Autonomous laboratories promise to accelerate chemical synthesis discoveries by automating measurements and decision-making. However, most current autonomous systems rely on bespoke equipment and limited characterization techniques, hindering their ability to mimic the multifaceted approach of human chemists. Human experimentation often involves a wider range of analytical instruments and decisions based on multiple measurements, a complexity that existing autonomous systems struggle to replicate. This research addresses this limitation by introducing a novel approach using mobile robots to integrate existing laboratory equipment into an autonomous workflow, thereby enabling more diverse and flexible experimentation. The study aims to demonstrate that a mobile robot-based system can perform exploratory synthetic chemistry, including supramolecular synthesis and photochemical reactions, in a manner that closely resembles human experimentation, leveraging multiple analytical techniques and making context-based decisions.

Previous research in autonomous laboratories has focused on bespoke automated equipment and single characterization techniques. Decision-making algorithms in these systems often operate on narrow data ranges, limiting their ability to handle the complex and diverse outcomes typical of exploratory synthesis. While progress has been made in automating synthesis platforms and enhancing autonomous capabilities, most platforms are expensive, complex, and monopolize analytical equipment. This often leads to a preference for single, fixed characterization techniques, limiting the analytical information available for decision-making and hindering the development of truly autonomous systems that mirror the complexity and flexibility of human experimentation. Prior work using mobile robots in chemical experiments has been limited in scope, focusing on specific chemistries and simpler analytical data.

This study developed a modular autonomous platform for general exploratory synthetic chemistry. The platform uses mobile robots to operate a ChemSpeed I synthesis platform, an ultrahigh-performance liquid chromatography-mass spectrometry (UPLC-MS), and a benchtop NMR spectrometer. The platform is divided into physically separated synthesis and analysis modules, with mobile robots facilitating sample transport and handling. Reactions are monitored by UPLC-MS and NMR, offering characterization comparable to manual experimentation. An algorithmic decision-maker, guided by chemist-defined heuristics, determines subsequent synthesis operations based on a binary pass/fail assessment of the MS and 1H NMR analysis. This heuristic approach allows for adaptability and the exploration of novel chemistry. The platform's modularity allows for expansion to include other equipment. The system was tested with several experiments, including parallel synthesis for structural diversity and autonomous discovery of host–guest assemblies, demonstrating its capabilities in complex chemical tasks. An offline photochemical synthesis module was also integrated to showcase the platform’s flexibility.

The developed platform successfully performed several complex chemical tasks. In the parallel synthesis experiment, the autonomous system successfully performed a multi-step synthesis of structurally diverse molecules, emulating a human medicinal chemist workflow, including screening, scale-up, and diversification. The system demonstrated the ability to make decisions based on multiple orthogonal analytical data (UPLC-MS and 1H NMR) and adapt to unexpected chemical outcomes. In the supramolecular chemistry experiments, the autonomous system successfully discovered new host–guest assemblies. The system successfully identified host-guest interactions by analyzing changes in 1H NMR spectra upon guest addition, despite the complexity of the supramolecular reactions and the potential for multiple product formation. The integration of an offline photochemical reaction module further demonstrated the platform’s flexibility and adaptability. Three photocatalysts were successfully identified for a decarboxylative conjugate addition reaction, showcasing the platform’s capability to handle diverse reaction conditions and characterization needs. The platform is shown to be extensible to a variety of chemical reactions and analytical methods.

The results demonstrate the feasibility of creating autonomous systems for exploratory synthetic chemistry that closely mirror human experimental workflows. The modular design and heuristic decision-making strategy overcome limitations of previous autonomous systems by enabling the integration of multiple analytical techniques and adaptable decision-making processes. The ability of the platform to handle unexpected outcomes and complex chemical reactions like supramolecular self-assembly and photochemical reactions highlights its potential for accelerating chemical discovery. The successful emulation of human decision-making processes, particularly in medicinal chemistry and supramolecular chemistry, signifies a significant advancement in the field of autonomous chemistry.

This research presents a significant advancement in autonomous chemical synthesis by introducing a modular, mobile robot-based platform capable of performing complex exploratory synthetic chemistry. The platform's flexibility, adaptability, and ability to integrate multiple analytical techniques represent a significant step towards truly autonomous laboratories. Future work could focus on refining the decision-making algorithms, incorporating more advanced AI techniques, expanding the range of supported chemical reactions, and further integrating the system with other laboratory equipment and techniques.

While the system demonstrates significant capabilities, some limitations exist. The current heuristic decision-making process relies on chemist-defined rules, potentially limiting its ability to discover truly unexpected or unconventional chemistry. The system's reliance on existing laboratory equipment could limit its adaptability to completely novel experimental setups. Further improvements in robotic dexterity and analytical capabilities could enhance the system's efficiency and capabilities. The current system's success criteria might need adjustment depending on the specific chemical reaction being explored.