(420c) An Autonomous Microreactor Platform for the Optimisation of Kinetic Parameters for Fast Liquid/Liquid Reactions
AIChE Annual Meeting
2023
2023 AIChE Annual Meeting
Computational Molecular Science and Engineering Forum
Automated Molecular and Materials Discovery: Integrating Machine Learning, Simulation, and Experiment II
Tuesday, November 7, 2023 - 4:00pm to 4:15pm
Batch processing is commonly used to obtain kinetic models. However, traditional stirred-tank reactors can be adversely affected by inefficient heat and mass transfer, safety concerns in case of exothermic reactions and long processing times overall because each batch needs to be processed separately in order to separate and analyse the products. Benefiting from their smaller reactor volumes, continuous flow microreactors can alleviate these challenges. The higher surface area-to-volume ratios allow for more efficient mass and heat transfer, safer usage of hazardous chemicals and precise control over reaction conditions.[2]
However, the effectiveness of model discrimination (i.e. the identification of the appropriate set of equations defining the model) between candidate models and the precise estimation of kinetic parameters depends on the conduction of properly designed experiments. Otherwise, less informative experiments may result in a waste of resources and time. Model-based design of experiments (MBDoE) is a tool for efficient development of deterministic models, which relies on prior knowledge of the model structure of the underlying system and initial parameter estimates to design an optimal experiment.[3] The further integration of MBDoE algorithms with automated flow platforms and inline analytical techniques is a powerful tool for obtaining intrinsic reaction kinetics in a very efficient way.[4] In such close-loop experimental systems, the MBDoE algorithm designs an experiment, which is executed automatically, and data collected in real time from the analytical instrument are interpreted by the algorithm to design the next experiment. This takes place iteratively until the optimization criteria are satisfied. This process aims to improve the way we obtain robust kinetic models by reducing experimentation time, overall cost and enhancing process sustainability.
This work shows the workflow for obtaining reliable and accurate kinetic models for fast liquid/liquid reactions by developing a close-loop experimental system. A reactive chemical system of interest is the heterogeneous nitration of 2-nitrotoluene (2-MNT), since it is a very fast reaction,[5] highly exothermic,[6] and reaction products, 2,4-dinitrotoluene (2,4-DNT) and 2,6-dinitrotoluene (2,6-DNT), are important intermediates in fine chemical industries.[7] In this work, we utilize an autonomous flow platform equipped with continuous flow micromixers to achieve efficient mixing and ensure process safety. Work-up steps are performed in-line and phases are separated based on membrane separation. Organic phase passes through a benchtop NMR for quantification of reaction products and data are processed by an MBDoE algorithm to design the next experiment with the objective of achieving a precise estimation of kinetic parameters for a pre-selected kinetic expression.[8]
References: [1] React. Chem. Eng., 2019, 4, 235, [2] Iscience, 2022, 25, 103892, [3] Chem. Eng. Sc., 2008, 63, 4846 â 4872, [4] React. Chem. Eng., 2019,4, 1623-1636, [5] React. Chem. Eng., 2022, 7, 111-122, [6] Beilstein J. Org. Chem., 2014, 10, 405-24, [7] Phys. Chem. Chem. Phys., 2021, 23, 4658â4668, [8] Chem. Eng. J., 2019, 377, 120346