This dissertation discusses the design and fabrication of micro-optical scanners that have high-quality optical-surface properties and are capable of previously unattainable high scanning rates. Scanners having high-quality optical surfaces are fabricated and characterized in both diffractive and reflective applications.
A first project investigates methods of creating scanning rectangular diffraction gratings using well-established fabrication methods of silicon surface-micromachining in a foundry process. We then introduce new methods to form an actuated blazed grating, a diffraction grating having a triangular surface profile that provides improved diffraction efficiency and wavelength resolution when compared to the rectangular grating. We use KOH anisotropic etching of single-crystal silicon to create a mold tailored by crystal planes in the substrate. The triangular shape of the mold shapes an LPCVD-deposited thin-film polysilicon plate. The polysilicon plate can be lifted out of the mold in the silicon substrate after an interposed thin film of silicon dioxide has been selectively etched by HF to release it. Hence, by integrating bulk-silicon micromachining steps into the surface-micromachining process, we create the high-quality diffractive surface properties of a blazed grating on a scanner structure. Improvements over the scanned rectangular grating in terms of diffraction efficiency and wavelength resolution are demonstrated to be significant. The measured diffraction efficiency of 60% in the blazed grating is four times that of the rectangular grating at an incident wavelength of 632.8 nm. Wavelength resolution for the high-order blazed grating is not limited by the linewidth of the fabrication process in contrast to the case of first-order rectangular gratings. Thus, the wavelength resolution of the blazed grating holds a fivefold increase from that of a comparably sized rectangular grating made in a minimum 2μm-linewidth process.
Some limitations of the surface-micromachined scanner, such as non-rigid optical plates that deform dynamically during scanning and weak actuators that are only useful for resonant scanning, prevent the use of these scanners in high-performance applications. Examples of these demanding applications are: in external-cavity tunable lasers (whichrequirestatic-positioningcapability),andindensewavelengthmultiplexing/demultiplexing (which requires minimum resolvable wavelengths of 1 nm or less). For our design of the surface-micromachined blazed diffraction grating, actuation was only satisfactory at resonance, and deformation of the optical plate during resonant scanning degraded the minimum resolvable wavelength of the grating from 1 nm to 2.7 nm. This performance limits the surface-micromachined grating to less demanding applications such as portable scanning spectrometers.
The limitations apply both to scanning gratings and mirrors. In the case of scanning mirrors for high-resolution display applications, much improvement is demonstrated in scanning mirrors that we have developed that use thick single-crystal silicon as structural material and which are built using high-aspect-ratio micromachining. With a new process, we have produced torsional scanners driven by high-force actuators. The mirrors are built on a thin film of polysilicon stretched across a thick, single-crystal silicon support rib. They are rigid and remain flat even under high-speed scanning stresses. We call the mirrors STEC/TOS (Staggered Torsional Electrostatic Combdrive / Tensile Optical-Surface). They have low inertia and are capable of high-speed resonant scanning without appreciable surface deformations (too small to compromise the optical resolution of the mirror).We demonstrate STEC/TOS micromirrors of differing designs, some have resonant frequencies as high as 81 kHz and others are capable of dc mechanical-deflection angles up to 8.3°. We have controlled the tensile stress in the polysilicon optical surface sufficiently to limit static- and dynamic-deformations to less than 70 nm across the polysilicon optical surface.
Scanning imaging applications such as a high-resolution scanning display require two scanning micromirrors having diverse requirements. One microscanner must be capable of large-angle, high-speed scanning and the other must be capable of large-angle, low-frequency mirror positioning. We have addressed the design constraints and tradeoffs for the STEC/TOS micromirrors in these two applications.
To achieve large-angle, high-speed scanning without significant static- and dynamic-surface deformations, the micromirror must be lightweight and have relatively large-force actuators and stiff torsional hinges. The polysilicon mirror surface should be moderately tensile-stressed (200-300 MPa) and supported by a stiff single-crystal silicon support rib. A tradeoff exists between stiffness of the silicon support rib, primarily determined by its width, and the overall mass of the mirror. Another tradeoff exists between achievable static- and dynamic-surface deformations. Higher tensile stress in the polysilicon surface produces lower dynamic-surface deformation but also applies more stress to the silicon support rib and thus contributes to increased static-surface deformation.
In the second case (that of the large-angle low-frequency scanner), the micromirror should have relatively large-force actuators and soft torsional hinges. The softer hinges provide lower restoring forces, thus allowing larger-angle scanning for a fixed amount of actuator force. As dynamic-surface deformations and overall mass are lesser issues for this scanner, its design is easier with only static-surface deformations to consider. A low tensile stress (under 100 MPa tensile) polysilicon-mirror surface supported by a very wide(and thus very stiff) single-crystal silicon support rib is sufficient to ensure minimal static-surface deformations.
The design of MEMS micromirrors has typically focused the use of existing fabrication technologies without giving equal attention to the optical performance of these devices, especially their performance under dynamic conditions. The research described here targets designs and fabrication technologies that can create optical MEMS scanners with high-quality optical surfaces that do not deform nor degrade during operation. We have pushed the limits by creating optical scanners to scan at previously unattainable speeds (as high as 81 kHz). In this work, we have developed a fabrication process that produces robust scanners and brought the technology to a level suitable for transfer to industry.