Gas Properties and Dierent Detector Geometries for Multi-GEM Detectors in the CMS Detector

Transcription

1 Gas Properties and Dierent Detector Geometries for Multi-GEM Detectors in the CMS Detector August 15, 2013 Travis Volz American University, Washington, DC CMS Collaboration, CERN, Geneva, Switzerland Advisor: Archana Sharma 1

2 Contents Abstract 1. Introduction 2. Background 2.1 The CMS Detector 2.2 Muon Detectors 2.3 GEM Detectors 2.4 Desired Gas Characteristics 3. Approach 4. Results 3.1 Simulation Methods 3.2 Data Analysis 5. Summary and Conclusion References Abstract In our research, we attempted to determine ways to increase the time resolution in a multi-gem (gas electron multiplier) muon detector for the CMS (compact muon solenoid). This would allow for faster and more accurate muon tracking in the CMS. We studied two ways of achieving this. First, we studied gas properties such as drift velocities and Lorentz angles in various gas mixtures. We also studied dierent geometrical set-ups in an attempt to maximize the time resolution. By using simulation and modeling programs such as Gareld, Magboltz, ROOT, and ANSYS, we tried to nd the most ecient set up, but ran into time constraints. In the following report, I give my contribution to this project. 1 Introduction To study the properties of dierent gas mixtures, Gareld++ in tandem with Magboltz was used. Gareld is is a computer program used for the simulation of two and three dimensional gas drift chambers and Gareld++ is a version of Gareld that uses the C++ coding language. Magboltz solves the Boltzmann transport equations for electrons in gas mixtures under the inuence of electric and magnetic elds. We specied the gas mixture percentages, the electric and magnetic elds, and the angles between E and B that would be used during the simulations. By saving the log les from the SSH Client PUTTY, ltering out the useful data, and analyzing it, we were able to make useful graphs in Excel to display the information and to help determine which gas combinations could be useful. To study the geometries of the multi-gem detectors, we used ANSYS, an engineering modeling and simulation software which can accurately calculate the electric eld of each geometry in question. This research was done for the multi-gem group at CERN's CMS detector. They are trying to create more ecient muon detectors for the CMS, though there are already several dierent types implemented by the CMS, such as the Resistive Plate Chambers, Cathode Strip Chambers, and drift tubes. [1] The results could be useful in determining which gas mixtures and the geometries to be use in the multi- GEM detector. For example, a gas that has a high drift velocity allowing the muons to be detected quickly 2

3 is desired, but a stable gas that will not spark under the presence of the large electric eld is also needed. By nding a good mixture, the detector can be faster and safer. Dierent geometries can also be implemented to increase the speed and eciency of the detector. More about this can be found in section Background 2.1 The CMS Detector The Compact Muon Solenoid is one of the four main detectors located on the Large Hadron Collider (LHC) at CERN. It is a general purpose detector aimed at studying a large range of high energy particle physics. As bunches of protons collide inside the CMS after having been accelerated in several stages, they create a variety of particles with dierent properties. The goal of the detector is to use the particle paths and energies to reconstruct the events at the center of the collision and to discover new things about the basic constituents of matter. In order to do this, the CMS has multiple layers that can detect dierent properties of the particles. Pictured in gure 1 is the layout of the CMS. One of the most important components of the CMS is the magnet, the third letter in its name (solenoid). The 4 Tesla solenoid magnet allows the detector to attain good momentum resolution for charged particles, since charged particles curve as they pass through a magnetic eld. By tracking their paths, the momentum can be determined. [2] Figure 1: Layers of the CMS detector [2] Charged particle tracking is an essential part of piecing together the collision, because the missing momentum and energy can either classify particles, or can determine if particles such as neutrinos are missing, since total momentum must be conserved. The tracking system is made of various detectors, the closest of which are the pixel detectors which has pixels of µm 2. Since fewer particles interact the further out you go in the detector, the layers of the tracking system that are further from the collision have larger detectors components. In the second region, from 22 to 55 cm, the sensors have a length of 10 cm and a width of 80 µm, and in the nal region, from 55 to 110 centimeters, the detecting strips are 25 cm by 180 µm. [2] 3

4 In the next two layers are the calorimeters. The electromagnetic calorimeter measures the energies of electromagnetically interacting particles, which are mainly electrons and photons. The hadronic calorimeter measures the energies of hadrons, which are things made out of quarks, such as pions or protons and neutrons. The hadronic calorimeter also gathers information indirectly about non-interacting particles such as neutrinos. The outermost sensor is the muon system. 2.2 Muon Detectors There are two broad categories of muon detectors. Both types work on the same basic principle. A chamber of some shape is lled with an ionizable gas. As the highly energetic muon passes through the gas volume, it ionizes some of the gas particles. A large electrical potential dierence is applied to various parts of the chamber causing the electrons which are created with ions, to accelerate in the presence of the strong electric eld. As the electrons are accelerated, they can collide with other gas particles ionizing them also. Then a large bunch of electrons, called an avalanche, is propelled towards the electric plate which reads the electric signal. The chain reaction creating more electrons is necessary, since the current due to a single electron would be too dicult to detect. The rst class of muon detectors is the high position resolution detector. As the title suggests, these detectors are able to trace with relatively small error the muons that pass through them. The detectors work in general by having many partitioned sections that can detect the electrons. As an avalanche occurs, many of the electrons will enter these small distinct regions, which will in turn allow the location to be closely approximated. For example, in cathode strip chambers at the end-caps of the CMS, there are many wires in one direction and cathode strips in another, which allow both the x and y components of the muon's path to be found. There are also multiple layers of these detectors which in turn can trace the z component, which completes the information required to reconstruct a the path. [1] The other type of muon detector is triggering detector. These types of detectors have a very fast response time on the order of 10s of nanoseconds in a Resistive Plate Chamber (RPC) in the CMS, but they have less position resolution. The fast time resolution for these types of detectors allow for the precise time of the muon transits to be determined. By pairing this accurate time data with the position data of the other types of detectors, the trajectory and hence the collision at the center of the CMS can be better determined. [1] 2.3 GEM Detectors A GEM detector is the title given to a muon detector that utilizes a GEM, though it is dierent from a GEM. A GEM is a thin composite sheet of an insulator sandwiched between two thinner pieces of metal. There are many tiny holes in the GEM which allow for the passage of electrons through it. A large electric potential dierence is supplied to the two metal plates of the GEM which creates a very large electric eld in the holes. As the electrons pass through the hole, they gain a large amount of energy and as a result collide with and ionize more atoms of the gas. This creates an avalanche of electrons, multiplying the number of electrons as they pass through the holes by up to 10 times. [3] GEMs can be implemented in other types of detectors to bypass other negative aspects of the detectors, or can be the main mechanism of a GEM detector. As an example of how they are implemented, they can be used in Micro-Strip Gas Chambers to allow the MSGC to operate well below the critical potential for discharges. [3] Under normal operation close to the maximum electron gain limit, the MSGC can be damaged beyond repair by a discharge triggered by heavily ionizing paths. By adding a GEM, the operating voltages can be signicantly decreased and as a result, the detector can operate safely with little chance of self destruction. [3] GEMs can also be applied several times within a given gas volume to increase the total number of electrons that will reach the signal receptor. A single GEM can be applied to a thin gas layer, or for more 4

5 amplication, several can be applied one after the next. In general, several GEMs would be applied in a row with the readout plate at one end and the drift gap at the other end. In one version of the double GEM, which is shown below, the readout, or induction plate, is between the drift gaps and the GEMs. This could allow for a faster detection time, since the electrons have to drift at most 2.5 mm as opposed to the 4 mm in a single GEM with 3 mm of drift gap. It is impossible to shrink this drift gap too much, because there needs to be a large enough volume so that when the muon passes through it, there will be a high probability of ionization and thus detection. If the drift gap is too thin, there will be no ionizations and the muon will pass undetected. 2.4 Desired gas characteristics In order for a GEM detector to be useful, it has to be able to detect muons quickly, eciently, and without wearing out too rapidly. Several characteristics are needed in order to make this happen. First, the gas mixture needs a gas in which the majority of the ionizations take place. Often noble gases such as helium, argon, or neon are used. In addition to the main ionizing gas, the gas mixture needs to have a quencher gas such as methane, carbon dioxide, or isobutane, to prevent the undesired secondary eects such as photon feedback. The nal gas mixture must also have adequate tracking and triggering properties. These properties are linked so several observables such as drift velocity, diusion coecients, and the Lorentz angle. Drift velocity is the mean velocity of electrons moving through a gas under the inuence of an electric eld. It is sensitive to changes in pressure, temperature, and pollutant concentrations in the gas. Diusion coecients describe the fact that when very small particles such as the electrons in the gas move through a gas mixture, they tend to deviate from eld lines due to scattering o the atoms of the gas. The scattering makes a variation in velocity, known as the longitudinal diusion, and a variation in lateral displacement, called transverse diusion. It is ideal for the gases to exhibit low diusion coecients so that the electrons move as close to predicted as possible. As a charge moves through a magnetic eld, it experiences a force: F = q v B. This has important implications in the case of electrons moving through a GEM detector. Since the electric eld moves the charges perpendicular to the plates, if any component of the magnetic eld is perpendicular to the electric eld, then electrons will experience a force that is perpendicular to the electric eld. This will cause the electrons to deviate from their otherwise straight average path. The angle that the new net drift velocity makes with the electric eld is dened as the Lorentz angle. Since the electrons must strike the plate below in order to be detected, it is best to have a small Lorentz angle to maintain a high eciency. 3 Approach 3.1 Simulation methods First, to study the specic gas mixture properties, Gareld++ was used, which in turn uses the libraries in Magboltz to solve the equations for the gas properties just discussed. By adjusting the electric and magnetic elds and the angle between them, drift velocity, the diusion coecients, and the Lorentz angle were attainable for each gas mixture that was used. First, to generate the graphs for the drift velocity and the diusion coecients, a zero Tesla magnetic eld was used, while varying the electric eld from 100 V/cm to 15,000 V/cm using logarithmic spacing with 30 specic electric eld values. Next, to determine the Lorentz angles, a variety of magnetic eld values, electric eld values, and the angle between the eld directions were needed. This is because the Lorentz angle depends on the angle between the direction the electron is traveling and the magnetic eld as well as the speed the electron is traveling. Three dierent electric eld values, 1 kv/cm, 5 kv/cm, and 10 kv/cm were used. Also, ve angles in increments of π 8 from 0 to π 2 were implemented. Finally, the magnetic eld was made to range from 0 to 5

6 4 Tesla in increments of 1 2 for each of the above parameters. There were therefore a total of = 135 trials, which takes a signicant amount of computing time. For a sample of the Gareld++ code which does this, look at APPENDIX A or something like that. By adjusting, according to the desired specications, the ANSYS code for a triple GEM, a model of a double GEM was made. A total drift gap of 3 mm was implemented so that the electrons don't have too far to travel, which would increase the time it takes to detect the muons. The geometry of the double GEM can be seen in gures 2 and 3 below. Figure 2: A side view of a double GEM muon detector. In the actual ANSYS le, where the geometry of the GEM was precisely specied, only the region in the rectangle, in gure 3 below, was built. ANSYS solved for the electric eld and then when the simulations were actually run later, Gareld++ patched many of the same regions together which resulted in the total grid that is shown in gure 3. In order to create a region like in the rectangle: Dene the kapton layer and the two copper layers on either side of it by each of their dimensions in µm. Dene ve sets of two cylinders that overlap with the copper layers and ve sets of two cones in the copper layer that taper down to the middle to give the shape in Figure 2. Subtract the volumes to get the desired shape, leaving only the copper with holes and the kapton with double tapered cone holes. Oset the working plane to a dierent z coordinate and make another GEM layer in the same way. Make a gas volume which encompasses everything including both gas GEMs and goes right up to the drift plates. See the above diagram for dimensions. Subtract the GEMs and the plates from the gas layer. Dene and apply voltages to each of the top and bottom layers of the GEMs along with the drift plates and the induction plate. 6

7 3.2 Data Analysis Figure 3: A detailed depiction of the GEM foil. Since each of the simulations took many hours to complete, they were often broken into smaller jobs and written into a log-le. Using this method, several jobs were completed at once or were left to run overnight so that the outputs didn't have to be monitored. After the simulation was completed, the data was gathered from the log les and transcribed it into Excel so that useful plots were obtained. The drift velocities and diusion coecients were each plotted directly, but the Lorentz angle was calculated. To do so, the z-component of the drift velocity which is in the direction of the electric eld, and also the x and y- components of the drift velocity were recorded. To determine the total amount which the electrons were displaced, d = x 2 + y 2, was used. tangent, the Lorentz angle was obtained. Namely Θ L = arctan ( d z 4 Results By dividing this by the z-component, ) and then taking the inverse. [4] Below are the graphs of the drift velocities and longitudinal and transverse diusion coecients for both gas mixtures, Ar:Isobutane 95:5 and Ar:CO 2 :CH 4 40:15:45. 7

8 Figure 4: Drift velocity data Figure 5: Longitudinal diusion coecients 8

9 Figure 6: Transverse diusion coecients According to gures 4 through 6, since in the drift region typical electric elds are around 3 5 kv/cm, the Argon, CO 2, and CF 4 mixture looks the most promising. This is problematic however, since freon, CF 4, is harmful to the ozone layer and will no longer be able to be used in detectors. [5] Aside from its detrimental eects, the rst reason that the freon mixture would work better than the isobutane one is, as you can see in gures 5 and 6, both the diusion coecients are smaller than that of the Ar and isobutane mixture. This would mean a more centralized charge distribution within the GEM, allowing for better position resolution. Although for the low end of the proposed electric elds, the Argon and isobutane mixture has a greater drift velocity, as the electric eld increases, the Argon, CO 2, and CF 4 mixture greatly overtakes it as seen in gure 4. Since the operating electric eld would likely be in the middle/upper range, the faster drift velocities of the Argon, CO 2, and CF 4 mixture would also allow for better time resolution. There are several ways to present the information of the Lorentz angles because the electric eld strength, magnetic eld strength, and the angle between them were all varied. First are the plots of the Lorentz angle versus magnetic eld for a constant electric eld while varying the angle between the electric and magnetic elds. The second group of plots are of the Lorentz angle versus magnetic eld for a constant angle while varying the electric eld strength. Here are the graphs for the gas mixture Ar:CO 2 :CH 4 40:15:45. 9

10 Figure 7: Lorentz Angle for Ar:CO2:CF4 40:15:45 (by drift electric eld) Figure 8: Lorentz Angle for Ar:CO2:CF4 40:15:45 (by angle between E and B) Here are the graphs for the gas mixture Ar:Isobutane 95:5. Figure 9: Lorentz Angle for Ar:Isobutane 95:5 (by drift electric eld) 10

11 Figure 10: Lorentz Angle for Ar:Isobutane 95:5 (by angle between E and B) The Argon:Isobutane 95:5 mixture doesn't look the best suited for the GEMs, because they have fairly large Lorentz angles compared to to the other gas mixture in the rst 3 graphs. Felipe Oviedo, one of the other students working on the project, did the simulations for the other gas mixture. Although the freon mixture looks better, it is an impracticable option. 5 Summary and Conclusion From the data we took, as far as these parameters go, the Argon, CO 2, and CF 4 mixture is the better gas to use for the multi-gem, but this is not possible since it has freon in it. Although the geometries were created, not enough tests of the drift times and eciencies were made to come to a clear conclusion on the most ecient detector set up. The number of gases used in our trials were also clearly incapable of determining the best gas for the multi-gem detectors, since so few were analyzed due to time constraints and learning the programs which were used. References [1] A. Sharma. Muon tracking and triggering with gaseous detectors and some applications. Physics Research Section A: Accelerators, Spectrometers, Detectors, and Associated Equipment, Print. [2] CMS Physics. Volume I: Detector Performance and Software. Technical Design Report, 2 Feb Print. 11

12 [3] R. Bouclier et al. The Gas Electron Multiplier (GEM). SLAC, Stanford U Web. 11 Aug < [4] K. Hoshina et al. Lorentz Angle Measurement for CO 2 /Isobutane Gas Mixtures. arxiv Web. 11 Aug < [5] M. Bellis. Freon. About.com Web. 11 Aug < 2013>. 12