ROSCAMAR

The ROSCAMAR project consists of the design and realization of demonstrators of RF-MEMS oscillators in the 100–160 MHz Frequency range. The first originality of the project lies in the achievement of elastic waves resonators on silicon localizing the acoustic energy to optimize the quality factor by cancellation of the anchor losses. Our goal consists in reaching the limits allowed by the visco-elastic behavior of the resonator material to grant the best spectral purity compatible with massively parallel fabrication processes stemming from microelectronics. The second originality of the project resides in the realization of demonstrators of resonant structures from stacked silicon wafers integrating MOS transistors in order to optimize the exploitation of the specific electrical response of the resonators based on electrostatic transduction. Those active components are aimed to sustain the achievement of the compensation of the feed–through capacitance arising from the use of sub-micronic electrostatic gaps. The project relies on recent but firm progresses of a theoretical and experimental research on Bulk Acoustic Waves resonators structures with electrostatic excitation, which led to a PhD defense [december 2012] and to the french patent FR 11 56250, now extended to EU and USA by the CNRS. After obtaining a new PhD grant in december 2012, a Research&Technology project at CNES, and a postdoc grant from the First-TF Labex in 2013, the Time-Frequency Dept. of FEMTO-ST is able to reinforce the ambition of the thema RF-MEMS resonators and oscillators by proposing an alternative solution to the approach followed by the companies actually producing MEMS oscillators, based on more conventional structures vibrating in the plane of the stacked wafers. The latter approach is altogether limited in terms of intrinsic performances and rather locked from the point of view of intellectual property. The demonstrators of smart capacitive resonators will investigate to distinct approaches: An assembly of stacked silicon wafers by thermo-compression through Au/Au contact layers achieving the air or vacuum gap used to generate the electrostatic force. Then, one silicon layer is thinned to reach the targeted frequency range through a thickness–extensional or plate waves trapped–energy mode. A mixed architecture stacking a silicon wafer on top of a SOI wafer. Then, the thickness of the inner silicon layer of the stack is used as the main mechanical parameter to control the propagation of guided elastic waves in this layer. The interconnexion of the conductive layers necessary to the electrostatic excitation with the outer oscillator circuit and the active components integrated in the stack will be achieved by a generic TSV (Through Silicon Via) technology. The project does not propose to deliver any demonstrator of integrated oscillator, for reasons pertaining to economical and technological realism. Through this hybrid approach associating simple active fonctionnalities with the implementation of a low loss MAR (Micro-Acoustic Resonator) in a stack of monocrystalline silicon layers, we aim to a drastic improvement of the compromise compactness / robustness / short term stability / cost to compensate the handicaps of MEMS resonators (thermal and structural stability, acoustic losses) with respect to piezoelectric resonators. The ROSCAMAR project has the vocation to child a further project of the ASTRID-maturation kind, aimed to demonstrate the feasibility of intermediate series production of such hybridized resonant structures on a pilot line of RF passive components.