11.1 Microcanonical, canonical, and grand canonical ensembles

Statistical Ensembles: Microcanonical, Canonical, and Grand Canonical

Statistical mechanics uses ensembles to model systems with different constraints. The microcanonical, canonical, and grand canonical ensembles represent isolated, closed, and open systems respectively, each with its own probability distribution and partition function.

These ensembles provide the framework for connecting microscopic states to macroscopic thermodynamic properties. The key idea: you maximize entropy subject to whatever constraints your system has, and Lagrange multipliers naturally produce the correct probability distributions for each case.

Defining Characteristics and Constraints

Microcanonical ensemble — represents an isolated system with fixed particle number ($N$), volume ($V$), and energy ($E$). Because no energy or particles flow in or out, every accessible microstate is equally probable. This equal-probability postulate is the foundation of equilibrium statistical mechanics.

Canonical ensemble — represents a closed system in thermal equilibrium with a heat bath. Energy can flow between the system and the bath, but particles cannot. The fixed quantities are $N$, $V$, and temperature $T$. The probability of finding the system in microstate $i$ is weighted by the Boltzmann factor $\exp(-\beta E_i)$, which exponentially suppresses high-energy states.

Grand canonical ensemble — represents an open system that exchanges both energy and particles with a reservoir. The fixed quantities are chemical potential ($\mu$), $V$, and $T$. The probability of a microstate now depends on both the energy and the particle number through the factor $\exp(-\beta(E_i - \mu N))$.

Probability Distributions and Lagrange Multipliers

Each ensemble's probability distribution comes from maximizing the Shannon/Gibbs entropy $S = -k_B \sum_i P_i \ln P_i$ subject to the ensemble's constraints, using Lagrange multipliers to enforce those constraints.

• Microcanonical: The only constraint (besides normalization) is that all microstates share the same energy $E$. Maximizing entropy under this condition yields equal probability for every accessible microstate: $P_i = 1/\Omega$, where $\Omega$ is the total number of microstates.
• Canonical: You add a constraint on the average energy $\langle E \rangle$. The Lagrange multiplier for this constraint turns out to be $\beta = 1/(k_B T)$. The result is the Boltzmann distribution:

$P_i = \frac{1}{Z} \exp(-\beta E_i)$

• Grand canonical: You add constraints on both average energy and average particle number. The energy multiplier is again $\beta$, and the particle-number multiplier is $-\beta\mu$. The resulting distribution is:

$P_{i,N} = \frac{1}{\Xi} \exp(-\beta(E_i - \mu N))$

The Lagrange multipliers aren't just mathematical tricks. They have direct physical meaning: $\beta$ connects to temperature, and $\mu$ connects to the cost of adding a particle to the system.

Ensemble Selection for System Types

Isolated Systems

The microcanonical ensemble applies when a system has fixed $N$, $V$, and $E$. Think of a gas in a perfectly insulated, rigid container, or a closed quantum system with a well-defined energy eigenvalue. No energy enters or leaves, so the total energy is strictly constant.

Closed Systems

The canonical ensemble applies when a system can exchange energy (but not particles) with a heat bath. A gas in a rigid container immersed in a large thermal reservoir at temperature $T$ is the standard example. The system's energy fluctuates, but its average is determined by $T$.

Open Systems

The grand canonical ensemble applies when both energy and particles can flow between the system and a reservoir. Examples include electrons in a metal in contact with a thermal and particle bath, or a small region of gas that freely exchanges molecules with its surroundings. Both energy and particle number fluctuate.

Problem-Solving Approach

When facing a statistical mechanics problem:

1. Identify whether the system is isolated, closed, or open.
2. Determine which quantities are fixed ($N$, $V$, $E$, $T$, or $\mu$).
3. Select the corresponding ensemble.
4. Write down the appropriate partition function and probability distribution.
5. Derive the thermodynamic quantities you need from the partition function.

Partition Functions and Thermodynamic Properties

Microcanonical Ensemble

The "partition function" here is simply $\Omega(N, V, E)$, the number of accessible microstates. Entropy follows directly from the Boltzmann relation:

$S = k_B \ln \Omega$

All other thermodynamic quantities (temperature, pressure, chemical potential) come from partial derivatives of $S$. For example, temperature is defined by $1/T = \partial S / \partial E \big|_{N,V}$.

Canonical Ensemble

The canonical partition function sums Boltzmann factors over all microstates:

$Z(N, V, T) = \sum_i \exp(-\beta E_i)$

This single function generates all equilibrium thermodynamics. The Helmholtz free energy is:

$F = -k_B T \ln Z$

From $F$, you can obtain entropy ($S = -\partial F / \partial T$), pressure ($P = -\partial F / \partial V$), and average energy ($\langle E \rangle = -\partial \ln Z / \partial \beta$). The partition function $Z$ acts as a generating function: derivatives of $\ln Z$ with respect to $\beta$ give energy moments and fluctuations.

Grand Canonical Ensemble

The grand canonical partition function sums over both microstates and particle numbers:

$\Xi(\mu, V, T) = \sum_N \sum_i \exp(-\beta(E_i - \mu N))$

The grand potential is:

$\Phi = -k_B T \ln \Xi$

From $\Xi$ you can extract the average particle number ($\langle N \rangle = k_B T \, \partial \ln \Xi / \partial \mu$), pressure, and particle number fluctuations. This ensemble is especially powerful for systems where particle number isn't fixed, such as quantum gases or adsorption problems.

Lagrange Multipliers in Ensemble Distributions

Role in Deriving Probability Distributions

Lagrange multipliers let you find the extremum of a function subject to constraints. In statistical mechanics, the function being maximized is the entropy, and the constraints are normalization ($\sum_i P_i = 1$) plus whatever physical quantities are fixed on average.

How It Works for Each Ensemble

Microcanonical: The constraint is that only microstates with energy $E$ are accessible. Maximizing entropy with just the normalization constraint over this restricted set gives $P_i = 1/\Omega$. The energy constraint is handled by restricting the sum rather than introducing a separate multiplier.

Canonical: You maximize $S = -k_B \sum_i P_i \ln P_i$ subject to:

• $\sum_i P_i = 1$ (normalization)
• $\sum_i P_i E_i = \langle E \rangle$ (fixed average energy)

The multiplier for the energy constraint is identified as $\beta = 1/(k_B T)$, yielding $P_i \propto \exp(-\beta E_i)$.

Grand canonical: You add a third constraint:

• $\sum_{i,N} P_{i,N} N = \langle N \rangle$ (fixed average particle number)

The multiplier for this constraint is $-\beta\mu$, yielding $P_{i,N} \propto \exp(-\beta(E_i - \mu N))$.

Why This Matters

This Lagrange multiplier approach gives a unified derivation for all three ensembles. You start from the same entropy-maximization principle each time and simply change which constraints you impose. The multipliers that emerge have direct thermodynamic interpretations: $\beta$ sets the energy scale through temperature, and $\mu$ governs the tendency of particles to enter or leave the system. This systematic method ensures that every probability distribution is internally consistent with the physical constraints of the system it describes.