Prof Dr Ingo Steinbach at Ruhr University, Bochum, leads an international research programme devoted to the use of advanced computational techniques to predict and tailor the properties of new technological materials. He has developed an approach that combines atomistic simulation, data mining and experimental verification to predict the chemical and mechanical properties of new materials and their performance during operation.
The development of new approaches for the creation of materials with well-defined physical and chemical properties is a pressing need that underpins a global effort to address the challenges posed by the rapid evolution of technology in advanced societies. Food and energy production, housing, transportation, microelectronics, functional materials and health are just a few of the fields that are increasingly and crucially being impacted by the development of new high-performance materials. In 2011, the US Government issued the Material Genome Initiative, designed to create the policy, resources and infrastructure required to discover and deploy new advanced materials and reduce their time-to-market. The key to achieving this ambitious goal is the exploitation of predictive computational techniques to drive the development and optimisation of new materials.
Virtual production and testing of materials: a multi-scale approach
The mechanical and chemical properties of matter are determined by its microstructure. In real materials, the microstructure consists of different elements: grains and phases, which determine the multi-crystalline structure of the sample, as well as dislocations, cracks, point defects and voids. All together, these elements determine the actual properties of materials, and controlling the microstructure, both during the preparation and during the operation of a material, therefore gives potential access to property control and optimisation.
The aim of Integrated Computational Materials Engineering (ICME) is to provide a unified computational and experimental framework to track the evolution of the microstructure during the whole production and lifetime cycle of a material, and to establish a complete chain of cause and effect from processing to materials properties. In addition to incorporating a fully atomistic description of matter (including, potentially, fundamental quantum mechanical principles) and a meso-scale description of the interaction between microstructural elements, existing information concerning materials’ structures and properties can be integrated using data-mining techniques. Data mining can be used to analyse and extract information from different types of sources by combining statistical analysis tools, machine learning and database systems. The data-mining process can be fully automated or optimised by human experts, and it can be used to obtain information concerning specific properties from large datasets of experimental data on natural or artificial materials.
The final crucial component of this framework is the experimental verification of the properties predicted by virtual design, to assess their accuracy or to propose improvements of the computational models. Experimental validation is an essential and high-risk step, owing to the potential complexity of the microstructure and of the interactions between its components, and it can be used to identify weak links and formulate an effective risk management strategy to resolve or circumvent problematic cases.
The phase-field method
The core of this virtual material design framework is represented by the phase-field method, a sophisticated computational approach that exploits the capabilities of modern high-performance computers to integrate all aspects contributing to the evolution of the microstructure in time. This provides a detailed description of how the microstructural elements interact during the production of a new material and while the material, once formed, operates in a device or process. The phase-field method is based on a thermodynamic description of the pathways that lead initial non-equilibrium states to a final equilibrium situation, corresponding to a thermodynamically stable sample in well-defined conditions (pressure, temperature, etc.). Crucially, this approach can be used to describe not only the formation of materials in their stable phases, but also their degradation over time. Prof Steinbach is a pioneer and a front-runner in the development and application of the phase-field method, and he is among the authors of OpenPhase, an open source software library that enables scientists to quickly develop phase-field simulation programs to describe the microstructure evolution in a wide variety of situations, including phase and structural transformations, like grain growth and recrystallisation. Once the complete chain of microstructure evolution has been identified using the phase-field method, dislocations and cracks, which cannot be resolved at phase-field level, can also be included in the model.
Knowledge of the entire microstructure history is important, because it clarifies how the final state of a material is achieved, which has important implications for its properties. For instance, a solidification structure that coarsens during heating and an initially coarse structure in which further coarsening is limited lead to the same final grain size, but also to a distinctly different distribution of slow diffusing solutes. The experimental distinction between the two mechanisms of formation is extremely difficult, and sometimes impossible, unless the full evolution of the microstructure is known. This is a very common occurrence in materials processing, and phase-field based methodologies have therefore the potential to bring about a breakthrough in our understanding of the microscopic mechanism of materials formation.
The future: challenges and outlook
Virtual design and virtual testing based on phase-field simulations, data mining and experimental verification are emerging as promising routes toward the development of new high-performance materials for technology, security and health. Some challenges remain to be addressed, however, before these methods can achieve the levels of robustness and reliability required in industrial-scale materials production. The first challenge arises from the difference in scales of different microstructures, which can range from ten to several hundred μm for solidification microstructures to a few nanometers for precipitation macrostructures. A ‘zoom-in’ strategy can in this case be applied to bridge different scales. A second challenge is related to the enormous amount of input data needed to simulate all aspects of chemical and mechanical stability of bulk phases. Finally, the behaviour of interfaces is a further source of complication in the model. In this context, the work initiated by Prof Steinbach represents a crucial step forward in the development of next-generation approaches to material design.
What are the technological fields that will benefit the most, in the coming years, from the virtual materials design approach you propose?
The field I am speaking of is “microstructure dominated materials”, i.e. materials whose properties depend crucially on their (internal) microstructure. These microstructures are adjusted during thermomechanical treatment of the material. This relates mostly to metallic materials of all kinds, but can be transferred to ceramics, polymers etc.
In which ways do virtual materials design and verification differ from materials discovery?
“Materials discovery” relates to new crystalline structures, while “materials design” relates to the design of an alloy and a treatment process for this alloy, which will in general decompose into different crystalline structures (patrix/precipitate, multiphase materials, multicrystalline materials), the composition of which determines the material’s properties.
How general is the phase-field method you developed? Are there problematic cases that need special handling?
The phase-field method is very general, can be applied to all kinds of multi-phase materials and also to biological systems. Difficult to handle are structures with huge differences in size, e.g. grain structures in the mue regime and precipitates at the nano regime. See “zoom in” strategy in final paragraph of article. Also input parameter as interfacial energies may be critical.
Will next-generation HPC facilities and the development of artificial intelligence (AI) have any impact on your work?
Supercomputers, GPUs yes, (direct relation to system size to be investigated), AI not so much.
How far are we from seeing your approach to materials design implemented as an everyday tool in industry and academia?
We are close to application in industry. The software is an open source project, available under www.OpenPhase.de. In addition, we are in the process of establishing a company, OpenPhase Solutions, which will provide customised solutions, user support and training.
Prof Dr Steinbach’s research interests include the microstructure evolution and phase transitions, thermodynamics in and at interfaces and phase-field theory.
- Thyssenkrupp Steel Europe AG
- Benteler Steel Tube GmbH
- Prof Dr Ralf Drautz, (ICAMS)
- Prof Dr Alexander Hartmaier, (ICAMS)
Prof Dr Steinbach studied physics at Ludwig-Maximilian University Munich, Bergische Gesamthochschule, and Ruhr-University Bochum before completing his doctorate at RWTH Aachen, Germany. He was Head of the Microstructure Simulation Group from 2000 at RWTH Aachen. Since 2008 he has held a Professorship and been Director of the Interdisciplinary Centre for Advanced Materials Simulation (ICAMS), Ruhr-University Bochum, Germany.
Prof Dr Ingo Steinbach
Ruhr University Bochum
Interdisciplinary Centre for Advanced Materials Simulation (ICAMS)
T: +49 234 322 9315