Table of Contents [ttc]

1. Equilibrium Thermodynamics I: Introduction

• Thermodynamics overview. [tln2]
• Preliminary list of state variables. [tln1]
• Physical constants. [tsl47]
• Equations of state. [tln78]
• Equation of state for ideal gas and real fluid. [tsl12]
• Classification of thermodynamic systems. Laws of thermodynam-
ics. [tln10]
• Thermodynamic processes (irreversible, quasi-static, adiabatic).
[tln79]
• Fast heat. [tex143]
• Expansion and compression of nitrogen gas. [tex144]
• Bathtub icebreaker. [tex145]
• Exact differentials. Applications to internal energy and entropy.
[tln14]
• Exact and inexact differentials I. [tex5]
• Exact and inexact differentials II. [tex146]

2. Equilibrium Thermodynamics II: Engines

• Carnot engine. [tln11]
• Maximum efficiency. [tln12]
• Absolute temperature. [tln13]
• Entropy change caused by expanding ideal gas. [tex1]
• Carnot cycle of the classical ideal gas. [tex3]
• Carnot cycle of an ideal paramagnet. [tex4]
• Reversible processes in fluid systems. [tln15]
• Adiabates of the classical ideal gas. [tex7]
 • Roads from 1 to 2: isothermal, isentropic, isochoric, isobaric.
[tex25]
• Room heater: electric radiator versus heat pump. [tex13]
• Mayer’s relation for the heat capacities of the classical ideal gas.
[tex12]
• Positive and negative heat capacities. [tex26]
• Work extracted from finite heat reservoir in infinite environment.
[tex9]
• Work extracted from finite heat reservoir in finite environment.
[tex10]
• Heating the air in a room. [tex2]
• Gasoline engine. [tln65]
• Idealized gasoline engine(Otto cycle). [tex8]
• Diesel engine. [tln66]
• Idealized Diesel engine. [tex16]
• Escher-Wyss gas turbine. [tln75]
• Joule cycle. [tex108]
• Stirling engine. [tln76]
• Idealized Stirling cycle. [tex131]
• Ideal-gas engine with two-step cycle I. [tex106]
• Ideal-gas engine with two-step cycle II. [tex107]
• Circular heat engine I. [tex147]
• Circular heat engine II. [tex148]
• Square heat engine. [tex149]

3. Equilibrium Thermodynamics III: Free Energies

• Fundamental equation of thermodynamics. [tln16]
• Free energy. [tln3]
• Retrievable and irretrievable energy put in heat reservoir. [tex6]
• Legendre transform. [tln77]
 • Thermodynamic potentials. [tln4]
• Alternative set of thermodynamic potentials. [tln9]
• Thermodynamic functions. [tln5]
• Maxwell’s relations. [tln17]
• Free energy stored and retrieved. [tln18]
• Useful relations between partial derivatives. [tln6]
• Response functions (thermal, mechanical, magnetic). [tln7] (2)
• Isothermal and adiabatic processes in fluid systems and magnetic
systems. [tln8]
• Conditions for thermal equilibrium. [tln19]
• Stability of thermal equilibrium. [tln20]
• Jacobi transformation. [tln21]
• Entropy of mixing. [tln25]
• Osmotic pressure. [tln26]

4. Equilibrium Thermodynamics IV: Applications

• Entropy and internal energy of the classical ideal gas. [tex14]
• Thermodynamic potentials of the classical ideal gas. [tex15]
• Chemical potential of the classical ideal gas. [tex17]
• Ideal gas heat capacity by design. [tex35]
• Sound velocity in the classical ideal gas I. [tex18]
• Sound velocity in the classical ideal gas II. [tex99]
• Absolute temperature from measurements. [tex134]
• Polytropic process of classical ideal gas. [tex138]
• Heavy piston. [tex141]
• Isothermal atmosphere. [tex150]
• Adiabatic atmosphere. [tex151]
• Homogeneous atmosphere. [tex152]
• Van der Waals equation of state. [tln22]
 • Cooling gases: Joule effect (free expansion) and Joule-Thomson
effect (throttling). [tln23]
• Joule-Thomson inversion curves. [tsl1]
• Heat capacities of the van der Waals gas. [tex27]
• Internal energy and entropy of the van der Waals gas. [tex38]
• Joule coefficient of the van der Waals gas. [tex31]
• Joule-Thomson coefficient of the van der Waals gas. [tex32]
• Assembling thermodynamic information. [tex29]
• How not to modify the ideal gas equation of state. [tex11]
• Reconstructing the equation of state of a fluid system. [tex42]
• Reconstructing the equation of state of a gas. [tex43]
• Effects of first virial correction on ideal gas properties. [tex33]
• Entropy due to electronic spins in iron ammonium alum. [tsl2]
• Adiabatic demagnetization. [tln24]
• Thermodynamics of an ideal paramagnet I. [tex19]
• Thermodynamics of an ideal paramagnet II. [tex20]
• Thermodynamics of an ideal paramagnet III. [tex21]
• Thermodynamics of a real paramagnet. [tex36]
• Thermodynamics of a classical ideal paramagnetic gas I. [tex22]
• Thermodynamics of a classical ideal paramagnetic gas II. [tex133]
• Hydrostatic pressure. [tex132]
• Rubber band heat engine. [tex39]
• Equation of state and adiabate of an elastic band. [tex40]
• Determining CV of condensed matter. [tex28]
• Thermodynamics of blackbody radiation. [tex23]
• Carnot cycle of thermal radiation. [tex24]

5. Thermodynamics of Phase Transitions I

• Phase diagram of a “normal” substance. [tsl3]
 • Phase diagram of H2 O. [tsl4]
• Ferrimagnetic phases. [tsl49]
• Liquid crystal phases. [tsl51]
• Ordering of surfactant molecules. [tsl50]
• Phase coexistence: Gibbs phase rule. [tln27]
• Classification of phase transitions. [tln28]
• Gibbs free energy and derivatives at discontinuous transition. [tsl7]
• Gibbs free energy and derivatives at continuous transition. [tsl8]
• Clausius-Clapeyron equation. [tln29]
• Entropy of a supercooled liquid. [tex30]
• Coexistence line of continuous phase transition. [tex37]
• Latent Heat and response functions. [tex124]
• Heat capacity of vapor in equilibrium with liquid phase. [tex41]
• Discontinuous transition: change in internal energy. [tex123]
• Dry ice. [tex125]
• Abnormal phase behavior. [tex54]
• Melting or freezing. [tex51]
• Triple-point phase changes. [tex52]
• Cooling down? Heating up? [tex153]
• Phase coexistence of ammonia. [tex55]

6. Thermodynamics of Phase Transitions II

• Van der Waals equation of state with coexistence curve. [tsl10]
• Law of corresponding states. [tln30]
• Maxwell construction. [tln31]
• Gibbs and Helmholtz free energies of the van der Waals fluid at
T < Tc . [tsl11]
• Condensation and evaporation. [tln32]
• Dieterici equation of state. [tex34]
 • Helium liquids. [tln33]
• Phase diagram of 4 He. [tsl13]
• Phase diagram of 3 He. [tsl14]
• Exotic properties of helium II. [tln34]
• Superconducting transition. [tln35]
• Thermodynamics of a ferromagnet. [tsl5]
• Structural transitions of iron. [tex53]
• Latent heat and heat capacities at superconducting transition.
[tex44]
• Thermodynamics of the mean-field ferromagnet I. [tex45]
• Thermodynamics of the mean-field ferromagnet II. [tex46]

7. Kinetic Theory I
• Statistical uncertainty and information. [tln37]
• Statistical concept of uncertainty. [tex47]
• Statistical uncertainty and information. [tln37]
• Statistical uncertainty and information. [tex48]
• Information of sequenced messages. [tex61]
• Kinetics of classical ideal gas. [tsl28]
• Pressure and mean square velocity in classical ideal gas. [tex49]
• Maxwell velocity distribution. [tln38]
• Maxwell velocity distribution (Maxwell’s derivation). [tex50]
• Maxwell distribution in D-dimensional space. [tex56]
• Boltzmannn equation. [tln39]
• Boltzmann’s H-theorem. [tln40]
• Energy distribution for N ideal gas atoms. [tex57]
• Maxwell velocity distribution (Boltzmann’s derivation). [mex58]
• Ideal-gas entropy and Boltzmann’s H-function. [tex59]
• H-theorem and entropy. [tln41]
 • Boltzmann’s H-function simulated. [tsl27]
• Maxwell distribution derived from minimizing the H-function. [tex60]
• Doppler broadening of atomic spectral lines. [tex63]

8. Kinetic Theory II
• Ideal gas atoms escaping from a container. [tex62]
• Toward thermal equilibrium via particle transfer. [tex64]
• Isotope separation via diffusion. [tex65]
• Kinematic pressure and interaction pressure. [tln42]
• Interaction pressure produced by Gaussian interparticle potential.
[tex66]
• Kinetic forces and mobility. [tln43]
• Average force of particle beam on heavy hard sphere. [tex68]
• Mobility of a hard sphere in a dilute gas. [tex69]
• Collision rate and mean free path. [tln44]
• Collision rate in classical ideal gas. [tex70]
• Mean free path of particle in classical ideal gas. [tex71]
• Rate of chemical reaction A + A → A2 in gas phase. [tex67]
• Effect of escaping particles on temperature of 1D ideal gas. [tex72]

9. Microcanonical Ensemble
• Classical Hamiltonian system. [tln45]
• Classical Liouville operator. [tln46]
• Quantum Liouville operator. [tln47]
• Gibbs entropy. [tln48]
• Microcanonical ensemble. [tln49]
• Classical ideal gas (microcanonical ensemble). [tex73]
• Array of classical harmonic oscillators (microcanonical ensemble).
[tex74]
• Quantum harmonic oscillators (microcanocal ensemble I). [tex75]
 • Quantum harmonic oscillators (microcanocal ensemble II). [tex126]
• Quantum paramagnet (microcanonical ensemble). [tex127]
• Entropy of mixing revisited. [tln50]

10. Canonical Ensemble I

• Canonical ensemble. [tln51]
• Classical ideal gas (canonical ensemble). [tex76]
• Ultrarelativistic classical ideal gas (canonical idela gas). [tex77]
• Ultrarelativistic classical ideal gas in two dimensions. [tex154]
• Array of classical harmonic oscillators (canonical ensemble). [tex78]
• Irreversible decompression. [tex136]
• Irreversible heat exchange. [tex137]
• Reversible decompression. [tex139]
• Reversible heat exchange. [tex140]
• Heavy piston. [tex141]
• Ensemble averages. [tln52]
• Classical virial theorem. [tln83]
• Systems of noninteracting particles. [tln54]
• Further ensemble averages. [tln55]
• Classical ideal gas in a uniform gravitational field. [tex79]
• Gas pressure and density inside centrifuge. [tex135]
• Relative momentum of two ideal gas particles. [tex80]
• Partition function and density of states. [tln56]
• Ideal gas partition function and density of states. [tex81]
• Vibrational heat capacities of solids. [tln57]
• Array of quantum harmonic oscillators (canonical ensemble). [tex82]
• Vibrational heat capacities of solids (Debye theory). [tsl29]
• Thermodynamic perturbation expansion. [tln80]
• Vibrational heat capacity of a solid. [tex83]
 • Anharmonic oscillator and thermodynamic perturbation. [tex104]

11. Canonical Ensemble II

• Paramagnetism. [tln58]
• Paramagnetic salts. [tsl30]
• Fluctuations in a magnetic system. [tln53]
• Fluctuations in a magnetic system. [tex109]
• Classical paramagnet (canonical ensemble). [tex84]
• Quantum paramagnet (two-level system). [tex85]
• Quantum paramagnet (Brillouin function). [tex86]
• Ising trimer. [tex142]
• Negative temperatures. [tsl31]
• Gases with internal degrees of freedom. [tln59]
• Classical rotational free energy of NH3 gas. [tex87]
• Classical rotational entropy of HCl and N2 gas. [tex88]
• Quantum rotational heat capacity of a gas at low temperature.
[tex89]
• Quantum rotational heat capacity of a gas at high temperature.
[tex90]
• Rotational and vibrational heat capacities. [tsl32]
• Orthohydrogen and parahydrogen. [tln81]
• Relativistic classical ideal gas (canonical partition function). [tex91]
• Relativistic classical ideal gas (entropy and internal energy). [tex92]
• Relativistic classical ideal gas (heat capacity). [tex93]
• Relativistic classical ideal gas (heat capacity). [tsl34]

12. Grandcanonical Ensemble

• Grandcanonical ensemble. [tln60]
• Classical ideal gas (grandcanonical ensemble). [tex94]
• Density fluctuations and compressibility. [tln61]
 • Density fluctuations in the grand canonical ensemble. [tex95]
• Density fluctuations and compressibility in the classical ideal gas.
[tex96]
• Energy fluctuations and thermal response functions. [tex103]
• Microscopic states of quantum ideal gases. [tln62]
• Partition function of quantum ideal gases. [tln63]
• Ideal quantum gases: grand potential and thermal averages. [tln64]
• Ideal quantum gases: average level occupancies. [tsl35]
• Occupation number fluctuations. [tex110]
• Density of energy levels for ideal quantum gas. [tex111]
• Maxwell-Boltzmann gas in D dimensions. [tex112]

13. Ideal Quantum Gases I: Bosons

• Bose-Einstein functions. [tsl36]
• Ideal Bose-Einstein gas: equation of state and internal energy.
[tln67]
• BE gas in D dimensions I: fundamental relations. [tex113]
• Reference values for T , V /N , and p. [tln71]
• Bose-Einstein condensation. [tsl38]
• Ideal Bose-Einstein gas: isochores. [tsl39]
• BE gas in D dimensions II: isochore. [tex114]
• BE gas in D dimensions III: isotherm and isobar. [tex115]
• Bose-Einstein gas: isotherms. [tsl40]
• Bose-Einstein gas: isobars. [tsl48]
• Bose-Einstein gas: phase diagram. [tln72]
• Bose-Einstein heat capacity. [tsl41]
• BE gas in D dimensions IV: heat capacity at high temperature.
[tex97]
• BE gas in D dimensions V: heat capacity at low temperature.
[tex116]
 • BE gas in D dimensions VI: isothermal compressibility. [tex128]
• BE gas in D dimensions VII: isobaric expansivity. [tex129]
• BE gas in D dimensions VIII: speed of sound. [tex130]
• Ultrarelativistic Bose-Einstein gas. [tex98]
• Blackbody radiation. [tln68]
• Statistical mechanics of blackbody radiation. [tex105]

14. Ideal Quantum Gases II: Fermions

• Fermi-Dirac functions. [tsl42]
• Ideal Fermi-Dirac gas: equation of state and internal energy. [tln69]
• Ideal Fermi-Dirac gas: chemical potentia.l [tsl43]
• FD gas in D dimensions: chemical potential I. [tex117]
• FD gas in D dimensions: chemical potential II. [tex118]
• Ideal Fermi-dirac gas: average level occupancy. [tsl44]
• Ideal Fermi-Dirac gas: isochores I. [tsl46]
• FD gas in D dimensions: statistical interaction pressure. [tex119]
• Ideal Fermi-Dirac gas: isotherms. [tln70]
• FD gas in D dimensions: isotherm and adiabate. [tex120]
• FD gas in D dimensions: ground-state energy. [tex102]
• Ideal Fermi-Dirac gas: heat capacity. [tsl45]
• FD gas in D dimensions: heat capacity at high temperature.
[tex100]
• FD gas in D dimensions: heat capacity at low temperature. [tex101]
• Ideal Fermi-Dirac gas: isochores II. [tln73]
• Ideal Fermi-Dirac gas: phase diagram in infinite dimensions. [tln74]
• Stable white dwarf. [tex121]
• Unstable white dwarf. [tex122]
Contents of this Document [ttc1]

1. Equilibrium Thermodynamics I: Introduction

• Table of contents (entire course) [ttc]

• Thermodynamics overview. [tln2]
• Preliminary list of state variables. [tln1]
• Physical constants. [tsl47]
• Equations of state. [tln78]
• Equation of state for ideal gas and real fluid. [tsl12]
• Classification of thermodynamic systems. Laws of thermodynamics.
[tln10]
• Thermodynamic processes (irreversible, quasi-static, adiabatic). [tln79]
• Fast heat. [tex143]
• Expansion and compression of nitrogen gas. [tex144]
• Bathtub icebreaker. [tex145]
• Exact differentials. Applications to internal energy and entropy. [tln14]
• Exact and inexact differentials I. [tex5]
• Exact and inexact differentials II. [tex146]
 Thermodynamics overview [tln2]

equilibrium state
quasi−static equations of state canonical
processes ensemble

irreversible and phase coexistence,

reversible changes phase transitions microcanonical grandcanonical
ensemble ensemble
isolated, closed,
and open systems

thermodynamic gateway to equilibrium

extensive and state variables potentials
intensive variables
statistical mechanics

thermodynamic Legendre transforms

limit thermodynamic functions
Maxwell relations
Gibbs−Duhem equation (first derivatives)
exact and inexact
differentials
laws of fundamental equation
thermodynamics of thermodynamics response functions stability conditions
(second derivatives)

first law:energy second law: entropy homogeneous functions

Preliminary list of state variables [tln1]

Thermodynamic functions
X: generic extensive variable Y: generic intensive variable
V : volume p: pressure
S: entropy T: temperature
M : magnetization H: magnetic field
N : number of particles µ: chemical potential
n: number of moles µ̄: chemical potential

Thermodynamic potentials
U: internal energy
E: enthalpy
A: Helmholtz free energy
G: Gibbs free energy
Ω: grand potential

Thermodynamic response functions

CV : isochoric heat capacity cV : isochoric specific heat
Cp : isobaric heat capacity cp : isobaric specific heat
χT : isothermal susceptibility
χS : adiabatic susceptibility
κT : isothermal compressibility
κS : adiabatic compressibility
αp : thermal expansivity
Physical Constants [tsl47]

[A. H. Carter 2001]

Equations of state [tln78]

The empirical specification of a thermodynamic system is traditionally ex-

pressed in the form of two kinds of equations of state.

• Thermodynamic equation of state:

Functional relation between thermodynamic variables.
For example: pV = nRT (classical ideal gas).
• Caloric equation of state:
Temperature dependence of internal energy or heat capacity.
For example: U = CV T with CV = const (classical ideal gas).

The complete thermodynamic information about a system is encoded in this

dual specification. In more complex systems the thermodynamic equation of
state consists of multiple relations.
The most concise way of encoding the complete specification of a thermo-
dynamic system is in the form of a thermodynamic potential. All thermo-
dynamic quantities of interest about a given system can directly be derived
from a thermodynamic potential.

Strategies commonly pursued:

• Equilibrium thermodynamics:
Construct a thermodynamic potential from the empirical information
contained in the thermodynamic and caloric equations of state. Then
derive any thermodynamic quantity of interest from the thermody-
namic potential.
• Equilibrium statistical mechanics:
Derive a thermodynamic potential (or partition function) from the mi-
croscopic specification of the system in the form of a many-body Hamil-
tonian. Then derive any thermodynamic quantity of interest from the
thermodynamic potential.
Equations of state for ideal gas and real fluid [tsl12]

classical ideal gas

argon

[from Kubo: Thermodynamics]

Classification of thermodynamic systems [tln10]

Criterion: Thermodynamic contact.

1. Mechanical interaction (with work source).

Exchange of energy via work performance.
2. Thermal interaction (with heat reservoir).
Exchange of energy via heat transfer.
3. Mass interaction (with particle reservoir).
Exchange of energy via matter transfer.

Isolated system: contact 1.

Closed system: contacts 1, 2.
Open system: contacts 1, 2, 3.

Laws of thermodynamics

Zeroth law: Two systems, each in thermal equilibrium with a third system,
are in thermal equilibrium with each other.

• Prerequisite for measurement of thermodynamic properties.

• Thermal equilibrium implies uniform temperature.
• Mechanical equilibrium implies uniform pressure.
• Chemical equilibrium implies uniform chemical potential.

First Law: Energy is conserved.

• Internal energy U is a state variable.

• Heat and work are not state variables.

Second Law: Heat flows spontaneously from high to low temperatures.

• Entropy S is a state variable.

• Efficiency of heat engines.
• Reversibility and irreversibility.
• Definition of absolute temperature T .

Third Law: ∆S → 0 as T → 0 for any process.

• No cooling to T = 0 in a finite number of steps.

Thermodynamic processes [tln79]

The study of equilibrium thermodynamics cannot do without processes that

connect equilibrium states. Processes necessarily disturb the equilibrium.

Generic process:
• During a generic process between equilibrium states some of the ther-
modynamic variables may not be defined.
• Information about changes in all thermodynamic variables during a
generic process can be obtained if we connect the same initial and final
equilibrium states by a quasi-static process.

Quasi-static process:
• A quasi-static process involves infinitesimal steps between equilibrium
states along a definite path in the space of state variables.
• The equations of state remain satisfied as the thermodynamic variables
change during a quasi-static process.

Adiabatic process:
• During an adiabatic process the system is thermally isolated. There is
no heat transfer. Changes are caused by work performance.
• An adiabatic process must not be too fast in order not to produce
entropy within the system.
• In some practical applications, an adiabatic process must not be too
slow in order to prevent significant heat exchange between the system
and the environment.

Reversible and irreversible processes:

• In an isolated system the entropy (to be defined) stays constant during
a reversible process and increases during an irreversible process.
• Quasi-static processes can be reversible or irreversible.
[tex143] Fast heat
A lead bullet of 20g mass traveling at 784mph is being lodged into a block of wood.
(a) How many calories of heat are generated in the process?
(b) If half that heat goes into the bullet, what rise in temperature will it experience?

Solution:
[tex144] Expansion and compression of nitrogen gas
One mol of N2 at 25◦ C undergoes isothermal expansion from 1.0bar to 0.132bar pressure.
(a) How much work does the gas perform during expansion?
(b) If the gas is then adiabatically compressed by the same amount of work, what will be its final
temperature?
Assume that both processes are quasistatic.

Solution:
[tex145] Bathtub icebreaker
How long does it take a heater coil operating at 500W to melt 1kg of ice initially at 0◦ F?

Solution:
Exact differentials [tln14]

∂F ∂F
Total differential of a function F (x1 , x2 ): dF = dx1 + dx2 .
∂x1 ∂x2

The differential dF = c1 (x1 , x2 )dx1 + c2 (x1 , x2 )dx2 is exact if dF is the total

differential of a function F (x1 , x2 ).
∂2F ∂2F ∂c1 ∂c2
Condition: = ⇒ = .
∂x1 ∂x2 ∂x2 ∂x1 ∂x2 ∂x1

Z (b1 ,b2 ) I
Consequences: dF = F (b1 , b2 ) − F (a1 , a2 ), dF = 0.
(a1 ,a2 )

Internal energy U
H
U is a state variable. dU = 0 for reversible cyclic processes.
dU = δQ + δW + δZ = T dS + Y dX + µdN is an exact differential.
δQ = T dS: heat transfer
δW = Y dX: work performance (−pdV + HdM + . . .)
δZ = µdN : matter transfer

Entropy S
|∆QL | TL ∆QL ∆QH
Carnot cycle: = ⇒ + = 0.
∆QH TH TL TH
Any reversible cyclic process is equivalent to an array of Carnot cycles run-
ning in parallel.
I I
δQ
⇒ ≡ dS = 0 for reversible cyclic processes.
T
S is a state variable.

Irreversible process:
|∆QL | TL |∆QL | TL ∆QL ∆QH
η =1− <1− ⇒ > ⇒ + < 0.
∆QH TH ∆QH TH TL TH
I I
δQ δQ
More general cyclic process: < 0, dS = 0 ⇒ dS > .
T T
Irreversible process in isolated system: δQ = 0 ⇒ dS > 0.
[tex5] Exact and inexact differentials I
2
(a)
H Show that the differential dF1 = xydx + x dy is inexact. Determine the value of the integral
dF1 along the closed path shown.
(b) Show that dF2 = (2y 2 −3x)dx+4xydy is an exact differential. Determine the function F2 (x, y).

y
3

1
x
0 1 3
Solution:
[tex146] Exact and inexact differentials II

(a)
H Show that the differential dF1 = (x + y)dx + ydy is inexact. Determine the value of the integral
dF1 along the circular path shown in counterclockwise direction.
(b) Show that dF2 = (x + y)dx + xdy is an exact differential. Determine the function F2 (x, y).

y0 r

x
0 x0

Solution:
Contents of this Document [ttc2]

2. Equilibrium Thermodynamics II: Engines

• Carnot engine. [tln11]

• Maximum efficiency. [tln12]
• Absolute temperature. [tln13]
• Entropy change caused by expanding ideal gas. [tex1]
• Carnot cycle of the classical ideal gas. [tex3]
• Carnot cycle of an ideal paramagnet. [tex4]
• Reversible processes in fluid systems. [tln15]
• Adiabates of the classical ideal gas. [tex7]
• Roads from 1 to 2: isothermal, isentropic, isochoric, isobaric. [tex25]
• Room heater: electric radiator versus heat pump. [tex13]
• Mayer’s relation for the heat capacities of the classical ideal gas. [tex12]
• Positive and negative heat capacities. [tex26]
• Work extracted from finite heat reservoir in infinite environment. [tex9]
• Work extracted from finite heat reservoir in finite environment. [tex10]
• Heating the air in a room. [tex2]
• Gasoline engine. [tln65]
• Idealized gasoline engine(Otto cycle). [tex8]
• Diesel engine. [tln66]
• Idealized Diesel engine. [tex16]
• Escher-Wyss gas turbine. [tln75]
• Joule cycle. [tex108]
• Stirling engine. [tln76]
• Idealized Stirling cycle. [tex131]
• Ideal-gas engine with two-step cycle I. [tex106]
• Ideal-gas engine with two-step cycle II. [tex107]
• Circular heat engine I. [tex147]
• Circular heat engine II. [tex148]
• Square heat engine. [tex149]
Carnot engine [tln11]

Second law: Heat flows spontaneously from high to low temperatures.

Thermal contact: Temperature differences disappear without producing work.
Heat engine: Part of the heat flowing from high to low temperatures is
converted into work via a cyclic process.
Carnot engine: All wasteful heat flows are eliminated (reversible processes).

The four steps of a Carnot process:

• 1 → 2: Isothermal absorption of heat: ∆Q12 > 0 at ΘH .

• 2 → 3: Adiabatic cooling: ΘH → ΘL with ∆Q23 = 0 and ∆W23 < 0.
• 3 → 4: Isothermal expulsion of heat: ∆Q34 < 0 at ΘL .
• 4 → 1: Adiabatic heating: ΘL → ΘH with ∆Q41 = 0 and ∆W41 > 0.

Total heat input: ∆Qin = ∆Q12 .

Use first law: ∆U = ∆Q12 + ∆W12 + ∆W23 + ∆Q34 + ∆W34 + ∆W41 = 0.
Net work output: ∆Wout ≡ −∆W12 −∆W23 −∆W34 −∆W41 = ∆Q12 −|∆Q34 |
∆Wout |∆Q34 |
Efficiency: η ≡ =1− .
∆Qin ∆Q12

Y
ΘH
1 ΘH
∆Q12 2
∆Wout
C ΘL
4
∆Q34 3

ΘL
X
Maximum efficiency [tln12]

Is it possible to construct a heat engine A which is more efficient than the

Carnot engine C?
Use engine A to drive engine C in the reverse i.e. as a refrigerator.

ΘH
∆ QA ∆Q12
∆W
A C

∆W ∆QA ∆Q34
ΘL

Heat transfers: ∆QA > 0, ∆Q12 < 0, ∆Q34 > 0.

(A) (C)
Work performance: ∆W = ∆Wout = ∆Win > 0.
∆W ∆W
Efficiencies: ηA = , ηC =
∆QA |∆Q12 |
Since engine C operates reversibly, ηC is the same in the forward and reverse
directions. Note: ηC is not an efficiency in the reverse mode.
ηA > ηC would imply ∆QA < |∆Q12 |.
The two engines combined would then cause heat to flow from low to high
temperature without work input, which is a violation of the second law.

Conclusions:

• Engine A cannot be more efficient than engine C.

• All Carnot engines operating between ΘH and ΘL must have the same
efficiency.
Absolute temperature [tln13]

Reservoir temperatures: ΘH , ΘM , ΘL .
|∆QL |
Efficiency: η = 1 − = 1 − f (ΘL , ΘH ).
∆QH
∆Q̄M |∆Q̄L |
Likewise: = f (ΘM , ΘH ), = f (ΘL , ΘM ).
∆Q̄H ∆Q̄M

ΘH

∆Q H
C ∆Q H

∆Q M
ΘM C

∆Q M
C ∆Q L
∆Q L

ΘL

Second law implies: If ∆Q̄L = ∆QL then ∆Q̄H = ∆QH .

|∆Q̄L | ∆Q̄M |∆QL |
⇒ = ⇒ f (ΘL , ΘM )f (ΘM , ΘH ) = f (ΘL , ΘH )
∆Q̄M ∆Q̄H ∆QH
g(ΘL ) TL TL
Functional form: f (ΘL , ΘH ) = ≡ ⇒ η =1− .
g(ΘH ) TH TH

TL |∆QL |
Definition of absolute temperature: = .
TH ∆QH
Kelvin scale is fixed by triple point of water: Ttrp = 273.16K.

Note: η = 1 implies TL = 0. However, the third law states δQ = T dS = 0 at

T = 0. Hence, all reversible processes at T = 0 are adiabatic. Heat cannot
be absorbed reversibly at T = 0.
[tex1] Entropy change caused by expanding ideal gas
Consider the amount n = 1mol of a classical ideal gas in a box of volume V1 with heat-conducting
walls. The gas is described by the equation of state pV = nRT and the internal energy U = CV T
with CV = const. Now we let the gas expand to the volume V2 = 2V1 via two different processes:
(a) by quasi-static isothermal expansion;
(b) by leakage through a hole in one wall.
Calculate the change in entropy ∆SG of the gas and ∆SE of the environment during each process.
Express the results in SI units.

(a) V1 (b) V1

Solution:
[tex3] Carnot cycle of the classical ideal gas
Consider the four steps of a Carnot engine with the operating material in the form of a classical
ideal gas [pV = nRT , U = CV T with CV = const].
(a) Determine the heat transfer, ∆Q, the work performance, ∆W , and the change in internal
energy, ∆U , for each of the four steps:
1 → 2 isothermal expansion: T = TH = const, V2 > V1 .
2 → 3 adiabatic expansion: S = const, V3 > V2 .
3 → 4 isothermal compression: T = TL = const, V4 < V3 .
4 → 1 adiabatic compression: S = const, V1 < V4 .
(b) Sketch the Carnot cycle in the (V, p)-plane and in the (U, S)-plane.
(c) Show that the efficiency is ηC = 1 − TL /TH .

Solution:
[tex4] Carnot cycle of an ideal paramagnet
Consider the four steps of a Carnot engine with the operating material in the form of an ideal
paramagnet. The equation of state is Curie’s law, M = DH/T , where H is the magnetic field, T
the absolute temperature, and D a constant. The internal energy is a monotonically increasing
function, U (T ), of temperature.
(a) Determine the heat transfer, ∆Q, the work performance, ∆W , and the change in internal
energy, ∆U , for each of the four steps:
1 → 2 isothermal demagnetization: T = TH = const, M2 < M1 .
2 → 3 adiabatic demagnetization: S = const, M3 < M2 .
3 → 4 isothermal magnetization: T = TL = const, M4 > M3 .
4 → 1 adiabatic magnetization: S = const, M1 > M4 .
(b) Sketch the Carnot cycle in the (M, H)-plane and in the (U, S)-plane.
(c) Show that the efficiency is ηC = 1 − TL /TH .

Solution:
Reversible processes in fluid system [tln15]

Isothermal process: T = const. δQ 6= 0 in general.

Isochoric process: V = const. δQ = CV dT, dU = CV dT .
Isobaric process: p = const. δQ = Cp dT .
Isentropic (adiabatic) process: S = const. δQ = 0.

Internal energy: dU = δQ + δW = T dS − pdV .

• V = const. ⇒ δW = 0 ⇒ dU = δQ (no work performed).

• S = const. ⇒ δQ = 0 ⇒ dU = δW (no heat transferred).

Classical ideal gas:

Equation of state: pV = nRT .
Internal energy: U = CV T, CV = αnR = const.
Isotherm: T = const. ⇒ pV = const.
Adiabate: S = const. ⇒ pV γ = const., γ = 1 + 1/α

• monatomic gas: α = 32 , γ = 53 .
• diatomic gas: α = 25 , γ = 75 .
• polyatomic gas: α = 3, γ = 43 .

p
p = const
V = const

T = const

S = const
V
[tex7] Adiabates of the classical ideal gas
The classical ideal gas is specified by the thermodynamic equation of state pV = nRT and by the
internal energy (caloric equation of state) U = CV T with CV = αnR = const [α = 32 (monatomic),
α = 52 (diatomic), α = 3 (polyatomic)]. A reversible process with S = const is called isentropic
or adiabatic and is characterized by the curve pV γ = const. No heat is exchanged in an adiabatic
process: dU = δW , δQ = 0. Find γ as a function of α.

Solution:
[tex25] Roads from 1 to 2: isothermal, isentropic, isochoric, isobaric
The amount n = 1mol of an ideal gas undergoes three different quasistatic processes (see Figure)
from the initial state (p1 , V1 , T1 ) to the final state (p2 , V2 , T2 ):
(i) 1 → A → 2; (ii) 1 → B → 2; (iii) 1 → C → 2.
Find the work ∆W done on the system and the heat ∆Q added to the system in each process.
Express all results in terms of (T1 , V1 , T2 , V2 ).

p
1

isothermal

isochoric
B

isentropic C

A isobaric 2
V
V1 V2

Solution:
[tex13] Room heater: Electric radiator versus heat pump
A room is to be kept at temperature TH = 294K, (21◦ C). The outdoor temperature is TL . Heat,
which leaks through the windows and doors at the rate Q̄leak = γ∆T , must be replaced by a
room heater at the same rate. The electric radiator converts electric power W̄el into heat with
100% efficiency. The electric heat pump uses the amount W̄sup of electric power to drive a Carnot
cycle in the reverse, which extracts heat Q̄L at temperature TL from the exterior and converts it
(reversibly) into heat Q̄H = Q̄L + W̄hp at temperature TH . In the relation W̄hp = (1 − λ)W̄sup , λ
represents the energy loss in the gears of the heat pump. Quantitites with overbars denote energy
transfers per time unit.
(a) Find W̄el as a function of γ, TH , TL , and W̄sup as a function of γ, λ, TH , TL .
(b) Plot W̄el /γ and W̄sup /γ versus tL ≡ TL − 273K (measured in ◦ C) for fixed TH = 294K and
λ = 0.8 (20% efficiency).
(c) Determine the range of TL where the heat pump is more economical than the radiator.

Solution:
[tex12] Mayer’s relation for the heat capacities of the classical ideal gas
The amount n = 1mol of a classical ideal gas [pV = nRT , U = CV T with CV = const] is initially
confined to a volume V1 at pressure p1 . In step 1 → 2 of Mayer’s cycle, the gas undergoes free
expansion to volume V2 while it is thermally isolated (δQ = 0, δW = 0) The pressure decreases
from p1 to p2 during this step. In step 2 → 3 the gas is quasi-statically compressed back to volume
V1 , while the pressure is maintained at p2 . With the temperature decreasing during this step, heat
is expelled. In step 3 → 1 the gas is heated up quasi-statically at constant volume V1 until the
pressure returns to p1 . Use the first law to derive Mayer’s relation, Cp − CV = R, between the
heat capacities of the classical ideal gas.

p
1
p
1

p 3 2
2
V
V1 V2

Solution:
[tex26] Positive and negative heat capacities
The p − V diagram shows an isotherm and an adiabate for the classical ideal gas. Show that a
quasistatic process of the type 1 → 2 is characterized by a positive heat capacity and a process of
the type 3 → 4 by a negative heat capacity.

adiabate
3
2

isotherm

1 4

V
Solution:
[tex9] Work extracted from finite heat reservoir in infinite environment
A (finite) heat reservoir with heat capacity C = const is initially at temperature TH and the (infi-
nite) environment at the lower temperature T0 . Now the reservoir is connected to the environment
by a heat engine, which absorbs an infinitesimal amount of heat δQ per cycle, converts part of
it into work δW , and dumps the rest into the environment. During each cycle the temperature
of the reservoir decreases infinitesimally: δQ = −CdT . Determine the maximum amount of work
∆W that can be extracted from the reservoir before its temperature has dropped to that of the
environment. The fraction of the excess internal energy Uex = C(TH − T0 ) that can be converted
into work is characterized by the quantity ∆W/Uex . Plot this quantity versus the reduced tem-
perature (TH − T0 )/T0 for T0 < TH < 3T0 . Set TH /T0 = 1 +  with   1 and find the dependence
of ∆W/Uex on  to leading order.

Solution:
[tex10] Work extracted from finite heat reservoir in finite environment
A (finite) heat reservoir with heat capacity CH = const is initially at temperature TH and the
(finite) environment with heat capacity CL at the lower temperature TL .
(a) When heat is allowed to flow from the reservoir to the environment, both will end up at the
temperature Tf = (CH TH + CL TL )/(CH + CL ) (arithmetic mean). Verify this and determine the
total amount of heat ∆Q that has been transferred.
(b) When the reservoir is connected to the environment by a Carnot engine which absorbs an
infinitesimal amount of heat δQ per cycle, converts part of it into work δW , and dumps the rest
into the environment, the final common temperature of the reservoir and the environment will be
C /(C +C ) C /(C +C )
Tf = TH H H L TL L H L (geometric mean). Verify this and determine the total amout of
work ∆W that has been extracted from the system. The fraction of the excess internal energy
Uex = CH (TH − TL ) that can be converted into work is characterized by the quantity ∆W/Uex .
Plot this quantity versus the reduced temperature (TH −TL )/TL for CH = CL and TL < TH < 3TL .
Discuss the properties of this function in the limit TH → TL .

Solution:
[tex2] Heating the air in a room
Calculate the amount of energy ∆Q that must be supplied to heat the air in a room from 0◦ C
to 20◦ C under three different circumstances. For each case, calculate also the change in internal
3
energy ∆U of the air in the room. Mass density of air at STP (0◦ C and 1atm): ρ = 0.00129g/cm .
Specific heats of air: cV = 0.169cal/gK, cp /cV ≡ γ = 1.41. Express all results in SI units.
(a) The room has rigid, insulating walls. The volume is 27m3 . The initial pressure is 1atm.
(b) The room has insulating walls. One wall is mobile. The process takes place at constant pressure
(1atm). The initial volume is 27m3 .
(c) The room has rigid, insulating walls. The volume is 27m3 . One wall has a small hole through
which air leaks out slowly. The process takes place at constant pressure (1atm).

Solution:
 Gasoline engine (Otto cycle) [tln65]

p p

3 3

2 4 2
1’ 4

5 1 5 1
V V
V2 V1 V2 V1

Four-stroke Otto cycle (left)

1-2: compression stroke

2-3-4: power stroke (spark plug ignites at 2)
4-1’-5: exhaust stroke (exhaust valve opens at 4)
5-1: intake stroke (intake valve opens at 5)

Idealized Otto cycle (right)

1-2: adiabatic compression of air-fuel mixture (S = const)

2-3: explosion of air-fuel mixture (V = const)
3-4: adiabatic expansion of exhaust gas (S = const)
4-1: isochoric release of exhaust gas (V = const).
1-5-1: intake stroke (thermodynamically ignored)

.
Parameter: K = V1 /V2 (compression ratio).
The compression ratio K must not be chosen too large to prevent the air-fuel
mixture from igniting spontaneously, i.e. prematurely.
[tex8] Idealized Otto cycle
Consider the four steps of the idealized Otto cycle for a classical ideal gas [pV = nRT , U = CV T
with CV = αnR].
(a) Determine the heat transfer, ∆Q, the work performance, ∆W , and the change in internal
energy, ∆U , for each of the four steps:
1 → 2 adiabatic compression of air-fuel mixture: S = const.
2 → 3 explosion of air-fuel mixture: V = const.
3 → 4 adiabatic expansion of exhaust gas: S = const.
4 → 1 isochoric release of exhaust gas: V = const.
(c) Calculate the efficiency η and express it as a function of the compression ratio K ≡ V1 /V2 .

p
3
S = const

4
2

S = const 1
V
V2 V1

Solution:

1
 Diesel engine [tln66]

p p
2 3 2 3

4
1’ 4

5 1 5 1
V V
V2 V3 V1 V2 V3 V1

Four-stroke Diesel cycle (left)

1-2: compression stroke (fuel injected and spontaneously ignited at 2)

2-3-4: power stroke (Diesel fuel burns more slowly than gasoline)
4-1’-5: exhaust stroke (exhaust valve opens at 4)
5-1: intake stroke (intake valve opens at 5)

Idealized Diesel cycle (right)

1-2: adiabatic compression of air (S = const)

2-3: isobaric expansion as fuel explodes (p = const)
3-4: adiabatic expansion of exhaust gas (S = const)
4-1: isochoric release of exhaust gas (V = const).
1-5-1: intake stroke (thermodynamically ignored)

. .
Parameter: K = V1 /V2 (compression ratio), L = V3 /V2
[tex16] Idealized Diesel cycle
Consider the four steps of the Diesel cycle for a classical ideal gas [pV = nRT , U = CV T with
CV = αnR].
(a) Determine the heat transfer, ∆Q, the work performance, ∆W , and the change in internal
energy, ∆U , for each of the four steps:
1 → 2 adiabatic compression of air: S = const.
2 → 3 isobaric expansion during explosion: p = const.
3 → 4 adiabatic expansion after explosion: S = const.
4 → 1 isochoric release of exhaust gas: V = const.
Calculate the efficiency η and express it as a function of the two parameters K ≡ V1 /V2 and
L ≡ V3 /V2 .

p
2 3
S = const

S = const 1
V

Solution:

1
Escher-Wyss gas turbine [tln75]

A gas flows in a closed system from the boiler via the turbine to the radiator
and then via the compressor back into the boiler. As the beam of hot gas
hits the blades of the turbine during the power stroke it expands with little
heat transfer. The compression of the cooled gas is also roughly adiabatic.
The gas is heated up inside the boiler and cooled down inside the radiator
at different but roughly constant pressures.

p
4 1
p
1

p
2 3 2
V

Idealized process (Joule cycle)

1-2: Adiabatic expansion of the hot gas after ejection from the boiler as it
drives the turbine (S = const).
2-3: Isobaric contraction as the gas flows through the radiator and cools
down further in the process (p = const).
3-4: Adiabatic compression of the cooled gas for injection into the boiler
(S = const).
4-1: Isobaric expansion of the gas as it heats up inside the boiler (p = const).

Notes:

• The pressure inside the boiler is regulated by the rates of gas injection
and ejection and the rate of heat transfer from the energy source to the
gas.
• The injection and ejection rates are the same in mass units but the
ejection rate is larger than the injection rate in volume units. This
accounts for the expansion of the gas inside the boiler as described in
step 4-1.
[tex108] Joule cycle
.
Consider the four steps of the Joule cycle for the classical ideal gas [pV = N kB T, CV = αN kB , γ =
Cp /CV = (α + 1)/α]. It represents an idealized version of the Escher-Wyss gas turbine.
(a) Calculate the work performance, ∆W , the heat transfer, ∆Q, and the change in internal energy,
∆U , for each step.
1 → 2 adiabatic expansion: S = const.
2 → 3 isobaric contraction: p = const.
3 → 4 adiabatic compression: S = const.
4 → 1 isobaric expansion: p = const.
(b) Calculate the efficiency η and express it as a function of the pressure ratio p2 /p1 .
(c) Sketch the Joule cycle in the (U, S)-plane.

p
4 1
p
1

p
2 3 2
V

Solution:

1
Stirling engine [tln76]

The Stirling engine is an external combustion engine. It isolates the working

fluid from the heat source. Combustion is better controlled than in internal
combustion engines.

R 3 TH
TH TL
4
2
D P 1
TL
V
V2 V1

Piston P expands gas at high temperature TH and compresses gas at low

temperature TL .
Displacer D moves gas between regions of high temperature TH and low
temperature TL through the regenerator.
Regenerator R acts as a heat exchanger. It stores heat when hot gas flows
from left to right and releases heat when colder gas flows from right to left.

Idealized Stirling cycle

1-2: Isothermal compression at temperature TL .

Displacer stationary at left. Piston moving left.
2-3: Isochoric heating up at volume V2 .
Piston stationary at left. Displacer moving right.
3-4: Isothermal expansion at temperature TH .
Displacer and piston moving right.
4-1: Isochoric cooling down at volume V1 .
Piston stationary at right. Displacer moving left.

Note: Some of the heat is recycled in the regenerator. This amount should
not be counted in the expression η = ∆Wout /∆Qin of the efficiency.
[tex131] Idealized Stirling cycle
Consider the four steps of the idealized Stirling cycle for the classical ideal gas [pV = N kB T, CV =
.
αN kB , γ = Cp /CV = (α + 1)/α].
(a) Calculate the work performance, ∆W , the heat transfer, ∆Q, and the change in internal energy,
∆U , for each step.
1 → 2 isothermal compression: T = TL ,
2 → 3 isochoric heating up: V = V2 ,
3 → 4 isothermal expansion: T = TH ,
4 → 1 isochoric cooling down: V = V1 .
(b) Calculate the efficiency η and express it as a function of TH and TL .

p
3 TH

4
2

1
TL
V
V2 V1

Solution:
[tex106] Ideal-gas engine with two-step cycle I
.
Consider the two-step cycle for a classical ideal gas [pV = N kB T, CV = αN kB , γ = Cp /CV =
(α + 1)/α] as shown. The first step (A) is an adiabatic compression and the second step (B) an
expansion along a straight line segment in the (V, p)-plane.
.
(a) Show that the difference in internal energy ∆U = U1 − U2 is determined by the expression
"  γ−1 #
∆U V1
=α 1− .
p1 V1 V2

(b) Show that the heat transfer δQ between system and environment during a volume increase
from V to V + δV along the straight line segment is given by the expression

1 − (V1 /V2 )γ
 
δQ V dV
= (1 + α)(1 + σ) − (1 + 2α)σ , σ= .
p1 V1 V1 V1 V2 /V1 − 1

(c) Show that along the straight-line segment the system absorbs heat if V1 < V < Vc and expells
heat if Vc < V < V2 , where Vc /V1 = [(1 + α)(1 + σ)]/[(1 + 2α)σ].

p
p
1

A
p
2
V
V1 V2

Solution:
[tex107] Ideal-gas engine with two-step cycle II
.
Consider the two-step cycle for a classical ideal gas [pV = N kB T, CV = αN kB , γ = Cp /CV =
(α + 1)/α] as previously discussed in [tex106]. For the following we assume that the compression
ratio is V2 /V1 = 2 and that the gas is monatomic (α = 32 ).
(a) Show that the net work output along the adiabate and along the straight line segment are
∆WA /p1 V1 ' 0.55506 and ∆WB /p1 V1 ' −0.65749.
(b) Show that the total heat absorbed during the cycle is ∆Qin /p1 V1 ' 0.39564.
(c) Determine the efficiency η2 of the two-step cycle and compare it with the efficiency ηC of the
Carnot engine operating between the same two temperatures.

p
p
1

A
p
2
V
V1 V2

Solution:
[tex147] Circular heat engine I
Consider 1 mol of a classical ideal gas [pV = RT ] confined to a cylinder by a piston. The cylinder
is in thermal contact with a heat bath of adjustable temperature. As the piston moves back and
forth between volume V = V0 (1 − r) and V = V0 (1 + r) quasistatically, the temperature of the gas
is being adjusted via thermal contact such that the cycle becomes circular in the (V, p)-plane and
proceeds in clockwise direction (φ from 0 to 2π).
(a) Calculate the net work output ∆Wout during one cycle.
(b) Set r = 0.5 and identify the segments along the circle where the temperature of the gas rises
and the segments where it falls.
(c) Repeat the previous part for r = 0.9. Note that there now are more segments.

p/p
0

r φ
1

0 V/V0
0 1

Solution:
[tex148] Circular heat engine II
Consider 1 mol of a monatomic classical ideal gas [pV = RT , U = 32 RT ] confined to a cylinder by
a piston. The cylinder is in thermal contact with a heat bath of adjustable temperature. As the
piston moves back and forth between volume V = V0 (1 − r) and V = V0 (1 + r) quasistatically, the
temperature of the gas is being adjusted via thermal contact such that the cycle becomes circular
in the (V, p)-plane and proceeds in clockwise direction (φ from 0 to 2π).
(a) Calculate the rate dW/dφ at which work is being performed, the rate dU/dφ at which the
internal energy changes, and the rate dQ/dφ at which heat is being transferred.
(b) Set r = 0.5 and identify the segments along the circle where each rate is positive or negative.
(c) Repeat the previous part for r = 0.9.
(d) Plot all three rates as functions of φ/π for r = 0.5 in one graph and then for r = 0.9 in a
second graph.

p/p
0

r φ
1

0 V/V0
0 1

Solution:
[tex149] Square heat engine
Consider 1 mol of a monatomic classical ideal gas [pV = RT , U = 32 RT ] confined to a cylinder by
a piston. The cylinder is in thermal contact with a heat bath of adjustable temperature. The gas
undergoes a quasistatic, cyclic process 1 → 2 → 3 → 4 → 1 as shown. Use p1 , V1 , T1 as units for
pressure, volume, and temperature, respectively.
(a) Find p2 , V2 , T2 , p3 , V3 , T3 , and p4 , V4 , T4 in these units.
(b) Find the work performance, ∆W12 , ∆W23 , ∆W34 , ∆W41 , the change in internal energy, ∆U12 ,
∆U23 , ∆U34 , ∆U41 , and the heat transfer, ∆Q12 , ∆Q23 , ∆Q34 , ∆Q41 , along the legs of the cycle.
Express these quantities in units of RT1 .
(c) Find the net work ∆Wnet performed during the cycle. Find also the heat ∆Qin absorbed and
the heat ∆Qout expelled by the gas during the cycle.
(d) Find the efficiency ηS of this cycle in the role of heat engine.

p/p
1

3
2 3

1 4
1

0 V/V1
0 1 3

Solution:
[tex155] Work performance and heat transfer
Consider 1mol of a classical ideal gas [pV = RT , U = 32 RT ] confined to a cylinder with a piston
and in thermal contact with a heat bath of adjustable temperature. Examine the three qua-
sistatic processes A, B, C, from point (2V1 , T1 ) to point (V1 , 2T1 ) as shown. Calculate the work
∆WA , ∆WB , ∆WC done by the piston on the gas and the heat ∆QA , ∆QB , ∆QC transferred be-
tween the gas and the environment in each process. Choose the signs for work and heat such that
∆U = ∆W + ∆Q. Independently of your calculation, formulate a verbal argument for the ranking
of ∆W and ∆Q of the three processes.

2T1
B

C
A
T1

V
V1 2V1

Solution:
Contents of this Document [ttc3]

3. Equilibrium Thermodynamics III: Free Energies

• Fundamental equation of thermodynamics. [tln16]

• Free energy. [tln3]
• Retrievable and irretrievable energy put in heat reservoir. [tex6]
• Legendre transform. [tln77]
• Thermodynamic potentials. [tln4]
• Alternative set of thermodynamic potentials. [tln9]
• Thermodynamic functions. [tln5]
• Maxwell’s relations. [tln17]
• Free energy stored and retrieved. [tln18]
• Useful relations between partial derivatives. [tln6]
• Response functions (thermal, mechanical, magnetic). [tln7] (2)
• Isothermal and adiabatic processes in fluid systems and magnetic sys-
tems. [tln8]
• Conditions for thermal equilibrium. [tln19]
• Stability of thermal equilibrium. [tln20]
• Jacobi transformation. [tln21]
• Entropy of mixing. [tln25]
• Osmotic pressure. [tln26]
Fundamental equation of thermodynamics [tln16]

The first and second laws of thermodynamics imply that

dU = T dS + Y dX + µdN (1)

with      
∂U ∂U ∂U
= T, = Y, =µ
∂S X,N ∂X S,N ∂N S,X

is the exact differential of a function U (S, X, N ).

Here X stands for V, M, . . . and Y stands for −p, H, . . ..
Note: for irreversible processes dU < T dS + Y dX + µdN holds.

U, S, X, N are extensive state variables.

U (S, X, N ) is a 1st order homogeneous function: U (λS, λX, λN ) = λU (S, X, N ).
∂U ∂U ∂U
U [(1 + )S, (1 + )X, (1 + )N ] = U + S + X + N = (1 + )U.
∂S ∂X ∂N
Euler equation:
U = T S + Y X + µN. (2)

Total differential of (2):

dU = T dS + SdT + Y dX + XdY + µdN + N dµ (3)

Subtract (1) from (3):

Gibbs-Duhem equation: SdT + XdY + N dµ = 0.
The Gibbs-Duhem equation expresses a relationship between the intensive
variables T, Y, µ. It can be integrated, for example, into a function µ(T, Y ).
Note: a system specified by m independent extensive variables possesses
m − 1 independent intensive variables.
Example for m = 3: S, V, N (extensive); S/N, V /N or p, T (intensive).
Complete specification of a thermodynamic system must involve at least one
extensive variable.
Free energy [tln3]

Mechanical system:
Consider a mechanical system in the form of a massive particle moving in a
potential (e.g. harmonic oscillator, simple pendulum).

• Energy can be stored in a mechanical system when some agent does

work on it.
• If the work involves exclusively conservative forces, then all energy
added can be retrieved as work performed by the system.
• If the work done also involves dissipative forces, then only part of the
energy added is retrievable as work performed by the system.
• The energy stored in and retrievable from the system is called mechan-
ical energy (kinetic energy plus potential energy).

Thermodynamic system:
Consider a thermodynamic system in contact with a work source and (op-
tionally) also with a heat reservoir and/or a particle reservoir.

• Energy can be stored in a thermodynamic system when some agent

does work on it.
• If the work is done reversibly, then all the energy added can be retrieved
as work performed by the system.
• If the work is done irreversibly, then only part of the energy added is
retrievable as work performed by the system.
• The energy stored and retrievable as work is called free energy.
• The free energy stored in or retrieved from a thermodynamic system is
expressed by some thermodynamic potential.
• The form of thermodynamic potential depends on the constraints im-
posed on the system during storage and retrieval of free energy.
[tex6] Retrievable and irretrievable energy put in heat reservoir
Consider the amount n = 1mol of a monatomic classical ideal gas inside a cylinder with a piston
on one side. This system is in thermal contact with a heat bath at temperature T0 = 293K. An
external work source pushes the piston from position 1 (V1 = 5m3 ) in to position 2 (V2 = 3m3 )
and then back out to position 1. Calculate the work ∆W12 done by the source during step
1 → 2 and the (negative) work ∆W21 done during step 2 → 1 under three different circumstances:
Compression and expansion of the gas take place (a) quasi-statically, i.e. isothermally; (b) rapidly,
i.e. adiabatically, and in quick succession; (c) adiabatically again, but with a long waiting time
between the two steps.
For each case calculate also the energy EW wasted in the heat bath after one full cycle. Find the
highest temperature TH and the lowest temperature TL reached by the gas in case (c).
The equation of state is pV = nRT , and the heat capacity is CV = 32 nR. During the adiabatic
process: pV γ = const with γ = 53 .

Solution:
Legendre transform [tln77]

Given is a function f (x) with monotonic derivative f 0 (x). The goal is to

replace the independent variable x by p = f 0 (x) with no loss of information.
Note: The function G(p) = f (x) with p = f 0 (x) is, in general, not invertible.
The Legendre transform solves this task elegantly.

• Forward direction: g(p) = f (x) − xp with p = f 0 (x).

• Reverse direction: f (x) = g(p) + px with x = −g 0 (p)

Example 1: f (x) = x2 + 1.

2 0 p p2
• f (x) = x + 1 ⇒ f (x) = 2x ⇒ x= ⇒ g(p) = 1 − .
2 4
p2 p
• g(p) = 1 − ⇒ g 0 (p) = − ⇒ p = 2x ⇒ f (x) = x2 + 1.
4 2

Example 2: f (x) = e2x .

1 p
• f (x) = e2x ⇒ f 0 (x) = 2e2x = p ⇒ x = ln
2 2
p p p
⇒ g(p) = − ln .
2 2 2
p p p 1 p
• g(p) = − ln ⇒ g 0 (p) = − ln = −x
2 2 2 2 2
⇒ p = 2e2x ⇒ f (x) = e2x .
Thermodynamic potentials [tln4]

• All thermodynamic potentials of a system are related to each other via

Legendre transform. Therefore, every thermodynamic potential has its
distinct set of natural independent variables.
• All thermodynamic properties of a system can be inferred from any
of the thermodynamic potentials. Any thermodynamic potential thus
yields a complete macroscopic description of a thermodynamic system.
• Any quantity U, E, A, G, Ω which is not expressed as a function of its
natural independent variables is not a thermodynamic potential and
thus contains only partial thermodynamic information on the system.
• In thermodynamics, we determine thermodynamic potentials from em-
pirical macroscopic information such as equations of state and response
functions.
• In statistical mechanics, we determine thermodynamic potentials via
partition functions from the microscopic specification (Hamiltonian) of
the thermodynamic system.
• Any spontaneous i.e. irreversible process at constant values of the natu-
ral independent variables is accompanied by a decrease of the associated
thermodynamic potential.
• The equilibrium state for fixed values of a set of natural independent
variables is the state where the associated thermodynamic potential is
a minimum.
• In the total differentials below, “=” holds for reversible processes and
“<” for irreversible processes.

Internal energy: (set X ≡ V, M and Y ≡ −p, H)

U (S, X, N ) = T S + Y X + µN, dU = T dS + Y dX + µdN .
Enthalpy:
E(S, Y, N ) = U − Y X = T S + µN, dE = T dS − XdY + µdN .
Helmholtz free energy:
A(T, X, N ) = U − T S = Y X + µN, dA = −SdT + Y dX + µdN .
Gibbs free energy:
G(T, Y, N ) = U − T S − Y X = µN, dG = −SdT − XdY + µdN .
Grand potential:
Ω(T, X, µ) = U − T S − µN = Y X, dΩ = −SdT + Y dX − N dµ.
Alternative set of thermodynamic potentials [tln9]

µ 1 p
Intensive variables: α = − , β = , γ =
T T T

The following four thermodynamic potentials are related to each other via
Legendre transform.

Entropy (microcanonical potential):

S(U, V, N ) = αN + βU + γV, dS = αdN + βdU + γdV

Massieu function (canonical potential):

Φ(β, V, N ) = S − βU = −A(T, V, N )/T, dΦ = αdN − U dβ + γdV

Kramers function (grandcanonical potential):

Ψ(β, V, α) = S − βU − αN = −Ω(T, V, µ)/T, dΨ = −N dα − U dβ + γdV

Planck function:
Π(β, γ, N ) = S − βU − γV = −G(T, p, N )/T, dΠ = αdN − U dβ − V dγ

Gibbs-Duhem relation: N dα + U dβ + V dγ = 0

Distinct properties:

• Any spontaneous i.e. irreversible process at constant values of the

natural independent variables is accompanied by an increase of the
associated thermodynamic potential.
• The equilibrium state for fixed values of a set of natural independent
variables is the state where the associated thermodynamic potential is
a maximum.
Entropy of mixing [tln25]

Consider two dilute gases in a rigid and insulat-

ing box separated by a mobile conducting wall: p T p T
n1 = x1 n (helium) and n2 = x2 n (neon) with
x1 + x2 = 1. Thermal equilibrium: p, T, V1 = n 1 V1 n 2 V2
x1 V, V2 = x2 V.
When the interior wall is removed, the gases mix spontaneously (irreversible
process). By how much does the entropy increase?
To calculate ∆S, consider a reversible mixing process involving an isothermal
expansion of semipermeable walls:
U = U1 + U2 with Ui (T ) = const ⇒ dUi = T dSi − pi dVi = 0.
ni RT
pi = are the partial pressures exerted on the semipermeable walls.
Vi
pi ni R dVi
⇒ dSi = dVi = dVi = nRxi .
T Vi Vi
X Z V dVi X
Entropy of mixing: ∆S = nR xi = −nR xi ln xi > 0.
i x i V Vi i

Gibbs paradox: No entropy increase should result if the two gases happen
to be of the same kind. This distinction is not reflected in the above deriva-
tion. The paradox is resolved by quantum mechanics, which requires that
distinguishable and indistinguishable particles are counted differently.
To calculate the change in Gibbs free energy during mixing, we use the
result for G(T, p, N ) of an ideal gas (see [tex15]) rewritten as G(T, p, n) =
−nRT [ln(T /T0 )α+1 − ln(p/p0 )].
"   #
α+1
X T p
Initially: p1 = p2 = p ⇒ Gini = − nxi RT ln − ln .
i
T 0 p 0
"   #
α+1
X T pxi
Finally: pi = pxi ⇒ Gf in = − nxi RT ln − ln .
i
T0 p0
X
Change in Gibbs free energy: ∆G = nRT xi ln xi .
i
 
∂G X
Use S = − to recover ∆S = −nR xi ln xi .
∂T p i
X
Change in chemical potential: use G = ni µi ⇒ ∆µi = RT ln xi .
i
Thermodynamic functions [tln5]

First partial derivatives of thermodynamic potentials with respect to natural

independent variables

Entropy:
     
∂A ∂G ∂Ω
S=− =− =−
∂T X,N ∂T Y,N ∂T X,µ

Temperature:    
∂U ∂E
T = =
∂S X,N ∂S Y,N

Volume/magnetization (X ≡ V, M ):
   
∂E ∂G
X=− =−
∂Y S,N ∂Y T,N

Pressure/magnetic field (Y ≡ −p, H):

     
∂U ∂A ∂Ω
Y = = =
∂X S,N ∂X T,N ∂X T,µ

Number of particles:  
∂Ω
N =−
∂µ T,X

Chemical potential:
       
∂U ∂E ∂A ∂G
µ= = = =
∂N S,X ∂N S,Y ∂N T,X ∂N T,Y
Osmotic pressure [tln26]

Consider a dilute solution. It consists of a sol-

vent (e.g. water) and a solute (e.g. sugar). Sys-
B
tem A (pure solvent) is separated from system h
water
B (solution) by a membrane that is permeable +
to the solvent only. At thermal equilibrium, this sugar
causes an excess pressure in system B, which is A
called osmotic pressure. water membrane

System A: pressure p0 , concentrations xS = 0, xW = 1.

System B: pressure p = p0 + π, concentrations xS = 1 − xW  1.
Observed osmotic pressure: π = ρB gh, where ρB is the mass density of B.

The osmotic pressure is a consequence of the requirement that the solvent

on either side of the membrane must be in chemical equilibrium:

µW (T, p, xW ) = µW (T, p0 , 1).

(1) Effect of solute concentration on chemical potential:

(1)
∆µW ≡ µW (T, p, xW ) − µW (T, p, 1) = RT ln xW = RT ln(1 − xS ) ' −RT xS .

(2) Effect of pressure on chemical potential: (use nW = nxW , nS = nxS )

   
(2) ∂µW ∂V
∆µW ≡ µW (T, p, 1) − µW (T, p0 , 1) = ∆p = ∆p
∂p T,nW ∂nW T,p
V Vπ
= ∆p ' .
nW n

(1) Vπ
(2)
At thermal equilibrium: ∆µW + ∆µW = 0 ⇒ − RT xS = 0.
n
RT nS
Van’t Hoff’s law for osmotic pressure: π = .
V
Maxwell’s relations [tln17]

inferred from second partial derivatives of thermodynamic potentials with

respect to two different natural independent variables

Fluid system:

   
∂T ∂p
dU = T dS − pdV ⇒ =−
∂V S ∂S V

   
∂T ∂V
dE = T dS + V dp ⇒ =
∂p S ∂S p

   
∂S ∂p
dA = −SdT − pdV ⇒ =
∂V T ∂T V

   
∂S ∂V
dG = −SdT + V dp ⇒ =−
∂p T ∂T p

Magnetic system:

   
∂T ∂H
dU = T dS + HdM ⇒ =
∂M S ∂S M

   
∂T ∂M
dE = T dS − M dH ⇒ =−
∂H S ∂S H

   
∂S ∂H
dA = −SdT + HdM ⇒ =−
∂M T ∂T M

   
∂S ∂M
dG = −SdT − M dH ⇒ =
∂H T ∂T H
Free energy stored and retrieved [tln18]

Consider a classical ideal gas confined to a cylinder with walls in thermal

equilibrium with a heat reservoir at temperature T .
Reversible cyclic process:

1. Push the piston in from position 1 to position 2 quasi-statically.

The work on the system is done reversibly.
Z 2 Z 2 Z 2
∆W12 = F dx = dA = (−SdT − pdV ) with dT = 0, dV < 0.
1 1 1

⇒ ∆W12 = A2 − A1 > 0.
The work done is equal to the excess Helmholtz potential (free energy).

2. Move the piston out from position 2 to position 1 quasi-statically.

Z 1 Z 1
∆W21 = dA = (−SdT − pdV ) with dT = 0, dV > 0.
2 2

⇒ ∆W21 = A1 − A2 = −∆W12 < 0.

All the energy stored (in the form free energy) is converted back into
work done by the system.

Irreversible cyclic process:

1. Push the piston in from position 1 to position 2 rapidly.

The initial and final equilibrium states are the same as previously, but
the process requires more work. The gas heats up, which produces a
larger pressure than in the quasi-static process.
⇒ ∆W12 > A2 − A1 > 0.
Only part of the work done on the system is stored as free energy.
2. (a) Move the piston out from position 2 to position 1 quasi-statically.
⇒ |∆W21 | = |A1 − A2 | < |∆W12 |.
All the free energy is converted back into work but that amount is
smaller than the work previously done on the system.
(b) Move the piston out from position 2 to position 1 rapidly.
⇒ |∆W21 | < |A1 − A2 | < |∆W12 |.
Only part of the available free energy is converted back into work, where
the full amount of free energy is only part of the work previously done
on the system.
Useful relations between partial derivatives [tln6]

Consider state variables x, y, z, w. Only two of the variables are independent.

  −1

∂x ∂y
#1 =
∂y
z ∂x z
     
∂x ∂y ∂z
#2 = −1
∂y z ∂z x ∂x y
     
∂x ∂x ∂y
#3 =
∂w z ∂y z ∂w z
       
∂x ∂x ∂x ∂w
#4 = +
∂y z ∂y w ∂w y ∂y z

Proof:
   
∂x ∂x
Start with total differential of x(y, z): dx = dy + dz.
∂y z ∂z y
   
∂y ∂y
Substitute total differential of y(x, z): dy = dx + dz.
∂x z ∂z x
     "      #
∂x ∂y ∂x ∂y ∂x
⇒ − 1 dx + + dz = 0.
∂y z ∂x z ∂y z ∂z x ∂z y

Expressions in brackets must vanish independently ⇒ #1 and #2.

Introduce parameter w: x(y, z) with y = y(w) and z = z(w).
       
∂x ∂x dx ∂x dy ∂x dz
dx = dy + dz ⇒ = + .
∂y z ∂z y dw ∂y z dw ∂z y dw
Specify path: z =const i.e. dz = 0: ⇒ #3.
Introduce parameter y: x(y, w) with w = w(y) and z = z(y).
       
∂x ∂x dx ∂x ∂x dw
dx = dy + dw ⇒ = + .
∂y w ∂w y dy ∂y w ∂w y dy
Specify path: z =const i.e. dz = 0: ⇒ #4.
Response functions [tln7]

Second partial derivatives of thermodynamic potentials with respect to nat-

ural independent variables. Response functions describe how one thermo-
dynamic function responds to a change of another thermodynamic function
under controlled conditions. Response functions are important because of
their experimental accessibility. Consider a system with N = const.

δQ
Thermal response functions (heat capacities): C ≡
δT
    
∂S ∂S
T dT + T dX for S(T, X)


∂T X ∂X T



δQ = T dS =    

 ∂S ∂S
 T dT + T dY for S(T, Y )


∂T Y ∂Y T
   2     2 
∂S ∂ A ∂S ∂ G
⇒ CX = T = −T , C Y = T = −T
∂T X ∂T 2 X ∂T Y ∂T 2 Y

where X ≡ V, M and Y ≡ −p, H.

Equivalent expressions of CX , CY are derived from δQ = dU − Y dX:

    
∂U ∂U
δQ = dT + − Y dX for U (T, X)
∂T X ∂X T
  
δQ  ∂U
⇒ CX ≡ =
δT X ∂T X
    
δQ  ∂U ∂X
⇒ CY ≡ = CX + −Y
δT Y ∂X T ∂T Y

 
∂E
Also, from δQ = dE + XdY we infer CY =
∂T Y

Note that U (T, X) and E(T, Y ) are not thermodynamic potentials.

1
Mechanical response functions

1 ∂2G
   
1 ∂V
Isothermal compressibility: κT ≡ − =−
V ∂p T V ∂p2 T
1 ∂2E
   
1 ∂V
Adiabatic compressibility: κS ≡ − =−
V ∂p S V ∂p2 S
 
1 ∂V
Thermal expansivity: αp ≡
V ∂T p
Relations with thermal response functions Cp , CV :

Cp κT T V αp2 T V αp2 κS
= , Cp = , CV =
CV κS κT − κS κT (κT − κS )

T V αp2
⇒ Cp − CV = >0
κT

Magnetic response functions

   2   2 −1
∂M ∂ G ∂ A
Isothermal susceptibility: χT ≡ =− 2
=
∂H T ∂H T ∂M 2 T
   2 
∂M ∂ E
Adiabatic susceptibility: χS ≡ =−
∂H S ∂H 2 S
 
∂M
“You name it”: αH ≡ −
∂T H
Relations with thermal response functions CH , CM :

2 2
CH χT T αH T αH χS
= , CH = , CM =
CM χS χT − χS χT (χT − χS )

2
T αH
⇒ CH − CM =
χT

2
Isothermal and adiabatic processes [tln8]

Fluid system:

Start from T dS = dU + pdV with U = U (T, V )

    
∂U ∂U (a) 1
T dS = dT + + p dV = CV dT + (Cp − CV )dV
∂T V ∂V T αp V
1 (b) κT
Isotherm: dT = 0 ⇒ T dS = (Cp − CV )dV = − (Cp − CV )dp
αp V αp
1 Cp − CV (c) κS Cp − CV
Adiabate: dS = 0 ⇒ dT = − dV = dp
αp V CV αp CV

     
∂U ∂V ∂V
(a) Use Cp − CV = +p , = V αp
∂V T ∂T p ∂T p
 
∂V
(b) Use dV = dp = −V κT dp
∂p T

(c) Use dV = −V κS dp

Magnetic system:

Start from T dS = dU − HdM with U = U (T, M ).

    
∂U ∂U 1
T dS = dT + − H dM = CM dT − (CH − CM )dM
∂T M ∂M T αH
1 χT
Isotherm: dT = 0 ⇒ T dS = − (CH − CM )dM = − (CH − CM )dH
αH αH
1 CH − CM χS CH − CM
Adiabate: dS = 0 ⇒ dT = dM = dH
αH CM αH CM
Conditions for thermal equilibrium [tln19]

Consider a fluid system in a rigid and isolated container that is divided into
two compartments A and B by a fictitious partition.
Conserved extensive quantities:
U = UA + UB = const, V = VA + VB = const, N = NA + NB = const.
Fluctuations in internal energy: ∆UA = −∆UB
Fluctuations in volume: ∆VA = −∆VB
Fluctuations in number of particles: ∆NA = −∆NB
1 p µ
Entropy: S(U, V, N ), dU + dV − dN
dS =
T T T
      
X ∂Sα ∂Sα ∂Sα
∆S = ∆Uα + ∆Vα + ∆Nα
α=A,B
∂U α Vα Nα ∂V α Uα Nα ∂N α Uα Vα
     
1 1 pA pB µA µB
= − ∆UA + − ∆VA + − ∆NA
TA TB TA TB TA TB

At thermal equilibrium, the entropy is a maximum. Hence the entropy change

due to fluctuations in U, V, N must vanish.
⇒ TA = TB , pA = p B , µA = µB

If the fictitious wall is replaced by a real, mobile, conducting wall, then

particle fluctuations in the two compartments are suppressed: ∆NA = 0. In
this case, µA 6= µB is possible at equilibrium.
If the mobile, conducting wall is replaced by a rigid, conducting wall, then
volume fluctuations in the two compartments are also suppressed: ∆VA = 0.
In this case, also pA 6= pB is possible at equilibrium.
If the conducting wall is replaced by an insulating wall, then energy fluctu-
ations in the two compartments are also suppressed: ∆UA = 0. In this case,
also TA 6= TB is possible at equilibrium.
Stability of thermal equilibrium [tln20]

Consider a fluid system with N = const in thermal equilibrium at tempera-

ture T0 and pressure p0 . Any deviation from that state must cause an increase
in Gibbs free energy:

G(T0 , p0 ) = U (S, V ) − T0 S + p0 V.

Effects of fluctuations in entropy and volume:

     
∂U ∂U
δG = − T0 δS + + p0 δV
∂S V ∂V S
 2   2   2  
1 ∂ U 2 ∂ U ∂ U 2
+ (δS) + 2 δSδV + (δV )
2 ∂S 2 ∂SδV ∂V 2

   
∂U ∂U
Equilibrium condition: − T0 = 0, + p0 = 0
∂S V ∂V S
 2   2   2 
∂ U 2 ∂ U ∂ U
Stability condition: 2
(δS) + 2 δSδV + (δV )2 > 0.
∂S ∂SδV ∂V 2

Condition for positive definite quadratic form:

2
∂2U ∂2U ∂2U ∂2U ∂2U

> 0, > 0, − > 0.
∂S 2 ∂V 2 ∂S 2 ∂V 2 ∂SδV

Implications:

∂2U
   
∂T T
= = > 0 ⇒ CV > 0.
∂S 2 V ∂S V CV
 2   
∂ U ∂p 1
2
=− = > 0 ⇒ κS > 0.
∂V S ∂V S V κS
 2   2   2 2  2
∂ U ∂ U ∂ U T ∂T
> ⇒ >
∂S 2 V ∂V 2 S ∂SδV V κS CV ∂V S
Jacobi transformations [tln21]

Change of independent variables in a state function. Unlike in a Legendre

transform, the state variable in question remains the same.
 
∂S
Task: Calculate Cp = T from entropy function S(U, V ).
∂T p
   
∂S 1 ∂S p
Use = , = and solve for T (V, U ), p(V, U ).
∂U V T ∂V U T
Solving T (V, U ), p(V, U ) for V (T, p), U (T, p) to obtain S(T, p) amounts to the
simultaneous solution of two nonlinear equations for two variables.
Solving
   
∂T ∂T
dT = dV + dU
∂V U ∂U V
   
∂p ∂p
dp = dV + dU
∂V U ∂U V
for
    
1 ∂p ∂T J1 (V, U )
dV = dT − dp ≡
J(V, U ) ∂U ∂U V J(V, U )
  V   
1 ∂p ∂T J2 (V, U )
dU = − dT + dp ≡
J(V, U ) ∂V U ∂V U J(V, U )
with Jacobian determinants
     
 ∂T ∂T 
 
 ∂V ∂U 
J(V, U ) =   U  V  ,
 ∂p ∂p 

 ∂V ∂U 
U V
       
 ∂T   ∂T 
 dT   dT 
 ∂U V  , J2 (V, U ) =   ∂V U
  
J1 (V, U ) =  ,
∂p   ∂p 
 dp   dp 
 ∂U V   ∂V
U


amounts to the solution of two linear equations for two variables.

1 p
⇒ dS = dU + dV
T  T          
1 ∂p ∂p 1 ∂T ∂T
= − +p dT + −p + dp
TJ ∂V U ∂U V TJ ∂U V ∂V U
       
∂S 1 ∂p ∂p
⇒ Cp = T = p −
∂T p J ∂U V ∂V U
Contents of this Document [ttc4]

4. Equilibrium Thermodynamics IV: Applications

• Entropy and internal energy of the classical ideal gas. [tex14]

• Thermodynamic potentials of the classical ideal gas. [tex15]
• Chemical potential of the classical ideal gas. [tex17]
• Ideal gas heat capacity by design. [tex35]
• Sound velocity in the classical ideal gas I. [tex18]
• Sound velocity in the classical ideal gas II. [tex99]
• Absolute temperature from measurements. [tex134]
• Polytropic process of classical ideal gas. [tex138]
• Heavy piston. [tex141]
• Isothermal atmosphere. [tex150]
• Adiabatic atmosphere. [tex151]
• Homogeneous atmosphere. [tex152]
• Van der Waals equation of state. [tln22]
• Cooling gases: Joule effect (free expansion) and Joule-Thomson effect
(throttling). [tln23]
• Joule-Thomson inversion curves. [tsl1]
• Heat capacities of the van der Waals gas. [tex27]
• Internal energy and entropy of the van der Waals gas. [tex38]
• Joule coefficient of the van der Waals gas. [tex31]
• Joule-Thomson coefficient of the van der Waals gas. [tex32]
• Assembling thermodynamic information. [tex29]
• How not to modify the ideal gas equation of state. [tex11]
• Reconstructing the equation of state of a fluid system. [tex42]
• Reconstructing the equation of state of a gas. [tex43]
• Effects of first virial correction on ideal gas properties. [tex33]
• Entropy due to electronic spins in iron ammonium alum. [tsl2]
• Adiabatic demagnetization. [tln24]
• Thermodynamics of an ideal paramagnet I. [tex19]
• Thermodynamics of an ideal paramagnet II. [tex20]
• Thermodynamics of an ideal paramagnet III. [tex21]
• Thermodynamics of a real paramagnet. [tex36]
• Thermodynamics of a classical ideal paramagnetic gas I. [tex22]
• Thermodynamics of a classical ideal paramagnetic gas II. [tex133]
• Hydrostatic pressure. [tex132]
• Rubber band heat engine. [tex39]
• Equation of state and adiabate of an elastic band. [tex40]
• Determining CV of condensed matter. [tex28]
• Thermodynamics of blackbody radiation. [tex23]
• Carnot cycle of thermal radiation. [tex24]
[tex14] Entropy and internal energy of the classical ideal gas
The classical ideal gas for a fixed number N of particles is specified by the equation of state
pV = N kB T and the constant heat capacity CV = αN kB [α = 32 (monatomic), α = 52 (diatomic),
α = 3 (polyatomic)].
(a) Use this information to calculate the internal energy U (T, V ) and show that the result is, in
fact, independent of V . Use the same information to calculate the entropy S(T, V ). Introduce
reference values T0 , V0 , U0 , S0 for the integrations.
(b) Determine the mechanical response functions αp (thermal expansivity), κT (isothermal com-
pressibility), and κS (adiabatic compressibility).

Solution:
[tex15] Thermodynamic potentials of the classical ideal gas
The classical ideal gas for a fixed number N of particles is specified by the equation of state
pV = N kB T and the constant heat capacity CV = αN kB [α = 32 (monatomic), α = 52 (diatomic),
α = 3 (polyatomic)]. From the functions U (T ) and S(T, V ) determined in [tex14] calculate the
thermodynamic potentials U (S, V ) (internal energy), E(S, p) (enthalpy), A(T, V ) (Helmholtz po-
tential) and G(T, p) (Gibbs potential). Use the reference values T0 , V0 , U0 , S0 from [tex14].

Solution:
[tex17] Chemical potential of the classical ideal gas
Calculate the chemical potential µ(T, p) of the classical ideal gas by integrating the Gibbs-Duhem
equation, SdT − V dp + N dµ = 0, with V (T, p) from the equation of state and S(T, p) from [tex14].
Compare the result with the Gibbs free energy per particle, G/N = µ, calculated in [tex15].

Solution:
[tex35] Ideal gas heat capacity by design
Consider 1mol of a classical ideal gas [pV = RT, CV = 32 R].
(a) If CX is the heat capacity of a reversible process in which the thermodynamic variable X is
kept constant, show that this process is described by a curve pV f = const in the (p, V )-plane with
f = (Cp − CX )/(CX − CV ).
(b) Discuss the result for three special cases: X = p, X = V, X = S.

Solution:
[tex18] Sound velocity in the classical ideal gas I
In a sound wave the classical ideal gas [pV = nRT , CV = αnR = const, molar mass M ] is
adiabatically compressed and expanded.
(a) Show that the sound velocity is related to the adiabatic compressibility κS and the mass density
ρ as follows:  
1 ∂p
c2 = = .
ρκS ∂ρ S
(b) Use thermodynamic properties of the classical ideal gas to infer from this expression the result
c2 = γRT /M , where γ = 1 + 1/α.

Solution:
[tex99] Sound velocity in the classical ideal gas II
Show that the internal energy per unit mass and the enthalpy per unit mass of a classical ideal
gas p
with heat capacity CV = const can be expressed as follows in terms of the sound velocity
c = (∂p/∂ρ)S and the ratio of heat capacities γ = Cp /CV :

c2 c2
Ū = Ū0 + , Ē = Ū0 + ,
γ(γ − 1) γ−1

where Ū0 is a constant.

Solution:
[tex134] Absolute temperature from measurements
Consider a compressible fluid in an insulating cylinder with a movable piston, a calibrated heat
source, a pressure gauge, and a thermometer with arbitrary temperature scale θ. The experiment
consists of measurements over a range of θ of the following quantities:
• rate A(θ) = (δV /δθ)p at which volume increases during isobaric heating up,
• rate B(θ) = (δQ/δp)θ at which heat is supplied during isothermal decompression.
Show that from the data of these two experiments we can infer the following differential relation
between the thermometer reading θ and the absolute temperature T :

d ln T A(θ)
=− .
dθ B(θ)

The (undetermined) integration constant of ln T (θ) can be used to fix the scale of the absolute
temperature (e.g. by using the triple point of H2 O).

Solution:
[tex138] Polytropic process of classical ideal gas
Consider an ideal monatomic gas [pV = N kB T, CV = 23 N kB ] confined to a cylinder by a movable
piston. The gas is compressed from volume V1 to volume V2 < V1 under circumstances such
that the relation pV x = a with a = const is satisfied. In this polytropic process, determine the
quantities ∆W (work done on the system), ∆U (change in internal energy), and ∆Q (heat added
to the system) as functions of V1 and V2 . Determine for which values of x (0 < x < 2) each of
these quantities is positive or negative.

Solution:
[tex141] Heavy piston
A cylinder of cross section A with insulating walls has two compartments separated by a disk of
mass m. The axis of the cylinder is vertical. A uniform gravitational field g is present. The disk is
initially held at a fixed position by an external agent. The upper compartment is evacuated and
the lower compartment contains 1 mol of a monatomic, classical, ideal gas [pV = RT , CV = 32 R]
at temperature T0 , volume V0 , and pressure p0 . When the disk is released, it moves without (wall)
friction and comes to rest at a lower position. Calculate the final values p1 , V1 , T1 of pressure,
volume, and temperature, respectively. The disk does not exchange heat. The only significant
action of the gravitational field is on the disk.

Hint: Use energy conservation and Newton’s third law. Assume thermal equilibrium for the inital
and final states.

p=0

p V0 g
0
T0

Solution:
[tex150] Isothermal atmosphere
Consider a column of air [molar mass M = 29g] treated as a classical ideal gas [pV = nRT ] in a
uniform gravitational field g = 9.81m/s2 . The column is assumed to be in thermal equilibrium.
(a) Calculate the dependence of pressure p on height z and (unifom) temperature T , assuming
that the pressure is p0 at z = 0.
(b) At what height z1 (in meters) has the pressure fallen to half of p0 and at what height z2 to one
percent of p0 if the temperature is 20◦ C everywhere?
Hint: Start from the relation, dp(z) = −ρ(z)dU(z), between pressure p, mass density ρ, and
gravitational potential U at height z. This relation expresses the increment of pressure caused by
the weight of a thin layer of air. The ideal-gas equation of state is assumed to hold locally at all
heights.

Solution:
[tex151] Adiabatic atmosphere
Consider a column of air [molar mass M = 29g] treated as a classical ideal gas [pV = nRT ,
Cp /CV = γ = 1.41] in a uniform gravitational field g = 9.81m/s2 . The column is assumed to be
in mechanical equilibrium but not (yet) in thermal equilibrium. The mechanical equilibrium is
established by gravitational pressure and governed by the adiabatic relation pV γ =const.
(a) Calculate the dependence on height z of the pressure p, the mass density ρ, and the temperature
T , assuming that p = p0 and T = T0 at z = 0.
(b) Find the height zm , expressed as a function of T0 , at which T , p, and ρ all reach zero. What
is that height (in meters) if T0 is room temperature?
Hints: (i) Infer from pV γ =const the differential relation dT /T = [(γ − 1)/γ]dp/p. (ii) Use
the relation dp(z) = −ρ(z)dU(z) from [tex150] linking pressure, mass density, and gravitational
potential to infer differential equations for T (z) and p(z).

Solution:
[tex152] Homogeneous atmosphere
Meteorological modeling uses the concept of homogeneous atmosphere with constant density over
some vertical distance. Consider a column of air [molar mass M = 29g] treated as a classical ideal
gas [pV = nRT ] in a uniform gravitational field g = 9.81m/s2 . The column is assumed to have
constant mass density ρ(z) = ρ0 = const.
(a) Calculate the dependence on height z of the pressure p and the temperature T , assuming that
p = p0 and T = T0 at z = 0.
(b) Find the height zh , expressed as a function of T0 , at which T and p both reach zero.
(c) What must be the temperature (in ◦ C) at sea level so that the homogeneous atmosphere just
reaches the tip of Mount Everest?
Hints can be gleaned from [tex150] and [tex151].

Solution:
Van der Waals equation of state [tln22]

Atoms of gases interact via short-range force. The ideal gas equation of state,
pV = nRT , neglects the interaction completely. The van-der Waals equation
of state takes it into account summarily:

an2
 
p+ 2 (V − nb) = nRT,
V
where a, b are empirical parameters.

• nb: excluded volume due to the repulsive core of the interaction at

short distances,
• −an2 /V 2 : pressure correction due to the attractive tail of the interac-
tion at long distances.

φ (r) φ (r)

repulsive core excludedvolume

Van der Waals

approximation
infinite−range
attractive tail attraction

r r
Cooling of gases [tln23]

During expansion, a gas does work against attractive intermolecular forces.

In the process, the average potential energy increases and, by virtue of en-
ergy conservation, the average kinetic energy decreases. The result is a drop
in temperature. We discuss two processes to illustrate this effect.

Joule effect: free expansion

Free expansion involves no heat transfer and no work performance:
∆Q = 0, ∆U = 0,.
Initial state: Vi , pi , Ti ; final state: Vf , pf , Tf with pf < pi .
The temperature change in the expanding gas is calculated for a quasi-static
process between the same equilibrium states.
       
∂U ∂U ∂S ∂p
Use = CV , =T −p=T −p
∂T V ∂V T ∂V T ∂T V
     
∂T (∂U/∂V )T 1 ∂p
Joule coefficient: =− = p−T .
∂V U (∂U/∂T )V CV ∂T V
 
∂p
Ideal gas: T = p ⇒ no effect.
∂T V

Joule-Thomson effect: throttling

The gas is forced through a porous wall between two chambers. During
the process the pressure is constant in both chambers. In the following we
consider quasi-static throttling.
Initial state: Vi , pi , Ti ; final state: Vf , pf , Tf with pf < pi .
Z Vf Z 0
∆Q = 0, ∆U = ∆W = − pf dV − pi dV = −pf Vf + pi Vi .
0 Vi

⇒ Ui + pi Vi = Uf + pf Vf = const. ⇒ E = const. ⇒ dE = T dS + V dp = 0.
       
∂E ∂E ∂S ∂V
Use = Cp , =T + V = −T +V.
∂T p ∂p T ∂p T ∂T p
  "   #
∂T (∂E/∂p)T 1 ∂V
Joule-Thomson coefficient: =− = T −V .
∂p E (∂E/∂T )p Cp ∂T p
 
∂V
Ideal gas: T = V ⇒ no effect.
∂T V
Joule−Thomson inversion curves [tsl1]

• Hydrogen gas: experimental data.

an2
 
• Van der Waals gas: p + 2 (V − nb) = nRT.
V
nRT  an 
• Dieterici gas: p = exp − .
V − nb RT V

[from Kubo: Thermodynamics]

[tex27] Heat capacities of the van der Waals gas
Consider the van der Waals equation of state for n = 1mol of a gas:
 a 
p + 2 (V − b) = RT.
V
(a) Show that the heat capacity CV may depend on T but is independent of V .
(b) Calculate the quantity Cp − CV as a function of p and V . Verify that Cp − CV → R in the
ideal gas limit.

Solution:
[tex38] Internal energy and entropy of van der Waals gas
Calculate the internal energy U (T, V ) and the entropy S(T, V ) of the van der Waals gas, specified
by the equation of state
aN 2
 
p + 2 (V − N b) = N kB T
V
and the specific heat CV = αN kB .

Solution:
[tex31] Joule coefficient of van der Waals gas
The cooling of a gas via free expansion is described by the Joule coefficient
     
∂T 1 ∂p
= p−T .
∂V U CV ∂T V

(i) Determine the Joule coefficient for 1 mol of the ideal gas [pV = RT, CV = αR] and for 1mol of
the van der Waals gas [(p + a/V 2 )(V − b) = RT, CV = αR].
(ii) Calculate the temperature change Tf − Ti when the van der Waals gas is freely expanded from
Vi to Vf > Vi .

Solution:
[tex32] Joule−Thomson coefficient of van der Waals gas
The cooling of a gas via throttling is described by the Joule-Thomson coefficient
  "   #
∂T 1 ∂V
= T −V .
∂p E Cp ∂T p

(i) Determine the Joule-Thomson coefficient for 1 mol of the van der Waals gas [(p+a/V 2 )(V −b) =
RT, CV = αR].
(ii) Throttling results in cooling only if (∂T /∂p)E > 0. In the (T, p)-plane, this region is bounded
by the inversion curve, which is determined by the condition (∂T /∂p)E = 0. Calculate the inversion
condition p∗ (T ) for the van der Waals gas.
(iii) Throttling is most efficient for cooling if it starts at the highest possible pressure. Find the
temperature at which the inversion curve has a maximum.

Solution:
[tex29] Assembling thermodynamic information
The following thermodynamic information is known about n = 1mol of a system:

• At constant temperature T0 , the work done on the system when it is compressed from V0 to
V is ∆W0 = −RT0 ln(V /V0 ).
• The entropy is S(T, V ) = R(V0 /V )(T /T0 )a , where V0 , T0 , a are constants.

Use this information to determine (i) the Helmholtz free energy A(T, V ), (ii) the equation of state
f (p, V, T ) = 0, and (iii) the work of compression ∆W done at an arbitrary temperature T .

Solution:
[tex11] How not to modify the ideal gas equation of state
Suppose we know empirically that for a real system with fixed n, the product of the pressure p and
the volume V is a function of the temperature alone, pV = f (T ), and that the internal energy is
also a function of the temperature alone, U = U (T ). These properties are realized in the classical
ideal gas, where f (T ) = nRT and U (T ) = CV T with CV = αnR = const. Show that the only
function f (T ) in the equation of state of the above form which is compatible with U = U (T ), i.e.
with (∂U/∂V )T = 0 is linear in T with zero intercept: f (T ) = rT with r = const.

Solution:
[tex42] Reconstructing the equation of state of a fluid system
A fluid system is found to have a thermal expansivity αp = (nR/pV ) + (na/RT 2 V ) and an
isothermal compressibility κT = (n/V )[T f (p) + b/p], where a, b are constants and f (p) is an
unknown function.
(a) Find the function f (p) which makes the two response functions thermodynamically consistent.
(b) Reconstruct the equation of state V = V (T, p) from the two response functions.

Solution:
[tex43] Reconstructing the equation of state of a gas
It is found for a gas that αp = RV /np + aV /nT 2 and κT = T (V /n)f (p), where a is a constant and
f (p) is an unknown function.
(a) Find the function f (p) which makes the two response functions thermodynamically consistent.
(b) Reconstruct the equation of state g(V, T, p) = 0 from the two response functions.

Solution:
[tex33] Effects of first virial correction on ideal gas properties
The ideal gas equation of state supplemented by the first virial correction reads p = (nRT /V )[1 +
(n/V )B(T )]. The associated heat capacity at constant volume can be written in the form CV =
3 2
2 nR−(n R/V )F (T ), where F (T ) and B(T ) are related by thermodynamic consistency conditions.
(a) Express F (T ) as a function of B(T ).
(b) Find the function U (T, V ) (internal energy) and S(T, V ) (entropy) for this system.
(c) Find the first virial correction B(T ) for the van der Waals gas, [p + a(n/V )2 ](V − nb) = nRT ,
and calculate the associated virial correction F (T ) to the heat capacity CV .

Solution:
Entropy due to electronic spins
in iron ammonium alum [tsl2]

Dependence of entropy on temperature and magnetic field for paramagnet.

(a) Carnot cycle operating on this system.

(b) Cooling of this system via adiabatic demagnetization.

Adiabatic demagnetization [tln24]

S
Equation of state for paramagnetic salt in a
weak magnetic field: M(T, H) = χT (T )H. H=0
1
Helmholtz free energy: H = H1
2
 
∂A M
dA = −SdT + HdM, =H=
∂M T χT

M2
⇒ A(T, M) = A(T, 0) + .
2χT T
T2 T1
   
∂A 1 2 d −1
Entropy: S(T, M) = − = S(T, 0) − M χ
∂T M 2 dT T

1 dχT dχT
⇒ S(T, H) = S(T, 0) + H 2 ; χT > 0, < 0 for paramagnet.
2 dT dT
dχT
Third law: lim S(T, H) = 0 independent of H ⇒ lim = 0.
T →0 T →0 dT

1 dχT 2
1. Isothermal magnetization: ∆S = H < 0.
2 dT 1
Heat expelled from system: ∆Q = T1 ∆S.

2. Adiabatic demagnetization: ∆S = 0 ⇒ S(T1 , H1 ) = S(T2 , 0).


1 2 dχT 
⇒ S(T2 , 0) = S(T1 , 0) + H1 ⇒ T2 < T1 .
2 dT T1

Consider the entropy function S(T, H) for iron ammonium alum.

• The sequence of steps 1 and 2 approaches absolute zero. As T →

0, adiabates and isotherms become increasingly parallel, implying a
diminishing efficiency of the cooling process.
• The sequence of steps requires heat reservoirs at various temperatures.
They can be established by employing a Carnot engine consisting of
steps 1 and 2 and their inverses.
• Magnetic refrigerators using paramagnetic salts attain ∼ 0.2K. Adia-
batic demagnetization attains mK temperatures.
[tex19] Thermodynamics of an ideal paramagnet I
For an ideal paramagnet specified by the equation of state M = H/T (Curie law) and internal
energy U = 0, find (a) the entropy S(T, H), (b) the thermodynamic potentials E(S, H), A(T, M ),
G(T, H), and (c) the response functions CM , CH , χT , χS , αH .

Solution:
[tex20] Thermodynamics of an ideal paramagnet II
For an ideal paramagnet specified by the equation of state M = H/T (Curie law) and heat capacity
CM = const, find (a) the internal energy U (T, H), the entropy S(T, H), and the enthalpy E(T, H);
(b) the thermodynamic potentials A(T, M ), G(T, H); (c) the response functions χT , χS , αH , CH .

Solution:
[tex21] Thermodynamics of an ideal paramagnet III
For an ideal Langevin paramagnet, which is specified by the equation of state M = tanh(H/T )
(Langevin function) and the internal energy U ≡ 0, find (a) the entropy S(T, H) and the enthalpy
E(T, H); (b) the thermodynamic potentials A(T, M ), G(T, H); (c) the response functions χT , χS ,
αH , CH . Determine the integration constant S0 in S(T, H) such that S → 0 for T → 0 and H 6= 0
in accordance with the third law of thermodynamics.

Solution:
[tex36] Thermodynamics of a real paramagnet
The magnetization M of a paramagnetic system was measured over a certain temperature range,
and it was found to depend only on the ratio H/T : M = f (H/T ).
(a) Show that the internal energy is then independent of H: U = U (T ).
(b) Show that the entropy then has the following functional form:

H/T T
U 0 (T 0 )
  Z Z
H H
S(T, H) = S1 (T ) − f + dx f (x), where S1 (T ) = S0 + dT 0 .
T T 0 T0 T0

Solution:
[tex22] Thermodynamics of a classical ideal paramagnetic gas I
Consider 1mol of a paramagnetic gas, specified by the equations of state pV = RT (classical ideal
gas) and M = H/T (Curie paramagnet), and by a constant heat capacity CV M = 32 R.
(a) Calculate the internal energy U (T, V, M ) by integration of the differential dU = T dS − pdV +
HdM . Show that U only depends on T . (b) Calculate the entropy S(T, V, M ) by integration of
the differential dS = (1/T )dU + (p/T )dV − (H/T )dM . (c) Calculate the Helmholtz potentials
. .
AM (T, V, M ) = U − T S and AH (T, V, H) = U − T S − M H. (d) Calculate the Gibbs potentials
. .
GM (T, p, M ) = U − T S + pV and GH (T, p, H) = U − T S + pV − M H.

Solution:
[tex133] Thermodynamics of a classical ideal paramagnetic gas II
Consider the fixed amount n = 1mol of a paramagnetic gas, specified by the equations of state
pV = RT , M = H/T , and by the heat capacity CV M = 3R/2.
(a) Characterize a general adiabatic process by a functional relation of the form f (T, p, H) = const.
Start from the function S(T, V, M ) derived in [tex22].
(b) Show that the heat transfer is path-independent for isothermal processes. Derive an expression
for the heat transfer ∆QT of a general isothermal process between (p0 , H0 ) and p1 , H1 .
(c) Show that the heat transfer for isobaric processes does depend on the path taken. Derive an
expression for the heat transfer ∆Qp of the isobaric process (T0 , H0 ) → (T0 , H1 ) → (T1 , H1 ) →
(T1 , H0 ) → (T0 , H0 ) along straight segments in the (T, H)-plane.
(d) Consider a process (T, P0 , H0 ) → (T, P1 , P1 ) with p0 > p1 that is both isothermal (no change
in energy) and adiabatic (no heat transfer). In this process mechanical work ∆Wp (T, p0 , p1 ) done
by the system is matched by the same amount of magnetic work ∆WH (T, H0 , H1 ) done on the
system. Find ∆Wp (T, p0 , p1 ) and ∆Wp (T, p0 , p1 ).

Solution:
[tex132] Hydrostatic pressure
The chemical potential in a homogeneous fluid is a function µ = µ0 (p, T ) such as calculated in
[tex17] for the classical ideal gas. What remains uniform throughout the fluid in the presence of a
uniform gravitational field is the potential

µ = µ0 (p, T ) + mgz = const.,

where m is the mass of the fluid particle, g is the acceleration due to gravity, and the z-direction
is against the field. For a fluid of negligible compressibility derive from the condition dµ/dz = 0
the familiar result for the hydrostatic pressure,

p(z) = p0 − ρgz,

where ρ = N m/V is the (average) mass density of the fluid.

Solution:
[tex39] Rubber band heat engine
Consider a heat engine that uses a rubber band in the three-step cycle shown. The equation of
state J = αLT with α = const relates the tension J in the band to the length L of the band and
to the absolute temperature T . The heat capacity of the band at constant length is CL = const.
Calculate the heat transfer ∆Q and the work performance ∆W in each of the three steps in the
cyclic process shown:
1 → 2 relaxation at T = const,
2 → 3 expansion at J = const,
3 → 1 heating up at L = const.
From these results calculate the efficiency η of the rubber band heat engine. Compare η with the
efficiency of a Carnot engine operating between the same temperatures.

J
2J0 1

2
J0 3

L
L0 2L 0

Solution:

1
[tex40] Equation of state and adiabate of an elastic band
Experimentally one finds the following response functions of a band of elastic material
  "  3 #   "  3 #
∂J aT L0 ∂J aL L0
= 1+2 , = 1− ,
∂L T L0 L ∂T L L0 L

where J is the tension, L is the length, T is the temperature, and a, L0 are constants. The heat
capacity at constant length is CL = const.
(a) Use this empirical information to reconstruct the equation of state J(T, L).
(b) Calculate the response function (∂L/∂T )J and discuss its physical meaning.
(c) If the band is stretched adiabatically from length L0 at temperature T0 to length 2L0 , what is
the final temperature?

Solution:
[tex28] Determining CV of condensed matter
The direct measurement of the heat capacity at constant volume CV for condensed matter is very
difficult. Therefore, express CV in terms of the following quantities, all of which are experimentally
more accessible: the heat capacity at constantp pressure Cp , the isothermal compressibility κT , the
mass density ρ, and the sound velocity c = (∂p/∂ρ)S .

Solution:
[tex23] Thermodynamics of black-body radiation
Electromagnetic radiation inside a cavity is in thermal equilibrium with the walls at temperature
T . The radiation has an energy density that depends only on the temperature, i.e. its internal
energy has the form U (T, V ) = V e(T ). The radiation pressure is determined by the energy density
alone: p = 13 e(T ). (a) Use the consistency equations for the total differential dS to show that
the energy density has the form e(T ) = σT 4 , where σ is a constant, now known as the Stefan-
Boltzmann constant. In this argument, the additional assumption enters that e(T ) → 0 for T → 0.
(b) Determine the entropy S(T, V ) and the thermodynamic potentials U (S, V ), E(S, p), A(T, V ),
G(T, p). (c) Determine the isotherms and adiabates in the (V, p)-plane. (d) Determine the response
functions CV , Cp , κT , κS , αp .

Solution:
[tex24] Carnot cycle of thermal radiation
Describe the four steps of a Carnot engine, where the operating material is black-body radiation.
The internal energy is given by Stefan’s law, U (T, V ) = σT 4 V . The equation of state is p = 13 σT 4 .
Determine the work performance ∆W and the heat transfer ∆Q during each of the four steps and
derive the Carnot efficiency from these results. Sketch the cycle in the (V, p)-plane.

Solution:
Contents of this Document [ttc5]

5. Thermodynamics of Phase Transitions I

• Phase diagram of a “normal” substance. [tsl3]

• Phase diagram of H2 O. [tsl4]
• Ferrimagnetic phases. [tsl49]
• Liquid crystal phases. [tsl51]
• Ordering of surfactant molecules. [tsl50]
• Phase coexistence: Gibbs phase rule. [tln27]
• Classification of phase transitions. [tln28]
• Gibbs free energy and derivatives at discontinuous transition. [tsl7]
• Gibbs free energy and derivatives at continuous transition. [tsl8]
• Clausius-Clapeyron equation. [tln29]
• Entropy of a supercooled liquid. [tex30]
• Coexistence line of continuous phase transition. [tex37]
• Latent Heat and response functions. [tex124]
• Heat capacity of vapor in equilibrium with liquid phase. [tex41]
• Discontinuous transition: change in internal energy. [tex123]
• Dry ice. [tex125]
• Abnormal phase behavior. [tex54]
• Melting or freezing. [tex51]
• Triple-point phase changes. [tex52]
• Cooling down? Heating up? [tex153]
• Phase coexistence of ammonia. [tex55]
Phase diagram of a “normal” substance [tsl3]

(a) pressure versus volume versus temperature

• 1-phase regions: solid, liquid, gas (vapor)

• 2-phase coexistence regions
• 3-phase coexistence line

(b) projection onto (T, p) plane (intensive variables)

• 2-phase coexistence lines

– vapor-pressure curve
– fusion curve
– sublimation curve
• 3-phase coexistence point
– triple point
Phase diagram of H2O [tsl4]

(a) solid – liquid – vapor

(b) solid phases

Ferrimagnetic phases of cerium antimonide [tsl49]

Magnetic ordering in phases 1-5, 7-14 is long-ranged and has

– uniform magnetization (↑, ↓),
– no magnetization (◦),
in successive crystalline planes.
Magnetic ordering is short-ranged (paramagnetic) in phase 15.
Magnetic ordering in phase 6 has been unknown at the time.

[from Yeomans 1992]

Liquid crystal phases [tsl51]

Most common liquid crystal phases.

• Nematic: Molecular orientational ordering. Continuous rotational

symmetry about director.
• Smectic A: Nematic ordering plus density wave along symmetry axis
(director).
• Smectic C: Smectic A ordering with broken rotational symmetry.
Density wave not perpendicular to director.
• Smectic B: Smectic A ordering plus density wave perpendicular to
director.

[Physics Today, May 1982]

Ordering of surfactant molecules [tsl50]

Surfactant molecules have highly soluble polar heads and barely soluble hy-
drocarbon tails.
Increasing the surfactant concentration produces a succession of structures.
Some structures exhibit short-range order, others exhibit long-range order.

[from Yeomans 1992]

Phase coexistence: Gibbs phase rule [tln27]

Consider a K-component fluid system in a state that allows P coexisting

phases (a pure pV T system has K = 1).
K
X
(i) (i) (i) (i) (i) (i) (i)
Internal energy: dU = T dS − p dV + µ` dN` , i = 1, . . . , P.
`=1

Each U (i) depends on K + 2 extensive variables for a total of P (K + 2).

Conditions of phase coexistence: [(P − 1)(K + 2) relations]

• thermal equilibrium: T (1) = T (2) = · · · = T (P ) ,

• mechanical equilibrium: p(1) = p(2) = · · · = p(P ) ,
(1) (2) (P )
• chemical equilibrium: µ` = µ` = · · · = µ` , ` = 1, . . . , K.

The number of independent extensive variables is independent of P :

P (K + 2) − (P − 1)(K + 2) = K + 2.
Of this set, P extensive variables, e.g. V1 , . . . , VP , are necessary to specify
the size of all coexisting phases. The remaining independent state variables
can be chosen to be intensive.
The number of independent intensive state variables required to describe the
thermodynamic state of P coexisting phases in a K-component system is
F =K +2−P (Gibbs phase rule).
The maximum number of coexisting phases is K + 2, which is realized for
F = 0. In this case, no intensive variables need to be specified. All are
determined by the conditions of phase coexistence.
Example: 1-component fluid (K = 1)

• P = 1 (one phase): F = 2 means that both p and T must be specified

to locate the thermodynamic state in the (p, T )-plane.
• P = 2 (two phases): F = 1 means that only p or T must be specified
to locate the thermodynamic state in the (p, T )-plane. The state lies
on a coexistence line.
• P = 3 (three phases): F = 0 means that the coexistence condition
determines the values of p and T . The state is located at the triple
point in the (p, T )-plane.
Classification of phase transitions [tln28]

Consider a 1-component fluid system with phases I and II.

The Gibbs free energy has a different functional dependence on its natural
variables in the two phases:

GI (T, p, n) = µI (T, p)n, GII (T, p, n) = µII (T, p)n.

For given values of T and p, the equilibrium state is the one with the lowest
Gibbs free energy, i.e the state with the lower chemical potential.
At the transition: GI = GII , i.e. G∗ = µ∗ (nI + nII ) with nI + nII = n.

Discontinuous transition:
The volume and the entropy change discontinuously:
 I  II
∂G ∂G
1. = ⇒ µI = µII = µ∗ .
∂n T,p ∂n T,p
 I  II
∂G ∂G
2. 6= ⇒ V I 6= V II .
∂p
T,n ∂p T,n
 I  II
∂G ∂G
3. 6= ⇒ S I 6= S II .
∂T n,p ∂T n,p

Latent heat (change in enthalpy): ∆E = ∆(G + T S) = T ∆S

Continuous transition:
The volume and the entropy change continuously. Discontinuities or diver-
gences occur in higher-order derivatives.
 I  II
∂G ∂G
1. = ⇒ µI = µII = µ∗ .
∂n T,p ∂n T,p
 I  II
∂G ∂G
2. = ⇒ V I = V II .
∂p
T,n ∂p T,n
 I  II
∂G ∂G
3. = ⇒ S I = S II .
∂T n,p ∂T n,p
Gibbs free energy and derivatives
at discontinuous transition[tsl7]
Gibbs free energy and derivatives
at continuous transition[tsl8]
Clausius-Clapeyron equation [tln29]

p
Consider the variation of the free energy in the
two phases I and II along their coexistence
line: dGI = dGII . II
⇒ V I dp − S I dT = V II dp − S II dT.
I
Slope of the coexistence line:
T
S I − S II
 
dp ∆S ∆E
= I II
= = (Clausius-Clapeyron equation).
dT coex V −V ∆V T ∆V

Latent heat (change in enthalpy): ∆E = T ∆S

Fact: ∆S > 0 if dT > 0.

Consequence: (dp/dT )coex > 0 if ∆V > 0.
The case of H2 O: The substance contracts upon melting. Therefore, the
fusion curve has negative slope. The highest melting temperature occurs at
the triple point. The density maximum at pressures 0.008atm . p . 270atm
occurs inside the liquid phase. Ice under pressure tends to melt, which makes
skating possible.

(i) Vaporization curve (liquid-gas coexistence line):

The equilibrium pressure is then the saturated vapor pressure p(T )coex . It
can be derived by integrating the Clausius-Clapeyron equation:
   
dp ∆E ∆E p p ∆E 1 1
= ' ' ∆E ⇒ ln =− −
dT coex T ∆V T Vvap nRT 2 p0 nR T T0
  
∆E 1 1
⇒ p(T )coex = p(T0 )coex exp − − .
nR T T0

(ii) Fusion curve (solid-liquid coexistence line):

The fusion curve does not terminate in a critical point. The solid and liquid
phases have different symmetries.

(iii) Sublimation curve (solid-gas coexistence line):

The sublimation curve ends in the triple point.
[tex30] Entropy of supercooled liquid
The heat capacity at constant pressure of a substance is Csol in its solid state and Cliq in its liquid
state. Both quantities can be treated as constants. When the substance melts (at T = TM ) it
absorbs the latent heat L. Find the entropy difference ∆S = Sliq − Ssol between the supercooled
liquid state and the solid state at some temperature T < TM .

liquid

∆S
solid

T
T TM

Solution:
[tex37] Coexistence line of continuous phase transition
Consider 1 mol of a fluid with two phases 1 and 2 in coexistence. The coexistence line is p(T )coex .
Suppose that the volume V and the entropy S vary continuously at the transition (∆S = 0
and ∆V = 0), but the response functions Cp (heat capacity at constant pressure), αp (thermal
expansivity), and κT (isothermal compressibility) are discontinuous. Now consider the differentials
.
dS and dV for each phase and for paths in the (T, p)-plane. Then calculate ∆S = dS (2) − dS (1)
. (2) (1)
and ∆V = dV − dV between points an infinitesimal distance across the coexistence line
(a) at constant p,
(b) at constant T .
In the limit where the distance between the two points shrinks to zero, the ratio ∆S/∆V stays
finite and expresses (via Clausius-Clapeyron) the slope (dp/dT )coex of the coexistence line in terms
of the discontinuities, ∆Cp , ∆αp , ∆κT , in the response functions.
(c) Derive a relation between ∆Cp , ∆αp , ∆κT from the consistency condition of the results obtained
in parts (a) and (b).

Solution:
[tex124] Latent heat and response functions
Consider a discontinuous transition between phases 1 and 2 of a simple fluid. The latent heat is
L(T ) and the coexistence curve is p = p(T )coex .
(a) Show that the rate at which the latent heat changes along the transition curve depends on the
heat capacity Cp and the thermal expansivity αp of the coexisting two phases as follows:
" #
  (2) (1)
dL (2) (1) 1 V2 αp − V1 αp
= Cp − Cp + L(T ) − .
dT coex T V2 − V1

(b) Simplify this expression when phase 2 is a classical ideal gas, in which case we have V2  V1 ,
(2) (1) (2)
αp  αp , and an explicit result for the T -dependence of αp .

Solution:
[tex41] Heat capacity of vapor in equilibrium with liquid phase
Consider 1 mol of a simple fluid. Use Vgas  Vliq , pVgas = RT . The heat capacities in the gas and
(gas) (liq)
liquid phases are Cp and Cp , respectively. The latent heat of vaporization is L(T ).
.
(a) Show that the heat capacity Ccoex = T (dS/dT )coex of vapor for a process along which the
vapor is maintained in equilibrium with the liquid phase can be expressed in the following two
alternative ways:  
L d L
Ccoex = Cp(gas) − = Cp(liq) + T .
T dT T
Use the result of [tex124] to derive the second expression.

Solution:
[tex123] Discontinuous transition: change in internal energy
Consider a discontinuous transition in a simple fluid system. The latent heat is L(T ) and the
coexistence curve is p = p(T )coex . Show that the change in internal energy during the phase
transition is "  −1 #
d ln p(T )coex
∆U = L(T ) 1 − .
d ln T

Solution:
[tex125] Dry ice
The T -dependence of the vapor pressure of CO2 below the triple point (Tt = −56.2◦ C) is well
represented by the empirical relation
 
p(T )coex 3116K
ln = 16 − .
1atm T

The molar heat of melting is Lsl = 8330J with negligible T -dependence.

(a) Find the pressure pt at the triple point.
(b) Find the latent heat of sublimation, Lsg , and the latent heat of vaporization, Llg .
(c) Find the vapor pressure p(T )coex at 20◦ C.

Solution:
[tex54] Abnormal phase behavior
The figure shows the two-phase coexistence lines near the triple point of a hypothetical substance.
Under the reasonable assumption that the densities of the liquid and solid phases are similar and
much larger than the density of the gas phase, this phase diagram implies a material property that
is abnormal under the circumstance described.
Show that in this substance the solid phase near the triple point has a higher entropy than the
liquid phase. Melting the substance along a specific path thus releases heat. A somewhat related
exotic property is actually observed in 3 He (see [tsl14]).

liquid

solid
gas
T
Solution:
[tex51] Melting or freezing?
A vessel with insulating walls of negligible heat capacity contains 7kg of ice at −20◦ C. Now we
pour 2kg of water at +30◦ C into the vessel and seal it.
(a) How much ice (in kg) will remain in the vessel when the system is in thermal equilibrium again?
(b) Find the entropy change that takes place inside the insulated vessel.
Specific heat of ice: cs = 2090J/kgK.
Specific heat of water: cl = 4180J/kgK.
Latent heat of liquid-solid transition: L = 3.34 × 105 J/kg.

Solution:
[tex52] Triple point phase changes
A vessel with insulating and rigid walls contains 1g of water, 1g of ice, and 1g of H2 O vapor in
thermal equilibrium. Now we add the amount ∆Q = 251J of heat to the vessel. Find the amount
(in grams) of ice, water, and vapor in the new equilibrium state. The densities of ice and water
are taken to be equal and much larger than the density of vapor.
Triple point temperature: T = 273K.
Triple point pressure: p = 611N/m2 .
Latent heat of fusion: Lsl = 335J/g.
Latent heat of vaporization: Llg = 2495J/g.
Latent heat of sublimation: Lsg = Lsl + Llg = 2830J/g.

Solution:
[tex156] Triple point phase changes II
A cylinder with vertical axis has insulating and rigid walls and is capped by an insulating piston.
(0) (0) (0)
The cylinder contains ms = 1g of ice, ml = 2g of water, and mg = 5g of H2 O vapor in thermal
equilibrium. All four processes described below start from this state. Assume that the difference
in densities between the liquid and solid phases is negligibly small.

(i) The piston stays at rest. How much heat ∆Q1 can be added before the temperature begins to
(1) (1) (1)
increase? What are the masses ms , ml , mg , of ice, water, and vapor, respectively, at that point?

(ii) The piston stays at rest. How much heat ∆Q2 can be extracted before the temperature begins
(2) (2) (2)
to decrease? What are the masses ms , ml , mg , of ice, water, and vapor, respectively, at that
point?

(iii) There is no heat transfer. How much (positive) work ∆W3 can the piston do onto the gas (by
(3) (3) (3)
moving in) before the temperature begins to increase? What are the masses ms , ml , mg , of
ice, water, and vapor, respectively, at that point?

(iv) There is no heat transfer. How much (negative) work ∆W4 can the piston do onto the gas (by
(4) (4) (4)
moving out) before the temperature begins to decrease? What are the masses ms , ml , mg , of
ice, water, and vapor, respectively, at that point?

Triple point temperature: T = 273K.

Triple point pressure: p = 611N/m2 .
Latent heat of melting: Lsl = 335J/g.
Latent heat of vaporization: Llg = 2495J/g.

Solution:
[tex153] Cooling down? Heating up?
A vessel with insulating walls of negligible heat capacity has two compartments. The first compart-
ment contains 1kg of ice and 1kg of water coexisting at 0◦ C. The second compartment contains
5kg of ice at −20◦ C. Now we open the wall between the compartments and wait until thermal
equilibrium has been reached. What are the final temperature, the final masses of ice and water,
and the change in entropy?
In a modified experiment the first compartment again contains 1kg of ice and 1kg of water coex-
isting at 0◦ C but the second compartment contains 5kg of water at +20◦ C. What are the final
temperature, the final masses of ice and water, and the change in entropy?
Specific heats of water and ice: cw = 4180J/kgK, ci = 2090J/kgK.
Latent heat of melting: Lm = 3.34 × 105 J/kg.

Solution:
[tex55] Phase coexistence of ammonia
The vapor pressure curve of solid NH3 (sublimation curve) is found to satisfy the relation ln p =
23.03 − 3754/T and the vapor pressure curve of liquid NH3 the relation ln p = 19.49 − 3063/T ,
where p is measured in units of mm Hg and T in Kelvin.
Consider n = 1mol of this substance. Assume that the densities of the solid and liquid phases are
much larger than the density of the gas phase. Treat the NH3 vapor as an ideal gas.
(a) Find the pressure p0 and the temperature T0 at the triple point.
(lg)
(b) Find the latent heat L0 (in units of Joule) of the liquid-gas transition at the triple point.
(sg)
(c) If the latent heat of the solid-gas transition (at the triple point) is L0 = 3.143 × 104 J what
(sl)
is the latent heat L0 of the solid-liquid transition?

Solution:
Contents of this Document [ttc6]

6. Thermodynamics of Phase Transitions II

• Van der Waals equation of state with coexistence curve. [tsl10]

• Law of corresponding states. [tln30]
• Maxwell construction. [tln31]
• Gibbs and Helmholtz free energies of the van der Waals fluid at T < Tc .
[tsl11]
• Condensation and evaporation. [tln32]
• Dieterici equation of state. [tex34]
• Helium liquids. [tln33]
• Phase diagram of 4 He. [tsl13]
• Phase diagram of 3 He. [tsl14]
• Exotic properties of helium II. [tln34]
• Superconducting transition. [tln35]
• Thermodynamics of a ferromagnet. [tsl5]
• Structural transitions of iron. [tex53]
• Latent heat and heat capacities at superconducting transition. [tex44]
• Thermodynamics of the mean-field ferromagnet I. [tex45]
• Thermodynamics of the mean-field ferromagnet II. [tex46]
Van der Waals equation of state [tsl10]

with spinodal curve

with coexistence curve

Law of corresponding states [tln30]

Use the critical-point values, pc , Vc , Tc , ρc , for the thermodynamic variables

and introduce reduced quantities:
p V T ρl ρg
p̄ ≡ , V̄ ≡ , T̄ ≡ , ρ̄l ≡ , ρ̄g ≡ .
pc Vc Tc ρc ρc

Empirical fact: Near the critical point, the relations between (reduced) ther-
modynamic quantities are universal.

Experimentally:
Guggenheim plot of liquid-vapor coexistence curves:
1 3 7
(ρ̄l + ρ̄g ) ' 1 + (1 − T̄ ), ρ̄l − ρ̄g ' (1 − T̄ )1/3 .
2 4 2

Theoretically:
an2
 
Van der Waals equation: p + 2 (V − nb) = nRT .
V
   2 
∂p ∂ p
Critical point condition: = = 0.
∂V T ∂2V T
a 8a
Critical-point values: pc = ,
Vc = 3nb, Tc = .
27b2 27bR
 
3
VdW equation in reduced units: p̄ + 2 (3V̄ − 1) = 8T̄ .
V̄
Maxwell construction [tln31]

Goal: Study the liquid-gas transition for a fluid system described by the van
der Waals equation of state.
Consider an isotherm at T < Tc with roadmarks A–I as shown in [tsl11].
Determine the variation of the Gibbs potential along the subcritical isotherm:
Z p
dG = −SdT + V dp = V dp ⇒ G(T, p) = G(T, pA ) + dp V (p).
pA

Consider the self-intersecting curve in the plot G versus p.

G(T, pC ) = G(T, pG ) implies that area 1 = area 2.
Discuss stability of thermodynamic state along all segments of the curve A–I:
 
∂p
• Segments ABC, GHI: < 0, G(T, p) is either the only branch
∂V T
or the lowest branch of a multi-valued function. The state is stable.
 
∂p
• Segments CD, FG: < 0, G(T, p) is not the lowest branch of
∂V T
a multi-valued function. The state is metastable.
 
∂p
• Segment DEF: > 0. This implies that the state is unstable.
∂V T

The physical isotherm includes only stable states. It is described by the curve
ABCGHI in the plot G versus p.
Stability requires that the Gibbs potential G(T, p) is a concave function p
and that the Helmholtz potential A(T, V ) is a convex function of V :
   2   2 
∂p ∂ G ∂ A
<0 ⇒ 2
< 0, > 0.
∂V T ∂p T ∂V 2 T

Variation of the Helmholtz potential along the subcritical isotherm:

Z V
dA = −SdT − pdV = −pdV ⇒ A(T, V ) = A(T, VA ) − dv p(V ).
VA

In the unstable and metastable regions, A(T, V ) can be made smaller if we

replace the homogeneous system by a system with two coexisting phases.
Note: the (shaded) metastable region is bounded by the coexistence curve
(solid line) and the spinodal curve (dashed line).
Gibbs and Helmholtz free energies
of the van der Waals fluid at T < Tc [tsl11]
Condensation and evaporation [tln32]

Supersaturated gases and superheated liquids

owe their metastable existence to the surface p g Vg Tg
tension σ. Consider a liquid droplet in equi-
librium with the surrounding vapor, implying
Tl = Tg , µl = µg , and pl > pg because of sur- p l Vl
face tension. 2R
Tl
For a vapor bubble surrounded by liquid, the
argument proceeds along analogous lines.

Work done if droplet expands or contracts: δW = −pl dVl − pg dVg + σdA.

Grand potential: Ω(T, V, µ) = −pl Vl − pg Vg + σA.
 
4π 3 4π 3
⇒ Ω(T, V, µ) = − R pl − Vtot − R pg + 4πR2 σ.
3 3

Mechanical equilibrium: (∂Ω/∂R)T,V,µ = 0 ⇒ 4πR2 (pg − pl ) + 8πRσ = 0.

Excess pressure in droplet: pl − pg = 2σ/R.
Gibbs-Duhem equations (with dT = 0), Nl dµl = Vl dpl , Ng dµg = Vg dpg .
Chemical equilibrium: dµl = dµg ⇒ (Vl /Nl )dpl = (Vg /Ng )dpg .
 
Vg /Ng − Vl /Nl 2σ
Differential excess pressure: d(pl − pg ) = dpg = d .
Vl /Nl R
 
Vg Nl Vg kB T kB T /pg 2σ
Use  , ' ⇒ dpg = d .
Ng Vl Ng pg Vl /Nl R
 
dpg Vl 2σ
Integrate = d from ∞ to R.
pg Nl kB T R
 
pg (R) 2σVl 2σm 2σm
⇒ ln = = ⇒ pg (R) = pg (∞) exp .
pg (∞) RNl kB T Rρl kB T Rρl kB T
Only liquid droplets of a particular radius Rc coexist with the supersaturated
gas phase. Droplets with R < Rc will shrink. Droplets with R > Rc will
grow. Hence the condensation process at pressure p = pg (Rc ) can be initiated
by the presence of droplets with radius R > Rc .
Metastability depends on the absence of droplets with radius R > Rc . The
boundary of the metastable region (spinodal line) corresponds to a value of
Rc comparable to the molecular radius. Supersaturation cannot be pushed
beyond that point.
[tex34] Dieterici equation of state
The Dieterici equation of state of a fluid system reads
nRT  an 
p= exp − ,
V − nb RT V
where a, b are phenomenological constants.
(a) Show that the pressure, volume, and temperature at the critical point are
a a
pc = , Vc = 2nb, Tc = .
4b2 e2 4Rb
(b) Rewrite the Dieterici equation of state as a relation between the dimensionless quantities
p̄ ≡ p/pc , V̄ ≡ V /Vc , T̄ ≡ T /Tc (law of corresponding states).

Solution:
Helium liquids [tln33]

Helium has a small atomic mass and a weak interatomic interaction. This
enhances quantum effects. Solid helium exists only at high pressure. Helium
at low T and moderate p is a quantum liquid with peculiar features.
The two helium isotopes, 3 He and 4 He, are chemically similar but physically
very different. The physical difference is governed by the difference in nuclear
spin ( 21 versus zero).

4
He Features

• The liquid-vapor coexistence line terminates in a critical point.

• The solid-liquid transition line is monotonic and ends at T = 0 and
p ' 25atm with zero slope (dp/dt = ∆S/∆V = 0).
• The λ-line separates the normal fluid (He I) from the superfluid (He II)
via a continuous transition.
• Each end of the λ-line is a triple point.
• 4 He was first liquefied in 1908 by Kamerlingh Onnes (at 4.2K, 1atm).
• The λ-transition has been interpreted microscopically as the condensa-
tion of interacting bosons.

3
He Features

• The relative abundance of 3 He in natural helium is 10−6 .

• 3 He can be produced artificially from tritium (3 H) via β-decay.
• 3 He has not been available in large quantities until 1940 (Manhattan
project).
• 3 He was first liquefied in 1948.
• A superfluid transition in 3 He was first observed in 1971.
• The superfluid phase in 3 He is akin to the superconducting phase. It
is described by bound pairs of quasi-particles with spin 21 .
• The A-phase and the B-phase differ by the orientation of the bound
pairs.
• The negative slope in the solid-liquid coexistence curve is attributable
to an entropy effect of nuclear spins (dp/dt = ∆S/∆V < 0).
Phase diagram of 4He [tsl13]
Phase diagram of 3He [tsl14]

[from Enss and Hunklinger 2005]

Exotic Properties of helium II [tln34]

Helium II behaves like a mixture of normal fluid and superfluid. The su-
perfluid portion increases with decreasing temperature at the expense of the
normal fluid portion. The superfluid has no viscosity and no entropy.
Consider two vessels A and B with rigid insulating walls, connected by a
capillary that allows unimpeded superfluid flow but prevents any normal
fluid flow.
In general, the thermal equilibrium of that system is characterized by the
following relations between intensive variables:

TA 6= TB , pA 6= pB , µA (TA , pA ) = µB (TB , pB ).

Consider situations in which system B is very large compared to system A.

Any process in which a change of pA or TA is forced in the smaller system
must then satisfy µA (TA , pA ) = µB (TB , pB ) = const i.e. dµA = 0.
   
∂µA ∂µA SA VA
⇒ dTA + dpA = − dTA + dpA = 0.
∂TA pA ∂pA TA NA NA

SA
⇒ dpA = dTA .
VA

Mechanocaloric effect: Thermomechanical effect:

Pressure increase Temperature increase
causes temperature increase. causes pressure increase.

A A
B
powder powder
Superconducting transition [tln35]

Meissner-Ochsenfeld effect:
Observation that the magnetic induction B = µr µ0 H vanishes inside a su-
perconductor (of type I). B is expelled by surface supercurrents. However, a
sufficiently strong external magnetic field H destroys superconductivity.

B H coex

T = const H0 normal
conductor

µ r µ0H c coexistence
super− line
conductor

H T
Hcoex(T) Tc

sc nc

Coexistence between the superconducting and the normal conducting phases

requires G(sc) (T, H) = G(nc) (T, H) (Gibbs free energy per unit volume).
Along the coexistence line: dG(sc) = dG(nc) .

⇒ − S (nc) dT − B (nc) dH = −S (sc) dT − B (sc) dH

with B (nc) = µr µ0 Hcoex (T ) and B (sc) = 0.

 
(nc) (sc) dH
Clausius-Clapeyron equation: S −S = −µr µ0 Hcoex (T ) .
dT coex

Latent heat: L = T S (nc) − S (sc) .

As H increases, G(sc) stays constant but G(nc) decreases:

Z H
(nc) (nc) 1
G (T, H) − G (T, 0) = − B (nc) dH = − µr µ0 H 2 .
0 2

On the coexistence line: G(nc) (T, Hcoex ) = G(sc) (T, Hcoex ).

1
⇒ G(sc) (T, 0) − G(nc) (T, 0) = − µr µ0 Hcoex
2
(T ).
2
Thermodynamics of a Ferromagnet [tsl5]

Materials with ions that have permanent electronic magnetic moments may
be paramagnetic, ferromagnetic, antiferromagnetic, or exhibit some other
kind of magnetic ordering.
In ferromagnetic materials there exists a continuous transition from the para-
magnetic phase to the ferromagnetic phase when the system is cooled through
the Curie temperature Tc .

Spontaneous magnetization of nickel and iron:

Heat capacity of nickel:

(a) total heat capacity, (b) contribution of lattice vibrations
(c) electronic contribution, (d) magnetic contribution.
[tex53] Structural transitions of iron
At constant atmospheric pressure, the stable phase of Fe below 900◦ C and above 1400◦ C is α-iron.
Between these temperatures, the stable phase is γ-iron. The specific heat of each phase can be
taken as constant: cα = 0.775J/gK, cγ = 0.690J/gK. Find the latent heat (per gram) at each of
the two phase transitions.

Solution:
[tex44] Latent heat and heat capacities at superconducting transition
The coexistence line between the normal and superconducting phases of some metallic material is
observed to be well approximated by the empirical formula,

Hcoex (T ) = H0 (1 − T 2 /Tc2 ), 0 ≤ T ≤ Tc ,

where H0 , Tc are constants.

(a) Use the Clausius-Clapeyron equation adapted to this situation (see [tln35]) to calculate the
latent heat L (per unit volume) at the transition. Plot L versus T for 0 ≤ T ≤ Tc .
(b) When the material is heated up along the coexistence line, different specific heats obtain for
the two phases. Use the results of (a) to calculate the discontinuity ∆C = C (nc) − C (sc) (per unit
volume) across the coexistence line. Plot ∆C versus T for 0 ≤ T ≤ Tc .

Solution:
[tex45] Thermodynamics of the mean−field ferromagnet I
The mean-field ferromagnet is specified by the heat capacity CM = 0 and by the equation of state
M = tanh([H + λM ]/T ), where λ is a constant. In zero magnetic field (H = 0), this system
undergoes a continuous transition at temperature Tc = λ between a paramagnetic phase (M = 0)
and a ferromagnetic phase (M 6= 0).
(a) Determine the spontaneous magnetization M (T, H = 0) in the ferromagnetic phase by numer-
ically solving the equation of state at H = 0. Plot M versus T for 0 ≤ T ≤ Tc .
(b) Show that the entropy depends only on M :

1+M 1+M 1−M 1−M

S(M ) = − ln − ln .
2 2 2 2
Plot S versus T at H = 0 for 0 ≤ T ≤ 2Tc .
(c) Calculate an analytic expression for the Helmholtz free energy A(T, M ).

Solution:
[tex46] Thermodynamics of the mean−field ferromagnet II
The mean-field ferromagnet is specified by the heat capacity CM = 0 and by the equation of state
M = tanh([H + λM ]/T ), where λ is a constant. In zero magnetic field (H = 0), this system
undergoes a continuous transition at temperature Tc = λ between a paramagnetic phase (M = 0)
and a ferromagnetic phase (M 6= 0).
(a) Calculate an analytic expression for the isothermal susceptibility χT (T, M ) from the equation
of state. Use the numerically determined M (T, H = 0) from [tex45] to plot χT (T, H = 0) versus
T for 0 ≤ T ≤ 2Tc .
2
(b) Determine the heat capacity CH (T, M ) = T αH /(χT − χS ) from αH = (∂M/∂T )H , χT =
(∂M/∂H)T , χS = (∂M/∂H)S , and plot CH versus T for 0 ≤ T ≤ 2Tc .
(c) Plot in the same diagram (with different symbols) the function T (∂S/∂T )H=0 by using the
data of S(T, H = 0) from [tex45].

Solution:
Contents of this Document [ttc7]

7. Kinetic Theory I

• Statistical uncertainty and information. [tln37]

• Statistical concept of uncertainty. [tex47]
• Statistical uncertainty and information. [tln37]
• Statistical uncertainty and information. [tex48]
• Information of sequenced messages. [tex61]
• Kinetics of classical ideal gas. [tsl28]
• Pressure and mean square velocity in classical ideal gas. [tex49]
• Maxwell velocity distribution. [tln38]
• Maxwell velocity distribution (Maxwell’s derivation). [tex50]
• Maxwell distribution in D-dimensional space. [tex56]
• Boltzmannn equation. [tln39]
• Boltzmann’s H-theorem. [tln40]
• Energy distribution for N ideal gas atoms. [tex57]
• Maxwell velocity distribution (Boltzmann’s derivation). [mex58]
• Ideal-gas entropy and Boltzmann’s H-function. [tex59]
• H-theorem and entropy. [tln41]
• Boltzmann’s H-function simulated. [tsl27]
• Maxwell distribution derived from minimizing the H-function. [tex60]
• Doppler broadening of atomic spectral lines. [tex63]
Statistical uncertainty and information [tln37]

An experiment has n possible outcomes that occur with probabilities P1 , P2 , . . . , Pn .

Properties that must be satisfied by any quantitative measure of uncertainty:

1. The uncertainty is a function of the probabilities of all possible out-

comes: Σ = Σ(P1 , P2 , . . . , Pn ).
2. The uncertainty is symmetric under all permutations of the Pi .
3. The maximum uncertainty occurs if all Pi are equal.
4. The uncertainty is zero if one of the outcomes has probability Pi = 1.
5. The combined outcome of two independent experiments has an uncer-
tainty equal to the sum of the uncertainties of the outcomes of each
experiment.
n
X
⇒ Σ(P1 , P2 , . . . , Pn ) = − Pi ln Pi = −hln P i.
i=1

Information comes in messages: A1 , A2 , . . .. A message carries information

only if it contains some news, i.e. something not completely expected.
P (A): probability that message A is sent.
I(A): information gain if message is indeed received.
The less likely the message, the greater the information gain if the message
is received:
If P (A) < P (B) then I(A) > I(B), if P (A) = 1 then I(A) = 0.
If two independent messages are received, then the information gain is the
sum of the information gains pertaining to each individual message:

P (A ∩ B) = P (A)P (B) ⇒ I(A ∩ B) = I(A) + I(B).

The information content of a message is equal to the change in (statistical)

uncertainty at the receiver:
A
P1 , P2 , . . . , Pn −→ P̄1 , P̄2 , . . . , P̄n ⇒ I(A) = Σ(P1 , P2 , . . . , Pn )−Σ(P̄1 , P̄2 , . . . , P̄n )

Information as used here refers only to the scarcity of events. Any aspects
of usefulness and meaningfulness are disregarded.
[tex47] Statistical concept of uncertainty
An experiment has n possible outcomes that occur with probabilities P1 , . . . , Pn . The uncertainty
about the outcome of the experiment is defined as
n
X
Σ(P1 , . . . , Pn ) = − Pi ln Pi .
i=1

(a) Prove that the maximum uncertainty occurs if all Pi are equal.
(b) The n2 combined outcomes of two independent experiments have probabilities Pij = PiI PjII .
Show that the uncertainty about the combined outcome of the two independent experiments is
equal to the sum of the uncertainties of the outcomes of each experiment: Σ({Pij }) = Σ({PiI }) +
Σ({PiII }).

Solution:
[tex48] Statistical uncertainty and information
The number of bird species living on some continent is known to be 100. An ornithologist visits
a small island off the coast of that continent to find out how many of the 100 bird species have
migrated to the island.
One month after her arrival on the island she sends a first message to the Ornithological Society,
stating that there exist only five of the 100 bird species on the island.
A month later she sends a second message stating that the relative abundance of the five bird
populations identified previously is 80%, 10%, 5%, 3%, 2%.
Determine the numerical value of the information contained in each message.

Solution:
[tex61] Information of sequenced messages
In which months of the year do Ellen, Nancy, and Susan have their birthdays? Three messages
X, Y, Z about their birthdays are received by persons a, b, c:
Person a receives the messages in the sequence X, Y, Z.
Person b receives the messages in the sequence Z, X, Y .
Person c receives the messages in the sequence Y, Z, X.
The three messages are the following:
X: Nancy’s birthday is in April.
Y : Ellen’s birthday is in a later month than Nancy’s birthday.
Z: Susan’s birthday is in the same month as Ellen’s birthday.
Find the numerical value of the information contained in the messages X, Y, Z as received by each
person a, b, c.

Solution:
Kinetics of Classical Ideal Gas [tsl28]

• Gas consists of a large number of atoms.

• Motion of each atom is rectilinear with constant speed.
• Interactions are limited to collisions with walls or between atoms.
• Motion is randomized by collisions.
• Thermal equilibrium is characterized by uniform spatial distribution of
atoms and by a velocity distribution f (v) to be determined.

Position and velocity distribution in two dimensions.

Properties of velocity distribution f (v):

Z
• d3 vf (v) = 1 (normalization),
Z
• d3 vf (v) v = 0 (symmetry),
 
1 2 1 3
Z
3 U
• d vf (v) mv = mhv 2 i = = kB T.
2 2 N 2

1N 1N
Pressure [tex49]: p = mhv 2 i = 3kB T ⇒ pV = NkB T.
3V 3V
[tex49] Pressure and mean square velocity in classical ideal gas
A classical ideal gas consisting of N atoms of mass m is confined to a container of volume V . The
gas in thermal equilibrium with the walls is described by a spatially uniform distribution of atomic
positions and an isotropic distribution of velocities f (v). Show that the pressure exerted on the
container walls is
Z
1N
p= mhv i, where hv i = d3 v v 2 f (v).
2 2
3V

Solution:
Maxwell velocity distribution [tln38]

Criteria used by Maxwell:

• statistical independence: f (vx , vy , vz ) = f1 (vx )f1 (vy )f1 (vz ).

q 
• spherical symmetry: f1 (vx )f1 (vy )f1 (vz ) = f1 2 2 2
vx + vy + vz f1 (0)f1 (0).
1 1
• equipartition: mhvα2 i = kB T, α = x, y, z.
2 2
Velocity distribution:
3/2
m(vx2 + vy2 + vz2 )
  
m
⇒ f (vx , vy , vz ) = exp − .
2πkB T 2kB T

Speed distribution:
p
integrate f (vx , vy , vz ) over shell v < vx2 + vy2 + vz2 < v + dv.
 3/2
4 m 2 /2k
⇒ fs (v) = √ v 2 e−mv BT
.
π 2kB T

Energy distribution:
 3/2
1 1 2
Use E = mv 2 , 2
v dv = E 1/2 dE.
2 2 m
2 √
⇒ fE (E) = √ (kB T )−3/2 E e−E/kB T .
π

r
p 3kB T
Root-mean-square speed: hv 2 i = .
m
r
8kB T
Mean speed: hvi = .
πm
 r
dfs  2kB T
Most frequent speed: = 0 ⇒ v0 = .
dv v0 m
[tex50] Maxwell velocity distribution (Maxwell’s derivation)
In the original derivation of the velocity distribution f (vx , vy , vz ) for a classical ideal gas, Maxwell
used the following ingredients: (i) The Cartesian velocity components vx , vy , vz (interpreted as
stochastic variables) are statistically independent. (ii) The distribution f (vx , vy , vz ) is spherical
symmetric. (iii) The mean-square velocity is determined by the equipartition theorem. Determine
f (vx , vy , vz ) along these lines.

Solution:
[tex56] Maxwell distribution in D−dimensional space
The Maxwell velocity distribution of an ideal gas in D-dimensional space is
 D/2
m 2
f (v) = e−mv /2kB T
,
2πkB T
2
where v = (v1 , . . . , vD ) and vp = v12 + · · · + vD
2
. Determine the associated speed distribution fS (v),
2
the root-mean-square speed hv i, the average speed hvi, and the most frequent speed v0 from
dfS /dv|v0 = 0.

Solution:
Boltzmann equation [tln39]

How does an arbitrary nonequilibrium velocity distribution f (~v , t) approach

equilibrium? Boltzmann’s kinetic equation takes into account elastic pair
collisions, characterized by a scattering cross section σ(~v1 , ~v2 ; ~v10 , ~v20 ) that de-
pends on the velocities of the two particles before and after the collision.
During the infinitesimal time interval τ , the number of particles with veloci-
ties ~v1 d3 v1 changes due to contributions A and B from two kinds of processes:

[f (~v1 , t + τ ) − f (~v1 , t)] d3 v1 = B − A,

where the number of collisions away from ~v1 d3 v1 is

Z Z Z
A = τ d v1 d v2 d v1 d3 v20 σ(~v1 , ~v2 ; ~v10 , ~v20 )f (~v1 , t)f (~v2 , t)
3 3 3 0

and the number of collisions into ~v1 d3 v1 is

Z Z Z
B = τ d v1 d v2 d v1 d3 v20 σ(~v10 , ~v20 ; ~v1 , ~v2 )f (~v10 , t)f (~v20 , t).
3 3 3 0

v1 v’1 v’1 v1
v2 v’2 v’2 v2

A B
Here we have made the assumption of molecular chaos, which neglects cor-
relations produced by the collisions: f (2) (~v1 , ~v2 , t) = f (~v1 , t)f (~v2 , t).
Symmetry properties: σ(~v1 , ~v2 ; ~v10 , ~v20 ) = σ(~v2 , ~v1 ; ~v20 , ~v10 ) = σ(~v10 , ~v20 ; ~v1 , ~v2 ).
Boltzmann equation for a spatially uniform velocity distribution:
Z Z Z
∂
⇒ f (~v1 , t) = − d v2 d v1 d3 v20 σ(~v1 , ~v2 ; ~v10 , ~v20 )
3 3 0
∂t
× [f (~v1 , t)f (~v2 , t) − f (~v10 , t)f (~v20 , t)] .
Boltzmann’s H-theorem [tln40]

Z
Boltzmann’s H-function: H(t) ≡ d3 v1 f (~v1 , t) ln f (~v1 , t).

Z  
dH 3 ∂f (~v1 , t) ∂f (~v1 , t)
⇒ = d v1 ln f (~v1 , t) + .
dt ∂t ∂t
Z Z
3 ∂f (~v1 , t) d
Use d v1 = d3 v1 f (~v1 , t) = 0 and use Boltzmann equation.
∂t dt

Z Z Z Z
dH
⇒ = − 3
d v1 3
d v2 d3 v10 d3 v20 σ(~v1 , ~v2 ; ~v10 , ~v20 )
dt
× ln f (~v1 , t) [f (~v1 , t)f (~v2 , t) − f (~v10 , t)f (~v20 , t)] .

Likewise: dH/dt = · · · {~v1 ↔ ~v2 }, {~v10 ↔ ~v20 },

dH/dt = · · · {~v1 ↔ ~v10 }, {~v2 ↔ ~v20 },
dH/dt = · · · {~v1 ↔ ~v20 }, {~v2 ↔ ~v10 }.

Z Z Z Z
dH
⇒ 4 = − 3
d v1 d v23
d3 v10 d3 v20 σ(~v1 , ~v2 ; ~v10 , ~v20 )
dt
× [f (~v1 , t)f (~v2 , t) − f (~v10 , t)f (~v20 , t)]
× {ln [f (~v1 , t)f (~v2 , t)] − ln [f (~v10 , t)f (~v20 , t)]} .

The function h(x, y) ≡ (x − y)(ln x − ln y) is non-negative for x, y > 0 and is

equal to zero if x = y.
dH dH
Properties of H(t): ≤ 0 and = 0 if f (~v1 , t)f (~v2 , t) = f (~v10 , t)f (~v20 , t).
dt dt
The (stationary) velocity distribution which makes H stationary is the Maxwell
distribution (Boltzmann’s derivation).
Boltzmann’s H-function is related to the uncertainty in our knowledge of the
particle velocities as contained in the distribution f (~v1 , t): H(t) = −Σf .
The stationary H-function is related to the entropy of an ideal gas at equilib-
rium: S = −N kB H(∞). Here the uncertainty in our knowledge of particle
velocities is a maximum.
[tex57] Energy distribution for N ideal gas atoms.
The equilibrium velocity distribution for N atoms of a classical ideal gas is
 3N/2
m 2 2
f (v1 , . . . vN ) = e−m(v1 +···+vN )/2kB T ,
2πkB T

where vi = (vix , viy , viz ).

(a) Determine the associated energy distribution fE (E), where E = 21 m(v12 + · · · + vN
2
).
(b) Define the function Fn (x) via Fn (x)dx = fE (E)dE with x = E/nkB T, n = 3N/2 − 1 and plot
n−1 Fn (x), 0 < x < 4 for N = 1, 2, 10, 20.
(c) How is the trend of this function for increasing N to be interpreted?

Solution:
[tex58] Maxwell’s velocity distribution (Boltzmann’s derivation)
The velocity distribution f (v) is guaranteed to be a stationary solution of the Boltzmann equation if
it satisfies the equation f (v1 )f (v2 ) = f (v10 )f (v20 ), where v1 , v2 and v10 , v20 are the velocities before
and after an elastic pair collison. Elasticity means that the four quantities px = m(v1x + v2x ), py =
m(v1y + v2y ), pz = m(v1z + v2z ), E = 21 m(v12 + v22 ) are conserved by the collision.
Boltzmann uses the following arguments: (i) The absence of any further conservation laws implies
that f (v1 )f (v2 ) = F (px , py , pz , E); (ii) in the relation ln f (v1 ) + ln f (v2 ) = ln F (px , py , pz , E) the
additivity of the two functions on the left-hand side implies that ln F is a linear function of its
variables: ln F (px , py , pz , E) = a1 px + a2 py + a3 pz + a4 E + a5 .
Show
R 3 that if the five coefficients a1 , . . . , a5 are determined1 such as to satisfy the requirements
d vf (v) = 1 (normalization), hvi = 0 (symmetry), and 2 mhv 2 i = 23 kB T (equipartition), then
f (v) is the Maxwell distribution.

Solution:
[tex59] Ideal−gas entropy and Boltzmann’s H−function
Consider N particles of a classical monatomic ideal gas confined to a box of volume V at tem-
perature T . Show that the entropy S(T, V, N ) = S0 + nR ln[(T /T0 )3/2 (V /V0 )] previously inferred
from the empirical relations pV = nRT, CV = 32 nR can be derived via S = −N kB H(∞) from the
stationary value of Boltzmann’s H-function,
Z Z
H(t) = d r d3 v f (r, v, t) ln f (r, v, t).
3

Solution:
H-theorem and irreversibility [tln41]

Q: How does the preferred time direction, selected by the monotonic time-
dependence of H(t), follow from the underlying microscopic dynamics, which
is invariant under time reversal?
A: The solution f (~v1 , t) of the Boltzmann equation is to be interpreted as rep-
resenting the properties of an ensemble of systems, i.e. the average behavior
of systems that are prepared equally (on a macroscopic level).
Z
Consider the function H̃(t) = d3 v1 f˜(~v1 , t) ln f˜(~v1 , t),
calculated via computer simulation, where f˜(~v1 , t) now represents the velocity
distribution of a single system.
Simulation data show that H̃(t) tends to decrease and approach an asymp-
totic value just as the function H(t) does.
Effect of velocity inversion at time tI : H̃(t) increases at t > tI for some time,
then decreases again and approaches the same asymptotic value as H(t) does.
We can interpret −H̃(t) as our uncertainty about the particle velocities in
the system. The information contained in f˜(~v1 , t) over and above the three
general properties from which the Maxwell distibution was derived is H̃(t) −
H̃(∞). However, this information is insufficient to carry out the velocity
inversion.
Performing the velocity inversion requires an influx of information beyond
what is contained in f˜(~v1 , t), which causes a discontinuous drop in uncertainty
of our knowledge about the particle velocities. At t = tI , where the velocity
inversion occurs, Boltzmann’s function H(t) jumps to a higher value and
then decreases gradually as the information injected gets lost gradually in
the wake of collisions.

~ ~
H H=H

t
tI
Boltzmann’s H-function simulated [tsl27]

Computer simulation of 100 hard disks moving in a 2D box and undergoing

elastic collisions. Initial state:
R positions on a regular lattice, velocities ran-
dom. Open circles: H(t) = d2 v f (v, t) ln f (v, t). Full circles: H(t) when
all velocities are inverted after 50 or 100 collisions.

[from Prigogine 1980]

[tex60] Maxwell distribution derived from minimizing the H−function
Minimize Boltzmann’s H-function
Z
H(t) = d3 v f (v, t) ln f (v, t)

for the spacially uniform velocity distribution f (v, t) of a classical ideal gas. Impose the integral
constraints Z Z
3 1 3
d v f (v, t) = 1, m d3 v v 2 f (v, t) = kB T,
2 2
dictated by normalization and equipartition, respectively. Show that the resulting velocity distri-
bution is Maxwellian.

Solution:
[tex63] Doppler broadening of atomic spectral lines
Consider a furnace containing a dilute gas at high temperature. Through a small window of
the furnace, we observe a particular spectral line of the gas atoms by means of a spectrometer.
The width of the observed spectral line is broadened due to the spread of velocities of the gas
atoms. This
p effect is called Doppler broadening. The relativistic Doppler shift of the wavelength
is λ = λ0 (1 + v/c)/(1 − v/c). For the case under consideration we can assume that v/c  1.
Show that the intensity profile is given by the expression

mc2 (λ − λ0 )2
 
I(λ) ∝ exp − ,
2λ20 kB T

where T is the temperature of the furnace, c is the speed of light, m is the mass of the gas atoms,
and λ0 is the wavelength of the radiation emitted by an atom at rest.

Solution:
Contents of this Document [ttc9]

9. Microcanonical Ensemble

• Classical Hamiltonian system. [tln45]

• Classical Liouville operator. [tln46]
• Quantum Liouville operator. [tln47]
• Gibbs entropy. [tln48]
• Microcanonical ensemble. [tln49]
• Classical ideal gas (microcanonical ensemble). [tex73]
• Array of classical harmonic oscillators (microcanonical ensemble). [tex74]
• Quantum harmonic oscillators (microcanocal ensemble I). [tex75]
• Quantum harmonic oscillators (microcanocal ensemble II). [tex126]
• Quantum paramagnet (microcanonical ensemble). [tex127]
• Entropy of mixing revisited. [tln50]
[tex62] Ideal gas atoms escaping from a container
A dilute gas is confined to a large vessel in thermal equilibrium at temperature T .
(a) Find the rate at which gas atoms escape into the vacuum through a tiny hole of area A in the
wall of the vessel.
(b) If the wall with the hole is perpendicular to the z-axis, find the distribution fz (vz ) for the gas
atoms escaping through the hole.

Solution:
[tex64] Toward thermal equilibrium via particle transfer
A vessel with insulating walls is divided into two compartments by an internal wall that is also
insulating but has a small hole of area A. The two compartments contain dilute gases of slightly
different densities, n± = n ± 12 dn, at slightly different temperatures, T± = T ± 21 dT .
(a) Show that the rates at which particles and energy are transferred through the hole are (in
leading orders of dn and dT ):
p  √  
dN A 1 n dE A 2 3 p
=√ kB T dn + √ d(kB T ) , =√ (kB T )3/2 dn + n kB T d(kB T ) .
dt 2πm 2 kB T dt πm 2

(b) If the compartment with the higher particle density is at the lower temperature it is possible
to create situations where either the particle flow or the energy flow is zero. Find the values of
dn/dT in terms of n and T for which we have either dN/dt = 0 or dE/dt = 0.

Solution:
[tex65] Isotope separation via diffusion
A vessel is divided into two compartments by a thin wall with many holes. The chamber on the
left contains a dilute gas mixture of two isotopes (masses mA , mB ; particle densities nA , nB ) of
some atom. As the gas diffuses into the chamber on the right, it is evacuated immediately by a
pump into the left chamber of an identical vessel.
(a) Find the ratio of the numbers of type B and type A particles that are pumped out of the first
vessel.
(b) Consider a battery of 20 vessels and pumps connected in series. If nA = nB in the left chamber
of the first vessel and if mA /mB = 0.8, find n′A /n′B of the gas when it is pumped into a container
by the last pump.

Solution:
Kinematic pressure and interaction pressure [tln42]

The kinematic pressure is dominant in gases and the interaction pressure is

dominant in liquids.
Kinematic pressure due to particles carrying net momentum across a sur-
face:
Impulse equals momentum transfer: F dt = Apkin dt = Pin − Pout .
Z Z
3
Pin = d vf (~v )(mvx )n|Avx dt|, Pout = d3 vf (~v )(mvx )n|Avx dt|.
vx >0 vx <0
Z
1
⇒ pkin = nm d3 vf (~v )vx2 = nmhv 2 i.
3

Interaction pressure due to interparticle force exerted across surface:

Consider a central-force potential φ(r).
Potential energy of particle at position (−x0 , 0, 0) due to the presence of all
particles at x > 0:
Z ∞ Z +∞ Z +∞ p 
U (−x0 ) = n dx dy dz φ (x + x0 )2 + y 2 + z 2 .
0 −∞ −∞

Force on particle at x = −x0 : F (x0 ) = −U 0 (−x0 ).

Here F > 0 means repulsive and F < 0 means attractive.
Total force exerted on particles at x < 0 by particles at x > 0:
Z ∞
Ftot
Ftot = nA dx0 F (x0 ) ⇒ pint = .
0 A

Note: For realistic interparticle potentials with repulsive core and attractive
tail, the interaction pressure is negative at low densities. This is effectively
taken into account by the van der Waals equation of state.

y (x,y,z)
A F=Ap A
x (−x0 ,0,0)
z

kinematic interaction
[tex66] Interaction pressure produced by Gaussian interparticle potential
Consider a dilute gas of density n, where the particles interact via a Gaussian central-force poten-
2 2
tial, φ(r) = φ0 e−r /a , with φ0 = 1eV = 1.6 × 10−19 J, a = 2 × 10−10 m.
(a) Calculate the interaction pressure pint under the assumption that the particles are distributed
randomly in space. Express the result as a function of φ0 , n, a.
(b) Compare the interaction pressure pint with the kinetic pressure pkin for a dilute gas at T = 293K
and n = 2.7 × 1025 m−3 .

Solution:
Kinetic forces and mobility [tln43]

Consider a single-velocity beam: a shower of particles with mass m, all with

velocity ~v0 , distributed randomly in space with particle density n0 .

Situation #1: A hard wall of area A and

A u
normal unit vector ~n moves with velocity
~u through the path of the single-velocity
beam. Find the kinetic force experienced n
by the wall.
v0

Rate of collisions (viewed from the rest frame of the wall):

dN
= −n0 A [~n · (~v0 − ~u)] if ~n · (~v0 − ~u) < 0.
dt

Momentum transfer per collision: ∆P~ = 2m~n [~n · (~v0 − ~u)] .

dN ~
Kinetic force: F~ = ∆P = −2mn0 A~n [~n · (~v0 − ~u)]2 .
dt

Situation #2: Consider a heavy hard sphere of radius R moving with veloc-
ity ~u in the path of a single-velocity beam of light particles (mass m, velocity
~v0 , density n0 ). The kinetic force experienced by the sphere is calculated in
exercise [tex68]:
F~ = πmn0 R2 |~v0 − ~u|(~v0 − ~u).

Situation #3: The mobility constant µ in the equation ~u = µF~app relates the
steady state velocity ~u of an object moving through a fluid to the external
force applied to the object. In steady-state motion, the external force is
balanced by the kinetic force F~ exerted by the fluid particles on the object:
F~app = −F~ . The kinetic force exerted by a dilute gas (density n, particle
mass m, temperature T ) on a slowly moving heavy hard sphere (radius R,
velocity ~u with u  hvi) is calculated in exercise [tex69]:
8 p
F~ = − 2πmkB T R2 n~u.
3
[tex68] Average force of particle beam on heavy hard sphere
Consider a heavy hard sphere of radius R moving with velocity u in the path of a single-velocity
beam of light particles (mass m, velocity v0 , density n0 ). Show that the average force exerted by
the beam on the sphere is
F = πmn0 R2 |v0 − u|(v0 − u).

Solution:
[tex69] Mobility of a hard sphere in a dilute gas
The mobility constant µ in the equation u = µFapp relates the steady state velocity u of an object
moving through a fluid to the external force applied to the object. In steady-state motion, the
external force is balanced by the average force F exerted by the fluid particles on the object:
Fapp = −F.
Show that the average force exerted by a dilute gas (density n, particle mass m, temperature T )
on a slowly moving heavy hard sphere (radius R, velocity u with u  hvi) is
8p
F=− 2πmkB T R2 nu.
3

Solution:
Collision rate and mean free path [tln44]

Consider two single-velocity beams of hard spheres with diameter d, mass m,

particle densities n1 , n2 , and velocities ~v1 , v~2 .
Find the collision rate of particles in a region of volume Ω at the intersection
of the two beams.

Ω v2
v1

View from rest frame of beam 1.

Number of particles in Ω: N1 = n1 Ω, N2 = n2 Ω.
Volume swept by one particle 2 inside Ω in time dt: ω2 = πd2 |~v2 − ~v1 |dt.
Volume swept by all particles inside Ω: Ω2 = N2 ω2 .
Number of particles 1 inside Ω that will be hit in time dt: dN = n1 Ω2 dt.
dN
Collision rate: Rcoll = = n1 n2 Ωπd2 |~v2 − ~v1 |.
dt

Collision rate in classical ideal gas:

From the above result, the rate of particle collisions within a region Ω of a
classical ideal gas with density n in thermal equilibrium at temperature T is
then calculated in exercise [tex70]:
p
R = 2Ωd2 n2 πkB T /m.

Mean free path of particle in classical ideal gas:

From the collision rate of particles in a classical ideal gas, the the mean
free path (average distance travelled between collisons) is then calculated in
exercise [tex71]:
1
`= √ .
2πd2 n
[tex70] Collision rate in classical ideal gas
Given the collision rate Rcoll = Ωn1 n2 πd2 |v2 − v1 | in a region of volume Ω in the path of two
single-velocity beams of particles (diameter d, mass m, velocities v1 , v2 , densities n1 , n2 ), show that
the collision rate within a region Ω of a classical ideal gas with density n in thermal equilibrium
at temperature T is p
R = 2Ωd2 n2 πkB T /m.

Solution:
[tex71] Mean free path of particle in classical ideal gas
Given that the collision rate of particles (diameter d, mass m) in a region of volume Ω
pof a classical
ideal gas with density n in thermal equilibrium at temperature T is R = 2Ωd2 n2 πkB T /m as
demonstrated in [tex70]. show that the average distance traveled by a particle between collisions
(mean free path) is
1
`= √ .
2πd2 n

Solution:
[tex67] Rate of chemical reaction A + A → A2 in gas phase
The rate at which a chemical reaction of the type A + A → A2 takes place in a dilute gas is
R = αNR , where α is a constant and NR is the density of pairs of atoms with a center-of-mass
kinetic energy Kcm in excess of some value 0 . Here Kcm is defined as the kinetic energy of the
two particles in a reference frame that moves with the center-of-mass velocity.
Show that
2αn2 ∞
Z
2 p
R= √ dx x2 e−x , x0 = 0 /kB T ,
π x0
where n is the particle density.

Solution:
[tex72] Effect of escaping particles on temperature of 1D ideal gas
A classical ideal gas (particle mass m, particle density n, temperature T , energy density u =
1
2 nkB T ) is contained in the region 0 < x < L. The particles can only move in ±x-direction. We
assume that the velocities satisfy a 1D Maxwell distribution at all times. All particles that hit the
wall at x = 0 are reflected elastically. The wall at x = L allows any particle that hits it to pass
through with a probability 0  1, independent of the particle’s energy. Otherwise the particle is
reflected elastically.
(a) Calculate the rate at which the system loses particles and energy. Express the rates in the form
dn/dt = fn (n, T ) and du/dt = fu (n, T ).
(b) The slowly varying particle density n(t) and energy density u(t) cause a slowly varying tem-
perature T (t) of the remaining gas. Derive from the results of (a) a differential equation for the
function T (t) and solve it.

Solution:
Contents of this Document [ttc8]

8. Kinetic Theory II

• Ideal gas atoms escaping from a container. [tex62]

• Toward thermal equilibrium via particle transfer. [tex64]
• Isotope separation via diffusion. [tex65]
• Kinematic pressure and interaction pressure. [tln42]
• Interaction pressure produced by Gaussian interparticle potential. [tex66]
• Kinetic forces and mobility. [tln43]
• Average force of particle beam on heavy hard sphere. [tex68]
• Mobility of a hard sphere in a dilute gas. [tex69]
• Collision rate and mean free path. [tln44]
• Collision rate in classical ideal gas. [tex70]
• Mean free path of particle in classical ideal gas. [tex71]
• Rate of chemical reaction A + A → A2 in gas phase. [tex67]
• Effect of escaping particles on temperature of 1D ideal gas. [tex72]
Classical Hamiltonian system [tln45]

Consider an autonomous classical dynamical system with 3N degrees of free-

dom (e.g. N particles in a 3D box with reflecting walls). The dynamics
is fully described by 6N independent variables, e.g. by a set of canonical
coordinates q1 , . . . , q3N ; p1 , . . . , p3N .
The time evolution of these coordinates is specified by a Hamiltonian function
H(q1 , . . . , q3N ; p1 , . . . , p3N ) and determined by the canonical equations:

∂H ∂H
q̇i = , ṗi = − ; i = 1, . . . , 3N
∂pi ∂qi

The time evolution of an arbitrary dynamical variable f (q1 , . . . , q3N ; p1 , . . . , p3N )

is determined by Hamilton’s equation of motion:
3N   3N  
df X ∂f ∂f X ∂f ∂H ∂f ∂H
= q̇i + ṗi = − ≡ {f, H}.
dt i=1
∂qi ∂pi i=1
∂q i ∂p i ∂p i ∂q i

df
Conserved quantity: = 0 ⇔ {f, H} = 0.
dt
dH
Energy conservation is guaranteed: = 0 because {H, H} = 0.
dt
The microstate of the system is specified by one point in the 6N -dimensional
phase space (Γ-space): X ≡ (q1 , . . . , q3N ; p1 , . . . , p3N ). As time evolves, this
point traces a trajectory through Γ-space.
The conservation law H(q1 , . . . , q3N ; p1 , . . . , p3N ) = const confines the motion
of any phase point to a 6N − 1-dimensional hypersurface in Γ-space. Other
conservation laws, provided they exist, will further reduce the dimensionality
of the manifold to which phase-space trajectories are confined.
Note: Within the framework of kinetic theory, the microstate of the same
system was described by N points in the 6D space spanned the position and
velocity coordinates of a single particle, (x, y, z; vx , vy , vz ).
Our knowledge of the instantaneous microstate of the system is expressed by
a probability density ρ(X, t) in Γ-space.
Z
Normalization: d6N X ρ(X, t) = 1.
Γ
Z
Instantaneous expectation value: hf i = d6N X f (X)ρ(X, t).
Γ

Maximum knowledge about microstate realized for ρ(X, 0) = δ(X − X0 ).

Classical Liouville operator [tln46]

To describe the time evolution of ρ(X, t) we consider a volume V0 with surface

S0 in Γ-space. The following equations relate the change of probability inside
V0 to the flow of probability through S0 and use Gauss’ theorem.
Z I Z
∂ 6N
d X ρ(X, t) = − ds · Ẋρ(X, t) = − d6N X ∇X · [Ẋρ(X, t)].
∂t V0 S0 V0

∂
Balance equation: ρ(X, t) + ∇X · [Ẋρ(X, t)] = 0.
∂t
Use ∇X · [Ẋρ] = ρ∇X · Ẋ + Ẋ · ∇X ρ and ∇X · Ẋ = 0.
∂
⇒ ρ(X, t) + Ẋ · ∇X ρ(X, t) = 0.
∂t
d ∂
Introduce convective derivative: ≡ + Ẋ · ∇X .
dt ∂t
d
Liouville theorem: ρ(X, t) = 0.
dt
3N   X3N  
X ∂ρ ∂ρ ∂ρ ∂H ∂ρ ∂H
Use Ẋ · ∇X ρ = q̇i + ṗi = − = {ρ, H}.
i=1
∂q i ∂p i i=1
∂q i ∂p i ∂p i ∂q i

3N  
X ∂H ∂ ∂H ∂
Liouville operator: L ≡ i{H, } = i − .
i=1
∂qi ∂pi ∂pi ∂qi

∂ρ
Liouville equation: i = i{H, ρ} = Lρ.
∂t
Formal solution: ρ(X, t) = e−iLt ρ(X, 0).
L is a Hermitian operator. Hence all its eigenvalues are real. Hence ρ(X, t)
cannot relax to equilibrium in any obvious way. The Liouville equation
reflects the time reversal symmetry of the underlying microscopic dynamics.
Obtaining the broken time reversal symmetry of irreversible processes from
the Liouville equation is a central problem in statistical mechanics (topic of
ergodic theory).
Nevertheless: the thermal equilibrium is described by a stationary (time-
independent) probability density:
∂ρ
= 0 ⇒ Lρ = 0 ⇒ {H, ρ} = 0.
∂t
A stationary ρ is an eigenfunction of L with eigenvalue zero. If ρ = ρ(H)
then {H, ρ} = 0. Hence ρ is time-independent.
Quantum Liouville operator [tln47]

The density operator ρ(t) is a positive definite Hermitian operator. Like its
classical counterpart, the phase-space density ρ(X, t), it describes what we
know about the state of the system.
Normalization: Tr[ρ(t)] = 1. Expectation value: hA(t)i = Tr[Aρ(t)].
X
Diagonal representation: ρ(t) = pi |πi (t)ihπi (t)|.
i
pi : probability of finding the system in the state |πi (t)i.
X X
⇒ hA(t)i = pi hπi (t)|A|πi (t)i = hn|A|n0 ihn0 |ρ(t)|ni.
i nn0

{|ni}: orthonormal basis. hn0 |ρ(t)|ni: elements of the density matrix.

∂
Schrödinger equation: H|πi (t)i = i~ |πi (t)i.
∂t
∂ρ X
⇒ i~ = pi [H|πi (t)ihπi (t)| − |πi (t)ihπi (t)|H] = Hρ − ρH = [H, ρ].
∂t i

1
Liouville operator: L ≡ [H, ].
~
∂ρ 1
Liouville equation: i = [H, ρ] = Lρ.
∂t ~
Formal solution: ρ(t) = e−iLt ρ(0) = e−iHt/~ ρ(0)eiHt/~ .
Time evolution carried by density operator or by dynamical variable:

hA(t)i = Tr[Ae−iHt/~ ρeiHt/~ ] = Tr[eiHt/~ Ae−iHt/~ ρ].

∂ρ
von Neumann equation: i~ = [H, ρ] ⇒ ρ(t) = e−iHt/~ ρ(0)eiHt/~ .
∂t
∂A
Heisenberg equation: i~ = −[H, A] ⇒ A(t) = eiHt/~ A(0)e−iHt/~ .
∂t
Density matrix in energy representation H|λi = Eλ |λi:
X
ρλλ0 (t) = hλ|ρ|λ0 ie−i(Eλ −Eλ0 )t/~ .
λλ0

∂ρ
Stationarity of density operator: i~ = 0 ⇒ [H, ρ] = 0.
∂t
P
⇒ ρ is diagonal in the energy representation: ρ = λ pλ |λihλ|.
Gibbs entropy [tln48]

At thermal equilibrium: ∂ρ/∂t = 0. This condition is satisfied by ρ = ρ(H).

Q: What is the functional dependence of ρ on H?
A: ρ(H) must maximize the entropy S(ρ) subject to the constraints related
to whether the system is isolated, closed, or open.
Q: What is the functional dependence of S on ρ?
A: The Gibbs entropy can be motivated by Boltzmann’s H-function and by
Shannon’s concept of uncertainty:Z
• classical system: S = −kB d6N X ρ(X) ln[CN ρ(X)],
• quantum system: S = −kB Tr[ρ ln ρ].
The additive constant CN in the classical expression allocates a certain phase-
space volume element to every microstate:
• distinguishable particles: CN = h3N , h ' 6.62 × 10−34 Js,
• indistinguishable particles: CN = h3N N !.
The factor N ! is needed to compensate for overcounting indistinguishable
permutations of identical particles. No correction is necessary in quantum
mechanics, where microstates have definite permutation symmetries.
Q: Why does one microstate require a nonzero phase-space volume element?
A: The Heisenberg uncertainty principle, ∆qi ∆pi ≥ 12 ~, must be satisfied to
accommodate quantum mechanical microstates.
Q: What is the precise size of that volume needed for one microstate?
A: The volume element is h3N for a system with 3N degrees of freedom.
3N  
6N 1 6N Y 1
Number of microstates in volume element d X: 3N d X = dqi dpi .
h i=1
h

Illustration: harmonic oscillator (2D phase space).

p2
 
1 2 2 hω 1
Hamiltonian: H = + mω q = n+ .
2m 2 2π 2
Classical trajectories are concentric ellipses with axes 2qmax , 2pmax .
Energy quantization implies quantized amplitudes:
s   s  
h 1 hmω 1
qmax = n+ , pmax = n+ .
πmω 2 π 2

Area of ellipse: A(n) = πqmax pmax = h(n + 1/2) ⇒ A(n + 1) − A(n) = h.

Microcanonical ensemble [tln49]

Consider an isolated classical system (volume V , N particles, internal en-

ergy U ). The goal is to determine the thermodynamic potential U (S, V, N )
pertaining to that situation, from which all other thermodynamic properties
can be derived.
Z
Maximize Gibbs entropy S = −kB d6N X ρ(X) ln[CN ρ(X)]
Z U ≤H(X)≤U +∆

subject to the contraint d6N X ρ(X) = 1 (normalization).

U ≤H(X)≤U +∆

Apply calculus of variation with one Lagrange multiplier:

Z
δ d6N X{−kB ρ ln[CN ρ] + α0 ρ} = 0
Z
⇒ d6N X δρ{−kB ln[CN ρ] − kB + α0 } = 0.
   −1
1 α0 Ω∆ U ≤ H(X) ≤ U + ∆
⇒ ρ(X) = exp − 1 = const =
CN kB 0 otherwise,
Z
where Ω∆ (U, V, N ) = d6N X is the volume of the energy shell.
U ≤H(X)≤U +∆
 
Ω∆ (U, V, N )
Thermodyn. potentials: S(U, V, N ) = kB ln ⇒ U (S, V, N ).
CN
The dependence of S or U on the energy width ∆ is undesirable and can,
in fact, be avoided. If we replace the shell volume Ω∆ (U, V, N ) by the entire
volume inside the shell, Ω(U, V, N ), the resulting expression for S(U, V, N )
differs only by a term ∝ ln N , which is negligible in macroscopic systems.
  Z
Ω(U, V, N )
⇒ S(U, V, N ) = kB ln , where Ω(U, V, N ) = d6N X.
CN H(X)≤U

Consider an isolated quantum system: H|ni = En |ni.

X X
Maximize S = −kB pn ln pn subject to the constraint pn = 1.
En <U En <U

1 X
⇒ pn = where N< (U ) = . ⇒ S = kB ln[N< (U )].
N< (U ) E <U n
[tex73] Classical ideal gas (microcanonical ensemble)
Consider a classical ideal gas of N atoms confined to an insulating box of volume V . The Hamil-
tonian of the system reflects the kinetic energy of 3N noninteracting degrees of freedom:
3N
X p2i
H= .
i=1
2m

(a) Calculate the entropy S(U, V, N ) in the microcanonical ensemble. Show that the result is the
Sackur-Tetrode equation:
"  3/2 #
5 V 4πmU
S(U, V, N ) = N kB + N kB ln .
2 N h3 3N

(b) Derive the internal energy U = 23 N kB T and the equation of state pV = N kB T from S(U, V, N ).

Solution:
[tex74] Array of classical harmonic oscillators (microcanonical ensemble)
Consider an array of N 3-dimensional classical harmonic oscillators, representing a system of 3N
uncoupled degrees of freedom:
3N  2 
X pi 1
H= + mω 2 qi2 .
i=1
2m 2

(a) Calculate the entropy S(U, V, N ) of this system in the microcanonical ensemble.
(b) Derive the internal energy U (T, V, N ), and the heat capacity C = (∂U/∂T )V N .

Solution:
[tex75] Quantum harmonic oscillators (microcanonical ensemble I)
Consider an array of N quantum harmonic oscillators:
N   
X 1
H= ~ω ni + , ni = 0, 1, 2, . . . .
i=1
2

(a) Calculate the entropy S(U, N ) of this system in the microcanonical ensemble via combinatorics
as follows: Set U = U0 + M ~ω, U0 = 21 N ~ω, M = n1 + · · · + nN . Next determine the number
N∆ (M, N ) of configurations (n1 , · · · , nN ) for fixed values of M, N . Then relate S to N∆ .
(b) Derive the internal energy U (T, N ), and the heat capacity C = (∂U/∂T )N from S(U, N ).

Solution:
[tex126] Quantum harmonic oscillators (microcanonical ensemble II)
Consider an array of N quantum harmonic oscillators:
N   
X 1
H= ~ω nj + , nj = 0, 1, 2, . . . .
j=1
2

(a) Calculate the entropy S(U, N ) of this system in the microcanonical ensemble via saddle point
method
P as follows. Express the number of distinct microstate at energy U in the form N∆ =
P
n1 · · · nN δ(U −H) with the δ-function replaced by its Fourier integral. Then use the asymptotic
Laplace expression for the integral and evaluate it retaining only contributions that are significant
in the thermodynamic limit.
(b) Derive the internal energy U (T, N ), and the heat capacity C = (∂U/∂T )N from S(U, N ).

Solution:
[tex127] Quantum paramagnet (microcanoncal ensemble)
Consider an array of N noninteracting localized magnetic dipole moments mi produced by localized
electron spins in a paramagnetic insulator. In the presence of a magnetic field H pointing in z-
direction, the Hamiltonian of this system represents the Zeeman energy:
N
X N
X N
X
H=− mi · H = −H mzi = −h σi .
i=1 i=1 i=1

where h = H/2 and σi = 2mzi = ±1.

(a) Calculate the entropy S(E, h, N ) of this system in the microcanonical ensemble via saddle
point method
P as
P follows. Express the number of distinct microstate at enthalpy E in the form
N∆ = σ1 · · · σN δ(E − H) with the δ-function replaced by its Fourier integral. Then use the
asymptotic Laplace expression for the integral and evaluate it retaining only contributions that
are significant in the thermodynamic limit.
(b) Calculate from S(E, h, N ) an explicit expression for the enthalpy E(T, h, N ) and derive from
it the magnetization M (T, h, N ).

Solution:
Entropy of mixing revisited [tln50]

Consider two dilute gases in a rigid and insu-

lating box separated by a mobile conducting p T p T
wall as in [tln25]: N1 atoms on the left and
N2 atoms on the right. N1 V1 N2 V2
At thermal equilibrium: N1 /V1 = N2 /V2 .
The removal of the internal wall initiates the mixing of particles 1 and 2.
Is the process reversible or irreversible? The answer depends on whether
particles 1 and 2 are of the same kind (indistinguishable) or of a different
kind (distinguishable).
Mixing occurs without changes in any of the following quantities:

• total internal energy: U = U1 + U2 ,

• total volume: V = V1 + V2 ,
• total number of particles: N = N1 + N2 .

Consider the Sackur-Tetrode formula for the entropy of an ideal gas [tex73]:
"  3/2 #
5 V 4πmU
S(U, V, N ) = N kB + N kB ln .
2 N h3 3N

Distinguishable particles: irreversible process

Initial entropy: Sinit = S(U1 , V1 , N1 ) + S(U2 , V2 , N2 )
Final entropy: Sf in = S(U1 , V1 + V2 , N1 ) + S(U2 , V1 + V2 , N2 )
V1 + V2 V1 + V2
Entropy change: ∆S = N1 kB ln + N2 kB ln > 0.
V1 V2

Indistinguishable particles: reversible process

Initial entropy: Sinit = S(U1 , V1 , N1 ) + S(U2 , V2 , N2 )
Final entropy: Sf in = S(U1 + U2 , V1 + V2 , N1 + N2 )
Entropy change: ∆S = 0.
Contents of this Document [ttc10]

10. Canonical Ensemble I

• Canonical ensemble. [tln51]

• Classical ideal gas (canonical ensemble). [tex76]
• Ultrarelativistic classical ideal gas (canonical idela gas). [tex77]
• Ultrarelativistic classical ideal gas in two dimensions. [tex154]
• Array of classical harmonic oscillators (canonical ensemble). [tex78]
• Irreversible decompression. [tex136]
• Irreversible heat exchange. [tex137]
• Reversible decompression. [tex139]
• Reversible heat exchange. [tex140]
• Heavy piston. [tex141]
• Ensemble averages. [tln52]
• Classical virial theorem. [tln83]
• Systems of noninteracting particles. [tln54]
• Further ensemble averages. [tln55]
• Classical ideal gas in a uniform gravitational field. [tex79]
• Gas pressure and density inside centrifuge. [tex135]
• Relative momentum of two ideal gas particles. [tex80]
• Partition function and density of states. [tln56]
• Ideal gas partition function and density of states. [tex81]
• Vibrational heat capacities of solids. [tln57]
• Array of quantum harmonic oscillators (canonical ensemble). [tex82]
• Vibrational heat capacities of solids (Debye theory). [tsl29]
• Thermodynamic perturbation expansion. [tln80]
• Vibrational heat capacity of a solid. [tex83]
• Anharmonic oscillator and thermodynamic perturbation. [tex104]
Canonical Ensemble [tln51]

Consider a closed classical system (volume V , N particles, temperature T ).

The goal is to determine the thermodynamic potential A(T, V, N ) pertaining
to that situation, from which all other thermodynamic properties can be
derived.
Z
Maximize Gibbs entropy S = −kB d6N X ρ(X) ln[CN ρ(X)]
Γ
subject to the constraints related to normalization and average energy:
Z Z
6N
d X ρ(X) = 1, d6N X H(X)ρ(X) = U.
Γ Γ

Apply calculus of variation with two Lagrange multipliers:

Z
δ d6N X{−kB ρ ln[CN ρ] + α0 ρ + αU Hρ} = 0
Γ
Z
⇒ d6N X δρ{−kB ln[CN ρ] − kB + α0 + αU H} = 0.
Γ
 
1 α0 αU
⇒ {· · · } = 0 ⇒ ρ(X) = exp −1+ H(X) .
CN kB kB

Determine the Lagrange multipliers α0 and αU :

Z   Z  
6N α0 1 6N αU
d X ρ(X) = 1 ⇒ exp 1 − = d X exp H(X) ≡ ZN .
Γ kB CN Γ kB
Z
d6N X ρ(X){· · · } = 0 ⇒ S − kB + α0 + αU U = 0.
Γ
1 kB 1
⇒ U+ S= ln ZN . Compare with U − T S = A ⇒ αU = − .
αU αU T
Helmholtz free energy: A(T, V, N ) = −kB T ln ZN .
Z
1 1
Canonical partition function: ZN = d6N X exp (−βH(X)) , β = .
CN Γ kB T
1
Probability density: ρ(X) = exp (−βH(X)) .
ZN CN

Canonical ensemble in quantum mechanics:

X 1 −βH
ZN = Tre−βH = e−βEλ , ρ= e , A = −kB T ln ZN .
λ
ZN
[tex76] Classical ideal gas (canonical ensemble)
Consider a classical ideal gas of N atoms confined to a box of volume V in thermal equilibrium
with a heat reservoir at temperature T . The Hamiltonian of the system reflects the kinetic energy
of 3N noninteracting degrees of freedom:
3N
X p2i
H= .
i=1
2m
p
(a) Show that the canonical partition function is ZN = V N /(N !λ3N T ), where λT = h2 /2πmkB T
is the thermal wavelength.
(b) Derive from ZN the Helmholtz free energy A(T, V, N ), the entropy S(T, V, N ), the pressure
p(T, V, N ), the internal energy U (T, N ), and the chemical potential µ(T, V ).
(c) Show that the pressure is equal to two thirds of the energy density and that the adiabates
satisfy p3 V 5 = const.

Solution:
[tex77] Ultrarelativistic classical ideal gas (canonical ensemble)
Consider a classical ideal gas of N atoms confined to a box of volume V in thermal equilibrium
with a heat reservoir at an extremely high temperature T . The Hamiltonian of the system,
N
X
H= |pl |c,
l=1

where c is the speed of light, reflects the ultrarelativistic energy of N noninteracting particles:
(a) Calculate the canonical partition function ZN of this system.
(b) Derive from ZN the Helmholtz free energy A(T, V, N ), the entropy S(T, V, N ), the pressure
p(T, V, N ), the internal energy U (T, N ), and the chemical potential µ(T, V ).
(c) Show that the pressure is equal to one third of the energy density and that the adiabates satisfy
p3 V 4 = const.

Solution:
[tex154] Ultrarelativistic classical ideal gas in two dimensions
Consider a classical ideal gas of N particles confined to a two-dimensional box of area V in thermal
equilibrium at extremely high temperature T . Most particles are moving at speeds close to the
speed of light c. We describe this system by a Hamiltonian of the form,
N q
X
H= p2x + p2y c.
l=1

(a) Show that the canonical partition function is

"  2 # N
1 kB T
ZN = 2πV .
N! hc

(b) Find the Helmholtz free energy A(T, V, N ), the entropy S(T, V, N ), the pressure p(T, V, N ),
and the internal energy U (T, N ).
(c) Find the adiabate (for constant N ) and express it in the form pν V = const.
(d) Infer from the given canonical partition function ZN (T, V ) an explicit expression for the grand
partition
R ∞ function Z(T, V, µ), where µ = kB T ln z is the
P∞chemical potential and z is the fugacity.
Use 0 dxxn e−ax = n!a−n−1 , ln n! ' n ln n − n, n
n=0 x /n! = e .
x

Solution:
[tex78] Array of classical harmonic oscillators (canonical ensemble)
Consider an array of N 3-dimensional classical harmonic oscillators, representing a system of 3N
uncoupled degrees of freedom:
3N  2 
X pi 1
H= + mω 2 qi2 .
i=1
2m 2

(a) Calculate the canonical partition function ZN for this model.

(b) Derive from ZN the Helmholtz free energy A(T, N ), the entropy S(T, N ), the internal energy
U (T, N ), and the heat capacity C ≡ (∂U/∂T )N .

Solution:
[tex136] Irreversible decompression
Consider an insulating box with two compartments. Each compartment initially contains N atoms
of a monatomic classical ideal gas in equilibrium at initial pressures p1 6= p2 and at the same initial
temperature T . Gas atoms are then allowed to leak through a hole in the dividing wall.
(a) Show that the temperature remains the same in the final equilibrium state.
(b) Find the uniform pressure p in the final equilibrium state as a function of p1 and p2 .
(c) Find the increase in total entropy, ∆S, between the initial and final equilibrium states.

Solution:
[tex137] Irreversible heat exchange
Consider an insulating box with two compartments. Each compartment initially contains N atoms
of a monatomic classical ideal gas in equilibrium at initial temperatures T1 6= T2 and at the same
initial pressure p. Gas atoms are then allowed to leak through a hole in the dividing wall.
(a) Find the uniform temperature T in the final equilibrium state as a function of T1 and T2 .
(b) Show that the pressure remains the same in the final equilibrium state.
(c) Find the increase in total entropy, ∆S, between the initial and final equilibrium states.

Solution:
[tex139] Reversible decompression
Consider a rigid, insulating box with two compartments of volumes V1 and V2 separated by an
internal wall. Each compartment contains N atoms of a monatomic classical ideal gas [pV =
N kB T, CV = 23 N kB ] in equilibrium at the same temperature Tini .
(a) Find the maximum work, ∆W (Tini , V1 , V2 , N ), that can be extracted from this system by any
means that keep the box rigid and insulating.
(b) Design a reversible process that employs the internal wall, which is movable by an external
agent in a controlled manner and which can be switched between heat-conducting and insulating
modes.

Solution:
[tex140] Reversible heat exchange
Consider a rigid, insulating box with two compartments of volumes V1 and V2 separated by an
internal wall. Each compartment contains N atoms of a monatomic classical ideal gas [pV =
N kB T, CV = 23 N kB ] in equilibrium at the same pressure.
(a) Find the maximum work, ∆W (T1 , T2 , N ), that can be extracted from this system by any means
that keep the box rigid and insulating.
(b) Design a reversible process that employs the internal wall, which is movable by an external
agent in a controlled manner and which can be switched between heat-conducting and insulating
modes.

Solution:
[tex141] Heavy piston
A cylinder of cross section A with insulating walls has two compartments separated by a disk of
mass m. The axis of the cylinder is vertical. A uniform gravitational field g is present. The disk is
initially held at a fixed position by an external agent. The upper compartment is evacuated and
the lower compartment contains 1 mol of a monatomic, classical, ideal gas [pV = RT , CV = 32 R]
at temperature T0 , volume V0 , and pressure p0 . When the disk is released, it moves without (wall)
friction and comes to rest at a lower position. Calculate the final values p1 , V1 , T1 of pressure,
volume, and temperature, respectively. The disk does not exchange heat. The only significant
action of the gravitational field is on the disk.

Hint: Use energy conservation and Newton’s third law. Assume thermal equilibrium for the inital
and final states.

p=0

p V0 g
0
T0

Solution:
Ensemble averages [tln52]

All thermodynamic quantities of a closed system can be inferred from the

canonical partition function ZN via the associated thermodynamic potential:
Z
1 1
ZN = d6N X exp (−βH(X)) , β = .
CN Γ kB T
Further properties of the system can be obtained from the canonical proba-
bility density ρ(X) via equilibrium expectation values of arbitrary dynamical
variables f (X):
Z
1
hf i = d6N X ρ(X)f (X), ρ(X) = exp (−βH(X)) .
Γ ZN CN
From such expectation values, we can recover thermodynamic quantities and
calculate fluctuations thereof, which are related to response functions, i.e.
different thermodynamic quantities. Other expectation values, e.g. correla-
tion functions, cannot be inferred directly from ZN .

• Uncertainty about microstate and entropy:

Z
S = −kB d6N X ρ(X) ln[CN ρ(X)].
Γ

• Average value H and internal energy:

Z Z
1
6N
hHi = d X ρ(X)H(X) = d6N X H(X) e−βH(X)
Γ Z C
N N Γ

1 ∂ZN ∂ ∂
⇒ hHi = − =− ln ZN = (βA)
ZN ∂β ∂β ∂β
 
∂ ∂T ∂ ∂ ∂A
Use = = −kB T 2 ⇒ hHi = A − T = A + T S = U.
∂β ∂β ∂T ∂T ∂T

• Energy fluctuations and heat capacity:

2
1 ∂ 2 ZN
  
2 2 1 ∂ZN ∂ 1 ∂ZN
hH i − hHi = − =
ZN ∂β 2 ZN ∂β ∂β ZN ∂β

∂2 ∂U ∂U
⇒ hH 2 i − hHi2 = 2
ln ZN = − = kB T 2 = kB T 2 CV .
∂β ∂β ∂T
Classical virial theorem [tln83]

Classical Hamiltonian system: H = T + V.

N interacting particles in 3D space represent 3N degrees of freedom.
Phase-space coordinates: {xi } = {(ql , pl )}, i = 1, . . . , 6N, l = 1, . . . , 3N .

Theorem in general form:

∂e−H/kB T
  Z Z
∂H . 1 6N ∂H −H/kB T kB T
xi = d x xi e =− d6N x xi .
∂xj Z ∂xj Z ∂xj

Integrate by parts:
  Z
∂H kB T
⇒ xi = d6N xe−H/kB T δij = kB T δij .
∂xj Z

Equipartition: average kinetic energy per degree of freedom

3N
p2l
   
X ∂H 1 2 1
T = ⇒ pl = hmp2l i ⇒ mp = kB T.
l=1
2m ∂pl 2 l 2

Virial: pair interactions

1X .
V= v(|ql − ql0 |). Set qll0 = ql − ql0 .
2 l6=l0
 
1X ∂v
⇒ qll0 = N kB T − pV.
6 l6=l0 ∂qll0

Anharmonic crystal in 1D: average potential energy per bond

N −1
X 1
V= u|ql − ql+1 |ν with ν > 0. Set p = 0.
l=1
2
 
1 kB T
⇒ u|ql − ql+1 |ν = .
2 ν

[adapted from Schwabl 2006]

Systems of noninteracting particles [tln54]

Consider a classical systems of N noninteracting particles.

N
X
Hamiltonian: H = hl (ql , pl ).
l=1

Canonical partition function of distinguishable particles:

Z N Z
1 −βH(X)
Y 1
ZN = d6N
Xe = Z̃l , Z̃l = 3 d3 ql d3 pl e−βhl (ql ,pl ) .
CN Γ l=1
h

Factorizing phase-space probability density:

N  
1 −βH(X)
Y 1 −βhl (ql ,pl )
ρ(X) = e = e .
ZN CN l=1
h3 Z̃
l

Identical one-particle Hamiltonians:

Z
1
h1 = · · · = hn ≡ h(q, p) ⇒ Z̃1 = · · · = Z˜N = Z̃ ≡ 3 d3 q d3 p e−βh(q,p) .
h
N  
N
Y 1 −βh(ql ,pl )
⇒ ZN = Z̃ , ρ(X) = e .
l=1
h3 Z̃

Indistinguishable particles:
1 N
ZN = Z̃ .
N!

Note: It is important that we discriminate between noninteracting subsys-

tems that are identical but distinguishable (e.g. atoms vibrating about rigid
lattice sites) and noninteracting subsystems that are identical and indistin-
guishable (e.g. atoms of an ideal gas).
Further ensemble averages [tln55]

Probability density in one-particle phase space:

ρl (q, p) = hδ(ql − q)δ(pl − p)i.

Position distribution and momentum distribution:

ρl (q) = hδ(ql − q)i, ρl (p) = hδ(pl − p)i.

Distribution of distances and relative momenta between pairs of particles:

flm (r) = hδ(r − |ql − qm |)i, Flm (P ) = hδ(P − |pl − pm |)i.

Average distance between pairs of particles:

Z ∞
hrlm i = h|ql − qm |i = dr rflm (r).
0

Average magnitude of relative momentum between pairs of particles:

Z ∞
hPlm i = h|pl − pm |i = dP P Flm (P ).
0

Applications to the classical ideal gas:

s
2
p V h2
Noninteracting particles: h(q, p) = ⇒ Z̃ = 3 , λT = .
2m λT 2πmkB T
2
⇒ ρl (q, p) = V −1 (2πmkB T )−3/2 e−p /2mkB T .
Z
1
⇒ ρl (q) = d3 p ρl (q, p) = .
V
Z
2
⇒ ρl (p) = d3 q ρl (q, p) = (2πmkB T )−3/2 e−p /2mkB T .
 −3/2
m 2 /2k
3 3
ρl (p)d p = f (v)d v ⇒ f (v) = e−mv BT
.
2πkB T

The spatial distribution of ideal gas particles in a uniform gravitational field

(law of atmospheres) is calculated in exercise [tex79].
The distribution of relative momenta between pairs of ideal gas particles is
calculated in exercise [tex80].
[tex79] Classical ideal gas in uniform gravitational field
Consider a column with cross-sectional area A of a classical ideal gas (N atoms of mass m) in a
uniform gravitational field of magnitude g. The gas is in thermal equilibrium at temperature T .
The Hamiltonian reads:
N  2 
X pl
H= + mgzl ,
2m
l=1

where zl is the height of particle l above sea level.

(a) Find the probability density ρ1 (z) for the vertical positions of individual gas atoms.
(b) Find the pressure distribution p(z).

Solution:
[tex135] Gas pressure and density inside centrifuge
Consider a hollow disk of width L and radius R filled with N particles of a dilute gas at temperature
T . The disk is in a state of rotation with angular velocity ω about its axis.
(a) Find the probability density ρ1 (r) for the radial position of a gas particle and find the particle
density n(r). Note that the unit of ρ1 (r) is [m−2 ] and the unit of n(r) is [m−3 ].
(b) Find the pressure p(r).
(c) In an experiment that measures p(0) and p(R) at various values of ω and fixed T , which two
quantities must be plotted against each other such that the data points are predicted to fall onto
a straight line with slope equal to the mass of the gas particles?

Solution:
[tex80] Relative momentum of two ideal gas particles
Consider a classical ideal gas of N atoms with mass m confined to a box of volume V in thermal
equilibrium with a heat reservoir at temperature T .
(a) Find the distribution Flm (P ) ≡ hδ(P − |pl − pm |)i of the magnitude of the relative momenta
of two ideal gas particles.
(b) Find the ratio of the average magnitudes hP i/hpi of the relative momentum of two particles
and the momentum of a single particle.

Solution:
Partition function and density of states [tln56]

Why do the microcanonical and canonical ensembles yield the same results?

(a) Derivation of ZN from Ω(U, V, N ).

Relation between the microcanonical phase-space volume Ω(U, V, N ) and the
number of microstates Σ(U, V, N ) up to the energy U :
Z
Ω(U, V, N ) ≡ d6N X = CN Σ(U, V, N ).
H(X)<U

Density of microstates:
∂Σ
g(U ) = .
∂U

The canonical partition function is then obtained via Laplace transform:

Z ∞ Z
−βU 1
dU g(U )e = d6N X e−βH(X) = ZN .
0 C N Γ

Here the energy scale has been shifted such that U0 = 0.

(b) Derivation of Ω(U, V, N ) from ZN .

Complex continuation of the canonical partition function:

ZN = Z(β) for β = β 0 + iβ 00 with β 0 > 0.

The microcanonical phase-space volume is the obtained via inverse Laplace

transform:
Z β 0 +i∞ Z U
1
g(U ) = βU
dβ e Z(β), Ω(U, V, N ) = CN dU 0 g(U 0 ).
2πi β 0 −i∞ 0

Both calculations are carried out in exercise [tex81] for the classical ideal gas.
[tex81] Ideal gas partition function and density of states
(a) Starting from the result of [tex73] for the phase-space volume Ω(U, V, N ) of a classical ideal gas
(N particles with mass m) in the microcanonical ensemble, calculate the density of microstates,
gN (u), and then, via Laplace transform, the result of [tex76] for the canonical partition function
ZN (β), where β = 1/kB T .
(b) Starting from the canonical partition function ZN (β) analytically continued into the complex
plane, calculate the density of state gN (U ) via inverse Laplace transform.

Solution:
Vibrational heat capacities of solids [tln57]

The interaction between atoms is attractive at long distances and repulsive

at short distances. The lowest-energy configuration of a macroscopic system
of N atoms is a perfect lattice. This is the equilibrium state at T = 0. It
has zero entropy. Heat input δQ = CdT causes lattice vibrations. In the
following we study vibrational heat capacities in successively improved ap-
proximations.

Atoms bound to rigid lattice by harmonic force (classical model):

The theory of Dulong and Petit considers an array of N classical 3D harmonic
oscillators with identical angular frequencies. The resulting vibrational heat
capacity, C = 3N kB , is T -independent and is calculated in exercise [tex74] for
a microcanonical ensemble and in exercise [tex78] for a canonical ensemble.
The main insufficiency of the Dulong-Petit result is that C does not approach
zero in the low-temperature limit, in violation of the third law.

Atoms bound to rigid lattice by harmonic force (quantum model):

The theory of Einstein considers an array of N quantum 3D harmonic os-
cillators with identical angular frequencies ω. The resulting vibrational heat
capacity, 2
3N kB eΘE /T

ΘE
C= 2, kB ΘE = ~ω,
kB T (eΘE /T − 1)
goes to zero exponentially in the low-T limit, C ∼ e−ΘE /T , and approaches
the Dulong-Petit result, C = 3N kB , at high T . Einstein’s result is derived
in exercise [tex75] for the microcanonical ensemble and in exercise [tex82] for
the canonical ensemble.
The main insufficiency of Einstein’s result is that it contradicts experimental
evidence, which suggests C ∼ T 3 at low T .

1
Atoms interacting via harmonic force:

3N 3N  2
p2i

X X X pi 1 2 2
H= + Aij qi qj = + mωi Qi .
i=1
2m ij i=1
2m 2

Here {Aij } is the dynamical matrix. The second equation results from a
transformation to normal-mode coordinates. In the present context the nor-
mal modes are sound waves (phonons).
Quantum mechanically, this system is an array of 3N independent harmonic
oscillators with normal mode frequencies ωi :
3N  
X 1
H= ~ωi ni + , ni = 0, 1, 2, . . .
i=1
2

The resulting Helmholtz free energy (in generalization to the result derived
in [tex82]) reads:
3N 3N
1X X
ln 1 − e−β~ωi .


A= ~ωi + kB T
2 i=1 i=1

In Debye’s theory, the normal modes, which, in general, consist of multiple

branches of acoustic and optical phonons, are replaced by a single branch of
sound waves with linear dispersion ω = ck as is expected in a continuous
isotropic elastic medium.
Total number of modes: 3N (same as in original lattice model).
Density of modes in k-space: V /(2π)3 .
Number of polarizations: 3 (2 transverse, 1 longitudinal).
V ω 2 dω 3V
Number of modes in dω: n(ω)dω = 3
(3)(4π) 2
= 2 3 ω 2 dω.
8π c c 2π c
ωD
6N π 2 c3
Z
3V 2 3
Debye frequency: dω ω = 3N ⇒ ωD = .
2π 2 c3 0 V
9N 2
Density of modes: n(ω) = 3
ω .
ωD
The resulting vibrational heat capacity is calculated in exercise [tex83] and
does show the experimentally observed ∼ T 3 behavior as T → 0:
3 Z ΘD /T
x4 ex

T
C = 9N kB dx , ΘD = ~ωD /kB .
ΘD 0 (ex − 1)2

2
[tex82] Array of quantum harmonic oscillators (canonical ensemble)
Consider an array of N 3-dimensional quantum harmonic oscillators:
3N   
X 1
H= ~ω ni + , ni = 0, 1, 2, . . . .
i=1
2

(a) Calculate the canonical partition function ZN for this model.

(b) Derive from ZN the Helmholtz free energy A(T, N ), the internal energy U (T, N ), and the heat
capacity C = (∂U/∂T )N .
(c) Show that U (T, N ) approaches the result of [tex78] for the classical oscillators.
(d) Calculate the quantity hni i for a single degree of freedom. It reflects the average number of
elementary energy quanta that are excited in one oscillator when it is in thermal equilibrium at
temperature T .

Solution:
Vibrational heat capacities of solids [tsl29]

Density of vibrational modes in aluminum

Debye predicition for heat capacity

in comparison with experimental data

[from Garrod 1995]

Thermodynamic perturbation expansion [tln80]

Consider a classical dynamical system in the canonical ensemble.

H = H0 + V.

The term H0 represents the dominant contribution to the energy of the sys-
tem under the circumstances of interest. We assume that the Helmholtz free
energy, A0 (T, V, N ), for that part alone can be calculated exactly:
Z
−βA0 . . 1
e = dΓ e−βH0 , dΓ = d6N X, β= .
kB T
We can then treat V perturbatively via the following expansion:
Z Z  
−βA −β(H0 +V ) −βH0 1 2 2
e = dΓ e ' dΓ e 1 − βV + β V .
2

This expression is then further expanded by using ln(1 − x) ' −x + x2 /2:

 Z Z 
−βA0 −βA0 1 2 2 −βH0
−βA ' ln e − β dΓ V e + β dΓ V e
2
 Z Z 
β(A0 −H0 ) 1 2 2 β(A0 −H0 )
' −βA0 + ln 1 − β dΓ V e + β dΓ V e .
2
Z   Z 2
1 2 β(A0 −H0 ) 1 β(A0 −H0 )
⇒ A = A0 + dΓ V − βV e + β dΓ V e .
2 2

With ensemble averages,

. RdΓ V e−βH0
R Z Z
hV i = −βH
= dΓ V eβ(A0 −H0 ) , 2
hV i = dΓ V 2 eβ(A0 −H0 ) ,
dΓ e 0

and the relation

hV 2 i − hV i2 = h(V − hV i)2 i,
we can write
1
A = A0 + hV i − βh(V − hV i)2 i.
2

The criterion of applicability for this expansion is hV i/N  kB T . Note that

if hV i = 0 then the leading-order perturbation always reduces the Helmholtz
free energy.
[tex83] Vibrational heat capacity of a solid
The vibrational Helmholtz free energy of a harmonic crystal of N atoms in thermal equilibrium at
temperature T is
3N 3N
1X X
ln 1 − e−β~ωi ,


A= ~ωi + kB T
2 i=1 i=1

where the ωi are the normal modes of transverse and longitudinal lattice vibrations (phonons). In
Debye’s theory, the density of modes is approximated by the functions n(ω) = 9N ω 2 /ωD 3
, where
the Debye frequency ωD is an undetermined parameter.
(a) Show that the internal energy U = A + T S in the Debye approximation reads
3 Z ΘD /T
x3

9 T
U = N ~ωD + 9N kB T dx ,
8 ΘD 0 ex − 1

where ΘD = ~ωD /kB is called the Debye temperature.

(b) Derive an expression for the heat capacity C = (∂U/∂T )N .
(c) At low
R ∞ temperatures the upper boundary ΘD /T in the above integral may be replaced by
infinity, 0 dx x3 /(ex − 1) = π 4 /15. Use this fact to determine the leading low-temperature term
of the heat capacity C.

Solution:
[tex104] Anharmonic oscillator and thermodynamic perturbation
Consider an array of N one-dimensional anharmonic oscillators,
N  2 
X pl
H= + V (ql ) , V (q) = cq 2 − gq 3 + f q 4 .
2m
l=1

(a) Evaluate the canonical partition function perturbatively by treating the quadratic term of V
exactly and considering only the leading nonzero corrections of the cubic and the quartic terms.
(b) Show that the heat capacity in this approximation is C/N kB = 1 + [15g 2 /16c3 − 3f /4c2 ]kB T .
(c) Show that the mean displacement in this approximation is hqi = (3g/4c2 )kB T .

Solution:
Contents of this Document [ttc11]

11. Canonical Ensemble II

• Paramagnetism. [tln58]
• Paramagnetic salts. [tsl30]
• Fluctuations in a magnetic system. [tln53]
• Fluctuations in a magnetic system. [tex109]
• Classical paramagnet (canonical ensemble). [tex84]
• Quantum paramagnet (two-level system). [tex85]
• Quantum paramagnet (Brillouin function). [tex86]
• Ising trimer. [tex142]
• Negative temperatures. [tsl31]
• Gases with internal degrees of freedom. [tln59]
• Classical rotational free energy of NH3 gas. [tex87]
• Classical rotational entropy of HCl and N2 gas. [tex88]
• Quantum rotational heat capacity of a gas at low temperature. [tex89]
• Quantum rotational heat capacity of a gas at high temperature. [tex90]
• Rotational and vibrational heat capacities. [tsl32]
• Orthohydrogen and parahydrogen. [tln81]
• Relativistic classical ideal gas (canonical partition function). [tex91]
• Relativistic classical ideal gas (entropy and internal energy). [tex92]
• Relativistic classical ideal gas (heat capacity). [tex93]
• Relativistic classical ideal gas (heat capacity). [tsl34]
Paramagnetism [tln58]

Paramagnetic salts contain localized ions with permanent magnetic dipole

moments associated with unpaired electron spins. The interaction between
the electron spins is negligibly small and there is no kinetic energy associated
with their orientational motion. Hence the internal energy vanishes: U = 0.
The microstate is specified by the instantaneous orientation of the magnetic
moments mi , i = 1, . . . , N relative to some coordinate system. The localized
moments can be treated as distinguishable particles. They do not need to
have a definite permutation symmetry. The macroscopic equilibrium state
in the canonical ensemble is characterized by P random orientations of the
moments mi . It has no magnetisation: M = i hmi i = 0.
An external magnetic field H causes a partial spin alignment. The interac-
tion of the magnetic moments with a field in z-direction is represented by
Hamiltonian (Zeeman energy) of the form:
N
X N
X
H=− mi · H = −H mzi .
i=1 i=1

Classical model: The permanent atomic magnetic moment is described as

a 3-component vector of fixed length:
mi = (mxi , myi , mzi ) = m(sin θi cos φi , sin θi sin φi , cos θi ).
Each mi represents one degree of freedom described by one pair of canon-
ical coordinates qi = φi , pi = m cos θi . The canonical partition function is
calculated in exercise [tex84].

Quantum model (spin 1/2): The permanent atomic magnetic moment

originates from a single electron spin. This is a two-level system, which
also has a host of realizations unrelated to paramagnetism. The magnetic
moment in appropriate units is quantized as follows:
1
mzi = ± .
2
The canonical partition function is calculated in exercise [tex85].

Quantum model (spin s): The permanent atomic magnetic moment origi-
nates from an effective spin of quantum number s = 12 , 1, 23 , . . . The magnetic
moment in appropriate units is quantized as follows:
mzi = −s, −s + 1, . . . , s − 1, s.
The canonical partition function is calculated in exercise [tex86].
Paramagnetic salts [tsl30]

Magnetization curves of paramagnetic salts

in comparison with Brillouin functions

[from Crangle 1977]

Fluctuations in a magnetic system [tln53]

Consider a system of N interacting magnetic moments mi positioned in an

external magnetic field of magnitude H.
N
X
Total magnetic moment: M = mi .
i=1

Hamiltonian: H = Hint − HM.

Canonical partition function: ZN = Tr e−βH , β = (kB T )−1 .

Gibbs free energy: G(T, H, N ) = −kB T ln ZN .

Magnetisation (average value of total magnetic moment):

1 ∂ .
Tr M e−βH = β −1
 
hM i = ln ZN = M.
ZN ∂H

Enthalpy (average value of Hamiltonian):

1 ∂
Tr H e−βH = −
 
hHi = ln ZN = U − HM = E
ZN ∂β

Energy fluctuations and heat capacity [tex109]:

∂2
hH2 i − hHi2 = ln ZN = kB T 2 CH .
∂β 2

Magnetisation fluctuations and susceptibility [tex109]:

∂2
hM 2 i − hM i2 = β −2 ln ZN = kB T χT .
∂H 2
[tex109] Fluctuations in a magnetic system
Consider a quantum magnet. The Hamiltonian is of the form H = Hint −hm, where Hint describes
the (unspecified) interaction between microscopic magnetic moments, h is the magnitude of the
external magnetic field (assumed constant) and m is the component of the total magnetic moment
in the direction of the field. Given the Gibbs free energy G(T, h, N ) = −kB T ln ZN as derived from
the canonical partition function ZN = Tr e−βH , where β = (kB T )−1 , derive the following relations
(a) between energy fluctuations and heat capacity at constant field,

∂2
hH 2 i − hHi2 = ln ZN = kB T 2 Ch ,
∂β 2

and (b) between magnetisation fluctuations and isothermal susceptibility,

∂2
hm2 i − hmi2 = β −2 ln ZN = kB T χT .
∂h2

Solution:
[tex84] Classical paramagnet (canonical ensemble)
Consider an array of N noninteracting localized magnetic dipole moments in the form of classical
3-component unit vectors mi = (mxi , myi , mzi ) = (sin θi cos φi , sin θi sin φi , cos θi ). In the presence of
a magnetic field H pointing in z-direction, the Hamiltonian of this system represents the Zeeman
energy:
X N XN
H=− mi · H = −H mzi .
i=1 i=1

(a) Calculate the canonical partition function ZN of this system.

(b) Calculate the Gibbs free energy G(T, H, N ), the magnetization M (T, H, N ) (Langevin func-
tion), the isothermal susceptibility χT (T, H, N ), and the heat capacity CH (T, H, N ).
(c) Plot M/N versus H for three values of T . Plot CH /N versus T for three values of H.
(d) Show that the leading term in an expansion of χT at small H is H-independent and represents
Curie’s law χT ' N/3kB T .

Solution:
[tex85] Quantum paramagnet (two−level system)
Consider an array of N noninteracting localized magnetic dipole moments mi produced by localized
electron spins in a paramagnetic insulator. In the presence of a magnetic field H pointing in z-
direction, the Hamiltonian of this system represents the Zeeman energy:
N N
X X 1
H=− mi · H = −H mzi , mzi = ± .
i=1 i=1
2

(a) Calculate the canonical partition function ZN of this system.

(b) Calculate the Gibbs free energy G(T, H, N ), the magnetization M (T, H, N ), the isothermal
susceptibility χT (T, H, N ), and the heat capacity CH (T, H, N ).
(c) Show that the internal energy U is identically zero.
(d) Show that the leading term in an expansion of χT at small H is H-independent and represents
Curie’s law χT ' N/4kB T .

Solution:
[tex157] Quantum paramagnet (three-level system)
Consider an array of N noninteracting localized magnetic dipole moments mi . Each magnetic
moment may be produced, for example, by two unpaired electron spins localized in a paramagnetic
insulator. In the presence of a magnetic field H pointing in z-direction, the Hamiltonian of this
system represents the Zeeman energy:
N
X N
X
H=− mi · H = −H mzi , mzi = 1, 0, −1.
i=1 i=1

(a) Calculate the canonical partition function ZN of this system.

(b) Derive the Gibbs free energy G(T, H, N ), the magnetization M (T, H, N ), and the entropy
S(T, H, N ).
(c) Show that the internal energy U vanishes identically.
(d) Calculate the isothermal susceptibility χT at H = 0 as a function of T .
(e) Sketch the shape of adiabates in the (H, M )-plane and in the (H, T )-plane.

Solution:
[tex86] Quantum paramagnet (Brillouin function)
Consider an array of N noninteracting localized magnetic dipole moments mi produced by localized
effective atomic spins in a paramagnetic insulator. In the presence of a magnetic field H pointing
in z-direction, the Hamiltonian of this system represents the Zeeman energy:
N
X N
X
H=− mi · H = −H mzi ,
i=1 i=1

where mzi can assume the 2s + 1 values (−s, −s + 1, . . . , s − 1, s) for fixed s = 21 , 1, 32 , . . ..

(a) Calculate the canonical partition function ZN of this system.
(b) Calculate the Gibbs free energy G(T, H, N ). Calculate the magnetization M (T, H, N ) (Bril-
louin function).
(c) Set s = 12 to recover the result of [tex85]. Take the limit s → ∞ and recover the result of
[tex84] for the rescaled quantities M̃ = M/s, H̃ = Hs.

Solution:
[tex142] Ising trimer
Three spins at the corners of an equilateral triangle interact with each other and with a magnetic
field. The Hamiltonian is of the form

H = −J(s1 s2 + s2 s3 + s3 s1 ) − H(s1 + s2 + s3 ),

where sn = ±1, n = 1, 2, 3, and J, H are energy units representing the interaction and the
magnetic field, respectively.
(a) Calculate the canonicl partition function Z and infer from it the Gibbs free energy G(T, H).
(b) Write detailed instructions for the derivation, from Z or G, of the magnetization M , the entropy
S, and the internal energy U .

Solution:
Negative temperatures [tsl31]

Consider N noninteracting 2-level systems with energies ±.

   
1 U 1 U
N = N+ +N− , U = (N+ −N− ) ⇒ N+ = N+ , N− = N− .
2  2 

N!
Degeneracy of state with energy U : NU (U, N ) = .
N+ !N− !
Entropy:
   
1 U U
S(U, N ) = kB ln NU (U, N ) = kB N ln N − kB N + ln N +
2  
   
1 U U
− kB N − ln N − .
2  

Inverse temperature:
   
1 ∂S kB N − U/
= = ln .
T ∂U N 2 N + U/

Inversion of level occupancy corresponds to negative temperature.

[from Greiner et al. 1995]

Applications: laser pumping to metastable states, nuclear magnetism.

Gases with internal degrees of freedom [tln59]

Assumptions: molecules are noninteracting; translational, rotational, and

vibrational degrees of freedom are independent:
N h
X (i) (i) (i)
i 1 N
H= HT + HR + HV ⇒ ZN = Z̃ , Z̃ = Z̃T Z̃R Z̃V .
i=1
N!

Translational motion (classical):

3/2
p2i

(i) 2πmkB T (T ) 3
HT = ⇒ Z̃T = V ⇒ C V = N kB [tex76].
2m h2 2

Rotational motion (classical):

(a) NH3 (multi-atomic molecule):
Euler angles θ, φ, ψ; canonical conjugate momenta pθ , pφ , pψ .
Uniaxially symmetric inertia tensor with principal moments I1 = I2 , I3 .

(i) p2iθ p2iψ (piφ − piψ cos θi )2

Hamiltonian: HR = + + .
2I1 2I3 2I1 sin2 θi
Ranges: 0 ≤ θi ≤ π, 0 ≤ φi , ψi ≤ 2π, − ∞ < piθ , piφ , piψ < +∞.
1 p (R) 3
⇒ Z̃R = 3
(2πI1 kB T )2 (2πI3 kB T ) ⇒ CV = N kB [tex87].
π~ 2

(b) HCl (two-atomic heteronuclear molecule):

The rotation about the molecular axis is suppressed due to quantum effect.

(i) p2iθ p2iφ

Hamiltonian: HR = + , 0 ≤ θi ≤ π, 0 ≤ φi ≤ 2π.
2I1 2I1 sin2 θi
2I1 kB T (R)
⇒ Z̃R = 2
⇒ CV = N k B [tex88].
~

(c) N2 (two-atomic homonuclear molecule):

Minor modification: range of one variable (0 ≤ φ ≤ π).
This change does affect the entropy but not the heat capacity ⇒ [tex88].

1
Rotational motion (quantum):
Consider a two-atomic molecule.
1 2
Angular momentum operator: L. Hamiltonian operator: HR = L.
2I
l(l + 1)~2
Energy levels: Elm = ; l = 0, 1, 2, . . . ; m = −l, −l + 1, . . . , l.
2I
Degeneracy: (2l + 1)-fold.
∞ X
+l ∞
2 /2I
X X
−βElm
⇒ Z̃R = e = (2l + 1)e−βl(l+1)~ ⇒ [tsl32]
l=0 m=−l l=0

~2
Characteristic temperature: kB ΘR = .
2I
Low-temperature analysis (T  ΘR ) ⇒ [tex89].
High-temperature analysis (T  ΘR ) ⇒ [tex90].

Vibrational motion (quantum):

f  2 
X pl 1 2 2
Hamiltonian: HV = + ml ωl ql .
l=1
2ml 2

Here f is the number of vibrational normal modes, each expressed by a pair

(ql , pl ) of canonical normal mode coordinates.
f 
e−β~ωl

(V ) T Θ
Y
⇒ Z̃V = −β~ω
⇒ CV −→V f N kB [tex82].
l=1
1−e l

Characteristic temperature: kB ΘV = ~ωl .

Vibrational modes require much higher temperatures to be activated:

~2 ~ωl
ΘR = ∼ 10K, ΘV = ∼ 1000K ⇒ [tsl32].
2IkB kB

2
Fine structure:
If the atomic ground state has zero orbital angular momentum (l = 0) and
nonzero spin angular momentum (s 6= 0), the entropy acquires an additive
constant, ∆S = N kB ln(2s + 1). The heat capacity remains unaffected.
In the presence of an external magnetic field, this system is a paramagnetic
gas. The thermodynamics of dilute paramagnetic gases are the theme of
[tex22] and [tex133].

If the atomic ground state has l 6= 0 and s 6= 0, then the L-S coupling
produces a fine-structure splitting of the ground-state degeneracy:
X
Z̃F S = (2j + 1)e−βj , |l − s| ≤ j ≤ l + s,
j

where j is the quantum number of the total angular momentum. If the lowest
level has j = j0 , then the entropy of the atomic gas increases by

(2s + 1)(2l + 1)
∆S = N kB ln
(2j0 + 1)

over a temperature range 0 < kB T . ∆ELS , where ∆ELS measures the total
L-S level splitting.
(F S)
The contribution to the heat capacity, CV , is a function of T that rises
from zero exponentially, exhibits a smooth maximum at kB T ∼ ∆ELS , and
then dips back down to zero algebraically.1
(F S)
The functional dependence of CV on T is very similar to that of the heat
capacity of a Langevin paramagnet as analyzed in [tex85] and [tex86].

1
In practical reality, atomic gases with l 6= 0, s 6= 0 tend to form molecules or condense at
temperatures far above kB T ∼ ∆ELS .

3
[tex87] Classical rotational free energy of NH3 gas
Under the assumption that the NH3 molecule is a rigid body with uniaxially symmetric inertia
tensor and principal moments I1 = I2 6= I3 , the one-particle Hamiltonian of the free rotational
motion reads
p2 p2ψ (pφ − pψ cos θ)2
HR = θ + + ,
2I1 2I3 2I1 sin2 θ
where (θ, pθ ; φ, pφ ; ψ, pψ ) are the Euler angles and their conjugate generalized momenta. The range
of these canonical coordinates is 0 ≤ θ ≤ π, 0 ≤ φ, ψ ≤ 2π, −∞ < pθ , pφ , pψ < +∞.
(a) Show that the canonical partition function for the rotational motion of N molecules is
N/2
N
ZR = π −N (2πkB T /~2 )3N/2 I1N I3 .

(b) Calculate the rotational Helmholtz free energy AR (T, N ), the rotational entropy SR (T, N ), and
the rotational internal energy UR (T, N ).

Solution:
[tex88] Classical rotational entropies of HCl and N2 gases
Under the assumption (to be justified quantum mechanically) that only the rotational modes
perpendicular to the axis of a two-atomic molecule are activated, its classical rotational motion is
described by a Hamiltonian with two degrees of freedom,

p2θ p2φ
HR = + ,
2I1 2I1 sin2 θ
where 0 ≤ θ ≤ π, −∞ < pθ , pφ < +∞. The range of the other angle is 0 ≤ φ ≤ 2π for heteronuclear
molecules (e.g. HCl) and 0 ≤ φ ≤ π for homonuclear molecules (e.g. N2 ).
N
(a) Calculate the canonical partition function ZR for the rotational motion of a gas of N HCl
molecules and a gas of N N2 molecules.
(b) Calculate the difference in rotational entropy and in rotational internal energy of the two gases
when both are at the same temperature.

Solution:
[tex89] Quantum rotational heat capacity of a gas at low temperature
The rotational spectrum of two-atomic molecules consists of energy levels

l(l + 1)~2
Elm = ; l = 0, 1, 2, . . . ; m = −l, −l + 1, . . . , l.
2I
Show that the leading term of the rotational heat capacity of a gas of N molecules at low temper-
ature (T  Θ) has the form
2
~2

Θ
CR (T ) ' 12N kB e−2Θ/T , Θ = .
T 2IkB

Solution:
[tex90] Quantum rotational heat capacity of a gas at high temperature
The rotational spectrum of two-atomic molecules consists of energy levels

l(l + 1)~2
Elm = ; l = 0, 1, 2, . . . ; m = −l, −l + 1, . . . , +l.
2I
(a) Use the Euler-McLaurin summation formula
∞ Z ∞
X 1 1 1 000
f (n) = dx f (x) + f (0) − f 0 (0) + f (0) + . . .
n=0 0 2 12 720

to calculate the first three terms of a high-temperature expansion of ZR = lm e−βElm .

P
(b) Use the result of (a) to show that the first two terms in a high-temperature expansion of the
rotational heat capacity read
"  2 #
1 Θ ~2
CR ' N kB 1 + + ... , Θ= .
45 T 2IkB

Solution:
Rotational and vibrational heat capacities [tsl32]

Rotational heat capacity of two-atomic gas:

 2
Θ ~2
T Θ: C ' 12N kB e−2Θ/T ; Θ ≡
T 2IkB
"  2 #
1 Θ
T Θ: C ' N kB 1 + + ...
45 T

Rotational and vibrational heat capacities of hydrogen molecules:

1
H: H (hydrogen)
2
D: H (deuterium)
3
T: H (tritium)

[from Greiner et al. 1995]

Orthohydrogen and parahydrogen [tln81]

Molecular hydrogen (H2 ) comes in two (nuclear) spin isomers.

The electronic and nuclear wave functions must both be antisymmetric. The
electronic ground state has a symmetric space part and an antisymmetric
spin part (spin singlet). Electronic excited states have much higher energies.
Nuclear wave functions with symmetric spin part and antisymmetric space
part or vice versa are energetically close to each other. The space part is
described by rotational modes with orbital quantum numbers l = 0, 1, 2, . . .
Vibrational modes have much higher energies.

• Orthohydrogen: Nuclear spin part is symmetric (spin triplet) and

nuclear space part is antisymmetric (odd l).
• Parahydrogen: Nuclear spin part is antisymmetric (spin singlet) and
nuclear space part is symmetric (even l).

At high T , the H2 gas at equilibrium contains 75% orthohydrogen and 25%

parahydrogen. The 3:1 ratio is a reflection of the nuclear spin degeneracy.
The lowest parahydrogen level is lower than the lowest orthohydrogen level
by ∆E/kB = ~2 /kB I ' 175K. Conversion is slow in the absence of catalysts.
Cooling and condensing hydrogen may leave the majority of molecules in a
metastable state. The transition to equilibrium at low T releases significant
amounts of energy.
Rotational factors of canonical partition function under two conditions:

• {eq} Slow temperature variation with catalysts present.

• {3:1} More rapid temperature reduction with catalysts absent.

N
3N/4
N/4
Z̃Req = Z̃o + Z̃p , Z̃R3:1 = Z̃o Z̃p .

2 /2I 2 /2I
X X
Z̃o = 3 (2l + 1)e−βl(l+1)~ , Z̃p = (2l + 1)e−βl(l+1)~ ,
odd l even l
 [from Wikipedia]
2
[tex91] Relativistic ideal gas (canonical partition function)
Consider a classical ideal gas of N atoms confined to a box of volume V in thermal equilibrium
with a heat reservoir at a very high temperature T . The Hamiltonian of the system,
N q
X 
2 4 2 2 2
H= m c + pl c − mc ,
l=1

reflects the relativistic kinetic energy of N noninteracting particles. Here c is the speed of light
and pl = |pl | is the magnitude of the momentum of particle l.
(a) Show that the canonical partition function can be expressed in the form
N
 mc 3 eu u ∞
 Z
1
ZN = 4πV K2 (u) , u ≡ βmc2 , Kγ (u) = dx sinh x sinh(γx)e−u cosh x
N! h u γ 0

where Kγ (u) is a modified Bessel function.

(b) Recover the result from [tex76] for
p ZN of−u the nonrelativistic ideal gas at kB T  mc2 by using
the asymptotic expression K2 (u) ' π/2ue for u  1.
(c) Recover the result from [tex77] for ZN of the ultrarelativistic ideal gas at kB T  mc2 by using
the asymptotic expression K2 (u) ' 2/u2 for u  1.

Solution:
[tex92] Relativistic ideal gas (entropy and internal energy)
(a) Derive from the result for the canonical partition function ZN of the relativistic classical ideal
gas as calculated in [tex91] the Helmholtz free energy A(T, V, N ), the equation of state p(T, V, N ),
the chemical potential µ(T, V, N ), and the entropy S(T, V, N ).
(b) Use the recursion relation, Kn0 (u) = −Kn−1 (u) − (n/u)Kn (u), for the modified Bessel functions
to derive the following expression for the internal energy:
 
K1 (u) 3 1
U (T, N ) = N kB T u + − 1 , u ≡ βmc2 , β = .
K2 (u) u kB T

Solution:
[tex93] Relativistic ideal gas (heat capacity)
(a) Derive from the result for the internal energy U (T, N ) of the relativistic classical ideal gas as
calculated in [tex92] the heat capacity in the form
  
3 K1 (u) K1 (u) 1
CV (T, N ) = N kB u u + − 3+u , u ≡ βmc2 , β = .
u K2 (u) K2 (u) kB T

by using the recursion relations Kn0 (u) = −Kn−1 (u) − (n/u)Kn (u) and Kn−1 (u) = Kn+1 (u) −
(2/u)Kn (u) for modified Bessel functions.
(b) Use the asymptotic results

K1 (u) 3 15 K1 (u) u
'1− + 2 (u  1), ' (u  1)
K2 (u) 2u 8u K2 (u) 2

to recover the the results CV = 23 N kB and CV = 3N kB in the nonrelativistic and ultrarelativistic

limits, respectively.

Solution:
Relativistic classical ideal gas [tsl34]

Heat capacity:

  
3 K1 (u) K1 (u)
CV = N k B u u + − 3+u
u K2 (u) K2 (u)

Kn (u): modified Bessel function; u ≡ βmc2 .

K1 (u) 3 15
Nonrelativistic limit (u  1): =1− + 2 + ...
K2 (u) 2u 8u
K1 (u) u
Ultrarelativistic limit (u  1): = + ...
K2 (u) 2

[from Greiner et al. 1995]

1
Contents of this Document [ttc12]

12. Grandcanonical Ensemble

• Grandcanonical ensemble. [tln60]

• Classical ideal gas (grandcanonical ensemble). [tex94]
• Density fluctuations and compressibility. [tln61]
• Density fluctuations in the grand canonical ensemble. [tex95]
• Density fluctuations and compressibility in the classical ideal gas. [tex96]
• Energy fluctuations and thermal response functions. [tex103]
• Microscopic states of quantum ideal gases. [tln62]
• Partition function of quantum ideal gases. [tln63]
• Ideal quantum gases: grand potential and thermal averages. [tln64]
• Ideal quantum gases: average level occupancies. [tsl35]
• Occupation number fluctuations. [tex110]
• Density of energy levels for ideal quantum gas. [tex111]
• Maxwell-Boltzmann gas in D dimensions. [tex112]
Grandcanonical ensemble [tln60]

Consider an open classical system (volume V , temperature T , chemical po-

tential µ). The goal is to determine the thermodynamic potential Ω(T, V, µ)
pertaining to that situation, from which all other thermodynamic properties
can be derived.
A quantitative description of the grandcanonical ensemble requires a set of
phase spaces ΓN , N = 0, 1, 2, . . . with probability densities ρN (X). The
interaction Hamiltonian for a system of N particles is HN (X).
X∞ Z
Maximize Gibbs entropy S = −kB d6N X ρN (X) ln[CN ρN (X)]
N =0 Γ
subject to the three constraints
∞ Z
X
• d6N X ρN (X) = 1 (normalization),
N =0 ΓN
X∞ Z
• d6N X ρN (X)HN (X) = hHi = U (average energy),
N =0 ΓN
X∞ Z
• d6N X ρN (X)N = hN i = N (average number of particles).
N =0 ΓN

Apply calculus of variation with three Lagrange multipliers:

" ∞ Z #
X
δ d6N X {−kB ρN ln[CN ρN ] + α0 ρN + αU HN ρN + αN N ρN } = 0
N =0 ΓN

∞ Z
X
⇒ d6N XδρN {−kB ln[CN ρN ] − kB + α0 + αU HN + αN N } = 0
N =0 ΓN
 
1 α0 αU αN
⇒ {· · · } = 0 ⇒ ρN (X) = exp −1+ HN (X) + N .
CN kB kB kB

Determine the Lagrange multipliers α0 , αU , αN :

  X ∞ Z  
α0 1 6N αU αN
exp 1 − = d X exp HN (X) + N ≡ Z,
kB N =0
C N ΓN k B k B

∞ Z
X
d6N X ρN (X){· · · } = 0 ⇒ S − kB + α0 + αU U + αN N = 0
N =0 ΓN

1 αN kB
⇒ U+ S+ N = ln Z.
αU αU αU
1
 1 µ
Compare with U − T S − µN = −pV = Ω ⇒ αU = − , αN = .
T T
Grand potential: Ω(T, V, µ) = −kB T ln Z = −pV.
∞ Z
X 1 1
Grand partition function: Z = d6N X e−βHN (X)+βµN , β = .
N =0
CN ΓN kB T
1
Probability densities: ρN (X) = e−βHN (X)+βµN .
ZCN

Grandcanonical ensemble in quantum mechanics:

1 −β(H−µN )
Z = Tr e−β(H−µN ) , ρ= e , Ω = −kB T ln Z.
Z

Derivation of thermodynamic properties from grand potential:

     
∂Ω ∂Ω ∂Ω
S=− , p=− , N = hN i = − .
∂T V,µ ∂V T,µ ∂µ T,V

Relation between canonical and grandcanonical partition functions:

∞
X ∞
X
µN/kB T
Z= e ZN = z N ZN , z ≡ eµ/kB T (fugacity).
N =0 N =0

Open system of indistinguishable noninteracting particles:

∞
1 N X 1 N N
ZN = Z̃ , Z= z Z̃ = ezZ̃
N! N =0
N !

⇒ Ω = −kB T ln Z = −kB T z Z̃.

Thermodynamic properties of the classical ideal gas in the grandcanonical

ensemble are calculated in exercise [tex94].

2
[tex94] Classical ideal gas (grandcanonical ensemble)
PN
Consider a classical ideal gas [HN = l=1 (p2l /2m)] in a box of volume V in equilibrium with heat
and particle reservoirs at temperature T and chemical potential µ, respectively.
(a) Show that the p grand partition function is Z = exp(zV /λ3T ), where z = exp(µ/kB T ) is the
fugacity, and λT = h2 /2πmkB T is the thermal wavelength.
(b) Derive from Z the grand potential Ω(T, V, µ), the entropy S(T, V, µ). the pressure p(T, V, µ),
and the average particle number hN i = N (T, V, µ).
(c) Derive from these expressions the familiar results for the internal energy U = 32 N kB T , and the
ideal gas equation of state pV = N kB T .

Solution:
Density fluctuations and compressibility [tln61]

Average number of particles in volume V :

∞ Z
X 1 1 ∂Z 1 ∂
N = hN i = d6N X N e−βHN (X)+βµN = = ln Z.
N =0
ZCN ΓN Zβ ∂µ β ∂µ

Fluctuations in particle number (in volume V ):

2
1 ∂2Z 1 ∂ 2 ln Z
  
2 2 1 ∂Z 1 ∂(βhN i) ∂N
hN i−hN i = − = 2 = 2 = kB T .
Zβ 2 ∂µ2 Zβ ∂µ β ∂µ2 β ∂µ ∂µ T V

Here we use Z = Z(β, V, µ).

   
V S ∂µ V ∂p
Gibbs-Duhem: dµ = dp − dT ⇒ = .
N N ∂(V /N ) T N ∂(V /N ) T

∂ ∂N ∂ N2 ∂
For V = const: = =− .
∂(V /N ) ∂(V /N ) ∂N V ∂N
∂ ∂V ∂ ∂
For N = const: = =N .
∂(V /N ) ∂(V /N ) ∂V ∂V

N 2 ∂µ
     
∂p ∂µ V
⇒ − =V ⇒ = 2 κ−1 .
V ∂N T V ∂V T N ∂N T V N T
 
1 ∂V
Compressibility: κT ≡ − .
V ∂p T N
N2
Fluctuations in particle number: hN 2 i − hN i2 = kB T κT .
V

An alternative expression for hN 2 i − hN i2 is calculated in exercise [tex95].

The density fluctuations for a classical ideal gas are calculated in exercise
[tex96].
At the critical point of a liquid-gas transition, the isotherm has an inflec-
tion point with zero slope (∂p/∂V = 0), implying κT → ∞. The strongly
enhanced density fluctuations are responsible for critical opalescence.
[tex95] Density fluctuations in the grandcanonical ensemble
Consider a system of indistinguishable particles in the grandcanonical ensemble. Derive the follow-
ing two expressions for the fluctuations in the number of particles N for an open system of volume
V in equilibrium with heat and particle reservoirs at temperature T and chemical potential µ,
respectively:
∂ ∂ ∂2p
hN 2 i − hN i2 = z z ln Z = kB T V ,
∂z ∂z ∂µ2
where z = exp(µ/kB T ) is the fugacity, p(T, V, µ) = −(∂Ω/∂V )T µ = −Ω/V is the pressure, and
Ω(T, V, µ) = −kB T ln Z is the grand potential.

Solution:
[tex96] Density fluctuations and compressibility of the classical ideal gas
(a) Use the results of [tex94] and [tex95] to show that the variance of the number of particles in a
classical ideal gas (open system) is equal to the average number of particles:

hN 2 i − hN i2 = hN i = N .

(b) Use this result to show that the isothermal compressibility of the classical ideal gas is κT = 1/p.

Solution:
[tex103] Energy fluctuations and thermal response functions
(a) Show that the following relation holds between the energy fluctuations in the microscopic
ensemble and the heat capacity of a system described by a microscopic Hamiltonian H:

h(H − hHi)2 i = kB T 2 CV .

(b) Prove the following relation in a similar manner:

   
∂CV
h(H − hHi)3 i = kB 2
T4 + 2T 3 CV .
∂T V

(c) Determine the relative fluctuations as measured by the quantities h(H − hHi)2 i/hHi2 and
h(H − hHi)3 i/hHi3 for the classical ideal gas with N atoms.

Solution:
Microscopic states of ideal quantum gases [tln62]

N
X
Hamiltonian: ĤN = ĥ` .
`=1

1-particle eigenvalue equation: ĥ` |k` i = ` |k` i.

N -particle eigenvalue equation: ĤN |k1 , . . . , kN i = EN |k1 , . . . , kN i.
N
X ~2 k2`
Energy: EN = ` , ` = .
`=1
2m

N -particle product eigenstates: |k1 , . . . , kN i = |k1 i . . . |kN i.

Symmetrized states for bosons: |k1 , . . . , kN i(S) .

1
• N = 2: |k1 , k2 i(S) = √ (|k1 i|k2 i + |k2 i|k1 i).
2

Antisymmetrized states for fermions: |k1 , . . . , kN i(A) .

1
• N = 2: |k1 , k2 i(A) = √ (|k1 i|k2 i − |k2 i|k1 i).
2

Occupation number representation: |k1 , . . . , kN i ≡ |n1 , n2 , . . .i.

Here k1 represents the wave vector of the first particle, whereas n1 refers to
the number of particles in the first 1-particle state.
∞
X
• energy: Ĥ|n1 , n2 , . . .i = E|n1 , n2 , . . .i, E= n k k .
k=1
∞
X
• number of particles: N̂ |n1 , n2 , . . .i = N |n1 , n2 , . . .i, N= nk .
k=1

` : energy of particle `. k : energy of 1-particle state k.

Allowed occupation numbers:

• bosons: nk = 0, 1, 2, . . .
• fermions: nk = 0, 1.
Partition function of ideal quantum gases [tln63]

0 ∞
!
X X
Canonical partition function: ZN = σ(n1 , n2 , . . .) exp −β nk k .
{nk } k=1

0
X ∞
X
: sum over all occupation numbers compatible with nk = N .
{nk } k=1

The statistical weight factor σ(n1 , n2 , . . .) is different for fermions and bosons:

• Bose-Einstein statistics: σBE (n1 , n2 , . . .) = 1 for arbitrary values of nk .


1 if all nk = 0, 1
• Fermi-Dirac statistics: σF D (n1 , n2 , . . .) = .
0 otherwise

What is the statistical weight factor for the Maxwell-Boltzmann gas?

∞
!N 0
1 N 1 X
−βk 1 X N!
n1 −β2
n2
ZN = Z̃ = e = e−β1 e ···
N! N! k=1
N! n1 !n2 ! . . .
{nk }
0 ∞
!
X 1 X
= exp −β n k k .
n1 !n2 ! . . .
{nk } k=1

1
• Maxwell-Boltzmann statistics: σM B (n1 , n2 , . . .) = .
n1 !n2 ! . . .

Grandcanonical partition function:

∞ ∞
!
X X X
N
⇒ Z= z ZN = σ(n1 , n2 , . . .) exp −β nk (k − µ) ,
N =0 {nk } k=1

∞
!
X
where we have used z N = (eβµ )N = exp βµ nk .
k=1

∞ X
∞ ∞
! ∞
X X Y
−1
• ZBE = · · · exp −β nk (k − µ) = 1 − ze−βk .
n1 =0 n2 =0 k=1 k=1
∞
1 1
! ∞
X X X Y
1 + ze−βk .


• ZF D = · · · exp −β nk (k − µ) =
n1 =0 n2 =0 k=1 k=1
∞ X
∞ ∞
! ∞
X 1 X Y
exp ze−βk .


• ZM B = ··· exp −β nk (k − µ) =
n1 =0 n2 =0
n1 !n2 ! . . . k=1 k=1
Ideal quantum gases:
grand potential and thermal averages [tln64]

Grand potential: Ω(T, V, µ) = −kB T ln Z = U − T S − µN = −pV.

∞
X ∞
X
−βk
• ΩM B = −kB T ze = −kB T e−β(k −µ) ,
k=1 k=1
∞
X ∞
X
−βk
ln 1 − e−β(k −µ) ,



• ΩBE = kB T ln 1 − ze = kB T
k=1 k=1
∞
X ∞
X
−βk
ln 1 + e−β(k −µ) .



• ΩF D = −kB T ln 1 + ze = −kB T
k=1 k=1

Parametric representation [a = 1 (FD), a = 0 (MB), a = −1 (BE)]:

∞
pV 1X
ln 1 + aze−βk .


ln Z = =
kB T a k=1

Average number of particles:

    ∞ ∞
∂Ω 1 ∂ ln Z X 1 X
N =− = = = hnk i.
∂µ T,V β ∂µ T,V k=1
z −1 eβk + a k=1

Average energy (internal energy):

  ∞ ∞
∂ ln Z X k X
U =− = = k hnk i.
∂β z,V k=1
z −1 eβk + a k=1

Average occupation number of energy level k :

∂ ln Z 1
hnk i = −β −1 = β( −µ) .
∂k e k +a

Fluctuations in occupation number [tex110]:

∂ 2 ln Z
hn2k i − hnk i2 = β −2 .
∂2k
Average occupation numbers
for MB, FD, and BE gases [tsl35]

Average occupation number of energy level ǫk :

1
hnk i =
eβ(ǫk −µ) +a
• a = 1: Fermi-Dirac gas,
• a = 0: Maxwell-Boltzmann gas,
• a = −1: Bose-Einstein gas.

Range of 1-particle energies: ǫk ≥ 0.

BE gas restriction: µ ≤ 0 ⇒ 0 ≤ z ≤ 1.

2.5

1.5 Maxwell-Boltzmann Bose-Einstein

<nk>

0.5 Fermi-Dirac

0
-2 -1 0 1 2 3
β (εk - µ)

The BE and FD gases are p

well approximated by the MB gas provided the
thermal wavelength λT = h2 /2πmkB T is small compared to the average
interparticle distance:

β(ǫk − µ) ≫ 1 ⇒ − βµ ≫ 1 ⇒ z ≪ 1.

[tex94] for D = 3 : ⇒ λT ≪ (V /N )1/3 .

[tex110] Occupation number fluctuations
Consider an ideal quantum gas specified by the grand partition function Z. Start from the expres-
sions 2 ∞
1 −2 ∂ 2 Z

1 −1 ∂Z 1X
2 2
hnk i − hnk i = β − β , ln Z = ln(1 + aze−βk ),
Z ∂2k Z ∂k a
k=1

where a = +1, 0, −1 represent the FD, MB, and BE cases, respectively, to derive the following
result for the relative fluctuations in the occupation numbers:

hn2k i − hnk i2 1
2
= − a.
hnk i hnk i

Note that in the BE (FD) statistics, these fluctuations are enhanced (suppressed) relative to those
in the MB statistics.

Solution:
[tex111] Density of energy levels for ideal quantum gas
Consider a nonrelativistic ideal quantum gas in D dimensions and confined to a box of volume
V = LD with rigid walls. Show that the density of energy levels is

LD  m D/2 D/2−1
D() =  .
Γ(D/2) 2π~2

Solution:
[tex112] Maxwell-Boltzmann gas in D dimensions
From the expressions for the grand potential and the density of energy levels of an ideal Maxwell-
Boltzmann gas in D dimensions and confined to a box of volume V = LD with rigid walls,
X LD  m D/2 D/2−1
Ω(T, V, µ) = −kB T e−β(k −µ) , D() =  ,
Γ(D/2) 2π~2
k

derive the familiar results pV = N kB T for the equation of state, CV N = (D/2)N kB for the heat
capacity, and pV (D+2)/D = const for the adiabate at fixed N .

Solution:
Contents of this Document [ttc13]

13. Ideal Quantum Gases I: Bosons

• Bose-Einstein functions. [tsl36]

• Ideal Bose-Einstein gas: equation of state and internal energy. [tln67]
• BE gas in D dimensions I: fundamental relations. [tex113]
• Reference values for T , V /N , and p. [tln71]
• Bose-Einstein condensation. [tsl38]
• Ideal Bose-Einstein gas: isochores. [tsl39]
• BE gas in D dimensions II: isochore. [tex114]
• BE gas in D dimensions III: isotherm and isobar. [tex115]
• Bose-Einstein gas: isotherms. [tsl40]
• Bose-Einstein gas: isobars. [tsl48]
• Bose-Einstein gas: phase diagram. [tln72]
• Bose-Einstein heat capacity. [tsl41]
• BE gas in D dimensions IV: heat capacity at high temperature. [tex97]
• BE gas in D dimensions V: heat capacity at low temperature. [tex116]
• BE gas in D dimensions VI: isothermal compressibility. [tex128]
• BE gas in D dimensions VII: isobaric expansivity. [tex129]
• BE gas in D dimensions VIII: speed of sound. [tex130]
• Ultrarelativistic Bose-Einstein gas. [tex98]
• Blackbody radiation. [tln68]
• Statistical mechanics of blackbody radiation. [tex105]
Bose−Einstein functions [tsl36]

∞
1 dx xn−1 X zl
Z ∞
gn (z) ≡ = , 0 ≤ z ≤ 1.
Γ(n) 0 z −1 ex − 1 l=1
l n

Special cases:
z
g0 (z) = , g1 (z) = − ln(1 − z), g∞ (z) = z.
1−z

Riemann zeta function:

∞
. X 1
gn (1) = ζ(n) = .
l=1
ln

Special values:
π2 π4 π6
ζ(1) → ∞, ζ(2) = , ζ(4) = , ζ(6) = .
6 90 945
Recurrence relation:
zgn′ (z) = gn−1(z), n ≥ 1.

Singularity at z = 1 for non-integer n:

∞
X (−1)ℓ .
gn (α) = Γ(1 − n)αn−1 + ζ(n − ℓ)αℓ , α = − ln z.
ℓ=0
ℓ!

2.5

g1/2 g1 g3/2

g2
1.5
gn(z)

g5/2

0.5

0
0 0.2 0.4 0.6 0.8 1
z
Ideal Bose-Einstein gas:
equation of state and internal energy [tln67]

Conversion of sums into integrals by means of density of energy levels [tex113]:

V  m D/2
D(ǫ) = ǫD/2−1 , V = LD .
Γ(D/2) 2π~2

Fundamental thermodynamic relations for BE gas:

Z ∞
pV X

V
=− ln 1 − ze −βǫk
=− dǫ D(ǫ) ln 1 − ze−βǫ = D gD/2+1 (z),
kB T k 0 λT

1
Z ∞
X D(ǫ) V
N = βǫ
= dǫ −1 βǫ = D gD/2(z), z < 1,
k
z e −1
−1 k
0 z e −1 λT
Z ∞
X ǫk D(ǫ)ǫ D V
U= = dǫ = k B T gD/2+1 (z).
k
z −1 eβǫk − 1 0 z −1 eβǫ − 1 2 λD
T

Warning: The range of fugacity is limited to the interval 0 ≤ z ≤ 1. At

z = 1, the expression for N must be amended by an additive term z/(1−z) to
account for the possibility of a macroscopic population of the lowest energy
level (at ǫ = 0). This amendment is only necessary for dimensionalities
D > 2, i.e. for the cases with limǫ→0 D(ǫ) = 0.

Equation of state (with fugacity z in the role of parameter):

pV gD/2+1 (z)
= , z < 1.
N kB T gD/2 (z)

1
∞

0.8 3
2
D=1
pV/NkBT

0.6

0.4

0.2

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
z
[tex113] BE gas in D dimensions I: fundamental relations
From the expressions for the grand potential and the density of energy levels of an ideal Bose-
Einstein gas in D dimensions and confined to a box of volume V = LD with rigid walls,
X V  m D/2
Ω(T, V, µ) = kB T ln(1 − ze−βk ), D() = D/2−1 ,
Γ(D/2) 2π~2
k

derive the fundamental thermodynamic relations at

p fugacity z < 1 in terms of the Bose-Einstein
functions gn (z) and the thermal wavelength λT = h2 /2πmkB T as follows:

pV V V D V
= D gD/2+1 (z), N = gD/2 (z), U= kB T D gD/2+1 (z).
kB T λT λD
T 2 λT

Solution:
Reference Values for T , V /N , and p [tln71]

The reference values introduced here are based on

s r
. h2 Λ h2
(i) thermal wavelength: λT = = , Λ= .
2πmkB T kB T 2πm
.
(ii) MB equation of state: pv = kB T , v = V /N .

The reference values for kB T , v, and p in isochoric, isothermal, and isobaric

processes are

Λ Λ
kB Tv = pv = (v = const.)
v 2/D v 2/D+1
 D/2  D/2+1
Λ kB T
vT = pT = Λ (T = const.)
kB T Λ
 D/(D+2)
 p 2/(D+2) Λ
kB Tp = Λ vp = (p = const.)
Λ p

These reference values are useful for bosons and fermions.

Universal curves for isochores, isotherms, and isobars:

• p/pv versus T /Tv at v = const.

• p/pT versus v/vT at T = const.
• v/vp versus T /Tp at p = const.

For fermions we will introduce alternative reference values based on the chem-
ical potential (Fermi energy).
Bose-Einstein condensation [tsl38]

Particles in the gas phase and in the Bose-Einstein condensate (BEC):

V z
N = g (z) +
D D/2
= Ngas + NBEC .
λT 1−z

Consider process at v = const.

Onset of macroscopic population of the lowest energy level begins when the
fugacity locks in to the value z = 1:

z  O(1), z < 1,
=
1−z 
O(N ), z = 1.

Ngas NBEC
T ≥ Tc : = 1, = 0.
N N
  D/2
 Ngas [V /λD
T ]ζ(D/2) T
= = ,


[V /λTc ]ζ(D/2)
D
Tc

 N

T ≤ Tc :
  D/2

 N BEC N gas T
=1− = 1− .


Tc

N N

0.8
NBEC/N

0.6

0.4 D = 2.5 3 6 ∞

0.2

0
0 0.2 0.4 0.6 0.8 1
T/Tv
Ideal Bose-Einstein gas: isochores [tsl39]

Isochore at T ≥ Tc [tex114]:
p gD/2+1 (z) T  −2/D
= 2/D+1 , = gD/2 (z) .
pv gD/2 (z) Tv

Isochore at T ≤ Tc (also valid asymptotically for T ≪ Tv in D ≤ 2):

 D/2+1
p T
= ζ(D/2 + 1).
pv Tv

Critical temperature:


 0 D =1
Tc 
0 D =2
= [ζ(D/2)]−2/D =
Tv 
 0.527 D =3
1 =∞

D

High-temperature asymptotic behavior:

"  D/2 #
p T 1 Tv
∼ 1 − D/2+1 .
pv Tv 2 T

1.2

0.8
p/pv

0.6

n
an
m
0.4
o ltz
ll-B
xwe
Ma
0.2 TC/Tv
1 2 3 D=∞

0
0 0.2 0.4 0.6 0.8 1 1.2
T/Tv
[tex114] BE gas in D dimensions II: isochore
(a) From the fundamental thermodynamic relations for the Bose-Einstein gas in D dimensions (see
[tln67]), derive the following parametric expression for the isochore at T ≥ Tc :

p gD/2+1 (z) T  −2/D

= 2/D+1 , = gD/2 (z) ,
pv gD/2 (z) Tv

.
where kB Tv = Λv −2/D and pv = Λv −2/D+1 with Λ = h2 /2πm are convenient reference values.
(b) Calculate the leading correction to the Maxwell-Boltzmann result at high temperature. (c)
Calculate the exact dependence of p/pv on T /Tv at T ≤ Tc in D > 2. Show that this result also
holds asymptotically for T  Tv in dimensions D = 1 and D = 2.

Solution:
[tex115] BE gas in D dimensions III: isotherm and isobar
(a) From the fundamental thermodynamic relations for the Bose-Einstein gas in D > 2 dimensions
(see [tln67]), derive the following expressions for the isotherm at v > vc and the isobar at T ≤ Tc :
p v
= gD/2+1 (z), = [gD/2 (z)]−1 ;
pT vT
 D/(D+2)
v gD/2+1 (z) T  −2/(D+2)
= , = gD/2+1 (z) .
vp gD/2 (z) Tp
where vT = (Λ/kB T )D/2 , pT = Λ(kB T /Λ)D/2+1 , kB Tp = Λ(p/Λ)2/(D+2) , vp = (Λ/p)D/(D+2) with
.
Λ = h2 /2πm are convenient reference values for temperature and pressure and reduced volume. (b)
Calculate the leading correction to the Maxwell-Boltzmann result for the isotherm at low density
and for the isobar at high temperature.

Solution:
Ideal Bose-Einstein gas: isotherms [tsl40]

For D > 2 we must again distinguish two regimes. At v > vc , all bosons are
in the gas phase. At v < vc , a BEC is present. Only the bosons in the gas
phase contribute to the pressure.
Isotherm at v ≥ vc = λD
T /ζ(D/2):

p v
= gD/2+1 (z), = [gD/2 (z)]−1 .
pT vT

Isotherm at v ≤ vc :


 2.612 D =1
p pc 
1.645 D =2
= = ζ(D/2 + 1) =
pT pT 
 1.341 D =3
1 =∞

D

Critical (reduced) volume:



 0 D =1
vc 
0 D =2
= [ζ(D/2)]−1 =
vT 
 0.383 D =3
1 =∞

D

2.5 D=1 Maxwell-Boltzmann

2
p/pT

1.5
3

1
∞

0.5

0
0 0.2 0.4 0.6 0.8 1 1.2
v/vT
Ideal Bose-Einstein gas: isobars [tsl48]

A phase transition at Tc > 0 takes place in all dimensions D ≥ 1. However,

the existence of a BEC requires vc > 0, which is realized only for D > 2.
Isobar at T > Tc :
 D/(D+2)
v gD/2+1 (z) T  −2/(D+2)
= , = gD/2+1 (z) .
vp gD/2 (z) Tp

Critical point:


 0 D =1
D/(D+2)
vc [ζ(D/2 + 1)] 
0 D =2
= =
vp ζ(D/2)  0.383
 D =3
1 =∞

D



 0.527 D =1
Tc 
0.779 D =2
= [ζ(D/2 + 1)]−2/(D+2) =
Tp 
 0.884 D =3
1 =∞

D

1.5

∞
v/vp

1
n
an
m
o ltz
-B 3
ll
xwe
Ma
0.5

2
D=1
0
0 0.5 1 1.5 2
T/Tp
Ideal Bose-Einstein gas: phase diagram [tln72]

D=1 D=3

p p

60 60
50 50
40 40
30 30
20 20
10 10
00 00
1 20 1 20
2 15 2 15
v 3 10 v 3 10
4 5 kBT 4 5 kBT
5 0 5 0

D=∞

80
70
60
50
40

kB T, T > Tc
30
pv =
20
10
0, T < Tc
00
1
2
20
. h2
v 3 10
15
kB Tc = Λ = .
4
5 0
5 kBT 2πm

• D = 1: Transition at T ≥ 0 and v = 0 (transition line = isochore).

• D = 3: Transition at T > 0 and v > 0.
• D = ∞: Transition at T > 0 and v > 0 (transition line = isotherm).
Ideal Bose-Einstein gas: heat capacity [tsl41]

Internal energy:

D gD/2+1 (z) T


 , T ≥ Tc ,
U  2 gD/2 (z) Tv
=  D/2+1
N kB Tv  D T
 ζ(D/2 + 1)
 , T ≤ Tc .
2 Tv

Heat capacity at T ≥ Tc [use zgn′ (z) = gn−1 (z) for n ≥ 1]:

gD/2+1 (z) D 2 gD/2+1 (z)

′
D D2
 
CV
= + − .
N kB 2 4 gD/2 (z) 4 gD/2′
(z)

Heat capacity at T ≤ Tc :

 D/2
D D2 D D 2 ζ D2 + 1
     D/2  
CV D T T
= + ζ +1 = +
.
N kB 2 4 2 Tv 2 4 ζ 2D
Tc

High-temperature asymptotic behavior:

"  D/2 #
CV D D/2 − 1 Tv
∼ 1 + D/2+1 .
N kB 2 2 T

2.5
5

2
(2/D)Cv/NkB

1.5

3
1
2
D=1
0.5

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4
T/Tv
[tex97] BE gas in D dimensions IV: heat capacity at high temperature
The internal energy of the ideal Bose-Einstein gas in D dimensions and at T ≥ Tc is given by the
following expression:
D gD/2+1 (z)
U = N kB T .
2 gD/2 (z)
Use this result to derive the following expression for the heat capacity CV = (∂U/∂T )V N :
0
D D2 gD/2+1 (z) D2 gD/2+1 (z)
 
CV
= + − 0 .
N kB 2 4 gD/2 (z) 4 gD/2 (z)

Use the derivative ∂/∂T of the result gD/2 (z) = N λD D

T /V with V = L to calculate any occurrence
0
of (∂z/∂T )V N in the derivation. Use the recursion relation zgn (z) = gn−1 (z) for n ≥ 1 to further
simplify the results pertaining to D ≥ 2.

Solution:
[tex116] BE gas in D dimensions V: heat capacity at low temperature
The internal energy of the ideal Bose-Einstein gas in D > 2 dimensions and at T ≤ Tc is given by
the following expression:
 D/2+1
U D T
= ζ(D/2 + 1)
N kB Tv 2 Tv
(a) Use this result to derive the following expression for the heat capacity CV = (∂U/∂T )V N :
D

 D/2
D D2
 
CV ζ 2 +1 T
= + D

,
N kB 2 4 ζ 2
Tc
.
where Tc = Tv [ζ(D/2)]−2/D is the critical temperature and kB Tv = Λ/v 2/D with v = V /N and
. 2
Λ = h /2πm a convenient reference temperature. (b) Show that the heat capacity is continuous
at T = Tc if D ≤ 4 and discontinuous if D > 4. Find the discontinuity ∆CV /N kB as a function of
D for D > 4. (c) Infer from the result of [tex97] the leading singularity of CV /N kB at T /Tv  1
for D = 1 and D = 2. Then show that these singularitues are consistent with the expression for
CV /N kB obtained here in part (a) provided we substitute (Tv /Tc )D/2 = ζ(D/2).

Solution:
[tex128] BE gas in D dimensions VI: isothermal compressibility
(a) Show that the isothermal compressibility, κT = −(1/V )(∂V /∂p)T N , of the ideal BE gas in D
dimensions at T > Tc is
0
gD/2 (z) v 1
pT κT = 0 , = ,
gD/2 (z)gD/2+1 (z) vT gD/2 (z)
. . . .
where v = V /N , vT = (Λ/kB T )D/2 , pT = kB T /vT , Λ = h2 /2πm, and gn (z) are BE functions.
0
Use zgn (z) = gn−1 (z) for n ≥ 1 to simplify the results in D ≥ 2. (b) Sketch pT κT versus v/vT
for v ≥ 0 in D = 1 and for v ≥ vc in D = 3, where vc /vT = [ζ(D/2)]−1 marks the onset of BEC.
(c) Determine the nature of the singularity of κT as v/vT → 0 in D = 1, 2. Determine the critical
compressibility pT κT at v = vc in D = 3, 5.

Solution:
[tex129] BE gas in D dimensions VII: isobaric expansivity
To derive the parametric expression of the isobaric expansivity of the ideal BE gas at T > Tc ,
0
" #
gD/2+1 (z)gD/2 (z) D

Tp D Tp  D/2+1
Tp αp = +1 0 − , = gD/2+1 (z) ,
T 2 gD/2 (z)gD/2+1 (z) 2 T
.
where kB Tp = Λ(p/Λ)2/(D+2) , Λ = h2 /2πm, and gn (z) are BE functions, establish first the
.
general thermodynamic relation αp = κT (∂p/∂T )v with v = V /N , the BE-specific relation
CV = N (D/2)v(∂p/∂T )v , and the results for CV and κT calculated in [tex97] and [tex128].

Solution:
[tex130] BE gas in D dimensions VIII: speed of sound
(a) Start from the relation c = (ρκS )−1/2 for the speed of sound as established in [tex18], where
ρ = m/v is the mass density and κS the adiabatic compressibility. Use general thermodynamic re-
lations between response functions to derive the following expression for c in terms of dimensionless
quantities:
mc2 (T /Tp )2 (v/vT )(Tp αp )2
 
(v/vT )
= 1+ ,
kB T (pT κT ) (pT κT )(CV /N kB )
where vT , pT , Tp are defined in [tln71]. (b) Use the expressions derived in [tex129] for αp , in [tex128]
for κT , and in [tex97] for CV to derive the result

mc2 gD/2+1 (z) 2

=γ , γ =1+ .
kB T gD/2 (z) D

(c) Relate the T -dependence of mc2 to that of the isochore for v = const and to that of the isobar
for p = const.

Solution:
[tex98] Ultrarelativistic Bose−Einstein gas
Consider a Bose-Einstein gas with ultrarelativistic one-particle energy k = c~k = cp in the
grandcanonical ensemble at temperature T and chemical potential µ = 0.
(a) Show that the one-particle density of states is D() = (4πV /h3 c3 )2 .
(b) Calculate the pressure p(T ), the internal energy U (T, V ), and the average number of particles
in excited states N (T, V ).
(c) Show that the heat capacity is CV /kB = [16π 5 /15h3 c3 ]V (kB T )3 .

Solution:
Blackbody radiation [tln68]

Electromagnetic radiation inside cavity in thermal equilibrium at tempera-

ture T . Grandcanonical ensemble of photons ( = ~ω = cp, p = ~k, spin
s = 1, bosonic, purely transverse).
4πV 2
Density of states: D() = g  with g = 2 independent polarizations.
h3 c3
1
Average occupation number: hn iBE = β .
e −1
Number of photons with energies between  and  + d:
8πV 2 1
dN () = hn iBE D()d = d.
h3 c3 eβ − 1

Spectral density inside cavity: [use dN () = V dn(ω) and  = ~ω]:

dn(ω) ~ dN () ω2 1
= = 2 3 β~ω .
dω V d π c e −1

Spectral energy density inside cavity: du = ~ωdn = ρ(ω)dω.

ω2 ~ω 8πν 2 hν
ρ(ω)dω = dω = dν.
π 2 c3 eβ~ω − 1 c3 eβhν − 1

Rate (per unit area) at which particles with (average) speed c escape from
cavity through small opening [tex62]: dN/dt = 14 (N/V )c.
c dn(ω) ω2 1
Spectral density of radiation: R(ω) = = 2 2 β~ω .
4 dω 4π c e −1
Spectral energy density of radiation:
ω2 ~ω
Q(ω) = ~ωR(ω) = 2 2 β~ω
(Planck radiation law).
4π c e −1

High frequencies: ultrarelativistic MB particles [use hn iM B = e−β ]:

~ω 3 −βω
Q(ω) = e (Wien radiation law).
4π 2 c2

Low frequencies: equipartition law applied to electromagnetic modes:

kB T ω 2
Q(ω) = (Rayleigh−Jeans radiation law).
4π 2 c2
[tex105] Statistical mechanics of blackbody radiation
Electromagnetic radiation inside a cavity is in thermal equilibrium with the walls at temperature
T . This system can be described by a grandcanonical ensemble (with µ = 0) of photons (massless
bosonic particles) with energy  = ~ω and density of states D̄(ω) = (V /π 2 c3 )ω 2 .
(a) Show that the internal energy can be expressed in the form

π 2 kB
4
U (T, V ) = σV T 4 , σ =
15~ c3
3

as postulated in a previous thermodydnamics problem [tex23].

1
(b) Show that the equation of state can be expressed in the form pV = 3 U (T, V ) as was also
postulated in [tex23].

Solution:
Contents of this Document [ttc14]

14. Ideal Quantum Gases II: Fermions

• Fermi-Dirac functions. [tsl42]

• Ideal Fermi-Dirac gas: equation of state and internal energy. [tln69]
• Ideal Fermi-Dirac gas: chemical potentia.l [tsl43]
• FD gas in D dimensions: chemical potential I. [tex117]
• FD gas in D dimensions: chemical potential II. [tex118]
• Ideal Fermi-dirac gas: average level occupancy. [tsl44]
• Ideal Fermi-Dirac gas: isochores I. [tsl46]
• FD gas in D dimensions: statistical interaction pressure. [tex119]
• Ideal Fermi-Dirac gas: isotherms. [tln70]
• FD gas in D dimensions: isotherm and adiabate. [tex120]
• FD gas in D dimensions: ground-state energy. [tex102]
• Ideal Fermi-Dirac gas: heat capacity. [tsl45]
• FD gas in D dimensions: heat capacity at high temperature. [tex100]
• FD gas in D dimensions: heat capacity at low temperature. [tex101]
• Ideal Fermi-Dirac gas: isochores II. [tln73]
• Ideal Fermi-Dirac gas: phase diagram in infinite dimensions. [tln74]
• Stable white dwarf. [tex121]
• Unstable white dwarf. [tex122]
Fermi-Dirac functions [tsl42]

1 dx xn−1
Z ∞
fn (z) ≡ , 0≤z<∞
Γ(n) 0 z −1 ex + 1

Series expansion:
∞
X zl
fn (z) = (−1)l−1 , 0 ≤ z ≤ 1.
l=1
ln

Special cases:
z
f0 (z) = , f1 (z) = ln(1 + z), f∞ (z) = z.
1+z

Recurrence relation:
zfn′ (z) = fn−1 (z), n ≥ 1.

Asymptotic expansion for z ≫ 1:

" #
(ln z)n
 
X 1 ζ(k)
fn (z) = 1+ 2n(n − 1) · · · (n − k + 1) 1 − k−1
Γ(n + 1) k=2,4,...
2 (ln z)k
(ln z)n π2

= 1 + n(n − 1) (ln z)−2
Γ(n + 1) 6
7π 4

+ n(n − 1)(n − 3) (ln z) + . . .
−4
360

2.5

f5/2
2
f2

f3/2

1.5
fn(z)

f1

f1/2
1

0.5

0
0 0.5 1 1.5 2 2.5 3 3.5 4
z
Ideal Fermi-Dirac gas:
equation of state and internal energy [tln69]

Conversion of sums into integrals by means of density of energy levels:

gV  m D/2 D/2−1
D(ǫ) = ǫ , V = LD .
Γ(D/2) 2π~2

Fundamental thermodynamic relations for FD gas:

Z ∞
pV X

gV
= ln 1 + ze−βǫk
= dǫ D(ǫ) ln 1 + ze−βǫ = D fD/2+1 (z),
kB T k 0 λT

1
Z ∞
X D(ǫ) gV
N = = dǫ = D fD/2 (z),
k
z −1 eβǫk +1 0 z −1 eβǫ
+1 λT

Z ∞
X ǫk D(ǫ)ǫ D gV
U= = dǫ = kB T D fD/2+1 (z).
k
z −1 eβǫk +1 0 z −1 eβǫ
+1 2 λT

Note: The range of fugacity has no upper limit: 0 ≤ z ≤ ∞. The chemical

potential µ is unrestricted. The factor g is included to account for any
existing level degeneracy due to internal degrees of freedom (e.g. spin) of the
fermions.

Equation of state (with fugacity z in the role of parameter):

pV fD/2+1 (z)
= .
N kB T fD/2 (z)

1.5

1.4

D=1
1.3

2
pV/NkBT

1.2
3

1.1

Maxwell-Boltzmann
1

0.9
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
z
Ideal Fermi-Dirac gas: chemical potential [tsl43]

Fugacity z from x = fD/2 (z), where

s
λD
T . gV h2
x= , v= , λT = .
v N 2πmkB T

µ T T
Chemical potential [tex117]: = ln z, = [fD/2 (z)]−2/D .
kB Tv Tv Tv

D=1
1
2
3
0 4
µ/kBTF

-1

-2

-3

-4
0 0.5 1 1.5 2 2.5 3 3.5 4
T/TF

Λ . h2
Reference temperature: kB Tv = , Λ= .
v 2/D 2πm
For a complete list of reference values see [tln71].

Fermi energy: lim µ = ǫF = kB TF .

T →0

TF D
= [Γ(D/2 + 1)]2/D
D≫1
Fermi temperature: .
Tv 2e
[tex117] FD gas in D dimensions: chemical potential I
(a) Start from the fundamental thermodynamic relation N = (gV /λD T )fD/2 (z) for the ideal Fermi-
. .
Dirac gas in D dimensions and use the reference temperature kB Tv = Λ/v 2/D , v = gV /N , Λ =
h2 /2πm to derive the following parametric expression for the dependence on temperature T of the
chemical potential µ:
µ T T
= ln z, = [fD/2 (z)]−2/D .
kB Tv Tv Tv
(b) Derive the following expression for the Fermi energy F and the Fermi temperature TF :

µ(T ) F TF 2/D
lim = = = [Γ(D/2 + 1)] .
T →0 kB Tv kB Tv Tv

(c) Show that this result includes the familiar result, F = (h2 /2m)(3N /4πgV )2/3 for D = 3.

Solution:
[tex118] FD gas in D dimensions: chemical potential II
Start from the results derived in [tex117] to infer the following expressions for the fugacity z and
the chemical potential µ at T  TF :
2
π2

TF µ T
ln z ∼ , ∼1− (D − 2) .
T kB TF 12 TF

Solution:
Ideal Fermi-Dirac gas:
average level occupancy [tsl44]

Average occupancy of 1-particle state at energy ǫ if system (with fixed N , V )

is at temperature T :

1
hnǫ i = with µ(T ) from [tex117].
eβ(ǫ−µ)+1

T/TF = 1 T/TF = 0

0.8

0.6
<nε>

0.4

0.2 T/TF = 5

-15 -10 -5 0 5 10
ε/εF

1
Note: hnǫ i = occurs at ǫ = µ(T ).
2

Limit T → 0: µ(T ) → ǫF , hnǫ i → Θ(ǫF − ǫ).

Ideal Fermi-Dirac gas: isochores I [tsl46]

Reference values for temperature and pressure:

Λ kB Tv . h2 . gV
kB Tv = , pv =
; Λ= , v= .
v 2/D v 2πm N
  2/D
TF pF D D≫1 D
= = Γ +1 .
Tv pv 2 2e

Isochore:
−2/D
T fD/2+1 (z)
  
p T D
= , = Γ + 1 fD/2 (z) .
pF TF fD/2 (z) TF 2

Low-temperature limit [tex119]:

 −1
p D
lim = +1 .
T →0 pF 2

High-temperature asymptotic regime [tex119]:

"   −1  D/2 #
pV T D TF
∼ 1 + 2D/2+1 Γ +1 .
N kB TF TF 2 T

The excess pressure relative to the Maxwell-Boltzmann line may be called a

manifestation of statistical interaction pressure.

2.5

D=1

1.5
2
p/pF

3
mann
B oltz
ell-
1
xw
Ma

0.5

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4
T/TF
[tex119] FD gas in D dimensions: statistical interaction pressure
Consider the isochore of an ideal Fermi-Dirac gas in D dimensions, as given by the parametric
relation
p T fD/2+1 (z) T  −2/D
= , = fD/2 (z) .
pv Tv fD/2 (z) Tv
. .
where kB Tv = Λ/v 2/D , pv = kB T /v, Λ = h2 /2πm, v = gV /N . The upward deviation of this
result from the Maxwell-Boltzmann result, p/pv = T /Tv , is a manifestation of repulsive statistical
interaction between fermions. (a) Calculate the high-T asymptotic dependence of p/pv on T /Tv
including the leading correction to MB behavior. (b) Calculate the low-T limit of p/pv . (c)
Calculate the low-T limit of p/pF , where TF = Tv [Γ(D/2 + 1)]2/D is the Fermi temperature and
pF = kB TF /v the associated reference pressure. (d) Compare the differently scaled statistical
interaction pressures p/pv and p/pF at T = 0 in the limit D → ∞.

Solution:
Ideal Fermi-Dirac gas: isotherms [tln70]

.
Reference values for reduced volume v = gV /N and pressure p:

vT = λD
T, pT = gkB T /λD
T.

Parametric expression for isotherm:

p v
= fD/2+1 (z), = [fD/2 (z)]−1 .
pT vT

Isotherm at low density [tex120]:

pv = const, v ≫ vT .

Isotherm at high density [tex120]:

pv (D+2)/D = const, v ≪ vT .

3.5 3 2 D=1

2.5
p/pT

1.5

1 Maxwell-Boltzmann

0.5

0
0 0.5 1 1.5 2
v/vT
[tex120] FD gas in D dimensions: isotherm and adiabate
(a) Show that the isotherm of an ideal Fermi-Dirac gas in D dimensions is described by the
parametric relation
p v
= fD/2+1 (z), = [fD/2 (z)]−1 ,
pT vT
where vT = λD D
T and pT = gkB T /λT are convenient reference values and v = gV /N is the reduced
volume. (b) Show that the adiabate is described by the relation pv (D+2)/D = const for all values of
v/v0 . (c) Show that the relation for the isotherm approaches Boyle’s law, pv = const, for v  vT
and that it approaches the adiabate, pv (D+2)/D = const, for v  vT .

Solution:
[tex102] FD gas in D dimensions: ground-state energy
Given are the following expressions for the average number of particles, the average energy, the
average occupation number at T = 0, and the density of states for an ideal Fermi-Dirac gas in D
dimensions:
 D/2
X X gV 2πm
N = hnk i, U = hnk ik , hnk i = Θ(F − k ), D() = D/2−1 .
Γ(D/2) h2
k k

Derive from these expressions the following results for the dependence of the ground-state energy
per particle, U0 /N , on the Fermi energy F and for the dependence of the ground-state energy
density U0 /V on the particle density N /V :
 (D+2)/D
U0 D U0 N
= F , ∝ .
N D+2 V V

Solution:
Ideal Fermi-Dirac gas: heat capacity [tsl45]

Internal energy:
D fD/2+1 (z)
U= N kB T .
2 fD/2 (z)

Heat capacity [use zgn′ (z) = gn−1 (z) for n ≥ 1]:

fD/2+1 (z) D 2 fD/2+1 (z)

′
D D2
 
CV
= + − .
N kB 2 4 fD/2 (z) 4 fD/2
′
(z)

Low-temperature asymptotic behavior:

CV π2 T
∼D .
N kB 6 TF

High-temperature asymptotic behavior:

"  D/2 #
CV D D/2 − 1 TF
∼ 1 − D/2−1 .
N kB 2 2 Γ(D/2) T

1.6

1.4

1.2 D=3

1
CV/NkB

D=2
0.8

0.6

D=1
0.4

0.2

0
0 0.5 1 1.5 2 2.5 3
T/TF
[tex100] FD gas in D dimensions: heat capacity at high temperature
The internal energy of the ideal Fermi-Dirac gas in D dimensions is given by the expression,

D fD/2+1 (z)
U = N kB T .
2 fD/2 (z)

(a) Use this result to derive the following expression for the heat capacity CV = (∂U/∂T )V N :
0
D D2 fD/2+1 (z) D2 fD/2+1 (z)
 
CV
= + − 0 .
N kB 2 4 fD/2 (z) 4 fD/2 (z)

Use the derivative ∂/∂T of the result fD/2 (z) = N λD D

T /gV with V = L to calculate any occurrence
0
of (∂z/∂T )V N in the derivation. Use the recursion relation zfn (z) = fn−1 (z) for n ≥ 1 to further
simplify the results pertaining to D ≥ 2. (b) Infer from this result the leading correction to the
Maxwell-Boltzmann result, CV = (D/2)N kB , at high temperature.

Solution:
[tex101] FD gas in D dimensions: heat capacity at low temperature
Use the results of [tex118] and [tex100] to determine the low-temperature asymptotic behavior,

CV π2 T
∼D ,
N kB 6 TF
of the heat capacity of the ideal Fermi-Dirac gas in D dimensions.

Solution:
Ideal Fermi-Dirac gas: isochores II [tln73]

Reference values for temperature and pressure:

Λ kB Tv . h2 . gV
kB Tv = , pv = ; Λ= , v= .
v 2/D v 2πm N

Noncommuting limits z → ∞, D → ∞:

• z < ∞, D → ∞:

p T fD/2+1 (z) D→∞ T

= −→ (ideal MB gas).
pv Tv fD/2 (z) Tv

• D → ∞, z → ∞ with D/2 = r ln z, r ≥ 0:

p fD/2+1 (z) D≫1 e−1

= ,
pv [fD/2 (z)]1+2/D 1 + 2/D

T  −2/D D≫1 D e−1

= fD/2 (z) (pure Fermi sea).
Tv 2 ln z

1.2

1 D=1

2
0.8
3

10
p/pv

0.6
40

300
0.4

0.2

0
0 0.2 0.4 0.6 0.8 1
T/Tv
Ideal FD gas: phase diagram in D → ∞ [tln74]

Equation of state:

kB T, T > Tc (ideal MB gas)
pv =
kB Tc , T < Tc (pure Fermi sea)

Transition temperature:

Λ . h2
kB Tc = , Λ= .
e 2πm

80
70
60
50
40
30
20
10
00
1 20
2 15
v 3 10
4 5 kBT
5 0
[tex121] Stable white dwarf
Consider a burnt-out white dwarf star. For simplicity we assume that it consists of equal numbers
N of electrons, protons, and neutrons. The electrons form a fully degenerate, nonrelativistic Fermi
gas that prevents the star from collapsing into a neutron star or a black hole.
(a) Under the assumption that the kinetic energy is predominantly due to the electrons and that
the potential energy is predominantly gravitational in nature, show that the total energy of the
star depends on N and R (radius) as follows:
2/3
3~2 N 5/3 12 2 N 2

9π
E = Ekin + Epot = − m G ,
10me 4 R2 5 n R

where me , mn are the electron and neutron masses, and G is the universal gravitational constant.
(b) Using a star of solar mass, m ' 1.99 × 1030 kg, find the radius Rwd in units of the solar radius,
R ' 6.96 × 108 m.

Solution:
[tex122] Unstable white dwarf
Consider a burnt-out white dwarf star of the same composition as described in [tex121] but with
N so large that most of the electrons are ultrarelativistic,  ' cp = ~kc, in the fully degenerate
state.
(a) Under similar assumptions as in [tex121] show that the expression of the total energy now reads
4/3
N 4/3 N2

~c 9π 12
E = Ekin + Epot = − m2n G ,
3π 4 R 5 R

where c is the speed of light.

(b) Find the critical mass in units of the solar mass, mc /m , beyond which this star is unstable
and thus prone to a gravitational collapse into a neutron star or a black hole.

Solution: