Superconductivity

• Introduction
• Thermal properties
• Magnetic properties
• London theory of the Meissner effect
• Microscopic (BCS) theory
• Flux quantization
• Quantum tunneling

Dept of Phys

M.C. Chang
A brief history of low temperature (Ref: 絕對零度的探索)
• 1800 Charles and Gay-Lusac (from P-T relationship) proposed G. Amontons
that the lowest temperature is -273 C (= 0 K) 1700
• 1877 Cailletet and Pictet liquified Oxygen (-183 C or 90 K)
• soon after, Nitrogen (77 K) is liquified
• 1898 Dewar liquified Hydrogen (20 K)
• 1908 Onnes liquified Helium (4.2 K)

• 1911 Onnes measured the resistance of metal at

such a low T. To remove residual resistance, he chose
mercury. Near 4 K, the resistance drops to 0.

ρ Au
Hg

ρR

ρR
T 1913
 0.03K

1.14K

0.39K 5.38K 0.88K 1.09K

0.55K 9.50K 0.92K 7.77K 0.51K 0.0003K 0.56K 3.40K 3.72K

4.88K 0.12K 4.48K 0.01K 1.4K 0.66K 0.14K 4.15K 2.39K 7.19K

1.37K 1.4K 0.20K 0.60K

Tc's given are for bulk, except for Palladium, which has been irradiated with
He+ ions, Chromium as a thin film, and Platinum as a compacted powder
http://superconductors.org/Type1.htm
 Superconductivity in alloys and oxides Applications of superconductor

• powerful magnet
HgBa22Ca
HgBa Ca22Cu
Cu33O
O99
160 (under
(under pressure) • MRI, LHC...
Superconducting transition temperature (K)

pressure)

• magnetic levitation
140 HgBa
HgBa22Ca
Ca22Cu
Cu33O
O99
• SQUID (超導量子干涉儀)
120 TlBaCaCuO
TlBaCaCuO
• detect tiny magnetic field
BiCaSrCuO
BiCaSrCuO
• quantum bits
100
YBa
YBa22Cu
Cu33O
O77 • lossless powerline
80 Liquid Nitrogen
•…
temperature (77K)
60

40 (LaBa)CuO
(LaBa)CuO
Nb Ge
Nb Sn Nb33Ge
Nb33Sn
NbN
20 NbC NbN
NbC
Hg Pb Nb V
V33Si
Si

1910 1930 1950 1970 1990

Bednorz
Muller
From Cywinski’s lecture note 1987
• Introduction
• Thermal properties
• Magnetic properties
• London theory of the Meissner effect
• Microscopic (BCS) theory
• Flux quantization
• Quantum tunneling
Thermal properties of SC: specific heat

C Sel ∝ e − Δ ( 0 ) k BT
For different superconductors, The exponential dependence with T is
CS − CN called “activation” behavior and implies
. at TC
~ 143 the existence of an energy gap above
CN Fermi surface.
Δ ~ 0.1-1 meV (10-4~-5 EF )
• Connection between • Temperature dependence of Δ
energy gap and Tc (obtained from Tunneling)

Universal behavior of Δ(T)

Δ‘s scale with different Tc’s

1/2
Δ (T ) ⎛ T ⎞
= 1.74 ⎜1 − ⎟ for T ≈ TC
2Δ(0) ~ 3.5 kBTc Δ(0) ⎝ TC ⎠
• Entropy Al

⎛ ∂S ⎞
C =T ⎜ ⎟
⎝ ∂T ⎠ H
Less entropy in SC state:
more ordering

• free energy
⎛ ∂F ⎞
Al S = −⎜ ⎟
⎝ ∂T ⎠ H
C ⎛ ∂S ⎞ ⎛ ∂2F ⎞
=⎜ ⎟ = −⎜ 2 ⎟
FN-FS T ⎝ ∂T ⎠ H ⎝ ∂T ⎠ H
= Condensation energy 2nd order phase
～10-8 eV per electron! transition
More evidences of energy gap
• Electron tunneling

• EM wave absorption

2Δ suggests excitations
created in “e-h” pairs
2Δ
ν= = 480 GHz (microwave)
h
Magnetic property of the superconductor

• Superconductivity is destroyed by a strong magnetic field.

Hc for metal is of the order of 0.1 Tesla or less.

• Temperature dependence of Hc(T) All curves can be collapsed onto

a similar curve after re-scaling.

{
Hc (T ) = Ho 1− (T Tc )
2
}
normal

sc
 Critical currents (no applied field)

Hi
Radius, a

Magnetic field
Current

G G 4π ca
∫ H ⋅ d A =
c
i so ic = Hc
2
The critical current density of a long
From Cywinski’s lecture note

thin wire is therefore

cHc
jc = (thinner wire has larger Jc)
2πa
jc~108A/cm2 for Hc=500 Oe, a=500 A

• Jc has a similar temperature Cross-section through a

dependence as Hc, and Tc is similarly niobium–tin cable
lowered as J increases.
Phys World, Apr 2011
Meissner effect (Meissner and Ochsenfeld, 1933)
A SC is more than a perfect conductor

Lenz law not only dB/dt=0

but also B=0!

Perfect
diamagnetism

different same
Superconducting alloy: type II SC
partial exclusion and remains superconducting at high B (1935)
(also called intermediate/mixed/vortex/Shubnikov state)

STM image
NbSe2, 1T, 1.8K

pure In

• HC2 is of the order of 10~100 Tesla (called hard, or type II, superconductor)
Comparison between type I and type II superconductors

B=H+4πM

Hc2

Lead + (A) 0%, (B) 2.08%, (C) 8.23%, (D) 20.4%

Indium
Areas below the curves (=condensation energy)
remain the same!

Condensation G G
dF = − M ⋅ dH FN ( H c ) = FS ( H c )
energy (for type I)
G G 1 FN ( H c ) = FN (0) for nonmagnetic material
For a SC, dFS =
( H is Ba in Kittel ) 4π
HdH
H c2
∴ΔF = FN (0) − FS (0) =
→ FS ( H ) − FS (0) =
H2 8π
8π (Magnetic energy density)
• Introduction
• Thermal properties
• Magnetic properties
• London theory of the Meissner effect
• Microscopic (BCS) theory
• Flux quantization
• Quantum tunneling
London theory of the Meissner effect (Fritz London and Heinz London, 1934)

Carrier density Two-fluid model:

nn
ns + nn = n = constant
ns
• Superfluid density ns σ=∞
nn e 2τ
Tc • Normal fluid density nn σn =
m
T

G
Assume G G 1 ∂B
like free Eq.(1) + ∇ × E = −
G G charges c ∂t
dJ s ns e 2 E G
(1) = G G n e ∂B
( )
2
d
dt
G
m
G ∇ × Js = − s
(2) Jn = σ nE dt mc ∂t
London G ns e 2 G
where proposed ∇ × Js = − B
mc It can be shown that
G G
J s = −ens vs G ns e 2 G ▽ψ=0 for simply
G G JS = − A + ∇φ connected sample
J n = −enn vn mc (See Schrieffer)
G G 4π G G
use ∇ × B = J s and G 4π ns e 2 G
B
c ∇2 B = B ≡
G G G
∇ × ( ∇ × v ) = ∇ ( ∇ ⋅ v ) − ∇ 2v
mc 2 λL2
• Penetration length λL Outside the SC, B=B(x) z

d 2B
λ 2
L 2
=B
dx
→ B( x) = B0 e − x / λL (expulsion of
magnetic field)

mc 2
λL = ≈ 170 A if nS =1023 /cm3
4π nS e 2

G G 4π G
∇× B = Js
• Temperature dependence of λL c
c dB cB0 − x / λL also
∴ J sy = − = e
tin
4π dx 4πλL decays

Predicted λL(0)=340 A,
measured 510 A
• Higher T, smaller nS
λ (0)
λ (T ) = 1/2
⎡1 − (T / TC )4 ⎤
⎣ ⎦
Coherence length ξ0 (Pippard, 1939)
• In fact, ns cannot remain uniform near a surface. ns
The length it takes for ns to drop from full value to

surface
0 is called ξ0
ξ0
• Microscopically it’s related to the range
of the Cooper pair.
superconductor
• The pair wave function (with range ξ0) is a
x
superposition of one-electron states with energies
within Δ of EF (A+M, p.742).

p Δp
Energy uncertainty
of a Cooper pair
≈Δ
m
• Therefore, the spatial range of the variation of nS

= =vF =vF
ξ0 ≈ = ↔ from BCS theory
Δp Δ πΔ
ξ0 ~ 1 μm >> λ for type I SC
Penetration depth, correlation length, and surface energy
Type I superconductivity Type II superconductivity
• ξ0 > λ, surface energy is positive • ξ0 < λ, surface energy is negative

From Cywinski’s lecture note

• smaller λ, cost more energy to • smaller ξ0, get more “negative”
expel the magnetic field. condensation energy.
• When ξ0 >> λ (type I), there is a • When ξ0 << λ (type II), the surface
net positive surface energy. Difficult energy is negative. Interface may
to create an interface. spontaneously appear.
Vortex state of type II superconductor (Abrikosov, 1957)
Normal
core
• the magnetic flux φ in a vortex is
isc
always quantized (discussed later).

• the vortices repel each other slightly.

• the vortices prefer to form a triangular

lattice (Abrikosov lattice). 2003

• the vortices can move and dissipate energy

(unless pinned by impurity ← Flux pinning)

0 H
Hc1 Hc2

-M

From Cywinski’s lecture note

Estimation of Hc1 and Hc2 (type II)

• Near Hc1, there begins with a single

vortex with flux quantum φ0, therefore
φ0
πλ 2 H c1 ≈ φ0 → H c1 ≈
πλ 2

• Near Hc2, vortex are as closely packed

as the coherence length allows, therefore
φ0
N πξ 0 2 H c 2 ≈ N φ0 → H c 2 ≈
πξ 0 2
2
H ⎛λ⎞
Therefore, c 2 ≈ ⎜ ⎟
H c1 ⎝ ξ 0 ⎠

Typical values, for Nb3Sn,

ξ0 ～ 34 A, λL ～ 1600 A
Origin of superconductivity?
• Metal X can (cannot) superconduct because its atoms can
(cannot) superconduct?
Neither Au nor Bi is superconductor, but alloy Au2Bi is!
White tin can, grey tin cannot! (the only difference is lattice structure)

• good normal conductors (Cu, Ag, Au) are bad superconductor;

bad normal conductors are good superconductors, why?
• What leads to the superconducting gap?
• Failed attempts: polaron, CDW...

• Isotope effect (1950):

It is found that Tc =const × M-α

α～ 1/2 for different materials mercury
lattice vibration?
Brief history of the theories of superconductors

• 1935 London: superconductivity is a quantum phenomenon

on a macroscopic scale. There is a “rigid” (due to the energy
gap) superconducting wave function Ψ.
2003
• 1950 • Frohlich: electron-phonon interaction maybe crucial.

• Reynolds et al, Maxwell: isotope effect

• Ginzburg-Landau theory: ρS can be varied in space.

G G
Suggested the connection ρ S (r ) =| ψ (r ) |2
and wrote down the eq. for order parameter Ψ(r) (App. I)

• 1956 Cooper pair: attractive interaction between electrons (with

the help of crystal vibrations) near the FS forms a bound state.

• 1957 Bardeen, Cooper, Schrieffer: BCS theory 1972

Microscopic wave function for the

condensation of Cooper pairs.

Ref: 1972 Nobel lectures by Bardeen, Cooper, and Schrieffer

Dynamic electron-lattice interaction → Cooper pair

+++

Effective attractive interaction ～ 1 μm

between 2 electrons (range of a Cooper pair;
(sometimes called overscreening) coherence length)
 Cooper pair, and BCS prediction

• 2 electrons with opposite momenta (p↑,-p↓) can form a bound

state with binding energy (the spin is opposite by Pauli principle)
1
−
Δ( 0) = 2=ω D e D ( E F )Vint
, see App. H
• Fraction of electrons involved ～ kTc/EF ～ 10-4
• Average spacing between condensate electrons ～ 10 nm
2Δ(0) ~ 3.5 kBTc • Therefore, within the volume occupied by the Cooper pair, there
are approximately (1μm/10 nm)3 ～ 106 other pairs.
• These pairs (similar to bosons) are highly correlated
and form a macroscopic condensate state with (BCS result)
1
−
. =ω D e
k B TC = 113 D ( E F )Vint

=ωD ≤ 500 K , D( EF )Vint ≤ 1 / 3

∴ Tc ≤ 500e −3 = 25 K (～upper limit of Tc)
Energy gap and Density of states

D(E)

~ O(1) meV

• Electrons within kTC of the FS have their energy lowered

by the order of kTC during the condensation.

• On the average, energy difference (due to SC transition)

per electron is
T 1
k BTC C  0.1 meV × 4  10−8 eV
TF 10
Families of superconductors

Cuperate

(iron-based)
T.C. Ozawa 2008

Conventional
BCS

Heavy fermion

F. Steglich 1978

wiki
• Introduction
• Thermal properties
• Magnetic properties
• London theory of the Meissner effect
• Microscopic (BCS) theory
• Flux quantization
• Quantum tunneling (Josephson effect, SQUID)
Flux quantization in a superconducting ring
(F. London 1948 with a factor of 2 error, Byers and Yang, also Brenig, 1961)

G q ⎛ = = ⎞
• Current density operator j= ⎜ψ * ∇ψ −ψ ∇ψ * ⎟ , q = −e
2m ⎝ i i ⎠
G q* ⎡ ⎛ = ⎤ q * = − 2e
*
q* K ⎞ ⎛= q* K ⎞
• SC, in the presence of B j= ⎢ψ * ⎜ ∇ − A ⎟ψ + ψ ⎜ ∇ − A ⎟ ψ *⎥
2m* ⎢ ⎝ i
⎣
c ⎠ ⎝ i c ⎠ ⎥⎦ m* = 2m
G
let ψ =|ψ |eiφ and assume |ψ | vary slowly with r

⎛ e= 2e 2 G ⎞ 2 London eq. with

then j = − ⎜ ∇φ + A ⎟ |ψ |
⎝ m mc ⎠ ns = 2|ψ |2
G G
• Inside a ring
v∫ j ⋅dA = 0
G G =c G =c
⇒ v∫ A ⋅ d A = −
2e ∫
v ∇ φ ⋅ d A = −
2e
Δφ

hc hc
∴ flux | Φ |= n = nφ0 , φ0 ≡ = 2 × 10−7 gauss-cm 2
2e 2e
• φ0 ～ the flux of the Earth's magnetic field
through a human red blood cell (~ 7 microns)
Single particle tunneling (Giaever, 1960)

• SIN
dI/dV

20-30 A thick

• SIS

For T>0
(Tinkham, p.77)

Ref: Giaever’s 1973

Nobel prize lecture
Josephson effect (Cooper pair tunneling) Josephson, 1962
1) DC effect:
There is a DC current through SIS in the absence of voltage.

1973
iθ 1 iθ 2
ψ 1 =|ψ 1 |e ψ 2 =|ψ 2 |e

Giaever

ψ = nS / 2 ( eiθ − K ( x + d / 2) + eiθ )
tunneling
1 2 + K ( x − d /2)
Josephson
tunneling
ie=nS
j= Ke − Kd ( −ei (θ1 −θ2 ) + ei (θ2 −θ1 ) )
2m
= j0 sin δ 0
j0 ≡ e=nS Ke − Kd / m, δ 0 ≡ θ1 − θ 2 2Δ / e
2) AC Josephson effect
Apply a DC voltage, then there is a rf current oscillation.

ψ = N − 1 ψˆ N ∝ e − i ( E N − EN −1 ) t / =
= e − iμt / =
→ θi (t ) = − μi t / = + θ i (i = 1, 2)

μ1 − μ2 = −2eV
2eV ⎛ 2eV ⎞ (see Kittel, p.290 for an
∴δ = t + δ0 ⇒ j = j0 sin ⎜ t + δ0 ⎟
= ⎝ = ⎠ alternative derivation)

• An AC supercurrent of Cooper pairs with freq. ν=2eV/h, a

weak microwave is generated.
• ν can be measured very accurately, so tiny ΔV as small as
10-15 V can be detected.
• Also, since V can be measured with accuracy about 1 part in
1010, so 2e/h can be measured accurately.
• JJ-based voltage standard (1990):
1 V ≣ the voltage that produces ν=483,597.9 GHz (exact)
• advantage: independent of material, lab, time (similar to the
quantum Hall standard).
3) DC+AC: Apply a DC+ rf voltage, then there is a DC current

V = V0 + υ cos ωt
⎡ 2e ⎛ υ ⎞ ⎤
j = j0 sin ⎢ ⎜ V0t + sin ωt ⎟ + δ 0 ⎥
⎣= ⎝ ω ⎠ ⎦
⎛ 2eυ ⎞ ⎛ 2eV0 ⎞
=j0 ∑ (−1) n J n ⎜ ⎟ ⎜
sin t − nω t + δ 0⎟
n ⎝ =ω ⎠ ⎝ = ⎠
=ω
⇒ there is DC current at V0 = n
2e
• Another way of providing a voltage standard

Shapiro steps (1963)

given I, measure V
SQUID (Superconducting QUantum Interference Device)
j = j0 sin δ a + j0 sin δ b
⎛ δ − δ ⎞ ⎛ δ + δb ⎞
= 2 j0 cos ⎜ a b ⎟ sin ⎜ a ⎟
⎝ 2 ⎠ ⎝ 2 ⎠
G G =c G
Similar to v∫ A ⋅ d A = − v∫ ∇θ ⋅ d A
2e
2e G G
=c C∫
We now have A ⋅ d A = θb1 − θ a1
1

2e G G
=c C∫
A ⋅ d A = θ a 2 − θb 2
2

2e G G φ
⇒ δ a − δb = ∫
=c v
C
A ⋅ d A = 2π
φ0
⎛ 2π φ ⎞ The current of a SQUID
∴ jmax = 2 j0 cos ⎜ ⎟ with area 1 cm2 could
⎝ 2 φ0 ⎠ change from max to min
by a tiny ΔH=10-7 gauss!

For junction with finite thickness

 SuperConducting Magnet

Non-destructive testing

MCG, magnetocardiography

MEG, magnetoencephlography
Super-sentitive photon detector
Transition edge sensor

semiconductor detector superconductor detector

科學人,2006年12月