Superconductivity

Masatsugu Sei Suzuki

Department of Physics, SUNY at Binghamton
(Date: April 19, 2013)

After getting a Ph.D. in physics from the University of Tokyo, and doing a research at
the Ochanomizu University (Tokyo, Japan), I came to the University of Illinois at
Urbana-Champaign (UIUC). I did my research, collaborating with Prof. Hartmut Zabal
(currently, Ruhr-Universität Bochum, Germany) in July 1984 (I stayed at UIUC for the
period between 1984 and 1985). First I really realized that a brilliant theory of BCS was
born from such a nice campus. I often saw Prof. Bardeen reading news paper at the
library of the Physics Department. Prof. William McMillan had worked hard using his
computers in basement of his home. At that times I was very interested in his ongoing
model of the spin glass (so-called domain model or droplet model). I also saw Prof.
David Pines (Bohm-Pines theory) and Prof. Anthony Leggett (Nobel laureate for
superfluidity of liquid 3He) in Physics Colloquium. In August, 1984, Prof. McMillan
unfortunately died because of car accident. So I missed the opportunity to learn about his
exciting model from him.
Recently I read a book titled "Genius, The life and science of John Bardeen, which
was written by Hoddeson and Daitch. I was very impressed by this book. It vividly
explained how the BCS theory was born from the hands of three distinguished scientists
in the Department of Physics (UIUC). I also noticed what happened to the fate of the
quantum theory of the charge density wave (CDW) proposed by Prof. Bardeen, around
1984 and 1985.
In 1986, the high temperature Tc superconductors were discovered by Dr. Georg
Bednorz and Dr. Karl Műller (IBM, Zűrich). Many physicists were so excited at the big
news. They attended the American Physical Society (APS) March Meeting 1987 (Spring)
(Hilton Hotel, NY City), in order to catch up with the ongoing research on the high Tc
superconductors such as YBa2Cu3O7-, so on, which cannot be explained by the BCS
theory. This March Meeting was called a Woodstock of Physics in New York Times.
Soon after I moved from Champaign (Illinois) to Binghamton (NY) (1986 Fall) as a
professor of Physics, Prof. Ivar Giaever (Rensslaer Polytechnic Institute, Troy, NY) came
to our campus (SUNY at Binghamton, NY) and gave an excellent and impressive talk on
his discovery of electron tunneling in superconductors, evidence of energy gaps in
superconductors (although his interest changed into the physics of brains). So I had a
chance to see him in the Colloquium.
________________________________________________________________________
John Bardeen (May 23, 1908 – January 30, 1991) was an American physicist and
electrical engineer, the only person to have won the Nobel Prize in Physics twice: first in
1956 with William Shockley and Walter Brattain for the invention of the transistor; and

1
again in 1972 with Leon Neil Cooper and John Robert Schrieffer for a fundamental
theory of conventional superconductivity known as the BCS theory.

http://en.wikipedia.org/wiki/John_Bardeen

Leon N Cooper (born February 28, 1930) is an American physicist and Nobel Prize
laureate, who with John Bardeen and John Robert Schrieffer, developed the BCS theory
of superconductivity. He is also the namesake of the Cooper pair and co-developer of the
BCM theory of synaptic plasticity.

http://en.wikipedia.org/wiki/Leon_Neil_Cooper

By late February or early March of 1956, it seemed clear that if somehow the entire
ground state could be composed of such pairs, one would have a ground state with
qualitatively different properties from the normal state. And this ground state - the state
of superconductivity - would be separated from the excited states by an energy gap.
(from True Genius, the life and science of John Bardeen, by L. Hoddeson and V. Daitch)

________________________________________________________________________
John Robert Schrieffer (born May 31, 1931) is an American physicist and, with John
Bardeen and Leon N Cooper, recipient of the 1972 Nobel Prize for Physics for
developing the BCS theory, the first successful microscopic theory of superconductivity.

2
http://en.wikipedia.org/wiki/John_Robert_Schrieffer

Schrieffer worked more on the expression that night at his friend's house. In the
morning, he did a variational calculation to determine the gap equation. "I solved the gap
equation for the cut off potential. It was just a few hours work." Expanding the
expression, he found he had written down a product of mathematical operators on the
vacuum that expressed adding electrons to the vacuum. In his sum of a series of terms,
each one corresponded to a different total number of pairs. He could hardly believe it.
The expression "was really ordered in momentum space" and the ground state energy
"was exponentially lower in energy," as required for the state to be stable.
(from True Genius, the life and science of John Bardeen, by L. Hoddeson and V. Daitch)
_______________________________________________________________________
Lev Davidovich Landau (Russian language: Ле́в Дави́дович Ланда́у; January 22 [O.S.
January 9] 1908– April 1, 1968) was a prominent Soviet physicist who made fundamental
contributions to many areas of theoretical physics. His accomplishments include the co-
discovery of the density matrix method in quantum mechanics, the quantum mechanical
theory of diamagnetism, the theory of superfluidity, the theory of second-order phase
transitions, the Ginzburg–Landau theory of superconductivity, the theory of Fermi liquid,
the explanation of Landau damping in plasma physics, the Landau pole in quantum
electrodynamics, and the two-component theory of neutrinos. He received the 1962
Nobel Prize in Physics for his development of a mathematical theory of superfluidity that
accounts for the properties of liquid helium II at a temperature below 2.17 K.

3
http://en.wikipedia.org/wiki/Lev_Landau

________________________________________________________________________

Vitaly Lazarevich Ginzburg ForMemRS (Russian: Вита́лий Ла́заревич Ги́нзбург;

October 4, 1916 – November 8, 2009) was a Soviet theoretical physicist, astrophysicist,
Nobel laureate, a member of the Russian Academy of Sciences and one of the fathers of
Soviet hydrogen bomb. He was the successor to Igor Tamm as head of the Department of
Theoretical Physics of the Academy's physics institute (FIAN), and an outspoken atheist.

http://en.wikipedia.org/wiki/Vitaly_Ginzburg

4
((Note)) Here I present the topics of superconductivity (Josephson effect will be
discussed in other chapter). I use the c.g.s. units. I use B for the magnetic induction
(internal magnetic field) and H for the external magnetic field.
________________________________________________________________________
1. Disappearance of resistivity below Tc
We show that the Meissner effect (B = 0) is the fundamental properties of the
superconductor. B = 0 cannot be derived from the property of  = 0.

E  J (Ohm's law).

where E is the electric field, J is the current density, and  is the electrical resistivity.
When   0 (J = constant), E must be zero. Using the Maxwell's equation, we have

1 B
E   0.
c t

The zero resistivity implies

B
 0,
t

which is different from B = 0 (Meissner effect). In other words, if B = 0, we can say that
B B
 0 . But even if  0 , we cannot say that B = 0.
t t

((Persistent current))

5
Fig. Schematic diagram for the persistent current. B is a thermal switch. It is closed so
that the current circulates in the superconducting contour inside cryostat D, filled
with liquid He. C is a superconducting ring which generates a magnetic field. The
persistent current is used for the generation of an extremely stable magnetic field
for the SQUID magnetometer.

2. Meissner effect: Meissner and Ochsenfeld (1933)

A bulk superconductor in a weak magnetic field will act as a perfect diamagnet with
zero magnetic induction in the interior.

6
 When a specimen is placed in a magnetic field and is then cooled through the critical
temperature for superconductivity, the magnetic flux originally present is ejected from
the specimen.
The demagnetiztion field contribution is negligible.

B  H  4M  0 ,

where M (emu/cm3) is the magnetization and H (Oe) is the external magnetic field.

H  4M .

((Note-1)) The unit of H is Oe. 1T = 104 Oe. 1 Gauss = 1 Oe.

((Note-2)) Experimentally we measure the magnetization M in the units of emu.

erg G 2  cm 3
emu    G  cm 3 .
G G

7
In the expression of B  H  4M  0 , the unit of M should be emu/cm3, since H and B
are in the units of G. In this case, the magnetic susceptibility is given by

M 1
  [emu/(cm3 Oe)]. (complete diamagnetism)
H 4

Conventionally we use emu/cm3, instead of emu/cm3 Oe.

-4pM

-4pM = H

O Hc H

Fig. (-4M) vs an external field H for the type-I superconductor. M is the

magnetization

Fig. B vs an external field H for the type-I superconductor. B is the magnetic

induction. B = H + 4M.

8
4. The distribution of B = 0 in the superconducting sphere
We consider a superconducting sphere of radius R placed in a uniform external
magnetic field H0. If H0 is small, the lines of force are expelled from the specimen. The
field configuration external to the sphere is determined by the equations,

B  0, B  0 , B → H0 as r →∞

where r is the distance measured from the center of the sphere. The Meissner effect
imposes the condition that no line of force can penetrate into the sphere. The normal
component of B vanishes on the surface of the sphere,

( Bn ) r R  0

The appropriate solution of B in the exterior region is

R 3 cos 
B  H 0e x  H 0  ( 2 )   
2 r

(P.G. de Gennes, Superconductivity of metals and alloys, .M. Tinkham, Introduction to

Superconductivity)

where

ex  coser  sine '

 1 
 ( r ,  )  e r  (r ,  )  e  (r ,  ) .
r r 

and

R3
 (r , )   H 0 (r  ) cos  .
2r 2

Note that

2  0

The field B is expressed by

9
 R3
B  H 0 (cos e r  sin e )  H 0 3 ( 2 cos e r  sin e ) .
2r

When r = R, we get

1
B  H 0 (cos e r  sin e )  H 0 (2 cos e r  sin e )
2
3
  H 0 sin e
2

Clearly the er component of B normal to the surface is equal to zero. We make a plot of
the distribution of B using the Mathematica (ContourPlot and StreamPlot).

-1

-2
-2 -1 0 1 2

Fig. Magnetic field distribution around a superconducting sphere of radius R. For an

external magnetic field H0 which is relatively low, there is a complete Meissner
effect. The length of arrows does not corresponds to the magnitude of B.

10
((Mathematica))
Clear"Global`";

V r_, _  :  H0 r  Cos  ;
H0 a3
2 r2

V2D  V r,  . r  x2  y2 ,   ArcTanx, y  Simplify;

rule1  a  1, H0  1; V2D1  V2D . rule1;

Hx   D V2D1, x  Simplify; Hy   D V2D1, y  Simplify;

g1  ContourPlotEvaluateTable V2D1  , ,  2, 2, 0.05, x,  2, 2, y,  2, 2,

ContourStyle  TableThick, Hue0.015 i, i, 0, 60,
RegionFunction  Functionx , y , x 2  y 2  1;

g2  StreamPlotEvaluateHx, Hy, x,  2, 2, y,  2, 2, StreamPoints  100,

StreamStyle  Blue, Thick, RegionFunction  Functionx , y , x 2  y 2  1;
g3  GraphicsBlue, Thick, Circle0, 0, 1;

Showg1, g2, g3, PlotRange  All

((Note)) Intermediate state


On the equatorial circle ( =  ) on the surface of sphere), the tangential component is
2
3
maximum and is H 0 , since the internal field B on the surface is given by
2

3 3
B   H 0 sin e   H 0e .
2 2

For B =2Hc/3, the field at the equatorial circle becomes equal to Hc. Thus For B>2Hc/3,
certain region of the sphere pass into the normal state. But there must be still
superconducting regions since B<Hc. In the domain

2H c
 B  Hc ,
3

there will be a co-existence of the normal and superconducting regions. This situation is
called the intermediate state.

For more detail, see also the article (pages 42 - 45, the superconducting sphere) in More
surprises in theoretical physics (R.E. Peierls, 1991, Princeton).

11
5. Parabolic law for critical field Hc vs T.
The relation between the critical magnetic field Hc(T) vs T for the transition between
the normal phase and the superconducting phase can be well described by a parabolic
law,

T2
H c (T )  H c (T  0 K )(1  2 ) ,
Tc

for T<Tc. In fact this relation is experimentally confirmed from the measurement of Hc(T)
vs T for the type I superconductors. The data of Hc(T) vs T for Sn is shown here.

Fig. Phase diagram of a transition from the normal to superconducting phase. The
critical field Hc vs T obeying a parabolic law

12
 250

Sn

200

150

100

50

0
1.5 2 2.5 3 3.5 4

T (K)

Fig. Critical field Hc of Sn (type-I superconductor) as a function of temperature

(Advanced Lab. SUNY at Binghamton). Hc(0 K) = 260 Oe and Tc = 3.7 K. The
data is fitted well to the parabolic law.

6. Experimental results from the Advanced Laboratory

In the Advanced laboratory (SUNY at Binghamton), we measured the magntization
of superconductors Pb and Sn (type-I superconductor) using SQUID magnetometer.

(a) Magnetization (emu/cm3) vs H (Oe) for Pb. T = 2.0 K. Lead (Pb): Type-I
superconductor. Tc = 7.193 K. Hc = 803 Oe.

13
 Pb (M vs H at T = 2.0 K)
0.5

-0.5

-1

-1.5

-2

-2.5

-3

-3.5
-200 0 200 400 600 800 1000 1200

H (Oe)

Fig. M vs H for Pb at T = 2.0 K. This is an original data obtained using SQUID

magnetometer (M.S.) at the Binghamton University.
a. T = 50 K, H = 0 Oe, annealing for 200 sec
b. Cooling the sample from 50 K to 2.0 K in the absence of H
c. Aging the system at 2.0 K for 100 sec.
d. Measure the magnetization at T = 2.0 K with increasing H from 0
to 1000 Oe.

(b) Zero-field cooled magnetization (emu/cm3) vs temperature T (K) for Pb. H is

changed as a parameter. Lead (Pb): Type-I superconductor. Tc = 7.2 K. HC = 803
Oe.

14
 M vs T for Pb
ZFC
0.5

0
720 Oe

700 Oe
-0.5
100 Oe

-1
650 Oe

200 Oe

-1.5
600 Oe
300 Oe
-2

500 Oe
-2.5 400 Oe

-3

-3.5
2 3 4 5 6 7 8

T(K)

Fig. The zero-field cooled magnetization MZFC (emu/cm3) vs T (K) for Pb for each H,
where H = 100 - 720 O2. The measurement was carried out as follows.
a. T = 50 K, H = 0 Oe, annealing for 200 sec
b. Cooling the sample from 50 K to 2 K in the absence of H
c. Aging the system at 2 K for 100 sec.
d Switch on the magnetic field H at 2 K.
e. Measure the ZFC magnetization with increasing T.
H = 1 Oe, 100, 200, 300, 400, 500, 600, 650, 700, 720 Oe

(c). Sn
Type-I superconductor: Tc = 3.722 K, Hc = 309 Oe

15
Fig. ZFC (zero-field cooled) magnetization M (emu/g) of Sn. Tc = 3.7 K. The critical
field is Hc = 300 Oe.

7. FC and ZFC susceptibility of superconductor

Recently, one can measure easily magnetic susceptibility of superconductor by using
SQUID magnetometer. There are two kinds of measurement of the susceptibility; zero-
field cooled (ZFC) susceptibility and field cooled (FC) susceptibility.
(i) ZFC susceptibility measurement.
The system is rapidly cooled from high temperatures well above the critical
temperature Tc to the lowest temperature T0 below Tc. After an external magnetic
field is applied to the superconductor at T = T0, the susceptibility is measured in
the presence of magnetic field with increasing temperature.
(ii) FC susceptibility
The FC susceptibility is measured in the presence of an external magnetic field as
the temperature is decreased from high temperature well above Tc to the lowest
temperature T0 well below Tc, in the presence of magnetic field.

16
 If the system exhibits a Meissner effect below Tc, the T dependence of ZFC
susceptibility is exactly the same as that of the FC susceptibility below Tc. They
exhibits a complete diamagnetic behavior,

1
 ZFC   FC   .
4

Figure shows a typical example of the ZFC and FC susceptibility for the
superconductor Rb3C60. The ZFC susceptibility (diamagnetic) is much smaller than the
FC susceptibility below Tc (= 28 K).

Fig. Magnetization of a sample of nominal composition Rb3C60. The data labeled ZFC
were obtained upon warming in a field of 2 Oe, after cooling the sample in zero
applied field. The FC data were obtained by cooling the sample in 2 Oe,
illustrating flux expulsion.
[M.J. Rosseinsky et al, Phys. Rev. Lett. 66, 2830 (1991); “Superconductivity at 28
K in RbxC60.”].

17
 H

FC susceptibility
H0
ZFC susceptibility

T
T0 Tc

Fig. H-T phase diagram, where the procedure of the ZFC and FC susceptibility
measurement for the type-I superconductor.

In the ZFC cooling process, the system undergoes a phase transition from the normal
phase to the Meissner phase, since no external magnetic field is applied. Even if the
external magnetic field (H0<Hc for a type-I superconductor) is applied at T=T0 (at the
lowest temperature), the Meissner phase with B = 0 remains maintained. In this sense, the
ZFC susceptibility reflects the nature of the Meissner phase.
How about the FC susceptibility? Suppose that the system has crystalline defects
inside it. The magnetic field is applied above Tc. Then the system is gradually cooled
from above Tc. The magnetic flux penetrates into the systems. A part of magnetic flux is
pinned to the defects. When the system is cooled down below Tc. Main part becomes in
the Meissner phase with B = 0. However, a part of the magnetic flux is pinned in the
defects and becomes frozen. In accompanying with this, the supercurrents flow around
the defect, in order to maintain the frozen magnetic flux. Such defects contribute to the
positive magnetization, leading to the FC susceptibility which is larger than the ZFC
susceptibility at the same temperature.

8. Isotope effect
Lattice vibrations (phonons) are responsible for the isotope effect in
superconductivity, which was discovered in 1950. The superconducting critical
temperature of Hg varie with isotopic mass M, from Tc = 4.185 K to 4.146 K, as the
isotopic mass M varies from 199.5 to 203.4. The critical temperature for superconductors
depends on the isotopic mass of the crystal, according to

Tc M   constant, (Isotope effect)

with  = 0.5. This indicates that lattice vibrations and hence electron-lattice interactions
are deeply involved in superconductivity. In BCS theory, it is predicted that

18
 Tc   D  M 1 / 2 .

where M is the mass of isotopes.

Fig. Isotope effect for Sn. The results of several authors are summarized. Maxwell (○);
Lock, Pippard, and Shoenberg (∎); Serin, Reynolds, and Lohman ().  = 0.47 ±
0.02 (from ISSP).

8. Type-II superconductor
The difference between the type-I and type-II superconductors are characterized by
the Ginzburg-Landau parameter ,


 ,


where  is the penetration length of magnetic field and  is the coherence length.

19
Fig. Boundary between the N-phase and S-phase for type-I superconductor.
 1
  .  is the Ginzburg-Landau parameter.
 2

20
Fig. Boundary between the N-phase and S-phase for type-II superconductor.
 1
  .
 2

There are two phases in the phase diagram of H vs T below Tc for the type-II
superconductor, the Meissner phase (H<Hc1) and the mixed phase (Shubnikov phase, or
vortex state, mixed phase) for Hc1<H<Hc2.

(a) Meissner phase (H<Hc1). (b) Mixed phase (Hc1<H<Hc2)

21
Fig. A single vortex which penetrates into the superconductor. The supercurrent flows
in a counterclockwise direction in this figure.

9. Surface energy
Consider the interface between a region in the S-state and a region in the N-state. The
interface has a surface energy that may be positive or negative .
Type II: The surface energy becomes negative.
Type I: The surface energy is positive.

10. Thermodynamics of superconductors

Fn: Free energy of superconducting phase

Fs: Free energy of normal phase

The Helmholtz free energy is

dF  d ( E  ST )  TdS  PdV  SdT  TdS

  PdV  SdT

Replacing

Intensive variable P→M

Extensive variable V→H

22
(see ISSP of Kittel),

we have the relation

dF  MdH  SdT

For T = constant,

dF  MdH (thermodynamics identity).

(a) The superconducting phase;

1
M  H (Meissner effect).
4

Then we have

H H
1 H2
Fs ( H )  Fs ( H  0)   dFS   dH  ,
0
4 0
8

or

H2
Fs ( H )  Fs ( H  0)  .
8

(b) The normal phase

Since M = 0,

Fn ( H )  Fn ( H  0 ) .

At the critical field Hc(T), the energies are equal in the normal and superconducting
states,

Fn ( H c )  Fs ( H c ) ,

or

[ H c (T )]2
Fn ( H  0)  Fs ( H  0)  .
8

Then it follows that

[ H c (T )]2
F  Fn (0)  Fs (0)  .
8

23
which is the stabilization free energy of the superconducting state ( or the condensation
energy).
F

FN FN
N-phase

S-phase

Hc H

Fig. The free energy density FN for the normal phase and Fs for the
superconducting phase. The S-phase is favorable for H<Hc and the N-
phase is favorable for H>Hc.

We define the entropy as

 F   F 
Sn   n  , S s   s  .
 T V  T V

The difference between these entropies is

  ( Fn  Fs )  1 dH c (T )
S n  S s     H c (T ) .
 T V 4 dT

The latent heat of the transition is

T dH c (T )
L  T (Sn  S s )   H c (T ) .
4 dT

24
The parabolic law:

T2
H c (T )  H c (0)(1  2
),
Tc

dH c (T ) 2T d 2 H c (T ) 2
 H c (0)( 2 ) . 2
 H c (0)( 2 ) .
dT Tc dT Tc

For T = Tc,

L = 0.

Since the entropy is continuous at T = Tc, the phase transition at T = Tc is the second
order.

11. Second order phase transition

The specific heat is defined by

 S 
CV  T   .
 T V

Then we get

25
  ( Ss  Sn ) 
(Cs  Cn )V  T  
 T V
T d dH c (T )
 [ H c (T ) ]
4 dT dT
T d 2 H c (T ) dH c (T ) 2
 {H c (T ) [ ]}
4 dT 2 dT

At T = Tc,

Tc dH c (T ) 2 1
(C s  Cn )V  [ ] |T Tc  [ H c (0)]2
4 dT Tc

Then the specific heat is discontinuous at T = Tc.

12. Thermodynamics for the type-II superconductor (de Gennes)

(a) Gibbs free energy per unit volume for the superconducting state

Gs can be written as

BH
Gs  Fs ( B)  ,
4

Gs has a minimum as a function of B for fixed H,

 Gs 
   0.
 B  H

(b) Gibbs free energy per unit volume for the normal state.

Gn can be written as

BH 0 B2 BH
Gn  Fn ( B )   ( Fn  ) .
4 8 4

((Note)) The first term of the right hand side can be derived from the discussion below.

Gn has a minimum as a function of B for fixed H,

26
  Gn 
   0.
 B  H

Then we have

BH,

leading to the expression of Gn

0 H2
Gn  Fn  ,
8

in the normal state.

________________________________________________________________________
(iii)

Let the field vary from H to H + H

Gs B Gn H
 ,  .
H 4 H 4

Thus we have

 BH
(Gn  Gs )  M .
H 4

We now integrate the relation between H = 0 and H = Hc2. We note that

(i) At H = Hc2 (the upper critical field)

Gn  Gs

(ii) At H = 0,

0 0
Gn0  Fn , Gs0  Fs ,

and by definition

27
 2
H
F F  c .
0 0

8
n s

Then we get

Hc2
1 2
 (4M )dH  4 [G
0
n  Gs ] |0H c 2  4 (Gn  Gs ) H  0 
2
Hc

13. The magnetization vs H for the type-II superconductor

(i)

Hc
1
 HdH  2 H
2
c
0

(ii)

28
 Hc2 H c1 Hc2

 (4M )dH   (4M )dH   (4M

0 0
1
H c1
2 )dH

H c1 Hc2

  HdH 
0
 (4M
H c1
2 )dH

Hc2
1 1 2
H c1   (4M 2 )dH  H c
2

2 H c1
2

or

Hc2
1
 (4M
2 2
2 )dH  ( H c  H c1 ) .
H c1
2

This means that the area (BCD) is equal to the area (DEF) in the magnetization vs H
curve for the type-II superconductor.

14. B vs H for type-II superconductor

B
B=H

Meissner phase Mixed phase

H
Hc1 Hc Hc2

The magnetic induction B (the mean magnetic field in the interior of a type II
superconductor is plotted as a function of an external magnetic field H. For H<Hc1, a
type-II super conductor behaves exactly like a type-I superconductor, exhibiting a perfect
diamagnetism (Meissner effect). At H = Hc1, normal cores with their associate vortices
form and pass into the system. The magnetic flux threading the vortices is in the same
direction as that due to H, so that the magnetic flux is no longer equal to zero. For

29
Hc1<H<Hc2, the number of vortices which occupy the system is governed by the fact that
vortices repel each other. The number of normal cores per unit area for a given strength
of H is such that there is equilibrium between the reduction in free energy of the system
due to the presence of each non-diamagnetic core and the existence of the mutual
repulsion between vortices. As H is increased, the normal cores pack closer, so the
average flux density in the system increases. At H = Hc2, there is a discontinuity change
in the slope of the flux density.
A good type-II superconductor excludes the field completely for H<Hc1.Above Hc1,
the field is partially excluded, but the specimen remains superconducting. For H>Hc2, the
magnetic flux penetrates completely and superconducting vanishes.

Figure shows the plot of -4M vs H for typical type-II superconductors, where M is the
magnetization and H is an external magnetic field. We note that the line BC for the
typical type-I superconductor is described by

 4M 1  H .

The line BDE for the typical type-II superconductor is described by

 4M 2  H  B

where B is the magnetic induction (the mean field inside the superconductor). Note that
The area (BCD) and area (DEF) are expressed by

Hc Hc Hc

Area (BCD) =  [(4M )  (4M

H c1
1 2 )]dH   [ H  ( H  B)]dH   BdH
H c1 H c1

30
 Hc2 Hc2

Area (DEF) =  (4M 2 )dH 

Hc
 ( H  B)dH
Hc

From the condition that area (BCD) is equal to area (DEF), we get

Hc Hc2

 BdH 
H c1
 ( H  B)dH
Hc

or

Hc Hc2 Hc2

 BdH   BdH   HdH

H c1 Hc Hc

or

Hc2 Hc2

 BdH   HdH
H c1 Hc

Since

Hc2 Hc2

 BdH = A2+A3,
H c1
 HdH = A1 + A2,
Hc

31
we have

A2 + A3 = A1 + A2,

or

A3 = A1.

15. Specific heat of type-I superconductor Al

The specific heat as a function of T changes discontinuously at the critical
temperature Tc. Figure shows the example of Al with Tc = 1.140 K and Hc = 100 Oe. The
specific heat undergoes a sharp jump at Tc with decreasing temperature. Then it sink to
below the value of the normal phase, at very low temperatures.

32
Fig. Specific heat of Al between 0.1 and 2.0 K. [N.E. Philips, Phys. Rev. 114, 676
(1959)]. Al (Tc = 1.140 K, Hc = 100 Oe). This figure is made using AppleDraw
based on the data obtained by Philips.

16. Specific heat in the normal state

The specific heat Cn of a normally conducting metal is composed of lattice part CL
and an electronic part Cne,

C n  C ph  C ne

where Ce is the electronic specific heat

Cne  T

with

2 2 k B
2
 N ( F )
3

Note that

D ( F )  2 N ( F )

and D(F) is the number of density (conventional) and N(F) is the number of density per
spin at the Fermi energy. Cph is the contribution of phonon to the specific heat.

3
T 
C ph    , ( = 427.7 K for Al)


which is negligibly small at low temperatures. Thus the normal specific heat in the
vicinity of Tc (T>Tc) is well described by

Cn  T

with

  1.35 mJ/(mol.K) for Al.

33
((Note))

 (mJ/mol K2) = 2.35715 DA ( F )

with

D( F )
DA ( F )  (1/eV atom)
N0

(see the detail for the notation in the Appendix-II)

17. Specific heat in the superconducting state

The specific heat of Al below Tc is well described by

Ces 1.34T c
 7.1 exp(  )
Tc T

for Al. Such an exponential form of 1/T in specific heat near Tc suggests an energy band
gap for the superconducting material. This band gap is one of the experimental evidence
which supports the BCS theory of superconductivity. For T0, the BCS theory predicts
that

Cs 1.5Tc
 9.17 exp( ),
Tc T

for the specific heat of conventional superconductors.

34
Fig. Electronic specific heat in the superconducting state, Ces for Al. The different
symbols distinguish the results of two completely separate experiments [N.E.
Philips, Phys. Rev. 114, 676 (1959)]. Al (Tc = 1.140 K, Hc = 100 Oe).

18. Prediction from BCS theory on specific heat

The BCS theory ( which will be discussed later) predicts that

8 2
C  (Cs  Cn ) |T Tc 0  k B Tc N ( F )
2
.
7 (3)

Then the ratio at T = Tc is predicted as

35
 8 2
k B Tc N ( F )
2

Cs  Cn 7 (3)
12
|Tc    1.43
Cn 2 k B
2 2
7 (3)
Tc N ( F )
3

For Al, this ratio is about 1.34, which is a little smaller that that predicted from the BCS
theory. For k BT   , it is predicted that

3/ 2
   
Cs    exp( ).
k T
 B  k BT

The energy gap is given by

1/ 2 1/ 2
 8   Tc  T 
  kBTc     .
 7 (3)   Tc 

((Note))
Comment by Richard Feynman, how to attack on the problem of superconductivity
["Richard Feynman and Condensed Matter Physics," D. Pines, Phys. Today 42, p.61
(1989)]

I decides it would be easiest to explain the specific heat rather than the electrical
properties. But we do not have to explain any entire specific heat curve; we only have to
explain any feature of it, like the existence of a transition, or that the specific heat near
absolute zero is less than proportional to T. I chose the latter because being near
absolute zero is a much simpler situation than being at any finite temperature. Thus the
property we should study is this: Why does a superconductor have a specific heat less
than T?

19. Quasi particle tunneling

Ivar Giaever (Giæver, born April 5, 1929, in Bergen, Norway) is a physicist who shared
the Nobel Prize in Physics in 1973 with Leo Esaki and Brian Josephson "for their
discoveries regarding tunnelling phenomena in solids". Giaever's share of the prize was
specifically for his "experimental discoveries regarding tunnelling phenomena in ...
superconductors". Giaever is an institute professor emeritus at the Rensselaer Polytechnic
Institute, a professor-at-large at the University of Oslo, and the president of Applied
Biophysics.

36
http://en.wikipedia.org/wiki/Ivar_Giaever

Giaever (1960) discovered that for the superconductor-insulator-metal sandwich, the

current-voltage characteristic changes from the straight line to the kinked curve. This
result indicates that the superconductor has a energy gap centered at the Fermi energy. At
T = 0 K, no current can flow until the applied voltage is V = /e. The energy gap
corresponds to the break-up of a pair of electrons in the superconducting state.

20. Semiconductor model

(a) S-I-N junction

37
Fig. Semiconductor model for the S-I-N junction at T = 0 K. V = 0. The Fermi energy
of the superconductor is at the same level as that of the metal.

Fig. Semiconductor model for the S-I-N junction at finite temperature below Tc of the
superconductor. V = 0. The Fermi energy of the superconductor is at the same
level as that of the metal. As a result of the breaking up of Cooper pair, a part of
electrons is excited to the upper level. There are holes in the lower level.

Fig. Semiconductor model for the S-I-N junction at finite temperature below Tc of the
superconductor. V = . The Fermi energy of the superconductor is higher than
that of the metal by V. The current flows between the superconductor and metal.

38
 The excitation of electrons is taken into account because of finite temperatures.
However, for simplicity, this effect is not described in this figure.

(b) S1-I-S2 junction

Fig. Semiconductor model for the S1-I-S2 junction at T = 0 K. V = . The energy gaps
are 2A for the superconductor S1 and 2B for the superconductor S2. The Fermi
energy of the superconductor S1 is the same as that of the superconductor S2.

Fig. Semiconductor model for the S1-I-S2 junction at T = 0 K. V = BA..

39
Fig. Semiconductor model for the S1-I-S2 junction at T = 0 K. V = BA.

21. Result of Giaevr and Megerle

The experimental results were obtained by Giaever and Megerle. Reference: I. Giaever
and K. Megerle, Phys. Rev. B 122, 1101 (1961).

Al: Tc = 1.140 K, Hc = 100 Oe;

Pb: Tc = 7.193 K, Hc = 803 Oe
Sn: Tc = 3.722 K, Hc = 309 Oe

40
Fig. I vs V characteristics of an Al-Al2O3-Sn sandwich at various temperatures. Fig.6
[I. Giaever and K. Megerle, Phys. Rev. B 122, 1101 (1961)].

41
Fig. The negative-resistance region traced out for different Al-Al2O3-Pb sandwiches
Fig.17 [I. Giaever and K. Megerle, Phys. Rev. B 122, 1101 (1961)].

22. Josephson junction which was missed by Giaever

In a series of his experiments, Giaever missed phenomena known as a DC Josephson
effect at zero voltage.

The detail of the DC Josephson junction will be discussed later.

42
Fig. Schematic diagram of quasiparticle I-V characteristic (usually observed in a S-I-S
Josephson tunneling-type). Josephson current (up to a maximum value Ic) flows at
V = 0.  is an energy gap of the superconductor . The DC Josepson supercurrent
flows under V = 0. For V>2/e the quasiparticle tunneling current is seen.

23. Energy gap

Fig. The energy gap of Pb, Sn, and In films as a function of reduced temperature,
which is compared with the BCS theory.Fig.11 [I. Giaever and K. Megerle, Phys.
Rev. B 122, 1101 (1961)].

43
24. Vortex state (Mixed state) Hc1<H<Hc2

Fig. Repulsive interaction between vortices (fluxoids).

There is a repulsive force between two vortices. due to the Lorentz force. The repulsive
force is given by

l lA
F  IB  JB ,
c c

or the force density is given by

F 1
f  JB
V c

44
Fig. Abrikosov lattice for Hc1<H<Hc2. The votices form a triangular lattice in the
above case. The arrows show the direction of the supercurrents.

At H = Hc2, the spacing of the Abrikosov lattice is on the order of the coherence length.
There is one vortex per lattice.

0
H c2 
2 2

At H = Hc2, the fluxoids are packed together as tightly as possible, consistent with the
preservation of the superconducting state.

45
 F0

Fig. The spacing of the Abrikosov lattice is close to the coherence length  just below
Hc2.

Fig. Scanning tunnel microscopy (STM) image of a type-II superconductor. Abrikosov

vortex lattice (H.F. Hess et al. 1989). This figure is copied from PPT of Cooper.

25. The critical field Hc1 and Hc2 for type-II superconductor
The vortex state describes the circulation of superconducting currents in vortices
throughout the bulk specimen. The vortex state is stable when the penetration of the
applied field into the superconducting material causes the surface energy to become
negative.

Estimation of Hc1:

At H = Hc1, there is one magnetic flux per the penetration depth.

46
 2 H c1   0

Hc1 is the field of nucleation of a single fluxoid

26. Movement of vortex leading to the disappearance of superconductivity

We consider Why the mixed state is still superconducting in spite of the penetration
of magnetic flux.

F 1
f  JB ,
V c

Because of the Lorentz force, the magnetic flux lines tend to move transverse to the
current density J.
If they do move, with velocity v, they essentially induce an electric field of magnitude

Fig. Movement of magnetic flux (fluxoid) when the magnetic field is applied to the
system in the mixed phase.

47
Fig. The generation of DC voltage through an AC Josephson effect, when the
magnetic flux (fluxoid) moves at the velocity vf.

The area is given by

 A  ( v f t ) x

The total change in the superconducting phase  is given by

  2n f A  2n f (v f t ) x

where nf is the area density of vortices (fluxoids) and vf is the velocity of vortices. The
magnetic flux penetrating in the area A is

  n f  0 A  n f  0 (v f t ) x

where

hc c
0  
2e e

27. Approach from the AC Josephson effect

We can evaluate the voltage V produced across the distance x along the x direction,
using the formula of the AC Josephson effect,

48
 d 2e
 V (AC Josephson effect)
dt 

Since

 2n f (v f t ) x 2e
  2n f (v f ) x  V
t t 

we get the electric field as

V   v
E  n f v f  n f v f 0  n f f  0 .
x e c c

Note that the effective field B0 is


B0   n f 0 .
A

Using this expression of B0, we have the final form of E as

vf B
E .
c

In summary, in the mixed phase of the type-II superconductors, the resistivity is

generated when the vortices which are pinned by impurities, dislocations, and so on,
starts to move due to the Lorentz force. This means the destruction of the
superconductivity.

28. Approach from the special relativity

49
Fig. Application of the special relativity when the magnetic vortex moves at the
velocity vf.
Fig. Configuration where the quantum magnetic flux moves at the velocity vf. The
internal magnetic field (magnetic induction) B is B = nf0. As a result, the electric
field is set up between the lines AB and

We consider the two frames K and K'. The frame K is the rest frame and the frame K' is
moving to the right at a velocity vf relative to the frame K. There is no electric field E in
the K. The magnetic field B is

n f  0 (x)(v f t )
Bz   n f 0
(x)(v f t )

In the frame K', a electric field is newly established as

1 1
E'  v f  B  v f  B'
c c

1
where B' (=  B) and   1
v2
1 2
c

Since B' is nearly equal to B, we have

1
E' vf B
c
__________________________________________________________________
29. Flux pinning preserving the superconductivity

50
 Suppose that there is a pinning force which cancels out the Lorentz force. Then the
velocity is equal to zero. Then we have V = 0, which means that the system is still
superconducting state where the resistivity is equal to zero.
Flux pinning is the phenomenon that magnetic flux lines do not move (become
trapped, or "pinned") in spite of the Lorentz force acting on them inside a current-
carrying Type II superconductor. Flux pinning is only possible when there are defects in
the crystalline structure of the superconductor (usually resulting from grain boundaries or
impurities).

Fig. Balance of the Lorentz force and pinning force. The system is still in the
superconducting phase.
____________________________________________________________________
30. Ginzburg-Landau (GL) Theory
It is surprising that the rich phenomenology of the superconducting state could be
quantitatively described by the GL theory,11 without knowledge of the underlying
microscopic mechanism based on the BCS theory. It is based on the idea that the
superconducting transition is one of the second order phase transition. In fact, the
universality class of the critical behavior belongs to the three-dimensional XY system
such as liquid 4He. The general theory of the critical behavior can be applied to the
superconducting phenomena. The order parameter is described by two components
i
(complex number    e ). The amplitude  is zero in the normal phase above a
superconducting transition temperature Tc and is finite in the superconducting phase
below Tc. In the presence of an external magnetic field, the order parameter has a spatial
variation. When the spatial variation of the order parameter is taken into account, the free
energy of the system can be expressed in terms of the order parameter  and its spatial
derivative of  . In general this is valid in the vicinity of Tc below Tc, where the
amplitude  is small and the length scale for spatial variation is long.

51
 The order parameter  is considered as a kind of a wave function for a particle of
charge q* and mass m*. The two approaches, the BCS theory and the GL theory, remained
completely separate until Gorkov13 showed that, in some limiting cases, the order
parameter  (r) of the GL theory is proportional to the pair potential (r) . At the same
time this also shows that q* = 2e (<0) and m* = 2m. Consequently, the Ginzburg-Landau
theory acquired their definitive status.
The GL theory is a triumph of physical intuition, in which a wave function  (r) is
2
introduced as a complex order parameter. The parameter  (r ) represents the local
density of superconducting electrons, ns (r) . The macroscopic behavior of
superconductors (in particular the type II superconductors) can be explained well by this
GL theory. This theory also provides the qualitative framework for understanding the
dramatic supercurrent behavior as a consequence of quantum properties on a macroscopic
scale.
The superconductors are classed into two types of superconductor: type-I and type-II
superconductors. The Ginzburg-Landau parameter  is the ratio of to , where  is the
magnetic-field penetration depth and  is the coherence length of the superconducting
phase. The limiting value   1 / 2 separating superconductors with positive surface
energy (  1 / 2 ) (type-I) from those with negative surface energy (  1 / 2 ) (type-
II), is properly identified. For the type-II superconductor, the superconducting and normal
regions coexist. The normal regions appear in the cores (of size ) of vortices binding
individual magnetic flux quanta  0  2c / q * on the scale , with the charge q *  2 e
appearing in 0 a consequence of the pairing mechanism. Since >, the vortices repel
and arrange in a so-called Abrikosov lattice. In his 1957 paper, Abrikosov14 derived the
periodic vortex structure near the upper critical field Hc2, where the superconductivity is
totally suppressed, determined the magnetization M(H), calculated the field Hc1 of first
penetration, analyzed the structure of individual vortex lines, found the structure of the
vortex lattice at low fields.

31. Ginzburg-Landau theory-phenomenological approach

We introduce the order parameter  (r) with the property that

 * (r ) (r )  ns (r ) ,

which is the local concentration of superconducting electrons. We first set up a form of

the free energy density Fs(r),

52
 2
12 4 1  q* B2
Fs (r )  FN        (   A )  
2 2m* i c 8

where  is positive and the sign of  is dependent on temperature.

We must minimize the free energy with respect to the order parameter (r) and the
vector potential A(r). We set

   Fs (r )dr 

where the integral is extending over the volume of the system. If we vary

 (r)  (r)   (r) and A(r)  A(r)  A(r) ,

we obtain the variation in the free energy such that

   

By setting   0 , we obtain the GL equation

2
1 
2 q* 
      *    A    0 ,
2m  i c 

and the current density

*2 2
q* * q 
Js  *
[    * ]  A
2m i m*c

or

q* *  q*  q*
Js  *
[ (   A )   (    A ) * ] .
2m i c i c

At a free surface of the system we must choose the gauge to satisfy the boundary
condition that no current flows out of the superconductor into the vacuum.

n  Js  0 .

53
32. GL free energy and Thermodynamic critical field Hc
A = 0 and     (real) has no space dependence. Why  is real? We have a gauge
transformation;

iq * 
A'  A   and  ' (r )  exp( ) (r ) .
c

We choose A'  A    A with  = 0 = constant. Then we have

iq  0
*
iq  0 *
 ' (r )  exp( ) (r ) , or  (r )  exp(  ) ' (r ) . Even if  ' is complex number,
c c
 can be real number.

1
Fs  FN       
2 4

F
When  0 , Fs has a local minimum at
 

   ( /  )1 / 2  (  /  )1 / 2 .

Then we have

2 H
2
Fs  FN    c
2 8

from the definition of the thermodynamic critical field:

1/ 2
 4 2  T2
H c     H c (0)(1  ) 
  
2
Tc

Suppose that  is independent of T, then

 T2  T T
  H c (0)(1  2 )  2 H c (0)(  1)   0 (  1) 
4 Tc 4 Tc Tc

where

54
 
0  2 H c (0) .
4

The parameter  is positive above Tc and is negative below Tc. Note that >0. For T<Tc,
the sign of  is negative:

1
Fs  FN  0 (t  1)     
2 4

where 0>0 and t = T/Tc is a reduced temperature.

f
3


-2 -1.5 -1 -0.5 0.5 1 1.5 2
-1

-2

-3

Fig. The GL free energy functions expressed by Eq.(5.7), as a function of   . 0 = 3.
 = 1. t is changed as a parameter. t = T/Tc. (t = 0 – 2) around t = 1.



1.4
1.2
1
0.8
0.6
0.4
0.2
t
0.2 0.4 0.6 0.8 1

Fig. The order parameter   as a function of a reduced temperature t = T/Tc. 0 = 3. 
= 1.

33. Coherence length from GL theory
We assume that A = 0. We choose the gauge in which  is real.

55
 2 d 2
    * 2   0 .
3

2m dx

We put     f .

2 d 2 f
  f  f 3 0.
2 m  dx
* 2

We introduce the coherence length

2ns  2
*
2 2
  *  *  * 2 ,
2

2m  2m  m Hc

where

 2 Hc2 
 ,  2   ns ,
2 8 

or

2 T
 | 1  |1 / 2 
2m  0
*
Tc

Note that the coherence length diverges as the temperature approaches Tc. Then we have

d2 f
2  f  f 3  0,
dx 2

with the boundary condition f = 1, df/dx = 0 at x = ∞ and f = 0 at x = 0.

df d 2 f df
2 2
 ( f  f 3)  0,
dx dx dx

or

2
d  2  df  d f4 f2
   (  ),
dx 2  dx  dx 4 2

56
or

2
 2  df 
1
   (1  f ) ,
2 2

2  dx  4
or

df 1
 (1  f 2 ) .
dx 2

The solution of this equation is given by

 x 
f  tanh  .
 2 

fx

0.8

0.6

0.4

0.2

x
1 2 3 4 5

Fig. Normalized order parameter f(x) as a function of x/.

34. Physical meaning of coherence length 

The coherence length is a measure of the distance within which the superconducting
electron concentration cannot change drastically in a spatially-varying magnetic field.
The coherence length  is a measure of the range over which we should average A to
obtain the current density J. It is also a measure of the minimum spatial extent of a
transition layer between the normal metal and superconducting phases.
We consider two kinds of wavefunctions given by

  k

1
  [ kq  k ]
2

57
where
1
 ( x)  x   x k  e ikx
2

and

 ( x)  x 
1
 [ x k  x k q ]
2
1 1 ikx
 e (1  eiqx )
2 2

where the probability amplitude are

2 1
 ( x) 
2

2 1 1
 ( x)  (1  e iqx )(1  e iqx )  [1  cos(qx)]
4 2

The kinetic energy;

2k 2
 Hˆ  
2m

1
 Hˆ   ( k  k  q ) Hˆ ( k  k  q )
2
2
 ( k  k  q )[k 2 k  (k  q) 2 k  q )
4m
2 2
 [k  (k  q) 2 ]
4m
2 2 2 2
 (2k  q  2kq) 
2 2
k  kq
4m 2m 2m

where we neglect q2 for q<<k.

58
 Probability density
2.0
Strongly modulated wave

1.5

1.0
Plane wave

0.5

qx
p 3p 5p 7p
0 p 2p 3p 4p
2 2 2 2

The increase of energy required to modulate is

2
kq .
2m

If this increase exceeds g, the superconductivity will be destroyed. The critical value q0
of the modulation wave vector is given by

2
k F q0   g
2m

The intrinsic coherence length 0 is defined as

1  2kF v
0    F
q0 2m g 2 g

where vF is the Fermi velocity.

((Note)) BCS theory

2v F v F
0  
 g 

In impure materials and in alloys,

  0

59
since in impure metals the electron eigenfunctions already have wiggles in them.

35. Lagrangian of particles with mass m* and charge q* in the presence of

magnetic field
The Lagrangian L for the motion of a particle in the presence of magnetic field and
electric field is given by

1 * 2 1
L m v  q* (  v  A) ,
2 c

where m* and q* are the mass and charge of the particle. A is a vector potential and  is a
scalar potential.

The canonical momentum is defined as

L q*
p  m* v  A .
v c

The mechanical momentum (the measurable quantity) is given by

q*
π  m* v  p  A.
c

The Hamiltonian H is given by

q* 1 1 q* 2
H  p  v  L  (m* v  A )  v  L  m* v 2  q *  *
( p  A )  q * .
c 2 2m c

The Hamiltonian formalism uses A and , and not E and B, directly. The result is that the
description of the particle depends on the gauge chosen.

36. Current density for the superconductors

We consider the current density for the superconductor.  is the order parameter of
the superconductor and m* and q* are the mass and charge of the Cooper pairs. The
current density is invariant under the gauge transformation.

q* q*
Js  Re[  ˆ
p  A ],
m* c

60
This can be rewritten as

q* *  q*
Js  Re[  (    A )]
m* i c
q* *  q*  q*
 [(    A *
 )  (    *
 A * )]
2 m* i c i c
*2 2
q* q 
 *
( *   * )  A
2m i m*c

The density is also gauge independent.

2
s  r  .

Now we assume that

 (r)   (r) ei (r ) .

We note that

q*  q* 2
 * (p  A ) (r )   * [ei (r )  (r ) ]  A (r )
c i c
 q* 2
  (r ) e i (r ) [i (r ) ei (r ) (r )  e i ( r ) (r ) ]  A (r ) .
i c
*
2 q
  (r ) [ (r )  A]  i (r )  (r )
c

The last term is pure imaginary. Then the current density is obtained as

q * 2 q* 2
Js  *
 (    A)  q*  v s ,
m c

or
q*
   A  m* v s .
c

Since

61
 q*
π  m vs  p  A ,
*

we have

q*
p A  m * v s    .
c

Note that Js (or vs) is gauge-invariant. Under the gauge transformation, the wave function
is transformed as

iq * 
 ' (r )  exp( ) (r ) .
c

This implies that

q* 
   '   ,
c

Since A'  A   , we have

q*
J s '  ( ' A' )
c
q*  q*
 [(  )  ( A   )] .
c c
*
q
 (  A )
c

So the current density is invariant under the gauge transformation.

37. London's equation

We start with

q*
p     A  m* v s ,
c

2
where p is the canonical momentum. We assume that   ns is independent of r. Then
we get

62
 2
J s  q *  v s  q * ns v s .

Using these two equations, we get

q* m*
p    A  * J s .
c q ns

Suppose that p   =0, which means that the phase  is independent of r. Then we
have a London’s equation,

q *2 ns
Js   A.
m *c

From this equation, we get

2 2
q * ns q * ns
  Js   *   A   * B .
mc mc

Using the Maxwell’s equation

4
B  Js , and B  0,
c

we get

2
4 4n q*
  (  B)    J s   *s 2 B ,
c mc

where

ns   = constant
2
(independent of r)

m*c 2
L 2  2 , (penetration depth).
4ns q*

Then

63
 1
  (  B)  (  B)  2B   B,
L 2
or

1
 2B  B.
L 2

Inside the system, B becomes zero, corresponding to the Meissner effect.

((Note)) The London penetration depth

m*c 2
L 2  2
4ns q*

In this equation, we put

n
e*  2 e , m*  2m , ns 
2

Then we get

( 2m ) c 2 mc2
L  
n
4 (2e) 2 4ne2
2

This equation can be used to estimate the order of the magnitude in the penetration depth.

38. Penetration depth and the surface current

We assume that the magnetic field is applied along the z direction.

B=(0, 0, Bz(x))

d 2 Bz ( x ) 1
 2 Bz ( x )
dx 2
L

The solution of this equation is given by

64
 x
Bz ( x)  Bz ( x  0) exp( )
L

N-phase S-phase

x
l
Fig. The distribution of the internal magnetic field B near the boundary between the
normal phase and the superconducting phase.  is the penetration depth. The
direction of B is into the page.

where L is the penetration depth. Then the surface super current Js is given by

c c B ( x ) c x
Js    B  ( )e y z  e y exp(  ) .
4 4 x 4 L L

B z ( x )
Since <0 for x>0 (inside the S phase), the surface supercurrent flows along the
x
positive y axis only in the region over the penetration depth  from the surface.

65
Fig. The distribution of the magnetic induction B(x) (along the z axis) and the current
density (along the y axis) near the boundary between the normal phase and the
superconducting phase. The plane with x = 0 is the boundary.

Fig. Suppose that a magnetic field is applied to the axis direction of the
superconductor [vacuum or N phase (green) and S phase (pink)]. The supercurrent
(red) flows around the surface of the S-phase (over the limited region of
penetration depth from the surface) in the clock-wise direction. The external
magnetic field is cancelled out by the magnetic field due to the supercurrent,
leading to the perfect diamagnetism (Meissner effect).

66
Fig. For T<Tc, the surface current flows near the surface, leading to the cancellation of
the external magnetic field. Meissner effect.

39. Flux quantization

We start with the current density

q * 2 q* 2
Js  *
 (    A)  q*  v s .
m c

2
Suppose that ns   =constant. Then we have

m* q*
  J s  A,
q*ns c

or

m* q*
   dl  q*ns*  J s  dl  c  A  dl .

The path of integration can be taken inside the penetration depth where J s =0.

67
 q* q* q* q*
    d l  c   A  d l  c   (  A )  d a  c   B  d a  c   ,

where  is the magnetic flux. Then we find that

q*
   2  1  2n  ,
c

where n is an integer. The phase  of the wave function must be unique, or differ by a
multiple of 2 at each point,

2c
 n.
q*

The flux is quantized. When |q*| = 2|e|, we have a magnetic quantum fluxoid;

2c ch
0   = 2.06783372 × 10-7 Gauss cm2
2e 2e

40. Parameters related to the superconductivity

Here the superconducting parameters are listed for convenience.

4 2
Hc  4 
2 4
Thermodynamic field Hc  ,


c
Hc 
2 q* 

Order parameter   2  ns*   /  .

2 2
Hc Hc
The parameters   ,  .
4ns *2
*
4ns
2c
Quantum fluxoid 0  .
q*

68
 m *c 2  m*c 2
Magnetic field penetration depth  2
  2 
4ns q*
*
4q * 

2ns 
*

Coherence length     
2m*  H c m*

 cm*  cm* H c
Ginzburg-Landau parameter     .
 2  q* 2 2ns q*
*

1
 for type-I superconductor
2
1
 for type-II superconductor
2

Note:

*
q *  2e (>0) m  2m ,
*
ns  ns / 2 .

41. Critical fields Hc1 and Hc2 for the type II supercinductor

0  Hc
Lower critical field (type II) H c1  ln( )  ln( ) .
4 2
 2

0
Upper critical field (type II) H c2  .
2 2

Surface-sheath field H c3  1.695H c 2 .

Relations between Hc, Hc1, and Hc2

 1  2 q* H c
 ,  .
 0 2 2H c 2 c

 2 22 H c
  2 2H c ,  .
2 0 0

69
 0 Hc 1
2H c  ,  .
2 Hc2 2

42. Vortex structure

0 x
Fig. Plot of the magnetic induction B(r )  K0 ( ) and the order parameter
2 2

 (x) /  as a function of x/. We use the GL parameter as  = / = 4.

  x 
f   tanh 
  2 

0 r
B( r )  K0 ( )
2 2


where Kn (x) is the modified Bessel function of the second kind.

70
For <r«,

0 r
B( r )  [ln( )  0.115932] .
2 2


For r»,
0  r
B (r )  exp( ) .
22 2r 

The magnetic flux (r) inside the circle with a radius r is

r r
0 r'
 (r )   2r ' dr ' B (r ' )   r ' dr ' K (  )
 2 0
0 0
r/
.
r r
 0  xdxK ( x)   [1   K (  )]
0
0 0 1

When r ∞, (r )  0

In Fig, for simplicity we assume that

B(r ) r
 K0 ( ) for r>
0 
2 2

and

B(r ) 
 K 0 ( ) =constant for r<.
0 
22

The current density is calculated as

c c d
J   B  e B( r ) .
4 4 dr

The vortex line energy is given by

71
 2
 0    
1    K 0 ( ) K1 ( )
 4    

The interaction between vortex lines is

2
 0 r
U 12  0 B1 (r2 )  K 0 ( 12 ) ,
4 8 
2 2


where r12  r1  r2

Fig. One dimesional Abrikosov-type structure of vortices.. The red line shows the
distribution of the normalized order parameter and the blue line shows the
distribution of the magnetic induction (the internal magnetic field). The center of
each vortex is denoted by the green lines.

43. Experimental verification of magnetic flux quantization (1961)

Two experimental measurements were published in 1961, by Deaver and Fairbank,
and by Doll and Näbauer. In both these experiments the flux trapped by a
superconducting hollow cylinder was found to be quantized, and the accuracy of the two
experiments was comparable. Doll and Näbauer found that the measured flux quantum
differed from London's work by as much as 60%, but Buyers and Yang were at Stanford,
and the Stanford group found that they agreed with the theoretically expected value of
hc/2e to within 20 %. (D.J. Thouless, "Topological quantum number in norelativistic
physics, " World Scientific, 1998).

72
 3 F0

6 F0

73
 8 F0

Fig. The magnetic flux lines (fluxoid) threading in the superconducting ring. Magnetic
flux is quantized. n = 1, 2, 3,..... Note that n = 3, 6, and 8 in these figures. The
direction of the supercurrent is denoted in the figure.

The magnetic field B can be measured inside the superconducting ring (with a radius r)

BA  n  0

or

n 0
B
r 2

where  0  2.06783366752 x 10-7 G.cm2. Suppose that r = 10-4 cm. Then we get

B = 6.582 n [Gauss] (n = 1, 2, 3,...).

which can be easily measured in the laboratories.

44. Cooper pair (1956)

44.1 Discussion by de Gennes
Even a weak attraction can bind pairs of electrons into a bound state. The Fermi sea
of electrons is unstable against the formation of at least one bound pair,
(i) regardless of how weak the interaction is,
(ii) so long as it is attractive.

74
((Assumption))
The extra electrons are added to a Fermi sea at T = 0 K, with the stipulation that the
extra electrons interact with each other but not with those in the sea except via the Pauli
exclusion principle.

We expect the lowest energy state to have zero total momentum. Two electrons have
equal and opposite momentum. The orbital wave function

 0 ( r1 , r2 )   g k exp( ik  r1 ) exp( ik  r2 )

k ,

which needs to satisfy the condition of anti-symmetry of the total wavefunction with
respect to the exchange of the two electrons.

exp(ik  r1 ) exp(ik  r2 )  cos[k  (r1  r2 )]  i sin[k  (r1  r2 )] .

First term (even function): anti-symmetric single spin function

1
 ( 0, 0 )  [ 1
 2
  1
 2.
2

Second term (odd function): symmetric triplet spin function

75
  (1,1)   1  2

1
 (1,0)  [ 1
 2  1
 2
2
 (1,1)   1
 2

Note that

D1/ 2  D1/ 2  D1  D0 . (angular momentum addition .s rule)

Anticipating an attractive interaction, we expect the singlet coupling to have lower

energy, because the cos[k  (r1  r2 )] gives a larger probability amplitude of the electrons
to be near each other. Thus we consider a two electron singlet wavefunction,

 0 (r1 , r2 )   g k cos[ k  (r1  r2 )]  (0,0)

k .

Here we follow the method which is used by de Gennes. He uses

 0 (r1 , r2 )   g k exp[ ik  (r1  r2 )]

k ,
where g k is the probability amplitude for finding one electron with k and the other
electron in the state -k. Note that

gk  0 for k<kF, (Pauli principal)

since the states with k<kF are already occupied.

Schrödinger equation:

2
( 1   2 ) 0 ( r1 , r2 )  V ( r1 , r2 ) 0 ( r1 , r2 )  E 0 ( r1 , r2 ) ,
2 2

2m

with

E  2 F   (the energy eigenvalue of two systems)

76
 2
F  2
kF .
2m

Since

2 2
(1   2 ) 0 (r1 , r2 )   g k (
2 2 2 2
 )(1   2 ) exp[ik  (r1  r2 )]
2m k 2m
2

m
kk
2
g k exp[ik  (r1  r2 )]

V ( r1 , r2 ) 0 ( r1 , r2 )   g k 'V ( r1 , r2 ) exp[ik '( r1  r2 )]

k'

  g k ' exp[ik '( r1  r2 )] Vk  k ' exp[i ( k  k ' )  ( r1  r2 )]

k' k k'

 Vk , k ' g k ' exp[ik  ( r1  r2 )]

k ,k '

we have

2 2
k g k  Vk , k ' g k '  (    2 F ) g k , (Bethe-Goldstone equation)
m k'

where we assume that the potential energy can be described by the Fourier series

V (r )  V
k k'
k ,k ' e i ( k  k ' ) r  V
k k'
k k' e i ( k  k ')r  Vq e iqr
q

Then we have

2 2
Vk ,k ' g k '  (    2 F 
k' m
k ) g k  (    2 k ) g k ,

with

2 2
k  k F .
2m

((Assumption))
For simplicity, for

77
 2 2 2 2
F  k   F   D . F  k '   F   D ,
2m 2m

we assume that

Vk ,k '  V , (V>0 for attractive and V<0 for repulsive interactions)

otherwise, Vk ,k '  0 .

This interaction is attractive and constant in an energy band  D above the Fermi level.
Then we get

(    2 k ) g k   V k , k ' g k '  V  g k ' .

k' k'

From this we obtain

V
gk 
  2 k
g
k'
k' ,

or

1
g
k
k V
k
 gk' ,
(  2 k ) k '

or

1 1
  .
V k (   2 k )
k kF

((Density of states)) N(); the density of states per spin

1
N ( )d  4k 2 dk .
( 2 ) 3

Noting that

78
 2m
k  F ,
2

2m d
dk  ,
 2  F
2

we get

3/ 2
1  2m  d
N ( ) d  4  2  (   F )
( 2 ) 3
  2  F
3/ 2
1  2m 
      F d
4 2   2 

or

3/ 2
1  2m 
N ( )     F .
4 2   2 

 D
1 1
V
 
0
2 N ( )d (
2  
).

If we assume  D   F , N() can be considered as constant and replaced by its value

N(0) at the Fermi level.

 D
 1     2 D 
1  2VN (0) 
0
d    VN (0) ln
 2     
.


In the limit of weak interaction (N(0)V<<1),

1 2D
exp[ ]  1 >>1.
VN (0) 

Then we get

1
  2 D exp[ ].
VN (0)

79
The energy of the Cooper pair relative to the state where the two electrons are at the
Fermi level.
There exists a two-electron bound state of energy >0. If we start from a free electron
gas, and turn on the interaction V, we predict that electrons will group themselves in pairs
giving up energy to the external world. The normal state is thus unstable. This means the
existence of a bound state for the two added electrons.
Cooper showed that, where the interaction is attractive, the system energy is reduced
by pairing. Therefore, the Fermi sea of single electrons is unstable, since any perturbation
that moves two electrons above F will lower the system energy. Note that the Cooper
model is not a model for the superconducting ground state, and that

1 1
  2 D exp[ ]  2k B  D exp[ ].
VN (0) VN (0)

44.2 Original discussion by Cooper

L.N. Cooper, Phys. Rev. 104, 1189 (1956).
J.R. Schrieffer, Theory of Superconductivity (W.A. Menjamin, New York, 1964).

Leon N. Cooper: Superconductivity and Beyond (25 th Army Science Conference,

Orlando, Florida, November 28, 2006).

The method is more direct compared to the above theory. We start with

(    2 k ) g k   V k , k ' g k '  V  g k ' ,

k' k'

where -V<0 for the attractive potential. Here we introduce E (the energy eigenvalue),

E  2 F   , k   k   F

Then we have

( E  2 k ) g k   g k 'V k , k '  V  g k '

k' k'

From this we get

V
gk  
E  2 k
g
k'
k' ,

80
or

1
g
k
k  V 
k E  2 k
g
k'
k' ,

or

1 1
   .
V k ( E  2 k )
k kF

where E is the energy eigenvalue of the system

44.3 Example
First we consider the simple example. Suppose that

k
N1 = 10, 2 k  ( 2  )
N1

where k = 1, 2, ..., 10. In other words,

2 k  2, 2.1, 2.2, 2.3,…, 2.8, 2.9, 3.

Then we get

N1 N1
1 1 1 1 1 1 1
  f (E)       ...  
k 1 E  2 k
k
V k 1
E  ( 2  ) E  2.1 E  2.2 E  2.9 E  3
N1

We make a plot of f(E) as a function of E.

81
 f E

40

20

0 E
0.5 1.0 1.5 2.0 2.5 3.0

- 20

- 40

Fig. Plot of f(E) as a function of E. The dashed line (-1/V = 40) for the repulsive
interaction. The dashed line (-1/V = -20) for the attractive interaction. The last
crossing on the left is the coherent state, split from the continuum by an energy
gap, displaying the essential singularity in the coupling constant.

1
What is the solution of  f (E) ?
V
(i) When V<0 (repulsive) the solution of E is always larger than 2.

(ii) When V>0 (attractive), we have a solution of E, which is lower than 2. This means that
the two electrons forms a Cooper pair (bound state).

((Note))
Leon N. Cooper: Superconductivity and Beyond (25 th Army Science Conference,
Orlando, Florida, November 28, 2006).

“Allow me (Cooper himself) to show you a page from my notes of that period with the
pair solutions as they first appeared to me.”

82
Fig. From the note of Cooper around 1956.

44.4 The derivation of energy gap

1 1
 
V k (  kF ) 2 k  E
 F   D
1
2

 2  E
N ( )d ,
F

 F   D
1
 2 N ( 0)

 2  E
d
F

or

1  2  2D  E 
 N (0) ln(2  E ) | FF  D  N (0) ln F  ,
V  2 F  E 

or

1 2  2D  E 2D
exp[ ] F  1
N (0)V 2 F  E 2 F  E

or

83
 1 2  D
exp[ ] 1  ,
N (0)V 2 F  E

where N ( F ) is the density of states per spin (per unit volume). Then we have

2  D
2 F  E  .
1
exp[ ] 1
N (0)V

In the weak coupling ( N (0)V <<1), we have

1
E  2 F  2 D exp[  ].
N (0)V
Note that

1
 0  E  2 F  2 D exp[  ],
N (0)V

is the bound state energy for the Cooper pair (V>0; attractive interaction).

44.5 Eigenvalue problem

Degenerate matrices have some very surprising properties. We return to the original
eigenvalue problem

( E  2 k ) g k   g k 'V k , k '  V  g k '

k' k'

Here we assume that

E  2 k  V     V  E1

84

is independent of k since k is nearly equal to F (degenerate states).

For simplicity we consider the matrix (N x N) for the example. The eigenvalue problem is
given by

 0 V V . . . . . V  V  g1   g1 
    
 V 0 V . . . . . .  V  g 2   g2 
 V V 0 . . . . . .  V  .   . 
    
 . . . 0 . . . . . .  .   . 
 . . . . . . . . . .  .   . 
 .    E1  
 . . . . . . . . . .  .   . 
 . . . . . . . . . .  .   . 
  
 . . . . . . . 0 V .  .   . 
    
 V . . . . . . V 0  V  g n 1   g n 1 
 V V V . . . . . V 0  g n   g 
  n 

where -V<0 (attractive interaction). When diagonalized, what happens to the energy
eigenvalue? The schematic diagram is shown below.

Fig. Schematic diagram of the energy levels.

The lower level is a coherent superposition of all the original states and is separated from
them by an energy gap.

44.6 Numerical calculation using Mathematica

We solve the eigenvalue problems for the (24x24) matrix with V = 0.1 (-V<0 means
attractive interaction). The results are as follows.
(i) The ground state (1 state) with the lowest energy; -(24-1) x 0.1 = -2.3
85
(ii) The degenerate states (23 states) with the energy 0.1.
(iii) The ground state is a coherent states; linear combination of the state with the same
amplitude.

________________________________________________________________________
((Mathematica))

Clear"Global`"; N1  24; a  0.1; A i_, j_ :  a;

Bi_, j_  a KroneckerDeltai, j; M1  Table A i, j, i, 1, N1, j, 1, N1;
M2  TableBi, j, i, 1, N1, j, 1, N1; M  M1  M2; eq1  Eigensystem  M ;
eq11; fn_ : Tablek, Normalizeeq12, n k, k, 1, N1;
gn_ : ListPlotfn  , PlotStyle  Thick, Hue0.08 n , Joined  True,
PlotRange  0, N1,  1, 1,
Epilog  TextStyle"E"  ToString eq11, n , Black, 12, 12, 0.6,
TextStyle"n"  ToString n , Black, 12, 12, 0.9;
K1  Tableg m ,  m, 1, 8; K2  Tableg m ,  m, 9, 16;
K3  Tableg m ,  m, 17, 24;
________________________________________________________________________

86
 1.0 1.0
n= 1 n=2
0.5 E=-2.3 0.5 E=0.1

0.0 0.0
5 10 15 20 5 10 15 20
- 0.5 - 0.5

- 1.0 - 1.0

1.0 1.0
n= 3 n=4
0.5 E=0.1 0.5 E=0.1

0.0 0.0
5 10 15 20 5 10 15 20
- 0.5 - 0.5

- 1.0 - 1.0

1.0 1.0
n= 5 n=6
0.5 E=0.1 0.5 E=0.1

0.0 0.0
5 10 15 20 5 10 15 20
- 0.5 - 0.5

- 1.0 - 1.0

1.0 1.0
n= 7 n=8
0.5 E=0.1 0.5 E=0.1

0.0 0.0
5 10 15 20 5 10 15 20
- 0.5 - 0.5

- 1.0 - 1.0

87
 1.0 1.0
n= 9 n=10
0.5 E=0.1 0.5 E=0.1

0.0 0.0
5 10 15 20 5 10 15 20
- 0.5 - 0.5

- 1.0 - 1.0

1.0 1.0
n=11 n=12
0.5 E=0.1 0.5 E=0.1

0.0 0.0
5 10 15 20 5 10 15 20
- 0.5 - 0.5

- 1.0 - 1.0

1.0 1.0
n=13 n=14
0.5 E=0.1 0.5 E=0.1

0.0 0.0
5 10 15 20 5 10 15 20
- 0.5 - 0.5

- 1.0 - 1.0

1.0 1.0
n=15 n=16
0.5 E=0.1 0.5 E=0.1

0.0 0.0
5 10 15 20 5 10 15 20
- 0.5 - 0.5

- 1.0 - 1.0

88
 1.0 1.0
n=17 n=18
0.5 E=0.1 0.5 E=0.1

0.0 0.0
5 10 15 20 5 10 15 20
- 0.5 - 0.5

- 1.0 - 1.0

1.0 1.0
n=19 n=20
0.5 E=0.1 0.5 E=0.1

0.0 0.0
5 10 15 20 5 10 15 20
- 0.5 - 0.5

- 1.0 - 1.0

1.0 1.0
n=21 n=22
0.5 E=0.1 0.5 E=0.1

0.0 0.0
5 10 15 20 5 10 15 20
- 0.5 - 0.5

- 1.0 - 1.0

1.0 1.0
n=23 n=24
0.5 E=0.1 0.5 E=0.1

0.0 0.0
5 10 15 20 5 10 15 20
- 0.5 - 0.5

- 1.0 - 1.0

Fig. Pattern of superposition in the amplitudes of thenormalized wavefunction.

E (= E1 = -+V) is the eigenvalue. n is the mode number ( n = 1 - 24). n =
1 corresponds to the ground state with E = -2.3. n = 1 is the coherent
ground state. -V = -0.1 (attractive interaction). In this calculation we mean
E by E1.

46.7 Classification of the superconductivity : s-wave, p-wave, and d-wave

(a). Introduction
We consider two fermions which are subject to a central field. In this case the wave function
can be described by

 ( r ' , r" )  r ' , r"    ( r  r 'r" ) .

This wave function can be decomposed into a radial part and a spherical-harmonics part, i.e.

89
  ( r )  r n, l , m  Rnl ( r )Yl m ( , )

Note that exchanging the two particles is equivalent to inverting the vector r ' r" (i.e. changing
its sign). With such an inversion, the spherical harmonics undergo the transformation

Yl m ( , )  ( 1) l Yl m ( , )

Suppose that the fermion is an electron with spin s = 1/2. In this case the total spin is S = 1
(triplet, symmetric state) and S = 0 (singlet, anti symmetric state). The wave function should be
antisymmetric under the exchange of the position. This requires the conditions for l = even and S
= 0, and l = odd and S = 1.

b. Parity operator and permutational operator

Pˆ12 r ' , r"  Pˆ12 r ' 1 r" 2  r" 1 r ' 2 ,

When

Pˆ , Hˆ   0
12

we have the simultaneous eigenket  ;

P̂12     , Hˆ   E 

Then we have

r ' , r" Pˆ12    r ' , r" 

or

r " , r '    r ' , r" 

We introduce the relative co-ordinate

r  r '  r"

r  r ' , r" ,  r  r" , r '  ˆ r

Then we have

r  r

90
or

r ˆ    r 

or

ˆ    

This implies that

Pˆ12  ˆ

We consider the relative motion for the two electrons. The Hamiltonian for these systems, can be
described by that for one electron with the reduced mass.

  nlm

We note that

ˆ l , m  ( 1) l l , m

Then the orbital state with even integer of l has the even parity, while the orbital state with odd
integer of l has the odd parity.

c. Classification of the symmetry for superconductivity

When we take into account of the spin states, the symmetry of the resultant eigenket should
be anti-symmetric. In other words,

S = 0 (antisymmetric) with l = even.

S = 1 (symmetric) with l = odd.

S-state (BCS Cooper pair)

1
l =0, S = 0 S j=1 (3 states)

P-state (liquid 3He superfluidity)

3
l =1, S = 1 P j = 2, 1, 0 (9 states; 5+3+1=9)

D-state (high Tc superconductor)

1
l =2, S = 0 D j = 2 (5 states)

91
The onset of superconductivity occurs with the condensation of electron pairs. These electron
pairs, called the Cooper pair, can be in a state of either total spin S=0 (spin singlet) or 1 (spin
triplet). Being fermions, electrons anticommute. Therefore the antisymmetric spin-singlet state is
accompanied by a symmetric orbital wave function (even parity) and vice versa, in order to
preserve the anti-symmetry of the total wave function.

________________________________________________________________________
45. BCS Hamiltonian
We start with a pairing Hamiltonian

H  N    k ak ak   Vk , q ak ak  a q  aq  .

k , k ,q

Fig. Two pairs of electrons (q, ; -q, ) and (k, ; -k, ), which are coupled with the
electron phonon interaction (Vk,q). The phonon has a momentum (k - q).

where  is the chemical potential;  = F (the Fermi energy).

Here we use the mean field theory. The expectation value is all one needs to
understand the nature of the ground state. This can be accomplished by setting

ak ak   (ak ak   ak ak  )  ak ak 

92
 a k ak   (a k ak   a k ak  )  a k ak 

by treating the first term as small compared to the second term (its average), and
expanding to first order in the small quantities. For simplicity we use

*
bk  a k ak  , bk  ak ak 

Using such an approiximation, we get

* *
akak a q aq  [(akak   bk )  bk ][(a q aq   bq )  bq ]
* * *
 bk (a q aq   bq )  bq (akak   bk )  bk bq
* * * *
 bk a q aq   bq akak   bk bq  bk bq  bk bq
* *
 bk a q aq   bq akak   bk bq
Then we have

H   F N    k ak ak
k ,

  Vk , q [bq a k  ak   bq ak ak   bk bq ]

* *

k ,q

   k ak ak   ( k ak ak    k a k  ak    k bk )

* *

k , k

where

k   k   F .

The gap parameter is defined by

 k  Vk ,q aq aq  Vk ,q bq ,

q q

or

 k   Vk , q aq aq    Vk , qbq .

* *

q q

The above Hamiltonian can be diagonalized using the Bogoliubov transformation,

93
   k    uk  vk  ak  
     *    ,

 k    vk u k  ak  

where we assume that uk is real and

2
| uk |2  | vk |2  uk  | vk |2  1 .

This condition can be derived from the commutation relation,

  
[ k  ,  k  ]   k  k    k   k   1 .

Note that

 
[ k  ,  k  ]  [uk ak   vk ak  , uk ak   vk ak  ]
* *


| uk |2 [ak  , ak  ]  | vk |2 [ak  , ak  ]
* * 
 uk vk [ak  , ak  ]  uk vk [ak  , ak  ]
2
| uk |2  | vk |2  uk  | vk |2  1

The inverse Bogoliubov transformation is given by

1
 ak    uk  vk    k    uk vk   k  
     *          .
 a  k    vk u k     k     vk
*
u k  k  

Using the Bogoliubov transformation, the above Hamiltonian can be rewritten as

H  N   k (uk k   vk   k  )(uk k   vk k  )

 *

   k (vk  k   uk k  )(uk  k   vk k  )

* 

  [ k (uk k   vk   k  )(uk k   vk  k  ) 

 * *

  k (uk  k   vk k )(uk k   vk k  )   k bk ]

* *

where  k   k . In order to diagonalize the Hamiltonian it is required that the coefficients

of  k  k  and  k    k  should be equal to zero. Then we have
 

94
 2
2 k uk vk   k vk*   k uk2  0 ,
* *

and its conjugate

2 k uk vk  k vk  k uk2  0 .
* 2

We use the second quadratic equation.

2
 uk   k  uk    k * 
   2     0.
 vk   k  vk    k 

The solution of this equation is obtained as

u k  k  Ek
 ,
vk k

where

2
Ek   k   k .
2

We will show later that the plus sign should be chosen to get the minimum energy. Then
the Hamiltonian H can be rewritten as

H  E g   Ek ( k   k  k   k  ) ,
 

where

Eg   [2 k vk   k uk vk   k uk vk   k bk ]
2 * * *

Eg is the ground state energy. The above Hamiltonian clearly indicates the energies of the
excitations above the ground state. Since

uk  k  Ek

vk k

95
uk and  k  Ek are real, which means that the phase of k is the same as that of vk. We
assume that

 k   k eik

 k  Ek 2  1  2 Ek ( Ek   k )
2 2
u k  vk 1
2
 2
 2 2
vk vk k k

2
Using the normalization condition ( uk  | vk |2  1 ), we have

2 ( Ek   k )
vk 
2 Ek

The energy of the ground state Eg can be rewritten as

Eg   [2 k vk   k uk vk   k uk vk   k bk ]
2 * * *

  [2 k vk  2 vk ( k  Ek )   k bk ]
2 2 *

  [2 vk Ek   k bk ]
2 *

  [ ( Ek   k )   k bk ]
*

where the double signs are in order. In order to get the minimum value of Eg, we need to
choose

2 ( Ek   k ) 1 
vk   (1  k ) ,
2 Ek 2 Ek

leading to

Eg  [ Ek  k   k bk ]
*

which is the ground state energy. Then we also have

96
 2 1 
uk  (1  k ) ,
2 Ek

and

2
2 2 1  2 
u k vk  (1  k 2 )  k 2 .
4 Ek 4 Ek

The condensation amplitude F is given by

k
F k  u k vk  .
2 Ek

We note that the quasi-particle occupation numbers follows the Fermi-Dirac function,

  1
  k   k     k   k   f ( E k )  ,
exp(  E k )  1

1
where f (Ek ) is the Fermi-Dirac distribution function and   . The parameter vk is
k BT
related to

u k  k  Ek
 ,
vk k

or

k  e i k
vk  uk  uk k .
 k  Ek  k  Ek

The phase of k is the same as that of vk.

97
 ED

2.0

1.5
ED

1.0

xD
0.5

x D
-2 -1 1 2

Fig. Excitation spectrum.  k   .

uk 2
0.8

0.6

0.4

0.2

vk 2
x D
-3 -2 -1 1 2 3

Fig. Plot of uk2 and |vk|2 as a function of /.  k   . In a normal phase, the
momentum distribution (vk|2) drops discontinuously at  = 0 (k = kF). In the
superconducting phase, this drop is smeared out on an interval k ≈1/0, where 0
is the coherence length.

0.5

0.4

0.3

0.2 vk u k

0.1

x D
- 15 - 10 -5 5 10 15

98
Fig. Plot of the condensation amplitude Fk = uk |vk| as a function of /.  k  

46. The gap equation

We start with the energy gap equation at finite temperature given by

 k  Vk , q a q aq 
q

,
where

q  Eq
a q  aq   tanh( )
2 Eq 2

Note that

a q  aq   (vq q  uq  q  )(uq q   vq q  )

 uq vq   q  q   uq vq  q q 

 uq vq [1   q   q  ]  uq vq  q q 

 uq vq [1  2  q q  ]
 uq vq [1  2 f ( Eq )]
2
uq  q  Eq q Eq
 tanh( ) tanh( )
 q  Eq 2 2 Eq 2

where we use

  q  q   q q  0 ,

  
[ q  ,  q  ]   q  q    q   q   1 .

Then we get

99
 q  Eq
 k   Vk ,q tanh( ).
q 2 Eq 2

In order to simplify this equation, we assume that

Vk ,q  V if k  D and  q   D .

Vk , q  0 otherwise.

where  D  k B  and  is the Debye temperature. Then  k becomes independent of k

and the above equation reduces

  Eq
 V tanh( ).
q 2 Eq 2

The final form of the gap equation is

1 1 Eq 1 
 )
2
tanh( tanh(  q
2
 q ) ,
V q 2 Eq 2 q 2  2  
2 2
q q

We assume that  q   .

(1) The energy gap at T = 0 K


1 D
d
 N (0) 
V   D 2   
2 2

 D
d
 N (0) 
0  2  2
D  2  D 
2

 N (0) ln[ ]

2D
 N (0) ln[ ]


where N (0) is the density of states per spin at the Fermi surface.

100
(2) Critical temperature Tc
The energy gap is equal to zero at the critical temperature Tc.

0 at T = Tc.

This equation can be rewritten as

D
1 d 
N (0)V
 
0

tanh(
2k BTc
)

D /( 2 k BTc )
d
 0

tanh( )

D /( 2 k BTc )
ln d

D /( 2 k BTc )
 [ln  tanh( )] 
cosh 2 ( )
0
0
,

ln d
 [ln  tanh( )]0D /( 2 kBTc )  
0
cosh 2 ( )
 D  D 
 ln( ) tanh( )  [ln( )   ]
2k BTc 2k BTc 4
 D 2e 
 ln( )
k BTc 

where  = 0.577216 is the Euler's constant. Here we use the fact that

 D 

k BTc Tc

is very large. For Al (type-I superconductor), in fact, Tc = 1.14 K and  = 428 K, The n
we have

 428  D 
  375.439>>1, tanh( )  tanh( ) 1
Tc 1.14 2 k BTc 2Tc

Then we get

1 D
 ln( )  0.81878
N (0)V 2k BTc

101
 
since  [ln( )   ]  0.81878 . This can be rewritten as
4

 D 1
 2 exp[ ] exp(0.81878)
k BTc N (0)V
1
 0.881939 exp[ ]
N (0)V

or

1
k BTc  1.13387 D exp[ ].
N (0)V

Since

1
 0  2 D exp[  ]
VN (0)

we have

2 0 4
  3.52774
k BTc 1.13387

or

20  3.52774kBTc = 0.303997 Tc[K] (meV)

2 0 is the energy gap for the breaking up of 1 Cooper pairs (two electrons).

102
 2D0meV
6 BCS theory
5

Tc K
5 10 15 20

Fig. 20 [meV] vs Tc [K], predicted from the BCS theory.

((Note))
2 0
The ratio 3.52774 (  ) is a 'universal' value independent of Tc and other variables. It
k BTc
should be noted , however, that this value is obtained within the so-called weak coupling
approximation in the BCS theory. The actual value of the ratio for ordinary
superconducting metals with low Tc is close to this weak coupling value, but for metals
with higher Tc, this ratio deviates from the universal constant.

Tc 20/kBTc C/Cn
__________________________________________
BCS 3.52774 1.43
Al 1.18 3.53 1.43
Cd 0.52 3.44 1.32
Sn 3.72 3.61 1.60
Hg 4.15 3.65 2.37
Pb 7.20 3.95 2.71
Nb 9.25 3.65 1.87

47. Temperature dependence of energy gap

We start with the energy gap equation

103
  D
1 d  2  2
N ( F )V


 2  2  2
tanh(
2 k BT
)
D

 D 
d  D
1  2  2 1 
 
  D
2
tanh( )  [
2k BT D 2  2  2
tanh(
2 k BT
)
2
tanh(
2 k BT
)]


 D 2e  D
1  2  2 1 
 ln( )  [ tanh( ) tanh( )]
k BT    D 2   
2 2 2 k BT 2 2 k BT
 D
 D 2e  1  2  2 1 
 ln(
k BT 
) 
0
[
  2 2
tanh(
2 k BT
)

tanh(
2 k BT
)]

Using the expansion formula


8x
tanh( x )  
n  0 4 x  ( 2 n  1) 
2 2 2

we get

 D
1  2e   4  
1
N (0)V
 ln( D
k BT 
)   
 k BT 
 
0 n 0
[
  2  2 
2

4   (2n  1) 2  2
 2k BT 
 
1
 2
]
  
4   (2n  1) 2  2
 2k BT 
 D
D 2e  
d
 ln( )  42 k BT  
k BT  n 0 0 [k B T (2n  1) 2  2   2 ]2
2 2

Here

 D  D / k BT
d 1 dx

0 [k B T (2n  1)    ]
2 2 2 2 2 2
 3 3
kB T 
0
[(2n  1) 2  2  x 2 ]2

1 dx
3 3 

k B T 0 [(2n  1)   x ]
2 2 2 2

1 1

k B T 4 (2n  1)
3 3 2 3

104
Then we have

1 D 2e 42 

1
N (0)V
 ln(
k BT 
) 2 2 2
4 k B T
 (2n  1)
n 0
3

D 2e 2 7
 ln( )  2 2 2  (3)
k BT   kB T 8

where


1 7
 (2n  1)
n0
3
  (3) .
8

and (3) = 1.2206. Using the relation

1 D 2e
 ln( ),
N (0)V kBTc 

we have

 D 2e   2e  2 7
ln( )  ln( D )  2 2 2  (3) ,
k BTc  k BT   kB T 8

or

T 2 7
ln( )   2 2 2  (3) .
Tc  kB T 8

The energy gap is obtained as

1/ 2
 8  T T
(T )  k BTc   ln( c )
 7 (3)  Tc T
1/ 2
 8  T
 k BTc   (1  )1/ 2
 7 (3)  Tc
T
 3.06326k BTc (1  )1/ 2
Tc

105
 T 1/ 2
in the vicinity of T = Tc. We make a plot of two functions (a) (1  ) and (b)
Tc
T T
ln( c ) as a function of T below Tc. It is found that these curves agree very well.
Tc T

D T 
D T = 0

0.4

0.3

0.2

0.1

T
t=
0.75 0.80 0.85 0.90 0.95 1.00 Tc

Fig. Plot of f (t )  (T ) /(3.06326kBTc ) as a function of reduced temperature t = T/Tc.

We show two curves, f(t) = (1  t )1/ 2 (red line) and t ln(1 / t ) (blue line).

((Mathematical note))
Suppose that

T
t 1 x
Tc

where x is close to x ≈ 1 (but x>0)

2
T Tc  1
 
 T ln( T )   t ln( t )
2

 c 
 t 2 ln(t )
 (1  x) 2 ln(1  x)
3 1
 x  x 2  x 3  0( x 4 )  x
2 3

Then in the vicinity of T = Tc (but T<Tc), we have

106
 2 2
T Tc  T Tc 
 ln( )   (1  t ) , or  ln( )   (1  t )1 / 2
T T   
 c  Tc T 
________________________________________________________________________
The energy gap is an order parameter for the superconducting transition. The critical
exponent  for the order parameter is  = 1/2 as is predicted from the mean field theory.
We note that

3.5277
0  k BTc .
2

Using this, we have

 (T ) 3.06326 k BTc T T
 2 (1  )1 / 2  1.73669 (1  )1/ 2 .
0 3.5277 k BTc Tc Tc

48. Density of states

The BCS spectrum is easily seen to have a minimum k for a given direction k on the
Fermi surface; k therefore, in addition to playing the role of order parameter for the
superconducting transition, is also the energy gap in the quasi-particle spectrum. To see
this explicitly, we can simply do a change of variables in all energy integrals from the
normal metal eigenenergies k to the quasiparticle energies Ek:

k       F ,

E   2  2 , or    E 2  2

N ( E)dE  N n ( )d  N n (0)d ,

since  ≈ 0. Then we get

d E
N ( E )  N n ( 0)  N n (0) ,
dE E 2  2

or

107
 E
N (E) E 
  .
N n ( 0) E  2
2
E
2

  1


ED

NENn 0
1

0
0 1 2 3 4

E N (E)
Fig. Plot of vs the normalized density of states . N(E) is the density of state
 N n (0)
per spin.

49. Evaluation of the energy of ground state, Eg

* * *
bk  uk vk , bk  uk vk  uk vk

Then we have

108
  ( E
k
k   k   k bk   k bk )   ( Ek   k   k uk vk   k uk vk )
* *

k
* * *

2 2
u 
  (  Ek   k  2 k k )
k  k  Ek
2

  (  Ek   k  k )
k Ek
2
  Ek
  ( k  k )
k Ek
k
   k (1  )
k Ek

Then we have

E g   ( Ek   k   k bk )
*

k
   k (1  )    k bk
*

k Ek k

We assume that k is independent of k. Then the energy gap equation can be expressed
by

 *
 bk   bk 
*
,
k V k V

Then the ground state energy Eg can be rewritten as

k *  2
E g    k (1  )    k (1  k ) 
k Ek V k Ek V

Here we assume that  is a positive real value.

2  2
E g ()    k    k
V k Ek k

k 2
     k   ( k   k )
k Ek k k

k 2
     k  En
k Ek k

109
where En is the energy of the normal state. Then the above equation can be rewritten as
 
2 D
2 D

E g ()   En   N (0)  d  N (0)   d

V   D  2  2 D
 D  D
2
 2 N (0) 
0
d
 2  2
 2 N (0)  d
0
 D /   D / 
x2
 2 N (0)   2 N ( 0)   xdx
2 2
dx
0 x 2  2 0
2 2
 D         
  N (0)2 1   D   N (0)2 sinh 1  D   N (0)2  D 
         
2 2
         
 N (0) [ D 1   D    D  ]  N (0)2 sinh 1  D ]
2

         
     
  N (0)2  D   N (0)2 sinh 1  D ]
     

Thus we have

2 2  D 2
E g ()   En   N (0) 
V  V
 
2 2
  N (0) 
2 V

or

2
E n  E g (  )  N ( 0)
2

We note that the energy gap  is related to the thermodynamic critical field Hc(0) as

2 1
N ( 0)  [ H c (0)] 2
2 8

((Note))
Here we use the formula

110
 x2 1
 1 x
dx  ( x 1  x 2  sinh 1 x) ,
2 2

D 2
d  D    
0  2  2
 sinh 1 (

)  ln[ D   D   1 ,
   

The energy gap equation id

 D  D
d d  D
1  N (0)V 
 2  
2 2
 N (0)V 
0  
2 2
 N (0)V sinh 1 (

),
D

or

1  D
sinh[ ] ,
N (0)V 

or

 D 1
  2 D exp[  ]  2  D .
1 VN ( 0 )
sinh[ ]
VN (0)

50. Physical meaning

The energy gap at T = 0 K is given by

3.5277
0  k BTc .
2

Then we get

[ H c (0)]2 N (0) 2 N (0) 3.5277

 0  ( k BTc ) 2  1.55558 N (0)( k BTc ) 2 .
8 2 2 2

Note that the unit of left-hand side is Oe2 = erg/cm3. Here we use the following relation
for the electronic specific heat in the normal state

Cen   eT ,

111
with

2 2
e  2
N ( 0) k B .
3

Then we have

[ H c (0)]2 3 2
 1.55558 e2 Tc .
8 2

Suppose that the atom (showing the superconductivity) has a molar mass M (g) and a
density  (g/cm3).

[ H c (0)]2 3  2
 1.55558 2  (10 3  10 7 )   ( mJ /( molK 2 ) Tc ,
8 2 M

where

M
is the volume (cm3) per mol; [mol/cm3].


3 
H c (0)  8  1.55558   (10 7  10 3 )   (mJ /( molK ) 2 ) Tc
2 2
M
3 
 8  1.55558   (10 7  10 3 )   (mJ /( molK ) 2 ) Tc
2 2
M

 243.76  (mJ /( molK ) 2 ) Tc
M

in the units of Oe.


(i) Hc(0) is proportional to  Tc .
M
(ii) If Tc is very high, then Hc(0) is also very high.

((Example))
(i) Pb,
= 11.34 g/cm3, M= 207.2 g/mol.
 = 2.98 mJ/(mol K ).
2
Tc = 7.193 K

112
 H c (0) =708.1 Oe (calculation)
H c (0) =803 Oe (experiment)

(ii) Sn
= 7.365 g/cm3, M= 118.692 g/mol.
 = 1.78 mJ/(mol K2). Tc = 3.722 K

H c (0) =301.5 Oe (calculation)

H c (0) =309 Oe (experiment)

(iii) Al

= 2.70 g/cm3, M= 26.9815 g/mol.

 = 1.35 mJ/(mol K2). Tc = 1.140 K

H c (0) =102.1 Oe (calculation)

H c (0) =105 Oe (experiment)

All the experimental data of g, Hc and Tc are obtained from the ISSP Kittel (8th-edition).
_______________________________________________________________________
51. BCS Ground state
According to BCS, the ground state can be expressed by

BCS   (u k  vk Bk )  vac ,

k

where Bk is a pairing operator, and is defined by

Bk  ak  ak  ,

and an operator Hermitian conjugate to the operator Bk is given by

Bk  ak ak    ak  a k .

The parameters uk and vk can be determined such that the BCS ground state is to be
minimized over the space of uk and vk. Another method (as is already discussed above) is
the diagonalization of the BCS Hamiltonian by using the Bogoliubov transformation. We
will show that these two methods are equivalent.

113
Since there is no quasi particle in the ground state, we have

 k  BCS  0 ,  k  BCS  0 .

 vac is the vacuum state. We use the Bogoliubov transformation;

  k    uk  vk  ak     k   uk  vk  ak  
* 
     *    ,     ,
     v uk  ak  
v
 k    k u k  ak    k    k

or

 ak    uk vk   k    ak   uk vk   k  
* 
        ,     .
a   v uk  k  
 k  
*
 a k     vk u k  k    k

Here we assume that u k*  u k (real number). vk is in general a complex number.

Then the above relations can be rewritten as

 k  BCS  (uk ak   vk ak  ) BCS  0 ,

and

 k  BCS  (uk ak   vk ak ) BCS  0 .

This means that BCS is a vacuum state for quasi-particles.

52. The formation of the BCS ground state by the successive addition of Cooper
pairs
The BCS state can be formulated from the successive addition of Cooper pairs to the
vacuum state.

 vac ;

114
 ky

kx
O

________________________________________________________________________
_
(uk1  vk1 ak1 ak1 )  vac ;

ky

k1 ,Æ>

kx
O

-k1 ,∞>

115
________________________________________________________________________
_
(uk2  vk2 ak2  ak2  )(uk1  vk1 ak1 ak1 )  vac ;

k2 ,Æ> k y

k1 ,Æ>

kx
O

-k1 ,∞>

-k2 ,∞>

________________________________________________________________________
(uk1  vk1 ak1 ak1 )(uk2  vk2 ak2  ak2  )(uk1  vk1 ak1 ak1 )  vac ;

116
 k2 ,Æ> k y

k1 ,Æ>

k3 ,Æ>
-k3 ,∞>
kx
O

-k1 ,∞>

-k2 ,∞>

BCS   (u k  vk Bk )  vac

k

117
53. The commutation relation

53.1. Fermion operators

The Fermion operators satisfies the anticommutation relation.

[ak , ak' ' ]  ak ak' '  ak' ' ak   k ,k '  , ' .

[ ak ak ' ' ]  [ ak , ak' ' ]  0 .

Using the vacuum state  vac , we have

[ak , ak ]  vac   vac ,

or

(ak ak  ak ak )  vac   vac ,

or
 
ak ak  vac  (1  ak ak )  vac   vac .

53.2 Paring operators Bk and Bk+

The application of Bk to  vac yields

Bk  vac  a k  ak   vac  0 ,

and


 vac Bk  0 .

Similarly, we get


Bk  vac  ak ak   vac  ak  ak  vac ,


The application of Bk Bk to  vac yields

118
 
Bk Bk  vac  a k  ak  ak ak   vac
 ak  a k  ak ak   vac
 ak  a k  ak  ak  vac
 ak  (1  ak  a k  )ak  vac
 ak  ak  vac  ak  ak  ak a k   vac
 ak  ak  vac
  vac

_____________________________________________________________________
54. Fermion operators and boson operators

(1)
[ak  , Bk ]  ak  ak ak   ak ak  ak 
 ak  ak  ak  ak  ak ak 
 ak  (1  2ak ak  )

(2)
[ak  , Bk ]  ak  ak ak   ak ak  ak 
  ak a k  ak   ak ak  ak 
  ak (1  2ak  a k  )

(3)

[ ak  , Bk ]  ak  ak ak   ak ak  ak 

  ak ak  ak   ak ak  ak 
0

(4)

[ak , Bk ]  ak ak ak   ak ak  ak

 ak ak ak   ak ak ak 
0

(5)

119
 [a k  , Bk ]  a k  ak ak   ak ak  a k 
 ak a k  ak   ak ak  a k 
 ak (1  2ak  a k  )

(6)

[ak  , Bk ]  ak  ak ak   ak ak  ak 

 ak  ak ak   ak ak  ak 
 ( ak  ak  ak ak  )ak 
 ak 

When these operators are applied to  vac , we have

(a)

[ a k  , Bk ]  vac  a k  Bk  vac  a k  (1  2 a k a k  )  vac  a k   vac ,

or

ak  Bk vac  ak  vac .

(b)

[ a  k  , Bk ]  vac  a  k  Bk  vac   a k (1  2 a k  a  k  )  vac   a k  vac .

or

a  k  Bk  vac   a k  vac .

(c)

[ a  k  , Bk ]  vac   a k (1  2 a k  a  k  )  vac   a k  vac .

or

a  k  Bk  vac   a k  vac .

(d)

[ a k  , Bk ]  vac  ( a k  Bk  Bk a k  )  vac  a k   vac ,

120
or

a k  Bk  vac  a k   vac .

________________________________________________________________________
55. The nature of the BCS ground state
Here we show that,

 k  BCS  0 .  k  BCS  0 .

((Proof))

 k  BCS   (ul  vl Bl )(u k ak   vk ak  )(u k  vk Bk )  vac

l k

  (ul  vl Bl )(u k ak   u k vk ak  Bk  u k vk ak   vk ak  Bk )  vac

2 2

l k

  (ul  vl Bl )(u k vk ak  Bk  u k vk ak   vk ak  Bk )  vac

2

l k

  (ul  vl Bl )(u k vk a k   u k vk ak   vac  0

l k

and

  k  BCS   (ul  vl Bl )(uk a k   vk ak )(u k  vk Bk )  vac

l k

  (ul  vl Bl )(uk a k   uk vk a k  Bk  uk vk ak  vk ak Bk )  vac

2 2

l k

  (ul  vl Bl )(u k vk ak  u k vk ak )  vac  0

l k

Here we note that

ak  Bk  vac  ak  ak ak   vac  0 ,

ak Bk  vac  ak ak ak   vac  0 ,

from the Pauli's exclusion principle; No two electrons can occupy the one state denoted
by k , , where k is the wavenumber and  is the spin up or down state.
____________________________________________________________________
56. Paring operators

Bk  a k  ak  ,

121
 Bk  a k a k 

The pairing operators obey commutation relations given by

[ Bk , Bk' ]  (1  nk   nk  ) k ,k '

with

nk   a k a k  , n k   a k  a  k  .

and

[ Bk , Bk ' ]  0 .

[ Bk , Bk ' ]  2 Bk Bk ' (1   k , k ' ) .

Bk Bk  vac  a k  ak  ak ak   vac

 a k  ak  ak  ak  vac
 a k  ak   vac
  vac

since

ak ak  vac   vac ,

we have


Bk  vac  0 ,  vac Bk  0 .

57. Normalization of the ground state:

 (u  vk Bk ) (u k  vk Bk )  vac
*
BCS BCS   vac k
k k

 (u
*
  vac k  vk Bk )(u k  vk Bk )  vac
k

 (u
2 * *
  vac k  u k vk Bk  u k vk Bk  vk vk Bk Bk )  vac
k

 (u
2 *
  vac k  vk vk )  vac
k

  vac  vac  1

122
since

2 *
uk  vk vk  1 .

________________________________________________________________________
58. Excited state: quasi particles

  k    uk  vk  ak     k   uk  vk  ak  
* 
     *    ,     .
     v uk  ak  
v
 k    k u k  ak    k    k

(i)
 k BCS   (ul  vl Bl )(u k ak  vk *a k  )(u k  vk Bk )  vac
l k

  (ul  vl Bl )(u k ak  u k vk ak Bk  u k vk a k   vk vk a k  Bk )  vac

2 * *

l k

  (ul  vl Bl )(u k ak  vk vk ak )  vac

2 *

l k

 ak  (ul  vl Bl )  vac

l k

since

a  k  Bk  vac   a k  vac ,

and

2 *
u k  vk vk  1 .

(ii)

 k  BCS   (ul  vl Bl )(vk *ak   u k ak  )(u k  vk Bk )  vac
lk

  (ul  vl Bl )(u k vk ak   vk vk ak  Bk  u k ak   u k vk ak  Bk  vac

* * 2

lk

  (ul  vl Bl )(vk vk ak   u k ak  )  vac

* 2

lk

 ak   (ul  vl Bl )  vac

l k

(iii)

123
  k k  BCS   (ul  vl Bl )(uk ak  vk *ak  )(vk *ak   uk ak  )(uk  vk Bk )  vac
l k

  (ul  vl Bl )(uk ak  vk a k  )ak   vac

*

l k

  (ul  vl Bl )(uk ak ak   vk a k  ak  )  vac

*

l k

 (uk Bk  vk ) (ul  vl Bl )  vac

*

l k

From these, it is found that

ground state Cooper pair (k, -k): (u k  vk Bk ) .

the first excited state (k): a k .
the first excited state (-k): ak  .
*
the excited state (k) and (-k): (uk Bk  vk ) .

________________________________________________________________________
59. Condensation amplitude

*
BCS nk  BCS  BCS ak ak  BCS  vk vk .

((Proof))

BCS ak ak  BCS   vac [ (ul  vl Bl ) (ul  vk Bl )](u k  vk Bk ) ak ak  (u k  vk Bk )  vac
* *

l k l k

  vac [ (ul  vl Bl ) (ul  vk Bl )](u k  vk Bk )vk ak ak  Bk )  vac

* *

l k l k

  vac [ (ul  vl Bl ) (ul  vk Bl )](u k  vk Bk )vk ak ak   vac

* *

l k l k

  vac [ (ul  vl Bl ) (ul  vk Bl )](u k  vk Bk )vk Bk  vac

* *

l k l k

  vac [ (ul  vl Bl ) (ul  vk Bl )](u k vk Bk  vk vk )  vac

* *

l k l k

 (u
* *
 vk vk  vac l  vl Bl )(ul  vk Bl ))  vac
l k
* 2
 vk vk  vk

using

ak  Bk  vac  ak   vac , Bk Bk  vac   vac .

124
Similarly, we have

BCS ak ak  BCS   vac [ (ul  vl Bl ) (ul  vk Bl )](u k  vk Bk ) ak ak  (u k  vk Bk )  vac
* *

l k l k

  vac [ (ul  vl Bl ) (ul  vk Bl )](u k  vk Bk )vk ak ak  Bk )  vac

* *

l k l k

  vac [ (ul  vl Bl ) (ul  vk Bl )](u k  vk Bk )vk ak ak   vac

* *

l k l k

  vac [ (ul  vl Bl ) (ul  vk Bl )](u k  vk Bk )vk Bk  vac

* *

l k l k

  vac [ (ul  vl Bl ) (ul  vk Bl )](u k vk Bk  vk vk )  vac

* *

l k l k

 (u
* *
 vk vk  vac l  vl Bl )(ul  vk Bl )  vac
l k
* 2
 vk vk  vk

using

a k  Bk  vac  a k   vac .

Fk is the condensation amplitude in the state k.

Fk  BCS Bk BCS .

Fk  BCS ak ak  BCS
  vac [ (ul  vl Bl ) (ul  vk Bl )](u k  vk Bk ) Bk (u k  vk Bk )  vac
* *

l k l k

  vac [ (ul  vl Bl ) (ul  vk Bl )](u k  vk Bk )(u k Bk  vk Bk Bk )  vac
* *

l k l k

  vac [ (ul  vl Bl ) (ul  vk Bl )](u k Bk  u k vk Bk Bk  u k vk Bk Bk

* 2 *

l k l k
*  
 vk vk Bk B B )  vac
k k
*
 u k vk

since

[ Bk , Bk ' ]  0 .

[ Bk , Bk ' ]  2 Bk Bk ' (1   k , k ' ) .

Bk Bk  vac   vac .

125
 [ Bk , Bk' ]  (1  nk   nk  ) k ,k ' .

BCS Bk BCS  BCS a k  ak  BCS

  vac [ (ul  vl Bl ) (ul  vk Bl )](u k  vk Bk ) Bk (u k  vk Bk )  vac
* *

l k l k

  vac [ (ul  vl Bl ) (ul  vk Bl )](u k  vk Bk )(u k Bk  vk Bk Bk )  vac

* *

l k l k

  vac [ (ul  vl Bl ) (ul  vk Bl )](u k Bk  u k vk Bk  u k vk Bk Bk

* 2 * 2

l k l k
* 
 vk vk Bk Bk B )  vac
k

 u k vk
________________________________________________________________________
60. The number operator:
The number operator is define by

N   ak ak   ( ak ak   ak ak  )   ( nk   nk  ) .

k , k k

The average of N for the BCS state is

N  BCS N BCS  2 vk .
2

N 2  BCS N 2 BCS  (2 vk ) 2   4uk vk .

2 2 2

k k

The fluctuation is given by

N 2    4uk vk .
2 2 2
N2  N
k

Note that

2 2 2
BCS a k a k  a k' a k ' BCS  vk  k , k '  vk vk ' (1   k , k ' ) .

The quantities uk and |vk| will typically be numbers of order 1, so since the numbers of
allowed k-states appearing in the k sums scale with the volume  of the system, we have

N  , and N 2   .

Therefore the width of the distribution of numbers in the BCS state is

126
 N  1 1
   .
N   

As N ≈ 1023 particles, this relative error implied by the number non-conservation in the
BCS state becomes negligible.

________________________________________________________________________
61. Relation between N state and BCS state.
Here we assume that

vk
hk  vk  vk ei ,
uk

Then we get

BCS     u k  (1  hk e i Bk )  vac ,

k k

or

2
1 1  
BCS ( )  {1  ei  hk Bk  ei 2   hk Bk  }  vac ,
1! k 2  k 

where  is the phase of the order parameter. This indicates that the BCS state can be
expressed as the superposition of N as


i
BCS ( )  
N 0
AN exp(
2
N) N ,

where N is the state having the N particles and N N '   N , N ' .


i
N BCS ( )  N 
N ' 0
AN ' exp(
2
N') N'

i
 
N ' 0
AN ' exp(
2
N') N N'

i
 AN exp( N)
2

or

127
 i
AN  exp( N ) N BCS ( ) .
2

We also have the expression

2
1 d i
N 
AN  2 exp( 2
0
N ) BCS ( ) .

AN is the probability for finding the system in the N . AN may be expressed by a

Gaussian distribution given by

1 ( N   N ) 2
AN  exp[ ,
2 N 2(N ) 2

where

N 2   4uk 2 vk
2
N  N *  2 v k ,
2
.
k k

AN
0.04

0.03 < N>=104

<DN>=10
0.02

0.01

N<N>
0.996 0.998 1.000 1.002 1.004

Fig. AN which has a Gaussian distribution with <N> = N* = 104 and N =10. The blue
arrow is the full-width at half maximum. In this figure,  ( N /  N ) = 2.355 x
10-3.

For an arbitrary operator F, the matrix element is given by

128
 BCS ( ) F BCS ( )  
N ,N '
BCS ( ) N N F N ' N ' BCS ( )

i i
 
N ,N '
AN exp(
2
N) N F N' AN ' exp(
2
N')

i
 
N ,N '
AN AN ' exp[
2
( N  N ' )] N F N '

where

i i
N BCS ( )  AN exp( N), BCS ( ) N  AN exp( N) .
2 2

(a) Suppose that F conserved the number of particles'

i
BCS ( ) F BCS ( )  
N ,N '
AN AN ' exp[
2
( N ' N )] N ' F N

  AN N F N
N

 N* F N* A
N
N

 N* F N*

(b) If F acting on N gives a state of N + p particles.

i
BCS ( ) F BCS ( )  
N ,N '
AN AN ' exp[
2
( N ' N )] N ' F N

ip
  AN exp( ) NpF N
N 2
ip
 exp(
2
) N*  p F N* A
N
N

ip
 exp( ) N*  p F N*
2

These expectation values do not vanish in the state BCS ( ) . It is clear that this state is
much simpler than N , and, moreover, manifestly shows essential features of
superconductivity.

62. BCS Theory

We start the so-called the BCS Hamiltonian

129
 H  N    k ak, ak ,   Vkl ak ak  al  al  .
k , k ,l

We assume that uk and vk are real.

WBCS  BCS Hˆ  Nˆ BCS  2  k vk   Vkl u k vk ul vl ,

2

k k ,l

which is to be minimized subject to the constant that

2 2
u k  vk  1 .

It is useful to put

u k  sin  k , v k  cos  k ,

1
WBCS  BCS Hˆ  Nˆ BCS  2  k cos k  Vkl sin(2 k ) sin(2 l ) .
2

k 4 k ,l

Then we get

 1
BCS Hˆ  Nˆ BCS  2  k sin( 2 k )   Vkl cos( 2 k ) sin( 2 l )  0 ,
 k k 2 k ,l

or

1
 k tan(2 k )  Vkl sin(2 l ) .
2 l

We define the energy gap equation by

1
 k    Vkl sin  l cos  l    Vkl vl u l    Vkl sin( 2 l ) .
l l 2 l

Thus we have

 k tan( 2 k )    k .

We find

130
 k
cos( 2 k )   ,
Ek

k
sin(2 k )  ,
Ek

where

Ek   k   k .
2 2

The gap energy is determined from the gap equation

l l
 k   Vkl  Vkl .
l 2 El l 2 l  l
2 2

We choose the simplified interaction (BCS interaction),

Vkl  V if k  D and l  D ,

Vkl  0 otherwise.

and

k   (independent of k) if k  D ,

k  0 otherwise.

Then we have

 D
d 
  VN (0) 
 2  2  2
,
D

or

131
  D
1 d  D
VN (0)
 
0  2  2
 sinh 1 (

).

This equation only has a solution for positive V,

 D 1
  2 D exp[  ].
1 VN ( 0 )
sinh[ ]
VN (0)

where N(0) is the density of states per spin at the Fermi level.
_______________________________________________________________________
63. Summary

vk2: probability that the Cooper pair (k, -k) is occupied.

uk2: probability that the Cooper pair (k, -k) is unoccupied.

where

2 2
u k  vk  1 ,

2 1  2 1 
uk  (1  k ) , vk  (1  k ) ,
2 Ek 2 Ek

where Ek is the excitation energy and is defined by

E k   k  2
2

Fk is the condensation amplitude and is defined by


Fk  uk vk  .
2 Ek

132
 1.0

uk 2
0.8

0.6

0.4

vk u k
0.2

vk 2
x D
-4 -2 2 4

Fig. Plot of uk2, vk2, and Fk= ukvk as a function of /. We assume that vk is real for
convenience.

There is an energy gap between the ground state and the excited state. The excitations
consists of the breaking of pairs. The excitation energy is the sum of Ek for each excited
electrons in each pair. The superconducting state is a condensed state in the sense that a
finite energy  is required to produce an excited state of the whole system.

64. Thermodynamic critical field

If we make the substitutions

 2 1 
2u k vk  , vk  (1  k )
Ek 2 Ek

into

WBCS  2  k vk  Vkl u k vk ul vl ,
2

k k ,l

then we get

k 1  
WBCS    k (1  )  Vk , k ' k k ' ,
k Ek 4 k ,k ' Ek Ek '

or

2 1 2 1
WBCS  ( E 0   N (0) 2 )   E 0  N (0) 2 .
V 2 V 2

133
Comparing this to the normal state energy ( = 0), measured relative to F,

E0   2
k kF
k .

we get

1
WBCS  E0   N (0)2 <0.
2

This indicates that

1 [ H c (0)]2
N ( 0)  
2
.
2 8

65. Mean field approximation at finite temperatures

At finite temperatures, quasi-particles are thermally excited. We need to take into
account of the effect in the mean field theory. To this end, the average over the BCS
ground state is replaced by the average in the thermal equilibrium, such that

BCS aka k ' BCS  ak a k ' .

BCS ak ak  BCS  ak ak  .

Here we use the Bogoliubov transformation given by

 ak    uk vk   k  
     *    ,
a
 k     vk uk  k  

The Fermi-Dirac distribution function is given by

  1
  k   k     k   k   f ( E k )  ,
exp(  E k )  1

where < > is the average in the thermal equilibrium at the finite temperature. We also
note that

134
   k  k     k  k    k k   k  k  ...  0 ,

since the number of particles are not conserved. Based on these rules, we can calculate
the following quantities.

(1)

ak  a k '  (uk k   vk k  )(vk ' k'  uk '  k ' ) 

 uk vk '   k  k'  uk 'vk   k   k ' 
 uk vk [   k  k     k   k  ] k , k '
 [1  2 f ( Ek )]uk vk k , k '

and

a k  ak '  [1  2 f ( Ek )]u k vk  k , k '

(2)

ak ak   (uk k  vk  k  )(u k k   vk k  ) 

*

 uk   k k   uk vk   k  k   uk vk  k k   vk vk  k  k 

2 * *

2
 uk   k k    vk [1   k  k  ]
2

2 2
 uk f ( Ek )  vk (1  f ( Ek )
2 2 2
 vk  (uk  vk ) f ( Ek )

(3)

ak ak   (uk k  vk   k  )(vk  k   uk k  ) 

* *

 uk vk   k k   uk vk    k  k  

* *

 uk vk   k k   uk vk [1   k   k  ]

* *

*
 uk vk (1  2 f ( Ek )

66. Fermi-Dirac function

The Fermi-Dirac distribution function is defined by

135
 1 1 1
f (E)    ,
exp(  E )  1 exp(   2  2 )  1 2
exp(    1)  1
2

where

3.5277 k BTc T
   1.76385 c ,
2 k BT T

and

2  3.5277k BTc . (BCS prediction)

f E

0.20 t= T  T c

t=1.4
0.15

1.2
0.10
1.0

0.05 0.8
0.6

x D
0.2 0.4
-3 -2 -1 1 2 3

Fig. Plot of f(E) as a function of /. t is the reduced temperature; t = T/Tc. t = 0.2, 0.4,
0.6, 0.8, 1.0, 1.2, and 1.4.

This function is an unusual Fermi function. The denominator features exp(E), rather
than the conventional form exp[(-F)]. Because of E>0, f(E) reaches its maximum at
the Fermi level, and it decays to zero when k lies either above or below. We also have the
relation

1 exp(  E ) 1
1  f (E)  1    '
exp(  E )  1 exp(  E )  1 exp(   E )  1

136
 1-f E
1.00 0.2
0.4
0.6
0.95 0.8

1.0
0.90
1.2

0.85 t= T  T c
t=1.4

0.80
x D
-3 -2 -1 1 2 3

Fig. Plot of 1-f(E) as a function of /. t is the reduced temperature; t = T/Tc. t = 0.2,
0.4, 0.6, 0.8, 1.0, 1.2, and 1.4.

67. Order parameter: coherence length and phase

The wavefunction is defined as

1 1
 (r ) 

ek
ik  r
ak  ,  (r ) 

e k
ik  r
ak 

where  is the volume of the system. Then the order parameter of the superconducting
phase is defined by the form of correlation of the wavefunctions at r and r'.

 (r  r ' )   (r ) (r ' )

1
 
 k ,k '
ei (k r  k 'r ') ak  a k '

1  E
 
2 k
eik (r r ') k tanh( k )
Ek 2

where we use the above formula

ak  a k '  u k vk [1  2 f ( E k )] k ,k '

 uk vk [1  2 f ( Ek )] k ,k '
k E
 tanh( k ) k ,k '
2 Ek 2

When T0, f ( Ek ) =0, and

137
 ak  ak   BCS ak  ak  BCS  uk vk  uk vk

since uk  uk . Then the order parameter  (r) at T = 0 can be evaluated as

1 1  sin( k F r ) r
 (r )  
 k
eik r uk vk  
2 k
e ik  r k 
Ek kF r
K0 (
 0
) , (1)

where K0(x) is the modified Bessel function of the second kind. 0 is the coherence
length and is defined by

v F v F
0  
 0  (1.763 k B Tc )

Note that in the limit of x   ,


K 0 ( x)  ex .
2x

The form of  (r) shows that the coherence length  0 represents the spatial extension of
the Cooper pair.

68. Phase of the order parameter

The order parameter at the finite temperature can be expressed by

 (r )    (r )  (0)
1
  ek r ak ak 
 k

The phase of the order parameter comes from the phase of vk ( vk  vk e i ) as

ak  ak   uk vk [1  2 f ( Ek )]
k Ek
 ei tanh( )
2 Ek 2

Then the order parameter can be rewritten as

138
 1 i  E
 (r)  ei  (r)  e  eik r k tanh( k ) .
 k 2 Ek 2

The parameter  is the phase of the order parameter. The order parameter has a wave-like
nature. Such character is found to appear in the Josephson effect.

________________________________________________________________________
69 Coherence length of Al and Sn
(1) Al (type-I superconductor)
kF = 1.75 x 108/cm.
vF = 2.02 x 108 cm/s.
F = 11.63 eV.
Hc = 105 Oe.
2  0 = 0.340 meV = 3.94553 K.
 0 = 1.973 K.
Tc = 1.140 K.

2  0 /kBTc = 3.461.

v F
0  = 2489.5 nm.
 0

((Experiment)) Al
Coherence length: 0 = 1600 nm.
London penetration depth  L = 16 nm.

(2) Sn (type-I superconductor)

kF = 1.62 x 108/cm.
vF = 1.88 x 108 cm/s.
F = 10.03 eV.
Hc = 309 Oe.
2  0 = 1.15 meV = 13.345 K.
 0 = 6.6726 K.
Tc = 3.722 K.

2  0 /kBTc = 3.585.

139
 v F
0  = 685.0 nm = 6850 Å.
 0

((Experimental values)) Sn
Coherence length: 0 = 230 nm.
London penetration depth  L = 34 nm.

_______________________________________________________________________
70. Approach from the Heisenberg's principle of uncertainty
In the BCS theory, the Cooper pairs are formed from electrons having the energy
close to the Fermi energy. Using the Heisenberg's principle uncertainty, we have

px   .

Since

0
  vF p   0 or p  ,
vF

x can be evaluated as

 v
x     F ,
p  0

where vF is the Fermi velocity. Note that the average distance between electrons d is
approximated by

v F
d .
F

since

2
1 2kF k
kF  , and F    F k F  v F k F .
d 2m m

When F>>, we have

  d .

140
71. Character in charge of quasi-particles
The charge density is defined by

Q  e ak ak

k ,

 2e{ vk  (uk  vk ) f ( Ek )]}

2 2 2

 Qs  Qqp

Here Qs is the contribution of the condensate of the Cooper pairs, and Qqp is the
contribution of the quasi-particle excitations,

Qs  2e vk ,
2

and

Qqp  2e (uk  vk ) f ( Ek )] .

2 2

Here we use the formula

2 2 2
ak ak   vk  (uk  vk ) f ( Ek ) .

We make a plot of

2 2
hk  (uk  vk ) f ( Ek ) ,

It s found that the sign of hk changes from positive (hole-like for <0) to negative
(electron-like for >0). This means that the quasi-particles behave like an electron for
>0 and like a hole for <0.

141
 t=T Tc
hk
0.04

hole-like 0.02 electron-like

t=0.2 x D
-4 -2 t=0.4 2 4

t=0.6
- 0.02

- 0.04 t=0.8

2
Fig. Plot of hk  (u k 2  vk ) f ( Ek ) , corresponding to the charge of quasiparticle, as a
function of /. t = T/Tc. t is changed as a parameter.

72. Condensation amplitude at finite temperature

The condensation amplitude at finite temperature is given by

*
F  akak   uk vk [1  2 f ( Ek )] .

Here we assume that vk is real. Then we have

k
Fk  uk vk [1  2 f ( Ek )]  [1  2 f ( Ek )] .
2 Ek

We make a plot of Fk as a function of k/k = /, where t = T/Tc is changed as a

parameter.

142
 Fk
0.5 t=0.2
t=0.4
t=0.6
0.4 t=0.8

0.3

0.2

0.1 t=T Tc

x D
-6 -4 -2 2 4 6

Fig. Plot of the condensation amplitude F as a function of /. /. t is the reduced
temperature; t = T/Tc. t = 0.2, 0.4, 0.6, and 0.8. We assume that the energy gap is
independent of t.

73. Specific heat

The electronic entropy in the superconducting phase is given by


Ses  2k B  dEN ( E ){ f ( E ) ln f ( E )  [1  f ( E )] ln[1  f ( E )]} ,
0

where f(E) is the Fermi Dirac function and the factor 2 comes from the degree of freedom
in spin.

1
f (E)  .
exp(  E )  1

Then the specific heat is calculated as

S 2 T d2 f ( E )
Ces  T   dEN ( E )( E 2  )[ ].
T T 2 dT E

(i) At low temperatures (T<<Tc), where  is independent of T.

f ( E )  exp( E ) .

The density of states is defined as

143
 E
N ( E )  N ( 0) .
2
E 2  0

Then we have


2 f ( E )
Ces   dEN ( E ) E 2 [  ]
T 0 E

2 N (0) 1 dEE 3e E

T k BT  (1  e 
0
E 2
) E 2  0
2


2 N (0) 1 dEE 3e  E

T k BT 
0 E 2  0
2

3
2 N (0)  0 1
 [ K1 (   0 )  K 2 (   0 )]
k BT 2
 0
3
2 N (0)  0
 K1 (   0 )
k BT 2

in the limit of  0 →∞, where Kn(x) is the modified Bessel function of the second kind
and has a symptotic form of

e x
K n ( x) 
2x


which is independent if n. Then we have

3/ 2
   0
Ces  2 N (0)k B  0  0  exp(  )
 k BT  k BT

and

144
 3/ 2
Ces 2 N (0)k B  0  0  0
   exp( )
Tc 2  2 k 2 N (0)T k T
 B  k BT
B c
3
3/ 2
3 2  0  0
 3/ 2 0   exp( )
2 k BTc  k BT  k BT
3/ 2
3 2  3.5277   0  0
   exp( )
4 3 / 2  k BT  k BT
3/ 2
 1.764Tc  1.764Tc
 0.671959  exp( )
 T  T
where

2 0  3.5277 k BTc .

0
We make a scaling plot of Ces /[ 2 N (0)k B  0 ] as a function of .
k BT

C 2 p N0k B D0

0.30

0.25

0.20

0.15

0.10

0.05

k B TD0
0.1 0.2 0.3 0.4

Fig. Plot of Ces /[ 2 N (0)k B  0 ] as a function of kBT/0 for T<<0/kB.

(ii) At T = Tc, where  = 0. The first term is continuous through the transition ( →0)

145
 S 2 f ( E )
Cn  T   dEN ( E ) E 2 [ ]
T Tc E

2 f ( )
 N (0)2  d [ 2 ]
Tc 0

2 2
 2 N ( 0)
Tc 6  2
2 2 2
 k B Tc N (0)  Tc
3

Note that E   in the limit of  →0. The specific heat jump at Tc is given by

C  (Cs  Cn )T Tc 0
 d2  f ( )
  
 dT T Tc 0
 dN ( )(

)

 d2 
  N (0) 
 dT T Tc 0

where we use

f ( )
   ( ) .


 ( ) is the Dirac delta function. Using the relation,

1/ 2
 8  T T
  k BTc   (1  )1/ 2  3.0626k BTc (1  )1/ 2 ,
 7 (3)  Tc Tc

we get

8
C   2 k B Tc N (0)  0.950751 2 k B Tc N (0)
2 2

7 (3) .

Then the ratio is evaluated as

8
 2 k B 2Tc N (0)
C 7 (3) 12
   1.42613 . (BCS prediction).
Cn 2 2
2 7 (3)
k B Tc N (0)
3

146
((Mathematica))

Integrate , x, ,  

x  x3

x2  2
Simplify ,   0,   0 &

2   BesselK1,    BesselK2,  



Integrate , x, 0,  

2x
Exp x  1
Simplify ,   0 &

2
6 2

_______________________________________________________________________
REFERENCES
1. J. Bardeen, L.N. Cooper, and J.R. Schrieffer, Phys. Rev. 108, 1175 (1957).
2. P.G. de Gennes, Superconductivity of Metals and Alloys (W.A. Benjamin, New
York, 1966).
3. M. Tinkham, Introduction to Superconductivity, Reprint edition (Robert E.
Krieger Publishing Company, INC, Malabar, Florida, 1980).
4. J.B. Ketterson and S.N. Song, Superconductivity (Cambridge University Press,
1999).
5. T. Tsuneto, Superconductivity and superfluidity (Cambridge University Press,
1998).
6. J.R. Schrieffer, Theory of Superconductivity, revised edition (Addison-Wesley,
Reading, 1983).
7. W. Buckel and R. Leiner, Superconductivity, Fundamentals and Applications,
Wiley-Vch Verlag GmbH & Co. KGaA, Weinheim, 2004).
8. C.G. Kuper, An Introduction to the Theory of Superconductivity (Clarendon Press,
Oxford, 1968).
9. G. Rickayzen, Theory of Superconductivity (Interscience, 1964).
10. D. Saint James, G. Sarma, and E. Thomas, Type II Superconductors (Clarendon
Press, Oxford, 1971).
11. S. Nakajima, Introduction to Superconductivity (in Japanese) (Baifukan, Tokyo,
1971).
12. C. Kittel, Quantum Theory of Solids, second revised printing (John Wiley & Sons,
New York, 1987).
13. H. Ibach and H. Lűth, Solid-State Physics, 4-th edition (Springer, Berlin, 2009).
14. R.P. Feynman, Statistical Mechanics (Benjamin, Reading, MA, 1972).
15. M.P. Marder, Condensed Matter Physics (John Wiley & Sons, New York, 2010).

147
16. L. Hoddeson and V. Daitch, True Ginus. The life and science of John Bardeen
(Joseph Henry Press, Washington DC, 2001).
17. D.J. Thouless, Topological quantum numbers in nonrelativistic physics (World
Scientific, Singapore, 1998).
18. M. Suzuki and I.S. Suzuki, Ginzburg-Landau theory for superconductivity,
http://www2.binghamton.edu/physics/docs/ginzburg-landau.pdf
________________________________________________________________________
APPENDIX-I
Formula related to the BCS theory

The magnetic quantum flux (fluxoid, or fluxon) is

hc 2c
0    2.06783372  10 7 (Gauss.cm2).
2e 2e

The coherence length at T = 0 K

v F
0  .
 0

The energy gap at T = 0 K:

1 1
 0  2 D exp[ ]  2k B  D exp[ ].
VN (0) VN (0)

The energy gap

20  3.5277kBTc .

The energy gap

 (T ) T
 1.73669 (1  )1/ 2 .
0 Tc

The specific heat discontinuity:

C s  Cn 12
|Tc   1.43 .
Cn 7 (3)
________________________________________________________________________
Bogoliubov transformation:

148
   k    uk  vk  ak     k   uk  vk  ak  
* 
     *  ,     .
uk  ak       v 
uk  k  
 k    vk  k    k a

or

 ak    uk vk   k    ak   uk vk   k  
* 
        ,     .
a   v uk  k  
k  k  
*
a
 k    k v u  k    k

The excitation energy of quasiparticle

2
Ek   k   k .
2

2 1 
uk  (1  k ) .
2 Ek

2 1 
vk  (1  k ) .
2 Ek

k   k   F .

k
vk  uk .
 k  Ek

k
u k vk  .
2 Ek

 k   k ei ,

uk is real. v k  v k ei .

Density of states for the quasi-particles

E
N (E) E 
  .
N n ( 0) E 2  2 E
2

  1


149
 1 [ H c (0)]2
N ( 0)  
2
.
2 8

________________________________________________________________________
The BCS ground state:

BCS   (u k  vk Bk )  vac .

k

*
BCS a k a k  BCS  u k vk .

k
BCS ak  a k  BCS  u k vk 
2 Ek .

2
BCS ak ak  BCS  vk .

 k  BCS  (uk ak   vk ak  ) BCS  0 .

 k  BCS  (uk ak   vk ak ) BCS  0 .

N 2    4uk vk .
2 2 2
N2  N
k

N  BCS N BCS  2 vk .
2

[ak , ak' ' ]  ak ak' '  ak' ' ak   k ,k '  , ' .

[ ak ak ' ' ]  [ ak , ak' ' ]  0 .

 
ak ak  vac  (1  ak ak )  vac   vac .

[ Bk , Bk ' ]  0 .

[ Bk , Bk ' ]  2 Bk Bk ' (1   k , k ' ) .

Bk Bk  vac   vac .

[ Bk , Bk' ]  (1  nk   nk  ) k ,k ' .

150
 [ a k  , Bk ]  ak  (1  2a k a k  ) .

[ a k  , Bk ]   a k (1  2 a k  a k  ) .

________________________________________________________________________

Bk  vac  a k  ak   vac  0 .


Bk  vac  ak ak   vac  ak  ak  vac .


Bk Bk  vac   vac .

________________________________________________________________________

a k a  k '  u k vk [1  2 f ( E k )] k ,k ' .

a  k  a k'  u k v k [1  2 f ( E k )] k , k '

2 2 2
ak ak   vk  (uk  vk ) f ( Ek ) .
________________________________________________________________________

H  E g   Ek ( k   k  k   k  )
 

Eg  [ Ek  k   k bk ]
*

APPENDIX-II The heat capacity of electrons per mol atom

The heat capacity of N electrons is given by

1
Cel   2 k B D( F )T
2

So the heat capacity per electron is given by

Cel 1 2 2 D( F )
  kB T.
N 3 N

Suppose that each atom has nv conduction electrons. The total number of electrons is N;

N  nv N 0 .

151
So each atom has the electronic heat capacity as

Cel Cel nv 1 2 2
nv    k B D ( F )T
N N0 N 3
1 2 D ( F )
  2kB T
3 N0
1
  2 k B D A ( F )T
2

where

D( F )
DA ( F ) 
N0

or

D ( F ) 3nv
D A ( F )  nv 
N 2 F

since

D ( F ) 3

N 2 F

The heat capacity per mol atom is

1
  2 k B DA ( F ) N AT
(M ) 2
Cel
3

where NA is the Avogadro number.

1 2
 N Ak B 2 =2.35715 mJ eV/K2.
3

 is related to DA ( F ) as

1
   2 N Ak B 2 DA ( F ) ,
3
or

 (mJ/mol K2) = 2.35715 DA ( F ) .

152
________________________________________________________________________

153