A self-driving laboratory advances the Pareto front for material properties

Abstract

Useful materials must satisfy multiple objectives, where the optimization of one objective is often at the expense of another. The Pareto front reports the optimal trade-offs between these conflicting objectives. Here we use a self-driving laboratory, Ada, to define the Pareto front of conductivities and processing temperatures for palladium films formed by combustion synthesis. Ada discovers new synthesis conditions that yield metallic films at lower processing temperatures (below 200 °C) relative to the prior art for this technique (250 °C). This temperature difference makes possible the coating of different commodity plastic materials (e.g., Nafion, polyethersulfone). These combustion synthesis conditions enable us to to spray coat uniform palladium films with moderate conductivity (1.1 × 105 S m−1) at 191 °C. Spray coating at 226 °C yields films with conductivities (2.0 × 106 S m−1) comparable to those of sputtered films (2.0 to 5.8 × 106 S m−1). This work shows how a self-driving laboratoy can discover materials that provide optimal trade-offs between conflicting objectives.

Autonomous optimization of non-aqueous Li-ion battery electrolytes via robotic experimentation and machine learning coupling

“…Research efforts on applying improved machine-learning methodologies to plan experiments in complex design spaces may unlock greater overall efficiencies than focusing on laboratory automation alone. Already, literature has demonstrated AFs above 20× to 1000× 11 , 12 , 41 , implying the potential for higher AFs when Clio searches complex design spaces.…”

“…The hope is that “closed-loop” approaches (i.e., the automated execution of experiments coupled directly to an experiment planner, working in tandem to achieve a goal without human operator influence) display the following traits when compared to the standard design of materials via human-operated experimentation: (1) closed-loop experiments are able to discover optimal material designs within a given design space; (2) closed-loop experiments discover optima faster and with fewer experiments; (3) closed-loop experiments offer a principled basis for design-of-experiments (DOE), balancing exploiting design regions likely to have optimal performance with exploring regions of unknown performance. These traits have been demonstrated in related fields 3 , 6 – 8 , 10 , 11 , 15 , but have not yet been demonstrated in battery material design outside of aqueous electrolytes 16 .…”

A self-driving laboratory designed to accelerate the discovery of adhesive materials

“…We performed this optimization by coupling our automated workow to a Bayesian optimizer. 15,16 Bayesian optimizers combine a surrogate model with an acquisition function. The surrogate model predicts the response of an experiment to the manipulated variables.…”

“…Such extensions would be synergistic with hardware upgrades. 12,16 For example, incorporating an oven into the robot would enable the optimization of thermal curing protocols. Expert knowledge (such as the range of resin ratios likely to yield high strength) or bond strength estimates based on simulations 21 could also be incorporated to increase the efficiency of the Bayesian optimization.…”

“…3a and j). We built on our previous self-driving laboratories, 15,16 by constructing the system described here using a 4-axis robotic arm (N9, North Robotics) (Fig. 4 and S1†).…”