1

Design, fabrication, and open-loop control of micro deformable mirrors for astronomical telescopes

Dr. Thomas Bifano Professor, Mechanical Engineering Director, Photonics Center Boston University, 8 Saint Mary’s Street, Boston, MA 02215 (617) 314-4083, tgb@bu.edu

Contents

The research product ...............................................................................................................................1Deformable mirrors: current state-of-the-art...........................................................................................2AO and MEMS DMs: relevance to astronomy and space science..........................................................3Summary of MEMS DM research and plans ..........................................................................................4

Design and modeling ..........................................................................................................................4Optical MEMS process innovation.....................................................................................................5Precise control approaches for MEMS DMs ......................................................................................6

Productive sectors involved in MEMS DMs and applications of the research.......................................7Astronomical imaging applications ....................................................................................................8Retinal imaging and bio-microscopy applications..............................................................................9Defense applications ...........................................................................................................................9

References .............................................................................................................................................10

The research product My research involves design, fabrication, and control of a new

class of deformable mirrors (DMs) for adaptive optics (AO). These DMs substantially extend the scientific capabilities of ground and space-based astronomical telescopes (Figure 1).

About ten years ago, my students and I at the Photonics Center at Boston University began a research program on silicon microelectro-mechanical (MEMS) DMs. Our primary goal was to find an alterna-tive to the expensive hand-assembled macroscale DMs that repre-sented the state-of-the-art in wavefront correction at that time. An ad-ditional goal was to use a new MEMS “foundry” for product research, which was unprecedented at the time. By conforming to predefined rules and process recipes, we were able to concentrate our research on device electromechanical design and optical performance, with less focus on materials science and proc-essing. This allowed us to design and build a family of DMs that proved particularly robust and manufac-turable. Within a little more than a year we produced our first working MEMS DMs.

Our DM architecture is based on a scalable array of parallel plate electrostatic actuators, fabricated in silicon through semiconductor batch processing. Each square actuator plate is rigidly connected to the substrate along two of its edges, and is suspended above an addressable electrode. Voltage applied to that electrode imposes an electrostatic attractive force on the electrically grounded actuator plate, causing it to bend toward the substrate in proportion to the square of applied voltage. Each actuator plate has a central post connected to a continuous or segmented mirror layer (Figure 2).

Figure 1: Packaged MEMS DM.

2

The success of this design and manufacturing approach led to devices that were in demand by re-searchers outside of BU, even as they were still being developed. As a result, in 1999 a former student and I founded a company to make the DMs commercially available. Boston Micromachines Corporation (BMC) is the exclusive licensee of BU’s MEMS DM technology. The family of innovative products that we have produced, in continuous collaboration with academic researchers and students from BU, share a heritage in the original design and manufacturing approach used. These products now include mirrors with 32 to 4096 actuators, mirror apertures from 1.5mm to 26mm, and stroke from 2µm to 8µm. BMC products in widespread use include the MultiDM,

with 140 actuators and a USB controlled driver, and the KiloDM and KiloSLM, continuous and seg-mented versions of a 1024 actuator device that features 20kHz frame rate. BMC also produces drivers and software for DMs and has custom-made several unique DMs for the astronomical imaging community.

The principal applications for my research are high-resolution, high-contrast astronomical telescope compensation with AO. Open-loop, precise, and high-resolution control, is possible uniquely with MEMS-DM, and is indispensible for the next generation of telescope instruments. Examples of devices currently being developed in my research program include a prototype 4096 actuator DM for the Gemini Planet Imager, an ultraprecise 993 actuator tip-tilt-piston DM for use in a visible nulling coronagraph ex-periment led by the Jet Propulsion Laboratory as part of NASA’s Terrestrial Planet Finder (TFP-C) mis-sion, and a 1024 actuator DM for a space-based AO experiment to launch in 2010 as part of NASA’s sounding rocket program. My continued research at BU and BMC will accelerate the availability of these devices to the international scientific community.

Deformable mirrors: current state-of-the-art Conventional macroscale DMs, made with either piezo-bimorphs or stacked piezo-actuators are lim-

ited as astronomical telescope wavefront correctors by a number of factors. They require closed loop con-trol due to inherent actuator hysteresis and temperature dependent performance, their manual assembly makes it costly and difficult to scale to devices with large actuator numbers, and their large mass results in relatively low temporal bandwidths. In potential space-based applications, large power requirements and bulky electronics drivers compound these limitations. Nevertheless, the architectures and algorithms for AO instrumentation has accommodated these limitations since these were the only wavefront correc-tors available for astronomical telescopes until the introduction of MEMS DMs. The design and manu-facturing approaches developed in our MEMS DM research offer inherent advantages: Design • The device is scalable: increasing the size of the DM and its spatial resolution is achieved simply by

adding identical actuators to the array (i.e. by copying and pasting lithographic mask features). • The device is mechanically stiff and has low mass, allowing control bandwidths of tens of kilohertz. • The actuation mechanism is repeatable to sub-nanometer precision, consumes almost no power, ex-

hibits no hysteresis, and is unaffected by billions of cycles of operation. Manufacturing • The production approach does not call for exotic materials or tolerances: it can be made using a

MEMS foundry and begins with an optically smooth, flat, and inexpensive substrate. • Devices are batch produced twenty wafers at a time, so that while development costs are high, com-

mercial production and replication costs are low. • Hundreds of devices can be produced on each wafer, allowing broad parameter variation in a single

batch production cycle. This accelerates research and prototyping.

Figure 2: Schematic cross section of continu-ous and segmented BU MEMS DMs. The mir-ror (gold) is connected by posts (green) to an array of electrostatic actuators (blue). Actua-tor deflection is approximately proportional to the square of voltage applied to the rigid elec-trodes (red) on the wafer substrate (black).

3

MEMS-DM research offers the rare opportunity to introduce technology that is both more economical and more capable than the state-of-the-art. MEMS-DMs already achieve stroke comparable to that of conventional systems (up to 8µm of physical stroke has been demonstrated using our devices), and ex-ceed all reported macroscale DMs in temporal bandwidth and actuator count. They reduce size, weight and power by considerably more than an order of magnitude, and are available commercially for a frac-tion of the cost of conventional DMs. In the following pages I summarize how AO and MEMS DMs are transforming the field of astronomical imaging, and detail how our MEMS DM research at BU and BMC has enabled new science in astronomy, biomedicine, microscopy, and communication.

AO and MEMS DMs: relevance to astronomy and space science Adaptive optics on large ground-based telescopes has resulted in a surge of new astronomical discov-

eries over the past few years, some examples of which are illustrated in Figure 3. Using AO, scientists around the world have produced what Science Magazine noted as a “breakthrough of the year” in 2008: the first direct observation of exoplanets in another solar system (1, 2). In our own solar system, AO has been used recently to image, with unprecedented clarity, the rings of Uranus (3), and the sodium tail of Mercury (by BU colleagues) (4). AO also allowed astronomers to find massive black holes at the merger of two distant galaxies (5). As a final example, an AO-enhanced military telescope permitted high resolu-tion images of the damaged wing of the ill-fated space shuttle Columbia as it orbited the earth (6).

Astronomical instruments are also undergoing rapid evolution, especially large ground-based tele-scopes. Promising and relatively new AO instruments include multi-conjugate adaptive optics (MCAO), multi-object adaptive optics (MOAO), extreme adaptive optics (ExAO), and space-based adaptive optics. MCAO employs two or more guide stars to enable tomographic wavefront error sensing, and then two or more deformable mirrors in series to correct those errors. The result is a wider corrected field of view (7). This technique was recently used to produce the sharpest whole-planet (Jupiter) picture ever taken from the ground (8). MOAO is an instrument concept that also uses multiple guide stars and multiple deform-able mirrors (9). However in MOAO, DMs would used in parallel to apply independent corrections for the turbulence-induced wavefront distortions in separate subregions of interest, allowing high-resolution imaging of small science objects distributed across a large field of view. ExAO refers to a planned ground-based instrument intended for observation of Jovian exoplanets, with the help of high spatial reso-lution DMs having thousands of control points. In addition to exoplanet detection, such an ExAO system will be capable of characterizing dust disks around stars and will allow high Strehl ratio imaging at visible wavelengths (10-12). Space-based imaging offers the opportunity to avoid the dynamic aberrations intro-duced by the atmosphere. Space based AO compensate for imperfections in the telescope optics, and is needed for high contrast coronagraphy.

Figure 3: Examples of science recently enabled by ground-based telescopes with AO. Clockwise from left: Exoplanet solar system; rings of Uranus; space shuttle Columbia; twin black holes in merging galaxies; sodium tail of Mercury.

4

These new instruments offer the potential to greatly advance astronomical science, but each requires DMs that improve upon those currently used in astronomical telescopes. Such DMs are now available as a result of our work at Boston University and Boston Micromachines Corporation. • MOAO requires open loop control of the DM: corrections made by the DM are not accessible to the

wavefront sensor, so closed feedback is not possible. That makes it important that the DM can be shaped to the required precision in a single step – something that is difficult or impossible to do using piezo-actuated conventional DMs. We have developed an analytical model coupled with a sparse calibration technique that allows open loop control to an accuracy of ~15nm RMS for MEMS DMs. This approach and these DMs offer a uniquely enabling path to MOAO.

• ExAO demands DMs with several thousand actuators, and shape control to within a few nanometers. Again, the prior generation of DMs fall short of this requirement, but MEMS-DMs have shown prom-ise. A 4096 actuator engineering grade DM has been produced for the Gemini Planet Imager, and a science grade version will be delivered in 2009. Additional work on packaging and through-wafer in-terconnects by my group promises paths to extend to mirrors with up to 10000 actuators. In this and other planet finding applications, control loop iterations could plausibly require hours of integration, so open loop “go-to” precision is again very valuable (13).

• Space based AO is highly constrained by size, weight, and power consumption of the AO system. MEMS DMs are well-matched to those constraints, because of their compact dimensions and low-power drivers, enabled by low capacitance (~100fF) electrostatic actuators. Overall, MEMS DMs re-duce by more than an order of magnitude the volumetric size, total weight, and consumed power of in comparison to commercially available alternative DMs. Our MEMS DMs have been built into a com-pact telescope scheduled for launch in 2010 using a NASA sounding rocket, and we developed a unique tip-tilt-piston segmented DM with segment flatness of better than 5nm RMS for a NASA pro-totype as part of a space based visible nulling coronagraph project.

Summary of MEMS DM research and plans My research involves design and modeling of DMs, optical MEMS process innovation, and demon-

stration of precise control approaches for MEMS DMs. Highlighted here are some of my important re-search achievements and plans for future work (14-26).

Design and modeling The basic architecture of the parallel-plate electrostatic actuator is depicted in Figure 4. We can gain

an understanding of its electromechanical behavior of the actuator by approximate the actuator as a fixed, planar lower electrode with surface area L2 and a compliant, planar upper electrode, separated by an ini-tial gap g. If we apply a voltage V across the gap, the center of the compliant upper electrode deflects by an amount w. With these assumptions, the electrostatic attractive force is:

€

FE =εL2V 2

2(g − w)2

where ε is the permittivity of free space (8.8pF/m). Assuming the compliant actuator electrode acts as an elastic beam with fixed ends, the actuator mechanical force that resists this actuation is:

€

FA =16Et 3wL2

where E is the elastic modulus of silicon, t is the actuator plate thick-ness, and L is the actuator span. An equilibrium force balance yields:

€

V =32Et 3w(g − w)2

L4ε

This equilibrium is unstable for deflections z that exceed one third of the initial actuator gap. Since the actuator gap is defined by the thick-

Figure 4: Simplified schematic model of the electrostatic actuator

5

ness of the first sacrificial layer in the device, and since processing constraints make it challenging to make sacrificial layers thicker than ~10µm without introducing cracks, the basic actuator design is limited to a stroke of ~3.3µm. This is sufficient for many AO applications, but inadequate for others. Through a series of actuator design changes anchored by these models and finite element electrostatic/elastic simula-tions, we altered the active area and shape of the lower electrode and perforated the compliant upper elec-trode to modify its flexural response. We were able to extend the range of stable actuation to 80% of the actuator gap, achieving >8µm stroke in prototype actuators. Our imminent research plan is to integrate that advanced actuator design into MEMS DM devices suitable for astronomical AO applications in visi-ble and near-IR wavelengths, while maintaining good optical quality and dynamic performance.

Optical MEMS process innovation To translate a MEMS design into a foundry fabrication process, one must decompose the device into

a series of thin film layers built up from the wafer surface. The cyclical process involves depositing a thin film, patterning it with a temporary mask, removing the film where it is unprotected by the mask, then removing the mask. This cycle is repeated until the full structure is produced. Generally, one alternates between polycrystalline silicon structural layers (e.g. for the base electrodes) and silicon dioxide sacrifi-cial layers (e.g. to form the actuator gap) . At the conclusion of thin film fabrication, sacrificial layers are all removed simultaneously through bulk etching in hydrofluoric acid (HF), leaving fully assembled MEMS DMs on the wafer. These combined processes, collectively know as silicon surface micromachin-ing, are performed with semiconductor fabrication tools and inherit most process steps from that industry, yielding high-quality fully-dense films measuring 0.1µm to 10µm thick, with lateral tolerances of a few hundredths of a micrometer and surface-normal tolerances of a few tens of nanometers.

We had to tackle two important process problems before we were able to produce useful MEMS DMs: reducing print-through and residual strain. Each patterned layer in-troduces unwanted topography in subsequent layers. This results in unacceptable print-through on the mirror. Our so-lution was to develop a new foundry design rule, now known as “design-based planarization,” or DBP (21), that constrains mask patterns to be made up only of narrow channels. Conformal deposition into narrow thin film chan-nels reduces print-through. Anchor cuts in a structural layer are allowed to capture and fully enclose regions of an under-lying sacrificial layer (gray areas in Figure 5) to reinforce the otherwise thin anchor structures.

With DBP, our devices featured acceptable surface figure before sacrificial release. However after sacrificial release the mirror film deformed by several micrometers, particularly on segmented devices, due to residual stress gradients in the film. We developed a physical model of the problem, based on mismatched strain introduced via dopant diffusion during a critical annealing step. We then invented a technique for flattening mirrors after structural release, using Argon ion machining at BU. This was shown to alter the contour shape of free-standing thin-film structures by affecting their through-thickness stress distribution. We demonstrated that post-release out-of-plane deformation could be reduced, and optically flat surfaces achieved. An analytical model incorporating the relevant mechanics of the problem was formulated and used to provide an understanding of the mechanisms behind the planarization proc-ess. The principal mechanisms identified were amorphization of a thin surface layer due to ion beam ex-posure and gradual removal of stressed material by continued exposure to the ion beam (19). The patent for this process (#6,929,721) is one of several granted for BU MEMS DM processing innovation. Con-tinuing process research in my group is aimed at improving optical quality of the MEMS surface, cur-rently exhibiting about 10-20nm RMS uncorrectable figure error. Through chemomechanical polishing, advanced annealing, and thick mirror film deposition with epitaxially grown silicon, we have reduced these errors on some DMs to <5nm RMS. Our plan is to translate those research outcomes into viable production processes at the foundry.

Figure 5: Schematic of design based planarization to eliminate print-through topography. Top: without DBP rules. Bottom: with DBP rules.

6

Precise control approaches for MEMS DMs One important feature of the BU/BMC MEMS DMs that have resulted from this decade of research is

that they can be shaped predictably in a single step to nanometer-scale precision, enabling open loop AO control. That characteristic allows profound advances in AO. If control can be achieved in open loop, then promising new astronomical instruments can be implemented for the first time, including especially multi-object adaptive optics (MOAO) and extreme adaptive optics (ExAO). Open-loop operation also enables separate correction of tip/tilt stars in a laser guide star system, which significantly improves sky coverage [12, 26].

Measurements made on a series of our 1024 actuator MEMS DMs by the Laboratory for Adaptive Optics (LAO) at UC Santa Cruz demonstrated sub-nanometer repeatability, sub-nanometer stability, and sub-nanometer hysteresis over a wide range of operating conditions (27). These characteristics, which are unachievable using conventional DMs that are intrinsically subject to motion hysteresis, led to the notion that MEMS DMs might enable unprecedented AO instrumentation. With closed loop control, the DM has been flattened in the LAO testbed to within a 0.54nm in its controllable band of spatial frequencies (and 13nm RMS overall) (28).

Controlling a MEMS DM, in open or closed loop, is compli-cated by the fact that independently addressed actuators are me-chanically coupled to one another through the mirror facesheet [5]. As a result, the displacement of any DM actuator alters the forces experienced by its neighboring actuators. An efficient open-loop control algorithm can be achieved if one first estab-lishes a DM model that consists of two coupled mechanical sub-systems: the continuous facesheet and the array of actuators con-nected to the facesheet via rigid posts. The rigid posts are the points of connection for the two subsystems. This can be seen in the free body diagram of Figure 6, where FM is the force imparted by the mirror facesheet, FA is the force imparted by the actuator diaphragm, and FE is the electrostatic attractive force associated with an applied actuator voltage.

The DM mirror layer (facesheet) is modeled as a thin plate free on all edges undergoing both stretching and bending. For dis-placements smaller than the facesheet thickness, the deformation

is dominated by linear elastic bending as described by the biharmonic thin plate equation. As displace-ment approaches the thickness of the facesheet, in-plane strain is no longer negligible and the facesheet begins to stretch as well as bend. The governing equation for out-of-plane deflection of a linear elastic plate experiencing bending and stretching is given by:

where q(x,y) is the surface-normal distributed load (N/m2) responsible for producing mirror facesheet deflection w(x,y), and D is the plate flexural rigidity, given by Eh3/12(1-ν2), where E is the modulus of elasticity, ν is Poisson’s ratio, and h is the facesheet thickness. Using this equation the generalized load q(x,y) necessary to create a desired (known) mirror shape w(x,y) can be calculated (Note that this is achieved by simple numerical integration). In reality however the mirror is loaded only at discrete points corresponding to the mirror post locations, and the generalized load can be represented by a collection of discrete forces FM acting at these post locations. FM is estimated at each post by integrating q(x,y) over each mirror subaperture, as shown in Figure 7. Once the mirror force FM is known at each actuator post, it is possible to reduce the open-loop control problem to one that is entirely local and uncoupled. As a re-sult, the electrostatic actuator response to a local mirror force FM can be modeled through a compact set of empirical measurements for a range of values of actuator post displacement w and applied voltage V.

Figure 6: Equilibrium force bal-ance at the actuator post.

7

Since all actuators in the array are identical, the calibration of one actuator is sufficient to achieve good open loop per-formance.

We recently implemented this computationally simple and inherently fast algorithm on a BU 140 actuator DM. Shapes at the limit of achievable mirror spatial frequencies with up to 1.5µm amplitudes were achieved with less than 15nm RMS error (14). Others have also implemented open loop AO control with our mirrors, demonstrating compara-ble results (29). That open loop control architecture and the BU DM was demonstrated in the first civilian use of MEMS-based astronomical AO in the Villages project led by Don Gavel et. al. in the past year (30, 31). In the coming year, we will make an effort to extend that model to DMs with larger achievable stroke and many more actuators as would be required in MOAO (9). We will expand the algorithm to account for small systematic differences among actuators in the array. A significant challenge in this work will be that larger stroke DMs can experience considerable stretching of the actuator membrane in addition to the bending that dominates behavior at small deflection. This stretching nonlinearity complicates the empiri-cal calibration step. We believe that these nonlinearities are predictable, and that extra computation will required only for advance calibration and not for real-time AO control.

While open loop control is a major focus of our current research, I have also established an active program of study for closed-loop controllers that can take advantage of the enhanced performance acces-sible with MEMS DMs. One, based on stochastic parallel gradient descent (SPGD), uses no wavefront sensor and instead optimizes a global image quality metric through iterative feedback based on simulta-neous perturbation of all DM actuators. Though somewhat inefficient, the SPGD algorithm has advan-tages in that it can achieve image improvement even in conditions where conventional wavefront sensing is difficult, such as in low elevation imaging through extended turbulence. Moreover, the SPGD algo-rithm can take advantage of the full 20kHz frame rate achievable with MEMS DMs. Our current plan is to employ SPGD and a 1024 actuator DM in an AO testbed at the US Army in a cascade with conventional AO, with the goal of testing this system on the Air Force 3.6m AEOS telescope in 2010. This program will be conducted in partnership with Dr. Vorontsov, who pioneered SPGD for AO at the Army Research Laboratory. I also work on more conventional AO control systems to characterize MEMS DMs for space-based telescopes. I am leading a project sponsored by the US Government to demonstrate closed loop compensation of a space-based reconnaissance telescope, using a custom fabricated hexagonal DM with 1024 actuators and controlled via interferometric feedback.

Productive sectors involved in MEMS DMs and applications of the research In addition to this comprehensive research program on MEMS

DMs, I work actively to translate these core research outcomes into useful products. The MEMS-DM technology that I developed at BU is licensed exclusively to Boston Micromachines Corporation (BMC), a company I co-founded with Paul Bierden. I continue to participate actively in BMC R&D programs, and serve as the CTO. BMC has been profitable since its first year of operation, and cur-rently employs 12 people in a facility located in Cambridge, MA. It has continued to use a foundry model for basic MEMS fabrica-tion. All design, modeling, mirror post-release processing, mirror coating, packaging, and testing are performed in-house at BMC. BMC also produces its own drivers for the MEMS DMs.

BU and BMC have produced mirrors for researchers world-wide, combining respective strengths in fundamental analysis at BU and MEMS prototyping/production at BMC.

Figure 7: Using the modified bihar-monic equation: one can rapidly de-termine mirror forces at each post, given the desired mirror shape w(x,y).

Figure 8: From left: P. Bierden, President; T. Bifano, CTO; S. Cornelissen, VP. Bierden and Cornelissen are BU graduates.

8

Applications for open-loop controllable MEMS-DMs spans three market sectors. First, and most rele-vant to astronomy of is high-resolution telescope imaging. Other important markets include retinal imag-ing and bio-microscopy, defense imaging systems, and laser communications.

Astronomical imaging applications Exemplifying its role as an international leader in MEMS-DM

development for astronomical AO, BMC produced the first 4096 ac-tuator 26mm aperture MEMS-DM for the Gemini Observatory, in a multinational project for planet imaging. An engineering-grade mir-ror, the first of its kind, has been delivered and is being evaluated (Figure 9). A science grade mirror will be delivered later this year.

BMC also produced the first MEMS-DMs ever used in civil-ian astronomical AO, at Lick Observatory, in a project led by Don Gavel and known as The MEMS-AO/Visible Light Laser Guidestar Experiments (Villages, Figure 10). The Villages experi-ment represents a pioneering effort to integrate MEMS DMs an in open loop AO control system for on-sky observation (28, 32-34).

BU and BMC collaborated on two projects to deliver advanced DMs for space-based astronomical imaging. The first, for NASA JPL, involved development of a segmented tip-tilt-piston mirror for a visible nulling coronagraph envisioned as part of the Terrestrial Planet Finder (TPF) program. The specifications for this mirror re-quired advances in both design and fabrication processes, to achieve 5nm RMS flatness on the segments and to ensure less than 2nm

bending of the 600µm diameter segment during actuation. The design ultimately developed by our team included an innovative flexure at the base of each of the three attachment posts connected to each mirror segment, and a 15µm thick post-polished silicon layer for the mirror, grown in a new epitaxial deposition process that we co-developed with our MEMS foundry. This device (Figure 11) was delivered to JPL last month, and is now undergoing testing at JPL’s High Contrast Imaging Testbed. In a second space-based DM project for astronomical imaging, BU and BMC have collaborated to produce a 1024 actuator DM for a NASA sounding rocket project, in collaboration with NASA JPL and Boston University’s Center for Space Physics.

In addition to these highlighted MEMS DM applications, the commercial 140 actuator and 1204 ac-tuator MEMS DM systems produced by BMC have been obtained by the following observatories and as-tronomical research laboratories over the past several years, and have resulted in dozens of publications at the forefront of the field. (12, 35-40)

Observatories: European Southern Observa-tory, Lick Observatory, Steward Observatory, Space Infrared Telescope for Cosmology and As-trophysics (SPICA), Gemini Observatory

Agencies/labs: European Space Agency, ONERA (France) Durham University, California Institute of Technology, NASA Jet Propulsion Laboratory, NASA Goddard NASA Ames, Japan Aerospace Exploration Agency (JAXA), National Astronomical Observatory of Japan, Princeton University, Lawrence Livermore National Labs, U. of California at Santa Cruz, U. of Hawaii, U. of Victoria, U. of Florida, Georgia Institute of Tech-nology, U. of Arizona.

Figure 10: Villages project in-strumentation, with MEMS DM AO. Picture credit (31).

Figure 11: Ultraflat (5nm RMS) tip tilt piston MEMS DM with flexure-relieved actuators. Left: measured surface for 1µm tip and 1µm piston ac-tuation of a single segment. Right: packaged DM with 993 actuators (331 segments).

Figure 9: Prototype 4096 actuator MEMS-DM for Gemini Planet Imager

9

Retinal imaging and bio-microscopy applications A second market sector that has been affected by the

introduction of MEMS DMs is retinal imaging and bio-microscopy. In a recent initiative supported by the Na-tional Science Foundation’s Center for Adaptive Optics and by a National Institutes of Health Bioengineering Research Partnership, a team of scientists produced more than a half dozen variations of retinal imaging instru-ments that are AO-compensated for aberrations of the eye. These include both scanning laser ophthalmoscopes (SLO) and optical coherence tomography (OCT) sys-tems, and they have resulted in the sharpest images of microscale features in the living retina ever obtained. AO has proved to be essential for high-resolution, high-contrast retinal imaging, enabled in these instruments by BMC’s 140 actuator MultiDMs. The results, an example of which is illustrated in Figure 11, might have significant impact on research associated with major retinal diseases and leading causes of blindness, including diabetic retinopathy, age related macular degeneration and glaucoma. In addition to other advantages already discussed, MEMS DMs are attractive in this application because their aperture diameter is closely matched to that of the eye’s pupil. AO sys-tems enabled by the BMC MEMS DM have been used for imaging photoreceptors, retinal vasculature, and blood flow velocity in foveal capillaries, and subsurface cell structure in the retina. A substantial body of discoveries made using instruments with our DMs has appeared in the recent scientific literature (41-64). Recent reviews highlight the impact of our device on the field (65, 66). MEMS DMs produced by other groups have also been used in retinal imaging (67-70), and a number of competing devices have been compared in the literature (71-73). Our group, with colleagues from LLNL, won two R&D 100 Awards for retinal AO technology: one in 2003 for a MEMS-based adaptive optics phoropter (MAOP), and one in 2007 for an adaptive optics scanning laser ophthalmoscope (MAOSLO).

In addition to retinal imaging, the BMC MEMS DM has had an impact on high-resolution bio-microscopy research, particularly in confocal, two-photon, and structured illumination microscopy for imaging subsurface tissue microstructures (54, 74). Based on one wide field microscope concept demon-strated using our DM, the first commercially available AO-enabled microscope was introduced by Thor-labs Corporation in 2008 (75). Thorlabs now also offers catalog sale of an AO kit with the MultiDM and a Thorlabs wavefront sensor.

Defense applications A final market sector for our

MEMS DMs is laser beam communica-tion and defense laser control systems. In a groundbreaking study conducted by colleagues at Lawrence Livermore National Laboratory in the California desert using a 1024 actuator DM, it was demonstrated that our DMs could com-pensate beam path aberrations in a sin-gle iteration over a 1km path using holographic control (76). Significant development of laser communication systems by US defense contractors us-ing BU/BMC mirrors has followed, including many DoD contractors and support from DARPA, the US Air Force, and the US Army.

Figure 11: Retinal image of the cone mo-saic measured without (left) and with (right) AO in an AOSLO using the BU/BMC DM. Photo credit: A. Roorda, UC Berkeley.

Figure 12: Schematic and cutaway photograph of the Adaptive Scanning Optical Microscope (ASOM), the first commercially available AO microscope, produced by Thorlabs Corporation, and built using the BMC MultiDM system.

10

References 1. in Science. (2008), vol. 322, pp. 1768. 2. C. Marois et al., Science 322, 1348 (Nov 28, 2008). 3. I. de Pater, H. B. Hammel, M. R. Showalter, M. A. van Dam, Science 317, 1888 (Sep 28, 2007). 4. J. Baumgardner, J. Wilson, M. Mendillo, Geophysical Research Letters 35, (Feb 2, 2008). 5. C. E. Max, G. Canalizo, W. H. de Vries, Science 316, 1877 (Jun 29, 2007). 6. in Spaceflight Now (2003). 7. R. Ragazzoni, E. Marchetti, G. Valente, Nature 403, 54 (Jan 6, 2000). 8. M. H. Wong et al., paper presented at the 40th Annual Meeting of the Division for Planetary Sciences, Ithaca, NY, 13 October 2008 2008. 9. R. G. Dekany et al., paper presented at the Advancements in Adaptive Optics, Glasgow, Scotland, United Kingdom, 2004. 10. J. W. Evans et al., paper presented at the Techniques and Instrumentation for Detection of Exoplanets II, San Diego, CA, USA, 2005. 11. J. W. Evans et al., paper presented at the MEMS/MOEMS Components and Their Applications III, San Jose, CA, USA, 2006. 12. B. Macintosh et al., Cr Phys 8, 365 (Apr-May, 2007). 13. B. Macintosh. (personal communication, 2009). 14. J. B. Stewart, A. Diouf, Y. P. Zhou, T. G. Bifano, Journal of the Optical Society of America a-Optics Image Science and Vision 24, 3827 (Dec, 2007). 15. J. B. Stewart et al., Sensors and Actuators a-Physical 138, 230 (Jul 20, 2007). 16. M. H. Miller, J. A. Perrault, G. G. Parker, B. P. Bettig, T. G. Bifano, J Micromech Microeng 16, 303 (Feb, 2006). 17. R. H. Webb, M. J. Albanese, Y. P. Zhou, T. Bifano, S. A. Burns, Appl Optics 43, 5330 (Oct 1, 2004). 18. J. A. Perreault, T. G. Bifano, B. M. Levine, M. N. Horenstein, Optical Engineering 41, 561 (Mar, 2002). 19. T. G. Bifano, H. T. Johnson, P. Bierden, R. K. Mali, J Microelectromech S 11, 592 (Oct, 2002). 20. T. Weyrauch et al., Appl Optics 40, 4243 (Aug 20, 2001). 21. R. K. Mali, T. Bifano, D. Koester, J Micromech Microeng 9, 294 (Dec, 1999). 22. M. Horenstein, T. Bifano, S. Pappas, J. Perreault, R. Krishnamoorthy-Mali, J Electrostat 46, 91 (Apr, 1999). 23. T. G. Bifano, J. Perreault, R. K. Mali, M. N. Horenstein, Ieee J Sel Top Quant 5, 83 (Jan-Feb, 1999). 24. R. K. Mali, T. G. Bifano, N. Vandelli, M. N. Horenstein, Optical Engineering 36, 542 (Feb, 1997). 25. M. N. Horenstein, T. G. Bifano, R. K. Mali, N. Vandelli, J Electrostat 42, 69 (Oct, 1997). 26. T. G. Bifano et al., Optical Engineering 36, 1354 (May, 1997). 27. K. M. Morzinski, D. T. Gavel, A. P. Norton, D. R. Dillon, M. R. Reinig, paper presented at the MEMS Adaptive Optics II, San Jose, CA, USA, 2008. 28. J. W. Evans et al., Opt Express 14, 5558 (Jun 12, 2006). 29. K. M. Morzinski, K. B. W. Harpsoe, D. T. Gavel, S. M. Ammons, paper presented at the MEMS Adaptive Optics, San Jose, CA, USA, 2007. 30. B. Grigsby et al., paper presented at the Advanced Optical and Mechanical Technologies in Telescopes and Instrumentation, Marseille, France, 2008. 31. D. Gavel et al., paper presented at the MEMS Adaptive Optics II, San Jose, CA, USA, 2008. 32. K. M. Morzinski et al., paper presented at the Adaptive Optics Systems, Marseille, France, 2008. 33. L. A. Poyneer, B. Bauman, B. A. Macintosh, D. Dillon, S. Severson, Opt Lett 31, 293 (Feb 1, 2006). 34. L. A. Poyneer, D. Dillon, S. Thomas, B. A. Macintosh, Appl Optics 47, 1317 (Mar 20, 2008). 35. R. D. Peters, O. P. Lay, A. Hirai, M. Jeganathan, paper presented at the Advances in Stellar Interferometry, Orlando, FL, USA, 2006. 36. R. D. Peters, A. Hirai, M. Jeganathan, O. P. Lay, paper presented at the Techniques and Instrumentation for Detection of Exoplanets II, San Diego, CA,

USA, 2005. 37. O. P. Lay, M. Jeganathan, R. Peters, paper presented at the Techniques and Instrumentation for Detection of Exoplanets, San Diego, CA, USA, 2003. 38. R. D. Peters, O. P. Lay, M. Jeganathan, Appl Optics 47, 3920 (Jul 20, 2008). 39. A. Give'on, R. Belikov, S. Shaklan, J. Kasdin, Opt Express 15, 12338 (Sep 17, 2007). 40. R. D. Peters, O. P. Lay, A. Hirai, M. Jeganathan, paper presented at the Techniques and Instrumentation for Detection of Exoplanets III, San Diego, CA,

USA, 2007. 41. F. Romero-Borja, K. Venkateswaran, A. Roorda, T. Hebert, Appl Optics 44, 4032 (Jul 1, 2005). 42. K. Grieve, A. Roorda, Investigative Ophthalmology & Visual Science 49, 713 (Feb, 2008). 43. D. W. Arathorn et al., Optics Express 15, 13731 (Oct 17, 2007). 44. J. L. Duncan et al., Investigative Ophthalmology & Visual Science 48, 3283 (Jul, 2007). 45. A. Roorda, Y. H. Zhang, J. L. Duncan, Investigative Ophthalmology & Visual Science 48, 2297 (May, 2007). 46. Y. H. Zhang, A. Roorda, J Opt Soc Am A 24, 1276 (May, 2007). 47. A. S. Vilupuru et al., J Opt Soc Am A 24, 1417 (May, 2007). 48. Y. H. Zhang, A. Roorda, J Biomed Opt 11, (Jan-Feb, 2006). 49. J. A. Martin, A. Roorda, Ophthalmology 112, 2219 (Dec, 2005). 50. S. Poonja, S. Patel, L. Henry, A. Roorda, J Refract Surg 21, S575 (Sep-Oct, 2005). 51. A. J. Roorda, Y. Zhang, Investigative Ophthalmology & Visual Science 46, (2005). 52. F. Romero-BOrja, K. Venkateswaran, A. Roorda, Investigative Ophthalmology & Visual Science 45, U1005 (Apr, 2004). 53. K. Venkateswaran, A. Roorda, F. Romero-Borja, J Biomed Opt 9, 132 (Jan-Feb, 2004). 54. R. J. Zawadzki et al., Opt Express 16, 8126 (May 26, 2008). 55. D. X. Hammer et al., Optics Express 14, 3354 (Apr 17, 2006). 56. T. Y. P. Chui, H. X. Song, S. A. Burns, J Opt Soc Am A 25, 3021 (Dec, 2008). 57. T. Y. P. Chui, H. Song, S. A. Burns, Investigative Ophthalmology & Visual Science 49, 4679 (Oct, 2008). 58. H. X. Song, Y. M. Zhao, X. F. Qi, Y. P. T. Chui, S. A. Burns, Optics Letters 33, 137 (Jan 15, 2008). 59. D. P. Biss et al., Optics Letters 32, 659 (Mar 15, 2007). 60. D. X. Hammer et al., Invest Ophth Vis Sci 49, 2061 (May, 2008). 61. C. E. Bigelow et al., Journal of the Optical Society of America a-Optics Image Science and Vision 24, 1327 (May, 2007). 62. Z. Zhong, B. L. Petrig, X. Qi, S. A. Burns, Opt. Express 16, 12746 (2008). 63. T. Y. P. Chui, H. Song, S. A. Burns, Invest. Ophthalmol. Vis. Sci., iovs.08 (June 14, 2008, 2008). 64. M. Pircher, R. J. Zawadzki, Expert Review of Ophthalmology 2, 1019 (2007). 65. K. M. Hampson, Journal of Modern Optics 55, 3425 (2008). 66. J. Porter, H. M. Queener, J. E. Lin, K. Thorn, A. Awwal, Adaptive Optics for Vision Science. (John Wiley & Sons, Inc., 2006). 67. N. Doble, D. R. Williams, Ieee J Sel Top Quant 10, 629 (May-Jun, 2004). 68. T. Y. Chen, C. W. E. Chiu, G. D. J. Su, Ieee Photonic Tech L 20, 830 (May-Jun, 2008). 69. I. Woong-Jung, Y. A. Peter, E. Carr, J. S. Wang, O. Solgaard, Ieee J Sel Top Quant 13, 162 (Mar-Apr, 2007). 70. D. Merino, C. Dainty, A. Bradu, A. G. Podoleanu, Opt Express 14, 3345 (Apr 17, 2006). 71. G. T. Kennedy, C. Paterson, Optics Communications 271, 278 (2007). 72. E. Daly, E. Dalimier, C. Dainty, paper presented at the MEMS/MOEMS Components and Their Applications III, San Jose, CA, USA, 2006. 73. T. Farrell, E. Daly, E. Dalimier, C. Dainty, paper presented at the MEMS Adaptive Optics, San Jose, CA, USA, 2007. 74. Y. Zhou, T. Bifano, C. Lin, paper presented at the MEMS Adaptive Optics, San Jose, CA, USA, 2007. 75. B. Potsaid, J. T.-Y. Wen, Journal of Micro/Nanolithography, MEMS and MOEMS 7, 021009 (2008). 76. K. Baker et al., Journal: Optics Letters, vol. 29, no. 15, September 1, 2004, pp. 1781-1783