J.M. Moss
Los Alamos National
Laboratory, Los Alamos, NM 87545
(phenix-muon-95-7; submitted 27 April 1995)
As we near the construction phase of both muon arms it is important to know
whether the muon identification requirements of the south arm, justified as an
upgrade for high-luminosity pp collisions, are different from those of
the north arm. Costs of the south arm alone are now approaching the $10M level.
If there is any chance to save enough money to instrument muon ID stations for
one or two of the central magnet arms, it seems likely that some of the savings
must come from the endcap muID.
I have made a comparison of the event rates from Drell-Yan (DY) production to
those from random coincidences in the region of pair masses greater than 2 GeV.
Both kinds of events were run in PISA using the south muon arm as currently
configured, 12-degree to 35-degree, with a shortened muon endcap, and no Pb or
neutron shield. Random events were generated with the UA1 pion generator,
restored to the original UA1 parameterization (for sqrt-s = 200 GeV collisions). The event multiplicity was
taken to be 14 charged pions/(pp event). These can be scaled to yield
results for ersatz AuAu collisions at the double HIJET value of 10^4/collision.
The UA1 events were run with a high threshold, 2 GeV and 4 GeV, in order to be
able to throw several million events before I retire. Pat McGaughey is
currently investigating background and multiplicity issues with a complementary
low-threshold configuration.
Momentum and mass spectra were generated from PISA ntuples using a very simple
output analysis. The UA1 events which reach the muon ID consist largely of
muons from pion decay before the nosecone of the south arm. A potentially
important background source is pions which do not shower in the nosecone,
central magnet pole, or muon arm backplate. Figure 1 shows the number of muons
and pions at two locations, at muID plane 3, and at muID plane 6, using the UA1
generator. The momentum is the "reconstructed" value; for the cognoscenti, that
means the momentum the particle has in ntuple 1010 prior to reaching the
nosecone (the momentum a perfect tracker would give). The left panels, with the
muID condition of a hit in plane 3, were generated from a 2 x 10^6 event data
set with p_min = 2 GeV, the right panels, with the muID condition of a
hit in plane 6,were generated from a 4 x 10^6 event data set with p_min
= 4 GeV. It is clear that decay muons are much more numerous than the
non-showering pions, but that the latter extend to relatively higher momenta,
as expected. This result is certainly familiar from the PHENIX CDR and from the
June 1993 TAC review document.
Figure 1: Muons from pion decay reaching station 3 of the muon identifier, top
left; non-showering pions which create a signal in plane 3 of the muon
identifier, bottom left; Muons from pion decay reaching station 6 of the muon
identifier, top right; nonshowering pions which create a signal in plane 6 of
the muon identifier, bottom right.
Pair mass spectra are readily created from the singles spectra by event mixing
with an assumed operating luminosity. We take this to be the enhanced
luminosity of polarized pp collision (RSC proposal value), L = 8
x 10^31 cm^-2 sec^-1. This corresponds to a pp collision rate of
r_pp=3.2 x 10^6/sec. The random rate is then F = N_iN_j, where i, j = , u and = 110 ns.
The singles rates from Fig. 1 (left panel) provide a good indication of the
level one trigger rates from single pions and muons. To penetrate to muID plane
3 particles require ~ 2 GeV. The ~ 1300 events of Fig. 2 (left panels)
translates to ~ 2000 /sec at L = 8 x 10^38 cm^-2, or 2 x 10^-4 per
bunch crossing.
The rate of DY events is easily calculated and run through PISA with the same
ntuple analysis as for the random events. Our calculations used the leading
order DY equation with a K-factor of 2 and the Duke S1.1 (leading order)
structure functions from the CERN PDF library. This gives excellent agreement
with the newly published absolute cross sections for DY production at
sqrt-s = 38.7 GeV from E772. The integrated cross section for > 2 (4) GeV at sqrt-s = 200 GeV
is 6.5 (1.5) nb, giving agreement with Table 3.3 of the PHENIX CDR. Figure 2
displays the results in terms of the absolute number of events recorded with an
integrated luminosity of 8 x 10^38 cm^-2 (10^7 sec of running time). For
simplicity of display, the random spectrum from a muon and a pion in
coincidence has been omitted.
It is clear, even at the assumed very high luminosity, that there is no
serious competition to the DY rate from pion or decay backgrounds. Further,
comparison of the the two panels of Fig. 2 shows that most of the pions have
showered by the time plane 6 is reached. Thus the requirement of a hit in plane
6 is sufficient to guarantee dominance of the background by decay muons.
Figure 2: Mass spectra from random coincidences of muons from pion decays and
nonshower-ing pions compared to Drell Yan events. The left frame spectra are
generated with the requirement of a signal in muID plane 3, while at the right
the requirement is muID plane 6. The solid, dashed, and dotted lines are
respectively, random muons, random pions, and DY events.
Figure 3 shows that there is no loss in efficiency for DY masses greater than
3.5 GeV from the requirement that the muons penetrate to the back of the
identifier. Recall that our muon identification algorithm is the simplest of
all imaginable --- penetration to plane x.
Figure 3: Comparison of DY spectra requiring signals in the 3rd (solid) and 6th
(dashed) muID planes.
Figure 4: Same as Fig. 2 but with random coincidences scaled for central AuAu
collisions.
The previous results may be extended to central AuAu collisions using
approximate scaling laws. For the DY process one has, . Happily the PHENIX/CDR luminosity of
AuAu collisions, 2 x 10^27 gives 7.7 x 10^31 cm^-2 sec^-1 for DY events, thus
requiring no correction for comparison to pp. Soft particle production
in central collisions scales as . While
pions in the region p > 2 GeV may have a nuclear dependence closer to
unity, we will assume 2/3 for the present case. This gives , so that the signal-background gets worse
by with respect to DY. Figure 4 shows the
spectra of Fig. 2 with the random backgrounds multiplied by . This is a somewhat more optimistic signal to background
than Fig. 3.6 of the PHENIX-CDR where the decay and DY signals are shown
crossing at about 5 GeV.
Taken at face value these calculations indicate for pp collisions that:
- There are no serious backgrounds toDY events above = 4 GeV from random pions or random muons
from pion decay,[1]
- The muon identifier provides ample hadron rejection without shower
identification,
- Fewer that 6 planes of depth segmentation could be tolerated in the
south arm muon ID system.
Before the above statements are taken as
gospel, high threshold PISA runs must be combined with the typical multiplicity
distributions from very low threshold runs. Additionally, one needs to look
into the background from beam gas interactions. My guess is that these further
refinements won't change the outcome much. They will, however, allow one to
define the required channel count in the muID for the south arm.
I would like to add a few more items to the list which require further scrutiny
for the south arm muID system. These are:
- Given the vastly simpler pattern recognition requirements for pp
compared to central AuAu collisions (multiplicities down by ~ , does one need pad readout?
- Could the r- road
finding of the PHENIX/CDR-Update be replaced with an x-y scheme?
- How much money is to be saved by designing the muID for pp
collisions? How well would it work for central AuAu?
These are some of
the issues which need to be on the table if we are to make intelligent use of
the funds for the spin upgrade.
Click here for a postscript version
Send mail to the author: Joel Moss, Los Alamos