Neuro-Electronic Interfacing Methods for High-Density CMOS-Compatible Microelectrode Arrays

Recently, CMOS-based microelectrode arrays containing a high-density of electrodes have emerged as a tool enabling recording the extracellular neural electrical activity of cell cultures at subcellular resolution. However, several improvements in areas such as the low-noise front-end readout electronics, post CMOS fabrication processing of the electrode array, or the cell-electronics co-simulation are still required in order to enable the widespread use of high-density active MEAs. An electrical model of the cell-electrode interface characteristics has been developed, enabling co-simulation of the cell-electrode environment with the front-end electronics at subcellular resolution. This model has shown to be adapted to determine the optimum electrode size, the amplitude of the sensed voltage, and the noise injected by the cell-electrode interface. Moreover, simulations performed using this model has also shown that the electrical coupling between a neural cell and an individual sensor is improved in the case of three-dimensional electrodes. Compared to planar electrodes, a 10-20 dB increase of the electrical coupling is observed for subcellular resolution three-dimensional tip electrodes, in simulation. The fabrication of high-density three-dimensional tip electrode array has thus been investigated. A novel post-processing technique which enables the systematic fabrication of very dense floating-gate field-effect transistor (FG-FET) arrays is presented. Several three-dimensional microstructures such as pillar, pyramidal, and inverted pyramidal microstructures have been manufactured. A maximum sensor density of 295'200 sensors/mm2, which corresponds to a sensor pitch dimension of 1.84 µm, has been obtained with pillar structures. Moreover, since no photolithographic steps are necessary for manufacturing the high-density FG-FET arrays, it has been shown that the proposed manufacturing technique is adapted in the case where die-level postprocessing is performed. An innovative readout architecture, where a single amplification stage simultaneously records the activity acquired from several electrodes, has also been developed. This new readout method, based on the amplitude modulation of the recorded signals, enables the design of a low-noise amplification stage while still reading the extracellular activity sensed by the whole electrode array. A theoretical analysis has demonstrated that a major physical limitation of the proposed readout architecture relates to the summation of the thermal noise of each recorded signal at the input node of the front-end amplification stage. After CMOS implementation of the proposed readout architecture, it has been shown that the maximum number of sensors which can simultaneously be recorded depends on the electrical characteristics of the recorded extracellular voltages, which depend on the experimental setup. If a typical case encountered during electrophysiological experiments is considered, the maximum number of electrodes which can be simultaneously recorded is approximately in the range of 5-10.