Modeling software for Materials 4.0

September 21, 2021

Society’s use of materials, with its humble beginnings in the Stone Age with natural materials, advanced through the Bronze Age and Iron Age with man-made alloys to the current Industrial Age and Modern Era. Now, in the 21st century, the functionality of society relies significantly on digital technologies in terms of integration of cyber-physical systems through digitization of knowledge, and the demand for new materials to enable and promote the digital age will continue to increase.

Penn State’s Zi-Kui Liu, the Dorothy Pate Enright Professor of Materials Science and Engineering, and his team are working to develop tools for digitization of knowledge for efficiently creating new materials and their robust manufacturing processes to fill this need. His team held virtual workshops, including the latest one this summer, which was supported by the IBM Academic Initiative, to train people on how to use these tools.

Industry 4.0 to Materials 4.0

Water and steam power mechanized production in the first industrial revolution, electricity powered mass production in the second, and computers and automation propelled the digital revolution in the third. The fourth, often referred to as Industry 4.0, fueled by data and machine learning, is building on these digital advances by bridging the physical and digital world through cyber-physical systems.

“After steam power, electricity, and computerization, the process of digitization — often referred to as Industry 4.0 — is now ushering in Materials 4.0,” Liu said.

Knowledge of materials has improved steadily over the last few hundred years and guided the advancement of manufacturing processes manifested through the changes of forms of various materials. The digitization of materials knowledge, particularly stability and functionality of phases in materials with respect to external stimuli, has made significant strides in the last 50 years and promoted the paradigm shift from empirical, serendipitous discovery of materials towards the computational design of materials, Liu said.

Using modeling software to design materials can drastically shorten the development time.

“Generation of data from experiments takes weeks and months, whereas computational tools reduce the time to hours and days and can produce results even in seconds to minutes, which can then be verified by experiments so that the full cycle of materials development and deployment can be significantly shortened,” Liu said. “The main driver of Materials 4.0 models is the development of digital databases containing the thermodynamic properties of materials, which dictate the stability and functionality of phases in materials, and the digitization of thermodynamic knowledge started by later Larry Kaufman who pioneered the field of the Calculation of Phase Diagrams (CALPHAD) method on the foundation of thermodynamics laid by Gibbs in 1870’s.”

The team created two software programs: PyCalphad and Extensible Self-optimizing Phase Equilibria Infrastructure (ESPEI) based on the research from two doctoral theses in Liu’s group. PyCalphad is a free and open-source Python computer language library for computational thermodynamics modeling that uses the CALPHAD method. ESPEI efficiently evaluates the thermodynamic model parameters within the CALPHAD method.

“The models can be continuously improved with new input data from computation and experiments in a manner that is analogous to the way humans learn from experience, capturing more and more fundamental building blocks of materials,” Liu said.

Partnering with IBM

The team hosts workshops to train participants on how to use the open-source software. With COVID-19, they switched to offer the training virtually.

“Before the pandemic, we would offer trainings in person,” Liu said. “We would have 10 to 20 people attend at a time. It took a lot of time from setting up the software to bringing the people together, and so forth. With the pandemic, we decided to offer them virtually and tested the viability in 2020.”

In 2021, a discussion occurred between the IBM Academic Initiative and the Department of Materials Science and Engineering at Penn State on potential support to provide complimentary access for faculty and students at academic institutions to use IBM’s resources for teaching, learning and noncommercial research purposes.

“Now, more than ever, IBM is pleased to partner with universities to lend IBM Cloud and Watson resources to support research, senior capstone projects and other learning opportunities such as this virtual workshop,” said Blain Dillard, operations manager at IBM’s Center for Advanced Studies. “IBM Cloud for Education and Penn State engineers worked closely together to quickly enable the platform tailored to the needs of the workshop participants.”

“Partnering with IBM made the summer workshop more economic,” Liu said. “We were able to focus on the delivery high quality content without the need to worry about the expense of the workshop since they were offered as a free service to people around the world including Asia, Europe and across the United States.”

With IBM providing the premier high performance computing services, the cloud-based workshop also was simpler for the participants.

“With digitization, you don't need to install any software anymore,” said Liu. “It's all available online – you just log in.”

IBM has also donated more than $1 million to Penn State since 2014 and is one of the University’s educational alliance partners. The partnership reached a new height in 2019, when IBM donated a cutting-edge system, known as the AC922, to Penn State, where it sits in a test environment of the University's supercomputer, ICDS-ACI, run by the Institute for Computational and Data Sciences. It’s the exact same hardware powering the world’s two biggest supercomputers, Summit and Sierra, both located in U.S. national laboratories.

Operational support for the workshop was provided by the Materials Genome Foundation, a nonprofit organization that promotes computational science and engineering, and who is also responsible for supporting infrastructure and public stewardship of the PyCalphad project.

“We were delighted to help Penn State make efficient use of IBM’s generous resource donation,” said Richard Otis, co-director of the foundation. “IBM Cloud’s enterprise networking and compute capability were essential in making this workshop event a complete success.”

(Media Contacts)

Last Updated September 21, 2021