A routine test leads to an extraordinary discovery.

by Gregory R. Stewart, Zachary Fisk, Jeffrey O. Willis, and James L. Smith

s ome experimental findings are sounexpected, so outside the limits ofprevious experience, that their in-terpretation lags well behind the

facts, awaiting new insight. Our finding ofSeptember 16, 1983, was certainly unex-

pected, but a possible interpretation was im-mediate—and exciting. We were measuringthe low-temperature electrical resistance of atiny whisker of the intermetallic compoundUPt 3 to see how defect-free its crystal latticewas, and, as the whisker slowly cooled, itsresistance suddenly fell to zero, a clear in-dication of superconductivity.

What was unexpected was not the super-conductivity per se but its occurrence in amaterial we were investigating as a likelycandidate for the greatly enhanced spin fluc-tuations characteristic of almost ferromag-netic materials. This phenomenon reflects atendency toward magnetism (and hence isoften called near magnetism). and a largebody of experimental evidence supports theview that. like ferromagnetism and supercon-

An attractive interaction between pairsof electrons with parallel spins maybe responsible for the observed but unex-pected superconductivity of Upt 3.(Adapted from a drawing by authorJames L. Smith.)

LOS ALAMOS SCIENCE Spring 1984

ductivity, substantial spin fluctuations andsuperconductivity are mutually incompatible.Not one of the thousands of known super-conductors had exhibited convincingevidence of enhanced spin fluctuations, norhad any of the few known “spin fluctuators”*exhibited superconductivity.

How, then, did we interpret what we hadseen? The idea immediately came to mindthat perhaps UPt3 is a “p-state” supercon-ductor (see “Superconductivity and SpinFluctuations”). This type of superfluidity,which would not be incompatible with spinfluctuations, had been considered twenty-oddyears ago as a generalization of the BCStheory and had been observed a decade agoin liquid helium-3 at millikelvin temperatures.Many other materials had been examined aspossible p-state superconductors because of

their relatively large magnetic susceptibilities,but all had failed a crucial test involvingextreme sensitivity of the superconductingtransition to lattice defects. Could UPt3 bethe first?

Before recounting the tale of our work onUPt3, we point out that it is but one of manyesoteric materials we investigate not only fortheir inherent scientific interest but also fortheir possible technological value. (Spin fluc-tuators, for example, are related to catalystsand hydrogen-storage media. ) The materialsare drawn from the alloys and intermetallic

compounds of the elements known as thetransition elements, the lanthanides, and theactinides. All of these elements are char-acterized by the presence of electrons in innerd or f shells, and the variable behavior ofsuch electrons is responsible, on an atomiclevel, for the diversity found in the crystallinesolids containing the elements. In some casesthe electrons are localized on the ions in thelattice; in others the electrons are itinerant,that is, are free to move about the lattice asdo conduction electrons in metals. Theseextremes of behavior can result in magnetismand superconductivity, respectively. Ofparticular interest are those materials inwhich the electrons are “indecisive,” easilypushed toward one or the other extreme.Among these materials had been found twoof the three known spin fluctuators, not tomention the two known “heavy-fermion”

superconductors (of which more later).

Why UPt3?

Our interest in UPt3 as a possible spinfluctuator was aroused in the fall of 1982,when J. J. M. Franse, Universiteit Amster-dam, sent us a collection of papers by his

*We use the term “spin fluctuator” as shorthandfor a material exhibiting enhanced spin fluctua-tions.

59

Spin Fluctuations

E lectrons in normal metals area normalFermi liquid in that, no matter howstrong their interaction, the particles

retain the essential properties of a non-interacting fermion system: the ground-statedistribution of particles can be characterizedby a Fermi surface in momentum space(which is spherical if one neglects anisotropyintroduced by the lattice); excited states canbe placed in one-to-one correspondence withthose of a free-electron gas: the specific heatvaries linearly with the temperature; and soon. In superconductors, on the other hand, asfirst shown by Bardeen, Cooper, and Schrief-fer in their microscopic theory developed in

by David Pines

1957, both the ground state and the excitedstates of the system are altered in a funda-mental way. A net attractive interaction be-tween pairs of particles near the Fermi sur-face gives rise to an instability of the normalstate, and the superconducting ground statebecomes a single quantum state, the con-densate, which is a coherent superposition ofbound particle pairs and which can flowwithout resistance. To produce single-particleexcitations from the superconducting groundstate requires a finite amount of energy, theenergy gap, so that the specific heat of asuperconductor is drastically altered fromthat of a normal metal.

In BCS theory the net attractive interac-tion between conduction electrons near theFermi surface arises from the exchange ofphonons, the quanta of crystal lattice vibra-tions. The coherent pairs that make up thecondensate are in °S states (that is, stateswith zero total spin and angular momentum),corresponding to pairs of particles with op-posite spins and momenta. Other pairings,such as p-state pairing (in which the con-densate would be a coherent superposition ofpairs of particles with parallel spins) or d-state pairing, are in principle possible; how-ever, both experiment and microscopic calcu-lations to date suggest that where electron-

group on various magnetic and nearly mag-netic systems, Among the papers was one byP. H. Frings and coworkers entitled "Mag-netic Properties of UXP ty Compounds ,”which had been presented during the summerat a magnetism conference in Kyoto thatnone of us had been able to attend. In this

paper were data on the specific heat andmagnetic susceptibility of UPt3 at temper-atures above I kelvin. These data clearlyhinted at enhanced spin fluctuations.

60

One sign of such behavior is a magneticsusceptibility! whose order of magnitude liesapproximately midway between that of a

unit per mole) and that of s ferromagnetic

ported a susceptibility of 0.8 x 10 -2

emu/mole for UPt3, a value of the right orderof magnitude and similar to those of the twoother known metallic spin fluctuators TiBe2

(discovered at Los Alamos) and UA12.

(Liquid helium-3, the other spin fluctuatorknown at the time, is nonmetallic.) Anothersign of near magnetism is an increase in thesusceptibility at some high magnetic field,indicating the transition from near mag-netism to magnetism. Such an increase oc-curs for UPt3, according to Frings et al.,

between 150 and 200 kilogauss. A final. suresign of enhanced spin fluctuations is an in-crease, rather than a steady decrease, in thespecific heat with decreasing temperature.

Spring 1984 LOS ALAMOS SCIENCE

p-state Superconductivity

phonon interactions are sufficiently strong asto bring about superconductivity, an s-statecondensate will be energetically favorable.

Under some circumstances a normalFermi liquid may become almost fer-romagnetic in that particle interactions giverise to internal magnetic fields that act toenhance substantially the usual Pauliparamagnetic susceptibility. In such a systemlow-frequency spin fluctuation excitationsare likewise greatly enhanced, and the strongcoupling of particles near the Fermi surfaceto these spin fluctuations (sometimes calledparamagnons) leads to an effective mass thatis frequency (and temperature) dependent. Asignature of this dependence is a term in thespecific heat that varies with temperature asT 3 1n T (compared to the T 3 variation char-acteristic of normal Fermi liquids). Threesuch almost ferromagnetic metallic Fermiliquids have thus far been discovered: TiBe2,UA12, and, most recently, UPt3.

Liquid helium-3 is an example of a fermionsystem that is both nearly ferromagnetic and,at temperatures less than 2 millikelvins,

superfluid (the analogue for neutral systems superconductivity of purely electronic origin.of superconductivity). Its specific heat con- UPt3 appears to be a particularly promisingtains a T 3 In T term, and neutron-scattering candidate for such an electronic analogue ofexperiments provide direct evidence for the liquid helium-3. Not only might it be the firststrongly enhanced low-frequency spin fluc- metal for which electron interactions alonetuation excitations responsible for that give rise to superconductivity, but its identifi-behavior. It is moreover a p- state superfluid; cation as an anisotropic superfluid couldthat is, the condensate is formed from open the way to a quite new family ofcoherent combinations of pairs of particles of superconducting phenomena, in much theparallel spins in 3P1 states. This p-state super- same way as the study of superfluid helium-3fluidity is not an accident: the short-range has vastly expanded our understanding ofrepulsion between helium-3 atoms is so neutral superfluid phenomena. ●

strong that s-state pairing is strongly sup-pressed, and the interplay between that Laboratory consultant David Pines is Professor of

strong repulsion and the Pauli principle is Physics and Electrical Engineering and a memberof the Center for Advanced Study at the Univer-

responsible for the almost ferromagnetic sity of Illinois. He has carried out pioneeringbehavior. Put another way, the particle cor- studies on classical and quantum plasmas, elec-relations responsible for the enhanced spin trons in metals, collective excitations in solids,

fluctuations tend to oppose s-state super-superconductivity, superfluidity, nuclear struc-ture, and, most recently, on superfluidity and the

fluidity and to favor formation of a p- state internal structure of neutron stars. the theory of

condensate, in part as a result of the particle- compact x-ray sources, and elementary excita-tions in liquid helium-3 and helium-4. He is the

spin fluctuation coupling. author of three books and serves as editor ofIt is natural therefore to hope that in Frontiers in Physics and Reviews of Modern

Physics. The latest of his many honors was receiptmetals exhibiting strongly enhanced spin in 1983 of the newly established Friemann Prize influctuations, one might possibly have p-state Condensed Matter Physics,

This increase follows from the presence of a Superconductors.” The intermetallic com- quality. By June of ’83 we had grown someterm proportional to T 3 1n T in the elec- pound CeA13 was regarded as a likely mem- crystals in the form of tiny whiskers (seetronic specific heat, Frings et al, reported an ber of this class and yet showed no supercon- “’Single Crystals from Metal Solutions”). Theupturn in (he Iow-temperature specific heat of ductivity. A study of UPt3 might help explain best measure of the quality of a metallicU P t3 but gave no detailed analysis of its why, since UPt3 and CeA13 have the same

temperature dependence. In light of these crystal structure.

suggestive data. we planned a more thoroughinvestigation of UPt3. The Serendipitous Experiment

We were also interested in UPt3 because of

our research on the new class of materials Before proceeding with our plans fordescribed in the sidebar “Heavy-Fermion we wanted single crystals of very

LOS ALAMOS SCIENCE Spring 1984

crystal is its chemical resistance near ab-solute zero. At such low temperatures theresistance is due primarily to scattering ofelectrons from lattice defects since scatteringfrom lattice vibrations is suppressed. The

UPt3, resistance of the whiskers was still droppinghigh at 1.3 kelvins, our lowest easily obtainable

61

temperature, and so we planned on furthermeasurements in our dilution refrigerator,which can attain temperatures as low as 0.01

kelvin (see “Getting Close to AbsoluteZero”). But first we worked on improvingsample quality and size.

In August we cooled the UPt 3 in therefrigerator but obtained no data because ofa problem with the electrical leads. Sinceexperiments in the refrigerator are extremelytime-consuming. we delayed another attempton UPt3 until samples of several other ma-terials were ready and could be cooled at thesame time.

On Friday, September 16 all the sampleswere in the refrigerator and well chilled. SinceUPt3 held (we thought) the least promise ofinteresting results, it was the last sample to bemeasured. So it was at 4:30 p.m. when wesaw the resistance of the whisker start toplummet at 0.54 kelvin (Fig, 1). Then began amonth of intensive effort to confirm whatwould be a remarkable discovery-the coex-istence of superconductivity and enhancedspin fluctuations.

Were We Right?

Our first concern was whether the ob-served zero resistance was due to UPt3 itselfor to some other, undetected superconduct-ing phase. Even if present as only a minorconstituent (say 1 percent), such a phase canproduce misleading indications of supercon-ductivity in measurements of both resistanceand magnetic susceptibility. The simplest testfor bulk superconductivity (that is, of themajor phase) is to measure the susceptibilityof the sample as a ground powder. Grindingbreaks up any field-excluding layers formedby a superconducting minor phase, and themeasured susceptibility more truly representsthe behavior of the major phase.

We immediately carried out this test on a

ground powder of UPt3, using an apparatuscooled by simple evaporation of liquidhelium-3 and thus much less time-consumingthan the dilution refrigerator. By Sunday,

●●

Fig. 1. Data obtained during the first measurement of (he low-temperature electricalresistane of a single crystal of UPt3. The abrupt disappearance of resistance, a sign ofsuperconductivity, was quite surprising since we regarded UPt3 as a likely spinfluctuator.

Fig. 2. Specific heat data for unannealed UPt3 whiskers at temperatures greater than1.5 kelvins. The curve is a least -squares fit to the temperature dependence predicted fora spin fluctuator. The extremely good fit constitutes strong evidence of enhanced spinfluctuations in UPt3.

62 Spring 1984 LOS ALAMOS SCIENCE

p-state superconductivity

T his descriptive name has recently been attached to a class of superconductors inwhich the effective mass of the fermions (in this context, electrons) responsiblefor the superconductivity is several hundred times greater than the mass of a

free electron. Such a large effective mass implies that the electrons behave less like agas of independent particles (the usual picture of conduction electrons) and more like aliquid of interacting particles The unmistakable sign of a heavy-fermion superconduc-tor is an extremely large value of y, the proportionality constant relating electronicspecific heat to temperature. (The effective mass is deduced from this parameter,which is determined experimentally by extrapolating low-temperature specific heatdata to absolute zero.)

Two heavy-fermion superconductors are now known: CeCu2Si2, the first, andUBe13. In each case the superconductivity is surprising (so much so that the fact wasinitially reported, almost apologetically, in a footnote) since near room temperaturethe magnetic susceptibility follows a temperature dependence like that of a materialwith local magnetic moments. Thus magnetism (an ordering of the moments), notsuperconductivity, is the expected response at lower temperatures.

We became involved in heavy-fermion superconductivity by being the first to growhigh-quality single crystals of CeCu2Si2. These crystals helped to dispel some of theconfusion about the properties of this material, which had varied wildly from sampleto sample. Then, in collaboration with H. R. Ott and H. Rudigier of EidgenossischeTechnische Hochschule-Honggerberg, we showed that UBe13 was another heavy-fermion superconductor. Its properties are very similar to those of CeCu2Si2 andfortunately vary little among different samples.

The existence of a second example has made heavy-fermion superconductivitymore appealing for study but as yet little better understood. More examples must befound before interesting questions about the phenomenon, such as whether p-state

September 18 we had some disappointing dication of superconductivity, this time downnews-the ground powder was not supercon- to 0.050 kelvin. We measured the specific

ducting down to 0.45 kelvin. More measure- heat of UPt3 at temperatures down to 1.5

ments followed. We cooled the powder in the kelvins. and the news from this front was

dilution refrigerator but again found no in- good, The data fitted beautifully to the

LOS ALAMOS SCIENCE Spring 1984

– T3 1n T dependence predicted for a spinfluctuator (Fig. 2).

We now knew that UPt3 was a bona fide

spin fluctuator and that the ground powderwas not a superconductor. Why, at thispoint. did we persist with further. perhapsfruitless. tests for superconductivity’? We hadseveral reasons. One was the lack of areasonable suspect for a superconductingsecond phase. Uranium is a superconductor,but its presence in UPt3 is not to be expectedsince two other phases of the uranium-platinum system (UPt and UPt2, neither ofwhich are likely superconductors) are closerin composition to UPt3, and a second phaseis usually adjacent to the major phase incomposition. In addition, crystals in the formof whiskers are generally free of other phases.A second reason was the behavior of a singlecrystal of UPt3 prepared by Franse’s groupin a totally different way than our samples.(Franse had sent this crystal to us earlier asan encouragement to measure its heat

capacity in a magnetic field.) We had

measured its susceptibility> in the dilutionrefrigerator along with that of the groundpowder and found a superconducting tran-sition at 0.35 kelvin. This fact made thenegative result from the ground powder moresuspect than the positive result from thewhiskers. The final reason for persistencewas the chance that our initial interpretationwas correct. If UPt3 was a p-state supercon-ductor, our measurements on a groundpowder could easily be misleading sincegrinding introduces defects into the latticethat would be extremely destructive of p-statesuperconductivity. (p- State superconductiv-ity is more strongly inhibited by lattice de-fects than is s-state superconductivity be-cause the effective diameter of the interactingelectron pairs is greater and thus encom-passes a greater number of defects.)

Fortified by these arguments (hopes?), weproceeded to look for the only sure sign ofbulk superconductivity in UPt3-a large up-ward step in its specific heat curve. A super-conducting second phase present at a con-

63

centration of less than about 5 percent (thelimit we had established by x-ray diffractiontechniques) would produce some increase,

depending on its concentration. but the in-crease would be nowhere near that expectedif UPt3 itself was a superconductor. (TheBCS theory predicts an increase of about 150percent.)

The whiskers we could gather at the timefor the specific heat measurement amountedto only 20 milligrams, but, fortunately. wehave developed techniques and equipment formeasuring specific heats of very small sam-ples. We spent nine days hovering over therefrigerator, and by Friday. September 30 thedata definitely showed a sizable discon-tinuity. However, because of experimentaldifficulties below 0.3 kelvin, there remained anagging uncertainty about its precise shape.

Such an important discovery deserved thebest possible data, so we decided to repeatthe heat capacity measurements, this timeusing annealed whiskers. (We had learnedfrom susceptibility measurements in thehelium-3 apparatus that annealed whiskershad much sharper superconducting tran-sitions, and this increased sharpness wouldbe reflected in the heat capacity curve.) Sincewe were running out of whiskers, we took theunannealed ones out of the refrigerator, an-nealed them, and had them cold again byMonday, October 3. That weekend turn-around was the fastest we had ever achieved.

As shown in Fig, 3. the specific heat of ourannealed single crystals of UPt3 increased byonly about 50 percent, and the transitionwas quite broad (and had been even broaderfor the unannealed crystals). Nevertheless. anincrease of this magnitude unequivocallyruled out the possibility that the supercon-ductivity was due to a minor second phase,We now felt confident that superconductivityand enhanced spin fluctuations coexisted inUPt3.

During these experiments we had re-peatedly attempted to produce better samplesand had significantly increased the size of thecrystals but not their lattice perfection. In

Single Crystalsfrom Metal Solutions

Given a free choice, any solid-state experimentalist would characterize amaterial by making measurements on a single crystal rather than apolycrystalline sample. A single crystal more accurately represents the

material (since it is free of grain boundaries at which impurities can hide) and is in factrequired for measuring the directional dependence of various properties. Yet growinga single crystal can be exceptionally difficult, and a large number of importantexperiments await the preparation of appropriate single crystals.

Numerous techniques exist for growing crystals, but finding one that works for aparticular material can be frustrating and time-consuming. A method we use quiteoften in our research is growth from slowly cooled solutions of the desired materiaI ina molten metallic solvent, (This method is an easy extension of the observed naturalgrowth of single crystals from aqueous solutions.) We have used as solvents suchmetals as aluminum, iridium, tin, copper, bismuth, and gallium, The solvent provides aclean environment for crystal growth, and the relatively low temperature at whichgrowth occurs often results in low defect concentrations. Offsetting these advantagesis the possibility that solvent atoms may appear at lattice sites and in voids of thecrystal. In addition, one must find a container that is not attacked by any componentof the solution and a chemical to remove the solvent without attacking the crystal Wehave built up a collection of workable “recipes” and are constantly including new“ingredients.” Still, success demands a certain flair.

When applying this technique to a new material, one unknown is always present:the material may be one that nature simply refuses to provide as nice crystals, Also,the appropriate phase diagram is usually lacking, Then we must rely on educatedguesses and hunches, since determining the phase diagram for a system of at leastthree elements is not a job to undertake merely for exploratory work on crystalgrowth.

To grow the single crystals of UPt3, we used bismuth (melting point: 280 degreesCelsius) as the solvent. As usual, the phase diagram for the system was not available.But we knew from published work that UPt3 has a melting point of 1700 degreesCelsius and is chemically quite stable, that reasonably large amounts of uranium andplatinum can be dissolved in bismuth at temperatures on the order of 1000 degreesCelsius, and that compounds of both uranium and platinum with bismuth exist. Butthe shapes of the uranium-bismuth and platinum-bismuth phase diagrams indicatedthat these compounds are not exceptionally stable, Our guess—that UPt3 wouldcrystallize preferentially-was correct, provided that the solution was not cooledbelow about 1100 degrees Celsius (where a competing crystallization takes place). Weobtained good yields by using atomic percentages of uranium, platinum, and bismuthin the ratio of 1:3:4 and an initial temperature of 1450 degrees Celsius. Since thattemperature is near the boiling point of bismuth, we sealed the crucible in a tantalumcan to prevent its evaporation, We used a crucible of BeO rather than the more usualAl2O3 because uranium might attack Al2O3 at such a high temperature.

As we improved the technique, we obtained crystals of UPt3 with a length of up to 1centimeter and a cross section of 1 millimeter by 1 millimeter, Nature shows her hand

64 Spring 1984 LOS ALAMOS SCIENCE

p -state superconductivity

fact, low-temperature resistance measure-ments indicated that the lattice perfectionwas near the limit expected for a compoundlike UPt3. Therefore we felt that the accuracyof’ the specific heat data could be increasedfurther only by better calibration of the calo-

Fig. 3. Specific heat data for annealed UPt3 whiskers near the temperature at whichthe resistance of the whiskers fell to zero. The sizable discontinuity rules out thepossibility that a minor superconducting phase was responsible for the zero resistance.

Fig. 4. The remarkably different resistance-versus-temperature curves of UPT3, a

possible p-state superconductor, and UBe13, a heavy-fermion superconductor. Exlain-ing this and other differences presents an interesting challenge to theory.

rimeter. We ran a piece of high-purity copperand then, as a check on the systematic errors.ran one-half of that piece. We finished thecalibration by October 13 and had the manu-script in the mail to Physical Review Letters

on October 18.

What Next?

We have now two new superconductors.UPt3 and UBe13 (see “Heavy Fermion Super-conductors”), as different from each other(Fig. 4) as they are from all other supercon-ductors (except CeCu2Si2). Many questions

come to mind about these materials: the mostintriguing is that of p -state superconductivity.Of the tests that have been proposed for thisphenomenon, we mention the more obvious.

One test we plan to carry out in collabora-tion with a group at the University of Califor-nia, Riverside. is to measure the shift of thenuclear magnetic resonance frequency ofplatinum-195 in UPt3. (A similar measure-ment is already in progress on beryllium-9 inUBe13. ) This “Knight shift” is due to shield-ing of the nucleus from an applied magneticfield by the counter magnetic field of theconduction electrons. The predicted tempera-ture dependence of the Knight shift in thevicinity of the transition temperature is quitedifferent for s- and p -state superconductors.

A test we have already mentioned is sensi-tivity to lattice defects. Our measurements onthe ground whiskers of UPt 3, although sug-gestive. need considerable elaboration. Inparticular, we must demonstrate that thesensitivity to magnetic defects is equal to(rather than greater than, as is the case withs-state superconductors) the sensitivity tononmagnetic defects. The difficulty with thistest is finding suitable magnetic impurities to

LOS ALAMOS SCIENCE Spring 1984 65

incorporate into the lattices of these materials junction between an s- and a p -state super- Clearly, much work remains to be done,

(nonmagnetic impurities come free). Another conductor. However, a poor junction would but (the data now available at least refute the

test is based on the fact that a “supercurrent” also kill a supercurrent, and good junctions conventional wisdom of a dichotomy be-

would not flow through a loop containing a are extremely difficult to prepare. tween superconductivity and a tendency

Getting Close to

L iquid helium-4 and helium-3 rank withvacuum as sine qua nons for manyscientific experiments. Some phenom-

ena occur only at temperatures achievablewith these unusual liquids, and others be-come much more tractable to theoreticalinterpretation.

Gaseous helium-4 occurs on the earth as aproduct of alpha decay and is found inreasonable concentrations in some naturalgas fields. It was first liquefied in 1908 byHeike Kamerlingh Onnes (whose discoveryof superconductivity soon followed). Tem-peratures between about 1 kelvin and theboiling point of liquid helium (4.2 kelvins)can be attained simply by pumping on theliquid. The atoms crossing the liquid-vaporphase boundary absorb heat, and the remain-ing liquid cools. Somewhat lower tem-peratures (routinely down to between 0.5 and0.3 kelvin, depending on the system) can bereached by pumping on liquid helium-3. (Thisstable but naturally extremely rare isotope isa by-product of the manufacture of nuclearweapons.) For both liquids the lower temper-

Absolute Zero

ature limit is set not by freezing (as it is fornormal liquids) but by a rapid decrease invapor pressure.

Even lower temperatures (down to about0.005 kelvin) can be reached with a “dilu-tion” refrigerator. This device exploits thenatural tendency of liquid helium-3 to“evaporate” into the “mechanical vacuum”of liquid helium-4. (These two liquids, despiteboth consisting of isotopes of the same ele-ment, interact very weakly because one(helium-4) follows Bose-Einstein statisticsand the other follows Fermi-Dirac statistics.)The atoms of helium-3 absorb heat (cor-responding to the heat of evaporation) asthey cross the phase boundary between thesetwo dissimilar liquids. The lower temperaturelimit is set not by a decrease in the “vaporpressure” as the temperature falls but by adecrease in the heat of “evaporation.”

The accompanying diagram illustratesschematically the continuous operation of adilution refrigerator. Liquid helium-3 dis-solves in liquid helium-4 in the mixingchamber, and the dilute solution is pumped to

a heated still where helium-3 evaporatespreferentially. For economy the helium-3 iscondensed and the liquid returned to thesystem. The photograph shows author Jef-frey O. Willis examining a UPt3 whisker inthe cryostat of the Physical MetallurgyGroup’s dilution refrigerator. A dewar con-taining liquid helium encloses the cryostatwhen the refrigerator is operating, Abouttwenty-four hours are required to cool asample to the desired temperature.

Temperatures in helium-3 and helium-4evaporation refrigerators are determinedsimply by measuring the vapor pressure.Thermometry in a dilution refrigerator in-volves use of a material whose magneticsusceptibility is known to be quite closelyinversely proportional to the temperature.The susceptibility versus temperature curvefor this material is calibrated against vaporpressure measurements in a helium-3 evapo-ration refrigerator. and lower temperaturesare obtained by extrapolation. ■

66 Spring 1984 Los ALAMOS SCIENCE

p-state superconductivity

toward magnetism. Granted, the two fluctuations and superconductivity are retical studies of superconductivity. Perhaps

phenomena had been found to coexist in known to coexist on the same electrons and David Pines’ interpretation is correct, and

ErRh 4B4, but in that material they originate at the same temperature. These results U P t3 i s a metallic analogue of liquid

on different electrons. Now in UPt3 spin breathe new life into experimental and theo)- helium-3. ■

LOS ALAMOS SCIENCE Spring 1984 67

AUTHORS

Gregory R. Stewart has shifted the emphasis of his research twice sincejoining the Laboratory in 1977: from superconductors, particularly thosewith high transition temperatures, to nearly magnetic materials and, finally, tothe recently discovered heavy-fermion superconductors. The connectionfound in UPt3 among these diverse interests joins the finding (with L. R.Newkirk and F. A. Valencia) of the new A-15 structure superconductorNb3Nb as his most exciting discoveries at Los Alamos. Author of more thanfifty refereed papers since 1977, Greg is a specialist in specific heatmeasurements but enjoys learning other research techniques, such as usinghigh magnetic fields to alter the electronic properties of materials. He earnedhis B.S. from Caltech and his Ph.D. from Stanford. A postdoctoral appoint-ment followed at the University of Konstanz in West Germany, where hisresearch included resistivity and Hall-effect measurements on solar-cellmaterials. Recently he spent a three months’ sabbatical at Kernforschungs-zentrum Karlsruhe.

Zachary Fisk was educated at Harvard University and theCalifornia, San Diego. He received his Ph.D. in physics at theunder Bernd Matthias. A postdoctoral year at Imperial College,

University oflatter in 1969London, was

followed by a year as Assistant Professor of Physics at the University ofChicago. He returned to UCSD, becoming Research Physicist and AdjunctProfessor of Physics before joining the Laboratory in 1981 as a staff memberin the Physical Metallurgy Group of what is now the Materials Science andTechnology Division. His research interests include the low-temperatureelectrical and magnetic properties of metals and the growth of single crystalsof these materials. The latter interest originally developed from a consultingagreement with Bell Laboratories.

Jeffrey O. Willis earned his B.S. and Ph. D degress in physics from theUniversity of Illinois, Urbana-Champaign, in 1970 and 1976, respectively.He then spent two years at the Naval Research Laboratory in Washington,D. C.. as a National Research Council Research Associate studying thesuperconductive properties of new materials at ultralow temperatures. In1978 he joined the Laboratory as a postdoctoral fellow in the CondensedMatter and Thermal Physics Group of the Physics Division. There heinvestigated the magnetic and superconductive properties of materials primar-ily using the Mossbauer effect. In 1980 he became a staff member in thePhysical Metallurgy Group of what is now the Materials Science andTechnology Division. He is currently engaged in the study of new materials atultralow temperatures. using the dilution refrigerator, and at very highpressures, using diamond anvil cell and other techniques. He is a member ofthe American Physical Society.

Spring 1984 LOS ALAMOS SCIENCE

p-state superconductivity

AUTHORS

James L. Smith received his B.S. in physics from Wayne State University in1965 and his Ph.D. in physics from Brown University in 1974. He has been astaff member in the Physical Metallurgy Group of what is now the MaterialsScience and Technology Division since 1973. In 1982 he was appointed aLaboratory Fellow for his scientific insight and experimental expertise. Hiswork began at Los Alamos on materials at dilution refrigerator temperatures.his special expertise then. That work evolved into addressing the question ofhow superconductivity in elements from the left side of the periodic tablecrosses over to magnetism in elements from the right side. This has led tointeresting speculation on such things as catalysts and the stability of stainlesssteel. He is now a leader in the field of actinide materials and gives severalinvited talks on the subject each year at various conferences and workshops.Despite his experience, he was completely surprised by the behavior of UPt3

reported in this article.

Further Reading

P. W. Anderson and P. Morel. “Generalized Bardeen-Cooper-Schrieffer States and the Proposed Low-Temperature Phase of Liquid He3.” Physical Review 123(1961):19 11-1934.

R. Balian and N. R. Werthamer. “Superconductivity with Pairs in a Relative p Wave.” Physical Review’131( 1963): 1553-1564.

W. A. Fertig, D. C. Johnston, L. E. DeLong, R. W. McCallum, M. B. Maple, and B. T. Matthias."Destruction of Superconductivity at the Onset of Long-Range Magnetic Order in the CompoundErRh4B4.” Physical Review Letters 38(1977):987-990.

F. Steglich, J. Aarts, C. D. Bredl, W. Lieke, D. Meschede, W. Franz, and H. Schafer. “Superconductivityin the Presence of Strong Pauli Paramagnetism: CeCu2Si1. ” Physical Review Letters 43(1979): 1892-1896.

Angelo L. Giorgi, Gregory R. Stewart, James L. Smith. and Bernd T. Matthias. “High TemperatureSuperconductivity: A Metallurgical Approach.” Los Alamos Science Vol. 1, No. 1(1980):28-39.

G. R. Stewart. “Measurement of Low Temperature Specific Heat.” Review of Scientific Instruments54( 1983) 1-11.

J. L. Smith and E. A. Kmetko. “Magnetism or Bonding: A Nearly Periodic Table of Transition Elements.”Journal of the Less-Common Metals 90(1983):83-88.

P. H. Frings, J. J. M. Franse, F. R. de Boer, and A. Menovsky. “Magnetic Properties of UXp ty

Compounds.” Journal of Magnetism and Magnetic Materials 31-34(1983):240-242.

H. R. Ott, H. Rudigier, Z. Fisk, and J. L. Smith. “UBe13: An Unconventional Actinide Superconductor.”Physical Review Letters 50(1983): 1595-1598.

G. R. Stewart, Z. Fisk, and J. O. Willis. “Characterization of Single Crystals of CeCu2Siz. A Source ofNew Perspectives.” Physical Review B 28(1983): 172-177.

“Second Heavy-Fermion Superconductor.” Physics Today, December 1983, pp. 20-22.

G. R. Stewart, Z. Fisk, J. O. Willis, and J. L. Smith. “Possibility of Coexistence of Spin Fluctuations andSuperconductivity in UPt3.” Physical Review Letters 52(1984):679-682.

LOS ALAMOS SCIENCE Spring 1984 69