Shrinking the future: why Nanoimprint Lithography will keep Moore’s Law alive

We’ve all been around the internet long enough to know that when something or someone is declared ‘dead’, it’s worth double-checking. And in the case of Moore’s Law (the nearly sixty-year-old theory governing the speed of computing) this happens around once a year – or every time a new processor rides into town. This might feel frequent enough to have us preparing for the worst, but is it really? Can it be even remotely possible for semiconductor chips to reach their maximum capacity, given how essential they are to… everything?

Generally speaking, Moore’s Law states that the number of transistors on a computer chip doubles approximately every two years. This naturally increases the speed and capability of the devices we use, but – and this is crucial – Gordon Moore (the founder of Intel, who created this theory) believed that the cost would only increase minimally. He predicted that mass production would ultimately bring the price down. In this respect, it's clear that, despite the name, it’s not actually a ‘law’ in the mathematical sense, more the result of an observation of the economics of semiconductors. Even so, Moore’s Law has borne out wave after wave of predictions of its demise.

However, we are living in a world that is already computationally intensive – can you think of a part of day-to-day life that doesn’t contain a semiconductor chip these days? And while the focus must lie creating smaller and more powerful chips, keeping their production at a reasonable cost is key in a compute-driven world. Nanoimprint Lithography (NIL) is likely to be key in achieving this. “It’s what is known as an ‘advanced lithography technique’,” explains Chris Howells, European Operations Director for Canon’s Semiconductor Equipment Division. “And ourversion derives from Canon’s expertise in inkjet technology.”

You might be surprised to learn that Canon has been in the business of supporting semiconductor manufacturers all over the world for close to fifty years, supplying state-of-the-art equipment for semiconductor lithography. This is the process used to print the tiny and precise patterns you see on computer chips, using light or radiation to transfer a pattern onto a wafer coated with a viscous liquid called photoresist. Therefore, NIL feels like the natural next step, combining Canon’s decades of extremely specialist knowledge from the worlds of print and photonics – the science of light. But what does it do? And how is it different?

The process is quite different to traditional semiconductor lithography and was fraught with complexity in the design. Firstly, instead printing a pattern onto a wafer that is completely coated with photoresist, NIL releases droplets of the liquid only where it is required. Using the same technology which can be found in Canon inkjet printers, each drop can be measured, controlled and dispensed with precision.

Then, a specially manufactured stamp called a ‘mask’ presses the desired pattern into the liquid. Sounds simple but remember this is at a minuscule scale and requiring absolute precision. Something as simple as air being trapped between the mask and the silicon wafer would completely derail the process, so the developers and designers of the machines had an exceptional challenge to avoid any external elements. As you might expect, more than one mask will be required over the lifetime of a NIL system, so these too are created using a machine that is also manufactured by Canon. “Essentially, the two machines together create an in-house sourced process or for nanoimprint technology,” explains Chris.

Finally, when the mask is removed, incredibly tiny structures remain, and these are cured with UV light. These intricate and quite beautiful geometric patterns are invisible to the naked eye, their size being only a few ‘nanometres’, hence the name. To put that into context, a nanometre is one billionth of a metre and a human hair will measure around 100,000 nanometres in diameter. “The smaller the ‘feature size’ [the tiny physical structures] on the silicon chip, the faster the device it’s used in will operate,” explains Chris. “Phones become faster; your PC is quicker. The more advanced your lithography equipment is, and able to print smaller feature sizes on the chips, the better the performance of that chip will be.”

Of course, this will require investment in this new technology on the part of chip manufacturers, but Canon considers this a wise move in the long term. “The overall cost of ownership shows that this is a technology worth investing in,” explains Chris “In terms of running costs, throughput and longevity.” Cost, of course, comes in many shapes and forms. So, from a machine perspective, it’s understood that NIL as a process will offer excellent value for money to manufacturers, not just in the initial investment, but by virtue of the way the technology operates.

For example, when you compare it to the closest alternative (‘Extreme Ultraviolet Lithography’ or EUV) or even traditional semiconductor lithography, both power consumption and waste are substantially lower. The precision nature of the process means that there is little in the way of excess material to be discarded and this too markedly reduces the use of chemicals. Both of which can only be a good thing, environmentally. Combined, we are looking at the kind of progress that will not only prolong the life of Moore’s Law in the traditional sense – in terms of processor speed and power – but adds a new sustainable aspect to the manufacture of semiconductor chips.

Meet the team behind the development of Canon’s Nanoimprint Lithography System.