Dual-readout calorimetry

In the past 20 years, dual-readout calorimetry has emerged as a technique for measuring the properties of high-energy hadrons and hadron jets that offers considerable advantages compared with the instruments that are currently used for this purpose in experiments at the high-energy frontier. The status of this experimental technique and the challenges faced for its further development are reviewed.

Figure 1

The energy dependence of the average em shower fraction measured for (a) copper- and lead-based sampling calorimeters, and (b) the event-to-event fluctuations in this fraction for showers induced by 150 GeV ${\pi }^{-}$ in lead. Experimental data from [3] (lead) and [8] (copper).

Figure 2

Illustration of the meaning of the $e/h$ and $e/\mathrm{m}\mathrm{i}\mathrm{p}$ values of a (generic) calorimeter. Shown are distributions of the signal per unit deposited energy for the electromagnetic and non-em components of hadron showers. These distributions are normalized to the response for minimum ionizing particles (“mip”). The average values of the em and non-em distributions are the em response (“$e$”) and non-em response (“$h$”), respectively.

Figure 3

Hadronic energy resolution in the TeV domain, calculated with Eq. (3). From [68].

Figure 4

The response of the uranium-based ZEUS calorimeter to low-energy hadrons. For comparison, the $e/\mathrm{mip}$ ratio for a copper-based calorimeter is shown as well. From [29].

Figure 5

Distribution of the fraction of the energy released by hadronically decaying (a) $Z^0$ and (b) $H^0$ bosons at rest that is carried by charged final-state particles with a momentum less than $5\mathrm{GeV}/c$. See text for details. From [65].

Figure 6

Results of tests of the dual-readout ACCESS calorimeter with 375 GeV pions. (a) Scatter plot of the signals recorded in the quartz fibers vs those in the scintillating fibers. The signal distributions from the scintillating fibers for (b) all events and for subsets of events with (c) a small or (d) average fraction of Čerenkov light. From [57].

Figure 7

Graphic representation of Eqs. (7) ([59]; [49]). The data points for hadron showers detected with a dual-readout calorimeter are located around the straight (red) line in this diagram. The data points for em showers in this calorimeter are clustered around the point where this line intersects the $C=S$ line, i.e., point (1,1). See text for further details.

Figure 8

The $S-C$ diagram of the signals from a (generic) dual-readout calorimeter ([49]). The hadron events are clustered around the straight (red) line, the electron events around point (1,1). Experimental signal distributions measured in the scintillation and Čerenkov channels for 200 GeV “jets” with the DREAM fiber calorimeter ([11]) are shown as well. Also shown is a typical (Čerenkov) response function measured for electrons in DREAM.

Figure 9

The $S-C$ diagram of the signals from a (generic) dual-readout calorimeter ([49]). The hadron events are clustered around the straight (red) line, the electron events around point (1,1). Experimental signal distributions measured in the scintillation and Čerenkov channels for 200 GeV “jets” with the DREAM fiber calorimeter, after applying the dual-readout transformation (10) are shown as well ([11]).

Figure 10

The $S-C$ diagram of the signals from a (generic) dual-readout calorimeter. The hadron events are clustered around the straight (red) line. Data points for protons and pions have different distributions, reflecting differences in the em shower fraction. From [49].

Figure 11

The $S-C$ diagram of the signals from a generic dual-readout calorimeter. The hadron events are clustered around the straight (red) line. Also shown is the effect of a rotation of this red line and the associated distribution of data points. From [49].

Figure 12

Signal distributions of the RD52 dual-readout lead-fiber calorimeter for 60 GeV pions. (a) Scatter plot of the two types of signals as recorded for these particles and (b) rotated over an angle $\theta =30°$ around the point where the two lines from (a) intersect. (c) Projection of the latter scatter plot on the $x$ axis. From [50].

Figure 13

(a) Scatter plots of the Čerenkov vs the scintillation signals from showers induced by monoenergetic hadrons. The arrow indicates the precision with which the em shower fraction, and thus the energy, of an individual particle can be determined on the basis of the measured ratio of the Čerenkov and scintillation signals, 0.7 in this example. The rotation procedure for an ensemble of monoenergetic pions leads to the scatter plot shown in (b). The precision of the measurement of the width of that distribution is indicated by a white arrow as well. From [50].

Figure 14

The DREAM detector. The basic building block is an extruded hollow copper rod. Seven optical fibers (four Čerenkov and three scintillating fibers) are inserted in the central hole. The two types of fibers are split into separate bunches as they exit the downstream detector end. The hexagonal readout structure is indicated. The Čerenkov fibers of the central tower and the six towers of the inner ring were filled with quartz fibers; in the 12 towers of the outer ring clear polymethyl methacrylate (PMMA) fibers were used for this purpose. From [11].

Figure 15

The DREAM calorimeter. (a) The fiber bunches exiting from the rear face of the detector and (b) a picture taken from the front face while the rear end was illuminated. From [11].

Figure 16

Signal distributions for 100 GeV ${\pi }^{-}$ recorded by the (a) scintillating and (b) Čerenkov fibers of the DREAM calorimeter, and (c) a scatter plot showing the correlation between both types of signals. The signals are expressed in the same units as those for em showers, which were used to calibrate the calorimeter (em GeV). From [11].

Figure 17

(a) The relationship between the ratio of the Čerenkov and scintillation signals from the DREAM calorimeter and the electromagnetic shower fraction, derived for the 100 GeV ${\pi }^{-}$ events on the basis of Eq. (7). The total Čerenkov signal distribution (b) for these events and (c) distributions for subsamples of events selected on the basis of the measured $f_{\mathrm{em}}$ value. From [11].

Figure 18

(a) The average Čerenkov and (b) scintillation signals for 200 GeV “jets” in the DREAM calorimeter as a function of the em shower fraction $f_{\mathrm{em}}$. From [11].

Figure 19

Effects of the dual-readout method applied on the basis of the observed Čerenkov-to-scintillation signal ratio. Čerenkov signal distributions for high-multiplicity 200 GeV “jets” in the DREAM calorimeter (a) before and (b) after the dual-readout method was applied. (c) The energy resolution for multiparticle “jets,” measured separately with the scintillation and Čerenkov signals, and after applying the dual-readout method, both for single pions and for multiparticle “jets.” (d) The electron response, which was the basis for calibrating the calorimeter signals, is shown as well. From [11].

Figure 20

Average values of the scintillation and Čerenkov signals from muons traversing the DREAM calorimeter as a function of the muon energy. Also shown is the difference between these signals. All values are expressed in units of GeV, as determined by the electron calibration of the calorimeter. From [10].

Figure 21

Principle of the asymmetry measurement that was used to establish the contribution of Čerenkov light to the signals from the ${\mathrm{PbWO}}_4$ crystals. Depending on the crystal orientation, this directionally emitted light contributed differently to the signals from the left and right photomultiplier tubes. From [13].

Figure 22

Left-right response asymmetry measured for 10 GeV electrons showering in a $2.5X_0$ thick ${\mathrm{PbWO}}_4$ crystal, as a function of the orientation of the crystal (the angle $\theta$). Results are shown for the early and late components of the showers. The latter measurements were done by placing 4 cm of lead upstream of the crystal. From [13].

Figure 23

Average time structure of the signals from a ${\mathrm{PbWO}}_4$ crystal doped with 1% Mo, generated by 50 GeV electrons. The angle $\theta$ was 30° in these measurements. Shown are the results obtained with UV and yellow filters, respectively. From [16].

Figure 24

The time structure of a typical shower signal from 50 GeV electrons measured in the BGO em calorimeter equipped with a UV filter. These signals were measured with a sampling oscilloscope, which took a sample every 0.8 ns ([17]). The UV signals were used to measure the relative contributions of scintillation light (gate 2) and Čerenkov light (gate 1).

Figure 25

Average time structure of the signals generated by 180 GeV ${\pi }^{+}$ traversing a BSO crystal in its center at $\theta =30°$ and passing through a U330 optical transmission filter, followed by a polarization filter. The transmission axis of the latter filter was oriented either (a) horizontally or (b) vertically. The time scale describes the time passed since the start of the time base of the oscilloscope. From [21].

Figure 26

(a) The Čerenkov signal distribution for 200 GeV “jet” events detected in the $\mathrm{BGO}+\mathrm{fiber}$ calorimeter system, together with (b) the distributions for subsets of events selected on the basis of the ratio of the total Čerenkov and scintillation signals in this detector combination. From [17].

Figure 27

(a) The scintillation signal distribution for 50 GeV electrons traversing a ${\mathrm{PbWO}}_4$ crystal at $\theta =30°$ and (b) the fractional width of the Čerenkov signal distribution as a function of the amount of energy deposited in the crystal, as derived from the scintillation signal. The crystal was doped with 0.3% Mo. See the text for more details. From [19].

Figure 28

Scatter plots showing the correlation between the kinetic energy carried by soft neutrons ($E<20\mathrm{MeV}$) and the nuclear binding energy lost when 5 GeV ${\pi }^{-}$ mesons are absorbed in depleted uranium (${}^{238}\mathrm{U}$). The distributions are shown separately for neutrons that originated from (a) evaporation processes and (b) nuclear spallation. Results from Monte Carlo simulations with the hetc/morse package. From [36].

Figure 29

The average time structure of the Čerenkov and scintillation signals measured for the showers from 200 GeV “jets” in the DREAM tower located on the shower axis. (a) The measured (oscilloscope) signals have been inverted. (b) Event-by-event distribution of the fraction of this scintillation signal attributed to neutrons. From [18].

Figure 30

(a) Scatter plot for 200 GeV “jets” in the DREAM calorimeter. For each event, the combination of the total Čerenkov-to-scintillation signal ratio and the fractional contribution of neutrons to the total scintillation signal is represented by a dot. (b) Distribution of the total Čerenkov signal for 200 GeV “jets” and the distributions for three subsets of events selected on the basis of the fractional contribution of neutrons to the scintillation signal. From [18].

Figure 31

(a) The energy resolution measured for the Čerenkov signals from 200 GeV “jets” with the same em shower fraction, as a function of the fractional neutron contribution to the scintillation signals. (b) Čerenkov signal distribution for 200 GeV “jets” with $0.70<C/S<0.75$ and $0.045<f_n<0.065$, together with the results of a Gaussian fit. Experimental data obtained with the DREAM calorimeter. From [18].

Figure 32

The structure of the RD52 fiber calorimeter (copper-based modules), compared to that of two other fiber calorimeters: DREAM ([11]) and SPACAL ([1]). From [24].

Figure 33

Front (top) and rear (bottom) view of one of the RD52 fiber calorimeter modules. The tower structure is made visible by shining light on two of the eight fiber bunches sticking out at the back end. See text for more details.

Figure 34

(a), (b) Comparison of the em response functions measured with the RD52 copper-fiber calorimeter ([24]) and (c), (d) the original DREAM copper-fiber calorimeter ([12]), for 40 GeV electrons. The angle of incidence of the electron beam with the fiber axis $(\theta ,\varphi )$ was $(1.5°,1.0°)$ for RD52 and $(3°,2°)$ for DREAM. Results are given separately for the scintillation and Čerenkov signals.

Figure 35

The em energy resolution measured with the Čerenkov fibers, the scintillating fibers, and the sum of all fibers in the RD52 copper-fiber calorimeter. From [24].

Figure 36

The signal linearity for electron showers, measured for the (a) RD52 copper and (b) lead modules. From [24].

Figure 37

Effects of four different shower characteristics that may be used to distinguish between electron and hadron showers in the longitudinally unsegmented RD52 fiber calorimeter. Shown are (a) the fraction of the total signal recorded by the tower in which the particle entered, (b) the ratio of the Čerenkov and scintillation signals of the event, (c) the starting time of the signal in the PMT, measured with respect to an upstream trigger signal, and (d) the ratio of the total integrated charge and the amplitude of the signal. Data obtained with 60 GeV particle beams. From [25].

Figure 38

The RD52 fiber calorimeter installed in the H8C beam area at CERN. The system of trigger counters and beam defining elements is visible in the left bottom part of the figure. The calorimeter is surrounded on four sides by “leakage counters,” the layout of which is shown in the bottom left inset. The other insets show the front face of the (lead)-fiber calorimeter (top left) and the tower structure of the readout (bottom right). From [50].

Figure 39

Signal distributions for ${\pi }^{-}$ beam particles of 20, 60, and 100 GeV showering in the RD52 lead-fiber calorimeter. The top row (a), (c), and (e) shows the signal distributions measured for the scintillation ($S$) and Čerenkov ($C$) signals. The $S$ signals are, on average, larger, and their distribution is less asymmetric.The bottom row (b), (d), and (f) shows the signal distributions that were obtained after combining the $S$ and $C$ distributions according to Eq. (10) with $\chi =0.45$. From [40].

Figure 40

The hadronic energy resolution of the RD52 lead-fiber dual-readout calorimeter for single pions. Shown are the results for the Čerenkov signals alone, and for the dual-readout signals, obtained with Eq. (10). From [50].

Figure 41

geant4 simulations of the response function to 100 GeV ${\pi }^{-}$ particles of a dual-readout fiber calorimeter with the RD52 structure and lateral dimensions of $65\times 65{\mathrm{cm}}^2$ ([23]). Results are shown for the (a) standard ftfp_bert hadronic shower simulation package, and with the (b) high-precision version of this package ftfp_bert_hp. From [70].

Figure 42

Scatter plots of the Čerenkov vs the scintillation signals from showers induced by (a) 80 GeV ${\pi }^{+}$ and (c) 80 GeV protons in the RD52 lead-fiber calorimeter. Projection of the rotated scatter plots on the $x$ axis for the (b) pions and (d) protons. The rotation procedure was identical to that used for 60 GeV ${\pi }^{-}$ (Fig. 12). From [50]

Figure 43

Signal distributions from the RD52 lead-fiber calorimeter for (a) 20 GeV ${\pi }^{+}$, (b) 125 GeV protons, and (c) 125 GeV multiparticle events obtained with the rotation method described in the text. The energy scale is set by electrons showering in this detector. From [50].

Figure 44

(a) The average calorimeter signal per GeV and the (b) fractional width of the signal distribution as a function of energy for single pions and multiparticle events (“jets”). Results are given for the RD52 dual-readout calorimeter signals obtained with the rotation method. From [50].

Figure 45

The lateral profile of 100 GeV electron showers in the RD52 lead-fiber calorimeter, measured with the scintillation signals. From [24].

Figure 46

The energy resolution measured for 20 GeV electrons in the scintillation and Čerenkov channels of the RD52 copper-fiber calorimeter as a function of the azimuth angle of incidence ($\varphi$) of the beam particles. The tilt angle $\theta$ was 1°. From [37].

Figure 47

The average signals for 20 GeV electrons in (a) the scintillation and (b) the Čerenkov channels of the RD52 copper-fiber calorimeter, as a function of the azimuth angle of incidence ($\varphi$) of the beam particles. The tilt angle $\theta$ was 1°, and the preshower detector was in the beam line. The shaded area represents a variation of $\pm 1\%$ with respect to the average signal. From [37].

Figure 48

Dependence of the starting time of the PMT signals on the average depth ($z$) inside the calorimeter where the light is produced (the dash-dotted line). This time is measured with respect to the moment the particles entered the calorimeter. Also shown are the time it takes the particles to travel to $z$ (the dashed line) and the time it takes the light to travel from $z$ to the PMT (the dotted line). From [25].

Figure 49

The measured distribution of the starting time of the DREAM calorimeter’s scintillation signals produced by (a) 60 GeV electrons and (b) 60 GeV pions. This time is measured with respect to the moment the beam particle traversed trigger counter $T1$, installed upstream of the calorimeter. (c) These data were used to determine the distribution of the average depth at which the light was produced in the hadron showers. The horizontal red line is an error bar that represents the precision with which the (average) depth of the light production in an individual event can be determined. From [25].

Figure 50

Light attenuation in the fibers. (a) The scatter plot shows the calorimeter signal for Čerenkov light from 80 GeV ${\pi }^{-}$ vs the average depth at which that light was produced inside the DREAM calorimeter. This depth was determined from the timing information. (b) The average signal, as a function of depth, provides the effective light attenuation curve of the fibers. From [25].

Figure 51

Average time structure of the signals measured in leakage counters surrounding the RD52 lead-fiber calorimeter for 40 GeV ${\pi }^{-}$ steered into the center of this calorimeter. (a) The signals measured in a counter located close to the shower maximum (not far from the front face of the calorimeter) and in a counter located near the shower tail, i.e., about 2 m from the front face of the calorimeter. (b) The signal from the upstream counter is unfolded into a “neutron” and a “prompt” component. From [70].

Figure 52

Energy resolutions reported for the detection of (a) electrons and (b) hadrons by RD52 ([24]; [50]), SPACAL ([1]), and ZEUS ([33]). From [49].

Figure 53

Distributions of the ratio of (a) the non-em energy and the nuclear binding energy loss and the ratio of (b) the total kinetic energy carried by neutrons and the nuclear binding energy loss for showers generated by 50 GeV ${\pi }^{-}$ in a massive block of copper. The limits on the hadronic energy resolution derived from the correlation between the nuclear binding energy losses and the parameters measured in dual-readout or compensating calorimeters as a function of the particle energy. The straight lines represent resolutions of $20\%/\sqrt{E}$ and $10\%/\sqrt{E}$ and are intended for reference purposes. Results from geant4 Monte Carlo simulations of pion showers developing in a massive block of (c) copper or (d) lead. From [49].