|
|
Researchers have taken the first step toward building silicon-based computers that use a fraction of the power of today's machines. A team has injected electrons into silicon in such a way that their spins, or magnetic orientations, tend to be aligned in one direction instead of the other. Although the reported effect is subtle, silicon has never before supported such attempts to implement spintronics-the manipulation of electrons by their spins instead of their charges. "To us this is the holy grail of semiconductor spintronics," says physicist and electrical engineer Ian Appelbaum of the University of Delaware in Newark, who worked on the experiment. "As long as Intel is making CPUs out of silicon, we're going to have to learn how to manipulate spin in silicon."
One promise of spintronics is to slash waste heat in computers. In normal electronics, electric fields propel electrons, which shed heat in the process. In contrast, spins can move around on their own or under a magnetic field without producing much heat. But researchers had only shown they could control electron spin reliably in more niche varieties of semiconductor such as gallium arsenide, which is used in cell phones. In the new device, Appelbaum and co-workers inject electrons from a layer of aluminum through a thin layer of ferromagnet (a permanent magnet) and into a pure silicon crystal. Aluminum has a 50-50 mix of spin up and spin down electrons--the two possible orientations. The ferromagnet, however, blocks electrons of one spin while letting the others flow into the silicon. The researchers found that their ferromagnet barrier gave silicon a one percent excess of one spin type versus the other, at a temperature of 85 kelvins, they report in a paper published online today in the journal Nature. A real spintronics device would need to produce a pure stream of one spin type, Appelbaum says, adding that he expects to achieve "vast improvements" on this front in the near future. "Right now we're focused on fundamental materials science," he says. "Integration obviously is a goal, but at the moment it's not on the near term horizon." The experiment brings the same spintronics tools to silicon that researchers have developed for other materials, says physicist David Awschalom of the University of California, Santa Barbara. In principle, he says, the effect should work at room temperature, which would allow researchers to study silicon spintronics in more realistic conditions. "It's a beautiful piece of work," he says. |
||||||||||||||||
KEY CONCEPTS
|
Diamond has a track record of extremes, including ultrahardness, higher thermal
than any other solid material and transparency to ultraviolet light. In addition,
diamond has recently become much more attractive for solid-state electronics,
with the development of techniques to grow high-purity, single-crystal synthetic
diamonds and insert suitable impurities into them (doping).
Pure diamond is an electrical insulator, but doped, it can become a semiconductor
with exceptional properties. It could he used for detecting ultraviolet light,
ultraviolet light-emitting diodes and optics, and high-power microwave
electronics. But the application that has many researchers excited is quantum
spintronics, which could lead to a practical quantum computer-capable of
feats believed impossible for regular computers - and
ultra-secure communication.
Spintronics is an advanced form of electronics that harnesses not just the
electrical charge of electrons (as in conventional electronics) but also
a property called spin that makes electrons act like tiny bar magnets. Your
computer probably already contains the first and most rudimentary commercial
application of spintronics: since 1998 hard-drive read heads have used a
spintronic effect called giant magnetoresistance to detect the microscopic
magnetic domains on a disk that represent the 1s and 0s of the data it
contains.
Another spintronic device, one that you may find in new computers in the
next few years, is magnetoresistive random-access memory (MRAM). As with
a hard drive, MRAM stores information as magnetization and therefore is
nonvolatile, meaning that the data are not lost when the device's power is
turned off. The read-out is done electrically, just like any other charge-
based memories [see "Spintronics," by David D. Awschalom, Michael F. Flatté
and Nitin Samarth; SCIENTIFIC AMERICAN, June 2002]. Free-scale Semiconductor,
a spin-off of Motorola, began selling the first MRAM in 2006.
Nonvolatile memory chips could lead to computers that will not need to reload
programs laboriously from a hard drive every time they are switched on. Instead
a computer would be ready within a fraction of a second to proceed from where
it left off (much like handhelds today) because all the necessary programming
and data would remain ready and waiting in the chip. More advanced spintronic
technologies that are in the early research stages-such as spin transistors,
which would make use of spin in controlling current flow-could enable computer
chips with logic circuitry capable of being reconfigured on the fly.
Quantum Electronics
Devices such as read heads and MRAM chips represent one class of spintronics,
in which the spins of large numbers of electrons are aligned the same way,
as with a collection of toy tops all spinning clockwise on the floor. These
so-called spin-polarized electrons typically flow through some part of the
device, forming a spin-polarized current, or spin current, that is highly
analogous to a polarized beam of light. Researchers have made many exciting
advances in this area in the past few years, including discovery of ways
to generate and manipulate spin polarization in semiconductors without relying
on a magnetic material or relatively bulky wiring to generate a magnetic
field. In particular, our group and others have observed a potentially very
useful phenomenon called the spin Hall effect [see "Spin Control for the
Masses," on page 62].
Much further from store shelves is the second class-quantum spintronics-which
involves the manipulation of individual electrons to exploit the quantum
properties of spin. Quantum spintronics could provide a practical way to
carry out quantum information processing, which replaces the definite 0s
and 1s of ordinary computing with quantum bits, or
qubits,
capable of being 0 and 1 simultaneously, a condition called a quantum
superposition [see "Rules for a Complex Quantum World," by Michael A.
Nielsen; SCIENTIFIC AMERICAN, November 2002].
Quantum computers, if they can be built, would exploit superpositions of
qubits to perform a kind of parallel processing that would be extremely effective
for certain tasks, such as searching databases and
factoring
large numbers. Efficient number factoring looms large because it
would render obsolete cryptographic codes that are widely used, including
for secure communication over the Internet. Anyone with a large enough
working quantum computer (say, an intelligence or law enforcement agency
or a corporation would he able to decode countless formerly secret messages
at will.
| SPIN AND ITS USES |
WHAT IS SPIN?
TWO KINDS OF SPINTRONICS
|
Perhaps the greatest impact of a future quantum computer will lie in its
unique capability to simulate, or model, other quantum systems, a task that
current computers are hopelessly bad at. For example, quantum simulations
will be required to understand the behavior of matter at the nanometer scale
and could therefore bring huge advances in physics, chemistry, materials
science and biology.
This exciting prospect has led to a worldwide race to find the most suitable
system for storing and processing quantum information. The most advanced
quantum information - processing units to date are arguably spins of ions
trapped in electromagnetic fields. But these systems have the disadvantages
of requiring an ultrahigh vacuum and complex trapping architectures to hold
the individual particles in place and isolated from disturbances. Developing
chips with large numbers of such traps on them is a major challenge. In contrast,
solid-state qubits, which reside directly in a solid substrate, could allow
developers to build on decades of experience fabricating semiconductor chips.
Yet many questions have loomed large for researchers hoping to implement
solid-state quantum computing: Can spins in solids be individually addressed
and controlled? Can scientists come up with suitable interactions to implement
quantum logic gates reliably? Can spins in solids maintain quantum information
long enough to perform a useful number of operations on that information?
In the past few years, all these questions have been answered positively.
Surprisingly, one of the most promising host materials for spins turned out
to be diamond.
ENABLED BY SPINTRONICS
|
Glitter of Diamonds
The diamond we use in our experiments looks very different from the sparkling
gemstones found in jewelry. Recent advances in materials science make it
possible to synthesize thin films of diamond-typically a few hundred nanometers
thick over areas as large as many square centimeters-by chemical vapor
deposition. In this process, a gas made up of carbon-containing molecules
(often methane) and hydrogen is broken down into individual atoms (for example,
by high-power microwave radiation), allowing the carbon atoms to deposit
on a silicon substrate. The diamond that forms can be extremely pure but
often consists of many small crystals, or grains, with grain sizes ranging
from nanometers to microns depending on the conditions in the chamber. The
best device performance usually comes from using single-crystal diamond,
in which diamond's characteristic tetrahedral lattice of carbon atoms is
uninterrupted by the disorderly grain boundaries, which degrade the quality
of the material for both optics and electronics. The ability to engineer
diamond into many forms will likely have a profound effect on electronics,
both conventional and quantum.
| A MAGICAL IMPURITY |
 As
with semiconductors in conventional electronics, the key to making diamond
functional for quantum spintronics is doping it with an impurity, in this
case a so-called nitrogen-vacancy (N-V) center. At an N-V center, two adjacent
sites in diamond's tetrahedral lattice of carbon atoms are altered. One has
a nitrogen atom instead of a carbon, and the other has an empty space. Electrons
orbit in the vacancy and around the adjacent four atoms and carry a spin
that quantum applications can exploit. For example, a laser can repeatedly
excite an electron at the N-V center, which each time emits a single photon
in a specific quantum
 state
when it decays back to its unexcited state. Researchers have used diamond
in this way to demonstrate quantum cryptography prototypes, which rely on
a steady supply of single photons. N-V centers in diamond show up as bright
spots (red) when pumped by a laser. Centers whose spin is in state 1 are
much brighter than centers whose spin is in state 0. Radio-frequency waves
tuned to a precise frequency change the N-V centers back and forth between
0 and 1, passing through transitional states that are quantum superpositions
of the two. Inserting a second nitrogen atom near the N-V center provides
a system of two coupled qubits that enables logic processing.
 The
frequency required to flip the N-V center's qubit is now slightly lower or
higher, depending on the state of the second nitrogen. Applying waves at
the higher frequency can therefore flip the N-V qubit only if the other qubit
is 1. That operation is known as a controlled NOT logic gate, which enables
arbitrary quantum computations. |
A key property of diamond for quantum electronics is the large amount of energy needed to dislodge an electron so that it can flow through the material. Physicists visualize the states that electrons can have in a solid as bands of different energy forming a ladder of unevenly spaced rungs. For semiconductors, the two important bands are the valence band, which is the highest band containing electrons, and the empty conduction band ,just above it, in which electrons can flow freely. The size of the energy gap, or band gap, between these two bands in diamond is 5.5 electron volts, about twice as much energy as present in a visible-light photon and five times as large as the band gap in silicon. Generally electrons in a semiconductor cannot have an energy that lies in the gap, but impurity atoms added to the material can introduce discrete states in the gap, like additional thin rungs to the ladder. Diamond's gap is big enough that two of these states can differ by an energy as large as that of a visible-light photon. Thus, optical-wavelength light can excite an electron at an impurity atom from one discrete state to another without knocking it all the way to the conduction band. When the electron falls back into its lower energy state, it emits a photon with a frequency corresponding to the energy-level difference- the process commonly known as fluorescence. Under continuous illumination, the optical excitation and relaxation process repeats over and over, and an impurity can emit millions of photons per second. In 1997 a group led by Jörg Wrachtrup, who was then at the University of Technology in Chemnitz, Germany, detected individual impurities in diamond fluorescing in this way; igniting a wave of research in the optical detection of single impurities. The particular impurity that Wrachtrup's group detected in those first experiments consisted of a nitrogen atom in place of one carbon atom and an adjacent void where another carbon usually would be, which is known as a nitrogen-vacancy (N-V) center. The N-V center in diamond has a number of remarkable properties that make it the favorite subject of research for many groups around the world. Interestingly, the void plays a crucial role: the N-V center is quite different from a single nitrogen atom without a neighboring vacancy. The electrons in the N-V center move in orbits that span the vacancy and its three neighboring carbons and spend only a little time near the nitrogen. Because of these molecular-like orbits, it is convenient to think of the N-V center as being a single impurity rather than a somewhat odd composite of a nitrogen atom and a vacancy. Single impurities, such as an N-V center, emit one photon at a time-a vital property for the burgeoning field of quantum cryptography [see Best-Kept Secrets," by Gary Stix; SCIENTIFIC AMERICAN, January 2005]
.
| SPIN CONTROL FOR THE MASSES |
By Yuichiro K. Kato
Spintronic devices exploit spin, a property of electrons that
makes them like tiny bar magnets. There are two classes of such devices -those
that manipulate the spins of single electrons [see main text] and those that
control large groups of spin-polarized electrons flowing en masse in
semiconductors (spin currents). Along with working toward single-electron
devices, researchers are making exciting discoveries in controlling spin
currents. I was fortunate enough to play a role in these advances while I
was a graduate student working in David D. Awschalom's group at the University
of California, Santa Barbara, from 2000 to 2005. In particular, we found
new ways to generate and manipulate spin polarization. We also observed for
the first time a phenomenon called the spin Hall effect, which may provide
a way to sort and route electrons based on the direction of their spins.
Because spins behave like tiny magnets, people control them by applying magnetic
fields. Producing magnetic fields usually requires magnetic materials or
external magnets, however. Instead, using electrical fields might enable
smaller, faster spintronic devices that are simpler to fabricate because
electric fields are easier to confine to small regions and easier to produce
with high frequencies (which enable faster operations). Unfortunately, spins,
like all magnets, do not respond to electric fields under normal circumstances.
A relativistic effect comes to the rescue: electrons that move perpendicular
to an electric field "see" a weak magnetic field mixed in with the electric
one. The magnetic field influences the electron's spin. This interaction
is called spin-orbit coupling because physicists first studied it in relation
to electrons "orbiting" in the electric field of atomic nuclei. The Santa
Barbara group initially studied this effect in gallium arsenide, a semiconductor
commonly used in electronics. We saw that when we moved packets of spin-polarized
electrons through this material, the spins rotated as if they were in a magnetic
field. The phantom magnetic field could also align the spins of unpolarized
electrons. Spin-orbit coupling also gives rise to the spin Hall effect, which
was predicted in 1971 by Michel D'yakonov and Vladimir Perel of the Ioffe
Institute in Leningrad. It is named by analogy with the Hall effect (discovered
in 1879 by Edwin Hall), in which opposite charges build up on each side of
a material that carries a current in a magnetic field (top right). In the
spin Hall effect, a small spin polarization accumulates on the edges of a
material carrying an electric current (bottom right), but without requiring
a magnetic field. This effect would be another nonmagnetic way to generate
spin polarization and to direct electrons according to their spin orientation.
In late 2004 Robert C. Myers (another graduate student), Arthur C. Gossard,
Awschalom and I reported seeing the expected spin polarization at the edges
of a slab of gallium arsenide chilled to 30 kelvins. A few months later a
group led by Jörg Wunderlich at Hitachi Laboratory in Cambridge, England,
published observations of the spin Hall effect involving holes (absences
of electrons). About a year ago the Awschalom group went on to demonstrate
the spin Hall effect at room temperature in the semiconductor zinc selenide.
Taken together, these discoveries offer exciting possibilities for developing
spin-based semiconductor technology. |
| Yuichiro K. Kato is associate professor in the Institute of Engineering Innovation at the University of Tokyo. |
Quantum cryptography systems transmit information in the form of single photons carrying one qubit apiece. The laws of physics guarantee that an eavesdropper cannot intercept the photons without disturbing the qubits in ways that the intended recipient can detect. In 2002 Philippe Grangier and his co-workers at the Institute of Optics in Orsay, France, demonstrated the first quantum cryptography prototype system based on a pulsed source of single photons. This breakthrough relied on having an extremely stable and reliable single- photon source-an N-V center in diamond. The N-V center electrons also carry a spin state, one which can be polarized conveniently with optical-wavelength light. And whereas other spin systems in the solid state typically must be cooled to very low temperatures to be polarized, the N-V center spin naturally goes into a specific spin state under optical illumination even at room temperature. Furthermore, experimenters soon discovered that one of the spin states fluoresces much more brightly than the others. Thus, fluorescence intensity can be used for spin-state readout-bright for state "1", dim for state "0."
 |
Diamonds Are Forever
During the past few years, our group at the University of California, Santa
Barbara, has developed a single-photon imaging technique to observe such
individual spins and their orientation in the diamond lattice and to manipulate
them. We have thereby studied how single spins interact with their environment-in
this case the diamond that surrounds them-a topic of fundamental importance
in developing quantum applications. The interactions of the N-V centers with
nearby atoms have allowed us to observe so-called dark spins in diamond-impurity
nitrogens without an associated vacancy that are invisible to optical detection
on their own. Crucially, as observed in these measurements, spins in diamond
are extremely stable against environmental disturbances. Indeed, one of the
most exciting aspects of the N-V center is that it exhibits quantum behavior
even at room temperature.
DIAMOND'S MANY FACETS
|
| DIAMOND MICROPROCESSOR |
| In the future, people wishing to carry out certain specialized tasks may use quantum computers based on diamond spintronics. |
 The
quantum chip that drives the computer's unique abilities contains millions
of optical cavities, each one consisting ofan array of holes etched into
the diamond. These cavities enhance the interaction between spins implanted
at the center of the cavity (purple dot) and photons that carry quantum
information to elsewhere on the chip. Voltages on electrodes control this
interaction. Gigahertz radio waves sent along "striplines" manipulate individual
spin states (qubits). |
 A
variety of spins in each cavity perform different functions: N-V centers
and nitrogen spins process data, the N-V centers interact with photons, and
carbon 13 spins store data for as long as seconds. |
Quantum phenomena tend to be washed out by thermal excitations, and many solid-state quantum effects require extremely cold temperatures, making them hard to study and even harder to turn into practical technology. In this regard, spins in solid materials typically suffer from two problems. The first is an interaction called spin-orbit coupling, which involves the electron's spin and its orbital motion. The second is magnetic interactions with other spins, such as the spins of the nuclei that make up the lattice. In diamond, both these effects are very weak. For example, the nuclei of carbon 12, which makes up 99 percent of natural carbon, have zero spin and thus no effect on the spin of an N-V center. Because it is so immune to outside disturbances of this kind, the quantum state of the N-V center spin can be used to encode quantum information even at room temperature. Of course, "immune" is a relative term. The quantum information stored in an N-V center spin state is lost after about one millisecond in high-purity diamond at room temperature. This loss is equivalent to a bit being flipped in a regular computer. As with such errors in ordinary computers, mistakes in qubits can he corrected provided the error rate is low enough. A rule of thumb for quantum error correction is that at most one in 10,000 operations may fail; any more than that and the procedure becomes a losing battle, with the extra data and operations needed to perform the correction themselves introducing too many new errors. How does the N-V center in diamond measure up against the one-in-10,000 criterion? Radio-frequency radiation guided to the N-V center through on-chip waveguides can deliberately change the N-V center spin within 10 nanoseconds. About 100,000 such operations can occur in the millisecond-long lifetime of the spin's quantum state, and thus the error rare will be very roughly one failure in 100,000 operations. This rate is well below the threshold and is better than any other system of solid-state qubits to date. Quantum cryptography requires only a sequence of individual qubits, but for quantum computation the qubits must interact to produce new qubits, a process that is analogous to how logic gates process pairs of input bits to produce an output in ordinary computers. For example, an AND gate produces an output of 1 if both inputs are 1 and produces 0 otherwise. Quantum logic gates must do similar operations and must also accept quantum superpositions of bits as inputs and produce superpositions as outputs. The next step toward quantum information processing with impurity spins is controlling the coupling between two spins to perform quantum logic.
| MORE TO EXPLORE A Hall of Spin. Vanessa Sih, Yuichiro Kato and David D. Awschalom in Physics World, Vol.18, pages 33-37:2005. Two Groups Observe the Spin Hall Effect in Semiconductors. Charles Day in Physics Today, Vol.58, No.2, pages 17-19; February 2005. Challenges for Semiconductor Spintronics. David D. Awschalom and Michael E. Flatté in Nature Physics, Vol.3, pages 153-159; 2007. Spins in Few-Electron Quantum Dots. R. Hanson, L. R Kouwenhoven, J. R. Petta, S. Tarucha and L.M.K. Vandersypen in Reviews of Modern Physics (in press). |
Our group and Wrachtrup's have studied an interaction that could carry out
quantum logic by using two spins that are near each other in the diamond
lattice. Specifically; we have measured how the spin on an N-V center interacts
with another spin on a nearby nitrogen impurity (with no vacancy). The
interaction is largely magnetic dipole coupling, essentially the same as
the force that makes two macroscopic bar magnets align with north poles facing
south poles. The interaction works as follows. The 0 and 1 states of an N-V
center have somewhat different energies. The energy difference, or splitting,
between the 0 and the 1 is much smaller than the energy of an optical photon,
and instead giga-hertz radio waves will drive the spins back and forth between
0 and 1 and superpositions thereof. When the N-V center is close to another
nitrogen atom, the splitting of its 0 and 1 states depends on the other
nitrogen's spin state. This dependence makes possible a controlled NOT (CNOT)
gate. in which one qubit is inverted if and only if the other qubit is a
1. The gate would work by using radio waves tuned to the frequency that will
flip the N-V center provided the nitrogen spin is a 1. If the nitrogen spin
is a 0, the N-V center's energy splitting will be different and the radio
waves will not affect it. The CNOT gate is quite special: we can compose
any arbitrary quantum operation on any number of qubits by combining CNOT
gates acting on pairs of qubits and rotations of individual quhits (which
can also be carried out by applying radio waves to spins; individual spins
could be addressed by bringing the radiation to them along special circuits
called striplines).
Demonstrations of a CNOT gate and qubit rotation are therefore major research
goals. Longer distance interactions between N-V spins in diamond may be possible
by using photons as mediators. On-chip optical devices such as waveguides
made of the same diamond substrate could route the photons. Integrating the
N-V centers in structures called optical cavities, in which light forms standing
waves, would enhance the strength of the interaction between the spins and
the photons. At Santa Barbara, in a collaboration with Evelyn Hu and her
students, we recently demonstrated proof-of-concept photonic crystal cavities.
Each "optical cavity" consists of a region of diamond with a honeycomb of
holes etched into it. The holes work to confine and amplify light at the
center of the structure [see "Photonic Crystals: Semiconductors of Light,"
by Eli Yablonovitch; SCIENTIFIC AMERICAN, December 2001]. Thus far, however,
this work is very preliminary: the N-V centers, which are randomly distributed
in the diamond instead of being precisely positioned in the cavities, are
bystanders in our studies.
Placing Impurities
Many of the experiments on N-V centers to date have been carried out using
synthetic diamonds like those used for our optical cavities: the N-V centers
were formed naturally but in random locations during the diamond growth process.
Now researchers at the Australian National University, the University of
Bochum in Germany and Lawrence Berkeley National Laboratory are making great
progress in placing individual impurities at specific locations. They use
advanced ion implantation techniques to insert single ions of nitrogen with
submicron accuracy. Then they heat the diamond to 550 degrees Celsius, which
causes the vacancies in the diamond to migrate. When a vacancy meets a nitrogen
atom it stays next to it, forming an N-V center. N-V centers seem a promising
technology for processing quantum information, but what about storage for
times longer than the millisecond-long decay time of their electronic spin
states? Researchers in Mikhail Lukin's group at Harvard University have explored
an approach that makes use of the spins of carbon nuclei. Because the nucleus
of the most common isotope of carbon, carbon 12, has zero spin in total,
the group used carbon 13 atoms, whose nuclei have the spin of their one extra
neutron. The scientists transferred the information encoded in a single N-V
center spin to a single nuclear spin of carbon 13 and retrieved it 20
milliseconds later. The nuclear spin showed no sign of decay, indicating
that the quantum state could survive for seconds. Thus, nuclear spins appear
to be a propitious route to qubit storage. The Harvard researchers have also
proposed a design for constructing a quantum repeater based on this work.
Quantum repeaters are a basic element needed for quantum communication
(transmitting qubits over longer distances). Iris an exciting time for quantum
information research, with many different computation architectures vying
for supremacy. Considering the rapid rise and successes of diamond-based
spin research over the past few years and with companies such as Hewlett-Packard
getting into the game, the prospect of room-temperature quantum information
processors is sounding less like science fiction. The diamond age of quantum
electronics could be just around the corner.
David D Awschalom Ryan Epstein and Ronald Hanson are associated with the Center for Spintronics and Quantum Computation at the University of California, Santa Barbara. Awschalom is director of the center and professor of physics and of electrical and computer engineering at Santa Barbara. His research group is primarily concerned with investigating electron-spin dynamics in a variety of semiconductor systems. Epstein obtained his Ph.D. in Awschalom's group, studying nitrogen-vacancy centers in diamond. He is now doing postdoctoral research on trapped ions at the National Institute of Standards and Technology in Boulder, Cob. Hanson was a postdoctoral researcher in the group and has just become assistant professor of physics at the Kavli Nanoscience Institute in Delft, the Netherlands. For his Ph.D., he studied single- electron spins in gallium arsenide quantum dots at the Delft University of Technology.
quantity
of angular momentum called spin, almost as if they were tiny spinning balls.
Scientists represent spin with a vector. For a sphere spinning "west to east,"
the vector points "north," or "up." It points "down" for the opposite spin.
Magnetic tunnel junction
