Computer chip technology is undergoing a revolution right now. New materials and new methods are changing the building block of our technological society.
We posted an article on Moore’s Law a couple of weeks ago. Today, we wanted to revisit this topic. We’re going to talk about some of the exciting discoveries being announced that herald the future of computer technology.
What Will Computer Chips Be Made Out Of?
We’re finally coming to the end of what it’s possible to do with silicon.
Intel has said that they’ll be moving away from silicon after two more generations of computer chips. Currently, Intel is producing chips with 14 nm transistors on them. They’re working on 10 nm transistors. After that will be 7 nm transistors. And after that? Well.
People are trying to find a material that might be faster and more energy efficient than silicon. We’re going to talk about two materials that are being researched for potential use with integrated circuits: gallium nitride and graphene.
Just saying, Silicon Valley might need a new name.
Gallium nitride (GaN) has been causing a lot of buzz lately. Why? According to a recent article by Rachel Courtland in IEEE Spectrum, GaN is “capable of faster switching speeds and can handle a higher voltage than a same-sized silicon device.” This sounds good.
Or as Dean Takahashi put it in an article in Venture Beat about Alex Lidow, head of Efficient Power Conversion and the most visible advocate for GaN around, gallium nitride “promises to unseat silicon because it has higher performance, less power consumption, and lower cost.” What?
With such eye-popping promises, it sounds like a fantasy material. But gallium nitride chips are already used by prominent companies. For instance, Google uses GaN in their self-driving car technology.
Even if it doesn’t reach the 1000x improvement over silicon that Lidow says is possible, GaN is a material to keep your eye out for.
We wouldn’t be in the least surprised to see gallium nitride popping up in the news a lot in the coming years.
Graphene is a one atom thick sheet of carbon atoms arranged in a hexagonal array. Which probably sounds pretty boring unless you’re really into material science.
But it’s not boring. Not at all.
Paul Rincon put it clearly in an article for the BBC in 2014: “The material is the subject of global research efforts aimed at harnessing the extraordinary electrical, optical, mechanical and thermal properties that potentially make it a cheaper and more energy efficient choice than silicon in electronics.”
While GaN is already in use (and is set to have a major impact on the market within the foreseeable future), graphene is still a long ways away. As you can imagine with such a material, it’s very difficult to make it pure enough for reliable use.
And while graphene has a whole host of incredible properties, just how these properties might work in an integrated circuit is not really known.
Sidenote—I can’t write about graphene without mentioning one of my favorite stories over the past week: graphene makes spiders spin super-silk. Science is awesome.
How Will Computer Chips Process Information?
New materials are definitely part of the future of computer chip technology. But they aren’t the only aspect that’s being revolutionized.
Researchers and engineers are experimenting with new ways of processing information. We’re going to talk about two of these ways: spintronics and quantum computing.
Spintronics is short for “spin-based electronics.” It’s a radically new way of processing information.
You can’t see electrons spinning because they’re so small. However, you can influence and measure spin with magnets.
This observation is the basis of spintronics. If you use electrons spinning in different directions to represent 1s and 0s instead of using electrical charges, then you could save on energy costs versus conventional electronics. A lot.
But spintronics, for now, is only feasible for the innermost processes. Stephen Shankland points out in CNET that “a computer would use spintronics within its deepest interior but rely on traditional electronics further out to communicate with memory, drives and networks. Translating data and instructions between the two zones takes time.”
So the advantages of spintronics could be offset by necessary translation.
The new method that’s received the most attention has been quantum computing. In a nutshell, the process uses “qubits” (quantum bits) that can be both 1 and 0 simultaneously.
You can see how this would help with problem solving. It would allow the computer to sift through possible answers in parallel—a significant advantage.
The technology also holds out the possibility for unbreakable quantum encryption.
However, there are two big problems with quantum computing right now.
First, the computers need be extremely cold to keep the qubits still enough to be usable. Too much energy and the system falls apart.
Second, error detection is extremely difficult. Jeremy Hsu explains in IEEE Spectrum: “Classical computers can detect and correct their bit-flip errors by simply copying the same bit many times and taking the correct value from the majority of error-free bits. By comparison, the fragility of quantum states in qubits means that trying to directly copy them can have the counterproductive effect of changing the quantum state.”
IBM has made some progress with error detection, but quantum computing is still a long ways away from being a practical technology.
Still, if Moore’s Law taught us anything, it’s that change can happen rapidly when there’s incentive.
We’re used to ever-increasing processing power. The market is demanding that companies solve the problems of the limits of silicon. How will the companies respond?
We now need to ask:
What will a computer chip be?