The next generation of crystal detectors

In high-energy and nuclear physics experiments, total absorption electromagnetic calorimeters made of inorganic crystals are known for their superb energy resolution and detection efficiency for photon and electron measurements. A crystal calorimeter is thus the choice for those experiments where precision measurements of photons and electrons are crucial for their physics missions. It is also known that the existing crystal detectors are neither bright nor fast enough nor radiation hard enough to survive severe radiation environment expected in future HEP experiments. Crystal detectors have also been proposed to build a homogeneous hadron calorimeter to achieve unprecedented jet mass resolution by duel readout of both Cherenkov and scintillation light, where the development of cost-effective crystal detectors is a crucial issue because of the 100 cubic meters crystal volume required. To develop novel inorganic crystal scintillator-based detector concepts for future HEP experiments at the energy and intensity frontiers. Optical and scintillation properties of novel inorganic crystal scintillators, such as excitation, emission and transmittance spectra, light output and decay time, are characterized before and after irradiation by ionization dose and hadrons. Their performance and radiation hardness are compared to the requirements, and feedback is given to the crystal manufacturers for quality improvement. As a result of this investigation, several inorganic crystal scintillator-based detector concepts are established for future HEP experiments, such as an LSO/LYSO crystal-based total absorption and/or sampling calorimeter concept, a barium fluoride crystal-based very fast crystal calorimeter concept, and a cost-effective inorganic scintillator-based homogeneous hadron calorimeter concept. Bright, fast and radiation hard LYSO/LSO crystals may be used for a total absorption ECAL. An LYSO/W Shashlik sampling calorimeter will survive the harsh radiation environment expected at the HL-LHC. With sub-ns decay time of its fast scintillation component and excellent radiation hardness, barium fluoride crystals would provide more than ten times faster rate and timing capability, provided that their slow scintillation component is effectively suppressed to avoid pileup. PbF2, PbFCl and BSO crystals may provide a foundation for a homogeneous hadron calorimeter with dual readout for both Cherenkov and scintillation light to achieve unprecedented jet mass resolution for future lepton colliders.


Introduction
Crystal detectors have been used widely in high-energy and nuclear physics experiments, medical instruments and homeland security applications. Novel crystal detectors are continuously being discovered and developed in academia and industry.
In high-energy physics (HEP) and nuclear physics (NP) experiments, total absorption electromagnetic calorimeters (ECALs) made of inorganic crystals are known for their superb energy resolution and detection efficiency for pho-R.-Y. Zhu ton and electron measurements [1]. A crystal ECAL is thus the choice for those experiments where precision measurements of photons and electrons are crucial for their physics missions. Examples are the Crystal Ball NaI:Tl ECAL, the L3 BGO ECAL and the BaBar CsI:Tl ECAL in lepton colliders, the kTeV CsI ECAL and the CMS PWO ECAL in hadron colliders and the Fermi CsI:Tl ECAL in space. Table 1 lists existing crystal calorimeters in high-energy physics. For future HEP experiments at the energy and intensity frontiers, however, the crystal detectors used in the abovementioned ECALs are either not bright and fast enough, or not radiation hard enough. Crystals have also been proposed to build a homogeneous hadron calorimeter (HHCAL) to achieve unprecedented jet mass resolution by duel readout of both Cherenkov and scintillation light [2], where development of cost-effective crystal detectors is a crucial issue because of the huge crystal volume required [3]. This paper discusses several R&D directions for the next-generation crystal detectors for future HEP experiments. Table 1 shows that the CMS lead tungstate (PbWO 4 or PWO) crystal calorimeter, consisting of 76,000 crystals of 11 m 3 , is the largest crystal calorimeter ever built. Because of its superb energy resolution and detection efficiency, the CMS PWO ECAL has played an important role for the discovery of the Higgs boson by the CMS experiment [4]. One crucial issue is crystal's radiation damage in the severe radiation environment at LHC, which requires precision monitoring to correct variations of crystal transparency [5]. After 2 years of operation, up to 70% loss of light output was observed in the CMS PWO crystals at large rapidity in situ at LHC as shown in Fig. 1 when the LHC was running at a luminosity of Fig. 1 Monitoring response of CMS PWO crystals observed during LHC run I up to 5 × 10 33 cm −2 s −1 and a half of its designed energy [6]. The damage in PWO crystals increases when the luminosity increases, and recovers during LHC stops.

Performance of PWO crystals
The radiation damage of PWO crystals shown in Fig. 1 is well understood. It is caused by radiation induces absorption, or color center formation, and is dose rate dependent [7]. Figure 2 shows that light output of a PWO crystal reached an equilibrium during irradiation under a defined dose rate, indicating a dose rate-dependent radiation damage caused by color center dynamics. At equilibrium the speed of the color center formation (damage) equals to the speed of the color center annihilation (recovery), so that the color center density, or the radiation-induced absorption, does not change unless the applied dose rate changes [7]. Figure 3 shows emission-weighted radiation-induced absorption coefficient (EWRIAC) as a function of the γray dose rate measured for a batch of mass-produced BTCP PWO crystals [8]. The EWRIAC values of all crystals are less    Figure 4 shows the expected light output loss as a function of luminosity for PWO crystals at different rapidity predicted in 2010 before LHC running [9]. This prediction was made according to the measured relation between light output and radiation-induced absorption assuming the average EWRIAC values shown in red color in Fig. 3. A detailed comparison shows that this prediction agrees well with the data shown in Fig. 1, indicating that radiation damage in PWO crystals observed so far is caused by ionization dose.
Additional damage caused by charged hadrons was also studied [10]. Because of these damages, the CMS endcap PWO ECAL is proposed to be replaced by using more radiation hard technologies [11], one of which is a LYSO crystal-based Shashlik sampling calorimeter [12].

Radiation hard LYSO/LSO crystals
Because of their high density (7.4 g cm −3 ), short radiation length (1.14 cm), fast (40 ns) and bright (4 times BGO) scintillation, cerium-doped lutetium oxyorthosilicate (Lu 2 SiO 5 :Ce, LSO) [13] and lutetium yttrium oxyorthosilicate (Lu 2(1−x) Y 2x SiO 5 :Ce, LYSO) [14,15] crystals have attracted a broad interest in the high-energy physics community pursuing precision electromagnetic calorimeter for future high-energy physics experiments [12,[16][17][18][19]. Their excellent radiation hardness against gamma rays [15,20], neutrons [17] and charged hadrons [10] also makes them a preferred material for calorimeters to be operated in a severe radiation environment, such as the HL-LHC.  The top plot of Fig. 5 shows the photoluminescence spectra of a LYSO crystal measured before (blue) and after (red) 1 Mrad γ -ray irradiation [20]. To facilitate a comparison, these spectra were normalized to the area between 380 and 460 nm under the spectra. The corresponding relative difference is shown in the bottom plot of Fig. 5. The bin-by-bin average of the absolute difference between the spectra mea-  sured before and after the irradiation is found to be 0.6% in the normalization region, less than the systematic uncertainty for this measurement. This indicates that the γ -ray irradiation does not affect the scintillation mechanism in LYSO crystals. Figure 6 shows the longitudinal transmittance spectra for five 200-mm-long LYSO crystals from different vendors before and after several steps of the γ -ray irradiation   EWLT represents crystal's transparency better than the transmittance at the emission peak since it runs through the entire emission spectrum. This is particularly important for crystals with self-absorption nature, i.e., a part of scintillation   is self-absorbed in the crystal [17]. Consistent damages are observed on the longitudinal transmittance of all LYSO crystal samples. It is noted that the degradation of longitudinal transmittance is small even after 10 6 rad. Figure 7 shows light output (LO) and light response uniformity (LRU) for a 280-mm-long LYSO crystal before and after several steps of the γ -ray irradiations with integrated dose of 10 2 , 10 4 and 10 6 rad [21]. About 14% LO loss is    Figure 9 shows a LYSO crystal-based Shashlik calorimeter detector concept with four evenly distributed wavelength shifting (WLS) fibers for readout and a monitoring fiber at the center [12]. This detector concept reduces the crystal volume and cost, and improves the radiation hardness of the calorimeter because of the much reduced light path. As shown in Fig. 9, the initial design consists of 30 LYSO plates of 1.5 mm thick and 29 W plates of 2.5 mm thick. Each tower has a depth of 25 X 0 to accommodate electrons and photons with energies up to the TeV range. The sampling fraction was chosen to be around 20% to provide an adequate stochastic term of the energy resolution at a level of 10%. Because of the high density of both the LSO/LYSO crystals and the absorber materials, the average radiation length (0.51 cm) and Moliere radius (1.3 cm) are much smaller as compared to commonly used crystal scintillators. This detector concept thus also provides a very compact calorimeter to mitigate the pileup effect expected at the HL-LHC. Figure 10 shows a photograph of LYSO/W Shashlik tower with four Y-11 WLS fibers coupled to a readout PMT. Beam tests for a LYSO/W Shashlik matrix consisting of sixteen towers were carried out at Fermilab and CERN with result reported in the Proceedings of the Calor 2016 conference at Daegu, South Korea. Figure 11 shows a setup used to measure longitudinal response uniformity for the LYSO/W Shashlik detector concept. A 14 × 14 × 1.5 mm 3 LYSO plate was moved along four Y-11 WLS fibers at five points of 2 cm apart. The LO was measured by using a γ -ray source shooting to the LYSO plate. Figure 12 shows the measured result and the corresponding fit for two straight lines using first and last three points. The corresponding slope of the fit is 0.2%/ X 0 at the front, which is less than 0.3%/ X 0 required to maintain good energy resolution [22]. The slope at the back is 0.7%/ X 0 , which is close to the optimum of 8% rise required in the last 10 X 0 .

Alternative fast crystals
The high cost of LYSO crystals caused by high Lu 2 O 3 price, however, may limit their use in future HEP experiments. Table 2 lists basic optical scintillation properties for alternative fast crystal detectors with scintillation decay time ranged from sub-nanosecond to a few tens nanosecond, and compared to plastic scintillator. Among the fast crystals listed in Table 2, mass production cost of barium fluoride (BaF 2 ) and pure CsI crystals is significantly lower than others because of their low raw material cost and low melting point. At this point, BaF 2 is baselined for the Mu2e experiment with pure CsI as an alternative option [23]. For applications in severe radiation environment, such as HL-LHC, one of the crucial issues for their application is radiation hardness. Investigation of radiation hardness for alternative fast crystals would provide important input for future HEP experiments at the energy and intensity frontiers. On the other hand, costeffective fast scintillating glass and ceramics may also be considered and developed. Figure 13 shows normalized emission-weighted longitudinal transmittance (EWLT, top) and light output (LO, bottom) as a function of integrated γ -ray dose for three 250-mm-long BaF 2 crystals. While SIC2012 and BGRI2012 were grown by SIC and Beijing Glass Research Institute (BGRI), respectively, in 2012, the crystal S302 was grown by SIC 20 years ago during nineties for the SSC. The result shows that SIC2012 is more radiation hard than other samples, indicating an improvement in crystal quality. It was also found that the slow component (300 nm) of all crystals is more radiation hard than the fast component (220 nm). BaF 2 crystals also show stable damage after 10 krad, indicating limited defect density in these crystals which was fully exhausted after 10 krad. This promises a stable BaF 2 crystal calorimeter in a severe radiation environment. Figure 14 shows radiation-induced absorption coefficient (RIAC) as a function of integrated dose for three BaF 2 crystals, indicating that RIAC of mass-produced BaF 2 may be controlled to less than 1.6 m −1 for the fast component. Note, RIAC is independent of sample size, so can be compared to other crystals. Figure 15 shows normalized EWLT and LO as a function of integrated γ -ray dose for large size CsI crystals. SIC2013 (30 cm long) and SIC2011 were grown by SIC in 2013 and 2011, respectively. The crystals CsI-3 (20 cm long) and CsI-4 (19 cm long) were grown by the Institute for Single Crystals, Kharkov, Ukraine. Consistent radiation damage between these crystals is observed. Figure 15 also shows that radiation damage in CsI is small below 10 krad, but degrades continuously with no sign of saturation at high dose, indicating high defect density in the crystals. Figure 16 shows RIAC as a function of integrated dose, which is longer than 3 m −1 after 1 Mrad. Figures 17 and 18 show normalized EWLT, LO and RIAC as a function of integrated dose for crystals with dose rateindependent radiation damage, i.e., BaF 2 , pure CsI and LYSO. Because of no recovery, these crystals promise a more stable calorimeter than that with dose rate-dependent radiation damage, such as PWO. LYSO crystals show clearly the best radiation hardness. Among two crystals of low cost, the radiation hardness of BaF 2 is good at high dose and that of CsI is good at low dose.

Time resolution of crystal scintillation
Crystal time resolution is important for many applications. It depends on the signal-to-noise ratio for the rise time mea-  surement. While scintillation light is known to have very different decay time for various crystals, the intrinsic rising time of most crystals is as fast as tens ps [24]. Figure 19 shows the rising time measured by using a Hamamatsu R2059 PMT for ten crystal samples of 1.5 X 0 size. The fast rise time of about 1.5 ns observed for BaF 2 , LYSO, CeF 3 and BGO is dominated by the PMT rise time 1.3 ns (2500 V) and the rise time of 0.14 ns of the Agilent MSO9254A (2.5 GHz) DSO. The measured rise time values are also reduced for the same crystal with black wrapping, indicating effect of the light propagation in the crystal [25]. We notice a 3 ns FWHM pulse width from BaF 2 crystal is the narrowest among all crystal scintillators, which may be reduced by using a aster photodetector. Table 3 lists the values of the figure of merit for time resolution for various crystal detectors, which is defined as the light output in the 1st, or the 1st 0.1 ns [25]. It is clear that the best crystal scintillators for ultra-fast timing are BaF 2 , LSO:Ca,Ce and LSO/LYSO:Ce. LaBr 3 is a material with high potential theoretically, but suffers from scattering centers in the crystal as well as its intrinsic hygroscopicity.

Crystals for the HHCAL detector concept
Aiming at the best jet mass resolution, cost-effective inorganic crystal scintillators are being developed for a homogeneous hadron calorimeter (HHCAL) detector concept with dual readout of both Cherenkov and scintillation light for future high-energy lepton colliders [2]. Because of the unprecedented volume (70-100 m 3 ) foreseen for the HHCAL Fig. 21 Decay kinetics of a PbFCl sample excited by γ -rays detector concept, cost-effectiveness is the most important requirement [3]. In addition, the material must be dense to reduce the calorimeter volume, UV transparent for effective collection of the Cherenkov light, and allow for a clear discrimination between the Cherenkov and scintillation light. The preferred scintillation light is thus at a longer wavelength, and not necessarily bright nor fast. Inorganic crystals being investigated are doped lead fluoride (PbF 2 ) [3,26], lead chloride fluoride (PbFCl) [27][28][29] and bismuth silicate (Bi 4 Si 3 O 12 or BSO) [30][31][32]. Figure 20 shows UV light-excited decay kinetics of doped PbF 2 samples [26]. The decay time of rare earth-doped PbF 2 crystals is several milliseconds which is too long to be used for HHCAL. Figure 21 shows decay kinetics of PbFCl crystal excited by gamma rays. Its 24 ns decay time is appropriate for the HHCAL application. Because of the cleavage planes  Because of its low UV cutoff wavelength (300 nm) and low raw material cost (< 50% of BGO), BSO crystals are under development at SIC. Figure 22 shows longitudinal transmittance (LT) and photoluminescence (PL) spectra for a 105-mm-long BSO crystal. Figure 23 shows relative light output of the BSO crystal comparing to BGO crystal. The LO of BSO with 100 ns decay time is approximately 12% of BGO crystal. See Reference [33] for recent development for this crystal.

Summary
Bright, fast and radiation hard LYSO/LSO crystals may be used for a total absorption ECAL. LYSO/W Shashlik calorimeter was of two options for the CMS FCAL upgrade technical report for the proposed HL-LHC. Crystal calorimeters with more than ten times faster rate/timing capability require very fast crystals, e.g., sub-ns decay time of the BaF 2 fast scintillation component. Crystals (PbF 2 , PbFCl and BSO) may provide a foundation for a homogeneous hadron calorimeter with dual readout for both Cherenkov and scintillation light to achieve unprecedented jet mass resolution for future lepton colliders.