Contents

1 Historical Prospective 3

2 How To Measure Zero Resistivity? 3

3 Defining Properties 3
3.1 The Meissner Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
3.2 Superconductors as Magnetic Shields . . . . . . . . . . . . . . . . . . . . . . . . 5
3.3 Heat Capacity of a Superconductor . . . . . . . . . . . . . . . . . . . . . . . . . 5
3.3.1 The Heat Capacity as We Approach Superconductivity . . . . . . . . . 6
3.4 Electronic Heat Capacity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
3.4.1 Normal Phase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
3.4.2 Superconductive Phase . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
3.5 The Order of Phase Transition . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
3.6 Defining The Thermodynamic Critical Field . . . . . . . . . . . . . . . . . . . . 7

4 A Superconductor’s Demagnetizing Field 8

5 A Two-fluid Model of Superconductor: The London Model 9

5.1 The London Parameter and Equations . . . . . . . . . . . . . . . . . . . . . . . 9
5.2 London Penetration Depth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
5.2.1 Typical Penetration Depths . . . . . . . . . . . . . . . . . . . . . . . . . 11
5.3 London Current Direction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
5.4 Pippard Coherence Length . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

6 Flux Quantization 12

7 Superconductor Phase Coexistences 13

7.1 Zero Demagnetising Factor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
7.1.1 Interface Energy between Normal and Superconductive Phase . . . . . . 14
7.2 Non-Zero Demagnetising Factor . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
7.3 The Intermediate State in a Flat Sheet (Interface Energies) . . . . . . . . . . . 16
7.3.1 Type 1 vs. Type 2 Interface Energies . . . . . . . . . . . . . . . . . . . . 18
7.4 Abrikosov Vortices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
7.5 Flux Line Pinning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

8 Ginzburg Landau Theory 20

8.1 Type 2 Superconductors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

9 AC Conductivity according to Ginzburg-Landau 24

10 Microscopic Theory of Superconductivity - The BCS Theory 24

10.1 Annihilation Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
10.2 BCS Coherent States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
10.2.1 Trial Ground State Wavefunction from Coherent States . . . . . . . . . 26
10.3 Attractive Potential Between Electrons . . . . . . . . . . . . . . . . . . . . . . . 27
10.3.1 Quick Rundown of Cooper Pair Properties . . . . . . . . . . . . . . . . 30
10.4 Excitation from BCS Ground State . . . . . . . . . . . . . . . . . . . . . . . . . 30
10.5 Density of States - Realising DELTA . . . . . . . . . . . . . . . . . . . . . . . . 31

11 The Josephson Effect & Junctions 32

1
 Superconductivity Module PHYS5300M This is a Level 5 Module run by Dr. G. Burnell,
the proposed syllabus is as follows:

• Phenomenalogical Aspects of Superconductivity

1. The discovery of superconductivity and its classification as a new state of matter.
2. Basic properties of superconductors - zero resistance, perfect diamagnetism, critical
fields and critical currents.
3. The Meissner effect.
4. The phenomenological London model, London penetration depth and Pippard coher-
ence length.
5. Demagnetisation factors.
6. Importance of surface energy in defining Type I and Type II behaviour. 7. The mixed
state and the intermediate state.
8. Flux penetration in Type II superconductors, flux pinning and Bean’s critical state
model., and the importance of flux pinning in applications

• Theoretical Approaches
1. Introduction to Ginzburg-Landau theory and the macroscopic wave function.
2. Flux quantisation.
3. Formation and character of Cooper pairs and the origin of the positive attraction
between electrons.
4. A description of BCS theory.
5. The superconducting gap and superconducting thermodynamics.
6. The isotope effect.
7. Excitations from the superconducting ground state and the BCS quasiparticle density
of states.

• Beyond Low Tc Superconductors

1. Superconducting materials and high temperature superconductors.
2. Principal differences between s and d wave superconductors and evidence for the d
wave symmetry in high temperature superconductors.
3. Superconducting electronics, dc and ac Josephson effects, analogues between Joesphson
critical current dependence on magnetic field and optical diffraction.
4. Applications of superconducting tunnel junctions and SQUID devices.
5. Andreev reflection.

The module is 100% exam and is lecture based.

IMPORTANT: We take, as Tinkin has, that h is the microscopic flux density and it may
be useful to identify that,
4πJ
h=
c
h (0.1)
2
∇ h=
λL
Furthermore, before actually defining what Type 1 and Type 2 Superconductors are we will
refer to them. Deal with it,

2
1 Historical Prospective
The two well-known properties of Superconductors are that they a) have virtually no electrical
resistance and b) they occur at only very low temperatures, there discovery was thus obviously
a mistake observed by low-termperature scientists. Sir James Dewar used a low-temperature
storage vessel to measure the resistivity of Au and Ag as low as 16K and retrieved data that sup-
ported Mathiessen theory of low-temperature metal electrical behaviour (as shown in Fig1.1),
Kamerlingh Omnes was measuring the resistivity of Hg at as low as temperature as possible
through the use of He.
Omnes, in 1908, observed that his system stuck at 4.2K - the resistivity of the Hg dropped to
an immeasurably small value and the system could not be cooled any further - the experimen-
talists were under the belief that the immeasurable resistivity was due to an electrical short,
however prior to a particular repeat of the experiment a valve had been set incorrectly and the
system began to increase in temperature beyond 4.2K and a measurable resistivity returned -
this could not be possible if the immeasurable resistivity at 4.2K was due a short. Thus Omnes
had discovered both the boiling point of He and the property of Superconductivity - a material
displaying an immeasurably small resistivity, essentially ZERO, at low Temperature.

Figure 1.1: Plot showing the competing theories of the low-temperature resistivity of the metals

2 How To Measure Zero Resistivity?

An experimenter cannot simply attach a voltmeter to a superconducting material and take a
reading of zero as proof of superconductivity; internal resistance and systematic error stand in
the way of this. If we construct a ring of superconducting material and cool it to below TC
with H-applied and then de-activate the applied field then a current will be induced to flow
within this ring and as T < TC this current should experience zero resistivity and generate a
stable external field as it flows around the ring. Various experimentalists have observed that
ρ ∼ 1 × 10−24 Ω meaning that the external field would be stable for decades (according to Dr.
Burnell but the longest actual observed time, according to Ashcroft & Mermin, is 2.5years)!
We see from Fig2.1 that superconductivity is not as rare as one may think, this is on
occasion bad thing: Tin is a superconductive material that does not need to be ultra-pure to
display superconductivity and this is an issue as solder typically comprises of largely Tin.

3 Defining Properties
Superconductors have a few defining critical parameters; the superconductive state may be
destroyed by an excessive external field, applied current density or by T > TC

3
Figure 2.1: Periodic Table with Elements currently known (2008) for displaying some Super-
conductive Behaviour

Figure 3.1: Diagram of the so-called ’JBT’ Surface of a Superconductor, note that in a realistic
Superconductor the critical parameters are interdependent.

3.1 The Meissner Effect

The Meisnner Effect was observed in 1933 and further defined the Superconductive Phase,
until this point Superconductors had been considered as delicate perfect conductors of electrical
power; now it was known that Superconductive materials actively exclude small magnetic fields
(see Fig 3.2). The Meissner Effect allows us to concretely define a Superconductor:

R=0
dB
=0 (3.1)
dt
B=0

This reversible Meissner Effect (excluding external field and expelling field that previously
penetrated the sample in the normal phase) meant the Superconducting phase could not be
described as perfect conductivity as such a device would tend to trap flux within, furthermore
this meant that the Superconductive state could be destroyed with a critical field Hc allowing
the phase transition to be considered as a true thermodynamic state - a description of the
thermodynamic ordering process is given later.
At this point we introduce some of the nomenclature of Magnetism in Condensed Matter in

4
Figure 3.2: Top: Material above TC with applied field, Middle: Material cooled below TC
with applied field, Left: The applied field is unperturbed, Right: Superconductor excludes the
applied field, Bottom: Applied field is removed, Left: Pure conductor acts as a bulk magnet,
Right: Circulating currents arise on the surface of the superconductor and generate an external
field

order that we better understand this new phase:

B = µ0 H
B = µ0 (H + Mv )
M Mv (3.2)
χ= v =
H Happ
B = µ0 Hχ
where Mv is the magnetic moment per volume.
The discovery of surface currents in Superconductors due to this Meissner Effect means that
with zero applied field we may define Mv = −H which should culminate in a Superconductor
being characterised by,
χ = −1 (3.3)
A superconductor is thus perfectly paramagnetic, however in practice we are forced to measure
some applied field from which we can calculate an effective susceptibility of the order 1 × 10−3
- this is more thoroughly (or rather, actually) explained in Section4.

3.2 Superconductors as Magnetic Shields

A cylindrical superconductor with one closed end will act as a highly efficient magnetic shield
as the field within the cylinder will act to oppose any applied field.

3.3 Heat Capacity of a Superconductor

dS
Recall the definition of heat capacity first, c = T dT and then the final line of Equ.6.6
 
d dHC
∆c = −T µ0 HC
dT dT
 ! (3.4)
d2 HC dHC 2

∆c = −T µ0 HC +
dT 2 dT

At T = TC we know HC and thus,

 2
dHC
∆c = −TC µ0 6= 0 (3.5)
dTC

5
3.3.1 The Heat Capacity as We Approach Superconductivity
Experimentalists observed a temperature dependence of the heat capacity of a material as it
approached TC that they approximated by a 4th Power Law,

T 4


nnormal
=
n TC
nsuper n
super
4
=1− (3.6)
n n
 4 !− 21
T
λL (T ) = λL (0) 1 −
TC

3.4 Electronic Heat Capacity

The standard electronic heat capacity is given by
Z ∞
dE df ()
Ce = = E() N ()d (3.7)
dT 0 dT

3.4.1 Normal Phase

Assumptions:

• E() =  − F

• f () = fermi function

• N () ∼ N (F

and thus,
π 2 N (F kB
2
Ce = T = γT (3.8)
3

3.4.2 Superconductive Phase

Assumptions:

• E() =  − F
 
−
• f () ∼ exp − k TF
B

• N () ∼ N (F for | − F > ∆

• N () = 0 for | − F < ∆

and thus,
F
 − F 2
   
 − F
Z
Ce = N (F ) exp − d+
0 kB T kB T
(3.9)
 − F 2
Z ∞    
 − F
N (F ) exp − d
F +∆ kB T kB T
Collect terms and simplify,
∞ 2  
 − F  − F
Z
Ce ∼ 2N (F ) exp d (3.10)
F kB T 2 kB T 2

6
 −F
Substitute x = kB T into the above,
Z ∞
Ce ∼ 2
2N (F )kB T (x)2 exp (−x) dx (3.11)
∆
k T
B

Ry 2 e−x dx
∆
= e−y 2 + 2y + 2y 2 -


Thus for kB T  1 the integral simplifies to according to, ∞x
the most significant terms being 2y 2 e−y and so:
 2     
2 2 ∆ ∆ 2N (F ) ∆
Ce ∼ 2N (F )kB T 2 T 2 exp − k T ∼ exp − (3.12)
kB B T kB T

3.5 The Order of Phase Transition

Cooling the superconductor requires no change in entropy and is thus not a 1st Order Transition,
however destroying the Superconductive phase does require ∆S and so is a 1st Order transition
- this is because the Heat Capacity of the Superconductive phase is not zero (see Section??)

3.6 Defining The Thermodynamic Critical Field

This is a bit of an aside but important for principles. Here we consider a flat Superconducting
sheet of thickness d subject to a parallel magnetic field Ha , solve Equ.0.1 with the condition
that at the two surfaces of the sheet h = Ha we obtain,

cosh(x/λ)
h = Ha (3.13)
cosh(d/2λ

We know from the Meissner Effect and from solving the above that h is at a minimum of
Ha
cosh(d/2λ) in the middle of the sheet. Now we posit the average microscopic field in the Super-
conductor,
2λ d
B ≡ h̄ ≡ Ha + 4πM = Ha tanh T hus, (3.14)
d 2λ
• when d  λ, B → − −H
− 0 as required by the Meissner Effect, and thus M → 4π
a

− Ha (1 − d2 /12λ2 ) and thus

• when d  λ a Taylor Expansion of the above provides B →
Ha d2
M→
− − (3.15)
4π 12λ2
As a field is applied to a Superconductor it becomes energetically unfavorable for the Super-
conductive phase to exist as the magnetic energy associated with the diamagnetic response of
the Superconductive phase becomes greater than the initial advantage in free energy in zero
field, we define this field strength as Hm via
Z Hm
(Fn − Fs ) = M (H)dH (3.16)
0

For the Meissner Case ( d  λ ) we term this field Hm the thermodynamic critical field and
identified as Hc via
Hc2
≡ (Fn − Fs ) |H=0 (3.17)
8π

7
Figure 4.1: A Superconducting pellet (Shaded Region) housed in a solenoid through which
some current I is applied, generating an external field, He . We consider the current loop from
A to F as shown

4 A Superconductor’s Demagnetizing Field

The Meissner Effect (Section.3.1) means that it is not possible to measure the internal field of
a superconductor, we only know that the internal flux is zero. In order to access this field we
consider a pellet of superconducting material in an infinite solenoid of n turns per length, as
shown in Fig.4.1
If we identify Hi as the superconductor’s internal field and He as the external field we can begin
to properly consider this problem using Ampere’s Law,
I I I
H · dl = nI = Hi dl + He dl (4.1)
AB CDEF

While in the normal phase (T > TC ) we observe Hi = Hi = H, however when the pellet enters
the Superconducting phase the flux density close to the pellet reduces, i.e. He reduces, the
properties of the Solenoid nI have not changed though and thus for T > TC the internal field
must be non-zero and indeed greater than that when in the normal phase. We may express Hi
in terms of a shape dependent demagnetising field HD .

Hi = H − HD
(4.2)
HD = nMv

If we assume the shape of the pellet to be ellipsoid - characterised by size fractions a and c
and n is the demagnetizing factor - the demagnetizing field may be approximated as shown in
Fig.4.2

Figure 4.2: Demagnetizing field as a function of a/c

From Equ.3.2 we can define the measurable susceptibility of a superconductor, χm

Mv Mv χ
χm = = = (4.3)
Happ Hi (1 + nχ) (1 + nχ)

We may therefore utilize this set-up as a measure of the susceptibility of a Superconducting

sample, however as mentioned previously this is not a direct measure and thus while a perfect
Superconductor has χ = 1 the susceptibility measured in this set-up is of the order of χm =
1 × 10−3 .

8
5 A Two-fluid Model of Superconductor: The London Model
Casimir and Gorter proposed a phenomenological model of the superconductive phase by invok-
ing two fluids within the material, one comprised of the normal phase charge carrying particle
(the electron) and the other the superconducting charge carrier. The model is relatively crude
as the Fermi-Dirac nature of the charge carriers (both normal and Superconductive) is ignored,
however Maxwell’s Equations and Thermodynamics were readily applicable.
The first question for the Model to address is what property of the Drude Model of Conduc-
tivity allow zero resistance, i.e. zero scattering events?
Naturally this would be an infinite mean free path for the Superconductive charge carriers, and
this is our one assumption in the entire model.

5.1 The London Parameter and Equations

Let us consider the Drude’s conductivity for a superconductor,

J = n s e ∗ vs (5.1)

Observe that the force experienced by these charge carriers under a field E is given by m dv
dt =
s

e∗ E in order for us to write the time derivative,

dJ n e∗2
= s E (5.2)
dt m
∗2
By choosing to define Λ = nsme as the so-called London Parameter we can re-arrange to
generate the 1st London Equation:
d
Eeff = Λ J (5.3)
dt
dB
Now recall that dt = −∇ × E and submit Equ.5.3:
 
dB dJ
= −∇ × Λ (5.4)
dt dt

Into which we may submit Ampere’s Law, ∇ × H = J + dv

dt ,
 
dB d
= −∇ × Λ ∇ × H (5.5)
dt dt

Which simplifies as,

dB d
= −Λ (∇ × ∇ × H) (5.6)
dt dt
∇2 B
From Vector Calculus identities and Equ.3.2, ∇ × ∇ × H = ∇ · ∇H − ∇2 H = 0 − µ0 , allows
us to identify London’s 2nd Equation:
Λ 2
B= ∇ B (5.7)
µ0

The one assumption of this Model, that there is an infinite mean free path for the Supercon-
ducting charge carriers, presents us with an issue: as vs is constant so too must E, however we
know from the Meissner Effect that this is not the case.
This led the London Brother’s to impose strict Boson Statistics on ns as an explanation for
their being no states into which the carriers could scatter into. Consider then the wavefunction
for the charge carriers,
√
Ψ(r) = ns | Ψ(r) | eiθ(r) (5.8)

9
We may actually retrieve the London Equations from this wavefunction by equating the Quan-
tum Mechanical momentum of the charge carriers (−i~∇) with the Classical change in mo-
mentum due to an applied magnetic field (mv + qA using tricks involving B = ∇ × A, and we
need to go through that step-by-step here.

hΨ | −i~∇ | Ψi = hρi = (mvs + e∗ A) ns (5.9)

Which we may simplify,

hΨ | Ψi~∇θ = (mvs + e∗ A) ns (5.10)
~∇θ = mvs + e∗ A (5.11)
We can then re-arrange Equ.5.1 and submit it into the above,
ms
~∇θ = J + e∗ A (5.12)
ns e∗

h
∇θ = ΛJ + e∗ A (5.13)
e∗
d
Observe that dt θ is the angular velocity of the carriers and that we may use this to calculate
d
the chemical energy, µ, as follows, ~ dt θ = ~ω = µ and in turn the Electric Field Strength
∇µ
- E = e∗ . Thus by taking time derivatives of Equ.5.13 we move towards the 1st London
Equation, Equ.5.3:
~ d d
∇θ = [ΛJ + e∗ A] (5.14)
e∗ dt dt
d
E= (ΛJ + e∗ A) (5.15)
dt
d
We need to define first Eeff = E − dt A before finally retrieving the 1st London Equation,

d
Eeff = Λ J (5.16)
dt
Lets return to Equ.5.12 and take the cross product in order to try and retrieve the 2nd London
Equation,  
ms ∗
∇ × ~∇θ = ∇ × J + e A (5.17)
ns e∗
Again submit our definition of the London Parameter Λ,
~
∇ × ∇~∇θ = ∇ × ΛJ + ∇ × A (5.18)
e∗
Observe that Vector Calculus requires that the Curl of a Grad is zero and that ∇ × A = B,

0 = Λ∇ × J + B (5.19)

Recall Ampere’s Law as ∇ × B = µ0 J and re-arrange to retrieve the 2nd London Equation:

Λ 2
−B= ∇ B (5.20)
µ0

10
5.2 London Penetration Depth
The first of many characteristic penetration depths, the London Penetration Depth λL defines
the extent to which a magnetic field will penetrate a Superconductor according to Equ.5.7,

Λ d2 B
−B= (5.21)
µ0 dx2

We thus define the London Penetration Depth as,

Λ ms
λ2L = = (5.22)
µ0 ns e∗2 µ0
 
in order that we may express the solutions to Equ.5.7 as B = B0 exp − λx
L

Figure 5.1: London Penetration Depth of a Superconductor

Figure 5.2: Diagrammatic view of the London Penetration Depth of a superconductor of various
thickness

5.2.1 Typical Penetration Depths

• Elemental Superconductors: 30-60nm

• Alloy-based Superconductors: 150-300nm

• High TC Oxides (such as Y B2 CuO7−δ ) have 150nm at HARD AXIS and 500nm at EASY
AXIS.

5.3 London Current Direction

Turns out that the 2nd London Equation (Equ.5.7) is the 1-D Poisson Equation, i.e. it describes
an infinite plane superconductor.
Λ d2 B
=B (5.23)
µ0 dx2

11
We may use this to identify the direction of the London Current in the thin skin into which a
magnetic field may penetrate a Superconductor - given by λL - as follows:
 
B0 x
exp − = Jy (5.24)
µ0 λL λL

i.e. the London Skin Current travels perpendicular to the applied magnetic field direction.

5.4 Pippard Coherence Length

The London Model stumbles when it has to deal with fields that change rapidly spatially, which
is in fact a fairly big flaw in the theory as the flux and electric field within the superconductor
are not spatially constant due to the Meissner Effect. Let’s consider the non-local version of
Ohm’s Law:
R(R · E)exp Rl
Z

3σ
J(r) = dr0 (5.25)
4πl | R |4
where R = r − r0 and l is some characteristic length scale.
Pippard suggested that a similar adaption to the London Equations could take into account
non-uniform electric fields, only those electrons within ∼ kTC of the Fermi energy would be
involved in the superconductivity - these electrons belong to the momentum band shown below,
kTC
∆p ∼ (5.26)
vf

From Heisenberg, ∆x ≥ ~
∆p , we can generate the intrinsic Pippard coherence length, ξ0

~vF 0.18~vF
ξ0 ∼ = (5.27)
kTC kTC

By recalling E − dA dJ A
dt = Eeff = Λ dt we arrive at J = − Λ which in turn allows us to approximate
the London Equivalent of Equ.5.25,
 
Z R[R · A(r0 )]exp R
3 ξ0
J(r) = 4
dr0 (5.28)
4πΛξ0 |R|

Further, we can define the actual Pippard Coherence length, ξ(l) in the presence of scattering
centres as,
1 1 1
= + (5.29)
ξ(l) ξ0 l
Thus in those Superconductors where the Pippard Coherence length ξ0 is greater than the
mean free path of normal electrons in the self-same material, l, the Superconducting charge
carriers may bleed into other materials provided they are within ξ0 of the Superconductor.

6 Flux Quantization

Figure 6.1: Superconductor with a hole through which a flux, B, passes

12
 Let us consider the Supeconductor shown in Fig.6.1, a hole is cut through the material
through which an applied flux penetrates, first let us recall earlier, when working towards the
1st London Equation, Equ.5.13 and take closed loop integrals around a curve deep within the
material, I I I
~
∇θ · dl = ΛJ · dl + A · dl (6.1)
e∗
Observe that as θ is the angular frequency the LHS must remain constant or vary by 2πn, also
recall Stoke’s Theorem:
I Z Z
A · dl = ∇ × AdS = BdS = Φ (6.2)

As the curve we are considering is deep in the Superconductor no current density is observed
and Equ.6.1 simplifies to define the Flux Quantum (note that it is experimentally observed
that e∗ = 2e and n = 1):
~ h
2π ∗ = Φ0 = (6.3)
e 2e
Thus flux entering a Superconductor is quantized, interestingly we can observe the ghost of
the upcoming Cooper Pairs, that the superconducting charge carriers look like pairs of normal
electrons (i.e. e∗ = 2e). Let’s now consider the partial derivatives of Equ.?? - dG = −SdT −

Figure 6.2: The Gibbs Free Energy cost of the transition from the normal to superconductive
phase

BdH - and by identifying that dGN = dGS we can consider the entropy changes due to the
Superconductive phase change.
− SS dT − BS dH = −SN dT − BN dH (6.4)
(SN − SN ) dT = (BS − BS ) dH (6.5)
As required by the Meissner Effect the flux density inside the Superconductor is zero, thus the
entropy change from the n to S is always positive as dHdT is always negative,
dHc
∆S = −µ0 Hc (6.6)
dT
We realise therefore that the Superconductive phase is more ordered than the normal phase,
destroying the Superconductive state requires ∆S and is thus a 1st Order Transition whereas
cooling from the normal to Superconductive phase requires no entropy and is thus a smooth
transition.

7 Superconductor Phase Coexistences

In this section we consider the influence of strong magnetic fields on a Superconductor, through
treatment of the demagnetizing factor we observe that phase coexistence occurs in some Su-
perconductive systems and we also consider practical concerns with flux quanta entering the
superconductor.
We begin by considering the influence of a strong magnetic field applied to a Superconductor.

13
Figure 7.1: Electrical Resistance of the Intermediate Transition: Apologies for the low-quality
image but I don’t have the internet to learn Inkscape tricks

7.1 Zero Demagnetising Factor

Consider the simplest geometry in which a magnetic field, Ha , applied to a Superconductor
induces screening currents in order to exclude flux, this geometry is a long, thin sheet of material
lying parallel to the field as the field measured at the surface of the material will be equal to
the applied field, Ha , at every point - the demagnetizing factor is thus zero. Recalling that a
Superconductor has χ = −1 and that B = µ0 Ha χ from Equ.3.2 we observe,
1
∆G = µ0 H2a (7.1)
2
Allowing us to define, Hc , the critical field for the Superconductive phase.
 1
2∆GN →
−S 2
Hc = (7.2)
µ0

Note that this applies only to the thermodynamic critical field, i.e. zero Kelvin, experimentally
we determine that for finite temperatures,
 2 !
T
Hc (T ) = Hc (0) 1 − (7.3)
Tc

7.1.1 Interface Energy between Normal and Superconductive Phase

If not already made clear by the heading order, we consider here the interaction energy in
a Superconducting sample with zero demagnetizing factor, the Helmholtz Free Energy of the
sample, where fn0 is the free-energy density of the normal phase in absence of any field. in the
normal phase is given by,
H2 H2
Fn = V fn0 + V a + Vext a (7.4)
8π 8π
In the Superconductive phase the Meissner Effect excludes the field from the interior, by con-
2
sidering a macroscopic sample we can ignore edge effects to define Fs = V fs0 + Vext H 8π , the
a

difference being,
H2
Fn − Fs = V (fn0 − fs0 ) + V a (7.5)
8π
Recall Equ.3.17 so we can define:

Hc2
Fn − Fs |Hc = V (7.6)
4π
As the sample transits from the Superconductive to the normal phase the energy source main-
taining the constant field Ha must do work against the back emf induced as flux (quanta!!)
enters the sample resulting in a Helmholtz Free Energy cost of Hc2 /4π per unit volume.
That’s all fair and good for the Helmholtz, where B is held constant, but in the case of a

14
constant H we need to use the Gibbs which differs from the Helmholtz as the work done by
the generator must be accounted for (NOTE: here h stands for the internal microscopic field
as mentioned previously, rather than Plank’s Constant):

hH
g=f− (7.7)
4π
• In the normal phase h = B = H and thus
Ha2 H2
Gn = V fn0 − V − Vext a (7.8)
8π 8π
• In the Superconducting phase h = B = 0 inside of the material and thus
Ha2
Gs = V fs0 − Vext (7.9)
8π
The difference is thus:
Ha2
∆Gn→ − S0 ) − V
− S = V (∆fn0→ 8π
(7.10)

7.2 Non-Zero Demagnetising Factor

For complicated geometries (i.e. anything that isn’t a long thin sheet or cylinder - even a
sphere!) the Meissner Effect responsible for the expelling flux from the Superconductor means
that over part of the surface the field will be greater than Ha and thus some of the material
will enter the normal phase even though Ha < Hc .
Quickly consider the case of the sphere given in Tinkin!
A sphere has a demagnetizing factor of 1/3 and radius R, we know that inside of the sphere
B = 0 and the external field must satisfy ∇B = ∇ × B = ∇2 B = 0 with boundary conditions,

B→
− Ha as r →
− inf inity
(7.11)
Bn = 0 at r = R

where Bn is the normal component of B. This problem is (apprently) a standard boundary-

value problem and has exterior solution, where θ is the polar angle measured from the direction
in which Ha is applied.
Ha R3
 
cos θ
B = Ha + ∇ (7.12)
2 r2
The surface tangential component of B at the surface of the sphere is given by
3
(Bθ )R = Ha sin θ (7.13)
2
The field measured from 42 ≤ θ ≥ 138 is thus greater than Ha . Considering the equator
specifically, Bθ = 3H2 a an so the equatorial field reaches the critical field when Ha ∼ 2H3 ,
c

thus above this field strength we expect the normal phase to emerge over some regions of the
sphere - however the entire sphere may not enter the normal phase as that would destroy the
diamagnetism, leaving H = Ha ∼ 2Hc /3 for the entire spheres interior which is insufficient to
prevent superconductivity reappearing.
2Hc
< Ha < Hc (7.14)
3
Thus Equ.7.14 defines the range of fields where a coexistence (or the intermediate state) of the
normal and Superconducting phases must exist, the size of these domains depends in the value

15
of the (positive, in the case of Type 1 Superconductors) interface energy.
More generally, we expect to observe coexistence whenever,
Ha
1−n< <1 (7.15)
Hc
As the demagnetizing factor of an infinite flat sheet perpendicular to the field approaches
unity, if a superconductor is set-up as such it will always show the intermediate state. Such a
configuration is thus useful for both theoretical and experimental observation.

7.3 The Intermediate State in a Flat Sheet (Interface Energies)

Figure 7.2: Surface Energies of Type 1 & 2 Superconductors: a) Type 1, positive interfacial
energy, b) Type2, negative interfacial energy

For clarity we are considering an infinite flat sheet of thickness λL  d upon which a field
is applied perpendicularly average flux per area far from the sheet (i.e. a displacement that
would average out any inhomogeneities due to the sheet) of Ba . Observe that while hs = 0
there is some flux density hn (note: if surface defects are neglected hn = Hc in the normal
phases, we may therefore calculate the fraction of the sheet that is in the normal phase,

Figure 7.3: Image stolen from Tinkham: Schematic diagram showing magnetic flux channeling
through the normal laminate in the intermediate state of a Type 1 Superconductor. Flux
Density is Ba at large distances and zero or hn (∼ Hc in the cross section of the slab. The
normal regions are macroscopic, in constrast to the vortices in a Type 2 Superconductor,
which contain only a single quantum of flux

Ba
ρn = (7.16)
hn
The size and shape of the domain walls is a current topic of interest in Superconductor Theory
but we may estimate the distribution of domains through considering the surface energies F1 ,

16
accounting for the interface between normal and Superconducting domains, and F2 , which
accounts for the interface between the bulk sample and the volume defined by λL .
We introduce a phenomenological surface-energy term for F1 such that there is an additional
energy per unit area of interface given by,

Hc2
γ= δ (7.17)
8π
Now, we have a full derivation of the Helmholtz Free Energy cost of the interface interactions
in the intermediate state as given by Tinkham and I explain that relatively fully, however in
the notes we are given the Gibbs Energy with no derivation and I don’t know how to convert
between the two - hopefully will find time to ask Dr. Brunell this week. For simplicity I’ll just
include the level of detail in the notes.
The demagnetizing factor, n, may be expressed in terms of the thickness, t, and the radius of
the normal phases, a, and equally the average internal field of the sample may be expressed in
terms of these parameters as well,
t
n∼1−
2a
(7.18)
Ha 2aHa
Hi = =
1−n t

Generally the domains are of the order of 10−2 to 10−1 cm, there is an energy cost to the
material due to the external field of
1
U = µ0 Hc2 ωA (7.19)
2
where ω is the width of a normal phase.
This implies that the ω is kept small, however the surface energy cost of these interfaces favours
increasing ω,
A t 1
2t δ∆G ∼ 2 Aδ µ0 Bc2 (7.20)
ω ω 2
The total energy of each interface is thus,
ω tA
U= Aµ0 Bc2 + δµ B 2 (7.21)
2 ω 0 c
note: Aω is of course ρn .
Minimizing this energy w.r.t to ω allows us to calculate the equilibrium of the intermediate
state! Whoop. √
ωequib ∼ 2tδ (7.22)
It is therefore energetically favourable to separate normal phase regions, but how small can
they go? We already know from Section6 that flux is quantized, and we can use our definition
of Φ0 . Let us take the extreme case where λ  ξ in order that we can use the London Model
- recall Equ.5.19 and assume one quanta of flux entering the sample,

Λ∇ × J + B(r) = ẑΦ0 δ(r)

(7.23)
µ0 λ2L ∇ × J + B(r) = ẑΦ0 δ(r)

Recall that ∇ × B = µ0 Js and submit into the above,

λ2L ∇2 B + B = ẑΦ0 δ(r) (7.24)

17
We may now cast expressions for the flux and current density in the intermediate state using
a modified Bessel function, K λr ,
L

 
Φ0
B(r) = ẑ K0 (r/λL )
2πλ2L
  (7.25)
Φ0
J(r) = θ̂ K1 (r/λL )
2πµ0 λ3L

Figure 7.4: Current and Flux Density profile of the Intermediate State: Have a look, fun stuff...
oh, and tell me what Ψ means here.

7.3.1 Type 1 vs. Type 2 Interface Energies

It will turn out (though I don’t know if you’ll see the derivation here) that δ ∼ ξ − λ

• (Typical) Type 1 Superconductors: δ is positive and of the order 10−5 to 10−4 cm

• Type 2: Interfaces proliferate and gain negative surface energy.

7.4 Abrikosov Vortices

Abrikosov considered the distribution of vortices within the intermediate state and observed
interesting phenomena like HCP packing of vortices etc., but we’re not interested in that here.
We’re simply looking at Superconductors generally as they transit through the intermediate
state.
hv
In comparing Type 1 with Type 2 recall that ξ0 ∼ 0.18 k Tf and therefore as the critical
b c
temperature increases the Pippard length decreases which implies that Superconductors with
higher critical temperatures are Type 2 as they require λL > ξ.

Figure 7.5: M-H Loops: a) Type 1 Superconductor : The demagnetizing field is zero implying
that Hi = H and we see the perfect diamagnetism of the Superconductor bulk when ξ > λL , b)
Type 1 Superconductor : There is a non-zero demagnetizing field (n=1/2 in this instance) and
χ
recall that χm = 1+nχ , c) Type 2 Superconductor Observe that the intermediate state survives
to a higher field than Type 1 materials, d) Vortex Phase Diagram: Note that around each flux
quantum (i.e. normal region) a supercurrent is developed.

18
It is evident that flux vortices will interact, indeed they may be modelled as phase of matter in
their own right - as indicated by Fig.7.5.c - a more complete impression of the vortices phase
diagram is given below... as soon as I get the internets.

7.5 Flux Line Pinning

Figure 7.6: Flux Line Pinning: Image stolen from Dr. Brunell with thanks.

If we desire to use Superconductors for practical purposes, in particular conveying super-

currents, we are largely forced to choose Type 2 Superconductors as both Tc and Hc are higher
than possible in Type 1 materials, however those situations where we require supercurrent also
require that these supercurrents are huge and will more than likely place the material between
Hc1 and Hc2 , i.e. in the intermediate state.
Consider a supercurrent sufficient to produce sufficient flux quanta to generate an array of
vortices, we may consider these as individual lines of flux - or more usefully as stiff rods of flux
- they will however experience a Lorentz force due to the supercurrent permeating the material,
h
recalling that Ψ0 = 2e :
fL = JT × Φ0 z (7.26)
h
recalling that Ψ0 = 2e : These flux lines will thus move perpendicular to this supercurrent and
will therefore induce an electric field (E = B × v) which will dissipate power - this essentially
means the supercurrent traveling through the material is no longer super.
We are able to solve this problem through ’pinning’ these flux lines in position, fortunately
material imperfections (interfaces and grain boundaries) are suitable in many Type 2 Super-
conductors as the Pippard Length, ξ, is of the order of a few nm - we’re told by Dr. Brunnell
that Nuclear Reactors are often used as a neutron source in order to deliberately damage the
structure of a Superconductor in order to produce pinning centres.
Consider now an idealised Type 2 Superconductor in the intermediate state with no super
current (i.e. Jext = 0 then we know H = Ha at every point, this should be recognized at
the equilibrium state and thus no net force acts on any vortices and thus the vortices only
experience a net force due to the nonequilibrium part of the current density, Jext . We may
therefore express the force density on the vortices as,
B B
α = Jext × = (∇ × H) × (7.27)
c 4π
As a specific example consider a Superconducting wire of diameter d, the amount of energy
required to move a flux line is given by ∆U ∼ πξ 2 d − 12 µ0 Hc2 and thus

∆U
Fpinning = JT × B ∼ 1 (7.28)
2d

Once the flux lines break free of the pinning force - as they are want to do due to thermal
effects or increasing supercurrent - they experience some force density as given by Equ.7.27.
Flux pinning is a relatively simple phenomenological description of why a Superconductor will

19
levitate above a magnet.
NOW, there are a great many graphs for the flux pinning section in the notes but I don’t really
see the point of them - not to say they’re useless, rather I don’t know what they are for. So
instead let’s move on.

8 Ginzburg Landau Theory

The Ginzburg-Landau Theory is a modification of the Classical Landau Theory.
Landau Theory is a powerful tool for describing Phase Transitions with virtually no infor-
mation about the particular system known, only some Order Parameter, Q, which is zero at all
Critical Points and in the chosen disordered phase and some knowledge of the physical symme-
tries of each phase. The method is essentially a mean-field theory and works best for continuous
phase transitions (i.e. from normal to Superconducting). Some function, f, characterised by Q
may be expressed as,
f = f0 + aa Q + a2 Q2 + a3 Q3 + ... (8.1)
We are usually uninterested in the origin of the function, f, and thus discard the the first time,
f0 leaving: f = f0 (T ) + FL (T, Q). Furthermore, if the Order Parameter, Q, is a vector then all
ODD terms must be set equal to zero in order for f to be retrieved as a scalar function - i.e.
if we were considering the magnetization vector, M · M is a scalar, whereas M · M · M is not,
apparently.entropy change from the n to S is always positive As a final point, as the Q is assumed
to be v. small in proximity to the Critical Point higher terms quickly loose significance and we
only need the minimum number of terms required to describe the behaviour of the system.
Now for Ginzburg!
The Superconducting state has already been identified as more ordered than the Normal
phase from Equ.6.6 and we keep the London assumption of two-fluids, the Order Parameter is
thus the local density of Superconducting charge carriers and defined as,

ns = |ψ(x)|2
(8.2)
Ψ = |ψ| exp(iθr)

The free energy density f can thus be expanded as follows, where the fourth term is the magnetic
self energy as seen in Section.??,

e∗ 2 h2
 
2 β 4 1  ~
f = fn0 + α|ψ| + |ψ| + ∇ − A + (8.3)
2 2m∗  i c 8π

Observe that at the critical point, i.e. where ψ = 0, we retrieve the free energy of the normal
h2
state as given by Equ.7.4 (i.e. fn0 + 8π ) and that the differential free density is given by,

1
∆f = fn − fs = α|ψ|2 + β|ψ|4 (8.4)
2
We’re told that analysis of the above requires that β is positive in order for the theory to be
useful, however α is more difficult to consider, in order for the Order Parameter to be non-zero
we require α must be negative for T < Tc and positive for T > Tc , i.e.

α = α0 = (T − Tc ) (8.5)
dFs
Minimizing Equ.8.3 with respect to the Order Parameter gives dφ = 2αψ + 2βψ 3 and we can
define the equilibrium position as,
2 α
ψequib =− (8.6)
β

20
note: this is essentially the value of the Order Parameter taken deep in the Superconductor far
from edge effects. Submitting ψequib2 into Equ.8.3 and recalling the Thermodynamic Critical
Field, Equ.3.17, we arrive at
H2c α2
∆f = fn − fs = − =− (8.7)
8π 2β
In order to derive the Landau-Ginzburg Equations we need to convert Equ.8.3 into Gibbs Free
Energy by integrating over the volume of the Superconductor to give the Ginzburg Landau
Energy function (only valid close to Tc ),
Z   Z Z
2 β 4 1 ∗ 2 1
2
G s = Gn + αψ + ψ dV + (−i~∇ − e A) ψ dV + B(r) − Bapp dV
V 2 V 2ms V 2µ0
(8.8)
So, I’m told that the differential Gibbs Free Energy is given by,
Z  2 !!
β 1 ~
∆G = αψ 2 + ψ 4 + ψ∗ ∇ − e∗ A ψ dV
V0 2 2ms i
Z    
~ ∗ ~ ∗ (8.9)
+ ψ ∇ − e A ψ · n̂ dS
2ims s i
Z
1
2
+ B − Bapp dV
2µ0
where 1st term is the real part, 2nd term is the imaginary component and 3rd term is equal
to zero! WeR obtain the 1st Ginzburg-Landau differential equation by minimizing the real
component ( V 0 ) with respect to ψ or ψ ∗
 
1 ~
2αΦ + 2βψ ∗ ψ 2 + ∇ − e∗ A ψ = 0 (8.10)
ms i
this may also be re-cast as
 2
1
∗ ~
αψ + βψ ψ + 2
∇ − e∗ A ψ=0 (8.11)
2ms i
The surface of the Superconductor, or rather the Ginzburg-Landau Boundary Condition is
∗


given by i ∇n − e An = 0 where the n subscript is normal to this surface. Now to generate
~

the 2nd Ginzburg-Landau Equation start here,

ie∗ ~ ∗ e∗2 A ∗
Z  
d∆G ∗
= = − (ψ ∇ψ − ψ∇ψ ) + ψ ψ + Js dV = 0 (8.12)
dA V ms ms
and so the superconducting charge density is given by:
ie∗ ~ ∗ e∗
Js = (ψ ∇ψ − ψ∇ψ ∗ ) − Aψψ ∗ (8.13)
ms ms
By recalling Equ.8.2 and assuming ψ is constant in space, we may express the Ginzburg-Landau
Superconducting Charge Density more succinctly as,
e∗2 A αe∗2 A
Js = − |ψequib |2 = (8.14)
ms βms
Now we move towards defining the Ginzburg-Landau Coherence Length, ξGL , through observing
that µ0 J = ∇ × B we move forwards:
e∗2 µ0 ns A
∇×B=−
ms
(8.15)
e∗2 µ0 ns e∗2 µ0 ns
∇×∇×B=− ∇ × A − ∇2 B = − B
ms ms

21
Observe through the definition of the London Equation, Equ.??, we can redefine the parameter
in terms of the Ginzburg-Landau Theory,
 2
Λ ms ms B
λ2L = = = (8.16)
µ0 ns e∗2 µ0 µ0 e∗2 |α|

note that this kind of reinforces our requirement that α is negative below Tc . Now recall
Equ.8.10 and set A and re-arrange,

~2 2
− ∇ ψ − ψ (α + βψ ∗ ψ) = 0 (8.17)
2ms
ψ
Through recalling ψequib = − αβ and f = ψequib we may re-arrange the above to a useful form,

~2 2
αf ψequib + βf 3 ψequib
3
− ∇ f ψequib = 0
2ms
β −α ~2
f + f3 − ∇2 f = 0 (8.18)
α β 2ms α
~2
∇2 f = f − f 3
2ms |α|
Recalling that as ψequib is the Order Parameter value at r →
− ∞ the value at every other point
is small in comparison f →3 − 0 we arrive at the Ginzburg-Landau Coherence Length,
 12
~2

ξGL = (8.19)
2ms |α|
note that Equ.8.5 suggests this diverges as T → − Tc
2 2 2
Now consider that λ , ξGL , Bc all have α dependence and multiply the square of the three
for the sake of it,
ms B~2 µ0 α2 ~2 h2 Φ20
λ2L ξGL
2
B2c = = = = (8.20)
µ0 e∗2 α2ms αB 2e∗2 8πe∗2 8π
Consider a slightly modified version of the 1st Ginzburg-Landau Differential Equation,
 2
1 ~ ∗
αψ + ∇−e A ψ =0 (8.21)
2ms i
This is essentially the Schrodinger Equation for motion of a charged particle in a magnetic field,
and we will now attempt to define α in terms of the circular frequency, ωc , and the particles’
energy, Ek
e∗ B
ωc =
m
2 2
s  (8.22)
~ k 1
Ek = + n+ ~ω
2ms 2
note that k is parallel to B. The maximum field, B, we may apply is when k = 0 and n = 0
allowing us to define α,
1 e∗ B~
− α = ~ωc = (8.23)
2 2ms
2 e∗ ~B
However we are already aware (apparently) that α = 2m~ξ2 = 2ms and this in turn allows us
s GL
to redefine the critical field, Bc
~ Φ
Bc = ∗ 2 = 2 0 (8.24)
e ξGL ξGL

22
From this we can calculate the field,
√ λ
B= 2Bc (8.25)
ξGL
λL
Turns out it is useful to define κ = ξGL

• If κ < √12 then magnetic fields smaller than Bc will sustain circulating Superconducting
charge carriers around flux vortices, this is a Type 1 Superconductor. Remember that
Type 1 Superconductors have a negative interface energy in the intermediate state.
• If κ > √12 then it is possible for magnetic fields greater than Bc to sustain circulating
currents of Superconducting charge carriers, this is a Type 2 Superconductor. Remember
that Type 2 Superconductors have a positive interface energy in the intermediate state.

8.1 Type 2 Superconductors

2 , we can define the field
Flux vortices have a radius of ξGL and will encapsulate an area of πξGL
at which the Superconducting currents may no longer be sustained around such a flux quanta
as Bc2 ,
Φ √
Bc2 = 02 = 2Bc κ (8.26)
πξ
for a diagrammatic explanation of these two critical points refer to Fig.7.5. Now let’s consider
Hc1 , at this critical point the intermediate state the Gibbs Free energy of the Meissner and
Mixed state must be equal and we define these quantities as such,
Z Z
Gmeissner = Fmeissner − Hc1 BdV Gmixed = Fmixed − Hc1 BdV (8.27)
sample sample

R know from the Meissner Effect in this phase B = 0 and thus Fmeissner = Fmixed −
As we
Hc1 sample BdV , we then calculate the free energy of one flux line, ,
Z
∆Fmix→− meissner =  = Hc BdV = Hc1 Φ0 (8.28)
sample

Including only those terms from the Helmholtz Free Energy that change during the transition
the free energy of each flux line may be expressed as,
2 !
B2
Z 
1 ~ ∗ 2
= ∇−e A Φ + dV (8.29)
2ms i 2µ0

Recalling that ~i ∇ − e∗ A is equivalent to the Superconducting current density, and then in turn
that µ0 Js = ∇ × B, Z 
1 
= λ2 (∇ × B)2 + B2 dV (8.30)
2µ0
Through vector identities we ca simply further,
λ2
Z I
1 2


= B + λ ∇ × ∇B BdV + B (∇ × B) · dS (8.31)
2mu0 2µ0
Consider the first term, we know from the Section.?? that ∇2 B = B and so this term equates
to zero and so our only consideration is the 2nd term around the area the flux encloses which
is a cylinder with , A = 2πξGL ,
λ2
 
dB
= B· · 2πr (8.32)
2µ0 dr r=ξ GL

23
Recall our Bessel functions for the field from earlier, Equ./refequ:Bessels, to identify:
Φ2
 
λ
= ln (8.33)
4πµ0 λ2 ξGL
Now consider Equ.8.20 in order to generate Hc1 in a more useful form,
Φ0 −1 Φ0
Hc1 = = lnχ (8.34)
 4πµ0 λ2

9 AC Conductivity according to Ginzburg-Landau

Figure 9.1: AC Conductivity of a Superconductor : Red dotted line: The conductivity profile of
a normal metal due to an AC current.

Consider the AC conductivity of a normal metal given by the Drude Model,

 ∗  
ne τ 1
j exp (−iωt) = E exp (−iωt) (9.1)
m 1 − iωτ
Let’s take the original London Assumption of infinite mean free path for Superconducting
ns e2
charge carriers in order to arrive at σ(ω) = iωm , for ω = 0 we have a finite real conductivity,
s

πns e2
Re[σ(ω)] = δ(ω) (9.2)
ms
A plot of this real contribution is shown in Fig.10.1, where ∆ is a parameter of the Supercon-
ductor to be reconciled later. The London Two-Fluid model requires that at finite temperatures
there is some nn and these normal charge carriers will be subject to the applied electric field -
the conductivity will therefore not be purely inductive. For ω < ∆ ~ this extra contribution to
the conductance is low, however as ω increases so does this non-inductive contribution due to
the normal charge carriers - the sharp jump in conductivity at |ω| = 2∆ ~ cannot, however, be
explained by the Two-Fluid Model.
Please note, it is important for later understanding that this is reference with Section 3.4!

10 Microscopic Theory of Superconductivity - The BCS The-

ory
Turns out that a thorough Microscopic Theory of Superconductivity reveals that phonons are
largely responsible for the Superconductive phase.
A cartoon picture of this is that electrons displace the phonon lattice and are over-screened
developing positive regions that interact with one another. But more on that later, hopefully.
From a treatment of the Superconducting charge carriers and the phonon lattice the Cooper
pair construction becomes apparent and responsible for supercurrents, however this takes a long
time and requires us to introduce many topics. In addition to topics we need some Quantum
Information Theory nomenclature to make our job easier, the so-called Annihilation Operators.

24
10.1 Annihilation Operators
First identify an empty electron k-state as |0k i and an occupied k-state as |1k i, now we may
identify:
The Creation Operator:
c+
k |0k i = |1k i (10.1)
The Annihilation Operator:
ck |1k i = |0k i (10.2)
As electrons are fermions we can identify a range of identities:
c+
k |1k i = 0
ck |0k i = 0
(10.3)
ck cj = −c∗j c∗k
+ +

∗
c+
k cj = −cj ck

Further we wish to identify,

c+ + +
k cj |0k i = 0ck ck |1k i = |1k i (10.4)
where c+
k ck = nk and thus ck c∗k = 1 − nk Now consider the Ground State of a Fermi Gas for
all k-states (both up and down spin) up to kF ,
Y
c+


kσ |0i (10.5)
k<kF

10.2 BCS Coherent States

The Microscopic Theory of Superconductors requires Coherent States, something we will see
later when we discover an attractive interaction between electrons due to virtual phonon cre-
ation, for ease we choose the well understood Quantum Harmonic Oscillator to generate our
Coherent States:
ρ̂2 mωc2 x̂2
H = + (10.6)
2m 2
We now propose two “Ladder Operators that take the system up and down an energy level:
  2 1 
ρ̂
− i mω
+ 1 2
• a = √ √
2
c
x̂ - CLIMBING THE LADDER (negative)
~ωc 2m
  1 
mωc2
√ρ̂
2
• a+ = √1 2m
+i 2 x̂ - FALLING DOWN THE LADDER (positive)
~ωc

These ladder operators are analogous to the Annihilation Operators we defined previously
(assuming fermions!) ,
cc+ + c+ c = n̂ + 1 − n̂ = 1 (10.7)
So that we may define:
1
a+ |Ψn (x)h= (n + 1) 2 |Ψn+1 (x)h
1
a|Ψn (x)h= (n) 2 |Ψn (x)h (10.8)
a+ a|Ψn (x)h= (n) |Ψn+1 (x)h
This allows us to express any energy level as a superposition of the ground state with some
excited state,
∞
α2 αn
 
α X
|αi = C ψ0 + √ ψ1 + √ ψ2 + ... = C √ ψn (10.9)
1! 2! n n!
 
|α|2
where C = exp − 2 in order to normalise the Coherent State Distribution.

25
10.2.1 Trial Ground State Wavefunction from Coherent States
Let us now consider the ground state |0i using Equ.10.9:
(αc+ ) (αc+ )2
 
|0i + ... = C exp αc+ |0i


|αi = C |0i + √ |0i + √ (10.10)
1! 2!
We should therefore be able to express the ground state wavefunction for all k as,
∞
!
X
+
ΨBCS = (const.) exp αk Pk |0i (10.11)
k

+ 2
where we define Pk+ = c+ +


k c−k and observe that Pk = 0. From our knowledge of exponentials
we can simplify this first as a product sum and the through the observation that higher order
terms quickly converge to zero,
∞
Y ∞
Y
exp αk Pk+ |0i = (const.) 1 + αk Pk+ |0i



ΨBCS = (const.) (10.12)
k k

Let us normalize this by setting some complex parameters:

1
uk =
1 + |αk |2
αk (10.13)
vk =
1 + |αk |2
Our trial wavefunction for the BCS Ground State is thus,
∞
Y
u∗k + vk∗ Pk+ |0i


ΨBCS = (10.14)
k

Let us now minimize the energy of this Ground State using the Schrodinger Equation where
the Hamiltonian is the Quantum Harmonic Oscillator Hamiltonian given before,
E = hΨBCS |H |ΨBCS i
(10.15)
X X X
K |Vk∗ |2 K 1 − |Uk |2 − V Vk Vk∗0 Uk Uk∗0


E=2 +2
0
k

where the first terms represents electrons, the second holes and the third the interaction energy
between the two. After minimization we find
 
1 
|Vk | = 1 − 

2 
2 ∗ 2 + c2k
P

V k uk vk
  (10.16)
1 
|uk | = 1 + 

2 
2 ∗ 2 + 2k
P

V k uk vk

Now define the parameter ∆ = V 2k u v∗ +c2k and recall that |uk |2 + |vk |2 = 1 Now consider
P
k k
the probability of finding an electron with momentum k in spin state ↑:
hnk↑ i = hΨBCS |Pk+ |ΨBCS i = hΨBCS |c+
k↑ ck↑ |ΨBCS i
 
= h0| uk + vk C−k↓ Ck↑ ck↑ ck↑ u∗k + vk∗ C−k↓

+ + +
Ck↑ |0i
(10.17)
= h0|vk c−k↓ vk∗ Ck↓
+
|0i
hnk↑ i = |vk |2

26
and so |uk | is the prob. of a HOLE with the same momentum and spin state! We can therefore
observe,
∆
uk vk∗ =
1 (10.18)
2 2k + ∆2 2
P ∆
and generate a self-consistent expression for ∆ = V 1 which looks similar to the
2(2k +∆2 ) 2
Cooper Pair Binding Energy that we will generate later,
 
1
|∆| = 2~ωD exp − (10.19)
N (F )V

Figure 10.1: The Emergence of paired Electrons as a function of the electron momentum: Note
how similar this plot appears to the Critical Temperature curve of the Superconductor, or not

10.3 Attractive Potential Between Electrons

Recall the electrostatic potential for two bear electrons, where R = r1 − r2 ,
e2
V (R) = (10.20)
4π0 |R|
Now consider the electrons are present within a lattice comprised of positive ion cores, these
ions will screen the electron interactions according to the Thomas-Fermi Equation,
e2
 
|R|
V (R) = exp − (10.21)
4π0 |R| rT F
where rT F ∼ a, the lattice parameter. Note that while this reduces the replusive potenital
between the two electrons it does not make them attract! Consider two electrons incident on
this lattice as shown in Fig.10.2, they may create a virtual photon! Momentum is conserved in

Figure 10.2: Creation of a Virtual Phonon:

0 0
this process, i.e. k1 = k1 + q and k2 = k2 − q, however total energy is not as,
0 0
|k1 |2 + |k2 |2 6= |k1 |2 + |k2 |2 (10.22)

However as scattering events occur over a very small time period Heisenburg’s Uncertainty
Principle comes to the rescue! We can find the scattering amplitude according to,
0 0
Vk0 ,k0 ,k = hk1 + k2 |V (R)|k1 + ks i (10.23)
1 2 1 ,k2

27
 Figure 10.3: k-state Overlap:

The size of the overlapping regions of k-states in Fig.10.3 is proportional to the number of
electrons that may scatter in such a way to generate a virtual photon, it is clear that when
the k-states overlap there is a maximum number of electrons to work with and this is acheived
0 0 0
when k1 = −k2 = k and k1 = −k2 = k , i.e. when electrons have opposite momentum. This
allows us to simplify Equ.10.23 to form:

e2
Vk,k0 = (10.24)
 (k2s + q2 )
To correct for taking into account only small changes in energy, and observing that ωq is the
phonon frequency, we arrive at
!
e2 ωq2
Vk,k0 = 1+ 2 (10.25)
 (k2s + q2 ) ω − ωq2

3n
where ks = 4 is the screening parameter... BUT need to check this value with Dr. Brunell or
F
a book or something. It is therefore clear that when ω ≤ ωq the potential between to electrons
becomes attractive.
In order to move on in theory we need to make a few assumptions, firstly the phonon
spectrum is given by the Debye-Model of Solids and that the only phonons of interest to us
occur at the Debye Frequence, ωD (this is essentially the highest vibrational mode that a solid
k 2


may support and is related to the speed of sound through the material and ωD ∼ m where
0
k is the spring constant). The potential Vk , k between the electrons is also assumed to be
constant for energies less than ~ωD and zero above the Debye Energy, as shown in Fig.10.4.
We can now construct a Hamiltonian to describe the scattering of electrons relative to the

Figure 10.4: Potential Assumptions for the BCS Theory:

Ground State given by Equ.10.5,

X
H = −V c+ c+ c c
k0 −k0 k −k
k −F
X
+2 |k |c+ c+ c c
k0 −k0 k −k (10.26)
k>kF
X
+2 |k |ck0 c−k0 c+ +
k c−k
k<kF

28
where the first term is the scattering term, the second is a Kinetic Energy Term found by
assuming a spherical Fermi Surface at 0K and all states up to F are full (and thus represents
the Kinetic Energy of any electron pairs above the Fermi Energy), and the third term represents
the Kinetic Energy of any hole pairs below the Kinetic Energy under the current assumptions.
In order to use this lovely Hamiltonian we need to construct the two-electron wavefunction,
we achieve this by considering a free-electron like system held t 0K to which we add 2 electrons
at an energy just above k , with opposite momentum, and spin states σ1 and σ2 :

Ψr1 ,r2 ,σ1 ,σ2 = exp (ikR) ρ (r1 − r2 ) ρspin (σ1 , σ2 ) (10.27)

observe that the preceding exponential is equal to 1 as we have already defined that k1 =
−k2 and k ≡ k1 + k2 = 0. As electrons are fermions we require that Ψr1 ,σ1 ,r2 ,σ2 = −Ψr2 ,σ2 ,r1 ,σ1
which means that: Either
• ρ(r1 − r2 ) = −ρ(r2 − r1 ) or

• ρspin (σ1 − σ2 ) = −ρspin (σ2 − σ1 )

As we are considering phonon scattering we require that ρ(r1 − r2 ) is large for r1 ∼ r2 and that
it should be symmetric, which therefore means the spin component must be anti-symmetric
(i.e. in the spin singlet state). The symmetric spatial component of the wavefunction may be
expanded as a Fourier series as an “s-wave”, and the spin-component as well!
X
ρ(r1 − r2 ) = ρk exp (ik (r1 − r2 ))
k
(10.28)
1

ρspin (σ1 − σ2 ) = √ ψ↑ ψ↓ − ψ↓ ψ↑
2
Allowing us to express the total wavefunction in terms of Bloch States: ψk↑ = exp (ik · r),
" #
X ψk↑(r ) ψk↓(r )
Ψr1 ,r2 ,σ1 ,σ2 = ρk 1 2 (10.29)
ψ−k↑(r ) ψ−k↓(r )
k 1 2

We want to calculate the complex coefficient ρk and this requires us to minimize H |ψi = E|ψi
with respect to E. Observe that hψk |H |ψi =iψk |E|ψh and move on to the next line:
X
Eρk = 2k ρk − V ρk0 (10.30)
k0
P
Now let C = k ρk
1
ρk = −CV (10.31)
E − 2k
and so,
X X 1
C= ρk = −CV
E − 2k
k k
X 1
1 = −V (10.32)
E − 2k
k
Z ~ωD
dE
1 = −V N (F )
0 E − 2k
and submit the above back into Equ.10.30 to finally obtain the binding energy for Cooper Pairs,
 
−1
E = −2~ωD exp (10.33)
N (F )V

29
10.3.1 Quick Rundown of Cooper Pair Properties
Electrons have opposite momentum and thus:

• Electrons occupy the spin singlet state, i.e. the spins are anti-symmetric

• The Cooper pair is an s-wave

• Cooper pairs appear independent of sample size (i.e. V )

10.4 Excitation from BCS Ground State

Figure 10.5: The Normal Electron Distribution in a Metal :

Let us consider excitation from the BCS Ground state as the breaking of electron pairs
(Cooper Pairs), we do this by setting the occupation probability for +k = 1, −k = 0 or vice
versa as so:
+
γk0 = Uk∗ Ck↑
+
− Vk∗ C−k ↓
(10.34)
+
γk1 = Uk∗ C−k↓
+
+ Vk∗ Ck ↑
We may ask what is the total energy of the Superconductor, to answer this we must assume
∗ | ∼ 1 and |V ∗ ∼ 0 which means the operator γ + is simply proportional to
that for k  kF , |UK k k0
+
Ck↑ . The Hamiltonian has two terms, EF due to the Fermi energy (the first and second term
in the expression below) and EP due to the pair interactions (the third term in the expression
below): X X X
K |Vk∗ |2 + 2 K 1 − |Uk |2 − V Vk Vk∗0 Uk Uk∗0


E=2 (10.35)
k0

We can thus define δEK = K 1 − 2|Vk |2 and we realise that δEP is convoluted (NOT formal

convolution) due to a great many interaction terms at the Fermi Energy,

X
δEP = −2V Uk Vk Uk∗0 Vk∗0 (10.36)
k0

and define: X
∆=V Uk0 Vk∗0 (10.37)
0
k

δEP = −2V Uk Vk ∆ (10.38)

Now this is jiggery-pokery but seemingly follows the route taken in Tinkham, identify that
δEF is truly Ek , the energy associated with excitation of each electron pair and then define
outright that at k = kF :

1
δEK = 2K + ∆2 2 (10.39)

30
 Figure 10.6: Electron or hole like charge carriers:

Figure 10.7: Quasi Particles generated under strange conditions:

10.5 Density of States - Realising DELTA

Note that in the transition from the normal to Superconducting phase we have not changed
the density of states, and assume that Nn (k ) ∼ Nn (F ) in order that the number density of
normal charge carriers is constant,
dEk d
NS (Ek ) dk = Nn (k ) k dk
dk dk
dk (10.40)
NS (Ek ) = Nn (k )
dEk

Recall and substitute Equ.10.39 into the above,

Ek
NS (Ek ) = Nn (k )
1 (10.41)
2K + ∆2 2

It is evident therefore that there is a a gap in the density of states for electrons in the Super-
conductive phase, this is shown diagrammatically in Fig.10.9. We see evidence of this energy
gap for electrons in the Heat Capacity, Electron Tunneling and the AC conductivity of the
Superconducting phase.
It is important to note that the pairing interactions terms were calculated at zero temper-
ature, at finite temperatures electrons will be removed from the pairing interaction term and
reduce the size of the energy gap in the density of states in terms of some function f (k, T ),
X
∆ (T ) = V Vk∗ Uk (1 − 2f (k, T )) (10.42)
k0

Now I do NOT understand the next line of the derivation so just take it, take it good. Oh, do
however note that Ek represents the Superconducting Charge Carriers excitation energies and
similarly k are the normal excitation energies!
 
X 1 2
∆ (T ) = V ∆ (T ) 1 −   
2Ek E
k exp k
kB T + 1
 
1 X 1 Ek (10.43)
= tanh
V 2Ek kB T
Z ~ω  
1 D 1 Ek

= tanh
V N F 0 2Ek kB T

31
Figure 10.8: The E-k Dependence of Different models: Section A: “Hole” Type Excitations,
Section B : A BCS Superconductor (i.e. Type 2) in the Mixed State, Section C : Electron Type
Excitations

Figure 10.9: The Electronic Density of States in the Sueprconducting Phase: Observe that the
gap here of 2∆ defines the region in which Cooper Pairs are generated

While we may evaluate this numerically we require that at Tc the Superconducting phase
disappears, i.e. ∆(T ) = 0 and Ek = k ,
 
Z ~ω tanh Ek
1 D kB Tc

= (10.44)
V N F 0 E
 
1
This solves to produce kB Tc = 1.13~ωD exp − N  V , now recall Equ.10.33 and Equ.10.39
( F)
in order to observe/calculate/marvel at,

2∆T =0 = 3.52kB Tc (10.45)

Thus the ∆ of the Ginzburg-Landau and BCS theory are shown to be the same by Garkov!

11 The Josephson Effect & Junctions

Let us start by considering the Superconducting Wave Function as
√
Ψ = Ψ0 exp i (θ(r) − ωt) = ns exp i (θ(r) − ωt) (11.1)
2 0
where ω = ~F . Furthermore we consider two Superconducting blocks separated by d, as d → −
the Superconducting wavefunctions of the two materials interact and eventually become phase
locked! We may therefore express the time evolution of the wavefunctions as so,
δΨ1
i~ = U1 Ψ1 + KΨ2
δt (11.2)
δΨ
i~ 2 = U2 Ψ2 + KΨ1
δt

32
 ∗
where U1 = −U2 = − e 2V and K is some coupling constant. Substitute the wavefunction
proposed at the start and separate the real and imaginary components!

• Imaginary Parts
δns1 2 1
= K (ns1 , ns1 ) 2 sin (θ2 − θ1 )
dt ~ (11.3)
δns2 2 1
= − K (ns1 , ns1 ) 2 sin (θ2 − θ1 )
dt ~
• Real Parts
1
e∗ V

δθ1 K ns2 2
=− cos (θ2 − θ1 ) +
dt ~ ns1 2~
1 (11.4)
e∗ V

δθ2 K ns1 2
=− cos (θ2 − θ1 ) −
dt ~ ns2 2~
δn δn
Now observe that dts1 = − dts2 represents the rate of Cooper pair tunneling between the two
Superconductors at a rate that depends on sin (θ2 − θ1 ) = sin (φ) and so we may define the
current between the two as,
δn2 Ke∗ 1
J = e∗ = (ns1 ∗ ns2 ) 2 sin (φ) (11.5)
dt ~
• D.C. Josephson Equation

Jc = J0 sin φ (11.6)

• A.C. Josephson Equation

d (θ2 − θ1 2e
= V (11.7)
dt ~
We can actually use the BCS Theory to derive
 
G π∆(T ) ∆(T )
J0 = N tanh (11.8)
A 2e kB T

where GN is the conductance of the junction for electrons of energy greater than the the energy
gap found in the density of states. The A.C. Josephson Equation Equ.11.7 implies that applying
a voltage V to the junction will impart a phase difference between the two Superconducting
wavefunctions, while the D.C. Josepshon Equation Equ.?? informs us that a constant voltage
1 2e 2e
will generate an oscillating current of 2π ~ V = h V , i.e. 483.T Hz per Volt.
Let us now consider a better define junction, more specifically the one shown in Fig.11.1!
We already know about flux quanta but let us briefly show that the phase difference picked
up around the current loop identified is single valued according to 2nπ by identifying that
points A, B, C, D are deep within the Superconductors far away from surface effects and then
considering the phase difference collected around the loop,

• φAB Josephson Current due to Cooper tunneling, i.e. φ0

∗
• φBC given by BC e~ Adx
R

• φCD Josephson Current due to Cooper tunneling, i.e. −φ0

∗
• φDA given by DA e~ Adx
R

33
Figure 11.1: A Josephson Junction: Two Superconductors are brought close enough together
for their independent wavefunctions to become phase locked through interaction, applied field
B induces surface screening currents (dotted lines, the diagram on the right indicates the closed
loop considered later).

e∗
Thus the total phase difference is related to the magnetic field as so, φ0 + ~B + 2nπ which
allows us to define the critical current (by recalling Equ.??),
R
Ic = J0 (x) sin φdx
e∗ (11.9)
Z  
Ic = J0 (x) sin φ0 + Bdx dx
~

Now consider only the imaginary component of the above and set eiφ0 = 1 as we want the
maximum current! Z  ∗ 
e
Ic = J0 (x) exp (iφ0 ) exp i Bdx dx
~
Z  ∗  (11.10)
e
Ic = J0 (x) i Bdx dx
~
We can therefore consider the Josephson junction as a diffraction experiment analogous to
Optics.

Figure 11.2: Characteristic Josephson Junction I-V Curve with SHUNT Resistor

34