Microgaskets for High-Channel-Density Reconnectable Implantable Packaging

Abstract

Demands for implantable bioelectronic devices to increase the number of channels for greater functional capacity and resolution, shrink implant size to minimize tissue response and patient burden, and support battery changes and electronics upgrades for long-term operational viability, cannot be met with existing implant-connector technology. In this paper we describe our novel approach to develop a rematable high-channel-density implant-connector technology, with a focus on the design, fabrication, and characterization of its microgasket. The microgaskets made of polydimethylsiloxane elastomer (PDMSe) have achieved much better electrical isolation for neural stimulation (~5 MΩ at 10 kHz) compared with conventional implant connectors (50 kΩ at 10 kHz), despite a 200-fold increase in channel density (conventional: ~0.0644 ch/mm², microgasket: ~12.8 ch/mm²). The microgaskets also achieved high electrical isolation for neural recording (i.e., ~35 MΩ at 1 kHz) at the same high channel density. When mechanically compressed the microscale vias in the PDMSe microgaskets deform laterally, which could damage or enhance gasket-traversing conductive spring elements in each microscale via depending on their design. We have demonstrated that by lowering the height-to-width aspect ratio of the gasket vias, they can maintain their shape under clamping pressures high enough to achieve high isolation.

I. Introduction

Neural implant systems with higher-resolution neural interfaces are being used to facilitate improved experiments that could advance fundamental knowledge of brain activity and function in normal, diseased, and injured states. The incentive to create higher channel count implantable neural interfaces for recording and stimulating neural activity has been driven by the need to increase the spatial resolution of the interfaces for different tissue targets [1]–[4]. Typically neural interfaces that have a high channel count and high channel density are connected to implanted hermetically packaged electronics by an array of permanent bonds [5]–[8]. However, since interfaces with a high channel count are frequently integrated intimately into delicate neural tissue, the explantation of the full implant system to replace batteries or the electronics typically leads to considerable damage to the neural tissue immediately surrounding the interface. To avoid this problem in the clinic, commercially successful bioelectronic implants, such as deep-brain-stimulation devices cardiac pacemakers, use implantable connectors that can be used numerous times without a degradation in performance [9]. Reusable implantable connectors enable the interface to be detached from the implanted hermetic electronics enclosure. Doing so allows for battery changes and electronics upgrades to be performed without disturbing the delicate engagement between the tissue and the implanted interface (i.e., electrode arrays).

Implantable connectors based on a one-dimensional array of cylindrical electrical contacts are commonly found in clinical devices (Fig. 1). With this connector design, cylindrical contacts are integrated into the proximal end of a flexible lead that itself is inserted into an enclosure-integrated header that has a one-dimensional array of toroidal-shaped springs [10]. Channel-to-channel isolation is achieved with a linear array of compressible insulating toroidal bushings that are integrated between the toroidal springs in the header. The individual bushings in the linear array are radially compressed sequentially when the lead is inserted. The tight fit that forms between the mating parts keeps the lead centered and resists the flow of electrical current between adjacent channels despite being immersed in a conductive saline solution. This approach can reliably maintain good electrical connections and electrically isolate neighboring channels, while also being easy to handle and reconnectable for the replacement of batteries and/or electronics. Although this approach for implantable connectors has been very successful and can be found in many commercial devices, the underlying technology has not advanced significantly for many years. The low density of channels (i.e., ~0.0644 ch/mm²) prevents the development of small high-channel-count implementations. To serve the needs of next-generation bioelectronic implants that require an increasingly large numbers of channels in an upgradable and miniature form factor, a new implant connector technology is needed that can achieve both high channel density and high channel-to-channel impedance (e.g., industry standard for DBS is more than 50 KΩ for stimulation and more than 1 MΩ for recording). Existing high-channel-count interfaces are typically directly connected with implant electronics by permanent-bonding techniques, such as those used prominently by the commercial microelectronics industry (e.g., aligned flip-chip bonding, anisotropic conductive adhesive, conventional wire bonding, etc.), which prevent battery changes or implant upgrades without explanting the entire implant, including the interfaces integrated into delicate neural tissue [11]–[14]. To ensure a high channel-to-channel impedance and minimize current leakage, connections made to implanted devices with permanent electrical bonds must be carefully packaged. Coating permanently bonded implant assemblies (e.g., hermetically packaged electronics connected to high-channel-count interfaces) with a polymer dielectric material, which provides electrical isolation from the surrounding conductive fluids of the body, is a commonly used packaging approach. Due to their high-impedance and moisture-resistance properties, the most commonly used polymer materials to encapsulate implanted electronics are PDMSe (polydimethylsiloxane elastomer), polyurethane, parylene-C, PEEK (polyetheretherketone), and, LCP (liquid crystal polymer) [15]–[17].

Fig. 1.

Schematic diagram of a commercial implantable device with a reusable connector

Since available reusable implant connectors and the use of polymer encapsulated permanent bonds do not allow small high-channel-count implants to be serviced (i.e., battery changes, upgrade electronics, etc.) without full explantation, we are working to create a new technology for scalable high-channel-density reconnectable implant connectors. In our design we use microgaskets to have excellent electrical channel-to-channel isolation, despite long-term immersion in conductive body fluids. We use electrochemical impedance spectroscopy (EIS) to assess the electrical isolation achieved with microgaskets over the frequency range of interested in neurotechnology (i.e., 0.1 Hz to 10 kHz). We have also demonstrated that thinner gaskets are mechanically superior at maintaining their shape under high clamping pressures.

II. Description

The envisioned implantable connector assembly (Figure 2) includes: an enclosure for electronics that is hermetically bonded to a ceramic header with an integrated high-density feedthrough array, a gasket with micromachined holes aligned to the header feedthroughs, a contact-pad array of a flexible neural interface aligned to the gasket holes, an array of conductive elements individually bridging the vertical gap between the aligned arrays of header feedthroughs and neural-interface contact pads, and finally a rigid plate to compresses the assembly together with screws and springs. Contacts pads in the interface will be electrically connected with corresponding feedthroughs in the header with mechanically compressible conductive elements.

Fig. 2.

Exploded (top) and assembled cross-section (bottom) views of the proposed connector-assembly design.

Connections made between individual contact pads on the neural interface, compressible conductive elements, and metal feedthroughs in the ceramic header will be electrically isolated by the microgasket. Pressure applied by the rigid plate and screws will be used to clamp the assembly, deform the microgaskets to achieve good electrical isolation, and compress the conductive elements to achieve good electrical connection.

Pressure-driven gaskets, which are widely used to contain fluids at the macroscale, have been adapted for use at the microscale (i.e., microfluidics) [18], [19]. Although gaskets have been used to provide electrical isolation at the macroscale, microgaskets have not yet been used for electrical isolation in miniature systems. To achieve good isolating seals with gaskets, enough pressure must be applied to clamp the assembly and force at least one of the parts (e.g., the mating parts or the gasket itself) to deform enough to confirm intimately to the other surfaces. However, even seals made with well-clamped gaskets are not completely hermetic because water is expected to diffuse through the gasket material over time. In spite of this under-appreciated limitation, pressure-clamped gaskets are designed to achieve seals that have sufficiently high electrical isolation impedances for various applications. In this paper we describe the design and implementation of a set of experiments to quantify the ability of compressed polydimethylsiloxane elastomer (PDMSe) microgaskets to provide electrical isolation to a device submerged in a conductive physiological saline environment.

Although the microgasket is designed to provide high electrical impedance between adjacent channels and the surrounding conductive solution (e.g., saline), the vertical vias through the microgasket enable opposing pairs of aligned interface-contact pads and header-feedthrough pads to be physically and electrically connected by individual compressible conductive elements in each via. The compressible conductive elements will be designed to achieve multiple points of contact for increased reliability maintain a contact force sufficient to reliably maintain low contact resistance. Figure 3 illustrates three candidate approaches for implementing compressible conductive elements: (a) conductive micro/nanoparticles embedded in a compressible material, (b) conductive electrospun fibers embedded in a compressible material and (c) conductive microfabricated spring elements using photolithography [20], [21]. Although the focus of this paper is on the gasket and not the compressible conductive elements, large geometric changes to the gasket observed during its compression could have profound impact on some approaches used to implement compressible conductive elements. Therefore, in this paper we quantify the geometric changes that occur when the gasket is compressed and should be considered when designing and implementing the overall connector assembly. In application, during replacement of batteries/electronics, a major priority is proper functioning of the connector after reconnection. Integrating the microgasket with the header and enclosure assembly, which gets replaced during such procedures, will prevent the microgasket from undergoing any loading-unloading cycles during reconnection and thus ensure the highest microgasket performance and reliability. The purpose of the microgasket is to achieve and maintain high connection/channel impedance once fully clamped.

Fig. 3.

Approaches to electrically bridge the gap formed by the gasket between interface and header: (a) conductive nanoparticles, (b) microscale fuzz button formed with electrospun conductors, and (c) cantilevers

III. Experiments

We designed and executed experiments to characterize the dependence of electrical isolation on the pressure used to compress silicone microgaskets onto the contact-pad array of a neural interface. Studies were performed dry (in air) and while immersed in a conductive physiological saline solution. We also performed experiments to quantify geometric changes to microgasket vias when compressed.

A. Experiments to Characterize Electrical Isolation of Microgaskets

When pressure is applied to the connector externally (Figure 2) the microgasket is compressed and conforms to adjacent surfaces (i.e., neural interface contact-pad array and ceramic feedthrough array). Soft gaskets under higher pressures will conform to adjacent surfaces more intimately (i.e., fill in the topographic surface features even more thoroughly). As a result, the number and size of unfilled voids between the microgasket and the adjacent surfaces (i.e., ceramic header and the neural interface contact-pad-array) will be reduced, the electrical paths formed by saline-filled voids between adjacent contact pads will have higher impedance, the leakage currents between channels will be reduced, and the channel-to-channel isolation in the connector will be increased.

To test the capability of microgaskets to electrically isolate neural-interface-contact-pad arrays under different levels of pressure, we replaced the neural interface with a test device made using the same fabrication process, materials, and layer thicknesses. The test device consists of 49 disc-shaped 100-μm- diameter platinum contact pads are arranged in a high-density (~12.8 ch/mm²) hexagonal configuration with pads having a center-to-center spacing of 300 μm (Figure 4). The contact pads are connected to zero-insertion-force (ZIF) contact pads by traces that are integrated between two layers of 5-μm-thick polyimide. The total thickness of the test device is ~10 μm. The test devices were microfabricated in class 100/1000 cleanroom using a set of photolithographic, thin-film deposition, and dry etching processes [8], [22]. First, 5 μm of BPDA-PDA polyimide (U-Varnish S, UBE Ind.) was spin coated on a 100-mm-diameter HMDS coated Si wafer and cured at 250 °C under N₂. Electrode sites, interconnect traces, test structures, and connector pads were formed by a lift-off process using negative photoresist (nLOF 2035, MicroChemicals GmbH). The PI surface was activated in a reactive-ion-etching (RIE) O₂ plasma and then coated by 50-nm-thick film of SiC formed by plasma-enhanced chemical vapor deposition (PECVD) at 100 °C as a measure to improve adhesion. Later, we sputtered a 400-nm-thick Ti/Pt/Au/Pt/Ti metal stack that was patterned by liftoff. The metal structures were sealed by a top layer of 5 μm of polyimide, deposited and cured at 350 °C. Electrode sites, connector pads, and thread-set geometry, shown in Figure 4 were patterned by RIE dry etching using O₂ (to remove PI) and SF₆ (to remove Ti) plasmas. The microgasket was fabricated from a ~50-μm-thick layer of PDMSe (BLUESIL RTV 3040, Elkem) that was spin coated onto a silicon wafer and then cured. Disks with a diameter of 1 cm were then cut from the cured PDMSe microgasket, peeled off the wafer, and placed on top of the contact-pad array on the test device. After an individual circular microgasket is aligned to the contact-pad array on the test structure, it is compressed between two rigid flat plates using a pneumatic piston to apply pressure uniformly (Figure 5). The rigid plates were used to spread the applied pressure over the compressed region uniformly. The compressed assemblage was then submerged in a mimic of physiological saline (i.e., 10 mM phosphate buffered saline (PBS)).

Fig. 4.

Illustration of contact pad array and the test structure (Left) and picture of laser milled via array in a microgasket (right)

Fig. 5.

Schematic diagram of the experimental set-up for testing the electrical isolation of thin PDMSe

We used electrochemical impedance spectroscopy (EIS) to quantify complex channel-to-channel isolation impedance over a wide frequency range. The measurements were performed between pairs of contact pads integrated into the thin-film polyimide test device, isolated by the microgasket, and submerged in room temperature PBS (~20 °C). To obtain experimental EIS measurements, the test devices were connected to a potentiostat (PGSTAT302N, Metrohm Autolab) via a ZIF connector. As shown in Figure 5, the EIS measurements were obtained with a two-electrode setup, in which one electrode acted as a combination counter/reference electrode and the rest of the 48 contact pads on the test structure were tested sequentially as independent working electrodes. The voltage signal using during EIS experiments had an amplitude of 100 mV and was swept from 0.1 Hz to 100 kHz. Using a lower amplitude can result in noisy measurements when working with high impedance values. Using a larger amplitude is limited by the potentiostat.

To test the dependence of electrical channel-to-channel isolation on the clamping pressure applied to the microgasket and test device, the assembly was subjected to three different clamping pressures: 0.07, 1.88 and 3.76 MPa in room-temperature PBS. Each experiment was performed for 4 days.

B. Mechanical Stability of Vias through the Microgasket

The application of clamping pressure is essential for implanted electrical connectors to function as intended. Specifically, clamping forces help achieve and sustain both good electrical contact and electrical isolation between adjacent channels and the surrounding conductive saline environment. Gaskets are often used between clamped mating part to form a seal that prevents leakage of fluids, gasses, and electrical current flow. When higher pressure is applied, gaskets undergo a larger mechanical deformation. In this work we used PDMSe, which can tolerate high strains, as the gasket material. For an optimized connector design, the mechanical behavior of the clamped parts that undergo large deformation needs to be understood.

The microgaskets tested here have vias, which will be needed in future work to accommodate conductive elements (Figure 3). Since the Poisson’s ratio of PDMSe is ~0.5, the clamping pressure that results in a vertical uniaxial compression strain will also yield a large transverse strain (Figure 6a–b). As a result, the diameter of the vias through the gasket will be reduced as the clamping pressure increases. If the clamping forces are high enough, the vias can be squeezed closed completely by this transverse gasket deformation (Figure 6c). Depending on the design of the conductive elements, which are located inside the vias and are responsible for electrically connecting opposing contacts pads, the shrinkage of vias due to transverse gasket forces could interfere with their operation (e.g., block the movement of micromachined springs). To understand the mechanical behavior of microgaskets and vias under large deformations, we have designed an experiment to study the dependence of these deformations on clamping pressure.

Fig. 6.

Schematic diagram of the geometric deformation of PDMSe as a function of applied pressures and gasket thickness.

In our experiment, each microgasket tested had 49 vias of the same size (100 μm diameter) and same positions (hexagonal-close-packed arrangement with 300 μm center-to-center spacing) as the contact pads on the electrical test devices (Fig. 4). The microgaskets were fabricated by spin coating and curing PDMSe (BLUESIL RTV 3040, Elkem) on a polycarbonate sheet that was attached to a 100-mm-diameter silicon wafer with adhesive tape. The polycarbonate sheet was used to facilitate the subsequent removal of the microfabricated silicone parts by mechanical peeling. A 30-μm-thick layer of photoresist (AZ9260, MicroChemicals) is then spin coated onto of the silicone layer. The vias are formed by laser ablation using a picosecond laser (Laser J-355-PS, Oxford). The photoresist sacrificial layer is then etched away in acetone to remove the considerable amount of debris that forms during the laser- ablation process. The microfabricated microgaskets are then peeled off the polycarbonate sheet as needed for testing.

To quantify the mechanical deformation of microscale vias in PDMSe microgaskets as a function of applied clamping pressure, the microgaskets are tested in an experimental setup that allows direct visual observation of the deforming microscale vias with a microscope. Specifically, as illustrated in Figure 7, the experimental setup positions the gasket between two vertically oriented rigid-borosilicate-transparent glass plates arranged horizontally on an experimental bench. The right plate is fixed while the left plate is movable and is pushed by a cylindrical rod connected to a pneumatic cylinder piston (SR-046-DPY, Bimba) to compress the microgasket. A microscope-based camera is used to directly observe the transverse deformation of the microscale vias. The via height-to-diameter aspect ratio is an important parameter that impacts the change in via geometry observed as a function of applied pressure. For our initial experiments, we chose to work with microgaskets made of two different thicknesses (i.e., ~100 and ~33 μm). Since all vias tested has a diameter of 100 μm, the aspect ratios tested were ~1 and ~1/3.

Fig. 7.

Schematic cross-sectional diagram of the experimental set-up for observing and quantifying the geometric deformation of microscale-vias in the PDMSe microgasket as a function of applied pressure.

IV. Experimental Results

A. Experiments to Characterize Electrical Isolation of Microgaskets

A summary of the microgasket-isolation experiments, which were performed to explore the impact of clamping pressure on channel-to-channel isolation, is illustrated in the plot of impedance as a function of frequency shown in Figure 8. The highest impedance measurements were obtained for the dry condition (i.e., when no saline is present) and was limited by the parasitic capacitances of the device and measurement setup. The measurements for the dry condition had a high signal-to-noise ratio (SNR) above 100 Hz and a very low SNR below 100 Hz. The SNR of the measurements below 100 Hz was limited by the very high impedance at these frequencies and the low drive signal (100 mV). The lowest impedance measurements were obtained for the wet condition (i.e., when the device was submerged in PBS without a microgasket). In this case the current flowed directly through the saline solution between the metal pads (i.e., representing complete isolation failure) and the impedance values were determined by the electrochemical interaction between the platinum pads and the saline solution. The measurements for the wet condition also had a low noise level over all frequencies tested (i.e., from 1 to 100,000 Hz).

Fig. 8.

Impedance spectra from microgasket-isolation experiments performed at different levels of microgasket-compression pressure. Plots of the impedance measured without a microgasket under dry (no saline) and wet (immersed) conditions are included for comparison.

In an earlier study [23], we compared the electrical isolation performance achieved by encapsulation coatings used with implantable devices (i.e., parylene-C or PDMSe) with PDMSe microgaskets clamped with high pressures [23]. That prior work demonstrated that reusable microgaskets compressed with high pressures could achieve a high level of electrical isolation that was essentially equivalent to permanently encapsulating devices with parylene-C or PDMSe. Here we present new results from microgasket experiments performed at both high and low pressures (Fig. 8). Although our new experiments also achieved high electrical isolation when clamped at high pressures (i.e., 1.88 to 3.76 MPA), when clamped at low pressure (i.e., 0.07 MPa) we observe that isolation performance varies significantly with frequency. Specifically, we observed that although the electrical isolation performance of lightly clamped microgaskets is as good as that achieved with strongly clamped microgaskets at frequencies >40 kHz, the insolation performance is greatly reduced (i.e., approaches that achieved without a gasket at all) at low frequencies.

Circuit diagrams can help explain the impedance results observed in Fig. 8 and can support the anticipation of impedance results at other clamping pressures. A high-level circuit diagram for the experiments is shown in Fig. 9 and diagrams for the different cases tested are shown in Figs. 10–12.

Fig. 9.

Two circuit diagrams illustrating the total impedance presented to the test equipment: (left) given as the parallel combination of the overall device impedance Z_device and the parasitic capacitance of the printed circuit board C_p(PCB) and the parasitic capacitance of the wires connecting the board to the test equipment C_p(wires); (right) with the device impedance broken down as the parallel combination of the impedance between the metal pads from the two channels tested Z_pad–pad and the parasitic impedance between the traces connecting each metal pad to the circuit board (Z_p(traces)). For simplicity, the parasitic capacitances of the PCB and wires between the PCB and test equipment, which do not change with experimental conditions, are combined as a parasitic impedance Z_parasitic.

Fig. 10.

Circuit diagrams of the wet case for (a) the pad-to-pad impedance Z_pad–pad, (b) the parasitic trace impedance Z_p(traces), and (c) the impedance presented to the test equipment.

Fig. 12.

Circuit diagrams of the case of a submerged device clamped at high pressure for (a) the pad-to-pad impedance Z_ch-ch, (b) the parasitic trace impedance Z_p(traces), and (c) the impedance presented to the test equipment.

1). Wet Electrodes without Gasket

When the device is submerged in saline, the metal pads are put in direct contact with saline and traces from each pad are isolated from the saline by the top and bottom polyimide layers. Circuit diagrams for the impedance between the metal pads Z_pad–pad, traces Z_p(traces), and that presented to the test equipment are shown in Fig. 10. The circuit for the pads includes elements associated with the electrochemical impedance of metal pads in saline (i.e., parallel charge-transfer resistance R_CT and a constant-phase element Z_CPE) as well as a series resistance between the pads through saline R_saline(pads) (Fig. 10a). The circuit for the traces includes capacitances between the traces through the polyimide C_PI, from the traces through the polyimide to the saline C_PI(saline), and a series resistance through saline between each trace R_{saline(traces)} (Fig. 10b).

The relative significance of each element and their impact on the overall impedance presented to the test equipment can vary significantly with frequency. Since R_CT (>>1 GΩ) is typically much higher than the Z_CPE (0.6 MΩ at 1 kHz) over the frequencies tested, R_CT can be omitted. Since Z_parasitic (96.4 MΩ at 1 kHz) is far larger than the impedance between the pads 2 ∙ Z_CPE + R_saline(pads) (1.34 MΩ at 1 kHz) Z_parasitic too can be omitted in this case. As a result, the magnitude of the impedance in this case will be equal to 2 ∙ Z_CPE until it approaches and falls below R_saline(pads) (0.14 MΩ). The frequency response for the wet case, with a CPE phase of ~78°, is clearly shown in Fig. 8.

2). Dry Electrodes without Gasket

In the dry case, the saline is replaced with air. Circuit diagrams for Z_pad–pad, Z_p(traces), and Z_parasitic are shown in Fig. 11. In this case the electrochemical impedance elements are replaced with parasitic air capacitances between pads C_air(pads) and traces C_air(traces).

Fig. 11.

Circuit diagrams of the dry case for (a) the pad-to-pad impedance Z_pad–pad, (b) the parasitic trace impedance Z_p(traces), and (c) the impedance presented to the test equipment.

In this case, the extremely high impedance of air and polyimide cause Z_pad–pad (>10 GΩ) to be much higher than Z_p(traces) (726 MΩ at 1 kHz) which is much higher than Z_parasitic (96.4 MΩ at 1 kHz) at the frequencies tested. As a result, the impedance measured in the dry case (85.1 MΩ at 1 kHz) is dominated by Z_parasitic. Since the impedance for the dry electrode case is dominated by parasitic capacitances (e.g., parasitic capacitance of the circuit board C_p(PCB), wires C_p(wires) attached to the device, and the test equipment), we would expect its magnitude to fall linearly with increasing frequency and its phase to be −90°. This expectation is confirmed by our high SNR experimental results (Fig. 8 at frequencies above 100 Hz).

3). Submerged with Gasket Clamped at High Pressure

When a microgasket is clamped at high pressures (1.88 and 3.76 MPa) to a submerged device, the thickness of the electrical pathway in saline between the metal pads tested is greatly reduced. Circuit diagrams for the physically constrained electrochemical impedance between pads, Z_{p(traces_wet)}, and Z_parasitic are shown in Fig. 12.

Since, the value of R_saline increases from 0.14 MΩ in the wet case to R_{saline(gasket)} now over ~10 GΩ, which is higher than the impedances measured in these experiments (Fig. 8), the pad-to-pad impedance can be ignored in this case. Replacing the air surrounding the traces with saline (i.e., they are not covered by the gasket), reduced the value of the parasitic trace impedance from Z_{p(traces_dry)} (726 MΩ at 1 kHz) to Z_{p(traces_wet)} (75 MΩ at 1 kHz). As shown in Fig. 12b, Z_p(traces) is a combination of C_PI + C_PI(saline)+R_{saline(traces)}. Since Z_{p(traces_wet)} is now less than Z_parasitic (96.4 MΩ at 1 kHz), the total impedance in the wet case (42.2 MΩ at 1 kHz) is less than the total impedance in the dry case (85.1 MΩ at 1 kHz). Since the impedance in this case is dominated by parasitic capacitances, we again expect its magnitude to fall linearly with increasing frequency and have a phase near to but less than −90°. These expectations are confirmed by our experimental results (Fig. 8) with phase being ~80°, except where observations are limited by the presence of noise.

4). Submerged with Gasket Clamped at Low Pressure

When a microgasket is clamped to a submerged device at low pressures (0.07 MPa), the thickness of the electrical pathway in saline between the metal pads is reduced but not as much as with high pressure (Fig. 12a). As a result, the series resistance R_{saline(gasket)} (2.26 MΩ) is higher than R_saline(pads) (0.14 MΩ) for the wet case but much lower than the clamped-at-high-pressure case (~10 GΩ).

Since at very high frequencies the parasitic impedance will dominate because it is less than the series resistance, this low-pressure-clamping case will behave like the high-pressure-clamping case above. As the frequency drops from very high frequencies, the parasitic impedance will increase until the impedance value of R_{saline(gasket)} begins to dominate, the phase trends towards 0°, and the slope of the impedance plot flattens out. Reducing the frequency further causes the electrochemical impedances at the metal pads, which are in series with R_{saline(gasket)}, to increase until they dominate, and this low-pressure-clamping case behaves like the wet case (i.e., no gasket). The above description is consistent with the experimental results shown in Figure 8, which were obtained for the case when submerged and clamped at low pressure (i.e., 0.07 MPa).

We expect that if the clamping pressure were gradually increased, the series resistance between the pads would also gradually increase. In addition, the plot of impedance would transition from the higher impedance values of the high-clamping-pressure case to the lower impedance levels of the wet case at lower frequencies and at higher impedance values.

An analysis of the impedance magnitude and phase data for all channels individually at low clamping pressures revealed that some channels transitioned from the higher-impedance behavior to the lower-impedance behavior at different frequencies between ~100 Hz and ~10 kHz. Since the change in impedance magnitude and phase between the two behaviors is very large and occur at different frequencies, the error bars for the low-clamping-force impedance and phase plots in Fig. 8 are also very large over this frequency range.

B. Experiments to Characterize the Mechanical Stability of Vias through the Microgasket

An example of the reduction in microscale via area observed when clamping pressure is applied to a PDMSe microgasket is shown in Figure 13. The data shown in a quantitative plot of the normalized microscale-via area as a function of microgasket clamping pressure (Figure 14), illustrates that the area of the microscale vias decreases as expected with increased applied pressure. At lower applied pressures, the reduction in the area of the 100-μm-diameter vias is independent of microgasket thickness (i.e., aspect ratio 1 or 1/3). However, as the pressure increases, thinner microgaskets (~33 μm) that have a lower thickness-to-diameter vias (i.e., aspect ratio of 1/3) perform better (i.e., a lower reduction in via area) than thicker microgaskets (~100 μm) with a higher aspect ratio. Specifically, at ~3.6 MPa, thinner microgaskets with an aspect ratio of 1/3 retain ~30% of the original via area, whereas the thicker microgaskets with a higher aspect ratio of 1 retain only <~10% of the original via area. This result indicates that at higher pressures, microgaskets with lower aspect ratios perform better at retaining via area.

Fig.13.

A series of microscope images that illustrate an example of the change in microscale via area as a function of clamping pressure applied to a PDMSe microgasket.

Fig. 14.

Plot of the changes in normalized microscale via area as a function of clamping pressure applied to the PDMSe microgaskets. The initial diameter of microscale vias is 100 μm. The initial thickness of microgaskets tested were ~33 μm and ~100 μm.

V. Conclusions

Our electrical experimental results have demonstrated that PDMSe microgaskets and high clamping pressures can provide excellent electrical isolation for neural stimulation (i.e., ~5 MΩ at 10 kHz) and neural recording (i.e., ~35 MΩ at 1 kHz) (Figure 8). These values are much higher (>60X) than worst-case isolation (i.e., wet case: bare metal pad in saline) and lower (~0.5X) than best-case isolation (i.e., dry case: measured outside saline). If lower clamping pressures are used, the isolation impedance performance can vary with frequency. Specifically, the isolation impedance achieved with low clamping pressure is nearly equivalent to the worst case (i.e., no gasket) at low frequencies and identical to the case of a gasket with high clamping pressure at high frequencies.

Circuit models based on the physical constraints and electrical properties of each case yield predictions that align with the experimental results and could guide future design optimizations.

Other than clamping pressure, several other factors (e.g., pad density, material hardness, pad size, surface roughness etc.) are also responsible for the electrical isolation performance of microgaskets. Further investigation of these factors on implant-connector performance is necessary for applications that require other values of these key design parameters. For example, applications requiring a higher or lower via-density will likely require a higher or lower clamping pressure to achieve the same level of electrical isolation.

Our mechanical experimental results have revealed that when PDMSe microgaskets containing 100-μm-diameter microvias are mechanically compressed, the gasket material expands laterally into the vias. As the material from the compressed gasket moves into the vias, the diameter of the vias can shrink significantly and can even close off completely with the application of enough clamping pressure (Figure 13). Although there is an inverse linear relationship between clamping pressure applied to the microgaskets and via area (i.e., higher microgasket clamping pressure results in less via area), the slope of this relationship is dependent on the aspect ratio. Specifically, we have observed that lower via thickness-to-diameter aspect ratios result in less reduction of via area with applied pressure. To avoid excessive via-area reduction without reducing clamping pressure, one can use larger diameter vias or thinner gaskets. Larger diameter vias will lead to lower channel count density and higher connector size and thinner gaskets could be more challenging to work with. For each application there likely exists a trade-off to be balanced between electrical isolation, clamping pressure, via density, channel count, connector size, and ease of use.

Future work focused on the design and implementation of via-spanning compressible conductive elements needs to take the results described here into account, because they have implications on the performance and ultimately the choice of the strategy used to achieve vertical conduction through the compressible conductive elements in the via from the pad array of the interface to the pad array of the package. Specifically, the dramatic reduction in via area with clamping pressure will cause the gasket material to mechanically interact strongly with the vertically conducting via elements. Although these experiments were not performed, we anticipate that for some approaches the result could be catastrophic failure and for other approaches the result could actually be improved performance.

For example, we expect that the vertical-via-conduction approach for the compressible conductive elements described in Fig 3(c), which achieves vertical conduction with delicate prestressed microcantilevers that curl up from the lower pad layer to touch the upper pad layer, could be adversely impacted severely by the strong encroachment of the compressed gasket material (e.g., the fragile microcantilevers could be broken and result in an open-circuit failure). In contrast, we expect that the performance of the vertical-via-conduction approach described in Fig 3(a), which achieves vertical conduction through the use of conductive particles in a compressible material matrix that is packed or otherwise integrated into the original uncompressed via volume, could be improved (e.g., lateral compression of the vertical-conduction plug made of conductive particles and compressible matrix by the gasket sidewalls will result in upward and downward forces on the opposing conductive pad arrays). We expect that the impact on the performance of the vertical-via-conduction approach described in Fig 3(b), which uses conductive electrospun fibers embedded in a compressible material, would be similar.

Biography

Paritosh Rustogi was born in New Delhi, India in 1991. He received Bachelor of Technology in Mechanical and Automation Engineering from Guru Gobind Singh Indraprastha University, New Delhi, India in 2013. Later, he finished his Master of Science in Mechanical Engineering from University of Florida, Gainesville, Florida, U.S.A.

Currently, he is pursuing PhD. In Electrical & Computer Engineering from University of Florida. His work is focused on developing implantable-rematable-backend connectors for neural implants. He interests are MEMS, micro-fabrication, and neural implants. Rustogi is part of IEEE since 2019.

Jack W. Judy received the B.S.E.E. degree in electrical engineering from the University of Minnesota, Minneapolis, USA, in 1989 and the M.S. and Ph.D. degrees in electrical engineering from the University of California, Berkeley, USA, in 1994 and 1996 respective. He is currently the Intel and Charles E. Young Nanotechnology Professor of Electrical and Computer Engineering at the University of Florida with courtesy appointments in Biomedical Engineering and Neurology. He is also the Directory of the Nanoscience Institute for Medical and Engineering Technology (NIMET).

From 1996 to 1997, he was with Silicon Light Machines, an optical MEMS startup company commercializing a projection-display technology. From 1997–2013 he was a professor in the Electrical Engineering Department at the University of California, Los Angeles, USA. There he served as the Director of the UCLA Instructional Microfabrication Laboratory and the UCLA Nanoelectronics Research Facility. He also served as the co-founder and Director of UCLA NeuroEngineering Training Program. From 2009 to 2013 he served as a Program Manager at DARPA in the Microsystems Technology Office, where he founded and led the Reliable Neural Technology (RE-NET) Program. The RE-NET Program was a large multidisciplinary effort to develop the neural interfaces needed to control advanced prosthetic limbs used by wounded warriors returning from battle with limb amputations. He has received the National Science Foundation Career Award, the Okawa Foundation Award, and the Office of the Secretary of Defense Medal for Exceptional Public Service.