Please note that the workshop has reached its maximum capacity, applications are now closed.

A shift away from fossil fuels is urgently needed and, as efficient energy storage systems, batteries and supercapacitors may become key components in this "green energy transition". Battery and supercapacitor production has been growing exponentially over the past decade [1,2], and these devices are now applied commercially, for instance, in electric vehicles [3] and large-scale energy storage systems [4,5]. Although the first batteries and supercapacitors were developed decades ago and several successful models were developed, there are still many unresolved challenges in capturing all the multi-physical and multi-scale processes at play during charge-discharge cycles. Developing better models can lead to a deeper understanding of these devices, which, in turn, will help in their rational design and further optimization.

Electrochemical devices usually store energy either (i) via electrical double-layer formation or (ii) via charge transfer reactions. The former is a purely capacitive process whereby ions adsorb and desorb at the interface between an active material and an electrolyte [6,7]. The latter corresponds to faradaic, i.e. redox, reactions and can involve ion insertion/intercalation, underpotential deposition, or material conversion [8]. Supercapacitors mainly operate through the first charging mechanism, which usually confers them high cyclability and power densities but mediocre energy density. Strategies to improve supercapacitor energy density include decreasing their mass or increasing their electrochemical window. Another strategy, by analogy with dielectric capacitors, is to increase the specific surface area of the electrodes—for instance, by nanostructuring them. However, several studies found that the capacitance does not scale linearly with the specific surface area [9]. Indeed, the capacitance of supercapacitors depends in an intricate way on the materials they are made of [10-12], especially when the active materials are filled with concentrated electrolytes such as ionic liquids or water-in-salt electrolytes [13]. Other promising energy storage materials under active development include [14,15]: (i) materials that enable intrinsic pseudocapacitance through fast redox reactions and transport; (ii) materials whose tuned nanoscale structure increases surface-controlled contributions, especially intercalation-based energy storage materials, to enable extrinsic pseudocapacitance.

Aiming at high-energy density applications, the modeling community further studied the factors determining the capacitance of supercapacitors [16-18]. State-of-the-art methods for constant potential simulations, developed for freely available molecular dynamics (MD) codes [19-22], have proven to be helpful for this task. With their efficient implementations for calculating the Coulomb interactions, these methods can represent the electrical double layer physically correctly [23,24] and account for semi-metallic character of electrode materials [25,26]. Classical MD and ab initio simulations are notoriously computationally expensive, limiting them to systems of a few nanometers at most. By contrast, mesoscopic and continuum approaches can describe much larger systems and full devices. Traditionally, electrolyte dynamics have been modeled through the Poisson-Nernst-Planck (PNP) equations, which continue to be widely used today [27].

Recently, machine learning approaches that bridge the gap between macroscopic properties and molecular structure have emerged as an additional scale in the computer-assisted design of advanced battery materials and are employed to predict effective electrode compositions and electrolyte additives and to estimate service life aspects of battery systems [28,29]. Another data-driven application field is the development of machine learned force fields which can be of ab initio accuracy [30] but which allow for a much faster evaluation of the forces and thus simulations of relatively large electrochemical systems. Such force fields can allow for accurate chemical reactions during the simulations (see the chapter on reactions in the recent review of ref. [31]), a clear improvement over traditional force fields.

None of the above models and methods alone can fully describe the charging of batteries, supercapacitors, and pseudocapacitors. The idea of this workshop is to bring together experts in atomistic modeling, multiscale and mesoscopic simulations, and artificial intelligence to overcome the obstacles posed by the multiple phenomena that occur at complex electrochemical interfaces and facilitate the combination of simulations at complementary scales.

As a consequence, questions that the workshop aims to address are both technical and fundamental:
• How to incorporate electrochemistry in simulations of relatively large systems?
• How to include nanoscopic mechanisms into existing analytical models and judge the validity of existing analytical models for nanoscopic systems?
• How can machine learning help to improve the design of batteries and supercapacitors with, e.g., machine learning potentials?
• How to improve machine-learned prediction of quantities based on structure-property relationships of electrolyte additives?
• How to improve constant potential methods for heterogenous electrodes?
• How can data-driven techniques be leveraged to predict the behaviour of electrode materials?