It doesn’t matter that “computing power is over 1.1 million times faster today than it was 40 years ago.” After decades of exponential growth, processor speeds are stagnating: we’re simply running out of atoms to shrink.
Something fundamental needs to change in the way we use technology to process information in order to progress. So what are the options?
Discovered under surprisingly low-tech circumstances (scientists rubbed a pencil across some Scotch tape), a few years ago, graphene looked like the answer to the silicon chip problem. It’s something of a wonder material: 300 times stronger than steel, harder than a diamond, impermeable to other elements and so transparent that 97% of white light passes through it. And at just one atom thick, electrons race across it much faster than silicon - in 2012 a team produced a graphene transistor that clocked at 427GHz.
However, there’s a problem: unlike silicon, graphene is not a semi-conductor. Its conductivity cannot be turned off. This means that whilst electrons move more quickly across it, the essential on-off switching operations electronic components depend on are impossible to achieve in its native state.
Scientists have recently had some success in synthesizing the “band gaps” necessary to switch between insulating and conducting states, but the subsequent transistors dissipate energy and current too quickly to be effective.
Though phosphorene shares many properties with graphene (even down to the primitive Scotch tape extraction technique) there is one crucial exception: like silicon, it is a natural semi-conductor. In fact, it’s been referred to as “graphene with a band gap”.
It’s less brittle than silicon, making it a viable candidate for flexible electronics, and it also emits light, making it possibly suitable for optical computing with lasers or LEDs. The main factor limiting experimentation with phosphorene is the difficulty in producing it at scale, so we’re probably some way off seeing commercially-available phosphorene-based chips.
More recently, scientists have found evidence for a massless quasiparticle known as a “Weyl fermion” (first theorised by Hermann Weyl in 1929) that could hypothetically carry an electric charge a thousand times faster than existing popular semiconductors, and at least twice as fast as graphene.
Other types of fermion particles (such as electrons) cannot share the same state at the same position at the same time. Being massless, Weyl fermions carry no such restrictions. Though we’re probably years away from seeing Weyl fermions in commercial computers, their discovery is particularly promising for the future of quantum computing, which has so far been plagued by stability challenges.
Though still in its infancy, quantum computing has progressed from theoretical concept to commercial debut in only a few years.
Quantum computers process information in a fundamentally different way to silicon transistors. Where the latter encodes and processes information using binary ones and zeroes (or bits), quantum computers use sub-atomic particles called qubits, whose value can be a one or a zero or a simultaneous superposition of the two at any given moment.
This doesn’t necessarily mean that quantum computers are faster than traditional silicon ones (in fact, they often aren’t) it’s that they’re suited for different tasks. Quantum computers can perform parallel operations at once because qubits aren’t constrained by the two-state bits.
Optical computing is a broad label. Whilst in principle it describes a shift away from electric current and towards light as a means of transmitting and calculating information, in practice it refers to a wide range of distinct (and often competing) technologies. It’s also largely theoretical at this point – controlling light with enough precision at a scale compatible with modern electronics has proved difficult so far.
It doesn’t matter that “computing power is over 1.1 million times faster today than it was 40 years ago.” After decades of exponential growth, processor speeds are stagnating: we’re simply running out of atoms to shrink.
Something fundamental needs to change in the way we use technology to process information in order to progress. So what are the options?
