Applied Materials Launches AIx

Applied Materials, Inc. announced AIx, a platform that accelerates the discovery, development and commercial deployment of new chip technologies.

AIx, which stands for Actionable Insight Accelerator, enables engineers to see into semiconductor processes in real-time, take millions of measurements across wafers and individual chips, and optimize thousands of process variables to improve semiconductor performance, power, area-cost and time to market (PPACt), according to Applied Materials.

The AIx platform works across all Applied Materials process equipment, eBeam metrology systems and inspection systems and is extendable from lab to fab. By providing engineers with the ability to fingerprint innovative recipes during R&D, AIx accelerates their transfer and ramp into high-volume manufacturing (HVM). 

“Accelerating the ‘t’ in PPACt is the biggest value driver for all the companies in our ecosystem,” said Prabu Raja, senior VP and GM of the Semiconductor Products Group at Applied Materials. “AIx connects all the capabilities of Applied in new ways with the goal of cutting development time in half and improving process windows by one-third. We have been developing AIx over the past three years to provide engineers with an entirely new kind of toolkit to solve the increasingly complex challenges of our industry.”

“AIx uses the power of big data and AI to give customers better outcomes at every stage of the semiconductor technology lifecycle, from R&D to ramp and HVM,” said Raman Achutharaman, group VP, Semiconductor Products Group at Applied Materials. “Engineers have thousands of process variables to choose from, and only a handful of elusive correlations provide the key to optimizing recipes for world-class results. AIx identifies and magnifies this actionable data, providing engineers with the actionable insights needed to accelerate PPACt.”

“Once again, Applied Materials is adding real value to the semiconductor industry with big data analytics in the form of its AIx platform for the process engineering ecosystem,” said Dan Hutcheson, CEO and chairman of VLSIresearch. “AIx moves beyond the decades-old statistical process control methods based on linear data streams to a new multidimensional world where data from 3D images, in-situ metrology and sensors can be stacked and then distilled into information that can be acted on. Applied’s AIx is a new toolkit that promises to accelerate R&D, thereby shortening time to results and ultimately time to money. I expect AIx algorithms will be ported to production to control the process with real-time chamber control.”

The AIx platform includes:

• ChamberAI: New sensors and machine learning algorithms for Applied Materials process chambers that provide engineers with real-time analytics of variables including chemistry, energy, pressure, temperature and duration;
• On-board metrology: Unique in-vacuum metrology that enables new films to be measured as they are being deposited, with angstrom-level precision;
• Inline metrology: Unique algorithms based on Applied eBeam metrology which can provide a 100-fold increase in measurement speed versus legacy approaches along with 50 percent higher resolution. Engineers can obtain over one million 3D wafer measurements per hour to make nanometer-scale assessments of how minuscule changes in recipes affect on-chip devices and structures;
• AppliedPRO: Process Recipe Optimizer generates digital process maps that help accelerate materials and recipe development, reduce variability and widen process windows. AppliedPRO can be used to optimize individual chambers and tools as well as accelerate matching across a fleet of systems;
• Digital twins: The AIx platform includes digital twin models of select Applied Materials chambers and systems that enable virtual experiments which accelerate recipe development, improve matching and ramp transfer, and optimize output and yield in high-volume production;
• Computing: The AIx platform includes the computing resources needed to store and analyze massive data using machine learning and AI algorithms.