In this paper, the compensation structure assisted the convex corner structures etch in inductively coupled plasma reactive ion etch (ICP-RIE) have been studied. The convex corner structures are widely used in many applications like micro optical devices or micro sensors. There are many researches to discuss the convex corner structure under wet etch, but rare researches for dry ICP-RIE etching. In anisotropic silicon etching, under the Bosch patent, sequentially alternating etch and passivation cycles can easily achieve high aspect ratio silicon structures. The feature size of the convex corner structures is difficult to maintain as original design at the bottom position in deep etch, due to some non-vertical movement plasma. The non-vertical movement plasma caused by collision between plasma ions, pollutants or rebounded from the etching mask. The compensation structure is design in front of the convex corner structure. The compensation structures can obstruct the non-vertical plasma to etch the convex corner structure and reduce the etch lag effect during the etch process leading to better profile at deep etch. The current study systematically investigates plasma condition to verify feasibility of the proposed method, and discusses effect of the gap between compensation structure and convex corner structure at three different gaps of 15, 10, 5µm. It demonstrate the convex corner structure have better profile with compensation structure at 5µm gap than other at deep ICP-RIE etching.

Keywords: Convex corner compensation, Inductively-Coupled-Plasma Reactive-Ion-Etch (ICP-RIE), High-aspect-ratio structure

The 1GHz lateral-field CMRs consist of a thin film of AlN patterned with interdigitated metal electrodes on top and a floating electrode on bottom. An AC signal to the electrodes excites the contour-extensional mode through the equivalent d31 piezoelectric coefficient of the AlN [4,5]. The devices are characterized by mechanical and electrical measurements. A Polytec UHF120 LDV captures phase and out-of-plane displacement data, with ±1.4pm resolution down to a noise floor of 4pm (Fig. 1). Spatial scans across the surface generate 3D mode shapes (Fig. 2). By performing frequency sweeps and fitting data to the equations of motion, we extract mechanical parameters such as the linear and nonlinear stiffness and damping coefficients. Electrical measurements further validate simulation results; through the electromechanical coupling coefficient the mechanical measurements can also be confirmed.

The 3D FEA models are developed in COMSOL and used to determine both mechanical and electrical responses. We compare simulated frequency response, out-of-plane displacement, and Q with experimental data to verify the models (Table 1). Fig. 2 shows the 3D mode shape, and the out-of-plane displacement profile of individual fingers is shown in Fig. 3. The maximum seen in experiments is 119.5pm, which agrees well with the simulated value of 117pm. To compare mechanical response, the wavelength λ in the lateral direction is studied. Simulations show a 3.2um and 7.9um wavelength at resonance. Experimentally, a 4um and an 8um wavelength appear. The multiple wavelengths at resonance are the result of a mismatch in acoustic velocity between AlN and Al and their existence in the simulation serves as further validation. Simplified five-finger device simulations show good fit with experimental admittance (Fig. 4). Full 33-finger device simulations, currently in progress, will improve this fit.

The experimental data gives further insight into the operation of the devices and is essential to the verification the 3D FEA models. With confirmed accuracy, these models can be used for predictive modeling of GHz-range CMRs. This allows optimization of geometric parameters such as electrode spacing and anchor performance.

The objective of this work is to develop a high frequency MEMS-based mass sensor capable of pushing the limits of sensitivity for the detection of explosive materials. The specific objective is to utilize a nonlinear microcantilever mass sensor functionalized with xerogel-based molecularly imprinted polymers (MIPs) for the selective detection of DNT. This paper reports the implementation of a novel control method for sensing which is based on the phase squeezing phenomenon present in the nonlinear dynamics of a parametrically driven microcantilever. It is expected to significantly improve the mass detection limits compared to bifurcation tracking [1], as well as reduce measurement time by over three orders of magnitude. Sensor speed, sensitivity, and selectivity are all important for ppt level DNT/TNT detection.

Previous work focuses on tracking the resonance frequency shift of the microcantilever due to DNT absorption using bifurcation tracking [1]. Bifurcation sensing is achieved by sweeping the frequency or voltage until a jump event occurs, and tracking is done by repeating the process to detect the frequency shift. However, bifurcation tracking is a time-consuming process which requires time to reset the system back to an appropriate initial condition after each jump event to avoid hysteresis before the next bifurcation occurs. This work reports a new method of sensing which concentrates on monitoring changes in the phase response as the device approaches the bifurcation point. Just prior to bifurcation, noise squeezing occurs due to a slowing down of the response component associated with the bifurcating eigenvalue near the bifurcation point. The statistical variance of the phase response serves as a precursor to activate the feedback control scheme, which employs frequency modulation to stabilize a parametrically-excited sensor at the edge of instability. By maintaining close proximity to the bifurcation, but not allowing the large amplitude growth necessary for existing bifurcation tracking methods, over three orders of magnitude measurement time improvements in the acquisition rate near the location of the bifurcation point can be made [2]. Initial results show that the controller is capable of tracking the resonance frequency based on the phase variance of the microcantilever with close proximity to the edge of instability and control parameters can be tuned to optimize the sensitivity of the sensor. DNT gas sensing using the controller is yet to be experimented and the sensitivity of noised squeezing bifurcation sensing of nonlinear MEMS microcantilever is expected to be presented at the conference.

Solder based self assembled (SBSA) structures are formed by transforming 2D patterns to 3D structures. SBSA processing involves conventional lithography, metal deposition, and etching methods in addition to dip soldering and solder reflow. The processing involves free metals around a fixed metal, when reflowed; the free metal pattern is rotated towards the fixed metal pattern till the solder reaches minimum surface energy. In previous work, we investigated two types of soldering: face and edge-soldered SBSA structures. The 3D structure produces an excess solder when the amount of the solder deposited is more than the 3D volume. The excess solder which is measured from the top of the metal structure to the top of the solder in face-soldered SBSA structures is known as solder standoff height (SSH). SSH is envisioned as being useful as a solder bump that could be used to stack hybrid layers in 3D integration schemes. Experimentally, SSH was found to be heat resistant. The simulation study we performed reflects the effect of processing parameters such as gap-size between metal patterns, solder thickness, and solder coverage on the formation of 3D structures. Formation of 3D structures is influenced by the tilt of free metal towards the fixed metal. Hence tilt angle is considered as one of the output parameters. The processing parameters are analyzed for truncated square pyramid (TSP) and open cube (OC) structures using an open source simulation tool, Surface Evolver.

Surface Evolver is an interactive program for the study of surfaces shaped by surface tension. Simulation results show that as the gap size decreases, the minimum solder energy is attained at lower tilt angles for face-soldered TSP and OC structures. Edge-soldered OC structures follow a similar trend. The tilt angle remains constant for a range of gap sizes for edge-soldered TSP structures. Varying solder thickness varies the minimum energy tilt angle for face-soldered structures. For edge-soldered structures, tilt angle remains constant irrespective of solder thickness. Optimum tilt angle remains constant for solder coverage of 70 % and higher in both TSP and OC structures.