Autonomous battery research: Principles of heuristic operando experimentation

Abstract

Unravelling the complex processes governing battery degradation is critical to the energy transition, yet the efficacy of operando characterisation is severely constrained by a lack of Reliability, Representativeness, and Reproducibility (the 3Rs). Current methods rely on bespoke hardware and passive, pre-programmed methodologies that are ill-equipped to capture stochastic failure events. Here, using the Rutherford Appleton Laboratory's multi-modal toolkit as a case study, we expose the systemic inability of conventional experiments to capture transient phenomena like dendrite initiation. To address this, we propose Heuristic Operando experiments: a framework where an AI pilot leverages physics-based digital twins to actively steer the beamline to predict and deterministically capture these rare events. Distinct from uncertainty-driven active learning, this proactive search anticipates failure precursors, redefining experimental efficiency via an entropy-based metric that prioritises scientific insight per photon, neutron, or muon. By focusing measurements only on mechanistically decisive moments, this framework simultaneously mitigates beam damage and drastically reduces data redundancy. When integrated with FAIR data principles, this approach serves as a blueprint for the trusted autonomous battery laboratories of the future.

Figures (14)

• Figure 1: Capturing stochastic events: Conventional vs. Heuristic Operando. (a) Conventional operando often misses stochastic failure precursors (red star) that occur between pre-programmed, fixed-cadence scans (blue circles). (b) Heuristic Operando uses an AI (trained on physics-based digital twins) to identify precursors in continuous low-resolution monitoring data (yellow diamonds). This detection triggers a targeted, high-cadence scan (pink pentagons), deterministically capturing the precursor event (green star).
• Figure 2: Resolving the 3R (Reliability, Representativeness, Reproducibility) Compromise. (a) Current Methodology: The fundamental hardware trade-off. Researchers are forced to choose between high-Representativeness commercial cells (e.g., Unmodified Pouch), which suffer from poor signal quality (low Reliability), or high-Reliability specialised cells (e.g., POLARIS, DRIX), which lack commercial relevance. (b) Heuristic Operando: The proposed framework breaks this compromise. The AI Pilot makes commercial cells viable by extracting faint precursor signals (boosting Reliability) and upgrades specialised cells by enabling efficient multiplexing and targeted scans (boosting Representativeness).
• Figure 3: Closed-loop architecture of the Heuristic Operando Framework.Exascale Zone: Physics-based Digital Twins generate synthetic training libraries for architecture-agnostic surrogates, deploying probabilistic priors ($P(M)$) to the edge. Edge Zone: The AI Pilot computes the Entropy-Scaled Measurement Efficiency ($E_{\mathrm{SME}}$) from real-time monitoring ($D_{mon}$). A control signal ($u_t$) is triggered only when the information gain exceeds the experimental cost ($C$). Experiment Zone: The instrument defaults to low-dose monitoring (State A) to mitigate beam damage. Signal $u_t$ activates targeted characterisation (State B), yielding high-fidelity data ($D_{high}$) and identifying anomalies ($m_{\emptyset}$) for model refinement.
• Figure 4: CAD drawing of POLARIS cell
• Figure 5: Schematic of the NIMROD cell, taken from Figure 15 of Shah et al.'s paper shah