Building a Foundry for the Future

The transformation of the 1980s-era fab is well underway. Engineers and technicians are actively installing and testing advanced chip manufacturing equipment, with full operational capacity anticipated by the first quarter of 2026. However, simply having the tools isn’t enough. The real challenge lies in establishing a predictable and reliable manufacturing process for these novel chip architectures.

One of the key hurdles is the inherent variability in non-silicon materials. Unlike standardized silicon wafers, these alternative materials often differ in size and exhibit varying thermal expansion properties. Successfully integrating these disparate materials requires micrometer-level precision. To address this, TIE is developing comprehensive process design kits (PDKs) and assembly design kits (ADKs). The ADK, in particular, is critical, defining the rules for 3D assembly and advanced packaging.

Exemplar Projects Driving Development

TIE will refine these kits through three initial “exemplar” projects identified by NGMM: a phased-array radar, an infrared focal plane array, and a compact power converter. These diverse applications will serve as a proving ground, establishing a roadmap for broader innovation. “These projects provide an initial on-ramp into a tremendous range of potential applications,” Holmes emphasizes.

The foundry is designed to operate as a “high-mix, low-volume” facility, capable of handling a wide variety of projects without focusing on mass production. This contrasts sharply with traditional silicon foundries, which thrive on economies of scale. To overcome the limitations of limited test runs, TIE is leveraging artificial intelligence, developed by Austin-based Sandbox Semiconductor, to predict the outcomes of process adjustments.

The NGMM program is also fostering collaboration with academic institutions. Researchers at institutions like UT Dallas, led by Ted Moise, are exploring advancements in thermal conductivity films, microfluidic cooling, and failure analysis – all crucial for the success of 3DHI.

Acknowledging the unconventional nature of the program, Whitney Mason, director of DARPA’s Microsystems Technology Office, admits, “NGMM is a bit of a departure for DARPA. We don’t typically stand up manufacturing facilities.” However, given Austin’s reputation for innovation and its unofficial motto, “Keep Austin Weird,” the partnership appears uniquely suited for success.

What impact will this new level of chip integration have on industries like aerospace, defense, and automotive? And how will the success of TIE influence global semiconductor supply chains?

Frequently Asked Questions About 3D Heterogeneous Integration

• What is 3D heterogeneous integration (3DHI)?

  3DHI is the process of stacking chips made from different materials – both silicon and non-silicon – to create a more powerful and efficient integrated circuit. This allows for functionalities that are impossible to achieve with traditional 2D chip designs.

• How does 3DHI differ from traditional chip stacking?

  Traditional chip stacking typically involves stacking multiple layers of silicon. 3DHI goes further by integrating a wider range of materials, like gallium nitride and silicon carbide, to optimize performance for specific applications.

• What are the potential benefits of 3DHI?

  3DHI promises significant performance gains – potentially up to 100x improvement – increased energy efficiency, and the ability to integrate more functionality into a smaller footprint. This opens doors to new possibilities in areas like AI, high-performance computing, and power electronics.

• What role does DARPA’s NGMM program play in advancing 3DHI?

  DARPA’s NGMM program is providing the funding and infrastructure – specifically the TIE facility in Austin – to accelerate the development and manufacturing of 3DHI technologies. It’s a crucial step towards establishing U.S. leadership in this critical field.

• How will the TIE facility support hardware startups?

  TIE offers startups access to advanced manufacturing capabilities and expertise that are often unavailable elsewhere. This helps bridge the “lab-to-fab” gap, enabling them to prototype and scale their innovations more effectively.