PHYSIK & ASTRONOMIE_Festkörperforschung

An ice-cold experiment: The samples researchers at the Max Planck Institute for Chemical Physics of Solids are investigating are cooled in this dilution fridge to below minus 273 degrees Celsius. This is done by carefully lowering the instrument into a well-insulated cryostat. Photo: Sven Döring

48 MaxPlanckResearch 2 | 11 PHYSICS & ASTRONOMY_Superconductivity

Superconductivity Is Pair Work

Electric cables that routinely conduct electricity without loss – physicists have been motivated by this idea ever since superconductivity was discovered 100 years ago. Researchers working with Bernhard Keimer at the Max Planck Institute for Solid State Research in Stuttgart and Frank Steglich at the Max Planck Institute for Chemical Physics of Solids in Dresden want to gain a detailed understanding of how unconventional superconductors lose their resistivity.

TEXT ROLAND WENGENMAYR

pril 8, 1911. At the Univer- electrical resistivity of the mercury Although superconducting technolo- sity of Leiden in the Neth- drops abruptly to zero. And there it re- gy has become indispensable in basic erlands, two men sit in a mains. The researchers’ first reaction is research, it is encountered on a routine darkened booth. Only their that it must be an error. However, sub- basis only in magnetic resonance im- instruments tell them what sequent experiments show that they agers, because the complex cooling is A is happening in the cryostat next door. have discovered a new phenomenon. expensive. The solution would be su- A man in his late fifties with a striking In 1913, Kamerlingh Onnes called this perconductivity at room temperature, walrus mustache fiddles with this effect superconductivity. but such materials have remained a enormous thermos flask: Heike Kamer- dream. Since 1993 it has not been pos- lingh Onnes has been famous ever WAITING FOR SUPERCONDUCTORS sible to raise the temperature record since he became the first person to suc- FOR EVERYDAY USE any higher. It currently stands at mi- ceed in liquefying helium in 1908. nus 138 degrees Celsius (135 Kelvin) This enabled him to reach the ex- Superconductors have excited the and is held by a high-temperature ce- tremely low temperature of 4.2 Kelvin, imagination ever since they were dis- ramic superconductor with the com-

which is around 4 degrees above abso- covered. Even Kamerlingh Onnes plicated name of HgBa2Ca2Cu3O8 (Hg: lute zero, at minus 273.2 degrees Cel- dreamed of transporting electricity in mercury, Ca: calcium, Ba: barium, Cu: sius. For this achievement, Kamerlingh power grids with no losses whatsoever. copper, O: oxygen). Onnes would be awarded the Nobel The discovery of so-called high-temper- Bernhard Keimer relates that compa- Prize for Physics in 1913. ature superconductors in 1986 at the ny representatives recently asked him The cryostat now contains a small IBM research laboratory near Zurich when superconductivity at room tem- glass tube filled with mercury. The created a great deal of euphoria. The perature could be expected. “They were Leiden-based physicists want to use it two men who made the discovery, Karl disappointed when I told them that no to observe, for the first time, how the Alex Müller and Johann Georg Bed- one had an answer yet,” he says, smil- electrical resistivity of metals behaves norz, were awarded the Nobel Prize for ing. The Director at the Max Planck In- when the temperature approaches ab- Physics as early as 1987. A more sober stitute for Solid State Research in Stutt- solute zero. At 4.2 Kelvin, minus 269 atmosphere has now returned, howev- gart and this year’s recipient of the degrees Celsius, something unexpected er, as has happened so often in the his- prestigious Leibniz Prize ought to know happens: the display indicating the tory of superconductors. – after all, he has been researching high-

2 | 11 MaxPlanckResearch 49 » PHYSIK & ASTRONOMIE 50 (pyramids). oxide planes copper the along move pairs Cooper the (blue), barium (green), copper (red) and oxygen (gray), yttrium of consists and YBCO of name the by goes which superconductor, Layers with zero resistivity: In the perconductors canbedeveloped. su- create thebasisonwhicheveryday terials losetheirresistivity–inorderto tailed understandingofhowthesema- years. Thescientistwantstoobtainade- temperature superconductors formany MaxPlanckResearch MaxPlanckResearch Onnes dreamed of transporting electricity in power grids with no losses whatsoever. have excitedSuperconductors imagination the since ever discovered. were they Even Kamerlingh 2 | 11

_Festkörperforschung This isalsothemotivationforre- temperatures. become superconducting low atvery called heavyfermionsystems,which Steglich’s teaminDresdenworksonso- ventional superconductors,” although searchers work inthefieldof“uncon- Keimer haveincommon:bothre- man scienceisnotallthatSteglichand awarded themostvaluableprizeinGer- Dresden. Thehonorofhavingbeen tute forChemicalPhysicsofSolidsin lich, aDirectorattheMaxPlanckInsti- search being undertaken byFrankSteg- regular spatiallattice.Nature lovesor- the atomsorganizethemselves intoa inwhichmaterials consistofcrystals the physicsofsolids. derstood onlybyimmersing oneselfin derives fromtheirinitials–canbeun- –thename this work.TheBCStheory received theNobelPrizeforPhysics to crackthistoughnut.In1972they Cooper andRobertSchrieffermanaged three AmericansJohnBardeen,Leon emphasizes. Itwas1957beforethe ity firstneedstobeexplained,”Steglich classical, conventionalsuperconductiv- conventional superconductivity means, “In ordertounderstandwhatun- Like mostsolids,superconducting es betweenthepeaksandtroughs ofthe rangement oftheatoms:if distanc- Steglich. Thisisduetotheuniform ar- meeting anyresistance,”explains Frank without move throughaperfectcrystal “The conductionelectrons caneven TWO ELECTRONS GENTLE VIBRATIONS BIND a positiveelectriccharge. This lossleavesthe“atomiccores”with theelectriccurrent. tion electronscarry in semiconductors.Thesefreeconduc- they requireacertainenergytodothis plete freedomthroughthelattice,but the atomsandmovewithalmostcom- Someofthemescapefrom the crystal. ever, notallelectronsactonlyasgluefor each other. they canbindneighboringatomsto cles, arealsospatiallyextendedwaves, ics: sinceelectrons,asquantumparti- only withtheaidofquantumphys- bond, betweentheatoms.Thisworks trons providetheglue,chemical trons amongeachother. Theseelec- that, astheycansharecertainelec- succeedindoingjustatoms incrystals der ifithelpstosaveenergy. Andthe In metalsandsemiconductors,how- 12

Photo: David Ausserhofer for the German Research Foundation (top, 2), graphic: MPI for Solid State Research PHYSICS & ASTRONOMY_Superconductivity

1 To gain a better understanding of high-temperature superconductivity, Bernhard Keimer and his colleagues use the apparatus in the far right of the picture to produce thin crystalline layers with customized superconducting properties. 2 Daniel Pröpper and Bernhard Keimer observe how the cryostat in the foreground cools the sample of a high- temperature superconductor that loses its resistivity only at temperatures far below freezing point.

electron wave match the distances be- binds it with the first electron, al- The properties of the Cooper pairs are tween the atoms, the electron can leap though the two particles violently re- radically different from those of the unhindered through the crystal. pel each other because they have the electrons. Pursuant to the rules of quan- But perfect crystals do not exist. Ev- same, negative charge. Only the large tum mechanics, electrons possess a ery crystal lattice has its flaws. They do temporal and thus also spatial separa- half-integer spin. They thus belong to not match the rhythm of the electron tion of up to one micrometer – a thou- the group of particles known as fermi- waves, which is why they literally rico- sandth of a millimeter – between the ons. These are quite acquisitive, each chet off them. And even if crystals with two allows the marriage in the soft lat- requiring a quantum state for it alone. perfectly ordered atoms did exist, the tice bed. For comparison: the distance In the Cooper pair, the two half-integer temperature would prevent the current between the atoms in the crystal lat- spins of the two electrons either sub- from flowing with zero resistivity, be- tice is around one thousand times tract to a total spin of zero or, more cause the higher it becomes, the stron- smaller, around one tenth of a nano- rarely, they also add up to one. ger and faster is the vibration of the meter – one nanometer is a millionth Quantum particles with integer crystal’s atomic lattice – like a mattress of a millimeter. spin, however, belong to the group of being jumped on with wild abandon. particles known as bosons. Bosons are This also interferes with the crystalline POWERFUL VIBRATIONS so sociable that they all like to con- perfection and hinders the wave of con- DESTROY THE COOPER PAIRS dense together into one quantum state duction electrons. if the heat energy in the system be- In conventional superconductors, This is how two free electrons form one comes low enough for this to happen. the vibrations of the atomic mattress be- Cooper pair, which is named after Leon For Cooper pairs, this is the supercon- come slow and gentle at low tempera- Cooper, one of the three fathers of BCS ducting state in which many Cooper tures – as happens in essentially all ma- theory. His idea provided the break- pairs permeate each other to form one terials. When a conduction electron through for the explanation of conven- “macroscopic” quantum state. This races through the lattice on its path, tional superconductivity in 1957. In or- large quantum object can simply “slip it electrically attracts the positively der to understand the properties of through” the many small flaws in the charged atomic cores. With a soft lattice, Cooper pairs, their “spins” are also re- crystal without noticing them. This is the atomic cores totter along sluggishly quired. To put it somewhat noncha- how the electrical resistivity disappears. behind the electron and only return to lantly, spins describe a sort of quantum “In superconducting cables, this mac- their original position much later. pirouette by the particles, turning them roscopic quantum state can extend “In conventional superconductors, into tiny quantum magnets. According over kilometers,” says Bernhard Keimer. this lattice vibration is extremely slow to the strict rules of quantum physics, “That’s quite remarkable!” compared to the motion of the elec- however, spins are not allowed to align In conventional superconductors, trons,” emphasizes Frank Steglich. themselves with each another arbitrari- the Cooper pairs, which have a wide “This is very important!” If a second ly. In the Cooper pairs, they can be ori- spatial extent, always have a total spin electron now passes, it is attracted by ented only parallel or antiparallel to of zero. Decisive for the properties of this more concentrated positive one another, like rotary switches snap- conventional superconductors are the charge. As a consequence, the lattice ping into place. lattice vibrations, which are slow, com-

2 | 11 MaxPlanckResearch 51 PHYSICS & ASTRONOMY_Superconductivity

» There is still no conclusive theory, but one thing is clear: superconductivity and magnetism work in close collaboration in unconventional superconductivity. This is surprising, because magnetism is pure poison for conventional superconductors.

pared to the speed of the electrons, and This is surprising, as magnetism is pure tional superconductivity was first dis- which bind the two conduction elec- poison for conventional superconduc- covered in 1979. The researchers in trons to a Cooper pair. Since crystal lat- tors. “A magnetic atom contamination Dresden know this forefather of the tices vibrate with ever-increasing speed of less than one percent in the crystal heavy fermion superconductors so well and amplitude as the temperature in- already destroys its superconductivi- in the meantime that they can adjust creases, this destroys the Cooper pairs. ty,” explains Frank Steglich. Magnetic its properties when growing crystals: a Conventional superconductivity can atoms introduce an electron into the tiny excess of copper turns it into a su- therefore be observed only at relatively crystal lattice, whose unshielded spin perconductor, a small deficit into an low temperatures. acts as a small magnet at the site of the antiferromagnet. High-temperature superconductors atoms. If, in a conventional supercon- In antiferromagnets, the electrons become superconducting at much ductor, an electron of an extended responsible for the magnetism, which higher temperatures. They must there- Cooper pair now passes by, the local in cerium are called 4f electrons after fore contain a much stronger Cooper- spin forces it to reverse its spin appro- their position in the electronic shell, al- pair glue in order for the lattice vibra- priately. But this love affair destroys ways orient themselves in the direction tions not to rip the pairs apart. Many the weak bosonic marriage with the opposite to that of the neighboring indications point to the fact that these distant “Cooper partner.” If this hap- atom. Their overall magnetic field thus quantum superglues function in a pens too often, the superconductivity cancels out perfectly on average. This completely different way than lattice breaks down. magnetic order is called antiferromag- vibrations. This is thus taken to be the netism, because it is the counterpart to characteristic of unconventional su- COOPER PAIRS WITH the ferromagnetism with long-range ef- perconductivity. There is still no con- PARTICULARLY STRONG GLUE fect, which takes its name from iron clusive theory, but one thing is clear: (lat. ferrum). superconductivity and magnetism act With unconventional superconductors, In CeCu2Si2, as a typical example, in close collaboration in unconven- in contrast, certain forms of magnetism the local 4f electrons develop a partic- tional superconductivity. seem to be really helpful. This also ap- ularly strong influence on the free con- plies to heavy fermion superconduc- duction electrons at very low tempera- tors, which have fascinated Frank Steg- lich for more than three decades: “We 1 Frank Steglich (top left), Oliver Stockert currently know almost 40 of them!” and Steffen Wirth (in the window) Most heavy fermion superconductors prepare measurements using a scanning lose their resistivity only at very low tunneling microscope for very low temperatures. “If we have around 2 Kel- temperatures. The microscope images atomic structures and stands on its vin, or minus 271 degrees Celsius, we own base that reaches down to the are very happy,” says Steglich’s col- rocky foundation in order to protect league Steffen Wirth, explaining the the apparatus from vibrations. challenge. Nevertheless, the Dresden- The soundproof room, whose roof was based researchers hope that such super- removed for the work, serves the same Ce purpose. conductors can also help them contrib- 2 Oliver Stockert (left) and Steffen Wirth ute to solving the mystery of the gluing Cu adjust a dilution fridge, which generates mechanism in high-temperature super- temperatures of below minus 273 Si conductors. But what exactly is a heavy degrees Celsius. fermion? 3 Experiments with a crane: Oliver The particles can be found in a crys- Stockert instructs a colleague who is tal lattice whose fundamental building lowering a superconducting magnet An unconventional superconductor: The blocks always include a magnetic atom: into a cryostat that sits on its own material made from cerium (blue), copper foundation. The researchers use the (turquoise) and silicon (yellow) owes its cerium, for example, takes on this role superconducting magnet to investigate special properties to specific electrons in in the cerium-copper-silicon com- how the physical properties of their samples depend on the magnetic field. the rare earth metal cerium. pound CeCu2Si2, in which unconven- Graphic: MPI for Chemical Physics of Solids, photos: Sven Döring (3)

52 MaxPlanckResearch 2 | 11 PHYSIK & ASTRONOMIE_Festkörperforschung

1

2 3 1

2 3

tures: put simply, they bind them nets. They thus become probes that re- against the normal metallic state, which together to form a quantum fluid with act to all electrons with a “free” mag- is magnetically unordered. “And if two the viscosity of honey. The electrons netic moment. are busy quarrelling, the third can take contained therein now sluggishly drag Oliver Stockert, another colleague advantage of the situation,” says Wirth. themselves through the lattice – up to of Frank Steglich, recently used this Stockert adds: “Very strong spin fluctu-

a thousand times slower than normal method to investigate CeCu2Si2 at the ations that produce Cooper pairs occur conduction electrons. “These new qua- neutron source of the Institut Laue- especially at the point where the two si-particles thus behave like electrons Langevin in Grenoble. In the process, cancel each other out.” that apparently have a mass up to one his team discovered that the magnetic With spin fluctuations, the particles thousand times greater,” says Steglich, interactions in the lattice were indeed involved waggle their spins and influ- “which is why we coined the term 20 times as strong as the gluing of the ence each other in this way – a disre- heavy fermions in the late 1970s.” Cooper pairs would require. “This is spectful way of putting it. The short- The heavy fermions also form the obviously how they make supercon- lived magnetic fields created in this Cooper pairs in such superconductors. ductivity possible,” says Stockert. The way are sufficient to strongly bond the Their sluggishness means that the two measurements with neutrons do not electrons to form Cooper pairs. Spin partners must cuddle up close in order provide proof, but the smoking gun, fluctuations also play a crucial role in to be able to bond with each other. so to speak. His colleague Steffen high-temperature ceramic supercon- However, this makes the electrical re- Wirth is employing further methods in ductors. Most, if not all, of the physi- pulsion between them all the stronger. parallel in order to conclusively com- cists conducting research in this field The glue between them must therefore plete the picture. are convinced of this, says Bernhard be much more powerful than in con- Keimer – and he includes himself here. ventional Cooper pairs. This is why it SHORT-LIVED MAGNETIC FIELDS “Although here we have only one has been clear for more than three de- BIND THE ELECTRONS electron system, unlike the heavy fer- cades that the conventional BCS mech- mion superconductors,” he explains. anism does not work here. This idea is now taking shape. The Coo- This means that there are no local 4f Although the theoretical physicists per pair glue is actually magnetic. Its electrons here to create a tendency for proposed alternative glue mechanisms force originates from spin fluctuations latent antiferromagnetism. Instead, it is early on, solids are complicated struc- of the heavy fermions. In the general only the conduction electrons that tures comprising enormous numbers chaos, they form small “bubbles” with bring about the interaction between of atoms and electrons that exert a re- short-range magnetic order that form magnetic order and superconductivity. ciprocal influence. This makes it very and disappear rapidly. These fluctua- Most copper oxide compounds, in- difficult to “take a look inside” them tions particularly exert their effect close cluding the high-temperature super- in experiments. Neutrons provide one to a state known as the quantum critical conductors, are even insulators. Their possibility: these nuclear particles are point. This point can be reached via low complex crystal lattice has a sandwich- small and electrically neutral, so they temperatures, the correct chemical mix- type structure. Layers of copper and ox- can penetrate matter almost unhin- ture in the crystal, magnetic fields or ygen atoms play a crucial role in this

dered. Like electrons, however, they pressure. In CeCu2Si2, two very strong ef- process, which is why these brittle ma- have a half-integer spin, which makes fects are competing with each other: the terials are called cuprates. The Cooper

them into well-matched little mag- antiferromagnetic order is fighting pairs move along these copper oxide Photos: Sven Döring (top left and right), MPI for Chemical Physics of Solids (bottom left)

54 MaxPlanckResearch 2 | 11 PHYSICS & ASTRONOMY_Superconductivity

1 Sample and carrier: The Dresden researchers use the box to store a sample carrier and the cerium copper silicide crystal after they have investigated its magnetic properties with a neutron beam. 2 In order to study the electric and magnetic properties – or more precisely the Hall effect in ytterbium rhodium silicide – the researchers fixed a platelet of the material with varnish and provided it with electric contacts. 3 A research object that can be handled: Frank Steglich holds a glass model of the cerium copper silicide crystal in his hands.

planes, so their superconductivity is “Their motion through such an anti- progress in high-temperature super- strongly two-dimensional – “flat,” as it ferromagnetic background is very conductivity with simple models,” were. The conduction electrons re- slow!” The electrons can save the en- states Keimer optimistically. These quired are provided by foreign atoms ergy they expend here by forming findings may one day contribute to with which the researchers intention- Cooper pairs – which are very small, as developing a material that conducts ally contaminate their crystals. they are in the case of the heavy fer- current at room temperature with zero Bernhard Keimer uses diagrams mion superconductors. With their to- resistivity. Bernhard Keimer keeps this to show how the superconductivity tal spin of zero, they become invisible, long-term objective in mind. “As a probably operates in cuprates. In the as it were, to the antiferromagnetic basic researcher, I have a vision, of copper oxide planes of the crystals, stubble field of half-integer spins. They course, that this research will benefit certain electronic states of the copper slip through without losing any ener- society one day.” atoms overlap each other. These “d-or- gy whatsoever, and the cuprate be- bitals” form the racetrack for the con- comes a perfect electrical conductor. duction electrons. The electrons tend GLOSSARY to a fluctuating antiferromagnetic or- RESEARCH SHOULD SOMEHOW der when approaching the supercon- BENEFIT SOCIETY Transition temperature ducting transition temperature: they The temperature below which a orient themselves antiparallel as they “It is still just an idea for a model,” says material becomes superconducting. get closer to it. Keimer with a smile. A Stuttgart-based Fermion If an electric current now flows, in- team headed by his colleague Vladimir A particle with half-integer spin; fermions such as electrons each require dividual conduction electrons must Hinkov used neutrons at the FRM-II a quantum state of their own. migrate through this spin landscape. research reactor in Garching and ob- Heavy fermion superconductors The rules of quantum physics require served that strong spin fluctuations The unpaired electrons they contain, them to continuously flip the spin of actually do occur. They input the mea- which in some metals stay very close the electrons they encounter to suit surement data without “fine-tuning” to the atomic nucleus, contribute to their own spin. “They leave behind a into a model being developed by the superconductivity. They therefore move extremely slowly, which is why they trail of flipped spins, which is very theoreticians in Stuttgart, explains appear heavier than they actually are. energy intensive,” explains Keimer: Keimer. This then computed a transi- Cooper pair tion temperature at which the sub- Two electrons that are fermions stance would theoretically become su- combine to form a Cooper pair and perconducting. The first shot at 170 become a boson. At sufficiently low Kelvin was almost twice as high as the temperatures, a very large number of Cooper pairs occupy a quantum state substance’s real transition temperature, in which they no longer experience the but Keimer considers this result to flaws and vibrations of the crystal – be very encouraging: “The closer one the resistivity disappears. looks, the simpler the picture gets.” Bosons The researchers in Stuttgart have Particles with integer spin; unlike also perfected the production of artifi- fermions, they all occupy a single quantum state in the ground state. cial crystals, which they now want to use to imitate the layer structure of the Spin fluctuations The spin or intrinsic angular momen- cuprates. They hope this will take tum gives the electron its magnetic them a few more steps along the learn- moment. The orientation of the spin ing curve. “It is possible to make great thus determines the magnetic order in a material. Even before it assumes a certain magnetic order – ferromagnetic Sample mosaic: Bernhard Keimer’s team or antiferromagnetic, for example – this mounts around 100 YBCO single crystals onto order can already form temporarily in a support in order to produce a sample with some regions because the spins fluctu- as large a volume as possible. The researchers ate – that is, change their orientation. use inelastic neutron scattering to investigate

Photo: MPI for Solid State Research the magnetic properties.

2 | 11 MaxPlanckResearch 55