Detailed evaluation of the performance of microfluidic T mixers using fluorescence and ultraviolet resonance Raman spectroscopy

A reliable device that produces efficient mixing with a short dead time has enormous utility in the kinetic analysis of biochemical and chemical processes. We have designed two different T mixers that use moderate flow rates $(0.2–0.4\mathrm{ml}∕\mathrm{s})$ , can monitor reactions up to several milliseconds, and achieve mixing times as low as $20\mu \mathrm{s}$ . The two mixers are easy to build and dismantle, reliable, and can perform hundreds of experiments without blocking. The first mixer comprises a stainless steel block, containing a microchannel, glued to a quartz cuvette, containing a $200\times 200\mu {\mathrm{m}}^2$ observation channel defining a conventional T mixer. The reactions are monitored by imaging the length of the observation channel onto a charge-coupled device camera. In the second mixer the entire T ($200\times 200\mu {\mathrm{m}}^2$ internal cross section) is contained within a $40\text{-}\mathrm{mm}$ -long quartz cuvette. We have adopted a novel approach to controlling the entrance channel bore by inserting a stainless steel wire in order to increase the linear speed of the impinging fluids. Using a dye to visualize the flow profile inside the second T mixer, it was shown that in this T geometry segregation of the reactants is observed in the junction between the inlet channels and the observation channel (T junction) and mixing occurs entirely in the observation channel. We thoroughly tested the two mixers through several kinetic reactions using both fluorescence and ultraviolet resonance Raman spectroscopy measurements. We show that both mixers provide efficient mixing with nominal dead times (using 1:10 $v∕v$ dilution), calculated using the quenching of the fluorescence of $N$ -acetyl-$L$ -tryptophanamide by $N$ -bromosuccinimide, of $200\pm 20$ and $100\pm 10\mu \mathrm{s}$ , for each mixer, respectively. However, the ability to monitor within the inlet channels and the entire observation channel of the second mixer shows that this standard approach to estimating the dead time is artifactual, since it relies on assuming a constant flow speed throughout the observation channel, a feature that we show is not adhered to at short distances from the T junction. Using both mixers the refolding of the $A$ state of cytochrome $c$ to the native state was followed by fluorescence and ultraviolet resonance Raman spectroscopy, revealing the ability of these instruments to provide insights into the early stages of protein folding using only milligrams of sample.