skip to main content
US FlagAn official website of the United States government
dot gov icon
Official websites use .gov
A .gov website belongs to an official government organization in the United States.
https lock icon
Secure .gov websites use HTTPS
A lock ( lock ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites.

Explore Research Products in the PAR
It may take a few hours for recently added research products to appear in PAR search results.


Title: AlloyGAN: Domain-Promptable Generative Adversarial Network for Generating Aluminum Alloy Microstructures
The global metal market, expected to exceed $18.5 trillion by 2030, faces costly inefficiencies from defects in alloy manufacturing. Although microstructure analysis has improved alloy performance, current numerical models struggle to accurately simulate solidification. In this research, we thus introduce AlloyGAN - the first domain-driven Conditional Generative Adversarial Network (cGAN) involving domain prior for generating alloy microstructures of previously not considered chemical and manufactural compositions. AlloyGAN improves cGAN process by involving prior factors from solidification reaction to generate scientifically valid images of alloy microstructure given basic alloy manufacturing compositions. It achieves a faster and equally accurate alternative to traditional material science methods for assessing alloy microstructures. We contribute (1) a novel Alloy-GAN design for rapid alloy optimization; (2) unique methods that inject prior knowledge of the chemical reaction into cGAN-based models; and (3) metrics from machine learning and chemistry for generation evaluation. Our approach highlights the promise of GAN-based models in the scientific discovery of materials. AlloyGAN has successfully transitioned into an AIGC startup with a core focus on model-generated metallography. We open its interactive demo at: https://deepalloy.com/  more » « less
Award ID(s):
2021871
PAR ID:
10523430
Author(s) / Creator(s):
; ; ;
Publisher / Repository:
IEEE
Date Published:
Journal Name:
Proceedings
ISSN:
1946-0759
ISBN:
979-8-3503-4534-6
Page Range / eLocation ID:
1649 to 1655
Subject(s) / Keyword(s):
Metals Generative adversarial networks Chemical reactions Aluminum alloys Manufacturing Numerical models Microstructure
Format(s):
Medium: X
Location:
Jacksonville, FL, USA
Sponsoring Org:
National Science Foundation
More Like this
  1. ; ; ; (, JOM)
    Abstract The solidification microstructures of plain and inoculated 6061 aluminum builds manufactured with gas metal arc-directed energy deposition were studied with a combination of models and experiments. Electron back-scatter diffraction (EBSD) showed that the plain 6061 build had large, columnar grains with intergranular solidification cracking, while the inoculated build had a near-equiaxed, fine grain microstructure with no solidification cracks. By combining EBSD and energy dispersive spectrometry, the inoculated build has been shown to have exhibited globular growth while the non-inoculated build displayed a dendritic microstructure. A combination of heat transfer and modified grain morphology models were employed to predict the solidification morphology of the 6061 builds, which closely matched experimental results. A modification is proposed to the criterion marking the transition from globular to dendritic growth that better matches experimental results in this work. The results of this study are expected to provide improved methods to predict solidification microstructure for the development of new materials and processing parameters for additive manufacturing. 
    more » « less
  2. ; (, Integrating materials and manufacturing innovation)
    High-throughput screening (HTS) can significantly accelerate the design of new materials, allowing for automatic testing of a large number of material compositions and process parameters. Using HTS in Integrated Computational Materials Engineering (ICME), the computational evaluation of multiple combinations can be performed before empirical testing, thus reducing the use of material and resources. Conducting computational HTS involves the application of high-throughput computing (HTC) and developing suitable tools to handle such calculations. Among multiple ICME methods compatible with HTS and HTC, the calculation of phase diagrams known as the CALPHAD method has gained prominence. When combining thermodynamic modeling with kinetic simulations, predicting the entire history of precipitation behavior is possible. However, most reported CALPHAD-based HTS frameworks are restricted to thermodynamic modeling or not accessible. The present work introduces CAROUSEL—an open-sourCe frAmewoRk fOr high-throUghput microStructurE simuLations. It is designed to explore various alloy compositions, processing parameters, and CALPHAD implementations. CAROUSEL offers a graphical interface for easy interaction, scripting workflow for advanced simulations, the calculation distribution system, and simulation data management. Additionally, CAROUSEL incorporates visual tools for exploring the generated data and integrates through-process modeling, accounting for the interplay between solidification and solid-state precipitation. The application area is various metal manufacturing processes where the precipitation behavior is crucial. The results of simulations can be used in upscale material models, thus covering different microstructural phenomena. The present work demonstrates how CAROUSEL can be used for additive manufacturing (AM), particularly for investigating different chemical compositions and heat treatment parameters (e.g., temperature, duration 
    more » « less
  3. ; ; ; ; ; ; (, Nature Communications)
    Abstract Controlling microstructure in fusion-based metal additive manufacturing (AM) remains a significant challenge due to the many parameters that directly impact solidification condition. Multiprincipal element alloys (MPEAs), also known as high entropy alloys, offer a vast compositional space to design for microstructural engineering due to their chemical complexity and exceptional properties. Here, we use the FeMnCoCr system as a model platform for exploring alloy design in MPEAs for AM. By exploiting the decreasing stability of the face-centered cubic phase with increasing Mn content, we achieve notable grain refinement and breakdown of epitaxial columnar grain growth. We employ a multifaceted approach encompassing thermodynamic modeling, operando synchrotron X-ray diffraction, multiscale microstructural characterization, and mechanical testing to gain insight into the solidification physics and its ramifications on the resulting microstructure of FeMnCoCr MPEAs. This work aims toward tailoring desirable grain sizes and morphology through targeted manipulation of phase stability, thereby advancing microstructure control in AM applications. 
    more » « less
  4. ; (, Springer Nature Switzerland)
    Material scientists have made progress in controlling alloy performance through microstructure quantification. However, attempts at numerically modeling microstructures have failed due to the complex nature of the solidification process. In this research, we present the AlloyGAN deep learning model to generate microstructures for castable aluminum alloys. This innovative model demonstrates its capacity to simulate the evolution of aluminum alloy microstructures in response to variations in composition and cooling rates. Specifically, it is successful to simulate various effects on castable aluminum, including: (1) the influence of Si and other elements on microstructures, (2) the relationship between cooling rate and Secondary Dendritic Arm Spacing, and (3) the impact of P/Sr elements on microstructures. Our model delivers results that match the accuracy and robustness of traditional computational materials science methods, yet significantly reduces computation time. 
    more » « less
  5. ; ; ; ; ; (, npj Materials Degradation)
    Abstract The rapid solidification associated with additive manufacturing (AM) leads to complex microstructures with peculiar features amongst which cellular solidification structures are the most remarkable. These metastable structures possess a clear segregation pattern dictated by the solidification pathway of the alloy and are bounded by dislocation walls. While they confer exceptional strength and ductility to AM 316L stainless steel, their effect on localized corrosion in chloride environments remains to be established. Here, we employ correlative electron microscopy to reveal coupled chemical, electrochemical, and crystallographic effects on localized corrosion attack and its development. We show that the Cr and Mo-depleted interior of the cellular solidification structures dissolves selectively, giving rise to an intricate damage morphology, that is directly related to the underlying crystallographic orientation. Whereas surface observations only reveal apparently shallow micrometer-size cavities, 3D tomography via focused ion beam serial-sectioning shows a high degree of connectivity between these features underneath the surface. We reveal this intricate morphology, propose a formation mechanism, and discuss alloy design guidelines to mitigate this phenomenon. 
    more » « less
  • Citation Formats
  • MLA
  • APA
  • Chicago
  • BibTeX
  • Save / Share this Record
  • Send to Email