Fig. 3: g-SGFET stability in-vivo.

a CNP vs. time over 4 weeks. The inset shows the I_ds–V_gs curves. Mean and standard deviation for n = 8 g-SGFETs (1 outlier excluded). b Signal from two DC-coupled channels. Indicated positions corresponding to map in panel h. The spectrogram of channel (5,7) is shown (bottom). c The phase-amplitude relation between the channels in panel b, for the noise measured in the beaker (left) and for the signals measured in vivo (right). d Boxplot of V_gs-rms vs. shifts in the effective gating (V_gs–V_CNP) of the 64 g-SGFETs. The colored area indicates the measured drift in the CNP referred to a Ag/AgCl electrode during the first 24 h of recording. The initial bias and the CNP are indicated by the red and green vertical lines respectively. e g_m (top) and V_gs-rms (bottom) measured over 4 weeks post implantation; g_m was obtained from the I_ds–V_gs curves of the DC-coupled channels (n = 8 g-SGFETs, 1 outlier excluded). f Current noise over 4 weeks after implantation (n = 64 g-SGFETs). The numeric values indicate the yield of working devices (see supplementary information S4). The boxes in panels a and d–f extend from the lower to the upper quartiles, with a line at the median. The whiskers extend 1.5 times the inter-quartile range and the data points beyond the whiskers are indicated by a dot. g Average and standard deviation of the frequency-dependent transconductance (|g_m | (f)) shown for different days after the implantation (n = 10 g-SGFETs). The inset shows the approximate position of the Pt-Ir electrode close to the array, the simulated equipotential contour lines in a conductive plane and the relative signal amplitude measured by each of the g-SGFETs in the array (see supplementary information S5). h Signals measured by all g-SGFETs on the array during a state of increased theta activity on day 1 and day 6 after implantation.