HPCWire covers Mat3ra AlphaFilm - A DOE Genesis Mission Project to Accelerate Thin Film Discovery

Mat3ra is leading AlphaFilm: Closed-Loop Agentic AI for Thin-Film Materials, one of over 300 Phase 1 projects selected this year under the U.S. Department of Energy's Genesis Mission, announced this past July.

Accelerating a notoriously hard problem

Thin films used in semiconductor manufacturing are difficult to develop. Unlike areas of materials discovery that lend themselves to pure simulation, thin films are full of real-world defects and irregularities that have to be scanned, analyzed, and modeled before a winning recipe is found. With AlphaFilm, we're aiming to compress that cycle — from what today can take weeks or months down to days or hours.

Our focus is on ferroelectric wurtzite nitride thin films, including aluminum scandium nitride and alloyed aluminum nitride — materials that could underpin a new generation of semiconductors for non-volatile memory, RF filters, neuromorphic computing, and sensors for jet engines and spacecraft.

How it works

AlphaFilm combines physics-based and AI-based simulation with two key experimental capabilities: National Lab of the Rockies' combinatorial co-sputtering deposition, which can vary film thickness and composition across a single wafer, and the University of Tennessee, Knoxville's automated, multimodal scanning probe microscopy for fast characterization. Our role is to tie it together — we're developing an AI surrogate model, built on top of the underlying physics model, that factors in variables like temperature, density, roughness, ferroelectric response, and oxygen concentration to predict outcomes far faster than physics-based simulation alone, then feeding that back into an experimental loop with AI agents deciding what to try next. Our goal is defect-free pathways toward manufacturing these materials at scale.

The team

Mat3ra's Sergey Barabash is the principal investigator on AlphaFilm, which carries a $750,000 grant covering nine months of work. The team also includes Professor Sergei Kalinin from UT Knoxville and Andriy Zakutayev from National Lab of the Rockies, with an unnamed defense contractor and chip company also involved as collaborators. We're now pursuing a Phase II Genesis Mission grant to continue the work.

Two more Genesis Mission projects

AlphaFilm is one of three Phase 1 Genesis Mission projects we're contributing to this round:

• High-Fidelity AI-Ready Excited-State Materials Datasets and AI Models for Accelerated Discovery, led by Lawrence Berkeley National Laboratory with USC, Carnegie Mellon, Yale, and Nvidia — building high-throughput GW and GW-BSE datasets and training AI surrogate models for excited-state phenomena in quantum and optoelectronic materials.
• SPOTTER-AI: Cybersecurity for Scientific AI Workflows, led by Argonne National Laboratory with Oak Ridge National Laboratory, Illinois Institute of Technology, and Hewlett Packard Enterprise — developing an AI-powered framework to detect and trace integrity attacks on scientific AI workflows.

‍

Source: Woodie, A. (2026, September 14). Genesis Mission Project Uses AI to Speed Thin Film Development. HPCwire. Read the original article

‍

Tags: AlphaFilm, Genesis Mission, Thin Film Discovery, Department of Energy, Timur Bazhirov, Sergey Barabash