Exclusive J/psi + photon Production



contact: Angela Wyatt 5th September 2003
acw@hep.ucl.ac.uk

Introduction


There has been a lot of interest in the possibility that, at the LHC, the Higgs boson could be produced by exclusive production and that this could be a competitive channel [1]:

p + p --> p + H + p

that is nothing else is produced at all in the interaction, except the Higgs boson (which then decays).

To test this prediction, a similar process can be searched for at the Tevatron run II:
p + pbar --> p + chi_c^0 + pbar
chi_c^0 --> J/psi + gamma, J/psi --> mu^+ mu^-

Events containing rapidity gaps, as these would, are said to be diffractive. Diffraction is often considered to be due to the exchange of a pomeron, which has vacuum quantum numbers and so its exchange can lead to rapidity gap formation.

This analysis uses 93pb-1 of di-muon triggered data to search for events containing a j/psi or j/psi+gamma and nothing else observed in any of the CDF detectors. These events are termed exclusive. The analysis consists of:
1. Selecting good di-muon pairs in the j/psi mass window: jpsi selection.
2. From these, selecting events with a forward rapidity gap on the proton and anti-proton side (double pomeron exchange [DPE]): DPE selection.
3. From the DPE sample, selecting events with nothing above noise in each detector component except for the j/psi and at most one photon candidate: exclusive selection.
4. Examining these events to see if they are compatible with exclusive chi_c^0 production: chic production.
Summary


[1] A. De Roeck, V. A. Khoze, A. D. Martin, R. Orava and M. G. Ryskin, Eur. Phys. J. C25 (2002) 391, hep-ph/0207042.

J/psi Selection


A sample of di-muon events is selected. The selected muons have opposite sign, P_t > 1.5 GeV and |eta| < ~0.6 (they are detected by the CMU). Other muon and track quality cuts are also used. The invariant mass of the di-muon is also required to be in the j/psi mass window (except for mass plots). Timing information is used to reject cosmic ray events.

The uncorrected distributions of the di-muon invariant mass, pt and rapidity of the selected di-muon sample (inclusive sample):


The plots as eps files: mass, pt, rapidity.

[top]

Double Pomeron Exchange Selection


From the inclusive j/psi sample, we now select events that are consistent with double pomeron exchange by selecting events with a rapidity gap in the region 3.5 < |eta| < 7.5 on both the proton and anti-proton side. This is done using the Beam Shower Counters (BSC), 5.5 < |eta| < 7.5 and the MiniPlug calorimeters (MP), 3.5 < |eta| < 5.1.

Since there is always some level of noise in the detectors (either electronic or beam related e.g. beam-gas collisions) we do not define a rapidity gap as a zero signal. Instead, we require that the signal measured in a detector is inconsistent with being from a real particle entering the detector and consistent with being noise. CDF is able to clearly separate these two possibilities.

An example of this separation is shown in the figure below. This shows the total energy in the MiniPlug detector on the anti-proton side, on a logarithmic scale. The energy distribution for events in which no interaction was detected in the central detector is shown by the black solid histogram. The events in which there was an interaction are shown by the red dashed histogram (non-diffractive interactions) and by the green dotted histogram (diffractive interactions). The diffractive interactions are consistent with the non-interactions i.e. in both cases the MP is "empty". These events are clearly separated from the non-diffractive interaction events. This means that although we don't cut at zero signal we are clearly able to isolate diffractive events and that the selection is not very sensitive to exactly where the cut is placed.



The plot as an eps file.

Having used information like this to select a threshold for a "hit" in the BSC or MP. We then select single diffractive (SD) events by requiring zero BSC and zero MP hits on the same side - this is done on both the proton and anti-proton side.

A clear peak at (0,0) is seen when the number of hits in the beam shower counters (BSC) is plotted against the number of hits in the MiniPlug (MP). This is shown for the anti-proton side and the proton side:


To search for double pomeron exchange events (DPE) we now look at the events that have a rapidity gap and see if there is also a rapidity gap on the other side. The figures below plot the number of BSC hits against the number of MP hits on the anti-proton (proton) side when there is a rapidity gap on the proton (anti-proton) side. A clear peak at (0,0,) is seen:


The plots as eps files: SD signal anti-p side, SD signal p side, DPE signal anti-p side, DPE signal p side.

The events with zero hits in the BSC and MP on both the anti-proton and proton side are selected as the DPE sample. To check that these events are really j/psi events the distributions of muon and di-muon quantities are compared to those of the inclusive sample:

Muon quantities:


di-muon quantities:


The plots as eps files: Muon pt, Muon eta, Muon track z0, Muon track d0, Muon N axial hits, Muon chi2, di-muon mass, di-muon pt, di-muon eta, di-muon phi.

[top]

Exclusive Event Selection


From the sample of 107 DPE events, we now search for exclusive events. These are events in which nothing is measured in the CDF detector above noise except for the tracks and towers associated with the j/psi and at most one photon candidate.

The following selections are made after excluding the tracks and towers associated with the j/psi muons:
No hits in the Cerenkov Luminosity detector (CLC) (covers similar region to MP): 99 events
No hits (tower Et > 100 MeV) in the fwd region of the plug calorimeter (2.61 < |eta| < 3.64): 76 events
No hits (tower Et > 100 MeV) in the rest of the plug calorimeter (1.10 < |eta| < 2.61): 45 events
No tracks observed (pt > ~300 MeV): 36 events
No muon stubs observed: 30 events
No hadronic central towers (after rejecting noise, equiv E > ~100MeV) (|eta| < 1.0) 25
0 or 1 EM tower (after rejecting noise, equiv E > ~100MeV) 23 events (13 0 EM tower, 10 1 EM tower)

The following plots show the distribution that is cut on in each case and the yellow histogram shows the events that survive in the final event sample:


The plots as eps files: CLC, fwd plug, plug, tracks, muons,hadronic towers, EM towers.

To check that the noise cuts made to select the exclusive events really select events that are consistent with the detector being empty, the distributions of what is allowed to remain in the detector as noise is compared to that of events in which there was no interaction. The jpsi tracks and towers and where appropriate the photon (there are 13 jpsi only, 10 jpsi+photon events) are excluded from the plots:



The plots as eps files: BSC, MP (anti-proton side), MP (proton side), fwd plug, plug, Central towers.

To check that the exclusive events are really j/psi events the distributions of muon and di-muon quantities are compared to those of the inclusive sample:

Muon quantities:


di-muon quantities:


The plots as eps files: Muon pt, Muon eta, Muon track z0, Muon track d0, Muon N axial hits, Muon chi2, di-muon mass, di-muon pt, di-muon eta, di-muon phi.

The main background is from cosmic ray events overlapping with non-interaction events. The timing information gives an extremely clean separation between cosmic rays and events originating from the interaction point. However, we double check that the events do not look like cosmic rays by looking at the acollinearity (3D opening angle between the muons), Delta(TDC) (timing information is also available from the hadronic TDC's.) the correlation between the muon tracks d0, given by rho and beta, the sum(eta) of tracks in the event and delta(phi) between the tracks:


The plots as eps files: acollinearity, Delta(TDC), rho, beta, Sum(eta), DeltaPhi.

[top]

Comparison to Exclusive chi_c Production



Having selected events that are compatible with containing nothing else over -7.5 < eta < 7.5 except a j/psi or a j/psi + photon, we now study whether or not these events are exceptional.

Firstly we look at all the DPE sample and plot the mass fraction: R_jpsi = M_jpsi/M_X. This is shown below:


In these plots the muon-stub cut is already made since events that are exclusive except for a muon-stub enter the distribution with R_jpsi = 1.
While there is a clear peak at R_jpsi=1, a distribution like this is in principle possible in any low multiplicity distribution since events with N=0 or 1 hit(s) tend to have R_jpsi > 0.9 and events with 2+ hits tend to have a low R_jpsi. The multiplicity in the DPE sample is shown below. Here it can be seen that the multiplicity could be compatible with an exclusive peak plus a flattish DPE distribution or just a low multiplicity DPE distribution with a long tail (shown). In order to estimate the contribution from low multiplicity DPE and whether this contribution could explain all of the events more data is needed.


The plots as eps files: mass fraction, N hits.


Since we cannot measure the background contribution we use the 10 exclusive j/psi + gamma events to set an upper limit on exclusive chic production. In order to do this we first show that the events are consistent with exclusive chi_c production.

The plots shown below are the invariant mass (left) and the Pt (right) of the di-muon and photon candidate compared to a sample of generated chic_0 events passed through a detector simulation:

The plots as eps files: Mass, Pt.


The invariant mass is consistent with that of the chic_0 although, as is seen in the mass fits below, the mean mass is higher and the distribution much broader in the data than in the simulation. This may be due to the simulation or there may be contributions from other chi_c mesons which have higher masses or, it may be that the events are not chi_c^0, but for example DPE j/psi + other pions --> gamma,gamma and one gamma is lost. Further data and better modeling of the DPE and detector is needed to resolve this. 3 events appear to have higher pt than expected, but do not appear otherwise different.


The plots as an eps file.

Other properties of the photon also look compatible with chi_c. However, we have no estimate of the background which could potentially be large since the photon e_t required is low. The plots below show the et, eta and phi of the photon and delta_et, delta_eta and delta_phi between the jpsi and photon.

The plots as an eps file.

The photons are not consistent with being from noise or additional soft interactions. The rate of photon candidates not associated with the hard interaction is estimated to be around 1%. The e_t distribution of these fake photons, measured in non-interaction events, is shown below and is not consistent with the jpsi+photon photon et distribution.

The plot as an eps file.

Finally, we can look at the invariant mass of all exclusive events (those with N_em towers <=1). The top-left plot below shows the invariant mass of all the exclusive events before any cosmic rejection cuts are made. The top-right shows the same distribution after the cosmic rejection cut is used. The bottom-left shows the effect of requiring 1 EM tower, and the bottom-right shows the distribution after both cosmic rejection and requiring 1 EM tower.

The plot as an eps file.

Comparing the bottom-left and top-left plots, we see that the photons are associated with a j/psi - the rate of the photons in the cosmic events is compatible with the measured fake rate (above) while the rate in the jpsi events is much higher. Comparing the top-right and bottom-right plots we see that all the di-muon+photon events occur in the jpsi peak. In the jpsi mass window there are 13 di-muon only events are 10 di-muon+gamma. Outside of this window there are 20 di-muon only events and 0 di-muon+photon events.

[top]

Summary




[top]