杏吧原创

IN the autumn of 1857, in a cluttered laboratory at the University of
Glasgow, the great physicist William Thomson was muttering to himself as he
stalked back and forth over a floor littered with wires and magnets. Thomson,
who later became Lord Kelvin, was vexed by his latest experiments. Their
purpose? To see whether a magnetic field would affect the flow of electricity
through a material.

It鈥檚 not that Kelvin was having problems doing the experiments. As he found,
a magnetic field does have an effect. There鈥檚 no doubt about it. The trouble
was, no matter what material he tried and how strong he made the magnetic field,
he just couldn鈥檛 alter the electrical resistance of anything by more than an
uninspiring 5 per cent. Disappointed, Kelvin dubbed the effect
鈥渕agnetoresistance鈥� and soon got on with something else.

But history might have been different. In 1988, physicists in Paris repeated
Kelvin鈥檚 experiments. They had an advantage over Kelvin, of course, in the form
of the modern materials laboratory. Instead of simple elements, they used thin
films made of more than one material, and found that at very low temperatures, a
magnetic field could cut by half the resistance of a sandwich of alternating
ultrathin layers of iron and chromium. The effect was called giant
magnetoresistance.

Then, in 1994, in a class of strange substances called perovskites, American
and British physicists stumbled over just the kind of inexplicable effect that
Kelvin had been looking for鈥攁 magnetically triggered reduction in
electrical resistance by 10 000 times. Kelvin had been poking around in fertile
ground after all. He just never had a chance to test the right stuff.

Colossal magnetoresistance (CMR), as this effect is known, is just the kind
of thing that gets scientists鈥� hearts racing. It鈥檚 nice when theory and
experiment agree, but it鈥檚 even better when an experiment produces completely
unexpected results. What is it about these perovskites that makes them respond
so dramatically to magnetic fields? What peculiarities of their composition make
them behave like nothing science has ever seen? Such questions have been driving
theorists to distraction, tempting materials scientists to bake up ever more
unusual samples, and prompting information technologists to dream about magnetic
discs that could pack every last letter of the 600 scientific papers Kelvin
wrote into a spot no bigger that a grain of sand.

So what is a perovskite? It is, in fact, a rather boring black and brown
mineral, named after a Russian statesman. But the name is given also to other
materials that have similar crystal structures. These include high-temperature
superconductors, as well as the substances that exhibit colossal
magnetoresistance. A compound of lanthanum, calcium, manganese and oxygen is an
example of the latter.

Cubic conundrum

What makes CMR so intriguing is that the structure of this material isn鈥檛
especially exotic (see Diagram).
It鈥檚 just simple cubes stacked
together like boxes in a supermarket. At the centre of each cube is a manganese
atom, which is surrounded by a diamond-like cage of six oxygen atoms. Calcium
and lanthanum atoms inhabit the cube鈥檚 corners. Whether a corner site has one
atom or the other is a matter of chance, but they appear in a precise numerical
ratio x which, when varied, leads to materials with different
properties. This structure is a bit more subtle than that of a material in which
the atoms are arranged in a regular, non-random way. Still, it seems rather
unremarkable. And yet this perovskite confounds some of physicists鈥� most
fundamental expectations.

In general, the electrical properties of a material depend on how its
electrons move. In a material with low resistance, electrons can slip through
without running into many of the material鈥檚 atoms. But in one with high
resistance, they move more like an army of disoriented soldiers running through
a forest, slamming into trees, bouncing off, and getting up to run again.
Resistance comes about because electrons get hung up or suffer deflections when
they interact with atoms.

A material鈥檚 magnetic properties depend on electrons too, but in a different
way. Electrons have spin as well as charge. As a result, they act like tiny
magnets, and their orientations determine whether a material is magnetic or not.
Ordinarily, the motions of electrons and the orientations of their spins don鈥檛
really affect one another very much. So whether or not a material is magnetic
usually has nothing to do with its electrical resistance. But things are
different with the perovskites.

Suppose you were to begin cooling a sample of the perovskite with x
= 0.3. For a while the resistance would creep up slowly, but nothing very
interesting would happen. Near 250 K, however, you would see the material鈥檚
electrical resistance suddenly plummet. At nearly the same temperature, the
material also becomes magnetic. This happens because the spins of the outer
electrons of the manganese ions make these self-same ions act like minuscule
magnets, which interact with one another and try to line up. At high
temperatures, constant thermal jostling forces the magnets to point in all
directions. But below 250 K, they overcome the reduced jostling and line up,
making the material magnetic. 杏吧原创s call this ordering temperature the
Curie temperature.

So this material鈥檚 electrical and magnetic properties change together. A
coincidence? In 1951, physicist Clarence Zener of Bell Laboratories in New
Jersey, didn鈥檛 think so. He suspected that something more interesting was going
on, and suggested that a subtle interaction between the motions and spins of
electrons in the perovskite might account for it.

In the x = 0.3 material, calcium atoms occupy 30 per cent of the
corner sites. Lanthanum atoms fill the rest. Both donate to the lattice
electrons, which can move and carry electrical currents. But the calcium atoms,
because they can donate fewer electrons than lanthanum atoms, cause a shortfall
in the sea of donated electrons. This gives the electrons some room to move
around. But these 鈥渋tinerant鈥� electrons interact with other 鈥渇ixed鈥� electrons
that are attached to manganese sites. These 鈥渇ixed鈥� electrons give these sites
their spins. And the itinerant electrons must move past these sites. The
question is: why, when the spins of the 鈥渇ixed鈥� electrons line up, can the
itinerant electrons then travel from site to site far more easily?

Zener鈥檚 explanation had two parts. He first pointed out that an 鈥渋tinerant鈥�
electron, in passing by a manganese site, tends to line up its spin with that of
the manganese ion. Spins interact, and this represents a savings in energy. He
then argued, using quantum theory, that such an electron would have a better
chance of moving past a nearby site if the spins of the first and second sites
were aligned. In loose terms, this is because electrons are not liable to change
their spin orientations as they travel. If the two sites have their spins
aligned, then the second will not resist the arrival of the electron since it
already has the 鈥渃orrect鈥� orientation. But if the electron were to try moving
past a site with a misaligned spin, a repelling force would push it away.

Spin thing

This so-called 鈥渄ouble exchange model鈥� attempts to explain how the manganese
perovskites behave in the absence of a magnetic field, and why the high to low
resistance and non-magnetic to magnetic transitions occur near the same
temperature. Above the Curie temperature, the spins on the manganese atoms are
all askew, and an 鈥渋tinerant鈥� electron cannot pass freely from site to site.
Below this temperature, the spins on the manganese sites line up, and the
鈥渋tinerant鈥� electrons move from site to site more easily
(see Diagram 1)
(see Diagram 2)
(see Diagram 3).

What does this have to do with CMR? Back in 1951, Zener was not thinking
about the effects of an applied magnetic field. But in 1994, when physicists
discovered that near 250 K an applied field cuts the resistance of the
x = 0.3 material by 10 000 times, they thought that Zener鈥檚 model might
explain it.

The thinking ran like this: without an applied magnetic field, the manganese
spins align below the Curie temperature. By applying a sufficiently large
magnetic field, however, the spins can be made to line up even above this
temperature. So, magnetoresistance in the perovskites isn鈥檛 all that surprising.
Forcing the spins of the manganese ions to align by applying a magnetic field
should allow the 鈥渋tinerant鈥� electrons to travel easily, dramatically reducing
the resistance.

A puzzle solved? Not quite. In 1995, Andy Millis, Peter Littlewood and Boris
Shraiman, then all at Bell Labs in New Jersey, put numbers into the model to
work out how high the resistance should be when the manganese spins are askew.
Their calculated resistance was far too small to explain the data. What鈥檚 more,
the predicted temperature dependence was wrong. Zener鈥檚 theory doesn鈥檛 explain
CMR.

So why is the magnetoresistance effect so huge? Or, to put it another way,
why is the resistance of the high resistance state so very large? Millis, now at
the Johns Hopkins University in Baltimore, has been working on the answer. He鈥檚
been trying to work out the consequences of a deceptively mundane
observation鈥攖hat the manganese ions aren鈥檛 actually all identical. That鈥檚
because when the calcium ions donate electrons to the lattice, and there aren鈥檛
enough to go round, some manganese sites bind an electron and some do not.
Accordingly, the material contains both Mn⁴⁺ and Mn³⁺ ions.

This is important, because when an electron moves from one manganese site to
another, it stays there for a time, altering the number of electrons associated
with each manganese ion and effectively converting Mn⁴⁺ into Mn³⁺.
So what? Well, the Mn³⁺ ion, with its extra electron, is significantly larger
than the Mn⁴⁺ ion. It is also less symmetrical. So, according to Millis鈥檚
theory鈥攁nd similar schemes proposed by theorists at Los Alamos and
elsewhere鈥攁n electron moving from one site to another carries a heavy
trail of distortions with it
(see Diagram 1)
(see Diagram 2).
This takes a lot of energy, which is why the resistance is so great.

In a magnetic field, however, if the manganese spins are aligned, then the
electrons move more freely between individual sites. As a consequence of their
quantum wave-like nature, these electrons are effectively 鈥渄e-localised鈥� and not
associated with any one manganese site. Spread out instead over many sites, such
electrons no longer cause strong distortions, or trail them behind. In this way,
it is possible to understand why the CMR response is so enormous. It鈥檚 all down
to a strange connection between the magnetic, electronic and structural
characteristics of the perovskites. Strong evidence for this theory comes from
physicists at the University of California and Los Alamos National Laboratory,
New Mexico, who have measured the distances between the manganese and oxygen
atoms (Physical Review Letters, vol 80, p 853).

Near and far

In a material without lattice distortions, the manganese-oxygen distances
should all be more or less the same. But if lattice distortions are occurring in
the high resistance state, the variation in these distances should be much
greater, because the distortions should increase the distance between some
atoms, and decrease the distance between others. And that is precisely what the
measurements show鈥攁 broad distribution of manganese-oxygen distances
in the material. So it looks as if this idea of distortions may well be
right.

The theory, however, still isn鈥檛 complete. In trying to pin down the
subtleties of electrons鈥� motions in a perovskite, physicists face a difficult
task. The particles鈥� movements and spins interact so strongly with each other,
and with the rest of the ions in the material, that everything needs to be taken
into account at once. The traditional mathematical approximations that
physicists have relied on for 70 years 鈥攊n which electrons move through a
fixed lattice impervious to their actions鈥攄on鈥檛 work. So, although the
basic picture is there, a detailed explanation will take some time. But that
hasn鈥檛 hindered attempts to put CMR to use.

Because CMR materials respond so strongly to magnetic fields, they should
make good sensors for use, for example, in computers. Information is stored on
computer disc drives in patterns of microscopic magnetised regions. To retrieve
this information, 鈥渋nductive read heads鈥� can scan a spinning disc. The changing
magnetic fields created by the information on the disc generate corresponding
electric currents in the read heads. That鈥檚 the traditional method. But heads
that use magnetoresistance work better.

These detect the regions of magnetisation on the disc and change their
electrical resistance accordingly. Even when made from 鈥渙rdinary鈥�
magnetoresistive materials, these heads work faster than inductive heads. In
January, however, IBM shipped the first disc drives to employ heads made from
giant magnetoresistive materials鈥攖he kind discovered in 1988 in Paris.
These drives can store 2.69 billion bits of information per square inch-which
translates to 16 billion bits of information per disc drive.

Putting CMR to work might one day permit far more information to be packed
onto a disc. But there鈥檚 some way to go before it becomes viable. For example,
even in the low resistance state, CMR perovskites are still not terribly good
conductors. This means that tiny fluctuations in current, which occur naturally,
cause relatively large fluctuations in voltage. So any devices made from the CMR
perovskites now available would suffer from high noise.

The hope is that the perovskite CMR materials are just the first of a host of
new materials yet to be discovered. In the past few years, physicists have also
found CMR in other compounds鈥攖hings known as pyrochlores, chalcogenide
spinels and silver chalcogenides. In these materials, the mechanism behind CMR
may be simpler than in the perovskites. In the pyrochlores, the magnetic and
electronic properties originate from two distinct ionic lattices that
interpenetrate. Distortions of these lattices seem to be unimportant, and there
are not many 鈥渋tinerant鈥� electrons, so it may be easier to gain a full
theoretical understanding of the pyrochlores before we ever come to terms with
the perovskites. Armed with such understanding, we might be able to design new
materials with better properties.

Then again, new materials may lurk out there. Like Lord Kelvin, we might be
sampling only a few materials and missing out on others yet more spectacular.