- Email: [email protected]
Readout of a Si strip detector with 200 μm pitch
Nuclear Instruments and Methods in Physics Research A 377 (1996) 409-411
NUCLEAR INSTRUMENTS & METHOI~ IN P H Y S I C S RESEARCH Section A
ELSEVIER
Readout of a Si strip detector with 200 txm pitch E Chochula a, V. Cindro b'*, R. Jeraj b, S. Ma~ek b, D. Zontar b, M. Krammer c, H. Pernegger c, M. Pernicka c, C. M a r i o t t i d "Comenius University of Bratislava, Bratislava, Slovakia bJo~ef Stefan Institute, Ljubljana, Slovenia ~Institut f iir Hochenergiephysik, Vienna, Austria ~INFN sezione Sanita, Rome, Italy
Abstract A new silicon detector for upgrading of DELPHI tracking in the forward region was developed. Due to space restrictions, the hybrids with the readout chips were placed directly on the detectors. The two different types of readout chips (Viking, MX-6) were used for signal readout in the prototype tests. An optimized hybrid design was required to reduce pick-up on the detector strips. The charge loss for hits on the intermediate strips was measured in the test beam and with a light source, as well as calculated from the electrostatic model.
1. Introduction
2. Calculation of the charge loss
Upgrading of the existing DELPHI detector in the forward region was motivated by the physics requirements at LEP 200. Its main aim is to improve the hermicity of the detector. As a part of the upgrade, a small forward silicon tracker ( V F l r - V e r y Forward Tracker), providing good measurement of the initial direction of tracks, is now under construction [1 ]. It consists of two cones of pixel detectors and two cones of silicon ministrip detectors, the latter being the subject of this paper. The VFT is placed in the space between the beam pipe (53 mm radius) and the inner detector ( l l 6 m m radius), and covers angles between 11° and 25 ° with respect to the beam direction. The space for detectors, front-end electronics, support, cooling and cables is very limited. Special care was taken in order to minimize the dead space and to provide efficient cooling of the detectors. On the other hand, multiple scattering in the material before and after the tracker limits the resolution required from the tracker. A simulation has shown that a position resolution of about 8 0 t , m would be enough to have the required track extrapolation accuracy. In order to minimize power consumption and the number of readout channels, a detector with 200 ~m readout pitch and one intermediate strip was developed and tested.
It is well known that the intermediate strips may improve the spatial resolution of the detector. Charge generated by an ionizing particle on an intermediate strip induces a charge on the adjacent readout strips, thus allowing to use an interpolation algorithm to determine the position with improved accuracy. However, a part of the signal is induced on more distant strips. Since this charge is not recognized in the cluster finding algorithm, the signal to noise ratio is reduced and in the worst case, the detector efficiency may be also reduced. In addition, some charge is capacitively induced on the backplane, which further reduces the signal. Therefore the strip geometry should be optimized in order to have the best possible signal to noise ratio. We performed electrostatic calculations in order to estimate the detector performance before its production. We used an electrostatic model with a thin wire approximation [3]. Three geometrical parameters important for the performance of the detector were varied in the calculations: readout pitch, number and width of the intermediate strips. The readout pitch is given by the detector size and the number of readout chips. We studied two possibilities: two or three chips per detector, thus having pitches of 200 and 133 Ixm, respectively. Results of the calculations are presented in Table 1. We see that the fraction of the charged induced on the two closest neighbouring strips increases with the strip width. On the other hand, the strip capacitance increases, giving a higher contribution to the noise.
* Corresponding author.
0168-9002/96/$15.00 Copyright © 1996 Elsevier Science B.V. All rights reserved PII S0168-9002(96)00202- 1
V. SILICON STRIP DETECTORS
410
P. Chochula et al. / Nucl. Instr. and Meth. in Phys. Res. A 377 (1996) 409-411
Table 1 Calculated fraction of charge induced on closest neighbours and strip capacitance for various detector geometries No. of inter. strips
Readout pitch Strip width
10 ixm
20 txm
40 txm
10 ~xm
20 Ixm
40 ixm
60 I~m
80 Ixm
0 1
Cap. [pF/cm] Signal Cap. [pF/cm] Signal Cap. [pF/cm] Signal Cap. [pF/cm]
0.80 0.58 0.80 0.60 0.80 0.59 0.81
0.96 0.66 0.96 0.69 1.00 0.68 1.03
1.24 0.77 1.33
0.74 0.52 0.74 0.54 0.74 0.52 0.75
0.88 0.58 0.88 0.61 0.89 0.59 0.90
1.09 0.67 1.1 I 0.71 1.16 0.73 1.28
1.27 0.75 1.36 -
1.46 0.82 1.76 -
2 3
133 txm
200 Ixm
-
Taking into account the results of this calculation, we tested detectors having 20 and 40 Ixm wide strips, with one intermediate strip• After analysing the test beam data (see Section 4), we increased the strip width to 60 txm for the detectors which will be installed in the DELPHI experiment.
3. The detector layout
The dimensions of the single sided detectors are 5.3 x 5.3 cm 2. Each detector has 256 readout strips, which are AC coupled to the readout electronics. AC coupling is achieved by a thin (200 nm) layer of SiO 2 between the p+ implant and the aluminum readout lines. The strips are biased with a FOXFET structure. In this way a high dynamic value (typically about 150MI~) of the bias resistances was achieved [2], giving a low contribution of this resistance to the total noise. Due to space limitation, the hybrids fabricated with a thick film technology on 0.5 mm thick BeO ceramics were glued directly on the passivated surface of the silicon detector. We used BeO due to its large radiation length and therefore small contribution to multiple scattering• Even more important is the high thermal conductivity which allows efficient cooling of the chips to be provided via the hybrid-support connection. Two different types of readout chips (VIKING, MX-6) with 128 readout channels were used in the tests• Pick-up of some digital signals by the readout strips caused a saturation of the chip input amplifiers and as a consequence, dead regions in the detector readout. This problem was seen only during the tests of MX chips. The reason is the specific timing of this chip, which requires digital signals (reset, sample one) a few microseconds before the signal of the charged particle is measured. This problem was solved by careful shielding of the critical signal lines on top of the hybrid. The pitch of the readout lines on the detector (200 txm} was matched to the input pitch of the chips (50 txm) by a special interconnection piece (53 × 5.2 mm 2) glued on the hybrid. It was produced by photolithography of thin (0.5 txm) AI sputtered on 200txm thick glass, or on a
polished 3001xm thick AI~O~ ceramic. The minimum width of the lines on the tan-in is 15 Ixm. The contribution of the fan-in to the total capacitance is 1-2 pF and thus negligible as concerns electronic noise. A detector module with two dimensional readout was obtained by glueing two single sided detectors with perpendicular strips back to back, thus forming a double sided module. Since only one hybrid glued on these two detectors has a connector, the connection between two adjacent hybrids is done with thin wires running through small holes (300 ixm diameter) on the edge of the hybrid.
4. M e a s u r e m e n t of c h a r g e loss
We determined the traction of the induced charge on intermediate strips in the test beam fbr detectors with 20, 40 and 60 I~m wide strips by measuring the most probable pulse height of single and double hit events (Fig. 1). The
160 '-! Interpolotlon strip hils 140
"!~j~eodout strip hits 120 loo
! = 60 ~tm
8O
6O
0
........
10
20
30
ii: i -H
40 50 60 ADC counts
70
80
90
100
Fig. I. Measured pulse height spectrum of readout strip hits (solid line) and intermediate strip hits (dashed line) for a detector with 60 ixm wide strips.
P. Chochula et al. / Nucl. Instr. and Meth, in Phys. Res. A 377 (1996) 409 411
Table 2 Comparison of the calculated traction of charge induced on closest neighbours with measurements in a beam test and with a light source Strip width
20 ~tm
40 I~m
60 i.tm
Calculation Test beam Light source
0.58 0.58 0.56
0.67 0.64 0.65
0.75 0.79
results are presented in Table 2. The highest fraction (0.79) was measured for 60 txm wide strips, which will be used for the VFT ministrip detectors. The measured signal to noise ratio for single strip hits is 28:1 using MX-6 chips. This was slightly worse compared to measurements with the Viking chip ( S / N = 40:1 ), which we used for detectors with 20 and 40 Ixm wide strips. One should be aware that the detectors equipped with Viking chips were from a
different producer and were approximately 20% thicker (350 txm compared to 300 ~m). In the final VFT detector the MX-6 chip will be used in VFT to take advantage of having the same chip for the VFT and the barrel microvertex detector.
411
The fraction of the charge induced on neighbouring strips was also measured with a pulsed light source for the 20 and 40 ~m strips. A light emitting diode emitting 1 I~s light pulses with 940 nm wavelength was used. The light spot on the detector had a diameter of about 10 Ixm. The absorption length (a few tens of microns in silicon) of this light is an order of magnitude longer than for visible diodes. In this way possible surface effects on charge collection were reduced compared to visible light. The measured results (Table 2) agree well with the test beam measurement. The spatial resolution of the detectors was also measured in a test beam [4]. It was found to be 35 t~m for perpendicularly incident particles, and improved with angle as shown in Fig. 2. The resolution improves with the track angle as the charge is shared between several strips and thus the analogue signal information can be used for a better estimate of the track position. The measured resolution is better than that required for tracking in the torward region of DELPHI.
5. Outlook The measured signal to noise ratio (28:1) and the fraction of the charge induced on the closest neighbours for hits on intermediate strips (0.79 as measured in the test beam) has encouraged us to start production of the modules for the DELPHI upgrade. We expect to install the complete detector during the 1995/96 shut-down.
E 50 c- 45 o 40 0
~o ,55 Q£ 50 25
References
20 15 10
0
0
1LO
20
310
[1] Proposal for the DELPHI Very Forward Tracker, The DELPHl Collaboration, CERN/LEPC/93-6 92-142. [2] W. Dabrowski et al., Nucl. Instr. and Meth. A 349 (1994) 424. [3] U. K6tz et al., Nucl. Instr. and Meth. A 235 (1985) 481. [4] C. Bosio et al., HEPHY-PUB 607/94, Proc. 6th Pisa Meeting on Advanced Detectors, May 1994, Isola d'Elba, Italy, Nucl. Instr. and Meth. A 360 (1995) 71.
Angle (degrees) Fig. 2. Spatial resolution as a function of the incident angle.
V. SILICON STRIP DETECTORS
NUCLEAR INSTRUMENTS & METHOI~ IN P H Y S I C S RESEARCH Section A
ELSEVIER
Readout of a Si strip detector with 200 txm pitch E Chochula a, V. Cindro b'*, R. Jeraj b, S. Ma~ek b, D. Zontar b, M. Krammer c, H. Pernegger c, M. Pernicka c, C. M a r i o t t i d "Comenius University of Bratislava, Bratislava, Slovakia bJo~ef Stefan Institute, Ljubljana, Slovenia ~Institut f iir Hochenergiephysik, Vienna, Austria ~INFN sezione Sanita, Rome, Italy
Abstract A new silicon detector for upgrading of DELPHI tracking in the forward region was developed. Due to space restrictions, the hybrids with the readout chips were placed directly on the detectors. The two different types of readout chips (Viking, MX-6) were used for signal readout in the prototype tests. An optimized hybrid design was required to reduce pick-up on the detector strips. The charge loss for hits on the intermediate strips was measured in the test beam and with a light source, as well as calculated from the electrostatic model.
1. Introduction
2. Calculation of the charge loss
Upgrading of the existing DELPHI detector in the forward region was motivated by the physics requirements at LEP 200. Its main aim is to improve the hermicity of the detector. As a part of the upgrade, a small forward silicon tracker ( V F l r - V e r y Forward Tracker), providing good measurement of the initial direction of tracks, is now under construction [1 ]. It consists of two cones of pixel detectors and two cones of silicon ministrip detectors, the latter being the subject of this paper. The VFT is placed in the space between the beam pipe (53 mm radius) and the inner detector ( l l 6 m m radius), and covers angles between 11° and 25 ° with respect to the beam direction. The space for detectors, front-end electronics, support, cooling and cables is very limited. Special care was taken in order to minimize the dead space and to provide efficient cooling of the detectors. On the other hand, multiple scattering in the material before and after the tracker limits the resolution required from the tracker. A simulation has shown that a position resolution of about 8 0 t , m would be enough to have the required track extrapolation accuracy. In order to minimize power consumption and the number of readout channels, a detector with 200 ~m readout pitch and one intermediate strip was developed and tested.
It is well known that the intermediate strips may improve the spatial resolution of the detector. Charge generated by an ionizing particle on an intermediate strip induces a charge on the adjacent readout strips, thus allowing to use an interpolation algorithm to determine the position with improved accuracy. However, a part of the signal is induced on more distant strips. Since this charge is not recognized in the cluster finding algorithm, the signal to noise ratio is reduced and in the worst case, the detector efficiency may be also reduced. In addition, some charge is capacitively induced on the backplane, which further reduces the signal. Therefore the strip geometry should be optimized in order to have the best possible signal to noise ratio. We performed electrostatic calculations in order to estimate the detector performance before its production. We used an electrostatic model with a thin wire approximation [3]. Three geometrical parameters important for the performance of the detector were varied in the calculations: readout pitch, number and width of the intermediate strips. The readout pitch is given by the detector size and the number of readout chips. We studied two possibilities: two or three chips per detector, thus having pitches of 200 and 133 Ixm, respectively. Results of the calculations are presented in Table 1. We see that the fraction of the charged induced on the two closest neighbouring strips increases with the strip width. On the other hand, the strip capacitance increases, giving a higher contribution to the noise.
* Corresponding author.
0168-9002/96/$15.00 Copyright © 1996 Elsevier Science B.V. All rights reserved PII S0168-9002(96)00202- 1
V. SILICON STRIP DETECTORS
410
P. Chochula et al. / Nucl. Instr. and Meth. in Phys. Res. A 377 (1996) 409-411
Table 1 Calculated fraction of charge induced on closest neighbours and strip capacitance for various detector geometries No. of inter. strips
Readout pitch Strip width
10 ixm
20 txm
40 txm
10 ~xm
20 Ixm
40 ixm
60 I~m
80 Ixm
0 1
Cap. [pF/cm] Signal Cap. [pF/cm] Signal Cap. [pF/cm] Signal Cap. [pF/cm]
0.80 0.58 0.80 0.60 0.80 0.59 0.81
0.96 0.66 0.96 0.69 1.00 0.68 1.03
1.24 0.77 1.33
0.74 0.52 0.74 0.54 0.74 0.52 0.75
0.88 0.58 0.88 0.61 0.89 0.59 0.90
1.09 0.67 1.1 I 0.71 1.16 0.73 1.28
1.27 0.75 1.36 -
1.46 0.82 1.76 -
2 3
133 txm
200 Ixm
-
Taking into account the results of this calculation, we tested detectors having 20 and 40 Ixm wide strips, with one intermediate strip• After analysing the test beam data (see Section 4), we increased the strip width to 60 txm for the detectors which will be installed in the DELPHI experiment.
3. The detector layout
The dimensions of the single sided detectors are 5.3 x 5.3 cm 2. Each detector has 256 readout strips, which are AC coupled to the readout electronics. AC coupling is achieved by a thin (200 nm) layer of SiO 2 between the p+ implant and the aluminum readout lines. The strips are biased with a FOXFET structure. In this way a high dynamic value (typically about 150MI~) of the bias resistances was achieved [2], giving a low contribution of this resistance to the total noise. Due to space limitation, the hybrids fabricated with a thick film technology on 0.5 mm thick BeO ceramics were glued directly on the passivated surface of the silicon detector. We used BeO due to its large radiation length and therefore small contribution to multiple scattering• Even more important is the high thermal conductivity which allows efficient cooling of the chips to be provided via the hybrid-support connection. Two different types of readout chips (VIKING, MX-6) with 128 readout channels were used in the tests• Pick-up of some digital signals by the readout strips caused a saturation of the chip input amplifiers and as a consequence, dead regions in the detector readout. This problem was seen only during the tests of MX chips. The reason is the specific timing of this chip, which requires digital signals (reset, sample one) a few microseconds before the signal of the charged particle is measured. This problem was solved by careful shielding of the critical signal lines on top of the hybrid. The pitch of the readout lines on the detector (200 txm} was matched to the input pitch of the chips (50 txm) by a special interconnection piece (53 × 5.2 mm 2) glued on the hybrid. It was produced by photolithography of thin (0.5 txm) AI sputtered on 200txm thick glass, or on a
polished 3001xm thick AI~O~ ceramic. The minimum width of the lines on the tan-in is 15 Ixm. The contribution of the fan-in to the total capacitance is 1-2 pF and thus negligible as concerns electronic noise. A detector module with two dimensional readout was obtained by glueing two single sided detectors with perpendicular strips back to back, thus forming a double sided module. Since only one hybrid glued on these two detectors has a connector, the connection between two adjacent hybrids is done with thin wires running through small holes (300 ixm diameter) on the edge of the hybrid.
4. M e a s u r e m e n t of c h a r g e loss
We determined the traction of the induced charge on intermediate strips in the test beam fbr detectors with 20, 40 and 60 I~m wide strips by measuring the most probable pulse height of single and double hit events (Fig. 1). The
160 '-! Interpolotlon strip hils 140
"!~j~eodout strip hits 120 loo
! = 60 ~tm
8O
6O
0
........
10
20
30
ii: i -H
40 50 60 ADC counts
70
80
90
100
Fig. I. Measured pulse height spectrum of readout strip hits (solid line) and intermediate strip hits (dashed line) for a detector with 60 ixm wide strips.
P. Chochula et al. / Nucl. Instr. and Meth, in Phys. Res. A 377 (1996) 409 411
Table 2 Comparison of the calculated traction of charge induced on closest neighbours with measurements in a beam test and with a light source Strip width
20 ~tm
40 I~m
60 i.tm
Calculation Test beam Light source
0.58 0.58 0.56
0.67 0.64 0.65
0.75 0.79
results are presented in Table 2. The highest fraction (0.79) was measured for 60 txm wide strips, which will be used for the VFT ministrip detectors. The measured signal to noise ratio for single strip hits is 28:1 using MX-6 chips. This was slightly worse compared to measurements with the Viking chip ( S / N = 40:1 ), which we used for detectors with 20 and 40 Ixm wide strips. One should be aware that the detectors equipped with Viking chips were from a
different producer and were approximately 20% thicker (350 txm compared to 300 ~m). In the final VFT detector the MX-6 chip will be used in VFT to take advantage of having the same chip for the VFT and the barrel microvertex detector.
411
The fraction of the charge induced on neighbouring strips was also measured with a pulsed light source for the 20 and 40 ~m strips. A light emitting diode emitting 1 I~s light pulses with 940 nm wavelength was used. The light spot on the detector had a diameter of about 10 Ixm. The absorption length (a few tens of microns in silicon) of this light is an order of magnitude longer than for visible diodes. In this way possible surface effects on charge collection were reduced compared to visible light. The measured results (Table 2) agree well with the test beam measurement. The spatial resolution of the detectors was also measured in a test beam [4]. It was found to be 35 t~m for perpendicularly incident particles, and improved with angle as shown in Fig. 2. The resolution improves with the track angle as the charge is shared between several strips and thus the analogue signal information can be used for a better estimate of the track position. The measured resolution is better than that required for tracking in the torward region of DELPHI.
5. Outlook The measured signal to noise ratio (28:1) and the fraction of the charge induced on the closest neighbours for hits on intermediate strips (0.79 as measured in the test beam) has encouraged us to start production of the modules for the DELPHI upgrade. We expect to install the complete detector during the 1995/96 shut-down.
E 50 c- 45 o 40 0
~o ,55 Q£ 50 25
References
20 15 10
0
0
1LO
20
310
[1] Proposal for the DELPHI Very Forward Tracker, The DELPHl Collaboration, CERN/LEPC/93-6 92-142. [2] W. Dabrowski et al., Nucl. Instr. and Meth. A 349 (1994) 424. [3] U. K6tz et al., Nucl. Instr. and Meth. A 235 (1985) 481. [4] C. Bosio et al., HEPHY-PUB 607/94, Proc. 6th Pisa Meeting on Advanced Detectors, May 1994, Isola d'Elba, Italy, Nucl. Instr. and Meth. A 360 (1995) 71.
Angle (degrees) Fig. 2. Spatial resolution as a function of the incident angle.
V. SILICON STRIP DETECTORS