The CMS Pixel Detector Upgrade and R&D Developments for the High Luminosity LHC On behalf of the CMS Collaboration INFN Florence (Italy) 11th 15th September 2017 Las Caldas, Asturias (Spain)
High Luminosity LHC schedule Nominal luminosity = 1x1034 cm-2s-1 CMS Phase II Tracker TDR CERN-LHCC-2017-009 Start of the CMS Phase II upgrade Installation of Phase I pixel detector 2
The HL-LHC numbers Peak instantaneous luminosity 5x1034 cm-2s-1 (max 7.5x1034 cm-2s-1); ~300 fb-1 (400 fb-1) of integrated luminosity per year delivered to ATLAS and CMS; Centre-of-mass energy of 14 TeV; 25 ns bunch spacing; 140 (200) pileup events on average; Hit rate of 3 GHz/cm2 in the inner layers; 2.3x1016 1 MeV neq/cm2 fluence and 12 MGy total ionizing dose in the innermost region of CMS. Challenging radiation environment, especially for the CMS inner tracker 3
Tracker upgrade requirements Radiation tolerance Full efficiency up to 3000 fb-1 Easy to be pulled out for maintenance New pixel sensors Increased granularity Occupancy around per mille level in the inner tracker Reduced material budget The amount of material affects the tracker performance Extended tracker acceptance in the forward region Readout electronics and services New inner tracker layout Improve the physics capabilities Good two track separation To be improved especially for track finding inside highly energetic jets Robust pattern recognition under high pileup conditions More difficult and time consuming track reconstruction Input to the Level 1 trigger Use the tracker information as input to the L1 trigger 4
Tracker layout Outer tracker See J. Luetic's talk on Tuesday Inner tracker Equipped with pixel modules. Thin silicon pixels of size of 25x100 μm2 or 50x50 μm2 are expected to have the desired resolution and radiation resistance. Total active surface of ~4.9 m2. 5
Inner tracker layout Green: 2 ROCs modules Orange: 4 ROCs modules Extended forward coverage Barrel detector with 4 layers (TBPX) First layer r = 2.9 cm Fourth layer r = 16 cm Max z = ±20 cm (±27 cm in Phase 1) 8 small double-discs per side (TFPX) First disk z = ±25 cm Eighth disk z = ±140 cm Simple mechanical structure 4 large double-discs per side (TEPX) First disk z = ±175 cm Fourth disk z = ±255 cm Simple installation and removal to allow for potential replacement of inefficient parts 6
Pixel modules The pixel module is the basic unit of the Inner Tracker It comprises a pixel sensor, ROCs, flex circuit, and mechanical support. Only two types of modules: 1x2 (2 ROCs) and 2x2 (4 ROCs). Sensors are bump-bonded to the ROCs. The flex circuit is glued onto the sensors and wire-bonded to the ROCs. Modules mounted on the TBPX. CO2 cooling pipes to remove the heat generated on the module and keep the temperature at -20 C. ~1 W/cm2 power dissipated by the ROCs. 7
Pixel sensor design parameters n-on-p module currently n-on-n Main challenge: prevent sparks between ROC and sensor Possible solutions: Coating of ROC wafer and sensors with a BCB layer Post-process deposition of a Parylene layer on the full module Small thickness (100-200 μm) currently 285 μm Before irradiation: smaller signal and prone to bow After irradiation: higher charge collection efficiency but exhibit a soft breakdown behaviour Small pitch pixel cells (2500 μm2) with rectangular (25x100 μm2) or squared (50x50 μm2) shape currently 15000 μm2 (100x150 μm2) Main challenge: limited space for p-stop and bias structures Alternatives: p-spray, temporary metal bump bond pads n+ implants contacts metal layers p-stops HPK submission designs 8
25x100 μm2 VS 50x50 μm2 25x100 μm2 vs 50x50 μm2 comparison using CMS simulation (barrel only) 50x50 μm2 25x100 μm Square pixels prone to cluster breakage for higher thresholds 2 Empty (full) markers = 100 (150) μm thickness 1000/1500/2000 electrons threshold Square pixels increase the bandwidth requirement at the barrel edge 50x50 μm2 25x100 μm2 9
Radiation hardness measurements Thin planar pixel sensors are the baseline for the outer TBPX layers (2-4) and TFPX rings (2-4), and all TEPX rings. 3D sensors are being considered for the inner layers. Radiation hardness studies limited to fluences of ~2x1015 neq/cm2 by the current available ROC (PSI46dig) Studies at higher fluences done using strip sensors or pad diodes Wire bonded to the ROC after irradiation Epi: Epitaxial strip sensors MCz: Magnetic Czochralski strip sensors 100 μm thick sensors collect the same charge as 200 μm ones after 1.3x1016 neq/cm2 irradiation but at lower bias voltage. 10
Planar VS 3D sensors 3D and thin planar sensors are promising candidates for the high radiation environment of the inner layers/rings. PLANAR 3D n+ p+ PROS: PROS: Cheaper Smaller bias voltage Mass production possible Smaller charge drift path Less leakage current than thick sensors CONS: CONS: Higher capacitance Thin sensors prone to bow Slightly higher cost Complex fabrication 11
Planar pixel submissions for R&D Punch through Several R&D submissions to mainly study: Pixel bias scheme: punch-through, polysilicon resistors, no bias scheme at all; Pixel isolation: p-stop versus p-spray; Metal overhangs: mitigate large electric fields at the pixel edges. Including sensors compatible with: PSI46dig, RD53A (RD53 p-stop collaboration), ROC4Sens (PSI) and FCP130 (FNAL). Metal bias grid Hamamatsu Photonics (HPK) 6 n-on-p wafer, 150 um active thickness Goals: pixel geometry, p-stop VS p spray, bias scheme, metal overhang HPK wafer INFN-FBK 6 n-on-p wafer, Si-Si DWB, 100-130 μm active thickness Goals: Small pitch pixels, punch through, spark protection with BCB, different wafer thinning procedures Bump bonding at IZM Berlin (SnAg) and Leonardo Rome (Indium) SINTEF n-on-n, 300 μm active thickness Goals: slim edge, slim pixels 25x600 μm2 INFN-FBK wafer 12
3D pixel submissions for R&D INFN-FBK n-on-p, 100-130 μm active thickness, single sided DRIE process Goal: small pitch pixels on thin substrate, different electrodes layouts Trento (Italy) CNM n-on-p 200-230 μm active thickness (double sided), 150 μm active thickness (single sided) Goal: Small pixel sizes, small aspect ratios, slim edges Spain 13
Status of INFN-FBK R&D Thin (130 um) n-on-p planar sensors with 100x150 μm2 pixel size, bump bonded to the PSI46dig ROC. Studied in a test beam at FNAL with 120 GeV protons Collected charge measured before and after irradiation 1.2 x 1015 neq/cm2 irradiation performed with 800 MeV protons at Los Alamos. Before irradiation After irradiation 40V bias T=+25 C 300V bias T=-20 C Thin planar sensors work as expected Charge collection efficiency comparable to that obtained with 100 μm strip sensors 14
Status of INFN-FBK R&D 100x150 μm2 3D sensors prove the feasibility of the production process Different layouts of the pixel cell Tested at test beams performed at the FNAL facility 130 μm active thickness 100x150 μm2 pixel size 30V bias voltage - T=+25 C Cell hit efficiency for 0 degrees incident tracks Cell hit efficiency for 5 degrees incident tracks All the cells are superimposed in the plots. Inefficiencies near the readout and ohmic columns for 0 degrees tilting angle Full efficiency (>99%) recovered already at 5 degrees 2 junction columns Bump pad on top of the junction column 15
Status of INFN-FBK R&D Small pitch 3D sensors proved to work as expected. Read out with PSI46dig only few channels connected to ROC, other channels grounded together. Irradiation of these modules with 24 GeV protons at the CERN IRRAD facility is ongoing. Three configurations tested: 1E: only one junction column 2E: 2 junction columns 2E BO: 2 junction columns and bump pad on top of one column 2E BO 40V bias voltage Effect of the charge sharing with not read out adjacent pixels No striking difference between the collected charge in the three 25x100 μm2 3D configurations. Need to test after irradiation 16
Status of SINTEF R&D Small pitch planar sensors, n-on-n, 300 μm thickness, 25X600 μm2 pixel size. Special bump bond pattern to match the PSI46dig layout. Test beam at FNAL showed the feasibility of fine pitch cells. Inactive edge of pixel sensor causes a hit finding inefficiency in layouts without modules overlap (e.g. TBPX along z). Slim edge 100x150 μm2 n-on-n prototypes produced by SINTEF distance from last pixel to dicing edge reduced to 210 μm 1150 μm in Phase I Charge reduction near the edge of the last pixel but significant decrease of the inactive area 17
Status of CNM R&D Special small pitch 3D sensors with rows of small pixels (25x100 μm2), 230 μm thickness The sensor has 128 strips, each consisting of 75 pixels shorted together. The sensor can be irradiated before bonding it to the ROC. ALIBAVA readout system used. Irradiation with 24 GeV protons to 5.7 x 1015 neq/cm2 Measurements performed at CERN test beam with 120 GeV protons/pions. Charge collection efficiency at 180 V is 94% of that of a nonirradiated device at 30V 18
Pixel readout chip 250 nm 8x10 mm2 100x150 μm2 Block diagram of the pixel ROC 400 MHz/cm2 1x1015 neq/cm2 See R. Beccherle's talk on Wednesday ~0.15 W/cm2 No opto-electronic device able to withstand the expected radiation level in the inner layers CURRENT DETECTOR Solution: remote low-power GBT cards (LpGBT) placed in the service cylinder (at r=20 cm) and connected to the modules via e-links. Modularity: Number of e-links matches the output rate of the module (depends mainly on the module position). 19
Services A service cylinder placed between the inner and outer tracker will house: LpGBT opto-electronic modules E-links from pixel modules to LpGBT Power cables Cooling pipes Material budget must be kept as low as possible Different options under investigation for the e-link cables and cooling pipes Required pixel modules power ~40 kw 65 nm technology requires low supply voltage (1.2 V) and high currents (2.2 A/chip) Traditional powering systems cannot be used due to high current, material budget and space constraints. Viable option: serial powering scheme across the pixel modules Intrinsically low material budget Negligible voltage drops Serial powering scheme demonstrated to work as expected using ATLAS FEI4 pixel chip. 20
Conclusions HL-LHC environment is highly demanding for the CMS Phase II pixel detector New layout with extended forward coverage Simple mechanics and easy installation Thin planar sensors (3D in the inner layers?) Small pitch pixels Radiation resistant ROC Several R&Ds ongoing to test the performance of new pixel detectors Planar: HPK, INFN-FBK (together with ATLAS), SINTEF 3D: INFN-FBK (together with ATLAS), CNM First studies show promising results R&D programs to study material budget and serial powering scheme. Phase II tracker TDR is now public: CERN-LHCC-2017-009 21
BACKUP
Physics motivations Weak vector boson scattering SM Higgs Yukawa couplings to per cent level precision H μμ Higgs boson self-coupling And many others: Increased discovery potential for BSM scenarios Indirect BSM searches through measurement of rare decays 23
Inner tracker layout TBPX layout Split in 2 parts along z of 4 and 5 modules. No projective acceptance gap at η~0. No modules overlap in z. Green: 2 ROCs modules Orange: 4 ROCs modules TFPX and TEPX layout Planar geometry with modules orthogonal to beam axis (no turbines nor blades). Double discs provide an hermetic detection plane. Overlaps both in r and r-φ 24
Forward discs structure Each double-disc is split in two half-discs, called dees 2 ROCs modules used at inner radii and 4 ROCs modules at higher radii A full TFPX double-disc holds 108 modules arranged in four discs. Embedded thin titanium cooling tubes carry two-phase CO2 at high pressure to keep sensors and ROCs at low temperature (~ -20 C). LpGBT modules mounted close to the outermost radii of the double-disc for TFPX and TEPX. 25
n-on-p pixels n-on-p: Several vendors Cost effective n+ E n-on-n: High cost for the production of thin double-sided sensors EE E WW e h p+ n-bulk p-bulk EW E p-on-n: Electrons have higher mobility than holes, advantage to collect electron at high weighting fields (EW ) Larger noise observed in p-on-n sensors n-in-p p-in-n 26
Material budget Phase 1 tracker Phase 2 tracker 27
Hit occupancy 28