Applied Materials on Monday unveiled 2 deposition systems specialised for sub-2-nanometre gate-all-around (GAA) transistor processes. As demand for AI computing expands, the semiconductor industry is shifting to GAA structures with higher energy efficiency. GAA transistors used for this require more than 500 process steps to form internal 3D structures, and many steps demand tolerances close to atomic scale, the company explained.

The 2 systems unveiled each handle different functions and structures that determine performance and power efficiency in the GAA process. The Producer Precision Selective Silicon Nitride PECVD (plasma-enhanced chemical vapour deposition) protects the shallow trench isolation (STI) structure that electrically separates adjacent transistors. Using the industry’s first selective bottom-up deposition method, it forms a silicon nitride film inside insulation trenches to prevent damage to insulating materials in subsequent processes, and reduces parasitic capacitance to raise chip performance per watt.

Outside the insulation structure, the Endura Trillium ALD (atomic layer deposition) controls the metal gate stack with atomic-level uniformity. The gate stack of a GAA transistor must fully wrap multiple horizontal nanosheets arranged at 10-nanometre intervals. The system integrates multiple metal deposition steps into a single platform to flexibly tune threshold voltage by transistor, and it newly adds functions to deposit work-function metals and dipole materials tailored to the GAA structure.

Prabu Raja (프라부 라자), president of Applied Materials' Semiconductor Products Group (SPG), said, "The semiconductor industry is entering a period of rapid and non-linear change where existing lithography-based chip scaling alone has reached its limits." He added, "At angstrom-class advanced logic nodes, performance and power efficiency are now determined by materials innovation."

He added, "The new deposition systems, based on Applied’s unmatched leadership in materials engineering, support customers in achieving the transition of key transistor technologies that underpin AI computing roadmaps."