- Email: [email protected]
Development of a Time Projection Chamber using CF4 gas for relativistic heavy ion experiments
ARTICLE IN PRESS
Nuclear Instruments and Methods in Physics Research A 564 (2006) 190–196 www.elsevier.com/locate/nima
Development of a Time Projection Chamber using CF4 gas for relativistic heavy ion experiments T. Isobea,, H. Hamagakia, K. Ozawaa, M. Inuzukaa,1, T. Sakaguchia,2, T. Matsumotoa, S. Kametania, F. Kajiharaa, T. Gunjia, N. Kuriharaa, S.X. Odaa, Y.L. Yamaguchib,3 a
Center for Nuclear Study, Graduate School of Science, University of Tokyo, 7-3-1 Hongo, Bunkyo, Tokyo 113-0033, Japan b Waseda University, 3-4-1 Okubo, Shinjuku-ku, Tokyo 169-8555, Japan Received 20 December 2004; received in revised form 16 February 2006; accepted 6 April 2006 Available online 12 May 2006
Abstract A prototype Time Projection Chamber (TPC) using pure CF4 gas was developed for possible use in heavy ion experiments. Basic characteristics such as gain, drift velocity, longitudinal diffusion and attenuation length of produced electrons were measured with the TPC. At an electric field of 900 V/cm, the drift velocity and longitudinal diffusion for 1 cm drift were obtained as 10 cm=ms and 60 mm, respectively. The relatively large gain fluctuation is explained to be due to the electron attachment process in CF4. These characteristics are encouraging for the measurement of the charged particle trajectories under high multiplicity conditions at RHIC. r 2006 Elsevier B.V. All rights reserved. PACS: 29.40.Cs Keywords: CF4; TPC; Gain; Drift velocity; Longitudinal diffusion
1. Introduction It is predicted from lattice QCD calculations that a phase transition from hadronic matter to a plasma of deconfined quarks and gluons may occur at high temperature and energy density. Such high energy density is expected to be created in heavy ion collisions produced at the Relativistic Heavy Ion Collider (RHIC) at Brookhaven National Laboratory (BNL). One of the most intriguing findings at RHIC is the suppression of mesons in the high transverse momentum region in central Au þ Au collisions [1,2]. The observed suppression is interpreted as an energy loss of Corresponding author. Tel.: +81 48 464 4156; fax: +81 48 464 4554.
E-mail address: [email protected] (T. Isobe). Current address: National Research Institute for Cultural Properties, Tokyo, 13-43 Ueno Park, Taito-ku, Tokyo 110-8713, Japan. 2 Current address: Brookhaven National Laboratory, Upton, NY 11973-5000, USA. 3 Current address: Center for Nuclear Study, Graduate School of Science, University of Tokyo, 7-3-1 Hongo, Bunkyo, Tokyo 113-0033, Japan. 1
0168-9002/$ - see front matter r 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.nima.2006.04.051
initially hard-scattered partons traversing the hot and dense matter. In order to study the energy loss quantitatively in the hot and dense matter, it is important to obtain detailed information of the produced partons. Such information is provided by the reconstruction of jets using charged particle tracks. For the reconstruction of momenta of all charged particles originating from jets, a tracking device with a large solid angle is needed. With such a tracking device, the energy loss of hard scattered partons in the dense matter can be studied through measurements of two jets or g-jet correlations. One of the most promising devices for such measurements is a vertex spectrometer in the form of a Time Projection Chamber (TPC) [6] with full coverage in azimuth. The hard-scattered partons can be reconstructed from secondary measured particles. A simulation study with PYTHIA [3] shows that on average, given an acceptance of jZjp 1 in pseudo-rapidity, the TPC can reconstruct 87% of the primary parton momentum. Such a vertex spectrometer has broad physics capabilities especially for the PHENIX experiment [4] at RHIC,
ARTICLE IN PRESS T. Isobe et al. / Nuclear Instruments and Methods in Physics Research A 564 (2006) 190–196
because the current PHENIX spectrometer has limited solid angle coverage as shown in Fig. 1. Also, such a vertex spectrometer can be helpful to measure particles of higher transverse momentum. With the current PHENIX tracking device, charged particles within a momentum range of 0.2–10 GeV=c can be measured [5]. It is, however, not possible to observe decay vertices, secondary interactions and conversions. The inability to detect these events results in increasing the background of false tracks at high transverse momenta. Thus, installation of a vertex spectrometer device would allow the PHENIX detector to reduce backgrounds and extend charged particle spectra to higher transverse momenta. High tracking capability in a high multiplicity environment is the most important feature of the TPC. In Au þ Au pffiffiffiffiffiffiffiffi central collisions at sNN ¼ 200 GeV, the average charged
PHENIX Detector PbSc
PC3 PC2
PC3 TEC
Central Magnet
PbSc
PbSc
PbSc DC
DC RICH
RICH BB
PbSc
PbSc
PbGl
PC1
PC1
191
particle multiplicity ðdN ch =dyÞ is 1000–1200 for the most central collisions, and the charged particle density is 0:26 cm2 at a radial distance of 20 cm from the beam line. To cope with such a high particle density, good twohit resolution as well as good position resolution are required. In order to reconstruct trajectories with an efficiency of 90% up to pT ¼ 10 GeV=c, the required hit occupancy in terms of space-time hits should be less than 0.1. High occupancy limits the two hit resolution in the wire direction and the drift direction. Two-hit resolution along an anode wire is estimated to be 1.2 cm governed by the distance between the anode wires and the readout pads (4 mm) of the prototype TPC. To achieve small hit occupancy, the two-hit resolution in the longitudinal direction should be less than a few millimeters. Thus, the longitudinal diffusion of electrons should be less than 100 mm for 1 cm drift. Here, a drift length of 50 cm is needed for the rapidity coverage. As a candidate drift gas, pure CF4 gas is chosen to satisfy the above requirements [7]. From a simulation study, CF4 was estimated toffi have a pffiffiffiffiffiffi longitudinal diffusion coefficient of 60 mm= cm at an electric field of 1 kV/cm. A prototype TPC was developed to measure the basic characteristics of CF4 such as drift velocity, the diffusion coefficient and attenuation length, and to study the overall performance of the TPC using pure CF4. In this article, the properties of CF4 with no magnetic field are reported. Studies with a magnetic field and further investigations of other known effects [8] are underway to realize the TPC using CF4.
PbGl
2. Development of a prototype TPC TOF
West
Beam View
The design view of the TPC is shown in Fig. 2. A gas vessel is made of aluminum, the dimension of which is 29 29 60 cm3 . Inside, a drift cage with the dimension of 16 16 36 cm3 is made of 116 multistage gold strips on a
East
Fig. 1. Schematic view of the PHENIX detector from beam side.
600 4
4
4
5
364
200
105
105
120
drift cage
grid frame anode frame cathode frame
top plate gas vessel
Fig. 2. Design view of prototype TPC. The dimension of the gas vessel is 29 29 60 cm3 . The drift cage with dimensions of 16 16 36 cm3 made of gold strips on G10 board is installed in the gas vessel. The large HV connector shown at the right is used for the field cage power supply. The left end cap shows the MWPC type readout.
ARTICLE IN PRESS T. Isobe et al. / Nuclear Instruments and Methods in Physics Research A 564 (2006) 190–196
192
G10 board. Strips are connected to each other through a ladder of 1 MO resistors to make a uniform electric field. Inhomogeneities of the electric field ðE transverse =E longitudinal Þ are estimated to be less than 103 in the fiducial volume at a radial distance of 3 cm from the field strips. The drift cage can hold up to 900 V/cm, sufficient to limit diffusion to less than 100 mm for 1 cm drift in CF4. The purity of CF4 is 99.999%. The flow rate of the gas to the chamber was adjusted to 200 cm3 = min at atmospheric pressure. Electro-negative pollutants are minimized as much as possible with the current gas system. Gas contamination with water and oxygen is 100 and 30 ppm, respectively. A multi-wire proportional chamber (MWPC) with cathode pads is used for the TPC readout. As shown in Fig. 3, the MWPC consists of three layers, i.e. grid wires, anode wires and cathode pads from top to bottom. The spacing between layers is 4 mm. The grid wire and anode wire pitch is 3.5 mm. The diameters of the anode wires and grid wires are 20 and 100 mm, respectively. The induced charge on the cathode pads of 13 mm 13 mm is read out with the pad size large enough to collect sufficient charge induced from an anode wire. The output signal is converted to a voltage signal by a charge-sensitive pre-amplifier with a 325 MHz op-amp (AD8058 by Analog Devices). The pre-amplifier has a gain of 1 V/pC with 1 pF feedback capacitance, and a slew rate of 1000 V=ms. The rise time of a MIP signal is expected to be 40 ns for CF4. Typical pulse height of a signal is 200 mV. The slew rate is therefore sufficient to handle the signals, and to assure fine two-hit resolution. The preamplifier is designed for differential output signals to reduce common noise in a cable. The analog signal is digitized by a flash analog-to-digital converter (FADC, RPV-160 by REPIC Co., Ltd), having 8 bit dynamic range with a scale of 0–1 V and a sampling rate of 100 MHz. For the gain measurement, a CAMAC ADC module (RPC-022 by REPIC Co., Ltd) is used allowing finer resolution. The dynamic range of this ADC is 12 bits, and full scale corresponds to 1000 pC.
3. Gain 3.1. Setup Gain values under several conditions were measured in the prototype TPC using 5.9 keV X-rays from 55Fe. A schematic view of the setup for gain measurements is shown in Fig. 4. The average energy required to produce one electron–ion pair in CF4 is 54 eV, and one X-ray on average generates 110 electrons through photoelectric absorption. 3.2. Test result Typical induced charge spectra obtained with CF4 along with P10 (Ar(90%)–CH4(10%)) as a reference are shown in Fig. 5. These charge spectra show clear peaks of 5.9 keV Xrays over background. Since only one pad is used to read out the charge, often only a fraction of the produced charge is collected by the pad due to the varying range of the 5.9 keV photons before conversion. Then only a fraction of the total charge from X-rays is present as the background. This phenomenon also explains the absence of the so-called argon-escape peak at 3 keV X-ray energy in the charge spectrum for P10. 3.2.1. Gain Relative gain values with respect to the calculated input primary charge were determined as a function of anode voltage. The charge spectra were fitted with a Gaussian and an exponential function (background). The amount of induced charge is obtained as the mean value of the fitted Gaussian function. The relative gain is obtained using the following relation: Gain ¼
Qinduced
charge
C anode2detector =C cathode2anode e 5:9 keV=W
(1)
where W stands for the average energy to generate an electron–ion pair, e is unit charge, Qinduced charge is the measured charge, C anode2detector is the capacitance between
55
Fe
3.5 mm
100 µm grid(0 V) 20 µm anode(up to 4 kV)
4 mm
13 mm square cathode pads
4 mm
0V
4 mm 4 mm
anode HV (up to +4kV)
4 mm
4 mm
Pre-amp
Fig. 3. Schematic view of the TPC readout. The end cap wire chamber consists of three layers, i.e. grid wires, anode wires and a cathode pad plane, from top to bottom with 3.5 mm wire spacing and 4 mm gaps.
Fig. 4. Cross-sectional view of the setup for the gain measurement.
ARTICLE IN PRESS T. Isobe et al. / Nuclear Instruments and Methods in Physics Research A 564 (2006) 190–196
CF4 anode HV: 3.5 kV
140
193
P10 anode HV: 2.0 kV
500
120 400 Count
Count
100 80 σ=32%
60
300 σ=10%
200
40 100
20 0
200
400
600 800 Charge (fC)
1000
1200
0
0
100 200 300 400 500 600 700 800 Charge (fC)
Fig. 5. Induced charge spectra in CF4 at anode HV of 3.5 kV (left) and in P10 at anode HV of 2.0 kV (right).
5
10
normalized number of electron
Gain (A.U.)
10
4
10
CF4 Gain P10 Gain 3
10
1.8
2
2.2
2.4 2.6 2.8 3 3.2 anode voltage (kV)
3.4
10
10
4
3
P10 (anode HV: 2150V) CF4 (anode HV: 3400V) 2
10
1
3.6 10
-1
10
Fig. 6. Relative gain values to calculated input primary charge as a function of anode voltage obtained with P10 and CF4.
-3
10
-2
10
-1
distance from wire [cm] Fig. 7. Calculated gain value as a function of distance from the anode wire for P10 and CF4.
the anode wire and the whole detector, and C cathode2anode is the capacitance between cathode pad and anode wire. The values of C anode2detector and C cathode2anode are calculated to be 0.9 and 0.4 pF, respectively. Then, it is expected that 44% of the induced charge is detected with one pad. The results of gain measurements are shown in Fig. 6. To get a clear signal in this configuration, a gain of 104 is needed for CF4 gas, which is achieved at an anode voltage of þ3:1 kV. In order to make comparable amplification as in P10, CF4 needs a 1.5 times higher electric field.
3.2.2. Gain fluctuation The gain fluctuation in CF4 is evaluated to be 32% at an anode voltage of 3.4 kV. The gain fluctuation in P10 is 10% at the same gain. While the fluctuation of P10 is consistent with the statistic error from the number of drift electrons, the fluctuation of CF4 is significantly larger than expected from statistics alone. This is most probably due to the absorption of electrons via dissociative attachment in the high field region near the anode wires [9,10]. These attachment processes in CF4 are: CF4 þ e ! F þ CF3
(2)
CF4 þ e ! F 2 þ CF2
(3)
CF4 þ e ! F þ CF 3.
(4)
The electron impact cross-sections for CF4 were evaluated for electron energies in the range of 0–1000 eV [11], which include attachment cross-sections derived from absorption processes. As well as the avalanche coefficient, the attachment coefficient is obtained from cross-sections of CF4 using the simulation program Magboltz [12]. As shown in Fig. 7, the number of electrons at each distance from the wire is calculated from the attachment and avalanche coefficients with the following relation: Z 1 number of electrons ¼ N 0 exp ðaðxÞ þ bðxÞÞ dx (5) r
where N 0 is the number of drift electrons, aðxÞ and bðxÞ stand for the avalanche and attachment coefficient, respectively. After a mean free path bðxÞ1 , a drift electron will be absorbed. The number of electrons is estimated to 1 become 10 before amplification. Thus, pffiffiffiffiffi the statistical fluctuations increase by a factor of 10. On the other
ARTICLE IN PRESS T. Isobe et al. / Nuclear Instruments and Methods in Physics Research A 564 (2006) 190–196
194
hand, if the drift electrons were not absorbed, the error would not become larger than in the case of P10.
ADC = a*TDC+b
fit line
100
4. Drift velocity and longitudinal diffusion 80
4.2. Test result The arrival time of a signal, t0 , is obtained with the following relation: ADCpedestal b (6) a where a and b are parameters of the fitted line as shown in Fig. 9, ADCpedestal being the pedestal value of the ADC. The drift time and the longitudinal diffusion were derived from the mean value and the standard deviation of the t0 -distribution after fitting it with a Gaussian function. t0 ¼
VH(~30kV) Drift region 1MΩ
N2 laser 337nm
ADC 40
5%
t2
20
pedestal
0 100 110 120 130 140 150 160 170 180 190
t0
TDC
Fig. 9. Typical shape of a laser signal. A t0 , start time of signal, is obtained from the fitted line between t1 (ADC ¼ 70%) and t2 (ADC ¼ 5%).
10 8 P10 measurement data P10 simulation data
6
CF4 measurement data CF4 simulation data
4 2 0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
electric field (kV/cm/atm) Fig. 10. The drift velocity as a function of the electric field obtained in P10 and CF4.
The drift velocity is obtained using the following relation,
Gas Vessel
f = 200
60
drift velocity (cm/µs)
The drift velocity and the longitudinal diffusion of electrons were measured using a nitrogen laser (YKN500 by Usho Co.). The experimental setup is shown in Fig. 8. A laser beam is reflected by two mirrors into a stationary drift cell through a quartz lens of 20 cm focal length. The focal point is in the middle of the active drift volume. For laser injection, there are two quartz windows which are located at a distance of 34.0 and 18.2 cm from the readout pads. A portion of the laser beam is reflected to a photo-diode (S1722-03: Si pin photo-diode by HAMAMATSU Co.) for start timing. The wave length and the energy of the laser beam are 337.1 nm and 2.5 mJ, respectively. This laser can generate 102 electrons around the beam focal point in CF4 by multiple photon absorption. By attenuating the laser beam, the number of generated electrons is reduced to the level of a single electron to measure drift velocity and diffusion.
70%
t1
4.1. Setup
mirror
0V
anode V Optical Trigger (photo diode)
Fig. 8. Experimental setup for the measurements of drift velocity, longitudinal diffusion and attenuation length.
vdrift ¼
z1 z2 t0 ðz1 Þ t0 ðz2 Þ
(7)
where z1 and z2 are the distances between the readout pads and the laser position, and t0 ðz1 Þ and t0 ðz2 Þ are the corresponding arrival times of the drift electrons. Results are shown in Fig. 10. The drift velocity in CF4 is 10 cm=ms, when the electric field is 900 V/cm. This high drift velocity of CF4 comes from the large Ramsauer dip in the elastic cross-section [12]. From the fluctuation (r.m.s.) of arrival times, the longitudinal diffusion coefficients of CF4 and P10 are obtained and shown in Fig. 11. Lines through the data points represent calculations using the Magboltz simulation program. Magboltz solves the Boltzmann transport equations to calculate transport properties of electrons in gas mixtures. The result of P10 is consistent with the
ARTICLE IN PRESS T. Isobe et al. / Nuclear Instruments and Methods in Physics Research A 564 (2006) 190–196
195
3
longitudinal diffusion P10 measurement data P10 simulation data CF4 measurement data CF4 simulation data
10 3 Attenuation Length (cm)
diffusion for 1cm (µm)
10
2
10
0
0.1
0.2
0.3 0.4 0.5 0.6 0.7 electric field (kV/cm/atm)
0.8
10 2
100
0.9
200
300
400 500 H2O (ppm)
600
700
800
Fig. 11. The longitudinal diffusion as a function of the electric field obtained for P10 and CF4. Lines are obtained from the Magboltz simulation [12].
Fig. 12. Attenuation length as a function of the quantity of water in the gas vessel. ðE ¼ 735 V=cmÞ The fraction of oxygen in this measurement was 30% of the water contents.
simulation. The longitudinal diffusion of CF4 is 60 mm for 1 cm drift, when the electric field is 900 V/cm. Thus, two electron clusters separated by 1.3 mm in longitudinal direction can be separated within 3-sigma even after 50 cm of drift. This is sufficient for operation under high multiplicity conditions in RHIC. The diffusion in CF4 is smaller than in CO2, for example, a popular chamber gas pffiffiffiffiffiffiffi with a small diffusion coefficient of 80 mm= cm at an electric field of 900 V/cm [13].
The results of the attenuation length measurements as a function of the contamination of water are shown in Fig. 12. If the contamination of water is less than 250 ppm, the attenuation length is more than 2 m, and more than 80% of the electrons will survive after 50 cm of drift in CF4. The solid line shown in Fig. 12 is a fitted line to obtain the attachment coefficient for electro-negative pollutants in CF4. Water does not contribute as much to attachment as oxygen [14]. The fraction of oxygen in this system was 30% of the water contents.
5. Attenuation length 6. Summary 5.1. Setup The attenuation length was measured in order to evaluate how much electro-negative contamination can be tolerated for adequate TPC operation. The attenuation length is obtained for CF4 by comparing the collected charge at two different laser beam entry points. The setup is the same as that for the drift velocity measurement. Water contamination is controlled by changing the flow rate of the gas to the chamber and is monitored by a moisture monitor (35-IS by PANAMETRICS Co., Ltd). The applied electric field is 735 V/cm. 5.2. Test result The attenuation length is obtained with the following relation: ðz1 z2 Þ attenuation length ¼ lnðQðz1 Þ=Qðz2 ÞÞ
(8)
where z1 and z2 are the distances between the readout pads and the laser beam position, and Qðz1 Þ and Qðz2 Þ are the collected charge at z1 and z2 . The amount of charge is obtained by fitting a charge spectrum with a Gaussian function.
A prototype Time Projection Chamber (TPC) using pure CF4 gas was developed for possible use in heavy ion experiments. Basic characteristics such as gain, drift velocity, longitudinal diffusion and attenuation length of electrons in CF4 were measured with this TPC. At an electric field of 900 V/cm, the drift velocity and longitudinal pffiffiffiffiffiffiffi diffusion were obtained as 10 cm=ms and 60 mm= cm, respectively. The relatively large gain fluctuation is explained to be due to the electron attachment process in CF4. Eighty percent of electrons will survive after 50 cm of drift with a water contamination of 250 ppm in CF4. These characteristics are encouraging for the measurement of the charged particle trajectories, using a TPC at RHIC. Acknowledgment Authors would like to thank C.L. Woody for his comments on the paper. References [1] [2] [3] [4]
S.S. Adler, et al., Phys. Rev. Lett. 91 (2003) 072301. C. Adler, et al., Phys. Rev. Lett. 90 (2003) 082302. T. Sjo¨strand, et al., Computer Phys. Commun. 135 (2001) 238. K. Adcox, et al., Nucl. Instr. and Meth. A 499 (2003) 469.
ARTICLE IN PRESS 196 [5] [6] [7] [8] [9]
T. Isobe et al. / Nuclear Instruments and Methods in Physics Research A 564 (2006) 190–196 K. Adcox, et al., Nucl. Instr. and Meth. A 499 (2003) 489. D.R. Nygren, Phys. Scripta 23 (1981) 584. J. Va’vra, et al., Nucl. Instr. and Meth. A 324 (1993) 113. T.C. Meyer, The Science and Culture Series—Physics, 2003, p. 180. L.G. Christophorou, et al., Nucl. Instr. and Meth. 163 (1979) 141.
[10] [11] [12] [13] [14]
R. Thun, Nucl. Instr. and Meth. A 273 (1988) 157. R.A. Bonham, Jpn. J. Appl. Phys. 33 (1994) 4157. S.F. Biagi, Nucl. Instr. and Meth. A 421 (1999) 234. V. Palladino, B. Sadoulet, Nucl. Instr. and Meth. 128 (1975) 323. M. Huk, et al., Nucl. Instr. and Meth. A 267 (1988) 107.
Nuclear Instruments and Methods in Physics Research A 564 (2006) 190–196 www.elsevier.com/locate/nima
Development of a Time Projection Chamber using CF4 gas for relativistic heavy ion experiments T. Isobea,, H. Hamagakia, K. Ozawaa, M. Inuzukaa,1, T. Sakaguchia,2, T. Matsumotoa, S. Kametania, F. Kajiharaa, T. Gunjia, N. Kuriharaa, S.X. Odaa, Y.L. Yamaguchib,3 a
Center for Nuclear Study, Graduate School of Science, University of Tokyo, 7-3-1 Hongo, Bunkyo, Tokyo 113-0033, Japan b Waseda University, 3-4-1 Okubo, Shinjuku-ku, Tokyo 169-8555, Japan Received 20 December 2004; received in revised form 16 February 2006; accepted 6 April 2006 Available online 12 May 2006
Abstract A prototype Time Projection Chamber (TPC) using pure CF4 gas was developed for possible use in heavy ion experiments. Basic characteristics such as gain, drift velocity, longitudinal diffusion and attenuation length of produced electrons were measured with the TPC. At an electric field of 900 V/cm, the drift velocity and longitudinal diffusion for 1 cm drift were obtained as 10 cm=ms and 60 mm, respectively. The relatively large gain fluctuation is explained to be due to the electron attachment process in CF4. These characteristics are encouraging for the measurement of the charged particle trajectories under high multiplicity conditions at RHIC. r 2006 Elsevier B.V. All rights reserved. PACS: 29.40.Cs Keywords: CF4; TPC; Gain; Drift velocity; Longitudinal diffusion
1. Introduction It is predicted from lattice QCD calculations that a phase transition from hadronic matter to a plasma of deconfined quarks and gluons may occur at high temperature and energy density. Such high energy density is expected to be created in heavy ion collisions produced at the Relativistic Heavy Ion Collider (RHIC) at Brookhaven National Laboratory (BNL). One of the most intriguing findings at RHIC is the suppression of mesons in the high transverse momentum region in central Au þ Au collisions [1,2]. The observed suppression is interpreted as an energy loss of Corresponding author. Tel.: +81 48 464 4156; fax: +81 48 464 4554.
E-mail address: [email protected] (T. Isobe). Current address: National Research Institute for Cultural Properties, Tokyo, 13-43 Ueno Park, Taito-ku, Tokyo 110-8713, Japan. 2 Current address: Brookhaven National Laboratory, Upton, NY 11973-5000, USA. 3 Current address: Center for Nuclear Study, Graduate School of Science, University of Tokyo, 7-3-1 Hongo, Bunkyo, Tokyo 113-0033, Japan. 1
0168-9002/$ - see front matter r 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.nima.2006.04.051
initially hard-scattered partons traversing the hot and dense matter. In order to study the energy loss quantitatively in the hot and dense matter, it is important to obtain detailed information of the produced partons. Such information is provided by the reconstruction of jets using charged particle tracks. For the reconstruction of momenta of all charged particles originating from jets, a tracking device with a large solid angle is needed. With such a tracking device, the energy loss of hard scattered partons in the dense matter can be studied through measurements of two jets or g-jet correlations. One of the most promising devices for such measurements is a vertex spectrometer in the form of a Time Projection Chamber (TPC) [6] with full coverage in azimuth. The hard-scattered partons can be reconstructed from secondary measured particles. A simulation study with PYTHIA [3] shows that on average, given an acceptance of jZjp 1 in pseudo-rapidity, the TPC can reconstruct 87% of the primary parton momentum. Such a vertex spectrometer has broad physics capabilities especially for the PHENIX experiment [4] at RHIC,
ARTICLE IN PRESS T. Isobe et al. / Nuclear Instruments and Methods in Physics Research A 564 (2006) 190–196
because the current PHENIX spectrometer has limited solid angle coverage as shown in Fig. 1. Also, such a vertex spectrometer can be helpful to measure particles of higher transverse momentum. With the current PHENIX tracking device, charged particles within a momentum range of 0.2–10 GeV=c can be measured [5]. It is, however, not possible to observe decay vertices, secondary interactions and conversions. The inability to detect these events results in increasing the background of false tracks at high transverse momenta. Thus, installation of a vertex spectrometer device would allow the PHENIX detector to reduce backgrounds and extend charged particle spectra to higher transverse momenta. High tracking capability in a high multiplicity environment is the most important feature of the TPC. In Au þ Au pffiffiffiffiffiffiffiffi central collisions at sNN ¼ 200 GeV, the average charged
PHENIX Detector PbSc
PC3 PC2
PC3 TEC
Central Magnet
PbSc
PbSc
PbSc DC
DC RICH
RICH BB
PbSc
PbSc
PbGl
PC1
PC1
191
particle multiplicity ðdN ch =dyÞ is 1000–1200 for the most central collisions, and the charged particle density is 0:26 cm2 at a radial distance of 20 cm from the beam line. To cope with such a high particle density, good twohit resolution as well as good position resolution are required. In order to reconstruct trajectories with an efficiency of 90% up to pT ¼ 10 GeV=c, the required hit occupancy in terms of space-time hits should be less than 0.1. High occupancy limits the two hit resolution in the wire direction and the drift direction. Two-hit resolution along an anode wire is estimated to be 1.2 cm governed by the distance between the anode wires and the readout pads (4 mm) of the prototype TPC. To achieve small hit occupancy, the two-hit resolution in the longitudinal direction should be less than a few millimeters. Thus, the longitudinal diffusion of electrons should be less than 100 mm for 1 cm drift. Here, a drift length of 50 cm is needed for the rapidity coverage. As a candidate drift gas, pure CF4 gas is chosen to satisfy the above requirements [7]. From a simulation study, CF4 was estimated toffi have a pffiffiffiffiffiffi longitudinal diffusion coefficient of 60 mm= cm at an electric field of 1 kV/cm. A prototype TPC was developed to measure the basic characteristics of CF4 such as drift velocity, the diffusion coefficient and attenuation length, and to study the overall performance of the TPC using pure CF4. In this article, the properties of CF4 with no magnetic field are reported. Studies with a magnetic field and further investigations of other known effects [8] are underway to realize the TPC using CF4.
PbGl
2. Development of a prototype TPC TOF
West
Beam View
The design view of the TPC is shown in Fig. 2. A gas vessel is made of aluminum, the dimension of which is 29 29 60 cm3 . Inside, a drift cage with the dimension of 16 16 36 cm3 is made of 116 multistage gold strips on a
East
Fig. 1. Schematic view of the PHENIX detector from beam side.
600 4
4
4
5
364
200
105
105
120
drift cage
grid frame anode frame cathode frame
top plate gas vessel
Fig. 2. Design view of prototype TPC. The dimension of the gas vessel is 29 29 60 cm3 . The drift cage with dimensions of 16 16 36 cm3 made of gold strips on G10 board is installed in the gas vessel. The large HV connector shown at the right is used for the field cage power supply. The left end cap shows the MWPC type readout.
ARTICLE IN PRESS T. Isobe et al. / Nuclear Instruments and Methods in Physics Research A 564 (2006) 190–196
192
G10 board. Strips are connected to each other through a ladder of 1 MO resistors to make a uniform electric field. Inhomogeneities of the electric field ðE transverse =E longitudinal Þ are estimated to be less than 103 in the fiducial volume at a radial distance of 3 cm from the field strips. The drift cage can hold up to 900 V/cm, sufficient to limit diffusion to less than 100 mm for 1 cm drift in CF4. The purity of CF4 is 99.999%. The flow rate of the gas to the chamber was adjusted to 200 cm3 = min at atmospheric pressure. Electro-negative pollutants are minimized as much as possible with the current gas system. Gas contamination with water and oxygen is 100 and 30 ppm, respectively. A multi-wire proportional chamber (MWPC) with cathode pads is used for the TPC readout. As shown in Fig. 3, the MWPC consists of three layers, i.e. grid wires, anode wires and cathode pads from top to bottom. The spacing between layers is 4 mm. The grid wire and anode wire pitch is 3.5 mm. The diameters of the anode wires and grid wires are 20 and 100 mm, respectively. The induced charge on the cathode pads of 13 mm 13 mm is read out with the pad size large enough to collect sufficient charge induced from an anode wire. The output signal is converted to a voltage signal by a charge-sensitive pre-amplifier with a 325 MHz op-amp (AD8058 by Analog Devices). The pre-amplifier has a gain of 1 V/pC with 1 pF feedback capacitance, and a slew rate of 1000 V=ms. The rise time of a MIP signal is expected to be 40 ns for CF4. Typical pulse height of a signal is 200 mV. The slew rate is therefore sufficient to handle the signals, and to assure fine two-hit resolution. The preamplifier is designed for differential output signals to reduce common noise in a cable. The analog signal is digitized by a flash analog-to-digital converter (FADC, RPV-160 by REPIC Co., Ltd), having 8 bit dynamic range with a scale of 0–1 V and a sampling rate of 100 MHz. For the gain measurement, a CAMAC ADC module (RPC-022 by REPIC Co., Ltd) is used allowing finer resolution. The dynamic range of this ADC is 12 bits, and full scale corresponds to 1000 pC.
3. Gain 3.1. Setup Gain values under several conditions were measured in the prototype TPC using 5.9 keV X-rays from 55Fe. A schematic view of the setup for gain measurements is shown in Fig. 4. The average energy required to produce one electron–ion pair in CF4 is 54 eV, and one X-ray on average generates 110 electrons through photoelectric absorption. 3.2. Test result Typical induced charge spectra obtained with CF4 along with P10 (Ar(90%)–CH4(10%)) as a reference are shown in Fig. 5. These charge spectra show clear peaks of 5.9 keV Xrays over background. Since only one pad is used to read out the charge, often only a fraction of the produced charge is collected by the pad due to the varying range of the 5.9 keV photons before conversion. Then only a fraction of the total charge from X-rays is present as the background. This phenomenon also explains the absence of the so-called argon-escape peak at 3 keV X-ray energy in the charge spectrum for P10. 3.2.1. Gain Relative gain values with respect to the calculated input primary charge were determined as a function of anode voltage. The charge spectra were fitted with a Gaussian and an exponential function (background). The amount of induced charge is obtained as the mean value of the fitted Gaussian function. The relative gain is obtained using the following relation: Gain ¼
Qinduced
charge
C anode2detector =C cathode2anode e 5:9 keV=W
(1)
where W stands for the average energy to generate an electron–ion pair, e is unit charge, Qinduced charge is the measured charge, C anode2detector is the capacitance between
55
Fe
3.5 mm
100 µm grid(0 V) 20 µm anode(up to 4 kV)
4 mm
13 mm square cathode pads
4 mm
0V
4 mm 4 mm
anode HV (up to +4kV)
4 mm
4 mm
Pre-amp
Fig. 3. Schematic view of the TPC readout. The end cap wire chamber consists of three layers, i.e. grid wires, anode wires and a cathode pad plane, from top to bottom with 3.5 mm wire spacing and 4 mm gaps.
Fig. 4. Cross-sectional view of the setup for the gain measurement.
ARTICLE IN PRESS T. Isobe et al. / Nuclear Instruments and Methods in Physics Research A 564 (2006) 190–196
CF4 anode HV: 3.5 kV
140
193
P10 anode HV: 2.0 kV
500
120 400 Count
Count
100 80 σ=32%
60
300 σ=10%
200
40 100
20 0
200
400
600 800 Charge (fC)
1000
1200
0
0
100 200 300 400 500 600 700 800 Charge (fC)
Fig. 5. Induced charge spectra in CF4 at anode HV of 3.5 kV (left) and in P10 at anode HV of 2.0 kV (right).
5
10
normalized number of electron
Gain (A.U.)
10
4
10
CF4 Gain P10 Gain 3
10
1.8
2
2.2
2.4 2.6 2.8 3 3.2 anode voltage (kV)
3.4
10
10
4
3
P10 (anode HV: 2150V) CF4 (anode HV: 3400V) 2
10
1
3.6 10
-1
10
Fig. 6. Relative gain values to calculated input primary charge as a function of anode voltage obtained with P10 and CF4.
-3
10
-2
10
-1
distance from wire [cm] Fig. 7. Calculated gain value as a function of distance from the anode wire for P10 and CF4.
the anode wire and the whole detector, and C cathode2anode is the capacitance between cathode pad and anode wire. The values of C anode2detector and C cathode2anode are calculated to be 0.9 and 0.4 pF, respectively. Then, it is expected that 44% of the induced charge is detected with one pad. The results of gain measurements are shown in Fig. 6. To get a clear signal in this configuration, a gain of 104 is needed for CF4 gas, which is achieved at an anode voltage of þ3:1 kV. In order to make comparable amplification as in P10, CF4 needs a 1.5 times higher electric field.
3.2.2. Gain fluctuation The gain fluctuation in CF4 is evaluated to be 32% at an anode voltage of 3.4 kV. The gain fluctuation in P10 is 10% at the same gain. While the fluctuation of P10 is consistent with the statistic error from the number of drift electrons, the fluctuation of CF4 is significantly larger than expected from statistics alone. This is most probably due to the absorption of electrons via dissociative attachment in the high field region near the anode wires [9,10]. These attachment processes in CF4 are: CF4 þ e ! F þ CF3
(2)
CF4 þ e ! F 2 þ CF2
(3)
CF4 þ e ! F þ CF 3.
(4)
The electron impact cross-sections for CF4 were evaluated for electron energies in the range of 0–1000 eV [11], which include attachment cross-sections derived from absorption processes. As well as the avalanche coefficient, the attachment coefficient is obtained from cross-sections of CF4 using the simulation program Magboltz [12]. As shown in Fig. 7, the number of electrons at each distance from the wire is calculated from the attachment and avalanche coefficients with the following relation: Z 1 number of electrons ¼ N 0 exp ðaðxÞ þ bðxÞÞ dx (5) r
where N 0 is the number of drift electrons, aðxÞ and bðxÞ stand for the avalanche and attachment coefficient, respectively. After a mean free path bðxÞ1 , a drift electron will be absorbed. The number of electrons is estimated to 1 become 10 before amplification. Thus, pffiffiffiffiffi the statistical fluctuations increase by a factor of 10. On the other
ARTICLE IN PRESS T. Isobe et al. / Nuclear Instruments and Methods in Physics Research A 564 (2006) 190–196
194
hand, if the drift electrons were not absorbed, the error would not become larger than in the case of P10.
ADC = a*TDC+b
fit line
100
4. Drift velocity and longitudinal diffusion 80
4.2. Test result The arrival time of a signal, t0 , is obtained with the following relation: ADCpedestal b (6) a where a and b are parameters of the fitted line as shown in Fig. 9, ADCpedestal being the pedestal value of the ADC. The drift time and the longitudinal diffusion were derived from the mean value and the standard deviation of the t0 -distribution after fitting it with a Gaussian function. t0 ¼
VH(~30kV) Drift region 1MΩ
N2 laser 337nm
ADC 40
5%
t2
20
pedestal
0 100 110 120 130 140 150 160 170 180 190
t0
TDC
Fig. 9. Typical shape of a laser signal. A t0 , start time of signal, is obtained from the fitted line between t1 (ADC ¼ 70%) and t2 (ADC ¼ 5%).
10 8 P10 measurement data P10 simulation data
6
CF4 measurement data CF4 simulation data
4 2 0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
electric field (kV/cm/atm) Fig. 10. The drift velocity as a function of the electric field obtained in P10 and CF4.
The drift velocity is obtained using the following relation,
Gas Vessel
f = 200
60
drift velocity (cm/µs)
The drift velocity and the longitudinal diffusion of electrons were measured using a nitrogen laser (YKN500 by Usho Co.). The experimental setup is shown in Fig. 8. A laser beam is reflected by two mirrors into a stationary drift cell through a quartz lens of 20 cm focal length. The focal point is in the middle of the active drift volume. For laser injection, there are two quartz windows which are located at a distance of 34.0 and 18.2 cm from the readout pads. A portion of the laser beam is reflected to a photo-diode (S1722-03: Si pin photo-diode by HAMAMATSU Co.) for start timing. The wave length and the energy of the laser beam are 337.1 nm and 2.5 mJ, respectively. This laser can generate 102 electrons around the beam focal point in CF4 by multiple photon absorption. By attenuating the laser beam, the number of generated electrons is reduced to the level of a single electron to measure drift velocity and diffusion.
70%
t1
4.1. Setup
mirror
0V
anode V Optical Trigger (photo diode)
Fig. 8. Experimental setup for the measurements of drift velocity, longitudinal diffusion and attenuation length.
vdrift ¼
z1 z2 t0 ðz1 Þ t0 ðz2 Þ
(7)
where z1 and z2 are the distances between the readout pads and the laser position, and t0 ðz1 Þ and t0 ðz2 Þ are the corresponding arrival times of the drift electrons. Results are shown in Fig. 10. The drift velocity in CF4 is 10 cm=ms, when the electric field is 900 V/cm. This high drift velocity of CF4 comes from the large Ramsauer dip in the elastic cross-section [12]. From the fluctuation (r.m.s.) of arrival times, the longitudinal diffusion coefficients of CF4 and P10 are obtained and shown in Fig. 11. Lines through the data points represent calculations using the Magboltz simulation program. Magboltz solves the Boltzmann transport equations to calculate transport properties of electrons in gas mixtures. The result of P10 is consistent with the
ARTICLE IN PRESS T. Isobe et al. / Nuclear Instruments and Methods in Physics Research A 564 (2006) 190–196
195
3
longitudinal diffusion P10 measurement data P10 simulation data CF4 measurement data CF4 simulation data
10 3 Attenuation Length (cm)
diffusion for 1cm (µm)
10
2
10
0
0.1
0.2
0.3 0.4 0.5 0.6 0.7 electric field (kV/cm/atm)
0.8
10 2
100
0.9
200
300
400 500 H2O (ppm)
600
700
800
Fig. 11. The longitudinal diffusion as a function of the electric field obtained for P10 and CF4. Lines are obtained from the Magboltz simulation [12].
Fig. 12. Attenuation length as a function of the quantity of water in the gas vessel. ðE ¼ 735 V=cmÞ The fraction of oxygen in this measurement was 30% of the water contents.
simulation. The longitudinal diffusion of CF4 is 60 mm for 1 cm drift, when the electric field is 900 V/cm. Thus, two electron clusters separated by 1.3 mm in longitudinal direction can be separated within 3-sigma even after 50 cm of drift. This is sufficient for operation under high multiplicity conditions in RHIC. The diffusion in CF4 is smaller than in CO2, for example, a popular chamber gas pffiffiffiffiffiffiffi with a small diffusion coefficient of 80 mm= cm at an electric field of 900 V/cm [13].
The results of the attenuation length measurements as a function of the contamination of water are shown in Fig. 12. If the contamination of water is less than 250 ppm, the attenuation length is more than 2 m, and more than 80% of the electrons will survive after 50 cm of drift in CF4. The solid line shown in Fig. 12 is a fitted line to obtain the attachment coefficient for electro-negative pollutants in CF4. Water does not contribute as much to attachment as oxygen [14]. The fraction of oxygen in this system was 30% of the water contents.
5. Attenuation length 6. Summary 5.1. Setup The attenuation length was measured in order to evaluate how much electro-negative contamination can be tolerated for adequate TPC operation. The attenuation length is obtained for CF4 by comparing the collected charge at two different laser beam entry points. The setup is the same as that for the drift velocity measurement. Water contamination is controlled by changing the flow rate of the gas to the chamber and is monitored by a moisture monitor (35-IS by PANAMETRICS Co., Ltd). The applied electric field is 735 V/cm. 5.2. Test result The attenuation length is obtained with the following relation: ðz1 z2 Þ attenuation length ¼ lnðQðz1 Þ=Qðz2 ÞÞ
(8)
where z1 and z2 are the distances between the readout pads and the laser beam position, and Qðz1 Þ and Qðz2 Þ are the collected charge at z1 and z2 . The amount of charge is obtained by fitting a charge spectrum with a Gaussian function.
A prototype Time Projection Chamber (TPC) using pure CF4 gas was developed for possible use in heavy ion experiments. Basic characteristics such as gain, drift velocity, longitudinal diffusion and attenuation length of electrons in CF4 were measured with this TPC. At an electric field of 900 V/cm, the drift velocity and longitudinal pffiffiffiffiffiffiffi diffusion were obtained as 10 cm=ms and 60 mm= cm, respectively. The relatively large gain fluctuation is explained to be due to the electron attachment process in CF4. Eighty percent of electrons will survive after 50 cm of drift with a water contamination of 250 ppm in CF4. These characteristics are encouraging for the measurement of the charged particle trajectories, using a TPC at RHIC. Acknowledgment Authors would like to thank C.L. Woody for his comments on the paper. References [1] [2] [3] [4]
S.S. Adler, et al., Phys. Rev. Lett. 91 (2003) 072301. C. Adler, et al., Phys. Rev. Lett. 90 (2003) 082302. T. Sjo¨strand, et al., Computer Phys. Commun. 135 (2001) 238. K. Adcox, et al., Nucl. Instr. and Meth. A 499 (2003) 469.
ARTICLE IN PRESS 196 [5] [6] [7] [8] [9]
T. Isobe et al. / Nuclear Instruments and Methods in Physics Research A 564 (2006) 190–196 K. Adcox, et al., Nucl. Instr. and Meth. A 499 (2003) 489. D.R. Nygren, Phys. Scripta 23 (1981) 584. J. Va’vra, et al., Nucl. Instr. and Meth. A 324 (1993) 113. T.C. Meyer, The Science and Culture Series—Physics, 2003, p. 180. L.G. Christophorou, et al., Nucl. Instr. and Meth. 163 (1979) 141.
[10] [11] [12] [13] [14]
R. Thun, Nucl. Instr. and Meth. A 273 (1988) 157. R.A. Bonham, Jpn. J. Appl. Phys. 33 (1994) 4157. S.F. Biagi, Nucl. Instr. and Meth. A 421 (1999) 234. V. Palladino, B. Sadoulet, Nucl. Instr. and Meth. 128 (1975) 323. M. Huk, et al., Nucl. Instr. and Meth. A 267 (1988) 107.