CHAPTER 6 FABRICATION
6.4 DEVICE ITERATIONS AND CONSIDERATIONS
The device has gone through 9 iterations. Each iteration shrunk the footprint and increased reliability or made the device more amenable to implant. The diffusor and the reservoir are built separately and combined after the reservoir was coated with Parylene. Differences between each iteration are covered in Figure 6.13.
Figure 6.13: Different versions of the oxygen generating device. Changes across different versions of the device. Note from the scale part, the device has shrunken to 25% of its size: from a 14mm diameter by 3mm tall, planar device (V1) to a 7mm wide by 8mm long by 2mm tall, curved device (V8).
Version 0 of the Oxygenerator was a prototype using a copper 44 Litz wire and a Parylene PCB. The reservoir and diffusor were filled with electrolyte. The electrodes activated in the diffusor. This oxygenerator had issues transporting electrolyte from the reservoir to the diffusor, and suffered adhesion issues where the Parylene PCB entered the cannula. Subsequent designs, versions 1-7, had the electrolyte separated from the diffusor. The air conduit referred in section 4.4.1 shuttled oxygen from the liquid reservoir to the diffusor. In versions 1-5 this air conduit contained a set of pillars to maintain the shape of the chamber (Figure 6.14) and prevent collapse—under pressure from electrolysis—of the membrane that separates the conduit from the liquid reservoir.
The coil changed along with the oxygenerator design and different fabrication methods. Versions 1 and 2 used a planar coil, whose fabrication is discussed in section 5.3.2.1. Version 3 and 4 used a hand spun coil and a metal spindle. This flat coil remained above the device but opened a window for injecting electrolyte from above. Versions 5-8 used an automatically wound coil as described in section 5.3.2.3. In versions 5 and 6, the coil surrounds the electronics pocket and the reservoir. These versions of the oxygenator are covered in a cast shell of silicone to give the device an appropriate shape.
Version 6 replaced the air conduit with nanoporous, hydrophobized Vycor glass. However, the nanoporous tack pushed the location of the cannula 2 mm back, which was problematic for implant, since the cannula needed to enter through the 3mm to 4mm wide pars plana. Version 7 reverted to an air conduit and moved the cannula to the very front.
Version 8 is a departure from the approach in earlier versions, as the reservoir region contains only the electronics. The diffusor contains the electrodes and performs electrolysis. This change was brought about by changing the method of replenishment from injections to osmosis. In doing so the liquid reservoir could shrink.
The fabrication steps for versions 1 through 7 are fairly similar. The description below is for version 1 of the device, but applicable for every version. Only one fabrication procedure is presented to avoid repetition.
Figure 6.14: Fabrication of version 1 of the oxygen generating device. The silicone for the reservoir was cast in dry-film photoresist molds, while the (D) diffusor was made by dip-coating a Parylene-coated wire in silicone. Subsequent models contained a cast diffusor made using dry-film molds. (B) Note in that there is a small conduit in the lower right side of the diagram where the cannula sits. Such a feature exists in versions 1 through 7 as the cannula is much larger than the 120µm height of the (A) air chamber. The final product can be found in Figure 6.13-V1.
Figure 6.15: Computer rendering of parts of version 5 of the device. Notice the commonality in the design process with version 1. Here placing the coil and electronics around the reservoir helps reduce its footprint to 10mm in diameter.
The reservoir was cast in silicone using MEMS fabricated molds. It was made in 3 sections, as seen in Figure 6.14a-c; each with a separate mold made by laminating negative dry film photoresist (DuPontTM WBR2120). To fabricate a mold, first, the dry film is laminated onto a silicon wafer at 95ᵒC. With all necessary layers for a given pattern applied, the dry film is baked for 20 minutes at 65ᵒC, and followed by UV patterning (~250-600mJ/cm2, depending on number of layers). The film is then post exposure baked to cure the pattern at 95ᵒC for 1 minute. Further layers can be applied, and the process repeated. When all patterns have been defined, the wafer is developed in an AZ340 develop-water solution (1:4) for approximately 40 minutes. The resultant mold is coated in Parylene- C as a release agent. The electrolyte chamber has a 5-layer lamination (600µm tall) for the walls of the chamber, and a 2-layer lamination (240µm tall) for the membrane separation from the air chamber. The air chamber, Figure 6.14a, has pillars to prevent its collapse when pressure builds in the adjacent electrolyte chamber. The final mold defines the top of the electrolyte chamber. Each of these molds is filled with a mixed two-part medical-grade silicone (MED4-4210 from NuSil Technologies, LLC) and degassed. After degassing, the excess silicone is wiped from the mold using a fresh razor blade, and placed in the oven at 100ᵒC for 3 minutes to partially cure. The three sections are released from their molds and bonded together using uncured silicone as a glue under a microscope. The assembled sections are cured in an oven at 100ᵒC for 3 minutes. Two platinum
wires (0.004in from California Fine Wire) are then inserted through the sidewall of the silicone into the electrolyte chamber to act as the electrodes. The 304 stainless steel cannula is glued to the opening in the air chamber using uncured silicone. The other side of the cannula is temporarily plugged with silicone, and then the oxygenerator is placed in an oven at 100ᵒC for 4 hours to fully cure. The oxygenerator is CVD (Chemical Vapor Deposition) coated with Parylene-C to reduce oxygen permeation through the reservoir into the conjunctiva.