1. Czochralski Silicon - a radiation hard material? Vertex 2005 November 7 – 11 Chuzenji Lake, Japan Alison Bates The University of Glasgow, UK.

2. 2 Main players: –Ljubljana, CERN, SMART, CNM, Helsinki, BNL and Hamburg. Cz Characterization – Overview Non- irradiated diodes Diodes protons Diodes neutrons Diodes π, e, γ Strip detector Irradiated strip detector Lab tests TCT DLTS, TSC,.. Test beams VA and LHC speed LHC speed Lab measurements study: CV IV Annealing studies TCT provides: Depletion voltage (QV) CCE Space charge determination Electric field profile Trapping times Defect characterization: Specific defect level concentrations Test beams: Towards detector grade components… CCE Resolution S/N,…..

3. 3 A crystalline silicon growth method. –The growth method used by the IC industry. –Recent developments (~3 years) has meant that the silicon is now of sufficient purity to allow use for HEP detectors. Pull Si-crystal from a Si-melt while rotating. Cz Silicon has an intrinsically high level of oxygen. MCz is Cz silicon grown in the presence of an magnetic field. Cheap production. Common production technique. Czochralski Growth What is Czochralski Silicon? IRST (Italy), CNM (Spain), CiS (Germany), Helsinki Institute of Physics (Finland) and BNL (USA) have all successfully produced Cz detectors.

4. 4 Float Zone silicon (FZ) -the usual growth method used to make HEP detectors Single crystal silicon Poly silicon RF Heating coil Float Zone Growth Start with a polysilicon rod inside a chamber either in a vacuum or an inert gas An RF heating coil melts ≈2 cm zone in the rod The RF coil moves through the rod, moving the molten silicon region with it This melting purifies the silicon rod Oxygen can be diffused into the silicon – called Diffusion Oxygenated Float Zone (DOFZ) (done at the wafer level)

5. 5 O in Diffusion Oxygenated FZ (DOFZ) ~ 1x10 17 cm -3 O in magnetic Cz (MCz) ~ 2-5 x10 17 cm -3 Why should Cz be any better? - Oxygen is important DOFZ: Saturation of reverse annealing (24 GeV/c p - only little effect after neutron irradiation observed !) DOFZ silicon has less variation in V fd with radiation compared to FZ – more radiation hard Adding carbon to silicon decreases the radiation hardness For hadron radiation only

6. 6 DOFZ: Saturation of reverse annealing (24 GeV/c p - only little effect after neutron irradiation observed !) Reverse Annealing Component Why should Cz be any better? - Oxygen is important O in Diffusion Oxygenated FZ (DOFZ) ~ 1x10 17 cm -3 O in magnetic Cz (MCz) ~ 2-5 x10 17 cm -3 DOFZ silicon has less variation in V fd with radiation compared to FZ – more radiation hard Adding carbon to silicon decreases the radiation hardness For hadron radiation only

7. 7 Cz Characterization – leakage current 300μm thick 5x5 mm 2 p + n silicon diodes (all 1 kΩcm) have been characterized before and after 24 GeV/c proton irradiation at the CERN PS. Diodes processed at the Helsinki Institute of Physics from FZ, DOFZ and MCz silicon manufactured at Okmetric Oyj. Silicon type α [A/cm] FZ4.96 x 10 -17 DOFZ4.85 x 10 -17 MCz4.73 x 10 -17 The leakage current of MCz silicon after proton irradiation follows the same behaviour as FZ and DOFZ silicon

8. 8 The gradient of the slope after the minimum is β, which is a measure of the radiation hardness Cz Characterization – radiation hard? β has been measured to be smaller for MCz than DOFZ, FZ silicon for 10MeV, 50 MeV and 24 GeV proton irradiations (E. Tuovinen, 4 th RD50 workshop, May 2004) MCz is more radiation hard than DOFZ or FZ silicon (with charged irradiation) The irradiation experiments which have been performed with Cz/MCz are; reactor neutrons 23 GeV protons 10, 20, 30 MeV protons 190 MeV pions 900 MeV electrons Co 60 gamma STFZ DOFZ Cz

9. 9 Illuminate front (p + ) or rear (n + ) side of detector with 660 nm photons –Light penetrates only a few  m depth Ramo’s theorem dictates signal will be dominated by one type of charge carrier I(t)=q E(  (t))  (t) drift –e.g. hole dominated current (hole injection) is produced by illuminating the rear (n + ) side of detector 660 nm laser light p+p+ n+n+ Hole movement Electron collection Low field High field Cz Characterization – Transient Current Technique Time [ns] Need signal deconvoluted from electronic shaping

10. 10 Cz Characterization – Transient Current Technique When the detector has been irradiated the drifting charge, Q e,h (t), will be lost with an exponential time dependence due to trapping in the defects To derive the electric field profile/SC sign you must take trapping effects into account The effective trapping probability, 1/  eff, is the probability that a carrier is lost due to trapping in the silicon. Injection time Injection charge Measured charge from detector Corrected charge Trapping compensation

11. 11 For V>V fd, then:  constant Q collected if no trapping  try various  eff values  correct  eff value is when gradient of this line is zero Charge [arb.] Sqrt V [Sqrt V] Cz Characterization – Transient Current Technique Charge Correction Method (CCM) Example of an electron injection signal collected before (dashed) and after (solid line) the correction for the trapped charge. Details: 15 kΩcm FZ 5.2x10 13 24 GeV/c p/cm 2 V fd = 30 V measured at 90 V. CCM method resulted in  eff,e = 37.8 ns.

12. 12 Type inversion in FZ silicon Electron injection  = 1.74x10 13 24 GeV/c p  = 3.61x10 14 24 GeV/c p Low field High field Time [ns] I [V/50Ohms]

13. 13 TCT in MCz Hole injection MCz silicon always has the high electric field on the structured side of the detector even after high fluences using standard p + - on-n MCz detectors Avoid the expensive double sided processing costs that arise from using n+-on- n silicon detectors 14 Alison Bates Time [ns] I [V/50Ohms] Low field High field NO TYPE INVERSION IN CZ SILICON UP TO 5x10 14 p/cm 2 Hole injection signal 5x10 14 p/cm 2

14. 14 MCz @5x10 14 p/cm 2 Confirmed by Gregor Kramberger et. al

15. 15 Effective doping concentration - FZ Donor Removal N D *exp(-cΦ) Build up of negative space charge Cluster and V 2 0 responsible Resultant N eff shape can be explained by the two processes

16. 16 Donor Removal N D *exp(-cΦ) Build up of negative space charge Cluster and V 2 0 responsible Resultant N eff shape can be explained by the three processes Build up of positive space charge Due to radiation induced donors (linked to O 2 ?) MCz has higher Oxygen content than FZ Effective doping concentration - Cz

17. 17 Trapping times –Effective trap introduction rate, β e, for electrons agree within experimental errors for FZ, DOFZ and Cz silicon. –Effective trap introduction rate, β h, for holes are 10-30% larger than β e for all of FZ, DOFZ and Cz.  Cz silicon has similar trapping to FZ and DOFZ silicon Cz Characterization –trapping

18. 18 Annealing studies - MCz shows excellent annealing behaviour Cz Characterization – annealing G. Pellegrini et. al “Annealing Studies of magnetic Czochralski silicon radiation detectors” NIM A, article in press Cz FZ DOFZ

19. 19 Test beam at the CERN SPS of a MCz detector* before and after irradiation –LHC speed electronics (40MHz) (3 SCTA (analogue) chips) –p + -on-n MCz material –Area read out = 6.1 x 1.92 cm –380  m thick –1150  cm (after processing) –50  m pitch parallel strips –V dep measured = 420 V (CV) * Many thanks to the Helsinki Institute of Physics for the MCz detector Cz Characterization – test beam NIM A 535 (2004) 428

20. 20 MCz test beam results S / N > 23.5 + 2.5 (380  m thick) Depleted the detector (~550 V) (CV measured Vdep ~ 420 V) 1.3 x 10 14 24 GeV p/cm 2 S/N = 15 4.3 x 10 14 24 GeV p/cm 2 S/N = 11 (under depleted) 7.0 x 10 14 24 GeV p/cm 2 S/N = 7 (under depleted) Signal [ADC Counts] Bias Voltage [V]] Signal [ADC Counts] Bias Voltage [V]] Unirradiated Detector Irradiated Detector

21. 21 Czochralski silicon is a cheap and standard industrial method for growing high purity silicon. Cz silicon –shows increased radiation hardness when compared to FZ or DOFZ with charged irradiation –does not type invert with charged particle radiation (up to a 24 GeV/c proton fluence of 5.10 14 p/cm 2 ) –has the same trapping behaviour as FZ and DOFZ –has small variation in N eff with annealing time Cz strip detector read out with LHC speed electronics shows promising results both before and after irradiation. Cz Characterization – Conclusions Is Czochralski silicon something to get excited about?

22. 22 Back up slides

23. 23 The evolution of the depletion voltage as determined by CV and IV methods for MCz silicon. The evolution of the depletion voltage as determined by CV and IV methods for DOFZ (d1) silicon

24. 24 IRST, CNM, CiS, HIP and BNL have successfully produced Cz detectors. Sumitomo is no longer accessible and Okmetric Oyj require large orders (>1000 wafers per order) Cz Characterization – procurement

25. 25 Confirmation of MCz depletion voltage behaviour of MCz after 24 GeV/c proton irradiation by G. Pellegrini et. al “Annealing Studies of magnetic Czochralski silicon radiation detectors” NIM A, article in press. MCz FZ DOFZSCSI? SCSI Cz Characterization – proton irradiation

26. 26 Confirmation of MCz depletion voltage behaviour of MCz after 190 MeV/c pion irradiation by G. Lindstroem et. al 1 st RD50 workshop, October 2002 MCz DOFZ FZ Cz Characterization – pion irradiation

27. 27  The minimum of V dep is reached at 1-1.5×10 14 n/cm 2.  V dep is  650 at 10 15 n/cm 2. Measurements after irradiation and before annealing Cz Characterization – neutron irradiation

28. 28 Cz Characterization – leakage current 5x5 mm 2 p + n silicon diodes have been characterized before and after 24 GeV/c proton irradiation at the CERN PS. Diodes processed at the Helsinki Institute of Physics from FZ, DOFZ and MCz silicon manufactured at Okmetric Oyj. Crystal orientation ρ (kΩcm)OxygenationThickness (μm) Initial V fd (V) FZ 1-295 ± 2235 ± 15 DOFZ 175h at 1100 o C295 ± 2269 ± 7 MCz 1-304 ± 2309 ± 5 Depletion voltages for FZ, DOFZ and MCz diodes before irradiation Single guard ring was always connected.

29. 29 Cz Characterization – proton irradiation Diodes measured after 24 GeV/c proton irradiation and 4mins/80 o C annealing with IV and CV techniques. Guard ring connected. CV measurements made with 10kHz in parallel mode. FZ silicon SCSI

30. 30 Diodes measured after 24 GeV/c proton irradiation and 4mins/80 o C annealing with IV and CV techniques. Guard ring connected. CV measurements made with 10kHz in parallel mode. MCz silicon SCSI SCSI? Cz Characterization – proton irradiation

31. 31 TCT in MCz 14 Alison Bates I [V/50Ohms] Low field High field Time [ns] I [V/50Ohms] Low field High field 14 Confirmed by Gregor Kramberger et. al Hole injection signal 5x10 14 p/cm 2 NO TYPE INVERSION IN CZ SILICON UP TO 5x10 14 p/cm 2 MCz silicon always has the high electric field on the structured side of the detector even after high fluences using standard p + -on-n MCz detectors

32. 32 CERN TCT set-up(1) Easy detector mounting Floating guard ring Front and back illumination possible Peltier cooled to ~-10 o C Temp. stability to +0.1 o C Flushed with N 2 gas Red 660 nm laser diode IR 1060 nm laser diode Amount of charge deposited can be tuned - laser diode output controlled by pulse generator signal Cu/Be spring contact to front pad Au PCB for ground plate Laser fibre for illuminating the top of the detector Water cooling and gas system

33. 33 CERN TCT set-up(2) Pulse duration min 1.5 ns FWHM Rise time of signal 1.5 ns Almost no detector shaping from electronics Custom written LabVIEW DAQ ROOT analysis of data* * Data analysis software courtesy from the wonderful Gregor Kramberger

34. 34 Signal treatment Deconvolution of the true signal from the measured signal Measured signal = detector signal  transfer function Transfer function: I(t)=  TCT /R x dU osc (t)/dt + U osc (t)/R R = 50  from input of preamp  TCT = RC d (C d = detector capacitance) Time [ns] I [V/50]

35. 35 Charge Correction Method The CCM assumes three conditions: –There exists one dominant trapping time –The detrapping effects are negligible in the readout time –All the lost charge is due to trapping The method requires no knowledge of the electric field profile in the detector or any information about the charge carrier distributions. All plots presented in this paper have been deconvoluted from the electronic transfer function of the TCT readout circuit and corrected for the trapped charge.

36. 36  parameter summary ee hh [10 -16 cm 2 /ns] FZ (f2)5.59 + 0.297.16 + 0.32 DOFZ (d1)5.73 + 0.296.88 + 0.34 MCz (n320)5.81 + 0.327.78 + 0.39 DOFZ (W317)5.48 + 0.226.02 + 0.29 Dortmund [2] DOFZ 5.08 + 0.164.90 + 0.16 Ljubljana [3] DOFZ and FZ 5.34 + 0.197.08 + 0.18 Lancaster/Hamburg [4] FZ5.32 + 0.306.81 + 0.29 Hamburg [5] FZ, DOFZ and Cz 5.07 + 0.166.20 + 0.54 Table 4. Comparison of βe and βh determined after 24 GeV/c proton irradiation. The top 4 rows are the values found in this work while the last four rows show data previously obtained by other groups. All values have been scaled to 5oC, for the temperature dependence of β (see section 4.5).

37. 37 What’s the limitations with FZ detectors p + n detectors deplete from the front segmented side before irradiation p + n FZ detectors type invert at a certain radiation level and then deplete from the back side of the detector The V fd is increasing every day hence at some point the detector will be operated under-depletion, in which case: –Charge spread – degraded resolution –Charge loss – reduced CCE p + n detector before type inversion and under-depleted.

38. 38 What’s the limitations with FZ detectors p + n detectors deplete from the front segmented side before irradiation p + n FZ detectors type invert at a certain radiation level and then deplete from the back side of the detector The V fd is increasing every day hence at some point the detector will be operated under-depletion, in which case: –Charge spread – degraded resolution –Charge loss – reduced CCE

39. 39 n-on-n silicon, under-depleted: Limited loss in CCE Less degradation with under-depletion Depletion fraction resolution n + on-n Charge spread for p + on-n Si For LHCb n + on-n detectors are the technology choice

40. 40 Macroscopic changes Shockley-Read-Hall statistics (standard theory) Trapping (e and h)  CCE shallow defects do not contribute at room temperature due to fast detrapping charged defects  N eff, V dep e.g. donors in upper and acceptors in lower half of band gap generation  leakage current Levels close to midgap most effective Increased NoiseReduced SignalEffects the operating voltage e e

41. 41 Depletion voltage in FZ silicon N eff – Effective doping concentration N eff positive – n-type silicon (e.g. Phosphorus doped – donor) N eff negative – p-type silicon (e.g. Boron doped – acceptor) Donor removal and acceptor generation –type inversion: n  p –depletion width grows from n + contact N eff (0) is the effective doping concentration before irradiation  = 0.025cm -1 measured after beneficial anneal before inversion after inversion n+n+ p+p+ n+n+

42. 42 Defects located close to the middle of the bandgap can generate current. Damage parameter  (slope)  independent of  eq and impurities  used for fluence calibration Reverse current and Carrier Trapping T dependence Defects can trap the charge carriers CCE = Charge Collection Efficiency CCE is reduced by radiation induced traps Problems arise if the de-trapping time becomes less than 25ns for the LHC t is the carrier transient time (e or h), β is a constant. Material independent