Laboratoire de Réactivité et Chimie des Solides

Mesoscale particle simulations play a crucial role in investigating the manufacturing processes of electrochemical energy devices. They enable physical insights into the processes while building datasets to train AI surrogates for electrode microstructure and manufacturing optimization. Hence, the automation of parameter optimization for mesoscale simulations remains imperative. Optimization algorithms that do not require prior knowledge or experience simulating a specific chemistry/system are the ideal solution. One such optimizer is the Particle Swarm Optimizer (PSO). However, in the classical serial implementation, the evaluation time varies across different parameter sets. This means the optimizer has to wait for all simulations to finish at each iteration, which can lead to unnecessary use of computational resources. This study presents an optimization framework that implements Asynchronous PSO (APSO), enabling more efficient resource use to find the global optimum, i.e., parameter values that yield model outputs that agree with experiments. The effect of different APSO configurations on convergence was tested. The framework was then used to find the optimal parameters for two different mesoscale Discrete Element Method simulations. The data acquired from the automated search were then analyzed to study the models' behavior within the chosen parameter search space.

Redox-Targeting Flow Batteries (RTFBs) are promising alternatives to classical vanadium-based batteries for large-scale and stationary energy storage. Typically, RTFBs are marked by higher energy densities thanks to the addition of solid boosters within aqueous systems, taking care to limit the use of critical raw materials. This work subsequently investigates the case of sodium titanium phosphate (NTP, NaTi2(PO4)3, 132.8 mAh/g) as a potential booster material for the negolyte side of aqueous RTFBs. Pure NTP and carbon-coated NTP (C-NTP) particles were synthesized and characterized by various techniques (X-ray, TEM, TGA, Raman). So-obtained NTP and C-NTP particles were found to be suitable for creating innovative porous composite boosters formed as centimeter-sized granules by dry processing. Porous composite granules with an open porosity of 65% and 50 wt % of immobilized NTP or C-NTP were successfully produced by an extrusion–dissolution process using a regular PVDF binder and PEO as a porogen agent. Subsequently, intensive electrochemical tests were performed using an innovative dual-mediator reaction system (Fe-Tiron and 2,7-AQDS). High NTP reactivity, with booster utilization rates of up to 84% of its theoretical capacity, can be achieved under flow conditions, with an increase in volumetric capacity by a factor of 1.5, from 4 Ah L–1 to 6 Ah L–1. The mediator concentration (10 – 100 mM) and the mediator/booster ratio (0.5 – 1) play key roles in NTP reactivity. The fundamental work also highlights the benefit of C-NTP, allowing higher reactivity at low mediator concentrations. The study consequently validates the potential of NTP as an interesting booster material in future RTFB applications, with its scalable extrusion–dissolution technique to create innovative porous booster granules.

Poly(ethylene oxide) (PEO)-based solid polymer electrolytes typically suffer from limited ionic conductivity at near-room temperature and often require inorganic reinforcement. Halide solid-state electrolytes such as Li3InCl6 (LIC) offer fast Li+ transport but are moisture-sensitive and typically require pressure-assisted densification. Here, we fabricate a flexible LIC–PEO composite electrolyte via slurry casting in acetonitrile with a small amount of LiPF6 additive. The free-standing membrane delivers an ionic conductivity of 1.19 mS cm−1 at 35 °C and an electrochemical stability window up to 5.15 V. Compared with pristine LIC, the composite shows improved moisture tolerance, and its conductivity can be recovered by mild heating after exposure. The electrolyte enables stable Li|LIC–PEO|Li cycling for >620 h and supports Li|LIC–PEO|NCM111 cells with capacity retentions of 84.2% after 300 cycles at 0.2 C and 80.6% after 150 cycles at 1.2 C (35 °C). Structural and surface analyses (XRD, SEM/EDX, XPS) elucidate the composite microstructure and interfacial chemistry.

This paper focuses on the coupling of the Differential Voltage Analysis (dV/dQ vs Q) and the Distribution of Relaxation Times (DRT), whose benefits to electrodes development are illustrated by the tracking of degradation effects in the cycling aging of NMC811/Gr coin cells, with and without an NMC electrode coating. The former technique allows monitoring of individual electrode loss of active material (LAM), chemistry change and state of charge (SOC) alignment of the electrodes, whose variation is related mainly to loss of Li inventory (LLI). DRT is a time domain representation of the impedance that offers a high resolution of the resistance processes, complementing the DVA analysis by bringing kinetics degradation of the cells and by identifying in which electrode the main resistance processes are originated. The protocol inserting those techniques in an aging regime is a non-invasive approach, and it indicated the consumption of Li + ions as the main reason for capacity decay in both cells, with reduced or minimal NMC LAM and chemistry change when coating is applied. From the perspective of power decay, the coating decelerated the impedance rise (IR), mostly related to resistance processes in the positive electrode. • DVA and DRT coupling approach tracked electrodes contribution to degradation. • NMC particle coating prevented loss of mass and slowed down cell impedance increase. • Non-invasive operando approach tracks the evolution of the degradation effects.

This work focuses on a novel human‐centered digital assistant combining Mixed Reality (MR), Computer Vision, and Machine Learning regression to guide professionals and students on how to operate and correctly parameterize battery manufacturing machinery. This article aims to provide a proof of concept of the digital assistant, for a process involving an operator interacting with a semi‐manual electrode calendering machine and examining how the intended calendering parameters will impact the electrode properties after calendering. The operator performs his/her actions while the digital assistant automatically detects the parameters entered by him/her on the machinery. Then, the digital assistant provides real‐time predictions to the operator, helping him/her in decision‐making through a minimalistic holographic interface minimizing visual hindrance. The ergonomics of the solution is optimized by acquiring feedback from users with various experience levels and evaluating their performance. As the necessity for advanced energy storage solutions rises, there is a strong need for modern training and guidance tools that aid in complex battery manufacturing processes at the prototyping stage, involving both semi‐manual and automatic activities. Thanks to MR, the digital assistant has strong potential to improve manufactured electrode and cell qualities by emphasizing the egocentric perspective along the battery cell prototyping process.

To improve the performance of high‐energy‐density electrode materials for all‐solid‐state fluoride‐ion batteries (ASSFIBs), it is important to understand the structure and phase evolution during operation, which is closely correlated to capacity fading. In this study, an operando cell is designed compatible with laboratory X‐ray diffraction (XRD) to monitor real‐time structural changes of bismuth trifluoride (BiF3) cathodes and degradation of the ionic conductor BaSnF4 under negative potentials at 100 °C. Supported by ex‐situ XRD, our results reveal a multi‐step defluorination of BiF3: from orthorhombic (o‐BiF3) to cubic (c‐BiF3), then to distorted orthorhombic (o’‐BiF3), and finally to metallic bismuth (Bi), indicating partial intercalation‐type character. Formation of bismuth oxidefluoride (BiOF) beyond 200 mAh g−1 is attributed to oxygen impurities introduced via solid‐state synthesis. operando X‐ray absorption spectroscopy (XAS) confirms a continuous reduction of Bi3+ to Bi0 with intermediate phases. Rietveld refinement quantifies the phase fractions and structural transitions, enabling a model for BiF3 defluorination. Comparison of operando XRD and XAS reveals that BaSnF4 contributes transport of both fluoride and oxygen impurities, leading to BiOF formation. BaSnF4 also exhibits a broad stability window, with degradation occurring below −200 mV, rather than the expected −50 mV vs. Sn/SnF2.

Laboratory practices are essential to prepare students and professionals to drive future innovations in the field of energy storage and conversion. However, universities and industries working in the battery field encounter challenges concerning effective training on battery production complexities, mostly due to the lack of access to battery prototyping facilities or its limited availability for training purposes. This concept introduces an innovative educational platform in virtual reality (VR) named battery manufacturing metaverse (BMM). BMM promotes accessibility and collaborative learning of lithium–ion battery (LIB) manufacturing through an interactive and flexible VR representation of a LIB manufacturing pilot line. It enables collaboration among individuals from different geographical locations. Users can explore electrode and cell chemistries, and adjust manufacturing parameters with informative feedback from a cell's composition to the functioning of the manufacturing equipment. BMM does it with real‐time collaboration using avatars and voice chat. This platform aims to connect students of Interdisciplinarity in Materials for Energy Storage and Conversion, an Erasmus+ MSc. Program, enabling seamless knowledge sharing and training. BMM represents a transformative step in battery research and education, offering an immersive, interactive environment without geographical barriers to pave the way towards global education and safe training in the energy sector.