Abstract
The development of the internet of things has led to an explosion in the number of networked devices capable of control and computing. However, whilst common place in remote sensing, these approaches have not impacted chemistry due to difficulty in developing systems flexible enough for experimental data collection. Herein we present a simple and affordable (<$500) chemistry capable robot built with a standard set of hardware and software protocols that can be networked to coordinate many chemical experiments in real time. We demonstrate how multiple processes can be done with two internet-connected robots collaboratively, exploring a set of azo-coupling reactions in a fraction of time needed for a single robot, as well as encoding and decoding information into a network of oscillating reactions. The system can also be used to assess the reproducibility of chemical reactions and discover new reaction outcomes using game playing to explore a chemical space.
Introduction
Traditional chemical automation is often prohibitively expensive, with systems typically costing between $50,000 and $500,000. These platforms often operate as isolated "silos," where data is not shared in a standardized or real-time manner. This lack of connectivity exacerbates the "reproducibility crisis" in chemistry and slows the exploration of vast combinatorial chemical spaces. To address this, researchers from the University of Glasgow developed a low-cost, networked robotic platform—the "Chemical Processing Unit" (ChemPU)—designed to perform, analyze, and share experimental results in real-time, moving toward a true "Internet of Chemical Things."
System Architecture and Networking
The team constructed six identical robots using affordable, off-the-shelf components. Each unit costs less than $500 and includes a pcDuino3 single-board computer, peristaltic pumps for liquid handling, and a standard webcam for visual sensing. The robots are controlled via Python-based software and communicate their states and results through social media (Twitter) or a dedicated server.
This architecture allows for real-time synchronization and data sharing. For example, a Support Vector Machine (SVM) using a "Bag of Words" approach was trained on a hand-annotated dataset of 1,295 images to recognize crystal formation with 95% accuracy, enabling autonomous 24/7 monitoring.
Collaborative Search Strategies
A key advantage of networking chemical robots is the ability to perform collaborative multitasking. The researchers implemented agent-based modeling to compare three search strategies: random, individual memory, and collaborative. In the collaborative mode, robots share their experimental history across the network, significantly reducing redundant experiments.
In practical tests involving an azo-dye grid (117 reagent combinations), the collaborative approach halved the number of experiments required to identify rare outcomes compared to robots working in isolation. This demonstrates that networking isn't just about increasing throughput, but about enabling "smarter" collective discovery.
Chemical Communication and Reproducibility
Beyond discovery, the network was used for chemical communication. By utilizing the oscillating Belousov–Zhabotinsky (BZ) reaction, the team synchronized two physically separated chemical oscillators with an uncertainty of only 2 seconds. By modulating the oscillation frequency, they successfully transmitted encoded characters across the internet.
The system also provided a tool to quantify the stochastic nature of chemical reactions. By running 120 polyoxometalate (POM) crystallization reactions three times each, the researchers found that some crystals formed only 11.8% of the time under identical automated conditions. This highlights the inherent variability in chemical processes and the value of automated, networked systems in identifying reproducibility issues.
Conclusion and Future Outlook
The development of $500 networked chemical robots democratizes access to automation and provides a framework for global collaboration. While current sensors are limited to webcams, future iterations could integrate benchtop NMR or IR spectrometers for deeper molecular insights. As the network scales, more robust communication protocols will be necessary to ensure security. Ultimately, this work paves the way for "chemical clouds" where researchers can download and execute complex syntheses on standardized hardware worldwide.
Summary updated