Alice TPC Upgrade Technical Design Report - CERN Document Server

ALICE-TDR-016

CERN-LHCC-???? February 2, 2015

CERN Graphic Charter: use of the outline version of the CERN logo Colour reproduction The badge version must only be reproduced on a plain white background using the correct blue: Pantone: 286 CMYK: 100 75 0 0 RGB: 56 97 170 Web: #3861AA

Clear space A clear space must be respected around the logo: other graphical or text elements must be no closer than 25% of the logo’s width. Placement on a document Use of the logo at top-left or top-centre of a document is reserved for official use.

Where colour reproduction is not faithful, or the background is not plain white, the logo should be reproduced in black or white – whichever provides the greatest contrast. The outline version of the logo may be reproduced in another colour in instances of single-colour print.

Addendum to the

This is an output file created in Illustrator CS3

Technical Design Report

Upgrade of the ALICE Time Projection Chamber The ALICE Collaboration∗

02/02/2015

CERN-LHCC-2015-002 / ALICE-TDR-016-ADD-1

for the

Copyright CERN, for the benefit of the ALICE Collaboration. This article is distributed under the terms of Creative Commence Attribution License (CC-BY-3.0), which permits any use provided the original author(s) and source are credited.

∗ See

list of authors in CERN-LHCC-2013-020

Minimum size Print: 10mm Web: 60px

Executive summary The upgrade of the ALICE Time Projection Chamber (TPC) was described in a detailed Technical Design Report (TDR) [1]. The key objective of the upgrade is the replacement of the present MWPC-based readout chambers by detectors that allow continuous operation without active ion gating. In the TDR, we propose a solution that employs stacks of four Gas Electron Multipliers (GEMs) and demonstrate that it fulfills the design specifications, in particular in terms of intrinsic position resolution, energy resolution, and ion backflow. A two-stage online reconstruction and calibration scheme was presented that allows efficient online track reconstruction and data compression, including correction of the spacecharge distortions that are induced by the residual ion backflow from the readout chambers into the drift volume. It was noted, however, that the performance figures that were achieved, although being within the limits of the design specification defined in [1], may leave little margin to be able to cope with possible unforeseen effects. In this document we demonstrate with additional results from both detector R&D and simulations that the technological solution chosen in the TDR has sufficient safety margin for a successful campaign with the upgraded detector in RUN 3 and beyond. A summary of simulation studies on the physics performance is presented in Chap. 1. As a test case, the sensitivity of a jet analysis to the charged-particle momentum resolution at high transverse momentum (pT ) was studied. It was found that the charged-jet momentum resolution deteriorates by about 50% if the momentum resolution at high pT is degraded by a factor of two. The charged-jet resolution in this case is still as good as 3.5% at 180 GeV/c which does not compromise any of the physics goals. − + A study of the Λ+ c →pK π measurement demonstrates that a dE/dx resolution of about 8.5% in the TPC is sufficient and would not lead to a notable loss in significance with respect to the nominal dE/dx resolution of about 7%. Similar conclusions arise from studies of J/ψ→e+ e− . Measurements with a fullsize quadruple GEM IROC prototype at the PS test beam confirm that the required dE/dx performance can be achieved with settings that imply a local energy resolution of 14% of the 55 Fe peak, as shown in Sec. 4.3.

In Chap. 2, simulations of the tracking performance of the upgraded TPC are shown. They extend similar studies presented in the TPC Upgrade TDR from a collision rate of 70 kHz to 200 kHz, and from an ion backflow of 1% to 2% at a gas gain of 2000, corresponding to an increase of the ion space-charge density parameter ε from 20 to 40. It is demonstrated that the TPC tracking performance remains unaffected for collision rates up to 100 kHz, which is twice the nominal interaction rate, and gradually decreases for higher collision rates. Moreover, the online data-compression scheme as well as the calibration of the space-charge distortions is validated for an ion backflow of 2% without a notable loss of performance, in particular of the combined ITS-TPC momentum resolution. It is expected that inclusion of the Transition Radiation Detector (TRD) into the global momentum fit will further reduce the sensitivity to residual TPC space-charge distortions. This scheme will already be employed in RUN 2. We conclude that the design specifications for the new TPC readout system as presented in the TDR, in particular the requirement of a local energy resolution better than 12% and an ion backflow of less than 1% at a gas gain of 2000, can be considered conservative in the light of the present studies and include a

i

ii

The ALICE Collaboration

substantial safety margin. Several different readout technologies have been studied to meet these requirements. These include mainly micro-pattern gas detectors, in particular GEMs of different types and Micromegas. Commonly used triple GEM detectors do not provide sufficient ion backflow suppression, as shown in [1]. Therefore, comprehensive studies of quadruple GEM systems were performed which lead to the baseline solution presented in the TDR. It consists of a combination of standard (S) and large hole pitch (LP) GEM foils, i.e. S-LP-LP-S. After optimization of the voltage settings, a working point was identified with an ion backflow of about 0.7% at an energy resolution of 12%, which is well within the design specifications. Further detailed studies of the S-LP-LP-S system are presented in this document and lead to a further optimization of the working point. In particular, it was found that the same performance figures can be reached with reduced transfer fields and GEM voltages, leading to the new baseline settings presented in Sec. 3.1. A large number of other configurations of quadruple GEMs containing foils with different hole pitch were studied and presented in Sec. 3.1.2. Indeed, a configuration including a foil with small hole pitch in layer four (S-S-LP-SP) was identified that provides a still lower ion backflow of 0.5% at an energy resolution of 12%. However, further qualification of this configuration is needed, in particular with respect to its discharge behaviour, before a final decision in favor of this configuration can be made. All other configurations under study showed a performance that is similar or worse than the baseline S-LP-LP-S system. COBRA-type GEMs were studied already in the ALICE Upgrade TDR and rejected because they provide insufficient energy resolution [1]. Detailed studies of the discharge behaviour of quadruple GEMs are presented in Sec. 3.2. Upper limits for the discharge probability of a S-LP-LP-S system upon irradiation with alpha particles, operated at a gas gain of 2000 with voltage settings that are optimized for low ion backflow, are found to be of the order of 10−10 . This number is compatible with that for standard triple GEM systems that are operated routinely in high-rate experiments. In a test-beam campaign at the SPS, a full-size quadruple GEM Inner Readout Chamber (IROC) was exposed to an intense hadron flux from a secondary production target (see Sec. 4.2). The measured discharge probability of (6.4 ± 3.7) × 10−12 translates to about 600 expected discharges in the whole TPC, or 4 discharges per GEM stack, during one month of Pb-Pb running at 50 kHz. Such small numbers are not expected to cause any damage to the detectors and are compatible with efficient and safe operation of the TPC in RUN 3 and beyond. Micromegas in combination with a double GEM pre-amplification stage (2GEM+MM) were also proposed in the TDR as an alternative technology for the new TPC readout chambers. Preliminary ion backflow values that are a factor 3-4 below the quadruple GEM results were reported, albeit not measured under the operational conditions of the ALICE TPC. More detailed studies of the ion backflow and energy resolution in a 2GEM+MM are presented in this document and indicate that the ion backflow is about 0.4% at an energy resolution of 12% (see Sec. 3.1.3), and therefore rather close to the performance of a quadruple GEM. Moreover, the discharge rate under irradiation with hadrons during the SPS testbeam campaign was found to be 2-3 orders of magnitude larger than that of a quadruple GEM system (see Sec. 4.2), implying the need for significant further R&D to minimize dead times and optimize spark protection. Solutions with Micromegas detectors were therefore not further considered as an alternative readout technology for the upgrade of the ALICE TPC. We therefore provide evidence that with the quadruple GEMs a solution exists which provides the performance needed for the ALICE TPC upgrade with considerable safety margins, and that the proposed solution is robust and capable of reliable long term operation.

Contents 1

Physics requirements and design specifications

1

2

Simulation and detector performance

5

2.1

Interaction rate dependence of the TPC tracking performance . . . . . . . . . . . . . . .

5

2.2

Space-charge distortion dependence of the TPC tracking performance . . . . . . . . . .

6

3

R&D with small prototypes

10

3.1

Studies of ion backflow and energy resolution . . . . . . . . . . . . . . . . . . . . . . .

10

3.1.1

Detailed characterization of S-LP-LP-S . . . . . . . . . . . . . . . . . . . . . .

10

3.1.2

Alternative quadruple GEM configurations . . . . . . . . . . . . . . . . . . . .

14

3.1.3

Prototype studies with a double GEM stack and a Micromegas . . . . . . . . . .

16

Discharge probability studies with small detectors . . . . . . . . . . . . . . . . . . . . .

18

3.2.1

Experimental setups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

18

3.2.2

Radiation sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

20

3.2.3

Study of single GEMs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

21

3.2.4

Study of triple GEMs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

21

3.2.5

Study of quadruple GEMs . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

25

3.2

4

Test beam studies with large-size prototypes

28

4.1

Prototype setups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

28

4.1.1

4-GEM IROC prototype . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

28

4.1.2

HIROC prototype . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

29

4.1.3

Yale prototypes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

30

Discharge studies with large-size prototypes . . . . . . . . . . . . . . . . . . . . . . . .

30

4.2.1

4-GEM IROC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

31

4.2.2

Yale prototypes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

32

4.2.3

HIROC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

33

4.2

iii

iv


4.3

4.4

Particle identification with large-size prototypes at the CERN PS . . . . . . . . . . . . .

34

4.3.1

Experimental setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

34

4.3.2

Gain uniformity and equalization . . . . . . . . . . . . . . . . . . . . . . . . .

35

4.3.3

PID analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

37

4.3.4

Gain scan with the 4-GEM IROC . . . . . . . . . . . . . . . . . . . . . . . . .

38

4.3.5

Ion backflow and energy resolution scan . . . . . . . . . . . . . . . . . . . . . .

39

Conclusions from large-size prototype tests . . . . . . . . . . . . . . . . . . . . . . . .

41

A TPC upgrade collaboration

43

B The ALICE Collaboration

47

References

55

List of Figures

57

List of Tables

59

Chapter 1

Physics requirements and design specifications Ungated operation of the ALICE TPC in the high-multiplicity environment of Pb–Pb collisions at 50 kHz after LS2 will lead to a considerable accumulation of positive ions in the drift volume that emerge from the amplification region. The resulting space-charge distortions must be kept within limits that allow efficient online track reconstruction and distortion corrections. As described in detail in [1], all specifications in terms of track reconstruction and matching efficiency as well as final momentum resolution are met if the number of back-drifting ions into the drift region per signal electron on the readout anode, the ion backflow, is ≤1%. In Chap. 2 we demonstrate that the same conclusion holds even if the ion backflow is 2%, and that in particular the space-charge distortion corrections can be performed without a notable degradation of the momentum resolution. In this chapter we discuss the impact of the transverse momentum (pT ) and dE/dx resolution on a few selected key observables of the ALICE physics programme after LS2. To this end, the nominal detector resolution was systematically degraded in the simulations to identify possible limits of feasibility for a given physics study. Observables at high transverse momentum are most sensitive to the charged-particle momentum resolution. While the study of inclusive charged-particle production at high pT is likely to be completed after RUN 2, the analysis of jets is one of the main physics goals of ALICE in RUN 3 and beyond. Fig. 1.1 shows the result of a study where the single-track momentum resolution at pT >10 GeV/c was degraded by a factor two with respect to the nominal combined momentum resolution of the present detector. In this case, the charged-jet momentum resolution would deteriorate by about a factor 1.5 above 120 GeV/c. However, the resolution at 180 GeV/c is still about 3.5%, which implies no limitation on the physics reach even in this extreme scenario. Since other observables on lower momentum scales are even less sensitive to the momentum resolution, we conclude that a deterioration of the transverse momentum resolution at high pT by a factor two would not impose a notable limitation on the ALICE physics programme after LS2. The result of a fast simulation is shown in Fig. 1.2 (left) to illustrate the role of the TPC in the momentum reconstruction. Clearly, inclusion of the TPC into a combined momentum fit improves the momentum resolution considerably as compared to a situation where only the ITS is used. On the other hand, omitting the track segments in the Inner Readout Chambers (IROCs) from the combined momentum fit hardly affects the momentum resolution. This relaxes the requirements for the precision of the distortion corrections in the IROCs, where the distortions are largest. It should be noted, however, that the IROC track information is important for dE/dx and matching to the ITS, even if part of the track is discarded for the final momentum fit. The right panel in Fig. 1.2 shows the relative performance deterioration in case 1


4 3.5

deteriorated p resolution T

3

nominal p resolution T

2.5 2

T

jet p resolution (GeV/c)

2

1.5 1 0.5 0

40

60

80

100

120

140

160 jet

p

T

180 (GeV/c)

jet

Figure 1.1: Charged-jet pT resolution as a function of pT . Results for the nominal jet pT resolution are compared to the case when the single-track pT resolution was degraded by a factor two for tracks with pT >10 GeV/c.

• Relative change due to the skipping of IROC in: • Primary tracks found In TPC with kTPCrefit flag • Same with correct match to ITS track • Fraction of correct ITS matches in all matches

pT (GeV/c)

OROC/IROC+OROC

σ(pT)/pT (%)

only the Outer Readout Chambers (OROCs) are used with respect to the full TPC including IROCs. No strong impact on the performance is observed for pT above 0.2 GeV/c. At low transverse momenta, the TPC reconstruction efficiency and accordingly the TPC-ITS matching efficiency show a strong decrease. This is explained by the track curvature since tracks below 0.2 GeV/c do not reach the outer radius of the TPC any longer. Below 0.1 GeV/c, tracks are fully confined in the IROCs. It should be noted that a typical•lower pT cut ofMC 0.15 GeV/c is applied in most analyses of ALICE. Same HijinParam (dN/deta=2100) reconstructed with physics and w/o IROC in the RefitInward step.

pT (GeV/c)

Figure 1.2: Left: Transverse momentum resolution for different combinations of ITS, IROC and OROC. Right: Ratio of the Probability of matching founs TPC track andtrack matching performances for OROC-only over full TPC (IROC+OROC). totracking correct ITS (e.g. TPC reconstruction efficiency excluded) NOT to be shown

The particle identification (PID) capability of the TPC via dE/dx is a crucial feature for the physics programme of the upgraded ALICE detector. In the original TPC TDR from the year 2000 [2], a chargedparticle multiplicity density dNch /dη = 8000 in central Pb–Pb collisions was assumed, and a relative dE/dx resolution of better than 10% was found to be sufficient to perform the initial ALICE physics programme. √ As a consequence of the significantly lower charged-particle density observed in central Pb–Pb at sNN = 2.76 TeV, a substantially better dE/dx resolution of about 7% was achieved in RUN 1. This number √ is consistent with Monte Carlo simulations. A similar dE/dx resolution is expected at sNN = 5.5 TeV at low collision rate, where a charged-particle multiplicity density dNch /dη =2000 (extrapolated from measurements in RUN 1) is assumed. At a collision rate of 50 kHz in RUN 3 and beyond, a degradation to about 7.5% at 50 kHz due to pile-up and cluster overlaps is expected, see Fig. 7.6 in [1].

TPC Upgrade TDR Addendum

3

Additional degradation may occur as a consequence of incomplete effective electron transparency (i.e. collection and extraction efficiencies) of the GEM system, which arises as a consequence of the optimization of the operational point with respect to ion backflow. The electron transparency can be characterized in terms of the energy resolution σ (55 Fe) of the 55 Fe photopeak. Simulations show that energy resolutions σ (55 Fe) of 12%, 14% and 16% correspond to a degradation of the dE/dx resolution from 7.5% to about 8%, 8.5% and 9%, respectively. In RUN 3 and beyond, the TPC dE/dx information will be essential to enhance the significance of rare observables that suffer from a low signal-to-background ratio. As a benchmark case, we study the im− + + pact of the TPC dE/dx resolution on the significance of the Λ+ c in the decay channel Λc → pK π , where TPC dE/dx information is used to enhance the purity of the decay track sample. The simulation and analysis details are described in [3]. Figure 1.3 shows the significance of the Λ+ c measurement in 1 central Pb–Pb events as a function of pT , for different TPC PID schemes . Clearly, the measurement greatly benefits from TPC dE/dx information. In comparison to the nominal dE/dx resolution, which corresponds to about 7% in this simulation at low interaction rate, the significance gradually decreases when the dE/dx resolution is degraded. We observe that the measurement is not significantly affected if the dE/dx resolution is 8.5%, i.e. does not exceed the nominal one by more than 20% (blue symbols in Fig. 1.3). Similar conclusions arise from the study of J/ψ → e+ e− which is shown in Fig. 1.4. The simulation and analysis details are described in [4]. In this analysis, the TPC dE/dx is used together with the TRD to select the electron sample. A degradation of the TPC dE/dx resolution by about 20%, i.e. to 8.5% will not lead to a notable decrease of the significance. To summarize, these arguments lead to design specifications of the upgraded TPC which imply that σ (dE/dx)/hdE/dxi ≤ 8.5%. This corresponds to an energy resolution of the GEM readout chambers of about 14% at the 55 Fe photopeak. As will be shown in Chap. 2, the proposed online reconstruction and calibration scheme was demonstrated to work without a notable loss in momentum resolution for an ion backflow of up to 2%. An onset of a breakdown of the procedure was not observed. Furthermore, a physics case that depends crucially on the momentum resolution of the present system could not be identified. Even with a momentum resolution that is a factor two worse at high pT , all physics goals can be reached. We conclude therefore that the design specifications of σ (55 Fe) ≤ 12% and an ion backflow ≤ 1% as stated in [1] include sufficient safety margin to ensure that the required physics performance of the upgraded ALICE detector can be reached.

1 The significance for the case with nominal PID performance (black markers) is slightly larger than reported in [3], because the decay topology selection was further optimized.


Significance

4

100

-1

No TPC PID ALICE upgrade, L int = 10 nb Degraded TPC PID(20%) Pb-Pb sNN = 5.5 TeV Degraded TPC PID (50%) Degraded TPC PID (100%) 0-20% centrality With full PID

90 80 70

Λ+c → pK π +

60 50 40 30 20 10

0

2

4

6

8

10

12

14

16

18

20

p (GeV/c) T

Significance [2.92-3.16 GeV/c2]

− + −1 as a function of p . Results are shown for a Figure 1.3: Significance of a Λ+ T c → pK π measurement at Lint = 10 nb successive degradation of the TPC dE/dx resolution compared to the nominal resolution of 7% (black). Also shown is the result without TPC dE/dx (red).

180

Pb-Pb, Centrality class 0-10%, L =10nb-1 int

160

Current TPC PID + TRD, LoI deteriorated TPC PID (20%) + TRD deteriorated TPC PID (50%) + TRD deteriorated TPC PID (100%) + TRD

140 120 100

ALICE upgrade

80 60 40 20 0 0

2

4

6

8 10 p (J/ψ )(GeV/c) T

→ e+ e−

Figure 1.4: Significance of J/ψ measurement in central Pb–Pb collisions at Lint = 10 nb−1 as a function of pT . Results are shown for a successive degradation of the TPC dE/dx resolution compared to the nominal resolution of 7% (red).

Chapter 2

Simulation and detector performance Simulation studies of the tracking performance and the online calibration scheme are described in this chapter. The present results extend similar studies in the TPC Upgrade TDR to higher interaction rates and larger ion space-charge densities. In summary, we observe that the tracking performance does not deteriorate for interaction rates up to 100 kHz in Pb–Pb, which is twice the nominal interaction rate assumed for the ALICE upgrade programme. Moreover, the proposed two-stage online reconstruction and calibration scheme can be successfully applied up to ion space-charge parameters of ε = 40, corresponding to an ion backflow of 2% at gain 2000, without a notable loss in momentum resolution. For the present simulations the standard AliRoot software is used. It implements a full detector description with microscopic transport based on GEANT3. The same software and strategy was used for the studies presented in the TPC upgrade TDR [1], details can be found there.

2.1

Interaction rate dependence of the TPC tracking performance

The design of the ALICE TPC and the reconstruction algorithms were optimized to cope with charged particle multiplicities of up to hdNch /dη i=8000 in central Pb–Pb collisions, see [2]. After the start of the LHC, it turned out that the charged-particle densities in Pb–Pb are substantially lower. The present estimates, which are the basis of the design studies for the ALICE upgrade, assume that charged-particle √ densities of hdNch /dη i=2000 (500) are reached in central (minimum bias) Pb–Pb collisions at sNN = 5.5 TeV. On the other hand, operation of the TPC at collision rates of 50 kHz in Pb–Pb implies significant event pile-up, i.e. on average 5 minimum bias events piling up within the drift time window of the TPC of 100 µs, and correspondingly increased detector occupancy. In order to study the robustness of the TPC and TPC-ITS track reconstruction scheme against event pile-up at high interaction rate, the interaction rate was varied from 20 kHz to 200 kHz in simulations. Different interaction rates are simulated by superimposing different numbers of minimum bias Pb–Pb pile-up interactions on top of a given central Pb–Pb event. Since the focus of this study is on the tracking performance at high occupancy, the effect of space-charge distortions is neglected here. Figure 2.1 shows the TPC standalone tracking efficiency (left) and the TPC-ITS track matching efficiency (right) as a function of 1/pT for different interaction rates. Only a small decrease of the performance is observed up to an interaction rate of 100 kHz which starts to become more pronounced for higher interaction rates. This decrease is due to the increasing occupancy and therefore decreasing efficiency to separate overlapping clusters from different tracks. However, the TPC standalone reconstruction efficiency at 200 kHz is still above 95%. In Fig. 2.2 the transverse momentum resolution is shown as a function of 1/pT for the different interac5

6


TPC tracking efficiency

tion rates. The left panel shows the performance of the combined TPC-ITS tracks and the right panel that of TPC standalone tracks. The conclusion is similar to that for the tracking performance. The pT resolution is almost unaffected for interaction rates up to 100 kHz and starts to deteriorate slightly for higher interaction rates. At high pT , the combined momentum resolution remains unaffected up to 200 kHz. 1

TPC-ITS matching efficiency

1

0.99

0.98

0.97

0.98 0.96 0.94 0.92 0.9

0.96

0.88

0.95

central event at 20 kHz


central event at 40 kHz central event at 60 kHz


0.86





0.84






0.94

0.82 0.93 0

0.5

1

1.5

2

2.5

0.8 0

3 3.5 4 1/p (GeV/c)-1


0.5

1

1.5

2

2.5

3

T

3.5 4 1/p (GeV/c)-1 T

0.025








TPC standalone

T

0.014


σ1/p (GeV/c)-1

0.016

T

σ1/p (GeV/c)-1

Figure 2.1: TPC standalone track reconstruction efficiency (left) and TPC-ITS track matching efficiency (right) for different interaction rates.

0.012

0.02


0.01

0.015

0.008 0.01

0.006 0.004

0.005 TPC + ITS

0.002










0 0

0.1

0.2

0.3

0.4

0.5

0.6

0.7 0.8 0.9 1 1/p (GeV/c)-1 (MC) T

0 0

0.1

0.2

0.3

0.4

0.5

0.6

0.7 0.8 0.9 1 1/p GeV/c)-1 (MC) T

Figure 2.2: Transverse momentum resolution for TPC-ITS combined tracks (left), and TPC standalone tracks (right) for different interaction rates.

2.2

Space-charge distortion dependence of the TPC tracking performance

The occurence of space-charge distortions of the drift field due to backdrifting ions from the amplification region poses a major challenge for the TPC reconstruction and calibration scheme after LS2. Moreover, the initial reconstruction steps need to be performed online to achieve sufficient data compression for permanent storage. To this end, a two-stage online reconstruction and calibration scheme was proposed in the TPC Upgrade TDR [1]. This scheme foresees a simplified distortion correction in the first reconstruction stage, that employs a static average correction map scaled by a single time-dependent scale parameter. The scale parameter is


7

proportional to the total number of charged particles produced in the preceeding 160 ms, i.e. the time it takes ions to drift from the readout chambers to the central electrode, and is straight forward to extract from integrals over readout chamber currents or detector raw data. The aim of the first reconstruction stage is to find clusters and associate them to tracks with high efficiency, thus allowing to reject raw data and clusters that do not belong to physics tracks. In the second reconstruction stage, residual track distortions are corrected and the final detector resolution is restored. For this purpose, the local residuals are determined in a fine grid of volume elements of the TPC by comparison of the TPC cluster positions to an external reference track. The reference track is obtained by interpolation of track segments from the ITS and the TRD. The precision of the procedure is limited by the temporal fluctuations of the space-charge distributions, which implies that a given spacecharge configuration can be considered as static over only 5 ms. This limits the number of events that are available to determine a given residual distortion pattern to about 250 as well as the granularity of the grid to about 72,000 volume elements. This approach and its implementation are described in detail in Secs. 8.3 and 8.4 in [1]. It was demonstrated that efficient online data compression is feasible and that space-charge corrections can be performed without a notable loss of momentum resolution for distortions that correspond to an ion backflow of 1% at a gain of 2000 (i.e. ε = 20) in Pb–Pb at 50 kHz. To investigate possible limits of the procedure and identify the safety margin for operational conditions beyond the design specifications, distortions according to ε parameters between 0 and 40 were simulated and the performance of the procedure was studied.

1.002



The results after the first reconstruction stage, where only a scaled average distortion correction map is applied, are shown in Fig. 2.3. The TPC track reconstruction efficiency (left) shows practically no variation up to ε = 40. The same is true for the average number of associated clusters per track (not shown). This is important because the proposed online data compression scheme implies that only clusters associated to tracks will be kept for further processing. These clusters will be written to temporary storage.1 The TPC-ITS track matching efficiency (right) exhibits a continuous degradation with increasing ε which can be explained by the residual distortions that are still present after the first reconstruction stage. It should be noted though that the algorithms are not optimized for tracks with residual distortions, which can be improved in the future. Moreover, the final TPC-ITS matching is performed in the second reconstruction stage, when residual distortion corrections are applied.

st

After 1 stage of reconstruction

1

0.998

0.996

1

0.95

0.9

0.85 no space-charge

no space-charge

space-charge, ∈=20


0.994

0.8





0.992


0.75


st

After 1 stage of reconstruction

0.99 0

0.5

1

1.5

2

2.5

3

3.5 4 1/p (GeV/c)-1

0.7 0

0.5

1

1.5

T

2

2.5

3

3.5 4 1/p (GeV/c)-1 T

Figure 2.3: TPC standalone reconstruction efficiency (left) and TPC-ITS track matching efficiency (right). Results are shown without distortions and with residual distortions after the first (synchronous) reconstruction stage. 1 The

option to keep all clusters at this stage for further processing is presently under study.

8


In Fig. 2.4 the transverse momentum resolution is shown as a function of 1/pT for different space-charge densities without and with residual distortions expected after the first reconstruction stage. The left panel shows the case for TPC-ITS combined tracks, the right panel for TPC standalone tracks. As expected, due to the residual distortions, a decrease with increasing space-charge density is observed. For calibration purposes the pT resolution after the first reconstruction stage is sufficient. 0.035

st

TPC + ITS, after 1 stage of reconstruction

T

0.03

σ1/p (GeV/c)-1

T

σ1/p (GeV/c)-1

0.035

no space-charge

0.025

st

TPC standalone, after 1 stage of reconstruction

0.03

0.025

space-charge, ∈=20 space-charge, ∈=25 space-charge, ∈=30

0.02

0.02


0.015

0.015

0.01

0.01

0.005

0.005

no space-charge space-charge, ∈=20 space-charge, ∈=25 space-charge, ∈=30 space-charge, ∈=40

0 0

0.1

0.2

0.3

0.4

0.5

0.6

0 0

0.7 0.8 0.9 1 1/p (GeV/c)-1 (MC)

0.1

0.2

0.3

0.4

0.5

0.6

T

0.7 0.8 0.9 1 1/p (GeV/c)-1 (MC) T

Figure 2.4: Transverse momentum resolution for TPC-ITS combined tracks (left), and TPC standalone tracks (right) for different space-charge densities without distortions and with residual distortions after the first (synchronous) reconstruction stage.

1.002 nd

After 2



Figure 2.5 shows the tracking performance after applying the correction for the residual distortions measured with the ITS-TRD interpolation method, as described in Sec. 8.4.1 of the TPC Upgrade TDR [1]. This corresponds to the expected performance after the second reconstruction stage. The TPC standalone track reconstruction efficiency is shown on the left, the TPC-ITS track matching efficiency on the right. In particular, the TPC-ITS matching efficiency is fully recovered by the residual distortion calibration scheme for all ε parameters under study.

stage of reconstruction

1

0.998

0.996

1

0.95

0.9 no space-charge space-charge, ∈=20

0.85

space-charge, ∈=25 space-charge, ∈=30

no space-charge space-charge, ∈=20

0.994


0.8

space-charge, ∈=25 space-charge, ∈=30

0.992

0.75


nd

After 2

0.99 0

0.5

1

1.5

2

2.5

3

3.5 4 1/p (GeV/c)-1 T

0.7 0

0.5

1


1.5

2

2.5

3

3.5 4 1/p (GeV/c)-1 T

Figure 2.5: TPC standalone track reconstruction efficiency (left) and TPC-ITS track matching efficiency (right). Results are shown without distortions and with residual distortions after the second (asynchronous) reconstruction stage.


9

0.035

nd

TPC + ITS, after 2


T

0.03

σ1/p (GeV/c)-1

T

σ1/p (GeV/c)-1

0.035

no space-charge

0.025


nd

0.03

TPC standalone, after 2


0.025


0.02


0.02


0.015

0.015

0.01

0.01

0.005

0.005

no space-charge space-charge, ∈=20 space-charge, ∈=25 space-charge, ∈=30 space-charge, ∈=40

0 0

0.1

0.2

0.3

0.4

0.5

0.6

0.7 0.8 0.9 1 1/p (GeV/c)-1 (MC) T

0 0

0.1

0.2

0.3

0.4

0.5

0.6

0.7 0.8 0.9 1 1/p (GeV/c)-1 (MC) T

Figure 2.6: Transverse momentum resolution for TPC-ITS combined tracks (left), and TPC standalone tracks (right) for different space-charge densities without distortions and with residual distortions after the second (asynchronous) reconstruction stage.

In Fig. 2.6 the transverse momentum resolution after correcting for residual distortions is shown, corresponding to the second reconstruction stage. The TPC standalone tracks (right panel) show a slight degradation of the pT resolution for increasing ε parameters. In the case of TPC-ITS combined tracks (left panel) which are relevant for physics analysis there is only a very small dependence of the momentum resolution observed for space-charge densities resulting from ε = 0 − 40. These small deteriorations do not limit the physics performance of the upgraded detector. It should be noted that the same tracking code is used as for undistorted tracks and no optimizations for the case of residual distortions are applied. A dedicated optimization will likely lead to further improvement in the future. Furthermore, the present momentum fit employs the present ITS, and no TRD. Inclusion of the upgraded ITS and the TRD will further improve the momentum resolution, and hence reduce the sensitivity of the combined momentum fit to residual distortions in the TPC. In conclusion, we demonstrate that the occupancy due to event pile-up has no notable impact on the tracking performance of the TPC for collision rates of up to 100 kHz, i.e. twice the nominal interaction rate after LS2. This finding is consistent with the fact that the TPC was initially optimized for chargeparticle densities of hdNch /dη i=8000. Furthermore, the performance of the two-stage TPC online track reconstruction and calibration procedure was studied for distortion parameters ε up to 40 and no significant deterioration was found. In particular, an onset of a sudden breakdown was not observed. This implies a safety margin of at least a factor two with respect to the design specifications for the maximal ion backflow of 1% stated in [1].

Chapter 3

R&D with small prototypes In this chapter, we present results of recent R&D efforts with small-size protype detectors. In the first part, detailed studies of ion backflow and energy resolution with the baseline S-LP-LP-S system as well as with alternative quadruple GEM configurations and a hybrid system employing a double GEM plus Micromegas readout are shown. The results demonstrate that the baseline S-LP-LP-S is a viable solution for the upgrade of the ALICE TPC, and that other technologies offer only little room for further improvement. In the second part, the results of a thorough study of the discharge behaviour of quadruple GEMs upon irradiation with alpha particles is presented. It is demonstrated that the discharge rates are compatible with those of commonly used “standard” triple GEM systems, and that the operational point of the baseline quadruple GEM implies sufficient margin for a safe operation of the detector.

3.1

Studies of ion backflow and energy resolution

In this section we summarize systematic studies of ion backflow and energy resolution in small prototypes of various quadruple GEM systems, notably combinations of GEMs with different hole geometries. In particular, standard GEMs with 140 µm hole pitch (S) as well as GEMs with larger (280 µm, LP) and smaller (90 µm, SP) hole pitches were investigated. All studies with quadruple GEMs presented here are performed with a dedicated test setup that consists of a relatively large drift volume with 60 mm drift length and a 10 × 10 cm2 quadruple GEM stack mounted inside a gas-tight aluminum body. To avoid drift field inhomogeneities, the drift volume is surrounded by three 16 mm wide aluminum strips, separated by 2 mm. The transfer gaps between GEMs and the induction gap between GEM4 and anode are 2 mm each. The anode is subdivided into 3 × 5 = 15 pads. One of them is connected to an analog readout chain with charge sensitive preamplifier, shaping amplifier (500 ns shaping time) and peak sensing ADC for the energy resolution measurement. The other pads are connected and commonly read out by a Keithley high-precision current meter with a resolution of about 2 pA. The cathode with typical potentials ∼6 kV is read out by a pico-amperemeter with a resolution of about 1pA. All potentials are provided by independent HV channels (12 in total) to allow for maximum flexibility. An 55 Fe source is used to irradiate the detector either from the top or from the side. 3.1.1

Detailed characterization of S-LP-LP-S

Ion backflow and energy resolution In [1] the S-LP-LP-S configuration was proposed as the baseline for the new TPC readout chambers. A working point was identified which complies with the specifications in terms of gas gain, ion backflow and energy resolution. In this document, we present a more detailed characterization of the S-LP-LP-S system. In Fig. 3.1 detailed scans as a function of the transfer fields ET2 and ET3 are shown. The gain of 10


11

the system is set to 2000 for each field setting by adjusting the voltages on GEM3 and GEM4 (∆UGEM3 and ∆UGEM4 ), while keeping the ratio ∆UGEM3 /∆UGEM4 = 0.8. The corresponding value for ∆UGEM4 is also shown in Fig. 3.1. The data confirm the working point presented in [1] at ET2 = 2 kV/cm and ET3 = 0.1 kV/cm, where an ion backflow of 0.66% and an energy resolution of 11% are observed. We observe that there is little variation of ion backflow and energy resolution for ET2 ≥ 2 kV/cm, which provides flexibility for further optimization. In particular, moving to values of ET2 = 3 kV/cm and ET3 = 1 kV/cm provides the same ion backflow, energy resolution and gas gain at reduced ∆UGEM3 and ∆UGEM4 , which may be favored in terms of operational stability. For these field settings, i.e. ET3 = 1 kV/cm, a further scan as a function of ET1 and ET2 was performed, see Fig.3.2. Little variation of ion backflow and energy resolution is observed for ET1 ≥ 2 kV/cm which would allow a reduction of ET1 to 2 kV/cm. This leads to new baseline settings for the readout chambers that are summarized in Tab. 3.1. A systematic variation of the transfer gap between GEM3 and GEM4 showed little effect on the observed ion backflow, as shown in Fig. 3.3. σ(55Fe) (%)

ET3(V/cm)

ET3(V/cm)

IB (%)

ET1 4 kV/cm Eind 4 kV/cm ΔUGEM1 275 V ΔUGEM2 240 V ET2(V/cm)


ET3(V/cm)

ΔUGEM4 (V)


Figure 3.1: Ion backflow (top left) and energy resolution (top right) in a quadruple S-LP-LP-S GEM in Ne-CO2 -N2 (90-10-5) as a function of ET2 and ET3 . The settings for ET1 , Eind , ∆UGEM1 and ∆UGEM2 are indicated in the figures. Also shown is ∆UGEM4 for each setting adjusted to provide a gain of 2000, while ∆UGEM3 /∆UGEM4 is kept at 0.8 (bottom).

GEM orientation and alignment Simulations indicate that there can be a strong dependence of the ion backflow on the relative orientation of the holes in consecutive GEM layers. In particular, a notable increase of ion backflow may occur if the GEM holes are aligned. This is illustrated in Fig. 3.4 where the results of a Garfield++ simulation of two GEM layers are shown. While the gas gain shows no dependence on the lateral displacement of

12


σ(55Fe) (%)

ET2(V/cm)

ET2(V/cm)

IB (%)

ET3 1 kV/cm Eind 4 kV/cm ΔUGEM1 275 V ΔUGEM2 240 V


ET1(V/cm)

ET1(V/cm)

ET2(V/cm)

ΔUGEM4 (V)


ET1(V/cm)

Figure 3.2: Ion backflow (top left) and energy resolution (top right) in a quadruple S-LP-LP-S GEM in Ne-CO2 -N2 (90-10-5) as a function of ET1 and ET2 . The settings for ET3 , Eind , ∆UGEM1 and ∆UGEM2 are indicated in the figures. Also shown is ∆UGEM4 for each setting to provide a gain of 2000, while ∆UGEM3 /∆UGEM4 is kept at 0.8 (bottom). The new baseline settings are marked by red circles.

New baseline settings TDR settings

∆UGEM3 /∆UGEM4

∆UGEM1 (V)

∆UGEM2 (V)

∆UGEM3 (V)

∆UGEM4 (V)

ET1 (kV/cm)

ET2 (kV/cm)

ET3 (kV/cm)

Eind (kV/cm)

0.8 0.8

275 270

240 250

254 270

317 340

2 4

3 2

1 0.1

4 4

Table 3.1: Typical voltage settings optimized for energy resolution and ion backflow. The voltages ∆UGEM3 and ∆UGEM4 are adjusted to achieve a gas gain of 2000. Also shown are the settings that were proposed in the TDR.

GEM holes in consecutive layers, a strong effect is observed for the ion backflow, in particular for strong transfer fields. This behaviour can be understood in terms of the low diffusion of ions which makes them tend to follow a given field line. In case of displacement, ions emerging from a GEM hole in the lower layer end on the bottom side of the upper GEM, while they enter a hole in the upper GEM in case of alignment. In the case of electrons, the much larger diffusion diminishes such correlations, and no effect on the gas gain is observed. Therefore, misalignment of GEM holes is preferred in terms of minimal ion backflow, or at least precautions should be taken to avoid accidental alignment. While accidental alignment of two GEM layers over the whole surface is quite improbable, a slight relative rotation of two consecutive layers in the GEM plane can lead to macroscopic interference patterns, featuring regions of high and low optical transparency, as illustrated in Fig. 3.5. Such patterns may lead to variations of the GEM characteristics over the detector surface. On the other hand, rotation of two consecutive layers by 90o prevents interference patterns, because the GEM hole pattern obeys a 60o rotational symmetry. This results in a randomization of relative hole positions over short distances, as shown in Fig. 3.6, and enhances the uniformity of the detector performance across the active area.

IB (%)


13

1 0.9 0.8 0.7 0.6 0.5 0.4 0.3

S-LP-LP-S, Ne-CO2-N2 (90-10-5)

0.2

1 mm

0.1

2 mm 3 mm

0 0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

ET3 (kV/cm)

Figure 3.3: Ion backflow in a quadruple S-LP-LP-S GEM in Ne-CO2 -N2 (90-10-5) as a function of ET3 for different sizes of the transfer gap between GEM 3 and GEM 4.

Figure 3.4: Garfield++ simulation of the gas gain (left) and ion backflow (right) in a double GEM system operated in Ne-CO2 N2 (90-10-5). The results are shown as a function of the lateral GEM hole offset between the two layers.

Electron and ion transport in a quadruple GEM system In the course of this R&D effort, a detailed discription of the propagation of single electrons and ions in the GEM stack has been achieved. A simulation of the electron and ion transport characteristics in an SLP-LP-S quadruple GEM system was performed with Garfield++. Typical voltage settings corresponding to the TDR settings (see Tab. 3.1) are applied. The total gain in this calculation is 1830, a summary of the results is presented in Tab. 3.2. For each GEM layer, the collection and extraction efficiencies (εcoll and εextr ), the gas multiplication (M) and the effective gain (G = εcoll · M · εextr ) are calculated. For a single electron arriving from the drift volume, also the average number of electrons entering and exiting the GEM (ne,in and ne,out ), the number of produced electron-ion pairs (ne−ion ) and the number of ions drifting back into the drift volume (nion,back ) are given for each layer. In this configuration and for typical voltage settings, most of the ions are produced in GEM4, which are efficiently blocked by the large-pitch

14

GEM alignment


!  small&'lts&lead&to&interference&pa2ern&in&„parallel“&GEM&orienta'on&

10

Figure 3.5: Left: GEM Optical alignment transparency of two standard GEM foils. Right: Illustration of the interference pattern that occurs when the foils are slightly rotated.

!  90°%orienta-on%leads%to%random%alignment%

Figure 3.6: Left: Optical transparency of two standard GEM foils after rotation of one foil by 90o . Right: Illustration of the randomization• of Random%alignment%prefered%in%terms%of%uniform%detector%response% the relative hole positions.

•  quan-ta-ve%studies%ongoing%

12

GEM2 and GEM3 foils. On the other hand, the large-pitch foils have rather low εcoll which requires a typical gain of about 10 on GEM1 to provide sufficient energy resolution. GEM2 and GEM3 are operated at low effective gains (G = 1 − 2) and essentially pass the pre-amplified electron cloud to GEM4, which is operated at an effective gain of about 150. Also shown in Tab. 3.2 is, for each layer, the number and fraction of produced ions that drift back into the drift volume. Most of the remaining back-drifting ions are produced in GEM1 and GEM2. Further ion backflow suppression could be achieved by reduction of the gain in GEM1, however, at the expense of a degraded energy resolution. The relative contribution from each GEM layer to the total ion backflow found in simulation is in fair agreement with differential measurements of the ion currents in a 10 × 10 cm2 prototype, as shown in the last two columns of Tab. 3.2. The final ion backflow results from the ratio of the total numbers of ions escaping into the drift volume, divided by the number of electrons reaching the anode, in this case IB = 9/1830 ' 0.5%. 3.1.2

Alternative quadruple GEM configurations

In the following we summarize the results of studies with alternative configurations of quadruple GEM systems with different hole pitch. Electric field scans of ion backflow and energy resolution as a function of ET2 and ET3 are shown in Figs. 3.7-3.9. The settings for ∆UGEM1 , ∆UGEM2 and ET1 were optimized in foregoing scans for each configuration, respectively. For each setting of ET2 and ET3 the gain was adjusted to 2000 by fine tuning of ∆UGEM3 and ∆UGEM4 , while keeping ∆UGEM3 /∆UGEM4 = 0.8.


15

εcoll

ne,in

M

ne−ion

εextr

ne,out

G

nion,back

fraction of total IBF (sim.)

fraction of total IBF (meas.)

1

1

14

13

0.65

9.1

9.1

3.6 (28%)

40%

31%

GEM2 (LP)

0.2

1.8

8

12.7

0.55

8

0.88

3.3 (26%)

37%

34%

GEM3 (LP)

0.25

2

53

104

0.12

12.7

1.6

1.3 (1.3%)

14%

11%

1

12.7

240

3053

0.6

1830

144

0.84 (0.03%)

9%

24%

1830

1830

9 (0.28%)

GEM1 (S)

GEM4 (S) Total

3183

Table 3.2: Electron and ion transport properties in a S-LP-LP-S configuration operated with TDR settings (see Tab. 4.1). For explanation see text. Also shown are the absolute and relative number of backflowing ions per layer, and their relative contribution to the total ion backflow. The last columns shows results from measurement on a 10×10 cm2 prototype.

S-LP-LP-SP As a possible improvement of the baseline S-LP-LP-S system, GEM4 was replaced by a foil with a smaller hole pitch of 90 µm (SP). The larger hole density in GEM4 may reduce the extraction of ions into transfer gap 3. Moreover, the distribution of the signal electrons into more holes in the final amplification stage could improve the operational stability. The result of an ET2 - ET3 scan of ion backflow and energy resolution is shown in Fig. 3.7. As in the baseline S-LP-LP-S configuration, the best ion backflow is found for high ET2 and low ET3 . The best values for the ion backflow of about 0.9% at an energy resolution of 12% are within the design specifications, but offer no improvement with respect to the baseline S-LP-LP-S configuration. σ(55Fe) (%)

ET3(V/cm)

ET3(V/cm)

IB (%)



Figure 3.7: Ion backflow (left) and energy resolution (right) in a quadruple S-LP-LP-SP GEM in Ne-CO2 -N2 (90-10-5) as a function of ET2 and ET3 . The settings for ET1 , Eind , ∆UGEM1 and ∆UGEM2 are indicated in the figures. The voltages ∆UGEM3 and ∆UGEM4 are adjusted to maintain a gain of 2000 in each setting.

S-S-LP-S In the S-S-LP-S system (Fig. 3.8) the best ion backflow at an energy resolution around 12% is achieved at low ET2 and high ET3 . This is consistent with previous findings presented in [1]. Again, the best achievable values for the ion backflow of about 0.8% at an energy resolution of 12% are within the design specifications but not better than the baseline S-LP-LP-S configuration. S-S-LP-SP The results for the S-S-LP-SP system are shown in Fig. 3.9. As for S-S-LP-S, the best ion backflow values are for low ET2 and high ET3 . However, replacement of GEM4 by an SP foil leads to a significant improvement with respect to S-S-LP-S. In the optimal settings, ion backflow values of about 0.5% at an energy resolution of 12% are achieved, a notable improvement even over the S-LP-LP-S baseline.

16


σ(55Fe) (%)

ET3(V/cm)

ET3(V/cm)

IB (%)



ET2(V/cm)

ET2(V/cm)

Figure 3.8: Ion backflow (left) and energy resolution (right) in a quadruple S-S-LP-S GEM in Ne-CO2 -N2 (90-10-5) as a function of ET2 and ET3 . The settings for ET1 , Eind , ∆UGEM1 and ∆UGEM2 are indicated in the figures. The voltages ∆UGEM3 and ∆UGEM4 are adjusted to maintain a gain of 2000 in each setting. σ(55Fe) (%)

ET3(V/cm)

ET3(V/cm)

IB (%)



Figure 3.9: Ion backflow (left) and energy resolution (right) in a quadruple S-S-LP-SP GEM in Ne-CO2 -N2 (90-10-5) as a function of ET2 and ET3 . The settings for ET1 , Eind , ∆UGEM1 and ∆UGEM2 are indicated in the figures. The voltages ∆UGEM3 and ∆UGEM4 are adjusted to maintain a gain of 2000 in each setting.

Summary of quadruple GEM studies Within an extensive R&D effort using small prototypes, four different configurations of quadruple GEMs with different hole pitch could be identified that fulfils the design specifications of ion backflow ≤ 1% and energy resolution ≤ 12%. For these systems, the characteristic correlation between ion backflow and energy resolution in optimized settings is summarized in Fig. 3.10. For the baseline system S-LPLP-S, the previously reported performance [1] has been confirmed. A notable improvement was found when a foil with smaller pitch was used for GEM4, i.e. S-S-LP-SP, however, a characterization in terms of operational stability of this configuration could not be done yet. It should be noted that ion backflow alone is not the driving criterion for the choice of technology, but rather a careful weighting of arguments including performance, stability, robustness, and risk of failure must be made. Further quadruple GEM configurations that were studied are S-S-S-S, S-S-SP-S, LP-S-LP-S, SP-S-SP-S, and SP-S-LP-S. In these configurations, no operational point could be identified that fulfils the design requirements. 3.1.3

Prototype studies with a double GEM stack and a Micromegas

As an alternative to a quadruple GEM system, a readout scheme employing a double GEM stack and a Micromegas amplification stage (2GEM+MM) was proposed. The ion backflow and energy resolution of

σ (%)


17

20 Ne-CO2-N2 (90-10-5) S-LP-LP-S S-LP-LP-SP S-S-LP-S S-S-LP-SP

18 16 14 12 10 8 6 0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8 2 IBF (%)

Figure 3.10: Correlation between energy resolution and ion backflow in various quadruple GEM systems in Ne-CO2 -N2 (9010-5), occuring by variation of ∆UGEM1 and ∆UGEM2 . All transfer fields are optimized in foregoing scans, respectively. The gain was adjusted to 2000 by fine tuning of ∆UGEM3 and ∆UGEM4 , while keeping ∆UGEM3 /∆UGEM4 = 0.8.

this system was studied by the Tokyo-CERN setup sketched in Fig. 3.11. A similar setup was used at Yale University for the measurements presented in the TDR [1]. Two standard GEMs (hole pitch 140 µm) are mounted on top of a Micromegas with a 400 LPI mesh and an amplification gap of 128 µm. The transfer gaps are 2 mm between GEM1 and GEM2, and 4 mm between GEM2 and Micromegas. Typical transfer fields are ET1 = 3 kV/cm and ET2 = 0.075 kV/cm. The setup has a drift gap of 8 mm, and the drift field is kept at Edrift = 400 V/cm. The Micromegas was produced at CERN in bulk technology. Cathode mesh Drift gap = 8mm Edrift=400V/cm

GEM1 Transfer gap 1 = 2mm

GEM2 Transfer gap 2 = 4mm

Mesh (400 LPI)

MM gap = 128µm

Anode pads Figure 3.11: Sketch of the Tokyo-CERN 2GEM+MM prototype setup.

The results of various scans of ET1 , ET2 , ∆UGEM1 , ∆UGEM2 and the Micromegas voltage VMM are shown in Fig. 3.12. At an energy resolution of about 12%, ion backflow values of about 0.4% are achieved in Ne-CO2 -N2 (90-10-5), which is only slightly better than for quadruple GEMs. In Ne-CO2 (90-10), the best values are about 0.7%. This can be understood in terms of the lower VMM required for a gain of 2000 in this gas mixture.


Energy resolution (%)

18

20

GEM1-MM scan (V GEM1-MM scan (V GEM1-MM scan (V

18

GEM2 GEM2 GEM2

=210V) (Ne/CO ) 2

=230V) (Ne/CO ) 2

=250V) (Ne/CO ) 2

GEM1 scan (at Yale) (Ne/CO /N2) 2

Et1 scan (at Yale) (Ne/CO /N2) 2

GEM1 scan (Ne/CO /N2)

16

2

Et1 scan (Ne/CO /N2) 2

GEM2-MM scan (V GEM2-MM scan (V

14

GEM2-MM scan (V GEM2-MM scan (V

=220V) (Ne/CO /N2)

GEM1

2

GEM1

2

GEM1

2

=230V) (Ne/CO /N2) =240V) (Ne/CO /N2)

GEM1-GEM2 scan (V

12

GEM1-GEM2 scan (V GEM1-GEM2 scan (V GEM1-GEM2 scan (V

10

=250V) (Ne/CO /N2)

GEM1

2

MM

=420V) (Ne/CO /N2)

MM

=440V) (Ne/CO /N2)

MM

=460V) (Ne/CO /N2)

MM

=465V) (Ne/CO /N2)

2 2 2 2

8 6 0

0.5

1

1.5

2

2.5

3 IBF (%)

Figure 3.12: Correlation between energy resolution and ion backflow in a 2GEM+MM system operated at a gain of 2000 in Ne-CO2 -N2 (90-10-5) and Ne-CO2 (90-10). Results for various scans of the GEM and Micromegas voltages and of the transfer fields are shown. The Tokyo-CERN measurements are shown in red and black, whereas the blue points indicate the results from the Yale group.

3.2

Discharge probability studies with small detectors

Conventional triple GEM systems operated with so-called standard settings, i.e. voltage settings in which the discharge probability is minimized, were operated successfully in high-rate experiments [5]. On the other hand, quadruple GEMs operated with settings that are optimized for low ion backflow may have different discharge behaviour. Therefore, a thorough investigation of the discharge properties of quadruple GEMs and a careful comparison to existing measurements with triple GEMs is required. The most comprehensive discharge studies with single and multiple GEM detectors were reported in [6] but concern mainly Ar-based gas mixtures and no quadruple GEMs. Therefore, discharge probability studies in triple and quadruple GEM structures in Ne- and Ar-based gas mixtures were performed in order to characterize the operational conditions of the upgraded ALICE TPC. 3.2.1

Experimental setups

Several small prototypes have been prepared by different groups to study the stability of GEM detectors against electrical discharges. Figure 3.14 shows a schematic drawing of the geometric configuration that is similar in all setups. Specific differences between the setups are summarised in the following. Munich setups Two setups have been prepared in Munich to measure the discharge probability. Both feature similar functionality and flexibility. The detector housing of each setup comprises a 10×10 cm2 GEM holder, a drift cathode and a readout anode. Both the cathode and the anode are made of a 1.5 mm thick PCB covered on one side with copper (∼10 × 10 cm2 ). In the middle of the cathode plate, an 8 mm diameter


19

Figure 3.13: Schematic picture of a typical setup used for the discharge probability studies. Different letters indicate possible chains to read out the anode signal. See text for more details.

hole is placed which allows for irradiation with a radioactive source perpendicular to the GEM plane. Both triple and quadruple GEM stacks can be installed for measurements. The setups do not employ field cages which gives the flexibility to adjust the drift gap length (distance between the cathode and the GEM stack) continuously between 3 and 71 mm in case of a 4-GEM setup. The gas gain of the setup for a given HV setting is measured by the usual method of recording the current at the pad plane and the rate of absorbed X-rays of known energy (a 55 Fe source is used). However, while running the detector with a collimated alpha source installed inside the gas vessel, the gain is measured via the ratio of the primary ionisation current, created by the alpha particles (and measured at the cathode with a drift field applied and a top electrode of GEM1 grounded) to the current measured at the pad plane, after the full amplification. Both methods give comparable results. The occurrence of a spark in a GEM foil (note that discharges between GEM foils or propagated to the pad plane were not measured) is detected according to option c) of the readout scheme shown in Fig. 3.13. The raw signal induced on the pad plane is attenuated (1-31 dB) and then directed into the discriminator unit. The threshold on the discriminator is set to 1 V to filter out signals induced by alpha particles ( 1 V) and to trigger on discharge signals with a much higher amplitude (see Fig. 3.14 for example). Since the raw signals are often modified by oscillations, a gate is created when the discriminator threshold is exceeded which is then counted by a scaler. In this way, multi-counting of the same signal can be avoided. It is also possible to split the signal just after the attenuator, using either a passive splitter or a Fan-In/Out unit, and monitor it on the oscilloscope.

Figure 3.14: A typical signal associated with a spark in a small prototype, recorded by the oscilloscope with a 50 Ω input impedance. The saturation effect comes from the electronic module (Fan-In/Out) used in the readout chain.

20


CERN setups The two setups used at CERN for the discharge probability measurements were previously used for the ion backflow and gain stability studies and were described already in the TPC Upgrade TDR [1]. Both have drift gaps of fixed length, i.e. 3 mm and 80 mm. The latter employs also a field cage. The setup with the shorter drift gap was used to measure the discharge probability with an internal gaseous 220 Rn source and a high-rate beta emitter 90 Sr (see Sec. 3.2.2). The setup with the longer drift gap was used in measurements where a collimated 241 Am source was employed. The housing of this detector employs a thin (1.5 µm) Mylar window which allows to irradiate with an alpha source from outside the detector. In both cases, discharges are detected by an amperemeter which measures the current continuously at the detector pad plane (see readout option b) in Fig. 3.13). In case of a discharge, a high current peak is expected at the readout anode. It may be induced by the charge liberated by a spark or (more likely) by the recharging current on the GEM foil which occurs after the discharge. 3.2.2

Radiation sources

Since discharges occur most likely when high local ionization densities are created in the detector, alpha particles are used for the present discharge studies. In addition, a high-rate beta emitter was used for comparison. In all cases, the discharge probability is defined by the ratio of the discharge rate to the source rate: N P= , (3.1) tR where N/t is the discharge rate measured by the number of sparks N recorded within the measurement time t, and R is the rate of a source. The statistical error of this quantity is calculated from: s √ 2 2 N N ∆R ∆P = + , (3.2) tR N R where ∆R is the statistical uncertainty of the source rate. The rates of all sources were measured by integrating their energy spectra (after background subtraction) obtained within a given time. Energy spectra were measured using option a) in the readout scheme in Fig. 3.13. Internal gaseous 220 Rn source An internal gaseous alpha emitter (220 Rn) was added to the gas flow which provides a uniform irradiation of the detector volume. The energy of the emitted alpha particle is Eα = 6.4 MeV. The low rate of about 0.5 Hz restricts the upper limit for the discharge probability that can be achieved within one day to O(10−5 ). This type of source was used both in Munich and at CERN. Collimated mixed 239 Pu+241 Am+244 Cm source The mixed nuclide source consists of 239 Pu, 241 Am and 244 Cm nuclides and was specified with an activity of 1 kBq for each component in January 2008. The energies of the most intense alpha particles emitted by the different constituents are: Eα,Pu = 5.2 MeV, Eα,Am = 5.5 MeV and Eα,Pu = 5.8 MeV. In Fig. 3.15, the mixed nuclide source and its energy spectrum are shown. The left figure shows the active material of the source deposited on a coin-like area. The rate of the source, measured in the Munich setup is R ≈ 600 Hz. Collimated 241 Am source An external, collimated 241 Am source was used in one of the CERN setups, sending 5.5 MeV alphas through the thin Mylar window, perpendicular to the GEM stack. The rate of the source is R ≈ 11 kHz.


21

Figure 3.15: Left: Picture of the mixed nuclide source with typical dimensions. Right: Energies and intensities of the most abundant alpha particles emitted by the different nuclides. Both pictures were taken from [7].

Collimated 90 Sr source In addition to the alpha sources, a beta emitter was used in the stability studies. The 90 Sr source emits electrons with energies up to about 2.3 MeV at a rate of R ≈ 60 kHz and was mounted outside the detector.

3.2.3

Study of single GEMs

A systematic study of the discharge properties of a single GEM detector was performed. By measuring the stability of the simplest configuration, including only one GEM foil, the intrinsic stability of the GEM foil against electrical discharges is decoupled from the influence of other effects, like transfer fields or charge sharing and spreading between the foils, which may occur in multi-GEM structures. Different gas mixtures were tested, the gaseous 220 Rn source was used for irradiation. Figure 3.16 shows the discharge probability measured in a single GEM detector in Ne-CO2 (90-10) and Ar-CO2 (90-10). In both cases the drift gap was set to 33 mm, drift field Edrift = 400 V/cm and induction field Eind = 3 kV/cm. Since in both mixtures the amount of quencher is the same, the only difference is the mass and atomic number of the noble gas component. The discharge probability depends strongly on the gas mixture, a higher discharge probability for a given gain has been measured in Ar-CO2 (90-10). A plausible explanation is that the range of a 6.4 MeV alpha particle in Ar-based mixtures is almost 40% shorter than in Ne-based mixtures, and the fact that the Bragg peak is narrower in argon mixtures (see Fig. 3.17d). This results in higher local charge densities. As a consequence, it is more likely to exceed the Raether limit in Ar-based mixtures, leading to higher discharge probabilities. In multi-GEM systems, the influence of transfer fields and a charge sharing between GEM foils may alter the dependence of the discharge probability on the shape of the Bragg curves, but the total charge density approaching a single GEM hole should remain an important factor to trigger sparks.

3.2.4

Study of triple GEMs

The discharge probability in triple GEM structures was measured as a function of the gas gain and the track inclination. To study these effects systematically, the detector was operated at high gains to collect enough statistics of discharges in a reasonably short time.


discharge probability

22

1

Ar-CO2 (90-10) Ne-CO2 (90-10)

220

Rn source

ED =400 V/cm; E

10-1

IND

=3000 V/cm

10-2 10-3 10-4 10-5 10-6

10-7

200

300

400 500

1000

2000

3000 4000

gain Figure 3.16: Discharge probability measured in a single GEM setup with a gaseous CO2 (90-10).

220 Rn

source in Ar-CO2 (90-10) and Ne-

Figure 3.17: GEANT4 simulation of the ranges in different gas mixtures of alpha particles from the sources used in the present measurements.

Comparison to published data As a first step, the results obtained for the triple GEM setup are compared to data available in the literature to validate our experimental methods. To this end, measurements of the discharge probability as a function of the gain in Ar-CO2 (70-30) are carried out using the gaseous 220 Rn source. Figure 3.18


23


shows the results of this measurement, compared to the data obtained for a similar setup using a 241 Am source from [6]. In both cases, the GEM detectors were operated at analogous HV settings, the so-called standard 3-GEM settings that are optimized for operational stability. This is achieved by a successive reduction of the gas gain from the first to the last GEM layer (see also Sec. 5.2.4 in the TPC upgrade TDR [1]). Very good agreement is observed between our results and those from the literature.

10-1

Ar-CO2 (70-30) Bachmann et al. - 1GEM, 241Am Bachmann et al. - 2GEM, 241Am Bachmann et al. - 3GEM, 241Am

10-2

TUM - 3GEM, 220Rn

10-3

10-4

10-5

10-6

3

10

104

5

10

6

10 gain

Figure 3.18: Discharge probability as a function of the effective gain for single, double and triple GEM detectors (from [6]) compared to the results obtained with the Munich setup (red points).

Discharge probability measurements in Ne-based gas mixtures Fig. 3.19 shows the results of a gain scan for two different Ne-based gas mixtures. The measurements are performed at high gas gains to acquire a sufficient number of sparks with the low-rate 220 Rn source (see Sec. 3.2.2). The detector is operated with the ”standard” HV settings that are commonly used for triple GEM structures, scaled in order to vary the total gain. Clearly, the addition of N2 to the gas mixture has a noticeable effect on the discharge behaviour. The discharge probability observed in Ne-CO2 -N2 (90-10-5) is one order of magnitude lower than in Ne-CO2 (90-10). This finding supports the choice of the baseline gas mixture for the upgraded TPC (see Sec. 3 in the TPC upgrade TDR [1]). Track inclination and track length studies Sending alpha particles parallel to the GEM stack (in the middle of a 25 mm drift gap) resulted in a 3 orders of magnitude lower discharge probability at a gas gain of 105 than previously measured with the 220 Rn source. This suggests that the primary charge density arriving at the GEM holes after drift may be too low (due to the track inclination and the diffusion) to affect the stability of the detector. Thus, the measurements were continued with the alpha source placed on the drift cathode, sending the alpha particles perpendicular to the GEM stack, through the hole in the cathode PCB. Figure 3.20 shows the discharge probability as a function of the distance dsource between the source and the GEM stack. The broad plateau clearly indicates that the discharge probability is higher when the alphas penetrate the GEM foils, or even get stopped there. In this case, the highest local primary charge densities in a single GEM hole can be reached. When the distance is increased such that the alphas



24

10-1 TUM - 3GEM, Ne-CO2-N2 (90-10-5),

TUM - 3GEM, Ne-CO2 (90-10),

220

220

Rn

Rn

10-2

10-3

10-4

10-5 104

5

10

gain

Figure 3.19: Discharge probability as a function of gain in a triple GEM setup, measured with a (90-10) and Ne-CO2 -N2 (90-10-5).

220 Rn

source in Ne-CO2


do no longer reach the GEM structure (see also Fig. 3.17), the discharge probability drops by orders of magnitude. Note that the primary ionization does still reach the GEMs from the drift field in this case, but this has clearly much less impact on the detector stability. This is due to the fact that the density of the charge that arrives at the GEM holes is reduced by diffusion and will therefore have less probability to create a spark. In all following measurements the source distance is adjusted to the position of the maximal discharge rate in Fig. 3.20, which can then be considered as a worst-case scenario for the detector stability.

10-3 Ar-CO2 (90-10), Mixed Src, G=30000 Ne-CO2-N2 (90-10-5), Mixed Src, G=45000

10-4

10-5

10-6

10-7

10-8

10-9

10

20

30

40

50

60

70 80 dsource (mm)

Figure 3.20: Discharge probability as a function of the distance between the alpha source and the GEM stack. In the measurement, the mixed nuclide alpha source was placed on top of the cathode, sending the alpha particles perpendicular to the GEM foils. Upper limits for the discharge probability for a given distance are indicated with arrows.

The track inclination and the track length scan indicates that the development of a spark in a GEM foil is


25

influenced mainly by the local charge in the single hole rather than total charge integrated over the whole foil area. This observation has an important implication when the stability of the TPC readout chambers during operation in ALICE is considered. We conclude that it is the number of particles that cross the GEM stacks which determine the discharge rate of the detectors, and that the primary charge that reaches the readout chambers from the drift volume has significantly less impact on the detector stability. This consideration will be used when the expected discharge rate during operation in ALICE is estimated in Chap. 4.2. Study of different HV settings All measurements presented so far were performed using the so-called ”standard” HV settings, put forward by the COMPASS experiment [8], which aim to maximize the stability of the triple GEM detectors. In these HV settings, the highest amplification occurs in the first (top) foil in a stack, where the total amount of charge is lowest. On the other hand, in HV settings that are optimised for low ion backflow, the sequence of amplification is usually reversed, which means that the highest amplification takes place in the last (bottom) foil. This amplification scheme, together with the pre-amplification of charge in the first GEM stages, may lead to charge densities inside the holes of the last foil that are higher than those in the first foil when the system is operated with “standard” settings1 . As a consequence, a noticable decrease of the detector stability against electrical discharges might occur. The discharge probabilities of a triple GEM detector were measured as a function of the gas gain, applying different HV settings to the stack. Figure 3.21 shows the results obtained in Ne-CO2 -N2 (90-10-5) using the 220 Rn source. Due to the low rate of the source, the measurements were performed at gains that are higher than the nominal gas gain of 2000. In addition to the “standard” settings, the detector was also operated in so-called “reversed” HV settings, where the amplification scheme is reversed with respect to the ”standard” configuration, but keeping the transfer fields as in the “standard” settings. Finally, the settings optimised for low ion backflow are applied, where, in addition to the ”reversed” amplification hierarchy, the value of the second transfer field ET2 is substantially reduced. The results shown in Fig. 3.21 clearly confirm the expectations. The discharge probability for the settings optimized in terms of low ion backflow is about 3 orders of magnitude larger than for the “standard” configuration. The results indicate a large influence both of the reversed amplification order and of the lowering of ET2 on the discharge probability. In order to get an estimate of the discharge probability at the nominal gas gain, the data points are fitted by power law functions and extrapolated to a gain of 2000, which yields ∼10−10 and ∼2 × 10−7 for the “standard” and the “IBF” settings, respectively. In conclusion, the results on the discharge probability of triple GEM detectors under alpha irradiation presented in this section are consistent with previous measurements. At a gas gain of 2000, discharge probabilities of about 10−10 are estimated in “standard” HV settings. Since triple GEM detectors in “standard” HV settings have been operated successfully in high-rate experiments before, a discharge probability of 10−10 under alpha irradiation can be considered as safe. Optimization of the voltage settings with respect to minimal ion backflow, however, leads to a significant increase of the discharge probability. In the next section, we study the discharge probability of quadruple GEM systems, where the additional layer is expected to lead to more stable operation than in a triple GEM system at comparable gain. 3.2.5

Study of quadruple GEMs

Stability tests of quadruple GEM detectors have been performed employing foils with different hole pitch, following the ion backflow and energy resolution optimization studies performed within the scope of the ALICE TPC upgrade R&D (see Sec. 3.1 in this document and Sec. 5.1.3 in the TPC Upgrade 1 it should be noted that also a charge spread between foils has a non-negligible influence on the charge density obtained in the foils, and thus on the stability of the system



26

1

3-GEM, Ne-CO -N2 (90-10-5), 2

220

Rn

Standard HV

10-1

Reversed HV IBF HV

10-2

10-3

10-4

10-5

10-6 3 10

104

5

10

6

10 gain

Figure 3.21: Discharge probability of a triple GEM prototype measured for different HV settings (see text). Dashed lines represent power law function fits. For the fit of the “standard” settings the data point at the highest gain was not used.

TDR [1]). All available radiation sources (see Sec. 3.2.2) have been used. The gas mixture used in all measurements is Ne-CO2 -N2 (90-10-5) which is the baseline gas mixture for the ALICE TPC upgrade. In measurements where collimated high-rate alpha sources were used, the alpha particles impinge on the GEM foils at normal incidence. Table 3.3 presents discharge probabilities measured with different sources for various quadruple GEM stack configurations that are optimised for low ion backflow (IB), including the baseline settings for the S-LP-LP-S stack (IB =0.63%, G=2000, see Sec. 3.1.1). For comparison, the result for a triple GEM operated with “standard” settings, as extrapolated from measurements at higher gains (see Sec. 3.2.4) is also shown. Most of the numbers quoted for quadruple GEMs are upper limits for the discharge probability (indicated by ” 3 times smaller than the HIROC prototype) follows the schematics shown in Fig. 4.3. In the Yale prototypes, the transfer gap between GEM2 and the Micromegas is 10 mm to enhance diffusion of the charge cloud and thus lower the charge density at the mesh with the aim to improve the stability against electrical discharges.

Figure 4.3: Schematic picture of a 2GEM+MM Yale prototype

4.2

Discharge studies with large-size prototypes

The discharge behaviour of the four prototypes described in Sec. 4.1 was evaluated in a test beam at the CERN-SPS. All detectors were operated with a Ne-CO2 -N2 (90-10-5) gas mixture. The discharge probability was measured using showers of hadrons produced by a high-intensity secondary pion beam with a momentum of 150 GeV/c impinging on a 30-40 cm thick iron absorber. The average beam intensity was ∼6×106 particles per spill (∼5 s), resulting in an average in-spill rate of ∼1.2 MHz. During the two-week RD51 test beam campaign, a total of 36 h of dedicated data taking was allocated for the ALICE TPC tests. The beam time was divided into two parts. In the first part, the 4-GEM IROC together with the Yale prototypes were tested. In the second part, the Yale prototypes were replaced by the HIROC prototype. Schematic drawings of both setups and a photo of the experimental area are shown in Fig. 4.4 and Fig. 4.5, respectively. The particle flux into the 4-GEM IROC chamber was calibrated by recording the current at the anode pad plane (without iron absorber) as a function of the counts measured in the beam scintillators. The IROC anode current was recorded continuously. Hadron showers were created by the pion beam hitting the 40 cm iron absorber (30 cm at the beginning of the beam time). On average, 20 shower particles were observed per incoming beam particle. They are emitted preferentially in the direction perpendicular to the detector plane. The integral of the chamber current over the whole beam period gives the total SPS = (4.7 ± 0.2) × 1011 . This number can be compared to the total number of accumulated particles Ntot


31

Figure 4.4: Schematic drawing of the setup at the SPS. In the middle of the beam time period, the two Yale prototypes were replaced by the HIROC prototype.

Figure 4.5: Photograph of the experimental area at the SPS showing the 4-GEM IROC and the Yale prototypes.

number of particles expected in the TPC during a typical yearly Pb-Pb run (106 s) at a collision rate of 50 kHz. Assuming hdNch /dη i=500 and a coverage of about one unit of pseudo-rapidity for each of the two readout planes of the TPC, we estimate 500 × 2 × 50000 × 106 = 5 × 1013 charged particles hitting the active surface of the TPC readout planes in a yearly heavy-ion run. Including a factor of two to account for background, this implies that each of the 144 GEM stacks accumulates about 7 × 1011 particles, which is comparable to the statistics accumulated at the SPS. As discussed in Sec. 3.2.4 we assume that mainly particles that cross the GEM foils are relevant for the discharge behaviour. The results obtained for the different prototypes are presented in the following subsections. 4.2.1

4-GEM IROC

The 4-GEM IROC was placed directly behind the iron absorber. The detector signals were read out according to the schematic picture shown in Fig. 4.6. All readout pads were connected together and the signal was divided into two branches: the first one is connected to a pico-amperemeter via a 10 kΩ resistor and the second one to an oscilloscope via a 100 kΩ resistor. The input impedance of the scope was set to Z = 1 MΩ. The pico-amperemeter was used to monitor continuously the anode current which is proportional to the particle flux into the detector (see Sec. 4.2). The oscilloscope was set to record high signals associated

32


Figure 4.6: Readout scheme for the 4-GEM IROC at the SPS. The signal from the entire pad plane is split between the ampere meter and the oscilloscope operated with an input impedance of1 MΩ.

with the occurrence of a discharge in the detector (see Fig. 4.7). The discharge signal recorded in an IROC is longer and larger than the typical signal associated with a discharge in a small detector (see Fig. 3.14). This can be explained by the different sizes of the detectors (i.e. ∼10 times higher capacitance of the IROC) and their readout configuration.

Figure 4.7: A typical signal associated with a discharge in an IROC recorded with the readout scheme presented in Fig. 4.6.

During the entire SPS beam time period, the 4-GEM IROC was operated with the new baseline HV settings (IB =0.63%, σ (55 Fe)=11.3%, see Tab. 4.1) at a gain of 2000. The HV was distributed to the subsequent GEM electrodes via a resistor chain. Due to the malfunctioning of a power supply, the drift field Edrift needed to be reduced to 275 V/cm, which is not expected to impact the measurement. In total, three discharges were detected in the 4-GEM IROC detector. This translates into a discharge probability of (6.4 ± 3.7) × 10−12 per incoming hadron. This result is of the same order of magnitude as the one obtained by the LHCb Collaboration where the discharge probability of triple GEM detectors operated with an isobutane-CF4 -based gas mixtures was measured under similar conditions [9]. The discharge rate per hadron measured at the SPS can be translated into the expected number of discharges in the upgraded TPC per unit time. We estimate about 650 discharges for the whole TPC, or 5 for each of the 144 GEM stacks per typical yearly heavy-ion run at 50 kHz. Such small numbers are not expected to create any damage to the GEM detectors and ensure efficient and safe operation of the TPC in RUN 3 and beyond. Further stability tests with the 4-GEM IROC are planned at the end of 2015 during the LHC Pb–Pb run, when the 4-GEM IROC will be placed inside the ALICE Miniframe, close to the interaction point in forward direction. This will allow to evaluate the operational stability under LHC conditions. 4.2.2

Yale prototypes

During the first half of the test beam, the two Yale prototypes were installed behind the 4-GEM IROC (see left picture in Fig. 4.4). Both were placed with their readout planes perpendicular to the beam direction. The detectors were operated at three different HV settings, summarised in Tab. 4.2. The first two settings result in a gas gain of ∼2000 and are typical settings in hybrid stacks, optimised for low ion


33

backflow values. It should be noted that the settings with higher VMM correspond to a lower ion backflow. The last setting was used to measure the stability of the Micromegas only. Here, the voltages across the GEMs and the transfer fields above the mesh were set to zero to prevent pre-amplification and charge drifting to the mesh. The signals from the Yale prototypes were read out via a preamplifier and then directed to a discriminator and a scope. High signals, associated with discharges, were recorded and counted with a scaler. To extract the discharge probability, the number of detected discharges was normalised to the total number of incoming hadrons as estimated from the 4-GEM IROC current measurement. It is assumed here that the number of shower particles in the 4-GEM IROC and the Yale prototypes is approximately the same. For each setting, both Yale prototypes recorded a similar number of discharges. The discharge probabilities are given in Tab. 4.2. #

∆UGEM1 (V)

∆UGEM2 (V)

VMM (V)

gain

Discharge probability

1 2 3

250 260 0

210 220 0

440 420 420

2050 2000 450

(2.0 ± 0.6) × 10−9 (3.5 ± 1.0) × 10−10 (1.7 ± 0.5) × 10−10

Table 4.2: Discharge probabilities measured with the Yale prototypes at the SPS for 3 different HV settings. For the first two settings Edrift = 400 V/cm, ET1 = 3875 V/cm and ET2 = 90 V/cm. For the third setting Edrift = 21 V/cm, ET1 = 3 V/cm and ET2 = 2 V/cm

The discharge probabilities observed in the Yale 2GEM+MM prototypes are 2-3 orders of magnitude larger than those of the 4-GEM IROC. The discharge probability is a strong function of the mesh voltage VMM , at higher voltages (which are favoured in terms of low ion backflow) a significant increase of the discharge probability is observed. A discharge probability of ∼ 10−9 would lead to an estimated number of about 100000 discharges in the TPC during a yearly heavy-ion run at 50 kHz. It is interesting to note that the discharge probability does not change much when the GEM voltages and transfer fields are switched off. This indicates that the discharge behaviour is mainly governed by particles crossing the mesh, and only to a lesser extend by the pre-amplification and charge that drifts to the mesh. 4.2.3

HIROC

During the second part of the beam time period, the HIROC prototype was placed behind the 4-GEM IROC (see the right picture in Fig. 4.4). The discharge probability of the HIROC was measured for 3 different HV settings, as listed in Tab. 4.3, resulting in a gas gain of 1600-2000. During the measurements, all pads of the HIROC pad plane were connected together and read out by the oscilloscope (1 MΩ input impedance) via the diode protection circuit. Table 4.3 summarises the discharge probabilities obtained for all 3 settings. The number of detected discharges in the HIROC was normalised to the total number of hadrons estimated from the 4-GEM IROC currents. #

∆UGEM1 (V)

∆UGEM2 (V)

∆UMM (V)

gain

Discharge probability

1 2 3

232 250 260

210 210 220

460 440 420

2000 1800 1600

(3.0 ± 0.3) × 10−9 (3.1 ± 0.7) × 10−10 (1.5 ± 1.1) × 10−11

Table 4.3: Discharge probabilities measured with the HIROC prototype at the SPS for 3 different HV settings. In all cases Edrift = 390 V/cm, ET1 = 3875 V/cm and ET2 = 150 V/cm. The gas gain for the HV settings was measured with a small-size 2GEM+MM prototype.

34


The discharge probability varies between 10−11 and 10−9 , depending on the HV setting. It is by 1-2 orders of magnitude larger than that of the 4-GEM IROC. In Sec. 4.2.2 it was argued that discharges are dominantly induced by particles crossing the mesh, and the discharge probability is mainly determined by the mesh voltage VMM . Thus one can compare the discharge probabilities of both 2GEM+MM prototypes at the same VMM . This comparison indicates that the discharge probability of the HIROC is approximately one order of magnitude smaller at a given VMM than that of the Yale prototypes. It must be noted however that there are differences in the chamber geometry which complicate a direct comparison of the detectors.

4.3

Particle identification with large-size prototypes at the CERN PS

Two full-size prototypes of TPC Inner Readout Chambers (IROCs) were built and tested at the CERN Proton Synchrotron (PS) in the fall of 2014. One of them is equipped with the baseline S-LP-LP-S quadruple GEM configuration (4-GEM IROC), while the other one (HIROC) employs a hybrid readout system with two GEMs and a Micromegas (2GEM+MM). Additionally, two smaller 2GEM+MM prototypes (Yale prototypes) were tested in the beam. More details on the different prototype setups are given in Sec. 4.1. The 4-GEM IROC prototype was operated during the whole PS beam period, while the HIROC was replaced by the Yale prototypes in the middle of the data taking. The experimental setup as well as the analysis techniques were identical to the 2012 test beam campaign. They are only briefly summarized below, more details can be found in Sec. 5.2 of the TPC Upgrade TDR [1]. 4.3.1

Experimental setup

The 4-GEM IROC and the HIROC (later the Yale prototypes) were operated simultaneously and placed behind each other in the T10 beam line of the PS. A common gas supply from a pre-mixed bottle of NeCO2 -N2 (90-10-5) is used. A picture of the setup showing the 4-GEM IROC and the HIROC is shown in Fig. 4.8.

Figure 4.8: Picture of the PS setup with 4-GEM IROC and HIROC.

The IROC prototypes were read out by 10 front-end cards each, corresponding to about 1200 readout channels. The readout electronics was borrowed from the LCTPC (Linear Collider TPC) collaboration. It is similar to the current TPC readout electronics (except for the PASA chip) and allowed to use the full ALICE readout chain. The readout electronics covers a 6 – 7 cm wide corridor over the full length of the detectors to allow for a systematic measurement of the dE/dx performance with beam particles. The readout system has an RMS noise of about 600 electrons. A zero suppression threshold of 2 ADC counts was used, corresponding to about 2000 electrons (120 ns peaking time and 12 mV/fC conversion gain). The sampling frequency was 20 MHz.


35

To discriminate electrons from pions, a Cherenkov counter was used. Moreover, the ambient temperature and pressure were monitored. This information was read out through a classic CAMAC system and associated to the events for later analysis.

15

10 10

2

5

charge (ADC)

pad

Event displays of the 4-GEM IROC and the HIROC are shown in Fig. 4.9. As expected in this setup with short drift length and therefore negligible diffusion, a typical track signal is visible mainly in a single pad per pad row. However, in the case of the HIROC prototype (lower panel in Fig. 4.9), a halo-like noise pattern occurs in the vicinity of the track, which extends over up to three neighboring tracks. The origin of this pattern has not yet been clearly identified, and more refined procedures will be needed for a proper treatment in the analysis. Similar noise patterns were also observed in the Yale prototypes. At the present stage of the analysis, we expect that the dE/dx resolution in the HIROC can be affected by this effect.

0 10 -5

-10 1 -15 0

10

20

30

40

50

60

10 10 5

2

charge (ADC)

pad

row 15

0 10 -5

-10 1 -15 0

10

20

30

40

50

60

row

Figure 4.9: Event display of the 4-GEM IROC (top) and the HIROC (bottom) at the PS. Both detectors are operated at a gain of about 2000.

4.3.2

Gain uniformity and equalization

To minimize the impact of gain variations over the readout plane on the dE/dx resolution, a gain map was extracted on a pad-by-pad basis. Gain variations may result from geometric imperfections of the GEM or Micromegas system as well as from channel-by-channel variations of the front-end electronics. The extraction of the gain map is done by calculating for each track the average energy loss hdE/dxi (see also Sec. 4.3.3) and monitoring on each pad the total cluster charge divided by hdE/dxi. With this procedure, the gain determination is independent of the gain and particle species. This results in a normalized-charge distribution per pad which follows a Landau distribution. The gain per pad is then determined by the truncated mean of the distribution, rejecting the largest 30% of charges. Figure 4.10 (left) shows the resulting gain map for the 4-GEM IROC prototype. Clearly visible are pad rows with lower gain at the HV sector boundaries, which also correspond to the spacer grid positions. Also visible is a low-gain region for pad rows >57. The reason is a short in one HV sector of one GEM foil, resulting in a lower ∆UGEM and therefore a lower amplification. In the right panel of Fig. 4.10 the gain distribution in the 4-GEM IROC is shown, excluding pad rows with relative gain < 0.6. The observed spread is ∼ 10.5%. In the current TPC with MWPC readout an average value of ∼ 9% is found, ranging from ∼ 6% to ∼ 14% for individual chambers. The values for the MWPCs were obtained by a calibration using radioactive Krypton gas, which may lead to smaller systematic variations.

36


Figure 4.11 (left) shows the gain map for the HIROC prototype. As for the 4-GEM IROC, the regions of the GEM-HV sector boundaries are visible as pad rows with lower gain. Regions with missing entries are caused by faulty or dead readout channels. It should be noted that the statistics available for this calibration is significantly lower than for that of the 4-GEM IROC, especially for pads at the edge of the active region. In the right panel of Fig. 4.11 the gain map distribution is shown, again excluding pad rows with low gain and single pads with a relative gain