When Moore's Law finally fails - and that's inevitable, physicists say - what will happen to computers? A technological breakthrough by two Ottawa physicists may provide the answer.
In 1965, Intel Corp. co-founder Gordon Moore observed that the number of transistors that can be packed into a microprocessor doubles every 18 months. Amazingly, Moore's Law, as it became known, has proved true, more or less, ever since.
But there must be a limit to how small transistors can get, and theoretical physicists say that Moore's Law should collapse in about 10 to 20 years. Over the past decade or so, the race has been on to re-imagine the microprocessor from the ground up.
And for the physicists at the Institute for Microstructural Sciences in Ottawa, a division of the National Research Council of Canada, "up" means where an electron is headed.
The breakthrough by Dr. Pawel Hawrylak and Dr. Andy Sachrajda has proved a five-year-old theory correct: That not only can data be stored in binary form using electrons (the on-off, yes-no language of microprocessors), but that it can be stored using the direction in which the electrons spin. If the spin is up, it's positive; if down, negative.
Not surprisingly, scientists have called the technique spintronics.
It doesn't sound too exciting expressed that way, but when it is performed on what Dr. Hawrylak calls "a nanoscale" level, it brings the concept of quantum computing a step closer to reality - and a step closer to multiplying computing power far beyond what can be done with traditional transistors.
The rush to make quantum computing a reality is important, Dr. Sachrajda said, because the effort to make ever-smaller chips in the traditional way will inevitably introduce atomic-level impurities into them. The impurities will make the behaviour of those transistors erratic.
That's one wall that Moore's Law is bound to crash into, Dr. Sachrajda said. "You will start to see elements of quantum computing creeping in, but that will make every transistor behave differently."
A second wall is cost. The so-called Moore's Second Law states that the cost of making silicon chips, as we do now, also grows at an exponential rate, to the extent that it will become economically prohibitive to keep trying to make them smaller based on current silicon technology.
With modesty typical of researchers into theoretical physics, Dr. Sachrajda pleaded that the accomplishment he and Dr. Hawrylak have achieved not be exaggerated, the way it was by Nature magazine, which hailed it as the creation of a new transistor.
"What we've done is much more modest," he said.
In fact, it is only a prototype of what is called a "single-spin transistor."
So far the experiment has worked only in a very controlled environment in the laboratory, and is still too far from practicable to be excited about, he said, speaking from Grenoble, France, where he is currently on sabbatical.
But if scientists can figure out how to solve a few problems, then it could be very exciting.
He did not use the word "revolutionary," though he could have.
What's revolutionary about this idea is that the two basic functions of a computer - logic processing and memory - can be performed together in a single place. Transistor-based processors need to store and retrieve information from a distant device, such as RAM chips, which makes its work agonizingly slow.
In its simplest form, the IMS breakthrough involves the creation of an artificial atom, called a "quantum dot." Dr. Sachrajda describes the dot as a box that has been emptied of its electrons so that electrons can be put back in - from zero to 50 of them - one at a time.
An electronic current is then run through the box. It will be influenced not only by the charge of the atom inside, but also by the direction of the electrons' spin. The current then leaves the box with a polarized spin.
Through this double duty, Dr. Sachrajda said, "you could include memory into the microchip. They would be in the same place. Of course, if you have more handles, or buttons, to use in the chip, then you can do more things."
More technically, the IMS described the process this way: "A quantum dot is an artificial atom which can be used in semiconductor lasers, where it exhibits quantum behaviour such as predictable and controllable energy levels. By connecting the dot to spin-polarized reservoirs, one can insist that the electrons flowing in or out have their spins aligned up or down, allowing a high or low current to flow through the dot. This combination of control at the single charge and single spin may play a role in the future solid state form of quantum computing where the unit of quantum manipulation, the qubit, might consist of specially prepared spin states.
"It's not merely the speed of the device components that is important for quantum computing," Dr. Sachrajda said, "but the ability to allow different bits - now called 'qubits' - to interact into what are called entangled states, and to perform operations on them while they are in that entangled state. It is this ability that most excites physicists and computer visionaries, allowing scenarios in which certain calculations to be performed in a dramatically shorter time."
Asked what it means for computing, Dr. Hawrylak said that "ultimately, people hope that semiconductor spintronics would lead to several new possibilities, such as the combination of memory ... and logic functions on a single semiconductor chip.
"As a trivial benefit," he said, "imagine your computer switching on almost immediately instead of waiting during that annoying long booting-up time. Very new functionalities will become possible, since now we will have two properties to play with, charge and spin.
"It should be pointed out," he added, "that many of the important issues in spintronics - how we get from these simple prototype demonstrations in laboratory devices to real useful devices - are really material science issues, and are being pursued by many groups worldwide. ... What we have done is to find a way to circumvent these issues to create prototype laboratory devices which will teach us how devices will operate once these material problems are solved."
Right now, however, Dr. Sachrajda plays down the development as little more than "a remarkable toy."
And it will remain a toy until physicists can hurdle another major obstacle. At the moment, the IMS breakthrough works only in a very chilly atmosphere - 100 millikelvin, about a sweater shy of absolute zero.
But still, he said, "we can start to play with the kind of effects that can be done when they get spintronics working at room temperature."
What's next? Dr. Hawrylak and Dr. Sachrajda are working on a set of more complex devices with several spin transistors that will make a prototype of a quantum computer.
If they succeed, they will be able to say goodbye to Moore's Law.
Not because technology finally failed to create faster processors, but because Mr. Moore will have been proved to be far too modest.