Proc. Int. Conf. Heavy Electrons (ICHE2010) J. Phys. Soc. Jpn.

80 (2011) SA014 c 2011 The Physical Society of Japan

Superconductivity in Rh17 S15 and Pd17 Se15 : A comparative study

H.R. Naren, Arumugam Thamizhavel, A.K. Nigam and S. Ramakrishan
Department of Condensed Matter Physics and Materials Science, Tata Institute of Fundamental Research, Mumbai-400005, India
(Received October 26, 2010)

The presence of strongly correlated superconductivity in Rh17 S15 has been recently established. In this work we estimate normal and superconducting parameters of a single crystal of Rh17 S15 and compare them with those of a polycrystalline sample of Pd17 Se15 which is reported here for the rst time to be a superconductor below 2.2 K. The presence of superconductivity in two iso-structural (space group Pm3m) compounds provides an opportunity to understand the strongly correlated superconductivity in Rh17 S15 . We see that large unit volume of Pd17 Se15 and the large separation of Pd-Pd atoms in its structure as compared those in a pure Pd metal lead to the absence of strong correlations in this compound. Finally Pd17 Se15 can be characterized as an intermediate type-II superconductor as opposed to the strongly correlated superconductivity in Rh17 S15 . KEYWORDS: Strong correlations, Superconductivity, Rhodium sulphide, . . .

1. Introduction Recently there has been intense activity in the study of transition metal sulphides and selenides due to a variety of their properties such as, strongly correlated superconductivity, spin and charge density waves.13) In our effort to understand the physical properties of noble metal chalcogenides, we have studied cubic Rh17 S15 4, 5) which has short Rh-Rh bonds with cage like structures. So far, Rh17 S15 has been mainly studied by chemists6, 7) since it has a potential usage in fuel cells and it can also serve as a useful catalyst for the industrial recovery of high-value chlorine gas. Naturally occurring Rh17 S15 (known as Miassite) is a mineral mainly found near the Miass river in Russia. It is a cubic compound (space group Pm3m) with large unit cell with 64 atoms in (two formula units) Rh atoms in four inequivalent crystallographic sites. The crystal structure is shown in the Fig. 1. Some of us had reported the observation of strongly correlated superconductivity (Tc = 5.4 K) in Rh17 S15 4, 5) via magnetization, resistivity, heat capacity and Hall coefcient measurements. The compound was remarkable for its unusually high upper critical eld (Hc2 (T)) values (around 13 T at 3 K) well above the Pauli paramagnetic limiting eld (9.9 T).8, 9) The observation of T2 dependence of resistivity, enhanced susceptibility, moderate Sommerfeld coefcient(), large value of the upper critical eld and large heat capacity jump (C /T c =2.6) suggest that Rh17 S15 is a strongly correlated system. From the determination of the extrapolated value of the normal state susceptibility at 0 K ((0)), coefcient A (in the expression for resistivity, = 0 + A T 2 ) and , we have obtained a value of 2 for the Wilsons coefcient (2 k2 (0)/32 ) and B B 5 x 105 for the Kadawaki-Woods ratio (A/2 ). These estimates further substantiate our claim that Rh17 S15 is a strongly correlated system. One way to achieve a moderate density of low-energy fermionic excitations (as seen by the appreciable value of ) is from the large density of states of the narrow 4d band of Rh at the Fermi level.810) This is supported by the structure11) where some of the Rh atoms (Rh atoms at the 3d and 6e positions) which are quite close to each other. The crystal strcuture in shown in Fig. 1. On the other hand, Pd17 Se15 is structurally similar to
Rh(6e) Rh(3d) Rh(6e) b ca Rh (1b) Rh (6e) Rh (3d) Rh (24m) S

Fig. 1. The crystal structure of Rh17 S15 which consists of two formula units with 64 atoms in the unit cell. The unit cell has 4 types of Rhodium atoms with position symmetry 24m, 6e, 3d and 1 b and 3 types of sulphur atoms with position symmetry 12j, 12i and 6f. The Rh(d) has two shorter Rh(e) neighbors at 0.258 nm which together with 4 S(j) atoms completes an octahedron about Rh(d). In the pure Rh metal which has the cubic closepacked structure the nearest neighbor Rh-Rh distance is 0.269 nm suggesting strong Rh bonds in Rh17 S15 . However, in the iso-structural Pd17 Se15 the shortest Pd-Pd distance (2.78 ) is more than that in the elemental Pd (2.75 ) .

Rh17 S15 and it is also a mineral (known as Palladseite mainly found in Brazil). However, in Pd17 Se15 the shortest Pd-Pd distance is more than that in the elemental Pd.12) In this work, we will establish that Pd17 Se15 is a conventional superconductor below 2.2 K as opposed to the strongly correlated behavior seen in Rh17 S15 .The existence of Pd17 Se15 and its superconductivity provide us a way to understand the strongly correlated superconductivity in Rh17 S15 . 2. Experimental details Both compounds were prepared by heating stoichiometric compositions of the two elements at the rate of 8 C/hr to a temperature of 1150 C in an alumina crucible. The alumina crucible was sealed in an evacuated quartz tube. The reacted mixture is annealed at 1050 C for two days and then cooled down rstly at 8 C/hr till 600 C and further at 30 C/hr. In this method we obtained a single crystal of Rh17 S15 with a residual resistivity ratio (RRR) of 7 and a polycrystalline

SA014-1

J. Phys. Soc. Jpn. 80 (2011) SA014

SA014-2

Fig. 2. A plot of the heat capacity(C p ) vs temperature(T) from 2 to 15 K of a single crystal of Rh17 S15 . The solid line is a t to the expression which is described in the text.

Fig. 3. The temperature dependence of the upper critical eld (H c2 ) of Rh17 S15 down to 70 mK. The inset shows the temperature dependence of H c2 near the superconducting transition. The H c2 values are estimated from dc and ac magnetization, heat capacity and and resistance measurements.

sample of Pd17 Se15 with an RRR of 7. The samples were characterized using powder XRD and Laue spectrometers. The single crystal of Rh17 S15 showed clear Laue spots identifying the space group Pm3m. The lattice constants were 9.912 for Rh17 S15 and 10.607 for Pd17 Se15 . Resistivity measurements were performed in a home-made system from 1.5 K to 300 K and very low temperature measurements are made using an adiabatic demagnetization refrigerator (Cambridge Magnetic Refrigeration, U.K.) down to 100 mK. Magnetization measurements were performed using SQUID (Quantum Design 7 T model, USA) and Vibrating Sample (Oxford instruments 12 T model, U.K.) magnetometers. Heat capacity measurements were made using a Physical Property Measuring System (PPMS, Quantum Design, USA). 3. Results and Discussion Fig. 2 shows a plot of the heat capacity(C p ) versus temperature(T) for the single crystal of Rh17 S15 in the absence of applied magnetic eld from 2 to 15 K. A t to the equation Cp = T + T 3 + T 5 (1) is also shown in the same gure where is due to the electronic contribution, is due to the lattice contribution and is due to the contribution from the unharmonicity. The estimated values of , and are 110 mJ/mol-K2 , 0.63 mJ/molK4 and 1.5 J/mol-K6 . The non-zero value of suggests complex phonon structure that could arise from the large cubic structure of this compound. From the value, one can estimate the Debye temperature D from the expression, D = ( 124 NrkB 1/3 ) 5 (2)

the McMillans theory13) which states that, = 1.04 + ln(D /1.45 T c ) (1 0.62 ) ln(D /1.45 T c ) 1.04 (3)

Assuming =0.13, we nd the value of to be 0.58 which puts Rh17 S15 as an intermediate electron-phonon coupling superconductor. However, the large heat capacity jump of C = 2.28 J/mol K (after subtracting the extrapolated value of C p from the low temperature region to the normal state value at 5.36 K). This gives a value of C/T c equal to 2.6 which suggests that the observed superconductivity is in the strong coupling regime. We address this apparent contradiction in the concluding section of our paper. We can also estimate the density of states at the Fermi level N (0) using the relation, N (0) = 0.4248 states/eV-formula unit where is expressed in mJ/mol K2 . We get a value of 46 states/eV-formula unit for N . If one uses the estimated value of (0.58), we get the value of bare density of states (N = N /(1 + )) as 29.1 states/eV-formula unit. Fig. 3 shows the temperature dependence of the upper critical eld measurement from 70 mK to 5.4 K. The very low temperature data (below 2 K) have been included from a recent study.9) From this one can estimate the value of the upper critical eld at H c2 at 0 K to be 20 T which is in agreement with the recent single crystal study of Settai al.8) The slope (dH c2 /dT) near the superconducting transition is estimated to be 3.6 T/K which is smaller than the value quoted by Settai al.8) Using this value of 3.6 T/K for the slope of the of the upper critical eld, one can an estimate a value of 13.3 T for H c2 (0) from the WHH theory.14) This theory (in the dirty limit) gives an expression, Hc2 (0) = 0.693 T c (dHc2 /dT )Tc (4)

where N is the Avogadro Number, r is the number of atoms per formula unit and kB is the Boltzmann constant. he estimated value of the Debye temperature is found to be 456 K. Using the value of D and Tc (Tc = 5.27 K as obtained by an equal entropy estimate at the superconducting transition), we can estimate the electron-phonon coupling parameter, , from

This value(13.3 T) is much smaller than the observed value of 20 T. Moreover, the Pauli paramagnetic limiting eld (H p (0) = 1.84T c ) is only 9.9 T suggesting that the observed upper critical eld is unaffected by H p (0) implying unconventional superconductivity in Rh17 S15 . We have also estimated the penetration depth 0 as 7000 from the SR measurements performed at the PSI, Switzerland recently.15) From the

J. Phys. Soc. Jpn. 80 (2011) SA014

SA014-3

Fig. 4. A plot of the temperature dependence of the resistivity of Pd17 Se15 from 1.5 K to 300 K. The lower inset (a) is the normal state low temperature data (4 K to 30 K) and the solid line is a t to the expression which is described in the text. The upper inset (b) shows the superconducting transition below 2.2 K.

Fig. 5. Comparison of the temperature dependence of the resistivity of the single crystal Rh17 S15 with that of the polycrystalline Pd17 Se15 .

estimation of the upper critical eld H c2 (0), one can calculate the Ginzburg-Landau coherence length (0 ) which is given by, 2 0 = 0 /2Hc2 (0). This gives a value of 40 for 0 . Using 0 and 0 , we can calculate the lower critical eld H c1 using the relation, 0 Hc1 (0) = ln(0 ) (5) 42 0 where 0 = 0 /0 . The value of the lower critical eld is 1.7 mT which makes Rh17 S15 as an extreme (=175) type-II superconductor. 3.1 Normal and superconducting state properties of Pd17 Se15 and its comparison with those of Rh17 S15 Fig. 4 shows the temperature dependence of the resistivity ((T)) of a polycrystalline sample of Pd17 Se15 from 1.6 to 300 K. The inset (a) shows the superconducting transition in this compound below 2.2 K. The inset (b) shows the low temperature data and the t to the T2 dependence as mentioned earlier in the text. This t gives a value of 43.2 cm for the residual resistivity and a value of 0.014 cm/K2 for the coefcient A. The value of A for Pd17 Se15 is comparable to that of Rh17 S15 .4) Fig. 5 compares the temperature dependent resistivity of the single crystal Rh17 S15 with that of the polycrystalline Pd17 Se15 . The two curves are distinctly different in the appearance of a knee-like feature at around 60 K in Rh17 S15 which is completely absent in Pd17 Se15 . We do not, at present, understand the reason for the knee-like feature. However, we note that there is a change of sign of majority charge carriers occurs around 60 K in Rh17 S15 which was seen in the Hall measurements reported earlier.4) Though the absolute value of the resistivity of both samples are comparable at 300 K, there is a distinct possibility that the strongly correlated behavior observed in Rh17 S15 could be related to this feature. Fig. 6 shows a plot of C p versus T for the polycrystalline Pd17 Se15 in the absence of applied magnetic eld from 2 to 15 K. The solid line in the same gure is a t to the equation (1) described earlier. The t yields a of 22 mJ/mol-K2 ,

Fig. 6. A plot of the heat capacity(C p ) vs temperature T from 2 to 15 K of a polycrystalline Pd17 Se15 . The solid line is a t to an expression which is described in the text.

of 4.3 mJ/mol-K4 and of 9.2 J/mol-K6 . From the value of we obtain a Debye temperature (D ) of 233.3 K (using equation (2)) which indicates that Pd17 Se15 is a much softer lattice as compared with Rh17 S15 . However, the unharmonicity term is much larger for Pd17 Se15 as compared to that of Rh17 S15 . More importantly, the enhanced density of states, N (0), as calculated from the earlier expression is around 9.35 states/eV-formula unit which is smaller than that of Rh17 S15 by a factor of 5 indicating that the correlations are much weaker in Pd17 Se15 . From the values of and D , we have also estimated a value of 0.54 for the electron-phonon coupling constant (using the equation (3)) for Pd17 Se15 which is very similar to the value of Rh17 S15 . We believe that Pd17 Se15 is truly an intermediate type-II superconductor while Rh17 S15 is a strongly correlated superconductor where one cannot use equation (3) to get the correct estimate of . The reduction in the density of states in Pd17 Se15 as compared to that of Rh17 S15 is also reected in the temperature dependence of normal state magnetic susceptibility as shown in the Fig. 7.

J. Phys. Soc. Jpn. 80 (2011) SA014

SA014-4

Fig. 7. The temperature dependence of the susceptibility of Pd17 Se15 and Rh17 S15 in a eld 0.1 T.

Fig. 8. The temperature dependence of the upper critical eld (H c2 ) of pure Pd17 Se15 . The H c2 values are estimated from the magnetization measurements.

Since none of Rh, Pd, S or Se carry a magnetic moment we expect to see a temperature independent susceptibility essentially sum of Pauli paramagnetism, Landau diamagnetism and core diamagnetism. However, although the values of susceptibility ((T)) are small, the data display a distinct temperature dependence for Rh17 S15 . There is a gradual increase in susceptibility as one goes to lower temperatures until the superconducting transition. The weak temperature dependence of (T) could arise due to a sharp density of states at the Fermi level which leads to the temperature dependence of the Pauli spin susceptibility. We believe that the mechanism is similar to the observed in V3 Si.16) This conjecture is recently supported by the observation of temperature dependent Knight shift from the 103 Rh-NMR experiment.10) In fact, the value of susceptibility per Rhodium atom in Rh17 S15 is of the same order as that of susceptibility per Vanadium atom in V3 Si.16) However, the (T) of Pd17 Se15 is diamagnetic, shows weak temperature independent and its (T) is much smaller that of Rh17 S15 reecting the small value of the density states which is conrmed by the heat capacity data. Finally, we show the temperature dependence of the upper critical eld (H c2 ) of Pd17 Se15 from 270 mK to 2.2 K in Fig. 8. The curvature of H c2 near T c is close to that of a conventional superconductor as opposed the one seen in Rh17 S15 .4) However, both show a tendency to saturate at lower temperatures. The saturation is better seen in Rh17 S15 because we have data upto 70 mK whereas, we have presented data only upto 270 mK in Pd17 Se15 . The Hc2 (0) is estimated to be 3.3 T by an extrapolation which is nearly 6 times smaller than that of Rh17 S15 . The Ginzburg-Landau coherence length 0 is 100 as estimated from the H c2 (0)(3.3 T). The slope (dH c2 /dT) near Tc is 1.5 T/K which yields a value of 2.3 T for H c2 (0) from the WHH theory. However, the Pauli paramagnetic limiting eld (H p (0)) is estimated to have a value of 4 T which suggests that that orbital critical led is limited in contrast to the situation in Rh17 S15 . 4. Conclusion Earlier,,4) we have conjectured that the strong correlations in Rh17 S15 could be due to a high density of states of Rh-

4d band at the Fermi level arising due to strong Rh-Rh interactions. On the other hand Pd17 Se15 has a large unit cell volume and Pd-Pd distances are larger than that of pure Pd metal. Hence, we see the absence of strong correlations from its small values of and the normal state susceptibility ((T)). More over, the superconducting state properties of Pd17 Se15 can be explained by the intermediate type-II superconductivity (=0.54) with its upper critical eld (3.3 T) is limited by the Pauli paramagnetic eld as seen in conventional superconductors. However, the large heat capacity jump at the superconducting transition of Rh17 S15 implies strong coupling superconductivity which contradicts the small value of the (0.56) obtained from the McMillans theory.13) The value of 20 T for H c2 (0) which is more than twice the value of H p (0)(9.9 T) also imply unconventional superconductivity for which McMillans theory cannot be applied. The normal and superconducting parameters are given in Table I. Finally, our comparative study supports our conjecture that strong correlations seen in Rh17 S15 comes from the strong RhRh interactions which lead a narrow 4-d band at the Fermi level. It will be interest to study other superconducting noble metal chalcogenides to see whether one can observe strongly corrleated superconductivity due to narrow 4d and 5d bands at the Fermi level.6, 17)

1) Fong-Chi Hsu, Jiu-Yong Luo, Kuo-Wei Yeh, Ta-Kun Chen, Tzu-Wen Huang, Phillip M. Wu, Yong-Chi Lee, Yi-Lin Huang, Yan-Yi Chu, Der-Chung Yan, and Maw-Kuen Wu, Proc. Nat. Acad. Sci. 105 (2008) 14262. 2) Yoshikazu Mizuguchi,Fumiaki Tomioka, Shunsuke Tsuda, Takahide Yamaguch and Yoshihiko Takano, Applied Physics letters 93 (2008) 152505. 3) Alaska Subedi, Lijun Zhang, D. J. Singh, and M. H. Du, Phys. Rev. B

J. Phys. Soc. Jpn. 80 (2011) SA014 78 (2008) 134514. 4) H. R. Naren, A. Thamizhavel, A. K. Nigam, and S. Ramakrishnan, Phys. Rev. Lett. 100 (2008) 026404. 5) H. R. Naren, A. Thamizhavel, S. Ramakrishnan and A. K. Grover, J. Phys.: Conf. Ser. 150 (2009) 052183. 6) Sandip Dey and Vimal K. Jain, Platinum Metals Review (Published by Johnson Matthey, Hatton Garden, London), 48 (2004) 16. 7) Oswaldo Di uez and Nicola Marzari, Phys. Rev. B 80 (2009) 214115. 8) Rikio Settai, Keisuke Katayama, Hiroshi Muranaka, Tetsuya Takeuchi, Arumugam Thamizhavel, Ilya Sheikin, Yoshisicko Onuki, Physics and Chemistry of Solids.71 (2010) 700. 9) M. Uhlarz, O. Ignatchik, J. Wosnitza, H. R. Naren, A. Thamizhavel and

SA014-5 S. Ramakrishnan, J. Low Temp. Phys. 159 (2010) 176. 10) T. Koyama, K. Kanda, K. Ueda, T. Mito, T. Kohara and H. Nakamura, J. Phys.: Conf. Ser. 200 (2010) 012095. 11) S. Geller, Acta Cryst. 15 (1962) 1198. 12) S. Geller, Acta Cryst. 15 (1962) 713. 13) W.L. McMillan, Phys. Rev. 167 (1968) 331. 14) N.R. Werthamer, E. Helfand and P. C. Hohenberg, Phys. Rev.147 (1966) 295. 15) H.R. Naren et al, To be published. 16) J. Labbe, Phys. Rev. 158 647. 17) S. Nagata and T. Atake, Journal of Thermal Analysis and Calorimetry, Vol. 57 (1999) 807.