5 Reasons to expect an 8 MeV line in the SN 1987A neutrino spectrum
EEvidence for an 8 MeV line in the SN 1987A neutrinospectrum and four reasons to expect one
Robert Ehrlich George Mason University, Fairfax, VA 22030 ∗ (Dated: February 24, 2021)Evidence is reported for an 8 MeV neutrino line associated with SN 1987A based on an analysisof 997 events recorded in the Kamiokande-II detector on the day of the supernova. Given that theenergy of the line nearly coincides with the peak of the background, it is important to note thatthere are four reasons to expect such an 8 MeV line based on other data.keywords: neutrino line, monochromatic neutrinos, SN 1987A, supernova, dark matter, galacticcenter PACS numbers:
I. INTRODUCTION
No neutrino lines have as yet been observed from astro-physical sources other than the sun. The sun, of course,does have lines in its neutrino spectrum, [1] and the Be line has been directly observed. [2] We here report evi-dence for an 8 MeV neutrino line in the SN 1987A spec-trum, which satisfies these criteria:1. The observed line has the right shape, i.e., a Gaus-sian having a width equal to the energy resolutionof the detector.2. Both the shape and height of the background atopwhich the line sits can be found independent of thedata.3. The statistical significance of the excess abovebackground is extremely high4. There is a theoretical basis for a prominent neutrinoline in the supernova spectrum.5. There are four reasons to expect the line to occurat the observed energy based on other data.6. There are no empirical or theoretical contradictionsruling out a prominent 8 MeV neutrino line fromSN 1987A.The data in support of an 8 MeV neutrino line arethe 997 events recorded on the date of SN 1987A by theKamiokande-II detector [3]. The analysis reported here,however, has not been endorsed or rejected by any mem-ber of the Kamiokande collaboration. The possibility ofan 8 MeV neutrino line was first raised in the Appendixto ref. [4], but the evidence for it has now gotten consid-erably stronger. Before considering the Kamiokande-IIdata, let us first deal with criteria 4 and 5 above. ∗ Electronic address: [email protected]
II. GALACTIC CENTER γ -RAYS The galactic center (GC) has been considered as a pos-sible place for dark matter (DM) annihilations to occur,in which case the γ − rays from the GC could be the resultof XX → e + e − followed by e + e − → γ. We can, there-fore, learn about the possible presence of DM near theGC by examing the spectrum of γ − rays from that source.There is also a direct connection between XX annihila-tions and neutrino lines. Thus, given the annihilation ofcold DM X particles in the reaction XX → ν ¯ ν, energyconservation requires the ν, ¯ ν be nearly monochromaticwith E = m X .Fig. 1 shows the predicted enhancement above back-ground for the GC γ − ray spectrum. The four curvescorrespond to different m X values. These are found byassuming that the e + are created in XX → e + e − withan initial energy E = m X . Of those e + , we assume 97%will annihilate at rest yielding the 511 keV line, while theremaining 3% propagate in a neutral medium before an-nihilating in flight [5]. Also shown in the figure is the GC γ − ray flux data from four instruments. Note that mostof the data and three of the four enhancement curves pre-viously appeared in refs. [6, 7], but the author has addedthe 8.3 MeV enhancement curve and the 7 OSSE pointsfrom Ref. [8]. It can be seen that the enhancement foreach m X above the straight line background power law(with index 1.55 [9]) extends from E min = 511 keV to E max = m X . The data in Fig. 1 can be seen to be con-sistent with m X = 10 MeV (black curve), with χ = 7 . ,p = 89% , dof = 13 . In contrast, the fit to the null hypoth-esis, i.e., the dashed line power law, is completely unac-ceptable: χ = 960 . Acceptable fits to the data in Fig.1 can only be found for the range: m X = 10 +5 − . MeV.Thus, the null hypothesis is excluded by N = 10 / . ∼ a r X i v : . [ a s t r o - ph . H E ] F e b FIG. 1: Flux, i.e., E × dFdE ( cm − s − ) versus energy for γ − rays from the inner galaxy, as measured by: SPI(opencircle), COMPTEL (open squares), EGRET (filled circles),and OSSE (filled triangles) fr. All but the OSSE data (fromRef.[8]) are from ref. [7]. The computed enhancements abovethe straight line are for positrons injected into a neutralmedium at initial energies E = m X = 5 , . , ,
50 MeVdisplayed respectively as: lower grey, dotted, black, and up-per grey. The sloped line is a power law (index 1.55) fit todata at high and low energies. for m X ∼ M eV, and hence a neutrino line with this en-ergy. An m X ∼ M eV for a Dirac fermion DM particleis, incidentally, within the bounds set by Planck satellitedata on light WIMPs that were in equilibrium with eitherthe neutrino or electron sector in the early universe. [10].
III. OTHER EVIDENCE FOR M X ∼
10 MEV
If cold dark matter X particles having m X ∼ M eV exist and their annihilation yields monochromatic e + e − pairs as suggested in the previous section, it is reasonableto suppose that the reaction proceeds via some media-tor particle Z’ as in XX → Z (cid:48) → e + e − , whose mass is m Z (cid:48) = 2 m X , by energy conservation. The natural placeto look for such a Z’ would be in a nuclear physics exper-iment, where the decay of some nuclear excited state N ∗ produced e + e − pairs via N ∗ → N + Z (cid:48) → N + e + e − . Ofcourse, most of the time when e + e − pairs are observed itwould be when the mediator particle is a photon, so theexistence of such a Z’ would be revealed by an enhance-ment to that reaction, i.e, an excess of e + e − pairs havinga specific opening angle, corresponding to m Z (cid:48) . In 2016exactly such an enhancement was reported by Kraszna-horkay et al. (the Atomki group) for e + e − emissions inthe reaction Li ( p, γ ) Be . [11] Their result implied an in-termediate short-lived Z’ particle (sometimes called X17)with mass m = 16 . ± . Be, i.e.: Be ∗ → BeZ (cid:48) , followed by Z (cid:48) → e + e − . In 2020 the Atomki group has reported the sameanomaly in the decay of excited helium atoms as theyearlier observed in Be . [12]. A particle having a 16.7MeV mass would also be expected to be found in someaccelerator experiments. However, the NA64 experiment(and others) at the CERN SPS have not observed it. [13].On the other hand, these negative results do not contra-dict those in refs. [11, 12] because they were not sensitiveto a small range of particle lifetimes consistent with thatreported in ref. [11] – see Fig. 1 in ref. [13]. Thus, theX17 claim is not yet contradicted by those of NA64 orany other experiments. On the other hand, Aleksejevs etal. [14] have cast doubt on the existence of the X17, byshowing the Atomki anomaly can be explained by addingthe full set of second-order corrections and the interfer-ence terms to the Born-level decay amplitudes.If the anomalies observed in refs. [11, 12] really aredue to a new Z’ boson, this particle has no place in thestandard model, and it could be the mediator of a fifthforce. [15] Moreover, as already noted, for cold DM, theend product of XX → Z (cid:48) → ν ¯ ν (which is the only otherZ’ decay mode according to ref. [16]) would be nearlymonochromatic ν and ¯ ν pairs having E ν = 8 . ± . Z (cid:48) particle is therefore the ideal candidatefor creating 8 MeV neutrino and antineutrino lines. Twoadditional reasons for expecting an 8 MeV neutrino linefrom SN 1987A are provided in the Conclusions section. IV. POSSIBLE SUPERNOVA NEUTRINO LINES
Many researchers have suggested that dark mattermight collect in the core of some stars. [17] DM annihila-tion triggering a supernova is plausible because withoutsuch an “extra” energy, shock wave stalling has been adifficulty with most supernova models, the best of whichhave elements that can still only be understood in quali-tative terms. [18] Moreover, Fayet et al. [19] have shownthat light ( <
10 MeV) dark matter particles can play asignificant role in core-collapse supernovae, if they haverelatively large annihilation and scattering cross sections.In this case, the DM in supernovae would cool on a timescale perhaps >
100 times that in the standard scenario,and they would emit neutrinos with significantly smallerenergies. [19]The preceding observations therefore provide a the-oretical basis for the possibility of a long-lasting O ( ∼ M eV ) neutrino line in the supernova spectrum createdby DM annihilation (criterion 4). It may be true that asof 2020 a self-consistent 3D simulation with detailed neu-trino transport has finally achieved a neutrino-driven ex-plosion with properties similar to SN 1987A without darkmatter. [20] However, even if DM may not be required totrigger a neutrino-driven explosion, the presence of largeamounts of DM in the stellar core could still play a role inthe explosion and be the source of significant long-lastingmonochromatic emissions.
FIG. 2: Histogram of N hit values for events in Fig. 4 inref. [3] The solid and dashed curves are two versions of thebackground for the detector that were extracted from pub-lished data on a search for B solar neutrinos. See AppendixA in ref. [4] for details on the computation of the two versionsof the background. N hit = 17 corresponds to 7 . ± . V. ANALYSIS OF KAMIOKANDE DATA
The largest of the four detectors operating at the timeof the SN 1987A observation was Kamiokande-II. [3] Thisdetector recorded neutrino arrival times and their ener-gies, which could be deduced from the “visible” energies, E vis based on E ν = E vis + 1 . ν e + p → n + e + . In addition toobserving the main 12-event burst, Kamiokande-II alsorecorded 997 events occurring during eight 17-min longintervals during several hours before and after the burst.Figs. 4 (a)-(h) of Ref. [3] show scatter plots for eachevent displaying the number of “hits,” N hit , (PMT’s ac-tivated) versus the event occurrence time during that∆ t = 8 × .
094 day time interval. Those eightdot plots were digitized by the author who counted thenumber of times various N hit values occurred. The N hit frequency distribution is shown in Fig. 2. Note that the N hit values are found to be proportional to the visibleenergy, i.e., E vis = cN hit M eV with c = 0 . ± , asshown in Fig. 4(a) of ref. [4]. Thus, Fig. 2 is actually aspectrum for the events observed over several hours, withthe peak at 17 hits corresponding to E ν = 7 . ± . M eV . A. Finding the background using B data Clearly, any claim of a peak above background in Fig.2 depends critically on how the background is deter-mined. If the background is correctly represented herethe peak would have very high statistical significance( ∼ σ ) – see section A2.6 of ref. [4] for details. The twodisplayed background curves for the detector have beenfound based on a 1989 publication by the Kamiokande-IICollaboration (K-II) on a search for solar neutrinos fromthe reaction B → Be ∗ + e + + ν e . [21]. The beginning of the 450 day data-taking period preceded the date ofSN 1987A, but most of it was many months afterwards.This B data to define the SN background spectrum –see Appendix A of ref. [4] for details on how the two ver-sions of the background are found. The excess countsabove either background in Fig. 2 can be well-fit by aGaussian curve centered on E max = 7 . ± . E/E = 22% / (cid:112) E/ , [3] consistent withwhat would be expected for an 8 MeV neutrino line B. Validity of using B data as background The most obvious flaw with using the data from the B neutrino search for the background on the day of SN1987A would be if the background count rate in the detec-tor were time dependent. Here we suggest some reasonswhy any such time variation of the background is likelyto have been small. Four other concerns about using the B data to find the background on the day of SN 1987Aare discussed in appendix A of ref [4].According to ref. [22], “The stability of the gain of thedetector was monitored using a γ − ray source. The rel-ative gain of the detector monitored with this source...is stable within ±
2% over four years of operation from1987 to 1990.” In addition, ref. [3] notes that there was notime dependence in the background count rate for sev-eral months prior to SN 1987A apart from “small per-turbations introduced by efforts to reduce the amount ofRn dissolved in the tank water.” Such ‘small perturba-tions’ in radon level and background count rate due tothe change from periodic to continuous cleaning (madeafter SN 1987A) would very likely be small compared tothe roughly ∼ ∼
700 count excess abovebackground in Fig. 2.
C. Finding the background without the B data Despite the arguments in the previous section, it couldbe risky to assume a constant background over time givenvarious improvements made to the detector. [22] If oneexpects to see a spectral line or a resonance at some spe-cific energy E , there is a standard procedure to find theprominence of the line in the data when the functionalform of the background is unknown. The procedure is tosee if the data at energies away from E can be fit withsome simple function, and then use that fitted functionto find the excess counts at energies near E = E . If theline exists, the excess counts must be found to have theGaussian or Lorentzian shape and width expected for aspectral line for a detector having some known energyresolution – see for example Fig. 4 in ref. [23]. Followingthis standard procedure we ignore the B data entirelyand fit a Gaussian curve to the ten data points in Fig. 2that lie outside the central range, 13 < N hit < . Theresulting fit is virtually indistinguishable from the solidbackground curve shown in Fig. 2 found using the B data. Of course, normally, when the preceding techniqueis used to find evidence for a signal above an unknownbackground, the two curves have very different shapes,and the evidence for the signal is much more certain. VI. POSSIBLE CONTRADICTIONS
Here we discuss three possible contradictions based ontheory or observations to an 8 MeV neutrino line fromSN 1987A.
A. Lack of time variation in line prominence
If the 8 MeV line is real, one might have expected thatof the eight 17-min time intervals for which ref. [3] pro-vided data, those time intervals closest to the 12-eventburst would show a greater excess above background,whereas no such variation is found in the data. Thelack of any such time variation in the strength of theline is very worrisome, but it would be consistent withthe long-lasting luminosity expected for neutrinos emit-ted via light dark matter annihilation. [19] If the DMannihilation in supernovae were strongly coupled to neu-trinos, with neutrinos and light DM particles decouplingat < ∼ . M eV rather than the usual 8 MeV for neu-trinos, the supernova cooling time scale would be largerby a factor of ∼
20 than in the standard scenario. [19]Furthermore, in the interior of the interior of the proto-neutron star, the cooling time scale would be a factor ≥
100 larger [19]. Such long cooling times would be ex-pected to be matched by long heating times for the darkmatter, so it would not be surprising to find that theemissions from DM annihilation are not simply confinedto times after the main 12 event burst. The neutrinoemission of massive stars before a supernova has beenstudied by many authors and they all agree on signifi-cant, potentially detectable neutrino luminosities, see thereview in Kato et al. [24]
B. Non-observation in diffuse supernova searches
The diffuse supernova neutrino background is a theo-retical population of neutrinos cumulatively originatingfrom all of the supernovae events which have occurredthroughout the Universe for which only upper limits cur-rently exist. Given the large number of excess eventsabove background comprising the 8 MeV neutrino line,one might expect it would have been detected in pre-vious searches for diffuse supernovae. However, thosesearches either focussed on neutrinos having energies wellabove 8 MeV [25, 26] or alternatively neutrinos emittedin seconds-long bursts. [27] Thus, despite those negativeresults, it might be possible to see such emissions even with today’s neutrino detectors, especially those whichhave: (a) operated over many years, (b) a low energythreshold, and are (c) sufficiently distant from a nuclearreactor – the main background at 8 MeV. In doing such asearch one might look for a significant excess of counts inany day-long time interval for events in an energy bandcentered on E ν = 8 M eV.
C. Impossible number of 8 MeV neutrinos
A third problematic aspect of the claimed 8 MeV neu-trino line from SN 1987A is the sheer magnitude of theexcess events in Fig. 2 being ∼
700 above background.Recall that these data were from eight 17 minute longtime intervals, i.e., roughly two hours duration chosen atrandom from a total time of ten hours. Therefore, it isreasonable that given the lack of time variation seen overthose ten hours, the total time during which monochro-matic neutrinos were emitted by SN 1987A would need tobe at least 12 hours duration. In that case, one would es-timate the number of monochromatic neutrinos in those12 hours to be 700 × / ∼ , , as compared to 12events seen in the main burst. In other words if wetake the usual estimate of 10 neutrinos emitted fromSN 1987A, the number associated with DM annihilationwould be ∼
400 times more or 4 × . Let us now seeif such a value is even remotely possible.The star giving rise to SN 1987A was estimated tohave 18 solar masses. Let us assume it consisted ofhalf DM, and that, as was assumed earlier, the DM Xparticles, had a mass of 8 MeV, yielding in total ∼ X particles in the star. We, therefore, find that atleast a half the DM in the star would need to annihilateproducing 8 MeV neutrinos to match the number ∼ × , of emitted monochromatic in an emissionduration of half a day. Thus, under our assumptions,finding as many as ∼
700 neutrinos constituting anobserved 8 MeV spectral line in the Kamiokande IIdata is not impossible. Lending further credence tothe notion of such a large monochromatic neutrinoemission, Brito et al. [28] have shown that stellar coresdo not necessarily collapse when they grow beyondtheir Chandrasekhar limit, and that large fractions ofDM in stellar cores are not ruled out. [29] In addition,the fact of a more long-lasting process would be lesslikely to result in instabilities and a large departurefrom spherical symmetry, which would result in a largerfraction of the DM particles annihilating.
VII. CONCLUSIONS
The spectrum of 997 events recorded by theKamiokande-II detector on the day of SN 1987A are con-sistent with an 8 MeV line atop a background. The ex-cess of ∼
700 events above background is about the samewhether the background is found from the SN 1987A datathemselves or alternatively from other data taken monthsbefore and after SN 1987A. If the neutrino line is real, ithas been well-hidden since 1987, because it occurs veryclose to the peak of the background. This fact might or-dinarily justify extreme skepticism about the existence ofsuch a line. However, in the present case, a less skepticalassessment is warranted based on the six criteria enu-merated at the start of the paper, all of which have beensatisfied. Moreover, even if one conservatively disregardsboth the Kamiokande data and the Atomki anomaly asevidence for an 8 MeV neutrino line, the γ − ray data fromthe galactic center alone provides strong support (6 σ ) forcold DM of mass m X = 10 +5 − . M eV, whose annihilationvia ν ¯ ν would yield an E ∼ M eV neutrino line.We conclude this paper by offering two further rea-sons to expect an 8 MeV neutrino line from SN 1987Ain addition to the Atomki anomaly and the GC gammaray data. The first is based on the controversial MontBlanc neutrino burst on the day of SN 1987A, which hasbeen disregarded by most physicists with a few excep-tions [33], owing to its 5 hour early arrival and its absencein the Kamiokande II detector. Unlike the bursts seen inthe other three detectors then operating which all hadmuch higher energies, the five Mont Blanc neutrinos allhave energies consistent with the value 8 MeV, [30, 31]in which case they could be the result of the 8 MeV neu-trino line. The absence of such a 5 hour early burstin the Kamiokande II data can be explained by the higher energy threshold of that detector (20% efficientat E ν = 8 M eV ), [3] and the synchronization of the twodetectors being no better than ∼ ± m ν < eV. [34] Acknowledgements
The author wishes to thank the following people whomade helpful comments on earlier versions of this paper:Vitor Cardoso, Alan Chodos, Robert Ellsworth, PierreFayet, Jonathan Katz, Yury Kolomensky, Andrey Yudinand Xilin Zhang. [1] J. N. Bahcall, Phys. Rev. D , 3923 (1994).[2] The Borexino Collaboration, M. Agostini, K. Al-tenm¨uller, et al., Nature , 505–510 (2018).[3] K. S. Hirata, et al. (Kamiokande collaboration), Phys.Rev. D , 448-458 (1988).[4] R. Ehrlich, Astropart. Phys., , 21-29 (2018);arxiv.org/1701.00488.[5] P. J. Jean, et al., Astron. and Astrophys., , 579(2006).[6] N. Prantzos, et al. Rev. Mod. Phys., , 1001 (2011).[7] P. Sizun, M. Cass´e, and S. Schanne, Phys. Rev. D ,063514[8] R. L. Kinzer et al., ApJ, , 282, (2001).[9] L. Bouchet, J. P. Roques, and E. Jourdain, ApJ ,1772 (2010).[10] R. Wilkinson, A. Vincent, C. Boehm and C. McCabe,Phys. Rev. D 94, 103525 (2016).[11] A. J. Krasznahorkay et al., Phys. Rev. Lett. , 042501(2016).[12] D. S. Firak et al., EPJ Web of Conferences , 04005(2020).[13] E. Depero et al., Eur. Phys. J. C , 071803 (2016).[16] C-S. Chen et al., Int. J. Mod. Phys. A , 1750178 (2017).[17] V. D. Barger, W. Y. Keung and G. Shaughnessy, Phys.Lett. B, , 190 (2008).[18] B. Muller, et al., MNRAS (1) 491–513 (2017). [19] P. Fayet, D. Hooper, and G. Sigl, PRL , 211302 (2006).[20] R. Bollig et al., submitted to ApJ, arXiv:2010.10506.[21] K. S. Hirata, et al. (Kamiokande collaboration), Phys.Rev. Lett., , 1 (1989).[22] K. S. Hirata, et al. (Kamiokande collaboration), Phys.Rev. , 8 (1991).[23] G. Aad et al. (Atlas Collaboration), Phys. Rev. D ,052004 (2014)[24] C. Kato et al., ApJ, 8¯48, 48 (2017).[25] The Super-Kamiokande Collaboration, K. Bays et al.,Phys. Rev. D, (2012) 052007 [arXiv:1111.5031].[26] B. Aharmim et al. (SNO Collaboration), Phys. Rev. D, , 062006 (2020).[27] Yu. F. Novoseltsev, et al., J. of Exp. and Theor. Phys., (1) 73–79 (2017).[28] R. Brito, V. Cardoso, and H. Okawa, Phys. Rev. Lett., , 111301 (2015).[29] R. Brito et al., Phys. Rev. D , 044045 (2016).[30] M. Aglietta et. al., Nucl. Phys. B (Proc Suppl.) (12) 1315 (1987).[32] R. Ehrlich, Astropart. Phys., (2013) 1–6;arxiv.org/1204.0484.[33] S. Giani, STAIF-99 Proceedings, American Institute ofPhysics (1999).[34] R. Ehrlich, Lett. in High En. Phys.,2