Physics MSc of FNS BME

I. Year, 1. semester

MODERN SOLID STATE PHYSICS

lecturer: Attila Virosztek

Problem set

1. Identical particles

2. Second quantization I. (bosons)

3. Second quantization II. (fermions)

4. Field operators

5. Phonons, magnons, bosons

6. Electrons I. (Fermi liquid, Hartree-Fock approximation, Wigner crystal)

7. Electrons II. (Wannier states, Hubbard model)

8. Condensates

1
1. Identical particles

1.1 Out of three bosons two is in the state ϕ1 (r), while one is in the state ϕ2 (r). Write
down the symmetrized wavefunction ψ(r1 , r2 , r3 ), then calculate the normalization
factor!

1.2 Prove that for√N fermions the normalization factor of the antisymmetrized wave-
function is 1/ N !!

1.3 Given N bosons

a. p
prove that the normalization factor of the symmetrized wavefunction is
N1 !N2 !.../N !, if during the symmetrization we only take into account per-
mutations leading to different terms!

b. what would be the normalization factor, if all permutations were included?

1.4 If the number of different states a particle can be in is Ω, then the system of N
such distinguishable particles can be in ΩN different states.

a. How many physically different states can be occupied by a system of N

fermions?

b. Suppose that Ω = 2N . What is the asymptotics of the number of many body

states of fermions and distinguishable particles for large N ?

2. Second quantization I.

2.1 Consider the three bosons of exercise 1.1.

a. Using ψ112 determined there, calculate the expectation value of the operator
F (1) = f (1) (r1 ) + f (1) (r2 ) + f (1) (r3 ) !

b. Determine the matrix element hψ122 |F (1) |ψ112 i!

PN
2.2 Prove the followings for N bosons and for the operator F (1) = i=1 f (1) (ri )

(1)
a. hψN1 ,N2 ,... |F (1) |ψN1 ,N2 ,... i =
P
j Nj fjj .

2
 (1) p
b. hψ...,Nj −1,...,Nk +1,... |F (1) |ψ...,Nj ,...,Nk ,... i = fkj Nj (Nk + 1).

2.3 Consider a single one particle state, which can host arbitrary number of bosons.
a+ and a are the usual creation and annihilation operators:
√
a|N i = N |N − 1i,
√
a+ |N i = N + 1|N + 1i,

[a, a+ ] = 1.

Calculate the following matrix elements between states with arbitrary N and N ′ :

a. hN ′ |a+ a|N i,

b. hN ′ |aa+ |N i,

c. hN ′ |a+ a+ |N i,

d. hN ′ |aa|N i,

e. hN ′ |a+ a+ aa|N i,

f. hN ′ |a+ aa+ a|N i.

2.4 Consider several one particle states, which are filled by N1 , N2 , ... bosons. The
creation and annihilation operators corresponding to the j-th state are a+ j and aj .
Calculate the following matrix elements:

a. hN1 , N2 , ...|a+
j ak |N1 , N2 , ...i,

b. h..., Nj − 1, ..., Nk + 1, ...|a+

l am |..., Nj , ..., Nk , ...i.

2.5 Utilizing the results of exercises 2.2 and 2.4 prove, that the second quantized form
of the general one particle operator F (1) is
X (1)
F (1) = fkj a+k aj .
k,j

2.6∗ Utilizing the rules for bosons Pprove, that the second quantized form of the general
N
two particle operator F (2) = i,j=1 f (2) (ri , rj ) is
X (2)
F (2) = fklmn a+ +
k al am an ,
k,l,m,n

where
Z Z
(2) 3
fklmn = d r1 d3 r2 ϕ̄k (r1 )ϕ̄l (r2 )f (2) (r1 , r2 )ϕm (r2 )ϕn (r1 ).

3
3. Second quantization II.

3.1 Prove that for N fermions the diagonal matrix elements of the one particle operator
F (1) are:
(1)
X
hN1 , N2 , ...|F (1)|N1 , N2 , ...i = Nj fjj .
j

3.2 Prove that for N fermions the off diagonal matrix elements of the one particle
operator F (1) are:
(1)
h..., Nj − 1, ..., Nk + 1, ...|F (1)|..., Nj , ..., Nk , ...i = ±fkj ,
where Nj = 1 and Nk = 0 of course, moreover the sign of the right hand side is
”+”, if there are an even number of particles in the states between states j and
k, and ”−”, if there are an odd number of particles in the states between states j
and k.

3.3 According to the definition of fermionic creation and annihilation operators:

P
N
ak |..., Nk , ...i = (−1) i<k i Nk |..., Nk − 1, ...i,
and P
Ni
a+
k |..., Nk , ...i = (−1) i<k (1 − Nk )|..., Nk + 1, ...i.
Prove that

a. (ak )+ = a+
k,

b. a+
k ak = Nk ,

c. ai a+ +
j + aj ai = δij .

3.4 ak and a+k are fermionic operators (k=1,2,...,5). Calculate the following matrix
elements:

a. h0, 1, 1, 0, 1|a+
5 a4 |0, 1, 1, 1, 0i,

b. h1, 1, 0, 0, 1|a+
1 a3 |0, 1, 1, 0, 1i,

c. h1, 0, 1, 1, 0|a5a2 a+ +
4 a3 |1, 1, 0, 0, 1i,

d. h1, 1, 0, 0, 1|a+ +
1 a3 a2 a4 |1, 0, 1, 1, 0i.

3.5 Using the results of exercises 3.1 - 3.3 prove that for fermions as well, the second
quantized form of a general one particle operator F (1) is:
X (1)
F (1) = fkj a+
k aj .
k,j

4
3.6∗ Prove, that the second quantized form of a general two particle operator F (2) for
fermions is identical to that for bosons!

4. Field operators

4.1 Based on the definition X

Ψ(r) = ak ϕk (r)
k
X
Ψ+ (r) = a+
k ϕ̄k (r)
k

of the field operators and on the known commutation rules for particle number
operators, verify the following commutation relations for field operators:

for bosons:
[Ψ(r), Ψ+ (r′ )] = δ(r − r′ ),

[Ψ(r), Ψ(r′ )] = [Ψ+ (r), Ψ+ (r′ )] = 0;

for fermions:
{Ψ(r), Ψ+ (r′ )} = δ(r − r′ ),

{Ψ(r), Ψ(r′ )} = {Ψ+ (r), Ψ+ (r′ )} = 0.

4.2 Prove, that the second quantized form of the one particle operator F (1) is expressed
by field operators as Z
F (1) = d3 rΨ+ (r)f (1) (r)Ψ(r).

4.3 Prove, that the second quantized form of the two particle operator F (2) is expressed
by field operators as
Z Z
(2) 3
F = d r1 d3 r2 Ψ+ (r1 )Ψ+ (r2 )f (2) (r1 , r2 )Ψ(r2 )Ψ(r1 ).

4.4 With the help of the particle density operator Ψ+ (r)Ψ(r) and the commutation
relations prove, that the state Ψ+ (r)|0i describes one particle in the point r in
space. The vacuum state is |0i = |N1 = N2 = ... = 0i.

5
4.5 Determine the second quantized form of the
X
j(r) = −e δ(r − ri )vi
i

current density operator! The field operator is

1 X ikr
Ψ(r) = √ e ak ,
V k

while the velocity is

pi + (e/c)A(ri )
vi = .
m
Watch out, the term containing the canonical momentum should be symmetrized!

4.6 Prove that for a system of identical particles described by the Hamiltonian
X
H= [−h̄2 ∇2i /2m + U (ri )],
i

the total particle number operator N commutes with H! Do it simultaneously for

fermions and bosons using the sign ±.

5. Phonons, magnons, bosons

5.1 The second quantized form of the Hamiltonian of an ionic lattice is

 
X
+ 1
H= h̄ωλ (k) aλ (k)aλ (k) + ,
2
k,λ

where ωλ (k) is the frequency of the phonon mode λ with wavenumber k, while
a+
λ (k) and aλ (k) creates and annihilates one phonon into or from this mode. If
the unit cell contains just a single atom, then the operator of the displacement
vector of the ion at lattice point R in second quantized form is
r
h̄ X eikR
eλ (k) aλ (k) + a+
 
u(R) = p λ (−k) ,
2N M ωλ (k)
k,λ

where M is the ionic mass, N is the number of ions and eλ (k) is the unit polar-
ization vector of the (λ, k) mode. Suppose, that the gas of phonons is in thermal
equilibrium, therefore hNλ (k)i = {exp[βh̄ωλ (k)] − 1}−1 .

6
 a. What is the expectation value of the displacement vector hu(R)i of an ion?

b. What is the expectation value of the square of the displacement vector

h[u(R)]2 i of an ion?

5.2 Suppose, that the three acoustic phonon branches in exercise 5.1 are degenerate,
and for all λ = 1, 2, 3
ωλ (k) = ck,
where k < kD = kB ΘD /h̄c, and ΘD is the Debye temperature. Here c is the
velocity of sound, and the Debye wavenumber kD ≈ π/a is of the order of the
inverse lattice spacing.
p
a. What is the relative average displacement h[u(R)]2 i/a at low temperatures
(T ≪ ΘD )?

b. What is the above ratio for T ≫ ΘD ?

5.3 Suppose that there are N ions with mass M each in an ionic lattice with one ion
per unit cell. We replace Ni (≪ N ) ions by their isotopes with mass M ′ . Due
to this replacement, the following perturbation is added to the Hamiltonian of
exercise 5.1:
Ni  
′
X 1 1 2
H = ′
− [P(Ri )] ,
i=1
2M 2M

where P(R) is the momentum of the ion at lattice point R, Ri is the site of the
i-th isotope, and these sites are random (the probability of each isotopes being at
any lattice site is the same).

a. Using the second quantized form of the momentum operator

r
1 h̄M X ikR p
ωλ (k)eλ (k) aλ (k) − a+
 
P(R) = e λ (−k)
i 2N
k,λ

write down H ′ in second quantized form, and give the diagrammatic repre-
sentation of the processes described by each term!

b. Collect the terms describing scattering of a phonon in H ′ , and bring them

to the form XX
′
Hscatt = g(k′ , λ′ ; k, λ)a+ ′
λ′ (k )aλ (k)
k,λ k′ ,λ′

(Make use of the commutation rules, and the relations ωλ (−k) = ωλ (k), and
eλ (−k) = eλ (k).) Write down the expression for the g(k′ , λ′ ; k, λ) scattering
amplitude!
′
c. Based on the form of Hscatt in point b., express the hk′ , λ′ |Hscatt
′
|k, λi matrix
′ ′
element with the help of the scattering amplitude g(k , λ ; k, λ), then use

7
 Fermi’s ”golden rule” in order to to calculate the w(k, λ → k′ , λ′ ) transition
probability per unit time! (Here, except for the states (λ, k) and (λ′ , k′ ), the
number of phonons in other modes do not change.)

d. The change of the number of phonons per unit time in the mode (λ, k) due
to scattering out of this mode is:

∂ X
Nλ (k)|out = − w(k, λ → k′ , λ′ ).
∂t ′ k ,λ
′

On the other hand phenomenologically

∂ Nλ (k)
Nλ (k)|out = − .
∂t τλ (k)

Express the 1/τλ (k) inverse lifetime of the (λ, k) phonon with the help of
g(k′ , λ′ ; k, λ)!

e. Utilizing the result for g(k′ , λ′ ; k, λ) obtained in point b., calculate the quan-
tity |g(k′ , λ′ ; k, λ)|2 averaged over the random positions of the isotopes. Use
the following definition of the average:

1 X 1 X 1 X
f¯(R1 , R2 , ..., RNi ) = ... f (R1 , R2 , ..., RNi ).
N N N
R1 R2 RN i

f. Calculate 1/τλ (k) (see point d.) based on the result obtained in point e.!
(Utilize that in thermal equilibrium Nλ (k) = {exp[βh̄ωλ (k)] − 1}−1 .)

g. Suppose that all acoustic branches are degenerate, therefore ωλ (k) = ck for
all λ = 1, 2, 3, where c is the velocity of sound. Then for each wavenumber k
the system of unit polarization vectors eλ (k) can be directed arbitrarily (as
long as they are mutually orthogonal). For example even eλ (k′ ) = eλ (k) can
be chosen for each k′ . Exploiting this, calculate 1/τλ (k) at zero temperature!
How does 1/τ (k) depend on k?

h. Evaluate the relative damping 1/ω(k)τ (k) also at zero temperature!

5.4 The spin operators can be expressed with the a, a+ bose operators (aa+ −a+ a = 1):

√ a+ a
S+ = 2Sa+ (1 − ),
2S
√
S− = 2Sa,
and
S z = a+ a − S.

8
 a. Show that with such a definition the spin operators satisfy the spin commu-
tation relations: [S + , S − ] = 2S z , [S − , S z ] = S − , [S z , S + ] = S + .

b. Express the X
H = −J Si Sj
<i,j>

Heisenberg Hamiltonian with the bose operators! < i, j > means nearest
neighbors.

c. Retaining terms with two operators only, diagonalize H with the following
transformation:
1 X iqRi
ai = √ e aq .
N q

Let the spins reside on a cubic lattice with lattice constant b. How does the
energy spectrum behave for small wavenumbers?

d. Calculate the low temperature specific heat!

5.5 The Hamiltonian of bosons scattered by impurities is

N N N
X h̄2 ∇2a X X i

H= − + V0 δ(ra − Rj ),
a=1
2m a=1 j=1

where the second term can be considered as perturbation (the number of impurities
is Ni ≪ N ).

a. Calculate the second quantized form of the above operators with the help of
the following field operator:
X 1
Ψ(r) = √ eikr ak !
k
V

b. Determine the transition probability per unit time w(k → k′ ) due to the
impurity scattering, with the help of Fermi’s golden rule!

c. Determine the inverse lifetime, then calculate its average due to the random
position of the impurities!

5.6 Consider the following Hamiltonian:

L/2 "  2 #
v ∂φ(x)
Z
H= Π2 (x) + dx,
2 ∂x
−L/2

9
 where the Π and φ fields obey canonical commutation relations: [Π(x), φ(x′ )] =
−iδ(x − x′ ). Using the above equations derive the equations of motion for the
fields Π and φ! Show that these satisfy a wave equation, and v is just the velocity
of the wave.

5.7 The density-density correlation function of the ideal gas of free noninteracting
spinless bosons is given by

hn(r1 )n(r2 )i = n2 ν(r1 , r2 ) + nδ(r1 − r2 ) + n2

in thermal equilibrium, with n = N/V being the density of the gas and with n(r)
being the operator of the particle density. Calculate the function ν in the following
steps:

a. Write down the field operator of the bose gas on the one particle basis which
fits best for the problem, then give the second quantized form of the Hamil-
tonian H of the system and that of the operator H − µN !

b. Write down the second quantized form of the particle density operator, then
using this result calculate the thermodynamic expectation value hn(r1 )n(r2 )i,
and the resulting ν(r1 , r2 ) function! Show that ν(r1 , r2 ) depends only on
the difference of its arguments! (Guidance: according to Wick’s theorem
ha+ + + + + +
i aj ak al i = hai aj ihak al i + hai al ihaj ak i.)

6. Electrons I.

6.1 The Hamiltonian of a spin S = 1/2 interacting fermi system is:

N
h̄2 X 2 1 X
H=− ∇ + V (ri − rj ).
2m i=1 i 2
i6=j

The wavefunction of the (k, σ) one particle eigenstate is

1
ϕk,σ (r, β) = √ eikr δσβ ,
V

where σ, β =↑, ↓, and these wavefunctions form an orthonormal system:

XZ
d3 r ϕ̄k,σ (r, β)ϕk′ ,σ′ (r, β) = δkk′ δσσ′ .
β

10
 Prove that the second quantized form of H is
X h̄2 k 2 1 X X
H= a+
k,σ ak,σ + Ṽ (q)a+ +
k+q,σ ak′ −q,σ ′ ak ,σ ak,σ ,
′ ′
2m 2V
k,σ k,k ,q σ,σ
′ ′

where Z
Ṽ (q) = d3 re−iqr V (r)

is the Fourier transformed pair interaction.

6.2 Prove that the expectation value of the Hamiltonian of exercise 6.1 in the many
body state characterized by theP set of occupation numbers {nk,σ } (often called
the distribution function) is ( k,σ nk,σ = N = const.):

E{nk,σ } = h{nk,σ }|H|{nk,σ }i =

X h̄2 k 2 1 X
= nk,σ − Ṽ (k − k′ )nk,σ nk′ ,σ ,
2m 2V
k,σ ′ k,k ,σ

if we disregard a constant independent of the {nk,σ } distribution.

6.3 The result of exercise 6.2 yields the total energy of interacting electrons up to first
order in the perturbation series with respect to the interaction, where we kept only
the exchange term omitting the structureless direct term.

a. In the spirit of Fermi liquid theory the one particle spectrum can be calcu-
lated from the change in total energy due to the change in the occupation
numbers: X
δE = εσ (k)δnk,σ ,
k,σ

where εσ (k) depends on the set {nk,σ } in general. Prove that the quasipar-
ticle spectrum is

h̄2 k 2 1 X
εσ (k) = − Ṽ (k − k′ )nk′ ,σ .
2m V ′
k

(Utilize that Ṽ (−q) = Ṽ (q) if V (−r) = V (r).)

b. By definition, the velocity of the quasiparticles is

1 ∂
vσ (k) = εσ (k).
h̄ ∂k
Prove that
h̄ 1 d3 k ′ ∂
Z
vσ (k) = k − 3
Ṽ (k − k′ ) ′ nk′ ,σ .
m h̄ (2π) ∂k

11
 1 d3 k
P R
(Utilize that V k → (2π)3
, and perform partial integration.)

c. At zero temperature nk,σ = 1, if k < kF , and nk,σ = 0, if k > kF , therefore

∂ k
nk,σ = −δ(k − kF ) .
∂k k

If the interaction is spherically symmetric, i.e. Ṽ (k − k′ ) = Ṽ (|k − k′ |), then

the velocity can be brought to the form v(k) = h̄k/m∗ , where m∗ depends
only on the length of k. Prove that on the Fermi surface (for k = kF )

1 1 kF
Z
= + dΩV̂ (ϑ) cos ϑ,
m∗ m (2π)3 h̄2

where V̂ (ϑ) = Ṽ [2kF sin(ϑ/2)].

6.4 Suppose that the interaction in exercise 6.3 is the unscreened Coulomb interaction:

4πe2
Ṽ (q) = .
q2

a. Prove that the quasiparticle excitation spectrum at zero temperature is

h̄2 k 2 2e2 kF
ε(k) = − F (k/kF ),
2m π
where
1 1 − x2 1+x
F (x) = + ln .
2 4x 1−x
Sketch the function F (x)!

b. Prove that the ground state energy per unit volume is

E h̄2 kF5 e2 kF4

= − .
V 10π 2 m 4π 3

c. Calculate the ground state energy per particle E/N ! (Use the relation be-
tween the particle density N/V and the Fermi wavenumber kF .)

d. The natural unit of atomic energies is the Rydberg. 1Ry= e2 /2a0 (= 13.6eV),
where a0 = h̄2 /me2 (= 0.529Å) is the Bohr radius. Express the result of point
c. in Ry! (The result depends only on the value of a0 kF .)

e. The radius rs defined by

V 4π 3
= r
N 3 s
12
 characterizes the average distance of the particles. What is the relation
between kF and rs ? Write down the result of point d. as a function of a0
and rs ! Evaluate the numerical coefficients!

6.5 We have seen in exercise 6.4 that the ground state energy per particle of the
interacting electron gas is
"  2 #
E a0 a0
= 2.21 − 0.916 Ry,
N rs rs

where a0 is the Bohr radius, while rs is the radius of a sphere, the volume of
which is equal to the system volume per electron number. The first term is the
kinetic energy and the second is the Coulomb exchange energy, i.e. the first order
correction from the perturbation series with recpect to the interaction. Obviously,
this expression is reliable only if the correction is relatively small, i. e. if a0 /rs is
large, in other words for dense electron gas. In case of a dilute electron gas (large
rs /a0 ) the interaction term dominates, and therefore the nature of the ground state
changes radically, the electrons may order themselves into a regular crystal lattice
(Wigner crystal). Calculate the energy of the Wigner crystal in the following steps:

a. Suppose that any single electron of the system is allowed to move within a
sphere of radius rs around its lattice point, and the compensating charge of
the ions is uniformly distributed within the sphere. Prove that if the electron
is located at a distance r from the center of the sphere, then the potential
energy of the system has the form
V (r) = α + βr 2 .
Evaluate the constants α and β! (Do not forget about the energy of the
charged sphere either!)

b. Supplementing the above potential energy with the kinetic energy of the
electron, the full Hamiltonian reads as
p2
H= + βr 2 + α,
2m
which defines a harmonic oscillator with respect to the electron. Calculate
the frequency ω of this oscillator using the result for β obtained in point a.!

c. Clearly, the ground state energy per electron of the Wigner crystal is given
by the ground state energy of the Hamiltonian in point b. How much is this
energy? (Do not forget, that there is a three dimensional oscillator in point
b.!) Express this energy in Ry with the help of a0 /rs !

d. The expectation value of the squared amplitude of a harmonic oscillator in

its ground state is
h̄
hx2 i = .
2mω
13
 p
How much is the relative expectation value hr 2 i/rs of the distance of the
electron from its equilibrium position? Express the result with the help of
a0 /rs !

6.6 Let hat be the spin independent Hamiltonian of an idealized two level atom:

hat ϕα (r) = εα ϕα (r),

where α = 1, 2. We turn on a homogeneous electric field E parallel to the z direc-

tion, in order to polarize the atom. The interaction of a single electron with the
field is described by h′ = −Ed, where d = −er is the dipole vector of the electron.
Write down the field operator for the electron, then give the second quantized form
of the full Hamilton operator H of the noninteracting atomic electron system (do
not forget about the spin). Is the atom polarizable, if the states ϕα have spherical
symmetry?

7. Electrons II.

7.1 The Hamiltonian of the one band Hubbard model in Wannier representation is
X X
H =t a+
R,σ aR+d,σ + U nR,↑ nR,↓ .
R,d,σ R

Introduce the following linear combinations of the Wannier operators:

1 X −ikR
ak,σ = √ e aR,σ .
N R

a. Prove that the inverse (Bloch → Wannier) transformation has the form

1 X ikR
aR,σ = √ e ak,σ .
N k

b. The kinetic energy term of H can be diagonalized by the above transforma-

tion as: X
H kin = ε(k)a+
k,σ ak,σ .
k,σ

Determine ε(k)!

14
 c. Perform the above transformation on the interaction term of H as well! Com-
paring this with the result of exercise 6.1, what difference can be seen?

7.2 Show that considering electron-phonon scattering there will be Umklapp pro-
cesses, meaning that the change in the wavenumber of the electron is equal to
the wavenumber of the phonon only up to a reciprocal lattice vector! Guidance:
the second quantized form of the operator eiqr should be obtained, the wavefunc-
tion of the electron, according to the Bloch theorem is ϕk (r) = eikr uk (r), where
uk (r) is lattice periodic.

7.3 Show that, similar to electron-phonon scattering, there are Umklapp processes
considering electron-electron scattering as well, during which the total wavenumber
is conserved only up to a reciprocal lattice vector! Guidance: according to the
Bloch-theorem ϕk (r) = eikr uk (r), where uk (r) is lattice periodic, and can be
expanded in Fourier series.

7.4 In the unit cell of a one dimensional lattice there are two atoms (with a single
energy level each). The energies of the levels are ε1 and ε2 , the values of the
two nearest neighbor hopping integrals are t1 and t2 (both real). The second
quantized form of the Hamiltonian of this system (without interaction) in Wannier
representation is
X X
H kin = ε1 a+
1,R,σ a1,R,σ + ε2 a+2,R,σ a2,R,σ +
R,σ R,σ

X
+t1 (a+ +
1,R,σ a2,R,σ + a2,R,σ a1,R,σ )+
R,σ
X
+t2 (a+ +
1,R,σ a2,R−a,σ + a2,R−a,σ a1,R,σ ),
R,σ

where a is the unit translation vector. Using the usual transformation (i = 1, 2)

1 X ikR
ai,R,σ = √ e ai,k,σ ,
N k

H kin can be brought to the following form:

XX
H kin = a+
i,k,σ hij (k)aj,k,σ ,
k,σ i,j

where the different k values have already decoupled, but the operator is not yet
fully diagonalized.

a. Calculate the matrix elements hij (k) (i, j = 1, 2)!

b. The dispersion relation of the two band model is given by the eigenvalues
E± (k) of the 2 × 2 matrix {hij (k)}. Determine the E± (k) band structure!

15
 c. Supposing that t1 and t2 have the same sign, how big is the gap in the
spectrum?

d. Sketch the band structure (E± (k)) for ε1 = ε2 = 0, t1 = t2 = t!

7.5 The most important part of the high temperature superconductors is the CuO2
plane, in which the copper atoms form a square lattice with lattice constant a,
while the oxigen atoms are sitting at the middle of the line segments connect-
ing nearest neighbor copper atoms. The simplest model takes into account the
hopping integrals between the copper and its neighboring oxigens only. Due to
symmetry reasons and to the properties of the atomic orbitals participating in the
hybridization (copper d and oxigen p orbitals) the hopping integral between the
copper and the oxigens located in the positive x and y direction from the copper
is t, while it is −t towards the oxigens located in the negative x and y direction
from the copper.

a. Sketch the structure of the CuO2 plane, identify the unit cell and the nearest
neighbor hopping integrals!

b. Let the site energy on the copper be ε, while that on the oxigen be 0. Write
down the Hamiltonian (self adjoint!) of the system in Wannier representa-
tion, if a+
i,R,σ and ai,R,σ creates and annihilates an electron in unit cell R
on the i-th atom (i = 1, 2, 3)! Use the formalism developed in the previous
exercise!

c. Applying the following transformation

1 X ikR
ai,R,σ = √ e ai,k,σ
N k

the Hamiltonian can be brought to the following form:

XX
H= a+
i,k,σ hij (k)aj,k,σ .
k,σ i,j

Give the matrix elements hij (k)!

d. The model’s dispersion is given by the eigenvalues of the 3×3 matrix {hij (k)}.
Determine the band structure!

7.6 The magnetization (for small magnetic fields) of the Hubbard model is given by

hMz i = V χ0 Hef f

in mean field approximation, where

2 X
χ0 = µ2B {−f ′ [ε(k) − µ]}
V
k

16
 is the susceptibility of the noninteracting system, and

U hMz i
Hef f = H +
2N µ2B

is the effective field. Then the magnetization can be brought to the form

hMz i = V χH,

which defines the susceptibility χ of the interacting system in mean field approxi-
mation. Suppose that the hopping integral t → 0, i.e. it is getting difficult for the
electrons to move. The band gets narrow, and the density of states

2 X 2
g(ε) = δ[ε − ε(k)] → δ(ε).
V Vc
k

Then at finite temperatures the derivative of the Fermi function in the expression
for χ0 can obviously not be approximated by the Dirac delta function.

a. Calculate the noninteracting susceptibility χ0 for this narrow band case! Give
the result as a function of n = Ne /N , the average electron number per lattice
site!

b. According to the theory of atomic paramagnetism, the Curie susceptibility

of S = 1/2 spins localized on lattice sites is:

µ2B 1
χCurie
0 = .
Vc kB T

How much is the susceptibility χ0 calculated in point a. in the correspond-

ing n = 1 case (half filled band)? What could be the explanation for the
difference between χ0 determined in such a way and χCurie
0 ? If the question
is too difficult, continue with point c.

c. Calculate the susceptibility χ of the interacting (U 6= 0) system in mean

field approximation for arbitrary n! Does the expression for χ indicate phase
transition at finite temperatures? If yes, what is the critical temperature Tc ?
What are χ and Tc for a half filled band?

7.7 The problem in exercise 7.6 (t → 0 Hubbard modell) is exactly solvable, since the
individual lattice sites decouple completely. This is a good occasion to demonstrate
the limits of the mean field approximation.

a. First prove that the magnetic term

X
HZeeman = µB H σa+
k,σ ak,σ
k,σ

17
 is expressed with the a+
R,σ and aR,σ Wannier operators as
X
HZeeman = µB H σa+
R,σ aR,σ .
R,σ

(Utilize the Fourier transformation encountered in exercise 7.1)

b. Then clearly, the full Hamiltonian is

X
H(t → 0) = H(R),
R

where
H(R) = µB H(nR,↑ − nR,↓ ) + U nR,↑ nR,↓
depends only on the occupation numbers of a single lattice site only. The
grand canonical partition function of the full system is Ξ = ΘN , where Θ is
the grand canonical partition function of a single site. The following states
are possible on a single lattice site:

0 electron (nR,↑ = nR,↓ = 0)

1 electron in the state ↑ (nR,↑ = 1; nR,↓ = 0)

1 electron in the state ↓ (nR,↑ = 0; nR,↓ = 1)

2 electrons (nR,↑ = nR,↓ = 1)

How much is the energy E of these states? Write down the partition func-
tion Θ! (Let the chemical potential be µ, and use the expression Θ =
P −β(E−µN)
e where N is now the number of particles on the given site in
the given state.)

c. The grand canonical thermodynamic potential of the full system is

1 N
Ω=− ln Ξ = − ln Θ.
β β
The number of electrons is Ne = −∂Ω/∂µ, while the magnetization of the
system is Mz = −∂Ω/∂H. Using the results of point b. determine the
quantities n = Ne /N and m = Mz /N as a function of µ and H!

d. What is the susceptibility

1 ∂Mz
χ=
V ∂H
of the system in the limit H → 0 as a function of µ?

e. Using the result obtained in point c. express the chemical potential µ with
the help of n, in the limit H → 0! What is µ for a half filled band (n = 1)?

18
 f. What is the susceptibility determined in point d. for n = 1? Does this exact
result indicate a phase transition at some finite temperature? Expand the
inverse of the exact susceptibility in powers of U ! What do you experience by
comparing this with the mean field result of point 7.6.c? What is the exact
susceptibility for U ≫ kB T ? Compare this with the Curie susceptibility of
point 7.6.b! Is the answer to the last question of point 7.6.b any easier now?

7.8 The half filled Hubbard model is described by the following Hamiltonian:
X X 1 1
H = −t (c+
i,σ ci+1,σ + h.c.) + U (ni↑ − )(ni↓ − ).
2 2
i,σ i

Let’s define
P the following operators: P
J z = 12 i (c+ +
1), J + = i (−1)i c+ + − i
P
c
i↑ i↑ + c c
i↓ i↓ − i↓ ci↑ , J = i (−1) ci↑ ci↓ .

a. Prove that these operators satisfy the usual SU (2) commutation relations!

b. Is the operator J a conserved quantity?

8. Condensates

8.1 In case of a spin density wave in mean field approximation the Hamiltonian can
be diagonalized by the quasiparticle operators

d+,k,σ = σe−iϕ uk ak,σ + vk ak−2kF ,σ

d−.k.σ = σe−iϕ vk ak,σ − uk ak−2kF ,σ

to obtain the following expression:
X q
H= α ξk2 + |∆|2 d+
α,k,σ dα,k,σ .
α,k,σ

Here v "
u #
u1 ξ k
uk = t 1 + p 2 ,
2 ξk + |∆|2
p
vk = 1 − u2k , α = ±, and k takes values from the reduced Brillouin zone corre-
sponding to the new periodicity. The ground state of the system can be described
by Y
|ψ0 i = d+
−,k,σ |0i,
k,σ

19
 and the ground state energy E0 = hψ0 |H|ψ0 i is independent of the phase ϕ of the
spin density wave. Consider a state, which is obtained from the ground state in
such a way, that we change the phase relations of one quasiparticle (−, k0 , σ0 ):
Y
|ψi = [σ0 e−i(ϕ+δϕ) vk0 ak0 ,σ0 − uk0 ak0 −2kF ,σ0 ]+ d+
−,k,σ |0i.
(k,σ)6=(k0 ,σ0 )

Prove that the energy expectation value of this state exceeds E0 by the following
amount:
|∆|2
δE = q [1 − cos(δϕ)].
ξk20 + |∆|2

8.2 The one band Hubbard model for spinless bosons is

X UX
H= ε(k)a+
k ak + nR (nR − 1),
2
k R

where nR = a+R aR , and the relation between operators acting in real and momen-
tum space has the usual form:

1 X −ikR
ak = √ e aR
N R

(N is the number of lattice points). The ground state of noninteracting (U = 0)

bosons is the Bose condensed state
1
|ψ0 i = √ (a+ )NB |0i,
NB ! k=0

where |0i is the vacuum state, NB√is the

Pnumber of bosons, and each boson is in
the state k = 0. Since ak=0 = (1/ N ) R aR , therefore the real space structure
of this ground sate is

1 1 X
|ψ0 i = √ √ NB a+ + +
R1 aR2 ...aRN |0i.
NB ! N R1 ,R2 ,...,RNB
B

Following Gutzwiller we can imagine the ground state of the strongly interacting
(U → ∞) system in such a way, that we omit those states from the above linear
combination, which contains two (or more) bosons on the same lattice site. (Then
clearly NB ≤ N has to be satisfied.) The trial wavefunction obtained in such a
way (by projection) is:

1 1 X
|P ψ0 i = √ √ NB a+ + +
R1 aR2 ...aRN |0i.
NB ! N R1 6=R2 6=...6=RNB
B

20
Prove that in this state the expectation value of the occupation number of the
k = 0 one particle state nk=0 = a+
k=0 ak=0 given by

hP ψ0 |nk=0 |P ψ0 i
hnk=0 i =
hP ψ0 |P ψ0 i

is evaluated as:  
NB − 1
hnk=0 i = NB 1 − ,
N
i.e. it is the total number of bosons reduced by a factor depending on the density
of bosons.

21