The purpose of the DEA cut is to veto the events that have photons that miss the LGD and the CSI.
As can be easily seen from a drawing of the E852 experimental setup, ideally, CSI, DEA and LGD give E852 an almost hermetic photon detection capability.
The photons either hit the LGD and are measured, or hit the CSI and cause the event to fail the CSI cut or hit the DEA and cause the event to fail the DEA cut. Ideally, after the DEA and CSI cuts, the final data sample will not contain events with unmeasured photons.
Many different approaches can be used to veto the events that have photons hitting the DEA.
The simplest DEA cut would veto events that have hits in any of the DEA segments. It's drawback is that it will also veto the events where charged patricles go through the DEA, leading to an excessive loss of events.
A more sneaky cut would try to recover some of these events by looking at hits in CPVC. In this approach it is assumed that a charged track will fire both a CPVC segment and the corresponding DEA segment. The cut would ignore all the DEA segments where the corresponding CPVC segment has a hit (assuming it was a charged particle) and only would only look at the DEA segments where the corresponding CPVC segment did not fire. If any of such DEA segments has a hit, the event is vetoed.
This cut was fairly successfully used to analyze the 1994 data. However a couple of problems were discovered- it turned out that there are "holes" between CPVC and DEA, so that charged particles manage to hit DEA segments without hitting the corresponding CPVC segments. Such events foul the cut and get erroneously vetoed, creating a "picture frame" effect on the CPVC Z-plane. The other problem was discovered while processing the 1995 data. It turned out that the bottom CPVC segment (channels 35 and 36) did not work very well. The ADC signals are not large enough and create a problem in telling real CPVC hits from pedestal fluctuations. For more details, see the report on the CPVC and DEA performance during the 1995 data run.
To solve these two problems, a more advanced cut was implemented. Instead of using CPVC, it relies on the reconstructed charged tracks to determine which segments of the DEA are hit by charged particles.
In addition to fixing (avoiding) the problems with CPVC, it also helps to veto events with missing charged particles. If the reconstruction program failed to find a charged track, and this track makes a hit in the DEA, the event will be vetoed by the DEA cut, unless the same segment is hit by another, correctly reconstructed track.
The details of the implementation and performance of this tracking-based DEA cut is discussed in the rest of this paper.