The Electrochemical Society, a prestigious nonprofit professional organization, has led the world in electrochemistry, solid state science and technology, and allied subjects since 1902, providing a rigorous and high-quality home for the whole community. The Society is dedicated to moving science forward by empowering researchers globally to leave their mark.
eing electrochemists, we often think we know everything about chemistry, but dealing with a pool brings challenges never addressed in P-Chem.
When we moved into our current home 16 years ago, our kids were thrilled that it had an inground pool. The move was in July of that year, so they were able to enjoy it for almost two months, given the warm Septembers in central Virginia. Note the pronoun, “they.” I never had a pool growing up, so the care and management needed was all new to me. In my usual overconfidence, I thought, “How hard can it be? It’s just chemistry, and I know that, kind of.” For a moment I thought I heard the gods laughing at that, but figured it was just the crows who nest nearby. Lesson 1: Trust your first instincts. Since that first day, my battles with the pool chemistry have been such that Heather refers to the pool as my “white whale,” akin to Ahab’s obsessive pursuit of Moby Dick. From April through September, I battle the elements of rain (which brings all kinds of biology into the pool), fire (the sun burns off the chlorine), and earth (the leaves and dirt that find their way into the pool).
Keeping the alkalinity, pH, hardness, free chlorine, and stabilizer in the appropriate ranges is harder than one might think. And if you don’t, watch out—a green pool appears, requiring one to set aside a month’s salary to pay for the chlorine to send the algae to their watery grave. Which you then need to vacuum out. Lesson 2: I only recently learned the genius of opening the pool early (i.e., in mid-April). Although the water temperature is a bracing 60 °F (15 °C), the clean-up of the tea made over winter from the leaves that snuck under the cover is so much easier than the black lagoon that is found if one opens in May. Lesson 3: Once stabilized, constant monitoring is key, and very explicit instructions must be left if you plan to travel. Lesson 4: Coagulants, under the pseudonym of clarifiers, can be magical in dealing with cloudy pool water, as can chelants in dealing with rust stains.
Biology is another player in the drama that is a pool. There is tree pollen season where the water’s surface has that dull reflectivity caused by zillions of pollen grains waiting to clog my sand filter. This season is followed by spider season, where each emptying of the strainer brings with it a confrontation with one or more generally very upset, very large spiders. Okay, they aren’t all large, but they are still scary. Nothing gets the heart beating like being surprised early in the morning by a wolf spider emerging from the pile of leaves in the skimmer, but that wasn’t me screaming. Overlapping with that season is the arrival of the toads who like getting into the pool, apparently at the end of a hard day at the DOGE office, not realizing that they are floating into the skimmer where they can’t get out by themselves. I should have a statue in Toad City for the number of lives I have saved, but they never even thank me. Lesson 5: An upset toad expresses its emotions by urinating on its hero. Classy
Lesson 6: Squirrels are the devil incarnate, but they do not like hot sauce. Once the pool season mercifully ends, squirrels like to chew holes in my safety cover to get nuts that rain down from the surrounding trees. So, I now spray the areas where nuts collect with hot sauce. They may take a nibble, but that is it. I do the same with our bird food and get perverse glee watching the squirrels’ reaction to spicy sunflower seeds. Fun fact: Birds don’t have taste buds for spice, but squirrels sure do.
I am sure that the pool has new chemistry lessons it is waiting to teach me, though, as often happens in life, it gives the test first and the lesson later. I’d love to tell more tales, but my white whale summons me. Where are those test strips?
Until next time, be safe and happy.
Rob Kelly Editor-in-Chief
Published by:
The Electrochemical Society (ECS) 65 South Main Street Pennington, NJ 08534-2839, USA Tel 609.737.1902, Fax 609.737.2743 www.electrochem.org
Editor-in-Chief: Rob Kelly
Guest Editors: Kang Xu, Y. Shirley Meng, Venkatasubramanian Viswanathan, William C. Chueh, and Qichao Hu
Contributing Editors: Christopher L. Alexander, Christopher G. Arges, Scott Cushing, Ahmet Kusoglu, Donald Pile, Alice Suroviec
Senior Director of Publications: Adrian Plummer
Senior Director of Engagement: Shannon Reed
Production Editor: Kara McArthur
Graphic Design & Print Production Manager: Dinia Agrawala
Staff Contributors: Frances Chaves, Francesca Di Palo, Genevieve Goldy, Maggie Hohenadel, Mary Hojlo, Christopher J. Jannuzzi, John Lewis, Anna Olsen, Fern A. Oram, Jennifer Ortiz, JaneAnn Wormann
Advisory Board: Jie Xiao (Battery Division)
Eiji Tada (Corrosion Division)
Vaddiraju Sreeram (Dielectric Science and Technology Division)
Luca Magagnin (Electrodeposition Division)
Chakrapani Vidhya (Electronics and Photonics Division)
Minhua Shao (Energy Technology Division) Cortney Kreller (High-Temperature Energy, Materials, & Processes Division)
Paul Kenis (Industrial Electrochemistry and Electrochemical Engineering Division) Eugeniusz Zych (Luminescence and Display Materials Division)
Jeff Blackburn (Nanocarbons Division)
Ariel Furst (Organic and Biological Electrochemistry Division)
Anne Co (Physical and Analytical Electrochemistry Division)
Praveen Sekhar (Sensor Division)
Publications Subcommittee Chair: Robert Savinell
Society Officers: James (Jim) Fenton, President; Francis D'Souza, Vice President; Robert Savinell, 2nd Vice President; Marca Doeff, 3rd Vice President; Gessie Brisard, Secretary; Elizabeth J. Podlaha-Murphy, Treasurer; Christopher J. Jannuzzi, Executive Director & CEO
Statements and opinions given in The Electrochemical Society Interface are those of the contributors, and ECS assumes no responsibility for them.
Authorization to photocopy any article for internal or personal use beyond the fair use provisions of the Copyright Act of 1976 is granted by The Electrochemical Society to libraries and other users registered with the Copyright Clearance Center (CCC). Copying for other than internal or personal use without express permission of ECS is prohibited. The CCC Code for The Electrochemical Society Interface is 1064-8208/92.
ISSN : Print: 1064-8208 Online: 1944-8783
The Electrochemical Society Interface is published quarterly by The Electrochemical Society (ECS), at 65 South Main Street, Pennington, NJ 08534-2839, USA. Subscription to members is part of membership service. © Copyright 2025 by The Electrochemical Society. *“Save as otherwise expressly stated.”
https://orcid.org/0000-0002-7354-0978
The Electrochemical Society is an educational, nonprofit 501(c)(3) organization with more than 9,000 scientists and engineers in over 75 countries worldwide who hold individual membership. Founded in 1902, the Society has a long tradition in advancing the theory and practice of electrochemical and solid state science by dissemination of information through its publications and international meetings.
Vol. 34, No. 2 Summer 2025
Editors’ Note: The Rise of AI and its Role in Revolutionizing Battery Technology
by Kang Xu, Y. Shirley Meng, Venkatasubramanian Viswanathan, William C. Chueh, and Qichao Hu
Searching for Ideal Electrolytes in the Molecular Universe
by Daniel Hannah, Yumin Zhang, Xinyu Li, Dengpan Dong, Joah Han, Gyuleen Park, Hong Gan, Bin Liu, Kai Liu, Qichao Hu, and Kang Xu
Artificial Intelligence for Electrolyte Design: Going Beyond the Molecular Paradigm
by Austin D. Sendek and Venkatasubramanian Viswanathan
Agentic Assistant for Materials Scientists
by Ruozhu Feng, Yangang Liang, Tianzhixi Yin, Peiyuan Gao, and Wei Wang
Evaluating and Interpreting the Predictive Power of Features in Battery Lifetime Prediction
by Xiao Cui, Shijing Sun, and William C. Chueh
Time-Series Approaches to Battery Cycling Data: Traditional and Emerging AI Methods by Weijie Mai and Shaoxiong Hu 33 3 5 39 4 5 49 55
3 From the Editor: Lessons in Chemistry
7 From the President: United through Science & Technology
10 Meet the New Society Officers
12 Society News 14 2025 Summer Fellowships 22 Podcasts of Note
People News
Reports from the Frontier
Tech Highlights 59 Section News
Awards Program 62 New Members 64 Student News
This month's cover is based on a figure from an article in
colleagues.
Cover design: Dinia Agrawala
Instituto Tecnológico y de Estudios Superiores de Occidente (ITESO)
y fellow members of the ECS community, what an enormous honor and privilege it is to be writing to you for the first time as president of The Electrochemical Society. I have been an ECS member for the last 42 years and counting, starting as a student in 1982, when I attended my first meeting in Detroit, MI.
From then on, ECS has been my professional home, providing guidance, inspiration, and direct access to the world’s finest minds in the fields of electrochemistry and solid state science. Similarly, my engagement in ECS has also provided me with the opportunity to serve. As someone wiser than I said, “The first and most important choice a leader makes is the choice to serve, without which one’s capacity to lead is severely limited.”* In that sense, I feel that my fourplus decades of engagement with ECS have prepared me well for the road ahead.
I served as Secretary (2017–2021), Vice President (2022–2024), officer of the ECS Boston Section (now the ECS New England Section), and officer of the ECS Industrial Electrolysis and Electrochemical Engineering Division. I have participated in countless ECS committees, including the Executive Committee, Board of Directors, Council of Local Sections, Education, Ethical Standards, Finance, and Individual Membership, and subcommittees including Awards, Interdisciplinary Science and Technology, New Technology, Publications, Technical Affairs, and Ways and Means. That is to say, my engagement with ECS has led me to where I am today, and I could not be more thrilled to serve as president of this outstanding organization.
I look forward to continuing to serve our member-led, member-driven society of world-class researchers from industry, academia, and government. And what a time to be part of ECS! Never before has science had the opportunity to positively affect the world on such a scale. And never before has it been so important that ECS advocate for science’s vital role in guiding decision-making, championing evidence-based policies, addressing misinformation, and tackling societal challenges like climate change and public health!
The Society is strongly positioned to serve its membership in these areas. We bounced back from the COVID-19 pandemic stronger than ever, with increased meeting attendance and technical content readership. Membership rose to well over 9,000 in 2024, the highest ever, and we are launching new
* Robert K. Greenleaf, “The Leader as Servant,” https://greenleaf.org/product/the-servant-as-leader/
membership opportunities and benefits to expand engagement at every career level. Our membership benefits—discounts on meeting registrations, education courses, and publishing; professional growth opportunities; digital library access; and honors & awards eligibility—are valuable resources for seasoned professionals while inspiring future pre-college, graduate, and undergraduate students to choose careers in electrochemical and solid state research. With our more seasoned members leading the way, this future workforce will develop the technologies that tackle important problems related to energy, health, education, the environment, national security, global development, and climate change.
Most of us believe that climate change is the most pressing challenge facing humanity and our planet. ECS members around the world are uniquely positioned to mitigate this existential threat to our planet. Rapid decarbonization by direct and indirect electrification of energy-producing and manufacturing processes will allow the world to limit the long-term increase in average global temperatures to less than 2 oC by 2050, the goal set by the Intergovernmental Panel on Climate Change. With ECS serving as the preferred global platform for continuing education, networking, collaboration, and support for career advancement, our members will succeed.
But ECS is more than just action at the international level. Our community is truly “United through Science & Technology” both globally and locally. While we continue to promote the global interconnectedness of solutions to global issues, I encourage divisions, sections, student chapters, and Institutional Partners to take action in their local communities. Reach out to your regional education systems to provide educational tools for K-12 teachers—and pave the way for advancing science & technology from the ground up, to higher levels. Promoting awareness of technical developments in electrochemistry and solid state science will extend the impact of research carried out by ECS members, and support climate change mitigation efforts by the public at large.
As president of ECS, my commitment is to collaborate with each of you, the officers, and our outstanding professional staff, to define and implement new visions and new initiatives to enable our members and future members to solve the global grand challenges before us today.
Won’t you join me in this effort?
James
(Jim)
M. Fenton ECS President
https://orcid.org/0000-0003-1996-4021
n May 23, 2025, the newly elected officers of The Electrochemical Society assumed their posts. We welcome James Fenton as president and Marca Doeff as 3rd vice president.
James (Jim) Fenton ECS President, 2025–2026
“As one gets older, one values relationships and family more than ever. In early 2022, when I was elected vice president of ECS, I had three grown children. My two daughters got married at the end of that year. While the twins, Scott and Abby (husband Grant), live far away in Pennsylvania and Minnesota, Lexi and her family (husband Bill, grandchild William, and soon-to-be grandchild Thomas) live nearby in Florida. I look forward to our get-togethers in Vermont over Christmas and during the summer. I also look forward to meeting with you—my ECS mentors, family, and long-term friends—at our biannual ECS Meetings!”
James (Jim) Fenton
James (Jim) Fenton is Professor of Materials Science and Engineering at the University of Central Florida and Director of the Florida Solar Energy Center (FSEC). He leads more than 70 faculty and staff who research, develop, and evaluate clean energy technologies, and prepare the current and future workforce for careers in clean energy. FSEC, which celebrates its 50th year in 2025, focuses on seven program areas: solar energy, high performance buildings, energy storage, decarbonization, transportation electrification, STEM education and workforce training, and energy policy research. Jim has more than 40 years of experience in electrochemical engineering and education areas, including redox flow batteries, hydrogen, PEM fuel cells, solar-to-PEM electrolyzer system analysis, fuel processing, high temperature corrosion, oxidizing agent generation, and metal recycling.
Jim holds a BS in Chemical Engineering from the University of California, Los Angeles (1979), and an MS (1982) and PhD (1984) in Chemical Engineering from the University of Illinois, Urbana-Champaign. The author of over 200 publications, he was named Fellow of The Electrochemical Society in 2007 and received the ECS Energy Technology Division Research Award in 2014 for his work on proton exchange membrane fuel cells.
An ECS member since attending his first meeting as a student in 1982, Jim’s prior service to the Society includes the offices of Secretary (2017–2021) and vice president (2022–2024), all offices of the ECS Boston Section (now the ECS New England Section) and the ECS Industrial Electrolysis and Electrochemical Engineering Division—as well as serving on many committees. He chaired the ECS student poster sessions for four years and chaired the Polymer Electrolyte Fuel Cells Student Poster Session from 2011 to 2022.
Marca Doeff is an affiliate with the Energy Storage and Distributed Resources Division (ESDR) at Lawrence Berkeley National Laboratory (LBNL). Prior to her retirement in June 2024, she was a senior scientist at LBNL and ESDR Deputy Division Director from 2019 to 2024.
Marca received her BA in Chemistry from Swarthmore College in 1978 and PhD in Inorganic Chemistry from Brown University in 1983. After postdoctoral work at the University of California, Santa Barbara, and University of California, Berkeley, she joined the Naval Ocean Systems Center in 1986 to research antifouling coatings, then in 1990 began research related to electric vehicle batteries at Lawrence Berkeley National Laboratory. Industry, the US Department of Energy, and the California Energy Commission were the primary funders of her research, which focused on materials for lithiumion batteries, sodium-ion batteries, and solid state batteries.
Marca has published approximately 170 peer-reviewed papers and has patented extensively. Her work has garnered the 2024 Lawrence Berkeley National Laboratory Director’s Lifetime Achievement Award, 2024 ECS San Francisco Section Award, 2023 US Department of Energy Office of Vehicle Technologies Distinguished Achievement Award, and 2020 R&D 100 Award. She is a Fellow of The Electrochemical Society and of the Royal Society of Chemistry. Since joining ECS in 1991, Marca has held various positions on the ECS Battery Division Executive Committee, culminating as Chair from 2019 to 2020. She has also served on several division committees, including Technical Affairs (2003–2007) and Honors & Awards (2014–2018). From 2020 to 2024 she served as Secretary of The Electrochemical Society.
Marca is married to Doug Taube. In her spare time, she enjoys outdoor activities, cats, yoga classes, traveling, and baking sourdough bread.
by Adrian T. Plummer, MPA, PMP, Senior Director of Publications
Spring is often seen as a season of renewal, and, for the ECS Publications team, the first quarter of 2025 has been a remarkable season of growth, momentum, and exciting transitions. At its heart, this progress reflects our community’s ability to embrace change as a pathway to excellence.
We began the year with important editorial leadership developments. Dr. Andrew Hillier, a long-standing ECS member (since 1992) and an engaged leader in the ECS Physical and Analytical Electrochemistry Division, was elevated to Technical Editor of the Journal of The Electrochemical Society (JES). His promotion is both a recognition of his scholarly contributions and a commitment to advancing the quality and rigor of JES in a rapidly evolving research landscape.
We also welcomed a major milestone for ECS Advances (ECSA), our fully Gold Open Access journal that launched in 2022. ECS
Robert G. Kelly, Editorin-Chief, Electrochemical Society Interface University of Virginia, Charlottesville, VA, US
Takayuki Homma, Technical Editor, Journal of The Electrochemical Society Waseda University, Tokyo, Japan Electrochemical/ Electroless Deposition
Advances is now indexed in Clarivate’s Web of Science. This achievement not only increases the visibility of the high-quality work published in ECSA but affirms ECS’s investment in broad, accessible, and ethical dissemination of electrochemical and solid state science.
Additionally, we are pleased to announce the reappointment of Dr. Robert Kelly as Editor-in-Chief of ECS Interface. His steady leadership and editorial vision have ensured that Interface remains a rich and accessible resource for ECS members across disciplines. Don’t miss his editorials—they are a highlight of every issue!
None of these accomplishments happen in isolation. They are the result of a dedicated team, forward-looking volunteers, and a shared belief that change embraced with purpose strengthens our mission and expands our impact.
Here’s to a year of continued progress, led by our commitment to adaptability, excellence, and community.
Paul Kenis, Technical Editor, Journal of The Electrochemical Society University of Illinois Urbana-Champaign, Champaign, IL, US Electrochemical Engineering
Andrew Hillier, Technical Editor, Journal of The Electrochemical Society Iowa State University, Ames, Iowa, US Physical and Analytical Electrochemistry, Electrocatalysis, and Photoelectrochemistry
The fall issue of Interface is a special issue on electrochemical impedance spectroscopy, guest edited by Masayuki Itagaki and Mark Orazem The issue includes articles on:
• Impedance of porous electrodes by Jianbo Zhang
• Impedance instrumentation by Burak Ülgüt
• Philosophy for interpretation of data by Bernard Tribollet and colleagues
Brett Lucht, Associate Editor, Journal of The Electrochemical Society University of Rhode Island, South Kingstown, RI, US Batteries and Energy Storage
John A. Staser, Associate Editor, Journal of The Electrochemical Society Ohio University, Athens, Ohio, US Electrochemical Engineering
• Use of COMSOL to model impedance by Christopher Alexander and Arthur Dizon
Fall 2025 also includes highlights from the 247th ECS Meeting, 2025 Toyota Young Investigator recipients, Pennington Corner, features favorites like Tech Highlights and Looking at Patent Law, and the latest news about people, students, and the Society.
19th International Symposium on Solid Oxide Fuel Cells (SOFC-XIX) July 13–18, 2025 | Stockholm, Sweden The Brewery Conference Center
Lester Eastman Conference on High Performance Devices (LEC
August 11–13, 2025 | Gainesville, FL University of Florida
The 2025 ECS Summer Fellowships were awarded to Mingyi Zhang (Colin Garfield Fink Fellowship), Jinhong Min (Edward G. Weston Fellowship), Ziqing Wang (Joseph W. Richards Fellowship), Monsuru Olatunji Dauda (F. M. Becket Fellowship), and Beauty K. Chabuka (H. H. Uhlig Fellowship). These fellowships provide support from June through August to students who are ECS members enrolled in a college or university—the Fink Fellowship for postgraduate researchers/engineers, and the Weston, Richards, Becket, and Uhlig Fellowships for students who are between their MS and PhD degrees—pursuing work of interest to the Society. Their fellowship reports appear in the 2025 Interface winter issue.
2026 ECS Summer Fellowships and the Colin Garfield Fink Fellowship applications are accepted from September 15, 2025, to January 15, 2026.
Mingyi Zhang received the Fink Fellowship for research on “Visualization of Cocatalyst-Induced Charge Separation via 3D Atomic Force Microscopy.” He is currently a Postdoctoral Research Associate in the laboratory of Dr. James J. De Yoreo, Chief Scientist for Materials Synthesis and Simulation Across Scales at the Pacific Northwest National Laboratory (PNNL).
Mingyi’s research focuses on understanding the structure and dynamics of the solid-liquid interface and its influence on crystal growth and nucleation processes and on photo/electrochemistry. By employing in situ atomic force microscopy and force spectroscopy to elucidate the structure and dynamics of solid-liquid interfaces, his work has advanced the fundamental understanding of crystallization and electrochemistry Mingyi joined PNNL after completing his PhD in Materials Science at Carnegie Mellon University in 2022, supervised by Prof. Gregory S. Rohrer and Prof. Paul A. Salvador He participated in the prestigious 2024 Rising Stars in Materials Science and Engineering workshop (a joint program of Stanford University, Carnegie Mellon University, the Massachusetts Institute of Technology, and the University of Illinois Urbana-Champaign), and in 2023 he received the PNNL Outstanding Performance Award and the Materials Research Society (MRS) Best Talk Award. Mingyi’s 11 published articles have an h-index of 4. He mentors four PhD candidates and two undergraduate students.
Jinhong Min will research “Operando Imaging of a Single LiFePO₄ Particle Using Interferometric Scattering Microscopy (iSCAT)” with support from the Weston Fellowship. Prof. Yiyang Li supervises Jinhong’s PhD studies in Materials Science and Engineering at the University of Michigan (U-M). Leveraging the concept of microelectrode arrays from biology, Jinhong developed a system for electrochemical measurements at the micrometer scale. Using these microelectrodes, he was able to extract kinetic parameters of individual micrometer-sized particles with high throughput.
Jinhong completed a BS in Materials Science and Engineering at the University of Seoul (2018) and an MS in Electrical and Computer Engineering at Sungkyunkwan University in 2021. His research has been recognized with a 2024 Rackham Predoctoral Fellowship at U-M, and he received a Silver Graduate Student Award at the 2023 MRS fall meeting.
Ziqing Wang’s Richards Fellowship-supported project, is titled “Unveiling Reaction Kinetics and Interfacial Electrochemistry in High-Entropy Electrolytes for Aqueous Zinc Metal Batteries.” Ziqing is a PhD student at the University of Texas at Austin (UT Austin) in the group of Prof. C. Buddie Mullins. Ziqing has developed multiple electrolyte systems for aqueous zinc-ion batteries to enable stable cycling at sub-zero temperatures and high voltages. His work also focuses on interfacial electrochemistry at the electrode surface to gain insights into reaction kinetics.
After completing a BE at Central South University in 2020, Ziqing joined UT Austin in 2021. He was awarded the University Graduate Continuing Fellowship (2024) and Allen J. Bard Center for Electrochemistry Student Scholar Fellowship (2023). The author of 21 articles (10 as first and co-first author) with an h-index of 12, Ziqing serves as a review editor for Frontiers in Chemistry. He is the co-founder and technical lead of StandUp Energy LLC. In his free time, Ziqing is an amateur body builder.
Monsuru Olatunji Dauda received the Becket Fellowship to research “Advancing CO₂ Electroreduction through Novel Electrocatalysts for Selective Carbon Products and Direct Carbonate Synthesis.” He is a PhD student at Louisiana State University (LSU) with Prof. John C. Flake as advisor. Monsuru’s research addresses the significant environmental threat of increasing levels of atmospheric carbon dioxide which necessitate sustainable technologies for greenhouse gas mitigation. He develops novel electrocatalysts that transform CO₂ into valuable C2 products and organic carbonates. Through systematic investigation and innovative synthesis approaches, this work offers a dual solution by turning a greenhouse gas into a feedstock for thousands of products that today are made from fossil fuels, while advancing understanding of the fundamental mechanisms governing product selectivity in membrane electrode assembly systems for high-performance CO₂ reduction.
Monsuru joined the PhD program at LSU after completing a BT with First Class Honors in Chemical Engineering at the Ladoke Akintola University of Technology in 2021, and an MS, also in Chemical Engineering, at LSU in 2024. He is the author of five journal articles (h-index 2), 14 conference proceedings, and two submitted patent applications. Monsuru received the ECS Industrial Electrochemistry and Electrochemical Engineering Division Student Achievement Award (2024) and served as a PRiME 2024 Meeting Student Ambassador. He has ECS-IOP Trusted Reviewer status.
Beauty K. Chabuka will study “Electron and Hole Upconversion as a New Design Paradigm for Combining Photochromism with Electrochromism” with Uhlig Fellowship support. She is a PhD candidate at Florida State University (FSU), with Prof. Igor V. Alabugin as advisor. Her research focuses on understanding the role of redox upconversion in reversible electrocyclic reactions to provide a theoretical framework for precise control over selectivity and directionality for designing molecular switches. Redox upconversion processes provide the key to using electrons as true catalysts for chemical transformations because they allow for chain propagation through electron transfer with the potential of using substoichiometric amounts of redox agents. This concept has broad implications for electrochemistry and photoredox catalysis.
Beauty completed an AAS in Secondary Education & Business Administration at the College of Southern Idaho in 2013, followed by a BS and MS in Chemistry at Idaho State University (2019). She worked as Clinical Assistant Professor at Idaho State University from 2019 to 2021. Beauty received FSU Department of Chemistry
& Biochemistry Graduate Student Travel Awards (2024 and 2025); an NSF ACCESS award (2024); an Inter-American Photochemical Society (I-APS) Winter Conference Graduate Student Best Poster Award (2023); an ACS-Florida Annual Meeting and Exposition (FAME) Student Travel Grant (2023); and an Association for Computing Machinery’s Special Interest Group on High Performance Computing Computer and Data Science Fellowship (2023). She is the author of nine articles with an h-index of 3.
The Society thanks the 2025 Summer Fellowship Subcommittee members:
• Peter Mascher, Subcommittee Chair, McMaster University
• Oumaïma Gharbi, Centre national de la recherche scientifique – Sorbonne Université
• Junsoo Han, Sorbonne Université
• Joey Kish, McMaster University
• Patrik Schmutz, Eidgenössische Materialprüfungs-und Forschungsanstalt
• Kody Wolfe, Ohio University
Praveen Sekhar, ECS Sensor Division Chair, received two prestigious awards in 2025—the Chancellor’s Award for Advancing Equity and the Students’ Award for Teaching Excellence—from the Washington State University (WSU) Vancouver, where he is Associate Professor in the School of Engineering and Computer Science.
The Chancellor’s Award for Advancing Equity honors a faculty or staff member for excellence in contributing to a community of equity, diversity, inclusion, and belonging at WSU Vancouver. The award recognizes the individual for helping to infuse equity-mindedness throughout the campus and/or helping to build and maintain a safe, welcoming campus environment. The award cites, “Praveen Sekhar has devoted his career and his life
to ensuring that access and opportunity are recognized, celebrated, and advanced on campus and beyond. His commitment to diversity pervades everything he does as teacher, scholar, research scientist, board and committee member, and mentor.”
The Students’ Award for Teaching Excellence recognizes a faculty member who commits time outside of the classroom to prevent students from falling through the cracks, demonstrates an enthusiasm for the subject matter, and instills enthusiasm and passion in students. Praveen’s students have nominated him twice (including this year) for his teaching excellence. One striking example of Praveen’s care for his students was how, on the first day of an 8:00 am class, he noticed that the students were tired. Many explained they hadn’t eaten yet. So, Praveen brought breakfast bars and fruit to class throughout the semester to make sure his students were alert to grasp the concepts.
Join us in congratulating Praveen!
The DS&T division congratulates Akash Hari Bharath, Alyx Wilhelm, and Aseel Zeinati, who won the division’s awards for travel to present at the 247th ECS Meeting in Montréal, QC, Canada. They received registration waivers, complimentary ECS membership, and cash.
Akash (University of Central Florida) and Alyx (Washington State University) presented the posters “Reactive Sputtering Studies of CuCrO2 Using Cu and Cr Targets” and “Impact of Processing on Non-
Volatile Memory and Synaptic Behaviors of Honey-CNT Memristor,” respectively. Aseel (New Jersey Institute of Technology) presented “A Comparative Study of H-Plasma Treated ZrO2 and HfO2 RRAM Devices for Low Power In-Memory Computing” in Symposium Z02—Materials, Devices, and Systems for Neuromorphic Computing and Artificial Intelligence Hardware 2.
Congratulations to all!
The results of the 2025 ECS officers’ election are:
PRESIDENT: James Fenton, University of Central Florida
3RD VICE PRESIDENT: Marca Doeff, Lawrence Berkeley National Laboratory
At the May 22 Board of Directors meeting, members voted to approve the ECS Nominating Committee’s recommended 2026 ECS Officer Election slate of candidates. The candidates
on the ballot for the election taking place from January to March 2026 are:
PRESIDENT: Francis D’Souza, University of North Texas
3RD VICE PRESIDENT: Uroš Cvelbar, Jožef Stefan Institute Jennifer Hite, University of Florida
TREASURER: Katherine Ayers, Nel Hydrogen Paul Kenis, University of Illinois at Urbana-Champaign
Full biographies and candidate statements are published in the winter 2025 issue
Executive Committee of the Board of Directors
James Fenton
Francis D’Souza
Robert Savinell
Marca Doeff
Gessie Brisard
Elizabeth Podlaha-Murphy
Alice Suroviec
President, Spring 2026
Senior Vice President, Spring 2026
2nd Vice President, Spring 2026
3rd Vice President, Spring 2026
Secretary, Spring 2028
Treasurer, Spring 2026
Community Inclusion Chair, Spring 2029
Christopher Jannuzzi Executive Director, Term as Executive Director (ED)
Community Inclusion Committee
Alice Suroviec, Chair Spring 2029
Natalia Bencomo Spring 2029
Anne Co Spring 2029
Damilola Daramola Spring 2029
Jeffrey Halpern Spring 2029
Yoshinao Hoshi Spring 2029
Nicole Iverson Spring 2029
Vivek Kamat Spring 2029
Eva Kovacevic Spring 2029
Fred Omenya Spring 2029
Nicola Perry Spring 2029
Kay Song Spring 2029
Natasa Vasiljevic Spring 2029
Justyna Zeler Spring 2029
TBD – student member Spring 2027
TBD – student member Spring 2027
Audit Committee
Colm O’Dwyer, Chair
Immediate Past President, Spring 2026
James Fenton President, Spring 2026
Francis D’Souza Senior Vice President, Spring 2026
Kevin Johnson Nonprofit Financial Professional, Spring 2028
Elizabeth Podlaha-Murphy Treasurer, Spring 2026
Education Committee
Scott Calabrese Barton, Chair Spring 2029
Damilola Daramola Spring 2027
Samantha Gateman Spring 2026
Dominika Grzesiak Spring 2027
Maria Eugenia Toimil-Molares Spring 2029
Wen Shen Spring 2026
Maureen Tang Spring 2027
Justyna Zeler Spring 2029
TBD Community Inclusion Representative, Spring 2026
Gessie Brisard
Secretary, Spring 2028
E. Jennings Taylor Chair, Individual Membership Committee, Spring 2026
Ethical Standards Committee
Colm O’Dwyer, Chair
Gessie Brisard
Immediate Past President, Spring 2026
Secretary, Spring 2028
Peter Fedkiw Past Officer, Spring 2026
Elizabeth Podlaha-Murphy
Treasurer, Spring 2026
Daniel Steingart Past Officer, Spring 2027
Alice Suroviec Community Inclusion Chair, Spring 2029
Finance Committee
Elizabeth Podlaha-Murphy, Chair
Treasurer, Spring 2026
Helmut Baumgart Spring 2028
Gessie Brisard Secretary, Spring 2028
Paul Kenis Spring 2026
Thorsten Lill Spring 2026
Kevin Johnson
Nonprofit Financial Professional, Spring 2028
Tim Gamberzky Chief Operating Officer, Term as COO
Honors and Awards Committee
Adam Weber, Chair Spring 2027
Katherine Ayers Spring 2029
Elizabeth Biddinger Spring 2027
Stanko Brankovic Spring 2027
Mikhail Brik Spring 2028
Dev Chidambaram Spring 2026
Uroš Cvelbar Spring 2028
Andrew Hoff Spring 2026
Y. Shirley Meng Spring 2026
Roberto Paolese Spring 2029
Stephen Paddison Spring 2029
Thomas Thundat Spring 2027
Siegfried Waldvogel Spring 2028
TBD Community Inclusion Representative, Spring 2026
James Fenton President, Spring 2026
Individual Membership Committee
E. Jennings Taylor, Chair Spring 2026
Jedidian Adjetey Adjei Spring 2027
Sarah Berlinger Spring 2028
Uroš Cvelbar Spring 2026
Joshua Gallaway Spring 2027
Hussain Abdul Jabbar Spring 2027
Sahand Serajian Spring 2027
Tomoyuki Yamamoto Spring 2028
Kent Jingxu Zheng Spring 2026
TBD .....................................................................................Community Inclusion Representative, Spring 2026
Gessie Brisard Secretary, Spring 2028
Alex Peroff Chair, Institutional Engagement Committee, Spring 2028
Institutional Engagement Committee
Alex Peroff, Chair Spring 2028
Vimal Chaitanya Spring 2026
William Cohen Spring 2027
Hanping Ding Spring 2026
Russell Freed Spring 2028
Hemanth Jagannathan Spring 2026
Vivek Kamat Spring 2028
Jacob Ketter Spring 2027
Dongping Lu Spring 2028
John Muldoon Spring 2027
Elizabeth Podlaha-Murphy Treasurer, Spring 2026
E. Jennings Taylor Chair, Individual Membership Committee, Spring 2026
Nominating Committee
Colm O’Dwyer, Chair
Marca Doeff
Immediate Past President, Spring 2026
3rd Vice President, Spring 2026
Stefan De Gendt Spring 2026
William Mustain Spring 2026
Sannakaisa Virtanen Spring 2026
Alice Suroviec
Christopher Jannuzzi
Community Inclusion Chair, Spring 2029
Executive Director, Term as ED Technical Affairs Committee
Francis D’Souza, Chair
Colm O’Dwyer
Senior Vice President, Spring 2026
Immediate Past President, Spring 2026
Robert Savinell Chair, Publications Subcommittee, Spring 2026
Gerardine Botte
2nd
Immediate Past President, Spring 2026
Maria Inman Chair, Interdisciplinary Science and Technology Subcommittee, Spring 2028
James Fenton President, Spring 2026
Marca Doeff Chair, Meetings Subcommittee, Spring 2026
Christopher Jannuzzi
Executive Director, Term as ED
Gessie Brisard-Secretary Spring 2028
Elizabeth Podlaha-Murphy Treasurer, Spring 2026
Alice Suroviec
Community Inclusion Chair, Spring 2029
Publications Subcommittee of the Technical Affairs Committee
Robert Savinell, Chair
Marca Doeff, Vice Chair
2nd Vice President, Spring 2026
3rd Vice President, Spring 2026
David Cliffel Journal of The Electrochemical Society EiC, 11/14/2026
Robert Kelly Interface EiC, 5/31/2026
Krishnan Rajeshwar ECS Journal of Solid State Science and Technology EiC, 12/31/2025
Sreeram Vaddiraju Spring 2027
Fan Ren 12/31/2025
TBD Community Inclusion Representative, Spring 2026
Netz Arroyo ECS Sensors Plus EIC, 12/31/2025
Rohan Akolkar ECS Advances EIC, Spring 2027
Xingfan Jin Member Spring 2026
Marc Secanell Member Spring 2026
Meetings Subcommittee of the Technical Affairs Committee
Marca Doeff, Chair
Robert Savinell, Vice Chair
3rd Vice President, Spring 2026
2nd Vice President, Spring 2026
Vito Di Noto Spring 2028
Xiaolin Li Spring 2026
Xiao Su Spring 2027
TBD Community Inclusion Representative, Spring 2026
Interdisciplinary Science and Technology Subcommittee of the Technical Affairs Committee
Maria Inman, Chair Spring 2028
Andreas Bund Spring 2028
Vidhya Chakrapani Spring 2026
Dev Chidambaram Spring 2028
Huyen Dinh Spring 2026
Alanah Fitch Spring 2026
Daniel Heller Spring 2028
David Hickey Spring 2027
Chockkalingam Karuppaiah Spring 2027
Rangachary Mukundan Spring 2027
David Reber Spring 2028
Alok Srivastava Spring 2027
Jianhua Tong Spring 2028
Sreeram Vaddiraju Spring 2026
TBD - Community Inclusion Member Spring 2026
Symposium Planning Advisory Board of the Technical Affairs Committee
Marca Doeff, Chair 3rd Vice President, Spring 2026
Jeffrey L. Blackburn Chair, Nanocarbons Division, Spring 2026
Vidhya Chakrapani Chair, Electronics and Photonics Division, Spring 2027
Anne Co Chair, Physical and Analytical Electrochemistry Division, Spring 2027
Ariel Furst Chair, Organic and Biological Electrochemistry Division, Spring 2027
Maria Inman Chair, Interdisciplinary Science and Technology Subcommittee, Spring 2028
Paul Kenis Chair, Industrial Electrochemistry and Electrochemical Engineering Division, Spring 2026
Cortney Kreller Chair, High-Temperature Energy, Materials, & Processes Division, Fall 2025
Luca Magagnin Chair, Electrodeposition Division, Fall 2025
Praveen Sekhar Chair, Sensor Division, Fall 2026
Minhua Shao Chair, Energy Technology Division, Spring 2027
Eiji Tada Chair, Corrosion Division, Fall 2026
Sreeram Vaddiraju Chair, Dielectric Science and Technology Division, Spring 2026
Jie Xiao Chair, Battery Division, Fall 2026
Eugeniusz Zych Chair, Luminescence and Display Materials Division, Fall 2025
Other Representatives
Society Historian Roque Calvo Spring 2026
American Association for the Advancement of Science
Christopher Jannuzzi Executive Director, Term as ED
National Inventors Hall of Fame
Adam Weber Chair, Honors & Awards Committee, Spring 2027
Four divisions held elections in April. These division executive committee members took office on May 23, 2025.
Electronics and Photonics Division
Chair
Vidhya Chakrapani, Rensselaer Polytechnic Institute
Vice Chair
Zia Karim, Yield Engineering Systems 2nd Vice Chair
Travis Anderson, University of Florida
Secretary
Jennifer Hite, University of Florida
Treasurer
Helmut Baumgart, Old Dominion University Members at Large
Albert Baca, Sandia National Laboratories
D. Noel Buckley, University of Limerick
Yu Lun Chueh, National Tsing Hua University
Stefan De Gendt, imec
M. Jamal Deen, McMaster University
Andrew M. Hoff, University of South Florida
Hiroshi Iwai, National Yang Ming Chiao Tung University
Hemanth Jagannathan, IBM Corporation Research Center
Soohwan Jang, Dankook University
Daisuke Kiriya, The University of Tokyo
Chung-Wei Kung, National Cheng Kung University
Yue Kuo, Texas A&M University
Jan Macák, Univerzita Pardubice
Junichi Murota, Tohoku University
Yaw Obeng, National Institute of Standards & Technology
Colm O’Dwyer, University College Cork
Takahito Ono, Tohoku University
Mark E. Overberg, Sandia National Laboratories
Harold Philipsen, imec
Fred Roozeboom, Universiteit Twente
Kay Song, Pivotal Systems
Tadatomo Suga, Meisei University
Yu-Lin Wang, National Tsing Hua University
Matthias Young, University of Missouri
Energy Technology Division
Chair
Minhua Shao, Hong Kong University of Science and Technology
Vice Chair
Hui Xu, Envision Energy USA
Secretary
Iryna Zenyuk, University of California, Irvine
Treasurer
Ertan Agar, University of Massachusetts Lowell
Members at Large
Christopher Arges, Argonne National Laboratory
Plamen B. Atanassov, University of California, Irvine
Siddharth Komini Babu, Los Alamos National Laboratory
Sarah Berlinger, Lawrence Berkeley National Laboratory
Scott Calabrese Barton, Michigan State University
Rod Borup, Los Alamos National Laboratory
Vito Di Noto, Università degli Studi di Padova
Huyen Dinh, National Renewable Energy Laboratory
James Fenton, University of Central Florida
Andrew Herring, Colorado School of Mines
Paul Kenis, University of Illinois Urbana-Champaign
Ahmet Kusoglu, Lawrence Berkeley National Laboratory
Matthew Mench, University of Tennessee, Knoxville
Sanjeev Mukerjee, Northeastern University
William Mustain, University of South Carolina
Mariappan Parans Paranthaman, Oak Ridge National Laboratory
Peter Pintauro, Vanderbilt University
Bryan Pivovar, National Renewable Energy Laboratory
Yuliya Preger, Sandia National Laboratories
Krishnan Rajeshwar, University of Texas at Arlington
James Saraidaridis, RTX Technology Research Center
Jacob Spendelow, Los Alamos National Laboratory
Jean St-Pierre, Cummins Technical Center
Adam Weber, Lawrence Berkeley National Laboratory
John Weidner, University of Cincinnati
Gang Wu, Washington University in St. Louis
Nianqiang Nick Wu, University of Massachusetts Amherst
Thomas Zawodzinski, University of Tennessee, Knoxville
Gaohua Zhu, Toyota North America
Chair
Organic and Biological Electrochemistry Division
Ariel Furst, Massachusetts Institute of Technology
Vice Chair
Jeffrey Halpern, University of New Hampshire
Secretary/Treasurer
David Hickey, Michigan State University
Members at Large
Mahito Atobe, Yokohama University
Graham Cheek, United States Naval Academy
David Cliffel, Vanderbilt University
Robert Francke, Leibniz-Institut für Katalyse
Matt Graaf, Corteva Agriscience
Seyyeadmirhossein Hosseini, University of South Carolina
Shinsuke Inagi, Institute of Science Tokyo
Jiří Ludvík, J. Heyrovský Institute of Physical Chemistry
Kevin Moeller, Washington University in St. Louis
Olja Simoska, University of South Carolina
Alice Suroviec, Berry College
Charuksha Walgama, University of Houston–Clear Lake
Chair
Physical and Analytical Electrochemistry Division
Anne Co, Ohio State University Vice Chair
Svitlana Pylypenko, Colorado School of Mines
Secretary
Iwona Rutkowska, Uniwersytet Warszawski
Treasurer
Valentine Vullev, University of California, Riverside Members at Large
Mario Alpuche-Aviles, University of Nevada, Reno
Plamen B. Atanassov, University of California, Irvine
D. Noel Buckley, University of Limerick
David Cliffel, Vanderbilt University
Abdoulaye Djire, Texas A&M University
Alanah Fitch, Loyola University
Burcu Gurkan, Case Western Reserve University
Andrew Hillier, Iowa State University
A. Robert Hillman, University of Leicester
Yasushi Katayama, Keio University
Paweł J. Kulesza, Uniwersytet Warszawski
Johna Leddy, University of Iowa
Robert Mantz, US Army Research Office
Shelley Minteer, University of Utah
Yue Qi, Brown University
Hang Ren, University of Texas at Austin
Joaquín Rodríguez-López, University of Illinois at UrbanaChampaign
Alice Suroviec, Berry College
Greg Swain, Michigan State University
Paul Trulove, United States Naval Academy
Petr Vanýsek, Northern Illinois University
Robert Warburton, Case Western Reserve University
Yingjie Zhang, University of Illinois at Urbana-Champaign
The ECS Electrodeposition Division; High-Temperature Energy, Materials, & Processes Division; and Luminescence and Display Materials Division nominated new officers for the fall 2025 to fall 2027 term. Elections take place from September 1–30, 2025, with results reported in the ECS Interface winter 2025 issue.
Chair
Electrodeposition
Andreas Bund, Technische Universität Ilmenau Vice Chair
Rohan Akolkar, Case Western Reserve University
Secretary
Adriana Ispas, Technische Universität Ilmenau
Treasurer
Massimo Innocenti, Università degli Studi di Firenze Members at Large
Faisal Alamgir, Georgia Institute of Technology
Antoine Allanore, Massachusetts Institute of Technology
Trevor Braun, ElectraSteel, Inc.
Jean-Yves Hihn, Université Marie et Louis Pasteur
Adam Maraschky, ElectraSteel Inc.
Toshiyuki Nohira, Kyoto University
Maria Eugenia Toimil-Molares, GSI Helmholtz Centre für Schwerionenforschung
BungUk Yoo, Korea Institute of Materials Science
Kent Zheng, University of Texas at Austin
Chair
High-Temperature Energy, Materials, & Processes
Xingbo Liu, West Virginia University
Vice Chair
Teruhisa Horita, National Institute of Advanced Industrial Science and Technology
Junior Vice Chair
Dong Ding, Idaho National Laboratory
Secretary/Treasurer
Xinfang Jin, University of Texas at Dallas Members at Large
Mohammed Hussain Abdul Jabbar, Nissan Group of North America
Stuart Adler, University of Washington
Mark D. Allendorf, Sandia National Laboratories
Jihwan An, Seoul National University of Science and Technology
Sean Bishop, Sandia National Laboratories
Fanglin (Frank) Chen, University of South Carolina
Zhe Cheng, Florida International University
Wilson Chiu, University of Connecticut
Chuancheng Duan, University of Utah
Jan Froitzheim, Chalmers tekniska högskola
Paul Gannon, Montana State University Bozeman
Fernando Garzon, University of New Mexico
Srikanth Gopalan, Boston University
Turgut Gür, Stanford University
Liangbing Hu, University of Maryland
Kevin Huang, University of South Carolina
Greg S. Jackson, Colorado School of Mines
Tatsuya Kawada, Tohoku University
KangTaek Lee, Korea Advanced Institute of Science and Technology
Wonyoung Lee, Sungkyunkwan University
Olga Marina, Pacific Northwest National Laboratory
Nguyen Minh, University of California, San Diego
Jason Nicholas, Michigan State University
Elizabeth Opila, University of Virginia
Nicola Perry, University of Illinois at Urbana-Champaign
Kannan Ramaiyan, University of New Mexico
Sandrine Ricote, Colorado School of Mines
Subhash Singhal, Pacific Northwest National Laboratory
Anna Staerz, Colorado School of Mines
Adnan Syed, Cranfield University
Hitoshi Takamura, Tohoku University
Jianhua Tong, Clemson University
Enrico Traversa, Università di Roma Tor Vergata
Eric Wachsman, University of Maryland
Yudong Wang, University of Connecticut
Leta Woo, Cummins, Inc.
Bilge Yildiz, Massachusetts Institute of Technology
Xiao-Dong Zhou, University of Connecticut
Chair
Luminescence and Display Materials
Chong-Geng Ma, Chongqing University of Posts and Telecommunications
Vice Chair
William Cohen, Current Chemicals
Secretary/Treasurer
Luiz Jacobsohn, Clemson University
Members at Large
Marco Bettinelli, Università degli Studi di Verona
Mikhail Brik, Tartu Ülikool
John Collins, Wheaton College
Won Bin Im, Hanyang University
Tetsuhiko Isobe, Keio University
Ru-Shi Liu, National Taiwan University
Kazuyoshi Ogasawara, Kwansei Gakuin University
Alok Srivastava, Srivastava Consulting LLC
Battery
Jie Xiao, Chair
Pacific Northwest National Laboratory
Jagjit Nanda, Vice Chair
Xiaolin Li, Secretary
Neil Dasgupta, Treasurer
Doron Aurbach, Journals Editorial Board Representative
Corrosion
Eiji Tada, Chair Institute of Science Tokyo
Rebecca Schaller, Vice Chair
Yaiza Gonzalez-Garcia, Secretary/Treasurer
Sannakaisa Virtanen, Journals Editorial Board Representative
Dielectric Science and Technology
Sreeram Vaddiraju, Chair
Texas A&M University
Eva Kovacevic, Vice Chair
Zhi David Chen, Secretary
Thorsten Lill, Treasurer
Peter Mascher, Journals Editorial Board Representative
Electrodeposition
Luca Magagnin, Chair
Politecnico di Milano
Andreas Bund, Vice Chair
Rohan Akolkar, Secretary
Adriana Ispas, Treasurer
Takayuki Homma, Journals Editorial Board Representative
Electronics and Photonics
Vidhya Chakrapani, Chair
Rensselaer Polytechnic Institute
Zia Karim, Vice Chair
Travis Anderson, 2nd Vice Chair
Jennifer Hite, Secretary
Helmut Baumgart, Treasurer
Aniruddh Jagdish Khanna, Journals Editorial Board Representative
Fan Ren, Journals Editorial Board Representative
Energy Technology
Minhua Shao, Chair
Hong Kong University of Science and Technology
Hui Xu, Vice Chair
Iryna Zenyuk, Secretary
Ertan Agar, Treasurer
Minhua Shao, Journals Editorial Board Representative
High-Temperature Energy, Materials, and Processes
Cortney Kreller, Chair
Los Alamos National Laboratory
Xingbo Liu, Vice Chair
Teruhisa Horita, Junior Vice Chair
Dong Ding, Secretary/Treasurer
Minhua Shao, Journals Editorial Board Representative
Industrial Electrochemistry and Electrochemical Engineering
Paul Kenis, Chair
University of Illinois at Urbana-Champaign
Elizabeth Biddinger, Vice Chair
Chockalingam Karuppaiah, Secretary/Treasurer
Paul Kenis, Journals Editorial Board Representative
Luminescence and Display Materials
Eugeniusz Zych, Chair
Uniwersytet Wrocławski
Chong-Geng Ma, Vice Chair
Marco Bettinelli, Secretary/Treasurer
Won Bin Im, Journals Editorial Board Representative
Nanocarbons
Jeffrey L. Blackburn, Chair
National Renewal Energy Laboratory
Ardemis Boghossian, Vice Chair
Yan Li, Secretary
Hiroshi Imahori, Treasurer
Dirk M. Guldi, Journals Editorial Board Representative
Organic and Biological Electrochemistry
Ariel Furst, Chair
Massachusetts Institute of Technology
Jeffrey Halpern, Vice Chair
David Hickey, Secretary/Treasurer
Janine Mauzeroll, Journals Editorial Board Representative
Physical and Analytical Electrochemistry
Anne Co, Chair
Ohio State University
Svitlana Pylypenko, Vice Chair
Iwona Rutkowska, Secretary
Valentine Vullev, Treasurer
Andrew Hillier, Journals Editorial Board Representative
Sensor
Praveen Kumar Sekhar, Chair
Washington State University
Dong-Joo Kim, Vice Chair
Leyla Soleymani, Secretary
Harshini Mukundan, Treasurer
Netz Arroyo, Journals Editorial Board Representative
Stefano Cinti, Journals Editorial Board Representative
ECS welcomes these great new additions to our Institutional Partner Program.
BMW Group is the world’s leading manufacturer of premium cars and motorcycles and a provider of premium financial and mobility services. Headquartered in Munich, Germany, BMW operates in over 30 production sites around the world.
To learn more, visit BMW Group’s website at https://www. bmwgroup.com/en.html
easyXAFS is the global leader in laboratory XAFS and XES instrumentation. Our proven laboratory X-ray spectrometers give synchrotronquality spectra and game-changing scientific freedom. Add to your existing research program or launch a new thrust. Imagine what you can do with reliable, easy access to advanced X-ray spectroscopy.
To learn more, visit easyXAFS’s website at http://www.easyxafs. com/
next Machinery Group is the exclusive North American reseller of Coatema® Coating Machinery, a global leader in coating, printing, and laminating technologies. We provide scalable equipment solutions for research, pilot, and full-scale production in high-tech industries such as batteries, fuel cells, electrolyzers, and printed electronics. With access to one of the world’s most advanced R&D centers in Germany and a large battery center at The Ohio State University, we support startups, research institutions, and manufacturers in accelerating innovation from lab to fab. Our expertise spans customized engineering, process development, and system integration—empowering customers to optimize performance, reduce time to market, and scale with confidence in the fast-evolving electrochemical and energy sectors. Learn more at next Machinery Group’s website: https://next-mg. com/
machiner y group
ECS’ Institutional Partner Program opens doors to a network that helps organizations meet business goals and objectives. Contact Sponsorship@electrochem.org for more information.
ECS acknowledges and thanks our long-term supporters and partners in electrochemistry and solid state science through Leadership Circle Awards. Since 2002, these awards have been granted when Institutional Partners reach milestone year levels. Congratulations to BASi, which reached a 10-year milestone in 2025. The Society is grateful for BASi’s support through these past years.
ECS acknowledges and thanks our long-term supporters and partners in electrochemistry and solid state science through Leadership Circle Awards. Since 2002, these awards have been granted when Institutional Partners reach milestone year levels. Congratulations to BASi, which reached a 10-year milestone in 2025. The Society is grateful for BASi’s support through these past years.
Silver Partner – 10 years BASi
Silver Partner – 10 years BASi
For more about BASi, visit their website at https://www.basinc.com/index.php
For more about BASi, visit their website at https://www.basinc.com/index.php
To learn more about the Institutional Partnership Program, contact Anna Olsen, Senior Manager, Corporate Programs.
To learn more about the Institutional Partner Program, contact Anna Olsen, Senior Manager, Corporate Programs
by Alice H. Suroviec
Patrick von Rosen and Lennart Peters meet with leading experts and scientists in the battery field. The podcast is produced by Helmholtz-Institut Ulm and the Kompetenzzentrum Elektrochemische Energiespeicherung Ulm & Karlsruhe (CELEST). Topics of the podcast range from AI in battery research to fundamentals of battery technology.
https://batterygeneration.podigee.io
Curious Cases is a science podcast hosted by mathematician Dr. Hannah Fry and comedian and science-enthusiast Dara Ó Briain. In each episode they explore science questions submitted by listeners. The hosts blend humor with the scientific method, making complex topics engaging and easy to understand. Recent episodes include exploring the science of deception and investigating if anything can truly be invisible.
https://bbc.com/audio/brand/b07dx75g
iLikeBatteries is a weekly podcast with experts in the area of battery technology. Jeff Lynch and Mike Muldoon cover topics that range from using EV cars to battery start-up companies. The episodes are a blend of casual conversations with scientific experts. With more than 200 episodes to choose from, there is a topic here for everyone to enjoy.
https://podcast.ilikebatteries.com
Alice Suroviec is a Professor of Bioanalytical Chemistry and Dean of the School of Mathematical and Natural Sciences at Berry College. She earned a BS in Chemistry from Allegheny College in 2000. She received her PhD from Virginia Tech in 2005 under the direction of Dr. Mark R. Anderson. Her research focuses on enzymatically modified electrodes for use as biosensors. She is a Fellow of the Electrochemical Society and Associate Editor of the PAE Technical Division for the Journal of The Electrochemical Society. She welcomes feedback from the ECS community.
https://orcid.org/0000-0002-9252-2468
Frances Chaves joined ECS as a fulltime employee on April 1, 2025, after five years as a part-time contractor. “Frances’ impressive background and diverse skill set will further boost our marketing and communications efforts and enrich the content we deliver to the ECS community. Known for her sharp eye and thoughtful approach, Frances’ contributions are sure to elevate everything we produce. It’s an exciting step forward to have a Content Creator/Editor as part of our team, and we look forward to all we will accomplish together!” says Fern Oram, Senior Manager, Marketing & Communications.
Frances has a BA (Reed College), MA (New York University), and “mini-MBA” in Digital Marketing (Rutgers). Her clients as a contractor included LexisNexis, Wiley, Princeton HealthCare System Foundation, and educational institutions. Prior to that, she was Director of Marketing & Special Events at Enable, Executive Director at The Lacoste School of the Arts and Anderson Ranch Art Center, and Curator for Reader’s Digest. Her extensive international experience includes living in Belgium, France, Spain, and Switzerland, and projects in Australia, Europe, Hong Kong, Mexico, New Zealand, and South Africa. Her husband, Michael Maloney, is a flavorist, and their son, Luke, is a chemical engineer at Regeneron. In her free time, Frances enjoys her rescued dogs, Leo and Miss Cassie Wiggle-Bottom, swimming at the Hopewell Quarry, and reading (lots of) fiction.
Former President of The Electrochemical Society Dr. Barry Miller passed away on February 7, 2025, in Cleveland, OH. Barry was born in Passaic, NJ, to Belle and Herman Miller on January 22, 1933. He graduated from Princeton University with an AB summa cum laude in 1955 and received his PhD in chemistry from the Massachusetts Institute of Technology in 1959. He began his professional career as an Instructor in Chemistry at Harvard University (1959–1962), then served as a Member of the Technical Staff of AT&T Bell Laboratories (1962–1993). In 1993 he was named the Frank Hovorka Professor of Chemistry at Case Western Reserve University, a title he held as emeritus professor after his retirement in 2000.
A member of The Electrochemical Society for more than 60 years, Barry served as the ECS President from 1997 to 1998 and as editorin-chief of the Journal of The Electrochemical Society from 1990 to 1995. He was a member of the Board of Directors of ECS from 1987 to 1989 and 1994 to 1998, Chairman of the Physical Electrochemistry Division from 1987 to 1989, a member of many Society committees, and co-organizer of numerous symposia for ECS, notably the first Symposium on Fullerenes: Physics, Chemistry, and New Directions
We are saddened to report the death of Prof. Dr. Hans Jürgen Schäfer (Hans), Professor Emeritus, University of Münster, on 17 August 2024. Hans obtained his PhD under the guidance of Ulrich Schöllkopf at the University of Heidelberg before carrying out postdoctoral studies at Yale with Kenneth Wiberg. Hans returned to Germany in 1966 to assume a position at the University of Göttingen. There, he initiated his independent career before moving to the University of Münster in 1973.
Hans lived a rich and meaningful life. He pioneered the development of organic electrosynthesis and demonstrated its utility long before it had attracted the attention it currently holds. Very
(1991) and the initial Symposia on High Temperature Superconductors (1988–1989). Outside of ECS he served as President of the Society for Electroanalytical Chemistry, Chairman of the Gordon Conference on Electrochemistry, Associate Member of the IUPAC Commission on Electrochemistry, and as National Secretary of the International Society of Electrochemistry. Additionally, he was named to multiple US Government Panels, including the Panel on the US Advanced Battery Consortium of the National Research Council and the Cold Fusion Panel of the Department of Energy.
The professional recognition he received over his long and distinguished career includes the ECS Physical and Analytical Electrochemistry Division David C. Grahame Award (1991), being named a Fellow of The Electrochemical Society (1992) and an Honorary Member of the ECS (1999), the Ernest B. Yeager Award of the ECS Cleveland Section (2004), and the ECS 2018 Edward Goodrich Acheson Award. He received the Charles N. Reilley Award from the Society of Electroanalytical Chemistry in 1994.
Barry met Sandra, who would become his wife of 59 years, when she was an executive chef, cooking teacher, and caterer at the Lake Tarleton Club in Pike, NH, in 1964. The two were married at the Chanticler in Millburn, NJ, on August 22, 1965. Barry is survived by Sandra; his two sons Jeff and David; daughters-in-law Amy and Michelle; and five grandchildren, Skyler, Riley, Maddie, Jason, and Kyle. Some of his fondest memories were whitewater rafting down the Colorado River through the Grand Canyon and traveling the world.
early he realized the potential of electrochemistry for syntheses based on redox reactions and championed the emergence of synthetic electrochemistry in the 70s and 80s of the last century. His contributions were formally recognized when in 1998 he was named the third recipient of the Manuel M. Baizer Award from the Organic and Biological Electrochemistry Division of the ECS.
Hans inspired scientists throughout the world, particularly those who wished to begin a career involving organic electrochemistry, or simply wanted to learn how to carry out a reaction electrochemically. He greatly supported organic electrosynthesis internationally through his original research, scholarly manuscripts and reviews, and his very active participation at conferences throughout the world. He will be dearly remembered by his friends and others in the organic electrochemistry community.
This remembrance was contributed by Dan Little.
edited by Scott Cushing, Interface Contributing Editor
This feature is intended to let ECS award-winning students and post-docs write primary author perspectives on their field, their work, and where they believe things are going. This month we highlight the work of Maha Yusuf and Noor Ul Hassan, 2024 recipients of the Energy Technology Division Graduate Student Award Sponsored by BioLogic.
by Maha Yusuf
Efforts to decarbonize are vital to reducing greenhouse gas (GHG) emissions by 45% by 2030 and achieving netzero emissions by 2050.1 Electrifying the transportation sector—which accounts for approximately 28% of GHG emissions in the United States—is central to meeting these goals.2 However, widespread adoption of electrified transport hinges on the development of next-generation batteries that are low-cost, safe, fast-charging, long-lasting, composed of abundant materials, and resilient under extreme conditions.3
Conventional lithium-ion batteries (LIBs) and emerging solidstate batteries (SSBs), particularly those using lithium-metal-free or anode-free configurations, face significant challenges related to interfacial degradation.4-6 These challenges stem from a limited mechanistic understanding of energy losses during Li plating and stripping.6 Key scientific gaps remain in understanding critical degradation mechanisms, such as electrode deterioration,4 electrochemo-mechanical instabilities,7 and Li behavior,8 including morphological changes and spatial heterogeneities. Addressing these gaps is crucial for the practical implementation of advanced battery technologies.
In situ 3D characterization is essential to the investigation of these mechanisms, as batteries remain chemically and electrochemically active even after cycling.9-10 A 3D approach enables the visualization of Li plating and stripping across the XY-plane and through the Z-axis, capturing spatial heterogeneities both laterally and in electrode depth. By combining 3D morphological insights with quantitative analysis, researchers can gain a deeper understanding of the electrochemical phenomena driving battery degradation.
In situ neutron and X-ray micro-computed tomography (µCT) offers a powerful tool for unraveling these degradation mechanisms in LIBs and SSBs. Neutrons are highly sensitive to low-Z materials like Li,11 while X-rays excel in detecting high-Z materials such as Cu12 (Table I). By combining these modalities, simultaneous neutron and X-ray tomography (NeXT) enables segmentation of multiple battery components from the same sample location, offering unprecedented insights.
This perspective highlights the value of NeXT as a non-destructive imaging modality through two case studies: (1) Imaging graphite electrode degradation in fast-charged LIBs,13 (2) Investigating electro- chemo-mechanical instabilities at solid-solid interfaces in anode-free SSBs.14 Additionally, I discuss high-resolution 3D neutron imaging (10–15 µm) to visualize Li morphologies and
spatial heterogeneities, emphasizing their role in LIB degradation following fast changing.15 (Fig. 1) Finally, I provide insights into the potential of these advanced neutron and X-ray imaging diagnostics for emerging battery chemistries and the challenges associated with leveraging these techniques.
This section highlights NeXT as a non-destructive imaging technique that overlaps the neutron and X-ray data from the same sample location for material-specific identification. Using a 2D bivariate histogram, NeXT integrates grayscale dual-tomography data to generate colorized segmented images.
In our study, NeXT facilitated ex situ 3D visualization of graphite electrode degradation in fast-charged LIBs, conducted at the BT-2 imaging beamline16 at the NIST Center for Neutron Research (NCNR), with a pixel size of approximately 6.5 µm and a resolution of 15–20 µm.13 We examined one pristine and two fast-charged graphite electrodes (9C charge for 450 cycles) harvested from singlelayer graphite/NMC532 pouch cells post-cycling. Segmented images revealed a qualitative correlation between electrode degradation and Li plating, with protrusions attributed to LiOH or its hydrated porous crystallites formed during sample preparation due to Li’s exposure to moisture. Without moisture or sample preparation effects, the degradation is primarily expected to involve Li deposits and potential delamination from the Cu current collector (CC). This
Table I. Total neutron cross-sections and total mass attenuation coefficients at 40keV X-ray energy for battery materials of interest: graphite, lithium, copper.
methodology establishes NeXT as a versatile tool for investigating battery chemistries beyond LIBs, such as Li-O₂, Li-metal, and Li-Si, by providing detailed insights into Li plating to address fundamental challenges in energy storage science.
Li plating and stripping on a metal (e.g., Cu) CC is the primary energy storage mechanism in anode-free SSBs, making NeXT an ideal tool for non-destructive investigation of 3D electro-chemomechanical instabilities in these systems. A key question is whether mechanical instabilities (fractures) in these batteries are driven by electrochemical forces, such as current density (CD) and Li penetration. Using the NeXT beamline at the Institut Laue-Langevin
(ILL)17 in Grenoble, France (neutron pixel size ~4 µm; X-ray pixel size ~6 µm), we imaged three cells in pristine and plated states at low (0.5 µA) and high (5 µA) CDs.14
X-ray-µCT revealed pre-existing fractures near the edges of the Li₆PS₅Cl solid electrolyte (SE) pellet in pristine cells, with larger fractures forming near the center after cycling at both CDs. NeutronµCT uniquely detected Li penetration within these fractures, which was absent in pristine cells. At high CD, Li exhibited “flow-like” morphology in larger fractures, while at low CD, “filament-like” Li formed in smaller cracks, indicating distinct plating behaviors. These findings suggest that fracture widening is driven by electrochemical forces at high CDs, with Li penetration into fractures attributed to viscoplastic deformation, where Li flows until equilibrium is reached with friction. Future work will extend phase-field modeling to explain the “filament-like” Li plating at the CC|SE interface.
(continued on next page)
Fig. 1. (A) (Left) Schematic of simultaneous neutron and X-ray tomography, illustrating the neutron and X-ray beams interacting with the sample simultaneously. (Right) 3D visualization of graphite electrode degradation in a fast-charged lithium-ion battery (9C charge for 450 cycles), showing LiOHcontaining protrusions and delamination from the Cu current collector (CC). Adapted from Reference 13 by permission of Elsevier Publishing. (B) Cross-sectional (XZ) reconstructed grayscale images from 3D volumes obtained via (Top) X-ray and (Bottom) neutron micro-computed tomography (µCT) of an anode-free solid-state battery after plating at 5.0 µA current density. X-ray-µCT reveals fractures of varying length scales within the Li₆PS₅Cl solid electrolyte (SE) pellet, while neutron-µCT visualizes “flow-like” Li penetration into these fractures, which is not visible in the X-ray-µCT image. © The Electrochemical Society. Adapted from Reference 14 by permission of IOP Publishing. All rights reserved. (C) 3D visualizations of plated, dead, and active Li in a lithium-ion battery cycled at 6C for 6 cycles. Dashed red circles highlight non-uniform density in plated and dead Li. Blue arrows indicate tip-like plated Li deposits formed at 6C, while dashed blue circles identify corresponding tip-like dead Li deposits. Regions of no dead Li deposition, variations in active Li, and spatial heterogeneities are evident, emphasizing 3D non-uniformities. Adapted from Reference 15.
instabilities (e.g., fracture widening) and electrochemical forces (e.g., Li penetration) interact at solid-solid interfaces in anode-free solid-state batteries.
In situ neutron and X-ray µCT offer non-destructive tools for investigating the physical and chemical mechanisms of battery degradation in LIBs and SSBs. By leveraging the complementary strengths of neutrons and X-rays, simultaneous neutron and X-ray tomography allows material-specific identification. This potential is highlighted in two case studies: imaging graphite electrode degradation in fast-charged LIBs and probing electro-chemomechanical instabilities at solid-state interfaces in anode-free SSBs. Neutron µCT, with its sensitivity to Li, is particularly effective for visualizing 3D Li morphologies and spatial heterogeneities, shedding light on their role in battery degradation.
Recent advances in neutron detector technology have improved spatial resolution to ~10 µm in state-of-the-art instruments like the NeXT beamline at ILL17 and the neutron microscope at PSI.19 Other facilities, such as the BT-2 beamline at NCNR (~15–20 µm),16 ICON at PSI (~10–15 µm),18 and CG-1D at Oak Ridge National Laboratory (~25–30 µm),20 provide unique capabilities for studying emerging battery chemistries, including solid-state, anode-free, Li-metal, LiSi, and Li-CO₂ systems.
A key challenge remains designing electrochemical cells that are compatible with both neutron and X-ray 3D imaging without compromising electrochemical performance. These cells should also enable high-throughput tomography, as neutron-µCT scans typically require prolonged acquisition times, often exceeding 10 hours, depending on reactor flux, sample composition, and the desired resolution. Lower flux sources or higher resolution demands can significantly extend the scanning duration. Rigorous electrochemical testing of imaging-compatible cells is crucial, and researchers should report comparisons with standard cells in their publications. Looking ahead, in situ 3D neutron and X-ray diagnostics are poised to play a pivotal role in uncovering fundamental degradation mechanisms, driving the design of low-cost, safe, long-lasting, and fastcharging batteries to support the widespread adoption of electrified transportation.
The author gratefully acknowledges financial support from the Stanford Office of the Vice Provost for Graduate Education through the DARE Doctoral Fellowship (2020–2022), the Electrochemical Society’s Edward G. Weston Summer Fellowship (2022), the Colin Garfield Fink Research Fellowship (2024), the Schlumberger Foundation’s Faculty for the Future Fellowship (2018–2023), and Princeton University’s Presidential Fellowship (2023–2026).
©The Electrochemical Society. DOI: 10.1149/2.F03252IF
Maha Yusuf, Presidential Postdoctoral Research Fellow, Princeton University Education: Bachelor’s in chemical engineering (National University of Sciences and Technology, Islamabad); Study abroad on a US Department of State Scholarship (University of Mississippi); Master’s and PhD in chemical engineering (Stanford University).
Research Interests: My work advances the fundamental understanding of physicochemical mechanisms that govern the performance and failure of electrochemical devices critical for decarbonization. At Stanford University and the SLAC National Accelerator Laboratory, I uncovered key insights into Li plating behavior, demonstrating how 3D morphologies and spatial heterogeneities accelerate Liion battery degradation during extreme fast charging. Currently, as a Presidential Fellow at Princeton, I investigate how mechanical
Honors & Awards: ECS Energy Technology Division Graduate Student Award (2024), ECS Colin Garfield Fink Fellowship (2024), ECS Edward G. Weston Fellowship (2022), American Chemical Society CAS Future Leader Award (2022), and Stanford Distinguished Student Energy Lecturer Award (2020). https://orcid.org/0000-0001-7908-2915
1. Net Zero Coalition | United Nations: https://www.un.org/en/ climatechange/net-zero-coalition
2. United States Environmental Protection Agency: https://www. epa.gov/greenvehicles/fast-facts-transportation-greenhousegas-emissions
3. C. Y. Wang, T. Liu, X. G. Yang, et al., Nature, 611(7936), 485 (2022).
4. J. S. Edge, S. O’Kane, R. Prosser, et al., PCCP, 23(14), 8200 (2021).
5. P. P. Paul, P. P., B. R. Chen, S. A. Langevin, Energy Storage Mater, 45, 969 (2022).
6. T. Krauskopf, F. H. Richter, W. G. Zeier, and J. Janek, Chem Rev, 120(15), 7745 (2020).
7. E. Kazyak, M. J. Wang, K. Lee, K., et al., Matter, 5(11), 3912 (2022).
8. Q. Liu, C. Du, B. Shen, et al., RSC Adv, 6(91), 88683 (2016).
9. P. P. Paul, E. J. McShane, A. M. Colclasure, et al., Adv Energy Mater, 11(17), 2100372 (2021).
10. S. Lou, Z. Yu, Q. Liu, et al., Chem, 6(9), 2199 (2020).
11. Neutron Scattering Lengths: https://www.nist.gov/ncnr/neutronscattering-lengths-list
12. XCOM calculator: https://physics.nist.gov/PhysRefData/Xcom/ html/xcom1.html.
13. M. Yusuf, J. M. LaManna, P. P. Paul, et al., Cell Rep Phys Sci, 3(11) (2022).
14. M. Yusuf, ECS Interface, 33(4), 36 (2024).
15. M. Yusuf, A. Kaestner, M. Wied, et al, ChemRxiv (2024).
16. J. M. LaManna, D. S. Hussey, V. H. DiStefano, NIST NeXT: a system for truly simultaneous neutron and x-ray tomography. In Hard X-Ray, Gamma-Ray, and Neutron Detector Physics XXII (Vol. 11494, pp. 61-71). SPIE (2020).
17. A. Tengattini, N. Lenoir, E. Andò, et al., NIM A, 968, 163939 (2020).
18. A. P. Kaestner, S. Hartmann, G. Kühne, et al., NIM A, 659(1), 387 (2011).
19. P. Trtik, J. Hovind, C. Grünzweig, et al., Phys Proc, 69, 169 (2015).
20. L. Santodonato, H. Bilheux, B. Bailey, et al., Phys Proc, 69, 104 (2015).
by Noor Ul Hassan and Shaun M. Alia
With the growth of renewable energy sources, hydrogen is attracting significant attention worldwide as an effective medium for energy storage. The US Department of Energy (DOE) has identified key strategic priorities for hydrogen production through the Hydrogen Shot initiative and the US National Clean Hydrogen Strategy and roadmap.1 The DOE aims to address these key challenges for the development of affordable clean hydrogen technologies with an interim hydrogen production cost target of $2/kg by 2026 and the Hydrogen Shot target of $1/ kg by 2031.2 This is an ambitious goal, especially considering the desire to simultaneously decarbonize hydrogen production. “Green hydrogen” is currently produced primarily by water electrolysis in which water is split into hydrogen and oxygen using power from lowcarbon energy sources such as wind, solar, and nuclear. However, current costs for hydrogen production by electrolysis are higher than $5/kg-H2 3 and multiple breakthroughs are required to advance water electrolysis to meet the Hydrogen Shot cost target.
Among the low temperature water electrolysis technologies, anion exchange membrane water electrolyzers (AEMWEs) have recently emerged as a promising competitor to traditional alkaline water electrolyzers (AWEs) and proton exchange membrane electrolyzers (PEMELs) due to their potential stack cost reduction in various cell components.4-5 Favorable aspects of AEMWEs include the use of PGM-free electrocatalysts as well as low-cost membranes, bipolar plates (BPs), and porous transport layers while offering high voltage efficiency and durability.
The anion exchange membrane (AEM) is the core of AEMWE cells. AEMs typically comprise an organic polymer backbone with side chains containing positively charged functional groups, which allow the AEM to move anions (typically OH-) from the cathode to the anode and also help transport water. Therefore, ion exchange capacity (IEC), water uptake, and conductivity are commonly measured AEM properties and contribute to AEMWE efficiency. The AEM also acts as a separator for produced hydrogen/oxygen gases, and it is important from both a safety and efficiency perspective to keep the gas crossover as low as possible.
During cell assembly, the AEM is sandwiched between two electrodes that are responsible for hydrogen and oxygen evolution reactions. The anode is generally composed of active materials such as Ir, Ni, Fe, and Co oxides (or combinations thereof) deposited onto a Ni or stainless-steel (SS) porous transport layer (PTL) using an ionomeric binder. An important current area of research is on avoiding ionomer degradation and catalyst layer loss when electrodepositing layers onto the PTL.16 Fig. 1(A) shows a microscopic image of an anode electrode. The anode composition includes Co3O4 nanoparticles deposited onto the SS PTL (which consists of Cr, Fe, Ni, and Mo) using an anion exchange ionomer as a binder. PTL structural properties such as porosity, fiber size, and pore diameter, as well as the catalyst layer structure, are important for efficient catalyst utilization and the transport of water/oxygen gas.6
Similarly, the cathode typically comprises Pt, Ni, or Mo (or alloys thereof) deposited on a carbon paper PTL using an ionomeric binder. In alkaline media, HER usually exhibits poor activity and stability due to the sluggish water cleavage step involved.19 Therefore, the
cathode requires the participation of electrocatalysts to facilitate the kinetics, transport OH– ions, and allow quick H2 escape for efficient device performance.
On the outside of the PTLs, flow fields/BPs are made of graphite or a metal (e.g., Ni, SS) and are responsible for feeding reactants and removing products, as well as providing mechanical compression to the membrane-electrode-assembly (MEA). Supporting electrolytes such as potassium hydroxide (KOH), potassium carbonate (K2CO3) or bicarbonate (KHCO3) are often used to increase the ionic conductivity of the catalyst layers, and possibly also the AEM, enhancing system efficiency. Water and electrical energy are provided as reactants to AEMWEs. The water then dissociates into positively charged hydrogen ions and hydroxide ions (OH–) at the cathode. While the H+ ions combine to produce hydrogen gas, the OH- ions migrate through the AEM, where they are oxidized to produce water and oxygen gas.
Apart from individual MEA material properties, their integration into AEMWE devices may further complicate the system, where understanding different types of losses is crucial for efficiency and durability enhancement. At its simplest, the total cell voltage required to operate the AEMWE device consists of the reversible thermodynamic cell voltage with added kinetic, ohmic, and mass transport overpotentials.
Where, ���������� is the total cell voltage, ��thermo the reversible thermodynamic potential (1.23 V at STP), ��kinetic the kinetic overpotential, ��ohmic the ohmic overpotential, and ������������������ the mass transport overpotential. While we expect the residual to be mass transport in part, the CLR and other losses that we can’t identify are grouped here as well. Kinetic losses are dictated by catalyst activity, its loading, utilization and electrochemical surface area. Moreover, interfaces (membrane / electrode or PTL) impact site access and cell kinetics. Ohmic losses are mainly caused by the ionic resistance of the membrane, the electrical resistance of catalyst layers, PTLs, BPs and interfacial resistances between the components. Mass transport is caused by the resistance of fluid transport across the catalyst layer and PTL plus ionic / electronic transport in the catalyst layers. Fluid transport is largely hindered by bubble accumulation within the catalyst layers and PTLs. The operating objective for AEMWEs is to maximize cell efficiency by reducing underlying kinetic, ohmic, and transport losses.
A current-voltage relationship, often called a polarization curve, is used to evaluate the performance of electrochemical systems. Potentiostatic or galvanostatic modes are typically used with either anodic (low to high voltage/current) or cathodic (high to low voltage/ current) scans. Fig. 1(B) (black curve) shows a polarization curve collected using a potentiostatic anodic scan. This polarization curve highlights kinetic and ohmic overpotential regions with no obvious transport losses. However, further characterization is needed to evaluate and understand those overpotential contributions, such as electrochemical impedance spectroscopy (EIS).
(continued on next page)
EIS is an essential tool for measuring different resistances in the cell and correlating losses caused by individual MEA components. EIS can be collected in either potentiostatic or galvanostatic modes. Fig. 1(C) shows Nyquist plots collected at various voltages (1.5 V to 2.0 V). From this, the high-frequency resistance (HFR) and charge transfer resistance can be obtained relatively easily. The HFR is used to calculate the cell’s ohmic resistance, with the main contribution being from the membrane ionic resistance. For example, decreasing the membrane thickness or adding a supporting electrolyte (1 M KOH) into the feed water would result in a decreased HFR, as reported in a comprehensive study on the effects of the feed electrolyte.7 Additionally, the charge transfer resistance helps to understand the contributions from the electrodes and transport phenomenon (in the low frequency region). Charge transfer resistance is inversely proportional to exchange current density, and lower charge transfer resistance means better kinetics, higher efficiency, and lower operating voltage at a fixed current (as shown in Fig. 1(C)). Moreover, fitting EIS data using a Randles cell equivalent circuit model or distribution of relaxation times provides the double layer capacitance (CDL) and insights into the electrochemical surface area. The CDL is also often determined via cyclic voltammetry (CV) collected in the non-faradaic region in a separate experiment, typically below 0.6 V vs RHE. In this region, no charge transfer reactions occur, and the observed current is assumed to be only from the charging of the double layer. This is a useful technique for quickly visualizing the electrochemical surface Hassan and Alia (continued from
1. (A) SEM image of Co3O4 deposited on SS PTL used as anode for oxygen evolution reaction, (B) Polarization curve with voltage breakdown for a single AEMWE cell, (C). Nyquist plots showing impedance spectra collected at various voltage holds. [Anode: Co3O4 was deposited on SS PTL, cathode: PtC deposited on carbon paper, AEM: PiperIon (80 um), Ionomer: PiperIon 5% dispersion, 1.0 M KOH electrolyte feed, cell temperature:80 oC, ambient pressure].
area and subsequent evaluation of CDL, but can be complicated by redox transitions and oxide contributions. Furthermore, EIS collected in a non-faradaic region close to the open circuit voltage is often used to evaluate catalyst layer resistance, which is discussed in a separate section below.
Understanding different types of voltage losses and linking them to individual MEA components is essential to developing mitigation strategies. Polarization and EIS measurements are used to evaluate the ohmic, kinetic, and residual (transport) contributions to overall cell voltage. Volk et al.8 provide a detailed discussion on the catalytic activity and stability of PGM-free electrocatalysts for the oxygen evolution reaction in AEMWEs using voltage breakdown analysis. For calculating the ohmic contribution, EIS is utilized to determine the HFR at each voltage tested in the potentiostatic polarization curve. The HFR multiplied by the cell current is assumed to be equal to the ohmic contributions to the overpotential.
The orange region in Fig. 1(B) shows the ohmic contribution to overpotential, which increases linearly with the current density. Kinetic losses are dictated by catalyst activity, loading, and utilization. Kinetic contributions to overpotential are determined by finding the iR-free potential (determined by ��iR-free = ��cell - ηohmic). Then, the linear region from a Tafel plot, prepared by plotting the calculated iR-free potential versus the log of the cell current (generally cell voltage below
1.50 V), is determined, and a Tafel slope is extracted. The equation for this Tafel slope is then used to find the kinetic voltage (����������������) at each cell current density. Kinetic contributions to overpotential can be obtained as follows:
The green region in Fig. 1(B) shows kinetic contribution to overpotentials. Finally, the residual contributions to overpotential are the remaining overpotential when ohmic and kinetic contributions are subtracted. Residual overpotential is also often called transport loss and is assumed to be caused by resistance to fluid transport across the catalyst layer and PTL plus ionic transport in the catalyst layers. The white + purple region in Fig. 1(B) shows the residual/transport contribution to the overpotentials. Residual loss can be further deconvoluted to extract/separate catalyst layer resistance (purple region) as discussed below.
High ionic or electronic resistances in the catalyst layer can lead to poor catalyst utilization, increased voltage losses, and high local overpotentials that can accelerate performance degradation. Therefore, understanding the role of catalyst layer resistance in AEMWEs is essential to develop strategies to avoid these losses. The catalyst layer resistance can be calculated using a transmission line model for porous electrodes, or approximated using following equation:
The HFR in the equation above was defined from the EIS and Rp is the intercept of the constant phase region at low frequencies. A typical process for calculating the catalyst layer resistance utilizes EIS at a non-faradaic condition (typically collected at 1.25 V). A detailed discussion of catalyst layer resistance has recently been reported by Padgett et al.9 The purple region in Fig. 1(B) shows the contribution to overpotentials caused by catalyst layer resistance.
Gas crossover is a phenomenon in which hydrogen from the cathode side diffuses through the membrane and mixes with oxygen at the anode side, reducing hydrogen efficiency loss and posing a safety hazard. The most common method for measuring H2 crossover is by using the limiting current, where hydrogen permeability is measured by the chronoamperometry technique in a fuel cell hardware feeding H2 on the anode and N2 on the cathode (fully humidified).10–12 These techniques may not work as well for PEMWEs as for AEMWEs due to the variety of catalyst materials and membrane properties employed. A novel procedure has been reported by Wrubel et al.13 for measuring the gas crossover using gas chromatography (GC). Understanding gas crossover is even more complex in AEMWEs due to the variety of electrolytes and operating configurations (dry vs wet cathode). Gas crossover can be reduced by using thicker membranes; however, this strategy will result in higher ohmic losses.
This final section features a couple of literature reports that demonstrate the tremendous recent progress in AEMWE performance. Klingenhof et al.14 recently reported high-performance AEMWEs (achieving >5 A cm−2 at around 2.2 V) using NiX (X = Fe,Co,Mn) catalyst-coated membranes, which demonstrate AEMWE
performance at values rivaling state-of-the-art PEMWE cell technology. Exceeding the limits of ultrahigh current density, Zheng et al.15 uses robust AEM, ionomer, catalyst, and porous transport layer materials, demonstrating successful operation of an AEMWE at 10 A/cm2 over 800 hours while remaining at ~2.3 V. Various chemistries of polymer membrane / ionomer, noble metal free electrocatalysts, and porous transport layers have been reported that show high performance and durability approaching that of well-established PEMWEs.
However, the majority of the demonstrations are lab scale, use materials of limited commercial availability, and have operational challenges. For instance, anion exchange ionomer oxidation,16,17 catalyst corrosion, dissolution, poisoning, and delamination18 during AEMWE operation have been reported in the literature. Therefore, there are several challenges to overcome in developing durable systems using commercial materials. The following areas should be emphasized to advance AEMWE technology for industrialscale deployment: (1) high performing, durable, and robust AEMs with negligible hydrogen crossover, especially under pressurized operation, and ionomers that are more resistant to anode oxidation; (2) high-performing PGM-free HER catalysts that can replace Pt-based catalysts (although industry feedback shows green signal with current Pt-loadings instead of compromising on performance); (3) PGMfree OER catalysts with improved activity and proven durability; (4) understanding ionomer-catalysts interactions and developing OER/HER catalysts that don’t oxidize the ionomer or membrane interface; (5) materials integration challenges and understanding of in-situ MEA degradation mechanisms; (6) and further exploration of corrosion mechanisms, which is essential for components such as the PTLs and bipolar plates as well as at their interfaces. Better understanding of crossover phenomena with mitigation strategies that do not compromise efficiency is also needed.
We would like to thank Prof. Dr. William Mustain (University of South Carolina), Prof. Dr. Paul A Kohl (Georgia Institute of Technology), and Dr. Bryan Pivovar (National Renewable Energy Laboratory) for their mentorship and guidance.
This work was authored by the National Renewable Energy Laboratory, operated by Alliance for Sustainable Energy, LLC, for the US Department of Energy (DOE) under Contract No. DE-AC3608GO28308. The views expressed in the article do not necessarily represent the views of the DOE or the US Government. The US Government retains and the publisher, by accepting the article for publication, acknowledges that the US Government retains a nonexclusive, paid-up, irrevocable, worldwide license to publish or reproduce the published form of this work, or allow others to do so, for US Government purposes.
©The Electrochemical Society. DOI: 10.1149/2.F04252IF
Noor Ul Hassan, Postdoctoral Researcher, National Renewable Energy Laboratory
Education: PhD, Chemical Engineering (University of South Carolina)
Research: Dr. Ul Hassan’s research in hydrogen and fuel cells technology explores the science behind anion/proton exchange membrane fuel cells and water electrolyzers, specifically focusing on improving the performance and durability of such systems. He has investigated various componentlevel aspects of fuel cell and water electrolyzer systems, such as electrode engineering/optimization, ionomeric binders, catalyst ink rheology, and porous transport layers.
(continued on next page)
Hassan and Alia
(continued from previous page)
Shaun M. Alia, Researcher V, National Renewable Energy Laboratory
Education: PhD in Chemical and Environmental Engineering (University of California, Riverside)
Research: Shaun Alia has worked in several areas related to electrochemical energy conversion and storage, including proton and anion exchange membrane-based electrolyzers and fuel cells, direct methanol fuel cells, capacitors, and batteries. His current research involves understanding electrochemical and degradation processes, component development, and materials integration and optimization. Within HydroGEN, a part of the US Department of Energy’s Energy Materials network, Alia has been involved in low-temperature electrolysis through NREL capabilities in materials development and ex- and in-situ characterization. He is further active within insitu durability, diagnostics, and accelerated stress test development for H2@Scale and H2NEW.
Scott Cushing, Assistant Professor of Chemistry, Caltech
Education: BS in Physics, emphasis in Material Science and Chemistry and PhD in Physics, under Nick Wu and Alan Bristow (West Virginia University).
Research Interests: With a multidisciplinary background spanning Chemistry, Materials Science, and Physics, his research focuses on the creation of new scientific instrumentation that can translate quantum phenomena to practical devices and applications. The Cushing lab is currently pioneering the use of attosecond x-ray, time-resolved TEM-EELS, and ultrafast beams of entangled photons for a range of microscopy and spectroscopy applications.
Work Experience: Past appointments include Dept. of Energy EERE Postdoctoral Fellow, Prof. Stephen Leone Group University of California, Berkeley with a Co-Appointment at Lawrence Berkeley National Laboratory. Currently Senior Research Advisor for Pacific Integrated (PI) Energy, San Diego, CA.
Pubs & Patents: >60 publications, 3 patents, h-index >30, cited ~8,000 times
Awards: 2022 Cottrell Scholar, 2022 Shirley Malcom Prize for Excellence in Mentoring, 2019–2021 Young Investigator awards for DOE, AFOSR, ACS, and Rose Hill Foundation.
Work with ECS: ETD Division: assist with organizing and chairing symposium. Member for >15 years.
Website: cushinglab.caltech.edu https://orcid.org/0000-0003-3538-2259
1. US National Clean Hydrogen Strategy and Roadmap, US Department of Energy (2023)
2. Hydrogen and Fuel Cell Technologies Office Multi-Year Program Plan, US department of energy (2024).
3. Hydrogen and Fuel Cell Technologies Office. Hydrogen Shot. US department of energy, (2021)
4. I. Vincent, D. Bessarabov, Renewable Sustainable Energy Rev, 81(2), 1690 (2018).
5. K. Ayers, N. Danilovic, R. Ouimet, et al., Annu Rev Chem Biomol Eng, 10, 219 (2019).
6. N. Ul Hassan, E. Motyka, J. Kweder, et al., J Power Sources, 555, 232371 (2023).
7. N. Ul Hassan, Y. Zheng, P. A. Kohl, and W. E. Mustain, J Electrochem Soc, 169, 044526 (2022).
8. E. K. Volk, S. Kwon, and S. M. Alia, J Electrochem So., 170, 064506 (2023).
9. E. Padgett, G. Bender, A. Haug, et al., ACS Appl Energy Mater, 6417 (2020).
10. M. Wang, J. H. Park, S. Kabir, et al., J Power Sources, 228344 (2019).
11. J. Wu, X. Z. Yuan, J. J. Martin, et al., J Power Sources, 84, 104 (2008).
12. J. A. Wrubel, C. Milleville, E. Klein, et al., Int J of Hydrogen Energy, 66, 28244 (2022).
13. M. Klingenhof, H. Trzesniowski, S. Koch, et al., Nature Catalysis, 1238 (2024).
14. Y. Zheng, W. Ma, A. Serban, A. Allushi, X. Hu, Angew Chemie Int Ed, e202413698 (2025).
15. A. W. Tricker, T. Y. Ertugrul, J. K. Lee, et al., Adv Energy Mater, 14(9), 2303629 (2023).
16. G. A. Lindquist, J. C. Gaitor, W. L. Thompson, et al., Energy Environ Sci, 10, 4373 (2023).
17. R. R. Raja Sulaiman, W. Y. Wong, and K. S. Loh, Int J. Energy Res, 1 (2021).
18. V. R. Stamenkovic, D. Strmcnik, P. P. Lopes, and N. M. Markovic, Nat Mater, 16, 57 (2017).
Electrolyte Design for Wide-Temperature Operation in Sodium Metal Batteries
This study addresses the challenges of electrolyte design in sodium metal batteries (SMBs), a promising alternative to lithiumbased storage. Conventional SMB electrolytes suffer from flammability, low stability, and poor performance across wide temperatures. The authors propose a low-concentration electrolyte (L-PFT) containing 0.3 M NaPF6 in a propylene carbonate (PC), fluoroethylene carbonate (FEC), and 1,1,2,2-tetrafluoroethyl 2,2,3,3-tetrafluoropropyl ether (TTE) mixture to enhance safety and efficiency. Electrochemical tests reveal that L-PFT exhibits high cycling stability, achieving 92.8% capacity retention at room temperature, 98.5% at 55°C, and superior low-temperature performance at −20°C. Raman spectroscopy and molecular dynamics simulations confirm a solvation structure with a higher fraction of solvent-separated ion pairs (SSIP), which improves ion diffusion and wetting properties. Despite slightly reduced initial performance compared to the standard 1.0 M NaPF6 electrolyte (S-PFT), L-PFT minimizes electrolyte decomposition at high temperatures, ensuring better long-term stability. Impedance spectroscopy and X-ray photoelectron spectroscopy indicate that L-PFT forms a more flexible and organic-rich interphase, reducing interfacial resistance and improving sodium deposition behavior. This work demonstrates that low-concentration electrolytes can maintain competitive performance while reducing costs and safety risks. By optimizing solvation structures, it provides a new approach to SMB electrolyte design.
From: Q. Zhang, X. Wang, et al., J Electrochem Soc, 172, 010501 (2025).
Electroreduction of Magnetite for Steel Production from Ceramic Suspensions Containing Metallic Iron Green steelmaking is an emerging goal for reducing energy consumption and carbon emissions. Electrochemical reduction of iron oxides in alkaline electrolytes has the potential to enable low temperature (<100 °C) electrowinning, but the effect of impurities is a significant challenge. A team from the Aveiro Institute of Materials in Portugal recently reported the effect of metallic iron impurity in magnetite (Fe3O4) suspensions in 10 M NaOH. This is important because the industrial byproduct “mill scale” could be used as a sustainable raw material for this type of iron recovery. Mill scale contains magnetite and other iron oxides, but also up to 10 wt% metallic iron. Using a cyclic voltammetry study in a three-electrode cell, they found that the presence of metallic iron in the electrolyte suspension significantly impacted electroreduction, leading to reduced faradaic efficiency and altered microstructures in the electrodeposited iron. The inhibitor effect of metallic iron impurity
may be due to sedimentation and entrapment of magnetite by larger iron particles or by magnetic interactions. Insights such as these are valuable in establishing electrochemical deposition as an innovative approach for recycling iron residues.
From: A. Fumo, D. V. Lopes, et al., J Electrochem Soc, 172, 012506 (2025).
A non-polarizable reference electrode is valuable to deduce electrode phenomena, especially in an operating device. In the case of a lithium ion battery, while the nonpolarizability of Li/Li+ couple is good, the complications with SEI formation make it a non-ideal choice. A team from General Motors addressed both the material stability issues and the placement through their novel reference electrode, a porous separator that is coated with a thin film of the cathode material lithium iron phosphate (LFP). With this LFP-coated porous separator, the critical problem of reference electrode placement is solved as well. The experimental procedure included the formation, current interrupt test for high frequency impedance measurement, electrochemical kinetic measurements, and fast charging lithium plating boundary condition testing. Control studies were done with and without the reference electrode. The team found the effect of their reference electrode to be minimal and susceptible to compensation. Through the application of the reference electrode, authors were able to identify a robust charging protocol. They also demonstrated that such a reference electrode can be used in large format cells, through the entire range of practical charging and discharging regime.
From: B. J. Koch, J. Gao, A. Zhang, et al., J Electrochem Soc, 172, 013507 (2025).
PCR-Free Self-Calibrated Ratiometric Electrochemical Genosensor Utilizing a Dual-Signal Amplification Approach for Genomic Detection of Mycobacterium Tuberculosis
As estimated by the World Health Organization, about a quarter of the global population has been infected with tuberculosis (TB). The standard method to diagnose TB is a time-consuming mycobacterial culture procedure. Extensive efforts have been made to develop alternative or supplemental methods based on the detection of nucleic acids. To achieve enough sensitivity, PCR is an essential step needed for most of these methods. In a recent report, researchers from the National Science and Technology Development Agency of Thailand described a PCR-free electrochemical sensor with additional features for mycobacterial DNA detection. The authors used silver nanoparticles to modify a screen-printed carbon electrode. Methylene blue-labeled, single-strand DNA probes were then selfassembled onto the electrode surface. Both
methylene blue and the silver nanoparticle displayed redox peaks in a voltametric scan. Hybridization of the bacterial DNA to the probes on the electrode led to a decrease in both peak currents. Interestingly, the authors found that the ratio of the two peak currents could serve as a more sensitive, accurate and robust sensor signal than either of the individual peak currents. Under optimized conditions, the sensor achieved a broad detection range of 3.5 fM–35 nM with a detection limit of 1.59 fM for non-amplified mycobacterial DNA.
From: S. Bunyarataphan and T. Prammananan, J Electrochem Soc, 172, 017520 (2025).
Adsorption of Copper and Chromium Mixed Solution on N-Doped BiomassActivated Carbon
Heavy metal ions such as Cu(II) and Cr(VI) are among the most concerning contaminants in wastewater due to their toxicity, persistence in the environment, and tendency to bioaccumulate in living organisms. Activated carbon is commonly used as for heavy metal ion adsorption in wastewater treatment systems. However, unmodified activated carbon typically exhibits relatively low adsorption capacity, slow adsorption kinetics, and limited selectivity. In this study two different methods were investigated to improve the performance of activated carbons prepared from waste biomass. Initially, sawdust was carbonized via annealing under inert gas conditions and acid treatment. The resulting activated carbons were further modified by nitrogen doping via ultrasonic and redox treatments. The researchers observed that the adsorption performance of the N-doped activated carbon prepared by the redox method demonstrated improved performance compared to samples prepared by ultrasonic treatment, in terms of adsorption capacity, adsorption strength, and equilibrium adsorption capacity. This study highlights the application of modified activated carbons derived from biomass as adsorbent materials to remove heavy metal ions (Cu(II) and Cr(VI)) from wastewater.
From: X. Yuan, L. Gao, A. S. Chen, et al., J Solid State Sci Technol, 14, 011002 (2025).
Tech Highlights was prepared by Joshua Gallaway of Northeastern University, David McNulty of University of Limerick, Chao (Gilbert) Liu of Shell, Zenghe Liu of Abbott Diabetes Care, Chock Karuppaiah of Vetri Labs and Ohmium International, and Donald Pile of EnPower, Inc. Each article highlighted here is available free online. Go to the online version of Tech Highlights in each issue of Interface, and click on the article summary to take you to the full-text version of the article.
by Kang Xu, Y. Shirley Meng, Venkatasubramanian Viswanathan, William C. Chueh, and Qichao Hu
Around the end of 2022, artificial intelligence (AI), which had long been a topic of academic curiosity confined to only a rather small circle, truly captured public attention in a Sputnik-like manner. This sudden surge in interest was driven by the rapid advancement of generative AI and its accessibility to the general public, symbolized by the release of a large-language-model (LLM) ChatGPT by OpenAI, followed by many other LLMs from traditional big tech companies (Google, Microsoft, Meta) as well as newly emerged start-ups (Mistral, Perplexity, DeepSeek). These models demonstrated unprecedented capabilities in natural language processing, content creation, problem-solving, and reasoning, making AI an increasingly daily tool for businesses and individuals alike.
As AI is being integrated into various aspects of everyday life, it is also transforming scientific research by accelerating the discovery, analysis, and optimization of new materials and chemistries and understanding of physical phenomena. As a recognition of this AI revolution, the 2024 Nobel Prizes for Physics and Chemistry were respectively awarded to the discoverers of artificial neural networks that enable machine-learning (ML) simulation of the human brain in deep-learning, and the developers of AlphaFold, an AI-model capable of accurately predicting protein 3D structures. These revolutionary tools have opened a new horizon on all scientific fronts, especially materials discoveries and designs, because materials research has traditionally relied on time-consuming trial-and-error experiments and computationally expensive simulations, while AI-driven methods and ML models pretrained with vast literature data and knowledge bring unprecedented speed and accuracy. The powerful computational capability enabled by graphics-processingchips (GPUs) has further accelerated the AI/ML models in their ability to handle increasingly sophisticated tasks. We are witnessing early successes of the pharmaceutical industry in discovering new biomedicines and antibiotics of higher efficacy, enabled by algorithms analyzing chemical spaces at unprecedented scales previously thought impregnable to human efforts. Such techniques are being rapidly generalized by scientists in wide domains, such as battery materials, semi-conductors, catalysis, and nanomaterials.
In addition to chemistries and materials, the powerful capabilities of AI algorithms in pattern recognition are being widely leveraged in domains that include structural characterization of materials and quality control in manufacturing of cells, as well as life prediction and safety early-warning of battery packs, because they can accurately capture the correlations that humans might overlook, while simulating the human brain in recognizing signals of special significance in a noisy background. On the other hand, the natural language processing capabilities of AI also provide powerful assistance in handling, categorizing, and analyzing vast numbers of scientific works, which, in the battery and materials domains alone, have increased in an exponential manner since the 2010s, reaching ca. 600~800 articles per week in 2024, and making it essentially unfeasible for any individual researcher to fully keep up with the knowledge produced. Here, LLMs have proven to be highly efficient and accurate in processing of natural languages, storing and extracting chemical structure and performance data, and memorizing literature, making it retrievable not only for content-generation but also for creative inferencing.
Finally, the pairing of AI with the capabilities of robotic experimentation systems in autonomous labs offers an unprecedented opportunity. New materials can be designed and synthesized with automated processes, then machine learning guides a set of experiments incorporating live processing of experimental data and accelerating discovery. We have seen early success of such systems in batteries for eletrolytes (Clio) and electrode material design (A-lab). ML models can analyze vast datasets from previous experiments, scientific literature, and simulations to help decide the material design space for synthesis or testing. They can forecast how changes in chemical composition, structure, or processing conditions might affect a material’s performance (e.g., reactivity/stability, conductivity, or energy storage capacity), allowing autonomous labs to prioritize promising candidates, reducing lengthy trial-and-error. Here AI serves as the “brain” for self-driving lab equipment, coordinating robotic arms, sensors, and analytical tools. It can interpret data from X-ray diffraction, spectroscopy, or microscopy instantly, deciding the next experimental step without human intervention. This closed-loop system enables continuous operation and rapid iteration. A promising future would be that an autonomous lab powered by AI might design a new battery electrode by predicting a high-capacity formula, synthesizing it with robotic precision, testing its performance, and refining the recipe—all in a fraction of the time it would take a human team. This synergy of AI and automation is revolutionizing fields like energy, electronics, and biomedicine, pushing the boundaries of what’s possible in materials science
To capture how AI technology impacts the battery field, in this issue, we organized a series of feature articles contributed by experts who have applied AI technology to diverse aspects of batteries, from discovering new molecular structures in the vast molecular universe (Hannah et al.) to the optimization of electrolyte formulation mixtures (Sendek and Viswanathan); from the agentic automation of selfdriving laboratory design (Feng et al.) to battery lifetime prediction by physics-based aging datasets (Cui, Sun, and Chueh) and by transformer-based frameworks (Mai and Hu). With these articles we hope to expose the general audience of The Electrochemical Society to this exciting new horizon. It will be critically important for us to realize the impact of AI on our research and career environment, and we all should be encouraged to embrace the change while preparing for fierce competition in the world in front of us.
© The Electrochemical Society. DOI:10.1149/2.F06252IF
Kang Xu, CTO, SES AI
Education: BS Chemistry (Southwest University); PhD in Chemistry (Arizona State University)
Research Interests: Electrolytes, Interfaces, Interphases
Work Experience: ARL Fellow (Army Research Lab), Team leader (Army Research Lab), Lead Scientist of JCESR, a DOE Hub at Argonne National Lab, Chief Scientist (SES AI), Chief Technology Officer (SES AI)
(continued on next page)
(continued from previous page)
Work with ECS: ECS Fellow, MRS Fellow, ARL Fellow (emeritus), Member-at-Large of ECS Battery Division, 20+ yrs ECS Member
Awards: IBA Technology Award, ECS Research Award, and multiple US DoD R&D Awards and medals.
Pubs + Patents: 350+ publications, 30+ patents, 5 book chapters, and 1 textbook.
Website: SES AI
Work Experience: Co-Founder, Aionics Inc.
Work with ECS: Active member of Electrochemical Society and winner of San Francisco Section Daniel Cubicciotti Student Award, Herbert H. Uhlig Summer Fellowship
Awards: MIT Technology Review Innovators Under 35, Office of Naval Research (ONR) Young Investigator Award, Alfred P. Sloan Research Fellowship in Chemistry, National Science Foundation CAREER award and American Chemical Society Energy Storage Lectureship.
Pubs + Patents: 150+ publications, 50+ patents.
https://orcid.org/0000-0002-6946-8635
Y. Shirley Meng, Liew Family Professor, Pritzker School of Molecular Engineering, University of Chicago
Education: PhD in Advanced Materials for Micro & Nano Systems (Singapore-MIT Alliance) BS in Materials Science (Nanyang Technological University of Singapore)
Work History: Chief Scientist of the Argonne Collaborative Center for Energy Storage Science (ACCESS) Argonne National Laboratory. Director of the Energy Storage Research Alliance (ESRA), an innovation hub funded by US Department of Energy, Office of Science. Principal investigator of the research groupLaboratory for Energy Storage and Conversion (LESC), established at University of California San Diego. Zable Chair Professor in Energy Technologies at UC San Diego (2017–2022), Founder of the Sustainable Power and Energy Center (SPEC) in 2016.
Pubs + Patents: >320 peer-reviewed journal articles, two book chapters, and 12 issued patents
Honors & Awards: ACS Electrochemistry Award, ECS Battery Division Research Award, C3E technology and innovation award, RSC Faraday Medal, IBA Research Award, Blavatnik Finalist, ECS Tobias Young Investigator Award, and NSF CAREER Award. ECS Fellow, MRS Fellow, and AAAS Fellow https://orcid.org/0000-0001-8936-8845
Venkatasubramanian Viswanathan, Associate Professor of Aerospace Engineering, University of Michigan
Education: BTech in Mechanical Engineering (Indian Institute of Technology, Madras); PhD in Mechanical Engineering (Stanford University).
Research Interests: Energy storage, Machine learning, Materials informatics
Website: https://aionics.io; https://eeg.engin.umich.edu/ https://orcid.org/0000-0003-1060-5495
William C. Chueh, Associate Professor of Materials Science & Engineering, Stanford University
Education: BS, Applied Physics, MS, PhD, Materials Science (Caltech)
Research Interests: Electrochemistry, Redoxactive solids, Reactions and transport
Work Experience: Director, Stanford Precourt Institute for Energy; Director, SLAC-Stanford Battery Center
Awards: David A. Shirley Award, Friedrich Wilhelm Bessel Research Award, MRS Outstanding Young Investigator Award, BASF/Volkswagen Science Award Electrochemistry
Pubs + Patents: 130+ publications, 15+ patents.
Website: http://chuehlab.stanford.edu/ https://orcid.org/0000-0002-7066-3470
Qichao Hu, CEO, SES AI Corp.
Education: BS, MIT; PhD, Harvard University
Research Interests: Battery technology and manufacturing, Physics, materials and chemistry, Artificial intelligence
Work Experience: CEO and Founder of SES AI Corp.
Work with ECS: ECS Member Awards: MIT Technology Review Innovator Under 35 TR35; Forbes 30 Under 30; IALB 2024 Innovation Award
Website: www.ses.ai
by Daniel Hannah, Yumin Zhang, Xinyu Li, Dengpan Dong, Joah Han, Gyuleen Park, Hong Gan, Bin Liu, Kai Liu, Qichao Hu, and Kang Xu
Advanced batteries such as lithium-ion batteries (LIB) based on graphitic carbon or silicon as the anode, as well as lithium-metal batteries (LMB), are complicated systems consisting of multiple components operating at extreme potentials. While all these components must work with each other in a highly synchronized manner, the electrolyte is undoubtedly a key element exposed to the most severe electrochemical constraints, because it must interface with every other component therein.1 In the history of battery development, the electrolyte was often the last piece of the puzzle to be figured out, as evidenced by the discoveries of thionyl chloride in primary LMBs, ethylene carbonate in graphite-based LIBs, and fluoroethylene carbonate in silicon-based LIBs.2,3 More often than not, discovery of the right electrolyte holds back the successful deployment of a new battery chemistry.4
Due to its significance, the field of electrolyte science has experienced exponential growth over the past three decades. Researchers around the world have been trying to find the ideal electrolyte materials that can work with the new battery chemistries emerging on the horizon, including the most challenging battery chemistry in the universe, LMBs.5,6 It has been well-established that to stabilize electrolytes at such extreme potentials, interphasial chemistries formed by the decomposition products of electrolyte components (solvent molecules, salt anions or additives) play the dictating role. Although the interphase formation mechanism has been under thorough investigation, yielding general knowledge about the “solvation-interphase” correlation7 and the preferred interphasial chemistry of inorganic nature (Li2O, LiF, Li3N, etc.),8,9 it is still impossible to directly tailor or manufacture effective interphases. So far, the most effective way of stabilizing the thermodynamic nonequilibria in advanced batteries remains through the empirical design and test of new electrolyte components, especially solvent molecules. The process of such quest has mostly relied upon trial-and-error serendipity rather than rational design.
Examining the 30 years of electrolyte research from a high level, one would find that there were only about 700 unique chemicals that have been investigated, reported, or patented in the open literature. Among those, fewer than 100 molecules are estimated to have been actually used in commercial battery systems as either solvents or additives, while an even smaller number of salt anions (< 10) has been used.10 A more concerning fact is that most of these chemicals were selected from existing chemical inventories that were not invented for battery applications in the first place! In other words, battery scientists have been confined to a rather narrow chemical space.
Then, what is the real size of chemical space that could provide candidates for ideal electrolytes? The answer is infinity, which is the product of the element diversity defined by the Periodic Table and the unlimited number of atoms in the molecule.
However, if we place certain restrictions, then the size of the chemical space becomes finite. For example, if we define the elemental diversity to be C, N, O, S, X (where X is halogen, i.e., F, Cl, Br. I) and heavy atom count (HAC, i.e., atoms heavier than H) remains under 30, then the size of the chemical space is estimated to be ~1060 , 11 which is already larger than the total number of stars in the observable universe (1024) according to the James Webb Space Telescope as of 2024 (Fig. 1).12 If we further shrink the HAC to 17,
an accurate number is available: 1.66X1011, which was provided by a pharmaceutical structural database that encompassed almost all possible small organic molecules in an exhaustive manner.13 This number is still larger than the number of stars in our galaxy, the Milky Way. On the other hand, the size of the public databases (PubChem, Zinc-20, etc.) are up to 10.14,15
To assess how sparsely the molecular universe is populated by those molecules actually studied by human scientists, we randomly selected 1.4X107 molecules from Zinc-20 and calculated their machine-learned molecular fingerprints using a pre-trained contrastive learning model. We then reduced the 512-dimensional embeddings to two dimensions by applying the uniform manifold approximation and projection (UMAP) technique (Fig. 2).16 In this two-dimensional graph, these 14 million molecules are distributed in distinct clusters according to their structural similarity, as represented by the fractallike patterns in purple/blue, whose color gradient actually reflects the number of molecules at a single pixel in the form of heat map. On these 14 million molecules we applied an “electrolyte filter,” which consists of a sequence of constraints such as “No active proton,” “No S-S, N-N or O-O linkage,” as well as threshold values on both energy levels of the highest-occupied-molecular-orbital/lowest-unoccupiedmolecular-orbital (HOMO/LUMO) and maximum/minimum electrostatic potential (ESP min/max). About 104 molecules passed this electrolyte filter, which are plotted as yellow dots and overlain on the background of the 14 million molecules in Fig. 2. Apparently, most of the yellow clusters contain possible electrolyte material candidates.
Furthermore, we summarized all electrolyte solvents and additives reported for LMB in the most recent 10 years (2015~2024), which are around 400,10 and plotted them in green by overlaying them on the possible electrolyte material candidates. One can immediately visualize that in this 14 million molecule subset of the molecular universe, the number of molecules touched by human scientists only occupy an infinitesimally small fraction. Expanding this view to the scale of 109~1012 that truly represents the whole molecular universe, one could only imagine how sparse the universe has been populated, given that fewer than 103 unique chemicals have been investigated by the battery and materials communities. It is beyond doubt that in the unexplored region, there must be some molecules that can serve as ideal solvents, additives, or salt anions for the emerging battery chemistries. Such a vast unexplored universe awaits us to discover
Given the large size, the only feasible approach would be to identify and discover useful candidates from the astronomical number of 109~1012 in-silico computation (Fig. 3).
In the past decade, density-function-theory (DFT) has matured,17 allowing fast and accurate calculation of single molecular properties such as HOMO/LUMO energy levels, minimum and maximum electrostatic potential (ESP min/max), electron affinity (EA), and ionization potential (IP) that are closely correlated to the molecules’ electrochemical stability, as well as molecular polarizability, dipole moment, and electrostatic potentials that dictate the molecule’s capacity to solvate working ions (solubility) or to mix with other molecules (miscibility). More recently, acceleration of DFT from various AI and ML techniques makes it possible to process large numbers of molecules on scales comparable to the size of the
# stars in Milky Way: 1011
Electrolyte
Total # of organic (CNOSX) molecules under 30 atoms
# stars in observable universe: 1023 Molecules
Fig. 1. The number of known electrolyte materials that have been explored by human beings is dwarfed by our observable universe and by the molecular universe, whose immensity is enabled by the combinatorial power of elements. Shown at the upper left corner in “102 order” are popular solvent molecules (ethylene carbonate, propylene carbonate, dimethyl carbonate) and salt anions (hexafluorophosphate, bisfluorosulfonyl imide), while in “103” order are typical electrolyte solution compositions and structures consisting of lithium-ion (Li+) solvated by carbonate molecules.
molecular universe, an unthinkable task just a few years ago.18 The increasingly powerful graphic-processing-units (GPUs) provide parallelism and scalability that accelerate calculations on a single GPU by 20~40 times compared with an equivalent number of central processing units (CPUs), while further acceleration comes from ML techniques that speed up the most time-consuming step in DFT calculation, geometry optimization, by another ~102 times (Table I). Ultimately, an ML property prediction model trained on DFT data is able to infer a molecule’s property in milliseconds. Combining acceleration in both hardware and software, one could completely calculate the whole molecular universe within a reasonable
Fig. 2. A subset of the molecular universe (14 million molecules, shown as purple to light orange pixels according to population as indicated by the legend bar on the right) plotted in UMAP, overlain by those molecules that passed the “Electrolyte Constraints” (ca. 14,000, shown as yellow dots) and those reported as electrolyte materials (ca. 400, shown as green dots) in the open literature including articles and patents during the past decade (2015~2024). The popular solvent molecules EC, PC, and DMC are marked as references. Note that EC and PC (both are cyclic carbonates) are located away from DMC (a linear carbonate).
timeframe. For example, based on the estimate we made, it will take a CPU cluster equivalent to 500 GPUs approximately 8,000 years to completely calculate 1011 molecules, while a cluster of 500 H100 GPUs would need only 2 months.
The vast DFT database thus obtained will serve as a solid foundation for constructing and training AI/ML models. As of the first quarter of 2025, we have constructed such a DFT database of 108 molecules, which is the largest in the world.
Meanwhile, molecular dynamics (MD) simulations enabled by proper force-fields can accurately predict many properties on the electrolyte solution level, including ion transport, solvent miscibility, salt solubility, and solvation structure as well as extended solution structures (Fig. 3).19 The cost and acceleration of MD simulations is expected to be fully addressed in the years to come, thanks to the rapid advances in both hardware (GPUs) and software (algorithms and ML accelerations).
Fig. 3. Multi-scale in silico chemistry supports the search for ideal electrolytes in the vast molecular universe, from the quantum chemistry expression of a single molecule (electronic states, optimized geometries, etc.), to the solvation and cluster behaviors in an electrolyte formulation (ion-solvent interaction, ion transport, solubility, etc.), to the electrodic behaviors (interfacial structures, reactions and interphasial chemistries).
The last remaining piece of this in-silico approach is the prediction of interphasial behaviors and chemistries, where methodologies such as reaction networks and ab-initio MD (AIMD) appear to be promising, but are still not mature (Fig. 3). 21, 22
Table I. The calculation of basic molecular properties via DFT has been significantly accelerated by advances in both hardware and software.
Assisted by NVIDIA, we have been able to leverage the AI-accelerations from both software and hardware to perform DFT calculations on small organic molecules. Thus far, we have established a giant DFT database encompassing ca. 121 million molecules, a substantial percentage of the entire molecular universe. This DFT database provides a solid foundation for the construction and training of various AI/ML models as well as a battery-domain large language model, which was pretrained with published battery literature (Fig. 4). Using these AI techniques, our company, SES, has developed a mature pipeline, through which we can precisely calculate, predict, and identify molecules that are promising for diverse battery chemistries. Already we have been able to discover several new co-solvents and additives whose incorporation into electrolytes enables LMBs and Si-based LIBs to outperform most human-designed electrolyte formulations via conventional design
approaches. This preliminary success highlights the great potential of AI-generated electrolyte materials.
What we have demonstrated here is just a preliminary effort to design, discover, and generate new electrolyte materials from the vast molecular universe in a systematic and exhaustive manner, which fundamentally differs from the conventional human-based design and discovery of new materials. Given the astronomical size of the molecular universe, such an approach would have been unthinkable just a few years ago without today’s state-of-the-art AI/ML techniques. The immense DFT database constructed in this work, together with the gigantic literature database, and the ever-evolving AI/ML models, is expected to benefit materials discoveries in general, including but also beyond applications in the electrochemical and energy storage domains.
© The Electrochemical Society. DOI:10.1149/2.F07252IF
All of the authors are from SES AI.
Kang Xu, CTO, SES AI
Education: BS Chemistry (Southwest University); PhD in Chemistry (Arizona State University)
Research Interests: Electrolytes, Interfaces, Interphases
Work Experience: ARL Fellow (Army Research Lab), Team leader (Army Research Lab), Chief Scientist (SES AI), Chief Technology Officer (SES AI)
(continued on next page)
Work with ECS: ECS Fellow, MRS Fellow, ARL Fellow (emeritus), Member-at-Large of ECS Battery Division, 20+ yrs ECS Member Awards: IBA Technology Award, ECS Research Award Pubs + Patents: 350+ publications, 30+ patents. Website: SES AI
8. M. S. Kim, Z. Zhang, P. E. Rudnicki, et al., Nat Mater, 21, 445 (2022).
9. G. M. Hobold, C. Wang, K. Steinberg, Y. Li, and B. M. Gallant, Nat Energy, 9, 580 (2024)
10. SES Internal Literature Database
11. R. S. Bohacek, C. McMartin, and W. C. Guida, Med Res Rev, 16, 3 (1996)
12. NASA “Star Basics.”
https://orcid.org/0000-0002-6946-8635
1. K. Xu, Electrolytes, Interfaces and Interphases: Fundamentals and Applications in Batteries, RSC Press, London (2023).
2. K. Xu, Chem Rev 104, 4303 (2004).
3. K. Xu, Chem Rev 114, 11503 (2014).
4. M. Winter, B. Barnett, and K. Xu, Chem. Rev. 118, 11433 (2018)
5. X. Fan, L. Chen, O. Borodin, et al., Nat Nanotechnol, 13, 715 (2018).
6. S. C. Kim, S. T. Oyakhire, C. Athanitis, et al., PNAS, 120, e2214357120 (2023).
7. K. Xu, Y. Lam, S. S. Zhang, T. R. Jow, and T. B. Curtis, J Phys Chem C, 111, 7411 (2007).
13. L. Ruddigkeit, R. van Deursen, L. C. Blum, and J.-L. Reymond, J Chem Inf Model, 52, 2864 (2012).
14. Pubchem Database
15. Zinc-20 Database
16. L. McInnes, J. Healy, and J. Melville, ArXiv e-prints 1802.03426 (2018).
17. E. Engel and R. M. Dreizler, Density Function Theory: An Advanced Course, Springer (2011).
18. D. Anstine, R. Zubatyuk, and O. Isayev, ChemRxiv
19. C. Massobrio, J. Du, M. Bernasconi, and P. S. Salmon, Molecular Dynamics Simulations of Disordered Materials, Springer, 2015.
20. X. Xie, E. W. C. Spotte-Smith, M. Wen, H. D. Patel, S. M. Blau, and K. A. Persson, JACS, 143, 13245 (2021)
21. F. A. Soto, Y. Ma, J. M. Martinez, D. L. Hoz, J. M. Seminario, and P. B. Balbuena, Chem Mater, 27, 7990 (2015).
Stand out to employers and step into your dream career with a professionally written resume, cover letter, LinkedIn profile, and interview prep with Job Seeker Premium Services.
Access to in-depth data on industry outlook, wages, qualifications, and more.
Elevate your career with personalized guidance from an experienced and certified career coach.
Discover new niche job opportunities tailored to your industry; you won’t find these anywhere else!
by Austin D. Sendek and Venkatasubramanian Viswanathan
Electrolytes are fundamental to the function of batteries. They transport ions between the electrodes, while remaining inert in the harsh electrochemical environment the electrodes create. Designing electrolytes is a challenging, multidimensional problem that is unique from typical materials discovery or molecular design efforts: for one, a candidate electrolyte must satisfy tens of properties simultaneously. Second, the design space is immense. Electrolytes are chemical mixtures consisting of organic molecules and inorganic salt; thus, designing an electrolyte formulation for Li-ion batteries involves the combinatorial challenge of selecting a small number of organic molecules and a lithium-containing inorganic salt from a seemingly endless number of options, while accurately predicting all the key properties for each candidate formulation.
Rick Nason of Dalhousie University famously distinguished between complicated and complex challenges: the former are processes where the solution is difficult to reach but a rational map to the solution exists.1 For the latter, the solution is difficult to reach, and no map exists to get you there. Historically, electrolyte design has been treated as a complex problem and much of the innovation has occurred through happenstance and trial-and-error. However, the rapid scaling of data and high-performance computing are poised to completely disrupt the process of electrolyte design, turning it from a complex and intractable problem into a complicated but deterministic and rational problem. The key to reducing the complexity of electrolyte design is to recognize that, though electrolytes are made up of molecules, this is not a molecular design problem. The most important information exists in the cross-terms: how molecules interact with each other, and how they interact with the electrodes. In this article, we discuss how designing an electrolyte formulation in the computing age requires a re-thinking of AI tooling beyond the standard molecular property prediction paradigm.
AI- and machine learning-guided materials discovery efforts have demonstrated extraordinary progress in the last decade. During that time, we have gone through three generations of machine learning models for material- and molecular property prediction. The first generation of models leveraged handcrafted descriptors to build models that can predict the properties of single-component systems (i.e., single molecules or homogeneous crystals), an extension of the quantitative structure–activity relationship (QSPR) model framework developed nearly 60 years ago. The second generation involved utilizing graphs, a natural representation for bond connectivity within molecules, pairing them with two core machine learning advances from image processing—convolution and deep neural networks. This family of models, referred to as graph neural networks (GNNs), provided unprecedented accuracy and revolutionized the ability to screen large libraries of molecules. The third-generation models are molecular foundation models which are inspired by the structure of natural language and can leverage large swathes of unlabeled data (i.e., training on large databases of synthetically accessible molecules), using the model to build a more robust representation which can be fine-tuned for downstream property prediction.
We, the authors, have been working on this problem for the past decade—initially as academic scientists, and then as cofounders of Aionics, Inc., a company that uses AI and high-performance computing to develop high-performance electrolytes for diverse
electrochemical systems. Beginning in 2014, we have been publishing studies developing data-driven and machine learning models for predicting battery material properties: first for liquid electrolyte electrochemical stability (VV),2 then for solid-state Liion conductivity (ADS),3,4,5,6 dendrite suppression properties (VV),5,7 then for cathode properties including voltage, capacity, strain upon lithiation, and anion redox (VV, ADS)8,9,10 and electrolyte-electrode interface kinetics (ADS, 2024).11 Riding this wave of innovation, we founded Aionics in 2020 and began developing and commercializing machine learning models for battery performance. Working with manufacturers of materials, batteries, and battery-powered devices, descriptor-based methods were initially found to excel in cases where training data was extremely limited due to the expense of its generation—notably, for predicting the cycle life of a battery based on information about the electrolyte.12 A few years later, Viswanathan and colleagues developed a suite of GNN-based electrolyte property prediction models with superior performance to descriptor-based methods.13 These models were then commercialized by Aionics and deployed to the public under the Advanced Material Property Prediction Model framework.14 Looking forward to the future, we are now developing molecular foundation models. Early data show unequivocally that these big-data models can vastly outperform all other models on some key tasks.15
Despite the algorithmic advances taking place in both the pure AI/ ML community and the materials informatics community, electrolyte design remains a major challenge. Consider the simple case study below: we perform an electrolyte screening study of a known chemical space, and apply filtering criteria that would seem natural: (i) the molecule has to be stable in the battery (i.e., having appropriate values for the single component molecular levels—highest occupied molecular orbital and lowest unoccupied molecular orbital), and (ii) the molecule is a liquid at room temperature, having appropriate melting and boiling points. This simple set of screening criteria would eliminate ethylene carbonate, one of the most widely used electrolyte components. The issue is that the properties of individual molecules by themselves do not translate to electrolyte performance: information on the rest of the system is required to determine whether the formulation will remain liquid, or whether the cell will remain electrochemically stable. Thus, methods for incorporating information on the intricacies of the formulation and its interfaces with the electrodes are required to make significant advances in electrolyte formulation design—not just simply predicting the properties of molecules alone, no matter how sophisticated the algorithms or how accurate the predictions.
In general, electrolyte performance is a function of two sets of properties: innate properties of the formulation itself (ex situ) and properties of the formulation interacting with the chemical environment of the battery (in situ). The former are mixture properties, whereas the latter are interface properties. In both cases, molecular information alone is insufficient. Predicting the properties of mixtures requires information on the interactions of molecules with each other, while predicting the properties of interfaces requires information on the interactions of molecules with the electrode. No amount of information on individual molecules in vacuum can infer these cross-terms. Thus, a new class of ML models is required to supplement the simplified “molecule-only” framework.
(continued on next page)
(continued from previous page)
The canonical electrolyte design challenge involves adding a small amount of an additive to a baseline electrolyte formulation that otherwise works well under most of the required operating conditions. The goal of the additive is then to provide improved functionality in one or more axes (e.g., fast charging or lower temperature operation) while not affecting any of the other properties (e.g., cycle life). Thus, to minimize the overall cost of the formulation as well as to ensure that the majority of the properties of the baseline electrolyte formulation are carried forward, additive amounts are limited to 5% or less. The question then arises: how will the addition of a small amount of a new molecule change the overall properties of the electrolyte?
The simplest theory to predict the properties of a chemical mixture is linear mixing (i.e., the mixture properties are a molar-weighted average of the component properties). However, assuming linear mixing creates a big challenge here: adding a mere 5% additive will not substantially alter any linear property. As an example, assume that the melting point of the baseline electrolyte is -20oC and an additive with a melting point of -100oC is added at the 5% molar level. If the additive and baseline electrolyte linearly mix, then the new electrolyte formulation will have a melting point of -24oC, a very small decrease of 4oC in the overall electrolyte performance despite identifying an additive component that has an 80oC difference in melting point. The goal then is to find additives which break the linear mixing rule for their given formulation. Fortunately, many properties are not naturally linear, though the direction and magnitude of the nonlinearity is often not obvious a priori. For example, in Fig. 1 we plot the predicted melting point of a set of electrolyte mixtures based on a linear mixing model versus the experimentally observed melting points. The correlation is poor; the components clearly do not linearly mix.
The deviation from linear mixing is called the excess property, and thus, designing functional electrolyte formulations requires identification of molecular mixtures that have large excess properties. We must understand how the additive and formulation interact, and then design new additives that mix super-linearly for the beneficial
2. Differentiable battery electrolyte optimization with DiffMix and robotic experimentation. Reproduced with permission from ref. [13]. In (a)-(c), optimization for ionic conductivity is performed over various liquid electroyte compositions. In each optimization case, a batch of four trajectories has been simulated starting from the dot sign and ending at the cross sign. The white arrows are the gradient information obtained by auto-differentiating the ionic conductivity against compositions through DiffMix. In (d) the optimization curves of ionic conductivities in (c) is shown along the four trajectories, including both DiffMix results and the robotic experimentation results generated by Clio. Reproduced with permission from ref. [13].
properties while mixing sub-linearly for harmful properties. Information on the individual molecules alone is an incomplete piece of the puzzle.
Excess properties have been at the heart of chemical mixtures but have not been widely used in battery electrolyte design until recently. Excess functions have been formulated, with Redlish-Kister polynomials being one of the popular polynomial expansions that leverage the permutation invariance property of mixtures. Even for simple mixtures, there can be significant excess properties as molecular interaction amongst the components plays a critical role.
This situation highlights the need for incorporating additional physics into otherwise “straightforward” molecular ML models. Recently, we found a way to merge the two worlds: (i) machine learning advances involving graph convolution and (ii) mixture physics. The core insight involves learning the coefficients of mixture physics laws using machine learning, and then using these coefficients to add corrections from linear mixing. These coefficients are directly linked to the molecular identity of each of the constituent molecules, leading to mixturespecific corrections.
To realize this idea, the mixture physics laws need to be written in a differentiable programming framework, and then we can chain this to all the machine learning advances. This novel idea, which we called DiffMix,16 has been put to test to design a high conductivity electrolyte formulation. DiffMix was used to navigate a ternary chemical space and the predictions were then validated using experiments in our robotic test stand, Clio.17 The optimization trajectory measured from the robotic electrolyte setup, Clio, matched the predictions of the DiffMix model (Fig. 2).
The design of an electrolyte mixture is a balancing act among the various components of the mixture. Satisfying ex-situ property requirements cannot be accomplished by single molecular prediction models; instead a
Fig. 3. Importance of mixture and interface information for cycle life prediction. A model is trained on battery cycling data to learn to predict cycle life as a function of electrolyte composition. Proprietary interface features, proprietary mixture features, and open source mixture features are used. The features in the trained model are ranked by their importance. The single most important feature, by far, is an interface descriptor. We acknowledge Drs. Noushin Omidvar, Handong Ling, and Mohamed Elshazly for generating this image and data.
sophisticated machine learning stack must be employed that can handle mixture physics and the underlying molecular interactions amongst the mixture components.
Of equal or greater importance to the ex-situ properties of electrolytes are the in-situ properties of the electrolyte when added to a battery. In this case, the properties are determined by the interactions of the electrolyte components with themselves but also primarily with the electrode surface. Again, the properties of the individual molecules alone are insufficient.
Relevant in-situ properties include cycle life, charging rate capability, discharging rate capability, internal resistance, coulombic efficiency, round-trip efficiency, and self-extinguishing time.
All these properties require some knowledge of the interactions at the electrolyte-electrode interface, where electrolyte molecules may reversibly bond with the electrode surface. By extending concepts from computational surface science and catalysis, we published the first series of studies showing that the properties of batteries can be predicted directly from density functional theory (DFT) simulations of electrolyte molecules on electrode surfaces.18,19 In these simulations, electron transfer is simulated explicitly and breakdown of the molecules under the electrochemical and chemical forces of the interface can be observed.
Consider cycle life as an example property. The cycle life of a battery is a simple metric of battery life that arises as a multidimensional function of many phenomena happening in the bulk of the electrolyte solution and at the electrode interface. However convoluted the process may be, cycle life is a deterministic function of the electrolyte and electrodes and thus should be learnable with sufficient examples. In a recent internal study, we built a model to predict the cycle life of a Li-ion battery as a function of its electrolyte based on a small dataset of under 200 examples. The resulting model had a cross-validation error of 41 cycles, meaning the model is expected to predict the cycle life of a cell with an arbitrary (in-distribution) electrolyte to within 41 cycles on average.
It is instructive to look at the information required to make these predictions. We trained the model on a set of proprietary interfacebased descriptors, proprietary mixture descriptors, and open-source descriptors; the relative importance of these various descriptors is
shown in Fig. 3. Instead of focusing on the exact definition of each descriptor, the emphasis is on whether the descriptors depend on mixture alone or of the mixture and electrode interface. The single most important descriptor in the model contains interface information.
Once the appropriate descriptors are identified for in-situ property prediction, they must be extracted for the entire chemical space of interest to enable large-scale screening. However, this screening quickly becomes computationally intractable with traditional density functional theory methods. Surface calculations require many atoms, some of which are best simulated with Gaussian-type basis sets (i.e., the molecules) and some of which require plane wave basis sets (i.e., the surface). Many molecular orientations and possible surface adsorption sites exist. The computational cost to exhaustively study the interfacial effects of all molecules on all relevant surfaces can easily reach into the trillions of dollars. Thus, some machine learning acceleration is required.
Using the output of a large, expensive model to train a simpler model with minimal accuracy loss is a common task in machine learning, and it is a helpful strategy here. This approach of building computationally cheap surrogate models is sometimes referred to as distillation, pseudo-labeling, or transfer learning, depending on the context. In the case of interfacial features, we have found that simple models can learn these features efficiently from hundreds to thousands of examples and can perform inference on new systems with high fidelity and a 106-fold increase in speed over DFT, enabling screening across the entire space of molecules for any given interface.
Currently, there are approximately 1010 molecules that can be synthesized commercially, and there may be a total of over 1050 synthetically accessible molecules. There are many efforts underway to study the properties of these known and hypothetical molecules for varied kinds of applications, particularly within medicine. For electrolyte design, however, this is just the tip of the iceberg. Electrolytes cannot be designed successfully without considering the effects of combining these molecules into mixtures and explicitly considering their interfaces with electrodes. This combinatorial effect makes the design space unfathomably large: there are 1055 ways to combine all 1010 commercially accessible molecules into twocomponent mixtures; when you mix three of them, you have 10165 options to consider. For comparison, the current age of the universe is about 1017 seconds. Like the astronomical universe, the sheer size of the mixture universe is staggering, but also a source of great opportunity and excitement if it can be successfully navigated.
There is one more important set of considerations in electrolyte design: commercial requirements. Fortunately, this is the only set of properties where information on each molecule individually is sufficient. Cost, synthetic accessibility, feasible vendors, supply chain resiliency and more can all be inferred directly from the structure of each component alone. This has been a historically overlooked area of electrolyte design, but some capabilities are beginning to emerge: for example, models have been trained to predict the cost to procure a small sample of a molecule by learning directly on its molecular structure and by considering its retrosynthetic pathway.20,21
In sum, the recipe for reducing electrolyte design from a complex problem to a merely complicated one then requires not just knowing the properties of molecules; it requires knowing the excess properties of their mixtures and understanding their interactions with electrode interfaces. Molecular property prediction can be done using existing methods, but mixture and interfacial properties require new methods that incorporate these complex interactions. Then, a second layer of machine learning models must be deployed to decrease the computational cost of these predictions such that a significant portion of the space of mixtures can be traversed.
This philosophy underpins the Aionics electrolyte design platform, which has been built and deployed to accelerate electrolyte design for many battery designs and applications. Our platform contains a library of over 1010 molecules and their predicted physical and commercial properties, plus the ability to predict values for new
(continued on next page)
(continued from previous page)
Fig. 4. High-throughput flash point optimization. Using a flash point model trained on experimental data and the DiffMix framework, the Aionics platform computes the predicted flash point of approximately 18 million modifications of a baseline formulation. Each modification contains one novel molecular additive and a unique ratio of the baseline components. Many of the sampled variations have mixture flash points around 30C, but there are notable outliers with substantially higher and lower flash points. Mixture models that contain excess property physics like DiffMix are necessary to identify these unintuitive and highly important outliers. We acknowledge Drs. Mohamed Elshazly and Shreyas Honrao for generating this image and data.
molecules that are not yet commercially known. Relationships with chemical suppliers have enabled the development of models for predicting how cost will scale with volume. Most importantly, the platform supports computationally combining these molecules into any possible mixture so that their emergent mixture and interfacial properties can be predicted. Millions of variants on existing mixtures can be simulated and assessed in the span of minutes. To illustrate this, in Fig. 4 we show predictions of the flash point of approximately 18 million unique electrolyte formulations, each of which is composed of one of 36 variants of a baseline formulation plus any one of approximately 500,000 commercially available additives. These computations execute at a rate of approximately 3,000 mixtures per second on modest CPU resources; this map of 18 million mixtures took less than two hours of wall time to generate. This capability represents one of the largest databases of electrolyte formulations and their physical properties, and to our knowledge is the only major database of electrolytes that also contains commercial properties. The methods described in this article—molecular property prediction, mixture excess property prediction, interface interaction surrogates, and price prediction—are among the key advances underlying this state-of-the-art electrolyte design platform. All relevant intellectual property that we developed in the university setting has been licensed exclusively to Aionics.22
Looking forward to the future, we encourage the development of new methods that capture the cross-terms that are inherent in electrolyte design solutions: molecule-molecule ex-situ interactions, and molecule-surface in-situ interactions. Despite the massive space of molecules available to us, we believe that approaching the problem as merely a molecular design challenge is an oversimplification. The space of mixtures made up of these molecules is orders of magnitude larger, raising the prospect that we can discover new, high performance electrolyte blends for any electrochemical cell. Existing efforts in molecular property ML models, particularly by the pharmaceutical chemistry community, have laid a solid foundation; it is now up to the computational electrochemistry community to develop the domain specific models that are required to push the field forward.
This article was enabled and supported by the following contributors: Mohamed K. Elshazly, Shreyas Honrao, Varun Kumar Karrothu, Thuy B. Kim, Handong Ling, Noushin Omidvar, Stephanie Tarczynski, and Jiayi Wu. We thank and acknowledge their direct contributions to the data, models, and figures. We also wish to thank the remainder of the team at Aionics, Inc. whose hard work helped bring this analysis to life. Additionally, we thank all our colleagues over the years who contributed substantially to the previous papers discussed here, including Drs. Yumin Zhang, Shang Zhu, Zeeshan Ahmad, Gregory Houchins, Vikram Pande, Emil Annevelink, Rachel Kurchin, Adarsh Dave, Hongyi Lin, Ekin Dogus Cubuk, Brandi Ransom, Nicholas Grundish, Qian Yang, and Evan J. Reed. We thank Meghan Hayes and the Center for Technology Transfer and Enterprise Creation at Carnegie Mellon University for supporting our commercialization efforts of the intellectual property developed in the Viswanathan lab.
© The Electrochemical Society. DOI:10.1149/2.F08252IF
Austin D. Sendek, Co-founder and CEO, Aionics, Inc.
Education: BS Applied Physics (UC Davis); PhD Applied Physics (Stanford University)
Research Interests: Energy storage, Machine learning, Materials informatics
Work Experience: Adjunct Professor of Materials Science and Engineering, Stanford University, Stanford, CA, USA
Pubs + Patents: 19 publications, h-index 12 Awards: CB Insights AI 100 (Aionics), Fast Company’s World Changing Ideas 2024 (Aionics), Forbes 30 Under 30 in Energy 2019 Website: aionics.io https://orcid.org/0000-0003-3338-1615
Venkatasubramanian Viswanathan, Co-founder and Chief Scientist, Aionics, Inc.
Education: BTech in Mechanical Engineering (Indian Institute of Technology, Madras); PhD in Mechanical Engineering (Stanford University)
Research Interests: Energy storage, Machine learning, Materials informatics
Work Experience: Associate Professor of Aerospace Engineering, University of Michigan
Work with ECS: Active member of Electrochemical Society and winner of San Francisco Section Daniel Cubicciotti Student Award, Herbert H. Uhlig Summer Fellowship
Awards: MIT Technology Review Innovators Under 35, Office of Naval Research (ONR) Young Investigator Award, Alfred P. Sloan Research Fellowship in Chemistry, National Science Foundation CAREER award and American Chemical Society Energy Storage Lectureship
Pubs + Patents: 150+ publications, 50+ patents
Website: https://aionics.io; https://eeg.engin.umich.edu/ https://orcid.org/0000-0003-1060-5495
The Electrochemical Society book series provides authoritative, detailed accounts on specific topics in electrochemistry and solid state science and technology. These titles are sponsored by ECS and published in cooperation with Wiley.
ECS members receive 20% off all Wiley books. ECS members use promo code ECS18 at checkout.
New
ISBN 978-1-119-78883-6 | July 2022
A practical, single-source reference on the unique nature of seawater as a corrosive environment. Explains practical corrosion control solutions via design, proper materials selection, and implementation of good corrosion control engineering practices.
INTERESTED IN PUBLISHING WITH US?
Please contact publications@electrochem.org to discuss your idea. LaQue’s Handbook on Marine Corrosion, 2nd Edition David A. Shifler
by Ruozhu Feng, Yangang Liang, Tianzhixi Yin, Peiyuan Gao, and Wei Wang
With the fast-paced development of artificial intelligence (AI), large language models (LLMs) have attracted the most attention and are being applied in a wide range of application scenarios, including programming and health care, among others. Research has shown that LLM does well in programming and may fall behind in other fields at the moment. It is reasonable to expect that advances in LLM’s capability in “reasoning” with less computing resources are in the near future with an extremely quickly advancing pace. For the scientific research field, many farsighted psychologists and social science researchers have attempted to propose several modes of LLM participating in human scientists’ behavior patterns.1,2,3 At some level, research itself can be considered a type of social event of the scientific community, forging a community understanding of a certain natural occurrence by using a certain discipline to explain it. Several research papers have attempted to use an LLM4,5 or other generative models6,7 in materials research discovery; however, the reliability and the limitations of the approach have not been thoroughly discussed from the perspective of domain scientists.
AI systems like AlphaFold8 are impressive, but the reality is that to train a model like AlphaFold is highly resource-intensive, including both data and computing. Completely relying on training a model to give a reliable prediction of the physics of every aspect of frontier knowledge discovery is challenging at the current stage, with the current understanding of AI. The reliability of using LLM directly for scientific discovery (even with fine-tuning9 or reflection10) at its current stage of development is not yet high enough compared to a traditional human researcher approach, which can be seen from two aspects. One is the “guessing machine” nature of an LLM.11Though statistical, the output is unpredictable and uncontrollable at the current phase of development. The other is convergence. As the LLMs are trained/fine-tuned and prompted with known knowledge, usually from literature, rigorous studies are needed to illustrate how the probability distribution is connected to a reasonable novel proposal/hypothesis/conclusion.11,12,13 Without in-depth study of these concerns, careless expansive applications of LLM into science discovery might lead to questions of monoculture and weaken the scientific creativity of the human researcher community.2
Programming is a field where impressive milestones have been achieved by LLM within a fairly short period of time. Outside the argument as to whether or not an LLM is capable of “reasoning,” the outcome of LLM’s current programming capability of generating
code blocks indeed proves its value where a general logic applies during the process.14,15 With that outcome-validated result, it is reasonable to propose that scientists use LLM as tools or as a certain kind of information processor in their research, powering up human scientists’ capability of vast information accessibility. As discussed above, using only the LLM itself is not enough. Pairing LLM with model context protocol16 or other function-calling17 modules would lead to a balance of dynamic decision-making and scientific rigor
With our current understanding of technology development, we propose that human scientists can take advantage of LLM as an automated-task node, facilitating task execution with an elevated level of autonomy and dynamic information processing (Fig. 1). From this point of view, computation/simulation and/or physical experimentation are being carried out in a traditional way with numerical strictness, with LLM being applied as an automation node for task dispatch in the envisioned autonomous workflow. This way, the balance of autonomy and scientific rigor can be maintained, analogous to human scientists using a simulation software tool like COMSOL or experimental tools like a potentiostat. In this article, we demonstrate three examples of virtual automation using agentic workflows and discuss the accessibility of physical automation in the self-driving lab (SDL) setting. A multi-agent framework has shown promise in solving complex tasks,18 many with sequential logic or hierarchical logic. In the demonstrations, we adopt sequential logic with these agentic workflows, where human scientists act as commanders designing tasks, and each agent acts as a soldier executing tasks sequentially.
This demonstration aims to analyze the cost of a user-inquired common redox active compound synthesis in a laboratory environment. Just asking a regular chatbot would give unreliable answers with concerns of hallucination. Mimicking what a human scientist would have done for this question, we segment the task into several sub-tasks, with each delivered by an agent with tools programmed (Fig. 2A). Human scientists designed the automation to retrieve the chemical synthesis procedures from the literature database in JSON (JavaScript Object Notation, an efficient datainterchange format) with details on the starting reagents and the required amount. The next step would be for the analyst agent to search the internet to find each reagent’s selling price from common vendors, and the calculator agent to calculate the cost for all involved reagents at the required amount. This way, the information on the synthesis requirement and the reagent price would be dynamic in real time. Then the final reporter would compile all the collected information to deliver the final cost analysis.
This automation analyzes the environmental impact of the synthetic procedure of a common redox active compound reported in the literature. Like the above automation, this task also mimics a human scientist’s thinking process for solving such a task. Starting off with a librarian, the sub-task is to retrieve the information on the synthesis procedures and all the required reagents during the synthesis process from the provided literature database in JSON. Then the analyst agent would search the internet for the environmental impact
(continued from previous page)
of each reagent being used during the synthesis. Lastly, the reporter would compile all the information and generate a report.
Human scientists have limited memory. Every day, thousands of pieces of literature are published. Many times, new ideas are generated by connecting distant concepts.13 A common way for human scientists to propose great ideas is through collaboration between different domain scientists with different knowledge storage.19 Mimicking such a process, a conversation between human scientists and a reliable literature assistant with an enormous store of relevant literature would be expected to produce a similar effect. Here we argue that the adoption of a domain-knowledge augmented LLM assistant can be useful for brainstorming during hypothesis generation. Retrieval augmented generation (RAG) (Fig. 2B) is the most basic method that can be implemented, with numerous advanced and developing methods being published every day to enhance such generative models. With this automation, we address the fact that domain-knowledge augmentation is vital in a research setting.
In summary, the three example virtual automations share the same theme: using LLM as a node to dispatch subsequent tools on information collection or compilation. Such an approach enables an elevated level of autonomy by using an AI agent for the automation process, while maintaining the accuracy of information by using tools or software. One can envision that by integrating with more advanced tools, these agentic workflows could solve more complex tasks, freeing the scientist to work in areas in which LLM or AI may not be as reliable.
The above discussion addresses pure virtual automation in research;20 in parallel, the self-driving lab (SDL) can be considered as physical automation. Emerging publications are discussing means of integrating LLM into materials science discovery on the software side21 and the hardware side.22 Materials researchers and chemists have been following the same mindset since the golden age of science discovery, back to the 1800s–1900s, where human labor was heavily involved. Even though scientific discovery claims to be a mental activity, many researchers’ creativity is greatly limited by human labor time. With the technological advancement of AI, modernizing scientific tools is on the horizon.
SDLs are at the forefront of modernizing scientific discovery. We propose that a serious design philosophy is needed from the community in order to avoid a monoculture of scientific discovery. Many current acclaimed SDLs are highly constrained to a very specific type of data generation or workflow and are hardly adaptable to other fields.23,24,25 Another concern is that the experimental process is usually guided by a fixed algorithm,26,27 which leads to convergence and hampers the creativity of novel knowledge discovery.
Like the above-discussed purely software-based automation, SDLs can be hardware-based automation, using a LLM as an automation node for a more flexible and autonomous toolset for human scientists. The Materials Innovation through Robotics and AI Laboratory (MIRAL) at Pacific Northwest National Laboratory (PNNL) has been experimenting with such an approach with flexible agent-operated experiment modules.28 For lightweight tasks where the goal is serving human scientists as an intelligent tool, instead of designing a closed-loop workflow for very specific heavy-duty data collecting and processing, human scientists can adopt agentoperated modular automation toolsets (Fig. 3). For instance, human scientists can speak with one agent, describing an experimental sample requirement, and such an agent can generate operable files for the next robotic agent to execute, preparing the physical samples. After that, the analytical agent can operate one piece of analytical equipment (such as a simple potentiostat for a cyclic voltammetry measurement) to analyze the prepared samples. Such agentic systems have been proposed by several pioneering research groups using LLMs for decision-making during a closed-loop SDL process.22,29 Here, we would like to address a human-centric SDL and its toolset character. In the envisioned human-centric SDL, human scientists are the decision makers who assemble the agent modules needed for a task at hand with a drag-and-drop type automation design. For instance, the envisioned SDL automation would be similar to the most popular automation platforms like Zapier30 or n8n,31 or the software we are most familiar with, like IKA LabWorld32 or AutoLab Nova.33 Such a modular approach would be flexible and adaptable to various types of lightweight tasks for researchers with different backgrounds. We propose that SDLs should have two categories in future development, analogous to a supercomputer and a personal computer for different use cases. One is designed for heavy-duty tasks for large projects, with an extensive dataset collection goal and the adoption of a domain-specific trained model for robotic data collection algorithm. The other is designed for researchers’ daily use as basic scientific toolsets in a laboratory setting for basic robotic handling integrated with AI/LLM. The latter should be affordable and accessible to the general academic and/or educational community.
LLMs can be useful for materials scientists in both virtual automation and physical automation. Agentic systems provide enhanced problem-solving capability with LLMs and tool use. Adopting a function calling/model context protocol in LLMs provides a balance of elevated autonomy and scientific rigor. We propose, as materials scientists, that the integration of LLMs and robotics should be human-centric and function as tools for human scientists. Extensive studies and discussions from the community are needed to provide philosophical design guidance for adopting LLM and robotics in research, to avoid a monoculture of discovery.
This work was supported by the Energy Storage Research Alliance (ESRA) (reconceptualization, experiment, manuscript writing, and revision) (under contract no. 82132), an Energy Innovation Hub funded by the US Department of Energy, Office of Science, Basic Energy Sciences, and by the Energy Storage Materials Initiative (ESMI) (ideation and initial experiment) at Pacific Northwest National Laboratory (PNNL), which is a Laboratory Directed Research and Development project. All experiments are performed in PNNL’s MIRAL (Material Innovation through Robotics & AI Laboratory). PNNL is a multi-program national laboratory operated by Battelle for the US Department of Energy under contract DE-AC05-76RL01830.
Website: https://www.pnnl.gov/projects/energy-storage-materialsinitiative-esmi/capabilities
The code for the three example virtual agentic automations is available at https://github.com/pnnl/RFB_Assistant_V2
The code in this repository is generated by materials scientists with the assistance of LLM coding. © The Electrochemical Society. DOI:10.1149/2.F09252IF
Ruozhu Feng, Materials Scientist, Pacific Northwest National Laboratory (PNNL)
Education: BS in Chemistry (Nankai University); PhD in Chemistry (Washington University in St. Louis).
Research Interests: Electro-organic synthesis, Redox flow battery, AI/automation for scientific research
Work Experience: Materials Scientist (PNNL), Team Lead (PNNL)
https://orcid.org/0000-0001-6712-3695
Tianzhixi Yin, Data Scientist, PNNL
Education: BS in Mathematics (Xiamen University); MS in Actuarial Science (Temple University); PhD in Statistics (University of Wyoming)
Research Interests: Machine Learning, Artificial Intelligence
Work Experience: Data Scientist (PNNL)
Work with ECS: ECS conference symposium organizer Awards: 2023 PNNL Ronald L. Brodzinski Award for Early Career Exceptional Achievement, 2023 Spring ACS Energy & Fuels Division Early Career Investigator Spotlight, 2022 PNNL Energy Processes & Materials Division Early Career Researcher Award Pubs + Patents: 20+, including first-author Science, Joule, JES, etc. Website: Ruozhu Feng PNNL https://orcid.org/0000-0003-1427-3571
Yangang Liang, Materials Scientist, PNNL
Education: BS and MS Chemistry (Fudan University); PhD in MSE (University of Maryland at College Park)
Research Interests: Dr. Liang specializes in AI/ML-driven high-throughput experimentation for energy storage materials, focusing on electrolytes and electrodes for lithium-ion and redox flow batteries
Work Experience: He is a Materials Scientist at PNNL leading the MIRAL, with prior industry roles at GE Research and Biolegend in energy, water, and biomedical applications
Work with ECS: Liang contributes actively to ECS through publications on high-throughput electrolyte screening in lithium-ion batteries and electrolytes development in RFBs Awards: He has received multiple honors, including GE Innovation and Technical Achievement Awards Pubs + Patents: Dr. Liang holds 20+ US patents and 20+ highimpact papers.
Pubs: 40+ publications https://orcid.org/0000-0003-0730-4309
Peiyuan Gao, Research Scientist, PNNL
Education: BS in Environmental Science (Harbin Institute of Technology); MS in Chemistry (Lanzhou University); PhD in Chemistry (Institute of Chemistry, Chinese Academy of Science)
Research Interests: Multiscale modeling and simulation of energy storage materials; AI for material discovery; Scientific machine learning of molecular systems
Work Experience: Research Scientist, PNNL
Work with ECS: ECS Member since 2021
Pubs: 70+ publications
Website: https://www.pnnl.gov/people/peiyuan-gao https://orcid.org/0000-0001-8950-2741
Wei Wang, Laboratory Fellow, Materials Scientist, PNNL
Education: BS Ceramic Engineering (East China University of Science and Technology); MS in Materials Science and Engineering (Clemson University); PhD in Materials Science and Engineering (Carnegie Mellon University)
Research Interests: Materials science and electrochemistry; Materials development and system integration of various energy storage systems; AI/automation for scientific research; Innovative energy storage technologies
Work Experience: Laboratory Fellow (PNNL), Materials Scientist (PNNL), Team Lead (PNNL), Deputy Director (Energy Storage Research Alliance, PNNL), Director (Energy Storage Materials Initiative, PNNL), Founder (Materials Innovation through Robotics and AI Laboratory, PNNL), Co-founder (International Coalition for Energy Storage and Innovation), Vice chair of the ECS Pacific Northwest Section
(continued on next page)
Work with ECS: Symposium and conference organizer specializing in energy storage topics and was recognized with the Electrochemistry Research Award from the ECS Pacific Northwest Section in 2021.
Awards: Elected to the Washington State Academy of Sciences, 2024; Awardee, DOE Advanced Research Projects Agency-Energy OPEN Program, DOE, 2021; Battelle Distinguished Inventor, 2020; 1% Highly Cited Researcher, Clarivate, 2018; Awardee, DOE Advanced Research Projects Agency-Energy OPEN Program, DOE, 2015; Award for Excellence in Technology Transfer, FLC, 2013; R&D 100 Award, 2012
13. B. C. Lee and J. Chung, Nat Hum Behav, 8(10), 1906 (2024).
14. Anthropic, “On the Biology of a Large Language Model,” Transformer Circuits. Accessed: Apr. 01, 2025.
15. Anthropic, “Circuit Tracing: Revealing Computational Graphs in Language Models,” Transformer Circuits. Accessed: Apr. 01, 2025.
16. “Introducing the Model Context Protocol.” Accessed: Mar. 31, 2025.
17. “Function calling - OpenAI API.” Accessed: Mar. 31, 2025.
18. Q. Wu et al., “AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversation,” Oct. 03, 2023, arXiv: arXiv:2308.08155. Accessed: Feb. 22, 2024.
19. I. Yanai and M. J. Lercher, Nat Biotechnol, 42(1), 18 (2024).
Pubs + Patents: i10-index 97, citations > 22K Website: https://www.pnnl.gov/people/wei-wang https://orcid.org/0000-0002-5453-4695
1. E. Fedorenko, S. T. Piantadosi, and E. A. F. Gibson, Nature, 630(8017), 575 (2024).
2. L. Messeri and M. J. Crockett, Nature, 627(8002), 49 (2024).
3. H. Wang et al., Nature, 620(7972), 47 (2023).
4. Z. Zheng et al., J Am Chem Soc, 145(51), 28284 (2023).
5. Y. Zheng et al., Nat Mach Intell, 7, 437 (2025).
6. C. Zeni et al., Nature, 639, 624 (2025).
7. B. Sanchez-Lengeling and A. Aspuru-Guzik, Science, 361(6400), 360 (2018).
8. J. Jumper et al., Nature, 596(7873), 583 (2021).
9. “Fine-tuning - OpenAI API.” Accessed: Mar. 31, 2025.
10. M. Renze and E. Guven, “Self-Reflection in LLM Agents: Effects on Problem-Solving Performance,” in 2024 2nd International Conference on Foundation and Large Language Models (FLLM), 476–483 (2024).
11. Y. A. Yadkori, I. Kuzborskij, A. György, and C. Szepesvári, Adv Neural Inf Process Syst, 37, 58077 (2024).
12. J. Requeima, J. Bronskill, D. Choi, R. E. Turner, and D. Duvenaud, Adv Neural Inf Process Syst, 37, 109609 (2024).
20. Y. Ye et al., “ProAgent: From Robotic Process Automation to Agentic Process Automation,” Nov. 23, 2023, arXiv: arXiv:2311.10751. Accessed: Feb. 22, 2024.
21. C. Lu, et al., “The AI Scientist: Towards Fully Automated Open-Ended Scientific Discovery,” Aug. 31, 2024, arXiv: arXiv:2408.06292.
22. K. Darvish et al., Matter, 8(2), 101897 (2025).
23. D. F. Nippa et al., Nat Chem, 16(2), 239 (2024).
24. R. Pollice, B. Ding, and A. Aspuru-Guzik, Matter, 7(3), 1161 (2024).
25. M. R. Carbone et al., Matter, 7(2), 685 (2024).
26. G. Tom et al., “Self-Driving Laboratories for Chemistry and Materials Science,” Jun. 18, 2024, ChemRxiv.
27. Z. Yao et al., Nat Rev Mater, 8(3), 202 (2023).
28. T. Yin et al., “Learning Advance: Robotics-LLM Guided Hypotheses Generation for the Discovery of Chemical Knowledge,” Apr. 02, 2025, ChemRxiv.
29. A. M. Bran, S. Cox, O. Schilter, C. Baldassari, A. D. White, and P. Schwaller, “ChemCrow: Augmenting large-language models with chemistry tools,” Oct. 02, 2023, arXiv: arXiv:2304.05376.
30. “Automate without limits Zapier.” Accessed: Mar. 31, 2025.
31. n8n-io/n8n. (Apr. 01, 2025). TypeScript. n8n - Workflow Automation. Accessed: Mar. 31, 2025.
32. “Laboratory software for process automation, equipment control and monitoring,” IKA. Accessed: Mar. 31, 2025.
33. “NOVA.” Accessed: Mar. 31, 2025.
by Xiao Cui, Shijing Sun, and William C. Chueh
Li-ion batteries are essential for powering electronics, electric vehicles, and renewable energy systems. However, accurately evaluating battery lifetime often takes months to years, which delays critical performance feedback and optimization efforts. To address this challenge, data-driven early prediction of battery lifetime is used to accelerate performance evaluation and facilitate effective management. Recent studies have demonstrated promising performance in predicting battery lifetime using data-driven models.1–8 Meanwhile, significant progress has been made in understanding battery degradation mechanisms through electrochemical and mechanical models.9–12 However, interpretability in data-driven prediction models is limited. Interpretability here refers to our ability to understand how features influence predictions in terms of the underlying physical processes or degradation modes. Bridging the gap between prediction and degradation understanding is crucial for translating predictive power into practical insights for battery design and management.
To understand which features accurately predict battery cycle life, we first need to identify these features. They may be physicsbased,13–15 manually7 or automatically16,17 derived from cycling curve transformations. Collectively evaluating features generated from various cycling types, such as high-rate, low-rate, and resistancebased measurements, is essential to compare their predictive power within a dataset. Battery datasets are commonly generated by varying aging protocols,6,7,18–21 and features generated from these variable aging cycles inherently encode the cycling protocol themselves, resulting in data leakage.22 To mitigate this issue, diagnostic cycles6,19,20 or consistent portions of the aging cycle across cells7,18 are employed.
We have collected a unique dataset where formation is varied but all the batteries are subject to identical post-formation cycling conditions.23 Uniform aging conditions ensure that the training data exclusively represents intrinsic degradation processes, enabling evaluation of the features’ predictive capabilities without conditionspecific interference.22 Additionally, low-rate and pulse resistance diagnostics were carried out periodically. Using this dataset, features generated from aging cycles, low-rate diagnostic cycles, and pulse resistance measurements can be evaluated together.
Considering the relatively small size of battery datasets (typically comprising several hundred batteries), a simple model with physicsbased features is more suitable for prediction and data interpretation than complex black-box neural network models. An example of a feature-based approach is the ∆Q(V) feature proposed by Severson et al.,7 defined as the difference in capacity (Q) vs. voltage (V) curves between an early cycle and a later cycle (e.g., cycle 10 and cycle 100). This feature and its variants have been utilized in other studies6,20,21 for various battery chemistries; however, its predictive power is not well understood.
Battery degradation is generally classified into resistance and open circuit voltage (OCV)-based degradation modes.24 Three degradation modes significantly affect the OCV: loss of lithium inventory (QLi), positive electrode (PE) capacity (QPE), and negative electrode (NE) capacity (QNE).25 Differential voltage analysis estimates these electrode parameters by reconstructing full-cell voltage curves from fresh half-cell potential curves.26,27 While past studies have incorporated calculated electrode specifics into predictive analyses,13,19,28 establishing a direct correlation between electrode
degradation and overall cell degradation remains challenging, since battery degradation is usually a combination of multiple degradation modes. Moreover, the fitted results may deviate from the true values due to fitting errors.
In this study, the predictive power of various features are compared and linked to specific degradation mechanisms using a unique dataset where all cells were cycled under the same aging condition. We begin by featurizing a dataset of 186 single crystalline Li[Ni0.5Mn0.3Co0.2] O2 (SC-NMC532)/artificial graphite pouch cells. We systematically compare the predictive power of features generated from aging cycles and periodic low-rate diagnostic tests and pulse resistance measurements. Notably, the same feature generated at different cycling rates varies substantially in predictive power. Forward simulations of OCV-based degradation modes and state-of-charge (SOC)-based resistance degradation reveal that low-rate features are particularly predictive since they effectively capture the primary degradation mode in this dataset: lithium inventory loss. Our findings highlight that predictive features reveal underlying degradation mechanisms, underscoring the importance of carefully designed diagnostic cycles for feature-based prediction.
The predictive power of various feature sets is evaluated for early battery cycle life prediction using random forest regression models. To assess model performance, mean absolute percentage error (MAPE) is employed as the primary metric. Our feature-based approach incorporates inputs extracted from different cycling types, including high-rate aging cycles (0.75C discharge), low-rate discharge tests (0.05C and 0.2C), and pulse resistance measurements. The low-rate discharge and pulse resistance tests constitute the diagnostic test, conducted every 100 aging cycles. The output is the battery cycle life, defined as when the 0.75C discharge capacity reaches 80% of its initial value. Examples of feature extraction include: capacity values at the end of the cycle, the ∆Q(V) feature,7 resistances at different states of charge (SOCs), average voltage during discharge, time spent at the current-based constant voltage hold, and electrode specifics. ∆Q(V) represents the capacity difference as a function of voltage between an early and a later cycle, which can be further summarized using the mean or variance. In this study, early cycle is the first cycle after formation, defined as cycle 0. Electrode-specific capacities are extracted from the 0.05C discharge cycles using differential voltage analysis.23 Resistance values are extracted from pulse resistance tests. We note that the time spent at constant voltage (CV) hold and the average voltage also implicitly contain kinetic information. All features, except for electrode-specific capacities and SOC-specific resistances, can be extracted from any type of cycling, irrespective of the cycling rate.
In this dataset, each formation protocol condition is tested with three repeats. To prevent data leakage and overfitting, we perform a 5-fold cross-validation using the Python scikit-learn tool GroupKFold,29 ensuring that cells with the same formation protocol condition are assigned to either the training or the test set. The final MAPE is calculated as the average MAPE across all validation folds. For baseline comparison, we use the dummy regressor from the scikit-learn package, which does not incorporate any features and simply predicts the mean cycle life of the training set.
(continued on next page)
Cui,
Sun, and Chueh
(continued from previous page)
We first establish baselines for benchmarking early prediction models. Specifically, we consider a non-early-prediction model that uses formation protocols as inputs, which were varied to generate a distribution of cycle life. The prediction model using only formation protocol conditions as features shows decent performance compared to the dummy model. Fig. 1 compares the prediction performance across several feature sets, including formation protocol conditions and aging cycle and diagnostic cycle features. Using only formation protocol conditions and not considering cell-to-cell variability, the MAPE decreases significantly from 19.5% (dummy model) to 11.0%. The cycle life variation originates primarily from the different formation protocols applied. The trained model is evaluating the formation protocols rather than predicting battery degradation.22
Next, we compare the prediction performance of features generated from diagnostic and aging cycles with that of the baseline formation protocol model which does not use any aging data (Fig. 1). Models trained on the aging cycle or diagnostic cycle features show an increase in prediction error (16.1% and 12.5%) compared to the baseline formation protocol model (11.1%). For models trained on aging cycle or diagnostic cycle features, prediction error decreases as batteries age from immediately after formation to 220 aging cycles (with the mean cycle life around 750 cycles). The spike in prediction error immediately after formation, followed by a decrease as cells
degrade, suggests that the features generated post-formation capture true battery degradation rather than reflect the formation protocols. Diagnostic cycle features are consistently more predictive than aging cycle features. Fig. 1 shows that models trained with cycle 0 diagnostic features have a smaller MAPE (12.5%) than that of the models trained with aging features from cycle 220 (13.9%). Only features generated from low-rate discharge cycles are predictive, and resistance features or features with kinetic information all correlate poorly with cycle life. As an example, Fig. 2(a) and (b) compare the correlation of the same ∆Q(V) feature generated from high-rate (0.75C) aging cycle and low-rate (0.05C) diagnostic cycle to battery cycle life. The low-rate ∆Q(V) exhibits a stronger correlation to cycle life (Pearson coefficient -0.73) than does the high-rate ∆Q(V) (-0.25). We will explain the dependence of ∆Q(V) on cycling rates in the following sections.
Observing that the same feature exhibits different performance under different cycling rates, we aim to understand the origins of these differences through simulation. Specifically, we simulate OCV degradation modes to generate the corresponding voltage curves. We model 4% loss in QPE, 4% loss in QNE, and 4% shift between the PE and NE curves, which is equivalent to a 4% loss of QLi, to evaluate their individual effects on the ∆Q(V) curve. ∆Q(V) curves under different simulated degradation modes exhibit distinct shapes, as shown in Fig. 3(a).
Cycles
Cycles
Cycles Diagnostic Cycles
Importantly, the low-rate ∆Q(V) curve (representing the difference between cycle 0 and cycle 120) closely resembles the ∆Q(V) curve associated with QLi dominated degradation (gray and purple curves in Fig. 3(a)), suggesting that QLi loss is the primary contributor to the degradation of low-rate capacity. As a further validation, the variance of low-rate ∆Q(V) is plotted against the Li inventory loss and electrode capacity loss calculated from the differential voltage analysis in Fig. 3(b), (c), and (d), and only Li loss shows a strong correlation with the variation in the low-rate ∆Q(V).
Fig. 1. Comparison of prediction errors for various feature sets: no-input features (dummy model, red), formation protocol conditions (blue), aging cycle features (gray), and diagnostic cycle features (light blue). The cycle number on the x-axis indicates the number of aging cycles after which the features are generated. A cycle number of 0 corresponds to the cycle immediately after formation.
Under high-rate cycling, overpotential becomes significant, necessitating an investigation into how different types of resistance impact the shape of the ∆Q(V) curve. Ohmic resistance is independent of the state of charge (SOC) of the battery, whereas other resistances such as charge transfer and mass transport resistance depend strongly on SOC.30 We investigate how SOC-dependent and SOC-independent resistances affect the shape of ∆Q(V). In Fig. 4(a), we model resistance growth as a parabolic function of battery SOC and vary the maximum resistance growth. Resistance increases dramatically at the SOC extremes, aligning with previous literature reporting poor electrode reaction kinetics at these points.31–33 The
2. Correlation of the mean (∆Q(V)) generated from (a) 0.75C aging cycle and (b) 0.05C low-rate diagnostic cycle to battery cycle life. The capacity difference is taken between cycle 120 and cycle 0.
∆Q(V) curves resulting from this simulation are shown in Fig. 4(c). These curves exhibit ∆Q(V) peaks around 3.5 V and 4.2 V, which correspond closely to the features observed in the ∆Q(V) curve obtained from high-rate cycling (representing the difference between cycle 0 and cycle 120). Conversely, we vary resistance but assume that it is independent of the SOC in Fig. 4(b), and the resulting ∆Q(V) curves are presented in Fig. 4(d). A mismatch in the peak features is observed when comparing the simulated curves to the high-rate ∆Q(V) curve. From Fig. 4, we observe that high SOC contributes
to the peak broadening at 3.5 V and the formation of the peak at 4.2 V. From this analysis, we conclude that ∆Q(V) curves generated under high-rate cycling effectively reveal resistance degradation, particularly the increase in resistance at high SOC levels.
Fig. 3. Simulated ∆Q(V) curves under different OCV-based degradation modes and the experimental ∆Q(V) curve generated from low-rate cycles. (a) The forward simulated ∆Q(V) curve for 4% QPE loss, 4% QNE loss, and 4% shift between the PE and the NE curve (equivalent to 4% QLi loss). The experimental ∆Q(V) curve generated from the diagnostic low-rate (0.05C) discharge curve between cycle 0 and cycle 120 is shown in purple. After 120 aging cycles, how the loss in QLi (b), QPE (c), and QNE (d) correlates with the variance of the low-rate ∆Q(V).
Predictive features reveal the underlying mechanisms of battery degradation. In the previous sections, we conclude that low-rate ∆Q(V) captures QLi loss while high-rate ∆Q(V) captures resistance growth, especially at high SOC. Low-rate ∆Q(V) features are more predictive than those generated from high-rate cycles. In this both high-temperature and fast formation protocols significantly improve battery cycle life. When using the variance to summarize the ∆Q(V) curve (var(∆Q(V))), two distinct groups emerge (Fig. 5(a)), corresponding to the high-temperature (purple) and fast-formation (blue) conditions. We have demonstrated the strikingly strong correlation between the variance of low-rate ∆Q(V) and QLi loss in Fig. 3(b). Despite the main degradation mechanism being QLi loss, different features exhibit different sensitivity to this loss, depending on the formation impact pathway. The correlation between the low-rate var(∆Q(V)) feature and cycle life for non-high-temperature formed cells is very strong (Pearson coefficient = -0.87; Fig. 5(b)). However, the most correlative feature for non-fast-formed cells, especially those formed at 55◦C, is resistance growth at low battery SOC (around 5 %). Our previous publication23 has shown that high-temperature formation and fast formation improve battery cycle life through different mechanisms. Fastformed cells experience significant Li loss during formation; however, this substantial Li loss shifts electrode utilization, counterintuitively slowing down subsequent Li loss. In contrast, the improved performance of high-temperature-formed cells are attributed to improved solid-electrolyte interphase (SEI) properties. We note that more resistance growth at low SOC after 120 aging cycles positively correlates with longer cycle life in non-fast-formed cells. We suspect that high-temperature formed cells experience less Li loss and the cathode remains more lithiated at low full cell SOC. Consequently, the low SOC resistance growth increases, because it is more sensitive to the Li loss.28 This effect also explains why only resistance growth at low SOC exhibits a decent correlation. The different predictive features of cells under different formation protocols further suggest differences in their underlying degradation mechanisms.
Fig. 4 Simulated ∆Q(V) curves under different resistance degradation modes. (a) Resistance varies with SOC, exhibiting higher resistance growth at both high and low SOC levels. (b) Resistance increases uniformly, independent of battery state of charge (SOC). (c) Simulated ∆Q(V) curve resulting from the SOC-dependent resistance in (a). (d) Simulated ∆Q(V) curve resulting from the SOC-independent resistance in (b). The experimental ∆Q(V) curve between cycle 0 and 100, obtained from high-rate cycling, is shown in purple for reference.
Our study addresses the critical need for interpretable models in early Li-ion battery cycle life prediction. We evaluate the predictive power of features derived from high-rate, low-rate, and resistance-based cycling measurements using a unique formation dataset with identical postformation aging cycles. By linking specific features to underlying degradation mechanisms, we find that the same ∆Q(V) generated from high and low-rate cycles corresponds to different degradation modes: at high rates, it captures resistance growth at high battery SOC, while at low rates, it reflects Li loss. Since Li loss is the inherent degradation mode in this
(continued on next page)
Cui, Sun, and Chueh (continued from previous page)
dataset, low-rate features are more predictive. Moreover, we demonstrate the importance of incorporating diagnostic cycles into battery dataset generation: not only do they standardize battery state of health evaluation, but they also capture various degradation modes present in the dataset through low-rate and highrate discharge cycles. By directly connecting predictive features to degradation mechanisms, our work facilitates more accurate and reliable battery life predictions.
Data used in this study is available at https:// data.matr.io/8/ Source code can be accessed at: https://github.com/cx26/degradation-mode.git
This work was supported by the Toyota Research Institute through the Accelerated Materials Design and Discovery program. X.C. acknowledges support from the Stanford Data Science Scholars Program.
Fig. 5 Different predictive features under different formation protocols. (a) Cycle life plotted against the variance of low-rate ∆Q(V) feature (var(∆Q0.05C(V))). Blue and purple data points correspond to fast and high temperature formation. (b) For cells formed below 55◦C, Var(∆Q0.05C(V)) strongly correlates with cycle life. (c) For non-fast formed cells, low-SOC resistance growth between cycle 120 and cycle 0 correlates with cycle life.
Conceptualization, X.C., W.C.C., S.S.; Methodology, X.C.; Writing – Original Draft, X.C.; Writing – Review & Editing, all authors; Funding Acquisition, W.C.C.; Supervision, W.C.C.
© The Electrochemical Society. DOI:10.1149/2.F10252IF
Xiao Cui, Materials Science & Engineering, Stanford University
Education: BEng in Nano Materials & Engineering (Soochow University); MS, PhD in Materials Science & Engineering, PhD minor in Computer Science (Stanford University)
William C. Chueh, Associate Professor of Materials Science & Engineering, Stanford University
Education: BS, Applied Physics, MS, PhD, Materials Science (Caltech)
Research Interests: Electrochemistry, Redoxactive solids, Reactions and transport
Work Experience: Director, Stanford Precourt Institute for Energy; Director, SLAC-Stanford Battery Center
Research Interests: Battery degradation, Electrochemistry, Battery informatics Awards: Stanford Data Science Fellowship Pubs + Patents: 5 publications, 2 patents https://orcid.org/0009-0008-7615-220X
Shijing Sun, Mechanical Engineering, University of Washington
Education: BA Natural Sciences (University of Cambridge); PhD in Materials Science (University of Cambridge)
Research Interests: Energy materials, AI for science, Self-driving labs
Work Experience: Senior Research Scientist (Toyota Research Institute), Research Scientist (MIT)
Awards: FACET Award, Scialog Fellow, NETI Educational Development Award
Pubs + Patents: 60+ publications, 1 patent Website: www.uwsunlab.com https://orcid.org/0000-0002-6179-1390
Awards: David A. Shirley Award, Friedrich Wilhelm Bessel Research Award, MRS Outstanding Young Investigator Award, BASF/Volkswagen Science Award Electrochemistry Pubs + Patents: 130+ publications, 15+ patents. Website: http://chuehlab.stanford.edu/ https://orcid.org/0000-0002-7066-3470
1. P. Gasper, N. Collath, H. C. Hesse, A. Jossen, and K. Smith, J Electrochem Soc, 169, 080518 (2022).
2. J. Hong, D. Lee, E.-R. Jeong, and Y. Yi, Appl Energy, 278, 115646 (2020)
3. M. R. Kunz, E. J. Dufek, Z. Yi, et al., Batter Supercaps, 4, 1186 (2021).
4. Y. Li et al., Renew Sustain Energy Rev, 113, 109254 (2019).
5. G. Ma et al., Energy Environ Sci, 10, 4083 (2022).
6. N. H. Paulson et al., J Power Sources, 527, 231127 (2022)
7. K. A. Severson et al., Nat Energy, 4, 383 (2019).
8. M.-F. Ng, J. Zhao, Q. Yan, G. J. Conduit, and Z. W. Seh, Nat Mach Intell, 2, 161 (2020).
9. I. Laresgoiti, S. Käbitz, M. Ecker, and D. U. Sauer, J Power Sources, 300, 112 (2015).
10. M. Johnen et al., J Energy Storage, 34, 102011 (2021).
11. J. Keil and A. Jossen, J Electrochem Soc, 167, 110535 (2020)
12. M. Rogge and A. Jossen, Batter Supercaps, 7, e202300313 (2023).
13. B.-R. Chen et al., Joule (2022).
14. M. Dubarry, N. Costa, and D. Matthews, Nat Commun, 14, 3138 (2023).
15. P. K. Jones, U. Stimming, and A. A. Lee, Nat Commun, 13, 4806 (2022).
16. J. Rhyu et al., arXiv [cs.LG] (2024).
17. X. Liu et al., J Am Chem Soc (2024).
18. P. M. Attia et al., Nature, 578, 397 (2020).
19. B. van Vlijmen, V. Lam, P. A. Asinger, X. Cui, and D. Ganapathi, ChemRxiv (2023).
20. T. Li, Z. Zhou, A. Thelen, D. A. Howey, and C. Hu, Cell Rep Phys Sci, 5, 101891 (2024)
21. J. Zhu et al., Nat Commun, 13, 2261 (2022)
22. A. Geslin et al., Joule, 7, 1956 (2023).
23. X. Cui et al., Joule, 8, 3072 (2024).
24. A. Barai et al., Prog Energy Combust Sci, 72, 1–31 (2019).
25. C. R. Birkl, M. R. Roberts, E. McTurk, P. G. Bruce, and D. A. Howey, J Power Sources, 341, 373 (2017).
26. M. Dubarry, C. Truchot, and B. Y. Liaw, J Power Sources, 219, 204 (2012).
27. A. Weng, J. B. Siegel, and A. Stefanopoulou, Front Energy Res, 11, 1087269 (2023)
28. A. Weng et al., Joule, 5, 2971 (2021)
29. F. Pedregosa et al., J Mach Learn Res, abs/1201.0490, 2825 (2011).
30. A. Barai, K. Uddin, W. D. Widanage, A. McGordon, and P. Jennings, Sci Rep, 8, 21 (2018).
31. J. Lim et al., Science, 353, 566 (2016).
32. V. J. Ovejas and A. Cuadras, J Power Sources, 418, 176 (2019).
33. J. Li et al., J Electrochem Soc, 165, A2682 (2018)
WA, US May 24-28, 2026
Washington State Convention Center
by Weijie Mai and Shaoxiong Hu
Battery cycling data contain rich information about battery performance and degradation and have been extensively explored for a variety of critical battery management tasks, including state-of-charge (SOC) estimation, state-of-health (SOH) monitoring, anomaly detection, fault diagnosis, and lifetime prediction. Among these, battery-life prediction is becoming particularly vital as lithium-ion batteries become increasingly indispensable in electric vehicles, grid-scale energy storage, and portable electronics. Accurate prediction of battery life improves battery management system (BMS) efficiency, reduces operational costs and safety risks, and accelerates the development of new battery chemistries and management strategies. Therefore, this article explicitly emphasizes battery-life prediction— one of the most fundamental yet challenging tasks—examining the associated traditional methods as well as emerging transformerinspired AI modeling frameworks.
Battery cycling—the repeated charge and discharge process— produces characteristic temporal data sequences that record electrical measurements such as voltage, current, and capacity under varying charge-discharge conditions. Such cycling data inherently contain complex temporal dependencies, nonlinear dynamics, and considerable variability across different chemistries and operational settings. Reliable battery-life prediction thus fundamentally becomes a challenging time-series modeling problem.
Historically, approaches to battery cycle data analysis began with simple statistical or physics-inspired models, later moving toward data-driven machine learning paradigms. These traditional methods (Fig. 1), though widely adopted, face difficulty in effectively capturing the long-range temporal dynamics embedded in battery cycles. Recently, the emergence of transformer-based large language models (LLMs), widely successful in modeling complex sequential patterns in language data, has provided inspiration for more sophisticated and context-aware (i.e., explicitly capturing long-range dependencies and historical cycling information) modeling for battery cycling datasets.
In this article, we provide an accessible overview of several traditional methods used in battery-life prediction tasks, subsequently describing recent conceptual frameworks inspired directly by largescale transformer-based language models, examining their potential advantages and the remaining open challenges.
Historically, predicting battery lifetime based on cycling data started by using traditional time-series modeling approaches. Broadly speaking, these methods can be classified into two categories: modelbased and data-driven.1–3
Model-based approaches rely on building mathematical models that explicitly describe battery behavior and degradation processes. These approaches aim to leverage established physical and/or chemical knowledge about batteries to predict aging trends directly.
Early examples include empirical or semi-empirical models, which use experimentally derived mathematical formulas to capture battery degradation behavior. 4–6 These empirical methods are intuitive and computationally simple; however, they typically require careful
parameter selection and may struggle when battery conditions differ markedly from those used to derive the model.
More physically rigorous approaches are electrochemical or physics-based models, which describe the internal electrochemical reactions and physical processes happening within a battery during charging and discharging.7–9 Although these models offer deeper insights into the battery’s internal mechanisms, their complexity often demands extensive expert domain knowledge, detailed characterization datasets, and intensive computational resources, making them less practical for real-time deployment.
Between these extremes are simplified approaches such as equivalent circuit models (ECMs), where the battery is modeled using simplified electrical circuits.10, 11 ECMs balance physical interpretability and computational simplicity, facilitating their practical implementation within battery management systems. Yet, like with empirical models, the accuracy of ECMs can degrade if operational conditions deviate from modeled assumptions.
Despite their widespread application, all these model-based methods rely upon predefined assumptions or simplifications about battery behavior. Consequently, their performance typically deteriorates when batteries experience significantly altered usage patterns or new operational scenarios, requiring frequent recalibration or adjustment.
Data-driven approaches utilize historical cycling measurements— such as voltage, current, temperature, and capacity profiles collected over charge-discharge cycles—to build predictive battery-life models. Rather than explicitly encoding physical or chemical battery mechanisms, these methods primarily rely on unveiling hidden patterns and trends embedded within recorded data.
Broadly speaking, data-driven methods can be classified into feature-based data-driven and purely data-driven categories,3 each involving different degrees of domain-specific knowledge. Featurebased methods begin by carefully extracting meaningful health indicators (HIs) from raw battery measurements. Health indicators often include statistical features, characteristic curve shapes, voltagerelated parameters, or capacity signal metrics specifically designed with electrochemical domain expertise. These extracted features are subsequently fed into traditional statistical or machine-learning algorithms—such as polynomial regression, Gaussian process regression (GPR), support vector regression (SVR), random forests, or gradient boosting—to perform life prediction.12–14
In contrast, purely data-driven approaches reduce reliance on explicit domain-specific feature extraction by directly utilizing raw measured signals as model inputs. In particular, deep neural networks—such as recurrent neural networks (RNNs), including long short-term memory (LSTM) and gated recurrent units (GRU)—have attracted substantial attention.15–18 The intrinsic capability of RNN architectures to process sequential temporal data enables automatic discovery of highly nonlinear degradation patterns hidden within multicycle battery measurements, facilitating end-to-end predictions.
Overall, both indirect and direct data-driven approaches have substantially improved battery-life modeling, yet each faces specific limitations. Indirect methods provide interpretability but depend heavily on expert-driven feature selection, reducing their adaptability.
Direct methods automatically capture complex patterns and are more adaptive but typically require large datasets and significant computational resources. Moreover, both approaches often struggle to effectively model subtle long-term dependencies within battery cycling data, motivating further research into alternative modeling frameworks.
The application of transformer-based models in time series prediction19 has gained significant attention in recent years, driven by their ability to capture long-range dependencies and model complex temporal patterns. Originally developed for natural language processing (NLP), transformers leverage self-attention mechanisms to dynamically weigh relationships between sequence elements,20 overcoming limitations of traditional recurrent neural networks (RNNs) and convolutional neural networks (CNNs) in handling sequential data.
Recent success stories of transformer-based large language models (LLMs), such as GPT models in NLP, have inspired researchers to reassess how sequential battery datasets are modeled. These LLM architectures rely fundamentally on context-dependent embeddings and attention mechanism—the powerful abilities that effectively capture context and long-range dependencies within complex sequences in textual data.
Inspiration comes from an intriguing parallel between battery cycling data and sequential language data. Battery cycling data
inherently represent sequences characterized by strong temporal dependencies, where the current cycle state significantly depends on past cycling history over long time horizons. Additionally, such data are typically multivariate, containing inherent correlations among multiple measured parameters (e.g., voltage, current, temperature, and capacity) at each time step. Transformer-based language models, which utilize powerful attention mechanisms, naturally excel at simultaneously capturing both long-range temporal dependencies and relationships among multiple variables, offering significant potential to improve battery-life predictions.
By adapting transformer models originally developed for natural language processing, battery cycling data can be analyzed analogously as sequences of words or phrases, as shown in Fig. 2. Specifically, segments of cycling data from defined intervals or cycles are first converted into numerical vectors, much like the embeddings used in language models. These embedded vectors are then processed through the transformer’s characteristic components—such as positional encodings and the attention mechanism—to capture relationships across both temporal dimensions (across multiple cycles) and multivariate dimensions (among multiple measured parameters). Finally, the processed information is utilized to predict battery life or other relevant battery states.
Following this fundamental intuition, several recent works have explored transformer-based approaches tailored explicitly to batterylifetime prediction tasks. 21–23 For instance, a temporal transformer network (TTN) 24 has been proposed that leverages self-attention mechanisms to simultaneously extract temporal dependencies across cycling data, significantly improving predictions for lithiumion battery remaining useful life. Another notable example is the attention-based recurrent algorithm for neural analysis (ARCANA),25 which, although not strictly a transformer architecture, utilizes the
core attention mechanism from transformers and—similar to large language models—highlights how extensive and diverse pretraining datasets can substantially improve model generalization across different battery chemistries.
Moreover, inspired by the success of standardized benchmarking and extensive pre-training in NLP, battery researchers have started establishing comprehensive evaluation frameworks and large-scale open-sourced datasets. For example, BatteryLife26 is a unified benchmark dataset that integrates 16 publicly available battery datasets which span diverse chemistries, battery formats, and operating scenarios, enabling systematic comparison of predictive models across previously unseen battery conditions. BatteryLife also introduces the CyclePatch technique, which segments cycling data into tokens analogous to word embeddings, effectively capturing recurring patterns within battery degradation sequences. Collectively, these efforts mirror the catalysts behind LLMs’ rapid progress— diverse open-sourced training resources, standardized performance evaluations, and innovative data embedding techniques—highlighting transformative opportunities for advancing battery lifetime modeling.
Despite these promising advances, challenges remain in adapting transformer-based approaches to battery analytics. Key issues include handling the high diversity and variability across battery chemistries, operating conditions, and cycling protocols, which pose difficulties for generalization. Additionally, publicly available battery datasets remain limited in scale and representativeness, complicating effective pretraining and systematic benchmarking efforts. Finally, improving the interpretability and uncertainty quantification of these models is crucial for adoption in safety-critical battery management applications. Addressing these challenges will be essential to fully realizing transformers’ potential in battery-lifetime prediction.
The empirical principles of scaling laws have established predictable relationships between model performance and computational resources. Pioneered by OpenAI,27 scaling laws quantify how increasing model size, dataset size, and computing budget systematically enhance capabilities such as reasoning, few-shot learning, and generalization. These insights highlight a critical trade-off: larger models require exponentially more data and computing resources to achieve sublinear performance gains. For battery-lifetime prediction, analogous principles could guide the design of predictive models by optimizing the interplay among data volume, model complexity, and computational constraints. For example, synthetic data generation27 might “scale” limited real-world battery degradation datasets, while lightweight architectures could balance accuracy and deployment feasibility in resource-limited settings.
Reinforcement learning (RL) has significantly advanced large language models (LLMs) such as ChatGPT and Claude by aligning their responses closely with human preferences. This approach, known as reinforcement learning from human feedback (RLHF),28 allows models to provide detailed, context-sensitive responses while decreasing reliance on manually labeled data. Similarly, RL can be applied to lifetime prediction to dynamically and adaptively improve prediction accuracy.29 For instance, an RL agent could learn to update battery degradation forecasts in real time by interacting
(continued on next page)
(continued from previous page)
with a simulated battery environment, receiving “rewards” based on prediction accuracy and operational safety.
Combining scaling laws and RL offers exciting opportunities for improving battery predictions. Scaling laws can help expand limited datasets, enhancing predictions even for uncommon battery degradation conditions. Meanwhile, RL techniques can help models continuously learn how to forecast battery life and while simultaneously optimizing control decisions. Inspired by the successful application of these methods in advanced language models, this hybrid approach could transform battery maintenance—allowing smarter, more adaptive, and efficient management of complex energy systems. Future research should further explore integrated solutions that combine battery physics knowledge with scalable neural-network methods, drawing inspiration from breakthroughs in language modeling.30
Battery lifetime prediction approaches have evolved significantly—from traditional statistical and physics-based methods to advanced data-driven techniques tailored for complex temporal data. Transformer-based models, inspired by recent breakthroughs in large language models, offer promising capabilities by effectively capturing long-range temporal dependencies and multivariate patterns within battery cycling datasets. Additionally, advances in scaling laws and RL provide synergistic opportunities to overcome challenges such as data scarcity, computational trade-offs, and limited adaptability. Future research should focus on integrating battery physics knowledge with scalable AI architectures, while leveraging insights from language modeling and RL to achieve more accurate, adaptable, and practically deployable battery lifetime prediction solutions.
© The Electrochemical Society. DOI:10.1149/2.F11252IF
Weijie Mai, Senior Director of Algorithms and Big Data, SES AI
Education: BS in Materials Science and Engineering (Zhejiang University); PhD in Materials Science and Engineering (the Ohio State University)
Research Interests: Electrochemistry, Batteries, Physics-based models, AI
Pubs + Patents: 20+ publications, 1+ patents
1. J. Zhao, L. Tian, L. Cheng, Y. Zhang, and C. Zhu, in I and CPS Asia 2022 - 2022 IEEE IAS Industrial and Commercial Power System Asia, p. 1501–1505, Institute of Electrical and Electronics Engineers Inc. (2022).
2. X. Hu, L. Xu, X. Lin, and M. Pecht, Joule, 4, 310–346 (2020).
3. V. Sulzer et al., Joule, 5, 1934 (2021)
4. I. Bloom, B. W. Cole, J. J. Sohn, et al., J Power Sources, 101, 238 (2001)
5. P. Ramadass, B. Haran, R. White, and B. N. Popov, J Power Sources, 123, 230 (2003)
6. J. Schmalstieg, S. Käbitz, M. Ecker, and D. U. Sauer, J Power Sources, 257, 325 (2014).
7. S. Atalay et al., J Power Sources, 478 (2020)
8. X. G. Yang, Y. Leng, G. Zhang, S. Ge, and C. Y. Wang, J Power Sources, 360, 28 (2017)
9. J. Keil and A. Jossen, J Electrochem Soc, 167, 110535 (2020)
10. A. Chu, A. Allam, A. Cordoba Arenas, G. Rizzoni, and S. Onori, J Power Sources, 478 (2020).
11. D, Zhang, S. Dey, H. E. Perez, and S. J. Moura, Remaining Useful Life Estimation of Lithium-Ion Batteries based on Thermal Dynamics, 2017 American Control Conference (ACC), (2017).
12. K. A. Severson et al., Nat Energy, 4, 383 (2019)
13. V. Sulzer, P. Mohtat, S. Lee, J. B. Siegel, and A. G. Stefanopoulou, (2020).
14. M. Lucu, E. Martinez-Laserna, I. Gandiaga, et al., J Energy Storage, 30 (2020).
15. M. Iftikhar, M. Shoaib, A. Altaf, et al., Sci Rep, 14, 25838 (2024)
16. Y. Z. Zhang, R. Xiong, H. W. He, and Z. Liu, A LSTM-RNN method for the lithium-ion battery remaining useful life prediction, IEEE Conference, (2017)
17. P. Xu and Y. Lu, Predicting Li-ion Battery Cycle Life with LSTM RNN
18. J. Lu, R. Xiong, J. Tian, et al., Energy Storage Mater, 50, 139 (2022).
19. A. Vaswani, N. Shazeer, N. Parmar, et al., (2017)
20. H. Zhou, S. Zhang, J. Peng, et al., (2020)
21. J. Zhao, X. Han, Y. Wu, Z. Wang, and A. F. Burke, J Energy Chem, 102, 463–496 (2025).
22. D. Chen, W. Hong, and X. Zhou, IEEE Access, 10, 19621–19628 (2022).
23. D. Chen and X. Zhou, J Energy Storage, 84 (2024).
24. W. Song, D. Wu, W. Shen, and B. Boulet, Procedia Comput Sci, 217, 1830 (2022)
25. F. Rahmanian, R. M. Lee, D. Linzer, et al., NPJ Comput Mater, 10 (2024)
26. R. Tan et al., (2025)
27. J. Kaplan et al., (2020).
28. L. Ouyang et al., (2022)
Website: https://www.linkedin.com/in/weijiemai/ https://orcid.org/0000-0001-8138-8104
Shaoxiong Hu, Senior AI Engineer, SES AI Education: BS in Mathematics (Hubei University) ; MS in Mathematics (University of Science and Technology of China); PhD in Statistics (Queen Mary, University of London)
Research Interests: Batteries, AI Work Experience: JP Morgan, Huawei, Alibaba Group
Pubs + Patents: 5+ publications, 1+ patents
29. J. Liu et al., (2023)
30. DeepSeek-AI et al., (2025)
Website: https://www.linkedin.com/in/ shaoxiong-hu-954637130/ https://orcid.org/0009-0005-6293-0048
ECS sections introduce and support activities in electrochemistry and solid state science within specific regions. Getting involved with a section is an excellent networking opportunity for those new to the field or advanced in their careers. Sections also bring technical news and activities within reach of those who are not able to attend ECS
meetings. Sections participate in overall ECS affairs, work hard to increase ECS membership, and help create awareness for the science. For more information on your region’s section, go to https://www. electrochem.org/sections. Contact ECS Section & Chapter Liaison, Maggie Hohenadel, for more information on joining.
Section Name
Section Chair
Arizona Section Candace K. Chan
Brazil Section Raphael Nagao
Canada Section Steen B. Schougaard
Chile Section José H. Zagal
China Section Open
Detroit Section Tobias Glossmann
Europe Section Jan Macák
Georgia Section Faisal Alamgir
India Section Sinthai Ilangovan
Israel Section Eran Edri
Japan Section Yasushi Idemoto
Korea Section Won-Sub Yoon
Mexico Section Norberto Casillas Santana
Mid-America Section Ahmed Farghaly
National Capital Section Chungsheng Wang
New England Section Sanjeev Mukerjee
Pacific Northwest Section April Li
Pittsburgh Section Open
San Francisco Section Xiong Peng
Singapore Section Zhichuan J. Xu
Taiwan Section Chi-Chang Hu
Texas Section Jeremy P. Meyers
Thailand Section Soorathep Kheawhom
Twin Cities Section Lifeng Dong
Learn more about ECS sections at www.electrochem.org/sections.
The ECS Honors & Awards Program recognizes outstanding technical achievement in electrochemistry, solid state science, and technology, and acknowledges exceptional service to the Society. Award opportunities are provided in the categories of Society Awards, Division Awards, Section Awards, and Student Awards.
Recognizing that today’s emerging scientists are our field’s next generation of leaders, ECS offers competitive fellowships and grants that make it possible for students and young professionals to make discoveries and shape our science far into the future.
Charles W. Tobias Early-Career Award recognizes a young scientist or engineer’s outstanding scientific and/or engineering work in fundamental or applied electrochemistry or solid state science and technology. Established in 2003, the award consists of a framed certificate; $5,000*; ECS Life Membership; complimentary ECS Meeting registration; and travel expenses.
Nomination period: April 15 – October 1, 2025
ECS Toyota Young Investigator Fellowship, launched in partnership with the Toyota Research Institute of North America in 2015, funds innovative electrochemical research in green energy technology. ECS Toyota Fellows receive ECS membership and restricted grants of no less than $50,000 to conduct their proposed research.
Materials deadline: January 31, annually
Edward Goodrich Acheson Award, established in 1928, acknowledges distinguished contributions to the advancement of ECS’s objects, purposes, or activities. Awardees receive a gold medal; plaque bearing a bronze replica of the medal; $10,000; ECS Life Membership; and complimentary ECS Meeting registration.
Nomination period: April 15 – October 1, 2025
Leadership Circle Awards, granted in the year that an institutional partner reaches a milestone level, were established in 2002 to honor and thank our electrochemistry and solid state science partners. Awardees receive a commemorative plaque and recognition in ECS Interface and on the ECS website.
Nominations not accepted
Pacific Northwest Section Electrochemistry Research Award Sponsored by Gamry Instruments, established in 2021, recognizes research excellence in electrochemistry and solid state science and technology by independent scientists or engineers working in Washington, Oregon, or Idaho. The award consists of a framed certificate and $1,000.
Nomination period: April 15 – July 15, 2024
Biannual Meeting Travel Grants provide complimentary ECS Meeting registration, travel expenses, luncheon/reception tickets, and more to undergraduates, graduate students, postdoctoral researchers, young professionals, and faculty presenting ECS biannual meeting papers. The divisions and sections providing the grants maintain their own application requirements.
248th ECS Meeting Travel Grant application period: March 28 – June 23, 2025
Colin Garfield Fink Fellowship, established in 1962, supports research by a postdoctoral scientist/ researcher from June through September in a field of interest to the Society. The award consists of $5,000 and publication of a summary report in ECS Interface
Materials deadline: January 15, annually
ECS General Student Poster Session Awards were established in 1993 to stimulate active student interest and participation in the Society. The Z01—General Student Poster Session enables undergraduate and graduate students to present research results of general interest to ECS. Accepted posters are eligible for General Student Poster Awards of 1st place: $1,500; 2nd place: $1,000; and 3rd place: $500. For award consideration, authors must submit abstracts and be accepted into the session; upload a digital poster; and be present for in-person judging.
Materials deadline: Relevant ECS Meeting abstract deadline
ECS Outstanding Student Chapter Award recognizes student chapters that demonstrate active participation in the Society’s technical activities; establish community and outreach activities in electrochemical and solid state science and engineering education; and create and maintain robust membership bases. The Outstanding Student Chapter receives a plaque; certificates; $1,000; and recognition in ECS Interface or other electronic communications. Up to two Chapters of Excellence are also awarded.
Materials deadline: April 15, annually
ECS Summer Fellowships support student research from June through August in fields of interest to ECS. Winners of the Edward G. Weston, Joseph W. Richards, F. M. Becket, and H. H. Uhlig Fellowships (established in 1928) receive $5,000 and publication of a summary report in ECS Interface.
Materials deadline: January 15, annually
Georgia Section Outstanding Student Achievement Award recognizes academic accomplishment by ECS Georgia Section region university PhD students in any area of science or engineering in which electrochemical and/or solid state science and technology is the central consideration. The $500 prize was established in 2011.
Nomination period: May 30 – August 15, 2025
*All prize amounts are in US dollars unless otherwise stated.
Journal submissions for recognition in the scholarly record
Abstract submissions for biannual Meetings
Poster submissions for unique student opportunities
Battery Workforce Development for today & tomorrow
ECS is proud to announce the new members for January, February, and March 2025 Members are listed alphabetically by family/last name.
Saya Ajito, Sendai, Miyagi, Japan
Hannah Barad, Ramat Gan, Ramat Gan, Israel
Jordi Jacas Biendicho, Barcelona, CAT, Spain
Soumyajit Mandal, Merrick, NY, USA
Paulina Martinez, Barcelona, CAT, Spain
RFelix Richter, Giessen, HE, Germany
Mansour Toorani, Boston, MA, USA
YAndrew Yeang, Louisville, CO, USA
Yasaman AbdiSobbouhi, Tucson, AZ, USA
Oluwafemi Abubakar, East Lansing, MI, USA
Meenal Agrawal, Trondheim, Trøndelag, Norway
Bamigbola Akindehinde, Zaria City, Kaduna, Nigeria
Yaaqoub Al Houqani, Abu Dhabi, Abu Dhabi, UAE
Cade Alaniz, Flower Mound, TX, USA
Nicolò Albanelli, Pianoro, EMR, Italy
Abdallah Wusu-Gim Ali, Athens, OH, USA
Ahmad Alkhalaileh, Abu Dhabi, Abu Dhabi, UAE
Abdel Rahman Allan, Al Bahyah, Abu Dhabi, UAE
Ali Alshamsi, Al Hammrah, UAE
Sergio Arias, Lubbock, TX, USA
Zachary Asawesna, Rowland Heights, CA, USA
Padmavathy Bagavathi, Palakkad, KL, India
Megha Bala, Chennai, TN, India
Oscar Ballantyne, Newcastle upon Tyne, England, UK
P M Anuradha Bandaranayake, London, England, UK
José Barrera, Guadalajara, Jalisco, MX
Tara Barwa, Maynooth, Leinster, Ireland
Stephanie Bazylevych, Montréal, QC, CA
Seifeddine Bdey, Trois-Rivières, QC, CA
Akzhan Bekzhanov, Vienna, Austria
Vansh Bhutani, Delhi, DL, India
Dibora Birusew, Johannesburg, Gauteng, SA
Mark Blyth, Bristol, England, UK
Sandra Boakye, Youngstown, OH, USA
Timo Böhler, Ulm, BW, Germany
Nellie Bowen, Mishawaka, IN, USA
Abraham Brink, Ankeny, IA, USA
Kyra Burton, Rock Hill, SC, USA
Mitzi Camargo Arellano, San Juan del Rio, Querétaro de Arteaga, MX
Muriel Carter, Huntsville, AL, USA
Sooyon Chang, West Lafayette, IN, USA
Prerna Chaturvedi, Abu Dhabi, Abu Dhabi, UAE
Chen-Yi Chien, Hsinchu City, Hsinchu County, Taiwan
Eric Colon, Bryan, TX, USA
Victoria Colon-LaBorde, Aiken, SC, USA
Ana de Isidro-Gómez, Getafe, MAD, Spain
Samrudh Devanahalli Bokkassam, Bangalore, KA, India
Connor Dixon, Bryan, TX, USA
Laura Donk, Eindhoven, North Brabant, NL
Thomas Dore, London, England, UK
Cornelius Dorsogilaa, Boston, MA, USA
Yu Dou, Kirkland, QC, CA
Maria Duarte, Querétaro, Querétaro de Arteaga, MX
Saly El Srouji, Abu Dhabi, Abu Dhabi, UAE
Ingeborg Ellingsen, Trondheim, Norway
Marius Engler, Ilmenau, TH, Germany
Luis Esteban Bravo, Querétaro, MX
Nastaran Farahbakhsh, Siegen, NRW, Germany
Abdulfatai Faro, Gainesville, FL, USA
Catarina Ferraz, New York, NY, USA
Nausir Firas, Trabuco Canyon, CA, USA
Hamed Fooladvand, Tehran, Tehran, Iran
Stephen Fuller, Austin, TX, USA
Leyla Gillett, College Station, TX, USA
Luis Glahn, Platteville, BY, Germany
Guilherme Gomes Saddy, Munich, BY, Germany
Christopher Gonzalez, Austin, TX, USA
Ram Gotame, Miami, FL, USA
Michael Gumoshabe, Athens, OH, USA
HZahra Hagheh Kavousi, Montpellier, France
Yuyang Han, Los Angeles, CA, USA
Md Shahriar Hasan, Socorro, NM, USA
Afaq Hassan, Gdansk, Pomerania, Poland
Sara Hernández Rizo, Guadalajara, Jalisco, MX
Avery Hooks, Athens, OH, USA
Md Farhan Hossain, Rapid City, SD, USA
Md Mosaraf Hossain, Stillwater, OK, USA
Yu-Yao Huang, Hsinchu City, Taiwan
Victoria Iannelli, South Bend, IN, USA
Ayooluwa Ilesanmi, Columbia, MO, USA
Kashif Iqbal, St Andrews, Scotland, UK
Reem Irshaid, Abudhabi, Abu Dhabi, UAE
Pulkit Jain, Amherst, MA, USA
Yinke Jiang, South Bend, IN, USA
Duane Johnson, Ames, IA, USA
Jeyaseeelan K, Athens, OH, USA
Amarsingh Bhabu Kanagaraj, Abu Dhabi, UAE
Kshitij Kathait, New Delhi, DL, India
Supreet Kaur, Trondheim, Norway
Mojtaba Khakpour Komarsofla, Oshawa, ON, CA
Osama Khan, Abu Dhabi, UAE
Seda Köksal Yeğin, İstanbul, Turkey
Nikhil Komalla, State College, PA, USA
Sandra Koshy, Oshawa, ON, CA
Aniket Kumar, Chennai, TN, India
Vivek Kumar, Meerut, UP, India
Nishchitha Kushalappa, Hyderabad, TG, India
Hala Kutkut, Abu Dhabi, Abu Dhabi, UAE
Nicholas Lam, San Diego, CA, USA
Nils Lamouche, Montréal, QC, CA
Adriana Lara, Wake Forest, NC, USA
Giwook Lee, Suwon, Gyeonggi-do, ROK
Jeongwon Lee, La Jolla, CA, USA
Sangyeon Lee, Troy, NY, USA
Michael Li, Cambridge, MA, USA
Kameron Liao, Austin, TX, USA
Zongqi Liu, Notre Dame, IN, USA
William Lvovich, Cleveland Heights, OH, USA
Santhoshini M, Chennai, TN, India
Jiushi Ma, Ann Arbor, MI, USA
Christoph Macho, Munich, BY, Germany
Abir Mahmud, Calgary, AB, CA
Kajetan Majdański, Wrocław, Dolnośląskie, Poland
Leónard Mamba, Nelspruit, Mpumalanga, SA
Charles McDaniel, Bozeman, MT, USA
Hugo Medina Lopez, Monterrey, Nuevo León, MX
Emily Medved, Burlington, ON, CA
Peter Mejia, Tampa, FL, USA
Joshua Melendez-Rivera, College Station, TX, USA
Benjamin Meyer, Oxford, England, UK
Bahar Molavi, Montréal, QC, CA
Matthew Monroy, College Station, TX, USA
Iztvan Monroy-Solis, San Pedro Tlaquepaque, Jalisco, MX
Neha Lyka Muttumthala, Oshawa, ON, CA
NNgoc Nguyen, Albuquerque, NM, USA
Nathan Novak, Madison, WI, USA
ONicholas Obeng, Gainesville, FL, USA
Magnolia Pak, Irvine, CA, USA
Yanlin Pan, Zurich, ZH, Switzerland
Jongyoon Park, Suwon, Gyeonggi-do, ROK
Ziming Peng, St Andrews, Scotland, UK
Mauricio Peregrina Loza, Guadalajara, Jalisco, MX
Sreerag Ponnarassery Suresh Babu, Monterrey, Nuevo León, MX
Shambhu Pulikkiri Krishnan, San Nicolas, Monterrey, Nuevo León, MX
QZhensong Qiu, Urbana, IL, USA
RKelsey Ramp, Belleville, MI, USA
Amund Raniseth, Trondheim, Sor Trøndelag, Norway
Seth Reed, Austin, TX, USA
Pablo Rojas Arcos, Cadereyta Jiménez, Nuevo León, MX
Adam Ronderos, Frisco, TX, USA
Clara Rubio, Guadalajara, Jalisco, MX
Janik Ruppert, Münster, NRW, Germany
SAritro Sarker, Gainesville, FL, USA
Kazi Araf Sayeed, Bryan, TX, USA
Sebastian Schaeffer, Garching bei München, BY, Germany
Simon Schlehuber, Münster, NW, Germany
Arianna Serrano, Albuquerque, NM, USA
Daniella Servin, Querétaro, Querétaro de Arteaga, MX
Hibah Shafeekali, Muroor, Abu Dhabi, UAE
Majid Shahsanaei, Siegen, NRW, Germany
Kaiya Shealy, Corpus Christi, TX, USA
Sujay Shekar, South Bend, IN, USA
Sripradha Shekhar, Mumbai, MH, India
Lola Shmeis, Austin, TX, USA
Ekaterina Sofina, München, BY, Germany
Sheril Soni, Chennai, TN, India
Maria Stefoni, New York, NY, USA
Luca Stegemann, Münster, NRW, Germany
Jinrong Su, Ann Arbor, MI, USA
Aldina Sultana, Raleigh, NC, USA
Sara Sumbal, Gdańsk, Pomerania, Poland
Wen Sun, Hsinchu City, Hsinchu County, Taiwan
Jacquelyn Sundstrom, Mishawaka, IN, USA
Uttung Surange, Montréal, QC, CA
TPreety Thokchom, Abu Dhabi, Abu Dhabi, UAE
Rafael Tomey, Getafe, MAD, Spain
Xavier Torres, Houston, TX, USA
Danae Torres Medina, Guadalajara, Jalisco, MX
Elena Toups, Seattle, WA, USA
Nhan Huu Huy Tran, Northampton, MA, USA
Maria Trejo Espinosa, Corregidora, Querétaro de Arteaga, MX
UAdit Upadhyay, Bryan, TX, USA
VVijay Vaiyadurai, San Antonio, TX, USA
Brandon van Veenhuyzen, Mbombela, Mpumalanga, SA
Jithin Varghese, Monterrey, Nuevo León, MX
Karen Vega Navarro, Juárez, Nuevo León, MX
WDakotah Wagner, Auburn Hills, MI, USA
David Waligo, Houston, TX, USA
Na Wang, London, ON, CA
Yihan Wang, Montréal, QC, CA
Alyx Wilhelm, Vancouver, WA, USA
Anna Woeste, Ilmenau, TH, Germany
Chi-Wei Wu, Hsinchu, Taiwan
XZhongyu Xie, Cleveland, OH, USA
YSara Yaseen, Rahim Yar Khan, Punjab, Pakistan
Shabnam Yousefi, Athens, OH, USA
Dan Yu, Aalborg East, North Jutland, Denmark
Fan Yu, Montréal, QC, CA
Aneesa Zafar, Abu Dhabi, Abu Dhabi, UAE Muhammad Zahradeen, Dhahran, El Hasa Province, Saudi Arabia
Tianyu Zhao, Kingston, ON, CA
Aleksandra Zieminska, Władysławowo, Pomerania, Poland
The ECS British Columbia Student Chapter is a collaboration of members from Simon Fraser University (SFU) and the University of British Columbia (UBC) in Metro Vancouver, Canada. At UBC in winter 2023, the chapter introduced a new event highlighting local companies in the field of electrochemistry through panel discussions and networking with invited industry members. The Electrochemistry Career and Networking Day generated great interest and a large turnout, with more than 60 attendees (UBC and SFU undergraduate and graduate students and post docs) participating in lively discussions and networking.
The chapter was excited to host their second Electrochemistry Career and Networking Day on December 4, 2024, at SFU. Four experts from local electrochemical companies were invited as panelists for the event: Joel Kelly, Manager, Cell Technology, Moli Energy, a manufacturer of high-performance lithium ion batteries; Devproshad Paul, Senior Scientist, Ballard Power Systems, which manufactures hydrogen fuel cells; David Novitski, Director of Government Relations and Funding, Mangrove Lithium, which specializes in lithium extraction and cathode manufacturing; and Mehdi Shahraeeni, Greenlight Innovation, an electrolyzer testing methods and equipment company.
Each panelist shared their unique background, from their education to the paths they took to reach their current positions in industry. Much of the panel discussion focused on advice for transitioning from academia to industry; particularly, the panelists discussed topics such as skills valued by industry, differences in how research is conducted in academia and industry, future growth in the field, and different roles in the workforce. Following the panel, students and panelists participated in an open networking space. Students were able to connect directly with the speakers and students from different research groups and universities. The ECS British Columbia Student
Chapter hopes to make the Electrochemistry Career Day a regular annual event in Metro Vancouver to continue fostering connections between academia and electrochemical industries in the province.
Next, the chapter began planning its 13th Annual Young Electrochemists Symposium, which takes place this summer at UBC. This yearly full-day symposium typically consists of research seminars from academia and industry and a student research poster competition, then concludes with a social event and the annual election.
The ECS CECRI Student Chapter organized the first session of the Redox Rendezvous: Conversations in Electrochemistry series which featured an engaging conversation with Prof. Arumugam Manthiram, George T. and Gladys H. Abell Endowed Chair of Engineering, University of Texas, Austin. The globally renowned expert in materials science and rechargeable battery technologies shared his deep insights into the evolving landscape of energy storage and electrochemical systems. The session was marked by chapter members’ dynamic interaction with Prof. Manthiram. He responded
enthusiastically to student queries and offered thoughtful guidance on both scientific and professional topics. His approachable demeanor and willingness to share his research journey resonated deeply with attendees. Events like Redox Rendezvous serve as a vital platform for knowledge exchange and inspiration within the electrochemical science community. The chapter is immensely grateful to Prof. Manthiram for his time and contributions. We look forward to continuing these interactive sessions with leaders in the field.
On March 29 and 30, middle and high school students from across Texas came to Texas A&M to compete in the Texas Science & Engineering Fair (TXSEF). Texas A&M departments, student organizations, and industry partners hosted TXSEF student participants on March 28 at the Night at the ZACH extravaganza. More than 3000 guests attended to learn more about A&M and participate in hands-on activities designed and run by chapter members, including Vivek Dalai, Karan Deshpande, Laura Hoagland, Autumn Kudlack, Wynn Miholits, and Khirabdhi Mohanty. Students made pictures with Lite-Brites while learning how the circuit underneath powers lights in series, and how batteries generate electricity. Additionally, the students discussed their own science projects and discoveries. It was a great event for inspiring students to study science.
On April 5, the ECS University of Texas (UT) Student Chapter visited Texas A&M. Students led presentations and lab tours throughout the day. A&M and UT students learned in a casual environment about electrochemical research at A&M and gave attendees ideas to take home. ECS Texas A&M Student Chapter President Laura Hoagland explained Dr. Abdoulaye Djire’s lab’s work on nitride MXene synthesis and characterization, as well as in situ electrochemical catalysis and energy storage. Dr. Hugo Yuset Samayoa Oviedo represented Dr. Lane Baker’s research group, showing how their lab designs and utilizes novel analytical tools for performing scanning ion conductance microscopy on electrochemical systems. Dr. Francisco Alejandro Ospina-Acevedo shared Dr. Perla Balbuena’s group’s research and discussed their work in computational electrochemistry modeling. Dr. Ahmed Badreldin presented Dr. Ying Li’s lab and showed their work in carbon dioxide reduction on the lab scale and beyond. The chapter’s Vice President Autumn Kudlack showed the work of Dr. Jodie Lutkenhaus’ research group in organic polymers used for polyelectrolytes and solid state batteries.
Tours of the Djire, Lutkenhaus, and Baker labs followed the student presentations, allowing attendees to ask more detailed questions about research conducted in these spaces. This event coincided with Family Weekend, which further emphasized the connections forged between Texas student chapter members. This, the fourth annual visit between Texas A&M and the University of Texas, demonstrates how ECS brings students and researchers together.
ECS Texas A&M University Student Chapter member Karan Deshpande shows his support with a Gig ’Em while chapter Vice President Autumn
Through a variety of talks, journal clubs, and social gatherings over the 2025 winter semester, the ECS University of Michigan (U-M) Student Chapter actively fostered academic growth, meaningful discussion, and community involvement among the university’s electrochemists.
The chapter hosted a talk by Dr. Krista Hawthorne of the Chemical and Fuel Cycle Technologies Division at Argonne National Laboratory. In “Electrochemical Recycling of Used Nuclear Fuels to Support Advanced Reactor Deployment,” she discussed her group’s efforts to address aspects of the process of electrochemistry and engineering at multiple scales to support the industrialization of nuclear fuel pyro-processing. Next, the chapter hosted a virtual talk by Prof. Mim Rahimi from the Department of Environmental Engineering, University of Houston, on “Electrochemical Carbon Capture: Technological Advancements and Economic Considerations.” Dr. Rahimi described the emerging science and research progress underlying electrochemical carbon capture processes and assessed their current maturity and trajectory for carbon capture from various sources. The chapter also hosted a networking dinner with Prof. Matthew McDowell, Associate Professor at the Woodruff School of Materials Science and Engineering, Georgia Institute of Technology. Over dinner, Dr. McDowell shared advice regarding research directions, looking for jobs in industry and academia, his thoughts on the development of batteries, and more.
The chapter has devoted much effort to connecting U-M students with industrial partners. This semester, the chapter hosted a tour of the Bosch Industrial Hydrogen Research Lab for roughly 30 undergraduates and graduate students. Collaboration with the ECS
Detroit Chapter has also strengthened, with students attending and presenting at the ECS Detroit Section’s “Detecting Internal Short Circuit in Li-ion Cells with the Sahraei Failure Model” seminar. Moreover, the chapter also introduced a new series of ECS Student Chapter Presentations, in which students are welcome to present their research or collective efforts in addressing issues related to electrochemistry. In the first presentation, chapter President Rachel Silcox described the United Nations Climate Change Conference (COP 29) in Azerbaijan, which she attended in November, and connected the visit to her own research on carbon removal technologies. Finally, the student chapter also hosts a social event every month to enhance collaboration among electrochemists in the U-M community.
The ECS University of Michigan Student Chapter (U-M) hosts “Electrochemical Carbon Capture: Technological Advancements and Economic Considerations,” a talk by Prof. Mim Rahimi of the University of Houston.
On April 3, the ECS University of Nebraska-Lincoln Student Chapter held its annual meeting in a friendly and vibrant atmosphere. Members came together to celebrate the chapter’s achievements, share lunch, and reflect on the past year. The meeting included a review of the last year’s activities, a discussion of plans for the upcoming year, and elections of new officers. Chapter President Sahand Serajian led the meeting and was unanimously re-elected for a third term. Elections for other leadership roles followed, with Oghenetega Allen Obewhere elected as Vice President, Sarang Ismail as Treasurer, and Micah Quirie as Secretary.
The chapter gives a big thank you for the continued support of all their passionate members and advisors. We look forward to another exciting year of collaboration and engagement in the field of electrochemistry, both within and beyond the university community.
BioLogic, Knoxville, TN, US
Duracell US Operations, Inc., Bethel, CT, US
Gamry Instruments, Warminster, PA, US
PalmSens BV, Houten, Netherlands
Pine Research Instrumentation, Durham, NC, US
Scribner, LLC , Southern Pines, NC, US
BASi, West Lafayette, IN, USA
Center for Solar Energy and Hydrogen Research Baden-Wurttemberg (ZSW), Germany
Central Electrochemical Research Institute, Tamil Nadu, India
Corteva Agriscience, Indianapolis, IN, US
DLR – Institute of Engineering Thermodynamics , Oldenburg, Germany
EL-CELL GmbH, Hamburg, Germany
Electrosynthesis Company, Inc., Lancaster, NY, US
Ford Motor Company, Dearborn, MI, US
GS Yuasa International Ltd., Kyoto, Japan
Honda R&D Co., Ltd., Tochigi, Japan
Medtronic, Inc., Minneapolis, MN, US
Nel Hydrogen, Wallingford, CT, US
Nissan Motor Co., Ltd., Yokosuka, Japan
NSF Center for Synthetic Organic Electrochemistry, Salt Lake City, UT, US
Pacific Northwest National Laboratory (PNNL), Richland, WA, US
Panasonic Energy Corporation, Osaka, Japan
Permascand AB , Ljungaverk, Sweden
Plug Power, Inc., Latham, NY, US
Teledyne Energy Systems, Inc., Sparks, MD, US
UL Research Institutes, Northbrook, IL, US
easyXAFS, LLC , Renton, WA, US
Energizer Battery, Westlake, OH, US
Faraday Technology, Inc., Clayton, OH, US
GE Aerospace Research, Niskayuna, NY, US
Hydro-Québec, Varennes, QC, Canada
Lawrence Berkeley National Laboratory (LBNL), Berkeley, CA, US
Toyota Research Institute of North America (TRINA), Ann Arbor, MI, US
BMW Group, München, Germany
Current Chemicals, Cleveland, OH, US
General Motors Holdings LLC , Warren, MI, US
Giner, Inc., Newton, MA, US
Ion Power, Inc., New Castle, DE, US
Los Alamos National Laboratory (LANL), Los Alamos, NM, US
Metrohm USA, Inc., Riverview, FL, US
Microsoft Corporation, Redmond, WA, U
next Machinery Group | Coatema® Coating Machinery GmbH , Chadds Ford, PA, US
Occidental Chemical Corporation, Dallas, TX, US
Sandia National Laboratories, Albuquerque, NM, US
Sensolytics GmbH, Bochum, Germany
Sherwin-Williams, Minneapolis, MN, US
Spectro Inlets ApS , Copenhagen, Denmark
Technic, Inc., Providence, RI, US
United Mineral & Chemical Corporation, Lyndhurst, NJ, US
Western Digital Corporation, Tokyo, Japan
Westlake Corporation, Monroeville, PA, US
Help us continue the vital work of ECS by joining as an Institutional Partner today.
To renew, join, or discuss institutional partnership options, please contact Anna Olsen, Sr. Manager, Corporate Programs, sponsorship@electrochem.org
19th International Symposium on Solid Oxide Fuel Cells (SOFC-XIX)
Stockholm, Sweden
July 13–18, 2025
The Brewery Conference Center
249th ECS Meeting
Seattle, WA
May 24–28, 2026
Washington State Convention Center
248th ECS Meeting
Chicago, IL
October 12–16, 2025
Hilton Chicago
250th ECS Meeting
Calgary, Canada
October 25–29, 2026
BMO Centre