Researchers from the Technical College of Munich have been utilizing GCS HPC assets to develop extra environment friendly strategies for producing graphene on the industrial scale.
Graphene could also be among the many most enjoyable scientific discoveries of the final century. Whereas it’s strikingly acquainted to us — graphene is taken into account an allotrope of carbon, which means that it primarily the identical substance as graphite however in a unique atomic construction — graphene additionally opened up a brand new world of prospects for designing and constructing new applied sciences.
The fabric is two-dimensional, which means that every “sheet” of graphene is just one atom thick, however its bonds make it as sturdy as among the world’s hardest metallic alloys whereas remaining light-weight and versatile. This beneficial, distinctive mixture of properties have piqued the curiosity of scientists from a variety of fields, resulting in analysis in utilizing graphene for next-generation electronics, new coatings on industrial devices and instruments, and new biomedical applied sciences.
It’s maybe graphene’s immense potential that has consequently prompted certainly one of its greatest challenges — graphene is tough to provide in massive volumes, and demand for the fabric is regularly rising. Current analysis signifies that utilizing a liquid copper catalyst could also be a quick, environment friendly means for producing graphene, however researchers solely have a restricted understanding of molecular interactions occurring throughout these temporary, chaotic moments that result in graphene formation, which means they can’t but use the strategy to reliably produce flawless graphene sheets.
As a way to handle these challenges and assist develop strategies for faster graphene manufacturing, a crew of researchers on the Technical College of Munich (TUM) has been utilizing the JUWELS and SuperMUC-NG high-performance computing (HPC) techniques on the Jülich Supercomputing Centre (JSC) and Leibniz Supercomputing Centre (LRZ) to run high-resolution simulations of graphene formation on liquid copper.
A window into experiment
Graphene’s attraction primarily stems from the fabric’s completely uniform crystal construction, which means that producing graphene with impurities is wasted effort. For laboratory settings or circumstances the place solely a small quantity of graphene is required, researchers can place a bit of scotch tape onto a graphite crystal and “peel” away atomic layers of the graphite utilizing a way that resembles how one would use tape or one other adhesive to assist take away pet hair from clothes. Whereas this reliably produces flawless graphene layers, the method is gradual and impractical for creating graphene for large-scale functions.
Business requires strategies that would reliably produce high-quality graphene cheaper and quicker. One of many extra promising strategies being investigated includes utilizing a liquid metallic catalyst to facilitate the self-assembly of carbon atoms from molecular precursors right into a single graphene sheet rising on prime of the liquid metallic. Whereas the liquid affords the power to scale up graphene manufacturing effectively, it additionally introduces a bunch of issues, such because the excessive temperatures required to soften the everyday metals used, corresponding to copper. When designing new supplies, researchers use experiments to see how atoms work together underneath a wide range of situations. Whereas technological advances have opened up new methods for gaining perception into atomic-scale conduct even underneath excessive situations corresponding to very excessive temperatures, experimental strategies don’t at all times permit researchers to look at the ultra-fast reactions that facilitate the proper modifications to a cloth’s atomic construction (or what elements of the response could have launched impurities). That is the place laptop simulations will be of assist, nonetheless, simulating the conduct of a dynamic system corresponding to a liquid shouldn’t be with out its personal set of issues.
“The issue describing something like that is you have to apply molecular dynamics (MD) simulations to get the correct sampling,” Andersen mentioned. “Then, in fact, there’s the system dimension — you have to have a big sufficient system to precisely simulate the conduct of the liquid.” Not like experiments, molecular dynamics simulations provide researchers the power to have a look at occasions occurring on the atomic scale from a wide range of completely different angles or pause the simulation to give attention to completely different elements.
Whereas MD simulations provide researchers insights into the motion of particular person atoms and chemical reactions that would not be noticed throughout experiments, they do have their very own challenges. Chief amongst them is the compromise between accuracy and price — when counting on correct ab initio strategies to drive the MD simulations, this can be very computationally costly to get simulations which can be massive sufficient and final lengthy sufficient to precisely mannequin these reactions in a significant means.
Andersen and her colleagues used about 2,500 cores on JUWELS in durations stretching over a couple of month for the latest simulations. Regardless of the huge computational effort, the crew may nonetheless solely simulate round 1,500 atoms over picoseconds of time. Whereas these could sound like modest numbers, these simulations have been among the many largest completed of ab initio MD simulations of graphene on liquid copper. The crew makes use of these extremely correct simulations to assist develop cheaper strategies to drive the MD simulations in order that it turns into potential to simulate bigger techniques and longer timescales with out compromising the accuracy.
Strengthening hyperlinks within the chain
The crew revealed its record-breaking simulation work within the Journal of Chemical Physics, then used these simulations to match with experimental knowledge obtained of their most up-to-date paper, which appeared in ACS Nano.
Andersen indicated that current-generation supercomputers, corresponding to JUWELS and SuperMUC-NG, enabled the crew to run its simulation. Subsequent technology machines, nonetheless, would open up much more prospects, as researchers may extra quickly simulate bigger numbers or techniques over longer durations of time.
Andersen obtained her PhD in 2014, and indicated that graphene analysis has exploded throughout the identical interval. “It’s fascinating that the fabric is such a latest analysis focus — it’s nearly encapsulated in my very own scientific profession that individuals have regarded carefully at it,” she mentioned. Regardless of the necessity for extra analysis into utilizing liquid catalysts to provide graphene, Andersen indicated that the two-pronged method of utilizing each HPC and experiment could be important to additional graphene’s growth and, in flip, use in business and industrial functions. “On this analysis, there’s a nice interaction between idea and experiment, and I’ve been on either side of this analysis,” she mentioned.
Reference: “Actual-Time Multiscale Monitoring and Tailoring of Graphene Development on Liquid Copper” by Maciej Jankowski, Mehdi Saedi, Francesco La Porta, Anastasios C. Manikas, Christos Tsakonas, Juan S. Cingolani, Mie Andersen, Marc de Voogd, Gertjan J. C. van Baarle, Karsten Reuter, Costas Galiotis, Gilles Renaud, Oleg V. Konovalov and Irene M. N. Groot, 1 June 2021, ACS Nano.
Funding for JUWELS and SuperMUC-NG was supplied by the Bavarian State Ministry of Science and the Arts, the Ministry of Tradition and Analysis of the State of North Rhine-Westphalia, and the German Federal Ministry of Training and Analysis by means of the Gauss Middle for Supercomputing (GCS).