We present the use of electrical square signals to excite an array of 3D carbon electrodes for the trapping of small latex particles from a flow in a microfluidic device using dielectrophoresis (DEP). DEP refers to particle motion induced by the interaction of that particle with an electric field gradient. Compared with the sinusoidal signals commonly used to induce DEP, square signals have a bigger RMS value which is also equal to the amplitude of the signal. Hence, one can expect to implement an electric field with stronger magnitude. The use of carbon-electrodes for DEP has enable high throughput, low voltages for particle manipulation, excellent biocompatibility and low cost fabrication [1]. Here we focus on studying the improvement of the DEP force magnitude when using a square signal to polarize an array of 3D carbon electrodes.

The experimental protocol focused on quantifying the number of particles that can be trapped from a flow when using sinusoidal or square signal and varying their frequency and magnitude.  1µm fluorescents particles suspended in Distillated Water were used here. Based on the modeling of their DEP force, frequencies up to 10 kHz can be used for trapping. The particle concentration in the experimental samples was 10⁸particles/ml while the electrical conductivity was 21 µS/cm. After the microchannel was filled up with the particles solution, the electrodes were polarized using a specific signal. The flow was started at the same time. A sample plug of 43.7 µl was processed followed by 60 µl of clean buffer (with no particles).  The amount of particles exiting the polarized carbon electrode array was determined by monitoring the fluorescence level with a microscope. After three minutes of experiment, the electrode array is turned off, releasing all previously trapped particles. If significant trapping was achieved a sharp, high peak on the fluorescence level at the exit of the array was detected. The magnitude of this peak decreases proportional to the level trapping achieved.

Results are shown in Figure 1 for voltages of 15 Vpp and frequencies below 10 kHz. Particle trapping was induced and an improvement on quantity of the particles trapped by the electrode array was achieved when using square instead of sinusoidal signals throughout all frequencies tested. The differences on the quantity of particles trapped differ for each frequency but the use of square provides a clear advantage across the entire range of frequencies tested. At 10 kHz, particles were only trapped when using a square signal, while frequencies above 100 kHz no particles were trapped regardless of the signal used. Similar results were obtained using an amplitude of 5 V_ppfor the same range of frequencies (see Figure 2). As expected, the amount of trapping is less when using signals with low voltage. At 5 kHz, trapping only occurred when square signals were used while no trapping was achieved at signal frequencies higher than 10 kHz.

This improvement in trapping efficiency is likely to be a result of the RMS value. For a square signal, the RMS value equals the amplitude of the signal while in a sinusoidal signal the RMS value is only √2 of the amplitude. Different phenomena were observed at frequencies lower than 1 kHz. Electrothermal flows were first observed, followed by eventual sample electrolysis, and the generation of bubbles around the carbon electrodes, at frequencies around 10Hz. In the case of electrothermal flows, a vortex-like movement is observed around the posts. The speed of the movement was inversely proportional to the signal frequency. As the frequency approximates 10 Hz, the particles can be seen to move back and forth between electrodes following the signal frequency. Frequency values below 10 Hz quickly led to bubbles. Although modeling of the DEP force for these latex particles predicted uniform trapping at frequencies below 10 kHz, it is clearly seen from our experimental results that the trapping efficiency decreases according to frequency. We believe the phenomena described for low frequencies are to blame for a decrease in trapping efficiency.

Here, we demonstrated the advantage of using square over traditional sinusoidal signals when trapping targeted particles. A square signal is easier to generate using portable electronics than a sinusoidal signal, hence facilitating future implementation of a hand-held lab-on-a-chip for particle trapping and sorting using carbon-electrode dielectrophoresis. Ongoing work is on implementing extraction of targeted cells from a heterogeneous population using square signals. Future work is on understanding why exactly a square signal leads to improved trapping and exploring other signal geometries.