Current and future uses of high density neural implant arrays for functional electrical stimulation systems

David J. Anderson

Departments of Electrical Engineering and Computer Science, Biomedical Engineering and Otolaryngology, University of Michigan, Ann Arbor, MI 48109-2122, USA

Introduction

A successful implant used for FES and sensor systems that support FES depends on the confluence of device fabrication technology, biomaterials electronics packaging and signal conditioning and processing. Over many years, we have been building a technology that embodies as much of this function into a system that is batch processed using silicon technology in much the same way a computer or communications chip is fabricated [1]. The icon for such a device is shown in Figure 1a. Much can be designed into the core silicon device such as functional form, efficient sensing and charge delivery electrochemistry [2-4], silicon cables [5], fluid delivery channels [6], electronics, pacification against the hostile environment of the body and signal routing. A key factor to the utility of such a system is the precision of the recording and stimulation sites.

Although the ideal is to have all function integrated into the batch fabricated core device, post-processing steps are still necessary to provide the core device with all the functionality to operate as an implant. These include a delivery system to the target tissue, treatment for biocompatibility and/or bioactivity, provision for communication with outside systems and additional pacification to prevent the environment from reducing function.

Process technology

The process technology used to fabricate the University of Michigan neural interface systems on silicon wafers is largely based on thin-film techniques used widely in the semiconductor industry. The benefit of this compatibility is the availability of high precision process equipment and services at a reasonable cost. The fundamental process difference is the method of device release from the silicon wafer by a controlled chemical etch. The chemical etch assures that the devices can be any arbitrary 2-dimentional shape and the edges are somewhat blunt, reducing damage during insertion into neural tissue. Figure 1b shows the steps in the process. The first and last steps are critical to the unique patterns we are able to achieve, and make this process different from usual silicon processing or even most MEMS processing. Boron is diffused into the silicon lattice by high temperature in a pattern which will later be the probe substrate shape. The nature of diffusion process assures that the boron concentration spatial derivatives are continuous in the bulk silicon. The EDP etchant has an etch rate that is highly nonlinear with boron concentration making it nearly a threshold effect. The result is a smooth device surface with no sharp cutting edges.

Fluid channels can be added to the devices by only one extra mask. A dense mesh of boron-doped silicon is formed on the surface of the probe over the future channel. The mesh is undercut with a timed EDP etch to form a channel with a scaffold over the top (Figure 1c). After additional boron diffusion to protect the bottom and sides of the channel from the final etch, the mesh can be filled in and the channel sealed with subsequent pacification layers.

The above process description is the Michigan ‘passive’ probe process which has been in existence for many years and has produced electrodes used in many physiological studies conducted in laboratories all over the world. The process has been extended to an ‘active’ process that includes electronics. More complex devices can be fabricated with the active process but the fundamental advantages of precise placement of recording and stimulation sites are the same for both systems.

Challenges met

While development continues on the passive process and the more complex active process, the bulk of the activity using these devices has been developing their usefulness in several areas. First the design rules for the process are versatile enough to allow us to design systems that can deliver recording and stimulation sites to cortex, deep brain, spinal cord, peripheral nerves, and to sensory systems. Users of the system have been successful in all of the brain areas mentioned using acute preparations. We face few problems with acute preparations because mechanical stabilization can be imposed easily and the duration of the applications are so short that the brain does not begin its response to invasion by a foreign body. Chronic preparations such as what would be required for a FES system face many problems including mechanical stability, tissue reaction and sensor fouling. The cortex has been our most frequent chronic application because it is a relativity stable environment for a neural implant. The other areas of the brain are more of a challenge but we have been successful in deep brain structures and spinal cord. As we engage more and more applications, issues other than silicon processing take more of role. Figure 2 illustrates the issues that must be faced simultaneously for success in long-term chronic brain implants.

Future developments

Technology development continues on silicon substrate neural implant devices in the areas of drug delivery and electronic function. The most conspicuous challenges are biological. To be successful, as a prosthetic device, an implant must be able to record single cells or at least clusters of near by cells over months or years. We have identified the processes that proceed after implantation and are engaging problems they cause. The build up of proteins that foul electrical recording sites are a large problem for the type of FES system which depends on cortical or other input areas of the brain but are not a significant problem for stimulation electrodes [7]. Because sensor fouling affects both scientific use and clinical use of recording electrodes of any variety, we are aggressively attacking this problem, as are several other laboratories.

There are several methods for the delivery of agents for tissue engineering that we are pursuing. Fluid channels with the appropriate valves and pumps are attractive options. We have also worked with site treatments, coatings [8] and small storage wells near the sites. While providing many options, much work remains before we have a set of design rules for these tools.

References

[1] Najafi, K., K.D. Wise, and T. Mochizuki, A high-yield IC compatible process for fabrication of microprobes. IEEE Trans. on Electron Devices, 1985. ED-35(7): p. 1206-1211.

[2] BeMent, S.L., et al., Solid-state electrodes for multichannel multiplexed intracortical neuronal recording. IEEE Transactions on Biomedical Engineering, 1986. 33(2): p. 230-41.

[3] Drake, K.L., et al., Performance of planar multisite microprobes in recording extracellular single-unit intracortical activity. IEEE Transactions on Biomedical Engineering, 1988. 35(9): p. 719-32.

[4] Anderson, D.J., et al., Batch-fabricated thin-film electrodes for stimulation of the central auditory system. IEEE Transactions on Biomedical Engineering., 1989. 36(7): p. 693-704.

[5] Hetke, J.F., et al., Silicon ribbon cables for chronically implantable microelectrode arrays. IEEE Transactions on Biomedical Engineering., 1994. 41(4): p. 314-21.

[6] Chen, J., et al., A multichannel neural probe for selective chemical delivery at the cellular level. IEEE Transactions on Biomedical Engineering, 1997. 44(8): p. 760-9.

[7] Weiland, J.D. and D.J. Anderson, Chronic Neural Stimulation with Thin-Film, Iridium Oxide Electrodes. IEEE Transactions on Biomedical Engineering, 2000. 47(7): p. 911-918.

[8] Cui, X., et al., Surface Modification of Neural Recording Electrodes with Conducting Polymer/Biomolecule Blends," , vol. 56, pp., . J. Biomed. Mat. Res., 2001. 56: p. 261-272.