1. 30 th March 2010 IoP HEPP/APP Annual Conference 2010 UCL Charge Collection Annealing in ATLAS SCT Silicon Sensors Craig Wiglesworth

2. The ATLAS Experiment and the SCT Craig Wiglesworth IoP HEPP/APP Annual Conference UCL March 2010 1 The ATLAS semiconductor tracker (SCT) is a silicon microstrip detector. Provides discrete space points for tracking and vertexing. Exposure to high levels of radiation damages sensors. Subsequent changes in sensor properties continue to change after irradiation. Important to understand SCT performance over full 10 year lifetime of ATLAS.

3. Evolution of Sensor Properties After Irradiation Evolution of V DEP shows dependence on both temperature and time spent after irradiation. Require full depletion of sensor in order to optimise CCE. I LEAK Shows strong temperature dependence but decreases with time. (Typical behaviour for a silicon microstrip sensor). Sensor leakage current (I LEAK ), depletion voltage (V DEP ) and charge collection efficiency (CCE) are all affected by irradiation. Craig Wiglesworth IoP HEPP/APP Annual Conference UCL March 2010 2 Need to avoid the sensors being warm (> 0 o C) for long periods of time! Low temperatures will suppress I LEAK but risk of thermal runaway (I LEAK not discussed further).

4. In the ATLAS Inner Detector Technical Design Report (1997) V DEP and I LEAK were predicted for SCT sensors. Assumed LHC luminosity profile: 3 years @ 10 33 cm -2 s -1 7 years @ 10 34 cm -2 s -1 (~ 1.4x10 14 n eq cm -2 ) ATLAS maintenance scenarios (days@temp yr -1 ): Standard Access Procedure (SAP) Many of the inputs to these calculations have since changed. Evaluating the Evolution of V DEP and I LEAK in the SCT In addition, the radiation damage model itself has evolved. Predictions re-evaluated with new radiation damage model and updated inputs. Craig Wiglesworth IoP HEPP/APP Annual Conference UCL March 2010 3

5. Re-Evaluating the Evolution of V DEP and I LEAK in the SCT Paul Dervan, Joost Vossebeld, Tim Jones (Liverpool), Taka Kondo (KEK), Graham Beck (QMUL), Georg Viehhauser (Oxford), Steve McMahon (RAL), Koichi Nagai (Brookhaven), Kirill Egorov (Indiana), Richard Bates, Alexander Bitadze (Glasgow). An updated LHC luminosity profile now exists. Maintenance/shutdown time and cooling temperatures reviewed in line with: Achievable coolant temperatures. Insertable B-Layer installation. Possibly longer maintenance. V DEP predictions suggest 450 V (max for SCT) is sufficient for at least 10 years operation. Craig Wiglesworth IoP HEPP/APP Annual Conference UCL March 2010 4

6. Hamburg model is now believed to be the best model available to predict V DEP. However, large differences observed between the predictions of the TDR model and the Hamburg model. Origin is in reverse annealing contribution  N Y to the predicted change in effective doping concentration  N EFF : Comparison of Hamburg Model and TDR Model Craig Wiglesworth IoP HEPP/APP Annual Conference UCL March 2010 5 TDR model parameterised reverse annealing as a second order process. Hamburg model parameterises reverse annealing as a modified first order process. Need high fluence + long annealing data to compare to predictions of both models.

7. Sensor performance traditionally studied by determining V DEP from CV measurements. In Liverpool the focus has been on measuring the annealing of CCE. Much data available for n-side readout sensors. Less detailed information for p-in-n sensors. ManufacturerHPK Wafer Tech.FZ Structurep-in-n Size1 cm x 1 cm Thickness285 µm Programme of Accelerated Annealing Measurements New programme of accelerated annealing measurements on ATLAS mini sensors. Pair of sensors irradiated with neutrons at Ljublijana (V. Cindro et al) to 2x10 14 n eq cm -2 (new prediction for SCT = 1.6x10 14 n eq cm -2 ). Craig Wiglesworth IoP HEPP/APP Annual Conference UCL March 2010 6 One sensor used for CCE measurements and one sensor used for V DEP measurements. Both sensors annealed together at same temperature for same length of time. All following plots by…. A. Affolder, H. Brown, G. Casse, P. Dervan, J. Vossebeld, C. Wiglesworth (Liverpool)

8. Charge Collection Measurements 90 Sr fast electron source used to generate signal. Readout triggered by scintillator. Charge collection measured using analogue electronics chip (SCT128) clocked at LHC speed (40 MHz clock, 25 ns shaping time). System calibrated to most probable value of MIP energy loss in non-irradiated 300  m thick sensor (~ 23000 e). Craig Wiglesworth IoP HEPP/APP Annual Conference UCL March 2010 7 Experimental Setup and Analysis Procedure Depletion Voltage Measurements V DEP Determined by standard method of measuring the V BIAS at which 1/C 2 saturates. Both sets of measurements performed in freezer at temperature of ~25 o C with N 2 flush. @ 1 kHz

9. Experimental Setup and Analysis Procedure Craig Wiglesworth IoP HEPP/APP Annual Conference UCL March 2010 8 Strip Number Noise (ADCs) Cluster Frequency Cluster Charge (ADCs) (1 ADC ~ 22 e) A landau convoluted with a gaussian is fitted to the resulting distribution of the collected charge. Most probable value is recorded. Hot channels are masked out in analysis Channels with low hit frequency are masked out Measure charge on strip (+ neighbours) with largest S/N ratio

10. Smooth fall off after ~ 100 days @ 20 o C Results of Charge Collection Measurements Craig Wiglesworth IoP HEPP/APP Annual Conference UCL March 2010 9 SCT Default threshold 1 fC (~ 6.2 ke) No sharp drop in the collected charge

11. Results of Charge Collection Measurements SCT Default threshold 1 fC (~ 6.2 ke) Still collecting significant charge after 1000 days @ 20 o C Craig Wiglesworth IoP HEPP/APP Annual Conference UCL March 2010 10

12. Results of Charge Collection Measurements Craig Wiglesworth IoP HEPP/APP Annual Conference UCL March 2010 11 Above ~ 500 V collected charge remains almost constant in time Largest increase in charge collection ratio @ 300 V

13. Comparison of Measured V DEP and Charge Collected Craig Wiglesworth IoP HEPP/APP Annual Conference UCL March 2010 12

14. Comparison of Predicted V DEP and Measured V DEP Data shows a slower annealing effect than the two models Data looks closer to Hamburg model Craig Wiglesworth IoP HEPP/APP Annual Conference UCL March 2010 13

15. Conclusions and Future Plans A programme for measuring the annealing of CCE in p-in-n microstrip sensors has started. No sharp drop in collected signal when sensor becomes under-depleted. Slow drop in collected signal with annealing time. Significant charge collected even when V BIAS is well below the V DEP predicted by the models. For ATLAS SCT this means operational range defined by V DEP prediction is pessimistic. Next…. Repeat measurements on pair of ATLAS mini sensors irradiated to 3x10 14 n eq cm -2. Make more detailed comparisons between results and predictions of the two models. (+ Perhaps an attempt at an updated model). Craig Wiglesworth IoP HEPP/APP Annual Conference UCL March 2010 14