Introduction

Superconductivity in quasi-two-dimensional crystal structures has attracted sustained interest in the past decades. The most outstanding examples include high-Tc cuprates with CuO2 superconducting layers1 and Fe-based superconductors with a Fe-square lattice2. Very recently, superconductivity of BiS2-based compounds which have similar layered crystal structure as those of cuprates and Fe-based materials has been reported. The first member of the BiS2-based superconducting family is Bi4O4S3 with a Tc of 8.6 K3. It was found that the characteristic BiS2 layers are responsible for the superconductivity3. So far, several ReBiS2O1-xFx (Re = La, Ce, Pr and Nd) and doped SrBiS2F superconductors have been discovered with the highest Tc of 10.6 K4,5,6,7,8,9,10. Band structure calculations indicate that the undoped parent compounds such as LaBiS2O and SrBiS2F are insulators with an energy gap of 0.82 and 0.80 eV, respectively11,12. Upon electron doping, both compounds exhibit metallic conducting behavior and a superconducting transition at low temperatures4,10. On the other hand, recent works demonstrate that the isostructural compounds EuBiS2F and Eu3Bi2S4F4 are metallic, and they even exhibit superconductivity without extrinsic doping, at temperatures below 0.3 K and 1.5 K respectively13,14, different from the other analogues. By various experimental approaches, it is pointed out that the self-doping nature of the observed superconductivity in both EuBiS2F and Eu3Bi2S4F4 is due to the mixed valence of Eu13,14. Currently, with 50 at.% Ce substitution for Eu in EuBiS2F, the Tc is enhanced up to 2.2 K15. It was suggested that the Eu valence is essentially divalent in Ce-doped system15. On the contrary, the average Eu valence with respect to the parent compound increases with the Se doping in Eu3Bi2S4-xSexF4 which has the highest Tc of 3.35 K16. How the Eu valence changes and its consequence on superconductivity in the parent and doped BiS2-based superconductors still remain unresolved.

Moreover, one of the important problems in the layered systems is the inter- and intra-layer interactions. Similar to Fe-based superconductors, the interactions between superconducting BiS2 layers and blocking layers can be revealed via the nanoscale atomic displacements17,18. Hence, in order to understand the origin of superconductivity, it is critical to investigate the Eu valence and the local atomic displacements in the parent and doped Eu-containing BiS2-based superconductors.

The X-ray absorption spectroscopy (XAS), consisting of the X-ray absorption near edge spectroscopy (XANES) and extended X-ray absorption fine structure (EXAFS) spectroscopy, is an ideal technique to retrieve the substantial information of both valence transition and nanoscale atomic displacements, thus XAS has been widely applied in physics and chemistry19,20,21. For example, based on the “fingerprint effect”, Eu L3-edge XANES for EuFe2As2 presents the visually experimental evidence for the pressure-induced valence changes of Eu ions22. In addition, Bi L3-edge EXAFS were performed to probe the local atomic structure of BiS2-based systems18. In this contribution, we investigated the local structure of EuBiS2F-based system as a function of Ce-doping by XAS, providing the atomic site-selective information of valence changes and nanoscale atomic displacements.

Results

Role of Eu valence changes in the parent and Ce-doped EuBiS2F

For the Eu-containing superconductors, detailed investigations of the Eu valence change may provide valuable information on the electronic structure, which is fundamental for a better understanding of their superconductivity22,23. Figure 1a shows normalized Eu L3-edge XANES data for EuBiS2F and Eu0.5Ce0.5BiS2F. The main peak (6975 eV) and the other feature (6983 eV) in the Fig. 1a are associated respectively to Eu2+ (4f7) and Eu3+ (4f6)22.

Figure 1
figure 1

Eu L3-edge XANES spectra and curve fitting for EuBiS2F and Eu0.5Ce0.5BiS2F.

(a) Normalized Eu L3-edge XANES spectra for EuBiS2F and Eu0.5Ce0.5BiS2F; (b) curve fitting for EuBiS2F; (c) curve fitting for Eu0.5Ce0.5BiS2F. The solid black line and red open circles correspond to the experimental data and the best fit, respectively.

Full size image

Now we determine quantitatively the valence of Eu for the parent and Ce-doped EuBiS2F by fitting the XANES spectra to an arctangent step function and a Lorentzian peak for each valence state. The mean valence was determined by using a widely used method24,25:

where I2+ and I3+ is integrated intensity of peaks corresponding to Eu2+ and Eu3+ on XANES spectrum. Based on the best curve fit in Fig. 1, we estimated the mean valence of Eu ions in EuBiS2F is +2.16(1), instead of +2, demonstrating the self-electron-doping nature in parent compound without any extrinsic doping. The mean valence of Eu in Ce-doped EuBiS2F is +2.05(1), basically consistent with previous crystallographic and magnetic structure data15. Therefore, these data confirm the Eu valence change, suggesting a potential relationship between the Eu valence and superconductivity.

In Fig. 2 we focus on the normalized Ce L3-edge XANES in Eu0.5Ce0.5BiS2F, in which three main structures A, B and C can be identified. The first peak A around 5728 eV is associated to the transition from the Ce 2p core level to the vacant Ce 5d state mixed with the Ce 4f 1 final state, i.e. Ce3+ state26. On the other hand, the weak feature B around 5745 eV is a characteristic feature of layered rare-earth systems26, and its intensity is generally sensitive to the F atom order/disorder in the Eu/CeF layers. The third peak C is the so-called continuum resonance, providing the information on the local lattice structures. It should be noted that the energy difference between the characteristic Ce3+ (4f1) and Ce4+ (4f0) absorption peaks is approximately 12 eV, which is independent and is mainly determined by the Ce 2p-4f Coulomb interaction26. But in Fig. 2 we found no obvious evidence of Ce4+ feature around 5740 eV, demonstrating that the Ce valence in the Eu0.5Ce0.5BiS2F sample is essentially trivalent. Considering the valence of Eu, 50 at.% Ce-doping could cause an increment of mean valence for Eu/Ce ions, which increases from +2.16 of parent EuBiS2F to +2.53 of Ce-doped system. Consequently, additional 17% charges were induced upon Ce-doping in EuBiS2F, which is believed to be crucial for the superconductivity enhancement.

Figure 2
figure 2

Normalized Ce L3-edge XANES data for Eu0.5Ce0.5BiS2F.

Full size image

Nanoscale atomic displacements in EuBiS2F and Eu0.5Ce0.5BiS2F

As is well known, material properties are in a close relationship with its nanoscale atomic structure. Analogous to cuprates and Fe-based superconductors, Ce impurity could alter the local atomic displacements of both blocking layers and BiS2 superconducting layers. Therefore, to gain an insight into the atomic displacements induced by Ce-doping, we have undertaken detailed structural study by means of Eu and Bi L3-edge EXAFS measurements. Figures 3 and 4 display the Fourier transform (FT) magnitudes of the EXAFS oscillations providing real space information at Eu and Bi L3-edge, respectively. We have to underline that the positions of the peaks in the FT are shifted a few tenths of Å from the actual interatomic distances because of the EXAFS phase shift27. In the BiS2 layer the in-plane and out-of-plane S atoms are denoted as S1 and S2, respectively. The Eu atom is coordinated with four nearest F atoms at ~2.52 Å and four S2 atoms at ~3.04 Å. Therefore, the broad structure (R = 1.5~3.0 Å) in the FT of Eu L3-edge EXAFS corresponds to the contributions of Eu-F and Eu-S2 bonds. On the other hand, the near-neighbor of Bi atoms are one out-of-plane S2 atom at ~2.50 Å and four in-plane S1 atoms at ~2.87 Å. Therefore, the broad structure (R = 1.4~2.6 Å) in Fig. 4 contains information on the Bi-S2 and Bi-S1 bonds. Obviously, large changes in the FTs of both Eu and Bi L3-edge can be seen with Ce-doping, indicating the atomic displacements in blocking layers and also in the electronically active BiS2 layers.

Figure 3
figure 3

Fourier transform (FT) magnitudes of the Eu L3-edge EXAFS measured on EuBiS2F and Eu0.5Ce0.5BiS2F.

Models fits to the FTs are also shown as triangles. The inset shows the local coordinate atomic clusters around Eu in cross-section view.

Full size image
Figure 4
figure 4

Fourier transform (FT) magnitudes of the Bi L3-edge EXAFS measured on EuBiS2F and Eu0.5Ce0.5BiS2F.

Models fits to the FTs are also shown as triangles. The inset shows the local coordinate atomic clusters encircled around Bi in cross-section view.

Full size image

The EXAFS amplitude depends on several factors and is given by the following general equation28:

where Nj is the number of neighboring atoms at a distance Rj, is the passive electron reduction factor, fj(k, Rj) is the backscattering amplitude, λ is the photoelectron mean free path, δj(k) is the phase shift and is the correlated Debye-Waller factor.

In order to obtain quantitative results, we firstly fit the peaks of EXAFS spectra at Eu L3-edge involving contributions of four Eu-F and four Eu-S2 bonds, which were isolated from the FTs with a rectangular window. The range in k space was 3~12 Å−1 and that in R space was 1.5~3.0 Å. Considering the absorption energy at Eu L3 (6977 eV) and L2-edge (7617 eV), the maximum wave-vector k for Eu L3-edge EXAFS is up to 12 Å−1. The spatial resolution 28 is about 0.13 Å with the kmax = 12 Å−1, which is sufficient to distinguish between Eu-F and Eu-S2 bonds. For the least-squares fits, average structure measured by diffraction on EuBiS2F system13 is used as the starting model. The backscattering amplitudes and phase shift were calculated using the FEFF code29. Only the radial distances Rj and the corresponding were allowed to vary, with coordination numbers Nj fixed to the nominal values. The passive electrons reduction factor and photoelectron energy zero E0 were also fixed after fit trials on different scans. The best values for the were found to be 0.9 and fixed to this value for all the shells. The number of independent parameters which could be determined by EXAFS is limited by the number of the independent data points Nind~(2ΔkΔR)/π, where Δk and ΔR are respectively the ranges of the fit in the k and R space28. In our case, Nind is 8 (Δk = 9 Å−1, ΔR = 1.5 Å), sufficient to obtain all parameters.

As shown in Table 1, upon Ce-doping the distance of Eu-S2 bond is essentially unchanged within the errors, while the Eu-F distance becomes slightly elongated from 2.51(1) Å to 2.54(1) Å, suggesting a thicker EuF layer induced by Ce-doping. Now we resort to the bond valence sum30 of Eu (Eu-BVS) using the formula , where R0 is an empirical parameter (2.04 and 2.53 Å for Eu-F and Eu-S bonds30, respectively) and dij denotes the measured bond distances between Eu and coordinate anions. Here, eight coordinate atoms (four F and four S2 atoms) were considered. Considering the bondlengths achieved from EXAFS fitting, the Eu-BVS value are +2.14(2) and +2.07(2) in EuBiS2F and Eu0.5Ce0.5BiS2F respectively, essentially in agreement with the valence information retrieved from our XANES data.

Table 1 The fitting result at Eu and Bi L3-edge EXAFS upon Ce-doping.
Full size table

Meanwhile, Ce-doping also affects the local atomic structure of superconducting BiS2 layers. In Fig. 4 the broad peaks at Bi L3-edge were modelled by two shells, involving contributions of one Bi-S2 and four Bi-S1 bonds, which were isolated from the FTs with a rectangular window. The range in k space was 3~15 Å−1 and that in R space was 1.4~2.6 Å. Spatial resolution is about 0.10 Å, while the number of independent parameters Nind is 9, sufficient to distinguish between Bi-S2 and Bi-S1 bonds and obtain all parameters.

Recently, it was reported that the enhancement of in-plane chemical pressure is responsible for the superconductivity in BiS2-based compounds31. Upon Ce-doping the sharp contraction of the in-plane Bi-S1 bond (, i.e. a higher in-plane chemical pressure) results in an enhancement of the packing density of Bi and S1 ions within the superconducting plane, which would enhance the hybridization of Bi 6px/6py-S 3p orbitals and result in an increase of Tc. In addition, the fact that in-plane Bi-S1 bondlength decreases with Ce-doping, while the Bi-Bi distance (i.e. a-axis, from 4.0508(1) to 4.0697(1) Å) showing a small increase, indicating the puckering and large in-plane disorder of the Bi-S1 layer. Further information on the atomic disorder can be provided by the correlated Debye-Waller factors (σ2), measuring the mean square relative displacement (MSRD) of the photoabsorber-backscatterer pairs32. Data point out that the σ2 for the in-plane Bi-S1 distance in EuBiS2F is anomalously large, demonstrating a large configurational disorder within the Bi-S1 plane. Here, it is worth recalling that the large configurational disorder in BiS2 plane is quite common in BiS2-based superconductors, consistent with the anomalously large diffraction thermal factor of in-plane S1 atom33. Upon Ce-doping, the σ2 for the Bi-S1 bond reduces by 25% with respect to the parent compound, demonstrating that puckering of the Bi-S1 layer seems to be getting reduced; that is to say, a flatter Bi-S1 plane is also responsible for a higher Tc. By contrast, the σ2 for the Bi-S2 bond is quite small and remains unchanged upon Ce-doping, indicating robust Bi 6pz-S 3p hybridizations. All these results suggest that Ce-doping can effectively tune the atomic displacements of BiS2 superconducting layers.

Discussion

The Ce-doping effect on the valence state and local atomic displacement in the EuBiS2F system is investigated by using XAS measurements. First of all, the valence of Eu ions in EuBiS2F is estimated to be about +2.16(1), demonstrating the self-electron-doping nature without any extrinsic doping. Upon 50 at.% Ce-doping, the mean valence of Eu reduces to +2.05(1) and that of Ce ions are essentially trivalent. The main effect of Ce-doping is to provide additional 17% electrons into the system, beneficial for the superconductivity enhancement. The local atomic displacements can be revealed by Eu and Bi L3-edge EXAFS: 1) the in-plane Bi-S1 distance is characterized by a large configurational disorder in EuBiS2F-based system, which is quite common in BiS2-based superconductors; 2) both the shortening of the in-plane Bi-S1 bond (i.e. a higher in-plane chemical pressure) and the flatter Bi-S1 plane are responsible for an enhancement of superconductivity.

In summary, we established a relationship between Ce-doping and the Tc enhancement in EuBiS2F-based superconductors, in terms of valence changes and nanoscale atomic displacements. The new findings are promising for providing insights on the interplay of charge, local structure and superconductivity.

Methods

Polycrystalline compounds of EuBiS2F and Eu0.5Ce0.5BiS2F were synthesized by solid-state reaction method13,15. The samples were well characterized for their phase purity, superconducting and other properties prior to the XAS measurements. The XAS spectra were collected at the BL-14W1 beamline of Shanghai Synchrotron Radiation Facility (SSRF). The storage ring was working at electron energy of 3.5 GeV, and the maximum stored current was about 250 mA. The energy of the incident energy was tuned by scanning a Si (111) double crystal monochromator with energy resolution about 10−4. The XAS spectra at Ce L3-edge, Eu L3-edge, and Bi L3-edge were collected with several scans in transmission mode at room temperature. Data reduction was performed using the IFEFFIT program package34.

Additional Information

How to cite this article: Cheng, J. et al. Role of valence changes and nanoscale atomic displacements in BiS2-based superconductors. Sci. Rep. 6, 37394; doi: 10.1038/srep37394 (2016).

Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.