Results concerning understanding and applications of timing GRPCs

Transcription

1 Nuclear Instruments and Methods in Physics Research A 58 (23) Results concerning understanding and applications of timing GRPCs Ch. Finck a, *, P. Fonte b, A. Gobbi a a Gesellschaft f.ur Schwerionenforschung, Darmstadt, Germany b Laborat!orio de Instrumentaç*ao e Fisica Experimental de Particulas, Coimbra, Portugal Abstract Results of beam tests are reported: the so-called efficiency puzzle is investigated using a :1 mm gap GRPC operated at different pressures for isobutane gas. The influence of edge effects on the time-response function is studied for a ð4 :3 mm 2 Þ-gap GRPC disk of 5 cm diameter and found to have negligible influence on resolution and tails. A design of large-scale array is elaborated for experiments requiring not only good time and position resolution but also good multi-hit capabilities. r 23 Elsevier Science B.V. All rights reserved. PACS: 29.4.Cs; 29.4.Cx Keywords: Time of flight array; Primary clusters; Surface electrons; Avalanches; Shielded detector strips; Cross-talks; Multi-hits 1. Introduction Glass resistive plate chambers (GRPC) have become, since about 2 years, very promising detectors for accurate time-of-flight (ToF) measurements in relativistic heavy ion collisions [1]. The intrinsic resolution was seen to be as good as 5 ps ðsþ for a four-gap single read-out GRPC of an area of 3 3cm 2 [2]. Resolutions of 6 7 ps were observed [3] for a one-strip counter as large as 16 5cm 2 built of normal float glass and with double read-out allowing for a position information along the strip with a resolution of 1:2 cmðsþ: *Corresponding author. Present address: SUBATECH, 4, rue A. Kastler, F-4437 Nantes Cedex 3, France. Tel.: ; fax: address: christian.finck@subatech.in2p3.fr (Ch. Finck). An accurate ðx; yþ position information was obtained using the charge division method [4]. The detector developments have profited from new refinements of the fast timing electronics [5] and understanding of the physical processes governing the counter operation has also received increased attention [6,7]. Here are reported further investigations in view of a realization of large-scale arrays to be operated in a multi-hit environment. Beam tests were performed which demonstrate that edge effects are negligible for single-cell (or single-strip) detectors and a design is elaborated which is optimized for a safe realization of a large array with good timing as well as position information, which avoids cross-talks and allows for double hit recognition. During the beam tests, a series of measurements was also devoted to an elucidation of the so-called efficiency puzzle aiming at an /3/$ - see front matter r 23 Elsevier Science B.V. All rights reserved. doi:1.116/s168-92(3)1278-6

2 64 Ch. Finck et al. / Nuclear Instruments and Methods in Physics Research A 58 (23) improved understanding of the avalanche generation mechanisms. 2. The efficiency puzzle A question which has been raised since many years in general but also in particular for RPCs [7], is whether the primary ionization clusters are produced by the detected particle exclusively in the gas or whether additional surface electrons are emitted from the cathode and generate avalanches. In order to elucidate this question, a detector of one gap, :1 mm wide, and with 3 3cm 2 surface, was used to measure the efficiency curve as a function of voltage for several pressures, 8oPo16 Torrs; using pure isobutane as gas. In order to be detected, a particle had to deliver in the detector electronic a slow charge larger than 2 2 fc (depending on amplifier gain). The high-voltage was raised up to about 4% above detection threshold for avalanches, where the efficiency curve tends to saturate. At such a voltage, an efficiency value, eff; was extracted from the data and, based on Poisson statistics, the average number of primary clusters n cl was deduced: n cl ¼ lnð1: effþ: The corresponding results are indicated in Fig. 1 as a function of pressure. The error bars consider statistical as well Primary clusters (n ) cl Pressure (torr) Fig. 1. Average number of primary clusters n cl as a function of pressure. A :1 mm gap GRPC was operated in pure isobutane gas. The lines refer to calculations assuming full efficiency of the gap for n cl ¼ 5 and 1 (dashed and dotted dashed line, respectively). The agreement with n cl ¼ 5 is misleading since only 3 5% of the gap are efficient, for details see text. as systematic errors. The deduced average number of primary clusters rises linearly with the pressure, as it may be expected for clusters produced in the gas. In the figure are also indicated, for comparison, two lines which correspond to the maximum number of clusters expected on the bases of: (a) the full gap (of size d) is efficient and (b) the primary number of clusters produced exclusively in the gas is, at normal pressure, n cl of 5 and 1 cl=mm; for the lower and upper line, respectively. Unfortunately, it is at present not possible to predict from literature a reliable value of n cl for minimum ionizing particles in isobutane. The lower value of 5 corresponds roughly to the value given in Ref. [8] and the higher one to Ref. [9]. The experimental values of Fig. 1 lie on a line with, if any, a small offset, what is indicative for a small number of surface electrons, less than.1. This estimation assumes that n s does not depend on the electric field strength or on the gas pressure: in such a situation, n s appears as an offset on the ordinate in Fig. 1. For a given pressure, the standard detector of :3 mm gap used for timing is operated at lower electric field when compared to one with :1 mm gap; in this sense, n s ¼ :1 can also be considered as an upper limit. In a comparison between calculations and data, the agreement of the line for n cl ¼ 5 with the data is misleading since it should be considered that, in reality, only a fraction of the gap is efficient. The fraction depends on how fast the effective amplification coefficient, a Z (where a is the first Townsend coefficient and Z the attachment coefficient) increases above threshold. For 4% overvoltage we estimate that 3 5% of the gap is efficient (depending on the reduced electric field E=P: at low pressure B3%; at high pressure B5%). As it was noticed before [7], a value of n cl ¼ 5 cl=mm would lead to unexpected large values for the Rather condition ðða ZÞdÞ: Such a discrepancy would disappear in the case of n cl ¼ 1 cl=mm: It would, therefore, be of considerable interest to settle in a near future the n cl values for isobutane as well as for the standard GRPC-gas mixture (to our knowledge, the large n cl -value of [9] is the only one substantiated by published experimental results).

3 Ch. Finck et al. / Nuclear Instruments and Methods in Physics Research A 58 (23) Edge effects The GRPC used in the test at CERN-PS had the shape of a disk (5 cm diameter) with two glass plates and four aluminum plates, providing four gaps of 3 mm each. The gas was the usual mixture of 85% freon ðc 2 H 2 F 4 Þ; 1% SF 6 and 5% isobutane. The induced signals are read-out from the inner aluminum plates on two opposite points of the circumference (defining the x-axis). The timing and amplitude informations was deduced using both signals. The secondary beam provided 7 GeV=c momentum negative pions. A pair of scintillators with fast photo-tubes gave the reference time information, additional finger scintillator led to an active trigger area of about 3 cm 2 : More informations about single-cell GRPC design, setup, electronics and data reduction used in this test can be found in Refs. [3,5,1]. Here are reported results of the beam test, followed by a discussion about edge effects. (1) Time response along the x-axis: The beam spot was moved from the center position ðx ¼ cmþ to the edge of the counter ðx ¼ 72:5cmÞ: The measured charge, time resolution and absolute tail (relative proportion of events outside the window 73 ps with respect to the center of the fitted Gaussian) are presented in Fig. 2. Illuminating the center Tails (%) Resolution (ps) Charge (a.u.) (a) (b) (c) X Position (mm) Fig. 2. Charge, resolution and proportion of tails as function of x-position, for a single-cell GRPC Entries 268 Double Readout σ int = 53 ps Tail =.1 % GRPC Beam ToF (ns) Fig. 3. Time response function for the single-cell GRPC when illuminated, in vertical direction, over the full counter diameter. of the GRPC, a resolution of 5 ps ðsþ with negligible tail and efficiency higher than 99% were achieved [1]. Even at the most outer position ðx ¼ 72:5 cmþ the resolution is still better than 6 ps and the amount of tail is below 1%. The maximum delay between the centroid of the time spectrum measured for the outer position ðx ¼ 72:5 cmþ compared to the central position ðx ¼ cmþ is less than 8 ps: (2) Time response along the y-axis: Using a larger trigger area ðb1 cm 2 Þ along the y-axis, the time spectrum plotted in Fig. 3 was obtained. Within errors, no deterioration ðs ¼ 53 psþ is observed compared to the nominal resolution ðs ¼ 5 psþ: The amount of tail is still negligible. (3) Discussion: For a discussion of the influence on the response function due to the edges in a single-cell RPC, several aspects need to be considered which originate from four different causes: Deformation of the electrical field: Where the parallel plate geometry of the gap ends, the field is not any more homogeneous, but still leads to valid signals, over a region of a given width around the circumference. As illustrated in Fig. 4(a), the electric field drops according to

4 66 Ch. Finck et al. / Nuclear Instruments and Methods in Physics Research A 58 (23) Fig. 4. Four possible edge effects: (a) deformation of the electrical field at the edge of one gap (cross view); (b) loss of signal amplitude for the charges induced by an avalanche at the edge of the HV-electrode (cross view); (c) top view of the counter with the signal propagation between the location of the avalanche and the pick-up point of the pre-amplifier: besides the direct path, there exist paths where the signal is reflected from the edge of the counter (echo); (d) in an overlapping geometry of single cells counters, due to variations in the angle of incidence of the particles there can be detection losses in the edge region. Garfield calculations [11] near the edge of the aluminum plate. The width of the inhomogeneous field can be estimated to be between 2 and 3 mm which is comparable to the gap size. The associated area, for a single cell of 5 cm diameter, represents.6.8% of the total counter surface. Correspondingly, the expected amount of tail remains under the 1% level and agrees with the results of Fig. 2(c). Loss of induced charges: The timing signal originates from the fast moving electrons of the avalanche drifting in the electric field and inducing charges on the readout electrode. At the edge of the gap, part of the charges are induced on frames outside the electrodes and lead to a drop of the measured pulse height (see Fig. 2(a)). As illustrated in Fig. 4(b), the loss is expected to be larger if the avalanche is produced in the upper gap which is further away from the read-out electrode. So the actual drop in amplitude is smeared out by this effect. The time response can, in principle, be affected (see Fig. 2) since the reduced amplitude influences the time to amplitude correction, especially for small pulses. Echo: As sketched in Fig. 4(c), all around the counter due to edges, the signal produced at the avalanche location is

5 Ch. Finck et al. / Nuclear Instruments and Methods in Physics Research A 58 (23) reflected and propagates in all directions with the velocity v sig of B13 cm=ns: The reflections interfere with the detected direct signal. The distortion depends on the relative timing and amplitude of the reflected signal with respect to the direct one. The relevance of the echo effect with respect to the time response function of the single cell is difficult to pin down since it depends on the signal termination all over the circumference of the counter. Angle of incidence: It is necessary to arrange GRPCs in an overlapping geometry as sketched in Fig. 4(d) in order to avoid dead area in a ToF array. In experiments using magnetic field, in order to identify charge particle, the incidence angle on the detector varies with their transverse momentum p > : The amount of overlap should be in accordance with the lowest p > to be measured. 4. A new design based on shielded single strips In the following, a design is proposed with the aim to satisfy the stringent requirements set by a sub-threshold strange meson identification in hadronic physics experiments, where a large particle multiplicity is expected. In order not to Fig. 5. Part (a) shows a schematic view of a shielded single GRPC strip. Part (b) shows a three layer arrangement of many strips.

6 68 Ch. Finck et al. / Nuclear Instruments and Methods in Physics Research A 58 (23) spoil the good intrinsic detector resolution of 5 ps; it seems to be safe and less expensive to adopt, for a multi-hit situation, a solution where single-strip detectors are shielded like in a coaxial transmission line so that the very fast pulses propagate along the counter under best conditions, undisturbed (what needs to be true also for all the analogue timing electronic that follows). Fig. 5 sketches the proposed solution: part (a) shows a cross-section for one strip while part (b) displays how the individual strips can be arranged. For one strip, a thin metallic shielding of rectangular profile houses a double gap GRPC strip. This consists of an aluminum high-voltage middle electrode, on both sides, with a :3 mm gap built by float glass plates, which on the outside are coated with a series of narrow long strips. A fast timing signal is read-out at both ends from the aluminum electrode. Sum and difference of the two measured times deliver a ToF and an x- position information, respectively. The slow charge Q i is read-out pairwise from upper and lower external narrow strips. The registration of individual Q i allows for a multi-hit recognition and rejection by software. A centroid determination of the induced charges delivers an accurate y- position information which is of interest for a hit association with the tracking device. It should be noticed that the price per channel of the slow charge measurement is more than an order of magnitude cheaper than the one of the timing channel so that the Q i measurement allows to keep the number of the expensive timing channels to a minimum, for an acceptable number of rejected double hits ðb5%þ and, in addition, with good position information. As illustrated in Fig. 5(b), the strips can be arranged in three layers with large overlap between strips. This has a number of advantages: (1) no loss of detected particles, even for trajectories bent in a magnetic field; (2) has large efficiencies using single two gap detectors which have, compared to four gaps, no floating electrodes; (3) ensures low tails, and (4) facilitates an accurate relative timing calibration between strips. 5. Conclusions Results from a recent beam test on timing GRPC were reported: one series of measurements addresses the question of the efficiency puzzle, while additional measurements aim at the verification of edge effects in view of large-scale applications, where high quality timing and good multi-hit capabilities are required. Concerning the efficiency puzzle, the results seems to exclude a relevant role of surface electrons ðn s o:1þ: The observations are consistent with the normal value of the Rather criterion ðb2þ; but only if for the primary number of clusters per mm of n cl ¼ 1 is assumed in isobutane (this would solve the puzzle). Apparently, one finds from literature conflicting values of n cl and it would be important to settle in a near future this open question. Concerning the large-scale applications of timing GRPC, the beam test results with the single four gap GRPC show that edge effects are not a limiting factor allowing for the proposition of a safe and economic solution, which is based on shielded single strip GRPC detectors. Acknowledgements H. Daues, K. Dermati, H. Gaiser, J. Klemm, B. Kindler and J. Luehning have been very helpful in the accurate preparation of the GRPC detectors. The excellent technical assistance of A. Blanco and N. Carolino during the beam test was highly appreciated. Many thanks are due to M. Abbrescia, V. Peskov, F. Sauli and W. Riegler for discussions concerning the number of primary clusters. We are grateful to A. Mangiarotti for suggestions and discussions. References [1] P. Fonte, A. Smirnitski, M.C.S. Williams, Nucl. Instr. and Meth. A 443 (2) 21. [2] P. Fonte, R. Ferreira Marques, J. Pinh*ao, N. Carolino, A. Policarpo, Nucl. Instr. and Meth. A 449 (2) 295.

7 Ch. Finck et al. / Nuclear Instruments and Methods in Physics Research A 58 (23) [3] A. Blanco, N. Carolino, Ch. Finck, P. Fonte, R. Ferreira Marques, A. Gobbi, M. Rosas, A. Policarpo, preprint LIP/ 1-4, Nucl. Instr. and Meth. A, accepted for publication. [4] A. Blanco, Ch. Finck, R. Ferreira Marques, P. Fonte, A. Gobbi, S.K. Mendiratta, J. Monteiro, A. Policarpo, M. Rozas, Nucl. Instr. and Meth. A 478 (22) 17. [5] A. Blanco, N. Carolino, P. Fonte, A. Gobbi, IEEE Trans. Nucl. Sci. NS-48 (4) (21) [6] P. Fonte, V. Peskov, Nucl. Instr. and Meth. A 477 (22) 17. [7] P. Fonte, Nucl. Instr. and Meth. A 456 (21) 6. [8] W. Farr, J. Heintze, K.H. Hellendrand, A.H. Walenta, Nucl. Instr. and Meth. A 154 (1978) 175. [9] F.F. Rieke, W. Prepejchal, Phys. Rev. A 6 (1972) 157. [1] Ch. Finck, et al., Proceedings of the XXXIX International Winter Meeting on Nuclear Physics, Bormio, 21. [11] Garfield, A drift chamber simulation program, version 5.4, updated , CERN.