MIT OpenCourseWare

http://ocw.mit.edu

5.60 Thermodynamics & Kinetics

Spring 2008

For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
 5.60 Spring 2008 Lecture #29 page 1

APPLICATIONS: CHEMICAL AND PHASE EQUILIBRIA

Apply statistical mechanics to develop microscopic models for problems you’ve

treated so far with macroscopic thermodynamics

Separated atoms
0

Products Reactants

Product & reactant

E energy levels

−εr = -D0,r
−εp = -D0,p

Chemical equilibria
Gas phase: Calculate Kp from microscopic properties

aA + bB F cC + dD

ΔG 0 = −RT ln K p = cGC0 + dGD0 − aGA0 + bGB0

Need G0 for each species

G = A + pV = −kT lnQ + pV = −kT lnQ + NkT

Q = (qtrans
N
N ! ) qint
N

lnQ = N ln (qtrans qint ) − ln N ! = N ln (qtrans qint ) − N ln N + N = N ln (qtrans qint N ) + N

G = −kT lnQ + NkT = −NkT ln (qtrans qint N ) = −NkT ln (q N )

For molecules in the gas phase, internal degrees of freedom are rotations and
vibrations and electronic states.
5.60 Spring 2008 Lecture #29 page 2

qint = qrot qvib qelec

qelec = e D 0 kT
Dissociation energy D0 from ground electronic level
Usually no other electronic levels matter

1
qvib = ∑ e −nε 0 kT
= 1 + e −ε 0 kT
+ e −2ε 0 kT
+ e −3ε 0 kT
+" =
n 1 − e −ε 0 kT

Note we’ve set the zero of vibrational energy as the lowest vibrational level.

The zero-point vibrational energy has been included in qelec by using the

dissociation energy rather than the bottom of the electronic potential energy.

kT (300 K) ↔ 200 cm-1

Most molecular vibrational frequencies > 500 cm-1

qvib ≈ 1 - Need to calculate it, but it’s not large

We have not treated qrot. Levels are NOT evenly spaced: εrot = J(J + 1)ε0,rot

where J = 0, 1, 2,…, and ε0,rot ≈ 1 cm-1.

High-T limit: qrot = kT/ε0,rot ≈ 100-1000.

We’ve seen that qtrans ≈ 1030, qtrans/N ≈ 106.

For multiple species in chemical equilibrium, need to use Ni for each species and
use partial pressure values in pV term.

Gi = −Ni kT ln (qi Ni ) = −ni RT ln (qi Ni )

Gi = −RT ln (qi Ni ) = Gi ° + RT ln ( pi p ° )
⎛ qi Ni kT ⎞ ⎛ qi kT ⎞
Gi ° = −RT ln ⎜ ⎟ = −RT ln ⎜ ⎟
⎝ Ni p °V ⎠ ⎝ p °V ⎠

Example: simple reaction at 300 K: A-B + C-D → A-C + B-D

Assume all molecules have high vibrational frequencies ⇒ qvib ≈ 1 for all species
Translational & rotation: high-T limits
Assume similar rotational energies (depends on moments of inertia)
⇒ qtrans & qrot contributions are all equal
5.60 Spring 2008 Lecture #29 page 3

⎛q q ⎞
ΔGi ° = −RT ln ⎜ AC BD ⎟ = −RT ln K p
⎝ qAB qCD ⎠
qAC qBD e D AC 0, kT
e D BD0, kT
= e(
D0,AC +D0,BD −D0,AB −D0,CD ) kT
Kp = =
qAB qCD e D AB 0, kT D CD
e 0, kT

Only dissociation energies matter since the other degrees of freedom

contribute equally.

Usually energy effects dominate in chemical equilibria.

Relatively simple case: Solid-state chemical reaction aA F bB

e.g. isomerization in an organic molecular crystal

NA molecules of A, NB molecules of B, NA + NB = N

No translation or rotation, no change in pV ⇒ all the change is in the electronic

and vibrational energy

If vibrational frequencies are the same, then only the difference in dissociation
energies is needed.

Q= ∑ i
e −Ei kT = ∑
Ei

Ω ( Ei ) e −Ei kT
microstates energies

Recall the entropy due to mixing of two species

N
N! N
N! N N

Q= ∑
Ei
Ω ( Ei ) e −Ei kT = ∑
NA = 0 NA !NB !
e ( NADA +NBDB ) kT
= ∑
NA = 0 NA !NB !
e( A ) e( B )
D kT A D kT B

energies
N
N!
∑N qANA qBNB = ( qA + qB ) = qAN (1 + s )
N
where s = qB qA = e (
N DB −DA ) kT
= = e ΔD0 kT

NA = 0 A !NB !

The binomial theorem gives a simple closed form for the sum.

It’s convenient here to redefine the zero of energy as –DA so qA = 1 and

N
Q = (1 + s ) = 1 + a1s + a2s2 + a3s3 + " + aN sN

The ground state of the system has all A molecules

5.60 Spring 2008 Lecture #29 page 4

System energy is E(NB) = NBΔD0.

Probability that the system has this energy is

p ( NA ) = aNA sNA Q

The average number of B molecules is

N
a1s + 2a2s2 + 3a3s3 + " + NaN sN s ⎛ ∂Q ⎞ ⎛ ∂ ln Q ⎞
NB = ∑ NBp ( NB ) = = ⎜ ⎟=
NB Q Q ⎝ ∂s ⎠ ⎝⎜ ∂ ln s ⎠⎟

and E = U = NB ΔD0

If vibrational energies differ then qvib also needs to be included.

In general, low vibrational frequency favors a species since the levels are closely

spaced, so there are many low-energy states of that species.

For chemical reactions involving covalent bonds, the bond energies dominate

over entropy. But the treatment we’ve given could also be used for much more

subtle changes like rotation of CO molecules in a CO crystal.

Note that this treatment assumes no interactions between different unit cells.

This gives rise to a fairly gentle T-dependence of the equilibrium.

5.60 Spring 2008 Lecture #29 page 5

Solid-solid phase equilibria

Phase transitions between different crystalline phases may be easier to treat
then chemical reactions, because there is just one state at any temperature:
the crystal in phase α or the crystal in phase β.

Problem set problem

Einstein model for crystalline phases α and β, frequencies να , νβ
Binding energy per atom: −εα, −εβ

Phase α Phase β
0

−εβ
−εα

Binding energy is like dissociation energy. It includes zero-point vibrational

energy, so no need to account for this separately.

You can predict the phase transition temperature based on a simple first

principles model.

Phase transition only occurs if the crystal with stronger binding energy, e.g. α

phase, also has higher vibrational frequency.

Note that this treatment assumes complete cooperativity: the crystal is either

all one or all the other phase. This gives rise to an abrupt T-dependence of the

equilibrium.

Many intermediate cases occur.

5.60 Spring 2008 Lecture #29 page 6

Example: helix-coil transition in biopolymers. Interactions make few long

segments of helix or coil more stable than many short segments. (Similar for

magnetic and nonmagnetic domains in a ferromagnetic crystal, and in all sorts of

interacting systems). In this case the T-dependence is not completely abrupt,

but not nearly as gradual as in the non-interacting limit.

Extra example (using QM result for qtrans)

Can also calculate solid-gas phase equilibria, i.e. vapor pressure over the crystal.

Assume monatomic ideal gas – only translational partition function.

N
⎡⎛ 2πmkT ⎞3 2 ⎤ ⎛ ∂Ag ⎞
Q =q g N
trans N! = ⎢⎜ ⎟ V⎥ N! A = −kTln Q
g g
μ =⎜g
⎟
⎣⎢⎝ h
2
⎠ ⎦⎥ ⎝ ∂N ⎠T,V
32
⎛ 2πmkT ⎞
ln Q = Nln ⎜
g
⎟ V − Nln N + N
⎝ h
2
⎠
32 32
⎛ ∂ln Q g ⎞ ⎛ 2πmkT ⎞ ⎛ 2πmkT ⎞ V
⎜ ⎟ = ln ⎜ ⎟ V − ln N − 1 + 1 = ln ⎜ ⎟
⎝ h ⎝ h N
2 2
⎝ ∂N ⎠T,V ⎠ ⎠
32
⎛ 2πmkT ⎞ kT
μ g = −kTln ⎜ ⎟ (ideal gas)
⎝ h p
2
⎠

Condition for phase equilibrium: μs = μg

32 52
⎛ kT ⎞ ⎛ 2πm ⎞ (kT )
μ = −ε − 3kT ln ⎜
s g
⎟ = μ = −kT ln ⎜ 2 ⎟
⎝ hνE ⎠ ⎝ h ⎠ p
3 32
⎛ 2πm ⎞
32
(kT )5 2 ⎛ hνE ⎞ (2πm ) ν3E
ε = kT ln ⎜ 2 ⎟ ⎜ kT ⎟ = kT ln
(kT ) p
12
⎝ h ⎠ p ⎝ ⎠
32
(2πm ) ν3E
p=
(kT ) e ε kT
12

gives vapor pressure p(T) over the crystal! Strong binding energy or low T give
low pressure as expected. Low vibrational frequency also gives low p, since this
allows higher entropy in the crystal.
So you can calculate the p-T phase diagram that you described before in
macroscopic terms only.