Machine learning (ML) drives efficiency in automated chemistry

    AI in flow chemistry

    The growing role and potential of machine learning in chemistry, and particularly within flow chemistry, is being shown by Vapourtec and Accelerated Materials at the upcoming ACHEMA process industry exhibition (Frankfurt, June 10-14).

    Machine learning delivers a number of benefits for chemical optimisation including enhancing the efficiency and accuracy of predictive modelling and improving decision-making processes. It brings the capability of the rapid analysis of vast datasets and identifying patterns and correlations into the experimental design, significantly improving the efficiency of exploration of potential experimental conditions.

    Accelerated Materials’ Chief Strategy Officer, Professor Alexei Lapkin, commented: “Bringing AI into chemistry accelerates the discovery and optimisation of new compounds, reduces experimental costs and shortens the timelines for development.
    “Traditional optimisation methods are based on variations of one-variable-at-a-time (OVAT) methods. The OVAT method frequently fails to capture any interaction effects between variables and is data inefficient.
    “Process chemists with advanced optimisation skills typically use Design of Experiments (DoE) methods which address many of the OVAT limitations. These include better efficiency in data acquisition, improved exploration of optimisation space and enhanced system understanding through modelling effects between variables.
    “DoE however also has its limitations when modelling complex non-linear, dynamic systems.
    “ML solutions address the limitations of DoE but are challenging to implement.”

    There have been a number of excellent ML solutions published utilising flow reaction applications.  A great example is the work of the Sustainable Reaction Engineering group at the University of Cambridge.  This group’s work has now moved on through the spinout company Accelerated Materials (AM).  AM is showing its software product AMLearn™ at the ACHEMA exhibition. The software integrates with the Vapourtec R-Series flow chemistry system and software to provide an off-the-shelf ML optimisation solution.

    Vapourtec’s founder Duncan Guthrie commented: “Vapourtec’s R-Series is the only commercially available flow chemistry system that provides the opportunity to integrate machine learning algorithms using high-level commands to specify reaction methods rather than by communication with individual system components.
    “The reaction method approach utilises Vapourtec’s industry-standard OPC UA interface and allows simple configuration.
    “This offers a selection of reagents or catalysts pre-loaded into an autosampler and can be used with any of Vapourtec’s 14 different types of reactor including our photochemical and electrochemical reactors.
    “It also features data retrieval from a range of pre-configured PAT (process analytical technology) devices including Raman, FTIR and UV detectors.
    “We are very excited about the impact machine learning is having on the optimisation of flow chemistry experimentation as it offers so many advantages over existing approaches and methods.”

    Vapourtec will be based at booth E64 in hall 9.0 at the ACHEMA Flow Chemistry Pavilion in Frankfurt. Accelerated Materials (AM) can be found on the ‘startups stand’ in Hall 6. You can also visit booth D36 in Hall 6 where Innovation Centre in Digital Molecular Technologies (iDMT@Cambridge) will have further information on AI, robotics, high-throughput, and other technologies for chemistry.

    To find out more about Vapourtec R-Series software with OPC UA, click here

    For ML resources, see Innovation Centre in Digital Molecular Technologies (iDMT@Cambridge)

    To find out more about Accelerated Materials (AM), click here

    Suggested reference publications:

    [1] A Machine Learning-Enabled Autonomous Flow Chemistry Platform for Process Optimization of Multiple Reaction Metrics
    M.I. Jeraal, S. Sung, A.A. Lapkin, Chemistry-Methods 1, 71 – 77 (2021)

    [2] Accelerated chemical reaction optimization using multi-task learning
    C.J. Taylor, KC. Felton, D. Wigh, M. I. Jeraal, R. Grainger, G. Chessari, C. N. Johnson, A. A. Lapkin, ACS Cent. Sci. 9, 5, 957–968 (2023)

    Get in touch

    For more information on flow chemistry systems and services please use the contact methods below.

    Call us on +44 (0)1284 728659 or Email us