The current and future capabilities of MCP based UV detectors

Abstract

At the heart of future space-based astronomical UV instruments will be a sensitive UV detector. Though there has been a death
of new UV mission opportunities, detector development has continued. Improvements have been made in spatial resolution, dynamic
range, detector size, quantum efficiency and background. At the same time the power and mass required to achieve these goals
have decreased. We review the current capabilities of microchannel plate based detectors at Berkeley, both in the laboratory
and aboard current on-orbit spacecraft. We also discuss what can be expected from the next generation of UV detectors over
the next decade.

Full text

Page 1

Astrophysics and Space Science
DOI 10.1007/sXXXXX-XXX-XXXX-X
The current and future capabilities of MCP based
UV detectors
John Vallerga, Jason McPhate, Anton Tremsin
and Oswald Siegmund
c ? Springer-Verlag ••••
Abstract At the heart of future space-based astro-
nomical UV instruments will be a sensitive UV de-
tector. Though there has been a dearth of new UV
mission opportunities, detector development has con-
tinued. Improvements have been made in spatial reso-
lution, dynamic range, detector size, quantum efficiency
and background. At the same time the power and mass
required to achieve these goals have decreased.
review the current capabilities of microchannel plate
based detectors, both in the laboratory and aboard cur-
rent on-orbit spacecraft . We also discuss what can be
expected from the next generation of UV detectors over
the next decade.
We
Keywords detectors:
crochannel plate
ultraviolet — detectors:mi-
1 Introduction
Imaging, photon counting detectors that use mi-
crochannel plates (MCPs) have been the detector of
choice for most UV astronomy missions over the last
two decade.In fact, the detectors on the SOHO,
EUVE, IMAGE, FUSE(Sahnow et al. 2000) , ALEXIS,
GALEX(Siegmund et al. 2004) , ROSSETA-ALICE,
and New Horizons-ALICE (Stern et al. 2007) missions
represent over 200+ detector years in space without a
failure (of the detectors). The Hubble servicing mission
in the fall of 2008 will install another UV instrument,
the Cosmic Origins Spectrograph (COS)(Green et al.
2003) .
John Vallerga, Jason McPhate, Anton Tremsin and Os-
wald Siegmund
Space Sciences Laboratory, University of California,
Berkeley, Berkeley, CA, 94720-7450, USA
Fig. 1.— COS-HST detector head with two adjoining
85 x 10 mm fields of view. Z stacks of three MCPs were
curved to match a 0.6m radius to match the focal plane
curvature. Cross delay line image readout anodes were
used.
The COS MCP detector (Vallerga et al. 2002) was
state of the art when it was fabricated in 2002 (Fig
1.). The format of the COS detector has two segments,
each 85 x 10 mm active area, separated by 9 mm. Each
detected photon event x,y position was digitized to 14
x 10 bits which could be transferred to the ground as
either a photon event list or a 2-d image ( 16384 x 128
pixels). The spatial resolution in the spectral (long)
dimension is 25 µm FWHM while the imaging (short)
dimension has a resolution of 50 µm FWHM. The event
electronics has a deadtime of 7.4 µm per event, resulting
in a deadtime fraction of 10% at an input count rate of
15,000.
Since the delivery of the COS detector, our group
at the Space Sciences Laboratory has been improving
the space-based capabilities of MCP detectors including
resolution, deadtime, uniformity, lifetime, power and
mass. Some of these improvements are the result of bet-
ter MCPs from the manufacturer. Other improvements
result from the adoption of new electronic technologies

Page 2

2
Fig. 2.— Readout anode types. The XDL consists of 2
serpentine delay lines and uses 4 amplifiers to determine
the time pulse arrival time difference. The XS anode
has many strips and amplifiers to measure the output
charge centroid. The pixellated anode ASIC detects
and counts events on a pixel scale
(ASICs, FPGAs) while still others result from special-
ized careful design to fit the observational application
(e.g. low power for a Pluto mission).
Table 1 summarizes some of the standard perfor-
mance parameters of the COS detectors as a baseline
and compares them to what is currently feasible in a
flight detector. There are certainly trade-offs among
these performance parameters. For example, larger for-
mat size usually requires more mass, and higher input
rates result in shorter MCP lifetimes for a fixed gain.
Yet there are often design paradigm shifts that can im-
prove certain parameters by orders of magnitude with-
out sacrificing other key parameters. Below we discuss
a few examples of these new designs, specifically better
MCPs and readout technologies, and refer the reader
to the references for more details.
2 MCP readouts
MCPs amplify a single photoelectron from a photocath-
ode into an exiting electron charge cloud whose centroid
represents the position of the incoming photon. Fig.
2 shows three methods of determining the centroid of
the electron cloud using a patterned anode: cross delay
line (XDL), cross strip (XS) and pixellated readout in-
tegrated circuit (ROIC). The XDL anode (Siegmund et
al. 1994) uses the temporal difference of the two charge
pulses at either end of a delay line to determine a linear
position in that axis. Though only 4 high bandwidth
amplifiers are required to read out the x,y position of
each event, large gains (≈107) are required to reach
high spatial resolutions. Typical (but well tuned) XDL
readouts can achieve resolutions of ≈ 20 µm FWHM.
As more than one event on an XDL anode at the same
time results in an incorrect position, the ultimate global
count rate limit for an XDL detector is approximately
1 MHz, though in practice 200 kHz at 10% deadtime is
typically achieved.
Fig. 3.— MCP flat field (zoomed to 1 x 1 mm) us-
ing a cross strip readout that resolves the MCP pores.
The pores are 12µm on 15µm centers. In this exam-
ple, the imaging spatial resolution would be limited by
the pore spacing rather than the readout anode, but it
is demonstrative of the capability of the new readout
technologies that could be combined with MCPs of a
finer pore pitch.
The XS anode (Siegmund et al. 2003) has a charge
amplifier stage on each strip, and the individually digi-
tized event signals are centroided to determine the sep-
arate x and y positions. By dividing up the anode area
into smaller strips, the capacitance of the electrodes,
and hence the noise, is decreased, allowing lower gain
operation even at better resolution. This capability of
many (≈64) parallel amplifiers became possible with
the advent of CMOS application specific integrate cir-
cuits (ASICs) with low noise and low power. XS anodes
can be easily scaled to larger formats by just adding
more amplifiers and ADCs, proportional to the linear
size of the array. We have achieved spatial resolutions
of the readouts of 5 µm at a gain of 5×105, even re-
solving the MCP pores (Fig. 3). The ultimate photon
spatial resolution would then be limited by the sam-
pling of the MCP pore spacing. Event rates using XS
anodes can be very high (> 1 MHz) if the digitization
of the ASIC amplifier outputs is also massively parallel
and the centroiding calculation is done with a digital
pipelined architecture using field programable gate ar-
rays (FPGAs) or digital signal processors.

Page 3

The current and future capabilities of MCP based UV detectors3
Fig.
(zoomed) taken with Timepix readout of a chevron
MCP stack. The charge cloud of every event was sam-
pled by the Timepix, centroided on sub-pixel scale, and
rebinned to 4096x4096. Pattern 5-6 (57 lp/mm) is just
resolved, consistent with the 8 µm pore spacing of the
MCPs. The width of one bar of the 5-1 pattern is 16
µm.
4.— Air force test pattern mask UV image
A recent development in MCP readouts is the
Medipix2 ROIC (Llopart & Campbell 2003; Vallerga
et al. 2004) . In this case, the anode is pixellated
(256x256), with every 55 µm pixel containing an am-
plifier, discriminator and counter. Events with signals
above a lower threshold of 1000 electrons are counted
and integrated on chip. As the image is digital, it can be
read out very fast, > 100MHz, allowing frame rates of
1 kilohertz with zero readout noise. Because the events
are integrated on chip , individual event time tagging
is lost, so event times are only accurate to the frame
time. What is gained are phenomenal input rates, up
to GHz rates. Of course, to operate at GHz rates for
extended periods would quickly deplete the gain of the
MCPs, unless the MCP can be operated at extremely
low gain, on the order of 104, which the low threshold
of the Medipix2 discriminator allows.
The Medipix2 pixel is a rather large (55 µm), as it
was originailly developed for x and gamma ray imaging.
The Timepix ASIC (Llopart 2007) is a modification
of the Medipix2 that mostly changed the discriminator
logic, allowing pulse amplitude determination for each
event, rather than just counting. Using this facility and
adjusting the detector parameters to spread the MCP
charge cloud over many pixels allows a very accurate
sub-pixel determination of the event centroid, similar to
the XS readout, but at lower gain. Fig. 4 demonstrates
the improved resolution performance when operated in
this mode, though sacrificing the GHz count rates to
200 kHz limit to avoid event collision between frame
readouts. Pattern 5-6 is just resolved corresponding to
57 lp/mm using an MCP with 8 µm pore spacing.
3 MCP uniformity
One of the consequences of the fabrication process for
MCPs is image modulation related to the multifiber
bundles. This is fairly typical of MCPs produced prior
to 2000 and in particular for large area MCPs that are
made by the block fabrication processes. Studies of the
physical presentation of these MCPs show that there
are distinct MCP pore misalignments and distortions
at the multifiber boundaries. One resulting effect are
changes of the local gain due to the pore distortions.
In addition, the pore distortions produce changes in
the direction of the emitted electrons at the output of
the MCP which produce the observed apparent spatial
deficit and excess of detected events across the multi-
fiber boundaries (Fig. 5) . At high resolution (Fig. 3)
the displacement and distortions of the pores at multi-
fiber boundaries are clearly observed. MCPs produced
by boule encapsulated fabrication have far less multi-
fiber edge modulation than block fabricated larger size
MCPs. Presumably this is because the misalignment
and stress at multifiber boundaries is worse for block
fabricated MCPs where the compression during high
temperature fusing is predominantly in one direction
rather than radial as in the case for a boule. Recent
production efforts have attempted to resolve some of
Table 1 Current MCP detector capabilities compared
to the COS detector
ParameterCOSCurrent Comment
Formats size (mm)90×10100×100
Spatial Resolution
(µm FWHM)
258MCP pore spacing
limited
MCP gain 107
107
105
104
XDL
XS
Medipix2
Global Count Rate
(kHz, 10% deadtime)
15200
1000
1000000
XDL
XS
Medipix2
Power (W)371.1Pluto-ALICE
Mass (kg) 330.66Pluto-ALICE

Page 4

4
Fig. 5.— Flat field images of two different detectors with two differenent sets of MCPs. On the left is a deep flat
field of a 10 x 10 mm section of the COS detector showing the image distortion at the MCP multifiber boundaries.
On the right is an optical image tube flat field (18mm diameter active area) using MCPs that had the multifibers
fused using a boule encapsulated fabrication method that shows little of the multifiber distortion.
these fabrication effects. Images obtained with smaller
33mm MCP stacks in XDL detectors (Fig. 5) show that
recently produced boule fabricated MCPs have no ob-
servable multifiber modulation in their intensity maps.
Acknowledgements
obtained through a series of NASA grants and con-
tracts, but the data was compiled under NASA grants
NAG5-12710 and NAG5-9149. The Medipix/Timepix
readout development was supported by NSF under
AURA AST-0132798-SPO6 (AST-0336888)
The studies presented here were

Page 5

The current and future capabilities of MCP based UV detectors5
References
Green, J. C., Wilkinson, E., & Morse, J. A. 2003,
Proc. SPIE, 5164, 17
Llopart, X., & Campbell, M. 2003, Nuclear Instruments
and Methods in Physics Research A, 509, 157
Llopart Cudie, X., Doctoral Thesis, Mid-Sweden University,
Sundsvall, Sweden, 2007
McPhate, J. B., Siegmund, O. H., Gaines, G. A., Vallerga,
J. V., & Hull, J. S. 2000, Proc. SPIE, 4139, 25
Sahnow, D. J., Gummin, M. A., Gaines, G. A., Fullerton,
A. W., Kaiser, M. E., & Siegmund, O. H. 2000,
Proc. SPIE, 4139, 149
Siegmund, O. H., et al. 1994, Proc. SPIE, 2280, 89
Siegmund, O. H. W., Tremsin, A. S., Vallerga, J. V.,
Abiad, R., & Hull, J. 2003, Nuclear Instruments and
Methods in Physics Research A, 504, 177
Siegmund, O. H. W., et al. 2004, Proc. SPIE, 5488, 13
Stern, S. A., et al. 2007, Space Science Reviews, 128, 507
Vallerga, J., Zaninovich, J., Welsh, B., Siegmund, O.,
McPhate, J., Hull, J., Gaines, G., & Buzasi, D.
2002, Nuclear Instruments and Methods in Physics
Research A, 477, 551
Vallerga, J. V., McPhate, J., Mikulec, B., Tremsin, A.,
Clark, A., & Siegmund, O. 2004, Proc. SPIE, 5490,
1256
This manuscript was prepared with the AAS LATEX macros
v5.2.