Ex -1.1. Calculate the ensemble-averaged kinetic energy of a 3D gas of N free electrons at zero temperature.

Let the N free electrons, each of mass m, be conned to a large

1 cubic volume V = L3 . The Schrödinger equation says that the
energy of a single such electron (which doesn't interact with any of the others) is,

2
1 −~2 (ik) ~2 k 2
ψ(x) = √ e−ik•x → ε = = ; ψ ∈ V = L3 → ψ(x + Lx̂) = ψ(x) → kL = 2πn; (1)
V ε·ψ=
−~2 2
∇ ψ=K̂ψ
2m 2m
2m

Furthermore, let this set of identical electrons be in thermal equilibrium. The ratio of the number of energy-ε − µ in their (n + 1)th
quantum state of excitation to the number in the nth quantum state at temperature T is,

Nn+1 /Nn = e−β(n+1)(ε−µ) /e−βn(ε−µ) = e−β(ε−µ) ; β −1 ≡ kB T ;

The formal statement of the Pauli-exclusion-principle for the above-mentioned electrons in thermal equilibrium is for a Gibbs-
Boltzmann factor
2 (1.2) to be assigned to the occupation-probability of n = 0 and n = 1, and for zero-deniteness to be to be assigned
to the occupation-probability of n > 0.
Combining Eq. (1) and (1.3), we calculate the ensemble-averaged (kinetic) energy using a discretization-approximation,

R ∞ ~2 k 2 3 3
( 2m )f · L(2π)
d k
D E P∞ E f R kF 2 3 5
n=0 n 0 3 T '0 ~2
0
k ·d k ~2 4π( 15 kF − 15 05 ) ~2 3 2 3 ~2 kF 2 3
K̂ = P∞ ' R∞ 3 3
= k
= 1 3 1
= kF = ≡ εF
f L d k 2m 2m − 3) 2m 5 5 2m 5
R
n=0 0
f· 3
F 3
d k 4π( k
3 F 3 0
(2π) 0

(1.4) 091 - 05-01-d - the Boltzmann factor: d

Consider a universe of total energy UT , a system A of energy UA , and a reservoir of energy UR . [totalenergy] = [bath − energy] +
[system − energy] → UT = UR + UA ; (1.1)
Let the number of arrangements permitting the partitioning of energy described above be a function W , W = W (UA , UR ) = 3
W (UA )W (UR ) = W (UA )W (UT − UA ) = W (UA , UT ); (1.2)
4
The existence of W as given by (1.2) immediately implies Boltzmann's state variable S = S(W ) = kB ln W .
5 ∂U


First law of thermodynamics, where we re-express δ̄Q = T · dS dU = dU (S, V ) = δ̄Q + δ̄W = T · dS + (−P · dV ) = dS +
∂S V
∂U


∂V S dV ; (1.3)
−1
The thermodynamic identity implied by (1.3) then states, dU (S, V ) = T · dS − P · dV ↔ dS(U, V ) = T dU + P · dV =
∂S ∂S



∂U V dU + ∂V U dV ; (1.4)
6 that −1
This implies T −1 = ( ∂U
∂S
)V = ( ∂U
∂S )V . Combining this statement with the denition of entropy as an equipartitioning
amidst phase-space-coordinate combinations corresponding to a given energy,
       
1 ∂S ∂(kB ln W ) canonicalensemble(reservoir) ∂(kB ln WR ) Gibbsrulefixingstate d(kB ln WR )
= = = = ;
T ∂U V ∂U V ∂UR V,T dUR V,T

(1.5)
That is, the Gibbs rule xing the state allows us to replace a partial-dierential with a total dierential because the reservoir is
dened to have a constant temperature. This, in turn, allows integration,

dUR
d ln WR = ↔ WR = W0 eUR /kB T = WR (UR )
kB T
(1.6)
The probability distribution (1.6) is higher about the reservoir energy, for the probability of Ei is,

W (Ei ) = WR (UT − Ei ) = W0 e(UT −Ei )/(kB T ) ;

(1.7)
The basic postulate of statistical mechanics is that all accessible quantum states are equally probable. The total number of
accessible quantum states of the combined system is

X X
Ntotal = W (Ej ) = W0 eUT /(kB T ) e−Ej /(kB T ) = [totalnumberofaccessiblequantum − states] ;
j j

1 Large" in this case means L  λ = 2π/k , where λ is the Fermi wavelength.

F F F
2 The Gibbs-Boltzmann factor is derived in 091 - 05-01-d - the Boltzmann factor (it has a surprisingly-large amount of underlying physical statements
inside of it!).
3 Example: Suppose that, out of all the quantum states of system A in environment Ĥ , a particular quantum state is chosen, ψi . This quantum state is an
eigenstate of the energy operator with eigenvalue Ei , i.e.,Ĥψi = Ei · ψi . Hence UA = Ei and WA (Ei ) = 1 since one quantum state is chosen. With system
A in quantum state ψi the total number of quantum states for the combined system, using (1.2), is W = W (UA , UR ) = W (Ei , UR ) = W (Ei )W (UT − Ei ) =
1 · W (UT − Ei ) = W (UT − Ei ).
4 See the work done in 068 - 04-01-d - Ergodic Hypothesis of Boltzmann and Entropy.
5 Proof: see 025 - 02-01-a - the entropy as a thermal integrating factor.
6 Neither (1.3) nor (1.4) by themselves imply the ability to naïvely" take the reciprocal of the partial-dierentials. However, the same variable V is held
constant"-is this what allows for such a naïvely" algebraic machination?
(1.8)
The sum is over all quantum states of system A. The classical probability that the system is in the quantum state characterized
by energy Ei is equal to the number of quantum states associated with Ei from (1.7) divided by the total number of states (1.8),

W0 e(UT −Ei )/(kB T ) e−Ei /(kB T ) e−Ei /(kB T )

 
W (Ei ) Boltzmann
pi = p(Ei ) = = = ≡ ≡ ;
W0 eUT /(kB T ) j e−Ej /(kB T ) −Ej /(kB T ) distribution
P P
Ntotal je Z

(1.9)
The occupation-expectation f of the distribution amidst the microstates of energy indexed by ε at macroscopic at temperature T
is,

P1 P1
nNn ne−n(ε−µ)/kB T 0 · e0 + 1 · e−(ε−µ)/kB T e−(ε−µ)/kB T 1 T '0
f = Pn=0
1 = Pn=0
1 = 0 −(ε−µ)/k T
= −(ε−µ)/kB T
= (ε−µ)/k T ' Θ(µ − ε)
n=0 Nn n=0 e
−n(ε−µ)/k B T e + e B 1 + e e B + 1

(1.3)
Alternate approach: use the grand potential,

1
−∂ΦG
 
(10.2.3b) −d − − d X −β(εk −µ) n d   1
nk = k = (−kB T ln Ξ) = kB T ln (e ) = kB T ln 1 + e−(εk −µ)/kB T =
(9.9.8a) ∂µ T,V (9.9.8b) dµ (10.2.1) dµ (10.2.2) dµ (10.2.3b) eβ(εk −µ)
n=0

Ex -1.2. (a) Derive a relation connecting the pressure and volume of an electron gas at 0 K.
Let us imagine the N electron gas to be conned to a volume V.
An experiment is performed to add an increment of internal energy dU by doing an amount of work δ̄W = −P · dV upon the
gas adiabatically (dS = 0 = δ̄Q/T where T = 0+ K ), so that the addition of dU is eected by nothing but this work. The 1st law of
thermodynamics gives the pressure as a derivative of internal energy U with respect to volume V over the course of this adiabatic
dS = 0 compression,

     
∂U ∂U ∂U
dU = dUS,V = dS + dV = δ̄Q + δ̄W = T · dS + (−P · dV ) = T · 0 − P · dV → P = −
∂S V ∂V S ∂V S

It was also found

7 that the internal energy of an electron gas having Fermi energy εF at T = 0K was U = 35 N εF . Imagining a
2π 3
k -space sphere of radius kF  2π/L wherein the N allowed levels are divided up into k -space unit cells of volume (
L ) , yielding the
Fermi energy as a function of state-variables N, V
2/3
2deg 34 πkF 3
r r
~2 kF 2 ~2

1 3 3 N N N
3π 2
3 3
N= 3 = L kF ↔ kF = 3π 2 3 = 3π 2 → εF = =
(2π/L) 3π 2 L V 2m 2m V

(1.2)
Combining (1.1) and (1.2), the adiabatic pressure measured by the above-described experiment
8 is,

2/3
~2 ∂(V −2/3 )
 2
3 1 ~2 2 ~2 k
        
∂U 3 2 2/3 3 ~ 2 2/3 − − 2 −5/3 2N 1 3
P =− = N (3π N ) − = N (3π N ) V = N 3π = N
∂V S 5 2m ∂V S 5 2m 3 5 3m V V 5 3V 2m

(1.3)
(b) Calculate the bulk-modulus of the electron gas.
Bulk modulus B σ = Bε = −P , where σ = −P is9 the pressure we calculated in (1.3) and ε = ∆`/` = ∆V /V . Since
appears in
B is now being measured by applying a P = σ and measuring ε, all other possible thermodynamic variables are being held constant,
and the total dierentials dP and dε are given solely by σ = Bε = −P , wherein we eventually use (1.1) and (1.3)
       
σ dσ −dP dP 2 d −1 2 dU −1 −2 2 ∂U U 2 U 2
B= = = = −V = −V (U V ) = −V V + U (−1)V =− − =− −P − = P
ε dε dV /V dV 3 dV 3 dV 3 ∂V S V 3 V 3

(1.4)
(c) Estimate for potassium, using Table 1, the value of the electron gas contribution to the bulk modulus.
10 U 10 35 N εiF 30 2 2 Zi i
Bie− = = = ni εiF = nεiF = ε
9 V 9 V 45 3 3 vi F
7 See the reasoning in 157 - pr 06-01 - Kinetic energy of Electron Gas.
8 Example of a pressure-measurement: The Bourdon pressure gauge uses the principle that a attened tube tends to straighten or regain its circular form
in cross-section when pressurized. A attened thin-wall, closed-end tube is connected at the hollow end to a xed pipe containing the uid pressure to be
measured. As the pressure increases, the closed end moves in an arc, and this motion is converted into the rotation of a (segment of a) gear by a connecting
link that is usually adjustable.
9 The sign-dierence in the statement σ = −P corresponds to σ being dened with the material (the electron gas) considered as the system, verses P
being dened with the universe/environment (the isothermal-bath made reference to when the Boltzmann factor was derived (i.e., when we drew upon the
results of 157 - pr 06-01 - Kinetic energy of Electron Gas)) considered as the system.
(1.5)
In (1.5),
Zi
22 e−
is the number of atoms per unit cell, and vi is the volume of the unit cell. For potassium, εK
F = 2.12eV , nK = 1.40 × 10 cm3 ,
vK = vBCC,K = aK 3
, aK = 5.328 × 10 −10
m and ZK = ZBCC = 2, yielding10 ,

e− 2 2 −19 J e− 2 2 −19
BCu = nK εK 28 e−
F = (1.40×10 m3 )(2.12×1.602×10 J) = 3.17×109 3 ; BLi = nLi εLi 28 e−
F = (4.70×10 m3 )(4.72×1.602×10 J) = 2.37
3 3 m 3 3
(1.6)

Ex -1.3. volume-elements of momentum-space dD k in dimensionalities D = 0, 1, 2, 3 are

d0 k
= 2πδ(ε − ε0 )
dε √
d1 k
r
1 d 2mε 1 m
= =
dε ~ dε ~ 2ε
2
√ r
d k dk 2mε 1 m m
= 2πk · = 2π · = 2π 2
dε dε ~ ~ 2ε ~
√
d3 k
r
dk 2mε dk 2mε 1 m 4πm 2mε
= 4πk 2 · = 4π 2 · = 4π 2 · = 2
dε dε ~ dε ~ ~ 2ε ~ ~

in zero dimensions (in which n is dimensionless since there are n electrons per no volume"),
+∞ +∞  
2πδ(ε − ε0 ) dε
Z Z
N dε 2 2
n= = 2deg gε fε< = = ↔ µ = µ(T ) = ε0 − k B T ln − 1
V 2π e(ε−µ)/kB T + 1 π e(ε0 −µ)/kB T + 1 n
−∞ −∞

in one dimension,

+∞ +∞
1
pm p Z∞ Z∞ −1/2
2ε Θε ε−1/2
Z Z r
N dε dε m/2 1 m x
n= = 2deg gε fε< = ~
= dε = dx
V 2π e(ε−µ)/kB T + 1 π π~e−βµ eβε + eβµ π~e−βµ 2β ex + a
−∞ −∞ 0 0

in two dimensions,

+∞ +∞
m
2π π~ 2 Θ(ε)
Z Z
N dε dε m 2
n= = gε fε< = = kB T ln(1 + eµ/kB T ) ↔ µ(T ) = kB T ln(eβnπ~ /m − 1) (*)
V 2π e (ε−µ)/k B T + 1 2π π~2
−∞ −∞

in three dimensions,

+∞ +∞ √
2mε
√ Z∞ √
~ Θε x · dx
Z Z
N < dε 4πm dε 4m 2m
n= = 2deg gε fε = 2 =
V 2π ~ e(ε−µ)/kB T + 1 π ~3 e−βµ β 3/2 ex + eβµ
−∞ −∞ 0

above, the integral functions are seen to be polylogarithm functions after summation of the Taylor-series expansions of the
integrand,

Z∞ Z∞  ∞ Z∞ −nx ∞ n−1 √
x−1/2 dx e−x e−2x a √ X
 X n−1 e (−a) π 1/2
Ia1 = = √ − √ + ... dx = (−a) √ dx = π √ =− Li
ex + a x x n=1
x n=1
n a −a
0 0 0

Z∞ Z∞ ∞ Z∞ √ X ∞ n−1 √
x1/2 dx √ −x −2x −3x 2
X n−1 √ −nx π (−a) − π 3/2
Ia3


= = x e −e a+e a − ... dx = (−a) xe dx = = Li−a
ex + a n=1
2 n=1 n3/2 2a
0 0 0
3/2
√
r 
m  1/2
 m µ/kB T 2m 3/2 1 1
n1D = 2
−Li−eβµ ; n2D = 2
ln(1 + e ); n3D = π 2 (−Li−eβµ ); β = ; [nkD ] = k ; (2)
2π~ β π~ β ~ β kB T L
10 The nK ∼ ZK 2 e−
= 1.32 × 1028 m 28 e−
estimate of nK using the indicated aK and vBCC is
aK 3
= 3 6= 1.40 × 10 (which, in turn, implies the
(5.328×10−10 m)3 cm3
same B, in agreement to the tenths decimal place).
 derive density of states in each dimensionality,

3 3 3/2
dNk3ε ( L ) d3 kε
r
4πkε 2 dkε
 r 
L V 2mεk d 2mεk V 2mεk 1 2m V 2m
Dε3k = = 2π = = 3 4π ~2 dε = 3 4π ~2 2 = εk 1/2
dεk dεk 2π dεk (2π) k ~2 (2π) εk ~2 2π 2 ~2

L 2 2  2
dNk2ε
r r r r
( 2π ) d kε L 2πkε dkε A 2mεk d 2mεk A 2mεk 1 2m mA
Dε2k = = = = 2 2π 2 dε 2
= 2 2 2
= 2
= [const]
dεk dεk 2π dεk (2π) ~ k ~ 2π ~ ε k ~ 2π~
s
1 L 1 1
r r
dN kε ( ) d kε L 2 deg dk ε L d 2mε k L 1 2m mL2
Dε1k = = 2π = = = =
dεk dεk 2π dεk π dεk ~2 π 2 εk ~2 2εk π 2 ~2
meanwhile, zero-dimensional density of states is just a single level ε0 ,
0
dNk0ε ( L ) d0 kε
Dε0k = = 2π = 2deg δ(εk − ε0 )
dεk dεk
Ex -1.4. (a) Given MS = 2 × 1033 g for the mass of the Sun, estimate the number of electrons in the Sun. assuming that electrons,
protons, and neutrons are equal in number, the number of electrons should be roughly half the number of nucleons,

1 MS 1 2 × 1036 kg 1
Ne− ≈ = ' × 1057
2 u 2 1.6749 × 10− 27 kg 2

In a white dwarf star this number of electrons may be ionized and contained in a sphere of radius 2 × 109 cm; nd the Fermi energy
of the electrons in electron volts,
2/3 1 2/3 2/3
~2 ~2 ~2 ~2 9πMsun
  
Ne− 2 2 Msun /u 1 1

2/3
εF = 2
3π n = 3π 2
= 3π 4 3
= 3u
≈ 10−27 ·1020 ≈ 10−7 ergs = 3.104 eV
2m 2m Vwhitedwarf 2m 3 πr 2m 8r 2 2

(b) The energy of an electron in the

√ relativistic limit ε & mc2 is related to the wavevector as k as ε ' pc = ~kc. Show that the
Fermi energy in this limit is εF ' ~c n = ~c N/V , roughly.
p
3
3

p p √
... = ~c3 n
εF = pF c ' ~kF c = ~ 3 3π 2 N/V c = ~ 3 4π 2 N/r3 c = ???

(c) If the above number of electrons were contained within a pulsar of radius 10 km, show that the Fermi energy would be ∼ 108 eV .
~2 2 ~2 3
√ ~4 c6 ~4 c6
εF = 2m kF = 2m (~c n)2 = 2m n = 3u
2m 4πrpulsar 3 ... athematica = 108 eV
= ???checkW ithM

This value explains why pulsars are believed to be composed largely of neutrons rather than of protons and electrons, for the energy
release in the reaction n → p+ e− is only 0.8 × 106 eV , which is not large enough to enable many electrons to form a Fermi sea. The
neutron decay proceeds only until the electron concentration huilds up enough to create a Fermi level of 0.8 × 106 eV , at which point
the neutron, proton, and electron concentrations are in equilibrium.
Ex -1.5. The atom He3 has spin 21 and is a fermion. The density of liquid He3 is ρHe3 = 0.081 cmg 3 near absolute zero. Calculate
the Fermi energy εF and the Fermi temperature TF .
The mass density ρ = ρHe3 is given in terms of the number-density n = nHe3 using the molar mass MHe3 and Avogadro's number
NA as nHe3 = NA ρHe3 /MHe3 . Meanwhile, by work done in -1.1 and -1.2, the Fermi energy of any ensemble of fermions
11 is given by,

~2 p NA ρHe3
εF = kF 2 ; kF = 3 3π 2 nHe3 ; m = 2mp+ + mn ; nHe3 = ; (a)
2m MHe3
In Eq. (a), using SI-units, we have,

−34
~ = 6.626
2π × 10 J · s; mp+ = 1.673 × 10−27 kg; mn = 1.675 × 10−27 kg;
3
3 kg −3 kg
NA = 6.02 × 1023 He
mol ; ρHe = 0.081 × 10 m3 ; MHe = 3.016 × 10
3 3
mol ;

Some SI-values of the useful intermediaries enumerated in Eq. (a),

1 He3
kF = 7.824 × 109 ; nHe3 = 1.618 × 1028 3 ;
m m
Combining the statements in Eq. (a), we have,

~2 p ~2
εF = ( 3 3π 2 nHe3 )2 = (3π 2 )2/3 (NA ρHe3 /MHe3 )2/3 = 4.232 × 10−4 eV
2(2mp+ + mn ) 2(2mp+ + mn )

This corresponds to a Fermi temperature of TF = εF /kB = 4.9125K .

11 Due to the spin-statistics theorem, which requires quantum eld theory to derive, but its consequences are simple.
Ex -1.6.
eE
X v(t)=ve−iωt j −env −en iωm−mτ −1 ne2 /m 1
mv̇(t) = F = −eE(t)−mv(t)τ −1 → −iωmv = −eE−mvτ −1 → σ = = = = −1 = σ0
E(t)=Ee−iωt E E E τ − iω σ0 =ne2 τ /m 1+
Ex -1.7. A metal with a concentration n of free electrons of charge −e is in a static magnetic eld Bz . The electric current density
in the xy plane is related to the electric eld by jk = σk` E` . Assume that the frequency ω  ωc , where ωc = eB/mc and τ is the
collision time,
(a) Solve the drift velocity equation (51) to nd the components of the magnetoconductivity tensor: σk=` = iωp2 /4πω and σxy =
σyx = ωc ωp2 /4πω 2 .
   P   e
    −1  
iωvx q(E + v × B) ωc =eB/mc e −ωc −e/m
−v̇
vt =ve−iωt
= = P Fx /m = = m Ex
e
+ ω c vy
↔v=
vx
=
iω Ex
=
iωvy Fy /m m m Ey − ω c vx vy m ωc iω Ey ω2 − ω
(b) Note from a Maxwell equation that the dielectric function tensor of the medium is related to the conductivity tensor as
εk` = 1̂ + i(4π/ω)σk` . Consider an electromagnetic wave with wavevector k = kẑ . Show that the dispersion relation for this wave in
the medium is12 (63) → c2 k2 = ω 2 − ωp2 ± ωc ωp2 /ω .
2
d2 Ext Ext =Eei(kz−ωt)
 
 εxx ω 2 − c2 k 2 εxy ω 2
    
2 2 2 Ex εxx εxy − − (iω) Ex  εxx =1−(ωp /ω)2 =ε
c ∇ Ext = ε → c(ik) = 2 →0=  2 2

2 2  =
dt 2 Ey εyx εyy − − (iω) Ey εyx ω εyy ω − c k εxy =iωc ωp 2 /ω3 =−

in which,
!2 !−2 !2 !2
ωc ωp2 (c2 k 2 + ωp 2 ) ωc ωp2 (c2 k 2 + ωp 2 ) ωc2 ωp4

−1 ω ω dimensionl
2 = αΩ =α p ↔α= 2 = 3 =
functionof
p
ω(c2 k 2 + ωp 2 ) c2 k 2 + ωp 2 ω(c2 k 2 + ωp 2 ) c2 k 2 + ωp 2 (c2 k 2 + ωp 2 )
this gives,

√ √ √
vp
r u
q
c 2 k 2 + ω 2 u 3 3 3 27α2 − 4α + 27α − 2 3
2
p t
2 2 2
ω± = ± Ω(c k + ωp ) = ± √3
+ p √ √ +2
3 2 3
3 3 27α2 − 4α + 27α − 2

∇ • E = ρ/ε0 ; ∇ • B = 0; ∇ × E + Ḃ = 0; ∇ × B − µ0 ε0 Ė = µ0 J; dF = dq(E + v × B); (3)

no charge ρ=0 and no current J = 0, so ∇•E = ∇•B = 0 and ∇ × B − µ0 ε0 Ė = ∇ × E + Ḃ = 0. taking the curl of both sides
of the vector equations,

∇(0) − ∇2 E c−2 Ë ∇2 E
           
∇ × (−Ḃ) ∇ × (∇ × E) ∇•E=0 −∂t ∇ × B c2 ∇×B=Ė
= → 1 = → =
∇ × ( c12 Ė) ∇ × (∇ × B) ∇•B=0 c2 ∂t ∇ × E ∇(0) − ∇2 B ∇×E=−Ḃ c−2 B̈ ∇2 B

∇2 E = µ0 ε0 Ë; ∇2 B = µ0 ε0 B̈; µ0 ε0 = c−2 ; (4)

Ex -1.8. We dene the dimensionless length rs, as r0/aH, where rn is the radius of a sphere that contains one electron, and aH is
the bohr radius h2/e2m.
(a) Show that the average kinetic energy per electron in a free electron Fermi gas at 0 K is 2.21/rf, where the energy is expressed
in rydbergs, with 1 Ry = me4/2h2. not in solution manual. don't care. (yet)
(b) Show that the coulomb energy of a point positive charge e interacting with the uniform electron distribution of one electron in
the volume of radius r0 is - 3(T/2r0, or -3/rs in rydbergs.
Z Z Z Z r0
3 q kq 2 4πkq 2 4πk(−e)2 2
Ue−/ion = − ρe− · Vion−core · d r = − [volume] · |r−0| · 4πr dr = − [volume] r· dr ' − (4/3)πr0
3 r · dr = − 3ke
r0 3
1
2 (r
r r r ∃screeningw/lengthr0 0

THe minus sign in my nswer I hve not accounted for.

(c) Show that the coulomb self-energy of the electron distribution in the sphere is 3e2 /5r, or 6/5rs in rydbergs.
q
[(4/3)πr0 3 ]
Z Z Z Z
kρ[vol] k (4/3)πr 3 (4/3)πr 3 q
Uself = Ve+ · dqe− = r · ρe− · d3 r = 0
r · (4/3)πr0 3 · 4πr2 dr = ( 4πr−e3
0 /3
)2 · 43 π·4π r3 r2 r−1 dr '
r r ∃screeningw/lengthr0

(d) The sum of (b) and (c) gives −1.80/rs for the total coulomb energy per electron. Show that the equilibrium value of rs is 2.45.
Will such a metal be stable with respect to separated IIatoms? The average Fermi energy per election is 3εF /5, from Problem 6.1;
because N/V = 3/4πr03 , the average is 3(9π/4)2/3 ~2 /10mr02 . Tlie sum of the Coulomb and kinetic contributions is U − − 1.80 2.21
rs + rs2
1.80 4.42
which is a minimum at
rs2 − rs3 , orrs − 4.42/1.80 − 2.45. The binding energy at this value of rs is less than 1Ry ; therefore separated
H atoms are more stable
12 At a given frequency there are two modes of propagation with dierent wavevectors and dierent velocities. The two modes correspond to circularly
polarized waves. Because a linearly polarized wave can be decomposed into two circularly polarized waves, it follows that the plane of polarization of a
linearly polarized wave will be rotated by the magnetic eld.
Ex -1.9. For the drift velocity theory of (51), show that the static current density can be written in matrix form as:
    
jx 1 −ωc τ 0 Ex
σ0  ωc τ
 jy  =
2
1 0   Ey 
jz 1 + (ωc τ ) 0 0 1 + (ωc τ )
2
Ez

In the high magnetic eld limit of ioct > 1 , show that: σyx = nec/B = −σxy .
ωc τ
 1 → σyx ' σ0 /ωc τ = ne2 τ /m (mc/eBτ ) = neB/c


ji = σij Ej → jy = σyx Ex = 2 σ0 Ex ; ωc τ
1 + (ωc τ )

In this limit or = 0, to order 1/ioct. The quantity crt is called the Hall conductivity.
Ex -1.10. Consider a square sheet of side L, thickness d. and electrical resistivity p. The resistance measured between opposite edges
of the sheet is called the surface resistance: Rsq = pL/JA = p/d ,which is independent of the area L2 of the sheet. (Rsl, is called the
resistance per square and is expressed in ohms per square, because p/d has the dimensions of ohms.) If we express p by (44), then
Rsq = m/nde2r. Suppose now that the minimum value of the collision time is determined by scattering from the surfaces of the sheet,
so that T ∼ d/vy , where vF is the Fermi velocity. Thus the maximum surface resistivity is Rsq ∼ mvh/nd2e2. Show for a monatomic
metal sheet one atom in thickness that K. ∼ h/e2 − 4.1kf l. It is possible to obtain the quantum of conductance within the simple
Drude-Sommerfeld model of a planar conductor if one is willing to settle for an order-of-magnitude estimate. Consider the resistance
Rsq across the opposite-edges of any square-faced orthorhobus made of a material well-described by the Drude-Sommerfeld model.
Let the square-sheet be of side-length L, thickness d, of electrical resistivity ρ, and of cross-sectional area A = Ld. The resistance
measured
13 between opposite edges of the sheet (what Kittel[?] calls the surface resistance") is,

ρL ρL ρ
Rsq = = = = Rsq (ρ, d)
A Ld d
Let an ohmmeter subject the material to a uniform electric eld E in the longitudinal dimension with Ohmic steady-state response
J = σE = E/ρ with dJdt = 0 and, for a number n charge-carriers per unit volume (themselves each of charge q = ±e), the current-
density is J = nqp/m. The response is furthermore modeled by the Drude-treatment: Newton's laws for a typical charge-carrier of
charge q = ±e and mass m are valid in the approximation of stochastic collisions happening on average every τ unit of time,

X dp p nqp p nqp p nq 2 ρ 1
F =m = m · 0 = 0 = Fe + Fcol = qE − = q ρ− ↔q ρ= ↔ =
dt τ m τ m τ m τ
This result shows that one knows σ if they can nd a good model for τ. Hence, one approximates this collision time τ using the
Fermi velocity (here, the Sommerfeld model enters), and note that, for a monatomic sheet, one has
14 d = n−1/3 ,
1/3
mn−1/3 3π 2

d d md m
τ∼ = = = × = (3π 2 )1/3
vF ~kF /m ~(3π 2 n)
1/3
~(3π 2 n)
1/3 3π 2 ~kF 2

Placing this estimate for τ into the equation of motion for ṗ, and using it to calculate the conductivity σ for use in Rsq ,
m 2/3
ρ nq 2 τ m m ~kF 2 ~(3π 2 n) ~
Rsq = = −1/3 = 2/3 2 ∼ 1/3
= 1/3
= 1/3
= 2 (3π 2 )1/3
d n n q τ m
n2/3 q 2 ~kF 2 (3π 2 ) n2/3 q 2 (3π 2 ) n2/3 q 2 (3π 2 ) q

One notices that (3π 2 )1/3 ≈ 3.094 < 101 , so they absorb it into the order-of-magnitude statement and nally estimate the
square-resistance as,
~ 6.626
× 10−34 J · s
Rsq ∼ 2 = 2π
2 = 4.109 × 103 Ω
(±e) −19
(±1.602 × 10 C)
This resistance Rsq is clearly half of the conductance-quantum. Taking the reciprocal and multiplying by 2 gives the conductance-
quantum for the conduction-channels of two electronic spins. Hence, dimensional analysis to estimate[?] the conductance within the
Drude-Sommerfeld theory gives the conductance-quantum.
13 Most ohmmeters work by passing a (usually small) current through the sample and measuring the voltage drop - thus nding the resistance by Ohm's
law
V = IR ↔ E = Jρ
. Any internal resistance of the ohmmeter should be well known and included in the calculation (usually automatically).
14 Furthermore, one uses the following: imagine a k-space kF  2π/L wherein the N allowed levels are divided
sphere of radius up into k-space unit cells
of volume ( 2π
L
)3 , yielding the Fermi energy as a function of state-variables N, V

2deg 34 πkF 3 N 2/3

r r
~2 kF 2 ~2
 
1 3 3 3 N 3 N
N = = L kF ↔ kF = 3π 2 3 = 3π 2 → εF = = 3π 2
(2π/L)3 3π 2 L V 2m 2m V
Ex -1.11. 053 - pr 01 - a free 2DEG: preliminaries,
Z +∞
(2.67) → n = dε · g(ε)f (ε)
−∞

(1.1)
(a) What is the relation between n and kF in two dimensions?

Ak 2deg πkF 2 L2 2 N kF 2
N= = 2 = kF ↔ n = n(kF ) = 2 =
Ak,0 ( 2π 2π L 2π
L )

(1.2)
(b) What is the relation between kF and rS , in two dimensions?

N 1 1 √
n= = 2
= kF 2 ↔ kF rS = 2
A πrS 2π
(1.3)
(c) Prove that in two dimensions the free electron density of levelsg = g(ε) is a constant independent of

for ε > 0, and 0 forε < 0. What is the constant?

√ √
1 d2 k 2 2πk · dk 1 2mε 2m 12 m T '0 m
gε = 2deg 2 dε = 4π 2 = √ = = Θ(εF − ε)Θ(ε)
(2π) dε π ~ ~ ε π~2 π~2

(1.4)
The density of states shuts o at ε > εF for some reason that I actually can't remember. It is not used in (e).
(d) Show that because g = g(ε) is constant, every term in the Sommerfeld expansion for n vanishes except the T = 0 term. Using
the Sommerfeld expansion (1.20) for the electron-density n from (1.21) (see Appendices),

+∞
Z Z∞ Z∞ ZεF ZεF ∞ 2n−1 
m m m X 2n d m 
I= H(ε)f (ε)dε = N = dε · g(ε)f (ε) = dε · 2 Θ(εF − ε)f (ε) = dε · f (ε) = dε− an (k B T ) =
π~ π~2 π~2 n=1
dε2n−1 π~2
−∞ −∞ 0 0 0

(1.5)
Deduce that µ = εF at any temperature.
This was one of the conclusions noted in 045 - pr 02-3-b - the Sommerfeld-approximated electronic Cv and mu dependence upon
T, wherein the following expression was obtained, (1.6)
Obviously, for a 2-dimensional electron gas,
gε0 F = 0
, resulting in µ(T ) = εF ∀T . This is wrong, as we will see in the next question.

One begins with an integral I over a nite band ε ∈ [−2W, 2W ] which is of the form of a band-density H = H(ε) (localized that
its leading dependence upon ε is the nth power of the same at the band-edges ±2W ) times a fermionic thermal function f ≶ (ε),
Z 2W  −1
I= H(ε)f ≶ (ε)dε; ; f < (ε) = e(ε−µ)/(kB T ) + 1 = 1 − f > (ε); ; lim H(ω) ∼ εn ;
−2W ω'±2W

One then uses the fundamental theorem of calculus to dene the antiderivative of H(ε), which is K(ε); i.e.,

Z ε
K(ε) ≡ H(ε0 )dε0 → H(ε) ≡ K 0 (ε) = ∂ε K(ε)
−2W

One subsequently integrates by parts (i.e., integrate both sides of Hf = (Kf )0 − (Kf 0 )) to transfer the derivative d/dε onto the
thermal function,

Z 2W    Z 2W Z 2W
≶ ≶ ≶
I= (K(ε)f ≶ (ε))0 − K(ε)f 0 (ε) dε = 0 − K(ε)f 0 (ε)dε = − K(ε)f 0 (ε)dε
−2W −2W −2W

Each thermal function's explicit derivative, using the abbreviation x = (ε − µ)/kB T , is a hyperbolic-secant-squared function which
is strongly-peaked a radius of the temperature kB T about the chemical potential µ,

−( kB1T e(ε−µ)/(kB T ) + 0) − kB1T

≶  2
dfε −1 sech(x/2) −1 ε−µ
± = 2 = 2 = k T = sech2
dε (e (ε−µ)/(kB T ) + 1) (e x/2 +e −x/2 ) B 2 4kB T 2kB T
 Because the thermal function derivative f 0≶ (ε) is strongly peaked about µ, the integral I must take its greatest support from its
integrand being sampled at µ ∈ [−2W, 2W ]. One thus Taylor-expands the antiderivative K(ε) about ε = µ,
+2W ∞
! +2W ∞ +2W
n ≶ ≶ ≶
(ε − µ) dn Kµ −dfε −dfε 1 dn Kµ n −dfε
Z X Z X Z
I= dε = Kµ dε + (ε − µ) dε
n=0
n! dεn dε |{z} dε n=1
n! dεn dε
−2W −2W ∂ε 0 Kµ −2W

Hence, by use of the antiderivative K0 = H and Taylor-expansion, the limit T ' 0 is built into the leading term (which some might
say is the genius of the Sommerfeld expansion). Ascending powers of the series are corrections due to nonzero temperature. Thus,
≶
the quantity I which is the integral of a thermal function fε and a band-density H(ε) (which is localized to a band ε ∈ [−2W, 2W ]
such that it has [at most] power-law behavior at the band-edges) due to Sommerfeld[?] may be represented as a zero-temperature
leading term plus corrections

Z2W Zµ (+2W
Z +µ)/T
∞
1 X 1 dn−1 Hµ n x
I= Hε fε≶ dε =± Hε dε ± (kB T ) xn sech2 dx (5)
4 n=1 n! dεn−1 2
−2W ∓2W (−2W +µ)/T

Convolution integrals by Sommerfeld expansion: A quantity Q which is a convolution-integral of H = H(ω) and Fermi
≶
function fω over a nite energy-band ω ∈ [−2W, 2W ] in a system having chemical potential µ and temperature T is representable as
a zero-temperature term plus corrections, as,

(µ−ω)Θ2W ±(µ−ω)
Z2W Z ∞
≶ (n−1)
X
Q≶
ω = dω 0 fω0 +ω Hω0 = (±H )dω +
ω0
0
Tα n Hµ−ω bn (6)

−2W ∓2W n=1

Z −µ)/T
(ω+2W
Tn x
bn ≡ dx · xn sech2
4(n!) 2
(ω−2W −µ)/T

The limits of integration have step-functions because the freely-varying (independent) variable ω in the convolution varies over
ω ∈ [−2W, 2W ]. Accordingly, the upper limit of integration must vanish if the integral takes its support from a peaked function that
is outside of the stepped-thermal-function.
The wide-band limit: Much simplication of the above result occurs for an innitely-wide band. These correction terms are
only those with even powers of ε − µ, because f 0 (ε) is even about µ and (ε − µ)n is even/odd alternatingly in n,
+∞
 +∞ 
∞ ∞
!
n ≶ ≶
(ε − µ) dn H(µ) −∂fε n
−∂f
Z Z
0
X X 1 d H(µ) n ε
I ≡ n
dε = 
n
 (ε − µ) dε
n=1
n! dε ∂ε n=1
n! dε ∂ε
−∞ −∞

∞ +∞ ∞ +∞
≶
1 dn H(µ) 1 d2n H(µ)

2n −∂fε
Z Z
X nodd → odd X
= [even] dε = (ε − µ)
n! dεn neven → even | {z } (2n)! dε2n ∂ε
n=1 −∞ n=1 −∞
-parity

ε−µ
One then gets rid of K(µ) in favour of the original H(ε), and also introduces the substitution x≡ kB T . One then gets,
 +∞

∞ 2n−1 ≶
−df
Z
X 1 d Hµ 2n (x)
I0 = 
2n−1
(kB T x)  · d(kB T x + µ)
n=1
(2n)! dε d(k B T x + µ)
−∞
 +∞ 
∞ 2n ≶ 2n−1 ∞ 2n−1
−df
Z
X x (x) 2n d Hµ X 2n d Hµ
=  dx · (kB T ) 2n−1
≡ [an ] (kB T ) 2n−1
n=1
(2n)! dx dε n=1
dε
−∞

The dimensionless numbers introduced above can be written in terms of the Riemann-zeta-function,

Z∞ ≶ Z∞ ∞ m+1
x2n −∂fx ±x2n x X ±(−1) 8(2n)!
an ≡ dx = sech2 dx =
(2n)! ∂x 4(2n)! 2 m=1
4(2n)! m2n
−∞ −∞
∞ m+1  
X (−1) 1
=2 ≡± 2− ζ2n
m=1
m2n 22(n−1)

This gives the familiar form of the Sommerfeld expansion, The integral I of some localized
15 function H = H(ε) (usually a density
<
of states) with a Fermi-Dirac distribution f is representable by Sommerfeld's expansion that converges for temperature T that is
lower than the energy-scale set by the derivative of H(ε). This expansion appears as follows,

15 The function H(ε) must decay at least as fast as ε−n .

 Z +∞ Z µ X∞ 2n
I= Hε fε≶ dε =± Hε dε ± an (kB T ) Hµ(2n−1) ; an = (2 − 2−2(n−1) ) · ζ(2n); (7)
−∞ ∓∞ n=1

Ex -1.12. 053 - pr 02 - thermodynamics of D dimensional electron gas: preliminaries,

2
(2.34) → P = (E/V ) ;
3
(1.1)
1
(2.56) → f = f (ε) = = [fermifunction] ;
exp( kε−µ
BT
) +1
(1.2)
d3 k
Z
(2.57) → u = E(k)f (ε(k));
4π 3
(1.3)
G
(B.5) → µ = ;G = U − TS + PV ;
N
(1.4)
(a) Take the following thermodynamic identities as given,

   
∂u ∂s S
cV = =T ;s = ;
∂T n ∂T n V

(1.5)
Also take as given (1.2), (1.3), and the third law of thermodynamics (i.e., s→0 as T → 0). Show that the entropy density, s =
S/V is given by,

d3 k < eβ(εk −µ)

Z
< > >

< 1 > >
s = s(ε, T, µ) = −kB f ln f + f ln f ; f = = 1 − f ; f = ;
4π 3 εk εk εk εk εk
eβ(εk −µ) + 1 εk εk
eβ(εk −µ) + 1

(1.6)
We'll start with the partition function, Z, for a free electron gas,

N N 1
! N 
Y Y X Y 
−βn(εi −µ)
ZN = Zi = e = 1 + e−β(εi −µ) ;
i=1 i=1 n=0 i=1

(1.7)
Energy from partition function and formula ε = −∂β ln Z ; cominbing this and (1.7),

N  N N N
∂ Y  X 0 + e−β(εi −µ) (−(εi − µ)) X (εi − µ)e−β(εi −µ) X
ε=− ln 1 + e−β(εi −µ) = − −β(εi −µ)
= −β(εi −µ)
= fε<i (εi − µ);
∂β i=1 i=1
1 + e i=1
1 + e i=1

(1.8)
Combining (1.7) and (1.8) and carefully using the properties of logarithms, we calculate the entropy as,

N  N N 
exi
 
S 1 ∂Ω X
−β(εi −µ) <
 xi ≡β(εi −µ) X
−x <

X 1
=− = ln Z+βε = ln(1 + e ) + fεi β(εi − µ) = ln(1 + e ) + fεi xi =
i
− ln × xi +
kB kB ∂T i=1 i=1 i=1
1 + e−xi e

We then rewrite the above calculations in the continuum limit, N ' ∞,

XN Z ∞ Z
d3 k
fε<i ln fε<i + fε>i ln fε>i ' −kB fε< ln fε< + fε> ln fε> g(ε)dε = −kB fε<k ln fε<k + fε>k ln fε>k




S = −kB 4π 3
i=1 −∞

(b) Since the pressure P satises (1.4), P = −(u − T s − µn), deduce from (1.6) that the pressure is,

Z 3
~2
 
d k −β( 2m k2 −µ)
P = P (T, µ) = kB T ln 1 + e ;
4π 3

(1.9)
Using (1.3), (1.6), P = −(u − T s − µn), the abbreviation

εk −µ
xk ≡ kB T
, the property
16
exk fε<k ≡ fε>k
, and the properties of logarithms (very carefully!),

d3 k d3 k εk −µ <
Z Z
P = −(u−T s−µn) = − (εk fε<k − T (−kB )(fε<k ln fε<k + fε>k ln fε>k ) − µfε<k ) = kB T (− kB T fεk − (fε<k ln fε<k + fε>k ln fε>k )) = kB T
4π 3 4π 3
Show that (1.9) implies that P is a homogeneous function of µ and T of degree 5/2; that is,

5/2
P = P (T, µ); P (λT, λµ) = λ P (T, µ) = λ5/2 P ∀λ;
(1.10)
Divide top and bottom of the exponential argument by
λ
√
, and eect a change of variables to y = k/ λ, which is not felt due to the integration over all of k -space,
! Z λ3/2 · d3 ( √k ) √ 2 !
2 2
~2 (k/ λ)
d3 k 1 ~ k 1 d3 y
Z Z
− ( −λµ) λ −k T ( −µ)
Pλ = P (λT, λµ) = kB λT 3
ln 1 + e kB (λT ) 2m = λ·kB T ln 1 + e B 2m = λ5/2 ·kB T
4π 4π 3 4π 3
(c) Deduce from the thermodynamic relations in Appendix B that,
   
∂P ∂P
= n; = s;
∂µ T ∂T µ

(1.11)
Calculating the indicated derivatives from the equation of state (1.9),

   Z 3  Z 3 Z 3 0 + e−β(εk −µ) −(εk −µ) Z 3

∂P ∂ d k  −β(εk −µ) d k  −β(εk −µ)
 d k kB (−T 2 ) 1 1 d k
= kB T ln 1 + e = k B ln 1 + e +k B T = P + (ε
∂T ∂T 4π 3 4π 3 4π 3 ~2 2 T T 4π 3
µ µ −β( 2m k −µ)
1+e
Z 3 Z 3 −β(εk −µ) Z 3
     
∂P ∂ d k 
−β(εk −µ)
 d k0+e (−β(0 − 1)) d k < ∂P
= kB T ln 1 + e = k B T = f = n → =n
∂µ T ∂µ 4π 3 T 4π 3 1 + e−β(εk −µ) 4π 3 εk ∂µ T
This conrms (1.11), and means that the equation of state (1.9) satises the dependence

P = P (T, µ)
.
(d) By dierentiating (1.10) with respect to λ, show that the ground-state relation (1.1) holds at any temperature, in the form
P = 23 u. Note that u = u(s, V ). At any temperature" means, Pµ,T = 23 us,V = Pµ,λT ;
Using
17
∂PλT ,λµ
( ∂(λT ) )λµ =s
and
∂PλT ,λµ
( ∂(λµ) )λT =n
, we use the chain rule to get,

∂PλT ,λµ ∂(λT ) ∂PλT ,λµ ∂(λµ)

d 5/2
dλ λ PT,µ = 52 λ3/2 PT,µ = d
dλ PλT,λµ =( ∂(λT ) )λµ ∂λ +( ∂(λµ) )λT ∂λ = sT + nµ ↔ PT,µ = 25 λ−3/2 (sT + nµ);
We then recall that PT,µ = −(u − T s − µn) implies PT,µ + u = T s + µn, and furthermore by (1.10) that an expression for λ comes
from PλT,λµ = λ5/2 PT,µ = −λ5/2 (u − T s − µn) so we write,
−3/2
PT,µ = 52 λ−3/2 (PT,µ + u) = 52 (λ) (PT,µ + u)
(e) Show that when kB T << εF , the ratio of the constant-pressure to constant-volume specic heats satises,

2
(cp /cv ) − 1 = 31 (πkB T /εF ) + O4
(1.13)
(f ) Show, by retaining further terms in the Sommerfeld expansions of u and n, that correct to order T3 the electronic heat capacity
is given by,
2 !
g 0 (εF ) g 00 (εF )

1 2 1 4
cv T = 3 (πkB T ) · g(εF ) − 90 (πkB T ) g(εF ) 15 − 21 ;
g(εF ) g(εF )
(1.14)

16 Notice this is in conformity with the uctuation-dissipation theorem for fermions.

17 These follow from the identities (1.11); since
∂PλT ,λµ ∂Px,y
( ∂(λT )
)λµ =( ∂x
)y =s
and
∂PλT ,λµ ∂Px,y
( ∂(λµ)
)λT =( ∂y
)x =n
.