Central Exclusive Production at the Tevatron

In CDF we have observed several exclusive processes: γγ → e+e− and μ+μ−, γ + IP → J/ψ,ψ(2S), and IP + IP → χc. The cross sections agree with QED, HERA photoproduction data, and theoretical estimates of gg → χc with another gluon exchanged to screen the color. This observation of exclusive χc, together with earlier observations of exclusive dijets and exclusive γγ candidates, support some theoretical predictions for p + p → p +H + p at the LHC. Exclusive dileptons offer the best means of precisely calibrating forward proton spectrometers. 1 Central Exclusive Production Central exclusive production at the Tevatron is the process p + p̄ → p + X + p̄, where “+” means a rapidity gap ∆y exceeding 3 units, and X is a simple system fully measured. Exchanges (t-channel) over such large gaps must be color singlets with spin J [or Regge intercept α(0)] ≥ 1.0. Only photons γ and pomerons IP qualify, apart from W and Z bosons which always cause the proton to break up. The gluon g would qualify apart from its color, but if another gluon is exchanged that can be cancelled, and IP = gg is often a good approximation. It cannot be exact; QCD forbids a pure gg state, and a qq̄ component certainly grows as Q2 increases. The IP has C = +1; in QCD one should also have a ggg state with C = -1, the odderon [1] O, not yet observed. The central masses MX are roughly limited to MX . √ s 20 with the outgoing protons having Feynman xF > 0.95. Hence MX . 3 GeV at the CERN ISR [2], appropriate for glueball spectroscopy, where M(π+π−) shows a broad f0(600), a narrow f0(980) and still unexplained structure possibly associated with f0(1710), a glueball candidate. The study ofX = hadrons, e.g. φφ and D◦D̄◦ to name two channels among many, has not been studied above ISR energies, but CDF is a perfect place to do it and hopefully it will be done. At the LHC MX can reach ≈ 700 GeV, into the electroweak sector, and we can have X = Z,H,W+W−, ZZ , slepton pairs l̃l̃, etc. Measuring the forward protons after 120m of 8T dipoles, in association with the central event, as the FP420 [4] proponents hope to do at ATLAS and CMS, one can measure MX with σ(MX) ≈ 2 GeV per event [5], and for a state such as H , also its width if Γ(H) & 3 GeV/c2. There are scenarios (e.g. SUSY) in which FP420 could provide unique measurements, e.g. if there are two nearby states both decaying to bb̄ or to W+W−. The quantum numbers of X are J = 0++ or 2++ (and these are distinguishable) for IPIP production. Two-photon collisions γγ → l+l−,W+W−, l̃l̃ become important at the LHC thanks to the intense high momentum photons, orders of magnitude more than at the Tevatron, giving > 50 fb for W+W− as a continuum background to H → W+W−. H → ZZ does not have this background. While there is a gold mine of physics in p+X+p at the LHC, we need to show that (a) the cross sections are within reach, and (b) one can build the spectrometers with resolution σ(MX) ≈ 2 GeV/c2 and calibrate their momentum scale and resolution, to measure Γ(H), and perhaps to distinguish nearby states. Both these issues are addressed by CDF in a “TeV4LHC” spirit, and they are also very interesting in their own right. The calculation of cross sections (e.g. [6]) involves, in addition to σ(gg → X), the unintegrated gluon distribution g(x1, x2), rapidity gap survival probability (no other parton interactions), and the Sudakov factor (probability of no gluon radiation producing hadrons). The Durham group predicts σ(SMH) for p+H + p at the LHC = 3 ÷3 fb. At the Tevatron p + H + p̄ is out of reach, but the process p + χc(χb) + p̄ is identical as far as QCD is concerned, as is p+γγ+p̄. Measuring these constrains the SMH cross section. In CDF we have looked for both exclusive γγ [7] and χc [8], without however having detectors able to see the p and p̄. Instead we added forward calorimeters (3.5 < |η| < 5.1) and beam shower counters BSC (5.5 < |η| < 7.4). If these are all empty there is a high probability that both p and p̄ escaped intact with small |t|. We also measured [9] exclusive dijets. For the exclusive γγ search we triggered on events with two electromagnetic (EM ) clusters with ET > 4 GeV in the central calorimeter, with a veto on signals in the BSC. This killed pile-up events and enabled us to take data without prescaling the trigger. We required all other detectors to be consistent with only noise; then our effective luminosity is only about 10% of the delivered luminosity. We found [7] 3 events with exactly two back-to-back EM -showers (assumed to be photons) with M(γγ) > 10 GeV/c2. From wire proportional chambers at the shower maximum we concluded that two were perfect p + p̄ → p + γγ + p̄ candidates and one was also consistent with being a p+ p̄→ p+ π◦π◦ + p̄ event. The Durham prediction [10] was 0.8 ÷3 events, clearly consistent. We have since accumulated more data, with a lower threshold, now being analysed. With the above trigger we also found [11] 16 p + p̄ → p + e+e− + p̄ events, with M(e+e−) > 10 GeV/c2 (up to 38 GeV/c2), the QED γγ → e+e− process [12]. Exclusive 2-photon processes had not previously been observed in hadron-hadron collisions; the cross section agrees with the precise theory prediction. This process has been suggested as a means of calibrating the LHC luminosity; then it must be done in the presence of pile-up, and one will need to know the acceptance etc. at the few % level. More interesting for FP420 is that measurement of an exclusive lepton pair gives both forward proton momenta, with a precision dominated by the incoming beam momentum spread ( δp p ≈ 10−4, or 700 MeV). One can do this with pile-up, selecting dileptons with no associated tracks on the l+l− vertex and ∆φ ≈ π. One can also cut on pT (l+l−) (correlated with ∆φ), but ∆φ has better resolution. In CDF we found that a cut π−∆φ < 0.8GeV M(l+l−) rads is suitable for QED-produced pairs. For each pair one can predict ξ1 and ξ2, and, if a proton is in the FP420 acceptance, compare ξi and ξ420. This can also possibly map the acceptance A(ξ, t ≈0), as the cross section shape is known from QED, and the (Coulomb) protons have very small t. CDF also used a “muon+track” trigger, again with BSC veto, to study p + p̄ → p + μ+μ− + p̄ with 3 GeV/c2 < M(μμ) <4 GeV/c2. This is a very rich region, with the J/ψ and ψ(2S) vector mesons that can only be produced exclusively by photoproduction γ + IP → ψ, or Fig. 1: Exclusive dimuon mass spectrum in the charmonium region, together with the sum of two Gaussians and the QED continuum, shown in the inset, excluding the 3.65 3.75 GeV/c bin (ψ(2S)). All line shapes are predetermined, with the normalization free. possibly by odderon exchange: O+ IP → ψ. We know what to expect for photoproduction from HERA, so an excess would be evidence for the elusive O. The spectrum [8] is shown in Fig. 1, together with the sum of three components: the vector mesons and a continuum, γγ → μ+μ−, which is again consistent with QED. These central exclusive spectra are exceptionally clean; in fact the biggest background (≈ 10%) is the identical process but with an undetected p → p∗ dissociation. The J/ψ and ψ(2S) cross sections dσ dy |y=0, are (3.92±0.62)nb and (0.54±0.15)nb, agreeing with expectations [13, 14]. Thus we do not have evidence for O exchange, and put a limit O γ < 0.34 (95% c.l.), compared with a theory prediction [15] 0.3 0.6. While the QED and photoproduction processes in Fig. 1 should hold no surprises, their agreement with expectations validates the analysis. We required no EM tower with E T > 80 MeV. If we allow such signals (essentially γ’s) the number of J/ψ events jumps from 286 to 352, while the number of ψ(2S) only increases from 39 to 40. The spectrum of EM showers is shown in Fig. 2. These extra J/ψ events are very consistent with being χc0(3415) → J/ψ + γ, from IPIP → χc, with about 20% of the γ being not detected (giving a background of 4% under the exclusive J/ψ). We measure dσ dy (χc)|y=0 = (75±14)nb. The existence of this process implies that p+H +p must happen at the LHC (assuming H exists), as the QCD physics is qualitatively identical. The χc cross section agrees with predictions: 150nb [16] and 130 ×4 ÷4nb [6]. It is therefore likely that σ(p + p → p + SMH + p) is of order 0.5-5 fb, within reach of FP420. In SUSY models the cross section can be much higher [4]. We are looking for p+p̄→ p+Υ+p̄ (by photoproduction, or byO+IP ), and IP+IP → χb. The Υ should be measurable in the presence of pile-up using nass = 0, ∆φ and pT cuts (nass is the number of additional tracks on the dilepton vertex). We have candidate events, with the Υ(1S), (2S) and (3S) states resolved; cross sections are now being determined. The χb → Υ+γ 0 100 200 300 400 500 600 0 2 4 6 8 10 12 14 16 18 20 / ndf 2 χ 0 / -3