A collaborative materials research team from NIMS and Nagoya University has introduced an innovative two-step thermal aging process, utilizing artificial intelligence (AI), to produce nickel-aluminum (Ni-Al) alloys with superior high-temperature strength compared to conventional thermal aging methods.
By employing AI techniques, the team identified multiple non-isothermal aging schedules that showed the potential to enhance the alloys’ strength at elevated temperatures. The mechanisms governing these schedules were subsequently elucidated through in-depth analysis, highlighting the potential of AI in generating new insights in materials research.
The study was reported in the journal Scientific Reports on August 4th, 2023.
The study performed focused on Ni-Al alloys with a γ/γ´ (gamma/gamma prime) two-phase microstructure. Enhancing the high-temperature strength of these alloys necessitates precise optimization of the size and volume fraction of the γ´ phase formed during thermal aging.
These parameters depend on how alloys undergo thermal aging—specifically, the temperatures and durations applied.
The potential combinations of temperature and duration are vast. To illustrate, segmenting thermal aging into 10 equal intervals with nine set temperatures yields around 3.5 billion potential combinations. This multitude of possibilities previously confined efforts to establish optimal thermal aging schedules using constant temperatures.
Earlier, the research team effectively cut down time and expenses by transitioning from experimental methods to computational simulations. However, with 3.5 billion combinations, the simulation of all options remained unfeasible.
Recently, the team implemented a Monte Carlo tree search (MCTS) system—an AI algorithm adept at condensing an extensive array of combinations into a select few optimal ones. Through this approach, they pinpointed 110 thermal aging schedules that outperformed traditional isothermal aging processes.
Initially, these patterns seemed intricate and divergent from the usual isothermal aging methods. Yet, upon meticulous scrutiny, the team uncovered the mechanisms underlying these patterns: commencing with a brief, high-temperature aging phase prompts γ´ precipitates to grow to nearly optimal sizes. Subsequent prolonged, low-temperature aging enhances their volume fraction while curbing excessive growth.
Armed with this revelation, the team devised a two-step thermal aging plan: a short stint of high-temperature aging followed by an extended period at low temperatures. This strategy proved to yield Ni-Al alloys with superior high-temperature strength compared to any identified by the AI algorithm as effective in thermal aging patterns.
The research team aims to leverage this AI-driven method in upcoming studies to enhance the high-temperature strength of complex nickel-based superalloys employed in gas turbines. This approach could significantly enhance their efficiency, marking a potential advancement in practical applications.
Nandal, V., et al. (2023) Artificial intelligence inspired design of non-isothermal aging for γ–γ′ two-phase, Ni–Al alloys. Scientific Reports. doi.org/10.1038/s41598-023-39589-2.