Transistors, the tiny switches that flip on and off inside computer chips, have long been the domain of electricity. But scientists are beginning to develop chip components that run on light. Last week, in a remarkable achievement, a team led by researchers at the Massachusetts Institute of Technology (MIT) in Cambridge reported building a transistor that is switched by a single photon.
Conventionally, photons are used only to deliver information, racing along fiber-optic cables with unparalleled speed. The first commercial silicon chip to include optical elements, announced last December, did little to challenge the status quo. The on-board beams of light in the device, developed at IBM’s research center in Yorktown Heights, New York, merely shuttle data between computer chips.
Now, Wenlan Chen of MIT and her colleagues have taught light some new tricks, using a cloud of chilled caesium atoms suspended between two mirrors. Their transistor is set to ‘on’ by default, allowing a beam of light to sail through the transparent caesium cloud unmolested. But sending in a single ‘gate’ photon turns the switch off, thanks to an effect called electromagnetically induced transparency. The injected photon excites the caesium atoms, rendering them reflective to light trying to cross the cloud (see ‘Turn off the light’). One photon can thus block the passage of about 400 other photons, says Chen, who presented the result on 7 June at a meeting of the American Physical Society’s Division of Atomic, Molecular and Optical Physics in Quebec City, Canada.
The ability to turn a strong signal on and off using a weak one fulfills a key requirement of an optical transistor. “Nothing even came close before,” says physicist Ataç İmamoğlu of the Swiss Federal Institute of Technology Zurich, who called the experiment “a true breakthrough”. In theory, the hundreds of photons, controlled by the triggering photon, could fan out and switch off hundreds of other transistors in an optical circuit.
With its exotic clouds of atoms and bulky equipment, the proof-of-principle transistor is unlikely to become a component in everyday computers. But it could be a useful tool for studying how photons interact at the quantum level — potentially leading to a quantum transistor that flips, not a one or a zero as in classical computing, but a fuzzy bit of quantum information.
A more practical optical transistor debuted in April 2012 at Purdue University in West Lafayette, Indiana, where electrical engineer Minghao Qi has made one that is compatible with the semiconductor industry’s existing manufacturing techniques. “The advantage of our device is that we have it on a silicon chip,” says Qi.
In this case, the beam of light to be switched on and off enters and exits along a channel, etched in the silicon, that sits next to a parallel channel. In between the two rails is an etched ring. When a weaker light beam courses through the second optical line, the ring heats up and swells, interfering with the main beam and switching off the transistor. This switch can flip on and off up to 10 billion times per second.
And the output beam can fan out and drive two other transistors, meeting one of the established requirements for an optical transistor set out in 2010 by David Miller, a physicist at Stanford University in California. Other criteria include matching the frequency of the exiting signal to the input frequency and keeping the output clean, with no degradation that could cause errors. “Making an optical transistor that really satisfies the necessary criteria is very hard,” says Miller.