Demonstration of coexistence conditions

M. Modell and R. C. Reid, “Thermodynamics and its applications”, Prentice Hall, New
Jersey, USA (1974).

Assume an isolated system (dU=0) of constant volume V, total number of particles Ni of

each component i (i=1,2….n) . Assume the system is in two phases α and β (note that
each phase is an open system, interchanging matter and energy between them).

According to the 2nd law for the system to be at equilibrium the total entropy must be
at a maximum; i.e. dS=0.

Phase α Phase β

Sα N1α N1β Sβ
Vα N 2α N 2β Vβ
Tα N iα N iβ Tβ

Since the entropy, the volume, and the number of particles are extensive variables, it is
possible to write
S = Sα + S β
V =Vα +V β (1)
N i = N iα + N iβ , i = 1,2,...
Using the entropy maximum principle (dS=0), and the condition of constant volume
(dV=0), and number of particles (dNi=0), and remembering that the internal energy is also
an extensive variable, and since the system is isolated (dU=0) the following relations can
be written:
dS = 0 = d ( S α + dS β )
dV = 0 = d (V α + dV β )
(2)
dN i = 0 = d (dN iα + dN iβ ), i = 1,2....
dU = 0 = d (U α + dU β ) ,
or,
 dS α = − dS β
dV α = −dV β
(3)
dN iα = −dN iβ , i = 1,2....
dU α = − dU β .
The entropy of a system can be expressed as a function of internal energy, volume and
number of particles as
dU PdV μ
dS = + − ∑ i dN i , (4)
T T i =1, 2...n T

where μi is the chemical potential of component i.

Using equation (3) for phases α and β, and since dS = dS α + dS β , the total entropy can
be written as
⎧1 1 ⎫ ⎧ Pα P β ⎫ ⎧ μα μ β ⎫
dS = ⎨ α − β ⎬dU α + ⎨ α − β ⎬dV α − ∑ ⎨ iα − iβ ⎬dN iα , (5)
⎩T T ⎭ ⎩T T ⎭ i =1, 2... n ⎩ T T ⎭
where P is the pressure. Since the internal energy, the volume and the numbers of
particles are independent variables in the expression above, dS=0, only if
⎧ 1 1 ⎫
⎨ α − β ⎬=0
⎩T T ⎭
⎧ Pα P β ⎫
⎨ α − β ⎬=0 (6)
⎩T T ⎭
⎧ μiα μiβ ⎫
⎨ α − β ⎬ = 0 ∀i .
⎩T T ⎭
so that the familiar conditions for phase equilibrium are obtained:

Tα = T β
Pα = P β (7)
α β
μ i = μi ∀i .

Important note:
At this point you may re-call that df = 0 is the mathematical condition for a
turning point in a function (a maximum, a minimum or a point of inflection), not just the
maximum. The conditions above (equation (7)) guarantee the system to be at equilibrium,
but do not guarantee that this equilibrium will be stable. Satisfying the relations above
may lead to solutions or states in the sytem that correspond to metastable (supercooled
and superheated states are examples you may have heard of before) and unstable
equilibria.
At constant P, T, the total Gibbs free energy of the mixture needs to be considered
together with equation (7) to ensure the equilibrium state obtained is stable. The stable
state will be that of lowest (minimum) Gibbs free energy.