Search for: All records

We describe an exciting new application domain for deep reinforcement learning (RL): droplet routing on digital microfluidic biochips (DMFBs). A DMFB consists of a two-dimensional electrode array, and it manipulates droplets of liquid to automatically execute biochemical protocols for clinical chemistry. However, a major problem with DMFBs is that electrodes can degrade over time. The transportation of droplet transportation over these degraded electrodes can fail, thereby adversely impacting the integrity of the bioassay outcome. We demonstrated that the formulation of droplet transportation as an RL problem enables the training of deep neural network policies that can adapt to the underlying health conditions of electrodes and ensure reliable fluidic operations. We describe an RL-based droplet routing solution that can be used for various sizes of DMFBs. We highlight the reliable execution of an epigenetic bioassay with the RL droplet router on a fabricated DMFB. We show that the use of the RL approach on a simple micro-computer (Raspberry Pi 4) leads to acceptable performance for time-critical bioassays. We present a simulation environment based on the OpenAI Gym Interface for RL-guided droplet routing problems on DMFBs. We present results on our study of electrode degradation using fabricated DMFBs. The study supports the degradation model used in the simulator. 

A prototype aerosol detection system is presented that is designed to accurately and quickly measure the concentration of selected inorganic ions in the atmosphere. The aerosol detection system combines digital microfluidics technology, aerosol impaction and chemical detection integrated on the same chip. Target compounds are the major inorganic aerosol constituents: sulfate, nitrate and ammonium. The digital microfluidic system consists of top and bottom plates that sandwich a fluid layer. Nozzles for an inertial impactor are built into the top plate according to known, scaling principles. The deposited air particles are densely concentrated in well-defined deposits on the bottom plate containing droplet actuation electrodes of the chip in fixed areas. The aerosol collection efficiency for particles larger than 100 nm in diameter was higher than 95%. After a collection phase, deposits are dissolved into a scanning droplet. Due to a sub-microliter droplet size, the obtained extract is highly concentrated. Droplets then pass through an air/oil interface on chip for colorimetric analysis by spectrophotometry using optical fibers placed between the two plates of the chip. To create a standard curve for each analyte, six different concentrations of liquid standards were chosen for each assay and dispensed from on-chip reservoirs. The droplet mixing was completed in a few seconds and the final droplet was transported to the detection position as soon as the mixing was finished. Limits of detection (LOD) in the final droplet were determined to be 11 ppm for sulfate and 0.26 ppm for ammonium. For nitrate, it was impossible to get stable measurements. The LOD of the on-chip measurements for sulfate was close to that obtained by an off-chip method using a Tecan spectrometer. LOD of the on-chip method for ammonium was about five times larger than what was obtained with the off-chip method. For the current impactor collection air flow (1 L/min) and 1 h collection time, the converted LODs in air were: 0.275 μg/m3 for sulfate, 6.5 ng/m3 for ammonium, sufficient for most ambient air monitoring applications.