superconductivity and mottness: exact results

52
Superconductivity and Mottness: Exact Results Luke Yeo Edwin Huang

Upload: others

Post on 05-Jan-2022

1 views

Category:

Documents


0 download

TRANSCRIPT

Superconductivity and Mottness: Exact Results

Luke YeoEdwin Huang

Cooper instability

px

py

�gb†pbp0

Eb / e1/g

�(g) = g2

FL ! BCS

µ = 0

no change in size of

Brillouin zone

Qq

Q q < qc

infall to black hole

q < qcMott

Insulator

use holography: competition between gravity and Coulomb

ε

µ

gap with no symmetry breaking!!

charge gap

U � tU

solve the Hubbard Model!!

Cooper instability??

Progress thus far?

DMFT QMC disputes

Sept. 1997 Nov. 1997

� = Mottness

zeros

DetReG(! = 0, p)

6= 0 = 0

counting particles

12

34

5 6

7 8

is there a more efficient way?

E > "p

E < "p

G(E) =1

E � "p

"p E

counting poles (qp)

"p E

zero-crossing

Luttinger’s theorem

n = 2X

k

⇥(<G(k,! = 0))

DetG(! = 0, ~p) = 0

strongly correlated gapped systems

what are zeros?

Im G=0µ = 0

= below gap+above gap = 0

DetG(k,! = 0) = 0 (single band)

( ) d!

Z 1

�1!

ReG(0, p) = Kramers -Kronig

poles+zeros (all sign changes)

singularities of lnG

=2i

(2⇡)d+1

Zddp

Z 0

�1d⇠ ln

GR(⇠,p)

G⇤R(⇠,p)

n

n = 2X

k

⇥(<G(k,! = 0))

are zeros important?

Fermi Arcs

Fermi arcs necessarily imply zeros exist.

Must cross A zero line (DetG=0)!!!

Re G Changes Sign across An arc

+

SC in Doped Mott Insulator: Cooper Instability

for a Luttinger (zero surface)??

C. Setty 2019

LS+ Pair fluctuations=

SYK

Is this true in general?

simplest model with

zeros

atomic limit of Hubbard

model

Ht=0 = �µ

X

i�

ni� +X

i

Uni"ni#

Exact Green Function

GRi (!) =

1 + x

! � µ+ U2

+1� x

! � µ� U2

,

x = 0

GRi (! = µ = 0) = 0

n = 2⇥

✓�µ

µ2 � U2/4

◆ µ > 0 n = 2

n = 1µ = 0

n = 0µ < 0

LSR violation

Can the same physics be

retained in the presence of a

kinetic energy?

Hatsugai-Khomoto Model

HHK = �t

X

hj,li,�

⇣c†j�cl� + h.c.

⌘� µ

X

j�

c†j�cj�

+U

Ld

X

j1..j4

�j1+j3,j2+j4c†j1"cj2"c

†j3#cj4#,

ck� =X

j

eikjcj�

HHK =X

k

Hk =X

k

(⇠k(nk" + nk#) + Unk" nk#) .

⇠k = ✏k � µ

General HK Model

X

k

(⇠k(nk" + nk#) + Unk" nk0#) .

[Ht, HU ] = 0

Solvable Mott transition

Gk�(i!n ! z) =1� hnk�̄iz � ⇠k

+hnk�̄i

z � (⇠k + U)

lower Hubbard band upper Hubbard band

⇣k� = c†k�(1� nk�̄) ⌘k� = c†k�nk�̄

Hubbard band operators

GR� (k,!) =

1

! + i0+ � (⇠k + U/2)� (U/2)2

! + i0+ � (⇠k + U/2)

Fermi liquid QP

{⌃(k,!)

! = ⇠k + U/2

<⌃ = =⌃ = 1

Mott transition: composite excitations

�E = U � 4dt = U �W

insulatorLuttinger Surface

metal

0

.5

1.0

| G; {�k}i =Y

k2⌦1

c†k�k

Y

k2⌦2

c†k"c†k#|0i,

metal is degenerate: SU(2) invariance

entropy across MIT

LY, PWP, PRD, 99, 094030 (2019)

what does the HK model leave out??

dynamical spectral weight transfer

[Ht, HU ] 6= 0

Cooper Instability

H = HHK � gHp

1 = � g

Ld

X

k2⌦0

h1� nk" + n�k#iE � 2⇠k � Uhnk# + n�k"i

| i =X

k2⌦0

↵kb†k|GSi

hnk�i = 0

1 = �g

Z W/2

µd✏

⇢(✏)

E � 2✏+ 2µ,

Eb = hGS|H|GSi � h |H| i 7 0

Eb = �E ⇠ W (1� (U/W )2)e�⇡Wp

1�(U/W )2/g

Cooper Instability

Pair Susceptibility

�(i⌫n) ⌘1

Ld

Z �

0d⌧ei⌫n⌧ hT�(⌧)�†ig

=�0

1� g�0

�0 = hT�(⌧)�†i0 =X

k,p

hTc�k#(⌧)ck"(⌧)c†p"c

†�p#i0

=X

k

G�k#(⌧)Gk"(⌧),

�0(0) =

Zd! N 0(!)

tanh �!2

2!

LdN 0(!) =X

k2⌦0

�(! � ⇠lk) +X

k2⌦2

�(! � ⇠uk )

Solution for T_c

+1

4

X

k2⌦1

�(! � ⇠lk) + �(! � ⇠uk )

1� g�0(0) = 0

Solve

Tc = (W � U)4/5U1/5 e�

⇡e�

45

Wg .

variational wave function

| BCSi =Y

k

(uk + vkb†k|0i

| BCSi =Y

k>0

(u2k + v2kb

†kb

†�k + ukvk(b

†k + b†�k))|0i

| i =Y

k>0

✓xk + ykb

†kb

†�k +

zkp2(b†k + b†�k)

◆|0i

| i =Y

k>0

✓xk + ykb

†kb

†�k +

zkp2(b†k + b†�k)

◆|0i

HK generalization

|xk|2 + |yk|2 + |zk|2 = 1

three variational parameters

h |H| i =X

k>0

(4⇠k + 2U)|yk|2 + 2⇠k|zk|2

�2gX

k,p>0;k 6=p

(x⇤kxk + y⇤kzk)(x

⇤pzp + z⇤pyp)

simplify with `gap’ equation

1 =g

2

Zd!

N 00(!)p!2 +�2

.

N 00(!) =X

k2⌦0

�(! � ⇠lk) +X

k2⌦2

�(! � ⇠uk )

+X

k2⌦1

�(! � ⇠lk) + �(! � ⇠uk )

⇠lk = ⇠k ⇠uk = ⇠k + U

gap equation

1 =g

Wsinh�1(

W � U

2�) +

↵g

Wsinh�1(

U

2�)

� ⌧ U,W

� = (W � U)1/2U1/2e�W2g

gap/T_c ratio

� = (W � U)1/2U1/2e�W2g

Tc = (W � U)4/5U1/5 e�

⇡e�

45

Wg .

limg!0

Tc! 1

non-BCS superconductivity

Bogoliubov excitations

�k� = ukck� � �vkc†�k�

Huangons excitations

�lk� /

p2xk⇣

†k� � �zk⇣�k�

�uk� / zk⌘

†k� � �

p2yk⌘�k�

�u/lk� | i = 0

Excitation spectrum

h |�u/lk� H�u/lk� )†| i = h |H| i+ E

u/lk

Eu/lk =

q⇠u/lk

2+�2

superconductivity affects both bands!

gap opening: superconducting state

Huangon band

spectral function: superconducting state

Doped MI half-filled metal

can we explain the color change?

Al =

Z ⌦

0�(!)d! ⌦/2⇡c = 10000cm�1

Ah =

Z 2⌦

⌦�(!)d! ⌦/2⇡c = 10000cm�1

�Al

Al/ 3%

condensation energy

UV-IR mixing

Bontemps, et al: FGT sum rule at 2eV

condensation energy:HK model

1.00

>1

�WL / O(2� 3%) as in experiments

why?

H = HHK +Hp

[HHK, Hp] 6= 0

dynamical spectral weight

transfer

atomic limit: x holes

2x1-x

density of

states

PESIPES

} }1 + x 1� x

1 + x + �(t/U, x)� =

t

U

ij

�c†i�cj�⇥ > 0

1� x� �

density of

states

beyond the atomic limit: any real system

Harris & Lange, 1967 dynamical spectral weight transfer

is this the general

mechanism of the color

change?

IR theory?

Ne↵ = N0 +N1x�

H =

Zddr

†(r)⇥K(�@2)� µ

⇤ (r) +Hother

cp2↵

Caffarelli-Silvestre extension theorem

(2006)

(��)�

Rn+1 g(x, y)

Dirichlet

Rn f(x)

Neumann

connection to holography

HK model

[Ht, HV ] = 0

Mott gap

[HHK, Hp] 6= 0

non-BCS superconductivity

Tc/ 1

�lk� /

p2xk⇣

†k� � �zk⇣�k�

�uk� / zk⌘

†k� � �

p2yk⌘�k�

WL(g)

WL(g = 0)> 1