Semiconductor manufacturing depends on specialized materials that can become vulnerable to geopolitical disruptions, limited suppliers, and environmental concerns. Researchers at the University of Michigan and Imec are addressing these risks through the Common Earth project, which seeks alternatives to critical materials and harmful chemicals used in chip production, tells IEEE Spectrum.
The project focuses partly on reducing dependence on rare earth elements and other critical materials. One example is hafnium, which is used in transistor gate dielectrics. Although hafnium is not technically a rare earth element, its availability is constrained because it is produced as a byproduct of zirconium mining associated with the nuclear industry. Increasing semiconductor production could therefore require scaling another industry as well.
Rare earth materials also appear in protective coatings for plasma-based semiconductor deposition equipment. Rather than depending on materials whose processing is concentrated in China, researchers are investigating more accessible alternatives that can withstand harsh plasma environments. They are also studying salt-based substitutes for hafnium.
Another priority is reducing PFAS, or per- and polyfluoroalkyl substances, generated during semiconductor manufacturing. The researchers are examining nitrogen-based precursors that could eliminate fluorine from some processes. Other University of Michigan teams are developing filtration materials that capture PFAS and remove them from manufacturing waste streams.
Replacing established semiconductor materials, however, can affect transistor performance. Alternative dielectrics might increase leakage or reduce other desirable characteristics. The project therefore combines materials research with circuit, architecture, and system-level approaches, including improved power gating and frequency throttling.
Researchers are also exploring chiplets as a way to reduce reliance on single-source designs. Reusable chiplet building blocks could be combined in different quantities to create lower- or higher-performance CPUs and GPUs.
Ultimately, Common Earth aims to make semiconductor manufacturing less vulnerable to material shortages, geopolitical disruptions, and environmental restrictions. Rather than treating these challenges separately, the project considers materials, manufacturing, architecture, sustainability, and supply chains as interconnected engineering problems.
