Superconductivity in Th3Ni5C5

The existence of a new metallic carbide of composition Th3Ni5C5 was reported in the literature in 1991. This compound is a new orthorhombic prototype structure. In this work we report a reinvestigation of the synthesis of this material and we find that the Th3Ni5C5 compound is a new bulk superconducting material. Despite the high concentration of Ni in this compound, we find bulk superconductivity with superconducting critical temperature of Tc = 5.0 K and an upper critical field of μoHc2 = 5.8 T. Details of the superconducting state with specific heat, magnetization, and resistivity measurements are discussed.


Introduction
Binary metal carbides have long been known to exist. Often the structures are binary composed of two elements carbon and T, RE, or A, where T designates a transition metal, RE a rare-earth metal, and A an alkaline-earth metal. The carboncarbon bond is a way of classifying these materials. CaC 2 [1] or UC 2 [2], ThC 2 [3] and Pu 2 C 3 [4] type structures belong to the first class. A common feature of this structure type is that there are distinct, strongly bonded C 2 pairs isolated from each other in the structures. The short contact between dimer and metal atoms also gives rise to strong metalcarbon interactions. Many members of this family exhibit interesting electric and magnetic properties [5][6][7][8][9][10]. Ternary carbides have also been receiving great attention, due to their potential importance in nuclear technology and in making permanent magnets. For example, many investigations on conductivity and magnetic properties have been made on R 2 Fe 14 C systems (R = Gd, Dy, Er, and Lu) [11,12]. It has been reported that Er 2 FeC 4 is paramagnetic at room temperature and Y 2 FeC 4 becomes superconducting at T c ∼ 3.6 K [13]. Like binary systems the ternary carbides can be classified into categories containing bonded carbon pairs and those containing isolated C atoms [14,15]. Solid state ternary transition metal carbides containing carbon, a transition metal, and a highly electropositive multivalent metal such as (Ln), Sc, Y, or Th, exhibit a number of structural features resembling those in metal carbonyls and other transition metal derivatives of π-acceptor hydrocarbon ligands. The complete ionization of the electropositive metal to the stable ions Ln 3+ , Y 3+ , or Th 4+ leads to a negatively charged transition metal-carbon subnetwork, which may be considered to be an organometallic net. Using this description, these compounds can be seen as negatively charged organometallic polymers embedded in a matrix of positive ions [16]. In many of these ternary compounds the transition metal atom can be assigned a low formal oxidation state reminiscent of the metal oxidation state in metal carbonyls and metalolefin complexes. Indeed, many of the ternary transition metal carbides contain C 2 structural units with carboncarbon distance inside 1.32-1.47Å intervals, suggestive of carbon-carbon double bonds. These structural units may be regarded as being derived from C 4− 2 anions obtained by the complete deprotonation of ethylene. To our knowledge the first nickel-based ternary carbide superconductor is the LaNiC 2 compound, which exhibits a superconducting critical temperature close to 2.7 K [17].
Moss [18] reported the discovery of the two new compounds Th 2 NiC 2 and Th 3 Ni 5 C 5 . Th 3 Ni 5 C 5 crystallizes with a new orthorhombic structure type (space group Cmcm) with the lattice constants a = 13.92Å, b = 7.14Å, and c = 7.04Å, and Z = 4, where Z means the number of chemical formula per unit cell. Th 3 Ni 5 C 5 compound contains twodimensionally infinite nickel-carbon sheets, while in Th 2 NiC 2 the NiC 2 units are separated from each other. Th 3 Ni 5 C 5 contains two C 2 pairs per formula unit, with carbon-carbon distance of about 1.37Å. The Th 2 NiC 2 compound crystallizes in the tetragonal symmetry with space group I 4/mmm and prototype structure Na 2 HgO 2 , with lattice parameters a = 3.75Å and c = 12.35Å. In Moss's paper the author concluded that both compounds are Pauli paramagnetic. However, in this work we find that Th 3 Ni 5 C 5 is a bulk superconductor, with superconducting critical temperature close to 5.0 K as substantiated by heat capacity, resistance, and magnetic measurements.

Experimental procedure
The samples were prepared from a stoichiometric mixture of Ni, graphite, and Th pieces (high purity). The constituent elements were placed together and melted in a Zr gettered arc furnace on a water-cooled Cu hearth under high purity argon. The sample was remelted five times to ensure good homogeneity. Due to the low vapor pressure of these constituent elements at melting temperature, the weight losses during arc melting were negligible (<0.5%). Some samples were annealed at 900 • C for two days and then quenched in liquid nitrogen. A microcomputer controlled diffractometer equipped with a copper target for Cu Kα (λ = 1.540 56Å) radiation was used to get the powder x-ray diffraction patterns. The lattice parameters were determined by using the PowderCell software [19]. Magnetic data were obtained using a commercial VSM-SQUID by Quantum Design. The temperature dependence was obtained using a zero field cooling (ZFC) and field cooling (FC) process, using applied magnetic field at 10 Oe. After both ZFC and FC processes, the M versus H measurement was made at 1.8 K. Electrical resistivity measurements were made between 1.8 and 300 K using a conventional four-probe method. The samples were of irregular shape and fine gold wires were spotwelded to the sample and served as the voltage and current leads. These measurements were made with and without an applied magnetic field in order to estimate the upper critical field in a physical property measurement system (PPMS) machine. The specific heat of a piece cut from the sample was measured in the range of 0.4-10 K with a He 3 calorimeter in PPMS (Quantum Design) using the relaxation method. These measurements were carried out in applied magnetic field between 0 T B 6.0 T. Figure 1 shows a comparison between experimental and simulated diffraction patterns. The simulation was carried by PowderCell software. The Miller indices have been omitted in order to avoid confusion when viewing the figure, due to the large number of peaks. There is excellent agreement between the two results, with an exception for minority peaks that can be indexed as a ThC 2 impurity, which is indicated by the * symbol. In the inset of this figure is shown the unit cell schematic of this compound. The Th atom is represented by a red sphere, Ni atom by a blue sphere and the black spheres represent carbon atoms. We do not observe any significant difference between as-cast and annealed samples. The refinement of the lattice parameter yields a = 13.92Å, b = 7.14Å, and c = 7.04Å in excellent agreement with results reported in the literature [18]. The magnetization as a function of temperature in ZFC and FC regimes reveals a superconducting transition close to 5.0 K, as shown in figure 2. The difference between ZFC and FC strongly suggests type II superconductivity. The Meissner flux expulsion (FC) is about 7% of the diamagnetic flux expulsion, a characteristic of relatively strong pinning. This is confirmed by the M versus H data shown in the inset of figure 2. About 85% is the superconducting fraction estimated at 1.8 K, calculated from the linear behavior on the M versus H curve, again strongly suggesting bulk superconductivity. No significant variation of the superconducting behavior was observed in the as-cast and annealed samples, consistent with results obtained for xray diffraction. The extrapolation of the linear behavior in M versus H indicates a lower critical field at 1.8 K close to 80 Oe. This is a rough estimate because the demagnetization  Resistance data between 1.8 and 300 K for the polycrystalline Th 3 Ni 5 C 5 sample are presented in figure 3.

Results and discussion
The inset of this figure shows the magnetoresistance behavior. We are presenting resistance instead of resistivity because the irregular shape of the sample precludes accurately determining the geometrical factors. The onset superconducting critical temperature is close to 5.0 K in zero magnetic field, consistent with magnetic measurement (shown in figure 2).
The sharp superconducting transition ( T c ∼ 0.2 K), indicates good sample quality. Magnetoresistance as a function of temperature, shown in the inset, suggests a relatively high upper critical field (H c2 ).
These results are consistent with the M versus T and M versus H curves, and also suggest bulk superconductivity. In order to confirm the bulk superconductivity of Th 3 Ni 5 C 5 we measured the heat capacity. An anomaly (jump) at 5.0 K is clearly observed in the temperature dependent heat capacity (C) measurement, shown in figure 4 with a temperature range of 0.4-8.0 K at zero magnetic field. This result is totally consistent with the M versus T and R versus T measurements, and represents clear evidence of bulk superconductivity in Th 3 Ni 5 C 5 . Figure 5 shows the C/T against T 2 in various magnetic fields. The inset shows C/T against T 2 at μ o H = 6.0 T which reveals the normal state (C n ). The normal state specific heat can be fitted to the expression C n = γ T + βT 3 by a least-square analysis, yielding the values γ = 38.84 mJ mol −1 K −2 and β = 1.012 mJ mol −1 K −4 . This β value corresponds to a Debye temperature of D ∼ 293 K and a Sommerfeld coefficient for the mole formula unit suggests a density of state at the Fermi level typical of transition metal superconductors. The subtraction of the phonon contribution allows us to evaluate the electronic contribution to the specific heat, plotted as C e /T versus T in figure 6. An analysis of the jump yields C e /γ n T c ∼ 1.1 which is smaller than the weak-coupling Bardeen-Cooper-Schrieffer (BCS) prediction (1.43). Indeed, the specific heat C s in the superconducting state shows marked deviations from conventional BCS theory as presented in figure 7. A great deviation can be observed at temperatures already close to T c . The origin of these deviations in both figures is not obvious. On the other hand, we note that the C e /T shows an unusual behavior at low temperature, i.e. an upturn for T < 1.0 K ( figure 6). This upturn may be due to the magnetic Schottky contribution and/or the paramagnetism of unreacted Ni impurities. In fact, similar behavior was observed in MgB 2 where Fe impurities lead to an upturn in C/T at low temperature [20]. However, we cannot disregard the fact that this kind of behavior (like exponential behavior)  could also represent the possibility of a second gap in low temperature, due to the nodal structure in the Fermi surface. Indeed the possibility of a second gap is observed in MgB 2 due to the π band in the Fermi surface [21]. The presence of the impurity phase (ThC 2 as a minority phase) could be related to the deviation observed, but this kind of behavior can be observed in the annealed and as-cast sample and ThC 2 has no effect on the superconductivity [22]. The deviation from BCS ( C e /γ n T c ∼ 1.1) would correspond to about 77% of the ideal value, but we estimate the impurity phase percentage ∼7% from the analyses of the powder x-ray diffraction pattern. This value is lower than the 23% from the analyses of the jump in specific heat. The deviation may be a real phenomenon in this material, but studies on phase pure material would be needed to determine this.
A comparison between the results shown in figures 3 and 5 shows excellent agreement. The shifts of the critical temperature as a function of applied magnetic field are consistent in both measurements. μ o H c and its temperature  Figure 8 shows the μ o H c as a function of reduced temperature (T /T c ), extracted from both figures (figures 3 and 5). These results are in good agreement. The upper critical field at zero temperature (μ o H c2(0) ) can be estimated using the WHH formula [23] in the limit of short electronic mean free path (dirty limit), Figure 8 shows the curve estimated by WHH which follows the data points very closely and gives a μ o H c2 (0) value of 5.8 T. On the other hand, the spin-orbit scattering counteracts the effect of the Pauli paramagnetism, giving an upper bound to H c2 where the pair breaking is only induced by orbital fields. The temperature dependence of the upper critical field can either be explained by Pauli paramagnetism with extremely strong spin-orbit scattering or with a completely dominating orbital field effect. The Pauli limiting field is described by Within the same weak-coupling BCS theory this gives an upper critical field of 9.22 T, which is much higher than μ o H c2 (0) in the absence of Pauli paramagnetism (5.8 T). Hence pair breaking in Th 3 Ni 5 C 5 is most probably determined by orbital fields.
The fitting of the figure 8 data allows an estimation of the coherence length, through the Ginzburg-Landau (GL) formula, μ o H c2 (0) = φ o 2πξ 2 0 , which yields ξ o ∼ 75Å. At 1.8 K the coherence length is about 85Å estimated from figure 8, whereas the penetration depth is about 207 nm. These values yield a GL κ (1.8) ∼ 25, which is much higher than 1 √ 2 . This κ is consistent with the behavior of a type II superconductor, as revealed by the M versus H curve (inset of figure 2).
The fact that superconducting compounds occur where so much Ni is present is always interesting, and Th 3 Ni 5 C 5 is a new example.

Conclusion
In summary, the excellent agreement of transition temperature as characterized by the magnetic, resistivity, and specificheat data unambiguously indicates that Th 3 Ni 5 C 5 is a type II superconductor with T c ∼ 5.0 K. Our results presented in this paper disagree with the previous results which reported the discovery of the Th 3 Ni 5 C 5 organometallic phase where the authors claimed that this material was a simple Pauli paramagnet at low temperature [18].