Strong correlation and massive spectral-weight redistribution induced spin density wave in \alpha-Fe{}_{1.06}Te

Strong correlation and massive spectral-weight redistribution induced spin density wave in -FeTe

Y. Zhang    F. Chen    C. He    L. X. Yang    B. P. Xie Department of Physics, Surface Physics Laboratory (National Key Laboratory), and Advanced Materials Laboratory, Fudan University, Shanghai 200433, People’s Republic of China    Y. L. Xie    X. H. Chen Department of Physics, University of Science and Technology of China, Hefei, Anhui 230027, People’s Republic of China    Minghu Fang Department of Physics, Zhejiang University, Hangzhou 310027, China    M. Arita    K. Shimada    H. Namatame    M. Taniguchi Hiroshima Synchrotron Radiation Center and Graduate School of Science, Hiroshima University, Hiroshima 739-8526, Japan.    J. P. Hu Department of Physics, Purdue University, West Lafayette, Indiana 47907, USA    D. L. Feng Department of Physics, Surface Physics Laboratory (National Key Laboratory), and Advanced Materials Laboratory, Fudan University, Shanghai 200433, People’s Republic of China
July 14, 2019

The electronic structure of -FeTe is studied with angle-resolved photoemission spectroscopy. We show that there is substantial spectral weight around and , and lineshapes are intrinsically incoherent in the paramagnetic state. The magnetic transition is characterized by a massive spectral-weight transfer over an energy range as large as the band width, which even exhibits a hysteresis loop that marks the strong first order transition. Coherent quasiparticles emerge in the magnetically ordered state due to decreased spin fluctuations, which account for the change of transport properties from insulating behavior to metallic behavior. Our observation demonstrates that FeTe distinguishes itself from other iron-based systems with more local characters and much stronger interactions among different degrees of freedom, and how a spin density wave is formed in the presence of strong correlation.


The discovery of iron-based high-temperature superconductors (Fe-HTSCs) has generated great interests Hosono2008 (). So far, two classes of Fe-HTSC have been discovered. They are iron pnictides, e.g., SmOFFeAs or BaKFeAs XHChen (); Johrendt1 (), and iron chalcogenides, e.g., FeTeSe Wu2 (). Although, both classes of materials share many common aspects, such as similarly high maximal superconducting transition temperature () (FeSe possesses a of 37 K under hydrostatic pressure of 7 GPa Prassides ()) and similar band structures from density-functional theory (DFT) calculations DJSingh1 (); DJSingh122 (). However, their parent compounds exhibit quite different spin density wave (SDW) states. A collinear commensurate antiferromagnetic order has been identified for the pnictides PCDai1111 (); XHChen122 (), while a bicollinear and 45-degree rotated antiferromagnetic order was identified for FeTe MaoN (); PCDai (). Furthermore, the transport properties of FeTe respond abruptly to the first order magnetic/structural transition. In the paramagnetic state, it shows insulator-like resistivity [Fig. 1(e)], and optical conductivity without a Drude peak, while the resistivity becomes metallic-like, and a Drude peak emerges in the SDW state NLWang (); MaoT ().

Like the cuprates, the nature of magnetic order and spin fluctuations in Fe-HTSC are most likely crucial for its superconductivity. Yet the origin of the magnetic ordering in iron pnictides/chacogenides is still under heated debate. For the iron pnictides, previous studies have shown that the large reconstruction of the band structure dominates the savings of electronic energy, and would be responsible for the SDW LXYang (); YZhang (); MYi (), while there are also suggestions that the SDW might be dominated by Fermi surface nesting Dong (). For the iron chalcogenides, a connection between the electronic structure and the bicollinear magnetic structure has not been established, except that Fermi surface nesting has been ruled out Hasan (); NLWang (). Many fundamental questions are yet to be addressed for iron chalcogenides: is there any connection between the electronic structure and magnetic ordering; and why is it different from the iron pnictides; what is responsible for the anomalous transport behaviors in iron chalcogenides? The answers of these questions will help build a general picture of iron-based superconductors.

In this Letter, we study the electronic structure of a prototypical parent iron chalcogenide, -FeTe, by angle-resolved photoemission spectroscopy (ARPES). We found that it is profoundly different from those of iron pnictides. The electronic structure of FeTe is dominated by strong correlation, which induces incoherent spectra over extended momentum region in the paramagnetic state. A large square shape of spectral weight unexpectedly appear around and extend to near the Fermi energy (). In the SDW state, with the spectral weight redistribution over a large energy scale of 0.7 eV, sharp quasiparticle peaks emerge near , indicating reduced spin fluctuations. Through detailed temperature-dependence studies, we prove that the massive redistribution of the spectral weight is responsible for the magnetic transition, unveiling a unique manifestation of SDW on electronic structure in the presence of strong correlation.

Figure 1: (Color online) Paramagnetic state electronic structure of FeTe measured at 135 K. (a) Photoemission intensity distribution integrated over the energy window of . (b) The spectral-weight distribution around , which are labeled by the filled orange circle (), the red circle (), the dashed black square (), the dashed blue squares (), and dashed purple squares (). (c) and (d) The energy distribution curves (EDCs) along cut #1 and #2 respectively. (e) The temperature dependence of the resistivity of FeTe. (f) The photoemission intensities along the high symmetry lines, and (g) the corresponding band structure based on DFT calculations. DJSingh1 ().

-FeTe single crystals were synthesized following the method in Ref. Transmeth (). Magnetic susceptibility measurements show that the SDW transition occurs at  = 70 K, accompanied by a structural transition, which is consistent with the neutron and transport reports PCDai (); NLWang (). ARPES data were taken with circularly polarized 24 eV photon at the Beamline 9 of Hiroshima synchrotron radiation center (HiSOR) with a Scienta R4000 electron analyzer. The energy resolution is 10 meV, and angular resolution is 0.3. The sample was cleaved in situ, and measured under ultrahigh vacuum of  torr. Aging effects are strictly monitored during the experiments.

Figure 2: (Color online) SDW state electronic structure of FeTe measured at 15 K. (a) Photoemission intensity distribution integrated over the energy window of . (b) and (c) The EDCs along cut #1 and cut #2 respectively. (d) and (e) are momentum distribution curves (MDCs) at -50 meV around and at different temperatures respectively. (f) The photoemission intensities along the high symmetry lines. Various Fermi crossings are denoted by the short arrows. (g) The photoemission intensity of Fe(TeSe) along . (h) The EDCs taken at 15 K for Fe(TeSe) at , and for FeTe at and .

The electronic structure of FeTe in the paramagnetic state is shown in Fig. 1. The spectra are characterized by broad incoherent feature, while the quasiparticle weight is negligibly small. Thus, Fermi crossings are not well-defined, analogous to the pseudogap in the cuprates. According to the spectral-weight distribution near [Figs. 1(a),  1(c),  1(d), and  1(f)], five main features are labeled as , , and . Their weight distribution is sketched in Fig. 1(b). Note that, contributes a straight dispersed feature at , thus it could not be recognized from EDCs. The features near around and qualitatively agree with calculations shown in Fig. 1(g). Thus, the spectral weight of and might be attributed to the hole and electron pockets respectively, as in FeTeSe ChenFeiPRB (). However, contradicting to the calculations, a fair amount of spectral weight around could be observed near in Fig. 1(d), which is an extension of the feature.

Many changes occur in the SDW state electronic structure (Fig. 2). Most notably, a dramatic reorganization of the spectral weight is observed in Fig. 2(a), where the weight suppression around is particularly strong. Such suppression is obvious by comparing the EDCs around (thick curves) in Fig. 2(c) and Fig. 1(d). Particularly, sharp quasiparticle peaks appear at around and in Figs. 2(b) and 2(c). The flat quasiparticle dispersion and small weight suggest a very low renormalization factor . Similar to BaFeAs, the SDW state of FeTe is accompanied by band shifts. As shown by MDCs in Fig. 2(d), the distance between two peaks decreases in the SDW state, indicating a change of dispersion below . The and bands also exhibit an abrupt momentum shift below , illustrating the enlargement of the electron pockets and band movement around [Fig. 2(e)].

The observed incoherent to coherent lineshape evolution explains the anomalous transport and optical properties of FeTe, particularly the absence of Drude peak in the paramagnetic state, and insulator-metal transition as shown in Fig. 1(e) NLWang (). Compared with the incoherent weight distribution of FeTe in Fig. 2(f), Fig. 2(g) illustrates the well defined band structure of FeTeSe, where the SDW is suppressed by the heavy Se doping. Since both systems contain similar amount of interstitial Fe ions, the broad overall lineshape of FeTe cannot be explained by the magnetic scatting of the excess Fe ions NLWang (). Furthermore, as illustrated in Fig. 2(h), the quasiparticle width of FeTe at low temperatures is much sharper than that of FeTeSe. This result together with the narrow transitions in resistivity [Fig. 1(e)] and magnetic susceptibility [Fig. 4(d)] confirm the high quality of FeTe crystals studied here. Therefore, the incoherent lineshape over a large energy scale should be an intrinsic property of FeTe.

Figure 3: (Color online) (a) The difference between the integrated spectral weight in 135 K and 15 K over the window for FeTe. (b) and (c) Temperature dependence of EDCs at various momenta as marked in panel a.

Due to the low weight of the coherent quasiparticles, and broad overall lineshape, the dominating effect on the electronic structure is not the band shift but the substantial spectral weight redistribution over a large energy scale. Figure 3(a) plots the difference between the integrated spectral weight over at 135 and 15 K in the Brillouin zone. It is clear that spectral weight is suppressed over extended momentum region, and enhanced around at low temperature. Figs. 3(b) and 3(c) compare the EDCs at various representative momenta in the paramagnetic and SDW states. The enhancement of spectral weight often happens within , while the suppression often occurs within . We note that due to the matrix element effects caused by different polarization, the difference map is not entirely symmetric. For example, the high energy part in EDC at momentum 4 is more prominent than that at momentum 3, although they are symmetric with respect to . Overall, a large amount of spectral weight is transferred from lower binding energies to higher binding energies, as a result, the electronic energy is significantly reduced. Such a suppression over a large energy scale is not relevant to Fermi surface instabilities like nesting. Consistently, no sign of gap opening is observed in all cases of Figs. 3(b) and 3(c). NLWang (); Hasan (). Early DFT calculations have predicted strong nesting instabilities with incorrect nesting wavevectors along the direction DJSingh1 (). Later on, it has been amended that the excess iron would significantly alter the electronic structure and produce the right wavevectorDJSinghdop (); LDAdop (). However, this is ruled out again by the absence of gap observed here.

Detailed temperature evolution of the spectral-weight redistribution near is shown in Fig. 4. The suppression at occurs abruptly below , and saturates at low temperatures [Fig. 4(a)]. Furthermore, the temperature cycling experiment with dense steps around in Fig. 4(b) gives a hysteresis loop in the integrated spectral weight in Fig. 4(c), which almost exactly follows the hysteresis loop in the susceptibility data in Fig. 4(d) of this first order transition. This establishes a direct relation between the suppression and the SDW transition, plus proving that our data reflect intrinsic and bulk properties. Similar behavior takes place at [Fig. 4(a)], noting that the difference between and might be caused by different ’s or matrix element effects.

Figure 4: (Color online) (a) Detailed temperature dependence of EDCs at the Fermi crossings of and for sample A. (b) The EDC at for sample B in a detailed temperature cycling experiment near . (c) The integrated spectral weight over as a function of temperature for data in panel b, data are normalized by the integrated weight at 75 K. (d) The magnetic susceptibility hysteresis loop.

Our observation of the intrinsically incoherent electronic structure of FeTe and the spectral weight redistribution associated with SDW transition suggests strong local magnetic fluctuations and their strong coupling to itinerant electrons. Consequently, carriers are more localized, causing local moments and insulating transport behavior NLWang (), and coherent quasiparticles are destroyed in the paramagnetic states. However in the SDW state, when the spin fluctuations are suppressed due to the opening of a spin gap as demonstrated by inelastic neutron scattering INS1 (), the sharp quasiparticles emerge. Consistently, it is found that the ordered moment in FeTe is about 2  PCDai (), much larger than the 0.87  in BaFeAs, or the 0.36  in LaOFeAs PCDai1111 (); XHChen122 (). Theoretically, the models based on magnetic exchange interactions between the nearest and next-nearest neighbor iron moments have successfully explained the bicollinear magnetic structure in FeTe JPHuEPL (); XTFeTe (). Our data will be a decisive support, if incoherent electronic structure and related spectral weight redistribution can be reproduced in these models.

Furthermore, an early ARPES experiment Hasan () has shown that FeTe ( K) exhibits an electronic structure close to that of the non-magnetic FeTeSe ChenFeiPRB (), with more coherent electronic structure. Since the magnetic order in iron chalcogenides could be strongly suppressed by just a small amount of Se, excess iron, or pressure MaoT (); FeTePre (); FeTePha (), our samples with higher of 70 K are in a more strongly ordered state. It is remarkable to observe that the strong correlation effect is enhanced so dramatically here, while is just slightly increased. It is sensible to study how the correlations in iron-based systems are affected by anions (P/As/Se/Te), doping, and pressure.

Similar behavior has been observed in charge density wave (CDW) systems like 2H-TaS, where strong electron-phonon interactions cause incoherent polaronic spectral lineshape, and spectral weight at the over the entire Brillouin zone. It was found that the massive spectral-weight suppression over a large momentum and energy phase space, instead of Fermi surface nesting, is responsible for the CDW in 2H-TaS and 2H-NbSe Shen1 (); Shen2 (). The analogous mechanism of SDW found here for FeTe indicates the density waves at the strong coupling limit share a universal theme, which makes them fundamentally different from the weak interaction systems.

To summarize, we have carried out a systematic photoemission investigation of high quality -FeTe single crystals. We observed an intrinsically incoherent electronic structure, and massive spectral weight redistribution that is responsible for the SDW transition. Our results demonstrate that correlations are probably the strongest in FeTe among all Fe-HTSCs and their parent compounds discovered so far, and reveal universal behaviors of density waves in the presence of strong interactions.

We thank Dr. Donghui Lu for helpful discussions. This work was supported by the NSFC, MOE, MOST (National Basic Research Program No. 2006CB921300), and STCSM of China.


  • (1) Y. Kamihara et al., J. Am. Chem. Soc. 130, 3296 (2008).
  • (2) X. H. Chen et al., Nature (London) 453, 761 (2008).
  • (3) M. Rottor et al., Phys. Rev. Lett. 101, 107006 (2008).
  • (4) K. W. Yeh et al., Europhys. Lett. 84, 37002 (2008).
  • (5) S. Margadonna et al., Phys. Rev. B 80, 064506 (2009).
  • (6) A. Subedi et al., Phys. Rev. B 78, 134514 (2008).
  • (7) D. J. Singh, Phys. Rev. B 78, 134514(2008).
  • (8) C. de la Cruz et al., Nature 453, 899 (2008).
  • (9) Q. Huang et al., Phys. Rev. Lett. 101, 257003 (2008).
  • (10) W. Bao et al., Phys. Rev. Lett. 102, 247001 (2009).
  • (11) S. L. Li et al., Phys. Rev. B 79, 054503 (2009).
  • (12) G. F. Chen et al., Phys. Rev. B 79, 140509 (2009).
  • (13) T. J. Liu et al., Phys. Rev. B 80, 174509 (2009).
  • (14) L. X. Yang et al., Phys. Rev. Lett. 102, 107002 (2009).
  • (15) Y. Zhang et al., Phys. Rev. Lett. 102, 127003 (2009).
  • (16) M. Yi et al., Phys. Rev. B 80, 174510 (2009).
  • (17) J. Dong et al., Europhys. Lett. 83, 27006 (2008).
  • (18) Y. Xia et al., Phys. Rev. Lett. 103, 037002 (2009).
  • (19) T. Taen et al., Arxiv: 0906.1951 [cond-mat.str-el]. (2009).
  • (20) F. Chen et al., Phys. Rev. B 81, 014526 (2010).
  • (21) L. Zhang, D. J. Singh, and M. H. Du, Phys. Rev. B 79, 134514 (2008).
  • (22) M. J. Han and S. Y. Savrasov, Phys. Rev. Lett. 103, 067001 (2009).
  • (23) A. Tamai et al., Arxiv: 0912.3152 [cond-mat.str-el].
  • (24) J. Zhao et al., Phys. Rev. Lett. 101, 167203 (2008).
  • (25) C. Fang, B. Andrei Bernevig and J. P. Hu, Europhys. Lett. 86, 67005 (2009).
  • (26) F. J. Ma et al., Phys. Rev. Lett. 102, 177003 (2009).
  • (27) H. Okada et al., J. Phys. Soc. Jpn. 78, 083709 (2009).
  • (28) P. L. Paulose, C. S. Yadav, and K. M. Subhedar, Arxiv: 0907.3513 [cond-mat.super-con].
  • (29) D. W. Shen et al., Phys. Rev. Lett. 99, 216404 (2007).
  • (30) D. W. Shen et al., Phys. Rev. Lett. 101, 226406 (2008).
Comments 0
Request Comment
You are adding the first comment!
How to quickly get a good reply:
  • Give credit where it’s due by listing out the positive aspects of a paper before getting into which changes should be made.
  • Be specific in your critique, and provide supporting evidence with appropriate references to substantiate general statements.
  • Your comment should inspire ideas to flow and help the author improves the paper.

The better we are at sharing our knowledge with each other, the faster we move forward.
The feedback must be of minimum 40 characters and the title a minimum of 5 characters
Add comment
Loading ...
This is a comment super asjknd jkasnjk adsnkj
The feedback must be of minumum 40 characters
The feedback must be of minumum 40 characters

You are asking your first question!
How to quickly get a good answer:
  • Keep your question short and to the point
  • Check for grammar or spelling errors.
  • Phrase it like a question
Test description