Thermally actuated deformable mirrors JONATHAN WEE1*, MENTOR: HAMID HEMMATI2*, CO-MENTOR: YIJIANG CHEN2 Cornell University, Ithaca, New York 14853 Jet Propulsion Laboratory, California Institute of Technology, 4800 Oak Grove Drive, Pasadena, California 91109 *e-mail: [email protected], [email protected]

1 2

September 22, 2006

A thermally actuated deformable mirror is presented as

a

low-cost

alternative

for

active

optical

compensation. The objective is to affix an optimized array of actuators to a thin, low-cost, continuous mirror and achieve a maximum displacement between 50 to 100 µm. The actuator cells operate by absorbing heat from an infrared laser on one end and expanding against an attached mirror on the opposite end. Partially filling the actuators with a low-boiling point liquid dramatically improves their expansion efficiency. A Mylar membrane and rod structure function as the interface

between

the

cells

and

the

mirror.

Temperature control of the cell array balances the internal and external pressures, equilibrating the Mylar for maximum displacement potential. Currently, a 2531 µm displacement was demonstrated for a 3 x 3 element deformable mirror powered by a 100 mW laser.

Although cylindrical cell dimensions of 5 mm

diameter and 7 mm depth displayed the greatest actuation capability, shallower depths produced faster response times in exchange for shorter displacement.

Space Grant | JPL Summer 2006

F

ree space optical communications is a technology that is based on line of sight transmission of modulated light. Mechanical, thermal and atmospheric environments can distort the reception of these signals from space, affecting the ability to focus properly on them.1 The optical imperfections vary as the surroundings change, so the goal is to dynamically counteract the effects of telescope aberrations by reducing the wavefront error. Such active optical systems consist of a deformable mirror, image quality detector, and real-time control. These mirrors utilize an array of actuators to deform their reflective surface, maintaining an optimal shape for correcting time-varying distortions. Since thin mirrors can be manipulated to produce similar performance compared to static, thicker ones at lower expense, development of thermally actuated deformable mirrors may provide a cost effective, high quality, active optics solution. The proof-of-concept previously demonstrated to produce more than 20 µm of actuation for a single, segmented mirror is further developed.2,3 Implementation of continuous faceplate mirrors was the most significant improvement, benefiting from many advantages over the segmented kind, including smooth wavefront control, large degrees of freedom, and no edge diffraction effects. An important factor in the enhanced actuator efficiency was the addition of a low-boiling point liquid. Partially filling the actuators with this substance increased their expansion over ones filled only with air. Temperature control increased the Mylar’s performance. Operating at an actuator’s optimum temperature balances its internal and external pressures, maximizing the Mylar’s displacement potential as it is neither stretched convexly nor concavely. Finally, applying small rods between the Mylar and the mirror improves the membrane-mirror interface, reducing pinning effects and membrane stiffness due to large epoxy footprints. Therefore, optimization of the deformable mirror’s geometry, actuating liquid, operating conditions, and membrane to mirror interface improved actuator efficiency for a continuous mirror by up to 5 times for a given power input.

1

Cell wall a

Glass

Membrane

Absorbing film b

Light

Expansion Heat

Figure 1 Golay cell concept. The underlying principle is volume change due to temperature fluctuations. a, Main components of a typical Golay cell. b, Operation of a Golay cell: Light enters through the glass and the absorbing film converts some of its energy into heat, which causes the gas or liquid in the cell to expand.

GOLAY CELLS Marcel Golay developed an infrared detector, named the Golay cell, based on the relationship between small temperature and volume changes inside a gas chamber. As a cell converts incoming light into heat, its internal gases absorb this warmth and expand against a membrane, which is displayed in figure 1. Golay reflected light off a mirror attached to the membrane into a photodiode, resulting in measurable voltage fluctuations associated with the mirror displacement. Golay cells are naturally appropriate for deformable mirrors since actuation on the order of microns is sufficient to compensate for optical distortions. The implementation consisted of a 100 µm microscope slide cover to allow light into the cell, 10 µm Mylar membrane, 70 µm copper absorber plate, 6061 aluminum cell wall, and 1064 nm infrared laser to power the cell. In addition, the cells were partially filled with 3M’s NOVEC HFE-7000, a relatively inert, environmentally friendly, low-boiling point liquid. It is a lowtemperature, heat transfer medium that has a greater coefficient of expansion than air does, and evenly distributes temperature for superior control precision. Cell geometries that were wider and shorter generally provided more actuation and faster response times. Ideally, a smaller volume of material should be easier to heat, but if a cell is too small, its expansion soon becomes limited by the Mylar and its elasticity. Small diameter cells require a higher power laser to achieve actuation similar to ones with larger diameter. Mylar’s elasticity is another limiting factor, because plastic deformations are irreversible and reduce repeatability. As a result of these limitations, small cell volumes resulted in faster response times, but lesser displacement. The amount of liquid in each cell, however, followed the opposite trend. With no fluid, there was hardly any actuation under low power. Higher levels of liquid resulted in more expansion at any given power input up to a certain height, past which more liquid did not produce any gain in actuation. These geometry restrictions allowed for maximum displacement only at an optimal cell diameter, cell depth, and amount of actuating liquid. Golay cell operation requires some moderate thermal control. If the cell wall temperature is not steady, then there 2

will be volume changes associated with the temperature fluctuations that will give a false measure of actuation. In addition to being thermally stable, each cell has an optimum operating temperature that is set when it is sealed. If a cell is hotter than its surroundings when it is sealed, its internal pressure will be greater than atmospheric and the membrane will be convex or “bubbled out”. Similarly, a cell sealed at a temperature less that that of its surroundings will have less internal pressure and the membrane will be concave or “bubbled in”. Since Mylar does not have high elasticity, it cannot actuate further once bubbled out. Therefore, the membrane has its maximum displacement potential when it is flat at the cell’s “flat temperature”, which is obtained through thermal control to balance the internal and external pressures. ROD INTERFACE A rod interface between the membrane and the mirror that diminishes stiffness and pinning effects is adopted.4,5 Originally, drops of 5 minute epoxy were placed on sections of Mylar covering the cells to connect the mirror to the actuators. When the epoxy cured, however, it had already spread over a significant portion of each cell’s diameter, stiffening the Mylar and reducing actuation potential. As a result of this thick connection, the areas of the mirror directly above the epoxy were rigid and flat, limiting the smoothness of the mirror and introducing surface errors. Therefore, the rod interface uses thin rods for their small footprints, which reduces membrane stiffness and pinning effects on the mirror as shown in figure 2. This interface method also thermally decouples the mirror from the cells, so structural heating will not cause the mirror to warp. Although the rods are still held in place by epoxy, their diameter is much smaller than that of the cells’, so there will be a significantly less stiffening effect on the membrane. The rods that were thinner and stiffer seemed to perform better. However, rods that were too thin either broke during application or buckled in actuation, and ones that were too stiff created a dimple in the membrane while actuating, reducing the Mylar’s displacement potential. Another consideration was choosing the rod length, desired to be as short as possible, which was limited by the tweezer application method. Selecting the material was also challenging since it had to be stiff even when cut into small cylinders. Certain types of thermocouple wire were good material choices, but there was not enough time to acquire them, so insulated copper wire was used instead.

a

Mirror

b

Rods Pinning Effect

Mylar

No Pinning Effect

Figure 2 Rod interface concept. Eliminates surface errors and membrane stiffness. a, With a thick mirror to membrane connection such as epoxy, the sections of the mirror directly over the connection are rigid and flat, introducing error. This also stiffens the membrane and reduces actuation. b, A thin interface reduces the surface errors and membrane stiffness, allowing for smoother wavefront control and further actuation potential.

Space Grant | JPL Summer 2006

DISCUSSION Optimization of the Golay cells improved actuation efficiency for a continuous mirror by up to 5 times at 100 mW of power input. The best cell geometry is cylindrical in shape, with a 5 mm diameter, 7 mm depth, and 7 mm spacing from center to center, or pitch. The pitch was designed to allow for a 2 mm gap between each cell for easy epoxy application to attach the membrane and microscope slide cover. The liquid level that provided the most actuation on average was 3/4 full. However, at a cell’s flat temperature, liquid levels ranging from 1/4 to 3/4 full performed equally well, so the amount of liquid matters less at Mylar’s equilibrium. The cells were manually sealed, so body heat made them hotter and increased their internal pressure. To compensate, the actuator structure had to be cooled with average flat temperatures around 21 or 22 degrees Celsius. The rod interface worked as planned, reducing Mylar stiffness and pinning effects. The final rod dimensions were .47 mm in diameter and 1.75 mm in length. Interferometry tests of the Golay cells with a 1064 nm, 100 mW infrared laser showed visible actuation using 270 µm thick segmented mirrors, and about 30 µm of displacement in the center cell of a 3 x 3 element, 100 µm thick, continuous mirror. With increased actuator efficiency and the implementation of continuous mirrors to the proof-of-concept done by previous Space Grant students, Golay cell actuated deformable mirrors have been shown to be a feasible, low-cost alternative for active optics compensation. However, this type of thermal actuation still needs many improvements before it can compete with current commercial deformable mirrors, including better structural temperature control at high laser powers, response times on the order of kilohertz, improved actuator coupling control, an elastic membrane for larger displacements, thinner and stiffer rod material such as thermocouple wire, better rod application techniques to keep the rods vertical, and XY laser scanner integration. METHODS CONSTRUCTION OF THE GOLAY CELLS

1) Cut a 6061 aluminum block 34 mm x 44 mm x 7 mm. 2) Sand the edges flat for easier clamping and remove edge burs. 3) Sketch a 3 x 3 array of holes with their centers 7 mm apart, 10 mm from the top and side edges, and 20 mm from the bottom edge for attaching to an optical setup. 4) Start the holes with a hole punch or center drill. 5) Drill out the holes to a 5 mm diameter and remove the edge burs. 6) Clean the structure with an all purpose cleaner such as Simple Green. 7) Paint a 70 µm thick copper sheet black on both sides and allow it to dry. 8) Cut the copper sheet into nine, 3.5 mm x 3.5 mm squares. 9) Glue the squares into the middle of the cell holes with 5 minute epoxy, allowing them to completely dry. 10) Cut a piece of 10 µm thick Mylar to fit over all the holes and glue it to one side of the block with 24 hour epoxy and allow it to completely dry.

Space Grant | JPL Summer 2006

11) Apply a thin layer of 5 minute epoxy on the clean side of the block for sealing the cells with the microscope slide cover. Be sure to fully cover the gaps between the cells for a tight seal. Do not seal the cells at this time. 12) Fill the cells 3/4 full with 3M NOVEC HFE-7000 using a syringe or pipette. Some of the liquid may evaporate and need to be refilled. 13) Seal the cells. Place a 100 µm thick microscope slide cover over the cells and hold it down only over the aluminum supported areas until it dries. Try to squeeze out any air bubbles which can lead to leakage. Do not press on the slide cover directly over the holes or it will break. 14) Find the array’s flat temperature by regulating the block’s temperature using a power resistor to heat it up or a Thermal Electric Cooler to cool it down. When the Mylar is neither bubbled up nor down, the cell array will be at equilibrium. 15) Secure an XYZ positioner to a table using suction, magnets, or screws. 16) Place a pair of tweezers in a small screw clamp, then tape or glue one finger of the tweezers to the clamp. 17) Attach this clamp to the XYZ positioner and orient it so the tweezers are vertical. 18) Cut 9 pieces of insulated copper wire 1.75 mm long with .47 mm diameters. Make sure they are all the same height and have a flat or rounded end so it does not poke through the Mylar. 19) Place one of the rods (wires) into the tweezers and screw the clamp down until the rod is secure and vertical as well. 20) Place the cell array with the membrane side up under the tweezers and practice with the XYZ positioner to ensure the rod is directly over the center of a cell and vertical. 21) Put a very small drop of 5 minute epoxy on the free end of the rod. The glue should only cover the end of the rod and not drip down the shaft, or there will be too much epoxy. 22) Lower the XYZ positioner so that the rod and epoxy just touch the Mylar. If this is done slowly and properly, the glue will spread to fill any gaps between the rod and Mylar without breaking the membrane. If the rod is pushed down too far, the Mylar will bubble in a small amount, but further than this the membrane will tear. Allow the glue to completely dry. 23) Unscrew the clamp and use the XYZ positioner to raise the tweezers out of the way. 24) Repeat steps 19 – 23 for the remaining rods. 26) Place a very small drop of 5 minute epoxy on the free end of the rods. Like step 21, make sure the glue does not drip down the shaft of the rods. 27) Attach a 100 µm thick mirror over the rods. We used an aluminized microscope slide cover. Be careful to glue the nonaluminized side to the rods. If the rods are not all the same length, the shorter ones will not connect with the mirror. Allow the glue to completely dry. You now have a 9 element Golay cell deformable mirror. A pictorial representation of important steps is presented in figure 3.

3

a

b

c

d

e

f

g

h

2 methods of applying rods Figure 3 Golay cell construction. a, Marked aluminum block with hole punched starters. b, Diced copper sheet painted black. c, Black absorbing sheets

glued in the center of each cell. d, Mylar membrane glued to one side. e, Sealed Golay cell array. f, XYZ positioner and hook clip for horizontal application of rod interface. g, XYZ positioner and tweezers for vertical application of rod interface. h, Completed 9 element, Golay cell deformable mirror.

TEST SETUP

through the TEC or power resistor to meet the array’s flat temperature, which was measured by the thermocouple.

A Michelson interferometer was used to measure actuation of the Golay cells. This device splits a beam of light into two paths and recombines them in an interference pattern. For the actual setup, one of these beams reflected off a reference mirror and the other reflected off the deformable mirror, forming a fringe pattern recorded by an oscilloscope. The Golay cell interferometer test setup is shown in figure 4. Since a 633 nm HeNe laser was used for the interferometry, multiplying the number of passing fringes by the half-wavelength (.3165) will return the microns of actuation.

Thermocouple Power resistor

Thermal Electric Cooler Reference mirror Figure 5 Thermal control. Sealed Golay cell array with power resistor,

Beam splitter Deformable mirror

Thermal Electric Cooler, and thermocouple in finding the flat temperature.

HAND WARMING EFFECT

Optical train (green arrows) Figure 4 Michelson interferometer. Golay cell test setup.

To keep the Golay cell array at its optimal temperature, a Thermal Electric Cooler (TEC), power resistor, and thermocouple were used to cool down, heat up, and measure structural temperature respectively as in figure 5. A TEC operates like a reverse thermocouple, using current to induce a temperature change between two dissimilar metals. The hot side of the cooler requires a small heat sink to wick heat away that would otherwise build up and backflow into the equipment being cooled. The cold side of the cooler was attached to one end of the cell array with Kapton tape because of its thermally insulating properties. Similarly, the power resistor was glued onto another free end of the structure and heated up as current passed through it. Electric current was slowly controlled 4

Before the Golay cells were reliably controlled using the TEC and resistor, they were warmed using body heat. As the interferometry tests were conducted, the surroundings cooled off the array, causing contraction of the cells that showed up as false actuation. If the structural temperature changes during a test, the fringes will register in a steady, constant period, constant amplitude fashion. This is a result of the array steadily cooling down from its warmed temperature to that of its surroundings. This problem was solved, however, through steady thermal control using the TEC and power resistor. References 1. 2. 3. 4. 5.

H. Hemmati and Y. Chen, Optics Letters, Vol. 31, No. 11 (2006). C. Scott, Actuation of Deformable Mirrors Using Laser Controlled Pistons (2005). A. Lint, Adaptive Optics: Arroyo Simulation Tool and Deformable Mirrors Using Laser Controlled Pistons (2005). R. Hamelinck, N. Rosielle, M. Steinbuch, N. Doelman, Large Adaptive Deformable Mirror with High Actuator Density, Proc. of SPIE Vol. 5490 (2004). R. Hamelinck, N. Rosielle, M. Steinbuch, N. Doelman, Large Adaptive Deformable Mirror: Design and First Prototypes, Proc. of SPIE 589418 (2005).

Acknowledgements I thank my mentor, Hamid Hemmati, for sharing his vision to pursue new technologies, my comentor, Yijiang Chen, for his guidance and undying motivation, and all of my collaborators in the Optical Communications Group for their support. This research was conducted at the Jet Propulsion Laboratory, California Institute of Technology, as part of a summer internship program sponsored by NASA’s Space Grant College.

Space Grant | JPL Summer 2006

APPENDIX PELLICLE MIRROR

One idea to increase the actuation of the continuous mirror involved using a thin, membrane mirror. Because commercial pellicle membranes are expensive, 10 µm thick Mylar was used as the membrane instead. The goal was to construct a Golay cell where a stretched Mylar membrane could act as a mirror. The construction of the pellicle version of the Golay cells is very similar to the general method described in the paper above, however, the aluminum block is much bigger so it can support the pellicle ring mount. Also, the rod interface is glued to the mirror before attaching it to the actuators, which is opposite of the normal method. The decision to attach the rods to the mirror first was motivated by an attempt to place more control on the delicate side of the interface, the rod to mirror bond, in exchange for some control on the rugged side of the interface, the rod to membrane bond. Special care had to be taken when applying the rod interface. In order to know where to place the rods on the mirror, a template of the cell centers had to be made from the aluminum block, reversed, and carefully copied onto the back of the mirror. This template was made from paper, with the cell centers defined by horizontal and vertical lines along with a notch that portrays the template’s orientation. The paper was hole-punched at the line intersections so that a marker could precisely blot points on the back of the mirror. The rods were cut long enough to match the depth of the pellicle ring mount, and the mirror was made by stretching Mylar over the ring mount, using 5 minute epoxy to secure it in place. If the Mylar had some wrinkles in it after the glue already dried, they were easily ironed out by using a heat gun or hairdryer and slowly moving it closer to the Mylar until the membrane appeared to stiffen up. After attaching the rod interface to the back of the mirror, the mirror was flipped around and the interface was glued to the aluminum block with the notch on the pellicle ring mount matching the one made on the block. Finally, a few drops of epoxy were added between the ring mount and the block to hold the mount in place. An outline of the general construction steps are shown in figure 6. Although a thinner mirror should produce more actuation, the Mylar pellicle mirror concept was not successful because the mirror quality was insufficient for interferometry tests. Instead of having a changing interference pattern, the fringes were fixed and did not move at all. Further development of methods to stretch and attach a membrane to the ring mount is required for interferometry validation. One possible path the pellicle mirror could take is to have interchanging parts. If the rod interface were connected to Mylar stretched on both sides of the ring mount, then there will be no need to glue the rods to the membrane on the aluminum block. The mount could be interchangeable and different mounts could have membranes of other thicknesses or rods of other materials. Having interchanging ring mounts on a standard, optimized Golay cell block would provide a robust testing apparatus for future research.

Space Grant | JPL Summer 2006

a

b

c

d

e

f

Figure 6 Pellicle mirror. a, Modified Golay cell aluminum block with black, copper absorbing squares. b, Sealed Golay cell with marks to indicate the cell centers. c, Paper template created with notch indicating the top, and holes at the cell centers. d, Template is copied onto the back of the pellicle mirror. e, Rod interface is attached to the back of the mirror. f, Completed, modified Golay cell structure with a Mylar pellicle membrane as the mirror. FRINGE PATTERNS

Fringe patterns were recorded by an oscilloscope when testing the actuation potential of the Golay cells. Most of these preliminary tests were done using segmented mirrors and only had 5 mW of power input. Throughout the series of tests done, there were effects noted on the recorded fringe patterns, and the following will show how to distinguish some of these effects. Acoustic noise was an effect common to all of the fringe patterns. It is discernable from fuzzy waveform lines that show a general trend, but have small amplitude fluctuations. The noise was not able to be controlled and varied in intensity from day to day, sometimes it even changed throughout the day. The hand warming effect is what originally caused the author to notice that different structural temperatures had an effect on actuation. Although this effect creates false actuation, a similar effect can be noticed even with the thermal control solution if interferometry tests were conducted while the structural temperature was reaching a steady state value. The hand warming effect is very easy to distinguish, and is characterized by constant period, constant amplitude waveform. While the temperature of the Golay cell is slowly and steadily equilibrating with that of its surroundings, the cell is also steadily contracting, creating false actuation that appears at a constant rate. True thermal actuation will slow down as it reaches completion so the amplitude should decay and the period should grow over time. A comparison of fringe pattern effects is depicted in figure 7.

5

a

b

c

d

Figure 7 Fringe patterns.

Typical fringe patterns recorded on an oscilloscope. a, Heavy acoustic noise. b, The hand warming effect with a constant amplitude, constant period waveform. c, Low-noise, decaying amplitude, and growing period. This is an example of a valid fringe pattern. d, Another valid fringe pattern with slightly more noise.

Figure 8 Golay cell CAD drawing. Golay cell continuous mirror design

using SolidWorks.

6

Space Grant | JPL Summer 2006

Block 6

row 1 1 1 2 2 2 3 3 3

row 1 1 1 2 2 2 3 3 3

7mm

5mm diam

Absorber

Depth of liquid

black black black black black black black black black

3/4 full 1/2 full 1/2 full 1/2 full 1/2 full 3/4 full 3/4 full 3/4 full Empty

col 1 2 3 1 2 3 1 2 3

col 1 2 3 1 2 3 1 2 3

No. of fringes Temp=22 12.0 10.0 11.0 11.0 7.5 8.0 13.0 5.0 8.0

Stroke (microns) Temp=22 3.8 3.2 3.5 3.5 2.4 2.5 4.1 1.6 2.5

No. of fringes

Strokes (microns)

Converge

Temp=23 7.5 17.5 7.0 4.5 6.0 5.0 8.5 4.5 8.0

Temp=23 2.4 5.5 2.2 1.4 1.9 1.6 2.7 1.4 2.5

Temp=23 Converging Converging Converging Yes Converging Converging Converging Yes Yes

P=5mW

Converge Temp=22 Yes Converging Yes Yes Converging Converging Yes Converging Yes No. of fringes

Stroke (microns)

Converge

Temp=25 3.5 4.5 5.0 2.0 3.0 3.5 4.5 3.0 7.0

Temp=25 1.1 1.4 1.6 0.6 0.9 1.1 1.4 0.9 2.2

Temp=25 Converging Converging Converging Yes Yes Converging Converging Yes Yes

Table 1 Table of tests. Tests on a typical 5 mm diameter by 7 mm depth Golay cell.

Space Grant | JPL Summer 2006

7