The Mighty Sensor In The Fab

The days of scheduled maintenance on fab equipment are coming to an end. In fact, the entire service model as we know it is about to undergo a mammoth change.

The addition of more sensors into manufacturing equipment may seem like an evolutionary step, but the impact is going to be much more significant than it might appear. Rather than just alerting fab managers or equipment makers when a part is not working, real-time data can be used to identify patterns of degradation and make adjustments prior to failures. That, in turn, affects everything from how and when that equipment is serviced, to how to compensate for variation and improve consistency and quality of results.

This is long overdue, although not entirely surprising. Change in complex manufacturing processes typically happens slowly in any industry. Entire supply chains build up around the successful processes, and any disruption can have significant ramifications.

What’s different about the introduction of more sensor data is that it isn’t terribly disruptive to the companies that directly benefit from that data—the fabs and the equipment makers—and there is clear economic benefit for both of them.

For fabs, the upside is servicing of equipment only when it’s needed, and they will be able to have more control over when that service is required. Some parts may last longer than expected, while others may need to be replaced more frequently. And with the addition of AI, they will be able to use that real-time data to maintain optimal consistency.

Equipment makers, meanwhile, can both sell more equipment and service that equipment on an as-needed basis, rather than a standard maintenance schedule, and they can charge more for both. For years, the service arms of equipment makers have been viewed as a profitable side business. This could bring them squarely into the mainstream because service will be required in real-time. So rather than providing maintenance when things break, they will be tasked with keeping a fab running at full speed.

In the chip manufacturing world, these are revolutionary concepts. But they also have far-reaching implications for a variety of other markets that depend on semiconductors. Process variation is a major factor in chip reliability, and that’s becoming much more important as the electronic content in cars, medical devices and industry increases, and as those devices are built using the most advanced nodes.

Having consistency in manufacturing is critical for all of these markets, and binning is not an option. These are highly customized, smaller-batch production runs, and inferior-grade chips can cause physical harm. Moreover, OEMs won’t tolerate fluctuations in quality because they are taking on greater liability as more intelligence is built into these end devices. Even in the consumer market, if consumers are spending more than $1,000 on a smart phone—prices for the new foldables are expected to be nearly twice as much—they won’t tolerate a degradation in performance.

More sensors and better use of that data can go a long way toward improving manufacturing quality and reducing defects. And while this may seem like a fairly basic change, the implications are enormous.