In This Issue
The Bridge: 50th Anniversary Issue
January 7, 2021 Volume 50 Issue S
This special issue celebrates the 50th year of publication of the NAE’s flagship quarterly with 50 essays looking forward to the next 50 years of innovation in engineering. How will engineering contribute in areas as diverse as space travel, fashion, lasers, solar energy, peace, vaccine development, and equity? The diverse authors and topics give readers much to think about! We are posting selected articles each week to give readers time to savor the array of thoughtful and thought-provoking essays in this very special issue. Check the website every Monday!

Predicted Advances in the Design of New Materials

Tuesday, December 22, 2020

Author: Susan B. Sinnott and Zi-Kui Liu

The prehistory and protohistory of humanity are divided into three ages in terms of materials: the Stone Age (~3.4 million years, until 8700–2000 BC), based on raw materials from nature; the Bronze Age (3500–300 BC), based on human-made copper (alloyed with 12 wt% tin); and the Iron Age (1200 BC–800 AD), derived from human-made iron-carbon alloys.

In the 21st century the functionality of society relies on digital technology built on silicon-based electronics. Digitization through the integration of cyberphysical systems with many autonomous subsystems will demand increasingly more efficient development of materials with emergent performance under extreme conditions, such as those required for the human ­colonization of other planets (Lambert 2018).

While knowledge of materials among engineers has improved steadily over the last few hundred years and especially since the start of the Industrial Revolution, most materials development has occurred through the intuition of experts, trial and error, or serendipitous discovery. The logical next step is the computational design of materials, first systemized in 1997 (Olson 1997) and given a big boost in 2011 with the launch of the Materials Genome Initiative by the US government (NSTC 2011).

The key enabler of this approach is the digitization of knowledge on materials stability in terms of thermodynamic  information (Gibbs 1873) stored in digital databases developed by the calculation of phase diagram (CALPHAD) method (Kaufman and Bernstein 1970), capabilities separated from each other by about 100 years. Fifty years have passed since the creation of CALPHAD, and we imagine here developments that will take place over the next 50 years.

Materials 4.0

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 2020). Digitization of materials knowledge progressed significantly in the 20th century, from the Schrödinger (1926) equation in quantum mechanics to its solutions based on the density functional theory (DFT; Kohn and Sham 1965), resulting in massive digital databases of materials properties predicted using high-performance com­puters. Known weaknesses in the DFT, such as consistent underestimation of band gaps in semiconductors, were addressed through theoretical improvements that were implemented in a computationally efficient manner.

Deep neural network
machine learning models
can be continuously
improved with new input
data in a manner analogous
to the way humans learn
from experience.

Data, empirical models, and mechanistic correlations (Cordero et al. 2016) are now leading to an era where artificial intelligence (AI) will be used to (i) interpret the knowledge that connects the data through machine learning (ML) algorithms and (ii) develop deep neural networks (DNNs) to predict new data and knowledge (Gubernatis and Lookman 2018).

Generation of data from experiments takes weeks and months, whereas DFT-based calculations reduce the time to hours and days, and DNN ML models can produce results in seconds to minutes. The models can also 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 2014).

The expected technical advances of this current ­trajectory include the design of materials with emergent properties (Liu et al. 2019), fulfilling the decades-long goal of “materials by design” (Gillespie 2019). In addition, the development of new experimental methods, such as the cold-sintering approach for producing complex metal oxides at very low temperatures (Guo et al. 2019), will further accelerate new material discovery and manufacturing.

Impacts and Applications

The transformative development will be the full integration of DNN methods into experimental and computational approaches used in materials synthesis and structure-property relationship determination. The integration of computational methods such as DFT, CALPHAD, and DNN in materials synthesis will continue to evolve to the point that human involvement will be greatly reduced. For example, optimizing the microstructure of materials may be achieved by rapidly analyzing many microstructures in multiple samples using a combination of electron microscopy with image recognition algorithms.

The biggest impact of these developments will be the speed with which new materials may be available for specific applications. More compositions and microstructures may emerge very rapidly, including pervasive applications of today’s nanotechnology, future quantum-scale manipulations, polymer materials that exist in nonequilibrium states across multiple scales (de Pablo et al. 2019), and metallic alloys optimized for new space applications (Lambert 2018).

It is further expected that experimental characterization, artificial intelligence, and ML will be seamlessly integrated with each other such that the line between computational methods and experimental characterization disappears.

Computational materials design will encompass the recycling of materials as the physical ecosystem interfaces with the data/cyber ecosystem throughout the materials lifecycle (Liu 2018). Initially, it is anticipated that this physical/cyber integration will result in efficient DNN ML models so that each step in a complex manufacturing process can be optimized according to the prior steps, starting from the inevitable fluctuations in the raw materials properties. This AI-guided interactive manufacturing system will be able to self-balance every subsequent step to ensure that materials remain on optimal pathways to final products with desired microstructures and properties, thus leading to zero-scrap manufacturing.

Ultimately, when this integrated system is fully implemented, the residuals from the design, manufacturing, service, and recycling of materials can be drastically reduced, thus lessening the impact of materials use on the environment.

References

Cordero ZC, Knight BE, Schuh CA. 2016. Six decades of the Hall-Petch effect: A survey of grain-size strengthening studies on pure metals. International Materials Reviews 61:495–512.

de Pablo JJ, Jackson NE, Webb MA, Chen L-Q, Moore JE, Morgan D, Jacobs R, Pollock T, Schlom DG, Toberer ES, and 14 others. 2019. New frontiers for the Materials Genome Initiative. npj Computational Materials 5(41).

Gibbs JW. 1873. Graphical methods in the thermo­dynamics of fluids. Transactions of the Connecticut Academy II:309–42.

Gillespie A. 2019. Materials by design. Featured story, May 22. Gaithersburg MD: National Institute of Standards and Technology.

Gubernatis JE, Lookman T. 2018. Machine learning in materials design and discovery: Examples from the present and suggestions for the future. Physical Review Materials 2:120301.

Guo J, Floyd R, Lowum S, Maria J-P, Herisson De Beauvoir T, Seo J-H, Randall CA. 2019. Cold sintering: Progress, challenges, and future opportunities. Annual Review of ­Materials Research 49:275–95.

Kaufman L, Bernstein H. 1970. Computer Calculation of Phase Diagrams with Special Reference to Refractory ­Metals. Cambridge MA: Academic Press.

Kohn W, Sham LJ. 1965. Self-consistent equations including exchange and correlation effects. Physical Review 140:A1133–38.

Lambert F. 2018. Tesla and SpaceX are partnering up to create new materials to use on Earth and in space. Electrek, May 10.

Liu ZK. 2014. Perspective on Materials Genome®. Chinese Science Bulletin 59:1619–23.

Liu ZK. 2018. Ocean of data: Integrating first-principles calculations and CALPHAD modeling with machine learning. Journal of Phase Equilibria and Diffusion 39:635–49.

Liu ZK. 2020. Materials 4.0 and the Materials Genome Initiative. Advanced Materials & Processes 178(2/3):50.

Liu ZK, Li B, Lin H. 2019. Multiscale entropy and its implications to critical phenomena, emergent behaviors, and information. Journal of Phase Equilibria and Diffusion 40:508–21.

NSTC [National Science and Technology Council]. 2011. Materials Genome Initiative for Global Competitiveness. Washington.

Olson GB. 1997. Computational design of hierarchically structured materials. Science 277:1237–42.

Schrödinger E. 1926. Quantisierung als eigenwertproblem. Annalen der Physik 384:361–76.

About the Author:Susan Sinnott is a professor and department head and Zi-Kui Liu is the Dorothy Pate Enright Professor, both in the Department of Materials Science and Engineering and Materials Research Institute at the Pennsylvania State University.