Introduction

Since the discovery of cuprate superconductors in 1986, one of the still open and challenging problems in condensed matter physics is the origin of the superconductivity in high-temperature superconductors (HTSCs)1. Two decades of efforts based on thousands of experimental and theoretical researches pointed out that electron-electron interaction (EEI) rather than the electron-phonon interaction (EPI) plays the key role to trigger the superconductivity in HTSCs materials2,3. However, iron-based superconductors the new family of HTSCs discovered in 2008 does not show an agreement with this assumption4,5,6. Indeed, a clear iron isotope effect on Tc was experimentally observed7,8,9,10. Chen et al.7 firstly reported a large and positive iron isotope effect (Fe-IE) on the Tc of both LnFeAs(O, F) and (Ba, K)Fe2As2 superconductors. Khasanov et al.10 found a larger Fe-IE exponent, well beyond the BCS value. Finally Shirage et al.8,9 observed an inverse iron isotope effect in the (Ba, K)Fe2As2 superconductor. Recently Yanagisawa et al.11 presented a model explaining the sign inversion of the IE exponent. In his model, the (Ba, K)Fe2As2system is characterized by a multiband superconductivity originating by two different paring channels: a phononic contribution and an antiferromagnetic (AF) fluctuation mechanism. However, his approach has been questioned by Bussmann-Holder and Keller12 that claimed that an inversion of the Fe-IE exponent could not occur. More recently, Khasanov et al.13 tried solving the discrepancy in the Fe-IE measurements by decomposing the exponent in two terms: one associated to structural changes and one intrinsic. They also pointed out that the Fe isotope substitution may induce small structural changes which, in turn, may affect Tc. Actually, the controversy still on points out that:

  1. 1

    lattice vibrations (i.e., phonons) play an important role in the superconductive mechanism(s) of iron-based superconductor materials;

  2. 2

    the understanding of the IE in these iron-based materials remains a challenging issue to clarify.

Therefore, additional and more accurate experimental methods to measure the IE, probing not only macroscopic averaged properties such as those based on Tc determinations, but also microscopic information of the system are required.

Here we present an investigation of the Fe-IE in the (Ba, K)Fe2As2 superconductor system by temperature-dependent extended x-ray absorption spectroscopy (T-EXAFS). Due to the correlation between the absorber and the backscatter through the emitted photoelectrons excited by x-rays, EXAFS probes the local structure around the selected photoabsorber, such as the number of neighboring atoms N, the average bond length between absorbing atoms and neighboring atoms <r> and the variance σ2 14,15. The Debye-Waller factor σ2 corresponds to the parallel mean square relative displacement (MSRD) among two atoms and it is the sum of two contributions:

The first term is temperature independent and associated to the static disorder, e.g., to the dispersion of bond lengths neglecting the vibrational dynamics. The second (dynamical) contribution is due to thermal vibrations and can be correlated to the vibrational spectrum of the system. The latter is usually well described by the Einstein model16

where μ is the reduced mass of the atom pair, T is the temperature, k is Boltzmann's constant and is the Plank's constant. In the Eq. 2, is the Einstein temperature and ωE is the corresponding Einstein frequency. The σvib2 contribution is different from zero also at 0 K.

Measurements of T-EXAFS may allow, via the analysis of the Debye-Waller factor, to separate the static and the thermal disorder contributions present in a sample. The first term is relative to the local structure and affects the electronic and the magnetic behavior of the system, while the second depends only by the local vibration behavior (i.e., phonons). By using the T-EXAFS technique, we may evaluate at the microscopic scale the contribution of the Fe-IE on both the static structure and the lattice dynamics in iron-based SCs. Because EXAFS is an element-specific technique, in these systems we may independently achieve information of the individual atomic pairs such as the Fe-As and the Fe-Fe pairs. Powder samples can be also used in an EXAFS experiment and single crystals are not necessary to accurately determine the mean-square relative displacement (MSRD).

Both Fe and As K-edge extended x-ray absorption fine structure spectra of the Ba0.6K0.4Fe2As2 system by replacing 56Fe (56Fe-sample) with the isotope 54Fe (54Fe-sample) were measured vs. temperature down to 10 K. After an accurate analysis and comparison of EXAFS spectra of the two isotopic substitutions we point out that it does not induce detectable change of the average distances of both Fe-As and Fe-Fe atomic pairs while slightly modulates the Fe-As and Fe-Fe pair distributions. Moreover, it affects the local vibration of both Fe-As and Fe-Fe bonds.

Results

Fig. 1(a) compares As K-edge EXAFS data at 10 K and at room temperature (RT) showing a clear damping of the EXAFS oscillations vs. temperature. This behavior of the local structure is also clear looking at the comparison of the Fourier transforms (FTs) of the EXAFS spectra in Fig. 2(b). In this iron-based superconductor with the 122-structure no anomalies are detected at low temperatures, especially around Tc. In addition to our investigation, to our knowledge other two manuscripts published temperature-dependent EXAFS measurements on iron-based superconductors. They are from the Oyanagi and Tyson teams. Both investigated the (1111) iron-based superconductors LaFeAsO1−xFx obtaining different results. Oyanagi et al. found an anomalous behavior across TSDW and Tc17. No such fine behavior has been observed in the data of Tyson18. We are convinced that in addition to always possible systematic errors or limited temperature points, the differences observed can be associated to the different F concentration: x = 0.07 and x = 0.11. Actually, the temperature-dependence of Fe-As and Fe-Fe MSRD of our system is similar to that observed by Tyson et al., but our work is investigating a different type of iron-based superconductors: the (Ba, K)Fe2As2 system belonging to the (122) family whose structure is well different from pnictides of the (1111) family. On the contrary, as we will discuss in the next, the iron isotope effect can be clearly seen looking at the magnitude of the FTs, which is correlated with the σ2 term, i.e., the EXAFS Debye-Waller (DW) factor.

Figure 1
figure 1

Temperature-dependent EXAFS data for the Fe isotope substituted samples.

(a) Representative experimental EXAFS oscillations at the As (upper panel) and Fe (lower panel) K-edges of the 54Fe-sample. (b) FTs of the As K-edge EXAFS data measured cooling the sample down to 10 K. (c) Comparison between the first-shell peak of the FTs of the 54Fe- and 56Fe-samples at 10 K. (d) The same of (c) at RT.

Figure 2
figure 2

EXAFS fits and result of the bond distances vs. T.

(a) Experimental (black circles) and fit (red lines) EXAFS oscillations of As-Fe (upper curve) and Fe-Fe (lower panel) shells, respectively, for the 54Fe-sample at RT. (b) Temperature dependence of Fe-As bond distances and Fe-Fe bond distances for the 54Fe- and the \\Fe-samples, respectively.

In Fig. 1(c) and (d) we compare the first-shell peak (As-Fe pair) of the FTs of the 54Fe- and 56Fe-samples at 10 K and at RT, respectively. The position of the first-shell peak, i.e., the As-Fe pair in the FTs, is constant in the two samples and the same occurs for the Fe-Fe pair (panel d). Data point out that the Fe-IE does not affect the average distances of both Fe-As and Fe-Fe atomic pairs in the time-domain of the EXAFS experiments. A difference in the FTs associated to the isotopic substitution appears only looking at the magnitude of the first-shell peak. As it is well know, for a given crystal structure the magnitude of the FT peak depends by the σ2 term. As shown in Fig. 1(c) a difference in the FT magnitude for the As-Fe pair is evident at the lowest temperature (10 K - panel c). It gradually reduces with the temperature and at RT the two curves almost overlap (panel d). To obtain a quantitative dependence of σ2 vs. T, we extracted the DW of both As-Fe and Fe-Fe pairs by fitting EXAFS signals vs. T from 10 K to RT at both As and Fe K-edges, respectively. For the As K-edge, we considered the EXAFS signals of the nearest As-Fe bond. For the Fe K-edge, both nearest neighboring (Fe-As) and next-nearest neighboring (Fe-Fe) were filtered out. We have to underline here that the As-Fe correlation is identical to the Fe-As correlation in an EXAFS process. Therefore, in order to obtain exact information on the Fe-Fe bond we fix the distance and the DW factor of the Fe-As bond using parameters obtained by the As K-edge EXAFS fit during the analysis at the Fe K-edge EXAFS. Fig. 2a shows the quality of the EXAFS fit of the As K-edge EXAFS oscillations at RT. In Fig. 2b we compare the average distances of both Fe-As and Fe-Fe bonds of 54Fe- and 56Fe-samples, respectively, in the whole temperature range (10 ~ 300 K). Data are consistent with what showed in Fig 1c and 1d. We may emphasize that the result is different from the report of Khasanov et al.10 on the FeSe1−x system that investigated the temperature dependence of the lattice structure by Neutron Powder Diffraction (NPD). This work also shows a large Fe-IE (αFe = 0.81) by means of SQUID measurements. As claimed by Khasanov et al., it is the tiny modulation of the structural parameters caused by the isotope substitution that induces the large Fe-IE, beyond the framework of the BCS theory. However, these findings indicate that a Fe isotope effect on Tc is really present and related with the atomic or lattice structure in the Fe-based superconductors of the (11) family. In the superconductor (Ba, K)Fe2As2 [(122) family] we detected an Fe-IE only on the atomic distributions and not on bond distances. Moreover, data points out that the isotope substitution has no effect on lattice distances in these samples of the (122) family. As a consequence, it represents an ideal model to characterize the isotope effect on the vibrational contributions, a fundamental issue to understand the electron-phonon interaction in a superconductor.

In Fig. 3 we also show the dependence of the DW factors of the As-Fe and the Fe-Fe bonds vs. temperature for both the 54Fe- and 56Fe-samples. A clear “isotopic” contribution appears at low temperature while at high temperature, the dependence of the DW factor approaches the classical linear behavior.

Figure 3
figure 3

DW factors σ2 for the bond of the 54Fe- and the 56Fe-samples vs. T.

(a) As-Fe bond; (b) Fe-Fe bond. Symbols are experimental data extracted from the EXAFS fits while solid lines are simulations based on the Einstein model. Bars refer to statistical errors.

As outlined above, the σ2 term contains two contributions: the static disorder and the thermal disorder. In order to evaluate the contribution of the static disorder and of the vibrational frequency of the As-Fe and Fe-Fe bonds of isotopic substituted samples, we used the correlation Einstein model to fit the σ2 function vs. temperature:

In Eq. 3 is the static contribution to the bond and the quality of our simulations is shown by solid lines in the same figure. The corresponding values of the static contribution and of the vibrational frequency of the two samples are summarized in table 1 for both As-Fe and Fe-Fe bonds. Data show that the isotopic substitution gives rise to a weak modulation of the distribution of the interatomic distances of both Fe-As and Fe-Fe pairs. Moreover, the Fe-IE affects slightly more the static structural disorder of the Fe-Fe pair (it changes from 0.0056 to 0.0053 Å2) than that of the As-Fe (or Fe-As) pair that decreases from 0.0025 to 0.0023 Å2. In addition, the static disorder of the Fe-Fe pair is much larger than that of the Fe-As pair. Bonds with a larger value (i.e., the Fe-Fe pair) and modulation due to the isotopic substitution, point out that these bonds are more flexible. This result is in agreement with other experimental data19 and because decreasing temperature down to the structural transition temperature (Ts) the Fe-Fe bond will undergo to a splitting20, data are also consistent with the occurrence of a structural phase transition in the iron layer of the BaFe2As2 system. Also by using T-EXAFS technique, in the K0.8Fe1.6Se2 [122-chalcogenide] a similar lattice property has been detected. Also this work points out that the Fe-Fe bond is more flexible than the Fe-As bond21. The iron isotope effect we observe on is due to the static structural modulation and corresponds to electronic and magnetic properties of the iron-based SCs.

Table 1 Values of local static distortions and Einstein temperatures associated to the As-Fe and Fe-Fe bonds in the 54Fe- and 56Fe-samples

A second important result we obtained by the analysis of the T-EXAFS experiments is that the Einstein temperature (TE) that is a probe of the vibrational frequency of the system, exhibits a shift for both the As-Fe and Fe-Fe bonds induced by the iron isotope substitution. The TE of the As-Fe bond changes about 1.3% from 331.5 to 327.1 K after the substitution with 54Fe in the Ba0.6K0.4Fe2As2. In the Fe-Fe bond, TE shifts ~1.5% from 169.0 to 166.5 K. Both Fe-Fe and Fe-As vibrations have then a similar, but not negligible contribution to the superconductive mechanism of iron-based SCs. Although the detected changes are small, similar changes of TE occur for both Fe-Fe and Fe-As bonds. Actually, the Einstein temperature TE can be associated to an effective force constant k0 normalized to the reduced mass μ:

Actually, because the Einstein temperature is lower for the Fe-Fe bond, the latter is supposed to be softer than the Fe-As one. This experimental result points out a greater flexibility of the iron layers in the (Ba, K)Fe2As2 system. To evaluate the contribution of the isotope effect on the local vibration of the Fe-Fe and As-Fe correlations, we introduced the coefficient α

very similar to the definition of the Tc isotope effect coefficient22. Here TE is the Einstein temperature obtained by the EXAFS experiments, regarding both Fe-Fe and Fe-As bonds (Table 1) while M is the atomic mass of the isotopic substitution. Using the coefficient α we may obtain an independent evaluation of the iron-isotope-effect for both As-Fe and Fe-Fe bonds. Using values listed in table 1 we got values of 0.37 and 0.41 for the As-Fe and Fe-Fe bonds, respectively, very close to the Tc iron-isotope-effect coefficient of 0.37 reported by the Chen's group7. Our analysis supports the hypothesis that the electron-phonon interaction plays a role in the superconductive mechanism of iron-based superconductors and that both Fe-Fe and Fe-As vibrations contribute to the superconducting mechanism.

Because Fe-Fe bonds lie in the superconducting layers (a-b plane) while Fe-As bonds are oriented along the c axis, our result confirms in the iron-based superconductors the presence of a weak bulk iron isotope effect. In addition, it points out the presence in iron-based superconductors of a not negligible 3D electron-phonon interaction as shown in Fig. 4. This result can be also used to interpret the nearly isotropic superconducting properties of iron-based superconductors23, systems well different from layered cuprate superconductors.

Figure 4
figure 4

Layout of Fe-Fe and Fe-As vibrations in the iron-based superconductors.

Fe-Fe bonds are in the superconducting layers and their vibrations attract electrons in the c direction, while Fe-As bonds are oriented along the c axis and their vibrations affect electrons in the a-b layers.

Discussion

An accurate EXAFS analysis of the (Ba, K)Fe2As2 system vs. temperature shows that an iron isotope effect correlated to the static structural distribution and with local vibrations can be detected. Changes of the static structural distribution may occur via changes of the inter-atomic distances and of the atomic distribution (or of the disorder). The Fe-IE does not induce appreciable changes in the inter-atomic distances while the atomic distribution is slightly affected. At present, it remains however extremely difficult identify a clear correlation among local disorder contributions (present in a powder sample) and superconducting mechanisms. In addition, we clearly showed the existence of a Fe-IE contribution to local vibrational term for both Fe-As and Fe-Fe bonds. Their isotopic exponents are in agreement with previous experiments7 looking at the iron isotope effect in these systems. This result represents the first experimental link between the Fe-IE of Tc experiments and the superconducting mechanism at a microscopic level.

Methods

Isotopically substituted polycrystalline samples were synthesized by the solid state reaction method described in detail in Ref. 7. Previous investigation of this superconductor showed a large iron isotope effect on the Tc7. As and Fe K-edge absorption spectra were collected using a Si (111) double crystal monochromator at the XAFS station of the 1W1B beamline of the Beijing Synchrotron Radiation Facility (BSRF). The storage ring was working at 2.5 GeV with an electron current decreasing from 240 to 160 mA within about 8 hours. Samples were ground into fine powders and brushed onto tapes that were stacked together to give approximately one x-ray-absorption lengths at their corresponding metal edges. Samples were cooled by a closed-cycle helium refrigerator with an error bar on the temperature determination of about 0.1 K. Data were processed and analyzed using both UWXAFS24 and FEFF 8.0 packages25,26 to generate scattering amplitudes and phase shifts starting from available XRD refinements (see supplementary materials of Ref. 7).