MEDEA: a multi element detector array for gamma ray and light charged particle detection at the LNS-Catania

Nuclear Instruments and Methods in Physics Research A314 (1992) 31-55 North-Holland

NUCLEAR INSTRUMENTS

&METHODS

IN PHYSICS RESEARCH

Section A

MEDEA: a multi element detector array for gamma ray and light charged particle detection at the LNS-Catania

E. Migneco

a,b,

C. Agodi

a,

R. Alba

a,

G. Bellia

b,

R. Coniglione

a,

A. Del Zoppo

a,

P. Finocchiaro

a,

C. Maiolino

a,

P. Piattelli

a,

G. Raia

a

and P. Sapienza

a

aIstituto Nazionale di Fisica Nucleare, Laboratorio Nazionale del Sud, Viale Andrea Doria (angolo ViaS. Sofia), Catania, Italy

bDipartimento di Fisica dell'Uniuersita, Catania, italy

Received 24 June 1991

A 411'highly granular Multi Element DEtector Array (MEDEA) for "I-rays and light charged particles is described. Its basic configuration consists of 180 barium fluoride scintillator crystals, arranged in the shape of a ball, plus a forward angle wall of 120 phoswich detectors. The inner radius of the ball (22 cm) and the distance of the wall from the target (55 cm) allow the placement of other detectors in the inner volume. The whole detection system operates under vacuum inside a large scattering chamber.

Dedicated electronics has been designed and realized. It includes a powerful hardware second level trigger and preanalysis system, which allows on-line event selection, and a modular VME-bus based data acquisition system. In-beam performances of the system are also described.

1. Introduction

Heavy ion collisions are a very useful tool to study the behaviour of nuclei under extreme conditions. In the low energy regime (E :::;10 MeV lamu) the nuclear mean field is responsible for the· main features of the reaction mechanism, while at high energies (several hundreds of Me V

I

amu) the two body collisions be- tween nucleons of the projectile and the target domi- nate. The intermediate energy region (20 MeV lamu:::;

E :::;200 MeV lamu), that has been made accessible in the last years due to the construction of new accelera- tors, is characterized by the competition andlor coex- istance of the two different behaviours described above and eventually by the appearence of some new phe- nomena (see for example ref. [1]). On the theoretical side some calculations [2,3] indicate that, in this energy regime, values of the nuclear matter density larger than the equilibrium value can be attained.

The evolution of the nucleus-nucleus collisions from the initial stage at high density and temperature to the expansion and decay stage is fast and most of the observables (angular distribution, energy spectra, mass distributions of the reaction products) are mainly re- lated to the final stage of the collision [4]. Therefore specific probes sensitive to the compression phase are needed. An interesting opportunity is given by the production of subthreshold pions or high energy ')'-rays

(E-y;::: 30 MeV) [5]. In fact kinetic approaches which correlate the particle emission to nucleon-nucleon col-

lisions indicate that the production of pions and hard photons occurs in the very first stage of the nucleus- nucleus collision. Photons are particularly interesting, since, at variance with pions that strongly interact with nuclear matter, they interact weakly and thus they constitute a nonperturbed probe.

Another field of interest is the study of nuclei at high temperature [6]. In particular the evolution of the giant dipole resonance (GDR) with increasing excita- tion energy may give useful information on the collec- tive behaviour of hot nuclei and on the possible occur- rence of phase transitions.

In any case the occurrence of different reaction mechanisms and the large number of nucleons involved in the collision create a very complex situation from the experimental point of view. From the first genera- tion of experiments realized in the intermediate energy domain the need for experimental devices capable to perform exclusive experiments, thus allowing the char- acterization of the various reaction mechanisms, has clearly appeared. In an ideal experiment one should detect, identify and characterize in energy and emis- sion angle all the reaction products in the whole solid angle. Presently this is an impossible task, due to the large variety of particles emitted in a collision event (')', pions, neutrons, light charged particles, heavy frag- ments) and to the wide energy range covered (from a few Me V to various hundreds of MeV).

To disentangle the complex experimental situation several multidetectors, devoted to the exclusive mea-

0168-9002/92/$05.00 © 1992 - Elsevier Science Publishers RV. All rights reserved

32 E. Migneco et at. / MEDEA: a multi element detector array

surements of some of the nuclear species produced in the collisions, have been realized or are presently un- der construction [7-13}.

At the Laboratorio Nazionale del Sud we have designed and constructed a multidetector system, in a 47T geometry, suitable for the detection of )'-rays (l MeVsE.,

s

300 MeV) and light charged particles, that allows the measurement of quantities related to the mechanisms of high energy )'-ray and neutral pion production, to the dynamics of nucleus-nucleus colli- sions and to the properties of hot nuclei. The detector, that has been named MEDEA (Multi Element DEtec- tor Array), has been designed to operate with heavy ion beams from the SMP tandem and superconducting cyclotron facility that will operate at the LNS.

2. General description

The MEDEA detection system consists of a ball, built up with 180 barium fluoride crystals, that covers the angular range between 30° and 1700in polar angle

(J, and a forward phoswich detector wall covering the angles between 10° and 30° in polar angle (J.Both the ball and the wall cover the full angular range of 360° in azimuthal angle c/J. A spherical polar coordinate system is employed hereafter, with the polar axis along the beam direction and the azimuthal angle ^c/J

=

0° in the vertical direction. To reduce the threshold on the charged particle detection the whole system operates under vacuum inside a large scattering chamber.

Barium fluoride (BaFz) is a scintillator crystal whose excellent properties have been extensively studied in the last years [14-25]. It has shown to offer a lot of advantages with respect to other materials, like NaI and BGO, which have suggested its application in our detection system and in various other systems devel- oped both for )'-ray and light charged particle detec- tion [16,26-28].

In the light emission of BaF2 two distinct compo- nents, with different wavelengths (A ""220 nm and

A "" 310 nm) and different decay times Cl' "" 600 ps and

T "" 600 ns), have been evidenced. The presence of a

very fast component allows for a very good time resolu- tion, comparable to that obtainable with fast plastic scintillators, which has been measured to be about 400 ps at the 662 keY 137Cs)'-line for large volume crystals [15]. This characteristic is particularly interesting in )'-ray experiments since it allows for a good separation of neutrons by means of time of flight measurements.

Moreover, the relative amount of the two light compo- nents emitted is a function of the incident radiation type. In fact the fast to slow component ratio is maxi- mum for )'-rays and decreases for light charged parti- cles with increasing Z. This property allows to discrim- inate between )'-rays and light charged particles using

a pulse shape analysis method as it is discussed in refs.

[19,20]. The studies carried out on this discrimination properties have also shown that the fast/slow ratio is also sensitive to the mass of the incident particle, thus allowing for a clear discrimination of p, d, t and et

particles [20).

The light yield of BaFz is rather low compared to NaI; nevertheless by optimizing the surface treatment and the reflector it is still possible to achieve quite a good energy resolution even with large volume crystals (- 9.5% at 662 keY for - 1 I crystals [17,21]).

Other advantages are offered by the excellent me- chanical stability ofBaF2 and by its nonhygroscopicity, which allows us to use the crystal in air without any canning, thus not introducing any threshold on the charged particle detection.

For the forward wall, which is devoted to the detec- tion of light charged particles and heavier fragments, a solution using a higher granularity array of phoswich detectors was preferred. A configuration using a thin fast scintillator NE-102A as I1E element optically cou- pled to a thick slow scintillator NE-llS working as E detector was realized. With this type of detector the shape of the anode signal gives results very similar to the BaF2 signal, although the meaning of the two components is different, allowing to use a similar elec- tronics for the two parts of the 47Tdetection system.

2.1. Geometry of the barium fluoride ball

The ball consists of an array of 180 closely packed barium fluoride scintilla tor crystals [29] covering the angular range between 30° and 170° in polar angle ((J)

and the full range in the azimuthal angle (c/J) from 0° to 360°. This geometry allows for the covering of 3.77T.

All the detector modules have the shape of trun- cated pyramids with trapezoidal cross section, with the four side faces converging towards the target position.

All the modules fit together to form a spherical shell with 22 cm of inner radius. The detector overall geom- etry is of course a compromise between different de- mands. From one point of view one has to maximize the crystal dimensions in order to reduce the incom- plete detection of the )'-ray energy due to shower leakage through the crystal faces. On the other side a large granularity to detect high multiplicity events and to achieve a reasonable angular resolution is necessary.

Moreover it is desirable to have modules with a cross section that optimizes the coupling to the photomulti- plier tube. In our case the geometry was defined in order to obtain equal solid angle modules. Almost all of the ball (between (J=30° and (J=150°) is divided in 24 azimuthal slices of constant 11

=

15°. The subdivi- sion in polar angle was made with cuts of constant

11cos (J

=

0.246 in the angular range 42° <(J < 138°, and with a11cos (J

=

0.128 in the angular ranges 30°<

E. Migneco et at. / MEDEA: a multi element detector array 33

/3049°

/25.00°

~20.00°

______1600°

___ 12.50°

___ 10°

beam 75.75°

/

/

60.50°

+

104.25°

11950°

\

\ 137.61°

~

1495l0~

100 cm

I I

Fig. 1. Schematic section of the array in the vertical plane containing the beam axis.

8<42° and 138°<8<150°. This last subdivision gives modules that cover exactly one half of the solid angle covered by the others. With this division four different types of modules are obtained, that were named type A, B, C and D (see fig. 1). The A, Band C type modules cover a solid angle of ~ 64 msr, while the D type covers ~ 32 msr. The angular range between 8

=

150° and 8

=

170° is filled with 12 modules with

t14J

=

30°, each one thus covering the same solid angle as the A, Band C type, that were named E type modules. The angles given here are only approximate:

the exact values of the polar angle position of each module are given in fig. 1, where a schematic section of

the array in the vertical plane containing the beam axis is shown. The characteristics of each module type are listed in table 1.

Several simulations were performed, using the EGS code, on realistic shapes of the BaF2 crystals, in order to define the optimal thickness for our purposes (de- tection of )'-rays up to 300 MeV). The crystal response was studied for several thicknesses (namely 12, 16, 20 and 22 cm) by taking into account an uniform illumina- tion of the crystal front face from a )'-source posi- tioned at a distance of 22 cm. In the calculations, the energies released into the crystal were folded with a Gaussian distribution in order to take into account

Table 1

Characteristics of the BaF2 crystal modules Detector type

M

efrometot:..</>

t:..ilCrystalCrystal [deg]

[deg][deg][deg]

[msr]thicknessvolume [cm]

[cm3]

D forward

11.90 30.4942.39 15 31.820 666.5 C forward

18.11 42.3960.50 63.320 1314.115 B forward

15.25 60.5075.75 63.620 1323.815 A forward

14.25 75.7590.00 15 63.820 1324.8 A backward

14.25 90.00 104.2515 63.820 1324.8 B backward

15.25 104.25119.5015 63.620 1323.8 C backward

18.11 119.5015137.61 63.320 1314.1 D backward

11.90 137.61149.5115 31.812 301.1 E backward

20.49 149.5130170.00 60.812 569.3

40

100 90

a)

80

~

70

~

60

"..

W 50

,'"

...•.

~

W 90b)

80 ~

70 6050

102101 103 100

for 20 and 22 cm thick crystals is greater than 90% in the whole investigated energy range.

Following these results the crystal thickness for MEDEA was fixed in 20 cm, that is about 10 radiation lengths (Xo

=

2.05 cm). Due to reasons linked to the crystal growing technologies available at the time of the commissioning, that did not allow to grow crystals

Ey (MeV)

Fig. 3. FWHM of the simulated response function vs incident energy. Symbols as in fig. 2. (a) Refers to the single crystal case, while (b) refers to a nine packed crystal cluster configu-

ration.

Ey (MeV)

Fig. 4. Deposited energy as percentage of the total incident energy. Symbols as in fig. 2. (a) Shows the <E> / E-yratio for the single crystal, (b) the <E> / E-yfor the cluster configura-

tion.

30a)

20~

'--' 10

L

I _:3

³⁰0^b)

I..L.

10 20

~

0

10-1

100101102103

E. Migneco et al. / MEDEA: a multi element detector array

~

100

>--

u

c

<I>

U _12-⁹⁰_cm ~

~ ^{16- cm <>}

w ^{20 -cm} •

c

22 -cm

8~0·1

10110°

102 103

Ey (MeV)

Fig. 2. Single crystal detection efficiency vs incident energy as simulated by the EGS code.

34

photon statistics and electronic noise. At high energy

(E-y:2: 20 MeV) this statistical contribution has been seen to be negligible and the FWHM of the response function appears to be essentially due to the incom- plete detection of the shower caused by the leakage out of the crystal back and side surfaces.

The results for the detection efficiency of the crys- tals are shown in fig. 2. This efficiency is quite constant on the whole investigated range. Up to 300 Me V the efficiency variation of the 20 cm thick crystal is con- tained inside a band of 5 points in percentage, against the 15 points of the 12 cm case.

The simulations were carried out also by consider- ing, besides the single crystal, a cluster configuration of nine packed crystals, the central one only being uni- formly illuminated on its front face by a '{-source. The coincidence between the central detector and the neighbouring ones allows partially recovering for the sideward leakage, thus improving the response func- tion. Fig. 3a shows, for different thicknesses, the FWHM of the response function as a function of the incident energy for the single crystal, while fig. 3b shows the FWHM of the energy spectra obtained by summing, on an event-by-event basis, the responses of all the crystals in the cluster. For a single crystal, the 20 cm thick case shows a FWHM that ranges up to about 20%, while in the 12 cm thick case the FWHM reaches a value of about 50% at 300 MeV. The situa- tion improves if the cluster is taken into account. In this case the 20 and 22 cm thick crystals have a FWHM of about 10% at 300 Me V, against the 20% and 38% of the 16 cm and 12 cm thick respectively.

Fig. 4 shows the behaviour of the average energy deposited in the crystal as a percentage of the total incident energy, for different detector thicknesses. As it can be seen in fig. 4a, while there is a steep increase of the quantity (E) IE-y from 12 to 20 cm, the same gradient is not present while increasing the thickness from 20 to 22 cm. This behaviour is enhanced in the cluster configuration (fig. 4a) where the ratio (E)

I

^E-y

E. Migneco et al. / MEDEA: a multi element detector array ³⁵

with such a large volume (~. 1.30, it was decided to obtain the desired thickness by coupling toghether two crystals, of the same tapered geometry, one 12 cm and one 8 cm thick.

The 20 cm thickness corresponds to the range of

~ 300 MeV protons and ~ 1 Ge V a-particles in the BaF2 crystal.

2. 2. The phoswich wall

The forward phoswich wall consists of 120 detectors covering the angular range between 100 and 300 in e and 3600 in <f;. Each phoswich detector is constituted by a 30 cm thick slow plastic scintillator (NE1l5) optically coupled to a 2 mm thick fast plastic scintilIa- tor (NE102A) [30] to form a !1E-E telescope. The light output of both scintillators is read by the same photo multiplier, which is coupled to the back surface of the slow scintillator. The different decay times of the light output of the two scintillators (~ 2.4 ns for the NE102A and ~ 230 ns for the NE1l5) allow one to separate the contribution of the two detectors by a pulse shape analysis [31]. With such a technique, by properly adjusting the PMT gain, one can get a good mass separation for Z

=

1 isotopes and charge separa- tion for heavier fragments. The total detector thickness was dimensioned to stop protons up to 200 Me V [32].

The threshold introduced by the !1E scintillator is about 13.5, 19, 23 and 54 MeV for p, d, t and a particles, respectively.

The detectors are placed at a distance of 55 cm from the target and occupy the full volume delimited by two conical surfaces of semi-aperture 100 and 300 displaced along the beam axis, and two spherical sur- faces with a radius of 55 and 85 cm centered on the target. The e range covered has been subdivided into five rings, of constant e, each one with 24 detectors.

The shape of each detector is similar to that of the barium fluoride crystals. The !1<f; acceptance of each module is still of 150, but in this case different consid- erations lead to the definition of the !1e acceptance of each module. In fact it was preferred to have different solid angles for the various e position, so that small solid angle modules were placed at the most forward angles, with the aim of reducing both the multiple firing and the count rate of each module due to the forward peaked angular distribution and high multi- plicity of the particles emitted in heavy ion collisions.

The characteristics of each module type are listed in table 2.

2.3. Detector mounting

For all the BaF2 detector modules, except for the E type and backward D type which are single 12 cm thick crystal, the 12 cm and the 8 cm thick crystals were

Table 2

Characteristics of the phoswich detector modules Detector type

(Jfrom°'06.</> !le

6.£2

[deg]

[deg][deg][deg]

[msr]

Wl

5.49 25.0030.4915 11.5 W2

5.00 20.0025.00158.6 W3

4.00 16.00 20.00155.6 W4

3.50 12.5016.00153.9 W5

2.50 10.0012.50152.2

coupled toghether using a thin layer of silicone oil (ViscasiI 600000 cstokes) [33], that has been measured to have a very good transmission even in the deep DV region [34], and wrapped with three layers of a 50 fLm thick PTFE foiL A thin adhesive aluminized Mylar tape 50 fLm thick was placed over the teflon to protect it. Another adhesive AI tape 200 fLm was used on the detector sides to enforce the mounting. The coupling of the photo multiplier tube was obtained by means of the same silicone oil mentioned above.

The energy resolution of each detector was mea- sured to be better than 12% at the 662 keY ')'-line of 137Csand the fast to slow amplitude ratio of the anode pulse is greater than 8: 1.

The two scintillators NEI02A and NE1l5 that com- pose each phoswich detector were coupled together, and coupled to the photomultiplier tube, by means of the same silicone oil mentioned above. The NE1l5 scintillator was supplied by the factory covered by titanium dioxide, a material that has a good reflection power in the wavelength domain of the light emission of these two scintilIators (A

=

420 nm for the NE102A and A

=

395 nm for the NE1l5). The front face of the fast NE102A is covered by an adhesive Al tape, 200 fLm thick, to stop low energy electrons.

2.4. Photomultipliers and voltage dividers

All the Photo multiplier Tubes are fast PMT with linear focused dinode structure. The tubes derive from the standard EMI 9127

(0

= 30 mm), EMI 9903

(0

⁼

40 mm), EMI 9954

(0 =

52 mm) and EMI 9821

(0 =

76 mm) [35]. In fact to optimize the light collection special diameter sizes (for example

0 =

62 mm) were chosen in order to better cover the rear surface of the detec- tors. All the tubes with standard high gain 12 stage structure were also modified to a 11 stage structure.

This was done in order to get the best performance of the photomultiplier tubes when operating in the opti- mal voltage range with intense scintillation photon flows incident on the photocatode. To extend the PMT spectral response down to the DV region, in order to match the scintillation light emission spectrum of the BaF2, all the phototubes have a quartz window. The

36 E. Migneco et al. / MEDEA: a multi element detector array

photocatode quantum efficiency is about 23% at 240 nm and 29% at 340 nrn.

Fig. 5 shows the scheme of a voltage divider chain designed for the EMI 9954 and EMI 9821 derived tubes. The voltage divider chains for each photomulti- plier type is designed in such a way to stabilize the cathode to first dinode voltage to 300 V by means of Zener diodes and the last three stages by means of active components. Also, the interdinode voltage in the last three stages is made to increase according to the factors 1.5, 2, 3 in order to reduce space charge effects and loss of linearity. The resistor values are chosen in order to have a divider current of the order of 1 mA while operating with a 2000 V voltage. With the EMI 9127 and EMI 9930 tubes, which are used with the smallest phoswich detectors, voltage dividers with a

slightly modified chain have been used. Each voltage divider is mounted on a printed circuit board con- nected, via a - 2 m long multiwire cable, with the PMT socket, where a compensating stage is mounted.

2.5 Mechanical support

Each A, Band C type detector is individually mounted on a steel arm that fits into the mechanical support and allows for the adjustment and eventual radial removement of the detectors one by one. Two of these arms (a B and a C type) are shown in fig. 6. The crystal holding is ensured by a textile of Kevlar [36]

fibers fixed to the arm. A spring system ensures the correct placement of the photomultiplier tube. Each arm has a system of wedges and control screws that

on PMT socket 08 07 010

05 09

03

01 06

04

02 anode

cathode tkO

4.7MO

47MQ

4.7MO lOnF

2.2nF

r

lkQ

Fig. 5. Scheme of the active voltage divider chain. This divider has been designed for the BaF2 detector photomultipliers: the phoswich detector photomultipliers are equipped with voltage dividers with a slightly modified chain.

E./vliRIlCCO etaf. /lviE[)[,A. aII/ulti clclI/ent dctector (lrrav

Fig. 6. Two detector mounting arms. In the upper part of the photograph a C type arm is shown and in the lower part a B type arm is shown. A vacuum flange with the llV fcedthrough (see text) is also visihle in the upper left.

Fig. 7. An A type detector mounted on the support system. Note the Kevlar string used to hold the crystal to the support arm.

Fig. 8. Particular of the arm positioning system. Two A type, a B and a C type are shown. The control screws that allow for a fine adjustment of the detector position are not shown.

E. Migneco et at. / MEDEA: a multi element detector array 38

1 m

Fig. 9. The scattering chamber. Windows and flanges are not shown. The ball in the upper left part of the figure is the shield for the"y calibration source (see text). Beam direction is from left to right.

E. Migneco et at. / MEDEA: a multi element detector array 39

Fig. 10. The backward half of the detector during the assembling. The rectangular windows for the introduction of the arms can be seen, together with the internal mechanical support and one half of the BaF2 ball assembled.

allows the correct positioning of the crystal. The arms are inserted in a support structure that is positioned inside a scattering chamber (figs. 7 and 8).

To allow for the placement of the complex mechani- cal support of the crystals, a large vacuum scattering chamber with cilindrical shape has been realized (fig.

9). This scattering chamber is composed of two sepa- rate symmetrical cilindrical sections (l m long), each one housing one half of the array, thus allowing for the opening of the detector array along the 8

=

90° plane.

Each one of the cilinders has been machined to obtain 48 rectangular windows, large enough to allow for the radial insertion of the detector arms (one window for each C detector plus one window for each couple of A and B detectors), as it can be seen in fig. 10 which shows one half of the detector in an initial assembling stage. Each window is closed by means of a rectangular flange that carries the high voltage feedthroughs. In fact, as it was mentioned above, a solution with the PMT divider chain placed outside the chamber has been preferred to avoid heat dissipation problems and to allow interventions on the voltage divider without opening the scattering chamber. The connection be- tween the active voltage divider and the PMT socket, which carries only a passive nondissipating compensa-

tion stage, is realized by means of a multiwire cable.

The vacuum feed through is a 15 pin connector through which the anode signal and relative ground are also carried.

Each of the two sections of the chamber is placed on a carriage with a ball bearing system that allows rotations around the beam axis. The carriage systems themselves are placed on 6 m long precision rails to allow the movement along the beam axis.

An internal view of the scattering chamber with the completely assembled BaFz ball is shown in fig. 11.

The phoswich wall is supported by a system of two coaxial cones of semi-aperture 10° and 30° with the axis coincident with the beam axis, in between which all the phoswich detectors are wedged. The detectors are kept in place by 24 rods with control screws that press on the detector back side. The whole block (support cones +detector wall) is fixed to a ring (15 cm thick X2 m

0)

which fits to the main body of the scattering cham- ber (fig. 12). This ring also supports the high voltage feedthroughs for all the wall detectors. The forward D type crystals (30°<8<42°) are individually mounted on short steel arms, which in their turn are fixed to the external part of the phoswich wall conical support.

The scattering chamber is closed on both sides by

40 E. Mi[;neco et af. / MEDEA: a mlllti clement detector array

Fig. 11. Internal view of the seattering chamber with the completely assembled SaFe ball.

Fig. 12. The forward wall support system. The 120 phoswich detectors are placed inside the conical support. Fixed to the external part of the support are the 24 D type SaFe detectors.

E. Migneco etat. / MEDEA: amulti element detectorarray 41

STEPPING MOTOR

Fig. 13. Scheme of the target holding system (see text for details).

two endcaps. The one on the beam entrance side supports also a rotating system to hold the backward D (138° < () <150°) and E type crystals.

A large circular window

(0

=60 cm) has been left

0,

ID:

101 101

t t

TARGET HOLDER

ABSOLUTE ENCODER

on the beam exit side to allow for the mounting of eventual forward dctectors.

Great care was taken in the design of the target holding system since it has to be as thin as possible to reduce the shadowing on the detectors. The motion system has to be inserted between two adiacent crystal faces without removing or displacing any module of the array. Moreover the system has to work under vacuum and all the positioning commands and information on the actual target position must be get via a remote control. Based on this specification a target holder, with dimensions of 100 mm x160 mm and 1.1 mm thick, has been designed (fig. 13). The system can carry up to seven target frames with a rectangular window of 15 mm x20 mm. The target holder is sustained and moved by means of two Kevlar wires kept in tension by means of a pulley system, two of which transmit the motion. The wires are allowed to slide between the faces of two adiacent A type crystals in the ()

=

90°

plane. The target position is monitored by a suitably realized 3 bit absolute encoder, and controlled by two security limit switches, while the motion is obtained by means of a small stepping motor.

A 9 mm internal diameter tube, that is fixed on the endcap at the beam entrance side and that can arrive with a telescopic movement close to the center of the ball, allows to insert a radioactive source for crystal

Fig. 14. External view of the completely assembled detector.

42 E. Migneco et at. / MEDEA: a multi element detector array

calibration keeping the source in air. When not needed, the source can be retired from the scattering chamber inside a lead and paraffin shield (fig. 9).

The vacuum system is constituted by a couple of rotor pumps and two molecular pumps (2200 I/s). Two cryogenic panels have also been installed inside the chamber, but their use has never revealed necessary up to now. Although the large volume of the scattering chamber and the large amount of material contained in it, the vacuum system has proved to work efficiently, attaining a pressure of 10-5 mbar in about 6 h.

An external view of the completely assembled scat- tering chamber is shown in fig. 14.

3. Electronics and data acquisition system

The anode output of each PMT is analysed by a weighted passive splitter, located close to the scattering chamber, which generates three output signals. For each detector one of the splitter outputs is sent, by means of a 3 m RG174 cable, to the input of a set of 38 8-channel GANELEC FCC8 CAMAC constant frac- tion discriminators (CFD) [37). By means of a 3 m RG 174 cable another output is sent, for the BaF2 detectors only, to the input of a set of 12 16-channels LRS 4413 CAMAC leading edge discriminators (LED) [38]. Both the discriminator sets are placed in the experimental area together with the photomultiplier power supply system. The two sets of discriminators constitute the first acceptance level filter for the events, this acceptance check being based on the multiplicity of detectors passing a preset threshold level. The con- trol of the discriminator threshold levels, multiplicity thresholds and output widths is performed remotely by means of a specific software implemented on a Macin- tosh-Plus personal computer located in the control room. The same software controls, by means of a special interface, the high voltage system, composed by 3 CAEN PS351100 power supplies and 6 CAEN SY170 64 channel voltage dividers [39].

The logic output generated by each CFD channel is used for the time correlation (T parameter) measure- ments with respect to the pulsed beam. The logic signals are transported to the control room through 100 m long twist and flat cables terminating into the stop input of 20 16-channels LRS 4303 time to FERA converters (TFC) +20 16-channels LRS 4300 FERA QDC system, this last one operating on 11 bits conver- sIon range.

For each detector the third splitter output is sent to the control room, by means of a 90 m long low attenua- tion RG213

ID

cable, and is used for the energy inte- gration. Such an analog signal is sent, for each detec- tor, to a weighted passive splitter which generates three outputs;

- F, which is used for the fast component conversion (60% of the input signa!);

- E,which is used for the low range energy conversion (35% of the input signa!);

- Ea' which is used for the high range energy conver- sion (5% of the input signal).

The relative weight of the three splitter outputs have been chosen in order to optimize the slope of the fast (F) versus attenuated energy (E) plots with re- spect to the QDC performance and the relative full scale range of E and Ea'

The splitter analog outputs are connected to the input of a system of 60 16-channels LRS 4300 FERA QDCs operating on a 11 bits conversion range.

The electronics associated to the multidetector sys- tem consists of two levels of fast logic triggers. The first level trigger operates fast decisions by the analysis of the logic signals produced by the two sets of discrimi- nators. The occurrence in an event of one out of a group of preselected conditions enables the analog to digital conversion of the parameters under study. The second level trigger operates on the converted data. It is described in detail in section 3.4.

3.1. First level trigger

First level decisions are based on the logic outputs generated by the sets of discriminators mentioned above. In experiments where the event trigger is given by the BaF2 ball, the use of the 12 LED discriminator set in combination with the CFD system allows a wider range of applications. In fact the use of such a second set of discriminators offers the advantage of event triggering at an electronic threshold level that can be different from the level selected on the CFDs. Thus, for instance, in the search and investigation of high energy photons (E'I2: 30 MeV) the threshold of the LED system is adjusted to correspond to the lowest energy desired (30 MeV in the example considered) and the threshold of the CFD system are kept low so as to accept also coincidence events with low energy photons and

lor

light charged particles. With reference to fig. 15 the LED system, daisy chained according to the trigger configuration selected, provides analog out- put pulses (~) whose amplitude is proportional to the number of channels activated in an event. Such a signal is further analysed by another CAMAC discriminator, whose threshold is also remotely adjustable from the control room, according to the fold coincidence level required. The logic output generated by such discrimi- nator, a few tenth of ns wide, is sent to the control room, where it is ANDed with a fast NIM logic signal generated by the HF. The occurrence of such a coinci- dence generates:

a) the common start to the TFCs;

E. Migneco et at. / MEDEA: a multi element detector array 43

Experimental area Data Acquisition area

12 modules

~

L R5 4 4 1 3

L R

5 4 4 31

L R5 4 4 1 3

L R5 4 4 1

,;.

F

CCl---+

8

chain multiplicity M

beam

M o

E L A Y

C F

o

gate E chain gate Ea chain

acquisition veto

startTFC

gate Tchain gate Echain

F C C 8 BaF,

signal 5 P L

+A

T E R

23 modules

~

ig ^!

module multiplicity 5 C A L E R

5

P L

+ B

T E R

to FERA (F chain) to FERA (E chain) to FERA (Ea chain)

Fig. 15. Simplified block scheme of the electronics and first level trigger for the BaF2 ball. (a) Scheme of the electronics installed inside the experimental area. (b) Scheme of the electronics installed in the data acquisition area. The phoswich detector electronics

is analogous, without the LED LRS4413 chain.

b) the common gate to the QDCs for the time-to- charge integration;

c) the common long gate (several hundreds of nanoseconds) for the integration of the E and Ea parameters;

d) the common short gate (about 20 ns of effective width) for the integration of the fast component;

e) the fast veto to the AND unit to be overlapped with the inhibit signal generated by the following elec- tronics operating data conversion, readout, second level decision taking, transfer to the data acquisition system and data storage.

3.2. Data conversion and readout

For the data conversion and readout the FERA system was preferred to others mainly for reasons of compactness and speed. In fact such QDCs can per- form the parallel conversion of all channels and the suppression of all zero contents in - 11fLS. Moreover, FERA QDCs can be connected together in a chain with two front-panel buses, one for data readout (that

can be performed at 100 ns/word) and one for the control signals. A common gate signal is distributed along the control bus, together with other control sig- nals, by a LRS 4301 FERA driver module. For this reason a chain is likely to be used for homogeneous parameter conversion, thus in our case the 1200 QDC channels needed (4 parameters/ detector X300 detec- tors) are displaced in four chains, each chain being physically located on a 1000 W CAMAC crate and associated to one of the four parameters T, F, E, Ea'

3.3. Dynamic range

The lower limit to the dynamic range for photons covered by the BaF2 detectors is determined by the lower level electronic threshold

(?

10 mV) selected on the CFD system.

Since the CFD accepts signals up to 5 V in input, the dynamic range for photons is, in principle, 500: 1.

The choice of the relative weights of the splitter out- puts in fig. 15 allows for the FERA chains (E and Ea) to cover almost the whole dynamic range.

E. Migneco et al. / MEDEA: a multi element detector array ⁴⁵

F

T

detector id# (DIN)

:l

E c_.,

>.,

Fig. 17. Block scheme of the second level trigger data processing (see text for details).

which one has to be taken into account in further steps.

- Particle and y identification. This step is based on the discrimination properties of the detector material used. In fact for the BaF2 crystals in a fast component versus total energy representation the "{events show a different slope with respect to the particles. The identi- fication is performed by determining in which region of the Ea-F or E-F plane the two parameters of each detector fall. The correspondence table that links each couple of E-F values to an identifier is mapped for each detector into a memory lookup unit (MLU). The identifier is stored for each detector in the highest bits of the DIN word for subsequent selective multiplicity computations. It is worth to mention that this kind of identification, based on a two parameter correlation, is not strictly bound to the particular type of detector used, but it could be performed for any type of detec- tor with discrimination properties, like for example the phoswich. In the version of the preanalysis system that is presently in operation the data related to the phoswich detectors are only synchronized but are not treated by the following steps described here.

- Neutron identification. This step is performed since the interaction of low energy neutron in BaF2 gives a signal identical to "{-rays.A low energy neutron is thus treated as a "{-rayin the preceding step, but can be identified by its time of flight. This identification is performed comparing the time information with a ref- erence time of flight (one for each detector), that is preloaded into a MLU.

- Selective multiplicity computation. At this point the five words related to the detector are temporarily stored in a data stack unit, waiting for the completion of the event treatment. At the same time the informa-

tion obtained in the preceding identification steps, that are stored in the higher bits of the DIN word, are used to increment selective counters (one for each kind of particle).

- Event validation. At the end of each event, that is recognized by a special end-of-event mark, a decision is taken, comparing the computed multiplicities with user selected windows, whether the event is to be rejected or transferred to the data acquisition system.

If the event is accepted it is transferred directly from the data stack unit to the data acquisition system.

Event transfer towards the data acquisition computer system is performed by means of a differential ECL transmission bus realized with a 34-wires twist-and-flat cable that allows a 16 bit word wide transmission. The transmitted signal width, and consequently the trans- mission speed, is strongly dependent on the cable length. In the current system setup, with a - 10 m long cable, the maximum transmission speed is about 5 Mwords/s.

The system is entirely realized using modules of the LeCroy ECLine. In particular, as it appears from the description given, most of the work is performed using memory lookup units, arithmetic logic units and data stacks.

The timing performances of the system are strongly dependent on the particular configuration as well as on the rate of accepted events. Nevertheless simple mea- surements were performed under particular conditions, and they have shown that the handling time for one detector, assuming a consistent input quadruple of parameters, is about 2.5f.LS. The FERA ADC conver- sion time is 11f.LS(including the zero suppression), and the time needed to transfer the data of one detector to the acquisition system is 2 f.LS(400 ns per word). Thus

46 E. Migneco et at. / MEDEA: a multi element detector array

we can estimate that the total time needed to handle one event, if no rejection takes place, with a multiplic- ity of 20 detectors fired, is - 100 /-Ls.

3.5. The data acquisition system

The data acquisition sytem is built around a modu- lar multiprocessor architecture, VME-bus based, using a one broadcasting/many receivers data passing scheme, in order to allow for different functions to be performed asynchronously and concurrently on sepa- rate receiver stations. A well tested software architec- ture is used, in order to keep the user interaction with the whole system [42,44] as simple as possible.

The guidelines that have been followed in develop- ing the system are:

- subdivision into independent subsystems that can be developed and tested separately;

- possibility of simple rearrangement of the subsys- tems configuration according to the specific experi- mental need;

- ease of synchronization by means of a data distribu- tion protocol implemented in hardware;

- ease of hardware and software upgrading;

- possibility to suit a wide variety of experimental conditions.

VME-bus was preferred to other solutions because of its wide diffusion in the nuclear physics electronic field. Coupling between the ECL broadcasting bus and the VME is realized by means of the special memory board TVM-FERA [45] that was conceived to perform the function of data receiver.

The general hardware structure (fig. 18) is with one data source and some independent consumers, each one realized with a number of VME boards. System consoles are implemented on one or more Apple Mac- intosh-Plus personal computers interfaced to VME-bus by means of MacVee/MacPlinth boards [46].

Each data consumer is constituted by an ECL input section, a processing section and an output section. No VME interrupt level is used, thus allowing more than one copy of the same receiver to work concurrently, when needed, without special software or hardware changes.

Software architecture of the various receivers is completely symmetrical, but a slight degree of asymme- try is introduced at run time for dead time handling reasons. In fact two kinds of receiver are allowed:

active listeners and pure listeners. A pure listener is a subsystem that samples data on the ECL broadcasting bus without affecting the data source, while an active listener is capable to stop the data source when busy, forcing it to broadcast at its own rate. The slowest active listener of the whole system will determine the data source effective rate.

Fig. 18. Block scheme of the data acquisition system. The EeLdifferential bus is the same as the one at the exit of the

second level trigger scheme in fig. 17.

The system is currently composed of three re- ceivers:

- Tape storage subsystem;

- Scatter plotting subsystem;

- On-line analysis and monitoring subsystem.

Tape storage is an active listener that uses two memory buffers (TVM-FERA) to receive and store data. It implements the most common magnetic tape functions (as write, rewind, skip, etc.) to allow for an easy data storage management. Its measured transfer rate was of 300 kwords/s in a true experimental envi- ronment. With an average multiplicity of 20 detectors fired per event this means an acquisition rate of about 3000 events/so

Scatter plotting is a pure listener that allows for immediate monitoring of any detector element, due to its capability to build live histograms and colour scatter plots onto a graphic screen. This subsystem has proved to be quite useful especially when tuning the detector system, because of its ease of defining and modifying in a little while the display set up [41]. Up to 32 his- tograms and 16 bidimensional scatter plots can be built at the same time on four virtual graphic pages, one of which can be shown in turn on the monitor screen, with the possibility to move a cursor on each histogram or scatter plot on a channel by channel basis, just like a multi channel analyzer. The measured event rate of this subsystem was of a few hundred events per second under average load conditions.

E. Migneco et at. / MEDEA: a multi element detector array ⁴⁷

On-line subsystem is a pure listener that allows to define, build and handle large memory histograms.

Data conditioning is performed by means of windows and contours, that can be interactively defined and modified by means of display functions. Special his- tograms, called multihistograms, can also be defined.

This type of histogram allows one to handle, for each event treated, not just one but a set of detectors. Thus, for example, a multimatrix can be used to build a histogram having the energy on the x axis and the detector number on the y axis. Such a matrix is simply the collection of the energy spectra of all the detectors.

Display functions are also provided to scan matrices in x-slice or y-slice modes, so giving the possibility to display the matrix described above as a whole energy matrix (for global monitoring purposes) as well as a collection of single energy spectra. A colour hardcopy is also provided to well exploit on-line graphical fea- tures. Histogram memory can be of any size, but the typical configuration has 8 Mbytes available to build

histograms with 8, 16 and 32 bit channel size and with whatever interest region limiting channels [47].

The whole acquisition system has currently two con- trol consoles: the first one is intended for the tape storage and scatter plotting control, while the second is dedicated to on-line analysis. However, free changes can be performed quite simply in this configuration, as like as in the active/pure listener status of each subsys- tem, if required.

4. Tests and performances

The individual BaFz detectors have been tested using monoenergetic photons and light charged parti- cles. The response function for high energy )'-rays was measured up to E.y

=

280 Me V using monochromatic )'-rays from the in flight annihilation of a positron beam delivered by the ALS Saclay accelerator. The response functions, especially in the high energy range,

1'-+-.••••--- •.•••--- ..•.•..-'---\-'$

~ ~

LE

<I

ESt

Fig. 19.Ea - Eacosmicrays correlation matrixwith the related projected spectra for two oppositeA type detectors.Abump due to events with cosmicmuons that have traversed the full detector thickness in both the crystals is clearlyvisible.The corresponding

"'I-energyhas been determined to be -130MeY.

48 E. Migneco et at. / MEDEA: a multi element detector array

are well reproduced by the Monte Carlo code EGS and the light output ofBaFz has been checked to be linear in the whole investigated energy range. The response to light charged particles was also measured at the GANIL laboratory using monoenergetic particle beams produced by exploiting a standard technique [48]. The details of the experimental procedure and data analy- sis, together with the measured performances, will be the subject of forthcoming papers.

In this section general performances of MEDEA deduced from preliminary tests with heavy ion beams at intermediate energy are presented. The tests have been performed with the heavy ion beams of the GANIL laboratory, where MEDEA is presently in- stalled for a first series of experiments.

4.1. Description of the measurements

The gains of the PMTs and the lower level thresh- old of the BaFz were adjusted by using a 6.13 MeY )'-ray PuC source. By means of the remote control of both the voltage of each PMT and the electronic threshold level of each CFD channel, and by making use of on-line multihistograms described in section 3.5, the gain and the threshold setting of the BaFz b'all takes typically a few hours. The lower threshold was selected so as to correspond to a level of about 2 MeY )'-rays. The gains as well as the electronic thresholds of the phoswich wall were roughly equalized in beam.

The energy calibration of the BaFz detectors was performed by using as reference points the low energy

s

C

PIL

E

7o

Ef'lERGV DETECTOR9-200 PARAMETER'4

Fig. 20. Energy spectra multimatrix. Detector number vs energy (Ea>. Each slice of this matrix for a fixed y-value (detector number) is the energy spectrum corresponding to that detector. Lower detector numbers correspond to forward detectors.

E.Migneco et at. / MEDEA: a multi element detector array 49

PuC "{source and the response to the cosmic rays. The MEDEA geometry allows one to select events due to cosmic rays traversing the full BaF2 detector thickness.

These events were selected off-line by looking at the coincidences between couples of opposite detectors.

The energy deposited in each crystal has been deter- mined to be ~ 130 MeY. An example is shown in fig.

19,where an Ea-Ea correlation matrix with the related projected spectra for two opposite type A detectors is shown. These· spectra were accumulated in about 48 h of acquisition.

The first level trigger configuration used in the tests was based on the logic multiplicity (M) output gener- ated by the CFD chain. Coincidence events between the multiplicity M:?:

1

signal and a fast logic signal

COLLECTION OF TIME SPECTRA 33-188

associated with the pulsed beam were accepted to occur within a time window of 30 ns.

Events of high "{-ray and light charged particle multiplicity were produced by bombarding a 5.7 mgjcm2 Au target with a 129Xebeam at 45 MeY jamu.

An on line check of the detection system as a whole was performed by looking at the multi histograms re- lated to the parameters of interest. As an example, fig.

20shows the energy spectra collected in beam with the BaF2 ball. The attenuated energy (Ea) on the x-axis is plotted versus the detector identification number (DIN); lower DINs correspond to forward placed de- tectors. Such energy spectra are dominated by the charged particles contribution. By updating and dis- playing on a screen one or several multihistograms, like

TIl'1E

,'-*-

,,"'!c~

DETECTOR 9-200 PARAMETER 1

s

C A

LE :56

Fig. 21. Time spectra multimatrix. Detector number vs time. Same as in fig. 20, but in this case the matrix is the collection of all the time spectra of theBaF2 detectors.

E. Migneco et al.

j

MEDEA: a multi element detector array 50

the one shown in fig. 20, one gets, on line, a complete view of the detection system allowing a fast check, for instance, on the response of all detectors, when in beam, or a fast monitoring of the PMT gain stability when using the external source. Another example of multihistogram is shown in fig. 21 where the time spectra of the BaF2 detectors are reported. Such a multihistogram can be scanned detector by detector to examine each one of the detector time spectra.

4.2. y-rays and light charged particles identification with theBaF2 ball

The method of analysis of the pulse shape produced byBaF2 detectors has been described elsewhere [20]. It is based on the separate integration of the total output signal and the fast component. In fig. 22 the raw data associated with the fast light output and with the total light output are shown in the typical fast (F) versus

attenuated energy (Ea) representation. Both the dis- crimination among the "yand protons and among Z

=

1 isotopes occur at an energy as Iow as a few MeV of energy deposited in the crystal. A discrimination ma- trix in the Iow energy range (E parameter), obtained with the short gate optimized for "y identification at any photon energy, is shown in fig. 23. The examples reported in figs. 22 and 23 clearly show that the short gate, and consequently the power of the discrimination matrix, can be adjusted according to the aims of the particular experiment. In the MEDEA electronics the short gate is common to all the BaF2 detectors and synchronous with the pulsed beam. Since detectors with one (or more) out of the three parameters T, F, E missing are discarded from the events by the data flow synchronization process performed by the second level electronics, the setting of the short gate is a compro- mise between the obvious requirement of a selective integration of the fast component only, from one side,

sC .RL E ,8

'.

....

Fig. 22. Discrimination matrix obtained with a BaF2 crystal placed at IJ'"82° (forward A type detector), in the reaction 129Xe+ 197Auat Ebeam=45 MeV jamu. The fast light output to the total light output are shown in the typical fast (F) versus attenuated energy (Ea) representation. The locus with the highest slope corresponds to 'Y~raysand cosmic muons. Then, with decreasing slope respectively, there are protons, deuterons, tritons and Z=2particles. It is easy to see that it is possibile to

separate, besides the Z ~ 1 and Z=2 regions, the three masses in the Z=1 region.

E. Migneco et al. / MEDEA: a multi element detector array 51 and the need of saving slower particles in the range of

mass and charge that can be identified.

4.3. y-neutron discrimination

Large volume BaFz crystals exhibit a non-negligible efficiency for neutrons. A good -y-neutron discrimina- tion can be obtained by combining the time of flight information and the pulse shape analysis. Neutrons interact with BaFz by (n, -y), (n, n' -y) and (n, charged particle) reactions, the first two processes being domi- nant for low energy neutrons [22]. The contribution of other reaction channels becomes increasingly impor- tant for increasing neutron energy above 20 MeY [49].

Thus we can separate high energy neutrons from -y-rays using the F vsE or Ea plot but it is not possible to do that for low energy neutrons. A means to discriminate -y-rays from low energy neutrons is provided by time of flight measurements.

In fig. 24 the raw time spectra collected at various detection polar angles are shown. At the backward angles the peak due to the -y-rays is well separated from the particle time spectrum. At forward angles the time spectra are dominated by the intense particle (mainly protons) yield. For these angles, by sorting the time spectra conditioned by a suitable contour around

I

••

the -y-Iocus in the F-E or F-Ea discrimination matrix, the -y-peak appears well resolved and 800 ps wide (FWHM), the contribution of the detector being about 600 ps. In such conditions the MEDEA geometry would allow, in principle, the discrimination of -y-rays from neutrons with energy up to ~ 100 MeY.

4.4. Light particle identification in the phoswich wall

The anode output of each phoswich detector con- sists of a fast component and a slow component. The fast component, extending over at most 20 ns, is pro- duced in the thin fast plastic scintillator acting as /lE detector. The slow component is produced in the thick slow plastic scintillator acting as detector of the resid- ual energy Eres

=

E - /lE. Anode signals associated to particles which are fully stopped in the thin plastic scintillator exhibit the fast component alone (Eres

=

0).

On the contrary anode signals associated to photons or neutrons interacting in the thick plastic scintillator exhibit the slow component alone (/lE

=

0). By analyz- ing the phoswich signals likewise to the BaFz signals one can construct the corresponding (/lE, E) discrimi- nation matrix [31,49]. In fig. 25 an example of such a matrix, obtained by adjusting the PMT gain so as to resolve the mass of Z

=

1 particles, is shown. With

s

C AL E 7

20<47

Fig. 23. Same as in fig. 22 but for the low energy range. The abscissa, in this case, is the non-attenuated energy (E). The line separates between -y-ray(above the line) and particle (below the line) contributions.

52 E. Migneco et al. / MEDEA: a multi element detector array

(f)

I-

z:

::0

o

U

TIME-D

(a)

I-

(f)

z:::0

U

o

TIME-A

(e)

650

tQ

I-

z:::0

o

U

(f)

I-

z:

::0

tl

(b)

(c)

(f)

I-

z:

::0

o

U

(f)

l-

S o u

(f)

(g) 650

700

I-

(f)

z:

8

::0

TIME-A

(d)

(f)

I-

::0z:

o U

(h)

550

Fig. 24.BaF1raw time spectra for various polar angles from e = 30° to e = 170°. (a) forward D type detector (30°.5<e<42°.4); (b) forward C type detector (42°.4<e<60°.5); (c) forward B type detector (60°.5<e<75°.7); (d) forward A type detector (75°.7<e<

90°); (e) backward A type detector (90°<e<104°.3); (f) backward B type detector (104°.3<e<119°.5); (g) backward C type detector (119°.5<e<137°.6); (h) E type detector (149°.5<e<170°). As can be seen, at forward angles the time spectra are dominated by the light particle (mainly proton) contribution, while at backward angles a well separated "'{-raypeak appears. It should be noted that the forward angle detector time spectra can be easily cleaned by sorting the data with a suitable contour

condition in the fast-energy discrimination matrix, to reduce the light charged particle contribution.

such a gain, light fragments up to Z

=

5 are accepted within the t:.t QDC range and identified in charge. In this example the E (Ea) parameter is integrated by the same gate used for the energy integration of the BaF1 detectors. Concerning the fast integral F,it is obtained by using a suitable short gate different from the one used for the BaF1 fast component. These two short gates are sent to the corresponding sets of FERA QDCs located on the same crate and distributed to each module by the same control bus. The gate associ-

ated to the phoswich detectors is distributed trough the FERA driver located on that crate to the 8 QDCs associated to the phoswich wall, while the gate associ- ated to the BaF1 is connected directly to the control bus. It should be noted that, due to the width of the short gate, the fast integral F is the sum of the energy loss t:.E plus a fraction of the residual energy Eres' Fig.

25 clearly shows the Eres

=

0 locus. The locus t:.E

=

0 is much less intense. Both are straight lines whose slopes are determined by the gate widths used for the

E. Migneco et al.j MEDEA: a multi element detector array 53 PHOSWICH 797 - FAST SLOW PLOT

s

C A

l

E

21

2039

Z=2

t

p

DETECTOR 797 PARAMETER 3 SLOW

",

" • &

o

r-(]I r- cc

o I-

U

W

I-

W

£:)

N

ccW

I-

W

I:

a:

cca:

CL.

I-

(fJ

a:LL

Fig. 25. Discrimination matrix for a phoswich detector placed at IJ "" 22°.5 (W2 type detector), in the reaction 129Xe+197Auat

Ebeam =45 MeV jamu. Z=1 and Z=2 contributions are well separated as well as the three masses forZ=1.

F and E integration as well as the weights of the second splitter. The knowledge of these two loci allows one to report the events in a !:lE vs Ere, representa- tion, which is more suitable for particle energy calibra- tion [51].

We are therefore convinced that this apparatus will be a very useful tool for a better understanding of various phenomena in heavy ion collisions.

Acknowledgements 5. Conclusion

A 4'Tf multi detector array has been designed and realized at the LNS in Catania. This apparatus aims at exclusive measurements of heavy ion collisions in the intermediate energy domain.

The tests exposed in this paper and the preliminary results of the first experiments performed at the GANIL laboratory confirm that the design characteris- tics have been achieved.

We would like to thank Prof. G. Russo for the fruitful discussions and the help given during the preparation and installation of the detector.

The help of the entire LNS staff in the installation and testing of the detector is greatly appreciated. In particular we would like to thank: G. Di Biasi, B.

Trovato and E. Zappala. of the LNS workshop, who have greatly contributed to the installation of the me- chanics; S. Marino, S. Salamone, of the LNS electronic staff, and F. Librizzi, of the INFN-Sezione di Catania,