CFEL inside MPI Halbleiterlabor MPI Semiconductor Laboratory Detectors for the experiments at FLASH, LCLS, SCSS and XFEL The fine art of high speed X-ray imaging UNIVERSITÄT SIEGEN
MPI für extraterrestische Physik (MPE) The MPI Semiconductor Laboratory High speed, low noise, low power, radiation hard, high Q.E.... Home made imaging X-ray detectors systems Feb 7, 2007 E. Lama, N. Kimmel, O. Hälker, S. Herrmann, T. Lauf, E. Hyde, N. Meidinger, D. Miessner, G. Hasinger, F. Schopper, G. Schaller, M. Porro, J. Treis, S. Wölfel, C. Zhang, P. Holl, P. Lechner, R. Eckardt, I. Radivojevic, R. Andritschke, L. Strüder A. Bechtel, O. Jaritschin, R. MPI für extraterrestrische Physik M. Schnecke, R. Richter, Hartmann, K. Heinzinger, C. Koitsch, A. Wassatsch H. Soltau, G. Lutz, C. Reich, G. Signeri, F. Hempelmann, Max-Planck-Institut für Physik
Prepared by 1. MPI-HLL (MPE and MPP) Lothar Strüder, Rainer Richter, Matteo Porro, Florian Schopper, Gabi Schächner, Danilo Miessner, Martina Schnecke, Thomas Lauf, Gerhard Schaller, Norbert Meidinger, Sven Herrmann, Laci Andricek, Gerhard Fuchs, Johannes Treis, Nils Kimmel, Robert Andritschke, Zdenka Albrechtskirchinger, Valentin Fedl, Giulio de Vita, Georg Weidenspointner, A. Wassatsch, Hans-Günther Moser, Admir Ramic, Gerhard Fuchs Daniel Pietschner, Johannes Elbs, Olaf Hälker, Toboas Panzner, Stefanie Ebermayer, Sebastian Hasinger, Florian Aschauer, Alexander Bähr, 2. PNSensor and PNDetector Heike Soltau, Robert Hartmann, Peter Lechner, Peter Holl, Rouven Eckhart, Adrian Nicolae, Klaus Heinzinger, Christian Koitsch, Andreas Liebel, Alois Bechteler, Uwe Weichert, Olga Jaritschin, Gerhard Lutz, Sebastian Ihle, Gabriele Signeri, Ivan Ordavo Christian Reich, Christian Thamm, Kathrin Hermenau, Markus Kufner Adrian Niculae, Armin Schön, Barbara Titze, Samantha Jeschke Melanie Schulze
OUTLINE Fully depleted, high speed, monolithic, large format pnccds and DePFETs are being or will be - used from 50 ev to 25 kev for spectroscopic and intensity imaging at the FLASH, Petra III, LCLS and XFEL synchrotrons: SDDs pnccds LSDDs DEPFET APS gatable DEPFETs RNDR
OUTLINE Fully depleted, high speed, monolithic, large format pnccds and DePFETs are being or will be - used from 50 ev to 25 kev for spectroscopic and intensity imaging at the FLASH, Petra III, LCLS and XFEL synchrotrons: SDDs pnccds LSDDs first part DEPFET APS gatable DEPFETs RNDR
OUTLINE Fully depleted, high speed, monolithic, large format pnccds and DePFETs are being or will be - used from 50 ev to 25 kev for spectroscopic and intensity imaging at the FLASH, Petra III, LCLS and XFEL synchrotrons: SDDs pnccds LSDDs DEPFET APS gatable DEPFETs RNDR second part
Requirements of the FLASH, LCLS and XFEL FLASH, LCLS + XFEL pnccd and DePFET system single photon resolution yes yes energy range 0.05 < E < 24 (kev) 0.05 < E < 25 [kev] pixel size (µm) 100 75 (250) sig.rate/pixel/bunch 10 3 (10 5 ) 10 3-10 4 quantum efficiency > 0.8 > 0.8 from 0.3 to 12 kev number of pixels 512 x 512 (min.) 1024 x 1024 and 2048 x 2048 frame rate/repetition rate 10 Hz - 120 Hz up to 250 Hz with pnccd XFEL burst mode 5 MHz (3.000 bunches) > 5 MHz (3.000 bunches) with DePFET system Readout noise < 150 e - (rms) < 30 e - (rms) (2 e - possible) cooling possible - 20 o C optimum room temperature possible
What is the challenge for Detectors @ XFEL Time structure: difference with others Electron bunch trains; up to 3000 bunches in 600 µsec, repeated 10 times per second. Producing 100 fsec X-ray pulses (up to 30 000 bunches per second). 100 ms 100 ms 600 µs 99.4 ms 200 ns X-ray photons 50-100 fs FEL process
What is the challenge for Detectors @ XFEL Time structure: difference with others Electron bunch trains; up to 3000 bunches in 600 µsec, repeated 10 times per second. Producing 100 fsec X-ray pulses (up to 30 000 bunches per second). 600 µs 100 ms 100 ms 99.4 ms 30 000 bunches/s but 99.4 ms (%) no photons 200 ns X-ray photons 50-100 fs FEL process
What is limiting the quantum efficiency? The thickness of Silicon!! Q.E. = 60 % @ 24 kev d = 2 mm Q.E. = 22 % @ 24 kev d = 0.5 mm
Monolithic Integration of optical blocking filters Thin entrance window Silicon entrance window with x nm of SiO 2 and y nm of Si 3 N 4 plus z nm of Al (optical shield) optical light attenuation: 5 10 6 50 ev 5 kev
I. the pnccd for operations up to 250 frames per sec
II. the DePFET active pixel sensor for 200 ns frame acquisition time
CCD basics full depletion (50 µm to 500 µm) back side illumination radiation hardness high readout speed pixel sizes from 36 µm to 650 µm charge handling: more than 106 e-/pixel high quantum efficiency SLAC, Menlo Park 27. 8. 2008 Lothar Strüder, MPI Halbleiterlabor and Universität Siegen
How many charges can be stored What determines the charge handling capacity in a pixel? pixel volume: 20x40x12 µm 3 1x10 4 µm 3 Doping: 10 2 P per µm 3 CHC = 1 x 10 6 per pixel can be increased by external voltages
Hard X-ray SASE Free Electron Lasers LINAC COHERENT LIGHT SOURCE LCLS 2009 European XFEL Facility 2013 2010 SCSS SPring-8 Compact SASE Source
Hard X-ray SASE Free Electron Lasers LINAC COHERENT LIGHT SOURCE LCLS 2009 2010 SCSS SPring-8 Compact SASE Source European XFEL Facility 2013 FLASH in operation
Hard X-ray SASE Free Electron Lasers LINAC COHERENT LIGHT SOURCE LCLS 2009 2010 SCSS SPring-8 Compact SASE Source European XFEL Facility 2013 FLASH in operation FLASH: the ice breaker for XFELs
Hard X-ray SASE Free Electron Lasers LINAC COHERENT LIGHT SOURCE LCLS 2009 2010 SCSS SPring-8 Compact SASE Source FLASH: 5 Hz, 10 Hz and 5 MHz LCLS: 120 Hz European XFEL Facility 2013 FLASH in operation FLASH: the ice breaker for XFELs SCSS: XFEL: 60 Hz 5 Hz, 10 Hz and 5 MHz
Detectors for FLASH+LCLS+XFEL 16 outputs device fabrication is finished now CMX CMX CMX CMX transfer of signal charges insensitive gaps: 800 µm Chip 1: area 29.5 cm 2 format: 1024 x 512 hole diameter: 3 mm Chip 2: area 29.5 cm 2 format: 1024 x 512 The full sensitive area of the system is 59 cm 2 with 75 µm pixels, 1024 x 1024 Full Frame imaging area per chip 512 x 1024 CMX CMX CMX CMX 16 outputs total area per chip: 29.5 cm 2 pixel size 75x75 µm 2 readout time per frame: 4 ms i.e. 250 fps can be triggered externally Total sensitive system area: 59 cm
pnccd: 1024 x 512, 30 cm 2 1024 pixel, 7.8 cm Area: 29.6 cm 2 512 pixel, 3.7 cm Imaging 7.8 x 3.7 cm 2 = 29.6 cm 2 75 x 75 µm 2 1024 parallel read nodes 2 e - @ 250 fps for 6 kev X-rays the system delivers 4k x 4k resolution points in all the area with less than one photon per pixel (typ. 90 %)
Flex lead ceramic carrier for thermal and structural support Flex lead CCD1 Flex lead CCD2 look at radiation entrance Flex lead for window side - power - control - signal out
pnccds overlapping in the center 800 µm insensitive gap Electrical components al components CCD1 Center hole CCD2 Electrical components Ceramic board Electrical components Mechanical thermal structure
The CFEL-ASG Chamber Reaction microscope: Many particle ion and electron imaging spectrometer Velocity map imaging Additional feautures: Integration of jet-targets,ports for lasers, special injectors support structures for fixed targets, etc...
The CFEL-ASG Chamber Imaging system: For inside view the housing of C1 and C2 is drawn transparently. Detector 1 can have any position between P min with d(p IA ;P min ) = 5 cm and P max with d(p IA ;P max ) = 300 mm. Detector 2 is mounted fixed at d(p ;P ) = 500 mm.
The Imager of the CFEL-ASG Chamber Imaging system: (a) format 1024x1024, pixel size 75x75 µm 2, 8x8 cm 2 focal surface (b) center hole, typically 3 mm Due to overlap of the two detectors, effective insensitive area can be reduced to 1.6 mm, insensitive gaps: 0.8 mm (c) movement in y-direction: up to 45 mm
The Imager of the CFEL-ASG Chamber System alignment: Detector 1 is movable in Y, Z and X (limited), 400 mm Ø Detector 2 is fixed, 250 mm Ø
Detectors for FLASH+XFEL+PETRA III devices are scheduled for fabrication end 2009 ready: end 2010 The full sensitive area of the system is 239 cm 2 with 75 µm pixels, 2048 x 2048 CMX CMX CMX CMX Chip 1: area 59 cm 2 format: 1024 x 1024 transfer of charges Full Frame imaging, format per chip 1024 x 1024 total area per chip: 59 cm 2, per system: 236 cm 2 readout time per frame: 8 ms i.e. 125 fps pixel size 75x75 µm 2 This system is 3 side buttable, can be extended to a 2048 x 2048 array 16 outputs
FLASH, PETRA III and XFEL 2048 x 2048 CCD array (resolution points: 8kx8k) pixel size: 75 x 75 µm 2 total area: 236 cm 2 readout time: < 8 ms read noise < 15 electrons Charge handling capacity: > 1000 photons pp Energy 0.1<E<24 kev thickness: 500 µm
Recent pnccds fabrications flight type erosita version format: 384 x 384 x 2 pixel area: 8.4 cm 2 + 5.6 cm 2 pixel size: 75 x 75 µm 2 prototype erosita version format: 256 x 256 x 2 pixel area: 3.7 cm 2 + 2.5 cm 2 pixel size: 75 x 75 µm 2
Measurements with 512x256 pnccds @ FLASH + BESSY 256 parallel low noise readout channels with 2 or 4 output nodes Frame store area 256 x 256 pixel Imaging area 256 x 256 pixel
CAMEX block diagram Sr1 S1... S8 Sr1 G2 G4 G1 Vdd G3 IN Vbst Sin Stst BW1 BW2 BW3 Ss&h Smux OUT- OUT+ Vsss Tst current source JFET- amplifier passive lowpass filter CDS- filter sample & hold serializer cable driver digital control Vref master reference DAC(s) communication
CAMEX erosita 128I-JD 9153 um 128 x current source 6080 um 128 x AMP & CDS-filter S&H MUX CMX internal SEQ (shiftregister 16x64) differential output buffer BIAS DACs BIAS master reference configuration registers
Development status: Flex lead + frontend PCB - connects CCD module to front-end PCB by wedge bonds - front-end PCB: analog output buffers (for the s, one per CAMEX) LVDS receivers generators of the analog CCD clock pulses (PHI drivers) connectors as electric interface to the camera electronics DUO flex lead: performance tested erosita flex lead: circuit diagram: under way layout: under way 29
Structure determination with a FEL Molecules atomic resolution Crystal J. Kirz, Nature Physics 2, 799 (2006) Lysozyme R. Neutze, J. Haidu et al., Nature 406, 752 (2000) Radiation damage and Coulomb explosion
Clusters in the Flight Xe clusters CFEL Detectors CCD camera nozzle skimmer cluster beam MCP detector + phosphor screen 550 nm visible light FEL beam aperture skimmer aperture VUV scattered light beam dump plane mirror
FLASH set-up in T.M. Beamline planar detector geometry in reflection mode CCD detector X-ray Beam @ 90 ev Multilayer X-ray Mirror
First observation of 32 nm Xe nanocluster simulation of 90 ev X-rays scattering at single Xe nanoclusters with 32 nm size (C. Bostedt et al.) field of view of the pnccd during the experiment at the FLASH X-ray free electron laser facility at DESY, Hamburg
90 ev X-rays in single photon counting mode!!! t r i g g e r t h re E = 90 ev 25 ± 1.7 e-h pairs ENC = 2.5 e - (rms) FWHM: 38.9 ev s h ol d 30 ev T = -50 C Spectrum from 4.000 frames with 0.05 photons/pixel/frame
Set-up in T.M. Beamline @ FLASH pnccd3 pnccd1 42 o 80 o 10 o 25 o pnccd2 20 o 40 o
Clusters in the Flight Collaboration: TU Berlin, MPI-HLL, MPI-K: Bostedt, Rupp, Adolph, Möller, Hartmann, Strüder, Rudenko, et al.
Free Electron Lasers Screen-Shots: To be evaluated!
Free Electron Lasers Screen-Shots: To be evaluated! SLAC, Menlo Park 27. 8. 2008 Lothar Strüder, MPI Halbleiterlabor and Universität Siegen
Clusters in the Flight Collaboration: TU Berlin, MPI-HLL, MPI-K: Bostedt, Rupp, Adolph, Möller, Hartmann, Strüder, Rudenko, et al.
Clusters in the Flight intensity / photons per pixel angle / deg
Clusters in the Flight intensity / photons per pixel angle / deg
Clusters in the Flight intensity / photons per pixel detector 1 detector 2 detector 3 angle / deg
Clusters in the Flight intensity / photons per pixel R = 83 nm detector 1 detector 2 detector 3 angle / deg
Clusters in the Flight intensity / photons per pixel R = 83 nm detector 1 detector 2 detector 3 R = 80 nm angle / deg
Clusters in the Flight intensity / photons per pixel Structure and Dynamics! detector 1 detector 2 detector 3 R = 83 nm R = 80 nm With strongly increased absorption! angle / deg
Bragg s law in energy space In angular space: λ=2 D sin α Peak distance : sinα n+1 sinα n 1/D Changing footprint area In energy space: E = hc/(2*d*sin α) E 6.2/(α*D) Peak distance : E n+1 E n 1/D
Useable spectral range Emission spectrum of BESSY II (1.7 GeV) Absorption by Air Usable energy range (measurement + calc.) Energy resolution of the detector < 150 ev 0.7 m air Useful spectrum 5... 30 kev
Langmuir-Blodgett-Film fixed angle of incidence Few minutes counting 1.000 to 10.000 photons per frame 400 to 1.000 frames per second typical measurement times: 100 s to 1.000 s position,time and energy resolved photons per spectrum: typ 10 8 α i = 0.8 deg
pnccd operating parameters Parameter Frame Store pnccd XMM type pnccd pixel size 36 µm, 48 µm, 51 µm and 75 µm 150 µm format 256 2 x2, 264 2 x2, 384 2 x2, 512x1024 64x200, 400x400 active area (image only) 3.9 cm 2, 1.9 cm 2, 8.3 cm 2, 29 cm 2 3 cm 2, 36 cm 2 sensitive depth 450 µm 300 µm readout noise 2 2.5 electrons (rms) 5 electrons (rms) Q.E. 90 % from 0.4 to 11 kev 90 % from 0.4 to 10 kev charge handling 5 x 10 5 electrons per pixel 5 x 10 5 e per pixel CTI @ 6 kev 1 x 10-5 50 x 10-5 E 40 ev @ 300 ev, 125 ev @ 6 kev --, 150 ev @ 6 kev readout time 10 µs or 20 µs per row 25 µs per row pixel rate 13 Mpix or 25 Mpix per second 2.8 Mpix per second frame rate up to 900 fps, 20 fps for erosita 14 fps out-of time events 0.2 % in the case of erosita 6%
What is the challenge for Detectors @ XFEL Time structure: difference with others Electron bunch trains; up to 3000 bunches in 600 µsec, repeated 10 times per second. Producing 100 fsec X-ray pulses (up to 30 000 bunches per second). 100 ms 100 ms 600 µs 99.4 ms 200 ns X-ray photons 50-100 fs FEL process
Original LSDD Concept: reduce number of readout channels by time multiplexing 128 channels 1 MHz frame 200µm pixel on-chip electr. E=2-20 kev 10 3 X-rays QE>80%@10keV ENC~50 el. Expandable to: 512x512 (monolithic) 1000 ns 1000 ns (Δx=200µm, Δt=15 ns) 1.28 cm (64 pixels) 1.28 cm (64 pixels) on-chip electr. V drift 13 µm/ns (i.e. ~3.5V/30µm bias) T drift, max = 1000 ns 128 channels Limitations of that concept: frame rate limited to 1 MHz limited dynamic range, limited occupancy system noise difficult to achieve @ 5 ns shaping
esa s, NASA s and JAXA s IXO mission in 2021
Circular DEPMOSFET pixels
Circular DEPMOSFET pixels
DEPFET Active Pixel Sensor matrix organisation common back contact operation philosophy» thin, homogeneous entrance window one active row» fill factor 100 % all other pixels turned off row-wise connection of gate, clear, clear gate» low power consumption column-wise connection of source / drain» individually addressable pixels all operations in a row in parallel SLAC, Menlo» Park windowing 27. 8. option 2008» fast processing Lothar Strüder, MPI Halbleiterlabor and Universität Siegen
DEPFET Active Pixel Sensor matrix organisation common back contact operation philosophy» thin, homogeneous entrance window one active row» fill factor 100 % all other pixels turned off row-wise connection of gate, clear, clear gate» low power consumption column-wise connection of source / drain» individually addressable pixels all operations in a row in parallel SLAC, Menlo» Park windowing 27. 8. option 2008» fast processing Lothar Strüder, MPI Halbleiterlabor and Universität Siegen
DEPFET Active Pixel Sensor matrix organisation common back contact operation philosophy» thin, homogeneous entrance window one active row» fill factor 100 % all other pixels turned off row-wise connection of gate, clear, clear gate» low power consumption column-wise connection of source / drain» individually addressable pixels all operations in a row in parallel SLAC, Menlo» Park windowing 27. 8. option 2008» fast processing Lothar Strüder, MPI Halbleiterlabor and Universität Siegen
DEPFET Active Pixel Sensor matrix organisation common back contact operation philosophy» thin, homogeneous entrance window one active row» fill factor 100 % all other pixels turned off row-wise connection of gate, clear, clear gate» low power consumption column-wise connection of source / drain» individually addressable pixels all operations in a row in parallel SLAC, Menlo» Park windowing 27. 8. option 2008» fast processing Lothar Strüder, MPI Halbleiterlabor and Universität Siegen
DEPFET Active Pixel Sensor matrix organisation common back contact operation philosophy» thin, homogeneous entrance window one active row» fill factor 100 % all other pixels turned off row-wise connection of gate, clear, clear gate» low power consumption column-wise connection of source / drain» individually addressable pixels all operations in a row in parallel SLAC, Menlo» Park windowing 27. 8. option 2008» fast processing Lothar Strüder, MPI Halbleiterlabor and Universität Siegen
SIMBOL-X-Hybrid 3.2 cm 3.2 cm
8.2 cm 8.2 cm
Hybrid Pixel Detector Approach DePMOS Active Pixel Sensor X-rays Connecting Bumps 1 per pixel CMOS Layer Signal processing Signal storage & output
DePFET system overview
Basic concept Internal gate extends below large area source Small signal charge is collected below transistor channel Large signal charges distributed also over outer regions of internal gate; only the fraction below channel steers the transistor current efficiently This arrangement leads to a non linear characteristics The shape of this characteristics can be tuned by the doping profile of the internal gate Clearing electrode is not shown in the picture
drift ring pixel size: 200x200 µm 2 drain can be realized from 75 x 75 µm 2 to 300 x 300 µm 2 gate clear 5 deep n implants for the graded internal gate source
37 consecutive steps of Non-linear DEPFET: Simulation 10-14 C charge deposition were simulated: i.e. 37 x 62.500 e - = 2.312.500 e - Resulting drain current I = 130 µa and
DePFET output characteristics this corresponds to 40 6.25 x 10 4 e - = 2.5 x i.e. 1.000 photons of 10 kev per pixel
Timing behaviour of the DePFET
Radiation damage tests Generation and saturation of surface states
Surface charge generation
Charge cloud expansion Plasma effects are not included Deformation of the local electric field is not included
electron dynamics potential electron concentration after 1.3 ns electron concentration after 50 ps injection of 400.000 electrons
Mechanical, thermal, electrical interconnections
What is achieved (by simulation) up to now: high dynamic range from single photon counting @ 1 kev up to 1.000 photons @ 10 kev in one single pixel the maximum voltage swing at the on-chip amplifier nev exceeds 1 V highest resolution @ low signals, lowest resolution @ the highest photon densities. Intensity resolution is matched with the precision of the Poisson statistics of the incoming photons the charge collection, the signal processing and the cha
Development system philosophy prototype system breadboard full quadrant with system with system with at least 64 x 64 or 512 x 128 512 x 512 pixel 128 x 128 pixel pixel
Full DePFET system (1k x 1k) one quadrant, composed by 4 ladders fill factor: 90 % a qualified prototype should be available in 3 years the breadboard system in 4.5 years and the full system in 6 years from now
Conclusions: 1. the proposed DePFET detector covers the dynamic range the required noise performance pixel size area fill factor charge collection, signal filtering and reset are within 200 ns 2. the anticipated pixel area of 200 x 200 µm 2 allows for the integration of a pixelwise analog signal processing, digitization and data storage with a 130 nm process 3. radiation hardness for X-rays seem to be sufficient
Conclusions Phase I + II fast pnccd systems will cover the parameter space for FLASH, LCLS, SCSS and XFEL except for the 5 MHz operation Phase III DePFET type active pixel sensors will cover the ultrafast imaging modes up to 5 MHz frame rates in 2014 Both systems are: highly efficient,