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 1. Si-type influence on charge collection time 2. Measurements on charge collection time 3. Measurements of the PSF 4. Summary AGIPD meeting, , PSI Julian Becker Uni-Hamburg 1/7

2 Pulses for different thicknesses Current pulse in assuming T=-20 C, continuous photon absorption, no plasma effects, 250V depletion voltage and 500V bias. Positive polarity for all pulses for easier comparison. Charge collection time dependent on thickness and type ~6ns/12ns for 500 µm plasma properties of 700 µm thickness not yet tested! ~10ns/21ns for 700 µm AGIPD meeting, , PSI Julian Becker Uni-Hamburg 2/7

3 Measurements on charge collection time current 1.68 x kev photons zoom small peaks and undershoot due to reflections in circuitry current pulses are limited (no pure exponentials) for 500 V applied bias 60 ns charge collection time 60ns AGIPD meeting, , PSI Julian Becker Uni-Hamburg 3/7

4 Measurements on charge collection time 1.68 x kev photons focus => σ ~3 µm -1 mm => σ ~40 µm -5 mm and -10 mm have the approximate charge collection time of low density pulses low density pulses AGIPD meeting, , PSI Julian Becker Uni-Hamburg 4/7

5 Measurements of the PSF tightly focused (3 µm) hard x-rays (~12 kev) 300 µm new parameterization: Gaussian (diffusion) circle (plasma) only of minor influence on imaging performance as plasma effects stay mostly confined in the pixels 450 µm scaling with thickness inconclusive, seems like downgrading one step (OD+1.0) for increased thickness AGIPD meeting, , PSI Julian Becker Uni-Hamburg 5/7

6 Measurements of the PSF 1015 nm 12 kev γ results are preliminary! PSF( x) = ( g c)( x) 1 g( x) = e σ 2π 2 2 c( x) = r 2 πr x 2σ parameterization: Gaussian (diffusion) circle (plasma) scaling with thickness inconclusive, seems like downgrading one step (OD+1.0) for increased thickness AGIPD meeting, , PSI Julian Becker Uni-Hamburg 6/7 x 2 2 2

7 Summary 500 µm 500 V seems possible AGIPD meeting, , PSI Julian Becker Uni-Hamburg 7/7

8 Backup AGIPD meeting, , PSI Julian Becker Uni-Hamburg 8/7

9 Measurements on charge collection time current 90x10 6 e,h pairs -> 3.24x kev photons charge no exponential shape! 60ns/100ns AGIPD meeting, , PSI Julian Becker Uni-Hamburg 9/7

10 Introduction The European XFEL will provide images with a huge dynamic range: from single photons to: 10 4 in the interesting physics range 10 5 routinely encountered 10 6 possible but probably not in one pixel >10 6 operating accidents (e.g. non-intended objects in the beam path) ASIC has to process regularly effects on input of ASIC have to be known to ensure ASIC functionality All results presented here are obtained with stronger focusing than expected at the XFEL. Results should be viewed as a worst case estimate. AGIPD meeting, , PSI Julian Becker Uni-Hamburg 10/7

11 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 AGIPD meeting, , PSI Julian Becker Uni-Hamburg 11/7

12 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 AGIPD meeting, , PSI Julian Becker Uni-Hamburg 12/7

13 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 AGIPD meeting, , PSI Julian Becker Uni-Hamburg 13/7

14 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 AGIPD meeting, , PSI Julian Becker Uni-Hamburg 14/7

15 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 in primary pixel will trigger switching to second gain stage in adjacent pixels Nγ > 6.8x10 4 in primary pixel 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 AGIPD meeting, , PSI Julian Becker Uni-Hamburg 15/7

16 Range switching in adjacent pixels sensor thickness voltage Nγ in primary pixel to switch to 2nd stage Nγ in primary pixel to switch to 3rd stage 500µm 500V x µm 1000V x µm 500V x µm 1000V x10 4 Having a primary pixel in third gain stage will result in switching in adjacent pixels even if less than 85 photons are absorbed AGIPD meeting, , PSI Julian Becker Uni-Hamburg 16/7

17 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 AGIPD meeting, , PSI Julian Becker Uni-Hamburg 17/7

18 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 AGIPD meeting, , PSI Julian Becker Uni-Hamburg 18/7

19 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 AGIPD meeting, , PSI Julian Becker Uni-Hamburg 19/7

20 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!) AGIPD meeting, , PSI Julian Becker Uni-Hamburg 20/7

21 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!) AGIPD meeting, , PSI Julian Becker Uni-Hamburg 21/7

22 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 AGIPD meeting, , PSI Julian Becker Uni-Hamburg 22/7

23 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 AGIPD meeting, , PSI Julian Becker Uni-Hamburg 23/7

24 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. AGIPD meeting, , PSI Julian Becker Uni-Hamburg 24/7

25 - + 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 AGIPD meeting, , PSI Julian Becker Uni-Hamburg 25/7

26 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 AGIPD meeting, , PSI Julian Becker Uni-Hamburg 26/7

27 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! AGIPD meeting, , PSI Julian Becker Uni-Hamburg 27/7

28 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 AGIPD meeting, , PSI Julian Becker Uni-Hamburg 28/7

29 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 AGIPD meeting, , PSI Julian Becker Uni-Hamburg 29/7

30 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 AGIPD meeting, , PSI Julian Becker Uni-Hamburg 30/7

31 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 AGIPD meeting, , PSI Julian Becker Uni-Hamburg 31/7

32 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] AGIPD meeting, , PSI Julian Becker Uni-Hamburg 32/7 current plasma delay (660 nm)

33 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. AGIPD meeting, , PSI Julian Becker Uni-Hamburg 33/7

34 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 AGIPD meeting, , PSI Julian Becker Uni-Hamburg 34/7

35 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) AGIPD meeting, , PSI Julian Becker Uni-Hamburg 35/7

36 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 AGIPD meeting, , PSI Julian Becker Uni-Hamburg 36/7

37 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 AGIPD meeting, , PSI Julian Becker Uni-Hamburg 37/7

38 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 AGIPD meeting, , PSI Julian Becker Uni-Hamburg 38/7

39 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 AGIPD meeting, , PSI Julian Becker Uni-Hamburg 39/7

40 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 = AGIPD meeting, , PSI Julian Becker Uni-Hamburg 40/7

41 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 AGIPD meeting, , PSI Julian Becker Uni-Hamburg 41/7

42 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 AGIPD meeting, , PSI Julian Becker Uni-Hamburg 42/7

43 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 AGIPD meeting, , PSI Julian Becker Uni-Hamburg 43/7

44 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 AGIPD meeting, , PSI Julian Becker Uni-Hamburg 44/7

45 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] AGIPD meeting, , PSI Julian Becker Uni-Hamburg 45/7

46 AGIPD meeting, , PSI Julian Becker Uni-Hamburg 46/7

47 AGIPD meeting, , PSI Julian Becker Uni-Hamburg 47/7