The Grand Canonical Ensemble

Introduction

An adsorption site, a small pore or a piece of a large fluid can exchange particles with a surrounding reservoir. Fixing its particle number would misrepresent that boundary. The grand canonical ensemble fixes temperature, volume and chemical potential while summing over possible particle numbers. This makes it a natural language for sorption, open subsystems and fluctuations in composition.

Core explanation

Let Q N(T,V) be the canonical partition function when the subsystem contains exactly N particles, including its appropriate state counting. For reservoir temperature T and chemical potential μ, the grand partition function is Ξ(T,V,μ) = Σ N=0∞ exp(βμN)Q N, with β = 1/(kT). A microstate with N and energy E has probability exp[−β(E − μN)]/Ξ. The factor μN rewards or penalises particle addition according to reservoir chemical potential, while E retains its thermal Boltzmann role. The weight is dimensionless because βμN is dimensionless when μ is energy per particle; a molar μ requires RT and mole amounts instead.

The grand potential is Ω G = −kT ln Ξ. Its natural variables are T, V and μ for a one-component system, and its differential is dΩ G = −S dT − p dV − N dμ for a simple homogeneous equilibrium phase. The mean particle number follows from ⟨N⟩ = (1/β)(∂ln Ξ/∂μ) T,V. A second derivative gives Var(N) = (1/β²)(∂²ln Ξ/∂μ²) T,V. These relations show that a reservoir fixes chemical potential but does not fix an open subsystem's instantaneous N.

A single adsorption site is an instructive model. If it can be empty at energy 0 or hold one molecule at energy ε, its two terms are Ξ = 1 + exp[−β(ε − μ)]. The occupancy probability is θ = exp[−β(ε − μ)]/Ξ = 1/{1 + exp[β(ε − μ)]}. Raising μ increases occupancy; raising ε at fixed μ decreases it. This toy site assumes at most one occupant and no interaction with other sites. For many independent equivalent sites, their grand partition functions multiply, while interactions require a more elaborate sum.

The grand canonical ensemble is not a claim that the whole universe exchanges particles with an external reservoir. One can divide a large closed system into a small subsystem and its much larger remainder. The remainder acts approximately as a reservoir, even though total particle number remains fixed. The ensemble describes the subsystem's equilibrium uncertainty about its own N.

Step-by-step reasoning

Identify a particle-permeable boundary and a reservoir with fixed T and μ. For each allowed N, write Q N and multiply it by exp(βμN). Sum those terms to obtain Ξ, normalise state probabilities with Ξ, then differentiate ln Ξ if a mean N or fluctuation is required. Keep μ per particle with k or μ per mole with R consistently.

Visual explanation

Draw a small pore connected by an opening to a vast reservoir. Above the pore show snapshots with N = 0, 1, 2 and so on, each with different energies. Put a weight exp[−β(E − μN)] beneath every snapshot. A bar chart of P(N) shows a distribution around a mean rather than one fixed occupancy.

Real-world analogy

A small shop restocked continuously from a huge warehouse has a fluctuating number of items on its shelves while the warehouse's supply conditions remain nearly constant. The shelf stock resembles N in a small open subsystem. The analogy helps show why fixed reservoir conditions do not imply a fixed local count, but molecules exchange according to energies rather than commercial decisions.

Real-world example

Gas adsorption in a porous solid depends on gas chemical potential and temperature. At low μ, empty sites dominate; as μ rises, occupancy increases. A grand canonical model can relate molecular binding energies to an adsorption isotherm, while measured deviations may indicate interactions among adsorbed molecules or a distribution of site energies.

Why?

When particles cross a boundary, equilibrium must compare states of different N. Their energy change alone is insufficient because removing a particle from the reservoir has a thermodynamic cost μ. The combination E − μN accounts for both subsystem energy and reservoir exchange, giving correct relative probabilities for different occupancies.

Common misconception

Using e^(−βE) alone for states with different N omits the chemical-potential cost of particle transfer. Another mistake is to take μ in J mol⁻¹ and put it into β = 1/(kT) with N counted as individual particles; the units then fail. Grand canonical N fluctuates, but its mean can be tightly defined for a large subsystem.

Worked example

For one adsorption site at ε = μ, the empty and occupied weights are both one, so Ξ = 2 and θ = 1/2. If ε − μ = kT ln 3, the occupied weight is e^(−ln 3) = 1/3. Then Ξ = 4/3 and θ = (1/3)/(4/3) = 1/4. The occupancy is a probability, not a fractional molecule sitting on one site at an instant.

Quick check

1. What happens to the mean occupancy of a one-site model when μ rises at fixed ε and T? Answer: The occupied weight exp[−β(ε − μ)] increases relative to the empty weight, so occupancy probability rises. The actual site remains either empty or occupied in any one microstate.

Exam focus

Write Ξ = Σ N e^(βμN)Q N with clear units and identify the N range. Use Ω G = −kT ln Ξ for the potential and do not confuse it with a microcanonical state count. If differentiating with respect to μ, hold T and V fixed. For a finite-capacity site, list all allowed occupancy states explicitly.

Advanced insight

For a uniform bulk one-component phase, Euler's relation G = μN and Ω G = A − μN = −pV. This provides a route from Ξ to pressure. In small or inhomogeneous systems, surface contributions can complicate the simple −pV identification, even though the grand partition function and derivative definitions remain useful.

Summary

The grand canonical ensemble fixes T,V,μ and permits N and E to fluctuate. Its Ξ sums e^(βμN)Q N over particle numbers, and Ω G = −kT ln Ξ. Differentiating ln Ξ gives mean N and its variance. This framework describes adsorption and open subsystems while keeping particle exchange thermodynamically consistent.

Practice questions

1. A pore permits N = 0 or 1 only, with occupied energy ε. Write Ξ and the empty probability. Answer: Ξ = 1 + e^(−β(ε − μ)). The empty state has weight 1, so P empty = 1/Ξ. 2. Why does a large reservoir approximately maintain μ while exchanging a few molecules with a small pore? Answer: The exchanged amount is negligible compared with the reservoir's total amount, so its intensive state changes only slightly. The pore's occupancy can change substantially while reservoir μ stays effectively fixed. 3. Which derivative of ln Ξ gives mean N at fixed T,V? Answer: ⟨N⟩ = (1/β)(∂ln Ξ/∂μ) T,V when μ is energy per particle and β = 1/(kT).