I. Introduction
Terahertz technology promises myriad opportunities for high data rate communications, biological spectroscopy or biomedical diagnostics, remote high resolution imaging of concealed weapon or threat detection, basic materials research, etc [1]. However significant challenges exist for the development of new compact radiation sources, based on physical principle of converting the kinetic energy of an electron current into electromagnetic field energy at terahertz frequencies [2]. Beam-wave interaction circuits above 0.1 THz where dimensional tolerances are around tens of microns have confronted many issues not only in designing and modeling interaction circuits, but also in improving RF return or insertion loss, enhancing electrical or thermal conductivity, reducing frequency shift, etc [3]. When the operating frequency reaches 0.1 THz on copper, for example, the surface roughness needs to be carefully controlled because of RF skin depth of only 210 nm. Conventional fabrication technologies such as mechanical machining or electrical discharge machining (EDM) are not suitable for such stringent requirements, but recent progresses in micro-electro mechanical systems (MEMS) and nano-electromechanical systems (NEMS) have allowed to achieve nano-scale precision of interaction circuits [3]. One of state of-art technologies is silicon microfabrication, which can be used to effectively fabricate complicated 3-dimensional microstructures to high accuracy [4], [5]. Particularly, a two-level, deep-reactive-ion etching (DRIE) and thermo-compressive, hermetic wafer bonding (TCB) technology was employed for 0.1 THz backward-wave oscillators [5]. In this work, however, the plasma loading effect of two-level DRIE at the narrow channel between resonant cavities changed aspect ratios, which eventually deteriorated RF return loss and frequency shift [5]. In our follow-up study of the silicon microfabrication, we propose multi-level microfabrication which overcomes such difficulties and enables more accurate RF characteristics in complicated slow-wave structures, including beam tunnel, cavities, waveguide transitions, and output port. This approach may allow a great deal of flexibility in design of various kinds of slow-wave circuits. It will be further discussed in next session.