Technological advancement has thrust MEMS design and fabrication into the forefront of modern technologies. It has become sufficiently self-sustained to allow mass production. The limiting factor which is stalling commercialization of MEMS is the packaging and device reliability. The challenging issues with MEMS packaging are application specific. The function of the package is to give the MEMS device mechanical support, protection from the environment, and electrical connection to other devices in the system. The current state of the art in MEMS packaging transcends the various packaging techniques available in the integrated circuit (IC) industry. At present the packaging of MEMS includes hermetic ceramic packaging and metal packaging with hermetic seals. For example the ADXL202 accelerometer from the Analog Devices. Study of the packaging methods and costs show that both of these methods of packaging are expensive and not needed for majority of MEMS applications. Due to this the cost of current MEMS packaging is relatively high, as much as 90% of the finished product. Reducing the cost is therefore of the prime concern. This Thesis explores the possibility of an inexpensive plastic package for MEMS sensors like accelerometers, optical MEMS, blood pressure sensors etc. Due to their cost effective techniques, plastic packaging already dominates the IC industry. They cost less, weigh less, and their size is small. However, porous nature of molding materials allows penetration of moisture into the package. The Thesis includes an extensive study of the plastic packaging and characterization of three different plastic package samples. Polymeric materials warp upon absorbing moisture, generating hygroscopic stresses. Hygroscopic stresses in the package add to the thermal stress due to high reflow temperature. Despite this, hygroscopic characteristics of the plastic package have been largely ignored. To facilitate understanding of the moisture absorption, an analytical model is presented in this Thesis. Also, an empirical model presents, in this Thesis, the parameters affecting moisture ingress. This information is important to determine the moisture content at a specific time, which would help in assessing reliability of the package. Moisture absorption is modeled using the single phase absorption theory, which assumes that moisture diffusion occurs freely without any bonding with the resin. This theory is based on the Fick's Law of diffusion, which considers that the driving force of diffusion is the water concentration gradient. A finite difference simulation of one-dimensional moisture diffusion using the Crank-Nicolson implicit formula is presented. Moisture retention causes swelling of compounds which, in turn, leads to warpage. The warpage induces hygroscopic stresses. These stresses can further limit the performance of the MEMS sensors. This Thesis also presents a non invasive methodology to characterize a plastic package. The warpage deformations of the package are measured using Optoelectronic holography (OEH) methodology. The OEH methodology is noninvasive, remote, and provides results in full-field-of-view. Using the quantitative results of OEH measurements of deformations of a plastic package, pressure build up can be calculated and employed to assess the reliability of the package.

Nesta dissertação, os principais conceitos de MEMS, suas aplicações, processos de fabricação, componentes e sistemas são abordados. O objetivo desta dissertação é o estudo detalhado de chaves MEMS para aplicações em RF, que apresentam bom comportamento em altas freqüências e com potencial de melhoria em sua banda de operação. Em particular, aprofundou-se o estudo para o caso de uma chave MEMS de membrana capacitiva paralela sobre um guia de onda coplanar ou CPW - Coplanar Waveguide. O objetivo foi o de ampliar sua banda de operação, mantendo-se outras especificações inalteradas. Partindo-se de uma chave com banda de operação nula para critérios de perda de retorno e isolação mínimas iguais a 20 dB, com alteração na geometria da chave foi possível obter-se uma banda de 28 GHz e posteriormente ampliá-la para 31 GHz, praticamente sem alteração nas demais características elétricas. / In this thesis, the main concepts of MEMS, their application, fabrication processes, components and systems are addressed. The objective of the thesis is a detailed study of MEMS switches for RF applications, that present good performance at high frequencies and with a potential for bandwidth improvement. More specifically, the study was deeply conducted for shunt capacitive membrane MEMS switches over CPW - Coplanar Waveguide. In this case, the objective was to enlarge the operation bandwidth, keeping the other specifications unchanged. Starting with a switch with null operational bandwidth for criteria of minimum return loss and isolation of 20 dB, after a modification in the switch geometry, it was possible to obtain an operational bandwidth of 28 GHz and then to enlarge it to 31 GHz, keeping almost unchanged the other electric characteristics.

Fabrication technology, based on MEMS processes, for constructing components for use in switched-mode power supplies are developed and presented. Capacitors, magnetic cores, and inductors based on sacrificial multilayer electroplating are designed, fabricated, and characterized. Characterization of the produced inductors' core losses at high frequency and high flux is presented, confirming the aptness of the featured microfabrication processes for reducing eddy current losses in magnetic cores. As well, the demonstration of the same inductors in DC/DC converters at high switching frequencies, up to 6 MHz, is presented. Initial work addressing the top-down development of a fully-integrated DC/DC converter is presented. As well, the comprehensive advancement of the central process - sacrificial multilayer electroplating - is presented, including the development of a second-generation automated multilayer electroplating system. The advanced sacrificial multilayer plating process is applied to produce microfabricated capacitors, which achieved in excess of 1.5 nF/mm² capacitance density, The fabrication of highly-laminated magnetic cores and power inductors based on sacrificial multilayer electroplating is presented, along with the design and development of a system for characterizing inductor behavior at high-frequency, high-flux conditions. The design and operation of both buck and boost DC/DC converters, switching at up to 6 MHz, built around these highly-laminated-core inductors are presented.

Delamination along thin film interfaces is a prevalent failure mechanism in microelectronic, photonic, MEMS, and other engineering applications. Current interfacial fracture test techniques specific to thin films are limited by either sophisticated mechanical fixturing, physical contact near the crack tip, non-representative test specimens, or complicated stress fields. Moreover, these techniques are generally not suitable for investigating fatigue crack propagation under cyclical loading. A fixtureless and noncontact experimental test technique is thus proposed and implemented to study interfacial fracture for thin film systems. The proposed test incorporates permanent magnets surface mounted onto micro-fabricated released thin film structures. An applied external magnetic field induces noncontact monotonic or fatigue loading to initiate delamination along the interface between the thin film and underlying substrate. Characterization of the film deflection, peel angle, and delamination propagation is accomplished through in situ optical techniques. Analytical and finite-element models are used to extract fracture parameters from the experimental data using thin-film peel mechanics. The developed interfacial fracture test has been demonstrated for Cu thin films on a SiO₂/Si substrate.

The scalability and cost of microelectromechanical systems (MEMS) optical switches are now the important factors driving the development of MEMS optical switches technology. The employment of MEMS in the design and fabrication of optical switches through the use of micromachining fabricated micromirrors expands the capability and integrity of optical backbone networks. The focus of this dissertation is on the design, fabrication, and implementation of a new type of MEMS optical switch that combines the advantages of both 2-D and 3-D MEMS switch architectures. This research presents a new digital MEMS switch architecture for 1×N and N×N optical switches. The architecture is based on a new microassembled smart 3-D rotating inclined micromirror (3DRIM). The 3DRIM is the key device in the new switch architectures. The 3DRIM was constructed through a microassembly process using a passive microgripper, key, and inter-lock (PMKIL) assembly system. An electrostatic micromotor was chosen as the actuator for the 3DRIM since it offers continuous rotation as well as small, precise step motions with excellent repeatability that can achieve repeatable alignment with minimum optical insertion loss between the input and output ports of the switch. In the first 3DRIM prototype, a 200×280 microns micromirror was assembled on the top of the electrostatic micromotor and was supported through two vertical support posts. The assembly technique was then modified so that the second prototype can support micromirrors with dimensions up to 400×400 microns. Both prototypes of the 3DRIM are rigid and stable during operation. Also, rotor pole shaping (RPS) design technique was introduced to optimally reshape the physical dimensions of the rotor pole in order to maximize the generated motive torque of the micromotor and minimize the required driving voltage signal. The targeted performance of the 3DRIM was achieved after several PolyMUMPs fabrication runs. The new switch architecture is neither 2-D nor 3-D. Since it is composed of two layers, it can be considered 2.5-D. The new switch overcomes many of the limitations of current traditional 2-D MEMS switches, such as limited scalability and large variations in the insertion loss across output ports. The 1×N MEMS switch fabric has the advantage of being digitally operated. It uses only one 3DRIM to switch the light signal from the input port to any output port. The symmetry employed in the switch design gives it the ability to incorporate a large number of output ports with uniform insertion losses over all output channels, which is not possible with any available 2-D or 3-D MEMS switch architectures. The second switch that employs the 3DRIM is an N×N optical cross-connect (OXC) switch. The design of an N×N OXC uses only 2N of the 3DRIM, which is significantly smaller than the N×N switching micromirrors used in 2-D MEMS architecture. The new N×N architecture is useful for a medium-sized OXC and is simpler than 3-D architecture. A natural extension of the 3DRIM will be to extend its application into more complex optical signal processing, i.e., wavelength-selective switch. A grating structures have been selected to explore the selectivity of the switch. For this reason, we proposed that the surface of the micromirror being replaced by a suitable gratings instead of the flat reflective surface. Thus, this research has developed a rigorous formulation of the electromagnetic scattered near-field from a general-shaped finite gratings in a perfect conducting plane. The formulation utilizes a Fourier-transform representation of the scattered field for the rapid convergence in the upper half-space and the staircase approximation to represent the field in the general-shaped groove. This method provides a solution for the scattered near-field from the groove and hence is considered an essential design tool for near-field manipulation in optical devices. Furthermore, it is applicable for multiple grooves with different profiles and different spacings. Each groove can be filled with an arbitrary material and can take any cross-sectional profile, yet the solution is rigorous because of the rigorous formulations of the fields in the upper-half space and the groove reigns. The efficient formulation of the coefficient matrix results in a banded-matrix form for an efficient and time-saving solution.

The scalability and cost of microelectromechanical systems (MEMS) optical switches are now the important factors driving the development of MEMS optical switches technology. The employment of MEMS in the design and fabrication of optical switches through the use of micromachining fabricated micromirrors expands the capability and integrity of optical backbone networks. The focus of this dissertation is on the design, fabrication, and implementation of a new type of MEMS optical switch that combines the advantages of both 2-D and 3-D MEMS switch architectures. This research presents a new digital MEMS switch architecture for 1×N and N×N optical switches. The architecture is based on a new microassembled smart 3-D rotating inclined micromirror (3DRIM). The 3DRIM is the key device in the new switch architectures. The 3DRIM was constructed through a microassembly process using a passive microgripper, key, and inter-lock (PMKIL) assembly system. An electrostatic micromotor was chosen as the actuator for the 3DRIM since it offers continuous rotation as well as small, precise step motions with excellent repeatability that can achieve repeatable alignment with minimum optical insertion loss between the input and output ports of the switch. In the first 3DRIM prototype, a 200×280 microns micromirror was assembled on the top of the electrostatic micromotor and was supported through two vertical support posts. The assembly technique was then modified so that the second prototype can support micromirrors with dimensions up to 400×400 microns. Both prototypes of the 3DRIM are rigid and stable during operation. Also, rotor pole shaping (RPS) design technique was introduced to optimally reshape the physical dimensions of the rotor pole in order to maximize the generated motive torque of the micromotor and minimize the required driving voltage signal. The targeted performance of the 3DRIM was achieved after several PolyMUMPs fabrication runs. The new switch architecture is neither 2-D nor 3-D. Since it is composed of two layers, it can be considered 2.5-D. The new switch overcomes many of the limitations of current traditional 2-D MEMS switches, such as limited scalability and large variations in the insertion loss across output ports. The 1×N MEMS switch fabric has the advantage of being digitally operated. It uses only one 3DRIM to switch the light signal from the input port to any output port. The symmetry employed in the switch design gives it the ability to incorporate a large number of output ports with uniform insertion losses over all output channels, which is not possible with any available 2-D or 3-D MEMS switch architectures. The second switch that employs the 3DRIM is an N×N optical cross-connect (OXC) switch. The design of an N×N OXC uses only 2N of the 3DRIM, which is significantly smaller than the N×N switching micromirrors used in 2-D MEMS architecture. The new N×N architecture is useful for a medium-sized OXC and is simpler than 3-D architecture. A natural extension of the 3DRIM will be to extend its application into more complex optical signal processing, i.e., wavelength-selective switch. A grating structures have been selected to explore the selectivity of the switch. For this reason, we proposed that the surface of the micromirror being replaced by a suitable gratings instead of the flat reflective surface. Thus, this research has developed a rigorous formulation of the electromagnetic scattered near-field from a general-shaped finite gratings in a perfect conducting plane. The formulation utilizes a Fourier-transform representation of the scattered field for the rapid convergence in the upper half-space and the staircase approximation to represent the field in the general-shaped groove. This method provides a solution for the scattered near-field from the groove and hence is considered an essential design tool for near-field manipulation in optical devices. Furthermore, it is applicable for multiple grooves with different profiles and different spacings. Each groove can be filled with an arbitrary material and can take any cross-sectional profile, yet the solution is rigorous because of the rigorous formulations of the fields in the upper-half space and the groove reigns. The efficient formulation of the coefficient matrix results in a banded-matrix form for an efficient and time-saving solution.

Since semiconductor fabrication processes require numerous steps, cost and yield are critical concerns. In-situ monitoring is therefore vital for process control. However, this goal is currently restricted by the shortage of available sensors capable of performing in this manner. The goal of this research therefore, was to investigate the use of acoustic signals for monitoring and control of semiconductor fabrication equipment and processes. Currently, most methods for process monitoring (such as optical emission or interferometric techniques) rely on "looking" at a process to monitor its status. What was investigated here involved "listening" to the process. Using acoustic methods for process monitoring enhances the amount and sensitivity of data collection to facilitate process diagnostics and control. A silicon acoustic sensor was designed, fabricated, and implemented as a process monitor. Silicon acoustic sensors are favorable because of their utilization of integrated circuit and micromachining processing techniques; thus, enabling miniature devices with precise dimensions, batch fabrication of sensors, good reproducibility, and low costs. The fabricated sensor was used for in-situ monitoring of nickel-iron electrochemical deposition processes. During this process, changes occur in its plating bath as the alloy is being deposited. It is known that changes in the process medium affect the acoustic response. Thus, the sensor was implemented in an electroplating set-up and its response was observed during depositions. By mapping the sensor response received to the film thickness measured at certain times, a predictive model of the plated alloy thickness was derived as a function of sensor output and plating time. Such a model can lead to real-time monitoring of nickel-iron thickness.

Microfabricated fuel cells have been designed and constructed on silicon integrated circuit wafers using many processes common in integrated circuit fabrication, including sputtering, polymer spin coating, reactive ion etching, and photolithography. Fuel delivery microchannels were made through the use of sacrificial polymers. The characteristics of different sacrificial polymers were studied to find the most suitable for this work. A polypropylene carbonate solution containing a photo-acid generator could be directly patterned with ultraviolet exposure and thermal decomposition. The material that would serve as the fuel cells proton exchange membrane (PEM) encapsulated the microchannels. Silicon dioxide deposited by plasma enhanced chemical vapor deposition (PECVD) at relatively low temperatures exhibited material properties that made it suitable as a thin-film PEM in these devices. By adding phosphorous to the silicon dioxide recipe during deposition, a phosphosilicate glass was formed that had an increased ionic conductivity. Various polymers were tested for use as the PEM or in combination with oxide to form a composite PEM. While it did not work well alone, using Nafion on top of the glass layer to form a dual-layer PEM greatly enhanced the fuel cell performance, including yield and long-term reliability. Platinum and platinum/ruthenium catalyst layers were sputter deposited. Experiments were performed to find a range of thicknesses that resulted in porous layers allowing contact between reactants, catalyst, and the PEM. When using the deposited glasses, multiple layers of catalyst could be deposited between thin layers of the electrolyte, resulting in higher catalyst loading while maintaining porosity. The current and power output were greatly improved with these additional catalyst layers.

Along with knowledge of the interactions unique to microscale devices, designers of microelectromechanical systems (MEMS) require information about complex fabrication and packaging techniques in order to fully complete a successful design. To that end, the successful design of MEMS requires the collaboration of experts and designers in a variety of engineering fields. From the decision-based design perspective, MEMS designers require a means to sort the input and information generated in a collaborative design process. While the potential for the use of languages and part libraries have been addressed in the literature as a means to solve this problem, a means to embody these principles has not been addressed. The use of modular, executable, decision-centric templates to rapidly compose, solve, archive, and reuse compromise Decision Support Problems (cDSP) for specific design problems has been proposed in the literature. The result of this work is a means of separating procedural design knowledge from declarative knowledge and parsing the cDSP into a set of computer-interpretable templates. A stated need in this work is the extension of the templates to accommodate the coupled solution of two cDSPs utilizing game theoretic principles. In this thesis, the theoretical structures of decision-centric templates are applied to the needs of MEMS designers. Computer interpretable, decision-centric templates, used to save, reuse, and aid in design decisions, are extended to permit MEMS designers and fabricators to collaborate via coupled cDSPs, using game theoretic principles of cooperative, noncooperative, and leader-follower games. This approach is illustrated through its application to the design and prototype fabrication of microscale gas chromatography separation channels. The outcome of this work is twofold: first, MEMS designers and fabricators will have a means to compose, collaboratively solve, archive, and reuse compromise Decision Support Problems in a computer interpretable manner, and second, decision templates will be extended through the use of game theoretic principles.

This dissertation presents an investigation of miniaturized sensors, designed to wirelessly measure pressure in harsh environments such as high temperature and biomedical applications. Current wireless MEMS pressure sensors are silicon-based and have limited high temperature operation, require internal power sources, or have limited packaging technology that restricts their use in harsh environments. Sensor designs in this work are based on passive LC resonant circuits to achieve wireless telemetry without the need for active circuitry or internal power sources. A cavity, which is embedded into the substrate, is bound by two pressure-deformable plates that include a parallel-plate capacitor. Deflection of the plates from applied pressure changes the capacitance, thus the resonance frequency varies and is a function of the applied pressure. The LC resonant circuit and pressure-deformable plates are fabricated into a monolithic housing that servers as the final device package (i.e. intrinsically packaged). This co-integration of device and package offers increased robustness and the ability to operate wirelessly in harsh environments. To intrinsically packaged devices, the fabrication approach relies on techniques developed for MEMS and leverage established lamination-based manufacturing processes, such as ceramic and flex-circuit packaging technologies. To demonstrate operation in high temperatures applications, LTCC and HTCC ceramic pressure sensors were fabricated and characterized, operating up to 450°C under 5 bars of pressure while HTCC devices demonstrated electrical functionality up to 600°C. To demonstrate operation in biomedical implantable applications, polymer-based and polymer-ceramic flexible designs were fabricated and characterized. Bench testing for > 300 millions pressure cycles (simulated 7 years of pulse pressure) confirmed the reduction of frequency drift for polymer-ceramic pressure sensors compared to purely polymer-based pressure sensors. Finally, LCP-based pressure sensors were delivered in vivo into canine models with mock abdominal aortic aneurysms and monitored wirelessly over 30 days. The animal results confirmed both catheter deliverability and wireless telemetry in real biomedical applications.