Robotics and automation can be helpful in chemical research, in particular, for large-scale screenings in which similar experiments are performed many times. While there are many examples, e.g., of automated organic synthesis, automation in the chemical research of materials has lagged behind. This is due to the many different types of samples and laboratory techniques that can be involved in this research.
Andrew I. Cooper and colleagues, University of Liverpool, Liverpool, UK, have developed a mobile, autonomous robot that can work in a conventional laboratory and combined it with an artificial intelligence (AI) approach to automate chemical experiments. The team used a robot arm that was mounted to a mobile base and equipped with a multipurpose gripper. The height and reach of the robot are similar to those of a human, which allows it to work in an unmodified, standard laboratory. It navigates within the lab using laser scanners and can perform precise manipulations based on data from additional touch sensors.
The researchers used the robot to search for improved semiconductor photocatalyst systems for water splitting. These catalysts usually require a sacrificial “hole scavenger” to produce hydrogen from water. Organic amines have generally been used for this, but they are irreversibly decomposed in the reaction. The team tried to find alternative hole scavengers. For this, the robot operated a solid-dispensing station that weighs the photocatalyst into sample vials, a liquid-dispensing station to add solutions of the hole-scavenger candidates, a capping station to close the vials, a sonication station that disperses the solid in the solution, and a photolysis station where the samples are irradiated with light. Finally, the robot used a gas chromatography station to quantify the produced hydrogen and then stored the samples.
After an initial screening of 30 candidates, the team used a Bayesian search algorithm to optimize several variables of the catalytic system at once. Based on this algorithm, the robot autonomously performed 688 experiments in eight days. The catalytic performance was improved by a factor of six. The researchers estimate that a human researcher using a manual approach might need months to perform a similarly detailed investigation. While the team’s workflow took a long time to set up (about two years), the researchers point out that the developed protocols could be transferred to other laboratories and research problems much faster. In many cases, the robot can use standard laboratory equipment, so it should be easy to add further stations and instruments.
- A mobile robotic chemist,
Benjamin Burger, Phillip M. Maffettone, Vladimir V. Gusev, Catherine M. Aitchison, Yang Bai, Xiaoyan Wang, Xiaobo Li, Ben M. Alston, Buyi Li, Rob Clowes, Nicola Rankin, Brandon Harris, Reiner Sebastian Sprick, Andrew I. Cooper,