1. Introduction

In the summer of 2005, there has been a proposal to install a shield in the upstream area adjacent to the STAR detector, roughly between DX and D0-Q3.

The assortment of notes, pictures and other documents can be found at:
http://starmac.lns.mit.edu/~rhicspin/shielding/.

2. Rationale and Proposal for the Shielding

2a. Rationale

Excesseive rate of triggers in the EMC was the prinary motivation to study the background. A set of background event were sudied by Josh Vredevoogd and Ted Hopkins using the Event Display, and the results can be found in:
http://www.physics.valpo.edu/studentResearch05/STARe-h/background.html

The study was not too technically advanced, but in general the angular distribution of the "stray" tracks in the TPC indicated that they come from the upstream area of the tunnel. It was presumed that they are correlated with the extraneous hits in the calorimeter. Hence, the idea was basically to shield the STAR detector from the tunnel and it's contents.

2a. Proposed Design

The proposed design calls for an installation of the shielding assembly between DX and D0. The shileding assembly consists of a concrete base resting on the floor of the tunnel, and layers of iron forming a wall that looks approximately like letter "L" in side view. In various versions of the proposal, the thickness of the iron is between 40" and 54", i.e. roughly 100-130 cm.

.

3. Simulating the Background and Effects of the Shielding

3a. C-AD simulations done to August'05

Kin Yip of C-AD performed a simplified standalone MARS-based simulation of the shielding assembly. The source of the background was a 100GeV proton experiencing a 1 mrad deflection inside Q3, which leads to a cascade process in the magnets iron core, with subsequent cascades elswhere. The results so far were quantified in a way that made it hard to interpert in the STAR event context, although it was indeed demonstrated that the shield will lead to background suppression. More work will follow.

3b. Initial setup of the STAR GEANT simulations related to shielding and background

It is imperative to undestand the effect of the background and it's variation due to the shielding, in a way that allows interpretation in terms of the observables we measure. The best way to achieve this is to run a full fledged GEANT simulation, integrated with the rest of the STAR software framework.

We have created a prototype layout of the magnets and shielding in the upstream area, This includes the shield itself, as well as D0, Q1, Q2 and Q3. Currently, we only model only the iron in the magnets as this is represents most of the materials there. We have also added a concrete floor in the geometrical model which was missing so far.

All this additions are contained in separate source files within the pams directory structure, and are activated only in the development geometry tag, DEV2005. A rough sketch of the top view is presented here:

For the initial survey, we used the same background source model as Kin Yip, which is a 1 mrad deflected 100 GeV proton hitting the magnet wall in Q3. Below are a few of the event displays generated in such simulation:

• Top view of 3 proton events
• Two events viewed from upstream
• A hypothetical 500 GeV event, view from top

3c. Energy and particle flux in the calorimeters

Executive summary: within the limits of the current crude geometry model, and lack of the magnetic field description for the upstream magnets, the following conclusions can be reached from the survey GEANT simulation runs:

• we can expect that the shielding constructed to the currently proposed specs will cut the number of background hits by roughly a half in both the barrel and endcap parts of the calorimeter
• total number of background tracks doesn't change by much when the shield is introduced, although the composition of the background does change showing an increased fraction of neutrons
• the mean energy per hit will remain roughly the same in the endcap but will be cut by about 50% in the barrel
• we were unable to reproduce a pronounced azimuthal assymetry of the background, pointing to the fact that the configuration of the source (i.e. loci of the beam loss) needs a better approximation

We have conducted two survey runs, with and without the shield. Immediately below, we present the numbers characterizing the number of hits (in absolute an relative terms) in the endcap and barrel calorimeter. The relative (i.e. per event numbers) are quoted in percent purely to improve readability. There was no energy cut on the hits, in these two tables.

No-shileding run, 6500 events - Hit Count Matrix:

Shileding run, 3500 events - Hit Count Matrix:

In the following, we plot hit and energy distribution in the barreal and endcap. The no-shield data is presented on the left and the shield-in data on the right.

Barrel Hits (no-shield vs shield)

Endcap Hits (no-shield vs shield)

Energy Per Hit in both Endcap and Barrel (no-shield vs shield)

Mean and RMS Energy Per Hit (no-shield vs shield)

Graphs for the above table: (no-shield vs shield)

Beam Background Particle Composition, per thousand background events

note: the following is based on 1,000 event runs. Only those tracks that left hits in the calorimeters were registered

3d. Action points

A service taks has been defined to address the background study issue.

• improve the quality of the geometrical model of the upstream area (with participation from Gerrit and others)
• model the magnetic field
• conduct simulation using a variety of assumptions about the location, nature and energy of the background event
• run full reco on the simulationed background event sample
• determine the efficacy of the shileding in terms of the observables, optimize shielding (?)