Towards LSI Vibrating Micromechanical Signal Processors

temperature stabilities better than 18 ppm over 27-107 o C [3] (c.f., Figure 3), putting them on par with the performance of most quartz crystals; (5) demonstrations of impedances approaching 50Ω allowing them to match to macroscopic antennas; and (6) demonstrations of medium-scale integrated (MSI) micromechanical circuits and functions, including low insertion loss bandpass filters (c.f, Figure 4), mixers, and even amplifiers. With many of the above attributes superior to those attainable by macroscopic counterparts, the technical argument for the use of vibrating RF MEMS as high-Q replacement components in wireless sub-systems is already quite strong. But the benefits of vibrating RF MEMS technology go far beyond mere component replacement. In fact, the extent of the performance and economic benefits afforded by vibrating RF MEMS devices grows exponentially as researchers begin to perceive them more as building blocks than as stand-alone devices. In particular, when integrated into micromechanical circuits, in which vibrating mechanical links are connected into larger, more general networks, previously unachievable signal processing functions become possible, such as reconfigurable RF channel-selecting filter banks, ultra-stable reconfigurable oscillators, frequency domain computers, and frequency translators. When further integrated together with other micro-scale devices (e.g., transistors, micro-ovens, micro-coolers, atomic cells), system-level benefits for portable applications abound, particularly those for which architectural changes allow a designer to trade high Q for lower power consumption and greater robustness, with potentially revolutionary impact. Indeed, circuit complexity and frequency range should only increase as MEMS technologies evolve into NEMS (or “nanoelectromechanical system”) technologies, with feature sizes that support frequencies exceeding 10 GHz. In fact, with knowledge of the micromechanical circuit concepts described above, one might now ponder whether or not the nanowires targeted for transistor functions might instead be better employed more naturally as vibrating resonators capable of doing mechanical signal processing when mechanically linked into circuit networks, such as that of Figure 5. Such nanomechanical networks would not only be completely passive, consuming substantially less power, but would also dispense with the need for the electrical contacts that presently inhibit large scale integration of nanowire transistors. This paper suggests the NEMS technologies and attributes most suitable to enabling such an integrated nanomechanical circuit technology.