Because the pandemic exacerbated what was already arguably a formidable provide chain downside, the semiconductor trade continued—because it has for years—to focus intensely on discovering methods to extend effectivity, precision, and yield. The pandemic and speedy shopper shopping for enthusiasm merely functioned as an accelerator to hammer residence what the trade already knew: swift advances are required that facilitate elevated productiveness.
Among the many most promising approaches for reaching increased yield inside the semiconductor trade is bigger reliance on software program and collaboration all through the trade to cohesively combine fragmented and siloed processes that inhibit progress.
Whereas software program for gear monitoring, downside troubleshooting, information evaluation, and report era is in place all through the trade, the software program options are usually both home-grown or based mostly on an assortment of options from a number of distributors. Because the sources accessible to any given producer are restricted, it may well take months to put in new options and enhancements. Even after repeated requests, the brand new software program capabilities typically by no means really seem. With options from a number of distributors in place, the onus is on the IC producer to contact every particular person vendor, replace all the software program at play on the manufacturing line, after which hope that all the pieces is suitable and works seamlessly collectively.
In keeping with members in a current survey (Determine 1), professionals within the semiconductor manufacturing chain need instruments which can be simpler to make use of, extra clever, in a position to assist all their gear, can automate the method of reporting and alerting, and are verifiably safe. But when producers try to combine gear monitoring efficiency on their very own, the outcomes will be lackluster.
Determine 1 A survey carried out by Gigaphoton supplies insights into the present state of the semiconductor manufacturing trade.
Prescriptive analytics
Among the many largest offenders are analytics instruments that provide primary descriptive and diagnostic options however don’t successfully deal with predictive and prescriptive upkeep. Predictive upkeep makes use of statistical and modeling strategies to find out what might occur sooner or later based mostly on historic information and makes an attempt to find out when a machine wants servicing. What it doesn’t do is decide what motion needs to be taken; it merely informs the consumer that upkeep is required. The predictive upkeep strategy is slowly being built-in into manufacturing sectors inside the context of Trade 4.0.
In distinction, prescriptive upkeep, whereas much less prevalent and properly developed, takes evaluation a notch increased. It not solely predicts failure occasions in gear, however can suggest actions as properly. The place predictive upkeep supplies info to find out whether or not to carry out or defer asset upkeep, prescriptive upkeep gives a collection of choices and outcomes from which to pick. Sooner or later, the prescriptive strategy might even schedule upkeep and assign the very best accessible technician or workforce to make the restore or alternative.
Prescriptive analytics powered by synthetic intelligence (AI) instruments might even help upkeep groups by offering detailed directions whereas the treatment is being deployed. With the huge quantities of apparatus and sensors inside a typical fab, each approaches can enhance each manufacturing course of as they reduce the likelihood that some very important element might fail or be degraded earlier than well timed motion will be taken.
Determine 2 The 4 ranges of doable insights are offered by analytics instruments that vary from descriptive to prescriptive. Supply: Gigaphoton
Frequent denominators
What’s attention-grabbing about this situation of predictive versus prescriptive upkeep is that whereas each semiconductor firm’s processes are totally different and proprietary, they’ve many widespread issues. To discover a resolution, it will make sense to have a single software program platform that might enable all gear producers to collaborate. That’s, it will give chip producers the flexibility to leverage the expertise of others who’ve solved related issues.
In fact, such a platform should be sure that members’ mental property (IP) could be inherently safe, and related efforts in different industries present that this purpose will be achieved. With this assurance in place, the platform would enable cross-vendor gear monitoring and evaluation to be carried out whereas enabling every vendor to resolve their particular issues. The power for personalization could be important to fulfill the distinctive wants of every participant. The platform additionally needs to be agnostic to varied programming languages and accommodate all or at the least essentially the most prevalent languages. One other good thing about the platform could be simpler use of AI integration inside crucial processes.
This collaborative strategy is greatest achieved on a single open platform that eliminates the necessity to depend on merchandise from a number of, typically incompatible, sources. It’s doubtless that downtime and repair calls may very well be diminished as a result of producers might rapidly and simply decide for themselves which device is inflicting the issue, and it will be doable to find out that some points could be cross-functional. The strategy would dramatically improve effectivity and yield that, in flip, might drive up the speed of manufacturing.
Briefly, developments in software program instruments have turn out to be important for the semiconductor trade to speed up time to market. However the sheer variety of these instruments, each developed in-house and equipped by third-party distributors, has additionally turn out to be a hindrance to the targets the trade is attempting to realize. As this text demonstrates, an environment friendly option to harness all these instruments is to share details about them amongst opponents with out revealing the “secret sauce” that’s the core of every firm’s success.
A single software program platform
The semiconductor trade’s present fragmentation of monitoring and evaluation instruments hinders its capacity to reply to challenges produced by provide chain interruptions that can virtually actually seem once more sooner or later. Rising options like Fabscape, a brand new open platform imagined by the laser producer Gigaphoton, supplies actionable insights that have been by no means recognized earlier than.
Determine 3 A rendering of Fabscape, a brand new open platform idea that goals to carry collectively streamlined information monitoring, administration, and analytics throughout the manufacturing line.
Such open, collaborative platform options will not be new, and even the Division of Protection (DoD) more and more employs them to encourage higher integration with the personal sector. Though the DoD initiative confronted early hurdles as corporations feared their IP could be compromised, assurances put into place slowly made them extra comfy with collaborating with their opponents—and with the federal government. The important thing to their success has been collaboration, belief and mutual profit, and there’s no cause why one thing related can’t be utilized by the semiconductor trade.
In abstract, to extend productiveness on an industrywide scale, semiconductor manufacturing could be properly served by a platform the place all analytics, fault detection, optimization, and communication happen in a single place and with the depth of perception that solely vendor-specific intelligence might present.
Yuji Minegishi is the overall supervisor at Gigaphoton Inc., a producer of sunshine sources for semiconductor lithography processes. He leads the engineering and improvement efforts for Fabscape.
Â
Associated Content material
