Accepted Manuscript Tracking with the MINOS Time Projection Chamber C. Santamaria, A. Obertelli, S. Ota, M. Sasano, E. Takada, L. Audirac, H. Baba, D. Calvet, F. Château, A. Corsi, A. Delbart, P. Doornenbal, A. Giganon, A. Gillibert, Y. Kondo, Y. Kubota, C. Lahonde-Hamdoun, V. Lapoux, D. Leboeuf, C.S. Lee, H.N. Liu, M. Matsushita, T. Motobayashi, M. Niikura, M. Nishimura, H. Otsu, A. Peyaud, E.C. Pollacco, G. Prono, H. Tokieda, T. Uesaka, J. Zenihiro
PII: DOI: Reference:
S0168-9002(18)30888-X https://doi.org/10.1016/j.nima.2018.07.053 NIMA 60999
To appear in:
Nuclear Inst. and Methods in Physics Research, A
Received date : 23 March 2018 Revised date : 23 June 2018 Accepted date : 18 July 2018 Please cite this article as: C. Santamaria, A. Obertelli, S. Ota, M. Sasano, E. Takada, L. Audirac, H. Baba, D. Calvet, F. Château, A. Corsi, A. Delbart, P. Doornenbal, A. Giganon, A. Gillibert, Y. Kondo, Y. Kubota, C. Lahonde-Hamdoun, V. Lapoux, D. Leboeuf, C.S. Lee, H.N. Liu, M. Matsushita, T. Motobayashi, M. Niikura, M. Nishimura, H. Otsu, A. Peyaud, E.C. Pollacco, G. Prono, H. Tokieda, T. Uesaka, J. Zenihiro, Tracking with the MINOS Time Projection Chamber, Nuclear Inst. and Methods in Physics Research, A (2018), https://doi.org/10.1016/j.nima.2018.07.053 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
*Manuscript Click here to view linked References
1
2 3 4 5 6 7 8
Tracking with the MINOS Time Projection Chamber C. Santamariaa,b , A. Obertellia,b,c , S. Otad , M. Sasanob , E. Takadae , L. Audiraca , H. Babab , D. Calveta , F. Chˆateaua , A. Corsia , A. Delbarta , P. Doornenbalb , A. Giganona , A. Gilliberta , Y. Kondof , Y. Kubotad , C. Lahonde-Hamdouna , V. Lapouxa , D. Leboeufa , C.S. Leeb , H.N. Liua , M. Matsushitad , T. Motobayashib , M. Niikurag , M. Nishimurab , H. Otsub , A. Peyauda , E.C. Pollaccoa , G. Pronoa , H. Tokiedad , T. Uesakab , J. Zenihirob a
17
IRFU, CEA, Universit´e Paris-Saclay, F-91191 Gif-sur-Yvette, France RIKEN Nishina Center, 2-1 Hirosawa, Wako, Saitama 351-0198, Japan c Institut f¨ ur Kernphysik, Technische Universit¨ at Darmstadt, 64289 Darmstadt, Germany d Center for Nuclear Study, University of Tokyo, RIKEN campus, Wako, Saitama 351-0198, Japan e NIRS-HIMAC, 4-9-1 Anagawa, Inage-ku, Chiba 263-8555, Japan f Tokyo Institute of Technology, 2-12-1 Ookayama, Meguro-ku, Tokyo 152-8550, Japan g Department of Physics, University of Tokyo, 7-3-1 Hongo, Bunkyo, Tokyo 113-0033, Japan
18
Abstract
9 10 11 12 13 14 15 16
b
We report on the performance of the MINOS Time Projection Chamber developed as a vertex tracker to study exotic nuclei produced from hydrogeninduced knockout via in-beam γ-ray and invariant-mass spectroscopy. Inbeam measurements with 4 He and 20 Ne beams at 200 and 350 MeV/nucleon, respectively, were performed at the HIMAC facility. The tracking algorithm for protons after quasi-free scattering is described. Realistic simulations and physics experiments are compared and show a good agreement. The vertex position resolution reaches 5 mm FWHM, mostly from re-scattering with the target and the Aluminum reaction chamber. The overall efficiency of vertex reconstruction is also benchmarked with the first experimental campaign data
Preprint submitted to Nuclear Physics and Methods A
June 23, 2018
performed at the RIBF, confirming an overall efficiency better than 90 % for physics experiments. 19
Keywords: Time Projection Chamber, tracking software, Hough
20
transform, Spectroscopy of exotic nuclei
21
1. Introduction
22
Away from the valley of stability, the nuclear structure has been shown
23
to evolve with respect to stable nuclei. However, many interesting short-
24
lived nuclei are hard to reach experimentally, even at state-of-the-art ra-
25
dioactive beam facilities. In-beam spectroscopy at intermediate energies has
26
been shown as a powerful method to perform the spectroscopy of the most
27
exotic species. The thickness of the target used for these experiments has
28
always been a compromise: the thicker it is, the more luminosity is provided
29
but also worsening the energy resolution due to uncertainty on the reaction
30
vertex position, i.e., the velocity of the projectile at the reaction vertex.
31
32
To address this limitation, the MINOS device [1, 2] was developed to re-
33
construct the interaction vertex within the target for knockout reactions at
34
the Radioactive Isotope Beam Factory (RIBF) of RIKEN, either coupled to
35
a gamma array for in-beam γ-ray spectroscopy or to particle detectors for
36
invariant-mass spectroscopy (of unbound states). MINOS was designed to
37
fulfill two criteria: (i) The detector offers a large geometrical acceptance for
38
a total efficiency of more than 80%. (ii) The vertex resolution should allow
39
a velocity reconstruction of a few ‰ along the beam axis. This corresponds
40
to a few millimeters resolution. Several experiments were performed with 2
41
MINOS at the RIBF since April 2014. First results from in-beam gamma
42
experiments in combination with the DALI2 NaI array [3] of neutron rich
43
nuclei have already been published [4–10]. Also, missing mass and invariant
44
mass spectroscopy experiments have been performed with MINOS at SAMU-
45
RAI [11, 12].
Figure 1: Principle of the MINOS active target detector system: a liquid H2 cryogenic target (blue cylinder) of 10-20 cm length and 40 mm diameter surrounded by a Time Projection Chamber (pink cylinder). The blueish grey parts are the cryogenic device with the cryostat on the left and the beam surrounding the target for thermal isolation. A (p, 2p) reaction induced by an incident A Z beam is schemed in the target, producing two protons tracked in the TPC. During the HIMAC tests, only the TPC was used. At the RIBF, the full device was implanted at the target focal point of the beam line, either in the F8 area of the BigRIPS line or in the SAMURAI area. 46
47
The full MINOS device is shown in Fig. 1. It is composed of a thick liquid
48
cylindrical hydrogen target [13, 14] (possible thicknesses ranging from 50 to
49
200 mm) and surrounded by an annular Time Projection Chamber (TPC)
50
equipped with a Micromegas [15, 16] pad detection plane to track recoil
51
protons from (p, 2p)-type reactions and to reconstruct the vertex position. 3
52
Thanks to this tracking concept, the beam energy at the interaction point
53
in the target is known and the system allows a resolution as good as with
54
thin solid standard targets. The MINOS system was developed from 2011 to
55
2013. In experimental conditions as illustrated in Fig. 1, beam particles go
56
through the liquid hydrogen target and a knockout reaction may take place
57
inside the target. The heavy fragment is scattered from the initial beam
58
direction at a small angle and is eventually identified downstream. The re-
59
coiling protons are ejected at a larger angle and in most cases at least one
60
of the two intersects the TPC surrounding the target. Charged particles,
61
while punching through the TPC, ionize the gas. The ionized electrons then
62
drift towards the anode, i.e. the Micromegas detection plane, in an electric
63
potential of typical value 120-250 V/cm2 . The tracks detected in the TPC
64
are exclusively straight lines as no magnetic field is applied and the charge
65
deposition along the track is uniform for energetic protons.
66
67
The signals from the detector are converted and read by dedicated read-
68
out electronics [17]. In the present study, they are composed of Front-End
69
Cards (FEC) equipped with AFTER chips, as for the T2K experiment [18].
70
Recently, the MINOS electronics have been equipped with state-of-the-art
71
AGET chips [19] from the GET project [20, 21]. They are read out in both
72
cases by the dedicated so-called Feminos cards [22]. All the cards are then
73
synchronized to a common clock and trigger by a Trigger Clock Module card
74
(TCM) [23, 24].
75
76
We present here the performance of the MINOS TPC, both with an in-
4
77
beam test at the HIMAC facility in Japan and with realistic simulations. In
78
section 2, we briefly introduce the experimental setup. Then the tracking
79
algorithm created for MINOS is detailed in section 3. In section 4 we expose
80
the experimental results and compare them with Monte Carlo simulations. In
81
section 5, we apply simulations for MINOS to realistic physics cases measured
82
at the RIBF.
83
2. Experimental setup
84
The TPC was tested at the HIMAC facility [25] for heavy ion therapy in
85
Chiba, Japan, in one of the two experimental rooms dedicated to physics ex-
86
periments. Beams of 20 Ne at 350 MeV/nucleon and 4 He at 200 MeV/nucleon
87
were used, the latter one being used in a parasitic way after the CAT-TPC
88
[26]. For the experimental setup illustrated in Fig. 2, CH2 or C targets of 30
89
mm diameter and 0.5 mm thickness were used, i.e., 5.73×1021 and 5.65×1021
90
atoms/cm2 respectively. The targets were mounted on Plexiglass frames for
91
an overall 40 mm diameter and placed about 120 mm apart along the beam
92
path inside an aluminum beam pipe identical to the reaction chamber used
93
during physics experiments. Note that several tests were performed during
94
the experiment and therefore the distance between the targets was modified
95
from one measurement to the other. Variations of the distance between the
96
two targets were of ±4 mm around 120 mm. Measurements were focusing
97
on the resolution, i.e., width of the reconstructed vertex position. For con-
98
venience, the beam pipe was kept in air and had 2 mm thickness and 72
99
mm internal diameter, with an equivalent thickness of 1.47 × 1021 atoms/cm2
100
inside. As a remark, the effect of straggling in the aluminum tube could be 5
101
reduced by cutting some part of the pipe for Mylar windows, for experiments
102
that really need to minimize angular straggling. The MINOS TPC had an
103
internal diameter of 80 mm, an external diameter of 180.8 mm, and a length
104
of 300 mm and was mounted on the beam pipe as shown in the picture on
105
the left side of Fig. 2.
106
Figure 2: (Left) MINOS TPC at the HIMAC facility. (Right) MINOS experimental scheme seen from above of the in-beam test at HIMAC, with distances in millimeters.
107
A MicroMegas detection plane with a ”projective” pad geometry (cf. Fig. 15
108
of Ref. [1]) was used on the anode plane of the MINOS TPC. It is composed
109
of 18 rings of 2 mm radius containing each 256 radial segments. Pads on
110
the outer ring of the TPC have therefore a larger area than those on the
111
inner ring. The 4608 pads for the projective geometry were readout with the
112
digital electronics described in the section above. Data taking was performed
113
with the following settings for the electronics: 50 MHz sampling frequency,
114
240 fC gain range and 200 ns shaping time. The gas used in the TPC was a 6
115
116
117
triple mixture of Ar (82%), isoC4 H10 (3%) and CF4 (15%). From simulations with the Magboltz software [27], the longitudinal and transverse diffusions √ were estimated at ∼200 µm ` (` being the drift length in cm). The cathode
118
was set to 5 (5.4) kV voltage leading to a field of 152 (166) V/cm and the
119
Micromegas stage was set to 450 (430) V on the mesh which corresponds to
120
an amplification gain around 5800 (2900) for the
121
the data taking, impurities were monitored online with O2 impurities main-
122
tained below 60 ppm and H2 O impurities below 1000 ppm.
20
Ne (4 He) beam. During
123
124
Plastic scintillators were positioned in the beam path upstream the TPC
125
for beam triggering and timing and two layers composed each of two plastic
126
scintillators of 220 mm width, 630 mm length, and 20 mm thickness were
127
placed on both sides of the TPC as shown in Fig. 2. Their position was set
128
to maximize the detection of the protons coming from the (p, 2p) quasi-free
129
scattering kinematics. The trigger was composed of a coincidence with the
130
beam plastic scintillator upstream and a coincidence on the left and right
131
scintillators on the sides of the TPC. Recoiling charged particles produced
132
at large scattering angles were detected by the plastic scintillator side layers
133
and provided a trigger information for two-particle events in the TPC. As
134
no particle identification was performed in the TPC itself or downstream our
135
setup, (p, 2p) events were not unambiguously identified. The vertex position
136
resolution and the detector efficiency detailed in the following sections were
137
consequently extracted with unidentified particles which could worsen the
138
vertex determination and impact the overall resolution.
7
139
3. Tracking algorithm
140
The tracking algorithm for MINOS has two main constraints from the
141
physics experiments. (i) The detector is intended to be used for experiments
142
with low beam intensity. A software with the highest tracking efficiency
143
should therefore be favored. (ii) A proper compensation of target effects, for
144
example a good Doppler reconstruction for in-beam gamma spectroscopy ex-
145
periments, implies a resolution in the beam direction for the vertex position
146
of 5 mm FWHM or better.
147
148
Several methods have been applied for tracking such as a Kalman filter for
149
the PANDA prototype TPC [28] or a back-tracking of the photons with
150
simulation-based pulse-shape analysis of the signals for the AGATA spec-
151
trometer [29]. As the TPC tracks are linear, the Hough Transform (HT) [30]
152
is a good compromise between computing time and reconstruction accuracy
153
in our case. The HT has been a constant source of ideas for pattern recog-
154
nition, first for the detection of straight lines [31] but later also applied to
155
the detection of arbitrary patterns [32]. For the recognition of straight lines
156
in two dimensions, the HT changes each point in the coordinate space into a
157
straight line in the parameter space. Each line is specified in the real space
158
by two parameters (r, θ), r being the algebraic minimum distance between
159
the line and the origin and θ being the angle from this orthogonal vector and
160
a given reference axis. The parametrization becomes unique once the angle
161
θ is restricted to [0, π], r being either negative or positive in this standard
162
representation.
163
A first objective of the tracking algorithm is to find the multiplicity of tracks 8
while Npads ≥ nmin
Raw event ( x, y, q(t) ) if qmax ≥ qmin Hough transform 2D
Hough transform k if Npads ≥ nmin & Cring
else
Keep (xnew, ynew) track i Discard pads for next iteration else
if Npads ≥ nmin
if 0 < Ntrack < 5 Fit of q(t) signals (xnew, ynew, znew, qmax) for every track
Hough transform 3D
Hough transform 3D if Npads ≥ nmin else & Rmin ≥ 15 Keep (xf, yf, zf) final track j
Discarded track if Nfinal track = {1,2}
3D Fit 4 parameters for each track if Nfinal track = 2 if Nfinal track = 1 Vertex finding with beam & track (xv, yv, zv) vertex position
Vertex finding with the 2 tracks (xv, yv, zv) vertex position
Figure 3: Scheme of the overall tracking algorithm with details on the applied conditions in Table 1.
9
164
per event in order to distinguish the one- and two-proton-like events. A first
165
selection of the number of particles per event is done, called event filtering,
166
by applying an adapted Hough filter to the Micromegas detection plane (xy).
167
Moreover, the tracking algorithm must filter off any possible noise contribu-
168
tions from the tracks (electronics noise, delta electrons) before being able to
169
reconstruct the vertex of interaction. The general algorithm scheme shown
170
in Fig. 3 is explained in details in the following paragraphs.
171
3.1. Event filtering
172
At a first stage, only the 2-dimensional information from the Micromegas
173
(charge projection on pad plane with no drift time information) are consid-
174
ered for each event. We therefore obtain a set of (x, y)j points for each event,
175
placed in the annular disk forming the detection plane. Physical tracks come
176
from the inner disk where the targets are located, imposing limits on the
177
Hough parameter space. We adapted the HT to the geometry of MINOS.
178
Each track segment in the TPC resulting from a reaction in the target is
179
determined by the crossing points on the inner and outer radii of the TPC.
180
The intersections are parameterized with the angles θint and θext respectively,
181
as shown in Fig. 4. For each (x, y)j pad on the Micromegas plane there is a
182
range of possible tracks coming from inside the beam pipe passing through
183
it with (θint , θext )i parameters. For every hit pad (x, y)j in the TPC, an iteration over all possible θint is made with a binning of ∆θint = 2◦ . For each possible value, there is one and only one line, crossing at the same time the hit point and the point located at the inner radius Rint at an angle θint , with coordinates (xint = Rint cos θint , yint = Rint sin θint ). The point (xext , yext ) at which this line crosses the outer radius 10
Figure 4: View of the HT parameterization used with the MINOS geometry. One possible track (in red) crossing the point (x, y)j (in purple) is shown. The range of possible angles θint and θext for this specific hit pad (x, y)j are shown in orange, they correspond to tracks coming from inside of the beam pipe.
is given by:
y = (y − p x ) + ext int 1 int x2 + y 2 = R 2 . ext ext ext
yint −y x , xint −x ext
(1)
184
Only one solution for (xext , yext ) corresponds to a track from a particle emit-
185
ted from the θint position of the inner tube of the TPC. This unique solution
186
determines θext .
187
We apply this iteratively and calculate all the possible (θint , θext )i combi-
188
nations for all the hit pads j in one event. Those values are stored in a
189
2-dimensional histogram. Physical tracks are identified by maxima in the
190
Hough space. Once a maximum is found in the Hough space, all the pads on
191
the line parameterized within the (θint , θext )max histogram binning are con-
192
sidered part of the same track and removed for the next iteration. The track
193
is considered only if it contains more than nmin = 10 pads as physical tracks
11
194
should all cross the detector given the transparent nature of the TPC. This
195
value for the minimum number of pads in a track was chosen from diffusion
196
considerations to remove the most number of stray pads from background.
197
An additional condition Σinner = 3 on the number of hit pads is applied in
198
the most inner rings of the TPC to rule out charged particles created outside
199
the target region. The procedure is applied several times until the number
200
of remaining pads with signal is smaller than nmin .
Figure 5: Two-dimensional HT applied to a two-particle event, with the treatment of the first track on the top and of the second track on the bottom. From left to right is shown the hit pads before the filter, the curves in the Hough representation of the event, and finally the hit pads after filtering out the points in the new-found track.
201
An example is given in Fig. 5, in which two tracks are recorded on the detec-
202
tion plane on the top left figure. In the top middle plot, all the possible cor-
203
responding (θint , θext )i couples are represented in the modified Hough space. 12
204
Two maxima can be distinguished, with angles around 350°and 160°which
205
correspond to the track circled in red and blue in the top left and bottom
206
left figures respectively. On the top right side is shown the pads considered
207
as part of this first cluster, which match the track seen at first sight in the
208
red circle. Those pads are removed for the next iteration presented in the
209
bottom part of the panel. The remaining track in the blue circle on the left
210
is parametrized in the Hough space and the second maximum leads to the
211
second cluster of pads in the bottom right figure.
212
3.2. Track filtering and time information In a second step all events with at least one track out of the previous HT are recovered and the software considers the previously discarded information on each pad. The charge deposited as a function of time or pulse shape gives the time and charge information. This signal is driven by several parameters in the electronics: the shaping time, the digitization frequency and the gain. An empirical formula for the signal q(t) as a function of the trigger time tpad relative to an external trigger (given by the beam plastic scintillator), the shaping time τ and the amplitude A can be written as: 3 t−t t − tpad t − tpad −3 τpad q(t) = A × e sin + qb τ τ
(2)
with the signal baseline qb , a constant fixed by the electronics. This analytical formula of the response function of the electronics has been obtained by fits to measured waveforms [33]. Therefore, by fitting with this function, one can extract the time tpad (in nanoseconds) at which the signal is produced in the TPC and the maximum of the function which is the charge qpad (in femtocoulombs) deposited on the 13
pad. tpad is the time it takes for the electrons to drift from the ionization at the track location to the detection plane relative to the trigger time tmin . The drift time is proportional to the position zpad at which the ionization took place inside the TPC along the beam direction: zpad = (tpad − tmin ) × vdrif t
(3)
213
Knowing the drift velocity vdrif t of the gas (see section 4.1 for its determina-
214
tion), a three dimensional picture of charge deposition can be deduced along
215
the track in the TPC.
216
217
Standard two-dimensional HT are then applied [31] in the (xy), (xz) and (yz)
218
planes once for each track of the selected events in order to filter the possible
219
delta electrons produced along the track (see Fig. 6). The track is taken into
220
account only if at least Rmin = 15 rings are hit when the charged particle is
221
not passing through the cathode. Otherwise for low scattering angle particles
222
that punch through the cathode, the final track is still required to contain
223
at least nmin = 10 pads to be registered. This value for the number of rings
224
allows for a few pads to be removed in the reconstruction and cleaning steps
225
from restrictive algorithms. It has been determined empirically to yield the
226
best reconstruction efficiency, and checked on a sample of data event-by-
227
event. As a result, one finds the final number of tracks for each event and
228
only the final one- and two-particle events are kept in the present analysis.
229
3.3. Track fitting and interaction vertex The fit of the tracks is performed with a minimization of the distance between every hit point and the reconstructed track weighted by charge in 14
Figure 6: HT showing the filtering of a track. From left to right the three 2-dimensional HT in the (xy), (xz) and (yz) planes are represented. From top to bottom the hit pads before the filter, the curves in the Hough space of the event, and finally the hit pads after filtering are shown.
the three dimensional space. In the present study, the determination of the vertex position in three dimensions is done for two-charged-particle events. As two straight lines in three dimensions generally do not intersect, the vertex position is determined as the midpoint of the line segment of minimum distance connecting the two tracks. To calculate the minimal distance between two tracks in three dimensions, we consider two vectors of coordinates → − − x and → x belonging to each line. The distance D between them is given a
b
− − by D2 = k→ xb − → xa k2 and can be analytically minimized. We can define the − vertex position of vector → x as the middle of this segment and we eventually v
15
get:
→ − − xa + → xb → − xv = 2 D=Dmin
(4)
230
The target being aligned along the beam for the physics experiments, the ver-
231
tex position zv along the beam direction is the most important parameter for
232
Doppler reconstructions of the gamma ray spectra or invariant mass measure-
233
ments. We therefore focus in the following on this parameter to extract the
234
resolution obtained during the HIMAC experiment. In physics experiments,
235
not only two-charged particle but also one-charged particle events are taken
236
into account for the reconstruction of the vertex using this particle track and
237
the beam trajectory in Eq. 4 instead. In this study, we will concentrate on
238
the two-charged particle events to extract the TPC performance.
239
4. Experimental results
240
In the following, the procedure to extract the drift velocity from the in-
241
formation collected by the TPC is described in Section 4.1. This allows us
242
to reconstruct the hit tracks in three dimensions and to apply the tracking
243
method described in Section 3 to calculate the resolution for the vertex po-
244
sition in Section 4.2. The pad multiplicity is studied in Section 4.3 as an
245
important parameter to check the gas transverse dispersion properties of the
246
TPC.
247
In the following experimental results are compared to simulations. They
248
are done in three consecutive steps: (i) a GEANT4 (v10.00) [34] simulation
249
including a reaction process (INCL v09.06.47) [35, 36] to generate events
250
and produce charged particles in the device, (ii) a Monte Carlo simulation to
251
reproduce the drift and the amplification of the ionized electrons towards the 16
Figure 7: (Left) Measured time in µs and extracted position or drift length z along the beam line inside the TPC for a
20
Ne run (350 MeV/nucleon). The time and distance 0
correspond to the Micromegas plane. Its calculated derivative is also shown in red. (Right) Deduced drift velocity as a function of the electric field applied in the MINOS TPC for several runs with different settings during the HIMAC measurements. Statistical error bars are smaller than the markers. The experimental points are compared to Magboltz simulations (blue line) with no impurities.
252
Micromegas detection plane, (iii) the reconstruction of the vertex position
253
via the above tracking software used also for experiment. Details on the
254
simulations are given in Ref. [1].
255
4.1. Drift velocity
256
The drift velocity vdrif t is a key parameter of the TPC and depends on
257
the electric field and gas composition including impurities such as H2 O and
258
O2 . It is determined by Eq. 3 from the distribution of trigger times tpad
259
inside the TPC for all pads and events, as shown on the left side of Fig. 7 in
260
blue. The histogram representing all possible drift times in the TPC should
261
reflect its length. As a convention, we choose the Micromegas plane of the 17
262
TPC to be the origin of the beam line axis, and therefore the statistics in
263
the plot must start at 0 mm and stop at zmax = 300.0(2) mm. With these
264
conditions, we can deduce the drift velocity by measuring the minimum and
265
maximum measured trigger times tmin and tmax inside the TPC such as in
266
Fig. 7: vdrif t =
267
mixture of Ar(82%) + CF4 (15%) + C4 H10 (3%), a drift velocity of 4.42(2)
268
cm/µs is found. We also show the calculated derivative of the drift time in
269
red in Fig. 7 as further reference for the reader. The rise and fall of the drift
270
times clearly correspond to peaks in the derivative.
271
During the performance measurements at HIMAC, several configurations of
272
electric field were taken, and the extracted drift velocities are plotted in Fig.
273
7. They are in very good agreement compared to Magboltz simulations [27],
274
consistent with no impurities in the TPC gas.
275
zmax . (tmax −tmin )
In the case of a 152 V/cm electric field and a gas
With these drift velocities, we can extract the vertex position along the
276
beam direction with the tracking algorithm.
277
4.2. Resolutions
278
With the tracking algorithm, we can extract the vertex position for (p, 2p)-
279
like events in the case of two-particle events with the fit of the tracks and the
280
reconstruction of the vertex point. The conditions applied for the tracking
281
are summarized in Tab. 1. The resolution of the vertex position is then ex-
282
tracted from the vertex position along the beam line for all filtered events as
283
shown in Fig. 8 on the left side for a run of
284
at 5 × 104 particles per second (pps) with two CH2 targets (relative distance:
285
120(4) mm). We obtain a resolution at full-width-half-maximum of 5.3(2)
286
mm and 6.6(3) mm for the first and the second CH2 targets, respectively. 18
20
Ne beam at 350 MeV/nucleon
Energy threshold of the pads
qmin ≈ 9000 e−
HT in 2D
∆θint/ext = 2◦
(binnings and intervals)
θint/ext ∈ [0◦ , 360◦ ]
Minimum number of points in track
nmin = 10
Condition Σinner
Npads ≥ 3
in the 4 first rings of the TPC Events with number of tracks
0 < Ntrack < 5
considered in 3D HT in 3D
∆ρxy/xz/yz = 2◦
(binnings and intervals)
∆ρxy/xz/yz = 3 mm ρxy ∈ [−45 mm, 45 mm] ρxz/yz ∈ [−300 mm, 300 mm] θxy/xz/yz ∈ [0◦ , 180◦ ]
Minimum number of rings touched
Rmin = 15
(if not crossing the cathode) Table 1: Conditions applied in the present study for the tracking algorithm.
19
Figure 8: (Left) (Filled histogram) Vertex position zv along the beam axis for a run with a 20
Ne beam at 350 MeV/nucleon. (Non-filled histogram) Results of a GEANT4 simulation
performed with the same conditions than the experiment and with the same tracking software, normalized to the number of counts in the experiment. (Right) Same with a 4 He beam at 200 MeV/nucleon.
287
The resolution of the vertex position is extracted from the vertex position
288
along the beam line for all filtered events as shown in Fig. 8 on the right side
289
for a run of 4 He beam at 200 MeV/nucleon at 1.5 × 104 pps with the same
290
setup. We obtain a resolution at full-width-half-maximum of 5.5(1) mm and
291
7.6(1) mm for the first and the second CH2 targets, respectively.
292
Simulations are performed in order to realistically compare our results. There
293
is an overall fairly good agreement between experimental data and simula-
294
tions. The experiment exhibits a slightly worse resolution for the second
295
target with both incident beams. This could come from the more forward
296
position of the created proton tracks: they need to travel farther in the TPC,
297
i.e., the ionized electrons experience a longer drift and therefore diffusion,
298
which affects the average multiplicity of the track. However, the simula-
299
tions exhibit a poorer resolution for the first target than for the second one.
20
300
Moreover, the integral of the two peaks were compared. The data shows an
301
equal amount of reactions occurring in the first and second target, while the
302
simulations exhibit a loss of counts in the second peak.
303
This difference between simulations and data was investigated to understand
304
the simulation bias. Plotting the proton angle distributions for the two differ-
305
ent beams, as seen in Fig. 9, the simulation discrepancy shows a difference in
306
the kinematics as compared to the experiment, especially for the 20 Ne beam.
307
The second target being placed at the middle of the TPC, the geometric effi-
308
ciency is smaller than for the first target as the tracks produced in the inner
309
tube outside of the TPC cover a wider solid angle in this case. However, the
310
simulations have taken into account the relative positions of the targets and
311
TPC, as demonstrated by the smaller number of counts in the simulations for
312
the second target. Comparing these considerations with the data exhibiting
313
a similar number of reactions in both targets, the supplementary counts in
314
the second targets could come from multiple scattering in the targets, which
315
are not accounted for in the simulations.
Figure 9: Angles (in degrees) of the charged particles emitted from the regions of the two CH2 targets, on the left for the first target with zv =[5,50] mm and on the right for the second target with zv =[125,170] mm. The data points (in black) are extracted from a run with a
20
Ne beam at 350 MeV/nucleon and compared to simulations (in blue).
21
316
4.3. Pad multiplicity
Figure 10: The top left panel shows the pad multiplicity as a function of the radius on the Micromegas plane. Pad multiplicities for the first, ninth and eighteenth (last) ring as a function of the position along the beam axis are shown in the other panels. The data points (in black) are extracted from a run with a 4 He beam at 200 MeV/nucleon and compared to simulations (in blue).
317
For the projective pad geometry, we have defined the pad multiplicity as
318
the number of pads hit for each ring of the TPC annular detector. As the
319
pads on the outer rings have a larger area, the ring multiplicity should de-
320
crease as one goes from the center of the TPC to the outer radius. The ring
321
multiplicity is extracted for each two-particle event which passed the track-
322
ing algorithm, and averaged over the number of events taken into account. 22
323
Its absolute value depends on the energy threshold taken for each pad, in
324
our case around 9000 electrons on the detection plane. The result for a 4 He
325
beam at 200 MeV/nucleon and a
326
a detector gain around 5800) is illustrated in the upper left part of Fig.10
327
and Fig.11 respectively as a function of the ring radius. The effect of the pad
328
size is clearly visible in this graph with a downward slope when going to the
329
exterior of the TPC, i.e., to larger radii.
20
Ne beam run at 350 MeV/nucleon (with
330
Figure 11: Same plot as Fig.10 but for a run with a
331
332
333
20
Ne beam at 350 MeV/nucleon.
There is a specific dispersion for a given gas mixture inside the TPC, here of √ about ∼200 µm `, which is a function of the distance the ionized electrons have to travel in the gas. As a result, the tracks coming closer to the cathode 23
334
are more dispersed when they get to the detection plane. We calculate the
335
ring multiplicities for the inner, middle and outer rings of the TPC as a
336
function of the position along the beam axis of the track points. The results,
337
as seen in the other three figures of Fig.10, reveal a slight slope in dispersion
338
as we go farther from the Micromegas plane, i.e., for longer drifts. They are
339
compared to simulations which show a very good agreement.
340
5. Physics cases simulations
341
Simulations were performed to obtain vertex position resolutions and de-
342
tection efficiencies for physics cases measured at the Radioactive Isotope
343
Beam Factory of RIKEN. In the following, we present cases that are part of
344
the RIKEN Proposal for Scientific Program (PSP) related to the search for
345
first 2+ states in neutron-rich nuclei [37]:
346
at an incident energy of 250 MeV/nucleon. Those two reactions were per-
347
formed in May 2014 and May 2017, respectively, and their data analysis
348
is ongoing. The simulations have been made in the same framework (see
349
Section 4) as the one benchmarked on HIMAC data in the present work.
350
The total efficiency of a (p, 2p) reaction to be analyzed can be decomposed
351
in three parts: the geometrical efficiency, the detection efficiency, and the
352
software efficiency. (i) The geometrical efficiency is given by the position and
353
size of the TPC as well as target. In other words, this is the angular coverage
354
by the TPC for which protons enter the detection area. (ii) The detection
355
efficiency is the probability for the protons crossing the TPC to actually
356
create a signal detected above the noise. We consider for this TPC a 100%
357
detection efficiency as for all tracks in (p, 2p) reactions we have completely 24
79
Cu(p, 2p)78 Ni and
53
K(p, 2p)52 Ar
358
continuous tracks without loss of signal. (iii) The software efficiency is then
359
determined as the efficiency of the tracking algorithm to reconstruct (p, 2p)
360
events, taking also into account the efficiency to detect only one of the two
361
tracks.
362
In the case of the spectroscopy of
363
at 250 MeV/nucleon on a 150 mm thick LH2 target is considered and 400
364
000 events are generated by GEANT4 simulations. We consider that (p, 2p)
365
reactions can be identified as it is usually the case in experiments by use of
366
a spectrometer downstream the target, either SAMURAI or the ZeroDegree
367
spectrometer at the RIBF. In the TPC, the gas ionization produced by proton
368
tracks is simulated as well as the electron drift towards the detection plane.
369
The algorithm described in Section 3 is used to extract the two proton events
370
and measure the total efficiency of the detector and the vertex resolution in
371
the beam direction. (p, 2p) events for which only one proton is detected in the
372
TPC are also treated and the vertex is determined from the intersection of the
373
found track and the beam trajectory from beam trackers. Out of all (p, 2p)
374
simulated reactions, 92(2)% were reconstructed with the successful tracking
375
of one (15(2)%) or two (77(2)%) protons in the TPC, demonstrating the
376
large efficiency provided by the MINOS design. The rare unidentified (p, 2p)
377
events correspond to a proton scattered at small angle and a second proton
378
scattered at low energy which does not reach the TPC, in very forward angles
379
of the (p, 2p) kinematics. Results for the efficiency are summarized in Tab. 2.
380
The reconstruction of the vertex are compared to the real position as shown
381
in Fig. 12. A vertex position resolution along the beam axis of 4.3(1) mm
382
at full-width-half-maximum is obtained.
52
Ar via (p, 2p) knockout, a beam of
25
53
K
53
K(p, 2p)52 Ar
79
Cu(p, 2p)78 Ni
Beam energy
250 MeV/nucleon
250 MeV/nucleon
Target length
150 mm
100 mm
(p, 2p) events simulated
2568 (100%)
2349 (100%)
(p, 2p) events in the TPC
2473 (96(2)%)
2295 (98(2)%)
(p, 2p) events analyzed
2361 (92(2)%)
2195 (93(2)%)
2 protons analyzed
1982 (77(2)%)
1787 (76(2)%)
1 proton analyzed
379 (15(2)%)
408 (17(2)%)
FWHM vertex position resolution
4.3(1) mm
4.5(1) mm
Table 2: Simulation results for the
53
K(p, 2p)52 Ar and
79
Cu(p, 2p)78 Ni physics cases.
Detected events correspond to (p, 2p) events with charge depositions in the TPC. Analyzed events correspond to events fully treated by the tracking algorithm and leading to a vertex position. Here the vertex position resolution is obtained from two proton tracks analyzed in the TPC, but it can also be determined with only one proton in the TPC by taking the beam position with a resolution defined by the position of the beam trackers and their resolution. The effect of the beam tracking resolution is determined in the simulations section of Ref. [1]. 383
79
Cu(p, 2p)78 Ni, a beam of
79
384
In the case of
385
ing on a 100-mm thick LH2 target is considered. A geometrical efficiency for
386
the detection of one or two protons of 98(2)% is found, with an efficiency
387
of 76(2)% to detect the two protons from (p, 2p) reactions. This detection
388
efficiency is comparable to the previous case with a slightly larger geomet-
389
rical efficiency understandable from the smaller target length. The results
390
are summarized in Table 2. The vertex position resolution is estimated to be 26
Cu at 250 MeV/nucleon imping-
Figure 12: Difference in millimeter between the vertex reconstruction and the real position of the (p, 2p) reaction along the beam direction for the simulation of (left) 53 K (p, 2p) 52 Ar and (right) 79 Cu (p, 2p) 78 Ni at a beam incident energy of 250 MeV/nucleon for both cases and with a 150 mm and 100 mm thick LH2 target, respectively. In both cases, a fit of the distributions is shown in red. 391
4.5(1) mm full-width-half-maximum along the beam direction, as illustrated
392
in Fig. 12.
393
394
The simulated total MINOS efficiency can be confirmed directly from data.
395
We used data from the
396
with a 150-mm long hydrogen target and DALI2 [3] at the RIBF during
397
one experimental campaign [38] in 2017. The 2+ 1 state of
398
to lie at 1178(18) keV [39] from a previous measurement. The statistics in
399
+ the 2+ 1 → 01 photopeak of the gamma spectrum measured by DALI2 for
400
events with and without the coincidence of a vertex reconstructed in MINOS
401
TPC were compared. Both spectra were Doppler corrected without using
51
K (p, 2p)
50
Ar reaction measured using MINOS
27
50
Ar is known
402
the vertex information. Therefore, their line shape is the same. Considering
403
the Doppler-corrected photopeak events ensures that the γ rays indeed come
404
from reactions inside the target. Fig. 13 illustrates the ratio of counts in
405
these two gamma spectra with an energy bin of 40 keV after background
406
subtraction. The mean value is 90(5)%. The quoted uncertainty mainly arose
407
from background subtraction and limited statistics. A detection efficiency
408
for tracking the two protons was found to be 80(6)% and only one proton
409
11(3)%. We also extracted the MINOS efficiency from the
410
channel in another experimental campaign in 2015 [4] using a 100-mm long
411
hydrogen target. A total MINOS efficiency of 92(1)% was found, with 75(1)%
412
efficiency to track the two protons and 17(2)% to reconstruct only one proton,
413
in agreement with the simulations. Note that for this last measurement we
414
only considered the statistical uncertainties in the errors.
415
6. Conclusion and perspectives
69
Co (p, 2p)
68
Fe
416
We reported on the in-beam validation of the MINOS cylindrical time
417
projection chamber by use of intermediate-energy heavy-ion beams at the
418
HIMAC facility. A set of two CH2 targets were inserted inside the TPC. Ne
419
beams at 200 and 350 MeV/nucleon and a He beam at 200 MeV/nucleon were
420
used for the measurement. The TPC performances were tested at different
421
drift fields and the drift velocities obtained from the measurements show a
422
very good agreement with Magboltz simulations in the domain of interest.
423
This benchmarks our method to monitor the changes in drift velocity due to
424
oxygen and water impurities.
425
Dedicated algorithms and software have been developed for track reconstruc28
Figure 13: The total MINOS efficiency as a function of the γ-ray Doppler corrected energy with respect to the photopeak position of (p, 2p)
50
50
Ar. It was extracted using the data from
51
K
Ar. The square points show the ratio of counts for an energy bin of 40 keV, and
the red solid line shows the mean value of these square points.
426
tion based on the Hough transform. A vertex resolution of 5-7 mm FWHM
427
has been found for (p, 2p)-like reactions that includes interactions on both
428
carbon and hydrogen of the target. Monte-Carlo simulations have been per-
429
formed and show a good agreement with experiment. Furthermore, the pad
430
multiplicity is correctly reproduced, which shows our correct simulation of
431
the electron drift and diffusion in the TPC gas.
432
To benchmark further the tracking algorithm in terms of efficiency, we have
433
performed simulations of physics cases in realistic RIBF conditions. The TPC
434
demonstrates a 100% detection efficiency of protons in the MINOS energy
435
domain. The total efficiency of MINOS to reconstruct a reaction vertex after
436
a (p, 2p) reaction is shown to be ∼92%, ∼16% and ∼76% from 1p and 2p
437
events respectively, mostly from the not totally complete angular coverage.
29
438
A vertex position reconstruction better than 5 mm FWHM is found. The
439
MINOS efficiency has also been benchmarked with the experimental results
440
of
441
+ the 2+ 1 → 01 photopeak in the Doppler-corrected gamma spectra with and
442
without the vertex reconstruction in the MINOS TPC.
443
To increase the vertex resolution, one could think of software improvements
444
on the tracking algorithm using for example a 3D Hough transform [40] or
445
hardware modifications of the setup to minimize the angular straggling of
446
the protons.
447
MINOS is a very efficient detector for gamma-ray spectroscopy of exotic nu-
448
clei thanks to thick-target induced reactions reconstructed with either the
449
same or better vertex resolution as standard thin targets. Concerning the
450
gamma-ray spectroscopy at the RIBF, the resolution limitation comes from
451
the DALI2 scintillator array. The combination of higher resolution gamma-
452
ray arrays with MINOS would open new possibilities. In particular, the
453
MINOS concept is expected to bring a large gain in sensitivity to in-beam
454
gamma experiment with new generation Ge tracking arrays [29, 41].
51
K (p, 2p)
50
Ar and
69
Co (p, 2p)
68
Fe reactions by using the statistics in
455
456
This work has been funded by the European Research Council through the ERC
457
Starting Grant MINOS-258567. C. Santamaria has been supported by the IPA
458
program at the RIKEN Nishina Center. A. Obertelli thanks the Japanese Society
459
for the Promotion of Science (JSPS) for its support.
460
[1] A. Obertelli et al., Eur. Phys. J. A 50, 8 (2014).
461
[2] http://minos.cea.fr.
30
462
463
[3] S. Takeuchi et al., Nucl. Instrum. Methods Phys. Res., Sect. A 763, 596 (2014).
464
[4] C. Santamaria et al., Phys. Rev. Lett. 115, 192501 (2015).
465
[5] N. Paul et al., Phys. Rev. Lett. 118, 032501 (2017).
466
[6] S. Chen et al., Phys. Rev. C 95, 041302 (R) (2017).
467
[7] F. Flavigny et al., Phys. Rev. Lett. 118, 242501 (2017).
468
[8] M. Lettmann et al., Phys. Rev. C 96, 011301 (2017).
469
[9] C. Shand et al., Phys. Lett. B 773, 492 (2017).
470
[10] L. Olivier et al., Phys. Rev. Lett. 119, 192501 (2017).
471
[11] T. Kobayashi et al., Nucl. Instrum. Methods Phys. Res., Sect. B 317,
472
473
474
294 (2013). [12] T. Nakamura, Y. Kondo, Nucl. Instrum. Methods Phys. Res., Sect. B 376, 156 (2016).
475
[13] A. Obertelli and T. Uesaka, Eur. Phys. J. A 47, 105 (2011).
476
[14] C. Louchart et al., Nucl. Instrum. Methods Phys. Res., Sect. A 736, 81
477
478
479
480
481
(2014). [15] I. Giomataris et al., Nucl. Instrum. Methods Phys. Res., Sect. A 376, 29 (1996). [16] I. Giomataris et al., Nucl. Instrum. Methods Phys. Res., Sect. A 560, 405 (2006). 31
482
[17] D. Calvet et al., IEEE Trans. Nucl. Sci. 61, 675-682 (2014).
483
[18] N. Abgrall et al., Nucl. Instrum. Methods Phys. Res., Sect. A 637, 25
484
485
486
(2011). [19] S. Anvar et al., in IEEE Nuclear Science Symposium and Medical Imaging Conference (NSS/MIC) 736, 745-749 (2011).
487
[20] GET (acronym for General Electronics for TPCs) is a joint project be-
488
tween CEA-IRFU, CENBG, GANIL (France) and NSCL (US) labora-
489
tories. The project has been funded by the French funding agency ANR
490
and the DOE (US). The spokesperson is E.C. Pollacco from CEA Saclay.
491
[21] E. Pollacco et al., Phys. Procedia 37, 1799 (2012).
492
[22] P. Baron et al., IEEE Transactions on Nuclear Science, vol. 64, pp.
493
494
495
496
497
498
499
1494-1500,(2017). [23] S. Anvar, IEEE Nuclear Science Symposium Conference, pp. 3558-3561 (2008). [24] P. Baron et al., IEEE Transactions on Nuclear Science, vol. 55 N◦ 33, pp. 1744-1752 (2008). [25] T. Furukawa et al., Nucl. Instrum. Methods Phys. Res., Sect. A 562, 1050-1053 (2006).
500
[26] S. Ota et al., J. Radioanal. Nucl. Chem. 305, 907-911 (2015).
501
[27] S.F. Biagi, Nucl. Instrum. Methods Phys. Res., Sect. A 241, 234 (1999).
32
502
503
504
505
506
507
[28] C. H¨oppner et al., Nucl. Instrum. Methods Phys. Res., Sect. A 620, 518 (2010). [29] S. Akkoyun et al., Nucl. Instrum. Methods Phys. Res., Sect. A 668, 26 (2012). [30] P.V.C. Hough, Methods and means for recognizing complex patterns, U.S. Patent 3.069.654 (December 1962).
508
[31] R.O. Duda and P.E. Hart, Commun. ACM 15, 11-15 (1972).
509
[32] D.H. Ballard et al., Pattern Recognition 13, 111-122 (1981).
510
[33] J. Giovinazzo et al., Nucl. Instrum. Methods Phys. Res., Sect. A 840,
511
512
513
15 (2016). [34] S. Agostinelli et al., Nucl. Instrum. Methods Phys. Res., Sect. A 506, 250 (2003).
514
[35] A. Boudard et al., Phys. Rev. C 66, 044615 (2002).
515
[36] A. Boudard et al., Phys. Rev. C 87, 014606 (2013).
516
[37] P. Doornenbal and A. Obertelli, RIKEN proposal for scientific program:
517
Shell evolution and search for two-plus states at the RIBF (SEASTAR),
518
2013 (unpublished).
519
520
521
[38] P. Doornenbal, A. Obertelli et al., experiment NP15-SAMURAI38 (2017). [39] D. Steppenbeck et al., Phys. Rev. Lett. 114, 252501 (2015).
33
522
523
524
525
[40] C. Dalitz, T. Schramke, M. Jeltsch, Image Processing On Line 7, 184196 (2017). [41] S. Paschalis et al., Nucl. Instrum. Methods Phys. Res., Sect. A 709, 44 (2013).
34