&_&L& Nuclear Instruments and Methods in Physics Research A 366 (1995) 100-l 14 -- *! 1:l!i!J NUCLEAR INSTRUMENTS A METHODS IN PHVSICS RESEARCH SectlcoA EISEVIER Igloo: a neutral pion spectrometer for low energy photoproduction studies J.M. Vogt*, J.C. Bergstrom, Saskatchewan Accelerator Laboratory, R. Igarashi, K.J. Keeter University of Saskatchewan, 107 North Road, Saskatoon, Saskatchewan S7N 5C6, Canada Received 27 April 199s Abstract A n” spectrometer constructed from 68 lead glass Cherenkov counters has been installed on the tagged photon beam line at the Saskatchewan Accelerator Laboratory. It is being used for the investigation of neutral pion photoproduction from light nuclei within about 25 MeV of threshold. It can be configured for total cross section measurements with a large acceptance, or for angular distribution studies with a reduced acceptance. 1. Introduction A 7~” spectrometer (“Igloo”) constructed from 68 lead glass Cherenkov counters has recently been installed on the tagged photon beam line at the Saskatchewan Accelerator Laboratory (SAL), for the purpose of studying 7~~’photoproduction from light nuclei at low energies, specifically within about 25 MeV of threshold. The combination of a pulse stretcher ring [I] and photon tagging spectrometer [2] at the 300MeV facility is well matched to this experimental program, providing a tagged photon beam with the necessary energy resolution and photon flux. For example, near threshold the user has optional photon energy resolutions of either 0.5 MeV over an extended range, or 0.2 MeV over a limited range. Several constraints guided the design of Igloo. First, it must cover a large fraction of the 47~ solid angle to accommodate total cross section measurements. Second, it must be sensitive to the pion angular distributions to accommodate differential cross section measurements. In addition, the spectrometer must readily accommodate cryogenic target assemblies, and finally for reasons described later, no charged-particle veto counters are to be employed. Since the spectrometer was to be constructed from existing glass blocks and photomultipliers, neither of which were ideally suited for the present purpose, our options for designing Igloo were somewhat limited. The resulting compromises lead to a rather simple * Corresponding 966 6058. author. Tel. + 1 306 966 6054, fax + 1 306 0168.9002/95/$09.50 @ 1995 Elsevier SSDZ 0168-9002(95)00557-9 Science B.V. All rights instrument with modest resolution, but sufficient for our purposes. To date, the spectrometer has been utilized in two photoproduction experiments, and ‘WY, n’)) “C(y, r”), both in a preliminary stage of analysis at the time of writing. Our purpose here is to report on the design of the device, to describe our solutions to some common problems, and finally to present briefly some preliminary experimental results to prove the performance of the hardware. 2. Igloo The photon calorimeter consists of four lead glass walls forming a symmetric box oriented with the corners in the vertical and horizontal planes (Fig. la). The two vertical “ends” of the box are unobstructed to accommodate the cryogenic target assembly and the incident tagged photon beam. Each wall is comprised of a 5 X 2 assembly of glass blocks, each block being 191 mm square and 305 mm long and viewed on the square end by a photomultiplier. Additional blocks are suitably situated to prevent shower leakage in the vicinity of the box corners. These are referred to as the “corner guard-blocks”. On each end of the detector there is a single layer of glass blocks which are referred to as the “end guardblocks”; they are used to define the active length of the spectrometer and to provide additional calorimetry for the exterior detector blocks. If a photon hits an exterior regular block, the energy deposited in that block and in the end guard-block are added together (Fig. 2, photon A). If the photon hits the guard-block reserved J.M. Vogt et al. I Nuct. Instr. and Meth. in Phys. Rex A 366 (1995) 100-114 101 tion of the geometric efficiency of Igloo (see Section Tagged Photon Beam J ,-.. -.; __ -.-“- .‘_ (b) Fig. 1. Igloo closed (a) and open (b). only and deposits no energy in the previous block, it is discarded in the data analysis (Fig. 2, photon B). That still leaves the active length of the spectrometer with a “soft edge”, because a photon can clip the corner of the regular block without converting, and deposit all its energy in the guard-block. That photon would also be discarded in the analysis (Fig. 2, photon C). This effect, however, is taken into account in the calcula- Fig. 2. The purpose of the guard-blocks (see text). 7). The above geometry describes the spectrometer when operating in the “closed” configuration, used for measuring total cross sections. To accommodate angular distribution measurements, the spectrometer is designed to separate at the horizontal corners and become, in effect, two L-shaped arms viewing the target symmetrically from above and below. Two rows of corner guard-blocks now become active components of the spectrometer, increasing the acceptance solid angle. In this operating mode, referred to as the “open mode”, each arm is retracted 424 mm along a line defined by the vertical diagonal of the original box (Fig. lb). In the closed and open modes, the distance between opposing walls is 420mm and 1020mm. respectively. Igloo was designed for the current (T,~F”) program. However, to accommodate future (y, P”X) experiments, the detector support has been designed so that the downstream, beam right quadrant permits an unobstructed view of the target by additional detectors. Initially, we considered using veto counters to reject charged particles (beam related background, electromagnetic showers from the target, cosmic rays) and those veto counters were to line the inside of the lead glass cave. The disadvantage of this scheme is that some photons would convert in the veto counters, and, because of the geometry, their probability of being rejected would be hard to determine and would be threshold dependent. In a test run with a few counters stacked in the same geometry as in the detector frame. we found that we could omit the veto counters, provided Igloo was adequately shielded. This is due to the fact that lead glass counters have a natural threshold of several MeV (low energy particles do not emit much Cherenkov light). According to the geometry of Igloo and the angular dependence of the atomic cross sections, only the downstream guard-blocks register a significant number of shower hits from the target. Those end guard-blocks are therefore shielded with 50mm of lead (Fig. 3) - enough to absorb the showers. This is feasible since the guard-blocks are not meant to detect direct hits from the 7~’ decay photons. In order to shield Igloo from photon beam related background that is produced upstream in the collimator or in the photon tagger outrun window. a 1OOmm thick lead wall is mounted on the detector frame at the upstream end (Fig. 3). The photon beam passes through a 35 mm clean-up collimator in that wall. With this shielding installed, in addition to the other lead walls that are normally situated in the experimental hall, neither the single rates in the counters nor the trigger rate from the whole detector are a problem. 102 J.M. Vogt et al. I Nucl. Instr. and Meth. in Phys. Res. A 366 (199-T) IOO-114 f Fig. 3. Lead shielding permanently Fig. 4 shows Igloo in Experimental Area 2 of the SAL facility. After careful alignment with respect to the photon beam line, the detector frame is anchored to the wall. The spot size of the photon beam is defined by a 10mm collimator near the tagger outrun window and is about 30mm in diameter at the target. The photon beam related background is greatly reduced by an arrangement of sweep magnets, clean up collimators and shielding walls. The tagger lead glass counter, which is shown just upstream of the photon beam dump, is used to measure the tagger efficiency (the percentage of tagged photons that pass through the collimator). It can be moved in and out of the photon beam on a remote controlled, movable table that is mounted downstream of Igloo. This is done routinely between production runs and no access to the experimental area is necessary. 3. The counters The majority of the glass blocks are 191 mm X 191 mm X 305 mm (6.2 X 10.0 radiation lengths) in size and are made of F-2 glass. The exceptions are the eight end guard-blocks which are made of SF-2 glass and are 191 mm x 191 mm x 352 mm in size. The individual counters were assembled as shown in Fig. 5. Five-inch EM1 9870B photomultipliers were coupled to the glass blocks via acrylic adapters. Optical silicone was used to couple the photomultipliers to the adapters and to couple the adapters to the glass blocks. While providing sufficient strength, the silicone allows removal of the photomultipliers from the adapters or of the adapters from the blocks without causing Lead Shielding for End Guard-Bkxks installed on Igloo. damage. For simplicity and to keep electrical noise to a minimum, the photomultiplier bases were soldered directly onto the pins of the sockets. An LED was mounted on each light guide adapter. The lead glass blocks and the adapters were then wrapped in aluminum foil. After mounting magnetic shielding shrouds, which are screwed to the acrylic adapters, the counters were wrapped in 0.125 mm brass foil. Brass was chosen because plastic foil was too weak and had a tendency to tear. All standard size counters and 6 of the 8 larger counters were assembled this way. Two of the end guard-blocks are viewed from the side by two photomultipliers each. Other than that they were assembled the same way. The sole function of the LEDs is to pulse individual counters for testing purposes. We did not install a pulser system to drive all LEDs in order to monitor the counter gains on-line. Our simpler approach to gain monitoring will be discussed later (Section 5). The EM1 9870B photomultipliers are hardly ideal for our application. Their performance has been studied in great detail in Ref. [3], where prepulsing, non-linearity and a time jitter of several nanoseconds are reported. However, those photomultipliers were available to us at no cost and monetary constraints precluded the possibility of using state-of-the-art 5 in. photomultipliers. After the first prototype counter was assembled, the detector response was studied by running the tagged photon beam into the front face of the lead glass block. Fig. 6 shows the pulse height as a function of the photon energy. To reduce the non-linearity evident in the figure, we amplified the analog signal by a factor of 10 and reduced the photomultiplier gain, so J.M. Vogt et al. I Nucl. Instr. and Meth. in Phys. Res. A 366 (1995) Primary Beam /-Primary Electron Beam Experimental that the anode signal pulse height was well below 1OOmV The dashed line in Fig. 7 shows the resulting pulse height as a function of the photon energy. The remaining non-linearity is a result of shower leakage out of the sides and the back of the lead glass block. The solid line in Fig. 7 shows the detector response after correcting for shower losses. It is linear for deposited energies of up to 150 MeV 103 Dump Electron Fig. 4. Igloo and the tagger in Experimental lCfSII4 Area 2 Area 2. Fig. 8 shows the time resolution between the photon tagger and the lead glass counter. using a constant fraction discriminator on the lead glass counter. The width of the coincidence peak is 2.4 ns FWHM. Fig. 9 is a scatter plot of the pulse height of the lead glass counter vs. the timing shown in Fig. 8. The lead glass counter has a significant timing jitter at low pulse heights. but other than that there is no pulse height J.M. Vogt et al. I Nucl. Insir. and Meth. in PhJas. Res. A 366 (199-f) lOt-114 FWHM = 2.4 ns hk;yk ShieldingJ Fig. 5. Lead ~PhotamuItipJier glass counter assembly. Tagger TDC (ns) Fig. 8. Timing spectrum between lead glass counters in a test run 2 600 3 500 5 400 2 300 the tagger .. .. . . ........ 800 - :‘.: ~..,I,.,I...I.,.I...I...I...I...I ‘0 20 40 60 100 80 120 140 and one of the . . . 160 EY WV) Fig. 6. Pulse height the photon energy. of the prototype counter as a function of . dependence of the timing. Fig. 10 shows the time resolution between the tagger and one of the lead glass counters in an actual experiment. The random background here is due to the fact that the tagged photon beam is not impinging directly into the lead glass counter, and also the count rate in the tagger is typically 3 to 4 orders of magnitude higher than in Fig. signal-to-noise ratio is impacted by the width of 20 ’ *. 30 40 50 Tagger TDC (ns) Fig. 9. Pulse height as a function of the timing between the tagger and one of the lead glass blocks using a constant fraction discriminator. 800 700 600 500 400 300 200 100 III.II..ll,.I.I.I...JII.I.. ‘0 20 40 60 80 100 120 140 E, (MeV) Fig. 7. Pulse height (dashed line) using phototubes. Detector leakage (solid line). as a function of the photon amplifiers and a lower gain response corrected for the energy in the shower Tagger TDC (ns) Fig. 10. Timing spectrum between the tagger lead glass counters in a real experiment. and one of the J.M. Vogt et al. I Nucl. Ins&. and Meth. in Phys. Res. A 346 (199s) 100-114 10s the coincidence peak and poses one of the practical limits to the count rate that Igloo can achieve (see Section 8). 4. The Electronics Fig. 11 shows the circuit diagram of the electronics for each channel and for the trigger definition. The analog signals of each counter are amplified by a fixed factor of 10 in LeCroy 612A amplifiers. Two of the downstream end guard-blocks have two photomultipliers each. However, due to the slow rise time of the phototubes it was possible to tie their anodes together without causing any ringing. These counters are subsequently wired up like all the other counters. One amplifier output is delayed, attenuated by a factor of 3 and connected to a LeCroy 2249A charge integrating ADC. The other output is connected to a EG&G CFSOOO constant fraction discriminator. The discriminator thresholds are set to 40 mV which corresponds to a deposited energy of 10MeV in the lead glass blocks. The ECL outputs of the discriminators are connected to LeCroy 4434 scalers and LeCroy 2277 pipeline TDCs which are daisy chained (the input terminators of the scalers are removed). The TDCs are run in common stop mode to avoid delaying the input signals. All 68 channels are wired up in this fashion. The discriminator OR-outputs are used to generate the trigger signals. There are 12 discriminator modules corresponding to the 12 rows of counters in Igloo as shown in Fig. 12. The upstream end guard-blocks are Fig. 11. Simplified circuit \-!YkZE Wall 4 Fig. 12. The rows and the walls of Igloo included in rows 2, 5, 8, 11, the downstream guardblocks are included in rows 3, 6, 9. 12. Each discriminator OR-signal is the OR of all counters in that row. The 12 row signals are ORed into four “wall” signals (l-3, 4-6, 7-9, 10-12). The trigger definition during a (-r, 7~“) experiment requires any two of the four walls to fire. This signal is sent to the tagger electronics as the X-trigger [2]. The tagger generates the interrupts to the data acquisition system. The tagger inhibit signal is used to inhibit the scalers. The tagger X-reference signal is used as the common stop for the TDCs and to gate the ADCs. However, a stand-alone interrupt system is diagram of the Igloo electronics. 106 J.M. Vogt et al. I Nucl. Instr. and Meih. included in the Igloo electronics for testing, which then provides scaler inhibits, ADC gates and TDC stops as well as generating the CAMAC LAMS. 5. Energy calibration and gain tracking All lead glass counters were calibrated individually with a tagged photon beam. During these calibration runs the counters were connected to the same channels of the electronics as in the final Igloo configuration. The high voltage was adjusted so that the peak of the pulse height distribution would be in the middle of the ADC spectrum for 100 MeV photons. Spectra such as displayed in Figs. 7-9 were obtained for all counters. In addition, the energy resolution of each counter was determined as a function of the photon energy (Fig. 13). The counters were installed in the Igloo detector frame immediately after their calibration. This was followed by a cosmic ray calibration run of about 12 h. An equivalent cosmic energy deposit was calculated for each counter from its average cosmic ray peak position and its in-beam calibration. This “reference energy” provided the benchmark for future gain monitoring. The procedure for determining the cosmic peak position, also used for tracking the gains of the counters during experimental runs, is as follows. In order to define the path of the cosmic rays through Igloo, counters 1, 4, 7, and 10 (see Fig. 12) in each belt of lead glass counters are used as cosmic trigger counters in the software. So called “good cosmics” spectra are accumulated for each counter if more than 40MeV is deposited in the corresponding trigger counters. If counters 1 and 4 fire, the “good cosmics” spectra are incremented for counters 1, 2, 3, and 4. If counters 4 and 7 fire, the “good cosmics” spectra are incremented for counters 5 and 6. The same principle is followed for counters 7 through 12. Thus, only 20 40 60 80 100 120 140 160 180 E, (MeV) Fig. 13. Energy resolution a function of the photon of one of the lead glass counters energy. as in Phw Rcs. A .766 (199.c) 100 114 cosmic rays that hit all blocks sideways are used for defining the reference energy and for gain monitoring. Conversely. a hit along the length of a block deposits too much energy and causes the ADC to overflow. Igloo has to be in closed mode to do a complete cosmic ray calibration, otherwise there is a gap between counters 6 and 7 as well as between counters 1 and 12. Valid “good cosmics” spectra can still be acquired for counters 1-4 and 7-10 in each belt when Igloo is open. To calibrate the guard-blocks at the ends of Igloo, opposing blocks are used in coincidence. At the beginning and at the end of each experimental run period, a 12 h cosmic run is taken to update the gain coefficients of the counters. This can be done with Igloo interfaced to the tagger or in stand-alone mode. However, note that the cosmic ray hits that satisfy the “good cosmics” cuts will always satisfy the trigger definition: “any two walls out of the four”. It is therefore possible to use cosmics for gain tracking during an experiment as long as Igloo is closed. This can be done by examining the cosmic events that occur in random coincidence with the tagger during the hardware coincidence resolving time of 50 ns (typically 50% of them), or by accepting each X-trigger from Igloo as an interrupt. To establish the gains for each counter, the reference energy of this counter is divided by its cosmic ray peak position. 6. Photon recognition and calorimetry One of the first steps in the data analysis is to process the information read from Igloo in a photon recognition algorithm which determines how many photons were detected in Igloo during a particular event, calculates their energies and assigns each photon to the lead glass counter in Igloo which it most likely hit. In order to calculate the calorimetry information for an event, the ADC values of the 68 counters are converted into deposited energies in MeV. The algorithm then finds the counter with the highest deposited energy. Using a map of the counter arrangement in Igloo, the deposited energies of all adjacent counters are added, provided each is clearly above the pedestal (i.e. greater than 1 MeV). All those counters are then removed from the list and the process is repeated until the calorimetry has been calculated for all photons that were detected. Most photons deposit energy in more than one counter. To obtain correct angular information, it is necessary to determine which block was originally hit. As illustrated by photon A in Fig. 2, the counter that was originally hit does not have to be the one with the highest deposited energy. A better approach than choosing the counter with the highest deposited J.M. Vogt et al. I Nucl. Instr. und Meth. in Phys. Res. A 366 (1995) IO&114 energy is to assign the photon to the counter closest to the target among those that detected some of its shower energy. On the other hand. in the case of photon B in Fig. 2, a small amount of shower energy can leak back into the counter closer to the target. We found that the best compromise is to assign the photon to the counter closest to the target in which more than 7 MeV of shower energy were deposited. This is true for all photons that are detected anywhere in Igloo. But if. according to that algorithm, the photon is assigned to one of the end guard-blocks, it is discarded in the subsequent analysis. 7. Computer simulations 7.1. Description of the detector of the Monte Carlo code The general response of Igloo, including the pion detection efficiency and angular resolution. has been evaluated by Monte Carlo simulation. Since caiculational speed and flexibility are essential in our application, the simulation code employs simple ray tracing of the photon trajectories together with geometric constraints imposed by the detector lattice. The complete detector has also been modelled in the shower code GEANT in order to provide certain calibration benchmarks. However, the requisite computing time limits the usefulness of the GEANT model in the analysis of experimental data. The simulation begins by creating a pion in the y-nuclear center of mass frame for a given incident photon energy. using a Monte Carlo method to choose the 7~”direction in accordance with a specified angular distribution. One then Lorentz transforms to the 7~” rest frame where the back-to-back decay photons are emitted with a random orientation. Finally, a Lorentz boost carries the photons to the lab frame. A valid event requires both photons to intercept any two walls (except the same wall) of the detector. The energy resolution of the lead glass is embodied by randomizing each photon energy about its true value using a gaussian distribution whose width is determined by the actual in-beam calibration of the glass blocks, as typified in Fig. 13. A threshold of 10 MeV is placed on each resulting energy to imitate the actual discriminator thresholds used to define a trigger. The model described so far assumes a 100% photon detection efficiency in the entire active volume of Igloo. In reality, however, it is necessary to take into account the photon shower-conversion probability. This has three important consequences. First, it modifies the effective length of Igloo along the zdirection since event C in Fig. 2 will be rejected in the data analysis. Second, it can shift photon “hits” to the wrong detector block, depending on how the shower 107 propagates among adjacent blocks. Finally, there is a small but finite probability that the photon can “punch through” the active volume of Igloo and escape detection. The conversion efficiencies are incorporated by means of a look-up table which was calculated as a function of glass thickness for three photon energies, 40, 70 and 100 MeV, using the shower code GEANT. In application, the appropriate path length of a photon in the glass is estimated using elementary geometry, and the appropriate conversion efficiency is obtained from the table by linear interpolation in energy and path length. A conventional random number procedure subsequently accepts or rejects the photon. as determined by the conversion probability above 710 MeV. In order to obtain IT” detection efficiencies which are reliable to l-2%, it is necessary to properly incorporate the physical dimensions of the lattice at the level of a few millimeters. especially in the spacing of opposing walls. The gaps between adjacent blocks (about 9 mm) affect the efficiency at the 1% level and are included. The position and physical extent of the target are also taken into account in the Monte Carlo simulations. As noted, the paramount feature of our simulation code, as compared to the more realistic GEANT calculations, lies in its ability to process events rapidly (10” pions require only a few minutes) permitting one to quickly explore the effects of differing initial conditions. as well as assemble a multitude of histograms such as missing-mass distributions, opening-angle distributions. and pion-detection efficiency as a function of pion angle, to mention only a few applications. The practical utility of the code of course depends on how it fares in comparison to the GEANT simulations, especially with respect to the photon detection efficiency. Consider, by way of example. Igloo in the closed mode (for total cross section measurements), with a point source of isotropic photons situated at the center of the detector. For photon energies between 30 and 110 MeV. the photon detection efficiencies from the Monte Carlo simulation agreed with the GEANT predictions to better than 1% in relative magnitude. The comparison is equally favourable when the photon source is shifted 10 cm along the detector axis. In the open mode, the Monte Carlo values are about l-3% low of the GEANT predictions. depending on the photon energy and source position. In view of the complexity of the Igloo geometry. especially in the open mode, these comparisons are acceptable. 7.2. Igloo pion detection efficiency The results described in this section have been calculated for the reaction 12C(y, a”)“C. The carbon J.M. Vogt et al. I Nucl. ltzstr. atd Melh. in Phw. Res. A .766 (lY9.T) lOO-11-I 108 target is 12.7mm thick and 30mm in effective diameter as determined by the photon beam size and is situated in the center of Igloo. This corresponds to the geometry used in the actual performance tests described later. For ‘*C the differential cross section in the y-nuclear cm frame is, to a good approximation, proportional to the quantity F’(Q) sin’@f where 0: is the pion angle and F’(Q) is a nuclear factor which tends to shift the angular distribution forward of the 90” maximum. (Further details on the cross section are given in Ref. [4].) The T” detection efficiency within 30 MeV of threshold (135.8 MeV) and with Igloo in the closed mode is illustrated by the solid curve in Fig. 14. The rapid decrease as one moves away from threshold is simply reflecting the rapidly decreasing minimum opening angle between the decay photons; at threshold they are back-to-back in the cm frame. For single photons, the spectrometer subtends about 0.86 of the 47~ solid angle when the “soft edges” at the ends are included. Let us consider how the detection efficiency is influenced by the pion angular distribution. As an extreme alternative to the sin’Bz distribution. consider the detection efficiency for pions emitted isotopically in the y-nuclear cm frame, shown as the dashed curve in Fig. 14. The divergence between the two curves reflects the different degrees to which photons escape through the open apertures of Igloo, as governed by the differences in the pion angular distributions. The significant point here is that the detection efficiency is not a strong function of the angular distribution. In other words, the total (y, v”> cross section deduced from Igloo is not unduly compromised by the incomplete 4~ coverage and the presumably-unknown angular distributions encountered in other measurements. ‘H( y. rr”) for example. Information on pion angular distributions is gained by reconfiguring Igloo to the open mode. as shown in Fig. 1. Each arm is retracted 424 mm, which means the perpendicular distance between opposing walls is now about 1020mm. The pion detection efficiency for “C is of course now somewhat reduced, spanning the range E = 28-10% over the incident photon energy range E, = 136-165 MeV In fact, the retraction of 424mm was chosen as a compromise between a rapidly decreasing detection efficiency versus improvement in the pion angular resolution. The angular resolution will be described later. In the open mode, the overall pion detection efficiencies for the sin’0: versus isotropic distributions are slightly more divergent than the results portrayed in Fig. 14. but even at EY = 16.5MeV they are within 5% of each other. However, the angular (i.e. 0,“) dependence of the efficiencies becomes more enhanced with increasing E., as illustrated in Fig. 15. The apparent contradiction between the angular dependence of the detection efficiency on the one hand, and the relative insensitivity of the gross efficiency to the 160 5- 7t’ 35 I I 140 145 I 1 150 155 Ey (MeV) , I I 160 165 170 0’0 TARGET : CARBON ’ 20 ’ 40 ’ ’ 60 80 0; Fig. 14. Total rr” detection efficiency as a function of the incident photon energy ET, for Igloo in the closed mode. The target is a carbon disk situated in the center of the spectrometer. The solid curve is based on the physical angular distribution which is proportional to sin’0:. For comparison, the dashed curve derives from an isotropic pion distribution. ’ 100 I I ! 120 140 160 180 (de@ Fig. 15. Angular dependence of the pion detection efficiencies for Igloo in the open mode, for a carbon target situated in the center of the spectrometer. These results are independent of the actual pion angular distribution. The angle 03 is the pion polar angle in the y-nuclear cm frame. J.M. Vogt et al. I Nucl. Instr. and Meth. in Phys. Res. A 366 (1995) two angular distributions on the other hand, is resolved when one realizes that P” spectrometers of this geometry directly yield de/de,*, not dcAd(cos 13:) (i.e. not the physical cross section da/da:). The sine,* factor relating these quantities effectively “damps out” the pion population at the extreme angles in du/dez, precisely where the detection efficiency is lowest (Fig. 15). Note that the sin 19: factor is not intrinsically contained in our definition of the pion detection efficiency, so the results in Fig. 15 apply to any pion angular distribution. In the closed mode, the angular variations of the efficiencies are much milder than this, as one might expect. For very light target nuclei, recoil effects cause the efficiencies shown in Fig. 15 to become skewed with respect to 0: = 90” (cm). Symmetry is restored by shifting the target position upstream with respect to the incident photons. Thus for hydrogen, the target center is situated 100 mm from the geometric center of Igloo. P,* = P, ) 7.3. Pion angular resolution (open mode) tan 0: = [Pl + P~]“‘lP~ The cm pion angles 0: and cp,* are reconstructed from the measured n” decay photon energies E,,, and their respective laboratory angles 0, ,Z and (p,,: as defined with respect to the target center. There are several ways of doing this, depending on the particular weight one wishes to give the individual photon parameters, and we refer the reader to Ref. [5] for a review of the various options. The particular algorithm we employ here strikes a balance between the limited resolutions in photon energy and angle as measured by Igloo in the sense that each contributes roughly equally to the reconstructed pion angular resolution. The photon angles 0, and p, are defined only to within the finite angular domains subtended by the square faces of the PbG blocks. Therefore, each photon is assigned angular coordinates within these domains by choosing ‘p and cos e,, with uniform probability, between the corresponding angular extremes defined by the appropriate block. The alternative is to place each photon dead-center of the block face, but this yields a discontinuous structure in the reconstructed distributions due the large facial dimensions of the blocks. The pion angles are deduced as follows. In the laboratory frame the Cartesian components of the pion momentum are given by P., = E, sin 0, cos ‘p, + E, sin 0, Pf = P, (2) PT =+; ‘ where The incident photon energy in the lab frame is E,, and M is the target nuclear mass (we employ c = 1). The pion total energy in the cm frame is given by E:= W2fm” 2; -M’ . where W’ = 2EYM + M2 describes the invariant mass. Finally, the pion angles in the cm frame are reconstructed using the expressions , (3) tan cp,*= P?lP, . (4) Only Eq. (3) is of interest for the present application. The angular response following from Eq. (3) has been investigated using the Monte Carlo simulation. The reconstructed O,* distributions for pions emitted with unique values of 0: (but arbitrary cpz) exhibit roughly gaussian shapes. The half-widths rFWHM, which are a measure of the angular resolution, are displayed in Table 1 as a function of the incident photon energy EY for some representative initial angles S_*. Two features of I&,, are of note here: the resolution is relatively insensitive to S,*, and the resolution improves quickly as one moves away from threshold (135.8 MeV for ‘*C). The latter effect can be traced to the increasing asymmetry in E, versus Ez away from threshold. However, the angular depenTable 1 The angular resolution of Igloo in the open mode, as deduced by Monte Carlo simulation. The resolution is defined as the half-width f,,,,, (degrees) of the distribution function of 02 as given by Eq. (4a). for specific values of the emitted pion angle 0: (but arbitrary ~05). The target is “C whose n”threshold is EI = 135.8 MeV, where EY is the incident photon energy in the lab frame cos cpz Pv = E, sin 0, sin ‘p, + E, sin e2 sin q2 P_ = E, cos e, 109 100-114 + Ez cos e, . A Lorentz transformation to the y-nuclear indicated by asterisks, gives 20 50 90 130 160 140 45” 1.50 160 170 29” 22” 20” 45” 28” 24” 22” 48” 32” 28” 27” 46” 29 23” 21” 45” 27” 23” 19” (1) cm frame, 110 J.M. Vogt et al. I Nd. lmtr. md Mrth. dence of I;-WHM is a more subtle phenomenon and requires comment. From the Monte Carlo simulations one may readily discover how c,v,,, is influenced by the uncertainties in the photon energies E, as returned by Igloo, as well as the corresponding uncertainties in the lab angles 0, and ‘p. The results are as follows: 1) The resolutions in E, and cp,have little influence on I&+, at 0: = 90” but both display increasing influence as the extremes in 0: are approached. 2) The resolution in 0, has a profound influence in r FWHM at 0: = 90” and least influence at the extremes in 0:. The uncertainties in E,, 0, and ‘p are such that these two opposing trends tend to complement each other across the whole angular range of 6:. 7.4. Pion angular distribution (open mode) Finally. we address the pion angular distribution measurements. specifically how to recover information on the differential cross section da/da,* in the 7~nuclear cm frame. Given the limits on energy and angular resolution, one cannot expect to reconstruct an accurate representation of da/do,* in the threshold region without resorting to a complicated deconvolution procedure. Instead, we will rely on the Monte Carlo code to provide a replication of the observed Igloo response by optimizing the input-model angular distribution. Comparison with the data is made on two levels: through the reconstructed angle 0,*, and through the *‘belt-hit” patterns. For purposes of illustration, let us suppose the cross section of interest can be expressed as in P/IJ.\. Res. A 366 (lYY_i) 100-11-l First. consider the cross section as given by the Hz reconstruction algorithm Eq. (3) and converted to du/dS2:. The reconstructed partial cross sections corresponding to the A, B and C terms of Eq. (5) are displayed in Fig. 16. No efficiency corrections have been applied to these simulations. so each curve represents the convolution of the input cross section with the angular resolution and the angular efficiency. The input cross sections are portrayed by the dashed curves. Each simulation has been renormalized to the same total cross section as given by the dashed curves. Ultimately, it is the coefficients A, B and C that one wishes to extract from the experimental data. In principle this could be accomplished by treating each simulation in Fig. 16 as a “template”. One then weights each template by the appropriate coefficient A-C until a fit to the experimental data over a small energy region is achieved. In practice one must fit 1.5 I 1 0.5 II 1.5 - B=l k da --=AAB(l-cos8,*)+Csin’B,*, 9 da,* where A, B and C are constants for a given energy. This expression is appropriate for describing the reaction ‘H(y, 7~“). and in fact we will use this particular reaction as our model since it represents a worst case scenario because of recoil effects. Writing the cross section in the above rather unconventional form has the advantage that, for hydrogen, all three coefficients are positive at low energy. Thus Eq. (5) can be considered as the sum of three positive-definite cross-section components which greatly simplifies the analysis. Let us now examine the Igloo response to each component of Eq. (5). The “target” cell will be assumed 100mm in length, 30mm in effective diameter, and is shifted 100 mm upstream from the center of the spectrometer. In order to display a “typical” response (i.e. neither the best nor worst), we will assume an energy 5 MeV above v5’ threshold, which for hydrogen corresponds to E, = 149.67 MeV o.5c / o//’ 0 20 \ ” 40 60 ’ 80 6; ’ ’ I ’ ‘._ 100 120 140 160 160 Ws) Fig. 16. Simulated reconstructions of the three components of the cross section Eq. (5). for ‘H(r, a”) at an energy 5 MeV above threshold with Igloo in the open mode. The reconstruction algorithm is defined by Eqs. (l)-(3). The angle 0; is the pion polar angle in the y-nuclear cm frame. The dashed curves describe the “input” to the Monte Carlo code used to simulate the “data“. while the solid lines are the reconstructed results. normalized to the total “input” cross sections. No detection efficiency corrections have been applied to the solid lines. J.M. Vogt et al. I Nucl. Imtr. and Meth. in Phys. Res. A 366 (199-T) 100-114 several energies simultaneously to maintain energy continuity in the coefficients, but that is beyond the scope of the present article. The second method for retrieving information on da/da_*, which requires no 0: reconstruction at all. is based on “belt-hit” patterns. Through a series of coplanar cuts made perpendicular to z-axis, one can conceptually dissect the active volume of Igloo into a set of five square rings or belts. These belts are situated side by side, and each consists of eight active PbG blocks and four corner guard-blocks. In the open mode these belts naturally contain dead sections, but the concept remains intact. The method relies on the fact that the relative distribution of photon hits across the five belts is a function of the pion angular distribution du/dnz. To illustrate, we again turn to the model cross section Eq. (5), with the same target geometry as before, and again observed 5 MeV above threshold. Let us consider the cross section under two extreme physical situations. S-wave dominance and P-wave dominance: a) S-wave dominance: A = 0,B = 1,C = 0. b) P-wave dominance: A = 1,B = 0,C = 1S. The corresponding distributions of belt-hits for these two cases, in the open mode of Igloo, are displayed in Fig. 17. We discern from this figure our ability to distinguish between symmetric and highly asymmetric cross sections, realizing that these indeed represent physical extremes and that physical reality would lie somewhere in between. Nevertheless, there is a 0.8 - 0.6 - 0.4 - I L A- (4 Om2- i Fig. 17. Simulated belt hit patterns for ‘H(y, n”) based on Eq. (5). evaluated 5 MeV above threshold with Igloo in the open mode. Pattern (a) represents the extremely asymmetric pion angular distribution given when A = 0, B = 1, C = 0, while pattern (b) reflects the symmetric pion distribution for A = 1, B = 0, C = 1.5. In both patterns, the forward belt is the left-most bin. 111 statistical advantage in this method in that all events are distributed over only five bins, and furthermore since no reconstruction is involved, the measured photon energies E,,,play no direct role. Finally, we will complete this discussion of belt-hit patterns on a technical note. For the method to be reliable, it is clearly important that the simulation code replicates the actual behavior of Igloo in situations where the initial shower in a given belt is subthreshold. in which case the photon hit might then be subscribed to the adjacent belt. The net effect is to transfer events outward, away from the target, to the outermost belts. This problem was previously alluded to in Section 7.1. The simulation code incorporates this leakage by resorting to the same conversion efficiency look-up table described therein, and employs a similar algorithm. Note that photons can be “lost” only to the extreme outer belts, otherwise they are simply transferred to the next-outermost belt. The transfer effect is not negligible, for example it can increase the population of the outer belts by 10% or more. The algorithm has proved to be reliable, as will be demonstrated when we discuss the spectrometer performance test. 8. The commissioning experiment The commission of Igloo was made using the reaction “C(y, n”) with an end-point energy of 206 MeV We chose this target for two reasons. First, there is an ample supply of total cross section data in the literature which provides a check on detection efficiencv. Second, the pion angular distribution can be predicted reliably at low energy and this provides a mechanism for gauging the angular response of the spectrometer. We report here some preliminary results of the commissioning runs. The photon tagging rate for these runs was about 100 kHz per tagger channel, or about 200 kHz per MeV of tagged photon energy. This modest rate was not limited by the singles rates in the tagger detectors, nor by the trigger rate from Igloo, but rather by random coincidences with the copious number of untagged pions due to the high end-point energy of the bremsstrahlung. With the installation of a new “end-point” tagger at SAL in the near future, the pion background should be further reduced because of the associated reduction in the bremsstrahlung end-point energy. Since the n” spectrometer employs no charged particle veto counters, another major source of background derives from efem pair production in the target, followed by large-angle rescattering within the target material. The random rate was exacerbated by the poor timing stability of the Igloo PMTs and the resulting compromise in coincidence resolving time. These backgrounds are a particular nuisance close J.M. Vogt et al. I Nucl. Instr. and Meth. in Phw. Res. A .166(199-T)lOC11-1 112 the pion threshold, where the cross sections are quite small. To deal with this problem we have devised a software technique we call “masking” which eliminates much of the background. It is most effective in the threshold region, precisely where needed. It is also most effective when Igloo is closed. Actually, the background is substantially reduced in the open mode in any case. The technique is based on the fact that the angular correlation between the decay photons becomes more tightly constrained as the pion energy decreases, and this reduces the available “phase space” for a background event that otherwise could pass as a legitimate event. Using the Monte Carlo simulation code, we compile a list of all combinations of detector pairs which fire in a simulated photopion run of several million pions. This is repeated for each tagged photon energy (typically 50-60 values), and for each target. From this we extract all pair combinations that did not fire, and these form the “masks” for background rejection in an actual Igloo run. Thus, any event that triggers a detector pair in the null-hit region of the masks is rejected, and this is done on a channel-bychannel basis of the photon tagger detector array, i.e. for each tagged photon energy. Because of the statistical nature of the maskgenerating procedure, there is always the possibility that certain block combinations, which are accessible to 3 t t =2 Ri T aJ ‘5 1 0 130 l*,*(1***1) ’ photon pairs but with very small probability. receive no hits in the Monte Carlo simulation and are assigned to the null-hit region of the mask. This in turn could lead to the rejection of legitimate pions. In order to reduce the chances of such an occurrence, we compute each mask 2MeV above the actual tagged photon energy, in effect reducing the number of null-hit patterns in the thinly populated regions. Of course. the price of this extra caution is an increase in the amount of background permitted to slip through the masks. Fig. 18 demonstrates the effectiveness of the background rejection technique, showing the raw yield curves within 10MeV of threshold, uncorrected for detection efficiencies, etc. The upper and lower curves represent the data without masking and with masking, respectively. A direct measure of the background contamination can be seen in the region below n” threshold, and here the masking has reduced the background to about l/7 of its original strength. As one moves to higher energies above threshold, the background rejection gradually becomes less effective, as is evident from Fig. 18. This simply reflects the increase in the available “phase space” for legitimate n” events and therefore a corresponding reduction in the null-hit patterns. Thus, unlike the unmasked situation, the residual target related background under the yield curve will now display an energy dependence. This dependence is readily deduced from the difference between the unmasked and masked data, together with the knowledge that the unmasked background is flat. In Fig. 19 we compare the total cross section for “C(y,n”), measured with Igloo in the closed mode, with a montage of available data from Mainz and Saclay (references are given in Ref. [4]). The visual agreement indicates that Igloo is operating according to expectations; in particular, the pion detection efficiencies from the Monte Carlo simulations are reliable. Turning to the open mode configuration, which tests the angular response, there are two quantities to be considered, the invariant-mass distribution and the belt-hit patterns. The invariant-mass distribution reconstructs the pion mass according to the general relation to t tt+, , , Ey , , 140 135 , , , m, , = [2E,&(l - cos @Ycy)]l” 14! WV Fig. 18. Preliminary yield curves for “C(y,n”) within 10 MeV of threshold, demonstrating background rejection using the software “masks” as described in the text. The upper data are the raw yields, while the lower data have been masked. Threshold is E, = 135.8 MeV. Near threshold, the procedure reduces the background by roughly a factor of 7. These data were taken with Igloo in the closed mode. where GY.,is the opening angle between the rr” decay photons, and E,, are the photon energies, This relation is useful because it incorporates all the measured quantities in a single quality factor, namely the FWHM of the distribution r_. To give an example, for EY = 160 MeV we obtain r (Igloo) = 47 MeV , J.M. Vogt et al. I Nucl. Instr. and Meth. in Phys. Res. A 366 (1995) 100-114 175 while the Monte Carlo simulation I I I I 1 113 predicts Tm(M. Carlo) = 44 meV, 150 in satisfactory agreement. As noted in Section 7.4, the belt hit patterns do not rely on the measured energies El,? to give information on the pion angular distribution, but they do rely on a correct understanding of the “shower leakage” phenomenon which can adversely effect the distribution. In Fig. 20 we compare the observed pattern for “C(y. n”) at E, = 160 MeV (solid line) with the simulated pattern (dashed line), normalized to the same total number of events. The model cross section includes excitation of the 4.43 MeV state of “C in the manner described in Ref. [4]. This comparison, while preliminary, is further confirmation that Igloo is operating according to design. i 125 9$LgwQm i I 4 140 145 9. Conclusion I I I 150 155 160 16!5 Ey We’4 Fig. 19. Total cross section for ‘*C(y, P”). Solid circles are the present results. Open squares are a montage of four measurements from Saclay and Mainz, as cited in Ref. [4]. I I I I I I __I Fig. 20. Belt hit patterns for “C(y. PO) at E, = 160 MeV with Igloo in the open mode. The solid line is the experimental pattern, where the vertical bars indicate the statistical errors for each belt. The dashed line is the Monte Carlo prediction, normalized to the same total number of events. The contribution of the 4.43 MeV (2’) state has been included in the simulation with strength as deduced in Ref. [4]. Note the distinction between these patterns and those of Fig. 17. A large-acceptance m” spectrometer, for taggedphoton measurements of nuclear photoproduction, has been constructed and commissioned. The instrument was designed to permit both total and differential cross section measurements in the low excitation region, roughly O-25 MeV above threshold. The pion detection efficiency, as well as the angular and energy resolutions, have been studied using Monte Carlo simulations based on a carefully modelled description of the spectrometer. Preliminary analysis of the commission of “C(y, -‘I) confirm the predicted efficiencies and energy resolution, while the belt-hit patterns have demonstrated the required angular sensitivity for future differential cross section measurements. Background from high energy (untagged) pions and genuinely tagged e+e- pairs are evident in the experimental data. In lieu of veto counters, the chargedparticle background was rejected in software by means of Monte Carlo generated “masks”. While this rejected much of the background, there is still room for improvement, particularly with respect to the untagged pion background. This situation should improve with the installation of the “end-point” tagging spectrometer in the near future. Acknowledgement We gratefully acknowledge the support given this project by Professor Ed Booth of Boston University, whose material assistance was instrumental to its successful completion. This work was supported in part by the Natural Sciences Canada. and Engineering Research Council of [Z] J.M. Vogt et al.. Nucl. Instr. and Mcth. A324 ( 1993) IYX [3] C.R. Wuest et al.. Nucl. Instr. and Meth. A239 (19X5) 467. [4] J.C. Bergstrom. Phys. Rev. C50 (1991) 2979. 1-51 H. Strciher et al., Nucl. Instr. and Meth. A269 (1988) 56X References [I] L.O. Dallin, Proc. 2nd Europe Part. Accel. Conf.. Nice, June 1990, eds. P. Marin and P. Mandrillor (Editions Frontitrs) p. 1260.
© Copyright 2026 Paperzz