A 3D intra-cortical electrode array is embedded for signal acquisition, processing, and wireless communication.
The ability to monitor the activities of ensembles of single neurons is critically important in understanding the principles of information processing in the brain that underlie perception, cognition, and action. Multiple microelectrode recording using appropriate neuronal implants provides this ability. The Telemetric Electrode Array System (TEAS) project aims at developing and embedding a three-dimensional intra-cortical electrode array with all electronics required for signal acquisition, processing, and wireless communication entirely into the head.
Although understanding the human brain with its several billion neurons is a formidable task, recent breakthroughs are leading the way to the realization of “brain-machine interfaces.” By encapsulating an entire brain implant into the skull and using wireless communication through the skull, pathways through the skin that could risk infections and electrical artifacts due to cable movement would be eliminated. This project aims at developing the minimum hardware and software required for an external computer to interface effectively with the brain, by adopting the brain-machine interface concept over the “total implant” concept.
The whole system to be embedded into the skull consists of a three-dimensional electrode array interconnected to an electronic block through a special flexible interconnection cable. The structure of the TEAS array is done using wire electro-discharge-machining (EDM) techniques (see figure). This numerically controlled approach enables machining of more complex microstructures than would be possible with a diamond saw. Tungsten carbide was initially used because of its fine grain and hardness. The entire array or just the tips of the needle electrodes are coated with platinum (for enhancing the charge transfer capabilities) using electron-beam deposition prior to applying an insulation coating of glass using electron-beam deposition or a biocompatible epoxy through a dipping process. In the case of dipping, the surface tension is often too high and heating the array above the annealing temperature of tungsten carbide prior to the dipping process to reduce the surface tension may also be required.
Providing independent electrical access to each electrode in the array is a flip-chip mounting based on stud bumping. With this high-density mounting method, an additional coating of aluminum or gold is applied for the wire bonding techniques over tungsten carbide, platinum, or iridium at the interconnection surface of each electrode to allow ultrasonic wire bonding. The process is the same as that of wire bonding IC die to lead frames.
The flexible attachment provides the electrical connections between the needle electrodes and the electronic block. The initial attachment consists of a single conductive layer of a polyimide-based flexible substrate with 64 conducting traces running in parallel on the top layer between the array and the electronic block. These specifications were chosen to ensure small dimensions while providing a good yield during manufacturing.
Unlike the rest of the system, the electronic block is fixed and attached to the skull. It consists of a front-end amplifier; the analog-to-digital (A/D) conversion and multiplexing; the triggering, control, and buffering subsystem; the wireless communication interface; and the power section.
Although through-hole assembly has been chosen for the first version of the implant to minimize the number of parts and simplify the assembly process, it is anticipated that the flip chip will be used as the feature sizes decrease. As the number of electrodes within the same array increases, the diameter of the needle electrodes will need to decrease to minimize the insertion force and damage to the brain. As such, other fabrication techniques will be required, as well as improvement in wireless communication.
This work was done by S. Martel, I. Hunter, J. Burgert, J. Malasek, C. Wiseman, and R. Dyer of the BioInstrumentation Laboratory at Massachusetts Institute of Technology; and N. Hatsapoulos and J. Donoghue of the Department of Neuroscience at Brown University for the Army Research Laboratory. ARL-0063