Ch.

3 Superconductivity Resistivity of a metal as a function of temperature:

(θD = Debye temperature)
Read one of the following references:
ρ ∝ T for T ≫ θ D
1. F. J. Owens and C. P. Poole, Jr., “The new
superconductors”, (QC611.95 .O89 2002 ebook). ρ
2. V L Ginzburg & E A Andryushin, “Superconductivity” 2/e
2
(QC611.92 .G56 2004 ebook)
3. G. Vidali, Superconductivity: the next revolution? (1993) (due to lattice
(QC611.92V53) vibration)
1 ρ → ρ0
4. P.F. Dahl, Superconductivity (QC611.92D34 1992)
for T → 0
( ρ − ρ 0 ) ∝ T 5 for T ≪ θ D
(Bloch-Gruneisen law)

0 T
(I) Introduction Origin of resistance:
• Solids have different conductivity: 1 ρ0 (residual resistivity) due to defects, impurities, etc.
Solid materials 2 temperature dependence of ρ depends on electron-ion
scattering & lattice vibrations.
conductor semiconductor insulator ρ ∝ T for T ≫ θ D
( ρ − ρ 0 ) ∝ T 5 for T ≪ θ D
Superconductor metal semimetal
θ D = 315K for Cu
• Superconducting materials: elements, alloys, compounds. θ D = 230 K for Pt
(classified as Type I & Type II)
What happens when T→ 0?
• Type I superconductor has properties:
perfect conductor (ρ = 0) If there are no defects and impurities then ρ = ρ 0= 0 at T = 0.
Meissner’s effect (B = 0) Ideal periodicity ⇒ No resistance!
• Applications: superconducting magnet, Maglev, SQUID, … (impossible!)

Origin of conductivity in metal Classical picture: At T = 0, all ions are standing still.
Quantum picture: Even at T = 0, electrons are still moving
inside atoms while the atoms can almost stand still.
metal
Conjectures near 1900:

ion James Dewar: lim ρ (T ) = 0

conduction electrons T →0
(immobile (mobile) (no lattice vibration ⇒ very pure metal is a perfect conductor)
but vibrating)
Lord Kelvin, Kamerlingh Onnes: lim ρ (T ) = +∞
T →0
Typically, the number of conduction electrons in a solid (freezing of electron gas)
n ≈ 1022 electrons/cm3.
Note: At higher pressure, n (and thus conductivity) increase. No theory was known at that time!
e.g. hydrogen can be forced to become metallic at Need experimental observation.
~ 1-10 megabars. (1 bar ≈ 1 atmospheric pressure)
(http://www.nature.com/news/1998/000622/full/news000622-6.html)
Motivation: (II) The big chill: low temperature experiments.
As T
0, gas becomes liquid, then solid in which atoms
1883 Olszewski & Wroblewski liquefied
have only minimum thermal vibrations.
oxygen (90.2 K) (boiling point at 1 bar) and
How about conduction electrons in metals at low nitrogen (77.3 K).
temperature? 1898 James Dewar liquefied hydrogen (20.5 K).
1908 K. Onnes liquefied helium (4.2 K),
To answer this question, need QM. obtained 1.7 K at reduced pressure.
Electron sees periodic potential and Onnes obtained Nobel prize in 1913.
has wave behavior (wave-particle duality).
Lowest temperature recorded on Earth: in Vostok (Russian)
-89.2 °C (confirmed): July 21, 1983
By solving Schrödinger’s equation (a wave equation), -91 °C (unconfirmed): 1997 (colder than dry ice!)
we can find allowed energy states for electrons and The coldest temperature achieved in physics labs:
electron in these states can travel through the solid more 100 pK (in 2000)
freely! (http://ltl.tkk.fi/wiki/LTL/World_record_in_low_temperatures)
Typical result (covered in PHY 4450) is shown in a graph: Temperature in deep space ~ 2.735 K.

H.K. Wong (CUHK 2011) PHY 3202 (3-1)

 Onnes then measured the resistance of pure substances at For more than 4 decades since Onnes’s discovery,
these low temperatures using 4-point probe technique (taught superconductors were
in PHY2822). • rare
• unpredictable
Onnes’s motivation:
• poorly understood
1. to find out lim ρ (T ) = ?
T →0 • little connection with normal state properties.
2. to develop new thermometer using R(T). 1929 Hans Meissner found the barely metallic compound
(At that time, gas thermometers (bulky) were used for low CuS to be superconducting whereas elemental Cu was
temperature experiments.) not.
1930s Hans Meissner found superconducting intermetallic
Results: borides.
Smaller ρ0 for purer samples, as expected. 1950s John Hulm & Bernd Matthias discovered
He also showed that lim ρ (T ) = +∞ is not correct. superconducting intermetallic alloys and compounds.
T →0

Note: Exceptional cases were found in some metals, Superconducting elements known today:
dρ
< 0 at very low temperatures. (New superconductors at high pressure)
dT
They are Kondo effect (1964) and H
weak localization effects (1978). Li Be B C N O F
Onnes thought ρ0 = 0 could be achieved by purification. He Na Mg Al Si P S Cl
looked for very pure samples and limited his studies to pure K Ca Sc Ti V Cr Mn Fe Co Ni Cu Zn Ga Ge As Se Br
metals only. Rb Sr Y Zr Nb Mo Tc Ru Rh Pd Ag Cd In Sn Sb Te I
Then he studied Hg because it was easier to purify Hg. The
Cs Ba La Hf Ta W Re Os Ir Pt Au Hg Tl Pb Bi Po At
experiment was done by G. Holst (a graduate student) and he
found a rapid drop of R to zero at very low temperatures. Fr Ra Ac Ku
Originally Holst thought it was due to short-circuit and thus
modified his sample. The resistance drop was confirmed by a superconducting
mistake on April 28, 1911. (J. de Nobel, “The discovery of superconducting at high pressure
superconductivity”, Physics Today, Sept. 1996, pp.40-42.) magnetic

TC = 4.15 K for Hg Features:

• 10-5 Ω is due to limit of • “Good metals”
instrument for (e.g. Na, K, Cu, Ag, Au, Pt, etc)
measurements. are NOT superconductors. (How to check?)
0.10
• Lower temperature can (Exception: Li was found to be superconducting at high
R pressure in 2002. See p.12.)
(Ohm) be reached by pumping
liquid helium bath. Recall: Electrons in “good metals” don’t scatter much.
0.05 • More superconductors ⇒ Need bad conductivity to enhance superconductivity?
were found later by his (No! Superconductivity needs appropriate lattice vibrations.)
group: • Magnetic elements are not superconducting.
In (TC = 3.4 K) (How does magnetic field affect superconductivity?)
Sn (TC = 3.7 K) (Exception: Fe was found to be superconducting at high
0
4.0 4.1 4.2 4.3 4.4 Pb (TC = 7.2 K) pressure in 2001. See p.12.)
TC T (K)
Onnes called it supra-conductivity and later superconductivity. (III) How to make sure that R = 0?
Nobel Prize (1913) For small resistance measurements, 4-point probe method
must be used to eliminate the contact problems.
Heike Kamerlingh Onnes
(1853 - 1926) The smallest resistance that can be detected depends on the
sensitivity of voltmeters or multimeters:
10 µV sensitivity in PHY2811/2822 lab
1 µV sensitivity in PHY3811/3822 lab.
Voltmeters with 1 nV or better sensitivity are commercially
“for his investigations on available. (www.keithley.com)
the properties of matter at The most sensitive voltmeter is made of a superconducting
low temperatures which device called SQUID (see p.10 & p.13).
led, inter alia, to the
production of liquid
helium”

H.K. Wong (CUHK 2011) PHY 3202 (3-2)

 A simple approach to test R =0 is to use a (IV) Magnetic property of superconductor
superconducting ring (first by Tuyn and Onnes). Onnes’ Dream
The current in the ring can be induced by Faraday’s law: Novel property ⇒ Novel application:
Ha R = 0 in superconductor ⇒ we can pass very high current
through a superconducting wire.
This suggests that we can use superconducting wire to build a
magnet (superconducting magnet, SM) to generate strong
I and stable magnetic field. SM is just a solenoid of
superconducting wire with high current.
1 First cool the ring to T < TC 2 Then remove the Onnes received a grant to build a 10 T SM.
in a magnetic field Ha. applied field Ha. But it is not that simple! He never attained the goal because
he worked only on elements (now classified as Type I
3 The induced current in the superconducting (SC) ring
superconductors). It took half a century for other scientists to
flows persistently. (If the ring is made of ordinary metal,
achieve this goal.
the induced current decays rapidly.) Tuyn and Onnes did
Instead he found critical current IC and critical field HC.
not observe any decay of current in 24 h.
This can be understood by Faraday’s law. The superconducting state can be maintained
The ring is equivalent to an RL circuit. only when I < IC and H < HC.
R = the total resistance of the ring (assuming R ≠ 0 first). For all pure metals Onnes studied,
L = the inductance of the ring HC ~ few 102 Oe (BC ~10-2 T).
 
In an external magnetic field Ba = µ0 Ha , Hc depends on temperature:   T 2 
d Φa dI H H C (T ) = H 0 1 −    [2]
the induced voltage in the ring is V = − = L + IR   TC  
  dt dt H (T )
where Φa = ∫ Ba ⋅ da
C
H0
ring
Normal state
dBa d Φa (R ≠ 0)
If = 0, then − = 0 or I (t ) = I0e
−( R / L)t
= I0 e− t / τ.
dt dt
L Superconducting
The decay time constant is τ = state (R = 0)
R
If R = 0, then I(t) = I0 (persistent current).
0 TC
T
Can R = 0 be verified by the experiment on SC ring?
For a ring of radius r and cross-sectional radius a, Current-carrying wire generates magnetic field:
2a
L = µ0 r[log e (8r / a ) − 2] r
r I
and R = 2r ρ / a 2
µ0 I
where ρ is the resistivity of the ring material. I The magnetic field on the wire surface is B = .
2π r
Thus ρτ = 12 µ0 a [log e (8r / a ) − 2]
2
[1] Assume there is no external magnetic field (Ha= 0).
For a Cu ring at room temperature In order to maintain superconducting state, the current must
(ρ ≈ 1.56 µΩ.cm, a = 1.0 mm and r = 10 cm), τ = 0.19 ms. not exceed a critical value IC so that the generated field on
For a superconducting ring with the same a and r, if, for over wire surface B ≤ µ0 H C .
µ0 I C
a period t, the supercurrent persists without appreciable Critical current: ≈ µ0 H C ⇒ I C ≈ 2π rH C
change (assuming <1% of drop in I, which can be easily 2π r
detected by a Hall effect sensor), We will show later that in a (Type I) superconductor, the
then τ is determined by I (t ) = I0e− t / τ . supercurrent is confined to the surface.
Assume I(t) > 0.99I0 in t = 1 year. Then τ > 3x109 s and due to I
 
thus by Eq. [1], ρ < 1x10-19 Ω.cm. If Ha≠ 0 then we need H a + H I < H c
This is how we estimate an For applied magnetic field Ha= 0, IC ∝ HC .
upper limit of resistivity of superconductor.   T 2 
∴
One can never show R = 0 by experiment. I C = I 0 1 −   
The measurement of I(t) in a ring sample is thus the most
  TC  
sensitive method to set an upper limit of resistivity.
In 1962, the measured upper limits are For different applied IC(Ha,T)
< 3x10-23 Ω-cm for conventional Type I superconductor field Ha, the
(J. Appl. Phys, 33, 748 (1962)), superconductor has
< 2x10-19 Ω-cm for conventional Type II superconductor different IC(T) curve.
In general, the IC(Ha,T)
(Phy. Rev. Lett. 9, 306 (1962)).
surface
 depends on
(The classification of superconductors will be given later.)
H a direction.


Ha

H.K. Wong (CUHK 2011) PHY 3202 (3-3)

 
(V) The Meissner effect Meissner & Ochsenfeld ’s discovery: B = 0 inside SC
conductivity σ = ∞
H
A:
• Is superconductor just a perfect conductor only?
No! The evidence: Some alloys and even some amorphous HC
metals (with low conductivity) are superconductors.
Superconductor has peculiar magnetic property. It is now B A
known as the Meissner effect. B:

• Meissner & Ochsenfeld (1933) used a solid Pb cylinder 

0
C TC
T B=0
in order to
1. look for any change in magnetic properties in From state A to state B, all magnetic field
association with the superconducting transition. inside superconductor is expelled.
2. find out whether the supercurrent shielded the Supercurrent is induced on the surface such
that the total magnetic field inside the sphere C:
magnetic field completely.
is zero. From state B to state C, the current

drops to zero in order to maintain B = 0
inside superconductor.
 H
What is B inside SC? Along another path:
A
D
C: A:
For perfect
 conductor,
  σ =∞.
Since J = σ E < ∞ , E = 0 . HC
(Note: in electrostatics, E = 0 ) A
 ∂B
Faraday’s law: ∇ × E = −
∂t
∂B D:
∴ = 0. TC T
∂t 0 C D

⇒ B inside a perfect conductor is time-independent.
 Now the state at C is path-independent.
But B = ?  C:
Are superconductors simply perfect conductors? The Meissner efect : B = 0
inside (Type I) superconductor.
Now consider the cooling process:

H 1 A (solid) lead sphere is Supercurrent in superconductors:

cooled from state A to A: Magnetic

The Meissner effect: B = 0 inside (Type I) SC. 
HC state B in a magnetic field. field lines   ∂E
B A By Maxwell-Ampere’s law, ∇ × B = µ0 J + µ0ε 0
∂t
or ∫ B ⋅ d ℓ = µ0 I enclosed + µ0ε 0 ∂ ∫ E ⋅ da
  
B: ∂t
(for a closed loop inside SC.)
0 T 
C TC Remember that we have shown that E = 0 inside SC .

2 ∴ J =0 At any point inside SC
C:
From state B to state C, the magnetic
field is reduced to zero at the same T. Current in a (Type I) SC is confined to its surface.
If a superconductor is just a perfect Therefore if we put a SC in a magnetic field H < HC ,
conductor, then by Faraday’s law, the then a supercurrent is induced
 on the SC surface to ensure that
induced supercurrent flows the total magnetic field B inside the SC is zero.
persistently on the surface of sphere
This current is called the screening current.
(similar to SC ring).
H Now we cool the Screening current distribution is unique & can be calculated:
sphere along another A: Consider a (Type I) superconducting sphere (at T < TC)
path A
D
C.  
HC
placed in a uniform magnetic field Ba = µ0 H = µ0 Hzˆ .
A
Because of the Meissner effect, a surface
 current is induced
so that inside the superconductor B = 0 .
D: ẑ
  
0
C TC D
T B = Ba + Bi = 0
 θ
From state A to state D, the magnetic field is Bi is due to the surface current density 
switched off. The induced current on the
 3  ↑ B a
surface of sphere decays quickly (because R ≠0). K =− Ba sinθφˆ ( B = 0)
No current is on the sphere in state C. C: 2 µ0
∴The final state in state C is thus (Here we use spherical coordinates ( r̂,θˆ,φˆ) .
path-dependent & this is not acceptable. For details, see Appendix 10.)
⇒ superconductor is not just a perfect conductor.

H.K. Wong (CUHK 2011) PHY 3202 (3-4)

 Penetration depth: (VIII) Classification of superconductors
Meissner effect: B = 0 inside (Type I) superconductor. Type I superconductors: The superconductors that are metal
elements except Nb & V. They have the properties
Actually the magnetic field can penetrate an ultra-thin depth listed in (VII).
underneath the surface of superconductor. (The thickness of Type II superconductors: Nb, V, alloys & compounds.
this layer is characterized by a penetration depth λ).
1930s: Schubnikow (Russian) discovered that some Pb
The supercurrents also flow within this layer.
alloys had HC much higher than HC (Pb). His paper
 was not noticed by the West due to the World War II.
Penetration depth depends on T: B
−1/ 2
1961: Scientists in Bell Lab discovered Nb3Sn (an
  T 4   intermetallic compound) with high IC and HC.
λ (T ) = λ0 1 −    B=0
These superconducting alloys and compounds have
  TC  
interesting properties: magnetic field can penetrate the
constant outside inside material in a novel way.
superoconductor
(Pb: λ0 =370 Å) (History: J.K. Hulm, J.E. Kunzler and B.T. Matthias, “The road to
superconducting materials”, Physics Today, Jan. 1981, pp.34-43.)
(VI) Superconductors are diamagnetic. For Type II
 superconductor,
By definition, Magnetization ( M ) is the (HC2)
   there are two
B = µ0 ( H + M ) magnetic moment per unit
 critical fields:
Inside SC, B = 0
 volume: 
M = lim i
∑ mi HC1(T) & HC2 (T)
  V →0 V
⇒ M = − H (diamagnetic)
   (H is the applied magnetic field.) mixed state
or B = Ba + Bi (HC1 < H < HC2)
magnetization due to is neither fully
applied field induced current S superconducting
magnet 
 (to generate Ba )
nor fully normal.
B = 0 inside SC N
   (HC1)
⇒ Bi = − Ba Ba  similar to Meissner state
N
Bi SC
(with induced
(0 < H < HC1)
∴SC can be levitated by a S supercurrent)
complete
magnetic field! (a strong force) Meissner state TC
Magnetic levitation for low temperature superconductors In mixed state, magnetic field penetrates the superconductor in
a form of regular triangular array of fluxoids (magnetic field
line bundles). Each fluxoid contains a normal core, surrounded
by a vortex of supercurrent.
magnet fluxoid
superconductor

low
temperature

superconductor ↑H
(applied supercurrent
at T < TC .
field) magnetic
field line
(Note: We will see that magnetic levitation for high superconducting normal
temperature superconductors is different.) region region
(VII) Summary Vortices in mixed
state when
 viewed
Superconductor possesses along H :
1. infinite conductivity (ρ = 0) 
2. perfect diamagnetism (the Meissner effect B = 0 )
provided that
T < TC (critical temperature or transition temperature)
H < HC (critical field)
I < IC (critical current)
The transition is an example of critical phenomenon.
Many properties change at TC: specific heat, thermo-emf, Hall h
effect, absorption of ultrasound, absorption of infrared, ... The magnetic flux of each fluxoid: φ0 = = 2.07 × 10−15 T-m 2
2e
For an average magnetic field B in the sample,
the number of fluxoids per unit area is B/φ0.
φ0 depends on h; indicating that we need QM to understand its
origin.

H.K. Wong (CUHK 2011) PHY 3202 (3-5)

 Immediately after superconductor Nb3Sn was found in 1961,
The mixed state allows very high HC2: high field superconducting magnet (SM) (Onnes’s dream)
was built at General Electric using Nb3Sn, and later in other
BC2=µ0HC2 TC labs using superconductor ~ 1 mm
Nb-Ti.
at T = 0
Modern SMs are
Nb3Ge 38 T 23 K built using
superconducting
Nb3Sn 24.5 T 18 K cable composed of
Ni-Ti filaments
V3Si 23.5 T 16.9 K embedded in pure
Cu.
Tesla
Like Type I superconductor, Type II superconductors have
small HC1.
Why is HC of Type I superconductors or Wire
HC1 of Type II superconductors so small? Cable (composed
(composed of of many
many wires) filaments)
Recall: Magnetic field stores energy with energy density
B 2 / 2 µ0 
SC ↑ Ba
Consider a superconductor (of volume V)
in a uniform magnetic field Ba. filament
In normal state, the energy of magnetic field
within the superconductor is ( Ba2 / 2 µ0 )V.

In superconducting state (at T < TC), B = 0 inside SC.
⇒ Need to exclude energy (( µ0 H C ) 2 / 2 µ0 )V
out of superconductor. (The New
That explains why HC & HC1 are so small (~ 102–103 Oe). Quantum
Universe,
Fig. 7.23)

• Fluxoids can move around: (IX) Decoration experiment (used to visualize the fluxoids)
When a current is passing through a superconductor, it may (first done in 1966): like mapping the magnetic field
hit the fluxoids and interact with the magnetic field in the distribution of a magnet using Fe dust in high school.
fluxoids. Or the magnetic field of this current interacts with The sample
the supercurrents surrounding the fluxoids. As a result, the (superconductor)
fluxoids can be moved by Lorentz force, resulting in energy must be in mixed
state first (T<TC & (to generate
loss & very small IC. magnetic
HC1< H <HC2).
field)
• Flux pinning:
Iron vapor
The fluxoids can be “pinned” (or fixed) by crystal (“smoke”) is (to block
imperfections (grain boundaries, defects, impurities, etc.), generated when direct vapor)
resulting in very high IC . the tungsten
filament is heated.
• Stable magnetic levitation demo with high temperature Vacuum is
superconductors (HTS) is due to strong pinning force. required for large mean
free path of vapor. (Scientific American, Feb 1993, p24.)

fluxoids

schematic photos
diagrams

Type II superconductor

H.K. Wong (CUHK 2011) PHY 3202 (3-6)

 The flux quantum (φ0) (X) Theories for superconductivity (Type I & Type II)
The flux quantum φ0 = h / 2e is not just for Type II SC. It Questions:
was measured by two groups using Type I SC: • Why TC so low?
1. Doll & Näbauer (1961) used a Pb tube (actually a Pb • Is superconductivity simply the state of a conductor with
coating on a quartz rod) (TC (Pb) = 7.19 K) . very high purity?
2. Deaver & Fairbank (1961) used a Sn tube (actually a Sn • Why are good metals not superconducting?
coating on a Cu wire) (TC (Sn)= 3.72 K). • Why is the transition so abrupt ?
Both samples had diameter ≈ 10 µm. • Do we need quantum mechanics (QM) developed in 20s-
Their experiments are quite similar. 30s to explain superconductivity?
1. At T > TC, a magnetic field Bf (called the freezing field) is Yes! Superconductivity is due to conduction electrons.
applied along the axis of the tube (in normal state). Bloch (1928) used QM to show that conduction electrons in
Therefore the magnetic flux in the tube ≠ 0. crystal move freely through the crystal like a wave (a
2. Then cooled the tubes in Bf to T = T0 < TC . quantum phenomenon).
3. Finally removed Bf at T0. As a result, a persistent current is • Why R = 0?
induced on the tube’s inner surface. (Why?)
The magnetic flux trapped in the tube is adjusted automatically (a) Cooper pair (1956)
to become quantized: φ = n φ0, where n is an integer. Cooper assumed a weak attractive force between two
(Note: φ < Bf A, where A is the cross-section area of the tube.) conduction electrons at 0 K, & then showed theoretically that
In other words, the tube (in superconducting state) becomes a
the two electrons:
tiny magnet. The next step is to measure its magnetic moment,  
from which φ is determined. Note: one with p ↑ (momentum p and spin up),

(1) The field is so small, even a Type II SC is used for this one with − p ↓ (momentum − p and spin down)
experiment, it is in Meissner state. can form “bound” state (just like one proton and one electron
(2) We can estimate the required Bf for these experiments: form a bound state, the hydrogen atom).
To get a flux = φ0 = 2 x10-15 T.m2,
the applied field is Bf = 2.5 x10-5 T. The weakly bound electron pairs (now called Cooper pairs) in
Recall: Earth field ≈ 5 x 10-5 T. (1 T = 104 Gauss) the superconductor are boson & can move cooperatively
without resistance.
(3) In mixed state of type II superconductor, the magnetic flux
of each fluxoid is φ0 in order to minimize the total energy. Note: Since the momentum of the pair is well defined, the
(Why? Hint: treat each fluxoid as a tiny magnet and think Cooper pair must spread out in space. (Uncertainty principle)
about holding a group of tiny magnets together.)
Doll & Näbauer Properties of Cooper pair:
(Phys. Rev. Lett. 7, 51 (1961)):
total charge = 2e (Note: φ0 = h / 2e )
used torsional pendulum
method to measure moment. total momentum = 0
and total magnetic moment = 0.
φ = 2φ0 Similar to two-fluid model, Cooper pairs appear at T < TC and
increase in number as T is lowered. Only those electrons near
Pb tube EF can form Cooper pairs.
light
φ = φ0 The Cooper pair has a binding energy per electron (∆(T),
called the pairing energy) ~ 10-4 – 10-3 eV.
The energy required to break up a Cooper pair (to become
two normal electrons) is 2∆(T).
Bm
The two electrons in a Cooper pair are separated by few
hundred Å. This size is much larger than the separation of
Bf two pairs. So all Cooper pairs overlap appreciably.
Bf Cooper pair is the basics of the BCS theory.
Deaver & Fairbank (Phys. Rev. Lett. 7, 43 (1961)): Recall: Energy diagram of electrons in normal metals

unoccupied states
no energy gap
allowed energy between filled and
Sn tube states for
energy

unoccupied states
conduction EF
electrons
At T = 0,
(densely spaced)
(conduction)
electrons fill
up states to EF
(Fermi energy)

Bf

H.K. Wong (CUHK 2011) PHY 3202 (3-7)

 The binding energy for superconductor introduces an energy Why the electron-lattice interaction supplies the attractive
gap 2∆(T) near the Fermi energy EF. force?
2∆(0) ~ 10-4 EF ~ eV “The mattress effect”: When a bowling ball falls onto a
normal state superconducting state mattress, it distorts the mattress (床褥). Another bowling ball
nearby may roll toward the first ball because of this distortion.
T>Tc 0<T<Tc T=0 It appears that the two bowling balls attract each other.
Now the analogy:
energy

-- --- - ------ 2∆(T) 2∆(0)

conduction electrons in superconductor ↔ bowling balls
EF
crystal lattice of superconductor ↔ mattress

normal electrons
States are filled up by Only those electrons
due to breakup
electrons. Some electrons near EF can form
(thermal excitation
just below EF are thermally Cooper pairs. They
of Cooper pairs
excited to slightly above EF. occupy the same state.

Low TC is related to small energy gap 2∆(T): The crystal lattice is

In superconducting state, the thermal energy of lattice polarized by an electron,
kBT < 2∆(T), resulting in an effective
otherwise most Cooper pairs are destroyed by thermal energy. positive charge to attract
another electron. These
two electrons form a
________ two normal electrons bound state: the Cooper
energy

pair.
________excited by thermal energy
Cooper pairs

Now 2∆(0) ~ 1 meV (see data shown later)

and for estimation, we take kBTC ~ 2∆(0)
(More exact calculation gives 3.5kBTC = 2∆(0))
⇒ TC ~ 12 K.  
Cooper pair: { p ↑, − p ↓}
What is the state of superconducting electrons in a Why R = 0?
superconductor? Recall:
Late 40s – early 50s, it was believed that electrons should
Conduction
have a new type of interaction in addition to the well-known band
repulsive Coulomb force.
-
Valence band
-

Fröhlich (1950) & Bardeen (1950) believed that the lattice

.. .. ..
. . .
vibration should play a major role. The idea was based on metal insulator semiconductor
some experimental results. e.g.: Cu NaCl Si

E.g., Isotope effect (1950) Why R = 0? The following argument is quite tricky.
We use the following energy diagram for both normal
For superconducting electrons and Cooper pairs:
element with more Sn isotopes
Thermally excited unoccupied states
than one isotope
normal electrons ------ 2∆(T)
TC ∝ m −1/ 2
log10 TC

normal electrons Cooper pairs

where m is the stay at the band edge
atomic mass of the
First, because all Cooper pairs (bosons) have the same
isotope.
binding energy, they are all gathering at the band edge.
If R ≠ 0 (with power dissipation), then when we apply an
Recall: For a spring-mass electric field, the Cooper pairs in the current can scatter with
system, log10 m
k
the lattice & change their states.
ω = k /m m
But there is an energy gap above the original state.
This confirmed the importance of lattice vibration and ⇒ Their states cannot be changed.
the electron-lattice interaction should be considered. ⇒ no scattering ⇒ R = 0.

H.K. Wong (CUHK 2011) PHY 3202 (3-8)

 (b) BCS theory of superconductivity The tunneling effect can be observed using thin film samples:
Many best minds in physics (including Einstein) tried to NIN, SIN or SIS junctions, where
understand superconductivity, but only 40 years after Onnes’s N = normal metal
discovery, the theory was established. S = superconductor I
BCS theory (1957) was derived using second quantization I = insulator.
technique (taught in QMII): Each sample consists of two
thin wires separated by an V
H BCS = ∑ ε k ( ck+ ck + c−+k c− k ) − ∑Vkk ' ck+' c−+k ' c− k ck ultrathin thin wire 1
k kk '
insulating layer. thin wire 2
They extended the Cooper model to all conduction electrons
near EF in the solid. The BCS theory explained successfully
almost everything that was known about low-temperature Each wire has a
superconductivity and provided the microscopic origin (i.e., width < 1 mm.
the mechanism).

Nobel Prize (1972) Consider NIN junction with layer thicknesses: x̂

layer 1: normal metal (N) <1 µm
N I
John Bardeen Leon Neil John Robert N

(1908-1991) Cooper Schrieffer layer 2: oxide (insulator) (I) < 3nm glass plate
(1930 - ) (1931- ) layer 3: normal metal (N) < 1 µm
N I N
The potential barrier for electron:

energy
The energy of electrons in metal is
limited (described by an energy
diagram by free electron model or the
conduction band in energy band x̂
diagram):
energy
EF -- --- - States are filled up by
electrons. Some electrons
just below EF are thermally
“for their development of a theory of superconductivity” excited to slightly above EF.
Nobel Prize (2003) The metal layers are connected to a voltage source.
x̂
Alexei A. Vitaly L. Anthony J. N I N
Abrikosov Ginzburg Leggett + N I N
energy
V I
The energy diagram of the sample -- --- - -- --- - EF

as a function of position x for V = 0 :

x a

x̂
Not all electrons can tunnel through the barrier,
e.g., electron a cannot tunnel because there is no available
state at the same energy in the left layer.
(We will ignore the lower parts of the energy diagrams
"for pioneering contributions to the theory of because no tunneling can occur in this region.)
superconductors and superfluids" Tunneling can occur for electrons above EF but I = 0. Why?
(A)
(XI)Tunneling
Tunnelingeffect::
effect When a voltage V is applied, the energy levels are shifted.
The energy gap 2∆(T) can be (Why?)
(Here e > 0, V > 0)
determined by tunneling experiments.
energy

Magician David Copperfield “walled

-----eV EF (tunneling
through” the Great Wall. current)
(http://www.youtube.com/watch?v=jtG6t6LcNAU) I
This is not allowed in classical mechanics. N I N
+ V≠ 0
Quantum mechanics: A particle can tunnel through the
barrier even if its energy is lower than the barrier height. x̂
It is due to Electrons in the right layer with energy
E F − eV ≤ E ≤ EF V
uncertainty barrier height (applied
principle. can tunnel through the insulating layer voltage)
initial energy into the left layer.
Result: Tunneling current increases linearly with increasing V.
x
(position) Tunneling is enhanced by electric field.

H.K. Wong (CUHK 2011) PHY 3202 (3-9)

 SIN junction: layer 1: superconductor (S) Energy gap 2∆(T)
layer 2: oxide (insulator) (I) can also be
layer 3: normal metal (N) measured by other
(A) (B)
methods.
(V = 0) (V = ∆ /e)
∆(T) depends on
I = 0. Why? temperature:
2∆ -- --- - EF BCS theory
S I N S I N I normal experimental
+ electron data
(C) tunneling
(V > ∆ /e)
(B) (C)

(?)

S I N (Townsend & Sutton 1962) T/TC

(A) ∆ /e V
I Nobel Prize (1973)
Ivar Giaever (1929 - )

(B) “for his experimental

Why is there a tail? discoveries regarding
tunneling phenomena in
∆ /e
superconductors”

Recall: At T > 0, some electrons can be found with E > EF

due to thermal excitation. The tunneling current in the tail is
due to these electrons.
The tail is weaker at lower temperature.
See curves 2 & 3 of data for Al-Al2O3-Pb junction below. You will enjoy reading his Nobel lecture:
www.nobel.se/physics/laureates/1973/giaever-lecture.html

SIS junction: layer 1: superconductor (S) (XII) The Josephson effect (1962)
layer 2: oxide (insulator) (I) Josephson further explored the theory for an SIS junction.
layer 3: superconductor (S) The two superconductors are the same, i.e., S1=S2 & ∆1=∆2.
(A) (B) (C) At that time, people already knew that all Cooper pairs are in
(V = 0) (V = ∆1 − ∆2 ) / e) (V = ∆1 + ∆ 2 ) / e) the same QM state, described by a single wavefunction (a

(∆1- ∆2) -- --- - complex number): ψ ( r ) = ρ eiθ
2∆1 2∆2
where ρ is the density of electrons and θ is the phase.
insulating layer
S I S S I S S I S
+ S1 I S2
I (V = 0)
(D) ψ1 ψ2
(V > (∆1+ ∆2) /e) (B)
(D)
E
due to thermal  iθ
excitation In superconductor S1: ψ 1 ( r ) = ρ1 e 1
 iθ 2
In superconductor S2: ψ 2 ( r ) = ρ2 e
(C)
V S I S
S I S ( ∆1 − ∆ 2 ) / e ( ∆1 + ∆ 2 ) / e
x̂
When a small U0 is applied, a dc
e.g. Al-Al2O3-Pb junction V
supercurrent I tunnels through the
is an SIS junction. S1 I S2
barrier with V = 0, and its direction
TC (Pb) = 7.2 K
R depends on the polarity of U0.
TC (Al) = 1.2 K I
U0 I = I max sin(θ 2 − θ1 )
I-V curves were measured at
different temperature T: I increases for higher U0 until I =
curve 1: T = 10 K Imax . For even higher U0, V ≠ 0
(NIN junction) and the V-I curve is
curve 2: T = 4.2 K
(SIN junction) This is the dc Josephson effect. Imax
curve 3: T = 1.64 K
(SIN junction)
curve 4: T= 1.05 K (For a simple derivation, see The
(SIS junction) Feynman lectures on physics, vol. 3,
(∆1- ∆2) /e (∆1+ ∆2) /e p.21-14, 1963.)
(Giaever & Megerle 1961)

H.K. Wong (CUHK 2011) PHY 3202 (3-10)

 Josephson also predicted the generation of a high frequency (b) New materials
alternating current at the junction: the frequency νJ = V/φ0. (i) Metallic hydrogen
This ac Josephson effect was confirmed by Giaever (1965).
H2 is the most common element in the universe.
The Josephson effect is a quantum behavior seen in
macroscopic object (the superconductor).
1898 liquefied hydrogen (20 K)
insulating
SQUID (superconducting quantum 1899 solidified hydrogen (14 K)
interference device) is composed of two 1935 Wigner & Huntington predicted the existence of
Josephson junctions. The two currents metallic hydrogen at 250 kbars pressure.
(I1, I2) passing through the two junctions (Scientific American, May 2000, pp.60-66.)
can have very interesting interference 1960s Ashcroft predicted that metallic hydrogen
behavior in magnetic field. Because of would be a room temperature superconductor.
this exotic property, SQUID is an I1↓ ↓I2 (Physics World 8 (7), p.43 (1995).)
extremely sensitive magnetic field
sensor.
(see applications)
Nobel Prize (1973) (ii) USO (unidentified superconducting objects)
Brian David Josephson 1973 TTF-TCNQ (an organic crystal):
(1940 - ) “superconducting fluctuations at 60 K”
1978 CuCl (a semiconductor): “diamagnetic anomaly at
high pressure at 150 K” (Rusakov & Chu)
1980 CdS: “flux exclusion at 77 K”
“for his theoretical 1980 TiB: “zero resistance at room temperature”
predictions of the properties 2000
of a supercurrent through a
tunnel barrier, in particular
those phenomena which are
generally known as the
Josephson effects”

(XIII) Limit of TC and new surprises (iii) HTS

(High temperature
• BCS theory was further perfected so that given the energy superconductors)
spectrum of lattice vibration and the electronic properties of
a material, one can predict the value of its TC.
• An example is McMillan equation (1968) from which the
maximum TC was estimated to be about 30-40 K. Before
1986, the record TC was found to be 23 K in Nb3Ge.
• By 80s, people believed that superconductivity is a “mature” J.G. Bednorz & K.A.
physics! Müller, “Possible high J=
• Is high TC superconductivity possible? TC superconductivity in
Some people are optimistic. the Ba-La-Cu-O
Theorists looked for new superconducting mechanisms. system”, Z. Phys. B64, TC
Experimentalists searched for new materials. p.289 (1986).
No data on Meissner effect.

(a) New mechanisms A few groups repeated their findings and achieved higher TC
at high pressure. This suggested new superconductors could
Little (1964) considered a hypothetical 1D
be found by chemical substitution. Chu then became famous:
organic molecule with side branches.
A conduction electron polarizes the side- 2
branches and induces positive charges at the
ends near the spine. A second electron is 7
attracted to this region of positive charge and is 3 1
therefore indirectly attracted to the first 4
electron. The polarization is due to electron-
electron interaction and thus 5
TC ~ me-1/2 ~ room temperature. 1

mass of electron << mass of ion 1: exceptionally rapid publishing. 6

(Ref.: Scientific American, Feb 1965, p.21) 2: Y or Yb?
3: Who (Wu or Chu) discovered YBCO?
(C.W. Chu, IEEE Tran. Appl. Sup. Vol. 7, p.80 (1997).

H.K. Wong (CUHK 2011) PHY 3202 (3-11)

 The 90 K (iv) New Surprises
superconductor was
(a)More superconducting elements at high pressures:
identified as YBa2Cu3O7
S (1997), O (1998), B (2001), Fe (2001), Li (2002)
with a perovskite crystal
(b)Ferromagnetic superconductors: Superconductivity is
structure
usually destroyed by high magnetic field but materials
were discovered with coexistence of superconducting and
ferromagnetic properties.
2001: Fe at high pressure (Nature 406, p.316)
2001: ZrZn2 alloy (Nature 412, p.58)
2001: UGe2, URhGe (Nature 413, p.613)
2000: κ-(BEDT-TTF)2Cu[N(CN)2]Br,
magnetic-field-induced superconductivity in
organic material. (Phys. Rev. Lett. 85, p.5420)
(c) Many semiconductors, oxides, hydrides, organic materials
were found to be superconducting.

Many more HTS

were found.

Record TC
= 164 K
(HgBa2Ca2Cu3O8
at 30 kbars
hydrostatic
pressure.)
(d) New non-copper high temperature superconductors:
2001: MgB2 (Jun Akimitsu)
• TC = 39 K
• MgB2 is available commercially.
(Physics World, Jan.2002, p.29)

2008: Fe-based superconductors

SmFeAsO1-xFx
• TC = 43 K

Nobel Prize (1986) (XIV) Applications (This part will not be covered in final exam.)
J. Georg Bednorz K. Alexander Müller (1) High current applications (based on R = 0):
(1950 - ) (1927 - ) electricity transmission, energy storage, …
(2) High field applications (based on superconducting
magnet): MRI, Maglev, motor, generator, accelerators,
research equipment, …
(3) Josephson applications (based on Josephson effects):
SQUID, supercomputers, …

“for their important breakthrough in the discovery of

superconductivity in ceramic materials” (Scientific American, Feb. 1989, p.45.)

H.K. Wong (CUHK 2011) PHY 3202 (3-12)

 (a) Superconducting magnet (SM) American Superconductor’s RABiTS 2G HTS tape
(ribbon-shaped YBCO tape)
Onnes’ dream: to build a 10 T SM. (2nd generation)
SM is a solenoid of SC wire. µ0nI
Features of SM:
1. generating a stable and
uniform magnetic field
2. light weight
What SC is required? B = µ0nI
10 m spool y
1. Type II
10 T 10 turns/cm 8 kA allo
%
2. with strong pinning force 5 at
i -W
dN
3. with high TC, high IC, high HC2 ure
text
4. operated at T << TC (why?)
5. can be shaped into wires or tapes
6. with good mechanical properties

Note: Other high field magnets:

Commercial SM: e.g. 15 T, 5 cm bore
Bitter magnets (for B up to 33 T)
Hybrid magnets (Bitter + superconducting)
(for B up to 45 T)
Pulsed magnets
(for B up to 70 T)
For an introduction, visit http://www.magnet.fsu.edu/
magnettechnology/research/magnetprojects/index.html

1968: RCA developed a 14 T magnet

using their superconducting Nb3Sn
tapes.
(See also “Superconducting
magnets above 20 Tesla”, Physics SM vessel
Today, Aug. 2002, p.37) for liquid
helium
HTS cable or tape: Applications of SM:
The most successful process for producing long-length tapes (i) Particle accelerator for high energy experiments:
for high current and high field applications at 77 K is called e.g. Tevatron
(proton-antiproton
RABiTS
collider) in Fermi
(the Rolling-Assisted
Lab near Chicago
Biaxially Textured
in USA
Substrate process)

Circumference
of the circular
ZrO2 accelerator is
~ 6.4 km.
(http://www.fnal.gov/)
Status of HTS cable/wire/tapes:
1995: The tapes were only 4 cm long .
JC ~ 106 A/cm2 (77 K)
(Science 268, 644 (1995))
2002: The tapes could not be made longer than 1 m.
2006: I < 140 A at 77 K for wire 100 m long and 4 mm wide
(http://www.amsuper.com)
2009: National High Magnetic Field Laboratory at Florida
State University was awarded nearly US$ 3 million to _
build a high-temperature superconducting magnet (32 T p p
with ~ 8 km of HTS cable).
(1 TeV)
Need about 1000 SMs to keep the
high energy particles moving in the
circular accelerator.

H.K. Wong (CUHK 2011) PHY 3202 (3-13)

 (b) SQUID for brain research
SQUID is a sensitive magnetic field sensor.

(magnetic field signal

due to brain activity)

SMs for Main Injector (Scientific American, Aug. 1994, p.36.)

磁浮火車)
磁浮火車 in Japan
(ii) Maglev (磁浮火車 SMs on board

Speed record:
518 km/h

Coils for guidance

Coils for
levitation using
Faraday’s law.

(iv) Magnetic Resonance Imaging (MRI) (1977) Note:

Earth’s magnetic field
Magnetic field level
Urban noise

magnetic field from the heart

magnetic field from brain activity

Require a strong magnetic field (0.5 – 2 Tesla), usually

generated by SM. (Physics World Dec. 2002, pp.31-35.) frequency (Hz)
Stronger field ⇒ higher resolution of images
Here is a high field MRI (4 Tesla) (for research only):

H.K. Wong (CUHK 2011) PHY 3202 (3-14)