A very forward luminosity monitor for the ALEPH detector at LEP

Nuclear Instruments and Methods in Physics Research A297 (1990) 153-162 North-Holland

153

A very forward luminosity monitor for the ALEPH detector at LEP E. Fernandez *, Ll. Garrido *, M. Martinez *, U.M. Mir and J.A. Perlas

Laboratori de Fisica d'Altes Energies, Universitat Autbnoma de Barcelona, E-08193 Bellaterra, Barcelona, Spain Received 10 April 1990

The design of a very small angle luminosity monitor for the ALEPH detector at LEP is described. The detector consists of four calorimeter modules made of alternating layers of tungsten and scintillators plus one layer of silicon strips for shower localization . Tests of the performance of the calorimeter and initial results obtained in LEP are also described .

1 . Introduction In e + e - colliders such as LEP, the traditional way of determining the luminosity is to measure the event rate of the e + e - - e + e - reaction (Bhabha scattering) where the outgoing particles are produced at small angles with respect to the beam . At these small angles the cross section is well known theoretically and therefore the luminosity can be obtained directly from the measured Bhabha event rate in a detector, which we refer to as luminosity monitor, appropriately located . In the ALEPH experiment at LEP [1] two such monitors

were installed. The ALEPH main luminosity monitor detects and measures the energy of electrons and positrons produced at angles of 55-90 mrad with respect to the beam [1] . At a luminosity of 10 31 cm -2 s -1 , which LEP is expected to reach during the 1990 run, the rate of Bhabha events in this monitor is about 0 .30 Hz . Its goal is to measure the luminosity precisely, after a careful processing and off-line analysis of the data . The main luminosity Bhabha rate is not sufficient for a quick monitoring of the luminosity during data taking, which is desirable for a proper running of the complete ALEPH detector . For this reason, a second

monitor, named BCAL, was also installed. BCAL de.twts electrons at very small poJCltrOnC., produced w eiwav~ and uu angles with respect to the beam, and the Bhabha rate is an order of magnitude higher than that of the main luminosity monitor . As described in section 2, BCAL is located very close to the beam line which makes the

system very sensitive to changes in beam parameters

* Now at CERN .

such as the exact location of the interaction point or the beam divergence . This makes a precise measurement of the absolute luminosity difficult . However, the beam parameters usually stay constant during a given fill of LEP, and under these conditions BCAL can measure the variations in the luminosity during a fill quite precisely. The information provided by BCAL can also be quite useful in those instances when only such a precise measurement of the relative luminosity is needed, as in the measurement of A LR using polarized beams in the proposed scheme for LEP [2] . A detailed study has been made in this context [3] of the dependence of the acceptance of BCAL on possible changes in beam parameters. After correction for these effects, at the level expected to be possible, the remaining error on the acceptance is of the order of 2 to 3 per thousand . A third area where BCAL is useful is in the monitoring of the LEP background conditions, especially of off-momentum beam particles . The information provided by BCAL on off-momentum electron rates is used, along with other background rates observed in other parts of the ALEPH detector, to decide on the

starting time of data taking . In this paper we describe the BCAL monitor and its operation during the first months of LEP running . Section 2 describes the BCAL position in the ALEPH region and its acceptance . Section 3 describes the overall design, as well as the expected performance based on a Monte Carlo simulation . Section 4 is a description of the expected response of one of the components of the calorimeter, namely the scintillators and phototubes. Section 5 describes the performance of the silicon strips used in BCAL as well as the silicon readout electronics . We conclude in section 6 by showing some of the initial data obtained in LEP.

0168-9002/90/$03 .50 © 1990 - Elsevier Science Publishers B .V . (North-Holland)

154

E. Fernandez et al. / A very forward luminosity monitor

2. BCAL location and acceptance The BCAL. detector consists of four identical calorimeters located symmetrically on each side of the beam pipe with respect to the horizontal (bending) plane of LEP, and on each side of the ALEPH interaction region (see fig. 1). This system was designed to detect Bhabha events at the smallest possible angles . To that effect the LEP beam pipe was made elliptical in the region from 7.66 to 7.91 m in z, in order to locate the monitors as close as possible to the beam line. (We use a reference frame centered at the interaction point, where the beam is in the z direction, the x axis is in the plane of the accelerator, pointing towards the center of the ring, and the y axis is vertical to the beam plane) . The ellipse's smaller inner radius (in the x direction) is 6 cm and the active area of the counters starts at 6.35 cm from the beam. Each calorimeter is built inside an aluminum box which also houses the front-end electronics. The box rests in a fixed position on a table supported by the mobile cantilever girder of the LEP superconducting minibeta quadrupoles . This table has adjustable devices so that the calorimeter box can be precisely located with respect to the nominal beam line. The table slides on rails so that the calorimeter can be moved away from the beam pipe when the pipe has to be baked out to improve the vacuum . The superconducting (minibeta) quadrupoles are located in the region 3.7 m < z < 5 .7 m and have a design focusing strength characterized by a k-value of k = 0 .16462428 m -2 . They defocus Bhabha electrons and positrons going towards the monitors from the interaction point, so that the effective minimum angle seen by the monitors is eproduction = 5.1 mrad. On the other hand the monitor is shadowed by beam-pipe

elements such that the maximum acceptance angle is eproduction = 9 mrad. The width of this angular region at the monitor position is only about 1 .3 cm. The acceptance of the monitor has been calculated by Monte Carlo taking into account complete QED radiative corrections to O(a) and Zo self-energy diagrams [4] . At the very small angles of the monitor, weak effects are completely negligible but QED radiative corrections are needed in order to take into account the acollinearity effects due to the emission of a photon, a relevant fact considering the small size of the monitors. The effects of the quadrupole fields have also been simulated . This calculation gives a total geometrical acceptance of 0.7 tLb . However, the efficiency for detecting Bhabhas will be about 70`70 as explained below in section 3. This corresponds to a Bhabha cross-section of about 500 nb. Taking this into account one would expect a counting rate of 5 Hz at a luminosity of 1031 cm - 2s-1 , which would give a measurement of the luminosity with a statistical precision of 2.5% in 5 min . In practice the acceptance depends strongly on the exact energy cut used to define a Bhabha event (see section 3) as well as in the exact location of the four monitors with respect to the actual interaction point . For these reasons we rely on the main luminosity monitor of ALEPH to calculate the acceptance. The ratio of Bhabhas seen in both monitors is found to be quite stable once the cuts used to define Bhabha events are fixed . This ratio, averaged over several runs, together with the acceptance of the main luminosity monitor of ALEPH, is used to define the BCAL acceptance. With the operating conditions that we have used during the 1989 run of LEP, the acceptance of BCAL was a factor of 12.4 that of the main monitor, or (329 ± 3) nb. At a luminosity of 1031 cm -2s-1, the statistical precision with this acceptance is 3.2% in 5 min of monitoring time.

Monitor Beam pipe

2.0 cm

1 6 .5 cm

770 cm Fig. 1. View of the beam pipe in the ALEPH region showing the approximate location of BCAL.

E. Fernandez et al. / A very forward luminosity monitor

15 5

3. BCAL design Each of the four counters consists of a sampling calorimeter made with tungsten converter sheets interspersed with sampling layers made of plastic scintillator, and a plane of vertical silicon strips as described below . The overall shape of the calorimeter is that of a rectangular box of 3 x 5 x 14 cm3 as shown in the schematic drawing of fig. 2. Tungsten was chosen for its small radiation length (3.8 mm for sintered tungsten) needed to locate the calorimeter in the limited space available . The first tungsten layer is 4 radiation lengths thick; this thickness is needed to protect the sampling layers from the high flux of synchrotron radiation photons. The next nine tungsten layers are each 2 radiation lengths thick. The ten sampling layers are made of 3 mm thick scintillators read in pairs by small photomultiplier tubes (section 3) . A plane of silicon strips (section 4) is located after the first 8 radiation lengths, near shower maximum for 45 GeV electrons . Finally, a thick plate of tungsten, 6 radiation lengths, protects the sam-

Fig. 2. View of BCAL showing its location with respect to the beam pipe. pling planes from synchrotron radiation photons entering the back of the calorimeter. Fig . 3 is a photograph of the two calorimeters that go in one side of ALEPH.

Fig. 3. Photograph of the two calorimeters that go in one side of the ALEPH detector. The beam pipe goes in between the two modules .

E. Fernandez et al. / A very forward luminosity monitor

15 6

is the simulated efficiency as a function of the threshold energy cut, expressed as a percentage of the beam energy . The rather poor efficiency reflects the fact that most of the electrons or positrons enter the calorimeters near the edge where both the average energy deposited

É W w W 0 .s

0 .6

0 .4

0 .2

0

1 .2 1 .4 1 .6 Distance to the edge (cm) Fig. 4 . Expected average energy deposited in the scintillator versus distance of point of entry from the edge of the calorimeter . 0

0 .2

0 .4

0 .6

0 .8

1

The energy resolution is dominated by the lateral leakage, since most of the electrons enter the counters very near the edge . Shown in fig . 4 is the expected average energy deposited in the scintillator by a 50 GeV shower versus the point of entry from the edge of the calorimeter. The showers were simulated by the GÉANT program [5] . A Bhabha event will be defined as a back-to-back shower with an energy above a certain threshold and therefore the effect of the leakage is to reduce the Bhabha detection efficiency . Shown in fig . 5 100

and the energy resolution are poor. The efficiency is improved by correcting the energy deposited according to the distance of the point of entry of the particle to the edge of the calorimeter . The determination of this point is the main goal of the silicon strips . The charge deposited in the strips can be clustered to find the centroid of the shower and thus the point of entry of the particle in the front face of the calorimeters . The strips run along the y direction, parallel to the inner edge of the modules. Therefore only the x coordinate of the point of entry, and thus the distance to the edge which is the relevant quantity, is deterYnined . From a simulation of the showers using GÉANT a single plane of strips located near shower maximum was found adequate. The simulation shows that at the silicon plane position the shower width is about 2.5 mm on the average. To have a reasonable sampling granularity, 480 Lm-wide strips with 500 ~Lm pitch, were chosen. In principle the resolution should improve as the width of the strips decreases. The simu-

lation shows that with 500 [,m pitch strips one can obtain a resolution on the entry point of about 0 .5 mm . More narrow strips do not improve significantly this resolution. After correcting the energy deposited by the position of the point of entry of the electron or positron in the calorimeter, the efficiency for detecting Bhabhas is considerably improved . The cut used in practice to define a

Bhabha is a pair of back-to-back showers with both of them depositing 85% (after correction) of the maximum energy . The efficiency without correcting would be only 40% (fig. 5) while, after correction, this efficiency increases to 70%, as mentioned earlier .

80 4. Photomultiplier and scintillator responses

60

The ten scintillator planes are read in pairs by five small photomutiplier tubes . We have chosen the 1.TT'If%INA 1__a_*_ -1 L  TT_rt~_arY~,..a___u T>11L7c 1V CIUG^ PLQJLIC NCadltlHULUI UHU Lite r1Qts R63J02 phototube together with the E1761-04 socket base. Each pair of scintillators is coupled to a small light guide which is in turn optically coupled to the photomultiplier by a small resin disk . The main physical

0

_

20

0

40

50

60

70

80

90

100

Energy of cut (% of E.-)

Fig. 5 . Efficiency of detecting a Bhabha electron versus energy threshold .

'll . . . .-,

constants of the scintillator can be found in table 1 and the most relevant photomultiplier characteristics are shown in table 2 . The small size of the phototubes makes possible a compact design of the monitors. From the graphs in figs . 6 and 7 it is clear that the spectrum of the light

E. Fernandez et al. / A very forward luminosity monitor

Table 1 Main physical constants of the NE102A plastic scintillator Density Refractive index Decay constant (main component) Wavelength of maximum emission

1.032 g cm -3 1.581 2.4 ns 423 nm

a E

157

10~ i

c a+ c e v

10

0

Table 2 Most relevant characteristics for the R1635-02 Hamamatsu photomultiplier 10 mm 300-650 nm 420 nm 82 mA/W 24% 8 -1500 V 2 . 10 6 50 nA 0.8 ns

Diameter Spectral response range Spectral response peak Radiant sensitivity at peak Quantum efficiency at peak Number of dynode stages Maximum supply voltage Current amplification (typ .) Anode dark current (max.) Anode rise time (typ.)

I .

10, '00

300

400

500

600

700 Wavelength (nm)

I

Fig. 7. Photocathode spectral response for R1635-02 photomultiplier. to reach such speed, which is a very important parameter in the trigger design . 5. Silicon strips; design and response

emitted by the scintillator matches perfectly with the spectral response of the photomultiplier. Another important fact is that this choice gives us a fast response (about 4 ns). With other detection techniques (e.g . silicon photodiodes) it would be impossible

120

c 0 0 r a 100 0 c

n

80

60

r

The planes of BCAL strips, build by Micron, have an active area of 5 x 2 cm and a thickness of 300 jim and consist of a parallel array of 40 strips of 480 Rm width and 500 p,m pitch. They reach full depletion typically at 35 V. At this voltage the leakage current per strip is about 40 nA and the capacitance is around 10 pF (see table 3). The electrical characteristics of each individual strip in each wafer received from Micron were tested and a very good uniformity in the leakage current and depletion voltage was found for all the wafers received . The wafers consist of p-type silicon strips deposited over a substrate of n-type silicon so that in each wafer the cathode for all the strips is common whereas each Table 3 Strip characteristics

40

20

0 I 400

5.1 . Strip characteristics

420

440

460

480 500 Wavelength (nm)

Fig. 6. Emission spectrum of NE102A plastic scintillator.

Active dimensions Thickness Number of channels Pitch Single strip area Full depletion voltage (typ .) Dark current (at 35 V) Junction capacitance (at 35 V)

5x2 crié 300 w m 40 0.5 mm 0.25 cm2 35 V = 40nA =10 pF

E. Fernandez et al. / A very forward luminosity monitor

158

Fig. 8. Photograph of a silicon strip plane.

strip has its individual anode connection . The wafers are mounted in a G-10 printed circuit board (PCB) which drives the signals to a standard multipin connector (see fig. 8). The average energy deposited in a single strip located at the center of the shower produced by a 50 GeV electron or photon hitting one monitor, turns out to be equivalent to the average energy deposited by about 64 minimum ionizing particles (MIPs) (about 7.5 MeV for 300 [tm silicon thickness) which corresponds to a charge of about 330 fC.

5.2 . Front-end electronics

The front-end electronics has been built around the AMPLEX Asic chip [6] which is a monolithic analog CMOS signal processor originally developed at CERN for the UA2 experiment. The chip contains 16 channels, each consisting of a charge amplifier, a shaper using continuous time filtering, and a track-and-hold stage (fig. 9). The channel outputs are connected to an analog multiplexer which is controlled by digital circuitry. For

Analog Oul Chwk-in

Clear

Clock-out

gm l

Adjust

gM2

Adjust

Fig. 9. AMPLEX block diagram .

7711

159

E. Fernandez et al. / A very forward luminosity monitor

a power consumption per channel of about 1 mW and a shaper peaking time of about 600-800 ns, the equivalent noise charge (ENC) is about 400 rms electrons plus 33 rms electrons per pF of input capacitance . The design of the charge amplifier, which uses a nonlinear feedback resistor, enables low noise as well as do stability even for increasing detector leakage current up to several hundred nA. The main reasons for choosing this chip were that it was well suited for our strip characteristics and for our readout scheme and that its utilization was rather simple. Additionally, its good do stability allowed a direct do coupling to the strips, therefore simplifying considerably the strip bias scheme, and guaranteeing at the same time good stability even with increasing leakage currents which could occur as a result of possible radiation damage . With strips do coupled and biased at maximum depletion voltage, we measured a typical power consumption per channel of about 2 mW, a shaper peaking

time of about 600 ns, an ENC of less than 1000 rms electrons and a gain of about 4.5 mV/fC, that is, an average signal for a MIP of about 23 mV over a rms noise of about 0.7 mV. Actually just eight channels per chip were used to take full advantage of the 8-event buffer of the LeCroy 1885F Fastbus ADC foreseen for the signal conversion. Since we did not have stringent space restrictions, the 5 AMPLER chips needed for each silicon plane together with all the filtering, clocking and signal conditioning circuitry (mainly SMD components) were placed in a 9 x 12 cm PCB carefully designed to minimize noise and cross-talk [8]. This board can be seen in the photograph of fig. 3. Each silicon strip is directly do coupled to an input of an AMPLER chip which provides the ground level (actually around 0.5 V) of the bias voltage for the depletion of the strip. Then the positive bias voltage is applied after suitable filtering to the common cathode pin of the strips PCB . Each AMPLER chip has its

X5

OUrn

Test IM

fan

n

Fig. 10. Silicon front-end electronics block diagram .

E. Fernandez et al. / A very forward luminosity monitor

160

output connected to a standard operational amplifier acting as a line driver and receives three steering signals which are common for all the chips (fig. 10) : - TRACK/HOLD : to freeze the voltage level of the signal in each channel at the signal peak and allow then a sequential readout. - CLOCK: to clock-in the multiplexer and allow then the sequential analog readout of the voltage level frozen in each channel . - CLEAR : to reset the multiplexer The digital circuitry in the front-end PCB receives two NIM signals (TRACK/HOLD and CLOCK) and produces the three steering signals needed by the AMPLEX chips at the proper voltage level (the CLEAR signal is obtained from the falling edge of the TRACK/HOLD input in order to eliminate control cables) . 5.3 . Readout

A custom NIM ~nodule called MASTER [7J provides the two control signals needed for the front-end electronics (TRACK/HOLD and CLOCK) and controls the dialog with the TRIGGER logic and the ADC (GATE and conversion-in-progress), allowing a simple connection of the detector to any standard DAQ system (fig. 11). The basic operation mode of this module is as follows (fig. 12): when a TRIGGER pulse is received the TRACK/HOLD signal is, after some delay, switched from TRACK to HOLD and simultaneously the TRIGGER input is disabled to avoid being interrupted by any other trigger while the present event is being handled. The delay time is adjusted in such a way that the signal can be frozen at the AMPLEX chip exactly when peak maximum occurs, thus taking into account all the delay time due to cables, trigger logic and steering circuitry . At the same time, this module initiates the dialog with the ADC, waiting until the DAQ system gives a "ready for a new conversion" signal. After this signal is received, the module sends a CLOCK pulse to

MASTER

Fig. 11 . Connections of the Master Module.

the front-end PCB to look for the next channel do level and a slightly delayed GATE pulse to the ADC for the conversion of that voltage. After that, the module waits again for the DAQ system to be ready to convert the next channel, and the process is repeated until all the channels are read out. Then the TRIGGER input in enabled again. Typically the whole readout time is dominated by the ADC conversion and not by the internal delays in the module. 5.4. Tests Purely electronic tests as well as a test using physical sources were done in the different pmts of the silicon detector system, separately for each component, and also with the complete electronics. The silicon strips were electronically tested to determine, for each, the leakage (dark current) and the curve of capacitance versus bias voltage C(Vb) (fig. 13) in order to compute the full depletion voltage. The whole system was tested using physical sources such as light spots and ß sources, and was calibrated using cosmic rays. The ENC of the whole setup was 0 1 0

TRIGGER TRACK/HOLD

1 0 1 0 1 0 1

CIP CLOCK GATE SIGNAL OUT

Fig. 1 2. Control signals of the Master Module .

E. Fernandez et al. / A very forward luminosity monitor u_ a

50

U

c

40

U

30

va vU a .c

20 10

. . 320 280 240 a 00 160 0 0 120

ô

80 40

0

10

20

30

40 50 Bios voltage (volt)

Fig . 13 . Silicon strip capacitance and depletion depth versus bias voltage.

Energy in monitor 4 (GeV) Fig. 15 . Energy deposited in one monitor versus energy deposited in the oposite-side monitor.

rr

â, 70

rN 6050 â

40 30 20 10 0

E Z10

0

10

20 30 40 Strip number

7

400 ,~ 300

c u+

c)

0

E 200 z

100

100 0

10

0

E

Z

200 300 ADC counts

12

10 8

d)

almost the beginning of the LEP operation. The initial analysis of the data taken by BCAL shows that its performance is very close to the expectations . Fig. 15 shows a scatter plot of the energy deposited in module 2 versus the energy deposited in coincidence in the opposite-side module (module 4) for a sample of BCAL events. The energy was determined by summing the signals of the phototubes in each monitor . One can see two clusters, one corresponding to Bhabha events and the other (of lower energy) to accidental coincidences of off-momentum particles hitting the modules

6 4

20

30 40 50 -0 200 400 ADC counts ADC counts Fig . 14. Cosmic rays signal as measured in the strips . measured to be less than 1500 rms electrons, while the typical average signal deposited by a MIP is about 32 500 electrons (fig . 14). Since the awnerage energy deposited to the strip located at the center of the shower is about 64 MIP, which corresponds to 2 x 106 electrons, the signal to noise ratio is almost 1400 .

6. Results from initial operation of BCAL in the ALEPH detector The system was installed in the ALEPH detector during the summer of 1989 . It has been taken data since

â

800

600

400

200

-

5

10

15

20

25

30

35 40 Strip number Fig. 16 . Comparison of the lateral shower profile with Monte Carlo .

E. Fernandez et al. / A very forward luminosity monitor

162 B-l Even t

R.n 5.510 Monitor 3 PM

a

4

3

2

f

Monitor 4 PM

a

4

3

.8

1

CJ D

1100

Aleph Event 3173 Monitor .3 PM

1

2

3

4

3

Monitor 1 PM

1

2

3

4

a

Fig. 17. Display of a BCAL event. The size of the bars is proportional to the energy deposited. which have an energy typically lower than the beam energy . The spectrum of these off-momentum particles intercepted by the two modules is not the same, as it can be seen in the figure . The origin of this asymmetry (which is not present in the other pair of modules) is not clear. The silicon signals are clean and show a shower profile in agreement with the GEANT prediction as can be seen in fig. 16. One can observe saturation at shower maximum, which however does not affect the cluster center determination. Fig. 17 is a graphical display of a BCAL event consisting in a coincidence of signals in back-to-back monitors . The display shows the energy deposited in each phototube and in the silicon strips . The initial analysis of the data also shows that the actual performance of BCAL as an on-line luminosity monitor is adequate . The detailed analysis of the data is in progress and will be the subject of a forthcoming paper.

Acknowledgements We thank the entire ALEPH collaboration for the help received at different stages of the BCAL construction, in particular P. March and H. Taureg for bringing the idea of building an online monitor to our attention and for helpful discussions, U. Shaeffer and H. Seywerd for the help provided with the slow-control hardware and J. Rothberg for the integration into the ALEPH background monitoring. The work of P. Mato was essential in integrating the monitor in the ALEPH data acquisition system, R. Alemany and E. Tubau helped on the installation in ALEPH and A. Pacheco wrote the BCAL event display program. References [1] D. Decamp et al ., Nucl. Instr. and Meth . A294 (1990) 121 . [2] A. Blondel, in : Polarization at LEP, eds G. Alexander et al ., CERN 88-06 (Geneva, 1988) vol. 1, p. 1. [3] H. Burkhardt et al ., in : Polarization at LEP, eds. G. Alexander et al ., CERN 88-06 (Geneva, 1988) vol. 2, p. 95 . [4] J.A . Perlas, A Bhabha event generator for luminosity studies, Universitat Autonoma de Barcelona Preprint UABLFAE 87-03 (1987) . [5] R. Brun et al., GEANT3, CERN, DD/EE/84-1 (1984). [6] E. Beuville et al ., AMPLEX : A low-noise, low-power analog CMOS signal processor for multi-element silicon particle detectors, CERN/EF 89-9 Rev. (1989) . [7] M. Martinez, The MASTER Amplex module, Universitat Autonoma de Barcelona Preprint UAB-LFAE 89-03 (1989) . [8] E. Fernandez, M. Martinez and J.A . Perlas, The use of silicon strips in a small angle luminosity monitor, ECFA STUDY WEEK Proc ., eds. E. Fernandez and G. Jarlskog, CERN 89-10/ECFA 89-124, vol. 1 (1989) .