Visual walkthrough — Bose-Einstein condensation — concept

We will only assume you know:

• Boson = an identical particle that is happy to share a state with any number of copies of itself (unlike its shy cousin, the fermion).
• Energy of a free particle grows with how fast its matter-wave wiggles.

Everything else — every symbol — we earn as we go.

---

Step 1 — The rule that lets bosons pile up

Let us name every piece right where it lives:

• $\bar{n}$ — the answer: how crowded this one state is (the bar means "average").
• $\epsilon$ — the energy of the state we are asking about.
• $\mu$ (mu) — the chemical potential: a dial that sets the overall crowd size. Bigger $\mu$ → every state gets fuller. Meet it properly in Chemical potential.
• $k_B$ — Boltzmann's constant, the fixed conversion "energy per degree of temperature".
• $T$ — temperature. $k_{B}T$ together is "the thermal energy budget".
• The $-1$ in the denominator is the whole personality of a boson — it comes from Bose-Einstein statistics and it is what lets $\bar{n}$ blow up to huge values.

WHY this formula and not something simpler? Because we need a rule that (a) gives some number for how full a state is, and (b) allows that number to become enormous when $\epsilon -\mu$ gets tiny. The $-1$ does exactly that: when the denominator approaches zero, $\bar{n}$ shoots to infinity. A classical gas uses $e^{-(\epsilon -\mu )/k_{B}T}$ (no $-1$), which never blows up — that is precisely why classical gases don't condense.

PICTURE. The blue curve is $\bar{n}$ versus energy. Notice how it rockets upward as $\epsilon$ drops toward $\mu$ — the low states are where all the action is.

---

Step 2 — Why $\mu$ can never climb above zero

Set the ground state at energy ${\epsilon }_0=0$. Its occupation is $N_0=\frac{1}{e^{-\mu /k_{B}T}-1}.$

• $N_0$ — number of atoms in the lowest state.
• $-\mu /k_{B}T$ — the exponent; for this to give a positive $N_0$ we need $e^{-\mu /k_{B}T}>1$, i.e. $-\mu >0$, i.e. $\mu <0$.

WHAT we just learned: $\mu$ is trapped at or below $0$. A negative-population state is nonsense, so nature forbids $\mu >0$.

WHY it matters: as $\mu$ creeps up toward $0^{-}$ (from below), the denominator $e^{-\mu /k_{B}T}-1$ approaches $0$, so $N_0\to \infty$. This is the escape valve: the ground state has unlimited capacity as $\mu$ nears its ceiling.

PICTURE. Watch $N_0$ (orange) rocket to infinity as $\mu$ slides toward $0$. The red wall at $\mu =0$ is the forbidden zone.

---

Step 3 — Counting the excited states (the density of states)

We want $N_{\text{exc}}$, the total number of atoms in all the excited states ($\epsilon >0$). There are astronomically many levels, so we don't sum them one by one — we ask "how many states lie in each little energy slice?" That count is the density of states $g(\epsilon )$ (see Density of states):

$g(\epsilon )d\epsilon =\frac{gV}{4{\pi }^2}{\left(\frac{2m}{{\hslash }^2}\right)}^{3/2}\sqrt{\epsilon }d\epsilon .$

• $g(\epsilon )d\epsilon$ — number of states with energy between $\epsilon$ and $\epsilon +d\epsilon$.
• $g$ (front) — spin degeneracy (how many spin flavours per level); don't confuse with the function $g(\epsilon )$.
• $V$ — box volume; more room, more states.
• $m$ — atom mass; $\hslash$ — reduced Planck constant.
• $\sqrt{\epsilon }$ — the crucial shape: the number of states grows with energy, and importantly it vanishes at $\epsilon =0$.

WHY the $\sqrt{\epsilon }$ is the villain-and-hero of this story: at $\epsilon =0$ there is zero density of states. So when we integrate over excited energies, the integral cannot see the ground state at all — it silently drops it. That is exactly why we had to track $N_0$ by hand in Step 2.

PICTURE. The green curve $\sqrt{\epsilon }$ starts at the origin. The ground state (red dot at $\epsilon =0$) sits outside the shaded integration region — it is invisible to the integral.

---

Step 4 — The excited states have a maximum capacity

Multiply "how many states are in a slice" ($g(\epsilon )d\epsilon$) by "how full each is" ($\bar{n}$) and add up all slices. Take the most generous case $\mu =0$ (as full as the excited states can ever get):

$N_{\text{exc}}^{\max }(T)={\int }_0^{\infty }\frac{g(\epsilon )d\epsilon }{e^{\epsilon /k_{B}T}-1}.$

• The integral ${\int }_0^{\infty }$ — sum over every excited energy slice.
• Numerator $g(\epsilon )d\epsilon$ — states available in that slice.
• Denominator $e^{\epsilon /k_{B}T}-1$ — this is $\bar{n}$ with $\mu =0$ plugged in.

WHY $\mu =0$? Because we want the ceiling. Raising $\mu$ raises occupancy; $\mu$ can go no higher than $0$; so $\mu =0$ gives the biggest the excited crowd will ever be. If the atoms we own exceed this ceiling, the extras have literally nowhere to sit but the ground state.

PICTURE. The area under the blue curve $g(\epsilon )\bar{n}(\epsilon )$ is $N_{\text{exc}}^{\max }$ — a finite shaded region. Finite area = finite bucket.

---

Step 5 — Squeeze out a pure number (the substitution)

The integral looks messy, but a change of variable cleans it up. Let $x=\epsilon /k_{B}T$ (dimensionless energy, measured in units of the thermal budget). Why? Because it pulls all the temperature dependence out to the front, leaving behind a naked number that no longer cares about $T$, $m$, or $V$:

$N_{\text{exc}}^{\max }=\frac{gV}{4{\pi }^2}{\left(\frac{2m{k}_{B}T}{{\hslash }^2}\right)}^{3/2}\underset{=\Gamma (3/2)\zeta (3/2)}{\underbrace{{\int }_0^{\infty }\frac{\sqrt{x}dx}{e^x-1}}}.$

• The bracket $(2m{k}_{B}T/{\hslash }^2{)}^{3/2}$ — the entire $T$-dependence, growing as $T^{3/2}$.
• $\Gamma (3/2)=\frac{\sqrt{\pi }}{2}$ — a standard integral factor (the gamma function).
• $\zeta (3/2)\approx 2.612$ — the Riemann zeta value; a fixed constant of the universe for this problem.

Using the Thermal de Broglie wavelength ${\lambda }_T=\frac{h}{\sqrt{2\pi m{k}_{B}T}}$ (the "size" of an atom's matter-wave at temperature $T$), the whole thing collapses to a jewel:

• $n_{\text{exc}}^{\max }$ — max excited atoms per unit volume.
• ${\lambda }_T^3$ — a "thermal volume", the room one matter-wave occupies.

WHAT it says: each thermal volume ${\lambda }_T^3$ can hold about $g\times 2.612$ excited atoms. That's the bucket size — and it shrinks as you cool, because ${\lambda }_T$ grows.

PICTURE. As $T$ falls, the thermal wavelength ${\lambda }_T$ (blue) grows, so the bucket per volume shrinks. Cross-section shows matter-waves swelling until they touch.

---

Step 6 — The overflow: transition temperature

The gas has a real density $n$ (atoms per volume) that you set. The bucket $n_{\text{exc}}^{\max }$ shrinks as you cool. The magic moment is when the bucket exactly equals your atoms: $n=g\frac{\zeta (3/2)}{{\lambda }_{T_c}^3}.$ Solve for the temperature at that instant — the critical temperature $T_c$:

• Higher density $n$ → higher $T_c$ (buckets fill sooner).
• Heavier mass $m$ → lower $T_c$ (heavy atoms have short matter-waves, hard to overlap).

WHY this is the condition: rewrite it as a single elegant inequality. BEC begins when $n{\lambda }_T^3=g\zeta (3/2)\approx 2.612,$ the phase-space density crossing $\sim 1$. Plain words: condensation starts the moment neighbouring atoms' matter-waves overlap — spacing $\approx$ wavelength. This is a genuine phase transition, with the condensate showing Superfluidity.

PICTURE. Two shrinking curves: your fixed atoms $n$ (orange line) and the falling bucket $n_{\text{exc}}^{\max }(T)$ (blue). They cross at $T_c$. Left of the crossing, the bucket can't hold everyone — overflow.

---

Step 7 — Where the overflow goes: the condensate fraction

Below $T_c$, $\mu$ pins itself at $\approx 0$, so the excited count follows its own $T^{3/2}$ shape: $\frac{N_{\text{exc}}(T)}{N}={\left(\frac{T}{T_c}\right)}^{3/2}.$ Everyone not in an excited state is in the ground state:

• $N_0/N$ — fraction of all atoms sitting in the single lowest state.
• At $T=T_c$: fraction $=0$ (nobody condensed yet).
• At $T=0$: fraction $=1$ (everyone condensed).
• Just below $T_c$ the curve leaves with infinite slope — the condensate switches on abruptly, the fingerprint of a phase transition.

PICTURE. The classic order-parameter curve: $N_0/N$ rising from $0$ at $T_c$ to $1$ at $T=0$, steep at the start.

---

Step 8 — The degenerate cases (never let the reader get stuck)

PICTURE. Side-by-side: 3D integrand ($\sqrt{x}$-weighted, finite area) versus 2D integrand (flat-weighted, area blows up near $0$). The 2D bucket has no bottom.

---

Worked examples (each answer machine-checked below)

---

The one-picture summary

This single frame stitches the whole logic: a finite excited-state bucket that shrinks as you cool, your fixed pile of atoms, the crossing at $T_c$, and the overflow pouring into the ground state to build the condensate.

Recall Feynman retelling — the whole walkthrough in plain words

Imagine each atom's matter-wave as a fuzzy blob whose size is ${\lambda }_T$. Warm gas → tiny blobs, far apart, plenty of "excited" seats to spread across. Now cool it. Each blob swells (${\lambda }_T$ grows as $T^{-1/2}$). We counted exactly how many atoms the excited seats can hold: about $2.612$ per blob-sized volume — a finite bucket whose size drops as blobs grow. Meanwhile you keep your atom count fixed. There comes a temperature $T_c$ where the shrinking bucket exactly matches your atoms — that's when the blobs start touching. Cool further and the bucket can't hold everyone; the leftovers have only one place to go, the single lowest state, whose capacity is infinite because its occupation diverges as $\mu$ kisses zero from below. Those leftovers are the condensate, and the fraction that has fallen in is $1-(T/T_c{)}^{3/2}$ — zero at $T_c$, everyone at absolute zero. No forces, no attraction — just identical super-social particles and a bucket that runs out of room.

Recall Predict then check

Q: If you double the density $n$ at fixed everything else, does $T_c$ go up or down, and by what factor? A: $T_c\propto n^{2/3}$, so doubling $n$ multiplies $T_c$ by $2^{2/3}\approx 1.587$. Denser gas condenses at a higher (easier) temperature because buckets fill sooner.