Chemical Potential from Particle Exchange
Adding particles in canonical and grand canonical descriptions
Lesson 3718 of 4,500 · Statistical Thermodynamics and Phase Equilibria
Learning objectives
- Interpret chemical potential as an incremental free-energy cost
- Connect canonical N derivatives to grand canonical particle-number response
Introduction
Chemical potential is often introduced as partial molar Gibbs energy. A particle-exchange viewpoint makes its meaning more concrete: it is the free-energy cost of adding material under specified constraints. In a fixed-volume canonical calculation this is a derivative of Helmholtz energy. In a grand canonical calculation the reservoir fixes μ, and particle-number fluctuations reveal how the subsystem responds.
Core explanation
At fixed T,V, the chemical potential per particle is μ = (∂A/∂N) T,V in the large-system continuous-N limit. For a finite model one may compare A (N+1) − A N, noting that discreteness and finite-size effects matter. Since A N = −kT ln Q N, the finite difference is μ add ≈ −kT ln(Q (N+1)/Q N) under a consistent energy reference and volume. If adding a particle creates many favourable states, Q (N+1) grows relatively strongly and the free-energy cost can be lower.
For the classical monatomic ideal gas, Q N = (V/Λ³)ᴺ/N!. The exact partition-function ratio Q (N+1)/Q N = [V/Λ³]/(N+1). Hence A (N+1) − A N = kT ln[(N+1)Λ³/V]. For large N this approaches μ = kT ln(NΛ³/V). The result rises with density: inserting a particle into a denser gas has a higher chemical potential at fixed T. The absolute value depends on the chosen energy zero and internal-state factors, but the density dependence is physical.
In the grand canonical ensemble, Ξ = Σ N e^(βμN)Q N. Differentiating gives ⟨N⟩ = (1/β)∂ln Ξ/∂μ, and ∂⟨N⟩/∂μ = βVar(N) ≥ 0. Raising reservoir μ cannot lower mean occupancy for this simple stable grand canonical setup; larger number fluctuations accompany a stronger response. For adsorption, these formulas connect binding-site energies and a gas reservoir to uptake curves.
At fixed T,p, the corresponding molar derivative is μ i = (∂G/∂n i) T,p,n j. The two definitions are compatible because A and G differ by pV and their derivatives hold different mechanical variables fixed. One must not insert a fixed-V formula into a fixed-p problem without transforming the potential. For a mixture, each species has its own μ i and a separate particle-number derivative.
Step-by-step reasoning
Identify whether the system is held at fixed V or p. For fixed T,V, write A N = −kT ln Q N and calculate a derivative or finite insertion difference. For open systems, write Ξ and differentiate ln Ξ to find mean occupancy and its response. Convert per-particle μ to molar units only by consistent use of N A, k and R. State the energy reference and any ideal-gas or noninteraction assumption.
Visual explanation
Draw a row of boxes labelled N, N+1 and N+2 with free energies A N. The slope or adjacent difference is μ. Next draw a reservoir feeding a small pore, with a graph of mean occupancy increasing as reservoir μ rises. Label the slope βVar(N) to connect response and fluctuation.
Real-world analogy
The marginal cost of adding one item to a constrained storage room can rise as the room fills. Chemical potential is a thermodynamic marginal cost: it incorporates both energy and entropy, not just physical crowding. A reservoir setting that marginal cost controls how many items enter on average.
Real-world example
Gas adsorption isotherms rise as reservoir chemical potential increases. In a simple site model, higher μ makes occupied states more probable. A steep uptake region corresponds to a strong occupancy response; for interacting adsorbates it can also signal cooperative behaviour or phase-like changes within pores.
Why?
Free energy balances energetic favourability against the number of accessible states at a set temperature. Adding a particle changes both. The ratio Q (N+1)/Q N measures their combined statistical effect, so its logarithm gives the incremental Helmholtz cost. Grand canonical weighting then compares all N values against the same reservoir μ.
Common misconception
Chemical potential is not simply the average energy per particle. It includes entropy and depends on what is held fixed. Also, a negative μ under one reference does not mean adding matter is always spontaneous; the relevant comparison is with the reservoir or another phase using a consistent reference. The finite N+1 difference need not equal a bulk derivative exactly.
Worked example
Let an ideal classical monatomic gas have NΛ³/V = 0.010 at one state. Its large-N μ per particle is kT ln 0.010 = −4.605kT. If the number density increases tenfold at the same T, NΛ³/V becomes 0.10 and μ becomes −2.303kT, an increase of kT ln 10. Both values still satisfy the classical dilute criterion only approximately for the denser state; the trend is the main point.
Quick check
1. What is the exact ratio Q (N+1)/Q N for the classical ideal translational gas model? Answer: It is (V/Λ³)/(N+1). The factorial changes from N! to (N+1)!, producing the denominator N+1.
Exam focus
Label held-fixed variables in μ derivatives. Use A at fixed T,V and G at fixed T,p. Include N! in Q N before forming an insertion ratio. Distinguish per-particle μ with k from molar μ with R. Interpret ∂⟨N⟩/∂μ = βVar(N) under grand canonical conditions, not as a fixed-N statement.
Advanced insight
For interacting fluids, particle insertion samples the energetic cost of placing a new particle into many possible configurations. The resulting excess chemical potential goes beyond the ideal density logarithm. Simulation methods can estimate it, but dense systems may have very low successful insertion probability, making direct sampling difficult.
Summary
Chemical potential is an incremental free-energy cost under specified constraints. At fixed T,V, μ = ∂A/∂N and ideal-gas counting gives μ ≈ kT ln(NΛ³/V). In the grand canonical ensemble, reservoir μ controls mean N, with response ∂⟨N⟩/∂μ = βVar(N). The fixed-T,p version uses a Gibbs-energy derivative.
Practice questions
1. If Q (N+1)/Q N = 2 under one finite model, what is A (N+1) − A N at temperature T? Answer: −kT ln 2, relative to the same energy reference and fixed T,V. The negative sign means the new state count lowers the finite insertion free-energy difference. 2. A pore's particle-number variance is nearly zero at one μ. What does that imply for its local uptake response to μ? Answer: The grand canonical relation ∂⟨N⟩/∂μ = βVar(N) gives a small local response. Occupancy is nearly fixed over a small change in μ. 3. Why are ∂A/∂N at fixed V and ∂G/∂N at fixed p not interchangeable without qualification? Answer: They hold different mechanical variables fixed. Adding material may change pressure in the fixed-V case or volume in the fixed-p case, and the appropriate potential includes the corresponding work term.