Nuclear Instruments and Methods in Physics Research A281 (1989) 517-527 North-Holland, Amsterdam
COMPOSITE CHARGED PARTICLE DETECTORS WITH LOGARITHMIC ENERGY RESPONSE FOR LARGE DYNAMIC RANGE ENERGY MEASUREMENTS M .M . FOWLER'1, T.C . SANGSTER 2), M.L . BEGEMANN-BLAICH ", T. BLAICH'), J.A . BOISSEVAIN'1, H.C . BRITT 2), Y.D . CHAN 3), A. DACAL ", D.J. FIELDS 2), Z . FRAENKEL 5), A. GAVRON'1, A. HARMON 3), B.V . JACAK' 1, R.G. LANIER 2), P.S . LYSAGHT' 1, G. MAMANE 5), D.J. MASSOLETTI 2), M.N . NAMBOODIRI 2), J. POULIOT 3), R.G . STOKSTAD 3), M.L . WEBB 6) and J.B . WILHELMY ' 1 1 ) Los Alamos National Laboratory, Los Alamos, NM 87545, USA -) Lawrence Livermore National Laboratory, Livermore, CA 94550, USA 31 Lawrence Berkeley Laboratory, Berkeley, CA 94720, USA °) Instituto de Fisica, UNAM, Mexico, DF 01000, Mexico 5) Weizmann Institute of Science, 76100 Rehovot, Israel h) Dynamics Technology, 21311 Hawthorne Blvd, Torrance, CA 90503, USA
Received 28 March 1989 We have developed an array of detectors to identify charged particles produced in heavy ion reactions. The array, which consists of eight individual detector modules and a forward hodoscope, subtends a solid angle of 0.58n and covers 62% of the reaction plane in laboratory coordinates. Each of the eight identical modules has an active area which extends 13 ° above and below the array plane with additional limited coverage between 13 ° and 26 °. Each module measures the position, energy and velocity of charged particles over a dynamic range which extends from minimum ionizing protons with energies up to 200 MeV to highly ionizing fission fragments with Coulomb-like energies . Position and time-of-flight are measured with low pressure multiwire proportional counters (MWPC) . Total energies for heavier ions are obtained from large ion chambers . Energy and position measurements for more energetic lighter ions which pass through the ion chambers are made with segmented phoswich arrays . The forward angle hodoscope is a 34-element array of phoswich detectors mounted symmetrically around the beam axis. These detectors are sensitive to beam velocity particles (EIA > 10-40 MeV/A) and capable of elemental resolution from protons to Z = 23 . 1. Introduction As studies of fission and charged particle emission induced by heavy ions collisions have moved into higher energy regimes, the requisite detector systems have become more complicated due to the wider ranges in energy, multiplicity and types of particles to be measured. Also, interest in high energy heavy ion reactions has shifted away from the heavy products associated with low energy fission and focussed on the lighter particles which provide insight into reaction mechanisms involving high energy deposition in the target nucleus. These reactions typically involve high charged particle multiplicities with particle energies ranging from a few to several hundred MeV and particle masses ranging from protons to heavy, target-like residues . Sensitivity to such a wide range of particle mass and energy is, perhaps, the most challenging aspect of detector design . The detector modules that have been jointly developed at Los Alamos National Laboratory (LANL), Lawrence Livermore National Laboratory (LLNL) and 0168-9002/89/$03 .50 © Elsevier Science Publishers B.V. (North-Holland Physics Publishing Division)
Lawrence Berkeley Laboratory (LBL) provide excellent particle identification over a broad dynamic range. To achieve a large dynamic range, these detector modules were designed to provide a logarithmic energy response . This response is obtained by linking together a series of progressively thicker (i .e., greater stopping power) detector elements so that only the most penetrating particles see the element(s) with the highest stopping power. With the proper selection of detector elements, such an arrangement is capable of triggering on and identifying several hundred MeV protons as well as several hundred keV/A heavy residue fragments . In our modules, three of the detector elements provide a fast rise-time signal for timing information . The combination of a time-of-flight an up to four dE/dx measurements allows us to identify virtually all of the incident particles. Recent experimental runs at the LBL Bevalac utilized an array of eight such detector modules. The full array is known as the "PAGODA" and the general detector layout is shown in fig. Ia. The PAGODA array consists
M. M. Fowler et al. / Composite chargedparticle detectors
b o E-E Phoswich Telescope Array J Ionization Chamber
Fig. 1 . (a) the PAGODA detector layout at the LBL Bevalac Low Energy Beam Line . The centerlines of the modules are 36o, 72°, 108° and 144° on either side of the beam . The hodoscope is located in the wedge-shaped target chamber extension downstream of the target . (b) A detailed layout of a single PAGODA detector module.
of eight identical modules (the modules and their components are interchangeable) mounted symmetrically around a target chamber in a planar geometry . Each detector module has four separate gas elements and the forward six modules are backed by identical arrays of nine (3 X 3) scintillator telescopes. Fig . l b provides a more detailed view of a single module and its components. The latest addition to the PAGODA is a forward angle hodoscope which consists of 34 scintillator telescopes mounted symmetrically around the beam axis downstream of the target . The position of the hodoscope can be seen in fig. l a ; it is located in the large wedge-shaped extension on the downstream side of the target chamber. The three gas elements closest to the target consist of two multiwire proportional counters (MWPC) separated by a drift region and operate in a common gas volume of isobutane at a pressure of 2 .5 Torr . The MWPCs provide timing and position information while the drift region is operated as a proportional counter (PC) giving a dE/dx measurement . The fourth gas element is a large volume, axial-field ion chamber (IC)  which, at present, is operated with 50 Torr of CF4 . At this pres-
sure, it provides primarily dE/dx information for lighter particles (Z<10 and total kinetic energies greater than 2-4 MeV/A) and residual energy information for slower, heavier particles . Behind each of the forward six ion chambers are identical nine element (3 X 3) plastic scintillator telescope arrays . The 54 scintillator telescopes are composed of fast/slow plastic and operated in a phoswich mode  . This combination allows the energy measurement and charge and mass identification of fast light particles which leave little or no signal in the gas elements, and residual energy measurement for particles which stop in the fast plastic . In the next section, we will discuss the individual gas counters, including trigger thresholds, MWPC timing and position resolution, d E/d x resolution in the PC and IC, and the characteristics of the IC counting gas, CF4 . The large angle phoswich arrays and the forward angle hodoscope will be described in section 3 . Relative light collection efficiencies using a number of different reflective materials to cover the polished plastic surfaces will also be presented. Finally, in section 4 we will discuss calibration methods for the individual detector elements and particle identification techniques using these elements in various combinations .
2. Gas counters 2.1 . Multiwire proportional counters
Particle timing and position information are obtained in each PAGODA module from a pair of multiwire proportional counters (MWPCs) . The active areas of the front (nearest the target) and the back MWPC are 8-by-16 and 16-by-16 cm, respectively . The MWPCs are positioned such that the 8-by-8 cm central region of the front MWPC subtends approximately the same solid angle (1 .67% of 41r) as the full 16-by-16 cm area of the back detector. In laboratory angular coordinates, the solid angle coverage corresponds to 26' in the array plane and ±13 ° out of the array plane . The two additional 4-by-8 cm regions of the front detector increase the out-of-plane coverage to ± 26 ° or to 3 .42% of 4ir . However, particles incident on these regions are identified only by the timing, position, and energy loss information from the front MWPC since they cannot reach the back MWPC or the ion chamber . Each of the MWPCs is a two-stage avalanche counter based on designs by Breskin  . Breskin counters are typically made up of several wire planes and operated with or without preamplification states, as illustrated in fig. 2. Such a configuration provides a fast timing signal as well as good sensitivity to light ions (small dE/dx in the detector). We have found, however, that the timing resolution can be improved without sacrificing light
M. M. Fowler et al. / Composite charged particle detectors
a b c d -v2 0 +V1 0 w PA A
a b c d e 42 0 +V1 0 42 w~r PA A PA
Fig. 2. (a) A MWPC amplifier (A) with a single preamplification (PA) step . The primary charges created in the PA region are preamplified, transferred to the A region and further amplified. The contribution to the total amplified charge from the primary electrons deposited in the A region is negligible . (b) A MWPC with two preamplifying steps; the total amplified charge is a factor of 2 greater than in scheme (a). This figure is reproduced from ref. . particle sensitivity by interchanging the cathode and anode planes and replacing the cathode wire plane with a 200 wg/cmz aluminized mylar foil, as shown in fig. 3 . Replacing the cathode wire plane with a foil slightly increases (2%) the transmission of each MWPC detector and, more importantly, creates a more uniform electric field in the preamplification (PA) region . The more uniform field results in a more uniform acceleration of the primary electrons in the PA region which gives rise to the improved timing characteristics . We should also note that having the two PA regions "back-to-back" reduces the timing jitter due to the particle flight time through the detector . We have found that the optimum operating potential in this configuration occurs when the anode and cathode voltages are symmetric with respect to the position wire planes at ground . With isobutane at 2.5 Torr, typical
a b c d e +V1 0 -V2 0 +V1 A PA PA A
Fig. 3. The "inverted" Breskin counter with a foil cathode. The more uniform electric field on either side of the cathode (plane c) results in a more uniform acceleration of the primary electrons in the PA regions. This leads to the improved timing characteristics without sacrificing light particle sensitivity.
voltages for the anode and cathode are + 320 V and - 320 V. All four wire planes are wound with gold-plated tungsten wire ; the wire diameter for the two anode planes is 10 p m and for the two position planes is 20 jim. The wire spacing on each plane is 1 mm giving each MWPC a transmission of 94%_ The wire is wound on 1 .6 mm thick epoxy-fiberglass (G-10) frames and the distance between the wire planes is 3.2 mm . Three of the wire planes have wires which run vertically, while the wires run horizontally on the fourth plane. The wire orientation is, of course, orthogonal for the position planes adjacent to the cathode foil . The eight front MWPCs extend into the target chamber (see fig. la) and are mounted directly on the front of each module body without a separate housing. This allows them to be positioned together as closely as possible. The MWPCs are made gas tight by stacking the wire planes with the appropriate spacers and then sealing the entire stack as a single unit. The sides of the stack are sealed using a silicone compound (RTV-21) manufactured by General Electric . The RTV-21 can be cut or peeled off and the MWPC disassembled for repairs. The front entrance window is 40-50 wg/cmz of stretched polypropylene mounted on an aluminum frame with support wires (> 99% transmission) which then attaches to the first spacer in the stack. Position measurements are made based on signal propagation time differences through delay lines mounted directly on the position wire planes . We use commercially available 50 0 tapped delay lines in single in-line packages. Each package is approximately 4 cm long and provides 10 taps with 5 ns of delay per tap. There are two 10 tap packages mounted along the 8 cm dimension and four packages mounted along the 16 cm dimension. Since the wire spacing is 1 mm, four wires are connected to each delay line tap. The delay line signals are coupled to hybrid amplifiers (Motorola MWA 110) which are also mounted directly on the wire planes . These amplifiers provide a voltage gain of four and are impedance matched to the output cables (RG174) . The position resolution is better than 1 .0 mm when constant fraction discriminators are used to develop the delay line signals. The anode wire diameter is considerably smaller than the position wire diameter to ensure the maximum possible electron gain at the anode wires. The diameter of the position wires is a compromise between particle transmission and electron transfer efficiency between the PA and the A stages . A smaller diameter would cause more of the preamplified charge to be lost on the position wires before entering the avalanche region . The two anode planes are electrically connected and capacitively coupled to an output connector. There is no on-board amplification as with the position signals. The rise time of the anode signals is typically about 5 ns .
M. M. Fowler et al. / Composite charged particle detectors
52 0 1 .25
2 .2 . Front proportional counter
The MWPCs and the drift region in between are all within a common gas volume . In order to obtain some useful energy loss information from the low pressure isobutane between the MWPCs, we installed a pair of collection wires (75 wm gold-plated tungsten) centered horizontally and strung vertically 1/3 and 2/3 of the way through the drift region to collect charge generated by ions passing between the MWPCs. With a potential of 300-500 V on the wires, the drift region operates as a proportional counter (PC) . In order to prevent charge injection from the MWPCs, we have isolated the drift region using thin (40-50 wg/cm2), aluminized (10 hg/cm2) polypropylene windows. In tests at the LANL Van de Graaff and the LBL 88 in . cyclotron, we have shown that the energy response of the PC is linear up to 10 MeV (see fig. 5) and that adequate signals can be obtained with an energy loss >_ 200 keV in the drift region. The energy resolution for particles on the high energy side of the Bragg peak is about 8% . Fig. 6 shows a counter plot of the potential gradient made over a median plane of the PC with a wire potential of 400 V. Although the counter has a high charge collection efficiency, the isopotential curves show that the potential gradient is not uniform throughout the detector . This results in a small position-dependent defect in the measured pulse heights. Close to the wires and near the walls, the pulse height is as much as 5% lower than in the main body of the detector. Fig. 7 shows a plot of the front proportional counter signal vs the respective MWPC-to-MWPC time-of-flight for all eight detector modules (the data have been added together) from a data run at the Bevalac using 100 MEV/A Nb + Au . The proportional counter signals have been gain matched and the individual detector module times have been aligned. The charge lines for Z = 2 to Z = 10 can be easily separated for ions with
0 .75 0 .50 0 .25 0 .00
Incident Energy (MeV)
Fig. 4. The position and anode efficiencies for light ions which reach the fast plastic elements of the phoswich arrays. The anodes are 100% efficient for ions with Z > 7 and the positions are 100% efficient for ions with Z > 8. The data are from a recent experimental run at the LBL Bevalac.
The two anode signals are used to measure the time-offlight between the two MWPCs (a distance of 18 .0 cm) and to start the position TDCs . The anode signals also provide one of the primary triggers for the data acquisition system . In the laboratory, the TOF resolution for 252Cf alpha particles is about 500-600 ps FWHM . This value does not reflect corrections for the detector module geometry and implies that the intrinsic resolution of a single MWPC in this configuration is about 400 ps FWHM . Although we have not performed an independent study of the anode and position efficiencies as a function of ion charge and energy, we do have a direct measure of the efficiencies for those particles which reach the fast plastic in the large angle phoswich detectors . These efficiencies are shown in fig. 4 as a function of the incident ion energy . For those light particles which do not reach the fast plastic but for which we have positive charge identification, we should be able to use the MWPC anode signals to estimate the efficiency by comparing dE/dx measurements . For instance, from the data in fig. 4, a 36 MeV Be ion has about the same anode efficiency as a 90 MeV B ion and a 160 MeV C ion (80-85%) . The calculated energy losses in the MWPC (i .e ., the anode pulse heights) for these ions and energies is between 20 and 22 keV, implying that the MWPC efficiencies are strictly a function of the ion dE/dx in the MWPC and that similar dE/dx values (or distribution of values) correspond to similar efficiencies. It should be noted that the position efficiencies are artificially low relative to the anode efficiencies in fig. 4. For this data run, the anode discriminators were located very close to the detectors while the position discriminators were about 100 ft away ; this caused the analog position signals to be attenuated by nearly a factor of two relative to the anode signals .
a) ~ . C W a>
e 4 2 .5 Torr Isobutane
2 as U
Channel No . Fig. 5 . Proportional counter pulse height versus calculated energy loss for Si, S, Ar, Fe and Kr ions at several different incident energies . The energy response of the proportional region is remarkably linear over a broad range of ion energies . The data was taken at the LBL 88 in. cyclotron .
M. M. Fowler et at. / Composite charged particle detectors
2.3. Ionization counter
0 2 4 6 8 10 12 14 16 Distance along detector axis (cm)
Fig . 6 . Isopotential plot of the field gradient in the front proportional counter with 400 V on the wires. The 40 V contours (10%) represent a median plane cross section of the detector.
energies less than about 3 MeV/A . The two bands in the upper right are fission fragments from the gold target and heavy (A > 100) target residues . Since these heavier ions must penetrate only about 500-600 wg/cmz of material to provide a satisfactory dE/dx signal and a TOF, they can be identified with energies as low as 200 keV/A .
The ionization chambers behind the back MWPCs were designed to provide only energy loss information . The anode and Frisch grid are perpendicular to the ion flight path and located at the back of the ion chamber. In this configuration, the windows and housing are operated at ground potential (cathode). The entrance window separating the low pressure isobutane in the forward counters from the higher pressure ion chamber CF, is a 200 l-Lg/cm z aluminized mylar foil supported by a mesh of 6 mil tungsten wires spaced 100 mil apart (88% transmission). The Frisch grid is fabricated from electroformed nickel mesh that has 20 lines per inch and a transmission greater than 90% . The anode is a 400 pg/cmz aluminized mylar foil stretched over a G-10 frame that spaces it 1 cm from the Frisch grid . The sensitive volume is about 20 cm long and is presently operated at 50 Torr . Fig. 8 shows the result of an electrostatic calculation made over a median plane of the IC with typical voltages for low pressure operation. Again, the isopotential curves show that the potential gradient is not uniform within the counter and is fairly small along the central axis . But, as with the PC, the chamber has a high charge collection efficiency and excellent energy resolution. However, the electrons have varying drift path lengths to reach the anode and the electric field is nonuniform over virtually all paths. Consequently, the ion chamber has quite poor timing characteristics and collects charge more slowly than conventional ion chambers where the electron drift is perpendicular to the incident ion path . Ions which stop near the anode or pass completely through the detector will produce signals more quickly than ions which stop near the entrance window. In some instances, we have used this
MWPC TOF (ns) Fig . 7 . A plot of the PC signal vs the TOF across the drift region for all eight detector modules. The individual PC signals have been gain matched and the times have been aligned using 252 Cf fission fragments . The individual spectra from the eight detectors have been added together. The data is from a 100 MeV/A Nb+Au run at the LBL Bevalac .
Fig. 8 . Isopotential plot of the field gradient over a median plane of the back ion chamber .
M. M. Fowler et al. / Composite charged particle detectors
scintillators behind the most backward two modules these are simply sealed with aluminum plates. 2.4. Gas supply system
L d a E s U C 0
0 MWPC TOF (ns)
Fig. 9. A plot of the IC signal vs TOF for detector modules located 72 ° from the beam . The IC signals have been gain matched using z52Cf fission fragments and the individual charge bands in each detector . The data are from a 75 MeV/A Nb+Au run at the LBL Bevalac.
difference in drift time to the anode as a measure of the ion range in the detector. Fig. 9 shows a plot similar to fig. 7. Here, the signals from ICs at 72 ° have been gain-matched, added together and plotted vs the measured TOF between the MWPCs. Charge identification can clearly be made up through Z = 10 for ions which stop in the IC and for a large fraction of the particles which actually pass through the IC anode. We have chosen to use CF, in the ion chamber based on its high stopping power and electron drift velocity . In section 4, we will discuss the behavior of some of the other gasses we have tested in this counter. With CF, we have found that the chamber typically exhibits about 100 keV of system noise due primarily to microphonic pickup by the large anode foils of cryo pump compressor vibrations. The FWHM resolution for 5.5 MeV alpha particles is 3.5% (200 keV) . The reduced electric field between the grid and anode gives the most efficient electron collection when adjusted to be about 2-3 times the field along the axis of the chamber. The forward four ionization chambers do not have exit windows. These modules are backed by large vacuum tight boxes which each contain a 3 X 3 array of fast/slow plastic phoswich detectors. These boxes will be described in more detail in the next section. The two modules located on either side of the beam at 108' have the remaining two phoswich arrays located immediately behind them operating in ambient air. The exit windows for these two modules are made of 0.9 mil aluminized mylar supported by a machined grid of 0.25 in. thick aluminum . The window is capable of withstanding a 1 atm differential pressure and the transmission of the support grid is 90%. Since there are no
The four modules on either side of the beam are operated as a single unit with two gas supplies for each unit ; one provides isobutane for the MWPC/proportional region and the other provides CF, for the ionization counters . The pressure in each part of a gas system is measured with capacitance manometers (MKS type 222B). Manometers mounted directly on the detectors themselves control individual pressure regulating valves in each gas system . The four detectors attached to each system are run in parallel with the regulating valve on the inlet (supply) side of the low pressure isobutane systems and on the outlet (pump) side of the higher pressure CF, systems. The flow rate of isobutane through the four MWPCs in each unit is adjusted to about 2 Torr 1/min while the flow through the four ICs is about 30 Torr I/min. Isobutane of commercial purity (98.5%) is passed through a molecular sieve (MS-5A) to remove moisture before delivery to the gas systems; high purity CF, (99.5%) is passed through an oxygen scrubber (Oxisorb) for the ionization counters . 3. Scintillator arrays 3 .1 . Large angle phoswich detectors
The Scintillator arrays located behind the forward six ionization chambers are made up of 54 identical 1 mm fast/26 cm slow plastic Scintillator telescopes operated in a phoswich mode. Each phoswich module is a truncated pyramid designed to be located 60 cm from the target and subtend 8 ° horizontally and vertically . The fast plastic is BC412 and the slow plastic is BC444, both commercially available from Bicron . We use Amperex (XP 2202-FL-B) and Hamamatsu (82154) phototubes which are optically coupled to machined Lucite light guides. The light guides are permanently attached to the slow plastic to form a single unit . The active length of the phoswich modules was chosen to stop a 200 MeV proton . The thickness of the fast plastic was chosen as a compromise between p, d, t resolution and the charged particle identification threshold; a thicker fast element would increase the p, d, t resolution but raise the energy threshold for positive charge identification in a AE--E mapping of the fast and slow signals . As operated in our experiments, the phoswich detectors provide unit charge resolution for virtually all of the charged particles which reach the slow plastic and p, d, t resolution up to a proton energy of about 20 MeV.
M. M. Fowler et al. / Composite charged particle detectors
Table 1 Results of the scintillator reflective covering tests using 120 MeV alpha particles
Millipore Teflon tape Aluminized mylar TiO-doped epoxy
ta c 0
Short gate (AE) Fig. 10 . A plot of the fast (A E) vs the slow (AE+E) plastic gates for all 54 large angle phoswich detectors. The individual spectra have been gain matched and added together . The data are from a 50 MeV/A Nb+Au run at the LBL Bevalac. In our experimental configuration, both the long and the short gates are started approximately 30 ns before the peak in the fast component of the phototube output . The short gate is 70 ns wide and includes the majority of the fast component and the leading edge of the slow component. The long gate is typically 325 ns and includes all of the fast component and about half of the slow component. By gating all of the detector charge integrating ADCs identically, we are able to gain match every detector/phototube combination onto a single spectrum and apply a single energy calibration . Fig. 10 shows a gain matched spectrum of all 54 large angle phoswich detectors. As was mentioned in the previous section, the forward four phoswich arrays are contained in a vacuum chamber which attaches directly to the back of the ion chambers . The vacuum seals are made on the lucite light guides so that the phototubes are operated outside the vacuum . The fronts of the phoswich detectors are 2-4 cm from the IC anode foil . We have conducted several tests in the lab and at the LBL 88 in . cyclotron and have determined that the presence of the plastic in the IC gas volume does not affect the performance or resolution of the IC . The threshold for a fast scintillator signal is only about 2-4 MeV/A . On the other hand, the remaining two arrays are located behind 0.9 mil mylar IC exit windows. In these modules, the fast plastic signal threshold is considerably higher, about 6-7 MeV/A . One of the other tests which we performed in the course of designing the large angle phoswich arrays was to determine the optimum covering for the polished plastic surfaces. We compared four commonly used materials: (1) millipore filter paper , (2) aluminized mylar, (3) teflon tape, and (4) TiO-doped epoxy. Table 1 summarizes the results of our tests at the LBL 88 in . cyclotron using a 120 MeV alpha beam . A single photomultiplier tube was used to test four identical phoswich
Relative pulse height in the slow plastic 1.00 0.84 0.64 0.30
scintillator telescopes, each covered with one of the materials. The pulse heights in the slow plastic were compared using the phoswich covered with millipore paper as the standard . While millipore paper is clearly the best choice, it is also by far the most expensive. Furthermore, millipore paper is quite brittle and virtually impossible to fold over an irregular surface without cracking . We were able to solve this problem by using distilled water to dampen small sections of individual sheets . Once wet, the paper can be folded quite readily and once dried, the paper retains whatever shape it was given when wet. In this way, all of the large angle phoswich detectors were wrapped with millipore paper. To protect the delicate millipore paper covering from damage during handling, we added a second covering of a heat-shrinkable plastic wrap called Monokote . This material is light weight and readily bonds to itself . It provides a light tight, durable covering and adds very little additional bulk. 3 .2 . Forward hodoscope array
The scintillator array located downstream of the target is composed of 34 identical 1 mm fast/40 cm slow plastic telescopes which are also operated in a
Fig. 11 . A schematic top (a) and side (b) view of the 34-element hodoscope array.
M. M. Fowler et al. / Composite charged particle detectors 8° Hodoscope element
Short gate (AE)
Fig. 12 . A plot of the fast (DE) vs the slow (AE +E) plastic gates for several of the outer 22 hodoscope elements. The spectrum does not show the full ADC range of either gate but rather a higher channel resolution . Charge bands can be identi fied in these elements from protons through Z =17. The data are from a 50 MeV/A Nb+Au run at the LBL Bevalac. phoswich mode . The design of this array was inspired by the plastic cube array (PCA) of Pouliot et al . . Fig. 11 shows a schematic of the array which is located approximately 40 cm from the target . The center module in the 5-by-7 array has been removed for the beam . Each module subtends 4 ° and the array covers from 2 ° to 10 ° on either side of the beam in the plane of the gas modules and from ± 2 ° to ± 14 ° out of plane. The size of these modules was chosen to provide a reasonably high granularity with a correspondingly small probability for multiple hits . The 40 cm long modules will stop 250 MeV protons . We use Hamamatsu phototubes (R580) which are optically coupled directly to the slow plastic . 2° Hodoscope element 50 MeV/A Nb + Au
Using Nb beams at energies between 50 and 100 MeV/A, only the inner 12 modules (i.e ., the 12 modules closest to the beam) see beam velocity fragments with a mass close to Nb . Therefore, these modules are typically run at gains which are a factor of two lower than the remaining 22 detectors. Fig. 12 shows a short gate (A E ) vs long gate (DE+E) plot of several outer modules which have been gain matched and added together . These gains are somewhat lower than the large angle phoswich detectors and consequently the proton line is below threshold . Fig. 13 shows the same plot for one of the inner 12 modules for which the phototube gain was adjusted to allow detection of elastically scattered Nb particles; there is clearly elemental resolution through Z=23 . 4. Calibrations During the last two years, we have performed a number of calibration experiments at the LBL 88 in . cyclotron and the LANL Van de Graaff using a wide variety of beams and energies in an effort to understand the response of each segment of a PAGODA detector module . Since the energy response of the detector module is logarithmic and sensitive to such a large dynamic range, the subset of detector elements we use to identify a particular ion is dependent on the initial ion kinetic energy . For instance, a 250 MeV carbon ion will trigger the MWPCs with considerably less than 100% efficiency and leave only a very small signal in the PC . Particle identification would have to be made using the IC and the fast plastic signals. On the other hand, identification of heavy, slow target residues can only be made based
Particle as Z > 48 with velocity Information Unit charge ID with velocity information
L .LM 300 t
Particle ID as fission-mass with no additional information
y 200 3 C. 100 U
Short gate (AE)
Fig. 13 . A plot of the fast (0 E) vs the slow (AE+E) plastic gates for one of the inner 12 hodoscope elements. The spectrum represents the full ADC range and shows the elastic beam spot (Nb) as well as individual charge bands from alphas (few protons trigger at these lower phototube gains) through Z = 23 . These data are also from a 50 MeV/A Nb+Au run at the LBL Bevalac.
MWPC TOF (ch#)
Fig. 14 . A plot of the PC vs TOF grid constructed from detailed calibration data taken at the LANL Van de Graaff. This grid calibration provides a unique charge given a measured PC pulse height and time-of-flight between the MWPCs. For particles where Z > 48, this method provides only an incident velocity . In other regions of the PC vs TOF grid, no particle identification can be made and we must rely on other detector elements of the PAGODA module to provide positive particle identification .
M. M. Fowler et al. / Composite charged particle detectors
on the TOF and the PC signal since they rarely reach the IC . However, except for fast, light particles (which are identified in the phoswich counters) all other charged particles leave a signal in the PC . Therefore, a thorough understanding and careful calibration of the PC can provide some mass identification for virtually all of the detected particles. The identification scheme using the PC is based on a series of velocity calibration measurements made at the LANL Van de Graaff . A number of light and heavy ion beams at various energies were scattered off several different target foils into a detector module which was mounted 30 ° from the beam . In this way, a nearly continuous range of projectile and target ion energy losses and TOFs were measured in the PC . These data have been reduced to a two-dimensional grid of PC pulse height vs measured TOF (see fig. 14). For a given PC pulse height and measured TOF, the grid determines the particle charge. A second grid of pulse height vs TOF has also been constructed which provides the initial ion velocity. 252Cf fission fragments are used to properly gain match and time align the experimental data from each module to the grid calibration . For a significant fraction of the lighter ions this PC vs TOF grid identification provides either no charge assignment or an unreliable value due to the lack of resolution . For these particles, improved or positive charge identification can be made from charge bands observed in the IC or the scintillator detectors. In this way, virtually all of the lighter particles (Z< 8) can be identified using some combination of detector elements to provide a DE-E or a AE-TOF mapping. Fig. 9 shows that charge identification for the lighter particles can be made using the 1C-TOF mapping. Fig. 15 shows a plot of the IC vs the fast plastic signals for the 36 phoswich telescopes behind the four forward ICs. In both of these figures it is clear that positive particle identification can be made for these lighter elements . Although fission fragment and target residue mass resolution in the PC-TOF grid is quite poor, in those instances where these larger fragments do reach the ion chamber, a residual energy measurement allows us to significantly improve the mass assignment . The IC has been studied in several different experiments under a number of different operating configurations. We have been able to demonstrate that the energy response for ions which pass completely through the detector is linear as a function of ion charge and chamber pressure for energy losses up to several hundred MeV (see fig. 16). We have also been able to demonstrate that there is no pulse height defect due to Zdependent changes in the ionization density through the detector . We have found, however, that when the detector is operated with CF, at pressures of 100 Torr and above, there is a considerable change in the charge collection
m a E to t
Phoswich short gate Fig. 15 . A plot of the IC pulse height vs the fast (0 E) plastic gates for the 36 phoswich elements contained inside the ion chamber gas volume (the four most forward ion chambers) . This spectrum is the sum of the 36 individual spectra. The particles in these spectra were required to have stopped in the fast plastic (i .e ., identical fast and slow gates) and multiple array (3 x 3) hits behind a single IC were rejected . The number of multiple hit rejections was significant (40%) for the forward two phoswich arrays . efficiency as a function of the ion range in the detector . Ions which stop near the front window have a smaller measured pulse height than the calibrations would indicate . This effect can cause ambiguities in the ion identification since two ions which deposit the same energy in the counter but have different path lengths produce different pulse heights. The effect is shown in fig. 17 where the pulse height measured for alphas and fission fragments from 252Cf is plotted as a function of the range of the ions in the counter. These data were obtained by measuring the pulse height at constant 400 300
w b 0 m U U
200 100 0
Channel No . Fig. 16 . Energy response of the ion chamber to a number of different light and heavy ions at various incident energies with 100 Torr of CF, The data were taken at the LBL 88 in . cyclotron and the ions included Si, S, At, Fe and Kr.
M.M. Fowler et al. / Composite chargedparticle detectors
5 10 15 Range in the Ionization Chamber (cm)
Fig. 17 . The response of the ionization chamber to 252 Cf alpha particles and fission fragments as a function of their range in the counter. The data are for CF4 and isobutane. The curve through the data points is to guide the eye.
reduced field while changing the pressure in the ion chamber. The range in the counter was then calculated from the pressure using energy loss codes. It should be pointed out that the ranges used were the residual ranges of the ions after passing through the MWPCs, the proportional counter and the IC entrance window . The pulse heights of the various ions were normalized to each other using pressures in the counter that allowed the ions to just reach the Frisch grid. No corrections were made for differences in the shape of the dE/dx curves for the different ions . As the path length of an ion decreases, the pulse height also decreases. Ions with a path length of 3 cm produce about 60% of the pulse height of ions with a path length of 17 cm . We have verified that this pulse height defect is a property of the CF, counting gas rather than an actual flaw in the design of the counter. If the CF, is replaced with isobutane or methane, the pulse height is nearly independent of the track length . This is also shown in fig. 17 where the curve for alpha particles in isobutane shows almost no change with path length . We have had some success in improving the performance of the CF4 by passing it through molecular sieve 5A and then through an oxygen scrubber . However, even when this is done, the pulse height is still a function of ion range. This difficulty can be minimized by operating the detectors with CF, at pressures of 100 Torr or less. Although we have considered using isobutane and P-10 in the ion chambers (these gasses have stopping powers which are similar to CF4 ), the electron drift velocities are far too slow for such large counters. Methane, on the other hand, has perhaps the best characteristics (high collection efficiency and drift velocity) of all the gasses we have considered . Unfor-
tunately, the stopping power of methane is only about one-third that of CF4 and running with a higher methane pressure would require a thicker entrance window . The calibration techniques for the phoswich detectors involves linearizing the fast and slow signals (i .e., decoupling the light outputs in the fast and slow gates), gain matching and applying an energy calibration to both signals. Figs . 10 and 12 show the gain matched spectra from both the large angle arrays and the forward angle hodoscope. Central to the calibration and understanding of all of these detectors has been and will be the use of a number of energy loss codes. We have compiled seven dE/dx codes which are applicable over different ranges of ion mass and energy into a single master code which integrates the output from any single code into time-offlight, range and energy loss information . We primarily use the DONNA code , a code based on the work of Ziegler et al . , and a code based on the Northcliffe and Schilling tables  to model the energy losses through our detectors.
5. Summary We have constructed an array of detectors which is capable of identifying heavy ion reaction products over a solid angle of about 13% of 4nr. The unique feature of this array is its large dynamic mass and energy range for particle identification. The individual detector modules are sensitive to 200 MeV protons, as well as 200 keV/A heavy residues (A > 100). The large dynamic range is achieved by using a logarithmic approach ; succeeding elements in the detector modules have progressively higher stopping powers . Each module is composed of two low pressure, position sensitive MWPCs which measure a time-of-flight over a low pressure drift region . The drift region is configured to operate as a proportional counter. A large, higher pressure, axial field ion chamber behind the second MWPC provides the primary dE/dx measurement for most of the heavier (Z >_ 6-8) particles. Light particle identification is made using segmented arrays of fast/slow phoswich counters mounted directly behind the ion chamber anodes. We have shown some of the typical detector responses from data runs at the LBL Bevalac using 50-100 MeV/A Nb + Au reactions. It is clear from these data that the logarithmic concept provides a superior tool for studying these types of reactions. Furthermore, we have shown that a fast/slow plastic phoswich hodoscope array can provide excellent particle identification for projectile-like fragments with energies between 50 and 100 MeV/A .
M. M. Fowler et al. / Composite charged particle detectors Acknowledgements We would like to thank Gary Westfall, Craig Thorn and Hans Sann for useful discussion concerning the development of the scintillator arrays and the concepts of a logarithmic detector. This work was performed under the auspices of the US Department of Energy by the Lawrence Livermore National Laboratory under contract number W-7405ENG-48 and by the Los Alamos National Laboratory under contract W-7405-ENG-36 . Additional support was furnished by the Director, Office of Energy Research, Division of Nuclear Physics of the Office of High Energy and Nuclear Physics of the US Department of Energy under contract DE-AC03-76SF00098. References  G.D. Westfall et al ., Nucl. Instr. and Meth. 238 (1985) 347.
 A.N . James et al ., Nucl . Instr. and Meth . 212 (1983) 545.  D.H . Wilkinson, Rev. Sci. Instr . 23 (1952) 414; D. Bodansky and S.F . Eccles, Rev. Sci. Instr. 28 (1957) 454; C. Pastor et al ., Nucl. Instr. and Meth . 212 (1982) 209. [41 A. Breskin, Nucl . Instr. and Meth . 141 (1977) 505; A. Breskin et al., Nucl . Instr. and Meth . 165 (1979) 125; A. Breskin, Nucl . Instr. and Meth . 196 (1982) 11 ; A. Breskin et al., Nucl . Instr. and Meth . 221 (1984) 363.  Millipore Corporation, Bedford, MA 01730, USA.  Manufactured by Top Flight Models, Inc., 2635 S. Wabash Ave., Chicago, IL 60616, USA . J. Pouliot et al ., Nucl . Instr. and Meth . A270 (1988) 69 . W.G . Meyer, private communication. The energy loss code is based on the Bethe formula. Code is based on : J.F. Ziegler, J.P . Biersack and U. Littmark, The Stopping and Range of Ions in Solids (Pergamon, New York, 1985).  Code is based on : L.C . Northcliffe and R.F . Schilling, Nucl . Data A7 (1970) 223.