USPAS June 2008 U.

Maryland

SUPERCONDUCTIVITY
OVERVIEW OF EXPERIMENTAL FACTS
EARLY MODELS
GINZBURG-LANDAU THEORY
BCS THEORY
Jean Delayen

Thomas Jefferson National Accelerator Facility

Old Dominion University
Historical Overview
 Perfect Conductivity

Kamerlingh Onnes and van der Waals in

Leiden with the helium 'liquefactor' (1908)
Unexpected result
Expectation was the opposite: everything should become an isolator at T Æ 0
 Perfect Conductivity
Persistent current experiments on rings have measured

ss
> 1015
sn

Resistivity < 10-23 Ω.cm

Decay time > 105 years

Perfect conductivity is not superconductivity

Superconductivity is a phase transition

A perfect conductor has an infinite relaxation time L/R
Perfect Diamagnetism (Meissner & Ochsenfeld 1933)

Perfect conductor Superconductor

∂B
=0 B=0
∂t
 Penetration Depth in Thin Films

Very thin films

Very thick films

 Critical Field (Type I)

Superconductivity is destroyed by the application of a magnetic field

È Ê T ˆ2˘
H c (T ) H c (0) Í1 - Á ˜ ˙
ÍÎ Ë Tc ¯ ˙˚

Type I or “soft” superconductors

Critical Field (Type II or “hard” superconductors)

Expulsion of the magnetic field is complete up to Hc1, and partial up to Hc2

Between Hc1 and Hc2 the field penetrates in the form if quantized vortices or
fluxoids
p
f0 =
e
 Thermodynamic Properties

Entropy Specific Heat

Energy Free Energy

 Thermodynamic Properties
When T < Tc phase transition at H = H c (T ) is of 1st order fi latent heat

At T = Tc transition is of 2nd order fi no latent heat

jump in specific heat

Ces (Tc ) ∼ 3Cen (Tc )

Cen (T ) = g T electronic specific heat

Ces (T ) ª a T 3 reasonable fit to experimental data
 Thermodynamic Properties
At Tc : S s (Tc ) = S n (Tc ) The entropy is continuous

∂S C
Recall: S (0) = 0 and =
∂T T
aT 3 gT 3g
Ú
Tc

Ú
Tc
T3
fi dt = dt Æ a = 2 Ces = 3g 2
0 T 0 T Tc Tc
T3 T
S s (T ) = g 3 S n (T ) = g
Tc Tc

For T < Tc S s (T ) < S n (T )

superconducting state is more ordered than normal state
A better fit for the electron specific heat in superconducting state is
bTc
-
Ces = a g Tc e T
with a ª 9, b ª 1.5 for T Tc
 Energy Difference Between Normal and
Superconducting State

U n (Tc ) = U s (Tc ) Εnergy is continuous

3 g g 2
( )
Tc
U n (T ) - U s (T ) = Ú (Ces - Cen )dt = (Tc
4
- T 4
) - Tc - T 2
T 4 Tc2 2
H c2
at T =0 () () 1 2 H c2
U n 0 - U s 0 = g Tc =
4 8p 8p
is the condensation energy

H c2
at T π 0, is the free energy difference
8p
2 2
2 È ÊT ˆ ˘
= DF = (U n - U s ) - T ( S n - Sc ) = g Tc Í1 - Á ˜ ˙
H c (T ) 1 2
8p 4 Í Ë Tc ¯ ˙˚
Î
1 È Ê T ˆ2˘
H c (T ) = (2pg ) 2 Tc Í1 - Á ˜ ˙
ÍÎ Ë Tc ¯ ˙˚

The quadratic dependence of critical field on T is

related to the cubic dependence of specific heat
 Isotope Effect (Maxwell 1950)

The critical temperature and the critical field at 0K are dependent on

the mass of the isotope

Tc ∼ H c (0) ∼ M -a with a 0.5

 Energy Gap (1950s)
At very low temperature the specific heat exhibits an exponential
behavior
cs μ e - bTc /T with b 1.5

Electromagnetic absorption shows a threshold

Tunneling between 2 superconductors separated by a thin oxide film
shows the presence of a gap
 Two Fundamental Lengths
• London penetration depth λ
– Distance over which magnetic fields decay in
superconductors
• Pippard coherence length ξ
– Distance over which the superconducting state decays
 Two Types of Superconductors
• London superconductors (Type II)
– λ>> ξ
– Impure metals
– Alloys
– Local electrodynamics

• Pippard superconductors (Type I)

– ξ >> λ
– Pure metals
– Nonlocal electrodynamics
Material Parameters for Some Superconductors
Phenomenological Models (1930s to 1950s)

Phenomenological model:
Purely descriptive
Everything behaves as though…..

A finite fraction of the electrons form some kind of condensate that

behaves as a macroscopic system (similar to superfluidity)

At 0K, condensation is complete

At Tc the condensate disappears

Two Fluid Model – Gorter and Casimir
T < Tc x = fractionof "normal"electrons
(1- x ) : fractionof "condensed"electrons (zero entropy)

Assume: F (T ) = x1/2 f n (T ) + (1 - x ) f s (T ) free energy

1
f n (T ) = - g T 2
2
1
f s (T ) = - b = - g Tc2 independent of temperature
4
4
ÊTˆ
Minimizationof F (T ) gives x=Á ˜
Ë TC ¯
È Ê T ˆ4˘
fi F (T ) = x1/2 f n (T ) + (1 - x ) f s (T ) = - b Í1 + Á ˜ ˙
ÍÎ Ë TC ¯ ˙˚
T3
fi Ces = 3g 2
TC
Two Fluid Model – Gorter and Casimir
È Ê T ˆ4˘
Superconducting state: F (T ) = x1/2 f n (T ) + (1 - x ) f s (T ) = - b Í1 + Á ˜ ˙
ÍÎ Ë TC ¯ ˙˚
2
g ÊTˆ
Normal state: F (T ) = f n (T ) = - T 2 = -2 b Á ˜
2 Ë TC ¯

H c2
Recall = difference in free energy between normal and
8p
superconducting state
2
È Ê T ˆ2˘ H c (T )
2
ÊTˆ
= b Í1 - Á ˜ ˙ fi = 1- Á ˜
ÍÎ Ë TC ¯ ˙˚ H c (0) Ë TC ¯

The Gorter-Casimir model is an “ad hoc” model (there is no physical basis for the
assumed expression for the free energy) but provides a fairly accurate
representation of experimental results
 Model of F & H London (1935)
Proposed a 2-fluid model with a normal fluid and superfluid components

ns : density of the superfluid component of velocity vs

nn : density of the normal component of velocity vn

∂u
m = -eE superelectrons are accelerated by E
∂t
J s = - ens u

∂J s ns e 2
= E superelectrons
∂t m

Jn = s nE normal electrons
Model of F & H London (1935)

∂J s ns e 2
= E
∂t m

∂B
Maxwell: —¥ E = -
∂t

∂Ê m ˆ m
fi Á 2 — ¥ J s + B˜ = 0 fi — ¥ J s + B = Constant
∂t Ë ns e ¯ ns e 2

m
F&H London postulated: 2
— ¥ Js + B = 0
ns e
 Model of F & H London (1935)
combine with — ¥ B = m0 Js

m0 ns e 2
— B-
2
B=0
m

B ( x ) = Bo exp [ - x / lL ]
1
È m ˘ 2
lL = Í 2 ˙
Î m0 ns e ˚

The magnetic field, and the current, decay

exponentially over a distance λ (a few 10s of nm)
 Model of F & H London (1935)

1
È m ˘ 2
lL = Í 2 ˙
m
Î 0 s ˚
n e

From Gorter and Casimir two-fluid model

È Ê T ˆ4˘
ns μ Í1 - Á ˜ ˙
ÍÎ Ë TC ¯ ˙˚

1
lL (T ) = lL (0) 1
È ÊTˆ 4
˘ 2

Í1 - Á ˜ ˙
ÍÎ Ë TC ¯ ˙˚
 Model of F & H London (1935)

B
London Equation: l 2 — ¥ J s = - = -H
m0
— ¥ A= H
choose —i A = 0, An = 0 on sample surface (London gauge)

1
Js = - A
l 2

Note: Local relationship between J s and A

 Penetration Depth in Thin Films

Very thin films

Very thick films

Quantum Mechanical Basis for London Equation

Ïe * ¸
Ú
e2
J (r ) = Â Ì È
Îy *
— n y - y — ny *
˘
˚ - A ( rn)y y ˝d (r - rn ) dr1 - drn
n Ó 2mi mc ˛

In zero field A = 0 J ( r ) = 0 , y =y 0

Assume y is "rigid", ie the field has no effect on wave function

r (r ) e2
J (r ) = - A (r )
me
r (r ) = n
Pippard’s Extension of London’s Model
Observations:
-Penetration depth increased with reduced mean free path
- Hc and Tc did not change
- Need for a positive surface energy over 10-4 cm to
explain existence of normal and superconducting phase in
intermediate state
Non-local modification of London equation

1
Local: J =- A
cl
R
-
3s R ÈÎ Ri A (r ¢ )˘˚ e x

4px l c Ú
Non local: J (r ) = - 4
du
0 R
1 1 1
= +
x x0
 London and Pippard Kernels
Apply Fourier transform to relationship between
c
J (r ) and A (r ) : J (k ) = - K (k ) A (k )
4p

Effective penetration depth

•
p
Ú
2 dk
Specular: leff = Diffuse: leff =
p K (k ) + k 2 •
È K (k ) ˘
Ú
o
ln Í1 + 2 ˙ dk
o Î k ˚
 London Electrodynamics
Linear London equations
∂J s E 1
=- 2 —2 H - H =0
∂t l m0 l 2

together with Maxwell equations

∂H
— ¥ H = Js — ¥ E = - m0
∂t

describe the electrodynamics of superconductors at all T if:

– The superfluid density ns is spatially uniform
– The current density Js is small
 Ginzburg-Landau Theory
• Many important phenomena in superconductivity occur
because ns is not uniform
– Interfaces between normal and superconductors
– Trapped flux
– Intermediate state

• London model does not provide an explanation for the

surface energy (which can be positive or negative)

• GL is a generalization of the London model but it still

retain the local approximation of the electrodynamics
 Ginzburg-Landau Theory
• Ginzburg-Landau theory is a particular case of
Landau’s theory of second order phase transition

• Formulated in 1950, before BCS

• Masterpiece of physical intuition

• Grounded in thermodynamics

• Even after BCS it still is very fruitful in analyzing the

behavior of superconductors and is still one of the
most widely used theory of superconductivity
 Ginzburg-Landau Theory

• Theory of second order phase transition is based on

an order parameter which is zero above the transition
temperature and non-zero below

• For superconductors, GL use a complex order

parameter Ψ(r) such that |Ψ(r)|2 represents the
density of superelectrons

• The Ginzburg-Landau theory is valid close to Tc

Ginzburg-Landau Equation for Free Energy
• Assume that Ψ(r) is small and varies slowly in space

• Expand the free energy in powers of Ψ(r) and its

derivative

2
b 1 Ê e ˆ
*
h2
f = fn0 + a y + y + — - A˜ y +
2 4
* Á
2 2m Ë i c ¯ 8p
 Field-Free Uniform Case
b
f - fn0 = a y + y
2 4

f - fn 0 2 f - fn 0

>0 <0
a
y• = -
2

b
•

Near Tc we must have b >0 a (t ) = a ¢(t - 1)

H c2 a2
f - fn0 =- =- fiy and H c μ (1 - t )
2
At the minimum 8p 2b
 Field-Free Uniform Case
b a
f - fn0 = a y + y y• = -
2 4 2

2 b

b >0 a (t ) = a ¢(t - 1) fi y•
2
μ (1 - t )

It is consistent with correlating |Ψ(r)|2 with the density of

superelectrons
ns μ l -2 μ (1 - t ) near Tc

a2 H c2
At the minimum f - fn0 = - =- (definition of H c )
2b 8p
fi H c μ (1 - t )

which is consistent with H c = H c 0 (1 - t 2 )

 Field-Free Uniform Case
Identify the order parameter with the density of superelectrons

lL2 (0) Y (T ) 1 a (T )
2
1
ns = Y ∼ 2 fi = = -
2

lL (T ) lL2 (T ) Y (0) 2 n b

1 a 2 (T ) H c2 (T )
since =
2 b 8p

H c2 (T ) lL2 (T ) H c2 (T ) lL4 (T )
na (T ) = - and n b =
2

4p lL2 (0) 4p lL4 (0)

 Field-Free Nonuniform Case

Equation of motion in the absence of electromagnetic field

1 2
- * — y + a (T )y + b y y = 0
2

2m

Look at solutions close to the constant one

a (T )
y =y • +d where y •
2
=-
b

1
To first order: — 2d - d = 0
4m* a (T )

2 r /x (T )
Which leads to d ª e-
 Field-Free Nonuniform Case

- 2 r /x (T ) 1 2p n lL (0)
d ªe where x (T ) = =
2m a (T )
* m* H c2 (T ) lL (T )

is the Ginzburg-Landau coherence length.

x0
It is different from, but related to, the Pippard coherence length. x (T )
( )
1/2
1- t2

lL (T )
GL parameter: k (T ) =
x (T )

Both lL (T ) and x (T ) diverge as T Æ Tc but their ratio remains finite

k (T ) is almost constant over the whole temperature range

 2 Fundamental Lengths
London penetration depth: length over which magnetic field decay
1/2
Ê mb ˆ
*
Tc
lL (T ) = Á
Ë 2e 2a ¢ ˜¯ Tc - T
Coherence length: scale of spatial variation of the order parameter
(superconducting electron density)
1/2
Ê 2
ˆ Tc
x (T ) = Á * ˜
Ë 4m a ¢ ¯ Tc - T

The critical field is directly related to those 2 parameters

f0
H c (T ) =
2 2 x (T ) lL (T )
 Surface Energy

1
s ÈÎ H c2x - H 2 l ˘˚
8p
H 2l
: Energy that can be gained by letting the fields penetrate
8p
H c2x
: Energy lost by "damaging" superconductor
8p
 Surface Energy
1
s ÈÎ H c2x - H 2 l ˘˚
8p

Interface is stable if s >0

If x >> l s >0
Superconducting up to Hc where superconductivity is destroyed globally

x
If l >>x s <0 for H 2 > H c2
l
Advantageous to create small areas of normal state with large area to volume ratio
Æ quantized fluxoids

More exact calculation (from Ginzburg-Landau):

l 1
k= < : Type I
x 2
l 1
k= > : Type II
x 2
Magnetization Curves
 Intermediate State

Vortex lines in Pb.98In.02

At the center of each vortex is a
normal region of flux h/2e
 Critical Fields
Even though it is more energetically favorable for a type I superconductor to
revert to the normal state at Hc, the surface energy is still positive up to a
superheating field Hsh>Hc → metastable superheating region in which the
material may remain superconducting for short times.

Type I Hc Thermodynamic critical field

Hc
H sh Superheating critical field
k
Field at which surface energy is 0

Type II Hc Thermodynamic critical field

Hc2 = 2 k Hc
H c2
H c1
Hc2
1
(ln k + .008) H c (for k 1)
2k
 Superheating Field

Ginsburg-Landau:
0.9 H c
H sh ∼ for k <<1
k
∼ 1.2 H c for k ∼ 1
∼ 0.75 H c for k >> 1

The exact nature of the rf critical

field of superconductors is still
an open question
Material Parameters for Some Superconductors
 BCS
• What needed to be explained and what were the clues?

– Energy gap (exponential dependence of specific heat)

– Isotope effect (the lattice is involved)

– Meissner effect
 Cooper Pairs

Assumption: Phonon-mediated attraction between

electron of equal and opposite momenta located
within w D of Fermi surface

Moving electron distorts lattice and leaves behind a

trail of positive charge that attracts another electron
moving in opposite direction

Fermi ground state is unstable

Electron pairs can form bound

states of lower energy

Bose condensation of overlapping

Cooper pairs into a coherent
Superconducting state
 Cooper Pairs
One electron moving through the lattice attracts the positive ions.
Because of their inertia the maximum displacement will take place
behind.
 BCS

The size of the Cooper pairs is much larger than their spacing
They form a coherent state
BCS and BEC
 BCS Theory

0 q,1 q :states where pairs (q ,-q ) are unoccupied, occupied

aq , bq : probabilites that pair (q ,-q ) is unoccupied, occupied

BCS ground state

(
Y = P aq 0 q + bq 1 q
q
)
Assume interaction between pairs q and k
Vqk = -V if x q £ w D and xk £ w D
= 0 otherwise
 BCS
• Hamiltonian
H = Â e k nk + ÂVqk cq*c-* q ck c- k
k qk

ck destroys an electron of momentum k

cq* creates an electron of momentum k
nk = ck*ck number of electrons of momentum k

• Ground state wave function

( )
Y = P aq + bq cq*c-* q f0
q
 BCS

• The BCS model is an extremely simplified model of reality

– The Coulomb interaction between single electrons is
ignored
– Only the term representing the scattering of pairs is
retained
– The interaction term is assumed to be constant over a
thin layer at the Fermi surface and 0 everywhere else
– The Fermi surface is assumed to be spherical

• Nevertheless, the BCS results (which include only a very

few adjustable parameters) are amazingly close to the real
world
 BCS
Is there a state of lower energy than
the normal state?
aq = 0, bq = 1for x q < 0
aq = 1, bq = 0 for x q > 0

xq
yes: 2b = 1 -
2
q
x q2 + D 02

where

wD
1
-
r (0)V
D0 = 2 wD e
È 1 ˘
sinh Í ˙
Î r ( 0 ) V ˚
 BCS
Critical temperature

È 1 ˘
kTc = 1.14 w D exp Í - ˙
Î VN ( E )
F ˚

D (0) = 1.76 kTc

Coherence length (the size of the Cooper pairs)

uF
x 0 = .18
kTc
 BCS Condensation Energy

r (0)V D 02
Condensation energy: E s - En = -
2
Ê D 0 ˆ H 02
- N D0 Á ˜ =
Ë e F ¯ 8p
D0 / k 10 K
eF / k 104 K
 BCS Energy Gap
At finite temperature:
Implicit equation for the temperature dependence of the gap:

( )
tanh ÈÍ e 2 + D 2 / 2kT ˘˙
12
wD

Ú Î ˚ de
1
=
V r (0 ) 0 (e 2 + D2 )
12
 BCS Excited States

Energy of excited states:

ek = 2 x k2 + D 20
 BCS Specific Heat

Specific heat
Ê Dˆ T
Ces exp Á - ˜ for T < c
Ë kT ¯ 10
Electrodynamics and Surface Impedance
in BCS Model
∂f
H 0f + H ex f = i
∂t
A (ri , t ) pi
e
H ex =
mc
Â
H ex is treated as a small perturbation
H rf << H c
There is, at present, no model for superconducting
surface resistance at high rf field
R
-
R [ R ◊ A] I (w , R, T ) e
Ú
l
Jμ 4
dr similar to Pippard's model
R
c
J (k ) = - K (k ) A (k )
4p
K (0) π 0 : Meissner effect
 Penetration Depth
Ú
2 dk
l= dk (specular )
p K (k ) + k 2

1
Represented accurately by l ∼ near Tc
4
ÊT ˆ
1- Á c ˜
ËT ¯
 Surface Resistance

Temperature dependence
t4
-close to Tc : dominated by change in l (t )
(1 - t )
3
2 2

Tc -D
- for T < : dominated by density of excited states ∼ e kT
2
A Ê Dˆ
Rs ∼ w 2 exp Á - ˜
T Ë kT ¯
Frequency dependence
w 2 is a good approximation
Surface Resistance
Surface Resistance
Surface Resistance