Title: Technical Status of Near Infrared Detectors for SNAP
1Technical Statusof Near Infrared Detectors for
SNAP
- Gregory Tarlé
- University of Michigan
- July 2002
2Outline
- State of NIR Detectors
- Dark Current
- Quantum Efficiency
- Read Noise
- Intrapixel Variation Dithering
- Michigan NIR Detector Lab
- Summary
3NIR Detectors
- Fully half of the SNAP focal plane is now NIR
sensitive. - At present there are two main technologies for
large megapixel arrays - InSb under development by the ALADDIN
(Advanced Large Area Detector Development in
InSb) project.? ORION - HgCdTe HAWAII (HgCdTe Array Wide Area Infrared
Imager) multiplexer developed by Rockwell Science
Center together with University of Hawaii. -
- need for passive cooling ? single technology
option (HgCdTe)
4State of HgCdTe NIR Detectors
- picture from RSC (progression)
- Here we discuss the state of the NIR detector
technology (RSC) Hawaii-2, Hawaii-2 RG, InSb
etc - discuss performance
- WFC3 , NGST, noise, Idark, QE
- Hg(1-x)CdxTe ?c determined by x this
determines operating temperature. - Rockwell Science Center (RSC) is the principal,
recognized source for Mercury Cadmium Telluride
(HgCdTe) infrared focal plane arrays (FPAs). - RSC has developed devices (NICMOS 256 x 256
FPAs, WFC3 1024 x 1024) for the IR channel on
HST. Also U Hawaii, ESO, Subaru, NGST
5HgCdTe Fabrication
Hybrid array has an array of HgCdTe detectors on
one layer and charge collection and addressing
circuitry (CMOS multiplexer) on another layer.
The two layers are connected with indium bump
bonds.
PACE Layer of CdTe is deposited on sapphire
substrate by chemical vapor deposition followed
by a layer of HgCdTe grown by liquid phase
epitaxy. MBE - Rockwell has developed a process
of fabricating detector layers in HgCdTe using
molecular beam epitaxy (MBE). This process
results in improved QE at short wavelength and
potential reduction in intrapixel variation.
6Hawaii RG Multiplexer
- NGST development
- (Hawaii-2RG)
- 0.25mm design rules
- 3-side close buttable
- 1, 4, or 32 outputs
- Bi-directional register to allow corner to
center scanning - Reference output R and internal reference
columns/rows - Guide window G output
- Fully selectable guide window within 2048 x 2048
architecture - Seamless guide mode/science mode readout
- Read noise 10 e- rms CDS ?N down to 3 e- rms
w/multiple samples achieved with lc 5 mm
material.
Four of these H1RG multiplexers are optically
stitched to produce a H2RG device
7SNAP NIR Detector Specifications
The Rockwell HgCdTe devices are our baseline
choice for the NIR system. WFC3 MBE material with
1.7 mm cutoff in the NGST 2k x 2k format (under
development) is a very good match for
SNAP. Performance Goals Read Noise 5
electrons (Fowler-4 read) Dark Current 0.1
e-/sec/pixel Quantum Efficiency gt 60
8NIR Detectors Dark Current
Dark current depends sensitively on the threshold
wavelength and the temperature.
SNAP requires dark currents below 0.1 e-/s/pix at
140?K
Measurements made at DCL (Goddard) on Hawaii-1R
with lc 1.7 mm at 150?K (courtesy of B. Hill).
Device Number
Achieved dark current meets SNAP specifications
9NIR Detectors Quantum Efficiency
Measurements made at Goddard-DCL on Hawaii-1R
(courtesy of B. Hill)
Early production lots have high QE but droops at
low wavelength.
- Rockwell seems to understand the MBE parameters
that control QE. - They have grown and tested new 1.7 mm material
- New devices have been tested at GSFC-DCL and have
uniformly high QE (slightly higher dark current).
We are confident that the SNAP QE solution will
be inherited from WFC3 development.
10NIR Detectors Read Noise
SNAP specification of 5e- assumes 10e- for a
single CDS and that 4 reads at the beginning and
end of an exposure will reduce this by a factor
of ?4. Recent development RSC has confirmed
earlier GSFC-DCL measurements with 1.7 mm
material that showed large read noise for the
WFC3 devices (25 e- with a single CDS reduced to
15 e- with 4 CDS compared to 10e-/5e- obtained
with 5 mm material). The cause remains unclear.
According to RSC this is a property of the 1.7 mm
material (not the MUX) and is expected to
significantly improve for devices with larger
wavelength cutoff (1.9 - 2.0 mm) and consequently
larger dark current. A RD program involving
additional production runs is required to solve
this problem. An additional run has been
recommended for WFC3 and should be complete at
the start of SNAP RD. If WFC3 fails to solve
this problem we have two options 1) We undergo
additional production lots at RSC 2) We modify
our readout strategy.
11Intra-Pixel Variations
- SNAP under-samples by factor of 3 at 1mm (worst
case). - Sensitivity variations on the scale of a single
pixel can significantly reduce photometric
accuracy of under-sampled images. - Accurate photometry ( 1 overall statistical,
1 relative systematic) is current NIR
performance target until SNAP science driven
requirements have been passed down to the
detector level. - Intrapixel response for the MBE HgCdTe has not
been measured yet. - Intrapixel response may be fine as it is or it
may not be. If not, Rockwell can work on
improving the deposition process to sculpt the
electric so as to more efficiently collect the
charges near the pixel edges. - Until we characterize the intrapixel response of
an actual device, simulations remain our only
tool to understand the impact of intrapixel
variations on photometry.
12Intra-Pixel Variation and Photometry
PACE process produced devices have ? 0.8 but by
controlling growth process MBE HgCdTe devices
should have ? closer to unity. Simulations show
that dithering is required for a lt 0.98. Even
for ? 0.8 a 2x2 ½ pixel dither pattern can
reduce photometry error to negligible levels for
all wavelengths.
Fractional Measurement Error
Fractional Sensitive Area (?) ? RdA/A
13Michigan Infrared Detector Lab
- We have begun to set up HgCdTe evaluation
facility at UofM - First measurements will focus on intrapixel
variation, read noise and dark current. - Rockwell Science Center has delivered a H1RG
multiplexer on a H2RG carrier. - ARC has produced a dewar and a DAQ system.
- Setup has been customized for the H2RG. Initially
we will use the H1RG multiplexer to debug the
system and establish noise floor. - The next step will be to acquire a science grade
FPA incorporating all the experience of the WFC3
and NGST developments to perform noise, QE and
intrapixel variation studies. - Eventually we will acquire additional science
grade device(s) to verify that we can meet all
SNAP NIR requirements. - The facility will serve as a prototype for large
scale HgCdTe characterization in Phase B.
14Ongoing HgCdTe Characterization Efforts
- Detector Characterization Lab (DCL) at Goddard
Space Flight Center (PI Ed Cheng) - The DCL currently supports the characterization
of HgCdTe detectors for the Hubble Space
Telescope Wide Field Camera 3 instrument. - NASA has funded four laboratories to develop and
assess the quality of NGST prototype detectors - The University of Hawaii Laboratory (PI Donald
Hall) will develop and characterize HgCdTe
detectors manufactured by Rockwell Scientific. - The University of Rochester Laboratory (PI
William Forrest) will develop and characterize
InSb detectors made by Raytheon Infrared
Operations. - The Independent Detector Testing Laboratory (PI
Don Figer) at Space Telescope Science Institute
and the Johns Hopkins University will
characterize both HgCdTe and InSb detectors in a
comparative hardware setup. - The Laboratory at NASA Ames Research Center (PI
Craig McCreight) is developing and characterizing
SiAs mid-infrared detectors.
15Multiplexer
16Mapping FPA sub-pixel photometric response
- At ? 1.3 ?m, a diffraction-limited 5 ?m spot
requires f lt 1.6 - Use precision molded aspherics, focal length
2.75 mm, f 0.76 - Pinhole light source at 200 mm, vary
longitudinal distance to compensate focal length
variation with IR wavelength - Move pinhole transversely to scan individual
pixel 1 mm per pixel - 4 ? 4 compound lens array or micro-lens array to
test multiple sensor regions at one time - Detailed design is underway and manufacture
should be achieved early in FY03.
17NIR Detectors Summary
- RD Phase NIR detectors
- Obtain and characterize a science grade HgCdTe
detector. - Assume that problem with anomalous read noise
will be solved by WFC3 and a solution compatible
with SNAP requirements will be found by the start
of SNAP RD. If not, we will need to find a
solution. This may require additional HgCdTe
production lot runs or a modified readout
strategy. - Noise reduction through multiple reads needs to
be verified. - Intrapixel variations will be studied.
- Demonstration that 2?2 dithering will produce
adequate photometry precision. - Perform detector modeling.
- Establish science driven requirements.
- Establish SNAP science grade specifications.
- Obtain and characterize a SNAP science grade
HgCdTe detector (optional 2nd detector). - Prepare for large scale FPA qualification.
- Demonstrate all SNAP NIR requirements can be met
with a SNAP science grade device. - Incorporate developments at RSC, WFC3 and NGST.