Performance of a shashlik calorimeter at LEPII

N

ELSEVIER

Btl[ll I I-'I ',IIIllh'l,,~[&"l

Nuclear

PROCEEDINGS SUPPLEMENTS

PhysicsB (Proc. Suppl.)78 (1999)220-225

www.elsevier.nl/Iocate/npe

Performance of a Shashlik Calorimeter at LEPII Presented by P.Ferrari S.J. Alvsvaag, A. Klovning, O.A. Maeland, B. Stugu ~ , A.C. Benvenuti, V. Giordano, M. Guerzoni, F.L. Navarria, M.G. Verardi b , T. Camporesi c , M. Bozzo, R. Cereseto d , G. Barreira, M.C. Espirito Santo, A. Maio, A. Onofre, L. Peralta, M. Pimenta, B. Tome e , H. Carling, E. Falk, V. Hedberg, G. Jarlskog, I. Kronkvist / , M. Bonesini, F. Chignoli, P. Ferrari, S. Gumenyuk, R. Leoni, R. Mazza, P. Negri, M. Paganoni, L. Petrovykh, F. Terranova g , D.R. Dharmasiri, B. Nossum, A.L. Read, T.B. Skaali h , L. Castellani, M. Pegoraro i , A. Fenyuk, Yu. Gouz, A. Karyukhin, A. Konoplyannikov, V. Obraztsov, N. Shalanda, E. Vlasov, A. ZaitsevJ , M. Bigi, V. Cassio, D. Gamba, E. Migliore, A. Romero, L. Simonetti, E. Torassa, P.P. Trapani k , M. Bari, G. Della Ricca, L. Lanceri, P. Poropat, M. Prest, E. Vallazza l

'~Bergen, bBologna, eCERN, dGenova, eLisbon,/Lund, gMilano, hOslo, iPadova, JSerpukov, kTorino, /Trieste The Small Angle Tile Calorimeter (STIC) is a sampling lead-scintillator calorimeter, built with "shashlik" technique. Results are presented from extensive studies of the detector performance at LEP.

1. I n t r o d u c t i o n STIC provides calorimetric coverage in the very forward region of the D E L P H I experiment at the C E R N L E P collider. The structure of the calorimeters, built with so-called "shashlik" technique, allows the insertion of tracking detectors within the sampling structure. Two silicon planes were installed inside STIC in order to make it possible to determine the direction of the showering particle. The detector is equipped with a chargedparticle veto system t h a t provides e-3' separation. A detailed study of the performance of the STIC detector has been made using a large sample of B h a b h a events collected at different L E P II center of mass energies.

beampipe t h a t are supported by the L E P lowbeta quadrupole. Each STIC module is composed

--~

Large Veto Scintillators

Silicon planes

Inner ma

I ill!

Small Veto Scintillators

2.

The Small Angle Tile calorimeter

The Small Angle Tile Calorimeter (STIC) [1,2] is a sampling lead-scintillator calorimeter and consists of two independent cylinders (called A and C) located at 220 cm from the interaction point and covering the angular region between 29 and 185 m r a d (Figure 1). Each cylinder consists of two independent half-cylinders around the

Figure 1. The STIC calorimeter layout.

of 47 layers, each made of a 3 m m thick lead plate, steel laminated on both sides for a total thickness of 3.4 mm, alternated with 3 m m thick scintillator

0920-5632/99/$ - see front matter © 1999 Elsevier Science B.V. All rights reserved. PII S0920-5632(99)00548-4

S.J. Alusuaag et aL /Nuclear Physics B (Proc. Suppl.) 78 (1999) 220-22~

tiles. The absorber plates are continuous to avoid non uniformities in the energy response and to provide full hermeticity. T h e scintillator tiles are made of injection molded polystyrene doped with 1,5% paraterphenyl and 0.05 % P O P O P . Tiles are optically insulated from each other by 120 # m thick white T y v e k paper and mounted on the converter plates by means of precision pins, with a mechanical accuracy of 50 #m. The tiles give a tower structure t h a t is projective towards the interaction point. Each calorimeter module is subdivided in 10 rings in radius and 16 sectors in azimuth. T h e total depth of the calorimeter is 26 X0. The radial width of the rings is 30 m m for ring 2 to 9 and 35 and 78 m m for the first and the last ring respectively. The azimuthal segmentation is 22.5 ° . The blue light produced in the scintillator by the electromagnetic shower is collected by means of 50 cm long green Wave Lenght Shifter (WLS) fibers t h a t ~re running perpendicularly to the active planes through holes drilled in the tiles. WLS fibers of type K u r a r a y Y7 with F - P M M A single cladding were used until the end of 1995 and then substituted with K u r a r a y Y l l with double cladding. T h e fibers absorb light at a wavelength of -~420 nm and have an emission peak at 500 nm. A large fiber density, about 0.8 per cm ~, was chosen to reduce considerably nonuniformities in the light collection t h a t otherwise is a well known problem of the wavelength shifting techniques [3]. The calorimeter towers are projective to the IP radially, but are twisted in azimuth by ~ 3 o in order to avoid losses of particles going through the fiber holes. The fibers of each tower are grouped together at the back of the calorimeter by a clamp fixed on a supporting plate. T h e light is read out by phototetrodes, the H a m a m a t s u R2149-03, designed to operate inside the 1.2 Tesla solenoidal magnetic field of DELPHI.

2.1. E n e r g y r e s o l u t i o n T h e energy response in calorimeters built with the "shashlik" technique is strongly dependent on t,he impact point of the incoming particle. At small radii the fiber readout structure produces non-uniformities which were less t h a n -t-3%

221

/X- 4 7 . 5 . * . . ' . :t ~ '. ,, .,. :,, .\., ,.., :, ~ :',

40

:.. '. :, :. :. ,. : . ' . ,

, .",

, ['. " ,~ .":'~ ,~ :' ', :. " :': :,, ,', "., t

' 0

50

100

150

200

250

300

350 ,degrees

SideA.

R =

105mm

o

42,5 40

.... 0

I , , , , I .... 50 1 O0

I .... 150

I .... 200

I .... 250

I ~ , , , I .300 350 ,

SideA.

> o

R =

155

rcm

degrees

5O

/~475 45 42.5 40 0

50

1 O0

150 Side

A.

200 R =

165

250 mm

300

350

rp , d e g r e e s

> 5O 0 /~ 4 7 . 5 Ld V 45 425

, , , , I J i , , I ,', , , I , , , , I J, , / II ,

40 o

50

100

150

200

250

i , I ~ h , , I / 300

350 ,degrees

SideA,

R =

195

mm

Figure 2. The energy response to 45.6 GeV electrons as a function of the azimuthal angle at four different radii. Kuraray Y7 fibers were used in this measurement.

(+2%) of the measured energy with Y7 ( Y l l ) fibers (Figure 2). T h e reason for the nonuniformities is t h a t the light collection is different in the proximity of the fibers. At high radii, where the towers are larger in azimuth, the fibers need to be bent more to be connected to the phototetrodes. Studies on the Y7 fibers demonstrate t h a t they begin to crack when the bending radius is smaller t h a n 8 cm. This results in a tower modulation causing nonuniformities of ± 6 % t h a t replaces the fiber modulation at large radii (lower plot in Figure 2). The

222

S.J. Al~vaag et at/Nuclear Physics B (Proc. Suppl.) 78 (1999) 220-225

new Y l l fibers are more flexible and less brittle and a a significant improvement of the energy resolution at large radius was observed after the change of fibers as can be seen in Figure 3. The

~ ~

55i

r

old, A = l b 6 ± 0 24%. B=I 07iO.06% new, A = ' 4 1 ±0.39%, B=0 6 8 ± 0 09%

ri Eo ~ A- e s

~g'

MC, A = ' 2 3 ± 0 25%, B=I 03±0.05%

5

L L
• old fibers +o • new fibers

906

* new ~ibers, corrected

60 ++

r

002

°1o

1.5

20

2.5

80

E, OeV

P T~-

50

Figure 4. Energy resolution at small radii after energy-map correction with the old Y7 and the new Y l l WLS fibers. The resolution from a G E A N T Monte Carlo simulation is also shown.

~, cr'o

Figure 3. The energy resolution at 45.6 GeV as a function of the radius for a calorimeter with the old Y7 and the new Y11 fibers.

remaining non-uniformities can be further corrected by mapping the energy response as a function of the impact point. The energy response after correction with the energy map has variations of less than +1%. The final energy resolution at small radii with the new fibers is =

(.68

0.09) •

(14.1 + 0.4)

and is in good agreement with a G E A N T prediction (Figure 4). 3. T h e s i l i c o n d e t e c t o r s . Two silicon planes [4] were installed at depths of 4 and 7 X0 in each arm of the calorimeter (Figure 1), covering the polar region between 32.5 nn'ad and 79 mrad. The detectors are made of 300 pm thick high resistivity n-type silicon, with ptype strips implanted on the front and an n + layer

on the back. The strips cover 22.5 ° in azimuthal angle, i.e, one calorimeter sector, and have a radial pitch of ~1.7 mm. The silicon planes are arranged in two concentric crowns. Each 45 ° sector consists of three silicon wafers: the inner covering 45 ° in azimuth, the two outer covering 22.5 °, composed respectively of 2 x 24 and 36 radial strips. Each of these sectors is read out by one Microplex MX4 amplifier chip with an input capacity of 128 channels. The strips are connected to the MX4 by eight Kapton cables, which are matched to the input of the MX4 by an hybrid fan-in card. The aim of the silicon planes in the STIC calorimeter is to provide a reconstruction of the shower axis, to improve the co-ordinate resolution and the two-showers separation. 3.1. T h e p o s i t i o n m e a s u r e m e n t . The impact position of the incoming particle can be evaluated both using the calorimeter information and the silicon detectors [2]. The radial and azimuthal position of the incoming particle are obtained from the energy sharing between nearby calorimeter towers. The resolution is better at the borders of a cell and worse in the cen-

S.J. Alvsvaag et al./Nuclear

Physics B (Proc. Suppl.) 78 (1999) 220-225

ter. Measurements of the distances from both the borders of the cell are performed and combined to improve the resolution near the center of the rings. The third plot of figure 5, shows the radial resolution obtained with 92 GeV electrons; the resolution varies from 0.2 m m at the border of a cell to 0.8 m m in the center. The azimuthal resolution is poorer t h a n the radial one since the cells are larger in the azimuthal than in the radial direction. In ring 4, as example, the resolution in the azimuthal direction varies between 1 m m at the border to 4 m m in the center of the tower.

16 14 I Angular resolution /2 8 6 0.6

L

,

~,

i

+

i

. . . . . . . .

i

. . . .

i

. . . .

,

. . . .

0.4 0.2 0 0.8 % o.6 tz 0.4 0.2

Calorimeter ~

o°4

Combined

.~ 0.3 ~- o.2 0.1

e

. . . .

90

i

. . . .

lO0

]

. . . .

llO

i

. . . .

120

~

. . . . . . . . .

130

140

i

,

,

150 160 170 Radius (ram)

Figure 5. The top plot shows the angular resolution obtained by the two silicon strip planes and the three remaining plots show the radial resolution of the second silicon strip plane, the resolution of the calorimeter and the resolution after combining the two silicon measurements with the calorimetric one. Asterixs correspond to the position of the fibers holes.

Using the silicon strips detectors both the azimuthal angle and the radial position of the shower can be estimated. The signals of 21 strips

223

are used for this measurement. In addition to the strip with the m a x i m u m signal, the +1 nearby strips in azimuth and the ~=3 strips in radius are also used. To calculate the radial co-ordinate of a shower, an estimator based on the logarithm of the fraction: ( E ai = l A i ) j - 1 3 (~i=lAi)j+l

are used where EyAj is the largest of the seven radial pulse height sums over the three azimuthal sectors. The azimuthal angle is evaluated using the same set of strips with a two dimensional barycenter method, where strips are weighted with the energy. The angular resolution of the shower was measured with 92 GeV B h a b h a electrons, dividing the sample into 2 m m radial bins and fitting a Gaussian to the distribution of the difference in reconstructed radius between the two planes. The top plot in Figure 5 shows t h a t the angular resolution is about 9 mrad except in the points marked by stars, t h a t correspond to the position where the holes, through which the WLS fibers pass, are located. There the resolution is poorer and varies between 11 and 15 mrad. The second plot illustrates the radial resolution of the second silicon strip plane. It shows a similar behaviour in the proximity of the fiber holes. The radial resolution from a weighted average of the radii determined from the calorimeter and the two silicon planes is shown in the b o t t o m plot. T h e combined measurement has a resolution which is between 100300 pm. The resolution of the reconstructed radii in both the calorimeter and the silicon detectors is energy-dependent and improves with increasing energy. 3.2. T h e v e r t e x m e a s u r e m e n t The straight line connecting the radial coordinates reconstructed by the two silicon planes can be extrapolated to the beamline to get the vertex of the showering particle. This allows the rejection of backgrounds not coming from the interaction point (IP), e.g., off-energy electrons, that are produced by interactions of the b e a m particles with residual gas atoms in the beampipe. Figure 6 shows the reconstructed vertex for a sample of 80 GeV B h a b h a electrons and

224

S.J. Alosvaag et al./Nuclear Physics B (Proc. Suppl.) 78 (1999) 220-225

~asooo 8h,,bh,,, /i

~20000 1.5000 ! 0000 5000 0 ........... -15 -I0 ~10000 V

sooo L

i

Otl:e,,~r~.,"

,ooo t

4000f 2000 ~ 0 I~, -15

J ./, ~,L,.,, ......... -5 0 5 10 15 Distance ]rom IP (m) ]

f'\

i

/

!

j ~ ..... -10

,

-5

....

,, 0

~ , ~ . . . . "FT;'~v 5 10 15 Distance fi'om IP (m)

Figure 6. Vertex reconstruction of the two silicon planes for 80 GeV B h a b h a electrons (upper plot) and for a sample of off-energy electrons (bottom plot).

tbr a sample of off-energy particles, which were triggered by showers in only one of the calorimeter halves. As expected the B h a b h a events are peaked around the IP, while the off-energy electrons have a much broader distribution. For purely geometrical reasons it is impossible for the silicon planes to reconstruct tile tracks which cross the beamline close to the position of the silicon planes themselves, and this causes the hole in the distribution of off-energy electrons at 2.5 m from the IP. The efficiency of the vertex reconstruction, which is defined as the fraction of events that are detected by both silicon planes, is about 90% for energies larger than 25 GeV. A cut rejecting particles with reconstructed vertex at a distance larger than 0.5 m from the IP, keeps 61% of the Bhabhas and rejects 89% of the off-energy electrons. 4. T h e l a r g e a n d s m a l l v e t o c o u n t e r s . The "large veto counters" consist of two planes of scintillators mounted in front of each calorimeter" (Figure 1), at 2010 and 2050 m m respectively fl'om the IP and covering the polar angular re-

gion from 2.5 ° to 9.8 °. Each scintillator plane is divided into 16 trapezoidal, 10 m m thick counters with an azimuthal segmentation of 22.5 ° , matching the calorimeter, for a total of 64 counters. The counters are made of Bieron BC-408 plastic scintillator. The "small veto" were installed directly on the beampipe. Each detector consists of two 10 m m thick scintillator counters shaped as half rings and they cover the angular region from 1.8 ° to 3.4 ° . The light is colleted by means of WLS fibers, of the same type as used for the calorimeter readout, glued in grooves which were machined on the edges of the large counters and on the outer perimeter of the small counters. The WLS fibers are connected to 10 stage H a m a m a t s u H3165 photomultipliers, located outside the magnetic field, by means of 10 m long clear optical fiber cables. 4.1. T h e n e u t r a l t r i g g e r A single photon trigger, the so-called neutral trigger, was made operational in 1995. The trigger allows a signal in at most one of the two planes of scintillators in front of the calorimeter sector with the shower. The veto counter sector directly in front of the shower, as well as its two neighbours, are considered since the scintillator sectors are overlapping. The small veto counters were also incorporated into the trigger by requiring no signal from the counters on the same side as the calorimeter producing the photon trigger. 4.2. E l e t r o n - p h o t o n s e p a r a t i o n The efficiency of the single photon trigger has been studied with a sample of radiative B h a b h a scattering events with the photon in STIC, one electron in the beampipe and the other electron in the forward lead glass calorimeter [5]. The selected photons had energies between 60 and 90 GeV. The efficiency of the single photon trigger for this sample is shown in the uppermost plot of Figure 7. The efficiency varies between 54% and 10% depending on the thickness of the material that the photons traverse before reaching the veto counter. In the offiine analysis, additional harder cuts on the veto counter pulse height were made. These had a fiat efficiency of 95% (Figure 7 center).

S.J. Alosvaag et al. /Nuclear Physics B (Proc. Suppl.) 78 (1999) 220-225

.....,.. ~ -°'~--°-'°'-°'-o......

0.4

~ :

02

....

.......

............................

,,i ................ 3 4 5

,

1

......

i .........

I I

i .... ',,, 6 7 8 9 Polar Angle (degrees)

225

of e+e - --* 77 events in the B h a b h a sample, it was not possible to measure directly the rejection by the veto counter when a requirement of no signal in both planes was made. However, from the electron efficiency one can estimate it to be 4. 10 -6 . 5. C o n c l u s i o n s

~0.75 - ~

...... i

.....i ..... i... ~......A,.:+....l

o.,~0.25 ~

....

i ...... i t

The "shashlik" technique provides a complete hermetic detector with uniform energy response. The STIC calorimeter has a good energy resolution given by:

~ / E = o.68% ~ 14.1%/~/E. Polar Angle (degrees) -2

2

3

4

5

6 7 8 9 Polar Angle (degrees)

The calorimeter together with the silicon strip detectors can measure the radial position of a shower with a resolution of 100-300 ~m, the shower direction can be measured with an angular resolution of 9 mrad at 92 GeV. The addition of a scintillator veto system provides an effective e - ~ separation. REFERENCES

1.

Figure 7. The trigger efficiency for photons (top), the efficiency of the offline photon identification (center) and the probability t h a t an electron will be misidentified as a photon (bottom). In all plots open circles denote a requirement of no hits in at least one scintillator plane, filled circles no hits in either plane and stars the efficiency of the small veto counter.

The rejection of electrons could be improved flu'ther by requiring no signal in either scintillator 1)lane but then 35% of the photons t h a t survived the trigger requirement and the pulse height cuts were lost (mainly due to albedo i.e. backsplash). The rejection of electrons was studied with a nonradiative B h a b h a sample. The probability t h a t an electron would give a signal in one plane of the veto counter was measured to be 99.8% and the probability t h a t an electron would be wrongly identified as a photon was measured to be (1 4) - 10 . 3 (Figure 7 bottom). Due to the presence

S.J. Alvsvaag et al., Proceedings of The 5th International Conference on Calorimetry in High Energy Physics, BNL, USA, September 1994, Ed. by H.A. Gordon and D. Rueger, World Scientific 1995; S.J. Alvsvaag et al., Proceedings of The Beiring Calorimetry Symposium, Beijing, China, October 1994, Ed. by H.S. Chen, IHEP, Beijing, 1995. 2. S.J. Alvsvaag et al., Preprint C E R N - E P / 9 8 132, September 1998. 3. H. Fessler et al., Nucl. Instr. and Meth. A 2 4 0 (1985) 284: B. Loehr et al., Nuel. Instr. and Meth. A 2 5 4 (1987) 26. 4. S.J. Alvsvaag et al., I E E E Trails. Nuel. Sci. 42 (4) (1995) 469; S.J. Alvsvaag et al.. Nucl. Instr. and Meth. A 3 6 0 (1995) 219. 5. P.Ferrari, V.Hedberg, D E L P H I 98-49 CAL 141, 18 May, 1998