Robert Klanner - Univ. of Hamburg - VCI 2013 Vienna - Feb Study of High-dose X-ray Radiation Damage of Si-Sensors + AGIPD Project Status and Recent Developments 1 R. Klanner for the Consortium (Uni-Bonn, DESY, Uni-Hamburg, PSI) AGIPD = A daptive G ain I ntegrating P ixel D etector Work supported by: Table of Content 1. Challenges for X-ray Sensors at the European XFEL 2. The AGIPD Project 3. Study of X-ray Radiation Damage of (p + n) Si Sensors 4. AGIPD-Sensor Design 5. X-ray Effects on the AGIPD ASIC 6. Summary

Robert Klanner - Univ. of Hamburg - VCI 2013 Vienna - Feb Challenges for X-ray the E-XFEL 2 -European X-FEL under construction in Hamburg  completion end Pulse trains of (e.g.12 keV) photons of <100 fs duration and 220 ns spacing  Pixel sensors for imaging: 0, 1 … > keV photons per 200 x 200 μm 2 pixel and pulses/sec 10 8 Comparison of peak brilliances of X-ray sources 220 ns Unique European-XFEL features: No.pulses x pulse duration x coherence x Å resolution  Challenges: Radiation damage Plasma effect/charge explosion Charge losses Pile-up from preceding pulse

Robert Klanner - Univ. of Hamburg - VCI 2013 Vienna - Feb The AGIPD Detector Project: One Science Goal 3 Just one example: Coherent Diffractive Imaging using nano-crystals  Single shot imaging of complex bio-molecules using nano-crystals (single molecules ???) S.Boutet, et al., Science Express (2012) Doi: /science X-ray pixel sensor

Robert Klanner - Univ. of Hamburg - VCI 2013 Vienna - Feb The AGIPD Project: Building Blocks 4 p + n – sensor: 128x512 pixels (200μm) 2 – 107.6x28 mm 2 Bump bonded “hybrid detector” 128x512 pixels Single detector assembly 1 Mpixel detector assembly (16 single det. assemblies) ASIC: 64x64 pixels (IBM CMOS:130nm) U.Trunk D.Pennicard Seungyu Rah J.Zhang

Robert Klanner - Univ. of Hamburg - VCI 2013 Vienna - Feb The AGIPD Detector Project: Read-out Concept 5 -E γ = 3 – 20 keV -Dynamic range: 0, 1 to >10 4 (12.4 keV γ‘s) -Adaptive gain switching to 3 ranges -~ 350 stored images/pulse train -Trigger + Fast Clear Adaptive-gain read-out concept Performance has been demonstrated ~100 ~3500 ~2x keV photons D.Greiffenberg

Robert Klanner - Univ. of Hamburg - VCI 2013 Vienna - Feb AGIPD: Detector Specifications 6 –X-ray energy range: 3 - ~20 keV –Dynamic range N γ /pixel: 0,1 to >~ keV for <100 fs pulse length –Distance pulses: 220 ns –No. pulses: 2700/(100ms) –Recording of >350 images/(100ms) d=500μm V > 500 V preferably ~ 1000 V radiation tolerant up to 1 GGy (SiO 2 ) For 3 years oper. non-uniformly over sensor pixels of (200μm) 2 + additional specifications:

Robert Klanner - Univ. of Hamburg - VCI 2013 Vienna - Feb Si-Sensor X-Ray Radiation Damage 7 XFEL requirements: 1 GGy (SiO 2 ) for 3 years operation (non-uniform !) Few data on X-ray damage for high-ohmic structures for such high doses  Work at UHH (PhD-theses of T.Pöhlsen, J.Schwandt and J.Zhang): -Irradiate test structures from different vendors to extract “microscopic” and “macroscopic” parameters due to X-ray radiation damage -“Understand” impact of above parameters on sensor performance, via measurements on irradiated sensors and detailed TCAD simulations -Optimize sensor design using TCAD simulations -Order “optimized” sensors (Aug.2012  delivery Feb.2013) Effects of X-ray radiation damage for p + n sensors: -No bulk damage for E γ < 300 keV  “Surface” damage: Build-up of oxide charges and Si-SiO 2 interface traps  Accumulation layers form (or increase)  High field regions appear reducing the breakdown voltage  Leakage currents increase due to interface states  Depletion voltage and inter-pixel capacitances increase  Charge losses close to the Si-SiO 2 interface occur (increase)

Robert Klanner - Univ. of Hamburg - VCI 2013 Vienna - Feb Sensor Radiation Damage: Microscopic Defects D it TDRC: Properties of interface traps (Thermal Dielectric Relaxation Current) - Bias MOS-C in e-accumulation  fill interface traps with electrons - Cool to ~10 K  freeze e in traps - Bias to inversion and heat up to 290 K  I TDRC due to release of trapped e’s  I TDRC (T)  D it (E) *)  (Energy levels + widths + densities) it *) Temperature T  E c – E it (T dependence of Fermi level) 8 Test structures (4 vendors + diff. crystal orientations, oxide thickness, … ) ???) : MOS Capacitor MOS-C Gate Controlled Diode GCD Parameterized by 3 states Interpretation not unambiguous !

Robert Klanner - Univ. of Hamburg - VCI 2013 Vienna - Feb Sensor Radiation Damage: Microscopic Defects N ox 9 C/G-V curves for CMOS-C: -D it (E) allows calculation of C/G-V curves as function of frequency (assuming values for trap cross sections) -Oxide charge density N ox just shifts curves along the V-axis  N ox Si-SiO 2 interface Interface traps Fair description of a large amount of data For details and (some of the) experimental complications, see: J.Zhang et al., JSR19/3(2012)340 (presently we are extending model  include interaction with valence band  relevant for region of weak inversion at low frequencies)

Robert Klanner - Univ. of Hamburg - VCI 2013 Vienna - Feb Sensor Radiation Damage: Macroscopic Defects J surf 10 Surface current density from GCD: -Measure I-V curve -J surf dominated by mid-gap traps J surf [A] Measured T-dependence of I surf : E-field at Si-SiO 2 interface: shielded non zero Comments on J surf measurements : For high J surf voltage drop along surface  Si-SiO 2 interface only partially depleted Si-SiO 2 interface states  decrease of mobility We do not take into account these effects  Measured I surf = lower limit of surface current We cannot explain J surf using D it  More work needed ! For simulations of highly-irradiated sensors we use “high” values of J surf (is not too critical)

Robert Klanner - Univ. of Hamburg - VCI 2013 Vienna - Feb Sensor Radiation Damage: Summary N ox and J surf 11 Vendors: CiS, Hamamatsu, Canberra, Sintef; Crystal orientations:, ; Insulator: SiO 2 ( nm) - with and without additional 50 nm Si 3 N 4 -Results reproducible (after some annealing) -Spread of a factor ~2 -N ox saturates for ~1 – 10 MGy -J surf peaks at 1-10 MGy, then decreases - Equilibrium h-trapping and eh-recombination ? - E-field effects due to oxide charges ?  Understanding needs more studies X-ray radiation damage saturates !!! J.Zhang et al., arXiv: (2012)

Robert Klanner - Univ. of Hamburg - VCI 2013 Vienna - Feb Sensor Radiation Damage: Impact on sensor I dark 12 AC-coupled CIS sensor: Interface current dominates - Current from depleted interface (E-field) - Interface area changes with V bias  seen by X-ray users  minimize depleted interface area (minimize gap between implants/Al) SYNOPSYS-TCAD Simulation (Ajay Srivastava) Important for sensor optimization 0 Gy

Robert Klanner - Univ. of Hamburg - VCI 2013 Vienna - Feb Sensor Radiation Damage: Impact on Sensor V depl 13 Effects of N ox  increase no.of electrons in accumulation layer - Step in 1/C 2 when non-depleted regions below SiO 2 separate - Voltage required to deplete entire sensor depends on N ox J.Zhang No significant impact – however, good to know J.Schwandt

Robert Klanner - Univ. of Hamburg - VCI 2013 Vienna - Feb Sensor Radiation Damage: Impact on Sensor V bd 14 Simulations 2-dim [x,z and r,z] and 3-dim N ox  accumulation layer  changes curvature p + -depletion  changes E-field Breakdown (V bd ) depends on N ox,t ox,p + -implant, Al-overhang, potential on top of sensor (passivation layer), technology, etc. z [μ] Major challenge to reach V bd > 500 V after irradiation J.Schwandt

Robert Klanner - Univ. of Hamburg - VCI 2013 Vienna - Feb Sensor Radiation Damage: Impact on Sensor V bd 15 Example: I-V characteristics for diode with 12 guard-rings (Sintef – design not optimized for X-ray radiation hardness)  Importance to understand radiation effects ! J.Zhang

Robert Klanner - Univ. of Hamburg - VCI 2013 Vienna - Feb AGIPD Sensor Design: Optimization Strategy 16 J.Schwandt et al., arXiv: (2012) Discuss only guard ring optimization – the hardest !

Robert Klanner - Univ. of Hamburg - VCI 2013 Vienna - Feb AGIPD Sensor Design: Guard-Ring Optimization Strong dose dependence: (V bd ~400V for N ox <10 11 cm -2 ) 17 2-D (x,y) simulations (for 0 guard ring - GR): (~ 10 kGy) (~ 100 MGy) Sudden decrease inV bd : - Si below Al overhang gets depleted  voltage drop over larger region  E smaller for a given (high) N ox : V bd increases with ↓ d oxide and ↑ p + -implant depth For high radiation damage optimization is very different than for non-irradiated sensor. V bd ~ 70 V (0 GR) can be reached J.Schwandt

Robert Klanner - Univ. of Hamburg - VCI 2013 Vienna - Feb AGIPD Sensor Design: Guard-Ring Optimization 18 Optimize GR layout 1 gap (0 GR)  V bd ~70 V  for V bd ~ 1000 V need 16 gaps (15 GR) Optimize spacing, width implant, Al overhang for equal max. E-field and minimal space + Assure that depletion region does not touch cut edge (critical for low N ox !) Result: - Distance pixel to cut edge: 1.2 mm Optimized pixel and guard ring layout meets all specifications n + scribe line implant GDS printout: J.Schwandt + J.Zhang J.Schwandt et al., arXiv: (2012)

Robert Klanner - Univ. of Hamburg - VCI 2013 Vienna - Feb AGIPD ASIC: Specifications and Technology 19 Specifications: 64x64 pixel for (200 μm) 2 pixel (2x8 ASICs with 128x512 pixels per sensor without dead area) 352 storage cells per pixel – possibility of veto (overwrite) Droop rate < few % for 10 ms Rms noise 300 electrons (1/10 signal of 12 keV) Maximum signal to measure: ~ keV X-rays (3.5x10 7 electrons) 4.5 MHz frame rate Power: ~2W per ASIC Max. X-ray dose: <100 MGy (protected by sensor) Technology: IBM 130 nm CMOS process Specifications met for non-irradiated ASIC Bonn-DESY-PSI cali- bration input transistor pre-amp CDS output driver Single-Pixel Layout 352 storage cells below feed-back Cs

Robert Klanner - Univ. of Hamburg - VCI 2013 Vienna - Feb AGIPD ASIC X-ray Radiation Effects: Noise 20 Irradiation up to 100 MGy: 100 MGy: ASIC stopped working, but recovered after annealing (reason under study) 10 MGy: Minor increase in noise: 320  420 rms-e No increase in power Change in droop rate of storage cells (by sub-threshold currents)  Radiation performance appears ok rms noise versus X-ray dose Specifications also met for irradiated ASIC A.Marras

Robert Klanner - Univ. of Hamburg - VCI 2013 Vienna - Feb AGIPD ASIC X-ray Radiation Effects: Droop 21 Time charge storage to read out variable <10ms  avoid discharge of storage Cs (<1%) Problem: sub-threshold currents (field oxide) Solution: V dd -rings around critical tran- sistors   further reduction to < 1% by cooling! Droop vs. FET layout/Vdd-ring for 0, 10 kGy and 1 MGy Droop well below the Poissonian limit U.Trunk p P+P+ n-well P+P+

Robert Klanner - Univ. of Hamburg - VCI 2013 Vienna - Feb Summary 22 AGIPD is getting ready for science at the European-XFEL For the individual building blocks + for a prototype system, it has (essentially) been demonstrated that the challenging speci- fications can be met, in particular: “Huge” dynamic range Radiation tolerance Next: Production of full-scale system X-Ray radiation damage: Complex multi-parameter problem Progress made towards understanding effect up to GGy regime (but far from an in-depth understanding !) Compared to bulk damage little efforts in the detector community on the study of X-ray damage for sensors (and there have been surprises in the past !) Many thanks to VCI 2013 organizers, and to my AGIPD colleagues

Robert Klanner - Univ. of Hamburg - VCI 2013 Vienna - Feb The Work Force 23 DESY PSI Uni-HHUni-Bonn Julian Becker Roberto Dinapoli Eckhart FretwurstMarkus Gronewald Laura Bianco Doninic Greiffenberg Robert KlannerHans Krueger Peter Goettlicher Beat Henrich Joern Schwandt Heinz Graafsma Aldo Mozzanica Jiaguo Zhang Helmut Hirsemann Xintian Shi (+ help Ioana Pintilie Stefanie Jack Bernd Schmitt and Thomas Pöhlsen) Alexandre Klyuev Sabine Lange Alessandro Marras Bjoern Nilsson Seungyu Rah Igor Shevyakov Segrej Smoljanin Ulrich Trunk Manfred Zimmer E-XFEL Jolanta Sztuk-Dambietz

Robert Klanner - Univ. of Hamburg - VCI 2013 Vienna - Feb The End 24