Arxiv:1812.00035V1

Home , Solar neutrino problem

arXiv:1812.00035v1 [astro-ph.SR] 30 Nov 2018 ftoedsusosaeete ocpuloverviews, conceptual either m are but textbooks discussions in those and of literature journal the in extensively tha evolution stellar topic. and other structure stellar grad on and courses undergraduate uate advanced in enthus students more from attract responses hav oscillations we neutrino reasons, solar related that Prizes and found Nobel these 2015 for and Presumably the 2002 Physics. of the in importance both the in that recognized then was fitting ideas is ne It that what be. for could clues physics possible suggested has and physics cle ahmtc n ai unu ehnc talvlcharac- level a w at initially mechanics familiar quantum is basic reader and the mathematics that only assumes fra It mathematical accessible work. an in within non-specialist alike student satisfy and will tor that solar level a the at of problem resolution trino the and oscillations neutrino solar others. to not perhaps but often cognoscenti, the speed, (relative to required are vious that the derivations in to steps up important th omitting come for to effort non-expert remedial systematicall substantial telligent more a theory requires beautiful typically the these but formulates of second ‘how’ the and them- much ‘why’ ideas; omits by mathematical but sources deep well the concepts ideal of the be introduces to first The neither in selves. find topic will this no incorporate courses, are and their who understand instructors to wish and stu- but problem, experts graduate this in interactions. beginning interested weak and dents the undergraduate of advanced theory Most field ba quantum substantial a in has reader ground the that assuming exten written o theory are neutrino but formulation solar treat or mathematical theory, neutrino systematic solar a omitting essarily h te adi a eosrtdcnlsvl htteei there that conclusively demonstrated the beyond has physics it o hand while other subfields, astrophysics the of number a in directions new the the On that demonstrated (SSM) physics. has on it particle impact hand, elementary one enormous and an astrophysics had both has oscillations flavor neutrino etiooclain nvcu n nmte r described are matter in and vacuum in oscillations Neutrino hsppratmt opoieasseai nrdcinto introduction systematic a provide to attempts paper This h eouino h oa etiopolmi em of terms in problem neutrino solar the of resolution The ok eakbywl,adhssgetdprofitable suggested has and well, remarkably works n lmnaydfeeta qain n matrices. wi and familiar equations this is differential learn reader elementary to and the like that would only who preparation li student theoretical courses inquisitive in the at students in aimed graduate to like beginning would but or theory field undergraduate quantum in experts not are who opeesv nrdcint h hoyo h oa neut solar the of theory the to introduction comprehensive A tnadMdl(SM) Model Standard .INTRODUCTION I. eateto hsc n srnm,Uiest fTenness of University Astronomy, and Physics of Department ai nrdcint h hsc fSlrNeutrinos Solar of Physics the to Introduction Basic A a ig ainlLbrtr,OkRde ense 37830 Tennessee Ridge, Oak Laboratory, National Ridge Oak tnadSlrModel Solar Standard feeetr parti- elementary of Dtd eebr4 2018) 4, December (Dated: 1–3 y ob- ly) sively struc- any n iastic ieGuidry Mike in- e a Billings Jay neu- nec- me- ck- ost 4–6 ith se to w y n e s f - t htebsc fqatmmcaisi ia notation Dirac in mechanics quantum of basics the th tnadSlrModel, in semi-phenomenological Solar the solar pro- from Standard the obtained are in be that may distribution These density neutrinos rior. flavor electron the the the of and spectrum are duced, and reactions type) neutrino-producing (neutrino solar these ing evolution. stellar and gen structure more stellar of for implication, models by eral and, ap- specifically stringent models core a solar solar and of detection, the to in prior detection minutes conditions 8.5 their of proximately Thus snapshot matter. a solar provides p the little Earth with with light, react of as- to speed in ability the an essentially core importance at the large Sun in of the produced leave is are these neutrinos This the that because Earth. and trophysics on neutrinos detected reaction produce be these that can of interactions some di- that of weak note be these are to not of except will details here, and The concern documented rect focus. well our by are be provided reactions will is nuclear they energy so Sun’s chains, PP the the of 99% almost temperature this: accomplish can the in mechanisms (1) energy different release Two and process. helium the into hydrogen convert that core w ed htaedrcl eeatt h aka hand. at task the to relevant the directly of are aspects that only fields two describing pragmatic, approa decidedly The be fixing. required will understandi both, perhaps our or or neutrino, Sun, the the of of implyi understanding problem” our neutrino either neu- “solar that why a of to and led observations particles, trino elementary of Model Standard the presentatio the of part as introduced as are assumed, concepts is required theory field part quantum elementary relativistic or astrophysics, physics, par solar No in major. background physics ticular undergraduate advanced an of teristic I OA ETIO N H TNADMODEL STANDARD THE AND NEUTRINOS SOLAR II. eatohsc rqatmmcais ti also is it mechanics; quantum or astrophysics ke o u pcfi upssalta srlvn concern- relevant is that all purposes specific our For its in place taking reactions nuclear by powered is Sun The ebgnwt ocs vriwo oa etio n of and neutrinos solar of overview concise a with begin We oi nteron h rsnainasmsas assumes presentation The own. their on topic † oprt hs da noisrcino advanced of instruction into ideas these corporate ∗ iopolmi ie hti ie tinstructors at aimed is that given is problem rino Pchains, PP .Nurnsi h tnadSlrModel Solar Standard the in Neutrinos A. e nxil,T 37996–1200 TN Knoxville, ee, n 2 the (2) and hc obnsteter fstellar of theory the combines which N cycle. CNO tispeetcore present its At rob- icle n. test the te- on ng ng ch se d s - - 2

Ga Cl Super-K 1012 ﬂux suppression from that expected indicates that the issue is real, and implies some combination of (1) the Standard Solar pp 1010 Spectrum of Model fails to describe properly the conditions in the Sun’s ) solar neutrinos core (we don’t understand the Sun), or (2) the properties of -1 13

s N 8 neutrinos differ from that assumed in the prediction of Fig. 1

-2 10 15O (we don’t understand the neutrino). To investigate these pos- 8 17 B 6 F sibilities, let us now consider the description of neutrinos and 10 the weak interactions in the Standard Model of elementary particle physics that is implicit in the prediction of Fig. 1. 104 pep hep Flux at 1 AU (cm Be 7 2 Be 10 7 B. The Standard Model and Weak Interactions 0.1 0.2 0.5 1.0 2.0 5.0 10.0 20.0 Neutrino energy (MeV) The Standard Model of elementary particle physics is the FIG. 1: The SSM neutrino spectrum, with fluxes scaled to the dis- most comprehensive and well-tested theory yet conceived in tance of Earth from the Sun.8 Labels refer to specific reactions in the theoretical physics, and it accounts for an extremely broad PP chains and CNO cycle. The sensitive regions for various experi- range of phenomena. However, our purposes are specific ments are indicated above the graph, with Ga denoting the GALLEX and require only four basic ingredients and one speculation and SAGE gallium detectors, Cl denoting the Homestake (Davis) from the SM: (1) the generational structure of its elementary chlorine detector, and Super K denoting the Super Kamiokande wa- fermions (matter fields) and associated lepton-number con- ter Cherenkov detector (see Table I). servation laws (see §IIB1), (2) the corresponding implication that neutrinos must be identically massless and that only 26 left-handed neutrinos and right-handed antineutrinos enter the Solar electron weak interactions (which builds the maximal breaking of par- 25 number density ity symmetry into the weak interactions by fiat), (3) the em- ) pirical observation that in the quark sector the weak eigen- -3 24 states are not congruent with the mass eigenstates, implying cm

e 23

n flavor mixing (see §IIB2 and §IIB3), (4) the long-standing ( Neutrino speculation10 that flavor oscillations might occur in the neu- 10 production 22 trino sector also (which would imply finite masses for neu- Log trinos and thus a violation of the minimal Standard Model) n 26 -10.54(R/R ) 21 e = 1.475 x 10 e 8B pp (see §III), and (5) that the weak interactions can occur through 7 Be charged currents mediated by the W ± bosons and neutral cur- 20 0 0.2 0.4 0.6 0.8 1 rents mediated by the Z0 boson (see §IIIA). R/R

FIG. 2: Electron number density as a function of fractional solar radius from the Standard Solar Model.8 The dashed line is an expo- 1. Generational Structure nential approximation that will be employed in discussing the MSW effect. Regions of primary neutrino production in the PP chains are indicated, using labels from Fig. 1. The fermions of the Standard Model are grouped into three families or generations, as displayed in Fig. 3. In the SM, interactions across family lines are forbidden. For the lep- structure and evolution with the values of key parameters in- tons this is implemented formally by assigning a lepton family ferred from measurements or theory to provide a comprehen- number to each particle and requiring that it be conserved by sive and predictive model for the structure and evolution of the interactions. The term flavor is used to distinguish the quarks Sun.7 The spectrum of solar neutrinos predicted by the SSM and leptons of one generation from another. Thus, (νe, νµ ,ντ ) is displayed in Fig. 1, and the electron number density as a are differentflavors of neutrinos, and (u, d, s, ...) are different function of radius from the SSM is displayed in Fig. 2. flavors of quarks. When solar neutrinos are detected on Earth by chemistry- Also displayed in Fig. 3 are characteristic mass scales for based or normal-water Cherenkov detectors, the flux is found quarks and neutrinos within each generation. The neutrino to be systematically less than that predicted by the SSM (see masses are upper limits, since no neutrino mass has yet been Fig. 1), in an amount that depends on the energy of the neu- measured directly. They are seen to be tiny on a mass scale set trinos; Table I illustrates. Typically for the highest neutrino by the quarks of a generation, which is a major unresolved is- energies the observed flux is around 30% of that expected sue. There is no known fundamental reason why the neutrino while for the lowest energies about 50% the expected num- masses should be identically zero (as assumed phenomeno- ber are seen. This is the famous solar neutrino problem. The logically in the SM), and if they are not identically zero, then reproducibility in independent experiments of the measured why are they so small? 3

TABLE I: Solar neutrino fluxes from various experiments compared with a Standard Solar Model (SSM) calculated in Ref. 9. All fluxes are in 36 solar neutrino units (SNU), except the result from Super Kamiokande. (1 SNU is the neutrino flux that would produce 10− interactions per target atom per second.) Experimental uncertainties include systematic and statistical contributions.

Experiment Observed ﬂux SSM Observed/SSM +1.2 Homestake 2.54 0.14 0.14 SNU 9.3 1.4 0.273 0.021 +12± +5 ± −+8 ± SAGE 72 10 7 SNU 137 7 0.526 0.089 − − +3.9 +−8 ± GALLEX 69.7 6.7 4.5 SNU 137 7 0.509 0.089 ±+0.14 − 6 2 1 −+0.93 ± Super Kamiokande 2.51 0.13 (10 cm− s− ) 6.62 1.12 0.379 0.034 − − ±

where d and s are mass-eigenstate quark ﬁelds and the 2 2 Generation × I II III γ G unitary matrix is parameterized by θc, which is termed the mixing angle or the Cabibbo angle. Data require the mixing ν ν ν e µ τ W Z angle to be rather small: θc 13◦. In the more general case of three generations of quarks,∼ weak eigenstates are described

Leptons e by a 3 3 mixing matrix called the Cabibbo–Kobayashi– µ τ Gauge bosons × (forces) Maskawa or CKM matrix, which has three real mixing angles and one phase. There is little fundamental understanding of u c t this quark ﬂavor mixing but data clearly require it. H

Quarks d s b Higgs boson (mass) 3. Flavor Mixing in the Leptonic Sector?

Fermions (matter) Bosons Observation of flavor mixing among quarks raises the ques- tion of whether similar mixing could occur among the neu- Quark and neutrino mass scales trino flavors. This is irrelevant in the Standard Model because to be observable the neutrino flavors undergoing mixing must Gen I II III have different masses, which is impossible in the SM where all neutrinos necessarily are identically massless. The possibility Leptons e-, ν µ-, ν τ-, ν e µ τ of flavor oscillations among neutrinos is intriguing because Quarks d, u s, c b, t it could have large implications both for astrophysics and for Quark particle physics. First, the Sun produces electron neutrinos νe mass ~0.006 ~1 ~100 and the initial neutrino detection experiments that found the (GeV) solar neutrino problem were sensitive essentially only to νe. ν mass <2.2x10-9 <1.9x10-4 <1.8x10-2 If solar neutrinos can undergo flavor oscillations, then some (GeV) νe produced by the Sun could reach the Earth as muon neutrinos νµ or tao neutrinos ντ that would not be detected, thus FIG. 3: Particles of the Standard Model and characteristic mass possibly accounting for the solar neutrino deficit exhibited in scales in the quark and neutrino sectors for each generation.11 Table I. Second, observation of neutrino oscillations would imply that at least one neutrino flavor has a finite mass, indicating that there must be physics beyond the Standard Model. 2. Flavor Mixing in the Quark Sector Thus motivated, let us develop a theory of flavor oscillations for neutrinos. The Standard Model is made richer and more complex by the experimental observation of flavor mixing in the quark sector. For quarks the mass eigenstates and the weak eigenstates III. NEUTRINO VACUUM OSCILLATIONS are not equivalent, which means that the quark states that enter the weak interactions (flavor eigenstates) are generally linear There are three known neutrino (and three antineutrino) combinations of the mass eigenstates (eigenstates of propa- flavors, but phenomenology suggests that solar neutrinos are gating quarks). For example, restricting to the first two gener- well understoodin terms of a simple mixingmodel for two fla- ations, it is found that the d and s quarks enter the weak inter- vors that will be employed here. We address first the simplest actions in the “rotated” linear combinations dc and sc defined case of flavor oscillations for neutrinos propagating in vac- by the matrix equation uum. Then the more difficult (and interesting) issue of how the coupling of neutrinos to matter alters flavor oscillations d cosθ sinθ d c = c c (1) will be taken up. In much of this development the convention sc ! sinθc cosθc! s! that is standard in particle physics of choosing natural units, − 4 where both h¯ and the speed of light c are set to one, will be with the sum of the two probabilities necessarily equal to employed. The Appendix illustrates the relationship between unity. natural units and standard units.

B. The Vacuum Oscillation Length A. Mixing for Two Neutrino Flavors Equations (7) define the essence of vacuum neutrino os- To be definite the two flavors will be assumed to correspond cillations in a 2-flavor model. However, it is normal to put to the electron neutrino with wavefunction νe νe and the these results into a more practical form through the follow- ≡ | i muon neutrino with wavefunction νµ νµ , but the formal- ing considerations. Neutrinos have at most a tiny mass mi, so ≡ | i 2 ism could be applied to the mixing any two flavors. By anal- Ei >> mic and the relativistic energy expression may be ap- ogy with Eq. (1) for quarks, the flavor eigenstates νe and νµ proximated using a binomial expansion, are assumed to be related to the mass eigenstates ν1 ν1 ≡ | i 2 and ν2 ν2 through the matrix transformation 2 2 1/2 mi ≡ | i Ei = (p + m ) p + . (8) i ≃ 2p ν ν cosθ sinθ ν e = U(θ) 1 = 1 , (2) νµ ! ν2! sinθ cosθ! ν2! Thus, assuming that E1 E2 E, − ∼ ≡ where θ is a phenomenological vacuum mixing angle chosen 2 ∆m 2 2 2 to lie in the range 0 45 , and the matrix U is parameter- E2 E1 ∆m m m . (9) ◦ − ≃ 2E ≡ 2 − 1 ized by the single real− angle θ. The matrix U is unitary (and orthogonal), so UU † = U †U = 1, where U † is the hermitian The probability for flavor survival and flavor conversion as a conjugate of U (in a matrix representation, interchange rows function of distance traveled r ct may then be expressed as and columns, and complex conjugate all elements). This may ∼ 2 2 πr be used to invert Eq. (2) and express the mass eigenstates as a P(νe νe, r)= 1 sin 2θ sin , (10a) linear combination of the flavor eigenstates: → − L 2 2 πr P(νe νµ , r)= sin 2θ sin , (10b) ν1 νe cosθ sinθ νe → = U(θ)† = − . (3) L ν ν sinθ cosθ ν 2! µ ! ! µ ! where θ is the (vacuum) mixing angle and the (vacuum) os- Assuming that the mass is non-zero for at least one neutrino cillation length L, defined by flavor, the different mass eigenstates will have slightly differ- 4πE 4πEhc¯ ent energies as neutrinos propagate. Thus, the probability of L = , (11) detecting a specific neutrino flavor will oscillate with time, or ≡ ∆m2 ∆m2c4 with distance traveled. From relativistic quantum field theory is the distance required for one one complete flavor oscilla- applied to left-handedneutrinos(see Section IVB1), the mass tion. [In Eq. (11) L is expressed first in h¯ c 1 units and eigenstates evolve with time t according to = = then in normal “engineering units” with factors of h¯ and c iEit νi(t) = e− νi(0) , (4) restored; see further discussion in the Appendix.] Neutrino | i | i where the index i labels mass eigenstates of energy E , so the oscillations for a 2-flavor model using these formulas are il- i lustrated in Fig. 4. time evolution of the νe state in Eq. (2) will be given by

iE1t iE2t ν(t) = cosθe− ν1(0) + sinθe− ν2(0) , (5) | i | i | i and from the transformation (3) this may be expressed as the C. Time-Averaged or Classical Probabilities mixed-flavor state In observations the oscillation length may be smaller than 2 iE1t 2 iE2t ν(t) = (cos θe− + sin θe− ) νe | i | i the uncertainties in position for emission and detection of neu- iE1t iE2t + sinθ cosθ( e− + e− ) νµ . (6) trinos. This will usually be the case for solar neutrinos, where − | i oscillation lengths are typically several hundred kilometers Taking the overlap of this expression with the flavor eigen- but there are variation of thousands of kilometers in the dis- states ν and ν , the probabilities that a neutrino initially e µ tance traversed between production in the Sun and detection in an electron| i neutrino| i flavor state will remain an electron neu- on Earth. If the oscillation length is less than the averaging trino, or instead be converted to a muon neutrino after a time introduced by the preceding considerations, the detectors will t, are given respectively by see a distance (or time) average of Eq. (10). Denoting the 2 P(νe νe, t)= νe ν(t) averaged probability by a bar gives → |h | i| 1 2 = 1 2 sin (2θ)[1 cos(E2 E1)t], (7a) 1 2 1 2 − − − P¯(νe νe)= 1 sin 2θ P¯(νe νµ )= sin 2θ, (12) 2 → 2 → 2 P(νe νµ , t)= νµ ν(t) − → |h | i| 1 2 for the probability (10) of detecting the two flavors distance- = sin (2θ)[1 cos(E2 E1)t], (7b) 2 − − averaged over the oscillating factor. This may also be viewed 5

1 e− ν P + e νe W 0.8 Charged current

− 0.6 νe e L Sin22θ 0.4 e− ν e− ν Probability e µ Z 0 Z 0 0.2 Neutral P current νµ e− e− 0 νee νµ 0 0.5 1 1.5 2 r/L

(a) 2 diagrams contribute to νe (b) 1 diagram contributes to νµ FIG. 4: Neutrino vacuum oscillations in a 2-flavor model as a func- scattering from electrons scattering from electrons tion of distance traveled r ct, with r in units of the vacuum oscil- ∼ lation length L. The probability as a function of r to be an electron FIG. 5: Feynman diagrams responsible for neutrino–electron scatter- neutrino is denoted by Pνe and that to be a muon neutrino by Pνµ . ing in a 2-flavor model with νe and νµ flavors. Such diagrams are pic- The period of the oscillation is L and its amplitude is sin2 2θ, where torial representations of quantum-mechanical matrix elements. For θ is the vacuum mixing angle. In this calculation θ = 33.5 , the example, the upper left diagram stands for the matrix element ( ◦ ∼ neutrino energy is E = 5 MeV, the difference in squared masses for probability amplitude) for a process where an electron neutrino ex- 2 4 5 2 + the two flavors is ∆m c = 7.5 10− eV , and the corresponding changes a virtual W intermediate vector boson with an electron, oscillation length L is 165.3 km.× with the neutrino converted to an electron and the electron converted to an electron neutrino. This is a charged-current process, because the W + virtual exchange particle is electrically charged. On the as the classical probability, since the same formula results if other hand, the lower left diagram describes a neutral-current process 0 the quantum interference resulting from squaring the sum of where the exchanged virtual Z carries no electrical charge. Elec- probability amplitudes is removed from the probability. For tron neutrinos can interact with electrons through both the charged- current and neutral-current diagrams shown in (a), but muon and tau two flavors the instantaneous probability to remain a ν can e neutrinos can interact with electrons only through the neutral-current approach zero if the mixing angle is large (see Fig. 4) but the 1 diagram in (b). average survival probability has a lower limit of 2 for two 1 flavors. The lower limit is n− for n flavors, but that limit is realized only for a flavor mixture tuned to maximal mixing. elastic forward scattering, mediated by both charged and neutral weak currents. It is instructive to view these interactions in terms of Feynman diagrams, which are pictorial representa- IV. NEUTRINO OSCILLATIONS IN MATTER tions of quantum-mechanical matrix elements. The Feynman diagrams relevant for our discussion are displayed in Fig. 5. The preceding formalism describes propagation of neutri- For muon or tau neutrinos the charged-currentdiagram is for- nos undergoing flavor oscillations in a 2-flavor model in the bidden by lepton family number conservation, so only the approximate vacuum from the surface of the Sun to the Earth, neutral current can contribute, as illustrated in Fig. 5(b). but those neutrinos must also propagate through matter from The neutral current interaction contributes to both electron the central regions of the Sun to the surface. Electron neu- and muon neutrino scattering, so let’s ignore it and concen- trinos couple more strongly to normal matter than do other trate on the charged-current diagram in Fig. 5(a) that con- flavor neutrinos because electron neutrinos and the particles tributes only to νe elastic scattering in a 2-flavor model. From making up normal matter all reside in the first generation of standard methods of quantum field theory4–6 the charged- the Standard Model, but the muon and tau neutrinos are in current diagram in Fig. 5 contributes a medium-dependent in- the second and third generations, respectively. It will now be teraction having the form of an effective potential energy V shown that this flavor-dependent interaction with the medium seen only by the electron neutrinos or antineutrinos, with a alters the effective mass of a propagating neutrino differently magnitude for electron neutrinos than for other flavors, which changes V = √2GFne, (13) the effective mass-square difference in Eq. (9) and influences ± the oscillation non-trivially. where the positive sign is for electron neutrinos (our primary concern here), the negative sign is for electron antineutrinos, ne is the local electron number density, and GF is the weak A. Propagation of Neutrinos in Matter (Fermi) coupling constant characterizing the strength of the weak interactions. The energy is given by E2 = p2 + m2. For Let us consider the effect that interaction with matter the electron neutrino subject to the additional potential V can have on neutrino flavor oscillations.4,7,12–14 For the low- energy neutrinos found in the Sun inelastic scattering is neg- V 2 p2 + m2 = (E V)2 = E2 1 E2 2EV, (14) ligible and electron neutrinos in the Sun interact only through − − E ≃ − 6

m2 but if the spin structure is eliminated from consideration the propagation of a (left-handed component of the) free neutrino may instead be described by the simpler free-particle Klein– m2 νe Gordon equation

∂ 2 ∂ 2 ∂ 2 ∂ 2 ( + m2) ν = 0 + + + , (16) | i ≡−∂t2 ∂x2 ∂y2 ∂z2 m2 (vac) νµ m2 where is termed the d’Alembertian operator and for n neu- νµ trino flavors ν is an n-componentcolumn vector in the mass- eigenstate basis| i and m2 is an n n matrix. × 2 Because of oscillations, the solutions to Eq. (16) of interest mν (vac) e correspond to the propagation of a linear combination of mass ne 0 eigenstates. For ultrarelativistic neutrinos one makes only small errors by assuming neutrinos of tiny mass and slightly FIG. 6: The effective mass-squared of electron neutrinos and muon different energies to propagate with the same 3-momentum p. neutrinos as a function of electron number density ne, neglecting fla- In that approximation a solution of Eq. (16) for definite mo- vor mixing. Because the νµ does not couple to the charged weak mentum is given by 2 2 current its m does not depend on ne but the effective m of νe in- 2 creases linearly with the electron density. Thus the m spectrum in iEit ip x 2 2 νi = e− e− · Ei = p + m . (17) vacuum can be inverted in matter at high electron density. | i · i q For ultrarelativistic particles p >> mi so | | where the last step is justified by assuming that V << E. Thus 2 1/2 2 2 2 mi mi mi 2 2 2 Ei = p 1 + p + p + , (18) E p + (m + 2EV). (15) p2 ≃ 2 p ≃ 2E ∼ | | Since V is positive, an electron neutrino behaves effectively where E p p . Thus, from Eq. (17) as if it is slightly heavier when propagating through matter ∼ ≡ | | than in vacuum, with the amount of increase governed by the i p t ip x ip x i(m2/2E)t νi(t) e− | | e− · e− · e− i . electron density of the matter, but a muon or tau neutrino is | i ≃ · · unaffected because they do not see the effective potential V . But the initial exponential factors involving the 3-momenta Fig. 6 illustrates. From this figure, an electron neutrino that provide only a common overall phase that does not affect ob- is less massive than a muon neutrino in vacuum will become servables, so they may be dropped to give effectively more massive than its oscillation partner in matter if the electron density is sufficiently high. i(m2/2E)t νi(t) e− i . (19) | i≃

B. The Mass Matrix Differentiating Eq. (19) with respect to time gives an equation of motion for a single mass eigenstate labeled by i,

To address in more depth neutrino oscillations in matter, d m2 let’s introduce a more formal derivation of the neutrino os- i ν (t) = i ν (t) , (20) dt i 2E i cillation problem that will prove to be useful for subsequent | i | i 4,7,12,14–16 discussion. which may be generalized for a 2-ﬂavor model to the matrix equation

1. Propagation of Left-Handed Neutrinos d ν m2/2E 0 ν i 1 = 1 1 , (21) dt ν 0 m2/2E ν First note that the full spin structure does not inﬂuence the 2! 2 ! 2! propagation of neutrinos in the absence of magnetic ﬁelds be- where the quantity cause only the left-handed component of the neutrino couples to the weak interactions. The Schrödinger equation of ordi- m2/2E 0 nary quantum mechanics is not relativistically invariant and M 1 2 so is not appropriate for relativistic particles. For relativistic ≡ 0 m2/2E! fermions the wave equation must be generalized to the Dirac equation, while for spinless particles the corresponding rel- is termed the mass matrix. In these equations elapsed time t ativistic wave equation is the Klein–Gordon equation. Rela- or distance traveled r may be used interchangeably as the in- tivistic fermions are generally described by a Dirac equation dependent variable, since r ct for ultrarelativistic neutrinos. ∼ 7

2. Evolution in the Flavor Basis 3. Propagation in Matter

Neutrinos propagate in mass eigenstates but they are pro- The flavor evolution equation (23) is just a reformulation of duced and detected in flavor eigenstates, so it is useful to ex- our previous treatment of neutrinos propagating in vacuum, so press the preceding equation in the flavor basis. The transfor- it contains nothing new. However, let us now add a charged- mations matrices are given in Eqs. (2) and (3), which permits current interaction with matter. By previous arguments the Eq. (21) to be written in the form charged current couples only elastically and only to electron neutrinos, so we add to the Klein–Gordon equation (16) an interaction potential given by Eq. (13), which modifies the equa- d ν1 d † νe i = i U tion of motion (23) to dt ν2! dt νµ ! m2/2E 0 ν d ν 0 0 V(t) 0 ν = 1 U † e . i e = U U † + e , 2 dt 2 0 m2/2E! νµ ! νµ ! " 0 ∆m /2E! 0 0!# νµ !

Multiplying from the left by U and using the unitarity condi- where the effective potential generated by the charged-current tion UU † = 1 gives coupling of the electron neutrino to the matter,

d ν m2/2E 0 ν V(t)= √2GF ne(t), (24) i e = U 1 U † e , (22) dt ν 0 m2/2E ν µ ! 2 ! µ ! depends on time (or equivalently on position), because it is a function of the electron density. Inserting the explicit forms where the ﬂavor and mass eigenstates are related by for the transformation matrices U and U † gives

νe ν1 ν1 † νe U U d νe = = , i = νµ ! ν2! ν2! νµ ! dt νµ ! and the transformation matrices between mass and flavor cosθ sinθ 0 0 cosθ sinθ bases may be written explicitly as 2 − " sinθ cosθ! 0 ∆m /2E! sinθ cosθ ! − cosθ sinθ cosθ sinθ V 0 νe U U † + = = − . 0 0 ν sinθ cosθ! sinθ cosθ ! !# µ ! − 2 2 ∆m 2 ∆m As is clear by substitution, Eq. (22) has a solution sin θ sinθ cosθ 2E 2E V 0 νe =  2 2  +  ∆m ∆m 2 0 0! νµ ! i m2 2E t sinθ cosθ cos θ ν (t) e ( 1/ ) 0 ν (0)    e = U − U † e . 2E 2E i(m2/2E)t    νµ (t) 0 e− 2 νµ (0)  2   ! ! ! ∆m 0 sinθ cosθ V 0 νe = 2 2 + " 2E sinθ cosθ cos θ sin θ! 0 0!# νµ ! Since no individual neutrino mass has been measured thus far, − it is convenient to rewrite Eq. (22) in terms of ∆m2, which ∆m2 V sin2θ has been measured. Adding a multiple of the unit matrix to 4E νe νe =   M , (25) the matrix in Eq. (22) will not modify quantum observables ∆m2 ∆m2 ν ≡ ν sin2θ cos2θ µ ! µ ! (a trick that will be employed several times in what follows),  4E 2E  so let’s subtract m2/2E times the unit 2 2 matrix from the   1 ×   matrix in Eq. (22) and use where in the second step (∆m2/4E)sin2 θ times the 2 2 unit matrix has been subtracted, in the final line the trigonometr× ic 2 2 0 0 m1/2E 0 m1/2E 0 identities 2 = 2 2 0 ∆m /2E! 0 m2/2E! − 0 m1/2E! sinθ cosθ = 1 sin2θ cos2 θ sin2 θ = cos2θ 2 − to replace Eq. (22) by the equivalent form have been used, and M is termed the mass matrix (in the flavor basis). d ν 0 0 ν i e U U † e (23) Eq. (25) is the required result but it is conventional to write = 2 , dt νµ ! 0 ∆m /2E! νµ ! the mass matrix M appearing in it in a more symmetric form. First define where the mass-squared difference ∆m2 m2 m2 was intro- ≡ 2 − 1 duced in Eq. (9). A 2EV = 2√2EGFne, (26) ≡ 8

1. Mass Eigenvalues for Constant Density e n

Density gradient At constant density the problem resembles vacuum oscillations, except with a different potential V = 0, and the time- m 6 m evolved mass states in matter, ν1 and ν2 , may be found ν by determining the eigenvalues| of thei mass| matrixi (27). The eigenvalues of an n n matrix A correspond to the values of × Log electron density λ that solve the characteristic equation det(A λI)= 0, with I the n n unit matrix and det indicating the determinant.− For Surface Center × R/R a general 2 2 matrix 0 × α β FIG. 7: Solar density gradient. Neutrinos are produced near the cen- A = , ter at high density and propagate out through regions of decreasing γ δ ! density. Within a given concentric layer, the density may be assumed to be nearly constant. the characteristic equation is quadratic with the two solutions,

λ = 1 tr(A) 1 tr(A)2 det(A), (which has units of mass squared) and then subtract A/4E + ± 2 ± 4 − (∆m2/4E)cos2θ multiplied by the unit matrix to give the q mass matrix in traceless form where trA is the trace and detA is the determinant of A. Hence the eigenvalues λ of the mass matrix (27) are given by A ± m2 cos2θ sin2θ 2 2 2 ∆ ∆m2 − m1 + m2 ∆m M =  A  λ = + χ 4E sin2θ cos2θ ± 2 2 2 (27) (29)  − ∆m  ∆m2   (cos2θ χ)2 + sin2 2θ. π χ cos2θ sin2θ ± 2 − = − , q L sin2θ cos2θ χ! − The second term gives the splitting between the two mass where the dimensionless charged-current coupling strength χ eigenstates and is minimal at the density where χ = cos2θ. has been introduced through By analogywith the vacuumcase, the mass eigenstates in matter, νm ν m and νm ν m at fixed time t, are assumed to 1 ≡ | 1 i 2 ≡ | 2 i L 2EV √2π 4πE be related to the flavor eigenstates by χ = ℓm L , (28) ≡ ℓ ∆m2 ≡ G n ≡ ∆m2 m F e ν νm e = U (t) 1 . (30) L m m with the vacuum oscillation length defined in Eq. (11) and νµ ! ν2 ! ℓm an additional contribution to the oscillation length in matter that is termed the refraction length because it is a character- where Um(t) is a unitary matrix depending on the density that istic distance over which there is significant refraction of the we must now determine. neutrino by the matter.5

2. The Matter Mixing Angle θm C. Solutions in Matter

The matrix Um(t) can be parameterized as for the vacuum For a fixed density the mass eigenstates in matter—which mixing matrix U, but now in terms of a time-dependentmatter generally will differ from the mass eigenstates in vacuum be- mixing angle θm(t) with cause of the interaction V—may be found by diagonalizing (finding the eigenvalues) of the mass matrix M at that den- cosθm sinθm † cosθm sinθm sity. However, since the interaction V depends on the density, Um = Um = − . sinθ cosθ sinθ cosθ in a medium with varying density such as the Sun the mass m m! m m ! − (31) eigenstates in matter at one time (or position) will generally The relationship of the matter mixing angle θ and vacuum not be eigenstates at another time. Let’s imagine dividing the m mixing angle θ at time t can be established by requiring that Sun up into concentric radial layers, as illustrated in Fig. 7, a similarity transform by the unitary matrix U t diagonalize and assume the density within each layer to be approximately m( ) the mass matrix, with the diagonal elements being the time- constant. Our approach will be to first understand how to cal- dependent eigenvalues in matter E t and E t , culate the mass eigenstates within a single layer assumed to 1( ) 2( ) have a constant density, and then to determine the evolution of neutrino states as they propagate through successive layers † E1(t) 0 Um(t)MUm(t)= diag[E1(t),E2(t)] = . (32) of decreasing density in the Sun. 0 E2(t)! 9

† Inserting the explicit values of the matrices Um, Um, and M π 2 2000 from Eqs. (27) and (31) in Eq. (32) gives a matrix equation, ο 1 (b) L (km) for which comparison of matrix entries on the two sides of the m equation requires the matter and vacuum mixing angles to be 1500 related by ο 5 45ο

sin2θ tan2θ π tan2θ = = , (33) ο 1000 m cos2θ χ 1 χ/cos2θ 4 35 ± ± ο 25 15ο where the plus sign is for m > m and the negative sign for ο 1 2 15 500 25ο m1 < m2. From Eq. (33), in vacuum θm = θ but in matter ο 5ο 35 the mixing angle will be modiﬁed from its vacuum value by ο ο (a) θm (radians) 45 a density-dependent amount governed by the charged-current 1 0 0 coupling strength χ. 0 1 2 3 0 1 2 3 χ χ

FIG. 8: (a) Mixing angle in matter θm(χ) and (b) oscillation length in matter Lm(χ), as a function of the dimensionless matter cou- 3. The Matter Oscillation Length Lm pling parameter χ. All calculations assumed E = 10 MeV and 2 5 2 ∆m = 7.6 10− eV , and curves are marked with the assumed vacuum mixing× angle θ. From Eq. (11) the vacuum oscillation length is proportional 2 to 1/∆m . In matter the νe effective mass is altered by interaction with the medium and the vacuum mass-squared differ- 4. Flavor Conversion Probabilities in Constant-Density Matter ence is rescaled, ∆m2 f (χ)∆m2, where from the splitting of the two eigenvalues→ in Eq. (29) In constant-density matter the electron neutrino state after a time t becomes 2 2 f (χ)= (cos2θ χ) + sin 2θ 2 iE1t 2 iE2t ν(t) = (cos θme− + sin θme− ) νe − | i | i (36) q iE1t iE2t = 1 2χ cos2θ + χ2. (34) +sinθm cosθm( e− + e− ) νµ , − − | i q which is analogous to the vacuum equation Eq. (6) with θ Hence the oscillation length in matter Lm is given by replaced by θm defined through Eq. (33). The corresponding flavor conservation and retention probabilities are given 4πE L by Eq. (10) with the replacements θ θm and L Lm, L = = → → m f (χ)∆m2 f (χ) 2 2 πr P(νe νe, r)= 1 sin 2θm sin (37) L → − L = , (35) m (cos2θ χ)2 + sin2 2θ with P(νe νµ , r) = 1 P(νe νe, r). The corresponding − → → q classical averages are − which reduces to the vacuum oscillation length L if χ 1 2 1 2 → P¯(νe νe)= 1 sin 2θm P¯(νe νµ )= sin 2θm, (38) 0. The variations of θm, Lm, and f with the dimensionless → − 2 → 2 charged-current coupling strength χ are illustrated in Fig. 8 which are appropriate when the uncertainty in distance be- for several values of the vacuum mixing angle θ. tween source and detection exceeds the oscillation length From Fig. 8(a), the matter mixing angle θm reduces to the (which is generally the case for solar neutrinos). π vacuum mixing angle θ as χ 0, but θm 2 for any value of the vacuum mixing angle as χ→ ∞, with→ the rate of approach π → to 2 being fastest for smaller vacuum angle. From Fig. 8(b) D. The MSW Resonance Condition the matter oscillation length is equal to the vacuum oscillation length at zero coupling, but increases to a maximum at the From Eq. (37), optimal flavor mixing occurs whenever π coupling strength where θm = [compare Fig. 8(a)], and then 2 π 4 sin 2θm 1, implying that θm = 4 . The most significant decreases again. The coupling strength at which Lm is maxi- property→ of Eq. (33) relative to| the| vacuum solution is that if 2 mal coincides with highest rate of change in θm, and with the ∆m and L are positive [requiring that m1 < m2, which se- minimum of the scaling function f (χ) displayed in Fig. 8(c). lects the negative sign in Eq. (33)]. Then tan2θm ∞ and π →± The rapid change of θm and the strong peaking of Lm near the θm whenever π → 4 density where θm = 4 are suggestive of resonant behavior, as will be elaborated shortly. χ = cos2θ, (39) 10

8 500 5 (a) θ = 33.5 ο 1.0 (a) θ = 33.5 ο

6 2 sin 2θm 400 4 sin L χ, Resonance m L cos(2 4 2 0.5 2 θ

m 300 3 m θ θ ), tan(2

2 (km) m L , tan 2 200 |tan(2θm)|/5 2 0 0 L θ

tan 2θ m m χ ) -2 100 1 cos 2θ -4 0 0 -4 -2 0 2 4 0 0.2 0.4 0.6 0.8 1.0 L/l n n0 m e/ e 8 2500 2.5 ο 1.0 (b) θ = 5 ο (b) θ = 5 |tan(2θm)| 6 2 sin 2θm sin 2000 2.0 Resonance χ, 2 4 cos(2 0.5 2

θ L m m m 1500 θ χ 1.5 2 θ ), tan(2 (km) m L tan 2

, cos 2θ

0 0 L 1000 1.0

tan 2θ θ m m ) -2 500 0.5 L

-4 -4 -2 0 2 4 0 0 L/l 0 0.2 0.4 0.6 0.8 1.0 m n n0 e/ e FIG. 9: The MSW resonance condition for two values of the vac- FIG. 10: Resonance parameters versus the electron number density uum mixing angle θ. When χ = L/ℓm cos2θ the denominator → π n in units of the central solar value n0 6.3 1025 cm 3 for θ = of Eq. (33) goes to zero, tan2θm goes to ∞ so that θm , and e e − ± | |→ 4 33.5 and 5 , with ∆m2 = 7.6 10 5 eV∼2 and× E = 10 MeV. The the flavor conversion probability sin2 2θ attains its maximum value. ◦ ◦ − m coupling strength χ = L/ℓ is× linear in the density. Intersection of Thus, at the resonance Eq. (37) indicates that it is possible to obtain m the dashed lines specifies the density giving the resonance condition. very large flavor conversion for any non-vanishing vacuum oscillation angle θ. This is the Mikheyev–Smirnov–Wolfenstein or MSW 12,13 where χ = L/ℓ , as illustrated in Fig. 9. From Eq. (28), this resonance. It implies that—no matter how small the m vacuum mixing angle θ—as long as it is not zero there occurs when the electron density satisfies R is some critical value ne of the electron density defined 2 by Eq. (40) where the resonance condition is satisfied and cos2θ∆m R ne = ne . (40) maximal flavor mixing ensues. The important resonance 2√2GFE ≡ parameters are plotted in Fig. 10 as a function of electron From Eq. (37), this correspondsto a resonance condition lead- density for two values of the vacuum mixing angle θ. ing to maximal flavor mixing between electron neutrinos and The effect of the MSW resonance on variation of the matter muon neutrinos, with a ν survival probability mixing angle θm and the oscillation length in matter Lm are e illustrated for a small and large angle solution in Fig. 11. The values of θm and Lm will vary with the solar depth since they 2 πr P(νe νe, r)= 1 sin (at resonance), (41) depend on the number density ne through χ. From Eq. (33), → − Lm θm θ as the electron density tends to zero, while in the → π opposite limit of very large electron density θm . From and, from Eqs. (35) and (39), an oscillation length at reso- → 2 nance LR given by Fig. 11, the manner in which the two limits are approachedde- m pends on whether the vacuum mixing angle is large or small. Case (a) corresponds to parameters valid for solar neutrinos. R L Lm = Lm(χ=cos2θ)= . (42) At the solar center (where χ 2.13) the value of the matter sin2θ ∼ 11

400 m2 MSW ο Surface (a) θ = 33.5 resonance νe L m of Sun π 2 θ = ∆m (ne) π m 2 2 300 θm m2 2 νµ νµ L m (rad)

(km) 2 m 200 ∆m Center θ Resonance vac of Sun νe nR π e 4 m2 n θm= θ 1 e Solar center 100 Solar FIG. 12: Schematic illustration of resonance ﬂavor conversion as interior an adiabatic quantum level crossing. Speciﬁc examples for different parameter choices are shown in Fig. 13. Solar surface 0 0 0 0 0.5 1.0 1.5 2.0 2.5 3.0

χ antineutrino ν¯e. Antineutrino oscillations could be important 2000 in environments such as core-collapse supernovae where all (b) L = 5ο ﬂavors of neutrinos and antineutrinos are found in copious m θ amounts, but the Sun produces mostly neutrinos so consid- π θ = π m 2 eration of oscillations for antineutrinos will be omitted from 2 1500 the present discussion. θm L m (rad)

(km) F. Resonant Flavor Conversion m

θ 1000 Resonance π If m m and the electron density in the central region 4 1 < 2 R of the Sun where neutrinos are produced satisfies ne > ne , a 500 solar neutrino will inevitably encounter the MSW resonance on its way out of the Sun. If the change in density is sufficiently slow that the additional phase mismatch between the θ = θ m νe and νµ components produced by the charged-current elas- 0 tic scattering from electrons changes very slowly with density 0 0.5 1.0 1.5 2.0 2.5 3.0 (the adiabatic condition discussed further below), the νe flux χ produced in the core can be almost entirely convertedto νµ by the MSW resonance near the radius where the condition (40) FIG. 11: Matter mixing angle θm as a function of the dimen- is satisfied. sionless coupling strength χ L/ℓ for vacuum mixing angles of m MSW flavor conversion can be viewed as an adiabatic (a) θ = 33.5 and (b) θ = 5 ≡. Also shown is the matter oscillation ◦ ◦ quantum level crossing,14,17 as illustrated schematically in length Lm, which has a maximum at the MSW resonance (see Sec- tion IV D), marked by the dashed vertical line. The oscillation length Fig. 12, and more realistically in Fig. 13. Solutions of the was computed from Eqs. (35) and (11) assuming E = 10 MeV and MSW eigenvalue problem illustrating this level crossing for 2 5 2 ∆m = 7.6 10− eV . Case (a) is realistic for solar neutrinos and various choices of the vacuum mixing angle are displayed in at the center× of the Sun χ 2.13. Hence the shaded region on the Fig. 13. At zero density on the left side of these plots the ∼ left side of (a) indicates the range of coupling strengths available to eigenvalues λ converge to the vacuum m2 values, but for electron neutrinos in the interior of the Sun. non-zero density± the masses are altered by the interaction of the electron neutrino with the medium and the mixing of the solutions by the neutrino oscillation. (Compare with the un- mixing angle is θm 76 , compared with a vacuum mixing ∼ ◦ mixed case in Fig. 6.) angle 33.5◦ at the solar surface. Conversely, for case (b) with A neutrino produced near the center of the Sun (high den- θ = 5◦ at the solar surface, the matter mixing angle at a density on right side) will be in a νe flavor eigenstate that co- sity corresponding to the solar center is θ 86 . ∼ ◦ incides with the higher-mass eigenstate, since V representing interaction with the medium increases the mass of the electron neutrino but not the muon neutrino. As the neutrino propa- E. Resonance in the Antineutrino Channel gates out of the Sun (right to left in this diagram) the density decreases so V and the effective mass of the neutrino de- If the condition m1 < m2 is not satisfied there is no MSW crease. Conversely, the lower-mass eigenstate (primarily νµ resonance for electron neutrinos in a 2-flavor model; there is flavor) remains constant in mass as ne decreases (no coupling instead a corresponding resonance condition for the electron to the charged current). Thus the two levels cross at the criti- 12

0.00020 26.0 (a) θ = 1ο MSW λ+ 0.00015 25.8 E min E = 2 MeV 0.00010 25.6 λ−

) n 0.00005 e ξ = 0.25 -3 25.4

0.00000 cm

e 6 MeV n

( 25.2 0.00020 ο 10 (b) θ = 5 MSW λ 10 MeV + 25.0 0.00015 Log 14 MeV 0.00010 24.8 18 MeV λ− 0.00005 24.6

) ξ = 6.2 2 0.00000 0.00 0.10 0.20 0.30 0.40 (eV

2 R/R 0.00020 ο (c) θ = 15 Mass λ+ 0.00015 FIG. 14: Radius where the MSW critical density for a 2-ﬂavor model is realized (dots at intersection of dashed lines with the curve for ne) 0.00010 for neutrinos of energies ranging from 2 to 18 MeV. A vacuum mix- 2 4 5 2 ing angle θ = 35◦ and ∆m c = 7.5 10− eV have been assumed. λ− MSW × 0.00005 The minimum energy of an electron neutrino Emin 1.6 MeV that ξ = 58.4 could be produced in the Sun and still encounter the MSW∼ resonance 0.00000 is indicated.

0.00020 ο (d) θ = 33.5 λ+ 0.00015 in the level-crossing region, as indicated by the arrows. It emerges from the Sun in a different flavor state than the one 0.00010 in which it was created in the core of the Sun because the λ− high-mass eigenstate is primarily νe in the dense medium but 0.00005 is primarily νµ in vacuum. Such adiabatic crossing of energy MSW ξ = 434.4 levels is common in a variety of quantum systems and is often 0.00000 0 40 80 120 160 termed avoided level crossing. ρ (g cm-3) The number density of electrons in the Sun computed in the Standard Solar Model is illustrated in Fig. 2, along with an ex- FIG. 13: Solutions λ of the MSW eigenvalue problem as a func- ponentialapproximationthat is rather good in the region of the tion of mass density± according to Eq. (29). Each case corresponds Sun where the MSW effect is most important. In Fig. 14, the 2 5 2 to the choices ∆m = 7.6 10− eV and E = 10 MeV, but to dif- approximate locations in the Sun where electron neutrinos of ferent values of the vacuum× mixing angle θ. The individual neu- various energies would encounter the MSW resonance condi- trino masses are presently unknown but for purposes of illustration tion [determined by solving Eq. (40) for each choice of en- 2 5 2 2 m1 = 5 10− eV has been assumed in vacuum, so that m2 = ergy] are illustrated. Only neutrinos having an energy larger 2 ×2 4 2 m + ∆m = 1.26 10− eV . The critical density leading to the than some minimum energy E can experience the MSW 1 × min MSW resonance (corresponding to minimum splitting between the resonance in the Sun because the neutrino must be produced r LR eigenvalues) and the value of the adiabaticity parameter ξ = δ R/ m at a density higher than the critical resonance density. The defined in Eq. (45) are indicated for each case. As will be discussed conditions used to obtain Fig. 14 give E 1.6 MeV. Thus, in Section V, realistic solar conditions imply the highly adiabatic min we may expect that the MSW effect is more∼ efficient at con- crossing exhibited in case (d). verting the flavor of higher-energy neutrinos. That flavor conversion is observed to be preferentially suppressed for lower- cal density where the effect of V exactly cancels the vacuum energy solar neutrinos will be an important piece of evidence 2 2 favoring the MSW mechanism over vacuum oscillations for mass-squared difference ∆mvac ∆m (ne = 0) between the eigenstates, and the neutrino remains≡ in the high-mass eigen- the observed flavor conversion of solar neutrinos. state and changes adiabatically into a νµ flavor state by the time it exits the Sun (left side), because in vacuum the high- G. Neutrino Propagation in Matter with Varying Density mass eigenstate approximately coincides with νµ . In summary, if the crossing is adiabatic the neutrino remains in the high-mass eigenstate in which it was created Consider now realistic neutrino propagation in the Sun, and follows the upper curved trajectory though the resonance where a neutrino produced in the core will encounter steadily 13 decreasing density as it travels toward the solar surface (see H. The Adiabatic Criterion Fig. 7). The neutrino flavor evolution will be governed by the analog of the coupled differential equations for vacuum The adiabatic condition for resonant flavor conversion can propagation, but with U Um(t) since the flavor–mass basis be expressed as a requirement that the spatial width of the res- transformation now depends→ on time. From Eq. (22) with this onance layer δrR (which is defined by the radial distance over replacement which the resonance condition is approximately fulfilled) be much greater than the oscillation wavelength in matter eval- 2 uated at the resonance, LR . This can be characterized by in- d ν1(t) 1 m1 0 ν1(t) m i Um(t) = Um(t) , troducing an adiabaticity parameter ξ through the definition dt ν (t) 2E 0 m2 ν (t) " 2 !# 2! 2 ! [see Eq. (42)] where both the wavefunctions and the transformation matrix δr nR L ξ R δr = e tan2θ LR = , (45) Um are indicated explicitly to depend on the time. Taking the R R m ≡ L (dne/dr)R sin2θ time derivative of the product in brackets on the left side and m † multiplying the equation from the left by Um gives where an index R labels quantities that are evaluated at the resonance, L is the vacuum oscillation length, and θ is the 2 vacuum oscillation angle.5,14 The adiabatic condition corre- d ν1 m1/2E 0 † d ν1 i = iUm Um sponds then to the requirement that ξ >> 1, which implies dt ν 0 m2 2E − dt ν 2! " 2/ ! # 2! physically that if many flavor oscillation lengths (in matter) 2 fit within the resonance layer the adiabatic approximation of m /2E iθ˙m ν1 = 1 − Eq. (44) is valid. iθ˙ m2/2E ν m 2 ! 2! Values of ξ computed from Eq. (45) are indicated in Fig. 13 2 ∆m /4E iθ˙m ν1 for several choices of vacuum mixing angle θ. In general very = − − , (43) iθ˙ ∆m2/4E ν sharp level crossings as in Fig. 13(a) are non-adiabatic, while m ! 2! avoided level crossings as in Fig. 13(d) are highly adiabatic. In the limit of no mixing (θ = 0) the levels cross with no in- ˙ where Eq. (31) was used, θm dθm/dt, and in the last step teraction, as illustrated earlier in Fig. 6. From Fig. 13 one sees 2 2 ≡ the constant (m1 +m2)/4E times the unit matrix has been sub- that the MSW resonance can occur under approximately adi- tracted from the matrix on the right side (which does not affect abatic conditions, even for relatively small values of the vac- observables). uum mixingangle [for example, case (b)]. As will be shown in Our earlier statement that mass eigenstates at some density Section V, solar conditions correspond to the highly-avoided R in the Sun generally will not be eigenstates at a different den- level crossing in Fig. 13(d), for which δrR >> Lm. Thus the sity may now be quantified. If the mass matrix in Eq. (43) MSW resonance is expected to be encountered adiabatically were diagonal the neutrino would remain in its original mass in the Sun, which optimizes the chance of resonant flavor con- eigenstate as it traveled through regions of varying density, so version. it is the off-diagonal terms proportional to θ˙m = dθm/dt that alter the mass eigenstates as the neutrino propagates. Gen- erally then, Eq. (43) must be solved numerically. However, if I. MSW Neutrino Flavor Conversion the off-diagonalterms are small relative to the diagonal terms, the mass matrix M may be approximated by dropping the off- Generally neutrino flavor conversion in the Sun must be ob- diagonal terms tained by integrating Eq. (43) numerically because of the solar density gradient, but it has just been argued that the adiabatic ∆m2/4E iθ˙ ∆m2 1 0 approximation (44) should be very well fulfilled for the Sun. M = m , (44) − ˙ −2 − Therefore, let’s now solve for flavor conversion of solar neu- iθm ∆m /4E! ≃ 4E 0 1! trinos by the MSW mechanism, assuming the validity of the adiabatic approximation. which affords the possibility of an analytical solution for neutrino flavor conversion in the Sun. This is called the adiabatic approximation, and corresponds physically to the assumption 1. Solar Flavor Conversion in Adiabatic Approximation that the matter mixing angle θm changes only slowly over a characteristic time for motion of the neutrino. A neutrino travels at near light speed so r ct and the adiabatic condition The adiabatic conversion of neutrino flavor in the Sun is il- also may be interpreted as a∼ limit on the spatial gradient of lustrated in Fig. 15. An electron neutrino is produced at Point θm. From Fig. 8(a) and Fig. 11, the most rapid change in θm 1 near the center of the Sun and propagates radially outward and thus the largest contribution of the off-diagonal terms oc- to Point 2. Detection is assumed to average over many oscilla- π curs near the MSW resonance (corresponding to θm = 4 ). Let tion lengths so that the interference terms are washed out and us now use these observations to quantify the conditions ap- our concern is with the classical (time-averaged) probability, propriate for the adiabatic approximation. as described in Section IIIC. The probability to be detected at 14

1 1 e n Resonance 2 λ layer

) + 2 (eV

2 Production 2 ν layer Mass MSW λ− resonance Log electron density

Surface Center Surface R/R Center ρ 0

FIG. 15: Adiabatic ﬂavor conversion by the MSW mechanism in the Sun. An electron neutrino is produced at Point 1 where the density lies above that of the MSW resonance and propagates radially outward to Point 2 where the density lies below that of the resonance. The width of the resonance layer is assumed to be much larger than the matter oscillation length in the resonance layer, justifying the adiabatic approximation of Eq. (44). Widths of resonance and production layers are not meant to be to scale in this ﬁgure.

4 Point 2 in the νe flavor eigenstate is then given by in Section IIIC for the classical average of vacuum oscilla- | i tions; indeed, Eq. (47) is equivalent to Eq. (12) in the limit 2 2 that θ θ. MSW flavor conversion in adiabatic approxi- ¯ cos θm(2) sin θm(2) m P(νe νe) = (1 0) mation is→ illustrated for four different values of the vacuum → sin2 θ (2) cos2 θ (2 m m ! mixing angle θ in Fig. 16. Figure 16(d) approximates the sit- 1 0 cos2 θ (1) sin2 θ (1) 1 uation expected for the Sun. m m , × 0 1 sin2 θ 1 cos2 θ 1 0 m( ) m( )! ! θ where θm(i) θm(ti) and the row vector (10) and correspond- 3. Resonant Conversion for Large or Small ≡ ing column vector denote a pure νe flavor state. Evaluating the matrix products and using standard trigonometric identities, As shown in Fig. 8(a), the matter mixing angle θm ap- π 1 proaches 2 near the center of the Sun and becomes equal to P¯(νe νe)= [1 + cos2θm(t1)cos2θm(t2)]. (46) → 2 θ at the surface of the Sun. Hence neutrinos are produced in a flavor eigenstate that is an almost pure mass eigenstate, but This result is valid (if the adiabatic condition is satisfied) for they evolve to a flavor mixture characterized by the vacuum Point 2 anywhere outside Point 1, but in the specific case that mixing angle θ by the time they exit the Sun. The most rapid Point 2 lies at the solar surface θm(t2) θ and the classical flavor conversion occurs around the MSW resonance where probability to detect the neutrino as an electron→ neutrino when the Pνe and Pνµ curves intersect. For smaller vacuum mixing it exits the Sun is angle θ almost complete flavor conversion occurs in the res- 1 0 onance, while for the large mixing angle case of Fig. 16(d) P¯(νe νe)= 1 + cos2θ cos2θ , (47) → 2 m 2 about 3 of 10-MeV νe produced in the core will undergo fla- 0 vor conversion before leaving the Sun. where θ is the vacuum mixing angle and θm θm(t1) is the matter mixing angle at the point of neutrino production.≡

4. Energy Dependence of Flavor Conversion 2. Dependence Only on the Mixing Angle Figure 14 suggests that flavor conversion has a significant Equation (47) has a simple physical interpretation. The energy dependence. For example, repeating the calculation mass matrix for a neutrino propagating down the solar den- of Fig. 16(d) for a range of neutrino energies E gives the elec- sity gradient is diagonal by virtue of the adiabatic assump- tron neutrino survival probabilities displayed in Table II, along R tion (44), so a neutrino produced in the λ+ eigenstate remains with the fractional solar radius R /R where the MSW reso- in that mass eigenstate until it reaches the solar surface, with nance occurs for that energy. From⊙ the spectrum in Fig. 1 the flavor conversion resulting only from the change of mix- and the experimental neutrino anomalies of Table I, one sees ing angle between the production point and the surface. Thus, an overall suppression of expected electron neutrino proba- in adiabatic approximation the classical probability P¯(νe νe) bilities to the 30%–50% range by the MSW effect, with the → depends only on the mixing angles at the point of production lower end of this range associated with higher-energy neutri- and point of detection, and is independentof the details of neu- nos. These results suggest a possible resolution of the solar trino propagation. This is reminiscent of the results obtained neutrino problem that will be elaborated further in Section V. 15

(a) θ = 5ο (b) θ = 15ο (c) θ = 25 ο (d) θ = 35ο 1.0 1.0 _ _ _ P P ν ν P µ 0.8 µ νµ _ 0.8 P νµ

0.6 0.6 _ P 0.4 0.4 _ P _ _ νe 0.2 P P _ 0.2 νe νe P νe 0.0 0.0 0.6 0.4 0.2 0.0 0.6 0.4 0.2 0.0 0.6 0.4 0.2 0.0 0.6 0.4 0.2 0.0 R/R R/R R/R R/R

FIG. 16: MSW flavor conversion versus fraction of solar radius for four values of the vacuum mixing angle θ in a 2-flavor model. All calculations are classical averages (Section III C) over local oscillations in adiabatic approximation using Eq. (47) with ∆m2 = 7.6 10 5 eV2 × − and E = 10 MeV. The exponential density approximation shown in Fig. 2 was used and neutrinos were assumed to be produced in a νe flavor ¯ ¯ state at the center (right side of diagram at R/R = 0). Solid curves show the classical electron-neutrino probability Pνe P(νe νe) and ⊙ ≡ → dashed curves show the corresponding classical muon-neutrino probability P¯ν P¯(νe νµ ). In these figures the MSW resonance occurs at µ → the radius corresponding to the intersection of the solid and dashed curves. ≡

TABLE II: Energy dependence of solar neutrino ﬂavor conversion for a vacuum mixing angle θ = 35◦ E (MeV) 14 10 6 2 1 0.70

Pνe (surface) 0.33 0.33 0.34 0.40 0.47 0.50 RR/R 0.28 0.25 0.20 0.10 0.03 0.0 ⊙

V. RESOLVING THE SOLAR NEUTRINO PROBLEM oscillations and thus of ﬁnite neutrino mass.

The two-flavor neutrino oscillation formalism described above may now be used in conjunction with a series of key B. SNO Observation of Neutral Current Interactions neutrino observations by Super Kamiokande, the Sudbury Neutrino Observatory (SNO), and KamLAND to resolve the The Super Kamiokande atmospheric neutrino results cited solar neutrino problem. As will now be summarized,this anal- above indicate the existence of neutrino oscillations and thus ysis indicates that electron neutrinos are being converted to of physics beyond the Standard Model. The neutrino oscilla- other flavors by neutrino oscillations, that the solar neutrino tions discovered by Super Kamiokande are not directly appli- “deficit” disappears if all flavors of neutrinos coming from cable to the solar neutrino problem because they do not ap- the Sun are detected, and that the favored oscillation scenario pear to involve electron neutrinos. However, a modified water is MSW resonance conversion in the Sun for a large vacuum Cherenkov detector operating in Canada has yielded informa- mixing angle solution. tion about neutrino oscillations that does have implications for the solar neutrino problem. A. Super Kamiokande Observation of Flavor Oscillation

Cosmic rays striking the atmosphere generate showers of 1. SNO and Heavy Water mesons that decay to muons, electrons, positrons, and neutrinos. The Super Kamiokande (Super K) detector in Japan The Sudbury Neutrino Observatory (SNO) differed from was used to observe neutrinos produced in these atmospheric Super Kamiokande in that it contained heavy water (water en- cosmic ray showers. These measurements found that the ra- riched in deuterium, 2H) at its core. In regularwater solar neu- tio of muon neutrinos and antineutrinos to electron neutrinos trinos can signal their presence only by elastic scattering from and antineutrinos was less than 64% of the value expected electrons, which typically requires about 5–7 MeV of energy from the Standard Model, and the∼ results were interpreted as to produce Cherenkov light for reliable detection. However, an indication that the muon neutrino was undergoing oscil- the deuterium (d) contained in the heavy water can undergo a lations with another flavor neutrino that was not the electron breakup reaction when struck by a neutrino through the weak neutrino.20 This was the first conclusive evidence of neutrino neutral current, where any flavor neutrino can initiate the re- 16 action nearby Japanese and Korean reactors. Antineutrinos are detected from the inverse β-decay in the scintillator: ν¯e + p ν + d ν + p + n (Neutralcurrent), (48) → → e+ + n. From the power levels in the reactors, the expected and the charged weak current reaction antineutrino flux at KamLAND could be modeled accurately. The experiment detected a shortfall of antineutrinos relative νe + d e− + p + p (Charged current), (49) → to the expected number and this could be accounted for as- which can be initiated only by electron neutrinos. These re- suming (anti)neutrino oscillations with a large-angle solution actions have much larger cross sections than elastic neutrino– having23,24 electron scattering, and the energy threshold can be lowered 2 +0.14 5 2 2 +0.10 to 2.2 MeV, the binding energy of the deuteron. Crucially, the ∆m = 7.58 0.13 10− eV tan θ = 0.56 0.07 (51) − × − neutral-current reaction (48) is flavor blind, which gave SNO the ability to see the total neutrino flux of all flavors coming (only statistical uncertainties are indicated), which corre- from the Sun. sponds to θ 36.8◦ for the vacuum mixing angle. The oscillation∼ properties of neutrinos and antineutrinos of the same generation are expected to be equivalent by CPT 2. Observation of Flavor Conversion for Solar Neutrinos symmetry, where C is charge conjugation, P is parity, and T is time reversal. Thus the large-angle KamLAND solution for electron antineutrinos may be interpreted as corroboration of The SNO results are summarized concisely in Table III and the large-angle solution found for solar neutrinos. Combining Fig. 17(a). The SNO observations confirmed results from the the solar neutrino and KamLAND results leads to a solution24 pioneering solar neutrino experiments summarized in Table I: a strong suppression of the electron neutrino flux is ob- 2 5 2 2 +0.06 21 ∆m = 7.59 0.21 10− eV tan θ = 0.47 0.05, (52) served relative to that expected in the Standard Solar Model. ± × − However, by analyzing the flavor-blind weak neutral current implying a vacuum mixing angle θ 34.4◦. The solar neu- events, the total flux of all neutrinos was found to be almost trino problem is resolved by neutrino∼ oscillations for which exactly that expected from the Standard Solar Model. Table the vacuum mixing angle is large (recall that θ is defined so 17 summarizes: although the ν flux is only 35% of that ex- e that its maximum value is 45◦). pected without oscillations, the flux summed over all flavors is 100% of that predicted by the Standard Solar Model for electron neutrino emission, within the experimental uncertainty. D. Large Mixing Angles and the MSW Mechanism The SNO case was strengthened by analysis of neutrino– electron elastic scattering data combined with data from the The large mixing angle solutions found by SNO and Kam- charged-current reaction (49). This also allows an estimate LAND indicate that the vacuum oscillations of solar neutri- of the total neutrino flux of all flavors: in elastic scattering nos are of secondary importance to the MSW matter oscil- from electrons, both charged and neutral currents contribute lations in the body of the Sun itself in reducing the electron for electron neutrinos but only neutral currents do so for other neutrino flux. The large-angle solutions imply vacuum oscil- flavors. Figure 17(a) illustrates the flux of neutrinos from the lation lengths of a few hundred kilometers, so the classical 8B reaction, based on SNO results. The best fit indicates that 2 average (12) applies and for θ 34◦ the reduction in νe flux 3 of the Sun’s electron neutrinos have changed flavor by the from averaging over vacuum oscillations∼ is by about 57%. time they reach Earth. Thus for vacuum oscillations the suppression of the electro∼ n neutrino flux detected on Earth would be by a factor of less than two, but the Davis chlorine experiment indicates a sup- 3. SNO Flavor-Mixing Solution pression by a factor of three. flavor conversion more severe than is possible from vacuum oscillations seems required, and Figure 17(b) shows the confidence-level contours for SNO this can be explained by the MSW resonance, as has been il- data, which suggest that the solar neutrino problem is solved lustrated in Fig. 16. Indeed, Fig. 16(d) indicates that for 10 by νe–νµ flavor oscillations with MeV neutrinos and parameters consistent with Eq. (50), the

2 +4.4 5 2 +2.4 ◦ MSW resonance gives a νe suppression by a factor of three. ∆m = 6.5 2.3 10− eV θ = 33.9 2.2 . (50) − × − ◦ Furthermore, vacuum oscillation lengths for the large-angle This large-mixing-angle solution means that νe is almost an solutions are much less than the Earth–Sun distance, which equal superposition of two mass eigenstates, separated by at would largely wash out any energy dependenceof the electron most a few hundredths of an eV. neutrino shortfall. Since the observations indicate that such an energy dependence exists (see Table I), and the MSW effect implies such an energy dependence (see Fig. 2 and Table II), C. KamLAND Constraints on Mixing Angles the MSW resonance is implicated as the primary source of the neutrino flavor conversion responsible for the “solar neutrino 1 2 KamLAND used phototubes to monitor a large container problem”. That is, the MSW resonance converts from 2 to 3 of liquid scintillator, looking specifically for electron antineu- (depending on the neutrino energy) of electron neutrinos into trinos produced during nuclear power generation in a set of other flavors within the body of the Sun, and these populations 17

21,22 6 2 1 TABLE III: Comparison of SNO results and Standard Solar Model (SSM) for solar neutrino ﬂuxes. Fluxes in units of 10 cm s− .

SSM νe flux SNO νe flux SNO νe/SSM SNOallflavors SNOall/SSM 5.05 0.91 1.76 0.11 0.348 5.09 0.62 1.01 ± ± ±

8 (a) (b) 7 Elastic Charged current scattering ) 10-4 -1 6 s * -2 5 ) cm 2 6 Neutral current

4 (eV 2 10-6 m

Standard ∆

flux (10 3 τ Solar Model ν +

µ 2 ν

1 10-8

0 0 1 2 3 4 5 6 0.01 0.10 1.0 6 -2 -1 2 νe flux (10 cm s ) tan θ

FIG. 17: (a) Flux of solar neutrinos from 8B detected for various flavors by SNO.21 The band widths represent one standard deviation. Note 2 8 that the bands intersect at the point indicated by the star, which implies that about 3 of the Sun’s B neutrinos have changed flavor between being produced in the core of the Sun and being detected on Earth. The Standard Solar Model band is the prediction for the 8B flux, irrespective of flavor changes. Note that it tracks the neutral current band, which represents detection of all flavors of neutrino coming from the Sun. (b) 2- flavor neutrino oscillation parameters determined by SNO22. The 99%, 95% and 90% confidence-level contours are shown, with the star at the most likely value. The best fit corresponds to the large-angle solution given in Eq. (50). are then only somewhat modified by the vacuum oscillations lar neutrino oscillations. However, additional resources are before the solar neutrinos reach detectors on Earth. available for those who wish to go further. The present paper is a synopsis of a more extensivediscussion that may be found in Chapters 10–13 of the book Stars and Stellar Processes E. A Tale of Large and Small Mixing Angles (Mike Guidry, Cambridge University Press, 2019). Those chapters contain 25 problems relevant to the present material that go into more technical depth than there is room for here, Initial theoretical prejudice favored a small vacuum mix- with complete solutions available online to instructors, and a ing angle. The MSW effect attracted initial attention because subset of solutions available online to students, Color lecture it was hoped that it might explain how a small mixing angle slides also are available to instructors from the publisher that could account for the solar neutrino deficit, since Fig. 16 indi- are suitable for teaching the present material. cates that the MSW resonance can generate almost complete flavor conversion even for small vacuum mixing angles. How- ever, data now indicate that the MSW resonance is indeed the solution of the solar neutrino problem, but for a large mixing VII. SUMMARY angle solution. Thus, in this tale an essentially correct physical idea, but with some initially incorrect specific assump- A concise introduction to neutrino vacuum and matter os- tions, led eventually to a surprising resolution of a fundamen- cillations in a two-flavor model has been presented. Sample tal problem. As a philosophical aside, this story representsa calculations with this formalism, in concert with observational beautiful example of the scientific method at work. data, demonstrate explicitly the resolution of the solar neutrino problem through neutrino flavor oscillations and the associated MSW matter resonance. The intent of this presenta- VI. ADDITIONAL TEACHING RESOURCES tion has been to make available to instructors in classes such as astrophysics and quantum mechanics at the advanced un- The material discussed here is a self-contained introduction dergraduate and beginning graduate level, and to motivated to learning and teaching the physics of vacuum and matter so- students through self-study, an introduction to the theory of 18 solar neutrino oscillations that assumes only a basic knowl- plied by a combination of h¯ and c having units of E T 4/L 3 edge of quantum mechanics, calculus, differential equations, to convert to normal units. Since in normal units and 2 2 matrices, and assumes no special prior knowledge of elementary× particle physics, quantum field theory, or solar [h¯]= E T [c]= L /T astrophysics. the required factor is h¯/c3 and

h¯ 4πE 4πEhc¯ Appendix: Natural Units L = = c3 × ∆m2 ∆m2c4 2 The formalism described here uses natural units, chosen 3 E eV = 2.48 10− km, × MeV ∆E2 such that h¯ = c = 1. Such units are particularly convenient for relativistic quantum ﬁeld theories and are ubiquitous in the literature of neutrino oscillations. Conversion between natu- where the neutrino energy E is in MeV and the energysquared difference ∆E2 corresponding to the mass squared difference ral units and more standard “engineering units” is basicallya 2 2 dimensional analysis problem. Let’s illustrate with an exam- ∆m is in eV . Transformation between natural units and ev- ple. Consider the neutrino oscillation length L. Multiply the eryday (engineering) units for other quantities may be carried ﬁrst expression in Eq. (11) (which is L expressed in h¯ = c = 1 out in a similar way. units) by c4/c4 = 1 to give

4πEc4 4πEc4 Acknowledgments L = = , ∆m2c4 ∆E2 Discussions of this material with Bill Bugg and Eirik En- where ∆E2 ∆m2c4 has units of energy squared. Let [x] de- deve, and with the students in Astronomy 411 and Astro- note the units≡ of a variable x and deﬁne our standard length physics 615 classes at the University of Tennessee are grate- unit as L , our standard energy unit as E , and our standard fully acknowleged. This work was partially supported by time unit as T . Then dimensionally, LightCone Interactive LLC. This work has been supported by the US Department of En- [E][c4] EL 4T 4 L 3 ergy and the Oak Ridge National Laboratory. ORNL is man- [L]= = − = L . [∆E2] E 2 E T 4 aged by UT-Battelle, LLC, for the US Department of Energy under contract no. DE-AC05-00OR22725. The ﬁnal expression should have units of length L in normal units, so the preceding result in natural units must be multi-

12 ∗ Electronic address: [email protected] L. Wolfenstein, “Neutrino oscillations in matter,” Phys. Rev. D17, † Electronic address: [email protected] 2369–2374 (1978). 1 C. Waltham, “Teaching Neutrino Oscillations,” Am. J. Phys. 72, 13 S. P. Mikheyev and A. Yu. Smirnov, “Resonant ampliﬁcation of 742–752 (2004). neutrino oscillations in matter and solar-neutrino spectroscopy,” 2 W. C. Haxton and B. R. Holstein, “Neutrino physics,” Am. J. Nuovo Cimento C9, 17 (1986). Phys. 68, 15–32 (2000). 14 H. Bethe, “Possible explanation of the solar-neutrino puzzle,” 3 W. C. Haxton and B. R. Holstein, “Neutrino physics: An update,” Phys. Rev. Lett. 56, 1305–1308 (1986). Am. J. Phys. 72, 18–24 (2004). 15 P. D. B. Collins, A. D. Martin, and E. J. Squires, Particle Physics 4 T. K. Kuo and J. Pantaleone, “Neutrino Oscillations in Matter,” and Cosmology, (Wiley Interscience, New York, 1989). Rev. Mod. Phys. 61, 937–979 (1989). 16 T. Morii, C. S. Lim, and S. N. Mukherjee, The Physics of the 5 A. Y. Smirnov, “The msw effect and solar neutrinos,” arXiv: hep- Standard Model, (World Scientiﬁc, Singapore, 2004). ph/0305106 (2003). 17 W. C. Haxton, “Adiabatic conversion of solar neutrinos,” Phys. 6 M. Blennow and A. Y. Smirnov, "Neutrino Propaga- Rev. Lett. 57, 1271–1274 (1986). tion in Matter," Advances in High Energy Physics 2013, 18 G. G. Raffelt, Neutrinos and stars, Proceedings ISAPP School, http://www.hindawi.com/journals/ahep/2013/972485/ (2013). Varenna, Italy; arXiv: 1201.1637 (2011). 7 J. N. Bahcall, Neutrino Astrophysics, (Cambridge University 19 S. F. King and C. Luhn, “Neutrino Mass and Mixing with Discrete Press, Cambridge, 1989). Symmetry”, Rep. Prog. Phys. 76, 056201 (2013). 8 J. N. Bahcall, M. H. Pinsonneault, and S. Basu, “Solar Models: 20 Y. Fukuda et al, “Evidence for Oscillation of Atmospheric Neu- current epoch and time dependences, neutrinos, and helioseismo- trinos,” Phys. Rev. Lett. 81, 1562–1567 (1998). logical properties, ”Astrophys. J. 555, 990 (2001). 21 Q. R. Ahmad et al, “Direct Evidence for Neutrino Flavor Trans- 9 J. N. Bahcall and M. Pinsonneault, “Solar models with helium formation from Neutral-Current Interactions in the Sudbury Neu- and heavy-element diffusion,” Rev. Mod. Phys. 67, 781 (1995). trino Observatory,” Phys. Rev. Lett. 89, 011301 (2002). 10 L. Bonolis, “Bruno Pontecorvo: From slow neutrons to oscillating 22 Q. R. Ahmad et al, “Measurement of Day and Night Neutrino neutrinos,” Am. J. Phys. 73, 487–499 (2005). Energy Spectra at SNO and Constraints on Neutrino Mixing Pa- 11 Particle Data Group at http://pdg.lbl.gov/. rameters,” Phys. Rev. Lett. 89, 011302 (2002). 19

23 K. Eguchi et al, “First Results from KamLAND: Evidence for Re- 24 S. Abe et al, “Precision Measurement of Neutrino Oscillation Pa- actor Antineutrino Disappearance,” Phys. Rev. Lett. 90, 021802 rameters with KamLAND,” Phys. Rev. Lett. 100, 221803 (2008). (2003).