DEVELOPMENT OF RADIATION HARD CZOCHRALSKI SILICON PARTICLE DETECTORS

Transcription

1 DEVELOPMENT OF RADIATION HARD CZOCHRALSKI SILICON PARTICLE DETECTORS Helsinki Institute of Physics, CERN/EP, Switzerland Microelectronics Centre, Helsinki University of Technology, Finland Okmetic Ltd., Finland Brookhaven National Laboratory, USA CERN RD39 & RD50 Accelerator Laboratory, University of Jyväskylä, Finland Ioffe PTI, Russia Electron Physics Laboratory, Helsinki University of Technology, Finland

2 OUTLINE -Motivation and background -What is the radiation hardness *Different particle radiations *Microscopic effects *Macroscopic effects -High resistivity Cz-Si -Cryogenic operation.solution for future HEP experiments? -Summary

3 BACKGROUND LHC is the first experiment to use Si detectors in large scale. Only very few proton-proton collisions produce Higgs. Luminosity of LHC beam is very high causing hostile radiation environment to Si devices. Radiation Hardness of Si devices is currently extensively studied topic. There are ~100 institutes within CERN s RD50 and RD39.

4 Different defects by different radiations RADIATION DAMAGE IN SILICON: microscopic effects. Point defects ( E > 25eV ) Cluster defects ( E > 5keV) M. Hutinen (ROSE TN 2001/02) Jyvaskyla proton beam > 4 X more Vacancies!! Ratio point/cluster defects depend on particle/energy 60Co-γ, E γ ~ 1MeV Displacement no clusters Electrons: E e > 255 kev displacement E e > 8 MeV clusters Neutrons: E n > 185 kev displacement E n > 35 kev clusters

5

6 WHAT IS THE RADIATION HARDNESS? In irradiation the particles create electrically active defects into silicon. These defects change the effective doping concentration fo Si Vdepl (V) [300µm] Point of n > p type inversion Breakdown... Typical depletion voltage evolution in traditional Fz-Si detector Every pn-junction device has a finite breakdown voltage... Major problem in HEP experiments!! 0 0 0,5 1 1,5 2 2,5 3 3,5 Dose Φ eq [10 14 cm -2 ] Adding oxygen into the Si lattice has proved to improve radiation hardness!

7 WHY CZ-Si AS A DETECTOR SUBSTRATE 1. Radiation hardness * Oxygen increases the radiation hardness of silicon detectors * Cz-Si intrinsically contains oxygen, cm Cost-effectiveness * Cz-Si wafers are cheaper than traditional Fz-Si wafers * Large area wafers available -> possibility for large detectors -> cost-effectiveness for frontend electronics, interconnection and module assembly There are reports of Cz-Si of resistivity 5kΩcm, "High resistivity CZ silicon for RF applications substituting GaAs, T.Abe and W.Qu, Electrochem.Society Proc. Vol

8 WHY CZ-Si AS A DETECTOR SUBSTRATE II 3. High oxygen concentration allows some additional benefits * Depletion voltage of detectors can be tailored by adjusting a) oxygen concentration in the bulk b) thermal history of wafers (Thermal Donor killing) * Possibility for internal gettering * Higher mechanical strengh * Less prone to slip defect formation WHY NOT BEFORE? * No demand for high resistivity Cz-Si -> No availability * Price for custom specified ingot 15,000-20,000 * Now RF-IC industry shows intrest on high resistivity Cz-Si (=lower substrate losses of RF-signal)

9 DEVICE PROCESSING The devices were processed at Helsinki University of Technology Microelectronics Center with simple 5 level mask process: 4 lithographies 2 ion implantations 2 thermal dry oxidations 3 sputter metal depositions on substrate grown by magnetic Czochralski nominal resistivity 900 Ωcm, thickness 380 um, orientation <100>, oxygen concentration <10 ppma Al WNx SiO2 p+ p+ p+ high resistivity n-type substrate

10 Pad detector Corner of Helsinki pad detector. -Multi-guard ring structure (16µm) -Wide guard ring (100µm) -Detector s active area. The distance between the active area implant and the first guard ring is 10µm.

11 Radiation hardness of Cz-Si FZ, E=10 MeV FZ, E=20 MeV FZ, E=30 MeV DOFZ, E=10 MeV DOFZ, E=20 MeV V dep (V) MCZ, E=10 MeV MCZ, E=20 MeV MCZ, E=30 MeV Φ eq (1 MeV equivalent neutrons/cm 2 )

12 RD50 - Development of Radiation Hard Semiconductor Devices for High Luminosity Colliders 280 Members from 55 Institutes 45 European and Asian institutes (34 west, 11 east) Belgium (Louvain), Czech Republic (Prague (2x)), Finland (Helsinki (2x), Laappeenranta), Germany (Berlin, Dortmund, Erfurt, Halle, Hamburg, Karlsruhe), Greece (Athens), Italy (Bari, Bologna, Florence, Milano, Modena, Padova, Perugia, Pisa, Torino, Trento, Triest), Lithuania (Vilnius), Norway (Oslo (2x)), Poland (Warsaw), Romania (Bucharest (2x)), Russia (Moscow (2x), St.Petersburg), Slovenia (Ljubljana), Spain (Barcelona, Valencia), Sweden (Lund) Switzerland (CERN, PSI), Ukraine (Kiev), United Kingdom (Exeter, Glasgow, Lancaster, Liverpool, London, Sheffield, University of Surrey) 6 North-American institutes Canada (Montreal), USA (Fermilab, New Mexico University, Purdue University, Rutgers University, Syracuse University, BNL) 1 Middle East institute Israel (Tel Aviv) Detailed member list: M.Moll 8/2003

13 CERN RD39 Collaboration * 66 members from 19 institutes Device Physics for radiation-hard silicon detectors Basic Research of heavily irradiated silicon Edgeless silicon detectors Cryogenic Module design, assembly and tests

14 SUPER-LHC * Luminosity cm -2 s -1 leading to fluence cm -2 Full Depletion Voltage Cz-Si Fz-Si * V fd will be very high * I leak will be very high Charge loss due to trapping will be severe 0 1,0E+13 1,0E+14 1,0E+15 1,0E+16 1 MeV eq. Fluence (cm^-2) * 10 years in LHC

15 Charge Collection Efficiency in S-LHC CCE = CCE GF CCE t = w d e t dr / τ t Trapping term CCE GF is a geometrical factor CCE t is related with trapping Depletion term 2εε V w w = 0 and = en d V eff V fd For fluence less than n/cm 2, the trapping term CCE t is insignificant For fluence n/cm 2, the trapping term CCE t is a limiting factor of detector operation!

16 1 τ t σv th N t TRAPPING = The trapping time-constant is not dependent on T The thermal velocity v th saturates at 20 kv/cm E-field to 10 7 cm/s cm -2 irradiation produces N T 3-5*10 13 cm -3 with σ cm 2 Particle generated charge carrier drifts 20-30µm before it gets trapped regardless whether the detector is fully depleted or not! In S-LHC conditions, 80-90% of the volume of d=300µm detector is dead space!

17 HOW A TRAP CAN BE NEUTRALIZED? -By filling them with current/charge or light injection. Q (arb. units) T(K) Qcol Qpol CID (Charge Injected Detector, irradiated with 3,8*10 15 n/cm 2 p + np + symmetric structure irradiated to cm 2 TCT measurement made with IR light pulses simulating MIPs Bias potential 300 V (Q col ) or 0 V (Q pol ; the electric field is induced by the polarization of the detector) Electric field is x 1/2 The p + np + detector with this fluence can be operated already at 100 V bias Standard p + nn + detector would deplete fully only above 1 kv bias at this fluence

18 DETRAPPING τ d = σ v th 1 N e C E t / kt If a trap is filled (electrically nonactive) the detrapping time-constant is crucial The detrapping time-constant depends exponentially on T For A-center (O-V at E c ev with σ cm 2 ) T(K) τ d 10ps 10ns 10µs 6ms 12.3s 5min 3,6 h 15 days 13 years

19 SUMMARY If a trap level is filled (say, by current or charge injection) and then frozen (very long detrapping time) at cryogenic temperatures, this trap level will no longer be able to trap free carriers again, and it becomes electrically inactive. In this case, the CCE t can be improved as well to a value close to 1 CCE GF can be increased close to 1 by manipulating the electric field in the detector via current and/or charge injection at temperatures from 130 K to 150 K. Feasible solution for very high luminosity colliders?

20 SUMMARY LOW T OPERATION At low temperature: No leakage current >> low electrical power from HV supply Low depletion voltage (original Lazarus effect) CCE increase without reduction of detector thickness (increase of charge collection depth) Readout electronics becomes faster and has lower noise

21 SUMMARY OF DEFECT ENGINEERING We have demonstrated first full-sized particle detectors ever made of Cz-Si Cz-Si shows superior radiation hardness properties in proton beams Cz-Si shows positive space charge built-up due to the Thermal Donor TD formation This gives possibility to further improvements in terms on compensation of SC of different sign