DOE-BES: Data-driven discovery of intermetallic catalysts with controlled active site nuclearity

Sponsor:  Department of Energy, Office of Science, Basic Energy Sciences

Main researcher(s): Shun-Li Shang

This project aims to develop computational tools and a workflow to guide selective intermetallic catalyst design and apply this approach to realize advances for specific selective hydrogenation reactions. We will combine density functional theory, statistical learning, and experimental testing to define the site ensembles that selectively target a specific unsaturated functionality within complex substrates and multi-component feedstocks. We will use data mining, crystal structure enumeration, and multi-factor screening to discover intermetallics exposing the targeted active site arrangements. We will expand the available intermetallic structures and determine their relative phase stability through CALPHAD (CALculation of PHAse Diagram) methods, using acceleration techniques that allow rapid generation of binary and ternary metal phase diagrams. High throughput, machine-learning accelerated approaches will be used to examine surface energies, adsorbate binding, and CO vibrational frequencies to further facilitate multi-factor screening efforts and guide experimental work towards high-potential intermetallic compositions. Solid state synthesis, a series of bulk and surface characterization methods, and kinetic testing will realize improved selectivity and refine active site targets.

NSF-CMMI: Accurate Prediction of TCP Phase Stability for Chemistry and Process Design of Ni-based Superalloys

Sponsor:  National Science Foundation, Civil, Mechanical and Manufacturing Innovation

Main researcher(s): Adam M. Krajewski, Yi Wang, Shun-Li Shang

This project aims to establish a new paradigm for reliable and effective assessments of the thermodynamic stability of intermetallic phases, especially the topologically closed packed (TCP) phases during processing. This objective will be achieved by: 1) performing high-throughput first-principles calculations of “sublattice stabilities” and atomic interaction energetics in individual sublattices of the complex TCP phases with multiple sublattices that cannot be directly measured experimentally (these energetics play vital roles in reliable modeling of Gibbs energy functions of the TCP phases); 2) exploring innovative and systematic strategies to enable facile incorporation of first-principles results into CALPHAD assessments – a long-standing yet unsolved issue in the field; 3) making high-throughput diffusion multiples to obtain reliable phase diagrams of about a dozen critical ternary  systems (that are critical to TCP phase stability evaluation) and employing the data to optimize the Gibbs energy parameters of the phases; and 4) expanding the open ESPEI platform (infrastructure) capabilities to seamlessly use both first-principles calculation results and experimental data to perform high-throughput CALPHAD assessments, including uncertainty quantification.

ONR: Design and Additive Manufacturing of Ni-based Alloys for Naval Applications

Sponsor:  Office of Naval Research

Main researcher(s): Hui Sun, Yi Wang, Shun-Li Shang

This project aims to use computational methods to design alloys for additive manufacturing (AM) that have a combination of high strength and superior seawater resistance. Success in this project relies on the development of reliable thermodynamic and diffusion knowledge, which can be used to tailor non-equilibrium phases in Ni-based AM alloys for naval applications. To accelerate the development of advanced AM alloys, the present project proposes an Integrated Computational Materials Engineering (ICME) approach that includes high throughput first-principles calculations and high throughput CALPHAD (calculation of phase diagram) modeling for database development and alloy design together with AM processing via the powder bed fusion (PBF) method and characterization for model verification and improvement.

DOE-ARPA-E: Design and Manufacturing of Ultrahigh Temperature Refractory Alloys

Sponsor:  DOE-ARPA-E

Main researcher(s): Shun-Li Shang, Yi Wang, and PhD students from several groups

The proposed project aims to develop an integrated computational and experimental framework for the design and manufacturing of ULtrahigh TEmperature Refractory Alloys (ULTERA). Penn State will generate alloy property data through high-throughput computational and machine learning models; design ultrahigh temperature refractory alloys through a neural network inverse design approach (where one first articulates the needed functionality/property, and then looks for the materials that haves the property); manufacture the designed alloys utilizing field assisted sintering technology and/or additive manufacturing; and demonstrate the performance through systematic characterization in collaboration with industry. The proposed platform with a sustainable data ecosystem could create fundamentally new approaches to understand and design a new generation of materials and provide pathways to improve existing materials to meet performance requirements.

DOE-NEUP: High Throughput Computational Platform for Predictive Modeling of Thermochemical and Physical Properties of Fluoride Molten Salts

Sponsor: DOE-NEUP

Main researcher(s): Jorge Paz Soldan, Yi Wang, Shun-Li Shang

The proposed project aims to (i) Understand critical salt characteristics of fluoride molten salts such as atomic structure, melting points, heat capacity, free energy of potential corrosion reactions, solubility of fission and corrosion products in a model system of F-(K-Li-Na)-(Cr-Ni)-(Pu-U) by means of the density functional theory (DFT) based molecular dynamics (MD) simulations and advanced experiments; (ii) Predict and optimize critical salt characteristics using thermodynamic database created by high throughput (H-T) thermodynamic modeling with validation and refinement by advanced experiments; (iii) Develop an open-source H-T computational platform with an automated process from DFT-based first-principles calculations, DFT-based MD simulations, machine learning (ML) predictions, to CALPHAD (CALculations of PHAse Diagrams) modeling of thermodynamic properties; and (iv) Provide a publicly accessible H-T computational platform with Python-based open-source codes and thermodynamic models that has a capability to integrate both experimental and computational data for continued advancement in understanding of complex molten salt systems.

NASA Fellowship: Computational design of additively manufactured functionally graded materials for structural applications

Sponsor: National Aeronautics and Space Administration (NASA)

Main researcher(s): Brandon Bocklund

Additive manufacturing (AM) has an enormous opportunity to become a leading technique for manufacturing high-value, mechanically optimized parts. Functionally graded materials (FGMs) prepared by additive manufacturing enable the use of composition as a spatial design parameter that can be used to control the stable phases in a part as well as transition between dissimilar materials to tailor the properties of materials as a function of position. However, a major challenge in the development of FGMs is the formation of brittle intermetallic and TCP phases and phase transformations that cause strain and cracking during the build process.

Due to the number of degrees of freedom in the composition path between two dissimilar multicomponent alloys, it is not feasible to design viable FGM gradient paths that avoid forming detrimental phases using experiments alone. A thermodynamic database that covers the entire composition range in the Cr-Fe-Ni-Ti-V can be used to develop FGMs between common materials used in AM such as commercially pure Ti, Ti-6Al-4V, stainless steels, Invar, and Inconel alloys. The database can be used to design multicomponent gradient paths between these alloys including any interlayers to facilitate the transition.

Large, multicomponent thermodynamic databases across different materials systems have not traditionally been explored using CALPHAD modeling due to the challenges with evaluating the energies of experimentally inaccessible phases and with the maintenance and improvement of large databases. First-principles calculations based on density functional theory can be used to predict the thermodynamic properties of stable and metastable phases including phases with large amounts of non-stoichiometry that are present in structural materials, such as the sigma, mu, or Laves phase. The open-source Python software ESPEI has been developed and used to automate the modeling of binary thermodynamic systems and will be extended to generate CALPHAD parameterizations for higher order systems in this work, which will allow for the rapid optimization and reassessments of large, multicomponent thermodynamic databases. The Cr-Fe -Ni-Ti-V database that will be created in this work will be able to be used for designing FGMs and other advanced structural materials relevant to NASA. The further development of ESPEI will allow this database to be maintained and continually improved so that it can be used in simulations requiring thermodynamic properties or phase relations, such as phase field modeling.