« WIRELESS PARYLENE-BASED RETINAL IMPLANT Thesis by Jay Han-Chieh Chang In Partial Fulfillment of the Requirements for the Degree of Doctor of ...»
(a) Particles left on parylene-C substrate during liftoff photoresist spinning will cause unwanted short circuits between the two metal lines. (b) Cotton rods were often used during classical metal liftoff process to help remove the metal residues which will easily damage the metal surfaces to affect the electrical properties
(a) Sample layout for soaking test. (b) Real fabricated device with 300μm trench........64 Figure 2.23.
(a) Trench of the initial soaking device (0 days). (b) Trench of the soaking device with hard-bake after 7 days. (c) Trench of the soaking device with micro-90 detergent treatment after 7 days. (d) Close-up view of (c); undercut can be easily observed
Fabrication process of the direct metallization on parylene-C film by sacrificial SU-8 masks. (a) Deposit parylene-C film on silicon wafer, (b) spin coat SU-8 with desired thickness on parylene-C film, followed by pre-soft bake, (c) expose by UV light and post bake, (d) develop the uncross-linked SU-8, (e) metal (Ti/Au)deposition by E-beam, and (f) peel off SU-8 mask................66 Figure 2.25. (a) 15μm thick SU-8 is highly flexible. (b) After metal deposition, sacrificial SU-8 mask can be easily peeled off by tweezers in seconds, and no visible residues were observed...........67 Figure 2.26. (a) Circles from 10μm to 300μm in diameter can be fabricated on parylene-C film by sacrificial SU-8 masks. (b) Squares from 10μm to 300μm in side length can also be fabricated on parylene-C film
(a) Image of 1cm long and 40μm wide metal line along with two metal pads. (b) Microscopic image of the pattern fabricated by liftoff process. (c) Microscopic image of the pattern fabricated by sacrificial SU-8 masks
Comparison of the resistance of the metal lines patterned by different methods............70 xiv Figure 3.1.
Former connection technology that requires an additional PDMS holder to house the IC chips; the adhesion between parylene-C and chips only relies on conductive epoxy, which occupies less than 2% of the total contacting area
Schematic representation of a clamp as the bonding tool on the testing samples.............79 Figure 3.3. Cross-sectional SEM image of the adhesive bonding between parylene sheet and silicon (2MPa, 130ºC). The shape of the microstructure keeps the same after the bonding process and the bonding pads are well defined
(a) Setup of the force gauge to measure the peeling force. (b) Real testing sample after bonding. (c) Schematic representation of the testing sample with top view and cross-sectional view
Peeling force v.s. bonding temperature for various photo-patternable adhesives.............82 Figure 3.6.
Peeling force v.s. bonding pressure for various photo-patternable adhesives
Adhesive interface before (a) and after (b) peeling test
(a) Maximum peeling forces of different photo-patternable adhesives. (b) Peeling force v.s. bonding time for different photo-patternable adhesives
Custom holder for chip assembly technique; all lithography was done on this holder which also served as the safety buffer zone for squeegee
(a) Dummy chip for assembly yield test; pads also served as alignment marks. (b)&(c) Metal pads were exposed; resolution of around 5μm could be achieved
(a) Unbaked AZ4620. (b) AZ4620 baked at 140°C for 30 minutes in vacuum oven;
slope formed by reflow will be beneficial for conductive epoxy to be fed through. (c)&(d) Side lengths show no change before and after baking
Gluing area was increased from 2% to 94% (2%+92%) by the extra photoresist used as glue. Note that unnecessary pads were covered to avoid shortage
(a) Dummy chip and discrete components integrated with surgical parylene flex. (b) Backside of the dummy chip and discrete components. (c) Close-up view of the high-density multichannel chip integration. (d) Retinal tack used to fix stimulating electrodes on retina
Setup of the measurement; the electrode array outputs (the electrode end that will be placed on macula) were probed to check the connection
Connection yields under 4 different conditions; reliability tests were carried out after squeegee connection, encapsulation by parylene-C coating, and accelerated soaking in 90°C saline.
The results show that our new technique combined with thick parylene-C coating do provides a high connection yield
xv Figure 3.16. (a) Dummy chips with 40μm by 40μm pad size and 40μm separation. (b) Connection between parylene substrate and dummy chip. (c) Yield v.s. separation of pads. (d) Yield v.s. side length of pads. The results show that high connection yield (90%) can be achieved for pads as small as 40μm by 40μm and with a 40μm separation in between
(a) Surgical parylene-C device connected with silicon chip and discrete components.
(b)(c) Only desired metal pads are exposed; other area is covered by photo-patternable adhesives.
The alignment accuracy can be improved to be around 5 µm
Gluing area is increased from 2% to 94 (2+92)%. Unnecessary pads are also covered to avoid electrical shortage during conductive epoxy squeegee process
Layout and wiring diagram of the real 268-channel retinal IC chip
Only desired metal pads were exposed; other area and pads for testing only were covered by photo-patternable adhesives
(a) Real 268-channel retinal IC chip integration by photo-patternable adhesives and conductive epoxy squeegee techniques. (b) Successful functional signal testing including oscillator assembly, clock level shifter, and low voltage rectifier
Layout and wiring diagram of the real 1024-channel retinal IC chip
Parylene flex integrated with 1024-channel retinal IC chip, and the connection of critical pads can be seen very clearly
The tested power coil from USC (center-tapped receiver coil) has similar inductance ratio and better Q factor compared to the litz-wire coil used in UCLA's demo system. It can successfully induce correct voltages for the 1024 channel retinal IC chip and was used as the power coil for the whole system
With correct induced voltage from the receiver power coil, all correct DC voltages including VrddL, VrddH, VrssL, and VrssH were successfully tested via wires through parylene flex with high yield
With correct induced voltage from the receiver power coil, the crystal oscillator output with 16MHz signal was successfully tested via wires through parylene flex with high yield...........102 Figure 4.1.
(Left) Parylene-based device protected by traditional silicone-parylene combination.
(Right)Parylene-based device protected by the proposed parylene-metal-parylene protection scheme
Schematic representation of the retinal implant connected with IC chips, coils, and discrete components
(a) Normal vision; (b) Blurry central vision for AMD patients; (c) Tunnel vision for RP patients
xvi Figure 4.4.
Commercial amplifier chips, discrete components, and resistor chips were first integrated and packaged before soaking test
Fabrication process of the parylene-C pocket structure. The sacrificial area is designed to be 2 mm×2 mm to accommodate the bare die (1.6mm×1.1mm×200 µm) and it leaves some space for chip insertion and movement. The pocket structure can provide support to the chip to make alignment
Schematics of the process flow to show how to open pocket and insert chip.................113 Figure 4.7.
Parylene pocket is opened by spatula after releasing the sacrificial photoresist. With wafer dicing tape as a fixation substrate, the pocket becomes easier to open
(a) A chip is inserted and aligned; conductive epoxy is applied to make connection. (b) Alignment accuracy of 10µm can be achieved. (c) The size of conductive epoxy drop is 200µm in diameter. (d) Signal is monitored by oscilloscope
(a) Top side of the series RLC circuit built by discrete components. (b) Resonant frequency is measured by impedance analyzer
Measurement setup of active soaking test for dummy conduction chip. Power supply, multi-meter, and dummy conduction chip soaked in saline are arranged in series. Door is opened to show the setup inside
Line resistance v.s. time of samples coated by 40µm parylene-C only soaked in high temperature saline solution. The subset shows the difference of chip size of dummy conduction chips
(a) After sandwich layer protection, the device is still highly flexible. (b) Device becomes inflexible after coating with thick silicone; thickness needs to be more than 5mm............118 Figure 4.13. Schematics of proposed packaging scheme and other protection for comparison........119 Figure 4.14. High-density multi-channel chip integration protected by different packaging schemes.
(Top left) No protection; (Top right) 40 µm parylene-C only; (Bottom left) 5mm silicone + 40 µm parylene-C; (Bottom right) 20 µm parylene-C + 0.5 µm metal + 20 µm parylene-C
Observed failure modes, including bubbles and corrosion, on parylene devices (a), corrosion of sandwiched metals (b), delamination of metal traces on dummy chips (c), and corrosion of conductive epoxy (d), respectively, are found after soaking
Samples without protection, water vapor can penetrate the thin parylene layer of device itself to damage the metals very quickly, especially under a bias field
Samples with thick parylene and other protection, uniform bubbles around the whole device are first observed and water vapor gradually diffuses through the protection to attack the device
Concept of high-density multi-channel chip integration and testing setup. (Top left) High-density multi-channel dummy chip; (Top right) Corresponding parylene flex; (Bottom left) Chip integration by squeegee process; (Bottom right) Testing setup
xvii Figure 4.19. New packaging scheme with a glass on top for further protection. Adhesive was used to cover the whole device. Adhesives to three different interfaces including glass, parylene, and silicon need to be well selected
Investigation of adhesives to different interfaces by undercut observation before and after interface treatments. N: no treatment; SAP: silicone adhesion promotor
(Left) Schematic of the new packaging scheme. (Right) High-density multi-channel chip integration protected by new packaging scheme
Failure mode on new packaging scheme: Saline can't diffuse through the glass directly to attack the electrodes and corrode the conductive epoxy. It can only slowly go through the interface
The 3-coil scheme for inductive power transfer
A model of the coil system is built using HFSS for the coil interference analysis.......131 Figure 4.25. S-parameters from coil 4 to coil 1 and from coil 3 to coil 1 with different inner diameters of coil 1
Figure 4.26. (a) A Notch filter. (b) Two Notch filters are connected to the system
A parylene flex mechanical model for 512-channel retinal IC chip was designed and fabricated
Demonstration of a parylene flex mechanical model before and after wrapping with silicon chips, discrete components, and coils
Successful in vivo implantation process in a pig's eye
Schematic design of the surgical mockup for the 1024-channel chip
Parylene surgical mockup integrated with IC chip and discrete components before and after folding
Successful in vivo implantation process in a dog's eye
Schematic diagram to show that the electrode array was positioned on the retina at around 5mm~8mm away from optic nerves
(Left) Parylene surgical mockup with metal connections integrated with IC chip, discrete components, and coils before implant. (Right) In vivo implantation trial in a dog's eye.....139 Figure 4.35. In vivo implantation trial in a dog's eye. (1) The electrode was first folded to fit the size of the cut on eyeball. (2) The electrode was then inserted into the cut. (3) Due to the rigidity of the electrode, the insertion process was blocked. (4) The insertion process also bent and damaged the device itself