Volume 75A, number 5
4 February 1980
SUPERCONDUCTIVITY OF RARE EARTH—IRON SILICIDES Hans F. BRAUN
1, University of California, San Diego,
Institute for Pure and Applied Physical Sciences La Jolla, California 92093, USA Received 11 October 1979
A series of RE 2Fe3Si5 ternary compounds is formed with scandium, yttrium and the rare earth elements samarium through lutetium except europium. The lutetium, scandium and yttrium members are superconducting, lutetium and scandium exhibiting the highest critical temperatures yet observed for iron compounds.
The recent discovery of superconductivity in ternary scandium—transition metal siicides Sc5T4Si10 with the transition metals T = iridium, rhodium and cobalt  made it worthwhile to look for possible superconductors in the scandium—iron—silicon system. This was especially intriguing, as there are only very few superconductors which contain iron as an essential constituent. As a solute in certain titanium and zirconium alloys, iron enhances their supercon. ductivity [2—5] Among compounds, only two superconducting iron binaries are known, Th7 Fe3  with a critical temperature of 1.86 K and U6Fe  with Tc = 3.9 K. The phase diagram scandium—iron—silicon has been studied by Gladyshevskii et al.  ,who determined the isothermal cross section at 800°Cand reported 14 ternary phases. For the present work, only the compounds with more than 40 at% silicon content were prepared by arc melting, in analogy to the silicon-rich Sc5T4Si10 superconductors. The alby with the composition Sc2 Fe3 Si5 was found to become superconducting with a broad transition around 3.5 K. Subsequent annealing of the specimen in a sealed quartz tube under a partial pressure of ultra high purity argon improved both the transition temperature and transition width. A heat treatment for 24 h at 1150°Cfollowed by 14 days at 1000°C .
sponsored by the National Science Foundation under contract NSF/DMR77-08469.
produced a superconducting transition ranging from 4.45 K to 4.18 K. Here, the two numbers represent the onset and completion of the transition, defined at 10% and 90% of the superconducting signal, respectively. A further heat treatment for 14 days at 800°Cresulted in a transition 4.52—4.25 K. These transition temperatures were determined by ac susceptibility measurements on a bulk sample. Measurements on the same specimen after powdering yielded a wider transition (4.5—3.5 K), but indicated superconductivity of the full volume. The X-ray powder pattern (Cr Ka, 2.2909 A) of Sc2 Fe3 Si5 could be indexed on the basis of a primitive tetragonal lattice with a = 10.225 A and c = 5.275 A (c/a = 0.5 16). The crystal structure of Sc2Fe3Si5, which belongs to spacegroup P 4/mnc, has been solved by Bodak et al. . However, the cell dimensions they reported (a = 10.05 A, c = 5.3 13 A, c/a = 0.529) differ considerably from those found in the present investigation. The atomic coordinates reported by Bodak et al. were used with the program “Lazy Pulverix”  to calculate powder intensities for Sc2Fe3 Si5. The calculated and observed intensities agree quite well, confirming that a phase of the Sc2Fe3 Si5 structural type was obtamed. Because differences in the phase composition could conceivably be responsible for the discrepancies in the lattice parameters, a comprehensive study of the homogeneity range and the dependence of the
Volume 75A, number 5
superconducting transition temperature on composition was conducted. Starting from a master alloy, five alloys encompassing the stoichiometric composition were prepared by addition of small measured amounts of scandium and silicon. All alloys were heat treated for two days at 1200°Cfollowed by 6 days at 800°C.From micrographs, none of the samples wasjudged single phase. The results are presented in table I. No strong variation of the lattice parameters was found. The critical temperatures are highest on the iron deficient side and drop sharply for iron rich samples. These results indicate a narrow homogeneity range and a significant dependence of the critical temperature on composition and most likely long range order. This latter effect is also suggested by the imtemperatures after low temperature The fact that dysprosium can replace scandium in the Sc2Fe3Si5 structure  made it very likely that at least the smaller rare earth elements would form isostructural compounds. RE2Fe3Si5 alloys were prepared with the complete lanthanide series (except Pm) and with yttrium. The Sc2 Fe3 Si5 structure was found for the yttrium compound and for the whole series samarium through lutetium, with the exception of europium. The lattice parameters for the lanthanides are given in fig. 1. While the yttrium and lutetium compounds are superconducting (table 1), all the other rare earth compounds are magnetic. This includes ytterbium, which, as can be seen from the lattice parameters, is in its trivalent state. The ac susceptibility measurements indicate magnetic ordering temperatures between 1 K and 11 K. The nature of
4 February 1980
053 0 52 0
0 30 -
A 5 40
62 64 66 68 70 Sm Gd Tb Dy Ho Er Tm Yb Lu Fig. 1. Lattice parameters of RE2Fe3Si5 compounds.
the magnetic ordering remains to be determined. While a detailed discussion has to await the cornpletion of magnetization and heat capacity measurements, a brief discussion of some structural properties is appropriate. One of the remarkable features of the Sc2Fe3Si5 structural type is the formation of iron “clusters”. In this structure, iron atoms occupy two sets of point positions. In both sets, each iron has two iron nearest neighbors. One of these sets forms iron chains along the twofold axes, the other,
Table 1 Lattice parameters and Tc for RE2Fe3Si5 ternaries. Composition
Sc19 5Fe31 0~’495 Sc21 9Fe301 Si48.0 Sc18 7Fe29 75i516 Sc202Fe292Si507 Sc209 Fe28 9Si502 Sc21 5Fe29 65i489 Sc2Fe3Si5 Y2Fe3Si5 Lu2Fe3Si5
10.222(8) a) 10.222(8) 10.222(8) 10.221(8) 10.224(8) 10.220(8) 10.225(5) 10.43(1) 10.34(1)
5.270(5) 5.272(5) 5.274(5) 5.269(5) 5.274(5) 5.273(5) 5.275(5) 5.47(1) 5.375(8)
3.9—2.8 3.0—2.6 3.9—3.5 4.5—3.7 4.5—3.6 4.0—3.6 4.52—4.25 2.4—2.0 6.1—5.8
a) Probable error in units of the least significant digit.
Volume 75A, number 5
ir~nsquares centered around the fourfold axes. The distances within the chains and squares are approximately equal, 2.64 A and 2.67 A, respectively, while the nearest iron distances between the two sets of “clusters” is considerably larger (4.10 A). However, these compounds are not “cluster compounds” in the most stringent sense  as the nearest neighbor distance is larger than in iron metal. The iron squares alternate with scandium squares to form the base planes of tetragonal antiprisms centered around silicon atoms. These tetragonal antiprisms are stacked in columns parallel to the c-axis, centered at (0, 0) and (1/2, 1/2). The columns provide a relationship between the Sc2Fe3Si5, the W5Si3 and the U6Fe ,
structural types. However, the way in which the columns are linked is different in all three structures. The high iron content of 30 at% does not cause ferromagnetism, because the iron—iron distances of 2.64 A are too long to permit ferromagnetic interactions  The rare earth in the RE2Fe3Si5 structure has 20 atoms in its coordination sphere , among them four rare earth atoms. The rare earth nearest neighbor distance is 3.70 A for the scandium compound, far shorter than in the ternary rhodium borides or in the Chevrel phases in which these distances are of the order of 5 to 6 A. It is therefore not surprising that no superconductivity is observed in RE2 Fe3 Si5 compounds containing magnetic rare earths. The superconducting critical temperatures of Sc2Fe3Si5 and Lu2Fe3Si5 surpass those of all previ. ously known iron compounds. This shows that 3d .
4 February 1980
elements offer prospects for superconductivity and illustrates the potential of ternary compounds for achieving high temperature superconductivity. The author wishes to thank Professor B.T. Matthias for his interest, and C.U. Segre for preparing the yttrium compound and fruitful discussions. References [11 H.F. Braun, Bull. Am. Phys. Soc. 24 (1979) 504; to be published.  B.T. Matthias and E. Corenzwit, Phys. Rev. 100 (1955)
626.  B.T. Matthias, 1. App!. Phys. 31(1960) 23S.  E. Raub, Ch.J. Raub, E. Rösche!, V.B. Compton, T.H. Gebal!e and B.T. Matthias, J. Less Common Met. 12
(1967) 36.  WE. Collings (1975) 157. and J.L. Ho, J. Less Common Met. 41  B.T. Matthias, V.B. Compton and E. Corenzwit, J. Phys. Chem. Solids 19 (1961) 130.  B.S. Chandrasekhar and .i.K. Hu!m, J. Phys. Chem. So!ids 7 (1958) 259.  E.I. G!adyshevskii, B.Ya. Kotur, 0.1. Bodak and V.P. Skvorchuk, Dop. Akad. Nauk Ukr. RSR, Ser. A (1977) 751.  0.1. Bodak, B.Ya. Kotur, V.!. Yarovets and E.I. Gladyshevskii, Soy. Phys. Crystallogr. 22 (1977) 217.  K. Yvon, W. Jeitschko and E. Parth~,J. App!. Cryst. 10 (1977) 73.  J.M. Vandenberg and B.T. Matthias, Science 198 (1977) 194. [121 For a recent review, see M.B. Stearns, Phys. Today 31 (1978) 34.