Impact of high photon densities on AGIPD requirements

Transcription

1 Impact of high photon densities on AGIPD requirements Julian Becker University of Hamburg Detector Laboratory new data 1. Heating estimations 2. Confined breakdown 3. Range switching in adjacent pixels 4. Measurements on charge collection time 5. Measurements of the PSF 6. Baseline sensor design 7. Summary Informal AGIPD meeting, , Hamburg Julian Becker Uni-Hamburg 1/14

2 Instantaneous local heating Basic estimation of local temperature increase due to photon absorption, neglecting heat conductance T = N V γ ρ E c 2.5mK γ = Si kev γ E γ = 12 kev V = 40x40x300 µm 3 To reach melting temperature > 5*10 11 photons are needed ρ c Si = 2.33 = 0.7 g cm J gk 3 Informal AGIPD meeting, , Hamburg Julian Becker Uni-Hamburg 2/14

3 Total heat load Heat sources: photon absorption, photocurrent, leakage current P P P γ abs, bunchtrain photo, bunchtrain leak = UI = U leak = = UI en N γ γ E E = U f 3.6eV v γ γ f 2 d drift = 10 dq dt 9 = U 9.6W 12 kev = U en γ N γ E 3.6eV T drift rep 67mW N keV γ 1W P total < Nγ + UI keV γ γ f γ t γ f v P P 5 drift, sat P leak Mhz = 1x10 bunchtrain = 6x10 T bunchtrain (1Mpix,1000V,10 9 γ)=2mk = = 3 t P 5 m / s bunchtrain bunchtrain f XFEL Informal AGIPD meeting, , Hamburg Julian Becker Uni-Hamburg 3/14

4 Local temperature increase Estimation of local temperature increase during bunch train assuming 10 6 photons in a single pixel for all 3000 frames and neighboring pixels at constant temperature (heat conductance only, no additional cooling) P Q& P 10 6 γ, total total = γ, bunchtrain T 10 80mW 200x500 (4 ) µm λ T 200 = Q& K 12 kev N γ γ λ Q& Q& Si neighbor total = 148W m* K A = λ T l = 4Q& neighbor Informal AGIPD meeting, , Hamburg Julian Becker Uni-Hamburg 4/14

5 Confined breakdown 7mA 1mA 8mA 450µm sensor 7mA linear scaling between last data points 280µm sensor 450µm sensor 450µm sensor Current peaks of up to 8mA have been observed in measurements so far, assuming linear scaling to 10 6 and 50 Ω impedance of the preamplifier 4V across the input of preamp killing threshold will probably be in the region photons/pixel Input protection will add noise to the measurements Discussion of trade off between noise and dead pixel count should be encouraged Informal AGIPD meeting, , Hamburg Julian Becker Uni-Hamburg 5/14

6 Range switching in adjacent pixels Current/charge pulse on adjacent pixel assuming continuous photon absorption, n-in-n pixels, no plasma effects and 500V bias. Range switching will be triggered if peak of red curve > 85 photons. Nγ > 2600 will trigger switching to second gain stage in adjacent pixels Nγ > 6.8x10 4 will trigger switching to third gain stage in adjacent pixels Having a primary pixel in third gain stage will result in switching in adjacent pixels even if less than 85 photons are absorbed Informal AGIPD meeting, , Hamburg Julian Becker Uni-Hamburg 6/14

7 Measurements on PSI strip sensor injection of 660 nm and 1015 nm light from rear side pn junction on front side (low field at injection side) position scan with spotsize ~3 µm Front electron hole charge carrier drift path structure (strip) readout PSI strip sensor <111> orientation thickness 450 µm U dep ~ 150 V pitch 50 µm width 11 µm attenuation length 3 µm (zone of injection) rear side bias 660 nm light fixed intensity x position Informal AGIPD meeting, , Hamburg Julian Becker Uni-Hamburg 7/14

8 Measurements on charge collection time practical cc limit recommended cc limit Zoom tightly focused (3 µm) soft x-rays (~1 kev) on 450 µm p-in-n strip sensor plasma delays for low bias voltages included => charge collection time stays below 60 ns even for very high Nγ if sufficient voltage is applied Informal AGIPD meeting, , Hamburg Julian Becker Uni-Hamburg 8/14

9 Measurements of the PSF tightly focused (3 µm) soft x-rays (~1 kev) new parameterization: Gaussian (diffusion) circle (plasma) => only of minor influence on imaging performance as plasma effects stay mostly confined in the pixels Informal AGIPD meeting, , Hamburg Julian Becker Uni-Hamburg 9/14

10 Measurements on charge collection time practical cc limit recommended cc limit determination algorithm stuck at electronic noise tightly focused (3 µm), hard x-rays (~12 kev) on 450 µm p-in-n sensor strongly dependent on threshold level (exponential pulse shape!) Informal AGIPD meeting, , Hamburg Julian Becker Uni-Hamburg 10/14

11 Measurements on charge collection time current 1.68 x kev photons charge 60ns 100ns long tails may be an issue, peak increasing with voltage no significant effects on current frame tail may spill over to next frame (0.1% of 10 5 photons is still 100 photons!) Informal AGIPD meeting, , Hamburg Julian Becker Uni-Hamburg 11/14

12 Measurements of the PSF tightly focused (3 µm) hard x-rays (~12 kev) new parameterization: Gaussian (diffusion) circle (plasma) => only of minor influence on imaging performance up to 1x10 4 photons as plasma effects stay mostly confined in the pixels Informal AGIPD meeting, , Hamburg Julian Becker Uni-Hamburg 12/14

13 Choice of technology n-in-n Baseline sensor design superior performance in high density case edges on ground Choice of bias voltage reasonably high (500 V? depends on resistivity, i.e, U dep ) less cumbersome than 1000 V (wire bonds etc.) charge collection time can be tuned to ASIC needs less peak voltage for the ASIC to stand Choice of temperature low (-20 C?) faster charge collection time (-13% for 20 C to -2 0 C) about 5% increase in spread for high voltages (20 C to -20 C) input from mechanical design welcome input from ASIC designers welcome Informal AGIPD meeting, , Hamburg Julian Becker Uni-Hamburg 13/14

14 Summary Heating (local, global) not a problem if leakage current is low Pixel killing threshold probably around photons/pixel Additional input protection needed? Range switching will occur in adjacent pixels if primary pixel is in third gain stage Bright pixels have long tails (no problem for soft x-rays) Memory effects Effective non-linearity under investigation next 500 V seems sufficient bias to prohibit plasma effects from deterioration the imaging performance Informal AGIPD meeting, , Hamburg Julian Becker Uni-Hamburg 14/14

15 Backup Informal AGIPD meeting, , Hamburg Julian Becker Uni-Hamburg 15/14

16 Measurements on charge collection time current 90x10 6 e,h pairs -> 3.24x kev photons charge no exponential shape! 60ns/100ns Informal AGIPD meeting, , Hamburg Julian Becker Uni-Hamburg 16/14

17 X-Ray Free Electron Laser (XFEL) 17.5 GeV linear electron accelerator producing 0.1 nm light (12.4 kev) through FEL process unprecedented peak brilliance of 5x10 33 (photons/s/mm 2 /mrad 2 /0,1% bandwidth) user facility: common infrastructure shared by many experiments pulse length < 100 fs bunch repetition rate 200 ns Informal AGIPD meeting, , Hamburg Julian Becker Uni-Hamburg 17/14

18 Introduction For the experiments at the European XFEL new detectors have to be build. The AGIPD consortium is building a detector using a hybrid pixel array and commonly available silicon. Under investigation by University of Hamburg Radiation damage due to very high intensities (up to γ/bunch -> 10 5 γ/pixel) -> integrated dose up to 1 GGy Instantaneous charge deposition -> charge explosion first talk this session this talk The multi channel Transient Current Technique (TCT) setup presented here allows studies of the charge explosion in a controlled laboratory environment by recording current pulses. Informal AGIPD meeting, , Hamburg Julian Becker Uni-Hamburg 18/14

19 - + TCT Principles peak due to holes recombining no plasma effects 1 kev γ at some XFEL experiments soft X-rays are used (SASE 3) - + electrons and holes drifting electrons drifting holes drifting 12 kev γ main XFEL X-ray energy (SASE 1) electron hole charge carrier drift path XFEL photons can be replaced by laser light with identical attenuation length 1 kev => 3 µm => 660 nm 12 kev=> 250 µm => 1015 nm Informal AGIPD meeting, , Hamburg Julian Becker Uni-Hamburg 19/14

20 TCT Principles Plasma - + When many photons are absorbed: charge carrier density > bulk doping O(10 12 cm -3 ) -> e,h plasmas are created resulting in: many 1 kev γ s charge carrier influence on electric field no longer negligible, dominating effect for high densities electron hole E undistorted d possible distortion field free regions inside the plasma -> ambipolar diffusion dominant plasma dissolved by diffusive processes -> time consuming -> effect on pulse shape -> charge carrier interactions (repulsion) increases spread of charge carriers at densities O(10 20 cm -3 ) (not reached at XFEL) electron hole recombination will occur Informal AGIPD meeting, , Hamburg Julian Becker Uni-Hamburg 20/14

21 Charge explosion Due to plasma effects lateral charge carrier spread is increased Effects on imaging performance Point spread function (PSF) becomes broader Increased charge sharing between adjacent pixels AGIPD pixel size beneficial if center of gravity algorithms can be used (low spatial frequency) disadvantageous if spreading of image points overlaps (high spatial frequency) masking of weak signals neighboring bright pixels => The PSF has to be known! Informal AGIPD meeting, , Hamburg Julian Becker Uni-Hamburg 21/14

22 Setup available for measurements Main goals: Determination of the pulse shape of individual pixels with XFEL type irradiation Agreement of experimental reference data with simulations (WIAS-Berlin) laser driver laser optics translation stages residual light control cooling 2.5 GHz scope attenuators, amplifiers DAQ and control PC * 660 nm -> 3 µm absorption length ~ 1 kev γ 1015 nm -> 250 µm absorption length ~ 12 kev γ 1052 nm -> 900 µm absorption length ~ mips Key features: laser pulses 100 ps FWHM duration up to kev γ equivalent 3 µm spot size 100 µm focus depth 660nm, 1015nm, 1052nm wavelength* multi GHz bandwidth electronics (<100 ps risetime) 0.1 µm position control 32 channels (4 simultaneously) temperature control (-35 C to 50 C) front and backside injection possible Informal AGIPD meeting, , Hamburg Julian Becker Uni-Hamburg 22/14

23 Measurements on planar sensors ~ 0.5 µm ~ 0.5 µm E not to scale not to scale front side and rear side injection of 660 nm and 1015 nm light (~ 1 kev/12 kev γ) rear side bias, U dep ~ 45 V Informal AGIPD meeting, , Hamburg Julian Becker Uni-Hamburg 23/14

24 Measurements on planar sensors Injection with 660 nm light (1keV γ) spot size ~ 2.5 µm scaled to same integral (charge) 100 V bias significant increase in charge collection time (red curve, rear side injection) less distortion for front side injection (blue curve) similar but less pronounced behavior for 1015 nm light (12 kev γ) => Junction-side injection favored! rear side injection front side injection 660 nm -> 1 kev γ impedance mismatch at amplifier impedance mismatch at amplifier Informal AGIPD meeting, , Hamburg Julian Becker Uni-Hamburg 24/14

25 Measurements on strip sensors injection of 660 nm and 1015 nm light from rear side pn junction on front side (low field at injection side) position scan with spotsize ~3 µm Front electron hole charge carrier drift path structure (strip) readout strip sensor same wafer as pad diode thickness 280 µm U dep ~ 63 V pitch 80 µm width 20 µm attenuation length 3 µm (zone of injection) rear side bias 660 nm light fixed intensity x position Informal AGIPD meeting, , Hamburg Julian Becker Uni-Hamburg 25/14

26 Results of transient evaluation investigated intensity (absorbed γ equivalent) charge collection time at 100 V/500 V applied bias expected collection time (no plasma) 100 V/500 V plasma delay (@100 V) peak current (<1 ns) 500 V sigma of PSF 100 V/500 V 660nm (1keV γ) ns / 11 ns 20 ns / 5.5 ns 15 ns 0.24 ma 55 µm / 28 µm All measured quantities show a strong dependence on the applied bias. 1015nm (12keV γ) >200 ns / 22 ns 20 ns / 5.5 ns none 8 ma 54 µm / ~25 µm Investigated in a 280 µm strip detector -> no guaranty for behavior in a 500 µm detector Sigma of PSF (660 nm -> 1 kev γ) number of photons time [ns] Informal AGIPD meeting, , Hamburg Julian Becker Uni-Hamburg 26/14 current plasma delay (660 nm)

27 Impact on imaging performance Two points can be separated if the distance between them exceeds the distance calculated from the Rayleigh criterion d 1 d 2 This distance is a function of the width of the PSF The minimal angular resolution achievable can be calculated from this distance The minimal angular resolution achievable is a function of intensity and bias voltage! Two Gaussian distributions with a sigma of 25 µm being d 1 =100 µm and d 2 =50 µm apart. Informal AGIPD meeting, , Hamburg Julian Becker Uni-Hamburg 27/14

28 Minimal angular resolution achievable 660 nm (1 kev γ) 1015 nm (12 kev γ) 2 m distance experiment-detector 280 µm p-in-n silicon detector, rear side injection minimal resolution stays below geometric resolution of 200 µm pixels measurements indicated by black lines Informal AGIPD meeting, , Hamburg Julian Becker Uni-Hamburg 28/14

29 Summary Better suppression of plasma effects if injection is done at the junction n-in-n technology recommended for AGIPD High bias voltage counteracts plasma effects aim for PSFs measured with 660 nm and 1015 nm light at s strip sensor (280 µm thickness / 80 µm pitch) PSF can be approximated as Gaussian distribution sigma of PSF up to ~55 µm measured, function of N γ and bias voltage Minimal angular resolution achievable calculated from PSF Outlook Detailed comparisons to simulations (WIAS-Berlin) Investigations on sensors of different thickness Reference data for sophisticated detector simulations (WIAS-Berlin) Informal AGIPD meeting, , Hamburg Julian Becker Uni-Hamburg 29/14

30 Simulations from WIAS Berlin Simulations using numerical solutions to the van Roosbroeck system of (partial) differential transport equations on a Delaunay grid Major features: arbitrary definition of input e/h distribution possible (space and time) diffusive transport model field dependent mobility model charge carrier interaction models (repulsion, recombination, etc) full 3D simulation code capable of modeling elaborate structures Open points: so far no simulation of electronics -> done with external circuit simulation (SPICE) comparison to measurement data for non-zero density Informal AGIPD meeting, , Hamburg Julian Becker Uni-Hamburg 30/14

31 Position sensitive transients central hit on readout strip neighbor strip charge on neighbor 9.1 x 10 6 e,h pairs -> kev γ 0.8 x 10 6 e,h pairs -> kev γ Informal AGIPD meeting, , Hamburg Julian Becker Uni-Hamburg 31/14

32 M-TCT ceramics 13x26 mm 2 space for d.u.t 32 Pads -> individual channels 2x 16 landing zones for wire bonds separate pads for ground contact separate pads for HV contact 90 rotation symmetry flip symmetry (vias at pads) and central hole for backside injection Informal AGIPD meeting, , Hamburg Julian Becker Uni-Hamburg 32/14

33 Laser beam characteristics minimum spotsize gaussian spot with minimum sigma of ~2 µm (for 660nm light) depth of focus of ~100 µm due to special optics with 75mm focal length Beam profile colored lines -> measurement black lines -> gaussian fit Informal AGIPD meeting, , Hamburg Julian Becker Uni-Hamburg 33/14

34 System calibration Injection of a step function (V) over a known capacitance (C) into the system instead of detector. When no charge is lost: is valid C V U osc ( t) / Roscdt Q = = inj Q meas V 50Ω C Amp Scope 50Ω system to calibrate instead of detector Measurement yields: K K = avg Q meas Q inj = Informal AGIPD meeting, , Hamburg Julian Becker Uni-Hamburg 34/14

35 Some words on electronics Element Bandwidth Attenuator: DC - 4 GHz Oscilloscope: DC GHz Amplifier*: 10 khz -1.0 GHz 2.5 m Cable? Stray L s, C s, R s? Pulse shape is electronically smeared! => SPICE Model actual detector * only necessary with low intensity injection Informal AGIPD meeting, , Hamburg Julian Becker Uni-Hamburg 35/14

36 Comparison to simple Sim. (HH) frontside (electron) injection Laser timing structure taken into account exponential injection decrease taken into account non focused laser beam no carrier interactions taken into account no diffusion taken into account Informal AGIPD meeting, , Hamburg Julian Becker Uni-Hamburg 36/14

37 Comparison to simple Sim. (HH) backside (hole) injection Laser timing structure taken into account exponential injection decrease taken into account non focused laser beam no carrier interactions taken into account no diffusion taken into account Informal AGIPD meeting, , Hamburg Julian Becker Uni-Hamburg 37/14

38 Mobility in simple Sim. (HH) µ = * µ 0 = (1 + µ ( min * µ * 0 µ 0E v sat + ) β ) 1 β µ µ 0 (1 + N C eff ref min α for holes µ 0 * was multiplied by to produce transients of the right time electric field was assumed linear and independent of charge carriers ) Jacoboni Selberherr Jacoboni α = C C µ µ µ µ v v ref, e ref, h 0, e 0, h min, e min, h sat, e sat, h h 0.72 = 1430 = = 80 = 1.45 T ( ) 300K 17 T ( ) 300K 17 T ( ) 300K 2 cm T 2.0 ( ) Vs 300K 2 cm T 2.18 ( ) Vs 300K 2 cm T 0.45 ( ) Vs 300K 2 cm T 0.45 ( ) Vs 300K 7 155K 10 tanh( ) T 6 312K 10 tanh( ) T 2 T ( ) 1K T ( ) = 1.12 = = = βe = 2.57 β = K Informal AGIPD meeting, , Hamburg Julian Becker Uni-Hamburg 38/14

39 Timing structure of 660 nm laser 660 nm linear 600,00 mw 500,00 mw provided by manufacturer opt. power 400,00 mw 300,00 mw 200,00 mw 11,51 mw 5,62 mw 2,50 mw 1,77 mw 660 nm log 100,00 mw 1000,00 mw 0,00 mw 100,00 mw 1 1,1 1,2 1,3 1,4 1,5 1,6 1,7 1,8 1,9 2 2,1 2,2 2,3 2,4 2,5 time [ns] opt. power 10,00 mw 1,00 mw 11,51 mw 5,62 mw 2,50 mw 1,77 mw 0,10 mw 0,01 mw 0 0,5 1 1,5 2 2,5 3 3,5 4 4,5 5 time [ns] Informal AGIPD meeting, , Hamburg Julian Becker Uni-Hamburg 39/14

40 Informal AGIPD meeting, , Hamburg Julian Becker Uni-Hamburg 40/14

41 Informal AGIPD meeting, , Hamburg Julian Becker Uni-Hamburg 41/14