TThe Upgrade of the CMS Tracker at HL-LHC
Alessandro L a R osa on behalf of CMS Tracker collaboration CERN, Geneva 23, CH-1211 SwitzerlandE-mail: [email protected]
In the high luminosity scenario of the LHC (HL-LHC), which will bring the instantaneous luminosityup to 7.5 × cm − s − , ATLAS and CMS will need to operate at up to 200 interactions per 25 nsbeam crossing and reaching up to 4000 fb − of integrated luminosity. To achieve their physics goalsthe experiments will need to improve the tracking and vertexing capability and the ability to selec-tively trigger on specific physics events at reasonable thresholds. The upgrade of the CMS Trackerrequires designing new inner and outer tracking detectors to cope with the increased luminosity andto implement first trigger level functionality. This paper describes the new layout and the technolog-ical choices together with some highlights of research and development activities. KEYWORDS:
HL-LHC tracking detectors, silicon pixel detectors, silicon strip detectors
1. Introduction
The High-Luminosity Large Hadron Collider (HL-LHC [1]) at CERN is expected to collide pro-tons at a centre-of-mass energy of 14 TeV and to reach the unprecedented peak instantaneous lumi-nosity of 7.5 x 10 cm − s − with an average number of pileup interactions up to 200. This will allowthe ATLAS [2] and CMS [3] experiments to collect integrated luminosities of 3000 fb − during theproject baseline lifetime, with the possibility to reach 4000 fb − as ultimate scenario. To cope withthis very challenging scenario the CMS detector will be substantially upgraded [4] before the startof the HL-LHC. The CMS tracking detector will have to be replaced in order to fully exploit the de-livered luminosity and cope with the demanding operating conditions. The new detector will providerobust tracking as well as input for the first level trigger (L1).The CMS Phase-2 tracker will consist of about 200 m of silicon modules and will be composedof two sub-detectors: the Inner Tracker (IT) made of silicon pixel modules and the Outer Tracker(OT) made of a combination of silicon modules with strip and macro-pixel sensors. A longitudinalview of one quarter of the new detector layout is shown in Fig. 1.The main requirements for the new tracker can be summarised as follows: radiation tolerance tobe fully e ffi cient up to the expected integrated luminosity; increased granularity to ensure excellenttracking performance in the presence of a high level of pileup (occupancy less than 1% for the OTand 0.1% for the IT); reduced material in the tracking volume; contribution of tracking informationto the L1 trigger (only for OT); large readout bandwidth and deep front-end bu ff ers for higher rate(750 kHz) and longer latency (12.5 µ s) of the L1 trigger system; extended tracking acceptance up to | η | ∼ ffi cient pile-up mitigation and better track reconstruction in the forward region.
2. Inner Tracker system
The Inner Tracker consists of four barrel layers (TBPX) plus eight small disks (TFPX) and fourlarge disks (TEPX) per side, covering a surface of 5 m with 3900 hybrid pixel modules featuringabout 2 billion readout channels. In addition to extending the tracking acceptance up to | η | ∼
4, the a r X i v : . [ phy s i c s . i n s - d e t ] F e b ig. 1. Sketch of one quarter of the CMS Phase-2 Tracker layout in r-z view. In the Inner Tracker the greenlines correspond to pixel modules with two readout chips (ROCs) and the yellow and brown lines to pixelmodules with four ROCs. In the Outer Tracker the blue and red lines represent the two types of modules,equipped with macro-pixel-strip and strip-strip sensors, respectively.
TEPX disks are employed to perform real time luminosity measurements. The innermost rings ofthe outermost TEPX disks are exclusively used for beam background and luminosity measurementswith independent readout and control systems. The expected hadron fluences and total ionising dosefor the innermost pixel layer are 3.5 × n eq cm − and 1.9 Grad, respectively. A replacement of theinnermost pixel layer and the innermost ring of the TFPX will be required during the long shutdown(LS) number 5 of the current HL-LHC schedule, irrespective of which sensor technology will beselected.The basic units of the IT detector are the hybrid pixel modules. As shown in Fig. 1, two types ofpixel modules will be employed in the detector: 1 × × Fig. 2.
Sketch of a 2 × An extensive silicon pixel sensors R&D for qualifying sensors capable to operate up to the10 n eq cm − range is currently on-going and 150 µ m thick n-in-p planar sensors are the current aseline for the IT detector. In order to guarantee a consistent charge collection e ffi ciency after irra-diation the technology requires a high bias voltage up to ∼ × µ m , as current baseline, and 50 × µ m .Initial studies show that, the relative di ff erence in term track parameter resolution between the twodesigns is rather small with a trade-o ff between primary vertex discrimination and resolution on theimpact parameter. For the innermost layer, square pixels would result in a very long cluster whichwould set more stringent requirements on the operational threshold of the chip, and it would alsodemand a larger bandwidth for reading the data.The readout chip for the ATLAS and CMS pixel detectors at HL-LHC is under design by theCERN RD53 collaboration [5]. The chip, designed in 65 nm CMOS technology, has been developedwith a cell size of 50 × µ m , a low threshold (below 1000 e − ), a high data rate (one 160 Mb / sinput link and up to four 1.28 Gb / s output links) and serial powering capability. The first half-sizeprototype chip, RD53A [6], featured a 400 ×
192 pixel matrix and has been used for extensive R&Dand qualification programs by ATLAS and CMS. Radiation hardness up to 500 Mrad was provenwith indications that operation up to 1 Grad would be possible under controlled conditions, mostimportantly by cooling the chip during the full lifetime. In the RD53A chip, three di ff erent analoguefront-ends (FE) are prototyped in three sub-matrices. These are the synchronous FE, linear FE anddi ff erential FE. After a dedicated review process CMS has selected the linear FE for the final chip.The common design framework of the final chip, known as RD53B, contains design improvementsand a few fixed bugs identified in the RD53A. A dedicated overview of the qualification results isgiven in [7]. The CMS final chip size is 16.8 × with a matrix of 336 ×
432 pixels. Thesubmission of the first full size chip prototype for CMS is expected at the end of 2020.
Fig. 3.
Typical ENC distribution in electrons of an individually powered readout chip, e.g. RD53A, (left)and per-pixel di ff erence after inclusion in a serially powered chain (right) [8]. The extreme rate requirements for the readout chip necessitate the use of a CMOS technologywith low supply voltage ( ∼ ∼ arallel with a constant voltage is not possible, due to the voltage drop along the cables and the pro-hibitively large cable cross section that would be required. Similarly point-of-load-DCDC conversioncannot be employed due to space constraints. The only viable powering scheme for such an environ-ment is serial distribution. In this scheme 8 to 12 modules are arranged in a chain. All chain elementsreceive the same current and the voltage is equally shared if all elements represent the same and con-stant load. This is possible due to the ShuntLDO implementation in the readout chip that combinesa linear regulator (LDO) and a shunt. A picture of a serially powered chain of three digital modulesis shown in Fig. 2, while Fig. 3 shows the noise distribution of a module in standalone operation andthe chip by chip di ff erence after inclusion in the serial chain [8].The IT modules will be connected with up to 1.6 m electrical links to optical-modules locatedat the periphery of the detector. The optical module hosts two LpGBT [9] transceivers and twoVTRx + [10] optical links. In total six 1.28 Gb / s up-links per module will be implemented for dataand monitoring information, and one 160 Mb / s down-link for bringing clock, trigger, fast commandsand configuration data to the module.
3. Outer Tracker system
The Outer Tracker consists of six barrel layers and five endcap disks per end, and it is subdividedin: TB2S (Tracker Barrel with 2S modules), TBPS (Tracker Barrel with PS modules) and two TEDDs(Tracker Endcap Double Disks). The OT covers a surface of 190 m with 13’200 modules featuringabout 213 million of readout channels.The modules have the ability to autonomously select track segments (stubs) above a selectedp T threshold and to send these to the backend electronics. The selection relies on the bending of thecharged particles in the magnetic field and a programmable selection window in the readout chips (seeFig. 4). The backend track finder system receives the stub data from the individual detector modulesand performs track finding in two steps, pattern recognition and track fitting, and sends the final tracksto the L1 trigger [4]. Fig. 4.
Sketch of the track stub finding principle (left) and the p T module concept (right). A track passes bothsensors of a module. A low momentum track falls outside the acceptance window and produces no stub [4]. An extensive R&D was undertaken to identify the sensor technology for the OT. At the end,only two options were left: Float Zone (FZ) silicon n-in-p with an active sensor thickness of 290 µ m(FZ290) or 240 µ m (FZ240). The sensor technology chosen for both sensors of the p T modules isthe FZ290. An extensive irradiation and characterization program showed that FZ290 provides su ffi -cient seed signal at the standard operation voltage of 600 V and the expected maximum fluence after3000 fb − . For scenarios up to and beyond 4000 fb − , an increase to an operation voltage of 800 Vwould allow FZ290 to maintain adequate performance even at the most exposed locations. Beam testperformance of strip sensors prototype are described in [11], [12]. An overview of the seed signal asa function of annealing time for the strip sensors after irradiation to the maximum expected fluencesis shown in Fig. 5. ig. 5. Seed signal as a function of annealing time for FZ290 (blue) and thFZ240 (green) strip sensors afterirradiation to the maximum expected fluences after 3000 fb − for the 2S (left) and PS (right) modules. Solid(open) points refer to signal collected with a bias voltage of 600 (800) V. The horizontal lines represent therequired signal for the 2S and the PS strip (PS-s) sensors. Both p T modules have two di ff erent types of high density interconnect hybrid circuits whichhouse the front-end and auxiliary electronics. The hybrids are stand-alone units that are connectedvia bidirectional optical links to the backend electronics with no intermediary aggregator system. Thefront-end hybrids host the readout chips and the concentrator chip while the service hybrids house theopto-electronics and the DC-DC converters for powering. Each side of the module is connected to afront-end hybrid. For both module types the signals from the top and bottom sensor are routed to onereadout chip to perform the track stub finding. This is achieved by using a flexible hybrid which isfolded over a spacer and which allows routing of the signal between the di ff erent parts of the module.A sketch of both module types with their front-end and other hybrids is shown in the Fig. 6. Fig. 6.
The 2S module (left) and PS module (right) of the Outer Tracker. Shown are views of the assembledmodules (top), and sketches of the front-end hybrid folded assembly and connectivity (bottom) [4].
The 2S module features two silicon strip sensors each with two columns of 1016 strips with singlestrip size of 5 cm × µ m. Each sensor side is read out by eight readout chips (CBC) implemented in
30 nm CMOS technology. A CBC chip reads 254 strips (127 from bottom and 127 from top sensorstrips), performs hit correlation between the two sensors and sends the stub data out at each bunchcrossing to the concentrator chip (CIC, designed in 65 nm CMOS technology), that performs datasparsification, formats the output data, and sends them to the service hybrid. The service hybrid hoststhe LpGBT, VTRx + optical link, DC-DC converters and HV distribution circuitry. The data from thefront-end hybrids are merged and sent via a single optical fibre to the back-end electronic system.The PS module is made of one silicon micro-strip sensor with two columns of 960 strips eachwith single strip size of 2.5 cm × µ m, and a macro-pixel sensor with a matrix of 32 ×
960 pixelswith a pixel size of 1.5 mm × µ m. The strip sensor is read out by two times eight short strip ASIC(SSA), while the macro-pixel sensor is read out by sixteen macro pixel ASIC (MPA); both chips areimplemented in 65 nm CMOS technology. The track stub finding in the PS modules is done by theMPA. The MPA chip receives the information about strip clusters from the SSA which sends themtogether with information about the bunch crossing in which a hit occurred. The MPA combinesthis information with the macro-pixel information to form track stubs. The transfer scheme and dataformats are very similar to those used in the 2S module, allowing the same CIC chip to be used forperforming the same data collection, sparsification and formatting functions as for the 2S module.For space reasons, the PS module has two service hybrids, one for the optical system and one for thepowering.The module design uses novel composite materials (Al-CF) and all structural components havebeen chosen for their high thermal conductivity, low CTE, and minimal material budget. Fig. 7.
Photo of a recent 2S module prototype.
The module prototyping (assembly and testing) has been successfully accomplished in di ff erentCMS institutes and, as an example, Fig. 7 shows a recent 2S module built with the latest prototypecomponents. Details on the performance of 2S modules in test-beam environment are described in[13].
4. Material budget and performance
Despite the increased number of readout channels in the new tracker the estimated material bud-get shows a significant reduction compared to the currently installed tracker as is shown in Fig. 8. Thekey features to achieve this reduction are: a reduced number of layers, an optimised routing of theservices, use of light weight material, low-mass CO cooling as well as the use of DC-DC converters OT) and serial powering (IT).
Fig. 8.
Material budget for the currently installed CMS tracker (left) and the new tracker (right). [4].
Also the tracking and vertexing capabilities of the new tracker will be better than for its prede-cessor. Figure 9 shows the p T resolution (left) and impact parameter resolution (right) comparisonbetween the currently installed tracker (Phase-1) vs. the new tracker (Phase-2), respectively. Fig. 9.
Relative resolution of the transverse momentum (left) and resolution of the transverse impact param-eter (right) as a function of the pseudorapidity for the Phase-1 (black dots) and the new Phase-2 tracker (redtriangles), using single isolated muons with a transverse momentum of 10 GeV [4].
As shown in Fig. 10, the new tracker is also expected to be capable of maintaining a high trackinge ffi ciency (about 90%) at the high pile-up and have a fake rate below few percent.
5. Summary
The CMS Phase-2 Tracker is an ambitious project that has to cope with a higher pile-up andradiation environment at the HL-LHC. The new tracker will consist of about 200 m of silicon and willbe made up of the Outer Tracker using modules containing pairs of closely spaced sensors, and theInner Tracker with silicon pixel sensors. The key features of the new tracker are the high granularity,radiation hardness, low material budget and the capability to provide tracking information to the first ig. 10. Tracking e ffi ciency (left) and fake rate (right) as a function of the pseudorapidity for tt-bar eventswith 140 pileup events (full circles) and 200 pileup events (open circles). The tracks are required to have p T higher than 0.9 GeV. The e ffi ciency is shown for tracks produced within a radius of 3.5 cm from the centre ofthe luminous region. [4]. stage of the CMS trigger system. The project is overall in good shape and on-track for installationin LS 3 with the Inner Tracker entering the prototyping phase and the Outer Tracker in prototypingphase with first items entering the pre-production phase. eferences [1] G. Apollinari et al., High-Luminosity Large Hadron Collider (HL-LHC), CERN Yellow Rep. Monogr. 4(2017) 1-516 [https: // cds.cern.ch / record / // cds.cern.ch / record / // cds.cern.ch / record / // cds.cern.ch / record / // cds.cern.ch / record / // cds.cern.ch / record / // cds.cern.ch / record / // indico.cern.ch / event / / contributions / + , an optical link module for data transmission at HL-LHC, PoS TWEPP-17(2017) 048 [https: // cds.cern.ch / record / // cds.cern.ch / record / ff erent silicon sensor options for the upgrade of the CMS OuterTracker, J. Inst. 15, IOP Publishing, (2020) P04017 [https: // cds.cern.ch / record //