Case D — Why $μ < ε_{m i n}$ for bosons. If $x < 0$ (meaning $μ > ε$ ), then $e^{- x} > 1$ and the geometric series of Step 5 diverges — the sum is nonsense, and $⟨ n ⟩$ would go negative, which is physically impossible (you can't have $- 2$ particles). So the boson chemical potential is forbidden from exceeding the lowest energy level. Fermions have no such bound; their series is just two terms and always finite.

PICTURE. All three curves on one axis — watch the boson curve stab upward at $ε = μ$ , the fermion curve step down through $\frac{1}{2}$ , and both hug the classical curve on the far right.

Step 7 — Why bosons "bunch" and fermions "avoid": the counting picture

WHAT. Recall the toy count from the parent note: 2 particles into 3 boxes.

Distinguishable: $3^{2} = 9$ labeled arrangements.
Bosons (unordered, repeats allowed): $(2 3 + 2 - 1) = 6$ .
Fermions (unordered, no repeats): $(2 3) = 3$ .

WHY. The fraction of configurations that are "both particles in the same box" reveals the personality:

Classical: $3$ same-box out of $9 = 33.3%$ .
Bosons: $3$ same-box out of $6 = 50%$ — they over-represent togetherness → statistical attraction (no real force!).
Fermions: $0$ out of $3 = 0%$ — togetherness is banned → statistical avoidance.

This is the same story the occupation numbers tell in the smooth limit, now visible in raw counting. See Gibbs Paradox & Entropy of Mixing for the classical $1/ N!$ patch that partly heals the gap.

PICTURE. The three columns of allowed arrangements side by side; the highlighted same-box states show the bunching bias at a glance.

The one-picture summary

Everything above compresses into a single master figure: one rung, one net price $(ε - μ)$ , one exponential weight, and one choice — cap the sum at 1 (fermion, gives $+ 1$ ) or sum to infinity (boson, gives $- 1$ ) or ignore the tail (classical, gives $0$ ). The unifier: $⟨ n ⟩ = \frac{1}{e ^{β (ε - μ)} + a}, a = ⎩ ⎨ ⎧ + 1 - 1 0 Fermi-Dirac Bose-Einstein Maxwell-Boltzmann$

Recall Feynman retelling — the whole walkthrough in plain words

We stared at a single shelf-rung and asked "how many particles usually stand here?" We hooked it to a giant bath that lends particles for a fee: the rung costs $ε$ to occupy but the bath refunds $μ$ , so the real price is $(ε - μ)$ . Cold baths and high prices make crowding unlikely — we measured that unlikeliness with $e^{- price \times n}$ , because independent costs must multiply and exponentials turn multiplying into adding. Then we let the particles' personality decide how far to sum. Grumpy fermions allow only 0 or 1, so we added two numbers and got a $+ 1$ downstairs — never more than one per rung. Friendly bosons allow any pile, so we summed forever, a geometric series that gave a $- 1$ downstairs — and that $- 1$ can hit zero, letting an infinite crowd pour into one state (that's condensation). When the rung is almost always empty, the $\pm 1$ stops mattering and everyone agrees on the plain classical $e^{- price}$ . One rung, one exponential, one choice of how far to sum — that's the entire quantum-statistics machine.

Recall Self-check

Why does the fermion formula have $+ 1$ and the boson formula $- 1$ ? ::: The $+ 1$ comes from a two-term sum $1 + e^{- x}$ stopping at $n = 1$ (Pauli cap); the $- 1$ comes from the infinite geometric series $\frac{1}{1 - e ^{- x}}$ summing all occupations. What must be true of $μ$ for bosons and why? ::: $μ$ must be below the lowest energy level ( $x > 0$ ), otherwise the geometric series diverges and $⟨ n ⟩$ goes negative — unphysical. At what energy is a fermion level exactly half-filled? ::: At $ε = μ$ (the Fermi level), where $⟨ n ⟩_{F D} = \frac{1}{2}$ . Where do all three statistics coincide? ::: In the dilute/hot limit $e^{β (ε - μ)} ≫ 1$ , giving $⟨ n ⟩ = e^{- β (ε - μ)}$ .