Development and optimization of a microfluidic module for biological reagent addition in microdroplets

Abstract

Microfluidics enables the miniaturisation of laboratory operations, where microlitre to picolitre volumes can be manipulated with laminar flow, high automation, and high throughput. Droplet microfluidics is a subfield of microfluidics in which picolitre-sized droplets serve as independent micro-reactors. One of the main challenges is to dynamically and precisely inject reagents into these droplets. While active and passive droplet merging strategies offer limited control, pico-injection has been reported as a promising method for precise, rapid, and tunable reagent injection into droplets. However, its practical implementation remains difficult, as the process is highly sensitive to pressure fluctuations, leading either to the disruption of injection or to unwanted dripping. Furthermore, no reproducible and widely accepted protocol has been established in the literature. This work, conducted in collaboration with LiveDrop, focused on the evaluation of two pico-injection chip designs. The study had two main objectives: first, to characterise a pico-injection chip and establish a practical method linking injection pressure to injected volume; and second, to determine whether the tested design met the criteria of a “good pico-injection,” defined as high-throughput, significant injected volume, and minimal dripping. To this end, a single-laser PMT probe was used to detect the fluorescent signal emitted by droplets and by the injection liquid. Injected volumes in droplets were quantified with the PMT data, and the percentages of injected droplets, non-injected droplets, and dripping events were also assessed. Complementary imaging measurements were also performed to validate the PMT results. Two chip designs were tested. The first chip failed to deliver efficient pico-injection. While some droplets reached injected volumes greater than 10 pL, this was systematically associated with high dripping and low droplet frequencies (15-30 Hz). Increasing inlet pressures did not improve performance, and the system could not be considered efficient or high-throughput. Consequently, this design was abandoned. In contrast, the second design showed promising performance. It achieved injected volumes of 15-20 pL with moderate dripping (generally <20%) at droplet frequencies around 620 Hz. These results suggest that the second chip fulfils the criteria of a good pico-injection in terms of effectiveness, efficiency, and throughput. However, differences were observed between PMT and imaging data, with the latter reporting smaller injected volumes at comparable injection pressures. Possible explanations include methodological limitations linked to the simplified protocol, the limited number of replicates and imaging data, and potential reproducibility issues of the pico-injection chip. Given these uncertainties, the study does not allow a definitive conclusion about the superiority of the second design over the first, and individually it remains difficult to assess whether either chip can be considered a successful pico-injection chip under the tested conditions. Experimental reproducibility needs to be addressed by adjusting the PMT setup, increasing the imaging dataset, and repeating experiments under consistent conditions. While purely speculative, inter- drop distance and flow regimes might play a role in increasing injection efficiency (dripping).