Germanium, doped with copper and cooled to the temperature of
liquid helium, is ordinarily a good insulator. But it becomes an astonishingly good
conductor of electricity when squeezed along one axis of its crystalline structure. The
surprising discovery was made by Oscar Dubon, Eugene Haller and Wladyslaw Walukiewicz of
the Material Sciences Division.
 Eugene Haller and Wladyslaw
Walukiewicz found that by compressing copper-doped germanium at low temperature,
resistance dropped a trillion- fold, permitting new studies of semiconductor
transitions. Photo by Roy Kaltschmidt
|
"A little pressure induces a change of over a dozen orders of magnitude--one
trillion fold!--in conductivity," says Haller, pointing to a plot that displays the
resistivity of a copper-doped germanium sample abruptly dropping as pressure increases.
The increase in conductivity is so rapid and so large that the effect may be useful in
designing instruments to detect extremely small changes of temperature or pressure. Even
more important, the phenomenon offers condensed-matter researchers a new way to perform
detailed studies of the insulator-metal transition in semiconductors.
What makes semiconductors so useful in the electronics industry and elsewhere is that
their conductivity can be changed from insulating to conducting behavior by adding minute
quantities of certain impurities, a process called "doping." In the well-ordered
crystalline structure of an undoped semiconductor, the overlapping orbitals of the
electrons form distinct energy bands. The lower-energy valence bands are full of
electrons--so full that the electrons cannot move--while the higher-energy conduction
bands, where electrons can move, are empty. If there are no extra charge carriers in these
bands, the crystal is an insulator.
In a semiconductor like germanium, the bands lie relatively close together on the
energy scale, and even a little heat can give valence-band electrons enough energy to leap
to the conduction band, which makes the crystal a conductor. The effect can be enhanced or
modified by doping. If some of the pure crystal's atoms are replaced by others, such as
phosphorus atoms, electrons are added to the conducting band. Doping with boron atoms
leaves positively charged "holes" in the valence band, which also allows charge
to flow.
At sufficiently high temperatures, doping germanium with copper can contribute holes to
germanium's valence band. But at low concentrations, the average separation of copper
atoms is much larger than the size of the outermost copper orbitals, and the charge can
move only by "hopping" between adjacent copper sites. Since this is an unlikely
process, the germanium crystal still acts as an insulator.
"We decided to take germanium that was doped with copper, but was still a good
insulator, and apply uniaxial stress," Walukiewicz explains. It was a simple matter
to apply over four kilobars of pressure [over 4,000 times atmospheric pressure at sea
level] to a one-by-one-by-five millimeter chip of copper-doped germanium by squeezing it
in a vise.
What Dubon, Walukiewicz and Haller found was that pressure applied along one direction
changed the electron orbitals around the copper atoms, effectively enlarging these
orbitals enough so that even at low concentrations they overlapped and permitted
conduction.
The overlapping orbitals of the dopant atoms themselves form their own separate band,
with properties that depend on charge and electron spin. Solid-state researchers have been
intrigued by these doping-induced bands since they were first proposed by theorist John
Hubbard almost 40 years ago.
In most semiconductors, however, the higher-energy Hubbard component merges with the
crystal's valence band near the insulator-metal transition point, making Hubbard bands
exceedingly difficult to study. Uniaxially stressed copper-doped germanium is free of
these problems and represents a pristine case of isolated Hubbard bands.
"You can say we have a new, smaller-gap semiconductor within the germanium
semiconductor," says Walukiewicz. "The placement of the copper is random, but
its electronic structure is a completely delocalized extended state."
Walukiewicz sees opportunities for many areas of research, such as studying random
systems "using germanium as a matrix to keep copper in place: random but well
defined. Random systems are much more difficult to study than crystals, but also much more
common."
Haller agrees that the possibilities are numerous. "We have seen this huge
physical effect, an increase in conductivity of many orders of magnitude achieved simply
by applying pressure. And it is intrinsically interesting to be able to study the Hubbard
band in a new state, clearly separated from the valence band of germanium. We can do that
for the first time."
Photo: Eugene Haller and Wladyslaw
Walukiewicz found that by compressing copper-doped germanium at low temperature,
resistance dropped a trillion-fold, permitting new studies of semiconductor transitions.
Photo by Roy Kaltschmidt (XBD9807-01754-01.tif)
