Crystal growth, transport, and magnetic properties of Ln 3 Co 4 Sn 13 ( Ln ¼ La, Ce) with a perovskite-like structure

Ln 3 Co 4 Sn 13 ( Ln ¼ La, Ce) have been synthesized by ﬂux growth and characterized by single crystal X-ray diffraction. These compounds adopt the Yb 3 Rh 4 Sn 13 -type structure and crystallize in the cubic space group Pm¯ 3 n (No. 223) with Z ¼ 2. Lattice parameters at 298 K are a ¼ 9 : 6430 ð 6 Þ ˚A, V ¼ 896 : 68 ð 10 Þ ˚A 3 , and a ¼ 9 : 6022 ð 5 Þ ˚A, V ¼ 885 : 34 ð 8 Þ ˚A 3 for the La and Ce analogues, respectively. The crystal structure consists of an Sn-centered icosahedron at the origin of the unit cell, which shares faces with eight Co trigonal prisms and 12 Ln -centered cuboctahedra. Magnetization data at 0.1 T show paramagnetic behavior down to 1.8 K for Ce 3 Co 4 Sn 13 , with m eff ¼ 2 : 56 m B per Ce 3+ , while conventional type II superconductivity appears below 2.85 K in the La compound. Electrical resistivity and speciﬁc heat data for the La compound show a corresponding sharp superconducting transition at T c (cid:1) 2.85K. The entropy and resistivity data for Ce 3 Co 4 Sn 13 show the existence of the Kondo effect with a complicated semiconducting-like behavior in the resistivity data. In addition, a large enhanced speciﬁc heat coefﬁcient at low T with a low magnetic transition temperature suggests a heavy-fermionic character for the Ce compound. Herein, the structure and physical properties of Ln 3 Co 4 Sn 13 ( Ln ¼ La, Ce) are discussed. r 2006 Elsevier Inc. All rights reserved.


Introduction
Many ternary stannides, Ln-T-Sn (Ln ¼ Lanthanide; T ¼ Transition metal), exhibit notable physical properties such as Kondo lattice behavior, superconductivity, or longrange magnetic order, and also adopt interesting structures [1]. These compounds display magnetic properties due to either the T or Ln substructure, or both. La 6 Co 13 Sn is a ferromagnet with T c ¼ 190 K, due to the Co substructure, with a Co magnetic moment of 1.1 m B [2]. LnCo 3 Sn (Ln ¼ Y, Gd-Yb) possess ordering temperatures of 117-238 K; however, the Co moments are lower than the expected values, a phenomenon which is attributed to the Co clusters found within the structure [3]. In the equiatomic LnRhSn group of compounds, LaRhSn becomes superconducting below 1.7 K [4]. The Kondo system, CeRhSn, shows valence fluctuations [5], and YbRhSn is a heavy fermion with an electronic specific heat coefficient g$1200 mJ mol À1 K À2 [6][7][8][9][10]. Heavy fermions are materials that possess large enhanced electronic masses (g$100 times that of a free electron) as a result of the interactions between the conduction electrons and the local magnetic ions at low temperatures [11,12]. Typically, the Ln atoms in ternary stannides are trivalent (with the exception of some Ce and Yb compounds), and the contribution of the transition metal to the magnetism is minimal. The compositions and structures adopted by these compounds, however, heavily influence their transport properties.
In our search for new Ce-based intermetallics by exploring the La-Co-Sn and Ce-Co-Sn systems, we have synthesized Ln 3 Co 4 Sn 13  Yb 3 Rh 4 Sn 13 -type stannides first reported by Hodeau et al. [13]. The coexistence of magnetism and superconductivity (T c $8 K) has been observed in Yb 3 Rh 4 Sn 13 [13,14], while heavy fermion behavior with two phase transitions is observed in Ce 3 Ir 4 Sn 13 [15][16][17]. Furthermore, the isostructural Ce 3 Pt 4 In 13 , is also a heavy fermion with g$1000 mJ mol À1 K À2 [18]. The crystal structure of the Ln 3 Co 4 Sn 13 (Ln ¼ La-Nd, Sm, Gd, Tb) compounds has been previously studied by powder X-ray diffraction methods [19]. Herein we report the synthesis, structure characterization by single crystal X-ray diffraction, magnetization, electrical resistivity and specific heat of Ln 3 Co 4 Sn 13 (Ln ¼ La, Ce) and compare the structures and physical properties to the Rh and Ir analogues [15,20].

Synthesis
Ln 3 Co 4 Sn 13 (Ln ¼ La, Ce) single crystals were grown using excess Sn. Ingots of La or Ce (99.99% purity, Materials Preparation Center, Ames Laboratory), Co powder (99.998% purity, Alfa Aesar), and Sn shot (99.8% purity, Alfa Aesar) were weighed and placed into alumina crucibles in a 1:1:20 (Ln:Co:Sn) ratio. The samples, weighing nearly 3.25 g, were covered with quartz wool and encapsulated into evacuated, fused-silica tubes. They were heated to 1273 K for 5 h, then cooled to 1123 K at a rate of 75 K h À1 , and finally cooled to 523 K at a rate of 33 K h À1 . At this temperature, the ampoules were removed from the furnace and the excess Sn was removed by centrifugation. The retrieved irregularly shaped crystals, with dimensions between 1 and 2 mm 3 , were slightly air and moisture sensitive, as surface oxidation appeared following exposure to the atmosphere for extended periods. The centrifugation process removed most of the flux contamination on the surfaces of the single crystals; however, where necessary, the remaining topical flux was etched using concentrated HCl. Powder X-ray diffraction data were collected using a Bruker D-8 X-ray diffractometer with monochromatized CuK a radiation, l ¼ 1:540562Å. Phase identification of the Ln 3 Co 4 Sn 13 (Ln ¼ La, Ce) compounds was determined by comparing the powder patterns (not shown) taken from ground single crystals with that of Yb 3 Rh 4 Sn 13 [13] and also with the calculated powder pattern from the refined crystal structure of Ce 3 Co 4 Sn 13 . Multiple crystals from the products of both reactions were also characterized by single crystal X-ray diffraction as a check for homogeneity.

Single crystal X-ray diffraction
Suitable crystal fragments with dimensions of $0.025 Â 0.050 Â 0.050 mm 3 (La 3 Co 4 Sn 13 ) and $0.025 Â 0.10 Â 0.10 mm 3 (Ce 3 Co 4 Sn 13 ) were mechanically separated and glued to the tip of a glass fiber with epoxy. Structural analysis was done using a Nonius Kappa CCD diffract-ometer equipped with graphite monochromated MoK a radiation (l ¼ 0:70173Å) at room temperature and at 140 K (to check for a structural phase transition in the Ce compound). Additional data collection parameters are presented in Table 1. The structures were solved by direct methods and refined using the SHELXL97 package [21]. The atomic data reported for Yb 3 Rh 4 Sn 13 [13] were used to further refine the structural models. Data were corrected for extinction and refined with anisotropic atomic displacement parameters. To accurately determine the composition of our compounds, the occupancy parameters were refined in separate sets of least-squares cycles, since Yb 3 Rh 4 Sn 13 -type stannides may show defects at the origin of the unit cell (Sn1 on the 2a site) as observed in Ce 3 Rh 4 Sn 13 and Ce 3 Ir 4 Sn 13 [20]. Unlike the Rh and Ir analogues, the Sn1 sites in Ln 3 Co 4 Sn 13 (Ln ¼ La, Ce) are fully occupied. The atomic positions and structural information for both compounds are detailed in Table 2. Table 3 lists selected interatomic distances. Lattice parameters obtained from the room-temperature data collections are a ¼ 9:6430ð6Þ and 9:6022ð5ÞÅ, for the La and Ce compounds, respectively. These values are in close agreement with data reported previously from powder X-ray diffraction: 9.635(1) Å (La) and 9.594(1) Å (Ce) [19]; and 9.721 Å (La) and 9.590 Å (Ce) [22]. The final least-squares refinement cycle gave RðF Þ ¼ 0:0305 (La) and 0.0561 (Ce) and R w ðF 2 o Þ ¼ 0:0560 (La) and 0.1424 (Ce). The largest differences in the Fourier map from the room-temperature data collections are 1.658/À1.905 e Å À3 from La/Sn2 in the La compound and 3.615/À2.506 e Å À3 from Ce/Sn1 in the Ce compound. Additional crystallographic information in CIF format is provided as Supporting Information.

Physical property measurements
Magnetic data were measured on single crystals using a Quantum Design MPMS Superconducting Quantum Interference Device magnetometer. The zero-field-cooled (ZFC) temperature-dependent susceptibility data were taken in an applied field of 0.1 T up to room temperature after being cooled to 1.8 K under zero magnetic field. The fielddependent susceptibility was measured at 2 K by sweeping the magnetic field to 7 T and back. The resistivity data have been measured using a standard four-probe method down to 0.4 K with a Quantum Design Physical Property Measurement System (PPMS) at ambient pressure. The specific heat was measured with a Quantum Design PPMS using a thermal relaxation method from 0.36 to 30 K in zero applied field; entropy was calculated by integrating the specific heat divided by the temperature.
The Sn1, Ln, Co, and Sn2 atoms occupy the 2a, 6d, 8e, and 24k sites, respectively. The structure of Ce 3 Co 4 Sn 13 is shown in Fig. 1 [23] and ACu 3 Ti 4 O 12 [24]. Ln 3 Co 4 Sn 13 (Ln ¼ La, Ce) have a perovskite-like arrangement where the Sn and Ln atoms occupy the A site of the perovskite and Co occupies the B site [13,25,26]. The Ln 3 Co 4 Sn 13 (Ln ¼ La, Ce) compounds extend the Yb 3 Rh 4 Sn 13 -type family of compounds with three substructures such that the Sn1 atoms form Sn1(Sn2) 12 icosahedra, the Ln atoms form Ln(Sn2) 12 cuboctahedra, and the transition metal forms TSn 6 trigonal prisms. Additionally, it can also be regarded as the sum of two interpenetrating structures of SnLn 3 and CoSn 3 .
The Sn1 atoms occupy the origin of the unit cell and form an infinite network of edge-sharing Sn1(Sn2) 12 icosahedra which pack in an CsCl arrangement. The faces of each Sn1 icosahedron in the network make contact with  (4) a U eq is defined as one-third of the trace of the orthogonalized U ij tensor.
The Co(Sn2) 6 trigonal prisms found in Ln 3 Co 4 Sn 13 (Ln ¼ La, Ce) are shown in Fig. 2. They are corner-sharing with a tilted three-dimensional arrangement that creates ''cages'' which encompass the Sn1 atoms. This feature is similar to the BO 6 octahedra in A 0 A 00 3 B 4 O 12 -type compounds and also to the TPn 6 (Pn ¼ pnictogen) octahedra found in the structures of skutterudites such as LaFe 4 P 12 [30]. The trigonal prisms contain Co-Sn interatomic distances of 2.6210(3) Å ( Â 6) and 2.751(10) Å ( Â 6) for Ln ¼ La and Ce, respectively. These distances are slightly less than the sum of the covalent radii of Co (1.25 Å ) and Sn (1.54 Å ); however, similar distances are found in the Co-Sn binaries CoSn (2.618 and 2.639 Å ) [31], CoSn 2 (2.737 Å ) [32], and Co 3 Sn 2 (2.703 Å ) [33]. The T-Sn interatomic distance in the isostructural Ce 3 Rh 4 Sn 13 and Ce 3 Ir 4 Sn 13 compounds are 2.659 and 2.667 Å , respectively, which are also slightly shorter than the sum of the metallic radii, although larger than the sum of the ionic radii [20]. In addition, the Rh-Sn2 distances in Yb 3 Rh 4 Sn 13 are also shorter than the corresponding intermetallic radii and it has been suggested that the bonding nature in this compound may be ''covalent-ionic'' [13]. Given the similar crystal chemistry found in the A 0 A 00 B 4 O 12 compounds, our Ce 3 Co 4 Sn 13 compound may be interpreted as the covalent counterpart to A 0 A 00 B 4 O 12 .

ARTICLE IN PRESS
3.4274 (4) 3.4080 (7) 3.3992 (6) Co trigonal prism Co-Sn2 ( Â 6) 2.6210 ( Fig. 3a. The data fit well with a doubly degenerate ground state yielding m sat ¼ gJ z m B of 0.64 m B /Ce. Fig. 3b shows the magnetic data for La 3 Co 4 Sn 13 . A superconducting signal is detected below 2.85 K with constant Pauli paramagnetic behavior above this temperature. The type II superconductor behavior with H c1 $20 G and an upper critical field (H c2 )$1 T (not shown), is most likely conventional BCS superconductivity mediated by the vibration of the lattice. Perfect diamagnetism is expected in the superconducting state below H c1 , which is represented by the susceptibility value (4pM/H) of À1 or 4pM is equal to ÀH. The data clearly show a full Meissner effect, and no superconductivity from Sn inclusions has been detected in this particular sample.
Electrical resistivity data for La 3 Co 4 Sn 13 are shown in Fig. 4a. The resistivity suddenly drops to zero in the La compound at T c $2.8 K due to the superconducting transition, consistent with the susceptibility data. (Israel et al. report a T c ¼ 2:4 K; however, no transport data were presented due to Sn flux contamination [22].) Above T c , the resistivity increases as T 3 , consistent with Wilson's theory of the s-d hybridization effect in transition metals [34]. In addition, there is a shoulder between 10 and 160 K. This can be attributed to the s-d scattering between the conduction electrons and electrons from the unfilled Co d-band [35,36]. At higher temperatures, the resistivity rises less rapidly than the usual linear dependence in temperature, which can be related to the high-temperature saturation of the resistance where the electron mean free path is similar to the interatomic distance [37]. The temperature dependence of the resistivity of Ce 3 Co 4 Sn 13 , shown in Fig. 4b, is very peculiar. With decreasing temperature, Ce 3 Co 4 Sn 13 displays metallic behavior down to $160 K. A sudden kink at 160 K is observed, which could be due to a structural change. A single crystal X-ray diffraction experiment at 140 K, however, shows no change in the crystal structure from that of the 298 K data as shown in Tables 1-3. Thus, it must be an intrinsic behavior which we do not fully understand at present. Below 160 K, the resistivity shows a rather complicated semiconducting behavior. There are two slope changes between 20 and 150 K which may have the same origin as in the La compound. Below 20 K, the resistivity increases dramatically and a kink is observed at T$0.6 K corresponding to the transition observed in the specific heat data ($0.65 K) [38]. We subtracted the resistivity of the La compound from the resistivity of the Ce compound to isolate the magnetic contribution of the Ce f-moment, assuming that lattice and electronic contributions are the same as in the La compound, other than f-electron contribution. As seen in Fig. 5a, above 160 K, one can clearly see the linear behavior in log T, which is indicative of Kondo scattering at high temperature. Below the slight kink at 160 K, only a limited region follows the log T dependence. This feature below 100 K may be due to Kondo scattering along with magnetic fluctuations. We note that there was some difficulty in collecting reliable resistivity data due to the Sn inclusions inside the crystal which led to significant changes in resistivity behavior proportional to the amount of the Sn inclusion. The resistivity data presented here for Ce 3-Co 4 Sn 13 were reproducible from measurements on several single crystals that had no Sn inclusions. The negative y cw from the susceptibility of Ce 3 Co 4 Sn 13 suggests an antiferromagnetic nature for this transition. The character of the magnetic transition, however, is not obvious as no long-range order has been detected at 0.8 K from powder neutron diffraction experiments [38]. Moreover, in the specific heat data under applied magnetic field of 2.5 T or more, this kink, originating presumably from short-range magnetic order, is suppressed, and Kondo impurity behavior appears [38]. Resistivity measurements with an applied magnetic field perpendicular to the current were performed to give better insight to the origin of this magnetic transition at low temperature and is shown in Fig. 5b. Interestingly, the kink at low temperatures disappears, and the resistance at 0.4 K drops linearly with applied magnetic field and decreases $20% with applied field of 9 T. It would be interesting to investigate with higher magnetic fields in order to suppress the lowtemperature resistivity to determine if the coherent scattering can be recovered.

ARTICLE IN PRESS
Temperature-dependent specific heat divided by the temperature (C p /T) for Ce 3  from previous reported data [22,38]. The linear specific heat coefficient (g) can be estimated by fitting the data with g þ bT 2 above the transition temperature, giving $4 mJ mol À1 K À2 for La 3 Co 4 Sn 13 . Below 5K in Ce 3 Co 4 Sn 13 , C p /T increases dramatically due to magnetic short-range fluctuations followed by a peak at $0.65 K, indicating a magnetic phase transition. Although g estimated from extrapolating C p /T to zero temperature is about 75 mJ mol À1 K À2 in the Ce compound, C p /T is enhanced up to 4280 mJ mol À1 K À2 at the transition peak temperature. This large enhancement of C p /T can be caused by large short-range magnetic fluctuations, but evidences of Kondo scattering in the susceptibility, resistivity, and entropy data (to be discussed in a later communication) support a heavy mass state involved in this low-temperature behavior. One can isolate the magnetic part of the specific heat of Ce 3 Co 4 Sn 13 by subtracting the corresponding La 3 Co 4 Sn 13 data in the normal state to exclude lattice contributions. The corresponding entropy is obtained by integrating C p /T after subtraction and is shown in Fig. 6. The magnetic entropy should have an R ln 2 value at the transition temperature if the spin is doubly degenerate, i.e., two spin degrees of freedom in its ground state. The magnetic entropy recovered at 10 K is about 0.85R ln 2. This value, close to R ln 2, implies the ground state is doubly degenerate as discussed earlier in relation to the magnetization results. The reduction of the entropy from R ln 2 around the transition temperature suggests the existence of the Kondo effect, as indicated from the resistivity data. This value is larger than the previously reported magnetic entropy at 20 K which is 60-70% of R ln 2 [22].

Summary
We have synthesized the Ln 3 Co 4 Sn 13 (Ln ¼ La, Ce) compounds in single crystalline form, characterized their structures by single crystal X-ray diffraction, and investigated their magnetic and transport properties. Specific heat data show evidence for heavy mass behavior in Ce 3 Co 4 Sn 13 and a magnetic phase transition at T$0.65 K. La 3 Co 4 Sn 13 becomes superconducting at T c $2.8 K, while transport behavior for the Ce analogue is more complex and contains both a metallic and semiconducting character. Ln 3 Co 4 Sn 13 (Ln ¼ La, Ce) is a phase worth further investigation because of the similarity of the structural units (cuboctahedra) found in this phase and the CeTIn 5 and Ce 2 TIn 8 (M ¼ Co, Rh, Ir) compounds. The heavy fermionic nature along with the low magnetic transition temperature makes Ce 3 Co 4 Sn 13 a particularly interesting candidate for the study of magnetic quantum critical phenomena. The behavior of the magnetic field dependence is worth further investigation. Also, thermal transport measurements, such as thermal conductivity and thermoelectric power, may prove interesting due to the caged structure of these materials.