Author:

Keywords:

neuro-electronics, neuron, cardiac, VLSI processing, design, electrophisiology, MEMS, micro-nail electrode

Abstract:

In vitro culturing and examination of electrogenic cells such as neurons is performed extensively in the fields of neurophysiological research and pharmacology. The patch clamp method is an established measurement technique, since it allows single ion channel current measurements and membrane potential recordings with excellent signal quality. However, the method is not suitable for long term experiments and allows only a limited number of simultaneous recordings. In order to cope with these issues, new measurement techniques based on extracellular recordings weredeveloped in the last decades.One of this measurement techniques is the use of multi-electrode arrays (MEAs) which have been extensively used in the last decades. MEAs are based on glass substrates and passive electrodes, and therefore, have a limited number of readout sites. In order to scale up the number of electrodes and improve the signal quality, integrated circuits on silicon chips have been developed. In such designs, the sensors are connected to an in situ amplifier and could be addressed either directly or indirectly by a matrix read-out scheme.Extracellular recording techniques have the advantage to be non-invasive, however, have some disadvantages. Firstly, the signals are usually much smaller in amplitude compared to whole-cell patch-clamp recordings. Secondly, depending on the size of the electrodes and density of the cell cultures, different neuronal signals are recorded by one electrode. Calculation intensive algorithms are required to identify the signal sources.In this work, we aimed to make a chip which allows high-density read-out and stimulation of neuronal networks with single-cell resolution. Such a chip would help to provide detailed insight in cell-to-cell interactions in complex neuronal networks. True single-cell addressability was not demonstrated before because of the limited cell/sensor coupling efficiency when electrodes smaller than the size of a cell body are used.In order to improve the cell/sensor coupling, we used the cell engulfment hypothesis, stating that in the right conditions, a cell is capable of engulfing a micro-nail shaped electrode. If true, this would enable the use of small electrodes and yet provide good cell/electrode interface, allowing true single-cell interaction.The work consisted of four phases. In the first phase, theoretical models were created in order to simulate the cell/sensor interface. Analysis of these models revealed a strong dependency on the electrode diameter and the seal resistance,which is defined by the gap between the cell membrane and the shaft of the micro-nail. The models predicted the possibility to measure small pores or single ion channel activity in the cell membrane on top of the electrode, and also predicted that electrical stimulation with small electrodes results in electroporation.In the second phase, based on the results of the modeling work, a novel chip architecture for high-density single-cell interfacing was developed. The chip was designed for a commercial mixed-signal 0.18 µm CMOS-technology, resulting in a matrix with 16384 addressable micro-electrodes, containing in-situ recording and stimulation circuits. The CMOS wafers were fabricated in Taiwan SemiconductorManufacturing Company (TSMC).The third phase comprised of process flow design and integration of a micro-electrode array. This processing pas performed in the IMEC 200 mm clean room on top of the TSMC wafers. Also, microfluidic syringes were integrated on the CMOS chips, allowing local chemical interaction with the cells. The chips were packaged by a custom flip-chip process. A measurement set-up was developed, consisting of a printed-circuit board and a mechanical system which allowed chip experimentsin combination with a patch-clamp set-up and an upright fluorecence microscope.In the last phase of the project, the chips were characterized and physiological experiments were performed. Using the on-chip circuits, electrochemical impedance spectroscopy measurements of the titanium nitride electrodes revealed oxide formation on the surface, causing problems with neuronal recordings. Therefore, we adapted the process flow to form tungsten micro-electrodes. Because thetungsten electrodes were toxic for neurons, the electrophysiological experiments were performed with cardiac cells and cell lines. By using the whole-cell patch clamp and fluorescent microscope techniques as references, extracellular recording and stimulation with single-cell resolution was demonstrated with the chips. The recordings showed slow negative potentials, which could not be explained by a linear transfer function. Therefore, a model including a slow cationic voltage-gated ion channel was implemented. This model demonstrateda good approximation of the recorded signals. This analysis indicated increased seal resistance and measurement of ion channel activity, which was predicted by the simulations.We believe that this thesis could be a first step towards a new generation of neuronal sensor arrays with high-density and single-cell addressability. Such neuronal sensor arrays could provide new insights in the field of neurophysiological research and might be a valuable tool for cost-efficient pharmaceutical development.