Development of Microelectrode Arrays for Artificial Retinal Implants Using Liquid Crystal Polymers

Abstract

Purpose.: To develop a liquid crystal polymer (LCP)-based, long-term implantable, retinal stimulation microelectrode array using a novel fabrication method.

Methods.: The fabrication process used laser micromachining and customized thermal-press bonding to produce LCP-based microelectrode arrays. To evaluate the fabrication process and the resultant electrode arrays, in vitro reliability tests and in vivo animal experiments were performed. The in vitro tests consisted of electrode site impedance recording and electrode interlayer adhesion monitoring during accelerated soak tests. For in vivo testing, the fabricated electrode arrays were implanted in the suprachoroidal space of rabbit eyes. Optical coherence tomography (OCT) and electrically evoked cortical potentials (EECPs) were used to determine long-term biocompatibility and functionality of the implant.

Results.: The fabricated structure had a smooth, rounded edge profile and exhibited moderate flexibility, which are advantageous features for safe implantation without guide tools. After accelerated soak tests at 75°C in phosphate-buffered saline, the electrode sites showed no degradation, and the interlayer adhesion of the structure showed acceptable stability for more than 2 months. The electrode arrays were safely implanted in the suprachoroidal space of rabbit eyes, and EECP waveforms were recorded. Over a 3-month postoperative period, no chorioretinal inflammation or structural deformities were observed by OCT and histologic examination.

Conclusions.: LCP-based flexible microelectrode arrays can be successfully applied as retinal prostheses. The results demonstrate that such electrode arrays are safe, biocompatible, and mechanically stable and that they can be effective as part of a chronic retinal implant system.

Electrical stimulation of the remaining retinal neurons of patients with degenerated photoreceptors has been studied as a potential method for providing artificial vision. To transfer and control such electrical stimulation, several research groups have developed polymer-based flexible microelectrode arrays. ^1–4 To date, several polymer materials, including polyimide, parylene, and silicone, have been used as substrate materials for retinal stimulation electrode arrays. These polymers are thin, flexible, and biocompatible—suitable characteristics for minimally invasive retinal electrode arrays. Although retinal stimulation electrode arrays fabricated on these polymers have been reported to be safe and effective in previous in vivo and in vitro studies, including animal and human trials, ^1–4 there is controversy about the long-term reliability of the polymers. These concerns are related to the polymers' relatively high water absorption and unstable interfacial adhesion properties in aqueous environments. ^5–7  

Liquid crystal polymers (LCPs) are flexible, mechanically stable, and biocompatible materials that have very low moisture absorption (<0.04%) compared with polyimide, parylene, and silicone. ^8–14 LCPs exhibit excellent barrier properties against various chemicals and can be thermally bonded to each other without adhesives. ^8–15 Because of their high reliability under harsh environmental conditions, LCPs have been investigated as long-term reliable substrate materials for high-performance printed circuit boards, ^8,15 microelectromechanical system sensors, ¹⁰ and neural prostheses. ^11,12  

In this article, we report the development of a novel retinal stimulation microelectrode array using LCPs and report on the electrode array's performance during in vitro and in vivo experiments. A simplified fabrication process for such LCP-based microelectrode arrays is also introduced. 

Materials and Methods

Fabrication Process

The fabrication process, which uses laser micromachining and customized thermal-press bonding, is shown in Figure 1a. Briefly, a 25-μm-thick, low-melting-temperature (280°C) LCP (Vecstar FA-25N; Kuraray Co., Ltd., Tokyo, Japan) film and a 25-μm-thick, high-melting-temperature (315°C) LCP (Vecstar OCL-25N; Kuraray Co., Ltd.) film were cut into 100-mm-diameter circles using a UV laser drilling system (Flex5330; Electro Scientific Industries, Inc., Portland, OR) to create a substrate and a cover, respectively. In the laser machining process, alignment marks were engraved on both the substrate and the cover pieces, and electrode site windows (500-μm diameter for the stimulation sites and 1400-μm diameter for a reference site) were created in the cover. The machined substrate LCP was attached to a similarly sized silicon wafer using photoresist (AZ4620; AZ Electronic Materials, Luxembourg, Luxembourg) as an adhesive before it was subjected to additional processes that required planar surface properties. ^10–12,14  

Figure 1.

 

(a) Representative schematic of LCP-based microelectrode array fabrication process. Laser machining was used for patterning the substrate and cover films and for cutting the electrode array outlines. Thermal-press bonding was performed to create the LCP multilayered structure. Total thickness of the structure is controllable from 50 to 75 μm, with a 25-μm-thick additional substrate. (b) Schematic diagram of LCP-based retinal electrode array. This structure has seven stimulation sites and one reference site. The diameters of the stimulation site and reference site windows are 500 μm and 1400 μm, respectively.

Subsequently, titanium (100-nm-thick), gold (400-nm-thick), and titanium (100-nm-thick) layers were consecutively deposited on the LCP substrate by a sputter machine (ALPS-C03; Alpha Plus Co., Ltd., Pohang, Korea). The Ti layers form biocompatible adhesion layers between the electrode site metals (Au, Pt, and IrOx) and the LCP films. ^11,12,14 Before metal patterning, photoresist (AZ1512; AZ Electronic Materials) was spin-coated on the metal-bearing substrate. Photolithography was then performed using a mask aligner machine (MA6/BA6; SUSS MicroTec, Garching, Germany). Subsequently, the metal microelectrode patterns were created by a conventional wet-etching process. 

After the patterning process, the substrate was released from the silicon wafer using acetone. The cover was then positioned on the substrate using the alignment marks, and the pair was placed into a custom aluminum mold comprising 100-mm-diameter planar plates and four alignment pins. Thermal-press bonding ¹⁵ was performed at 300 psi (2.1 MPa) and 285°C for 45 minutes using a heated press (model CH, press no. 4386; Carver, Inc., Wabash, IN). Subsequently, the laminated structure was cut into the final microelectrode array shape (Fig. 1b) using the aforementioned UV laser machining system (Electro Scientific Industries). 

In Vitro Reliability Tests

To evaluate the reliability of polymer-based electrode arrays, various testing methods have been used. ^6,7 For electrical reliability testing, electrical leakage current measurement between adjacent leads has been used. ⁷ For mechanical reliability testing, interlayer adhesion strengths have been measured. ⁶ We considered both these testing methods to evaluate the overall reliability of the fabricated microelectrode arrays; however, in this study, we focused on tests that indicate long-term structural reliability. 

To assess the long-term structural reliability of the LCP-based electrode arrays within a relatively short time, electrode site impedance and adhesion strength of the LCP multilayered structures were monitored during in vitro accelerated (75°C) and nonaccelerated (37°C) soak tests. The soak tests were performed in phosphate-buffered saline (PBS) solution (Gibco 10010; Invitrogen Life Technologies, Carlsbad, CA). Monitoring of electrode site impedance provided information on electrical connectivity and the exposed site metal status. We selected five sites (stimulation channels 1, 2, 3, 4 and the reference electrode; Fig. 2b) among the eight available sites and, during the soak tests, regularly measured their impedance (magnitude) at 1 kHz 5 mV amplitude sine waveform using an impedance analyzer (IM6e; Zahner-Elektrik, Kronach, Germany), as shown in Figure 2a. 

Figure 2.

 

In vitro reliability tests. (a) Schematic diagram of electrode site impedance measurement apparatus. (b) Schematic diagram of electrode site arrangement showing channel numbers 1 to 4 and the reference electrode. (c) Cross-sectional diagram showing layers in the blister test samples for determination of adhesion strength between LCP/LCP and Ti/LCP interfaces. (d) Images of samples for blister testing. (e) Conceptual diagram of soak and blister testing process. (f) Image of blister test apparatus.

Monitoring of adhesion strength provided information about the durability of the multilayered structure. Adhesion strength was measured using a customized blister test. ¹⁶ To compare the results, previously reported data of polyimide adhesion strengths, comprising polyimide/polyimide and titanium/polyimide interfaces, were used (Lee SW, et al. IOVS 2007;136:ARVO E-Abstract 664). Figure 2c shows a cross-section of the structure of the blister test samples, which were fabricated similarly to the structure of the aforementioned LCP-based microelectrode arrays. The tested structures consisted of bonded high- and low-melting-temperature LCPs with metal at the LCP interface. Given that a Ti layer was used as the adhesion layer in the Au microelectrode arrays, we focused on testing LCP/LCP and Ti/LCP interface adhesions. The LCP/LCP sample consisted of a high-melting-temperature LCP substrate, a low-melting-temperature LCP interlayer, and a high-melting-temperature LCP cover (Fig. 2c). On the interlayer and the cover, 4-mm ⊘ and 2-mm ⊘ holes, respectively, were created by UV laser machining (Electro Scientific Industries). The Ti/LCP sample consisted of the same LCP/LCP structure but with a Ti (100-nm) layer deposited on the LCP cover. 

A conceptual diagram of the customized blister test is presented in Figure 2e. Subsequent to soak testing, adhesion strength was measured as the critical pressure (MPa), which can initiate crack propagation between the films. To perform the blister test, the cover side of the sample was attached to a metal holder, which had a 4-mm ⊘ hole, using an acrylate adhesive (Uni-401; Dong Sung Uni-Tech, Pocheon, Korea) with a bonding strength of 200 kg/cm² (19.6 MPa). The metal holder was positioned in the apparatus (Fig. 2f), and pressure was applied to the sample using N₂ gas supplied through the 4-mm ⊘ hole in the metal holder. The applied pressure was controlled by a precision regulator (Harris Products Group, Mason, OH). The maximum available pressure was 1.1 MPa. During the crack initiation test, changes in blister diameter and height were monitored with two charge-coupled device cameras (MTV-7266ND; Mintron Enterprise Co., Taipei, Taiwan). 

In Vivo Animal Experiments

To demonstrate the application feasibility of the fabricated electrode arrays, in vivo animal experiments were performed using New Zealand White rabbits. All procedures conformed to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. General anesthesia was induced by intramuscular injection of tiletamine/zolazepam (Zoletil; Virbac Laboratories, Carros, France) and xylazine (Rompun; Bayer AG, Leverkusen, Germany) in a 1:1 mixture. The rate of injection was 0.6 mL/kg. A conjunctival incision was made along the limbus at the 1 o'clock position, and a 4-mm scleral incision parallel to the limbus was made with a crescent knife (Sharptome 74–1010; Surgical Specialties Co., Reading, PA). The fabricated LCP microelectrode arrays were implanted into the suprachoroidal space under pan funduscopic examination assisted by an indirect contact lens (Quad Pediatric; Volk Optical Inc., Mentor, OH) to locate the electrode array under the visual streak adjacent to the optic disc. The external part of the electrode array was stabilized by placement over the sclera and under the extraocular muscles, similar to the fixation achieved in circumferential scleral buckling. 

After implantation, electrophysiological tests were performed to determine electrode array functionality. Biphasic current stimulation pulses were applied to the rabbit's retina through the electrode arrays, and electrically evoked cortical potentials (EECPs) were simultaneously recorded from a stainless needle electrode (Rochester Electro-Medical Inc., Tampa, FL) in the visual cortex using a multichannel neurophysiological workstation (Tucker-Davis Technologies, Alachua, FL). These EECPs recordings were acute and performed using a previously reported system. ¹⁷ Briefly, the recording electrode (a single-needle electrode) was inserted into a fine hole drilled in the skull (without craniotomy). The hole was located 6 mm anterior and 4 mm contralateral to lambda, an area previously reported as a good position for EECP recordings. ¹⁷ The reference electrode (a single-needle electrode) was inserted into a hole located 20 mm anterior to the lambda. The counter electrode (ground electrode) was inserted in the ipsilateral ear. 

For the long-term biocompatibility and stability evaluation, the LCP microelectrode arrays were implanted in five rabbit eyes and monitored using optical coherence tomography (OCT; Cirrus OCT, Carl Zeiss, Dublin, CA) for 3 months. During the test period, fundus examinations were performed to evaluate any inflammatory changes or other complications in vitreous and retinal areas. After 4 months, two rabbits were euthanatized, and their eyes were enucleated to determine cataract or other morphologic changes in the eye. Histologic changes in the retina were evaluated using light microscopy and a hematoxylin-eosin stain. 

Results

Fabrication Results

The LCP-based microelectrode arrays were fabricated using the process described, and their morphologies were examined with a field emission scanning electron microscope (FE-SEM; S-4800 UHR FE-SEM; Hitachi High-Technologies, Tokyo, Japan). Figure 3 shows the array outline, the Au electrode site windows, the LCP cover surface, and the overall structure. The FE-SEM images indicate that laser cutting produced a smooth, rounded edge on the array outline (Figs. 3a, b). The Au electrode site/LCP window edges were smooth, distinct, and without misalignment (Figs. 3c, d). In addition, no burrs or residues were observed in the surrounding areas. 

Figure 3.

 

Photographs of LCP-based Au microelectrode array. (a) FE-SEM image of microelectrode array (top view). (b) Oblique view of a portion of the array edge. (c) 500-μm diameter Au site. (d) Portion of the Au site window edge. Laser-machined site windows and structure outlines exhibited clear, smooth, and rounded edge features. (e) Image of the overall structure.

In Vitro Reliability Tests

The site impedance of the fabricated microelectrode array was monitored during 9-week soak tests at 37°C and 75°C. For the first week, impedance was measured daily, and in the remaining weeks impedance sampling was performed once a week. As shown in Figure 4, the electrode impedance showed an initial drop before reaching steady values. The impedance from the 75°C soak stabilized more quickly than that from the 37°C soak. Such a decrease in impedance has been observed by other groups who used various types of neural probes, ^18–21 and the change has been attributed to metal-fluid interface equilibration. ²⁰ This change could have been accelerated at higher temperature. After the stabilization period, the impedance of each of the electrode sites was maintained over the 8 weeks, and there were no marked differences between the soak test results from the two soak temperatures (Fig. 4). These results showed that the electrical connections of all test channels were sustained and indicated that the exposed electrode sites on LCP were well preserved during the test period. 

Figure 4.

 

Electrode site impedance monitoring data. (a) Magnitude of impedance of Au electrodes on LCP under 37°C PBS soak test. (b) Magnitude of impedance at 75°C. The electrode impedance showed an initial drop before reaching steady values. Impedance from the 75°C soak stabilized more quickly than that from the 37°C soak. After the stabilization period, the impedance of each of the electrode sites was maintained over the 8 weeks, and there were no marked differences between the soak test results from the two soak temperatures.

The blister test results (Fig. 5) showed that the LCP/LCP and Ti/LCP adhesions were strong and reliable in comparison to polyimide/polyimide (PI/PI) and titanium/polyimide (Ti/PI) during 8-week soak tests. Initial adhesion strength data revealed that the LCP/LCP (1.0897 ± 0.0138 MPa) and Ti/LCP (1.0097 ± 0.0807 MPa) interfaces were stronger than the PI/PI (0.9862 ± 0.0712 MPa) and Ti/PI (0.4414 ± 0.0253 MPa) interfaces (Fig. 5a). In nonaccelerated soak tests at 37°C, the LCP/LCP and Ti/LCP interfaces showed no change in adhesion strength (Fig. 5b) during the 8-week test period. In the accelerated soak tests at 75°C, during the same test period the PI/PI and Ti/PI adhesion strengths markedly decreased by 58.7% and 63%, respectively (Lee SW, et al. IOVS 2007;136:ARVO E-Abstract 664), but the LCP/LCP and Ti/LCP adhesion strengths decreased by only 8.1% and 11.5%, respectively. 

Figure 5.

 

Blister test results. (a) Initial adhesion strengths without soaking. (b) Adhesion of LCP/LCP and Ti/LCP interfaces under 37°C PBS soak test. (c) Adhesion of LCP/LCP and PI/PI interfaces under 75°C PBS soak test. (d) Adhesion of Ti/LCP and Ti/PI interfaces under 75°C PBS soak test. The measurement limit (1.1 MPa) was the upper limit of applied pressure in the test apparatus. Error bars represent ±1 SE (n = 5). The data show that the LCP/LCP and Ti/LCP interfaces were stronger and more reliable than the PI/PI and Ti/PI interfaces during the 8-week soak test.

In Vivo Animal Experiments

The fabricated microelectrode arrays were successfully implanted in the suprachoroidal spaces of the rabbit eyes. During insertion, no guide tools were needed because the fabricated structure exhibited an adequate amount of flexibility. The stimulation sites were successfully located near the retina's visual streak, and the reference site was located at the outer wall of the sclera. 

Acute in vivo electrical stimulation experiments were performed to record the EECPs from the rabbit visual cortex. Cathodic-first biphasic current pulses of 0 to 100-μA amplitude, 1-ms duration, and 1-Hz period with 1-ms interphase delay were applied among the four stimulation sites and the reference site (Fig. 2b), and EECP waveforms were simultaneously recorded (Fig. 6a). The waveforms exhibited the typical characteristics of EECPs, which have discernible negative and positive waves following the stimulus artifact components. The threshold current amplitude was estimated at approximately 40 μA under four-channel simultaneous stimulation (40 μA × 4 channels simultaneously), and the threshold charge density was calculated as 20.4 μC/cm² (500-μm diameter). Because the stimulus artifact (Fig. 6a) component might have distorted or reduced the amplitude of the negative wave (Fig. 6a, N1), the implicit time of the first negative peak was estimated at less than 16 ms. The first positive peak (Fig. 6a, P1) was clearly observed, and its implicit time was 26 ms. This relatively slow wave was similar to those observed in previous suprachoroidal stimulations ^17,22 and clearly different from that resulting from a stimulus artifact. In addition, the first positive peak (P1) amplitude had a nearly linear relationship with the stimulation amplitude (Fig. 6b). 

Figure 6.

 

EECP recording. (a) Representative EECP waveforms measured in visual cortex of a rabbit. (b) Relationship between stimulation intensity and first positive peak amplitude. The first positive peak (P1) was clearly observed, and its implicit time was 26 ms. P1 had a nearly linear relationship with the stimulation amplitude.

The implanted electrode arrays were monitored by fundus observation and OCT for 3 months. Subsequently, after 4 months, histologic examinations were performed, and FE-SEM images were taken to evaluate the array condition. The representative OCT images in Figures 7a and b showed that the retina structures containing the LCP-based retinal electrode arrays were well preserved during the postoperative 3-month period without observation of any chorioretinal inflammation or structural deformities. Moreover, the fundus images in Figures 7c and d showed that the implanted arrays had not migrated, induced haziness, or resulted in vitreous inflammation. The histologic examinations (Fig. 7e) revealed no evidence of retinal neural cell loss or inflammation around the space where the arrays were implanted after 4 months. The FE-SEM image of the explanted array (Fig. 7f) showed no sign of degradation such as delamination of sites or site windows. 

Figure 7.

 

Suprachoroidally implanted microelectrode array in the rabbit eye. (a) OCT image, 2 weeks after operation. (b) OCT image, 12 weeks after operation. (c) Fundus image, immediately after operation. (d) Fundus image, 7 weeks after operation. (e) Histology of the retina, 16 weeks after operation. Asterisk: site of microelectrode array implantation. (f) Field emission scanning electron microscope image of the microelectrode array explanted after 16 weeks. The retina structures with LCP-based microelectrode array were well preserved at the end of the 4-month period. No migration or deformation of the implanted array was found.

Discussion

Characterization of the Fabrication Process

The LCP fabrication process is different from existing polymer fabrication processes for polyimide and parylene. First, LCP is a thermoplastic polymer that is supplied as a thin, film-type product. ¹⁴ Therefore, no spin-coating and curing processes, which are generally used in thermosetting polymer fabrication, are needed to fabricate the substrate and the insulation layer. Moreover, LCP films can be thermally bonded to other LCP films without adhesives ^10,12,13,15 ; accordingly, seamless, monolithic encapsulation of microelectrode arrays is available. 

Although LCPs are a physically stable and chemically inert materials, they have disadvantages in their compatibility with conventional photolithography alignment and plasma dry-etching methods. Conventionally, alignment is performed using metal-patterned alignment marks on the substrate. However, such marks cannot be observed through an LCP film because of its opacity; thus, a conventional alignment process is not suitable. Moreover, plasma dry etching of LCP results in a slow etching rate and in irregular surface morphologies; therefore, additional time is required to create smooth site windows and electrode array outlines. To overcome these difficulties, modifications to conventional fabrication procedures were needed. 

In our work, laser micromachining was fully exploited to improve fabrication productivity. Laser-drilled alignment marks were useful for precise alignment, and laser machining produced a fast etching rate with high flexibility. Although laser machining is a serial process often disadvantageous to batch fabrication, it is suitable for simplified fabrication of LCP material. 

Long-term Reliability Test Methodology

A potential source of failure of polymer-based electrode arrays is the high electrical leakage between channels attributed to water absorption and unstable interfaces. Such failure can occur when the array structure experiences environments with high humidity. ^5–7 Because of moisture and ion influences, the adhesion strength of the electrode array's interlayer can decrease, allowing electrical leakage through the damaged interface. Once leakage paths are created, interelectrode cross-talk increases significantly, and eventually the electrode arrays lose stimulation or recording selectivity. 

In this study, for detection and analysis of such potential failures within a shortened period, we used accelerated soak tests. At temperatures higher than 37°C, the structure degradation process caused by moisture and ion influences may be accelerated. ^5–7 During our 75°C accelerated soak tests, we monitored electrode site impedance and interlayer adhesion strength to evaluate site and interlayer reliabilities. However, those measures could not provide direct information about the electrical reliability such as cross-channel leakage. Therefore, we are performing experiments to measure electrical current leakage through LCP interfaces using customized multi-interdigitated electrodes. Preliminary results have shown minute (approximately 2.1–38 pA) interface leakage currents during a 1-week experiment at 75°C under 5-V DC bias voltage (data not shown). These experimental results support the site impedance and interlayer strength reported here. 

Although accelerated soak tests are convenient for fast analysis of possible failure mechanisms, some potential pitfalls should be considered. First, the test temperature has to be carefully selected to avoid material transition or decomposition that may not occur under normal temperature conditions. In addition, unknown repair or stabilization processes can occur under accelerated conditions. ^6,7 For these reasons, we performed the soak tests at two temperatures, 37°C and 75°C. 

Conclusions and Future Work

In this study, we fabricated and tested a prototype LCP-based Au microelectrode array. Although Au was used, other site materials, such as Pt and IrOx, could be applied to our fabrication process using well-established sputtering methods. Similar studies into their long-term reliabilities with LCP substrates will be reported in the future. 

In vitro accelerated reliability tests showed that such LCP-based microelectrode arrays have excellent stabilities in a high-humidity environment. Furthermore, even under high-temperature (75°C) PBS soak tests, Au site conditions and interlayer adhesion strengths of the electrode arrays showed no degradation for periods of 9 weeks and 8 weeks, respectively. These results can be explained by the very low moisture absorption (<0.04%) and thermal bondable interface characteristics of LCPs. 

The feasibility and long-term biocompatibility of LCP-based microelectrode arrays were examined by in vivo animal experiments, including EECP recording, OCT imaging, and histologic examination. Typical EECP waveforms were recorded, and OCT images and histology after 3 months of implantation showed good biocompatibility. 

Footnotes

 Supported by the Korea Science and Engineering Foundation (KOSEF) through the Nano Bioelectronics and Systems Research Center (NBS-ERC) at Seoul National University and by a grant from the Korea Health 21 R&D Project (A050251), Ministry of Health and Welfare, Republic of Korea.

Footnotes

 Disclosure: S.W. Lee, None; J.-M. Seo, None; S. Ha, None; E.T. Kim, None; H. Chung, None; S.J. Kim, None

Footnotes

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.