Abstract: This dissertation addresses the key issues related to the development of dry-etch Benzocyclobutene (BCB) based neural-electronic interface for intra-cortical and epidural neural recording using surface micromachining techniques. First, the processing and characterization of Dry-etch BCB thin films are studied. This includes study of BCB film thickness, surface uniformity control through spin-coating, curing recipe development and optimization, dry etch rate and aspect ratio control, planarization level study, and degradation effect investigation. The results reveal that the dry-etch BCB film uniformity and pin-hole density are mainly determined by the ramping speed and spin-duration. Dry etch BCB film (25um) with surface roughness less than 1000A and pinhole density less than 1.5x10-3mm-2 has been acquired. The BCB etch rate and via angle are determined by the pressure, gas concentration, and RF power. The optimized plasma etch test shows that greater than 1μm/min etch rate and 65 degree via angle are best for the packaging and patterning of the neural probes. The degree of planarization of BCB decreases as the thickness of underlying pattern and the etch rate and film thickness of BCB drops 10% as time after 12 months. Second, the design and fabrication process for various neural electrode arrays using silicon surface micromachining technique are investigated. This includes structure and process design of single side neural electrode array (SNEA) and double side neural electrode array (DNEA). Several approaches have been investigated to fabricate the BCB based neural electrodes, and the one with higher FAB yield and device performance is identified. The related processing techniques were also studied, which including the mask making, metallization, wet etch, photolithography, silicon etch, and flip chip bonding technique. Penetrating and impedance test reveals the dry-etch BCB based neural implant with 1.5um tungsten incorporation or enough thickness can penetrate into the brain without damage to the electrode and brain. The electrode with 20um size recording site shows 2M-Ohm impedance at 1kHZ. The process yield and device failure mechanism are analyzed. Several approaches have been performed to improve the electrode yield from 40% to 70%, however, long term effort is still required to improve the overall process yield.