Superconductivity: Ginzbrug-Landau, BCS and

II
Beyond

P
CM
Prof. Dr. Narayan Prasad Adhikari
Professor
Central Department of Physics
Tribhuvan University Kirtipur, Kathmandu, Nepal

December 7, 2022
PA

1
 Introduction General properties Revisited

II
P
CM
• Discuss from separate pdf/ppt file. Here start from GL theory.
PA

2
 Ginzburg Landau Theory
-Superconductivity

II
P
CM
PA
Figure: Ginzburg (Nobel prize-2003) and Landau (Nobel prize-1962)- Two
Russian scientists who first gave theory of superconductivity, even though
it is Phenomenological
3
 Ginzburg Landau Theory
-Superconductivity

II
• Some how; N ∼ 1023 intercating fermions, with ψ(~r1 , ~r2 ...~rN )
antisymmetric many body wave function; behaves like a giant

P
single, zero spin, bosonic quantum state
ψ(~r) = |ψ(~r)|exp (iφ(~r)) (1)

CM
with a common phase φ(~r) = φ = a constant in equilibrium
and possible slow variations.
• Order parameter: |ψ(~r)|2 represents ’superconducting electron
density’. The mean value, or most probable (ψ̄) value must be
nonzero below Tc and zero abve Tc . ψ̄ is the order parameter.
We try to describe the simplest case, of ψ ”small” and slowly
PA
varying, ∇ψ(~r) ’small’ and near transition temperature
T
||≡ −1 1 (2)
Tc
4
 GL Theory -Superconductivity

II
• Roughly speaking ψ(~r) corresponds to the center of mass wave
function of the BCS pairs.

P
• Variational free energy density: We assume the most
probable or mean or best value (ψ̄) is that which minimizes a

CM
variational free energy;
Z
F = d3 rf (ψ(~r), ∇ψ(~r)) (3)

where f is the free energy density. ψ(~r) =⇒ complex;

∇ψ(~r) =⇒ complex vector. But f is real, scalar so;
PA
 
f ≡ f |ψ(~r)|2 , |∇ψ(~r)|2 (4)

5
 GL Theory -Superconductivity

II
• Inclusion of the magnetic field:

P
• In QM =⇒ P~ → h̄i ∇ in zero magnetic field.

CM
∗
 
~ r) in the presence of magnetic field
• And P~ → h̄i ∇ − qc A(~
• This is called ”Minimal Coupling”.
• The ’minimum coupling’ refers to a coupling between fields
which involves only the charge distribution and not higher
multipole moments of the charge distribution
PA

6
 GL Theory -Superconductivity

II
• For a Gauge transformation (Hamiltonian OR Langrangian left
invariant under a continuous group of local transformations. An

P
invariant is a model that holds true no matter the mathematical
procedure applied to it. This is the concept behind Gauge

CM
invariant) to be valid,
~ r) → A(~
(i) em vector potential A(~ ~ r) + ∇χ
(ii) Charge particle wave function; ψ(~r) → ψ(~r) exp ( iqχ
h̄c
) Then
p2
the Hamiltonian H = 2m + V (~r); Hψ(~r) = Eψ(~r) is left
invariant. Above discussion is for QM of a single particle. Whats
PA
about for the GL, ψ(~r) ”Effective wave function” that comes
from many particle? (SOME HOW! Assume same!)

7
 GL Theory -Superconductivity

II
 2
 
iq ∗ ~
 2 ~ r)
∇ × A(~
∴ f = f |ψ(~r)|2 , ∇ − A(~r) ψ(~r) + (5)
h̄c 8π

P
Note the addition of magnetic field
Where q ∗ is the effective charge of some ”clump” of electrons that

CM
plays a role in superconductivity; q ∗ = −2|e| paired electrons. Free
energy Density
We assume ψ(~r), p~ψ(~r) small so that only first few terms come in
the expression of f . Therefore,
f = fn + a|ψ(~r)|2
PA
2
(6)

b h̄2

iq ∗ ~
 2 ~ r)
∇ × A(~
+ |ψ(~r)|4 + ∇− A(~r) ψ(~r) +
2 2m∗ h̄c 8π
 
h̄2
where  = TTc − 1  1 , a, b > 0 and 2m∗
to make this term like
Quantum KE. 8
 GL Theory -Superconductivity

II
We need to search for minimum f by variational method. Here

P
b
∆f = aψ(~r)ψ ∗ (~r) + ψ(~r)ψ ∗ (~r)ψ(~r)ψ ∗ (~r)
2

CM
h̄2 iq ∗ ~ iq ∗ ~
   
+ ∇− A(~r) ψ(~r) ∇ − A(~r) ψ ∗ (~r) (7)
2m∗ h̄c h̄c
~ r) 2
 
∇ × A(~
+
8π
We consider two different cases:
PA
Case (i) Absence of the field
Case (ii) Presence of the field

9
 GL Theory -Superconductivity

II
Case (i) Absence of the field
In this case;

P
b h̄2
∆f = aψ(~r)ψ ∗ (~r)+ ψ(~r)ψ ∗ (~r)ψ(~r)ψ ∗ (~r)+ ∗ (∇ψ(~r)) . (∇ψ ∗ (~r))
2 2m

CM
(8)
and Z
F = d3 r∆f (ψ(~r), ψ ∗ (~r)) (9)

Further for variational method ψ ∗ (~r) → ψ ∗ (~r) + δψ ∗ (~r). Then

Z
δF = d3 r [∆f (ψ(~r), ψ ∗ (~r) + δψ ∗ (~r)) − ∆f (ψ(~r), ψ ∗ (~r))] (10)
PA
Then make δF = 0 for ψ(~r) = ψ̄.

10
 GL Theory -Superconductivity

II
Then we get

h̄2

P
Z   Z
∗
3 2
d r aψ̄ + b|ψ̄| ψ̄ δψ + d3 r (∇ψ(~r)) . (∇δψ ∗ )ψ=ψ̄ = 0
2m∗
(11)

CM
Now using
~ = ∇φ.A
∇.(φA) ~ + φ(∇.A)
~ (12)
~ = ∇ψ we get
with φ = δψ ∗ ; A

(∇ψ(~r)) . (∇δψ ∗ ) = ∇.(δψ ∗ ∇ψ) − δψ ∗ (∇2 ψ) (13)

PA
Therefore second part of above eq 11 is;

11
 GL Theory -Superconductivity

II
h̄2
Z
d r ∗ (∇ψ(~r)) . (∇δψ ∗ )ψ=ψ̄
3

P
2m
2
Z
h̄ Z
h̄2
= d3 r ∗ ∇.(δψ ∗ ∇ψ) − d3 r ∗ δψ ∗ (∇2 ψ) (14)
2m 2m

CM
Z
h̄2
= − d3 r ∗ δψ ∗ (∇2 ψ)
2m
The first temr of RHS can be changed to surface integral and it
vanishes. Therefore we get;

h̄2
PA
− (∇2 ψ̄) + aψ̄ + b|ψ̄|2 ψ̄ = 0 (15)
2m∗
is some called non-linear Schrödinger Equation.
12
 GL Theory -Superconductivity

II
• The boundary conditions on ψ is very different from that of the

P
usual SE, however, and may understood as guaranteering that
ĵ.n̂ vanishes at the surface of the sample.

CM
• The boundary conditions on these equations are found from the
condition that the surface integrals in the variation of δF are
zero. n̂ is normal vector at the surface of the body. As a result
of this condition, the normal component of the current is also
zero (see next lecture...) as it should be ĵ.n̂ = 0.
PA

13
 GL Theory -Superconductivity

II
• With this B.C. ψ it self is not zero, as the wave function
apparently ought to be at the boundary of the body. This is

P
because ψ actually falls to zero only at the distances ξ0 from the
surface, but such distances are regarded as negligible in the GL
theory. n̂.(−ih̄∇ψ) = 0 has been derived here for a boundary

CM
with an insulator, but it is not correct for an interface between
different metals (one SC, another normal), since it does not take
into account the partial penetration of superconducting electrons
into the normal metal. In this case, above BC is replaced by a
more general condition compatible with ĵ.n̂ = 0 and
n̂.∇ψ = iψ/λ where λ is real constant (with the dimensions of
PA
length); however, an estimate of this constant would need a
more detailed microscopic investigation.

14
 GL Theory -Superconductivity

II
Case II:
~ 6= 0, ψ nonunif orm (e.g. in the interf ace/External H
A ~ =
6 0)
Then we have;

P

Z
b
r)ψ ∗ (~r) + ψ(~r)ψ ∗ (~r)ψ(~r)ψ ∗ (~r)

CM
∆F = d3 r 
aψ(~
2
h̄2 iq ∗ ~ iq ∗ ~
   
+ ∗
∇ − A(~r ) ψ(~
r ) ∇ − A(~r) ψ ∗ (~r)
2m h̄c h̄c (16)
 2 
~
∇ × A(~r) 
+
8π

PA
Z h i
≡ ~ r)
d3 rf ψ(~r), ψ ∗ (~r), A(~

15
 GL Theory -Superconductivity

II
~ r) fixed.
• 1. First vary ψ(~r) OR ψ ∗ (~r) keeping A(~

• 2. Then vary A(~~ r) i.e. A(~

~ r) → A(~~ r) + δ A(~
~ r) keeping

P
ψ(~r) and ψ ∗ (~r) fixed.
• Use the BC;

CM
iq ∗ ~
Z  
n̂. ∇ − A(~r) ψ(~r) dS = 0 (17)
h̄c surf ace
and Z  
∇. δ A ~×∇×A ~ d3~r = 0 (18)
Further use;
Z     Z  
~ . ∇ × δA
~ d3 r = − ~ .δ Ad
~ 3 r (19)
PA
∇×A ∇×∇×A
After algebra (for details see my lecture notes a few years back
.... OR get photocopy from me)
16
 GL Theory -Superconductivity

II
h̄2 iq ∗ ~
 
2
aψ(~r) + b|ψ(~r)| ψ(~r) − (∇ − A(~r))ψ(~r) = 0
2m∗ h̄c

P
!2 (20)
2π
OR ξ 2 i∇ + A ψ − ψ + ψ|ψ|2 = 0
φ0

CM
h̄
which is called Ginzburg-Landau first equation. Here ξ = √2am ∗ is
hc
the ”Coherence Length”. It is intrinsic coherence length. φ0 = 2e is
the ”Fluxiod Quantization”.
HW: Compare it with Aharonov- Bohm effect you studied in
QM
2
~ = −i φ0 (ψ ∗ ∇ψ − ψ∇ψ ∗ ) − |ψ| A
PA
∇ × ∇A ~ (21)
4πλ2 λ2
is called Ginzburg-Landau second equation. These equations
introduce two length scales: (i) Coherence length (ξ) and (ii)
Penetration depth (λ.) 17
 GL Theory -Superconductivity

II
P
Figure: λ =

CM
(a) Magnetic field penetration length (λ)

bm∗c2
−4πaq ∗2
1/2
(b) Coherence length &
λ

and ξ in a superconductor
PA
Ratio κ = λξ defines whether the given superconductor is a Type I OR
it is Type II.
18
 GL Theory -Superconductivity

II
P
CM
(a) Type I Superconductor

Figure
(b) Type II Superconductor
PA

19
 GL Theory: Consequences

II
GL theory is phenomenological theory still it plays crucial role to
understand the basic properties of a superconductor. There are many

P
consequences i.e. it can explain however here in this lecture we just
consider a few of them. for practical applications of a Supercondutor

CM
GL theory is more important than microscopic one.
Coherence length
In zero field GL first equation becomes

2 h̄2 2
aψ + b|ψ| ψ − ∇ ψ=0 (22)
2m∗
PA
Linearzing we get

20
 GL Theory: Consequences...

II
h̄2 2

P
− ∇ ψ = −aψ
2m∗
a2m (23)
OR ∇2 ψ = ψ

CM
h̄2
=⇒ ψ ≡ exp (ix/ξ)
h̄
where ξ = √2am ∗ is the intrinsic coherence length. Considering
nonlinear term also we have;

h̄2 2
PA
aψ + b|ψ|2 ψ − ∇ ψ=0 (24)
2m∗

21
 GL Theory: Consequences...

II
P
Figure: N-SC System. In the left there is Normal metal and right there is
a Superconductor. In Normal metal ψ = 0 [Link]

CM
ψ(x = 0) = ψ0 = 0 and ψ(x = ∞) = ψ∞ = −a b is the solution for
superconducting state.

The solution of 24 becomes;

!
a x
r
ψ= − tanh √ (25)
PA
b 2ξ

HW: Verify that above solution satisfies 24.

22
 GL Theory: Consequences...

II
Penetration depth for A(~ ~ r) variations

P
Consider ψ = |ψ0 |exp (iθ(~r)) and A ≡ A(~~ r). Then we get after some
calculations

CM
1 4πq ∗2 (−a)
= (26)
λ2 c2 m ∗ b
OR v
u c2 m∗ b
u
λ=t (27)
4πq ∗2 (−a)
PA

23
 GL Theory: Consequences...

II
Similarly it explains Type I & Type II superconductors, Cirtical

P
magnetic field, critical current density. Also interesting phenomena is
the negative surface energy, Fluxiod quantization and Probably the

CM
most interesting is ”GOLDSTONE MODES” - Anderson-Higgs
Mechanism ... Higgs Bosons ... All these stories started from
Superconductivity. For Anderson-Higgs Mechanism we need to
consider Fluctuations in GL theory. Another phenomena more
interesting could be ”Abrikosov Vortex Lattice in Type II
superconductors (I just give idea of this intriguing phenomena).
Superconductivity could be a course of 4 CH itself .....
PA

24
 GL Theory: Consequences...

II
P
CM
(a) Abrikosov lattice

Figure
(b) Abrikosov
lattice in YBCO
PA

25
 GL Theory: Consequences...

II
P
CM
(a) Triangular is cheaper or
favorable than rectangular
(b) Vortex antivortex pair in a
supercoducting thin films
Homework: Can you think similarities between XY model and
superconducting thin films? Its astonishing to see similarities
PA
between such different systems.... Thats why this is physics
and interesting!!! So devote our whole life to understand
these mystries....
26
 BCS Theory

II
P
CM
Figure: Nobel Prize for Physics in 1972: The theory is one of the most
important scientific achievements of 20th century. It explains the so called
PA
”conventional superconductors”

27
 BCS Theory

II
• Limitations of Phenomenological theory
• GL theory in many respect is quite complete. However, certain

P
questions they canot answer
• Why are some materials superconducting and others not? What

CM
determines Tc and Hc ?

• What sets ξ and λ? These are the questions to be answered by

microscopic theory. Because the single electron Hamiltonians
that are amenable to exact or numerical solution do not produce
superconductivity and because the fully many electron
PA
Hamiltonian is intractable, decades passed after the experimental
discovery of superconductivity with little theoretical progress.

28
 BCS Theory...

II
P
• ”Bloch in a famous theorem later extended by Bohm to many
body systems, showed that in the absence of a magnetic field

CM
the most stable state of an electron system is that of zero
current. Because of the frustrations which the many theorists
who worked on the problem encountered, Bloch jokingly
proposed a second theorem -that ”Any theory of
superconductivity can be refuted - Bardeen (1963)””.
PA

29
 BCS Theory-Crucial steps to path to
BCS

II
• Discovery of the isotope effects (E. Maxwell 1950 & Reynolds)
The idea: If Tc could be depressed by changing the mass of the

P
ions, then the interaction of electron and ions must be crucial.

CM
PA

Figure: Isotope effects in [Link] Tc varies with reciprocal of

square root of mercury.
30
 Crucial steps to path to BCS

II
• Actually, Frölich began a theory of superconductivity based upon

P
an interaction between electrons and phonons prior to the
experimental observations shown in above figure.

CM
• Energy/Length Scale: The energy per electron involved in the
SC transition can be estimated from Thermodynamical critical
field:
−Hc2 (T )
fs − fn = −
8π
−22
also kB Tc ' 10 J f or 1K. ∴ energy ' 10−22 J i.e. mev
PA

31
 Crucial steps to path to BCS

II
• And Hc ' 1000 Oersted;
2
∴ −H8π c (T )
' 105 erg/cm3 .
n ' 1022 cm−3 .

P
Therefore the energy responsible for superconductivity is
105 −17
' 10 22 ' 10 erg/electron ' 10−5 eV /electron

CM
• Electronic Coulomb energies are ' 1eV.
Therefore the energy responsible for SC can be neglected in
many applications of quantum theory of metals without affecting
the results.
• Phonon energies are smaller ' meV. Could it be that? Also
Tc ' M −α depends on ionic mass for some elements at least. So
PA
again, may be some phonon related interactions. It is clear that
the lattice of ions in a metal is an active participants in creating
the Superconducting state.
32
 Crucial steps to path to BCS

II
P
CM
Figure: Phase diagram of Helium. Normal to Superfluid for T < TC
PA
He4 is a boson. could the fermion somehow stick together to give a
spin-integer ’boson’ - that could Bose Condensate?
33
 Crucial steps to path to BCS

II
• So electron pairing? If pairing ↑↑ OR ↑↓?

P
• In fact, flux quantization confirms that the basic
superconducting object is 2e. But what is the size or length scale

CM
involved in SC pair?

• Coherence? ξ(T ) ' ξ(0)||−1/2 ? Pippard from uncertainty

argument estimated ξ(0).
Only electrons ∼ kB Tc around the Fermi surface can participate,
in a momentum range
∆p ∼ kBvFTc =⇒ ξ0 ∼ kh̄v F
.
PA
B Tc

34
 Cooper pair and instability of Fermi Sea

II
• Cooper recognized that the ground state (T= 0 K) of an e gas
is unstable if one adds a weak attractive interaction between

P
each pair of electrons. Such an interaction had been obtained by
Frölich in the form of phonon-mediated interaction.

CM
• When an e passes through solid, due to negative charge leaves
behind a deformation trail affecting the ion cores. This trail is
associated with an increased density of positive charge due to
the ion cores and thus has an attractive effect on a second
electron. The lattice deformation therefore causes a weak
PA
attraction between pairs of electrons.

35
 Cooper pair and instability of Fermi Sea

II
P
CM
PA

36
 Cooper pair and instability of Fermi Sea

II
P
CM
PA
(a) Deformation of lattice (b) Displacement of ion
cores as a function of
distance from first
37
electron
 Cooper pair and instability of Fermi Sea

II
• Thi attractive e-e interaction is retarted because of the slow
motion of the ions in comparison with the almost instantaneous

P
Coulomb repulsion between electrons, at the instant when an e
passes, the ions receive a pull which, only after the e has passed,

CM
leads to a displacement and thereby to a polarization of the
lattice. The lattice deformation reaches its maximum at a
distance from the first e which can be estimated from the e
velocity (vf ∼ 108 cm/s) and the phonon vibrational period
( ω2πD ' 10−13 s). The two e corrlated by lattice deformation have
an approximate separation of about ' vF . ω2πD ' 108 .10−13 cm ∼
PA
10−5 cm ' 1000Å. ∴ size of Cooper pair ' ξ ' 1000Å.

38
 Cooper pair and instability of Fermi Sea

II
• The extremely long range of interactions of the two electrons
correlated by a lattice deformation explains why the Coulomb

P
repulsion is insignificant; it is completely screened out over
distances of just a few Angstroms.

CM
• Quantum mechanically, the lattice deformation can be
understood as the superposition of phonons which the e, due to
its interaction with the lattice, continuously emits and absorbs.
to comply with energy conservation, the phonon constituting the
lattice deformation may only exist for a time τ ' 2π ω
deformed
by the uncertainty relations; thereafter they must be absorbed.
PA
One thus says ”Virtual Phonons”.

39
 Cooper pair model

II
P
CM
PA
Figure: Upto K = KF completely occupied. Add just two electrons
(11, E(11) and 12, (E(k2))) in the states just above EF (by necessity.)

40
 Cooper pair model

II
• These two electrons do not interact with the inert Fermi-Sea.
They do however interact with each other, but in a very peculiar

P
manner. The interaction is such that the two electrons attract if
they both are within a small, energy (Ed ) from the Fermi

CM
surface, otherwise they do not interact at all. due to phonon
exchange the two additional electrons continually change their
wave vector, however, momentum must be conserved.
~k1 + ~k2 = k~0 1 + k~0 2 = K
~ (28)

As the interactions in k space is restricted to a shell with an

PA
energy thickness of h̄ωD above EF the possible ~k states are
given by the shaded area in figure (see next figure)

41
 Cooper pair model

II
P
~
CM
Figure: Two spherical shells with Fermi radius KF and thickness ∆k
describe the pair of wave vectors ~k1 and ~k2 . All pairs for which
~k1 + ~k2 = K
~ end in the shaded volume (rotationally symmetric about K).
~
~
The number of pairs k1 , k2 is proportional to this volume and is maximum
for K = 0. Therefore it is sufficient to consider the case in which
PA
~k1 = −~k2 = K(say).
~

42
 Cooper pair model

II
The Schrödinger equation for the pair of electrons;
h̄2  2
!

− ∇1 + ∇22 + V (~r1 − ~r2 ) ψ(~r1 , ~r2 )

P
2m (29)
= Eψ(~r1 , ~r2 )
= ( + 2EF ) ψ(~r1 , ~r2 )

CM
Here  is the energy of the electron pair relative to interaction free
state (V=0) and 2EF is as each electron is just above Fermi level.
1
∴ ψ(~r1 , ~r2 ) = 3 exp (i~k.(~r1 − ~r2 )) (30)
L
The two electrons have opposite spins since ψ is symmetric w.r.t.
PA
~r1 , ~r2 . Hence,
~ ↔ −K
K ~
. It is because the space part is symmetric from the form of the wave
functions above. This means spin part must be antisymmetric.
43
 Cooper pair model

II
~ ↑, −K
=⇒ Pairing is (K ~ ↓) (31)
The most general representation of a two-particle state for (V 6= 0) is

P
given by the series:
1 X ~

CM
ψ(~r1 , ~r2 ) = g(k) exp (i~k.(~r1 − ~r2 )) (32)
L3 ~
k

The summation is confined to pairs with ~k1 = −~k2 = K

~ such that

h̄2
EF ≤ < (EF + h̄ωD ) (33)
2m
PA
The interaction is assumed to be in a shell of thickness 2h̄ωD ,
positioned symmetrically about EF . for the case of one extra Cooper
pair considered here only the k-space wirh E > EF is of interest.
44
 Cooper pair model

II
˜
g(~k) = Prob. of finding one e at k̃ & another at −k (34)
This means

P
2
g(~k) = 0{for k < kF Pauli principle}

CM
s
2m (35)
= 0{for k > (EF + h̄ωD ) due to energy}
h̄2

substituting 32 in 29, multiplying by (exp −ik~0 .~r) and integrating

over the normalization volume yields;
PA
h̄2 k 2 1 X ~0
g(~k) + 3 g(k )V~k~k0 = ( + 2EF ) g(~k) (36)
m L ~0
k

45
 Cooper pair model

II
The interaction matrix element
Z
V~k~k0 = V (~r) exp (i(~k − ~k 0 ).~r)d3 r (37)

P
describes the scattering of the e pair from (~k, −~k) to (~k 0 , −~k 0 ) and

CM
vice versa. in the simplest model, V~k~k0 is assumed to be independent
of ~k and attractive i.e.; V~k~k0 < 0 :
0
h̄2 k 2 h̄2 k 2
! !
V~k~k0 = −V0 V0 > 0 f or EF < , < (EF + h̄ωD )
2m 2m
= 0 otherwise
PA
(38)

46
 Cooper pair model

II
From eq. 36;
h̄2 k 2
!
− +  + 2EF g(k) = −A (39)

P
2m
where
V0 X ~ 0

CM
A= g(k ) (40)
L3 ~ 0
k

is independent of ~k. summing 39 over ~k and comparison to 40, we

get (from consistency) that;

V0 X 1
1= (41)
PA
 
L3 ~ − + h̄2 k2
k m
− 2EF

47
 Cooper pair model

II
X L3 Z Z dSE
→ dE (42)
~k
(2π)3 ∇~k E~k

P
Therefore from 41 we get

V0 L3 Z Z dSE

CM
1
1 = 3. 3
 2 2
 dE
L (2π) h̄ k
∇~k E~k − + m − 2EF
(43)
V0 Z Z dSE 1
= dE
(2π)3
∇~k E~k (− + 2E − 2EF )
h̄2
where E = 2m
PA
Further
1 Z dSE
D(E) =
(2π)3 ∇~k E~k

48
 Cooper pair model

II
Approximation: We use:
Z Z EF +h̄ωD
dE → dE = a constant (44)

P
EF

Further

CM
 
1 Z
dSE 
D(E)dE =  dE (45)
(2π)3 E(k)=constant ∇~k E~k

We represent  
1 Z dSE 
N (0) =
PA
(2π)3 ∇~k E~k
E=EF

as DoS at the Fermi level for one spin type.

49
 Cooper pair model

II
Z EF +h̄ωD
dE
∴ 1 = V0 N (0)
EF (− + 2E − 2EF )

P
Z EF +h̄ωD
d1
= V0 N (0)
EF (21 − ) (46)

CM
V0 N (0) EF +h̄ωD
= ln (21 − )|E
2 "
F
#
V0 N (0) 2EF + 2h̄ωD − 
= ln
2 2EF − 

We use 1 = E − EF in above integration. With ∆ = 2EF −  is the

PA
energy difference between the states of two non-interacting particles
on the Fermi surface and the exact energy eigen value . Therefore
above equation can be rewritten as:

50
 Cooper pair model

II
P
" #
V0 N (0) 2h̄ωD
1= ln 1 +
2 ∆

CM
" # !
2h̄ωD 2
OR 1 + = exp
∆ V0 N (0)
! (47)
2h̄ωD 2
OR = exp −1
∆ V0 N (0)
!
2
' exp
V0 N (0)
PA

51
 Cooper pair model

II
!
∆ 2
∴ = exp −

P
2h̄ωD V0 N (0)
!
2
OR 2EF −  = 2h̄ωD exp −

CM
V0 N (0)
! (48)
2
∴  = 2EF − 2h̄ωD exp −
V0 N (0)
!
2
OR  − 2EF = −2h̄ωD exp −
V0 N (0)
PA
This means a lot!!!!! This is one of the most important
results for superconductivity!!! Now you see ”bloch’s second
theorem” what he was jokingly proposing...
52
 Cooper pair model

II
• Therefore there exists a two-electron bound state, whose energy
is lower than that of the fully occupied Fermi sea (T= 0 K) by
an amount

P
!
2
−2h̄ωD exp − ≡  < 0.

CM
V0 N (0)
• This shows that the g.s. of the non-interacting free e gas
becomes unstable when a minute attractive interaction between
electrons is ”switched on”. It should be noted that the energy
reduction  results from a ”GEDANKENEXPERIMENT”
(THOUGHT EXPERIMENT) in which the Fermi sea for states
PA
2 2
with h̄2mk < EF is assumed to be fixed and only the effects of
the attraction between two additional electrons in the presence
of the Fermi sea is treated.
53
 Cooper pair model

II
• In reality, the instability leads to the formation of a high density
of such e pairs, so called Cooper Pairs (~k, −k)
~ via which the

P
systems tries to achieve a new lower-energy ground state. This

CM
new g.s. is identical to the superconducting phase.

• Since ∆ has an essential singularity as (N (0)V0 ) → 0. This

means that the above result for ∆ could never have been
obtained to any finite order in perturbation theory in (N (0)V0 ).
Hence, an arbitrary weak interaction is a singular perturbation to
a free e gas. It destabilizes a Fermi-liquid.
PA

54
 Cooper pair model

II
• Secondly, we see that (N (0)V0 > 0) only if N (0) > 0. for a
2 2 √
parabolic band (~k) = h̄2mk , in 3D we have N () ∼ . Hence, if

P
the Fermi sea vanishes by draining electrons out of the system,
(N (0)V0 → 0) and ∆ → 0. In other words, the tendency to

CM
pairing of electrons that we see above vanishes when the Fermi
sea vanishes. We should not expect a two-electron bound state
for arbitrarily weak attraction in vacuum.
  
• Thirdly ∆ ∼ 2h̄ωD exp − V N2 (0) shows that the pairing is
0
also a quantum effect, since the bound state energy ∆ vanishes
PA
when h̄ → 0 i.e. it vanishes in the classical limit.

55
 Cooper pair model

II
• Lastly, the two e bound state resides at an energy which
apparently is prohibited by the Pauli principle, in a vague way

P
hints at the fact that perhaps bound pairs of electrons are not as
much slaves to the Pauli principle as are individual electrons. In

CM
fact this is precisely true. By forming Cooper pairs, electrons
shed their fermion behaviour. The resulting quantum fluid of
Cooper pairs is not as protected from interactions as are the
Pauli prnciple loyal Fermionic fluids. This is basically the root
cause of why attractive interactions are able to destabilize the
Fermi-liquid while repulsive interactions are virtually incapable of
PA
doing so in metallic states.

56
 Cooper pair model

II
• In the microscopic foundations of Landau-Fermi liquid theory (to

P
be discussed after SC lectures), it is stated that the Pauli
principle sevrely suppressed the effects of repulsive interactions.

CM
For attractive interactions, the Pauli principle is essentially
sidelined and no longer acts as a suppressor of interaction
effects. This gives the possibility of g.s. qualitatively different
from the free e gas.

• Now we can discuss the BCS ground state.

PA

57
 The BCS Ground State

II
• Due to energy reduction of a pair ever more Cooper pairs are
formed. The new g.s. of the fermi sea after pair formation is

P
achieved through a complicated interaction between the
electrons. The total energy reduction is not obtained by simply

CM
summing the individual pair contributions. Since the excitation
above EF is necessary, the pairing is associated with an increase
in kinetic energy (KE).

• The kinetic component can be given immediately; if ωk ≡ vk2 is

the probability that the pair state (~k ↑, −k
~ ↓) is occupied, the
PA
kinetic component Ekin is;

58
 The BCS Ground State

II
P
CM
PA

59
 The BCS Ground State

II
vk2 (~k)
X
Ekin = 2 (49)

P
~k

where

CM
~ ~ h̄2 k 2 h̄2 kF2
(k) = E(k) − EF = − (50)
2m 2m
The energy reduction can be calculated by considering the
Hamiltonian (H) which explicitly takes account of the fact that
’annihilation’ of a pair (~k ↑, −k
~ ↓) and the simultaneous creation of a
0
pair (~k 0 ↑, −k
~ ↓) i.e. scattering from (~k ↑, −k ~ 0 ↓)
~ ↓) to (~k 0 ↑, −k
PA
leads to an energy reduction of V~k~k0 .

60
 The BCS Ground State

II
• Since a pair state ~k can be either occupied or unoccupied, we
choose a representation consisting of two orthogonal states
|1i~k and |0i~k where |1i~k is the state in which (~k ↑, −k
~ ↓) is

P
occupied and |0i~k is the corresponding unoccupied state. The
most general state of the pair (~k ↑, −k~ ↓) is thus given by;

CM|ψi~k = u~k |0i~k + v~k |1i~k

This is Cooper pair wave function.
(51)

• In the representation 51, the g.s. of the many body system of all
PA
the Cooper pairs can be approximated by the product of the
state vectors of the single pairs:

61
 The BCS Ground State

II
 
|ΦBCS i ' Π~k u~k |0i~k + v~k |1i~k (52)
with u~k and v~k both real. With 52 the many-body state is

P
non-interacting pairs or interactions between the pairs are neglected
in the state vector

CM
( have you seen many body wave function product of individual?
what does it mean? Hartree-Fock? ). In 2D representation;
! !
1 0
|1i~k = and |0i~k = (53)
0 1
One can use the Pauli matrices;
! !
PA
(1) 0 1 (2) 0 −i
σ~k = ; σ~k = (54)
1 0 ~k i 0 ~k

to describe the ”creation” and ”annihilation” of a Cooper pair.

Similarly the operator
62
 The BCS Ground State

II
1  (1) 
σ~k† =
(2)
σ~k + iσ~k (55)
2

P
transforms the unoccupied state |0i~k into the occupied state |1i~k
while

CM
1  (1) (2)

σ~k− =
σ~k − iσ~k (56)
2
transforms the occupied state |1i~k into the unoccupied state |0i~k .
HW Prove
! !
1 0
PA
σ~k† |0i~k = ≡ |1i~k and σ~k− |1i~k = ≡ |0i~k (57)
0 ~k 1 ~k

63
 The BCS Ground State

II
From above representation of operators one ca deduce the following
properties:

P
σ~k† |1i~k = 0; σ~k† |0i~k = |1i~k ; σ~k− |1i~k = |0i~k ; σ~k− |0i~k = 0 (58)

CM
Scattering from (~k ↑, −k ~ 0 ↓) is associated with an
~ ↓) to (~k 0 ↑, −k
energy reduction by an amount of V~k~k0 . In BCS model, V~k~k0 is
independent of ~k and ~k 0 . We consider unit volume.
∴ L3 = volume = one unit volume.

Therefore the operator that describes the corresponding energy

PA
reduction is
(−V0 σ~k† 0 σ~k− ).

64
 The BCS Ground State

II
Therefore, the total energy reduction due to pair collisions ~k → ~k 0
and ~k 0 → ~k is given by summing over all collisions and may be
expressed in operator terminology as;

P
X †
H = (−V0 σ~k0 σ~k− ) (59)
~k,~k0

CM
V0 is restricted to ±h̄ωD around EF hence the sum over ~k, ~k 0 also
includes only pair states in this shell. Therefore, energy reduction
=Expectation value of H in the BCS state
energy reduction
= hΦBCS |H|ΦBCS i
PA
(60)
   

−V0 πp~ (up~ hp~ 0|+vp~ hp~ 1|)  σ~k† 0 σ~k− πq~ (uq~|0i + vq~|i)
X
=
   
~k,~k0

65
 The BCS Ground State

II
• In evaluating 60, one should note that the operator σ~† (σ~− ) acts
k k
only on the state |1i~k (|0i~k ). The explicit rules are given in 58.
Further we have orthonromality relations;

P
h~k 1|1i~k = h~k 0|0i~k = 1; h~k 1|0i~k = 0 (61)

CM
• From 60, after a bit calculations (HW);
X
hΦBCS |H|ΦBCS i = −V0 v~k u~k0 v~k0 u~k (62)
~k,~k0

This is energy due to interactions (PE). Therefore, the total

energy of the system of Cooper pairs can be represented as:
PA
v~k2 (~k)−V0
X X
EBCS = Ekin +hΦBCS |H|ΦBCS i = 2 v~k u~k0 v~k0 u~k
~k ~k,~k0
(63)
66
 The BCS Ground State

II
• We should find the identity of the function v~2 such that he total
k
energy EBCS is a minimum.

P
CM
PA
Figure: v~k2 versus electron energy compare Fermi distribution at T= Tc
and this v~k2 at T= 0 K.

67
 The BCS Ground State

II
∂EBCS
=⇒ =0 (64)
∂v~k2

P
From equations 63 and 64 we get;
1 − 2v~k2 X

CM
2~k − V0 u~ 0 v~ 0 = 0
u~k v~k ~ 0 k k
k
u~k v~k ∆0 (65)
=⇒ 2
=
1 − 2v~k 2~k
X
where ∆0 = V0 u~k v~k
~k
PA
From above eq 65 we get
∆20 (1 − 4v~k2 + 4v~k4 ) = 4~2k u~2k v~k2 (66)

68
 The BCS Ground State

II
OR
∆20

P
∆20 (v~k4 − v~k2 ) + − ~2k u~2k v~k2 = 0
4
∆2

CM
OR v~k4 − v~k2 + 02 = 0 (67)
4E~k
!
1 ~
where E~k2 = ~2k + ∆20 =⇒ v~k2 = 1− k
2 E~k

The negative sign in last expression of above equation stems from a

general argument that as ~k → 0 we ought to have v~k2 → 1, while,
PA
~k → −F .

69
 The BCS - Ground State Energy

II
We have
X
∆0 = V0 u~k v~k
~k

P
" ! !#1/2
X 1 ~ 1 ~
= V0 1− k 1+ k

CM
~k
2 E~k 2 E~k
#1/2
V0 X E~k2 − ~2k
"
= (68)
2 ~ E~k2
k
V0 Xh i−1/2
= ∆0 ~2k + ∆20
2 ~ k
PA
V0 X h 2 i−1/2
OR 1 =  ~k + ∆20
2 ~k

70
 The BCS - Ground State Energy

II
R h̄ωD
→
P
Now, ~k −h̄ωD N ()d.
V0 Z h̄ωD h i−1/2
∴1= N (~k )d~k ~2k + ∆20
2 −h̄ωD

P
Z h̄ωD h i−1/2
= V0 N (0) ~2k + ∆20
0

CM
(∵ N (~k ) → N (EF + h̄ωD ) ' N (EF ) ' N (0)) (69)
!
1 h̄ωD
=⇒ = sinh−1
V0 N (0) ∆0
!
h̄ωD 1
OR = sinh
∆0 N (0)V0
PA
Recalling that for the majority of superconductors (N (0)V0 ) ≤ 0.3 we
obtain !
1
∆0 ' 2h̄ωD exp − (70)
N (0)V0
71
BCS - Calculation of Ground State Energy

II
P
CM
If h̄ωD ' 100 K; N (0)V0 ' 0.3 =⇒ ∆0 ∼ 4K
PA

72
BCS - Calculation of Ground State Energy

II
The general expression for energy of a superconductor is given by;
2~k~k 2 +
X X
Es = V~k~k0 u~k v~k u~k0 v~k0 (71)
~k ~k,~k0

P
In the normal state when the interaction between the electrons is

CM
switched off and all states below the Fermi level are full, the ground
state energy is given by;
X
En = 2~k (72)
~k<~kF

In above equation 2 is due to the sum is taken over pairs of state

(~k, −~k). The g.s. energy of the superconductor is measured from the
PA
g.s. energy of the normal metal, we seek the quantity;
Esn = Es − En

73
BCS - Calculation of Ground State Energy

II
2~k~k 2 − V0
X X X
Esn = u~k v~k u~k0 v~k0 − 2~k

P
~k ~k,~k0 ~k<~kF
X X X (73)
= (2~k − 1) + 2~k2 − V0 u~k v~k u~k0 v~k0

CM
~k<~kF ~k>~kF ~k,~k0

We have; !
1 ~
v~k2 = 1− k
2 E~k
and
h̄2 k 2
PA
~k =− EF
2m
For k < KF ; ~k < 0. Therefore for k < kF ;

74
BCS - Calculation of Ground State Energy

II
!
1 |~ |
vk2
= 1+ k

P
2 E~k
!
2 |~k |
∴ 2v~k ~k = 2.(1/2).(−)|~k | 1 +

CM
E~
" ! k #
1 |~ |
2~k (v~k2 − 1) = −|~k | 1+ k −1 (74)
2 E~k
" #
1 1 |~k |
= −2 − |~k | − +
2 2 E~k
!
PA
|~k |
= |~k | 1 −
E~k

75
BCS - Calculation of Ground State Energy

II
! !
X |~ | X ~ X

P
∴ Esn = |~k | 1 − k + ~k 1 − k − V0 u~k v~k u~k0 v~k0
~k<~kF
E~k ~ ~
E~k ~k,~k0
k>kF
!

CM
X ~ X
=2 ~k 1− k − V0 u~k v~k u~k0 v~k0
~k>~kF
E~k ~k,~k0
(75)

Now,
∆20 X
= u~ v~ u~ 0 v~ 0
V02 ~ ~ 0 k k k k
PA
k,k

Therefore,

76
BCS - Calculation of Ground State Energy

II
!
X ~ ∆20
∴ Esn = 2 ~k 1− k −
~k>~kF
E~k V0

P
Z h̄ωD 
 ∆2

= 2N (0)  1− d − 20
E V0

CM
0
 
Z h̄ωD
 ∆20
= 2N (0)  1 − q  d −
0 2 + ∆20 V02
 
!2  !2 1/2 !
h̄ωD h̄ωD  h̄ωD h̄ωD 
= N (0)∆20  − 1+  + sinh−1
∆0 ∆0 ∆0 ∆0
 
PA
∆20
−
V0
(76)
77
BCS - Calculation of Ground State Energy

II
Since !
−1 h̄ωD 1
sinh = and h̄ωD ≫ ∆0
∆0 N (0)V0

P
we get
 

CM
!2 !2  !2 1/2
h̄ωD h̄ωD ∆0 1 
Esn = N (0)∆20  − 1 +  +
∆0 ∆0 h̄ωD N (0)V0
 

∆20
−
V0
 !2 !2  !2  
h̄ωD h̄ωD 1 ∆0 1 
= N (0)∆20 − 1 + +
PA

∆0 ∆0 2 h̄ωD N (0)V0
∆20
−
V0
78 (77)
BCS - Calculation of Ground State Energy

II
N (0)∆20
∴ Esn = Es − En = −
2

P
• This shows that the superconducting state is more stable at
T < Tc . Now from thermodynamics of a superconductor

CM
Hc2 (0) N (0)∆20 q
= En − Es = =⇒ Hc (0) = ∆0 4πN (0)
8π 2
• Estimation of Hc (0) in a typical SC 1 cm3 of metal ∼ 1022
electrons. width of conduction band ∼ 10 eV = 10X1.6X10−12
1022 33 3
erg ∴ N (0) ' 10X1.6X10 −12 ' 10 /erg − cm
PA
This gives ∆0 ' 10K√' 10−15 erg
∴ Hc (0) ' 10−15 erg( 4πX1033 erg −1 cm−3 ) ' 100 Oersted is
the reasonable value for thermodynamic critical field.
79
Elementary excitation spectrum of SC

II
• The Energy Gap: Lets us focus on an arbitrary pair of states
(~k, −~k) in a SC in momentum space. First find the contributions
of this pair, w~k , to the total energy of the SC:

P
w~k = 2~k v~k2 − 2V u~k v~k
X
u~k0 v~k0 (78)

CM
~k0

first term - KE of the pair and second term is the pairs

contributions to the negative part of the g.s. energy and it arises
from various interactions in which the pair participates. The pair
can go to any other state (~p, −~p) or conversely any other pair
~ −K).
(~p, −~p) can go to the state (K, ~ The coefficient 2 in the
PA
2nd term takes into account that the pair (~k, −~k) is encountered
twice, first when the sum runs over ~k and the other when it runs
over −~k.
80
Elementary excitation spectrum of SC

II
We have
X
∆0 = V u~k0 v~k0
~k0

P
q
E~k = ~k2 + ∆20 (79)
!
1 ~

CM
v~k2 = 1− k
2 E~k
From Eq 78 and 79 we get
! " ! !#1/2
1 ~ 1 ~ 1 ~
w~k = 2~k 1− k − 2∆0 1− k 1+ k
2 E~k 2 E~k 2 E~k
 !2 1/2
(80)
!
PA
~ ~k
= ~k 1− k − ∆0 1 − 
E~k E~k
= ~k − E~k
81
Elementary excitation spectrum of SC

II
• Consider a pair state (~k ↑, −~k ↓) in the g.s. of a SC. Let the
state be empty. how will the energy of the system change if we
add one more electron (e) to it from outside and place it in the

P
state ~k? Since we have only one e, the pair (~k ↑, −~k ↓) cannot
take part in the scattering events i.e. it cannot contribute to the

CM
g.s. energy of the SC. Its potential contributions is known to w~k .

• Hence the energy of the SC with one extra e in the state ~k will
be
W~k = Esn − w~k + ~k (81)
PA
Tis extra uncoupled e is referred as an elementary excitation of
our system or quasiparticle. Esn is g.s. of SC and ~k is the KE
of extra e.
82
Elementary excitation spectrum of SC

II
from above 2 equations we get

P
W~k = Esn − ~k + E~k + ~k = Esn + E~k (82)

CM
q
Further E~k = ~2k + ∆20 . By adding one extra e to a SC in the g.s.
we increase the energy of the system by at least the value of ∆0 (the
minimum increase corresponds, ~k = 0, ~k = ~kF ). Therefore the
spectrum of elementary excitations of the SC is separated from the
g.s. enenergy level by an energy gap.
PA

83
Elementary excitation spectrum of SC

II
P
CM
PA

84
Elementary excitation spectrum of SC

II
P
The energy gap separates the energy levels of elementary excitations
from the g.s. level. Therefore, if a pair (~q ↑, −~q ↓) breaks it must go

CM
to ~k, p~ or elementary excitation levels which requires an energy at
least 2∆0 . The existence of this energy gap is a very important
property of the SC and explains many aspects of its behavior.
PA

85
DoS of Elementary excitations in SC and
Coherence length -ξ

II
P
CM
PA

86
DoS of Elementary excitations in SC and
Coherence length -ξ

II
P
CM
PA

87
DoS of Elementary excitations in SC and
Coherence length -ξ

II
r 
2 2
q
We have, E~k = ~2k+ ∆20
= h̄ k
2m
− h̄2 kF2 2m ∆20 As E~k → ∆0 ,
the energy levels of elementary excitations become denser.

P
dN
D(E) =

CM
dE
=no. of energy levels in the energy interval dE in the vicinity of the
level E. Letting dN |
dE E=EF
≡ N (0) for normal metal= dNd
Therefore

dN d E
D(E) = = N (0) q (83)
d dE E 2 − ∆20
PA
Therefore at E~k → ∆0 , the DoS of elementary excitations in the SC
is D(E) → ∞ as shown in the figure.
88
DoS of Elementary excitations in SC and
Coherence length -ξ

II
P
CM
The g.s. of a SC can be represented by the distribution of e pairs in
momentum space given by the function v~k2 .
PA

89
Persistent current - BCS theory

II
P
CM
PA

90
Persistent current - BCS theory

II
P
CM
PA

91
DoS of Elementary excitations in SC and
Coherence length -ξ

II
The large variations of v~k2 can occur only within the region around kF
and ∆k ' kF 2∆ 0
. Therefore in the real space large variations of the

P
EF
wavefunction is in the region ∆x given by ∆x∆k ' 1. Here ∆x is
the coherence length ξ defined in GL theory.

CM
2∆0
∴ ∆[Link] '1
EF
EF h̄2 kF2 h̄pF h̄vF
OR ∆x ' = = = (84)
kF .2∆0 [Link] .2∆0 4m∆0 4∆0
h̄vF h̄vF
ξ' ' 0.25
PA
4∆0 kB Tc
More rigorous calculations give
10−27 .108
ξ ' 0.18 kh̄vF
B Tc
' 0.18. 1.38X10 −16 X1 ' 10
−4
cm. which is large distance.
92
Temperature dependence of energy gap

II
At T= 0 K; ∆ = ∆0 ≡ ∆(T = 0) ≡ ∆(0). to break a Cooper pair,
we need 2∆ energy and ∆ ∼ ∆(T ). If kB T ' 2∆(T ), many cooper
pairs will be broken through thermal processes. A large number of

P
states in momentum space will be filled by elementary excitations.
These electron do not reduce the g.s. energy of SC. Therefore the

CM
energy of the SC has to increase.
As the no. of broken pairs grows, there is an increase of the no. of
elementary excitations and a decrease of the energy gap. The
elementary excitations and a decrease of the energy gap. The
elementary excitations obey FD statistics;
1
f~k = (85)
PA
E
exp ( kB~kT ) +1

where E~k is the energy of elementary excitations.

93
Temperature dependence of energy gap

II
• If at least one of the states (~k) or (−~k) is occupied, the pair
state (~k ↑, −~k ↓) cannot take part in creating the SC state. The
probability of this is 2f~k . this means the (~k ↑, −~k ↓) state can

P
take part in SC state is (1 − 2f~k .)

CM
• The expression for the total energy of the SC at T 6= 0 can then
be written in the Xform
~k (1−2f~k )v~k2 −V0
X X
W = 2|~k |f~k +2 v~k u~k v~k0 u~k0 (1−2f~k )(1−2f~k0 )
~k ~k ~k,~k0
(86)
The first term is the KE of elementary excitations 2nd term is KE of
SC electrons and third term is the energy of phonon mediated e
PA
interaction.
The free enrgy density of the SC is;
F = W − TS (87)
94
Temperature dependence of energy gap

II
v~k2 can be obtained from
∂F
=0 (88)
∂v~k2

P
From above eq 85, 86, 87 and 88 we get

CM
v~k u~k ∆
2
= (89)
1 − 2v~k 2~k
where X
∆ = V0 v~k u~k (1 − 2f~k ) (90)
~k
OR  
PA
X 2
∆(T ) = V0 v~k u~k 1 − E
 (91)
~k exp ( kB~kT ) + 1
gives the temperature dependence of the energy gap.
95
Temperature dependence of energy gap

II
Now using v~k2 and E~k we get

P
 
X ∆  2
∆ = V0 1− E~k

2E~k exp ( kB T ) + 1

CM
~k

1 Z h̄ωD
N (0) E~k
 
= q tanh d (92)
V0 0 2 + ∆2 (T ) 2kB T
q 
1 Z h̄ωD
d 2 + ∆2 (T )
= q tanh  
N (0)V0 0 2 + ∆2 (T ) 2kB T
PA
where we converted sum to integration.

96
Temperature dependence of energy gap

II
 
Therefore near Tc we get; ∆(T ) ' (Tc − T )1/2 . At T = Tc ; ∆ = 0
replacing T with Tc and setting ∆ = 0 in above eq yields

P
1 Z h̄ωD
d 
 
= tanh
N (0)V0  2kB Tc

CM
0
!
1
=⇒ kB Tc = 1.14h̄ωD exp −
N (0)V0 (93)
!
1
=⇒ ∆0 = 2h̄ωD exp −
N (0)V0
T hen 2∆0 = 3.52kB Tc
PA
above relations are verified experimentally.

97
Temperature dependence of energy gap

II
P
CM
PA

98
BCS concluding ....

II
Not only above a few experimental relations BCS theory has
explained all the results we discussed (mentioned only for many
properties like Josephson effects .....) in GL theory. However there

P
are some other types of SC so called unconventional one which
cannot be understood by BCS. There are heavy fermions (will be

CM
discussed later ...), high Tc SC, pnictide SC, oxide SC .... Its a really
rich physics. I have not seen any other physics more interesting
for me than in SC... Still we have not understood any other
satisfactory theory other than BCS even thoguh it can explain only
certain type of SCs. Its perfect for them. Others are beyond BCS.
Here we consider only the persistent current as a representative
example to discuss applications of BCS to understand the properties
PA
of a SC. Pl read BCS original papers and their Nobel Lectures if you
want ... OR if you want to be more serious in theory of SC you can
read book entitled ”Theory of Superconductivity” by Schreiffer
99
Persistent current - BCS theory

II
• The existence of the gap between the ground and excited states
in the superconducting state is the key factor in understanding

P
the disappearance of electrical resistance at temp below TC .

CM
• consider a current density caused by some rather small (to be
explained below) external electric filed E.~ The current is given
by J~ = −en~v where m~v = h̄δ~k and n is the density of the e gas.
~
As we know Fermi sphere moves in k space by δ k = neh̄ m ~
J.
• Therefore each electron in the cooper pair will experience this
shift so that the pair (~k ↑, −~k ↓) becomes
PA
(~k + δ~k ↑, −~k + δ~k ↓) = (~k + 21 K,
~ −~k + 1 K)
2
~ = 2h̄δ~k
~ where h̄K
is the change of the total momentum of the pair.

100
Persistent current - BCS theory

II
The wave function of the pair
1 X
ψ0 (~r1 , ~r2 ) = C~ exp (i~k.(~r1 − ~r2 )) (94)
V ~ k

P
k
becomes
1 X

CM
ψK~ (~r1 , ~r2 ) = C~ exp (i~r1 .(~k + K/2).
~ exp (i~r2 .(−~k + K/2)
~
V ~ k
k
1 X
= C~ exp (i~r.~k) exp (iR.
~ K)
~
V ~ k
k
~ K)ψ
= exp (iR. ~ 0 (~r1 , ~r2 )
(95)
PA
where ~r = (~r1 − ~r2 ) is the relative position of the electrons in the pair
and R~ = 1 (~r1 2~r2 ) is their center of mass. Note that we have
2
neglected here that the coefficient C~k will also be slightly changed by
the field. 101
Persistent current - BCS theory

II
Thus, the wavefunction of the cooper pair to a first approximation
chnages in an electric field only by a phase factor. This does not
affect the prob. density given by

P
|ψK~ (~r1 , ~r2 )|2 = |ψ0 (~r1 , ~r2 )|2 .
Also it does not affect the energy of the pair. Indeed, the matrix

CM
element of the electron-electron interaction U (~r1 , ~r2 ) mediated by the
lattice, does not change:
Z
~ 6= 0) =
U~k~k0 (K ~ exp (−i(~r1 − ~r2 ).~k)U (~r1 , ~r2 )
~ K)
d~r1 d~r2 exp (−iR.
~ exp (i(~r1 − ~r2 ).~k 0 )
~ K)
exp (iR.
~ = 0)
PA
= U~ ~ 0 (K
kk
(96)
and therefore the leectric field will not change the binding energy of
the cooper pairs.
102
Persistent current - BCS theory

II
P
As a result, the excitation energies of the SC in the current carrying

CM
state will be the same as in the non-current carrying one. In
particular, the gap between the ground and excited states will not
change.
PA

103
Persistent current - BCS theory

II
P
CM
PA

104
Persistent current - BCS theory

II
• This is the main point. If a current is excited in a normal metal,
then the Fermi sphere is shifted by δ~k in the K~ space as shown

P
in figure. When the source of the field is switched off, scattering
processes (due to phonons and impurities) will ”return” the

CM
sphere back to the original position i.e. th e with ~k vectors in
the right hatched area will return to the left hatched area (note
that ll other states are occupied.)

• Due to energy conservation, the energy of the initial e (on the

right) is larger then the energy of the state it occupies in
PA
equilibrium on the right. This is easily achievable for any current
(any sphere shift) for a normal metal.

105
Persistent current - BCS theory

II
• In a superconducting metal, the initial state should have energy
at least 2∆ higher than the final one since there are no states

P
allowed in the gap. In addition, as has been explained above, the
energy of a SC does not change appreciably in the field E ~ since
~ is sufficiently
the e are copper pairs. Therefore, if the field E

CM
enough ensuring a rather small shift of the Fermi sphere (the
corresponding criteria given below) , then there will be no simple
mechanism which could return the Fermi sphere back to the
equilibrium position. The current will flow forever.

• An alteration of the state, by inelastic e scattering at least

PA
(phonons) can only occur by excitation across the gap 2∆, i.e.
by breaking up at least one Cooper pair.

106
Persistent current - BCS theory

II
• There is however limit to the current which is to flow w/o
resistance. Indeed the electron in the right hatched area will be
able to get scattered in the left hatched area (and therefore to

P
destroy the superconductivity) provided that

CM
h̄2 ~ h̄2 ~
(kF + δ~k)2 − (kF − δ~k)2 ≥ 2∆ (97)
2m 2m
 
where δ~k = m
h̄en
~ This gives the critical current density
J.

2ne∆
Jc = (98)
h̄kF
PA
from which typical value of Jc is ' 2 × 107 A/cm2 . Compare this
critical current with the current flowing in your house hold wire.

107
Persistent current - BCS theory

II
• The criteria for the critical current obtained above is easy to
understand: the flowing current J~ increases the KE of the

P
electrons at the Fermi sphere by;

CM
h̄2 ~ h̄2 ~ h̄2~kF .K
~
(kF + δ~k)2 − (kF − δ~k)2 ' (99)
2m 2m 2m
If twice this energy (due to both electrons in the pair) is larger
than the energy which is necessary to break one cooper pair,
2∆, the current will collapse. this gives exactly the same
PA
expression for the critical current as above expression.

108
Strong Coupling

II
• Although the BCS theory explains type I superconductivity well,
deviations occur, especially in the case of type II. Hc (T ) of Pb

P
deviated from the BCS parabolic variation as much as a few
percent. There are other deviations as well. for example BCS
predicts;

CM
2∆0
= 3.52 (100)
kB Tc
but the ratio is as large as 4.6 in Hg and 4.3 in Pb. Eexc = 2∆0
represents the minimum energy necessary to excite a pair of
electrons from the ground state. The ratio of superconducting
to normal thermal conductivities of these SC has a limiting slope
PA
of order 10 at Tc in comparison with the value of order 1.5
expected from the BCS theory.

109
Strong Coupling

II
• Similar deviations in many other properties necessitate
reexamination of the theory.

P
∆0 ' h̄ωD (BCS)
(101)
−V~k~k0 ' −V0

CM
• However, e-phonon interaction causes a mass enhancement near
the EF and a finite life time of quasiparticle. With the excitation
energy of the order Debye frequency the life time of a
quasiparticle is short and its level width is of the order of the
excitation energy. That is its damping is very strong and a well
PA
defined quasiparticle no longer exists. Hence, the quasiparticle
picture becomes invalid.

110
Strong Coupling

II
• More detailed consideration of e-e interactions, frequency
dependency in energy transfers, and other refinements are
needed. This is theory of ”strongly Coupled Superconductors”.

P
The finite lifetime of a quasiparticle is expressed by the
imaginary part of the e self energy, its real part being the gap

CM
∆(ω). Note ∆ ' ∆(ω).
Pl make at least a brief concept of self energy
• When an e of momentum p~ is added to the g.s. of a normal
system, the zeroth order change in energy relative to the
chemical potential is simply:
PA
!
p2
p~ = −µ (102)
2m

111
Strong Coupling

II
P
• The lowest order energy shift due to the virtual emission and
reabsorption of a phonon by the added e is given by the Golden

CM
rule:
1 − f (~p0 )
|hp~p~0 |2
X
(103)
~0
p
p~ − p~0 − ωp~p~0
where hp~p~0 is the dressed e-phonon coupling and ωp~p~0 is the energy of
the virtual dressed phonon. A typical process is shown in figure.
PA

112
Strong Coupling

II
P
CM
Figure: a: Fluctuations in which the e in state p virtually emits a phonon,
goes to an unoccupied state p’, absorbs the phonon and returns to p. (b)
PA
Similar virtual transitions of an e from the Fermi sea to the state p which
is blocked by the addition of an e of momentum p.

113
Strong Coupling

II
The Fermi factor (1 − f (~p0 )) in above equation 103 ensures that the
state to which the e fluctuates is not occupied. A further
consequence of the Pauli principle is that the added e in state p~

P
blocks transitions of the type illustrated in (b), in which an e in the
Fermi sea emits a phonon and goes virtually to p~. These transition

CM
previously contributed to lowering the g.s. energy by an amount

f (~p0 )
|hp~p~0 |2
X
(104)
~0
p
p~0 − p~ − ωp~p~0

• Because they no longer can occur, the energy ω required to add

PA
a particle of momentum p~ is given by taking p~ minus the
blocking energy plus the transition energy:

114
Strong Coupling

II
f (~p0 ) 1 − f (~p0 )
!
2
X
ω = p~ + |hp~p~0 | − (105)
p~ − p~0 + ωp~p~0 p~0 − p~ + ωp~p~0

P
~0
p

The correction to p~ represents an approximation to the e self-energy

CM
Σ.
q An expression for the normal self energy which is accurate to order
( m/M ) can be obtained by making the Brillouin-wigner like
substitution p~ → ω in 105. Therefore, self energy is;

f (~p0 ) 1 − f (~p0 )
!
2
X
Σ (~p, ω) = |hp~p~0 | −
~0
ω + ωp~p~0 − p~0 − iδ −ω + ωp~p~0 + p~0 − iδ
PA
p
(106)
A small imaginary part (iδ) has been subtracted and it has to be
brought to zero.
115
Strong Coupling: Notes on Eq 106

II
The sum can be converted to to integration.

P
X Z Z
dΩp~0 1
= dp~0
(2π)3 vF

CM
~0
p
0
Z Z
dp 2 1 (107)
= dp~0
SF (2π)3 vF
Z Z
dS 0 1
= dp~0
SF (2π)3 vF

Now changing (p~0 ) to (−p~0 ) in the first term we get

PA

116
Strong Coupling: Notes on Eq 106

II
Z
dΩp~0 1 Z
Σ (~p, ω) = dp~0 |hp~p~0 |2
(2π)3 vF

P
(108)
f (~p0 ) 1 − f (~p0 )
!
−
ω + ωp~p~0 − p~0 − iδ −ω + ωp~p~0 + p~0 − iδ

CM
Since ω, ωp~p~0 are less than or of order ωD , the dominant
contributions to the p~ integrations comes for p~ ≤ ωD and therefore
p0 is essentially restricted to the Fermi surface.
Z
dS 0 1
Z
Σ (~p, ω) = dp~0 3
|hp~p~0 |2
SF (2π) vF
PA
0 (109)
1 − f (~p0 )
!
f (~p )
−
ω + ωp~p~0 − p~0 − iδ −ω + ωp~p~0 + p~0 − iδ

117
Strong Coupling

II
• The real part of Σ is the energy shift and the imaginary part of
Σ is given by;

P
|hp~p~0 |2 (1 − f (~p0 )) δ (p~0 + ωp~p~0 − ω) (ω > 0)
X
ImΣ (~p, ω) = −π

CM
~0
p
(110)
The level width parameter of ImΣ (~p, ω) is
1
= 2ImΣ (~p, ω) (111)
τ
where τ is lifetime of added e. The bare e-phonon coupling
PA
associated with a momentum change ~q = p~ − p~0 is given by;

118
Strong Coupling

II
!1/2
N
hq~ = −i ~q.~nq~V (~q) (112)
2M Ωq~

P
where Ωq~ is phonon energy, ~nq~ is a unit vector representing
polarization, M is the ion mass, V (~q) is an effective electron-rigid ion

CM
interaction, and N is the number of ions.
Strong e-phonon coupling means that not only is the e affected by
the phonon but also the phonon is affected by the e. Accordingly, the
Green’s functions of an e and a phonon are coupled with each other.
The ideal green’s function of an e is given by;
1
G0 (ω) = (113)
PA
ω − p~ + iδω
where δω = δsign(ω).δω is the sign f unction. for above eq pl see
separate reference may be by Carr...
119
Strong Coupling

II
• In the presence of a screened coulomb interaction between the

P
electrons, the Green’s function is modified such that;
1

CM
G(ω) = (114)
ω − p~ − Σ(~p, ω)
• The self energy depends on processes that can conveniently be
represented by diagrams. For the present Coulomb correction
alone, we need only the mechanism in which the e emits and
reabsorbs momentum representing the screened coulomb
PA
interaction.

120
Strong Coupling

II
• The self energy Σ(~p, ω) can be expanded at pF and ω = 0. For
p ' pF ;

P

p = pF +
vF
p~ ∂Σ ∂Σ (115)

CM
Σ(~p, ω) = Σ(p~F , 0) + +ω
vF ∂p ∂ω
Therefore Dr of 114 can be written as;
" #
p~ ∂Σ ∂Σ
ω − p~ − Σ(p~F , 0) + +ω
vF ∂p ∂ω
(116)
PA
! !
∂Σ 1 ∂Σ
=ω 1− − p~ 1 + − Σ(p~F , 0)
∂ω vF ∂p

121
Strong Coupling

II
• Now one can introduce a renormalization paramter;
∂Σ
Z =1− (117)
∂ω

P
and a renormalized quasiparticle energy;

CM
!
1 ∂Σ m
 
Zp∗~ = 1+ = p~ (118)
vF ∂ω m∗
where m∗ is an effective mass.
• Eqs. 117 and 118 represent how renormalization due to e-e
interactions take place. In terms of the renormalization
parameter Z and the effective mass m∗ , and with the
PA
replacement
−Σ(p~F , 0) → iδ,
the Green’s function is modified into;
122
Strong Coupling

II
1
G(ω) = (119)
Z(ω − p∗~ + iδ)

P
• In the above approximation the renormalization factor Z is
constant, but more elaborate calculations with frequency

CM
dependent Z(ω) are possible.

• As the e is affected by phonon, the later is also affected by the

former since the lattice vibration takes place in a medium of
dielectric constant (~q) due to the cloud of electrons. As a
result, the phonon frequency Ωq~ is renormalized as;
PA
Ωq~
ωq~ = q (120)
1 (~q)
where 1 (~q) is real part of the dielectric constant of electrons.
123
Strong Coupling

II
• The imaginary part of the dielectric function causes damping of
lattice vibrations. Also, similarly to the Coulomb correction Z in

P
the e propagator, the phonon propagator will have acorrection
Zp~ .

CM
• These and some other corrections lead us to a renormalized
e-phonon coupling function;
!1/2
N λ(~q) V (~q)
h̃q~ = −i ~q.~nq~ (121)
2M ωq~ q~ Z
PA
where q~ is dielectric constant and λ(~q) is a vortex correction.

124
Strong Coupling

II
• Two more steps are necessary to cast 109 this into the standard
form used in the gap equations. First, a δ-function is inserted to

P
separate the phonon contributions in a convenient manner;
Z ∞ Z ∞
dS 0 1

CM
Z
Σ (~p, ω) = dp~0 dω 0 |hp~p~0 |2 δ(ω 0 − ωp~p~0 )
0 0 SF (2π)3 vF

f (~p0 ) 1 − f (~p0 )
!
−
ω + ωp~p~0 − p~0 − iδ −ω + ωp~p~0 + p~0 − iδ
(122)

and second, Σ (~p, ω) is averaged over the Fermi surface, giving a

PA
self-energy Σ(ω) appropriate to an isotropic or dirty material:

125
Strong Coupling

II
Z ∞ Z ∞

P
Σ (~p, ω) = dp~0 dω 0 α2 (ω 0 )Np (ω 0 )
0 0
1 1
! (123)

CM
−
p~0 + ω + ω − iδ p~0 − ω + ω 0 − iδ
0

where Np (ω 0 ) (in some text its written as F(ω)) is the phonon DoS;

0 1 Z
d~p0 δ(ω 0 − ωp~p~0 )
X
Np (ω ) = 3
(124)
λ (2π)
PA
where α2 (ω 0 ) is an effective e-phonon coupling defined by

126
Strong Coupling

II
P
h i
1
dS 0 v1F |h̃p~p~0 |2 δ(ω 0 − ωp~p~0 )
R R
dS (2π)3
α2 (ω 0 )Np (ω 0 ) = R (125)
dS

CM
Eq 123 which is based on a simple mechanism, represents the
pertinent self-eenrgy expression. The first integral over p~0 can be
converted into a frequency integral similarly to the 2nd integral over
ω 0 . The frequency integral may then be expressed in terms of the
DoS associated with superconductivity that is characterized by an
energy gap. The DoS relative to that in the normal state is given for
PA
an energy dependent complex gap by

127
Strong Coupling

II
!
|ω|
Ns (ω) = g(EF )Re (126)
(ω − ∆2 (ω))1/2
2

P
where g(EF ) is the DoS at the Fermi surface of the normal state. in
the immediate vicinity of ∆(ω) ≡ ∆ = Constant, the DoS can be

CM
approximated by
" !#
1 ∂∆
Ns (ω) = NsBCS 1+ (f orω ≥ ∆) (127)
2 ∂ω

This represents that the DoS is slightly enhanced over the BCS case.
For Pb, the enhancement factor is 1.025.
PA
• We must note that the renormalization processes as discussed
above are necessary not only for the superconducting state but
also for the normal state.
128
Strong Coupling

II
• The existence of a gap ∆ causes the renormalization parameter
Zs of the SC state to differ from the Zn of the normal state.
However, if the e-phonon is weak, they must be nearly the same

P
Zs ' Zn (128)

CM
• Therefore strong coupling is characterized by the difference
(Zs − Zn ). According to calculations of Bardeen and Stephen
based on an einstein-phonon model with a characteristic
frequency ωE ;
!2
∆ ωE
 
Re[Zs (ω) − Zn (ω)] ' ln (f or ω < ωE )
ωE ∆
PA
2 !2 (129)
ωE ∆ ωE
  
' ln (f or ω < ωE )
ω ω ∆

129
Strong Coupling

II
Here, ∆ is a frequency gap independent of frequency at T= ). For
type I SC; !2
∆

P
∼ 10−3
ωE

CM
so that the difference

Re[Zs (ω) − Zn (ω)]

is approximately 0.1-1%. On the other hand for type II SC:

!
∆ 1
∼ .
PA
ωE 3

Therefore, difference is in range (10-15)%.

130
High Tc Superconductors and Theoretical
Models...

II
P
CM
You just go thro’ the TEXT.....
PA

131