Late News Poster Session: Materials Design
Program Organizers: TMS Administration
Tuesday 5:30 PM
March 1, 2022
Room: Exhibit Hall C
Location: Anaheim Convention Center
M-25: AI-based Design of Daytime Radiative Cooling Materials with Record Temperature Reduction: Quang-Tuyen Le1; Sih-Wei Chang2; Huyen-Anh Phan1; An-Chen Yang3; Nan-yow Chen3; Hsueh-Li Chen2; Yu-Chieh Lo1; Dehui Wan4; 1National Yang Ming Chiao Tung University; 2National Taiwan University; 3National Center for High-Performance Computing, NARLabs, Taiwan; 4National Tsing Hua University
In this work, we propose an AI-based deep generative model combined with a one-dimensional convolutional neural network to performs the inverse design process of the extraordinary passive daytime radiative cooling materials in a probabilistic way. The AI-enabled strategy really delivers a comprehensive solution for the one-to-many mapping problem of inverse design. The prediction results are then validated by Kramers-Kronig relations and Lorentz-Drude model, and a new record-breaking PDRC material with a reduction of ~79 K compared with ambient temperature, and ~12 K lower than the temperature of the conventional ideal selective emitter is discovered. The AI-extrapolated extraordinary PDRC materials provide a new guideline for designing PDRC materials in the future and connect the gap between ideal selective emitter and real materials.
M-26: Combined Experimental and DFT Computational Study of NiTiZrCu High Temperature Shape Memory Alloy: Tapasendra Adhikary1; Bharat Charan Marupalli1; Gourab Bhattacharya1; Akash Oraon1; Banty Kumar1; Shampa Aich1; 1IIT Kharagpur
The effect of Zr and Cu additions on the phase stability, microstructure and martensitic transformation temperature of binary NiTi alloys has been investigated with a combined approach of experimental measurements and first principle density functional theory calculations. In this work, multilayer Ni/Ti/Zr/Cu films have been deposited using the magnetron sputtering on Si substrates at room temperature and subsequently annealed at 350 C for complete inter-diffusion of multilayers to create amorphous phase followed by high temperature annealing at 600 C for 5 min to achieve Ni44Ti35Zr15Cu6 shape memory alloys. The high temperature annealing is required for the formation of crystalline phases. The formation energy results indicate that the Zr and Cu additions to NiTi favor the formation of monoclinic B19' phase. The differential scanning calorimetry results show that the martensitic transformation temperature increases with Zr and Cu additions.
NOW ON-DEMAND ONLY – M-27: Machine Learning-enabled Framework for the Screening of Hydrogen Storage Materials: Amit Bundela1; Rahul R1; 1Indian Institute of Technology (Indian School of Mines) Dhanbad
Design of new materials with enhanced hydrogen storage properties is needed for future energy applications. Recently the applicability of high entropy alloys (HEA) for hydrogen storage is getting wider attention due to their unique structure and properties. Identifying the right composition from the huge compositional space of HEAs is challenging. A database was developed based on the reported hydrogen storage materials and identified a HEA system in the current study. High throughput studies were carried out to develop the data to identify the right composition in the system. The data distribution was analyzed using various data analysis methods. Machine learning algorithms were used to identify the right composition from the high throughput data by incorporating the design parameters etc. The ML prediction is validated with experimental results, and the developed framework can accelerate the identification of Hydrogen storage materials.
Cancelled
M-28: Uncertainty Quantification Using Thermo-Calc’s TC-Python Package: Giancarlo Trimarchi1; Masoomeh Ghasemi1; Qing Chen1; 1Thermo-Calc Software AB
We present the uncertainty quantification (UQ) analysis of the phase diagrams of selected binary systems calculated using the calculation of phase diagrams (CALPHAD) method as implemented in the Thermo-Calc Software package. The analysis is performed according to the Bayesian approach taking the binary systems Cu-Mg and Zn-P as case studies. Using a Markov chain Monte Carlo method we infer the posterior probability distribution of the Gibbs free energy model parameters starting from datasets that comprise a wide range of phase boundary and thermochemical data. We then show how the parameter distribution propagates to the predicted phase diagrams. The analysis is performed with a software pipeline that leverages the TC-Python interface to the Thermo-Calc thermodynamic solver to perform all the needed thermodynamic calculations.