Objective. Ensuring the longevity of implantable devices is critical for their clinical usefulness. This is commonly achieved by hermetically sealing the sensitive electronics in a water impermeable housing, however, this method limits miniaturisation. Alternatively, silicone encapsulation has demonstrated long-term protection of implanted thick-film electronic devices. However, much of the current conformal packaging research is focused on more rigid coatings, such as parylene, liquid crystal polymers and novel inorganic layers. Here, we consider the potential of silicone to protect implants using thin-film technology with features 33 times smaller than thick-film counterparts. Approach. Aluminium interdigitated comb structures under plasma-enhanced chemical vapour deposited passivation (SiO x , SiO x N y , SiO x N y + SiC) were encapsulated in medical grade silicones, with a total of six passivation/silicone combinations. Samples were aged in phosphate-buffered saline at 67 °C for up to 694 days under a continuous 5 V biphasic waveform. Periodic electrochemical impedance spectroscopy measurements monitored for leakage currents and degradation of the metal traces. Fourier-transform infrared spectroscopy, x-ray photoelectron spectroscopy, focused-ion-beam and scanning-electron- microscopy were employed to determine any encapsulation material changes. Main results. No silicone delamination, passivation dissolution, or metal corrosion was observed during ageing. Impedances greater than 100 GΩ were maintained between the aluminium tracks for silicone encapsulation over SiO x N y and SiC passivations. For these samples the only observed failure mode was open-circuit wire bonds. In contrast, progressive hydration of the SiO x caused its resistance to decrease by an order of magnitude. Significance. These results demonstrate silicone encapsulation offers excellent protection to thin-film conducting tracks when combined with appropriate inorganic thin films. This conclusion corresponds to previous reliability studies of silicone encapsulation in aqueous environments, but with a larger sample size. Therefore, we believe silicone encapsulation to be a realistic means of providing long-term protection for the circuits of implanted electronic medical devices.

We are developing an electrode array for epidural spinal cord stimulation and a thin integrated circuit (IC) is to be embedded in it. This paper focuses on the development and characterization of a manual process for thinning individual IC die and discusses the issues associated with thinning small dice by a manual process. The procedure allows easy and controlled post-separation thinning of small (about 1 mm²) silicon chips by grinding. A systematic approach was followed to characterize the technique and repeatability of the results. With the setup we introduced we were able to control the final thickness of the IC with a standard deviation of 9.2 μm. Although no chemical processing is used, a small grit size film can create smooth surfaces, with roughness comparable to reported values after etching, acting as the so-called 'stress-relief' step. Electrical tests performed on a thinned stimulator output stage IC indicated that no die damage was caused by the procedure. Some issues regarding the integration of thinned ICs on flexible substrates and the reliability of gold ball rivet bonds on the ICs' aluminium pads are also discussed.

We are developing an active implant for epidural spinal cord stimulation. A thin application specific integrated circuit (ASIC) (~80 μm) is to be embedded within it. The laser patterned tracks are electrically and mechanically thermosonically bonded on the ASIC pads using gold ball studs, forming micro-rivets through holes in the foil of the tracks, an interconnection method called electrical rivet bonding, or microflex [1]. In this work, we sought to characterize and optimize the technique, with respect to its bonding strength. The technique is relatively new and, so far, the mechanical strength of the bonds has only been investigated for interconnection on gold tracks. Standard ASICs however, normally come with aluminium pads. We ran a series of pull tests on the bonds between the metal tracks and aluminium ASIC pads. In these tests, we were concerned with the effect of the different parameters on the bond strength, and more specifically the size of the gold balls and the size of the holes in the foil. We recorded the maximum force (stress) before bond failure for different combinations of parameters. Our results indicate that average stress values can vary between 9.6 and 60 cN, depending on the process parameters. Different failure mechanisms have been identified and these are discussed. Overall, we conclude that larger holes provide larger contact areas with the substrate and generally result in stronger bonds, but the right combination of ball and hole sizes, could lead to strong bonds even with smaller holes.

This paper discusses the design of an application-specific integrated circuit (ASIC) suitable for mounting on a multi-electrode array for epidural spinal cord stimulation in rats. The ASIC acts as a demultiplexer, driving 12 electrodes on the array in any configuration. It is capable of routing biphasic constant current pulses of up to 1 mA to high impedance loads (with a maximum output voltage swing of approximately 25 V) and is small enough to be implanted into a rat's spinal column. Communication with its driver is achieved via 3 wires to minimize the number of interconnections. The circuit was implemented in a 0.18-μm high-voltage CMOS technology occupying a core area of 0.36 mm ². Power dissipation is about 110 μW. Post-layout simulations are presented which show the correct operation of the system.

The design of a very small low-power IC driving 16 electrodes in any combination, for testing different epidural stimulation patterns in freely moving rats using a two wire interface is presented. A scheme is proposed, to receive power, data, clocking information and stimulus current pulses from an external source via the two wires. This is achieved by a combination of time multiplexing and pulse width modulation. The architecture of the system is discussed and simulated results of some building blocks are presented.