SEMI Strategic Materials Conference (SMC) 2025

This year, the event will feature keynote sessions from us, Microsoft, NVIDIA, and TSMC:

• Our Surésh Rajaraman, PhD | EVP, Thin Films "Integrated Material Development Modality Enabling the AI Revolution"
• Nvidia John Hu, PhD | Director, Advanced Technology Group "Accelerating AI Computing with Innovations from Materials to AI Factories"
• TSMC Carlos H. Diaz, PhD | Senior Director "Semiconductor Industry – A Logic Technology Outlook"
  
• Microsoft Chetan Nayak | PhD Technical Fellow "Reinventing the Transistor for the Quantum Age"

Integrated Material Development Modality Enabling the AI Revolution 

The rapid advancement in the hardware needed to enable to burgeoning AI revolution continues to require new deposition materials that enable increased performance by reducing energy utilization and increasing data processing capacity. Research required to develop such novel materials for the semiconductor industry therefore needs significant acceleration to meet the stringent timelines of the anticipated technology nodes.

By employing internally developed functional device proxy models, we can screen newly developed precursors and deposition processes more quickly and effectively to avoid following non-productive leads and thereby more quickly introduce viable solutions to our business partners. This process development modality, when coupled with precursor innovation capabilities, has proven to be a winning model.

In this talk, we will showcase examples of this development model in materials for both advanced logic and memory. For capacitive memory, combining machine learning with MIMCAP screening has accelerated material discovery and improved failure assessment, enabling faster transitions to new high-k dielectrics.

To overcome mobility limitations in shrinking devices, we have developed ALD 2D-materials as high-mobility channel solutions, demonstrating their integration into 3-terminal TFT devices. Additionally, to address the challenges of work-function (WF) materials for AI computing, we employed computational modeling and combinatorial deposition, narrowing down thousands of candidate materials for precursor development.