In the late ‘80s and ‘90s, when our magazine staff gathered for dinner we often made a toast: “Here’s to chip silicon!” I really believed (and still do) that making electronics more affordable would increase their use and make our lives better and the world a better place to be.
I haven’t toasted to cheap silicon for a while. Why? Because that mission has been accomplished.
At SEMI’s ISS, Paul Farrar, manager of the G450C consortium put the industry progress over the last 40+ years in perspective. “1 Megabyte of memory in 1970 was $750,000. It was sold as an IBM add-on,” he said. “The great technology was made of 57mm wafers, five masking levels, and one level of metal. Today, it’s is less than a penny. That is a 100 million X improvement.”
Of course, most people would like to see this trend continue, but it’s highly unlikely that we’ll see such dramatic progress. Scaling is getting too expensive. The transition to 450mm looks feasible from a technical standpoint (see my column on pg. 10) but it’s not yet clear if it will be less expensive than 300mm, particularly when you factor in 450mm lithography.
So if the scaling mission is accomplished, what’s next? There’s exploding interest in the “Internet of Things” where almost everything is tagged and connected. That will require some big upgrades in the server/network infrastructure, but that can be done with existing technology. It will also require inexpensive sensors and wireless communication. By some estimates, the technology to achieve that is not ready. We need about a 10X improvement in price/performance. Ditto for wearable electronics and a whole host of applications in medical, automotive and the smart grid.
In the future, perhaps electronics will be printed like potato chip bags on roll-to-roll machines with ink-jet-like deposition of materials. Perhaps tiny MEMS with integrated sensors, thin-film batteries, energy harvesting, microprocessors and other functions will be produced for less than a penny. Perhaps everyone will have inexpensive body area networks embedded in their clothing that constantly monitor their health.
It’s all possible, but it will take some innovation in processing equipment and materials.
The answer to the question in the last (actually next to last) paragraph is already in the statement in the first (ok, third). if 1M+ transistors cost less than a penny, how many do the number required to effect a wireless sensor cost?
The technical problem is in the packaging, for which Terepac has developed the solution (and it is not potato chip bag printing, which is not nearly as cheap as that 1Mb of memory). There is also a business problem, however: not even RFID tags (much simpler and potentially cheaper than wireless sensors) are selling for a penny each, because unless one makes trillions of them annually, the business is not there by today’s standards (a $1M annual revenue is not catching anyone’s attention). So there has to be a growth route from millions of devices to the trillions. That is by far the more complex problem.