Companies Put Their Heads Together to Make Chips that Stack Up
IBM will work with materials manufacturer 3M to develop the necessary mortar to build much more complex three-dimensional computer chips. The companies announced this week that they will aim to develop microchips made of 100 chip layers stacked on top of each other. Stacking chips in this way could make all sorts of electronics faster and more power-efficient.
Three-dimensional chips have already found their way into some niche applications, but they are expensive to make, and can only be stacked about a dozen layers high before they overheat.
Three-dimensional chips can handle data more efficiently because data has to travel less distance to reach a different component. Stacked chips with connections running through them vertically like pipes in a skyscraper should be able to process more data faster, and with lower power requirements.
Eby Friedman, a professor of electrical and computer engineering at the University of Rochester who is not involved with either company, says that because of heat-management issues, today’s three-dimensional chips max out at around a half dozen layers even in research labs. “These chips are burning a lot of power, very close to one another, and thermal effects will become dominant,” he says.
What’s needed is a material that sits between each layer—mortar that 3M hopes to create—to glue them together but also rapidly shunt heat off the chips. “We need a material that absorbs mechanical stress, conducts the heat away very rapidly, and is fantastically electrically insulating so you don’t get shorts,” says Bernard Meyerson, vice president of IBM Research.
Ming Cheng, technical director of 3M’s electronic materials division, says the company will aim to find such a material by expanding its existing group of adhesive and electronic materials through a combination of computational simulation and trial-and-error work in the lab.
New chip designs, which IBM is working on as its part of the collaboration, are also key to making three-dimensional chips work. Friedman says heat problems have remained in part because chipmakers have, for the most part, seen making three-dimensional chips as a packaging problem, not a chip-design problem. “You have to change the design of the chips themselves to have heat sinks and other features—we have to think about heat production in a primary way, not just think about designing for performance,” he says.
The three-dimensional stacks will be built on a base of a relatively conventional chip, topped with chips that have been thinned to half their normal size so that the entire structure doesn’t become too thick, says IBM’s Meyerson. A major design challenge for IBM, he says, is figuring out a way to make these stacks not chip by chip, but wafer by wafer. Manufacturing at this scale is the only way to take three-dimensional chips from a niche product to a commercial one. “I think IBM is going to be the first to have high-volume commercial products coming out,” predicts Friedman.
Geoffrey Hinton tells us why he’s now scared of the tech he helped build
“I have suddenly switched my views on whether these things are going to be more intelligent than us.”
Meet the people who use Notion to plan their whole lives
The workplace tool’s appeal extends far beyond organizing work projects. Many users find it’s just as useful for managing their free time.
Learning to code isn’t enough
Historically, learn-to-code efforts have provided opportunities for the few, but new efforts are aiming to be inclusive.
Deep learning pioneer Geoffrey Hinton has quit Google
Hinton will be speaking at EmTech Digital on Wednesday.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.