From (03C0, 0) magnetic order to superconductivity with (03C0, 03C0) magnetic resonance in Fe1.02Te12212xSex

From (03C0, 0) magnetic order to superconductivity with (03C0, 03C0) magnetic...

(Parte 1 de 2)

From (pi,0) magnetic order to superconductivity

The iron chalcogenide Fe1+y(Te1−xSex) is structurally the simplest of the Fe-based superconductors1–3. Although the

Fermi surface is similar to iron pnictides4,5, the parent compound Fe1+yTe exhibits antiferromagnetic order with an in-plane magnetic wave vector (pi,0) (ref. 6). This contrasts the pnictide parent compounds where the magnetic order has an in-plane magnetic wave vector (pi,pi) that connects hole and electron parts of the Fermi surface7,8. Despite these differences, both the pnictide and chalcogenide Fe superconductors exhibit a superconducting spin resonance around (pi,pi) (refs 9–1). A central question in this burgeoning field is thereforehow(pi,pi)superconductivitycanemergefroma(pi,0) magnetic instability12. Here, we report that the magnetic soft mode evolving from the (pi,0)-type magnetic long-range order is associated with weak charge carrier localization. Bulk superconductivity occurs as magnetic correlations at (pi,0) are suppressed and the mode at (pi,pi) becomes dominant for x>0.29. Our results suggest a common magnetic origin for superconductivityinironchalcogenideandpnictidesuperconductors.

Unconventional superconductivity in cuprates, heavy-fermion intermetallics and strontium ruthenate all occur in close proximity to magnetic instabilities and seem to be mediated by spin fluctuations. The newly discovered iron pnictide superconductors13–16 follow the paradigm of superconductivity achieved by suppressing a long-range magnetic order through charge carrier doping or pressure. The long-range antiferromagnetic (AFM) order in the parent compounds of iron pnictide superconductors is charac- terized by the in-plane Fermi surface nesting wave vector Qn = (pi,pi) (refs 7,8). (Here and throughout this Letter, we refer to wave vectors in units of the inverse tetragonal lattice parameters.) Iron chalcogenide Fe1+y(Te1−xSex) superconductors, discovered more recently1–3, have a similar Fermi surface to iron pnictides, accord- ing to both density functional calculations4 and photoemission measurements5. However, the undoped parent compound of this system, Fe1.02Te, exhibits AFM order characterized by an in-plane wave vector Qm = (pi,0) (ref. 6), which distinguishes this compound from the iron pnictide parent materials. Yet both doped iron chacolgenide11 and iron pnictide9,10 superconductors exhibit a magnetic resonance in the spin excitation spectra below Tc around the wave vector (pi,pi) consistent with s± pairing symmetry17–19. Resolution of the dichotomy between (pi,0) magnetic order in undoped

1Department of Physics and Engineering Physics, Tulane University, New Orleans, Louisiana 70118, USA, 2Department of Physics, Renmin University of China, Beijing 100872, China, 3Helmholtz-Zentrum Berlin für Materialen und Energie, Hahn-Meitner Platz 1, D-14109 Berlin, Germany, 4Institut Max von Laue-Paul Langevin, 6 rue Jules Horowitz, BP 156, F-38042, Grenoble Cedex 9, France, 5Advanced Materials Research Institute and Department of Physics, University of New Orleans, New Orleans, Louisiana 70148, USA, 6NIST Center for Neutron Research, National Institute of Standards and Technology, Gaithersburg, Maryland 20899, USA, 7Department of Materials Science and Engineering, University of Maryland, College Park, Maryland 20899, USA, 8Institute for Quantum Matter and Department of Physics and Astronomy, The Johns Hopkins University, Baltimore, Maryland 21218, USA, 9NSSD, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831, USA. *;

FeTe and superconductivity with (pi,pi) magnetic resonance in Se-doped samples is a key challenge to our emerging understanding of iron-based superconductivity12.

Here, we address the challenge through systematic investigation of transport, magnetic and superconducting properties in various regions of the phase diagram of Fe1.02(Te1−xSex) using resistivity, Hall coefficient, magnetic susceptibility, specific heat and neutron scattering measurements. We find that magnetic correlations of the (pi,0) variety survive as short-range magnetic correlations after the long-range AFM phase has been suppressed by partial Se substitution for Te. Our results link these correlations to weak charge carrier localization in underdoped samples. Bulk superconductivity occurs only when magnetic correlations near (pi,0), although still present, are strongly suppressed and spin fluctuations near (pi,pi) become dominant for x > 0.29. The latter exhibitaspingapandaspinresonanceinthesuperconductingstate. These results indicate that short-range magnetic correlations near (pi,0) are antagonistic to metallicity and superconductivity, and that iron chalcogenide and iron pnictide superconductivity have similar origins associated with (pi,pi) spin fluctuations. Using a variety of techniques we have constructed a detailed electronic and magnetic phase diagram of Fe1.02(Te1−xSex) with 0 ≤ x < 0.5, which is shown in Fig. 1a. In summary, we find three composition regions with distinct physical properties. Region I (0 ≤ x < 0.09) exhibits long-range AFM order with a wave vector (pi,0). Region I (0.09 < x < 0.29) exhibits neither longrange AFM order nor bulk superconductivity. Only in region I (x ≥ 0.29) do we find evidence of bulk superconductivity. More specifically in Fig. 1a, the squares, downtriangles, uptriangles and crosses represent the Néel temperatures TN determined by neutron diffraction, d.c. susceptibility, Hall coefficient and resistivity measurements, respectively. Importantly, these disparate measurements are entirely consistent with each other. TN gradually decreases with increasing Se content, approaching zero for x ∼0.09.

A trace of superconductivity is observed for 0.04 ≤ x < 0.09, which will be examined in greater detail later as a non-bulk phenomenon. The open diamonds represent the onset of the superconducting transition Tcρ as indicated by the resistivity data. In region I (0.09 < x < 0.29), although long-range AFM order is fully suppressed, non-bulk superconductivity remains with a volume fraction, VSC, that is less than 3% for all samples with


ρ Tc _ ρ

AFM Non-bulk-SCBulk SC

Weakly localized Metallic

Tc ρ

Tc χ

dρ χ

60TN _ neutron

TN _ RH TN _Tcbulk_

/dT at 35 K

/d T (arb. units)

(0≤x<0.5). a, The phase diagram. The Néel temperature, TN, of the AFM phase, determined by neutron scattering (green squares), susceptibility

(orange triangles), Hall coefficient (blue triangles) and resistivity (black crosses) measurements. Tcρ, onset of the superconducting transition

probed by resistivity (open diamonds); Tcχ, bulk superconducting transition temperature (filled diamonds) probed by susceptibility. Bulk superconductivity (SC) exists when sufficient Te is replaced by Se, with the superconducting volume fraction >75% for x≥0.29. For x<0.29, only non-bulk-superconductivity exists with the superconducting volume fraction <3%. The bulk superconductivity and non-bulk superconductivity concentration regions also differ in their normal-state transport property: metallic in the former, non-metallic in the latter. b, The superconducting volume fraction (−4piχ) and the derivative of normalized resistivity (ρ(T)/ρ(300K)) with respect to temperature as a function of Se content.

x <0.29. In region I (x ≥0.29), however, bulk superconductivity is found. The filled diamonds represent the bulk superconducting transition temperature Tcχ probed by susceptibility. VSC rises to above 75% for x ≥ 0.29, as shown in Fig. 1b. In region I, the transport properties above Tcρ indicate weak charge carrier localization, which contrasts with the metallic behaviour seen in the normal state of region I and the AFM phase of region I. The cross-over is clearly indicated by the sign change in the derivative of resistivity with respect to temperature dρ/dT (see Fig. 1b). The absence of bulk superconductivity in region I makes the phase diagram of Fe1.02(Te1−xSex) distinct from those of iron pnictide superconductors where bulk superconductivity either appears immediatelyfollowingsuppressionoflong-rangeAFMorder20,21,or coexistswith(pi,pi)AFMorderinacertaincompositionrange22–24.

We shall now address properties of each region of the phase diagram (Fig. 1a) in greater detail. Throughout the AFM phase (region I), elastic neutron scattering measurements reveal the same commensurate (pi,0) magnetic structure as reported for the parent compound6. The ordered magnetic moment of iron MFe depends strongly on the Se content as shown in Fig. 2a. MFe is approximately 2.09(3)µB/Fe for the x =0.04 sample, but decreases to 0.3(2)µB/Fe for x = 0.08. The saturated staggered magnetic moment for Fe1.02Te (ref. 6) is much larger than that of iron pnictide parent compounds (0.36µB/Fe for LaOFeAs (ref. 7) and 0.87µB/Fe for BaFe2As2 (ref. 8)). The Hall effect measurements x = 0.04

RH (10

¬9 m

M Fe (

B /Fe)

M Fe S (

B /Fe)

Figure 2 | Evolution of the long-range AFM order and Fermi-surface variation across the AFM transition in Fe1.02(Te1−xSex). a, Temperature dependence of the ordered magnetic moment MFe. Inset: The saturated moment MFe as a function of Se content. The magnetic order is suppressed when x>0.09. b, Hall coefficients as a function of temperature. The Fermi surface changes significantly across the AFM transition.

shown in Fig. 2b indicate that the AFM transition in region I is accompanied by a remarkable change of the Fermi surface. For x < 0.08 the Hall coefficient RH exhibits a sharp drop from a positive to a negative value across the transition. This indicates that the Fermi surface is dominated by holes above TN and by electrons below TN.

Figure 3a presents the in-plane resistivity ρab(T) as a function of temperature for typical samples in region I. ρab(T) exhibits an anomaly at TN, which is marked by a downward arrow in the figure. Each sample in this region also shows a trace of superconductivity belowthe AFMtransition.Thiscan beseenfrom theseconddrop of ρab(T) at low temperatures (denoted by upward arrows in Fig. 3a).

AlthoughTcρ showsasystematicincreasewithincreasingSecontent in this region, the superconducting volume fraction is nearly zero

(Figs 1b and 3d) because we did not observe any diamagnetism below Tcρ for these samples (see Fig. 3d). Despite the complete suppression of long-range AFM order, superconductivity remains a non-bulk phenomenon throughout region I. Although all samples in this region exhibit zero resistance below Tcρ, their susceptibility fails to exhibit significant diamagnetism when the resistivity vanishes (Figs 1b and 3d). The superconducting volume fraction of these samples estimated from −4piχ is below 3%. Furthermore, the specific heats of samples in both regions I and I are free of anomalies near the resistive superconducting transition. They can approximately be described byC =γT+βT3 atlowtemperatures,whereγT andβT3 represent the electron and phonon specific heat, respectively. Figure 3e shows


x = 0.42 x = 0

TN Tcρ b d e c x = 0.19 ab (m

Ω cm) ρ ab (m

Ω cm) ρ ab (m

Ω cm)

C/T (J mol

¬1 K

C/T (J mol

¬1 K

(J mol

¬1 K

Figure 3 | Evolution of superconductivity as a function of Se content for Fe1.02(Te1−xSex). a, In-plane resistivity ρab(T) as a function of temperature for samples in the AFM region (0≤x<0.09). The downward arrows mark the AFM transition and the upward arrows mark the onset of a trace of superconductivity. b, ρab(T) for samples with 0.09<x≤0.29. c, ρab(T) for samples with x>0.29. d, Magnetic susceptibility data measured with a zero-field-cooling history and a field of 30Oe for typical samples. e, Specific heat divided by temperature C/T as a function of temperature for various samples. The left inset is the electronic specific heat coefficient γ as a function of Se content x. The right inset is C/T as a function of T2 for the x=0.19 sample. Both magnetic susceptibility and specific heat data show that bulk superconductivity occurs only in samples with x≥0.29. The samples with bulk superconductivity exhibit a metallic temperature dependence in ρab in the normal state, whereas those samples without bulk superconductivity exhibit a non-metallic temperature dependence in ρab.

datafortypicalsamples.Theelectronicspecificcoefficientγ derived from linear fitting for various samples is given in the left inset of Fig. 3e. The right inset of Fig. 3e shows an example of the fit for the x = 0.19 sample where we observe a slight deviation from linearity below 4.5K that may be due to non-bulk superconductivity. The significant increase of γ for x ≥ 0.09 is associated with enhanced magnetic fluctuations as shown below.

In contrast, samples in region I exhibit characteristics of bulk superconductivity. Their susceptibility exhibits significant diamagnetism below Tcχ and anomalous peaks near Tcχ in specific heat data also indicate bulk phase transitions (see Fig. 3d,e).

The inferred superconducting volume fraction rises to above 75% for x ≥ 0.29 (Fig. 1b). In addition, samples with bulk superconductivity in region I differ from samples with non-bulk superconductivity in region I in their normal-state properties as noted above. As shown in Fig. 3b,c, the normal state in region I exhibits metallic behaviour in ρab(T). However, samples in region I show a noticeable non-metallic upturn before the superconducting transitions in ρab(T). Indeed, for Tc < T < 20K, ρab(T) is characterized by a logarithmic temperature dependence (see Supplementary Information), indicating weak charge carrier localization in region I. Overall, our phase diagram in Fig. 1a is consistent with previous reports of bulk superconductivity in

Fe1+y(Te1−xSex) single crystals25–27. Why is bulk superconductivity suppressed and charge carriers weakly localized in region I? It is a critical question to understand the difference between the phase diagrams of iron chalcogenide and iron pnictide superconductors, so we examine several possible explanations in the following. One possibility is that quencheddisorder-induced charge localization suppresses superconductivity. As this is an alloy, structural disorder is present throughout regions I and II and does not readily account for the fact that bulk superconductivity occurs only in the latter region. There is however evidence that magnetic correlations are changing substantially from region I to region II. Early neutron scattering measurements revealed that whereas long-range AFM order at (pi,0) is suppressed, short-range static correlations remain near (pi,0) (refs 6,27,28). To clarify the role of such short-range magnetic correlations, we carried out neutron scattering measurements on two high-quality single-crystalline samples with x = 0.19 and 0.38. The x = 0.19 sample resides in region I and has a superconducting volume fraction of ∼0.6%, whereas the x = 0.38 sample is in region I

(Parte 1 de 2)