Grand Potential

Every closed system in statistical mechanics has its master function — the canonical partition function $Z$ , whose logarithm gives the Helmholtz free energy $F=-k_BT\ln Z$ and, through derivatives, all of thermodynamics. But an enormous class of real systems is open: gas adsorbed on a surface trades molecules with the vapor above it, electrons in a metal wire flow in and out, a droplet exchanges particles with its own vapor. For these, the controlling variables are not $(T,V,N)$ but $(T,V,\mu)$ — temperature, volume, and the reservoir's chemical potential. The grand potential $\Omega=-k_BT\ln\Xi$ is their master function: the bridge from the grand partition function $\Xi$ straight to thermodynamics. Its power comes from one differential. Legendre-transforming the energy to trade away both the entropy (for $T$ ) and the particle number (for $\mu$ ) gives $\mathrm d\Omega=-S\,\mathrm dT-p\,\mathrm dV-N\,\mathrm d\mu$ , so entropy, pressure, and mean particle number each drop out as a single derivative of $\Omega$ . Better still, because $\Omega$ is extensive yet its only extensive natural variable is $V$ , Euler's theorem forces the astonishingly compact identity $\Omega=-pV$ . Computing $\ln\Xi$ is computing the equation of state — for a classical gas, a Fermi gas, a photon gas, all at once. This article derives the grand potential from the grand-ensemble probabilities, proves the owned result $\Omega=-pV$ from first-order homogeneity in $V$ , extracts $S$ , $p$ , and $N$ as its derivatives, and shows how $N=-\partial\Omega/\partial\mu$ reproduces the Fermi–Dirac and Bose–Einstein occupation numbers on the ideal quantum gas. It closes with the number-fluctuation relation that links $\Omega$ to light scattering. Every relation here is python-verified.

Key takeaways

The grand potential $\Omega(T,V,\mu)=-k_BT\ln\Xi$ is the master function of the grand canonical ensemble — the open-system analogue of $F=-k_BT\ln Z$ .
Its differential $\mathrm d\Omega=-S\,\mathrm dT-p\,\mathrm dV-N\,\mathrm d\mu$ makes $(T,V,\mu)$ the natural variables and gives $S=-\partial_T\Omega$ , $p=-\partial_V\Omega$ , $N=-\partial_\mu\Omega$ mechanically.
Owned result: because $\Omega$ is extensive with only one extensive natural variable $V$ , Euler's theorem forces $\Omega=-pV$ — so $pV=k_BT\ln\Xi$ is the equation of state directly.
It relates to Helmholtz by a Legendre transform, $\Omega=F-\mu N$ , and is minimized at equilibrium for a system held at fixed $T,V,\mu$ .
For ideal quantum gases $\Omega=\mp k_BT\sum_k\ln(1\mp z\,e^{-\beta\varepsilon_k})$ ; one derivative $N=-\partial_\mu\Omega$ delivers the Fermi–Dirac and Bose–Einstein occupations.
Key idea: $\Omega=-k_BT\ln\Xi$ is the double Legendre transform of the energy (trading $S\to T$ , $N\to\mu$ ); its natural variables are $(T,V,\mu)$ , extensivity collapses it to $\Omega=-pV$ , and $\ln\Xi$ thereby generates the whole equation of state.

The core insight

The grand potential $\Omega=-k_BT\ln\Xi$ is the double Legendre transform of the internal energy, trading entropy for temperature and particle number for chemical potential. Its natural variables are $(T,V,\mu)$ ; its first derivatives give $S$ , $p$ , and $\langle N\rangle$ ; and because it is extensive in the single extensive variable $V$ , Euler's theorem collapses it to $\Omega=-pV$ — making $\ln\Xi$ a direct generator of the equation of state and of open-system fluctuations.

Where this lives today

The grand potential is the everyday tool of quantum many-body physics because it turns "count the states" into "write down the pressure." Every ideal-quantum-gas result — the electronic heat capacity of metals via the Sommerfeld expansion, the critical temperature of Bose–Einstein condensation, the $T^4$ energy density of blackbody radiation, the degeneracy pressure that holds up a white dwarf — is a differentiation of $\Omega=\mp k_BT\sum_k\ln(1\mp z\,e^{-\beta\varepsilon_k})$ . Because the modes of a non-interacting gas are independent, this single sum factorizes, and one grand-potential calculation covers fermions and bosons at once by a sign. Beyond textbook gases, the grand potential is the central object of density-functional theory (DFT), where the ground-state energy of electrons in a solid is obtained by minimizing a grand-potential functional of the electron density at fixed chemical potential — the computational backbone of modern materials science and quantum chemistry. In statistical field theory and lattice QCD, the "grand potential" at finite chemical potential governs the phase diagram of dense nuclear matter and the physics of neutron-star interiors. And the number-fluctuation relation $\langle(\Delta N)^2\rangle=k_BT\,\partial N/\partial\mu$ that follows from the second $\mu$ -derivative of $\Omega$ is exactly what light-scattering and small-angle-scattering experiments measure: as a fluid approaches its critical point, $\kappa_T$ diverges, density fluctuations grow to optical wavelengths, and the fluid turns milky — critical opalescence, all encoded in the curvature of the grand potential.

The physical picture

Picture the system wrapped inside an enormous bath that supplies or absorbs both heat and particles at no cost to itself. The bath fixes two "prices": temperature $T$ sets the price of energy, and chemical potential $\mu$ sets the price of a particle. Held at fixed $T$ , $V$ , and $\mu$ , the system settles into the state that minimizes a cost function — and that cost function is the grand potential $\Omega$ . Where the Helmholtz free energy $F$ is the right potential when the particle number is fixed, $\Omega$ is the right potential when the particle number is free to fluctuate and only its price $\mu$ is fixed. The cleanest way to feel what $\Omega$ is: it is $F$ with the particles "bought out" by the reservoir. The relation $\Omega=F-\mu N$ subtracts the chemical energy $\mu N$ that the bath has effectively paid for. Once that bookkeeping is absorbed into the reservoir, what remains is purely mechanical — and indeed $\Omega=-pV$ is nothing but the $pV$ work the bulk fluid can do. A homogeneous fluid at fixed $T$ and $\mu$ has no thermodynamic content beyond its pressure, because there is only one extensive variable ( $V$ ) left to carry information.

Figure

The four thermodynamic potentials as Legendre transforms of the internal energy U(S,V,N). Trading S→T gives F; trading N→μ as well gives Ω, whose all-intensive prices leave only V extensive — forcing Ω = −pV.

The single tidy identity $\Omega=-pV$ is the punchline of the whole grand-ensemble machinery. It says that to find the equation of state of any system — dilute gas, dense liquid, electron gas in a metal, photons in a cavity — you compute one logarithm, $\ln\Xi$ , and read off $pV=k_BT\ln\Xi$ . The next figure shows why $\Omega$ is the natural coordinate for the particle number: $\Omega$ as a function of $\mu$ is a smooth curve whose slope is exactly $-\langle N\rangle$ , so raising the reservoir's chemical potential (making particles "cheaper") makes the system fill up, and the filling rate is the curvature — which is the number fluctuation.

Figure

Ω versus μ at fixed T,V: a smooth curve whose slope is −⟨N⟩ (the system fills as μ rises) and whose curvature is the number fluctuation ⟨ΔN²⟩/kT. Near a critical point the curvature blows up — opalescence.

The core equations

The grand potential is defined from the grand partition function exactly as the Helmholtz free energy is defined from $Z$ . This is the bridge equation: compute the microscopic sum $\Xi=\sum_N e^{\beta\mu N}Z_N$ over all particle numbers and states, take its logarithm, and you have a full thermodynamic potential.

\Omega(T,V,\mu)=-k_BT\ln\Xi(T,V,\mu),\qquad \Xi=\sum_N e^{\beta\mu N}Z_N=\mathrm{Tr}\,e^{-\beta(\hat H-\mu\hat N)} \qquad (1)

where: $\Omega$ — grand potential (also called the Landau free energy or Landau grand potential); $\Xi$ — grand partition function; $\beta=1/k_BT$ ; $\mu$ — reservoir chemical potential; $\hat N$ — particle-number operator. $\Omega$ carries the entire equilibrium content of the open system in one function of $(T,V,\mu)$ . The differential of $\Omega$ tells you which variables it "wants" and what each derivative returns. It comes from Legendre-transforming the first law to trade away both $S$ and $N$ :

\mathrm d\Omega=-S\,\mathrm dT-p\,\mathrm dV-N\,\mathrm d\mu \qquad (2)

where: $S$ — entropy; $p$ — pressure; $N=\langle N\rangle$ — mean particle number. The three natural variables are $(T,V,\mu)$ ; each carries a minus sign because $\Omega$ is a doubly-transformed lower potential. Reading equation (2) off directly gives the three observables as single first derivatives — the practical payoff of the whole formalism:

S=-\left(\frac{\partial\Omega}{\partial T}\right)_{V,\mu},\qquad p=-\left(\frac{\partial\Omega}{\partial V}\right)_{T,\mu},\qquad N=-\left(\frac{\partial\Omega}{\partial\mu}\right)_{T,V} \qquad (3)

where: each partial holds the other two natural variables fixed. Once $\ln\Xi$ is known, entropy, equation of state, and particle number follow mechanically — no re-derivation needed. Extensivity then collapses $\Omega$ to a single term. Because $\Omega(T,V,\mu)$ is extensive but $T$ and $\mu$ are intensive, its only extensive natural variable is $V$ , so $\Omega$ must be strictly proportional to $V$ :

\Omega=-pV,\qquad \text{equivalently}\qquad pV=k_BT\ln\Xi \qquad (4)

where: $p$ — bulk pressure. This holds for any homogeneous bulk system; it fails only when surface or finite-size terms are non-negligible. It is the most direct route to any equation of state. Finally, $\Omega$ connects back to the Helmholtz free energy by a single Legendre step, and the second derivative gives the number fluctuation — the open-system fluctuation–response relation:

\Omega=F-\mu N,\qquad \langle(\Delta N)^2\rangle=k_BT\left(\frac{\partial\langle N\rangle}{\partial\mu}\right)_{T,V}=\frac{\langle N\rangle^2 k_BT}{V}\,\kappa_T \qquad (5)

where: $F=-k_BT\ln Z$ ; $\kappa_T=-\frac1V(\partial V/\partial p)_{T,N}$ — isothermal compressibility. The relative fluctuation $\langle(\Delta N)^2\rangle/\langle N\rangle^2=k_BT\kappa_T/V\sim1/N$ away from criticality but diverges as $\kappa_T\to\infty$ at a critical point.

Derivation AThe grand potential is minus kT log Xi — the statistical-thermodynamic bridge

The goal is to prove that the statistical object $-k_BT\ln\Xi$ , built from the grand-ensemble probabilities, is literally the thermodynamic $\Omega=U-TS-\mu N$ — not merely analogous to it. The strategy is to compute the Gibbs entropy $S=-k_B\sum_i p_i\ln p_i$ of the grand-canonical distribution, substitute the explicit probabilities, and watch $\ln\Xi$ appear as the free-energy term. This is the grand-ensemble twin of the canonical result $F=-k_BT\ln Z$ , and it is what justifies calling $\Omega$ the master function.

Step 1. Write the Gibbs entropy of the grand-canonical distribution: $S=-k_B\sum_i p_i\ln p_i$ , where the sum runs over all microstates $i$ (each with energy $E_i$ and particle number $N_i$ ) and $p_i=\Xi^{-1}e^{-\beta(E_i-\mu N_i)}$ . Why: the Gibbs entropy is the correct information-theoretic entropy of any ensemble, and it equals the thermodynamic entropy at equilibrium.

Step 2. Take the logarithm of the probability: $\ln p_i=-\beta(E_i-\mu N_i)-\ln\Xi$ . Why: turning the exponential weight into an additive expression separates the "energy" part from the constant normalization $\ln\Xi$ .

Step 3. Average, using $\sum_i p_i=1$ , $\sum_i p_i E_i=\langle E\rangle$ , $\sum_i p_i N_i=\langle N\rangle$ : $S=-k_B\sum_i p_i[-\beta(E_i-\mu N_i)-\ln\Xi]=k_B\beta(\langle E\rangle-\mu\langle N\rangle)+k_B\ln\Xi$ . Why: the ensemble average is linear, so each term is replaced by its mean; the normalization constant $\ln\Xi$ pulls out untouched.

Step 4. Solve for $\ln\Xi$ and multiply by $k_BT$ : $k_BT\ln\Xi=TS-\langle E\rangle+\mu\langle N\rangle$ , hence $-k_BT\ln\Xi=\langle E\rangle-TS-\mu\langle N\rangle$ . Why: rearranging isolates $-k_BT\ln\Xi$ on one side, and the right-hand side is a combination of the mean energy, entropy, and particle number.

Step 5. Identify the right-hand side as the thermodynamic grand potential: $\Omega=U-TS-\mu N$ with $U=\langle E\rangle$ , $N=\langle N\rangle$ . Therefore $\Omega=-k_BT\ln\Xi$ exactly. Why: $U-TS-\mu N$ is the definition of $\Omega$ from the double Legendre transform, so the statistical and thermodynamic objects coincide — the bridge is proven. *Verified: the identity $-k_BT\ln\Xi=\langle E\rangle-TS-\mu\langle N\rangle$ is confirmed symbolically, and $\Omega=F-\mu N=U-TS-\mu N$ closes, in verify_aq-sm-grand-potential.py.*

Go deeper · Derivation B · 1 minOmega equals minus pV from first-order homogeneity — the owned result

Derivation BOmega equals minus pV from first-order homogeneity — the owned result

The goal is the compact identity $\Omega=-pV$ , and the point is to derive it not by quoting the Euler relation as a black box but by using the single structural fact that makes it inevitable: $\Omega$ is extensive, yet among its natural variables $(T,V,\mu)$ only $V$ is extensive. A function that is degree-one homogeneous in exactly one of its arguments is forced by Euler's theorem to equal that argument times its own derivative — and here that derivative is $-p$ . This is why a bulk fluid at fixed $T$ and $\mu$ carries no thermodynamic information beyond its pressure.

Step 1. State the homogeneity. Doubling the system ( $V\to\lambda V$ ) at fixed intensive $T,\mu$ doubles $\Omega$ : $\Omega(T,\lambda V,\mu)=\lambda\,\Omega(T,V,\mu)$ . So $\Omega$ is a degree-one homogeneous function of $V$ alone. Why: extensive quantities scale linearly with system size, and here $V$ is the only size-carrying (extensive) variable, since $T$ and $\mu$ are intensive and stay fixed under the doubling.

Step 2. Apply Euler's theorem for a degree-one homogeneous function of $V$ : $V\,(\partial\Omega/\partial V)_{T,\mu}=\Omega$ . Why: Euler's theorem states that for $f(\lambda V)=\lambda f(V)$ one has $V\,f'(V)=f(V)$ ; differentiate $\Omega(T,\lambda V,\mu)=\lambda\Omega$ with respect to $\lambda$ and set $\lambda=1$ .

Step 3. Insert the pressure. From the differential (2), $(\partial\Omega/\partial V)_{T,\mu}=-p$ , so $V(-p)=\Omega$ , i.e. $\Omega=-pV$ . Why: the volume-derivative of $\Omega$ is the (negative) pressure by the fundamental differential, so Euler's identity turns directly into $\Omega=-pV$ .

Step 4. Read off the equation of state. Since $\Omega=-k_BT\ln\Xi$ , we get $pV=k_BT\ln\Xi$ . For the classical ideal gas $\ln\Xi=zV/\lambda_T^3$ , giving $pV=k_BT\,zV/\lambda_T^3=N k_BT$ once $N=zV/\lambda_T^3$ — the ideal-gas law falls out. Why: substituting the microscopic $\ln\Xi$ into $pV=k_BT\ln\Xi$ and using $N=-\partial_\mu\Omega=zV/\lambda_T^3$ recovers $pV=Nk_BT$ , confirming the machinery end-to-end. *Verified: Euler's identity $V\,\partial_V\Omega=\Omega$ for the degree-1 form, $\Omega=-pV$ , and the ideal-gas $pV=Nk_BT$ are all checked symbolically in verify_aq-sm-grand-potential.py.*

Limiting cases

The grand potential reduces to, or connects with, simpler descriptions in the standard regimes:

Case	Result	Meaning
Classical ideal gas	$\Omega=-k_BT\,z\,V/\lambda_T^3$ , $z=e^{\beta\mu}$	$pV=-\Omega=Nk_BT$ recovers the ideal-gas law; $N=zV/\lambda_T^3$ .
Quantum ideal gases	$\Omega=\mp k_BT\sum_k\ln(1\mp z\,e^{-\beta\varepsilon_k})$	Upper sign bosons, lower sign fermions; one master formula for BE and FD thermodynamics.
Photon gas	$\mu=0$ , $\Omega=-\tfrac13 a V T^4=-pV$	Radiation pressure $p=\tfrac13 u$ ; $\Omega$ delivers the Stefan–Boltzmann equation of state.
Closed-system limit	$\Omega=F-\mu N$ , invert $N(\mu)$	Recovers the Helmholtz description; ensembles equivalent as $N\to\infty$ .
Bulk extensive system	$\Omega=-pV$	Exact when surface and finite-size terms are negligible (Euler relation holds).

Common trapThe identity Ω = −pV holds for any system, and dΩ has no −N dμ term

Two traps live here. First, $\Omega=-pV$ is not universal: it relies on bulk extensivity through the Euler relation, and it fails for small clusters, surface-dominated systems, or inhomogeneous ones where finite-size and surface contributions matter — there $\Omega$ still equals $-k_BT\ln\Xi$ , but it is no longer simply $-pV$ . Second, when writing the differential you must keep the full $\mathrm d\Omega=-S\,\mathrm dT-p\,\mathrm dV-N\,\mathrm d\mu$ ; dropping the $-N\,\mathrm d\mu$ term (a common slip when copying the Helmholtz differential) loses the entire particle-number physics and gives the wrong entropy and pressure. Remember: $\Omega$ has three natural variables, all three appear in $\mathrm d\Omega$ , and each derivative carries a minus sign because $\Omega$ is a doubly-transformed potential.

Worked examples

1 · The equation of state of a classical ideal gas, straight from Ω (a real number)

One mole of an ideal gas is held at $T=300\ \mathrm K$ in a volume $V$ . Use $\Omega=-pV=-k_BT\ln\Xi$ with $\ln\Xi=zV/\lambda_T^3$ to find the pressure, and determine the volume at which $p$ equals one atmosphere.

Step 1. Write the grand potential: $\Omega=-k_BT\,\dfrac{zV}{\lambda_T^3}$ , with $z=e^{\beta\mu}$ the fugacity and $\lambda_T=h/\sqrt{2\pi m k_BT}$ the thermal wavelength.

Step 2. Get the particle number from $N=-\partial\Omega/\partial\mu$ . Since $\partial z/\partial\mu=\beta z$ , we find $N=k_BT\,\dfrac{V}{\lambda_T^3}\,\beta z=\dfrac{zV}{\lambda_T^3}$ . So $\ln\Xi=N$ and $\Omega=-Nk_BT$ .

Step 3. Apply $\Omega=-pV$ : $-pV=-Nk_BT$ , hence $p=\dfrac{Nk_BT}{V}$ — the ideal-gas law, obtained without ever writing a canonical partition function.

Step 4. Put in numbers for $N=N_A=6.022\times10^{23}$ , $T=300\ \mathrm K$ , $k_B=1.381\times10^{-23}\ \mathrm{J/K}$ . Setting $p=1\ \mathrm{atm}=101{,}325\ \mathrm{Pa}$ gives $V=Nk_BT/p=(6.022\times10^{23})(1.381\times10^{-23})(300)/101{,}325$ . The molar volume is $V=0.02462\ \mathrm{m}^3=24.62\ \mathrm{L}$ , the familiar molar volume of an ideal gas at $300\ \mathrm K$ and $1\ \mathrm{atm}$ — extracted purely from the grand potential.

2 · Langmuir adsorption: coverage from a per-site grand potential (exam level)

A surface has $M$ independent adsorption sites, each of binding energy $\varepsilon$ , in equilibrium with a gas reservoir at chemical potential $\mu$ . Compute the fractional coverage $\theta$ from the grand potential and find the value of $\mu$ at which the surface is half-covered.

Step 1. Each site is either empty (energy $0$ , $N=0$ ) or occupied (energy $\varepsilon$ , $N=1$ ), so its per-site grand partition function is $\xi=1+e^{\beta(\mu-\varepsilon)}$ and the total is $\Xi=\xi^M$ (independent sites).

Step 2. The per-site grand potential is $\Omega/M=-k_BT\ln\xi=-k_BT\ln(1+e^{\beta(\mu-\varepsilon)})$ .

Step 3. Coverage is the mean occupation per site: $\theta=-\dfrac1M\dfrac{\partial\Omega}{\partial\mu}=\dfrac{e^{\beta(\mu-\varepsilon)}}{1+e^{\beta(\mu-\varepsilon)}}$ — the Langmuir isotherm, delivered directly by $\Omega$ .

Step 4. Half coverage means $\theta=\tfrac12$ , which requires $e^{\beta(\mu-\varepsilon)}=1$ , i.e. $\mu=\varepsilon$ . The surface is half-covered exactly when the reservoir's chemical potential equals the binding energy. For reference, $\theta=\tfrac34$ when $\beta(\mu-\varepsilon)=\ln3$ , i.e. $\mu=\varepsilon+k_BT\ln3\approx\varepsilon+1.10\,k_BT$ — every occupation number the surface can show is one derivative of the grand potential.

Origins · 1 mingrand potential and how this was discovered

grand potential

thermodynamic potential — Gibbs 1902, Landau tradition · open-system free energy

The grand potential $\Omega=-k_BT\ln\Xi$ is the thermodynamic potential natural to open systems, and it wears two names for two lineages. As a member of the grand canonical ensemble it descends from J. Willard Gibbs, who introduced and named the "grand ensemble" in his 1902 Elementary Principles in Statistical Mechanics, the treatise that founded modern statistical mechanics on the ensemble concept. As a free energy it is often called the Landau free energy or Landau grand potential, after the treatment in Landau and Lifshitz's Statistical Physics, where $\Omega(T,V,\mu)$ is developed as the potential whose minimization governs open-system and quantum-gas equilibrium. Its defining virtue is economy: with only one extensive natural variable, the Euler relation collapses it to the single term $\Omega=-pV$ , so a bulk fluid's entire equation of state is one logarithm. This makes $\Omega$ the workhorse of quantum-gas thermodynamics — the Fermi–Dirac and Bose–Einstein occupation numbers, the Sommerfeld expansion of metals, Bose–Einstein condensation, and blackbody radiation ( $\mu=0$ ) all flow from a single grand-potential calculation. In modern condensed-matter theory the same object reappears as the free-energy functional of density-functional theory and as the effective potential of field theory at finite chemical potential.

History

1875

— Gibbs introduces the chemical potential in On the Equilibrium of Heterogeneous Substances.

1902

— Gibbs's Elementary Principles in Statistical Mechanics defines and names the grand ensemble.

1924

— Bose and Einstein develop quantum statistics, for which the grand potential is the natural potential.

1926

— Fermi and Dirac give the exclusion-obeying statistics, whose Ω is a single log sum.

1937

— Landau's work on phase transitions establishes the free-energy-functional viewpoint associated with the Landau grand potential.

Going further

References. Pathria & Beale, Statistical Mechanics, 3rd ed., Ch. 4 (The Grand Canonical Ensemble) · Kardar, Statistical Physics of Particles, Ch. 4 & 7 · Reif, Fundamentals of Statistical and Thermal Physics, Ch. 6 & 9 · Sethna, Statistical Mechanics: Entropy, Order Parameters, and Complexity, Ch. 6 · Landau & Lifshitz, Statistical Physics Part 1, §35 (The thermodynamic potential Ω) · Gibbs, Elementary Principles in Statistical Mechanics (1902).

Provenance. All relations — the bridge $\Omega=-k_BT\ln\Xi=U-TS-\mu N$ , the differential $\mathrm d\Omega=-S\,\mathrm dT-p\,\mathrm dV-N\,\mathrm d\mu$ , the owned identity $\Omega=-pV$ from first-order homogeneity, the Legendre relation $\Omega=F-\mu N$ , and the fluctuation–compressibility relation — are standard (Pathria Ch. 4, Landau–Lifshitz §35). A verifier should reproduce: the Gibbs-entropy identity giving $-k_BT\ln\Xi=U-TS-\mu N$ ; the Euler collapse $V\,\partial_V\Omega=\Omega=-pV$ ; the ideal-gas $pV=Nk_BT$ ; the quantum-gas occupations $N=-\partial_\mu\Omega$ giving $1/(e^{\beta(\varepsilon-\mu)}\mp1)$ ; and the worked numbers (molar volume $24.62\ \mathrm L$ at $300\ \mathrm K,1\ \mathrm{atm}$ ; Langmuir half-coverage at $\mu=\varepsilon$ ). These are recomputed in verify_aq-sm-grand-potential.py. $k_B$ , $N_A$ , $h$ (CODATA).

Builds on: Grand Canonical Ensemble · Grand Partition Function · Gibbs Factor and Chemical Potential · Helmholtz Free Energy from the Partition Function. Leads to: Thermodynamic Quantities from the Partition Function · Heat Capacity and Fluctuation-Response Relation · Energy Fluctuations in the Canonical Ensemble.

Why is Ω = −pV so simple, when F and U are not simply proportional to any single variable?

Because of how many extensive natural variables each potential has. The internal energy $U(S,V,N)$ has three extensive natural variables ( $S$ , $V$ , $N$ ), so Euler's theorem gives the three-term relation $U=TS-pV+\mu N$ — no single-term collapse. The Helmholtz free energy $F(T,V,N)$ still has two extensive ones ( $V$ and $N$ ), giving $F=-pV+\mu N$ . But the grand potential $\Omega(T,V,\mu)$ has traded away $N$ for the intensive $\mu$ , leaving $V$ as the only extensive natural variable. A quantity that is extensive and depends on exactly one extensive variable must be strictly proportional to it, and Euler's theorem pins the proportionality constant to $\partial\Omega/\partial V=-p$ . That is the whole reason $\Omega=-pV$ and no analogous single-term identity exists for $F$ or $U$ : $\Omega$ is the potential whose transforms have removed every extensive variable but volume.

What is the difference between Ω and the Helmholtz free energy F, in practice?

$F$ is the right potential when the particle number $N$ is fixed (a closed system in contact with a heat bath); $\Omega$ is the right potential when $N$ is free to fluctuate and only its price $\mu$ is fixed (an open system in contact with a heat-and-particle bath). They are Legendre transforms of each other: $\Omega=F-\mu N$ . In practice you reach for $\Omega$ whenever the particle number is not naturally fixed — adsorption on a surface, electrons in a metal exchanging with contacts, or ideal quantum gases, where treating each single-particle level as an independent open subsystem makes the whole calculation factorize. You reach for $F$ when $N$ is genuinely conserved and you want $\mu$ as an output rather than a control knob. In the thermodynamic limit the two give identical thermodynamics (ensemble equivalence), so the choice is one of convenience: whichever variable — $N$ or $\mu$ — is the one you actually control.

How does Ω give the Fermi–Dirac and Bose–Einstein distributions?

For a gas of non-interacting particles, the single-particle levels are independent, so the grand partition function factorizes over modes: $\ln\Xi=\mp\sum_k\ln(1\mp z\,e^{-\beta\varepsilon_k})$ , with the upper sign for fermions and the lower for bosons. The grand potential is then $\Omega=\mp k_BT\sum_k\ln(1\mp z\,e^{-\beta\varepsilon_k})$ . The total mean particle number is $N=-\partial\Omega/\partial\mu$ , and because the sum is over independent modes, this derivative acts mode by mode, giving $\langle n_k\rangle=1/(e^{\beta(\varepsilon_k-\mu)}\pm1)$ — the Fermi–Dirac ( $+$ ) and Bose–Einstein ( $-$ ) occupation numbers. So the two great quantum distributions are a single derivative of a single grand potential, distinguished only by a sign. (The deeper quantum-mechanical reason for the sign — symmetric versus antisymmetric wavefunctions — belongs to the Quantum Statistics chapter; here $\Omega$ supplies the grand-canonical bookkeeping that turns that microscopic input into thermodynamics.)

Why does every first derivative of Ω carry a minus sign?

Because $\Omega$ is a doubly lowered potential — it sits at the bottom of the Legendre ladder, having traded away both $S$ (for $T$ ) and $N$ (for $\mu$ ). Each Legendre transform that swaps an extensive variable for its intensive conjugate flips the sign of the corresponding term in the differential. Starting from $\mathrm dU=T\,\mathrm dS-p\,\mathrm dV+\mu\,\mathrm dN$ (where $\mathrm dS$ and $\mathrm dN$ appear with $+$ ), transforming $S\to T$ turns $+T\,\mathrm dS$ into $-S\,\mathrm dT$ , and transforming $N\to\mu$ turns $+\mu\,\mathrm dN$ into $-N\,\mathrm d\mu$ . The $-p\,\mathrm dV$ term is untouched because $V$ was never transformed. The result, $\mathrm d\Omega=-S\,\mathrm dT-p\,\mathrm dV-N\,\mathrm d\mu$ , has all three minus signs, so $S$ , $p$ , and $N$ are all obtained as negative derivatives of $\Omega$ . The minus signs are the fingerprint of the double transform.

When does Ω = −pV break down?

It breaks down whenever the system is not bulk-extensive — that is, whenever surface, edge, or finite-size contributions to the free energy are not negligible compared to the bulk. The derivation of $\Omega=-pV$ used first-order homogeneity: doubling the volume doubles $\Omega$ . That is only true if the system has no important surface. For a small droplet, a thin film, a nanoparticle, or any system where the surface-to-volume ratio is appreciable, $\Omega$ acquires a surface term (roughly $\gamma A$ with $\gamma$ the surface tension and $A$ the area), so $\Omega=-pV+\gamma A+\dots$ and the clean identity fails. Similarly, systems with long-range interactions (gravitating matter, unscreened Coulomb systems) are non-extensive and do not obey $\Omega=-pV$ . The safe statement is that $\Omega=-k_BT\ln\Xi$ is always true, but $\Omega=-pV$ requires the additional assumption of bulk extensivity — exactly the assumption that also underlies ensemble equivalence and the ordinary thermodynamic limit.

How is Ω related to measurable light scattering?

Through its second derivative with respect to $\mu$ . The curvature $\partial^2\Omega/\partial\mu^2=-\partial N/\partial\mu$ controls the number fluctuation $\langle(\Delta N)^2\rangle=k_BT\,\partial N/\partial\mu$ , and via a thermodynamic chain this equals $\langle N\rangle^2 k_BT\kappa_T/V$ , where $\kappa_T$ is the isothermal compressibility. Density fluctuations of exactly this size are what scatter light: the scattered intensity in a fluid is proportional to $\langle(\Delta N)^2\rangle$ , hence to $\kappa_T$ . Ordinarily $\langle(\Delta N)^2\rangle/\langle N\rangle^2\sim1/N$ is tiny and the fluid is transparent. But as a liquid–gas critical point is approached, $\kappa_T$ diverges, the density fluctuations grow until their correlation length reaches optical wavelengths, and the fluid scatters visible light strongly and turns milky — critical opalescence, first explained by Einstein and Smoluchowski. So a curvature of the grand potential, an abstract thermodynamic quantity, is directly visible to the eye.