First World Congress on Artificial Intelligence in Materials and Manufacturing (AIM 2022): AI-Assisted Development of New Materials/Alloys I
Program Organizers: Taylor Sparks, University of Utah; Michael Dawson-Haggerty, Kerfed, Inc.; Elizabeth Holm, University of Michigan; Jin Kocsis, Purdue University; Adam Kopper, Mercury Marine; Benji Maruyama; James Warren, National Institute of Standards and Technology
Monday 9:30 AM
April 4, 2022
Room: William Penn Ballroom
Location: Omni William Penn Hotel
Session Chair: Alex Kitt, Edison Welding Institute
9:30 AM Break
10:00 AM Invited
3rd Wave AI to Accelerate Materials Discovery: Andrew Detor1; Kareem Aggour1; Aida Amroussia1; Scott Weaver1; Abha Moitra1; Alfredo Gabaldon1; Paul Cuddihy1; Sharad Dixit1; 1Ge Research
With the widespread availability of open-source tools and commercial platforms, machine learning is changing the way new materials are developed; but a revolution in materials discovery remains elusive. We posit that “3rd wave AI” techniques allowing for the modeling of different forms of knowledge and the abstraction of concepts for reasoning and contextual adaptation will be necessary to make significant advancements in machine-assisted material discovery. We present recent progress at GE Research toward our vision to capture factual materials data in a knowledge graph augmented with analytical models and experiential knowledge codified from domain experts. By fusing these forms of knowledge together, we are enabling (i) reasoning with uncertainty for question-answering, and (ii) inferencing to propose novel materials outside the boundaries of the original dataset. A proof-of-concept is demonstrated through a case study involving the optimization of a physical vapor-deposited coating for power generation steam path applications.
10:30 AM
Discovery of New Periodic Inorganic Crystals Via GANs: Taylor Sparks1; Michael Alverson1; 1University of Utah
Moving away from materials screening to true materials discovery will require accurate generative models for both organic and periodic inorganic systems. In this work, we investigate and analyze the performance of various Generative Adversarial Network (GAN) architectures to find an innovative and effective way of generating theoretical crystal structures that are synthesizable and stable. The space group number, atomic positions, and lattice parameters are parsed from the CIFs and used to construct an input tensor for each of the different network architectures. Several different GAN layer configurations are designed and analyzed, including Wasserstein GANs with gradient penalty, in order to identify a model that can adequately recognize and discern symmetry patterns. This work will detail the process and techniques that were used in an attempt to generate never-before-seen crystal structures that are both stable and synthesizable, as well as reveal a plethora of guiding questions for future work.
10:50 AM
Finding Superhard Materials Through Machine Learning: Jakoah Brgoch1; 1University of Houston
Superhard materials with a Vickers hardness >40 GPa are essential in applications ranging from manufacturing to energy production. Finding new superhard materials has traditionally been guided by empirical design rules derived from classically known materials. However, the ability to quantitatively predict hardness remains a significant barrier in materials design. To address this challenge, we constructed an ensemble machine-learning model capable of directly predicting load-dependent hardness. The predictive power of our model was validated on eight unmeasured metal disilicides and a hold-out set of superhard materials. The trained model was then used to screen compounds in Pearson’s Crystal Data (PCD) set and combined with our recently developed machine-learning phase diagram tool to suggest previously unreported superhard compounds. Finally, industrial materials often experience tremendous heat during application; thus, we are building a method for predicting hardness at elevated temperatures.
11:10 AM
Integrating Data-Driven and Experimental Techniques for the Design and Development of New Corrosion-Resistant Coating Alloys for Lightweight Automotive Steels: Rohit Bardapurkar1; John Speer1; Sridhar Seetharaman2; 1Colorado School of Mines; 2ASU Ira A. Fulton Schools of Engineering
Data-driven design, discovery, and development (D5)TM was utilized in this study to identify novel corrosion-resistant coating alloys with low liquidus temperature (TL) for galvanizing of new lightweight automotive sheet steels. Machine-learning (ML) algorithms trained on a database containing TL data (computed via CALPHAD modeling) and experimental corrosion data (collected from the literature) were employed to predict properties of new alloy coatings. A “Materials Selection Map” was developed to visualize the current state of design space and potential future opportunities related to the key performance criteria: corrosion-current (Icorr), corrosion-potential (Ecorr), and TL. Based on computed and predicted results, ZnMgAl, ZnMgAlSn, ZnMgAlSnBi, ZnMgAlSnGa alloys were selected for experimental verification of the selected performance criteria. Differential scanning calorimetry (DSC) was used for TL validation, and potentiodynamic polarization testing was performed to study the corrosion behavior of alloys. Finally, sequential learning was utilized to optimize the ML models.