GossamerSuperconductivity
R.B.Laughlin
DepartmentofPhysics,StanfordUniversity,Stanford,CA94305
(Dated:September1,2002)
AnnewsuperconductinghamiltonianisintroducedforwhichtheexactgroundstateistheAn-
dersonresonatingvalencebond.Itdiffersfromthet-Jandhubbardhamiltoniansinpossessingapowerfulattractiveforce.Itssuperconductingstateischaracterizedbyafullandintactd-wavetunnelinggap,quasiparticlephotoemissionintensitiesthatarestronglysuppressed,asuppressedsuperfluiddensity,andanincipientMott-Hubbardgap.[PublishedasPhil.Mag.86,1165(2006).]PACSnumbers:74.20.-z,74.20.Mn,74.72.-h,71.10.Fd,71.10.Pm
IthasbeenknownsincetheearlyworkofUemura
1
that
thesuperfluiddensityandtransitiontemperatureofun-derdopedcupratesuperconductorsbothvanishmore-or-lesslinearlywithdopingandareproportional.Thecon-stantofproportionalityisconsistentwiththetransitionbeinganorder-parameterphaseinstabilityanalogoustotheKosterlitz-Thoulesstransitionin2dimensions
3
.This
ideaissupportedbynumerousotherexperiments,includ-ingtheopticalsumrulestudiesofUchida
2
,thegiant
proximityeffectreportedbyDeccaetal
4
,andtherecent
heat-transportmeasurementsofWangetal
5
showing
superconductingvortex-likeeffectsabovethetransitiontemperature.Thetransporttrendscontinueintothein-sulator,whereAndo
6
reportsthathigh-temperaturehall
effectconsistentwithacarrierdensityroughlypropor-tionaltodoping.Atthesametime,however,thed-wavegapinthequasiparticlespectrumgrowsmonotonicallyasthedopingdecreasesandsaturatesatavalueofabout0.3eV
7–10
.Thishasledtospeculationsthatthetunnel-
ingpseudogapistheenergytomakeapre-formedCooperpair,whichthencondensesintoasuperfluidatalowertemperature.Thereisnodirectevidencethatthed-wavenodalstructuresurvivesintotheinsulator,butthereiscircumstantialevidenceforthis,notablytheobservationinLa2xSrxCuO4byYoshidaetal
11
ofdispersingquasi-
particlebandsnearthed-wavenodethatbecomefainterasdopingisreducedbutdonotshiftorchangetheirvelocityscale.ThesebandsarealsodetachedfromthelowerHubbardband,andsimplymaterializeinmid-gapasthedopingisincreasedfromzero.
Allofthisbehaviorisconsistentwiththeideathat
thesuperconductivitypersistsdeepintothe“insulating”state,coexistswithantiferromagnetismthere,andfailstoconductonlybecauseitslong-rangeorderisdisrupted,presumablyonaccountofitslowsuperfluiddensity
12
.
Therearemanywaysthelattercouldoccur,includingimpuritylocalizationorcrystallizationoftheorderpa-rameter,forsuchagossamersuperconductorisphysically
equivalenttoadilutegasofbosonsandthushighlyun-stable.
Theideathatthe“insulator”mightactuallybeathin,
ghostlysuperconductorisimplicitinthemathematicsoftheAndersonresonatingvalencebond(RVB)
13
worked
outbyvariousauthorsinthelate1980s
14–16
andfur-
therextendedrecentlybyParamehanti,Randeria,and
Trivedi
17
.Unfortunately,thisideahasalwaysrunafoul
ofabasicpremiseofRVBtheorythatsuperconductiv-ityshouldbeauniversalaspectofquantumantiferro-magnetism.Thispremiseisbothconfusingandfun-damentallyincorrect,astheconventionalspindensitywavegroundstate,whichcontainsnosuperconductivity,isaperfectlygoodprototypeforaquantumantiferro-magnet.Therealissueisnotwhetherallantiferromag-netsaresuperconductorsbutwhethersomeofthemare-i.e.whetherthereexistsasecondkindofantiferro-magnetismdistinguishedfromthefirstbyatinyback-groundsuperfluiddensity.Onewouldalsoliketoknowwhichhamiltoniansfavorthissecondkindofstateoverthefirst.Itisnotjustthehamiltoniansthatstabilizean-tiferromagnetism,forthesesimplyemphasizetheaspectsofthevacuathatarethesameandde-emphasizetheas-pectsthataredifferent.IthasbeenknownsincetheearlyworkofHsu
18
,forexample,thatthesetwokinds
ofvacuumhavealmostidenticalvariationalenergiesinthecontextofthet-Jhamiltonian.Thisisconsistentwiththerecentnumerical-variationalstudiesofSorellaetal
19
,whoreportthatthet-Jmodelsuperconductsinare-
gionofitsparameterspace,eventhoughpreviousnumer-icalworkonthesamemodelreportedantiferromagneticstripeordering
20
.Beccaetal
21
havearguedthatthe
latterisanartifactlatticeanisotropy.Howevertheim-portantpointitthatsensitivitytoalgorithmicdetailandtheinherentdifficultyofdeterminingtheorder,reectedinlackofagreementamongdifferentgroups,demonstratethatmodelsofthiskindarehighlyconflictedandclosetoaquantumphasetransition
22
.Inotherwords,byex-
aggeratingthemagnetismthesemodelsconfusetheissueratherthanclarifyingit.Thereisnopersuasiveevidenceforsuperconductivityinthehubbardmodel
23–26
.
Thepurposeofthisletteristoproposeanewstrategyforresolvingthecupratedilemma.Ratherthanstruggletodiagonalizeaconflictedhamiltonian,weshallmodifytheequationsofmotiontostabilizethegossamersuper-conductor.Theeasiestwaytodothisisbypostulatingapowerfulattractiveforcebetweenelectrons,justasonewouldinanyothersuperconductor.Thissolutionisnotunique,butitisexperimentallyfalsifiablethroughtheex-citationspectrumofthestate,whichismodel-dependent.Insofarasthesepropertiesmatchexperiment,whichhasyettobeseen,itwouldsuggestthatcoulombinteractions
2
arenotsucienttoexplaincupratesuperconductivity.
Weconsideraplanarsquarelatticeofsitesjonwhich
electronsmaysit.Thesuperconductingvacuumwewishtostabilizeisj>=j>,where
j>=
NY
k
(uk+vkcy
k"cy
k#)j0>;(1)
withck=N1=2PN
jexp(ikrj)cjasusual,and
=
NY
j
z
(nj"+nj#)=2
0(10nj"nj#):(2)
TheBardeen-Cooper-Schrieffer(BCS)pairingampli-tudessatisfy
kk
kk
uk
vk
=Ek
uk
vk
(3)
andarenormalizedbyu
2k+v
2
k=1.Theyarerelatedto
theholedopingby
1
N
X
k
v
2
k=1
1
N
X
k
u
2k=
1
2
:(4)
Theimportantpositiveeigenvalue
Ek=
q
(k)2+2
k(5)
istheenergytomakeaquasiparticle(eitheranelectronorahole)whentheparameter0iszero.Theparameter
z0isafugacityrequiredtokeeptheelectrondensitythe
sameasonevaries0.Assumingthesuperconducting
orderparametertobed-wave,sothereisnoon-sitepair-ingamplitude,thechargestatesofasitearestatisticallyindependentandcharacterizedbyafugacityz.Thecon-ditionthatthetotalchargeonthesitebe1,whereisthedoping,is[2z+2(1)z
2
]=[1+2z+(1)z
2
]=1,
where1=(10)
2
,or
z=
p
1(12)
(1)(1+)
=(
1
1+
)z0:(6)Theparameterz0isthefactorbywhichzexceeds(1)=(1+),itsvaluefor0=0.
ThehamiltonianisconstructedusingtheBCSannihi-
lationoperators
bk"=ukck"+vkcy
k#bk#=ukck#vkcy
k":(7)
Solongas06=1,thepartialprojectorhasaninverse
1
=
NY
j
z
(nj"+nj#)=2
0(1+0nj"nj#);(8)
where0=0=(10).Thisenablesustoconstruct
themodifiedannihilationoperators
~b
k"=bk"1
=
1
p
N
NX
j
e
ikrj
z
1=2
0uk(1+0nj#)cj"+z
1=2
0vk(10nj")cy
j#
;(9)
andlikewisefor~b
k#,forwhich~b
kj>=0.Thusj>is
aneigenstateofthehamiltonian
H=
X
k
Ek~by
k
~b
k(10)
witheigenvalue0.However,thishamiltonianhasonlynon-negativeeigenvalues,sinceforanywavefunctionj>
<|Hj>=
X
k
Ek<~b
kj~b
k>0:(11)
Thusj>isagroundstateofH.However,itisalsothe
groundstatebyvirtueofadiabaticcontinuity.ThestateinquestionmaybecontinuouslydeformedintoaBCSstatebytaking0slowlytozero.Sinceitdoesnotcross
aphaseboundaryintheprocess,thegroundstateandlow-lyingexcitationsmusttrackinaone-to-oneway.
Letusnowconsiderthequasiparticleexcitationsof
thissuperconductor.Theoperators~b
knolongeran-
ticommuteproperlywiththeirhermitianadjointsandthuscannotbeusedtocreatequasiparticles.Thephys-icalmeaningofthisisthatthequasiparticlesinter-act.Insteadwewillusethevariationalwavefunctionsjk>=by
kj>borrowedfromtheRVBliterature.
Theexpectedenergyis
<k|Hjk>
<kjk>
=Ek
<j>
<kjk>'Ek:(12)
Thelaststeprequiresevaluatingtherelevantnorms,whichcanbedonebyhandonlyapproximately
17,18
.We
repeattheargumentshereforcompleteness.From
by
k"j>=
1
u2
k
cy
k"j>=
1
v2
k
ck#j>(13)
wefindthat
1
N
NX
k
u
2k
<jbk"
2by
k"j>
<j2
j>
=
<jcj"
2cy
j"j>
<j2
j>
3
=z0
1+(1)z
1+2z+(1)z2
=
1+
2
(14)
and
1
N
NX
k
v
2
k
<jbk"
2by
k"j>
<j2
j>
=
<jcy
j"
2cj"j>
<j2
j>
=
1
z0
z+z
2
1+2z+(1)z2
=
1
2
:(15)
Forj6=j0weassumethattheamplitudeforagivenconfigurationisweightedbythesquarerootofitscorre-spondingprobability,andthattheseweightsaddequally.Wethenhave
<jcj"
2cy
j0"j>
<j2
j>
<j>
<jcj"cy
j0"j>
'
4
12(zz0)
1+(1)z
1+2z+(1)z2
2
=1(16)
and
<jcy
j"
2cj0"j>
<j2
j>
<j>
<jcy
j"cj0"j>
'
4
12(
z
z0
)
1+z
1+2z+(1)z2
2
=1:(17)
Letusnowconsiderthelow-energyspectroscopicprop-
ertiesofthismodel.Reasoningasabove,wendthematrixelementforphotoemissiontobe
<k#jck"j>p
<k#jk#><j>
=gvk;(18)
where
g
2
'
20
1
0
1p
1(12)
12
]}
:(19)
Inversephotoemissionisthesameexceptwithuksub-
stitutedforvk.Thesuppressionofthephotoemissionintensityismatchedbyasimilarsuppressionofthesu-perfluidorderparameter:
<jcj"cj0#j>
<j>'g
2<jcj"cj0#j>
<j>
:(20)
Weneednotconsiderthecaseofj=j0forad-wavesuperconductor.Thusunderstrongprojectionnearhalf-fillingthismodelexhibitsthepseudogapphenomenon:
ThequasiparticleenergiesremainattheirunperturbedvaluesEkas0increasesfrom0to1,butthesuperfluiddensitydecreasesfrom1to2jj=(1+jj).
LetusnowconsidertheformationoftheMott-
Hubbardgap.Photoemissionofaquasiparticleaccountsforonlyasmallfractionofthesumrule
<jcy
kckj>
<j>'g
2
v
2
k+(1g
2
)
1
2
:(21)
Therestmustoccuratahigherenergyscale,thevalueofwhichmaybeestimatedbycomputingtheexpectedenergyofahole.Fromtheanticommutators
f~b
k";cj"g=
1
p
N
e
ikrj
z
1=2
00vkcj"cy
j#
(22)
f~b
k#;cj"g=
1
p
N
e
ikrj
z1=2
00ukcj"cj#
+z
1=2
0vk(10nj#)
(23)
weobtain
<jcy
kHckj>=z0
10
1
2
2
v
2
kEk
+
1
N
NX
q
Ek+q
z0
2
0Aqv
2
k+q+
2
0
0
Bqu
2k+q
;(24)
where
Aq=
NX
j
<jS0Sjj>
<j>
+
14
<jn0njj>
<j>
(1)
2
4
e
iqrj(25)
Bq=
NX
j
<jcy
0"cy
0#cj#cj"j>
<j>
e
iqrj:(26)
Proceedingsimilarlywiththestatecy
k"j>,wefindthat
theenergytoinjectanelectronistheexactparticle-holeconjugateofthisexpression,producedfromitbyinter-changingukandvk,negating,andsubstitutingof1nj
fornj.Thusathalf-fillingtheelectronspectralfunction
issymmetric.Thedensityandsuperfluidcorrelationsbecomesuppressedathalf-fillinginthe0!1limit,
whilethemagneticcorrelationsbecomeenhanced.Ap-proximatingthelatterbythecorrelationfunctionoftheNeelvacuum,weobtainat=0
4
lim
0!1
<jcy
kHckj>
<jcy
kckj>
=lim
0!1
<jckHcy
kj>
<jckcy
k>
'
1
10
1
N
NX
q
Eq+
1
2
Ek
:(27)
ThusthespectralfunctionconsistsofMott-Hubbard“lobes”athighenergieswithafaintbandofstatesatmid-gapassociatedwiththegossamerquasiparticles.Thechemicalpotentialdoesnotjumpinthismodelwhenistunedfromnegativetopositive,asitdoesinthehub-bardmodel
26
.
Letusnowconsiderantiferromagnetism.Athalf-lling
thelargestterminEq.(10)isanon-sitecoulombre-pulsionofmagnitude2
PN
kEk=N(10).Ifthisre-
pulsionismadeslightlylarger,thesystembecomesun-stabletospindensitywaveformationontopofthesuperconductivityatthenestingwavevectorofthed-wavenodes.Thecaseofhalf-llinghasbeenworkedoutindetailbyHsu
18
andweonlyquotetheresult
here.ThequasiparticledispersionrelationEq.(5)be-comesmodifiedtoEk=
p
(k)2+2
k+2
0,where
theenergygap0isrelatedtothesitemagnetization
<jS
z
jj>=<j>=mby
m'
m0
10(14m2
0)
m0=0=
2
N
NX
k
Ek:(28)
ItwasobservedbyHsu
18
thatstrongprojectionenhances
themagnetizationbyroughlyafactorof2overitsun-projectedvalue,enablingtheHeisenbergmodelvalueofm=0:307tobeachievedwitharathermodestvalueof0,about1/3thezoneaverageofEk.Incontrastto
thecaseworkedoutbyhim,however,thegapherehasphysicalmeaningandcanbedetectedinatunnelingorphotoemissionexperiment.Thissmallquasiparticlegapisakeycharacteristicofthegossamersuperconductor.
PhasefluctuationshavebeenleftoutofEq.(10)on
thegroundsthattheyareirrelevanttothefermispec-trum,whichischaracterizedbyanenergyscalemuchhigherthanthesuperconductingTc.However,theyare
essentialforaccountingforboththeUemuraplotandstrange-metaltransportaboveTc.Thetransportofadilutegasofbosonsformedwhentheorderparameterde-phaseswouldnotexhibitanytraditionalmetallicbehav-iorandindeedwouldtendto“shortout"conductionbythefermions.Lattice-mediatedcrystallizationofagos-samersuperconductorwouldalsoprovideareadyexpla-nationforwhythestaticstripes
27
areobservedat=1=8
issomecupratesandnotothers,whythestripecommen-surationwavevectorissopeculiar,whystripescanbede-stroyedbymoderatepressure
28
,andwhythematerials
insulatewhensubjectedtostrongmagneticelds
29
.
IwishtoexpressparticularthankstoZ.-X.Shenand
myotherNEDOcollaboratorsH.Eisaki,A.Fujimori,S.Maekawa,N.Nagaosa,N.P.Ong,Y.Tokura,andS.Uchidafortheircollegialityandinnumerablehelpfuldiscussions.ThisworkwassupportedbytheDepartmentofEnergyundercontractNo.DE-AC03-76SF00515.
R.B.Laughlin:http://large.stanford.edu
1
Y.J.Uemura,Phys.Rev.Lett.62,2317(1989).
2
S.Uchidaetal.,Phys.Rev.B43,7942(1991).
3
V.J.EmeryandS.A.Kivelson,Nature374,434(1995).
4
R.S.Deccaetal.,Phys.Rev.Lett.85,3708(2000).
5
Y.Wangetal.,Phys.Rev.Lett.88,257003(2002).
6
Y.AndoandK.Segawa,Phys.Rev.Lett.88,167005
(2002).
7
D.S.Marshaletal.,Phys.Rev.Lett.76,4841(1996).
8
H.Dingetal.,Nature382,51(1996).
9
R.B.Laughlin,Phys.Rev.Lett.79,1726(1997).
10
T.TimuskandB.Statt,Rep.Prog.Phys.62,61(1999).
11
T.Yoshida,X.J.Zhou,T.Sasagawa,W.L.Yang,P.V.Bogdanov,A.Lanzara,Z.Hussain,T.Mizokawa,AFujimori,H.Eisaki,Z.-X.Shen,T.Kakeshita,andS.Uchida,“NodalMetallicBehaviorofLightly-DopedLa2xSRxCuO4”,cond-mat/0206469(SubmittedtoPhys.Rev.Lett.).
12
L.Merchantetal.,Phys.Rev.B63,134508(2001).
13
P.W.Anderson,Science235,1196(1987);G.Baskaran,
Z.Zou,andP.W.Anderson,SolidStateComm.63,973
(1987).
14
C.Gros,R.Joynt,andT.M.Rice,Phys.Rev.B36,381(1987);C.Gros,Phys.Rev.B38,931(1988).
15
H.YokoyamaandH.Shiba,J.Phys.Soc.Jpn.57,2482
(1988).
16
F.C.ZhangandT.M.Rice,Phys.Rev.B37,3759(1988).
17
A.Paramekanti,M.Randeria,andN.Trivedi,Phys.Rev.Lett.87,217002(2001).
18
T.C.Hsu,Phys.Rev.B41,11379(1990).
19
S.Sorellaetal.,Phys.Rev.Lett.88,117002(2002).
20
S.R.WhiteandD.J.Scalapino,Phys.Rev.Lett.80,1272
(1998).
21
F.Becca,L.Capriotti,andS.Sorella,Phys.Rev.Lett.87,167005(2001).
22
S.R.WhiteandD.J.Scalapino,Phys.Rev.Lett.84,3021(2000).
23
G.Su,Phys.Rev.Lett.86,3690(2001).
24
R.Preuss,W.Hanke,C.Gr¨ober,andH.G.Evertz,Phys.Rev.Lett.79,1122(1997).
25
F.F.Assaad,M.Imada,andD.J.Scalapino,Phys.Rev.Lett.77,4592(1996).
26
N.Bulut,D.J.Scalapino,andS.R.White,Phys.Rev.Lett.73,748(1994);ibid.72,705(1994).
27
J.M.Tranquada,et.al.,Nature375,561(1995).
28
S.Arumuganetal.,Phys.Rev.Lett.88,247001(2002).
29
S.Onoetal.,Phys.Rev.Lett.85,638(2000);G.S.Boe-
bingeretal.,ibid.77,5417(1996).