Scalable readout IC based on wideband noise cancelling for full-rate scanning of ultra-high-density microelectrode array (UHD-MEA)

Abstract

Microelectrode array is the pivotal platform in neural science and engineering because of its ability to read out the electrical activity of living neurons. The recent development of high-density and highly scalable electrode arrays enables sub-cellular recording and wider coverage, respectively. However, it is extremely challenging to achieve both metrics in a single array. The speed requirement in highly scalable arrays induces wideband-noise folding, and this property precludes meeting the strict noise requirements in high-density arrays. Previous studies contended that circuit thermal noises should be reduced to improve the signal-to-noise ratio of the recording system. By investigating the cell-electrode interface, however, we showed that the sealing resistance between a cell and an electrode turns out to contribute over 90% of total wideband noise power. In line with this finding, wideband-noise canceling strategy was presented to cope with the condition and support reliable recording. The recording system supported full-frame scanning of 100,000 electrodes with a $10- \mu m$ pitch. As a result, an 85% reduction of the noise contribution by the sealing resistance was achieved. Input-referred noise was reduced from $80 \mu V_{rms}$ to $18 \mu V_{rms}$. Spikes in weak-sealed cases that could not be detected without the cancellation were disclosed, and signal-to-noise ratio was significantly increased for all sealing conditions.