Integrated CMOS sensor technologies for the CLIC tracker Magdalena Munker (CERN, University of Bonn) On behalf of the collaboration International Conference on Technology and Instrumentation in Particle Physics 27, Beijing Work sponsored by the Wolfgang Gentner Programme of the Federal Ministry of Education and Research
CLIC - the Compact Linear Collider Possible multi-tev linear e + e - collider in the post LHC phase at CERN: CLIC layout: 3 TeV stage: Bunch separation [ns].5 # bunches / train 32 Train duration [ns] 56 Repetition rate [Hz] 5 Bunch size σ x / σ y [nm] ~ 45 / σ z [µm] 44 High centre of mass energies up to 3 TeV Dense bunches to achieve high luminosity See talk by E. Sicking: Detector challenges for future high-energy e + e colliders High background rates time-stamping of ~ ns needed to reject background hits Note: significantly lower radiation levels of ~ neq/cm 2 /y compared to hadron colliders Magdalena Munker (CERN, University of Bonn) - TIPP 27 p. / 2
Integrated technologies for the CLIC tracker Physics needs & environment: Tracker requirements: Technology choice: See talk by A. Nurnberg: Momentum resolution σ PT / p T 2 = 2 5 / GeV Suppression of high beam beam background occupancies } 7 µm single } point resolution ~% radiation } length per layer } ns time } stamping. Highly granular / fine pitch 2. Thin / low material budget 3. Fast signal A vertex and tracking detector system for CLIC Needed: Multi purpose technology Challenge to meet requirements simultaneously: Benefit from integrated CMOS technologies: No separate ASIC material: Lower material budget No sensor-asic interconnect: Large scale production (m 2 CLIC tracker) Finer pitch Crucial for integrated CMOS technologies: Full depletion for fast and fully efficient operation Achievable with CMOS circuitry on High Resistivity HR epitaxial layer (epi)? Magdalena Munker (CERN, University of Bonn) - TIPP 27 https://en.wikipedia.org/wiki/multitool#/media/ File:Lezyne_Carbon collage.jpg p. 2 / 2
The Investigator Chip (W. Snoeys, J. W. van Hoorne et. al.) HR-CMOS process: 8 nm High Resistivity (HR) CMOS process, 5-4 µm thick epitaxial layer (-8 kωcm): Developed as part of ALPIDE development for ALICE ITS upgrade Fully monolithic ALPIDE chip developed in this process Test-chip: Various mini-matrices with different pixel layouts: Investigator: Investigator & readout board: Optimisation of pixel layout: Minimising size of collection diode Minimise capacitance (~ ff) Large signal/noise fast timing (~ ns) External readout board (designed by K. M. Sielewicz): 64 ADCs to read out full analogue waveform of 8 x 8 active pixel matrix 65 MHz sampling clock limits achievable timing resolution Two different submissions: Changes in modified process to achieve full depletion: Better timing performance Radiation hardness Magdalena Munker (CERN, University of Bonn) - TIPP 27 Schematic of process cross section of standard process: N-well Schematic of process cross section of modified process: N-MOS P-MOS N-well N-MOS P-MOS Deep P-well Deep P-well NP- P++ backside P p. 3 / 2
Test-beam studies Test-beam setup: Timepix3 telescope at SPS beam line: Investigator data-taking & reconstruction: If at least one pixel crosses a seed threshold: Readout board Investigator Full analogue waveform of all 8 x 8 active pixels read out Timestamp send to telescope planes for offline synchronisation Telescope planes Amplitude [ADC] Example of single pixel waveform: 56 54 Data taking threshold Hit time t(hit) Signal Rise time t(rise) defines slope 5 Excellent timing resolution ~ ns: Benefit for studies of fast Investigator timing Magdalena Munker (CERN, University of Bonn) - TIPP 27 52 Timepix3 telescope: Excellent track prediction resolution ~ 2 µm: Benefit for sub-pixel performance studies for small pixel sizes of Investigator Pedestal 58 48 33 34 35 36 37 38 Time [/65 [65 MHz] MHz] Waveform reconstructed by exponential fit: p. 4 / 2
Efficiency over pixel matrix / modified process Local y-coordinate [mm] Local x-coordinate [mm].2. -. Efficiency over pixel matrix: Pitch = 28 µm, bias voltage = 6 V, epi thickness = 25 µm, modified process: Single pixel Fiducial region Seed threshold ~ 5 e -.2 - -.2 -...2 Local x-coordinate [mm] Local x-coordinate [mm].9.8.7.6.5.4.3.2. Analysis of efficiency of standard process currently ongoing Efficiency > 99 % over fiducial region (masking half of edge pixels to account for limited track precision) Magdalena Munker (CERN, University of Bonn) - TIPP 27 p. 5 / 2
Cluster size & resolution / standard & modified process Standard process: Pitch = 28 µm Bias voltage = 6 V Epi thickness = 8 µm Neighbour threshold ~ 7 e - Arbitrary units.5.4.3.2. μ ~.5 Arbitrary units.8.6.4.2 σ ~ 5 μm Gauss fit Modified process: 2 3 4 5 6 Cluster size X -.4 -.2.2.4 xdut X residual - xtrack [mm] Pitch = 28 µm Bias voltage = 6 V Epi thickness = 25 µm Neighbour threshold = 5 e - Arbitrary uunits.6.4.2 μ ~.3 Arbitrary units.8.6.4.2 σ ~ 6 μm Gauss fit 2 3 4 5 6 Cluster size X -.4 -.2.2.4 X residual [mm] xdut - xtrack [mm] Despite thinner epi and larger threshold for standard process: Larger cluster size and better resolution, as expected from more diffusion Position resolution matching well requirement of 7 µm for CLIC tracker (t.b.c. with fully integrated chip). Magdalena Munker (CERN, University of Bonn) - TIPP 27 p. 6 / 2
Timing / standard & modified process Standard process: Pitch = 28 µm Bias voltage = 6 V Epi thickness = 8 µm Neighbour threshold ~ 7 e - Seed threshold ~ 2 e - Modified process: Pitch = 28 µm Bias voltage = 6 V Epi thickness = 25 µm Neighbour threshold = 5 e - Seed threshold ~ 5 e - Gauss fit Arbitrary units.3.25.2.5 Gauss fit. σ ~ 7 ns.5 σ ~ 5 ns -. -.5.5. t Hit - t track [µs] Faster timing for modified process, as expected from full depletion Measured timing resolution limited by readout sampling frequency of 65 MHz Timing resolution matching well requirement of ns for CLIC tracker (t.b.c. with fully integrated chip). Magdalena Munker (CERN, University of Bonn) - TIPP 27 p. 7 / 2
Sub pixel studies / modified process Y-coordinate [pixel] Mean cluster size in pixel cell:.9.8 Work in.7.6.5.4.3.2...2.3.4.5.6.7.8.9 X-coordinate [pixel] 3 2.8 2.6 2.4 2.2 2.8.6.4.2 Mean cluster size Y-coordinate [pixel] Mean seed signal in pixel cell:.9.8 Work in.7.6.5.4.3.2...2.3.4.5.6.7.8.9 X-coordinate [pixel] 8 6 4 2 8 6 Mean seed signal [e - ] Y-coordinate [pixel] Mean efficiency in pixel cell:.9.8 Work in.7.6.5.4.3.2...2.3.4.5.6.7.8.9 X-coordinate [pixel].95.9.85.8.75.7.65 Mean efficiency Results shown for modified process: Pitch = 28 µm, epi thickness = 25 µm, bias voltage = 6V, neighbour threshold ~ 5 e - More charge sharing in pixel edges and corners Higher cluster size and Lower seed signal in pixel edges and corners Low seed threshold of ~ 5 e - during data taking: No significant efficiency loss in pixel corners Sub-pixel performance in qualitative agreement with expectations. Magdalena Munker (CERN, University of Bonn) - TIPP 27 p. 8 / 2
Simulation Simulation chain: GEANT4 simulation: Energy that particle deposits while traversing the sensor. 2-dimensional TCAD simulation: Simulate sensor geometry, doping and bias voltage application Transient simulation using particle with energy deposit from GEANT4 Fast parametric model: Energy fluctuations Threshold application Telescope resolution Reconstruction TCAD simulation of standard & modified process: Electrostatic potential for standard process: Electrostatic potential for modified process: Depletion Depletion Magdalena Munker (CERN, University of Bonn) - TIPP 27 p. p. 9 / 2 3
Comparison data - simulation / standard process Mean X cluster size in pixel cell: Residual in X-dimension: Mean Mean X cluster size 2.5 8 μm epi Arbitrary Arbitray units. Simulation Data.8.6.4.2 8 μm epi 6 V 6 V Neighbour threshold ~ 7 e - Neighbour threshold ~ 7 e -.2.4.6.8 X-coordinate [pixel] -. -.5.5. xdut X residual - xtrack [mm] 55 Fe- calibration applied to define threshold in simulation Excellent agreement between simulation and data on sub-pixel level. Magdalena Munker (CERN, University of Bonn) - TIPP 27 p. / 2
Comparison data - simulation / modified process Cluster size distributions for different thresholds: Arbitrary units.9.8.7.6.5.4 Neighbour threshold ~ 5 e - : Simulation Data 25 μm epi 6 V.3.2..5.5 2 2.5 3 3.5 4 4.5 5 Cluster size Arbitrary units.9.8.7.6.5.4.3.2. Simulation Data Neighbour threshold ~ 2 e - : 25 μm epi 6 V.5.5 2 2.5 3 3.5 4 4.5 5 Cluster size Arbitrary units Residual distributions for different thresholds:.8.7.6.5.4.3.2. Neighbour threshold ~ 5 e - : Simulation Data 25 μm epi 6 V -.3 -.2 -...2.3 Residual [mm] xdut - xtrack [mm] Arbitrary units.6.5.4.3.2. Neighbour threshold ~ 2 e - : 25 μm epi 6 V Simulation Data -.3 -.2 -...2.3 Residual [mm] xdut - xtrack [mm] Mean cluster size & resolution for different thresholds: Mean cluster size [%] (sim.-data)/sim. 2.9.8.7.6.5.4.3.2. 25 μm epi 6 V Simulation Data 5-5 - 4 6 8 2 4 6 8 2 - Threshold [e ] Resolution [µm] [%] (sim.-data)/sim. 9 8 7 6 5 4 3 25 μm epi 6 V Simulation Data Resolution defined as RMS of residual histogram 5-5 - 4 6 8 2 4 6 8 2 - Threshold [e ] 55 Fe-calibration applied to define threshold in simulation Expected trend of lower cluster size and worse resolution visible in data & simulation Good agreement of data and simulation for different thresholds within a few percent. Magdalena Munker (CERN, University of Bonn) - TIPP 27 p. / 2
Summary Study of Investigator HR-CMOS test-chip with respect to requirements for the CLIC tracker: Test-beam study of two different submissions standard & modified process Spatial and timing resolution matching requirements of 7 µm single point resolution & ns time stamping for CLIC tracker: Single point resolution ~ 6 µm } Modified process Time resolution < 5 ns 28 µm pitch (t.b.c. with fully integrated chip) Efficiency > 99% Local x-coordinate [mm].2. -. Single pixel Efficiency over pixel matrix, modified process, -.2 -.2 -...2 Local x-coordinate [mm].9.8.7.6.5.4.3.2. Studies used as input for design of fully monolithic tracker chip for CLIC (see talk by A. Nurnberg: A vertex and tracking detector system for CLIC ) Explore Investigator HR-CMOS technology: Detailed understanding of charge sharing on sub-pixel level Simulation of standard and modified process show agreement between simulation and data within a few percent showing a good understanding of the studied technology Electrostatic potential, TCAD simulation standard process, Magdalena Munker (CERN, University of Bonn) - TIPP 27 p. 3 2 / 3 2
Andreas Nürnberg Dominik Dannheim Wolfgang Klempt Walter Snoeys Jacobus W. v. Hoorne Krzysztof M. Sielewicz THANK YOU!
The CLIC detector CLIC detector for high precision measurements: Physic aims Detector needs CLIC detector model: E.g. Higgs recoil mass, Smuon endpoint Momentum resolution σ PT / p T 2 = 2 5 / GeV W/Z/H seperation E.g. Higgs couplings (b/c-tagging) Jet energy resolution σe / E = 3.5% - 5% Impact parameter resolution σ rϕ ~ 5 * 5 / p sin 3/2 Θ µm ~ 4.5 m ~.5 m Tracker requirements: Momentum resolution at high p T Momentum resolution at low p T Reduce occupancies from beam-beam interactions 7 µm single } point resolution ~% radiation length per layer ns time stamping Large area (~m 2 ) silicon tracker
Analysis & definition of observables Investigator event reconstruction: Signal defined as magnitude of amplitude drop Noise defined as RMS of fluctuation around pedestal Analysis cut on Signal/Noise > 5 for each single pixel (Note: higher data taking threshold corresponds to cut on seed signal while lower analysis cut corresponds to cut on neighbour pixel signal) Fit exponential function f(t) to waveform of each pixel to extract exact timing and signal: f(t) = { Pedestal if t t Hit Pedestal + Signal (e -(t - t Hit) / t Rise - ) if t > t Hit Example of single pixel waveform: Further analysis cuts: Event size of µs Distance track-investigator hit position < 2 x pixel pitch Amplitude [ADC] 58 56 54 52 Signal Pedestal Hit time t(hit) Rise time t(rise) defines slope Data taking threshold Masking of half of edge pixels to avoid bias by edge effects due to limited tracking resolution and/or charge sharing 5 48 33 34 35 36 37 38 Time [65 MHz] [/65 MHz]