The INFN R&D: new pixel detector for the High Luminosity Upgrade of the LHC

Size: px

Start display at page:

Download "The INFN R&D: new pixel detector for the High Luminosity Upgrade of the LHC"

Melvin Allen
6 years ago
Views:

results: D before irradiation ATLAS results: D before irradiation Summary and prospects /4/17 Mauro Dinardo

1 IFAE 17: XVI Incontri di fisica delle alte energie The INFN R&D: new pixel detector for the High Luminosity Upgrade of the LHC INFN Pixel R&D: main design features CMS results: thin-planar before & after irradiation CMS results: D before irradiation ATLAS results: D before irradiation Summary and prospects /4/17 Mauro Dinardo Università degli Studi di Milano Bicocca and INFN, Italy on behalf of the INFN (ATLAS-CMS)-FBK Pixel R&D Collaboration

2 INFN Pixel R&D: main design features High Luminosity upgrade of the CERN-LHC: operation conditions Luminosity 5x1 4 /(cm s), up to events/5 ns bunch crossing Radiation level for first pixel layer at fb 1 ~x1 16 neq/cm (~1 years) carriers lifetime ~. ns, mean free path ~ μm for electrons at saturation velocity Sensor design contraints Maintain occupancy at % level and increase the spatial resolution pixel cell size ~ 5x1 μm or 5x5 μm (currently 1x15 μm CMS, 5x5 μm ATLAS) Reduce electrodes distance to increase electric field and thus the signal thin planar or D columnar technologies Joint ATLAS-CMS INFN collaboration, partnership with Fondazione Bruno Kessler-FBK (Trento, Italy), for the development of thin planar and D columnar n-in-p sensors on 6 FZ wafers with Direct Wafer Bond( 1 ): Planar process options: p-spray and/or p-stop periphery design: standard and active-edge D columnar single sided process, optimised by FBK ( 1 ) IceMos Technology, Belfast

3 INFN Pixel R&D: main design features Pixel cell Pixel cell 1-1 μm High resistivity 85 μm Direct Wafer Bond( 1 ) or μm Low resistivity or 18 μm Two wafers, high and low resistivity, bonded together Backplane at HV p-type high resistivity (active layer) > kohm cm Active layer thickness: two choices, 1 μm and 1 μm Thinning process after fabrication: final total sensor thickness from 85 μm down to 18 μm Doping concentration profile measurement Effective thickness reduced by Boron diffusion from wafer carrier: ~1 μm MIP Most Probable Value expected at: ~6 e for 1 μm thick sensors ~8 e for 1 μm thick sensors NIMA 84 (16) 88 ( 1 ) IceMos Technology, Belfast

4 INFN Pixel R&D: main design features Prototypes assembly main features Hybridisation with chip via bump-bonding: SnAg( 1 ) and Indium( ) Spark protection (only for thin planar sensors): on sensor (periphery) and chip (periphery or whole area) by BCB (Benzo-Cyclo-Butene) layer ~ μm thick( 1 ) (= Planar and Active Edge device) D single sided process, optimised by FBK (more details in backup slides) Thin sensors on support wafer: SiSi or SOI Ohmic columns/trenches depth > active layer depth (for bias) Junction columns depth < active layer depth (for higher Vbreakdown) Reduction of columns diameter to ~5 μm Holes (at least partially) filled with poly-si ( 1 ) IZM Fraunhofer, Berlin ( ) Leonardo Finmeccanica, Rome 4

INFN Pixel R&D: main design features 5x5

with 1 junction electrode (1E) and 5x1 μm

ohmic columns under test bumps on columns

5 INFN Pixel R&D: main design features 5x5 μm 1E 5x1 μm 1E 5x1 μm E!"#$%# (#)*+,# L ~5 μm -.%,# L~51.5 um &''$%# 4#)*+,# L ~8 μm /%1# 1# 5x5 μm with 1 junction electrode (1E) and 5x1 μm with 1 junction electrode (1E) enough space for bump pad 5x1 μm with junction electrodes (E) has bump pad too close to ohmic columns under test bumps on columns 1x15 and 5x5 μm cell sizes made for compatibility with available readout chip 5

CMS results: testbeam setup Detectors Under Test (DUTs) Testbeam carried out at Fermilab MTest area (NIM-A 811

rows and 5 columns) ~8 μm resolution on each coordinate { { { 4 telescope planes downstream 4 telescope planes

than 5 hits on each plane track Χ / d.o.f.

6 CMS results: testbeam setup Detectors Under Test (DUTs) Testbeam carried out at Fermilab MTest area (NIM-A 811 (16) ) 1 GeV protons from Main Injector 8 pixel planes based on PSI46 analog chip (1x15 μm pixel cell, 8 rows and 5 columns) ~8 μm resolution on each coordinate { { { 4 telescope planes downstream 4 telescope planes upstream Base requirements for testbeam data analysis tracks with 8 associated hits (one per plane) no more than 5 hits on each plane track Χ / d.o.f. 5 only one track per event Other specific requirements might be requested depending on the analysis Sensors bump-bonded to CMS pixel readout chip PSI46 digital chip (1x15 μm cell size) 6

7 CMS results: thin-planar before irradiation Requirements single pixel clusters predicted track impact point located μm far from cell edges Pixel cell Fiducial area μm Number of entries Number of entries Data χ / ndf 699 / 69 Width 76.8 ± 6. MPV 814 ± 8. Noise 1115 ± 1.5 Fit: Landau Gaussian Charge (electrons) Data χ / ndf 81.1 / 48 Width ± 5. MPV 614 ± 7. Noise ± 9. Fit: Landau Gaussian Bias: 4 V Thickness: 1 μm p-stop Charge (electrons) Bias: 5 V Thickness: 1 μm p-stop Landau MPV (electrons) μm μm µm, with p-stop Bias Voltage (V) 1 1 µm, no p-stop 1 µm, with p-stop 1 µm, no p-stop Averaging over full sample the ratio MPV@1 μm / MPV@1 μm is ~1.8 (expected 1.) μm 7

Divide Fiducial area Fiducial area μm Pixel cell μm μm efficiency 1.95.9.85.8.75

8 Pixel cell CMS results: thin-planar before irradiation Efficiency scan across cell s divide (long pitch (row direction) and short pitch (column direction)) Efficiency of pointed pixel Efficiency of pointed + closer pixel on same row (col) Divide Pixel cell μm Pixel cell Divide Fiducial area Fiducial area μm Pixel cell μm μm efficiency Bias: 4 V Thickness: 1 μm p-stop long pitch (um) efficiency Bias: 4 V Thickness: 1 μm p-stop short pitch (um) The sensors are fully efficient in both views 8

CMS results: thin-planar irradiation campaign Irradiation performed at Los Alamos with 8 MeV protons Irradiation done after flip-chip assembly Constraints from radiation tolerance of PSI46 digital

9 CMS results: thin-planar irradiation campaign Irradiation performed at Los Alamos with 8 MeV protons Irradiation done after flip-chip assembly Constraints from radiation tolerance of PSI46 digital chip ~5 Mrad Non uniform irradiation Fluence: x115 neq / cm Radiation level measured by in situ dosimetry and cross checked with D MATLAB simulation predictions Finite element analysis to solve Poisson equation Beam halo Electrodes Increasing irradiation fluence End-of-column logic Sensor bulk Weighting potential 9

10 CMS results: thin-planar after irradiation entries (#) Bias: 4 V Fluence: none χ / ndf 81.1 / 48 Width ± 5. MPV 614 ± 7. Noise ± 8.9 Landau distribution before and after irradiation Fluence: ~1x1 15 neq / cm Sensor thickness: 1 μm 1 entries (#) charge (electrons) χ / ndf / Width 66. ± 6.1 MPV 499 ± 8.9 Noise ± 1. Bias: 8 V Fluence: ~1x1 15 neq / cm charge (electrons) Landau MPV (electrons) Before irradiation 15 After irradiation ~1x1 n eq / cm Bias Voltage (V)

11 CMS results: thin-planar after irradiation Number of entries (#) χ / ndf 1.1 / 48 Width 7.6 ± 6.8 MPV 7871 ± 1.1 Noise 171 ± 1.9 Bias: 4 V Fluence: none Landau distribution before and after irradiation Fluence: ~1.x1 15 neq / cm Sensor thickness: 1 μm 11 Number of entries (#) Charge (electrons) χ / ndf 75.1 / 9 Width ± 5.4 MPV 587 ± 9.6 Noise ± 9.4 Bias: V Fluence: ~1.x1 15 neq / cm Charge (electrons) Landau MPV (electrons) Before irradiation 15 After irradiation ~1.x1 n eq / cm Bias Voltage (V)

CMS results: thin-planar after irradiation Efficiency affected by punch through structure (some modules affected before irradiation, all modules affected after irradiation) 1

1 4 5 Bias: 4 V Fluence: none 6 4 4 6 long pitch (um) 1.99.98.97.96.95.94.9.9.91.9 short pitch (um) 5 4 1 1 4 5 Bias: V Fluence: ~1.

12 CMS results: thin-planar after irradiation Efficiency affected by punch through structure (some modules affected before irradiation, all modules affected after irradiation) 1 Thickness: 1 μm Punch through p-stop Thickness: 1 μm Punch through no p-stop short pitch (um) Bias: 4 V Fluence: none long pitch (um) short pitch (um) Bias: 4 V Fluence: none long pitch (um) short pitch (um) Bias: V Fluence: ~1.x1 15 neq / cm long pitch (um) Simulation studies ongoing to optimise geometry/process of bias structure For small pitch pixel design can be critical (i.e. common 4-fold bias dot)

13 CMS results: D before irradiation D columnar, 1 μm thick sensors, 1x15 μm cell size with (E) and (E) junction electrodes Ohmic column 1 μm E 15 μm Junction column Bump pad 1 μm E 15 μm For bias voltage > V E and E are compatible within calibration uncertainty At saturation the collected charge is compatible with planar sensors At low bias voltage (< 15 V) E shows greater charge collection efficiency than E as expected E E 1

14 CMS results: D before irradiation Efficiency map on cell for orthogonal tracks Bias voltage: 4 V 1 μm E 15 μm Bias: V Overall efficiency: (99.4 ±.4)% Visible efficiency deterioration on both junction and ohmic columns 14

CMS results: D before irradiation Charge collection Cell size: 5x5 μm 1E 1 μm thick sensors 1 μm Bump pad Ohmic column Number of entries 5 4 1 Fit: Landau Gaussian χ / ndf.4 / 18 MPV 8799 ± 15.

15 CMS results: D before irradiation Charge collection Cell size: 5x5 μm 1E 1 μm thick sensors 1 μm Bump pad Ohmic column Number of entries Fit: Landau Gaussian χ / ndf.4 / 18 MPV 8799 ± 15.8 Noise 1117 ±.4 Bias: V Charge (electrons) MPV [electrons] Junction column 15 μm Two adjacent readout cells MPV vs bias voltage Charge shared with adjacent pixels which are not being readout Bias Voltage [Volt] 15

CMS results: D before irradiation Charge collection Cell size: 5x1 μm 1E and E and E with bump on column 1 μm thick sensors Junction column Ohmic column Bump pad 5 μm 1 μm MPV [electrons] 1 95 9 85 8

16 CMS results: D before irradiation Charge collection Cell size: 5x1 μm 1E and E and E with bump on column 1 μm thick sensors Junction column Ohmic column Bump pad 5 μm 1 μm MPV [electrons] MPV vs bias voltage 5x1 E 5x1 EBO 5x1 1E (= bump on column) 1 μm Bias voltage [Volt] Two adjacent readout cells bump on column collects ~ e less further studies ongoing 5x1 μm collects ~7 e less than 5x5 same voltage probably sharing with nearby cells not readout 16

17 HC Phase- upgrades, when the ATLAS tracker will be replaced to cope with the higher luminosities. The sensor technology and design are optimized for extreme radiation 16 neq cm ) and pixel layout is compatible with the present FE-I4 chip of ATLAS and 5A. While waiting for a new small pixel cell readout chip, some devices have been h FE-I4 readout electronics. Telescope Planes Module Module Module 1 Test Beamresults: Setup - (CERN SPS H6A, Aconite) m MPP ATLAS D before irradiation conﬁgura?on Telescope Planes (Front) (5X5) (5X1) (5X5) D first batch: 9 wafers produced, best two bumped in Leonardo (former Selex). Wafer 76: 6 inch, 1 µm active thickness. 1µm n+-column depth, with poly-cap, ~1ke are expected for a MIP. Ohmic columns/trenches depth > active layer depth (for bias) Junction columns depth < active layer depth (for high Vbd) Reduction Beam line of hole diameters to ~5 um Holes (at least partially) filled with poly-si (Back) eme direc,on emparature Pixel cells Column onsidering shold and TOT e high D wafer layout Row Mean noise measurements Ref. Ref.Module Module Average noise measurements have been performed (single planar) with ST Control software, recording the (Planar) modules with different pixel size I-V curve for every module and selecting a suitable range of values for bias voltage. sured Aconite and characterized with theh6a SPS beam SPS line, Aconite telescope. Noise vs bias voltage for the three DUTs telescope on CERN H6A beam line tup. This test has been performed for the three 1 GeV pions CIS4-W8-4 (single planar) as reference. types of modules at the three different F1_76_66 pixel planes threshold values, at a fixed ToT tuning A tricky mechanical configuration due to the sub-optimal dependence of pitch, 1BC@1ke. Measurements based on Mimosa6 chip (18.4 μm square pixel shape ofofthe readout boards -HV needs improvements in the future. threshold scans with on, give noise cells, 576 rows and 115 columns) values related to bias voltage selected. Tilt angle iscoordinate not very accurately controlled (around 5 in row direction) ~ μm resolution on each Results showed are related to the threshold at 5e. Acquiredset dataset: As presented the graph, modules of, Three DUTs 5x5 1E, 5x1 1E, andin 5x5 E μm pixel are slightly 5x5µm and 5x5µm cycles of HV scan at a fixed tuning (coarse/fine steps) F1_76_8 1 μm thick + planar module as reference (15 thickthis device. less noisy than the 5x1µmμm setatlas of tuning variation (threshold, ToT). result isfe-i4 consistent measurements sensor) bump-bonded1 to chip with performed at threshold values of e and Hideyuki Oide e. Mauro Dinardo, Universita` degli Studi di Milano and INFN Threshold à 5e, ToT àbicocca 1BC@1ke 17 TestBeam Schema,cs

18 ATLAS results: D before irradiation Efficiency maps vs bias voltage (thr = 15 e, ToT = 1 BC / 1 ke ) 5x5µm 5x5µm 1x5µm V Efficiency 5V 1V 15V 18

Pixel Hit Hit Map Map vs. vs. Cluster Cluster size size 5x5 5x5 (F1-76-6) (F1-76-6) 5x5 Pixel Pixel Hit Map vs.

cellsmasked on 1 out of 4 sides ineﬃcient 5x5 μm cells have neighbouring cells on 4 out of 4 sides ineﬃcient ineﬃcient Masking is necessary to compute

8 Hideyuki Oide Sharing Sharing Sharing Efficiency size=1 size=1 size=1 Hit Efficiency Efficiency for size 1 clusters Inefficiency map ATLAS results: D

7 Efficiency for size clusters 1 17--1.99 5 1 15 HV [V] size= size= size= Leaking Leaking Leaking.6.

to<~1 5x5 μm: almost 5x5µm flat at ~98% efficiency for~1 V.VRamping < HV up bias 15V.

19 Pixel Hit Hit Map Map vs. vs. Cluster Cluster size size 5x5 5x5 (F1-76-6) (F1-76-6) 5x5 Pixel Pixel Hit Map vs. Cluster size 5x5 (F1-76-6) Due to chip-sensor pitch mismatch 5x5 μm cells have neighbouring cells on out of 4 sides 5x1 μm cells have neighbouring cellsmasked on 1 out of 4 sides ineﬃcient 5x5 μm cells have neighbouring cells on 4 out of 4 sides ineﬃcient ineﬃcient Masking is necessary to compute unbiased efficiency 1x5 mask for 5x1 μm sensor Masked HV Scan: Hit efficiency (after masking) Hideyuki Oide Sharing Sharing Sharing Efficiency size=1 size=1 size=1 Hit Efficiency Efficiency for size 1 clusters Inefficiency map ATLAS results: D before irradiation Efficiency for size clusters HV [V] size= size= size= Leaking Leaking Leaking µm Threshold = 15e µm Sharing Sharing ToT = 1BC/1ke Sharing µm Efficiency vs bias voltage HV [V] : Efficiency is >99% above up to<~1 5x5 μm: almost 5x5µm flat at ~98% efficiency for~1 V.VRamping < HV up bias 15V. V Leaking 5x5µm : Keeping almost flat ~98% efficienty in < HV < 15 V. Slight increasing (slight increase of ~1%) Leaking Similar story for 1x5 Leaking similar 1x5µmsimilar : shown just for reference. Qualitatively trend to 5x5µm. Similar story for 1x5 5x1 μm: qualitatively trend to 5x5 μm Similar story for 1x x5 μm : efficiency > 99% above ~1 V (ramping up to ~1 V) 1 Bias: 1 V Hideyuki Oide Column Side [μm] Hideyuki Oide Cell size: 5x5 μm Hideyuki Oide Hideyuki Oide 17--1

20 T [BC] Clusters ATLAS results: D before irradiation Peak at ~875 e (15 um thick) 14ke 5µm 1µm Charge collection at bias voltage = 1 V Average: 5.6 RMS:.9 Cluster ToT [BC] Clusters ke 5µm 5µm Peak at Cluster ~7 ToT [BC] e Average: 7.1 RMS:. Clusters Reference Planar 1 5µm 5µm T [BC] Peak at ~8 e Clusters 6 5 1ke Cluster ToT [BC] Average: 9. Hideyuki Oide Clusters Peak at ~7 e 5 1ke Cluster ToT [BC] Average: RMS:.6 RMS: m 1 5µm 1µm 5 5µm 5µm ot [BC] Cluster ToT [BC] Cluster ToT [BC] Collected charge MPV of D modules ~ as expected from sensor thickness 14ke Average: 5.6 RMS:.9

21 Summary and prospects First prototype sensors, both thin planar and D columnar, developed within INFN Pixel R&D collaboration with partnership FBK, show good data quality and behave as expected Further studies, especially at higher radiation dose, are ongoing CMS D and thin planar modules will be irradiated ~summer 17: with 4 GeV protons at CERN up to 5x1 15 neq / cm with neutrons at Lubiana up to 1 16 neq / cm CMS testbeam campaign foreseen ~fall 17 ATLAS D modules are being irradiated up to ~1 16 neq / cm : 6 modules with 4 GeV protons at CERN modules with MeV protons at KIT New wafers made with Direct Wafer Bonding (DWB) with SOI are being processed and will be tested in comparison with SiSi DWB As soon as RD5 chip (joint ATLAS-CMS collaboration to develop high radiation tolerant readout chip in 65 nm CMOS technology) will be available the plan is to bump-bond new sensors with RD5 chip (higher radiation resistance, capability to readout both 5x5 μm and 5x1 μm cells) 1

22 Backup

INFN Pixel R&D: main design features Devices single sided processed: it can be thinned Pixel Unit Cell design with variety of collecting electrodes Location of pad for bump-bonding optimised 1-1 μm

23 INFN Pixel R&D: main design features Devices single sided processed: it can be thinned Pixel Unit Cell design with variety of collecting electrodes Location of pad for bump-bonding optimised 1-1 μm Standard pixel prototype 1x15 μm pitch Readout by PSI46dig chip (phase-1) 1 μm active thickness Small pitch pixel prototype 5x5 μm and 5x1 μm cell size Readout cell adapted for PSI46dig chip (phase-1)

24 INFN Pixel R&D: main design features After column filling with poly-silicon, residual remains outside the columns (on the surface) Two possibilities for removal with mask: a poly-cap remains on the columns without mask: entirely removal of the poly-cap some consequences in damaging the aperture of the columns (i.e. higher leakage current)! 4!

25 INFN Pixel R&D: main design features Full depletion voltages Vfull dep ~16V for 1 μm thick Vfull dep ~V for 1 μm thick 5

26 INFN Pixel R&D: main design features We performed a simulation of the charge transport with MATLAB: D simulation velocity dependence from electric field taken into account space charge corresponding to nominal/un-irradiated resistivity e-h pairs per micron: 6 radiation damage effects simulated with finite carrier lifetimes from carrier lifetimes we evaluate irradiation fluence: 1/τe/h = βe/h Φ (16 JINST 11 P4) (in the following the errors on fluence are derived only from propagation of error on βe/h) Electrodes Sensor bulk Finite element analysis to solve Poisson equation The irradiation was not uniform we concentrate our modelling on a small region in the highly irradiated zone Weighting potential 6

27 CMS results: D before irradiation Column θ Charged particle Sensor bulk L LC w=1 μm L1 { L = L1 + L + LC L cos(θ) = w LC sin(θ) = d LC / L < Threshold / MIP d=5 μm tan(θ) > (d MIP) / (w Threshold) 7

28 CMS results: D before irradiation Efficiency map on cell for orthogonal tracks Bias voltage: 4 V 1 μm E 15 μm Overall efficiency: (99.19 ±.5)% Visible efficiency deterioration on both junction and ohmic columns 8

CMS results: D before irradiation Efficiency map on cell Bias voltage: V 1 μm E 15 μm Short pitch (μm) Orthogonal tracks Short pitch (μm) short pitch (um) 5 4 1 5 degrees Cell efficiency Duttilt 1

29 CMS results: D before irradiation Efficiency map on cell Bias voltage: V 1 μm E 15 μm Short pitch (μm) Orthogonal tracks Short pitch (μm) short pitch (um) degrees Cell efficiency Duttilt 1 degrees Cell efficiency Dut tilt Short pitch (μm) short pitch (um) Long pitch (μm) long pitch (um) Long pitch (μm) long pitch (um) Long pitch (μm) Angle (degree) Efficiency E Efficiency E Column inefficiency shadowed by tilt, compatible with simple geometric considerations (tilt angle > 9 o, see backup slides) 9

Very high average signal efficiency Significant variations of

30 ATLAS results: D simulation Simulated signal efficiency Simplified simulation domain (D slice), no pixel edge effects Very high average signal efficiency Significant variations of signal efficiency with hit position Possible impact ionisation effects at high field

Pixel sensors with different pitch layouts for ATLAS Phase-II upgrade

Pixel sensors with different pitch layouts for ATLAS Phase-II upgrade Different pitch layouts are considered for the pixel detector being designed for the ATLAS upgraded tracking system which will be operating