ebook img

Interface-induced superconductivity and strain-dependent spin density wave in FeSe/SrTiO3 thin films PDF

0.97 MB·English
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 Interface-induced superconductivity and strain-dependent spin density wave in FeSe/SrTiO3 thin films

Interface-induced superconductivity andstrain-dependent spindensity waveinFeSe/SrTiO thin 3 films S. Y. Tan,1,2 M. Xia,1 Y. Zhang,1 Z. R. Ye,1 F. Chen,1 X. Xie,1 R. Peng,1 D. F. Xu,1 Q. Fan,1 H. C. Xu,1 J. Jiang,1 T. Zhang,1 X. C. Lai,2 T. Xiang,3 J. P. Hu,3 B. P. Xie,1,∗ andD. L. Feng1,† 1State Key Laboratory of Surface Physics, Department of Physics, andAdvanced MaterialsLaboratory, FudanUniversity, Shanghai 200433, People’sRepublicofChina 2Science and Technology on Surface Physics and Chemistry Laboratory, 3 Mianyang 621907, Sichuan, People’s Republic of China 1 3Instituteof Physics, ChineseAcademyofSciences, Beijing100190, People’sRepublicofChina 0 (Dated:January15,2013) 2 Therecordofsuperconductingtransitiontemperature(T )haslongbeen56Kfortheiron-basedhigh c n temperature superconductors (Fe-HTS’s). Recently, in single layer FeSe films grown on SrTiO sub- 3 a strate,signsforanew65KT recordarereported. Herewithin-situphotoemissionmeasurements,we c J substantiatethepresenceofspindensitywave(SDW)inFeSefilms,akeyingredientofFe-HTSthatwas 3 missedinFeSebefore,whichweakenswithincreasedthicknessorreducedstrain. Wedemonstratethat 1 thesuperconductivityoccurswhentheelectronstransferredfromtheoxygen-vacantsubstratesuppress the otherwise most pronounced SDW in single layer FeSe. Besides providinga comprehensive under- ] standingofFeSefilmsanddirectionstofurtherenhanceitsT ,weestablishthephasediagramofFeSevs. n c latticeconstantthatcontainsalltheessentialphysicsofFe-HTS’s. Withthesimpleststructure,cleanest o compositionandsingletuningparameter,itisidealfortestingtheoriesofFe-HTS’s. c - r PACSnumbers:74.25.Jb,74.70.Xa,79.60.-i,71.20.-b p u s FeSe is the simplest and arguably most environmental- the Fermi surface of the K Fe Se (ref 9), except that the . x 2−y 2 t friendly Fe-HTS. The T of bulk FeSe is only about 8 K smallelectronpocketaround(0, 0, π)inK Fe Se isabsent a c x 2−y 2 m (ref 1), however,it reachesas high as 37 K underpressure2. here. Fig.1dshowsthetemperaturedependenceofthesym- Theoretically,a collinear2×1 SDW ordersimilar to that in metrizedphotoemissionspectrumtakenataFermicrossingon - d theironpnictideswaspredictedtobepresentinFeSe(ref3). the Fermi surface aroundM. The measuredsuperconducting n However, partly due to the lack of high quality single crys- gap(∼15meV)islargerthanalltheotherknownbulkiron- o tals,suchSDWorderwasnotfoundinFeSeorcloselyrelated based superconductors,and it closes at a highertemperature c compounds before, and little is known about its electronic of about 65±5 K (Fig. 1e), confirming the previousex-situ [ structure. Onlyspinfluctuationsaroundthe SDWwave vec- ARPESmeasurement7. ThissuggestsanewrecordofhighT c 1 tor were foundin Fe(Te,Se) by inelastic neutron scattering4. forFe-HTS’s,assumingitisnotapseudo-gap. v Themagneticstructureinitsclosestsiblingcompound,FeTe, Intriguingly, the Fermi surface topology changes dramat- 8 4 isbicollinear5,differentfromtheSDWorderobservedinthe ically for the 2 ML film (Fig. 1a2). One not only observes 7 ironpnictides. the Fermi surface of the 1 ML film, but also a new “cross”- 2 Recently, in singlelayerFeSe thin filmsgrownonSrTiO like Fermi surface at the zone corner. Some spectral weight 3 . (STO) substrate by molecular beam epitaxy (MBE), both alsoappearsatthezonecenter. WhenthethirdFeSelayeris 1 0 ScanningTunnellingSpectroscopy(STS)andangle-resolved grown,the1MLFermisurfaceishardlydetected. Asshown 3 photoemission spectroscopy (ARPES) experiments have ob- in Figs. 1a-1c, the electronic structures of multilayer FeSe 1 servedthelargestsuperconductinggapinFe-HTS’s6,7,which filmsarequalitativelysimilar,butwithsubtledifferences.The : likelyclosesabove65K.Althoughfurthertransportmeasure- spectral weight at the zone center is contributed by several v i mentsareneededtoconfirmwhetherthelongstanding56K hole-likebands. TheydonotcrossEF inthe2MLfilm;with X record of T is broken, the remarkable properties of FeSe increasedthickness,theybecomestrongerandcross E , and c F r film, such as the role of the substrate and the superconduct- theholepocketareaincreaseslightly.Thebandstructurenear a ingmechanism,callforfurtherexploration8. M is rather complicated, several bands cross and give four WehavegrownFeSethinfilmslayerbylayeronSTOsub- smallelectronpocketsthatmakethecrossshape. Suchsmall strate with MBE as described in the Methods sections, and pocketshavebeenobservedinBaFe2As2 (ref10), EuFe2As2 it was transferred back and forth between the MBE cham- (ref 11) in the SDW state. Low energy electron diffraction ber and the ARPES chamber under ultra high vacuum for (LEED)patterninFig.1fshowsthatthereisnonoticeablylat- the continuous in-situ growth and measurement of the elec- ticeorchargesuperstructures,thusthecomplicatedelectronic tronic structure as a function of thickness. The electronic structureisnotduetosomebandfoldingbychargeordering. structures of FeSe thin films are presented in Fig. 1. Since In the photoemission intensity map of the 2 ML film thephotoemissionintensityattheFermienergy(E )reflects (Fig.1a2),thecenterofthecircularFermisurfacemismatches F the Fermi surface, one found that the Fermi surface of the the center of the cross-shaped Fermi surface, indicating the 1 ML (monolayer) FeSe film is composed of four electron different Brillouin zone sizes for the two FeSe layers, and Fermipocketsat the zone corners(Fig. 1a1). It is similar to thus different in-plane lattice constant a along the Fe-Se-Fe 2 FIG. 1: The electronic structure of FeSe films as a function of thickness. a1-a6, The thickness dependence of the Fermi surface as representedbythephotoemissionintensityattheFermienergyat30K.b,c,Thethicknessdependenceofthebandstructurearound(b1-b6) (0,0)(cut#1)and(c1-c6)(π,0)(cut#2)respectively. Alldataaretakenat30K.d,TemperaturedependenceofthesymmetrizedEDCatthe Fermicrossingofcut#2for1MLFeSefilm,wheretheclosingofthegapisshownbythefillingupthestatesatE . e,Thesuperconducting F gapvs.temperaturein1MLFeSefilm.Thegapisobtainedfollowingthestandardfittingproceduredescribedinref.9. f,LEED(lowenergy electrondiffraction)patternsfor1MLand4MLFeSefilmrespectively. g,ThelatticeconstantaofthetopFeSelayerasafunctionoffilm thickness,whichisderivedbasedonthephotoemissionmapsinpanelsa1-a6. direction.acouldbederivedfromtheinversedBrillouinzone film is severely expandedfrom the bulk value of 3.765 Å to size determined by high symmetry points of photoemission 3.905 Å enforced by the STO lattice. With increased film maps, and we found that the lattice constant of the 1 ML thickness, a relaxes rapidly (Fig. 1g), and reaches the bulk valueat35MLandabove. The Fermi surface, band dispersion, and even supercon- aa b c g h ductinggapoftheinterfaciallayermeasuredonthe2MLfilm STO STO (treated) 1ML FeSe 0 are the same as those measuredon the 1 ML film (Figs.1a2- eV) -0.1 1c2), which indicates that the electronic structure of the in- E-E(F terfacialFeSe layerisnotaffectedbythesurfaceFeSe layer. -0.2-d10 ST0O 10 -eS1T0O (t0reate1d0) -1f 01ML0 FeS1e0 HighLow nsity (arb.units)i-8E-EF-(4eV) 01SSTTMOOL (Ft-er8eSae0t eEd-)-4E0F(m0eV)40 Twcoheueaprkelibfnoegrtewa,entehdneitithnsetsemtrol.aiAcyhesiroacmnoeeuxtprplyie,ncgttheaednctdoopcnhsleaaqrygueeerntircsaencoshffaetrrhgieeswvneeeaurky- 0 nte j tral within 0.01e− per Fe accuracy based on the calculated E-E(eV)F-0.1 I He lam p LpeurttFineg[edrevriovleudmfero.mThthereeLfourteti,ntgheer0v.o1l2ume−eeaxscseusmsiivnegethleecterloencs- -0.2 x10 Laser tron pockets are made of two degenerate ones like those of -10 0 10-10 0 10-10 0 10 -80-40 0 40 -0.2 -0.1 0 K Fe Se (ref9)]inthebottomFeSelayerisaninterfacial Emission angle (deg.) E-E(meV) E-E(eV) x 2−y 2 F F effect. Based on theLEED patternin Fig. 1f, the 1 ML film areofhighquality,andthereisnosignforanyorderingofSe FIG.2:Electronicstructureevolutionduringthegrowthof1ML vacancies, consistent with the STM measurements6. There- FeSe. a,bandc,Thephotoemissionintensitytakenaroundnormal emission for (a) the STOsubstrate afterdegassing at 550 ◦C for 3 foretheexcessiveelectronsshouldcomefromthesubstrate. hours, (b)STOsubstrateafter30minutesheattreatmentat950 ◦C To answer how such charge transfer is induced at the underSe-flux,and(c)1MLFeSerespectively.Dataweretakenwith interface8,wehavetakenthephotoemissiondataattheendof 21.2eVphotonsofaheliumlamp. d,eandfarethesameaspanels eachofthethreestagesduringthegrowthof1MLFeSefilm a,bandc,exceptthedataweretakenwith7eVphotonsfromalaser. (Fig. 2). The STO substrate was first degassed in ultra high g,andh,Thecomparisonof(g)thevalencebandspectraand(h)the spectranear E takenwithaheliumlampintheabovethreecases. vacuumat550◦Cfor3hours;thenitwasfurtherheat-treated F i,Thesameaspanelhexceptthedataweretakenwitha7eVlaser. at950◦CunderSe-fluxfor30minutes;andfinally1MLFeSe The1MLFeSedataareamplifedtentimesforclarity.jcomparesthe wasgrownonthesubstrate. Attheendofthefirststage, the 1MLFeSedatatakenwithlaserandheliumlamp,afternormalized photoemission data are insulator-like, no states are detected at-0.1eV.Alldataweretakenat30K. near E as shown in Fig. 2a (taken with 21.2 eV photons) F 3 a1 a2 a3 a4 a5 a6 a7 0.05 0.00 gh Hi V) -0.05 E-E(eF -0.10 w Lo -0.15 30K 54K 94K 105K 115K 125K 135K -0.4 0.0 0.4 -0.4 0.0 0.4 -0.4 0.0 0.4 -0.4 0.0 0.4 -0.4 0.0 0.4 -0.4 0.0 0.4 -0.4 0.0 0.4 b1 b2 b3 b4 b5 b6 b7 0.05 0.00 gh Hi E-E(eV)F --00..1005 w Lo -0.15 30K 54K 94K 105K 115K 125K 135K -0.4 0.0 0.4 -0.4 0.0 0.4 -0.4 0.0 0.4 -0.4 0.0 0.4 -0.4 0.0 0.4 -0.4 0.0 0.4 -0.4 0.0 0.4 c d e k(Å-1) f g // 1 1.0 0.0 135K 0.5 0.5 V)-0.1 Intensity (arb.units) 11190124555KKKK E-E(eF-00..02-03.04K 0.0 0.4 -01.345K 0.0 0.4 -1k(Å)y --10001.....05500 30K E-E(eV)F-0-.150 54K E-E(eV)F-0.1 High -00..50 High -1.5 30K -0.2 -80 -40 0 40 -03.04K 0.0 0.4 -01.345K 0.0 0.4 Low -1.0-1.1535K-1.0 -0.5 0.0 Low -2 M Γ E-EF(meV) k// (Å-1) kx (Å-1) k// (Å-1) FIG.3: Temperaturedependenceoftheelectronicstructureforthe50MLFeSefilm. aandb,(a1-a7)Thephotoemissionintensityand (b1-b7)thecorrespondingsecondderivativewithrespecttoenergyaround(π,0). c,ThesecondderivativeoftheEDCat(π,0). Thedipof thesecondderivativecouldhighlightthebandpositionmoreclearly.dande,Theelectronicstructureandthecorrespondingsecondderivative withrespecttoenergyaround(0,0)at30Kand135Krespectively. f,Thephotoemissionintensitymapat30Kand135Krespectively. g, ThecalculatedbandstructureofFeSe3,thethickenedpartofthebandshavebeenobservedinourdata. andFig.2d(takenwith7eVphotons). Inthesecondstage,it Fig.2i. Asthechargehastobeconserved,we concludethat is knownthatSTO loses oxygenat high temperatures12, and the electronsin the localized oxygen-vacancy-inducedstates metallicconductivitywasobservedforSTOannealedinvac- are transferedto the FeSe layer, and thusare responsiblefor uumat800◦C13. TheSe-fluxactsasacleaningagentduring theelectrondopingin1MLFeSe. thisstage,andnoSewouldbeadsorbedatsuchhightempera- A weak but dispersive band just below E becomes visi- F ture.Metal-likeFermistepwithoutmomentumdependenceis ble in the laser data in Fig. 2f, and it dominatesthe spectral observedafterthesecondstage(Figs.2band2e),mimicking weight in the first 40 meV below E . Since compared with F the typical photoemission spectrum taken on polycrystalline theFermistepintensityoftheoxygenvacancystates,therel- gold. The electronsin these states aremostlikelycausedby ativeintensityoftheweakdispersivefeatureismuchweaker the oxygen vacancies. After the 1 ML FeSe was deposited, in the more bulk sensitive laser data than that in the Helium Figs.2c,and2fshowthetypicaldispersionofFeSefilm. To lamp data. Therefore, it can not be from STO. In fact, it is identifythedestinyofthenon-dispersiveSTOstates,Fig.2g visibleintheHeliumlampdataaswellwhenthespectraare compares the valence bands taken at normal emission after normalized at low energies (Fig. 2j), but was missed in the these three stages. The features around-4.5 and -6.7 eV are previous ex-situ ARPES study7. Moreover, because its dis- theSTOvalencestates, whichweretakenatthesamecondi- persionisdifferentfromthehole-likebandinthe2MLfilm, tionandtheirintensitiesarecomparable. Therefore,thepho- itshouldnotbecontributedbytheappearanceofsmall2ML toelectrons from the substrate are not reduced noticeably by FeSe regions. Judgingfromits intensity, itis likely madeof thetopFeSelayer,thusthenon-dispersiveSTOstateswould thed orbital,whosephotoemissionmatrixelementisweak xy have been observedif they still existafter 1 ML FeSe is de- near the zone center, while the stronger band at -0.1 eV is posited. However, in Fig. 2h, the Fermi step is clearly ab- madeofd /d orbitals. xz yz sentinthe1MLdata.Themorebulk-sensitivephotoemission Forthe multi-layerFeSe films, the complicated electronic datatakenwith7eVlasershowsthiscontrastmoreclearlyin structure around the zone corner in Figs. 1a2-a6, and 1c2- 4 c6 is different from that of Fe(Te,Se) (ref 14), but it is very a b c d e similar to those observed in BaFe2As2 (ref 10) and NaFeAs T(K)= T(K)= (ref15)intheirSDWstate. Infact, suchdramaticelectronic 165 T(K)= T(K)= 165 T(K)= SdrmtieuuocaccnorehknidnsoatbfraeuyltllhcedeticiryotoSrnnnoDanhpWmiancisicocsbtrtmiredcueeoecnasltlnuibpnreefireofeaowvlrrdeoeanu1nt0hltt,dio1ef1oe,db1rri5ryes–oa2mtb0ph,apaenaegdneandxrecptihaineclarcsitoumhrbledaeetenenirontoafnnlorsemrh2pm1aa,r2lgaol2----. Intensity (arb.units) 111111145679555550 111111103456785555555 11119432145555 11111432109555554 11111532109555554 netic state. To examine this, Fig. 3 presents the temperature 65 65 54 54 54 3 ML 30 13 ML 30 15 ML 30 25 ML 30 35 ML 30 dependence of electronic structure of the 50 ML film. In- -100 0 -100 0 -100 0 -100 0 -100 0 deed,withincreasingtemperature,theseparatedbandsgradu- f g E-EF(meV) h allybecomedegenerateagainabove125K(Figs.3aand 3b). aSdSprcathihuDunffaocrWgerhteaeorecdeTaontmeNrnbtdreiadieosmsnisrbtsdiiaispocnteensitrcereBvsipmflseiaaodFupranaceebsttv2ryiuaAaeoatlrnnusotyi2ernoheuf(gnaatrTsrsestothAfobnep1rethr0seFSoca)nMDbanatesWntth,aehdonrietitwdNhnscagntaAoat1Futt5Feioe,lcm1MAd7b,oe1sser8rdedg(,i2cnrree0eaesrs.fuecinastT1itgte5ohhda)nete.erhbmoseSiyhffgipoonhtterhhrceerteee--r, Intensity (arb.units)34611235 35550MMM MMMMMLLLLLLLL Band separationT(K)A (meV)11176548640000000 Band separation (meV)1842000 FSSCNSBNerrraaaa00FSFF..FF98eeee2Kee2K 22A00AfiAA.0.19.slsF18ssm282e22F5C2eAo2As02.0s1275As ourresultsprovethatatleastshort-rangedSDWexistsinthe T=30K 120 0 50 ML FeSe thin films below 125 K. Practically, T is the -200 -100 0 0 40 80 0 50 100 150 200 A temperaturethattheseparatedbandsbecomedegenerate,and E-EF(meV) Film thickness (ML) TA (K) here it could be determined by the merging of two dips in FIG. 4: Thickness dependence of the SDW behavior for the the secondderivativeofthe EDCat(π,0)inFig. 3c. On the multi-layerFeSefilms. a-e, Temperatureevolutionofthespectral otherhand,wefoundthatthebandstructurenearzonecenter lineshapeatMforvariousthicknesses. Thetwobandsmergeabove does not change much across the transition, and band fold- acharacteristictemperature T . f,Themaximalbandseparationfor ingduetoSDWisnotobserved(Fig.3d). TheSDW-induced A variousthickness. g,BandseparationandT asafunctionofthick- A foldingis often veryweakforironpnictides, whichis likely ness. h,BandseparationasafunctionofT . Thedashed lineisa A caused by the lack of long range coherence. Consequently, guidefortheeyes. theFermisurfaceismorestronglyreconstructednearthezone corner(Fig. 3f). At high temperatures, the band structureof the 50 ML FeSe film is determined from the second deriva- arationfollows T ratherlinearly in bulk samples. ForFeSe A tiveofthephotoemissiondatawithrespecttoenergy(Figs.3e films,theamplitudeofbandseparationisslightlysmallerthan and 3b7), which shows a qualitativeagreementwith the cal- thoseoftheironpnictides,andthereisamonotonicbutnon- culationofbulkFeSeinthenonmagneticstate(Fig.3g). The linearcorrelationbetweenT andbandseparation. Formany bandrenormalizationfactorisabout2∼3, similarto thoseof A iron pnictides such as BaFe As (refs 10,16,19), T and T 2 2 A N mostironpnictides.SincehighqualtyFeSesinglecrystalwas coincidewith the structuraltransitiontemperature T , how- S not available, and particularly, the natural crystal surface is ever, for compounds like NaFeAs (ref 15), T > T > T . the (110) plane, it was very difficult to obtain the electronic A S N SincetheT isabout105K forbulkFeSe(ref1), andT of S A structureofbulkFeSe. Ourdataonthethickfilmgivethefirst the thick films are about125 K, we expectFeSe follows the experimentalbulkelectronicstructureofFeSe. lattercase,namely,T issomewherebelowT .Thiscertainly N S Thebandreconstructionsobservedinthe50MLfilmhave needstobeconfirmedbymoredirectmeasurementsinthefu- beenobservedinallthefilmswithmorethan1MLthickness ture. (their data are presented in the supplementary Figs. S1-S5). Our results in Fig. 3 indicate that the SDW order ex- Using the temperature dependence of the EDC’s at (π,0) as ists in thick FeSe films. The T determined for the thick A representative, Fig. 4 examines how such signature of SDW films (∼ 125 K) agree well with the ∼130 K temperature order evolves with film thickness. As found similarly to scale determined by recent static and transient optical spec- Fig. 3c, the two features merge into one above the tempera- troscopieson460nmthick(1,0,1)FeSefilmgrownonMgO ture T in all cases (Figs. 4a-4e). The band separation satu- substrate26, wherethe phase transitionnaturewas associated A ratesatlowtemperatures,andFig.4fcollectsEDC’sat(π,0) withnematicity-inducedorbitalorchargeorderingabovethe takenat30K.AssummarizedinFig.4g,themaximalsepara- structural transition. Our data suggest that the optical data tion decreases with increasing thickness, and asymptotically couldbeattributedtotheSDWfluctuations/orderinginFeSe. reaches a constant above 35 ML. Such separation character- The large electronic structure reconstruction observed here izes the strength of the SDW order23, thus as expected, T explainsthespectralweighttransferandpartialgapopeningin A alsodecreasesandbecomesflatinthethickfilms. InFig.4h, theopticaldata. Infact,thedensityfunctionaltheorycalcula- the relation between the maximal band separation and T is tionsbyMaandcoworkers3haveshownthatthegroundstate A shown togetherwith all those existing onesmeasuredon the of bulk FeSe is in the SDW state, instead of the bi-collinear SDW state of various iron pnictides16,17,24,25. The band sep- antiferromagneticorderasforFeTe,becausethethirdnearest 5 200 FeSe crystal FeSe/STO film theelectronicstructureofthe1MLFeSefilmresemblesthat 180 Increasing hydrostatic pressure IncreasingF ethSiec k(finlemsss) ofKxFe2−ySe2. WespeculatethattheTc ofKxFe2−ySe2 could T A have been much higherand closer to the dash-dottedline, if 160 T C werenotforitsseverephaseseparation30,31.Thegeneraltrend 140 ure (K) 120 threornedsuogpgedesFtsetSheatcohmigphoeruTndccthoautldhabselaacrgheiervleadttiicnehceoanvsiltyanetl.ec- mperat 100 SDW IntheSTMstudiesoftheFeSe thinfilms6, itwassurpris- Te 80 ingthatthe1MLFeSeissuperconductingwithhighT ,while FeSe(1ML) c 60 the2MLFeSelayerappears“semiconducting”withmuchre- Li(NH)(NH) FeSe x 2y 31-y 2 2 duceddensityofstatesatE . Besidestheweakcouplingbe- 40 FeSe (Bulk) KFeSe F 2 2 tweenthesetwolayers,themainreasonisthestrongappear- 20 FeTe0.5Se0.5 Superconducting anceofSDWorderinthesurfaceFeSelayer,whichstrongly 0 reconstructstheelectronicstructureandcausesasuppression 3.5 3.6 3.7 3.8 3.9 of spectral weight at E . Moreover, it eliminates any prox- Lattice constant a (Å) F imityof the superconductivityfromthe interfacialsupercon- FIG. 5: Phase diagram of FeSe. The T and T for FeSe are ductingFeSelayer. BecausetheSDWorderisenhancedwith c A ploted against the latticeconstants. The right side isbased on our the increasing lattice constant in thinner films, it is reason- thinfilmARPESdata,andtheleftsideisbasedonthetransportdata able to deduce that if the 1 ML FeSe were not so heavily ofFeSesinglecrystalunderhydrostaticpressuretakenfromref. 2. doped by electrons transferred from the substrate, it would ThedashedlinerepresentstheextrapolatedTA’s,suggestingtheex- havebeeninthestrongestSDWstate. Inagreementwithour istenceofSDWorderinbulkFeSeunderpressure. T forotheriron c results,arecentfirst-principaldensityfunctionalcalculations selenidesarealsoplottedintheellipticalregion. show that the ground state of 1 ML FeSe is a SDW state if itisundoped32. Thesamecalculationalsosuggestedthatthe 2MLfilmisaslightlydopednarrowgapsemiconductor. neighborantiferromagneticexchangeinteractionmediatedby the 4p bands of Se is much smaller than that in FeTe. Our To summarize, we have shown that the property of FeSe resultssubstantiatethisprediction. is sensitive to the lattice, which provides a clean canonical The asymptotic behavior in the thick films indicate that system to study iron-based high superconductors. We have the film property is similar to its bulk behavior when the providedcompellingevidenceforSDWorder(atleastshort- lattice is relaxed to its bulk value. Therefore the thickness ranged SDW order) in FeSe thin films and established the dependence is actually a negative pressure (tensile strain) phase diagram of the FeSe system similar to the iron pni- dependence of the FeSe system. In Fig. 5 we plot the cides. The bulk FeSe electronic structure is revealed. Our phase diagram of FeSe as a function of lattice constant, to- insitumeasurementsnotonlyconfirmthepossible∼65Ksu- gether with the bulk FeSe phase diagram under hydrostatic perconductivityinthe1MLFeSefilm,butalsoshowthatthe pressure2, where the T maximum is 37 K at 7 GPa. The superconductivityof 1 ML FeSe is induced, when the SDW c resulting phase diagram possesses all the same generic fea- initissuppressedbythechargetransferredfromtheoxygen tures as those of the iron pnictides, eg., superconductivity vacancy states of the substrate. We suggest that one might arises when the SDW is weakened, except the tuning pa- further enhance T by expanding the lattice and/or optimiz- c rameter is lattice constant instead of doping. Particularly, ingtheelectronconcentrationviaoptimizedsubstrateprepa- it resembles the phase diagramof BaFe (As P ) (ref 27), ration. Moreover,establishingthesimplestpossibleandpro- 2 1−x x 2 Ba(Fe Ru ) As (ref28),wherephysicalorchemicalpres- totypicalmodel system is often a critical step toward under- 1−x x 2 2 surecouldinducesimilareffects.SeveralotherFeSebasedsu- standing complex phenomena in condensed matter physics; perconductorshavebeenplottedonthesamephasediagram, byidentifyingthemissingSDWinFeSeanditsevolutionwith K Fe Se (ref9)andLi (NH ) (NH ) Fe Se (ref29)are lattice constant, we show that FeSe is the simplest/cleanest x 2−y 2 x 2 y 3 1−y 2 2 heavilyelectron-doped,likethe1MLFeSe/STO.Intriguingly, modelsystemforstudyingFe-HTS. ∗ Electronicaddress:[email protected] (2009). † Electronicaddress:[email protected] 4 Phelan,D.etal.Neutronscatteringmeasurementsofthephonon 1 Hsu, F. C.et al. Superconductivity inthe PbO-typestructure α- density of states of FeSe superconductors. Phys. Rev. B 79, FeSe.PNAS105,14262–14264(2008). 014519(2009). 2 Medvedev,S.etal.Electronicandmagneticphasediagramofβ- 5 Li,S.L.etal.First-ordermagneticandstructuralphasetransition Fe Sewithsuperconductivityat36.7Kunderpressure.Nature inFe Se Te .Phys.Rev.B79,054503(2009). 1.01 1+y x 1−x Mater8,630-633(2009). 6 Wang, Q.Y.etal.Interfaceinducedhightemperaturesupercon- 3 Ma, F.etal. First-principlescalculations of theelectronicstruc- ductivityinsingleunit-cellFeSefilmsonSrTiO .Chin.Phys.Lett 3 tureoftetragonal α-FeTeandα-FeSecrystals:evidence forabi- 29,037402(2012). collinear antiferromagnetic order. Phys. Rev. Lett 102, 177003 7 Liu, D.F. et al. Electronic origin of high-temperature supercon- 6 ductivity in single-layer FeSe superconductor. Nat. Commun 3, 28 Albenque,F.R.etal.Holeandelectroncontributionstothetrans- 931(2012). portpropertiesofBa(Fe Ru ) As singlecrystals.Phys.Rev.B 1−x x 2 2 8 Xiang, Y. Y. et al. High-temperature superconductivity at the 81,224503(2010). FeSe/SrTiO interface.Phys.Rev.B86,134508(2012). 29 Matthew, B.L.etal.Enhancement of thesuperconducting tran- 3 9 Zhang, Y. et al. Nodeless superconducting gap in sitiontemperatureofFeSebyintercalationofamolecularspacer A Fe Se (A=K,Cs) revealed by angle-resolved photoemeis- layer.NatureMater28,3464(2012). x 2 2 sionspectroscopy.NatureMater10,1038(2011). 30 Sun, L. L. et al. Re-emerging superconductivity at 48 kelvin in 10 Yang,L.X.etal.Electronicstructureandunusualexchangesplit- ironchalcogenides.Nature483,67-69(2012). tinginthespin-densitywavestateofBaFe As parentcompound 31 Chen, F. et al. Electronic identification of the parental phase 2 2 ofionbasedsuperconductors.Phys.Rev.Lett102,107002(2009). andmesoscopicphaseseparationofK Fe Se superconductors. x 2−y 2 11 Zhou, B. et al. High-resolution angle-resolved photoemission Phys.Rev.X1,021020(2011). spectroscopystudyoftheelectronicstructureofEuFe As .Phys. 32 Liu,K.,Lu,Z.Y.&Xiang,T.Atomicandelectronicstructuresof 2 2 Rev.B81,155124(2010). monolayerandbilayerthinfilmsonSrTiO (001):First-principles 3 12 Zvanut, M. E. et al. An annealing study of an oxygen vacancy study.Phys.Rev.B85,235123(2012). related defect in SrTiO3 substrates. J. Appl. Phys 104, 064122 Methods: (2008). FeSethinfilmwasgrownontheTiO terminatedandNb- 13 Szot,K.,Speier,W.,Carius,R.,Zastrow,U.&Beyer,W.Local- 2 doped SrTiO (0.5% wt) substrate with the molecular beam izedmetallicconductivityandself-healingduringthermalreduc- 3 epitaxy (MBE) method following the previous report(ref 6). tionofSrTiO .Phys.Rev.Lett88,075508(2002). 3 14 Tamai, A. et al. Strong electron correlations in the normal state Thesubstratewasfirstultra-sonically-cleanedbyusingdeion- oftheion-basedFeSe0.42Te0.58superconductorobservedbyangle- ized water, followed by drying in N2 gas flow. This pro- resolvedphotoemissionspectroscopy.Phys.Rev.Lett104,097002 duces a surface of strontium hydroxide. Then it was subse- (2010). quentlyetchedwithbuffered-oxideetchantinordertoremove 15 He, C. et al. Electronic-structure-driven magnetic and structure thestrontiumhydroxideandformaTiO terminatedsurface. 2 transitions in superconducting NaFeAs single crystals measured After that, annealing treatment in air at 950 ◦C for 2 hours by angle-resolved photoemission spectroscopy. Phys. Rev. Lett was carried out. Further cleaning by trichloroethylene was 105,117002(2010). needed to remove any dust before loading the substrate into 16 Zhang, Y. et al. Unusual doping dependence of the electronic thegrowthchamberwithabasepressureof5x10−10mbar. In structureandcoexistenceofspin-density-waveandsuperconduc- vacuum,thesubstratewasdegassedat550◦Cfor3hours,and torphaseinsinglecrystallineSr K Fe As .Phys.Rev.Lett102, 1−x x 2 2 thenheatedto950◦C undertheSefluxfor30minutes. Dur- 127003(2009). 17 Zhang,Y.etal.Symmetrybreakingviaorbital-dependentrecon- inggrowth,thesubstratewaskeptat490◦Cwiththeselenium structionofelectronicstructureindetwinnedNaFeAs.Phys.Rev. fluxtwentytimesmorethantheFefluxbyco-deposition.The B85,085121(2012). film thickness was monitored by crystal oscillator and con- 18 Zhang,Y.etal.Strongcorrelationsandspin-density-wave phase firmedbyx-rayreflectivitymeasurements. Aftergrowth,the inducedbyamassivespectralweightredistributioninα-Fe1.06Te. film was annealed at 600 ◦C for 3 hours, and directly trans- Phys.Rev.B82,165113(2010). ferredintotheARPESchamberwithatypicalvacuumof1.2x 19 Yi, M. et al. Symmetry-breaking orbital anisotropy oberved for 10−11 mbar. ARPES was conducted with 21.2 eV photons detwinnedBa(Fe Co ) As abovethespindensitywavetransi- 1−x x 2 2 fromaheliumdischargelamp,anda7eVlaser. WithaSCI- tion.PNAS108,6878-6883(2011). ENTA R4000 analyser, energy resolutions of 6 meV (lamp) 20 Yi,M.etal.Electronicreconstructionthroughthestructuraland and 4 meV (laser) and an angular resolution of 0.3◦ were magnetic transitions in detwinned NaFeAs. New J. Phys. 14, achieved. Aging effects were strictly monitored during the 073019(2012). 21 Yin,Z.P.,Haule,K.&Kotilar,G.Magnetismandchargedynam- measurement. icsinironpnictides.NaturePhys7,294-297(2011). Acknowledgement: We gratefully acknowledge Prof. 22 Yin, Z. P., Haule, K. & Kotilar, G. Kinetic frustration and the QikunXue,XiChenandDr. WeiLiforsharingtheirthinfilm natureofthemagneticandparamagnetic statesinironpnictides growthprocedures,Prof.ZhongyiLuforthefileofFeSeband andironchalcogenides.NatureMater10,932-935(2011). structure,andtheenlighteningdiscussionwithProf. Chandra 23 Jiang, J. et al. The distinct in-plane resistivity anisotropy Varma.ThisworkissupportedinpartbytheNationalScience in the nematic states of detwinned NaFeAs and FeTe sin- Foundation of China, and National Basic Research Program gle crystals: evidences for Hund’s rule metal. Preprint at ofChina(973Program)underthegrantNos. 2012CB921400, http://arxiv.org/abs/1210.0397(2012). 24 Yi,M.etal.Unconventionalelectronicreconstructioninundoped 2011CB921802,2011CBA00112,2011CB309703. (Ba,Sr)Fe As acrossthespindensitywavetransition.Phys.Rev. Author contributions: S.Y.T., M.X., X.X., D.F.X., 2 2 B80,174510(2009). H.C.X., and R.P. built the MBE system and grew the films, 25 Ge,Q.Q.etal.Anisotropicbutnodelesssuperconductinggapin S.Y.T., Y.Z., Z.R.Y., F.C., Q.F., J.J. and B.P.X. performed thepresenceofspindensitywaveinion-pnictidesuperconductor ARPESmeasurements.S.Y.T.,D.L.F.,Y.Z.,Z.R.Y.,T.Z.,T.X. NaFe Co As.Preprintathttp://arxiv.org/abs/1209.1967 1−x x and J.P.H. analyzed the ARPES data. D.L.F. wrote the pa- 26 Wen,Y.C.etal.Gapopeningandorbitalmodificationofsuper- per. D.L.F.andX.C.L.are responsiblefortheinfrastructure, conducting FeSe above the structural distortion. Phys. Rev. Lett projectdirectionandplanning. 108,267002(2012). 27 Kasahara, S. et al. Evolution from non-Fermi-to Fermi-liquid AdditionalInformation: Theauthorsdeclarenocompet- transport via isovalent doping in BaFe (As P ) superconduc- ing financialinterests. Correspondenceand requestsforma- 2 1−x x 2 tor.Phys.Rev.B81,184519(2010). terialsshouldbeaddressedtoD.L.F.([email protected]).

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.