Table Of ContentSU(2)symmetryinarealisticspin-fermionmodelforcupratesuperconductors
T. Kloss1,2, X. Montiel1,2, C. Pépin1
1IPhT, L’Orme des Merisiers, CEA-Saclay, 91191 Gif-sur-Yvette, France and
2IIP,UniversidadeFederaldoRioGrandedoNorte,Av.OdilonGomesdeLima1722,59078-400Natal,Brazil
(Dated:May22,2015)
WeconsiderthePseudo-Gap(PG)stateofhigh-Tc superconductorsinformofacompositeorderparameter
fluctuating between 2p -charge ordering and superconducting (SC) pairing. In the limit of linear dispersion
F
and at the hotspots, both order parameters are related by a SU(2) symmetry and the eight-hotspot model of
Efetov et al. [Nat. Phys. 9, 442 (2013)] is recovered. In the general case however, curvature terms of the
5
dispersionwillbreakthissymmetryandthedegeneracybetweenbothstatesislifted.Takingthefullmomentum
1
dependenceoftheorderparameterintoaccount,wemeasurethestrengthofthisSU(2)symmetrybreakingover
0
thefullBrillouinzone. ForrealisticdispersionrelationsincludingcurvaturewefindgenericallythattheSU(2)
2
symmetry breaking is small and robust to the fermiology and that the symmetric situation is restored in the
y large paramagnon mass and coupling limit. Comparing the level splitting for different materials we propose
a ascenariothatcouldaccountforthecompetitionbetweenthePGandtheSCstatesinthephasediagramof
M
high-Tcsuperconductors.
2 PACSnumbers:74.40.Kb74.20.-z74.25.Dw74.72.Kf
2
] I. INTRODUCTION attempttosuchaunificationhasbeentheSO(5)theorywhich
n
relatesthed-waveSCstatetotheAForder30.Notlessfamous
o
c istheSU(2)symmetrywhichrelatestheSCd-waveorderto
Reflecting our rather poor understanding of the physics
- the π-flux phase of orbital currents1. In both cases, the key
r of cuprate superconductors, two kinds of theories are still
p question was to argue that the energy splitting between the
debating whether the final solution for this problem will
u two orders was small enough so that thermal effects would
be a “bottom-up” approach based on a strong coupling
s restore the symmetry above T and below T∗, which are the
. theory1–3 or rather a “top-down” approach, where symme- c
t SCandthePGcriticaltemperatures.Thesamequestionholds
a triesandproximity toaQuantumCritical Point(QCP)plays
here for the SU(2) symmetry relating d-wave and QDW or-
m a dominant role4–7. The recently proposed Eight Hot Spots
der. Although the EHS model in its linearized version ver-
- (EHS)modelisapromising“top-down”approachtocuprate
d superconductors8,9. ItreducestheFermisurfacetoonlyeight ifies the symmetry exactly, it is not clear if a more realistic
n Fermisurface,includingcurvatureandthewholebandstruc-
pointsontheanti-ferromagnetic(AF)zoneboundaryandtak-
o ture,willinducesmallorlargeenergysplitting.Moreover,the
ing long-range AF fluctuations between them into account.
c EHSmodelreliesonlong-rangeAFfluctuationswhichmedi-
[ When the dispersion is linearized at the hot spots, one ob-
ate the interactions, but the experimental observation points
serves surprisingly that an SU(2) symmetry relating the d-
2 outtoshort-rangeAFcorrelationswhichpotentiallywillgap
wave SC channel (Cooper pairing) to the d-wave bond or-
v a whole part of the Fermi surface, as depicted in Fig. 1. A
der, or Quadrupolar Density Wave (QDW), (charge channel)
4 theoryfor“hotregions”insteadof“hotspots”isthusneeded.
2 ispresent. Moreover,animposantpre-emptiveinstability(of
3 order of 0.6J, where J is the AF energy scale) in the form
Inthispaperweaddresscarefullyalltheseissuebyevaluat-
5 ofacompositeSU(2)orderparameteremerges,thathasbeen
ingtheSU(2)splittingonrealisticFermisurfaces,forthetwo
0 identifiedasagoodcandidateforthepseudo-gap(PG)stateof
distinctcomponentsofthecompositeorderparameter: thed-
1. thosecompounds9–12. Motivatedbyanimpressivesetofnew
waveχ-fieldinthechargesector(whichformstheQDWorder
0 experimentalresults13–22,thistheorypointsouttotheemerg-
intheEHSmodel)andthed-wave∆-fielddescribingtheSC
5 ing idea that charge order is most certainly a key player in
pairingsector. WefindthatthesplittingoftheSU(2)symme-
1
thephysicsofcupratesuperconductors,inadditiontoAFor-
: tryincreaseswiththemassoftheparamagnons,butdecreases
v der, d-wave SC state and the Mott insulator phase. Angle-
withthestrengthofthecouplingconstantbetweenAFfluctu-
i resolved photoemission spectroscopy (ARPES) experiments
X ationsandconductionelectrons. Thisopensawideregimeof
confirm as well the presence of modulations in the SC state
parameters where the splitting is minimal – of the order of a
ar forunderdopedBi220123,24andhasbeeninterpretedeitherin fewpercents–andwheretheSU(2)symmetryisexpectedto
termsofchargeorderorPairingDensityWave(PDW)inside
berecoveredthroughthermaleffectsinaregimeoftempera-
thePGphase25–28. Finally, weliketomentiontherecentin- tures T <T <T∗. Above the PG temperature T∗ all traces
c
terpretationofRamanresonancesbyacollectiveSU(2)mode
of the short-range charge and SC field have disappeared. Of
whichisthefirstexperimentalresultthatsupportstheideaof
course the SU(2) symmetry holds in the whole temperature
acompositePG29. range between T and T∗, hence subleading charge instabil-
c
TheideaofanemergingSU(2)symmetrybelongstoawide ities which occur below T∗ do have their SU(2) partners in
class of theories which explain the PG phase of the cuprates theformofPairingDensityWaves(PDW)31,32. Wealsostudy
throughthenotionofdegeneratesymmetrystatesbetweenthe the effects of the Fermi surface shape in breaking the SU(2)
d-waveSCorderandanotherpartner.Maybethemostfamous symmetry–whichisonlypreservedintheEHSmodelwitha
2
300
(cid:114) YBCO
(cid:54) YBCO
(cid:114) Bi2201
(cid:54) Bi2201
200 (cid:114) Bi2212
(cid:54)| (cid:54)(cid:114) HBig21221021
(cid:114)|, | (cid:54)(cid:114) eHl.g d1o2p0e1d
|100 (cid:54) el. doped
0
0.001 0.01 0.05 0.1
a) m
1
FIG.1. (Coloronline)SchematicFermisurfaceofhole-dopedsu-
YBCO
perconductorsinthefirstBrillouinzoneofasquarelattice.Theorder 0.8 Bi2201
parametersspatiallyextentoverhotregions,thatarecenteredaround (cid:54)| Bi2212
Hg1201
cthoeuhpolitnsgpoQtstpooospitpioonssedanrdegwiohnics.harecoupledbythe2pFandtheAFM (cid:54) | / |0.6 el. dope
(cid:60)
(cid:114) 0.4
|
0.2
0
0.001 0.01 0.05 0.1
b) m
FIG.3. (Coloronline)Panela): Variationofthemaximumvalue
ofthegapfunctions|χ|and|∆|asafunctionofthemassm. Note
thatinallmaterialsthe2pFpairingintermsof|χ|vanishesabruptly,
whereastheSCpairingintermsof|∆|approacheszeroasymptoti-
callywhentheparamagnonmassmisincreased.Panelb):Variation
of the level splitting as a function of the mass m. In both panels
FIG. 2. Schematic phase diagram of hole-doped cuprate super-
λ =44.
conductors. Theeffectivemassoftheparamagnonpropagatorm
eff
serves as as measure for the distance to the quantum critical point
QCP.IntheSCphasem ispositiveandvanishesattheQCPtobe-
eff
comenegativeintheAFMphase.Thebaremassm isdefinedfar Thefermionicfieldψ describestheelectronswhicharecou-
bare
away from criticality at some higher temperature, indicated by the pled via Lφ to spin waves described by the bosonic field φ.
dashedline. Theeffectivespin-wavepropagatorisD−1=γ|ω|+|q|2+m
q
wheremistheparamagnonmasswhichvanishesattheQCP
andγ aphenomenologicalcouplingconstant, whichweesti-
linearizeddispersion. Wefindthatthesplittingissmallaway matefromitsformintheEHSmodel9tobeoftheorder10−5.
from the points where χ and ∆ are maximal. In the physi- Fornotationalreasonswealsowriteq≡(iω,q). Neglecting
calsituationofalargeparamagnonmassandalsostrongcou- the spinwave interaction (u=0) one can formally integrate
pling,themaximumof χ movestowardsthezoneedgelead- outthebosonicdegreesoffreedom. Inthespinbosonmodel,
ing to bond order parallel to the x-y axes. All these findings thisgeneratesaneffectivespin-spininteractionoftheform
pointtotherealizationthat,whilebeingasecondaryinstabil-
itytoAFordering,chargeorderisakeyplayerinthephysics S =−∑J¯(cid:126)S (cid:126)S . (2)
int q q −q
ofthePGphaseofthecuprates. q
It is convenient to use a fermionic representation of the spin
operator(cid:126)Sandconsiderinthefollowingonlytheparamagnet-
II. MODEL icallyorderedphaseinz-directio´n,sothatJ¯=3/2J. Thepar-
tition function then writes Z = D[Ψ]exp(−S −S ) with
0 int
Westartfromthespin-fermionmodel8,9,31withLagrangian
L=L +L ,where
ψ φ
S =∑Ψ†G−1Ψ , (3a)
0 k 0,k k
Lψ =ψ∗(∂τ+εk+λφσ)ψ , (1a) k,σ
L = 1φD−1φ+u(cid:0)φ2(cid:1)2 . (1b) Sint =− ∑ Jqψk†,σψk+Q+q,σ¯ψk†(cid:48),σ¯ψk(cid:48)−q−Q,σ. (3b)
φ 2 2 k,k(cid:48),q,σ
3
800 withk¯ =k+QandthematrixMˆ is
(cid:114), m = 0.001
600 (cid:54)(cid:114)(cid:54),,, mmm === 000...000000515 Mˆk=(cid:18)mˆ†k mˆk(cid:19), mˆk=(cid:18)−−∆χ†k+kp −χ−∆kk(cid:19). (8)
(cid:54)| (cid:114)(cid:54),, mm == 00..0011 ThefermionsinEq.(6)cannowbeintegratedoutsothatthe
(cid:114)|, |400 (cid:114)(cid:54),, mm == 00..0055 partitionfunctionbecomes
| ˆ
(cid:104) 1 1 (cid:105)
200 Z= D[Mˆ]exp − ∑TrJ−1Mˆ Mˆ + ∑TrlogGˆ−1 ,
4 q k¯+q k 2 k
k,q k
(9)
0 40 80 120 with Gˆ−1=Gˆ−01−Mˆ. After functional differentiation of the
a) (cid:104) freeenergyF =−TlnZwithrespecttoMˆk weobtaintheMF
800 equationsinmatrixform
(cid:114) YBCO
(cid:54)(cid:114) BYiB2C20O1 Mˆk=∑Jk¯−k(cid:48)Gˆk(cid:48). (10)
600 (cid:54) Bi2201 k(cid:48)
(cid:114) Bi2212
(cid:54)| (cid:54)(cid:114) HBig21221021 The matrix equation can now be projected onto the different
(cid:114)|, |400 (cid:54)(cid:114) eHl.g d1o2p0e1d coordmepropnaeranmts.etWereswwihllicchoncasindenrohtebreenthoen-czaesreooaftttwheoscaommeppeotiinngt
| (cid:54) el. doped
in k space. Therefore, we consider the equation for ∆ with
200 χ =0andviceversa. Thegapequationsfollowas
0 40 80 120 ∆k=T ∑ Jk¯−k(cid:48)∆2 +ε∆2k(cid:48)+ω(cid:48)2, (11a)
ω(cid:48),k(cid:48) k(cid:48) k(cid:48)
b) (cid:104)
FIG.4. (Coloronline)Panela):Variationofthemaximumvalueof χk=−ℜTω∑(cid:48),k(cid:48)Jk¯−k(cid:48)(iω(cid:48)−εk(cid:48))(iωχ(cid:48)k−(cid:48) εk(cid:48)+p)−χk2(cid:48). (11b)
thegapfunctions|χ|and|∆|inYBCOasafunctionofthecoupling
λ fordifferentmassesm. Panelb):Variationofthemaximumvalue To solve these equations numerically, εk is parametrized
ofthegapfunctions|χ|and|∆|fordifferentcompoundsasafunction in tight-binding approximation with the following parame-
ofthecouplingλ forthemassm=0.5. ters: YBCO33 (parameter set tb2), Bi220134, Bi221235 and
Hg120136 and for electron doped cuprates37. The momen-
tumsumsinEq.(11)arethencarriedoutbydiscretizingthe
wherethebarepropagatoris k-spacebyrectangularandequidistantgrids. Tokeepthenu-
merical computations tractable we neglect the frequency de-
Gˆ0−k1=diag(iω−εk,iω+ε−k−p,iω−εk+p,iω+ε−k), (4) pendence of χ and ∆38. The Matsubara sums are then car-
ried out exactly in the limit T →0 and the momentum sums
and the spinor field Ψ = (ψ ,ψ† ,ψ ,ψ† )T. are performed over 200×200 points and over one Brillouin
k k,σ −k−p,σ¯ k+p,σ −k,σ¯
Furthermore, J−1 = 4D−1/3λ2, σ ∈ {↑,↓} labels the spin, Zone (BZ). Moreover, note that the 2pF vector which con-
q q
Q=(π,π)T istheAFMorderingvectorandpstandsforthe nectstwoopposedFSpointsat±pF dependsontheexternal
momentumkinEq.(11)andisonlyproperlydefinedonthe
2p vector, asdepictedinFig.1. Notethatthechemicalpo-
F FS.Sinceweexpectthatthemaincontributiontothemomen-
tential µ is implicitly subtracted from the dispersion ε . We
k tumsuminEq.(11)comeshoweverfromthehot-spotregion,
selecttheSCandthe2p channelbyintroducingthetwoorder
F we make the approximation to take the 2p vector constant
parameters F
and take the 2p from the hotspot for arbitrary points in the
F
firstBZ,asdepictedinFig.1. Throughoutthisarticle,weuse
∆ =(cid:104)ψ† ψ† (cid:105), χ =(cid:104)ψ† ψ (cid:105). (5)
k k,σ −k,σ¯ k k,σ k+p,σ 10%holefilling(respectively10%electronfillingintheelec-
tron doped case) and the bandgap is 104K. To evaluate the
TheinteractionS isnowdecoupledbymeansofaHubbard-
1
strength of the SU(2) symmetry, we study the level splitting
Stratonovichtransformation. Thepartitionfunctionbecomes
|χ−∆|/∆. Thisparameteraffordthestudyoftherelativeam-
(uptoanormalizationfactor)
plitudebetweentheQDWandSCorderparameter, χ and∆.
ˆ
ItvanishesforaperfectSU(2)symmetryandbecomescloser
Z= D[Ψ]D[∆,χ]exp[−S −S ]. (6)
0 1,eff to one for a complete SU(2) symmetry breaking. From the
Lagragian we find that φλ has dimension of an energy and
Theeffectiveinteractionis φ ∼m−1/2. To estimate the coupling strength, we evaluate
an effective energy via E =φλ ∼m−1/2λ and away from
(cid:104) (cid:105) eff
S1,eff = ∑ Jq−1χk†χk¯+q+Jq−1∆†k∆k¯+q −∑Ψ†kMˆkΨk, (7) criticalityE0∼m0−1/2λ wherewetakem0=1asbaremass,
k,q,σ k,σ compareFig.2.
4
0.4 0.4 0.6 0.4
0.4
Π0 0.2 Π0 0.2 Π0 Π0 0.2
0.2
0
0Π 0Π 0Π 0Π
FIG.5. (Coloronline)Gapfunctions|χ|,|∆|and|χ−∆|/|∆max|(fromuptodown)fordifferentmaterialsinthefirstBZ.Thecompounds
are(fromlefttoright):YBCO,Bi2201,Bi2212andHg1201. Thefigurescorrespondtothesmallmassandsmallcouplinglimit(λ =40and
massm=10−3;notethatthebarecouplingissmallw.r.t.therenormalizedoneE (cid:39)40KandE =1265K)thatismostunfavorableforthe
0 eff
SU(2)symmetry.
III. RESULTSANDDISCUSSION becomesoforderoneareclearlyseen. Withinthenonlinear
σ-modelassociatedtothepresenttheory9, theSU(2)regime
isasignatureofthePGofthesystem,whiletheenergysplit-
A. Massandcouplingdependenceonlevelsplitting
tingofthetwolevelsisassociatedtothesuperconductingT .
c
WeseethatFig.3a)mimicsthegenericphasediagramofthe
cuprateswherethePGlineT∗abruptlyplungesinsidetheSC
Inordertotesttheeffectofthecurvatureontheleveldegen-
domeatsomevalueofoxygendoping.
eracy,wehaveplottedinFig.3a)thevariationofthemaxima
of χ and∆withtheparamagnonmassmforafixedvalueof
thecouplingconstantλ. Weobserveasimilaritybetweenthe
various compounds that we have tested. In a wide range at AlthoughitisveryencouragingtoseethattheSU(2)regime
low value of the mass, the SU(2) degeneracy between χ and has a non-zero probability to exist, one can wonder whether
∆ is verified within a few percents. The existence of such a theparamagnonmassinholedopedcupratesuperconductors
regime is an indication that a PG driven by SU(2) symme- is small, since typically the AF correlation length is of few
tryispossibleincupratesuperconductors. Asthemassisin- lattice constants40. The issue is addressed in Fig. 4a), where
creased we progressively lose the level degeneracy with the the values of χ and ∆ are shown for a fixed mass as a func-
parameter χ abruptly dropping down while the paring ∆ is tionofthecouplingconstantλ. Hereagainagenericpattern
asymptotically going down to zero when the mass increases. emerges. Forsmallλ theSU(2)symmetryisbroken,butsur-
ItisinterestingtoseethattheSU(2)symmetryisweakforthe prisingly,aboveacertainthresholdofλ,theSU(2)symmetry
electrondopedandHg1201compoundwhichexperimentally isalmostcompletelyrestored. AsseeninFig.4a),thebigger
show much weaker signs of charge order22,39. We also find themassis,thestrongerthecouplingconstantneedstobefor
that the compound Bi2212 behaves sightly different than the the symmetry to be restored. Figure 4b) shows that this be-
othercompoundsinFig.3and4,althoughitisnotclearatthe haviorisquitegeneralamongthedifferentcompounds. Note
currentstagewherethisdeviationcomesfrom. InFig.3b)the however that the electron -doped compound is less sensitive
levelsplittingisdirectlyshownforallthecompoundsandthe totheeffectofincreasingthecouplingconstant,comparedto
two regimes, the one at low mass where the SU(2) symme- the other hole-doped ones, for which the SU(2) symmetry is
try is obtained and the higher mass regime where |χ−∆|/∆ restoredforlargeenoughλ.
5
B. Spacialdependenceofthesplitting
The SU(2) symmetry is not only broken due to the curva-
ture of the Fermi surface at the hot spots, but it is typically
broken in the BZ away from the eight hot spots. Figure 5
shows the typical shape of |χ| and |∆| for four compounds
underinvestigationforthemostunfavorablecaseforthesym-
metry, thatisforsmallvaluesofthemassandcouplingcon-
stant. The level splitting is shown as a density plot in the
bottom. It is rather small almost everywhere in the BZ and 0.1
atthehotspotpositionswithmaximaoftheorderof20-40%
aroundthe“shadow”Fermisurface. Themainlearningfrom
Π0 0.05
theseplotsisthatthevariationsoftheFermisurfacegeometry
givesarathersmalldepartingfromtheSU(2)-degeneracyfor
0
a various range of compounds. In all cases, the SU(2) sym- 0Π
metryiswellrespectedatthehotspotpositions.
FIG. 6. (Color online) Generic picture of the gap functions |χ|,
In Fig. 6 we place ourselves in the strong coupling and |∆|and|χ−∆|/|∆max|inthefirstBZforhole-dopedcuprates,here
strong mass regime and plot the variation of |χ| , |∆| and explicitlyshownforYBCO.Thefigurescorrespondtothelargemass
|χ−∆|/∆. The level splitting is also shown in Fig. 6 and andlargecouplinglimit(λ =160andmassm=0.5, sothatE (cid:39)
0
foundtobemuchsmallerthanthepreviouscaseinFig.5,and 160KandEeff=226K)wheretheSU(2)symmetryiswellrespected.
hasdroppedtoanorderof5-10%41. Interestingly,thetypical
shapeof|χ|and|∆|intheBZhaschangedcomparedtoFig.
5, withmaximanowaroundthezoneedge. Thisisthejusti-
ficationthat“hotregions”insteadof“hotspots”isthecorrect
descriptionofholedopedcupratesuperconductorswithinthe
spin-fermion model. Note that since the maximum of |χ| is
now at the zone edge, the wave vector corresponding to the
associatedchargeorderisnowparalleltothex/yaxesofthe
system,insimilaritywiththefindingsofRef.[12].
0.4
Π0
C. Globaltrendsincupratesuperconductors 0.2
The interplay between mass and coupling allows us to re- 0Π
lateglobaltrendsinthephasediagramforcupratesupercon-
FIG. 7. (Color online) Generic picture of the gap functions |χ|,
ductors with the strength of the SU(2) symmetry breaking.
|∆|and|χ−∆|/|∆max|inthefirstBZforelectrondopedmaterials.
In the electron doped compounds the coupling between AF
The figures correspond to the small mass and small coupling limit
modesandconductionelectronsisbelievedtobeweakerthan (λ =40andmassm=10−3,sothatE (cid:39)40KandE =1265K)
0 eff
fortheholedopedcase. Fromtheaboveanalysiswefindthat thatismostunfavorablefortheSU(2)symmetry.
theSU(2)symmetryislessrespectedinthatcaseleadingtoa
smallerPGdomeandoverallsmallerSC,seeFig.7. Forthe
same reasons La compounds where the coupling is also be-
IV. CONCLUSION
lievedtobesmallbehavesimilarly. Ontheotherhand,hole-
doped cuprates like YBCO live in the large mass and large
Inconclusion,thispapergivesfirmgroundtotheintuition
couplingregime. Thisresultsinbroadgappedregionsinthe
thatthechargesectorisakeyplayerinthephysicsofcuprate
BZwherethesymmetryiswellrespectedsothatboththePG
superconductors. While the main instability is still the AF
andSCdomearelarge,asshowninFig.6.
ordering, the d-wave bond order relates to the d-wave pair-
Finally, let us mention that a major effect of a magnetic ing through an SU(2) symmetry. We have shown that there
field is to invert the order of the level splitting between the exists a wide range of parameters where the SU(2) degen-
CDWandtheSCcomponents10,11. Thiswillfavorthecharge eracy is fulfilled, which gives a natural explanation for the
ordercomparedtotheSCpairing. WebelievethatUmklapp large PG regime observed in certain compounds. We argue
scatteringcanhaveasimilareffecttoinvertthelevelordering, that compounds like electron doped cuprates or the La com-
butleavedetailedinvestigationforfurtherstudies. pounds are outside the regime of SU(2) degeneracy, and the
6
morepronouncedenergysplittingisthereasonfortheweaker Carvalho. WethanktheKITP,SantaBarbaraandtheIIP,Na-
PGregime. tal for hospitality during the elaboration of this work. This
workwassupportedbyLabExPALM(ANR-10-LABX-0039-
PALM),oftheANRprojectUNESCOSANR-14-CE05-0007,
V. ACKNOWLEDGMENTS as well as the grant Ph743-12 of the COFECUB which en-
abledfrequentvisitstotheIIP,Natal. Numericalcalculations
WeacknowledgediscussionswithA.Chubukov, S.Kivel- werecarriedoutwiththeaidoftheComputerSystemofHigh
son, H. Alloul, P. Bourges, Y. Sidis, A. Sacuto, and V. S. de PerformanceoftheIIPandUFRN,Natal,Brazil.
1 P.A.Lee,N.Nagaosa, andX.-G.Wen,Rev.Mod.Phys.78,17 Goldman, T. Schmitt, G. Ghiringhelli, N. Barišic´, M. K. Chan,
(2006). C.J.Dorow, G.Yu, X.Zhao, B.Keimer, andM.Greven,Nat.
2 E. Gull, O. Parcollet, and A. J. Millis, Phys. Rev. Lett. 110, Commun.5,5875(2014).
216405(2013). 23 R.-H.He,M.Hashimoto,H.Karapetyan,J.D.Koralek,J.P.Hin-
3 S. Sorella, G. B. Martins, F. Becca, C. Gazza, L. Capriotti, ton, J. P. Testaud, V. Nathan, Y. Yoshida, H. Yao, K. Tanaka,
A.Parola, andE.Dagotto,Phys.Rev.Lett.88,117002(2002). W.Meevasana,R.G.Moore,D.H.Lu,S.-K.Mo, M.Ishikado,
4 M.R.NormanandC.Pépin,Rep.Prog.Phys.66,1547(2003). H.Eisaki,Z.Hussain,T.P.Devereaux,S.A.Kivelson,J.Oren-
5 A. Abanov, A. V. Chubukov, and J. Schmalian,Adv. Phys.52, stein,A.Kapitulnik, andZ.-X.Shen,Science331,1579(2011).
119(2003). 24 I. M. Vishik, M. Hashimoto, R.-H. He, W.-S. Lee, F. Schmitt,
6 A.Chubukov,D.Pines, andJ.Schmalian,inSuperconductivity, D. Lu, R. G. Moore, C. Zhang, W. Meevasana, T. Sasagawa,
editedbyK.BennemannandJ.Ketterson(SpringerBerlinHei- S.Uchida,K.Fujita,S.Ishida,M.Ishikado,Y.Yoshida,H.Eisaki,
delberg,2008). Z.Hussain,T.P.Devereaux, andZ.-X.Shen,PNAS109,18332
7 Closetoamagneticinstability,theSpin-Fermionmodelshould (2012).
capture the relevant low-energy physics of the Hubbard model 25 Y.WangandA.Chubukov,Phys.Rev.B90,035149(2014).
witharbitraryinteractionstrength,seee.g.Ref.[6]. 26 D.F.Agterberg,D.S.Melchert, andM.K.Kashyap,Phys.Rev.
8 M.A.MetlitskiandS.Sachdev,Phys.Rev.B82,075128(2010). B91,054502(2015).
9 K.B.Efetov,H.Meier, andC.Pépin,Nat.Phys.9,442(2013). 27 P.A.Lee,Phys.Rev.X4,031017(2014).
10 H.Meier,M.Einenkel,C.Pépin, andK.B.Efetov,Phys.Rev.B 28 E. Fradkin, S. A. Kivelson, and J. M. Tranquada,
88,020506(2013). arXiv:1407.4480[cond-mat.supr-con] (2014).
11 M.Einenkel,H.Meier,C.Pépin, andK.B.Efetov,Phys.Rev.B 29 X. Montiel, T. Kloss, C. Pépin, S. Benhabib, Y. Gallais, and
90,054511(2014). A.Sacuto,arXiv:1504.03951[cond-mat.supr-con] (2015).
12 H.Meier,C.Pépin,M.Einenkel, andK.B.Efetov,Phys.Rev.B 30 E.Demler,W.Hanke, andS.-C.Zhang,Rev.Mod.Phys.76,909
89,195115(2014). (2004).
13 J. E. Hoffman, E. W. Hudson, K. M. Lang, V. Madhavan, 31 C.Pépin,V.S.deCarvalho,T.Kloss, andX.Montiel,Phys.Rev.
H.Eisaki,S.Uchida, andJ.C.Davis,Science295,466(2002). B90,195207(2014).
14 N. Doiron-Leyraud, C. Proust, D. LeBoeuf, J. Levallois, J.-B. 32 Y. Wang, D. F. Agterberg, and A. Chubukov, Phys. Rev. B 91,
Bonnemaison,R.Liang,D.A.Bonn,W.N.Hardy, andL.Taille- 115103 (2015); Y. Wang, D. F. Agterberg, and A. Chubukov,
fer,Nature447,565(2007). arXiv:1501.07287[cond-mat.str-el] (2015).
15 W.D.Wise,M.C.Boyer,K.Chatterjee,T.Kondo,T.Takeuchi, 33 M.R.Norman,Phys.Rev.B75,184514(2007).
H.Ikuta,Y.Wang, andE.W.Hudson,Nat.Phys.4,696(2008). 34 Y. He, Y. Yin, M. Zech, A. Soumyanarayanan, M. M. Yee,
16 S.E.Sebastian,N.Harrison,R.Liang,D.A.Bonn,W.N.Hardy, T.Williams,M.C.Boyer,K.Chatterjee,W.D.Wise,I.Zeljkovic,
C.H.Mielke, andG.G.Lonzarich,Phys.Rev.Lett.108,196403 T. Kondo, T. Takeuchi, H. Ikuta, P. Mistark, R. S. Markiewicz,
(2012). A.Bansil,S.Sachdev,E.W.Hudson, andJ.E.Hoffman,Science
17 G. Ghiringhelli, M. Le Tacon, M. Minola, S. Blanco-Canosa, 344,608(2014).
C. Mazzoli, N. B. Brookes, G. M. De Luca, A. Frano, D. G. 35 K.Fujita, C.K.Kim, I.Lee, J.Lee, M.Hamidian, I.A.Firmo,
Hawthorn,F.He,T.Loew,M.M.Sala,D.C.Peets,M.Salluzzo, S.Mukhopadhyay,H.Eisaki,S.Uchida,M.J.Lawler,E.A.Kim,
E.Schierle,R.Sutarto,G.A.Sawatzky,E.Weschke,B.Keimer, andJ.C.Davis,Science344,612(2014).
andL.Braicovich,Science337,821(2012). 36 T.Das,Phys.Rev.B86,054518(2012).
18 T. Wu, H. Mayaffre, S. Krämer, M. Horvatic´, C. Berthier, P. L. 37 D.SénéchalandA.-M.S.Tremblay,Phys.Rev.Lett.92,126401
Kuhns, A. P. Reyes, R. Liang, W. N. Hardy, D. A. Bonn, and (2004).
M.-H.Julien,Nat.Commun.4,2113(2013). 38 Wehavecheckedthatthisapproximationisreliabletocalculate
19 D.LeBoeuf,S.Kramer,W.N.Hardy,R.Liang,D.A.Bonn, and thespacialdistributionofthegapfunctions.
C.Proust,Nat.Phys.9,79(2013). 39 R. Comin, R. Sutarto, F. He, E. da Silva Neto, L. Chauviere,
20 M. Le Tacon, A. Bosak, S. M. Souliou, G. Dellea, T. Loew, A.Frano,R.Liang,W.N.Hardy,D.Bonn,Y.Yoshida,H.Eisaki,
R. Heid, K.-P. Bohnen, G. Ghiringhelli, M. Krisch, and J.E.Hoffman,B.Keimer,G.A.Sawatzky, andA.Damascelli,
B.Keimer,Nat.Phys.10,52(2014). arXiv:1402.5415[cond-mat.supr-con] (2014).
21 L.E.Hayward,D.G.Hawthorn,R.G.Melko, andS.Sachdev, 40 V.Hinkov,S.Pailhes,P.Bourges,Y.Sidis,A.Ivanov,A.Kulakov,
Science343,1336(2014). C.T.Lin,D.P.Chen,C.Bernhard, andB.Keimer,Nature430,
22 W.Tabis,Y.Li,M.LeTacon,L.Braicovich,A.Kreyssig,M.Mi- 650(2004).
nola,G.Dellea,E.Weschke,M.J.Veit,M.Ramazanoglu,A.I. 41 Thegapfunctionsandlevelsplittingoftheothercompoundsare
verysimilarandthereforenotshowninFig.6.