Classical and Quantum Statistics: Ideal Bose and Fermi Gases
CSIR NET statistical physics: Maxwell–Boltzmann, Bose–Einstein and Fermi–Dirac statistics, density of states, Fermi gas, Sommerfeld, Bose–Einstein condensation.
- 17 min
On this page · 14 sections
Prerequisites
- Canonical and Grand Canonical Ensembles: the grand partition function, from which all three distributions follow in two lines.
- Identical Particles, Pauli Principle and Spin-Statistics: which particles are bosons and which are fermions, and how to count their states.
- One-Dimensional Problems: Box, Oscillator and Tunnelling: particle-in-a-box levels, the basis of the density of states.
- The integral \( \int_0^\infty\dfrac{x^{s-1}}{e^x - 1}dx = \Gamma(s)\zeta(s) \), with \( \zeta(3/2) \approx 2.612 \), \( \zeta(5/2) \approx 1.341 \), \( \zeta(3) \approx 1.202 \).
The Hook: Electrons That Will Not Hold Heat
Three Statistics from One Formula
Treat each single-particle state \( k \) (energy \( \varepsilon_k \)) as a subsystem exchanging particles with the rest of the gas at chemical potential \( \mu \). Its grand partition function is \( \xi_k = \sum_ne^{-\beta n(\varepsilon_k - \mu)} \), with \( n \) restricted by the statistics:
| Statistics | Allowed \( n \) | \( \xi_k \) | Mean occupancy \( \bar n(\varepsilon) \) |
|---|---|---|---|
| Fermi–Dirac (fermions) | 0, 1 | \( 1 + e^{-\beta(\varepsilon - \mu)} \) | \( \dfrac{1}{e^{\beta(\varepsilon - \mu)} + 1} \) |
| Bose–Einstein (bosons) | 0, 1, 2, … | \( \dfrac{1}{1 - e^{-\beta(\varepsilon - \mu)}} \) | \( \dfrac{1}{e^{\beta(\varepsilon - \mu)} - 1} \) |
| Maxwell–Boltzmann (classical limit) | — | — | \( e^{-\beta(\varepsilon - \mu)} \) |
and \( \ln\Xi = \pm\sum_k\ln\left(1 \pm e^{-\beta(\varepsilon_k - \mu)}\right) \) (upper sign FD, lower BE). \( \mu \) is fixed by \( N = \sum_k\bar n(\varepsilon_k) \).
The same result by counting
The older route groups single-particle states into cells of \( g_i \) states near energy \( \varepsilon_i \) holding \( n_i \) particles, counts arrangements \( W \), and maximises \( \ln W \) at fixed \( N \) and \( E \) with Lagrange multipliers \( \alpha \), \( \beta \):
| Statistics | Arrangements \( W \) | Most probable \( n_i/g_i \) |
|---|---|---|
| MB (distinguishable, Boltzmann counting) | \( N!\prod_i\dfrac{g_i^{n_i}}{n_i!} \) | \( e^{-\alpha - \beta\varepsilon_i} \) |
| BE | \( \prod_i\dfrac{(n_i + g_i - 1)!}{n_i!(g_i - 1)!} \) | \( 1/(e^{\alpha + \beta\varepsilon_i} - 1) \) |
| FD | \( \prod_i\dfrac{g_i!}{n_i!(g_i - n_i)!} \) | \( 1/(e^{\alpha + \beta\varepsilon_i} + 1) \) |
with \( \alpha = -\mu/k_BT \) and \( \beta = 1/k_BT \).
Key features
- FD: \( 0 \leq \bar n \leq 1 \) (Pauli). At \( \varepsilon = \mu \), \( \bar n = 1/2 \) at any temperature. At \( T = 0 \) it is a step: 1 below \( \mu \), 0 above.
- BE: \( \bar n \geq 0 \) requires \( \mu < \varepsilon_0 \), the lowest single-particle energy (\( \mu \leq 0 \) for a free gas with \( \varepsilon_0 = 0 \)). \( \bar n \) diverges as \( \varepsilon \to \mu \).
- Photons and phonons: number not conserved, so \( \mu = 0 \) and \( \bar n = 1/(e^{\beta\hbar\omega} - 1) \), the Planck distribution.
The classical limit
All three agree when \( e^{\beta(\varepsilon - \mu)} \gg 1 \) for every state, i.e. \( \bar n \ll 1 \): states are sparsely occupied and the \( \pm1 \) is negligible. Since \( e^{\beta\mu} \approx n\lambda^3/g_s \) in that limit, the condition is
$$\boxed{n\lambda^3 \ll 1, \qquad \lambda = \frac{h}{\sqrt{2\pi mk_BT}}}$$the mean spacing between particles much larger than their thermal de Broglie wavelength. Quantum effects appear at high density, low temperature and low mass. The first correction to the equation of state is
$$P = nk_BT\left[1 \pm \frac{n\lambda^3}{2^{5/2}g_s} + \cdots\right]$$with + for fermions (an effective repulsion from the Pauli principle) and − for bosons (an effective attraction).
Density of States
For free particles in a box of volume \( V \) (\( L^d \) in \( d \) dimensions) with spin degeneracy \( g_s \), counting wavevectors on a grid of spacing \( 2\pi/L \):
| Dimension | \( g(\varepsilon) \) (states per unit energy) | Dependence |
|---|---|---|
| 3D | \( \dfrac{g_sV}{4\pi^2}\left(\dfrac{2m}{\hbar^2}\right)^{3/2}\varepsilon^{1/2} \) | \( \propto\sqrt\varepsilon \) |
| 2D | \( \dfrac{g_sAm}{2\pi\hbar^2} \) | constant |
| 1D | \( \dfrac{g_sL}{2\pi\hbar}\sqrt{\dfrac{2m}{\varepsilon}} \) | \( \propto\varepsilon^{-1/2} \) |
| 3D, massless (\( \varepsilon = \hbar ck \)) | \( \dfrac{g_sV\varepsilon^2}{2\pi^2\hbar^3c^3} \) | \( \propto\varepsilon^2 \) |
In \( d \) dimensions with \( \varepsilon \propto k^s \), \( g(\varepsilon) \propto \varepsilon^{d/s - 1} \). The dimension and the dispersion relation decide almost every qualitative result below, including whether BEC happens.
A useful general result: for any ideal gas (MB, BE or FD) with \( \varepsilon \propto k^s \) in \( d \) dimensions,
$$\boxed{PV = \frac{s}{d}U}$$so \( PV = \tfrac{2}{3}U \) for non-relativistic particles in 3D and \( PV = \tfrac{1}{3}U \) for photons or ultra-relativistic particles.
The Ideal Fermi Gas
At absolute zero
All states up to the Fermi energy \( E_F \) are filled, one particle per state. In 3D with \( g_s = 2 \):
$$N = 2\times\frac{V}{(2\pi)^3}\times\frac{4}{3}\pi k_F^3 \Rightarrow \boxed{k_F = (3\pi^2n)^{1/3}, \qquad E_F = \frac{\hbar^2}{2m}(3\pi^2n)^{2/3}}$$| Ground-state property (3D, spin 1/2) | Value |
|---|---|
| Mean energy per particle | \( \tfrac{3}{5}E_F \) |
| Total energy | \( U_0 = \tfrac{3}{5}NE_F \) |
| Degeneracy pressure | \( P_0 = \tfrac{2}{3}U_0/V = \tfrac{2}{5}nE_F \) |
| Density of states at \( E_F \) | \( g(E_F) = \dfrac{3N}{2E_F} \) |
| Fermi temperature | \( T_F = E_F/k_B \) |
Other dimensions (spin 1/2): 2D, \( E_F = \pi\hbar^2n/m \) (\( n \) per area); 1D, \( E_F = \pi^2\hbar^2n^2/8m \) (\( n \) per length).
At low temperature: the Sommerfeld expansion
For \( k_BT \ll E_F \), only electrons within about \( k_BT \) of \( E_F \) can be excited. The Sommerfeld expansion gives
$$\mu(T) \approx E_F\left[1 - \frac{\pi^2}{12}\left(\frac{k_BT}{E_F}\right)^2\right], \qquad U \approx \frac{3}{5}NE_F\left[1 + \frac{5\pi^2}{12}\left(\frac{k_BT}{E_F}\right)^2\right]$$$$\boxed{C_v^{\text{el}} = \frac{\pi^2}{2}Nk_B\frac{T}{T_F} = \frac{\pi^2}{3}g(E_F)k_B^2T}$$A crude estimate gives the same physics: a fraction \( T/T_F \) of electrons each gain about \( k_BT \), so \( U - U_0 \sim N(k_BT)^2/E_F \) and \( C \propto T \).
The low-temperature heat capacity of a metal is therefore \( C = \gamma T + AT^3 \) (electrons plus Debye phonons). Plotting \( C/T \) against \( T^2 \) gives a straight line with intercept \( \gamma \) and slope \( A \).
The same idea explains Pauli paramagnetism: only electrons near \( E_F \) can flip their spins in a field, so the spin susceptibility is about \( \mu_0\mu_B^2g(E_F)/V \), nearly independent of temperature, instead of the Curie \( 1/T \) law.
Where degenerate Fermi gases matter
Conduction electrons in metals; electrons in white dwarfs (degeneracy pressure supports the star; when the electrons become relativistic the support weakens, giving the Chandrasekhar mass limit of about 1.4 solar masses); neutrons in neutron stars; liquid \( ^3 \)He; nucleons in nuclei.
The Ideal Bose Gas and Bose–Einstein Condensation
Why condensation happens
For spinless bosons in 3D with \( \varepsilon_0 = 0 \), \( \mu \leq 0 \). The number in excited states is
$$N_{\text{ex}} = \int_0^\infty\frac{g(\varepsilon)\,d\varepsilon}{e^{\beta(\varepsilon - \mu)} - 1} \leq \int_0^\infty\frac{g(\varepsilon)\,d\varepsilon}{e^{\beta\varepsilon} - 1} = \frac{V}{\lambda^3}\zeta(3/2)$$The excited states can hold at most \( 2.612\,V/\lambda^3 \) particles. When \( N \) exceeds this, \( \mu \) sticks at (essentially) zero and the surplus goes into the single ground state, giving it a macroscopic occupation. The transition temperature satisfies \( n\lambda_c^3 = \zeta(3/2) \):
$$\boxed{T_c = \frac{2\pi\hbar^2}{mk_B}\left(\frac{n}{\zeta(3/2)}\right)^{2/3}, \qquad \frac{N_0}{N} = 1 - \left(\frac{T}{T_c}\right)^{3/2}\ (T < T_c)}$$Thermodynamics below \( T_c \)
With \( \mu = 0 \), the condensate carries no energy or pressure, and
$$U = \frac{3}{2}k_BT\frac{V}{\lambda^3}\zeta(5/2) \propto T^{5/2}, \qquad P = \frac{k_BT}{\lambda^3}\zeta(5/2) \propto T^{5/2}$$- \( P \) is independent of density: compressing the gas just moves particles into the condensate. Isotherms are flat, like a first-order transition in the \( P \)–\( v \) plane.
- \( C_v \propto T^{3/2} \) below \( T_c \), rising to a cusp of \( \tfrac{15}{4}\dfrac{\zeta(5/2)}{\zeta(3/2)}Nk_B \approx 1.93\,Nk_B \) at \( T_c \), then falling to the classical \( \tfrac{3}{2}Nk_B \) at high \( T \). \( C_v \) is continuous at \( T_c \) but its slope jumps.
Dimension and trap dependence
| Situation | BEC at \( T > 0 \)? | Reason |
|---|---|---|
| Free bosons, 3D | yes | \( \int\sqrt\varepsilon\,d\varepsilon/(e^{\beta\varepsilon} - 1) \) converges |
| Free bosons, 2D | no | \( g \) constant; \( \int d\varepsilon/(e^{\beta\varepsilon} - 1) \) diverges at 0 |
| Free bosons, 1D | no | integrand even more divergent |
| Harmonic trap, 3D | yes | \( g \propto \varepsilon^2 \); \( k_BT_c = \hbar\bar\omega(N/\zeta(3))^{1/3} \) |
| Photons | no condensation | number not conserved, \( \mu = 0 \) fixed |
Dilute alkali-atom gases in magnetic or optical traps, cooled by laser and evaporative cooling, were the first clean realisations of BEC; their transition temperatures lie in the sub-microkelvin range.
Comparison at a Glance
| Property | MB | BE | FD |
|---|---|---|---|
| Occupancy limit | none | none | 1 per state |
| Allowed \( \mu \) (free gas) | any (very negative) | \( \mu \leq 0 \) | any; \( \mu \to E_F > 0 \) as \( T \to 0 \) |
| \( T = 0 \) state | — | all in ground state | Fermi sea up to \( E_F \) |
| \( C_v \) at low \( T \) (3D) | \( \tfrac{3}{2}Nk_B \) (violates third law) | \( \propto T^{3/2} \) | \( \propto T \) |
| Pressure vs classical at same \( n, T \) | — | lower | higher |
| Classical limit | always | \( n\lambda^3 \ll 1 \) | \( n\lambda^3 \ll 1 \) |
Memory Tricks
Don’t Confuse
| This | vs | That |
|---|---|---|
| Fermi energy \( E_F \): \( \mu \) at \( T = 0 \) | Fermi level / \( \mu(T) \): slightly below \( E_F \) at \( T > 0 \) (3D) | |
| \( \bar n_{FD}(\mu) = 1/2 \) at every \( T \) | \( \bar n_{BE} \) diverges as \( \varepsilon \to \mu \) | |
| BEC: condensation in momentum space, no latent heat in the usual sense | Liquefaction: condensation in real space | |
| Electron \( C_v \propto T \) | Phonon \( C_v \propto T^3 \) (Debye) | |
| \( PV = \tfrac{2}{3}U \) (non-relativistic) | \( PV = \tfrac{1}{3}U \) (photons, ultra-relativistic) | |
| Photon gas: \( \mu = 0 \), no BEC | Massive bosons: \( \mu \to 0^- \) at \( T_c \), BEC in 3D |
Common Mistakes
Practice
All questions below are original practice questions written for this page. They are not from any CSIR NET paper. Use \( \hbar = 1.055\times10^{-34} \) J s, \( m_e = 9.11\times10^{-31} \) kg, \( k_B = 1.381\times10^{-23} \) J K\( ^{-1} \).
Level 1: Foundation
- What is the Fermi–Dirac occupancy of a state with energy equal to the chemical potential?
- What is the chemical potential of a photon gas?
- By what factor does the Bose–Einstein condensation temperature of an ideal 3D gas change when its density is doubled?
Answers
- 1/2, at any temperature.
- Zero. Photon number is not conserved.
- \( 2^{2/3} \approx 1.59 \). \( T_c \propto n^{2/3} \).
Level 2: Part B Style
Practice Q1. A two-dimensional gas of free electrons (spin 1/2) has areal density \( 1.0\times10^{16} \) m\( ^{-2} \). Its Fermi energy is closest to:
(a) 1.2 meV (b) 2.4 meV (c) 4.8 meV (d) 24 meV
Answer: (b) 2.4 meV
\( E_F = \pi\hbar^2n/m = \pi\times(1.055\times10^{-34})^2\times10^{16}/(9.11\times10^{-31}) \approx 3.84\times10^{-22} \) J \( \approx 2.4 \) meV. (c) is the spinless (or spin-polarised) value \( 2\pi\hbar^2n/m \); (a) halves the correct answer.
Practice Q2. An ideal Bose gas is held at \( T = T_c/2 \). The fraction of particles in the ground state is closest to:
(a) 0.29 (b) 0.50 (c) 0.65 (d) 0.75
Answer: (c) 0.65
\( N_0/N = 1 - (1/2)^{3/2} = 1 - 0.354 = 0.646 \). (d) uses the exponent 2; (b) uses exponent 1.
Practice Q3. For an ideal Fermi gas of \( N \) spin-1/2 particles in 3D at \( T = 0 \), the pressure in terms of the number density \( n \) and Fermi energy \( E_F \) is:
(a) \( \tfrac{3}{5}nE_F \) (b) \( \tfrac{2}{5}nE_F \) (c) \( \tfrac{2}{3}nE_F \) (d) 0
Answer: (b) 2/5 n E_F
\( U_0 = \tfrac{3}{5}NE_F \) and \( PV = \tfrac{2}{3}U \), so \( P = \tfrac{2}{3}\times\tfrac{3}{5}nE_F = \tfrac{2}{5}nE_F \). (a) is the energy density; (d) is the classical answer at \( T = 0 \).
Practice Q4. For non-interacting particles in a three-dimensional harmonic trap, the single-particle density of states varies with energy as:
(a) \( \varepsilon^{1/2} \) (b) \( \varepsilon \) (c) \( \varepsilon^2 \) (d) constant
Answer: (c) ε²
Levels \( \hbar\omega(n_x + n_y + n_z) \): the number of states below \( \varepsilon \) is the number of non-negative integer triples with sum \( \leq \varepsilon/\hbar\omega \), about \( (\varepsilon/\hbar\omega)^3/6 \). Differentiating, \( g(\varepsilon) = \varepsilon^2/(2(\hbar\omega)^3) \). This is why trapped bosons condense even where free 2D bosons do not.
Level 3: Part C Style
Practice Q5. A monovalent metal has conduction-electron density \( 8.5\times10^{28} \) m\( ^{-3} \). Treat the electrons as a free Fermi gas. Consider:
- The Fermi energy is about 7.0 eV.
- The Fermi temperature is about \( 8\times10^4 \) K.
- At 300 K the electronic heat capacity satisfies \( C_{\text{el}}/(Nk_B) \approx 0.018 \).
- At 300 K the electronic heat capacity is comparable to the lattice value \( 3Nk_B \).
Which are correct?
(a) 1 and 2 only (b) 1, 2 and 3 only (c) 2 and 4 only (d) 1, 2, 3 and 4
Answer: (b) 1, 2 and 3 only
- \( E_F = \frac{\hbar^2}{2m}(3\pi^2n)^{2/3} \approx 1.13\times10^{-18} \) J \( = 7.05 \) eV. True.
- \( T_F = E_F/k_B \approx 8.2\times10^4 \) K. True.
- \( C_{\text{el}}/Nk_B = \frac{\pi^2}{2}\frac{T}{T_F} = 4.93\times300/81800 \approx 0.018 \). True.
- Lattice \( 3Nk_B \) (one electron per atom) is about 170 times larger. False.
Practice Q6. Two non-interacting identical particles occupy non-degenerate single-particle levels 0, \( \varepsilon \), \( 2\varepsilon \). Write \( x = e^{-\varepsilon/k_BT} \). Consider:
- For spinless fermions, \( Z_F = x + x^2 + x^3 \).
- For spinless bosons, \( Z_B = 1 + x + 2x^2 + x^3 + x^4 \).
- \( Z_B + Z_F = (1 + x + x^2)^2 \), so the corrected classical result \( (1 + x + x^2)^2/2! \) is their average.
- At \( k_BT = \varepsilon \), the probability that both bosons are in the lowest level is about 0.44.
Which are correct?
(a) 1 and 2 only (b) 1, 2 and 3 only (c) 2 and 4 only (d) 1, 2, 3 and 4
Answer: (b) 1, 2 and 3 only
Fermion pairs: \( (0,\varepsilon) \), \( (0,2\varepsilon) \), \( (\varepsilon,2\varepsilon) \): energies \( \varepsilon, 2\varepsilon, 3\varepsilon \). True.
Bosons add the doubly occupied states \( 0, 2\varepsilon, 4\varepsilon \): \( 1 + x + 2x^2 + x^3 + x^4 \). True.
Sum \( = 1 + 2x + 3x^2 + 2x^3 + x^4 = (1 + x + x^2)^2 \). True.
\( x = e^{-1} \): \( Z_B = 1 + 0.368 + 0.271 + 0.050 + 0.018 = 1.707 \), so \( P = 1/1.707 \approx 0.59 \). The value 0.44 is \( 1/(1 + x + x^2)^2 \), the distinguishable-particle answer. False.
Practice Q7. Consider an ideal gas of spinless bosons of mass \( m \) at number density \( n \) in three dimensions. Consider:
- Below \( T_c \), the pressure depends on temperature but not on density.
- The heat capacity at constant volume is about \( 1.93\,Nk_B \) at \( T_c \), larger than its classical value.
- Above \( T_c \), at the same \( n \) and \( T \), the pressure is slightly lower than \( nk_BT \).
- If the same particles were confined to a plane (free 2D gas), they would condense at a lower but finite temperature.
Which are correct?
(a) 1 and 2 only (b) 1, 2 and 3 only (c) 3 and 4 only (d) 1, 2, 3 and 4
Answer: (b) 1, 2 and 3 only
- \( P = k_BT\zeta(5/2)/\lambda^3 \): no \( n \). True.
- \( \tfrac{15}{4}\zeta(5/2)/\zeta(3/2) = 1.926 \), above \( \tfrac{3}{2} \). True.
- First quantum correction for bosons is negative: \( P = nk_BT(1 - n\lambda^3/2^{5/2} + \cdots) \). True.
- A free 2D Bose gas has no BEC at \( T > 0 \). False.
Quick Revision Table
| Idea | Formula or fact |
|---|---|
| Occupancy | \( 1/(e^{\beta(\varepsilon - \mu)} \pm 1) \); MB: \( e^{-\beta(\varepsilon - \mu)} \) |
| Classical limit | \( n\lambda^3 \ll 1 \) |
| First correction | \( P = nk_BT[1 \pm n\lambda^3/(2^{5/2}g_s)] \), + FD, − BE |
| 3D density of states | \( \propto V\sqrt\varepsilon \); 2D constant; 1D \( \propto\varepsilon^{-1/2} \) |
| \( PV \) | \( \tfrac{2}{3}U \) (non-relativistic), \( \tfrac{1}{3}U \) (photons) |
| Fermi energy (3D) | \( \frac{\hbar^2}{2m}(3\pi^2n)^{2/3} \); 2D \( \pi\hbar^2n/m \); 1D \( \pi^2\hbar^2n^2/8m \) |
| Fermi sea | \( U_0 = \tfrac{3}{5}NE_F \), \( P_0 = \tfrac{2}{5}nE_F \), \( g(E_F) = 3N/2E_F \) |
| Electron \( C_v \) | \( \tfrac{\pi^2}{2}Nk_BT/T_F \); metal \( C = \gamma T + AT^3 \) |
| BEC | \( n\lambda_c^3 = 2.612 \); \( N_0/N = 1 - (T/T_c)^{3/2} \) |
| Below \( T_c \) | \( P, U \propto T^{5/2} \); \( C_v \propto T^{3/2} \); cusp \( 1.93Nk_B \) |
| 2D free bosons | no BEC; 3D trap: \( k_BT_c = \hbar\bar\omega(N/1.202)^{1/3} \) |
Key Takeaways
What’s Next
- Blackbody Radiation, Planck’s Law and Detailed Balance: the Bose gas with \( \mu = 0 \), and the principle that links emission to absorption.
- Canonical and Grand Canonical Ensembles, to revisit the grand partition function.
- Identical Particles, Pauli Principle and Spin-Statistics, for few-particle ground states and state counting.
- Thermodynamics and Statistical Physics: Quick Revision.