Juha Saarinen: Honey, I can't shrink the chips any further

Just how small can chips get? Photo / Getty Images

OPINION:

Taiwanese and Korean silicon chip foundries are gearing up to produce silicon semiconductors made with extremely fine processes, just a few nanometres in size, an amazing technological feat that's by and large going unnoticed apart from hardware geeks.

As the digitisation of everything continues, we'll need more powerful yet less power hungry chips, and billions have been invested in shrinking the transistors inside them to achieve this.

We are getting close to how small you can go with existing materials and technology, and the race is on to develop the next generation of chips, and New Zealand researchers are involved in that.

What will replace the current - and excellent - silicon-based technology is up in the air at the moment, even though as the co-director at The MacDiarmid Institute for Advanced Materials and Nanotechnology, physics professor Nicola Gaston says, we're getting close to the limits of how much smaller transistors can be made.

"Regardless of what materials or their atoms you make your transistors of, when you get down to 2nm you are talking about devices that are just a few - less than ten - atoms across in size.

"This fundamental limit – which has been known for a long time as Moore's law – is not going anywhere; it has been with us for decades, and technologies that have gotten us down as far as 2nm are significant, but fundamentally, we are going to need different device structures to get us past this size-related barrier," Gaston said.

That could be different materials, such as graphene, which Gaston says is "definitely a wonder material".

Single-atom two-dimensional layers of graphene, exfoliated from graphite, show amazing properties and promising future applications under controlled experimental conditions, and yes, that includes very fast-switching transistors.

Although graphene has been mooted as a future silicon replacement, Gaston notes that it's difficult to incorporate the material into devices. It's not the only material being researched currently, with many others such as molybdenite and indium selenide thought to take us past the upcoming silicon chip barrier.

When that will happen isn't clear, however, Gaston believes that forms of computing that rely on very different mechanisms than what we have currently will be more significant than, say, graphene-based transistors.

In other words, we need to rethink computing devices quite drastically.
"Quantum computation, where you can think of a simulation that encompasses multiple possibilities at once – which is super relevant to materials science simulations, but also to other probabilistic applications – is one example that is already demonstrating its capabilities, and is currently in the process of being engineered to be more practical," Gaston said.

Another is neuromorphic computing with systems that process information in similar ways to the brain. This gets us very low energy information processing abilities that are much better than traditional computers at some tasks, such as image recognition.

How about no energy wasted or lost as heat? Gaston points to solutions such as superconducting components without resistance, and spin-based electronic communication for devices that are close to making an impact.

The year is 2022, and there's a geopolitical aspect to developing the electronics of the future. As electronics in everything became the norm, it was a good idea to have semiconductor foundries as near assembly lines and factories as possible.

Now however, keeping the tech that will boost our fortunes in the immediate future safe from authoritarian nations that can't develop it themselves has become a consideration.

Recently, a Chinese government-connected economist proposed that authorities in the giant nation simply seize Taiwan Semiconductor Manufacturing Company in case of sanctions.

Could such threats mean that the tyranny of distance would work to New Zealand's advantage, along with our relative political stability, to attract chip makers?

"Well that is a big question, and subject to a lot of different variables," Gaston replied, as a scientist would.

Apart from the abovementioned move away from a less hegemonic world in the near future towards one in which different types of computing are complementary for solving different problems, Gaston thinks it may be possible for NZ to make computational technology for specific markets and niche applications.

In principle, that is, and there are manufacturing challenges to be overcome if we get to that stage.

Gaston thinks some big shifts are definitely coming, and she's right. Computing will be transformed over the next few years.

This time, NZ shouldn't wait to see what happens like when personal computing and the internet developed.

Instead, we need to literally chip in as much as possible with materials science, research and manufacturing efforts, or we won't keep up with the world.