Commissioning of the new ALICE Inner Tracking System
PPrepared for submission to JINST
Commissioning of the new ALICE Inner Tracking System
J. P. Iddon a , b on behalf of the ALICE ITS Collaboration a Department of Physics, University of Liverpool, Liverpool, UK b CERN, 1211 Geneva 23, Switzerland
E-mail: [email protected]
Abstract: The upgrade of the Inner Tracking System (ITS) of ALICE (A Large Ion ColliderExperiment) will extend measurements of heavy-flavour hadrons and low-mass dileptons to a lowertransverse momentum than currently achieved and increase the readout capabilities to incorporate thefull interaction. Furthermore, the tracking efficiency will be improved at low transverse momentum.To achieve this, the new ALICE ITS is comprised of seven layers of a custom Monolithic ActivePixel Sensor design known as ALPIDE, with a spatial resolution of 5 µ m. The use of the ALPIDE-based detector design will reduce the material budget to 0 .
35% X per layer for the innermost threelayers, and to 1 .
0% X per layer for the outermost four layers, compared to 1 .
14% X per layer inthe previous ITS. The construction effort in numerous sites around the world has resulted in a fullyassembled and connected detector, which is currently undergoing on surface commissioning beforeits installation in the ALICE cavern. This contribution discusses the design and the current status ofthe commissioning of the new ITS detector, including the methods used to characterise the detectorand the results obtained so far.Keywords: Detector design and construction technologies and materials; Particle tracking detec-tors (Solid-state detectors) a r X i v : . [ phy s i c s . i n s - d e t ] M a y ontents ALICE is the dedicated heavy ion experiment at CERN [1]. The primary physics goal of ALICEis to study the state of matter at the highest energy densities reached in heavy-nucleus collisions.Under these conditions, colour confinement no longer occurs within the spatial confines of hadrons,meaning quarks and gluons are free to roam within the medium, known as the Quark Gluon Plasma(QGP).The innermost sub-detector of ALICE is the Inner Tracking System (ITS). In Runs 1 and 2 ofthe Large Hadron Collider (LHC), the ITS consisted of two layers of Silicon Pixel Detectors (SPD),two layers of Silicon Strip Detectors (SSD) and two layers of Silicon Drift Detectors (SDD). ForRun 3, an entirely new ITS has been constructed. The main goals of this upgrade are to improvethe physics reach for low-mass dielectrons and rare probes at low transverse momentum ( p T ) byincreasing vertex resolution, tracking efficiency and readout rate [2]. An integrated luminosity of13 nb − will be delivered during Run 3 for Pb-Pb collisions, which is a factor 100 larger comparedto Runs 1 and 2 for minimum bias events.The new ITS, a schematic of which is shown in Fig. 1, consists of seven layers of a CMOSMonolithic Active Pixel Sensor (MAPS) design known as ALice PIxel DEtector (ALPIDE) [3]which is discussed in Section 2.1. The innermost three layers of the ITS are known as the InnerBarrel (IB), whilst the outermost four layers are known as the Outer Barrel (OB). The OB is furthersegmented into the Middle Layers (ML) and Outer Layers (OL) which are the innermost two layersof the OB and the outermost two layers of the OB respectively.The readout rate of the upgraded ITS will be improved to 100 kHz which is twice the Pb-Pbinteraction rate. In addition, the radius of the first layer of the ITS will be reduced from 39 mm to23 mm and the pixel size reduced from 425 µ m × µ m in the SPD, to O ( µ m ) × O ( µ m ) . The– 1 –seudorapidity region of | η | < .
22 for the 90% most luminous area will be covered by the tracker,which has an active region of roughly 10 m , segmented in 12.5 billion pixels.The upgrade will result in an improved impact parameter resolution, which will be reduced bya factor six in the direction along the beam axis (from 240 µ m to 40 µ m) and by a factor three in thetransverse plane (from 120 µ m to 40 µ m) at a transverse momentum of 500 MeV/ c . An example ofthe anticipated performance of the upgraded ITS detector, with an integrated luminosity of 10 nb − ,is the measurement of the nuclear modification factor and anisotropic flow down to p T of 2 GeV/ c and 3 MeV/ c respectively for the Λ c baryon [2]. Figure 1 : Layout of the new ITS. Taken from [2].
The ALPIDE chip, shown in Fig. 2, is manufactured by TowerJazz with their 180 nm CMOS imagingprocess [4]. The ALPIDE chip is 15 mm ×
30 mm and consists of 512 × µ m × µ m and can be masked if necessary. The chips are thinned down to 50 µ mand 100 µ m for the IB and OB respectively.A key feature of the design is the deep p-well shielding of the n-well, allowing the use ofPMOS transistors inside the pixel matrix and therefore full CMOS logic. This makes it possible tohave an amplifier, signal-shaper, discriminator and multiple event buffers in-pixel. The front-endis continuously active and has a power consumption of 40 mW / cm . The pixels are arranged indouble columns and read out by a priority encoder. Only the addresses of hit pixels are sent to thechip periphery.A 2 µ m diameter, low capacitance n-well diode is used together with an epitaxial layer resis-tance of 1 kΩ · cm and a reverse bias voltage of -3 V. This contributes to a radiation tolerance of270 krad TID and 1 . × eq NIEL which is the expected dose after 10 years of operationof the ITS. The n-well diode is roughly 300 times smaller than the pixel size, which combined withthe reverse bias and hence reduced capacitance of the n-well diode, increases the signal to noiseratio. This is important for good detection efficiency at low power consumption.The peaking time of the output of the front end is approximately 2 µ s, and the discriminatedpulse has a duration of 5 − µ s. For a more detailed insight into the ALPIDE chip, see [5].– 2 – igure 2 : Cross section schematic of the ALPIDE chip. Taken from [2]. A diagram of IB and OL staves can be seen in Fig. 3. Each IB stave consists of one IB module aswell as a carbon fiber cold plate and a space frame. The cold plate has a continuous pipe runningalong the base, through which 20 °C water is circulated. The full system is under atmosphericpressure and the outlet pressure is lower than the inlet pressure (a ‘leakless’ system). Each IBmodule consists of 9 ALPIDE chips bonded structurally via glue and electrically via aluminiumwirebonds to a Printed Circuit Board (PCB). The IB layers have an average radial position of 23 mm,31 mm and 39 mm, and a total number of staves per layer of 12, 16 and 20 respectively. The weightof an IB stave is about 25 g.Each OL stave consists of two half staves that overlap. Each half stave consists of 7 OBmodules, a power bus, as well as a carbon fiber cold plate and a space frame. ML staves have thesame design as the OL staves but have 4 OB modules instead of 7. Each OB module consists of 14ALPIDE chips arranged in two parallel rows of 7, bonded to a PCB. Power is distributed to eachmodule via the power bus. The OB layers have an average radial position of 194 mm, 247 mm,353 mm and 405 mm, and a total number of staves per layer of 24, 30, 42 and 48 respectively. Thetotal number of chips per stave is 112 for the ML and 196 for the OL. The weights of ML and OLstaves is about 200 g and 400 g respectively.Each chip in the IB is read out in parallel via its own high speed data link. In the OB, eachmodule is split into two strips of 7 chips, which are segmented into 1 master and 6 slaves. Eachmaster in the OB is read out in parallel via its own high speed link. The bandwidths of the IB andOB are 1 . / s and 400 Mbit / s per link respectively.– 3 – igure 3 : Schematic diagrams of an IB stave (left) and OL stave (right).The construction of the IB, ML and OL including spares finished in July 2019, October 2019and December 2019 respectively. The overall yield for stave construction was 73% for the IB and94% for both the ML and OL. The assembly and connection to the readout system (discussed inSection 2.3) and cooling of the IB and OB were completed in mid and late 2019 respectively.Figures 4a and 4b show the fully assembled half barrels for the OB and IB respectively. (a) OB half barrel. (b) IB half barrel. Figure 4 : One fully assembled ITS half barrel.– 4 – .3 Readout System
Each chip in the IB or each master chip in the OB receives clock and control signals from, and sendshit pixel data to, a Readout Unit (RU) via an 8 m long Samtec twinax copper differential link. Everystave is connected to a dedicated RU, a total of 192 units [6]. The RUs are located in a lower radiationenvironment than the detector, resulting in less than 10 krad TID and 10 eq / cm NIEL.The RUs are connected via CERN GBT [7] link to a Central Trigger Processor (CTP), from whichthey receive triggers, as well as a Central Readout Unit (CRU), to which data from the RUs areshipped. The CRUs are housed within the Front-end Level Processor computers which are in abackground level radiation environment, the Counting Room. The detector is configured via thesame datapath, which is shown in Fig. 5.
Figure 5 : Architecture of the readout system. From [8].
Each stave in the ITS was tested with the full readout system after installation in the ITS. In thefollowing, the focus lies on the commissioning of the OB. The basic test of each stave consisted of averification of the power consumption, the control communication and the high speed links. Afterthis, threshold tuning was conducted (see Section 3.1). Finally, a fake-hit rate measurement wascarried out (see Section 3.2).
The threshold value of each chip is defined as the charge for which a pixel fires with a probabilityof 50%. In ALPIDE, every pixel has an injection capacitor allowing the stimulation of the input ofthe front-end circuitry.The threshold value of each chip can be adjusted chip-wide by augmenting two on-chip digitalto analogue converters within the front end circuitry, VCASN and ITHR. The threshold is increasedby decreasing VCASN or increasing ITHR [2]. VCASN acts exponentially on the baseline, whileITHR acts in a linear fashion on the pulse height. For this reason, VCASN is used for the main tuneand ITHR is used for a finer tune.For the threshold tuning, the concerned parameter is swept while a fixed charge correspondingto the target threshold is injected. The data are then analysed to see which parameter gives athreshold value closest to the injected charge. After this, a threshold scan is run with the optimumparameter. The threshold scan simply injects charge of increasing amplitude into the pixels, whilstkeeping the chip configuration constant. Figure 6 shows the threshold values of 7 randomly selectedOL staves after tuning VCASN. The average threshold value over the staves shown after tuning is– 5 –12 e − , with a standard deviation of 4 e − . The standard deviation of threshold values from chip tochip within each stave varies from 3 e − to 6 e − . To improve the threshold uniformity, scanning overITHR can be performed. Likely this will further reduce the range in threshold values from chip tochip.Figure 7 shows the threshold values of each chip in one stave. The average threshold acrossthis stave is 113 e − and the root mean square of thresholds per chip is 20 e − .The full services, including the full data path described in Section 2.3, as well as the finalcooling system, were used to obtain these results. Figure 6 : Tuned threshold scan of a selection of OL staves. Red chips are excluded from datataking. Threshold value units are number of electrons.
Figure 7 : Probability density function of chip thresholds for each chip in one OL stave (the laststave shown in Fig. 6). Each line represents a chip. The average threshold is 113 e − and the rootmean square is 20 e − . The fake hit scan involves reading out hit pixels without any stimulation of the chip. A hit is dueto either noise or a cosmic ray. The fake hit scan shown in Fig. 8 was performed after thresholdtuning. The scan was performed over 5 minutes with 9 . × triggers. Without masking pixels,– 6 – able 1 : Number of pixels with a given number of hits. Data from a single OL stave shown inFig. 7. Number of hits Pixel firing probability Number of pixels0 0% 1027569751 to 10 to 10 to 10 to 10
1% to 10% 10110 to 9 . ×
10% to 99% 117 ≥ . × . × − /pixel/event, comparable to the requirementof 10 − / pixel / event. This leads to an occupancy of 0 . , roughly the same as theparticle hit density in the OB for central Pb-Pb collisions [2]. The expected cosmic muon rateis 1 cm − min − [9] which leads to a cosmic muon hit density of roughly 5 . × − / event / cm ,10 times smaller than the fake hit occupancy. Some vertical lines of hits can be seen. Thesecorrespond to bad double columns, read out by a single broken priority encoder.The number of hits varies across each chip within each stave. For the stave shown in Fig. 7 forexample, the total number of hits was roughly 25 × . These hits were from only 3700 pixels,that is a fraction of 10 − of the total pixels on the stave. Of the hit pixels, 60% had less than 3hits and 80% had less than 100 hits. See Table 1 for an overview of the number of hits seen byeach hit pixel. 19 pixels had a number of hits roughly equal to the number of triggers. These areknown as stuck pixels, as they read out a hit on every trigger. The fake hit rate of this stave was3 × − / pixel / event with all pixels included and 8 × − / pixel / event after the 19 stuck pixelswere removed. The number of hit pixels per chip for this stave was on average 15 with a spread of11. Figure 8 : Fake hit scan of a selection of OL staves. Red markers denote hits. Each marker isenlarged by a factor of 40 for visibility. – 7 –
Summary
The replacement of the ITS during Long Shutdown 2 (LS2) will extend the physics reach of ALICEto lower transverse momentum, allowing the characterisation of the QGP via measurements ofunprecedented precision. The use of a sole MAPS design, ALPIDE, is a huge step forward in termsof material budget, readout rate and spatial resolution. The new ITS is now fully constructed aftera huge effort from numerous sites around the world. Notably, a yield of over 94% was achievedfor the OB staves. The readout chain for both the IB and OB has been demonstrated to work and acampaign to gather the first particle tracks from cosmic muons with the OB is ready to begin. Thedetector is on track for an installation in the ALICE cavern during LS2.
References [1] ALICE Collaboration,
The ALICE experiment at the CERN LHC , JINST 3 (2008) S08002 .[2] ALICE Collaboration,
Technical Design Report for the Upgrade of the Inner Tracking System , J. Phys.G 41 (2014) 087002 [CERN-LHCC-2013-024] (2013).[3] M. Mager on behalf of the ALICE Collaboration,
ALPIDE, the Monolithic Active Pixel Sensor for theALICE ITS upgrade , Nucl.Instrum.Meth.A 824 (2016) 434-438 .[4] Tower Jazz,
Specialty IC Manufacturing for a Smart World
The ALPIDE pixel sensor chip for theupgrade of the ALICE Inner Tracking System , Nucl. Instrum. Meth. A 845 (2017) 583 .[6] J. Schambach et al. on behalf of the ALICE Collaboration,
ALICE inner tracking system readoutelectronics prototype testing with the CERN ‘Giga Bit Transceiver’ , JINST 11 (2016) C12074 .[7] P. Moreira et al.,
The GBT project, in proceedings of Topical Workshop on Electronics for ParticlePhysics , Paris, France, September 21-25 2009, pp. 342-346.[8] A.Szczepankiewicz on behalf of the ALICE Collaboration,
Readout of the upgraded ALICE-ITS , Nucl. Instrum. Meth. A 824 (2016) 465-469 .[9] K. Nakamura et al. (Particle Data Group),
J. Phys. G 37, 075021 (2010) ..