Extending the scope of the Dutch integrated-photonics foundry with microtechnologies benefits both domains, argues Paul van Gerven.
Intel’s announcement that it will build a fab in Germany won’t have come as a surprise to anyone who had been paying the slightest attention to the run-up of the European Chips Act. But the new strategy has produced some surprises as well. STMicroelectronics and Globalfoundries are looking to breathe new life into FD-SOI technology as a CMOS competitor, at least in a number of applications. The Spanish 12.3-billion-euro semiconductor ambitions have raised some eyebrows, too. And most recently, a plan has surfaced to build a foundry in the Netherlands. In Twente, to be precise.
It won’t be a hardcore CMOS foundry, of course. The initiators, a consortium of local (semi-)public organizations, target integrated photonics, microsystems such as MEMS and microfluidics, and combinations of these and other technologies through heterogeneous or back-end integration. The facility would double as a proving ground for – locally developed – equipment to manufacture such systems.
I think it’s a clever way of giving substance to Photondelta’s ambition to make our country a global hub for integrated-photonics manufacturing. That goal, backed by the Dutch government, involves constructing two foundries, one for silicon nitride (SiN) and one for indium phosphide (InP)-based integrated photonics. Although these two platforms are highly complementary, they can’t be manufactured at the same facility. In fact, there’s no particular reason why the foundries should be located close to one another. Given the pioneering role of Enschede-based Lionix in SiN integrated photonics, it would be only fair to award that foundry to Twente. InP, championed by Smart Photonics, is at home in the Eindhoven area.
As long as we’re building that SiN foundry, it’s a good idea to extend the scope beyond integrated photonics – not in the least because all but the most basic need electronics (and in the future, supposedly, electronics will need photonics as well). Given the compatibility of SiN’s manufacturing process with CMOS, that’s no problem at all. I assume there’s a significant overlap with microtechnology manufacturing, too.
Adding products to the mix offers opportunities to fill up production lines, which is quite useful since it’s unclear how demand for integrated photonics will develop. Balancing capacity with demand is tricky for an emerging technology, so it makes sense to hedge bets. A broader technology portfolio will put the foundry on more solid ground.
At this point, it should be noted that high-volume manufacturing of integrated photonics is still at least a decade away. I haven’t met a stakeholder who thinks otherwise. Consensus on the right way forward is a stepped approach, scaling up when changes in demand call for it. At the same time, you’d like to scale as fast as possible, since this increases efficiency and drives down cost. As different technologies ride their S-curves, the leaders of the pack give the rest a little tow.
Honestly, though, I don’t know what volumes to expect from microsystems and related technologies. East Netherlands is home to respectable companies in that domain, one of which was even acquired by Lam Research recently, as well as a university that maintains close ties with industry, but realistically you’d have to say that there wouldn’t have been a business case to build a foundry just for that. By hitching their ride to integrated photonics, there’s an excellent opportunity to see if there’s any untapped potential being wasted. Since we’re building a foundry anyway, let’s give it a try.
Main picture credit: University of Twente