Ultra-flat micromirror arrays with tip, tilt, and piston control
The space-based observatories planned within the framework of NASA's Terrestrial Planet Finder (TPF) mission are being designed to perform high-contrast diffraction-limited imaging of extrasolar Earth-like planets. However, the required optical systems face a variety of unconventional wavefront control challenges. One example is provided by the visible nulling coronagraph under development at the Jet Propulsion Laboratory, which uses an interferometric approach to block parent starlight while enhancing planet light in the visible spectrum for detection. For successful planet imaging, parent starlight must be suppressed by a factor of 1010, which requires the nulling coronagraph to control both the phase and amplitude of recombined wavefronts to a very high precision.1–3
This technological challenge provided the impetus for the development of a new class of segmented deformable mirrors (DM) based on microelectromechanical systems (MEMS) technology. The DM has three adjustable degrees of freedom per subaperture that control out-of-plane rotation (tip and tilt) and surface normal motion (piston) over a continuous range of ±3mrad and 1μm, respectively. To perform wavefront phase and amplitude control on the collected planet and star light, the nulling coronagraph pairs this tip-tilt-piston (TTP) DM with a coherent array of optical fibers, coupling each mirror segment or subaperture to a fiber. Phase errors between the recombined interferometer wavefronts are corrected on a subaperture-by-subaperture basis using mirror segment piston motion. Amplitude errors are controlled by introducing tilt to unbalanced wavefront subapertures, which couple more or less light into their corresponding optical fibers (see Figure 1).
To produce interference nulls with the level of contrast required for planet detection, the TTP DM segments must have a deviation from flatness of less than 1nm RMS over their complete range of motion. This requirement translates into significant design and microfabrication challenges due to the tendency of thin polysilicon components to bend during actuation and curl as a result of the residual stress gradients embedded in the deposited materials. Furthermore, the surface micromachining process used to fabricate DMs is prone to print-through effects, meaning that successive polysilicon layers adopt the topography of prior depositions, thus reducing surface flatness. To address these issues, we have developed a new mechanical design and microfabrication process capable of producing micromirrors with the desired surface planarity.
Our micromirror design consists of a 600μm hexagonal mirror segment attached to three individual electrostatic actuators, which provide tip, tilt, and piston motion through differential actuation. The actuators are made of a compliant diaphragm rigidly attached on two opposing ends and suspended over a fixed electrode (see Figure 2). A short video of mirror segment motion is available online.4
To maintain the stringent surface flatness requirements of the nulling coronagraph, two techniques are used to oppose the bending moments experienced by the mirror surface during actuation. First, an increase in mirror segment thickness relative to that of the actuator diaphragm provides resistance to the bending moments imparted to the mirror surface by the post attachments. This is because the flexural rigidity of the mirror segment is proportional to its thickness cubed. Second, flexure cuts embedded in the actuator diaphragm are made around the post connections, as illustrated in Figure 2. These also reduce the bending moments applied to the mirror segment by lowering the torsional stiffness of the post connection.
DM microfabrication primarily consists of a micromachining process that uses epitaxial polysilicon (epipoly) deposition to achieve a mirror segment thickness greater than that provided by MEMS foundries using conventional polysilicon deposition techniques.5 The epipoly silicon provides the mirror segment with additional rigidity while allowing polishing to better than 1nm RMS local surface roughness. This eliminates print-through effects. Figure 3 compares polishing results obtained for both conventional polysilicon and epipoly surfaces.
Unfortunately, stress gradients in the epipoly layer can create significant curvature in released devices, making them unsuitable for the nulling coronagraph (see Figure 4, left). However, research has shown that the curvature of released epipoly segments can be successfully controlled using high temperature furnace annealing in which the mirror segment is simultaneously doped with phosphorous. The high temperature helps to relieve some of the residual stress, while the dopant creates a counteracting stress gradient that acts to flatten the mirror. Using this process, desired curvatures have been obtained for mirror segment thicknesses in the 7 to 11μm range (see Figure 4, right). To prevent mirror segment bending during actuation, the actuator flexure dimensions (width and length) are accordingly designed to accommodate this thickness range.
A closed-packed hexagonal array of 331 TTP micromirrors with optimized mirror thickness and actuator flexure dimensions is currently being fabricated.5 This DM will be used in a visible nulling coronagraph to demonstrate the feasibility of the instrument for the TPF Coronagraphic observatory. The technology is also being extended to a NASA Discovery mission for detecting extrasolar giant planets, which will require an hexagonal array of 1027 mirror segments. We are currently exploring the feasibility of applying this micromirror technology to laser-guide star tracking in adaptive optics for extremely large telescopes.
This work was initially supported by NASA/JPL award number 1254441. It is now supported by an SBIR Phase II award (number NNC07CA31C) to Boston Micromachines Corporation. Dr. Bifano acknowledges a financial interest in Boston Micromachines Corporation.
Jason B. Stewart is a PhD candidate in the Department of Electrical and Computer Engineering at Boston University. He received a BA in Physics from Colgate University in 2000 and completed his MS at Boston University in Electrical Engineering in 2004. He is presently a research assistant with Professor Thomas Bifano in the Precision Engineering Research Laboratory.
Thomas G. Bifano received his BS and MS in Mechanical Engineering and Materials Science from Duke University in 1980 and 1983, respectively, and a PhD in Mechanical Engineering from North Carolina State University in 1988. He is currently the Director of the Boston University Photonics Center, a professor in the Department of Mechanical Engineering and the Chief Technical Officer of Boston Micromachines Corporation.