1: \section{Data analysis}
2: In the data analysis 258k events taken by a minimum bias trigger during Run2
3: were used. The trigger requires coincident hits
4: in Beam-Beam Counters(BBC) and Zero-Degree Calorimeters.
5: The charged particles were reconstructed by a drift chamber and two multi-wire
6: chambers with pad readouts without a magnetic field in order to statistically
7: enhance the low $p_t$ charged particles. It is essential to focus on the
8: low $p_t$ charged pions for the discussion of the phase transition.
9: The charged particles were measured by the east arm of the PHENIX detector
10: which covers $\pm 0.35$ in pseudo-rapidity and $\pi/2$ in azimuthal angle
11: $\phi$ around the beam line.
12: Events with collision points within $\pm 5$cm from the center of the detector
13: along the beam line were analyzed.
14: The collision centralities were classified by the correlation between
15: BBC charge sum and ZDC energy sum\cite{Milov}.
16: We have checked the detector stability rigorously for a run range
17: we have analyzed.
18:
19: In addition, the dead or inefficient areas in the detector have been
20: identified and the effects on the NBD parameters were carefully
21: studied. We have estimated the relation between input $k$ values of NBD
22: and biased $k$ values due to dead or inefficient areas
23: for each size of pseudo-rapidity interval in each centrality.
24: As long as the baseline distribution is known as NBD, one can generate
25: the number of particles based on the probability distribution
26: in all fine $\eta-\phi$ bins respectively.
27: Dead maps can be produced from the track projection points
28: in $\eta-\phi$ plane in the real data and the maps were used to
29: suppress particles which hit the dead areas defined in the maps.
30: After applying the dead maps, the biased multiplicity distributions
31: were fit with NBD in given rapidity intervals again,
32: which provides correction factors between the true $k$ values and
33: the biased ones.
34: Therefore systematic errors arise from the definition of the dead maps.
35: As a general tendency, the dead maps scales only the absolute values of $k$,
36: but not change the correlations between $k$ and
37: rapidity interval sizes drastically.
38: