Bose-Einstein condensation and the critical temperature

Intuition Beginner

A Bose-Einstein condensate is what happens when you cool a gas of bosons until a macroscopic fraction of the particles all pile into the same quantum state — the single lowest-energy state of the system. They stop behaving like a crowd of independent atoms and start behaving like one big quantum object, a single coherent matter wave.

Why does this happen for bosons and not for other particles? Bosons are the family of particles (photons, helium-4 atoms, rubidium-87 atoms) that have no restriction on how many can share a state. Fermions (electrons, protons, neutrons) obey the Pauli exclusion principle: at most one fermion per state. Bosons are the opposite extreme — they actively prefer to bunch up. At very low temperature, that preference wins out, and the ground state fills up macroscopically.

The temperature where this transition turns on is called the critical temperature $T_c$. Above $T_c$ the bosons are spread across many states, jostled by thermal energy. Below $T_c$ a finite fraction collapses into the single ground state. The fraction that has condensed grows as the temperature drops, reaching essentially 100 percent near absolute zero.

A useful picture is the thermal de Broglie wavelength ${\lambda }_T=h/\sqrt{2\pi m{k}_{B}T}$. At high temperature this wavelength is much smaller than the spacing between atoms, and quantum statistics is irrelevant. As $T$ drops, ${\lambda }_T$ grows. When ${\lambda }_T$ becomes comparable to the inter-atom spacing, the wave packets of neighbouring atoms start to overlap, and the gas becomes quantum-degenerate. For an ideal Bose gas in three dimensions, condensation begins when $n{\lambda }_T^3\approx 2.6$, where $n$ is the number density.

The first three Bose-Einstein condensates in dilute atomic gases were produced in 1995. A group at JILA (Cornell, Wieman, and collaborators) cooled rubidium-87 to about 170 nanokelvin and observed the characteristic narrow peak emerge in the velocity distribution. Months later, groups at Rice (Hulet, with lithium-7) and MIT (Ketterle, with sodium-23) followed. Cornell, Wieman, and Ketterle shared the 2001 Nobel Prize. Superfluid helium-4 below 2.17 K, discovered in 1938 by Kapitsa and by Allen and Misener, is a strongly interacting cousin — a Bose-Einstein condensate dressed in heavy interatomic forces.

Visual Beginner

Three momentum-distribution panels arranged left to right, each showing the velocity profile of a trapped gas of rubidium-87 atoms at a different temperature. At the highest temperature (around 400 nanokelvin) the distribution is broad and Gaussian — a thermal cloud. At intermediate temperature (around 200 nanokelvin) a sharp narrow peak emerges at zero velocity on top of the broad pedestal — the first signature of the condensate. At the lowest temperature (around 50 nanokelvin) the broad thermal background has nearly vanished, leaving an almost pure condensate spike.

A second panel shows the condensate fraction $N_0/N$ plotted against $T/T_c$. Above $T=T_c$ the fraction is essentially zero. Below $T_c$ it rises smoothly, following the prediction $N_0/N=1-(T/T_c{)}^{3/2}$ for an ideal three-dimensional Bose gas. At $T/T_c=1/2$ about 65 percent of the atoms have condensed; at $T/T_c=1/4$ about 88 percent have.

Worked example Beginner

Estimate the critical temperature for Bose-Einstein condensation of rubidium-87 atoms in a typical magnetic trap at density $n=1.0\times 10^{20}{\text{m}}^{-3}$. Use the formula $$ k_B T_c ;=; \frac{2\pi \hbar^2}{m} \left(\frac{n}{\zeta(3/2)}\right)^{2/3}, $$ with $\zeta (3/2)=2.612$ and the rubidium-87 mass $m=1.443\times 10^{-25}\text{kg}$.

Step 1. Compute $n/\zeta (3/2)=1.0\times 10^{20}/2.612=3.83\times 10^{19}{\text{m}}^{-3}$.

Step 2. Raise to the $2/3$ power: $(3.83\times 10^{19}{)}^{2/3}=(3.83{)}^{2/3}\times 10^{19\cdot 2/3}=2.46\times 10^{12.67}\approx 1.15\times 10^{13}{\text{m}}^{-2}$.

Step 3. Multiply by $2\pi {\hslash }^2/m$: $$ \frac{2\pi \hbar^2}{m} ;=; \frac{2\pi \times (1.055 \times 10^{-34})^2}{1.443 \times 10^{-25}} ;\approx; 4.85 \times 10^{-43},\text{J}\cdot\text{m}^2. $$ So $k_B{T}_c\approx 4.85\times 10^{-43}\times 1.15\times 10^{13}=5.58\times 10^{-30}\text{J}$.

Step 4. Divide by $k_B=1.381\times 10^{-23}\text{J/K}$: $$ T_c ;\approx; \frac{5.58 \times 10^{-30}}{1.381 \times 10^{-23}} ;\approx; 4.0 \times 10^{-7},\text{K} ;=; 400,\text{nK}. $$

What this tells us: the critical temperature for a typical rubidium magnetic trap is a few hundred nanokelvin — about ten million times colder than your refrigerator, and five billion times colder than the cosmic microwave background at 2.725 K. Reaching this regime in the laboratory required the development of laser cooling and evaporative cooling, both of which won later Nobel Prizes (1997 for laser cooling, 2001 for the condensation itself).

Check your understanding Beginner

Formal definition Intermediate+

Consider an ideal gas of $N$ identical non-interacting bosons of mass $m$ in a box of volume $V$ at temperature $T$, with periodic boundary conditions. The single-particle energy levels are ${\epsilon }_{\mathbf{p}}=p^2/(2m)$ for $\mathbf{p}=(2\pi \hslash /L)(n_1,n_2,n_3)$, $n_i\in \mathbb{Z}$, with ${\epsilon }_0=0$ the ground-state energy.

Definition (grand-canonical occupation). The average occupation of the single-particle state with energy ${\epsilon }_{\mathbf{p}}$ in the grand canonical ensemble at temperature $T$ and chemical potential $\mu$ is $$ \langle n_\mathbf{p}\rangle ;=; \frac{1}{e^{(\varepsilon_\mathbf{p} - \mu)/k_B T} - 1}. $$ For a non-negative occupation in every state we must have $\mu <{\epsilon }_0=0$ for free particles.

The total particle number is $$ N ;=; \sum_\mathbf{p} \langle n_\mathbf{p}\rangle ;=; \langle n_0\rangle + \sum_{\mathbf{p} \neq 0} \langle n_\mathbf{p}\rangle. $$ In the thermodynamic limit $V\to \infty$ with $n=N/V$ fixed, the non-zero-momentum sum becomes an integral over the single-particle density of states $g(\epsilon )=(V/4{\pi }^2)(2m/{\hslash }^2{)}^{3/2}\sqrt{\epsilon }$ in three dimensions. Writing $z=e^{\mu /k_{B}T}$ for the fugacity and ${\lambda }_T=h/\sqrt{2\pi m{k}_{B}T}$ for the thermal de Broglie wavelength, the thermal-excitation number reads $$ N_T ;=; \frac{V}{\lambda_T^3}, g_{3/2}(z), $$ where $$ g_s(z) ;=; \sum_{\ell=1}^\infty \frac{z^\ell}{\ell^s} ;=; \frac{1}{\Gamma(s)} \int_0^\infty \frac{x^{s-1}}{z^{-1} e^x - 1}, dx $$ is the polylogarithm of order $s$. The function $g_{3/2}(z)$ is monotonically increasing on $0<z\le 1$, with maximum $g_{3/2}(1)=\zeta (3/2)\approx 2.612$.

Definition (critical temperature). The critical temperature $T_c$ for Bose-Einstein condensation at fixed density $n$ is the temperature at which $\mu =0$ (equivalently $z=1$) saturates the thermal excitation: $$ n ;=; \frac{\zeta(3/2)}{\lambda_{T_c}^3}, \qquad k_B T_c ;=; \frac{2\pi \hbar^2}{m}\left(\frac{n}{\zeta(3/2)}\right)^{2/3}. $$ Equivalently, the BEC criterion is the phase-space-density condition $n{\lambda }_T^3\ge \zeta (3/2)\approx 2.612$: condensation begins when the thermal wavefunctions of neighbouring particles overlap densely enough that one cannot continue to absorb additional thermal excitations.

Below $T_c$, the thermal-excitation pool can hold only $N_T(T)=(V/{\lambda }_T^3)\zeta (3/2)$ particles. The remainder $$ N_0 ;=; N - N_T(T) ;=; N\left[1 - \left(\frac{T}{T_c}\right)^{3/2}\right] $$ must therefore occupy the ground state $\mathbf{p}=0$. This is the condensate fraction $N_0/N=1-(T/T_c{)}^{3/2}$, a hallmark of three-dimensional ideal-gas BEC.

Counterexamples to common slips

• Ignoring the ground state. Replacing the full sum over $\mathbf{p}$ by an integral over the density of states sets $⟨n_0⟩=0$ because $g(\epsilon )\sqrt{\epsilon }\to 0$ as $\epsilon \to 0$. The ground-state occupation $⟨n_0⟩=1/(e^{-\mu /k_{B}T}-1)$ is finite for $\mu <0$ but diverges as $\mu \to 0^{-}$, so the integral approximation hides macroscopic occupation of the ground state. The thermodynamic-limit decomposition $N=N_0+N_T$ must be carried out by hand.
• Confusing $T_c$ with the spacing-set temperature. The naive temperature scale where $k_{B}T$ equals the level spacing ${\hslash }^2/(m{L}^2)$ is much smaller than $T_c$. Condensation is a thermodynamic phenomenon driven by the density of states near $\epsilon =0$ and the inability of the polylogarithm $g_{3/2}$ to exceed $\zeta (3/2)$ — not by individual level spacings.
• Forgetting the dimensional restriction. In two dimensions the analogue integral involves $g_1(z)=-\ln (1-z)$, which diverges as $z\to 1$. There is no finite saturation, and ideal BEC in 2D at $T>0$ does not happen. In one dimension the divergence is even stronger: $g_{1/2}(z)\to \infty$ as $z\to 1$. Bose-Einstein condensation of an ideal gas in a uniform box requires $d\ge 3$.

Key theorem with proof Intermediate+

Theorem (Einstein 1924–25 critical temperature). Let an ideal non-interacting gas of $N$ identical bosons of mass $m$ occupy a volume $V$ in three dimensions at temperature $T$ with periodic boundary conditions. In the thermodynamic limit $V,N\to \infty$ with density $n=N/V$ fixed, there exists a critical temperature $$ k_B T_c ;=; \frac{2\pi \hbar^2}{m}\left(\frac{n}{\zeta(3/2)}\right)^{2/3} $$ below which a macroscopic fraction $$ \frac{N_0}{N} ;=; 1 - \left(\frac{T}{T_c}\right)^{3/2} $$ occupies the single-particle ground state. Above $T_c$, the chemical potential is strictly negative ($\mu <0$); at $T=T_c$, $\mu =0$; below $T_c$, $\mu$ remains pinned at zero while the condensate fraction grows.

Proof. The proof is in four steps: density of states, saturation, condensate identification, and continuity of $\mu$.

Step 1 (density of states). In three dimensions with periodic boundary conditions on a box of side $L=V^{1/3}$, the allowed wave vectors are $\mathbf{k}=(2\pi /L)(n_1,n_2,n_3)$ for $n_i\in \mathbb{Z}$. The number of states with momentum magnitude between $p$ and $p+dp$ is the volume of a spherical shell in momentum space divided by the volume per state $(2\pi \hslash /L{)}^3$: $$ dN_{\text{states}} ;=; \frac{4\pi p^2 , dp}{(2\pi\hbar/L)^3} ;=; \frac{V p^2}{2\pi^2 \hbar^3}, dp. $$ Substituting $\epsilon =p^2/(2m)$, $d\epsilon =pdp/m$, and $p=\sqrt{2m\epsilon }$: $$ g(\varepsilon), d\varepsilon ;=; \frac{V (2m\varepsilon)^{1/2} \cdot m}{2\pi^2 \hbar^3}, d\varepsilon ;=; \frac{V}{4\pi^2}\left(\frac{2m}{\hbar^2}\right)^{3/2} \sqrt{\varepsilon}, d\varepsilon. $$

Step 2 (saturation of the thermal pool). The expected number of bosons in the non-condensate states is $$ N_T(T, \mu) ;=; \int_0^\infty \frac{g(\varepsilon), d\varepsilon}{e^{(\varepsilon - \mu)/k_B T} - 1}. $$ Substitute $x=\epsilon /k_{B}T$ and $z=e^{\mu /k_{B}T}$: $$ N_T ;=; \frac{V}{4\pi^2}\left(\frac{2m k_B T}{\hbar^2}\right)^{3/2} \int_0^\infty \frac{\sqrt{x}, dx}{z^{-1} e^x - 1}. $$ The integral evaluates to $\Gamma (3/2)g_{3/2}(z)=(\sqrt{\pi }/2)g_{3/2}(z)$, where $g_s(z)={\sum }_{\ell \ge 1}{z}^{\ell }/{\ell }^s$. Tidying up with ${\lambda }_T=h/\sqrt{2\pi m{k}_{B}T}$: $$ N_T ;=; \frac{V}{\lambda_T^3}, g_{3/2}(z). $$ Because $g_{3/2}(z)$ is monotonically increasing on $0<z\le 1$ with maximum $g_{3/2}(1)=\zeta (3/2)$, the saturation value is $N_T^{\max }(T)=(V/{\lambda }_T^3)\zeta (3/2)$, attained at $\mu =0$ (i.e., $z=1$).

Step 3 (identification of the condensate). For fixed density $n=N/V$ define $T_c$ by $n=\zeta (3/2)/{\lambda }_{T_c}^3$: $$ k_B T_c ;=; \frac{2\pi \hbar^2}{m}\left(\frac{n}{\zeta(3/2)}\right)^{2/3}. $$ For $T>T_c$ the saturation $N_T^{\max }(T)>N$, so $z<1$ is fixed by $N=N_T(T,z)=(V/{\lambda }_T^3)g_{3/2}(z)$ — every particle lives in the thermal pool. For $T<T_c$ the saturation $N_T^{\max }(T)<N$. The chemical potential cannot move beyond $\mu =0$ without producing negative occupations, so it stays pinned at $\mu =0$ and the excess $$ N_0(T) ;=; N - N_T^{\max}(T) ;=; N\left[1 - \frac{V}{N\lambda_T^3}\zeta(3/2)\right] ;=; N\left[1 - \left(\frac{T}{T_c}\right)^{3/2}\right] $$ condenses into the ground state $\mathbf{p}=0$. The last equality uses $V/{\lambda }_T^3=V/{\lambda }_{T_c}^3\cdot (T/T_c{)}^{3/2}$ and $V\zeta (3/2)/{\lambda }_{T_c}^3=N$.

Step 4 (continuity). At $T=T_c$ both formulas give $N_0=0$, so the condensate fraction is continuous across the transition. The chemical potential is continuous (both branches give $\mu =0$ at $T_c$) but its derivative $d\mu /dT$ is not — $\mu$ approaches zero from below for $T>T_c$ and stays at zero for $T<T_c$. The specific heat $C_V$ is continuous at $T_c$ with a discontinuous derivative, classifying the ideal-gas BEC as a third-order phase transition in the Ehrenfest scheme. $\square$

Bridge. The critical-temperature formula builds toward the Pitaevskii-Stringari treatment of harmonic-trap BEC, which appears again in cold-atom experimental analysis. The foundational reason that ideal-gas BEC happens is the saturation of the polylogarithm $g_{3/2}(z)$ at $z=1$, and this is exactly the missing ingredient that classical Maxwell-Boltzmann statistics lacks — the Boltzmann factor $e^{-(\epsilon -\mu )/k_{B}T}$ has no maximum-occupation cutoff. Putting these together with the photon-gas analysis in 11.05.03, the structure that identifies the two cases is the same Bose-Einstein occupation number; for photons the chemical potential is exactly zero because photon number is not conserved, while for a massive Bose gas $\mu$ slides up to zero as $T$ drops and gets stuck there — the photon spectrum is the BEC of conserved-charge zero at the bottom of every temperature. The pattern generalises to relativistic bosons (where ${\epsilon }_{\mathbf{p}}=\sqrt{(pc{)}^2+(m{c}^2{)}^2}$) and to confined geometries; the bridge between three-dimensional BEC and the lower-dimensional cases via the polylogarithm convergence is the load-bearing piece of the proof.

Exercises Intermediate+

Advanced results Master

Theorem 1 (Einstein 1924–25 — ideal-gas critical temperature). For an ideal gas of $N$ non-interacting bosons of mass $m$ in a box of volume $V$ in three dimensions, Bose-Einstein condensation occurs below the critical temperature $$ k_B T_c ;=; \frac{2\pi \hbar^2}{m}\left(\frac{n}{\zeta(3/2)}\right)^{2/3}, $$ and the condensate fraction is $N_0/N=1-(T/T_c{)}^{3/2}$.

Einstein 1924–1925 published two papers extending Bose's 1924 photon-statistics derivation to massive particles. The key insight is that the chemical potential cannot exceed the ground-state energy ($\mu \le {\epsilon }_0=0$ for free bosons), so the thermal-excitation pool $N_T(T)$ has a maximum value $(V/{\lambda }_T^3)\zeta (3/2)$ that depends only on $T$. When $N$ exceeds this maximum, the excess particles condense into the ground state. Einstein's 1925 paper anticipated the discovery 70 years before experimental confirmation; the dilute-gas BEC programme of the 1990s was a direct consequence of this argument.

Theorem 2 (London 1938 — the lambda transition). The lambda transition in liquid helium-4 at $T_{\lambda }=2.17\text{K}$ is Bose-Einstein condensation of an interacting Bose system, with the ideal-gas formula predicting $T_c\approx 3.13\text{K}$ — within 30 percent of the measured value despite the strong interactions.

London 1938 ^[London1938] proposed the BEC interpretation of superfluid helium as a counter-argument to Tisza's "two-fluid" phenomenological model. The condensate fraction in superfluid helium is only about 7 percent at low temperature (much smaller than the ideal-gas prediction of 100 percent) because strong interactions deplete the ground state into virtual excitations — the condensate depletion computed in the Bogoliubov framework. Despite this, the superfluid fraction reaches 100 percent at $T=0$, with the "missing" condensate atoms locked into the superfluid by the coherence of the order parameter.

Theorem 3 (Bogoliubov 1947 — weakly-interacting spectrum and superfluidity). The elementary excitations of a weakly-interacting dilute Bose gas at zero temperature have dispersion $$ E(\mathbf{p}) ;=; \sqrt{\varepsilon_\mathbf{p}(\varepsilon_\mathbf{p} + 2nU_0)}, \qquad U_0 ;=; \frac{4\pi\hbar^2 a_s}{m}, $$ linear at small $p$ with sound speed $c=\sqrt{n{U}_0/m}$ and quadratic at large $p$. The Landau criterion gives critical superfluid velocity $v_c=c$.

Bogoliubov 1947 ^{[Bogoliubov1947]} introduced the c-number replacement of the condensate-mode operators and the Bogoliubov transformation that diagonalises the resulting bilinear Hamiltonian. The linear low-momentum dispersion is the foundational reason a weakly-interacting BEC is a superfluid: an obstacle moving below $c$ cannot emit excitations and therefore experiences no friction. The dispersion has been confirmed by Bragg-spectroscopy measurements in dilute alkali BECs (Stamper-Kurn et al. 1999, Steinhauer et al. 2002) and inelastic neutron scattering in superfluid helium-4 (with the additional Landau roton feature characteristic of the strongly-interacting case).

Theorem 4 (Gross 1961, Pitaevskii 1961 — Gross-Pitaevskii equation). In the mean-field limit, the condensate wave function $\Psi (\mathbf{r},t)$ of a dilute Bose gas with contact interaction $U_0=4\pi {\hslash }^2{a}_s/m$ and external potential $V_{\text{ext}}(\mathbf{r})$ satisfies $$ i\hbar,\frac{\partial \Psi}{\partial t} ;=; \left(-\frac{\hbar^2 \nabla^2}{2m} + V_{\text{ext}}(\mathbf{r}) + U_0 |\Psi|^2\right) \Psi. $$

The Gross-Pitaevskii equation ^{[Gross1961; Pitaevskii1961]} is a non-linear Schrödinger equation with the cubic non-linearity $U_0|\Psi {|}^2\Psi$ arising from the mean-field interaction. The stationary solution in a harmonic trap is the Thomas-Fermi profile $n(\mathbf{r})=(\mu -V_{\text{ext}})/U_0$ for $\mathbf{r}$ inside the cloud and zero outside, a parabolic density profile sharply different from the Gaussian profile of the non-interacting condensate. The healing length $\xi =\hslash /\sqrt{2mn{U}_0}$ is the scale over which the condensate wave function recovers from a perturbation — the characteristic length scale for vortex cores and condensate boundaries.

Theorem 5 (Onsager 1949, Feynman 1955, Madison-Chevy-Wohlleben-Dalibard 2000 — quantised vortices). Vortices in a superfluid BEC carry quantised circulation $\oint \mathbf{v}\cdot d\mathbf{\ell }=2\pi \hslash k/m$ for integer $k$, with vortex cores of size $\xi$ (the healing length). In a rotating cloud the equilibrium vortex configuration is an Abrikosov triangular lattice.

Onsager 1949 and Feynman 1955 proposed quantised circulation as the explanation for the rotation-bucket experiments in superfluid helium-4 (Vinen 1958, Hall 1960). The 2000 experiments at ENS Paris ^{[MadisonChevyWohllebenDalibard2000]} visualised individual vortices in a rotating sodium BEC and confirmed the triangular Abrikosov lattice predicted in 1957 for type-II superconductors. Higher-circulation vortices ($k\ge 2$) are unstable to decay into multiple unit vortices, and antivortex-vortex pairs annihilate, behaviour central to the modern theory of two-dimensional quantum turbulence.

Theorem 6 (Mermin-Wagner-Hohenberg theorem and BKT transition). Continuous symmetries cannot be spontaneously broken in two dimensions at finite temperature for systems with short-range interactions. Ideal-gas BEC in $d\le 2$ does not exist at $T>0$. The 2D superfluid transition is replaced by the Berezinskii-Kosterlitz-Thouless transition (1971–73) with quasi-long-range power-law correlations and vortex-pair unbinding.

The Mermin-Wagner theorem (Mermin-Wagner 1966 Phys. Rev. Lett. 17, 1133; Hohenberg 1967 Phys. Rev. 158, 383) excludes off-diagonal long-range order in 2D Bose systems. The BKT transition (Berezinskii 1971; Kosterlitz-Thouless 1973 J. Phys. C 6, 1181) at the universal jump in the superfluid density was confirmed in 2D atomic BECs by Hadzibabic et al. 2006 Nature 441, 1118 via matter-wave interference. In a harmonic trap the finite system size restores the appearance of BEC even in two dimensions (Petrov-Holzmann-Shlyapnikov 2000), with quasi-condensate phase coherence developing at $T_c$ but no true symmetry breaking.

Theorem 7 (BEC-BCS crossover at unitarity). A two-component Fermi gas with attractive contact interaction tuned via a Feshbach resonance interpolates smoothly from a BCS superfluid of Cooper pairs (weak attraction, $1/k_F{a}_s\to -\infty$) to a BEC of tightly bound dimers (strong attraction, $1/k_F{a}_s\to +\infty$), passing through the unitary regime ($1/k_F{a}_s=0$) where the only length scale is the inverse Fermi wave number.

The crossover scenario was anticipated by Eagles 1969 (in the context of low-density semiconductors), Leggett 1980, and Nozières-Schmitt-Rink 1985. Experimental observation in cold-atom Fermi gases (Regal-Greiner-Jin 2004, Zwierlein et al. 2004) demonstrated the smooth interpolation by adjusting the magnetic field across a Feshbach resonance. The unitary regime is universal: thermodynamic quantities depend only on $T/T_F$ and $\mu /k_{B}T$, with all microscopic details of the interaction absorbed into the divergent scattering length. The Bertsch parameter $\xi \approx 0.376$ measures the universal energy density at unitarity in units of the non-interacting Fermi gas.

Theorem 8 (Photon BEC — Klaers et al. 2010). Photons confined in a curved-mirror microcavity filled with a dye exhibit Bose-Einstein condensation in two spatial dimensions, with a critical photon number per area set by an effective mass given by the cavity curvature.

Klaers et al. 2010 ^[Klaers2010] reported the first thermalised photon condensate, exploiting the dye-mediated thermalisation that allows photon number to be effectively conserved on relevant timescales. The 2D nature of the cavity geometry escapes Mermin-Wagner because the harmonic confinement of the cavity mirrors breaks translation invariance — the same loophole that allows finite-trap-size BEC in 2D dilute atomic gases. Magnon BECs (Demokritov et al. 2006 Nature 443, 430) and exciton-polariton BECs (Kasprzak et al. 2006 Nature 443, 409) round out the family of quasi-particle condensates discovered in the past two decades.

Synthesis. The Bose-Einstein condensation framework is the foundational reason that quantum statistics has macroscopic experimental consequences in conditions ranging from nanokelvin atom traps to liquid helium at a few kelvin to photon microcavities at room temperature. The central insight is that the polylogarithm $g_{3/2}(z)$ saturates at $z=1$, and this is exactly the structural feature that distinguishes Bose-Einstein statistics from the unsaturable Maxwell-Boltzmann form. Putting these together with the Bogoliubov spectrum and the Landau superfluid criterion, the connection between condensation and superfluidity becomes the unifying observable: a dilute BEC at $T=0$ is a superfluid with critical velocity equal to the sound speed $c=\sqrt{n{U}_0/m}$, an experimental signature distinct from any classical analogue. The pattern recurs across particle species: the same statistical-mechanical structure governs photon BEC in microcavities, magnon BEC in pumped ferromagnets, exciton-polariton BEC in semiconductor microcavities, and the cold-atom BECs of alkali gases. The bridge is between the ideal-gas saturation argument (Einstein 1924–25) and the interacting-gas mean-field treatment (Bogoliubov 1947, Gross-Pitaevskii 1961): the order parameter is the condensate wave function $\Psi (\mathbf{r},t)$, the U(1) symmetry it breaks is the global phase symmetry, and the Landau-Ginzburg framework identifies the BEC universality class with the 3D XY model. This pattern generalises further: the BEC-BCS crossover in two-component Fermi gases identifies the dimer-BEC limit with the Cooper-pair BCS limit, and the cosmological-scale Bose-Einstein condensate of cold axion dark matter — if confirmed — would extend the BEC framework to galaxy-scale halos.

Full proof set Master

Proposition 1 (saturation of the polylogarithm). The function $g_s(z)={\sum }_{\ell \ge 1}{z}^{\ell }/{\ell }^s$ for $s>1$ is strictly increasing and continuous on $0\le z\le 1$, with $g_s(0)=0$ and $g_s(1)=\zeta (s)$, the Riemann zeta value. For $s=3/2$, $\zeta (3/2)\approx 2.6124$; for $s=5/2$, $\zeta (5/2)\approx 1.3415$.

Proof. For $0\le z<1$, the series $\sum z^{\ell }/{\ell }^s$ converges absolutely (comparison to a geometric series), so $g_s$ is analytic on the open disk. Termwise differentiation gives $g_s^{\prime }(z)=\sum z^{\ell -1}/{\ell }^{s-1}=g_{s-1}(z)/z$, which is positive on $(0,1)$. Therefore $g_s$ is strictly increasing on $(0,1)$. At $z=1$, $g_s(1)={\sum }_{\ell }1/{\ell }^s=\zeta (s)$ converges for $s>1$ (Riemann's classical result). Continuity at $z=1$ follows from Abel's theorem: the radius of convergence is 1, the series converges absolutely at $z=1$, and the partial sums converge uniformly on $[0,1]$. $\square$

Proposition 2 (continuity of the chemical potential at $T_c$). In the thermodynamic limit, the chemical potential $\mu (T,n)$ of the ideal Bose gas is continuous at the critical temperature, with $\mu (T_c,n)=0$, but $\partial \mu /\partial T$ has a finite discontinuity.

Proof. For $T>T_c$, $\mu$ is determined implicitly by $n{\lambda }_T^3=g_{3/2}(e^{\mu /k_{B}T})$, with $n{\lambda }_T^3<\zeta (3/2)$. By the inverse-function theorem, $\mu$ is a smooth function of $T$ in this regime, with $\mu \to 0^{-}$ as $T\to T_c^{+}$. For $T\le T_c$, the formula $n{\lambda }_T^3\ge \zeta (3/2)$ forces $z=1$ identically, hence $\mu =0$. Continuity at $T_c$ follows from continuity of $g_{3/2}$ at $z=1$ (Proposition 1).

For the derivative: differentiating $n{\lambda }_T^3=g_{3/2}(z)$ implicitly, $$ n \cdot \frac{d\lambda_T^3}{dT} ;=; g_{1/2}(z) \cdot \frac{1}{z} \cdot \frac{dz}{dT}. $$ As $T\to T_c^{+}$, $z\to 1^{-}$, and $g_{1/2}(z)\to +\infty$ (since $\zeta (1/2)$ is divergent). Therefore $dz/dT\to 0$, but $d\mu /dT=(k_{B}T/z)(dz/dT)-(k_B/z)\ln z$ tends to a finite negative value at $T_c^{+}$. For $T<T_c$, $\mu \equiv 0$ gives $d\mu /dT=0$. The two one-sided derivatives differ, producing the kink. $\square$

Proposition 3 (Bogoliubov ground-state energy). In the Bogoliubov approximation, the ground-state energy per particle of the weakly-interacting Bose gas is $$ \frac{E_0}{N} ;=; \frac{nU_0}{2}\left[1 + \frac{128}{15\sqrt\pi}(n a_s^3)^{1/2}\right], $$ where the second term is the Lee-Huang-Yang 1957 quantum correction.

Proof. The Bogoliubov ground state is the vacuum of the quasi-particle operators ${\hat{b}}_{\mathbf{p}}$. Its energy relative to the unperturbed mean-field ground state is $$ \Delta E_0 ;=; \frac{1}{2}\sum_{\mathbf{p} \neq 0}\left[E_\mathbf{p} - \varepsilon_\mathbf{p} - nU_0 + \frac{(nU_0)^2}{2\varepsilon_\mathbf{p}}\right], $$ where the last subtraction removes a divergent contact-interaction self-energy in the second-order Born series (a standard ultraviolet renormalisation). Substituting the Bogoliubov spectrum and converting the sum to an integral over momenta: $$ \Delta E_0/V ;=; \frac{1}{2}\int \frac{d^3 p}{(2\pi\hbar)^3}\left[\sqrt{\varepsilon_p(\varepsilon_p + 2nU_0)} - \varepsilon_p - nU_0 + \frac{(nU_0)^2}{2\varepsilon_p}\right]. $$ Substitute $x=p\sqrt{2m/n{U}_0}/(2\hslash )=p\xi /\hslash$ (the dimensionless momentum in units of the inverse healing length): $$ \Delta E_0/V ;=; \frac{(nU_0)^{5/2}}{(2\pi\hbar)^3} \cdot \left(\frac{2m}{nU_0}\right)^{3/2} \cdot I, $$ where $I={\int }_0^{\infty }4\pi x^2[\sqrt{x^2(x^2+2)}-x^2-1+1/(2x^2)]dx=32\sqrt{2}/15$. Combining, $$ \frac{\Delta E_0}{N} ;=; \frac{nU_0}{2} \cdot \frac{128}{15\sqrt\pi}\sqrt{n a_s^3}, $$ which is the Lee-Huang-Yang 1957 result. The leading term $n{U}_0/2$ is the mean-field Hartree energy. The correction has been measured in dilute alkali BECs via radio-frequency spectroscopy (Papp et al. 2008 Phys. Rev. Lett. 101, 135301), with experimental values in agreement with the prediction to about 5 percent. $\square$

Proposition 4 (vortex circulation quantisation). In a Bose-Einstein condensate described by a single-valued condensate wave function $\Psi (\mathbf{r})=\sqrt{n(\mathbf{r})}{e}^{i\varphi (\mathbf{r})}$, the circulation along any closed loop $\mathcal{C}$ is quantised in units of $h/m$: $$ \oint_\mathcal{C} \mathbf{v} \cdot d\boldsymbol{\ell} ;=; \frac{2\pi\hbar}{m} k, \qquad k \in \mathbb{Z}. $$

Proof. The superfluid velocity is $\mathbf{v}(\mathbf{r})=(\hslash /m)\nabla \varphi (\mathbf{r})$, the gradient of the phase. Therefore $$ \oint_\mathcal{C} \mathbf{v} \cdot d\boldsymbol{\ell} ;=; \frac{\hbar}{m} \oint_\mathcal{C} \nabla\phi \cdot d\boldsymbol{\ell} ;=; \frac{\hbar}{m} \Delta\phi, $$ where $\Delta \varphi$ is the net phase change around $\mathcal{C}$. Single-valuedness of $\Psi$ requires $e^{i\Delta \varphi }=1$, so $\Delta \varphi =2\pi k$ for integer $k$. Combining, $\oint \mathbf{v}\cdot d\mathbf{\ell }=(2\pi \hslash /m)k$.

The integer $k$ is the winding number of the phase field around $\mathcal{C}$. A non-zero $k$ requires the phase to be undefined somewhere inside $\mathcal{C}$ — a vortex core where $|\Psi |=0$. The core size is set by the healing length $\xi =\hslash /\sqrt{2mn{U}_0}$: the density rises from zero at the core centre to its bulk value over distance $\xi$, and the kinetic energy of the phase circulation is logarithmically divergent in the cloud size, $E_{\text{vortex}}\sim (n{\hslash }^2/m)\ln (R/\xi )$ per unit length of the vortex line. $\square$

Connections Master

• Bose-Einstein distribution 11.05.01. Supplies the underlying single-state occupation $⟨n_{\mathbf{p}}⟩=1/(e^{({\epsilon }_{\mathbf{p}}-\mu )/k_{B}T}-1)$ from which the BEC analysis proceeds. The condensation phenomenon emerges as the failure of the integrated occupation to absorb all particles when the chemical potential approaches the ground-state energy from below; the same distribution evaluated at $\mu =0$ on photons yields the Planck spectrum.

• Blackbody radiation 11.05.03. The Planck spectrum is the photon-gas case at $\mu =0$ identically (photon number is not conserved). BEC of a massive gas is the analogous structure at $\mu \to 0^{-}$ with conserved particle number: the same Bose-Einstein occupation, the same density-of-states integral, the same polylogarithm — different constraint on $\mu$. The shared mathematical structure is the foundational reason both phenomena rely on the saturation of $g_{3/2}(z)$ or $g_3(z)$ at $z=1$.

• Bosonic Fock space and second quantisation 12.13.01. Provides the second-quantised framework needed for the Bogoliubov treatment of the weakly-interacting Bose gas. The $c$-number replacement ${\hat{a}}_0\to \sqrt{N_0}$, the Bogoliubov transformation ${\hat{a}}_{\mathbf{p}}=u_{\mathbf{p}}{\hat{b}}_{\mathbf{p}}-v_{\mathbf{p}}{\hat{b}}_{-\mathbf{p}}^{†}$, and the resulting spectrum $E_{\mathbf{p}}=\sqrt{{\epsilon }_{\mathbf{p}}({\epsilon }_{\mathbf{p}}+2n{U}_0)}$ all live in the bosonic Fock-space construction; the canonical-commutation algebra is the load-bearing piece of the diagonalisation argument.

• First and second laws of thermodynamics 11.01.01. The BEC transition is constrained by the Clausius-Clapeyron relation for the latent-heat structure of the lambda transition, and by the third law for the condensate fraction approaching unity at $T=0$. The thermodynamic-identity derivation of the heat capacity discontinuity at $T_c$ relies on the standard Maxwell relations and entropy continuity across the transition.

Historical & philosophical context Master

Bose 1924 ^[Bose1924] submitted a manuscript deriving Planck's radiation law by counting indistinguishable photon arrangements in phase-space cells — a counting procedure that differed from the Maxwell-Boltzmann hypothesis on what to treat as distinct microstates. Einstein translated the paper into German, communicated it to the Zeitschrift für Physik, and recognised that the same counting applied to massive particles would predict a peculiar low-temperature accumulation. Einstein 1924 ^{[Einstein1924]} and Einstein 1925 ^{[Einstein1925]} developed the analysis in two follow-up papers, predicting condensation for a non-interacting ideal Bose gas and computing the critical temperature in essentially the modern form. The 1925 paper noted the unusual phase-transition character (the condensation is a thermodynamic singularity that develops only in the limit of infinite volume) and offered no clear physical realisation; Einstein regarded the result as an interesting mathematical curiosity.

Helium-4 was first liquefied by Kamerlingh Onnes in 1908; its superfluid behaviour was discovered by Kapitsa 1938 Nature 141, 74, and independently by Allen and Misener 1938 Nature 141, 75. London 1938 ^[London1938] proposed that the superfluid transition at $T_{\lambda }=2.17\text{K}$ is Bose-Einstein condensation, with the ideal-gas formula giving $T_c\approx 3.13\text{K}$ — within 30 percent of the observed value despite the strong inter-atomic interactions. Tisza 1938 C. R. Acad. Sci. 207, 1035 introduced the phenomenological two-fluid model; the modern reconciliation of the two pictures (BEC of a small condensate fraction, with the superfluid component encompassing both condensate and quasi-particle excitations) was completed by the 1947–1959 work of Bogoliubov, Landau, and Penrose-Onsager.

Bogoliubov 1947 ^{[Bogoliubov1947]} developed the microscopic theory of the weakly-interacting Bose gas, replacing the condensate-mode operators by c-numbers and diagonalising the resulting bilinear Hamiltonian. The Bogoliubov transformation became the prototype for analogous diagonalisations in BCS superconductivity (Bogoliubov 1958), antiferromagnetism (Holstein-Primakoff 1940 plus Bogoliubov diagonalisation), and the quantisation of fields in curved spacetime (Hawking radiation derivation). Gross 1961 ^[Gross1961] and Pitaevskii 1961 ^{[Pitaevskii1961]} independently derived the non-linear Schrödinger equation for the condensate wave function, completing the mean-field framework. The 1957 Lee-Huang-Yang correction (Phys. Rev. 106, 1135) provided the first quantum correction to the Bogoliubov ground-state energy, later confirmed in dilute alkali BECs.

The dilute-gas BEC programme of the 1990s built on three preceding decades of laser-cooling and atomic-physics development. Letokhov 1968 JETP Lett. 7, 348 and Hänsch-Schawlow 1975 Opt. Commun. 13, 68 proposed laser cooling of atoms; the technique was realised by Chu, Cohen-Tannoudji, and Phillips (1997 Nobel Prize). Evaporative cooling of magnetically trapped atoms (Hess 1986 Phys. Rev. B 34, 3476) provided the final factor of 100 needed to reach the BEC regime. Anderson, Ensher, Matthews, Wieman & Cornell 1995 ^{[AndersonEnsherMatthewsWiemanCornell1995]} reported the first BEC at JILA: $2\times 10^4$ rubidium-87 atoms cooled to $\sim 170\text{nK}$ in a magnetic trap, with the characteristic narrow peak in the velocity distribution emerging as the temperature was lowered below the predicted $T_c$. Davis et al. 1995 ^{[DavisMewesAndrewsVanDrutenDurfeeKurnKetterle1995]} reported sodium-23 BEC at MIT six months later with $5\times 10^5$ atoms, and Hulet's group at Rice reported lithium-7 BEC the same year (with the additional twist that lithium-7's negative scattering length led to collapse instabilities later analysed by Pérez-García et al. 1997 Phys. Rev. A 56, 1424). Cornell, Ketterle, and Wieman shared the 2001 Nobel Prize.

The subsequent two decades produced an enormous experimental and theoretical literature. Vortices in BECs were observed in rotating sodium clouds (Madison-Chevy-Wohlleben-Dalibard 2000 ^{[MadisonChevyWohllebenDalibard2000]}) with the Abrikosov triangular lattice confirmed in 2001 (Abo-Shaeer et al. Science 292, 476). The BEC-BCS crossover was experimentally realised in cold-atom Fermi gases at unitarity (Regal-Greiner-Jin 2004 Phys. Rev. Lett. 92, 040403; Zwierlein et al. 2004 Phys. Rev. Lett. 92, 120403). Photon BEC was achieved by Klaers et al. 2010 ^[Klaers2010] in a dye-filled microcavity, extending the BEC framework to the two-dimensional case via finite cavity confinement. Magnon BEC (Demokritov et al. 2006 Nature 443, 430) and exciton-polariton BEC (Kasprzak et al. 2006 Nature 443, 409) further populated the family of quasi-particle condensates. The Mott-insulator transition in optical-lattice BECs (Greiner et al. 2002 Nature 415, 39) opened the field of quantum simulation with cold atoms, leading to the modern cold-atom platforms for quantum-computing and lattice-gauge-theory simulation.

The conceptual lineage of Bose-Einstein condensation links pre-quantum classical statistics (Boltzmann 1872), the Bose-Einstein 1924–25 statistical-mechanical prediction, the London 1938 superfluid-helium identification, the Bogoliubov-Gross-Pitaevskii mean-field theory of the late 1940s and early 1960s, and the 1995 experimental confirmation in dilute alkali gases. The 70-year gap between Einstein's prediction and the first laboratory realisation was filled by the development of laser and evaporative cooling, the discovery of magnetic and optical traps, and the establishment of the dilute-gas regime where the ideal-gas formula is quantitatively accurate. Modern BEC physics now spans atomic, molecular, optical, and condensed-matter regimes, with theoretical extensions to relativistic Bose-Einstein condensates (axion dark-matter models, neutron-star pion condensates) and cosmological-scale objects. The 1925 Einstein paper remains foundational reading 100 years after its publication.

Bibliography Master

Primary literature:

• Bose, S. N. (1924). "Plancks Gesetz und Lichtquantenhypothese." Zeitschrift für Physik 26, 178–181.

• Einstein, A. (1924). "Quantentheorie des einatomigen idealen Gases." Sitzungsberichte der Preussischen Akademie der Wissenschaften, Berlin 22, 261–267.

• Einstein, A. (1925). "Quantentheorie des einatomigen idealen Gases. Zweite Abhandlung." Sitzungsberichte der Preussischen Akademie der Wissenschaften, Berlin 1, 3–14.

• London, F. (1938). "The λ-phenomenon of liquid helium and the Bose-Einstein degeneracy." Nature 141, 643–644.

• Bogoliubov, N. N. (1947). "On the theory of superfluidity." Journal of Physics USSR 11, 23–32.

• Lee, T. D., Huang, K., & Yang, C. N. (1957). "Eigenvalues and Eigenfunctions of a Bose System of Hard Spheres and Its Low-Temperature Properties." Physical Review 106, 1135–1145.

• Gross, E. P. (1961). "Structure of a quantized vortex in boson systems." Nuovo Cimento 20, 454–477.

• Pitaevskii, L. P. (1961). "Vortex lines in an imperfect Bose gas." Soviet Physics JETP 13, 451–454.

• Mermin, N. D. & Wagner, H. (1966). "Absence of Ferromagnetism or Antiferromagnetism in One- or Two-Dimensional Isotropic Heisenberg Models." Physical Review Letters 17, 1133–1136.

• Hohenberg, P. C. (1967). "Existence of Long-Range Order in One and Two Dimensions." Physical Review 158, 383–386.

• Kosterlitz, J. M. & Thouless, D. J. (1973). "Ordering, metastability and phase transitions in two-dimensional systems." Journal of Physics C 6, 1181–1203.

• Anderson, M. H., Ensher, J. R., Matthews, M. R., Wieman, C. E., & Cornell, E. A. (1995). "Observation of Bose-Einstein condensation in a dilute atomic vapor." Science 269, 198–201.

• Davis, K. B., Mewes, M.-O., Andrews, M. R., van Druten, N. J., Durfee, D. S., Kurn, D. M., & Ketterle, W. (1995). "Bose-Einstein condensation in a gas of sodium atoms." Physical Review Letters 75, 3969–3973.

• Madison, K. W., Chevy, F., Wohlleben, W., & Dalibard, J. (2000). "Vortex formation in a stirred Bose-Einstein condensate." Physical Review Letters 84, 806–809.

• Greiner, M., Mandel, O., Esslinger, T., Hänsch, T. W., & Bloch, I. (2002). "Quantum phase transition from a superfluid to a Mott insulator in a gas of ultracold atoms." Nature 415, 39–44.

• Klaers, J., Schmitt, J., Vewinger, F., & Weitz, M. (2010). "Bose-Einstein condensation of photons in an optical microcavity." Nature 468, 545–548.

Modern textbook anchors:

• Pitaevskii, L. & Stringari, S. (2016). Bose-Einstein Condensation and Superfluidity. Oxford University Press.

• Pethick, C. J. & Smith, H. (2008). Bose-Einstein Condensation in Dilute Gases, 2nd ed. Cambridge University Press.

• Pathria, R. K. & Beale, P. D. (2021). Statistical Mechanics, 4th ed. (Elsevier), §7.1–7.2 "The Bose-Einstein gas and Bose-Einstein condensation."

• Landau, L. D. & Lifshitz, E. M. (1980). Statistical Physics, Part 1, 3rd ed. (Course of Theoretical Physics Vol. 5, Pergamon), §62 "Degenerate Bose gas and Bose-Einstein condensation."

• Huang, K. (1987). Statistical Mechanics, 2nd ed. (Wiley), Ch. 12.

• Reif, F. (1965). Fundamentals of Statistical and Thermal Physics (McGraw-Hill), §9.7–9.10 "Quantum statistics and the Bose-Einstein gas."

• Griffiths, D. J. & Schroeter, D. F. (2018). Introduction to Quantum Mechanics, 3rd ed. (Cambridge University Press), §5.4 "Quantum Statistical Mechanics."

• Eisberg, R. & Resnick, R. (1985). Quantum Physics of Atoms, Molecules, Solids, Nuclei, and Particles, 2nd ed. (Wiley), Ch. 11.