Low-Noise Low-Power Acquisition ICs for Neural Recording with Microelectrode Arra

With the problems of aging moving to center stage in many countries, there is an emerging demand for novel point-of-care diagnostic testing (POCT) with portable/wearable/implantable devices. Among the widespread chronic diseases, the brainrelated diseases (neurodegenerative diseases, neural disorders, etc.) have raised significant attention due to their large patient populations and the lack of effective diagnostic methods for long-term monitoring at the early stages.;Real-time diagnosis and mechanism study of brain-related diseases requires the recording of electrophysiological/electrochemical neural signaling at the cellular level. Besides the clinical EEG and traditional patch-clamp methods, integrated with electronics, the microelectrode array (MEA) platform provides another viable method of diagnosis with great benefits that the other methods do not afford: high spatial/temporal resolution, high throughput and being less invasive.;Although MEA devices have been widely studied, great technical challenges exist for future POCT applications. The integration of the MEA device and the acquisition IC for highresolution recording is a major constraint at the moment. The performance of existing acquisition ICs is not suitable for POCT devices, which limits the POCT applications greatly. Therefore, the design of acquisition ICs has to be altered for the technology to advance.;This thesis focuses on developing a set of novel acquisition IC designs for different modes of neural signals, which include neural potentials and neurotransmitters, as well as neuron-electrode adhesion. In the research scope of this thesis, two chips have been developed with voltage channels for sensing of neural potentials, current channels for sensing of neurotransmitter release and impedance channels for monitoring the adhesion. Novel circuit techniques to reduce the noise and improve the power efficiency, the linearity and the processing dynamic range techniques are developed.;The first chip is designed to capture dual-band neural potentials, which include the local-field potential (LFP, 0.1-200Hz) and the spike potential (SP, 200-10kH). To achieve high sensitivity and avoid the cross-talk between dual-band signals, the acquisition IC uses a continuous-time (CT) front-end with chopping to suppress the flicker noise in the LFP band and a discrete-time (DT) back-end to achieve good linearity. A new feedback loop is also proposed to linearize the chopped low-noise amplifier (LNA) in the front-end.;The second chip is designed to record the chemical neurotransmitter release and electrical neural potentials simultaneously for high-density MEA. As the MEA technology is migrating to more electrodes (up to tens of thousands), power and area constraints are becoming the bottleneck of the acquisition IC design. Under the constraints of noise, linearity and settling performance, this design enables the most efficient usage of power and area. The current channel for chemical sensing saves power and lowers the noise with a novel current buffer biased in the discrete-time (DT) before a TIA (trans-impedance amplifier). The voltage channel for electrical sensing maximizes its power and area efficiency with a new methodology proposed to optimize the major noise sources.;Impedance is another important signal from biosensors. This thesis also presents the development of a novel impedance channel for fast electrochemical impedance spectroscopy (EIS) with the best power efficiency. The design implements heterodyne conversion for the first time. It allows multiple input frequencies with the FFT methods for fast response, and it preserves the amplitude and phase, which can be extracted simultaneously without another quadrature channel. The impedance front-end is designed with low noise, low settling power and minimal interference.