Superconductivity in Single-Crystal YIn3
aa r X i v : . [ c ond - m a t . s up r- c on ] D ec Superconductivity in Single-Crystal YIn S.D. Johnson, J.R. Young, and R.J. Zieve
Physics Department, University of California Davis, Davis, CA 95616
J.C. Cooley
Los Alamos National Laboratory, Los Alamos, NM 87545 (Dated: November 10, 2018)We measure the superconducting transition of YIn by resistivity, susceptibility, and specific heat.Despite using high-quality single-crystal samples, the transitions detected by the three techniquesare shifted from each other in temperature, suggesting a region of filamentary superconductivity.We discuss the possible implications for filamentary superconductivity in unconventional supercon-ductors. I. INTRODUCTION
One hundred years after the discovery of supercon-ductivity, the goal of predicting whether a material willsuperconduct from its atomic composition and crystalstructure remains elusive. Even more difficult is esti-mating the transition temperature T c . Over the decades,researchers have often examined families of related ma-terials, trying to quantify the effects of replacing a singleatomic constituent with a related atom. One such family,investigated 40 years ago, had the cubic AuCu structure.Much work followed the belief that T c directly correlatedto the valence electron concentration [1, 2]. During thistime a large number of compounds with AuCu structurewere synthesized and tested for superconductivity at am-bient pressure, e.g. MIn (M = Y, La, Lu) at reported T c of 0.78 K, 0.71 K, and 0.24 K, respectively [3].Subsequently the discovery of superconductors thatcontain magnetic ions in their crystal structures over-turned the traditional view that magnetism and super-conductivity cannot coexist. The field of potential su-perconductors vastly broadened with the inclusion ofmagnetic ions, and several new families have appearedsince the early heavy-fermion and high-temperature su-perconductors. In heavy-fermion materials, conductionelectrons acquire an effectively “heavy” mass, often hun-dreds of times greater than the bare electron mass [4],from interaction with the magnetic ions. Furthermore,it appears that the superconductivity is unconventionalin that magnetic interactions actually cause the electronpairing required for superconductivity, a role played byphonons in conventional superconductors.One of the heavy-fermion superconductors is CeIn ,with T c peaking at 0.2 K under 26.5 kbar of hydro-static pressure [5]. Related heavy-fermions CeMIn (M= Co, Ir, Rh) and PuXGa (X = Co, Rh) are also su-perconducting with ambient-pressure T c reaching 18.5 Kin PuCoGa . CeIn has the AuCu structure, which hasinspired new studies of the AuCu family of materials.Normal state properties have been measured in YIn viade Haas-van Alphen experiments [6], thermal conductiv-ity [7], and heat capacity [8], leaving it well characterizedand a promising candidate for further exploration of the interplay between crystal structure and magnetic super-conductivity.Here we present measurements on much higher-qualitysamples of YIn than those previously explored below 1K [3]. We confirm the earlier ac susceptibility measure-ments of the superconducting transition and also pro-vide resistivity and specific heat data. We find an un-usual shift in the transition temperature using the differ-ent techniques, which may be connected to behaviors inrelated heavy-fermion compounds. II. SAMPLE PREPARATION
YIn crystallizes in a cubic AuCu structure with lat-tice constant 4.593 ˚A [6]. The single crystal samples weregrown from an excess Indium flux at a ratio of approxi-mately 50:1 in an alumina crucible with an integral alu-mina frit. The crucible was sealed in a quartz tube underargon, held at 1150 ◦ C for 72 hours and slow cooled to450 ◦ C. The 450 ◦ C crucible was removed from the fur-nace, inverted, and placed in a centrifuge to spin off theexcess flux. After cooling to room temperature cruciblewas broken out of the quartz tube and the samples re-moved from the crucible. The starting materials were99.9999% purity indium and 99.9% purity yttrium.The measurements were performed on a He/ He di-lution refrigerator, except that the ac susceptibility mea-surement on sample A was done on a He sorbtion cryo-stat. The sample was mounted in close proximity to thecryostat thermometer. For resistivity and susceptibilitywe took data on both warming and cooling through thetransition, with no sign of hysteresis. For the susceptibil-ity and heat capacity measurements we did not polish orclean excess indium from the sample. For the resistivitymeasurement we cut down and polished all sides, bothto improve the sample geometry and to ensure a goodconnection between the leads and the YIn . T (K) -1-0.50 χ Temperature (K) χ FIG. 1: Susceptibility of sample B, with a sharp jump at 950mK. The inset shows susceptibility of sample A, measured to3.65 K. The values are scaled to 0 at the highest temperatures(sample fully normal) and -1 at the lowest temperatures. Thelarge jump above 3 K indicates the superconducting transi-tion of the remnant indium in and around the sample. Thecontribution of YIn , barely visible near 1 K, makes up only3% of the entire transition. III. RESULTS AND DISCUSSION
We performed ac susceptibility measurements at a fre-quency of 12 Hz and an ac field of about 0.4 gauss. Asshown in the main graph of Figure 1, the onset of thedrop in susceptibility is at 960 mK, and the sample isfully superconducting by 825 mK. We scale the data to 0in the normal state and -1 in the superconducting state.The inset shows susceptibility extended to higher temper-ature. A much larger transition appears at 3.4 K fromthe excess indium in and around the sample. The YIn transition makes up only ∼
3% of the total susceptibilitychange. A simple estimate of the expected signal, basedon the sizes of the sample and the measurement coils andassuming the sample becomes entirely superconducting,gives about 1.5 times the observed signal. This reason-ably good agreement suggests that in fact the indium isshielding almost the entire sample volume. The smallmagnitude of the YIn transition then stems from theshielding by the indium, and does not indicate that onlya small portion of the YIn is superconducting. We be-lieve the shielding arises from remnant indium on thesurface of the sample; subsequent cutting into the in-terior confirmed that only a few small isolated veins ofindium reside there.Figure 2 shows the specific heat of sample B, with a fea-ture at 825 mK. We use a relaxation method with a 50:50AuCr heater, a Cernox film thermometer, and a graphite T (K) C / T ( m J / m o l e - K ) FIG. 2: Heat capacity of YIn , sample B. thermal link. The sample mass is 57.55 mg, and our heatpulses produce a temperature change of about 20 mK. Inthe normal state, C/T is by no means constant, but itis the same order of magnitude as in conventional super-conductors. This is consistent with de Haas-van Alphenmeasurements reporting a small effective electron mass[6, 9]. The relative size of the transition is ∆ CC ( T c ) = 0 . samples.These were the same two samples shown in Figure 1, butcut down and polished to increase the aspect ratio andremove excess indium. We used a four-wire techniquewith platinum wires 0.051 mm in diameter spot-weldedto the four corners. The other ends of the wires were sol-dered to a small copper plate which in turn connected toa resistance bridge via fixed wiring on the cryostat. Atroom temperature the measured resistivities of the twosamples differ by about 10%, with ρ ≈ µ Ω-cm. Theresistivity ratios from room temperature to 4 K are 21and 36 for samples A and B, respectively. A sharp dropin resistivity with decreasing temperature indicates thesuperconducting transition. For sample A the drop be-gins about 1.2 K and is completed above 1 K; for sampleB it occurs between 1.08 K and 0.98 K.An earlier report [3] of superconductivity in YIn wasbased solely on ac susceptibility measurements. Thetransition midpoint was 780 mK, with half-width of 210mK. The half-width was calculated by taking the tangentline to χ at the transition midpoint, and then finding T(K) ρ ( µ Ω - c m ) FIG. 3: Resistive transitions for two samples of YIn . Solidpoints are sample A; open are sample B. the temperatures where it attained the full normal andsuperconducting susceptibilities. Using this same defini-tion, the susceptibility transition shown in Figure 1 hashalf-width 36 mK. Our sample has a higher transitiontemperature, a much smaller width, and a fairly highresidual resistivity ratio of 36. All of these suggest bet-ter sample quality than in the previous work.Strangely, measurements by resistivity, susceptibility,and heat capacity give significantly different transitiontemperatures. For sample B, the transition onsets are at1.08 K, 0.95 K, and 0.90 K, respectively, with the threetechniques. The transition widths, from onset to com-pletion, are less than 100 mK in resistivity, and less than150 mK with the other techniques. This means that theresistive transition does not overlap the other two, andthere is a gap of nearly 100 mK between the comple-tion of the resistive transition and the onset of the heatcapacity transition. Although discrepancies in T c canresult from sample quality issues, a clear separation be-tween the transitions such as we observe is unusual inconventional superconductors. However, it appears in avariety of heavy-fermion and other unconventional super-conductors, including CeIrIn , URu Si , and Fe(Te,Se).The discrepancy indicates a regime of filamentary super-conductivity, where despite a complete transition in resis-tance the superconducting volume is too small to registerwith bulk probes.In CeIrIn the resistive T c occurs at 1.2 K, while thesusceptibility and heat capacity signals occur only at 0.4K [10]. Improved samples have not brought the two closer together; if anything, they have raised the resistive T c .URu Si also shows a large discrepancy in the supercon-ducting phase boundary as mapped out by resistivity orspecific heat [11]. At ambient pressure the values of T c are 1.4 K and 0.8 K, respectively. Both decrease withapplied pressure, but the specific heat transition disap-pears by 0.5 GPa, while the resistive transition persiststo 1.8 GPa. Similarly, at certain Se concentrations x ,Fe y (Te − x ,Se x ) exhibits a resistive transition withoutany specific heat signal. The onset of partial diamag-netism may [12] or may not [13] accompany the resis-tive transition. These materials all have different crystalstructures, but they have in common a large and highlyanisotropic response to strain. One possible source of thefilamentary superconductivity is this extreme sensitivityto uniaxial pressure combined with internal strain withinthe sample [14].One of these unconventional superconductors, CeIrIn ,has a crystal structure closely related to that of YIn .As noted previously, its parent compound CeIn has theAuCu lattice structure. In both these Ce compounds,the magnetic cerium atom imparts a heavy effective massto the charge carriers. Both become superconducting,and the sizes of their specific heat discontinuities showthat the heavy particles themselves form the pairs nec-essary for superconductivity. By contrast, with no mag-netic ions, YIn has quasiparticle masses not far abovethat of free electrons. Its superconductivity is likely con-ventional, with electron pairing mediated by phonons.Our finding filamentary regime in a conventional super-conductor supports the internal strain explanation. Fur-ther comparative studies of YIn , its cerium-based rel-atives, and (Y,Ce)In alloys may pinpoint how stronglythe filamentary superconductor is linked simply to thecrystal structure, rather than to any magnetic behaviors. IV. CONCLUSION
We have measured ac susceptibility, specific heat, andresistivity on two single-crystal samples of YIn and con-firmed superconductivity by all three techniques. We finda T c somewhat higher from that previously reported andvarying based on measurement technique. The tempera-ture range between the resistive and bulk transition tem-peratures supports filamentary superconductivity, whichmay share the same origin as non-bulk superconductivityin several unconventional superconductors. V. ACKNOWLEDGEMENTS
This work was funded by the NSF though grant DMR-0454869 and the REU grant PHY-0649297. [1] E. E. Havinga, “W-like dependence of critical tempera-ture on number of valence electrons in non-transitional metal Cu Au-type alloys,”
Phys. Lett. , 350 (1968). [2] B.T. Matthias, “Empirical relation between supercon-ductivity and the number of valence electrons per atom,”
Phys. Rev. , 74 (1955).[3] E. E. Havinga, H. Damsma, and M H. van Maaren, “Os-cillatory dependence of superconductive critical temper-ature on number of valency electrons in Cu Au-type al-loys,”
J. Phys. Chem. Solids , 2653 (1970).[4] F. Steglich, J. Aarts, C.D. Bredl, W. Lieke, D. Meschede,W. Franz, H. Sch¨afer, “Superconductivity in the presenceof strong Pauli paramagnetism: CeCu Si ,” Phys. Rev.Lett. , 1892 (1979).[5] I. R. Walker and F. M. Grosche and D. M. Freye and G.G. Lonzarich, “The normal and superconducting states ofCeIn near the border of antiferromagnetic order,” Phys-ica C , 303 (1997).[6] V.B. Pluzhnikov, A. Czopnik, I.V. Svechkarev, “de Haas-van Alphen effect in ScGa , LuGa and YIn ,” PhysicaB , 375 (1995).[7] J. Mucha, “Thermal conductivity of REIn compounds,” J. Phys.: Condens. Matter , 1427 (2006).[8] L. Hale, K.A. Gschneidner Jr., V.K. Pecharsky, andY. Mudryk, “Low temperature properties of some RIn compounds,” Journal of Alloys and Compounds , 24(2009).[9] I. Umehara, N. Nagai and Y. Onuki, “High field magne-toresistance and de Haas-van Alphen effect in LaIn ,” J.Phys. Soc. Japan , 591 (1991). [10] C. Petrovic, R. Movshovich, M. Jaime, P. G. Pagliuso, M.F. Hundley, J. L. Sarrao, Z. Fisk, J. D. Thompson, “Anew heavy-fermion superconductor CeIrIn : A relativeof the cuprates?” Europhys. Lett. , 354 (2001).[11] E. Hassinger, G. Knebel, K. Izawa, P. Lejay, B. Salce, andJ. Flouquet, “Temperature-pressure phase diagram ofURu Si from resistivity measurements and ac calorime-try: Hidden order and Fermi-surface nesting,” Phys. Rev.B , 115117 (2008).[12] P.L. Paulose, C.S. Yadav, and K.M. Subhedar, “Mag-netic phase diagram of Fe . Te − x Se x : A compar-ative study with the stoichiometric superconductingFeTe − x Se x system,” Europhys. Lett. , 27011 (2010).[13] T.J. Liu, J. Hu, B. Qian, D. Fobes, Z.Q. Mao, W. Bao,M. Reehuis, S.A.J. Kimber, K. Proke˘s, S. Matas, D.N.Argyriou, A. Hiess, A. Rotaru, H. Pham, L. Spinu, Y.Qiu, V. Thampy, A.T. Savici, J.A. Rodriguez, and C.Broholm, “From ( n ,0) magnetic order to superconduc-tivity with ( n , n ) magnetic resonance in Fe . Te − x Se x ,” Nature Mater. , 718 (2010).[14] A. Bianchi, R. Movshovich, M. Jaime, J.D. Thompson,P.G. Pagliuso, and J.L. Sarrao, “Origin of the zero-resistance anomaly in heavy fermion superconductingCeIrIn : A clue from magnetic-field and Rh-doping stud-ies,” Phys. Rev. B64