Development and modelling of a versatile active micro-electrode array for high density in-vivo and in-vitro neural signal investigation

The electrophysiological observation of neurological cells has allowed much knowledge to be gathered regarding how living organisms are believed to acquire and process sensation. Although much has been learned about neurons in isolation, there is much more to be discovered in how these neurons communicate within large networks. The challenges of measuring neurological networks at the scale, density and chronic level of non invasiveness required to observe neurological processing and decision making are manifold, however methods have been suggested that have allowed small scale networks to be observed using arrays of micro-fabricated electrodes. These arrays transduce ionic perturbations local to the cell membrane in the extracellular fluid into small electrical signals within the metal that may be measured. A device was designed for optimal electrical matching to the electrode interface and maximal signal preservation of the received extracellular neural signals. Design parameters were developed from electrophysiological computer simulations and experimentally obtained empirical models of the electrode-electrolyte interface. From this information, a novel interface based signal filtering method was developed that enabled high density amplifier interface circuitry to be realised. A novel prototype monolithic active electrode was developed using CMOS microfabrication technology. The device uses the top metallization of a selected process to form the electrode substrate and compact amplification circuitry fabricated directly beneath the electrode to amplify and separate the neural signal from the baseline offsets and noise of the electrode interface. The signal is then buffered for high speed sampling and switched signal routing. Prototype 16 and 256 active electrode array with custom support circuitry is presented at the layout stage for a 20 μm diameter 100 μm pitch electrode array. Each device consumes 26.4 μW of power and contributes 4.509 μV (rms) of noise to the received signal over a controlled bandwidth of 10 Hz - 5 kHz. The research has provided a fundamental insight into the challenges of high density neural network observation, both in the passive and the active manner. The thesis concludes that power consumption is the fundamental limiting factor of high density integrated MEA circuitry; low power dissipation being crucial for the existence of the surface adhered cells under measurement. With transistor sizing, noise and signal slewing each being inversely proportional to the dc supply current and the large power requirements of desirable ancillary circuitry such as analogue-to-digital converters, a situation of compromise is approached that must be carefully considered for specific application design.EThOS - Electronic Theses Online ServiceGBUnited Kingdo