1. CBM workshop – GSI, April 18th – 20th 2007. A. Rivetti Pixel detector development for PANDA A.Rivetti INFN – Sezione di Torino

2. CBM workshop – GSI, April 18th – 20th 2007. A. Rivetti Major requirements  In PANDA the MVD is expected to contribute also to particle ID via dE/dx.  Dynamic range: up to 2.5 MeV of deposited energy (700000 electrons).  Low momentum particles => low material budget.  About 120 modules and 1300 front-end chip will be needed.  Asymmetric hit rate distribution.  Global estimated rate: 1.025 GHz (50 Gbit/s)  Maximum hit rate per module: 34 MHz (1.7 Gbit/s)  Maximum hit rate per front-end chip: 5.8 MHz (290 Mbit/s)  Triggerless read-out  Front-end power < 200mW/chip.  Pixel size: 100  m x 100  m

3. CBM workshop – GSI, April 18th – 20th 2007. A. Rivetti Motivations for R&D  Custom front-end design motivated by the combination of:  Large dynamic range.  Pixel form factor.  Triggerless read-out.  Material budget  Eg. Tests done in Juelich show that the ATLAS chip used in “ PANDA mode ” is at the limit of rate required and would dissipate 330 mW/chip.  In this presentation:  Architecture of the front-end.  Results from a first prototype.  R&D work on epitaxial silicon sensors.

4. CBM workshop – GSI, April 18th – 20th 2007. A. Rivetti Pixel read-out chain

5. CBM workshop – GSI, April 18th – 20th 2007. A. Rivetti Front – end chip Pixel matrix Read – out logic Basic features:  Chip size O(1 cm 2 ).  Technology 0.13  m CMOS.  Adequate radiation tolerance.  Buttable on three sides.  Capable of handling sensors of both polarities.  In each pixel: hit time and charge

6. CBM workshop – GSI, April 18th – 20th 2007. A. Rivetti 3 tasks run simultaneously: Hit detection in pixel and storage of time stamp Readout of pixel data into the column FiFos Readout of FiFos out of the FE  Column read-out ATLAS-like.  Chip read-out optimized for triggerless environment A first architecture

7. CBM workshop – GSI, April 18th – 20th 2007. A. Rivetti Event ordering  Three circuits in the chip ensures the proper time ordering of the hits in case of counter overflow.  The column controller, which generates a counter-overflow signal when the counter overflows after all “old” data in a column was read out.  The read-out controller. If a FIFO shows a CO event, its read-out is stopped till all the FIFO show a CO event.  The EoC event generator. The EoC is generated when all the FIFO show a CO event. The EoC mark is sent outside the chip and the CO flags are cleared.

8. CBM workshop – GSI, April 18th – 20th 2007. A. Rivetti Architecture simulations  Combined Monte-Carlo – VHDL simulations.  Simulation based on a sample of 120000 events on pbar on copper with a beam momentum of 4 GeV/c.  Full read-out cycle simulated on 68000 events.  800 hit lost (1.18 %)  Reasons for losses:  hit below threshold (set at 1200 electrons)  Pile-up on the ToT (0.5 %)

9. CBM workshop – GSI, April 18th – 20th 2007. A. Rivetti Front-end readout time (1)

10. CBM workshop – GSI, April 18th – 20th 2007. A. Rivetti Front-end readout time (2)

11. CBM workshop – GSI, April 18th – 20th 2007. A. Rivetti First prototype for PANDA  A first prototype was designed and tested.  Aim of the exercise: gain experience with the technology (very complex!) and explore the analogue performance.  32 pixel cells with preamplifier with ToT and comparator.  Goal: keep the analogue power consumption below 12  W pixel at 1.2V supply

12. CBM workshop – GSI, April 18th – 20th 2007. A. Rivetti Inside the pixel cell

13. CBM workshop – GSI, April 18th – 20th 2007. A. Rivetti Preamp schematic  Single stage preamplifier design to minimize power consumption  Feedback capacitor of 10fF nominal value  Calibration capacitor of 30 fF

14. CBM workshop – GSI, April 18th – 20th 2007. A. Rivetti Comparator schematic

15. CBM workshop – GSI, April 18th – 20th 2007. A. Rivetti Test board  The chip has been wired bonded to a small sensor provided by ITC – IRST (Trento).

16. CBM workshop – GSI, April 18th – 20th 2007. A. Rivetti Chip and sensor detail Fe chip IRST sensor

17. CBM workshop – GSI, April 18th – 20th 2007. A. Rivetti Performance example

18. CBM workshop – GSI, April 18th – 20th 2007. A. Rivetti Preamp response to 0.5fC Noise@ 0pF: 98 e- rms

19. CBM workshop – GSI, April 18th – 20th 2007. A. Rivetti ToT linearity  Only the region up to 32fC can be explored through the injection capacitor  Preamplifier saturates at 12 fC.

20. CBM workshop – GSI, April 18th – 20th 2007. A. Rivetti A first spectrum ENC: 440 rms electrons Input capacitance about 1.5 pF

21. CBM workshop – GSI, April 18th – 20th 2007. A. Rivetti Epitaxial silicon sensor Highly doped Cz substrate:  = 0.01 – 0.02  cm Lightly doped epitaxial layer  = 50  cm  Several studies indicate that sensor build on epi wafer have good radiation tolerance.  Interest for PANDA: radiation tolerance adequate with standard p-type design. Wafers supplied by ITME and processed by ITC-IRST in Trento  Substrate properties:  Diameter: 100 ± 0.5 mm.  Doping type: n/Sb.  Thickness 525 ± 25  m.  Resistivity: 0.01 – 0.02  cm.  Epi layer properties:  Doping type n/P.  Thickness 50/75/100  m.  Variation: <4%/8%/8%  Resistivity: 2500 – 5000  cm.

22. CBM workshop – GSI, April 18th – 20th 2007. A. Rivetti Epi assemblies  Low cost R&D reusing existing electronics and sensor design:  Electronics: ALICE front-end chip provided by CERN  Sensor: masks developed by INFN-Ferrara for the P326 project.  Three types of epi assemblies:  3 with 150  m total thickness (100  m epi)  7 with 120  m total thickness (75  m epi)  1 with 100  m total thickness (50  m epi)  Several wafers cracked during thinning (problem more severe on thicker wafer)  Problem probably due to stresses on the epi layer: work in progress to fix it  Good assemblies were however delivered and will be tested in May.

23. CBM workshop – GSI, April 18th – 20th 2007. A. Rivetti Summary  The R&D for a custom pixel detector for PANDA has been started.  The architecture of front-end chip is under development.  Design is not yet frozen, many changes still possible!  A first analogue prototype has been designed and tested in 0.13  m.  Good performance obtained for noise, dynamic range, linearity.  The analogue part has been operated with 12  W/pixel.  The results are in reasonable agreement with simulations.  The technology is significantly more complex than the 0.25  m, but it offers several interesting options (triple-well NMOS, analogue components, thicker oxide devices for I/Os, etc..)

24. CBM workshop – GSI, April 18th – 20th 2007. A. Rivetti Credits  Thanks to the many people working on the project in several ways… D. Calvo, P. Deremigis, G. Mazza, S. Martoiu, M. Mignone, R. Wheadon, K. Brinkmann, F. Huegging, T. Stocksmann, R. Jäkel, M. Mertens, J. Ritman…  Thanks to the organizers for the invitation!

25. CBM workshop – GSI, April 18th – 20th 2007. A. Rivetti Discussion slides

26. CBM workshop – GSI, April 18th – 20th 2007. A. Rivetti Some design considerations in 0.13  m  Mobility very different between PMOS and NMOS devices (k N /k P ~6).  The use of NMOS transistor to control small currents becomes problematic (problem much more severe than in 0.25  m).  NMOS input with triple-well and splitted PMOS current sources is probably the best compromise.

27. CBM workshop – GSI, April 18th – 20th 2007. A. Rivetti Issues related to ToT technique  The preamplifier saturates at 12 fC, but the ToT preserves good “linearity” at least up to 40 fC.  Caveat: in the saturated region the open-loop gain of the input stage drops down, making the system more prone to cross-talk. Cin=Cf(1+A0)>>Cd Cf=10fF, Cd=200fF Cin=Cd for A0=20

28. CBM workshop – GSI, April 18th – 20th 2007. A. Rivetti Mastering cross talk  The cross-talk with a saturated preamplifier can go up to 20%.  Possible remedies:  1. Guard-ring connected to ground  2. Two stage preamplifier, with first stage working always in linearity hit event no hit event Pixel Matrix Adjacent Coupled DetectorsC J in2 = 0 J in1  0 Particle hit J in2 CDCD -+-+ CfCf V o2 CSA 02 detector 02 J in1 CDCD -+-+ CfCf V o1 CSA 01 detector 01 Equivalent Circuit hit event no hit event Injected charge C

29. CBM workshop – GSI, April 18th – 20th 2007. A. Rivetti Energy deposited in pixel

30. CBM workshop – GSI, April 18th – 20th 2007. A. Rivetti Hits per module

31. CBM workshop – GSI, April 18th – 20th 2007. A. Rivetti Time to store hit in a FIFO