Introduction

Superconductivity discovered in iron chalcogenides has stimulated great interests since it is a possible candidate to break the superconducting transition temperature record (Tc ~ 55 K) in the iron-based superconductors. Although the initial Tc in FeSe is just 8 K, it increased up to 14 K1,2 with appropriate Te substitution and 37 K3,4 under high pressure. Furthermore, by intercalating spacer layers between adjacent FeSe layers, Tc has reached ~ 32 K5 in AxFe2−ySe2 (A = K, Cs, Rb and Tl) and 43 K6 in Lix(NH2)y(NH3)1−yFe2Se2 (x ~ 0.6; y ~ 0.2). By applying pressure to AxFe2-ySe2, Tc can even reach ~48 K7. Recent angle-resolved photoemission spectroscopy (ARPES) revealed unexpected large superconducting gap ~19 meV in a single-layer FeSe, which suggests a Tc as high as 65 K8. Among iron chalcogenides, Fe1+yTe1−xSex are unique in their structural simplicity, composing of only iron-chalcogenide layers, which is favorable for probing the mechanism of superconductivity. Band structure calculation9,10 and ARPES11,12,13 results show that Fermi surface of FeTe1−xSex is characterized by hole bands around Γ point and electron bands around M point, which is similar to iron pnictides. Moreover, the competition between magnetism and superconductivity, similar to iron pnictides, is also observed in iron chalcogenides14. The parent compound Fe1+yTe is not superconducting, but exhibits a spin-density wave (SDW) ground state. With Se doping, superconductivity emerges and Tc goes up to 14 K accompanied by the suppression of SDW. Iron chalcogenides also manifest some differences compared with iron pnictides. In particular, Fe1+yTe1−xSex exhibits a bi-collinear antiferromagnetic ordering and it can be either commensurate or incommensurate depending on the Fe content15,16, different from the common collinear commensurate antiferromagnetic ordering observed in all iron pnictides.

Although much attention has been paid to this compound, some crucially fundamental physical properties are still controversial. In optimally-doped Fe1+yTe1−xSex, ARPES13 shows an isotropic superconducting gap, while anisotropic or two-gap features were suggested by angle-resolved specific heat measurement17, muon spin rotation measurements18,19 and optical conductivity measurements19,20. Recently, London penetration depth measurement down to 50 mK indicates a possible nodal gap21. On the other hand, although Liu et al.14 reported that bulk superconductivity resides only in the region of Se level higher than 29%, it was observed in the Se doping level between 10 and 50% by Noji et al.22,23. As for the Hall coefficient, conflicting low-temperature behaviors have been observed in crystals with nominally the same amount of Se24,25,26. Even in the case of resistivity, both the metallic and nonmetallic behaviors were reported and the absolute value just above Tc has a spread from 200 to 1500 μΩ cm1,2,23,24,25,26. These controversies are believed to come from the sample-dependent Fe nonstoichiometries15,27, which originate from the partial occupation of the second Fe site (excess Fe site) in the Te/Se layer. The excess Fe with valence near Fe+ will provide electron doping into the 11 system. The excess Fe is also strongly magnetic, which provides local moments that interact with the adjacent Fe layers28. The magnetic moment from excess Fe will act as a pair breaker and also localize the charge carriers24. Thus the existence of excess Fe complicates the study of iron chalcogenides from the mechanism of superconductivity to normal state properties. Unfortunately, the excess Fe is easily formed in the standard growth technique employing slow cooling29. Specifically, crystals with higher Te content usually contain more excess Fe and the magnitude of excess Fe decreases with the increase in Se doping29. The excess Fe may be also responsible for the unexpected lagged application research despite its simpler structure, less toxic nature and easier synthesis technology compared with iron-based 122 compounds30,31. Thus, eliminating the influence of excess Fe is essential to clarify the intrinsic mechanism of superconductivity as well as to implement applications of iron chalcogenides.

Some previous studies indicated that annealing in O232,33,34, I235 and chalcogen36,37 atmosphere is effective to induce superconductivity in the as-grown Fe1+yTe1−xSex single crystal. In particular, our previous studies34 have demonstrated that proper O2-annealing is very effective in inducing bulk superconductivity in Fe1+yTe0.6Se0.4 single crystal. In this report, we have systematically studied the optimum condition and dynamics of O2-annealing to induce bulk superconductivity. O2-annealing is proved to be effective to remove the excess Fe and a pure crystal without excess Fe was obtained. In addition to the evolution of superconductivity under O2-annealing, changes in some fundamental properties after removing the excess Fe were also studied and presented in the paper.

Results

Optimum condition and dynamics of O2-annealing in Fe1+yTe0.6Se0.4

To study the optimum condition and dynamics of O2-annealing, we firstly fixed the annealing temperature at 400°C and annealed the crystals by successively increasing amount of O2 for 20 hours (enough time for the crystals to absorb the O2 totally). Shown in Figure 1a is the temperature dependence of zero-field-cooled (ZFC) and field-cooled (FC) magnetization at 5 Oe for Fe1+yTe0.6Se0.4 single crystal annealed with increasing the amount of O2. The as-grown crystal usually shows no superconductivity or very weak diamagnetic signal below ~3 K. After annealing, superconductivity emerges and Tc is gradually enhanced with increasing the cumulative amount of O2. The diamagnetic signal also increases with increasing the amount of O2 even faster than the increase of Tc. Its magnitude at low temperatures reaches the maximum value when the onset Tc reaches ~7 K. However, it is well known that even the sample is mostly non-superconducting, the diamagnetic signal becomes significant when the non-superconducting region is covered by superconducting region. Thus, to prove the bulk superconductivity induced by O2-annealing, we need to refer to other quantities. Although specific heat is ideal for this purpose, limited size of the crystal optimized for uniform annealing does not allow reliable evaluation of the volume fraction of superconductivity. As an excellent alternative to probe bulk superconductivity, we propose to use the critical current density Jc. The method used for the calculation of Jc can be seen in the supplement. Self-field Jc at 2 K vs. the amount of O2 are summarized in Figure 1b together with the change in Tc. It is obvious that both Tc and Jc are gradually enhanced by annealing with increasing amount of O2 and reach the optimal values of Tc higher than 14 K and Jc larger than 1 × 105 A/cm2 when the ratio of oxygen to the total Fe is ~1.3% to 1.5%. Bulk superconductivity with high Tc and large Jc is induced by the removal of excess Fe as we will show later. Here we should point out that when the ratio of oxygen is higher than 1.3%, the value of Jc slightly decreases, which may come from the reaction of Fe in Fe layers with O2. Actually, we found the crystal will be decomposed into oxides such as Fe2O3, TeO2 and SeO2 after annealing with too much O2, as shown in the X-ray diffraction (XRD) pattern of Figure S1.

Figure 1
figure 1

(a) Temperature dependence of zero-field-cooled (ZFC) and field-cooled (FC) magnetization at 5 Oe for Fe1+yTe0.6Se0.4 single crystal annealed at 400°C with increasing amount of O2 (molar ratio of the oxygen to the nominal Fe ranging from 0.1% to 1.5%). (b) Changes in Tc and self-field Jc(2 K) for Fe1+yTe0.6Se0.4 annealed at 400°C with increasing amount of O2.

Full size image

Next we study the effect of annealing temperature and time. In this experiment, the crystals were annealed with the optimal amount of O2 (ratio of oxygen to the nominal Fe is 1.5%) at different temperatures with increasing time. It should be noted that the total amount of O2 reacted with the crystal is larger than 1.5% since we fix the amount of O2 in all the annealing processes. Shown in Figure 2a is the typical result of annealing at 200°C. With increasing annealing time, Tc gradually shifts to higher temperatures together with the sharpening of the transition width. Jc is also measured to probe the evolution of bulk superconductivity, which is shown in Figure 2b together with the change in Tc. Both Tc and Jc are increased with increasing annealing time and reach the optimal values after 10 and 100 hours, respectively. To know the influence of temperature, the crystals were also annealed at 250, 300 and 400°C with increasing time and one more typical results of a crystal annealed at 400°C are shown in Figure 2c. At all these annealing temperatures, both Tc and Jc are gradually increased with annealing time similar to the case of 200°C. Changes in Jc with annealing time at different temperatures were further scaled by Jc/Jc(max) and plotted in Figure 2d, which manifests that the rate of inducing superconductivity is accelerated at higher annealing temperatures. It also proves that O2-annealing is very effective to induce bulk superconductivity in Fe1+yTe0.6Se0.4. In particular, in the case of 400°C, Tc and Jc reach the optimal values only after 5 and 60 min annealing, respectively. We also tried to anneal the crystal at higher temperatures, like 500°C. In this case, the crystal is damaged during the annealing and superconductivity is not induced.

Figure 2
figure 2

(a) Temperature dependence of magnetization at 5 Oe for Fe1+yTe0.6Se0.4 single crystal annealed at 200°C with increasing time. Changes in Tc and self-field Jc(2 K) with increasing annealing time at (b) 200°C, (c) 400°C. (d) Changes in Jc2K(0)/Jc, max2K(0) with increasing the annealing time at temperature from 200°C to 400°C.

Full size image

Spatial evolution of superconductivity during annealing

The above results manifest that O2-annealing can gradually induce bulk superconductivity in Fe1+yTe0.6Se0.4, while the spatial evolution of superconductivity during the annealing and distribution of Jc are still unknown, which is crucial to the understand of the mechanism of O2-annealing. In order to reveal these unresolved issues, we employ the magneto-optical (MO) imaging for a crystal annealed for different periods of time, which allows direct observations of the spatial evolution of superconductivity and Jc. Figure 3a and b show the MO images of flux expulsion under 5 Oe at 10 K for the crystal annealed at 400°C for 1 and 3 min, respectively. When the sample is annealed for 1 min, only the edge part shows Meissner state, while the central part still remains non-superconducting. When the annealing time is increased to 3 min, all parts of the sample turn into Meissner state. It indicates that the superconductivity mainly evolves from edges of the crystal to the center.

Figure 3
figure 3

Meissner state magneto-optical (MO) images under 5 Oe at 10 K for Fe1+yTe0.6Se0.4 annealed at 400°C for (a) 1 and (b) 3 min, respectively. (c) Local magnetic induction profiles at temperatures from 5 K to 14 K taken along the dashed lines in Figure S2a. (d) Spatial distribution of Jc/Jc(max) at 10 K for sample annealed at 400°C for 3 min.

Full size image

As we know from Figure 2c, Jc of the crystal annealed at 400°C for 3 min is still lower than the optimal value, which means if the superconducting state evolves from the edge to the center, the edge of the crystal should maintain larger Jc than the central region. To examine this assumption, we also took MO images on the crystal annealed for 3 min in the remanent state at temperatures ranging from 5 to 14 K, which are shown in Figure S2 in the supplement. Figure 3c shows profiles of the magnetic induction at different temperatures along the dashed line shown in Figure S2a. Both the MO images and profiles show that the magnetic field cannot totally penetrate the sample at 5 K. At higher temperature, the field penetrates deeper into the center. At 8 K, the sample is totally penetrated and the MO image manifests a typical roof-top pattern up to 10 K. After that, from 11 to 14 K, the edge of the crystal traps higher magnetic field than the center, which indicates Jc in the edge of the crystal is larger. To directly observe the distribution of Jc in the crystal, we converted the MO images into the current distribution with a thin-sheet approximation and by using the fast Fourier transform calculation method38. A typical image of the distribution of the modulus of Jc at 10 K is shown in Figure 3d, which manifests that the edge of the crystal maintains a larger current and the Jc decays from the edge to the center. All these results testify that superconductivity mainly evolves from the edge of the crystal to the center. It should be noted, however, that we can observe a total expulsion of magnetic flux at 5 K even in the sample annealed for only 1 min (not shown). This fact means that weak superconductivity also evolves from the top and bottom surfaces.

Discussion

The above results on the O2-annealing dynamics prove that it is effective to induce bulk superconductivity in Fe1+yTe0.6Se0.4 single crystal and the superconductivity evolves mainly from the edge of the sample to the central part. Now we turn to the discussion of the mechanism of O2-annealing. Single crystal XRD measurements show that lattice constant c keeps almost constant value during the annealing process as shown in Figure S3a. On the other hand, the lattice constant a increases from 3.802 ± 0.002 Å for the as-grown crystal to 3.812 ± 0.002 Å for the annealed one. Such an increase in the in-plane lattice constant after removing excess Fe is similar to previous reports on structure refinement of crystals containing different amount of excess Fe35,39. However, it is different from the structure change of FeTe0.8S0.2 annealed in O240. Actually, in FeTe0.8S0.2, the superconductivity induced by O2-annealing is reported to be removed by further vacuum annealing40. We also annealed our O2-annealed crystal again in vacuum at 400°C for more than two weeks. The magnetization results for a crystal before and after vacuum annealing were compared in Figure S3b, which shows that the superconductivity induced by O2-annealing is stable under further vacuum annealing. This result together with the unchanged lattice constant c all indicates that the oxygen is not simply doped into the sample.

As is well known, the presence of excess Fe is unavoidable in the growth of Fe1+yTe1−xSex single crystals, which suppresses the superconductivity and causes magnetic correlations27. In our as-grown single crystals, the amount of excess Fe is ~14% as analyzed by Inductively-coupled plasma (ICP) atomic emission spectroscopy. The O2-annealing may induce superconductivity by removing excess Fe from the as-grown crystal. Although the excess Fe may be removed after O2-annealing, it should still remain in the crystal, mainly on the surface, in some form of oxides. Thus, traditional compositional analysis methods like ICP, energy dispersive X-ray spectroscopy (EDX) and electron probe microanalyzer (EPMA) can hardly detect the change of excess Fe. Actually our ICP analyses on the O2-annealed crystal just show a slight change in the Fe content compared with the as-grown one. One important information is the fact that the color of the surface layer of the crystal gradually changes during the O2-annealing and turns into blue after annealing with sufficient O2 as shown in the Figure S4, which is similar to the previous report on the crystal annealed in air41. EDX results show that the blue surface contains more Fe, which is interpreted as a signature that removed excess Fe forms oxide on the surface after annealing. To precisely determine the change in the number of excess Fe, we employ the scanning tunneling microscopy (STM) measurement, which has atomic resolution. The excess Fe occupies the interstitial site in the Te/Se layer42 and the previous report proved that the cleaved Fe1+yTe1−xSex single crystal possesses only the termination layer of Te/Se43, which guarantee the STM can directly observe the excess Fe in Te/Se layer without the influence of neighboring Fe layers. Shown in Figure 4a is the STM image for the as-grown Fe1+yTe0.6Se0.4 single crystal. There are several bright spots in the image, which represent the excess Fe according to the previous STM analysis43,44,45,46,47. After optimally annealing in O2, the bright spots, i.e. the excess Fe, disappeared in STM images as shown in Figure 4b and a larger region in Figure 4c. Actually, we searched several different regions and almost no bright spots can be found, which directly proves that the excess Fe is almost totally removed by O2-annealing. It may be also the reason why our O2-anneald Fe1+yTe0.6Se0.4 single crystal is of high quality.

Figure 4
figure 4

STM images for (a) as-grown, (b) – (c) O2-annealed Fe1+yTe0.6Se0.4 single crystal. The bright spots in (a) correspond to the excess Fe, which disappear in the optimally-annealed crystal.

Full size image

Figure 5a shows the temperature dependence of in-plane resistivity for the as-grown and O2-annealed Fe1+yTe0.6Se0.4 single crystals. For both as-grown and O2-annealed crystals, resistivity maintains a nearly constant value above ~150 K. On the other hand, from 150 K down to the superconducting transition, the as-grown crystal shows a nonmetallic behavior (dρ/dT < 0), which is replaced by a metallic behavior (dρ/dT > 0) after removing the excess Fe by O2-annealing. Similar results are also reported in Fe1+yTe1−xSex with different amounts of excess Fe24,48. To probe the influence of excess Fe to the superconductivity, specific heat is carefully measured in both the as-grown and O2-annealed Fe1+yTe0.6Se0.4 single crystals. Specific heat divided by temperature C/T as a function of T for both crystals are shown in Figure 5b. Obviously, no specific heat jump can be observed in the as-grown crystal, which proves that the superconducting transition observed in the resistivity measurement comes from the filamentary superconductivity. After O2-annealing, a clear specific heat jump can be observed, indicating the bulk nature of the superconductivity in O2-annealed Fe1+yTe0.6Se0.4. Specific heat jump ΔC/Tc was obtained as ~66.3 mJ/molK2 with Tc = 14.5 K determined by entropy balance consideration. The normalized specific heat jump at Tc, ΔC/γnTc, is ~3.0 by using the fitted Sommerfeld coefficient of γn = 22 mJ/molK2. (Details about the fitting of the specific heat data and more discussions can be seen in the supplement.) The value is larger than those of 2.1 ~ 2.8 in previous reports22,49 and is much larger than BCS weak-coupling value of 1.43 indicating that superconductivity in O2-annealed Fe1+yTe0.6Se0.4 is strong coupling in nature. To further reveal the influence of excess Fe on the electronic behavior, Hall coefficient RH of the as-grown and well-annealed samples were measured and shown in Figure 5c. RH is almost temperature independent above 150 K and keeps a constant value ~1 × 10−9 m3/C for both the as-grown and O2-annealed samples. When temperature decreases below 150 K, an obvious divergence in RH is observed. For the as-grown crystal, RH gradually increases with decreasing temperature. On the other hand, for the O2-annealed crystal, RH keeps nearly temperature independence above 50 K, followed by a sudden decrease and finally changes sign from positive to negative before approaching Tc. The sign reversal in Hall coefficient can be attributed to the multiband structure12, indicating the dominance of electron in the charge carriers before the occurrence of superconductivity. We should point out that the sign reversal in RH is also observed in high-quality polycrystal and thin films26,50, usually absent in single crystals with more excess Fe24, which may be the intrinsic nature of FeTe0.6Se0.4. Figure 5d shows the comparison of high-temperature magnetic susceptibility, χ, for Fe1+yTe0.6Se0.4 single crystals before and after the O2-annealing. At high temperatures, susceptibilities for both crystals maintain almost constant values and the value of χ in the O2-annealed crystal is appreciably larger than that of the as-grown crystal, distinct from the previous report51, which may come from the contribution of iron oxide on the surface. When the temperature approaches Tc, χ slightly decreases in the annealed crystal, while it dramatically increases in the as-grown sample. The differences in resistivity, specific heat, Hall effect and magnetic susceptibility in Fe1+yTe0.6Se0.4 before and after the O2-annealing can be attributed to the delocalization of excess Fe by the O2-annealing. Density functional study shows that the excess Fe in the interstitial sites is magnetic and interacts with magnetism of Fe in Fe planes28. The magnetic moment from excess Fe will act as a pair breaker, so that the specific heat jump associated with the superconducting transition is suppressed in the as-grown crystal and recurs after removing the excess Fe by O2-annealing. On the other hand, the magnetic moment will also localize the charge carriers, which causes the nonmetallic resistive behavior and the increase of RH at low temperatures. Moreover, neutron scattering measurements revealed that the coupling between the excess Fe and adjacent Fe layers will tune the AFM order from commensurate to incommensurate15,16. Furthermore, the resistivity, Hall coefficient and magnetic susceptibility for the as-grown crystal all undergo a steep increase below 50 K, which is consistent with the neutron scattering result that the short-range magnetic order in a crystal with rich excess Fe enhances noticeably below 40 K15,52. On the other hand, the divergence of transport properties between the as-grown and O2-annealed samples occurs below 150 K, which may come from the increasing correlation below that temperature53.

Figure 5
figure 5

Temperature dependence of (a) in-plane resistivity scaled by the value at 300 K, (b) specific heat plotted as C/T vs T, (c) Hall coefficient and (d) magnetization under 50 kOe with field parallel to c-axis in the as-grown and O2-annealed Fe1+yTe0.6Se0.4 single crystals.

Full size image

In summary, we have systematically studied the optimal condition and dynamics of O2-annealing in Fe1+yTe0.6Se0.4 single crystal. Bulk superconductivity was gradually induced by annealing with increasing amount of O2 and annealing time at temperatures from 200 ~ 400°C. MO images show that superconductivity mainly evolves from the edge of the crystal to the center. STM observations directly demonstrate that the excess Fe is removed after the optimal annealing. Transport, heat capacity and magnetization measurements on the sample before and after the O2-annealing manifest that the O2-annealing induces bulk superconductivity and delocalize the charge carriers by removing the excess Fe. The controllable way of obtaining high-quality Fe(Te,Se) crystals without excess Fe is hopeful to solve the controversies in iron chalcogenides and to pave the way to overcome the obstacles for applications.

Methods

Sample growth and O2-annealing

Single crystals with a composition Fe1+yTe0.6Se0.4 were grown by using the self-flux method as described in detail elsewhere34. The obtained crystals were then cut and cleaved into thin slices with dimensions about 2.0 × 1.0 × 0.05 mm3, weighed and loaded into a quartz tube with d ~ 10 mm ϕ for further annealing. The quartz tube was carefully baked and tested not to emit any gas under the same condition as the sample annealing. The quartz tube was carefully evacuated by a diffusion pump and the pressure was measured by a diaphragm-type manometer with an accuracy better than 1 mTorr. After totally pumping out the gas, we filled Ar/O2(1%) mixed gas into the quartz tube and sealed it into the length of 100 mm. During the sealing process, the manometer monitored the pressure in the system in real-time to prevent gas leakage and controlled the O2 pressure in the quartz tube. Because of the fixed volume, the amount of O2 filled into the quartz tube can be evaluated by the pressure. Then the crystals were annealed at selected temperatures (ranging from 200°C to 400°C) for different periods of time, followed by water quenching. We also confirmed that almost all O2 in the quartz tube were consumed by the crystal by breaking the quartz tube in a larger quartz tube while monitoring its pressure. In this case, we used pure O2 gas. Furthermore, we confirmed that the increase in the crystal weight after the annealing is equal to the mass of O2 filled into the quartz tube. Our careful experiments described above guarantee that the O2 was totally absorbed by the crystal.

Measurements and verifications

Magnetization measurements were performed to check the superconducting transition temperature Tc and critical current density Jc by using a commercial superconducting quantum interference device (SQUID). Measurements on sample annealed at certain temperature with increasing amount of O2, or increasing annealing time were performed on a given crystal piece to avoid piece-dependent influence. Longitudinal and transverse (Hall) resistivity measurements were performed by the six-lead method. Specific heat was measured with a Quantum Design physical property measurement system (PPMS). Magneto-optical images were taken by using the local-field-dependent Faraday effect in the in-plane magnetized garnet indicator film employing a differential method. Details of the lattice constant change during the annealing was characterized by means of X-ray diffraction (MAC Science M18XHF) with Cu-Kα radiation. Inductively-coupled plasma (ICP) atomic emission spectroscopy and energy dispersive X-ray spectroscopy (EDX) were used for chemical analyses. Scanning tunneling microscopy (STM) images were obtained by a modified Omicron LT–UHV–STM system54. The sample was cleaved in situ at 4 K in an ultra-high vacuum chamber of ~10−8 Pa to obtain fresh and unaffected crystal surface.