Inductive coupling and matching networks are used to increase the bandwidth of filters realized with aluminum nitride contour-mode resonators. Filter bandwidth has been doubled using a wirebonded combination of a wafer-level-packaged resonator chip and a high-Q integrated inductor chip. The three-pole filters have a center frequency near 500 MHz, an area of 9 mm × 9 mm, insertion loss of < 5 dB for a bandwidth of 0.4%, and a resonator unloaded Q of 1600.
This paper, the second of two parts, reports the measurement and characterization of a fully integrated oven controlled microelectromechanical oscillator (OCMO). The OCMO takes advantage of high thermal isolation and monolithic integration of both aluminum nitride (AlN) micromechanical resonators and electronic circuitry to thermally stabilize or ovenize all the components that comprise an oscillator. Operation at microscale sizes allows implementation of high thermal resistance platform supports that enable thermal stabilization at very low-power levels when compared with the state-of-the-art oven controlled crystal oscillators. A prototype OCMO has been demonstrated with a measured temperature stability of -1.2 ppb/°C, over the commercial temperature range while using tens of milliwatts of supply power and with a volume of 2.3 mm3 (not including the printed circuit board-based thermal control loop). In addition, due to its small thermal time constant, the thermal compensation loop can maintain stability during fast thermal transients (>10 °C/min). This new technology has resulted in a new paradigm in terms of power, size, and warm up time for high thermal stability oscillators. [2015-0036].
This paper, the first of two parts, reports the design and fabrication of a fully integrated oven controlled microelectromechanical oscillator (OCMO). This paper begins by describing the limits on oscillator frequency stability imposed by the thermal drift and electronic properties (Q, resistance) of both the resonant tank circuit and feedback electronics required to form an electronic oscillator. An OCMO is presented that takes advantage of high thermal isolation and monolithic integration of both micromechanical resonators and electronic circuitry to thermally stabilize or ovenize all the components that comprise an oscillator. This was achieved by developing a processing technique where both silicon-on-insulator complementary metal-oxide-semiconductor (CMOS) circuitry and piezoelectric aluminum nitride, AlN, micromechanical resonators are placed on a suspended platform within a standard CMOS integrated circuit. Operation at microscale sizes achieves high thermal resistances (∼10 °C/mW), and hence thermal stabilization of the oscillators at very low-power levels when compared with the state-of-the-art ovenized crystal oscillators, OCXO. A constant resistance feedback circuit is presented that incorporates on platform resistive heaters and temperature sensors to both measure and stabilize the platform temperature. The limits on temperature stability of the OCMO platform and oscillator frequency imposed by the gain of the constant resistance feedback loop, placement of the heater and temperature sensing resistors, as well as platform radiative and convective heat losses are investigated. [2015-0035].
Radio frequency microelectromechanical system (RF MEMS) devices are microscale devices that achieve superior performance relative to other technologies by taking advantage of the accuracy, precision, materials, and miniaturization available through microfabrication. To do this, these devices use their mechanical and electrical properties to perform a specific RF electrical function such as switching, transmission, or filtering. RF MEMS has been a popular area of research since the early 1990s, and within the last several years, the technology has matured sufficiently for commercialization and use in commercial market systems.
Eliminating spurious modes in Aluminum Nitride (AlN) microresonators improves their insertion loss and quality factor by reducing acoustic energy leakage. Spurious modes that result from transverse wave propagation, termed fine-frequency modes, leak energy and propagate in the electrical busing and appear near the fundamental resonance. Although these modes can be predicted using three-dimensional (3D) finite element methods (FEM) for devices with very short acoustic length (e.g. 1 acoustic wavelength), 3D FEM is very slow and memory intensive when compared to a two-dimensional (2D) simulation. A fast 2D coupling-of-modes (COM) model was developed to predict, identify and implement strategies to suppress the fine-frequency modes.
The goal of this project was to develop high frequency quality factor (fQ) product acoustic resonators matched to a standard RF impedance of 50 {Omega} using overmoded bulk acoustic wave (BAW) resonators. These resonators are intended to serve as filters in a chip scale mechanical RF spectrum analyzer. Under this program different BAW resonator designs and materials were studied theoretically and experimentally. The effort resulted in a 3 GHz, 50 {Omega}, sapphire overmoded BAW with a fQ product of 8 x 10{sup 13}, among the highest values ever reported for an acoustic resonator.
In this work, we demonstrated engineered modification of propagation of thermal phonons, i.e. at THz frequencies, using phononic crystals. This work combined theoretical work at Sandia National Laboratories, the University of New Mexico, the University of Colorado Boulder, and Carnegie Mellon University; the MESA fabrication facilities at Sandia; and the microfabrication facilities at UNM to produce world-leading control of phonon propagation in silicon at frequencies up to 3 THz. These efforts culminated in a dramatic reduction in the thermal conductivity of silicon using phononic crystals by a factor of almost 30 as compared with the bulk value, and about 6 as compared with an unpatterned slab of the same thickness. This work represents a revolutionary advance in the engineering of thermoelectric materials for optimal, high-ZT performance. We have demonstrated the significant reduction of the thermal conductivity of silicon using phononic crystal structuring using MEMS-compatible fabrication techniques and in a planar platform that is amenable to integration with typical microelectronic systems. The measured reduction in thermal conductivity as compared to bulk silicon was about a factor of 20 in the cross-plane direction [26], and a factor of 6 in the in-plane direction. Since the electrical conductivity was only reduced by a corresponding factor of about 3 due to the removal of conductive material (i.e., porosity), and the Seebeck coefficient should remain constant as an intrinsic material property, this corresponds to an effective enhancement in ZT by a factor of 2. Given the number of papers in literature devoted to only a small, incremental change in ZT, the ability to boost the ZT of a material by a factor of 2 simply by reducing thermal conductivity is groundbreaking. The results in this work were obtained using silicon, a material that has benefitted from enormous interest in the microelectronics industry and that has a fairly large thermoelectric power factor. In addition, the techniques and scientific understanding developed in the research can be applied to a wide range of materials, with the caveat that the thermal conductivity of such a material be dominated by phonon, rather than electron, transport. In particular, this includes several thermoelectric materials with attractive properties at elevated temperatures (i.e., greater than room temperature), such as silicon germanium and silicon carbide. It is reasonable that phononic crystal patterning could be used for high-temperature thermoelectric devices using such materials, with applications in energy scavenging via waste-heat recovery and thermoelectric cooling for high-performance microelectronic circuits. The only part of the ZT picture missing in this work was the experimental measurement of the Seebeck coefficient of our phononic crystal devices. While a first-order approximation indicates that the Seebeck coefficient should not change significantly from that of bulk silicon, we were not able to actually verify this assumption within the timeframe of the project. Additionally, with regards to future high-temperature applications of this technology, we plan to measure the thermal conductivity reduction factor of our phononic crystals as elevated temperatures to confirm that it does not diminish, given that the nominal thermal conductivity of most semiconductors, including silicon, decreases with temperature above room temperature. We hope to have the opportunity to address these concerns and further advance the state-of-the-art of thermoelectric materials in future projects.
This work demonstrates a lateral overtone bulk acoustic resonator (LOBAR), which consists of an aluminum nitride (AlN) transducer coupled to a suspended thin silicon carbide (SiC) film fabricated using standard CMOS-compatible processes. The LOBAR design allows for high transduction efficiency and quality factors, by decoupling the transduction and energy storage schemes in the resonator. The frequency and bandwidth of the resonator were lithographically defined and controlled. A LOBAR operating at 2.93GHz with a Q greater than 100,000 in air was fabricated and characterized, having the highest reported f×Q product of any acoustic resonator to date.
This paper demonstrates silicon carbide phononic crystal cavities for RF and microwave micromechanical resonators. We demonstrate design, fabrication, and characterization of Silicon Carbide/air phononic crystals used as Bragg acoustic mirrors to confine energy in a lateral SiC cavity. Aluminum nitride transducers drive and sense SiC overtone cavities in the 2-3GHz range with fxQ products exceeding 3×1013 in air. This approach enables decoupling of the piezoelectric AlN material from the SiC cavity, resulting in high Q resonators at microwave frequencies. The SiC cavities are fabricated in a CMOS-compatible process, enabling integration with wirelesss communication systems.
Recently reported narrow bandwidth, <;2%, aluminum nitride microresonator filters in the 100-500 MHz range offer lower insertion loss, 100x smaller size, and elimination of large external matching networks, when compared to similar surface acoustic wave filters. While the initial results are promising, many microresonators exhibit spurious responses both close and far from the pass band which degrade the out of band rejection and prevent the synthesis of useful filters. This paper identifies the origins of several unwanted modes in overtone width extensional aluminum nitride microresonators and presents techniques for mitigating the spurious responses.
A two-dimensional phononic crystal (PnC) that can operate in the GHz range is created in a freestanding silicon substrate using NanoFIBrication (using a focused ion beam (FIB) to fabricate nanostructures). First, a simple cubic 6.75 x 6.75 ?m array of vias with 150 nm spacing is generated. After patterning the vias, they are backfilled with void-free tungsten scatterers. Each via has a diameter of 48 nm. Numerical calculations predict this 2D PnC will generate a band gap near 22 GHz. A protective layer of chromium on top of the thin (100 nm) silicon membrane confines the surface damage to the chromium, which can be removed at a later time. Inspection of the underside of the membrane shows the vias flaring out at the exit, which we are dubbing the 'trumpet effect'. The trumpet effect is explained by modeling the lateral damage in a freestanding membrane.
An AlN MEMS resonator technology has been developed, enabling massively parallel filter arrays on a single chip. Low-loss filter banks covering the 10 MHz--10-GHz frequency range have been demonstrated, as has monolithic integration with inductors and CMOS circuitry. The high level of integration enables miniature multi-bandm spectrally aware, and cognitive radios.
In this work we describe a new parallel lattice (PL) filter topology for electrically coupled AlN microresonator based filters. While 4th order, narrow percent bandwidth (0.03%) parallel filters based on high impedance (11 kΩ) resonators have been previously demonstrated at 20 MHz [1], in this work we realize low insertion loss PL filters at 400-500 MHz with termination impedances from 50 to 150 Ω and much wider percent bandwidths, up to 5.3%. Obtaining high percent bandwidth is a major challenge in microresonator based filters given the relatively low piezoelectric coupling coefficients, kt2, when compared to bulk (BAW) and surface (SAW) acoustic wave filter materials.
This paper reports post-CMOS compatible aluminum nitride dual mode resonator filters that realize 4th order band-pass filters in a single resonator device. Dual mode filters at 106 MHz operating in their fundamental mode are reported with insertion losses as low as 5.5 dB when terminated with 150 Ω. A notching technique is demonstrated for varying the 3 dB bandwidth of these filters from 0.15 to 0.7%, overcoming a significant limitation of previous work. Dual mode filters operating at their 5th and 10th overtones are reported scaling the operating frequencies of this class of device to 0.55 and 1.1 GHz.
This work presents a new type of MEMS resonator based on launching an acoustic wave around a ring. Its maximum frequency is set by electrode spacing and can therefore provide a means for developing resonators with center frequencies in the GHz. In addition since the center frequency is dependent on the average radius it is not subject to lithographic process variations in ring width. We have demonstrated several Ring Waveguide (RWG) Resonators with center frequencies at 484 MHz and 1 GHz. In addition we have demonstrated a 4th order filter based on a RWG design.
This work utilized advanced engineering in several fields to find solutions to the challenges presented by the integration of MEMS/NEMS with optoelectronics to realize a compact sensor system, comprised of a microfabricated sensor, VCSEL, and photodiode. By utilizing microfabrication techniques in the realization of the MEMS/NEMS component, the VCSEL and the photodiode, the system would be small in size and require less power than a macro-sized component. The work focused on two technologies, accelerometers and microphones, leveraged from other LDRD programs. The first technology was the nano-g accelerometer using a nanophotonic motion detection system (67023). This accelerometer had measured sensitivity of approximately 10 nano-g. The Integrated NEMS and optoelectronics LDRD supported the nano-g accelerometer LDRD by providing advanced designs for the accelerometers, packaging, and a detection scheme to encapsulate the accelerometer, furthering the testing capabilities beyond bench-top tests. A fully packaged and tested die was never realized, but significant packaging issues were addressed and many resolved. The second technology supported by this work was the ultrasensitive directional microphone arrays for military operations in urban terrain and future combat systems (93518). This application utilized a diffraction-based sensing technique with different optical component placement and a different detection scheme from the nano-g accelerometer. The Integrated NEMS LDRD supported the microphone array LDRD by providing custom designs, VCSELs, and measurement techniques to accelerometers that were fabricated from the same operational principles as the microphones, but contain proof masses for acceleration transduction. These devices were packaged at the end of the work.
The goal of this LDRD project was to evaluate the possibilities of utilizing Stochastic resonance in micromechanical sensor systems as a means for increasing signal to noise for physical sensors. A careful study of this field reveals that in the case of a single sensing element, stochastic resonance offers no real advantage. We have, however, identified a system that can utilize very similar concepts to stochastic resonance in order to achieve an arrayed sensor system that could be superior to existing technologies in the field of inertial sensors, and could offer a very low power technique for achieving navigation grade inertial measurement units.