Availability of Detectors for Space-Based Astronomy
A Discussion by the NASA Astrophysics Working Group
1999 February 17
In the autumn of 1998, the Discipline Scientists of the NASA Office of Space Science requested
that the Astrophysics Working Group (AWG) look into the issue of detector availability from X-ray
to infrared wavelengths. We have solicited comments from individuals involved in instrument
development in order to make a preliminary assessment of the problem on the basis of ancedotal
evidence. We conclude that the issue of detector availability is a chronic problem that needs to be
addressed by NASA. Availability of detectors is not yet a critical problem, but this potential exists,
and it will certainly occur without action by the Office of Space Science.
Most X-ray astronomy missions projected for the next decade rely on either CCD or cryogenic
(microcalorimeter or STJ) technology. Several groups are actively engaged in development of
cryogenic detectors and a substantial amount of funding is currently being invested in the
technology not requiring substantial funding for its maintenance. Concerns with this inaccurate
view are addressed here.
X-ray CCDs are currently the state-of-the-art detectors of choice for many applications. They have
produced tremendous science returns from ASCA and are poised for launches on Chandra and
XMM. At least half a dozen missions proposed in the last MIDEX round rely on X-ray CCD
detectors for either spectroscopy or imaging spectroscopy, and CCDs are projected for use as
grating readout devices on Constellation-X in the next decade. There is clearly a continuing role for
these detectors, which provide a unique combination of high quantum efficiency, large format, and
moderate spectral resolution.
The photon-counting X-ray CCDs needed for all these missions provide the most difficult
challenge for detector manufacturers of any CCD type. They place very stringent requirements on
several areas of device performance, including read noise (must be less than 4 electrons rms, and
current state-of-the-art detectors achieve less than 2 electrons) and charge transfer efficiency (less
than 0.001% of charge lost on each pixel transfer). Operation at energies up to 10-15 keV requires
the use of either high resistivity bulk silicon (Chandra/ACIS) or high resistivity thick epitaxial
silicon wafers (XMM/EPIC MOS), neither of which has commercial applications. Similarly,
operations at low energies (below 0.5 keV) requires the use of special gate structures (thin gates or
open gates) or backside illumination with special passivation. Neither of these technologies have
commercial applications. X-ray astronomy must therefore rely on manufacturers who are willing to
pursue specially designed CCDs with applications solely for X-ray purposes.
Over the years, a number of CCD manufacturers have built detectors that were evaluated for use in
X-ray astronomy. Because of the extreme technical requirements and device complexity, these
devices can only be manufactured at highly specialized fabrication plants with enormous capital
investments and very high production quality standards. Over and over again, these sources have
dried up as the companies decided to consolidate operations or to cut unproductive research
projects. For example, the Chandra/ACIS CCDs were originally developed at Texas Instruments
(as an offshoot of their HST/WFPC CCD development) until they closed their research
development lab, shifted CCD development to Japan, and lost the technological capability of
producing state-of-the-art X- ray detectors.
Another example of this type of problem is provided by the Penn State University experience with
Loral. Ford Aerospace was building state-of-the-art optical CCDs in the mid-1980s. Ford
Aerospace was purchased by Loral and continued to develop CCDs for use on missions including
CRAF/Cassini and WF/PC 2. The Penn State group began working with this foundry on the design
of a new CCD optimized for use as a low-energy X-ray spectrometer on its CUBIC instrument.
Unfortunately, Loral decided to close down their research operations (where this work was being
done) and consolidate CCD production to their fabrication plant. Penn State's devices were
manufactured at the research plant in the midst of this consolidation. Because of decisions made at
corporate headquarters, Loral research labs was unable to obtain good quality silicon wafers, and
because of morale problems caused by the imminent closure of the plant, processing mistakes were
made. Out of three lots producing hundreds of CCDs, only one CCD was good enough for use as an
X-ray detector. This single device was the best CCD ever evaluated by Penn State, but with only
one sample, they were unable to use it for a mission. Shortly after the Penn State CCD lots were
finished, the Loral research lab was closed, and the Loral production lab has not been able to repeat
Only two CCD manufacturers in the world currently produce devices built on high resistivity
silicon that can be used as photon-counting X-ray detectors: Lincoln Laboratory and EEV. (A third
group was set up by MPE in Germany to build pn-CCD detectors for XMM/EPIC and ABRIXAS,
but it is rumored that this plant has been or is about to be closed down.) Lincoln Lab CCDs are
excellent, state-of-the-art devices. Unfortunately, to date only MIT has had access to these
detectors. In spite of an enormous financial investment by the Chandra project, Lincoln Lab was
not very successful in producing devices that perform at low energies — only two backside
illuminated CCDs built for Chandra were of flight quality, and these have significant and poorly
understood problems in terms of charge transfer efficiency and energy resolution. EEV is the only
manufacturer producing X-ray CCDs on a commercial basis and is the only viable competitor to
Lincoln Laboratory in this area. They have an excellent track record in development of new devices
for X-ray use (financed by XMM), including several technologies that extend low energy quantum
efficiency (using open gates or backside illumination).
While the situation regarding X-ray CCDs may not be desperate today, since there are two (or
three) sources of these devices at the present time, past experience suggests that this situation could
change at any time. Lincoln Labs is not subject to corporate takeovers, but it is difficult to predict
their long-term commitment to continued X-ray CCD development and fabrication. EEV has
provided assurances on several occasions that they have a long-term commitment to CCD
fabrication, and in particular, to X-ray CCD development, but in the absence of program funding to
keep both their interest and their expertise alive, how long will EEV's commitment survive? As the
end of the XMM program approaches, the company is not pursuing any other major photon
counting X-ray CCD program. It would be prudent to continue funding X-ray CCD development at
a level sufficient to retain EEV's interest in this technology, lest we find ourselves at the mercy of a
sole vendor for Constellation-X and other future missions.
The recent explosive interest in high spatial resolution and high dynamic range imaging and
interferometry puts new requirements on IR detectors. Scientific goals such as extra-solar planet
detection, imaging of stellar disks and dust rings, detection of zodiacal light around other stars
require interferometers and chronographs that operate in the near IR with virtually noiseless
detectors. Existing or new photon counting detectors are ideal for these applications if their
sensitivity can be extended into the near IR with high quantum efficiency. GaAs photocathodes
now have greater than 40% quantum efficiency to 0.9 microns and some work has been done on
developing such materials as InGaAs, which can extend the sensitive range to nearly 2 microns.
Support for IR photocathode development and its integration into existing or new photon-counting
detectors could be crucial for such NASA missions as TPF and NGST. Any discussion of IR
detectors is complicated by the fact that coverage of the spectrum from the near-IR to 200 microns
requires five different types of detectors.
Detectors used for IR astronomy have typically resulted from useful collaborations between small
divisions of aerospace/defense companies and university researchers. The former wanted these
detectors for high-background applications; the latter were tasked with making them suitable for
low-background applications, astronomy in particular. The large companies fabricate the detectors
and the NASA-supported researchers provide the electronics and the critical testing. With the end
of the Cold War, consolidation of defense contractors began apace, and many of the corporations
began to realize that their detector fabrication operations were not central to their "mission."
The result has been a rapid decrease in vendors capable of producing viable detectors, less
motivation for considering the needs of the astronomical community, and a pervasive concern
among astronomy users that the situation will only get worse in the future. This concern stems
partly from the recent trend for large companies to take over detector houses (Boeing absorbed
Rockwell, and Raytheon absorbed Santa Barbara Research Corporation [SBRC]), and partly from
the need for the very best multiplexers for future projects. The best foundries (producers of the
multiplexers/electronic readouts, also known as "muxes") are either closed or being sold. Even the
very good SBRC has sometimes resisted progress to the most sensitive devices. The large
aerospace companies do not have small, unique customers such as space astronomy in mind when
they consider about productivity and profit.
SBRC is a highly competent vendor, but strategic decision-making is now done by their parent
company Raytheon. This has led to numerous difficulties in procuring suitable detectors for
astronomical applications. Some examples are the following:
The University of Rochester (UR) insisted on the starting material purity to get good response
below 50K with InSb (Indium Antimonide) arrays. SBRC initially resisted using the purest
starting material, although they had found the reason (impurities) for the decline in QE below
SBRC developed various front-side and backside passivations for InSb; the first front-side
passivation they produced had no image persistence, but small wells (50000 e-). Later versions
have image persistence and larger wells, but the persistence problem did not attract their
attention until it became an issue in a military program.
SBRC changed the backside passivation without informing UR about it. SBRC was still
delivering arrays that seemed very good, but they had a dark square of lower response in the
center. UR tests showed the region to have a quantum efficiency that declined with temperature.
This was ultimately traced to the passivation. Once SBRC agreed to strip it off, and retest it, the
quantum efficiency behaved as expected. But considerable resistance at SBRC had to be
overcome for the correction to be made.
SBRC developed an excellent cryogenic mux for IRAC at Hughes Carlsbad. That was closed,
so the mux technology was transferred to the Hughes Newport Beach Foundry with many
Carlsbad people involved. But the new muxes exhibited what UR calls "mux bleeding". SBRC
had delivered arrays that they thought were perfect, but these had to be eliminated because of
the mux bleeding. SBRC had not noticed the problem. Eventually IRAC went back to the
residuals from the Carlsbad lot and got a space qualification exemption.
These examples, which can be regarded as typical experiences, underscore two points: (1) for
sensitive devices, research astronomers, the end users of these devices, are more attentive to the
subtle effects which need to be overcome for space missions, and (2) technology transfer is in
practice difficult. From our point of view, the specific problem with SBRC is not a question of
competence, but that Raytheon now makes the strategic decisions, which are likely to be
disadvantageous to astronomy. SBRC has neither the facilities nor the appropriate personnel to do
the testing required when pushing the envelope, which SIRTF/IRAC does, and which NGST will do
to an even greater degree.
In the energy band above ~100 keV, where focussing optics are no longer practical, large areas
(>0.5 m2) of pixelated detectors are required to achieve interesting sensitivities. Although coded
aperture instruments may have useful sensitivity near the lower end of this energy range, it is
generally conceded that the Compton-scatter telescope technique provides the best combination of
wide field of view, sensitivity and angular/energy resolution. However, both techniques measure
their performance in terms of stopping power of the detector material, pixelization or position
resolution, and energy resolution. Detector options include scintillators, germanium, silicon,
CdZnTe and liquid Ar/Xe. New developments in solid state detectors have potential for enabling
much more capable gamma-ray instruments in the future. The realization of this potential will not
occur without the support of the scientific communities for there are few commercial applications
for these costly detectors.
CdZnTe is the newest and one of the most exciting new detector materials applicable to gamma ray
instruments. It, along with Si, has the possibility of operation at or near room temperature, thus
making implementation of a large detector system vastly simpler. It is configured as either a strip
detector with parallel electrodes on each face, as a pixel detector with pad electrodes on one face, or
as a single channel device up to ~1 cm on a side. The main areas of development for astrophysical
applications are: 1) better quality material for larger volumes; 2) improved electrode design for
position readout and hole-signal rejection; 3) greater electrode reliability for large arrays; 4) hole-
trapping correction to improve energy resolution; 5) understanding and minimizing the activation
background; and 6) evaluation of radiation damage effects and annealing to restore performance.
CdZnTe detector development has been largely driven by the medical imaging community but
there is still no reliable source for large quantities of high quality CdZnTe. Thick silicon strip
detectors provide both good spectral and spatial resolution for a scattering detector in a Compton
telescope but have limited stopping power for other applications.
Germanium detectors have been available and used in space missions for years to provide the best
spectral resolution possible. However, these detectors were in the form of large co-axial detectors
without positioning resolution appropriate for good imaging. Recent developments in contact
technology have resulted in planar germanium strip detectors similar to the similar to the silicon
strip detectors used in high-energy physics experiments. The thick (1 - 2 cm) planar germanium
detectors with strips on both cathode and anode provide the potential for both high-resolution
spectroscopy and good spatial resolution critical to Compton telescope designs.
There is currently no US source for these detectors. Due to the cost of these detectors and the
requirement for cryogenic temperatures, there is no commercial market to drive the development.
The technology has been developed in small demonstration programs at a few US research
institutions and one French company. The nuclear physics community of the DOE has recently
expressed interest in these germanium strip detectors for the Greta detector which will replace
Gamma sphere. Without the investment by the scientific communities, these technologies, which
are critical to the next generation of gamma ray space instruments will not be available.
While the situation is in some ways better than in other wavelength regimes, there are still great
difficulties in procuring suitable detectors for UV/optical applications.
Timely delivery of astronomical-quality CCD imaging devices remains a problem. In particular, the
backside illumination required for optimal sensitivity generally lies far outside the routine
processing capabilities of the mass manufacturers of commercial devices, leaving us with a small
number (sometime even zero) of vendors who at any given time can realistically deliver even a few
such imaging devices. This situation is especially acute for proposed survey instruments that need
large focal plane mosaics of 10-20 or more large-format CCDs; most vendors simply cannot
promise the delivery of such large numbers in less than two years or more. This puts any
accelerated mission (especially Explorer-class missions, which are intended to launch within 40
months of selection) at grave risk.
There are many potential missions that require solar-blind photon-counting detectors. Delay-line
detectors, the current detectors of choice, have limits on resolution, pixel size, count rate, dynamic
range, and stability of its point spread function which either do not meet or barely meet the
scientific requirements of these missions. In the recent past, the support environment has not been
conducive to developing new detectors due mostly to the large investment in one or two existing
technologies. Development of new UV detector technologies that have extended capabilities should
be encouraged in tandem with support for existing technologies. We note that electron-bombarded
CCDs are currently showing some promise as alternatives to delay-line detectors, but they are still
The Astrophysics Working Group is deeply concerned about future availability of flight-quality
astronomical detectors at all wavebands. The situation is not yet acute, though the problem is
chronic and the situation is increasingly unstable. It may be that NASA will have to engage in long-
term strategies that will assure availability when detectors are actually needed. For this reason, we
believe that this is primarily a strategic issue that ought to be considered by the Space Science
Advisory Committee (SScAC) rather than the AWG; the AWG will be happy to assist the SScAC if
we can be of assistance. A temporary (band-aid) partial solution might for NASA to carry out a
fairly complete inventory of already existing NASA detector assets from previous and current
missions. For instance, projects like STIS (completed) or ACS (under development) should
generate a fair number of spare imaging devices of performance comparable to the flight detectors.
In addition, there should be a large number of rejected devices that nevertheless might well serve
the needs of a less-demanding mission. These valuable assets should be preserved and a permanent
inventory system established to document their characteristics for future missions, particularly
Explorer-class missions. NASA might also consider small, low-cost follow-on procurements where
possible, once a vendor is geared up to produce as space-qualified detectors suitable for
astronomical applications. This might also be carried out in partnership with other agencies with
similar if not identical concerns, such the Department of Defense, the National Reconnaissance
Office, and the European Space Agency.
In any case, NASA needs to guard against allowing an existing hardware capability to die; trying to
duplicate lost capabilities years later is not only a false economy, it sometimes proves to be
virtually impossible to accomplish in a limited time and is potentially fatal to any mission requiring
delivered state-of-the-art detectors on time scales shorter than several years.
Based on contributions from Michael D. Bicay (Caltech), David N. Burrows (Penn State Univ.),
John Geary (CfA), F.Rick Harnden, Jr. (CfA), W. Neil Johnson (NRL), and Judy Pipher (Univ. of
Rochester). Compiled by Bradley M. Peterson (Ohio State Univ.).