The Year in Science: Technology 1997

It had been several decades since physicists Neil Gershenfeld of MIT and Isaac Chuang at Los Alamos National Laboratory last felt such a powerful sense of accomplishment simply by adding together two numbers. Of course, they didn’t do it the easy way, with pencil and paper, but by inventing a quantum computer made out of complex liquids. Last March they demonstrated such a device capable of adding one plus one, and by year’s end, they were able to search a list of seven telephone numbers. That’s a far cry from the Pentium, but it’s at least a first step towards actually building quantum computers, which in theory could far surpass conventional computers in certain applications. Whereas present-day computers represent data as a series of ones or zeros, which correspond to electrical charges held in electronic circuits, quantum computers would use particles such as atoms instead of circuits, and they would use the atoms’ quantum states, such as spin or polarization, to represent data. The main challenge in building such a device is to be able to read the data and manipulate it without disturbing the delicate quantum states of the atoms. To get around this problem, Gershenfeld and Chuang took atoms embedded in large complicated molecules and used the direction of their spins to represent data. Not only is spin a much hardier quantum state than ones that other physicists have used, but it also allowed them to use a proven technology—nuclear magnetic resonance (NMR) machines—to read and manipulate data. NMR machine use magnetic fields and radio waves to probe the spins of atoms, and to change their orientation. The two physicists started with trillions and trillions of atoms in a thimble-full of the organic chemical alanine. Most of the atoms in the thimble are pointing every which way, but, for the same reason that a dozen tosses of a die will tend to turn up one number more than the others, a tiny plurality of the atoms in a drop will usually be found pointing in one specific direction. These atoms stand out to the NMR machine because the other, randomly pointing ones cancel each other out. The plurality of atoms constitutes one quantum bit, or qubit. The alanine molecule, it turns out, yields three qubits because it has three atoms of carbon that each respond to a different NMR frequency, and yet that are still linked in such a way that they can be used for addition (the two digits have to be linked for the carry operation). To add one plus one, the physicists used two of the carbon atoms and zapped them with a series of radio pulses. Each pulse, depending on its duration and the orientation of its partner bit, makes the atoms shift the orientation of their spins — say, one hundreth of a second for a quarter turn and two hundredths of a second for a half turn. The series of pulses, then, constituted the quantum computer’s program. When the program was finished, all that remained was to read the final spin state of the collective atoms to get the answer. Since each qubit was actually the average of trillions of atoms, it didn’t matter if a few atoms got disturbed by a wayward molecule here and there. The great potential of quantum computers lies in their theoretical ability to perform lots of calculations at once, which stems from the quantum mechanical nature of the atoms to exist in an infinite number of spin states at the same time. [Rob: that’s my best shot — Fred] Before they can tap this potential, however, Gershenfeld and Chuang need to find molecules that will give them more than two qubits. With ten qubits, Chuang says, they should be able to do some simple factoring—figuring out, for instance, that 3 times 5 is 15. To do something really useful—-to factor numbers large enough to crack encryption codes, for instance—one would need thousands if not hundreds of thousands of quantum bits, says Chuang. That’s a ways off. —Jeffrey Winters