AI-enabled RF-to-terahertz chip design beyond human intuition [Invited]

Area of Science:

• Electrical Engineering
• Computer Science
• Materials Science

Background:

• Traditional design of radio frequency integrated circuits (RFICs) and electromagnetic components is labor-intensive and relies on human intuition.
• Existing design methodologies often limit exploration of novel circuit architectures.

Purpose of the Study:

• To introduce a novel AI-enabled approach for the synthesis of complex RF-to-THz passive components and circuits.
• To demonstrate the capability of AI to automate and accelerate the design of advanced microelectronic chips.
• To expand the design landscape beyond traditional human-driven paradigms.

Main Methods:

• Development of an AI-driven framework for automated circuit synthesis.
• Application of the methodology to design complex RF-to-THz passive components.
• Comparison of AI-generated designs with traditional human-driven designs.

Main Results:

• The AI-enabled approach successfully synthesized complex RF-to-THz passive components and circuits.
• The methodology enabled the design of non-intuitive circuit architectures surpassing traditional paradigms.
• Automated design significantly reduced the labor and time associated with manual workflows.

Conclusions:

• The proposed AI framework represents a novel and effective method for RF-to-THz component and circuit synthesis.
• This approach moves away from manual, labor-intensive workflows, heralding a new era in RFIC and electromagnetic design.
• The ability to generate non-intuitive designs opens new possibilities for advanced microelectronic chip development.