Detecting the Diffuse Supernova Neutrino Background in the future Water-based Liquid Scintillator Detector Theia
DDetecting the Diffuse Supernova Neutrino Background in the future Water-basedLiquid Scintillator Detector Theia
Julia Sawatzki ∗ Physik-Department, Technische Universit¨at M¨unchen,James-Franck-Straße 1, 85748 Garching, Germany
Michael Wurm † Institut fr Physik, Johannes Gutenberg Universitt Mainz, Staudinger Weg 7, 55128 Mainz, Germany
Daniel Kresse ‡ Max Planck Institute for AstrophysicsGarchingPhysics Department, TU MunichMax-Planck-Institut f¨ur Astrophysik, Karl-Schwarzschild-Straße 1, 85748 Garching, Germany andPhysik-Department, Technische Universit¨at M¨unchen,James-Franck-Straße 1, 85748 Garching, Germany (Dated: July 30, 2020)A large-scale neutrino observatory based on Water-based Liquid Scintillator (WbLS) will be ex-cellently suited for a measurement of the Diffuse Supernova Neutrino Background (DSNB). TheWbLS technique offers high signal efficiency and effective suppression of the otherwise overwhelm-ing background from neutral-current interactions of atmospheric neutrinos. To illustrate this, weinvestigate the DSNB sensitivity for two configurations of the future Theia detector by developingthe expected signal and background rejection efficiencies along a full analysis chain. Based on astatistical analysis of the remaining signal and background rates, we find that a rather moderate ex-posure of 150 kt · yrs will be sufficient to claim a (5 σ ) discovery of the faint DSNB signal for standardmodel assumptions. We conclude that, in comparison with other experimental techniques, WbLSoffers the highest signal efficiency of more than 80 % and best signal significance over background. I. INTRODUCTION
Core-collapse Supernovae (SNe) are intense sourcesof low-energy neutrinos ( E ν (cid:46)
50 MeV). For SNe oc-curring within the Milky Way, bright signals are ex-pected for several current-day neutrino detectors, per-mitting detailed analyses of the astrophysics of the explo-sion and superimposed effects caused by neutrino prop-erties (for a comprehensive review, see e.g. Ref. [1]).However, even compared to the decades-long operationtimes of large-volume neutrino observatories, galacticSNe are rare. Therefore, the search for the faint butconstant signal predicted for the Diffuse Supernova Neu-trino Background (DSNB), i.e. the integral neutrino fluxfrom past core-collapse SNe at cosmological distances, isespecially appealing [2–10]. A measurement of the DSNBflux and spectrum will provide valuable information onthe redshift-dependent SN rate as well as on the proper-ties of stellar core collapse, such as the nuclear equationof state of the emerging neutron stars or the fractionof core-collapse events leading to the formation of blackholes in faint or failed explosions.Given the minute expected flux of O (10 ) per cm s andlow energy of DSNB neutrinos and antineutrinos of all ∗ [email protected] † [email protected] ‡ [email protected] flavors, an experimental measurement is very challeng-ing. Detector target masses on the order of ∼
10 kilo-tons are required to obtain one signal event per year. To-day, the Super-Kamiokande (SK) experiment holds thebest upper limit on the DSNB’s ¯ ν e flux component of(2.8 − − s − above 17 . σ ) of the DSNB sig-nal. However, given that the respective target masses of ∼
20 kt are relatively low by standards of the DSNB, ac-cumulation of event statistics will be slow. Moreover, thepresence of background events caused by neutral-current(NC) interactions of atmospheric neutrinos complicatesdetection [9, 13, 14].The present paper studies the potential of an advanceddetection concept for a definitive 5 σ -detection of a ”stan-dard” DSNB signal [10]. As has been laid out in the whitepaper of the future Theia detector [15], Water-basedLiquid Scintillator (WbLS) in combination with ultra-fast light sensors (LAPPDs) and/or high PMT granu-larity permits the simultaneous detection of Cherenkovand scintillation light. Different from pure water or or-ganic scintillator detectors, the evaluation of the dualCherenkov/scintillation signal provides superior back- a r X i v : . [ phy s i c s . i n s - d e t ] J u l ground discrimination. Correspondingly, a WbLS detec-tor will feature an excellent signal-to-background ratio:For Theia at Homestake, we expect ∼ +40 − DSNB signalevents over 9 ± · yrs (sec. VI)and an observation window ranging from 8 to 30 MeV.Note that a similar study has been performed inRef. [16] in the context of the Jinping Neutrino Experi-ment. However, the present paper goes substantially be-yond the earlier study by including a realistic detectorsimulation to evaluate the expected ratio of Cherenkovand scintillation photons (C/S ratio) detected. Moreover,we investigate not only background discrimination basedon the ratio of detected Cherenkov and scintillation pho-tons but as well the possibility to add further backgroundtags exploiting Cherenkov ring counting and delayed de-cays of excited final-state nuclei.The paper is structured as follows: Sec. II sets out thebasic concept of DSNB detection in WbLS and the lay-out of the Theia detectors. The recent DSNB flux mod-els by [10] employed in this work are briefly described insec. III. Relevant backgrounds and several conventionaltechniques for their suppression in liquid-scintillator de-tectors are shortly reviewed in sec. IV. Instead, sec. Vplaces particular emphasis on the discrimination tech-niques specific to WbLS that are instrumental to thevery effective reduction of the dominant NC backgroundfrom atmospheric neutrinos: delayed decays from oxygenspallation, ring counting, and − most importantly − theC/S ratio. Sec. VI provides signal and background ratesas well as the corresponding DSNB discovery potential,while sec. VII puts these results into context with otherexisting and planned neutrino detectors. II. DETECTOR TECHNOLOGY
When neutrinos interact in a conventional (i.e. organic)liquid scintillator detector, the final state particles willcreate not only scintillation but also Cherenkov photons.However, both organic solvent and fluorophores featurestrong absorption bands in the dominant UV/blue partof the Cherenkov spectrum. Upon arrival at the photo-sensors surrounding the scintillator neutrino target, theremaining direct (i.e. unscattered) Cherenkov photonsare effectively hard to distinguish from the overwhelmingscintillation signal.
Water-based Liquid Scintillator (WbLS).
The pri-mary motivation for the use of WbLS is to provide a verytransparent scintillator by adding ultrapure water to theorganic compounds. The resulting liquid is an emulsionthat consists of mycels (nanoscopic droplets) of organicmaterial surrounded by a surfactant and dissolved in thebulk water. The main characteristics of state-of-the-artWbLS samples have been investigated in [17, 18]: Thelight yield of the resulting WbLS is roughly proportionalto its organic fraction. The attenuation length reached inthe blue spectral range depends primarily on the proper-ties of the diluted organic fraction and is assumed here
FIG. 1. Basic detector geometries of Theia100 (left) andTheia25 (right). very conservatively to ∼
20 m [15]. This permits the de-tection of a sizable fraction of the Cherenkov photons.
Cherenkov/scintillation separation.
To make use ofthe information carried by the Cherenkov signal, it mustbe distinguished from the scintillation photons. For this,there are three basic approaches: • fast light sensors with (sub-)nanosecond time reso-lution (e.g. LAPPDs [19]) permit the identificationof a front of Cherenkov light arriving nanosecondsearlier than the delayed scintillation emission, • the characteristic angular dependence of Cherenkovemission will cause a ring-shaped local enhance-ment in the detected light intensity on top of theisotropic scintillation signal, and • wavelength-sensitive photosensors, e.g. dichroi-cons [20], can distinguish near-UV scintillationlight from the blue-green lower end of theCherenkov spectrum that has been traveling un-perturbedly through the WbLS bulk volume. Theia.
A large-volume WbLS detector has been firstproposed as part of the Advanced Scintillator DetectionConcept [21]. It has been later on developed further tothe Theia detector project [15]: The full physics potentialof the WbLS technique could be exploited with a detec-tor mass of 100 kt, named Theia100. More recently, asmaller detector realization, Theia25, has been discussedas a possible contender for the fourth DUNE detectormodule with a WbLS mass of 25 kt [15]. The correspond-ing detector geometries are displayed in fig. 1.
DSNB detection.
The detection potential of the DSNBin WbLS has been first described in [21]. As will bedemonstrated in sec. V, the possibility to detect and dis-tinguish Cherenkov and scintillation photons providesa great advantage in terms of background discrimina-tion when compared to conventional water Cherenkovor fully organic scintillation detectors: Large-volumeCherenkov detectors like SK I-IV feature low tagging effi-ciency for the delayed neutron capture in inverse beta de-cays (IBDs), making event selection much more suscepti-ble to single-event backgrounds [11, 22–24]. On the otherhand, liquid scintillator detectors like JUNO suffer fromneutral-current (NC) interactions of atmospheric neutri-nos on carbon that may break-up the nucleus and createIBD-like signatures from quenched final-state hadronsand the delayed capture of a single neutron [13, 14, 25–27]. A similar background from NC reactions on oxygenis present as well in SK [28]. It should be noted that bothdetector technologies can mitigate those backgrounds:SK will soon start a detection run with gadolinium addedto the water, named SK-Gd, greatly enhancing the de-layed neutron tag [29]. On the other hand, pulse-shapediscrimination may be used in organic scintillators to sup-press the NC background [13, 27].
DSNB in Theia.
For both water Cherenkov and or-ganic scintillator experiments, the residual detection effi-ciency for the DSNB signal is in the order of 50 %. Con-trariwise, we expect a detection efficiency of more than80 % in WbLS. As in an organic scintillator, the addi-tional scintillation light enables a reliant detection of thedelayed neutron capture. Crucially, background discrim-ination capabilities are greatly enhanced by exploitingCherenkov ring counting (sec. V B) and the Cherenkov-scintillation ratio (sec. V C). As a result, the expectedsignal-to-background ratio will be excellent. Moreover, adetailed study of NC atmospheric neutrino backgroundevents with the diagnostic possibilities of WbLS detec-tion is likely to provide SK-Gd and JUNO with a bettersystematic understanding of their respective NC back-ground levels and event topologies.
Detector configuration.
In the Geant4 (Ver-sion 9.4.p4) simulation underlying this study [30, 31], weused a generic spherical detector geometry on the scaleof Theia25. We assume a WbLS with a 10 % organicfraction, corresponding to a scintillation light yield of ∼ photons per MeV and thus on the same scale as theCherenkov light emission. Light transport and scatteringin the target medium have been implemented. The ab-sorption and Rayleigh scattering length were set to 77 mand 27 m, respectively, to obtain an attenuation lengthof ∼
20 m (for λ = 430 nm). Since data on quenching inWbLS is scarce [17], and none is available for the rele-vant energy range, we assumed the quenching factors tobe similar to that of organic scintillators.Furthermore, dense instrumentation of the detectorsurfaces is required to obtain both sizable Cherenkov andscintillation signals. In Theia, this will be realized by amixture of high-QE 10”-PMTs and LAPPDs (ratio 25:1).Here, we assume a generic coverage of 70 % (25 %) ofthe detector surface for Theia100 (Theia25) and a peakdetection efficiency of ∼
30% for the PMT-like photo-sensors [15]. The resulting photoelectron yield for thescintillation signal is 130 p.e./MeV, while the Cherenkovcomponent provides 80 p.e./MeV. Note that, in the fol-lowing, all visible energy spectra are based solely on thephotons collected for the scintillation component of thesignal. This implicitly assumes close-to-perfect separa-tion of the scintillation and Cherenkov components. The same assumption is made for the event selection in sec. V.
Inverse beta decay.
Since it has the largest cross-section at low energies, the primary detection channelfor the DSNB in water and organic liquids is the inversebeta decay (IBD) of electron antineutrinos on free pro-tons: ¯ ν e + p → e + + n . Due to the kinematics of the IBDreaction, the kinetic energy of the positron translates al-most directly to the incident neutrino energy but is re-duced by the reaction Q-value of 1 . E ν ≈ E vis + 0 . ∼ µ s,producing a 2 . III. DSNB MODEL
DSNB spectrum.
Given the vast multitude anddistance of SNe that contribute to the DSNB, its fluxis nearly isotropic. The energy spectrum averages overthe entire population of stellar core collapses from awide range of progenitor stars (including failed explo-sions leading to the formation of black holes) and is sub-stantially red-shifted for sources at far-out distances. Thedetectable signal above ∼
10 MeV is thus dominated byrelatively close-by core-collapse events up to redshifts of z ≈ N ν d E ν = N p × dΦ ν d E ν × σ ν ( E ν ) , (1)where dΦ ν / d E ν is the differential DSNB flux, σ ν ( E ν )is the energy-dependent cross-section for the IBD reac-tion [32] and N p = 6 . × / (10 kt) is the number ofprotons in the target volume. As the Earth is virtuallytransparent to low-energy neutrinos, detectors see a fullyisotropic signal. The differential number flux dΦ ν / d E ν can be computed via a line-of-sight integral of the averageSN neutrino number spectrum d N ( E (cid:48) ν ) / d E (cid:48) ν (weightedby an initial mass function), multiplied by the evolvingcore-collapse SN rate R SN ( z ) over cosmic history [2, 3]:dΦ ν d E ν = c (cid:90) ∞ d N ( E (cid:48) ν )d E (cid:48) ν × d E (cid:48) ν d E ν × R SN ( z ) × (cid:12)(cid:12)(cid:12)(cid:12) d t d z (cid:12)(cid:12)(cid:12)(cid:12) d z, (2)where c is the speed of light, E (cid:48) ν is the neutrino energyat emission, and E ν = E (cid:48) ν / (1 + z ) denotes the neutrinoenergy upon detection, corrected for the redshift. Theterm | d t/ d z | is given by an underlying cosmologicalmodel; it accounts for the expansion history of theUniverse and relates z to the cosmic time t . Model inputs.
As our DSNB model, we employthe recent flux predictions by [10]. They are based ona large set of spherically symmetric SN simulations withthe
Prometheus-HotB code [33–36] over a wide rangeof progenitor stars with birth masses between ∼ (cid:12) ) and include cases of black holeformation in failed explosions. Their modelling approachof using “calibrated neutrino engines” follows the worksby [36–38]. For most parts of our analysis, we take the“fiducial” model of [10] (see their Section 4). However,to account for the large uncertainties which are still un-derlying any theoretical prediction of the DSNB, we ad-ditionally consider a “low-flux” and a “high-flux” modelfrom [10]. The parameters of the three models are as fol-lows: • Fiducial model:
This model with “Z9.6 & W18”neutrino engine assumes a maximum baryonic neu-tron star mass of 2 . (cid:12) (which corresponds to ∼ . (cid:12) of gravitating mass) and a spectral shapeof the SN neutrino emission according to [39], ad-justed to more sophisticated SN simulations withdetailed microphysics. The core-collapse SN ratescales with the cosmic star-formation history forwhich the parametric description by [40] is takentogether with the best-fit parameters by [41]. • Low-flux model:
This model makes for a ratherconservative flux prediction and is taken as a lowerlimit in our study. It is based on SN simulationswith the “Z9.6 & S19.8” neutrino engine and as-sumes a maximum baryonic (gravitational) neutronstar mass of 2 . (cid:12) ( ∼ . (cid:12) ). The same spectralshapes as in the fiducial model are taken, whereasthe − σ lower-limit parameters by [41] are em-ployed to the cosmic star-formation history. • High-flux model:
This model is chosen such thatits integrated flux above 17 . . (cid:12) ) as inthe fiducial model are taken. However, a spectralshape parameter according to [39] of α = 2 . σ upper-limit parame-ters by [41] for the functional form of the cosmicstar-formation history.The reader is referred to [10] for a more detailed descrip-tion of the DSNB modelling and the motivation of theentering parameters and assumptions. Using these models, the expected number of DSNB¯ ν e events in a WbLS detector can be computed accord-ing to eq. (1). Assuming the fiducial model to be re-alised in Nature, we expect 35 events per 100 kt · yrs (for The flux models can be downloaded from the Garching Core-Collapse Supernova Archive ( ) upon request. scintillation p.e. · p . e . ) · e v en t s / ( k t y r - - -
10 110 visible scintillation energy (MeV)0 5 10 15 20 25 30 35 40 45 DSNBReactor BGAtmCC BGAtmNC BGLi9 BGFastN BG
FIG. 2. The visible scintillation energy spectrum expectedfor the fiducial DSNB signal and its ample backgrounds.The presented spectra include reactor neutrinos, cosmogenic Li, fast neutrons as well as atmospheric neutrino charged-current(CC) and neutral-current(NC) interaction rates. Weassume a basic event selection of IBD-like coincidence sig-nals (with only a single accompanying neutron capture). Theenergy scale is based on the number of scintillation photonsdetected. The upper axis lists the corresponding visible scin-tillation energy. (cid:54) E ν (cid:54)
40 MeV). The more conservative low-fluxmodel would yield 17/(100 kt · yrs), whereas the high-fluxmodel (roughly corresponding to the current SK limit)would translate to 68/(100 kt · yrs). This wide spread ofvalues illustrates the large uncertainties which are stillpreventing more precise DSNB predictions at the presentday. Nonetheless, it also shows the big potential of fu-ture DSNB measurements to independently constrain the(yet wide) parameter space. If not stated otherwise, werefer to the fiducial DSNB model throughout our workbut comment on the model dependence of our results insec. VI. IV. BACKGROUND MODELING
A variety of backgrounds besets the DSNB signal.They can be divided into three categories: • The terrestrial flux of ¯ ν e ’s from reactors and atmo-spheric neutrinos causes an irreducible backgroundof real IBD coincidences and reduces the detectionwindow to the range of about 8 to 30 MeV. • Cosmic muons are penetrating the rock shieldingabove the detector. In spallation processes, thesemuons generate β n-emitting isotopes (e.g. Li) inthe target material as well as fast neutrons whenpassing through the rock surrounding the detector.Both can mimic IBD coincidence signals. • High-energy atmospheric neutrinos undergoneutral-current (NC) interactions in the targetmedium. In case neutrons are released in thesereactions, they can create coincidences with aprompt signal in the visible energy range of theDSNB.In the following, we give a detailed account of the mod-eling of these background sources. The expected eventrates are listed in the first column of tab. II, while fig. 2displays the corresponding visible energy spectra.
A. Reactor and Low-Energy AtmosphericNeutrinos
Reactor neutrinos . ¯ ν e ’s emitted by nuclear reac-tors provide a high background flux at energies below10 MeV. The total reactor neutrino rate (including itsuncertainty) and the oscillated energy spectrum are de-rived from [42]. For the Sanford Underground ResearchFacility (SURF) in South Dakota, we expect (2240 ± · yrs. Atmospheric neutrinos.
At low energies, the fluxof atmospheric ¯ ν e ’s increases with energy and startsto surpass the DSNB signal at around 30 MeV. Sincetheir flux depends on the geographic (geomagnetic) lat-itude [43], we adopt the HKKM atmospheric neutrinofluxes between 100 MeV and 10 GeV that have beencalculated for the DUNE experiment at the same loca-tion [44]. For lower neutrino energies, we extrapolate theFLUKA simulations performed for the location of theGran Sasso National Laboratories [45] that are at nearlythe same geographical latitude (42 . ◦ N vs. 44 . ◦ N).The FLUKA fluxes are scaled to match the atmosphericHKKM spectrum between 100 MeV and 200 MeV. Us-ing the IBD cross-section from [32], the rate of atmo-spheric neutrino reactions below 100 MeV is calculatedto (48 ± · yrs). The relative error of 35 % re-flects the uncertainty of low-energy atmospheric flux pre-dictions [45, 46]. B. Cosmogenic Backgrounds
In-situ production of Li.
Cosmic muons create avariety of radioisotopes by spallation on the oxygen (andcarbon) nuclei of the WbLS target. Of those, only βn -emitters can mimic the IBD signature and are thus po-tential contributors to the background. The only isotopeproduced with a relevant cross-section and sufficientlyhigh endpoint energy ( Q = 13 . Li. In ∼
50 % of all cases, Li de-cays to an excited state of Be that de-excites via emis-sion of a neutron [47]. The decay scheme of the β − -decayof Li was implemented according to [48, 49]. Since thisbackground scales to first order with the muon flux, adeep location like SURF translates to a substantial reduc-tion in the Li production rate. We estimate the expectedbackground rate by adopting the Li yield measured for water and the organic component of the WbLS in SK[50] and Borexino [51], respectively. According to [52], wescale this rate to the lower muon flux and higher muonmean energy at SURF, i.e. R Li ∝ Φ µ ·(cid:104) E µ (cid:105) . , assuminga muon flux of Φ µ = 4 . × − cm − s − and mean muonenergy of (cid:104) E µ (cid:105) = 293 GeV [53]. The resulting IBD-likebackground rate is approximately (530 ± / (100 kt yr),the relative uncertainty of 20 % in line with the measuredvalues [50, 51]. Fast Neutrons.
High-energy neutrons induced by cos-mic muons can mimic the IBD signature. The promptsignal is provided by the elastic scattering of the neu-tron on a free proton in the target material, followed bythe thermalization of the neutron and its capture on hy-drogen. Muons crossing the target volume can be clearlyidentified, permitting a suppression of the trailing neu-tron signals by a short veto of the detector following eachmuon event (∆ t ∼ µ s). On the other hand, neutronsinduced by muons passing through the rock layer sur-rounding the detector provide no immediate backgroundtag. The reconstructed vertex positions will usually beclose to the verge of the detection volume, though.We estimate the fast neutron production rate basedon prior work for the LENA experiment at the Pyhsalmimine [27]. At this location, rock overburden (4 km.w.e.)and mean muon energy of 300 GeV are comparable toSURF, so that we scale the rate by a factor 0.5 to takeinto account the lower muon flux. Due to the relativelyshort mean free path of the neutrons ( λ n < . / (100 kt yr) for E vis <
40 MeV.Measured neutron yields in Borexino are accurate upto 5% [51], but uncertainties on muon flux and neutronpropagation will further increase uncertainty on the pre-dicted background rate. For Theia25, the resulting fastneutron rate is 60 % lower. In line with earlier publica-tions from reactor neutrino experiments, we expect analmost flat visible energy spectrum for the neutron re-coils (compare e.g. [55]). Even if this simplified assump-tion may not be fully justified, the impact on this studyis negligible since the fast neutron background can bereduced very efficiently (sec. V).
C. Neutral-Current Reactions of AtmosphericNeutrinos
The neutral-current(NC) interactions of high-energyatmospheric (anti-)neutrinos of all flavors pose the mostserious background to DSNB detection. Events mimick-ing IBDs originate from reactions with a single neutron inthe final state, while the prompt signal can be composed ν x + O −−→ ν x +Reaction Channel Branching Ratio [%]n + O 49.4n + p + N 20.0n + 2 p + C 12.6n + p + d + C 7.7n + p + d + α + Be 2.1n + α + He + Be 1.7n + 3 p + B 1.6n + p + α + B 1.3n + 2 p + α + Be 1.3other reaction channels 2.3 R NC (cid:39) · yrs)TABLE I. Branching ratios of inelastic NC scattering reac-tions of atmospheric neutrinos on O with one neutron inthe final state, including the de-excitation channels, sortedby their prevalence for visible energies below ∼
46 MeV. of a multitude of different combinations of nuclear frag-ments and gamma-rays. First recognized by the Kam-LAND experiment [14], it dominates the DSNB signalby more than one order of magnitude.
Expected event rates.
To obtain the event ratesin WbLS as a function of final-state particle compo-sition, we use the GENIE Neutrino Monte-Carlo gen-erator (Version 2.12.4) [56, 57] and feed it with theHKKM atmospheric neutrino spectrum considering amaximum neutrino energy of 10 GeV [44]. The propa-gation of Cherenkov and scintillation signals created bythe final-state particles are obtained using the GEANT4detector simulation. However, we pre-filter the reactionsand include only events that feature at least one neutronin the final state, which is valid for ∼
79% of all interac-tions. Since NC reactions on C contribute only 8 % ofall NC reactions, they are taken into account by increas-ing the event number resulting from O interactions by10 %. The overall NC event rate is thus 7 . · per100 kt · yrs below 10 GeV of neutrino energy.Tab. I displays the branching ratios for different con-figurations of nuclear fragments in the final state. Sincewe are only interested in reactions resulting in a sin-gle neutron, the event rate is reduced to a residual of1270/(100 kt · yrs).While an expectation for the background event ratecan be quoted, it is not straight-forward to determinethe corresponding uncertainties on both rate and spectra.The underlying atmospheric neutrino flux is relativelywell known up to an uncertainty of less than 10% in therelevant energy region [44]. However, there is still verylittle experimental data on the relevant NC cross-sectionson oxygen in this energy region [58].In the near future, corresponding data on cross-sections in water can be expected from the ANNIE exper- iment at Fermilab. ANNIE will provide a high-statisticssample ( O (10 )) of GeV neutrinos from the Booster Neu-trino Beam [59] that will be used to constrain CC andNC cross-sections. A future WbLS phase of ANNIE iscurrently discussed and would provide even more valu-able data in the present context. Nuclear de-excitations.
When modeling the light out-put of the prompt events, i.e. of the nuclear fragmentscreated in the NC interactions, it is essential to includede-excitation gammas-rays. For the present study, we aremost interested in NC interactions were a single neutronis knocked out from an O nucleus in the WbLS (tocreate an IBD-like signature).In the simple shell model, the neutron configuration of O is [(1 s / ) (1 p / ) (1 p / ) ]. As the energy of atmo-spheric neutrinos is large compared to the binding en-ergies of individual nucleons, it is valid to assume thatthe probability of neutrino interaction does not dependon the nucleon’s binding state [60]. When a neutron isknocked out, the interacting nucleus is left with a hole inthe corresponding shell. Reordering of the neutron con-figuration leads to a release of binding energy via parti-cle emission. No rearrangement is needed in the 25 % ofall cases for which the neutron was located in the 1 p / shell. However, with a probability of 50 %, the neutron isknocked out from the 1 p / shell. In this case, the excited O nucleus undergoes a direct transition to its groundstate, causing the emission of a single 6 .
18 MeV gamma-ray (with a branching ratio of ∼
87 %) [61]. Finally,knock-outs from the 1 s / shell leave an excited state of O with a potential transition energy that exceeds theseparation energy of both protons and neutrons. Thus,de-excitation proceeds mainly via the emission of pro-tons, nucleons and α -particles and only sub-dominantlyvia γ -rays [62].Note that there is a chance that the knock-off nucleonwill scatter on other nucleons before leaving the nucleus,leading to the emission of additional nucleons. For sim-plicity, it is assumed that these intra-nuclear scatteringreactions only emit nucleons from the 1 p / shell. More-over, gamma emission can be as well induced in sec-ondary interactions of knock-out protons and neutronson neighboring O nuclei (sec. V) that may in turn con-tribute to the overall prompt signal.Since neither GENIE nor GEANT4 provides detailson particles emitted in the de-excitation of atomic nucleiin neutrino interactions, the nuclear reaction programTALYS (Version 1.4) [63] was used to describe the parti-cle emission from residual excited nuclei ( O, N, N, C, C).
V. BACKGROUND REDUCTION
As shown in tab. II, several of the background ratesin a WbLS target significantly exceed the DSNB sig-nal rate. Like in water Cherenkov and organic liquid-scintillator detectors, the IBD interactions induced bythe reactor and low-energy atmospheric ¯ ν e ’s prove an ir-reducible background and effectively limit DSNB detec-tion to the ∼ [8 −
30] MeV range. However, within theobservation window, the enhanced event discriminationcapabilities of a WbLS detector enable an effective re-duction of background rates to a level significantly lowerthan that of the DSNB signal (sec. VI). In the following,we discuss the corresponding event selection in detail.
A. Basic Selection Cuts
IBD selection.
The analysis of signal and back-ground events starts with the imposition of the selectioncriteria for the inverse beta decay (IBD). We limit the ac-ceptance of prompt events to the energy range from 8 to30 MeV (corresponding to 10 − · p.e. in Theia100).The lower energy threshold excludes the reactor neutrinobackground, which is thus neglected in the following. Forsimplicity, we assume that in this energy region, spatialand time difference cuts applied for selection of the IBDcoincidence can be chosen sufficiently wide so that thecorresponding efficiency will be close to unity. This as-sumption is justified since accidental backgrounds will benegligible based on the high energy of the prompt event;the scintillation signal of the 2.2-MeV signal from delayedneutron capture will be well above threshold; and finally,we impose a substantial fiducial volume cut (see below)that will mitigate as well border effects for the IBD selec-tion. It is important to note that the IBD selection onlyaccepts events with exactly one delayed neutron event.As pointed out in sec. IV C, this results in a significantreduction of the NC atmospheric background rate, wheremultiple neutrons are often generated in the final state. Cosmogenic background veto.
Background from cos-mogenic radioisotopes, especially of the βn -emitter Li,can be effectively reduced by the formation of a coinci-dence veto with the preceding parent muon. We choosea time veto of 2 s following each muon and a cylindricalcut around the muon track of 5 m radius. This providesexcellent efficiency when compared to the Li life timeof τ ( Li) = 257 ms and the expected lateral productionprofile [64]. Given the deep location at SURF, the corre-sponding loss in exposure is less than 1 %. In the follow-ing, we regard any possible residuals of Li as negligible.
Fiducial volume cut.
Finally, we impose a fiducial vol-ume cut to reduce surface backgrounds, border effects indetection efficiency and, most crucially, the backgroundimposed by fast neutrons. Given the difference in detec-tor geometries, we choose individual cuts for Theia100, scintillation p.e. · p . e . ) · e v en t s / ( k t y r - - -
10 110 visible scintillation energy (MeV)0 5 10 15 20 25 30 35 40 45 DSNBReactor BGAtmCC BGAtmNC BGFastN BG
FIG. 3. The visible scintillation energy spectra expected forthe fiducial DSNB signal and backgrounds after the appli-cation of basic discrimination techniques: We apply a 2 . Li based on the coincidence with the precedingparent muons. 100 kt · yrs exposureBefore Cuts Muon Veto Volume CutDSNB signal 35.0 34.6 34.6Reactor neutrinos 2240 2218 2218Atmospheric CC 9.0 8.9 8.9Atmospheric NC 1270 1253 1253 βn -emitters ( Li) 529 − − fast neutrons 131 (57) 129 (56) 2 (4)TABLE II. Rates of DSNB signal and backgrounds at en-ergies below ∼
46 MeV (6000 p.e.) normalized to a live ex-posure of 100 kt · yrs. The fast neutron rates displayed inthe last row assume a 2 . . where an outer layer of 2 . . B. Cherenkov Ring Counting
The number of Cherenkov rings created by the promptevent provides a handle to discriminate the DSNB signalagainst background events that feature multiple parti-cles in the final state. A positron induced by an IBDinteraction will create a single Cherenkov ring. In con-trast, NC interactions of atmospheric neutrinos often re-sult in several particles above the Cherenkov thresholdand thus multiple rings. Fig. 4 displays the number ofCherenkov rings in NC events as a function of the vis-ible scintillation energy. Inside the observation window scintillation p.e.0 500 1000 1500 2000 2500 3000 3500 4000 4500 500050100150200250300350 visible scintillation energy (MeV)0 5 10 15 20 25 30 35
One-RingMulti-RingNo-Ring
AtmNC events
FIG. 4. Number of Cherenkov rings reconstructed for atmo-spheric NC events as a function of the prompt visible scintil-lation energy. The frequent occurrence of two or more ringsallows for an efficient discrimination against the single-ringDSNB positrons. The grey boxes indicate the limits of theobservation window. (1 , − ,
600 p.e.), about half of the NC events featureeither no or more than a single ring.To use this approach for event discrimination, the indi-vidual Cherenkov rings must be sufficiently bright for re-construction. While a detailed event reconstruction goesbeyond the possibilities of this study, we impose herethe rather conservative condition that a Cherenkov ringmust contain at least 300 p.e. to be discernible in thepresence of other rings. This condition is immediatelymet by all signal events, translating to no relevant lossof signal efficiency. Moreover, our analysis code assignsall Cherenkov photons emitted by secondary particlesto the initial particles created in the interaction vertex.Thus, secondary particles cannot be discerned as individ-ual Cherenkov emitters and do not enter the ring count.Despite this very conservative assumption, single-ring se-lection rejects 57 % of atmospheric-neutrino NC interac-tions as multi-ring background without any relevant lossin signal efficiency.
C. Cherenkov/Scintillation Ratio
The discrimination technique unique to WbLS de-tectors is the evaluation of the ratio of reconstructedCherenkov to scintillation photons, named C/S ratio inthe following. This parameter depends crucially on theparticle type: While the relativistic prompt positrons ofIBD events feature a high C/S ratio, the final statesof atmospheric NC reactions are mostly hadronic andemit no or comparatively little Cherenkov light. Thisholds even more for the proton recoils induced by fastneutrons in the WbLS target. The resulting discrimina-tion power for IBDs and atmospheric NC interactionsis demonstrated in fig. 5. It displays the C/S ratio as a scintillation p.e. C / S r a t i o visible scintillation energy (MeV)0 10 20 30 40 50 60 70 DSNBAtmNC scintillation p.e. C / S r a t i o visible scintillation energy (MeV)0 10 20 30 40 50 FIG. 5. The Cherenkov-to-scintillation (C/S) ratio offersa powerful tool to discriminate prompt positrons of DSNBevents (blue) and hadronic prompt events of atmospheric NCreactions (black). Atmospheric NC events lead to a signifi-cantly reduced emission of Cherenkov photons. The lower plotpresents a zoom-in for C/S values greater than 0.5. The grayshaded area indicates the limits of the observation window.The red line corresponds to the C/S cut threshold reaching82% signal efficiency. function of visible scintillation energy. In the range above8 MeV (10 p.e.), there is a clear separation of the eventdistributions. The residual NC background contamina-tion arises from NC reactions with one or more γ -rays inthe final state (sec. IV C).The prevalence of γ -rays of ∼ S + BS / yr) · AtmNC Rate /(100 kt y r) · D S N B R a t e / ( k t
82% signal efficiency
Residual Background [%]1 2 3 4 5 6 7 8 9 10
FIG. 6. The optimum choice of a cut on the C/S ratio dependson the optimization of the signal-to-background (S/B) ratio.The rate of surviving DSNB events as a function of the resid-ual rate of atmospheric NC events is indicated by the solidline. While not shown, the corresponding C/S cut thresholdis increasingly relaxed from left to right. The dashed line in-dicates the corresponding significance of the signal over back-ground S/ √ S + B (scale on the right y-axis). The maximumof the curve (82% signal efficiency at 3.5 % residual back-ground, indicated by the grey line) is chosen for the furtheranalysis. a single Cherenkov ring, it is likely that these events − though not discernible by their C/S ratio − could beidentified as background by Cherenkov ring counting (seeabove). However, since the current study lacks a suf-ficiently detailed modeling and reconstruction of theseevents, we conservatively assume that they cannot bediscriminated.In order to utilize the C/S ratio in the analysis, wedefine an energy-dependent lower threshold for signal se-lection. It is exemplified by the red line inserted in fig. 5.The exact threshold values imposed depend on the focusof the DSNB analysis. Here, we optimize for the de-tection potential of the DSNB, i.e. we apply an energy-dependent selection cut that maximizes the significanceof the DSNB signal S over the NC background B , rep-resented by the signal significance S/ √ S + B and visu-alized in fig. 6. For the shown configuration, we reachan optimum in DSNB signal acceptance of 82 %, leavinga residual of only 3.5 % for atmospheric NC backgroundevents and ∼
1% for fast neutron events (not shown).Given the importance of this discrimination techniquefor the DSNB detection, we also investigated its depen-dence on the light collection. Since Theia25 plans for aninitial coverage of 25 %, the number of photo electronscollected would be reduced by a factor three comparedto 75 % coverage of Theia100. While the larger uncer-tainty in photon statistics translates to a slight weaken-ing of the discrimination power, the effective S/B valuesare only mildly affected: for the optimum threshold, thesignal acceptance is 78 % while a background residual of 3.7 % is permitted.
D. Delayed Decays
As displayed in tab. I, a significant fraction of the NCreactions of atmospheric neutrinos on O (with a sin-gle neutron in the final state) leaves behind a radioactiveisotope. In principle, its decay can be used to reject theoriginal IBD-like interactions by means of a delayed co-incidence tag. In water, the dominant isotope created is O with a branching ratio of ∼ O undergoes a β + -decay with an endpoint of 2 . . O and low de-cay energy, a refined selection condition must be appliedfor the delayed coincidence to prevent a high rate ofaccidental coincidences of DSNB neutrino events withradioactive decays intrinsic to the scintillator. In accor-dance with [15], we assume a contamination of the waterwith elements of the U/ Th chains on the level of10 − g/g. To reduce the rate of accidental coincidences,we impose a maximum delay time of 10 min (about 3 lifetimes of O) and a spatial distance of 1 m between thevertices. This is sufficient to reduce the probability ofaccidental coincidences of signal events to 1 %, while theveto efficiency for delayed-decay NC interactions is stillat 95 %.Beyond O and the low-yield isotope (in water) B,none of the other isotopes created permits a similar ve-toing technique. While N, C, C, B, and Be arestable, Be almost immediately decays into two alphas.Arguably, WbLS can be regarded as the only tech-nique that will provide reasonably high efficiency for thedelayed-decay veto. For pure Cherenkov detectors, tag-ging efficiency for O will be very low. In organic scin-tillators, the main product of NC interactions on C is C. While these decays are above the threshold, thelifetime is a long ∼
30 min, making a veto based on thedelayed decay as proposed in [66] quite challenging.
VI. DETECTION POTENTIAL
The excellent background discrimination capabilitiesof WbLS mean that a detector based on this techniquecan hope for fast acquisition of statistics on the DSNBsignal at comparatively low background levels. Tab. IIIillustrates the impact of the sequence of event selectioncuts that have been introduced in sec. V. A great reduc-tion of the overall background levels is evident. Thisis especially true for the rejection of the background byatmospheric-neutrino NC interactions that is reduced byalmost two orders of magnitude. At the same time, theDSNB signal acceptance is only mildly affected.
Signal and background rates.
Overall, ∼ (9 ±
2) back-ground events per 100 kt · yrs remain in the observation0
100 kt · yrs exposureSpectral component basic cuts single-ring C/S cut delayed decaysDSNB signal 25.1 25.1 20.5 (20.1) 20.3 (19.9)Atmospheric CC 2.0 2.0 1.7 (1.6) 1.7 (1.6)Atmospheric NC 682 394 13.6 (14.6) 7.4 (7.9)fast neutrons 0.8 0.8 − − Signal efficiency 1 1 0.82 (0.81) 0.81 (0.80)Background residual 1 0.58 0.022 (0.024) 0.013 (0.014)Signal-to-background 0.04 0.06 1.3 (1.2) 2.2 (2.1)Signal significance 1.0 1.2 3.4 (3.3) 3.7 (3.7)TABLE III. Integral rates of DSNB signal and backgrounds within the observation window ( ∼ −
30 MeV) for a live exposure of100 kt · yrs in Theia100 (Theia25). The first column represents the rates applying the basic selection cuts described in sec. V A,where reactor and cosmogenic backgrounds have already been reduced to a negligible level. The following columns correspondto the selection of single-rings, application of C/S ratio cut, and a delayed-decay veto (secs. V B-V D). Furthermore, the fractionof DSNB event and the residual background retained in the observation window, signal-to-background (S:B) ratio as well asthe signal significance over background S/ √ S + B , is given. window, while the expected signal rate is ∼ +40 − eventsper 100 kt · yrs. The corresponding visible-energy spec-trum of signal and background events is shown in fig. 7.The exact signal and background numbers depend sub-stantially on the C/S ratio cut imposed. When compar-ing to the data sample left after basic event selection, wefind a final signal efficiency of >
80 % and a backgroundresidual of 1.3 %. The direct signal-to-background (S:B)ratio is 2.2 (2.1) for Theia100 (Theia25). As describedin sec. V C, we optimize the signal significance over thebackground instead, which amounts to S/ √ S + B ≈ . · yrs. Sensitivity of detection.
When calculating the ex-pected significance for the discovery of the DSNB sig-nal, we regard only the total event numbers of signaland background detected in the observation window (i.e.no spectral information is used). The measured DSNBrate is determined by statistical subtraction of the back-ground rates from the overall number of detected events.The corresponding confidence interval is calculated ac-cording to [67]. Obviously, this method requires to im-pose priors on the relevant background rates, mostlybased on control windows: • True IBDs:
The (negligible) contribution of re-actor neutrinos can be constrained in the energyrange below the observation window, IBDs by at-mospheric neutrinos in the energy region above.For the latter, we assume a scaling uncertainty of20 %. • βn emitters: The efficiency of the spatial cut sur-rounding muon tracks can be calculated based onthe acquired lateral distribution. The contributionis, in any case, negligible. • Fast neutrons can be constrained based on the ex-pected exponential radial profile and the rate mea-sured in the WbLS target outside the fiducial vol-ume. scintillation p.e. · p . e . ) · e v en t s / ( k t y r - - -
10 110 visible scintillation energy (MeV)0 5 10 15 20 25 30 35 40 45 DSNBReactor BGAtmCC BGAtmNC BG
FIG. 7. The visible scintillation energy spectrum expected forfiducial DSNB signal and backgrounds after all selection cuts.The background components include IBDs from reactor andatmospheric neutrinos as well as a residual of IBD-like NCinteractions of atmospheric neutrinos. The signal dominatesw.r.t. to the backgrounds over the entire observation window(white region). • Atmospheric NC events:
Due to the large reductionfactor for this background, it is hard to estimatefrom first principles how well the relative uncer-tainty on the residual rate can be estimated. How-ever, both the event rates before and after the ap-plication of background rejection cuts can be con-strained by extrapolation from the energy windowabove ∼
30 MeV.While it is most likely that the uncertainty on the NCbackground rate will remain the dominant source of sys-tematic uncertainty, it is hard to constrain its value basedon present knowledge. Therefore, we decided to leave thisquantity as an open parameter in the sensitivity studies,varying the relative uncertainty in a range from 5 to 20 %1 yr] · exposure [kt
20 40 60 80 100 120 140 160 180 ] s s i gn i f i c an c e [ FIG. 8. Significance of DSNB detection as a function of ex-posure. The curves correspond to a variation of the relativeuncertainty in the atmospheric NC background rates from 5to 20%. The upper black (blue) horizontal axis scale indicatesthe operation time of Theia25 (Theia100). yr] · exposure [kt ] s s i gn i f i c an c e [ FIG. 9. Significance of DSNB detection as a function ofexposure for the different DSNB flux models considered inour work. The black line corresponds to the fiducial model,whereas the high-flux and low-flux models are given in gray.The maximum exposure required for a 3 σ -detection (in caseof the low-flux model) is ∼
200 kt · yrs. The upper black (blue)horizontal axis scale indicates the operation time of Theia25(Theia100). of the predicted NC rate value. Fig. 8 displays the DSNBdetection significance as a function of the acquired expo-sure, varying the systematic uncertainty on the atmo-spheric NC background rates. Even under quite unfavor-able conditions, a 5 σ -discovery can be achieved based ona live exposure of 150 kt · yrs. This translates to ∼ Model dependence.
The measuring time required for apositive detection of the DSNB depends strongly on theunderlying flux model realized in Nature. This depen-dence is depicted in fig. 9, where the uncertainty of the predicted atmospheric NC rate is fixed to 20%. While amoderate exposure of only 40 kt · yrs suffices for a discov-ery in case of a high-flux model (see sec. III), 200 kt · yrswould be required for a detection at 3 σ level if the DSNBis best described by a low-flux model. This is still wellwithin the proposed operation times of both Theia100and Theia25. VII. COMPARISON WITH OTHERTECHNIQUES
The detection potential for the DSNB in a WbLS de-tector is best evaluated in comparison to the capabil-ities of other large-scale neutrino observatories comingonline during the next decade. At the time of writing,there is no experiment offering sufficient sensitivity for apositive detection, with the Water Cherenkov Detector(WCD) Super-Kamiokande (SK) providing the currentbest limit on the DSNB flux. Contrariwise, the currentlyprepared SK-Gd phase featuring the addition of gadolin-ium to the water target as well as Hyper-Kamiokandewith (HK-Gd) or without gadolinium (HK) all promisegenuine sensitivity to the DSNB signal [70, 73]. More-over, JUNO will offer the possibility for DSNB search inan organic liquid scintillator (oLS) [13], while the DUNEliquid-argon (LAr) TPCs promise sensitivity to the ν e flux component of the DSNB [72]. Signal over background.
Tab. IV permits a coarsecomparison of these experiments concerning theirprospects for a DSNB detection: The experiments aresorted by the underlying detector technology. The centercolumns list the fiducial masses foreseen, the scheduledstart of operation, and the observation window. To im-prove comparability, we used the cross-sections, massesand efficiencies given in the relevant experimental ref-erences [68–71] and calculated the expected event ratesfor the fiducial DSNB model [10] used throughout thepresent paper. Background rates were scaled for expo-sure. Based on this, we show the number of signal andbackground rates for a uniform exposure of 100 kt · yrs inthe third and second to last column. Finally, the sig-nificance of the signal over background, S/ √ S + B , isdisplayed in the last column.This comparison illustrates the exceptional perfor-mance of WbLS as a target material: With ∼
20 eventsdetected in 100 kt · yrs, WbLS is leading in signal accep-tance, i.e. the detection efficiency per unit exposure, andthus permits a fast accumulation of statistics. More im-portantly, WbLS features also the largest signal signifi-cance over background of ∼ Time projections of exposure.
While WbLS showsthe best performance per unit exposure, the sensitivity ofa given experiment depends effectively on the total targetmass it commands and the start of operation. The corre-2
Technology Experiment FM Start Energy Window Signal Signal BG S √ S+B [kt] [MeV] Efficiency [/(100 kt yr)]WbLS Theia25 20 2030 8 −
30 0.8 19.9 9.5 3.7Theia100 80 20.3 9.1 3.7LS JUNO 17.0 2021 10.2 − −
30 0.7 15.0 14.0 2.8HK 187 2027 20 −
30 0.9 4.6 39.3 0.7HK-Gd 187 2033 10 −
30 0.67 14.4 14.0 2.7LAr DUNE 20+20 2026 16 −
40 11.4 6.0 2.7TABLE IV. Comparison of neutrino observatories aiming at the detection of the DSNB signal. The first columns list experi-mental technique, abbreviation of the experiment ( see text ), fiducial mass (FM), projected start of data taking, and observationenergy window. For better comparability, the expected signal rate was recalculated for the DSNB fiducial model assumedthroughout the paper, using information from [68–71]. Background (BG) rates are scaled from the same sources. The givenrates for DUNE are based on [72]. The corresponding signal significance over background is calculated in the last column.For comparison, we show the expected performance of Theia25 and Theia100. The WbLS technique offers the best signalacceptance and highest signal significance. year D S N B e v en t s
10 sumTheia25Theia100JUNOSuperK+GdDUNEHyperKHyperK+Gd year S + BS / FIG. 10. Projections for the signal rates (left panel) and signal significance (right panel) of the relevant DSNB observatoriesover the next two decades. Optimistic scenarios correspond to dashed lines. The optimistic sum includes Theia100, and asecond tank for Gd-loaded HyperK. DUNE is not added to the overall sum, due to different neutrino channel. Assuming astart of data taking in 2030, Theia100 soon dominates the scene regarding both collected signal statistics and significance ofthe detection. Theia25 makes a slower start but provides an increasingly relevant contribution over ten years of data taking.See the text for a more detailed discussion. sponding projections for the individual experiments aredisplayed in fig. 10: The left panel shows the DSNB signalstatistics as a function of calendar year, the right panelthe corresponding signal significance. Again, we base ourprojections on the information given in Refs. [68–71].In the initial phase, the scene will be dominated by SK-Gd and JUNO since both will start data taking around2021. While SK-Gd will collect almost twice the statis-tics per year compared to JUNO, the JUNO S:B-ratiois significantly better, leading to a comparatively smalllead of SK-Gd in signal significance. DUNE is estimatedto start data taking 2026 with two caverns ( ∼
20 kt · yrs),while the third and fourth chamber will be ready one andthree year(s) later, respectively [71]. Around 2027, HKwill start data taking. It is worth noting that despite of the huge target mass and hence fast accumulation of sig-nal statistics, its sensitivity will not surpass that of SK-Gd. However, DSNB signal significance will sky-rocket assoon as gadolinium is added for HK-Gd. DUNE, on theother hand, will reach signal statistics and significancecomparable to SK-Gd about ten years into its operationphase (with the usual caveat that this concerns only the ν e signal of the DSNB).Both realizations of Theia have to be regarded againstthis background. In the following, we assume an opti-mistic start of operation in 2030. Set on this time scale,Theia100 would almost immediately emerge as the lead-ing DSNB observatory, providing both fast collection ofsignal statistics and excellent signal significance. Onceinitiated, HK-Gd could match the collected event num-3bers but at a considerably worse S:B ratio. Hence,Theia100 bears the promise of a speedy discovery ofthe DSNB signal and an important contribution to thefollow-up phase of DSNB spectroscopy.Even the smaller Theia25 will be able to make a rel-evant contribution. Despite a relatively long ramp-uptime ( >
10 yrs), Theia25 will eventually surpass JUNOin collected event statistics and approach the event num-bers of SK-Gd. The same is true for signal significance.Before the arrival of HK-Gd, arguably the most likelyscenario for a further exploration of the DSNB is a com-bined analysis of the data sets of all running experiments(i.e. SK-Gd and JUNO for the ¯ ν e component). Espe-cially in this scenario, a contribution of Theia25 wouldprove very important, in terms of both event statisticsand understanding of the crucial NC atmospheric back-ground. While not as spectacular as Theia100, Theia25would thus provide a substantial improvement of globalDSNB sensitivity. VIII. CONCLUSIONS
As laid out in the present paper, WbLS will provide anexcellent target material for the detection of the DSNB.We investigated this potential for two possible configura-tions of the future neutrino observatory Theia. Perform-ing a full analysis including IBD event selection, basicdiscrimination cuts as well as a selection on events fea-turing single-rings, high C/S ratio and no delayed decays of final-state radio-nuclei, we find a remaining signal ef-ficiency of >
80 % and a background residual of 1.3 % ofthe high-level selection cuts. Based on a statistical anal-ysis, we conclude that an exposure of ∼
150 kt · yrs willbe sufficient to claim a 5 σ discovery of the DSNB un-der standard assumptions. Longer measuring times maybe required in case detector uncertainties are unexpect-edly large or the DSNB is best described by a low-fluxmodel [10].While such comparisons are always difficult, we alsotried to evaluate the detection potential of a WbLS de-tector in the context of other large-scale neutrino ob-servatories (present and future) that feature sensitivityfor the DSNB. We conclude that per unit detector vol-ume, WbLS outperforms all other detection techniques.However, such an evaluation must take into account aswell the time scales on which measurements can be per-formed and the target masses that can be realized: Ifrealized a decade from now, Theia100 would still outper-form all other observatories within less than two years.The more modest Theia25 would take another ten yearsbefore having collected sufficient data to add significantlyto a global analysis of the DSNB flux and spectrum. ACKNOWLEDGMENTS
This work was supported by the Collaborative Re-search Center SFB1258 of the Deutsche Forschungsge-meinschaft (DFG), the Excellence Cluster Universe andExcellence Cluster ORIGINS. DK is grateful to ThomasErtl and Hans-Thomas Janka for their continued supportand guidance during the project. [1] A. Mirizzi, I. Tamborra, H.-T. Janka, N. Sa-viano, K. Scholberg, R. Bollig, L. Hudepohl, andS. Chakraborty, Supernova Neutrinos: Production, Os-cillations and Detection, Riv. Nuovo Cim. , 1 (2016),arXiv:1508.00785 [astro-ph.HE].[2] S. Ando and K. Sato, Relic neutrino background fromcosmological supernovae, New J. Phys. , 170 (2004).[3] J. F. Beacom, The Diffuse Supernova Neutrino Back-ground, Ann. Rev. Nucl. Part. Sci. , 439 (2010).[4] F. Vissani and G. Pagliaroli, The diffuse supernova neu-trino background: Expectations and uncertainties de-rived from SN1987A, Astron. Astrophys. , L1 (2011),arXiv:1102.0447 [astro-ph.HE].[5] C. Lunardini and I. Tamborra, Diffuse supernova neu-trinos: oscillation effects, stellar cooling and progeni-tor mass dependence, JCAP , 012, arXiv:1205.6292[astro-ph.SR].[6] K. Nakazato, E. Mochida, Y. Niino, and H. Suzuki, Spec-trum of the Supernova Relic Neutrino Background andMetallicity Evolution of Galaxies, Astrophys. J. , 75(2015), arXiv:1503.01236 [astro-ph.HE].[7] S. Horiuchi, K. Sumiyoshi, K. Nakamura, T. Fischer,A. Summa, T. Takiwaki, H.-T. Janka, and K. Ko-take, Diffuse supernova neutrino background from ex-tensive core-collapse simulations of 8-100M (cid:12) progeni- tors, Mon. Not. Roy. Astron. Soc. , 1363 (2018),arXiv:1709.06567 [astro-ph.HE].[8] A. Priya and C. Lunardini, Diffuse neutrinos from lu-minous and dark supernovae: prospects for upcom-ing detectors at the O (10) kt scale, JCAP , 031,arXiv:1705.02122 [astro-ph.HE].[9] K. Møller, A. M. Suliga, I. Tamborra, and P. B. Den-ton, Measuring the supernova unknowns at the next-generation neutrino telescopes through the diffuse neu-trino background, JCAP (05), 066.[10] D. Kresse, T. Ertl, and H.-T. Janka, Stellar Collapse Di-versity and the Diffuse Supernova Neutrino Background,to be submitted to ApJ (2020).[11] K. Bays et al. , Supernova Relic Neutrino Search at Super-Kamiokande, Phys. Rev. D85 , 052007 (2012).[12] H. Sekiya, The Super-Kamiokande Gadolinium Project,PoS
ICHEP2016 , 982 (2016).[13] F. An et al. , Neutrino Physics with JUNO, J. Phys.
G43 ,030401 (2016).[14] A. Gando et al. , A study of extraterrestrial antineutrinosources with the KamLAND detector, Astrophys. J. ,193 (2012).[15] M. Askins et al. , Theia: an advanced optical neutrino de-tector, The European Physical Journal C , 416 (2020).[16] H. Wei, Z. Wang, and S. Chen, Discovery potential for supernova relic neutrinos with slow liquid scintillator de-tectors, Phys. Lett. B769 , 255 (2017).[17] L. J. Bignell et al. , Characterization and Modeling of aWater-based Liquid Scintillator, JINST (12), P12009,arXiv:1508.07029 [physics.ins-det].[18] J. Caravaca, B. Land, M. Yeh, and G. Orebi Gann,Characterization of water-based liquid scintillatorfor cherenkov and scintillation separation (2020),arXiv:2006.00173 [physics.ins-det].[19] A. V. Lyashenko et al. , Performance of Large Area Pi-cosecond Photo-Detectors (LAPPD), Proceedings, 15thVienna Conference on Instrumentation (VCI2019):Vienna, Austria, February 18-22, 2019 , Nucl. In-strum. Meth.
A958 , 162834 (2020), arXiv:1909.10399[physics.ins-det].[20] T. Kaptanoglu, M. Luo, B. Land, A. Bacon, and J. Klein,Spectral Photon Sorting For Large-Scale Cherenkov andScintillation Detectors, Phys. Rev. D , 072002 (2020),arXiv:1912.10333 [physics.ins-det].[21] J. Alonso et al. , Advanced scintillator detector concept(asdc): A concept paper on the physics potential ofwater-based liquid scintillator (2014), arXiv:1409.5864[physics.ins-det].[22] M. Malek et al. , Search for supernova relic neutrinos atSuper-Kamiokande, Phys. Rev. Lett. , 061101 (2003).[23] H. Watanabe et al. , First Study of Neutron Tagging witha Water Cherenkov Detector, Astropart. Phys. , 320(2009).[24] H. Zhang et al. , Supernova Relic Neutrino Search withNeutron Tagging at Super-Kamiokande-IV, Astropart.Phys. , 41 (2015).[25] M. Wurm, F. von Feilitzsch, M. Goeger-Neff, K. A.Hochmuth, T. M. Undagoitia, L. Oberauer, andW. Potzel, Detection potential for the diffuse supernovaneutrino background in the large liquid-scintillator de-tector LENA, Phys. Rev. D75 , 023007 (2007).[26] M. Wurm et al. , The next-generation liquid-scintillatorneutrino observatory LENA, Astropart. Phys. , 685(2012).[27] R. M¨ollenberg, F. von Feilitzsch, D. Hellgartner, L. Ober-auer, M. Tippmann, V. Zimmer, J. Winter, andM. Wurm, Detecting the Diffuse Supernova NeutrinoBackground with LENA, Phys. Rev. D91 , 032005 (2015).[28] L. Wan et al. , Measurement of the neutrino-oxygenneutral-current quasielastic cross section using atmo-spheric neutrinos at Super-Kamiokande, Phys. Rev.
D99 , 032005 (2019).[29] J. F. Beacom and M. R. Vagins, GADZOOKS! Anti-neutrino spectroscopy with large water Cherenkov de-tectors, Phys. Rev. Lett. , 171101 (2004).[30] S. Agostinelli et al. , GEANT4: A Simulation toolkit,Nucl. Instrum. Meth. A506 , 250 (2003).[31] J. Allison et al. , Geant4 developments and applications,IEEE Trans. Nucl. Sci. , 270 (2006).[32] A. Strumia and F. Vissani, Precise quasielastic neu-trino/nucleon cross-section, Phys. Lett. B564 , 42 (2003).[33] H.-T. Janka and E. Mueller, Neutrino heating, convec-tion, and the mechanism of Type-II supernova explo-sions., A&A , 167 (1996).[34] K. Kifonidis, T. Plewa, H. T. Janka, and E. M¨uller, Non-spherical core collapse supernovae. I. Neutrino-drivenconvection, Rayleigh-Taylor instabilities, and the forma-tion and propagation of metal clumps, A&A , 621(2003). [35] L. Scheck, K. Kifonidis, H. T. Janka, and E. M¨uller,Multidimensional supernova simulations with approxi-mative neutrino transport. I. Neutron star kicks and theanisotropy of neutrino-driven explosions in two spatialdimensions, A&A , 963 (2006).[36] T. Ertl, H. T. Janka, S. E. Woosley, T. Sukhbold, andM. Ugliano, A Two-parameter Criterion for Classify-ing the Explodability of Massive Stars by the Neutrino-driven Mechanism, ApJ , 124 (2016).[37] M. Ugliano, H.-T. Janka, A. Marek, and A. Arcones,Progenitor-explosion Connection and Remnant BirthMasses for Neutrino-driven Supernovae of Iron-core Pro-genitors, ApJ , 69 (2012).[38] T. Sukhbold, T. Ertl, S. E. Woosley, J. M. Brown, andH. T. Janka, Core-collapse Supernovae from 9 to 120 So-lar Masses Based on Neutrino-powered Explosions, ApJ , 38 (2016).[39] M. T. Keil, G. G. Raffelt, and H.-T. Janka, Monte CarloStudy of Supernova Neutrino Spectra Formation, ApJ , 971 (2003).[40] H. Y¨uksel, M. D. Kistler, J. F. Beacom, and A. M. Hop-kins, Revealing the High-Redshift Star Formation Ratewith Gamma-Ray Bursts, ApJ Letters , L5 (2008).[41] G. J. Mathews, J. Hidaka, T. Kajino, and J. Suzuki, Su-pernova Relic Neutrinos and the Supernova Rate Prob-lem: Analysis of Uncertainties and Detectability ofONeMg and Failed Supernovae, ApJ , 115 (2014).[42] A. Barna and S. Dye, Global antineutrino modeling:A web application (2015), arXiv:1510.05633 [physics.ins-det].[43] T. K. Gaisser, T. Stanev, and G. Barr, Cosmic Ray Neu-trinos in the Atmosphere, Phys. Rev.
D38 , 85 (1988).[44] M. Honda, M. Sajjad Athar, T. Kajita, K. Kasahara, andS. Midorikawa, Atmospheric neutrino flux calculation us-ing the NRLMSISE-00 atmospheric model, Phys. Rev.
D92 , 023004 (2015), arXiv:1502.03916 [astro-ph.HE].[45] G. Battistoni, A. Ferrari, T. Montaruli, and P. R. Sala,The atmospheric neutrino flux below 100-MeV: TheFLUKA results, Astropart. Phys. , 526 (2005).[46] W.-L. Guo, Low energy neutrinos from stopped muonsin the Earth, Phys. Rev. D99 , 073007 (2019).[47] R. B. Firestone and V. Shirley, eds.,
Table of Isotopes ,8th ed. (A Wiley–Interscience Publication, John Wileyand Sons Inc., 1999).[48] D. R. Tilley, C. M. Cheves, J. L. Godwin, G. M. Hale,H. M. Hofmann, J. H. Kelley, C. G. Sheu, and H. R.Weller, Energy levels of light nuclei A=5, A=6, A=7,Nucl. Phys.
A708 , 3 (2002).[49] D. R. Tilley, J. H. Kelley, J. L. Godwin, D. J. Millener,J. E. Purcell, C. G. Sheu, and H. R. Weller, Energy levelsof light nuclei A=8,9,10, Nucl. Phys.
A745 , 155 (2004).[50] Y. Zhang et al. , First measurement of radioactive isotopeproduction through cosmic-ray muon spallation in Super-Kamiokande IV, Phys. Rev.
D93 , 012004 (2016).[51] G. Bellini et al. , Cosmogenic Backgrounds in Borexino at3800 m water-equivalent depth, JCAP , 049.[52] T. Hagner, R. von Hentig, B. Heisinger, L. Oberauer,S. Schonert, F. von Feilitzsch, and E. Nolte, Muon in-duced production of radioactive isotopes in scintillationdetectors, Astropart. Phys. , 33 (2000).[53] D. Mei and A. Hime, Muon-induced background studyfor underground laboratories, Phys. Rev. D73 , 053004(2006).[54] M. Aglietta et al. , Measurement of the neutron flux pro- duced by cosmic ray muons with LVD at Gran Sasso, in Proceedings, 26th International Cosmic Ray Conference(ICRC) (1999) pp. 44–47.[55] F. An et al. , Improved Measurement of Electron An-tineutrino Disappearance at Daya Bay, Chin.Phys.
C37 ,011001 (2013).[56] C. Andreopoulos et al. , The GENIE Neutrino MonteCarlo Generator, Nucl. Instrum. Meth. A , 87 (2010),arXiv:0905.2517 [hep-ph].[57] C. Andreopoulos, C. Barry, S. Dytman, H. Gallagher,T. Golan, R. Hatcher, G. Perdue, and J. Yarba, Thegenie neutrino monte carlo generator: Physics and usermanual (2015), arXiv:1510.05494 [hep-ph].[58] M. Tanabashi et al. (Particle Data Group), Review ofParticle Physics, Phys. Rev. D , 030001 (2018).[59] A. Back et al. (ANNIE), Accelerator Neutrino NeutronInteraction Experiment (ANNIE): Preliminary Resultsand Physics Phase Proposal (2017), arXiv:1707.08222[physics.ins-det].[60] H. Ejiri, Nuclear deexcitations of nucleon holes associ-ated with nucleon decays in nuclei, Phys. Rev. C48 , 1442(1993).[61] K. Abe et al. , Measurement of the neutrino-oxygenneutral-current interaction cross section by observing nu-clear deexcitation γ rays, Phys. Rev. D90 , 072012 (2014).[62] Y. A. Kamyshkov and E. Kolbe, Signatures of nucleondisappearance in large underground detectors, Phys.Rev.
D67 , 076007 (2003).[63] A. Koning and D. Rochman, Modern Nuclear Data Eval-uation with the TALYS Code System, Nucl. Data Sheets , 2841 (2012).[64] M. Grassi, J. Evslin, E. Ciuffoli, and X. Zhang, Vetoingcosmogenic muons in a large liquid scintillator, J. HighEnergy Phys. (10), 32.[65] F. Ajzenberg-Selove, Energy levels of light nuclei A =13-15, Nucl. Phys.
A523 , 1 (1991).[66] R. M¨ollenberg,
Monte Carlo Study of Solar B Neutri-nos and the Diffuse Supernova Neutrino Background inLENA , Phd, Technical University Munich (2013).[67] W. A. Rolke, A. M. Lopez, and J. Conrad, Limits andconfidence intervals in the presence of nuisance parame-ters, Nucl. Instrum. Meth.
A551 , 493 (2005).[68] M. R. Vagins, Preparing for a gd-loaded super-k, Talk atRevealing the history of the universe with undergroundparticle and nuclear research 2019 (2019).[69] Y. Takatomi, Astrophysical neutrinos at hyper-kamiokande, Talk at Topics in Astroparticle andUnderground Physics (2019).[70] K. Abe et al. (Hyper-Kamiokande), Hyper-kamiokandedesign report (2018), arXiv:1805.04163 [physics.ins-det].[71] B. Abi et al. (DUNE), The dune far detector interim de-sign report volume 1: Physics, technology and strategies(2018), arXiv:1807.10334 [physics.ins-det].[72] A. G. Cocco, A. Ereditato, G. Fiorillo, G. Mangano, andV. Pettorino, Supernova relic neutrinos in liquid argondetectors, JCAP , 002.[73] K. Abe et al.et al.