The TOF System for BNL-AGS E896
K. Kainz, W.J. Llope, C. Stokely
T.W. Bonner Nuclear Laboratory
Rice University
Houston, TX 77005-1892
This page was last updated on May 16, 1996...
If you have any questions concerning what's shown here, just click on
a name above.
Introduction
Go to the top of this document...
Go to the Introduction...
Go to the Figure list...
Go to the Rates Table...
This is a presentation of results from detailed simulations that lead to our
favorite definition of a TOF system for E896. Such an array will be used for
the following reasons, which are listed more or less in order of importance.
- to aid the definition of tracks from Distributed Drift Chamber (DDC) space-points by providing the positions of particle hits after a long flight path.
- to provide direct Particle Identification (PID) information on the basis of the Time-of-Flight (TOF) for tracks for which the momentum and path length is known via the DDC information.
- to provide fast information to E896 for new triggers. Given the fact that E896 has a strong sweeping field, charged particle hits in particular regions of the E896 TOF system are predominantly due to particles that are the daughters of unstable neutral particles such as the H0, Lambda, Antilambda, and K-short.
In general, the system is composed of sub-units that are each mobile, which
allows relatively easy revisions to the configuration of the TOF System
should such changes become necessary. Nonetheless, the primary goal of this
study is to determine the geometry of a TOF System that can satisfy the three
goals above for a number of different physics analyses simultaneously. The goal is to go after tracking information, time-of-flight information, and trigger information for H dibaryons (ctau >= 4cm), Lambda and Antilambda hyperons, and K-short mesons, all at the same time.
The present simulations were all performed with the most recent version of the E896Geant. Very few revisions to this code were necessary during the course of this study. We have taken considerable care to stay current with the latest (and presumably most realistic) definitions of the geometry of the other parts of the E896 experiment and its magnetic field maps.
Simulations of complete events are the most realistic. In Geant, every
particle in the final state is stepped through the apparatus, and along the
way all of the expected secondary particles are produced via the various background processes. However, the simulation of full events is also very CPU-intensive. Specific questions can often be addressed using events obtained from realistic Monte Carlo approaches. These events contain fewer particles, e.g. just the
lambdas, but these have multiplicity and kinematic distributions that are consistent with the complete events. This kind of event generation is useful where high statistics are needed, for example to study the efficiencies of cuts on the information from the TOF system that are used for the measurement of particular (rare) particles.
Here, the event generation was done in three different ways. Each Figure below is labelled with the type of events that were used to produce it.
- Full Au+Au events were obtained from Hijet. The event files studied include 1000 b=0 events w/out the fragment afterburner, minimum bias events with the fragment afterburner, and 0.le.b.le.4 fm events with the fragment afterburner.
We intend to also study RQMD events to get a feel for the model dependence of the simulations results.
- Monte Carlo events (MC) include only a particular "interesting" parent. The multiplicity of these parents per event is sampled from probability distributions obtained from the complete events. All of the particles are emitted from the target. For each particle, the rapidity and transverse momenta are sampled from the two-dimensional (y,Pt) distributions that are also obtained from complete events. Any correlations between the particles in each event is washed out. This approach is most useful to maximize the geometrical efficiencies for measuring the daughters in the TOF system, and to understand the efficiencies of cuts on the TOF system's information to positively identify these daughters.
- Fiducial events (Fid) contain only one particle per event, and this particle is either an H0, a Lambda, an Antilambda, or a K-short (which are all uncharged). A (y,Pt) for each particle is sampled from distributions obtained from the complete events. This particle is allowed to decay with its proper lifetime. The momentum vector of the particle, the sampled lifetime, and the relative positions of the target and the fiducial volume of the DDC are used to determine whether this particle decays inside this fiducial volume. If so, the three-vector momentum and three-vector decay vertex are saved. The overall probability for these fiducial decays is also saved to allow the scaling of the numbers of these events studied to the equivalent number of complete events. The single particle fiducial events are then read into Geant, in which the lifetime of the parent under study is made arbitrarily small. These events are worth studying as the presence of the decay vertex in the fiducial volume of the DDC should make for the cleanest possible measurements of the parent. However, the requirement that the decay vertex occurs inside the DDC hardens the momentum spectra of the parents compared to the MC or complete events, which changes the performance of the TOF system.
We will show the results for a certain progression of cuts on the simulated data. These cuts are placed on the information from the DDC and the TOF system to produce samples of particular parent particles (H0, Lambda, Antilambda, and K-short).
CUT 1
This cut is performed inside the E896Geant code. On the occurence of a hit in the TOF System, all of the particle information is saved to the "TOF N-Tuple". This cut therefore imposes the geometrical efficiency of the TOF system for measuring a particular particle (daughter or otherwise).
CUT 2
This cut is a minimal requirement on the "trackability" of the particles that passed Cut 1. The track associated with the hit in the TOF wall is required to pass through at least 10cm of the fiducial volume of the DDC. For primary particles or daughters that pass this cut, it is implicitly assumed that the momentum and path length of the particle can be obtained from the DDC (each with a resolution that can be adjusted). The fact that the momentum, momentum resolution, path length, and path length resolution for particular tracks are all related in a generally complicated way is clearly ignored here. However, detailed studies of these effects are of lesser import for determining the optimal geometry of the TOF system, so these questions are postponed and appropriate but uncorrelated numbers for the various DDC-related resolutions are employed (See Figures 4 and 5).
CUT 3
This cut requires that both siblings pass Cuts 1 and 2. It is at this level where the simulation results are specific to the measurement of a particular "interesting" parent particle.
CUT 4
This cut requires that the daughter momentum is less than the maximum momentum for which the daughter can be positively identified on the basis of its time-of-flight. These cut-off momenta were located using complete events and canonical numbers for the DDC momentum resolution (dP/P in the range 1-2%) and the TOF resolution (in the range 80-200 ps).
It should be stressed that the Cut 4 events correspond to the "full cuts" that result in direct PID of both daughters from the decay of a particular parent. It is clearly true that some fraction of the tracks that fail Cut 4 are recoverable by using kinematic correlations, i.e. by cutting on Armenteros plots and/or pair invariant masses. We intend to pursue this in the near future, as these approaches will increase the rates of measurable parents above those obtained from the full cuts used herein.
We studied a number of different geometrical configurations of the TOF system. The specific restrictions on the possible geometries are as follows.
- We have to leave free a few meters downstream of the analyzer for the DDC mount.
- The MUFFINs plastic is already being cut for the smaller disk size. This more or less freezes the position of the MUFFINs in the beam-line relative to the analyzer. Salvo has agreed that any possible future changes in the MUFFINs positioning will be done only after consultation with the TOF Group.
- There are potential obstructions in the beam area, e.g. the mechanical structure of EVA.
- The path of the Au beam must be kept in mind. During the actual experiment, we will have to remove any slats that are struck at high rates by uninteracting Au beams. Otherwise, the high rate Au ions in the beam-line produce copious secondaries in the the experiment. Also, the scintillators that are struck would
soon be damaged.
- The resources of the E896 trigger and DAQ are not infinite, so the total channel count must be limited to values that can reasonably be handled by these systems. Also, we intend to borrow as many of the components for the TOF system as possible, which will also limit the total channel count.
At the three previous collaboration meetings, we have shown simulations results for several different configurations of an E896 TOF system. Since the last collaboration meeting, McGill University officially joined the experiment, and in so doing made available their beautiful TOF wall that was once part of E877. This is a high-granularity double-ended plastic scintillator slat system that has been shown to provide a TOF resolution on the order of 80 ps. With the addition of this array to the experiment, our focus then shifted to the study of the optimal positioning of this array in E896, and the possible addition of one or two more scintillator walls to improve the efficiency of the E896 TOF system as a whole. The McGill Array alone is not sufficient.
We compared the performance of three different configurations called Options A, B, and C. These configurations are sketched in the following figure.
For each configuration separately, the geometrical efficiencies, the multiple-hit probabilities, the backgrounds, and the efficiencies of direct PID on the basis of the TOF, were compared.
To make a long story short, the basic problems with Options A, B, and C are as follows. At the previous collaboration meetings, we showed that the product of the geometrical efficiency and the TOF efficiency is relatively flat versus the length of the flight path for particular parents. This makes the product of these efficiencies for the Options A, B, and C rather similar. However, Option A has a noticably poorer geometrical efficiency for the low momentum daughters (the pi- from measurable H0's, the pi- from measurable Lambdas, and the pi+ from measurable Antilambdas), and a poor TOF efficiency in general. The DSW in Option A would have to be very granular (well exceeding any realistic number of read-out channels)
because of the multiple-hit probabilities. The DSW in Option C is rather large, which has several unattractive implications. Furthermore, and most importantly for Options B and C, the DSW in these two options gets pounded by secondaries produced in the MUFFINs.
Our simulations implied that a fourth configuration, called A', would easily outperform the A, B, and C configurations. For the remainder of this write-up, we will concentrate on variety of results on the performance of the A' wall. The A' wall looks like this:
Finally, we note that the A' wall does not fit! There is an immovable vertical beam that is part of EVA in the way. The most adiabatic change from the A' geometry is shown below, this is unimaginatively called Option A''. We expect absolutely negigible differences between the performance of the A' and the A'' options. It is this A'' geometry that we propose for the "Day-1" configuration of the E896 TOF system. We are still looking into several different options for evading the
EVA obstruction, but we consider all of these to be "epsilon" changes from the A'
option.
Figures
Go to the top of this document...
Go to the Introduction...
Go to the Figure list...
Go to the Rates Table...
Figure 1
In the upper frames, the probabilities for two or more hits per slat versus the local-X distribution. Complete central Au+Au events were used, and all secondaries produced in interactions downstream of the target are included. Note the bin widths in the upper two frames are the slat size. In the lower two frames, the local-X distributions as shown for the daughters of all four interesting parents. Note the bin widths are not the slat size. The thick vertical lines in the lower frames indicates the local-X position of the uninteracting Au beam (see the following figure)
Figure 2
The local-X distributions of 197-Au ions on the E896-TOF wall, including the beam-spot size, the angular divergance of the beam, and beam momentum smearing. Note that the bin-width is not the slat size.
Figure 3
The local-X distributions on the E896 TOF wall for various primaries and secondaries in
b=0 Hijet events.
Figure 4
The 1/velocity versus momentum distributions for negatively charged particle hits on the E896 TOF wall. The various frames show the results for different values of the momentum and TOF resolution.
Figure 5
The 1/velocity versus momentum distributions for positively charged particle hits on the E896 TOF wall. The various frames show the results for different values of the momentum and TOF resolution.
Figure 6
The total path length (target to TOF wall) for all interesting daughters.
Figure 7
The momentum distributions at the E896 TOF wall for the daughters of interesting parents in the MC events. The distributions for Cuts 2, 3, and 4 are shown, as is the cut-off momentum for PID via TOF obtained using a DDC momentum resolution of dp/p=1% and an 80 ps TOF resolution.
Figure 8
The momentum distributions at the E896 TOF wall for the daughters of interesting Fiducial parents. The distributions for cuts 2, 3, and 4 are shown, as is the cut-off momentum for PID via TOF obtained using a DDC momentum resolution of dp/p=1% and an 80 ps TOF resolution.
Figure 9
The local-X distributions on the E896 TOF wall for daughters passing Cuts 2, 3, and 4 in the MC events. These distributions have been normalized to give the probabilities as numbers/event that pass each level of cuts. Note that the bin-width is not the slat size.
Figure 10
The local-X distributions on the E896 TOF wall for daughters passing Cuts 2, 3, and 4 for Fiducial parents. These distributions have been normalized to give probabilities as numbers/event that pass each level of cuts. Note that the bin-width is not the slat size.
Figure 11
The parent rapidity versus the daughter local-X position on the E896 TOF wall for Fiducial events and for those daughters passing Cut 1.
Figure 12
The parent rapidity versus the daughter local-X position on the E896 TOF wall for Fiducial parents and for those daughters passing Cut 2.
Figure 13
The parent rapidity versus the daughter local-X position on the E896 TOF wall for Fiducial parents and for those daughters passing Cut 3.
Figure 14
The parent rapidity versus the daughter local-X position on the E896 TOF wall for Fiducial events and for those daughters passing Cut 4.
Figure 15
The parent (y,Pt) distributions from Fiducial parents for daughters that pass Cut 1 on the E896 TOF wall information.
Figure 16
The parent (y,Pt) distributions from Fiducial parents for daughters that pass Cut 2 on the E896 TOF wall information.
Figure 17
The parent (y,Pt) distributions from Fiducial parents for daughters that pass Cut 3 on the E896 TOF wall information.
Figure 18
The parent (y,Pt) distributions from Fiducial parents for daughters that pass Cut 4 on the E896 TOF wall information.
Figure 19
The parent rapidity distributions for Fiducial parents and for those daughters passing Cuts 2, 3, and 4 as labelled.
Figure 20
The parent rapidity distributions for MC events and for those daughters passing Cuts 2, 3, and 4 as labelled.
Figure 21
The parent (y,Pt) distributions from MC events for daughters that pass Cut 1 on the E896 TOF wall information.
Figure 22
The parent (y,Pt) distributions from MC events for daughters that pass Cut 2 on the E896 TOF wall information.
Figure 23
The parent (y,Pt) distributions from MC events for daughters that pass Cut 3 on the E896 TOF wall information.
Figure 24
The parent (y,Pt) distributions from MC events for daughters that pass Cut 4 on the E896 TOF wall information.
Figure 25
The parent rapidity versus the daughter local-X position on the E896 TOF wall for MC events and for those daughters passing Cut 1.
Figure 26
The parent rapidity versus the daughter local-X position on the E896 TOF wall for MC events and for those daughters passing Cut 2.
Figure 27
The parent rapidity versus the daughter local-X position on the E896 TOF wall for MC events and for those daughters passing Cut 3.
Figure 28
The parent rapidity versus the daughter local-X position on the E896 TOF wall for MC events and for those daughters passing Cut 4.
Figure 29
The parent transverse momenta for Fiducial parents and for those daughters passing Cuts 1, 2, 3, and 4. These distributions are not corrected for the various measurement efficiencies. The acceptance extends to Pt=0 for all parents and all levels of the cuts.
Figure 30
The parent transverse momenta for MC events for those daughters passing Cuts 1, 2, 3, and 4. These distributions are not corrected for the various measurement efficiencies. The acceptance extends to Pt=0 for all parents and all levels of the cuts.
Figure 31
The comparison of the numbers of parents that pass the full cuts from the MC events (black), and the numbers that pass the full cuts from the Fiducial events (red). The scale factors relating the Fiducial parent generation to the equivalent number of complete or MC events were obtained as described in the Introduction. The differences in the relative heights of the two histograms for each parent are related primarily to the different proper lifetimes. Of these, the K-short has the smallest lifetime, hence the largest difference in the rates between the MC events (decay vertex not required to be in the DDC) and the Fiducial events (decay vertex inside the DDC). MC events were not generated for the H0.
Figure 1
Go to the top of this document...
Go to the Introduction...
Go to the Figure list...
Go to the Rates Table...
Figure 2
Go to the top of this document...
Go to the Introduction...
Go to the Figure list...
Go to the Rates Table...
Figure 3
Go to the top of this document...
Go to the Introduction...
Go to the Figure list...
Go to the Rates Table...
Figure 4
Go to the top of this document...
Go to the Introduction...
Go to the Figure list...
Go to the Rates Table...
Figure 5
Go to the top of this document...
Go to the Introduction...
Go to the Figure list...
Go to the Rates Table...
Figure 6
Go to the top of this document...
Go to the Introduction...
Go to the Figure list...
Go to the Rates Table...
Figure 7
Go to the top of this document...
Go to the Introduction...
Go to the Figure list...
Go to the Rates Table...
Figure 8
Go to the top of this document...
Go to the Introduction...
Go to the Figure list...
Go to the Rates Table...
Figure 9
Go to the top of this document...
Go to the Introduction...
Go to the Figure list...
Go to the Rates Table...
Figure 10
Go to the top of this document...
Go to the Introduction...
Go to the Figure list...
Go to the Rates Table...
Figure 11
Go to the top of this document...
Go to the Introduction...
Go to the Figure list...
Go to the Rates Table...
Figure 12
Go to the top of this document...
Go to the Introduction...
Go to the Figure list...
Go to the Rates Table...
Figure 13
Go to the top of this document...
Go to the Introduction...
Go to the Figure list...
Go to the Rates Table...
Figure 14
Go to the top of this document...
Go to the Introduction...
Go to the Figure list...
Go to the Rates Table...
Figure 15
Go to the top of this document...
Go to the Introduction...
Go to the Figure list...
Go to the Rates Table...
Figure 16
Go to the top of this document...
Go to the Introduction...
Go to the Figure list...
Go to the Rates Table...
Figure 17
Go to the top of this document...
Go to the Introduction...
Go to the Figure list...
Go to the Rates Table...
Figure 18
Go to the top of this document...
Go to the Introduction...
Go to the Figure list...
Go to the Rates Table...
Figure 19
Go to the top of this document...
Go to the Introduction...
Go to the Figure list...
Go to the Rates Table...
Figure 20
Go to the top of this document...
Go to the Introduction...
Go to the Figure list...
Go to the Rates Table...
Figure 21
Go to the top of this document...
Go to the Introduction...
Go to the Figure list...
Go to the Rates Table...
Figure 22
Go to the top of this document...
Go to the Introduction...
Go to the Figure list...
Go to the Rates Table...
Figure 23
Go to the top of this document...
Go to the Introduction...
Go to the Figure list...
Go to the Rates Table...
Figure 24
Go to the top of this document...
Go to the Introduction...
Go to the Figure list...
Go to the Rates Table...
Figure 25
Go to the top of this document...
Go to the Introduction...
Go to the Figure list...
Go to the Rates Table...
Figure 26
Go to the top of this document...
Go to the Introduction...
Go to the Figure list...
Go to the Rates Table...
Figure 27
Go to the top of this document...
Go to the Introduction...
Go to the Figure list...
Go to the Rates Table...
Figure 28
Go to the top of this document...
Go to the Introduction...
Go to the Figure list...
Go to the Rates Table...
Figure 29
Go to the top of this document...
Go to the Introduction...
Go to the Figure list...
Go to the Rates Table...
Figure 30
Go to the top of this document...
Go to the Introduction...
Go to the Figure list...
Go to the Rates Table...
Figure 31
Go to the top of this document...
Go to the Introduction...
Go to the Figure list...
Go to the Rates Table...
Rates Table
Go to the top of this document...
Go to the Introduction...
Go to the Figure list...
Go to the Rates Table...
Notes
- These rates are the numbers/event that pass the Cut level listed. For a description of these cuts, see the Introduction above.
- The Lambda-bar rates are calculated assuming the same multiplicity/event and pt-y spectra as lambdas. One should mentally include factors on the order of a few times 10^-3 for the actual production rate of Lambda-bars as compared to Lambdas.
- The H0 rates are calculated assuming 0.1 H0 per central event.
- The rates per week are calculated assuming 250 central events per second are written to tape, and that there are 100 hours of beam per week.
Rates from the MC events |
|
L->pi- |
L->p |
Lbar->pbar |
Lbar->pi+ |
Ks->pi- |
Ks->pi+ |
|
cut 2: |
0.169 |
0.433 |
0.941 |
0.108 |
0.120 |
0.0443 |
|
cut 3: |
0.104 |
0.104 |
0.0955 |
0.0955 |
0.0119 |
0.0119 |
|
cut 4: |
0.0106 |
0.0106 |
0.0111 |
0.0111 |
0.00170 |
0.00170 |
|
|
Rates from the MC events
(per week) |
|
L->pi- |
L->p |
Lbar->pbar |
Lbar->pi+ |
Ks->pi- |
Ks->pi+ |
|
cut 2: |
1.52e7 |
3.90e7 |
8.47e7 |
9.72e6 |
1.08e7 |
3.99e6 |
|
cut 3: |
9.36e6 |
9.36e6 |
8.60e6 |
8.60e6 |
1.07e6 |
1.07e6 |
|
cut 4: |
9.54e5 |
9.54e5 |
9.99e5 |
9.99e5 |
1.53e5 |
1.53e5 |
|
|
Rates from the Fiducial events |
|
L->pi- |
L->p |
Lbar->pbar |
Lbar->pi+ |
Ks->pi- |
Ks->pi+ |
Ho->pi- |
Ho->p |
|
cut 2: |
0.0620 |
0.0969 |
0.0950 |
0.602 |
0.00161 |
0.00155 |
0.0000216 |
0.0000218 |
|
cut 3: |
0.0564 |
0.0564 |
0.0564 |
0.0541 |
0.00755 |
0.00755 |
0.00000617 |
0.00000617 |
|
cut 4: |
0.00687 |
0.00687 |
0.00578 |
0.00578 |
0.000137 |
0.000137 |
0.000000242 |
0.000000242 |
|
|
Rates from the Fiducial events
(per week) |
|
L->pi- |
L->p |
Lbar->pbar |
Lbar->pi+ |
Ks->pi- |
Ks->pi+ |
Ho->pi- |
Ho->p |
|
cut 2: |
5.58e6 |
8.72e6 |
8.55e6 |
5.42e7 |
1.45e5 |
1.40e5 |
1.94e3 |
1.96e3 |
|
cut 3: |
5.08e6 |
5.08e6 |
5.08e6 |
5.08e6 |
6.80e5 |
6.80e5 |
5.55e2 |
5.55e2 |
|
cut 4: |
6.18e5 |
6.18e5 |
5.20e5 |
5.20e5 |
1.23e4 |
1.23e4 |
2.18e1 |
2.18e1 |
|