ASIC-Integrated, High Density Flexible Electrode Array with Long-Term Reliability

Abstract

For several decades, extracellular recording has been widely used as a significant tool to understand brain function and behaviors. Extracellular recording enables the isolation of the activity of a single neuron. Therefore, the physiology and connectivity of individual neurons can be characterized by monitoring their activities and, in turn, with the correlation with the behaviors and physiological events. For a broader understanding of brain circuits, more individual neurons have to be simultaneously monitored and thus a larger scale and density of recording sites are required. Another important aim of the neuroscientist is chronic recordings to monitor the plastic change in neuronal network, but the instability of recording over long periods of time has not been fully addressed yet mainly due to biological failure caused by an acute damage and chronic tissue response, and device failure caused by adhesion failure, electrode delamination and insulation failure. This work is concentrated on three objectives to address the challenges mentioned above and to improve neural interfaces: 1) to minimize acute damage during insertion of a flexible array, 2) to enhance the reliability of polymer-based neural electrode arrays, and 3) to increase the channel count within a limited shank width and with a small system form factor. A novel T-shaped diamond shuttle to deliver a flexible array with less tissue damage was designed and fabricated. T-diamond provides a 56% reduction in cross-sectional area compared to a planar silicon shuttle with equivalent insertion strength, and stiff enough to penetrate through tough layers such as with the dura and epineurium. It will theoretically reduce micro-vascular damage by 37% compared to standard probes. In vivo experiments with cat DRG and rat brain with dura verified the penetration capability of UNCD shuttle and the neural signals were successfully recorded. A long-term reliable, high-density polyimide based neural electrode array was developed by submicron patterning and metal-polyimide adhesion improvement for traces. Also, a wafer level impedance lowering process was developed. 64 channels with 0.5µm line and 0.5µm space could be implemented in 85µm width. A novel adhesion layer of TiO2 showed more than 6 months of lifetime through bench top test. A method for wafer level impedance lowering by substrate roughening was developed. The measured impedance of the roughened Pt electrode demonstrated 3X reduction compared with plain Pt electrodes. Integrating the presented advances, polyimide based neural probe with 64 channel was fabricated and tested by chronic in vivo recordings. The probe was validated until 14 days after implantation reliably showing 25 single units in average. An ASIC integrated, 256-channel recording system with a 4-shank polyimide neural probe was presented. A 256-channels neural probe with Ni/Au bumped silicon backend was interconnected with 256-channel high density ASIC by anisotropic conductive film bonding. The form factor of the proposed system demonstrated 4X increase in channel count density per package volume. The system was characterized by an in vitro recording setup measuring 17µVrms of noise. An acute In vivo validation presented local field potential from 198 channels and 20 clusters from spikes. Overall, a compact package of 256-channel neural probe with high compliance was developed. Both the channel density per volume and the compliance of the probe were improved compared with the state-of-the-art active flexible probe system. The compactness of high density probe recording system is a promising solution for a long-term recording.