Superconductivity induced by Ni doping in BaFe$_2$As$_2$

A series of 122 phase BaFe$_{2-x}$Ni$_x$As$_2$ ($x$ = 0, 0.055, 0.096, 0.18, 0.23) single crystals were grown by self flux method and a dome-like Ni doping dependence of superconducting transition temperature is discovered. The transition temperature $T_c^{on}$ reaches a maximum of 20.5 K at $x$ = 0.096, and it drops to below 4 K as $x$ $\geq$ 0.23. The negative thermopower in the normal state indicates that electron-like charge carrier indeed dominates in this system. This Ni-doped system provides another example of superconductivity induced by electron doping in the 122 phase.


Introduction
Following the discovery of superconductivity at T c of 26 K in F-doped LaFeAsO [1], a so-called 1111 phase family of high-T c superconductors with ZrCuSiAs-type structure, i.e., LnFeAsO 1−x F x [2,3,4,5,6,7,8], LnFeAsO 1−δ [9] and Ln 1−x Th x FeAsO [10,11] (Ln = La, Ce, Pr, Nd, Sm, Gd, Tb) have been reported. Similar to the cuprate high temperature superconductors, the (Fe 2 As 2 )-layers are conducting layers and essential to the occurrence of superconductivity, while the (R 2 O 2 )-layers inject charge carriers to the former via chemical doping and also retains structural integrity of the (Fe 2 As 2 )-layers. Furthermore, a 122 phase family of high-T c superconductors, AFe 2 As 2 (A = Ca, Sr and Ba) [12,13,14,15,16] with the ThCr 2 Si 2 -type structure (space group I4/mmm), which possess the same (Fe 2 As 2 )-layers as LnFeAsO but separated by simple A-layers, have been discovered. Superconductivity in a so-called 111 phase compound LiFeAs with a similar structure has also been reported [17]. In the 1111 phase family, superconductivity is mostly induced by electron doping, i.e., partial substitution of O by F, Ln by Th, or O-vacancy though superconductivity with T c of 25 K has indeed been reported in hole-doped La 1−x Sr x FeAsO [18]. In contrast, superconductivity is mostly induced by hole doping in 122 phase, i.e., partial substitution of Ca, Ba, or Sr by K or Na etc. No superconductivity has been observed by means of A-site electron doping, such as the partial substitution of divalent ions (Ba 2+ or Sr 2+ ) by trivalent ions (La 3+ ) [15].
In contrast to high-T c cuprates, superconductivity can also be induced by partial substitution of Fe in the conducting layers by other transition metal elements like Co [19,20] and Ni [21] in the 1111 phase, and the measurements of Hall effect and thermopower indicate that the electron-type charge carriers dominate in the Co doped case. For the 122 phase, superconductivity with T c as high as 25 K has also been observed in Co-doped BaFe 2−x Co x As 2 [22] and SrFe 2−x Co x As 2 [23] systems. As implied by the negative thermopower value in the normal state [15,20,24], the Co doping appears to donate the extra electrons into Fe layers as itinerant charge carriers, and thus induce superconductivity for either 1111 phase or 122 phase. In this paper, we report that superconductivity is induced in the Ni-doped BaFe 2−x Ni x As 2 single crystals. A dome-like Ni doping dependence of T c is established and the highest T on c of 20.5 K is realized at the optimal doping level x = 0.1. Compared with the Co doped 122 phase, the superconducting window is narrower and the optimal Ni doping concentration is only about half. This Ni-doped system provides another example of superconductivity induced by electron doping in the 122 phase.

Experimental
Single crystals with nominal formula BaFe 2−x Ni x As 2 (x = 0, 0.05, 0.1, 0.16, 0.2) were prepared by self flux method [22]. All the starting materials, Ba rods, Fe powders, Ni powders, As pieces, are of high purity(≥99.99%). First, FeAs and NiAs binaries were prepared by reacting Fe/Ni powder and As powder in evacuated silicon tube. They were heated slowly to 873 K, kept for 10 hours, then cooled to room temperature. A ratio of Ba: FeAs/NiAs = 1:5 was used and the extra FeAs materials acted as the flux. The mixtures of Ba and FeAs/NiAs were loaded in a finely designed thin corundum crucibles. These processes were all done in a glove box filled with high purity argon. Then the crucibles were sealed in evacuated quartz tubes. Finally the quartz tubes filled with the starting materials were slowly heated to 973 K, dwelled for 5 hours and then heated to 1453 K, dwelled for 10 hours. Then the temperature decreased to 1363 K at a rate of -2 K/hour, followed by decanting the quartz tube at 1363 K, and finally cooled down to room temperature slowly. Single crystals with a diameter of about 5 mm were obtained.
X-ray diffraction (XRD) was performed at room temperature using a D/Max-rA diffractometer with Cu-K α radiation and a graphite monochromator. Lattice parameters were refined by a least-squares fit. Chemical analysis by Energy Dispersive of X-ray (EDX) microanalysis was performed on an EDAX GENESIS 4000 X-Ray analysis system affiliated to a Scanning Electron Microscope (SEM, model SIRION). For each doped single crystal, EDX microanalysis was performed at more than three different spots to check the distribution of Ni content. The electrical resistivity was measured using a standard four-probe method. The temperature dependence of dc magnetization was measured on a Quantum Design Magnetic Property Measurement System (MPMS-5). The thermopower was measured by a steady-state technique.

Results and discussion
The XRD patterns of BaFe 2−x Ni x As 2 crystals are shown in Fig.1. Only (00l) reflections appear, indicating that the c-axis is perpendicular to the cleaved surface. The (00l) refections with only even l were observed because the system belongs to the bodycentered space group. The c-axis lattice constant was calculated as 1.303 nm for the undoped parent compound BaFe 2 As 2 , consistent with previous reports [25]. The inset of Fig.1 shows the variation of the c-axis lattice constant with nominal Ni content x. The c-axis lattice constant decreases monotonically with increasing x. The EDX measurements show that the actual average Ni content (x) is slightly different from the nominal composition, but the variation of Ni content is less than 5% in each sample except the sample with nominal x of 0.05, which indicates that the distribution of Ni doping is quite uniform in these samples. A representative EDX spectrum for a single crystal with nominal x of 0.1 is shown in Fig.2 and the inset shows a SEM photo of this sample. According to the EDX reports, the average x value is 0.055, 0.096, 0.18, and 0.23 for the samples with the nominal x of 0.05, 0.1, 0.16, and 0.2, respectively. Fig. 3 shows the temperature dependence of in-plane resistivity. Similar to the previous report [15,25], there is an anomalous decrease in the resistivity of the parent compound around T an of 140 K, which is associated with a structural phase transition. Neutron studies have confirmed that the spin-density wave order occurs simultaneously with the structural phase transition in 122 phase [26,27,28]. The resistivity anomaly is shifted to a lower temperature, about 90 K, for the x = 0.055 sample. However, the resistivity exhibits a sharp increase instead of a decrease around T an . Such anomalous increase in the resistivity has been also observed in other doped 122 phase [23]. With further decreasing temperature, a superconducting transition occurs below T on c (defined as the onset point in the resistive transition) of 12.5 K for the same x = 0.055 sample, but zero resistivity can not be reached even for T as low as 4 K. It is hard to distinguish whether there is a microscopic phase separation (inhomogeneous distribution of Ni content) or there is a co-existence of SDW order and superconducting order according to current measurements. With increasing Ni content, the resistivity anomaly disappears. At the optimal doping level x = 0.096, T on c reaches a maximum of 20.5 K, T mid c is 20.2 K, and the transition width △T c is less than 1.0 K, suggesting a sharp superconducting transition. The temperature dependence of susceptibility as shown in Fig.2 (b) also indicates very sharp superconducting transitions for x = 0.096 and 0.18. However the superconducting transition in magnetic susceptibility of the x = 0.055 sample is rather broad, which means that the distribution of Ni dopants could be very inhomogeneous in this sample. Fig.4 shows the variation of T on c and T mid c with Ni content x. In the superconducting window (0.055 ≤ x ≤ 0.23), a dome-like T c (x) curve is roughly established, similar to that of Ni doped 1111 phase. It should be noted that the optimal doping level x = 0.096 corresponds to about 5% of Fe substituted by Ni in 122 phase, which is very close to the optimal Ni doping level in 1111 phase (4% of Fe substituted by Ni) [21]. However, compared to Co-doped 1111 and 122 phases, the superconducting window in Ni-doped system is much narrower, and the optimal Ni doping concentration (about 5% Ni content) is only about half of that of Co-doped systems [20,24]. Compared to Fe 2+ ion, Co 2+ (3d 7 ) has one more 3d electron, but Ni 2+ (3d 8 ) has two more 3d electrons. Then it is expected that each Ni dopant induces two extra itinerant electrons while each Co dopant only induces one extra itinerant electron. Thus the fact that the optimal doing content of Ni is only about half of that of Co can be understood. Actually both Ni doped and Co doped systems show maximum T c at the same effective doping level. A universal doping (charge carrier concentration) dependence of T c has been established in high-T c superconducting cuprates. Our result implies that there might exists a universal doping dependence of T c in iron-based arsenide superconductors. For the higher Ni-doping levels (x ≥ 0.23), superconductivity is no longer observed for T > 4 K, but the resistivity becomes even more metallic. It should be noted that the other end member, BaNi 2 As 2 , is a superconductor with T c of only about 0.7 K [29]. However, the first principles calculations suggested that BaNi 2 As 2 could be a conventional phononmediated superconductor [30]. Fig. 5 plots the thermopower (S) as a function of temperature. All the samples show negative thermopower, which means that electron-type charge carriers dominate. For the undoped parent compound (x = 0), thermopower exhibits an anomalous enhancement just below the structural phase transition temperature T on of about 140 K. Similar behavior has been observed in undoped parent compounds such as LaFeAsO [31], TbFeAsO [11] and SmFeAsO [24]. With Ni doping, this anomaly is suppressed quickly. For x = 0.055, only a slight enhancement can be observed below about 80 K (indicated by the arrow in Fig.4). Meanwhile the absolute value of normal state thermopower (|S|) increases remarkably with increasing Ni content, and reaches a maximum at the optimal doping level x = 0.096, then decreases with further Ni doping. It has been suggested that there is a correlation between the normal state thermopower and T c according to the studies on the thermopower of Co-doped SmFeAsO system [24]. In the Ni-doped 122 phase, such a correlation should also exist although the absolute value of thermopower is much smaller than that of F-doped 1111 phase and Co-doped 1111 and 122 phases.

Conclusion
In conclusion, a series of Ni-doped 122 phase single crystals BaFe 2−x Ni x As 2 (x = 0, 0.055, 0.096, 0.18, 0.23) were successfully synthesized by self flux method and superconductivity with T c as high as 20 K is observed. The thermopower is negative in the normal state, suggesting that the electron-like charge carrier dominates. This Ni-doped system provides another example of superconductivity induced by electron doping in the 122 phase.     (Color Online) Temperature dependence of thermopower (S) for BaFe 2−x Ni x As 2 single crystals. The arrows indicate the anomaly in thermopower associated with structural phase transition and/or SDW ordering.