Impact of radiation on electronics and opto-electronics
Abstract
In this chapter we will present the results of the impact of radiation on electronics and opto-electronics for two of the LHC experiments during Run 1 and Run 2. ATLAS results are presented in Section 6.1; CMS in Section 6.2. In Section 6.3 we will present a comparison between the two experiments, highlighting operational guidelines and proposing solutions to build the electronics and opto-electronics of the future LHC experiments.
Full text
CERN Yellow Reports: Monographs, CERN-2021-001 6Impact of radiation on electronics and opto-electronics Editors: M. Bindia, E. Butzb. Contributing authors: M. Backhausc; M. Bindia; P. Buttid; E. Butzb; E. Chaberte; S. Chenf;J. Dickinsong; W. Erdmannh; J. Garcia Pascuali, R. Gerosaj; B. Haneyf; H. Hillemannsk; K. Lantzschl, A. LaRosak; K. Padekenl; F. Pintom; G. Pownalln; D. Robinsono; A. Rozanovp; J. Troskak; T. Weidbergq. aGeorg August University of Göttingen, Germany bKarlsruhe Institute of Technology, Germany cETH, Zürich, Switzerland dSLAC, Stanford, USA eHubert Curien Pluridisciplinary Institute (IPHC), Strasbourg, France fUniversity of Pennsylvania, USA gFermilab, Batavia, USA hPSI, Villingen, Switzerland iIHEP, Beijing, China jUniversity of California San Diego, USA kCERN, Geneva, Switzerland lUniversity of Bonn, Germany mUniversity of Seville, Spain nDESY, Hamburg, Germany oUniversity of Cambridge, United Kingdom pCPPM, Marseille, France qUniversity of Oxford, United Kingdom In this chapter we will present the results of the impact of radiation on electronics and opto-electronics for two of the LHC experiments during Run1 and Run2. ATLAS results are presented in Section 6.1; CMS in Section 6.2. In Section 6.3 we will present a comparison between the two experiments, highlighting operational guidelines and proposing solutions to build the electronics and opto-electronics of the future LHC experiments. 6.1 ATLAS Radiation effects from TID on the IBL front-end electronics will be described in Section 6.1.1; SEU/SET effects from highly ionizing particles in IBL and SCT detectors will be shown in Section 6.1.2 whilst impact on opto-electronics from SCT will be described in Section 6.1.3. Finally, results from TRT electronics will be presented in Section 6.1.4. 6.1.1 TID effects in the IBL front-end chip The IBL consists of 14 carbon fibre staves instrumented along 64cm, 2cm wide, and tilted in by 14 surrounding the beam pipe at a mean radius of 33mm from the beam axis and providing a pseudo-rapidity coverage of ±3. Each stave, with integrated CO2cooling, is equipped with 32 front-end chips bump bonded to silicon sensors. The IBL detector was designed to be operational until the end of the LHC Run3, where the total integrated luminosity was expected to reach 300 fb1. The detector components are qualified to work up to 250Mrad of TID. This chapter should be cited as: Impact of radiation on electronics and opto-electronics, Eds. M. Bindi, E. Butz, DOI: 10.23731/CYRM-2021-001.87, in: Radiation effects in the LHC experiments: Impact on detector performance and operation, Ed. I. Dawson, CERN Yellow Reports: Monographs, CERN-2021-001, DOI: 10.23731/CYRM-2021-001, p. 87. © CERN, 2021. Published by CERN under the Creative Commons Attribution 4.0 license. 87
6. Impact of radiation on electronics and opto-electronics The IBL front-end chip, namely FE-I4B [1], was designed in 130 nm CMOS technology which features an array of 80⇥336 pixels with a pixel size of 50⇥250µm2. Each pixel contains an independent, free running amplification stage with adjustable shaping, followed by a discriminator with independently adjustable threshold. The FE-I4B keeps track of the time-over-threshold (ToT) of each discriminator with 4-bit resolution, in counts of an external supplied clock of 40MHz frequency. The FE-I4B operates by feeding the common power supply to analogue signal amplifiers and digital signalprocess circuits, referred to as the low-voltage (LV) power supply and the clock input. 6.1.1.1 Observations during 2015 data taking During the first year of the IBL operation in 2015, a significant increase of the LV current of the front-end chip and a detuning of its parameters (threshold and time-over-threshold) have been observed in relation to the received TID. The average LV current of the IBL module groups (serving four FE-I4B chips) during data taking was stable at a value of 1.6–1.7A until the middle of September 2015. Then, the current started to rise up significantly (see Fig. 71), and the change of the current during the period September–November 2015 was even more than 0.2A within a single LHC fill, depending on the luminosity and duration of the fill. Due to safety concerns, after ⇠2 fb1of integrated luminosity corresponding to ⇠580krad of TID, the IBL was switched off in early October 2015. The detector was kept off during two LHC fills for a total of ⇠240pb1of integrated luminosity, corresponding to ⇠70 krad irradiation from the LHC beam. A complete return to the baseline of the LV currents was observed when the detector was powered up again. However, as soon as the LHC beam irradiation restarted with the detector being powered, the raising behaviour of the LV current appeared again with similar characteristics. Fig. 71: LV current drift of an IBL module group (serving four FE-I4B chips) during 2015 data taking. There are two levels of the current depending on the front-end configuration: STANDBY (lower level, with pre-amplifiers off) and READY (higher level, with pre-amplifiers on). There are two long power-off periods on October 5–6 (with the LHC beam ON ) and November 3–4 (during a LHC technical stop). With the increase of the LV current, the temperature of IBL modules also changes (see Fig. 72). In addition, as shown for example in Fig.73, the calibration of the FE-I4B chips for the analogue discriminator threshold and the target ToT drifted rapidly despite a frequent update of the calibration. The increase of the LV current of the FE-I4B chip and the drifting of its tuning parameters were traced back to the generation of a leakage current in NMOS transistors induced by radiation. The radiation induces positive charges that are quickly trapped into the shallow-trench-isolation (STI) oxide at the edge of the transistor. Their accumulation builds up an electric field sufficient to open a source-drain channel where the leakage current flows. If the accumulation of positive charges is relatively fast, the formation of interface states is a slower process. The negative charges trapped into interface states start to compete with the oxide-trapped charges with a delay. This is what gives origin to the TID effect at low dose [3] which is discussed in more detail in Section2.2.1. 88
CERN Yellow Reports: Monographs, CERN-2021-001 Average Module Current per 4 FEs [A] 1.6 1.8 2 2.2 2.4 2.6 C]°Average Module Temperature [ 4− 3.5− 3− 2.5− 2− 1.5− Runs Sep 26 - Oct 6 2015 Runs Oct 9 - Nov 2 2015 Linear Correlation Line ATLAS Pixel Preliminary Fig. 72: Performance of the IBL modules during high luminosity proton–proton collision runs from September to November 2015, separated into the periods before (red circles) and after (black triangles) the long power-off on October 6. The average IBL module current (4 FE-I4B) is displayed against the average module temperature [2] and compared to a linear dependence. 0 500 1000 1500 2000 2500 3000 3500 4000 4500 ToT [Bunch Crossing] 8 9 10 11 ATLAS Pixel Preliminary ] -1 Luminosity [pb 0 500 1000 1500 2000 2500 3000 3500 4000 4500 RMS ToT [Bunch Crossing] 0.4 0.6 0.8 TID [Mrad] 0 0.2 0.4 0.6 0.8 1 1.2 1.4 Fig. 73: The time-over-threshold (ToT) and its RMS as a function of the integrated luminosity or TID [2]. The detector was regularly retuned, and each marker type corresponds to a single tuning of the detector. 6.1.1.2 Irradiation test results Dedicated laboratory measurements [4] of irradiated single transistors in 130nm CMOS commercial technologies showed that the increase of the leakage current reaches its peak value between 1Mrad and 3Mrad. For higher TID the current decreases to a value close to the pre-irradiated one. To reproduce and analyse the effects described above during the IBL operation, several irradiations and electrical tests were performed [2]. Since the current increase in NMOS transistors depends on dose rate and temperature, measurements under different temperature and dose rate conditions have been 89
6. Impact of radiation on electronics and opto-electronics carried out to qualify this dependency. The first irradiation test aimed at measuring the boundary current (at a given temperature and dose rate) that the chip always approaches after annealing periods and re-irradiation. Figure 74 shows the increase of the current consumption of a single FE-I4B chip in data taking condition as a function of the TID. The temperature of the chip was 38C and the dose rate 120kradh1. After reaching the maximum of each peak the chip was annealed for several hours resulting in the observed partial recovery. Fig. 74: Current consumption of a single FE-I4B chip in data taking condition as a function of the TID. The temperature of the chip was 38C and the dose rate 120kradh1. After reaching the maximum of each peak the chip was annealed several hours resulting in the observed partial recovery [2]. The fit performed on the first set of data (first peak) has been carried out by using the current parametrization described in Ref. [5]. In order to study the dependence of the LV current increase on temperature and dose rate several irradiation tests were performed by setting one of those variables and changing the other. Figure 75 shows the results of three different measurements, performed with three different and previously not irradiated chips. The dose rate was 120kradh1and the temperatures were 38C, 15C, and 15 C. Before irradiation, the LV current of the three chips was 400mA (38C), 360mA (15 C), and 380 mA (15C). For comparison, Fig.76 shows the result of two different measurements where the temperature was kept fix at 15 C, while the dose rate set to 120 krad h1or 420krad h1. Also in this case the tests were performed with different and previously not irradiated chips. The measurements described above revealed two facts: – at a given dose rate the LV current increase is stronger at lower temperatures; – at a given temperature, the LV current increase is stronger at higher dose rates. To simulate the dose rate conditions of the 2015 and 2016 data taking, a first irradiation was performed at 15C and 120 krad h1. This was followed by several hours of annealing and a second irradiation this time performed at 5C and 420 krad h1. As shown in Fig.77, the second LV current peak is lower than the first one, i.e., by increasing the operational temperature of the chip it was possible to keep the increase of the LV current below the boundary current given by the first irradiation. To verify that a temperature of 5C was safe for the IBL detector operation, a measurement at 10kradh1was performed. The maximum LV current increase was of the order of 250mA, which gives a LV current 90
CERN Yellow Reports: Monographs, CERN-2021-001 TID [Mrad] 1− 10 1 10 Current increase [A] 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 -15C +15C +38C ATLAS Pixel Preliminary Fig. 75: Increase of the LV current of three single FE-I4B chips in data taking condition as a function of the TID (x-axis in log scale). Test measurements were carried out at 38 C (blue), at 15 C (black), and at 15C (red) with a dose rate of 120 krad h1. A dose rate up to 10kradh1is expected in the experiment. The LV current of the single FE-I4B chips before irradiation were 400 mA (38 C), 360 mA (15C), and 380 mA (15 C) [2]. TID [Mrad] 1− 10 1 10 Current increase [A] 0 0.2 0.4 0.6 0.8 1 1.2 1.4 120krad/h 420krad/h ATLAS Pixel Preliminary Fig. 76: Increase of the LV current of two FE-I4B chips in data taking as a function of the TID (x-axis in log scale). Test measurements were carried out at 15C with a dose rate of 120kradh1(red), and 420krad h1(black). A dose rate up to 10 krad h1is expected in the experiment. The LV current of the single FE-I4B chips before irradiation were 380mA (420 krad h1) and 360 mA (120 kradh1)[2]. increase of 1A for a four-chip unit; this would not exceed the safety limit of the LV current originally set to 2.8A. In principle, lower operational temperatures are favourable for the sensor performance and properties after irradiation and therefore preferred. Consequently, irradiation and electrical tests were 91
6. Impact of radiation on electronics and opto-electronics TID [Mrad] 1− 10 1 10 Current increase [A] 0 0.2 0.4 0.6 0.8 1 1.2 1.4 120krad/h @-15C 420krad/h @+5C ATLAS Pixel Preliminary Fig. 77: Increase of the LV current of a single FE-I4B chip in data taking condition as a function of the TID (x-axis in log scale) during two consecutive irradiation campaigns in a lab measurement. Between the two irradiations several hours of annealing period at room temperature was performed, and resulted in the observed recovery. The TID of both irradiations is summed up. The LV current of a single FE-I4B chip before irradiation was 380mA (first step) and 550mA (second step) [2]. also performed at a temperature of 0 C to investigate the feasibility for a colder operation. In addition it was investigated the evolution of the maximum of the LV current peak under several irradiation steps followed, interleaved with periods of annealing. In this case the first two consecutive peaks of the LV current increase exceeded the maximum current allowed for a safe detector operation. Therefore, it was decided to set 5C as minimum temperature for a safe and successful data taking. Another effect was confirmed during the irradiation campaign of one FE-I4B chip with 16 MeV protons at room temperature (see Fig.78). The chip was left without powering for 30 minutes in the beam, integrating a 2Mrad dose (on top of the 9.2 Mrad of dose previously collected). As a result, the chip was completely annealed; this behaviour confirmed the observations with the IBL detector at LHC, with the return to baseline of the LV current consumption. The LHC data were taken with a module T⇠7C on the 6 October 2015 (see Fig.71). To be noticed that a similar effect was also observed during X-ray irradiation of CMS DC-DC converters (see Section 6.2). 6.1.1.3 Detector operation guidelines Based on the observations during the first year of data taking in 2015 with the IBL detector, it was decided to raise the safety limit for the IBL LV currents from 2.8A to 3A for module groups of four chips, which means a maximum current consumption of 750mA per chip. Since the average current consumption for a single FE-I4B chip is about 400mA before irradiation, the increase of the current due to the TID effects cannot be higher than 350mA per chip. Given the above results it was decided to increase the IBL operation temperature from 10C to 15C. In addition, the digital supply voltage (VD) was lowered from 1.2V to 1 V to decrease the LV current. Thanks to dedicated measurements at 5C and at a dose rate comparable to the LHC in 2016 (10krad h1), it is proven that the current increase is of the order of 250 mA. With this a module group of four chips does not exceed the safety limit of 3A. Therefore operating the IBL at 5C is safe with respect to the expected luminosity in 2016. The temperature of the IBL cooling system was 92
CERN Yellow Reports: Monographs, CERN-2021-001 Fig. 78: LV current in a single FE-I4B chip during proton irradiation as a function of the TID delivered. The LV current is shown for the analogue and digital part of the FE-I4B chip, the sum of the two is also reported (total current). The LV current was monitored both during the irradiation and for about an hour afterward, showing an annealing effect. The FE-I4B chip was at room temperature and the dose rate was 3Mrad·h1for the first irradiation and 4Mrad·h1for the following campaign. The measurement was made in two steps, separated by one week in which the chip was not powered and left at room temperature. During the second irradiation, the chip was switched off for half an hour while the irradiation continued. In this measurement, the chip was irradiated with a 18 MeV proton beam [6]. lowered to a set point of 5C. The digital supply voltage (VD) was raised from 1V to 1.2 V, after an accumulated dose of ⇠4.3Mrad which, as the measurements show, is well beyond the high peak region for the current consumption. An overview of the mean LV current of the IBL FE-I4B chips as a function of integrated luminosity and TID during stable beam is shown in Fig. 79. The LV currents are averaged for all modules across 100 luminosity blocks (⇠100 minutes), and the changes in digital supply voltage (VD) and the temperature (TSet) are highlighted. In addition, since the shift of the tuning parameters can be seen even at low dose rates and warmer temperatures, a re-tuning on a regular basis was performed. 6.1.1.4 Observations collected at the end of Run 2 During the late years of Run 2 (2016–2018), the LHC instantaneous and integrated luminosity grow considerably respect to the previous years. The effect of the detector occupancy on the IBL LV currents became visible. Figure80 shows the average IBL LV current at the start of each stable beam period, whenever the FE-I4B pre-amplifiers were turned ON, for the entire Run2 period. A clear correlation of the current consumption with the LHC peak luminosity was observed. This correlation became ⇠100% for the difference between IBL LV currents at the start of each LHC fill (Istart) and right before the fill, with amplifiers OFF (Iampoff), for 2017–2018 data (see Fig.81). In this figure the TID effects are very small and the current difference is dominated by the chip digital activity proportional to the LHC peak luminosity. Figure82 shows the IBL LV current consumption during the inter-fill periods, when no beams were present in the machine and the FE-I4B pre-amplifiers were off. In this case, the only contribution comes from the integrated TID effect; such contribution became less and less effective, showing a sort 93
6. Impact of radiation on electronics and opto-electronics ] -1 Integrated Luminosity [fb 0 5 10 15 20 25 30 35 LV Current [A] 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 Mean Range IBL Total Ionising Dose [Mrad] 0 2 4 6 8 10 ATLAS Pixel Preliminary 2015 1.2V 2016 1.0V 2016 D 1.2V = V C o -10 C o 15 Set C = T o 5 Fig. 79: Mean IBL LV current during stable beam against integrated luminosity and TID. Changes in digital voltage (VD) are highlighted. The set temperatures (TSet) of the modules correspond to actual module temperatures of about 5C, 20C, and 10 C. There were significant increases in LV current during 2015; this was addressed in 2016 by increasing the module temperatures and decreasing the digital voltage. The digital voltages were later increased back to decrease readout error frequency. 100 200 300 400 500 600 Relative fill number 2015 2016 2017 2018 =1.2V d V 1.0V 1.2V C°=-10 set T C°15 C°5 C°-20 ATLAS Preliminary Data 2015-2018 1.5 2 2.5 LV current at the start of the fill [A] 0 2 4 ] -1 s -1 cm 34 Peak luminosity [10 Peak Luminosity Fig. 80: Mean IBL LV current at the start of each stable beam fill (pre-amplifiers turned on) of saturation on the LV currents in 2018 respect to the previous years. Figure83 shows the average IBL LV current consumption for FE-I4B chips with planar sensors that shares the same zcoordinates. In total there are 56 FE-I4Bs per each zlocation. Higher current is present in the central part of the detector whilst a smaller impact is visible in the forward region. This increase of the current in the central zregion is the sum of the TID effects and occupancy effects. The shape of the distribution is present during the entire Run 2 period. 6.1.1.5 Summary During the first year of data taking in 2015, a peculiar increase of the LV current of the FE-I4B chip and the detuning of its parameters (threshold and time-over-threshold) was observed in relation to the received total ionizing dose. It was tracked back to the generation of a leakage current in NMOS transistors induced by radiation. Dedicated irradiation and electrical tests of FE-I4B chips showed that the leakage 94
CERN Yellow Reports: Monographs, CERN-2021-001 300 350 400 450 500 550 600 650 Relative fill number ATLAS Preliminary 0.25 0.3 0.35 0.4 0.45 Istart-Iampoff [A] 1− 0 1 2 3 ] -1 s -1 cm 34 Peak luminosity [10 Data 2017-2018 Peak Luminosity Fig. 81: Mean IBL LV current at the start of each stable beam fill (Istart) (pre-amplifiers turned on) subtracted of the mean current right before the fill injection (Iampoff) (pre-amplifiers turned off). 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 2.1 2.2 LV current amplifiers OFF [A] 2015 2016 2017 2018 =1.2V d V 1.0V 1.2V C°=-10 set T C°+15 C°+5 C°-20 1 10 2 10 ] -1 Integrated luminosity [fb 1 10 Dose [MRad] ATLAS Preliminary Fig. 82: Mean IBL LV current during the Run 2 inter-fill periods (pre-amplifiers turned off) current reaches its peak value when the total ionizing dose is in the range of 1–3Mrad, and above this the current decreases to a value close to the pre-irradiation one. This effect was shown to be temperature and dose rate dependent. Due to safety concerns, the IBL was temporarily powered off in October 2015 (during two LHC fills). A complete annealing of the LV currents was observed; this effect was confirmed in a dedicated test during the proton irradiation campaign. Thanks to the intensive irradiation studies it was possible to propose special detector settings (temperature and digital supply voltage) that could guarantee a successful data-taking in the following months. With the increase of the integrated luminosity/dose at the end of Run2 operation, a clear saturation of the LV current was observed. Furthermore, a minimal contribution of the LV current from the detector occupancy was also visible during the high luminosity fills (peak lumi >1034 cm2s1) in the years 2017–2018. A clear zdependence on the IBL LV currents was observed during Run2, similar to the data obtained from sensor leakage current measurements and simulation presented in Chapters 5and 7. 95
6. Impact of radiation on electronics and opto-electronics Pixel latch# 0 2 4 6 8 10 12 -1 Rate of bit flips per fb -3 10 -2 10 -1 10 1→0 0→1 = 13 TeVs2018, LHC Fill 7334 Shift Register = 1 ATLAS Pixel Preliminary Pixel latch# 0 2 4 6 8 10 12 -1 Rate of bit flips per fb -3 10 -2 10 -1 10 1→0 0→1 24 GeV protons PS 1→0 Shift Register = 0 (a) (b) Fig. 90: Average rate of SEU/SET bit flips in pixel memory of FE-I4B per fb1in LHC fill 7334, as a function of bit number (0-12). In (a), the SR was set to 1, and 0!1flips dominate. In (b), the SR was set to 0, and 1!0flips dominate. The extrapolation of the measurement of the SEU rate with 24 GeV protons at CERN PS is shown with a blue line. ] -1 Integrated luminosity [pb 0 100 200 300 400 500 600 Cumulative fraction of GCM bit flips 0 0.0005 0.001 0.0015 0.002 0.0025 0.003 0.0035 0.004 0.0045 0.005 GCM 0 -> 1 GCM 1 -> 0 = 13 TeVs2018, LHC Fill 7333 RunMode=OFF, Register = 1 ATLAS Pixel Preliminary Fig. 91: Cumulative fraction of SEU/SET bit flips in the global configuration memory (GCM) of FE-I4B as a function of integrated luminosity in LHC fill 7333. 102
CERN Yellow Reports: Monographs, CERN-2021-001 Table 12: Summary of measured rates and SEE cross-sections Beam, SEE type Transition SR Rate per fb1Cross-section value (%) (stat./syst.) 1015 cm2 LHC: mainly SEU 0!10 0.22 ±0.01 ±0.09 0.24 ±0.13 LHC: mainly SEU 1!01 0.46 ±0.01 ±0.19 0.51 ±0.26 LHC: mainly SET 1!00 3.07 ±0.02 ±0.80 3.39 ±1.34 on LOAD line LHC: mainly SET 0!11 4.68 ±0.03 ±1.21 5.16 ±2.04 on LOAD line 24 GeV protons 0!10 n.a. 1.10 Mostly SEU with some SET admixture Fig. 92: SEU rate per module versus module cluster occupancy. The line shows the result of a linear fit through the origin. tons at CERN PS. So for future HL-LHC experiments, measurement of SEE cross-sections using the 24 GeV proton PS beam is recommended, and special attention should be paid to the optimization of chip design not only against SEU, but also against SET effects. 6.1.2.2 SEU in SCT front-end chips From test beam data single event upsets (SEU) are expected in the on-detector p-i-n diodes that receive the TTC data and in the ABCD ASICs. Extrapolations of the test beam results were used to compare with the measured in situ results [17]. p-i-n diodes While no direct measurements of SEU rates could be performed in situ, the occurrence of SEUs led to a characteristic signature of synchronization errors [17]. The characteristic signature for a genuine SEU is the occurrence of a burst of events which fails this synchronization test. The correlation between the measured SEU rate and the cluster occupancy in a module in a given run is shown in Fig.92. The cluster occupancy is a good proxy for particle flux, as the rate of noise hits is negligible. The measured number of SEUs in a data set corresponding to an integrated luminosity of 8.69 fb1 was 2504. The value of the SEU cross-sections determined from the test beams was used to predict the 103
6. Impact of radiation on electronics and opto-electronics Fig. 93: Rate of SEUs in the ABCD DAC registers versus the module occupancy in a run number of SEUs. The value determined was 1900 which is in reasonable agreement given the uncertainties in the extrapolation [17]. ABCD registers SEUs in the ABCD registers cannot be directly identified during ATLAS operation. However, an indirect determination of SEUs in the ABCD register that sets the DAC value for the threshold is possible. The MSB is normally set to ‘1’ and if an SEU causes it to be flipped to a ‘0’ the threshold would become below the baseline. This would then result in a very high occupancy until the register was reset. This allowed the detection of SEU bursts. To demonstrate that these candidates are genuine SEUs, the rate was measured as a function of fluence. The results are illustrated in Fig. 93 and also show the expected linear relationship. The measured number of SEU events in a data set corresponding to 20.3 fb1was 3046 ±100. Simple extrapolation of the measured SEU cross-sections at test beams gave a prediction of 1000 [17]. The discrepancy is probably a reflection of the uncertainties in the extrapolation of the SEU cross-sections with energy. Mitigation strategies including regular resets have reduced the data loss due to SEU to a negligible level. 6.1.3 Optical links studies in SCT The radiation damage in the optical links was measured in situ [18] and compared with extrapolations of test beam data [19], [20]. The radiation induced damage expected at the end of Run3 can be accommodated in the links power budgets. 6.1.3.1 VCSELs The optical power of the on-detector VCSELs was measured in special runs by measuring the photocurrents in the off-detector p-i-n diodes which received the light. The optical power decreased linearly with luminosity. The decrease in optical power was measured in different regions of the SCT that were exposed to different fluences. The results of these measurements [18] are summarized in a plot of the change in VCSEL output versus the fluence (see Fig. 94). Assuming that the threshold current and slope efficiency change linearly with fluence the results can be compared to test beam data with and without annealing. As these links have had more annealing time than used in the test beam studies, the plot indicates that the radiation damage is slightly larger than expected. 104
CERN Yellow Reports: Monographs, CERN-2021-001 250 300 350 400 ) -2 cm eq 1MeV n× 11 Region total fluence (10 10− 5− 0 Change in output power (%) Barrels Endcap A Endcap C Linear fit Test beam with annealing Test beam before annealing ) -1 1MeV n_eq/cm^2 /fb× 11 0.001) % per (10±0.009 −Gradient: ( / DOF = 65.91 / 8 = 8.24 2 χ 0.32), ± = 0): (0.05 LIntercept( Fig. 94: Plot of the change in VCSEL output versus fluence for the different regions of the SCT 6.1.3.2 The p-i-n diodes The radiation damage in the p-i-n diodes was studied in situ by measuring the photo-currents for the on-detector devices. The decrease in responsivity is not linear and tends asymptotically to a fixed value. The results of these fits are shown for one layer of the barrel in Fig. 95. The other regions gave similar results [18] and the weighted mean value for the asymptotic decrease of responsivity was Re↵ 1=0.731 ±0.027 ±0.046, which is in good agreement with the value of 0.71 obtained in test beam measurements [19]. ] 1− [fbLIntegrated Luminosity 0 20 40 60 80 100 120 140 160 180 > 0 I / PIN I< 0.9 0.95 1 1.05 1.1 1.15 0.01± = 0.25 0 R 0.0007± = 0.8803 ∞ R 0.007± = 0.782 (eff) ∞ R / DOF = 0.90 2 χ Measurements 0 L/L^2) + L+B/(A −) = L(fFit Barrel 3 Fig. 95: On-detector p-i-n diodes. Each point represents the mean value for one day averaged over all modules in this layer. The value is the normalized mean value of the photo-current IPIN. 105
6. Impact of radiation on electronics and opto-electronics 6.1.4 Irradiation effects in TRT electronics The signals coming from the TRT straws are detected by custom designed electronics [24]. The analogue part of the electronics ‘amplifier, shaper, discriminator and baseline restorer’ (ASDBLR) performs the amplification, shaping, discrimination and base-line restoration against the long trailing signal caused by the drifting ions. ASDBLR is capable of detecting signals at two different thresholds: a tracking threshold about 200 eV (low-level level threshold), and the transition radiation threshold about 5 keV (high-level level threshold). The output is a ternary signal, with the first level indicating the time the input signal exceeded the low threshold and the second level indicating the time the input signal exceeded the high threshold. The digital readout chip for the TRT is the ‘drift time measurement/read out chip’ (DTMROC). A DTMROC receives the signals from two ASDBLR’s, that is from 16 straws. It samples each input eight times per 25 ns bunch crossing in 3.125 ns intervals. It also determines whether the high threshold was exceeded in each bunch crossing. The high and low thresholds for a ASDBLR are set by two registers in a DTMROC. DTMROCs are also capable of injecting test charge to ASDBLRs which is shaped similar to the integral of the expected ionization signal in a xenon filled straw in TRT. When a L1A signal is received, the data is sent from DTMROCs to the back-end consisting of the TRT Read-out Driver and subsequently the ATLAS central ROS system [25]. The TRT front-end electronics is susceptible to radiation damage caused by the intense particle flux coming from beam collisions. There are two important effects: irreversible radiation damage on the gain and the offset of the ASDBLR transistors, and the temporary radiation damage (SEU) in the DTMROCs. 6.1.4.1 Effective threshold shift in ASDBLR Transistors are sensitive to displacement damage from neutrons, protons, pions, and other interacting particles such as kaons. In ASDBLRs, this causes an increase in current at the base of the transistors of the pre-amplifiers, which lowers the gain [24]. The loss of gain can manifest itself both as a decrease in the gain of the amplifier and as a shift in the offset. Both of these effects can contribute to an effective shift in the low or high threshold. Extensive irradiation tests, to many tens of Mrad, were carried out on the electronics prior to the construction of ATLAS and no significant effects were to be expected in either Run1 or Run 2. However, during Run 1, it was observed that the detection efficiency for Z!ee decreased slowly during the run [26]. This was traced to a small shift in the effective high threshold. While similar shifts in the effective low threshold also occurred, it did not cause observable effects since the low threshold was adjusted regularly to maintain a constant noise occupancy. On the other hands, the high-level thresholds were not tuned time to time as it was expected that any small shifts that might occur would not affect the efficiency for electron identification. This phenomenon, which was not observed in the former irradiations tests since they were performed only with higher dose, was investigated using a cobalt-60 source that produces gamma-rays at 1.17 MeV and 1.33 MeV at Brookhaven National Lab (BNL). Sets of TRT endcap front-end boards (triplet with 12 DTMROCs, 24 ASDBLRs, 192 ASDBLR channels) were irradiated with various doses: 30krad which is roughly equivalent to the total Run 1 dose, 60 krad, and 500 krad. The metric of interest is the effective low-level threshold defined as the at the low-level threshold which the ASDBLR outputs would produce a 50% occupancy; measurements were made both with zero injected charge and with an injected charge from the DTMROCs with a pulse height corresponds to an ionization energy of typical low-level threshold hits (400 eV). Figure 96 shows the shift in the effective low-level threshold measured with and without charge injection. The observed shift, corresponding to an ionization energy of about 30 eV, after 30 krad is similar to that observed in the detector during Run1. Tests to higher total dose rates showed no further shift up to 500krad. Measurements of the shift in the high-level threshold after irradiation at BNL were not recorded. However, a similar examination is performed in the ATLAS TRT by similarly measuring the 50% occupancy high-level threshold regularly over the time of Run 1 and Run 2. Figure 97 summarizes the 106
CERN Yellow Reports: Monographs, CERN-2021-001 0 50 100 150 200 Channel index 0 10 20 30 50% occ. LT [DAC]Δ- Noise: 0 rad - 30k rad Noise: 30k rad - 500k rad Test pulse: 0 rad - 30k rad Test pulse: 30k rad - 500k rad ATLAS Preliminary Fig. 96: The magnitude of negative shift in the effective low-level thresholds (LT) of ASDBLR chips after being irradiated by a Co60 source at BNL. The effective low-level threshold is determined as the DAC value that yields 50% occupancy for noise or a test pulse which corresponds to an ionization energy of 400 eV. The individual shifts of 24 ASDBLRs (192 channels) on a test board are shown. A shift of 12–13 DAC counts, corresponding to an ionization energy of about 30 eV is observed for both threshold definitions after an irradiation of 30krad, which corresponds to the average dose accumulated in the inner layer of the TRT over the course of Run1. No significant shift is seen for additional irradiations up to 500krad, except for the three chips which saturated at about 60krad. chronological change of effective high-level threshold measured using the test pulse with the height equivalent to an ionization energy of 1.9 keV. The shift is clearly observed during Run1 and was sufficient to cause the observed decrease in electron efficiency. The effect saturates after about 30 krad and there is no further shift during Run 2, which is consistent with what is found in the low-level threshold in the BNL experiment. Nonetheless, the high-level thresholds used for operation have been updated since the beginning of the Run2 to compensate for any possible shift, calculated based on the extrapolated 50% occupancy threshold shift at the test pulse height equivalent to the transition radiation signal which is about 5 keV ionization energy equivalence. 6.1.4.2 Single event upset in DTMROCs When a charged particle or electromagnetic radiation traverses digital micro-circuitry, secondary ionization charges can cause internal logic elements to change their state (SEU ‘bit flips’). In the DTMROCs the bits that are susceptible to SEUs are: (1) straw hit data bits in the buffer; (2) event identifier bits associated with the straw data; (3) the bit flips in the registers storing the configuration information. While the former two have relatively limited impact on the data taking either because only small subset of data is read out on the L1A or because they are frequently reset according to the LHC orbit, the SEUs on the configuration registers can cause serious impact. The general configuration and the threshold information are the most critical part of the configuration registers. Therefore they are triplicated and 107
6. Impact of radiation on electronics and opto-electronics 0 50 100 150 Integrated luminosity [1/fb] 40 60 80 Avg. 50% occ. HT [DAC] ATLAS Preliminary 2012 2016 2017 2018 2011 2015 Run1 Run2 Barrel-A layer 1/2/3 Barrel-C layer 1/2/3 Endcap-A type A/B Endcap-B type A/B 0 50 100 150 Integrated luminosity [1/fb] 5 10 15 (avg. 50% occ. HT) [DAC]Δ- ATLAS Preliminary 2012 2016 2017 2018 2011 2015 Run1 Run2 Barrel-A layer 1/2/3 Barrel-C layer 1/2/3 Endcap-A type A/B Endcap-B type A/B Fig. 97: (a) The effective high-level threshold (HT) of ASDBLR chips and (b) its decrease since the beginning of 2012, as function of integrated luminosity recorded in the ATLAS over the last year of the Run1 (2012) and Run2 (2015–2018). The HT is determined as the DAC value that yields 50% occupancy for a test pulse which corresponds to an ionization energy of 1.9 keV. Values are shown separately for the A side and C side of the detector for the three barrel layers and groups of endcap layers with type A and type B front-end boards, which largely correspond to the low-/high-zpart of the endcaps. All shown values are averaged over all channels in the respective detector part. The TRT active gas mixture of barrel layers 1 and 2 have been changed between Run 1 and 2015; and between 2015 and 2016. This required a change of the ASDBLR shaping, which resulted in a large change in the HT. The other changes to the HT during Run2 are still being investigated, however, they are unrelated to radiation as the changes are independent of integrated luminosity. automatically corrected using the parity logic in the DTMROCs. A polling scheme is also implemented in the TRT DAQ system to monitor and correct the DTMROC configuration registers. This is done by the TRT-TTCs requesting read back during the 199 bunch-crossing long beam gap booked at the end of the orbit, and compare the register information to the originally written configuration from the TRT-TTC. Any mismatch is logged in the database. The registers regarding to the general configuration and the thresholds are additionally re-written once the bit changes are detected. The polling is done sequentially and each DTMROC channel is fetched at about 10 Hz. A SEU rate measurement has been performed utilizing this polling system during the ps=7TeV operation in 2011, or the ps= 13 TeV operation in 2015-16. After removing the events recorded due to the other known reasons such as bad communication between the TRT-TTCs and the DTMROCs, 46 and 1016 SEU events are observed in total during the ps=7TeV operation in 2011 and the ps= 13 TeV operation in 2015-16 respectively. Figure 98 shows the calculated SEU rate for the entire TRT as function of the average instantaneous luminosity in the various selected run period. A good linearity with respect to the instantaneous luminosity at the same centre-of-mass energy is found. The SEU cross-section per DTMROC chip is derived as (1.25.7) ⇥1014 cm2given the particle fluence in the TRT (charged and neutral hadrons with E>20 MeV: 4.0⇥1011 1.8⇥1012 cm2per 1 fb1of integrated luminosity depending on the position [27]), and the total number of chips (22 016). This is largely consistent with the measured SEU cross-section (0.81.2) ⇥1014 cm2for the DTMROC prototypes obtained using the 24 GeV proton test beam at the CERN PS irradiation facility [28]. 6.2 CMS The CMS tracking detector consists of the silicon pixel and silicon strip detectors. Together they provide charged particle tracking in the pseudo-rapidity range of |⌘|<2.5. The original CMS pixel detector [30] was running during the years 2009–2012 and 2015–2016 and was replaced with the CMS Phase-1 pixel detector [31] during the extended year-end technical stop 2016/17. 108
CERN Yellow Reports: Monographs, CERN-2021-001 0 2 4 6 8 10 12 ] -1 s -2 cm 33 10×Instantaneous luminosity [ 0 0.05 0.1 0.15 0.2 0.25 SEU rate [mHz] ATLAS Preliminary =7 TeVsRun1, =13 TeVsRun2, Fig. 98: Single event upset (SEU) rate in the DTMROCs measured as a function of the instantaneous luminosity observed in the ATLAS detector, using 2011 Run 1 data, collected at ps=7TeV (hollow dots) and 2015–2016 Run2 data, collected at ps= 13 TeV (closed dots). The SEU rate is calculated for selected run periods for which the instantaneous luminosity is determined as the average over the respective period. The x-axis error bars represent the luminosity range from the different runs in the period. Only one point is shown for the ps=7TeV result due to the limited data statistics. The observed SEU rate depends linearly on the instantaneous luminosity at the same centre-of-mass energy. 6.2.1 Radiation effects in the CMS pixel detector The CMS pixel detector and its readout electronics have been described in Section 3.2.2.1. In the following some of the radiation effects observed in both the Phase-0 and Phase-1 pixel detectors will be described. Evolution of low voltage currents Figure99 shows the evolution of the analogue low voltage currents in the CMS Phase-0 barrel pixel detector over its full lifetime. The current is averaged for all ROCs belonging to the layer in question. To be comparable, the current measurements are always taken 10 minutes into an LHC fill after it has reached stable proton collisions. Several distinct features can be seen in the plot. The current rises most steeply in the innermost layers in early Run1 necessitating a recalibration of layers 1 and 2 already after few fb1. Another two recalibrations were performed during Run1 on all layers afterwards. It can be seen that the increase per luminosity decreases over the course of the run from initially around 0.4mA/fb1/ROC for layers 1 and 2 gradually decreasing to 0.1 mA/fb1/ROC. For layer 3 the initial increase is lower at around 0.2mA/fb1. The analogue circuitry is powered via a bandgap reference voltage which increases with dose. This effect has been measured for the Phase-1 PSI46dig in Ref. [34] to be around 11% for TID expected for the lifetime of the Phase-1 detector with a 10% increase below 300kGy. The increase reaches a plateau of around 14% for very high TIDs. Radiation and SEU effects The SEU cross-section in 250nm technology which is used for the readout ASICs has been measured and protection measures have been put in place [33]. The large majority of SEUs in the ROCs will not be visible as they only affect small parts of the ASIC like single-pixel trim bits and do not affect the detector 109
6. Impact of radiation on electronics and opto-electronics Fig. 99: Evolution of the analogue supply currents per ROC for the CMS Phase-0 pixel detector as a function of the integrated luminosity. Recalibrations, full or partial, are indicated as well as short technical stops (TS) and year-end technical stops (YETS). response in a significant way. The data taking will be affected by SEUs when vital parts of the ASIC get affected which can happen with some probability. Also the TBM chips can be affected by SEUs in which case the readout to a group of ROCs is stopped. Both SEUs for single ROCs and TBMs get detected in the front-end driver (FED) that keeps a list of channels which do not send data. Once a programmable threshold of channels has been reached, a recovery procedure is initiated which will be described in more detail below. The threshold is adapted depending on the position in the detector and hence the potential impact on the data quality. The rate at which ROCs become fully inefficient in the Phase-0 detector was estimated from data to be around 1/2–4pb1for the PSI46 and about 1/73 pb1for the TBM. In the Phase-1 detector the TBM ASICs have been found to be more susceptible to SEUs because of a single non-protected transistor in the readout control. This condition cannot be recovered with a reprogramming of the chip but instead a power-on reset has to be issued. In 2017 the power cycling was performed using the DCDC converters of the Phase-1 detector. This procedure led to the discovery of the DCDC converter issue caused by a problem in the FEAST ASIC. The description of this problem is beyond the scope of this report, more details can be found in [35]. The rate at which TBM cores got stuck because of this problem was about 0.7%/100pb1for BPIX layer 1, the worst affected region of the detector. During LS2, the innermost barrel pixel layer will be replaced. Improved versions of the readout ASICs will be deployed and the increased SEU susceptibility of the TBM will be removed for the newly installed modules. Figure100 shows the evolution of inactive ROCs in the barrel layer 1 over a long (11h) LHC fill with peak instantaneous luminosity around 1.5⇥1034 cm1s1. The recovery from SEUs or SEU like effects is done by completely reprogramming the front-end ASICs. This can either be done specifically targeting affected detector regions or by performing a full reprogramming of the full detector. For the year 2018 the firmware of the front-end controllers, responsible for reprogramming of the detector modules, underwent a major revision. In the original firmware the amount of parallelism was limited by FIFO sizes and the structuring of the on-board memory. The new firmware features a segmented DDR memory and multiple segments per module with a maximum of 28 modules per channel. With the new firmware the time needed to perform pixel-level reprogramming of the detector modules decreased dramatically to only few seconds. This allowed to deploy a more complete recovery scheme for the 2018 run where a full pixel-level reprogramming of the modules was possible also in short interruptions of the data taking (making use of the CMS soft error recovery mechanism). In Fig.101 the recovery time before and after the firmware update are compared. Here it is important to note that before the firmware update only a partial recovery was performed while with the 110
CERN Yellow Reports: Monographs, CERN-2021-001 Fig. 100: The number of inactive ROCs as a function of time in BPIX layer 1 during a typical LHC fill in 2017. The number of inactive ROCs increases until a programmable threshold is reached, at which point the SEU recovery mechanism is activated and the ROCs are recovered. The SEU recovery mechanism can be activated several times during an LHC fill. The SEU rate depends on the instantaneous luminosity, which decreases over the time of the fill. In the fill used for this plot, the peak luminosity was around 1.5 ⇥1034 cm2s1. The typical rate for inactive ROCs is one in five minutes at an instantaneous luminosity of 1.0 ⇥1034 cm2s1for BPIX layer 1 modules. See Ref. [32]. Fig. 101: Comparison of fast module configuration time before and after the PixelFEC firmware upgrade in 2018. new firmware the full pixel-level reprogramming is carried out. The time needed for the full procedure with the new firmware is less than for the partial recovery with the old. For Run3 a new scheme is being developed that will make use of fast reset commands to perform continuous reprogramming of pixel registers during a run. 6.2.2 Radiation effects in the CMS strip detector The CMS strip tracker and the main components of its front-end electronics have been described in Section 3.2.2.2. In the following some of the radiation effects observed in the Front-End electronics of the system will be described. 111
6. Impact of radiation on electronics and opto-electronics Fig. 110: Single strip occupancy from a run directly after dump of the LHC stable beams. Both sets of power groups in the positive and negative end caps which have been left without sensor bias only (TEC+) and without both sensor bias and front-end low voltage (TEC-) exhibit high occupancies. The drift was characterized and reproduced during an X-ray irradiation campaign at CERN, where the dependency on parameters like dose rate, temperature and digital supply voltage was studied. Important guidelines were given to operate the detector safely, running temporarily at warmer temperature and lower digital supply voltage until the ‘TID bump’ was reached in 2016. A similar but much less pronounced behaviour was observed in the analogue currents of the CMS pixel ROCs (PSI46v2, 250 nm CMOS technology), during Run1 and at the beginning of Run 2 (before the replacement with the Phase-1 pixel detector chips (PSI46dig and PROC 600). For the CMS pixel ROCs in the Phase-0 detector, a regular calibration was performed in order to contain the drift of the currents. For both cases, a plateau of the TID effects seems to be reached for high dose values >300 kGy (reached by the IBL detector during 2018); the increase was quantified to be ⇠14% for the CMS Phase-1 pixel detector. Furthermore, a beam annealing effect was observed in the ATLAS IBL FE-I4B chips in early October 2015, when the IBL was kept off and irradiated for a short time period (during two LHC fills). Such effect was reproduced during an irradiation campaign with 16 MeV protons and was observed during X-ray irradiation of CMS DC-DC converters. More systematic studies are needed for better understanding of this effect. CMS observed SEU effects in the pixel detector. The most pronounced effect on data taking came from SEUs in the TBM chips of the Phase-1 detector. Because of an unprotected transistor their rate was greatly enhanced and in addition SEUs could result in the need for a power-on reset. These SEUs subsequently led to the discovery of the DC-DC converter problem during 2017. SEUs in the ROCs are also observed and the rates were quantified and found to be consistent with expectations. Particular emphasis was placed on the recovery mechanism. Actions were taken to mitigate the impact on the data quality and to limit the dead time in the DAQ system whenever a certain fraction of detector was affected. An upgrade of the control firmware for the 2018 run greatly reduced the recovery time needed. The entire ATLAS tracker was affected by SEU, with different level of severity depending on the proximity and peculiarity of some of the electronics components. In the ATLAS IBL, SEUs were observed at rates close to the expectation based on previous tests in 24 GeV proton beam. SEUs in the global and local pixel configuration memories of FE-I4B chip were observed during LHC fills which had several consequences, including: module de-synchronization, current jumps, quiet modules, quiet pixels, noisy pixels, and broken clusters. Even in this case, a recovery mechanism was put in place in 118
CERN Yellow Reports: Monographs, CERN-2021-001 2017, performing a regular reconfiguration of the FE-I4B global registers in the entire detector without introducing additional DAQ dead time. Furthermore, by read-back measurements and simulations of the electronics, it was demonstrated that SETs on the LOAD line of the DICE latch dominate the memory flips. The ATLAS SCT observed and analysed the effects of SEU in their on-detector p-i-n diodes and in ROCs (ABCD). The macroscopic effect of such issues were respectively de-synchronization (bit flip in the trigger transmission) and noise (bit flip in the threshold value). Both measurements show a good agreement with the extrapolations from test beam results. The ATLAS TRT also observed an effect on their ROCs (DTMROCs), in particular in the configuration registers. For this reason a continuous polling and rewriting of such registers was implemented in the DAQ system using the abort gap timing in the LHC orbit. For what concerns the damage from NIEL, ATLAS TRT analogue electronics (ASDBLR, BiC- MOS DMILL radiation hard process) experienced a shift of the effective threshold cause by a loss of gain. This effect could be reproduced during an irradiation campaign at BNL and shows a saturation effect at about 30kGy. ATLAS SCT observed an effect on the opto-electronics, measuring the decrease of power of the on-detector VCSELs and a decrease of photo-current of the on-detector p-i-n diodes. Both effects are in good agreement with the test beam results even if the VCSEL optical power degradation seems a bit higher than expected considering the longer annealing periods during the LHC operation respect to the test beam campaigns. CMS strips has also observed an impact of radiation on its optical links. The evolution of the optical link properties is monitored regularly during operations using optical link set up runs which scan gain and bias of the LLDs. As expected, a clear shift of the threshold current during Run 2 was seen. Important correction factors in this measurement come from the temperature and the fluence seen by the device. The data are compared to simulation for periods with constant operating temperature and very good agreement between data and simulation was found. CMS observed occupancy dependent inefficiencies in the Strip Tracker during 2015 and 2016. This was eventually traced to saturation effects in the pre-amplifier of the APV25 chip caused by slower than anticipated discharge times at low temperatures. A change of a configuration register cured the problem. Residual dynamic inefficiencies which are observed as function of increasing instantaneous luminosity are likely caused by heavily ionizing particles (HIP). The rate of HIP-like events was measured in special runs with non-zero suppressed readout. References [1] M. Garcia-Sciveres et al.,Nucl. Instrum. Methods Phys. Res A636 (2011) S155, doi:10.1016/j.nima.2010.04.101. [2] ATLAS Collaboration, Radiation induced effects in the ATLAS Insertable B-Layer readout chip, ATL-INDET-PUB-2017-001 (CERN, Geneva, 2017). [3] F. Faccio, G. Cervelli, IEEE Trans. Nucl. Sci. 52 (2005) 2413, doi:10.1109/TNS.2005.860698. [4] L. Gonella, et al.,Nucl. Instrum. Methods Phys. Res A582 (2007) 750, doi:10.1016/j.nima.2007.07.068. [5] M. Backhaus, JINST 12 (2017) P01011, doi:10.1088/1748-0221/12/01/P01011. [6] Braccini et al., The new Bern Cyclotron laboratory for radioisotope production and research, Proc. 2nd Int. Particle Conf., IPAC 2011, San Sebastian, Spain, September 4–9, 2011, pp. 3619–3621. [7] G. Balbi et al.,JINST 6(2020) P06023, doi:10.1088/1748-0221/15/06/P06023. [8] M. Menouni et al.,JINST 8(2013) C02026, doi:10.1088/1748-0221/8/02/C02026. [9] T. Calin et al.,IEEE Trans. Nucl. Sci. 43 (1996) 2874, doi:10.1109/23.556880. 119
6. Impact of radiation on electronics and opto-electronics [10] F. Márquez et al.,IEEE Trans. Nucl. Sci. 62 (2015) 1609, doi:10.1109/TNS.2015.2456831. [11] E. López-Morillo et al.,IEEE Trans. Nucl. Sci. 65 (2018) 1943, doi:10.1109/TNS.2018.2840326. [12] F.R. Palomo et al., Simulation of SET/SEU in microelectronics: AFTU software, Radiation Effects in the LHC Experiments and Impact on Operation and Performance, CERN, 11–12 February 2019, proc. in preparation (2020), Indico. [13] M. Huhtinen and F. Faccio, Nucl. Instrum. Methods Phys. Res A450 (2000), 155, doi:10.1016/S0168-9002(00)00155-8. [14] T. Sjöstrand et al.,Comput. Phys. Commun. 191 (20115) 159, doi:10.1016/j.cpc.2015.01.024, arXiv:1410.3012 [hep-ph]. [15] I. Dawson, Pos Vertex2012 (2013) 015, doi:10.22323/1.167.0015. [16] A. Ferrari et al., FLUKA: A multi-particle transport code, CERN-2005-010, INFN/TC-05/11, SLAC-R-773, (CERN, Geneva, 2005), doi:10.5170/CERN-2005-010. [17] A. Dafinca et al.,JINST 9(2014) C01050, doi:10.1088/1748-0221/9/01/C01050. [18] I. Dawson et al.,JINST 14 (2019) P07014, doi:10.1088/1748-0221/14/07/P07014. [19] D.G. Charlton et al.,Nucl. Instrum. Methods Phys. Res A456 (2001) 300, doi:10.1016/S0168-9002(00)00666-5. [20] J. Beringer et al.,Nucl. Instr. Meth. A435 (1999) 375. doi:10.1016/S0168-9002(99)00570-7. [21] ATLAS TRT Collaboration, JINST 3(2008) P02013, doi:10.1088/1748-0221/3/02/P02013. [22] ATLAS TRT Collaboration, JINST 3(2008) P02014, doi:10.1088/1748-0221/3/02/P02014. [23] ATLAS TRT Collaboration, JINST 3(2008) P10003, doi:10.1088/1748-0221/3/10/P10003. [24] ATLAS TRT Collaboration, JINST 3(2008) P06007, doi:10.1088/1748-0221/3/06/P06007. [25] G. Crone et al.,Nucl. Instrum. Methods Phys. Res A623 (2010) 534, doi:10.1016/j.nima.2010.03.061. [26] ATLAS Collaboration, Particle identification performance of the ATLAS Transition Radiation Tracker, ATLAS-CONF-2011-128, (CERN, Geneva,2011). [27] ATLAS Collaboration, Interactive web pages for radiation environment exploration of ATLAS ATL-SOFT-PUB-2020-003, 2020. [28] T. Åkesson et al., Implementation of the DTMROC-S ASIC for the ATLAS TRT detector in a 0.25 µm CMOS technology, 2002 IEEE Nuclear Science Symposium Conference Record, Norfolk, VA, USA, 2002, pp. 549-553 vol.1, doi: 10.1109/NSSMIC.2002.1239373. [29] F. Faccio et al., SEU effects in registers and in a dual-ported static RAM designed in a 0.25µm CMOS technology for applications in the LHC, Proc. 5th Conference on Electronics for LHC Experiments, Snowmass (CO), USA, 20–24 Sep 1999, CERN-LHCC-99-033, CERN-99-09, (CERN, Geneva, 1999), pp.571-575, doi:10.5170/CERN-1999-009.571. [30] S. Chatrchyan et al. [CMS Collaboration], JINST 3(2008) S08004, doi:10.1088/1748-0221/3/08/S08004. [31] A. Dominguez et al. [CMS Collaboration], CMS technical design report for the pixel detector upgrade, CMS-TDR-011, CERN-LHCC-2012-016, (CERN, Geneva, 2012), doi:10.2172/1151650. [32] CMS Collaboration, JINST 14 (2019) P10017, doi:10.1088/1748-0221/14/10/P10017. [33] H.-C. Kästli, Nucl. Instrum. Methods Phys. Res A565 (2006) 188, doi:10.1016/j.nima.2006.05.038,arXiv:physics/0511166. [34] J. Hoss et al.,JINST 11 (2016) C01003, doi:10.1088/1748-0221/11/01/C01003. [35] F. Faccio et al.,Failure of DC-DC modules in the CMS pixel system during the 2017 run, (CERN, Geneva, 2018). [36] R. Macias et al.,IEEE Trans. Nucl. Sci. 52 (2005) 1488, doi:10.1109/TNS.2005.855812. [37] K. Gill et al.,IEEE Trans. Nucl. Sci. 49 (2002) 2923, doi:10.1109/TNS.2002.805422. 120
CERN Yellow Reports: Monographs, CERN-2021-001 [38] M. Jansova, Search for the supersymmetric partner of the top quark and measurements of cluster properties in the silicon strip tracker of the CMS experiment at Run2 Ph.D. thesis, University of Strasbourg, 2018. [39] CMS Tracker Collaboration, Nucl. Instrum. Methods Phys. Res A543 (2005) 463, doi:10.1016/j.nima.2004.11.049. 121