Post on 10-Mar-2020
The Emergent Fermi LiquidFermions, Fermions, Fermions...
Koenraad Schalm
Institute Lorentz for Theoretical PhysicsLeiden University
Mihailo Cubrovic, Jan Zaanen, Koenraad Schalmarxiv/0904.1993
Lee: arxiv/0809.3402Liu, McGreevy, Vegh: arxiv/0903.2477Faulkner, Liu, McGreevy, Vegh: arxiv/0907.xxxxCubrovic, Sadri, Schalm, Zaanen: arxiv/090x.xxx
AdS/CMT July 2009 KITP UCSB
Fermions at finite density
• Electrons in the real world/in the “mundane” [Hartnoll (Monday)]
Experiment Theory
Fermi Liquid adiabatic continuation Fermi-gasLandau quasi-particles
BCS superconductor BCS PairingSpontaneous symmetry breaking
Non Fermi Liquids- high Tc cuprates- heavy fermions- ...
?
Fermions at finite density
• Electrons in the real world/in the “mundane” [Hartnoll (Monday)]
Experiment Theory
Fermi Liquid adiabatic continuation Fermi-gasLandau quasi-particles
BCS superconductor BCS PairingSpontaneous symmetry breaking
Non Fermi Liquids- high Tc cuprates- heavy fermions- ...
?
Non-Fermi Liquids and the fermion-sign-problem
• The non-Fermi liquid: two important characteristics– Physics controlled by a QCP [conjecture]– Adjacent phase is regular Fermi Liquid
• Fundamental problem:
Fermion-sign-problem [original]
[Computer]
[not weakly interacting electrons:interacting CFT?strongly coupled?]
Non-Fermi Liquids and the fermion-sign-problem
• The non-Fermi liquid: two important characteristics– Physics controlled by a QCP [conjecture]– Adjacent phase is regular Fermi Liquid
• Fundamental problem:
Fermion-sign-problem [here]
How can a Fermi Liquid (defined by EF , kF )disappear into a quantum critical state?
[not weakly interacting electrons:interacting CFT?strongly coupled?]
Non-Fermi Liquids and the fermion-sign-problem
• The non-Fermi liquid: two important characteristics– Physics controlled by a QCP [conjecture]– Adjacent phase is regular Fermi Liquid
• Fundamental problem:
Fermion-sign-problem [here]
How can a Fermi Liquid (defined by EF , kF )emerge from a quantum critical state?
[not weakly interacting electrons:interacting CFT?strongly coupled?]
⇒difficult for standard QFT......”natural fit” for AdS dual
The fermion-sign-problem II
High Tc superconductors Heavy fermion systems
@@@R
@@@R
The fermion-sign-problem II
High Tc superconductors Heavy fermion systems
@@@R
@@@R
AdS-to-ARPES
• Essential: Single fermionic quasi-particle spectrum encodesFermi Liquid ground state
A(ω, k) = − 1π
ImGR(ω, k)
• Experimental probe: ARPES [Source: A. Damascelli CIAR 2003]
[Source: A. Damascelli CIAR 2003]
AdS-to-ARPES
• Essential: Single fermionic quasi-particle spectrum encodesFermi Liquid ground state
A(ω, k) = − 1π
ImGR(ω, k)
• Experimental probe: ARPES [Source: A. Damascelli CIAR 2003]
!!"#$#%&'(('#)*')$%)('$+"!!!"#$!%&!'()*"#$!%)+,"!%(-"#$!"!
,')-.%/012$'3.%*(0%3$21.*45)3$'#)
#!"#!$%6*$"%*7(%.4/%)%2-+8*/*310$52%(*$"%*%44%3$(*#4*')$%213$'#)(
9)-.%/:%(#.;%<*!"#$#%&'(('#)*,0%3$2#(3#0+
[Source: A. Damascelli CIAR 2003]
AdS-to-ARPES
• Essential: Single fermionic quasi-particle spectrum encodesFermi Liquid ground state
A(ω, k) = − 1π
ImGR(ω, k)
• Experimental probe: ARPES [Source: A. Damascelli CIAR 2003]
!"
#
$
!
!
!
!"#"$%&'()*$+,-$!"*$./01$2.3304$
!"#$%&'(#)*+"&,)&'#-.(/
0
!
$ !
"
#$% #$#
"
$
&'()'*+(,-./$0$1
2+'3+'4-5')64,-.)71
!"#$%&'&()*(+&,--.
!
$ !
"
5"$!67689:;;&$:<$6;"*$+,=$#!*$0.3>$2?@@@4
A"B"$%):'$:<$6;"*$+,-$$%*$.1@0@?,$2?@@.4
!"#$%&'&/*01023&4+5
#$8 #$#
!
"
$
&'()'*+(,-./$0$1
-2+'3+'4-5')64,-.)71
$
!"#$%&'(#)*+"&,-.-/-01&-)&'#23(45
!
!
!
!
$ !
"!
[Source: A. Damascelli CIAR 2003]
The AdS set-up
• Spectral density from charged AdS BH [e.g. McGreevy’s lectures]
– GΨΨR (ω, k) : infalling b.c. at BH Horizon ⇔ T > 0
– GΨΨR (ω, k): finite µF? ⇔ µU(1)
⇒ Charged AdS BH in 3+1 dim dual to 2+1 dim relativistic CFT:
ds2 =α2
z2
(−f(z)dt2 + dx2
1 + dx22
)+
1f(z)
dz
z2
A0 = 2qα(z − 1)f(z) = (1− z)(z2 + z + 1− q2z3)
4πT = α(3− q2) , µ0 = −2qα
Is µU(1) = µF?
• AdS Phenomenology:
– Do not know exact quantum theory (CFT or AdS)i.e. do not know constituents of the BH. There could be manyU(1) charged particles (fermions and bosons)
It is not a priori guaranteed that µU(1) will act as µF
• Empirical approach:
– QFT: In an interacting system µF renormalizes:µ
(IR)F is empirically determined by the pole in GR
– AdS: We will follow the same approach.Expectation µ(UV )
U(1) ≡ µ0 induces a µF ≡ EF , whose value we
read off from the spectrum. A priori µF 6= µ(UV )U(1)
Caveat:dynamicalinstabilityin full theory
Fermions in AdS/CFT I
• Action: Einstein YM + charged fermions
S =∫ √−g [R+ 6− 1
4F 2 − Ψ̄eMA ΓA (DM + igAM ) Ψ−mΨ̄Ψ
]+ Sbnd
• Fermions:[1. ] constrained (first order) system:⇒ reduce to physical d.o.f.
ΓzΨ± = ±Ψ±
[2. ] action vanishes on shell.⇒ add boundary term
Sbnd =∫ √−hΨ̄+Ψ−
⇒ variation∂S
∂Ψ−well-defined
[L = κ = 1; also set g = 1 from hereon]
[Henningson,SfetsosMueck,ViswanathanHenneauxContino,Pomarol]
Fermions in AdS/CFT II
• Standard AdS/CFT: .... with some subtleties– to solve Dirac Equation...
(∂z +Az)Ψ± = ∓/T Ψ±
– ... write as second order equation
(∂2z + P∂z +Q)Ψ+ = 0
– ... compute on-shell action(bdy term only; bdy Ψ0 determined via EOM)
Ψ± = F±(z)F−1+ (z0)Ψ0
+ ⇔ Ψ0− = F−(z)F−1
+ (z0)Ψ0+
– Green’s function
G(z0) = F−(z0)F−1+ (z0)
Fermions in AdS/CFT III
• Unitarity bound Fermion mass [Contino,Pomarol;CZS;Iqbal,Liu]– 2nd order equation (∂2
z + P∂z +Q)Ψ+ = 0 has asymptoticsolutions
Ψ+ = zd+1
2 −|m+ 12 |(A+ . . .) + z
d+12 +|m+ 1
2 |(B + . . .)
– CFT Green’s function (T = 0) has a pole unless
−12< m (Sbnd breaks m⇔ −m)
⇒ Unitarity bound on CFT fermionic operator dimension
〈OΨ(−k)OΨ(k)〉 = k2∆Ψ−d ∆Ψ =d− 1
2+ |m+
12|
d− 12
< ∆
Which ∆/m?
• IR emergence of a new state– Recall from holographic superconductors
[Gubser;Hartnoll,Herzog,Horowitz]
relevant perturbations = scalars with −d2
4 < m2φ < 0
⇒ For fermions: choose mass near unitarity bound: − 12 < mΨ
– Upper bound on mass?Green’s function marginal at: mΨ = 0, ∆Ψ = d
2⇒ Interesting range to obtain new physics
−12< mΨ < 0 − d− 1
2< ∆ <
d
2
[mΨ = 0: Lee:0809.3402; Liu, McGreevy, Vegh:0903.2477][Qualitative argument (g = κ = 1): quantitative cf. Denef,Hartnoll]
Unitarity bound
��
�
The emergent Fermi Liquid from AdS/CFT
• Our expectation:
T !0 µ0 !
!! d
2
Fermi-Liquid
CFT
0d! 1
2
T
|µ0| !
Remnant scaling behavior
• Boundary AdS Green’s function invariant under
ω → λω , k → λk , α→ λα
– (recall that 4πT = α(3− q2) , µ0 = −2qα)
• in terms of CFT quantities
Teff (µ0) = T
(12
+12
√1 +
µ20
T 2
)
Gµ0/T
Ψ̄Ψ(ω, k) = T 2∆Ψ−df
(ω
Teff,k
Teff;µ0
T
)
Spectral functions and a Quasiparticle peak
A(ω, k) = − 1π ImTr (iγ0GR(ω, k))
2 4 6 80
0.02
0.04
0.06
0.08
!/Teff
A (!
, k)
k = 0k = 1.5k = 1.9k = 2.3!2"−d fit
µ0
T→ 0
GR = 1(√−ω2+k2)d−2∆Ψ
∆ = 1.25
Spectral functions and a Quasiparticle peak
A(ω, k) = − 1π ImTr (iγ0GR(ω, k))
2 4 6 80
0.02
0.04
0.06
0.08
!/Teff
A (!
, k)
k = 0k = 1.5k = 1.9k = 2.3!2"−d fit
µ0
T→ 0
GR = 1(√−ω2+k2)d−2∆Ψ
0−2 2 40
0.01
0.02
0.03
0.04
!/Teff
A (!
, k)
k = 0k = 1.5k = 1.9k = 2.3
k < kF k > kF
(!−µ0)2"−d tail
EF = − 0.65
k = 0
µ0
T= 30.9
µ0
T>(µ0
T
)c
QP Peaks
∆ = 1.25
Fermi Liquid phenomenology
• Single particle Green’s functions encodes the ground-state
G(ω, k) =1
ω − µ0 − k2
2m + Σ(ω, k)
≡ Z
(ω − EF )− vF (k − kF ) + . . .
– Quasiparticle peak (QP) ω = EF , k = kF and a sharp Fermisurface |k| = kF
• Spectral function
A(ω, k) =ImΣ(ω, k)
|ω + µ0 + (k−kF )2
2m + ReΣ(ω, k)|2 + |ImΣ(ω, k)|2
List of Fermi Liquid tests:
• Quasiparticle peaks
• Zero density of states at the Fermi surface A(EF , k) = 0
• Quadratic scaling of the QP-widths with temperature
• Analytical structure of the self-energy Σ(ω, k)
• Linear dispersion of QP-excitations
Identifying EF , kF
−0−1−2−3−4−50
0.05
0.1
A(!
, k)
−1−2−3−4−50
0.05
0.1
A(!
, k)
−1−2−3−4−50
0.05
0.1
!/Teff
A(!
, k) k > kF
k " kF
k < kF
T = 0: A(EF , k) = Zδ(k − kF )⇔ ImΣ(ω, k) =12
(ω − EF )2 ∂2ImΣ∂ω2
∣∣∣∣ω=EF
Identifying EF , kF
−0−1−2−3−4−50
0.05
0.1
A(!
, k)
−1−2−3−4−50
0.05
0.1
A(!
, k)
−1−2−3−4−50
0.05
0.1
!/Teff
A(!
, k) k > kF
k " kF
k < kF
T = 0: A(EF , k) = Zδ(k − kF )⇔ ImΣ(ω, k) =12
(ω − EF )2 ∂2ImΣ∂ω2
∣∣∣∣ω=EF
Dip in the spectrum:zero density of statesat EF for k 6= kF
@@@@@R
AAAAAAAAAAAU
BBBBBBBBBBBBBBBBN
Identifying EF , kF
−0−1−2−3−4−50
0.05
0.1
A(!
, k)
−1−2−3−4−50
0.05
0.1
A(!
, k)
−1−2−3−4−50
0.05
0.1
!/Teff
A(!
, k) k > kF
k " kF
k < kF
T = 0: A(EF , k) = Zδ(k − kF )⇔ ImΣ(ω, k) =12
(ω − EF )2 ∂2ImΣ∂ω2
∣∣∣∣ω=EF
Dip in the spectrum:zero density of statesat EF for k 6= kF
Finite T effect:dip position“smeared”
@@@@@R
AAAAAAAAAAAU
BBBBBBBBBBBBBBBBN ��
Temperature Dependence/Self-energy structure
−4 −3 −2 −1 00
0.2
0.4
0.6
0.8
1
1.2
1.4
!/Teff
A(!
, k)
µ0/T = 10.25µ0/T = 1.04µ0/T = 0.73µ0/T = 0.47µ0/T = 0.23µ0/T = 0.12
EF
(!−EF)/Teff
−5
−2.4 −1.4 −0.4 1.6 2.6
Peak width δ vs. T : δ ∼ T 2
0 0.1 0.20.000
0.001
0.002
0.003
0.004
0.005
0.006
0.007
0.008
0.009
0.010
(T/Teff)2
!
Tc2
210 −1 −2 0
1
2
3
4
5
6
7
8
9
!/Teff
Im"(
!, k
)
k = 1.7k = 1.8k = 1.9
EF = − 0.24
−0.24
2.241.240.24−0.76−1.76(!−EF)/Teff
0
No constant term in ∂ImΣ/∂ω
210−1−2
0.08
0.06
0.04
0.02
0
−0.02
−0.04
−0.06
−0.08 !/Teff
d(Im
")/d
!
k = 1.8k = 1.9k = 1.7
− 0.24
−0.76 0 0.24 1.24 2.24− 1.76(!−EF)/Teff
EF = − 0.24
∆Ψ = 1.05 ∆Ψ = 1.40, µ0/T = 30.9
“ µT
”c' 4
XXXXXy
Dispersion relation
Quasiparticle dispersion: ω − EF = vF (k − kF ) +O((k − kF )2)
−8 −6 −4 −2 0 2 4 6 8
6
4
2
0
−2
−4
−6k/Teff
4.48
2.48
6.48
0.480
FL region
−2.92−3.52
−5.52
−1.52
(!−EF)/Teff
−3.4
!/Teff
µ0 = − 3.4
EF = − 0.48
kF = 1.82
vF = 0.67
Renormalized QP velocityPPPPPq
Dispersion relation
Quasiparticle dispersion: ω − EF = vF (k − kF ) +O((k − kF )2)
−8 −6 −4 −2 0 2 4 6 8
6
4
2
0
−2
−4
−6k/Teff
4.48
2.48
6.48
0.480
FL region
−2.92−3.52
−5.52
−1.52
(!−EF)/Teff
−3.4
!/Teff
µ0 = − 3.4
EF = − 0.48
kF = 1.82
vF = 0.67
Bare (Lorentz dispersion) ω = k at high ω
Renormalized QP velocity
����������������
�����
������:
PPPPPq
Dispersion relation
Quasiparticle dispersion: ω − EF = vF (k − kF ) +O((k − kF )2)
−8 −6 −4 −2 0 2 4 6 8
6
4
2
0
−2
−4
−6k/Teff
4.48
2.48
6.48
0.480
FL region
−2.92−3.52
−5.52
−1.52
(!−EF)/Teff
−3.4
!/Teff
µ0 = − 3.4
EF = − 0.48
kF = 1.82
vF = 0.67
Bare (Lorentz dispersion) ω = k at high ω EF is not AdS zero – a new scale
Renormalized QP velocity
����������������
�����
������:
AAAAAAAAK
PPPPPq
List of Fermi Liquid tests:
• Quasiparticle peaks X
• Zero density of states at the Fermi surface A(EF , kF ) = 0 X
• Quadratic scaling of the QP-widths with temperature X
• Analytical structure of the self-energy Σ(ω, k) X
• Linear dispersion of QP-excitations X
Charge density dependence
0 6 12 18 24 300
0.5
1
1.5
2
µ0/T
−EF/T
eff, k
F/Tef
f, vF
0 6 12 18 24 30
−10
−5
µ0/T
log
Z
− EF/TeffkF/TeffvFmF
0
2
4
(µ0/T)c
(µ0/T)cmF/Teff
0 5 10 150
2
4
6
8
10
12
kF/T
µ0/T− EF/TvFmF/T
− EF(ren)/T
36
30
24
18
6
12
µ0/T,
mF/T
−EF/T,
−EF(ren)/T,
vF
∆Ψ = 1.25
E(ren)F ≡ k2
F
2mF
mF ≡ kFvF
Fermi LiquidZ = mbarevF
kF
vF = Z(1 + ∂kReΣ|kF ) + . . .Z = 1/(1− ∂ωReΣ|EF ) + . . .
restored Lorentzinvariance[cf. Randeria et alcond-mat/0307217]∂kReΣ−∂ωReΣ ∼ 0
XXXXXXXz
Coupling strength dependence
Scaling dimension ∆Ψ proxy for coupling strength
1 1.1 1.2 1.3 1.4 1.5−10
−5
0
!
log
Z
1 1.1 1.2 1.3 1.4 1.50
0.5
1
1.5
2
2.5
3
3.5
!
kF/TeffvFEF/Teff
µ0/T = 30.9
• kF ' constant for all ∆⇒ Luttinger theorem!?
• Pole strength Z dropsexponentially as ∆→ 1.5...[cf. Lawler et alcond-mat/0508747]
• ...but vF ,mF stay finite⇒ emergence of Lorentzinvariance[cf. Randeria et alcond-mat/0307217]
Comparison with previous results
• Always two peaks (are there more?) :
– standard FL,marginal FL [Faulkner,Liu,McGreevy,Vegh] ,. . .
– CFT pole = BH quasinormal mode [Kovtun,Starinets]AdS BH: No a priori reason ∃ only one Dirac quasinormal mode.
– What is the true IR? “Band structure”?
−1.5 −1 −0.5 0 0.50
1
2
3
4
5
6x 10−3
!/Teff
A(!
, k)
" = 1.35" = 1.40" = 1.45" = 1.49
µ0
T= 30.9
marg-FLstandard FL QP
@@R
@@@R
Including a magnetic field
Landau levels in a magnetic field[Preliminary....] 8
0 100 200 300 400 500 600 700 8000
0.5
1
1.5
2
2.5
3
3.5
4
4.5x 10
5
! (H
)
Erel
FIG. 7: Level broadening.
1 2 3 4 5 6 7 8 9 10!1.5
!1
!0.5
0
0.5
1x 10
9
1/H
"#
$ = 1.45
$ = 1.25
$ = 1.05
FIG. 8: Oscillating behavior of the thermodynamical quantities is shown for µ/T = 30.9 and H/T between 0.1 and 1.0. Theconformal dimension is ! = 1.25. The curves correspond to three di"erent values of the conformal dimension. It is seen thatthe period remains the same (count the number of peaks).
!2.5 !2 !1.5 !1 !0.5 0 0.5 1 1.5 2 2.50
0.5
1
1.5
2
2.5x 10
!4
%/Teff
A (
%,k
)
FIG. 9: Intermediate field high charge density system, with |µ0|/T = 30.9, ! = !5/4 and H/T = 3.7. Shown is the dispersion(EDCs) along kx with ky/Teff = 1. The x-momentum takes the values 0.58, 0.72, 0.87, 1.15, 1.44, 1.73, 1.88, 2.02, 2.17, 2.30.We see the splitting of the quasiparticle as well as two branches of gapless modes. The curves suggest disappearance of theFermi surface in favor of a set of disconnected sheets.
ρ(ω) =dN
dω, N(ω) =
∫d2~k nFD(ω, k − kF ;EF )A(ω, k)
[Cubrovic,KS,Zaanen,Sadri]
Conclusion and Outlook
• Theoretical Calculation of an emergent Fermi Liquid(CFT) = (AdS)
probing a charged black hole with electrons
• What is kF on the gravity side?– in string theory geometry depends on the probe...
• Transition to BCS superconductor• Physical interpretation of the ∆ = 3/2 transition?• Interpretation of/co-existence with ω = 0 pole?
[Faulkner,Liu,McGreevy,Vegh]
• Inclusion magnetic field/.../(better understanding of theemergence)
• ... direct connection with experimental data?! ...
Conclusion and Outlook
• Theoretical Calculation of an emergent Fermi Liquid(CFT) = (AdS)
probing a charged black hole with electrons
• What is kF on the gravity side?– in string theory geometry depends on the probe...
• Transition to BCS superconductor• Physical interpretation of the ∆ = 3/2 transition?• Interpretation of/co-existence with ω = 0 pole?
[Faulkner,Liu,McGreevy,Vegh]
• Inclusion magnetic field/.../(better understanding of theemergence)
• ... direct connection with experimental data?! ...
Conclusion and Outlook
• Theoretical Calculation of an emergent Fermi Liquid(CFT) = (AdS)
probing a charged black hole with electrons
Thank you.
• ... direct connection with experimental data?! ...
Boundary behavior of Ψ±
• Asymptotic behavior near z = 0:
Ψ+(z) = zd+1
2 −|m+ 12 |(ψ+ + . . .) + z
d+12 +|m+ 1
2 |(A+ + . . .)
Ψ−(z) = zd+1
2 −|m−12 |(ψ− + . . .) + z
d+12 +|m− 1
2 |(A− + . . .)
• The boundary value of Ψ− is not independent.
(∂z − d/2−mz
)Ψ+ = −/T |z=0Ψ− + . . .
Thus ψ− ∝ ψ+ and A− ∝ A+.• The scaling behavior of G(ω, k):
G(ω, k) =1
N F−F−1+ ∼ z
d+12 −|m−
12 |(ψ− + . . .) + z
d+12 +|m− 1
2 |(A− + . . .)
zd+1
2 −|m+ 12 |(ψ+ + . . .) + z
d+12 +|m+ 1
2 |(A+ + . . .).
• Three different regimes:
G(ω, k) ∼
8>>><>>>:z“ψ−ψ+
+ . . .”
+ z2m“A−ψ+
+ . . .”
m > 12,
z2m“ψ−ψ+
+ . . .”
+ z“A−ψ+
+ . . .”
12> m > − 1
2,
1z
“ψ−ψ+
+ . . .”
+ 1z2m
“A−ψ+
+ . . .”
− 12> m .
Boundary behavior of Ψ±
• Asymptotic behavior near z = 0:
Ψ+(z) = zd+1
2 −|m+ 12 |(ψ+ + . . .) + z
d+12 +|m+ 1
2 |(A+ + . . .)
Ψ−(z) = zd+1
2 −|m−12 |(ψ− + . . .) + z
d+12 +|m− 1
2 |(A− + . . .)
• The boundary value of Ψ− is not independent.
(∂z − d/2−mz
)Ψ+ = −/T |z=0Ψ− + . . .
Thus ψ− ∝ ψ+ and A− ∝ A+.• The scaling behavior of G(ω, k):
G(ω, k) =1
N F−F−1+ ∼ z
d+12 −|m−
12 |(ψ− + . . .) + z
d+12 +|m− 1
2 |(A− + . . .)
zd+1
2 −|m+ 12 |(ψ+ + . . .) + z
d+12 +|m+ 1
2 |(A+ + . . .).
• Three different regimes:
G(ω, k) ∼
8>>><>>>:z“ψ−ψ+
+ . . .”
+ z2m“A−ψ+
+ . . .”
m > 12,
z2m“ψ−ψ+
+ . . .”
+ z“A−ψ+
+ . . .”
12> m > − 1
2,
1z
“ψ−ψ+
+ . . .”
+ 1z2m
“A−ψ+
+ . . .”
− 12> m .