固体電子論特論 - riken · h. seo, j. merino, h. yoshioka, m. ogata, j. phys. soc. jpn. 75...
TRANSCRIPT
奈良女子大学大学院集中講義
固体電子論特論Introduction to strongly-correlated material science: with a focus on molecular conductors
Hitoshi Seo (妹尾仁嗣)
RIKEN (理化学研究所)
レポート
・自分の研究(の計画でも)の紹介 A4 1枚程度・講義の感想。
オプション:以下の論文を1本読み、A4半分~1ページくらいのレジュメにまとめよ。1. H. Seo, H. Fukuyama, J. Phys. Soc. Jpn. 66, 1249 (1997) (mean-field)
2. H. Shimahara, J. Phys. Soc. Jpn. 58, 1735 (1989) (RPA)
3. M. Tsuchiizu, H. Yoshioka, H. Seo, 3人の共著論文であればどれでも (bosonization)
plan
固体中の電子の互いにクーロン相互作用のために避けあいながら量子
力学の法則に則って運動している。相互作用の効果が特に強い「強相
関電子系」では、モット転移、高温超伝導、近藤効果などの様々な異常
な振る舞いを示し活発に研究されている。この講義では、分子性導体を
中心に強相関電子系を理論的に記述するための枠組みとその基本的
性質を解説し、最近のトピックを紹介する。学部レベルの量子力学と統
計力学を前提として、初学者にも分かりやすく説明したい。
plan
0. イントロダクション
強相関電子系とは.遷移金属酸化物と分子性導体
1.多フェルミオン系の量子力学とバンド理論
1-1.多体系の量子力学の復習1-1-1. 行列表示1-1-2. フェルミオン系の第2量子化1-1-3. 補足:クーロン相互作用の期待値、場の演算子
1-2. 強束縛モデルとバンド理論1-2-1. 強束縛モデル1-2-2. 1次元鎖の例とバンド理論1-2-3. 2量化とパイエルス不安定性1-2-4. d-pモデルと軌道混成
plan
2.強相関系における金属絶縁体転移
2-1. バンド理論の破綻:モット絶縁体と電荷秩序2-1-1. ハバードモデルと反強磁性の平均場理論2-1-2. 拡張ハバードモデルと電荷秩序
2-2. 強結合からのアプローチと数値計算2-2-1. ハイゼンベルグモデルと擬スピン2-2-2. 数値計算手法とその結果の紹介(スライド)
3.超伝導
3-1. BCS理論3-1-1. 平均場理論3-1-2. 引力の起源3-1-3. BCS波動関数と変分法、および擬スピン表示
3-2. 非従来型超伝導3-2-1. 感受率と金属状態の不安定性3-2-2. 分子性導体における超伝導発現機構(スライド)
goals of this lecture
simple picture of band theory: what it can/cannot explain
what are Mott insulators, charge ordering
why do we study them
understand what (extended) Hubbard model is
general idea about what comes out from these models
connections to experimental systems = what is explained, what is not
band theory led to semiconductor technology. “success story of electron theory of solids”but…
Although the band theory of solids had been very successful in describingvarious electrical properties of materials, in 1937 Jan Hendrik de Boer andEvert Johannes Willem Verwey pointed out that a variety of transitionmetal oxides (TMOs) predicted to be conductors by band theory (becausethey have an odd number of electrons per unit cell) are insulators.[3] NevillMott and Rudolf Peierls then (also in 1937) predicted that this anomaly canbe explained by including interactions between electrons.[4]
Wikipedia : “Mott insulator”
carrier doping to Mott insulator → high-Tc superconductivity
layered perovskiteCuO2 plane
OCu
1986: discovery of high-Tc cuprates by Bednorz and Muller
Mott insulator http://for538.wmi.badw.de/projects/P4_crystal_growth
metal-insulator transition induced by doping
layered perovskiteCuO2 plane
OCu
1986: discovery of high-Tc cuprates by Bednorz and Muller
insulating state due to“strong correlation”
spin degree of freedom is left → magnetic order
Mott insulator http://for538.wmi.badw.de/projects/P4_crystal_growth
OCu
carrier doping in transition metal oxides (TMOs)
La
La2CuO4 : (La3+)2 Cu2+ (O2-)4
(3d)9
layered perovskite structure
La2-xSrxCuO4 :
(La3+)2-x(Sr2+)x Cu(2+x)+ (O2-)4
(3d)9-x
Sr
hole doping“filling control”
metal-insulator transition induced by pressure“band-width control”
metal vs insulator: how can we detect?
temperature dependence of electrical resistivity rr = R×A/l R: resistance from Ohm’s law V=RI
K. Kanoda, J. Phys. Soc. Jpn. 75 (2006) 051007
(Mott)insulator
metal(& SC at low-T)
metal-insulator transitions in TMOs
“Metal-insulator transitions”, M. Imada, A. Fujimori, Y. Tokura, Reviews of Modern Physics 70, 1049 (1998)
metal-insulator transitions in TMOs
A B:TM
O
3-D perovskite structuregeneral formula: ABO3
Mott insulator
metal
A site: Y3+ → Ca2+
B site: Ti3+ → Ti(3+d)+
(3d)1 → (3d)1-x
metal-insulator transition induced by doping
metal-insulator transitions in TMOs
metal-insulator transitionat a certain temperature Tc
metal-insulator transitions in TMOs
metal-insulator transition induced by magnetic fieldCMR manganites
molecular conductors
TM
ET
TM2X ET2X
Se
Se
Se
Se
CH3
CH3
H3C
H3C S
S
S
S
S
S
S
S
DCNQI
DCNQI2X
(TMTSF)
(BEDT-TTF)
“X”
“X”
“X”
spin-density-wave (SDW) in TMTSF2PF6
Review: Denis Jerome, Chemical Reviews 104 5565 (2004)
metal-insulator transitions in molecular conductors (1)
“Peierls instability”~ band theory
metal-insulator transition induced by pressure
“band-width controlled Mott transition”
K. Kanoda, J. Phys. Soc. Jpn. 75 (2006) 051007
Mottinsulator
metal(& SC at low-T)
metal-insulator transitions in molecular conductors (2)
k-(BEDT-TTF)2X
K. Kanoda, J. Phys. Soc. Jpn. 75 (2006) 051007
metal-insulator transitions in molecular conductors (2)
electron localizeson each dimer
“dimer Mott” insulators
k-(BEDT-TTF)2X
metal-insulator transitions in molecular conductors (3)
q-ET2RbZn(SCN)4 a-ET2I3
K.Bender et al., Mol. Cryst. Liq. Cryst. 108 (1984) 359N.Tajima et al. J. Phys. Soc. Jpn. 69 (2000) 543
H.Mori, S.Tanaka, T.Mori, Phys.Rev.B 57 (1998) 12023
Vij
“crystal of electrons”
strong correlation
charge ordering
metal-insulator transitions in molecular conductors (3)
effective model for molecular conductors
・ transfer integrals tij~ 0.1-0.3 eV
from Q. chem. / 1st principles band calc.
・ on-site Coulomb energy U~ 0.5-1 eV
・ inter-site Coulomb energy Vij ~ 0.2-0.5 eV
・ electron-phonon interaction
tij
extended Hubbard model based on molecular orbitals
=
U
Vij
妹尾仁嗣、鹿野田一司、福山秀敏, 物理学会誌2004年11月号解説
H. Seo, C. Hotta, H. Fukuyama, Chemical Review 204 (2004), 5005
H. Seo, J. Merino, H. Yoshioka, M. Ogata, J. Phys. Soc. Jpn. 75 (2006), 051009
systematic study on molecular conductors
H = S tij ( cis† cjs + h.c. ) + U S ni↓ni↑ + S Vij ni nj
H. Kino, H. Fukuyama, J. Phys. Soc. Jpn. 64 (1995), 1877; 2726; 4523; 65 (1996), 2158
reviews
end of chapter 0
2. 強相関系における金属絶縁体転移
2-2-2 数値計算手法とその結果の紹介
☆ exact diagonalization of large matrix
・ power method
in number operator basis, state vector ~ 4N, matrix 4N×4N
what we want: ground state = eigenstate with largest absolute value
・ Lanczos methodwe don’t stock the Hamiltonian matrix, only vectorsdo recursive matrix * vector calculations matrix elements are calculated each time
=1019 for N=16
☆ variational Monte-Carlo method
cf) https://github.com/cmsi/MateriAppsLive/wiki/MaLiveTutorial
☆ DMFT = dynamical mean-field theory(outside the scope here)
theories on Mott transition: numerical
DMFTReview: Georges et alReviews of Modern Physics 68, 13 (1996)
paramagnetic insulator: reproduced.
Ground state phase diagram of 1/2-filled t-t’-U Hubbard model.
DMFT Kyung-Tremblay PRL 2006
theories on Mott transition: numerical
PIRG Morita et al JPSJ 2002 VMC Watanabe et al PRB 2008
VCA Laubach et al. PRB 2015
theories on charge ordering: numerical
Mila-Zotos, Europhysics Letters 1993Lanczos
Ejima et al, Europhysics Letters 2005DMRG (density matrix renormalization group)
Ground state phase diagram of 1-dimensional 1/4-filled t-U-V model.
theories on charge ordering: analytical
Yoshioka-Tsuchiizu-Suzumura JPSJ 2000
Bosonization
Yoshioka-Tsuchiizu-Seo JPSJ 2006
extension to q1D
Ground state phase diagram of 1-dimensional 1/4-filled t-U-V model.
theories on charge ordering: numerical
finite-T phase diagrams
Otsuka et al, JPSJ 2005, Quantum Monte-Carlo
Quasi-1D model coupled to classical phonons (like SSH model)
theories on charge ordering: numerical
ground state of 2-dimensional t-U-V model on square lattice
Hanasaki-Imada, JPSJ 2005 correlator projection method (extension of DMFT?)
Merino PRL 2007 DMFT
H = tpS ( cis† cjs + h.c. )
+ tcS ( cis† cjs + h.c. )
+ U S ni↓ni↑
+ Vp S ni nj + Vc S ni nj
tp
U
Vc
tcVp
H. Seo, J. Phys. Soc. Jpn. 69 (2000) 805
tp = -0.094 eV, tc = 0.021 eV (extended Huckel)
・mean-field calculations(U=0.7 eV, vary Vp ,Vc as parameters)
competing “stripe-type” CO patterns ~ frustration
q-ET2X
“vertical” “horizontal” “diagonal”
theories on charge ordering: numerical
3. 超伝導
3-2-2. 分子性導体における超伝導発現機構
high-Tc cuprates
TMTSF2PF6 k-(BEDT-TTF)2X
2-dimensional Hubbard model:
1/2-filling : Mott insulator, usually AF
+ doping → d-wave SC
Misawa-Imada, PRB 2014, VMC
2-dimensional Hubbard model:
1/2-filling : Mott insulator, usually AF
+ doping → d-wave SC
Misawa-Imada, PRB 2014, VMC
Quasi-1-dimensional Hubbard model
(tx >> ty >> tz) at 1/4-filling:
as increasing ty, SDW → SC
Shimahara, JPSJ 1986, RPA
Figure from H. Oike, Ph. D. thesis
superconductivity near Mott transition
similarities between high-Tc cuprates and organic SC?
T(K)hole
doping
electron
doping
AF
Mott ins.
SCSC
Mott insulator
doping rate
single-band Hubbard model
superconductivity near Mott transition
similarities between high-Tc cuprates and organic SC?
Paramagneticinsulator
Paramagneticmetal
Super--conductor
AF insulator
k-ET2X
Theory: Is SC reproduced
by 1/2-filled Hubbard model ??
K. Kanoda
Figure from H. Oike, Ph. D. thesis
Mott insulator
single-band Hubbard model
H = t S (cis†cjs + h.c.) + t’ S (cis
†cjs + h.c.)
+ U S ni↓ni↑
stable superconductivity (in theories) ??
anisotropic triangular lattice
𝒕𝒕𝒕′
𝑼
H. Morita, S. Watanabe, and M. Imada
JPSJ 71, 2109 (2002)
PIRG CDMFT VMC
B. Kyung and A.-M. S. Tremblay
PRL 97, 046402 (2006)
T. Watanabe et al.
PRB 77, 214505 (2008)
VCA
M. Laubach et al.
PRB 91, 245125 (2015)
AF (Mott insulator) → OK
SC → controversial (no SC phase in many works…)
cf) doped Hubbard model : stable SC phase
early works: d-wave superconductivity
H. Kino, H. Kontani, JPSJ 67, 3691 (1998); J. Schmalian, PRL 81, 4232 (1998);
H. Kondo, T. Moriya, JPSJ 67, 3695 (1998); K. Kuroki, H. Aoki, PRB 60, 3060 (1999)
K. Miyagawa et al., Chem. Rev. 104, 5635 (2004)
in experiments, ubiquitous superconductivity
k-ET
H. Kobayashi et al., Phys. Rev. B 56, R8526 (1997).
AF ?
l-(BETS)2GaBrzCl4-z
l-BETS
K. Kanoda, JPSJ 75, 051007 (2006)
• 4 molecules in a unit cell (4-band model)
• transfer integral:
• Coulomb interaction:
C. Hotta, PRB 82, 241104(R) (2010); M. Naka and S. Ishihara, JPSJ 79, 063707 (2010)
K. Kuroki et al., PRB 65, 100516(R) (2002): V=0
A. Sekine et al., PRB 87, 085133 (2013): V≠0
D. Guterding et al., PRB 94, 024515 (2016)
Strong coupling theory (effective Model)
Weak coupling theory (superconductivity)
Mean field approximation
H. Kino, H. Fukuyama, JPSJ 64 (1995) 2726; 65 (1996) 2158: V=0; H. Seo, JPSJ 69, 805 (2000): V≠0
let’ go back to the original 3/4-filled model
competition between 2 SC states
dxy (“high-Tc type”) vs extended-s+dx2-y2
H = S tij ( cis† cjs + h.c. )
+ U S ni↓ni↑ + S Vij ni nj
• 4 molecules in a unit cell (4-band model)
• transfer integral:
• Coulomb interaction:
H = S tij ( cis† cjs + h.c. )
+ U S ni↓ni↑ + S Vij ni nj
・Charge Jastrow factor :
・One-body part :(Slater determinant)
・Spin Jastrow factor :
up to 78th neighbor
(within ~ 4Rx radius)
Jastrow-type Trial Wave Function
variational Monte-Carlo method: ground state T=0 properties
stable superconducting phaseextended-s+dx2-y2 type symmetry
we need intersite V’s to stabilize it
3-fold CO-1(non-magnetic)
dimer-AFMott insulator
polar CO (magnetic)
1152 sites (24×24×2)VMC ground state T=0 phase diagram
paramagnetic
metal
H. Watanabe, H. Seo, S. Yunoki, J. Phys. Soc. Jpn. 86, 033703 (2017)
1st order Mott transition
(no SC along this line)
wide region where SC condensation energy is
finite along the border of charge instability
importance of charge correlation
3-fold CO-1(non-magnetic)
polar CO (magnetic)
1152 sites (24×24×2)VMC ground state T=0 phase diagram
paramagnetic
metal
H. Watanabe, H. Seo, S. Yunoki, J. Phys. Soc. Jpn. 86, 033703 (2017)
dimer-AFMott insulator