Closed-loop AI enables time- and resource-efficient optimization of processing pathways without microstructural input

Over the past decade, the application of artificial intelligence (AI) tools has accelerated the discovery of novel material compositions, but has done little to speed the discovery of novel material processing paths, particularly when sequential processing operations are required. Current process optimization often involves extensive efforts to characterize microstructures and properties to fully understand process-microstructure-property relationships. However, characterizing both microstructures and properties slows and limits the scalability of materials optimization. Here, we demonstrate that processing pathways can be optimized within an AI-driven closed loop \textit{without} the need for intermediate microstructural characterization. By varying processing steps and parameters in a controlled manner and focusing on properties that can be quantified rapidly, our multi-objective framework simultaneously targets two competing properties: resistance to deformation (hardness) and resistance to corrosion (corrosion rate). We demonstrate this approach in ZX10, a lean, biodegradable magnesium alloy with potential applications in temporary orthopedic implants. Within three closed-loop iterations, the framework identifies processing pathways that move beyond the hardness–corrosion trade-off observed in the initial dataset, achieving high hardness and low one-day corrosion rates. Subsequent mechanical testing and extended corrosion experiments confirm that the optimized materials also exhibit increased tensile strength, enhanced ductility, and favorable corrosion performance over 14 days. Post-optimization microstructural characterization reveals that the selected materials share similar microstructural features, including ultrafine grains, comparable dislocation densities and crystallographic textures, and finely dispersed second-phase particles. These findings demonstrate that rapidly measured properties can guide the identification of favorable processing conditions without requiring microstructural information during optimization, while subsequent characterization provides insight into the underlying structure–property relationships. By delaying resource-intensive microstructural characterization until promising processing conditions are identified, this framework substantially reduces the experimental burden and offers a scalable strategy for AI-driven materials processing.

Authors

Institutions

Publication Details

Journal
Research Square
Published
2026-11-19
DOI
https://doi.org/10.21203/rs.3.rs-11719391/v1
Primary Topic
Machine Learning in Materials Science
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Closed-loop AI enables time- and resource-efficient optimization of processing pathways without microstructural input

Sreenivas Raguraman, Paulette Clancy, Maitreyee Sharma Priyadarshini, Andrew Kim et al.
Research Square
Machine Learning in Materials Science
preprint

Closed-loop AI enables time- and resource-efficient optimization of processing pathways without microstructural input

Sreenivas Raguraman, Paulette Clancy, Maitreyee Sharma Priyadarshini, Andrew Kim, Adam J. Griebel, Timothy P. Weihs, Ryan McGovern, Veronica Ivanovskaya, Tunde Ayodeji, Camryn Byrum
preprint en

Abstract

Over the past decade, the application of artificial intelligence (AI) tools has accelerated the discovery of novel material compositions, but has done little to speed the discovery of novel material processing paths, particularly when sequential processing operations are required. Current process optimization often involves extensive efforts to characterize microstructures and properties to fully understand process-microstructure-property relationships. However, characterizing both microstructures and properties slows and limits the scalability of materials optimization. Here, we demonstrate that processing pathways can be optimized within an AI-driven closed loop \textit{without} the need for intermediate microstructural characterization. By varying processing steps and parameters in a controlled manner and focusing on properties that can be quantified rapidly, our multi-objective framework simultaneously targets two competing properties: resistance to deformation (hardness) and resistance to corrosion (corrosion rate). We demonstrate this approach in ZX10, a lean, biodegradable magnesium alloy with potential applications in temporary orthopedic implants. Within three closed-loop iterations, the framework identifies processing pathways that move beyond the hardness–corrosion trade-off observed in the initial dataset, achieving high hardness and low one-day corrosion rates. Subsequent mechanical testing and extended corrosion experiments confirm that the optimized materials also exhibit increased tensile strength, enhanced ductility, and favorable corrosion performance over 14 days. Post-optimization microstructural characterization reveals that the selected materials share similar microstructural features, including ultrafine grains, comparable dislocation densities and crystallographic textures, and finely dispersed second-phase particles. These findings demonstrate that rapidly measured properties can guide the identification of favorable processing conditions without requiring microstructural information during optimization, while subsequent characterization provides insight into the underlying structure–property relationships. By delaying resource-intensive microstructural characterization until promising processing conditions are identified, this framework substantially reduces the experimental burden and offers a scalable strategy for AI-driven materials processing.

Research Square
Johns Hopkins University (US), ETH Zurich (CH), Fort Wayne Metals (United States) (US), Virginia Tech (US)
Machine Learning in Materials Science
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon