Chemical Engineering Seminar

The high specific surface areas and small dimensions in microchannel microreactors can enable essentially ideal isothermal, plug flow performance in small scale continuous systems. Thus, these features enable characterization of reaction systems under carefully controlled conditions, with the potential to extract accurate reaction kinetic parameters using a simple mathematical model. To confirm the plug flow assumption for a particular commercially available microreactor, we conducted experiments to characterize the residence time distribution and characteristic mixing time. Experiments in this microreactor gave a characteristic mixing time as low as 6 seconds for a model hydrolysis reaction at 2.4 mL/min total flow, while reaction-diffusion calculations for this device predict minimum mixing times near 1 second at flow rates of 10 mL/min or higher. In contrast, the literature reports mixing times as low as 0.01 seconds for other microchannel systems and 0.002 seconds for a Mettler RC1 calorimeter. We hypothesize that one or more of the following three factors caused the relatively slow mixing in this reactor. (1) The initial mixing of the interdigital feeds takes place while the lamellae are being thinned from 4500 µm down to 33 µm in a 0.095 mL volume. (2) Each of the mixing and reaction channels contains only 6 lamellae, so the characteristic mixing length approaches that of the channel height for reactions operated near the stoichiometric ratio. (3) For reactions that are extremely sensitive to stoichiometry, uneven distribution of the two feeds to the multiple channels can lead to anomalously high apparent mixing times. Building on the foundation of the microreactor characterization case study, we developed a system to characterize homogeneous liquid phase reactions using serial screening in a microreactor. The system used a single microchannel reactor with reactants injected as finite pulses. Fundamental reaction engineering principles (including equations and constraints for heat transfer, mixing, axial dispersion, and pressure drop) were used to design or select the system components to enable analysis of the reactor effluent at undiluted, steady-state concentrations while injecting only microliters of the reactant solutions per experiment. The custom single channel reactor was shown to give the expected low axial dispersion of injected reactants, consistent with the analysis of axial dispersion of Newtonian fluids in circular tubes. However, experiments with different materials of construction demonstrated a variable lag time between the injection of a reactant pulse and its arrival at the analytical sample point as a function of the material of construction. We hypothesize that this effect is the result of adsorption or absorption of the reactants on or in the tubing walls. The impact of this observation is that only some materials of construction may be used for high throughput serial screening by injection of reactant pulses, dependent upon the chemistry employed.