Doping-insensitive density-of-states suppression in polycrystalline iron-based superconductor SmO 1−x F x FeAs
H. W. Ou, Y. Zhang, J. F. Zhao, J. Wei, D. W. Shen, B. Zhou, L. X. Yang, F. Chen, M. Xu, C. He, R. H. Liu, M. Arita, K. Shimada, H. Namatame, M. Taniguchi, Y. Chen, X. H. Chen, D. L. Feng
aa r X i v : . [ c ond - m a t . s up r- c on ] M a y Doping-insensitive density-of-states suppression in polycrystalline iron-basedsuperconductor SmO − x F x FeAs
H. W. Ou , Y. Zhang , J. F. Zhao , J. Wei , D. W. Shen , B. Zhou , L. X. Yang , F. Chen , M. Xu , C. He , R. H.Liu , M. Arita , K. Shimada , H. Namatame , M. Taniguchi , Y. Chen , X. H. Chen , and D. L. Feng ∗ Department of Physics, Surface Physics Laboratory (National Key Laboratory),and Advanced Materials Laboratory, Fudan University, Shanghai 200433, P. R. China Hefei National Laboratory for Physical Sciences at Microscale and Department of Physics,University of Science and Technology of China, Hefei, Anhui 230026, P. R. China and Hiroshima Synchrotron Radiation Center and Graduate School of Science,Hiroshima University, Hiroshima 739-8526, Japan (Dated: October 30, 2018)We investigated the temperature dependence of the density-of-states in the iron-based supercon-ductor SmO − x F x FeAs ( x = 0 , . , . , .
2) with high resolution angle-integrated photoemissionspectroscopy. The density-of-states suppression is observed with decreasing temperature in all sam-ples, revealing two characteristic energy scales (10 meV and 80 meV). However, no obvious dopingdependence is observed. We argue that the 10 meV suppression is due to an anomalously doping-independent normal state pseudogap, which becomes the superconducting gap once in the super-conducting state; and alert the possibility that the 80 meV-scale suppression might be an artifact ofthe polycrystalline samples.
The recent discovery of superconductivity in the ironoxypnictides LnO − x F x FeAs (Ln=La, Sm, Nd, etc. )with transition temperatures ( T c ’s) well beyond theMcMillan limit for BCS superconductors has rejuvenatedintensive research on unconventional superconductivity[1, 2, 3, 4, 5, 6, 7]. In some ways, the layered struc-ture of a iron-based superconductor resembles the high-T c cuprates: LnO − x F x layers act as the charge reservoir,and FeAs layers act as the conducting layers. Moreover,neutron diffraction measurements show that the groundstate of the parent compound LaOFeAs is a spin den-sity wave (SDW) [8]. This makes the phase diagrams ofcuprates and iron oxypnictides share some common fea-tures as well. Whether or not the superconductivity iniron oxypnictides follows the same physics as in cupratesthus becomes a fundamental question to ask.One prominent feature for cuprates is the pseudogap, i.e. suppression of density-of-states (DOS) in the normalstate. For LaO − x F x FeAs, several nuclear magnetic res-onance experiments have argued the existence of a pseu-dogap [9, 10, 11]. Recently, Sato et al. [12] reported theangle integrated photoemission spectroscopy (AIPES) re-sults on polycrystalline LaO . F . FeAs ( T c =24K), andthey observed a pseudogap of 15 ∼
20 meV far above T c up to 130K, together with a finite DOS at the Fermienergy ( E F ). Almost the same time, Ishida et al. [13]reported AIPES evidence for a 100 meV pseudogap-likefeature at 250K in LaO . F . FeAs (LOFFA, T c =26K),which shrinks to 20 meV at 70K. Moreover, their data onthe low- T c LaO . F . FeP (LOFFP, T c =5K) shows aless pronounced 20 meV pseudogap-like feature. There-fore, these experiments on polycrystals do suggest theexistence of a pseudogap of 9 ∼ k B T c , and it is verysimilar to the 10 ∼ k B T c pseudogap scale observed incuprate superconductor[14]. At the lowest temperature, both groups observed a spectral weight suppression whichwas related to superconducting gaps of about 4 ∼ meV for LOFFA’s, and a superconducting gap of about 1 meVfor LOFFP. Intriguingly, they are all about 2 k B T c .Photoemission is sensitive to surface and sample qual-ity, systematics thus needs to be collected before reach-ing a robust conclusion when dealing with polycrys-tals. However, so far only one doping of each mate-rial was measured in individual studies. In this pa-per, we report systematic AIPES measurement of poly-crystalline SmO − x F x FeAs (SOFFA) with a variety ofdopings (x=0, 0.12, 0.15, 0.2). At high temperatures,we observed a DOS suppression over a large energyscale of 80 meV, whose onset temperature is even higherthan 300K. At low temperatures, a V-shaped lineshapearound the energy scale of 10 meV was observed in thesymmetrized spectra. Surprisingly, we found that both
SmO F x FeAs ρ ( m Ω * c m ) x=0x=0.12x=0.15x=0.2, FIG. 1: (color online). Temperature dependence of Resistiv-ity for SmO − x F x FeAs (x=0, 0.12, 0.15, 0.2)(data are takenfrom Ref.[15]). -1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 x=0x=0.12x=0.15x=0.2 I n t en s t i y ( a r b . un i t s ) E-E F (eV) -8 -6 -4 -2 0-10 @10K FIG. 2: (color online). AIPES spectrum near E F as a functionof doping for SmO − x F x FeAs. The inset shows AIPES spec-trum of SmO . F . FeAs over a large energy window. Datawere taken with 22.7eV photons at 10K. kinds of suppressions are doping-independent, very dif-ferent from the pseudogap behaviors in cuprates. Thelow energy suppression scale of 10 meV corresponds to2 k B T maxc , where T maxc is the maximum T c of 54K forSOFFA’s. Considering similar correlations in LOFFAand LOFFP, it suggests that the 10 meV suppression re-flects an intrinsic pseudogap in the normal state, whichevolves into the superconducting gap at low tempera-tures. However, it is not clear at this stage whether the80meV scale is related to an intrinsic pseudogap effect, orcaused by extrinsic effects such as inhomogeneities anddomain boundaries of the polycrystal.Polycrystalline SmO − x F x FeAs (x=0, 0.12, 0.15, 0.2)have been synthesized through solid state reaction, thedetailed information about the synthesis and character-ization of the sample has been described elsewhere[15].Resistivity data (Fig.1) clearly indicate the supercon-ducting transition occurs at 25K, 42K, 47K, and 54Kfor x=0.12, 0.15, 0.2( ∼
60% accord-ing to the susceptibility measurements, which is amongthe best for the polycrystalline samples synthesized sofar. The resistivity of the parent compound exhibits asimilar peak at around 150K similar to that observed inLOFFA, indicative of a possible SDW and/or structurephase transition in SOFFA [8].Photoemission measurements were performed at beamline 5-4 of Stanford Synchrotron Radiation Laboratory(SSRL) and beam line 9 of Hiroshima Synchrotron Ra-diation Center (HiSOR), data were taken with ScientaR4000 electron analyzers. The overall energy resolutionwas set to 7 meV. The sample rod was cracked in-situ andthen measured in ultra-high vacuum ( ∼ × − mbar ). We emphasize that all the sample surfaces prepared inthis way show consistent results, and the data measuredwith 22.7 eV photons agree with those measured withmore bulk sensitive 8eV photons (shown in Fig.4), in-dicative of the high sample quality.The AIPES spectra of SmO − x F x FeAs are shown inFig. 2, which measure the DOS. There is a broad low en-ergy feature around 0.22 eV below E F , whose positionseems to be quite doping independent. The low energyfeature was attributed to Fe 3 d states in various bandcalculations[16, 17, 18, 19, 20]. Our data, especially thepeak at 0.22 eV and the flat DOS between 0.5 eV to 3 eVbinding energy (see inset of Fig.2)[21], agree best with thecalculation that considers an antiferromagnetic groundstate[19, 20]. On the other hand, this low energy fea-ture does not show up in dynamical mean field theorycalculation[22].The evolution of DOS near E F is studied as a functionof doping and temperature. As shown in Fig.3(a1-d1),the lineshape evolution above and below E F clearly dif-fers from the symmetrical Fermi-Dirac distribution func-tion normally observed on a polycrystalline metal. Thespectra [ n ( ω )] are symmetrized to remove the thermalbroadening effects[23], and the resulting n ( ω )+ n ( − ω ) areshown in Fig.3(a2-d2) respectively. This has been com-monly practiced in the study of gap in cuprate supercon-ductors. In this way, a spectra weight suppression withdecreasing temperature is clearly revealed, which occursat the highest measured temperature over the energyrange of ± meV . With decreasing temperature, furthersuppressions of the DOS happen in a smoothly shrinkingenergy window around E F , as shown by the arrows andsummarized in Fig.3e. This resembles the anisotropicpseudogap opening behavior in cuprates, where gap oflarger size opens at higher temperature[23]. It also indi-cates the maximum gap could well exceed 80 meV. How-ever, unlike the cuprates, the suppression has no dopingdependence. All AIPES spectra exhibit similar behaviorin the entire investigated doping range. As quantified inFig.3f, the DOS at E F has almost the same linear tem-perature dependence when normalized by their values at300K.To study the superconducting state, high resolutionAIPES spectra were measured with 1 meV steps nearthe Fermi energy at low temperatures in Fig. 4(a1-d1).The insets in Fig.4(a1-d1) show the enlargement near E F , the cross-points indicate a leading-edge gap of 1 ∼ meV fluctuating with sample. The corresponding sym-metrized angle-integrated spectra are shown in Fig. 4(a2-d2). There is clearly an additional abrupt drop of DOSat a fixed characteristic energy scale of 10 meV at lowtemperatures (see Fig.3e), which eventually causes a V-shaped DOS near E F at the lowest temperatures. The10 meV energy scale itself is quite intriguing, since it isabout 2 k B T maxc for SmO − x F x FeAs. Similar low energyscales of 2 k B T c have been observed for LOFFA[12, 13], -100 -50 0 50 100-100 -50 0 50 100 I n t en s i t y ( a r b . un i t s ) E-E F (meV) (a2)(a1) (b2)(b1) (c2)(c1) (d2)(d1) T (K) n ( ω = , T ) / n ( ω = , K ) ( % ) (e) D O S s upp r e ss i on S c a l e ( m e V ) (f)806040200 x=0x=0.12x=0.15x=0.2 x=0x=0.12x=0.15x=0.2 FIG. 3: (color online) Temperature dependence of the SmO − x F x FeAs AIPES spectrum near E F for (a1) x = 0, (b1) x = 0 . x = 0 .
15, and (d1) x = 0 . E F normalized by their values at 300K. Data were taken with 22.7 eVphotons. and LOFFP[13], therefore it is likely an intrinsic pseu-dogap related to pairing fluctuation in the normal state.This gap might have started to open above 100K, butthe thermal broadening prevents it being observed. If thepseudogap/superconducting gap transition in optimallydoped cuprates could apply here, it naturally evolves intothe superconducting gap once the system enters the su-perconducting state. It is remarkable that such a pseu-dogap is doping independent and very small in SOFFA,compared with the large and doping dependent pseudo-gap in cuprates. Furthermore, the V-shaped DOS withfinite value at E F indicates an anisotropic gap with nodesor Fermi arc in the superconducting state[21].The doping-independent DOS suppressions do raisethe question whether they are artifacts from the polycrys-talline nature of the sample. Although the Meissner ratiois 50 ∼
60% for the superconducting samples, as high ascurrently one could get, still there are 40 ∼
50% non-superconducting part in the polycrystal, which might come from the domain boundary or inhomogeneity of thefluorine dopants or oxygen vacancies. It is not surprisingif this portion of the polycrystal could be independentof doping, and cause the observed 30% spectral weightsuppression at E F . Furthermore, the quite large DOSaround E F at 10K could be also attributed to the non-superconducting portion besides the possible nodes orFermi arc of the superconducting portion, which natu-rally explains the fluctuating leading edge gaps as well.However, for the suppression at 10 meV energy scale, itscorrelation with T maxc in several systems indicates thatit should be an intrinsic effects from the bulk part of thepolycrystal. For the suppression at 80 meV energy scale,there is no strong evidence to exclude the extrinsic poly-crystalline effects, except the fact that this energy scaleis smoothly connected to the 10 meV scale as shown inFig.3e.To summarize, we have found doping indepen-dent behavior of the spectral weight suppression in -100 -50 0 50 100 -100 -50 0 50 100 SmO1-xFxFeAsx=0.2, I n t en s i t y ( a r b . un i t s ) E-E F (meV) (a2)(a1) (b2)(b1) (c2)(c1)
60K 10K (d2)(d1) -10 -5 0 5 10
FIG. 4: (color online) Temperature dependence of theSmO − x F x FeAs AIPES spectrum near E F at low tempera-tures for (a1) x = 0 . T c = 54 K , (b1) x = 0 . T c = 47 K ,(c1) x = 0 . T c = 25 K and (d1) x = 0 respectively. Thecorresponding symmetrized spectra are shown in (a2-d2) re-spectively. Data in (b1-d1) were taken with 22.7 eV photonsat SSRL, while data in (a1) were taken with 8 eV photons atHiSOR. SmO − x F x FeAs. A large “pseudogap” of 80 meV is ob-served from the highest measured temperature, whoseorigin is currently unclear, and debatably, could be ex-trinsic. A smaller gap of 10 meV becomes observable be-low 100K, which is likely an intrinsic pseudogap in thenormal state, and would become a superconducting gaponce in the superconducting state. Therefore, the pseu-dogap behavior in iron oxypnictides is very different fromthat in cuprate superconductors. Moreover, our dataalert that the results obtained from polycrystalline iron oxypnictides have to be interpreted with caution.The authors thank Dr. D. H. Lu and R. H. He for theirkind help at SSRL, and thank Prof. Z. D. Wang, and S.Y. Li for stimulating discussions. This work was sup-ported by the Nature Science Foundation of China andby the Ministry of Science and Technology of China (Na-tional Basic Research Program No.2006CB921300 and2006CB922005), and STCSM of China. SSRL is oper-ated by the DOE Office of Basic Energy Science underContract No. DE-AC03-765F00515.Note added: During the preparation of thismanuscript, we noticed another work on SmO − x F x FeAsis posted online[24]. ∗ Electronic address: [email protected][1] Y. Kamihara, et al. , J. Am. Chem. Sco. , 3296(2008).[2] X. H. Chen et al. , Nature advance online publication,25/05/2008 (doi:10.1038/nature07045)[3] G. F. Chen et al. , arXiv:0803.3790 [cond-mat.supr-con].[4] Zhi-An. Ren et al. , arXiv:0803.4283v1 [cond-mat.supr-con][5] Zhi-An. Ren et al. , arXiv:0804.2053v2 [cond-mat.supr-con][6] W. L. McMillan, Phys. Rev. 167, 331 (1968).[7] C. Day, Physics Today 61, 11(2008)[8] Clarina de la Cruz et al. , arXiv:0804.0795v1 [cond-mat.supr-con][9] K. Ahilan et al. , arXiv:0804.4026v1 [cond-mat.supr-con][10] H. J. Grafe et al. , arXiv:0805.2595v1 [cond-mat.supr-con][11] Y. Nakai et al. , arXiv:0804.4765v1 [cond-mat.supr-con][12] T. Sato et al. , arXiv:0805.3001v1 [cond-mat.supr-con][13] Y. Ishida et al. , arXiv:0805.2647v1 [cond-mat.supr-con][14] A. Damascelli, Z. Hussain, and Z.-X. Shen, Rev. Mod.Phys. 75, 473 (2003).[15] R. H. Liu et al. , arXiv:0804.2105v3 [cond-mat.supr-con][16] D. J. Singh and M. H. Du, arXiv:0803.0429 [cond-mat.supr-con][17] G. Xu, W. Ming, Y. Yao, X. Dai, and Z. Fang,arXiv:0803.1282 [cond-mat.supr-con][18] I. I. Mazin, D. J. Singh, M. D. Johannes, and M.H. Du,arXiv:0803.2740 [cond-mat.supr-con][19] C. Cao, P. J. Hirschfeld, and H.-P. Cheng,arXiv:0803.3236 [cond-mat.supr-con][20] F. Ma and Z. Y. Lu, arXiv:0803.3236 [cond-mat.supr-con][21] H. W. Ou et al. , Chin. Phys. Lett. , 2225(2008)[22] K. Haule, J. H. Shim, and G. Kotliar, arXiv:0803.1279[cond-mat.str-el][23] M. R. Norman et al. , Nature(London) 392, 157 (1998)[24] H. Liu et al.et al.