At-Scale Data-Driven Exploration of High-Voltage Cathode-Active Materials for Sodium Batteries

Suchona Akter, Mohammad R. Momeni · 2026

Read At-Scale Data-Driven Exploration of High-Voltage Cathode-Active Materials for Sodium Batteries by Suchona Akter, Mohammad R. Momeni (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-27229.txt.

Introduction

Akter and co-authors hunt for high-voltage cathodes for sodium-ion batteries using only charged structures for training. A committee of their four best models predicts mean voltage, and uncertainty scores steer new costly first-principles data toward structures the committee finds least certain. Applying the trained models to large host databases yields NaCAM-Atlas, with 14843 candidate materials above 3 volts. A top subset was then checked with explicit voltage, stability, and electronic calculations.

Catalog record

CREATORSuchona Akter, Mohammad R. Momeni
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE45097
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.27229

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

At-Scale Data-Driven Exploration of High-Voltage Cathode-Active Materials for Sodium Batteries Suchona Akter, Mohammad R. Momeni Abstract Sodium-ion batteries (SIBs) share similar electrochemistry with Li but offer several advantages, including high abundance in nature and low cost, as well as suitability for fast charging due to a Na-ion mobility higher than that of Li. The development of high-voltage SIBs heavily relies on the discovery of novel, robust cathode-active materials (CAMs). All-inorganic materials represent the most mature and practical choice as CAMs for next-generation SIBs; however, their family spans a vast and chemically diverse space. In this work, we present a large-scale, chemically validated database of stable materials for SIB cathode discovery, curated from four major databases: Materials Project, AFLOW, OQMD, and GNoME. Generalizable and transferable descriptor-based machine learning (ML) models are developed based on a dataset of charged-only structures rather than charged/discharged pairs. Using a committee of the top four trained ML models, average voltage and specific capacity are predicted as target properties. Finally, a subset of top-ranked candidate CAMs is validated through explicit, high-throughput first-principles calculations of voltage profiles, phase stability, structural robustness upon sodiation/desodiation, and electronic properties. Together, this integrated data curation, ML ranking and predictions, and first-principles validation strategy establishes a scalable and transferable framework for accelerating the discovery of stable, high-voltage CAMs for SIBs and beyond. † † email: [email protected] † † affiliation: Division of Energy, Matter and Systems, School of Science and Engineering, University of Missouri - Kansas City, Kansas City 64110, MO United States † † suppinfo: D 1 1. INTRODUCTION 1. INTRODUCTION Building a low-carbon society based on clean and sustainable energy has become a critical scientific challenge of the 21 st century. Tarascon and Armand 2001 ; Armand and Tarascon 2008 ; Goodenough and Kim 2011 In this quest for renewable energy, electrochemical energy storage plays a central role given the intermittent nature of solar and wind. He et al. 2012 Among available technologies to date, lithium-ion batteries (LIBs) have become the leading platform for energy storage, thanks to their high energy density, long cycle life, and high operating voltages. Manthiram 2020 ; Goodenough and Park 2013 ; Tarascon and Armand 2001 ; Li et al. 2018 Despite their commercial success, Li is considered a semi-rare element with a highly uneven distribution worldwide. You and Manthiram 2018 The environmental impact of Li extraction is of significant concern as well. Phogat et al. 2025 Simultaneously, the rapid growth in demand for consumer electronics and stationary grid energy storage is driving the evolution of new rechargeable ion battery technologies toward lower cost, higher capacity and voltage, longer cycle life, faster charging, and improved safety, among others. Song et al. 2021 In this regard, sodium-ion batteries (SIBs), which share similar electrochemistry with LIBs, offer several advantages, particularly their high abundance in nature and low cost. Jeong et al. 2024 Na, the sixth most abundant element in the Earth’s crust, is distributed almost without limits. de la Llave et al. 2016 The significantly lower cost of the trona mineral ( \approx $135-165 per ton) used to produce sodium carbonate, compared with lithium carbonate ( \approx $5000 per ton in 2010), strongly supports the development of SIBs as viable alternatives to LIBs. Reddy and Linden 2019 ; Hwang et al. 2017 To enable practical implementations, next-generation SIBs must combine higher energy and power densities with improved cycle life, superior safety, and reduced cost. These parameters are intrinsically linked to the crystal structure and chemical composition of the electrode active materials, in which intercalation-deintercalation of working ions occurs during discharging-charging. CAMs, as the principal determinant of a battery’s core performance metrics, account for approximately 40% of the device’s overall cost. Huang et al. 2024 Specifically, SIB cathodes face two major challenges. First, SIBs have lower gravimetric energy density than LIBs due to Na’s higher atomic mass. Satrughna et al. 2023 Second, the Na + ion has a larger ionic radius (1.02 Å) compared to that of the Li + (0.76 Å), causing larger volume changes during charge and discharge cycles, resulting in possible deterioration of the structural stability. Hwang et al. 2017 As such, cathode host materials for Na + ions are often prone to phase transitions and deformations, leading to capacity decay and low power density. Gupta et al. 2022 ; Satrughna et al. 2023 Therefore, the identification and optimization of new CAMs capable of reversibly intercalating/deintercalating Na + ions with high capacity and stability remain a major research focus. Extensive research efforts to date have led to the development of diverse families of CAMs for SIBs, most notably sodium layered transition-metal oxides (LTMOs, Na x TMO 2 ) and polyanionic compounds. Rudola et al. 2021 ; Bauer et al. 2018 ; Broux et al. 2019 ; Peng et al. 2023 ; Yadav et al. 2023 Layered Na x TMO 2 materials (TM = Ti, V, Cr, Mn, Fe, Co, Ni, and a mixture of 2 or 3 elements) Han et al. 2015 are distinguished by their comparatively high energy densities, making them widely applicable in energy storage systems. The feasibility of electrochemistry in single transition-metal layered oxides was first demonstrated in the 1980s. Delmas et al. 1981 The specific stacking sequence of the oxide layers and the coordination environment of the alkali ion give rise to different polymorphs. In contrast to octahedral O-type Li-ion layered oxides, Na-ion oxides exhibit greater structural versatility and are generally classified into two main groups: the O3 type, where Na ions occupy octahedral sites, and the P2 type, where Na ions reside in prismatic sites. Wang et al. 2025 ; Yabuuchi et al. 2012 The P2-type oxides are found to offer better structural stability, higher Na-ion conductivity, and more open channels for ion diffusion compared to O3 structures, resulting in lower diffusion barriers and better power density and cycling stability. Lavela et al. 2024 Figure 1: Adapted workflow in this work. A dataset comprised of charged structures from the MP battery database is used to train descriptor-based ML models. The trained models are then applied to a carefully curated materials database collected from MP, AFLOW, OQMD, and GNoME to predict voltage and specific capacity using a committee of top-ranked ML models. In addition to LTMOs, polyanionic materials are pivotal as CAMs for SIBs and can be classified by their crystal structures as olivine Ali et al. 2016 , alluaudite Trad et al. 2010 , tavorite Boivin et al. 2017 , Na 3 V 2 (PO4) 2 F 3 Massa et al. 2002 , and NA-Super-Ionic-CONductor Goodenough et al. 1976 (NASICON)-type structures. Phosphate-based materials can be categorized into four subgroups: (i) single-phosphate compounds (NaMPO 4 , M = transition metal), (ii) NASICON-type phosphates (Na x M 2 (PO 4 ) 3 ), (iii) pyrophosphates and mixed polyanions (Na 2 MP 2 O 2 and Na 4 M 3 (PO 3 ) 2 P 2 O 7 ), and (iv) fluorophosphates. Li et al. 2017 Among these, NASICON-type materials have been extensively studied as CAMs for SIBs. First reported in 1976 Goodenough et al. 1976 , NASICONs have the general formula A n M 2 (XO 4 ) 3 , where A denotes an alkali-metal ion or vacancy, M is a transition metal or

Read the complete text