SiD Calorimeter R&D Report - UTA HEP WWW Home Page
SiD Calorimeter R&D Report
I. Overview of SiD Calorimeter R&D.
The ILC physics program places strong requirements on the performance of calorimeter systems. Many production channels to be studied require efficient identification and separation of jets in multi-jet final states (e.g. in e+e- → t tbar → 6 jets), and the measurement of jet energies and jet-jet invariant masses with unprecedented precision. Our goal is to measure jet energies at the level of σ/E ~3-4%. Event-by-event separation of hadronic decays of W- and Z-bosons and Higgs particles will be necessary to measure cross-sections and branching ratios at the few percent level of error. The calorimeter system is a critical element of the SiD detector concept design for achieving these goals. The calorimeter will also provide excellent electromagnetic energy resolution and efficient π0 reconstruction, charged hadron/photon shower separation, and tau lepton reconstruction.
The SiD detector design is based on an all-silicon tracking system which allows for a limited inner radius for the central calorimeter system. This, in turn, constrains the cost of the silicon-tungsten electromagnetic calorimeter, and limits the outer radius of the hadron calorimeter, thereby also constraining the size and cost of the superconducting coil. It is taken as a basic design parameter that both the electromagnetic and hadron calorimeters should lie inside the coil so as not to degrade their abilities to make particle-energy cluster associations. A schematic view of part of the SiD concept is shown in Fig.1, which indicates the dimensions and location of central and endcap components of the calorimeter system.
[pic]
Fig. 1 Quadrant view of the SiD detector design concept.
The forward calorimeter systems for SiD will be covered in a separate report by the FCal collaboration, and so are not discussed here.
A promising approach for achieving the unprecendented jet energy resolution lies in the use of a Particle Flow Algorithm (PFA). We have invested considerable time and effort in the development of such algorithm(s), and in understanding the implications of a PFA for the SiD detector design. This approach uses the tracking system to measure the energy of the charged hadron component of a jet and the electromagnetic calorimeter to measure photon energies. Accordingly, the neutral hadron energy is the only component of jets to be measured directly in the hadron calorimeter. Leakage from the rear of the hadron calorimeter could be estimated by the use of a “tail-catcher”, which would also be the first part of the muon system.
The measurement of jet energies using data from several subsystems leads to a very integrated view of the design of the SiD detector. The tracker must find the charged particles with high efficiency and ensure accurate projection of the tracks into the calorimeter to make the correct track/energy cluster associations. The electromagnetic calorimeter must be sufficiently fine grained to provide efficient charged particle/photon shower separation. The hadron calorimeter must be sufficiently fine grained to allow efficient track following for charged hadrons and their hadronic shower products, and to provide an approximately linear hits vs. energy for the digital hadron calorimeter case.
Simulations of PFA performance for light jet production at the Z-pole are beginning to show excellent performance. Intense work is underway to optimize the algorithms for the higher jet energies expected for higher center of mass energies. Prudence suggests we also remain open to the investigation of more traditional calorimetry or more radical, alternative ideas.
In the PFA approach, the electromagnetic energy resolution does not impose limits on jet energy resolution: the current Si-W design is expected to have an electromagnetic shower resolution of 17%/√E. However, the segmentation, both longitudinal and transverse, is a critical consideration for the use of PFAs for hadronic jets. Transversely, not only should the effective Moliere radius be kept small, but also the transverse cell size should be about half of this radius. Longitudinally, the active gap between absorber layers should be kept small to achieve a small effective Moliere radius, limiting transverse shower spread. A sufficient number of longitudinal layers need to be provided to contain electromagnetic showers and allow efficient recognition and reconstruction of these showers. The resulting highly segmented electromagnetic calorimeter, in addition to having the correct characteristics for use with PFAs, will then have outstanding performance for the reconstruction of electrons, photons, and tau leptons and for tracking charged hadrons and muons.
The choice of the silicon-tungsten combination for the electromagnetic calorimeter is a centerpiece of the original SiD design. The Moliere radius for tungsten is 9mm, and our present design has a pixel area of 12mm2. Longitudinally, there are 30 alternating layers of tungsten and silicon giving a total thickness at normal incidence of 29X0. The goal for the active gaps is 1mm width or less. Each group of approximately 1000 silicon pixels is read out via one KPiX-ASIC mounted on the wafer. The first version of the silicon wafers have been delivered and evaluated. The medium-term plan is to assemble a prototype calorimeter section based on the second generation silicon wafers in conjunction with the 1024-channel KPiX readout. The electromagnetic calorimeter is discussed in detail in Section II.
The KPiX ASIC is being developed for the readout of the electromagnetic calorimeter, and the central silicon tracker for SiD, and has applicability to several technology options for the hadron calorimeter. The high energy density for electromagnetic showers requires a large dynamic range for the readout. The KPiX chip uses a novel method by which the feedback path on the front end amplifier can be switched between two capacitors, switching in the large ((10 pF) capacitor only when it is required. This allows the amplified charge for smaller input signals to be well above the noise. There is an event threshold, which can hold off bunch crossing resets in order to allow a fairly long integration time of (1 (s. The calculated noise level is about 1000 electrons, to be compared with a MIP signal charge of 25 times this. Charge digitization uses two overlapping 12-bit scales. The chip also allows up to four hits per bunch train to be stored for each pixel. Several prototype versions of KPiX have been fabricated and tested, with encouraging results for the present 64-channel version. A version of a 64-channel KPiX has also been developed to read out the gas electron multiplier based version of the hadron calorimeter active layers. Details of the KPiX development are given in section III.
The hadron calorimeter in our PFA-oriented, baseline design consists of 35-40 layers of alternating absorber plates and active sampling gaps. The critical feature of this calorimeter is the small transverse cell size, O(1-3cm), required for the imaging requirements imposed by a PFA. Depending on the approach taken, this device could be analog, semi-digital, or fully digital. There is a premium on a small active gap size as this has a large impact on the overall detector cost. The full depth of the baseline hadron calorimeter is 4λI. The basic mechanical structure is foreseen to have three main sections: a barrel and two endcaps. The barrel is subdivided along the beam direction into three sections, and there are twelve modules in azimuth.
The baseline material choice for the absorber is steel, although we have also considered tungsten and brass, with plates having thickness of 20mm or 1.1 X0. In contrast four technologies, three gaseous and one plastic, are under consideration for the active layers of the hadron calorimeter. These range from resistive plate chambers, gas electron multipliers, and micromegas for digital implementations, to small scintillator tiles for semi-digital and analog implementations. Prototypes of each technology have been built and evaluated.
We have had extensive discussions on the process for selection of the best technology for the SiD hadron calorimeter. We have enumerated a set of requirements for the calorimeter and selection criteria for the technologies, and the steps that should be followed to reach a decision. These requirements and criteria incorporate the demands of PFAs. For instance, our PFA studies are now focusing on center-of-mass energies beyond the Z-pole, and consequently, higher energy, more collimated, jets. This helps us understand the requirement for the transverse cell size. Also our simulations have highlighted the pros and cons of neutron detection. The issue is whether we gain in the PFAs from the use of scintillator to see neutron clusters, or whether it is better to use a gas active medium to suppress the neutron signal and thereby simplify the track-cluster association in the PFA. Also at issue is the comparative ease of construction and cost differences between the various technologies. A full discussion of the hadron calorimeter is given in Section IV.
The option of installing a “tail-catcher” after the ~1.5 λI of the superconducting coil and cryostat, to potentially identify and measure the last few percent of hadron shower energies, is under discussion. Recent prototype results from CALICE data taking at CERN show improved energy resolution when the tail-catcher data is included. The technology for implementation of the tail-catcher would follow that of the muon system: currently scintillator strips and resistive plate chambers are being considered. The tail-catcher is also discussed in Section IV.
Basic simulation studies of hadron calorimetry are being carried out in the SiD context. The parameters studied include the absorber material, the active layer material, the longitudinal and transverse segmentation, the total thickness of the hadron section, the thickness (λI) per layer, the inner radius, and the magnetic field. The idea is to work from a basic understanding of single particle energy resolutions, through jets of known energy(s), to full dijet mass resolution in physics events. The single particle studies already reveal interesting and significant differences in energy resolution between the combinations of steel and tungsten absorber and scintillator and gas active media that will be essential input to decisions on the hadron calorimeter. Studies of jet energy resolutions and neutral energy measurement are ongoing. Section V discusses this simulation work.
Initially the study and development of PFAs for SiD was undertaken by a number of independent physicists at several different locations. In order to bring some structure into these studies, and provide a common basis for comparison of algorithms within the various PFAs, a framework has been produced by a collaboration between SLAC and ANL. Initial studies for SiD, as for the other concepts, have focused on the Z-pole region. Based on physics processes requiring the separability of W and Z bosons, the initial goal for the PFAs was set at achieving an energy resolution at the Z-pole of σ/E ~30%/√E. However, more recent considerations particularly at higher center-of-mass energies have revised this goal to σ/E ~3-4%. The current state-of-the-art for SiD PFAs is σ/E ~35%/√E. We are approaching the level of understanding of the PFA performance that will provide the basis for meaningful comparisons of technology choices for the hadron calorimeter,
DAQ, fiber cables, data rates?
Backgrounds in the electromagnetic calorimeter, mainly low energy photons from the beamcal, and photons from gamma-gamma interactions, produce an occupancy rate of at most 1 x 10-4/BX in the highly segmented silicon pixel detectors. This should be an inconsequential effect in calorimeter pattern recognition, and the electronic buffer size is more than adequate to ensure full efficiency.
We are forming a SiD Engineering Group, with initial participants from SLAC and Fermilab, to carry out a preliminary engineering study of the SiD detector design. This study will consider the basic design of the calorimeter modules, materials, support of the barrel and endcap calorimeters and the superconducting coil, assembly procedures and magnetic force effects .This generic study will assume a calorimeter inside the coil and active gaps of order 1cm.
Stability, calibration, push-pull ?
In this report, we present the electromagnetic calorimeter and its readout in detail. For the hadron, we summarize the R&D relevant to the SiD calorimeter design and technology selection. The details of the hadron calorimeter and TCMT R&D will be discussed in the CALICE report submitted to this review. We include a section on basic simulation studies as these offer valuable insight into the relative importance of various materials and size parameters for the calorimeter design. Finally, we summarize the results from SiD-based PFA studies that also bear on the optimization of this design.
II. Electromagnetic Calorimeter
1. Introduction
We expect pattern recognition to be critical for identifying and quantifying final states, which may include hints of new physics in complicated high-energy events at the ILC. A dense, imaging ECal will have excellent capabilities for identifying and measuring EM showers in three dimensions, for separating these from other particles, for tracking charged particles in the ECal, and for identifying charged and neutral hadrons which begin showering in the ((1 interaction length) ECal. In addition to its critical usefulness for reconstruction of hadronic final states using the PFA, as discussed in Section I, a highly-segmented, imaging ECal such as we propose will have important advantages for carrying out other critical elements of the broad ILC physics program. We outline some of these benefits in the next section.
Hence, the silicon-tungsten (Si-W) ECal we propose is not only a necessary element of a detector which uses PFA methods for jet reconstruction, but its general ability to aid in the identification and measurement of many types of particles – by direct identification or by reconstruction of individual decay products – makes, we believe, a dense, imaging ECal the optimal choice for a multi-purpose ILC detector. The main questions are (1) can we afford it? and (2) is there a technological solution? The answer to the first question may depend on the eventual cost of silicon sensors. Nevertheless, our goal is to design an excellent instrument for exploring and quantifying new physics, but with constrained costs. We believe we have found a positive response to the second questions using the Si-W technology which we discuss in Section 3.
Table II-1 gives the global parameters of the ECal in the SiD baseline model. A cartoon depicting a possible overall structure of the barrel is shown in Fig. II-1.
Table II-1. Main parameters of the SiD silicon/tungsten ECal
|inner radius of ECal barrel |1.27 m |
|maximum z of barrel |1.7 m |
|longitudinal profile |(20 layers x 0.64X0 ) + (10 layers x 1.3X0 ) |
|silicon segmentation |13 mm2 (hexagonal) pixels |
|readout gap |1 mm |
|effective Moliere radius |13 mm |
|pixels per readout chip |1024 |
[pic]
Figure II-1. Possible overall mechanical layout of the ECal barrel. The thermal load and digital readout signals are extracted at the exterior regions at the module edges.
2. Physics Requirements and Design Goals
As discussed in Section I, a generic physics requirement for ILC detectors is that they provide excellent reconstruction of hadronic final states. This allows access to new physics which is complementary to the LHC, where reconstruction of hadronic (multi-jet) final states is problematic, and where QCD backgrounds dominate new physics signatures involving jets.
How well do we need to reconstruct multi-jet final states? Of course, we do not know what the new physics will be, but a rather generic requirement is that intermediate particles which decay into jets, such as W, Z, or top, can be identified and isolated. This places unprecedented requirements on 2-jet or 3-jet mass resolution, typically at the level of 3-5%. The PFA technique, discussed separately, is capable of providing this level of performance. However, the PFA does make challenging demands on the calorimeters, which we believe we can meet.
With a PFA, electromagnetic energy resolution is not expected to limit jet resolution. However, particle separation – photon-photon and charged hadron-photon – is crucial. In addition, if one provides this kind of imaging calorimeter to meet the PFA needs, these same features will also be put to good use for reconstruction of specific tau decay modes (to enable final-state polarization measurement), to “track” photons, to track MIPS, and so forth.
The few mm segmentation possible with a Si-W ECal provides outstanding capability for reconstructing individual photons and for separating the photons from other photons and from charged hadrons. At the same time, good electromagnetic energy resolution is achievable -- designs to date have used sampling fractions which give EM resolution of approximately 15%/sqrt{E}. So far, a strong physics argument for providing better EM resolution has not emerged. In addition, such an ECal will provide excellent lepton reconstruction. In particular, tau final states, so crucial for many signatures of new physics, will greatly benefit from such a highly-segmented ECal. It is likely that we will continue to see benefits from these designs as the physics simulation studies become more detailed, for example in flavor tagging. Physics opportunities with a dense, imaging ECal include:
• Enabling application of PFA technique for reconstruction of multi-jet final states at a level which allows precision measurements, for example for by cleanly separating W and Z decays to jets.
• Cleanly identifying tau final states is crucial in SUSY and Higgs physics. Taus can be cleanly identified.in many decay modes.
• Separation of [pic] from [pic] allows measurement of tau polarization, thus providing a handle on the electroweak couplings of the parent particle, such as a [pic]. Figure II-2 shows a simulated [pic] decay in SiD with its Si-W ECal.
• Tracking of photons. Gauge-mediated SUSY models, for example, include decays of long-lived particles to a photon plus invisible. The ability to accurately measure photons and determine their distance of closest approach to the IP (resolution of ~1 cm) would be crucial.
• Exceptional electron identification in a multi-jet final state.
• Precise measurement of Bhabha and radiative Bhabha events. In addition to the usual physics implications, these measurements provide crucial information on the luminosity spectrum (via the Bhabha acollinearity) and the beam energy (via radiative returns to the Z).
• Tracking of MIPs. This is especially important in V0 decays in a silicon tracker, where the tracks in the ECal provide crucial track seeds,
• Tracking of MIPs as an integral part of charged hadron identification in PFAs and for muon identification.
• (o decays can be accurately reconstructed. Placing a mass constraint improves the jet energy resolution.
[pic]
3.1 Transverse segmentation requirements
Based on past experience with silicon-tungsten luminosity calorimeters at LEP and SLD, two electromagnetic showers can be usefully resolved when separated transversely by about half of the effective Molière radius. Here, “effective” includes the shower spreading in the readout gaps. In our case, this implies that two showers can be resolved when their separation is about 6 mm or more. Although the current algorithms have not yet attained this level of performance, this will presumably be readily achieved. .
Charged track – shower separation should be possible when the tracks are on average separated from the shower core by about 10 mm or more. Due to bending in the field, it may be possible to do better than this. Simulations involving PFAs or simply examining particle densities indicate that a segmentation of 5 mm or less will not limit PFA performance for jets of about 100 GeV energy. We have chosen a smaller segmentation to accommodate higher energies, to aid pattern recognition by using pixels well below the Moliere radius, and to provide improved MIP resolution. As we explain in the next section, a smaller segmentation (to no smaller than a few mm) comes at no cost.
3.2 Longitudinal segmentation requirements
One of the issues pertaining to longitudinal segmentation is to identify and separate hadrons which begin to shower in the ECal. The ECal represents about 1 hadron interaction length, so we must have sufficient sampling throughout the ECal to identify these showers. This has not yet been optimized, either in detector design or in pattern recognition studies using simulations.
The longitudinal sampling has been chosen to provide moderate energy resolution, while attaining the smallest practical Moliere radius and providing the sampling needed to provide the high-quality imaging required by the physics. More frequent sampling provides better energy resolution and imaging, but with degraded Moliere radius and higher cost. Both pattern recognition and energy resolution point toward putting more sampling in the inner layers where the EM showers are narrower and deposit most of their energy. So for the baseline design, we have chosen 30 layers, with the first 20 layers having tungsten of 2.5 mm thickness, and the back 10 layers being twice as thick. The tungsten is actually a 93% tungsten alloy, which makes it possible to readily roll and machine. Hence, the radiation length is slightly larger than pure tungsten, YY X0, giving the thicknesses in radiation lengths shown in Table II-1. The total thickness is about 25Xo.
This baseline configuration gives an EM resolution of [pic] for photons up to 50 GeV. We have verified that both EGS4 and Geant4 simulations give consistent results, after ensuring that they are set up to handle the thin silicon sampling. Figure II-3 provides the EM resolution (for 1 GeV photons) as a function of silicon and tungsten thicknesses. The variation with silicon thickness is due to straggling – non-gaussian fluctuations in ionization loss. Figure II-4 shows that the energy resolution at low energy does not suffer relative to a design with 30 layers of 2.5 mm thickness. The low-energy resolution is most important for PFA performance. The EM resolution need only remain better than ~20%/sqrt(E) to remain a negligible contributor to jet resolution. So while the low-energy performance does not suffer, the containment of high-energy showers is significantly better at high energy with the 20+10 configuration – about a factor two better at 250 GeV. We haven’t yet optimized the longitudinal sampling function, with one of the open questions being whether the HCal can correct for shower leakage in a useful way.
[pic]
Figure II-3. Geant-4 simulations of 1 GeV photons giving contours of equal resolution as functions of silicon and tungsten thickness. The yellow contour represents a resolution of 14%/sqrt(E), for example.
[pic]
Figure II-4. Energy resolution for photons as a function of photon energy for two longitudinal configurations. The “20+10” configuration is the baseline design.
4. Proposed Technical Solution
Silicon-tungsten is the ideal choice to meet the requirements stated above, as it is dense, compact, and easily segmented. Si-W calorimeters have been successfully employed as luminosity calorimeters at SLD and LEP. However, the scale here (~1000 m2 of silicon) is significantly larger. An outstanding technical question, then, is how to integrate a silicon detector wafer with its readout electronics. Since the number of detector pixels for the ECal design is on order 80 million, a solution to the integration issue, along with the cost of the silicon detectors themselves, is likely to determine the overall viability of the Si-W approach. A few years ago, we proposed a possible solution to the integration problem and have received U.S. LCDRD support for three years to pursue this. The integrated approach can accommodate the design discussed in the previous sections.
The thrust of our project is to integrate detector pixels on a large, commercially feasible silicon wafer, with the complete readout electronics, including digitization, contained in a single chip which is bump bonded to the wafer. Our baseline design uses a pixel size of 13 mm2, which corresponds to 1024 pixels per silicon sensor, based on a 6-inch wafer. We take advantage of the low beam-crossing duty cycle ((10-3) to reduce the heat load using power cycling. Our design then has several important features:
1. The electronics channel count is effectively reduced by a factor N=1024.
2. A transverse segmentation down to a few mm can be naturally accommodated.
3. The cost, to first order, will be independent of the transverse segmentation.
4. Readout gaps can be small ((1 mm), thus maintaining the small Molière radius intrinsic to tungsten.
The first property, we feel, is necessary for any realistic highly-segmented ECal. In this case, the electronics is likely to be relatively small fraction of the ECal cost. The third point makes the design flexible, so that one can optimize to meet the physics goals. The fourth is an optimization of the physics capability of the ECal at a given (barrel) radius. For example, the angle subtended by the Molière radius for an ECal at radius 1.27 m with our design is smaller than one with 3 mm readout gaps at 1.7 m. Hence, this has a significant impact on both performance and overall detector cost. We note that for a Si-W ECal, the features above remain unique to this R&D. Figure II-5 is a schematic of the silicon sensor, showing the positioning of the readout ASIC (called KPiX) and a few representative signal traces which bring the signals from individual pixels to the ASIC, where the signals are digitized. The signal traces are metallized on the detectors as part of fabrication. Figure II-6 shows a cross section of the silicon sensor, indicating the signal trace metallization. In order to keep our design saimple, and therefore reduce costs, we use DC-coupled readout. Figure II-7 depicts a readout gap cross section. A kapton cable for the digitized signals and KPiX power and control signals helps maintain the small 1 mm gap. The KPiX R&D is described in more detail in Section III of this report.
[pic] [pic]
[pic]
Figure II-7. Readout gap schematic near the KPiX ASIC. A few of the 1024 bump bond connections are shown for illustration,
4.1 Dynamic range
The readout of the Si pixels in the ECal must accommodate a very large dynamic range. The EM showers in tungsten are exceedingly dense. Based on EGS4 simulations, the largest signals in a single pixel – arising from 500 GeV Bhabha electrons – correspond to about 2000 MIPs at shower max. At the low end, one requires measuring MIPs with very good signal to noise. As discussed in the next section, the KPiX design incorporates this large dynamic range in a novel way, using on-the-fly range switching. Figure 4 of Section III shows this range-switching function in action in the lab. In the plot, as the injected charge is increased, we see the range switch at about 700 fC. For 320 micron silicon, 1 MIP is equivalent to about 4.1 fC. Thus the upper end of the plot corresponds to about 2500 MIPs, more than the expected maximum. Besides providing for the maximum charge, the range switching also maintains good signal/noise for MIPs.
4.2 Thermal management
If the KPiX heat load is kept below about 40 mW, the temperature gradient across an ECal module can be kept at an acceptable level ( 14 bit |
|Noise (ECAL mode) |< 2,500 e |< 3,000 e |
|Max Signal (ECAL, 5 mm pixels) |10 pC |> 10 pC |
|Auto Self-trigger |Yes |Yes |
|Cross-talk |< 1% |tbd |
|Power (for 1,024 channels) |< 40 mW |~ 20 mW |
|Positive or negative input signals |yes |yes |
|Noise (GEM mode) |< 40,000 e |< 40,000 e |
|Nearest neighbor logic |yes |yes |
|Leakage current compensation |> 4 uA |> 5 uA |
In Figure 4 a calibration measurement result showing the low and high gain auto switching are shown.
[pic]
Figure 4: Calibration injection transfer function, low and high auto-select gain range response.
A prototype of the flex-cable interconnecting up to 12 KPiX/detector assemblies has also been designed and is waiting to be fabricated. Tests to bump-bond KPiX chips to detectors are in progress.
3. Plans
Evaluation of the 64-channel version is still in progress. Additional performance tests are planned, e.g. more cross-talk and noise measurements, forced neighbor trigger analog performance, and more measurements using different kind of radiation sources.
A measurement setup has been completed for a test-beam investigating the GEM mode. In June 07 the performance of the device is expected to be measured in the test beam.
Layout changes to further reduce voltage drops on power traces on the chip and cross-talk between channels have been completed. The current plan is submit another 64-channel device with subsequent evaluation before submitting the full 32x32 channel KPiX.
The current KPiX will be assembled to a detector and integrated with a flex-cable to investigate system performance
VI. Hadron Calorimetry
1. Introduction
The Hadron Calorimeter (HCal) is a sandwich of absorber plates with gaps instrumented with active detector elements. To satisfy the stringent imaging requirements of the PFA algorithm transverse segmentation is required to be small of order a few cm2 and each layer is read out separately. This in principle allows an analog or digital treatment of the signals.
The current baseline uses steel for the absorber and resistive plate chambers as the detector. Alternative detector possibilities, consisting of Gas Electron Multipliers (GEMs), Micromegas or scintillator are also under study. One of the design criteria for the HCal is to minimize the size of the active gap, because an increase in the gap size has a large impact on the overall detector cost. The current gap size is 12mm.
The absorber consists of steel plates with a thickness of 20 mm or approximately 1.1 X0.
The cell structure, which is identical for the barrel and the endcaps, is repeated 34 times, leading to an overall depth of the HCal corresponding to four interaction lengths. Tungsten has been and will also be considered as a absorber, but is currently not the
baseline.
The inner radius of the HCal is 138 cm, the outer radius of the barrel is 233.7 cm and its length is 554.0 cm. The endcaps start at a distance of 179.65 cm from the interaction point. The SiD baseline design uses steel as absorber material in the HCAL. Steel with a radiation length Xo = 1.8 cm and an interaction λI = 16.8 cm offers the smallest X0/λI of all commonly used absorber materials. A small ratio permits a one radiation sampling, while keeping the number of active layers manageable for a hadron calorimeter with a depth equivalent to four interaction lengths at 90o to the beam pipe.
The first attempts at a mechanical structure foresee a barrel and two endcaps, which are
inserted into the barrel structure. The barrel is subdivided along the beam direction into
three sections (Fig. XX), and there are 12 modules to make a complete ring in azimuth.
[pic] [pic]
Fig. XX Schematic of SiD Hadron Calorimeter (left), and single module (right).
The steel plates are held in place by a set of picture frames located at each end (Fig. XX). Readout cables, high voltage cables and the gas lines are routed to the outer radius of the barrel structure through the openings in the picture frames. In azimuth the barrel structure is subdivided into twelve modules. Each module weighs approximately ten tons and will
be held in place through supports in the cryostat of the solenoid. Deflections have been calculated and do not exceed 0.5 mm at any point of the structure.
Prototyping and testing of active gap technology options has been ongoing for several years. The basic characteristics and initial viability of each option have been established through a series of cosmic ray and beam tests. For the gas-based technologies the next stage will involve a “slice test” of a number of RPC and GEM chambers with absorber layers. The scintillator based medium has been validated with test beam studies and focus has moved to production issues. Beyond this we plan to build one or more ILC prototypes of approximately 1m3 size to contain hadron showers, and read out using these technologies. This scale of testing, involving up to 400,000 channels is presently limited by funding in the US.
We have defined a set of criteria and requirements for the selection of the final active medium. We have established a time schedule for the convergence of the hardware tests and the PFA results. We anticipate making a selection on the timescale of the SiD CDR – around middle to late 2008.
2. Technologies under Consideration
Digital Hadron Calorimeter with Resistive Plate Chambers
Resistive Plate Chambers (RPCs) with small readout pads are an ideal candidate for the active medium of a hadron calorimeter optimized for the application of PFAs. RPCs can provide the segmentation of the readout pads, of the order of 1 to 4 cm2, which is necessary to keep the ‘confusion term’ small. They can be built to fit small active gaps (less then 10 mm) to maintain a small lateral shower size and to keep the longitudinal extend of the hadron calorimeter as short as possible. Glass RPCs have been found to be stable in operation over long periods of time [1], especially when run in avalanche mode, and their rate capabilities are adequate for the ILC and for test beam studies of hadronic showers. RPCs are inexpensive to build since most parts are available commercially. Signals in avalanche mode are large enough (in the range of 100 fC to 2 pC) to simplify the design of the front-end electronics.
Figure H1 shows a schematic diagram of a single-gap RPC. The chamber consists of two plates with high electrical resistance. Readily available window glass of thickness 0.8 to 1.1 mm is used. High voltage is applied to a resistive coating on the outside of the glass plates. The resistance of this coating must be low enough to allow for a fast local re-charge of the glass plate after a hit, and high enough to avoid screening of the electron avalanche in the gas from the readout pads located on the outside of the chamber. The glass plates enclose a gas volume in which ionization and electron multiplication takes place. Particles traversing the gas gap ionize the gas, creating an avalanche of electrons drifting towards the glass plate at positive high voltage. The signal is picked up inductively with pads located on the outside of the glass.
[pic]
Detailed simulation studies have shown that the energy response is preserved with a simple one-bit (or digital) readout of the pads, see Fig. H2. The single particle energy resolution with one-bit readout is found to be significantly better at lower energies, due to its insensitivity to Landau fluctuations. At higher energies the resolution with analog readout is superior, due to the high density of particles in the core of the shower. However, for the application of PFAs, only the response in the energy range below say 20 GeV is important. Using a simple digital readout, pads with an area of 1 x 1 cm2 appear to provide the best possible resolution without unnecessarily increasing the channel number.
Gas Electron Multipliers
The essential requirements for DHCAL include a robust design with stable, reliable operation; a thin sensitive layer for compact calorimeter design; on-board amplification/discrimination/digitization for digital readout; high efficiency for minimum ionizing particle (MIP) tracking in a hadron calorimeter; flexible design for implementation of varying cell sizes; in addition to the basic requirements of minimal supports/intrusions for hermeticity, ease of construction, and cost containment. As a member of the SiD concept group, the University of Texas at Arlington (UTA) group have been exploring an implementation based on the Gas Electron Multiplier (GEM) [H5] technology developed at CERN by Fabio Sauli and the GDD Group [H6], shown in Fig. H3. As shown in Fig. H3a – c, thanks to its structure of having 70μm holes separated by 140μm, GEM detectors can provide wide range of granularity, starting from microstrips to macro-pads. Figure H3d shows a conceptual design of a GEM based hadronic calorimeter in which double GEM active layers are alternated with layers of absorber to form a sampling calorimeter.
Figure H3. (a) A field line diagram of a GEM foil. (b) A simulated electron avalanche process through a hole on a GEM foil. (c) A microscopic photograph of a GEM foil. (d) A schematic diagram of a conceptual DHCAL using double GEM foil sensitive layers.
In order to study the viability of using GEMs as a sensitive medium in a DHCAL, the GEM layer geometry has been implemented into Mokka [H7], a GEANT 4 [H8] based simulation package, replacing the scintillation counter sensitive layers in the TESLA TDR hadronic geometry (stainless steel/ scintillation counter) with the double GEM layer structure. All other detector structure as in TESLA TDR detector design [H9] was retained. Using this simulation package, nearly linear behavior was observed for a wide variety of single particle incident energies and for the number of hits above the threshold. Much more detailed simulation studies of GEM DHCAL have been documented in two master’s students’ theses [H10, H11].
Figure H4. (a) GEM DHCAL single pion energy resolution. (b) GEM DHCAL jet energy resolution using a “perfect” PFA.
Simulated performance studies show that GEM response for analog and digital calorimetry very close to one another. Figure H4a shows single pion energy resolution for GEM DHCAL (green) comparable to the TESLA TDR detector (red) except at low energies. On the other hand, the GEM analog mode resolution is significantly worse than other detectors. Figure H4b compares jet energy resolution for a GEM DHCAL using a “perfect” PFA (blue), which has a sampling term of 30%/(E, to other detector technologies, using the single pion energy resolution obtained for GEM DHCAL. While the geometry is not that of SiD, the above studies give us confidence that GEM will perform well in a digital hadron calorimeter, providing the high jet energy resolution required for ILC physics.
Micromegas
A micromegas detector was proposed few years ago by Y. Giomataris, P. Rebourgeard, J.P. Robert, and G. Charpak. As in the case of Gas RPC it is a very good candidate for the active medium of a fine granularity hadronic calorimeter. The Micromegas detector is a gaseous detector also, based on the micro-pattern detector technology and widely used by many experiments: COMPASS, CAST, NA48, n-TOF, ILC TPC project, and the T2K TPC project.
A schematic view of the detector is shown in Figure H5: a commercially available fine mesh supported by pillars of 400 μm diameter separates the drift gap (~ 3mm ) from the amplification gap (~ 100 μm). A thin copper plate glued to the calorimeter absorber medium, defines the cathode, without adding extra space. Assuming a thin PCB for the anode pixel read-out (~0.8mm), this compact structure provides an economic active medium with a total depth of about 3.1mm, reducing the total radius of the HCAL. The cathode and the mesh are connected to high voltage (~800 and ~400 V respectively). The anode PCB is segmented into pads of 10x10mm2.
This simple structure allows full efficiency for MIPs and thanks to the thin pillars provides a good uniformity over the whole surface. The rate obtained with Micromegas chamber is not constrained. Another advantage of such a chamber is the tiny size of the amplification avalanche, resulting in fast signals without physical cross talk.
[pic]
Fig. H5 : Schematic view of a Micromegas chamber Fig H6 : T2K Micromegas TPC Prototype
A new promising technology ‘Bulk Micromegas’ has been recently developed in collaboration with the CERN PCB workshop. The basic idea is to build the whole detector in one process: the anode PCB plane with the copper pads, two photo-resistive films summing up to the correct pillar thickness, and the cloth mesh are all laminated together at high temperature, forming a single object. By a photolithographic method the photo resistive material is then etched which produces the pillars supporting the mesh. The new industrial process allows easy implementation and provides uniform, light, low cost and robust detectors. Employing the bulk technology, large detectors have been built for the TPC prototype of T2K experiment as shown in Fig. H6.
Scintillator/SiPM
The scintillator-based hadron calorimeter option capitalizes on the marriage of proven detection techniques with novel solid-state photo-detector devices called silicon photomultipliers (SiPMs). SiPMs are multi-pixel photo-diodes operating in the limited Geiger mode. Their high gain (106), small size and low operating voltages ( neutral hadron energy mapping, with particle type and energy distributions from a physics process. Results are shown in fig. 9.
The calibrations could be improved, most notably by correcting for depth of interaction (losses beyond the HCal) and using a neutral hadron energy distribution from a 500 GeV cm physics process for the inversion. The weighted neutral hadron energy distributions in Z-pole events and 500 GeV ZZ events are shown in fig. 10.
[pic]
Fig. 9 – Effective resolution of neutral hadrons in Zpole events
[pic]
Fig. 10 – Fraction of neutral hadron energy vs energy in Zpole and 500 GeV ZZ events
4. Perfect pattern recognition
The calibrations were used to analyze full detector simulations with perfect pattern recognition, i.e. all calorimeter hits are correctly associated with the final state particle from which they originate. Energy sums in mono-energetic jets were studied in 2 light quark (uds) events at 91, 200, 500 GeV cm energy. Results are shown in fig. 11. For comparisons, the quantity alpha = (rms90/mean90)*sqrtE is shown, where 90 means the 90% of the events with the smallest rms were used.
The transverse spreads were also studied in these events. A cylinder was defined through the HCal for each neutral hadron. For a given radius of the cylinder, the number of hits from the neutral hadron inside the cylinder divided by the total number of hits in the HCal from the neutral hadron (summed over all neutral hadrons) was used to define an efficiency. Similarly, the number of hits inside the cylinder from charged hadrons over the number of hits inside the cylinder (again summed over all neutral hadrons) defined a purity. The results are shown in fig. 12 and13.
Also studied were full detector simulations of ZZ events at 500 GeV, where one Z->nu nu, and the other to 2 light quarks. Using perfect pattern recognition for both the neutral hadrons and the photons, and requiring 2 jets from the jet finder, both the dijet mass and delta (dijet mass – generated Z mass) are shown in fig. 14 and 15. The direction of both quarks was required to have abs(cos(theta)) < 0.8 to minimize losses down the beampipe. The effective resolution for neutral hadrons in these events is shown in fig. 16. For a significant difference in neutral hadron resolution, little or no difference is seen in the width of the dijet mass distribution.
The same events were analyzed with a different level of cheating. Instead of perfectly associating the hits with the correct particle, first all the hits in the calorimeters were clustered using a weighted density clustering algorithm. (Directed Tree ). The clusters were then associated with the particle contributing the most hits to the cluster. Dijet mass and delta mass are shown in fig. 17and 18.
[pic]
Fig. 11 – Energy sums in uds events at 91, 200, 500 GeV
[pic]
Fig. 12 – Efficiency/Purity curves for neutral hadrons using a fixed radius cylinder at 91 GeV
[pic]
Fig. 13 – Efficiency/Purity curves for neutral hadrons using a fixed radius cylinder at 200,500 GeV
[pic]
Fig. 14 – Dijet mass distribution using “perfect” pattern recognition
[pic]
Fig. 15 – Perfect pattern recognition dijet mass – generated Z mass
[pic]
Fig. 16 – Effective resolution of neutral hadrons in 500 GeV ZZ events
[pic]
Fig. 17 – Dijet mass cheating on cluster association
[pic]
Fig. 18 – Cluster association cheating dijet mass – generated Z mass
5. Summary
Simulations of single neutral hadrons in isolated detectors have been studied varying several design parameters, yielding an overall picture of tradeoffs in resolution and shower spread. With a limited set of parameters, the effect of intrinsic HCal resolution on jet energy and dijet mass resolution has been quantified. With both perfect pattern recognition and real clustering, no significant gain in dijet mass resolution is seen with better intrinsic neutral hadron resolution. Attempts to quantify the effect of shower spread on confusion in PFAs have been made, but the algorithms themselves will probably be needed.
VI. Use of PFAs in SiD Development
1. Introduction
The SiD detector is designed to take advantage of jet reconstruction using the Particle Flow Algorithm (PFA) approach. A PFA attempts to reconstruct all of the individual particles in an event according to their detector signatures – tracks for charged particles and calorimeter clusters for photons and neutral hadrons. Separation and identification of calorimeter energy deposits is, therefore, crucial for the PFA approach to work. Charged hadrons normally comprise ~65% of the typical jet energy, photons ~25%, and neutral hadrons the remaining ~10%. By separating the calorimeter clusters made by these particles, better jet energy resolution and also dijet mass resolution can be obtained by optimal use of detector components – tracking detectors for charged particles, the ECAL for photons, and the ECAL and HCAL for neutral hadrons. A PFA can be used to optimize choices for detector components – mainly the ECAL and HCAL. The SiD will take advantage of the PFA approach to optimize many of the detector parameters, resulting in improved jet reconstruction performance over existing detector designs along with optimized cost.
2. PFAs in Physics Processes
The goals for PFA performance have been expressed in many ways. For an ILC detector based on PFA jet reconstruction, the goal is to be able to separate on an event-by-event basis W and Z bosons using all decay modes including the dijet mode. For no loss in luminosity, this means that the W and Z masses must be measured with a resolution of ~3 GeV, or to ~3-4%. Typical calorimeters designed without PFA optimization cannot achieve this goal – even if they are perfectly compensating. Since the performance goal depends on (QCD) jet reconstruction, the optimization of the detector will ultimately depend on jets in simulation.
The PFA approach for the SiD detector has been studied with this goal in mind. The dijet mass resolution performance of the SiD has been studied as a function of several variables – notably the solenoidal B-field, the inner radius of the ECAL, jet energy, and jet multiplicity. Our approach has been to start with the simulated process e+e- -> ZZ where one Z -> 2 jets and the other Z -> nunu. This process produces jets of energy ~120 GeV – the typical jet energy of many 4-jet processes at e+e- annihilation at 500 GeV. The dijet mass can be determined without any jet combinatoric ambiguity for these events. Then, the all-hadronic decay mode is studied, where ZZ -> 4 jets. Thus, the detector is filled with more 120 GeV jets. Another test is e+e- -> ttbar – 6 jets with reduced jet energies. Finally, we can test the limitations of PFAs with e+e- -> qqbar at 500 GeV. With this analysis, we can investigate the jet energy dependence of the dijet mass resolution. As the jet energy increases, confusion in the PFA eventually dominates the mass resolution, degrading the PFA performance.
3. Tools for PFA Development
To develop the PFA, several tools are needed which are common to all detector models and PFA implementations. A common basis for PFA applications is required – in particular, a common standard detector calibration is needed along with a standard Perfect PFA definition in order for useful comparisons to be made between PFA algorithms and detector models.
Calorimeter Calibration
A method of calorimeter calibration has been developed which could be used in a real test beam for a detector prototype. The PFA approach results in separated charged hadrons, photons, and neutral hadrons. Therefore, a full calibration should include beams of these species if possible. Neutral hadron beams are especially useful. In simulation, pions (+-), photons, and a mixture of n, nbar, and Klong are used to calibrate the ECAL and HCAL. The calibration method gives corrected energies for cheated (all hits) clusters. Additional corrections are needed for the application of real cluster algorithms, analysis techniques, etc. A standard calibration method exists and SiD variants all use this method to provide standard calibrations for detector components.
Perfect PFA Definition
For PFA analysis, it is important to have a standard definition of the perfect application of the PFA – no confusion in the identification of particle-cluster associations. It is especially important for the useful cases in which perfect definitions are used – for example, when perfect charged particle tracks are used to test the PFA performance in the calorimeter only. When used with the standard detector calibration procedure, resolutions of individual particle energy distributions match those of the particles used in the calibration method. The Perfect PFA provides a check not only of the optimal PFA performance, but also a check of the calorimeter calibration method.
Cluster Algorithms
Many cluster algorithms have been developed for use in PFAs. More importantly, most of these have standardized inputs and outputs, so easy evaluation and comparison of multiple cluster algorithms can be done. Procedures exist to illustrate purity and efficiency of cluster algorithms, so that optimal choice can be made of algorithms to use in various stages of the PFA. For example, when choosing a cluster algorithm for track-cluster matching, overall purity performance is preferred over efficiency, while for photons, a more balanced performance is desired.
4. PFA Approaches for SiD
Several PFA approaches have been investigated for the ILC detector concepts. These can be divided into basically 2 types – a modular approach designed to perform particle identification and separation with methods optimized for each particle species, and cluster algorithm approaches where a specific cluster algorithm is used to define all particle clusters in the calorimeter.
One approach to PFA development is to use an optimized single cluster algorithm to try to form clusters of all particles in the event. These cluster algorithms must be able to reconstruct clusters of many varied types – mips, electromagnetic showers (photons and electrons), and hadronic showers (both neutral and charged). Cluster algorithms based on hit density, fragment distances, and shower shapes have been developed for this type of PFA.
Another approach is to develop individual algorithms optimized for the particular particle being tested. This type of PFA has a natural modular structure with each step consisting of a cluster algorithm and topological evaluation algorithms optimized for a particular particle species. The modular structure allows easy comparisons to be made at each stage and for fast identification of PFA performance.
5. PFA Results
The following plots show results for application of a PFA on single particle species. The ultimate performance limit of the PFA can be illustrated for each type of particle in this way and estimates of the minimum confusion introduced by the PFA can be made. Plots for charged hadrons (pions), photons, and neutral hadrons here.
Single Particle PFA Response
Charged Hadron/Cluster Association
Photon Algorithm Performance
Neutral Hadron Algorithms
Close, but not ready yet – will have by Monday
Using plots such as these along with other related analyses, the combination of PFA cluster algorithms and analysis methods can be optimized to reconstruct the individual particles which ultimately are used as inputs to jet algorithms.
Initially, PFAs were developed using simulated e+e- -> Z production at the Z pole (91.5 GeV). No jet algorithms were needed and the PFA performance could be shown as a simple energy sum of resulting reconstructed particles. The following plot is an example of PFA results on the particle energy sum at the Z pole for the SiD model. In this case, the HCAL is SS/RPC.
[pic]
Figure xx. PFA results for energy sum at the Z pole. The RMS90 performance measure corresponds to 3.6 GeV, or ~4% when compared to the Z mass.
Moving on from Z pole results, the dijet mass resolution is the most important measure of PFA performance for the e+e- collider at 500 GeV CM. For the process e+e- -> ZZ -> qqbar((, differences between the dijet mass and the initial qqbar mass are shown in the following plots for 4 different SiD variants with different HCAL technologies.
[pic]
Figure xx. Dijet mass residuals for SiD with a) W/Scintillator HCAL, b) SS/Scintillator HCAL, c) W/RPC HCAL, and d) SS/RPC HCAL.
The results show that a single PFA operating on various models can give consistent results. Further studies using the PFA template will be able to separate the PFA confusion contributions from the detector model effects.
The dijet mass resolution is not constant for jets of varying energies in a real PFA. When plotted as a function of jet energy, the increasing contribution of confusion in the real PFA is seen.
Dijet results vs jet energy will show up here next week
6. Future SiD Optimization with PFAs
Based on the performance obtained so far, PFAs will be used to optimize several parameters of the SiD detector, helping to choose these parameters for the final SiD design. In particular, the PFA performance can be used to evaluate various choices of inner radius (distance from IP to inner face of ECAL), and B-field (central solenoidal field strength). Along with the cost parameters, the PFA can then be used to optimize the size of the SiD detector. Also, various technologies for an optimized HCAL can be tested with PFA applications. For example, the PFA performance of analog scintillator designs can be compared to digital gas calorimeters. Shown here is a first study of the effect of varying the central solenoidal B-field for a SiD model with Z pole events.
[pic]
Figure xx. Results of PFA performance at Z pole with different B-fields.
In this illustration, both the number of events in the central peak and the resolution of that peak are improved with higher field.
In the future, more detailed studies will be done various detector parameters, optimizing them for final inclusion in the SiD model.
-----------------------
Figure II-9. Prototype (version 1) silicon sensor.in the lab.
Figure II-6. Sensor cross section. The top layer is a signal trace metalization. Dimensions are approximate.
Figure II-5. Sensor schematic showing placement of the KPiX ASIC (center) and a representative signal trace.
Figure II-2. A simulated [pic] decay in SiD. Note that the graphical squares in the Si-W ECal are larger than the detector pixels. The individual measurement of the SiD silicon tracker are visible on the charged pion track.
Figure H1 : Schematic diagram of a typical Resistive Plate Chamber
Resistive paint
Resistive paint
Mylar
1.2mm gas gap
Mylar
Aluminum foil
1.1mm glass
1.1mm glass
-HV
Signal pads
Analog
Digital (0.5x0.5)
Digital (1.4x1.4)
Digital (2.5x2.5)
Digital (3.0x3.0)
E [te Chamber
Resistive paint
Resistive paint
Mylar
1.2mm gas gap
Mylar
Aluminum foil
1.1mm glass
1.1mm glass
-HV
Signal pads
Analog
Digital (0.5x0.5)
Digital (1.4x1.4)
Digital (2.5x2.5)
Digital (3.0x3.0)
E [GeV]
σ/E
Figure H2 : Single particle resolution versus particle energy for different readout segmentations and resolutions. (Plot by V.Zutshi, NIU)
Figure H8. (a) A schematic diagram of 30cmx30cm double GEM chamber (b) 30x30 anode board.
4 T B-field
3.78 GeV 89.2 GeV 54% in central fit
5 T B-field
3.63 GeV 89.3 GeV 63% in central fit
b)
a)
c)
d)
Figure II-8. KPiX-2 in the lab.
Figure II-11. CV curve for a pixel in region a.
Figure II-10. Specific signal traces in region a.
Figure II-13. Measured position of peak versus total pixel capacitance.
Figure II-12. Signals from Am241 60 keV photons observed in a typical Hamamatsu silicon sensor pixel.
................
................
In order to avoid copyright disputes, this page is only a partial summary.
To fulfill the demand for quickly locating and searching documents.
It is intelligent file search solution for home and business.