Phase 1 upgrade of the CMS pixel detector, INFN & University of Perugia, On behalf of the CMS Collaboration. IPRD conference, Siena, Italy. Oct 05, 2016 1
Outline The performance of the present CMS pixel detector. Schedule for the LHC and the pixel detector upgrade. The reason and concept behind the upgrade of present pixel detector. Comparison between the current and upgraded pixel detector geometries. The improvements in the performance for the upgraded detector. Present status of pixel phase 1 upgrade and next steps for the future. 2
Barrel Pixel (BPIX): 3 layers, 48 M pixels R=4.4, 7.3 and 10.2 cm. Forward pixel (FPIX): 2+2 disks, 18 M pixels Z=34.5, 46.5 cm. n+-in-n sensors with pixel size of 100x150 μm2 (rφ-z). Present pixel detector The detector performed well in Runs 1 and 2 of the LHC Excellent resolution: 10μm (r-φ), 20-40μm (z) High efficiency (> 99%). Plays a crucial role in tracking, vertex finding, b-tagging, µ, τ reconstruction and e/γ separation. 3
Reasons behind upgrade Present pixel detector is designed to cope up with instantaneous luminosity of 10 34 cm -2 s -1. Instantaneous luminosity already exceeded 10 34 cm -2 s -1. While we expect 2x10 34 cm -2 s -1 and 300 fb -1 by the year 2022. This exceeds the capability of the present Readout chips (ROC). Dynamic inefficiencies and dead time caused insufficient readout, limited size of bandwidth affects the detecor performance when the instatntaneous luminosity reaches close to 2x10 34 cm -2 s -1. Simulated tt events with PU 2x10 34 cm -2 s -1 @ 25ns, PU 50 4
Upgraded pixel detector 4-hit layout to cover maximum η range up to 2.5, with first layer closer to the primary vertex, more pixels, BPIX 48M 79M, FPIX 18M 45M. First layer closer to the interaction point with new beam pipe of smaller radius (30 22.5 mm) Gain : Improved vertex resolution and b-tagging efficiency. Upgrade 16.0 cm 10.9 cm 6.8 cm 2.9 cm 4.4 cm 7.2 cm 10.2 cm Current BPIX FPIX 5
More layers but less material With the insertion of new layers and disks, number of pixels are almost doubled. Material budget is less: Moving passive materials like electronic connections etc., out of tracking volume. Lighter mechanical support. Lightweight support for two-phase CO 2 cooling. Radiation length Interaction length 6
CMS pixel system FPIX Service cylinders and BPIX supply tubes host power, readout electronics with optical converters and cooling lines BPIX FPIX service cylinder FPIX BPIX Supply tube Move electronics away from interaction point Reduced mass in tracking volume Space for more extensive FPiX system Require long (~1m) cables to read out modules 7
New Read-Out Chip design New readout chip with same architecture based on PSI46roc: 40 MHz analog 160 Mbit/s digital Increase of hit (32 80) and time stamp (12 24) buffer depth Additional readout buffer Smaller cross-talk, improved comparator Threshold reduced from 3200 e- to 1800e-, Better efficiency, resolution and radiation hard Good for L2 to L4 for the pixel detector. 8
Pixel Modules Pixel modules contains Sensor, ROCs, HDI with token bit manage(s) and readout cable, 66K pixel/module. Twisted cables The number of modules required: 1856 for BPIX 672 for FPIX TBM HDI ROC BPIX module FPIX module Sensor Base Strips 9
Module qualification The test procedure, grading criteria are all the same over all the production centers. Detector modules are tested at +17 and -20 C in cold box with 10 temperature cycles. Measure IV and grading criteria based on leakage current, bump bond defects, number of dead pixel, noise. Test with X-ray: Check performances at high rate Absolute energy calibration with different fluorescence sources. 10
Module production status Enough module available for L3 and L4. Spares are still being continued. Production of L2 modules almost complete. L1 modules to be completed by mid-october. FPIX production complete, qualification will end in October. BPIX: L1+L2: Switzerland L3: CERN+Finland+Taiwan / Italy L4: Germany production centers FPIX: USA 11
Mechanical support Barrel Lightweight support structure made from CFRP/Airex compund for barrel. Cooling tubes are the backbone of the structure. L1 and L2 half shells are ready with cooling pipes embedded Module mounting started for L2 L3 and L4 half shells are in preparation. Supply tubes are ready and electronics installation and testing ongoing. 12
Mechanical support Forward Service half-cylinders are completed. All the half disks are completed with tubing embedded, and 482 of the total 672 modules are already installed. The half disks installed in 2 half cylinders. After testing, disks will be removed for the transport to CERN. 13
Power system and DAQ The upgraded detector will have 1.9 times more channel, power loss in cables are factor 4 larger. Present power supplies and cables can not power the upgraded detector. Use of DC-DC converters, with power efficiency more than 80% and good performance. 1200 DC-DC converters to be used in total. DAQ developement helped by availability of a 0.5% system already installed in CMS Pilot System. Even with limited number of channels the operational experience under realistic conditions are gained. It helped spotting some problems. 14
Summary The pixel phase 1 upgrade project targets to build a light-weight, radiation hard detector to replace the current one. With the extra pixel layer better performance is expected in every aspect like tracking, b-tagging, vertex resolution. The upgraded detector can work with high PU and luminosity. Detector installation in extended end of year shutdown with a minimal impact on the data taking. Looking forward for the installation and commissioning of new pixel detector during Feb Mar 2017. 15
Thank you! For more details see CMS TDR for pixel upgrade: http://cds.cern.ch/record/1481838?ln=en 16
BACK-UP SLIDES 17
CMS Luminosity for Run 1 and 2 The instantaneous luminosity reaching the peak luminosity for which CMS pixel was designed for! Peak luminosity achieved 1.5x10 34 cm -2 s -1. 18
ROC design in digital version The changes in the read-out-chips in the digital version, shown at pixel unit cell level. 19
Sensor technology and ROCs Same design technology (n+-in-n) as used in the present detector. Pixel size 100µm x 150µm with thickness of 250 µm. p-spray (BPIX) or p-stop (FPIX) for n-side isolation. While sensor technology remains the same, we need to change the design of ROCs due to huge data loss due to dynamic inefficiencies. Data loss in present pixel at Layer1: 4% data loss at design luminosity. 22% data loss at 2 times the design luminosity. 50% data loss @ 50 ns bunch spacing! Require upgraded ROC and improved readout chain. 20
Pixel Modules Upgraded BPIX module components for Layer 1 and Layers 2 4. Layers 2-4 TBM Twisted cables HDI Sensor ROC Base Strips 21 Layer 1
Special arrangement BPIX L1 Reduction of mean radius of L1 from 4.4 to 2.9 cm. Inner-most layer of the upgraded pixel detector will be exposed to very high particle flux 580 MHz cm -2. Modified ROC design for L1 (PROC600) with additional increased buffer: Requires additional space on the periphery, chip 3% larger Column drain cluster algorithm, transfer of 2x2 pixel cluster instead of individual hits. New chip can sustain 600 MHz cm -2. Radiation hard to 500 Mrad. Two Token Bit managers (TBMs) per module are required to transmit data from ROC. 22
Activities in different places in Italy Electrical connections for bare modules are tested at Pisa. Shear and pull test of the bump bonded chip at Pisa. ROCs are tested after dicing at Padova. TBM is glued to HDI and they are tested for electrical connections at Catania. Assembly of HDIs with bare module is done at Bari, and preliminary tests are done there. The final qualification tests with thermal cycle, X-ray calibration are done at Perugia, the test results are uploaded to data-base hosted at Pisa : http://cmspixelprod.pi.infn.it/ This is an example related to the Italian consortium, other consortia have a different organization. The final assembly of the pixel detector will be done at PSI and the installation during the extended winter shutdown in 2016-17. 23
Radiation dose for 4 layers Layer 1 : 1 MGy Layer 2 : 400 kgy Layer 3 : 200 kgy Layer 4 : 130 kgy 500 fb -1 100 kgy = 10 Mrad Fluence wise: 3 x 10 13 cm -2 MIPS ~ 1MRad 24
PERFORMANCE PLOTS 25
Tracking efficiency, fake rate Track efficiency and fake rate for tt-bar monte-carlo sample. The efficiency is better at lower fake rate for the upgraded pixel detector. 26
Tracking efficiency for Muon The tracking efficiency gets much better for muon enriched montecarlo samples, with negligible fake rate. Improves physics potential a lot. Current detector Upgraded detector 27
Primary Vertex resolution Primary Vertex resolution is estimated in tt-bar monte-carlo Becomes more and more important specially at high PU. The PU 50 scenario is better than the current detector in the same condition. 28
b-tagging performance Performance of the Combined Secondary Vertex b-tagging algorithm depends on track impact parameter and primary vertex resolution. Jets with pt > 30 GeV is considered for tt-bar monte-carlo sample for current and upgraded detector scenario. For upgraded detector PU of 50 is even better than current detector with 25 PU. 29
b-tagging @ <PU> = 100 Efficiency gain as well as lower fake rate. Efficiency gain Fake rate reduction 30