The vision of MOSIS 2.0

| June 5, 2024

The next-generation of USC ISI’s semiconductor fabrication service is taking shape.

Photo of a semiconductor wafer in a person's hands.

Photo credit: Northrop Grumman

Established at USC Viterbi’s Information Sciences Institute (ISI) in 1981, MOSIS revolutionized the world of silicon-based semiconductor manufacturing. At the time, most high-performance chips were made in large-scale manufacturing efforts. But MOSIS provided critical access to semiconductor manufacturing resources to small-scale researchers.

Now, as the industry faces new challenges and opportunities, MOSIS is embarking on its next phase: MOSIS 2.0, a pioneering initiative at the heart of California DREAMS, the Defense Ready Electronics and Microdevices SuperHub.

The Legacy of MOSIS 1.0 

The original MOSIS, created with support from DARPA, transformed silicon-based semiconductor manufacturing by introducing the multi-project wafer model. This innovative approach allowed multiple projects to be placed on a single wafer, making it cost-effective for startups, universities, government agencies, and research institutes to experiment with chip designs. 

At its peak, MOSIS fulfilled around 3,000 orders per year, generating up to $10 million in sales. “MOSIS has been extremely successful,” Mike Haney, Deputy Director of the DREAMS hub, said. “Even though it was run by a university, it was a self-sustaining business for 40 years.”

Advancing Compound Semiconductors

MOSIS 2.0 is building upon the success of its predecessor by focusing on the rapidly growing field of compound semiconductors. These materials, such as gallium nitride and indium phosphide, offer superior performance characteristics at high frequencies, making them ideal for next-generation applications in radio frequency devices, defense systems and 5G/6G communication technologies.

“Compound semiconductors, experimental processes and heterogeneous integration all require higher levels of support from an organization like MOSIS 2.0 because of their immaturity relative to commercial silicon-based processes,” Steve Crago, Director and Principal Investigator of DREAMS, said.

Unlike silicon-based semiconductors, which have well-established and standardized manufacturing processes, compound semiconductors present new fabrication challenges that require specialized expertise and resources. MOSIS 2.0 aims to address this by not only continuing the multi-project wafer service for these materials but also advancing their fabrication processes. 

One of the key obstacles in advancing compound semiconductor technology is the slow lab-to-fab transition, which involves moving semiconductor device designs from the research and development phase in lab settings to the mass production environment of a semiconductor fabrication facility, or fab. “There’s a lot of R&D done in universities on new devices and materials to make analog chips,” Haney said. “But this process has been hindered because there was no mechanism for university labs to readily work with full fabs.” 

The strength of collaboration

The DREAMS coalition solves this problem by bringing together university and industry partners. Led by USC, the 16 funded partners work like one organization, allowing microelectronics resources, including teams and tools, to be optimized, standardized, and accelerated across the hub.  

“We are creating a web of communication, not just a direct line,” Damon McCaskill, Senior Project Manager for MOSIS 2.0, said. “When there is an advancement of technology at one of our universities, we can share it across the whole hub at once.” 

“Cleaning up the pipes”

The driving force behind MOSIS 2.0 is the PIES team, an acronym for Prototype Integration and Engineering Services, formed by engineers from each hub organization. Together, they work on projects called “pipe cleaners” to remove bottlenecks and streamline processes within semiconductor fabrication, just like a pipe cleaner clears blockages in a pipe. 

“Pipe cleaners are the fuel driving the MOSIS 2.0 operation,” said Haney. “The engine is the PIES team itself.” 

So far, the projects span a range of topics, with researchers from each partner institution bringing their unique expertise to the table. For instance, researchers at the University of California, Irvine are exploring the application of AI in equipment maintenance, aiming to improve efficiency in semiconductor manufacturing. 

Meanwhile, the University of California, Santa Barbara, home to one of the best university nanofabs in the country, is focusing on integrating advanced data analytics software into the manufacturing process. This could reduce system down time during product development, accelerating the lab-to-fab transition for semiconductor technologies.

The collaborative power of the DREAMS hub also shines through in several pipe cleaner projects. Northrop Grumman, an industry fab partner, has worked both with USC on chiplet packaging and the University of California, Los Angeles on supply chain issues, avoiding months in delays for each project. 

Launching the MOSIS 2.0 storefront

The PIES team will also have another role in the near future: serving MOSIS 2.0 customers, with a storefront that emulates the success of the original MOSIS. 

Set to launch in 2025, the MOSIS 2.0 storefront will offer end-to-end prototyping services for compound semiconductor devices. Customers, ranging from startup founders to established defense clients, will be able to submit their project requirements and be matched with a dedicated PIES team member. This expert will guide them through the entire process, from initial consultation to design, fabrication, and testing.

“You’ve heard the term white glove service. Well, that’s what we have here,” Haney said.

The storefront’s key advantage lies in its ability to provide custom solutions that leverage the entire catalog of tools and processes available across the California DREAMS hub. The PIES team is currently compiling a searchable database of these resources, which will enable them to efficiently identify the most suitable facilities and expertise for each customer’s specific needs.

For instance, if a customer approaches MOSIS 2.0 with an idea for a new compound semiconductor device that requires a novel deposition technique, the assigned PIES team member will work closely with the customer to develop and optimize this specific process at one of the hub’s university labs. On the other hand, if a customer requires a complete chip fabrication service, the PIES team will connect them with a fab partner, such as HRL, to leverage their manufacturing capabilities through a dedicated chip run.

Communication and cross-hub standardization — two main goals of MOSIS 2.0 — will ensure that each project can travel between hub institutions seamlessly.

“The way we operate as a hub is going to be a key part of our effectiveness as a business,” Zach Lemnios, a DREAMS Advisor and Consultant, said. 

A self-sustaining future

Over the next year, the MOSIS 2.0 team will focus on developing their storefront, refining the hub’s tools and processes, and establishing clear metrics to gauge success. The ultimate goal is to transform MOSIS 2.0 into a self-sustaining platform by the end of the five-year Microelectronics Commons program in 2028.

If MOSIS 2.0 succeeds, it has the potential to accelerate compound semiconductor research and development, providing a streamlined system for innovative technology to come to fruition. Additionally, a thriving MOSIS 2.0 could generate a significant revenue stream for the California DREAMS hub, ensuring its long-term sustainability and growth. This model is the first of its kind.

“​​We have such a superb team, combined with the extensive resources within our nano fabs and fabs,” said Haney. “If we can’t do it, nobody can.”  

Published on June 5th, 2024

Last updated on June 6th, 2024

Share this Story