0810.4084/ecl.tex
1:  \section{ECL performance}
2:  \label{sec:ecl}
3:  For the electromagnetic calorimeter, ECL, overlapping beam-background
4:  events will degrade the performance. The mean shift of energy distribution due to this
5:  background source can be calibrated in principle, but the energy
6:  resolution will be worse and cannot be recovered unless we resolve the
7:  contribution from the background.  Therefore, we will upgrade the
8:  readout electronics so that we can sample and fit the signal
9:  waveform. With waveform fitting two overlapping waveforms can be
10:  resolved if they are separated by more than about one
11:  peaking time. From a toy MC study, in which we assume a flat background
12:  distribution in time, the background reduction factor is estimated to
13:  be seven in the energy range above 20\,MeV where the background hits
14:  create fake clusters.
15: 
16:  On the other hand, in the energy range below
17:  20\,MeV, the effect of the beam-background is similar to electric
18:  noise; we call this low energy background ``pileup noise''.  The
19:  waveform fitting does not have great suppression power for the pileup
20:  noise, compared to its suppression of fake clusters; the
21:  suppression factor is two for the pileup noise.
22:  The magnitude of the pileup noise, $\sigma_p$, is proportional to the
23:  fluctuation of the background, which is,
24:  \[
25:   \sigma_p\propto\sqrt{\nu\tau (\overline{\epsilon^2}-\overline{\epsilon}^2)},
26:  \]
27:  where $\nu$ is the rate of the photons, $\tau$ is effective duration
28:  of the signal, $\overline{\epsilon}$ and $\overline{\epsilon^{2}}$ are
29:  the mean and the square mean of the photon energies.
30:  Therefore, the background suppression factor for the pileup noise by
31:  waveform fitting is four ($\sqrt{N}/2 = \sqrt{N/4}$).
32:  A concern is only in the
33:  low energy photon reconstruction, which should be examined in the
34:  analysis of $B\to D^\ast K, D^\ast\to D\gamma$ or $D\pi^0$.
35: 
36:  For higher background levels when the suppression by waveform fitting
37:  is not sufficient, CsI(Tl) has to be replaced
38:  with pure CsI crystals. This is planned for both endcaps. 
39:  Because of the shorter time constant for pure CsI ($\sim 30$\,ns)
40:  compared to CsI(Tl) ($\sim 1\,\mu$s), a background reduction of about 30 is expected.
41:  Since the light output will be lower and the wavelength of the 
42:  scintillation light will be in the UV region, photomultiplier tubes 
43:  have to be used instead of PIN photodiodes.
44:  The biggest issue for pure CsI crystals is the cost. The 
45:  number of crystals that we can replace will depend on the available budget.
46:  In the following subsections, effects of a partial upgrade with pure
47:  CsI are discussed.
48: 
49:  \subsection{Partition of endcaps}
50:  If only a part of the endcaps is upgraded with pure CsI crystals, a 
51:  choice on the location of the pure CsI part has to be made.
52:  There are several possible partitions of the two regions.  Two ways of
53:  partitioning are under consideration, 
54:  \begin{itemize}
55:   \item {\bf (A)}: 
56: 	Forward innermost 4 rings (224), middle 5 rings (448) and
57: 	outermost 4 rings (480)\\
58: 	Barrel (6624)\\
59: 	Backward innermost 6 rings (480), and outermost 4 rings (480),
60:   \item {\bf (B)}: 
61: 	Forward innermost 5 rings (288), next 2 rings (192), next 2
62: 	rings (192) and outermost 4 rings (480)\\
63: 	Barrel (6624)\\
64: 	Backward innermost 3 rings (192), next 1 ring (96), next 2 rings
65: 	(192) and outermost 4 rings (480),
66:  \end{itemize}
67:  where the numbers in parentheses are the numbers of crystals in each partition.
68:  The difference between the two turn out to be minor, 
69:  and are not crucial  for the conclusion of the studies in this
70:  section. 
71:  %Just to record which study is based on what, those
72:  % are as follows, :
73:  
74:  \subsection{Study of $B\to\PKzS\Pgpz\gamma$ decays}
75:  Assuming that the readout electronics will be upgraded and that
76:  waveform fitting will give us a background suppression by a factor of 7 as
77:  expected, 20 times more background corresponds to 3 times more
78:  background for the case with CsI(Tl), while we will be essentially free from
79:  background for the case with pure CsI. 
80:  The effect of a partial upgrade with pure CsI on $B\to\PKzS\Pgpz\gamma$ is studied
81:  with partitioning scheme A.
82:  Using exp 55 MC, we overlaid 3 times more background to the part with
83:  CsI(Tl), while no background is overlaid on the part with pure CsI.
84:  For detectors other than ECL, 3 times more background is overlaid in
85:  all cases for technical reasons.
86: 
87:  Several combinations of partitions are tested. Each configuration is
88:  identified with a 6 digit identifier like 033000, where each digit
89:  represents the background multiplication factor; the leftmost
90:  digit refers to the most backward (largest in theta) partition,
91:  and the rightmost digit to the most forward (smallest in theta)
92:  partition. Hence in the example given above, 033000, corresponds to 
93:  no background in the backward innermost 6 rings,
94:  3 times more background in the backgward outermost 4 rings, 3 times
95:  more background in the barrel, and no background in the forward endcap.
96:  Note that the present conditions correspond to 111111.
97: 
98:  As shown in Fig.~\ref{fig:de_kspi0gam}, in the case of 333333 (green,
99:  with filled circles),
100:  the mean of the $\Delta E$ distribution is shifted towards the positive
101:  side, the width becomes larger, and the signal yield is lower compared
102:  to 111111 (black solid histogram); these are consequences of the
103:  background superposition.  We observe a similar deterioration also in
104:  the case of 003000 (red, with open circles), namely
105:  the case with both forward and backward endcaps fully upgraded to pure
106:  CsI. Other configurations (003300, 003333, 033000, 033300, 033330,
107:  033333, 333000, 333300 and 333330) are also tested; it is found that the 
108:  corresponding performance is between the performances for 003000 and 333333.
109:  This deterioration corresponds to a 10 to 15 percent loss of
110:  luminosity, depending on the number of pure CsI crystals.
111: 
112:  \begin{figure}
113:   \resizebox{0.46\columnwidth}{!}{\includegraphics{epsfiles/ecl_bkg_kspi0gam/de3_kspi0gam.eps}}
114:   \caption{ $\Delta E$ distributions for $B\to\PKzS\Pgpz\gamma$ decays for several
115:   background conditions.
116:   }
117:   \label{fig:de_kspi0gam}
118:  \end{figure}
119: 
120:  Note that in this decay photons with relatively high energy
121:  are produced as  shown in Fig.~\ref{fig:energies_kspi0gam}.
122:  The analysis can be performed using the barrel ECL.
123:  Therefore, the impact of having pure CsI in the endcaps on this mode
124:  is not very large, although it is preferable to have a larger number of 
125:  pure CsI crystals. 
126: 
127:  \begin{figure}
128:   \resizebox{0.46\columnwidth}{!}{\includegraphics{epsfiles/ecl_bkg_kspi0gam/energies_bw.eps}}
129:   \caption{ Laboratory energy distribution of three photons in the
130:   $B\to\PKzS\Pgpz\gamma$ decay sequence: the primary photon from the $B$
131:   decay (top) and the two photons from the $\Pgpz$ decay ordered by energy (middle and bottom).}
132:   \label{fig:energies_kspi0gam}
133:  \end{figure}
134: 
135:  % I. Nakamura-san
136:  \input{ecl_frec}
137: 
138: % \subsection{$B\to D^\ast K, D^\ast\to D\gamma, D\Pgpz$}
139: % is an important decay mode, and contains low energy photons, however,
140: % no study has been done yet.
141: