U.S. patent application number 11/260999 was filed with the patent office on 2006-02-23 for deformable mirror method including bimorph flexures.
Invention is credited to Micheal Albert Helmbrecht.
Application Number | 20060038103 11/260999 |
Document ID | / |
Family ID | 32872718 |
Filed Date | 2006-02-23 |
United States Patent
Application |
20060038103 |
Kind Code |
A1 |
Helmbrecht; Micheal Albert |
February 23, 2006 |
Deformable mirror method including bimorph flexures
Abstract
An apparatus comprising a substrate; and a platform elevated
above the substrate and supported by curved flexures. The curvature
of said flexures results substantially from variations in intrinsic
residual stress within said flexures. In one embodiment the
apparatus is a deformable mirror exhibiting low
temperature-dependence, high stroke, high control resolution, large
number of degrees of freedom, reduced pin count and small
form-factor. Structures and methods of fabrication are disclosed
that allow the elevation of mirror segments to remain substantially
constant over a wide operating temperature range. Methods are also
disclosed for integrating movable mirror segments with control and
sense electronics to a produce small-form-factor deformable
mirror.
Inventors: |
Helmbrecht; Micheal Albert;
(Lafayette, CA) |
Correspondence
Address: |
STATTLER JOHANSEN & ADELI
P O BOX 51860
PALO ALTO
CA
94303
US
|
Family ID: |
32872718 |
Appl. No.: |
11/260999 |
Filed: |
October 28, 2005 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
10703391 |
Nov 7, 2003 |
|
|
|
11260999 |
Oct 28, 2005 |
|
|
|
60425049 |
Nov 8, 2002 |
|
|
|
60425051 |
Nov 8, 2002 |
|
|
|
Current U.S.
Class: |
248/346.01 ;
310/309 |
Current CPC
Class: |
G02B 26/0825 20130101;
G02B 26/06 20130101 |
Class at
Publication: |
248/346.01 ;
310/309 |
International
Class: |
H02N 1/00 20060101
H02N001/00; A47B 91/00 20060101 A47B091/00 |
Claims
1. (canceled)
2. A method for fabricating a microelectromechanical (MEMS)
structure, the method comprising: forming a platform connected with
a set of one or more bimorph flexures; and for each bimorph flexure
in the set of bimorph flexures: forming a first layer comprised of
a first material; and forming a second layer comprised of a second
material, the first and second materials having particular
intrinsic residual stress (IRS) characteristics and coefficients of
thermal expansion (CTEs), each bimorph flexure having a curvature
resulting from a first component proportional to the difference in
IRS characteristics of the first and second materials and a second
component proportional to the difference in CTEs of the first and
second materials, the first component being larger than the second
component.
3. The method of claim 2, wherein the curvature of each formed
bimorph flexure results predominantly from the first component.
4. The method of claim 2, wherein the first material comprises
silicon and the second material comprises silicon nitride, or the
first material comprises polysilicon and the second material
comprises ceramic, SiC, or silicon nitride (SixNy).
5. The method of claim 2, wherein forming the second layer
comprises forming the second layer external to the first layer.
6. The method of claim 2, wherein forming the second layer
comprises forming the second layer to extend over a portion of the
first layer that is less than the entire length of the first layer
and forming the second layer to be affixed to the first layer along
the entire length of the second layer.
7. The method of claim 2, wherein the first and second layers are
formed under conditions that produce substantially different
intrinsic residual stress (IRS) characteristics in the first and
second materials.
8. The method of claim 7, wherein: forming the first layer
comprises tuning the residual stress of the first layer; and
forming the second layer comprises forming the second layer under a
specific ratio of the reactant gasses, deposition pressure, and
deposition temperature to produce a desired residual stress of the
second layer.
9. The method of claim 7, wherein the first and second materials
comprise polysilicon.
Description
CROSS-REFERENCE TO RELATED APPLICATIONS
[0001] This application claims the benefit of U.S. Provisional
Patent Application No. 60/425,049 entitled Reduced Rotation MEMS
Deformable Mirror Apparatus and Method, and U.S. Provisional Patent
Application No. 60/425,051 entitled Deformable Mirror Method and
Apparatus Including Bimorph Flexures and Integrated Drive, both
filed Nov. 8, 2003.
BACKGROUND OF THE INVENTION
[0002] 1. Field of the Invention
[0003] This invention relates to a methods and structures for
elevating a platform above a substrate and for producing a
controlled motion of that platform. It also relates to MEMS
deformable mirror ("DM") arrays, and more particularly to
long-stroke MEMS deformable mirror arrays for adaptive optics
applications.
[0004] 2. Description of the Related Art
[0005] Adaptive optics ("AO") refers to optical systems that adapt
to compensate for disadvantageous optical effects introduced by a
medium between an object and an image formed of that object. Horace
W. Babcock proposed the concept of adaptive optics in 1953, in the
context of mirrors capable of being selectively deformed to correct
an aberrated wavefront. See John W. Hardy, Adaptive optics for
astronomical telescopes, Oxford series in optical and imaging
sciences 16, Oxford University Press, New York, 1998. Since then,
deformable mirrors (DM) have been proposed for a variety of AO
applications, although they have yet to be implemented in many such
proposed applications.
[0006] The general operation of a DM is shown schematically in FIG.
1, in which a DM 100 reflects an aberrated wavefront 105, resulting
in a desired planar wavefront 110. The DM shape is dynamically
adapted to correct the path-length variations of the inbound
aberrated wavefront. That is, by selectively deforming the mirror
to decrease or increase the path length for specific portions of
the aberrated wavefront, the aberrations in the reflected wavefront
are corrected. The amount of local displacement needed of the DM
surface is generally approximately equal to half the path-length
variations in the aberrated wavefront. The exact scale factor
depends on the angle at which the aberrated wavefront strikes the
deformable mirror.
[0007] A prior art AO system is shown schematically in FIG. 2. This
example is particularly related to an astronomical telescope
application, but the general principles of AO shown here are
illustrative of other applications. In FIG. 2, an aberrated
wavefront 105 enters the optical system 205 where it is modified as
it reflects off a DM 100. Aberrations in the wavefront reflected
from the DM are the error signal for a computer-controlled feedback
loop. The reflected wavefront 110 enters a dichroic beam splitter
220; the infrared wavelengths pass to a science camera 225 and the
visible wavelengths reflect toward a wavefront sensor 230. The
wavefront sensor measures the wavefront slope at discrete points
and sends these data to a wavefront reconstructor 235. The
wavefront reconstructor 235 determines the remaining wavefront
aberrations in the corrected wavefront. An actuator control block
240 calculates actuator drive signals to correct the remaining
wavefront errors, which are sent from the block 240 to the DM 100,
thus dosing the feedback loop. In this way, the DM is continuously
driven in such a way as to minimize the aberrations in the
reflected wavefront, thereby improving image resolution at the
science camera.
[0008] AO systems have been proposed and demonstrated for improving
resolution in a number of imaging applications. In astronomy, for
example, AO has been used to correct aberrations introduced by
motion of the atmosphere, allowing ground-based telescopes to
exceed the resolution provided by the Hubble Space Telescope under
some observing conditions. In the field of vision science, AO has
been shown to offer benefits, for example, for in-vivo retinal
imaging in humans. Here, AO systems can compensate for the
aberrations introduced by the eye, improving lateral image
resolution by a factor of three and axial resolution by a factor of
ten in confocal imagers. This has allowed individual cells to be
resolved in living retinal tissue, a capability that was not
present before the advent of AO.
[0009] In addition to improving image resolution, AO systems can be
used to improve confinement of a projected optical beam traveling
through an aberrating medium. Examples of applications in this
category are free-space optical communication, optical data storage
and retrieval, scanning retinal display, and laser-based retinal
surgery.
[0010] A number of characteristics are commonly used to compare
performance of DM designs. Fill-factor is the fraction of the DM
aperture that is actively used to correct wavefront aberrations.
Mirror stroke is the amount of out-of-plane deformation that can be
induced in the DM surface. The number of degrees-of-freedom is a
measure of the spatial complexity of the surface shapes the DM is
capable of assuming and is related to the number of individual
actuators that are used to deform the mirror surface. DM aperture
diameter, DM device size, control resolution, operating temperature
range, power consumption, frequency response and price are also
generally considered when selecting a DM for a given application.
For example, astronomical imaging typically requires mirror stroke
in the range of a few micrometers, frequency responses in the
kilohertz range and aperture sizes on the order of a few
centimeters to a few meters. Systems for imaging structures in the
human eye, by contrast, generally require mirror stroke on the
order of 10 micrometers or greater, frequency responses in the tens
to hundreds of Hertz range, and aperture sizes on the order of one
centimeter or less.
[0011] Despite the advantages outlined above, AO has not been
universally adopted, even in the aforementioned applications. Two
important factors that have impeded the widespread adoption of AO
are the high cost and limited stroke of available DMs.
[0012] DM designs can be broadly divided into two classes;
continuous-face-sheet designs and segmented designs.
Continuous-face-sheet DMs have a reflective surface that is
continuous over their whole aperture. The surface is deformed using
actuators, typically mounted behind it, that push or pull on it to
achieve a desired deformation. This type of DM has been
implemented, for example, by mounting an array of piezoelectric
actuators to the rear surface of a somewhat flexible glass or
ceramic mirror. Because the optical surface is continuous and
rather inelastic, large actuation forces are required to deform the
mirror, and the resulting mirror stroke is small, typically less
than 5 micrometers. The continuous surface also means that the
deformation produced by each actuator is not tightly confined to
the area of the mirror directly connected to it, but instead may
extend across the whole mirror aperture, making precise control of
the overall mirror deformation problematic. Because of the way they
are constructed, such DMs are also comparatively large, having
apertures on the order of 50 mm or greater. This large size
precludes their deployment in many optical systems that might
otherwise benefit from AO. Their fabrication methods also make
these DMs expensive to manufacture and do not permit easy
integration of control electronics into the DM structure.
[0013] A number of continuous-face-sheet DMs using microfabrication
techniques that offer the potential to reduce DM size and cost have
been created. Vdovin and Sarro, in "Flexible mirror micromahined in
silicon", Applied Optics, vol. 34, no. 16 (1995), disclose a DM
fabricated by assembling a metal-coated silicon nitride membrane
above an array of electrodes that are used to deform the membrane
by electrostatic attraction.
[0014] Bifano et al. disclose an alternative microfabricated
continuous-face-sheet DM in "Microelectromechanical Deformable
Mirrors", IEEE Journal of Selected Topics in Quantum Electronics,
vol. 5 no. 1 (1999). Their design relies on the removal of a
sacrificial layer to create cavities underneath the mirror surface
that define the maximum travel range of each mirror actuator.
[0015] U.S. Pat. No. 6,384,952 to Clark et al. (2002) discloses a
continuous-face-sheet DM that employs a mirrored membrane
fabricated, for example, from metal-coated silicon nitride and
actuated by an array of vertical comb actuators disposed underneath
the membrane. Use of vertical comb actuators can provide higher
force for a given applied voltage than the parallel plate
electrostatic actuators used in other continuous-face-sheet
designs.
[0016] In contrast to the continuous-face-sheet designs discussed
above, segmented DM designs divide the DM aperture into a number of
generally planar mirror segments, the angle and height of each
segment being controlled by a number of actuators. Segmented
designs are advantageous in that they allow the area of influence
of each actuator to be tightly confined, simplifying the problem of
driving the mirror to a particular desired deformation. Segmenting
the mirror surface also eliminates the need to deform a
comparatively inflexible optical reflector to produce a desired DM
surface shape. Rather, the individual mirror segments are tilted,
raised and lowered to form a piecewise approximation of whatever
deformation is required to correct the aberrations of the incoming
wavefront. Segmenting the surface can therefore result in a lower
force requirement for a given surface deformation, enabling the
high-stroke DMs that are needed for many AO applications.
[0017] A number of inventors have disclosed segmented DM designs
that may be constructed using microfabrication techniques. U.S.
Pat. No. 6,175,443 to Aksyuk et al. (2001) discloses an array of
conductive mirror elements, connected together by linking members
that act as supports, suspending the mirror array above an
actuating electrode. These linking members also serve to keep the
mirror array in an approximately planar configuration when no
actuating voltage is applied. Energizing the electrode results in
an attractive force between it and the mirror segments, deforming
the array into a curved configuration.
[0018] U.S. Pat. No. 6,028,689 to Michalicek et al. (2000)
discloses an array of mirror segments attached to a substrate by
posts, each segment capable of tilting about two axes and also
moving vertically, perpendicular to the array, under the influence
of applied control voltages.
[0019] U.S. Pat. No. 6,545,385 to Miller et al. (2003) discloses
methods for elevating a mirror segment above a substrate by
supporting it on flexible members that can bend up out of the
substrate plane. This provides a large cavity underneath the mirror
segment, not limited by the thickness of the sacrificial materials
used in its fabrication, and offering the potential for large
mirror stroke.
[0020] Helmbrecht, in "Micrmirror Arrays for Adaptive Optics", PhD.
Thesis, University of California, Berkeley (2002), discloses a
segmented DM for use in AO applications, that exhibits high
fill-factor, high mirror quality and offers the potential for high
mirror stroke.
SUMMARY OF THE INVENTION
[0021] It is therefore an object of the present invention to
provide improved methods and structures for elevating a platform
above a substrate and for precisely controlling the tip, tilt and
piston motion of that platform.
[0022] A further object of the invention is to provide a
high-degree-of-freedom DM which can be used to compensate for large
optical wavefront aberrations, without the need for temperature
control or monitoring.
[0023] Another object of the invention is to provide a
high-degree-of-freedom, high-stroke DM with integrated control
electronics in a small form-factor configuration.
[0024] A further object of the invention is to provide a
high-degree-of-freedom, high-stroke DM with integrated sense
electronics in a small form-factor configuration.
[0025] Yet another object is to provide a high-degree-of-freedom DM
with a greatly reduced control-pin count.
[0026] A further object of the invention is to provide a
small-form-factor DM that can be used in clinical ophthalmic
instruments to correct wavefront aberrations of the human eye.
[0027] A further object of the invention is to provide a
high-degree-of-freedom, high-stroke DM that can be fabricated at
low cost.
[0028] A further object of the invention is to provide a
temperature-insensitive, high-fill-factor, segmented
piston-tip-tilt DM, having segments with improved optical
flatness.
[0029] Yet another object of the invention is to provide a
highly-reliable DM, capable of operating over many millions of
actuation cycles.
[0030] A further object of the invention is to provide a
high-degree-of-freedom DM comprising actuators that may be operated
largely independently, in order to provide correction for different
areas of an optical wavefront.
[0031] A further object of the invention is to provide a DM that
can be batch fabricated using IC-compatible fabrication methods and
materials.
[0032] A further object of the invention is to provide a
high-degree-of-freedom DM with reduced power consumption.
[0033] In accordance with the above objects, the invention, roughly
described comprises an apparatus including a substrate and a
platform elevated above the substrate and supported by curved
flexures, wherein the curvature of said flexures results
substantially from variations in intrinsic residual stress within
said flexures.
[0034] In another embodiment, the invention comprises a tiled array
of mirror segments, each supported by a number of curved flexures
attached, at one end, to the underside of the segment and, at the
other end, to a substrate. A number of independently addressable
actuators are used to apply forces to each mirror segment, causing
it to move in a controlled manner. The application points of the
actuating forces and the locations of the support flexures are
placed so as to allow each segment to be tilted about two distinct
axes substantially parallel to the substrate and translated along
an axis substantially perpendicular to the substrate. The invention
may optionally include electronic circuits embedded in the
substrate for the purpose of addressing the individual actuators
and/or sensing the state of a given mirror segment. The invention
includes methods and structures for improved flexures for
supporting and elevating the segments above the substrate. More
particularly, the invention provides methods and apparatus for
fabricating mirror segments supported by curved flexures, the
curvature of which is induced, principally or entirely, by
variations in intrinsic residual stress through the thickness of
the flexure material or materials. The invention also includes
methods for separately fabricating the MEMS portion of the
inventive apparatus and the electronics portion, and then
integrating the two to form the apparatus.
BRIEF DESCRIPTION OF THE DRAWINGS
[0035] FIG. 1: Illustration of prior art use of a deformable mirror
to correct an aberrated wavefront.
[0036] FIG. 2: Illustration of prior art adaptive optic (AO)
system.
[0037] FIG. 3A: Partial cutaway perspective view of a first
embodiment of the invention.
[0038] FIG. 3B: Perspective view of the improved flexure according
to a first embodiment of the invention.
[0039] FIG. 4A: Flow diagram of the process steps required to
fabricate a first embodiment of the invention.
[0040] FIG. 4B: Schematic cross-sections through structures
fabricated at various process steps in a first embodiment of the
invention.
[0041] FIG. 5: Schematic cross-section through the MEMS structures
for a single mirror segment in a first embodiment of the
invention.
[0042] FIG. 6: Schematic cross-section through the portion of the
CMOS substrate underlying a single mirror segment in a first
embodiment of the invention.
[0043] FIG. 7A: Schematic cross-section through a single mirror
segment and underlying structures before MEMS release and
passivation layer removal in a first embodiment of the
invention.
[0044] FIG. 7B: Schematic cross-section through a single mirror
segment and underlying structures after MEMS release and
passivation layer removal, in a first embodiment of the
invention.
[0045] FIG. 8: Table of coefficients of thermal expansion for
several candidate materials for construction of curved
flexures.
[0046] FIG. 9: Partially exploded perspective view of a third
embodiment of the invention.
DETAILED DESCRIPTION
[0047] Methods and structures for elevating one or more platforms
above a substrate and for controlling the tip, tilt and piston
motion of those platforms with high precision are hereinafter
described. Several embodiments are described in which a plurality
of such platforms are tiled to form a large-stroke segmented
piston-tip-tilt deformable mirror
[0048] FIG. 3A shows a partial cutaway perspective view of a first
embodiment of a DM incorporating the improved methods and
structures. The DM is formed on a substrate 300, which may be a
silicon wafer or chip containing embedded addressing and sensing
circuits (not shown). On top of the substrate 300 are formed a
number of control electrodes 370 that are electrically isolated
from one another and electrically connected to the embedded
addressing and sensing circuits. In the first embodiment, the
control electrodes 370 are arranged in groups of three and are
rhombic in shape, so that the footprint of each group is
essentially hexagonal. Disposed around each group of three control
electrodes 370, are three conductive ground pads 310, fabricated
from the same material as the control electrodes 370. The ground
pads 310 are electrically isolated from the control electrodes 370
and electrically connected to a ground plane or to circuits
embedded in the substrate 300. Attached to one end of each ground
pad 310 is a first anchor portion 350 of a flexure 320. The
flexure, in the first embodiment comprises two layers, a first
flexure layer 330 formed from conductive polycrystalline silicon
and a second flexure layer 340 formed from silicon nitride (SixNy).
The first anchor portion 350 is both mechanically and electrically
connected to the ground pad 310 so that the conductive first
flexure layer 330 is held at the same electrical potential as the
ground pad 310. The second flexure layer 340 is rigidly attached to
the underside of the first flexure layer 330 and extends over a
portion of the length of the flexure 320. The purpose of the second
flexure layer is to provide a residual stress difference between
the top and bottom portions of the flexure 320, causing the flexure
320 to bend up out of the plane of the substrate 300.
[0049] The end of the flexure 320 opposite the first anchor portion
350 terminates in a second anchor portion 360. FIG. 3B is a detail
perspective view of one flexure 320, showing the first anchor
portion 350, the second anchor portion 360, the first flexure layer
330, the second flexure layer 340, and the ground pad 310
underlying the flexure.
[0050] Referring again to FIG. 3A, the second anchor portion 360 is
mechanically and electrically connected to the underside of a
mirror segment 380. The mirror segment is any one individual mirror
of the DM device. Thus the mirror segment 380 is held at some
elevation above the substrate 300. In the first embodiment, this
elevation is on the order of 50 micrometers. The mirror segment is
electrically conductive and therefore is held at the same potential
as the ground pad 310. In the first embodiment, the mirror segment
380 is hexagonal in shape and is formed from a 20 micrometer-thick
layer of single crystal silicon and is coated on its top surface
with an optical coating, which may be a highly reflective metal
layer. The mirror segment diameter in the first embodiment is on
the order of 500 micrometers.
[0051] For the sake of clarity, FIG. 3A shows only three mirror
segments 380. However, an exemplary embodiment of the DM comprises
an array of 121 nominally identical elevated mirror segments 380
disposed over the substrate so as to form a larger, segmented
mirror surface, approximately circular in outline and having
inter-segment gaps of 5 micrometers.
[0052] The following is a general overview of the process of the
current invention for fabricating the first embodiment of the DM.
The process involves separately fabricating the MEMS structure and
the addressing and sensing circuits on two separate wafers, then
assembling them together as shown in FIGS. 4A and 4B. FIG. 4A is a
process flow diagram and FIG. 4B illustrates the corresponding
structure at each step. As shown at step 400, each mirror segment
380 is fabricated by reactive ion etching (RIE) the top single
crystal silicon "device" region of a bonded silicon-on-insulator
wafer (BSOI). At step 405, the wafer is then coated with a
sacrificial layer to fill the trenches left by the previous etch,
provide a temporary support for various mechanical structures of
the DM, and optionally to act as a dopant source for undoped
polysilicon regions. This sacrificial layer might typically be
phosphorus-doped silicate glass deposited by low pressure chemical
vapor deposition (LPCVD). Alternatively, in cases where the
sacrificial layer is not required to act as a dopant source,
silicon oxide deposited by a tetraethoxysilane (TEOS) process might
be used.
[0053] As shown at step 410, the PSG region is next patterned to
define the attachment points for the second anchor portions 360 of
the flexures; in some instances the patterning may include an
etching step. At step 415, a one micron undoped amorphous
polysilicon layer and a PSG layer are deposited by LPCVD and
annealed at 950.degree. C. for six hours to dope and tune the
residual stress of the polysilicon layer to approximately -40 MPa,
where the negative sign denotes compressive stress. The top PSG
layer is then removed at step 420 using a wet hydrofluoric (HF)
acid etch and the polysilicon layer is patterned and etched to
define the first flexure layer 330 at step 425. Silicon nitride
(SixNy) is then deposited by LPCVD at step 430, and patterned and
etched to define the second flexure layer 340 at step 435. At step
437, conductive metal pads are deposited, for example by
electroplating, on to the first anchor portion 350 of the flexures.
These metal pads will serve as the electrical and mechanical
attachment points between the flexures and the substrate 300.
[0054] FIG. 5 schematically illustrates a cross-section through the
MEMS structure 500 supporting a single mirror segment, completed up
to this point and including the mirror segment 380, flexure 320 and
the sacrificial layer 515, typically phosphorus-doped silicate
glass (PSG). As compared with the structure shown at the last step
437 of FIG. 4B-1, the structure shown in FIG. 5 has been inverted
in preparation for bonding to the electronics chip. In the first
embodiment, the flexure 320 is a two-layer structure with a first
flexure layer 330 of phosphorus-doped polysilicon, and a second
flexure layer 340 of SixNy. Although not required in all
embodiments, the MEMS device in the first embodiment includes a
temporary handle wafer (not shown in FIG. 5), typically 300 to 500
micrometers thick, used to support the MEMS structure prior to
release in a manner known in the art.
[0055] Continuing again with reference to process steps 445
onwards, shown in FIGS. 4A and 4B, drive circuitry in the form of
an integrated circuit is now introduced. This integrated circuit is
the substrate 300 on which the flexures and mirror segments will be
mounted. The substrate 300 is typically fabricated through separate
processing in a conventional manner, for example using silicon CMOS
techniques not shown here, and well known in the art. As shown at
step 445 in FIG. 4, the substrate 300 is typically coated with a
passivation layer to protect it from the MEMS release agent, which
may for example be hydrofluoric acid. As shown in step 447 of FIG.
4, the passivation layer is patterned and etched to expose bond
sites on the substrate 300 that are electrically connected to a
ground plane or to underlying circuits. An electrically conductive
bonding agent 610 is then deposited on these bond sites. FIG. 6 is
a schematic cross-section through the substrate at the end of step
447, showing the locations of the control electrodes 370, the
bonding agent 610 and the wiring layer 625.
[0056] Continuing to refer to FIG. 4, at step 450 the MEMS
structure 500, constructed as described above, is disposed over the
substrate 300 and the two are then bonded together. At this point
the MEMS structure 500 still includes the sacrificial layer
515.
[0057] At step 455 the handle wafer of the BSOI wafer is etched
away from the MEMS mirror segment, after which the sacrificial
layer is released from the MEMS structure as shown at step 460. The
IC passivation layer is removed at step 465, typically using an O2
plasma or appropriate solvent. Finally, an optical coating is
deposited on the top surface of the mirror segments, for example
using a shadow-masked metal evaporation, in step 470. The resulting
device is a completed, integrated DM. FIG. 7A shows a cross-section
through a single mirror segment and underlying structures, after
removal of the handle wafer at step 455. FIG. 7B shows a cross
section through the same structure at the end of the fabrication
process, after the MEMS sacrificial layer 515 and circuitry
passivation layer have been removed. The device includes the
following elements: IC portion or substrate 300 and MEMS structure
500; on the IC portion are shown a control electrode 370, the
bonding agent 610 and a wiring layer 625. On the MEMS portion 500
are shown the mirror segment 380 and flexure 320, comprising the
first flexure layer 330 and second flexure layer 340.
[0058] One important aspect of the present invention is the
above-described passivation layer. In the first embodiment of the
invention, an electrically-conductive contact must be established
through the passivation layer at the points where the MEMS
structure 500 is bonded to the substrate 300. The bonding process
can be any suitable process that results in a conductive bond, for
example gold to gold bonding. To allow the bond material to be
deposited onto the IC substrate 300, the passivation layer is
preferably patternable. In an exemplary arrangement, the
passivation layer is completely removable after the MEMS structure
is released in a manner that will not damage the MEMS structure.
This passivation material may be a protective polymer material such
as a polyimide or parylene.
[0059] Alternatively, the passivation material can be conductive so
that upon removal from the exposed surfaces, electrical contact
between the ICs and MEMS element is maintained. The passivation
material need not be patterned before bonding as it is selectively
removed, where not bonded to the MEMS structures, in the
passivation layer removal process. A conductive polymer or epoxy
can be used, for example, EPO-TEK OH108-1 or other similar
conductive epoxy made by Epoxy Technology, Inc., of Billerca,
Mass.
[0060] The present invention differs significantly from the prior
art in that it relies on the influence of IRS (as opposed to CTE)
in the flexures to elevate the mirror segments above the substrate
plane, to a much greater degree than has been found in the prior
art. The "Coefficient of thermal expansion" ("CTE") describes the
linear change in size of a material as a function of temperature,
while "Intrinsic residual stress" (IRS) describes the stress in a
material, which is dependent on the grain morphology and
crystalline defects of a material. This means that the elevation of
the segments above the substrate can be far less sensitive to
changes in temperature than for comparable prior art devices. The
deflection at the elevated end of each flexure is essentially
proportional to the curvature of the flexure, which may be written
as the sum of two components; a first component proportional to the
intrinsic residual stress in the flexure and a second component
proportional to the CTE mismatches in the flexure. In the first
embodiment of the invention, the first flexure layer is composed of
polysilicon and the second flexure layer is composed of silicon
nitride. This provides a flexure for which the IRS component is
larger than the CTE component by a factor of approximately one
thousand at normal operating temperatures, for example in the range
0-100 degrees Celcius.
[0061] Many alternative embodiments of the flexure are possible in
which the second flexure material is one with a CTE similar to that
of the first flexure material. If that first material is
polysilicon, the second material can be a ceramic, such as SiC, or
silicon nitride (SixNy), or even polysilicon itself, deposited
under different conditions so as to induce a different grain
structure and crystal defect concentration, and thus different IRS.
FIG. 8 is a table that lists the CTE of some example materials.
[0062] In contrast to the prior art usage of nickel, SixNy is
advantageous because it does not contaminate etchers as Ni does.
SixNy is also easier to process because it is a standard IC
material deposited by LPCVD. The residual stress of SixNy can be
controlled by varying the ratios of the reactant gasses, deposition
pressure, and the deposition temperature. For example, a layer
deposited with a gas flow ratio of 1:3 dichlorosilane to ammonia at
125 mTorr and 800 will yield a stoichiometric film (Si3N4) with
approximately 1 GPa of residual tensile stress, while 4:1 gas ratio
at 140 mTorr and 835.degree. C. will yield a film composition near
Si3N3 with approximately 280 MPa of residual tensile stress. To
achieve the desired radius of curvature of the flexure, different
SixNy stoichiometries can be used, the appropriate choice for which
may be application-specific.
[0063] The first embodiment of the DM comprises a tiled array of
mirror segments, supported on flexures and elevated approximately
50 micrometers above the substrate. As described, the substrate
contains electronic circuits used for controlling and sensing the
tip, tilt and piston motion of the segments. The circuits are
controlled via electrical signals transmitted, for example, through
bond pads on the substrate and generated, for example, by a
microprocessor in a manner well known in the art. The control
signals typically contain information, generated by a wavefront
reconstructor, about the combination of tip, tilt and piston
motions for each mirror segment needed to compensate for the
wavefront aberrations at a given time. "Piston movement" is one of
three types of movement used to describe actuation of a mirror
segment, and describes translation normal to the plane of the DM
aperture. "Tilt", the second type of movement, is movement about
any first axis that is parallel to the plane of the DM aperture.
"Tip", the third type of movement, is movement about any second
axis (not parallel to the first axis) that is also parallel to the
substrate.
[0064] The circuits embedded in the substrate 300 decode this
information and translate it into a corresponding set of voltages
that are applied to the control electrodes disposed under each
mirror segment. The electrical potential difference and resulting
electrostatic force between each mirror segment and its three
control electrodes causes it to move in tip, tilt and piston, and
assume a position and orientation determined by the voltages
applied to the three electrodes. This ability to independently
orient and position each segment allows spatially complex wavefront
aberrations to be corrected by the DM. In some implementations of
the first embodiment, the substrate also contains sense electronics
that detect the tip, tilt and piston of each segment, for example
by measuring the capacitance between the segment and its three
control electrodes. Incorporation of sense electronics can improve
the resolution with which the segments can be controlled. Because
the attractive force between a segment and its control electrodes
increases rapidly as the gap between them diminishes, the control
voltages must be limited to avoid pulling segments into contact
with the electrodes. Typically, the maximum operation voltage is
chosen to be the voltage that causes a segment to travel 25% of the
elevation produced by the flexures. Therefore, the flexure
elevation of 50 micrometers described in the first embodiment
results in a useable mirror stroke of approximately 12
micrometers.
[0065] In a second embodiment of the invention, the structure of
the DM is identical to the structure of the first embodiment,
except that the ground pads and control electrodes are formed on
the MEMS part rather than the CMOS part. The appearance of the
completed device is essentially identical to that of the first
embodiment, illustrated in FIG. 3A.
[0066] Fabrication of the second embodiment proceeds in a manner
identical to that used for the first embodiment up to step 435 of
FIG. 4B. A sacrificial layer is then deposited, patterned and
etched to open up anchor points where the ground pads 310 will
attach to the first anchor regions 350 of the flexures. A layer of
polysilicon is then deposited, patterned and etched to define the
ground pads 310 and the control electrodes 370. A layer of metal is
then deposited, patterned and etched so that it coats the surfaces
of the ground pads 310 and control electrodes 370, but does not
bridge unconnected structures.
[0067] The CMOS portion 300 of the device is fabricated in the same
way as for the first embodiment, but has bond sites in locations
that correspond to both the ground pads 310 and the control
electrodes 370 of the MEMS structure. The ground pad bond sites are
electrically connected to a ground plane or to circuits in the
substrate 300, while the control electrode bond sites are connected
to the appropriate control and sense circuits within the substrate
300. The MEMS portion and the CMOS portion are bonded together
using a film of anisotropic conductive polymer that conducts only
in a direction normal to the plane of the film. In this embodiment,
the anisotropic conductive polymer acts as both a bonding agent and
a CMOS passivation layer. After bonding, the MEMS structures are
mechanically released, for example by HF etching, as in the first
embodiment. Because of the anisotropic nature of the polymer, it
does not need to be removed from the DM and so the passivation
layer removal step is omitted for this embodiment. As for the first
embodiment, the final step is the deposition of an optical coating
on the top surface of the mirror segments.
[0068] The method of operation for the second embodiment is
identical to that for the first embodiment.
[0069] FIG. 9 shows the mechanical structure of a DM according to
the third embodiment of the invention, in a partially exploded
perspective view. For the sake of clarity, FIG. 9 shows only a
single piston-tip-tilt mirror segment. However, it will be clear to
one skilled in the art that multiple such mirrors may be fabricated
side-by-side on a single substrate to form a segmented DM, as was
described for the first embodiment.
[0070] The third embodiment of the DM comprises a substrate 900,
which may be a silicon wafer. On top of the substrate 900 are
formed a number of control electrodes 960 that are electrically
isolated from one another and electrically connected to conductive
traces (not shown in FIG. 9) that may either be embedded in the
substrate 900 or attached to the surface of the substrate 900.
These traces electrically connect the control electrodes 960
directly to bond pads (not shown in FIG. 9) that may be disposed
around the perimeter of the DM chip. The control electrodes 960 are
arranged in groups of three and are rhombic in shape, so that the
footprint of each group is essentially hexagonal.
[0071] Disposed around each group of three control electrodes 960,
are three conductive ground pads 910, fabricated from the same
material as the control electrodes 960. The ground pads 910 are
electrically isolated from the control electrodes 960 and
electrically connected to a ground plane embedded in the substrate
900. Attached to one end of each ground pad 910 is a first anchor
portion 950 of a flexure 920. The flexure, in the third embodiment
comprises two layers, a first flexure layer 930 formed from
conductive polycrystalline silicon and a second flexure layer 940
formed from silicon nitride. The first anchor portion 950 is both
mechanically and electrically connected to the ground pad 910 so
that the conductive first flexure layer 930 is held at the same
potential as the ground pad 910. The second flexure layer 940 is
rigidly attached to the top side of the first flexure layer 930 and
extends over a portion of the length of the flexure 920. The
purpose of the second flexure layer is to provide a residual stress
difference between the top and bottom portions of the flexure 920,
causing the flexure 920 to bend up out of the plane of the
substrate 300.
[0072] The end of the flexure 920 opposite the first anchor portion
950 is electrically and mechanically connected to a hexagonal
platform 980. A platform bond site 990, fabricated from a metal, is
electrically and mechanically connected to the platform. This
platform bond site matches up with a corresponding segment bond
site, also fabricated from a metal, on the underside of a mirror
segment 970. The segment bond site is not visible in FIG. 9, since
it is on the underside of the mirror segment 970. In the fully
assembled DM, the mirror segment 970 is mechanically and
electrically connected to the platform 980 via these bond sites.
Thus the mirror segment 970 is held at some elevation above the
substrate 900. In the first embodiment, this elevation is on the
order of 50 micrometers. The mirror segment is electrically
conductive and therefore is held at the same potential as the
ground pad 910. In the third embodiment, the mirror segment 970 is
hexagonal in shape and is formed from a 20 micrometer-thick layer
of single crystal silicon and is coated on its top surface with an
optical coating, which may be a highly reflective metal layer. The
mirror segment diameter in the third embodiment is on the order of
500 micrometers.
[0073] In the third embodiment, the DM does not incorporate drive
and sense electronics, but does incorporate the improved bimorph
flexure. The actuator substrate 900 is fabricated in a method
similar to that used to fabricate the MEMS portion of the first
embodiment, but where the starting material is a standard silicon
wafer rather than a bonded SOI wafer. The ground pads 910, control
electrodes 960, electrical traces and bond pads are defined in a
first undoped polysilicon layer, deposited on an insulating silicon
nitride layer. Alternatively, the traces could be fabricated in a
buried layer beneath the electrodes that is electrically isolated
in all regions except areas that contact the electrical traces to
electrodes and bond pads. A phosphorous-doped silicate glass (PSG)
sacrificial layer is then deposited, patterned and etched to open
up regions where the first anchor portion 950 of the flexures will
connect to the ground pads 910. A second undoped amorphous
polysilicon layer is then deposited followed by a PSG layer. The
wafer is annealed at 950.degree. C. for six hours to dope and tune
the residual stress of the second polysilicon layer to
approximately -40 MPa. In this step, the sacrificial PSG layer also
dopes the first layer of polysilicon. The top PSG layer is then
removed using a wet HF acid etch and the second polysilicon layer
is patterned and etched to define the first flexure layers 930 and
platforms 980. A layer of silicon nitride is then deposited,
patterned and etched to define the second flexure layers 940, after
which a low-temperature oxide (LTO) is deposited by LPCVD to
protect the structures from a later etch. The LTO layer is etched
and a metal layer is selectively deposited, for example by
electroplating, to form the bond sites 990 and bond pads disposed
around the perimeter of the DM chip.
[0074] The mirror segments 970 are formed on a separate wafer,
typically a BSOI wafer with a 20 micrometer thick device layer. The
mirror segments are defined using deep reactive ion etching,
followed by deposition of a sacrificial layer (typically PSG) that
refills the trenches between the segments. The sacrificial layer is
then patterned and etched to clear access holes for bond sites that
match those deposited on the actuator substrate 900. A metal layer
is then selectively deposited, for example by evaporation and
lift-off, to form the segment bond sites that will be joined to the
corresponding platform bond sites 990.
[0075] The actuators and mirror segments are then assembled and
bonded together, for example using gold to gold bonding. The
mirror-segment handle wafer is then removed in a manner known to
those skilled in the art, and the sacrificial layers are removed,
for example by HF etching, to allow the flexures to lift the mirror
segments 970 above the substrate 900. Finally, an optical coating
is deposited on the top surface of the mirror segments.
[0076] The third embodiment is operated in a manner similar to the
first embodiment, with the exception that the control voltages used
to set the orientation and piston of the mirror segments are
generated by driver electronics on a chip or board that is
physically separate from the DM chip. The control electrodes for
each mirror segment are connected to the outputs of the drive
electronics for example via bond wires electrically connected to
the bond pads disposed around the edges of the DM chip.
[0077] Accordingly, the invention provides improved methods and
structures for elevating a number of platforms above a substrate
and for controlling the piston, tip and tilt motions of those
platforms. The resulting structures feature low temperature
dependence, small size and power consumption and high control
precision. The methods and structures may be used to construct an
improved deformable mirror (DM) that features low temperature
dependence, high fill-factor, high control resolution and large
stroke, and which can be fabricated in a small form-factor at low
cost. The ability to integrate drive and sense electronics on the
same chip as the mirror segments allows DMs with large numbers of
actuators to be realized. The structures and methods for producing
temperature-insensitive elevated mirror segments and the structures
and methods for assembling the mirror segments on to control and
sense electronics can be applied separately or in combination.
[0078] Although the description above contains many specificities,
these should not be construed as limiting the scope of the
invention but as merely providing illustrations of some of the
possible embodiments of this invention. For example, the mirror
segments can have other shapes, such as square, rectangular,
triangular etc.; the mirror segments can be supported by different
numbers of flexures; the flexures can be constructed from any
number of materials and comprise any number of layers, provided
their curvature is predominantly caused by IRS, rather than CTE
differences; the tip, tilt and piston of the mirror segments can be
controlled by varying the duty cycle of an AC signal applied to the
control electrodes rather than the magnitude of an applied DC
signal; the thicknesses of the layers that comprise the DM can be
varied; the diameters or widths of features such as the mirror
segments, flexures and control electrodes can be varied; the number
and placement of the control electrodes under each segment can be
changed; the elevation of the mirror segments above the substrate
can be altered; the actuators need not be electrostatic but could
be, for example, piezoelectric or magnetic; the gaps between mirror
segments can be changed; different reflective coatings including
both metallic and dielectric coatings can be deposited on the top
surface of the segments; different materials and methods can be
used to bond the MEMS portion to the CMOS portion; different
passivation materials can be used to protect the CMOS circuits
during MEMS release; the number of mirror segments comprising the
DM can be varied, etc.
[0079] While numerous specific details have been set forth in order
to provide a thorough understanding of the present invention,
numerous aspects of the present invention may be practiced with
only some of these details. In addition, certain process operations
and related details which are known in the art have not been
described in detail in order not to unnecessarily obscure the
present invention.
[0080] Thus the scope of the invention should be determined by the
appended claims and their legal equivalents, rather than by the
examples given.
[0081] The foregoing detailed description of the invention has been
presented for purposes of illustration and description. It is not
intended to be exhaustive or to limit the invention to the precise
form disclosed. Many modifications and variations are possible in
light of the above teaching. The described embodiments were chosen
in order to best explain the principles of the invention and its
practical application to thereby enable others skilled in the art
to best utilize the invention in various embodiments and with
various modifications as are suited to the particular use
contemplated. It is intended that the scope of the invention be
defined by the claims appended hereto.
* * * * *