University of Huddersfield Repository
Akimov, D, Alner, G, Araujo, H, Bewick, A., Bungau, Cristian, Burenkov, A.A., Carson, M,
Chagani, H., Chepel, V. and Cline, D.
The ZEPLIN-III dark matter detector: Instrument design, manufacture and commissioning
Original Citation
Akimov, D, Alner, G, Araujo, H, Bewick, A., Bungau, Cristian, Burenkov, A.A., Carson, M,
Chagani, H., Chepel, V. and Cline, D. (2007) The ZEPLIN-III dark matter detector: Instrument
design, manufacture and commissioning. Astroparticle Physics, 27 (1). pp. 46-60. ISSN 09276505
This version is available at http://eprints.hud.ac.uk/14186/
The University Repository is a digital collection of the research output of the
University, available on Open Access. Copyright and Moral Rights for the items
on this site are retained by the individual author and/or other copyright owners.
Users may access full items free of charge; copies of full text items generally
can be reproduced, displayed or performed and given to third parties in any
format or medium for personal research or study, educational or not-for-profit
purposes without prior permission or charge, provided:
The authors, title and full bibliographic details is credited in any copy;
A hyperlink and/or URL is included for the original metadata page; and
The content is not changed in any way.
For more information, including our policy and submission procedure, please
contact the Repository Team at: E.mailbox@hud.ac.uk.
arXiv:astro-ph/0605500v1 19 May 2006
The ZEPLIN-III dark matter detector:
instrument design, manufacture and
D. Yu. Akimov a, G. J. Alner b, H. M. Araújo c,b, A. Bewick c,
C. Bungau c,b, A. A. Burenkov a, M. J. Carson d, V. Chepel e,
D. Cline f , D. Davidge c, J. C. Davies d, E. Daw d, J. Dawson c,
T. Durkin b, B. Edwards c,b, T. Gamble d, C. Ghag g,
R. J. Hollingworth d, A. S. Howard c, W. G. Jones c, M. Joshi c,
J. Kirkpatrick d, A. Kovalenko a, V. A. Kudryavtsev d,
I. S. Kuznetsov a, T. Lawson d, V. N. Lebedenko c , J. D. Lewin b,
P. Lightfoot d, A. Lindote e, I. Liubarsky c, M. I. Lopes e,
R. Lüscher b , J. E. McMillan d, B. Morgan d, D. Muna d ,
A. S. Murphy g , F. Neves e, G. G. Nicklin d, S. M. Paling d,
D. Muna d , J. Pinto da Cunha e, S. J. S. Plank g, R. Preece b,
J. J. Quenby c, M. Robinson d, C. Silva e, V. N. Solovov e,
N. J. T. Smith b, P. F. Smith b, N. J. C. Spooner d,
V. Stekhanov a, T. J. Sumner c,∗, C. Thorne c, D. R. Tovey d,
E. Tziaferi d, R. J. Walker c, H. Wang f , J. White h & F. Wolfs i
a Institute
b Particle
for Theoretical and Experimental Physics, Moscow, Russia
Physics Department, Rutherford Appleton Laboratory, Chilton, UK
c Blackett
d Physics
e LIP–Coimbra
f Department
Laboratory, Imperial College London, UK
and Astronomy Department, University of Sheffield, UK
& Department of Physics of the University of Coimbra, Portugal
of Physics & Astronomy, University of California, Los Angeles, USA
g School
of Physics, University of Edinburgh, UK
h Texas
i University
A&M University, USA
of Rochester, New York, USA
Preprint submitted to Elsevier Science
4 February 2008
We present details of the technical design and manufacture of the ZEPLIN-III dark
matter experiment. ZEPLIN-III is a two-phase xenon detector which measures both
the scintillation light and the ionisation charge generated in the liquid by interacting particles and radiation. The instrument design is driven by both the physics
requirements and by the technology requirements surrounding the use of liquid
xenon. These include considerations of key performance parameters, such as the
efficiency of scintillation light collection, restrictions placed on the use of materials
to control the inherent radioactivity levels, attainment of high vacuum levels and
chemical contamination control. The successful solution has involved a number of
novel design and manufacturing features which will be of specific use to future generations of direct dark matter search experiments as they struggle with similar and
progressively more demanding requirements.
Key words: ZEPLIN-III, dark matter, liquid xenon, radiation detectors, WIMPs
PACS: code, code
ZEPLIN-III is a two-phase (liquid/gas) xenon detector developed and built by
the ZEPLIN Collaboration, 1 which will try to identify and measure galactic
dark matter in the form of Weakly Interacting Massive Particles, or WIMPs
[1,2]. Upon completion of physics testing now underway at Imperial College,
the system will join the ZEPLIN-II [3] and DRIFT-IIa [4] experiments already
operating 1100 m underground in our laboratory at the Boulby mine (North
Yorkshire, UK).
Two-phase emission detectors based on the noble gases date back several
decades [5]. In last decade, this technology has gained a new momentum in
view of increasing interest for searching rare events, WIMPs in particular,
requiring both large detection masses and high discrimination against background. In its previous work, the ZEPLIN Collaboration has explored the potential of high-field xenon systems to enhance sensitivity and background discrimination [6,7,8]. The operating principle relies on different particle species
∗ Corresponding author; address: Astrophysics Group, Blackett Laboratory, Imperial College London, SW7 2BW, UK
Email address: t.sumner@imperial.ac.uk (T. J. Sumner).
1 Edinburgh University, Imperial College London, ITEP-Moscow, LIP-Coimbra,
Rochester University, CCLRC Rutherford Appleton Laboratory, Sheffield University, Texas A&M, UCLA.
generating different amounts of vacuum ultra-violet (VUV) scintillation light
and ionisation charge in liquid xenon (LXe). The ratio between these two signal channels provides a powerful technique to discriminate between electron
and nuclear recoil interactions. WIMPs are expected to scatter elastically off
Xe atoms, much like neutrons, and the recoiling nucleus will produce a different signature to γ-ray interactions and other sources of electron recoils.
WIMP detectors differ from more traditional detectors of nuclear radiation in
that they require: i) extremely low radioactive and cosmic-ray backgrounds,
addressed by the use of radio-pure materials and operation deep underground;
ii) excellent discrimination of the remaining background events, especially
for electron recoils; iii) a low energy threshold for nuclear recoils, since the
kinematics of WIMP-nucleus scattering results in a very soft recoil spectrum
(.100 keV).
Monte Carlo simulations [9,10] were essential in key areas to inform the design
of the instrument. Acceptable levels of trace contamination must be set for all
detector materials, requiring simulations of internal and external backgrounds
expected from each component. Cosmic-ray-induced backgrounds also need
careful calculation, since experimental measurements would require nothing
short of a dedicated WIMP detector. These simulations establish the residual
electron/photon and neutron event rates and spectra. Detailed detector simulations leading to predicted data timelines can be used to find the level of
discrimination and energy threshold which can realistically be achieved. Feedback from this process into the design process has been essential for ZEPLIN
III. In addition, the data produced by two-phase detectors are often complex, and particular simulations are required to help extract actual physics
parameters. Finally, realistic datasets help with planning the data acquisition
electronics and the data analysis software.
In this paper we describe the instrument design philosophy, the engineering
design solutions and the manufacturing processes adopted. In a separate paper
[11] we present full performance Monte Carlo simulations for the final, as built,
The ZEPLIN-III instrument
There are four important design requirements for a dark matter detector: a
low energy threshold, good particle discrimination, 3-D position reconstruction
and a low background within the fiducial volume. The ZEPLIN III approach,
as shown in figure 1, tries to push the boundaries of the two-phase xenon
technique to simultaneously achieve the best performance possible in these
four aspects.
Fig. 1. Cross-sectional views of the ZEPLIN III instrument showing the key system
design concepts. The rendered CAD representation shows the copper parts.
ZEPLIN-III achieves a low threshold for the primary scintillation by placing
its photo-detectors, photomultipliers (PMTs) in the liquid phase and by using
a flat planar geometry. Using PMTs in the liquid removes two interfaces, both
with large refractive index mismatches and puts in an additional interface at
which total internal reflection also works to improve the light collection for the
primary scintillation. The planar geometry gives a large solid angle acceptance
and lessens the dependance on surface reflectivities. A low threshold for the
electroluminescence from the gas phase which provides the secondary signal
is achieved by using a high electric field in the gas region to produce high
levels of photon emission per electron emitted from the surface and by using
refraction at the liquid surface to produce a ‘focusing’ effect for the light onto
the immersed PMT array.
ZEPLIN-III achieves good particle discrimination between the nuclear recoil
signals expected from WIMPs and the electron recoils from photon backgrounds by employing a two-phase design which allows both scintillation and
ionisation to be measured for each event. The ratio of these two signals depends on the particle species. The effectiveness of this discrimination depends
on the width and separation of the distributions for each species. It turns out
[12] that the discrimination is improved by working at moderate electric fields
which increases the separation between the two distributions and improves the
statistical uncertainties of the ionisation signal. Some discrimination against
nuclear recoil signals from neutron elastic scattering is obtained by having
good 3-D position reconstruction which can identify the multiple scattering
expected from the much higher cross-sections for neutron scattering than for
WIMP scattering. Efficient measurement of the ionisation relies on achieving
a long lifetime against trapping for free electrons in the liquid. This requires
ultra-pure xenon as free from electronegative impurities as possible. The target volumes must be constructed as high vacuum vessels and a dedicated gas
purification system is needed.
ZEPLIN-III achieves good 3-D position reconstruction by using an array of 31
2” diameter photomultipliers. These provide sub-cm 2-D spatial resolution in
the horizontal r, θ plane. Resolution in the z co-ordinate at the ∼ 50 µm level
is obtained from the timing between the primary and secondary scintillation
ZEPLIN-III achieves a low background partly by operation underground and
partly by using a very restricted range of materials for its construction. Although the PMTs are the largest specific contributors to the background budget it is important that careful attention is paid to all materials used as these
exceed the PMT mass by two orders of magnitude. In addition it is planned to
eventually replace the PMTs by low-background versions which are currently
in development.
In the following sections we detail the design and manufacture of the individual parts of the ZEPLIN III experiment. These include the target volume, the
cooling system, the outer vacuum jacket, the gas handling system, including
the safety reservoirs, and the data acquisition system. In the final section we
provide data from surface commissioning tests which validate the key performance parameters for ZEPLIN III
The target volume
The detailed design of the inner components within the target volume is shown
in figure 2.
3.1 The PMT array
Inside the xenon vessel is the array of 31 PMTs, immersed in the liquid phase,
looking up to a ≃40 mm-thick liquid xenon layer on top of which is a 5 mm
xenon gas gap. The figure shows a cross-sectional view through a centre line
passing through 5 of the 52 mm diameter PMTs. The others are arranged in
a hexagonal close-packed array with a pitch spacing of 54 mm. A pure copper
‘screen’ has an array of 53 mm holes into which the PMTs fit. This provides
Fig. 2. Cross-sectional assembly drawing of the internal chamber volume of ZEPLIN
both light screening and electrical isolation between the PMTs. It has an
outer diameter of 340 mm and a height of 128 mm. For ease of manufacture
the total height of the ‘screen’ was made in four sections. Each PMT hole
through the copper ‘screen’ has a diameter of 53mm giving a 1 mm wall
minimum thickness between each PMT. Two techniques were used to produce
such thin wall section through such a thickness of copper; wire erosion and
boring. Both worked but the boring produced a better surface finish. Sitting
directly on top of the ‘screens’ is another copper disc with holes in it. This
time the thickness is 7 mm and the holes are finished with highly polished
conical sections to improve the light collection; this plate is hence referred to
as the ‘PMT mirror’.
Each PMT has 15 pins to which connections must be made (12 dynodes, anode,
cathode and focus). However it would be impractical to bring all 465 connections out through individual UHV electrical feeds through the bottom thick
copper flange. Instead all the PMTs are run from a common high voltage supply and dynode distribution system which reduces the amount of feedthroughs
to just 47. The corresponding dynode pins on each PMT are connected together using a stack of 16 thin copper plates, held apart with small quartz
spacers, below the PMT array. Each 2mm thick plate has a different pattern
of holes (see figure 3) allowing connection to each pin in turn whilst the others
pass through with clearance. Connections between the copper plates and the
PMT contacts were done by first cold welding a pin into the copper plate and
then using spring loaded tubes to join the two pins together (see figure 4).
The pins used in the copper plates were made in copper with a gold coating
and these were inserted into tight fitting holes in the plates using a drill press.
The spring contacts were made from stainless steel tubing with reduced wall
sections and slots. These contacts provide enough friction for retention of the
PMT against buoyancy forces during immersion in liquid xenon. Connection
between each plate and its single UHV coaxial feedthrough was again made
by a direct spring loaded tube but with the addition of gold-plated copper
wires with silver-plated copper adaptors to provide the extensions between
end contacts. The anode connection from each PMT is brought out separately
on a dedicated coaxial UHV feedthrough in a similar way. The specific arrangement of the 16 copper plates can be seen in figure 4. The upper and
lower plates are connected to ground. The second lowest plate is connected
to the PMT cathodes, the next 11 are connected to dynodes 1 to 11 in turn.
Above that there is then another grounded plate and between this and the top
ground plate is dynode 12. The two grounded plates either side of the dynode
12 plate deliberately provide both extra capacitance to ground for that dynode and prevent cross-coupling with other connections. Copper tubes provide
shielding along the run of each anode output connection. Shielded cables pass
across the outer vacuum jacket space to connectors in its base plate. A single
external voltage divider chain is used to provide all the common dynode voltages. To ensure reasonably well matched gains when running from a common
HV supply, PMTs were procured with gains within prescribed limits. Once
selected the batch of 35 PMTs (ETL D730/9829Q) was tested and calibrated
at low temperature with Xe scintillation UV light prior to installation in the
detector [13]. The PMTs were customised specifically for ZEPLIN III in two
ways: firstly a conductive pattern of so-called ‘fingers’ was deposited on the
inside of the window to avoid saturation at high count rates, and secondly
to provide a modified pin-out arrangement to facilitate the use of the copper
interconnection plates. The PMTs are operated with the cathode at ground
3.2 The electric field
Proper operation in two-phase mode requires that there be a sufficiently high
electric field in three distinct regions. In the active volume of the detector the
electric field helps to separate ionisation charge released from the track of the
interacting particle before it can recombine. This field must be directed such
that the electrons start ‘drifting’ towards the liquid surface. Hence the field
in this first region is called the ‘drift’ field. The second critical region is at
the liquid/gas interface. Here the field in the liquid must be high enough to
efficiently extract the electrons into the gas phase. This not only increases the
signal strength but also prevents charge build-up at the surface. This field is
called the ‘extraction’ field. Finally in the gas phase the field must be high
enough for the accelerated electrons to produce excitation in the gas atoms.
The excited atoms then form excited dimers followed by dissociative radiative
emission in the usual way, which produces the signal seen by the PMTs. This
Fig. 3. The 16 2-mm copper plates used to make the internal PMT dynode interconnections.
last field is called the ‘electroluminescence’ field. These three fields can either
be produced by setting up a segmented electrode structure producing distinct
regions, as is done in ZEPLIN II [14], or, as in the case of ZEPLIN III, a
single pair of outer electrodes can be used to produce all three at once. The
advantage of the latter is the absence of any physical electrode structure in the
liquid which could then be a source of background and/or feedback. However
it does mean that a single much higher individual voltage is required and the
fields can not be controlled independently. The two electrodes used are the
PMT Screen
PMT Cathode Pin
PMT Anode Pin
PMT Dynode 12 Pin
S.S. Tube
Copper Silver
Plated Slotted
Copper Gold
Plated Wires
Copper Silver Plated
Slotted Pins
Fig. 4. Various spring loaded contacts used to connect the PMT pins to the copper
solid flat plate (‘anode mirror’) above the gas gap and a wire plane (‘cathode
grid’) 40 mm below it in the liquid. The 8-mm top plate is made from copper
and its bottom surface has been lapped using optical techniques and left highly
polished. Up to 40 kV can be applied between the two ‘electrodes’.
A second wire grid (‘PMT grid’) is located 5 mm below the cathode grid and
just above the PMT array. This defines a reverse field region just above the
PMTs which suppresses secondary signals from low-energy background photons from the PMTs and also helps isolate the internal PMT photocathode
fields from the external high electric field. The diameter of the electrode structure is ∼ 40 cm, whilst that of the PMT array is 34 cm. The fiducial volume
will be defined by a combination of primary to secondary timing and position
recovery from the PMT hit pattern, and it will be well inside the PMT array
diameter. This ensures that the electric field will be very uniform over the
fiducial volume region. Field and electron trajectory simulations, produced
using ANSYS [15], are shown in figure 5.
The stainless-steel wire grids were strung from continuous lengths of 100 µm
diameter wire wound around copper formers. The position of each wire was
controlled by slots machined into the formers (see figure 6). The wires were
tensioned using two techniques. Firstly the formers were elastically deformed
whilst the wire was wound and secondly the winding jig tensioned the wire
as it was wound. Once the grid winding was complete the wire was anchored
and the formers were then released from their restraining jigs.
Fig. 5. Electric field distribution within the target volume as computed using ANSYS [15]. On the right is shown an expanded view of the drift paths of electrons
near the right-hand gap between the two electrodes.
Some consideration was given to whether the anode mirror should be coated to
enhance its reflectivity. The performance of polished copper is quite uncertain
at VUV wavelengths, depending on the surface finish, oxidation state and
possible LXe condensation onto the cold surface in the gas phase. Only a
single measurement has been found, indicating R=25% for normal incidence
for a clean-cut surface [16]. However the simplicity of leaving this surface as is,
the uncertainty of using coatings in a high-field application and the desire not
to compromise the spatial reconstruction argued for not using any coating.
Fig. 6. On the left are the copper formers for the wire grids. The inset detail shows
an expanded view in which the slots cut to control to wire positioning can be seen.
On the right is a view of the assembled PMT array in which the PMT grid can be
3.3 The xenon transport system
Two copper access pipes are included for movement of xenon in and out of the
target vessel (see figure 7). One surfaces above the liquid level in an unconfined
volume as is used as a ‘Gas Inlet’. The second has a double tube structure with
an open ended inner pipe connected directly to the main xenon liquid volume,
and an outer pipe which vents to the outside through the ‘LXe Outlet’. The
outer pipe is sealed at the top and the inner opens above the liquid surface
and essentially allows a ‘syphon’ action during emptying. Transfer of xenon in
and out of the target vessel is independent of the cooling system.
LXe Vessel Shell
LXe Vapour P 1
Anode Mirror
Copper Rings
LXe Vessel
Base Flange
LXe Outlet
Gas Xe
Fig. 7. Arrangement of the pipework used for xenon transfer in and out of the target
3.4 The target vessel
The containment vessel for the xenon target must perform as both a high
vacuum vessel, for purity reasons, and a pressure vessel for safety reasons.
The pressure vessel design was done following the relevant British Standard
(BS5500:1997). This safety standard dictates the cylindrical wall, dome and
bottom flange thicknesses which are dependent on the material and processes
used. The vessel was required to be certified to 6 bar absolute. The material of
choice was determined by requiring the product of total mass times radioactive
content be a minimum. Added to this prime requirement was then the need
for the material to be suitable for manufacture of the vessel. OFHC copper
type C103 was selected. This required 4 mm wall thickness on the cylindrical sections, 3 mm for the spun domes and 25 mm for the flat bottom flange.
To minimise the likelihood of inclusion of any impurities electron-beam welding was used throughout and the number of welds was kept to a minimum.
In particular the cylindrical section was rolled in one piece. Stainless steel
parts were used for some specialist components which would have been very
difficult to make out of copper, such as vacuum knife edge flanges and vacuum HV feedthroughs for which commercial parts were used. Where necessary
these stainless parts were also electron-beam welded to the copper. Welding
techniques adapted to our requirements were developed by The Welding Institute, UK[17], in close cooperation with us. This included setting the welding
parameters and optimising the structural/thermal design of the weld joints.
All safety critical welding was done by certified processes and copper witness
plates were used to ensure proper and complete breakthrough as all welds were
required to show full-depth penetration. Special jigging was required to hold
all seams for welding securely in place during the process. On completion all
joints were leak-tested down to the level of ∼ 10−10 mbar.l.s−1 .
The electrical feedthroughs for the PMT dynode connections were fitted in
with screw threads with thick indium coated onto them using an ultrasonic
soldering iron. The demountable vacuum seal between the cylindrical section
and the bottom flange was done using a stainless steel gasket with double knife
edges and an indium wire at both copper surfaces (see figure 8). All copper
Vessel Shell Flange
S.S. Gasket Cu Spacer
Indium Wires
Vessel Base Flange
I Assembling Before Sealing
Fig. 8. Vacuum seal between the
cylindrical wall section and the bottom flange
II Assembling After Sealing
before (I) and after (II) sealing.
parts were cleaned, starting with a coarse hand polishing with stainless wire
wool, a fine polishing with copper wire wool, an ultrasonic bath using 2%
CITRANOX [18] in de-ionised water and a high-pressure wash using pure
water. The polishing phase was done using a powered rotation table specially
built for the purpose and polishing was always applied along the line of existing
machining marks.
The cooling system
Cooling is done using liquid nitrogen (LN2). The internal reservoir, located
under the target vessel, holds 36 litres. There are two thermal links between
this reservoir and the target vessel (see figure 9). The first link is a conduction
path provided by flexible bundles of thick copper wires thermally anchored to
a hollow copper cooling flange (‘heat exchange ring’) attached to the underside of the target vessel. The flexibility helps decouple acoustic/mechanical
noise in the LN2 reservoir from the LXe chamber. The other end of the bundle dips into the liquid nitrogen and the thermal impedance depends on the
depth of the liquid. The bundle is welded and polished at both ends for good
thermal matching. A second thermal path is provided by a direct connection
between the nitrogen reservoir and the hollow cooling flange. This allows liquid or boil-off gas to be used as additional coolant and provides the means
for active thermal control with minimum cryogen usage, which is important
during stand-alone operation underground.
LXe Vessel Base Flange
Heat Exchange Ring
N2 Gas Outlet Port
S.S. Hose
LN2 Vapour Brai d Outlet Port
LN2 Vapour Outlet Tube
LN2 Inlet Tube
Flexible Welded
Silver Plated
Cu w ire braid
LN2 Brai d 2 Inlet Tube
LN2 Brai d 1 Inlet Tube
LN2 Vessel
Vacuum Vessel
Base Flange
LN2 Inlet Port
Fig. 9. Thermal control system elements.
Three external pipes are connected to the cooling system. The first is the
liquid nitrogen delivery line and this terminates inside the reservoir close to
the top. A second pipe also opens to the top of the reservoir, whilst a third pipe
connects to the delivery to the cooling flange. The last two pipes are fitted with
control valves which regulate the internal pressure and the flow rate through
the hollow cooling flange. During initial cool-down the flow through the cooling
flange is increased to allow bulk liquid flow into it. Once cold, the gas flow
through the cooling flange provides a fine temperature control mechanism
whilst the copper cable bundles provide the main thermal link balancing the
average heat load. The heat load is reduced by the use of thermal insulation
around both the target vessel and the nitrogen chamber (see figure 10). The
nominal operating temperature is around −100o C and the heat load is, as
expected, ∼ 40 W, giving a design hold-time between refills of ∼2 days.
Fig. 10. The assembled instrument without its vacuum jacket giving a view of the
target vessel (top) and liquid nitrogen reservoir both covered with thermal insulation.
The outer vacuum jacket
The design principles for the vacuum jacket were much the same as for the
target vessel, except that the pressure rating was reduced to 4.3 bar absolute.
The safety standard for pressure vessels dictated the material thicknesses and
process standards and the same attention to background and cleanliness was
imposed. Hence OFHC copper was used, with electron beam welding and
minimisation of the number of seams; the cylindrical section of this larger
vessel was also made from just one rolled plate. The bottom flange has an
included domed section and the vacuum seals were all done in the same way
as for the target vessel. Figure 11 shows the underside view of the bottom
Fig. 11. The bottom flange of the vacuum jacket showing the inner domed section
and the arrangement of feedthroughs and ports around the outer skirt.
The gas purification system and safety reservoirs
Figure 12 shows a schematic of the xenon gas purification system. The main
requirement is to be able to remove electronegative contaminants which will
prevent the ionisation electron drift and suppress the secondary signal. Typically this requires liquid xenon purity down to the parts per billion level,
beyond that available through commercial purchase. In addition the level of
radioactive krypton needs to be kept as low as possible as the beta-decay
of 85 Kr gives a continuum energy deposit down into the level expected from
elastic scattering of WIMPs. An all-metal bakeable gas system has been used.
The system is pumped by a combination of oil-free scroll and turbo-molecular
pumps. The xenon gas is contained in two large stainless steel cylinders fitted
with high purity all-metal UHV valves and regulators. These two cylinders
stand in cooling jackets allowing them to be cooled to liquid nitrogen temperatures. Two SAES[19] getters are used. Fine particle filters (0.5 µm) are fitted
to all gas delivery lines. The gas system is fitted with a mass spectrometer
which is used both for helium leak testing and residual gas analysis. The base
vacuum attainable in the system is ∼ 10−8 torr, dominated by H2 ; a partial
H2 O pressure of ∼ 10−10 torr was achieved prior to the xenon input. The detector itself is connected without valves to a port on the main volume of the
gas purification system. Another port is connected to the large volume safety
reservoirs with only a burst disk between them. This is not only to guard
against the safety risk associated with catastrophic failure of the target vessel
under overpressure, but also to avoid loss of xenon. The two gas cylinders contain 50 kg of xenon supplied by ITEP from stock collected from underground
sources between 20 and 40 years ago. This xenon is expected to have a very
low radioactive krypton content. A final component of our gas purification
system is a novel portable chamber for electron lifetime measurements which
will be described elsewhere[20].
0 V16
F4 1
turbo T4
dump D2
turbo T3
turbo T2
6 bar
6 bar
6 bar
clean room
6 bar
dump D1
6 bar
6 bar
F3 1
14 bar
6 bar
6 bar
100 bar
purity monitor
bottle B2
getter G2
turbo T1
getter G1
bottle B1
bottle B3
UHV valves
cylinder valves
pressure regulators
bursting disks
particle filters
mass spectrometer head
getter valves
100 bar
270 bar
Fig. 12. Schematic diagram of the gas purification system.
The data acquisition system
The 31 PMT signals are fed into wideband amplifiers and split into a dual
dynamic range data acquisition system (DAQ). This ensures sensitivity to
very small primary scintillation signals containing only a few photoelectrons
(phe) as well as to large secondaries without saturation. All 62 channels are
sampled at 500 MS/s by 8-bit AQIRIS digitisers. For the collection of ‘dark
matter’ data a PMT gain of 2 × 105 will be used. Such a low gain should avoid
internal PMT saturation effects following very large secondary scintillation
signals. Wideband amplifiers add electronic gain in two stages. The first stage
is (x10) with a noise referred to the input of 30 µV rms. They then feed into
adjustable attenuators which are used to equalise the single photoelectron
response for each PMT. The outputs from this stage then feed into the 31
low-gain digitisers as well as into the next stage x10 wideband amplifiers. The
high and low-gain input channels thus have a factor of 10 gain difference which
can be further expanded by adjusting the full-scale ranges on the digitisers.
A simple threshold trigger signal is derived from a summing amplifier, with
inputs from all PMTs, fed into a discriminator whose output provides an
external trigger for the AQIRIS digitisers. This trigger can not differentiate
between primary and secondary scintillation signals. A more sophisticated
trigger using a time to amplitude converter can provide a width measure
and differentiate the very short primary scintillation signals (∼ 30 ns time
constant) from the much more extended secondary scintillation signals (∼ 1 µs
duration). The maximum delay between primary and secondary scintillation
signals in ZEPLIN III is ∼ 17 µs. A LINUX-based software application reads
out the digitiser crates. A FIFO-type memory buffer, accessed independently
by two CPUs for data transfer and write-out, reduces the overall dead time.
An acquisition rate of 100 events/s can be sustained.
Commissioning cool-down tests
The first cool-down test was designed to verify the thermal control system
and to test out the PMT array. For this test the anode and cathode electrodes
were replaced by a copper plate located just 8 mm above the PMT array.
31 241 Am radioactive sources were vacuum-sealed into this plate with a thin
copper foil overlay to prevent leakage of radioactivity and to stop α-particles
from interacting in the xenon. These then provided a source of low-energy
(mainly 59.6 keV) photons. For subsequent cool-down tests the radioactive
sources had been removed and the full electric field system installed in its
final configuration.
8.1 Cooling system
The cooling system performance during the first cool-down was as expected.
The initial cool-down period used 200 litres of liquid nitrogen and progressed
at ∼ 5 o C/hour. An array of temperature sensors was used to monitor critical
points within the instrument. One of these, on the lower face of the cooling
flange on the bottom of the target, is used as the control temperature and its
reading is compared with a set temperature in the controller to automatically
operate two valves: one which exhausts straight from the gas volume of the
nitrogen reservoir, and one which exhausts through the cooling ring. Once
down at the nominal operating temperature (∼ −100 o C) the temperature of
the target vessel is stable to better than 0.2o C and the liquid nitrogen usage
drops to ∼ 20 litres/day as expected. Figure 13 shows some key engineering
parameters monitored over a 24 hour period during the second cool-down test.
The upper trace is from the temperature sensor on the cooling flange and the
periodic behaviour is due to the control system. The lower trace is then the
temperature of the base plate of the target vessel itself.
heat exchanger
temperature (oC)
∆C (pF) or ∆P (bar/10)
baseplate temperature
time (hours)
Fig. 13. Some key system engineering parameters monitored over a 24 hour period.
8.2 PMT array
Pulse height spectra, pulse waveforms and single photoelectron spectra were
collected from all PMTs during the first cold-run both with the DAQ electronics just described and with a pulse height analysis (PHA) set-up using
a multichannel analyser (MCA). These confirmed correct operation of all 31
PMTs in the array, including ∼1000 crimped connections! A typical primary
scintillation pulse from a 59.6 keV photon interaction is shown in figure 14.
This shows the characteristic decay time of ∼ 40 ns. The single photoelectron
spectra show well resolved peaks and these were used to set the amplifier gains
in order to normalise all channels to the same overall gain.
voltage, V
time, ns
Fig. 14. A typical primary scintillation pulse from a 59.6 keV photon interaction
and a single photoelectron spectrum from one of the PMTs. Both the spectrum and
the scintillation pulse were obtained with +2kV on the PMT anode; however there
was an additional x10 amplifier present for the spe measurement.
8.3 Scintillator performance
LXe scintillates in the vacuum ultraviolet (VUV), near 175 nm, with a yield
comparable to the best scintillator crystals. The VUV luminescence is produced by the decay of singlet and triplet states of the Xe∗2 excimer. These can
be formed directly by excited atoms left by the interacting particle or as a
result of recombination into an excited state along the particle track [21,22].
Figure 15 shows a typical spectrum taken from one PMT when the internal
Am sources were in place. The plot shows two spectra. The bottom spec19
trum was taken with the whole arrangement covered with liquid xenon. The
two spectral features are the 59.6 keV line from 241 Am and a blend of the
26.3 keV 241 Am γ-ray with a 30 keV line resulting from escape of Xe K-shell
fluorescence photons. Using the measured single photoelectron spectrum from
this PMT gives a signal level of ∼12 phe/keV. For this measurement there is
no applied electric field. The top spectrum was taken with the liquid level between the source and the PMT window. The interaction occured in the liquid
phase and the improved light collection (up to ∼ 17 phe/keV is a result of
total internal reflection at the liquid gas interface due to the refractive index
mismatch. The resolution from the two-phase spectrum is ∼ 13% FWHM.
More detailed and extensive physics results from the first cold run will be
published separately[23].
Fig. 15. Pulse height spectra obtained from the 241 Am primary scintillation signals
during the first cool-down test. The two panels correspond to two different liquid
xenon levels. The highest energy peaks in both spectra are at 59.6 keV.
8.4 Two-phase operation
Once the radioactive sources used for the measurements in the previous section had been removed the second and subsequent cold-runs have successfully
loaded the detector with liquid xenon. A capacitive level-sensing system probes
the liquid xenon height with sub-mm accuracy at three locations in the chamber. A signal from one of these coaxial capacitor structures, readout to 0.03 pF
accuracy, is shown in figure 13. In underground operation, these sensors will be
integrated with an active levelling system in order to maintain the electrodes
parallel to the liquid surface, guarding the heavily shielded detector against
any structural deformation of the underground cavern.
With the xenon filled to its nominal depth, but with no applied electric field,
figure 16 shows 57 Co spectra obtained with an external uncollimated source
placed above the detector. Two spectra are shown, both reconstructed using
the outputs from all the PMTs. The shaded one, however, only includes events
in which the peak signal occurred in one of the inner seven PMTs. This ‘collimated’ spectrum has a FWHM of ∼ 25%, which we expect will improve once
final corrections for PMT sensitivities are done. The broad shoulder on the low
side for the uncollimated spectrum is purely due to light collection variation
towards the edge of the xenon volume (way outside the fiducial volume).
Fig. 16. Pulse height spectra obtained from using an external 57 Co source above the
After the zero-field tests, 13.5 kV were applied between the cathode grid and
the anode mirror, setting up a field of 3.0 kV/cm in the liquid. Figure 17 shows
a typical signal from a γ−ray interaction in the liquid. The fast scintillation is
the primary signal, S1, caused by direct excitation created by the photoelectron. The second, broader signal, S2, occurs when the ionisation also created
has drifted to the liquid surface and has been extracted into the gas phase.
Once in the gas phase the electric field is strong enough to cause excitation
leading to a burst of additional photons. The time delay depends on the depth
at which the interaction happened and the drift velocity at our operating fields
is ∼2.5 mm/µs. The width of the secondary depends on the gas gap and the
electric-field in the gas. The secondary emission shows a flat plateau as the
charges drift across the gap. The rise and fall time is due to a combination
of extraction dynamics, diffusion and the gas scintillation time-constant. The
ratio of the two areas (S2/S1) is ∼150 as expected at this field.
One of the key design drivers of ZEPLIN-III was the ability to resolve each
interaction point in the three dimensions. A position reconstruction algorithm
was developed from simulated datasets which will provide sub-cm resolution
in the horizontal plane [24]. Even before this is applied to real data, this
spatial sensitivity is well demonstrated in figures 18 and 19, showing an event
in which two interactions have overlapped in time. Moreover there are at least
four secondary signals. Without position sensitivity it would not be possible
to separate these two events just from the summed signals. However, looking
at the individual PMT traces (left-hand panel in figure 19) it is immediately
obvious that these two events have happened in very different parts of the
detector (right-hand panel) and they can be unambiguously separated. They
are both double-Compton scatters.
amplitude, V
time, ns
Fig. 17. Summed waveforms from a γ-ray event showing a fast primary pulse followed
by the secondary wider pulse from electroluminescence in the gas phase caused by
ionisation drifted from the event site. The two traces shown are from the dual range
DAQ. The low-sensitivity data have been multiplied by 10.
The key design features of the ZEPLIN III instrument have been described.
The challenging and pioneering aspects of the manufacturing technologies and
amplitude, V
time, ns
Fig. 18. Summed waveforms from two overlapping γ-ray events both showing fast
primary signals followed by secondary signals.
Fig. 19. Individual waveforms from the 31 PMTs showing how the two overlapping
events can be separated using position reconstruction. The reconstructed positions
are indicated on the right-hand panel and the size of the symbols is representative
of the position resolution.
procedures have been detailed and first commissioning data have been presented to demonstrate the successful completion of build of this instrument.
Further extensive calibration work is now needed to prepare this instrument
for use as a dark matter detector. To fully charaterise ZEPLIN III much of
this will need to be done underground in a lower background environment.
This work has been funded by the UK Particle Physics And Astronomy Research Council (PPARC). We would like to acknowledge the superb copper
machining achieved within the Imperial College Phyiscs Department workshop led by R. Swain, and the development of new welding techniques by The
Welding Institute.
[1] T. J. Sumner (2005), New Astronomy Reviews 49, 277–281.
[2] H. M. Araújo (2005), Proc. Int. Conf. Dielectric Liquids, Coimbra, Portugal,
IEEE Press (IEEE Cat. 05CH37643), 305.
[3] G. J. Alner et al. (2005), New Astronomy Reviews 49, 259–263.
[4] G. J. Alner et al. (2005), Nucl. Instrum. Meth. A555, 173–183.
[5] B. A. Dolgoshein, V. N. Lebedenko & B. U. Rodionov (1970), JETP Lett. 11,
[6] T. J. Sumner et al. (1999), Proc. 26th Int. Cosmic Ray Conf., 2, ed. D. Kieda,
M. Salamon & B. Dingus, 516.
[7] A. S. Howard et al. (2001), Proc. 3rd Int. Workshop on the Identification of Dark
Matter, ed. N. J. C. Spooner & V. Kudryavtsev, Singapore: World Scientific,
[8] D. Yu Akimov et al. (2003), Proc. 4th Int. Workshop on the Identification
of Dark Matter, ed. N. J. C. Spooner & V. Kudryavtsev, Singapore: World
Scientific, 371.
[9] D. Davidge (2003), PhD Thesis, University of London.
[10] J. Dawson (2003), PhD Thesis, University of London.
[11] H. M. Araújo et al. (2006), Astropart. Phys. accepted. (arXiv:astro-ph/0603243)
[12] E. Aprile et al. (2006), (ArXiv: astro-ph/0601552)
[13] H. M. Araújo et al. (2004), Nucl. Instrum. Meth. A521, 407–415.
[14] N. J. T. Smith et al., in preparation.
[15] ANSYS Inc. (www.ansys.com)
[16] CRC Handbook of Chemistry & Physics, pp. 12-120
[17] http://www.twi.co.uk/
[18] http://www.alconox.com
[19] http://www.saesgetters.com/
[20] R. Walker et al. in preparation
[21] S. Kubota et al. (1979), Phys. Rev. B 20(8), 3486–96.
[22] A. Hitachi et al. (1979), Phys. Rev. B 27(9), 5279–85.
[23] V. Chepel, C. Thorne et al. (2006), in preparation.
[24] A. Lindote et al. (2005), Nucl. Instrum. Meth. A, in press.
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF