Abstract
Recently, research on the neuroscience is becoming a potential subject and holds a lots attraction, such as Brain Machine Interface (BMI) or the Brain-Computer Interface (BCI). The development of neuroscience not only promotes the progress of the medicine and pharmacology but also the understanding of brain function. Furthermore, the neural prosthesis is useful for patients with neural diseases to restore part of their physiological functions. The bidirectional communication is based on recording neural activity and stimulating neurons. Intracellular recording and extracellular recording are both approaches to monitored neural activity. Compared with Intracellular recording, extracellular recording has the ability to record large amount of samples and the synchronized activities. This thesis proposes a low-noise, CMOS-compatible, neuro-silicon interface for extracellular recording. For the multi-channel array system, CMOS-compatible sensor can be integrated with the recording circuit to decrease the routing complexity. The interface is based on the OSFET, which is a MOSFET with several micromachined process steps. In the experimental results, we find that OSFETs have some unideal characteristics, process variation、threshold-voltage drift and increased noise after post process. Finally, the automatic Vth-compensation feedback loop is used to overcome the biasing complex for array system. The post-CMOS process, characteristics of neuro-silicon interface, the circuit design, the influence of gate to bulk voltage for OSFET noise，and the results of biological tests are introduced in the thesis.