The T2K ND280 Off-Axis Pi-Zero Detector
S. Assylbekovb, G. Barrd, B.E. Bergerb, H. Bernsh, D. Beznoskoc, A. Bodekf, R. Bradfordf,1, N. Buchananb, H. Buddf, Y. Caffarib, K. Connollyh, I. Dankoe, R. Dasb, S. Davish, M. Dayf,
S. Dytmane, M. Dziombah, R. Flightf, D. Forbushh, K. Giljec, D. Hansene, J. Hignightc, J. Imberc, R.A. Johnsona, C.K. Jungc, V. Kravtsovb, P.T. Lec, G.D. Lopezc, C.J. Malafisc, S. Manlyf, A.D. Marinoa,∗, K.S. McFarlandf, C. McGrewc, C. Metelkog, G. Nagashimac, D. Naplese, T.C. Nichollsg, B. Nielsenc, V. Paolonee, P. Paulc, G.F. Pearceg, W. Qiang, K. Ramosc, E. Reinherz-Aronisb, P.A. Rodriguesf, D. Ruterboriesb, J. Schmidtc, J. Schwehrb,
M. Siyadg, J. Steffensc, A.S. Tadepallic, I.J. Taylorc, M. Thorpeg, W. Tokib,c, C. Vaneka, D. Warnerb, A. Weberd,g, R.J. Wilkesh, R.J. Wilsonb, C. Yanagisawac,2, T. Yuana
aUniversity of Colorado at Boulder, Department of Physics, Boulder, Colorado, U.S.A.
bColorado State University, Department of Physics, Fort Collins, Colorado, U.S.A.
cState University of New York at Stony Brook, Department of Physics and Astronomy, Stony Brook, New York, U.S.A.
dOxford University, Department of Physics, Oxford, United Kingdom
eUniversity of Pittsburgh, Department of Physics and Astronomy, Pittsburgh, Pennsylvania, U.S.A.
fUniversity of Rochester, Department of Physics and Astronomy, Rochester, New York, U.S.A.
gSTFC, Rutherford Appleton Laboratory, Harwell Oxford, United Kingdom
hUniversity of Washington, Department of Physics, Seattle, Washington, U.S.A.
Abstract
The Pi-Zero detector (PØD) is one of the subdetectors that makes up the off-axis near detector for the Tokai-to-Kamioka (T2K) long baseline neutrino experiment. The primary goal for the PØD is to measure the relevant cross sections for neutrino interactions that generateπ0’s, especially the cross section for neutral currentπ0 interactions, which are one of the dominant sources of background to theνµ→νeappearance signal in T2K. The PØD is composed of layers of plastic scintillator alternating with water bags and brass sheets or lead sheets and is one of the first detectors to use Multi-Pixel Photon Counters (MPPCs) on a large scale.
Keywords: Neutrinos, Neutrino Oscillation, Long Baseline, T2K, J-PARC, Pi zero Detector
∗Corresponding author
Email address:[email protected](A.D. Marino)
1Presently at Argonne National Laboratory, Argonne, Illinois, U.S.A
2Also at BMCC/CUNY, New York, New York, U.S.A.
Preprint submitted to Nuclear Instruments and Methods in Physics March 19, 2012
1. Introduction
1
The Tokai-to-Kamioka (T2K) experiment is a long-baseline neutrino oscillation experiment
2
designed to probe the mixing of the muon neutrino with other neutrino species and to shed light
3
on the neutrino mass scale. The T2K neutrino beam is generated using the the new high-intensity
4
proton synchrotron at J-PARC, which has a Phase-I design beam power of 0.75 MW. T2K uses
5
Super-Kamiokande [1] as the far detector to measure neutrino rates at a distance of 295 km from
6
the beam production point, and near detectors to sample the unoscillated beam. The neutrino
7
beam is directed 2.5◦away from the Super-Kamiokande detector and travels through the Earth’s
8
crust under Japan, as illustrated in Fig. 1.
9
295 km
280 m J-PARC Near Detector Super-
Kamiokande
1000 m
Neutrino Beam
Figure 1: A schematic showing the path of the neutrinos in the T2K experiment, from the start of the neutrino beamline at J-PARC to Super-Kamiokande, 295 km away.
The T2K experiment [2] near detector complex (ND280), located 280 m from the start of the
10
neutrino beam, contains the on-axis INGRID detector and an off-axis detector.
11
The off-axis detector, shown in Fig. 2, is situated at the same off-axis angle as Super-Kamiokande
12
and contains the Pi-Zero detector (PØD) a plastic scintillator-based detector optimized forπ0de-
13
tection followed by a tracking detector comprising two fine grained scintillator detector modules
14
(FGDs) sandwiched between three time projection chambers (TPCs). The PØD and tracker are
15
surrounded by electromagnetic calorimeters (ECALs), including a module that sits immediately
16
downstream of the tracker. The whole detector is located in a magnet with a 0.2 T magnetic
17
field, which also serves as mass for a side muon range detector (SMRD). This paper describes
18
the PØD subdetector in greater detail.
19
1.1. Goals of the PØD
20
The primary physics goal of T2K is to measure the mixing angleθ13or to improve the existing
21
limit by an order of magnitude if the angle is too small to measure directly. This is done by
22
2
Figure 2: An exploded view of the off-axis near detector.
looking for the appearance ofνein aνµbeam. Additional physics goals include the precision
23
determination of the∆m223andθ23 parameters through aνµdisappearance measurement, where
24
the parameters will be measured to a precision ofδ(∆m223) ∼10−4eV2 andδ(sin22θ23)∼0.01
25
respectively. In addition to neutrino oscillation studies, the T2K neutrino beam (a narrow-band
26
beam with a peak energy of about 600 MeV) will enable a rich physics program of neutrino
27
interaction studies at energies covering the transition between the quasi-elastic and resonance
28
production regimes.
29
To achieve the required precision for theνeappearance measurement (observed via the pro-
30
cessνe+n → e−+p), the neutral currentπ0 rate (νµ+N → νµ+N+π0+X) must be mea-
31
sured at the J-PARC site near the neutrino beam production point using the off-axis near detector.
32
Events containingπ0’s are the dominant physics background to theνeappearance signal at Super-
33
Kamiokande. The PØD sits at the upstream end of the off-axis detector and has been designed
34
to precisely measure the neutral current processνµ+N → νµ+N+π0+X on a water (H2O)
35
target. In addition the PØD will constrain the intrinsicνecontent of the beam flux which is an
36
irreducible background to theνeappearance measurement.
37
Early design studies demonstrated that understanding theπ0 andνe backgrounds required
38
sensitivity to interactions containingπ0with momentum greater than 200 MeV/c. This requires
39
a photon reconstruction threshold of well below 100 MeV. Both of the background processes to
40
3
be constrained by the PØD are a relatively small fraction of the total PØD interaction rate, and
41
must be measured on a water target, forcing a large water mass. In addition, sufficient energy
42
resolution is needed to demonstrate the presence of aπ0 through reconstruction the invariant
43
mass. The eventual design of the PØD realizes these goals by interleaving water target between
44
scintillator layers which both measure charged particles and support the water target. The rate on
45
water is determined using statistical subtraction with data collected during periods having water
46
in the detector and out of the detector.
47
1.2. Description of the PØD
48
The main features of the PØD design are shown in Fig. 3. The electronics supports and
49
detector mounting system are visible surrounding the active regions of the detector. In addition
50
the regions of the detector are also labeled. Figure 4 shows a schematic of the active regions of the
51
PØD. The central region, composed of the ”upstream water target” and ”central water target,”
52
is made from alternating scintillator planes, water bags, and brass sheets. The front and rear
53
sections, the “upstream ECal” and “central ECal” respectively, use alternating scintillator planes
54
and lead sheets. This layout provides effective containment of electromagnetic showers and a
55
veto region before and after the water target region to provide rejection of particle interactions
56
that enter from outside the PØD.
57
Figure 3: 3D drawing of the roughly 2.5 m cube PØD outside of the basket. Downstream face of detector shown. See Section 5 for a description of the TFB and RMM electronics.
4
Upstream ECal
Upstream Water Target
Central Water Target Central ECal
Legend
Lead Light-tight Cover
Brass Water Scintillator Wavelength-shifting Fiber
Figure 4: A schematic of the four PØD Super-PØDules as installed in the detector. Beam direction: left to right.
There are a total of 40 scintillator modules in the PØD. Each PØD module, or PØDule, has
58
two perpendicular arrays of triangular scintillator bars, forming a plane. There are 134 horizontal
59
bars (2133 mm long) and 126 vertical bars (2272 mm long) in each PØDule. Each bar has a
60
single coaxial hole through which is threaded a wavelength-shifting (WLS) fiber. Each fiber has
61
a mirrored coating applied on one end while the other end is optically coupled to a Hamamatsu
62
Multi-Pixel Photon Counter (MPPC) [3] for readout, as shown in Figure 5. Each photodetector
63
is read out with Trip-t Front-end electronics (Section 5). There are a total of 10,400 channels for
64
the entire PØD.
65
The PØDules were assembled into four units called Super-PØDules. The two ECal Super-
66
PØDules each consist of a sandwich of seven PØDules alternating with seven stainless steel-clad
67
lead sheets (4.5 mm thick). The water target is formed from two units, the upstream and central
68
water target Super-PØDules. The upstream (central) water target Super-PØDule is comprised
69
of 13 PØDules alternating with 13 (12) water bag layers (each of which is 28 mm thick), and
70
13 (12) brass sheets (1.28 mm thick), as shown in Fig. 6. The dimensions of the entire PØD
71
active target are 2103 mm×2239 mm×2400 mm (width×height×length) and the mass of
72
the detector with and without water is 15,800 kg and 12,900 kg respectively. The PØD is housed
73
5
Scintillator Bar
Fiber
Ferrule
MPPC Signal Wires
External Shell PVC Frame
Figure 5: A close-up view of the edge of a PØDule showing how the WLS fibers exit the scintillator bars and couple to the MPPCs. The optical connectors will be described on more detail in Section 2.2.5.
inside a detector basket structure that supports the central off-axis detectors inside the magnet.
74
Figure 6: Expanded view of water target PØDule, brass radiator and water bladder containment frame.
The remainder of this paper describes in detail the design, fabrication, and performance of the
75
PØD. The production of the scintillator bars and their assembly into planks and PØDules will be
76
presented followed by a description of how the individual PØDules were combined into the four
77
Super-PØDules, and are read out using photosensors. The detector component performance,
78
starting with scans of the PØDules using a radioactive source, dark noise measurements, and
79
tests with the light injection system, is presented. The paper concludes with a description of the
80
calibration and performance of the full detector.
81
6
2. Design and Construction of the PØDule
82
The PØDule is the basic structural element of the PØD active region, and is constructed
83
of scintillator bars sandwiched between sheets of high-density polyethelene (HDPE, thickness
84
6.4 mm). The entire structure is surrounded by PVC frames that support the PØDule as well as
85
providing mounts for the required services such as the MPPC light sensors, and the light injection
86
system.
87
The polystyrene triangular scintillating bars that make up the PØDules were fabricated by
88
co-extruding polystyrene with a reflective layer of TiO2 and a central hole for the WLS fiber.
89
The light seal for the tracking plane is maintained by light manifolds that collect the WLS fibers
90
into optical connectors. These manifolds also provide access to the fibers for the light injection
91
system. Because of the large number of scintillating bars and the available space limitations, it
92
was impractical to route the fibers outside the magnetic volume therefore the Hamamatsu MPPC
93
photosensors, which are immune to the magnetic field, were attached directly to each WLS fiber
94
just outside the PVC PØDule frame, as shown in Figure 5.
95
2.1. Design of the PØDule
96
The PØDule was designed to both provide the active tracking region and to serve as a struc-
97
tural element. This was achieved using a laminated structure of crossed scintillator bars between
98
polystyrene skins. The final PØDule has been shown to have a rigidity similar to a solid mass
99
of polystyrene of similar thickness. The edge of the central scintillator and skin structure of the
100
PØDule is surrounded by a machined PVC frame. Each PØDule is instrumented on one side
101
(both y and x layers) with MPPCs and on the other a UV LED light injection system. The bottom
102
PVC frame supports the weight of the PØDule within the ND280 detector basket. The frames
103
also provide the fixed points needed to assemble the PØDule into the four Super-PØDules via
104
two precision holes located in the four corners of each PØDule as well as a set of seven holes
105
spaced along each side through which tensioning rods were passed.
106
The PØDules, after installation into the finished PØD, are oriented such that the most up-
107
stream layer of scintillator has the bars oriented approximately along the vertical axis while the
108
downstream layer has its bars oriented along the horizontal axis. This arrangement results in a
109
local coordinate system defined such that the x, y and z axes are approximately congruent with
110
the global coordinate system where x is horizontal, y is vertical, and z points downstream toward
111
7
Super-Kamiokande. The external dimensions of the PØDule are 2212 mm (x) by 2348 mm (y)
112
by 38.75 mm (z).
113
To facilitate assembly of the PØDule (described in Section 2.3), all of the components were
114
prefabricated with holes that allowed alignment during assembly. The assembly tolerance was
115
less than 0.5 mm on all internal dimensions, and less than 1 mm on the thickness. The rela-
116
tive dimensions of the PØDules were maintained using precisely located holes in the PØDule
117
assembly table.
118
2.2. Assembly of PØDule Components
119
The construction and assembly of the PØDule components was distributed across several
120
institutions. This allowed a supply chain that could produce the required components in parallel
121
and optimized use of facilities, local expertise and available personnel.
122
2.2.1. PØDule Scintillator Preparation
123
The polystyrene scintillator for the PØDule was manufactured in the extrusion facility at
124
Fermi National Accelerator Laboratory [4] using an extrusion die and process originally devel-
125
oped for the inner detector of the MINERνA experiment [5, 6]. The blue-light emitting scintil-
126
lator base material was Dow Styron 663 W doped with 1% PPO and 0.03% POPOP to shift the
127
UV scintillation light emitted by the styrene into the blue. The bars are triangular in cross section
128
with nominal dimensions of 17±0.5 mm height and 33±0.5 mm width. Each bar also had a
129
nominal 2.6 mm diameter hole centered at 8.5±0.25 mm above the widest part of the triangle
130
for fiber insertion. To reflect the produced light and therefore maximize the possibility of capture
131
by the wavelength shifting fiber in the center hole, a thin, 0.25 mm on average, layer of styrene
132
mixed with 25% TiO2was coextruded on the outside of the bar, and both ends of the scintillator
133
bar were painted with white EJ-510 TiO2Eljen paint.
134
During production, physical characteristics of the scintillator were monitored by taking fre-
135
quent samples and measuring their outer dimensions, the location and dimensions of the center
136
hole, and the thickness and coverage of the coextrusion. At approximately twenty different
137
equally spaced times during production, samples were also taken and used to characterize light
138
output using a radioactive source counting setup with a reference piece of scintillator from the
139
MINERνA production. Physical dimensions were held well within the tolerance, and no ev-
140
idence was observed for variation in light output beyond the uncertainties in the monitoring
141
8
measurement, roughly 5%.
142
2.2.2. PØDule Plank Assembly
143
The extruded scintillator bars were bundled into manageable sized “planks” to be used in the
144
assembly of the full-sized PØDules. There were two sizes of planks for each of the bar lengths
145
and a special jig was constructed for each of the four plank types.
146
Each of the triangular scintillating bars was prepared for the plank assembly and subjected
147
to quality assurance (QA) procedures prior to assembly into a plank. As each bar was unpacked
148
it was inspected for signs of visible damage, such as nicks or cuts in the TiO2coating and any
149
damaged bars were removed from plank production. Once a bar passed the visual inspection it
150
was cut to length using a jig to ensure proper length. A mounted pneumatic drill was used to
151
bore out the ends of the holes running down the center of each bar. A long stiffwire was passed
152
through each bar to ensure that no debris was lodged in the hole that would prevent insertion of
153
the WLS fiber. An additional check was made to ensure that the hole for the fiber was centered
154
on the end of the triangular bar.
155
Four separate jigs were set up on two optical tables for the gluing of the bars into planks.
156
The short bars were made into two types of planks, one type containing 16 bars and one type
157
containing 17 bars. The long bars were made into 15 bar and 17 bar planks. Prior to application
158
of the epoxy to the bars, the bars were placed in the jig and a heavy straight-edge was used to
159
ensure that the thickness of bars were within the 0.25 mm tolerance of the nominal 17.25 mm
160
plank thickness. A log document, or traveler, was kept with each plank during the entire assembly
161
process. The traveler contained details such as the plank serial number, the identification number
162
of all bars contained in the plank, and any measurements made on the plank during assembly and
163
quality assurance.
164
Once the bars had been test fitted into the plank gluing jig, they were removed and epoxy
165
was applied to each using an automated gluing machine (Fig. 7) that mixed the two epoxy parts
166
and applied a steady stream of glue to two sides of the bar. The epoxied bars were placed back
167
into the gluing jig and a vacuum sealed frame was used to apply pressure to the plank for about 2
168
hours while the epoxy set. A final QA inspection was made to ensure that the planks were within
169
the thickness tolerance.
170
9
Figure 7: Automated glue machine used to apply epoxy to the triangular bars before they were placed into the plank gluing jig.
2.2.3. PØDule WLS Fiber Preparation
171
The wavelength shifting fibers that are inserted into the holes in the scintillator bars are
172
Kuraray multi-clad, S-35, J-type, doped with Y-11 (175 ppm) with a diameter specification of
173
1.00+0.02−0.03 mm. The fibers were placed into the holes in the scintillator, but were not glued into
174
place. Studies done for the Minerνa experiment indicated that the light yield for glued fibers was
175
approximately 2 times greater than the light yield for unglued fibers [7], but unglued fibers are
176
considerably easier to install. The same study also showed that the light yield did not strongly
177
depend on the fiber-to-hole diameter ratio over the range of 0.3 to 0.9 [7], so the decision to use
178
a 1.0 mm fiber in a 2.6 mm hole does not have a large impact on the light collection.
179
The WLS fiber was delivered in unspooled “canes” pre-cut to a rough length 67 mm longer
180
than the bar length in order to avoid memory effects of spooled fiber.
181
The first step in processing the delivered fiber was to mirror one end. This work was per-
182
formed in the Thin Film Coatings facility in Lab 7 at Fermi National Accelerator Laboratory. One
183
end of the fiber was first “ice polished” with the ice providing mechanical support for a group of
184
approximately 800 fibers polished with a diamond polisher in a single batch. The polished end
185
was then coated with aluminum using a sputtering vacuum deposition process. After completion
186
of the mirroring, each fiber was coated with a thin layer of epoxy to protect the mirror.
187
10
The reflectivity of three fibers from each batch of 800 was determined by measuring the light
188
output of a fiber with the mirror end placed into a piece of scintillator with an attached radioactive
189
source, and then remeasuring the light output after cutting offthe mirror with a 45◦ cleave and
190
painting the cleft end with black paint. For individual fibers, an average reflectivity of 86% was
191
measured, with a root mean square of the ensemble of measurements of 6%.
192
After mirroring, the fiber was glued into one end of a ferrule (see Section 2.2.5 that was
193
injection molded from Vectra A430, a Teflon-filled liquid-crystal polymer. The length between
194
the ferrule end and the fiber mirror end was required to be kept at 1 mm tolerance over the
195
several meter length of the fiber. The ferrule was designed to mount into a housing that contained
196
the MPPC and kept the fiber end in contact with the pliable optical layer covering the MPPC.
197
Because the Teflon in the Vectra plastic clogged the diamonds in the fiber polisher, the fiber was
198
glued into the ferrule a few mm proud of the end of the ferrule, a 1 mm thick layer of optical
199
epoxy was deposited for mechanical support, and then the epoxy with the embedded fiber was
200
polished. The resulting finish was inspected with a microscope to establish when diamond wear
201
adversely affected the finish. Typically, 1500 to 2000 fibers could be polished with a single
202
diamond.
203
2.2.4. MPPC Acceptance Testing
204
The MPPCs used by the PØD, as well as by the ECAL, SMRD, and INGRID detectors,
205
are solid-state photosensors manufactured by Hamamatsu Photonics. The active sensor for the
206
MPPC is a 1.3 mm×1.3 mm array of 50-micron pixels, totaling 667 pixels (including a small
207
inactive area for electrical contact). Each pixel operates in Geiger mode, producing a well-
208
defined pulse when a photoelectron generates a cascade. The MPPC output is the sum of all
209
pixel outputs, the pixels having very similar responses. As a result, the output spectrum from
210
the MPPC shows well-separated peaks corresponding to the number of pixels fired, which at low
211
light levels is a good measure of the number of photoelectrons.
212
A photosensor quality-control (QC) testing program was performed on all 10,400 MPPCs
213
installed in the PØD, as well as 1,100 spares. The goals of the testing were to measure the
214
operational characteristics of the MPPCs, to verify that their performance was acceptable, and to
215
set the initial operating voltages to be used in the PØD. In particular, the gain of each device was
216
measured as a function of bias voltage. A linear fit was used to extract the breakdown voltage –
217
the minimum bias voltage to produce a Geiger cascade. The MPPC characteristics were found to
218
11
be largely consistent from device to device as a function of overvoltage (the bias voltage above
219
the breakdown voltage), but there is a significant variation in breakdown voltages from device
220
to device, particularly between devices originating in different batches. All tests were done at a
221
controlled temperature of 20◦C, controlled to better than 0.2◦C, as the MPPC breakdown voltage
222
depends on the operating temperature.
223
The testing protocol required the measurement of a number of different MPPC operating
224
parameters. First, a scan of gain vs bias voltage was performed to measure the breakdown voltage
225
and to establish the operating voltage. Measurements were made over a 2 V wide overvoltage
226
range around the predicted gain range of 5×105 and 7.5×105. Dark noise rates and relative
227
detection efficiencies at these gains were also measured.
228
Physics signals in the PØD range from a few to hundreds of photoelectrons. Eight light levels
229
across this range were used to characterize the photosensors, with measurements made at four
230
different bias voltages for each light level.
231
Production testing of the photosensors began in late September 2008 and was completed by
232
January 2009. Photosensors passing all QC tests were shipped for installation into the completed
233
PØDules. Only 230 out of 11,500 photosensors were rejected by the quality control procedure,
234
despite stringent acceptance criteria. Of those rejected, 74 were broken during assembly, 76 were
235
rejected due to dirt on the photosensor surface. Only 80 were rejected for abnormal behavior in
236
the photosensor testing data.
237
A dedicated NIM paper describing the photosensor quality-control testing and characteriza-
238
tion procedure in more detail can be found at [8]. A paper describing the PØD MPPC testing in
239
greater detail is under preparation.
240
2.2.5. Optical Connectors
241
Custom optical connectors were fabricated to provide optical fiber alignment to the MPPC
242
active area and to reduce the signal rate due to light contamination from external sources to ac-
243
ceptable levels to well below the intrinsic dark noise of the MPPCs. The connector system, shown
244
in Fig. 8 consists of three injection-molded components: a fiber-alignment tube or “ferrule”, a
245
housing that holds the MPPC and provides alignment to the fiber within 150µm, and an external
246
shell to provide mechanical protection and lock the connector on the ferrule. The material for all
247
three molded components is Vectra®A130, a 30% glass fiber loaded liquid-crystal polymer with
248
very low shrinkage, excellent dimensional stability, and very good mechanical properties.
249
12
Ferrule
MPPC Foam Disk
PCB External Shell
Figure 8: Optical connector: exploded (left) and assembled (right).
The MPPC is held in place against the fiber by a 3 mm thick closed-cell polyethylene foam
250
disk, acting as a spring. To ensure good optical contact between fiber and MPPC, the fiber is
251
allowed to protrude from the ferrule end by 0.5 mm after polishing, so the entire compression
252
force from the foam spring is applied between the fiber end and MPPC face. Electrical con-
253
nection between the MPPC and the front-end electronics is provided via a small circular printed
254
circuit board (PCB), with spring-loaded pin sockets making contact to the MPPC leads and a
255
Hirose Electric Co. micro-coax connector.
256
Connectors of this type are used in ND280 for the P0D, ECAL and INGRID sub-detectors.
257
Approximately 40,000 total connectors were produced (including spares).
258
2.3. Assembly of the PØDule
259
The PØDule assembly was one of the main fabrication stages to produce the forty PØDules
260
(plus 10% spares). This step involved gluing the main mechanical components: the pre-glued
261
scintillators planks, the four outer PVC frames, and the two outer HDPE plastic skins. A key
262
requirement for this process was to keep within the tolerances of the thickness of the PØDule (28
263
mm) and to have the alignment of the scintillator bars match the wavelength shifting fiber holes
264
in the scintillators with the holes in the PVC frames.
265
PØDule construction was performed on a specially designed gluing table. This flat table
266
(∼2.5×2.5 m) had precision alignment holes to keep the outer four PVC frames in the same
267
location for all the PØDules. The first assembly step was the placement of the HDPE plastic
268
sheet or skin on the gluing table, followed by painting HYSOL epoxy glue with paint roller
269
brushes over the entire top side of the sheet. Next, the four outer PVC frames were placed on
270
the edges of the glued sheet, and the position of the PVC frames were fixed by steel pins pressed
271
13
into the holes in the table. Next, the scintillator x-layer planks were lowered onto the bottom
272
HDPE sheet. They were aligned with respect to the PVC frame holes, then steel alignment pins
273
were pressed through the frame holes and into the holes in the scintillator. After all the x-planks
274
were pinned into place, epoxy glue was painted over the upper surface of the x-plank scintillator.
275
Then the y-planks were lowered onto the epoxied x-planks, and again the y-plank positions were
276
aligned with steel alignment pins. Next the upper HDPE outer sheet or “skin” was lowered on to
277
the glue-painted y-plank surface.
278
After all the gluing was completed, a vacuum sheet was lowered onto the PØDule and vacuum
279
was applied so the PØDule was uniformly compressed on all sides with about 0.5 atmosphere
280
of pressure. The PØDule was left for about 12 hours to cure overnight. After the epoxy had
281
cured, the vacuum sheet and the pins were removed and the PØDule was loaded with 260 WLS
282
fibers that were each attached to the snap-on optical connector containing an MPPC. Finally the
283
260 MPPCs were connected via mini-coax cables to the scanner readout electronic boards. The
284
PØDule was scanned overnight with a60Co source and read out as described in Section 6.2.
285
During the peak of the construction phase, five PØDules were produced per week. After the
286
PØDules were constructed and successfully scanned, they were stacked together vertically into
287
the four Super-PØDules.
288
2.4. The PØDule Light Injection System
289
The purpose of the PØD Light Injection System (LIS) is to provide monitoring for all 10,400
290
channels in the PØD. The LIS is capable of exposing the MPPC photosensors to light intensities
291
covering a range of more than two orders of magnitude with flash-to-flash intensity stability of
292
less than 2% [9]. This allows for monitoring of the photosensor response over the full range of
293
energy deposition expected for neutrino interactions in the PØD.
294
The main design challenge of the LIS was the geometrical constraint that it be embedded
295
within the 3 cm×4 cm×220 cm PØDule layer support frame for each of the 80 scintillator
296
planes. Each PØDule support frame has a cavity that allows 5 mm segments of WLS fibers to
297
be exposed to light produced by a pair of LEDs (Fig. 9). The LIS uses 80 pairs of fast-pulsed
298
400 nm UV LEDs as light sources; each pair illuminates a cavity in a single PØDule support
299
frame. LEDs exhibit minimal pulse-to-pulse fluctuations in intensity, so the temporal response
300
of an MPPC photosensor is dominated by photoelectron statistics.
301
14
The LIS provides the capacity to vary the LED light intensity over the required dynamic
302
range through control of both the height and the width of the pulses, by varying the current pulse
303
applied to the LED. The LEDs are driven by electronics originally designed for the MINOS
304
experiment LIS system [9]. The LIS electronics consists of four pulser boxes, a control box, a
305
distribution box and a power supply. The pulser boxes are mounted on the bottom and the north
306
sides on the PØD in close proximity to the LEDs. The control box is situated under the PØD ,
307
mounted on the detector basket. The distribution box and the power supply are located in a rack
308
about 10 meters from the PØD. Each pulser box contains two LED driver boards, a controller
309
board, and an LVDS to TTL converter board. Each driver board has ten channels that can be
310
programmed to pulse simultaneously in group of of 5, 10, or 20 channels. In contrast to the
311
MINOS light injection system, each channel is connected to a pair of LEDs by a 60 cm long
312
shielded cable. Communication between a PIC16F877 microprocessor located on the controller
313
board and a control computer is handled via ASCII commands over a serial RS232 link. Signals
314
from the control computer are carried over an Ethernet link and converted for the pulser boxes
315
by an Ethernet-RS232 converter.
316
During normal operation the PØD LIS control computer instructs the system to pulse a spe-
317
cific combination of LEDs at a specific height, width, and frequency to monitor the temporal
318
behavior of the PØD and associated readout. Dedicated LIS runs are taken periodically to allow
319
more precise measurements of properties, such as the relative timing between the PØD channels.
320
Figure 9: Cross-section view of the LIS cavity. Not shown are transparent strips with an opaque band to shadow nearby fibers.
15
3. Water Targets
321
3.1. Water Target Design
322
The water target modules were designed to provide layers of water approximately 3 cm thick
323
in the beam direction, with transverse dimensions of approximately 2 m×2 m. In order to
324
provide such large but thin layers without excessive container mass, high-density polyethylene
325
(HDPE) bladders were fabricated using materials and techniques derived from experience with
326
the Pierre Auger Observatory Water Cherenkov detector liners [10]. Adjacent scintillator bar lay-
327
ers provided the structural support required, with water pressure transmitted through successive
328
PØDules to end plates attached to the ND280 basket frame.
329
Since the water bladders are not rigid, it is necessary to protect against excessive deflection of
330
adjacent scintillator bar layers due to unbalanced pressure, such as the case where one water layer
331
is emptied while the others remain full. A vertical HDPE center strut to limit such deflections
332
was required, so each water layer consists of two side-by-side bladders, each approximately 1 m
333
wide by 2 m high by 3 cm thick.
334
The bladders were heat-welded and sealed into HDPE frames that provide ports for fill, drain,
335
and sensor tubes. The frames also accommodate the mounting and support hardware used to as-
336
semble the Super-PØDules. The frames were edged with silicone sponge gasket material (Stock-
337
well Elastomerics R-10470), such that each bladder was effectively encased within a waterproof
338
seal when compressed between adjacent PØDules. This provides a second level of water con-
339
tainment to protect against leaks. The water bladder frames were equipped with drain ports to
340
direct any water captured within the gaskets into two drip pans mounted below the PØD and
341
basket, one on either side of the basket centerline. Each target layer was filled with water and left
342
overnight prior to acceptance and shipment for integration with scintillator layers and assembly
343
into Super-PØDules. The upstream target Super-PØDule had 13 water layers (26 bladders) and
344
the central target Super-PØDule had 12 layers (24 bladders), for a total of 25 water target layers
345
and 50 bladders. After integration, water fill/drain testing was repeated and repeated a third time
346
when the Super-PØDules arrived at J-PARC.
347
In order to meet the physics analysis requirements, a water pump and monitoring system was
348
designed to allow individual bladders to be filled, drained, and also provide water depth data.
349
A pump rack with 50 self-priming bellows pumps (Gorman-Rupp Industries (GRI) 16001-005
350
F-009 T-007), one for each bladder, was designed with valves arranged to allow each pump to
351
16
be used either to fill or drain a single bladder. The pump rack is located outside the magnet.
352
The pumps are connected to fill and drain tubes on the bladders using Polyflo 66P 3/8-inch OD
353
polyethylene tubing.
354
Water is supplied from a 3000 liter high-density polyethylene tank (DenHartog model VT0900-
355
46), next to the pump rack, coupled to the pumps through two 20-liter buffer tanks on the rack.
356
The main tank is filled with filtered tap water, with commercial chlorine bleach added as a biocide
357
at an effective concentration of 0.025% sodium hypochlorite by volume.
358
Each bladder is equipped with separate drain and fill tubes made of 3/8-inch schedule 40
359
PVC pipe, and has two additional PVC pipes containing water monitoring sensors of two kinds:
360
binary level sensors, which simply report their state as wet or dry, and depth (pressure) sensors.
361
3.2. Fabrication of Water Targets
362
The water target bags are made from a 1 meter wide continuous roll of seamless polyethylene
363
plastic tubing. The plastic sheet is 0.1 cm thick, so the combined thickness of 2 walls of the water
364
bag is 0.2 cm. The tubing is cut to the length of the PØDule height, and is heat welded to make
365
a bag with a leak proof seal at the bottom with no seams on the sides of the bag. Two HPDE
366
frames are then attached, one on the top and one on the bottom section of the bag. The top
367
frame is slotted to provide entrance holes from the top for the water tubes and the pressure and
368
level sensor pipes. The top HDPE frame length is held in position by the HDPE gasket frames
369
on the PØDule. The bottom frame is threaded to allow for through-screws, protruding from the
370
bottom HDPE frame of the gasket frame, to pull the bag taut in the enclosure. This arrangement
371
allows for a leaky bag to be pulled straight from the top of the water target Super-PØDule and
372
replaced by lowering a new bag from the top without removing the Super-PØDule out of the
373
basket frame.
374
3.3. Instrumentation of the water targets
375
In order to properly monitor the water level in each water target layer, sensors are inserted
376
into each water target bladder. Accurate water level readings are needed not only for safety and
377
engineering concerns, but also to determine the water mass inside the fiducial volume of the PØD
378
to the desired accuracy of 3%.
379
17
The monitoring system consists of sensors that are inserted into the bladder and operated in
380
the water, an external monitoring sensor for the environment, and a DAQ system. The entire
381
system is designed to be independent of the ND280 Global Slow Control system in order to be
382
independent of shutdowns and to provide additional flexibility needed during PØD fill or drain
383
operations.
384
Two sensor pipes are installed in each of the 50 water bladders. In each bladder, both pipes
385
have a Global Water WL400 water depth sensor at the bottom end of the pipe. Each pipe also
386
have one Honeywell LLE series Liquid Level Sensor binary wet-dry level sensors placed near its
387
top, for calibration and back-up purposes. All new sensor pipes were water tested overnight or
388
longer prior to deployment. The depth sensors are calibrated by filling test pipes to a series of
389
measured heights, and logging repeated depth readings. These data are then used to fit calibration
390
curves relating depth to current-loop mA. Results are consistent with factory calibration data.
391
The calibration process is checked with a 1-point measurement using an identical test stand after
392
shipment to J-PARC.
393
The water sensors required an auxiliary, custom built, connection board to distribute DC
394
power and convert current-loop signals to voltage signals to be read out by the sensor boards.
395
The sensor boards are mounted on the top of the PØD, above the water target bladders. This
396
board contained a SenSym ICT series ASDX Pressure Transducer, a Texas Instruments TMP275
397
Digital Temperature Sensor, 12-bit ADCs to read the sensor outputs, and digital logic to com-
398
municate with the rest of the DAQ. Communication between the DAQ and system hardware
399
components is performed using I2C. Signals are converted to RS-232 serial data for transmission
400
due to the distance between the PØD and the DAQ computer. Sensor signals were digitized on
401
the Sensor Boards, which were in turn connected to a custom built, 8 port Multiplexer (Mux)
402
Board using the I2C bus. A total of four Mux boards are required. The Mux Board converted
403
the I2C signal to RS232 and also supplied power to the sensors and electronics. The sensors are
404
controlled and monitored by a graphical interface using the Qt 4.0 application and user interface.
405
The monitoring program controls the sensor readout, interprets the data, feeds the data to the
406
GSC system, and then stores the data locally.
407
An identical WL400 sensor and readout board is installed in the main water storage tank in
408
order to monitor its water level during filling and draining.
409
18
4. Super-PØDules
410
The full PØD detector is constructed of four Super-PØDules assembled from PØDules. Fig-
411
ure 4 shows the arrangement of the four Super-PØDules in the assembled detector as well as
412
details of the structure within each type of Super-PØDule.
413
4.1. The Super-PØDule Design and Tooling
414
The four Super-PØDules that make up the PØD were designed to simplify the installation
415
and shipping from the assembly site to the installation site at J-PARC while allowing for effi-
416
cient assembly. This was achieved by assembling all four modules from a set of standardized
417
components (PØDules, water target modules, brass radiators, and lead radiators) that could be
418
preassembled prior to the Super-PØDule assembly. The components were assembled onto their
419
final mounting hardware, which was held on custom rolling carts. The carts were designed to sup-
420
port the Super-PØDules and allow them to be moved. After arrival in Japan, the Super-PØDules
421
were kept on their individual carts which allowed them to be moved and tested independently.
422
Table 1 shows the mass, dimensions and the depth in radiation lengths of each Super-PØDule.
423
During assembly the component masses were sampled and used to estimate the dry mass of
424
each Super-PØDule. The mass of the water added to the target Super-PØDule is accounted for
425
separately. We estimate that the dry mass uncertainty is approximately 0.8%. The length of
426
each Super-PØDule along the neutrino beam axis was measured after assembly and is estimated
427
to have an accuracy of 0.5 mm. The width and height dimensions of the Super-PØDule are
428
perpendicular to the beam axis and have a tolerance of 5 mm. These dimensions include space
429
for the TFB read out electronics.
430
4.2. ECal Super-PØDule Assembly
431
An ECal Super-PØDule is assembled from seven PØDules interleaved with seven layers of
432
lead plates. The assembly began with the construction of the lead panels. The first step was
433
the preparation of a 0.05 cm thick, or 0.03 radiation lengths (R.L.), stainless steel (S.S.) sheet
434
which was mounted on a flat assembly table. A thin aluminum frame was then epoxied and
435
screwed on all four sides of the sheet. Epoxy was applied to the surface of the S.S. sheet and
436
25 3.45 mm thick (0.67 R.L.) lead strips were gently positioned side-by-side onto the epoxied
437
surface. The lead plates were precut to size to minimize any gaps. The lead plates were then
438
19
Super-PØDule Mass Dimensions Depth (kg) (mm×mm×mm) in R.L.
Upstream ECal 2900 2298×2468×305 4.946 Upstream Water 2298×2468×888
Target:
Empty 3600 1.370
Filled 5100 2.379
Central Water 2298×2468×854 Target:
Empty 3500 1.356
Filled 4900 2.287
Central ECal 2900 2298×2468×304 4.946
Table 1: The mass, dimensions, and depth in radiation lengths for each Super-PØDule.
painted with epoxy and two stainless steel half-width sheets were placed onto the pre-epoxied
439
lead plates, creating a panel that contained lead plates sandwiched between a pair of S.S. sheets.
440
A vacuum cover was placed over the entire assembly and evacuated prior to allowing the epoxy
441
to set overnight. Figure 10 is a photo of a S.S. plate being placed onto a layer of lead panels that
442
have been painted with black epoxy.
443
Once the epoxy had set overnight, a U-channel beam was placed across the top of the com-
444
pleted lead panel and screwed into position. The U-channel beam was then lifted with a forklift
445
to position the lead panel and rotated to hang vertically. The lead panel was then moved to a
446
rollabout cart and mounted to a frame. Once the panel was positioned and held into place by the
447
cart frame, the U-channel beam was removed. The completed PØDule, was then lifted vertically
448
and mounted onto the lead panel. This process was repeated seven times to complete an ECal
449
Super-PØDule. After the seventh PØDule was added, threaded stainless steel rods were screwed
450
through clear holes around the periphery of the PØDules and secured in place with a special flat
451
nut.
452
Finally the electronics mounting rail assemblies were attached to the top and on one side of
453
the ECal Super-PØDules where the MPPCs were mounted. Once the readout electronics (see
454
20
Figure 10: A stainless steel plate being placed onto a lead panel
Section 5) were installed, the mini-coax cables from 1820 MPPCs were connected to the TFBs
455
and labeled, completing the assembly. Figure 11 is a photo of a fully assmbled PØDule, mounted
456
vertically on a rollabout stand.
457
4.3. The Water Target Super-PØDule Assembly
458
The water target bag Super-PØDule assembly on the PØDules consisted of two water blad-
459
ders that were mounted on the face of one PØDule and encased by HDPE frames on four sides
460
(top, bottom, left, and right). This procedure was performed on 25 PØDules to form two separate
461
water target Super-PØDules that held 12 or 13 water layers.
462
The assembly began with a PØDule placed flat on an assembly table. Two brass sheets of
463
thickness, 0.15 cm (0.1 R.L.) were placed on the PØDule. A thin weather strip gasket was placed
464
along the PØDule edges on two sides and the bottom. Three HDPE frames were then placed over
465
the weather strips to form a water tight gasket seal and then a vertical HDPE center divider strip
466
was placed in the middle of the PØDule. Next, the two water bladders were placed on the brass
467
sheet. Each water bag had a top and bottom HDPE frame. The next step was to place the weather
468
strip gasket in a groove in the three HDPE frames.
469
Finally a thin HDPE cover sheet was placed over the entire surface to form a water-tight seal
470
and to keep the water bladders from protuding out when the PØDule was mounted vertically and
471
moved. The upper water bladder HDPE frame was supported by slots in the upper HDPE gasket
472
21
Figure 11: The fully-assembled upstream ECal Super-PØDule including the TFB readout electronics boards.
frame. The lower water bladder HDPE frame had mounting holes that were screwed down with
473
screws extending through the bottom HDPE gasket frame. The bottom HDPE gasket frame had
474
drain holes or ports, which were connected to a drain hose to allow any water leak to drain out in
475
a controlled manner.
476
4.4. Shipping and Installation of the Super-PØDules
477
Each of the four Super-PØDules was mounted on a solid wooden base and enclosed in a
478
wooden crate for shipping. The four Super-PØDules were flown from JFK airport in New York
479
to the Narita International Airport near Tokyo by the Nippon Express shipping company in April
480
2009. The crated Super-PØDules were then delivered to J-PARC on two flatbed trucks and
481
22
offloaded into the LINAC building where they were unpacked, removed from their wooden bases
482
and checked out in preparation for installation. All electrical and plumbing utilities were installed
483
in the detector hall and attached to the off-axis detector basket while the Super-PØDules were
484
being checked out in the LINAC building, which simplified the installation of the Super-PØDules
485
into the basket.
486
Each Super-PØDule remained bolted to custom rolling carts until just prior to installation.
487
Aluminum covers were attached to the side and top electronics of the Super-PØDule to prevent
488
damage during installation into the detector basket. Super-PØDules were installed in order from
489
the most upstream side of the detector basket. For installation each Super-PØDule was discon-
490
nected from its cart and lifted by a crane using a custom lift fixture (Fig. 12) that allowed precise
491
positioning of the Super-PØDule as it was lowered into the basket about 20 m below the staging
492
area. Figure 12 shows the first Super-PØDule, the upstream ECAL, being lowered into position
493
with guidance from a local contractor.
Figure 12: The left figure shows an engineering drawing of an ECAL module being lifted by the custom lift fixture, while the right picture shows the first Super-PØDule (upstream ECAL) being lowered into the detector basket.
494
Once a Super-PØDule was set into position in the basket, it was bolted into place and util-
495
ity connections were made to it. Precise positioning of each Super-PØDule in the basket was
496
accomplished using adjustment screws. Following the utility connections and the filling of the
497
water targets, aluminum covers were attached to the PØD in preparation for data taking.
498
23
5. Readout Electronics and DAQ
499
The PØD, like the SMRD, ECal, and INGRID detectors in the ND280, uses Trip-t-based
500
front-end electronics [11] to read out its 10,400 photosensors. Each Trip-t Front-end Board
501
(TFB) contains four 32-channel Trip-t ASIC chips [12], originally developed at Fermilab for the
502
DØ experiment, and can serve up to 64 MPPC sensors, which are connected by miniature coaxial
503
cables to the TFB. The signal from the photosensor is divided capacitively in the ratio of 10:1
504
and routed into separate Trip-t channels to provide a high and a low gain response to the same
505
input, thereby increasing the dynamic range of the electronics. The high gain channel provides
506
measurement for up to a 50 photoelectron (p.e.) signal with∼10 ADC/p.e. resolution, while the
507
low gain channel can be used to measure larger signals up to about 500 p.e.
508
The Trip-t chips integrate the charge in 23 consecutive integration cycles that are synchro-
509
nized with the accelerator beam spill such that each 58 ns wide beam bunch, separated by∼
510
580 ns, falls into a separate integration window 100 ns away from the start of each integration
511
cycle. The length of the integration cycle and the reset period between them can be programmed
512
and are set to 480 ns and 100 ns respectively for beam operation. These values will virtually
513
eliminate deadtime for in beam spill interactions but will result in some deadtime for out of spill
514
events such as Michel electrons. The Trip-t stores the results of the 23 integration cycles in its
515
analog pipeline, which are digitized using a 10-bit ADC when the data are transferred offthe
516
board following the end of the integration period. The high gain channels also feed an internal
517
discriminator that measures the time when the integrated signal exceeds a programmable preset
518
threshold of about 2.5 p.e. in each integration cycle. The operation of the TFB is controlled
519
by a Field Programmable Gate Array (FPGA), which also provides time stamping for the dis-
520
criminator output with a 2.5 ns resolution, and moves the data from the Trip-t to the back-end
521
electronics.
522
The TFB operation requires four low voltage levels: 5.0 V, 3.3 V, 2.5 V, and 1.2 V. The bias
523
voltages for the MPPCs are also provided through the TFB with the central core of the mini-
524
coax cables connecting the sensors to the board being at the same uniform voltage level (∼70 V).
525
Individual voltage adjustments are achieved by setting the voltage level of the shield sheath of
526
the coax cable between 0 to 5 V in 20 mV increments using an 8-bit DAC.
527
The entire PØD readout requires 174 TFBs; 29 for each of the upstream and central ECal
528
Super-PØDules and 58 for each of the water target Super-PØDules. The back-end electronics of
529
24
the PØD consist of 6 read-out merger modules (RMM), a cosmic trigger module (CTM) shared
530
with the SMRD and the downstream ECal (DsECal), a slave clock module (SCM), and one mas-
531
ter clock module (MCM) for the whole ND280 detector. Each RMM serves as a communication
532
interface between the 29 TFBs and the data acquisition (DAQ) system, by passing control com-
533
mands, clock, and trigger signal in one direction, and data in the other. Signals between the
534
RMM and TFBs are transmitted using the LVDS protocol via shielded Ethernet cables, while the
535
RMMs are connected to the DAQ computers with optical Gigabit Ethernet links.
536
Cosmic trigger primitives are formed from the 29 TFBs on the upstream ECal Super-PØDule
537
based on coincidences between some number of MPPCs. These trigger primitives are transmitted
538
to the CTM and combined with other trigger primitives from the SMRD and the DsECal to create
539
a global cosmic trigger decision when any side pairs of the ND280 detector are traversed by a
540
cosmic ray muon. The MCM receives the accelerator timing signals when a spill happens and
541
transmits trigger as well as periodic clock synchronization signals to the RMMs and TFBs via
542
the SCM. The MCM can also generate pedestal and calibration triggers, such as for synchronous
543
operation of the PØD light injection system. The SCM duplicates most of the functionality of
544
the MCM and allows the configuration and operation of each sub-detector, such as the PØD, in
545
standalone mode independent of the other detectors.
546
The global ND280 DAQ [2] [13] uses the MIDAS framework [14], developed at TRIUMF,
547
operating on computing nodes running Scientific LINUX operating system. The TFBs are con-
548
trolled and read out by the front-end processing nodes (FPN), each of which serves two RMMs
549
(3 FPN are used for the PØD). The read-out and configuration is provided by the read-out task
550
(RXT), while the raw data from the TFBs are decoded and formatted for output by the data
551
processing task (DPT). The DPT also performs per-channel histogramming of specific trigger
552
types before zero suppression and inserts the histograms to the output data stream periodically.
553
An event builder process collects the fragments from the sub-detectors, and writes the fully-
554
assembled events to a buffer after basic consistency checks. Finally, an archiver process transfers
555
the completed files to mass storage and creates a preview copy on a local semi-offline system for
556
fast-turnaround calibration and data quality checks.
557
A global slow control (GSC) complements the global DAQ using the same MIDAS-based
558
software framework. Front-end tasks running on the sub-detector computers connect to various
559
equipment of the sub-detectors and collect monitoring data that are stored in a MySQL database
560
25