Finite-Element Analysis Simulations of Separation Electrode Array

Since the COMSOL particle tracing software can only calculate DEP forces exerted on particles, we made custom modifications to it so that the effects of TWDEP forces could be included. The electrode array is symmetric about the axis that runs the length of the electrodes thus it is only necessary to simulate the fields in two dimensions. Figure 5.10 shows the 2D finite-element model we created for the separation electrode array. The electrodes in this model are 30μm wide with equally sized gap spaces. The surface plot shows the electric-field magnitude profile that is generated in the fluid containment region when voltage magnitudes are 2Vpp @ 1MHz and the

phase of the signals on each electrode is increased by 90⁰ with respect to its neighbor (going from left to right), resulting in a positive phase gradient of ( ). The buffer in the containment area is assumed to be the 5mS/m reference medium and the containment dimensions are determined by the microfluidic designs described later in section 5.2.

Figure 5.10 Simulation results of electric field magnitude profile generated in fluid containment region when

traveling-wave voltages are applied. The electrode gaps and spacing are 30μm and the voltages are 2Vpp, 1MHz signals. The phase of each electrode is shifted 90⁰ with respect to its neighbor (going left to right), resulting in a positive phase gradient.

Figure 5.11 shows the results of the transient particle motion simulation. The top of figure 5.11 shows the initial random distribution of particles in a zoomed in region of the array. The bottom of figure 5.11 shows that after 60 seconds, all particles almost reach their steady state levitation depth of 60μm, the point at which the vertical DEP forces balance out with gravity, and travel in the negative x-direction. The particles travel opposite the phase gradient (right to left) because Im{KCM}is negative for these particles at 1 MHz. The average TWDEP x- velocity for the particles is approximately 0.5 μm/s which is

Figure 5.11 Top shows 2D particle motion simulation model and initial random distribution of particles. Bottom

shows particles reaching a steady-state levitation height of 60μm due to DEP forces balancing out with gravity and traveling in the negative x-direction due to the TWDEP field created by the electrodes.

slightly below our desired threshold of 1μm/s. However the difference in velocity is well within the range of being able to be compensated for by slightly increasing the voltage or switching to the 15μm electrode array.

As a result of these simulations we were able to determine the electrode geometries and dimensions that give us a reasonable ability to test our methodology. We made the decision to fabricate multiple devices, covering the size ranges we determined to be operable via simulation. Sets of 5mm x 5mm quadrupole electrodes with gaps of 15μm and 30μm and filet distances ranging from of 25μm to 150μm were fabricated as well as linear electrode arrays with gaps and widths of 15μm and 30μm. In the next section we provide the fabrication details of our electrode designs.

5.1.3 Electrode Fabrication and Packaging

The electrodes were fabricated in batch, using a metal on glass photolithography process (TRICR Corp.). The minimum electrode dimension fabricated was 5μm, allowing the use of a simple metal lift-off process. Two sample wafers that have been fully processed are shown in figure 5.12.

Figure 5.12 Gold and aluminum processed wafers with multiple copies of both electrode structure types

Four-inch soda lime glass wafers were selected for substrates, as glass has a number of desirable qualities in this application. Since there are no active circuit components at this level, expensive silicon substrates are not needed. In addition, the transparency of glass allows for illumination of the electrode region from many different angles, which is critical for recording data. During the metal lift off fabrication process, a sacrificial layer of photoresist is deposited onto the wafers and photolithographically patterned according to our electrode designs, using a chrome photo-mask. Vapor deposition is then used to pattern a 200 Å adhesion layer of titanium, followed by a 200nm metal electrode layer. Wafers with electrode layers made of both gold and aluminum were fabricated. The tradeoff between the two metals is cost versus durability. After experimenting, it was found that gold electrodes were better suited for this

platform in the long run, as one gold chip can be reused for weeks if properly cleaned, whereas the aluminum devices oxidized after a few uses.

However the most critical fabrication parameter in this process is the thickness of the electrode layer. If this layer is made to be too thin, the manufactured device becomes unreliable. However if it is too thick, the electrodes become difficult to use for dielectrophoresis. When the electrodes heights are on the same order of magnitude as the size of the particles, then dielectrophoresis ‘dead zones’ will be created between electrodes, as there is little to no electric field gradient in that region, if particles fall into those valleys they remain trapped, as no dielectrophoretic forces can be exerted on the particles.

Each processed wafer contains 8 separation electrode arrays and 8 sets of characterization electrodes of varying sizes. A wet-saw is then used to dice the wafer into individual chips. Figure 5.13 shows a close up view of one of the gold quadrupole electrode chips that had a gap spacing of 15μm and filet radius of 75μm. Figure 5.14 shows a close-up view of one of the gold 30μm separation electrode arrays after processing.

Figure 5.13 Close-up view of one of the gold quadrupole characterization chips after processing