Contents 1 1 0 2 1 2.4A modern,but waytooshorthistoryofthetheoryofsuperconductivityatahigh n a temperature. 1 J JanZaanen 4 1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.2 PushingBCStothelimits: thespinfluctuationsuperglue. . . . . . . . . . . . . . 3 ] 1.3 ThelegacyofPhilipW.Anderson(I):Mottness. . . . . . . . . . . . . . . . . . . 8 n o 1.4 Thelegacyof PhilipW. Anderson(II):ResonatingValenceBondsanditsdescen- c dants. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 - r 1.5 Theorythatworks:thecompetingordersofthepseudogapregime. . . . . . . . . 16 p 1.6 Quantumcriticalmetalsandsuperconductivity. . . . . . . . . . . . . . . . . . . . 20 u s . t a m - d n o c [ 2 v 1 6 4 5 . 2 1 0 1 : v i X r a i 1 2.4 A modern, but way too short history of the theory of superconductivity at a high temperature. JanZaanen Instituut-LorentzforTheoreticalPhysics,LeidenUniversity,Leiden,TheNetherlands 1.1 Introduction ............................................................................ 1 1.2 PushingBCStothelimits:thespinfluctuationsuperglue. ................................ 3 1.3 ThelegacyofPhilipW.Anderson(I):Mottness. ......................................... 8 1.4 ThelegacyofPhilipW.Anderson(II):ResonatingValenceBondsanditsdescendants. .... 10 1.5 Theorythatworks: thecompetingordersofthepseudogapregime. ....................... 16 1.6 Quantumcriticalmetalsandsuperconductivity. .......................................... 20 1.1 Introduction J.Zaanen-LeidenUniversity,TheNetherlands Onecannotwritethehistoryofawarwhenitisstillragingandthisiscertainlyapplyingtothe task we are facing dealing with the theory of superconductivity in the post BCS era. The BCS theory was of course a monumentalachievement that deserves to be counted among the greatest triumphsinphysicsofthetwentiethcentury.Withthediscoveryofhigh-T superconductivityinthe c cupratesin19861aconsensusemergedimmediatelythatsomethingelsewasatworkthantheclassic (i.e. phonondriven)BCSmechanism. Aquarterofacenturylaterthissubjectisstillcontentious, perhapsevenmoresothanatanyinstanceinthepast. Itisplainlyimpossibletowritean’objective’ history of any kind and this piece has no pretense in this regard. I thoughtit would be useful to writeinsteadsomesortofeyewitnessaccount,servingtwopotentialreaderships: inthefirstplace thenewcomerswhowonderabouttheprehistoryofsomestandardnotionsthatappearasestablished andcontroversialatthesametime. Theothercustomersarerightnownomorethanapotentiality. Imaginethat the theoryof high-T superconductivitywill turn truly gloriousat some pointin the c future. The era that I am discussing here will then have a similar status as the ’dark ages’ that precededthediscoveryofBCStheoryasdescribedelsewhereinthisvolume. Icanimaginethatan eyewitnessaccountlike the present one mightbe entertainingfor the professionalhistorianswho wanttochroniclethistriumphof21-thcenturyphysics. Iamawarethatthisexpositionisfarfromcomplete–therearequiteanumberofideasthatare very interesting by themselves, while they might be on the long run more consequentialthan the mattersIdiscusshere.However,giventhelengthrestrictionsIjusthavenotthespacetogiveatruly 1K.A.Mu¨ller,G.Bednorz,Z.Phys.B64(1986)189. 1 2 Superconductivity comprehensiveoverviewofeverythingthathasbeenexplored. Ihavetorestrictmyselftoasketch ofthehistorybehindthenotionsthatarerightnowmostontheforeground.Similarly,Idonotclaim rigorwith regardto distributingthe creditsto the variousindividualsinvolved. Althoughatsome instancesIwilldrawattentiontocontributionsthatIfindareunderexposed,thismaterialissurely notintendedasbackgroundmaterialforawardcommittees. Thiseyewitnessaccountisalsocoloredwithregardtothesubstancesthatarediscussed.Whether we like it or not, humanityis a religiousspecies and belief systems play a big role when science isstillinthemaking. OnebetterbeawareofitandinphysicsIliketobemyselfa’denomination tourist’.Ihavedwelledaroundamongthevariousbeliefsystems,gettingintoittoadegreethatIfelt myselftobeabeliever,tohopatthatpointtoanextsystem. InthecourseoftimeIhavesettledinto a faith that is rightnow quite popular: the ’agnostic denominationin high-T superconductivity’. c Their dogma is as follows: ”the research in unconventional superconductivity is flourishing. It is howeverowned by the experimentalists. Energizedby the high-T puzzle, the experimentation c in quantum matter leaped forward in the last twenty years, due to a variety of highly innovative instrumentaldevelopments.Thesenewexperimentalfactshavehoweverjustdeepenedthemystery forthe theorists. The agnosticzealis anchoredin the convictionthatthese fantastic new data are beggingforan explanationin termsof mathematicalequationsof a beautyandelegancethatwill measure up to the productsof Einstein and Dirac. Right now we have not the faintest idea what theseequationsarebutthatshouldnotkeepusawayfromattemptingtofindthem.” Surely this has imprinted on the organizationof this text. I will discuss matters in a sequence thatisinahistoricalorder(fromoldtonew),whileatthesametimethesenseoftheawarenessof dealingwithamystery(theagnosticdogma)isincreasing. The pointof departureis the highlyinfluential’denomination’thatrests on the believethatthe theoreticalframeworkasemergedinthe1950’sinessencestillsufficestoexplainthephysicsofthe cupratesandsoforth.ThenormalstateisaFermi-liquidthatisperhapsabitobscuredbyperturba- tiveprocesses. However,theessence ofBCS theoryisstill fullyintactandto explainthe high-T c superconductivityoneneedsa ’superglue’thathasmorepunchthanthetraditionalphonons. This superglueis than supposedto be embodiedby spin fluctuations. Since this ’fifties paradigm’has beenaroundforalongtimeithasturnedintoaframeworkthatpretendstobequitequalitative.This ”holytrinity”(LocalDensityApproximation(LDA)bandstructure,RandomPhaseApproximation (RPA)spinfluctuationsuperglue,Eliasbergequationsforthesuperconductivity)isrightnowinthe middle of a revival in the context of the superconductivityin the iron pnictides. It surely should betakenquiteseriouslybutmorethananywhereelsethedangerofreligiousillusionsisaroundthe corner. InthisregardIperceiveitasbeneficialtobeawareofthelong,andcontentioushistoryof thespinfluctuationsuperglueidea. Thisisthenarrativeofsection2.4.2. Althoughithasbeenabitontheretreatrecently,thenextinfluentialschoolcanbecalled’RVB, andrelatedphenomena’. Thisreferstoahighlyimaginativesetofideas,thatemergedinthewild early years of high-T superconductivityunder the inspired intellectual leadership of Phil Ander- c son. Regardlesswhetherithasanythingtosayaboutthereallifeofhigh-T superconductorsthese c deserveattentionifnotonlybecauseitdevelopedintoaquiteinterestingandinnovativebranchof theoreticalphysics. The spin fluctuation theoriesare rooted in the perturbativeschemes of the fermiologythat was establishedinthe1950’s. TheResonatingValenceBond(RVB)ideaseventuallyrestontherecog- nitionthatdopedMottinsulatorshavetobedifferent. Thisstorystartswiththelastskirmishesof a war that is much older: the understanding of the Mott-insulating state itself. As I will discuss in section 2.4.3, this pursuitstarted a long time ago, while it settled in the late 1980’swhen, as a byproductofthehigh-T research,aratherdetailedandcompleteunderstandingofthereallifeMott c insulatorswas established. Section2.4.4is dealingwith the RVB schoolof thought. Thisstarted from an appealing idea by Phil Anderson that evolved during the wild early years of high-T in c something of a mathematicalscience by itself. Differentfrom the 1950’s spin fluctuations, these ’gaugetheoriesofstronglycorrelatedfermions’restconceptuallyonthediscoveriesinthe1970’s Amodern,butwaytooshorthistoryofthetheoryofsuperconductivityatahightemperature. 3 inhighenergyphysics.Althoughitstilllacksbothconvincingexperimentalsupportandrealmath- ematicalcontrol,thesetheoriessurelyservewellthepurposeofconfusingthehighenergytheorists. Atstakeisthemeaningofthegaugeprinciple. Infundamentalphysicsitisaprimordialprinciple. The’RVB’gaugetheoriesareinsistinghoweverthatthegaugeprincipleisgovernedbycollectivity, tellinghowlargenumbersof’un-gauged’microscopicdegreesoffreedomco-operateinunisonto generate’gauged’collectivebehaviors. Keepingthehistoricalorder,insection2.4.5Iwillfocusonthesubjectthathasbeendominating the field duringthe last ten yearsor so: the physicsof the underdoped’pseudogapregime’. This has been so much on the foregroundbecause it turned out that there was much to discover using themodernexperimentaltechniques. Itisnowclearthatthereisacollectionofexoticcompeting orders at work, including stripes, quantum liquid crystals and spontaneous diamagnetic currents. Apparentlytheoreticalphysicsasweknowitdoesgiveagriponorderlythings,becausetheorists didmakeabigdifferenceinthisparticularpursuit. InthelastsectionIwilldiscusswhatmaybethebeginningofthehistoryofthefutureofhigh-T c superconductivity.Thisrevolvesaroundtheideathatthequantumphysicsofthemysterioussuper- conductorsisinthegripoftheoverarchingsymmetryprincipleofscaleinvariance.Inanalogywith the classical criticalstate, this ’quantumcriticality’ is believedto be caused by a phase transition happeningatzerotemperature,drivenbyquantumfluctuations. Giventheincreasedunderstanding of the competing orders at work in the pseudogapphase this notion is at present becomingquite popular in the cuprates. In the context of the heavy fermions it has been however obvious for a long time that their superconductivityis closely related to quantumphase transitionsand at stake is whether these share the essential physics with the cuprates. This is the stronghold for the ag- nostictheoristsbecausenowhereelseitissoobviousthatwearelackingmathematicalmachinery thatworksasinthecontextofquantumcritical’fermionic’metals. Iwillendthischronicleonan optimistictonebyexpressingmyhopethatthestringtheoristsmightactuallyhavesuchmachinery intheoffering. 1.2 Pushing BCS to thelimits: the spinfluctuation superglue. The BCS theory is a remarkable achievement. A constant factor in the debate during the last twentyfiveyearshasbeenwhetheritisatallpossibletogetaroundtheessenceofBCStoexplain superconductivityat a high temperature. Its centralwheel is the Cooper instability which in turn rests on the physicsof nearly free fermions. A first conditionis that the normalstate closely ap- proachesthe Fermi-liquid fixed point near the superconductingtransition temperature. A second conditionisthatthenetresidualquasiparticleinteractionsturnattractiveintheappropriatepairing channel. AtthisinstancetheFermi-surfaceexertsitpower,bysinglingoutthepairingchannelas uniquelysingular. Feedinga genericfermionsystem with a verystrongattractiveinteractionthat overwhelmstheFermi-energywouldjustcauseaclumpingoftheparticlessincein(nearly)classical systems thereis no specialstability associated with two particle boundstates2. The remainderof BCS is simple. Cooperpairsare bosonsandthe CooperpairsBose condenseat themomentthey form,ofcoursemodulothepossibilityof’dangerous’thermalorderparameterfluctuations. ThequestioniswhetherthenormalstateissufficientlylikeaFermi-liquidfortheBCSmechanism toapplydirectly. Thishasbeendividingthemindsmorethananythingelse–the’radical’theories discussedinthenextsectionsdepartfromtheassertionthatthenormalstateissomethingelsethan aFermiliquid. AssumingthegovernanceoftheFermi-liquid,theonlyissueisthentoexplainthe 2Althoughitisusuallyworkedundertherug,thisproblemhuntsideasinvolvinglocalpairing,likethebipolarons. 4 Superconductivity originoftheattractiveinteractions. Thisisfarfromtrivialsinceelectronsactuallyrepeleachother ratherstronglyonthemicroscopicscalethroughtheCoulombinteractions. Muchofthehardwork ofdiminishingthesebareinteractionscanbeascribedtothemagicoftheFermi-liquid.Theubiqui- tousnatureofthequasiparticlegasismuchmoreofamysterythanisoftenrealized.Theparticular limitingcasewhereonthemicroscopicscaletheinteractionsaresubstantiallysmallerthanthebare Fermienergyisnowadayswellunderstood,althoughthisinvolvesthefanciful’Polchinski-Shankar’ functionalrenormalizationgroup.Byprincipleonehastokeeptrackofaninfinityofcouplingcon- stants’spanningup’theFermi-surface,whenintegratingoutshortdistancedegreesoffreedom. In therenormalizationprocess,onefindsthatbesidesthe’harmless’marginalityoftheLandauFfunc- tions, onehas onlyto watch a possible (marginal)relevancyin the Cooperchannel.3 The bottom line is that with the exception of the BCS superconductivitythe Fermi-liquid is generically (i.e., awayfromspecialnestingconditions)stableasaninfraredfixedpoint. Theproblemishoweverthatthesystemsofinterestinthepresenteraareinvariablyofthekind wherethemicroscopicinteractionscaleexceedsthebarekineticenergybyorder(s)ofmagnitude. One can no longer rely on the perturbative renormalization group and it is just an experimental fact that Fermi-liquids are nevertheless ubiquitous. A classic example is 3He. At microscopic distancesthisislikeaquantumversionofadense,highlycorrelatedVanderWaalsliquid.However, uponcoolingit down to milli-kelvinsone findsa long wavelengthphysicscorrespondingwith an impeccableFermi-gasof quasiparticles. The miracleis thatthe ’hardHelium balls’ havebecome effectivelycompletelytransparentrelativetoeachother,onlycommunicatingviathePauliprinciple, whiletheeffectsofinteractionsisthatthese’quasiHeliumatoms’arethreetimesheavierthanthe realatoms. IlearnedfromRobertSchriefferthatin the1960’s-70’stheoriststriedhardtoexplain thiswonder,notgettinganywhere.4 Fermi-liquidsareunreasonablyresilientandthisisperhapsthe bestreasontotakethe’conservative’theoriesofhigh-T superconductivityserious. c Thenextquestionbecomes:howtoturntheresidualweakquasiparticlerepulsionsoftheFermi- liquid into serious attractive interactions? The classic BCS answer is of course phonons. The exchange of bosons that are not governed by the gauge principle will generically give rise to an inducedattractiveinteraction.Thephononsarejustperfect:theyareexternaltotheelectronsystem since they originate in the lattice, their interactions with the quasiparticles emerge naturally and theyhavetheaddedbenefitofcausingretardedinteractions. Phononenergiesaretinycomparedto theFermienergyofconventionalmetalsandtheassociated’Migdalparameter’canbeexploitedto simplifytheperturbationtheoryintoavirtuallyexact’resummed’Migdal-Eliashbergtheory. Thiscontrolovertheclassicphonon-drivenBCStheoryexplainsthesociologicalearthquakethat rumbledthroughthephysicscommunityfollowingtheBednorz-Mu¨llerdiscoveryofsuperconduc- tivity at 34K in La2−xBaxCuO4 in 1986. Back then it was believed that phonons could not be responsible for a T exceeding 30 K and a consensus emerged immediately that something new c was at work. In hindsight this is actually a bit embarrassing. It is a number game and numbers aredangerous. Thestrongestclaimrestedonstabilityconsiderations5thatweresubsequentlychal- lenged by theorists of Ginzburg’s high-T superconductivity group at the Lebedev institute6. In c fact, in 2001superconductivityat 40 K was discoveredin the simple MgB system. In no time a 2 consensusemergedthatthesuperconductivityisphonondriven,witha numberofmaterialdetails (high phononfrequencies, multi-bands)conspiringto optimize the phononmechanism, while the phononsresponsibleforthesuperconductivityareunrelatedtothefactorsrenderingthecrystaltobe 3ForafancifulrecentdiscussionseeS.A.Parameswaran,R.ShankarandS.L.Sondhi,arXiv:1008.2492. 4Remarkably,asaveryrecentdevelopmentitappearsthatthequantum-fieldsgravityholographicdualitiesofstringtheory shedanewlightonthisoldproblem: theFermiliquidisrelatedtothegravitational stabilityofspecialblackholeswith ’fermionhair’,seesection2.4.5. 5M.L.CohenandP.W.Anderson,inSuperconductivityind-andf-bandMetals,ed.D.H.Douglass,AIP,NewYork(1972), p.17. 6O.V.Dolgov,D.A.KirzhnitsandE.G.Maksimov,Rev.Mod.Phys.53(1981)81. Amodern,butwaytooshorthistoryofthetheoryofsuperconductivityatahightemperature. 5 stable.Thisdevelopmentgavealsocredibilitytotheclaimthatsuchphonondrivensuperconductiv- itycanbequantitativelystudiedusingLDAtypebandstructuretheory. ComputingT ’sfromfirst c principlesis very demandingand has only become possible recently by the increase of computer power. MgB andanumberofothercaseshavelendcredibilitytothecasethatthesenumberscan 2 be trusted. This support the ’computational’case that 40K need not be the limit of conventional phonondrivensuperconductivity7. Despite these caveatsI am notaware ofanyexpertbelievingthatconventionalphononscan be responsibleforthesuperconductivityincupratesorironpnictides. Perhapsthebestreasonisthat the extreme electron-phonon coupling required to explain a T of 150K should somehow reveal c itself in band structure calculations, while these just indicate quite moderate couplings. This is not to say that phononsdo not play any role. In fact, there are striking experimentalindications that especially at very low dopingthe electron-phononcoupling is so strong that some variety of small polaron physics is at work. This appears to invoke polar couplings that are ignored in the LDAcalculationswhereimplicitlymetallicscreeningisassumed. Itcannotbeexcludedthateven atoptimaldopingsuch a ’poormetal’polaronicphysicsis atworkandclaimsthatbipolaronsare behindthesuperconductivitycannotbedismissedbeforehand.8 The main stream thinks however differently. Staying within the ”BCS paradigm” the quest to explainthe high-T ’s turnsintothe issue whetheronecan identifynon-phononstuffthatcan take c theroleofthephonons,mediatingverystrongattractiveinteractions.Theonlyoptiononthetableis tolookfor’glue’thatemergesfromtheinteractingelectronsystemitself. Therearequiteanumber of options here, like the excitonic mechanism first proposed by Little, emphasizing special units in the crystal structure characterized by a high electronic polarizability.9 Another old idea is the Kohn-Luttingeroverscreeningeffect in 2D systems10 and the closely related idea of the plasmon takingthe role of glue. However,by far the mostpopularview has beenup to the presentday to pointatspinfluctuations.ThebasicideadatesbacktoBerkandSchriefferinthe1960’swhoaimed atexplaningwhymetalsthatareclosetoatransitiontoanitinerantferromagneticstateusuallydo notsuperconduct11. Theyintroduceda particularapproximatewaytocomputemattersthatforms the conceptualcore of the myriadof improvedspin fluctuationtheoriesthathave been developed sincethen. A first hurdle is that in the electron system at microscopic distances there is no such thing as a system of spins that has a separate existence of the electrons – the electrons just carry around the spins themselves. The ’spin glue’ has thereforeitself to be a highlycooperativephenomenon emergingat’large’distancesin the interactingelectronsystem. Uponapproachinga transitionto amagneticstateinanitinerantmetalonedoeshoweverexpectprecursorsofthemagnetisminthe paramagneticstate. Theseareintheformofshortlivedmagneticfluctuationsthatbecomeincreas- ingly better defined upon approachingthe magnetic transition where they turn into real magnons –theparamagnons. Usingtimedependentmeanfieldtheory(RPA)theparamagnonscanbecom- puted perturbatively assuming weak interactions. Assuming in addition residual interactions be- tweenthesespinfluctuationsandtheconductionelectrons,theirpropagatorscanbejustfeededinto theMigdal-Eliasbergformalismtotaketheroleofphonons. Theparticularbenefitofthisapproachisthatitisquitepredictivewithregardtothepairingsym- metries.ForthecaseofferromagneticfluctuationsSchriefferandBerkshowedthatthesegenerically causerepulsionsinthes-wavechannelexplainingwhynearlyferromagneticmetalsusuallydonot 7See, e.g., Marvin Cohen’s talk at the 2009 KITP superconductivity conference: http://online.kitp.ucsb.edu/online/highertc09/cohen/ 8A.S.AlexandrovandN.F.Mott,PolaronsandBipolarons,WorldScientific,Singapore(1995). 9Thisidearevivedinthecontextofpnictides:seeG.A.Sawatzkyetal.,Eur.Phys.Lett.86(2009)174409. 10W.KohnandJ.M.Luttinger,Phys.Rev.Lett.15(1965)524. 11N.F.BerkandJ.R.Schrieffer,Phys. Rev. Lett. 17(1966)433;animportantfollowupisS.DoniachandS.Engelsberg, Phys.Rev.Lett.17(1966)750,focussingontherenormalizationoftheFermiliquid. 6 Superconductivity superconduct. In the early 1970’s superfluidity in fermionic 3He was discovered. It was already knownthatthenormalstateischaracterizedbyastronglyenhancedmagneticsusceptibilityindicat- ingthattheHeliumFermi-liquidisonthevergeofbecomingferromagnetic.Giventhatthereareno phononsaround, the idea that this superfluidityhad dealingswith paramagnonswas immediately realized. Dealingwithphonons,theinducedattractiveinteractionsurelyfavorsaspinsingletorder parameter,whileingoodmetalstheeffectivecouplingislocalizedinspaceandtherebymomentum independent, favoring s-wave superconductivity. However, ferromagneticfluctuationsfavor spin- tripletpairing. Thisinturnimpliesthatthepairsneedtobeinanorbitalstatewithunevenangular momentum,havingthe addedbenefitthatthe real space pair wave functionhas to have a nodeat theorigin. Thefermionsinthepairtherebyavoideachotherautomaticallysuchthatthehardcore microscopic repulsions are circumvented. The bottom line is that although there might be a net repulsionbetweenthequasiparticlesthismightwellturnouttosumupintoanattractiveinteraction inahigherangularmomentumchannel. The superfluidity in 3He is of the triplet variety and a discussion doing justice to the richness of this subject deserves a separate chapter in a book like this.12 With regard to the microscopic pairing mechanism a worthwhile lesson is the use of the Berk-Schrieffer logic by Anderson and BrinkmantogetwithgreateffectthedetailsoftheA-phaseright.13 Thegrossmessageisthat3He demonstratesthataninteractingfermionsystemasdominatedbymicroscopicrepulsiveinteractions caneventuallyturnintoasuperconductor,withthepairingmechanismhavingundoubtedlydealings with the spin fluctuations. However, the Anderson-Brinkmanresult is just showing that the spin fluctuationscontributeto the attractive interactions, stabilizing a particular superconductingstate, butitdoesnotprovethattheRPA style spinfluctuationsdoallthepairingworkexplaininginde- tailwhyT iswhatitis. Forinstance,muchattentionwaspaidaswelltotheroleoftheattractive c tail of the microscopic Van der Waals interaction. Another example of this lack of microscopic understandingis the questionwhetherthe mass enhancementsofthe 3He Fermiliquidcan be ac- countedforbythecouplingtotheferromagneticspinfluctuationsorwhethertheyarerelatedtothe proximity of the crystallization transition14. It appears that the quest to explain the nature of the residualattractiveforceresponsibleforthepairingin3Heendedinastalemate. Manyfactorscan contributeanditisjustanumbergame. Giventhattherearenocontrolledmathematicalmethods thattellushowtoconnectthestronglyinteractingmicroscopytotheweaklyinteractinginfrared,it isinfactplainlyimpossibletoaddressthissubtlequantitativematterwiththeavailabletheoretical technologies. Withaneyeontheremainder,oneshouldbeawarethat3Heisinawaya bestcase scenarioforglueideas,sincethenormalstateisawelldevelopedFermi-liquid,withasuperfluidity thatisclearlyintheweakcouplinglimit,whilecomplicationslike(lattice)Umklappandphonons areabsent. Thenotionthatthesuperconductivityofelectronsystemsinsolidscouldbecausedbyintrinsic electronicinteractionseffectsgotonthemainstageforthefirsttimebythediscoveryoftheheavy fermionsuperconductorsinthelate1970’sand1980’s.Thegiganticmassrenormalizationsinthese systemsleavenodoubtthatelectron-electroninteractioneffectsaredominant.Inadditionitbecame gradually clear that the superconducting order parameters in these system are unconventionalby default while magnetism is usually around the corner. During the last twenty years a consensus emergedthatinthesesystemsthesuperconductivityiscloselyrelatedtothepresenceofaquantum phasetransitionwhereatzerotemperatureamagneticallyorderedstategetsdestroyedbyquantum fluctuations. Insection2.4.6Iwilltakeupthisthemeindetail,includingadiscussionoftheHertz typetheoriesthatarecloselyrelatedtothespinfluctuations. 12Theclassic,hardtoimprovetreatiseisA.J.Leggett,Rev.Mod.Phys.47(1995)331. 13P.W.AndersonandW.F.Brinkman,Phys.Rev.Lett.30(1973)1108. 14D.Vollhardt,Rev. Mod. Phys. 56(1984)99;thisinturnrestedonthediscoverybyMauriceRiceandBillBrinkmanof thedescriptionoftheMotttransitionintermsofGutzwillerprojectedwavefunctions:W.F.BrinkmanandT.M.Rice,Phys. Rev.B2(1970)4302.TheseideaswereinturninspirationalinthedevelopmentoftheRVBtypetheoriesofsection2.4.4. Amodern,butwaytooshorthistoryofthetheoryofsuperconductivityatahightemperature. 7 The spin fluctuation idea acquired directly prominency in the early history of cuprate high-T c superconductivity. Obviouslyan agentdifferentfromphononscausingthe superconductivitywas atwork,whileinthefrenzyofthelate1980’sitbecameinnotimeclearthatcupratesuperconduc- tivityemergedfromdopinganantiferromagnet. Inaddition,bothinelasticneutronscattering-and NMR experimentsgave away that the metals at superconductingdoping concentrationsshow the signaturesof pronouncedspin fluctuations. The spin-fluctuation-glueidea was lying on the shelf anditwaspredictablytakenupbypartofthecommunity. ThemostvisibleadvocateswereDou- glasScalapinoandDavidPines,representingsomewhatcomplementaryviewsthatarein2010still discernible.Pinesembodiedthe’bottom-up’phenomenologicalapproach,analyzingingreatdetail experimentalinformationonthespinfluctuations,aimingforaquantitativephenomenologicalde- scription of the superconductivity. Scalapino approachedit from a more theoretical angle, trying to cast the (incomplete) information from quantum Monte Carlo computations and so forth in a diagrammaticframeworktayloredaccordingto the Berk-Schriefferframeworkand improvements thereofflikethe”fluctuation-exchange”(FLEX)perturbativetheory. Infact,thespinfluctuationadvocatesfacedaheavybattleintheearlyhistoryofhigh-T super- c conductivity. The predictivepower with regardto the pairing symmetrywas their stronghold. In the presence of Umklapp associated with the scattering of the electrons againstthe crystal lattice the pairingsymmetrystorybecomesmore interestingandspecific. The lattice warpsthe electron dispersionsandtheFermi-surfaceshapewiththeramificationthatonthelevelofRPAcalculations the spin fluctuations tend to get quite structured in momentum space. The cuprate fermi-surface computedbyLDAbandstructure(andconfirmedbyphotoemissionmeasurements)shows’nesting’ featuresthatpromotesastrongenhancementofspinfluctuationsatlargemomenta.Thisinturnim- pliesastrongmomentumdependenceofthepairinginteractionsmediatedbythesefluctuationsthat stronglyfavorad superconductingorderparameter.Inthe1987-1990eraitwashowevertaken x2−y2 forgrantedbythecommunityatlargethatcupratesuperconductivitywass-wave. Inhinsightthese strongbelievesarenoteasy to explainonrationalgrounds: the datawere ofa poorqualityandit appearstoberootedinsteadinaconservativesociologicalreflex–evend-wavewastoorisky.After ahardfight,thatlastedforacoupleofyearsScalapinoandPinesetal. wereprovenrightwiththeir d-waveclaim bythe phasesensitivemeasurementsofTsui, Kirtley,vanHarlingenandcoworkers discussedelsewhereinthisbook. ThisisremarkableandIfinditrespectablethatthissuccesshas motivateda stubbornattitude in the spinfluctuationcommunity: there hasto be somethingto the notiongiventhatitgotthed-waveright. However,afteritbecameclearthatthepairingsymmetry isd-waveinnotimeitturnedoutthatthisisalsothenaturalgroundstateinforinstancetheRVB context (section 2.4.4), and even invoking the spontaneouscurrent phases of Varma as discussed insection2.4.5. Althoughd-wavesuperconductivityisratherunnaturaldealingwithconventional phonons,theconsensushasemergedthatitisjusttoogenericdealingwithelectronicmechanisms forittoberegardedasconclusiveevidenceforanyspecificmechanism. Atthismomentintime the ideaofa spin-fluctuation’superglue’isverymuchalive– counting headsinthecommunityitmightbethemostinfluentialhypothesisalltogether.Partofthisappealis surelyassociatedwiththefactthatitposesacrispchallengetotheexperimentalists: thesuperglue islikeaholygrail,andonejusthastodigwiththemodernexperimentalmachineryofcondensed matterphysicsintotheelectronmattertoisolateit. Factofthematterishoweverthataftertwenty yearsof concertedeffortin thecupratesthe ’evidence’is asindirectas ithasbeenallalong. The stateoftheartisperhapsbestrepresentedbyatourdeforcethatwaspublishedin2009.15 Empir- icalinformationobtainedfromdirectmeasurementsof the spin fluctuationsbyneutronscattering andtheelectronself energyeffectspickedupin photoemissioniscombinedwiththeresultsfrom numericalcalculations,arrivingattheclaimthatthespinfluctuationgluecanfullyaccountforthe superconductivity.Butthisworkalsorevealstheweaknessesofthis’paradigm’.Theevidencesare 15T.Dahmetal.,NaturePhysics5(2009)217. 8 Superconductivity highlyquantitative,whileitdepartsfromoversimplifiedtoymodelswhereoneneverthelesslooses mathematicalcontrolbecauseoftheintermediatestrengthofthecouplingsinvolved. Equallycon- vincingcaseshaveappearedarguingthattheselfenergykinksinphotoemissionareentirelydueto ratherweakelectron-phononcoupling.Themagneticresonancethatdominatesthespinfluctuation spectrum in optimally doped cuprates can be interpreted rather as an effect than as the cause of thesuperconductivity. Perhapsthemostdevastatingcounterevidenceavailableatpresentforsuch aninterpretationisthestaggering’electroniccomplexity’oftheunderdopedpseudogapregimeas discussedinsection2.4.5.Thesimplefermiology’RPAview’onthiscompetingorderappearsasa grossoversimplificationofthisreality. The adventof the pnictide superconductorsappeared initially as a shot in the arm for the spin fluctuation idea. The LDA calculations indicate nesting properties of the Fermi-surface pockets thatarebeneficialfortheRPAparamagnons,whiletheyareconsistentwiththe”(p ,0)”magnetism foundintheparentcompounds.Thecasewasmadevigorouslythatsincetheparentsystemsarenot Mottinsulatingithastobethattheironsuperconductorsarequiteitinerantandtypicalcandidates forspinfluctuationsuperconductivity.Lastbutnotleast,apeculiar”s±”pairingsymmetrywaspre- dictedbasedontheLDA-RPA-Eliashberg”trinity”.16 Itisnowwidelybelievedthatthissymmetry isindeedrealizedinatleastsomeofthesesystems,althoughdefinitiveevidenceisstillmissing. Muchoftheearlyexperimentalactivityinthisyoungfieldwasaimedattryingtofindevidencefor thespin-fluctuationsuperglue.Remarkably,whenthemomentseemedfittocryvictorythebuilding startedtocrumble. Dataappearedindicatingthatunexpected’nematic’phenomenaarehappening in the underdopedpnictides that seem to parallel the pseudogapcomplexities of the cuprates. In addition, the pnictide metals at optimal doping have weird properties like a linear resistivity that againsuggeststhatthesehaveuncannycommonalitieswiththemysterystuffofthecuprates. This is a very fresh development and it is much too early to say where it will land. However, I do expectthatitisbeneficialforthenewcomersinthepnictidefieldtobeawareoftheverylongand contentioushistoryofthespinfluctuationglueidea. 1.3 The legacyofPhilipW.Anderson (I):Mottness. WheninsomefuturetherealhistoryofT canbewritten,theoddsarehighthatPhilipAnderson c willhavealegendaryroletoplay.TheinfluentialAndersoniantraditioninhigh-T superconductiv- c ityrevolvesaroundtheassertionthatthegroundrulesofmetalphysicsaslaiddowninthe1950’sdo notapplytotheelectronsincuprates,forthereasonthathigh-T superconductivityemergesfrom c aMottinsulatorthatisdoped. Thisisinstarkcontrastwiththespin-fluctuationschoolofthought asdiscussed in the previoussection. Thisdepartsfromthe notionthatonecan getaway with the ideathatthenormalstateisinessenceaFermi-liquid. Althoughshakenhardbythepowerfulglue bosons,thequasiparticlesformingaFermi-gasareeventuallyrulingthewavesanditisjustamatter ofsortingouttheFeynmangraphsgoverningtheperturbationsofthequasiparticlegas.Whenhigh- T was discoveredin 1986thiswastakenasa self-evidenttruismbynearlyeverybody. However, c Andersonpublishedoneofthefirst paperseveronthe subjectandthisis still themostinfluential paperinthehistoryofthesubjectintroducingthe’resonatingvalencebond’(RVB)idea17 thatwill be the focuspoint of the nextsection. This section is a preliminary: the success story of nailing downthespecificsoftheparentcuprateMottinsulatorinthelate1980’s. 16Thisappearstohavebeenacloserace,withwinners: I.I.Mazinetal.,Phys. Rev. Lett. 101(2008)057003,beingcited already500times. 17P.W.Anderson,Science235(1987)1196.