| About this Abstract |
| Meeting |
2020 TMS Annual Meeting & Exhibition
|
| Symposium
|
Additive Manufacturing: Materials Design and Alloy Development II
|
| Presentation Title |
A-53: Automatically Quantifying Phase Information from HRTEM for Additively Manufactured Inconel 718 |
| Author(s) |
Sen Liu, Behnam Aminahmadi, Branden Kappes, Aaron Stebner , Xiaoli Zhang |
| On-Site Speaker (Planned) |
Sen Liu |
| Abstract Scope |
Additive manufactured (AM) metals have a unique initial microstructure that requires custom heat treatments. Inconel 718 precipitation strengthening relies on the coprecipitation of γ' and γ'' phases. High resolution transmission electron microscopy (HRTEM) has been used to track precipitation growth behavior during post processing. Identifying nanoprecipitates evolution during heat treatments requires hundreds of images and thousands of precipitates. Computer vision (CV) coupled with machine learning (ML) segmentation automatically extract and quantify phase information with angstrom-scale resolution, allowing for quantitative correlation between nanostructure formation and processing conditions. We introduce a sliding Fast Fourier Transform (FFT) to automatically segment unique phases from HRTEM. The image processing used is largely insensitive to image variations. An unsupervised ML was used to automatically group similar phases, which is unique to composition and orientation of constituent phases. This study shows the promise of ML for enabling high-throughput materials characterization to accelerate AM materials development. |
| Proceedings Inclusion? |
Planned: Supplemental Proceedings volume |