We demonstrate the consistent and manageable nature of surface adhesion and stiction forces in MEMS devices fabricated using the high-temperature epitaxial encapsulation process. In this encapsulation process (commercialized by SiTime), there are no chemical anti-stiction films or getters. Data from more than 2000 test structures with more than 80 design variations from three different fabrication runs were gathered in this study. Surprisingly, the adhesion force is shown to be independent of design geometry. The measured adhesion forces (18-25uN) are small enough for inertial sensors. In addition, we demonstrate anti-stiction bump stops with springs for a sliding contact, which reduce the probability of stiction by over 50%.
Research on MEMS Resonators began over 50 years ago. In just the last 10 years, there has been a series of important technological developments, and (finally!) success at commercialization. The presentation will highlight some key milestones along this path, describe some of the critical technology steps, and outline some of the important non-technological events within SiTime - all of these factors contributed to the successful outcome.
This paper presents a capacitive X/Y and Z-axis accelerometer, pressure sensor, and resonant thermometer, co-fabricated in a single die using an ultra-clean, high-temperature, wafer-scale, production-compatible encapsulation process. This process is free of bond rings and getter areas, thereby reducing overall die size to a minimum. Utilizing a process that allows for nitride and silicon etch stops, we show that it is possible to design high-sensitivity pressure sensors and accelerometers with very little cross-sensitivity. In addition, remaining sensitivities to environmental effects, such as temperature, can be compensated to reveal a suite of on-chip high-performance sensors that is accurate over temperature. All of this is accomplished in a process similar to current high-volume production processes in industry.
This paper examines previously unmeasured properties of atomic layer deposited Pt films relevant to sensor design, including wafer-scale uniformity, 1/f noise, Young's modulus, and the dependence of all these parameters on ALD processing.We observe an increase in the 1/f noise Hooge parameter of nearly two orders of magnitude with decreasing thickness, which introduces constraints on the design of ALD Pt sensors with low noise requirements.Additionally, we measure the variation in waferscale electrical uniformity and Young's modulus with thickness, highlighting tradeoffs associated with integrating ALD Pt films into wafer-scale processing to achieve freestanding structures with high yield.
In this work we present migration phenomenon of throughholes in silicon membranes.The sealing of through-holes in hydrogen ambient at high temperature (1130°C) with various dimensions and annealing time durations was investigated in both singlecrystalline silicon (sc-Si) and poly-crystalline silicon (poly-Si) membranes.The sealing process in silicon was observed as highly dependent on local crystal grain geometry, leading to more distributed, unpredictable migration rates and shape evolutions in poly-Si compared to holes in sc-Si.These findings can be leveraged in fabrication processes that require a balance between silicon migration and deposition.
We report on the modeling, fabrication and characterization of a new serpentine geometry for platinum bolometers realized with plasma-enhanced atomic layer deposition (PEALD). The measured responsivity of 7.5 · 107 VW-1 A-1 of the proposed structure is almost four times higher than that of previously published PEALD deposited platinum bolometers with a traditional square absorber design [1]. This improvement was achieved without increasing the sensor's footprint or fabrication complexity.
We provide a cost calculation of running a pair of example nanofabrication processes in the Stanford Nanofabrication Facility and the University of California, Berkeley's Marvell Nanofabrication Laboratory. Cost analysis for a MEMS membrane process and a photonic waveguide process for each facility will be shown at the symposium. In this abstract the outline for the MEMS membrane process at SNF is provided as an example.
Epitaxially grown polycrystalline silicon (epi-poly) has shown great promise as a MEMS material, offering isotropic material properties with minimal residual stress and stress gradients.While epi-poly has been used previously for MEMS inertial systems or sensors, its use in high precision resonator applications has been restricted to an encapsulation layer for resonators fabricated in single crystal silicon.In this work, encapsulated resonators fabricated with epi-poly as the functional layer were tested alongside single crystal silicon devices with regard to resonant frequency, quality factor, temperature dependence, and stability.Test results indicate that epi-poly resonators have very similar mechanical properties and stability to single crystal silicon resonators when operated in a clean package.
This paper presents an uncooled infrared bolometer using a metal thin film that is formed by atomic layer deposition (ALD). Nanometer-thick freestanding layers enabled by ALD have the potential to improve the performance of bolometers with achieving low thermal conductance and near optimal optical properties. The fabrication and characterization of the first implementation are described as well as the electrical properties of ALD platinum films.
A method of using electrostatic tuning to compensate the phase noise due to external vibrations in a MEMS oscillator is presented in this paper. An accelerometer measures the acceleration applied to a resonator, and a compensation signal generated by the accelerometer is added to the bias voltage. We achieve 91% reduction of the acceleration sensitivity for sinusoidal accelerations from 100 Hz to 300 Hz using a double-ended tuning fork resonator. This is the first demonstration of active acceleration compensation for MEMS resonators.
This paper investigates dielectric charge in oxide-coated composite-beam silicon electrostatic resonators.The effects of charge in the oxide layers on resonant frequency are theoretically modeled.Evidence for the presence of both fixed and mobile charge is presented through experimental studies.It is shown that the motion of charge is controllable through the bias voltage applied to the resonator.These studies are relevant for improving the reliability of non-contacting MEMS where dielectrics may be present such as timing references, inertial sensors, and optical actuators.
In this work, we study the effect of non-isothermal operation on beam-type micromechanical resonators. A double ended tuning fork (DETF) resonator was studied under a linearly varying temperature profile along the length of the beam. We present models which show that for a constant center temperature of the beams, increasing the linear temperature gradient along the DETF substantially increases the quality factor (Q), with marginal effect on frequency. We also find that it is possible to use non-isothermal operation to improve quality factor in MEMS resonator based ovenized reference oscillators.
MEMS devices must be packaged to be used. Unfortunately, MEMS packages are challenging to develop, and the packaging of MEMS devices often dominates the cost of the product. in recent years, our group has worked with a team from Bosch to develop and demonstrate a novel wafer-scale encapsulation approach for MEMS. This process uses MEMS fabrication steps to build the device and the package at the same time. The main advantage of this approach is that the wafers emerge from the fabrication facility with all the fragile MEMS structures completely buried within the wafer, allowing all existing standard handling and packaging approaches, such as wafer-dicing, pick/place, and injection mold packaging to be used. This encapsulation process enables CMOS integration, embedding, and extreme miniaturization of complete systems. In this paper, we describe some advantages for performance, size and cost that can come from this approach.
Micromechanical Resonator based oscillators are a promising technology for replacing quartz crystal based oscillators. In this work, we will report the effects of mechanical vibrations and bias voltage noise on the phase noise performance of electrostatic MEMS resonator based oscillators. Accurate models for both these effects are discussed along with their experimental verification using a 1.3MHz, epi-silicon encapsulated Single Anchored Double Ended Tuning Fork (DETF) resonator. The acceleration sensitivity of the resonator was found to be < 10ppb/g which is better than many low cost crystal resonators, and shows potential for improvement to get performance which is at par or better than quartz crystal oscillators.
Silicon MEMS resonators have great potential for on-chip high frequency signal applications. This paper compares methods of sensing the temperature of an encapsulated silicon MEMS resonator and using this temperature measurement to stabilize the temperature, and hence the resonant frequency, of the resonator. The use of external Pt RTDs, integrated Si thermistors, and the use the Quality factor (Q) of the resonator are explored. Use of the Q as a temperature sensor is explored in detail as it is a nearly ideal temperature sensing method. Characterisations of the temperature sensors and preliminary temperature control results are presented.
Encapsulated micromechanical resonators have been fabricated from single crystal silicon and a method for stabilizing their resonant frequency over temperature has been demonstrated.The method uses high-frequency AC signals applied directly to the resonant structure in order to heat it.By adjusting the power of the heating signal, varying amounts of heat could be added to the resonant structure in order to compensate for the effects of changing external temperature and the variation in resonant frequency was reduced significantly.This method requires minimal changes to existing resonator designs and is applicable to any resonator with a current path through its structure.The resonators examined here exhibited temperature coefficients of resonant frequency of over 210 Hz/°C (170 ppm/°C) without temperature compensation.With compensation, the temperature coefficient was reduced to less than 5 Hz/°C (4 ppm/°C).
Micromachined accelerometers have been one of the most successful MEMS devices. Once a successful MEMS device is developed, packaging is the most critical step commercializing the device. Micromachined accelerometers are currently packaged with a separate lid made of glass or silicon bonded on the accelerometer device wafer. In this study, we have developed a new advanced packaging scheme using a thick film epitaxial grown polysilicon encapsulation. With this method, we can reduce the total die size and reduce an additional "cap" wafer; both result in cost reduction. Furthermore the encapsulation can withstand conventional post processing, such as dicing, wire bonding and even injection molding. This paper will discuss the packaging process and show performance results from encapsulated piezoresistive and capacitive accelerometers.
MEMS packaging has always been a field of great importance since it can dominate the cost and size of a final working device. Considering this, we have concentrated on developing a wafer-scale encapsulation scheme which uses a thick epi-poly (epitaxially deposited poly silicon) layer as the sealing layer. This approach allows the use of conventional post processing, such as dicing, wire bonding, and other standard handling and mounting techniques. We also can minimize the chip area used for packaging, in some cases reducing the chip size by ×5 from what was required for silicon fusion bonded covers. This packaging scheme can be used for various MEMS devices and can be integrated with other electronics. This paper will discuss the packaging process and show some preliminary results.
Micromechanical resonators with resonant frequencies from 500 kHz to 10 MHz were built and examined for several energy loss mechanisms. Thermoelastic damping, clamping loss and air damping were considered. The devices were shown to be limited by thermoelastic damping, providing experimental verification of this phenomenon at the microscale. Resonators with scaled dimensions also matched well with scaling theory of damping at a given pressure. An energy loss mechanism other than thermoelastic dissipation, most likely clamping loss, was shown to be dominant for resonators whose ratio of length to width was less than 10:1. The devices were fabricated using a single-wafer encapsulation process.