PSU Logo
NSF-ITR: Computational Tools for Multicomponent Materials Design
General Info  |  People  |  Research and Reports  |  Links  |  Internal

Project Overview

The development of new materials and the capability of tailoring existing materials to meet new and demanding applications is critical for continued improvements in the quality of human life.  Traditionally, the field of materials science and engineering predominantly focus on the processing of materials, establishing structure-property relations, and measuring materials properties.  This empirical approach is increasingly shifting towards the design of materials to achieve optimal functionality, driven largely by advances in information technology and computational materials science.

In the present information technology research proposal, we propose an interdisciplinary effort, including two materials scientists (Liu, Chen), a computer scientist (Raghavan), a mathematician (Du) and two physicists (Langer, Wolverton) from industry, academics, and national laboratory, to computationally design materials through multicomponent, multi-scale modeling including the following (See Figure 1):
  1. Databases of fundamental materials properties from the first-principles atomistic calculations to the CALPHAD modeling of multi-component materials systems through efficient information exchange;

  2. A phase-field model for materials microstructure simulation in complex multicomponent systemts through advanced algorithms and parallel computing schemes;

  3. Object-oriented programming for finite element analysis of mechanical response of simulated microstructures;

  4. An asynchronous, distributed software architecture (and its implementation) for end-to-end materials design suitable for wide area distributed grid-computing.

image003.png The rationale for proposing such a multiscale computational approach is as follows.  Recent advances in first-principles calculations have made it possible to predict accurate thermodynamic properties, such as formation energies and enthalpies, of unary and binary alloys using only the atomic numbers as the input.  However, it is not computationally tractable, for the foreseeable future, to use first-principles calculations to accurately determine the configurational and vibrational entropy contribution to the total free energy directly for a multicomponent system.  On the other hand, semi-empirical methods based on the CALPHAD approach have been very successful in predicting the phase equilibria of multicomponent commercial alloys, often with ten components or more .  The CALPHAD approach builds thermodynamic databases for multicomponent systems using data obtained in unary, binary and ternary systems.  Therefore, a marriage of first-principles calculations of simple low-order systems and the CALPHAD approach will allow one to develop thermodynamic databases for multicomponent systems from first-principles.  A similar strategy can be adopted for developing kinetic databases and databases for lattice parameters, elastic constants and interfacial energies as a function of composition and temperature.

A recent important development in computational materials science is the emergence of the powerful phase-field approach to modeling phase transformations and microstructure evolution (see brief reviews [1, 2] ).  It has a number of advantages over other microstructure models, which turn out to be critical to microstructure modeling in complex multicomponent alloy systems.  First of all, the phase-field approach does not explicitly track the positions of interfaces, and hence the temporal evolution of any arbitrary microstructures can be predicted without any a priori assumptions about their evolution path.  Secondly, the phase-field approach is based on the fundamental thermodynamic, kinetic, and crystallographic information, it can be directly used to predict the microstructure evolution in complex systems by coupling a phase-field model with reliable thermodynamic and kinetic databases, thus contributing to materials design.  Finally, there is no technical difficulty in extending from two-dimensional (2D) to three-dimensional (3D) simulations except a significant increase in computational time and memory requirements.  Therefore, more efficient algorithms and high performance computer clusters are to be developed to provide much more sophisticated computational tools. The ultimate goal of scientists and engineers is to predict the mechanical behavior of materials as a function of their chemistry, service temperature and time.  Therefore, it is critical to link predicted microstructures to the properties of materials.  Therefore, one of the goals of the proposed research is to collaborate with NIST to further extend the object-oriented finite element analysis of material properties based on the OOF program developed at NIST, and integrate it with the simulated microstructure.  OOF is designed to calculate macroscopic properties from images of microstructures and perform virtual experiments on realistic microstructures.  It was named one of the top 25 Technologies of the Year by Industry Week magazine in December 1999.  Its true innovation is how the meshes of microstructure are generated.  The primary difficulties in extending the current 2D simulations to 3D simulations will involve image based mesh generation and the interfaces between the computer program and users.

In addition to the development of new parallel algorithms, a significant aspect of our project is the design and development of software for end-to-end materials design that can be deployed on a “computational grid”.  A computational grid is an infrastructure that enables the integrated, collaborative use of a geographically distributed ensemble of powerful clusters of workstations, raid disk arrays, and other instruments on a high-speed network managed by multiple organizations (see the TeraGrid project).  We expect that our prototype grid-enabled software will serve as useful IT infrastructure for the broader community of scientists engaged in materials design.