50 years of BCS School Basics of pairing Ilya Vekhter Louisiana State University, USA July 2007 Cargese
50 years of BCS School
Basics of pairing
Ilya Vekhter
Louisiana State University, USA
July 2007 Cargese
July 2007 Cargese
Timescale
H. Kamerlingh-Onnes1911, Nobel Prize 1913
G. Holst“carefully carried out
all the measurements”
J. Bardeen, L. Cooper and R. Schrieffer1957, Nobel Prize 1972
Why so long?
July 2007 Cargese
Solids: 1900 -- 1925• 1897 J.J. Thomson: discovery of an electron• 1900 P. Drude: transport theory
– classical electrons scatter on atoms. Wiedemann-Franz ratio.
mne τσ
2
=
2
2v
3v
nemc
T=
σκ
3v v
2 τκ c=
July 2007 Cargese
Solids: 1900 -- 1925• 1897 J.J. Thomson: discovery of an electron• 1900 P. Drude: transport theory
– classical electrons scatter on atoms. Wiedemann-Franz ratio.
mne τσ
2
=
Tkm B3v2 =
2/3v Bnkc =
2
2v
3v
nemc
T=
σκ Classical gas
3v v
2 τκ c=
July 2007 Cargese
Solids: 1900 -- 1925• 1897 J.J. Thomson: discovery of an electron• 1900 P. Drude: transport theory
classical electrons scatter on atoms. Wiedemann-Franz.
mne τσ
2
=28
2
2
2v /1011.1
23
3v KW
ek
nemc
TB Ω×≈⎟⎠⎞
⎜⎝⎛== −
σκ
3v v
2 τκ c=
•1904 H. Lorentz: Maxwell-Boltzmann to treat electron-atom collisions. Wrong temperature dependence, wrong WF•No understanding of metal physics•1911: superconductivity… what is it about?
July 2007 Cargese
Missing specific heat1912: P. Debye: specific heat of lattice vibrations (phonons),
http://hyperphysics.phy-astr.gsu.edu/hbase/solids/phonon.html
No trace of classical electronic specific heat
July 2007 Cargese
Metals: 1925 -- 1930
• W. Pauli exclusion principle, beginning of solid state theory
W. Pauli
Then he said, "Who are you?" "I am Weisskopf, you asked me to be your assistant." "Yes," he said, "first I wanted to take Bethe, but he works on solid state theory,which I don't like, although I started it."
V. Weisskopf
July 2007 Cargese
Metals: 1925 -- 1930• W. Pauli: exclusion principle• Fermi-Dirac statistics
filled states
empty states• Fermi energy EF ~ 10000K
EF
July 2007 Cargese
Metals: 1925 -- 1930• W. Pauli: exclusion principle• Fermi-Dirac statistics
filled states
empty states• Fermi energy EF ~ 10000K
• Absorbed energy ∆E~n(kBT)2/EF~(kBT)2N0
• Specific heat CV ~ nkB (kBT/EF)<< nkB
kBT
July 2007 Cargese
Metals: 1925 -- 1930• W. Pauli: exclusion principle• Fermi-Dirac statistics:• Sommerfeld theory of conductivity
CV ~kB2N0kT<< nkB
Tkm B3v2 =
2/3v Bnkc =
2
2v
3v
nemc
T=
σκ
July 2007 Cargese
Metals: 1925 -- 1930• W. Pauli: exclusion principle• Fermi-Dirac statistics:• Sommerfeld theory of conductivity
CV << nkB
Tkm B3v2 =
2/3v Bnkc =
2
2v
3v
nemc
T=
σκ
)/(v FBB ETknkc ≈
)/2(v2F TkETkm BFB=
July 2007 Cargese
Metals: 1925 -- 1930• W. Pauli: exclusion principle• Fermi-Dirac statistics:• Sommerfeld theory of conductivity
CV << nkB
2822
/1044.23
KWe
kT
B Ω×≈⎟⎠⎞
⎜⎝⎛= −π
σκ
measurable at low T
July 2007 Cargese
Metals: 1925 -- 1930• W. Pauli: exclusion principle• Fermi-Dirac statistics:• Sommerfeld theory of conductivity
CV << nkB
2822
/1044.23
KWe
kT
B Ω×≈⎟⎠⎞
⎜⎝⎛= −π
σκ
• Problem: Mean free path too long
• Bloch’s theorem for a crystal
al F >>= τv
)()( rr kkr
k uei=ψ
• Infinite mean free path in a periodic lattice
measurable at low T
MM
Me a
July 2007 Cargese
Metals: summary
2
22
maae h
≈=ε2
2
meaB
h=MM
Baa 54 −≈
Mm
D εω ≈h2
222
221
maaM D
h≈⎟
⎠⎞
⎜⎝⎛ωDebye frequency:
cmaF
210v −≈≈h
aNaN
LnkF
122 max =≈≈ππ
Electron momenta:
300≈=mMβK
maEF
42
2
10≈≈h
Energy scales KD 300≈ωh
DFc EKT ω,10 <<≤ Very small scale
July 2007 Cargese
Second Bloch’s theorem
Flaw: thinking of a superconductor as
a state of a metal with finite current
July 2007 Cargese
Superconductivity: 1930-1937
J. C. McLennan 1934
Sn
• 1932 Keesom, Kok: specific heat anomaly in tin
July 2007 Cargese
Superconductivity: 1930-1937• 1932 Keesom, Kok: specific heat anomaly in tin
Superconductivity is a stable thermodynamic
phase different from the usual metallic state
J. C. McLennan 1934
Sn He
superfluid
July 2007 Cargese
Superconductivity: 1930-1937
• 1933 Meissner-Ochsenfeld: Meissner effect
Superconductors are not just perfect conductors
July 2007 Cargese
Superconductivity: 1930-1937• Gorter: B=0 inside a superconductor • London equations: zero resistivity and Meissner effect
Ev edtdm =
tc
∂∂
−=×∇ − HE 1vj ens=
Hj 12
−−=×∇ cen
m
sEj =⎟⎟
⎠
⎞⎜⎜⎝
⎛∂∂
2enm
t s
• Penetration depth
jHcπ4
=×∇
022 =−∇ − HH Lλ H0
)/exp(0 LxHB λ−=
2
22
4 enmc
sL πλ =
July 2007 Cargese
Rigidity of wave function and gap
Ej =⎟⎟⎠
⎞⎜⎜⎝
⎛∂∂
2enm
t s tc ∂∂−=
×∇=− /1 AE
AHHj 12
−−=×∇ cen
m
s
00=⋅∇=⋅∇
Aj
Ajmc
ens2
−=gauge invariance?but
F.London 1937
July 2007 Cargese
Rigidity of wave function and gap
Ej =⎟⎟⎠
⎞⎜⎜⎝
⎛∂∂
2enm
t s tc ∂∂−=
×∇=− /1 AE
AHHj 1
2−−=×∇ c
enm
s
00=⋅∇=⋅∇
Aj
Ajmc
ens2
−=gauge invariance?but
•Rigidity: wave function does not respond to A (phase coherence over large distances)
[ ]∗Ψ∇Ψ−Ψ∇Ψ−∝ *
2me
phj Aj 2
2
Ψ−∝mce
d
•Energy gap between continuum and superconducting state
July 2007 Cargese
Superconductivity: 1939-1949
• Pippard: there is another scale besides λL
• Non-local electrodynamics: current averaged over ξ0
• Ginzburg-Landau: effective theory for order parameter
• Discussed in other lectures
July 2007 Cargese
Isotope effect: 1950
Reynolds, Serin, Nesbitt 1950Maxwell 1950
Ionic motion is important for superconductivity
July 2007 Cargese
Phonon attraction: formalismAmplitudes of processes:
1p 2p
kp h−1 kp h+2
kh
kh−
)()()(||
11
2
kkpEpEV
ωhh −−−+
)()()(||
22
2
kkpEpEV
ωhh −+−
2211
2
)]([)]()([)(||
kkpEpEkV
ωω
hh
h
−−−)()()()( 2121 kpEkpEpEpE hh ++−=+)()()()( 2211 pEkpEkpEpE −+=−− hh
Phonon-mediated attractive e-e interaction
for electrons within ωD of the Fermi surfaceH. Fröhlich 1952
July 2007 Cargese
Phonon attraction: simple pictureFa v/≈τTime electron spends near an atoma
MMe ττ )/( 22 aeFp ≈≈Typical momentum transfer
MpxM D /222 ≈ωTypical atomic displacement
βωωa
aeMmM
ae
maa
Maea
Mpx
DFD≈≈≈≈
)/(/
)/(1
v 22
2
2
2 h
h
July 2007 Cargese
Phonon attraction: simple pictureFa v/≈τTime electron spends near an atoma
MM e
aττ )/( 22 aeFp ≈≈Typical momentum transfer
β/ax ≈ mM /=βTypical atomic displacement
Dae
ax
aeU ω
βh−≈−≈−≈∆
22Change in potential
July 2007 Cargese
Phonon attraction: simple picture
Fa v/≈τTime electron spends near an atom
MM e
MM
ττ )/( 22 aeFp ≈≈Typical momentum transfer
β/ax ≈ mM /=βTypical atomic displacement
DU ωh−≈∆Change in potentialM M1−≈ DM ωτTime before atoms relax back
M Mavl MF βτ ≈≈Displacement trail length
July 2007 Cargese
Phonon attraction: simple picture
Fa v/≈τTime electron spends near an atom
ττ )/( 22 aeFp ≈≈Typical momentum transfer
β/ax ≈ mM /=βTypical atomic displacement
DU ωh−≈∆Change in potential
l βωτ )v/(1FDM a≈≈ −
Time before atoms relax back
avl MF βτ ≈≈Displacement trail length
“Tube” of attractive potential: most advantageous for electrons moving in opposite directions
V. Weisskopf 1979
July 2007 Cargese
Cooper phenomenon
Fpair EE 2<Bound states of two electrons near the Fermi surface:
( ) ( ) ),(2),(),(2 222
22
21
2
rrrrrr 111 Ψ+=Ψ⎥⎦
⎤⎢⎣
⎡+∇+∇− FEEV
mh
[ ])(exp)(),( 22 rrkkrr 1k
1 −⋅=Ψ ∑ iψOur “interaction tube”: CM at rest
( ) )(2)()(22
kkkk
kk ψψψ FEEVmk
+=′+∑ ′h
[ ] rkkrrkk diVLV ∫ ′−⋅= −′ )(exp)(3What is the matrix element?
July 2007 Cargese
Matrix element
al β≈
Both electrons near the Fermi surface: rapidly oscillating phases
[ ]drirLUVl
∫ ′−⋅−≈ −′
0
30 ||exp kkkk
DmaakkEE ω
ββh
hhh ≈≈≤′−≈′− 2
2F
Fv||v||
July 2007 Cargese
Bound State( ) )(2)()()(2 kkkk
kkk ψψψ FEEVE +=′+∑ ′
e
e
Fkk <= for0)(kψKey: filled Fermi sea
))((2
)()(
FEEE
V
−−
′=
∑ ′
k
kk k
kk ψψSolution:
1
00 )2(
))((211 −−≈
−−= ∫∑ EdVN
EEEV
D
Fξξ
ωh
k kSelf-consistency:
Bound state of two electrons with opposite momenta: Cooper pair ]/2exp[2 00 VNE D −−= ωh
L. N. Cooper 1956
00 ∆≈ENB:
July 2007 Cargese
Importance of the Fermi surface• Same problem without FS:
)(2)(2
)()(
kk
kk k
kk
EEC
EE
V
−=
−
′=∑ ′ψ
ψ
• Self-consistency on C is equation for energy E
∑+
=k k ||/
1122 EmV h
Is there a bound state for a small attractive potential?
For small E integral finite: need critical potential value∫ +
=||/
4122
2
Emkdkk
V h
π3D
)/1exp( 0VNWE −−=Fermi surface makes it “2D”
||ln
||/21
022 EWN
Emkkdk
V≈
+= ∫
h
π2D
July 2007 Cargese
Comments
• Non-perturbative result: no expansion V~0• Cooper pairs: bosonic objects
• Bose condensation of pairs leads to superconductivity.M. Schafroth, S. Butler, J. Blatt 1957
4He is also a collection of bound fermions (6, not 2).
What is special about gas of Cooper pairs?
July 2007 Cargese
Cooper pair gas
3−≈ an
pF∆≈∆ v0
)/(/v 000 ∆≈∆≈ FF EahξFF Ep
p 0∆≈∆
Spread of momentum
Cooper pair size
Electron density
Cooper pair density
03/1
03/1 )/( ξ<<∆≈≈ −
Fc EandDistance between pairs
10
20
3 )/( −−− ≈∆≈ ξaEan Fc
Strongly overlapping gas of Cooper pairs
July 2007 Cargese
Dilute Bose gas
13 <<ξnSingle-particle states Ensemble states
k≠0 excited
k=0 ground state
ξ condensatecondensate
Gapless spectrum: when one particle is excited (starts moving) it has little chance of scattering off the particles in the condensate
July 2007 Cargese
Dense neutral Bose gas
1~3ξnSingle-particle states Ensemble states
k≠0 excited
k=0 gap ∆
ground statecondensate
Low energy excitation: rotation of atoms relative to one another: typical energy
22 / ξMh≈∆ Rotons
July 2007 Cargese
Cooper pair gas
Ensemble states Overlapping charged pairs: roton energy high
excited
Energy gap=energy to dissociate a pair
gap ∆
ground state
How to describe such a state? What is the “right” combination?
July 2007 Cargese
BCS wave function
kkkk
kkkk
k bbVnH +′
′′∑∑ += σ
σεReduced Hamiltonian:
electron pairing
( )( ) 0
02
2
=
=+
k
k
b
b+↓−
+↑
+
+
=
=
kkk
kkk
ccb
ccn σσσ
↓−↑+
′′′′+
−−=
=
kkkk
kkkk
nnbb
cc
1],[
, σσσσ δδNot the usual bosonicoperators!
July 2007 Cargese
BCS wave function
kkkk
kkkk
kk bbVbbH +′
′′
+ ∑∑ +=σ
ε2Reduced Hamiltonian: electron pairing
( )( ) 0
02
2
=
=+
k
k
b
b+↓−
+↑
+
+
=
=
kkk
kkk
ccb
ccn σσσ
↓−↑+
′′′′+
−−=
=
kkkk
kkkk
nnbb
cc
1],[
, σσσσ δδNot the usual bosonicoperators!
Variational approach: unoccupied / doubly occupied states, no single occupancy
Inspiration: polaron
July 2007 Cargese
BCS wave function
kkkk
kkkk
kk bbVbbH +′
′′
+ ∑∑ +=σ
ε2Reduced Hamiltonian: electron pairing
( )( ) 0
02
2
=
=+
k
k
b
b+↓−
+↑
+
+
=
=
kkk
kkk
ccb
ccn σσσ
↓−↑+
′′′′+
−−=
=
kkkk
kkkk
nnbb
cc
1],[
, σσσσ δδNot the usual bosonicoperators!
Variational approach: unoccupied / doubly occupied states, no single occupancy
Inspiration: polaron Static distortion of the lattice around an electron: emission of multiple phonons into the same state
[ ] vacaagk
kkk∏ −+ +∝Ψ )(exp
e
T.D. Lee, F. Low, D. Pines 1953
July 2007 Cargese
BCS wave function IIvacbgvace kk
kk
bg kk )1( +∏∏ +≡∝Ψ+
vacccvu kkkk
kBCS )( +↓−
+↑∏ +=ΨNormalization:
122 =+ kk vuProbability empty
Probability occupied
July 2007 Cargese
Ground state energykkkk
kkkk
kkkBCSBCSg vuvuVvNHE ′′
′′∑∑ +=Ψ−Ψ=
σεµ 22
Choice of coefficients: variational method
P. Hirschfeld talk
kk
kk
vu
θθ
cossin
== 22 ∆+= kkE ε
⎟⎟⎠
⎞⎜⎜⎝
⎛−=
⎟⎟⎠
⎞⎜⎜⎝
⎛+=
k
kk
k
kk
Ev
Eu
ε
ε
121
121
2
2 excitation energy
kkk
kkk vuV ′′′
′∑−=∆
( ) 00/1exp2 EVND ≈−≈∆ ωh
energy gap
0=∂∂θ
gE
2/)0()( 20∆−==∆−∆ NEE ggCondensation energy
July 2007 Cargese
Occupation numbers
⎟⎟⎠
⎞⎜⎜⎝
⎛−=
k
kk E
v ε1212Probability that the states with
momenta k and –k are occupied
2kv
FE
∆~ At T=0 looks like thermally smeared Fermi distribution
No discontinuity in occupation numbers: no gap in momentum space
Gap in excitations that are not coherent with the pairs: different phase
kε
July 2007 Cargese
Particle number
vacccvu kkkk
kBCS )( +↓−
+↑∏ +=Ψ particle number is not
conserved!
∑∑ =ΨΨ= ↑+↑
kk
kBCSkkBCS vccN 222
NNN 1
∝δ
( ) ∑=−=k
kk vuNNN 2222 4δ
Small relative fluctuations of the particle number
July 2007 Cargese
Particle number II
∑ Ψ=Ψ NNBCS λProject on a state with a definite # of electrons
State with N electrons has phase 2/ϕiN vacccevu kk
ik
kk )( +
↓−+↑∏ +=Ψ ϕ
ϕ
ϕϕ
πϕϕ
π
ϕϕ Ψ=+=Ψ −+↓−
+↑
− ∫∏∫ 2/2
0
2/2
0)( iN
kki
kk
kiN
N edvacccevued
All phases needed for definite particle number
NiiN∂∂⇔∂∂−⇔
//
ϕϕNumber and phase are
conjugate variables1≈∆∆ ϕN
July 2007 Cargese
Summary I
• Reduction of the problem to essential physics, and then trial for new ground states/excitations
• Solve problems even seemingly difficult ones: electrons form pairs despite Coulomb repulsion
July 2007 Cargese
Summary I
• Reduction of the problem to essential physics, and then trial for new ground states/excitations
• Solve problems even seemingly difficult ones: electrons form pairs despite Coulomb repulsion
• Works for other complicated problems: Laughlin wave function for fractional quantum Hall.
• Does the reduction method always work?
iyxzezzzz z
ijjin +=−=Ψ −∏ ,)(),...,(
2||31
July 2007 Cargese
Extension of BCS: AFM, RVB and all that
• Doped Hubbard systems (High-Tc, cobaltates)
ji
ijij
i JctcH SS∑∑ += +σσ
tU >>double occupancy expensive
C. Gros 2007
• Superexchange: AFM interaction
• If not 1 electron per site: vacancies that move
↓↑+ ∑∑ += i
iij
iji nUnctcH σσ
July 2007 Cargese
Resonating valence bonds
21
=• Form singlets on neighboring sites
• Look at 1D chain, compare energy
valence bond wins in 1D
energies may be close in 2D, with frustration…4
3J−
43J
−4J
−4J
−4J
−
• Make a linear superposition of bonds: allow them to resonate to lower the kinetic energy: mobile singletsAnderson, Anderson and Fazekas, 1973
July 2007 Cargese
RVB trial wave function
• Two electrons on a mobilevalence bond of length r
• Vary bond length/ no double occupancy
• Bonds condense/resonate
• Project out doubly occupied (Gutzwiller)
ikrk
kkri
iir eccccb +
↓−+↑
+↓+
+↑
+ ∑∑ ==
0)( =∑k
ka+↓−
+↑
+ ∑= kk
k cckab )(
( ) vacbN 2/+ double occupancy
( )( ) vacbnnN
iiiRVB
2/1 +
↓↑∏ −=Ψ
Anderson, 1987
July 2007 Cargese
BCS and RVB+↓−
+↑
+ = kkk cckab )(• Consider an unusual pairing
vacbgvace kkk
bg kk )exp(∑=∝Ψ +∏+
• Construct a BCS wave function
• Project a fixed number vacbvN
kkkNBCS
2/
, ⎥⎦⎤
⎢⎣⎡=Ψ ∑ +
• Prohibit double occupancy ( ) NBCSi
iiRVB nn ,1 Ψ−=Ψ ∏ ↓↑
Resonating valence bonds and Cooper pairs look similar: basis for theories of exotic SC
Anderson, 1987-2007,Gros et al. 2007,Paramekanti et al. 2004…
July 2007 Cargese
Summary and challenge
• (Correctly) guessing a wave function remains the best way to solve a strongly correlated problem
• Challenge: wave function describing an electronic system with most exotic properties.
• Bonus points: Hamiltonian for which this wave function is a good variational ground state
• Prize: TBD