1. 24/02/2010Richard Bates, 5th Trento workshop, Manchester1 Irradiation studies of CNM double sided 3D detectors a. Richard Bates, C. Parkes, G. Stewart b. G. Pellegrini, C. Fleta, M. Lozano, M.Ullan c. U. Parzefall, M. Koehler a) University of Glasgow, Department of Physics and Astronomy, Glasgow, UK, b) Instituto de Microelectronica de Barcelona, IMB-CNM-CSIC, Barcelona, Spain, c) Physikalisches Institut, Universität Freiburg, Germany

2. 2 24/02/2010 Richard Bates, 5th Trento workshop, Manchester Contents Introduction –Motivation –CNM Double-sided 3D detectors Alibava system –Set-up –Calibration Charge collection – Response to Sr-90 electrons –High bias operation –Comparison with testbeam Some notes on fabrication Summary

3. 3 24/02/2010 Richard Bates, 5th Trento workshop, Manchester Radiation-hard detectors for the (s)LHC ATLAS Forward Physics programme (2012) –Fluence ~ 1 x 10 15 1MeV n eq /cm 2 Inner pixel layer (IBL) replacement for ATLAS (2014) –Fluence of 4.4 x 10 15 1MeV n eq /cm 2 Super-LHC (2017) –10 * luminosity upgrade on present LHC –Fluence up to ~ 1 x 10 16 1MeV n eq /cm 2 LHCb VELO upgrade –Integrated fluence ~ 1 x 10 16 1MeV n eq /cm 2 Design fluences for ATLAS sensors (includes 2x safety factor) : Innermost Pixel Layer ~ 5cm radius : 1.6 x 10 16 1MeV n eq /cm 2 = 500 Mrad Outer Pixel Layers ~ 30cm radius: 3 x 10 15 1MeV n eq /cm 2 = 150 Mrad

4. 4 24/02/2010 Richard Bates, 5th Trento workshop, Manchester CNM double-sided 3D detectors Array of electrode columns passing through substrate Electrode spacing << wafer thickness (e.g. 30  m:300  m) Benefits: –Low full depletion voltage –Fast collection times → Reduced charge trapping –Reduced charge sharing due to E-field shape → Higher signal in one pixel More complicated fabrication - micromachining Double sided 3D p-type detector

5. 5 24/02/2010 Richard Bates, 5th Trento workshop, Manchester Hole aspect ratio up to 25:1 (columns are 10µm diameter, 250µm deep) Poly 3D Electrodes Electrode fabrication: 1.ICP etching of the holes: Bosch process, ALCATEL 601-E 2.Holes partially filled with 3 µm LPCVD poly 3.Doping with P or B 4.Holes passivated with 1 µm TEOS SiO 2

6. 6 24/02/2010 Richard Bates, 5th Trento workshop, Manchester 4 inch wafer CNM 3D-wafer Medipix2 ATLAS Strips Pads CNM Double-sided 3D fabrication runs ICP is a reliable and repeatable process (many test runs always successful). Yield: Strip detectors = 88%. ATLAS pixels = 50% Devices designed at Glasgow & CNM Fabricated at CNM 3D guard ring 80 µm 50 strips DC coupled 50 electrodes/strip 4mm long strips Devices » n + strips » p - bulk » p + back contact DC coupled devices Only one guard ring 10um wide Substrate p- and n-type 285µm thick

7. 7 24/02/2010 Richard Bates, 5th Trento workshop, Manchester Short strip detectors –Electron collecting, n + readout columns Irradiated at Karlsruhe with 26 MeV protons. –Irradiated cold –Not biased No intentional annealing –max 5 days RT during module assembly Detectors glued to Ceramic base boards –RC pitch adaptors from VTT/Helsinki Institute of Physics Irradiated strip detectors 26 MeV protons scaled to 1 MeV neutron equivalent fluence with a hardness factor of 1.85 Fluence (1 MeV n eq /cm 2 ) 5E14 1E15 2E15 5E15 1E16 2E16

8. 8 24/02/2010 Richard Bates, 5th Trento workshop, Manchester Alibava system - Hardware Daughter board Mother board Detector Bias Trigger Input PMT or Laser TTL Mother Board Control Beetle chips Trigger on PMT signal Process analogue data from readout chips Communicate with a PC via USB. Daughter board Two Beetle readout chips Re-bondable fan-ins H.V. filter operational to 1kV Beetle front end (from LHCb) –LHC speed bi-polar amplifier (25ns peaking time) –Full analogue readout Rest of hardware –Collimated Sr-90 source above the detector –Scintillator with PMT trigger below the detector –Everything in a freezer with dry air and monitoring

9. 9 24/02/2010 Richard Bates, 5th Trento workshop, Manchester Alibava system - Software RAW data Pedestal corrected Common mode noise corrected S/N cuts applied Pulse shape Signal spectrum Raw data : ADC counts vs channel number Raw data with pedestal value for each channel corrected (ADC counts vs channel number) Signal corresponding to an event (trigger): ADC counts vs channel number Pulse shape: collected charge vs. delay Signal spectrum with in time cut: number of events vs. collected charge

10. 10 24/02/2010 Richard Bates, 5th Trento workshop, Manchester Calibration Planar strip detector –SCT Barrel miniature p + readout strips, n - bulk 300 µm thick 80 µm pitch, 1 cm long AC coupled strips –Hole collection –Plateau value taken as full charge collection in planar device

11. 11 24/02/2010 Richard Bates, 5th Trento workshop, Manchester Zero fluence data Pulse height spectra Average waveform noise floor at ~20ACD counts Note: This is a 4mm long strip detector The noise is large due to the capacitance. Pixels detectors are only 50 - 500µm long 100% Charge Collection from all non-irradiated devices –c.f.1 st production run where low CCE of 85% observed –Scaled to 285 µm full device thickness

12. 12 24/02/2010 Richard Bates, 5th Trento workshop, Manchester Collected charge for irradiated devices Bias Voltage fixed at 150V for all irradiated samples Non-irradiated sample biased at 18V Detector’s ceramic based board temperature between -10°C to -15°C Electron collecting strip detectors

13. 13 24/02/2010 Richard Bates, 5th Trento workshop, Manchester Simulation – ISE TCAD n+n+ n- 250  m 50  m Charge collection simulation 300  m substrate NB – simulation assumes n+ readout for detectors, and 55  m pixels, experimental results have n+ readout and 80µm pixels D. Pennicard et al., IEEE Trans. Nucl. Sci. 54, 4 (2007) D. Pennicard et al., Nucl. Instr. Meth. A 592 (2008) 16 10  m column radius p+p+

14. 14 24/02/2010 Richard Bates, 5th Trento workshop, Manchester CCE with increasing bias voltage Detector response after a fluence of 1 x 10 15 1 MeV n eq cm -2

15. 15 24/02/2010 Richard Bates, 5th Trento workshop, Manchester CCE with increasing bias voltage Charge multiplication? Detector response after a fluence of 1 x 10 16 1 MeV n eq cm -2

16. 16 24/02/2010 Richard Bates, 5th Trento workshop, Manchester Spectra after 1 x 10 16 Bias Voltage is 300V

17. 17 24/02/2010 Richard Bates, 5th Trento workshop, Manchester Spectra after 2 x 10 16 Bias Voltage is 250V

18. 18 24/02/2010 Richard Bates, 5th Trento workshop, Manchester High bias operation Bias voltage applied maximum possible before excess current or noise typically 250 to 350V Increased CCE for high fluences > 5 x 10 15 Close to 100% CCE for 10 16 1 MeV n eq cm -2 More than 100% CCE for fluences 0.5 to 2 x 10 15 1 MeV n eq cm -2 Possible charge multiplication observed (?) Observed in heavily irradiated planar devices with kV bias

19. 19 24/02/2010 Richard Bates, 5th Trento workshop, Manchester Comparison with Planar devices 3D results compared to Planar results complied from published work 3D measured at 150V Planar measured at 1000V 3D measured at ~300V Planar measured at 1000V

20. 20 24/02/2010 Richard Bates, 5th Trento workshop, Manchester Strip testbeam work Planar and 3D strip detectors tested with APV frontend electronics last summer in testbeam at CERN Same type of CNM strip devices tested as with lab tests Irradiated doses: –0 –1 x 10 15 1 MeV n eq /cm 2 –2 x 10 15 1 MeV n eq /cm 2 Details in M. Koehler’s talk. Just showing charge collection as a function of bias.

21. 21 24/02/2010 Richard Bates, 5th Trento workshop, Manchester Collected Charge vs Voltage Saturated signal of unirradiated detector: ~35 ADC 1x10 15 : max. signal ~70 ADC 2x10 15 : max. signal ~65 ADC Charge collected in testbeam is very close to lab tests e.g. for 1x10 15 : 100% cce at 150V, 140% cce at 200V unirradiated 2x10 15 1x10 15 Strong charge multiplication seen!

22. 22 24/02/2010 Richard Bates, 5th Trento workshop, Manchester Fabrication RD50 wafer with FE-I3 device did not work –This was due to a fancy punch-through biasing structure that was washed out during the 3D processing New FE-I3 wafers (6 off) ready for end March 2010 –Punch-through structure removed Made Medipix3 and Medipix3 quad devices –Medipix is 256 x 256 pixels (50um x 50um) giving an area of 2cm 2 –Quad is 4 devices together –These will be tested soon –Yield data will follow.

23. 23 24/02/2010 Richard Bates, 5th Trento workshop, Manchester Summary Non-irradiated 3D double sided devices show 100% CCE Post irradiation CCE, measured at 150V, falls –CCE of 62% : 14000 e - after 5x10 15 –CCE of 47% : 11000 e - after 1x10 16 Charge collection agrees reasonably with simulation prediction Enhanced charge collection with high bias Possible charge multiplication in high electric fields –30000 e - for 200V bias up to a fluence of 2x10 15 –CCE of 98% : 22500e - after 5x10 15 measured at 250V –CCE of 94% : 21500e - after 1x10 16 measured at 350V –CCE of 53% : 12000e - after 2x10 16 measured at 300V Summary of noise and currents to be done Similar charge collection seen in testbeam –Detailed studies on going and simulation to start Sufficient charge for ATLAS front-end chip Operation at -10°C to -15°C fine FE-I3 devices ready soon for test beam and irradiation studies