Would IMEC Qualify As A Technology Provider For Commercial Silicon #FabInIndia?

Interuniversity MicroElectronics Centre (IMEC) of Belgium is what can be called an intermediate research organisation in the world of semiconductor fabs. Dr Santhosh Onkar has covered its history in detail in this article.

First, let us look at what the policy says. As per the gazette notification of 21 December 2021, a screenshot of which is given below, the ‘operational experience’ requirement has two options.

The low-volume production capability of approximately 500 wafers per year is much less than the 40,000 wafers per month that the government of India notification has asked for. Moreover, that capacity at IMEC too does not seem to be a constantly running one, but more of a "need-to" basis when there is a partner needing product development. So it may not be possible to count IMEC as a fab as mentioned in option A of the notification.

How about option B — licensing a technology ? A paper titled "Models Of Technology Development In Intermediate Research Organisations", says that "With respect to its overall strategic positioning in technologies’ lifecycles, IMEC’s core research activities are concentrated in early phases where potential commercial value starts to emerge out of basic science".

To understand technology lifecycle in the context of semiconductor fabs and to look at what IMEC can offer from a 'technology readiness' perspective, I will use the terminologies mentioned in this presentation (slide 4) which were used in IBM's high volume manufacturing fabs:

Checkpoints include T0, T1, T2, S0, S1 and S2.

> T0 and T1 includes basic development with increasing levels of product qualification.

> S0 and beyond involves release to manufacturing and transition to quality control.

What I have come to understand is that IMEC can provide at best a T0 or equivalent technology. Ramping up from T0 to the S0/S1/S2 stages and achieving the typical yield percentages that are needed for high volume manufacturing can take close to three years — in the Indian context, after the new fab is built and equipped. This may not then qualify for the “production grade licensed technology" that is mentioned in option B.

Given that new commercial silicon fabs in India will need a quick ramp up in production in order to be able to capture the market, the ideal (and what the policy seems to ask for) scenario will be an established commercial fab bringing a technology that is already running in high volume production to a 'branch' that it opens in India on its own or as a joint venture (option A) or licence that to a business partner wanting to open a fab in India (option B).

Established commercial fab running high volume production are the likes of TSMC, Samsung, Intel, UMC, GlobalFoundries, Tower Semiconductors etc. IMEC's strengths as a leading R&D organisation are different — India should make use of those in the right areas. For example (but not limited to) help boost the R&D ecosystem that already exists in India through R&D fabs at IISc, IITs etc — more on that later.