SUPERCONDUCTORS & SUPERFLUIDS

Andrei Golov
September 2023

1 Quantum gases, BEC

Novel Phenomena

Bose-Einstein Condensation (BEC) – the condensation of many identical bosons

in the same single-particle quantum state of the lowest energy, in which all of
them are described by the same wavefunction with a meaningful complex phase
θ(r) (we call this macroscopic phase coherence).

Superfluidity is the ability of a fluid to have a long-lived current-carrying

state. Observed in liquid 4 He below Tλ = 2.17 K (1938) and in liquid 3 He below
Tc ≈ 1 mK (1972).

Superconductivity is the superfluidity of an electron liquid; first observed in

mercury below Tc = 4 K (1911).

Quantum vs. classical gas (or liquid)

In an ideal monoatomic gas, the mean kinetic energy of particles is

p2 3
= kB T.
2m 2
At low temperature when their de Broglie wavelength,
h
λdB = ∼ h(3mkB T )−1/2 ,
p

exceeds the average distance between them, n−1/3 , particles’ individual trajec-
tories lose sense and we have to describe the gas in terms of a many-particle

1
wave function. This characteristic temperature of “quantum degeneracy” is thus

h2 n2/3
T∗ ∼ . (1.1)
3mkB
Below T ∗ the quantum statistics (whether bosons or fermions) is important.
Bosons make “Bose-Einstein condensate”; fermions make “Fermi sea”.
The particle’s total spin, sh̄, defines the statistics. Bosons (photon, phonon,
4
He, 1 H, 23 Na, 87 Rb) have s integer. Fermions (quark, electron, neutron, proton,
3
He, 86 Rb) have s half-integer.

Quantum liquids: metastable cold gases, liquid helium,

electrons in metals.

Interacting particles normally solidify at temperatures higher than T ∗ , hence,

can’t be made a quantum fluid. There are two ways around this constrain:

a) Quantum effects can keep some of them liquid down to low temperatures.
Two types of quantum liquids, stable at T = 0, are available:
(i) liquid 3 He and 4 He. Without applying substantial external pressure, helium
does not solidify down to T = 0 – because its zero-point kinetic energy is large
(due to the very weak interatomic attraction and low atomic mass m) and is
comparable with the energy of interatomic attraction;
(ii) electrons in metals (conduction electrons can be thought of as a liquid filling
the metal; here T ∗ ∼ 105 K is huge compared to even room temperature T ∼
300 K).

b) Metastable cold gases can be prepared.

How to prepare a metastable (long-lived) cold gas:

a) Reduce the particle density to ∼ 10−5 of that of air to slow down their re-
combination into molecules (which requires three-particle collisions to take away
energy from recombination); this should be done in a sealed chamber.
b) Prevent atoms from reaching the walls of the chamber (by a magnetic trap
– this only works with particular spin states of atoms).
c) Cool down below T ∗ in two steps:
(i) laser precooling (red-shifted laser tricks atoms into absorbing less and then
emitting more energy);
(ii) evaporative cooling by gradual reduction of the depth of the trapping poten-
tial well: the fastest particles keep escaping leaving slower ones behind (however,

2
this crucially relies on the interaction between particles for their thermalization;
i.e. will not work with a truly ideal gas).

Non-interacting Bose gases. BEC.

As a simple example, consider the quantum states of a free particle in a cubic

box of volume V with periodic boundary conditions, described by wavefunction:

Ψk = V −1/2 exp(iθ1 ) exp(ik · r). (1.2)

They are labeled by the allowed wavenumbers k that match the boundary con-
2 2
ditions. The energy levels are ϵ(k) = h̄2m
k
, where k = |k|. The state of the
lowest energy ϵ = 0 corresponds to k = 0, and its wavefunction is

Ψ0 = V −1/2 exp(iθ1 ). (1.3)

For a single particle, the arbitrary choice of the phase θ1 has no observable
consequences.

For N bosons (density n = N/V ), below the critical temperature for ‘Bose-
Einstein condensation’ (BEC)

2πh̄2  n 2/3
Tc = , (1.4)
mkB 2.612
a macroscopic number of particles N0 condenses into the same quantum state
with the lowest energy so that
 3/2
N0 (T ) T
=1− , (1.5)
N Tc

leaving the rest (‘excitations’) in higher-energy states

3/2
N′

T
= . (1.6)
N Tc

This is a phase transition into a more ordered state (in momentum space).

Macroscopic wavefunction

In the Bose condensate at rest (i.e. k = 0), all N0 particles are in the same
quantum state described by wavefunctions (1.3) in which the phases θ1 become
−1/2
close to each other, with a very small uncertainty ∆θ1 ∼ N0 ≪ 1 when their

3
number N0 is large. Hence, we can use their mean phase θ =< θ1 > as another
macroscopic variable alongside with their density n0 = N0 /V . Thus properties
of all particles in the condensate are described by just two parameters, n0 and
θ, through the ‘macroscopic wavefunction’
 1/2
N0 1/2
ψ= exp(iθ) = n0 exp(iθ), (1.7)
V

Its interpretation for an ensemble of N0 indistinguishable particles is the same

as for the single-particle wavefunction (1.3): the probability of finding a particle
in a small volume dV near point r is equal to |ψ(r)|2 dV .

More generally, n0 (r) and θ(r) may depend on position in space r:

(i) when the condensate’s density n0 (r) varies with r, then so does |ψ(r)| (for
instance, while approaching the trap’s wall).
(ii) when the condensate moves, as a whole, at velocity vs relative to the labo-
ratory reference frame, each its particle acquires the momentum mvs = h̄k (de
Broglie’s relation), and the phase of their wavefunctions (1.2) becomes depen-
dent on position r through the factor exp(ik·r) = exp(i m
h̄ vs ·r). Thus, the phase
of the macroscopic wavefunction θ acquires gradient ∇θ = m m
h̄ ∇(vs · r) = h̄ vs ,
which results in the relation between the superfluid velocity vs and ∇θ,
h̄
vs = ∇θ. (1.8)
m

Exercises

1. Repeat the derivation of expression (1.1).

2. Estimate Tc for cold gas of 87 Rb at a realistic density (do some research).
3. Sketch N0 (T ) and N ′ (T ) from (1.5) and (1.6).
4. Repeat the derivation of (1.8).
5. Work through Section 1 (Quantum Gases) of Problem Sheet 1.

4
2 Superfluid helium and macroscopic wavefunc-
tion

Superfluid helium (4 He) and macroscopic coherence

At low temperatures and pressures, helium liquefies but never solidifies – thanks
to the low mass of its atoms and very weak attraction between them (and hence a
relatively substantial kinetic energy of zero-point motion). We thus have a liquid
at arbitrarily low temperatures! At Tλ = 2.17 K (at P = 0) liquid 4 He undergoes
a sharp phase transition into a state with a new internal order. Historically, this
state was called “He II” (and the normal state at T > Tλ is “He I”) but we will
refer to it as “superfluid 4 He” or simply “superfluid helium”. In this state,
steady flow is possible without any losses, heat transport is extremely fast,
oscillating flow behaves as if there were some viscous (“normal”) component in
the liquid but of a reduced density ρn (T ) < ρ (where ρ is the density of all
liquid); at T = 0 this “normal component” vanishes: ρn (0) = 0. The remaining
component of density ρs (T ) = ρ − ρn (T ) behaves as an inviscid ordered liquid
(called the “superfluid component”).

Macroscopic coherence

Neutron-scattering experiments revealed the presence of a Bose-Einstein con-

densate (i. e. helium atoms with truly zero velocity), although their fraction
makes only about 7% of the total number of atoms (this is because of the
strong repulsion between atoms in the liquid state). Yet, the number of these
atoms making BEC is macroscopic and can hence control the dynamics of all
superfluid component. In thermodynamic equilibrium, the centre of mass of all
the superfluid component has to be at rest with the condensate (i. e. all super-
h̄
fluid component moves with the condensate’s velocity vs = m ∇θ) – thanks to
the presence of the coherent phase θ(r) of the condensate wavefunction. Indeed,
experiments (e. g. with rotating helium, see below) revealed the presence of a
h̄
coherent phase θ(r) that governs the velocity vs (r) = m ∇θ (1.8) at which the
superfluid component flows.

One immediate consequence of vs ∝ ∇θ (1.8): vs is irrotational everywhere

except on the axis of a quantized vortex line,

∇ × vs = 0. (2.9)

5
Quantization of circulation

The circulation of vs is quantized:

I I  
h̄ h̄ h
vs · dl = ∇θ · dl = ∆θ = N = κN. (2.10)
m m m

Hence in a multiply-connected volume (such as an annulus), once the superflow

is induced it can’t slow down continuously because only discrete values of ve-
locity are allowed. This is based on the condition that the total change of the
phase of the macroscopic wave function after going round the annulus, ∆θ, can
only be equal to an integer number of 2π. I the presence of quantized vortices
the flow might become dissipative: each vortex takes away a phase different of
2π (phase slip) when passes across the flow – thus reducing the flow velocity vs .

Quantized vortices

To allow rotation, the magnitude of the order parameter |ψ(r)| becomes inho-
mogeneous, creating continuous lines along which ψ = 0 (hence, making the
phase θ undefined on those lines). This makes the superfluid formally multiply-
connected, so non-zero circulation around such a line becomes possible. The
superfluid velocity around the vortex line is vs (r) = κ/2πr.
2
κ
The vortex energy per unit length is ρs 4π ln(R/a0 ), where R is the radius of
container (or half the distance to the nearest vortex in an array). Rotation of
the superfluid component at angular velocity Ω in a simply-connected volume
takes forms of an array of quantized vortex lines, each carrying one quantum of
circulation, κ = h/m, and the areal density of lines is 2Ω/κ (‘Feynman’s rule’).

Exercises

1. Look up the lecture materials and review main properties of superfluid he-
lium.
2. Repeat derivation of the quantization of circulation.
3. Calculate the value of the quantum of circulation κ.
4. Sketch the succession of flow patterns (velocity and phase maps, as seen from
above) while one vertical vortex traverses a flow across the channel (phase slip,
see lecture materials).
5. Estimate the distance between vortex lines in a bucket of superfluid helium
rotating at Ω = 1 rad s−1 .

6
3 Excitations. Two-fluid description
Ground state and excitations. Landau criterion.

Particle-like excitations of all liquid have the momentum p and energy ϵ(p).
Landau criterion for superfluidity: non-dissipative motion at a speed v past an
object (wall) is possible if v < vL :
 
ϵ
vL = min . (3.11)
p

The ground state of the non-interacting Bose gas is when all particles are in
the lowest energy state. Hence, excitations are associated with free particles
in higher energy states. In contrast, weakly repulsive particles have a wave-
like collective mode (sound) as low-energy excitations, with dispersion ϵ = cp.
Hence, a non-interacting Bose-Einstein condensate (ϵ = p2 /2m) is not superfluid
(vL = 0)! But a weakly repulsive Bose gas (ϵ = cp for p → 0) is superfluid
(vL = c > 0). So is liquid helium below Tλ .

Excitations in superfluid helium have a complicated spectrum ϵ(p), however

there are two regions of special importance: phonons at small p, ϵ = c1 p (sound
velocity c1 = 240 ms−1 ), and rotons near the minimum at p = p0 , ϵ = ∆ + (p −
p0 )2 /2mrot (∆/kB = 8.65 K, p0 /h̄ = 19.2 nm−1 , mrot = 0.16m4 ). The Landau
critical velocity is hence vL ≈ ∆/p0 = 60 ms−1 .

The normal component, thermodynamics

The normal component is a gas of thermally excited particle-like excitations
(phonons and rotons) with momentum p and energy ϵ(p). Their energy and
entropy determine the thermodynamics of He-II. When the gas of excitations
is moving, as a whole, relative to the stationary condensate at a velocity vn ,
the excitations carry the total momentum P ∝ vn . The density of the normal
component ρn is thus defined through P = ρn vn .
The rest is the superfluid component: ρs = ρ − ρn . It can carry momentum
(when moves at non-zero vs ) but no heat and entropy.

In thermal equilibrium, excitations obey the Bose-Einstein distribution,

 −1
ϵ(p)
f (p) = e kB T − 1 , (3.12)

∂U


in which the chemical potential, µ ≡ ∂N S,V
, is set to zero (µ = 0) – because
the number of excitations is not conserved (their number becomes a free param-
eter with which the system minimizes its free energy when in thermodynamic
equilibrium).

7
 The temperature T reflects the density of excitations,
Z ∞
d3 p
n(T ) = f (p) . (3.13)
0 (2πh̄)3

This integral results in the density of phonons,

 3
6 kB T
nph (T ) = , (3.14)
5π 2 h̄c1

and of rotons,
nrot (T ) ∝ T 1/2 exp(−∆/kB T ). (3.15)
Phonons dominate at T < 0.6 K and rotons – at T > 0.7 K. The phonon con-
tribution to the normal component, specific heat and entropy, ρn (T ) ∝ T 4 ,
Cph (T ) ∝ T 3 , Sph (T ) ∝ T 3 , scale as power laws of T ; while the leading term of
the roton contributions is exponential ∝ exp(−∆/kB T ).

Exercises

1. Look up the lecture materials and review main properties of excitations in

superfluid helium.  
2. Derive the formula vL = min ϵ(p)p .
3. There are two mechanisms for the dissipation of energy of flow at veloc-
ity v through a channel of width d: microscopic (emission of excitations when
v > vL ) and macroscopic (phase-slips due to quantized vortices moving when
v > vc , where vc is of order ∼ κ/d). Which mechanism is usually relevant and
why?

8
4 Two-fluid hydrodynamics

The hydrodynamic description of a system of molecules is an approximation

that deals with averaged macroscopic properties such as local pressure, density,
temperature, velocity of fluid. It only works on length scales large compared
to the mean free path of molecules, and when the local thermal equilibrium
is established. In the case of superfluid helium, the gas of excitations can be
treated this way when it is sufficiently dense (the mean free path of phonons and
rotons is small comparable to the wavelength of perturbation), i.e. typically at
temperatures above ∼ 0.6 K.

Classical Fluids

Let us firstly refresh our knowledge of the hydrodynamics of classical (one-

component) fluids. The macroscopic variables, all depending on position and
time (r, t), are density ρ, pressure P , velocity v.

The conservation of matter (mass): continuity equation,

∂ρ
+ ∇ · (ρv) = 0. (4.16)
∂t
The conservation of momentum (Newton’s 2nd law): Navier-Stokes equation,
 
∂v
ρ + v · ∇v = −∇P + η∇2 v. (4.17)
∂t

For sufficiently small velocities, the LHS can be linearized (only term linear in
v left):
∂v
ρ = −∇P + η∇2 v. (4.18)
∂t

A solution for a small-amplitude compression wave, i. e. sound, when vis-

cosity is small (η = 0, in this case the Navier-Stokes equation becomes Euler’s
equation): pressure P (t) and density ρ(t) wave, while T = const and s = const.
Harmonic oscillations of ρ, P , v propagate with velocity c,
 
∂P
c2 = . (4.19)
∂ρ s

The relaxation time τ (mean free path l = v̄τ ) is set by collisions with
other particles of fluid (Poiseuille regime when this is dominant), volume de-
fects (Drude regime when this is dominant – often for electrons in metals) and

9
ultimately (for either low density of particles or pure metals) by boundaries such
as pipe walls or surface of wire (Knudsen regime).

For the Poiseuille regime in a cylindrical pipe of radius R and pressure drop
∆P over distance ∆x (∇P = ∆P ∆x ), the total volume throughput per unit time
in the steady-state is
dV πR4
= ∇P (4.20)
dt 8η
(i. e. requires pressure gradient). For a gas with the mean velocity of particles
v̄, the viscosity is η = 31 ρlv̄. Similarly, the thermal conductivity is κ = 13 CV lv̄.

Equations of Two-fluid Hydrodynamics

The conservation of total mass (continuity equation),

∂ρ
+ ∇ · (ρs vs + ρn vn ) = 0. (4.21)
∂t
The conservation of entropy S (per unit volume of all fluid) for reversible pro-
cesses (only the normal component carries entropy, at velocity vn ):

∂S
+ ∇ · (Svn ) = 0. (4.22)
∂t
The dynamics of the superfluid component (the force on atoms of the frictionless
superfluid component is due to the gradient of their potential energy µ),
 
∂vs
m + v · ∇vs = −∇µ = −v∇P + vS∇T,
∂t

where the Gibbs-Duhem relation, dµ = vdP −vSdT , was used (µ being chemical
potential per atom, v = m/ρ atomic volume, m mass of a 4 He atom). For small
velocities vs it can be linearized and re-written as
∂vs 1 S
= − ∇P + ∇T. (4.23)
∂t ρ ρ

The conservation of the total momentum (here ηn is the viscosity of the normal
component):
∂vs ∂vn
ρs + ρn = −∇P + ηn ∇2 vn . (4.24)
∂t ∂t

Additionally, the superfluid velocity is irrotational:

∇ × vs = 0. (4.25)

10
Fountain effect

With superfluid helium in a pipe, one can administer a pressure gradient along
it by maintaining a difference in temperatures of pipe’s ends – and hence push
the fluid through the pipe without any pump. This is called fountain effect.

Suppose two reservoirs with superfluid helium are connected by a pipe. If

temperatures of helium in them (T1 and T2 ) are different, what will be the
difference between pressures at the opposite ends of the pipe (P1 and P2 )? To
have a steady-state situation, ∂vs /∂t = 0, from Eq. 4.23:

∇P = S∇T. (4.26)

To maintain this pressure difference, the vertical levels of free surfaces of helium
in two reservoirs have to be different by ∆H given by:

ρg∆H = P2 − P1 = S(T2 − T1 ). (4.27)

Two sound modes

Remember problems with one-dimensional harmonic oscillators? A single body

on a spring has a single natural frequency. But two coupled bodies would nor-
mally have two normal modes with different frequencies and different symmetries
of oscillations (say, in one both bodies oscillate in phase, while in another they
oscillate out-of-phase). In a similar fashion, the presence of two components
in superfluid helium results in two oscillatory modes (sounds), not one as in a
classical one-component fluid.

One can see that the superfluid and normal components are accelerated in
the same direction when acted upon by a pressure gradient ∇P , but are accel-
erated in the opposite directions when acted upon by a temperature gradient
∇T . This results in two different types of oscillating modes (sounds).

First sound is a wave of pressure and density, while T = const and S = const,
vn = vs . Oscillations of ρ(x, t) and P (x, t) propagate with the velocity c1 , given
by the adiabatic compressibility like in any other fluid :
 
2 ∂P
c1 = . (4.28)
∂ρ S

Second sound a wave of temperature and entropy, while P = const and

ρ = const, ρn vn = −ρs vs (no mass flow). Oscillations of T (x, t) and S(x, t)

11
propagate with the velocity c2 ,

ρs S 2 T
c22 = , (4.29)
ρn ρCP

where CP is the specific heat (per unit volume) at constant pressure.

Exercises

1. Look up the lecture materials and review main differences between the equa-
tions describing the hydrodynamics of classical fluids and of superfluid helium.
2. Sketch instantaneous velocities of the superfluid and normal components and
the associated variations of:
(i) pressure and density in a plane wave of first sound;
(ii) temperature and entropy in a plane wave of second sound.
3. Sketch the temperature dependence of the velocity of second sound c2 (T ).
4 (optional). Derive the formulas for the velocities of first and second sound in
superfluid helium.

12
 Mid-term checklist
(superfluid helium means ‘liquid 4 He below Tλ = 2.17 K’):

1. Quantum vs. classical fluids, estimate of relevant temperatures; what

keeps helium liquid down to T = 0?
2. Bosons and fermions, Bose-Einstein condensation, temperature depen-
dence of the density of condensate and non-condensate.
3. Macroscopic wave function of the ground state, interpretation of its am-
plitude and phase, superfluid velocity and quantization of its circulation.
4. Simply-connected and multiply-connected geometries; persistent flow in
an annular channel and its stability.
5. Quantized vortices, their motion as a mechanism of dissipation of the
energy of superflow.
6. Excitation spectrum in energy-momentum space; phonons and rotons in
superfluid helium, excitations in a gas of non-interacting and weakly-
repulsive bosons.
7. Disspation of energy of superflow through creation of excitations, Landau
velocity and criterion for superfluidity.
8. Normal component as a gas of thermal excitations; temperature depen-
dence of the density of the normal component and specific heat of super-
fluid helium.
9. Two-fluid hydrodynamics, fountain effect, general properties of the first
and second sound.

13
5 Electrodynamics of Superconductors
Revision: Electromagnetic Fields (valid for all materials)
Fundamental fields E and B can be determined by measuring the force F
they exert on a charge, q, moving at a speed v:

F = q(E + v × B)

Electrostatic and vector potentials, ϕ and A, are introduced so that

E = −∇ϕ

B=∇×A
For the same fields, E and B, ϕ and A are not uniquely defined. We often fix
the uncertainty by choosing ϕ(∞) = 0 and ∇ · A = 0

For a material with macroscopic dielectric polarization, P, and magnetiza-

tion, M, quantities D and H are defined:

D = ϵ0 E + P,

H = B/µ0 − M,
−12
where ϵ0 = 8.85 · 10 F/m (or A2 s2 /Nm2 ) and µ0 = 4π · 10−7 H/m (or N/A2 ).

Maxwell’s equations (q and j are free charge and current density):

∇·B=0
∂B
∇×E+ =0
∂t
∇ · D = qn
∂D
∇×H=j+ .
∂t

Gauge invariance. All observable quantities, such as:

- field B = ∇ × A,
- field E = −∇ϕ − ∂A ∂t ,
- momentum p (the eigenvalue of the operator p̂ ≡ −ih̄∇ − qA),
∂
- energy E (the eigenvalue of the operator of energy Ê = ih̄ ∂t − qϕ),
2 ∗
- current density js = q|ψ| v (where v = p/m ),
- gradient of the ‘gauge-invariant phase γ’, ∇γ = ∇θ − h̄q A,
do not change after simultaneous transformations
from A1 , ϕ1 , θ1 , |ψ1 | (where ψ = |ψ| exp(iθ) is a wavefunction)
to A2 = A1 + ∇χ, ϕ2 = ϕ1 − ∂χ q
∂t , θ2 = θ1 + h̄ χ, |ψ2 | = |ψ1 |,
where χ(r, t) is an arbitrary real-valued (scalar) function.

14
 Properties of Normal Metals (reference)

Macroscopic variables, all depending on position and time (r, t): number
density of charged particles n, their charge q (usually, q = −e), electrostatic
potential ϕ, drift velocity v, current density j = qnv.

Conservation of charge: continuity equation,

∂n
q + ∇ · j = 0. (5.30)
∂t

On average, charge carriers (usually, electrons) respond to applied force qE:

 
dv v
m + = qE, (5.31)
dt τ

where the relaxation time τ = l/vF is introduced to account for their finite
mean free path before exchanging acquired momentum with the crystal lattice
and other electrons (vF = pF /m is the Fermi velocity).

In the Drude regime, particles mainly collide with defects randomly dis-
tributed in the volume of a metal wire. A non-zero potential gradient along the
wire, ∇ϕ = −E, is required to maintain the steady flow of charge of current
density j = qnv – hence the Ohm’s law (from Eq. 5.31):

j = σE, (5.32)

where the conductivity σ (the reciprocal of resistivity) is proportional to the

mean free path l (which can never become infinite),

ne q 2 τ ne q 2 l
σ= = , (5.33)
m mvF
where ne is the density of conducting electrons.

In metals, the behavior of electrons is similar to that of free fermions, but

with effective mass, m, that is different from the free electron mass, me , and
can be directly determined from measurements of the electronic heat capacity:
mpF
Ce = V T, (5.34)
3h̄3
where pF is the momentum of an electron with Fermi energy and its dependence
on electron density, ne = N/V , is the same as for a non-interacting Fermi gas:

pF = h̄(3π 2 ne )1/3 (5.35)

15
 ne σ −1 (77 K) σ −1 (273 K) m/me ΘD TF vF
10 cm−3
23
µΩ cm µΩ cm K 105 K 108 cm/s
Cu 0.85 0.2 1.56 1.3 310 0.82 1.57
Pb 1.32 4.7 19.0 1.9 88 1.10 1.83

Table 1: In this table, copper is an example of a good conductor which is normal

down to T = 0, and lead is a superconductor below Tc = 7.2 K.

Properties of superconductors

In some metals below a certain critical temperature, Tc , the resistance suddenly

drops to zero – transition into the superconducting state. A persistent current
can be induced in a loop of superconducting wire. The superconducting state
is a new equilibrium phase of metals: below Tc its free energy, Fs (T ), is lower
than that of the normal state, Fs (T ).

In metallic alloys Tc can be as high as 40 K. In “high-Tc superconductors”

(1986) critical temperatures can reach 130 K (even 160 K under pressure). They
generally consist of a few elements (like Ta2 Ca2 Ba2 Cu3 O10 , Tc = 125 K), with
layers made of Cu and O atoms. Fairly recently (2008) a new type of high-
temperature superconductors was discovered; they contain ions of iron, not
copper. However, most spectacular advances in raising the Tc towards room
temperatures is associated with dense assemblies of hydrogen atoms, which so
far can only be realized at very high pressures in compounds such H3 S (200 K,
2015), LaH10 (250 K, 2019) and mix of H3 S and CH4 (288 K, 2020).

Another important property of superconductors – Meissner effect, i. e. com-

plete expulsion of dc magnetic flux from the interior of the superconducting
phase. In the boundary region of a thickness, λ ∼ 300 − 500 Å, called “penetra-
tion depth”, the field rapidly decays as a result of precise compensation of the
external field by the magnetic field of an induced surface current.

- Specific heat of superconductors, at T ≪ Tc , depends on temperature as

C(T ) ∝ exp(∆0 /kB T ), (5.36)
that points at the existence of an energy gap for thermal excitations.
- Absorption of infrared photons only at frequencies higher than some threshold,
ωc (T ) – when the photon energy is sufficient to break a Cooper pair:
h̄ωc = 2∆(T ). (5.37)

- The experimentally measured gap, ∆0 = ∆(0), is comparable with Tc

∆0 ∼ kB Tc . (5.38)

16
Two-fluid model and London equations

Görter and Casimir (1934):

- The total density of electrons is formally divided into two components,
superconducting ns (T ) and normal nn (T ), n = ns + nn , carrying currents
js = (−e)ns vs and jn = (−e)nn vn in parallel.
ns (Tc ) = 0 and nn (Tc ) = ne , while ns (0) = n and nn (0) = 0.
- The normal electrons have resistance and entropy. They obey the Drude’s
formula (from Eq. 5.31 with finite τ and d/dt = 0 for frequences ω ≪ τ −1 ):

nn e2 τ
jn = σE = E. (5.39)
m

- The superconducting electrons have zero resistivity and entropy. They obey
the London equations (F. and H. London, 1935):

djs (r) ns (T )e2

= E(r), (5.40)
dt m
ns (T )e2
∇ × js (r) = − B(r). (5.41)
m
The former can be seen as a direct consequence of Eq. 5.31 with τ = ∞ (i. e.
for a perfect conductor), while the latter results from the coherent macroscopic
wavefunction (see below).

Meissner effect and penetration depth

The following equation, explaining complete screening of the external magnetic

field by the surface supercurrent (Meissner and Ochsenfeld, 1933), can be de-
rived (Example Sheet for week 5) from the second equation (5.41):

ns e2 µ0
∇2 B = B. (5.42)
m
Take a plane surface of a superconductor, x-axis perpendicular to the surface,
parallel external field Bz = B. The decaying solution is:

B(x) = B0 exp(−x/λL ), (5.43)

where
m
λ2L = . (5.44)
µ0 ns e2

In superconductors like lead λL (0) ∼ 300 − 500Å, but in transitional metals

and compounds with narrow conducting band (large m) it reaches ∼ 2000Å.

17
Macroscopic wave function for superconducting electrons

[With hindsight, we can think of charge carriers called Cooper pairs of mass
m∗ = 2m, charge q = −2e and density n∗ (T ) = ns /2]:

ψ(r) = (n∗ )1/2 exp(iθ(r)). (5.45)

Apply the gauge-invariant operator of momentum:

p̂ψ = (−ih̄∇ − qA)ψ = pψ.

Then the momentum of a pair (in magnetic field B = ∇ × A):

p = m∗ vs = h̄∇θ − qA. (5.46)

The supercurrent js = qn∗ vs = q|ψ|2 vs :

qh̄ q2
js = ∗
|ψ|2 ∇θ − ∗ |ψ|2 A. (5.47)
m m
This is the London equation in the most general form: apply ∇× to both sides
and you will recover the second Londons equation (because ∇ × ∇θ = 0).

Flux quantizaion. Consider a closed contour deep inside a superconductor

(it can be a loop of superconducting wire). Inside the superconductor
H js = 0
(due to Meissner effect B = 0 and js = ∇ × B/µ0 ). From ∇θdl = 2πN follows
flux quantization:
Φ = N Φ0 , (5.48)
where
h h
Φ0 = = = 2.07 × 10−15 Wb (5.49)
|q| 2e
is the flux quantum.

Exercises

1. Go through all derivations in the lecture notes.

2. Estimate the radius of a contour, the magnetic flux from the Earth’s field
through which is equal to one flux quantum.

18
6 Cooper Instability

Electron-electron interaction

For the Cooper effect (formation of pairs) we need some effective attractive
interaction between electrons. This should overcome the Coulomb repulsion,
which is fortunately screened by other electrons on pretty short distances.
Positively charged ions are attracted to the nearest electron and slowly (be-
cause they are heavy) accelerate towards it. Hence, a moving electron leaves
a wake of enhanced positive charge density. Other electrons are attracted to
this wake. We thus have a short-range electron-electron attraction mediated
by the deformation of the crystal lattice (i. e. emission and absorption of a
phonon with momentum h̄q and energy h̄ωq , during which one electron changes
its momentum from h̄k1 to h̄k1 − h̄q and another – from h̄k2 to h̄k2 + h̄q). In
quantum mechanics, such interaction is described by the matrix elements Vkk′
connecting all possible initial and final electron states:

g 2 h̄ωq
Vkk′ = , (6.50)
(ϵk′ − ϵk )2 − (h̄ωq )2

where g is the amplitude of the electron-phonon interaction and q = k − k′ .

For the transitions such that |ϵk′ − ϵk | < h̄ωq , this interaction is attractive
(Vkk′ < 0). The effect is averaged over all available phonons, and is only effective
for the transitions between the states with energies within some ∼ h̄ωD from
the Fermi energy (h̄ωD is the typical phonon energy, where ωD is the Debye
frequency). If this attraction overcomes the Coulomb repulsion, we have net
attractive interaction between electrons.
The following simplified matrix element captures this:


 −V0 EF < ϵk < EF + h̄ωD
EF < ϵk′ < EF + h̄ωD

Vkk′ = (6.51)


0 otherwise.


The parameter V0 > 0 is a measure of the strength of attractive interaction. The

shell of electron states, engaged into this process, is very thin: EF /(h̄ωD ) ∼ 103 .
The number of electron states available for the interaction has a sharp max-
imum for the total momentum K ≡ k1 + k2 = 0. The effect of the interaction
is therefore the strongest between electrons of nearly opposite momenta – both
before (k and −k) and after (k′ and −k′ ) the scattering.

19
Cooper pairs
In the presence of attractive interaction between electrons the normal ground
state is unstable against the formation of bound states (Cooper 1956). Consider
the “Fermi sea” at T = 0: non-interacting electrons fill all levels (each specified
by a wave vector k and spin projection σ ≡ 1/2 or −1/2) with kinetic energy
ϵk = h̄2 k 2 /2m up to the Fermi energy EF = h̄2 kF2 /2m. Add two special elec-
trons (coordinates r1 , r2 ) that interact via a pair potential V (r1 − r2 ). Their
coordinate wave function is Ψ(r1 , r2 ). The Scrödinger equation:
 
h̄2
− 2m (∇21 + ∇22 ) + V (r1 − r2 ) Ψ(r1 , r2 ) = EΨ(r1 , r2 ), (6.52)

where E is the eigenvalue of their energy. Our goal is to calculate the lowest
possible E.

After switching to the coordinate of the centre of mass R = (r1 + r2 )/2 and
∂R ∂r
the relative coordinate r = (r1 − r2 ) it becomes (we use ∇1 = ∇R ∂r 1
+∇r ∂r 1
):

h̄2 2 h̄2
 
− ∇R − ∇2r + V (r) Ψ(R, r) = EΨ(R, r). (6.53)
4m m
After separating the centre-of-mass and relative coordinates we get:

Ψ(r1 , r2 ) = Φ(R)Ψ(r), (6.54)

where Φ(R) = eiK·R is a plane wave, and Ψ(r) obeys

 2
h̄2 K 2

h̄
− ∇2r + V (r) Ψ(r) = (E − )Ψ(r). (6.55)
m 4m
The lowest energy E corresponds to K = 0, so we consider this case:
 2

− h̄m ∇2r + V (r) Ψ(r) = EΨ(r). (6.56)

Let us express Ψ(R) as a sum over plane-wave states which are still available

a(k)eik·r .
P
Ψ(r) = |k|>kF (6.57)

The Scrödinger equation becomes

X
(2ϵk + V (r) − E) a(k)eik·r = 0. (6.58)
|k|>kF

′
We multiply it by e−ik ·r and integrate over volume:
Z
X ′
(2ϵk′ − E) a(k′ )Ω + a(k) V (r)ei(k−k )·r d3 r = 0, (6.59)
k>kF

20
where we used the relation (Ω is the volume of the system)
Z
′
ei(k−k )·r d3 r = Ωδ(k − k′ ). (6.60)

After defining the matrix element

′
1
V (r)ei(k−k )·r d3 r,
R
Vkk′ = Ω (6.61)

we arrive at “Scrödinger equation in momentum space”:

(2ϵk′ − E) a(k′ ) = − k>kF a(k)Vkk′ .
P
(6.62)

We take the simplified Vkk′ from (6.51), so (6.62) becomes

(2ϵk′ − E) a(k′ ) = AV0 , (6.63)
where X
A= a(k) (6.64)
EF <ϵk <EF +h̄ωD
is a constant (i. e. independent of k). Hence,

a(k) = AV0 (2ϵk − E)−1 . (6.65)

After plugging Eq. (6.65)) into Eq. (6.64)), A can be eliminated:

X
A = AV0 (2ϵk − E)−1 . (6.66)
EF <ϵk <EF +h̄ωD

The sum can be expressed as an integral over the energies in terms of the density
of states D(ϵ) = dN/dϵ . In the narrow range from EF to EF + h̄ωD , D(ϵk ) is
nearly constant: D(ϵk ) ≈ D(EF ):
Z EF +h̄ωD  
dϵ 1 2(EF + h̄ωD ) − E
1 = V0 D(EF ) = V0 D(EF ) ln . (6.67)
EF 2ϵ − E 2 2EF − E
Without interaction the energy of the pair would be 2EF . It is convenient to
define the energy difference Eb ≡ E − 2EF . We thus have
2h̄ωD
Eb = − . (6.68)
exp(2/V0 D(EF )) − 1
In the weak-coupling limit V0 D(EF ) ≪ 1 we finally have
 
Eb ≈ −2h̄ωD exp − D(E2F )V0 . (6.69)

For repulsive interactions (V0 < 0) Eb > 0; the normal state (Fermi sea) is
the ground state. But for any attraction (V0 > 0), there is a bound state with
Eb < 0. More Cooper pairs will form (all in state K = 0) to reduce total energy.
At finite temperature, the bound state is stable at T < Tc ∼ |Eb |/kB . Eb
is much smaller than h̄ωD . We thus have three well-separated energy scales:
kB Tc ∼ |Eb | ≪ h̄ωD ≪ EF .

21
7 BCS theory
BCS theory at T = 0
(Bardeen, Cooper, Schriffer 1957) Let us treat all electrons on the same foot-
ing. The Hamiltonian is Ĥ = K̂ + V̂ . Variational method: we will vary the
parameters of the trial wave function Φ to find the state at which the energy is
a minimum.

The ‘BCS wave-function’: only pairs (k, −k) are allowed. For any k, a pair
of electrons with (k, ↑) and (−k, ↓) is either present (|11 >) or absent (|00 >):
Q
Φ= k ψk , (7.70)

ψk = u∗k |00 > +vk |11 > (7.71)

with normalized variational amplitudes:

|uk |2 + |vk |2 = 1. (7.72)

For comparison, the normal ground state has sharp step-like functions:
uk = 0, vk = 1 k < kF
(7.73)
uk = 1, vk = 0 k > kF .

In the presence of attractive interaction one can decrease the total energy by
rounding these distributions – similar to the Cooper problem, Eq. (6.65).

In this approach we do not fix the total number of particles N (as only
the probabilities of occupation are specified). Even though there exists a final
probability to find the system with any N , for a macroscopic system the main
contribution arises from configurations with the number of particles nearly equal
to the expectation value < Φ|N̂ |Φ >. To find the minimum of < Φ|Ĥ|Φ >
while keeping < Φ|N̂ |Φ > constant, we will minimize the expectation value of
< Φ|Ĥ − µN̂ |Φ >, where the chemical potential is µ = EF , and the single
particle kinetic energies ξk ≡ ϵk − EF are measured relative to EF .

The kinetic energy of pairs is

X
< Φ|K̂ − EF N̂ |Φ >= 2ξk |vk |2 . (7.74)
k

The potential energy is (summation over all processes Vkk′ each scattering the
state (k occupied, k′ empty) with the amplitude u∗k′ vk into the state (k empty,
k′ occupied) with the amplitude vk′ u∗k ):
X
< Φ|V̂ |Φ >= Vkk′ uk vk∗ ′ u∗k′ vk . (7.75)
k,k′

22
All vk and uk have the same phase, so we assume they are real. After combining
the previous two equations, we get the expectation value of the total energy

2ξk vk2 +
P P
< Φ|Ĥ − µN̂ |Φ >= k k,k′ Vkk′ uk vk′ uk′ vk . (7.76)

We will vary vk . Variations of uk are related through Eq. (7.72):

∂uk vk
=− (7.77)
∂vk uk
The condition for a minimum is zero derivative:
∂
< Φ|Ĥ − µN̂ |Φ >=
∂vk
X X ∂uk
4ξk vk + 2 Vkk′ uk vk′ uk′ +2 Vkk′ vk′ uk′ vk = 0. (7.78)
∂vk
k′ ′ k

It is convenient to define
P
∆k ≡ − k′ Vkk′ uk′ vk′ , (7.79)

so we can rewrite (7.78) using (7.77) and (7.79) :

2ξk vk uk − ∆k (u2k − vk2 ) = 0. (7.80)

The normalization equation Eq. (7.72) can be rewritten

vk = sin θk
uk = cos θk , (7.81)

and (7.80) becomes

2ξk sin θk cos θk − ∆k (cos2 θk − sin2 θk ) = 0, (7.82)

or
ξk sin 2θk − ∆k cos 2θk = 0. (7.83)
If we define
Ek ≡ (ξk2 + ∆2k )1/2 , (7.84)

the solutions of Eq. (7.83) are

sin 2θk = ∆k /Ek

cos 2θk = ξk /Ek . (7.85)

Because 2vk′ uk′ = sin 2θk′ , we can now rewrite (7.79) as

X ∆k ′
∆k = − Vkk′ . (7.86)
2Ek′
k′

23
BCS used a simplified form of Vkk′ :


 −V0 |ξk | < h̄ωD ,
|ξk′ | < h̄ωD

Vkk′ = (7.87)


0 otherwise.


For this choice of Vkk′ , ∆k is independent of k (∆k ≡ ∆ = const):

X 1
1 = V0 (7.88)
2Ek′
k′

with summation over the narrow interaction shell (7.87) |ξk | < h̄ωD , |ξk′ | < h̄ωD .
We can reduce summation to integration (D(ξ) = dN/dξ):
X Z h̄ωD
= D(0) dξ (7.89)
k −h̄ωD

Then
Z h̄ωD  
dξ h̄ωD
1 = 2D(0)V0 = D(0)V0 sinh−1 , (7.90)
0 2(ξ 2 + ∆2 )1/2 ∆
and hence,
h̄ωD
∆= . (7.91)
sinh(1/D(0)V0 )
For many superconductors, D(0)V0 < 0.3 (except for Pb and Hg where: D(0)V0 =
0.39 and 0.35, respectively). For the weak-coupling limit, D(0)V0 ≪ 1,
1
∆ = 2h̄ωD exp(− D(0)V 0
). (7.92)

We can now find the final expressions for uk and vk . Using Eq. (7.84) and
Eq. (7.85) we get
u2k − vk2 = ξk (ξk2 + ∆2 )−1/2 . (7.93)
Together with the normalization condition Eq. (7.72) this gives:
1 
u2k = 1 + ξk (ξk2 + ∆2 )−1/2 ,
2
1 
2
vk = 1 − ξk (ξk2 + ∆2 )−1/2 . (7.94)
2

Condensation Energy
Let’s evaluate the total energy of the BCS state, using Eq. (7.76) and Eq. (7.79):
2
− ξk2 /Ek − ∆2k /2Ek ).
P P
E0 = k (2ξk vk − ∆k uk vk ) = k (ξk (7.95)

24
We expect E0 to be less than the energy of the normal state EN :
P
EN = k<kF 2ξk . (7.96)

Replacing sums by integrals gives

Z h̄ωD 
ξ2 ∆2

E0 = −2D(0) + dξ, (7.97)
0 (ξ 2 + ∆2 )1/2 2(ξ 2 + ∆2 )1/2
Z h̄ωD
EN = −2D(0) ξdξ. (7.98)
0
After integrating:
2D(0)(h̄ωD )2
E0 − EN = − , (7.99)
exp(2/D(0)V0 ) − 1
that in the weak coupling limit simplifies to

2
E0 − EN = −2D(0)(h̄ωD )2 exp(− D(0)V 0
). (7.100)

This is the condensation energy at T = 0 which for weak coupling limit can be
rewritten using Eq. (7.92):

E0 − EN = − 21 D(0)∆2 . (7.101)

BCS at finite temperature

If an electron with momentum h̄k is unpaired (a “broken pair”), from Eq. (7.76)
one can show that the total energy has increased by:

EBP − E0 = ξk − 2ξk vk2 − 2∆k uk vk = (ξk2 + ∆2k )1/2 = Ek . (7.102)

These excitations are called Bogoliubov quasiparticles. At finite temperature, the

Fermi-Dirac occupancy number for these excitations (now with a temperature-
dependent ∆k (T )), should be introduced,

−1
fk = (1 + exp(Ek /kB T )) , (7.103)

Ek = (ξk2 + ∆2k (T ))1/2 ≈ (h̄2 vF2 (k − kF )2 + ∆2 (T ))1/2 . (7.104)

For T ≪ ∆0 /kB – few excitations. For T ∼ ∆0 /kB – very many. In their

presence the BCS state benefits less from electron-electron attraction (less scat-
tering is allowed), so this weakened onding is described by the temperature
dependence: X
∆k = Vkk′ uk′ vk′ (1 − 2fk′ ). (7.105)
k′

25
The gap equation Eq. (7.86) becomes
X ∆k′
∆k = − Vkk′ (1 − 2fk′ ) . (7.106)
′
2Ek′
k

Replacing sum with an integral and using the simplified BCS form of Vkk′ gives:
Z h̄ωD
dξ
1 = D(0)V0 (1 − 2f ((ξ 2 + ∆2 )1/2 )). (7.107)
0 (ξ + ∆2 )1/2
2

This is an implicit relation between T and ∆. As T increases, the value of ∆

decreases and eventually disappears, ∆(Tc ) = 0, at a temperature Tc given by:
Z h̄ωD
dξ ξ
1 = D(0)V0 tanh . (7.108)
0 ξ 2kB Tc

In the weak coupling limit D(0)V0 << 1:

1
kB Tc = 1.14h̄ωD exp (− D(0)V 0
) = 0.57∆(0) (7.109)

As Tc ∝ ωD , this equation contains the isotope effect:

ωD ∝ M −1/2 ⇒ Tc ∝ ωD ∝ M −1/2 . (7.110)

In the weak coupling limit the dependence ∆(T ) is universal, and near Tc :
 1/2
T
∆(T ) ≈ 1.74 1 − Tc ∆(0). (7.111)

In the limit T ≪ Tc the heat capacity is C ∝ exp(−∆(0)/kB T ). The electrons in

the BCS ground state (the condensate) are ordered and do not possess entropy,
hence give no contribution in the specific heat and thermal conductivity. This
is why at T ≪ Tc superconductors can conduct heat only via phonons (like
insulators) and are thus much more poor conductors of heat than normal metals.

Unconventional superconductors
In many metals electrons make spherically-symmetric spin-less Cooper pairs
(both total orbital momentum and total spin are zero), the collective motion of
which is described by a common complex order parameter – very much like in
superfluids made of bosonic atoms and molecules. It turned out that in some
fermionic systems (like electrons in high-Tc and ‘heavy-fermion’ superconductors
but also atoms in liquid 3 He), Cooper pairs are anisotropic and can have internal
degrees of freedom (orientation of the orbital and spin angular momenta). This
greatly enriches the physics of their superconducting state, allowing variations,
in space, of the values of parameters describing their multi-component wave-
function; for instance, several types of quantized vortices and other topological
defects of the order parameter are possible. The microscopic BCS theory can
often be modified to account for the anisotropy of Cooper pairs and non-phonon
mechanism of pairing in these materials.

26
8 Thermodynamics of superconductors in the
Meissner state

Two alternative descriptions of a superconductor in the

Meissner state

1. Microscopic description: screening currents js (r) and no magnetization. B =

µ0 H. Deep inside the sample js , B, H = 0 (except within the penetration depth).

2. Macroscopic description: screening currents replaced by equivalent mag-

netization M: js = 0. B = µ0 (H + M). Inside the sample B = 0. Outside:
B = µ0 H.

Hence, inside M = −H, i. e. χ = M/H = −1 : in this description a supercon-

ductor is a perfect diamagnet.

To describe the microscopic electrodynamics we used the microscopic de-

scription (and London equations). For thermodynamics we will use the macro-
scopic description (i.e. accounting for the screening currents by the effective
macroscopic magnetization M).

Thermodynamics of the Meissner state: thermodynamic

critical field, entropy, latent heat of phase transition

When a system is held at constant applied magnetic field, Ba , (with use of, say,
an external magnet and its current supply) – the equilibrium state corresponds
to the minimum of the Gibbs free energy (here g is its density per unit volume):
g ≡ U − T S + P − Ba · M, (8.112)
where U and S are the densities of energy and entropy. After using the first law
of thermodynamics, for constant T and P the differential is
dg = −M · dB. (8.113)
We thus have (valid for any phase)
Z Ba
g(Ba ) = g(0) − M · dB. (8.114)
0

- In a superconductor in the Meissner state, Ms = −Ba /µ0 , so that (from

now on B = |Ba |):
B2
gs (T, B) = gs (T, 0) + . (8.115)
2µ0

27
 - While in the normal phase (almost non-magnetic), Mn ≈ 0, hence:

gn (T, B) ≈ gn (T, 0). (8.116)

In field B up to Bc , the superconducting phase prevails: gs (T, B) < gn (T ).

The condition for their coexistence at B = Bc (T ) is gs (T, Bc ) = gn (T, Bc ):
Bc2
gs (T, 0) + = gn (T, 0). (8.117)
2µ0
The condensation energy is related to Bc , the thermodynamic critical field, as
Bc2 (T )
(gn0 − gs0 )(T ) = . (8.118)
2µ0
Empirically, the border between the superconducting and normal states, Bc (T )
(or Tc (B)) follows
 2 !
T
Bc (T ) ≈ Bc (0) 1 − . (8.119)
Tc

The values of Tc (0) and Bc (0) for several superconductors are:

Tc , K Bc (0), mT
Al 1.19 10
Hg 4.16 39.5
Nb 9.2 195
Pb 7.2 80
Sn 3.72 30.4
Zn 0.86 5.2

 
∂g
The entropy is S = − ∂T , hence:
B

Bc dBc
Sn (T, 0) − Ss (T, 0) = − . (8.120)
µ0 dT
For T < Tc , the RHS is always positive (because dBc /dT < 0), hence: Sn ≥ Ss ,
i. e. the superconducting state is more highly ordered.

The latent heat of a phase transition is

Bc dBc
L = Tc (Sn − Ss ) = −Tc (B) . (8.121)
µ0 dT
In zero field (B = 0), L = 0 – a second order phase transition at Tc (0).
But when B ̸= 0: La ̸= 0 – a first order phase transition at Tc (B).

28
Exercises

1. For a superconductor in an external magnetic field Ba , sketch the dependence

of the magnetic field B(x) and density of screening current jc (x) on depth x
within several London penetration lengths λL from the surface.
2. From the table provided, check that for different materials the value of Bc is
roughly (within a factor of two or so) proportional to Tc .
3. Look up the definitions of the first-order and second-order phase transitions.

29
9 Ginzburg–Landau Theory

Ginzburg-Landau expansion of free energy

After a phase transition into a more highly ordered state, it usually differs from
the old one by a new symmetry; this new symmetry can often be quantified
using an observable quantity (called order parameter), which is non-zero only
in the more highly ordered phase. Continuous phase transitions (such as the
superconducting transition in zero magnetic field) have the benefit of the fact
that the order parameter is continuous through the transitions (i.e. it grows
from zero as one crosses into the ordered phase) – hence, one might guess the
analytical dependence of the order parameter from some general principles. A
typical example of the order parameter in a ferromagnet is its magnetization M
– a vector pointing in a certain spontaneously-chosen direction (while magneti-
zation was zero in the paramagnetic phase above the critical temperature). In
the superconducting state, for the order parameter it is convenient to take the
complex-valued macroscopic wavefunction ψ(r) (which is non-zero only below
Tc ); the square of its magnitude gives the density of Cooper pairs: |ψ(r)|2 = n∗
while the phase θ is chosen spontaneously.

The Helmholtz free energy of a superconductor, as function of its order

parameter ψ(r) and magnetic field B(r) = ∇ × A(r), is given by the integral
over volume, Z
Fs (ψ(r), A(r)) = Fn0 + (fs − fn0 )(r)d3 r, (9.122)

where Fn0 is the free energy of the normal state in zero field, and the difference
in the density of free energy,

1 2 2 1 4 (∇ × A)2
fs − fn0 = |(−ih̄∇ − qA)ψ| + α(T )|ψ| + β(T )|ψ| + , (9.123)
2m∗ 2 2µ0

can be expanded in powers of the order parameter ψ(r) just below Tc (see
below). In equilibrium the superconductor distributes ψ(r) and B(r) (i. e. n∗ (r)
and js (r) = µ10 ∇ × B) to minimize the total free energy Fs .

[Comment: here we are minimizing the Helmholtz free energy F = U − T S

because the magnetic field B and density of current js are treated as local vari-
ables (‘microscopic description’), subject to certain boundary conditions from
the external field. In fact, the same results (Eqs. 9.133–9.134) would have been
obtained if we minimized the Gibbs free energy in terms of the applied external
field Ba and magnetization of superconductor M (‘macroscopic description’)]

30
 Justification of the above expression for fs − fn0 :

1. Let’s begin with a spatially-uniform case ∇ψ = 0 without currents and

magnetic field: js = 0, B = 0. The simplest form:
β ∗2
fs − fn0 = α(T )n∗ + n . (9.124)
2
For T > Tc , it should have the minimum at n∗ = 0, and for T < Tc its minimum
should be at a finite n∗ (T ). We can satisfy these conditions with the choice:
α > 0 in normal phase, α < 0 in superconductor and α = 0 in between (T = Tc ).

α(T ) ≈ −a(Tc − T ), a = const, a > 0; (9.125)

β(T ) ≈ β = const, β > 0. (9.126)

Now let’s find the wave function |ψ| = const with which fs − fn0 = min:

β 4
fs − fn0 = α(T )|ψ|2 + |ψ| , (9.127)
2
dfs
0= = α(T ) + β|ψ|2 , (9.128)
d|ψ|2
α a
|ψ|2 = − = (Tc − T ), (9.129)
β β
a2
(fs − fn0 )min = − (Tc − T )2 . (9.130)
2β
B2
This is the condensation energy, − 2µc0 , hence Bc ∝ (Tc − T ).

2. Let’s now allow spatial variations of ψ(r), i. e. ∇ψ ̸= 0. A new term (the

kinetic energy of moving Cooper pairs p2 /2m∗ ) should be added:
1
KE = | − ih̄∇ψ(r)|2 . (9.131)
2m∗

3. Finally, let’s add magnetic field B = ∇ × A:

- was (B = 0): p̂ = −ih̄∇ (non-gradient-invariant momentum),
- now (B ̸= 0): p̂ = −ih̄∇ − qA (gradient-invariant momentum).
Hence, the gradient contribution (kinetic energy) becomes:
1
|(−ih̄∇ − qA)ψ(r)|2 . (9.132)
2m∗

31
Ginzburg-Landau equations

We have to find the conditions for a minimum of Fs (ψ(r), A(r)). Bu taking the
variational derivative with respect to ψ we arrive at:
1
(−ih̄∇ − qA)2 ψ + αψ + β|ψ|2 ψ = 0. (9.133)
2m∗
And taking the variational derivative with respect to A, we get:

ih̄q ∗ ∗ q2
js = − (ψ ∇ψ − ψ∇ψ ) − A|ψ|2 . (9.134)
2m∗ m∗
These two are called the Ginzburg-Landau equations. The first one looks like the
Shrödinger equation with a non-linear, superfluid density dependent, potential
energy. And the second one can be rewritten as the London equation,
q
js = |ψ|2 (h̄∇θ − qA). (9.135)
m∗

The boundary condition at the surface of superconductor is

n · (−ih̄∇ − qA)ψ = 0, (9.136)

i. e. no supercurrent is leaving the SC.

Remember that Ginzburg-Landau equations are only valid just below Tc ,

where |ψ| → 0, i.e. when Tc − T ≪ Tc .

Let us consider a flat surface of a superconductor with the x-axis normal to

the surface pointing into the superconductor, with x = 0 at the surface. Assume
there is no magnetic field (A = 0). We expect |ψ(x)| = 0 for x < 0 and small
|ψ(x)| for x just above 0. We hence can linearize (neglect the last term on the
LHS) the first equation (9.133):

h̄2 d2 ψ
− + αψ = 0. (9.137)
2m∗ dx
For the given boundary condition ψ(0) = 0, the general solution of this ODE,
for x > 0, is ψ = A sin(x/ξGL ). Plugging it into (9.137) yields the characteristic
length scale of the first Ginzburg-Landau equation:

2 h̄2 h̄2
ξGL =− = . (9.138)
2m∗ α 2m∗ a(Tc − T )

ξGL is the “Ginzburg-Landau coherence length” which characterizes the shortest

distance over which ψ(r) can vary without undue energy increase.

32
 In the Londons theory, there was no allowance for any variations of ψ inside a
superconductor); only the magnetic field could vary over distances greater than
the London penetration depth λL , which also arises from the second Ginzburg-
Landau equation (because it is basically equivalent to the second Londons equa-
tion).

Exercises

1. Sketch the magnitude of the order parameter |ψ|(T ) of a superconductor

near Tc .
2. Check the derivations of expressions (9.129) and (9.130).
3. Derive (9.129), (9.130) and (9.138) from the first Ginzburg-Landau equation
(9.134).

33
10 Type II superconductors. Vortex state.

N-S interface energy, type II superconductors

Consider a flat boundary of area LW (L – length, W – width) between the

normal (N) and superconducting (S) phases in equilibrium at applied field B =
Bc . The bulk condensation energy is −Bc2 /2µ0 per unit volume. Within the
depth ∼ ξ below inside the superconductor, there are fewer superconducting
electrons, hence a reduced contribution to the condensation energy: the energy
is higher than that in bulk superconductor by
Bc2
≈ +LW ξ . (10.139)
2µ0
On the other hand, the expulsion of the magnetic field, B, costs B 2 /2µ0 per
unit volume. As the field ∼ Bc is not quite expelled from the depth ∼ λ near
the interface, the energy is lowered by the amount
Bc2
≈ −LW λ (10.140)
2µ0
relative to bulk superconductor. After combining these two terms, the excess
energy of a surface area LW is
Bc2
≈ LW (ξ − λ). (10.141)
2µ0
The surface energy, per unit area, is then
Bc2
σ≈ (ξ − λ). (10.142)
2µ0
Depending on the interplay between ξ and λ for a particular material, σ can be
either positive (type-I) or negative (type-II).

For type-I superconductors (if λ < ξ, hence σ > 0) – to minimize the total
energy, the N-S surface area should not be too large. At certain values of Ba
some macroscopic parts of the sample go normal and allow the magnetic flux
through: this is the intermediate state (not really a homogeneous phase, but
just coexisting macroscopic regions of the superconducting and normal phases),
where the particular shape of the N-S boundary depends on the sample’s shape.

For type-II superconductors (when λ > ξ), hence σ < 0 – the N-S surface
becomes unstable with respect to increasing its area (by increasing the number
of smaller normal islands penetrated by magnetic field) until flux quantization
will limit this growth. The magnetic flux now is arranged in a dense regular
lattice of quantized flux lines, each carrying a single flux quantum, Φ0 = h/2e.
This is the vortex state – a new homogeneous phase in which an average magnetic
field co-exists with superconductivity.

34
Intermediate state in type-I superconductors (optional)

Take a thin superconducting rod exposed to the external magnetic field Ba

directed along the rod: B ≈ Ba everywhere on the rod’s surface. When Ba =
Bc all rod goes normal. Now let us take a sphere instead of a rod. On its
equator: Beq = 23 Ba . Hence, at Ba = 23 Bc equator goes normal. In the range
2
3 Bc < Ba < Bc – there will be coexistence of macroscopic regions of normal
(with field B = Bc ) and superconducting (B = 0) states. This is called the
intermediate state. Its behavior depends on the shape of the sample. Do not
confuse it with the vortex state for type-II superconductors.

Type-II superconductors

Material Tc Bc1 Bc Bc2 ξ0 λ0 κ≡

K T T T nm nm λ/ξ
Al 1.19 - 0.010 - 550 40 0.03
Pb 7.20 - 0.080 - 82 39 0.48
Nb 9.3 0.18 0.37 2.0 39 50 1.28
Pb-In 7.0 30 150 5.0
Nb3 Sn 18.2 0.035 0.44 23 3 65 22
Rb3 C60 29.6 0.012 0.44 57 2.0 247 124
YBa2 Cu3 O7 89 1.8 170 95
HgBa2 CuO4+δ 99 103 > 35

The G-L equations have two characteristic lengths

1/2 1/2
h̄2 m∗ β
 
ξ= and λ = . (10.143)
2m∗ a(Tc − T ) q 2 µ0 a(Tc − T )
B2
The N-S surface energy is σ ∼ 2µc0 (ξ − λ). The exact solution of G-L equations
√
for the border line between type I (σ > 0) and type II (σ < 0) is λ/ξ = 1/ 2.

The “G-L parameter”, κ, is a characteristics of the material:

s
λ 2β m∗
κ≡ = . (10.144)
ξ µo qh̄

In the vortex state, the flux penetrates as quantized flux lines – each with the
smallest possible flux Φ0 . The field is B0 at the axis of a line but is screened
within radius ∼ λ by superconducting current (hence, another name for this
line defect: Abrikosov vortex); hence πλ2 B0 ∼ Φ0 . On the axis, the superfluid
order parameter is suppressed (|ψ| = 0) but recovers to the bulk value at the

35
distance from the axis ∼ ξ (“core size”); the phase of the order parameter θ
winds up 2π round the axis.

The energy per unit length of a flux line is (solution of G-L equations):
 2
1 Φ0
Γ= (ln κ + ϵ), (10.145)
4πµ0 λ

(the small parameter ϵ represents the contribution from the core region; it is
negligible for κ ≫ 1). Γ is positive, that means the lines are under tension.

Two parallel flux lines repel each other by a short-range force of radius ∼ λ.
This mutual repulsion stabilizes a 2-d hexagonal lattice of flux lines.

Type II critical fields (derivations are given below)

ln κ Φ0 2


Because flux lines have positive energy Γ ≈ 4πµ 0 λ (for κ ≫ 1), the vortex
state can only become energetically favorable over the Meissner state if the
applied magnetic field Ba exceeds a finite value Bc1 :
√
Bc1 = Bc ln κ/ 2κ ∼ Bc κ−1 , for κ ≫ 1, (10.146)

And approaching Bc2 , non-superconducting cores (of cross-section ∼ ξ 2 ) of flux

lines (of areal density Bc2 /Φ0 ) begin to overlap and thus leave no room for
superconductivity (|ψ| ≠ 0) between them:

2|α|m∗ h̄ √
Bc2 = = 2 = 2Bc κ. (10.147)
qh̄ qξ
√
In the limit κ = 1/ 2, the difference between type-I and type-II superconduc-
tors disappears, and Bc1 = Bc2 = Bc . And for κ → ∞, Bc2 → ∞, i. e. extreme
type-II superconductors with high κ ≫ 1, the vortex state survives up to fields
B ∼ Bc2 ≫ Bc .

Flux pinning

In type-II superconductors, magnetization curves are often irreversible. Flux

lines can be pinned by inhomogeneities in the sample (non-equilibrium states).

The Lorentz force, per unit volume, on electrons:

FL = j × B. (10.148)

36
Without pinning, the flux lines would move across the current; this would gen-
erate emf (voltage in the directions of current) and hence energy dissipation.
In a particular material, there exists a certain maximum pinning force per unit
volume F p (max). Hence, there is a critical current density,

jc = F p (max)/B, (10.149)

beyond which there is effective dissipation.

Certain flux lines are normally pinned to the regions of weakened supercon-
ductivity, i. e. impurities or crystalline defects (as the vortex core is normal, the
loss of condensation energy is smaller if the core passes through such a region).
The other lines are restrained because of the repulsion from other lines, i. e. the
rigidity of the flux lattice – this effect is called collective pinning: the critical
pinning force F p (max) becomes dependent on field B.

Flux creep is a slow relaxation of pinned arrays of lines from one metastable
pinned configuration to another. Creep is always associated with finite voltage
and dissipation.

In type-I superconductors, the critical current is normally determined by Bc

(superconductivity breaks when field B, generated by the current, reaches Bc ).
In type-II superconductors in the vortex state, creep (and hence dissipation) is
always possible; however, it can be minimized by strong pinning while main-
taining relatively high densities of current One thus can compare pros and cons
of using different types of superconductors:
• Type-I: perfect Meissner effect up to Bc ∼ 0.01 T (good shielding), true
persistent current, absolutely no dissipation. But Bc is low.
• Type-II: Inevitable resistive losses due to flux lines motion. To reduce
dissipation and increase Bc2 – pinning is needed with associated inhomo-
geneity, metastability and creep. But Bc2 and jc are high (for high-Tc SC
Bc2 can be as high as 100 T or higher).

Applications

Because of high jc and high Bc2 , superconducting magnets made of type-II wires
can produce very stable B ∼ 20 T at zero power. To maximize pinning, one can
use a combination of cold working and heat treatment (for NbTi: material is
divided into cells by dislocation walls, small particles of Ti-rich phase precipitate
on these walls or pinning to grain boundaries (for Nb3 Sn). To make the wire
stable against sudden local temperature increase caused by creep (“flux jump”),
they are made of thin filaments (∼ 30 µm) enclosed into a good heat conductor
(usually copper).

37
 Why would one want to have high magnetic fields?
1. NMR imaging of various elements in tissue, lungs and bloodstream.
2. Physics research:
• condensed matter research (e. g. of graphene);
• particle physics: to guide charged particles in accelerators by strong
magnetic field;
• to contain plasma in fusion reactors (tokamaks);
• cooling to ∼ 10−6 K by demagnetizing a paramagnet from field ∼ 8 T.
3. Magnetic levitation.
4. Magnetic separation: e.g. Cu-Ni ore.
5. Energy storage.
6. Compact and powerful motors, generators, etc.

Lower critical field Bc1 (optional)

ln κ Φ0 2


Because flux lines have positive energy Γ ≈ 4πµ 0 λ (for κ ≫ 1), the vortex
state can only become energetically favorable over the Meissner state if the
applied magnetic field Ba exceeds a finite value Bc1 . At this field, Ba = Bc1 ,
the Gibbs free energies of these two competing states become equal:

gMeissner (Ba ) = gvortex (Ba ), (10.150)

where g = f − Ba · M. The Helmholtz free energy density of the vortex state

is greater than that for the Meissner state by N Γ, where N is the areal density
of flux lines in the vortex state:

fvortex = fMeissner + N Γ. (10.151)

The mean magnetic flux densities, B = µ0 (H + M ), are B = 0 in the Meissner

state and B = N Φ0 in the vortex state (as each vortex carries flux Φ0 ). Hence,
Ba
MMeissner = − (10.152)
µ0
and
N Φ0 − B a
Mvortex = . (10.153)
µ0
After plugging all these into Eq. 10.150, we arrive at
µ0 Γ Φ0
Bc1 = ≈ ln κ. (10.154)
Φ0 4πλ2

38
 √
2Φ0
Using Bc = 4πλξ (which can be obtained from the expressions for condensation
α2 Bc2
energy 2β = 2µ0 , and for ξ and λ in terms of α and β), we get

Bc
Bc1 = √ ln κ, (10.155)
2κ

i. e. roughly Bc1 ∼ Bc κ−1 .

Upper critical field Bc2 (optional)

As the applied field Ba approaches the upper critical value Bc2 , the supercon-
ducting order parameter vanishes, |Ψ| → 0. We thus can use the first Ginzburg-
Landau equation without its non-linear term:
1
(−ih̄∇ − qA)2 ψ = |α|ψ. (10.156)
2m∗
Our task is to find the highest possible Ba that still allows a non-trivial (i. e.
with |Ψ| =
̸ 0) solution of this equation. Fortunately, this equation is equivalent
to the Schroedinger equation for a stationary state Ψ of a particle of charge q
in magnetic field B = ∇ × A and with the eigenvalue of energy Ec = |α|. The
solutions of the latter, for a uniform field along z-axis, B = (0, 0, B), corresponds
qB
to quantized cyclotron rotation in xy-plane at frequency ωc = m ∗ )while moving

along z-axis at any velocity. The energy of such solutions is

h̄2 kz2
 
1 h̄qB
Ec = n + + , (10.157)
2 m∗ 2m∗

where n is any integer, including 0, and kz is any wavenumber. We will recy-

cle this solution. After equating |α| = Ec , and noticing that the solution of
Eq. 10.157 with the largest value of B corresponds to kz = 0 and n = 0,
1 h̄qB
|α| = , (10.158)
2 m∗
we arrive at
2|α|m∗
Bc2 = . (10.159)
h̄q
Again, using expressions for Bc and ξ in terms of α and β, one can get
h̄ Φ0 √
Bc2 = 2
= 2
= 2Bc κ. (10.160)
qξ 2πξ

The relation, Bc2 /Φ0 ∼ ξ −2 , tells that, at B = Bc2 , the non-superconducting

cores (of cross-section ∼ ξ 2 ) of flux lines (of areal density Bc2 /Φ0 ) begin to
overlap and thus leave no room for superconductivity (|ψ| = ̸ 0) between them.

39
Exercises

1. Sketch |ψ(r)| and B(r) as function of the distance r from the axis of a flux
line when the value of an applied field is:
(i) just above Bc1 , (ii) just below Bc2 .
2. For the Pb-In alloy and YBa2 Cu3 O7 , using the parameters given in the table
in the notes (and inventing a way to estimate the value of Bc ), make estimates
on Bc1 and Bc2 at T = 0.
3*. Complete the derivations of the formulae for Bc1 and Bc2 , given in the notes.

40
11 Josephson junctions

Weakly coupled superconductors

The long-range order of the phase of the order parameter means that fixing its
value at one point determines its value at any other point of the same piece of
superconductor. Consider two isolated pieces of superconductor with phases θ1
and θ2 . If Cooper pairs have a small (but finite) probability of tunneling between
two bulk pieces, the superconducting current I between the pieces will depend
on the phase difference ∆θ ≡ θ2 − θ1 – and can hence be used to determine
the value of ∆θ. The function I(∆θ) should be periodic with a period 2π. The
simplest one is I ∝ sin ∆θ which is generally valid for very weak links (in the
solutions to today’s example sheet, see a derivation of this formula using the
G-L equations). Such links are called Josephson junctions and are typically
made of a thin (∼ 20 Å) insulating film between two superconductors.

Josephson equations

DC equation. The equation for the current through a weak link,

Is = Ic sin ∆θ. (11.161)

It tells us that for V = 0 a DC supercurrent Is can flow across the junction if

Is < Ic , where Ic is the critical current of the particular junction. We call this
“DC Josephson effect”. The phase difference ∆θ will adjust to the value of I.

RF equation. In the presence of a non-zero voltage V across the junc-

tion separating two bulk superconducotrs with the order parameters ψ1 =
ψ0 exp(iθ1 ) and ψ2 = ψ0 exp(iθ2 ) at chemical potentials µ1 and µ2 (here ∆µ ≡
µ2 −µ1 = qV , where q = 2e), the phases θ1 (t) and θ2 (t) evolve at different rates.
The time-dependent equations,
∂ψ1 ∂ψ2
ih̄ = µ1 ψ1 , ih̄ = µ2 ψ2 , (11.162)
∂t ∂t
result in
∂∆θ q
= − V. (11.163)
∂t h̄
Integration gives the second Josephson equation (or “RF Josephson effect”):
q
∆θ = ∆θ0 − V t. (11.164)
h̄
Hence, in the presence of voltage the phase difference keeps changing with time.

41
 After plugging this result in the first equation we obtain
 q 
Is = Ic sin ∆θ0 − V t , (11.165)
h̄
which tells us that an application of a constant voltage to the Josephson junction
generates an AC current with frequency

|qV | |V |
νJ = = . (11.166)
h Φ0
This current doesn’t dissipate energy because the time average of Is V is zero!

Since 1990 the volt has been maintained using the Josephson effect, where
the Josephson constant is fixed by the 18th General Conference on Weights and
Measures as: νJ /V = 2e/h = 0.4835979 GHz/µV. This is typically used with
an array of several thousand or tens of thousands of junctions at microwave
frequences between 10 and 80 GHz.

Magnetic field and phase

In the presence of magnetic field, ∆θ is not gauge-invariant, i.e. it doesn’t have

a unique value for different choices of A (gauges) and hence cannot define the
current. The gauge-invariant gradient of phase is
q
∇γ ≡ ∇θ − A, (11.167)
h̄
and the gauge-invariant phase difference is
Z
q
∆γ ≡ ∆θ − A · dl. (11.168)
h̄
Then the experimentally measured current is
 Z 
q q
Is = Ic sin ∆γ = Ic sin ∆θ0 − V t − A · dl . (11.169)
h̄ h̄

Without magnetic field, θ was the same in all bulk parts of the superconductor
where Is = 0. With field, only γ is the same everywhere, while θ(r) becomes a
function of position r and subject to the choice of gauge (A).

42
Interference effects in magnetic field. DC SQUID

DC SQUID: Consider two Josephson junctions in parallel (both having the same
critical current Ic ): one from point (a) to (b) conducts current I1 , the other from
point (d) to (c) conducts current I2 . The total current
   
1 1
I = I1 +I2 = Ic (sin ∆γ1 +sin ∆γ2 ) = 2Ic cos (∆γ1 − ∆γ2 ) sin (∆γ1 + ∆γ2 ) .
2 2
(11.170)

We will now show that the value of ∆γ1 −∆γ2 is determined by the magnetic
flux through the loop made by the wires connecting the junctions.
q b
Z
∆γ1 = θb − θa − A · dl (11.171)
h̄ a
and
q c
Z
∆γ2 = θc − θd − A · dl. (11.172)
h̄ d
q a q c
Z Z
∆γ1 − ∆γ2 = (θd − θa ) + (θb − θc ) + A · dl + A · dl. (11.173)
h̄ b h̄ d
Note that inside bulk superconducting wires, js = 0, hence (from the second
G-L equation, which is basically a definition of the value of js ),
q
js = ∗ |ψ|2 (h̄∇θ − qA) = 0, (11.174)
m
we get
q
∇θ = A. (11.175)
h̄
Hence,
Z d
q d
Z
θd − θa = ∇θ · dl = A · dl, (11.176)
a h̄ a
Z b
q b
Z
θb − θc = ∇θ · dl = A · dl. (11.177)
c h̄ c

We can now re-write the expression (11.173), ∆γ1 − ∆γ2 =

Z d Z c Z b Z a ! I
q q q
= A · dl + A · dl + A · dl + A · dl = A · dl = Φ,
h̄ a d c b h̄ h̄
(11.178)
where Φ is the magnetic flux through the loop.

q 2π
Finally, as h̄ = Φ0 ,
   
1 Φ
cos (∆γ1 − ∆γ2 ) = cos π , (11.179)
2 Φ0

43
and    
πΦ 1
I = 2Ic cos sin (∆γ1 + ∆γ2 ) . (11.180)
Φ0 2
The first term, 2Ic cos( πΦ
Φ0 ), gives the total critical current for a given flux Φ.
The second term, sin( 12 (∆γ1 +∆γ2 )), automatically adjusts the parameters ∆γ1
and ∆γ1 to sustain the given total supercurrent I when I ≤ Ic .

Usually, the magnetic field under investigation, B, is made to generate the

flux inside SQUID, Φ ∝ B, and the critical current through SQUID (propor-
tional to cos( πΦ
Φ0 )) is monitored. As the interval between the two adjacent zeros
of the critical current, Φ0 = 2.07 × 10−15 Wb, is very small, one can detect
extremely small changes in B. Typically, the sensitivity to the change of flux of
order 10−3 Φ0 can be achieved.

SQUID and other applications of superconductors

Superconducting QUantum Interference Device. DC SQUID has two (usually

identical) junctions. It can detect magnetic field ∼ 5×10−15 T (when integration
time is ∼ 1 s).

As a result, SQUIDs are also sensitive to even a tiny electromagnetic noise.

To reduce the background electrical and magnetic fields, magnetic shielding is
used. A superconducting cage in the Maeissner state is a perfect shield of a
weak magnetic field because it expels it completely.

The flux from a pick-up coil (which is near or around the sample) is often
transferred to the SQUID loop by the “flux transformer” – a closed loop of a su-
perconducting wire. The total flux in this loop is quantized (hence, conserved).
If the flux through one (pick-up) part of such loop has changed, a supercurrent
is induced to compensate for it. Hence, the flux through another part of the
transformer (which is next to the SQUID) reflects the change in the signal –
and can thus be monitored by the SQUID.

Because of their unmatched sensitivity to small magnetic fields, SQUIDs are

widely used:

– in experimental physics – measurements of small magnetic field, suscepti-

bility, current, voltage, etc.;

– in medicine and biophysics – detecting tiny currents due to brain activity,

heart, etc.;

– in geology – mapping Earth’s magnetic field to detect iron, nickel ore, etc.

44
Exercises

1. What value of the Josephson frequency corresponds to the voltage across a

Josephson junction of V = 1 × 10−6 V? Does this value depend on temperature?
2. How small should be the diameter of a circular contour of a DC SQUID, so
the flux from the Earth’s magnetic field BE = 5 × 10−5 T is equal to one flux
quantum?

45
 End-term (second-half of the course) checklist
(for the first-half checklist go to page 13)

1. Fermions, Fermi-Dirac distribution, general properties of electrons in nor-

mal metals.
2. General experimental properties of superconductors, evidences for the en-
ergy gap and pairing of electrons.
3. Role of electron-phonon interaction in superconductivity, Cooper pairing,
main results of BCS theory: Bogoliubov excitations, energy gap, excitation
spectrum and pair-breaking critical velocity, critical temperature, isotope
effect, coherence length (size of Cooper pair).

4. Order parameter for Cooper pairs, interpretation of its magnitude and

phase, expression for the density of superconducting current.
5. Two-fluid model for superconductors, London equations and their relation
with perfect conductivity and macroscopic coherence (order), Meissner
effect, penetration depth and its temperature dependence, macroscopic
wave function for Cooper pairs, flux quantization, flux trapping.
6. Magnetization in the Meissner state, thermodynamic critical field, con-
densation energy, entropy, latent heat of phase transition into the super-
conducting state.

7. Ginzburg-Landau equations (without derivation), Ginzburg-Landau co-

herence length, Ginzburg-Landau parameter κ, energy of the interface
between a normal and a superconducting region, type-I and type-II su-
perconductors.
8. Quantized flux lines in type-II superconductors, vortex core sizes, vortex
state, first and second critical field in type-II superconductor, magneti-
zation in the vortex state, pinning of flux lines and its importance for
applications.
9. Josephson effects in superconductors, Josephson frequency, effect of mag-
netic field, DC SQUID, applications.

46
12 Numerical answers to exercises

1.2. Tc ∼ 34 nK (with n ∼ 2.5 × 1012 cm−3 as in the original paper by Anderson

et al., 1995).

2.3. κ ≡ h/m4 = 1.00 × 10−3 cm2 s−1

2.5. ℓ ∼ n−1/2 ∼ 0.2 mm q

FYI: the exact result for a hexagonal lattice is d = √23 n−1/2 = 0.24 mm.

3.3. Even in very narrow channels, the critical velocity for phase slip vc is
smaller than vL . Hence, dissipation due to phase slips always prevails.

5.2. With Earth magnetic field being BE ∼ 5 × 10−5 T, R ∼ 3.6 µm.

√
2Φ0
10.2. Using Bc = 4πλξ (formula above Eq. 10.155), we get:
Pb-In: Bc ∼ 5.2 × 10−2 T, Bc1 ∼ 1.2 × 10−2 T, Bc2 ∼ 0.37 T.
YBa2 Cu3 O7 : Bc ∼ 0.76 T, Bc1 ∼ 2.6 × 10−2 T, Bc2 ∼ 100 T.

11.1. ν ≈ 0.48 GHz, independent of temperature.

11.2. d ≈ 7.3 µm.

47