Electrons are pretty good at processing information but not so good at carrying it over long distances. Photons, on the other hand, do a grand job of shuttling data round the planet but are not so handy when it comes to processing it.
As a result, transistors are electronic and communication cables are optical. And the world is burdened with a significant amount of power hungry infrastructure for converting electronic information into the optical variety and vice versa.
So it’s no surprise that there is significant interest in developing an optical transistor that could make the electronic variety obsolete.
There’s a significant problem, however. While various groups have built optical switches, optical transistors must also have a number of other properties so that they can be connected in a way that can process information.
For example, their output must be capable of acting as the input for another transistor–not easy if the output is a different frequency from the input, for instance. What’s more, the output must be capable of driving the input for at least two other transistors so that logic signals can propagate, a property known as fanout. This requires significant gain. On top of this, each transistor must preserve the quality of the logic signal so that errors do no propagate. And so on.
The trouble is that nobody has succeeded in making optical transistors that can do all and can also be made out of silicon.
Today, Leo Varghese at Purdue University in Indiana and a few pals say they’ve built a device that take a significant step in this direction.
Their optical transistor consists of a microring resonator next to an optical line. In ordinary circumstances the light supply enters the optical line, passes along it and then outputs. But at a specific resonant frequency, the light interacts with the microring resonator, vastly reducing the output. In this state, the output is essentially off even though the supply is on.
The trick these guys have perfected is to use another optical line, called the gate, to heat the microring, thereby changing its size, its resonant frequency and its ability to interact with the output.
That allows the gate to turn the output on and off.
There’s an additional clever twist. The microring’s interaction with the gate is stronger than with the supply-output line. That’s significant because it means a small gate signal can control a much bigger output signal.
Varghese and co say the ratio of the gate signal to the supply is almost 6 dB. That’s enough to power at least two other transistors, which is exactly the fan out property that optical transistors require.
These guys have even built a device out of silicon with a bandwidth capable of data rates of up to 10 GHz.
That’s an impressive result, particularly the silicon compatibility.
Nevertheless, there are significant hurdles ahead before an all-optical computer made with these devices can hope to compete against its electronic cousins.
The biggest problem is power consumption. Much of the power consumption in electronic transistors comes from the need to charge the lines connecting them to the operating voltage.
In theory, optical transistors could be even more efficient–their lines don’t need charging at all. But in practice, lasers burn energy as if it were twenty dollar bills. For that reason, it’s not at all clear that optical transistors can match the efficiency of electronic chips.
And with the computer industry now responsible for almost 2 per cent of global carbon dioxide emissions, almost as much as aviation, power consumption may turn out to be the overarching factor for the future direction of information processing.
Ref: arxiv.org/abs/1204.5515: A Silicon Optical Transistor
These weird virtual creatures evolve their bodies to solve problems
They show how intelligence and body plans are closely linked—and could unlock AI for robots.
A horrifying new AI app swaps women into porn videos with a click
Deepfake researchers have long feared the day this would arrive.
Chinese hackers disguised themselves as Iran to target Israel
But they left a few clues that gave them away.
DeepMind says it will release the structure of every protein known to science
The company has already used its protein-folding AI, AlphaFold, to generate structures for the human proteome, as well as yeast, fruit flies, mice, and more.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.