By Lefei Chen, Junior Project Manager - Academic Relations
Materials are at the core of key solutions to challenges we are facing today, from energies and environment to health and biomedical engineering, etc. When Steve Jobs published the first iPhone using a transparent conductor to remove the keyboard, people knew all smart changes were possible. Thanks to the advancement of computational science, the long and painstaking process of new material discovery has been largely automated with an additional sense of creativity. In order to know more about this promising area, on June 22, 2021, in the second edition of nexFrontier series, we invited experts from NCCR MARVEL (Swiss National Centre of Competence in Research on Computational Design and Discovery of Novel Materials), EPFL, Fudan University and Beihang University to introduce us into this wonderland.
First, Professor Nicola Marzari, director of NCCR MARVEL gave us a funny example of how computational science can help scientists to predict the properties of materials from the First Principle without actually doing the experiments. For example, in Swiss watch industry, the quantum mechanical simulation could help to predict the colors and shapes of metal Aurum into metallics in a fast and authentic way without actually making the object.
Professor Marzari said that as the computational simulation gets increasingly faster with the development of information and communication technologies, a trillion-dollar market is prospected over the next decade. Seizing the trends of future development, Switzerland started the National Centre on Computational Design and Discovery of Novel Materials in 2014, engaging 41 experts from 12 swiss institutions with an idea that a number of principal investigators could work for 12 years around this ground-breaking area. With a generous budget around 50 million CHF, the centre could start ambitious initiatives including harvesting materials for energy collection, storage and conversion, materials for ICT, high tech and pharmaceuticals etc.
Professor Marzari then elaborated on a project in the computational exfoliation of all known inorganic materials. Since low-dimensional materials are very different from 3-dimensional materials regarding their physics and chemistry aspects, the exfoliation of inorganic materials into layered structures would greatly help to define their individual properties. Powered by a big database encompassing known inorganic mechanical materials with one million different structures, scientists could use the computer to figure out whether these materials are geometrically layered rather than 3 dimensional so that characterization using quantum mechanical simulations are possible. At the same time, a powerful calculation system can automatically calculate the various properties of around a million materials in the database without any human intervention.
Secondly, Professor Xingao Gong from Key Laboratory of Computational Physics Science of Fudan University, elaborated on how to understand solar cells, the energy harvesting materials, from the perspective of computational studies. Based on the photoelectric effect, the solar cell has become very popular in striving toward carbon neutrality. In the next 50 years, it is expected that solar cell would provide over 30% of electricity in China.
In order to get the most efficient solar cell in Photoelectric conversion, we need to find materials that have good properties in VBM and CBM to improve the process of light absorption, separation and transport of photogenerated electron-hole pairs, and collection of electric charge. In this aspect, computational science helped to design a system with multi ternary semiconductors for absorber with more elements and flexible properties. Currently, Cu2ZnSnS4 and Cu2ZnGeSe4, were figured out as the two best gap materials.
Another example is the simulation of the solar cell’s interface consisting of CdS, the buffer and CdTe, the absorber to understand the function of this combined structure. However, the lattice mismatch between the two layers is too big for real-life simulation and the supercell has too many atoms for the application of First Principle. Professor Gong and his team solved this problem by using the Neural Network Potential using computational science, and then compared it with the real VASP potential. They are delighted to find that the difference between the two is very small.
Computational science not only can help design a new structure but also can test why a widely applied structure is the better one than many others, for example, by calculating the capture rate of Cu2ZnSnS4:Snzn and CH3NH3PbLPbi, that is, the defect in solar cell, we can know that the former one has a capture rate ten to the third power higher than that of the latter one, which means it can waste a lot of efficiency of the cell. Needless to say, computational science can contribute to the solar cell, absorber materials, interface properties defect and its related properties, which could play an important role in the application of solar energy.
In the area of chemistry, Professor Corminbeuf from EPFL, showed us computational chemistry in the age of big data with a focus on the theoretical framework in the concept of catalysis. It is surprising to know that instead of watch (10%), the largest share of export in Switzerland is chemical and pharmaceutical products, accounting for 44.8% of total Swiss export, and 90% of these chemical compounds are synthesized by the catalyst. Therefore, computational approaches do contribute significantly to this field. According to Sabatier’s principle, the active catalyst will balance the energy requirements for all steps, so it is very useful to build up maps that cover the different properties of catalysts and allow scientists to navigate upon in searching for the best catalyst.
Relying on big data, different capabilities and regions of catalysts can be cast into very simple plugs - volcanic plugs which have only one descriptor that should be computable using the First Principle approach or accelerated using machine learning techniques. Then these different types of catalysts can be categorized into three types: too strong, too weak, and the ones that reach the Sabatier optimum. This smart technology has enabled interactive exploration of catalysts by screening, and 557 catalysts were identified in the small plateau region out of nearly 20000.
In the area of pharmaceuticals, Professor Wang from Beihang University presented to us how computational science could help in searching for the effective design of drug product. Usually, the drug particles we eat are wrapped in a wax capsule which can dissolve in our stomach; however, during the manufacturing process, the machine usually mixes the several substances together into a liquid form without control of the structure of the drugs. Traditional direct modelling usually requires a very sophisticated Multiphysics process and is multiscale in time and space. In this case, computational science can help to simplify the process through a surrogate model created between Local Geometry and the Global Property, the effective diffusion coefficient of the drugs, while using Minkowski functionals to present the geometric description of the porous network.
This model could also provide sensitivity test to the manufacturer to understand which parameters have the most impact on the probability density function and help them to improve their drug design. Professor Wang stressed that thanks to computational science multidisciplinary collaborations between Physics, Mathematics, and pharmaceuticals are largely enabled and facilitate the exploration and discovery of new materials.
In the Q&A session, topics such as real-life simulation, collaboration with experimentalists, and advice for young talents were discussed.
Finally, great appreciation to our speakers Prof. Nicola Marzari, Prof. Xingao Gong, Prof. Clémence Corminboeuf, and Prof. Peng (Koby) Wang, for their availability and great insights. We also would like to thank the audience who joined our second nexFrontier webinar and participated in the dynamic conversations. Looking forward to meeting you at our next events!