Autonomous experiments that integrate machine learning and robotics are reshaping materials research. By automating experimental workflows and efficiently searching high-dimensional parameter spaces, these approaches markedly accelerate materials discovery and process optimization.
Here, we report a modular self-driving laboratory (SDL) for solids and thin films [1–4]. The SDL orchestrates all stages of the experimental cycle—including sample transfer, synthesis, characterization, and iterative optimization. Data acquisition spans X-ray diffraction, scanning electron microscopy, Raman spectroscopy, electrical conductivity and optical transmittance measurements. A Bayesian optimization enables autonomous exploration of the parameter space and rapid identification of optimal conditions.
We demonstrate the platform by synthesizing thin films of TiO₂ and LiCoO
2
. We further show that the same workflow supports the discovery of new ionic conductors. These results highlight the potential of autonomous experimentation to accelerate research in solid-state materials. Ongoing efforts extend the SDL to bulk-materials synthesis, aiming to unify thin-film and bulk workflows within a single autonomous framework.
[1] "Autonomous experimental systems in materials science" N. Ishizuki, R. Shimizu, and T. Hitosugi, STAM Methods 3, 2197519 (2023).
[2] "Autonomous materials synthesis by machine learning and robotics" R. Shimizu, T. Hitosugi
et al.
, APL Mater. 8111110 (2020).
[3] “Autonomous exploration of an unexpected electrode material for lithium batteries“ S. Kobayashi, T. Hitosugi
et al.
, ACS Materials Lett. 5, 2711–2717 (2023).
[4] “Digital laboratory with modular measurement system and standardized data format” K. Nishio, T. Hitosugi
et al.
, Digital Discovery 4, 1734-1742 (2025).
Figure 1
Energetic materials store large amounts of chemical energy that can be rapidly released and directed in applications such as propellants, explosives, and pyrotechnics. Traditional experimental methods provide insight into bulk powder behavior; however, averaging effects in heterogeneous particle systems often obscure the fundamental mechanisms controlling energy release at the micro- and nanoscale. This dissertation resolves this limitation by developing automated and fully autonomous experimental platforms for high-throughput characterization of individual particle behavior. An initial automated framework was developed to investigate laser-induced reactions in aluminum microparticles using a combination of scanning electron microscopy and custom optical microscopy. This system achieved over a 100x increase in experimental throughput and enabled the determination of reaction thresholds as a function of particle size. Building upon this foundation, a fully autonomous optical platform was designed and implemented, integrating custom optics, computer-visionbased particle detection, machine-learning-based outcome classification, and precise motion control with synchronized laser excitation. This system enables the identification, targeting, and testing of individual particles without human intervention, permitting rapid data collection and statistical characterization of reaction behavior. A thermomechanical model was developed to interpret these results and investigate the roles of energy deposition, heat transfer, and material properties in governing particle reactivity. Together, these experimental and modeling efforts provide insight into laser-driven particle spallation and demonstrate how automated, data-driven experimentation can accelerate materials research.
Discovering and improving new semiconductor nanomaterials made in solutions is often a slow process that relies on trial and error. Traditional methods using batch reactors can be inconsistent, especially with heating and mixing, making it hard to explore all the possible ways to create and process these materials. Even though these nanomaterials have remarkable properties and are widely used in energy and chemical technologies as well as photonic devices, we need better approaches to speed up their discovery and development. Recent advances in reaction miniaturization, automated experiments, in-situ multi-modal characterization, and using machine learning (ML) for experimental planning offer exciting new opportunities to accelerate nanomaterials discovery and development. In my talk, I'll present how combining continuous-flow reactors with autonomous experimentation (what we call a Fluidic
Self-Driving Lab
) can accelerate research in colloidal nanoscience. By breaking down the steps of making nanomaterials and processing into separate modules, using methods that can run up to 100 experiments per minute, and applying ML to help model the processes in real-time and make informed decisions about future experiment(s), we can efficiently navigate complex and high-dimensional experimental spaces. Specific examples will be shared to show how these self-driving fluidic labs can autonomously and precisely create metal halide perovskites, as well as II–VI and III–V semiconductor nanocrystals, reducing the development timeline from more than a decade to just a few weeks.
Most computationally predicted materials are never synthesized because conventional synthesis optimization is slow, expertise-dependent, and iterative. Here we present a closed-loop framework that automates this expert workflow by placing human tacit knowledge in the loop through a large language model (LLM) that distills synthesis knowledge from the literature, high-throughput hyperspectral imaging for rapid film evaluation, and multi-objective Bayesian optimization guided by experimental feedback. In a paired optimization campaign, LLM-assisted initialization produced more Pareto-optimal samples and higher hypervolume than a Latin hypercube sampling baseline at matched trial counts, and this advantage persisted throughout iterative optimization. We demonstrate the framework by synthesizing the previously unreported perovskite-inspired compound Rb3BiI6 as thin films and validating the optimized films by optical bandgap analysis and X-ray diffraction. The framework transforms synthesis prediction from single-shot recommendation to iterative learning, providing a generalizable strategy to accelerate automated and fully autonomous experimental materials discovery.
Fang Sheng, Steven B. Torrisi, Amanda A. Volk et al.· 0 citations
The development of solid electrolytes with high ionic conductivity is crucial for advancing all-solid-state batteries. However, conventional materials discovery approaches are hindered by experimental inefficiencies and analytical bottlenecks. Here, we present a high-throughput experimental platform integrating composition-gradient thin-film synthesis, automated structural and electrochemical characterization, and machine-learning pipelines for exploring pseudo-ternary systems. Composition-gradient thin films on 4-inch Si wafers were fabricated by co-sputtering of three targets. Synchrotron X-ray diffraction (SXRD) at SPring-8 BL28XU, equipped with automated sample exchange and XY-stage positioning, allows rapid structural mapping. Non-negative matrix factorization (NMF) first extracts latent phase information as basis patterns with corresponding phase fractions. These basis patterns are subsequently clustered using DBSCAN with dynamic time warping (DTW) distance metrics, which effectively groups solid solutions exhibiting continuous peak shifts into single clusters. Electrochemical impedance spectroscopy was performed using an automated XY-stage with a Z-axis contact probe system. EIS analysis employs Bayesian-navigated equivalent-circuit model (ECM) fitting to ensure consistent and automated extraction of bulk and grain-boundary conductivities. To validate the platform, we investigated the CeF
3
–LaF
3
–SrF
2
pseudo-ternary system for fluoride-ion conductors. SXRD analysis revealed distinct formation regions for tysonite and fluorite structures. The ionic conductivity mapping revealed that Ce-rich tysonite exhibited bulk conductivities exceeding 10
–4
S cm
–1
, with values decreasing sharply in the two-phase region and reaching approximately 10
–8
S cm
–1
for the fluorite phase. This integrated approach establishes a framework for accelerated discovery and optimization of solid electrolytes.
Acknowledgements: This study was conducted using a grant from the project (JPNP21006) commissioned by the New Energy and Industrial Technology Development Organization (NEDO).
Naoki Matsui, Y. Shimo, So Fujinami et al.· ECS Meeting Abstracts· 0 citations
The Simulation-Calibrated Active Learning Estimator (SCALE), a closed-loop framework uniting high-throughput molecular dynamics, machine learning, and robotic synthesis to bridge the gap between simulation and experiment, is introduced.
Felix Arendt, T. Waurischk, Stefan Reinsch et al.· npj Computational Materials· 0 citations
This review examines how AI methodologies, ranging from machine learning‐assisted first‐principles simulations to deep‐learning analysis of experimental data, are reshaping the study of HfO‐based ferroelectrics to enable predictive design and autonomous optimization of next‐generation hafnia‐based ferroelectrics.
Faizan Ali, D. Lehninger, F. Sánchez et al.· Advanced Electronic Material...· 0 citations