The Gibbs-Duhem Equation: Why Component Activities Are Not Independent

What the Equation Says and Where It Applies

The Gibbs-Duhem equation is a constraint on the intensive variables of a single homogeneous phase. In its full form it reads:

S dT − V dP + Σᵢ nᵢ dμᵢ = 0

where S is entropy, V is volume, T temperature, P pressure, nᵢ the moles of component i, and μᵢ its chemical potential (partial molar Gibbs energy, in J/mol). At constant temperature and pressure the first two terms vanish and, dividing by total moles, it collapses to the celebrated form Σᵢ xᵢ dμᵢ = 0.

• It applies to any single phase — gas, liquid solution, or solid solution.
• It links the components: for a binary mixture, once you know how μ₁ varies with composition, μ₂ is fully determined.
• It reduces the number of independently variable intensive properties, and underlies the Gibbs phase rule F = C − P + 2.

Because chemical potentials are hard to measure directly, the practical payoff is enormous: measure the easy component, integrate, and recover the hard one.

Derivation from Euler's Theorem, Step by Step

The equation drops out of a single deep fact: Gibbs energy G is a first-order homogeneous function of the mole numbers. Double every nᵢ at fixed T and P and G doubles.

Step 1. Euler's theorem for such a function gives G = Σᵢ nᵢ (∂G/∂nᵢ)_{T,P,nⱼ} = Σᵢ nᵢ μᵢ, since μᵢ ≡ (∂G/∂nᵢ) is exactly the partial molar Gibbs energy.

Step 2. Take the total differential of G = Σ nᵢ μᵢ using the product rule: dG = Σᵢ μᵢ dnᵢ + Σᵢ nᵢ dμᵢ.

Step 3. But the fundamental equation of thermodynamics already says dG = −S dT + V dP + Σᵢ μᵢ dnᵢ.

Step 4. Subtract Step 3 from Step 2. The Σ μᵢ dnᵢ terms cancel, leaving:

Σᵢ nᵢ dμᵢ = −S dT + V dP, i.e. S dT − V dP + Σᵢ nᵢ dμᵢ = 0.

That is the Gibbs-Duhem equation. No approximation was made — it is exact, a pure consequence of extensivity.

Key Quantities and a Worked Binary Example

For a two-component liquid at constant T and P, write chemical potential as μᵢ = μᵢ° + RT ln aᵢ, where aᵢ = γᵢxᵢ is the activity, γᵢ the activity coefficient, and R = 8.314 J·mol⁻¹·K⁻¹. Substituting into Σ xᵢ dμᵢ = 0 and dividing by RT gives:

x₁ dln a₁ + x₂ dln a₂ = 0

Since dln xᵢ terms satisfy x₁ dln x₁ + x₂ dln x₂ = 0 on their own, this reduces to x₁ dln γ₁ + x₂ dln γ₂ = 0.

Worked example. Suppose a solution follows the one-parameter Margules model ln γ₁ = A x₂². Then dln γ₁ = 2A x₂ dx₂. Gibbs-Duhem forces ln γ₂ = A x₁², a symmetric partner — you did not get to choose it. With A = 0.6 at x₁ = 0.3 (so x₂ = 0.7): γ₁ = exp(0.6·0.49) = 1.34 and γ₂ = exp(0.6·0.09) = 1.06. In the dilute limit each γ → 1, recovering Raoult's and Henry's laws automatically.

How It Is Measured and Used in Practice

The workhorse application is vapor-liquid equilibrium (VLE). The Duhem-Margules equation, x₁(∂ln p₁/∂x₁) = x₂(∂ln p₂/∂x₂), lets chemists check or complete partial-pressure data.

• Recovering an immeasurable activity: the vapor pressure of a nonvolatile solute (a salt, a polymer) cannot be measured, but the solvent's can. Integrating x₁ dln a₁ = −x₂ dln a₂ from pure solvent gives the solute activity.
• Thermodynamic consistency tests: Redlich-Kister and Herington area tests use ∫₀¹ ln(γ₁/γ₂) dx₁ ≈ 0 to flag bad VLE datasets before they enter process simulators like Aspen Plus.
• Electrolytes: osmotic coefficients of aqueous salts are converted to mean ionic activity coefficients (Debye-Hückel, Pitzer models) via Gibbs-Duhem integration.

In metallurgy the same integration yields the activity of a dissolved metal in an alloy from the measured activity of the solvent metal — critical for predicting slag-metal reactions and phase diagrams.

How It Relates to Neighboring Concepts

The Gibbs-Duhem equation is easy to confuse with its relatives, so distinguish them:

• vs. the fundamental equation (dG = −S dT + V dP + Σ μᵢ dnᵢ): that describes how G changes when you add matter; Gibbs-Duhem describes the constraint among the intensive variables at fixed composition scaling.
• vs. Raoult's and Henry's laws: those are limiting behaviors; Gibbs-Duhem proves they are thermodynamically linked — if the solvent obeys Raoult's law over a range, the solute must obey Henry's law there.
• vs. the Gibbs phase rule: the phase rule counts degrees of freedom; Gibbs-Duhem is one of the equations that removes a degree per phase.
• vs. the Clausius-Clapeyron equation: both constrain phase behavior, but Clapeyron governs a pure substance's P-T coexistence line, while Gibbs-Duhem governs composition within a mixture.

All flow from the same root: chemical potential equality at equilibrium plus the extensivity of energy.

Exceptions, Limits, and Historical Significance

The equation is exact within a single homogeneous phase. Its 'limits' are really conditions of applicability:

• It says nothing across a phase boundary; each phase has its own Gibbs-Duhem relation, and at equilibrium μᵢ is equal across phases separately.
• The T,P-constant form fails if T or P vary — you must retain the S dT and V dP terms (relevant in gradients, centrifuges, or gravitational fields).
• For charged species the 'chemical potential' becomes the electrochemical potential μ̃ᵢ = μᵢ + zᵢFφ; the constraint still holds with μ̃.

Historical significance: Gibbs published the relation buried in his monumental 1875-1878 memoir On the Equilibrium of Heterogeneous Substances — arguably the founding document of chemical thermodynamics. Pierre Duhem, whose 1886 thesis championed and extended it, gave it independent visibility in Europe. Together their names mark one of the most-used consistency constraints in all of physical chemistry, from brewing to blast furnaces.

General and specialized forms of the Gibbs-Duhem relation

Form	Equation	Conditions / use
General	S dT − V dP + Σᵢ nᵢ dμᵢ = 0	Any phase, variable T and P
Constant T, P	Σᵢ xᵢ dμᵢ = 0	Isothermal-isobaric mixtures
Binary (activity form)	x₁ dln a₁ + x₂ dln a₂ = 0	Two-component solutions
Duhem-Margules	x₁(∂ln p₁/∂x₁) = x₂(∂ln p₂/∂x₂)	Vapor pressures of a binary
Activity-coefficient form	x₁ dln γ₁ + x₂ dln γ₂ = 0	Consistency test for γ data
Ternary	x₁ dln a₁ + x₂ dln a₂ + x₃ dln a₃ = 0	Multicomponent electrolytes