AMAZEMET automates ultrasonic atomization: rePOWDER with AI process control

AMAZEMET is integrating an AI model into its rePOWDER platform, aiming for largely unattended ultrasonic atomization for powder production. The solution targets a typical bottleneck in laboratories and pilot manufacturing: metallurgical processes require continuous operator presence and tie up personnel.

Technically, rePOWDER relies on machine vision: a welding camera provides a live stream of the melt pool, which the AI model evaluates in 120-millisecond cycles. On this basis, the system adjusts torch position and power, wire feed, as well as ultrasonic amplitude and frequency. The goal is stable wetting of the sonotrode, as this is crucial for high yields in the desired particle size distribution. In addition, the control monitors gas atmosphere, overpressure, and flow velocities, and manages a gas scrubber to reduce overspray.

Introducing the AI required a new control center. The Advanced Control Cabinet integrates an industrial GPU for edge inference, a recirculation and passivation system for the process gases, a plasma power source tailored to the platform, and a fast PLC. Via an API, the system can be integrated into industrial networks; material feeders record throughput. In benchmarking with Ti-6Al-4V wire, rePOWDER achieves up to 0.5 kg/h and at least four hours of unattended operation, with eight hours envisioned. Additional autonomous recipes for NiTi and the Nb alloy C-103 are in the works.

For materials research, the combination of automated powder production and data-driven process control is significant. It complements high-throughput approaches in CALPHAD-based ICME and ML-assisted alloy discovery, shortens feedback loops, and facilitates reproducibility.

“In most institutions, it is far easier to buy new equipment than to hire new technical staff”, said Dr. Łukasz Żrodowski, CEO of AMAZEMET, Adjunct Professor at Carnegie Mellon University.” AMAZEMET focused its efforts on limiting the time users must commit to atomizing novel alloys. Our new Ai process control delivers much more autonomy, allowing researchers to focus on discovery or supervise more devices and processes at the same time. “