New battery technology will be crucial to the electrification of transportation and aviation 1, 2 , but battery innovations can take years to deliver. For battery electrolytes, the many design variables present in selecting multiple solvents, salts, and their relative ratios [3][4][5][6][7] mean that optimization studies are slow and laborious, even those restricted to small search spaces. The key challenge is to lower the number and time-cost of experiments needed to formulate an electrolyte for a given objective.

Developing high-energy and efficient battery technologies is a crucial aspect of advancing the electrification of transportation and aviation. However, battery innovations can take years to deliver. In the case of non-aqueous battery electrolyte solutions, the many design variables in selecting multiple solvents, salts and their relative ratios make electrolyte optimization time-consuming and laborious. To overcome these issues, we propose in this work an experimental design that couples robotics (a custom-built automated experiment named "Clio”) to machine-learning (a Bayesian optimization-based experiment planner named "Dragonfly”). An autonomous optimization of the electrolyte conductivity over a single-salt and ternary solvent design space identifies six fast-charging non-aqueous electrolyte solutions in two work-days and forty-two experiments. This result represents a six-fold time acceleration compared to a random search performed by the same automated experiment. To validate the practical use of these electrolytes, we tested them in a 220 mAh graphite∣∣LiNi0.5Mn0.3Co0.2O2 pouch cell configuration. All the pouch cells containing the robot-developed electrolytes demonstrate improved fast-charging capability against a baseline experiment that uses a non-aqueous electrolyte solution selected a priori from the design space.

High-concentration aqueous electrolytes have shown promise as candidates for a safer, lower-cost battery system. Ionic conductivity is a key property required in high performing electrolytes; the Advanced Electrolyte Model (AEM) has previously shown great accuracy in predicting ionic conductivity in highly-concentrated non-aqueous electrolytes. This work provides extensive experimental data for mixed and highly concentrated aqueous electrolyte systems, rapidly generated via a robotic electrolyte testing apparatus. These data demonstrate exceptional accuracy from AEM in predicting conductivity in aqueous systems, with the accuracy being maintained even in highlyconcentrated and mixed-salt regimes.

<div> <div> <div> <p>Innovations in batteries take years to formulate, requiring extensive experimentation during the design and optimization phases. We approach the design of a battery electrolyte as a black-box optimization problem. We report here the discovery of a novel battery electrolyte by a robotic electrolyte experiment guided by machine-learning software. Motivated by the recent trend toward super-concentrated aqueous electrolytes for high-performance batteries, we utilize Dragonfly - a Bayesian machine-learning software package - to search mixtures of commonly used lithium and sodium salts for super-concentrated aqueous electrolytes with wide electrochemical stability windows. Dragonfly autonomously managed the robotic test-stand, recommending electrolyte designs to test and receiving experimental feed- back in real time. Within 40 hours of continuous experimentation, Dragonfly discovered a novel, high-performing aqueous sodium electrolyte that a human-guided design process may have missed. This result demonstrates the possibility of integrating robotics with machine-learning to rapidly and autonomously discover novel battery materials.</p></div></div></div>

A fully automated, computer-controlled test stand capable of rapidly creating and electrochemically characterizing any arbitrary liquid electrolyte solution is described. Hundreds of different electrolytes were studied, and the results were used to verify the precision and accuracy of the system. To test the functionality of the approach, several 2-dimensional co-solvated electrolyte solutions containing blends of aqueous sulfates and nitrates were rapidly created and examined automatically. The test stand took less than a day to conduct these searches, while conventional manual methods would have taken much longer. The demonstrated standard error of the test-stand was 0.5 mS/cm on conductivity and 0.02 V for voltage stability window measurements, and several of the combinations studied revealing surprisingly high voltage stability and conductivity values. The demonstrated success of the test-stand in a 2-dimensional search spaces shows the promise of conducting high speed co-optimization studies of liquid electrolytes in particular when used in concert with a machine learning-based real time/in-loop data assessment computational package.