Comment on: “Search for direct photons from S-Au Collisions at 200 GeV/nucleon” by the CERES Collaboration

In this note we comment on a recent publication in this journal by the CERES (NA45) Collaboration [1]. The authors report to have determined an upper limit on the direct photon yield relative to the decay photon yield in S + Au collisions of 14 % and 7 % by two different methods, respectively. We argue that these limits are unsupported by the results and analysis of the CERES data. The systematic error estimates quoted in the CERES analysis are consistently overly optimistic. Using more realistic estimates of the various error contributions and propagating them appropriately we arrive at a direct photon upper limit which at best is 20 % of the inclusive photon yield, and most probably is much higher. The authors of Ref. [1] have presented an analysis in which they report to demonstrate an upper limit on the production of excess direct photons in 200 AGeV S + Au reactions. They obtain upper limits, at the 90 % confidence level, of 14 % and 7 % of the inclusive photon yield derived respectively by two different methods. This is a remarkable achievement by an experiment optimized to measure electron pairs. Dedicated photon experiments attempting to make similar measurements [2] have hardly achieved similar accuracy. The reported precision is even more remarkable in light of the fact that, as the authors report, only 2.3% of the photons produced are converted and available for measurement, of these less than 20% are identified in the experimental apparatus, and of those photons identified, 18% are background photons resulting from Dalitz decays. Furthermore, in the search for excess photons, the measured inclusive yield is to be compared to the expected photon yield arising from conventional background sources, predominantly radiative decays of π and η mesons, none of which have been measured for the CERES experimental acceptance and event selection. Since the low p⊥ direct photon excess is expected to be small, if observable at all, in S+Au reactions [2] it is not surprising that this experiment can only set an upper limit. In setting an upper limit, the crucial issue for the experimental measurement is an accurate understanding and estimate of the systematic errors. As the quality and importance of the measurement is to be judged by its level of accuracy, we consider in detail the analysis of the systematic errors presented in Ref. [1] and we shall demonstrate that the authors have decidedly underestimated their systematic errors. Apart from raising concerns about this particular measurement, our comments should be viewed in a broader context, since preparations are presently being made for the first generation of LHC experiments, notably ALICE, where the detection of direct photons is one of the aims. To decide, wether the “direct” or the conversion method is better suited for this purpose, a detailed knowledge of the advantages and disadvantages of the different methods is indispensible. For these reasons we have put considerable effort into the comparison of the methods and to study the corresponding systematic errors. In the first method used by the authors an upper limit on the integrated direct photon production is extracted by studying the ratio rγ, given in Eq.(1) of Ref. [1], which is the ratio of the integrated photon p⊥ distribution, integrated from 0.4 GeV/c ≤ p⊥ ≤ 2.0 GeV/c, to dNch/dη. This ratio is calculated from the experimental data, rdata, and compared to the same ratio calculated from simulation for photons arising from hadron decays, rhadr. The ratio of these two ratios provides a determination of the possible direct photon yield in this p⊥ region. A slight direct photon excess of 4 % is determined in Ref. [1]. The systematic error on these two ratios is then used to set a direct photon upper limit.