Optomechanical Uncooled Infrared Imaging System

1
Project Review 09/11/01
Optomechanical Uncooled Infrared Imaging
System
DARPA Program Manager Ray Balcerak SPAWAR
Program Monitors Cindy Hanson, Randy
Shimabukuro
University of California, Berkeley Arun
Majumdar Roberto Horowitz IR Vision Paul
Norton CFD Research Corporation Andrej
Przekwas Mahesh Athavale NIST-Boulder Joh
n Kitching Raytheon John Varesi
2
Program
9-910 Introductions 910-920 General Outline
(Arun Majumdar) 950-1020 Thermal Design
Control (J. Yamaguchi J.
Choi) 920-950 Microfabrication (Yang
Zhao) 1020-1050 Optical Readout (Yang
Zhao) 1050-1110 Simulations (Arun
Majumdar) 1110-1130 Wrap up and feedback
3
System Diagram
Low Power
Visible Laser
Partial
Beam
Reflective
Piezo
Spliter
Mirror
Imaging
CCD
Lens
Infrared
Lens
Infrared (IR)
Image
Source
FPA Side
Plane
View
Focal Plane
Array(FPA)
Visible
of
Bimaterial Cantilever
Reflector
Temperature Sensors
IR Radiation
IR Absorber
IR Antireflection
Image Process System
Coating
4
Previous Work
Single Level Pixel
Fourier optics based optical readout
Noise-Equivalent Temperature Difference NETD ?
1-2 K
5
Current Work

• 2-Level Pixel
• - Bimaterial Length, Lm 4 x L (Pixel Size)
• - SiNx IR absorption pad separated lIR/4 away
  from Au reflector
• - Single thermal isolation
• - Fill factor ? 50 (thermal isolation and IR
  absorption pad in level 1)
• Optical readout based on interferometry. Pixel
  size scalable down to 25 mm.

6
Berkeley Team Members
Yang Zhao Microfabrication Optics Joji
Yamaguchi J. Choi Thermal Control S. Morales
7
Design and control of a thermal stabilizing
system for an Optomehanical Uncooled Infrared
Imaging camera J. Choi, J.Yamaguchi, S.
Morales, R. Horowitz, Y. Zhao and A.
Majundar Department of Mechanical
Engineering University of California at Berkeley
8
Objective and Design

• Objective
• Maintain FPA temperature in steady state of 100
  ?K(std.) to achieve the NETD of 1.7mK(std.)
• (Noise Equivalent Temperature Difference)
• Design approach
• Design of a thermal shield consists of Copper
  rings and a Sapphire plate
• Design of a temp. controller to stabilize the
  thermal shield and the FPA temperature

9
Temperature Control Setup
Tch Cantilever temp.
Laser light
Tcu Copper temp.
Tsi Si substrate temp.
Ta Ambient temp.
Cantilevers
Si substrate
Copper ring
Thermistors
q
Thermoelectric heater/cooler
I
Vacuum chamber
Temperature Controller
IR radiation
Te Exhaust temp.
10
Objective and Design

• Objective maintain FPA temperature in steady
  state of 100 ?K(std.) to achieve the NETD of
  1.7mK(std.)
• Design approach
• Design of a thermal shield consists of Copper
  rings and a Sapphire plate
• Design of a temp. controller to stabilize the
  thermal shield and the FPA temperature
• Model based Linear Quadratic Gaussian temp.
  control using a lumped parameter model for the
  thermal system

11
Lumped Parameter Thermal Model
Ta
Tcu Copper ring temp. Tsi Si substrate
temp. Tch Chip temp. Ta Ambient
temp. Rcxx Thermal resistance of
conduction Rrxx Thermal resistance of
radiation Cxx Thermal capacitance
Rrpl
Plate
Rcpl
Cpl
Copper ring
Rrch1
Rccu
Cantilever (chip)
Rleg
Tch
Rrcu
Si substrate
Tcu
Tsi
Ta
Csi
Rcsi
Ccu
Rccu
Rrch2
Thermoelectric heater/cooler
q
TT
Vacuum chamber
12
Objective and Design

• Objective maintain FPA temperature in steady
  state of 100 ?K(std.) to achieve the NETD of
  1.7mK(std.)
• Design approach
• Design of a thermal shield consists of Copper
  rings and a Sapphire plate
• Design of a temp. controller to stabilize the
  thermal shield and the FPA temperature.
• Model based Linear Quadratic Gaussian temp.
  control using a lumped parameter model for the
  thermal system
• Copper temperature TCu measuring using a
  thermistor with an expected resolution below 1
  milli Kelvin
• Actuation of heat q via thermoelectric
  heater/cooler

13
Temperature Control Setup
Tch Cantilever temp.
Laser light
Tcu Copper temp.
Tsi Si substrate temp.
Ta Ambient temp.
Cantilevers
Si substrate
Copper ring
Thermistors
q
Thermoelectric heater/cooler
I
Vacuum chamber
Temperature Controller
IR radiation
Te Exhaust temp.
14
Design

• Design strategy
• The thermal shield system provides a large
  attenuation for the open loop from Ta to Tch
  (-50-100dB) for .002ltflt100Hz
• LQGR temp. Controller improves the attenuation by
  more than 50dB for low frequency ( flt.002Hz
  )range

15
Design

• Open and closed loop system Bode plot

(1) Ta to Tch
(2) Tt to Tch
16
Design

• Design strategy
• The thermal shield system provides a large
  attenuation for the open loop from Ta to Tch
  (-50-100dB) for .002ltflt100Hz
• LQGR temp. Controller improves the attenuation by
  more than 50dB for low frequency ( flt.002Hz
  )range
• No noticeable changes appear in the transfer
  function from Tt to Tch for with or without
  controller

17
Design

• Open and closed loop system Bode plot

(1) Ta to Tch
(2) Tt to Tch
18
Design

• Design strategy
• The thermal shield system provides a large
  attenuation for the open loop from Ta to Tch
  (-50-100dB) for .002ltflt100Hz
• LQGR temp. Controller improves the attenuation by
  more than 50dB for low frequency ( flt.002Hz
  )range.
• No noticeable changes appear in the transfer
  function from Tt to Tch for with or without
  controller
• A Reference model for the desired trajectory is
  included in the controller modeling to insure no
  overshoots during the transient response
• Integral action for the rejection of constant
  disturbances

19
Design

• Controller Design
• Set Point Cr
• Ur Xr
• Ta Tcu Tcu hat
• q Xa
• qx
• qe
• error
• qr

20
Experiment Result (1)

• Simulation vs. Experimental results

(Tcu, Tcu, Tsi, Tch)
(Tcu, Tcu, Tsi, Tch)
(control q)
(control q)
21
Experiment Result (2)
(Tcu, Tcu, Tsi, Tch)
Steady State Experiment Result
- Std. noise in measurement1mK
- Std. Tcu measured 1.1m K
- Std. Tcu estimated 61.5 µK
Estimation at FPA

• Std. Tch estimated117.3µK (experi)
• -Std. Tch estimated120.0µK (simula)

22
Conclusion and Future work

• Thermal Control system successfully achieved
  steady
• state FPA temperature of 100 ?K(std.).
• The regulated FPA temp. guarantees the NETD of
• 1.7mK(std.) (by considering only thermal
  noise)
• Future work is to improve the thermal model and
• measurement system
• - Measurements from FPA dynamics
• - Including dynamic of Tch
• - Integrating sensors with the FPA

23
Design, Fabrication, and Preliminary Results

• Yang Zhao
• Mechanical Engineering Department
• UC Berkeley

24
Outline

• Pixel design
• Microfabrication
• FPA layout
• Preliminary tests
• Low temperature Process

25
Current Design

• Material selection
• SiNx/Al
• IR optics ( SiNx )
• Resonant cavity (ml / 2)
• Thermal design
• IR absorption area
• Minimum thermal conductance
• Thermomechanical design
• Long bimaterial beam
• Pixel tiling - spatial resolution

26
Microfabrication Process
Si substrate

• High temperature process
• Metal deposition simultaneously on both layer
  after release

27
FPA Layout
Bare SiNx cantilevers
After metal (Al) deposition
28
Temperature Response
110 mm pixel
Thermomechanical Response 2 mm/K
29
Single Pixel Calibration
Image plane of L3
L4
Mirror
CCD
Pinhole3
L3
Image plane of L2
Pinhole2
L2
Laser (670 nm)
BS
FPA and Mounting Plate
L1
Pixel Array
Mirror
PZT
Pinhole1

• did with John Kitching at NIST

30
Beam Steering

• Block the Reference Mirror Arm
• Use smaller pinhole2

DT 0.4 oC
31
Interference Measurement

• Calibration temperature effect on mounting system
• 0.2 0.25 mm/K
• Feedback PZT control
• Keep the intensity constant
• 0.2 0.25 mm/K ( 65 mm pixel)
• ? 0.4 0.5 mm/K
• (0.5 mm/K theoretical analysis)
• Considering the temperature difference between
  the cantilever and mounting Cu plate, the actual
  temperature sensitivity should be higher than this

Photo Diode
Cu Plate
B.S.
FPA
Mirror
PZT
32
Optical Readout Setup
Mirror
Cooking Stove Coil
Beam Splitter
Laser
Partial Reflective Plate
FPA
Vacuum Chamber
f1 IR lens
I I0 cos2( 2pd /l )
33
Blurred Image?

• IR absorption on the thermal isolation leg

• Solution thin metal layer underneath the leg
  part to block the IR

34
Alternative Process

• Low temperature Process
• Sacrificial layer Photo Resist
• Room temperature PECVD SiNx
• Advantage can be dry released
• Disadvantage stress

35
Optical Readout System Characterization

• John Kitching
• NIST

36
High-Quality CCD Installed
Replaced 8-bit CCD with high-quality, 12-bit CCD
camera
CCD Properties Full-well capacity 500,000
e- 12-bit digitization Low dark current ?
Camera S/N 700
High quality CCD
Magnifier
Photodiode
ND Filter
M
F50cm
BS
PZT
Laser
Translation
50 cm
FPA
Can block interferometer arm here
12-bit image with new CCD
37
Imaging Individual Pixels
Single interferometer arm
Both interferometer arms

• Some pixels are bright, some dark
• Pixel intensity changes with mirror pos.
• Fringe contrast 80

• Reflection from pixels clearly visible
• Magnification

I0
0.8I0
Intensity
Mirror Position
38
Signal-to-Noise

• Zoom in on single pixel
• Take 10 frames, each binned into 9 sections

• For each frame add the
• values of the 9 sections
• Calculate mean and
• std. dev. of 9-section sum

Mean ltxgt Std. Dev. ltDx2gt1/2
10 frames

• Measure these quantities as a function of
  exposure time, and therefore ltxgt
• If ltxigt0,4096 and full-well capacityFWC,
  then shot noise?

39
Signal-to-Noise Measurements I
Noise measurements for single arm of
interferometer
Shot noise

• Noise is close to theoretical value based on
  shot noise
• S/N 1000 obtained for ltxgt12,000.
• Indicates that other optical readout noise
  sources (laser noise, etc.)
• will be smaller than shot noise when a single
  CCD pixel is used (ltxgt2000)

40
Signal-to-Noise Measurements II
Noise measurements for both arms of interferometer

• When interferometer is on, noise increases
  dramatically
• Large increase in interferometer noise is due to
  long-term drift of interferometer
• (see next viewgraph)
• Note this will be much less of a problem
  with a Fabry-Perot interferometer
• Best S/N with interferometer on 100

41
FPA Pixel Drift

• Measuring CCD image from each pixel over 100
  seconds

• Slow drift of pixels is causing degraded S/N
• Drift could be due to pixel temperature change
  or interferometer
• path length change

42
Interferometer Locking Scheme

• To improve the interferometer S/N
• Include auxiliary beam ( ), which
  hits edge of FPA (not on pixels)
• Lock mirror arm to FPA arm using auxiliary beam
  
• pixel motion will still be detected since aux.
  beam hits side of FPA
• Interferometer path length stabilized (w.r.t.
  air currents, vibrations, etc.)

High quality CCD
Photodiode 2
Magnifier
Photodiode
ND Filter
M
F50cm
BS
PZT
Laser
Translation
50 cm
FPA
43
FPA Pixel Drift

• Interferometer locked up
• Measuring CCD image from each pixel over 100
  seconds

• Much improved S/N but some drift still present
• Drift could be due to pixel temperature change
• DT required to cause drift 25 mK over 100 s

If we remove slow drift S/N 500 ?

• Non shot-noise sources will allow us to get to
  S/N500

44
Signal-to-Noise III

• Improved interferometer shows much better S/N
  500 for ltxgt10,000
• Shot noise limit is about 1000 at ltxgt10,000
  within a factor of 2

Shot noise
45
Conclusions

• Signal-to-noise of 500 has been obtained for
  interferometric readout of pixel motion
• Since we have expanded a single FPA pixel onto
  many CCD pixels, we are not measuring the CCD
  shot noise but measure the sum of all the other
  noise sources
• CCD shot noise should result in
• This is very close to our goal for DTopt
• Main things remaining to be done
• 1) Demonstrate high S/N measurement with no
  image magnification
• 2) Test stabilized interferometric readout on
  highly temperature stabilized FPA to ensure that
  drifts in current system are in fact due to FPA
  temperature changes.
• 3) Integrate into UCB system for high S/N
  measurement

46
CFD Research Corporation
215 Wynn Dr. , Huntsville, AL 35805 (256)
726-4800 FAX [phone] www.cfdrc.com
FURTHER ANALYSIS OF BASELINE PIXEL
DESIGN Mahesh M. Athavale CFDRC, September 9,
2001
47
WORK ACCOMPLISHED

• Evaluation of the Thermal Characteristics of
  Baseline Pixel Design Continued.
• - analysis of the baseline pixel (65mmX65mm pad)
  was
• continued
• - two groups of IR absorption areas in the
  pixel pad and
• underside of cantilever on the pad level
  (65-35 area
• ratio)
• - Thermal (steady-state) simulations of the
  pixel with
• irradiation on either of the absorption areas
  considered

48
COMPUTATIONAL GRID AND BOUNDARY CONDITIONS
65 m Pad Pixel, 13500 cells
49
RESULTS (FULL PIXEL IR FLUX)
65 m Pixel, 1 mm SiN and 0.6 mm Al as
bimaterial Two Types of Surface Conditions Used
on Pixel Surfaces that do not have Incident IR
Flux - adiabatic (higher T limit) and radiative
transfer (low T limit) Specified IR Flux 1
W/m2 on Both the Pad and Cantilever
z
BC Type Dtmax mK Adiabatic 31 Radiative
19
y
x
Maximum Temperature Rise in the Pixel
T 300K
(Scale xyz 115)
50
TEMPERATURES (IR FLUX ON PAD)
Incident IR flux of 1 W/m2 on the Pad Area (65
of total)
With Adiabatic Walls DTmax 20.6 mK
With Radiating Walls DTmax 12.6 mK
51
TEMPERATURES (IR FLUX ON CANTILEVER)
Incident IR flux of 1 W/m2 on the Underside of
the Lower Level Cantilever (35 of total)
With Adiabatic Walls DTmax 10.7 mK
With Radiating Walls DTmax 6.6 mK
52
THERMOMECHANICAL RESPONSE
Predicted Maximum Deformation is 0.0118 mm for
an Incident IR Flux of 1 W/m2 and Adiabatic
Walls Predicted Maximum Deformation is 0.0073
mm for an Incident IR Flux of 1 W/m2 and
Radiating Walls
Adiabatic Walls
Radiating Walls
53
DEFORMATION (IR FLUX ON PAD)
Incident IR flux of 1 W/m2 on the Pad Area (65
of total)
With Adiabatic Walls
With Radiating Walls
54
DEFORMATION (IR FLUX ON CANTILEVER)
Incident IR flux of 1 W/m2 on the Underside of
the Lower Level Cantilever (35 of total)
With Adiabatic Walls
With Radiating Walls
55
PARTIAL IRRADIATION RESULTS
The Temperature Patterns in Both Partial
Irradiation Cases are Similar Maximum
Temperatures in the Pixel Scale According to
the Relative Absorption Areas of the Cantilever
and Absorber Pad (about 35-65) The
Temperature Increases in the Two partial Cases
Can be Added to get the Full Flux Case
(Superposition)- So Can the Deformations.
56
PLANNED EFFORTS
Continue Support of the Design and Development
Activity at UCB Via Simulations Thermal and
Optical