Search

Residual stress gradients often negatively affect the performance of MEMS devices, causing film curvature and changing the designed gaps of released structures. In this work, we built folded beams designed to compensate for the film curvature and keep the actuator gaps of sensitive resonant switches constant. While the average stress gradient is cancelled by our designs, we find that random variations in the stress gradient (rather than random variations in device dimensions) cause the majority of the observed variation in actuator gap. To our knowledge, this has not previously been reported, and represents an important limitation on MEMS designs using sputtered films. The standard deviation of the 400 nm contact gap for a folded beam of total length 152 µm and width 108 µm was measured to be about 134 nm. Using parameters measured from test cantilevers, our simulations predict that about 98% of the variation in contact gap is due to stress gradient variation, rather than variations in device geometry.

In this paper we present a 3D nonlinear dynamic model which describes the transient mechanical analysis of an ohmic contact RF MEMS switch, using finite element analysis in combination with the finite difference method. The model includes real switch geometry, electrostatic actuation, the two-dimensional non-uniform squeeze-film damping effect, the adherence force, and a nonlinear spring to model the interaction between the contact tip and the drain. The ambient gas in the package is assumed to act as an ideal and isothermal fluid which is modeled using the Reynolds squeeze-film equation which includes compressibility and slip flow. A nonlinear contact model has been used for modeling contact between the microswitch tip and the drain electrode during loading. The Johnson–Kendall–Roberts (JKR) contact model is utilized to calculate the adherence force during unloading. The developed model has been used to simulate the overall dynamic behavior of the MEMS switches including the switching speed, impact force and contact bounce as influenced by actuation voltage, damping, materials properties and geometry. Meanwhile, based on a simple undamped spring–mass system, a dual voltage-pulse actuation scheme, consisting of actuation voltage (Va), actuation time (ta), holding voltage (Vh) and turn-on time (ton), has been developed to improve the dynamic response of the microswitch. It is shown that the bouncing of the switch after initial contact can be eliminated and the impact force during contact can be minimized while maintaining a fast close time by using this open-loop control approach. It is also found that the dynamics of the switch are sensitive to the variations of the shape of the dual pulse scheme. This result suggests that this method may not be as effective as expected if the switch parameters such as threshold voltage, fundamental frequencies, etc. deviate too much from the design parameters. However, it is shown that the dynamic performance may be improved by increasing the damping force. The simulation results obtained from this dynamic model are confirmed by experimental measurement of the RF MEMS switches which were developed at the Northeastern University. It is anticipated that the simulation method can serve as a design tool for dynamic optimization of the microswitch. In addition, the approach of tailoring actuation voltage and the utilization of squeeze-film damping may provide further improvements in the operation of RF MEMS switches.

We have designed and characterized a MEMS-based Fabry–Perot device (MFPD) to measure vibration at high temperatures. The MFPD consists of a micromachined cavity formed between a substrate and a top thin film structure in the form of a cantilever beam. When affixed to a vibrating surface, the amplitude and frequency of vibration are determined by illuminating the MFPD top mirror with a monochromatic light source and analyzing the back-reflected light to determine the deflection of the beam with respect to the substrate. Given the device geometry, a mechanical transfer function is calculated to permit the substrate motion to be determined from the relative motion of the beam with respect to the substrate. Because the thin film cantilever beam and the substrate are approximately parallel, this two-mirror cavity arrangement does not require alignment or sophisticated stabilization techniques. The uncooled high-temperature operational capability of the MFPD provides a viable low-cost alternative to sensors that require environmentally controlled packages to operate at high temperature. The small size of the MFPD (85–175 µm) and the choice of materials in which it can be manufactured (silicon nitride and silicon carbide) make it ideal for high-temperature applications. Relative displacements in the sub-nanometer range have been measured and close agreement was found between the measured sensor frequency response and the theoretical predictions based on analytical models.

As an essential part of an optical imager project, there was the need for a very high aspect ratio MEMS optical scanning mirror (5 mm ? 150 ?m clear aperture), capable of a ?2? sweep at 1 kHz with an applied voltage less than 200 V. This paper reports on the design, fabrication, modeling and testing of such an electrostatically actuated MEMS mirror. Fabrication involves using a 1?0?0 n-type double side polished silicon wafer, along with surface and bulk micromachining techniques, to produce a mirror with nickel torsion springs and nickel electrostatic actuators. There were 8 masks and 20 processing steps required. The performance of these devices was measured and found to be within the required specifications. Analysis involved developing models to predict the dynamic behavior of these MEMS micromirrors. A basic parallel plate capacitor model was adjusted with finite element analysis to account for fringing fields. Young's modulus of the electroplated nickel was determined to be 110 GPa from a comparison of the model with the results of dynamic testing, and the residual strain test structures led to a value of 0.0029 for the residual strain. Each of these values is within the rather wide range of published values. Once these values were determined, the model agreed very well with the measurements of the dynamic angular response of the mirror.

A technique for performing nanomechanical testing of free-standing structures using contact scanning mode atomic force microscopy (AFM) has been developed and implemented. The technique consists of performing a contact mode scan of an area of a cantilever test structure as well as of an area that does not deform. The use of two different values of the force during the scan allows the effect of the topography of the test structure to be eliminated. Also note that in this constant force scan, the bending moment in the test cantilever changes continuously with the position of the applied force, allowing both the stress and strain to change even though the force remains constant. With the dimensions of the structure known, material properties (i.e. the elastic modulus) can be determined from the relationship between the applied force and the structural deflection. This method is not susceptible to nonlinearities in the AFM photodetector and allows for piezo drift during testing to be quantified and corrected. 50 nm thick chromium cantilever structures were fabricated for testing using electron-beam lithography. The determination of the elastic modulus gave consistent results as the AFM tip was scanned along the cantilever. As is the case with AFM fixed position force–displacement measurements, the accuracy of the results is affected by uncertainties in the measured AFM cantilever stiffness and by uncertainties in the test structure dimensions. This testing method can be performed by any AFM capable of scanning in the contact mode without requiring specialized software. Fabrication of the cantilevers requires only a single-mask process.

The ion energy distribution during plasma immersion ion implantation (PIII) has been computed including charge transfer collisions in a dynamic sheath. The boron ion energy distribution has been derived experimentally from sheet resistance and SIMS measurements. The experimental results are consistent with theoretical predictions. Both the experimental and theoretical results indicate that during the PIII doping experiments boron ions have a relatively wide energy distribution and that many of the ions have low energies because of collisions in the sheath.

One of the most important methods for selective and repeatable assembly of carbon nanotubes (CNTs) is alternating current dielectrophoresis (DEP). This method has been demonstrated experimentally as a viable technique for nano-scale manufacturing of novel CNT based devices. Previous numerical analyses have studied the motion of nanotubes, the volume from which they are assembled, and the rate of assembly, but have been restricted by various simplifying assumptions. In this paper we present a method for simulating the motion and behavior of CNTs subjected to dielectrophoresis using a three-dimensional electrostatic finite element analysis. By including the CNT in the finite element model, we can accurately predict the effect of the CNT on the electric field and the resulting force distribution across the CNT can be determined. We have used this information to calculate the motion of CNTs assembling onto the electrodes, and show how they tend to move towards the center of an electrode and come into contact at highly skewed angles. Our analysis suggests that the CNTs move to the electrode gap only after initially contacting the electrodes. We have also developed a model of the elastic deformation of CNTs as they approach the electrodes demonstrating how the induced forces can significantly alter the CNT shape during assembly. These results show that the CNT does not behave as a rigid body when in close proximity to the electrodes. In the future this method can be applied to a variety of real electrode geometries on a case-by-case basis and will provide more detailed insight into the specific motion and assembly parameters necessary for effective DEP assembly.

Using an AFM-based test setup, experiments were performed on Ru microcontacts under a variety of leading and trailing edge hot switching conditions, including different voltages, different currents, different polarities (including bipolar and ac up to 20?MHz), and different approach and separation rates. It was found that hot switching damage is a complex phenomenon for microcontacts. It consists of a number of different mechanisms occurring simultaneously to different degrees depending on the hot switching conditions. It was determined through a combination of experiments and models that the mechanisms leading to contact erosion operate when the electrodes are separated by less than a few ? or are barely touching. For leading edge hot switching, i.e. hot switching when the contacts are closing, the main damage mechanism was found to be associated with currents less than 0.15?mA. Pre-contact currents were observed on uncleaned contacts and were not found to contribute to contact damage. Despite the damage caused by hot switching, it was found that unless the contact material is almost or completely eroded, hot switching does not lead to high contact resistance or high adhesion on Ru contacts. Under bipolar hot switching conditions, microcontacts with a 400??N contact force maintained a contact resistance of less than 1?? and a pull-off force less than 60??N for more than 100 million cycles.

Carbon nanotubes are desirable components of nanoelectromechanical (NEM) devices due to their excellent mechanical and electrical properties. In this study, dielectrophoresis, a potential high-rate nanomanufacturing process, was used to assemble single-walled carbon nanotube (SWCNT) bundles suspended over a trench. The intent was to assemble a single SWCNT bundle between two electrodes. However, it was observed that when two or more SWCNT bundles assembled across the trench, the bundles were attached together in a portion of the suspended section. This study models the separation and re-adhesion processes of two adhered SWCNT bundles as their internal tensions are varied using an atomic force microscope (AFM) tip. Two devices were selected with distinct SWCNT bundles. Observation of the force–distance measurements through applying an AFM tip at the middle of the suspended SWCNT bundles, in conjunction with continuum mechanics modelling, allowed the work of adhesion between the two nanotube bundles to be determined. As the force was applied by the AFM tip, the tension induced in each bundle increases sufficiently to partially overcome the adhesion between the bundles, thereby decreasing the adhesive length. The adhesive length then recovers due to the decrease in the induced tension during the unloading process. The average value of the work of adhesion between two adhered SWCNT bundles was determined to be 0.37 J m−2 according to the experimental data and modelling results.

Since their discovery in 1991, carbon nanotubes (CNTs) have been attractive components for nanoelectromechanical (NEM) devices because of their excellent mechanical and electrical properties. Several CNT-based NEM devices have been reported. Experimental investigations have discovered that the force required for a CNT to slip on a SiO2 substrate is almost constant in the range of contact lengths from 140 to 246?nm and increases with the contact length for greater lengths. This study develops a theory based on the continuum mechanics to explain this observed phenomenon. The critical force needed to cause the CNT to slip over the whole substrate is determined by applying a beam theory which includes the bending, shear, and axial deformations of the CNT. At the stick?slip transition, the work of adhesion contributes a concentrated force and a concentrated moment to produce discontinuities in the internal forces and moments of the CNT. The modelling results obtained here provide the external force required for complete slip versus a dimensionless parameter related to the shear stress and the contact length. A comparison of the modelling result and the experimental data from the literature shows good agreement.

Using a custom built contact testing system, direct current micro contact damage under hot-switching conditions was explored in ruthenium-on-ruthenium contacts operated at a contact force of approximately 400 µN. For the first time, contact damage on making and breaking contact under bias (leading and trailing edge hot switching) is compared. Trailing-edge hot switching leads to significantly higher adhesion (1.5–4 times higher) than leading-edge hot switching. The high-voltage tests (3.5 V) lead to polarity-dependent material transfer, with material moving in the direction of the electric field. The amount of material transfer does not depend strongly on the current limit from 0.78 to 380 mA. The low voltage (0.71 V) tests result in much less damage, and the material transfer does not have a clear directionality. However, the amount of damage does increase significantly as the current limit is increased from 16 to 78 mA. Also observed for the first time is a new type of high-current, short duration current spike associated with hot switching events at voltages above 1.5 V. The fact that these spikes occur at the higher voltage but not at the lower voltage suggests (but does not prove) that they are associated with the polarity dependent material transfer.

An improved scanning probe microscope based contact tester has been designed, constructed and used for cyclic testing of various metal contacts. The tester is designed for contact material evaluation, especially for metal-contact MEMS switch applications, and is capable of simultaneous measurements of contact force and resistance. The tester uses a specially designed silicon force sensor with an integrated contact bump and a mating silicon pillar in order to simulate switch operation. The sensor and the pillar are coated with contact materials, allowing a wide range of contact metals and metal pairs to be evaluated. The testing takes place within a custom-built test chamber in which both plasma and UV-ozone treatments are available for contact cleaning and surface modification. It was found that a dissimilar contact pair of Au?Ru with O2?plasma cleaning can provide low contact adhesion and low resistance as compared with Au, Ru and Ir contact pairs. Layered structures with a thin layer of Ru on top of Au were also modeled and tested. Ru layers between 50 and 100 nm were effective in reducing adhesion and contact resistance provided the contact force is greater than 300 ?N.

Several electromechanical devices, each consisting of a small bundle of single-walled carbon nanotubes suspended over an actuation electrode, have been fabricated and operated electrically. The nanotubes are assembled on the electrodes using dielectrophoresis, a potential high-rate nanomanufacturing process. A large decrease in the threshold voltage was seen after the first actuation. This is a result of the nanotubes sliding inward on their supports as they are pulled down toward the actuation electrode, leaving slack in the nanotube bundle for subsequent actuations. The electrical measurements agree well with an electromechanical model that uses a literature-reported value of the shear stress between the nanotubes and the SiO2 shoulders. Electrical measurements were performed in dry nitrogen as a large build-up of contamination was seen when the measurements were performed in lab air. We present measurements as well as a detailed mechanics model that support the interpretation of the data.