Electrode array design and fabrication for implantable systems
Many medical implants use electrodes to stimulate nerves or muscles, or sense their activity. As more advanced implants such as functional electrical stimulators, cochlear implants and retinal prostheses are developed, requirements for implantable electrodes have become more demanding. Whereas implants such as pacemakers can be made using simple, large, noble-metal electrodes, newer systems such as retinal implants require a complex array of electrodes within a very small enclosure. This necessitates micro-fabrication techniques for both electrodes and substrates with the final assembly needing to retain its biocompatibility. The manufacturing techniques for these new types of arrays push the boundaries of current technologies, both in the formation of microelectrodes and interconnects. This chapter describes processing methods for different types of electrode arrays and discusses their limitations.
Implantable electrodes are used in a variety of medical applications, some of which have been described in this book. For example, retinal implants, cochlear implants and functional electrical stimulation systems all use electrodes to stimulate nerves or muscles (actuators) (Loizou, 1998; Spelman, 2006; Hodgins et al, 2008). Other implants use electrodes to record nerve signals for diagnostic purposes, for example, neural implants (Qin et al., 2009) or as the feedback loop for a stimulator implant (recording electrodes) (Cogan, 2008). For each of these applications, the requirements for the electrodes vary significantly. At one end of the spectrum, when stimulating or recording from muscle, the electrode may be a large single device. At the other end of the spectrum, for activating or recording from a group of nerves, an electrode array of microscopic proportions that can be inserted into a particular location is required.
In retinal and cochlear applications, the electrode array is replacing a large number of natural stimulators that have failed; photoreceptors in the retinal implant and sensory hairs in the cochlear implant. With technology where it is today, we fall a long way short of being able to restore full functionality for either ear or eye, and this limitation is at least partly due to our inability to make very small electrode arrays with a large number of individually addressable electrodes. For example, the eye has 130 million photoreceptors positioned over the retina and the ear has thousands of hair cells in the cochlea, whereas current technology limits electrode arrays to between tens and hundreds of contacts.
For recording applications, the electrode is detecting signals produced by nerves. These can be electrodes providing the feedback system for a bladder stimulator (Donaldson et al., 2008), or they may be more complex arrays detecting signals, for example those generated in the brain, where size, form and rigidity are critical aspects of the design (Kawano, 2004).
Stimulation electrodes can either be in direct contact with the nerves or close by. The former approach requires less power than the latter, but in many cases may be impossible to achieve in practice. Examples of indirect stimulation are the retinal and cochlear implants. The retinal implant (described in more detail in Chapter 15 of this book) stimulates the retinal layer that transmits electrical pulses to the ganglion cells, which feed the optical path (Feucht et al., 2005). Current technology for the cochlear implant uses an electrode array inserted into the scala tympani of the cochlea to stimulate the neurotransmitters, which then send the electric pulses into the auditory nerve bundle (Deman et al., 2003). Systems where the electrode is in direct contact with the nerve include the Finetech-Brindley™ bladder control system and STIMuSTEP™, a system for Dropped Foot, both produced by Finetech Medical (Finetech Medical). Investigations are also underway to provide direct stimulation into the auditory nerve bundle using an intra-modiolus electrode array by CTC (Volckaerts, 2007). Recording electrodes need to be in direct contact with the nerve as the signal strength is very low and, hence, the signal-to-noise ratio is an issue, an example of which is given in another chapter of this book.
Implantable electrodes have been used for many years. However, with the development of microsystem technology, electrode arrays that can be implanted can now be realised. These arrays are far more complex to fabricate than a single electrode and are, hence, the focus of this chapter.
This chapter describes different methods for producing a complete electrode array, including electrode contacts, substrate and coating. A substrate-based approach has been chosen, because the most challenging implantable electrode arrays today require a large number of electrodes in a very small space. Macroscopic electrodes, formed from sheet or wire, cannot be made on the microscopic scale, whereas metals deposited onto a substrate can produce electrode arrays in the micron scale and at an acceptable manufacturing cost. The fabrication methods presented focus on platinum for the electrode, as it has good electrical properties, can be processed using a variety of methods and offers excellent biocompatibility properties. Substrates serve as a base onto which the electrode material is deposited. The substrate can subsequently be removed, leaving a freestanding electrode material, or can remain and provide both the mechanical support and location for attaching wires where appropriate. Silicone, silicon and polymer foil substrate materials are investigated, and different coating materials are described.
The chapter starts by describing general requirements for implantable electrode arrays, materials that are suitable for electrodes and substrates, and fabrication processes for electrodes, substrates and coating materials. The chapter then goes on to describe the design and fabrication of a complete electrode array using the materials and processes described and, finally, presents drawbacks and improvements as well as the risks involved in the manufacturing of electrode arrays.
In all applications, electrodes need to be biocompatible, and in many applications, this means over a lifetime. Electrode arrays need to be located so that they cannot move throughout their lifetime and do not break during initial insertion. Furthermore, electrodes should discourage an excessive immune response, as this can cause very high impedance at the electrode surface as well as physiological problems such as inflammation. In addition, a stimulating electrode array must also be able to provide the necessary stimulation current to exceed the threshold on the specific nerve(s) and provide sufficient differentiation for the given application. For example, with the retinal implant, there needs to be an adequate number of electrodes to be able to form an image, whilst the cochlear implant requires a sufficient number of electrodes to differentiate sounds. In specific cases where the stimulation current has to be relatively high while the electrode size is very small (micrometre range), the shape of the electrode has to be optimised to prevent damage of the tissue due to excessive current density. For recording electrodes, they need to provide sufficient signal, such that the signal-to-noise ratio after electronic amplification is adequate to detect and record nerve activity.
There are three noble metals often used for implantable electrodes because of their excellent corrosion resistance and biocompatibility properties (Geddes and Roeder, 2003): platinum (Pt), gold (Au), and iridium (Ir). For most long-term implant applications, where the electrodes are used to either record or stimulate, the preferred electrode material is platinum. This metal can be processed using various micro-fabrication techniques to form electrodes, has a relatively low resistivity, can be joined to wires using conventional joining technologies, can be deposited onto a variety of surfaces and is biocompatible. It also suffers only slight corrosion when used for stimulating or recording purposes. Gold is suitable for use as a recording electrode but not as a stimulating electrode as it suffers from corrosion (Myllymaa et al, 2009). Iridium oxide is also considered an excellent material for stimulating electrode arrays (Slavcheva et al, 2004) as it can be sputtered onto a substrate, has low impedance, and has good long-term mechanical stability and corrosion resistance. There are other materials being investigated for stimulation applications, including titanium nitride (TiN), but this was found to have higher impedance and lower charge storage that iridium oxide (Weiland et al, 2002).
The substrate onto which the electrode material is deposited must be suitable for micro-fabrication, as many implantable systems have size constraints. If the substrate is going to remain as part of the final electrode array structure then there are further, often conflicting requirements that must be met: compliant when in the body, but rigid during insertion. It may either be made of a biocompatible material, or if not, then it must be coated with one. However, it must be stressed that even a biocompatible material can be affected by the processing method, so care must be taken to ensure that these processes do not render the material non-biocompatible. One material that is frequently used is silicone, as it is compliant, biocompatible and can be processed using a range of micro-fabrication techniques. The processing of the substrate material to provide a suitable surface for the electrode material is crucial, and techniques are still being developed. The precise patterning of biocompatible silicone rubber is a key issue for the manufacture of electrodes for active implantable medical devices (AIMD), and one approach is to use plasma etching.
Silicone plasma etching is a method by which the surface of the silicone substrate can be prepared for the deposition of platinum electrodes at the micron scale. There are a number of different medical-grade silicones available that can be processed using this method, including polysiloxane (Nusil MED-6215). If silicone plasma etching is chosen, then there are a number of parameters that can be varied, which affect etching time and surface finish. Both of these are important considerations for the manufacturing of electrode arrays for medical implants.
In this section, some information on the etching of polysiloxane under various high-density plasma conditions is provided. It is dependent upon the temperature of the substrate holder above 40 °C and etching of polysiloxane in high-density plasma source (Surface Technology Systems Multiplex equipped with Inductively Coupled Plasma – ICP) is faster (ca. 70% at 50 °C) than in conventional Reactive Ion Etching (RIE) plasma (Szmigiel et al, 2006b, 2008b). Additionally, using an ICP tool the etched profiles have improved anisotropy (Szmigiel et al., 2008a ).
Post-etch morphology (Szmigiel et al., 2008a) is also an issue as the electrode material has to be deposited onto the silicone surface. Under RIE plasma conditions, the specific grass-like post-etch residue appears on the surface (see Fig. 5.1). X-ray Photoelectron Spectroscopy (XPS) analysis has shown a high concentration of carbon and fluorine on plasma-etched silicone surfaces, which indicates chemical bonding between the carbon and fluorine. The grass-like residue is hardly removable, and it is present on the surface even if the substrate surface is visibly attacked by plasma. That is why the ability to control the morphology of the plasma-treated polysiloxane is a key factor in obtaining an entirely clean surface of metal electrodes after the removal of silicone layer in plasma.
5.1 Polymer post-etch residues visible on the surface of platinum substrate: (a) Scanning Electron Microscope (SEM) graph showing the surface of the Pt electrode after the polysiloxane removal in SF6 + O2 plasma; (b) image of polymer post-etch residues taken with an optical microscope. (Source: Szmigiel et al., 2008a.)
The post-etch residues are not present on non-masked wafers. It has also been found that the polysiloxane morphology becomes smooth if a processed wafer is hotter (see Fig. 5.2). Therefore, the following factors tend to decrease or even eliminate post-etch residues:
5.2 SEM graphs showing the polymer morphology after SF6 plasma exposure at 5 m Torr pressure, 800 W power, 100 W bias power and substrate holder temperature of 20 °C (a) or 40 °C (b). (Source: Szmigiel et al., 2008a.)
• increasing the substrate holder temperature (up to ~ 50 °C or higher) in case of polysiloxane dry-etching using an aluminium mask. (It is very important to provide good temperature uniformity of a processed wafer.) (Szmigiel et al., 2008a ).
Laser ablation has been investigated as a method to remove post-etch residue and open contact pads with some success. Irradiation using laser light at 248 nm showed poor selectivity for silicone compared with platinum, so the pads were damaged in the cleaning process. However, at 193 nm the threshold for ablation of Pt was higher, so it was possible to remove silicone with a greater degree of success.
Concerns about biocompatibility of plasma-etched silicone layers have arisen from observations that plasma treatment of polysiloxane might cause changes in the specific surface composition and in the morphology of this material. For medical implants, it is important to understand whether this could render the material non-biocompatible. For this reason, samples have been produced and the surfaces have been analysed using a range of tests.
Samples of silicone elastomer (MED-6215) from Nusil Silicone Technology were investigated by XPS analysis, wetting test, crystal violet test (cytotoxicity), and MTT(3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) tests (Szmigiel et al., 2006a). All samples were prepared by mixing two liquid components of MED-6215 in a 10:1 ratio. Silicone layers of thickness of 20 μm were formed by spin coating, while thicker structures were cast into plastic moulds. Layers were dried, sputtered with aluminium (Al) and etched through the Al mask using SF6 or CF4 plasma chemistry. Directly after the plasma treatment, the samples were immersed in the mixture of acids (H3PO4, CH3COOH, HNO3 and H2O) to remove the Al mask.
XPS analysis indicated compositional changes in the silicone surface etched with the CF4 + O2 and the SF6 + O2 plasma process. The fresh material (MED) contained silicon (Si), oxygen (O) and carbon (C). The analysis performed for the specimen etched in CF4 + O2 plasma (MED-CF4) revealed the presence of fluorine (F), nitrogen (N) and aluminium (Al), while the analysis of the specimen etched in SF6 + O2 plasma (MED-SF6) showed the presence of fluorine (F) only. The results of a quantitative chemical analysis by X-ray photoelectron are compared in Table 5.1. The initial composition of the elastomer Si:O:C of 1.0:1.0:1.7 changed into 1:1.1:2.0 and 1.0:0.9:1.4 for the specimens etched in CF4 + O2 and SF6 + O2, respectively. The silicon surface content of 27.24% in the fresh sample decreased to 22.54% after CF4 + O2 treatment and increased to 30.02% in the sample exposed to SF6 + O2 plasma. The investigated specimens did not differ significantly in surface oxygen content. Compared to the fresh material MED (27.14%) the surface contribution of oxygen was less in MED-CF4 (24.43%) and in MED-SF6 (26.0%). The surface carbon content in the MED sample had changed slightly from the value of 45.62% to 43.77% and 42.07% after etching in CF4 + O2 and SF6 + O2 plasma, respectively. Each element produces a characteristic set of XPS peaks, which corresponds to electron configuration within the atoms. In order to indicate the electron configuration in atoms a special notation consists of a sequence of atomic orbital labels is used: 1 s, 2 s, 2p, etc.
Source: Szmigiel et al. (2006a).
The fluorine contribution to the chemical composition of the surface was about three times higher for the MED-CF4 specimen than for MED-SF6. The XPS studies have indicated the presence of traces (< 0.5%) of nitrogen on the MED-CF4 surface. Additionally, about 3% of aluminium has been detected on the surface of this sample. The surface of the specimen etched in SF6 plasma (MED-SF6) has been aluminium free.
The contact angles measured in the wetting tests have shown that the raw polysiloxane surface has highly homogenous and hydrophobic characteristics (water contact angles above 100°). Both fluorine-based plasma treatments have made the polymer surface less uniform and the initial contact angle has decreased. The contact angle values after the etching were scattered significantly (see Fig. 5.3). Furthermore, an asymmetric drop shape has been observed in the case of both plasma-exposed surfaces (see Fig. 5.4). Nevertheless, the polysiloxane surface has preserved its hydrophobic properties after the plasma processing.
5.3 The contact angle (CA) values on a fresh (MED) and plasma treated (MED-SF6 and MED-CF 4) polysiloxane surface. (Source: Szmigiel et al., 2006a.)
5.4 Drop shapes observed on the fresh polysiloxane surface (MED), after SF6 + O2 (MED-SF6) and CF4 + O2 (MED-CF4) plasma treatment. (Source: Szmigiel et al., 2008b.)
Silicone samples prepared for cytotoxicity tests were steam-sterilised and placed in 24-well culture dishes. A suspension of mouse L929 fibroblasts in RPMI 1640 medium containing 10% foetal bovine serum and antibiotics was seeded on each sample and on control glass. The cells were cultured at 37 °C in an incubator with humidified 5% CO2 atmosphere. The adhesion of cells was measured by crystal violet test (CV) while the viability of cells was estimated by MTT assay method. The MTT (Methyl Thiazol Tetrazolium Bromide) test is a colorimetric assay that is indicative of enzymatic activity in living cells. Hence, the degree to which the (yellow) MTT is converted to (purple) formazan is a measure of cell viability. The results have shown that the adhesion of cells to studied materials has been considerably lower than to control glass. Fluorine-based plasma has significantly improved the adhesion of cells in comparison to raw polysiloxane material (see Fig. 5.5; RAW, SFO and CFO samples). Adhesion of cells was stronger for the sample not exposed to plasma but sputtered with an aluminium layer, which was removed by wet etching (see Fig. 5.5; ALU sample). Viability of cells on raw and plasma-treated polysiloxane was high and comparable to control (see Fig. 5.6; RAW, SFO, CFO and control samples). Comparable viability has been found after aluminium mask removal from polymer surface (see Fig. 5.6; ALU sample – it should be noted that such areas are the largest one in case of a real electrode structure).
5.5 Adhesion by crystal violet of L929 fibroblasts grown on control glass, polysiloxane prior to (RAW), after SF6 + O2 (SFO) and CF4 + O2 (CFO) plasma treatment, after aluminium removal (ALU); OD – optical density. (Source: Szmigiel et al., 2006a.)
5.6 Viability by MTT test of L929 fibroblasts grown on control glass, polysiloxane prior to (RAW), after SF6 + O2 (SFO) and CF4 + O2 (CFO) plasma treatment, after aluminium removal (ALU); OD, optical density. (Source: Szmigiel et al., 2006a.)
The results of the cytotoxicity tests indicate that fluorine-based plasma processing of originally biocompatible silicone is not likely to cause long-term biocompatibility problems. This is an important finding and indicates that this form of biocompatible silicone can undergo plasma etching prior to the deposition of the electrode material and remain biocompatible.
Applying a coating layer onto implantable microelectrodes is essential for ensuring reliability, and it is one of the most important components of the electrode structure. The most obvious function of the coating is for protection of the fragile device against mechanical damage during surgical insertion. In many applications, the coating also provides a hermetic barrier against possible leakage of contamination from the electrode to the human body. The coating also protects electronic circuits against humidity (e.g. body fluids) and corrosion.
It should be mentioned that a perfect hermetic coating is required only for electrode arrays where the electrodes are deposited onto a non-biocompatible material such as silicon or aluminium that could contact the human body. Perfect hermeticity is much less critical for devices that consist only of biocompatible metals and polymers. Silicone rubber, polyimide and Parylene C have been selected as basic coating materials due to their biocompatibility and sufficient chemical stability (Lago et al, 2007; Hassler et al, 2010; Kumar et al, 2010).
• multi-layer coatings for implantable electrodes that need to provide a hermetic seal for electronics: silicone and Parylene C (e.g. a thin parylene layer deposited on the electronic circuit as an internal hermetic barrier and a much thicker silicone layer as an external coating ensuring mechanical protection of the device and acting as ion barrier),
Silicone coatings provide mechanical and electrical isolation and are an effective ion barrier but are highly permeable to water. Therefore, when using silicone on its own, it is important to consider its water permeability.
The hermeticity of any coating can be verified by electrical measurement of capacitive humidity detectors used in accelerated ageing tests, composed of a set of inter-digit electrodes. Tests have been performed using two types of devices designed and manufactured for this purpose: relatively big humidity detectors fabricated on flat rectangular chips (see Fig. 5.7) and small humidity detectors fabricated on three-dimensional (3D) profiled electrodes (see Fig. 5.8).
5.7 Flat silicon chips with capacitive humidity detectors used in accelerated ageing tests to evaluate the hermeticity of Parylene C-silicone coatings. (The higher magnification view shows part of the gold comb electrodes of the capacitor.)
It was found that structures coated with Parylene C with additional edge protection or with a double Parylene C-silicone layer had a longer time to failure than structures with a single Parylene C coating. This means that the problem of bad hermeticity results from poor adhesion between the Parylene C coating and the humidity detector and not the Parylene C layer.
In conclusion, there are a number of different biocompatible coating materials that can be used for electrodes and the tracks that carry the signal to the electronics. The choice is dependent upon the time that the electrode will be used and the substrate. For short-term use, a single-layer coating can be used. For long-term use, a double Parylene C-silicone coating has been shown to be better than a single coating of either material.
Platinum is one of the most commonly used materials for implantable electrodes and there are a number of processing techniques that can be applied for dense, microfabricated arrays. In the following sections some of these techniques have been described.
The deposition of the platinum electrode is important in the overall performance of the electrode array. The material has to be deposited to the required thickness and provide a good bond to the substrate. There are two techniques commonly used to deposit thick film platinum: electrodeposition (or electroplating) and sputtering. The former is suitable for producing thick films, whereas the latter is only suitable for thin films. In the case of implantable electrode arrays, there is a need to minimise the impedance of the electrodes, so thick films are preferred. The following analysis covers the electrodeposition of thick platinum films, which could be applied to electrodes and elastic wire lead-outs, where the basic requirements for the platinum electroforming process apply:
Table 5.2 compares basic process parameters and technology issues for the different plating solutions.
Two plating solutions (high pH and low pH) can be used to deposit layers thicker than 8 μm, but cracking or delamination of the plated tracks thicker than 4 μm may occur. Bright, grey, porous Pt strips of 5 μm thickness can be deposited with a constant current density or 8 μm thickness using a unipolar pulse-plating technique. In both cases, a current density in the range 1.5–2.5 mA/cm2 gives the thickest layer. Pulse-plating does not offer any improvement over a constant current in the mechanical properties of thick Pt structures.
Suppliers of plating solutions generally recommend the most suitable process conditions. For the high pH solution, 20 μm thick, crack-free, light grey platinum layers have been produced, but these could easily be delaminated after the process, making them unsuitable for use. Deposition processes have been carried out on a 4-inch wafer with a pattern of 25 μm or 50 μm wide strips (see Fig. 5.9). The maximum thickness of the crack-free Pt strips that did not delaminate was 4 μm.
The elastic modulus and hardness of the Pt layers either just as deposited or after annealing can be investigated by the nano-indentation method using a Hysitron Ubi I Indenter system (Oliver and Pharr, 1992, 2004). The results shown in Table 5.3 relate to the examples described. The best mechanical parameters have been achieved for Pt layer deposited in high pH solution and annealed at 350 °C. It should be noted that it is only for these samples that the extracted value of effective elastic modulus is comparable to that of bulk or sputtered platinum. The experiment on annealing of samples with electroformed Pt elements has shown significant improvement in the mechanical properties of Pt structures after heating the structure to 350–400 °C.
The resistivity of Pt deposited in a low pH solution (22 × 10–8 Ωm) and deposited in a high pH solution (28 × 10–8 Ωm) was over two times higher than the resistivity of bulk Pt (10.6 × 10–8 Ωm). An increase in resistivity compared to bulk material is expected for any deposited material, since it can only ever be as dense as bulk. Typically, it will be a lower density film with more sub-structure, leading to increased resistance through the inter-grain contacts.
The most important requirements in the selection of a Pt electroplating process for microelectrodes are good mechanical properties and low resistance of Pt paths. PLATINART 100 seems to be the best candidate for deposition of Pt layer of good mechanical properties. However, there are some limitations of this solution:
In summary, whilst electrodeposition of Pt electrodes is the most attractive in terms of providing a thick layer and, hence, low impedance, the current processes do not produce a sufficiently robust electrode. Specifically, the Pt layer has a high resistivity when compared to bulk Pt and high stresses can cause delamination. An alternative strategy is to sputter or electrodeposit a thinner (2 μm) layer and reduce the overall impedance by increasing the width where possible and minimising the electrode length.
Having ascertained that fabricating a Pt electrode thicker than 2 μm is not feasible with currently available methods, a number of schemes for 2 μm thick electrodes with wide tracks were produced on a silicon wafer. All of these were designed to meet the resistivity and size requirements for a stimulation electrode not in direct contact with a nerve.
A range of electrode designs, which would ultimately have a double Parylene C-silicone coating were fabricated on a 4-inch silicon wafer: short electrodes and electrodes integrated with an elastic lead-out and equipped with connectors of different of shapes, as shown in Fig. 5.10.
5.10 ( a) A silicon wafer with manufactured microelectrodes and (b) an example of microelectrode structure mechanically released from the wafer. (Source: Szmigiel et al., 2008b.)
The resistances of wires in the short passive electrodes and in electrodes integrated with the elastic lead-out were in the range of 60–140 Ω and 120–220 Ω, respectively. The resistance essentially exceeded 100 Ω required by some implant manufacturers. This was due to the significant lateral over-etching of Pt paths in the wet etching process (Fig. 5.11a). An alternative is to use a Pt dry-etching process with a very thick silicon dioxide (SiO2) layer (4 μm) as a hard mask. The Pt pattern has been etched in a pure argon (Ar) sputtering process at room temperature. Sufficient selectivity to the mask has been achieved to etch through the whole Pt layer and there is minimal narrowing of the Pt tracks in the etching process (Fig. 5.11b). In this case, the resistances of electrode wires measured on the wafers from the third run have been kept below 100 Ω.
This analysis of the fabrication of different devices highlights the importance of each process stage in the fabrication of electrodes. The wet etching process produced significant over-etching of the Pt electrodes, which resulted in unacceptably high electrode impedance. The over-etching of the plasma-etched devices was considerably less resulting in electrode impedance below 100 Ω.
This section describes one method of producing an array using sputtered Pt as the electrode material and no final substrate. The array was designed for stimulation within the cochlea, although the design approach could also be used for recording purposes. The final design is fully encapsulated with a double layer of Parylene C and silicone, except for where the electrodes are exposed and the connection points. This processing method meant that the structure had to be transferred from one substrate to another during the fabrication process, and ultimately both substrates were removed.
First, a 2 μm thick layer of sputtered Pt was deposited onto a temporary 4-inch silicon wafer. This was used as mechanical support for photolithography-based patterning of platinum, silicone and Parylene C layers. The whole fabrication sequence uses standard technological processes commonly used in silicon technology, complemented by unique processes of coating, patterning and releasing of thick layers of biocompatible silicone rubber where necessary.
The final technological sequence of the manufacturing process is shown schematically in Fig. 5.12. A temporary substrate was oxidised (step 1) to create an oxide etch-stop layer for later process of Si etching. A Pt layer was then sputtered and patterned using a wet or plasma etching process (steps 2 and 3). The silicon dioxide (SiO2) and silicon nitride (Si3N4) layers deposited in Plasma Enhanced Chemical Vapour Deposition (PECVD) process served as mask material for the both type of processes for Pt etching. Next, a 5 μm thick layer of Parylene C was deposited on the wafer and then etched in plasma process using an oxide mask (steps 4 and 5). Reducing the size of the area of the Parylene C layer by etching of the unwanted part of this layer minimises the risk of future delamination of Parylene C from the wafer. Two 20 μm thick layers of biocompatible silicone rubber (MED-6215 from Nusil Technologies) were coated on the wafer and then cured (step 6). The temporary substrate was bonded using the third layer of silicone to another substrate covered with a sacrificial Al layer (steps 7 and 8). The first temporary substrate was removed in two steps using long etching of Si by RIE process (step 9) and final short plasma etching of the SiO2 etch-stop layer (step 10). The next layer of 5 μm thick Parylene C was then deposited on the wafer (step 11) and etched in plasma process to define the outer shape of each structure and openings to the Pt electrodes and pads (step 12). The top 60 μm thick layer of silicone has been formed by triple coating and curing of 20 μm thick silicone layers (step 13). The final shape of the device and openings to the pads have been defined by the final photolithography step and plasma etching of a 120 μm thick silicone layer (step 14) using the Al mask, which is removed after the process by wet etching. It should be noted that in the several experiments performed with this process it has been split into two stages separated by an additional photolithography step. The outer shape of the device has been etched before the openings to the pads, but it has been found that such a solution causes some difficulties in the photolithography process (especially for coating of photoresist on 120 μm deep trenches around the structures and for exposure of fine pattern close these trenches). The openings to the electrodes have not been etched, because the released structures should be finally over-moulded with an additional silicone layer, which would cover all surfaces. After rinsing in deionised water for three hours, the structures were ready for the release process based on mechanically detaching the structures from the wafer (step 15).
5.12 Sequence of fabrication of cochlear microelectrode: 1 – cleaning and oxidation of Si wafer. 2 – sputtering of Pt layer and deposition of mask layer on temporary substrate. 3 – photolithography of metal (wet or plasma etching of Pt layer). 4 – deposition of Parylene C. 5 – photolithography of Parylene C layer. 6 – multiple coating of silicone rubber layers. 7 – deposition of sacrificial layer on final substrate. 8 – wafer bonding. 9 – etching of temporary substrate. 10 – removing of etch-stop layer. 11 – deposition of Parylene C. 12 – photolithography of Parylene C layer (outer shape and openings to pads and electrodes). 13 – multiple coating of silicone rubber layers. 14 – photolithography of silicone layer (outer shape of microelectrodes and openings to pads). 15 – removing of Al mask, rinsing and releasing of structures.
A simple modification of the technological sequence allows for the manufacturing of devices with a textured surface over the stimulation electrodes if required. The modification of the technology has consisted of an additional photolithography process and plasma etching of silicon creating a set of shallow grooves on the surface of temporary substrate (step 1, Fig. 5.12). The Pt layer sputtered on the temporary wafer follows the shape of grooves etched in the wafer (see also Fig. 5.13) and, in that way, the surface of the stimulation electrode is textured in the device fabricated according to the modified technological sequence.
The design of a microelectrode array will be determined by the space envelope, the density and the amount of power required. For example, the array may be flexible or rigid and could incorporate two-dimensional (2D) or 3D electrodes. The following sections describe designs for different implantable applications.
In some medical applications, there is a need to stimulate or record data from a number of nerves, within a very small space envelope. One example for a stimulating array is the retina implant described in another chapter of this book. The requirements for such electrodes are extremely challenging from a manufacturing aspect for the following reasons:
A microelectrode array device with a thin and flexible polymer foil with integrated circuit paths, metal electrodes and bonding pads enables electrical stimulation using a dense matrix of Pt electrodes. The stimulation electrodes may be 3D shaped to increase their surface area and to enable more power to be applied to the nerves. The microelectrode design can also serve as a carrier for microelectronic components such as an ASIC or an antenna. These additional components can receive data and energy from external components.
The design described here is based on the formation of a multi-layer structure consisting of polyimide layers and metal circuits on a 4-inch silicon wafer coated with a sacrificial layer used for releasing of the devices. The design of the electrode matrix has 16 electrically connected stimulation sites in each square (see Fig. 5.14).
5.14 The design of the photolithographic masks and the layout of the matrix of the stimulation electrodes (dotted areas – cavities in silicon; light grey – polyimide; dark grey – metal 1 (platinum); black – metal 2 (gold); hatched areas – contact via).
The electrode matrix consists of 108 or 252 electrodes (16 stimulation sites per electrode leads to 1728 and 4032 stimulation sites, respectively) connected to bonding pads with 10 μm wide metal paths. Since the distance between electrodes is small, the circuit paths have to be routed both between and over the electrodes. It has to be noted that circuit paths crossing a large number of relatively deep cavities (etched in the substrate to form the shape of electrode) make the photolithography process difficult. Another photolithography problem has been anticipated due to arrangement of each electrode to form the separate stimulation sites located close to each other, which required patterning of the polyimide layer close to edges of cavities in silicone. An example of a device with 252 stimulation electrodes is shown in Fig. 5.15.
Stimulation electrodes with alternative profiles, planar and 3D shaped with sharp and rounded edges, can be produced using similar fabrication techniques. Recording electrodes would generally only be planar as they are not required to put energy into the nerve and, hence, do not require a high contact surface area. Examples of the shapes of electrodes are shown in Fig. 5.16 and these can be manufactured as high-density arrays (Fig. 5.16d).
The previous section described an array with a large number of electrodes. A more practical option for fabrication and integration into a complete implant system has fewer electrodes and interconnects. The example in this section describes an array comprising 231 Pt electrodes formed on a polyimide base, where each electrode is a 3D structure and consists of a 3 × 3 array of sub electrodes. Two gold conducting layers are used to connect to the 231 electrodes and the pitch of the circuit paths is 10 μm lines and 10 μm spaces. The connections for counter electrodes, chip pads and photodiode pads can be either on the first or the second conducting layer and will use gold bond pads. Tracing of 231 connection paths over 3.2 mm wide polyimide cable has been achieved by using two levels of metals for the connection.
The fabrication sequence (shown in Fig. 5.17) started with oxidation and then deposition of a Si3N4 layer on a silicon wafer with < 100 > crystal orientation. The first photolithography created openings in the nitride and oxide layers, and cavities for electrodes were etched into silicon in potassium hydroxide (KOH) followed by isotropic etching solution. After removing the oxide-nitride mask, the silicon surface was covered with thermally grown SiO2 and a 2 μm thick layer of deposited SiO2 highly doped with phosphorus (step 1). These layers served as a sacrificial layer for the final releasing process. Next, the first layer of polyimide (Pyralin PI 2611) was coated on the wafer and patterned by plasma etching (using double photolithography process – step 2). Then, 0.5 μm thick Pt was patterned using lift-off process (step 3). The next layer of metal (gold) was deposited and patterned by wet etching (step 4). In the next step, a second polyimide was coated and photolithography was used to make the contact vias (step 5). The third layer of metal (gold) was deposited and patterned by wet etching (step 6). After deposition of a thin titanium layer, photolithography and titanium etching, 5 μm thick gold structures were electrodeposited using the same photolithographic mask. Removal of the temporary titanium layer (step 7) and polyimide coating was followed by photolithography of the outer shape of the device (step 8). The bonding pads were opened in the last photolithography process (step 9). Finally, chemical etching in diluted HF solution was used to release the structures (step 10), which were rinsed for 24 hours.
5.17 Sequence of fabrication of Pt microelectrode array: 1 – photolithography of cavities, anisotropic etching of silicon and deposition of a sacrificial layer. 2 – polyimide coating and photolithography of openings for electrodes. 3 – deposition and photolithography of first layer of metal (Pt lift-off process). 4 – deposition and photolithography of second layer of metal (wet etching of gold). 5 – polyimide coating and photolithography of via contacts. 6 – deposition and photolithography of third layer of metal (wet etching of gold). 7 – deposition of Ti seed layer, photolithography, electrodeposition of fourth layer of metal and removal of Ti layer. 8 – polyimide coating and photolithography of outer shape of microelectrodes. 9 – photolithography of pads. 10 – structure release.
This array differs from the one described previously in that the bond pads are gold, which is deposited onto a titanium seed layer. This provides an easier method to attach bond wires than the platinum in the earlier design, although it must be remembered that gold suffers from corrosion.
For some implants, the electrode can be in direct contact with the nerve, and thus enable direct interaction between the electrodes and the nerves. This approach has the potential to reduce the power consumption of a stimulating implant, by decreasing stimulation thresholds and thus, reducing potential risk of the damage resulting from the relatively high stimulating current, which is required in the case where the distance between the electrode and the nerve is high. For recording electrodes, this approach would improve the signal-to-noise ratio of the system.
In the example provided in this section, the electrode array is designed to be rigid enough to be inserted and maintain its form to keep the electrodes in contact with the nerves. This is done by maintaining the substrate beneath the electrodes, which differentiates this design from the previous ones described.
The design is based on a double-sided bulk silicon micromachining process supplemented by formation of biocompatible coating. The main advantages of the silicon version of this microelectrode are: stiffness of the silicon beam sufficient for inserting the electrode without additional stylet and possible integration with active devices, like a deflection sensor for precise driving the electrode during the insertion or a multiplexer for addressing of a large number of electrodes. However, the manufacturing process is far more complicated than in the case of polyimide-platinum electrode and silicon, which introduces an additional risk regarding the biocompatibility of the device.
The design comprises 22 platinum electrodes arranged in a single row on a silicon beam. To facilitate the monitoring of a beam deflection during the surgical insertion of these electrodes, a piezo-resistive force/deflection sensor has been designed at the base of the cantilever beam.
The process flow enables the entire structure of the electrode (the top and bottom surfaces and all sidewalls) to be coated with a thin biocompatible layer before making the last photolithography process on the wafer, as illustrated in Fig. 5.18 :
A silicon beam with a thickness of approximately 20 μm should be stiff enough to pierce a soft-tissue membrane and flexible enough to survive a relatively large bend during insertion. These electrodes can be integrated with a piezoresistive deflection sensor and protected with single-layer coating (Silicone NUSIL MED-6215). This provides a stiffer electrode array than those previously described, where the platinum electrodes are coated in silicone.
Further development of the fabrication technology for Pt stimulation electrodes of the crucial technology processes or modifications to the technology sequence could improve the quality and performance of the final device. The up-to-date results of the most important technology processes, and the impact of current solutions on device performance and manufacturing costs as well as the comments on potential technology improvement are summarised in Table 5.4.
The Pt patterning step seems to have a major technology impact on the cost drivers for the product. Therefore it is recommended to concentrate on plasma etching of sputtered platinum layer and further improve the Pt electroforming process as this will reduce the manufacturing cost. Moreover, the electroplating process could be a solution for forming of thick textures on stimulation electrodes. The transferring of Pt paths between temporary and final substrates is essential for all methods of Pt patterning. It enables the formation of Pt wires before silicone and Parylene C coating and allows for high temperature annealing of Pt wires. Etching of the Pt layer in an aggressive plasma process and etching or electrodeposition of Pt in highly corrosive solutions is also feasible.
The major risks regarding the use of implantable electrodes in medical devices are related to the electrical and mechanical properties of the micro-electrodes, interconnects and biocompatibility issues. The manufacturing costs may also impinge on the successful implementation of a developed technology into a commercially viable product platform. One major concern for the silicon-based electrode arrays described in this chapter is biocompatibility. Any array that includes materials that are not biocompatible relies on the coating providing a hermetic seal, and it must be noted that a double Parylene C-silicone coating is not fully hermetic. Furthermore, there is a risk that processes needed to form the array can affect the biocompatibility of some materials.
Laser cleaning of the stimulation electrodes could be replaced by a plasma etching process, and prevention against coating during the over-moulding step. If mechanical detachment of the electrodes from the wafer is unsuccessful, the process of electrochemical etching of the sacrificial Al layer could be used for releasing these devices.
Joining electrodes to the electronic circuitry also involves risk. Micro-welding of Pt wires to the bond pads, which may only be 2 μm thick, can cause delamination or micro-fractures to the pad or the electrode.
The design and manufacturing procedures for the structures described in this chapter are examples of typical Pt based implantable electrodes. Several conclusions important for commercialisation have been outlined in the Table 5.4, many of which relate to the patterning of thick platinum and silicone layers. Another important issue is the manufacturing cost of the electrodes, especially integrated with the connections, since they are quite big compared to 4-inch wafers. Therefore, it is strongly recommended that larger, and possibly even rectangular substrates, are used to reduce the unused substrate area. Using thick electroplated Pt wires could be more beneficial from an economic point of view than patterning thick sputtered Pt layers; further development of platinum electroforming processes is required.
Silicon-based electrode arrays offer the potential for the future construction of active electrodes, which could be easily equipped with multiplexers for addressing a large number of stimulation or recording electrodes and with sensors facilitating precise manipulation of the electrode during its surgical insertion.
The hermeticity of a parylene-silicone coating is considered a high-risk area for long-term medical implants. A possible solution may be an additional layer (like titanium or Diamond-Like Carbon, DLC) beneath the Parylene C coating, which could significantly improve the adhesions of Parylene C to the surface of the structure (e.g. by roughening of this surface).
Large microelectrode arrays, consisting of 231 platinum stimulation electrodes, have been produced with both flat surfaces and 3D structures. However, the stability of 3D structures during stimulation at high current density has not been fully demonstrated to date and should be considered a high-risk area. The combination of gold and platinum also introduces complex processes and therefore one option may be to replace the gold metallisation with platinum, enabling the manufacture of multi-layer polyimide-platinum structures.
With the development of new implants for nerve stimulation or signal recording, the need for denser, smaller electrode arrays is increasing. The first multi-channel cochlear implant was produced in the 1980s and since then there has been a rapid increase in the number of channels required for advanced implants, such as the cochlear and retinal implants. This need for dense arrays, where the electrodes can each provide adequate charge to exceed the nerve threshold, is encouraging research and development in both the design and manufacture of implantable electrode arrays.
Each of these aspects and technologies require further development before electrode arrays that can come close to the target specification for systems like cochlear and retina implants can be realised.
Fuller L, Probes and Electrodes http://people.rit.edu/lffeee/mem_probes.pdf
Kipke DR, Advanced Neural Implants and Control http://www.philipnute.com/pdfs/darpa.pdf
Li W, Integrated Retinal Implants, Thesis, CALTECH, Pasadena, California, 2009 http://thesis.library.caltech.edu/648/1/Dissertation_Wen.pdf
Taranovich S, Medical sensors in biomedical electronics, part 1: the eye and ear http://www.edn.com/article/520122-Medical_sensors_in_biomedical_electronics_part_1_the_eye_and_ear.php
Taranovich S, Medical sensors in biomedical electronics, part 2: the brain, heart and lung http://www.edn.com/article/520108-Medical_sensors_in_biomedical_electronics_part_2_the_brain_heart_and_lung.php
Deman, P., Kaiser, T., Dirckx, J., Offeciers, F., Peeters, S. Design, construction and mechanical optimisation process of electrode with radial current flow in the scala tympani. Journal of Neuroscience Methods. 2003; 128(1–2):143–150.
Feucht, M., Laube, T., Bornfeld, N., Walter, P., Velikay-Parel, M., Hornig, R., Richard, G. Entwicklung einer epiretinalen Prothese zur Stimulation der humanen Netzhaut (in German) [Development of an epiretinal prosthesis for stimulation of the human retina]. Der Ophthalmologe. 2005; 102:688–691.
Hassler, C., von Metzen, R., Ruther, P., Stieglitz, T. Characterization of parylene C as an encapsulation material for implanted neural prostheses. Journal of Biomedical Materials Research B: Applied Biomaterials. 2010; 93b:266–274.
Hodgins, D., Bertsch, A., Post, N., Frischholz, M., Volckaerts, B., Spensley, J., Wasikiewicz, J., Higgins, H., Von Stetten, F., Kenney, L. Healthy Aims: developing new medical implants and diagnostic equipment. IEEE Pervasive Computing. 2008; 7(1):14–21.
Kumar, A., Achyuta, H., Polikov, V., White, A., Pryce Lewis, H., Murthy, S. Biocompatibility assessment of insulating silicone polymer coatings using an in vitro glial scar assay. Macromolecular Bioscience. 2010; 10:872–880.
Lago, N., Yoshida, K., Koch, K., Navarro, X. Assessment of biocompatibility of chronically implanted polyimide and platinum intrafascicular electrodes. IEEE Transactions on Biomedical Engineering. 2007; 54(2):281–290.
Oliver, W., Pharr, G. An improved technique for determining hardness and elastic modulus using load and displacement sensing indentation experiments. Journal of Materials Research. 1992; 7(6):1564–1583.
Oliver, W., Pharr, G. Measurement of hardness and elastic modulus by instrumented indentation: Advances in understanding and refinements to methodology. Journal of Materials Research. 2004; 19(1):3–20.
Qin, G., Liu, Y., Liu, Y., Ding, Y., Qi, X., Du, R. Fabrication of bio-microelectrodes for deep-brain stimulation using microfabrication and electroplating process. Journal Microsystem Technologies. 2009; 15(6):933–939.
Slavcheva, E., Vitushinsky, R., Mokwa, W., Schnakenberg, U. Sputtered Iridium oxide films as charge injection material for functional electrostimulation. Journal of the Electrochemical Society. 2004; 151(7):226–E237.
Szmigiel, D., Domanski, K., Prokaryn, P., Grabiec, P., Sobczak, J. The effect of fluorine-based plasma treatment on morphology and chemical surface composition of biocompatible silicone elastomer. Applied Surface Science. 2006; 253:1506–1511.
Szmigiel, D., Hibert, C., Bertsch, A., Pamula, E., Domanski, K., Grabiec, P., Prokaryn, P., Scislowska-Czarnecka, A., Plytycz, B. Fluorine-based plasma treatment of biocompatible silicone elastomer: the effect of temperature on etch rate and surface properties. Plasma Process Polym. 2008; 5:246–255.
Weiland, J., Anderson, D., Humayun, M. In vitro electrical properties for iridium oxide versus titanium nitride stimulating electrodes. IEEE Transactions on Biomedical Engineering. 2002; 49(12 Pt 2):1574–1579.