The Aurora Supercomputer will arrive at the Argonne Leadership Computing Facility (ALCF) in 2021. Credit: Argonne National Laboratory
Exascale is coming. At Argonne National Laboratory, the Aurora supercomputer will lead the charge as the nation’s first system exceeding an exaflop, or a billion-billion calculations per second. Upon Aurora’s arrival at the Argonne Leadership Computing Facility (ALCF) in 2021, the scientific community will gain a level of computing performance and processing power not before possible. David Bross, an assistant computational chemist in Argonne’s Chemical Sciences and Engineering Division, is ready to embrace Aurora’s stunning prowess to enable cutting-edge science.
Bross currently leads a project supported by the Aurora Early Science Program (ESP), an ALCF program designed to prepare critical applications and tools for the architecture and scale of the new system before deployment. His team will use pre-production time on the exascale system to develop software tools that promise to enable revolutionary advances in heterogeneous catalysis for research areas, such as chemical transformations of bio-derived feedstock and heavy fossil fuels, and the conversion of small molecules, such as CO, CO2, or methane, into larger and more valuable compounds.
This Aurora Early Science Program project will use data science techniques in combination with quantum chemistry simulations to explore the otherwise intractable phase space resulting from gas phase molecules on catalyst surfaces to find relevant configurations and the lowest transition states between them. The team’s results will be used to produce the thermodynamic and kinetic parameters necessary to fully simulate the complex chemistry occurring in these heterogenous environments. (Image courtesy of Eric Hermes, Sandia National Laboratories)
This Aurora Early Science Program project will use data science techniques in combination with quantum chemistry simulations to explore the otherwise intractable phase space resulting from gas phase molecules on catalyst surfaces to find relevant configurations and the lowest transition states between them. The team’s results will be used to produce the thermodynamic and kinetic parameters necessary to fully simulate the complex chemistry occurring in these heterogenous environments. (Image courtesy of Eric Hermes, Sandia National Laboratories)
The Aurora ESP effort is part of a larger project called Exascale Catalytic Chemistry (ECC) funded by the U.S. Department of Energy’s Basic Energy Sciences, Computational Chemical Sciences Program. The ECC team is led by Judit Zádor at Sandia National Laboratories and includes partners across the United States at Pacific Northwest National Laboratory, Brown University, Northeastern University, and, of course, at Argonne. ECC’s goal is to develop a comprehensive and open-source computational framework, including publicly accessible databases such as Argonne’s Active Thermochemical Tables, that can explore chemical reaction pathways in heterogeneous catalysis. Keeping their work open-source makes their data and software solutions available to the broader scientific community, which can then put these to use for their own research endeavors using the Aurora system.
Exploring Chemical Reactions
Catalysts are at the heart of virtually every chemical process carried out in industry. Catalysts can change the rate of a chemical reaction by modifying subtle details of the reaction path. Well-designed catalysts speed up the chemical reactions that produce the desired products while slowing down undesired processes. Catalysts come in many forms and phases. They can exist as nanoparticles, a solution phase in a solvent, a metal like platinum, or as complex proteins called enzymes.
The ECC team is developing a computational framework that accelerates the characterization and optimization of metal and metal-oxide catalysts for the transformation of gas-phase molecules. The framework is composed of three related research efforts which involve developing automated methods of exploring catalytic reaction pathways and assembling them into complete reaction mechanisms that others can use in the future to optimize catalysts and operating conditions. Bross describes the three-stage process. “Using Aurora, our team’s first step is to characterize reaction pathways for individual reactions by performing quantum chemistry simulations. After that, we will seek to understand the overall reaction mechanism for a given process that consists of many such reactions. In this process, we are creating related thermochemical and kinetic information that others can use. Finally, we will analyze the mechanisms to find the parameters that would allow the improvement of a given catalyst. We have a lot of work ahead, but it’s an inspiring project.”
Bross’s expertise in databases makes his skills uniquely valuable for the effort. “As we speak, chemical transformation technologies and catalysts drive many industries,” he says, reinforcing the importance of catalyzed reactions. As one example of a catalyst-driven chemical reaction, Bross cites the Haber-Bosch process. The process was developed by two German scientists, Carl Bosch and Fritz Haber, in 1910. “The Haber-Bosch process generates ammonia from nitrogen, which is critical for manufacturing fertilizer. If that process didn’t exist, we could not support the seven-and-a-half billion people on the earth today. Simply put, we would not have enough food,” he states.
Catalyzing Innovation
As one practical goal of their endeavor, the computational tools developed in the project will enable the team and others in the catalysis community to develop new catalysts which can improve existing industrial processes and thus reduce energy usage and waste. Therefore, the Early Science Program team needs to establish mechanisms to understand the nuances of existing catalysts and find ways to design new, even more effective catalysts. Notes Bross, “To develop new industrially valuable catalysts, we need to develop a molecular-level understanding of how they work. Once we have a complete description of the thermochemistry and reaction kinetics underlying nall those mechanisms, we can use that knowledge to tailor and engineer catalysts which can advance many technologies that we use.”
He emphasizes that the project’s goal is developing tools that are open source and have broad applicability within gas/solid heterogeneous catalysis. There are many processes and related catalysts to be optimized. Rather than focus on one particular problem, the project seeks to create a capability to solve many possible questions in this field using exascale computers. As an example, Bross says that one potential application of this effort involves processing of fossil fuels.
Exascale Computing Powering Chemistry
“Larger molecules, like those we wish to explore in our work with catalysts, create incredible amounts of complexity for simulations. Each additional atom in these molecules adds three additional independent variables to the amount of space to be explored, and catalysts have various facets with different catalytic sites and activities. Gathering information requires an enormous amount of computing power to attack the ‘scaling’ problem.” Continuing, Bross adds, “The big advantage of exascale computing is that we’ll be able to explore more chemical reaction pathways than ever possible before. In doing so, we can develop ‘global’ mechanisms rather than focusing on a single reaction step. With that, we gain more profound insights and a complete mechanistic understanding of identifying new, and even improved, catalysts.”
For complex endeavors like this, the team must find ways to condense the scope of the challenge to be accommodated by the most powerful computer systems. Even a system as powerful as Aurora might choke on such an intimidating workload. The process requires a powerful combination of efficient algorithms to accomplish the feat.
Achieving Aurora’s performance level requires the latest advanced hardware under the hood. Built by Cray, the system will incorporate a future generation of Intel Xeon Scalable processors, next-generation Intel Optane DC Persistent Memory, and future Intel Xe technologies. Bross reinforces the importance of the group effort required to realize a vast project like the development of Aurora: “We feel very fortunate to have Intel and Cray supporting Aurora’s success. It takes a village to enable a system of Aurora’s size and scope. Without the combination of technologies and human expertise and working in tandem, a system like Aurora could not exist.”
Elaborates Bross, “We are also extremely thankful for the Department of Energy’s support, and their grant to fund the research we are doing in ECC. That funding allows us to develop the needed software and tap Aurora to complete our simulations.”
Empowering New Science
The major part of the overall effort involves software development. Bross and the larger team are working to create the necessary code to assist in the effort. While the software has immediate applicability for the catalyst identification process, it can also apply to other types of research in the future.
“Exascale will allow us to solve yesterday’s insurmountable scientific problems. Until now, we have been forced to navigate tradeoffs between the fidelity of a simulation, the time necessary to derive it, and the number of simulations we can perform with a finite number of cycles. With a greatly increased computational resource like Aurora, we can perform vastly more high-fidelity simulations. Therefore, we can get much better quantitative descriptions to fuel our work.”
“I’m extremely excited to be a part of the Early Science team and have an opportunity to apply Aurora’s power to our joint effort.” With a laugh, Bross adds, “It’s also thrilling that Aurora is only a ten-minute walk from my office!”
Rob Johnson spent much of his professional career consulting for a Fortune 25 technology company. Currently, Rob owns Fine Tuning, LLC, a strategic marketing and communications consulting company based in Portland, Oregon. As a technology, audio, and gadget enthusiast his entire life, Rob also writes for TONEAudio Magazine, reviewing high-end home audio equipment.
This article was produced as part of Intel’s HPC editorial program, with the goal of highlighting cutting-edge science, research and innovation driven by the HPC community through advanced technology. The publisher of the content has final editing rights and determines what articles are published