Thermal Physics Part 1: Steve Bramwell
Thermal Physics Part 1: Steve Bramwell
Thermal Physics Part 1: Steve Bramwell
Steve Bramwell
1
Contents
1 Thermodynamic Systems . . . . . . . . . . . . . . . . . . . . . . . . . 4
2 Equilibrium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
3 Thermal Contact . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
4 Mechanical Contact . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
5 Mass Contact . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
6 The Basic Problem of Thermodynamics . . . . . . . . . . . . . . . . . 6
7 Heat Capacity C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
8 The Final Temperature Problem . . . . . . . . . . . . . . . . . . . . . 8
9 Thermal Reservoir . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
10 Latent Heat . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
11 Thermal Conduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
12 Joule’s Paddle-Wheel Experiment (1843) . . . . . . . . . . . . . . . . 11
13 Interal Energy and the First Law . . . . . . . . . . . . . . . . . . . . 11
14 pV Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
15 Redefinition on Heat Capacity . . . . . . . . . . . . . . . . . . . . . . 13
16 The Case of an Ideal Gas . . . . . . . . . . . . . . . . . . . . . . . . . 14
17 Temperature Dependence on U and H . . . . . . . . . . . . . . . . . 15
18 State Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
19 Notes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
20 Thermodynamic Processes in an Ideal Gas . . . . . . . . . . . . . . . 17
21 Reversible Processes with a Fixed Parameter ( constant n) . . . . . . 17
22 Reversible Adiabatic Expansion / Compression . . . . . . . . . . . . 19
23 Interpretation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
24 Heat Engines and the Second Law . . . . . . . . . . . . . . . . . . . . 20
25 The Carnot Cycle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
26 Carnot’s Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
27 Kelvin’s Absolute Thermodynamic Temperature . . . . . . . . . . . . 23
28 Clausius Inequality . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
29 Entropy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
30 Law of Increase of Entropy . . . . . . . . . . . . . . . . . . . . . . . . 26
31 Equivalent Reversible Process . . . . . . . . . . . . . . . . . . . . . . 27
32 Statistical Interpretation of Entropy . . . . . . . . . . . . . . . . . . . 27
33 Temperature Dependence of Entropy at Fixed Pressure . . . . . . . . 28
34 The Fundamental Equation . . . . . . . . . . . . . . . . . . . . . . . 29
35 The Fundamental Equation for an Ideal Monatomic Gas . . . . . . . 30
36 Helmholtz and Gibbs Free Energies . . . . . . . . . . . . . . . . . . . 31
2
37 Natural Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
38 Maxwell Relations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
39 Thermodynamics – A Final Word . . . . . . . . . . . . . . . . . . . . 33
40 Kinetic Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
41 ‘Heuristic’ Argument for Gas Pressure . . . . . . . . . . . . . . . . . 34
42 Molecular Velocity Distribution . . . . . . . . . . . . . . . . . . . . . 37
43 Molecular Speeds Distribution . . . . . . . . . . . . . . . . . . . . . . 38
44 Mean Square Speed and Internal Energy . . . . . . . . . . . . . . . . 39
45 The Energy Distribution . . . . . . . . . . . . . . . . . . . . . . . . . 40
46 Equipartition Principle . . . . . . . . . . . . . . . . . . . . . . . . . . 40
47 Degrees of Freedom . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
48 Application to Molecular Energies . . . . . . . . . . . . . . . . . . . . 41
49 Recipe for Calculating Energy . . . . . . . . . . . . . . . . . . . . . . 42
50 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3
1 Thermodynamic Systems
The system is the part of the universe in which we are interested ( e.g. an object, a
bulb of gas, a human being, the Earth, a galaxy etc). The surroundings ( or thermal
surroundings) are the rest of the universe. The universe is the systems plus the
surroundings (Fig. 1.1).
Several systems of interest may be treated as a single composite system (Fig.
1.2). An isolated system is one that has no interaction ( exchange of energy, matter,
momentum, etc.) with other systems or the surroundings.
Surroundings
A B C
System Universe
Figure 1.2
Figure 1.1
2 Equilibrium
Consider an isolated composite system containing gas (Fig. 2.1). No matter how
complex the initial state, at long times the system will reach a state of equilibrium
with time independent macroscopic properties (Fig. 2.2). Two types of property:
• Intesive - property of part of the system is equal to that of the whole system,
e.g, p, T .
pA VA = nA RTA ; pB TB = nB RTB
4
pA pB
A B TA TB
VA VB
nA nB
3 Thermal Contact
If A, B are in thermal contact then at equilibrium TA = TB and A, B are said to be in
thermal equilibrium with each other. Thermal contact is allowed by diathermal walls
(Fig. 3.1) and prevented by adiabatic walls (Fig. 3.2). Diathermal walls allow the
transfer of energy in the form of heat ( thermal conduction, convection, radiation).
Adiabatic walls forbid it. At thermal equilibrium all heat flow ceases.
If two systems (A, C) are in thermal equilibrium with a third (B) then they are
in thermal equilibrium with each other. This is the zeroth law of thermodynamics
which ensures temperature can be consistently defined. Temperature is the property
that determines if systems are in thermal equilibrium with each other.
TA TB TA TB
TA = TB TA 6= TB
Figure 3.1 Figure 3.2
A B C TA = TB
⇒ TA = TC
TB = TC
Figure 3.3
5
4 Mechanical Contact
If A, B are in mechanical contact then at equilibrium pA = pB and A, B are said to
be in mechanical equilibrium with each other. Mechanical equilibrium is allowed by
movable walls (or flexible walls) and disallowed by immovable walls ( or inflexible
walls). Movable walls allow the transfer of energy in the form of work which is force
times displacement. All forces are balanced at equilibrium (pA = pB ).
Movable walls
pA pB
pA = pB
Figure 4.1 Figure 4.2
5 Mass Contact
If A, B are in mass contact then mass can be exchanged between them, and they
come to ‘mass equilibrium’. A wall permeable to matter allows mass contact ( Fig.
5.1); a wall impermeable to matter prevents it.
At mass equilibrium the gas densities ρ become equal (Fig. 5.2). Note that
equilibrium ( in these simple cases) involves homogeneity of intensive properties -
T, p, ρ, etc.
Permeable walls
ρA ρB
ρA = ρB
Figure 5.1 Figure 5.2
6
What is the new equilibrium state reached after the removal of
internal constraints in an isolated composite system (Fig. 6.1)?
Later in the course we will give a completely general answer to this question.
But to start off with one will consider replacing an adiabatic wall with a diathermal
one so that heat can be exchanged.
? ?
Figure 6.1
7 Heat Capacity C
System
T
Q
Figure 7.1: Energy added to the system in the form of heat (Q)
Say energy Q is added to the system in the form of heat (Fig. 7.1). The
temperature changes from T to T + ∆T and ∆T ∝ Q so that Q = C∆T . As a
differential,
Q d̄Q
C = lim =
∆T →0 ∆T dT
where the symbol d̄ will be explained later. Specific heat c is the heat capacity per
unit mass ( or per unit mole, etc):
Q = mc∆T
Specific heats are only approximately temperature independent. For gases we must
specify them for constant pressure or volume.
7
8 The Final Temperature Problem
Object A at temperature TA is placed in thermal contact with object B at temperature
TB . What is the final temperature (Fig. 8.1)?
We can formulate this in terms of the ‘basic problem of thermodynamics’.
Solution: If TA > TB , heat flows from A to B:
)
QB > 0
heat added to the system is positive.
QA < 0
and thus
CA TA + CB TB
Tf =
CA + CB
or equivalently
mA cA TA + mB cB TB
Tf =
mA cA + mB cB
which is a kind of ‘weighted averaged’ and can be used to compare heat capacities.
TA TB Tf Tf
Figure 8.1
9 Thermal Reservoir
Suppose
CB mB cB
= = 1.
CA m A cA
Then
TA + TB
Tf = .
1+
8
As → 0, Tf → TA .
A acts as a ‘thermal reservoir’ for B, or a ‘heat bath’. An object placed in contact
with it will adopt its temperature - this is how we control temperature!
Often the thermal surroundings are pictured as a thermal reservoir, and also as a
‘pressure bath’ which fix {p, T } for the system.
p, T, Surroundings
Figure 9.1
10 Latent Heat
s l
1 2
1 atm
g
T (K)
273 373
Figure 10.1: 1. Solid to liquid Lfus (melting) 2. liquid to gas Lvap (boiling).
At phase changes ( solid to liquid, liquid to gas) it takes finite heat to melt or
boil a given mass with no change in temperature.
Q = Lm or d̄Q = L dm .
9
T
Phase coexistence
100◦ C
1/c
0◦ C
333kJ 2260kJ Q
Figure 10.2
11 Thermal Conduction
Consider a rectangular parallelepiped (Fig. 11.1). We have
d̄Q dT Thot − Tcold
= −kA = +kA
dt dt l
at steady state. Here k is the thermal conductivity in Wm−1 K−1 . Suppose a hot
source is a cooling object at temperature T , heat capacity C. Then
d̄Q = −C dT
d̄Q dT
= −C .
dt dt
So
dT kA
= − (T − Tcold ).
dt Cl
kA
Objects cool according to e− Cl t . The ‘relaxation method’ for comparing heat capacities
or specific heats.
area A l
dQ
flux
dt
x
Thot Tcold
Figure 11.1
10
12 Joule’s Paddle-Wheel Experiment (1843)
The work done by weight w = mgh converted to heat in water W = Q, Q = C∆T
which implies
mgh = C∆T
This established the equivalence of heat and work. Doing work ( mechanical or
electrical ) gives a third method of measuring heat capacity - but this time on an
absolute scale.
∆T
Paddle wheel h
water
Figure 12.1
pivot
friction
11
involves a succession of equilibrium states. A reversible process is a quasistatic process
in which no dissipative (e.g. frictional) forces act (Fig. 13.1). The opposite is an
irreversible process (Fig. 13.2).
During a process, energy may be transferred between system and surroundings in
different ways. We consider heat and work. A statement of energy conservation:
∆U = Q + W
This is the first law of thermodynamics. Here Q the is heat added to the system, W
is the work done on the system, and ∆U is the change in the system’s interal energy.
U is extensive and a function of the thermodynamic variables, called a state
function:
U = U (p, T, n) = nu(p, T )
or nu(p, V )
where u is the internal energy per mole. For an ideal gas internal energy per mole
depends on one independent variable– T :
u = u(T ), U = nu(T )
Using differentials:
dU = d̄Q + d̄W.
Here d̄ denotes an inexact differential (Sect. 19).
14 pV Work
x
gas p, V
F
F0
Figure 14.1
12
friction). Then
d̄W = −F dx
= −F 0 dx
= −pA dx
= −p dV .
Note the sign on F acts to reduce V . Work is done on the gas when dx < 0 so
d̄W = 0. Note we have expressed d̄W in terms of the system properties p, V .
The expression works equally in expansion, so dV > 0 so d̄W < 0 – the gas
does work. We can call this “pV ” work. It applies to liquids and also solids under
‘hydrostatic’ pressure ( equal from all sides) – not just gases.
therefore
∂U d̄QV
= .
∂T V dT
So
∂U
CV =
∂T V
The heat capacity Cp is at constant pressure
d̄Qp
Cp = .
dT
The first law:
dU = d̄Q + d̄W
= d̄Qp − p dV
13
so
d̄Qp = dU + p dV = cp dT .
Now we can define the enthalpy
H = U + pV,
dH = dU + p dV + V dp .
At constant p: dH = dU + p dV = Cp dT . So
∂H
Cp = .
∂T p
H = nh(p, V ), U = nu(p, V ).
therefore
dU = CV dT =
= ncV dT .
dU = d̄Q + d̄W
CV dT = d̄Q − p dV
so
d̄Q = CV dT + p dV
and
d̄Qp ∂V
Cp = = CV + p . (16.1)
dT ∂T p
14
Now, for an ideal gas, pV = nRT , so V = nRT /p and
∂U nR
=
∂T p p
From 16.1,
nR
Cp = CV + p
p
and so
Cp = CV + nR
This is true for all gases at sufficiently low pressure. In fact, for a monatomic gas,
U = 32 nRT so
3 5
CV = nR, Cp = nR (heat capacity)
2 2
3 5
cv = R, cp = R (specific heat per mole)
2 2
and
d̄Qp ∂H
Cp = =
dT ∂T p
so Z T2
Cp dT = H(T2 ) − H(T1 ) = ∆H.
T1
15
H
cp
latent heat L = ∆H
T
solid liquid
Figure 17.1
18 State Functions
At fixed n, U = U (p, V ). Changes in a state function like U depend only on the start
and end points, not on the path or process:
∆U = U (pB , VB ) − U (pA , VA ).
R
But ∆U = Q + W and W does depend on the path, so Q = d̄Q must also depend
on the path (Fig. 18.1). Therefore Q and W are not state functions. We cannot
write Q(p, V ) or W (p, V ). We can discuss “energy in a system” but not “heat or
work in a system”. Heat and work are forms of energy transfer.
p
A
process A → B
R R V
area path
p dV = − d̄W = −W
Figure 18.1
16
19 Notes
In mathematics if f = f (x, y) (an ‘explicit function’ of x and y), then
∂f ∂f
df = dx + dy .
∂x y ∂y x
R
Here df is an “exact differential”: df is path-independent. Hence dU is an exact
differential of U (p, V ). But d̄Q and d̄W are not. They are inexact differentials –
hence the bar.
On (p, V ) diagrams we will represent reversible processes as full lines and irre-
versible processes as dashed lines.
p
p
irreversible
reversible
V
V
Figure 19.2: Only end points equilibrium
Figure 19.1: All states equilibrium states.
states.
17
Adiabatic walls
gas empty
gas
Figure 20.1
Isovolumetric: Z
W =− p dV = 0.
Isothermal: Z Vf
nRT Vf
W =− dV = −nRT ln .
Vi V Vi
Since U = 32 nRT = 32 pV for a monatomic ideal gas,
3
∆U = (pf Vf − pi Vi ).
2
We then use ∆U = Q + W to work out Q.
Isothermal case:
U = U (T ), so ∆U = 0; Q + W = 0, so Q = −W
and so
Vf
Q = +nRT ln .
Vi
In an expansion the gas absorbs heat and converts it all to work. In a compression
work is all converted to heat.
18
22 Reversible Adiabatic Expansion / Compression
Proof that pV γ is a constant for an adiabatic expansion/compression, where γ =
Cp /CV . From the first law:
dU = d̄Q + d̄W,
dU = d̄W (adiabatic),
CV dT = −p dV (ideal gas),
CV dT + p dV = 0 (reversible).
Since pV = nRT ,
dV
CV + nRT = 0.
V
Dividing by T ,
dT dV
CV + nR = 0.
T V
As Cp − CV = nR,
dT dV
CV + (Cp − CV ) = 0.
T V
This becomes, by γ = Cp /CV ,
dT dV
+ (γ − 1) =0
T V
Integrating
ln T + (γ − 1) ln V = K 00
so
T V γ−1 = K
where K is a constant. Using T = pV /nR,
p
· V · V γ−1 = pV γ = K
nR
for fixed n. For a monatomic ideal gas: CV = 32 nR, Cp = 52 nR, so
5
γ=
3
and so
pV 5/3 = K.
19
p adiabat adiabat
TA isotherm
pV δ = T V δ−1 = K
TB isotherm
V
Figure 23.1
23 Interpretation
Adiabats are steeper than isotherms (Fig 23.1). The gas cools on adiabatic expansion
and gas warms on aiabatic compression. Since U = U (T ) ( increasing function),
∆U < 0 for an expansion and by the first law ∆U = Q + W with Q = 0,
(
W < 0 for an expansion
∆U = W
W > 0 for a compression.
Therefore, for an adiabatic expansion (W < 0) internal energy is converted to work.
For an adiabatic compression work is converted to internal energy.
Work
Cold
Figure 24.1
Heat flows from hot to cold. A heat engine (Fig 24.1) (or engine, steam engine,
etc.) diverts some of this energy to do work. Note the non-standard sign convention
in the figure. Question: how efficient can a heat engine be?
20
Hot
Hot
Qh
cyclical Qh
process cyclical
W process
W
Qc
Qc
Cold
Cold
Figure 24.2: Qh heat extracted from hot
reservoir per cycle, Qc heat delivered to Figure 24.3
reservoir per cycle, W work done per cycle
Carnot (1824) realised that the most efficient possible engine must be reversible.
Otherwise friction diverts energy straight into the surroundings and misses an op-
portunity to do work. A Carnot engine is a hypothetical heat engine that (Fig.
24.2)
• has 2 thermal reservoirs (hot/cold)
• operates in a cycle with a working substance
• is reversible
An engine can be run in reverse to divert heat from cold to hot (Fig. 24.3)!
21
p isotherm
Qh → W
adiabat
W → ∆V Th
adiabat
∆V → W
Tc
isotherm
W → Qc
V
Figure 25.1
26 Carnot’s Theorem
Hot Hot
Qh Qh
W = Qh
Qc = Qh
Cold Cold
Historically this implied the second law. Here we postulate the second law and
derive Carnot’s theorem. Second law statements:
The application to cyclical processes is shown in Fig. 26.1 and Fig. 26.2. It can
be shown the two statements are equivalent. Suppose we have two identical Carnot
22
engines, one driving the other in reverse (Fig. 26.3). The net effect is |Qh | = |Q0h |.
Now suppose that we have an engine that is more efficient than a Carnot engine (Fig.
26.4) it supplies the same work but extracts less heat Q00h to do so. The net effect is
Qh − Q00h > 0 and Qc − Q00c > 0 which is in violation of the Clausius statement!
Carnot’s theorem:
No engine operating between two thermal reservoirs can be more
efficient than a Carnot Engine operating between the same two reservoirs
Corollary: all Carnot (reversible) engines operating between (Tc , Th ) are equally
efficient.
There is a fundamental limit on the efficiency of heat engines that depends on
(Tc , Th ) and not on the details of engineering or the type of working substance:
Hot
Qh Q0h
W
Qc Q0c
Cold
Figure 26.3
23
Hot
Q00h Q0h
W
Q00c Q0c
Cold
Figure 26.4
Qh > 0 free Qh = 0
Q=0 1 1
Q=0
4 4
2 2
Q=0 Q=0
3 3
Qc < 0 Qc < 0
28 Clausius Inequality
Consider two Carnot cycles (Fig. 19.1 and 28.2), one with the isothermal expansion
replaced by an adiabatic free expansion ( irreversible). Use conventional signs:
Qh Th
− =
Qc Tc
and so
Qh Qc
+ = 0.
Th Th
Therefore X Qi
= 0.
i
Ti
24
Clausius argued that any cycle can be represented as the sum of Carnot cycles (Fig.
28.3). Making the sum an integral:
I
d̄Q
≤0
T
which is the Clausius inequality for a system doing work. The equality applies to a
reversible cycle.
29 Entropy
Now consider a reversible process (Fig. 29.1) ( d̄Qrev )
Z f Z i
d̄Qrev d̄Qrev
+ = 0,
i T f T
| {z } | {z }
Path 1 Path 2
Z t Z t
d̄Qrev d̄Qrev
− = 0,
i T i T
so Z t Z t
d̄Qrev d̄Qrev
=
i T path 1 i T path 2
and so the integral is path independent! There exists a state function for which
changes are given by: Z f
d̄Qrev
∆S = Sf − Si =
i T
and S is called the entropy. We can write
d̄Qrev
dS =
T
where dS is an exact differential.
Figure 28.3
25
p
path 2 i
path 1
f
V
Figure 29.1
dS ≥ 0.
This answer the ‘basic problem’ (Sect. 6). The final equilibrium state in an isolated
system is that which maximises the entropy. Extending the isolated system to
encompass the universe:
dS universe ≥ 0
for an real process (Clausius).
26
p irreversible
i
reversible
f
V
Figure 30.1
27
2V
V
V → 2V
∆U = 0
Figure 32.1
Figure 32.2
Hence from Cp (T ) measurements S(T ) can be found (Fig. 32.3). There is experimental
evidence that S(T = 0) = 0. Hence S(T ) can be put on an absolute scale ( unlike
U (T ) or H(T )). Also, S is extensive so we can define the molar entropy s = S/n. In
general, s(gas) > s(liquid) > s(solid) > s(soft solid) > s(hard solid). E.g. for lead,
s = 15.5JK−1 mol−1 and for diamond s = 0.6JK−1 mol−1 .
28
S
T
phase changes
Figure 32.3
which is our final and best (?) definition of temperature. The fundamental equation
can also be written
S = S(U, V ).
All thermodynamic relations can be derived from this, but the actual form of the
equation depends on the system. In practice 34.2 is more useful.
29
35 The Fundamental Equation for an Ideal Monatomic Gas
From the first law
d̄Q = dU − d̄W,
T dS = dU + p dV
dU p
dS = + dV .
T T
Using the equations of state pV = nRT , U = 32 nRT and so
3 dU dV
dS = nR + nR .
2 U V
Integrating we have
3 U V
S = nR ln + nR ln + S0
2 U0 V0
where U0 , S0 , and V0 define a reference state. This is the entropy fundamental equation
for an ideal monatomic gas. Rearranging:
−2/3
U V
= e(2/3nR)(S−S0 ) .
U0 V0
Application:
∂U 2
=T = U
∂S V 3nR
and so
3
U = nRT.
2
Also,
∂U 2
= −p = − U
∂V S 3V
and so
nRT
.p=
V
Now, instead of two equations of state, we have one single fundamental equation for
the gas.
30
36 Helmholtz and Gibbs Free Energies
Consider a finite process in a system in contact with a thermal reservoir at temperature
T . From second law, ∆Ssys + ∆Ssurr ≥ 0. But as T = Tsurroundings is fixed the addition
of heat to the surroundings is reversible:
Z
d̄Qrev Qsurr
∆Ssurr = = .
T T
∆Usys
∆Ssurr = − .
T
Then
∆Usys
∆Ssys − ≥0
T
or
∆[Usys − T ∆Ssys ] ≤ 0
for the system with constant T . Define F = U − T S as the Helmholtz free energy and
then ∆F ≤ 0. This is the system focused version of the second law for constrained T
and V .
The case of fixed pressure:
and so
∆H
∆Ssurr = − .
T
Thus
∆[H − T S] ≤ 0
for a system with T constrained. Define G = H − T S and then ∆G ≤ 0. This is the
Gibbs Free energy. It is the system focused version of the second law for fixed p and
T.
We could also write
∆G = ∆(U + pV − T S)
= ∆U + p∆V − T ∆S
= ∆F + p∆V
= ∆F + W
31
and so ∆G ≤ 0 implies ∆F + W ≤ 0 or
W ≤ −∆F.
The maximum work available from a process ( at constant T ) is −∆F , not −∆U .
Hence the name “free” energy – the energy available to do work.
37 Natural Variables
In W = −p dV we have to express p as a function of V to integrate. Similarly, in
dU = T dS − p dV , S and V are ‘natural variables’ but usually we would prefer (V, T )
or (p, T ).
Start with dU = T dS − p dV , and substitute:
F = U − T S,
dF = dU − T dS − S dT
and so
dF = −S dT − p dV
so T and V are the natural variables. For the Gibbs free energy
G = H − T S = U + pV − T S,
dG = dU + p dV + V dp − T dS − S dT
and so
dG = V dp − S dT
and here p and T are the natural variables.
These are useful equations when we required natural variables V, T and p, T ,
respectively.
Transformations like
F = U − T S, G = H − T S, H = U + pV
32
38 Maxwell Relations
If we have an exact differential of f (x, y)
∂f ∂f
df = dx + dy = A dx + B dy
∂x y ∂y x
dU = d̄Q + d̄W,
d̄Q
dS ≥ .
T
For reversible ( or equivalent reversible) processes we can write dS = d̄Qrev /T and
derive ( at fixed n):
dU = T dS − p dV .
To be clear - this equation is true for all changes between equilibrium states, not just
reversible ones, because it follows from U = U (S, V ).
Using Legendre transforms like H = U + pV , F = U − T S, G = H − T S, we can
derive equations like
dF = −p dV − S dT
etc. From these, we can get other usual equations. Finally, note the similarity between
d̄Qrev = T dS and d̄W = −p dV . Just as pressure is the thermodynamic variable
conjugate to volume, so temperature is the thermodynamic conjugate to entropy.
33
40 Kinetic Theory
From thermodynamics, the details of any system are inserted by specifying either
the form of the ‘fundamental equation’ or ( equivalently) two equations of state. For
example, for an ideal gas ( monatomic)
3
pV = nRT, U = nRT.
2
Let’s write these in the form of a ‘mechanical’ equation
2
pV = U
3
and a ‘thermal’ equation
3
U = nRT
2
and try to derive these from a microscopic theory of gas ‘molecules’ ( strictly atoms
in this case).
l
c1
A
Figure 40.1
34
• There are negligible intermolecular interactions,
N 2m hc2 i 1 N m hc2 i
F̄ = =
6 l 3 l
so
1 nm hc2 i 1 nm hc2 i
p = F/A = = .
3 l3 3 V
Finally
1
pV = nm c2 .
3
But U is the total energy which is N times the average kinetic energy per molecule.
So
1
U = nm c2
2
or
2
pV = U.
3
We have derived pV = 3 U , but not U = 2 N kT. However, if we accept U = 32 N kT
2 3
3
hEi = kT
2
where hEi is the average kinetic energy per molecule. We have
1
2 3
m v = kT
2 2
or
1
2 3
m vx + vy2 + vz2 = kT.
2 2
Now because gas is isotropic
2
2
2 kT
vx = vy = vz =
m
35
and
1
hEx i = hEy i = hEz i = kT
2
where Ex = 21 mvx2 etc.
To calculate averages we need a probability distribution. Here we give a simple
argument for the Boltzmann distribution, based on S = k ln Ω. Imagine N atoms,
each with 2 energy levels E = 0 and E = ε. With n exited atoms U = nε. Hence the
probability an atom is excited is P (E = ε) = n/N . Now, excite one extra atom, n
becomes n + 1 and
N! N!
Ωbefore = ; Ωafter =
n!(N − n)! (n + 1)!(N − n − 1)!
so
Ωafter N −n N −n N
= ≈ = −1
Ωbefore n+1 n n
if n 1. So
N
∆S = k ln Ωafter − k ln Ωbefore = k ln −1 .
n
Thus, as ∆U = ,
∆U ε
T = =
∆S k ln(N/n − 1)
so that
N
− 1 = eε/kT
n
and
n 1 e−ε/kT eE1 /kT
= = = P
N 1 + eε/kT 1 + e−ε/kT ie
−Ei /kT
where E0 = 0, E1 = ε. Thus
n e−Ei /kT
= P (E = Ei ) = P −Ei /kT .
N e
i=0,1
e−Ei /kT
P (E = Ei ) = P −Ej /kT .
je
36
This is the Boltzmann distribution for the probability of occupation of the ith level
out of a set of j levels. More simply
Pi ∝ e−Ei /kT
or
Pi = C(T )e−Ei /kT
where C(T ) is a constant that normalises the distribution.
Armed with the Boltzmann distribution, we will go on to prove our second equation
of state
3
U = nRT
2
vz
v
vz
vx vy
vx
vx
Figure 41.1
v = (vx , vy , vz )
37
2 2 2
P (vx , vy , vz ) = ce−(1/2)m(vx +vy +vz )/kT
The constant c can be found by ‘normalising’ the distribution ( density function),
such that the integral over all ‘velocity space’ is unity:
Z
p(vx , vy , vz ) dvx dvy dvz = 1
all space
so that m 3/2
c=
2πkT
The physical interpretation is p(vx , vy , vz ) dvx dvy dvz proportionla to the number of
molecules with velocities in the range (vx , vy , vz ) to (vx + dvx , vy + dvy , vz + dvz ).
This is the “Maxwell-Boltzmann distribution of molecular velocities” (Fig. 42.1).
p(vx )
vx
Figure 42.1: Gaussian or normal curve.
38
so that 1/2
0 2 m 3/2
c = .
π kT
A fraction p(s) ds of molecules have speeds in the range s to s + ds. This is the
“Maxwell speeds distribution” (Fig. 43.2).
vz p(s)
most probable speed (mode)
mean speed hsi
vy root mean square speed hs2 i
vx
s
Figure 43.1: 4πs2 ds ‘velocity states’ share
the same speed. Figure 43.2
So
2
s = 3kT /m
Now energy U = N × ( 12 m hs2 i) and thus
3 3
U = N kT = nRT
2 2
as required. We have finally proved both of the ideal gas equations of state, assuming
that S = k ln Ω and the concept of ‘velocity states’ ( later justified by quantum
mechanics).
39
45 The Energy Distribution
Is the energy distributed as broadly as the speed?
1 ds 1
E = ms2 ; =√ ; s2 = 2E/m
2 dE 2mE
and so
dE −1
P (E) = p(s) = c 2E √ 1 e−E/kT .
0
ds m 2mE
Thus
P (E) = c0 E 1/2 e−E/kT .
This is for one molecule, it is a broad distribution, but U = E1 + E2 + E3 . . . + En .
For N molecules, the curve sharpens with respect to its most probable value, a
consequence of the central limit
√ theorem of statistics (Fig. 45.1). In fact, the ‘width’
∆U is given by ∆U/ hU i ∼ 1/ N .
p(U )
1 molecule
N molecules
U/Umax
Figure 45.1
46 Equipartition Principle
By integrating the speeds distribution we showed that the average kinetic energy of a
molecule is 32 kT . If we write
1
47 Degrees of Freedom
Consider a molecule of M atoms. The number of degrees of freedom ( d.o.f.) is
the number of coordinates needed to describe the positions of all the particles, i.e.
(x, y, z) × M = 3M .
• The centre of mass needs three coordinates, so there are three translational
d.o.f.
• For a linear molecule, rotation requires two angles (θ, φ) angles, so three
rotational d.o.f.
• For a nonlinear molecule, rotation requires one more angle, so three rotational
d.o.f.
• Translations
41
• Rotations
• Vibrations
Figure 49.1
42
Linear Nonlinear
Translational 3/2kT (3/2)kT
Rotational kT (3/2)kT
Vibrational (3m − 5)kT (3m − 6)kT
Example of water vapour. We have M = 3 so 9 d.o.f.We have
3 3
hEtrans i = kT, Utrans = nRT
2 2
3 3
hErot i = kT, Urot = nRT
2 2
hEvib i = 3kT, Uvib = 3nRT,
So that U = 6nRT . Also, CVtrans = 32 nR, CVrot = 32 nR, CVvib = 3nR. An experiment
is shown in Fig. 49.2.
CV /nR
vib
3
rot
3
2
trans
T
Room
temp
Figure 49.2
The vibrational d.o.f. are dormant at room temperature. This is not in classical
physics! Errors comes from approximating discrete energy levels as a continuous
spectrum.
According to Pi ∝ e−Ei /kT , level will accept no energy if kT Ei .
50 Discussion
Such breakdowns of classical equipartition were discovered in specific heats of gases,
crystals ( ‘Dulong-Petit Law’) and thermal radiation ( Planck spectrum). It con-
tributed to the birth of quantum mechanics.
43
In quantum mechanics, the more ‘confined’ a motion, the larger its ‘quantum’
(energy-level spacing) and the higher a temperature is required to ‘turn on’ the specific
heat.
Specific heat of internal motions of an atom ( electronic, nuclear) is measurable
only at T 300K. The complexity of U (T ) for the ‘average’ molecule explains why
physicists were so slow to embrace the chemists’ atomic model.
Final note:
hkinetic energyi ∝ T
is only true of an ideal gas- not general!
44