K. Gill, G. Cervelli, R. Grabit, F. Jensen, and F. Vasey. CERN, Geneva Radiation damage and annealing in 1310nm InGaAsP/InP lasers for the CMS Tracker K. Gill, G. Cervelli, R. Grabit, F. Jensen, and F. Vasey. CERN, Geneva August 2000
Background CMS Tracker readout and control project Complex system with >50000 optical links Harsh radiation environment Extensive use of commercial off-the-shelf components (COTS) Part-of series of on-going validation tests required before components integrated into final system Previous tests reported at SPIE and RADECS 97- 99
CMS Tracker optical link technology lasers single-mode fibre + array connectors photodiodes Transmitter - 1310nm InGaAsP edge-emitter Fibres and connectors - single-mode Ge-doped fibre Receivers - InGaAs p-i-n photodiode Electronics - rad-hardened 0.25mm in radiation zones COTS issues: radiation damage: up to 1014particles/cm2 + 100 kGy reliability: 10 year lifetime in radiation environment Add figure re-emphasise widespread use in LHC Tx, fibres, connectors exposed in readout links all types of element exposed in control links
CMS Experiment
CMS Tracker radiation environment Numbers come from: Show Mika’s plots (Ref) show fig for other environments, put LHC in context space: - optocouplers - data bus (lightweight, low power fibre optics) nuclear: - sensors, fibroscopy - diagnostics charged hadrons (p, p, K) (courtesy M. Huhtinen, CERN)
CMS Tracker readout and control links Analogue Readout 50000 links @ 40MS/s FED Detector Hybrid Tx Hybrid 96 Rx Hybrid processing MUX A APV 4 buffering 2:1 amplifiers D DAQ 12 12 pipelines C 128:1 MUX Timing PLL Delay DCU TTCRx TTC Digital Control 2000 links @40MHz FEC Control 4 64 TTCRx CCU CCU 8 processing buffering CCU CCU Front-End Back-End
System specifications Analogue readout links Last 2 columns filled in for each device type after testing
Objectives Compare damage from different particles 0.8MeV n and 6MeV n, 330MeV p, 24GeV p, 60Co g Measure annealing characteristics Temperature and current dependence Make prediction for damage expected in CMS tracker 10 years at -10°C, including LHC luminosity profile
Experiment Devices Italtel/NEC 1310nm edge-emitting InGaAsP/InP MQW lasers mounted on Si-submounts compact mini-DIL packages, single-mode fiber pigtails no other components in the package, e.g. lenses Pre-irradiation characteristics at 20°C : Laser threshold currents 8-13mA Output efficiencies (out of the fibre) 30-70mW/mA This type of device previously studied 6MeV n, 330MeV p, 24GeV p, 60Co g
DCPBH-MQW lasers double-channel-planar-buried-heterostructure laser Emphasise similarities and differences with LED’s
Test Procedures Test A: Irradiate 0.8MeV n - compare damage with other particles 4 lasers, irrad room T, biased 5-10mA above threshold, 1015n/cm2 in 6.5 hrs. Anneal at room T, biased 5-10mA above threshold for 115 hrs Test B: Irradiate 0.8MeV n - anneal at different T 12 lasers, cooled -13°C, unbiased, 1015n/cm2 in 6.3 hrs. Anneal in groups of 3 at 20, 40, 60, 80°C for 300 hrs. Test C: Irradiate 0.8MeV n - anneal at different bias currents 8 lasers, irrad room T, unbiased, 1015n/cm2 in 6.5 hrs. Anneal in groups of 2 at 0, 40, 60, 80mA bias for 115 hrs.
Test setup for in-situ measurement of radiation damage and annealing
Test A - 0.8MeV irradiation at room T Damage approximately linear with fluence
Test A - Comparison with other particles Data averaged over devices then normalized to 96 hour irradiation with 5x1014particles/cm2. Relative damage factors for 0.8MeV n with respect to ~6MeVn (1/3.1), 330MeV p (1/11.4), 24GeV p (1/8.4).
Test B - cooled irradiation Irradiation fluence 1015 (0.8MeV n)/cm2 Test made at -10°C, then devices stored at -35°C
Test B - Annealing versus temperature Devices split into 4 groups of 3 and annealing at different temperatures. Threshold damage assumed to be proportional to number of defects Annealing generally linear with log (time)
Test C - Annealing versus current Irradiation to 1015n/cm2 at room T, unbiased, then anneal in 4 groups of 2 at different bias currents Enhancement caused by: (i) ‘recombination enhanced annealing’ (?) - supposed to be unlikely in InGaAsP/InP (ii) thermal acceleration due to power dissipation. At 80mA DTjunction ~ 8C. Up to factor 10 enhancement in terms of annealing time
Annealing model Assume 1st order (exponential) annealing obeying Arrhenius law: remaining fraction of defects: where For defects with a uniform distribution of activation energies r = N/(tmax-tmin), the annealing is linear with log (time)
Activation energy spectrum Data points for each group of 3 devices averaged. Fit annealing model to Test B data. Activation energy spectrum for best fit is 0.66<Ea<1.76eV
Damage prediction in 1yr in CMS tracker Use model to predict annealing of defects at -10°C over 1 LHC year LHC/CMS running LHC shutdown damage + annealing annealing 32% of total defects introduced during 1 year are annealed
Damage prediction 10yrs in CMS tracker Extend to 10 years, taking into account LHC luminosity profile Based on damage of 0.8MeV n at -10C (Test B) and relative damage factors (Test A), possible to estimate damage to laser threshold in CMS Tracker: in worst case, at low radii (and no bias-enhancement included), DIthr = 14mA
Conclusions ‘Calibrated’ damage from 0.8MeV neutrons relative to 6MeV n, 330MeV p, 24GeV p Determined annealing dependence temperature and forward bias current Constructed a model to describe the annealing v T uniform distribution of activation energy 0.66<Ea<1.76eV Based on data, applied model to CMS Tracker to predict laser threshold damage In the worst case, at low radii: DIthr = 14mA Further work: extension of the study to include lasers from other manufacturers.