Degenerate Fermi Gases and BEC-BCS Crossover

Degenerate Fermi gases take two ideas wholesale from quantum statistical mechanics: the Pauli exclusion principle, which forbids two identical fermions from sharing a single-particle state, and the Fermi–Dirac distribution that follows from it. Run a dilute gas of fermionic atoms — usually ⁶Li or ⁴⁰K — through the very same magneto-optical, dipole and evaporative cooling that makes a Bose–Einstein condensate, and instead of collapsing into one mode the atoms are forced to stack one per momentum state, filling every level below the Fermi energy $E_F$ and building a Fermi sea with a razor-sharp surface. Degeneracy sets in when the thermal de Broglie wavelength grows to the interparticle spacing, that is when the temperature drops below the Fermi temperature $T_F$ ; today's experiments routinely reach $T/T_F\approx0.05$ , deep in the quantum regime. To make the sea pair and flow, one uses two internal hyperfine states as a pseudo-spin $\uparrow,\downarrow$ . Atoms in opposite states collide via an s-wave scattering length $a$ that a magnetically tuned Feshbach resonance can set to any value — positive, negative, or infinite. That one knob drags the gas continuously from tightly bound bosonic molecules that Bose-condense on the BEC side, through a strongly interacting universal regime at unitarity, to weakly attracted, spatially overlapping Cooper pairs on the BCS side. No phase transition marks the journey — a single superfluid ground state connects the ends, making the gas a pristine, fully tunable quantum simulator for correlated fermionic matter, from neutron-star crusts to high- $T_c$ superconductors.

Key takeaways

A gas of fermionic atoms cooled below its Fermi temperature $T_F$ fills a Fermi sea: quantum statistics forbid double occupancy, so even at $T=0$ atoms occupy every state up to the Fermi energy $E_F=k_BT_F$ .
In a two-component mixture the s-wave scattering length $a$ is tuned across a Feshbach resonance, sweeping the interaction from weak attraction ( $a<0$ ) through the unitary point ( $|a|\to\infty$ ) to weak repulsion with bound molecules ( $a>0$ ).
On the BCS side loosely bound Cooper pairs live in momentum space and overlap heavily; on the BEC side atoms form tight diatomic molecules that condense — one gap-preserving ground state connects the two.
At unitarity the scattering length drops out entirely; the only scales are $k_F$ and $T$ , so thermodynamics is universal and fixed by the Bertsch parameter $\xi\approx0.37$ , with $T_c/T_F\approx0.17$ .
The whole crossover is a superfluid: gap, condensate fraction and $T_c$ evolve smoothly, making cold atoms a clean analog for BCS superconductors and neutron matter.
Key idea: Tuning the scattering length through a Feshbach resonance connects a Bose–Einstein condensate of diatomic molecules to a BCS superfluid of Cooper pairs as a single, gap-preserving crossover — with a universal strongly interacting fixed point at unitarity.

Where this lives today

The unitary Fermi gas is now a precision metrology tool. In 2012 the MIT group of Martin Zwierlein measured the full equation of state of a unitary ⁶Li gas by imaging a trapped cloud slice by slice, pinning $\xi=0.376$ and $T_c/T_F=0.167$ — numbers that quantum Monte Carlo must reproduce and that stand in for dilute neutron matter. Radio-frequency spectroscopy, pioneered in Jin's lab, still reads out the pairing gap across the crossover, and rotating clouds nucleate vortex lattices that are the unambiguous fingerprint of superfluidity even at unitarity. The newest frontier is the fermionic quantum-gas microscope: single-atom-resolved imaging of a two-component gas in an optical lattice, realizing the Hubbard model site by site and letting cold atoms emulate the doped antiferromagnet believed to underlie high- $T_c$ superconductivity — with the Feshbach knob and lattice depth setting interaction and hopping independently.

The physical picture

Picture pouring identical marbles into a bowl whose slots each hold exactly one: they pile up to a definite height, the Fermi surface. Bosons, by contrast, all crowd into the lowest slot. This single difference — Pauli exclusion — is why a spinless Fermi gas can never condense into one mode and why you need two spin states before anything can pair. Cool the gas below $T_F$ and the sea forms; turn on an attraction between the two spins and pairs appear. What kind of pair depends entirely on how hard you turn the Feshbach knob, and that dial is the whole story.

Figure

One ground state, two limits: sweeping the interaction dial 1/(k_Fa) across a Feshbach resonance runs the gas from weak attraction on the BCS side through the unitary point where the scattering length diverges to tightly bound molecules on the BEC side, with the universal numbers ξ ≈ 0.37 and T_c/T_F ≈ 0.17 fixed at unitarity

On the BCS side the pairs are enormous — far larger than the atomic spacing, so many overlap and only momentum-space correlations betray them: a Cooper pair is a ripple on the Fermi surface. On the BEC side the pairs are tiny molecules, smaller than the spacing, behaving as point bosons. The pair size passes through the interparticle spacing exactly at unitarity, where the scattering length is infinite, $k_F$ is the only scale, and every energy is a pure number times $E_F$ — the hallmark of a universal, scale-invariant quantum fluid. All of this pairing physics only becomes visible once the sea itself is cold and sharp, the subject of the second picture: how the Fermi–Dirac occupation freezes into a filled sea as $T/T_F$ falls.

Figure

The Fermi–Dirac occupation f(E) sharpening into a filled sea: at T/T_F = 0.5 the step is thermally rounded, at T/T_F = 0.1 it is nearly square, and at T = 0 it is a perfect step at the Fermi energy E_F — today's cold-atom experiments reach T/T_F ≈ 0.05, deep in the degenerate regime

The core equations

The whole subject is calibrated by one energy. Fill momentum space up to the Fermi wavevector $k_F$ , one fermion per state, and the topmost occupied level is the Fermi energy — the chemical potential at $T=0$ and the natural unit for every other scale. For a uniform two-component gas of total density $n$ it reads:

E_F=k_BT_F=\frac{\hbar^2k_F^2}{2m}=\frac{\hbar^2}{2m}\left(3\pi^2n\right)^{2/3} \qquad (1)

where: $E_F$ — Fermi energy, the chemical potential at $T=0$ ; $k_F$ — Fermi wavevector, $=(3\pi^2n)^{1/3}$ ; $n$ — total number density of both spin states; $m$ — atomic mass; $T_F$ — Fermi temperature, the energy $E_F$ expressed in kelvin. A larger density stiffens the sea: $E_F$ grows as $n^{2/3}$ . Cold-atom experiments never hold a uniform gas; they hold a cloud in a harmonic trap, so the relevant Fermi energy counts oscillator states, not plane waves. For a spin-balanced mixture of total atom number $N$ it takes a compact form.

E_F=\hbar\bar\omega\,(3N)^{1/3} \qquad (2)

where: $\bar\omega$ — geometric-mean trap frequency $(\omega_x\omega_y\omega_z)^{1/3}$ ; $N$ — total atom number in the balanced two-spin mixture. Counting per spin instead, with $N_\sigma=N/2$ , gives the equivalent $E_F=\hbar\bar\omega(6N_\sigma)^{1/3}$ — always check which convention a source uses. At finite temperature the sharp edge of the sea softens. The mean occupation of a state of energy $E$ is the Fermi–Dirac function; the plus sign in the denominator caps every occupation between zero and one, enforcing Pauli exclusion.

f(E)=\frac{1}{e^{(E-\mu)/k_BT}+1} \qquad (3)

where: $f(E)$ — mean occupation of a state of energy $E$ ; $\mu$ — chemical potential, with $\mu\to E_F$ as $T\to0$ ; $k_BT$ — thermal energy. The step is rounded over a width $\sim k_BT$ about $\mu$ , which is why deep degeneracy ( $T\ll T_F$ ) is needed to see pairing. Interactions enter through a single dimensionless dial. The scattering length $a$ is set at will by a magnetic Feshbach resonance, and comparing it to the spacing $1/k_F$ gives the one knob that carries the gas across the entire crossover.

\frac{1}{k_Fa},\qquad k_F=(3\pi^2 n)^{1/3} \qquad (4)

where: $a$ — s-wave scattering length, tunable via Feshbach resonance; $1/(k_Fa)$ — the crossover parameter, running from $-\infty$ (BCS) through $0$ (unitarity) to $+\infty$ (BEC). It is the horizontal axis of the whole phase diagram. Right at the resonance the scattering length is infinite and cannot appear in any formula, so $k_F$ is the only scale left and the ground-state energy must be a pure number times the ideal-gas value. That number is the Bertsch parameter.

E_0=\xi\,\frac{3}{5}N E_F,\qquad \xi\approx0.37 \qquad (5)

where: $E_0$ — ground-state energy at unitarity; $\xi$ — Bertsch parameter, a universal constant $\approx0.37$ (quantum Monte Carlo and MIT/ENS experiment); $\tfrac{3}{5}NE_F$ — ground-state energy of the ideal Fermi gas. The chemical potential at unitarity is then $\mu=\xi E_F\approx0.37\,E_F$ . Past resonance on the repulsive side a genuine two-body bound state exists: a weakly bound "halo" dimer whose size is the scattering length itself, setting the molecular scale of the BEC side.

E_b=\frac{\hbar^2}{ma^2}\qquad(a>0) \qquad (6)

where: $E_b$ — binding energy of the diatomic molecule; $m$ — single-atom mass (the reduced mass $m/2$ is already folded in); $a$ — scattering length, also the dimer radius. As $a\to\infty$ the dimer unbinds ( $E_b\to0$ ) — the molecule dissolves into the sea exactly at unitarity. On the opposite, weakly attracting side the pairs are fragile and the energy to break one is exponentially small in the coupling — the signature of BCS superfluidity, derived below.

\Delta_{\rm BCS}=\frac{8}{e^2}\,E_F\,\exp\!\left(-\frac{\pi}{2k_F|a|}\right) \qquad (7)

where: $\Delta_{\rm BCS}$ — energy gap to break a Cooper pair; $8/e^2\approx1.08$ ; the exponent $-\pi/(2k_F|a|)$ diverges as $k_F|a|\to0$ . The transition temperature tracks the gap: $T_c\simeq0.28\,T_F\,e^{-\pi/(2k_F|a|)}$ on the deep BCS side.

Derivation AFermi energy and density of states of a uniform gas

The Fermi energy is the single number that organizes everything downstream — the gap, the transition temperature, the size of a trapped cloud — so we build it from scratch by counting plane-wave states in a box and dropping one fermion into each, per spin, until the atoms run out. The topmost filled state defines $E_F$ , and the same counting hands us the density of states $g(E)\propto\sqrt{E}$ that fixes the ideal-gas energy at $\tfrac35 NE_F$ .

Step 1. Quantize plane waves in a cube of side $L$ : allowed wavevectors sit on a grid of spacing $2\pi/L$ , so each occupies a $k$ -space cell of volume $(2\pi)^3/V$ . Fill a sphere of radius $k_F$ with two spin states per cell, giving $N=2\,V\,(4\pi k_F^3/3)/(2\pi)^3=V k_F^3/3\pi^2$ . Why: one fermion per state is the entire content of Pauli exclusion, and it forces the gas up to a finite momentum even at $T=0$ .

Step 2. Invert in terms of the total density $n=N/V$ to get the Fermi wavevector $k_F=(3\pi^2 n)^{1/3}$ , the radius of the filled Fermi sphere. Why: $k_F$ is the only length scale an ideal Fermi gas possesses, so every energy in the problem must be built out of it.

Step 3. The topmost occupied state carries energy $E_F=\hbar^2 k_F^2/2m=(\hbar^2/2m)(3\pi^2 n)^{2/3}=k_BT_F$ . Why: this is the chemical potential at $T=0$ and the yardstick against which temperature, the gap $\Delta$ , and the binding energy $E_b$ are all measured.

Step 4. Differentiate the state count with respect to energy to get $g(E)=(V/2\pi^2)(2m/\hbar^2)^{3/2}\sqrt{E}$ ; integrating $E\,g(E)$ up to $E_F$ yields a total energy $\tfrac35 NE_F$ . Why: the $\sqrt{E}$ density of states is exactly why the average energy per particle is $\tfrac35 E_F$ rather than $E_F$ , and it is the ideal-gas benchmark the Bertsch parameter rescales.

Go deeper · Derivation B · 1 minContinuity of BEC and BCS from the pairing wavefunction

Derivation BContinuity of BEC and BCS from the pairing wavefunction

The claim that BEC and BCS are two ends of one road rather than two distinct phases rests on a single object: the BCS variational wavefunction. We write it down, read off the pair wavefunction it secretly encodes, and then follow that pair from a fat Fermi-surface ripple to a compact molecule as the chemical potential $\mu$ slides continuously from $+E_F$ down through zero to $-E_b/2$ — with no symmetry ever changing, hence no phase transition anywhere along the way.

Step 1. Write the paired ground state $|\Psi\rangle=\prod_{\mathbf k}(u_k+v_k\,c^\dagger_{\mathbf k\uparrow}c^\dagger_{-\mathbf k\downarrow})|0\rangle$ with $|u_k|^2+|v_k|^2=1$ ; the ratio $\phi(\mathbf k)=v_k/u_k$ is the momentum-space wavefunction of a single pair. Why: one wavefunction describes the entire many-body gas, and it is precisely this pair amplitude that morphs across the crossover.

Step 2. On the BCS side $\mu>0$ and $\Delta\ll\mu$ , so $v_k^2$ is a slightly smeared step at $k_F$ : pairs are ripples on the Fermi surface, far larger than the spacing $n^{-1/3}$ , and overlap heavily. Why: a pair size much greater than the spacing is exactly the weak-coupling, momentum-space picture of Cooper pairing — real-space pairs you cannot even tell apart.

Step 3. As the attraction strengthens $\mu$ turns negative toward $-E_b/2$ , the Fermi surface vanishes, and $\phi(\mathbf k)\propto 1/(k^2+a^{-2})$ — the Fourier transform of the halo-dimer wavefunction $e^{-r/a}/r$ . Why: a negative $\mu$ means there is no sea left to pair on; the pairs have become genuine bound molecules of size $a$ , and the gas is a BEC of them.

Step 4. Because $\mu$ moves continuously and the $U(1)$ pairing symmetry is broken throughout, no thermodynamic quantity is singular anywhere between the limits. Why: an order parameter $\Delta$ that never vanishes and a symmetry that never changes is the definition of a crossover, not a phase transition.

Go deeper · Derivation C · 1 minThe exponentially small BCS gap

Derivation CThe exponentially small BCS gap

On the BCS side the pairing gap is not algebraic in the coupling — it is exponentially small, and seeing why demands the mean-field gap equation regularized so that the bare, divergent contact attraction is traded for the physical scattering length $a$ . That single renormalization removes an ultraviolet divergence and leaves a clean logarithm, which we invert to expose the tell-tale factor $e^{-\pi/2k_F|a|}$ that dominates weak-coupling superfluidity.

Step 1. The $T=0$ gap equation, with the contact interaction replaced by $a$ through the vacuum two-body T-matrix, reads $-m/(4\pi\hbar^2 a)=\int d^3k/(2\pi)^3\,(1/2E_k-m/\hbar^2 k^2)$ , where $E_k=\sqrt{(\epsilon_k-\mu)^2+\Delta^2}$ and $\epsilon_k=\hbar^2k^2/2m$ . Why: subtracting the $m/\hbar^2 k^2$ term is what cancels the divergence of the bare gap equation and makes $a$ the sole interaction input.

Step 2. For weak attraction $\Delta\ll E_F$ and $\mu\approx E_F$ , so the integrand is sharply peaked at the Fermi surface; evaluating it there produces a logarithm $\ln(\Delta/E_F)$ plus finite constants. Why: pairing is a Fermi-surface phenomenon, so the logarithm comes entirely from the thin shell where $\epsilon_k\approx\mu$ — the same log that gives superconductivity its exponential $T_c$ .

Step 3. Collecting the constants gives $\ln(\Delta e^2/8E_F)=-\pi/(2k_F|a|)$ , i.e. $\Delta=(8/e^2)E_F\exp(-\pi/2k_F|a|)$ with $8/e^2\approx1.08$ . Why: the coupling sits inside an exponent, so halving $k_F|a|$ does not halve $\Delta$ — it slashes it by orders of magnitude, which is why weak-coupling pairing needs absurdly low temperatures.

Limiting cases

The gas is organized by two ratios: the degeneracy ratio $T/T_F$ (is the sea filled?) and the interaction dial $1/(k_Fa)$ with $k_F=(3\pi^2n)^{1/3}$ (how strong is the pairing?). Together they sweep the following regimes.

Case	Result	Meaning
Classical gas	$T\gg T_F,\ n\lambda_{\rm dB}^3\ll1$	Maxwell–Boltzmann statistics; quantum degeneracy irrelevant, no Fermi sea
Ideal degenerate	$T\ll T_F,\ a\to0$	Filled Fermi sea, energy $E=\tfrac35 NE_F$ ; sharp Fermi surface but no pairing or superfluidity
BEC limit	$1/(k_Fa)\gg+1$	Tight bosonic dimers of size $\sim a$ , binding $E_b=\hbar^2/(ma^2)$ ; they Bose-condense, dimer-dimer length $a_{dd}\approx0.6\,a$
Unitarity	$1/(k_Fa)=0,\ \|a\|\to\infty$	Scale-invariant universal gas; $E_0=\xi\,\tfrac35 NE_F$ with $\xi\approx0.37$ , $\mu=\xi E_F$ , $T_c/T_F\approx0.17$
BCS limit	$1/(k_Fa)\ll-1$	Weak attraction; exponentially small gap $\Delta\simeq\tfrac{8}{e^2}E_F\,e^{-\pi/2k_F\|a\|}$ , pairs far larger than the spacing

Common trapthe BEC–BCS crossover is a phase transition from bosons to fermions

It is not a phase transition at all. The superfluid ground state and its broken $U(1)$ pairing symmetry are the same throughout; nothing becomes singular as the Feshbach knob is turned. What changes continuously is the pair size relative to the spacing, and the sign of $\mu$ (positive on the BCS side, negative on the BEC side, zero near unitarity). A second, common slip: $T_F$ is a Fermi temperature — the energy $E_F/k_B$ of the filled sea — not a transition temperature. Superfluidity sets in far below it; even at the strongly paired unitary point $T_c/T_F\approx0.17$ , not at $T_F$ itself. Calling $T_F$ "the critical temperature" overestimates $T_c$ by nearly an order of magnitude.

Worked examples

1 · Fermi temperature of a trapped lithium gas (confidence builder)

A balanced two-spin cloud of ⁶Li holds $N=1.0\times10^5$ atoms per spin in a trap of geometric-mean frequency $\bar\omega=2\pi\times200$ Hz. Find the Fermi energy and temperature, and the transition temperature if the gas is tuned to unitarity.

Step 1. Use the per-spin harmonic result $E_F=\hbar\bar\omega(6N)^{1/3}$ . First the counting factor: $(6\times1.0\times10^5)^{1/3}=(6.0\times10^5)^{1/3}=84.3$ . This is how many oscillator quanta deep the sea reaches.

Step 2. Multiply by the level spacing $\hbar\bar\omega=(1.0546\times10^{-34})(2\pi\times200)=1.325\times10^{-31}$ J: $E_F=1.325\times10^{-31}\times84.3=1.118\times10^{-29}$ J.

Step 3. Convert to a temperature: $T_F=E_F/k_B=(1.118\times10^{-29})/(1.381\times10^{-23})=\mathbf{0.81\ \mu K}$ . This is a typical cold-atom Fermi temperature — the sea is "hot" only in the sense that it sits nearly a microkelvin above absolute zero.

Step 4. At unitarity the transition temperature is $T_c\approx0.17\,T_F=0.17\times0.81\ \mu\text{K}=\mathbf{0.14\ \mu K}$ . Superfluidity requires cooling from $T_F$ to roughly a sixth of it — well within reach of evaporative cooling, which is why the unitary gas is the easiest strongly paired superfluid to reach. $T_F\approx0.81\ \mu$ K, and the unitary gas becomes superfluid below $T_c\approx0.14\ \mu$ K.

2 · The exponentially small BCS gap

A gas is tuned to the BCS side with $k_F|a|=0.4$ (so $1/(k_Fa)=-2.5$ ). Estimate the pairing gap $\Delta/E_F$ , and compare with the gap at unitarity, where quantum Monte Carlo gives $\Delta\approx0.5\,E_F$ .

Step 1. Evaluate the exponent: $-\pi/(2k_F|a|)=-\pi/0.8=-3.93$ . This large negative number is why the gap is small — the coupling lives inside an exponential.

Step 2. Exponentiate: $e^{-3.93}=0.0197$ . Multiply by the prefactor $8/e^2=1.08$ : $\Delta=1.08\times0.0197\,E_F=\mathbf{0.021\ E_F}$ . If $E_F/k_B\approx1\ \mu$ K, this is a gap of only $\Delta/k_B\approx21$ nK.

Step 3. Compare with unitarity, $\Delta\approx0.5\,E_F$ . The ratio is $0.5/0.021\approx24$ : turning the dial from $1/(k_Fa)=-2.5$ to $0$ boosts the gap more than twentyfold, even though $k_F|a|$ only changes from $0.4$ to $\infty$ . On the BCS side $\Delta\approx0.021\,E_F$ , roughly $24\times$ smaller than the $\approx0.5\,E_F$ gap at unitarity — the exponential suppression in action.

3 · Universal energy at unitarity

Using the Bertsch parameter $\xi=0.37$ , find the ground-state energy per particle of a unitary Fermi gas relative to $E_F$ , and its chemical potential. Contrast with the ideal (non-interacting) gas.

Step 1. The ideal Fermi gas has energy $E_{\rm ideal}=\tfrac35 NE_F$ , i.e. $0.60\,E_F$ per particle — the benchmark set purely by Pauli filling, with no interactions.

Step 2. At unitarity the attraction lowers this to $E_0=\xi\,\tfrac35 NE_F=0.37\times0.60\,NE_F=\mathbf{0.222\ NE_F}$ , i.e. $0.222\,E_F$ per particle — the interacting gas sits at only 37% of the ideal energy.

Step 3. Since $E_0\propto N E_F\propto N^{5/3}$ for a uniform gas ( $E_F\propto n^{2/3}$ ), the chemical potential is $\mu=\partial E_0/\partial N=\xi E_F=\mathbf{0.37\ E_F}$ . Both numbers are pure multiples of $E_F$ with no reference to $a$ — the concrete meaning of "universal." $E_0/N=0.222\,E_F$ and $\mu=0.37\,E_F$ , set by the single constant $\xi$ .

Physics of the portrait

Degenerate atomic Fermi gases went from theory to laboratory reality through one experimentalist's insistence that fermions could be cooled as deeply as bosons — and then paired.

Origins · 1 minDeborah Jin and how this was discovered

Deborah Jin

1968–2016 · ultracold fermionic quantum gases at JILA

Deborah Jin created the field this article describes. In 1999, working at JILA in Boulder with her student Brian DeMarco, she cooled a gas of ⁴⁰K atoms below its Fermi temperature and produced the first quantum-degenerate atomic Fermi gas — the fermionic counterpart of the 1995 Bose–Einstein condensates, and far harder, because identical fermions refuse to collide and so resist the evaporative cooling that works for bosons. Jin's group solved this using two spin states and sympathetic cooling. In 2003–2004 her team used a Feshbach resonance to bind ⁴⁰K atoms into molecules and then observed the condensation of fermionic atom pairs on the BCS side of the resonance — the first "fermionic condensate," direct evidence that Cooper-like pairs of atoms could condense. This established cold fermions as a clean, tunable realization of the BEC–BCS crossover long imagined for superconductors, and her measurements of the crossover, the pairing gap, and the paired-atom momentum distribution became templates the whole community adopted. Widely expected to share a Nobel Prize, Jin died of cancer in 2016 at 47; the degenerate Fermi gas, the unitary superfluid, and the atomic quantum simulator of correlated fermions are her enduring legacy.

History

1926

— Fermi and Dirac derive the quantum statistics of identical half-integer-spin particles

1957

— Bardeen, Cooper and Schrieffer explain superconductivity as a condensate of paired electrons

1980

— Leggett shows the paired ground state connects BCS superfluidity smoothly to a molecular condensate

1999

— DeMarco and Jin cool potassium-40 into the first quantum-degenerate atomic Fermi gas

2004

— Regal, Greiner and Jin observe condensation of fermionic atom pairs across a Feshbach resonance

Going further

Texts. Foot, Atomic Physics (2005), Ch. 10–11 (quantum gases and degeneracy) · Ketterle and Zwierlein, Making, probing and understanding ultracold Fermi gases, Riv. Nuovo Cimento 31 (2008) · Giorgini, Pitaevskii and Stringari, Theory of ultracold atomic Fermi gases, Rev. Mod. Phys. 80, 1215 (2008) · Zwerger (ed.), The BCS–BEC Crossover and the Unitary Fermi Gas, Lecture Notes in Physics 836 (Springer 2012) · Pethick and Smith, Bose–Einstein Condensation in Dilute Gases, 2nd ed.

Provenance. Definitions and equations follow Foot Ch. 10–11 and Giorgini–Pitaevskii–Stringari, Rev. Mod. Phys. 80, 1215 (2008). A verifier should recompute: (1) $E_F=(\hbar^2/2m)(3\pi^2n)^{2/3}$ from momentum-state counting; (2) the trapped $E_F=\hbar\bar\omega(3N)^{1/3}$ for total $N$ , equivalently $(6N_\sigma)^{1/3}$ per spin; (3) the weak-coupling gap prefactor $8/e^2\approx1.08$ and exponent $-\pi/(2k_F|a|)$ ; (4) the worked numbers $T_F\approx0.81\ \mu$ K and $\Delta\approx0.021\,E_F$ at $k_F|a|=0.4$ . Standard cited values: Bertsch parameter $\xi\approx0.37$ (measured $0.376$ , Ku et al., Science 2012) and $T_c/T_F\approx0.17$ at unitarity; dimer–dimer length $a_{dd}\approx0.6\,a$ (Petrov, Salomon, Shlyapnikov 2004). No non-standard numbers introduced.

Prerequisites. Bose–Einstein Condensation in Dilute Gases · Feshbach Resonances · Evaporative Cooling · Dipole Force and the Optical Dipole Potential. Leads to. Feshbach Resonances · Bose–Einstein Condensation in Dilute Gases.

Ask the tutor: Ask me why the same BCS wavefunction describes both a cloud of molecules and a superfluid of overlapping Cooper pairs — and what makes the point in between universal.

Is a degenerate Fermi gas already a superfluid?

No. A degenerate but non-interacting Fermi gas is just a filled sea with a sharp surface — it has quantum statistics but no phase coherence and no superfluid flow. Superfluidity requires an attractive interaction between the two spin states so that atoms bind into pairs that share a single macroscopic wavefunction. This is why the Feshbach resonance is essential: it supplies and tunes the attraction. Cooling ⁴⁰K below $T_F$ in 1999 gave a degenerate gas; a paired, superfluid gas took the tunable interactions of 2003–2004. Degeneracy is necessary for pairing but not sufficient on its own.

What is the Bertsch parameter, physically?

It is the ratio of the ground-state energy of the unitary Fermi gas to that of the ideal (non-interacting) gas at the same density: $E_0=\xi\,\tfrac35 NE_F$ with $\xi\approx0.37$ . Because the scattering length is infinite at unitarity it cannot enter any formula, so the energy must be a pure number times the only scale available, $E_F$ — and $\xi$ is that number. Its smallness shows how strongly the attraction lowers the energy relative to the ideal gas. George Bertsch posed the problem it answers in 1999, and it is now one of the most precisely computed and measured numbers in cold-atom physics.

How is the pairing gap actually measured?

By radio-frequency spectroscopy. One drives a transition that flips a paired atom into a third, empty hyperfine state that does not feel the attraction. Breaking the pair to free that atom costs an extra energy set by the gap $\Delta$ , so the RF resonance is shifted from the bare atomic line by an amount that maps $\Delta$ . Sweeping the Feshbach field traces how pairing evolves: on the BEC side the shift reflects the molecular binding $E_b$ , on the BCS side the many-body gap. Developed in Jin's group, it is the fermionic analog of tunneling spectroscopy in superconductors.

If the pairs change so drastically, why call the crossover smooth?

Because the ground state never becomes singular. Across the crossover the pair size shrinks from far larger than the spacing (BCS) to far smaller (tight molecules, BEC), and the chemical potential slides from $+E_F$ through zero to $-E_b/2$ — but the order parameter $\Delta$ stays nonzero and the broken $U(1)$ symmetry never changes. A phase transition is marked by a non-analyticity in thermodynamic quantities; here there is none. The BCS variational wavefunction contains the molecular BEC as an exact limit — the mathematical statement that the two are one continuous family of the same state.

What does unitarity mean and why is it universal?

Unitarity is the point where the s-wave scattering length diverges, $|a|\to\infty$ , so the cross-section saturates the maximum quantum mechanics allows (the "unitarity limit"). With $a$ gone from the problem, the only scales left are the spacing $\sim1/k_F$ and the thermal wavelength, so every dimensionless thermodynamic ratio must be a pure number independent of the atomic species. That is what "universal" means: the equation of state of a unitary ⁶Li gas, a unitary ⁴⁰K gas, and dilute neutron matter are the same function, fixed by constants like $\xi$ and $T_c/T_F$ .

How cold is degenerate in real numbers?

For a typical trapped alkali Fermi gas the Fermi temperature is around a microkelvin — worked example 1 gives $T_F\approx0.81\ \mu$ K for $10^5$ ⁶Li atoms per spin in a 200 Hz trap. Degeneracy begins once $T$ drops below $T_F$ , and modern experiments push to $T/T_F\approx0.05$ , so the actual temperature is tens of nanokelvin. At unitarity superfluidity appears below $T_c\approx0.17\,T_F$ , about $0.14\ \mu$ K here — among the coldest temperatures ever produced, reached by combining evaporative and sympathetic cooling. The same strongly interacting physics recurs in the superfluid crust of a neutron star, which is why the unitary gas doubles as a tabletop analog for nuclear matter.