New technique advances carbon-fiber composites.
Researchers at MIT and Delft University of Technology have come up with a way to use standard integrated circuit technology to make a very basic component of a quantum computer.
While other efforts to create a quantum computer have focused on manipulating systems found in nature, this MIT approach, published in the August 13 issue of Science, is one of the first in which the device can be fabricated.
The discovery, led by Associate Professor of Mechanical Engineering Seth Lloyd, Visiting Professor Johan Mooij and Professor of Electrical Engineering Terry P. Orlando may make it possible to place large numbers of superconducting qubits (bits of information in the quantum realm) on a single chip, thereby moving a quantum computer one step closer to reality.
Because a quantum computer would be capable of power and efficiency far beyond that of any conceivable conventional computer, researchers are attacking from all angles the problems associated with building one.
Atoms trapped in small electromagnetic cavities, nuclear-magnetic resonance on large molecules and photons have all been put forth as possible bases for a quantum computer. Professors Lloyd, Orlando and their colleagues -- Professor of Physics Leonid Levitov, physics graduate student Lin Tian, visiting scientist Juan Mazo, Professor Mooij and Caspar van der Wal at Delft -- propose using superconducting circuits to store and manipulate information.
"Compared with other proposals for quantum computation, this one, if it can be realized, allows much greater flexibility in circuit design," Professor Lloyd said. "Rather than being stuck with what nature deals you, this kind of qubit allows quantum circuits to be designed in the same way you would design integrated circuits. With it, you have the ability to create your own 'designer' quantum computation system."
A superconductor is a metal that, when cooled to very low temperatures, shows no resistance to the flow of electrical current. Electrical current can flow forever in a loop made of a superconducting wire.
Professor Orlando said that one example of such a "persistent current" is in the electromagnets used in magnetic resonance imaging (MRI). MRI machines are made of large coils of superconducting wire.
By making the coil or loop very small -- only the diameter of a human hair -- the researchers have proposed that these circulating currents will obey the laws of quantum mechanics.
CONTROLLING THE SPIN
Researchers in what Professor Lloyd calls the emerging field of "quantum mechanical engineering" are facing the same problems engineers have faced for a long time, but they are forced to rethink them for the tiny quantum scale.
The physics theory called quantum mechanics is based on the observed features of subatomic particles. One of these features is spin, or angular momentum.
Similar to a compass, spins align in a magnetic field to be either up or down. This is the equivalent of the conventional computer's "on" and "off" states, represented by ones and zeroes. Currents flowing in one direction and the opposite direction have often been proposed as the zero and one states, called bits, that are needed for computation and communications in a quantum computer. Orlando's team is the first to suggest the use of these persistent currents as qubit states.
A single bit of information in the quantum realm is known as a qubit. A qubit can represent one, zero or the two states at once.
The goal is to cause a controlled interaction among two or more qubits, producing a coherent change in the state of one qubit that is contingent on the state of another.
By using superconducting wire to create persistent currents, Professor Orlando says the currents in the loop can now be in states that are not only zero and one, but more important, in a superposition of these states (both at the same time). The quantum computer's ability to represent many states at once gives it its enormous power to perform many computations simultaneously.
It is also this ability that makes it so fiendishly difficult to make qubits cooperate: atoms "talk" to each other and to materials in their environment, thereby losing their coherence or ability to concentrate on the task at hand. The information on the qubits has to be entangled, but in a controlled way.
One must also do the quantum measurements required to read out the result of a quantum computation.
Quantum computers based on ion traps or controlling the spin of large numbers of identical molecules are good at staying coherent. On the other hand, these systems seem difficult or impossible to scale up to a useful size. For a quantum computer to be of practical value, it must have at least 80 to 100 qubits.
Meanwhile, solid-state circuits lend themselves to being scaled up in size, but their coherence is relatively low. The MIT researchers' qubit stays coherent for much longer than similar work along these lines.
"What is important about the proposed superconducting qubit is that it uses the technology of integrated circuits, so that it has the potential to have large numbers of superconducting qubits on a single chip," Professor Orlando said. "The transfer of information among qubits is done by magnetic coupling. In particular, we will use a larger superconducting loop to couple the magnetic field produced by one circulating current loop to the other loop."
The qubit of Professors Lloyd, Orlando and their colleagues has persistent electrical currents of opposite signs (one positive, one negative) as its basic states. The qubits can be driven individually by magnetic microwave pulses. Measurements can be made with superconducting magnometers.
Although Professor Orlando said a quantum computer based on this technology is still years away, the major advantages of this approach are that the devices can be fabricated into desired configurations rather than relying on materials available in nature, and fabrication can be done by existing superconducting technology.
"This approach provides a bridge between the weird new world of quantum mechanics and the world of engineering systems that we are familiar with," Professor Lloyd said. "We're still playing with the same cards, but we get to deal our own hand."
A version of this article appeared in MIT Tech Talk on September 25, 1999.