寻找开启高温超导机理的钥匙 胡江平 中科院物理研究所 & 美国普渡大学 7/22/2011 Acknowledgements • Materials: Z.X Zhao, X.R.Chen (IOP), X.H Chen(USTC), H.H. Wen(Nanjing), G.F. Chen(People), F. M. Fang, Z.A Xu (ZJU) … • ARPES: H. Ding, X. J. Zhou (IOP) , D. L. Feng (Fudan) …. • Neutron: P. C. Dai, S.L Li(IOP), W. Bao(People) • STM: Q.K Xue, X. Chen, Y.Y Wang (Tsing), S.H. Pan(IOP) • NMR: G.Q. Zheng (IOP), W.Q.Yu (People) • Transport: H.Q. Yuan(ZJU), S.Y. Lee(Fudan) • Optics: N.L. Wang (IOP), Q.M. Zhang(People) • Theory: T. Xiang, Z. Fang, X. Dai (IOP) Z.Y. Wen, G. M Zhang, H. Zai(Tsing), Z.Y. Lu ( People) Q.H. Wang, J.X. Li (Nanjing) J. H. Dai(ZJU), F.C. Zhang(HKU)… • Students & Postdoctor: Chen Fang, Kangjun Seo, Wei-Feng Tsai • S.A. Kivelson (Stanford), B.A. Bernevig( Princeton), Cenke Xu(UCSB) • Lu Yu, D.H. Lee, F. Wang, Q. M. Si History of Superconductivity Tc (K) HgBaCuO 140 TlSrBaCuO BCS Theory 1957 Ginzburg-Landau Theory 1950 BiCaSrCu2O9 Josephson Effect 1962 YBa2Cu3O7 London (two fluid model) 1934 High Tc SC Theory ? 77 CeFeAsO1-x 55 LaBaCuO4 MgB2 35 26 Onnes 1911 Mercury(Hg) 1910 1920 Meissner effect 1933 1930 1940 V3Si Nb3Sn 1950 1960 Bednorz & MÜller Nb3Ge Cuprates 1986 1970 1980 1990 SrFe2As2 FeSe Fe-based 2008 2000 2010 Outline • Theory of conventional superconductors • High Tc Superconductors (cuprates) • Iron-based superconductors and its connections to cuprates The focus of the talk: Conceptual development! Conventional Superconductivity 超导现象的发现: Vanishing of Resistivity R/ 0.10 0.05 零电阻特性 * * * * 超导的转 变温度 TC 4.10 4.20 4.30 T/K Heike Kamerlingh Onnes 1908年荷兰物理学家H.开默林-昂内斯液化氦成功,从而达到一个新的低 温区(4.2K以下)。 1911年,他发现,当温度降到4.2K附近时,汞样品的电阻突然降到零。 他把这种性质称为超导电性。 该工作获1913年诺贝尔物理学奖 超导态基本特性:(1) Meissner effect Walter Hans. Meissner Robert Ochsenfeld Perfect Diamangetic 1933年W.Hans. Meissner 和Robert Ochsenfeld 发现超导体的完全 抗磁性,磁化率χ =-1,即完全抗磁性,又称为迈斯纳效应。 超导态基本特性 : ( 2) Flux quantization h F0 = 2e London equation (1934) London模型是基于两流体模型的超导宏观唯象理论, 引入了London穿透深度(Penetration depth)概念从 超导电动力学角度来描述完全迈森纳效应 唯象解释排磁通效应:超导体体内磁通密度为零,使 得任意电流流过超导体只能在表面,这会使得表面电 流密度无穷大,因而必须引入穿透深度概念。 两流体模型 London穿透深度λ London’s Equation ns e2 ¶J s ns e2 Ñ ´ Js = B, = E mc ¶t m Ñ ´ (Ñ ´ B) = l B 2 ns e 2 Js = A mc Gauge Fixed ! Gauge symmetry Breaking! London’s gap argument i e * e2 J =(Y ÑY - YÑY*) - | Y |2 A 2m m ÑY = 0 If it is viewed as a single particle, the ground state has a single wavefunction who does not mix with other states when small magnetic field is applied. Therefore, this state must be separated from excited states with an energy gap. London Brother’s Contribution: • Meissner Effect is fundamental property of superconductivity • Superconductivity is a macroscopic quantum phenomenon • Superconductivity state is protected by a gap • Gauge symmetry breaking Ginzburg-Landau theory (1950) 基于Landau二级相变理论的唯象理论,描述Tc附近 的现象,想法是要引入一个序参量|Ψ|2=ns/2来的得 到自由能的表达式。 V. Ginzburg L. Landau 自由能表达式: 零场: 有场情况: G-L参数:κ=λL/ξ Type-I超导体: 1 / 2 Type-II超导体: 1 / 2 Ginzburg-Landau theory (1950): order parameter theory • Provide explanation of many properties (thermal, electrodynamic) • Order parameter is complex scalar • Prove superconductivity are macroscopic quantum phenomena • Both phase and amplitude of order parameter are very important • Provide another length scale: coherent length • Predict vortex lattice, type-I and type-II superconductors The Version of Gaps before BCS: • Barden (1930): gap produced by small lattice displacements • CDW gap (Heisenberg, Koppe, 1947): electron wave-packets localized. • SDW( Overhauser): Spin density wave gap …. Isotropic effect: T M 1/ 2 M Frohlich, 1950: indirect attraction between electrons due to exchange of virtual phonons. Bardeen & Pines, 1955: combined treatment of screened Coulomb repulsion and phonon-induced attraction net interaction at low (w ≲ wD) frequencies may (or may not) be attractive. Energy Saving in superconducting state H = Hk + Hv H k = å Pi,e2 / 2m + å q 2j,l / 2m i j e2 Hv = [å + ...] 4pe 0 ij | ri - rj | 1 • • Electron kinetic energy was paid in superconducting Total interaction energy was saved in superconducting state Chester: Phys. Rev. 103, 1693 (1956) BCS理论(1957):超导电性微观理论 • What is superconductivity: H = å e (k)c c + å (D(k)c c + ks ks ks + + k­ -k¯ k Y = Õ (m (k) + n (k)c c + + k- -k¯ k )|0 > E(k) = e (k) + | D(x) | 2 + h.c.) 2 Quantitative Prediction: e Retarded attractive force • BCS ratio: D / kBTc = 1.57 • Tunneling spectrum 2 • Electron Phonon Coupling: a F(w ) • Josepheson Effect e Cuprates 目前: 瞎子摸象和战国时代 Is it a good time? 漫漫长夜还是黎明前的黑暗? To be good and successful Good Timing theorist good problem • • Fundamental questions Conceptual challenge Opportunity successful theory • Identify right problem • Identify most important phenomenon • Ask right question • Solve it at least selfconsistently • • Quantitative results Powerful predictions Does not need to be consistent as a theorist even if physics has to be consistent and novel. Fundamental questions • What is superconductivity? • Why do they become superconductors? • What are the fundamental differences between low Tc and high Tc superconductors? • Why do they become high Tc? Appealing Differences • Cuprates (LaCuO2): • Complicated lattice structure • Layered structures: Two dimensional • Transition metal: 3d electrons • Strong magnetism • Superconductivity induced by doping • Not a good metal • Very short coherent length • Complicated phase diagram • Low superfluid density • Intrinsic dirty materials Conceptual Challenges Good for superconductor in BCS theory: • Normal state: Metal with large density of states • No Magnetism : Magnetism: pairing breaking • Less disorder: especially for d-wave SC In cuprates and iron-pnictides, all of above conditions are violated and SC is robust: Bad metal Strong magnetism Intrinsic strong disorder Difficulty I: What are the fundamental phenomena • Which phases should we focus on? Superconducting? Normal state: Pseudogap? Strange metal? Insulating state: Magnetism? Hidden competing states? Quantum critical phenomena? Difficulty II: Separating different energy scale • Spin, lattice, orbital, charge: mixed strongly! How to rule out other possibilities? Difficulty III: Lacking of quantitative results • Weak coupling: BCS, physics dominated by electrons near Fermi surface. • Strong interactions: physics is dominated locally. • How to compromise? Good questions ruled out • Novel superconducting state: anyon superconducting (SC breaks time reversal) R.B. Laughlin • Electron Fractionalization in pseudogap state: Fisher and senthil • Kinetic energy saving: Anderson Still working • What causes pseudogap or the nature of pseudogap? • Time reversal symmetry breaking, orbital current states • Relation between magnetism and superconductivity • Is superconductivity state much more normal? What do the iron-based superconductors bring to the high Tc table? Fe-based Supercondcutors • Iron-Pnictides: a. 1111 Series: Electron doped: CeO1-xFxFeAs: 41K SmO1-xFxFeAs: 55K PrO0.89F0.11FeAs: 52K SmFeAsO1-x 55k, CaFFeAs: 36K Hole Doped: La[1-x]SrxOFeAs ? b. 122 Series: (both Hole and Electron Doped) Ba1-xKxFe2As2, 38K, BaFe2-xCoxAs2 BaFe2As2-xPx, BaFe2-xRuxAs2 ( isoelectronic doping ) c. 111 Series: Li(Na)FeAs 16k d. 42622: Sr4V2O6Fe2As2 37K • Iron-Chalcogenide : a. 11 Series: FeSe, 8k - 37k, FeSexTe1-x b. 122 Series: K(Cs,Rb)Fe2Se2, 42K Structure of LaOFeAs Fe As above the plane As below the plane Key Question • Are the Fe-based superconductors siblings of cuprates? Similarity Between Cuprates and Oxypnictides • Oxypnictides: • Cuprates (LaCuO2): Transition Metal: 3d electron Layer structure: two dimensions Magnetic ordered state in parent compounds Superconductivity induced by doping Comparable transition temperature (single layer) Very similar phase diagrams Very short coherent length Differences Between Cuprates and Oxypnictides • Cuprates: • Oxypnictides: Cu: 3d9 Fe: 3d6 Spin 1/2 Spin: 0-2 Single d orbits Multi d orbits Simple band structure More complicated band structure Parent compounds: insulator Parent compounds: bad metal Antiferromagnetic pairing symmetry d-wave Collinear-AFM magnetic order pairing symmetry (s-wave ?) Fundamental Questions in High Tc Why are they high Tc? Why are the superconducting states so robust? Theory of High Tc Superconductivity Should Not be so Fancy! Induction in Math VS Physics • Mathematical Induction • Physics Induction Step 1: n=1, Correct Step 1: n=1, Correct Step 2: Assume n=m, Correct Step 2: n=2, Correct Step 3: n=m+1, Correct Step 3: n=3, Correct For any n, it is correct ! 事不过三 • Curpates • Ferropnictides • Ferrochalcogenites Repeat good things three times: 1 = Maybe 2 = Possible 3 = Infinite = Truth Comparison of Phase Diagrams The Basic Problems in Cuprates vs. p-types La2-xSrxCuO4 Nd2-xCexCuO4 AFM SC AFM Temperature (K) n-types SC Dopant Concentration x Magnetism Superconductivity Case I: Cuprates Magnetic Order in Cuprates • Magnetism J H J Si S j ij a. J>0, Antiferromagnetic b. Superexchange: kinetic energy saved c. Between nearest neighbor sites Cu O Superconducting states in curpates D-wave D-wave (kx , k y ) (k y ,kx ) Ck Ck Form in momentum space (k ) 0 (coskx cosk y ) D-wave configuration in real space: pairing between two nearest neighbor sites + + - Pairing Symmetry From Antiferromagnetic Exchange J Si S j ij d (coskx cosk y ) J | d | 2 + - J | s | 2 + + s (cosk x cosk y ) + + + Which one will win? Selection Rules of Pairing Symmetry d (coskx cosk y ) + + - J Si S j ij + s (cosk x cosk y ) + + + + AFM exchange provides pairing force and possible choices of pairing symmetries. Fermi surface topology selects the pairing symmetry. Local AFM exchange interaction in real space + Fermi Surface topology in reciprocal space • • • Doping destroys long range AFM order Doping does not kill short range AFM coupling Effect of electron-electron correlation causes strong renormalization Determine High Tc and pairing symmetry!!! Effectiave t-J model ~ H t Ci C j J Si S j ij ij A: Doping destroys the long range AFM order B: Magnetic exchanges provide the force gluing electron pairs. a. D-wave pairing is favored over S-wave pairing. b. D-wave was really a prediction from the meanfield solution of t-J model. D. Scalapino et al, PRB 34 8190 (1986) Kotliar and Liu (1988), Gros, C.(1988) Susumura, Hasegawa and Fukuyama(1988) Yokoyama and Shiba(1988) Afflect,et al (1988) Zhang F.C and T.M.Rice (1988) Anderson et al, J Phys.Cond. Mat 16 (2004) R755 Van Harlingen DJ. Rev. Mod. Phys, 67, 515, 1995 Case 2: Ferropnictides Parallel Paradigm of Magnetism in Oxypnictides Fe H J1 As J1 J2 S S ij NN i j J2 S S ij NNN i j a. As bridges four nearest neighbor Fe atoms b. Two magnetic exchange coupling parameters T. Yildirim, Phys Rev. Lett 101, 057003; F. Ma et al, arXiv: 0804.3370 Q.Si and E. Abrahams, Phys. Rev. Lett 101, 076401 C. Fang et al, Phys. Rev. B 77 224509; C. Xu et al, Phys. Rev. B 78 020501 Magnetic Order in J1-J2 Model H J1 S S ij NN J1>2J2 , E=-2J1+2J2 AFM i j J2 S S ij NNN i j J1<2J2, E=-2J2 Collinear-AFM Spin wave 1. Spin wave excitation is observed almost in entire zone of reciprocal space. 2. Spin wave excitation is described well by a short-range J1a-J1b-J2 model 3. No clear evidence of stoner continuum J. Zhao et al, Nature Physics 5, 555 - 560 (2009) Band Structure in Fe-Based Superconductors Electron and Hole pockets Pairing symmetry in two band-{t}-J1-J2 model S wave pairing coskx+cosky J1 d wave pairing coskx-cosky + + + + + - Symmetry factors + - S wave pairing coskxcosky J2 d wave pairing sinkxsinky K. Seo, A. B. Bernevig, J. Hu PRL 101, 206404 (2008) Function peaks at Fermi surfaces Pairing strength in two band-{t}-J1-J2 model • S-wave coskxcosky dominates over other symmetry pairing. • There is a small component of dx2-y2 (coskx-cosky). • The interband pairing is very small ( only dxy). Properties of S-wave coskxcosky Pairing Symmetry Order parameters have different signs at electron and hole pockets. • S-wave pairing is the strongest if both electron and hole pockets are small and have close sizes! Superconducting gaps are larger in smaller pockets. - - + + - - Fermi surfaces are generally gapped unless heavy doping crosses gapless line. The transition temperature should be very sensitive to J2( explain the dependence of Tc on the angle dependence of structure) Gapless line Case 3: Ferrochalcogenides Ferrochalcogenides • Magnetism: FeTe: Bicollinear Antiferromagnetic State K0.8Fe1.6Se2: Block AFM Minimum Magnetic Model H = J1 å SS i <ij> NN j + J2 å <ij> NNN Si S j + J3 å Si S j - K <ij> NN NN • J1 is Ferromagnetic • J3 is significant and AFM å (S S ) i <ij> NN j 2 Phase Diagram FeTe Fe1+yTe BCAF IC1 CaFe2As2 CAF J.P.Hu, et al, Arxiv:1106.5169 IC3 AF IC2 FM CAF cuprates 1 K Vacancy Order KFe2Se2: Magnetically driven (c) • Magnetic Energy Save: •Moment reduction from spin wave | J1 | +J2 + 2J3 m 0.1B 2J3 d m » 0.4 mB Vacancy and Magnetic frustration couple strongly! Chen Fang, et al arXiv:1103.4599 J3 Robust S-wave in KFe2Se2 S wave pairing coskx+cosky + + + + J1 + + - S wave pairing coskxcosky J2 S wave pairing Cos2kx+cos2ky J3 d wave pairing sinkxsinky D µ coskx cosky - d (cos2kx + cos2ky ) C. Fang, et al, arxiv:1105.1135 d wave pairing cos2kx-cos2ky J1 is ferromagnetic. J3 enhance s-wave pairing • Curpates: J1 • Ferropnictides J1 , J2 • Ferrochalcogenites J2 , J3 J.P. Hu and H. Ding, arxiv 1107.1334 Prediction of Future High Tc • Strong AFM • Strong Band Renormalization • Collaboration between AFM and Fermi Surface. Superconducting Gaps Measured from ARPES • NdFeAsO0.9F0.1 • Ba1-xKxFe2As2 • NaFe0.95Co0.05As • BaFe2-xCoxAs2 • FeSexTe1-x • LiFeAs • (K,Cs)Fe2Se2 • Almost isotropic gaps around each Fermi pockets. A strong support for local magnetic exchange driving SC ? Determining the pairing symmetry in KFe2Se2 will be critical! Conclusions: Can we break the last rule of Matthias? High sysmetry is best; Peaks in density of state are good; Stay away from oxygen; Stay away from magnetism; Stay away from insulator; Bernd Matthias The Last Matthias Rule: Stay away from Theorists! Fact: None of superconductor was predicted by theorists! Two Solutions: • Convert to a half theorist + a half experimentist • Break it ( a task for young generation)