Optical MEMS Switches: Theory, Design, and Fabrication of a New Architecture

Abstract

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.