MIT Combines GaN and Silicon Chips for Faster, Efficient Electronics

Published  June 18, 2025   0
User Avatar Abhishek
Author
Hybrid GaN-CMOS Chip Enables Faster, Efficient Electronics

Researchers from MIT and elsewhere have developed a new way to combine two types of chips: gallium nitride (GaN) and silicon-based CMOS chips into a single chip. The solution is cheaper, scalable, and more energy-efficient. Gallium nitride chips are not cheap or easy to work with, but they are faster and better at handling power. Silicon chips, on the other hand, are cheap and common but lack the same power. This method gets the best of both.

They created a power amplifier with this tech, a crucial component in wireless communication, such as in phones and Wi-Fi. Tiny GaN transistors were created as opposed to using a larger piece. These tiny transistors were cut out one by one into pieces called dielets and are bonded directly to a silicon chip using copper. The bonding process takes place at under 400°C, a safe temperature for both chip materials, reducing damage and costs, making the process compatible with existing chip factories.

The hybrid chips achieve higher signal strength, better efficiency, reduced heat buildup, lower energy use, and possibly longer smartphone battery life. This could improve current electronics, support quantum computing, and even be of use in data centers, radar systems, and high-speed communications.

Making whole chips from GaN can turn out expensive, with a lot of material ending up wasted. This method uses only the absolutely required portions of GaN. To make this work, the team built a machine that uses vacuum suction to pick up tiny transistors and place them on the chip with nanometer precision. The chip was successfully used to build a radio frequency (RF) amplifier and displayed better performance than traditional ones. This work addresses limits in Moore’s Law by showing how multiple chips can be stacked or combined through 3D integration

Add New Comment

Login to Comment Sign in with Google Log in with Facebook Sign in with GitHub