1. Radiation damage to electronic devices for LHC and Super-LHC experiments 1 Presented by Julien Mekki IES, University Montpellier II, France CERN, Geneva, Switzerland Seminar IPNL – Lyon – France 14 th January 2011

2. 2Outline I.Introduction II.Packaging effect on RadFET sensors for the radiation monitoring project III.Forward biased p-i-n diodes used as dosimeters IV.Perspectives and outlook on future studies V. Conclusion

3. 3 Who am I ? Actual position Assistant professor at University of Montpellier 2 – CERN USER. Study of silicon detector performances for LHC and Super-LHC experiments. PhD in Electronics (Nov. 2009) – (CERN, Université Montpellier 2) Characterization and performance optimization of radiation monitoring sensors for high energy physics experiments at the CERN LHC and Super-LHC Master thesis in Science and Technology (2006) (CNES, EADS Astrium) Radiation hardness of electronic components used for space applications. February 2011: Senior fellow at CERN – Emerging Energy Technologies department Project: Radiation to electronics (R2E)

4. 4 CERN Technologies Three keys technologies at CERN Accelerating particle beams Detecting particles Large-scale computing (Grid) Concorde (15 Km) Balloon (30 Km) CD stack with 1 year LHC data! (~ 20 Km) Mt. Blanc (4.8 Km) GRID

5. 5 CMS The L arge H adron C ollider (LHC)

6. 6 7 TeV Mixed radiation field H adrons (n, p, k +,k -, π +, π - ) L eptons (e -, e +, μ -, μ + ) P hotons Intense close to the interaction point General design principle Sub-detectors: 1)Inner detectors → Trackers 2)Calorimeters → Energy deposited 3)Muon gas chamber ATLAS

7. 7 Radiation monitoring project – Why ? The effect of radiation on electronic and detector components → Major issue All equipments → Affected by radiation damage LHC experiments are designed to operate for 10 years. → Radiation level survey needed for damage and failure analysis. Different radiation field parameters have to be monitored… Different sensitivity and range are required… Different small active devices have been investigated !

8. 8 Radiation monitoring project – What ? r (cm)Z (cm) Dose (Gy/y) Annual Ф eq (×10 13 (n eq /cm 2 )) A 20-30 80-90173822.9 B40-50340-35056252 C80-90340-35012491 Ionization effect TID (Total Ionizing Dose) e.g. accumulation of charge in SiO 2 : damage to microelectronic components Unit Gray: 1 Gy = 1 Joule released in 1 kg of matter = 1 J/kg Non-Ionizating effect NIEL (Non Ionizing Energy Loss) causing e.g. crystal defects in semiconductor crystals: silicon detector damage Unit: 1 MeV neutrons/ cm 2 “equivalent fluence” (Ф eq ) In space (Geostationary orbit) 10-30 Gy/y ATLAS Ideal: Measure the full radiation spectrum (particle type, energy and intensity at all locations) → Impossible (there is no such device) Luminosity: 10 34 cm -2.s -1

9. Many radiation sensors tested, only few of them was selected and installed in the LHC experiments 2 major issues: 9 Radiation monitoring project – How ? Radiation monitoring project – How ? 2 types of RadFET: 250 nm oxyde thickness → REM,UK 1600 nm oxyde thickness → LAAS, France BPW34 Commercial Silicon p-i-n diodes Measure of the 1-MeV Ф eq Measure of the TID

10. Many radiation sensors tested, only few of them was selected and installed in the LHC experiments 2 major issues: 10 Radiation monitoring project – How ? Radiation monitoring project – How ? 2 types of RadFET: 250 nm oxyde thickness → REM,UK 1600 nm oxyde thickness → LAAS, France BPW34 Commercial Silicon p-i-n diodes Measure of the 1-MeV Ф eq Measure of the TID Packaging can induce possible dose enhancement in the measurements. The only freedom remaining in the design is the chip carrier cover. But, like the chip carrier, it has an effect on the TID measurement.

11. 11 Packaging effect on RadFET sensors for the radiation monitoring project Packaging effect on RadFET sensors for the radiation monitoring project

12. 12 RadFETs General (1) e - /h + pair generation; (2) e - /h + pair recombination; (3) e - / h + transport; (4) hole trapping; (5) Interface state. Build-up of charge in SiO 2  increase of the p-MOS Threshold Voltage  integrated Dose Measurement Exposure: “zero bias” Readout: i DS V GS ∝ TID

13. γ -neutron Irradiation Chip carrier was placed into the reactor core Various materials and thicknesses Measurement: dose Slight increase of TID was measured for thicknesses exceeding 1 mm. 13 Ref : F. Ravotti. Phd thesis, University Montpellier II, France.

14. 14 Packaging Effect on RadFET sensors Packaging Effect on RadFET sensors How the RadFETs response is influenced by the cover ? and also …. How much dose is deposited by different particles with different energies in the RadFETs ? RadFET response studied using the simulation toolkit:

15. 15 What is What is C++ based / Object Oriented Toolkit for the simulation of particle interactions with matter. Geant4 provides the possibility to describe accurately an experimental setup. (Geometry and Materials) GEometry ANd Tracking The program provides the possibility of generating physics events and efficiently track particles through the simulated detector. The interactions between particles and matter must be simulated by taking into account all possible physics processes, for the whole energy range.

16. Geant4 Model 16 Packaging REM-TOT-500 LAAS-1600 Without coverWith ceramic cover Chip carrier has been hit perpendicularly in the front side. Result of the simulation is the total energy deposited by primary and secondary particles. First set of simulation: → Full dies size are taken as sensitive volume Second set of simulation: → sensitive volume: thin oxide Layer (SiO 2 )

17. 17 Packaging comparison Results for Pions: Charged hadrons are dominated by pions close to the interaction point. Most important contribution on the total energy deposited in a mixed field. Low energy pions are absorbed in the cover. Simulation have been carried out for all particles and energies present in the LHC radiation field.

18. RadFET sensors in the ATLAS detector Provide information about the TID in the LHC experiments 260 µm cover has been investigated and compared to uncovered RadFETs 2 locations are taken as example: Inner detector (1) Liquid Argon Calorimeter (2) Estimation of the total energy deposited in the RadFETs as well as the cover effect for each particle type. 1 2 18

19. % of the total number of particles Annual dose without cover in units of kGy/year and (contribution %) Annual dose with cover in units of kGy/year and (contribution %) Dose enhancement (%) (260µm /no-cover) Protons1.2 1.73×10 0 (26.7)1.65×10 0 (22.9) -4.3 Photons54.9 2.46×10 -2 (0.4)8.29×10 -2 (1.1) 237.1 Electrons5.9 1.13×10 0 (17.4)1.34×10 0 (18.4) 19.4 Pions10.8 2.89×10 0 (44.7)3.39×10 0 (46.4) 17.2 Neutrons25.2 3.17×10 -2 (0.5)3.72×10 -2 (0.5) 17.2 Muons1.9 6.66×10 -1 (10.3)7.42×10 -1 (10.5) 11.5 Total dose enhancement: 12.1±1 % About 45 % of the energy is deposited by pions. Significant dose enhancement for photons due to secondary particles. Photons deposit less than 2 % of the overall energy. Detailed results for the Inner Detector :Results 19

20. 20Results Pions: Pions are charged hadrons: heavy particles Mass 270 times higher than e -. Energy deposited → Bragg peak Photons: Secondary particles (e -, e + ) → Compton, pair production effects → Photonuclear absorption ( α ) Energy deposited in the medium (MeV.cm -2.g -1 ) Depth (cm) Bragg peak Compton e - Alpha (e - ; e + )

21. 21Results Results for the Liquid Argon Calorimeter: Total Dose enhancement = 23.6 ± 2.4% Pions represent 0.1 % of particles → contribution to dose ≈ 7 % Protons deposit about 35% of the overall energy (represent only 0.08 % of particles, but mass 1800 times higher than e -.) Annual dose values in the covered and uncovered RadFET sensors for both locations. Inner DetectorLiquid Argon Calorimeter Simulated TID (SiO 2 ) Without coverWith coverWithout coverWith cover 6.5 kGy/year7.3 kGy/year5 Gy/year6.1 Gy/year

22. 22 Conclusion of this study Dose enhancement as TID was simulated using Geant4 for all particles and energies present within the LHC radiation field. Understanding of each particle and energy influence. 260 µm thick Alumina cover can alter the measured dose up to 25 %. The choice of RadFET packages is thus important for measuring the TID in High Energy Physics Experiments. Study published in J. Mekki et al, IEEE TNS, vol. 56, no. 4, pp. 2061-2069, 2009.

23. 23 Forward biased p-i-n diodes used as dosimeters Forward biased p-i-n diodes used as dosimeters

24. 2 major issues: 24 Radiation Monitoring at the LHC Experiments Radiation Monitoring at the LHC Experiments 2 types of RadFET: 250 nm oxide thickness → REM,UK 1600 nm oxide thickness → LAAS, France BPW34 Commercial silicon p-i-n diodes Measure of the 1-MeV Ф eq → 10 8 ≤ Ф eq ≤ 10 14 -10 15 n eq /cm 2 for LHC Measure of the TID

25. 25 p-i-n diodes (NIEL) Displacement damage in high  Si-base  Resistivity increases vs Ф eq FORWARD BIAS Fixed I F  V F  Ф eq V F =  (material parameters, geometry [W], readout current [J], pulse length) VFVF iFiF BPW34 p-i-n diode: Thickness ≈ 300 µm, Area = 2.65×2.65 mm 2, ρ ≈ 2.7 kΩ.cm

26. 26 Hadron sensitivity range from 2×10 12 to 4×10 14 n eq /cm 2. Perspectives for the future Super-LHC: Luminosity and radiation level (×10). Detectors will be exposed to fluences up to 10 16 1-MeV equivalent neutrons. A solution to measure very high fluences has to be found Readout protocol for LHC BPW34 diode FORWARD BIAS Fixed Readout Current I F  V F  Ф eq I F = 1 mA with a short duration pulse F. Ravotti et al., IEEE TNS, vol. 55, no. 4,pp. 2133-2140, 2008

27. 27 First study New readout protocol Different current steps of 50ms pulse duration Current used: 10µA – 100µA – 1mA – 5mA – 10mA – 15mA – 25mA Increase of bulk resistivity with Ф eq Thyristor - like behavior (F. Ravotti et al, IEEE TNS, vol. 55, no. 4, pp. 2016-2022, 2008.) Self-heating of the diode

28. 28 Detailed study of the detectors behavior Second Study Modifications of the electrical properties of the material Development of 2 tests benches for the detector characterization

29. 29 2 differents regimes can be distinguished: At low fluences: 1)At low voltages a linear region can be observed. 2)As V F increases: linear region → sharp increase of I F. Second study(1/2) I-V curves f rom very low voltages (=1mV), to high voltages. I-V curves f rom very low voltages (=1mV), to high voltages. Up to 6.26×10 15 n eq /cm 2 (60% of the expected Super-LHC fluences) Up to 6.26×10 15 n eq /cm 2 (60% of the expected Super-LHC fluences) First regime: Forward current (A)

30. 30 Rise of I F vs Ф eq increases up to ≈ 1 × 10 13 n eq /cm 2 Second regime: Second study (2/2) 1)For Ф eq > 1 × 10 13 n eq /cm 2, I-V characteristics are linear at low voltages. 2)With further increase of the radiation level, this linear behaviour extend to higher V F. Forward current (A)

31. 31 New formulation (1/3) This new formulation is based on the relaxation material theory This new formulation is based on the relaxation material theory Relaxation materials have a large density of g-r centers near E g /2. Relaxation materials have a large density of g-r centers near E g /2. Recombination pins the fermi level at minimum conductivity Recombination pins the fermi level at minimum conductivity Maximum resistivity: (see references in my PhD thesis) http://jmekki.web.cern.ch/jmekki/2009-11-27-Thesis-Mekki.pdf Forward current (A) Ф eq

32. 32 New formulation (2/3) Relaxation materials were experimentally fitted as : For I F > 1mA, possibility to have thyristor-like behavior 1 and/or self-heating effect. For I F > 1mA, possibility to have thyristor-like behavior 1 and/or self-heating effect. 1 F. Ravotti et al., IEEE TNS, vol. 55, no. 4,pp. 2133-2140, 2008 Forward current (A) I F ≥ 1mA I F ≤ 1mA FIT Ф eq = 6.3×10 14 n eq /cm 2 Ф eq = 6.3×10 15 n eq /cm 2 I F ≥ 1mA I F ≤ 1mA FIT

33. 33 New formulation (3/3) At the LHC experiments, BPW34FS diodes are operated in forward bias. A new formulation to predict and monitor values of V F versus Ф eq : For Ф eq ≥ 1×10 13 n eq /cm 2 For I F ≤ 1mA Based on: I F = 1 mA I F = 100 μA I F = 10 μA LambertW(x) function is the inverse function of:

34. 34 Qualitative evaluation of the temperature dependence Temperature Coefficient < 0 n i increases with T°, so ρ max decreases when T° increases.

35. 35 Conclusion of this study Effects on radiation damage up to 6.3×10 15 neq/cm 2 on the OSRAM BPW34FS silicon p-i-n diode have been studied. Comparison with relaxation materials. New formulation to predict V F versus Ф eq for: Ф eq ≥ 1×10 13 n eq /cm 2 I F ≤ 1mA Sensitivity is increased, and Ф eq measurement range can be expanded when diode is measured at lower temperature. Summary: Allow to extend the existing readout protocol. (I F = 1 mA) Permit to predict radiation response for expected SLHC fluences. Study published in J. Mekki et al, IEEE TNS, vol. 57, no. 4, pp. 2066-2073, 2010.

36. 36 Perspectives and outlook on future studies Perspectives and outlook on future studies

37. 37 Perspectives and outlook BPW34 p-i-n diode can be used for monitoring LHC and Super-LHC fluences from 2×10 12 n eq /cm 2. 2 possibility already exists: → Pre-irradiation allows to measure Ф eq > 8×10 9 n eq /cm 2. → CMRP diode (Thickness = 1 mm; Area = 1.2 mm 2, ρ ≈ 10 kΩ.cm): 1×10 8 < Ф eq (n eq /cm 2 ) < 2×10 12 With the intention to develop our specific dosimeter → An investigation on custom made devices (high resistivity silicon detector)

38. 38 Silicon Detectors Tested devices were made from n-type FZ and MCz silicon wafers. Geometry dependence on the detector’s radiation response has been evaluated. → 2 different active area: 2.5×2.5 cm 2 and 5×5 cm 2 → 2 different thicknesses: 300 µm and 1000 µm Outcome: The device thickness is the main parameter which influence their radiation response.

39. 39 Silicon Detectors Readout Current Detector A (300 µm) Detector B (1000 µm) 100 µA 9.1×10 9 cm -2 /mV3.2×10 8 cm -2 /mV 1 mA4.2×10 9 cm -2 /mV1.9×10 9 cm -2 /mV Sensitivity is increased by a factor ≈ 25 Thick detector Thin detector Study published in J. Mekki et al, IEEE TNS, vol. 57, no. 6, pp. 3483-3488, 2010.

40. 40 Silicon Detectors Readout Current Detector A (300 µm) Detector B (1000 µm) 100 µA 9.1×10 9 cm -2 /mV3.2×10 8 cm -2 /mV 1 mA4.2×10 9 cm -2 /mV1.9×10 9 cm -2 /mV Sensitivity is increased by a factor ≈ 25 Ф eq = 2×10 10 n eq /cm 2 Thick detector Ф eq = 2×10 12 n eq /cm 2 Thin detector Ф eq ≈ 8×10 12 n eq /cm 2 Thick detector Study published in J. Mekki et al, IEEE TNS, vol. 57, no. 6, pp. 3483-3488, 2010.

41. 41 General Conclusion Monitor the LHC radiation field: 2 devices → RadFET (TID) → p-i-n diodes (Ф eq ) RadFETs: Evaluation of packaging configurations Evaluation of the TID and package impact on a real LHC experiment. → Dose enhancement up to 25 % p-i-n diodes: New formulation for monitoring very high fluences (Super-LHC). At low temperature → expand to higher fluences Custom made devices : Sensitivity for low Ф eq can be improve using thicker p-i-n diodes or detectors.

42. Thank you for your attention 42 Thank you for your attention The Atlas Detector

43. 43

44. 44 Normal readout protocol: Wait for temperature stabilization inside the diode after each measurement: Outcome: Problem for measuring at high injection level due to self-heating. VF at IF = 100µA VF at IF = 10µA 50ms VF at IF = 10µA VF at IF = 1mA VF at IF = 25mA 50 ms Measurement VF 1 at IF = 10µA VF 2 at IF = 10µA After measurement VF 2 < VF 1 (self-heating) Wait intil VF 2 =VF 1 Self heating Self heating effect

45. 45 Summary of the relaxation materials theory (1/3) Relaxation theory occurs when the material has high resistivity, and contains defects due to impurities or damage which enhance the G-R rate. Definition of the dielectric relaxation time: Time to restore charge neutrality to a region when excess carrier are suddently introduced. When excess carriers are injected across the PN junction, at the instant of injection (t=0), there will be an excess charge (Δ n,p ), so that charge neutrality is disturbed. It is assumed to be the bulk equivalent of a RC time constant : τ D = ρεε 0

46. 46 Summary of the relaxation materials theory (2/3) Example: Injection of minority carriers in the n side (Δ p ): p(x) x p0p0 ΔpΔp n(x) x n0n0 Diffusion of holes (gradient of holes) At t = 0 → there are excess holes but no excess e - e - (Δ n ) are attracted in this region by drift because of the field induce by Δ p. p(x) x p0p0 ΔpΔp n(x) x n0n0 ΔnΔn Δ n flow in from the contact to neutralize Δ p This neutralization occurs in a dielectric relaxation time ( τ D ). While neutrality is quickly established, Δ p diffuse slowly and recombine with e - so that there is still excess charges in the material : The conventionnal carrier lifetime τ 0 Resistivity is decreased by the enhancement of carrier in the material. In conventionnal lifetime material, neutrality is restored before excess carrier recombine. τ 0 >> τ D The np product is equal to: np = n i 2 ×exp[( Ф n -Ф p )/kT]; Ф n and Ф p are the quasi-fermi levels for e - and h +, and is dependent on the applied voltage.(V = Ф n -Ф p )

47. 47 Summary of the relaxation materials theory (3/3) For irradiated diodes, the material becomes highly recombinative do to high density of recombination centers. Minority carrier injection increases the resistivity since the concentration of minority and majority carriers is reduced by recombination. τ D = ρεε 0 increases. τ D >> τ 0 Injected minority carrier lead to a depletion of majority carriers through the g-r centers activity. Therefore the carrier equilibrium is rapidly reached → no possible to influence it by externally applied voltage. Recombination pins the fermi level at minimum conductivity (defect near Eg/2) → np = ni 2 as for the steady-state condition in lifetime diode. Maxiumum resistivity of Silicon : Maxiumum resistivity of Silicon :