Superconductivity: From Physics to TechnologyTheodore H. Geballe Citation: Phys. Today 46(10), 52 (1993); doi: 10.1063/1.881384 View online: http://dx.doi.org/10.1063/1.881384 View Table of Contents: http://www.physicstoday.org/resource/1/PHTOAD/v46/i10 Published by the American Institute of Physics. Additional resources for Physics TodayHomepage: http://www.physicstoday.org/ Information: http://www.physicstoday.org/about_us Daily Edition: http://www.physicstoday.org/daily_edition

Downloaded 18 Mar 2013 to 136.159.235.223. Redistribution subject to AIP license or copyright; see http://www.physicstoday.org/about_us/terms

SUPERCONDUCTIVITY: FROMPHYSICS TO TECHNOLOGY

It took half o century to understand Kamerlingh Onnes's discovery,and another quarter-century to make it useful. Presumablywe won't have to wait that long to make practical useof the new high-temperature superconductors.

Theodore H. Geballe

Superconductivity was discovered in 1911 by HeikeKamerlingh Onnes in Leiden.1 That discovery can betraced to the steady advance in laboratory techniques forobtaining ever lower temperatures that began when LouisCailletet in France and Raoul Pictet in Switzerland suc-ceeded in liquefying trace amounts of the "permanentgases," nitrogen, air and hydrogen. These gases were sonamed because previous attempts to liquefy them, byMichael Faraday among others, had been unsuccessful.2

Low-temperature physics emerged at about the sametime as Physical Review itself, when Z. F. Wroblewski inCracow succeeded in condensing experimentally usefulquantities of liquid air in 1891. He found that theresistivities of pure metals had curious temperature de-pendences: It looked as if their resistances would vanishat nonvanishing temperatures. This intriguing possibil-ity generated theories of the limiting low-temperaturebehavior that predicted everything from zero resistanceto infinite resistance. Interestingly enough, today's theo-ries still do that!

The following year James Dewar in England inventedthe vacuum-insulated, silver-plated glass vessel thatbears his name, which enabled him to obtain experimen-tal quantities of liquid hydrogen and proceed fartherdown the temperature scale. There he found that themetallic resistivities did not vanish; their temperaturedependences simply flattened out.

Finally, less than two decades after William Ramsaydiscovered that helium exists on Earth, Kamerlingh On-nes succeeded in liquefying it. Liquid helium extendedthe range of temperatures available for experiment down-ward by another order of magnitude. Three years laterKamerlingh Onnes and his student G. Hoist discovered

Theodore Geballe is a professor emeritus in StanfordUniversity's department of applied physics.

the remarkable discontinuous disappearance of the elec-trical resistance of a mercury sample at a critical tem-perature Tc as it was cooled in liquid helium. In furtherexperiments, a persistent current was induced thatshowed no measurable decay. The new phenomenon wascalled "supraconductivity," meaning literally "beyond con-ductivity." That name seems to me more appropriatethan the one we use today.

The Leiden experimenters had high hopes that theywere witnessing the birth of a new electromagnetic tech-nology. They knew that the energy cost of refrigeratinga superconducting electromagnet would be vastly lessthan the Joule heat dissipated in operating a conventionalone. It was soon found, however, that normal resistancereturned below Tc in the presence of a magnetic fieldabove a critical limit Hc or a current above a critical limitJc. Both of these critical parameters, alas, were lowenough to thwart any hopes of prompt technologicaldevelopment.

In 1933 the key discovery of perfect diamagnetismby Walter Meissner and R. Ochsenfeld in Berlin showedsuperconductivity to be a reversible thermodynamic phe-nomenon. There soon followed the two-fluid model ofCornelis Gorter and Hendrik Casimir, and Fritz andHeinz London's equations describing the electrodynamicsof the phenomenon in terms of surface currents that limitthe magnetic field to a penetration length A. BrianPippard's investigation of A as a function of the electron'smean free path led to the nonlocal extension of theLondon theory.

In Russia, investigations of the intermediate state inwhich superconducting regions coexist with normal re-gions over a range of external magnetic fields led in 1950to the Ginzburg-Landau theory,3 which generalized LevLandau's theory of phase transitions to include a spatiallydependent, complex order parameter. The remarkableintuition underlying this new theory became evident

52 PHYSICS TODAY OCTOBER 1993 1990 Americon Insriture of Physics

Downloaded 18 Mar 2013 to 136.159.235.223. Redistribution subject to AIP license or copyright; see http://www.physicstoday.org/about_us/terms

A single crystal of thesuperconducting layered-cuprateceramic YBa,Cu3O7_A, whosecritical temperature is 92 K.The crystal, about 200 micronson a side, was grown by DebraKaiser at the National Instituteof Standards and Technology.The diagonal striations in thispolarized-light micrograph (byL. C. Smith and F. Cayle)indicate twinningstructure. Figure 1

shortly after the historic paper by John Bardeen, LeonCoooper and Robert Schrieffer4 appeared in 1957, whenLev Gorkov derived the Ginzburg-Landau equations fromthe microscopic BCS theory, thus showing the equivalenceof the order parameter with the pair-condensate wave-function.

Phys. Rev. takes center stageAfter World War II a number of factors conspired to shiftthe center of gravity of superconductivity research towardthe US and the Physical Review, though of course muchimportant work continued in other countries. Heliumbecame available commercially as a byproduct of thenatural gas industry. Liquefiers built in specialized low-temperature labs were now supplemented by commer-cially available Collins liquefiers; that opened the fieldto nonspecialists. The helium isotope 3He, a decay prod-uct of the tritium produced for thermonuclear weapons,became available for refrigeration below 1 K, as well asfor study in its own right as a Fermi liquid with fasci-nating ground states.

Research proceeded along two initially separate linesof investigation. The first approach used theory to guideresearch and to explore new quantum phenomena inwell-characterized elemental and simple alloy supercon-ductors. The second approach involved searches for newsuperconductors and studies of the variation of Tc withcomposition. The goal of that effort was to discern apattern in the occurrence of superconductivity. It washoped that such a pattern would reveal the essentialsuperconducting interactions, and that it might also leadto technologically more useful superconductors withhigher critical temperatures.

The more traditional first approach depended upona close coupling between theory and experiment, butthere was as yet no satisfactory theory. Many prominenttheorists, among them the creators of quantum mechan-

ics, had tried to explain superconductivity. Bardeen'sdeep insight into the problem is already evident in a 1941abstract in which he suggests that superconductivitymight result from a sufficiently strong electron-phononinteraction that would temporarily cause the electrons toexperience Bragg reflections from the phonon waves, thusproducing large diamagnetic contributions.5 "The germof an idea was there, and John never ceased to befascinated by it," wrote Conyers Herring in the Bardeenmemorial issue of PHYSICS TODAY (April 1992, page 29).

The discovery of the inverse square-root dependenceof Tc upon mass for the different isotopes of tin, lead andmercury in 1950, by four different groups in the US andBritain, restimulated Bardeen's intense interest in theelectron-phonon interaction. Evidence for this isotopeeffect, predicted in a model proposed in that same yearby Herbert Frohlich,6 had actually been sought experi-mentally at Leiden in the 1920s, when Pb isotopes firstbecame available. In the early 1960s it was shown thatthe isotope effect could vary widely, and even vanish, asa result of the retarded character of the electron-phononinteraction.

The exponential behavior of the heat capacity andthermal conductivity were suggestive of an energy gapin the density of states above the superconducting groundstate. Those clues, among others, guided Bardeen,Cooper and Schrieffer to the inspired discovery of thepairing theory of superconductivity, which they pub-lished in the Physical Review 36 years ago.4

Directly thereafter Physical Review and Physical Re-view Letters published a whole raft of papers that usedthe BCS theory to account quantitatively for a variety ofobservations, including the heat capacity anomalies, theMeissner effect, persistent currents, the temperature de-pendence of the penetration length, and other thermal,optical and acoustic properties. A particularly pleasingaspect of the theory was the significant role of the

PHYSICS TODAY OCTOBER 1993 5 3

Downloaded 18 Mar 2013 to 136.159.235.223. Redistribution subject to AIP license or copyright; see http://www.physicstoday.org/about_us/terms

coherence factor in negating the effect of the large densityof states in acoustic, but not electromagnetic, attenuation.

Two independent measurements of the magnetic fluxquantum in 1961 provided another nice verification ofthe pairing theory.7 Long before the BCS theory, FritzLondon had predicted that if the superconducting wave-function is to be single-valued, magnetic flux must bequantized in multiples of hlq, where q is the charge ofthe electron. The experiments found the flux quantumto be just half that predicted by London, indicating thatthe relevant charge unit is that of a pair of electrons.This important result is true for all the known supercon-ductors, including the high-temperature superconductingcuprate ceramics discovered in the 1980s. (See figure 1.)

TunnelingIvar Giaever was inspired to look for evidence of theenergy gap in the excitation spectra of superconductorsby studying quantum tunneling through barriers. Tun-neling was first proposed by Robert Oppenheimer in1928, to account for the field ionization of atomichydrogen. Leo Esaki's invention of the semiconductingtunnel diode demonstrated that quantum tunnelingwas not only interesting physics but also that it couldbe the basis for technologically important devices.Giaever found direct evidence for the gap in the non-linear current-voltage curves he measured for tunnel-ing through A12O3 barriers obtained simply by allowingaluminum to oxidize in air.8

For superconductors with strong electron-phononcoupling, William McMillan and John Rowell at Bell Labsin 1965 used the BCS theory to develop a tunnelingspectroscopy that demonstrated unequivocally that theattractive pairing interaction was mediated by phononsin all known superconductors. This is still true, exceptfor heavy-fermion superconductors such as UPt3, whichhave more exotic pairing mechanisms, and for the high-Tccuprate ceramics, where the issue remains unclear. Un-fortunately, technical details have thus far prevented thefabrication of good tunnel junctions from the high-Tccuprates.

In 1962, thinking about the possibility of pairedelectrons tunneling through a barrier, Brian Josephsonpredicted the unusual dc and ac properties and quantuminterference effects that came to bear his name.9 Pairtunneling was regarded as improbable until Rowell and

Philip Anderson experimentally confirmed Josephson'spredictions.10 (See figure 2.) Interference between twoparallel superconducting tunnel junctions, which isanalogous to double-slit optical interference, was firstobserved by scientists at Ford.11 They introduced theacronym SQUID for this new superconducting quantuminterference device.

Technology for superconducting high-speed comput-ing was developed at IBM, which eventually abandonedthe effort, in part because of the materials limitations ofthe Pb alloy used for electrodes. Ironically, that was justwhen new techniques were being developed for producingmore rugged and satisfactory Nb/A]/Al2O3/Nb Josephsontunnel junctions. Nowadays these junctions form thebasis of commercial Josephson-junction detectors, mag-netometers, mixers, switches and voltage standards.SQUID arrays are even being used to study noninvasivelythe magnetic signals from neurons firing in the brain.

Searching for new superconductorsThe second approach to the understanding of supercon-ductivity after World War II involved a search for newfamilies of superconductors. Such an approach was nec-essarily empirical. Bernd Matthias and John Hulm pio-neered this enterprise in the early 1950s. Extensiveinvestigations in this spirit led to the discovery of abouta thousand new superconductors. Of course the mostspectacular payoff of this approach came three decadeslater, with the discovery of superconductivity in thelayered cuprates by Georg Bednorz and Karl-Alex Miillerat IBM Zurich.12

Many interesting binary metal-metalloid compoundswere found in the years before the Bednorz-Muller dis-covery, particularly those, such as V3Si and Nb3Sn, thathave cubic A15 structure and therefore linear arrays ofnonintersecting transition metal atoms.13 The highestsuperconducting transition temperatures known at thetime (above 20 K) were found in this family. NbTi, asimple body-centered cubic alloy, has become the work-horse of present-day magnet technology.

Matthias found a useful rule that qualitatively pre-dicts, simply from the average number of valence elec-trons per constituent atom, whether or not a given alloyor binary compound will exhibit superconductivity.14 Ob-served exceptions to the rule turned out to be signals ofeither new physics or unexpected materials science. Ex-

5 4 PHYSICS TODAY OCTOBER, 1993

Downloaded 18 Mar 2013 to 136.159.235.223. Redistribution subject to AIP license or copyright; see http://www.physicstoday.org/about_us/terms

500

amples include the dramatic reduction of Tc by scatteringoff magnetic impurities that violates time-reversal sym-metry and breaks pairs.

It was an unexpected, pleasant surprise in 1961 whenEugene Kunzler and coworkers at Bell Labs found thatNb3Sn samples would support large supercurrents, ex-ceeding 150 kiloamps, in strong magnetic fields (8.8tesla).15 (See figure 3a.) It had taken 50 years todiscover real materials that had the critical currents andfields necessary for making useful magnets and electricpower machinery. It would be another two decades beforethe physics and materials science were understood wellenough for practical production of long multistrand cablesthat could be wound into coils for large, powerful magnetsor rotating machinery. (See figure 3b.)

In the Soviet Union, experimental work by Lev Shub-nikov had already stimulated Alexei Abrikosov in 1957to extend the Ginzburg-Landau theory to include systemsin which the interface energy between normal and su-perconducting regions is negative.16 Abrikosov predicteda new state, now known as type-II superconductivity, inmagnetic fields above the Meissner critical field. Theimplications of Abrikosov's work were not pursued at thetime because, I believe, the two paths being followed inthe study of superconductivity in those days were decou-pled from each other. Nb3Sn research was, in fact,sometimes referred to as "schmutz physics." (Schmutz isGerman for dirt.)

Abrikosov's model is a reversible thermodynamic de-scription that says nothing about the pinning of the fluxlines needed to account for high critical currents in highfields. Charles Bean's critical-state model17 accounts forthe current flow pattern by positing that each volumeelement carries either its maximum critical current or nocurrent at all.

In 1962 Young B. Kim and coauthors and Anderson18

introduced the concepts of flux creep and flux flow. Afterthat the studies became interdisciplinary, because met-allurgical defects such as precipitates, dislocations, voids,grain boundaries, surfaces and interfaces serve as flux-pinning centers. All these imperfections present spatialvariations of free energy that prevent flux quanta fromresponding to the Lorentz force and causing dissipation.Flux-jump instabilities resulting from mechanical or elec-

Josephson pair tunneling. This plot of critical current acrossa Josephson junction vs magnetic field is from the 1963Physical Review Letter by John Rowell that reported the firstconvincing proof of pair tunneling.1" The decrease of thecritical current by three orders of magnitudes at specificsmall field values, as predicted by Brian Josephson,eliminated the possibility that the current was simply beingcarried across the junction by superconducting shortcircuits. Figure 2

tromagnetic fluctuations can lead to catastrophic quench-ing of the stored magnetic field energy. These instabili-ties were brought under control largely through researchimpelled by the high-energy physicists' need for big,powerful magnets for accelerators and particle detectors.

Nowadays large coils of superconducting compositewire capable of safely carrying kiloamp currents arereadily available. A 1-mm-diameter high-current wirecontains as many as 105 filaments of NbTi with a fineprecipitate of hexagonal Ti that provides the pinningcenters. Last year NbTi wire was used to construct morethan $1 billion worth of clinical magnetic resonanceimaging systems. The wire is ready for other technolo-gies, such as magnetic energy storage, rotating machineryand levitated trains, if and when the markets develop.The Superconducting Super Collider and CERN's pro-posed Large Hadron Collider would be inconceivable with-out NbTi superconducting magnets.

High-temperature superconductorsBednorz and Miiller's discovery of superconductivityabove 30 K in ternary perovskite-related cuprate struc-tures12 was soon followed by the discovery of supercon-ductivity above 90 K by Paul C. W. Chu and his colleaguesat the Universities of Houston and Alabama, in a ceramicthat turned out to be the layered compound YBa2Cu307.This discovery of superconductivity above the boilingtemperature of nitrogen generated enormous excitementand research challenges that are still at the forefront ofphysics. (See, for example, the June 1991 special issueof PHYSICS TODAY on high-temperature superconductivity.)

Since the pioneering work of Bednorz and Miiller, arather large family of layered copper oxide superconduc-tors with complicated unit cells has been discovered. Theparent compounds are Mott insulators with propertiesthat are not understandable in terms of the conventionalone-electron theory that has been so successful in ex-plaining metallic, semiconducting and insulating behaviorin most materials. With small changes in composition,these insulators can be doped to form conductors in whichthe carriers are highly correlated. In addition to theirspectacular superconductivity, they also have unusualnormal-state properties. All the evidence indicates thatthe action is on the doped CuO2 planes. The conventionalelectron-phonon interaction does not suffice to explainthe superconductivity of these materials, but there is as

PHYSICS TODAY OCTOBER 1993 5 5

Downloaded 18 Mar 2013 to 136.159.235.223. Redistribution subject to AIP license or copyright; see http://www.physicstoday.org/about_us/terms

20

108

o. 42zo

1

0.8

0.6

0.4

Nb3Sn

• -4.20 / 0 ' 0 5 0 C M x 0.063 CMA - 1 Ci® \

A -4^2 0 / ° - 0 3 0 C M x 0-053 CM

R . 4_|° )• 0.025 CM X 0.025 CM

10 20 30 40 50 60H IN KGAUSS

70 80 K

Niobium superconductors have critical currents and fieldshigh enough to make them useful for large, powerfulmagnets, a: Plots of critical current vs magnetic field forshort Nb3Sn samples of various cross sections at differenttemperatures, from the 1961 Physical Review Letter^announcing the discovery of high-current superconductivityin very high magnetic fields, b: After three more decades ofdevelopment we have Nb-Ti cable for the 6.6-tesla magnetsof the Superconducting Super Collider. The micrographshows the cross section of a strand 0.8 mm in diameter,fabricated at the University of Wisconsin. Each strandconsists of many individual superconducting filaments,6 microns thick, embedded in a copper matrix. (Courtesyof David Larbalestier.) Figure 3

yet no consensus as to the microscopic origin of theelectron pairing interaction. Nowadays, new models areproposed in almost every issue of Physical Review B andPhysical Review Letters.

There is, of course, a much larger potential marketfor devices that can operate at liquid nitrogen tempera-tures than for those that need liquid helium. Electronicand electrical power technologies based on the supercon-ducting cuprates are expected to grow rapidly in the nearfuture. SQUID magnetometers operating at 77 K havealready been used to make magnetocardiograms. Simple77 K SQUIDS are now available commercially. Signal-to-noise ratios in magnetic resonance imaging and incommunications technology can be increased signifi-cantly by the use of superconducting detector coils andfilters. The detection of magnetic anomalies associatedwith petroleum deposits and other geological featurescan benefit from liquid-nitrogen-cooled supercon-ducting gradiometers. There are potentially large mar-kets for superconducting oscillators, antennas and mul-tichip interconnections. Possibilities for powertransmission lines are bright, but they will requirefurther study of dissipation processes.

The history of superconductivity, so much of whichis to be found in the volumes of the Physical Review andPhysical Review Letters, teaches us that the new quantumphenomena of today may well become the new technologyof tomorrow. Presumably we won't have to wait as longfor the tomorrow of liquid nitrogen superconducting tech-nology as we did for the helium-based superconductingtechnology of today.

References1. H. Kamerlingh Onnes, Commun. Phys. Lab. U. Leiden 124C

(1911).2. See, for example, J. Dewar, Encyclopedia Britannica, 11th ed.

(1911), vol. XVI, p. 744.3. V. L. Ginzburg, L. D. Landau, Zh. Eksp. Teor. Fiz. 20, 1064

(1950).4. J. Bardeen, L. N. Cooper, J. R. Schrieffer, Phys. Rev. B 108,

1175 (1957). See also R. Ogg, Phys. Rev. 69, 243 (1946).5. J. Bardeen, Phys. Rev. 59, 928 (1941).6. H. Frohlich, Phys. Rev. 79, 845 (1950).7. B. Deaver, W. Fairbank, Phys. Rev. Lett. 7, 43 (1961). R. Doll,

M. Naebauer, ibid., 51.8. I. Giaever, Phys. Rev. Lett. 5, 147 (1960).9. B. D. Josephson, Phys. Lett. 1, 251 (1962).

10. P. W. Anderson, J. M. Rowell, Phys. Rev. Lett. 10, 230 (1963).J. M. Rowell, Phys. Rev. Lett. 11, 200 (1963).

11. R. Jaklevic, J. Lambe, A. Silver, J. Mercereau, Phys. Rev. Lett.12, 159 (1964).

12. J. G. Bednorz, K.-A. Miiller, Z. Phys. B 64, 189 (1986).13. G. F. Hardy, J. K. Hulm, Phys. Rev. 93, 1004 (1954). B. T.

Matthias, T. H. Geballe, S. Getter, E. Corenzwit, Phys Rev.96, 1435(1954).

14. B. T. Matthias, Phys. Rev. 97, 74 (1955).15. J. E. Kunzler, E. Buehler, F. S. L. Hsui, J. H. Wernick, Phys.

Rev. Lett. 6,89 (1961).16. A. A. Abrikosov, Zh. Eksp. Teor. Fiz. 32, 1442 (1957) [Sov.

Phys. JETP 5, 1174 (1957)].17. C. P. Bean, Phys. Rev. Lett. 8, 250 (1962). H. London, Phys.

Lett. 6, 162(1963).18. P. W. Anderson, Phys. Rev. Lett. 9, 309 (1962). Y. B. Kim,

C. F. Hemstead, A. R. Strnad, Phys. Rev. Lett. 9, 306 (1962).See also M. R. Beasley, R. Labusch, W. W. Webb, Phys. Rev.181,682(1969).

19. M. K. Wu, J. Ashburn, C. J. Torng, P. N. Hor, R. L. Meng, L.Gao, Z. J. Huang, Y. Wang, C. W. Chu, Phys. Rev. Lett. 58,908(1987). •

56 PHYSICS TODAY OCTOBER 1993

Downloaded 18 Mar 2013 to 136.159.235.223. Redistribution subject to AIP license or copyright; see http://www.physicstoday.org/about_us/terms