ebook img

Superfluidity in atomic Fermi gases PDF

0.27 MB·
by  Yi Yu
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Superfluidity in atomic Fermi gases

Superfluidity in atomic Fermi gases YiYua,QijinChenb,∗ aCenterforMeasurementandAnalysis,ZhejiangUniversityofTechnology,18ChaowangRd.,Hangzhou310014,China bZhejiangInstituteofModernPhysicsandDepartmentofPhysics,ZhejiangUniversity,38ZhedaRd.,Hangzhou310027,China 1 1Abstract 0 In a trapped atomic Fermi gas, one can tune continuouslyvia a Feshbach resonance the effective pairing interaction between 2 fermionicatomsfromveryweaktoverystrong. Asa consequence,thelowtemperaturesuperfluidityevolvescontinuouslyfrom n the BCS type in the weak interaction limit to that of Bose-Einstein condensation in the strong pairing limit, exhibiting a BCS- a JBEC crossover. In this paper, we review recent experimentalprogressin atomic Fermi gases which elucidates the nature of the 4superfluidphaseastheinteractioniscontinuouslytuned. Ofparticularinterestistheintermediateorcrossoverregimewherethe 1s-wavescatteringlengthdiverges.Wewillpresentanintuitivepairingfluctuationtheory,andshowthatthistheoryisinquantitative agreementwithexistingexperimentsincoldatomicFermigases. ] s Keywords: a gAtomicFermigases,superfluidity,BCS-BECcrossover,pairingfluctuations -PACS:03.75.Hh,03.75.Ss,74.20.-z t n a u 1. Introduction temperature[4]. ThediscoveryofhighTcsuperconductivityin q 1986gaveastrongboosttotheinterestinBCS-BECcrossover . t Ultracold atomic Fermi gases have been a very exciting, a [6, 7, 8, 9, 10, 1]. It was suggested that the unusual pseudo- rapidly developing field, which has emerged within the past m gap phenomena in the cuprate superconductorsmight have to severalyears,bridgingcondensedmatterandatomic,molecular dowithBCS-BECcrossover. Experimentaleffortsinthisarea - dandopticalphysics[1]. UsingaFeshbachresonanceinamag- fell far behind, because it had been difficult to find a system nneticfield,onecantunetheeffectivepairinginteractionstrength where the attractive pairing interaction is tunable. Thanks to obetween fermionic atoms from very weak to very strong [2]. thelasercoolingandtrappingtechniquein1990’s,oneisable c Astheinteractionstrengthvaries,thenatureofthelowtemper- [ tocreate“artificial”many-bodysystemsoffermionicatomsin ature superfluidity of these Fermi gases evolves continuously a laboratory. The existence of a Feshbach resonance in these 1fromtheBCStypeintheweakcouplinglimittoBose-Einstein Fermigasesmakesitpossibletotunetheinteractionstrength. v condensation (BEC) in the strong pairing limit, exhibiting a 6 For ease of control, the Feshbach resonances for the two BCS-BECcrossover,whichhasbeenofgreattheoreticalinter- 4 widely studied species, 6Li and 40K, are both very wide. The 8estsince1960’s[3,4,5,1]. Ofparticularinterestistheunitary interaction in both cases are of the short-range, s-wave type. 2regime,wherethe s-wave scatteringlengthdiverges. Thisisa They are often taken to be a contact potential in theoretical .strongly correlated regime where no small parameter is avail- 1 treatments. 0ableforperturbativeexpansions. Ithasbeenexpectedthatthis 1regime provides a prototype for studying both high T super- The first experimental realization of BCS-BEC crossover c 1conductorsandstronglyinteractingFermigaseswhicharealso wasachievedin2004byJinandcoworkers,[11,12]andalmost v:ofinteresttonuclearandastro-physicists. thesametimebytheGrimmgroup[13]andtheKetterlegroup i Inthispaper,wefirstreviewexperimentalprogressinatomic [14]. Due to the difficulty in tuningtemperatureT, the Fermi X Fermigases,withanemphasisonrecentradiofrequencyspec- gaseswereeitherinthesuperfluidornormalstateatgiveninter- artroscopymeasurements.Thenwewillpresentapairingfluctua- action strength (or the magnetic detuning). Continuousvaria- tiontheorycomparewithexperiment.Weshowthatthistheory tionofthesystemasafunctionoftemperaturewasfirstrealized successfullyexplainexperimentalmeasurements. bytheThomasgroup[15]atunitarity.Incollaborationwiththe theory groupat Chicago [16], Thomaset al [17] observedfor thefirsttimecontinuousphasetransitionfromthenormaltosu- 2. Experimentalprogress perfluidstateinaunitary6Ligas. Onecouldargue,ofcourse, ThefirsttheoreticalstudyofBCS-BECcrossoverdatesback thatthevortexmeasurementoftheKetterlegroupprovidedthe to1960’s,althoughitdidnotgetmuchattentionuntilthesem- mostdefinitivesmokinggunforasuperfluidstate. [18] inal work of Leggett in 1980 on BCS-BEC crossover at zero Besides the interaction strength, another great tunability is populationimbalancebetweenthetwofermionicspeciestobe ∗Correspondingauthor:[email protected] paired [19]. It adds a whole new dimension to the phase dia- PreprintsubmittedtoPhysicaC:SuperconductivityanditsApplications January17,2011 gramandmakesthephysicsmuchricher. Italso generatesin- terest[20]inpossibleobservationofthe Larkin-Ovchinnikov- Fulde-Ferrell(LOFF)state[21]. Experimentalworkinpopula- tionimbalancedFermigaseswaspioneeredbytheHuletgroup [22] and the Ketterle group [23]. Experiment in the extreme Figure1: Schematicdiagramsforthefermionicself-energyΣ(K). Thedotted population imbalancedlimit by the Ketterle groupmanifested and (red) double lines represent the condensate and finite momentum pairs, [24] the importance of Hartree-like correlationeffects besides respectively. BCS-typeofpairing. Unlikeanelectronsystem,ithasbeendifficulttomeasurethe mation,givenbythegrandcanonicalHamiltonian excitation gap in the Fermi gas superfluid. Amongall experi- mentaltechniques,Radio frequency(RF) spectroscopy[13] is H− µ N = (ǫ −µ )a† a σ σ k σ k,σ k,σ arguablythemostdirectprobe. UsingatunableRFfieldtoex- Xσ Xk,σ cite oneofthetwo pairingatomsfromalowerhyperfinestate + U(k,k′)a† a† a a , (1) q,k,k′ q/2+k,↑ q/2−k,↓ q/2−k′,↓ q/2+k′,↑ (level2)toahigherhyperfinelevel3whichdonotparticipate P in pairing, a higher frequency will be needed if the atoms in whereǫ =~2k2/2misthefreeatomdispersion.Thedifference k level2arepaired. Suchafrequencyshift(detuning)providesa betweenEq.(1)anditsBCScounterpartisthatBCSkeepsonly goodmeasureoftheexcitationgap. Previousmeasurementby the q = 0 term in the interactions. The inclusion of finite q Grimm and coworkers[13], and later repeatedby the Ketterle termsallowsincoherent,finite momentumpairing. Forclarity group [24], was done in a momentum integrated fashion. At ofpresentation,wewilltakeacontactpotential,U(k,k′) = 1, low T, the RF spectra displayeddouble-peakstructure, with a and use a 4-momentum notation, K = (k,iω ), Q = (q,iΩ), n l sharppeakatzerodetuningandabroadpeakatpositivedetun- = T , and set ~ = 1. Population imbalance can be K k n ing. This double-peakfeature was nicely interpreted [25, 26] PdescribedPbyPµ↑ , µ↓. However,herewewillonlypresentthe as transitionsfrom unpairedatoms the trap edge (correspond- equationsforthecaseofequalspinmixture. Generalizationto ing to the sharp peak)andfroma distributionof pairedatoms populationimbalancecanbefoundinRef.[33]. intheinnerpartofthetrap(broadpeak). However,doubtwas Weassumethat(i)thefermionicselfenergyΣhasapairing castabouttheoriginofthetwopeaksastowhethertheyreflect origin,(ii)pairscanbeeithercondensedorfluctuatingwithafi- pairingofboundstateeffects[24]orsimplyaresultoftrapin- nitemomentum,and(iii)condensedandnoncondensedpairsdo homogeneity[27]. Recently, attentionwasalso drawnto final notmixatthelevelofT-matrixapproximation.Figure1shows stateeffectsboththeoretically[28,29]andexperimentally[24]. diagrammaticallythecontributionstotheself-energy,wherethe A big step in the RF technique was the recent momentum- double(red)linesindicatefinitemomentumpairsandthedot- resolved RF spectroscopy experiment in 40K by the Jin tedlineindicatesthecondensate. Thesubscripts“sc”and“pg” group [30]. With momentum resolution, RF spectroscopy is stand for superfluid condensate and pseudogap contributions, equivalent to the angle-resolved photoemission spectroscopy respectively. (ARPES) for an electron system, In fact, it is cleaner than To tackle this problem, we use a Green’s function method. ARPESinthatARPESisonlyatwo-dimensionalprobe,which We derive the equations of motion for one- and two-particle isoftenplaguedbytheexistenceofsurfacestates,surfacecon- Green’s functionsG and G , which will involve higher order, 2 taminations, work function, and the complication of energy three particle Green’s functions G : iG˙ = [H,G] ∼ G,G , 3 2 dispersion in the third dimension. In comparison, of course, iG˙ = [H,G ] ∼ G,G ,G . We then truncate the equations 2 2 2 3 the signal-to-noise ratio in a Fermi gas experiment is much of motion at the level ofG , factorizeG into a sum of prod- 3 3 lower, as limited by the (low) total number of atoms in the ucts of G and G , and treat G and G on equal footing. For 2 2 gas. although the trap inhomogeneity adds complication to G , we focus on the particle-particle channel, neglecting the 2 the interpretation of the spectrum. Like ARPES, momentum- particle-hole channel which normally only provides a chemi- resolvedRFspectroscopymeasuresthefermionspectralfunc- calpotentialshift. Weemphasizethatitistheparticle-particle tion, A(k,ω), which is of central importancein characterizing channelthatgivesrisetosuperfluidity. Aftersomelengthybut thesystem. straightforwardderivation,weobtaintheselfenergy: Σ(K) = Σ (K)+Σ (K), (2) sc pg 3. TheoreticalFormalism Σ (K) = −∆2G (−K), (3) sc sc 0 Σ (K) = t (Q)G (Q−K), (4) pg pg 0 In this section, we now presenta simple pairingfluctuation X Q theory, which was first developed [10] to explain the pseudo- gapphenomenainhighTcsuperconductors.Fermigasesinthe where presenceofaFeshbachresonancecanbeeffectivelydescribed U t (Q)= (5) by a two-channelmodel [2]. It has now been known that the pg X1+Uχ(Q) closed-channelfraction[31,32]isverysmallforboth6Liand Q 40K,throughouttheBCS-BEC crossover. Therefore,forthese is the (pseudogap)T-matrix,and χ(Q) = G (Q−K)G(K) K 0 systems,aone-channelmodelisoftenusedasagoodapproxi- isthepairsusceptibility. HereG0 isthebaPreGreen’sfunction. 2 40 0.5 Equations(9)and(7),alongwiththenumberequation V) (a) ∆ (b) B E C n=2 G(K), (10) me TF XK s ( /c p ∆ T formaclosedsetofequationsforthehomogeneouscase,which Ga ∆pg sc CS can be used to solve for µ, Tc, and the gaps at T ≤ Tc. Tc is B determinedby setting ∆ = 0. Typicalbehaviorsof the gaps sc 0 0 0 0.5 1 1.5 -2 0 2 4 areshowninFig.2(a). T/T 1/k a To address Fermi gases in a trap, we use the local density F F approximation,byreplacingµ → µ−V (r). Thenthenum- trap Figure2: Typicalbehaviorof(a)thetemperaturedependenceofthegapsina ber equation becomes N = d3rn(r). In Fig. 2(b) we show pseudogappedsuperfluid,andof(b)Tcasafunctionof1/kFainatrap,where the BCS-BEC crossoverbehaRviorof Tc in a trap. Here 1/kFa kFisthenoninteractingFermimomentum,andaisthes-wavescatteringlength. parametrizestheinteractionstrength. The RF response can be derived using the linear response theory.TheRFinteractionisdescribedby A detailed derivation of this result can be found in Ref. [34]. NotethattheT-matrixiseffectivelyarenormalizedpairingin- teraction. Itsharesexactlythesamepolestructureasthetwo- Hrf =eiΩtZ d3xψ†3ψ2+h.c., (11) particle Green’s function,G . Througha Taylor expansionof 2 itsdenominator,onecanextractthepairdispersion: andtheresponseKernelby t−pg1(Q)≈Z(iΩl−Ωq+µpair). (6) D(iΩl)=XG(2)(K)G(3)(K+Q). (12) K The superfluid instability is given by 1+Uχ(0) = 0 ∝ µ , pair We assume hyperfine level 3 was initially empty. In the ab- whichistheBECconditionforpairs. Notethatχ(Q)involvesa sence of final state interactions, as in 40K, we obtain [35] the mixofbareandfullGreen’sfunctions. Weemphasizethatthis RFcurrent is a natural consequence of the equation of motion technique sleinacdesibtainckvotlovethsethBeCoSp-eforartmoroGfˆ−0g1a.pIteiqsutahtiisonG0inGtfhoermabosfenχctehaotf I(k,ν) = −π1ImDR(ν+µ−µ3) finitemomentumpairs. 1 QW=e0.fDoceufisnoinngthe superfluid phase where tpg(Q) diverges at = 2πXk A(k,ω)f(ω)(cid:12)(cid:12)(cid:12)(cid:12)(cid:12)ω=ǫk−µ−ν. (13) (cid:12) InordertoaddressA(k,ω)= −2ImG(cid:12)(k,ω+i0+)properly,we ∆2pg ≡− tpg(Q), (7) need to include the lifetime effects of finite momentum pairs XQ,0 andaddanincoherenttermiΣ in(andonlyin)Σ , reflecting 0 pg the residuetermδΣ whichwe dropin solvingthe setof equa- wehave tions,i.e., Σ (K) = − t (Q) G (−K)+δΣ ∆2 pg  pg  0 Σ (k,ω)= pg −iΣ . (14) = −∆XQ2 G (−K)+δΣ. (8) pg ω+ǫk−µ+iγ 0 pg 0 WhileaboveT thespectralfunctionwithapseudogapconsti- c tuteadoublepeakstructurewithsuppressedspectralweightat NeglectingtheresiduetermδΣ,Σ takesthesameformasΣ . pg sc theFermilevel,belowT ,thereisazeroatω = −(ǫ −µ). As Thus we have immediately the BCS form of total self energy, c k ∆ increases with decreasing T below T , the spectral peaks Σ(K) = −∆2G (−K), with ∆2 = ∆2 +∆2 . Thisthenleadsto sc c 0 sc pg sharpenrapidly. Thisisaphasecoherenceeffect. Theparame- theBCSformofgapequation, tersγandΣ canbeestimatedfromexperimentalRFspectra. 0 1−2f(E ) 1+U k =0, (9) Xk 2Ek 4. Comparisonbetweentheoryandexperiment where E = (ǫ −µ)2+∆2 is the quasiparticle dispersion. In Fig. 3, we compare between theory (curves) and experi- k k Different frompthe BCS mean-field theory, we emphasize that ment(symbols)(a)the densityprofile[36] and(b)system en- here ∆2 containscontributionsfrom both condensedandnon- ergy [17] for 6Li in the unitary limit. Both experimental and condensed pairs so that it in general does not vanish at T . theoreticaldensityprofilesareverysmooth,ingoodagreement c Note thatthefiniteqpairsaredifferentfromthe orderparam- witheachother.Alternativetheoriespredictseitherakinkatthe etercollectivemodes;thelatterrepresentacoherentmotionof edgeofthesuperfluidcoreornonmonotonicradialandtemper- the condensate. Here ∆2 and ∆2 are loosely proportionalto aturedependence. Theenergycomparisonalsorevealsaquan- sc pg thedensityofcondensedandnoncondensedpairs,respectively. titativeagreement. Thefactthattheunitaryandnoninteracting 3 1.5 Theory, noninteracting 20 2 T/T =0.19 F 4 Theory, unitary noninteracting 10 1 1 n(x)0.5 (a) E/EF2 unitary (b) ω (kHz) −100 ωµ+)/EF−01 ( −20 −2 T=0.29 0 -1 x (1000µm) 1 00 c 0.5 T/TF 1 1.5 −300 5 10 15 −30 0.5 1 1.5 k (µm−1) k/k F Figure3:Comparisonof(a)densityprofileand(b)energyE/EF foraunitary 6Ligasbetweentheory(curves)andexperiment(symbols). Alsoshownin(b) Figure4:ComparisonofspectralintensitymapI(k,ν)k2/(2π2)betweenexper- iscomparisonforthenoninteractingenergy. HereEF =kBTF isthenoninter- iment(left)fromRef.[30]andtheory(right). Thewhitedashedcurveisan actingFermienergy. experimentalextractedquasiparticledispersion,andthewhitesolidlineisob- tainedtheoreticallyfollowingthesameexperimentaldataanalysisprocedure. curvesmergeatT ≈ 0.6T ≫ T manifeststhe presenceof a F c pseudogap.Itshouldbenotedthatthereisnofittingparameter [5] P.Nozie`resandS.Schmitt-Rink,J.LowTemp.Phys.59,195(1985). inourtheoreticalcalculations. [6] Y.J.Uemura,PhysicaC282-287,194(1997). ShowninFig.4isacomparisonofthespectralintensitymap [7] R.FriedbergandT.D.Lee,Phys.Lett.A138,423(1989). [8] Randeria,PhysicaB198,373(1994). as a function of k and single-particle energy ω + µ between [9] B.Janko´,J.Maly,andK.Levin,Phys.Rev.B56,R11407(1997). experiment[30]andtheory[35]foraunitary40Kgasatatem- [10] Q.J.Chen,I.Kosztin,B.Janko´,andK.Levin,Phys.Rev.Lett.81,4708 peratureslightlyaboveT . Thewhitedashedcurveistheexper- (1998). c imentally extractedquasiparticledispersion, whereasthe solid [11] C.A.Regal,M.Greiner,andD.S.Jin,Phys.Rev.Lett.92,040403(2004). [12] Q.J.Chen,C.A.Regal,M.Greiner,D.S.Jin,andK.Levin,Phys.Rev. curveisobtainedtheoreticallyfollowingtheexperimentalpro- A73,041601(2006). cedure.Itisevidentthattheoreticalandexperimentalresultsare [13] C. Chin, M.Bartenstein, A.Altmeyer, S.Riedl, S.Jochim, J.Hecker- rather close to each other. Indeed, as T decreasesfrom above Denschlag,andR.Grimm,Science305,1128(2004). [14] M.W.Zwierlein, C.A.Stan, C. H.Schunck, S.M.F.Raupach, A.J. to below T , the spectral intensity map evolves [35] from an c Kerman,andW.Ketterle,Phys.Rev.Lett.92,120403(2004). upwarddispersingbranchathighT toabifurcationaroundT , c [15] J.Kinast,A.Turlapov,andJ.E.Thomas,arXiv:cond-mat/0409283. and finallyto a downwarddispersingbranchat T ≪ Tc. This [16] Q.J.Chen,J.Stajic,andK.Levin,Phys.Rev.Lett.95,260405(2005). result establishes the actual single particle dispersions which [17] J. Kinast, A. Turlapov, J. E. Thomas, Q. J. Chen, J. Stajic, and K. Levin, Science 307, 1296 (2005), published online 27 January 2005; contribute to the RF current, revealing that the broad peak in doi:10.1126/science.1109220. previous momentum-integrated RF spectra [13] indeed has a [18] M.W.Zwierlein,J.R.Abo-Shaeer,A.Schirotzek,andW.Ketterle,Na- pairingorigin. Furthermore,italsoshowsthat,despitethetrap ture435,170404(2005). inhomogeneity,momentumresolvedRFspectroscopycanstill [19] Q.J.Chen,Y.He,C.-C.Chien,andK.Levin,Phys.Rev.A74,063603 (2006). provideaquantitativemeasureofthespectralfunctionandsin- [20] Y.He,C.-C.Chien,Q.J.Chen,andK.Levin,Phys.Rev.A75,021602(R) gleparticledispersion.ItalsolendssupportforthepresentG0G (2007). schemesincealternativeNSR-basedtheoriesdonot[37]seem [21] P.FuldeandR.A.Ferrell,Phys.Rev.135,A550(1964);A.I.Larkinand to generate the two-branch-like feature observed in Ref. [30]. Y.N.Ovchinnikov,Zh.Eksp.Teor.Fiz.47,1136(1964)[Sov.Phys.JETP 20,762(1965)]. The downwarddispersion around (and above)T providesdi- c [22] G.B.Partridge,W.Li,R.I.Kamar,Y.A.Liao,andR.G.Hulet,Science rectevidencefortheexistenceofapseudogapaboveTc atuni- 311,503(2006). tarity. Ourtheoryservesasabasisformomentum-resolvedRF [23] M.W.Zwierlein,C.H.Schunck,A.Schirotzek,andW.Ketterle,Nature spectroscopyanalysis. (London)442,54(2006). [24] C.H.Schunck,Y.Shin,A.Schirotzek,M.W.Zwierlein,andW.Ketterle, In summary, we have presenteda pairingfluctuationtheory Science316,867(2007). wherefinitemomentumpairingplaysaprogressivelymoreim- [25] J.Kinnunen,M.Rodriguez,andP.To¨rma¨,Science305,1131(2004). portantroleasthepairingstrengthincreases,leadingtoapseu- [26] Y.He,Q.J.Chen,andK.Levin,Phys.Rev.A72,011602(R)(2005). [27] E.J.Mueller,arXiv:0711.0182(unpublished). dogap in the single particle excitation spectrum. This theory [28] S.BasuandE.Mueller,Phys.Rev.Lett.101,060405(2008). hasbeensuccessfullyappliedtomultipleexperimentsinatomic [29] Y.He,C.C.Chien,Q.J.Chen,andK.Levin,Phys.Rev.Lett.102,020402 Fermigases. (2009). ThisworkwassupportedbyZhejiangUniversityandNSFof [30] J. T. Stewart, J. P. Gaebler, and D. S. Jin, Nature (London) 454, 744 (2008). ChinaGrantNo.10974173. [31] G.B.Partridge,K.E.Strecker,R.I.Kamar,M.W.Jack,andR.G.Hulet, Phys.Rev.Lett.95,020404(2005). [32] Q.J.ChenandK.Levin,Phys.Rev.Lett.95,260406(2005). References [33] C.-C.Chien,Q.J.Chen,Y.He,andK.Levin,Phys.Rev.Lett.97,090402 (2006). [1] Q.J.Chen,J.Stajic,S.N.Tan,andK.Levin,Phys.Rep.412,1(2005). [34] Q.J.Chen,Ph.D.thesis,UniversityofChicago, 2000,(available inthe [2] J.N.Milstein,S.J.J.M.F.Kokkelmans,andM.J.Holland,Phys.Rev. ProQuestDissertations&ThesesDatabaseonline). A66,043604(2002). [35] Q.J.ChenandK.Levin,Phys.Rev.Lett.102,190402(2009). [3] D.M.Eagles,Phys.Rev.186,456(1969). [36] J.Stajic,Q.J.Chen,andK.Levin,Phys.Rev.Lett.94,060401(2005). [4] A. J. Leggett, in Modern Trends in the Theory of Condensed Matter [37] E.Mueller,privatecommunication. (Springer-Verlag,Berlin,1980),pp.13–27. 4

See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.