September 20, 2023
News Release

PNNL Collaborates with Microsoft, Micron to Bring Computational Chemistry to the Masses

DOE-funded program brings cloud, memory innovations together to accelerate technology transfer

Artist's render of computational chemistry, showing orbs connected to each other

Computational chemistry problems are incredibly complex but the payoffs for energy research and other applications are enormous.

(Illustration by Nathan Johnson | Pacific Northwest National Laboratory)

RICHLAND, Wash.—Pacific Northwest National Laboratory is collaborating with leading technology companies Microsoft Corp. and Micron Technology to make computational chemistry—a challenging subject but one with far-reaching significance for our lives—broadly available to applied researchers and industrial users.

The project, known as TEC4 (Transferring Exascale Computational Chemistry to Cloud Computing Environment and Emerging Hardware Technologies), is part of a broad effort announced by the Department of Energy to quicken the transfer of technology from fundamental research to innovation that can be scaled into products and capabilities that support the economic health and security of the nation.

Earlier this month, DOE funded 11 projects around the country with a total of $73 million to reduce the time it takes for new findings in the laboratory to become relevant in everyday life. The projects will be conducted with an eye toward the innovation’s end application and commercialization.

Led by PNNL, TEC4 aims to bring the sophisticated tools of computational chemistry to more people, effectively delivering computational chemistry as a service, also known as CCaaS. The two-year project is slated to receive up to $8 million from the DOE, supporting approximately 30 researchers. The project aims to deliver significant advancements toward critical goals like sustainability, energy security and environmental stewardship.

“This will mark the democratization of access to high-performance computational chemistry capabilities,” said PNNL scientist Karol Kowalski, who leads the effort. “There are many people with very good ideas, with their codes already written, and we intend to give them the resources they need to put their ideas into action.”

Computational chemistry problems are incredibly complex but the payoffs are enormous. Understanding the happenings of molecules in exceptional detail helps scientists address problems related to energy—for instance, developing better fuels or new molecules to grab and store carbon. Chemistry is at the core of a process known as catalysis, which the world relies on to create fertilizers, medicines and a myriad of other materials.

While researchers draw on computing power every day to study chemical processes, there are barriers. One is a shortage of powerful yet cost- and energy-efficient computing facilities able to perform the trillions upon trillions of calculations needed to explore molecular interactions, such as electronic structures, in the detail needed.

That’s where TEC4 comes in. Instead of using a centralized uber-powerful computer, the team will use Microsoft’s Azure Quantum Elements, which features simulation workflows augmented by artificial intelligence. The project will incorporate Micron’s advanced memory technology, which is specialized to store and move tremendous amounts of data quickly. And PNNL brings decades of experience developing specialized programs and computer codes designed to force even the toughest chemical reactions to yield their secrets.

As described by Matthias Troyer, technical fellow and corporate vice president of Microsoft Azure Quantum, “Innovations in chemistry and material science are estimated to have an impact on 96% of all manufactured goods, which impact 100% of humanity. It’s why Microsoft is committed to making investments in HPC and AI to empower researchers to accelerate scientific discovery now.”

The goal of the project is to bring computational chemistry as a service to a significantly broader group of researchers and practitioners—something that has never been done before at this scale. TEC4 will put the capability of crunching immense chemistry calculations into more hands and speed scientists’ ability to move their ideas from the laboratory to the real world.

Right now that sort of power exists only in a handful of places—a bottleneck for an ever-increasing queue of scientists with good ideas waiting for an opportunity to access those facilities. Distributing the problems to collections of already-existing powerful machines equipped with the latest computing technology, accessible in the project through Azure Quantum Elements, sidesteps these bottlenecks and gives scientists access to powerful tools that many assumed they could never access.

Discoveries in the project around machine learning and AI will also benefit from Microsoft and Micron technologies. Working from the cloud makes it easier to collect the enormous amounts of data needed to take a machine learning approach. The project’s objectives include enabling ML-guided workflows and AI in the cloud, trained by high-accuracy methods for large-scale simulations.

Micron considers its high-performance memory technology to be well suited to manipulate those large datasets. PNNL will be using Micron’s expansion modules based on the Compute Express Link (CXL™) interface. These expansion modules will give researchers the flexibility to scale memory as required by computational chemistry workloads—allowing more data to be processed more quickly by supporting the dynamic allocation of additional memory on an as-needed basis.

“Memory continues to be a bottleneck in data-intensive research projects where trillions of calculations are needed,” said Balint Fleischer, senior director of Near Data Computing at Micron. “This exciting collaboration allows Micron to leverage powerful emerging CXL technology in scalable, shared memory systems to overcome these critical bottlenecks and accelerate time to insights, and ultimately innovation.”

The project is not without challenges. It’s no easy task to connect many servers and develop ways for them to “speak” to each other—to exchange large amounts of data in an orchestrated fashion. How that data is packaged, presented, and transferred continuously, and then brought together from far-flung sites to solve a complex problem will put the team to the test.

The team intends to accomplish the task while keeping its system competitive in energy and cost efficiency when compared to today’s largest systems. The time is ripe for the effort: the world’s most powerful computers are no longer the purview of only large governments. For instance, several of Microsoft’s computers are now ranked in the top 50 worldwide.

The project builds on long-standing regional collaborations across PNNL in eastern Washington; Microsoft, based in the Seattle area; and Boise-based Micron, with team members contributing from Texas, California and Massachusetts. The project also taps expertise from several other institutions, including some historically minority-serving institutions. Partners include the University of Texas at El Paso, Louisiana State University, Central Michigan University, and Lawrence Berkeley National Laboratory.

“It’s a bold project,” said Kowalski. “Distributing complex computational processes across machines in the cloud to this degree is a new kind of high-performance computing that has not been attempted before.”


About PNNL

Pacific Northwest National Laboratory draws on its distinguishing strengths in chemistry, Earth sciences, biology and data science to advance scientific knowledge and address challenges in sustainable energy and national security. Founded in 1965, PNNL is operated by Battelle for the Department of Energy’s Office of Science, which is the single largest supporter of basic research in the physical sciences in the United States. DOE’s Office of Science is working to address some of the most pressing challenges of our time. For more information, visit For more information on PNNL, visit PNNL's News Center. Follow us on Twitter, Facebook, LinkedIn and Instagram.