Sei sulla pagina 1di 7

PHYS 352 Homework 2 Solutions

Aaron Mowitz (1, 2, and 3) and Nachi Stern (4 and 5)

Problem 1
The purpose of doing a Legendre transform is to change a function of one or more variables into a
function of variables conjugate to the original ones. In other words, we want to trade the dependence
on one set of variables for another. The entropy and its differential as a function of energy and
volume are written as
E pV
S(E, V ) = +
T T
   
∂S ∂S 1 p
dS = dE + dV = dE + dV
∂E V ∂V E T T
∂S
E = E

To find the Legendre transform with respect to energy J, we subtract ∂E V T from the
entropy:
E F
J ≡S− =−
T T
where we recall that F ≡ E −T S. To find dJ, we simply take the differential of the above expression
 
E p E
dJ = dS − d = dV + 2 dT
T T T

So we see that J is a function of T and V . This isn’t surprising, since the Helmholtz free energy
is a function of temperature and volume. Now to find the Legendre transform with respect to
both energy and volume Y , we subtract ∂V E V = pV
∂S

T from the entropy, in addition to what we
subtracted before. Equivalently, we can just subtract this from J:
pV F + pV G
Y ≡J− =− =−
T T T
where we recall G ≡ E − T S + pV = F + pV . To find dY , we again just take the differential, which
yields  
pV E + pV V H V
dY = dJ − d = 2
dT − dp = 2 dT − dp
T T T T T
where we use the definition of the enthalpy, H ≡ E + pV . So we see that Y is a function of T and
p. Again, this makes sense, since the Gibbs free energy is a function of temperature and pressure.

Problem 2
The distribution of velocities of a classical ideal gas is given by the Maxwell-Boltzmann distribution:
3/2 " #
m |v|2

m
f (v) = exp −
2πkB T 2kB T

1
This can be found in many ways (e.g. using symmetry considerations or the canonical ensemble),
and the derivation is in most standard textbooks, so we will not reproduce it here. We can use
this distribution to calculate the average speed. Since the M-B distribution only depends on the
magnitude of velocity, the integration is most easily done in spherical coordinates:
Z
h|v|i = |v| f (v) d |v| dΩv
3/2 ∞
mv 2
 Z  
m
= 4π v 3 exp − dv
2πkB T 0 2kB T
r
8kB T
=
πm
q
≡ hv i = 3km
BT
p
Note that this is different from both the root-mean-square speed, vrms 2 , and the
q
most probable speed, vp = 2km BT
, the speed at which the M-B distribution is a maximum.

Problem 3
(i) Since the Hamiltonian of a classical spin is H = −µz B = −µ0 B cos θ, the Boltzmann factor is
given by eµ0 Bβ cos θ . To find the canonical partition function, we integrate this over the sphere:
Z
Z = eµ0 Bβ cos θ dΩ
Z 2π Z 1
= eµ0 Bβ cos θ d (cos θ) dφ
0 −1
2π  µ0 Bβ  4π
= e − e−µ0 Bβ = sinh (µ0 Bβ)
µ0 Bβ µ0 Bβ

(ii) The average magnetization is given by


Z
1
hµz i = µ0 cos θeµ0 Bβ cos θ dΩ
Z

We want to show this equals T ∂∂B


ln Z
= 1 ∂Z
βZ ∂B . We see that
Z Z
∂Z ∂ µ0 Bβ cos θ
= e dΩ = µ0 β cos θeµ0 Bβ cos θ dΩ
∂B ∂B
If we divide this by βZ, we get exactly what we wrote down for the average magnetization.

(iii) To find the magnetic susceptibility, we first compute the derivative of the average magne-
tization with respect to B:
 Z 
∂hµz i ∂ 1 µ0 Bβ cos θ
= µ0 cos θe dΩ
∂B ∂B Z
Z Z
1 1 ∂Z
= µ20 β cos2 θeµ0 Bβ cos θ dΩ − 2 µ0 cos θeµ0 Bβ cos θ dΩ
Z Z ∂B
Z Z 2
β 2 2 µ0 Bβ cos θ β µ0 Bβ cos θ
= µ0 cos θe dΩ − 2 µ0 cos θe dΩ
Z Z

2
We now evaluate this at zero magnetic field. We note that Z|B=0 = 4π, which can be seen by
either setting B = 0 in the original definition of the partition function, or by using the fact that
limx→0 sinh
x
x
= 1. Using this, we find
" Z Z 2 #
∂hµz i 1 1
χ= =β µ20 cos2 θdΩ − µ0 cos θdΩ
∂B B=0 4π 16π 2

which is inversely proportional to T , since the quantity in brackets is independent of temperature.


The quantity in brackets above is the proportionality factor χ0 . The second term vanishes, since
the average of cosine over the sphere is zero. So we are left with

µ20
Z
1
χ0 = µ20 cos2 θdΩ =
4π 3

3
4) Fluctuations

i) Show that, in the canonical ensemble


𝜕𝐶𝑉
〈(Δ𝐸)3 〉 = 𝑇 4 ( ) + 2𝑇 3 𝐶𝑉
𝜕𝑇 𝑉

Let us first consider the LHS:

〈(Δ𝐸)3 〉 ≡ 〈(𝐸 − 〈𝐸〉)3 〉 = 〈𝐸 3 − 3𝐸 2 〈𝐸〉 + 3𝐸〈𝐸〉2 − 〈𝐸〉3 〉 = 〈𝐸 3 〉 − 3〈𝐸 2 〉〈𝐸〉 + 2〈𝐸〉3

Now, we also have the relation between moments of the energy and the partition function
𝜕𝑛 𝒵
〈𝐸 𝑛 〉 = (−1)𝑛 𝒵 −1
𝜕𝛽 𝑛

𝜕3𝒵 𝜕 2 𝒵 𝜕𝒵 𝜕𝒵 3
〈(Δ𝐸)3 〉 = −𝒵 −1 + 3𝒵 −2
− 2𝒵 −3
( )
𝜕𝛽 3 𝜕𝛽 2 𝜕𝛽 𝜕𝛽
This looks quite complicated, but consider the following
𝜕 ln 𝒵 𝜕𝒵
= 𝒵 −1 = −〈𝐸〉
𝜕𝛽 𝜕𝛽

𝜕 2 ln 𝒵 𝜕 −1
𝜕𝒵 −2
𝜕𝒵 2 −1
𝜕2𝒵
= (𝒵 ) = −𝒵 ( ) + 𝒵 = −〈𝐸〉2 + 〈𝐸 2 〉 = 〈(Δ𝐸)2 〉
𝜕𝛽 2 𝜕𝛽 𝜕𝛽 𝜕𝛽 𝜕𝛽 2
And very similarly:

𝜕 3 ln 𝒵 𝜕 −2
𝜕𝒵 2 −1
𝜕2𝒵 −1
𝜕3𝒵 −2
𝜕 2 𝒵 𝜕𝒵 −3
𝜕𝒵 3
= [−𝒵 ( ) + 𝒵 ] = 𝒵 − 3𝒵 + 2𝒵 ( ) = −〈(Δ𝐸)3 〉
𝜕𝛽 3 𝜕𝛽 𝜕𝛽 𝜕𝛽 2 𝜕𝛽 3 𝜕𝛽 2 𝜕𝛽 𝜕𝛽

𝜕〈𝐸〉
That is rather useful! Let us further note that 𝐶𝑉 = ( 𝜕𝑇 ) . One may change variables (𝑘𝐵 ≡ 1):
𝑉

𝜕 𝜕𝑇 𝜕 1 −1 𝜕 𝜕
= = (− 2 ) = −𝑇 2
𝜕𝛽 𝜕𝛽 𝜕𝑇 𝑇 𝜕𝑇 𝜕𝑇
Now we can write

𝜕 3 ln 𝒵 𝜕 2 〈𝐸〉 𝜕 𝜕〈𝐸〉
〈(Δ𝐸)3 〉 = − = = −𝑇 2
{−𝑇 2
}
𝜕𝛽 3 𝜕𝛽 2 𝜕𝑇 𝜕𝑇
𝜕 𝜕𝐶𝑉 𝜕𝐶𝑉
= 𝑇2 {𝑇 2 𝐶𝑉 } = 𝑇 2 {2𝑇𝐶𝑉 + 𝑇 2 ( )} = 𝑇 4 ( ) + 2𝑇 3 𝐶𝑉
𝜕𝑇 𝜕𝑇 𝜕𝑇

Which is what we wanted to show.

4
ii) Show that, in the grand canonical ensemble

𝜕〈𝑁〉
〈(Δ𝑁)2 〉 = 𝑇 ( )
𝜕𝜇 𝑇,𝑉

Although this proof can be done using a trick similar to the one shown on part (i), let us do it in brute
force, which is quite easy in this case. Using the relations of the partition function and the average
particle number

𝜕𝒵 𝜕2𝒵
〈𝑁〉 = 𝑇𝒵 −1 ; 〈𝑁 2 〉 = 𝑇 2 𝒵 −1 2
𝜕𝜇 𝜕𝜇
We may write

𝜕〈𝑁〉 𝜕 𝜕𝒵 𝜕2𝒵 𝜕𝒵 2
𝑇( ) = 𝑇 ( 𝑇𝒵 −1 ) = 𝑇 2 𝒵 −1 ( 2 ) − 𝑇 2 𝒵 −2 ( ) = 〈𝑁 2 〉 − 〈𝑁〉2 = 〈(Δ𝑁)2 〉
𝜕𝜇 𝑇,𝑉 𝜕𝜇 𝜕𝜇 𝑇,𝑉 𝜕𝜇 𝑇,𝑉 𝜕𝜇 𝑇,𝑉

And we got the expected relation.

ii) Compute √〈(𝚫𝑵)𝟐 〉 for an ideal gas.

The grand canonical partition function of an ideal gas is

𝛽𝜇
2𝜋𝑚 1.5
𝒵 = exp [𝑉𝑒 ( 3) ]
𝛽ℎ

Using this we calculate the average particle number

𝜕𝒵 𝜕 ln 𝒵 2𝜋𝑚 1.5
〈𝑁〉 = 𝛽 −1 𝒵 −1 = 𝛽 −1 = 𝑉𝑒 𝛽𝜇 ( 3 )
𝜕𝜇 𝜕𝜇 𝛽ℎ
Taking another derivative will give us the fluctuation

𝜕〈𝑁〉 2𝜋𝑚 1.5


〈(Δ𝑁)2 〉 = 𝛽 −1 ( ) = 𝑉𝑒 𝛽𝜇 ( 3 ) = 〈𝑁〉 → √〈(Δ𝑁)2 〉 = √〈𝑁〉
𝜕𝜇 𝛽,𝑉 𝛽ℎ

This result tells us that we should expect no fluctuations compared to the equilibrium particle number in
the thermodynamic limit:

√〈(Δ𝑁)2 〉
= 〈𝑁〉−1/2 → 0
〈𝑁〉

5
5) Adsorption Centers

i) Find fraction of occupied adsorption sites.

The partition function of one adsorption site is


1

𝒵1 = ∑ 𝑒 −𝛽(𝐸−𝜇)𝑛 = 1 + 𝑒 −𝛽(𝐸−𝜇)
𝑛=0

Different adsorption sites are independent, so the total partition function is


𝑁
𝒵 = (𝒵1 )𝑁 = (1 + 𝑒 −𝛽(𝐸−𝜇) )

We can relate the average adsorbed particle number to this partition function ,as we did in the previous
question:
𝑁−1
−1 −1
𝜕𝒵 𝑁𝛽𝑒 −𝛽(𝐸−𝜇) (1 + 𝑒 −𝛽(𝐸−𝜇) ) 𝑒 −𝛽(𝐸−𝜇)
〈𝑁〉 = 𝛽 𝒵 = =𝑁
𝜕𝜇 𝛽(1 + 𝑒 −𝛽(𝐸−𝜇) )𝑁 1 + 𝑒 −𝛽(𝐸−𝜇)
The fraction of adsorbed sites is thus

〈𝑁〉 𝑒 −𝛽(𝐸−𝜇) 1
𝑓= = =
𝑁 1 + 𝑒 −𝛽(𝐸−𝜇) 1 + 𝑒𝛽(𝐸−𝜇)

ii) Binding energy of 𝑶𝟐 molecules.

Apparently this part gave you guys a lot of trouble. To solve it efficiently it is useful to relate the
chemical potential to the given quantities in the problem.

We are given that 𝑓 = 0.5, which allows us to relate 𝐸 and 𝜇:

1 𝑒 −𝛽(𝐸−𝜇)
= → 𝐸=𝜇
2 1 + 𝑒 −𝛽(𝐸−𝜇)
The chemical potential of an ideal gas can be derived from the Sackur-Tetrode relation

𝑉 4𝜋𝑚 𝐸 3/2 5
𝑆 = 𝑘𝑁 ln [ ( 2 ) ] + 𝑘𝑁
𝑁 3ℎ 𝑁 2
3/2
𝜕𝑆 𝑉 4𝜋𝑚 𝐸 3/2 𝑁 3ℎ2 𝑁
𝜇 = −𝑇 ( ) = −𝑘𝑇 ln [ ( 2 ) ] = 𝑘𝑇 ln [ ( ) ]
𝜕𝑁 𝐸,𝑉 𝑁 3ℎ 𝑁 𝑉 4𝜋𝑚 𝐸

Note that for an ideal gas 𝑁/𝑉 = 𝑃/𝑘𝑇. Further, in room temperature, 𝑂2 molecules have translational
+ rotational degrees of freedom (linear molecule), so = 0.5(3 + 2)𝑁𝑘𝑇 = 2.5𝑁𝑘𝑇 . Using these facts
we may write

6
3/2
𝑃 3 ℎ2
𝐸 = 𝜇 = 𝑘𝑇 ln [ ( ) ]
𝑘𝑇 10𝜋 𝑚𝑘𝑇

The mass of a 𝑂2 molecule is 𝑚 = 5.4 ⋅ 10−26 𝑘𝑔. In STP conditions 𝑘𝑇 = 𝑘(273[𝐾]) ≈ 1/43 𝑒𝑉, and
the pressures is 𝑃 = 1𝐴𝑡𝑚 ≈ 105 𝑃𝑎. Plugging these numbers, the binding energy is
1 3 1 16.4
𝐸≈ ln [2.6 ⋅ 1025 × (2 ⋅ 10−22 )2 ] ≈ ln[7.8 ⋅ 10−8 ] ≈ − 𝑒𝑉 ≈ −0.38𝑒𝑉
43 43 43

Potrebbero piacerti anche