1: \section{Electromagnetic Calorimeter}
2: \label{sec:emc}
3: \newcommand{\CsI}{\rm CsI(Tl)}
4:
5:
6: \subsection{Purpose and Design}
7:
8: The electromagnetic calorimeter (EMC) is designed to measure
9: electromagnetic showers with excellent efficiency, and energy and
10: angular resolution over the energy range from 20\mev\ to 9\gev.
11: This capability allows the detection of photons from \piz\ and
12: $\eta$ decays as well as from electromagnetic and radiative
13: processes. By identifying electrons, the EMC contributes to the
14: flavor tagging of neutral $B$ mesons via semi-leptonic decays, to
15: the reconstruction of vector mesons like \jpsi, and to the study
16: of semi-leptonic and rare decays of $B$ and $D$ mesons, and $\tau$
17: leptons. The upper bound of the energy range is set by the need
18: to measure QED processes, like $\epem \to \epem (\gamma)$ and
19: $\epem \to \gamma \gamma$, for calibration and luminosity
20: determination. The lower bound is set by the need for highly
21: efficient reconstruction of $B$-meson decays containing multiple
22: $\piz$s and $\eta^0$s.
23:
24: \subsubsection{Requirements}
25:
26: The measurement of extremely rare decays of $B$ mesons containing
27: $\piz$s (\eg\ $B^0 \to \piz \piz$) poses the most stringent
28: requirements on energy resolution, namely of order 1--2\%. Below
29: energies of 2\gev, the \piz\ mass resolution is dominated by the
30: energy resolution. At higher energies, the angular resolution
31: becomes dominant, and therefore it is required to be of the order
32: of a few mrad.
33:
34: Furthermore, the EMC has to be compatible with the 1.5\tesla\
35: field of the solenoid and operate reliably over the anticipated
36: ten-year lifetime of the experiment. To achieve excellent
37: resolution, stable operating conditions have to be maintained.
38: Temperatures and the radiation exposure must be closely monitored,
39: and precise calibrations of the electronics and energy response
40: over the full dynamic range must be performed frequently.
41:
42:
43: \subsubsection{Design Considerations}
44:
45: The requirements stated above lead to the choice of a hermetic,
46: total-absorption calorimeter, composed of a finely segmented array
47: of thallium-doped cesium iodide (\CsI) crystals. The crystals are
48: read out with silicon photodiodes that are matched to the spectrum
49: of scintillation light. Recent experience at CLEO~\cite{cleo_emc}
50: has demonstrated the suitability of this choice for physics at the
51: \FourS\ resonance.
52:
53: The energy resolution of a homogeneous crystal calorimeter can be
54: described empirically in terms of a sum of two terms added in
55: quadrature
56: \begin{equation}
57: \frac{\sigma_{E}}{E}= \frac{a}{^{4}\sqrt{E (\gev)}}\oplus b,
58: \label{caleqn::res}
59: \end{equation}
60: where $E$ and $\sigma_{E}$ refer to the energy of a photon and its
61: rms error, measured in \gev. The energy dependent term $a$ arises
62: primarily from the fluctuations in photon statistics, but it is
63: also impacted by electronic noise of the photon detector and
64: electronics. Furthermore, beam-generated background will lead to
65: large numbers of additional photons that add to the noise. This
66: term is dominant at low energies. The constant term, $b$, is
67: dominant at higher energies ($>1$\gev). It arises from
68: non-uniformity in light collection, leakage or absorption in the
69: material between and in front of the crystals, and uncertainties
70: in the calibrations. Most of these effects can be influenced by
71: design choices, and they are stable with time. Others will be
72: impacted by changes in the operating conditions, like variations
73: in temperature, electronics gain, and noise, as well as by
74: radiation damage caused by beam-generated radiation.
75:
76: The angular resolution is determined by the transverse crystal size
77: and the distance from the interaction point. It can also be
78: empirically parameterized as a sum of an energy dependent and a
79: constant term,
80: \begin{equation}\label{caleqn::posres}
81: \sigma_{\theta}=\sigma_{\phi} = \frac{c}{\sqrt{E (\gev)}} + d,
82: \end{equation}
83: where the energy $E$ is measured in \gev. The design of the EMC
84: required a careful optimization of a wide range of choices,
85: including the crystal material and dimensions, the choice of the
86: photon detector and readout electronics, and the design of a
87: calibration and monitoring system. These choices were made on the
88: basis of extensive studies, prototyping and beam
89: tests~\cite{beamtest}, and Monte Carlo simulation, taking into
90: account limitations of space and the impact of other \babar\
91: detector systems.
92:
93: Under ideal conditions, values for the energy resolution
94: parameters $a$ and $b$ close to 1--2\% could be obtained. A
95: position resolution of a few mm will translate into an angular
96: resolution of a few mrad; corresponding parameter values are
97: $c\approx 3\mrad$ and $d\approx 1\mrad$.
98:
99: However in practice, such performance is very difficult to achieve
100: in a large system with a small, but unavoidable amount of inert
101: material and gaps, limitations of electronics, and background in
102: multi-particle events, plus contributions from beam-generated
103: background.
104:
105: Though in \CsI\ the intrinsic efficiency for the detection of photons
106: is close to 100\% down to a few \mev, the minimum measurable energy in
107: colliding beam data is expected to be about 20~\mev, a limit that is
108: largely determined by beam- and event-related background and the
109: amount of material in front of the calorimeter. Because of the
110: sensitivity of the \piz\ efficiency to the minimum detectable photon
111: energy, it is extremely important to keep the amount of material in
112: front of the EMC to the lowest possible level.
113:
114: \subsubsection{CsI(Tl) Crystals}
115:
116: Thallium-doped CsI meets the needs of \babar\ in several ways. Its
117: properties are listed in Table~\ref{tab:csiproperties}. The high
118: light yield and small Moli\`{e}re radius allow for excellent
119: energy and angular resolution, while the short radiation length
120: allows for shower containment at \babar\ energies with a
121: relatively compact design. Furthermore, the high light yield and
122: the emission spectrum permit efficient use of silicon photodiodes
123: which operate well in high magnetic fields. The transverse size of
124: the crystals is chosen to be comparable to the Moli\`{e}re radius
125: achieving the required angular resolution at low energies while
126: appropriately limiting the total number of crystals (and readout
127: channels).
128:
129:
130: \begin{table}
131: \caption{Properties of \CsI\ .}
132:
133: \vspace{.5\baselineskip}
134: \begin{center}
135: \begin{tabular}{ll}\hline
136: Parameter\rule[-5pt]{0pt}{17pt} & Values \\ \hline
137: \rule{0pt}{12pt}Radiation Length & 1.85 \cm\\ Moli\`{e}re Radius &
138: 3.8 \cm \\ Density & 4.53 $\mathrm{g/cm^{3}}$ \\ Light Yield &
139: 50,000 ${\rm \gamma/\mev}$ \\ Light Yield Temp. Coeff. &
140: 0.28\%/\degc\ \\ Peak Emission $\lambda_\mathrm{max}$ & 565 \nm \\
141: Refractive Index ($\lambda_\mathrm{max}$) &1.80 \\ Signal Decay
142: Time & 680 \ns\ (64\%) \\ \rule[-5pt]{0pt}{0pt}& 3.34 \mus\
143: (36\%)\\ \hline
144: \end{tabular}
145: \end{center}
146: \label{tab:csiproperties} \vspace{-.5\baselineskip}
147: \end{table}
148:
149:
150:
151: \subsection{Layout and Assembly}
152: \label{calsec::mechanics}
153:
154: \begin{figure*}
155: \begin{center}
156: \includegraphics[width=12.5cm]{8572A03.eps}
157: \vspace{-2pc}
158: \caption{A longitudinal cross section of the EMC (only the top half is
159: shown) indicating the arrangement of the 56 crystal rings. The
160: detector is axially symmetric around the $z$-axis. All dimensions are
161: given in mm.}
162: \label{calfig::detector}
163: \end{center}
164: \end{figure*}
165:
166: \subsubsection{Overall Layout}
167:
168: The EMC consists of a cylindrical barrel and a conical forward
169: endcap. It has full coverage in azimuth and extends in polar angle
170: from 15.8\degrees\ to 141.8\degrees\ corresponding to a
171: solid-angle coverage of 90\% in the c.m. system (see
172: Figure~\ref{calfig::detector} and Table~\ref{tab:csigeometry}).
173: The barrel contains 5,760 crystals arranged in 48 distinct rings
174: with 120 identical crystals each. The endcap holds 820 crystals
175: arranged in eight rings, adding up to a total of 6,580 crystals.
176: The crystals have a tapered trapezoidal cross section. The length
177: of the crystals increases from 29.6\cm\ in the backward to
178: 32.4\cm\ in the forward direction to limit the effects of shower
179: leakage from increasingly higher energy particles.
180:
181: \begin{table}[!htb]
182: \caption{Layout of the EMC, composed of 56 axially symmetric
183: rings, each consisting of CsI crystals of identical dimensions. }
184: \vspace{.5\baselineskip}
185: \begin{center}
186: \begin{tabular}{cccc} \hline
187: \rule{0pt}{12pt}$\theta$ Interval & Length & \# & Crystals \\
188: \rule[-5pt]{0pt}{0pt}(radians) &($X_{0}$) &Rings & /Ring \\ \hline
189: \multicolumn{4}{c}{\rule[-5pt]{0pt}{17pt}Barrel}\\ \hline
190: \rule{0pt}{12pt}$2.456-1.214$ & 16.0 & 27 & 120 \\
191: $1.213-0.902$ & 16.5 & 7 & 120 \\
192: $0.901-0.655$ & 17.0 & 7 & 120 \\
193: \rule[-5pt]{0pt}{0pt}$0.654-0.473$ & 17.5 & 7 & 120 \\ \hline
194: \multicolumn{4}{c}{\rule[-5pt]{0pt}{17pt}Endcap} \\ \hline
195: \rule{0pt}{12pt}$0.469-0.398$ & 17.5 & 3 & 120 \\
196: $0.397-0.327$ & 17.5 & 3 & 100 \\
197: $0.326-0.301$ & 17.5 & 1 & 80 \\
198: \rule[-5pt]{0pt}{0pt}$0.300-0.277$ & 16.5 & 1 & 80 \\ \hline
199: \end{tabular}
200: \end{center}
201: \label{tab:csigeometry}
202: \end{table}
203:
204:
205: To minimize the probability of pre-showering, the crystals are
206: supported at the outer radius, with only a thin gas seal at the
207: front. The barrel and outer five rings of the endcap have less than
208: 0.3--0.6\Xrad\ of material in front of the crystal faces. The SVT
209: support structure and electronics, as well as the B1 dipole shadow the
210: inner three rings of the endcap, resulting in up to 3.0\Xrad\ for
211: the innermost ring. The principal purpose of the two innermost rings
212: is to enhance shower containment for particles close to the acceptance
213: limit.
214:
215: \subsubsection{Crystal Fabrication and Assembly}
216:
217: The crystals were grown in boules from a melt of CsI salt doped
218: with 0.1\% thallium~\cite{salt}. They were cut from the boules,
219: machined into tapered trapezoids (Figure~\ref{calfig::crystal}) to
220: a tolerance of $\pm 150$\mum, and then polished~\cite{vendor}.
221: The transverse dimensions of the crystals for each of the 56 rings
222: vary to achieve the required hermetic coverage. The typical area
223: of the front face is $4.7 \times 4.7 \cma$, while the back face
224: area is typically $6.1 \times 6.0 \cma$. The crystals act not
225: only as a total-absorption scintillating medium, but also as a
226: light guide to collect light at the photodiodes that are mounted
227: on the rear surface. At the polished crystal surface light is
228: internally reflected, and a small fraction is transmitted. The
229: transmitted light is recovered in part by wrapping the crystal
230: with two layers of diffuse white reflector~\cite{tyvek,gerd}, each
231: 165\mum\ thick. The uniformity of light yield along the wrapped
232: crystal was measured by recording the signal from a highly
233: collimated radioactive source at 20 points along the length of the
234: crystal. The light yield was required to be uniform to within
235: $\pm 2$\% in the front half of the crystal; the limit increased
236: linearly up to a maximum of $\pm 5$\% at the rear face.
237: Adjustments were made on individual crystals to meet these
238: criteria by selectively roughing or polishing the crystal surface
239: to reduce or increase its reflectivity.
240:
241: \begin{figure}
242: \begin{center}
243: \includegraphics[width=6.2cm]{8572A02.eps}
244: \vspace{-2pc}
245: \caption{A schematic of the wrapped \CsI\ crystal and the
246: front-end readout package mounted on the rear face. Also
247: indicated is the tapered, trapezoidal CFC compartment, which is
248: open at the front. This drawing is not to scale.}
249: \label{calfig::crystal}
250: \end{center}
251: \end{figure}
252:
253: Following these checks, the crystals were further wrapped in 25\mum\
254: thick aluminum foil which was electrically connected to the metal
255: housing of the photodiode-preamplifier assembly to provide a Faraday
256: shield. The crystals were covered on the outside with a 13\mum-thick
257: layer of mylar to assure electrical isolation from the external
258: support.
259:
260:
261: \subsubsection{Photodiodes\\ and Preamplifier
262: Assembly}
263:
264: The photon detector consists of two $2 \times 1\cma $ silicon PIN
265: diodes glued to a transparent 1.2\mm-thick polysterene substrate
266: that, in turn, is glued to the center of the rear face of the crystal by
267: an optical epoxy~\cite{epoxy} to maximize light
268: transmission~\cite{readout}. The surrounding area of the crystal face
269: is covered by a plastic plate coated with white reflective
270: paint~\cite{paint}. The plate has two 3\mm-diameter penetrations for
271: the fibers of the light pulser monitoring system.
272:
273: As part of the quality control process, the 1.836\mev\ photon line
274: from a $^{88}$Y radioactive source was used to measure the light
275: yield of every crystal-diode assembly, employing a
276: preamplifier with 2\mus\ Gaussian shaping. The resulting signal
277: distribution had a mean and rms width of 7300 and
278: 890~photoelectrons/MeV, respectively; none of the crystals had
279: a signal of less than 4600~photoelectrons/MeV~\cite{readout,emc_footnote}.
280:
281: Each of the diodes is directly connected to a low-noise
282: preamplifier. The entire assembly is enclosed by an aluminum
283: fixture as shown in Figure~\ref{calfig::crystal}. This fixture is
284: electrically coupled to the aluminum foil wrapped around the
285: crystal and thermally coupled to the support frame to dissipate
286: the heat load from the preamplifiers.
287:
288: Extensive aging tests were performed to ascertain that the diodes and
289: the preamplifiers met the ten-year lifetime requirements. In
290: addition, daily thermal cycles of $\pm$5\degc\ were run for many
291: months to assure that the diode-crystal epoxy joint could sustain
292: modest temperature variations.
293:
294: \subsubsection{Crystal Support Structure}
295:
296: The crystals are inserted into modules that are supported individually
297: from an external support structure. This structure is built in three
298: sections, a cylinder for the barrel and two semi-circular structures
299: for the forward endcap. The barrel support cylinder carries the load
300: of the barrel modules plus the forward endcap to the magnet iron
301: through four flexible supports. These supports decouple and dampen
302: any acceleration induced by movements of the magnet iron during a
303: potential earthquake.
304:
305:
306: \begin{figure*}
307: \begin{center}
308: \includegraphics[width=15.0cm]{8572A06}
309: \vspace{-2pc}
310: \caption{The EMC barrel support structure, with details on the
311: modules and electronics crates (not to scale).}
312: \label{calfig::construction}
313: \end{center}
314: \end{figure*}
315:
316: The modules are built from tapered, trapezoidal compartments made from
317: carbon-fiber-epoxy composite (CFC) with 300\mum-thick walls
318: (Figure~\ref{calfig::construction}). Each compartment loosely holds a
319: single wrapped and instrumented crystal and thus assures that the
320: forces on the crystal surfaces never exceed its own weight. Each
321: module is surrounded by an additional layer of 300\mum\ CFC to
322: provide additional strength. The modules are bonded to an aluminum
323: strong-back that is mounted on the external support. This scheme
324: minimizes inter-crystal materials while exerting minimal force on the
325: crystal surfaces; this prevents deformations and surface degradation
326: that could compromise performance. By supporting the modules at the
327: back, the material in front of the crystals is kept to a minimum.
328:
329: The barrel section is divided into $280$ separate modules, each
330: holding 21 crystals ($7 \times 3$ in $\theta \times \phi$).
331: After the insertion of the crystals, the aluminum readout frames,
332: which also stiffen the module, are attached with thermally-conducting
333: epoxy to each of the CFC compartments. The entire 100\kg-module is
334: then bolted and again thermally epoxied to an aluminum strong-back.
335: The strong-back contains alignment features as well as channels that
336: couple into the cooling system. Each module was installed into the
337: 2.5\cm-thick, 4\m-long aluminum support cylinder, and subsequently
338: aligned. On each of the thick annular end-flanges this cylinder
339: contains access ports for digitizing electronics crates with
340: associated cooling channels, as well as mounting features and
341: alignment dowels for the forward endcap.
342:
343: The endcap is constructed from 20 identical CFC modules (each with 41
344: crystals), individually aligned and bolted to one of two semi-circular
345: support structures. The endcap is split vertically into two halves
346: to facilitate access to the central detector components.
347:
348: The entire calorimeter is surrounded by a double Faraday shield
349: composed of two 1\mm-thick aluminum sheets so that the diodes and
350: preamplifiers are further shielded from external noise. This cage
351: also serves as the environmental barrier, allowing the slightly
352: hygroscopic crystals to reside in a dry, temperature controlled
353: nitrogen atmosphere.
354:
355: \subsubsection{Cooling System}
356:
357: The EMC is maintained at constant, accurately monitored
358: temperature. Of particular concern are the stability of the
359: photodiode leakage current which rises exponentially with
360: temperature, and the large number of diode-crystal epoxy joints
361: that could experience stress due to differential thermal
362: expansion. In addition, the light yield of \CsI\ is weakly
363: temperature dependent.
364:
365: The primary heat sources internal to the calorimeter are the
366: preamplifiers ($2 \times 50$\mw/crystal) and the digitizing electronics
367: (3\kw\ per end-flange). In the barrel, the preamplifier heat is
368: removed by conduction to the module strong backs which are directly
369: cooled by Fluorinert (polychlorotrifluoro-ethylene)~\cite{fluorinert}.
370: The digitizing electronics are housed in 80 mini-crates, each in
371: contact with the end-flanges of the cylindrical support structure.
372: These crates are indirectly cooled by chilled water pumped through
373: channels milled into the end-flanges close to the inner and outer
374: radii. A separate Fluorinert system in the endcap cools both the 20
375: mini-crates of digitizing electronics and the preamplifiers.
376:
377:
378: \subsection{Electronics}
379: \label{calsec::electronics}
380:
381:
382: \begin{figure*}
383: \begin{center}
384: \includegraphics[width=12cm]{8572A04.eps}
385: \vspace{-2pc} \caption{Schematic diagram of the EMC readout
386: electronics.} \label{calfig::electronics}
387: \end{center}
388: \end{figure*}
389:
390:
391: The EMC electronics system, shown schematically in
392: Figure~\ref{calfig::electronics}, is required to have negligible
393: impact on the energy resolution of electromagnetic showers
394: from 20\mev\ to 9\gev, while accommodating the use of a 6.13\mev\
395: radioactive source for calibration. These requirements set a limit of
396: less than 250\kev\ equivalent noise energy (ENE) per crystal and
397: define an 18-bit effective dynamic range of the digitization
398: scheme. For source calibrations, the least significant bit is set to
399: 50\kev, while for colliding beam data it is set to 200\kev. To
400: reach the required energy resolution at high energies, the coherent
401: component has to be significantly smaller than the incoherent noise
402: component. In addition, the impact of high rates of low energy
403: ($<$5\mev) beam-induced photon background needs to be minimized.
404:
405:
406: \subsubsection{Photodiode Readout\\ and Preamplifiers}
407:
408: The ENE is minimized by maximizing the light yield and collection,
409: employing a highly efficient photon detector, and a low-noise
410: electronic readout. The PIN silicon photodiodes~\cite{hamamatsu}
411: have a quantum efficiency of 85\% for the $\CsI$ scintillation
412: light~\cite{diodes}. At a depletion voltage of 70\volt, their
413: typical dark currents were measured to be 4\nA\ for an average
414: capacitance of 85\pf; the diodes are operated at a voltage of
415: 50\volt. The input capacitance to the preamplifier is minimized by
416: connecting the diodes to the preamplifier with a very short cable.
417: The preamplifier is a low-noise charge-sensitive amplifier
418: implemented as a custom application specific integrated circuit
419: (ASIC)~\cite{freytag}. It shapes the signal and acts as a
420: band-pass filter to remove high- and low-frequency noise
421: components. The optimum shaping time for the $\CsI$-photodiode
422: readout is 2--3\mus, but a shorter time was chosen to reduce the
423: probability of overlap with low-energy photons from beam
424: background. The commensurate degradation in noise performance is
425: recovered by implementing a realtime digital signal-processing
426: algorithm following digitization.
427:
428: To achieve the required operational reliability~\cite{reliability} for
429: the inaccessible front-end readout components, two photodiodes were
430: installed, each connected to a preamplifier. In addition, all
431: components were carefully selected and subjected to rigorous tests,
432: including a 72-hour burn-in of the preamplifiers at 70\degc\ to avoid
433: infant mortality. The dual signals are combined in the
434: postamplification/digitization circuits, installed in mini-crates at
435: the end-flanges, a location that is accessible for maintenance.
436:
437:
438: \subsubsection{Postamplification, Digitization\\ and Readout}
439:
440: The two preamplifiers on each crystal, A and B, each provide
441: amplification factors of 1 and 32 and thus reduce the dynamic
442: range of the signal that is transmitted to the mini-crates to
443: 13-bits. A custom auto-range encoding (CARE)
444: circuit~\cite{freytag} further amplifies the signal to arrive at a
445: total gain of 256, 32, 4 or 1 for four energy ranges, 0--50\mev,
446: 50--400\mev, 0.4--3.2\gev, and 3.2--13.0\gev, respectively. The
447: appropriate range is identified by a comparator and the signal is
448: digitized by a 10-bit, 3.7\mhz\ ADC. Data from 24 crystals are
449: multiplexed onto a fiber-optic driver and sent serially at a rate
450: of 1.5\gbps\ across a 30\m-long optical fiber to the ROM. In the
451: ROM, the continuous data stream is entered into a digital
452: pipeline. A correction for pedestal and gain is applied to each
453: sample. The pipeline is then tapped to extract the input to the
454: calorimeter trigger.
455:
456: Upon receipt of the L1 \emph{Accept} signal, data samples within a
457: time window of $\pm$1\mus\ are selected for the feature
458: extraction. Up to now, the calorimeter feature extraction
459: algorithm performs a parabolic fit to the peak of the signal
460: waveform to derive its energy and time. In the future, it is
461: planned to employ a digital filter prior to the signal fit to
462: further reduce noise. For this filter algorithm, the frequency
463: decomposition of an average signal pulse and the typical noise
464: spectrum are measured for all channels and subsequently used to
465: derive an optimum set of weights that maximizes the
466: signal-to-noise ratio. These weights are then applied to
467: individual samples to obtain a filtered waveform.
468:
469:
470: \begin{figure}
471: \begin{center}
472: \includegraphics[width=\columnwidth]{8583A09.eps}
473: \vspace{-1.25pc} \caption{The distribution of equivalent noise
474: energy (ENE) or all channels of the EMC with and without digital
475: filtering. The data were recorded in the absence of beams by a
476: random trigger.} \label{calfig::enoise}
477: \end{center}
478: \end{figure}
479:
480: The magnitude of the electronic noise is measured as the rms width
481: of the pedestal distribution as shown in
482: Figure~\ref{calfig::enoise}. The observed distribution for all
483: channels translates to an ENE of 230\kev\ and 440\kev\ with and
484: without digital filtering; this result is comparable to design
485: expectations. Measurements of the auto-correlation function
486: indicate that the coherent noise component is negligible compared
487: to the incoherent noise, except for regions where the
488: preamplifiers saturate (see below).
489:
490: During data-taking, the data acquisition imposes a single-crystal
491: readout threshold in order to keep the data volume at an
492: acceptable level. This energy threshold is currently set to
493: 1\mev\ and during stable colliding beam conditions on average
494: 1,000 crystals are read out (measured with 600\mA\ of \en\ and
495: 1100\mA\ of \ep\ and a random clock trigger), corresponding to an
496: average occupancy of 16\%. The electronic noise accounts for about
497: 10\%, while the remaining signals originate from beam-generated
498: background (see Section~\ref{sec:ir}). A typical hadronic event
499: contributes signals in 150 crystals.
500:
501:
502: \subsubsection{Electronics Calibration\\ and Linearity}
503:
504: To measure pedestal offsets, determine the overall gain, and to
505: remove non-linearities the FEE are calibrated by precision charge
506: injection into the preamplifier input. Initially, residual
507: non-linearities of up to 12\% in limited regions near each of the
508: range changes were observed and corrected for
509: offline~\cite{nonlin}. These non-linearities were traced to
510: oscillations on the ADC cards that have since been corrected. The
511: correction resulted in markedly improved energy resolution at high
512: energies. Residual non-linearities (typically 2--4\%) arise
513: primarily from cross-talk, impacting both the electronics
514: calibrations and the colliding-beam data. The effect is largest at
515: about ~630\mev\ (950\mev) in a high (low) gain preamplifier
516: channel, inducing a 2\mev\ (6\mev) cross-talk signal in an
517: adjacent channel. The implementation of an energy dependent
518: correction is expected to significantly reduce this small,
519: remaining effect, and lead to a further improvement of the energy
520: resolution.
521:
522: \subsubsection{Electronics Reliability}
523:
524: With the exception of minor cable damage during installation
525: (leaving two channels inoperative), the system of 13,160 readout
526: channels has met its reliability requirements. After the
527: replacement of a batch of failing optical-fiber drivers, the
528: reliability of the digitizing electronics improved substantially,
529: averaging channel losses of less than 0.1\%.
530:
531:
532:
533: \subsection{Energy Calibration}
534:
535: The energy calibration of the EMC proceeds in two steps. First,
536: the measured pulse height in each crystal has to be translated to
537: the actual energy deposited. Second, the energy deposited in a
538: shower spreading over several adjacent crystals has to be related
539: to the energy of the incident photon or electron by correcting for
540: energy loss mostly due to leakage at the front and the rear, and
541: absorption in the material between and in front of the crystals,
542: as well as shower energy not associated with the cluster.
543:
544: The offline pattern recognition algorithm that groups adjacent
545: crystals into clusters is described in detail in
546: Section~\ref{calsec::reco}.
547:
548:
549: \subsubsection{Individual Crystal Calibration}
550: \label{calsec::calcrystal}
551:
552: In spite of the careful selection and tuning of the individual
553: crystals, their light yield varies significantly and is generally
554: non-uniform. It also changes with time under the impact of
555: beam-generated radiation. The absorbed dose is largest at the
556: front of the crystal and results in increased attenuation of the
557: transmitted scintillation light. The light yield must therefore
558: be calibrated at different energies, corresponding to different
559: average shower penetration, to track the effects of the radiation
560: damage.
561:
562: The calibration of the deposited energies is performed at two energies
563: at opposite ends of the dynamic range, and these two measurements are
564: combined by a logarithmic interpolation. A 6.13\mev\ radioactive
565: photon source~\cite{source} provides an absolute calibration at low
566: energy, while at higher energies (3--9\gev) the relation between
567: polar angle and energy of \epm\ from Bhabha events is
568: exploited~\cite{bhabha1}.
569:
570: A flux of low-energy neutrons ($4\times 10^{8}/s$) is used to
571: irradiate Fluorinert~\cite{fluorinert} to produce photons of
572: 6.13\mev\ via the reaction $ ^{19}F + n \to ^{16}N+\alpha$,
573: $^{16}N \to ^{16}O^{*} + \beta$, $^{16}O^{*} \rightarrow ^{16}O +
574: \gamma$. The activated $ ^{16}N $ has a half-life of 7 seconds and
575: thus does not cause radiation damage or long-term activation. The
576: fluid is pumped at a rate of 125\liter/\s\ from the neutron
577: generator to a manifold of thin-walled (0.5\mm) aluminum pipes
578: that are mounted immediately in front of the crystals. At this
579: location, the typical rate of photons is 40\hz/crystal.
580:
581: Figure~\ref{calfig::source} shows a typical source spectrum that was
582: derived from the raw data by employing a digital filter algorithm.
583: For a 30-minute exposure, a statistical error of 0.35\% is obtained,
584: compared to a systematic uncertainty of less than 0.1\%. This
585: calibration is performed weekly.
586:
587: \begin{figure}
588: \centering
589: \includegraphics[height=7.5cm]{sourcepeak.eps}
590: \vspace{-2pc}
591: \caption{A typical pulse-height spectrum recorded with the
592: radioactive source to calibrate the single-crystal energy scale
593: of the EMC. The spectrum shows the primary 6.13\mev\ peak and
594: two associated escape peaks at 5.62\mev\ and 5.11\mev.
595: The solid line represents a fit to the total spectrum, the dotted
596: lines indicate the contributions from the three individual photon
597: spectra.}
598: \label{calfig::source}
599: \end{figure}
600:
601:
602: At high energies, single crystal calibration is performed with a
603: pure sample of Bhabha events~\cite{bhabha1}. As a function of the
604: polar angle of the \epm, the deposited cluster energy is
605: constrained to equal the prediction of a GEANT-based Monte Carlo
606: simulation~\cite{geant}. For a large number of energy clusters, a
607: set of simultaneous linear equations relates the measured to the
608: expected energy and thus permits the determination of a gain
609: constant for each crystal. In a 12-hour run at a luminosity of
610: $3\times 10^{33}$\cms\ some 200 \epm\ per crystal can be
611: accumulated, leading to a statistical error of 0.35\%. This
612: calibration has been performed about once per month, and will be
613: fully automated in the future.
614:
615: \subsubsection{Cluster Energy Correction}
616: The correction for energy loss due to shower leakage and
617: absorption is performed as a function of cluster energy and polar
618: angle. At low energy ($\mathrm{E} < 0.8\gev$), it is derived from
619: $\piz$ decays~\cite{pizero}. The true energy of the photon is
620: expressed as a product of the measured deposited energy and a
621: correction function which depends on $\ln E$ and $\cos\theta$. The
622: algorithm constrains the two-photon mass to the nominal $\piz$
623: mass and iteratively finds the coefficients of the correction
624: function. The typical corrections are of order $6 \pm 1\%$. The
625: uncertainty in the correction is due to systematic uncertainties
626: in the background estimation and the fitting technique.
627:
628: At higher energy ($ 0.8 < E < 9\gev $) the correction is
629: estimated from single-photon Monte Carlo simulations. A second
630: technique using radiative Bhabha events~\cite{radbhabha} is being
631: developed. The beam energy and the precise track momenta of the
632: $e^+$ and $e^-$, together with the direction of the radiative
633: photon, are used to fit the photon energy. This fitted value is
634: compared to the measured photon energy to extract correction
635: coefficients, again as a function of $\ln E$ and $\cos \theta$.
636:
637:
638: \subsection{Monitoring}
639: \subsubsection{Environmental Monitoring}
640:
641:
642: The temperature is monitored by 256 thermal sensors that are
643: distributed over the calorimeter, and has been maintained at
644: $20\pm 0.5$\degc. Dry nitrogen is circulated throughout the
645: detector to stabilize the relative humidity at $1\pm 0.5$\%.
646:
647:
648: \subsubsection{Light-Pulser System}
649:
650: The light response of the individual crystals is measured daily
651: using a light-pulser system~\cite{kocian,pulser}. Spectrally
652: filtered light from a xenon flash lamp is transmitted through
653: optical fibers to the rear of each crystal. The light pulse is
654: similar in spectrum, rise-time and shape to the scintillation
655: light in the CsI(Tl) crystals. The pulses are varied in intensity
656: by neutral-density filters, allowing a precise measurement of the
657: linearity of light collection, conversion to charge,
658: amplification, and digitization. The intensity is monitored
659: pulse-to-pulse by comparison to a reference system with two
660: radioactive sources, $^{241}\mathrm{Am}$ and $^{148}\mathrm{Gd}$,
661: that are attached to a small $\CsI$ crystal that is read out by
662: both a photodiode and a photomultiplier tube. The system is stable
663: to 0.15\% over a period of one week and has proven to be very
664: valuable in diagnosing problems. For example, the ability to
665: accurately vary the light intensity led to the detection of
666: non-linear response in the electronics~\cite{kocian}.
667:
668:
669: \subsubsection{Radiation Monitoring and Damage}
670:
671: \begin{figure}
672: \centering
673: \includegraphics[width=7.0cm]{8583A10}
674: \vspace{-1.5pc}
675:
676: \caption{ Impact of beam-generated radiation on the CsI(Tl)
677: crystals: a) the integrated dose measured with RadFETs placed in
678: front of the crystals, b) the degradation in light yield measured
679: with the radioactive-source calibration system.}
680: \label{calfig::radiation}
681: \end{figure}
682:
683: The radiation exposure is monitored by 56 and 60 realtime
684: integrating dosimeters (RadFETs)~\cite{radfets} placed in front of
685: the barrel and endcap crystals. In Figure~\ref{calfig::radiation},
686: the accumulated dose is compared to the observed loss in
687: scintillation light, separately for the endcap, the forward, and
688: the backward barrel. The dose appears to follow the integrated
689: luminosity, approximately linearly. The light loss is greatest in
690: the forward region corresponding to the area of highest integrated
691: radiation dose. The size of the observed light loss is close to
692: expectations, based on extensive irradiation tests.
693:
694: \subsection{Reconstruction Algorithms}
695: \label{calsec::reco}
696:
697: A typical electromagnetic shower spreads over many adjacent
698: crystals, forming a \emph{cluster} of energy deposits. Pattern
699: recognition algorithms have been developed to efficiently identify
700: these clusters and to differentiate single clusters with one
701: energy maximum from merged clusters with more than one local
702: energy maximum, referred to as a \emph{bumps}. Furthermore, the
703: algorithms determine whether a bump is generated by a charged or a
704: neutral particle.
705:
706: Clusters are required to contain at least one seed crystal with an
707: energy above 10\mev. Surrounding crystals are considered as part of
708: the cluster if their energy exceeds a threshold of 1\mev, or if
709: they are contiguous neighbors (including corners) of a crystal with at
710: least 3\mev. The value of the single crystal threshold is set by the
711: data acquisition system in order to keep the data volume at an
712: acceptable level, given the current level of electronics noise and
713: beam-generated background. It is highly desirable to reduce this
714: threshold since fluctuations in the effective energy loss at the edges
715: of a shower cause a degradation in resolution, particularly at low
716: energies.
717:
718: Local energy maxima are identified within a cluster by requiring
719: that the candidate crystal have an energy, $E_\mathrm{LocalMax}$,
720: which exceeds the energy of each of its neighbors, and satisfy the
721: following condition: $ 0.5 (N-2.5) >
722: E_\mathrm{NMax}/E_\mathrm{LocalMax}$, where $E_\mathrm{NMax}$ is
723: the highest energy of any of the neighboring $N$ crystals with an
724: energy above 2\mev.
725:
726: Clusters are divided into as many bumps as there are local maxima.
727: An iterative algorithm is used to determine the energy of the
728: bumps. Each crystal is given a weight, $w_{i}$, and the bump
729: energy is defined as $E_\mathrm{bump} = \sum_{i} w_{i} E_{i}$,
730: where the sum runs over all crystals in the cluster. For a
731: cluster with a single bump, the result is $w_{i} \equiv 1$. For a
732: cluster with multiple bumps, the crystal weight for each bump is
733: calculated as
734: \begin{equation}
735: w_{i} = E_{i} \frac{\exp(-2.5 r_{i}/r_M)}{\sum_{j} E_{j} \exp(-2.5
736: r_{j}/r_M)},
737: \end{equation}
738: where the index $j$ runs over all crystals in the cluster. $r_M$
739: refers to the Moli\`{e}re radius, and $r_i$ is the distance of the
740: $i$th crystal from the centroid of the bump. At the outset, all
741: weights are set to one. The process is then iterated, whereby the
742: centroid position used in calculating $r_{i}$ is determined from the
743: weights of the previous iteration, until the bump centroid position is
744: stable to within a tolerance of 1\mm.
745:
746: The position of a bump is calculated using a center-of-gravity
747: method with logarithmic, rather than linear
748: weights~\cite{brabson,otto}, $ W_i = 4.0 + \ln
749: {E_i/E_\mathrm{bump}}$, where only crystals with positive weights,
750: \ie\ $E_i > 0.0184 \times E_\mathrm{bump}$, are used in the
751: calculation. This procedure emphasizes lower-energy crystals,
752: while utilizing only those crystals that make up the core of the
753: cluster. A systematic bias of the calculated polar angle
754: originates from the non-projectivity of the crystals. This bias is
755: corrected by a simple offset of $-2.6$\mrad\ for $\theta
756: >90$\degrees\ and $+2.6$\mrad\ for $\theta < 90$\degrees.
757:
758: A bump is associated with a charged particle by projecting a track to
759: the inner face of the calorimeter. The distance between the track
760: impact point and the bump centroid is calculated, and if it is
761: consistent with the angle and momentum of the track, the bump is
762: associated with this charged particle. Otherwise, it is assumed to
763: originate from a neutral particle.
764:
765: On average, 15.8 clusters are detected per hadronic event, of
766: which 10.2 are not associated with charged particle tracks. At
767: current operating conditions, beam-induced background contributes
768: on average 1.4 neutral clusters with energies above 20\mev. This
769: number is significantly smaller than the average number of
770: crystals with energies above 10\mev\ (see Section~\ref{sec:ir}).
771:
772:
773: \subsection{Performance}
774:
775: \subsubsection{Energy Resolution}
776:
777: \begin{figure}
778: \centering
779: \includegraphics[width=6.5cm]{8583A32}
780: \vspace{-2pc}
781: \caption{The ratio of the EMC measured energy to the expected
782: energy for electrons from Bhabha scattering of 7.5\gevc. The
783: solid line indicates a fit using a logarithmic function.}
784: \label{calfig::bhabha}
785: \end{figure}
786:
787:
788:
789: \begin{figure}
790: \centering
791: \includegraphics[width=6.5cm]{8583A41}
792: \vspace{-2pc} \caption{The energy resolution for the ECM measured
793: for photons and electrons from various processes. The solid curve
794: is a fit to Equation~\ref{caleqn::res} and the shaded area denotes
795: the rms error of the fit.} \label{calfig::eres}
796: \end{figure}
797:
798: At low energy, the energy resolution of the EMC is measured
799: directly with the radioactive source yielding $\sigma_{E}/E = 5.0
800: \pm 0.8\%$ at 6.13\mev\ (see Figure~\ref{calfig::source}). At
801: high energy, the resolution is derived from Bhabha scattering,
802: where the energy of the detected shower can be predicted from the
803: polar angle of the \epm. The measured resolution is $\sigma_{E}/E
804: = 1.9 \pm 0.07\%$ at 7.5\gev\ (see Figure~\ref{calfig::bhabha}).
805: Figure~\ref{calfig::eres} shows the energy resolution extracted
806: from a variety of processes as a function of energy. Below 2\gev,
807: the mass resolution of $\piz$ and $\eta$ mesons decaying into two
808: photons of approximately equal energy is used to infer the EMC
809: energy resolution ~\cite{pizero}. The decay $\chi_{c1} \to \jpsi
810: \gamma$ provides a measurement at an average energy of about
811: 500\mev, and measurements at high energy are derived from Bhabha
812: scattering. A fit to the energy dependence results in
813: \begin{equation}
814: \frac{\sigma_{E}}{E}= \frac{(2.32 \pm 0.30)\%}{^{4}\sqrt{E
815: (\gev)}} \oplus (1.85 \pm 0.12)\%. \label{caleqn::resfit}
816: \end{equation}
817: Values of these fitted parameters are higher than the somewhat
818: optimistic design expectations, but they agree with detailed Monte
819: Carlo simulations which include the contributions from electronic
820: noise and beam background, as well as the impact of the material
821: and the energy thresholds.
822:
823:
824: \subsubsection{Angular Resolution}
825:
826: \begin{figure}
827: \centering
828: \includegraphics[width=6.5cm]{8583A42}
829: \vspace{-2pc} \caption{The angular resolution of the EMC for
830: photons from \piz\ decays. The solid curve is a fit to
831: Equation~\ref{caleqn::posres}.} \label{calfig::pres}
832: \end{figure}
833:
834: The measurement of the angular resolution is based on the analysis
835: of \piz\ and $\eta$ decays to two photons of approximately equal
836: energy. The result is presented in Figure~\ref{calfig::pres}. The
837: resolution varies between about 12\mrad\ at low energies and
838: 3\mrad\ at high energies. A fit to an empirical parameterization
839: of the energy dependence results in
840: \begin{eqnarray}
841: \sigma_{\theta} &= &\sigma_{\phi}\nonumber\\
842: &= &(\frac{3.87 \pm
843: 0.07}{\sqrt{E (\gev)}}~+~0.00 \pm 0.04)~\mrad.
844: %\label{caleqn::posres}
845: \end{eqnarray}
846: These fitted values are slightly better than would be expected
847: from detailed Monte Carlo simulations.
848:
849:
850: \subsubsection{$\piz$ Mass and Width}
851:
852: \begin{figure}
853: \centering
854: \includegraphics[width=6.5cm]{pi0_egam30_epi0300.eps}
855: \vspace{-2pc} \caption{Invariant mass of two photons in \BB\
856: events. The energies of the photons and the \piz\ are required to
857: exceed 30\mev\ and 300\mev, respectively. The solid line is a fit
858: to the data.} \label{calfig::pi0}
859: \end{figure}
860:
861:
862: Figure~\ref{calfig::pi0} shows the two-photon invariant mass in
863: \BB\ events. The reconstructed $\piz$ mass is measured to be
864: 135.1\mevcc\ and is stable to better than 1\% over the full photon
865: energy range. The width of 6.9\mevcc\ agrees well with the
866: prediction obtained from detailed Monte-Carlo simulations. In
867: low-occupancy $\tau^{+}\tau^{-}$ events, the width is slightly
868: smaller, 6.5\mevcc, for \piz\ energies below 1\gev. A similar
869: improvement is also observed in analyses using selected isolated
870: photons in hadronic events.
871:
872:
873:
874: \subsubsection{Electron Identification}
875:
876:
877: Electrons are separated from charged hadrons primarily on the
878: basis of the shower energy, lateral shower moments, and track
879: momentum. In addition, the \dedx\ energy loss in the DCH and the
880: DIRC Cherenkov angle are required to be consistent with an
881: electron. The most important variable for the discrimination of
882: hadrons is the ratio of the shower energy to the track momentum
883: $(E/p)$. Figure~\ref{calfig::electronpure} shows the efficiency
884: for electron identification and the pion mis-identification
885: probability as a function of momentum for two sets of selection
886: criteria. The electron efficiency is measured using radiative
887: Bhabhas and $e^+ e^- \to e^+ e^- e^+ e^- $ events. The pion
888: misidentification probability is measured for selected charged
889: pions from \KS\ decays and three-prong $\tau$ decays. A tight
890: (very tight) selector results in an efficiency plateau at 94.8\%
891: (88.1\%) in the momentum range $0.5 < p < 2\gevc$. The pion
892: misidentification probability is of order 0.3\% (0.15\%) for the
893: tight (very tight) selection criteria.
894:
895: \begin{figure}
896: \centering
897: \includegraphics[width=6.5cm]{8583A43}
898: \vspace{-2pc} \caption{The electron efficiency and pion
899: mis-identification probability as a function of a) the particle
900: momentum and b) the polar angle, measured in the laboratory
901: system.} \label{calfig::electronpure}
902: \end{figure}
903:
904:
905: \subsection{Summary}
906:
907: The EMC is presently performing close to design
908: expectations. Improvements in the energy resolution are expected from
909: the optimization of the feature-extraction algorithms designed to
910: further reduce the electronics noise. Modifications to the
911: electronics should allow for more precise calibrations. The expected
912: noise reduction should permit a lower single-crystal readout
913: threshold. However, this decrease in noise might be offset by an
914: increase in the beam background that is expected for higher
915: luminosities and beam currents.
916: