1. Zero Read Noise Detectors for the TMT Don Figer, Brian Ashe , John Frye, Brandon Hanold, Tom Montagliano, Don Stauffer (RIDL), Brian Aull, Bob Reich, Dan Schuette, Jim Gregory, Erik Duerr, Joseph Donnelly (MIT/LL)
This talk describes the Moore Foundation-funded project to develop photon counting imaging detectors for astronomical (and other) applications. The goal of the project is to develop zero read noise detectors for the Thirty Meter Telescope and other applications. It is a collaboration between personnel at RIT and Lincoln Lab. The duration of the project is four years, and the first year is over. At this stage, a 256x256 readout circuit has been sent out for fabrication.
MIT LL No. MS-43282, ESC No

2. Outline Motivation Moore Detector for TMT Heritage: LIDAR Conclusions
Why pursue photon-counting technology?
Why use Geiger-mode avalanche photodiodes (APDs)?
Moore Detector for TMT
Heritage: LIDAR
Conclusions
This slide gives the outline of the whole talk.

3. Outline Motivation Moore Detector for TMT Heritage: LIDAR Conclusions
Why pursue photon-counting technology?
Why use Geiger-mode avalanche photodiodes (APDs)?
Moore Detector for TMT
Heritage: LIDAR
Conclusions
The following section will motivate our technology development efforts.

4. Why pursue photon-counting technology?
Photon-counting detectors effectively have zero read noise.
In low light applications, read noise can dominate signal-to-noise ratio.
Many applications can become low light applications with higher resolutions.
spectroscopy
time-resolved photometry
fast wavefront sensing and guiding
This slide provides the motivation for photon-counting detectors in low light applications.

5. Detectivity (higher is better)
Sensitivity and detectivity are metrics for evaluating the performance of detectors. In low source and background flux cases, detectivity goes as one over the read noise.

6. Exposure Time to SNR=1
Another way to evaluate performance is by estimating the time it would take for a system to deliver a particular signal-to-noise ratio. Again, for low flux cases, read noise is critically important.

7. Example for Planet Imaging
The exposure time required to achieve SNR=1 is dramatically reduced for a zero noise detector compared to detectors with state of the art read noise.
The table gives exposures times to produce SNR=1 for a typical planet detection space mission. Note the dramatic reduction in time required for a zero noise detector. R=100 is assumed, and the effective planet flux is about 10 photons/hour per pixel. The nearby starlight is suppressed at the 10^-10 level by optical nulling. Zodiacal light contributes about electrons/s/pixel.

8. Why use Geiger-Mode Avalanche Photodiodes (GM-APDs)?
produce easily distinguishable high voltage pulse per photon
have zero “excess noise factor”
allow for hybridization and bonding to non-optical detecting materials
allow photon counting inside each pixel for high frame rates and time tagging
have demonstrated excellent performance for LIDAR applications
This slide gives the reasons why we use Geiger-Mode Avalanche Photodiodes (GM-APDs) for photon counting.

9. Gain of an APD M Ordinary photodiode Linear-mode APD Geiger-mode APD
100 10 1
An avalanche photodiode will generate a cluster of free charges as the voltage across it exceeds the breakdown value. In the linear regime, the charge flow will eventually stop without any external quenching circuitry. In the Geiger mode regime, the avalanche will continue as long as current is supplied.
Breakdown
Response to a photon ∞
I(t) M 1

10. Geiger-Mode Imager: Photon-to-Digital Conversion
Pixel circuit
Digitally encoded photon flight time
Digital
timing
circuit
photon
Lenslet
array
APD/CMOS array
Focal-plane
APD
We are developing Geiger-mode APD arrays integrated with CMOS timing electronics in each pixel. The left-hand diagram depicts the physical implementation of the receiver array. The APD-array wafer is bonded to the CMOS timing-circuit wafer in order to provide the necessary electrical connections in a compact and sturdy integrated package. A lenslet array will also be incorporated to enhance the effective detector fill factor. The right-hand figure is a block diagram of the pixel circuit. A CMOS interface circuit converts the analog pulse from the APD into a CMOS-compatible digital transition that gates the counter in the digital timing circuit.
Quantum-limited sensitivity
Noiseless readout
Photon counting or timing

11. Outline Motivation Moore Detector for TMT Heritage: LIDAR Conclusions
Why pursue photon-counting technology?
Why use Geiger-mode avalanche photodiodes (APDs)?
Moore Detector for TMT
Heritage: LIDAR
Conclusions
This section describes the Moore Foundation detector project.

12. Moore Detector Project Goals
Operational
Photon-counting
Wide dynamic range: flux limit to 108 photons/pixel/s
Streaming readout
adaptive optics imaging
multiple target tracking
Time delay and integrate
Technical
Backside illumination for high fill factor
Demonstrate 25 mm pitch imager with streaming, single photon, readout
The Moore detector will have extraordinary capabilities in a competitive device and pixel size format. It will be capable of delivering high dynamic range and ultra-fast imaging.

13. Moore Photon Counting Imager
Optical (Silicon) Detector Performance
Parameter
Phase 1
Goal
Phase 2
Format
256x256
1024x1024
Pixel Size
25 µm
20 µm
Read Noise
zero
Dark Current K)
<10-3 e-/s/pixel
QEa Silicon (350nm,650nm,1000nm)
30%,50%,25%
55%,70%,35%
Operating Temperature
90 K – 293 K
Fill Factor
100%
aProduct of internal QE and probability of initiating an event. Assumes antireflection coating match for wavelength region.
The proposed performance for the optical (silicon) Moore detector is shown in the table. In Phase 1, we will fabricate and test a moderate-size detector, and in Phase2, we will do the same for a large detector.

14. Moore Photon Counting Imager
Infrared (InGaAs) Detector Performance
Parameter
Phase 1
Goal
Phase 2
Format
Single pixel
1024x1024
Pixel Size
25 µm
20 µm
Read Noise
zero
Dark Current K)
TBD
<10-3 e-/s/pixel
QEa (1500nm)
50%
60%
Operating Temperature
90 K – 293 K
Fill Factor NA
100% w/o mlens
aProduct of internal QE and probability of initiating an event. Assumes antireflection coating match for wavelength region.
The proposed performance for the infrared (InGaAs) Moore detector is shown in the table. In Phase 1, we will fabricate and test a single test structure, and in Phase2, we will do the same for a large detector. Note that we will likely only pursue an optical OR infrared detector in Phase 2, depending on the maturity of both paths.

15. Moore Detector Project Status
A 256x256x25mm readout integrated circuit is being fabricated.
InGaAs test diodes are being fabricated.
Silicon GM-APD arrays have been fabricated and will be bump-bonded to the new readout circuit.
Photon-counting electronics are being built.
Testing will begin later in 2009.
Depending on results, megapixel silicon or InGaAs arrays will be developed.
The slide describes the status of the Moore project.

16. Overview of Pixel Operation
Pixel Architecture
The slide describes the basic operation of a pixel.
Active quench circuit converts the APD firing into a 1.8V logic transition
Global “record” clock latches APD state, and then a global rearm is asserted
Latched states of the APDs are read out directly by addressing a column of pixels and putting the states onto the core data bus
7-bit pseudorandom counter increments if the APD has fired
Overflow latch is set just before the 7-bit counter reaches turnover
The overflow bit is read out by addressing a row of pixels to put this data onto the peripheral data bus

17. ROIC Pixel Layout (2x2 pixels)
metal bump bond pad
core
(active quench, discriminator, APD latch)
2 pixels, 50 mm
counters (4 pixels)
counter rollover latch
The slide describes the ROIC pixel architecture. It includes 2x2 pixels. Key features are highlighted in color boxes with labels from the schematic picture.
2 pixels, 50 mm

18. InGaAs Development 3 APD designs grown and fabricated
2-mm-wide avalanche region (all InP)
3-mm-wide avalanche region (all InP)
2-mm-wide avalanche region (InGaAs absorber)
Room-temperature CV measurements made
Devices in packaging for low temperature measurements
The key challenge for developing infrared versions of the photon counting detector is the dark current. LL are now developing InP test structures to test different designs in this regard. Several devices have been made, and they are now going to be packaged, cooled, and measured.

19. Outline Motivation Moore Detector for TMT Heritage: LIDAR Conclusions
Why pursue photon-counting technology?
Why use Geiger-mode avalanche photodiodes (APDs)?
Moore Detector for TMT
Heritage: LIDAR
Conclusions
The following section describes the heritage that Lincoln Lab has in making photon-triggering LIDAR devices.

20. Si APD/CMOS Development History
1996
2009
APD’s
Discrete 4x4 arrays
4x4 arrays
wire bonded to 16-channel
CMOS readout
32x32 arrays
fully integrated with 32x32 CMOS readout
64 x 64 arrays 3D-integrated with
2 tiers of SOI CMOS
256 x 256 arrays
not to scale
This chart summarizes the development of these sensors over the past ten years.
We developed our own design of a silicon APD, and are currently developing sensors for 1.06 and 1.55 micron wavelength radiation.
The first brass board system slide used an external readout circuit coupled to a discrete 4×4 array
We next produced an integrated 4×4 array which had integrated CMOS readout circuits and wire bonded the small APD array to it.
We then developed a 32×32- and 64×64- pixel APD arrays integrated with a CMOS readout circuit with one circuit for each pixel.
The next design is a 256x256 array. The ROIC for this part is in fabrication.

21. LIDAR Imaging System
Microchip laser
Geiger-mode APD array
Imaging system photon starved. Each detector must precisely time a weak optical pulse.
A flash ladar system is like a flash camera, except the flash bulb is a laser that puts out short light pulses, and the pixel measures the arrival time of the light rather than its intensity.
Color-coded range image

22. A LIDAR Imaging Detector for NASA Planetary Missions
Parameter
Current
Goal
Space-Qualifiable NO
YES
Scalable to Large Format
CMOS ROIC Timing Resolution
250 ps
Pixel Size
50 mm
Multiplied Dark Current K)
unknown
<10-3 e-/s/pixel
QE (350nm,650nm,1000nm)a
45%,65%,5%
45%,65%,10%
Operating Temperature
293 K
90 K – 293 K
Radiation Limit
50 Krad(Si)b
Technology Readiness Levelc 2 4
Low field
High field
multiplier
Medium low field
absorber
These arrays will be fabricated for back-illumination with bump bonding, enabling high performance in a space-qualifiable focal plane.
The design of the ROIC will be finished by the end of 2009, with fabrication starting in early 2010.
Funding: $546,000
Duration: 3 years ( )
RIT and LL are funded to develop a LIDAR imaging detector for planetary missions. “QE” is a product of internal QE and probability of initiating Geiger mode avalanche; it does not assume anti-reflection coatings.

23. 32x32 APD/CMOS Array with Integrated GaP Microlenses
The bridge bonded focal plane has been incorporated into a hermetically sealable package with built-in thermoelectric cooler and temperature sensor. Also mass transported microlens arrays made of GaP have been attached to the focal plane.

24. Laser Radar Brassboard System (Gen I)
This slide shows a photo of the Gen I ladar system, and a 3D image of a Chevy van taken by the Gen I at noontime on a sunny day.
4  4 APD array
External rack-mounted timing circuits
Doubled Nd:YAG passively Q-switched microchip laser
(produces 30 µJ, 250 ps pulses at  = 532 nm)
Transmit/receive field of view scanned to generate 128  128 images
Taken at noontime on a sunny day 1 1

25. Conventional vs LIDAR Image
3D images enable:
easy image segmentation based on range,
object recognition signatures that are robust to lighting or reflectivity variations,
and object recognition through camouflage.
3D images can also be used to align intensity images from multiple sensors in different locations.
Here is an example comparing an intensity image of two vehicles with a 3D image of the same scene. One vehicle is obscured by a camouflage net. The 3D image, when rotated using rendering software, reveals the camouflaged vehicle clearly. It is not possible to make out the camouflaged vehicle at all in the intensity image, even with over 250 times as many photoelectrons per pixel.
Conventional image

26. 3D Imaging of Model Airplane
Single Frame
3D Display of Processed Image,
Probability of Detection Color-code
Color-code:
1 m range display
Airplane hanging on 6 mm rope
Multiple-frame coincidence processing of ~3-4 frames removes isolated dark counts
Image quality excellent due to low optical cross-talk between pixels

27. Rotatable 3D Images of Multiple Objects
Color-coded by Distance
Color-coded by Detection Probability
128x128 images recorded with scanned 4x4 array at 1.06 mm
Coincidence processed to remove background/dark counts
Dark blue equivalent to <2 photon average return (right image)

28. Outline Motivation Moore Detector for TMT Heritage: LIDAR Conclusions
Why pursue photon-counting technology?
Why use Geiger-mode avalanche photodiodes (APDs)?
Moore Detector for TMT
Heritage: LIDAR
Conclusions
Section on the motivation behind our technology development efforts.

29. Conclusions
Large-format photon-counting imaging detectors are within reach.
We are funded to make 256x256 and megapixel devices.
A 256x256 detector silicon-based array should be in testing by the end of the year.
The devices will be implemented in a broad range of low light level and LIDAR timing applications.
The slide describes the status of the Moore project.