STAN FO RD LIN EAR ACCELERATO R CEN TER

Fall 2001, Vol. 31, No. 3

Guest EditorMICHAEL RIORDAN

EditorsRENE DONALDSON, BILL KIRK

Contributing EditorsGORDON FRASER

JUDY JACKSON, AKIHIRO MAKIMICHAEL RIORDAN, PEDRO WALOSCHEK

Editorial Advisory BoardPATRICIA BURCHAT, DAVID BURKE

LANCE DIXON, EDWARD HARTOUNIABRAHAM SEIDEN, GEORGE SMOOT

HERMAN WINICK

IllustrationsTERRY ANDERSON

DistributionCRYSTAL TILGHMAN

A PERIODICAL OF PARTICLE PHYSICS

CONTENTS

FALL 2001 VOL. 31, NUMBER 3

The Beam Line is published quarterly by the Stanford Linear Accelerator Center, Box 4349,Stanford, CA 94309. Telephone: (650) 926-2585.EMAIL: beamline@slac.stanford.eduFAX: (650) 926-4500

Issues of the Beam Line are accessible electroni-cally on the World Wide Web at http://www.slac.stanford.edu/pubs/beamline. SLAC is operated byStanford University under contract with the U.S.Department of Energy. The opinions of the authorsdo not necessarily reflect the policies of theStanford Linear Accelerator Center.

Cover: The Sudbury Neutrino Observatory detectsneutrinos from the sun. This interior view frombeneath the detector shows the acrylic vesselcontaining 1000 tons of heavy water, surrounded by photomultiplier tubes. (Courtesy SNOCollaboration)

Printed on recycled paper

2 FOREWORDDavid O. Caldwell

FEATURES

4 PAULI’S GHOSTA seventy-year saga of the conceptionand discovery of neutrinos.Michael Riordan

15 MINING SUNSHINEThe first results from the SudburyNeutrino Observatory revealthe “missing” solar neutrinos.Joshua Klein

24 NEUTRINO EXPERIMENTSAT FERMILABNeutrino physics is thrivingin the Nation’s heartland.Paul Nienaber

32 THE ENIGMATIC WORLDOF NEUTRINOSTrying to discern the patterns ofneutrino masses and mixing.Boris Kayser

42 THE K2K NEUTRINOEXPERIMENTThe world’s first long-baselineneutrino experiment is beginningto produce results.Koichiro Nishikawa & Jeffrey Wilkes

50 WHATEVER HAPPENEDTO HOT DARK MATTER?Neutrinos may have had asignificant impact on thestructure of the Universe.Joel R. Primack

58 CONTRIBUTORS

DATES TO REMEMBER

2 FALL 2001

HE PAST FEW YEARS have been anexceptionally exciting time for physicistsinvolved in research on neutrinos. We now havetwo different confirmed (and one unconfirmed)pieces of evidence that neutrinos oscillate from

one type into another, which implies that they possess mass.These experiments have provided the first convincing experimen-tal evidence for new physics beyond the Standard Model, today’sdominant theory of elementary particles and the interactionsamong them. For two decades, ever since this theory took firmhold within the community during the late 1970s, particle physi-cists have been looking in every possible corner and under everyaccessible rock for this new physics. Now that we have suchevidence, neutrino studies will lead the research into what liesbeyond the Standard Model.

As a probe for new physics, neutrinos are unique amongparticles. As far as we know today, they are elementary particles,but unlike their fellow leptons and quarks, neutrinos areunencumbered by electrical charge and do not take part in thecomplicated strong interactions. They seem about as elementaryas particles can be—possessing just the tiniest bits of mass andengaging only in weak interactions with other leptons and quarks.Neutrinos’ exceptionally small mass may in fact provide a valu-able window into the very high-energy scales that are otherwisecompletely inaccessible at today’s particle accelerators. Anddespite the smallness of their mass, neutrinos may have played animportant role in establishing the structure of the Universe duringthe years immediately after the Big Bang.

FOREWORDDAVID O. CALDWELL

T1930 Wolfgang Pauli

predicts that

neutrinos exist.

1956 Frederick Reines and

Clyde Cowan discover

the electron neutrino.

1961 Muon neutrinos are

discovered at the

Brookhaven National

Laboratory.

1968 Ray Davis and

colleagues begin the

first solar neutrino

experiment.

1989 Experiments at SLAC

and CERN prove that

there are only three

kinds of light neutrinos.

1998 Super-Kamiokande

experiment reports

conclusive evidence for

neutrino oscillations.

2000 DONUT experiment

reports direct observa-

tion of tau neutrinos.

2001 SNO experiment finds

evidence that solar

neutrinos indeed

oscillate.

NEUTRINO TIMELINE

BEAM LINE 3

This special issue of the Beam Line appears at a pivotalmoment in the evolution of the neutrino physics, when we takestock of the new advances and try to glimpse what discoveriesmay lie just over the horizon. A historical article by MichaelRiordan helps familiarize readers with Wolfgang Pauli’s 1930 con-ception of the neutrino and its evolution over seven decades ofresearch. Joshua Klein, Paul Nienaber, Koichiro Nishikawa, andJeffrey Wilkes bring us up to date on recent neutrino experimentsas well as others planned for the near future. Theorists BorisKayser and Joel Primack attempt to interpret the pattern ofneutrino masses that seems to be emerging and assess whatimpact these ghostly particles might have had on the structure ofthe Universe.

Despite all the recent advances, there appears to be no end ofquestions about these elusive, fascinating particles and the effectsthat they may produce. This special Beam Line issue on neutrinophysics, edited by Michael Riordan, captures the excitement ofthis field.

David O. Caldwell has been doingresearch on neutrinos since 1980.He earned his Ph.D. in nuclearphysics at UCLA in 1953 and since1965 has been a member of thephysics department faculty at theUniversity of California, SantaBartbara. A Guggenheim Fellowand a Fellow of the AmericanPhysical Society, he is editor of therecently published book, CurrentAspects of Neutrino Physics.

AS THIS ISSUE was going to press, we learned about the

extensive damage to most of the phototubes in the Super-

Kamiokande experiment. This accident is not only devastating for

that research group but also for all of physics. Super-Kamiokande

has yielded so many insights about neutrinos and into physics

beyond the Standard Model that the inevitable delays are a severe

blow. We wish them as rapid a recovery as possible.

UPDATE

4 FALL 2001

HEN WOLFGANG PAULI conceived hisidea of the neutrino in 1930, it was substan-tially different from the ghostly particles rec-ognized today. That December he proposedthis light, neutral, spin-1/2 particle as a “des-perate remedy” for the energy crisis of that

time: that electrons emitted in nuclear beta decay had a con-tinuous, rather than discrete, energy spectrum. The crisishad grown so severe by the late 1920s that Niels Bohr beganto contemplate abandoning the sacrosanct law of energy con-servation in nuclear processes. Pauli could not countenancesuch radical unorthodoxy. Instead, he suggested that verylight poltergeists might inhabit the nucleus along with pro-tons and electrons. They should have a mass “of the sameorder of magnitude as the electron mass.” Normally boundwithin nuclei, they would have “about 10 times the pene-trating capacity of a gamma ray” after their emission. Hecould account for continuous beta-decay spectra by assumingthat this particle “is emitted together with the electron, insuch a way that the sum of the energies . . . is constant.”

Pauli clearly thought of his ghosts as actual constituentsof atomic nuclei, with a small mass and substantial interac-tion strength. He was trying not only to preserve energy con-servation in nuclear processes but also to dodge the severeproblems with spin and statistics that cropped up in nuclei—then imagined to consist only of protons and electrons. Thenitrogen nucleus, for example, was widely pictured as having14 protons and 7 electrons, but it did not seem to obey Fermistatistics, as expected of any object containing an odd num-ber of fermions. By adding seven more fermions to the heap,

by MICHAEL RIORDAN

Adapted from “Pauli’s Ghost: TheConception and Discovery ofNeutrinos,” by Michael Riordan, inCurrent Aspects of Neutrino Physics,D. O. Caldwell, Editor (Springer-VerlagBerlin, 2001).

WA Seventy-Year Saga of the Concep

The idea

of the

neutrino

has evolved

substantially

over the

seven decades

of its

existence.

BEAM LINE 5

Wolfgang Pauli, left, Niels Bohr, center,Erwin Schrödinger, and Lise Meitner at the1933 Solvay Congress (Courtesy Niels BohrArchives).

tion and Discovery of Neutrinos

6 FALL 2001

Pauli could explain why it behavedlike a boson. But nobody could fig-ure out how to cloister such light,speedy particles within the narrowconfines of a nucleus.

James Chadwick’s 1932 discoveryof a heavy fermion he dubbed theneutron resolved most of these prob-lems. Composed of seven protonsand seven almost equally massiveneutrons, the nitrogen nucleus nowhad an even number of fermions in-side and could easily behave like aboson. Enrico Fermi’s famous theoryof beta decay put the capstone on thegrowing edifice. Instead of inhabit-ing the nucleus as constituents, theelectron and “neutrino” (a namecoined by Fermi in 1931 to mean“little neutral object”) were to be cre-ated the moment a neutron trans-formed into a proton. Fermi evenwent so far as to indicate how theenergy spectrum of beta-decayelectrons depends critically on theneutrino’s mass. By comparing histheoretical curves with the measuredspectra near their high-energy endpoint, he concluded that “the restmass of the neutrino is either zero,or, in any case, very small in com-parison to the mass of the electron.”

Shortly thereafter, Hans Bethe andRudolf Peierls used Fermi’s theory toshow that the interaction of neutri-nos with matter had to be essentiallynegligible. In the energy range char-acteristic of beta-decay neutrinos,they would have a mean free path inwater of more than 1000 light years!Bethe and Peierls concluded “thereis no practically possible way ofobserving the neutrino.” Pauli himselfwas dismayed. “I have done a terri-ble thing,” he said. “I have postulateda particle that cannot be detected.”

Thus was the idea of the neutrinoborn, but it remained mostly anintriguing possibility for years. Evenafter reading Fermi’s paper, Bohr wasstill not convinced of its reality. “Inan ordinary way I might say that I donot believe in neutrinos,” Sir ArthurEddington remarked, “Dare I say thatexperimental physicists will not havesufficient ingenuity to make neu-trinos?”

WHEN George Gamowwrote “The Reality ofNeutrinos” in 1948, how-

ever, he could discourse about themwith confidence that they indeedexisted. Although nobody had yetdetected one directly, there were sev-eral indirect experimental proofs oftheir reality. Sensitive measurementsof the energy and momentum ofbeta-decay electrons and theirrecoiling nuclei in Wilson cloudchambers indicated that substantialquantities of energy and momentumwere missing. “This means someother particle must have been eject-ed at the same time as the electron,”he wrote. “These single-processexperiments leave little doubt that athird particle must be involved.”

Gamow even speculated that neu-trinos might be involved in thelethargic disintegration of the re-cently discovered pi-mesons andtheir lighter counterparts, thenknown as mu-mesons. After all,some kind of invisible entity wasspiriting energy away from thesetwo-body and three-body decayprocesses. Why not the same elusiveparticle involved in beta decays?

But little could be said conclu-sively about a particle that had thusfar evaded direct detection. Whether

Graph from Fermi’s paper on the theoryof beta decay, showing how the shapeof the emitted electron’s energy spec-trum varies with the possible mass of theneutrino.

BEAM LINE 7

the idea of placing a detector with-in 100 meters of an atomic-bomb ex-plosion, they decided instead to putit close to one of the nuclear reactorsthen in operation. Their first exper-iment, placed near one of the Han-ford Engineering Works reactors usedto breed plutonium for the Manhat-tan Project, involved a 300-liter tankof liquid scintillator. But the mar-ginal increase in signal they observedwith the reactor in operation was

its antiparticle was a completely dis-tinct entity or just a different spinstate of the same poltergeist couldnot then be determined. And the bestattempts at measuring its mass couldonly establish an upper limit of aboutone-twentieth the electron mass. Asthe 1950s began, however, this situ-ation was about to change dramati-cally.

In 1951, following atomic-bombtesting at Eniwetok atoll, Los Alamosphysicist Frederick Reines began con-templating experiments in funda-mental physics he might attempt.The Manhattan Project had providedintense new sources of neutrinos thatcould be used to ascertain moreabout them. Reines and ClydeCowan recognized that the recentlydeveloped organic scintillating liq-uids would allow them to build themassive detector required to observesuch ghostly particles. Together withthe intense neutrino fluxes gener-ated by atomic blasts or close to a fis-sion reactor, such a large detectormight finally overcome the daunt-ingly miniscule probability of a neu-trino interacting.

Reines and Cowan elected tosearch for evidence of the interaction

ν + p → n + e+,

which should yield a prompt lightflash in the organic scintillator dueto the positron’s annihilation withan atomic electron, followed a fewmicroseconds later by another flashdue to neutron capture. (Savvy read-ers will protest that antineutrinos,not neutrinos, participate in such an“inverse beta-decay” interaction, butthis distinction was not clear at thetime.) After considering and rejecting

Frederick Reines at work on an under-ground experiment in a South Africanmine in 1966. (Courtesy University ofCalifornia, Irvine)

8 FALL 2001

nearly swamped by cosmic-ray back-grounds.

Reines and Cowan then did a sec-ond experiment at the SavannahRiver reactor, which could generatea flux of 10 trillion antineutrinos persquare centimeter per second at aposition 11 meters away. (By then itwas becoming recognized that theneutrino and antineutrino are dis-tinct particles, with the latter be-ing produced in tandem with an elec-tron in beta-decay.) The detector,positioned 12 meters underground toreject cosmic-ray backgrounds, con-sisted of three tanks of organic scin-tillator, each viewed by 110 photo-tubes; between them sat two tanksof water with dissolved cadmiumchloride to promote neutron capture(see illustration on next page).

An antineutrino from the reac-tor occasionally interacted with aproton in the water, producing apositron and a neutron. The positronannihilated almost immediately withan atomic electron, yielding twogamma rays that were detected in thescintillator; about 10 microsecondslater, neutron capture by a cadmiumnucleus resulted in another burst of

gamma rays (see illustration at left).A delayed coincidence between thefirst and second gamma-ray burstswas taken as the signature of an an-tineutrino event; Reines and Cowanobserved three events per hour withthe reactor operating—much greaterthan backgrounds due to cosmic raysor accidental coincidences.

Elated by their discovery, Reinesand Cowan sent Pauli a telegram onJune 14, 1956: “We are happy toinform you that we have definitelydetected neutrinos from fission frag-ments by observing inverse beta-decay of protons.” According toReines, Pauli drank a whole case ofchampagne with his friends to cel-ebrate the discovery and drunken-ly penned a reply. “Thanks for themessage,” he wrote; “Everythingcomes to him who knows how towait.”

REVIEWING the status ofneutrino physics a year later,Reines and Cowan could cite

a variety of major improvements inthe understanding of this previouslyinvisible poltergeist. Delicate mea-surements of the electron spectrumin tritium beta decays had estab-lished that the neutrino’s mass wasless than 1/2000th that of the elec-tron. The lack of evidence for doublebeta-decay (in which two electronsare emitted) indicated that it wasmost probably a Dirac particle likethe electron, with neutrino and an-tineutrino distinctly different enti-ties. And the failure of Brookhavenscientist Ray Davis to find any Cl37

→ Ar37 conversions in a tank con-taining a thousand gallons of carbontetrachloride placed near the Savan-nah River reactor could also be

Diagram of the antineutrino-detectionscheme used in the Savannah Riverexperiment.

BEAM LINE 9

AMAJOR MYSTERY inthe late 1950s was whetherthe neutral particles emitted

in pion and muon decays were thesame neutrino and antineutrino asobserved in nuclear decays—or some-thing else. Because the strengths ofthese interactions were similar, andbecause they also violated parity, itwas widely assumed that the verysame particles were involved. Butif this were the case, then muonsshould occasionally have been seendecaying into an electron and aphoton (µ → e + γγ). If so, the neutrinoand antineutrino generated in three-body decays of muons could oc-casionally annihilate each other,yielding a photon in addition to thedeparting electron. Theorists calcu-lated that such processes shouldoccur about once in every 10,000muon decays, but accurate mea-surements indicated that nothinglike this occurred in many millions

explained in the same way: antineu-trinos could not induce such tran-sitions, while neutrinos should have.

The most striking advance inunderstanding neutrinos had comethe previous year in the wake of theearthshaking discovery of parityviolation. Tsung-Dao Lee and Chen-Ning Yang (among others) proposedto rescue the deteriorating situationby invoking a peculiarity of neutri-nos. If they were indeed Dirac par-ticles with absolutely no mass, neu-trinos themselves would violateparity because their spin vectorswould always be aligned along theirdirection of motion, while the spinsof antineutrinos would only pointthe opposite way. We say neutrinosare “left-handed” and antineutrinos“right-handed.” “Since this newmodel for the neutrino does not obeythe simple parity principle, no re-action involving such a neutrino canbe expected to conserve parity,”wrote Reines and Cowan.

A further consequence of thishypothesis was that the probabilityof reactor-produced antineutrinosinteracting with protons had to betwice as large—an effect that Reinesand Cowan had already begun toobserve. But the most convincingproof came from a sensitive experi-ment at Brookhaven led by MauriceGoldhaber. He determined the spindirection of the recoiling nucleus thatemerged after a europium-152nucleus had captured one of itsatomic electrons and emitted a neu-trino. From this he concluded thatthe neutrino is always left-handed,just as Lee and Yang had suggested.Thus the neutrino and antineutrinoappeared to be distinctly different,massless entities.

Artist’s conception of the detector usedby Reines and Cowan in their SavannahRiver experiment. Tanks I, II, and III con-tained liquid scintillator viewed on eachend by 55 phototubes. Tanks A and B,containing 200 liters of water with dis-solved cadmium chloride for neutroncapture, served as the target volume.

10 FALL 2001

of events. One way to accommodatethis apparent discrepancy was to saythat two different kinds of neutrinoswere involved in muon decay.

Intrigued by these questions andthe possibility of resolving them bymaking neutrino beams, MelvinSchwartz, Leon Lederman, JackSteinberger and their colleagues be-gan planning an experiment atBrookhaven. Spurred by his discus-sions with T. D. Lee, Schwartz rec-ognized that intense, high energybeams of protons soon to be availablefrom its Alternating Gradient Syn-chrotron would allow them to gen-erate neutrino beams with sufficientintensity (see illustration above).

With energies ranging up to sev-eral billion electron volts, these neu-trinos had interaction probabilitiesmore than a hundred times greaterthan reactor-born neutrinos, but alarge, massive detector was stillrequired to observe a sufficient num-ber of them. Schwartz and his col-leagues elected to build a 10 ton spark

chamber from aluminum plates. Ifneutrinos produced in pion or kaondecays (e.g., π→ µ + ν) were distinctfrom those in beta-decays, theyexpected to see only the long, pene-trating tracks of muons generated byneutrinos that interacted in the alu-minum. However, Schwartz recalled,“If there had been only one kind ofneutrino, there should have been asmany electron-type as muon-typeevents.”

In the initial run of this experi-ment, which began in late 1961, theyrecorded 34 events in which thereappeared a single muon track origi-nating in the aluminum plates. Therewere another 22 events having amuon and other particles, plus sixambiguous events that might havebeen interpreted as electrons. Butcomparisons with actual electronevents from a separate run showedlittle similarity.

Thus the neutrinos produced intandem with muons in pion andkaon decay are distinct from those

Plan view of the two-neutrino experimentat Brookhaven National Laboratory.Pions and kaons produced by protonshitting a beryllium target at far leftdecayed, yielding neutrinos (and anti-neutrinos) that traveled from left to right,penetrating the massive steel shieldingand striking the detector at the far right.

BEAM LINE 11

produced together with electrons innuclear beta decay. After Schwartz’spivotal experiment, particle physi-cists began calling the former “muonneutrinos” (or νµ) to distinguish theseparticles from the latter, known as“electron neutrinos” (or νe). When-ever a positive muon decays, forexample, it yields a positron, an elec-tron neutrino and a muon anti-neutrino (µ+ → e+ + νe + ν–µ). By 1962physicists recognized four distinct“leptons,” or light particles: the elec-tron, the muon, and their two respec-tive neutrinos (plus antiparticles).

THEORETICAL and exper-imental advances over theensuing decade resulted in a

revolutionary new picture of the sub-atomic realm that came to be knownas the Standard Model of particlephysics. In this theory, leptons andother particles called “quarks” areregarded as elementary, point-likeentities. The electromagnetic andweak interactions, previously thoughtof as distinct forces with vastly dif-fering strengths, are now consideredto be just two different aspects of oneand the same “electroweak” inter-action. The extreme feebleness of theweak interaction arises because it oc-curs via the exchange of ponderousspin-1 particles known as “gaugebosons,” which are difficult to con-jure up out of sheer nothingness. Inbeta decay, for example, a neutroncoughs up a massive, negativelycharged W boson and transforms intoa proton (n → p + W−); the W− imme-diately converts into an electron plusits antineutrino (W− → e− + ν–e). Onlythe left-handed electrons and neu-trinos (and their right-handed anti-particles) participate in these weak

interactions, thereby yielding theircharacteristic parity-violating prop-erty.

An inescapable requirement ofthis unification of the electromag-netic and weak interactions is theexistence of “neutral currents” thatoccur due to exchange of anothermassive, but neutral, boson Z.Instead of converting into a muonwhen it interacts with a nucleus viathe exchange of a W boson, forexample, a muon neutrino can in-stead glance away unaltered, re-maining a muon neutrino.

After years of searching, neutralcurrents were discovered in 1973 byan international collaboration ofphysicists working at the EuropeanCenter for Nuclear Research (CERN)on the Gargamelle bubble chamber.Filled with liquid freon, it wasexposed to beams of muon neutrinosand antineutrinos from CERN’s pro-ton synchrotron. The initial evidencecame from a few rare events in whichthese spookinos rebounded elasti-cally from atomic electrons (forexample, νµ + e– → νµ + e–), andimparted energy to them. These

One of the first neutral current eventsobserved in the Gargamelle bubblechamber at CERN. A muon neutrinotraveling invisibly from left to right strikesan atomic electron, which makes thegently arcing track in this photograph.(Courtesy CERN)

12 FALL 2001

electrons left wispy tracks in thechamber, whereas the neutrinos orantineutrinos crept away undetected.Subsequent confirmation came fromGargamelle and two experiments atFermilab in which muon neutrinosscattered inelastically from nuclei.

By the mid-1970s, it was clear thatneutrinos (and, in fact, all the otherleptons and quarks) were capable ofa new kind of weak interaction inwhich they maintained their iden-tity instead of transforming into apartner lepton (or quark). The dis-covery of these neutral currents,together with conclusive evidencefor a fourth, or charm, quark c toaccompany the initial trio—up u,down d, and strange s—providedstrong support for the Standard Mod-el. Quarks and leptons come in pairs:u and d, e and νe; c and s, µ and νµ.What’s more, quark and lepton pairscan be grouped into families of four,often called “generations.” Two such

families were recognized by 1976: thefirst includes the up and down quarksplus the electron and electron neu-trino, while the second contains thecharm and strange quarks plus themuon and muon neutrino. Particlephysicists could now discern a highlysatisfying symmetry among the ele-mentary entities in their newontology.

THREE YEARS earlier, twoJapanese theorists had sug-gested that a third family of

quarks and leptons might exist.Makoto Kobayashi and ToshihideMaskawa were seeking a way to in-corporate the mysterious phenom-enon of CP violation within theemerging structure of what was soonrecognized as the Standard Model.Discovered in 1964 by James Cronin,Valentine Fitch, and their colleagues,this phenomenon indicated that—atleast in certain decays of kaons—Nature is asymmetric under thecombined operations of charge con-jugation (C) and parity inversion (P).They observed that kaons behavedifferently, that is, if one replaces par-ticles by antiparticles and views theirinteractions in a mirror. Kobayashiand Maskawa could not obtain thisCP violation using only the twoknown families of quarks and lep-tons. But if they added a third familyto the mix, including two morequarks plus another charged leptonand its neutrino, they discovered thatit arose naturally.

At about the same time, MartinPerl and his colleagues were begin-ning their search for another heavy,charged lepton using the SLAC-LBLdetector at the new electron-positroncollider SPEAR. If such a heavy lepton

One of the first anomalous eµ eventsobserved after a “muon tower” had beenadded in 1975 to the SLAC-LBL detectorat the SPEAR collider. The muon travelsupwards through the tower in this com-puter reconstruction of the detector’scross section, while the electron movesdownward.

BEAM LINE 13

difficulties of making a sufficientlyintense beam of tau neutrinos anddetecting them unambiguously.Meanwhile, the two quarks expectedin the third family—the bottom band top t quarks—were isolated atCornell and Fermilab, leaving onlythe elusive tau neutrino remainingto be discovered. So dominant hadthe Standard Model become that fewparticle physicists seriously ques-tioned its existence.

Cosmological arguments aboutnucleosynthesis of light elementsduring the first few minutes ofthe Big Bang required that a thirdneutrino—and perhaps even afourth—ought to exist. The abun-dance of primordial helium-4 syn-thesized during this process isdetermined by the expansion rateof the Universe at that time, whichis sensitively related to the numberof different kinds of light neutrinos.As measurments became moreaccurate during the 1980s, primordialhelium-4 was observed to contributeabout a quarter of the visible mass inthe Universe, suggesting a third (anda remotely possible fourth) kind ofneutrino.

λ existed, he reasoned, pairs of themshould be produced in high-energyelectron-positron collisions:

e+ + e− → λ+ + λ−.

Depending on their mass, such lep-tons could have a variety of decays,two of which would be similar (byanalogy) to the muon’s. By searchingfor events in which one of thesehypothetical motes decayed into anelectron and the other into anoppositely-charged muon (plus un-seen neutrinos and antineutrinos),they hoped to find evidence for theexistence of a new heavy lepton—and (again by analogy) a thirdneutrino.

By 1974 Perl’s group began to findsuch “anomalous eµ events” in thedata samples being collected on theSLAC-LBL detector, and by 1975 theyhad dozens (see diagram on the left).But convincing their colleagues thatthese events were conclusive evi-dence for a heavy lepton took longer.In 1977, confirmation of the SLACresults began to come in from theDORIS electron-positron collider inHamburg. Further experiments onSPEAR and DORIS showed the massof this tau lepton τ to be around 1.78GeV and identified its decay into apion and tau neutrino (τ− → π−+ ντ).By the following summer, there waslittle doubt among particle physicistsregarding the existence of the taulepton.

STILL, IT TOOK physicistsover two more decades toachieve the direct detection of

a free tau neutrino, well separatedfrom its point of production. Theproblem occurred due to the

The quarks and leptons in the StandardModel. There are three, and only three,families of quarks and leptons. The lastto be isolated, by the DONUT experimentat Fermilab, was the tau neutrino,ντ.

The issue was settled in 1989by terrestrial experiments at newelectron-positron colliders—the SLCat SLAC and LEP at CERN—that werecapable of producing hordes of Zbosons. Within the structure of theStandard Model, additional kinds ofneutrinos give this particle additionalpathways to decay, thereby shorten-ing its lifetime. By making precisionmeasurements of the Z resonancepeak, physicists at the two facilitiesconcluded that there are three, andonly three, kinds of “conventional”light, weakly interacting neutrinos.Confirming the cosmological pre-dictions, these experiments put afirm lid on the complexity of theUniverse. Its table of fundamentalentities could have only threefamilies of quarks and leptons (seeillustration below).

Particle physicists

could now discern

a highly satisfying

symmetry among

the elementary

entities in their

new ontology.

14 FALL 2001

Direct experimental evidence forthe tau neutrino came finally in 2000,seventy years after Pauli conceivedhis novel idea. By examining millionsof particle tracks left on special three-dimensional photographic film,physicists working on the DONUTexperiment at Fermilab found fourevents that could only be interpretedas a tau neutrino colliding with anucleus and producing a tau lepton.(For further details, see the article byPaul Nienaber, this issue.) The thirdand final neutrino had finally leftsubtle, indirect footprints that con-firmed its long-anticipated existence.

THE IDEA of the neutrinohas therefore evolved sub-stantially since its concep-

tion in the early 1930s. Pauli’s ten-uous hypothesis was just the startingpoint for a lengthy process of theo-retical and experimental elaborationthat continues today. Where he atfirst regarded neutrinos as nuclearconsituents, Fermi showed how theycan instead be created in nucleartransformations. Where Pauli soughta minimal way to preserve the con-servation of energy and angularmomentum in individual betadecays, physicists have recentlyestablished that at least two—andmost likely all three—of the neutri-nos have a tiny bit of mass. Andwhere he speculated that this massmight well be greater than an elec-tron’s, physicists now think it mustbe at least a million times smaller.

But despite the great transforma-tions that have occurred in the ideaof the neutrino since 1930, Paulideserves due credit for being the onedaring enough to take the greatconceptual leap of introducing

another fundamental entity into theminimalist ontology of his day. Nodoubt the enduring success of hisbold scheme has encouraged theo-rists of later decades to repeat thisexercise whenever a cherished sym-metry or conservation law appears tobe violated. In this subtle way, theghost of Wolfgang Pauli still hauntsthe way particle physics is practicedtoday.

For Further Information

Good summaries of the history ofneutrinos can be found at:

http://wwwlapp.in2p3.fr/neutrinos/aneut.html

http://www.ps.uci.edu/~superk/neutrino.html

On Wolfgang Pauli, see:

http://www.physicstoday.com/pt/vol-54/iss-2/p43.html

http://www-groups.dcs.st-andrews.ac.uk/~history/Mathematicians/Pauli.html

BEAM LINE 15

MINING SUNSHINE

THE COMMUTE is only four kilometers, but it cantake an hour each way. The problem is neither trafficnor mass-transit delays, but the fact that the trip takes

you 6800 feet below the surface of the Earth. Much of the time istaken up by the required safety rituals: donning the reflective cov-eralls and hardhat, lacing up the steel-toed work boots, bucklingthe headlamp battery into the safety belt and placing your nametag on the peg board.

The schedule of a busy mine is driven by the need to moveequipment—not people—from level to level. You wait by the mineshaft until the “on-shift boss” decides it is convenient to bringyou down and calls for your levels: “Sixty-two to seven-thousand!—going down.” Both miners and physicists then climbinto an open steel box (the “cage”), which is suspended by a cableas thick as a man’s leg. The cage comfortably fits twenty people;it usually holds closer to sixty. On a good day, you can place bothfeet on the floor.

While the drop downward starts slowly, it eventually reaches atop speed of about 2200 feet/minute. Headlamps provide the onlylight, and you can just make out the rock wall of the shaft as itrushes upward past your face. Occasionally the darkness is brokenby a brief glimpse of one of the shallower mining levels blinkingby so fast that you barely register an image of a lighted tunnelreceding away. By the 4000-foot level, your ears start to feel theadditional pressure, and even the most seasoned miners snort andwork their jaws trying to adjust.

by JOSHUA KLEIN

The first

results

from the

Sudbury

Neutrino

Observatory

reveal the

“missing”

solar neutrinos.

16 FALL 2001

If they don’t stop to let off minerson one of the shallower levels, thetrip down to the 6800-foot level takesless than four minutes. As you stepoff the cage, a sign long-ago coveredin grime declares that this is “TheHome of the Sudbury NeutrinoObservatory.” To the miners, it isjust another level rich with nickel.

The commute is not yet over,however, as there is still over a mile’sworth of hiking left at this level. Thedark and the dust and the mud aretypical of any of the mine’s other tun-nels (called “drifts”), as is the factthat the rock above your head—near-ly a mile and a half of it—is kept fromcollapsing only by a brute-force com-bination of bolts 30-feet long, steelscreening, and concrete. Withoutthat support, the pressures at this

depth are so great that the rockwould quickly crumble.

A stiff breeze blows at your back,ventilation that keeps the airthroughout most of the mine near70°F all year round—much coolerthan the natural 100°F of the rock. Insome pockets where the ventilationdoesn’t reach, the air is as hot andhumid as a summer day in Phila-delphia.

The (quite literal) light at the endof this tunnel shines on two plain-looking blue doorways—one for peo-ple and one for railcars. The accu-mulated dirt and mud on your bootsmust be washed off before you enter,and once inside, the contrast withthe conditions in the drift could notbe more striking: bright lights, dust-free air, spotless floors, and whitepainted walls suddenly make it hardto remember that you’re at the bot-tom of one of the world’s deepestmines. To go further inside, you needto clean off even more: the boots,hardhat, and coveralls come off, andeveryone must shower and changeinto clean clothes and a hairnet.

Computers, fax machines, coffeemakers, telephones—they make italmost impossible to convince your-self that there is a mile and a halfof rock above your head! And thereis more: a water purification plant, acontrol room where shift operatorsmonitor and alter the way data aretaken, and a darkened, domed areawith rack upon rack of flickeringdata-acquisition electronics. Onlythe fact that there are no windows—no way to see the Sun from a mileand a half underground—remindsyou of where you are. Which is per-haps ironic, given that the princi-pal purpose of the Sudbury Neutrino

The SNO detector just before completionof the bottom of the acrylic vessel. Theacrylic used for this vessel is over 2.5inches thick and built to hold 1000 tonsof heavy water.

BEAM LINE 17

probe the details of the solar core.Yet, while Davis did prove that theSun emits neutrinos, his measure-ments also produced a surprise: com-pared to the detailed calculationsmade by John Bahcall and others, hedetected far fewer neutrinos thanexpected. The five experiments thatcame after his also saw similardeficits. The mystery of these “miss-ing” neutrinos is known as the solarneutrino problem, the oldest puz-zle in experimental particle physics.

Essentially there are three possi-ble ways in which neutrinos from theSun might seem to be missing. First,the measurements may simply beincorrect: the neutrinos are present,but we have undercounted them. Butthe different detectors and ap-proaches used by the various solarneutrino experiments make a com-mon error between them veryunlikely. Second, the theory of theSun may be wrong in some way;perhaps the Sun’s interior propertiesare different from what is expected,and fewer neutrinos are being cre-ated. The theory of the Sun correctlypredicts other solar characteristics,such as the way in which the Sun’ssurface vibrates in response to seis-mic activity, and it is thereforeunlikely that we would grossly mis-calculate the number of neutrinos.We are left with a third possibility:that the problem is not due to ourmeasurement of neutrinos or ourunderstanding of the Sun, but tosomething about the neutrino itself.

Davis’s experiment therefore pro-vided much more than a new way tostudy the Sun: it raised questionsabout a part of fundamental particlephysics we thought we alreadyunderstood. Our best theory of

Observatory (SNO) is to study sun-shine—not the sunshine visible tothe human eye, but sunshine madeof neutrinos, which travel to the bot-tom of the mine more easily thansunlight through a pane of glass.

THE SOLAR NEUTRINOPROBLEM

The sunshine that flows throughwindows—and down to the bottomof mines—is born in a tremendousnuclear reactor that rests at the cen-ter of the Sun. Sir Arthur Eddingtonfirst suggested that the energy storedin atomic nuclei provides the Sun’spower, pointing out in 1920 that thissource was “well-nigh inex-haustible”and “sufficient . . . tomaintain [an] output of heat for 15billion years.” (See John Bahcall’sarticle in the Winter 2001 Beam Line,Vol. 31, No. 1.) But proving that theSun is tapping nuclei for energy is noteasy—it requires the detection ofsome specific signature other thanthe light and heat that we see and feel(most of which is coming directlyfrom the boiling surface of the Sun,not its center). Fortunately for us, theneutrino is a common by-productof nuclear reactions, and thereforeobservation of solar neutrinos onEarth can prove Eddington’s hypothe-sis.

It was not until 1967, however,that Brookhaven National Labora-tory scientist Ray Davis and his co-workers constructed the first detec-tor to look for neutrinos coming fromthe Sun. Davis’s experiment was notonly expected to be a triumphantconfirmation of the theory but alsoto usher in a new era of solar physicsin which neutrinos would be used to

Ray Davis, circa 1966. The photographwas taken in the Homestake gold minein South Dakota shortly before the firstsolar neutrino experiment began opera-tion. (Courtesy John Bahcall)

18 FALL 2001

elementary particles, the StandardModel, holds that there are three dif-ferent types (or flavors) of neutrinos—electron neutrinos (written νe), muonneutrinos (νµ), and tau neutrinos (ντ)each closely associated with the cor-responding charged particle. The Suncan produce only electron neutrinos,and therefore the six solar-neutrinoexperiments prior to SNO havesearched primarily for this one fla-vor. But if the Standard Model iswrong, and the three flavors of neu-trinos are not completely distinct butcan change from one type into an-other, we may have an explanationfor the solar neutrino problem. If theelectron neutrinos born in the cen-ter of the Sun change into one of theother types on their way to Earth,then the earlier experiments willhave recorded fewer neutrinos thanexpected; the changelings would sailright through the detectors withoutbeing noticed.

The changing of neutrinos backand forth from one flavor to anoth-er is usually referred to as neutrinooscillations (see Maury Goodman’sarticle in the Spring 1998 Beam Line,Vol. 28, No. 1). Strong evidence thatneutrinos can oscillate from one fla-vor to another was reported in 1998by the Super-Kamiokande experi-ment in Japan, which observed muonneutrinos produced in the Earth’s up-per atmosphere by cosmic rays. WhatSuper-Kamiokande observed wasthat the number of muon neutrinosmeasured depends on where themuon neutrinos were produced—above the detector, where theyneeded to travel only 10–100 km be-fore observation, or below it all theway on the other side of the Earth,where they would have to travel

Cut-away view of the SNO detector, showing the acrylic vessel containing 1000 tonsof heavy water at its core.

BEAM LINE 19

thousands of kilometers to reach thedetector. If muon neutrinos can os-cillate into another type of neutrino,they would produce exactly the typeof up-down, distance-dependent dif-ference in the number of neutrinoswitnessed by Super-Kamiokande (seeJohn Learned’s article in the Winter1999 Beam Line, Vol. 29, No. 3).

For solar neutrinos, however, thedistance from the production point(the Sun) to the detection point (theEarth) does not vary enough to allowthe kind of measurements made withSuper-Kamiokande’s atmosphericmuon neutrinos. Rather than lookfor distance-dependent effects, theway to prove that the electron neu-trinos produced in the Sun areoscillating into muon neutrinos ortau neutrinos is to search directly forevidence of these neutrinos. SNO isdesigned to do just that: determinewhether or not solar neutrinos otherthan electron neutrinos are arrivingfrom the Sun. If this were true, itwould not only solve the long-stand-ing solar neutrino problem, but alsoprovide the most direct evidence yetthat neutrinos do indeed oscillate.

THE SUDBURY NEUTRINOOBSERVATORY

Like an invisible man on thebeach who can be “seen” only bytracking his footprints, neutrinos canbe observed only by the traces theyleave behind as they pass through adetector. We cannot see themdirectly. In Davis’s experiment, thesetraces came in the form of atoms ofthe element argon, which were cre-ated when neutrinos collided withatoms of chlorine. Every so often thephysicists would flush out the detec-

tor and count the number of argonatoms that had accumulated, deduc-ing from this evidence how manysolar neutrinos must have traversedthe detector. Other experiments thatused gallium rather than chlorineoperated in a similar way: essentiallyvisiting the beach now and then justto see if any additional footprints hadbeen made.

A different approach was taken bythe Kamiokande experiment—andsubsequently by Super-Kamiokande—which was able to view the neu-trino “footprints” as they were beingmade. In these experiments, thedetectors were made primarily ofwater, but rather than changing theatoms of water into something else,neutrinos scattered their electrons inthe same way that one billiard ballcollides with another. Each time thishappened, the scattered electron pro-duced a cone-shaped flash of bluelight called Cerenkov light, whichwas quickly converted into electri-cal signals by photon detectors placedoutside the water volume.

The footprint metaphor howeverfails to highlight the most criticalproblem of neutrino detection. Neu-trino interactions with matter are soweak that the chances of a footprintoccurring are incredibly small; a typ-ical neutrino can travel easilythrough a light year’s worth of lead.Even then, every so often—essentiallyjust by accident—one does collidewith something, such as a chlorineatom in Davis’s experiment or anelectron in the Super-Kamiokandeexperiment. To increase the chancesof observing neutrinos, a detectormust therefore contain lots ofmaterial—lots of chlorine or lots ofelectrons, for example—thus giving

the neutrinos many opportunities foran accidental collision. Solar neu-trino detectors therefore tend to bevery big: the Super-Kamiokandeexperiment, for example, holds over50,000 tons of water.

There is one further difficulty intrying to detect neutrinos from theSun. To return one last time to ourinvisible man, imagine that the beachis crowded with swimmers and sun-bathers each of whom also leavesfootprints. Trying to figure out whichfootprints are meaningful (especiallygiven that the number made by theother beachgoers is far larger thanthose made by our invisible stroller)is an extremely difficult task. Cosmicrays are one example of this problem,since they can create Cerenkov lightin the same way neutrino interactionsdo. And we are being bombarded bycosmic rays all the time: if each cos-mic ray were a raindrop hitting thesurface of the Earth, we would besoaked by a steady downpouramounting to about one inch everyfew hours. The bottom of a mine istherefore an ideal place to study neu-trinos because most cosmic rays arestopped by the rock before they reachthe detector, while neutrinos are un-affected. To solar neutrinos, a kilo-meter of rock hardly matters at all,nor do the thousands of kilometersbelow the detector through whichthey travel upward during the night.

The Sudbury Neutrino Observa-tory uses the same basic techniqueas the Kamiokande experiments—looking for Cerenkov light producedby neutrinos interacting in water. Atotal of 7000 tons of water sits in acavern 22 m wide by 34 m high,carved out of rock 2000 m beneaththe surface of INCO Ltd.’s Creighton

20 FALL 2001

Mine in Sudbury, Ontario (see illus-tration on page 18). At SNO’s depth—deeper than any other solar neutrinoexperiment—only three cosmic rayspass through the detector each hour.The Cerenkov light created by theneutrino interactions is detected byan array of 9500 photomultipliertubes (PMTs), each capable of regis-tering a single photon of light. Thissensitivity to single photons is nec-essary because a neutrino event typ-ically produces only 500 photons. ThePMTs are supported by a stainlesssteel geodesic sphere 17.8 m indiameter.

Nested inside the PMT supportsphere is a second sphere 12 m indiameter, built from 5.5-cm thicktransparent acrylic. This vessel andits contents are what makes SNOunique: rather than holding ordinarywater, which has two hydrogenatoms bound to one oxygen atom(H2O), the acrylic vessel holds 275,000gallons (1000 metric tons) of heavywater, which has two deuteriumatoms bound to one oxygen atom(D2O). Deuterium is the same ashydrogen in all respects except thatit has a proton and a neutron in itsnucleus (called a deuteron). It is thisextra neutron that allows SNO to dis-tinguish different types of neutrinosand thus determine whether there areneutrinos other than electron neu-trinos reaching Earth from the Sun.

Cosmic rays are not the onlysource of false signals that SNO hasbeen designed to avoid. Just aboutany material one can find on Earth—water, steel, mine dust—has a tinyamount of naturally occurring radio-active contamination. These radio-active nuclei produce charged par-ticles that can generate the same

Outside view of the SNO detector after construction was complete. Supported by ageodesic sphere, 9500 photomultiplier tubes view the acrylic sphere within.(Courtesy Lawrence Berkeley National Laboratory)

BEAM LINE 21

Cerenkov light as neutrino interac-tions or cosmic rays. For this reason,nearly everything used in the con-struction of the detector—even theglass used in the photomultipliertubes—was designed specifically forSNO out of materials that are lowin radioactivity. Locating the detec-tor in an active mine adds an extrachallenge, since the dust and mudhave relatively high levels of radio-active contamination. Just a table-spoon of mine dust dropped into the275,000 gallons of heavy water wouldhave enough radioactivity in it tomask the neutrino signals.

The entire underground lab istherefore operated as a clean room,the air is continuously filtered andeverything and every person enter-ing it must be cleaned off to removeany telltale remnants of the outsidemine. The heavy water in the detec-tor is also purified, removing not justdust but radon, a radioactive gas. Andas a last line of defense, ordinarywater fills the space outside theacrylic vessel, shielding the heavywater from radioactive signals orig-inating in the surrounding rock. ThisH2O shielding, the great depth, andthe insistence on ultrapure materi-als as well as dust-free air and veryclean people, has made the centerof the SNO detector the point of low-est radioactivity on Earth.

DETECTING NEUTRINOS

When a neutrino enters the SNOdetector, several things can happen.By far the most likely is that itsails straight through unobserved;although about five million solarneutrinos pass through each squarecentimeter of the detector every

second, only about five of them pro-duce a signal in the entire detectorin any given day. For those few neu-trinos that do interact, SNO’s heavywater allows three possible scenar-ios. For one, an electron neutrino canbe absorbed by a neutron inside adeuteron; the neutron then becomesa proton and emits an electron (seetop diagram at right) which then pro-duces Cerenkov light that is detect-ed by the PMTs. This neutrino-ab-sorption reaction is special becauseit only occurs for electron neutrinos.

SNO’s second reaction is the sameelectron-scattering interaction as wit-nessed by the Kamiokande andSuper-Kamiokande experiments. Asolar neutrino collides with anelectron and kicks it out of its D2Omolecule. As with the neutrino-absorption reaction, the scatteredelectron produces Cerenkov lightthat is detected by the photomulti-plier tubes. SNO can distinguishthese electrons from those created inthe neutrino-absorption reaction bythe fact that, like a billiard ball struckdead on by the cue ball, the scatteredelectron emerges close to the direc-tion of the incoming neutrino. There-fore knowledge of the electron’s di-rection and the Sun’s position whenthe event occurred are required toidentify it as an electron-scatteringevent. Although this reaction canoccur for either muon neutrinos ortau neutrinos, it happens most often(6.5 times more often) for the elec-tron neutrinos.

In the third reaction, which is alsosensitive to muon and tau neutrinos,the neutrino breaks the deuteron upinto a separate neutron and proton.The neutron itself cannot produceCerenkov light, but eventually it is

d e

p

p

νe pn

eeν

ν

d

p

ν νp

n

n

Scattering of an electron by a neutrino.This process occurs most often withelectron neutrinos, but can also occurfor muon and tau neutrinos.

Absorption of a neutrino by a neutron ina deuteron. The neutron changes into aproton, emitting an energetic electron.

Breakup of a deuteron by a neutrino.This process happens with equalprobability for any neutrino type.

22 FALL 2001

the Sun. For this to be true, neutrinosborn as electron neutrinos in thesolar center must be changing theirflavor somewhere on their way to theEarth.

THE FIRST RESULTS

The SNO experiment therefore hastwo good ways of demonstrating thepresence of muon and tau neutrinoscoming from the Sun. The deuteron-breakup reaction has the advantagethat it occurs for all neutrino flavorsequally, and thus it directly measuresthe Sun’s total neutrino output.Because of this, we would expect tofind a big difference between thenumber of neutrinos measured throughdeuteron breakup (all kinds) and thenumber measured through neutrinoabsorption (just electron neutrinos).Using the electron-scattering reac-tion is not as easy. Because it occursless than one sixth of the time formuon and tau neutrinos as for elec-tron neutrinos, the difference be-tween the total number of interac-tions measured with this reactionand the number measured with theneutrino-absorption reaction will notbe very large unless most of the neu-trinos coming from the Sun havechanged from electron neutrinos intomuon or tau neutrinos.

But the electron-scattering reac-tion does have one great advantage:the Super-Kamiokande experimenthas already used this reaction toobserve neutrinos from the Sun. Overseveral years, this experiment has ob-served that the number of neutrinosproducing electron-scattering reac-tions in the detector was 45 percentof what had been required by theory.If the theory of the Sun is correct,

captured by one of the other deu-terons in the heavy water, producinga gamma ray that scatters an electronin the water. This secondary electronthen produces Cerenkov light. In thisdeuteron-breakup reaction, all neu-trino flavors are on an equal footing—a muon neutrino or a tau neutrino hasthe same probability of interactingthis way as has an electron neutrino.

Using these three reactions, prov-ing that neutrino oscillationsexplain the solar neutrino problembecomes very simple: if the numberof neutrinos measured with eitherthe electron-scattering reaction orthe deuteron-breakup reaction isgreater than the number measuredwith the neutrino-absorption reac-tion, their must be neutrinos otherthan electron neutrinos coming from

10/16/1999 14:31:22 Run: 5566 GTID: 375971

Computer reconstruction of a typicalsolar neutrino event in the SNO detector.The filled-in dots correspond to photo-multiplier tubes that have detectedphotons; the resulting cone of Cerenkovlight produces a ring when projectedonto the sphere.

BEAM LINE 23

then the missing 55 percent are muonor tau neutrinos that the electron-scattering reaction has failed todetect. What is needed, therefore, isa separate determination of the fluxof electron neutrinos alone; once thisis known, we can conclude that anyadditional flux is due to these otherkinds of neutrinos.

SNO’s first order of business wasthus to use the neutrino-absorptionreaction to measure the number ofelectron neutrinos and to comparethis to Super-Kamiokande’s mea-surement of the number of neutrinosobserved with the electron-scatter-ing reaction. In 241 days of takingdata, SNO detected 1169 neutrinoevents, about 950 of which weredetermined to have been due toneutrino-absorption reactions. Thetheory of the Sun predicted that thereshould have been over 2700 electronneutrinos detected this way (assum-ing no oscillations) during thistime—in other words, SNO onlyobserved 35 percent of the expectednumber. The fact that Super-Kamiokande measures 45 percent ofthe expected number using a reac-tion that can detect all neutrinoflavors, while SNO measures just 35percent of the expected number usinga reaction that detects only electronneutrinos, can mean just one thing:there are other kinds of neutrinoscoming from the Sun.

For the first time, therefore, thereis clear evidence that neutrinos fromthe Sun can change from one typeinto another. The Standard Model,which predicts that the three neutri-no flavors are separate and distinct,must be altered to accommodate thiseffect. But there is an added bonus,one that starts to bring to an end the

long search begun over 30 years agoby Ray Davis. The additional neutri-nos that Super-Kamiokande observeswith the electron-scattering reactionare the first confirmation that thetheory of the Sun is indeed correct!

Recall that the electron-scatteringreaction only detects 1/6.5 of themuon and tau neutrino flavors, so the10 percent extra flux that Super-Kamiokande sees must mean thatactually about 65 percent of neu-trinos coming from the Sun aremuon or tau neutrinos. Adding this65 percent to the 35 percent worth ofelectron neutrinos which SNO mea-sures gives us 100 percent of theexpected total. After 30 years, we findthat there really aren’t any neutrinos“missing” after all. They were justmore difficult to detect! Our theoryof the Sun therefore appears to bemore correct than our fundamentaltheory of matter.

THE FUTURE

There is still much left for SNO to do.In particular, measurements with thedeuteron-breakup reaction willprovide a much more accuratedetermination of the total number ofneutrinos coming from the Sun.Comparing the number of neutrinosmeasured with the deuteron-breakupreaction to those that are measuredwith the neutrino-absorption reac-tion will provide a much betterdetermination of the degree to whichneutrinos oscillate. In addition tothis, SNO should be able to look forchanges over time in the Sun’s out-put of neutrinos, observe theirenergies, and measure other char-acteristics that can tell us aboutproperties of the solar core.

Gazing at sunshine from thebottom of Creighton mine, theSudbury Neutrino Observatory hasalready learned two importantthings: that neutrinos born in theSun do change into other kinds ontheir journey toward Earth, and thatthis change is responsible for the neu-trino deficit first noticed by RayDavis. He started out trying to learnsomething about the Sun, but whathe and the rest of us discovered wasthat we first needed to learn some-thing more about the neutrinosthemselves. And while there are stillmany ways yet in which we will usethe Sun to help us understand neu-trinos, we can now begin anew theproject that was postponed afterDavis’s first observations—and startusing neutrinos to understand theSun itself.

For Further Information

To learn more about the SudburyNeutrino Observatory, go to:

http://owl.phy.queensu.ca

24 FALL 2001

ARTICLE PHYSICS EXPERIMENTS oftenget lumped under the general heading of “bigscience.” Particle accelerators are gargantuan,detectors are huge and complex, and today’smulti-institutional collaborations generatesocial dynamics whose unraveling would

baffle the most sophisticated supercomputer. Neutrinoexperiments are no exception to this observation; thedetectors come in three sizes: large, extra-large, andpositively Bunyanesque. Yet, like the graceful hippopotamiin Walt Disney’s Fantasia, a “little” neutrino experimentcan on occasion dance briskly in, do some beautiful tour-de-force physics, and take a bow—all for a relatively modestticket price. Two premier danseurs of this sort on theFermilab stage are the DONUT and BooNE experiments.DONUT (Direct Observation of the Nu-Tau) has completedits run and is now accepting bravos for being the firstexperiment to detect the tau neutrino. BooNE (BoosterNeutrino Experiment) is waiting in the wings, warming upto delve more deeply into the peculiar pas de deux ofneutrino oscillations.

According to the Standard Model, our current bestworking theory of elementary particles and fundamentalforces, a neutrino can interact and change into its electricallycharged lepton partner by emitting or absorbing the chargedcarrier of the weak force, the W boson. This type of exchangeis called a “charged-current” interaction (to distinguish it

Neutrino Experiments at by PAUL NIENABER

PNeutrino physics

is thriving

in the

Nation’s

heartland.

BEAM LINE 25

Fermilab

from the exchange of a neutral Z boson). The weak force also obeys a two-fold lepton-number conservation rule: both the number of leptons and theirkind remains unchanged. Interacting electron neutrinos yield only electrons,muon neutrinos only muons, tau neutrinos only taus. Neutrinos and theircharged partners can switch back and forth within a Standard Model genera-tion, but they cannot jump between generations. This conservation rulemakes the charged-current interaction a useful “generation tagger”: if yousee a muon zipping through your neutrino detector, and you can ascertainthat it came directly from a neutrino interaction, you can be sure it wasproduced by a muon neutrino.

Because neutrinos are immune to the strong and electromagneticinteractions, they provide a unique “weak force scalpel” for investigating theproperties of particles. Electron neutrinos are difficult to concentrate into abeam whose direction and energy can be varied, so muon neutrinos have be-come the scalpel of choice. The recipe is fairly straightforward: take a protonbeam, smack it into a block of material, magnetically focus and direct thecharged pions and kaons spewing forth from the collision, allow them to de-cay into muons and muon neutrinos, filter out the charged decay productsand any other unwanted debris, and voila! You have a beam of muonneutrinos.

But because neutrinos interact so weakly, the chances of any singleone interacting in a detector are minuscule. The probability of witnessingan interaction depends on the number of neutrinos, their energy, and thenumber of other particles available for them to hit. Getting a reasonablenumber of events therefore requires lots of neutrinos, lots of energy, and lots and lots of detector material—hence the Brobdingnagian scale of neutrinoexperiments.

26 FALL 2001

WITH THE DISCOVERYof the top quark at Fer-milab in 1995, the set of

six fundamental constituents of thestrongly interacting particles in theStandard Model was complete. Thelepton six-pack, however, still hadone gaping hole—the tau neutrino re-mained the most elusive of its stand-offish cousins. Experiments at SLACand CERN had determined that therewere at most three conventional,lightweight neutrinos, and there wasindirect evidence from the missingmomentum observed in decays of thetau lepton, but no definitive sight-ings (see “Pauli’s Ghost,” by MichaelRiordan, this issue).

Understanding why the tau neu-trino is so camera shy takes us backto the charged-current interaction.Electron neutrinos make electrons,

which interact almost immediatelyin detector material and produce adistinctive shower of particles. Muonneutrinos engender muons, whichtravel relatively undisturbed throughmatter before decaying and thusleave long tracks in detectors. Tauneutrinos make taus—and there’s therub. The tau lifetime is less than onethird of a trillionth of a second; mul-tiply this brief instant by the speedof light and you get a decay lengthliterally the size of a gnat’s whisker,about 90 microns. If a tau is ex-tremely relativistic, it may travel afew millimeters before decaying, buttracing these microscopic paths todiscover any kinks in them demandsextremely high-precision tracking.This requirement, coupled with thesmall likelihood of tau neutrinos everinteracting, makes detecting theman extraordinary challenge. To riseto that challenge came the DONUTexperiment, which ran at the Fermi-lab Tevatron in 1997.

There are two hurdles to over-come if you want to bag a tau neu-trino: first you gotta make ‘em, andthen you gotta take ‘em. To make tauneutrinos, you employ a variation onthe muon-neutrino recipe, but in-stead of using pions and kaons as theparents, you must employ muchheavier mesons. The lepton-numberconservation rule says you can onlymake a tau or its neutrino in tandemwith an antilepton of the third gen-eration. To get tau neutrinos, there-fore, you have to produce a parentparticle that includes a tau (whosemass is a whopping 1777 MeV, almosttwice the mass of a proton) in itsdecay chain. DONUT’s tau neutrinoscome from the decay of DS mesons,produced by slamming the Tevatron’s

IncomingNeutrino

OutgoingLepton

ChargedWeak Boson

Target

Calorimeter

Drift Chambers

Magnet

Emulsion Target

Steel Shield

15 meters

Muon Detector

NeutrinoBeam

A neutrino charged-current interaction.

DONUT detector for direct observation oftau neutrinos.

BEAM LINE 27

800 GeV protons into a tungsten tar-get. These mesons are also heavy(1970 MeV) and short-lived, so noattempt to focus them is made.

The aim of the experiment is tolook for charged-current scatteringof the resulting tau neutrinos. Thisprocess will make final-state taus,but how does one corral theseevanescent heavyweights? The keyto this experiment, the factor thatmakes it both beautiful and ratherdifficult, is the use of a large targetcomposed of photographic emulsionplates. These plates are just standardblack-and-white film medium, silverbromide, suspended in a gel. Insteadof a thin layer of this substance coat-ing a 35 mm piece of plastic film,however, DONUT’s plates containthe bulk emulsion deposited onsquare sheets about the size of a largepizza box (50 cm square). Chargedparticles traversing the emulsionleave tracks of exposed grains, andfinding a vertex (where the tau neu-trino interacted) that spawns akinked track (where the resulting taulepton subsequently decayed) pro-vides the “smoking gun,” evidencefor a tau neutrino.

Isolating such a vertex with atrack that kinks a scant few milli-meters downstream of it is no easyfeat. The emulsion stack is only 36meters downstream of the tungstentarget, which means that extensiveshielding must be used to keep theemulsion from being swamped bysuperfluous tracks. The usual pas-sive shielding (which filters outunwanted particles by letting theminteract in material) is not enough;DONUT’s emulsion stack needed ac-tive elements (magnetic fields todivert charged detritus) as well.

DONUT was also in the unusualposition of having to deal with neu-trino backgrounds, since the protoninteractions in the tungsten produceplenty of electron and muon neutri-nos, too. These extra neutrinos caninteract in the emulsion as well andrender the stack awash in tracks.

To help pick out a few specificbent needles from this huge haystackof tracks, the experimenters posi-tioned a series of scintillating-fibertracking detectors among the emul-sion plates. These trackers, togetherwith other detectors downstream ofthe emulsion array, helped identifythe particles and trace out theirpaths. The vertex and track recon-struction enabled a tiny, preciselyselected region of the emulsion to bepinpointed for closer scanning andmeasurement. The actual digitiza-tion and scanning of the emulsion,

One of the four tau neutrino eventsrecorded by the DONUT detector. Thetrack with a kink is a tau lepton decayinginto an invisible tau neutrino and anotherparticle.

28 FALL 2001

direct sightings of the tau neutrino,these events confirmed at last that itindeed exists.

AT ABOUT THE TIME thatthe DONUT collaborationwas looking for telltale

kinks, evidence was accumulatingfrom other quarters that the StandardModel might not be quite right. Sev-eral experiments had seen hints thattwo fundamental features thought tohold for neutrinos—the expectationthat they are massless and that theycannot change from one kind toanother—were in fact incorrect.Rather loose upper bounds had beenset on the neutrino masses by care-fully adding up all the visible energyin certain particle decays that pro-duce neutrinos. And the prohibitionon leptons jumping from one gen-eration to another had been probedby searches for exotic processes suchas µ → eγ.

One way to establish more strin-gent limits is to examine the curiouspossibility of neutrino oscillations(see box at left). Understandinghow such oscillations might oc-cur requires us to step into thesometimes-counterintuitive worldof quantum mechanics. According tothe Standard Model, neutrinos aremassless and conserve lepton num-ber. But if this is not in fact the case,then neutrinos can have differentmasses, and a neutrino created as onekind (say an electron neutrino,produced in the nuclear reactionsoccurring in the Sun’s core) couldevolve, or oscillate, into another kindas it zooms along through space.

Quantum mechanics tells us thatthe characteristics of any neutrinooscillation—its amplitude and repeat

section by tiny section, was done ata unique facility in Nagoya, Japan,one of the few of its kind in theworld. In July 2000, after data reduc-tion and a careful analysis to screenout backgrounds, the DONUT teamshowed the world four events with adistinctive kink in one track, indi-cating a tau lepton decay. The first

Neutrino Oscillations

HERE’S A GOOD WAY to think about neutrino oscillations. Sup-pose you could only hear a single pitch (or frequency) of sound at atime: That is, your “sound detector” could only tell that a note was a G,

or a B, or a D; it could not detect a mixture of these notes. If a note (you hear)starts out as a G (or a B or D for that matter), it would stay that way forever.This is the way that charged leptons behave.

Neutrinos play by different rules, however, and admit the possibility that anote originating as a G (an electronneutrino, say) can “detune” as it trav-els, developing B or D components (a muon or tau neutrino). This kind ofbehavior is analogous to that of thestrings in a guitar, which—onceplucked—can excite lesser vibrationson the other strings.

Does this mean that one wouldhear chords in neutrino beams? No—remember, you can detect only one“pitch” at a time. What it does mean,however, is that if you create a purebeam of 1000 Gs, say, and set up alistening post some distance away,you might get 990 Gs and 10 Bs. What fluctuates in neutrino oscilla-tions is the probability that a neutrinocreated as one specific kind will bedetected as that same kind (oranother kind) of neutrino.

All GNo B

Distance Traveled

Pro

babi

lity

Mostly GSome B

G

B

1

0

Oscillation probability: even though theneutrino beam starts out as 100 per-cent G’s, as it moves through space,the beam can gain (and lose, and thenregain) some B’s. What oscillates is theprobability.

BEAM LINE 29

length—are governed by four factors.Two of them are experiment-specific:the neutrino energy E and the source-to-detector distance L. The other twoare intrinsic properties of how theneutrinos fluctuate back and forthfrom one kind to another: the oscil-lation strength itself (the amplitudeof the wave) and the absolute differ-ence between the squares of theirtwo masses ∆m2 = |m1

2 − m22|. One of

the current problems in neutrino-os-cillation physics has to do with thislast parameter, ∆m2. Since there arethree known kinds of neutrino, therecan only be two independent differ-ences between their mass states. Thethree classes of experiments cur-rently yielding positive indicationsof neutrino oscillations—thosestudying neutrinos from the Sun,those observing neutrinos frommuons produced in Earth’s atmos-phere, and two accelerator-basedexperiments—do not yield a coher-ent set of ∆m2 values (see article byBoris Kayser, this issue). This dis-crepancy could point to a new, fourthkind of neutrino (called a “sterile”neutrino because it cannot interactwith ordinary matter). Or it couldimply that one (or more) of theseexperiments is not really seeingoscillations but another effect.

The one accelerator-based exper-iment that reported a neutrino-oscillation signal is the Liquid Scin-tillator Neutrino Detector (LSND),which detected neutrinos from pionsproduced by a medium-energy pro-ton beam at the Los Alamos NationalLaboratory. The collaboration ob-served an excess of events abovebackground that can be explained bythe oscillation of muon neutrinosinto electron neutrinos, with a ∆m2

of about 1–2 eV2. A result of no smallimportance, it cries out for indepen-dent confirmation. Enter the secondof Fermilab’s lissome leviathans, theBooster Neutrino Experiment.

BooNE’s main raison d’etre is totest the LSND result. It will eitherprovide confirming evidence (and getenough events to establish the νµ toνe transformation once and for all) orprove LSND wrong—and set morestringent limits on this particulartype of oscillation. There are somesurface similarities between LSNDand BooNE: both use large tanks filledwith mineral oil and lined withphotomultiplier tubes. But the detec-tor geometries are different (LSND iscylindrical, while BooNE is spheri-cal), the source-to-detector distanceL and beam energy E are different(although the ratio L/E for the two

Some of the 1200 photomultiplier tubeson the inside of the BooNE detector.

30 FALL 2001

The BooNE detector is a 12 mdiameter steel sphere filled with 770tons (about 250,000 gallons) of min-eral oil (see illustration on this page).A second light-tight inner walldivides the tank into an active cen-tral sphere and an outer “veto” shellused to flag exiting or extraneous in-coming particles. Charged particlesproduced by neutrino interactions inthe oil are detected by means of theblue Cerenkov light they generate,which will be detected by 1280 photo-multiplier tubes lining the inner sur-face of the central sphere or the 330tubes in the veto shell. Again, thecharged-current interaction willallow physicists to identify the kindof neutrino involved; a long muontrack (from the interaction of a νµin the mineral oil) will generate a pat-tern of Cerenkov light distinct fromthat produced by an electron shower(from a νe interaction).

As this article goes to press, con-struction of the BooNE detector isnearing completion, and the proton-extraction beam line and target-hornsystems are well underway. Datataking is expected to commence inearly 2002, after which the experi-ment should run for one to two years.If a signal is observed, a seconddetector can be built downstreamof the first—to make further detailedmeasurements of the parameters forνµ → νe oscillations. Such a resultwould add to the growing body ofevidence that neutrinos do indeedhave mass, and point to the excit-ing possibility of new physics beyondthe Standard Model.

NEUTRINO PHYSICShas long been a featuredperformer in Fermilab’s

experiments is the same), and thebackgrounds for BooNE will be dif-ferent from those of LSND.

The BooNE experiment will fol-low the standard recipe for making amuon neutrino beam, starting with8 GeV protons from the FermilabBooster, a low-energy synchrotron inthe Tevatron acceleration chain usedto boost protons from 400 MeV to 8GeV before passing them on to theMain Injector. Protons extractedfrom the Booster will strike a beryl-lium target; pions generated by thesecollisions will be focused by a mag-netic horn, and their decay will pro-duce the muon neutrinos speedingoff toward the detector.

Signal Region

Veto Region

Artist’s conception of the BooNEdetector, showing some of the photomul-tiplier tubes lining its 250,000-gallon tank.

BEAM LINE 31

multifaceted particle physics pro-gram. DONUT has made a major con-tribution, and the upcoming BooNEexperiment bids fair to continue thathistory. Future neutrino-beam op-portunities might include upgradingthe Booster to allow it to provideeven greater fluxes of neutrinos.There are also the exciting prospectsof “long-baseline” neutrino experi-ments where the distance betweenthe source and detector is hundredsof kilometers rather than hundredsof meters.

One such Bunyanesque experi-ment stretches all the way from Fer-milab to the Soudan mine in north-ern Minnesota, where the detectorfor the MINOS (Main Injector Neu-trino Oscillation Search) experimentwill be located. The MINOS beamstarts with 120 GeV protons from theMain Injector, which feeds the Fermi-lab Tevatron. The neutrino beam pro-duced by these protons hitting a tar-get will be directed north and slightlydownward, to intercept a six-kilotonsteel-and-scintillator detector in aniron mine some 750 km away—making MINOS one of the largestneutrino experiments on the planet.When this football field-length detec-tor is complete, it will capture neu-trinos beamed from Fermilab andfurther probe the mysteries of neu-trino oscillations. The combinationof neutrino energies and the longbaseline between source and detec-tor make MINOS sensitive to valuesof ∆m2 a hundred times smaller thanBooNE. This is the same range ofvalues that has been observed foroscillations of neutrinos in the up-per atmosphere (see article by JohnLearned in the Winter 1999 BeamLine, Vol. 29, No. 3).

Peering still further into thefuture, one can glimpse the in-triguing possibility of a “neutrinofactory”—a storage ring for muonswhose decay would provide a copi-ous source of muon and electron neu-trinos. Such a high-energy source ofelectron neutrinos would open wholenew vistas for further explorationof the weak interaction. But what-ever scenarios the future contains,neutrinos will surely play leadingroles in them.

Members of the MINOS collaboration be-fore one of its enormous steel plates.

For Further Information

DONUT home page:http://www-donut.fnal.gov

BooNE home page:http://www-boone.fnal.gov/

Juha Peltoniemi’s “ultimateneutrino” page:

http://cupp.oulu.fi/neutrino/

32 FALL 2001

NEUTRINOS ARE AMONG THE MOST ABUNDANTelementary particles in the Universe. They are a billion timesmore abundant than nucleons or electrons. If we would like to

understand the Universe, we must therefore understand neutrinos. Learn-ing about the world of neutrinos is not easy, however, because their inter-action with other matter is extremely feeble.

In the last few years, however, dramatic progress has been made. Anexperiment has now confirmed (see article by Paul Nienaber, this issue)that neutrinos come in three different varieties, or “flavors”: νe , νµ , andντ . These three neutrino flavors correspond to the three charged leptons:the electron e, its heavier cousin the muon µ, and its still heavier cousincalled the tau lepton τ. Neutrinos of different flavor are distinctly differentobjects, but it has been convincingly demonstrated that a neutrino canspontaneously change its flavor! This metamorphosis is known asneutrino flavor oscillation, or simply neutrino oscillation. The evidencethat neutrinos oscillate between flavors is particularly compelling in thecase of atmospheric neutrinos—neutrinos made in the Earth’s atmosphereby cosmic rays. In 2001, the evidence also became rather strong in the caseof solar neutrinos—neutrinos made in the Sun by nuclear processes (seearticle by Joshua Klein, this issue).

Any neutrino oscillation necessarily implies that neutrinos have mass.Thus, even though neutrino masses are extremely tiny, the observationthat neutrinos oscillate tells us that these masses are not zero. Theseneutrino masses could have very interesting consequences for bothparticle physics and cosmology.

How big are the neutrino masses? Unfortunately, oscillation experi-ments cannot tell us, because oscillation depends only on the differences—or “splittings”—between the squared masses of different neutrinos. Of course,

by BORIS KAYSER

The Enigmatic World of

Trying

to

discern

the

patterns

of

neutrino

masses

and

mixing.

BEAM LINE 33

Neutrinos

these splittings can teach us about the pattern of neutrino masses but not about the absolute scale of this pattern. To discover this scale, we will have to carry out very different kinds of experiments.

While finding experimental evidence for neutrino masses has been a challenge, many theoretical physicists have strongly suspected that these masses do indeed exist. Neutrinos are close relatives of the other con-stituents of matter—the quarks and the charged leptons—and interact withthem via the weak nuclear force. Moreover, the quarks, charged leptons, andneutrinos have very similar weak interactions. The very successful theory ofthe elementary particle interactions known as the Standard Model relatesthe quarks, charged leptons, and neutrinos to one another by placing theminto families. Each family contains two quarks, or else one neutrino and onecharged lepton, and the weak force can turn one member of any family intothe other. With these hints that neutrinos are related to the charged leptonsand quarks, and the observation that all of the charged leptons and all of thequarks have nonzero masses, one expects that the neutrinos probably havenonzero masses as well.

The masses of the charged leptons and quarks are incorporated, althoughnot explained, in the Standard Model—which may also be able to includeneutrino masses. But because the neutrino masses are very tiny, any suchinclusion would have to involve some extremely small numbers that are notterribly likely to be ingredients of a genuine physical explanation. Thus, thecomplete explanation of neutrino masses will almost certainly take us to newphysics beyond the Standard Model. Such physics is almost universallybelieved to exist because of the many questions this theory leaves open. How-ever, this new physics has proved remarkably elusive. By opening a windowon the new physics, neutrino masses could become very interesting indeed.

34 FALL 2001

Furthermore, when a neutrinointeracts with matter and gives birthto a charged lepton, that leptonalways has the same flavor as theneutrino. A νe always begets an e, aνµ a µ, and a ντ a τ. This is how weknow that νe, νµ and ντ must be dis-tinctly different objects.

THE MOST convincing sin-gle piece of evidence for neu-trino oscillation in Nature

comes from the behavior of theatmospheric neutrinos. These neu-trinos have been detected by severalunderground detectors and studiedmost extensively by the Super-Kamiokande detector in a zinc minein Japan. Atmospheric neutrinosstudied by an underground detectorcan originate in the atmospheredirectly above the detector, on theopposite side of the Earth, or aboveany other point on Earth (see illus-tration on the left). By detectingmuons created in the Super-Kamiokande detector by incomingatmospheric muon neutrinos, theSuper-Kamiokande collaboration hasdetermined how the flux of themdepends on the direction from whichthey are coming. The upward-goingflux, coming up from all directionsbelow the horizon at the location ofthe detector, is only about half thecorresponding downward-going flux.Of course, the upward-going neutri-nos originated in distant parts of theatmosphere, while the downward-going ones started much closer andtraveled a much smaller distance.

The flux of cosmic rays energeticenough to produce atmospheric neu-trinos with energies above a few bil-lion electron volts is known to beisotropic. The atmospheric muon

Owing to their great abundance,the neutrinos might well play a sig-nificant role in shaping the evolutionand destiny of the Universe (seearticle by Joel Primack, this issue).Exactly what role they play dependson their masses, and on how bigthese masses are.

How is it possible for neutrinos tooscillate between different flavors,and why does this phenomenon im-ply that neutrinos have mass? Tohelp explain the physics of neutrinooscillation (see box on next page), letme first sharpen the concept of neu-trino flavor. By the neutrino ν of fla-vor (= e, µ, or τ), physicists meanthe neutrino produced together witha particular charged lepton in anyprocess where there was no neutrinoor charged lepton to begin with; oneneutrino plus one charged lepton arecreated. For example, the νµ is theneutrino produced together with amuon in the pion decay π+ → µ+ + νµ.

Earth

Atmosphere

Neutrino Made

in the Atmosphere

D

Atmospheric neutrinos incident on anunderground detector (labeled D). Theneutrinos originate from all around theworld. Upward-going neutrinos (coloredlines) have much more time to oscillatethan downward-going ones (black lines).

BEAM LINE 35

neutrinos in this energy range aretherefore being produced at the samerate everywhere around the Earth.Now one can easily formulate aneutrino-flux version of the Gauss’sLaw familiar to any student of elec-tricity and magnetism. From this lawand the equality of neutrino produc-tion rates around the Earth, it followsthat as long as atmospheric muonneutrinos neither appear nor disap-pear within the Earth, the upwardand downward νµ fluxes studied bySuper-Kamiokande should be equal.Thus, independent of any theoreti-cal assumptions, the observed in-equality of these two fluxes impliesthat something is adding or takingaway muon neutrinos within theEarth. The dominant candidate forthe culprit is neutrino oscillation.Since the upward neutrinos comefrom more distant parts of the at-mosphere than the downward onesand consequently have more time tooscillate, the upward muon neutri-nos can oscillate away into neutri-nos of another flavor, while thedownward ones simply do not get thechance. This mechanism wouldexplain why fewer muon neutrinosare observed going upward thandownward. From upper limits on theoscillation of electron neutrinos gen-erated by nuclear reactors, we knowthat the oscillating atmospheric νµ

are not (or at least not appreciably)turning into νe. Thus they must beturning into ντ—or perhaps into somenew kind of neutrino not seen before.High-precision experiments at CERNand SLAC have taught us that onlythree species of neutrinos—νe, νµ andντ—interact with matter through theweak nuclear force. Consequently,any new, fourth kind of neutrino

Neutrino Mass and Mixing

IF NEUTRINOS HAVE MASSES, then the neutrino, ν , producedtogether with a specific charged lepton, , need not be a particle with adefinite mass. Instead, it can be a quantum-mechanical superposition ofsuch particles. This perplexing sleight-of-hand is called neutrino mixing.Let us call the neutrinos that do have definite masses—that is, theneutrino “mass eigenstates”—ν1, ν2 and ν3, and so on. Then ν can be asuperposition of these eigenstates. Quantum mechanically, the ν at itsmoment of birth is described by the wave labeled “At Birth” in the figurebelow, where we have made the simplifying assumption that ν has onlytwo significant components of definite mass, ν1 and ν2. Each of thesemass-eigenstate components is represented by a traveling wave. Whenν is born, the peaks of the waves coincide, and the two waves then addup to a resultant wave with theproperties of ν . If ν1 and ν2 havedifferent masses M1 and M2 andour neutrino has a certain averagemomentum, then its ν1 and ν2pieces will travel at differentspeeds. Consequently, after theneutrino has traveled awhile, thepeak of the wave corresponding to the lighter mass-eigenstatecomponent (ν1 say) will haveforged ahead of the peak of thewave corresponding to the heavier component.

Now, at any given time, thewave representing our neutrino isthe sum (obtained using the usual rules for adding waves characterizedby both amplitudes and phases) of the ν1 and ν2 contributions. Clearly,this sum is not the same “Later” as it was “At Birth.” After the neutrinohas traveled for awhile, it therefore cannot be a pure ν anymore. If onewere to analyze this neutrino in terms of its flavor content, one wouldquickly find that it has developed components bearing flavors other thanits original flavor. Thus, if the neutrino interacts in a detector and pro-duces a charged lepton, there is some probability that this charged lep-ton is of another flavor. For example, our neutrino can be born togetherwith a muon, so that it starts life as a νµ, but after developing compo-nents with other flavors, it can interact to produce a τ. We describe thissequence of events by saying that our νµ had turned into a ντ. This ishow a neutrino can spontaneously change flavor.

The possibility of this flavor change depends intrinsically on havingneutrino mass eigenstates with unequal masses, which in turn requiresthat at least one of these masses be nonzero. Thus, neutrino oscillationimplies neutrino mass. A calculation shows that when only two neutrinomass eigenstates ν1 and ν2 play a significant role, the probability for aneutrino to change flavor is proportional to sin2[(δM 2

21/4)(L/E)]. Here,δM 2

21 = M 22 − M 2

1 is the splitting between the squared masses of the con-tributing mass eigenstates, L the distance traveled by the neutrino be-tween its birth and detection, and E its energy. Note that the probabilityfor neutrino oscillation really does oscillate, as the name implies. Thisprobability depends only on the splitting between the neutrino squaredmasses, and not on the individual masses. This remains true when morethan two mass eigenstates (hence more than one splitting) are involved.

ν2

ν1

ν2

ν1

At Birth Later

A propagating neutrino wave consistingof ν1 and ν2 pieces. The vertical linepasses through the peak of the ν2 wave.When the neutrino is born (At Birth), thisline coincides with the peak of the ν1wave, but at a later time (Later) it doesnot.

36 FALL 2001

from which the neutrinos are com-ing—extremely well. The Super-Kamiokande results are supported bythose of the MACRO experiment inthe Gran Sasso underground labo-ratory in Italy, and by those of theSoudan-2 experiment in the Soudanmine in Minnesota. The neutrinosquared-mass splitting δM2 requiredby the data is approximately 0.003eV2. This value allows neutrinoswith typical atmospheric neutrinoenergies to oscillate significantly ifthey travel upward to the detectorfrom the other side of the Earth, butnot if they travel downward from justabove.

The neutrino mixing required bythe atmospheric oscillation data isquite large. That is, a νµ is a super-position of mass eigenstates in whichno single one of them dominates.Rather, the νµ is an almost 50–50mixture of two mass eigenstates.And while it is still possible thatsome of the atmospheric νµ are os-cillating into sterile neutrinos, Super-Kamiokande analyses tell us that nomore than 25 percent of them aredoing so. Oscillation into tau neu-trinos is the favored path.

Compelling as the evidence foratmospheric muon neutrino oscilla-tion is, one would like to confirmthis observation by showing thatmuon neutrinos produced by a par-ticle accelerator also undergo thesame oscillation. Several experi-ments aim to do just that, and one ofthem, K2K, has already reported pre-liminary results (see article byKoichiro Nishikawa and JeffreyWilkes, this issue). These results arenot yet conclusive, but they do seemto indicate that muon neutrinos areoscillating into something else.

would be even more of a ghost, sinceit would not interact with matterthrough any known force save grav-ity. Such a neutrino is called a “ster-ile” neutrino, νs.

The hypothesis that the atmos-pheric muon neutrinos are oscillat-ing into tau neutrinos fits a host ofdata—including the detailed depen-dence of the νµ flux on the direction

The Super-Kamiokande detector duringfilling in 1996. Physicists in a rubber raftare polishing the 20-inch photomultipli-ers as the water slowly rises. (CourtesyICRR, University of Tokyo)

AN INDEPENDENTstrongpiece of evidence for oscil-lation is the behavior of so-

lar neutrinos, which are generated inhuge numbers by the nuclearreactions that power the Sun. Atbirth, all of these neutrinos are elec-tron neutrinos, since they are cre-ated in association with electrons (or,more precisely, positrons). Solar neu-trinos have been detected here onEarth by several underground detec-tors, all of which are sensitivelargely—or exclusively—to electronneutrinos. Interestingly, all thesedetectors report solar neutrino fluxesthat are only 30 to 60 percent of thoseexpected from the theoretical calcu-lations of neutrino production in theSun. Valiant attempts to modify thetheory of the Sun, without invok-ing neutrino mass or oscillation, havebeen notably unsuccessful in explain-ing this deficit. Hence, it has beenhypothesized that the deficit iscaused by the oscillation of electronneutrinos into another, hard-to-seeflavor. Perhaps this oscillation isoccurring in outer space between theSun and the Earth. Or perhaps amatter-enhanced neutrino oscilla-tion, known as the Mikheyev-Smirnov-Wolfenstein (MSW) effectafter the theorists who first exploredits physics, is occurring within theSun itself as the neutrinos producedby the solar core stream outward. Ei-ther way (if we neglect the possibil-ity of sterile neutrinos), νµ or ντ (orboth) are eventually produced. Toconfirm that the solar neutrinos areundergoing oscillation, one wouldtherefore like to verify that νµ or ντ ,which are not being forged in the so-lar furnace, are nevertheless arrivinghere on Earth from the Sun.

This past June, the Sudbury Neu-trino Observatory (SNO) reported anew result that, when compared toan earlier Super-Kamiokande result,implies that νµ or ντ are indeed ar-riving here from the Sun! The SNOcollaboration has succeeded in mea-suring the solar νe flux arriving atEarth, independently of whether anyother flavors of neutrinos might be

BEAM LINE 37

The heart of the solar neutrino detectoroperated by Brookhaven/University ofPennsylvania chemist Raymond Davisand his collaborators. The vessel con-tains about 615 tons of perchlorethyleneand is located almost a mile under-ground in the Homestake Gold Mine inLead, South Dakota. (CourtesyBrookhaven National Laboratory)

38 FALL 2001

smaller than what is called for by theatmospheric-neutrino oscillation.This means that the solar oscillationdoes not involve the same two neu-trino mass eigenstates. To a certainextent, the solar-neutrino data sug-gest that the MSW effect is indeedinvolved, that δM2 falls in the range10−5 eV2 to 10−4 eV2, and that the neu-trino mixing that is playing a roleis large—like its analogue in theatmospheric-neutrino oscillation.

The MSW effect, like neutrinooscillation in empty space, requiresneutrino mass. Thus, whether or notthe solar νe are turning into νµ or ντthrough the MSW effect, the obser-vation that they are also undergo-ing a metamorphosis is another pow-erful piece of evidence that neutrinosindeed have masses.

THERE IS A THIRD pieceof evidence that neutrinososcillate, but it is uncon-

firmed. This evidence comes fromthe Liquid Scintillator NeutrinoDetector (LSND) experiment at LosAlamos, which has recorded theappearance of electron antineutrinosin a neutrino beam produced via thedecays of positively charged muons.As these decays do not yield electronantineutrinos directly, their appear-ance has been interpreted as result-ing from the oscillation of the muonantineutrinos that µ+ decays docreate.

The squared-mass splitting de-manded by the reported LSND oscil-lation is around 1 eV2, much largerthan what is called for by theatmospheric and solar oscillations,which gives rise to an interesting sit-uation. Suppose there are only threedifferent flavors of neutrino: νe , νµ ,

arriving as well (see article by JoshuaKlein, this issue). In contrast, the ear-lier Super-Kamiokande result (fromthe same detector that studied theatmospheric neutrinos) measuredthis same νe flux plus an admixtureof one-sixth of the νµ and ντ fluxesarriving from the direction of theSun. By subtracting from this totalthe νe flux measured by SNO, physi-cists can extract the νµ+ ντ flux. Theyfind that, at a strong—albeit not yetcompletely conclusive—level of con-fidence, the νµ + ντ flux arriving atEarth from the Sun is not zero. Thisis an important result.

The neutrino squared-mass split-ting δM2 required by the solar-neu-trino data falls somewhere between10−11 eV2 and 10−4 eV2, depending onwhether or not the MSW effect isinvolved. Either way, the splitting is

The LSND experiment was designed tosearch for the presence of electron anti-neutrinos. Over 1200 photomultipliertubes line the inner surface of themineral oil tank, shown above with LSNDphysicist Richard Bolton of Los Alamos.(Courtesy Los Alamos NationalLaboratory)

BEAM LINE 39

and ντ. Then there are only three cor-responding neutrino mass eigen-states: ν1, ν2, and ν3. To accommo-date the solar and atmosphericoscillations, the neutrino squared-mass spectrum must be as picturedin the illustration on the right. Theremust be a closely-spaced pair of neu-trinos we shall call ν1 and ν2, whosesplitting δM2

solaris the small onecalled for by the solar data. This pairmust be separated from the thirdeigenstate, ν3, by the bigger splittingδM2

atm required by the atmosphericdata. Since both δM2

solar and δM2atm

are much smaller than 1 eV2, thisthree-neutrino spectrum cannotinclude any pair of neutrinos whosesplitting is the large 1 eV2 demandedby LSND. No three-neutrino spec-trum that fits both the solar andatmospheric oscillations can also fitthe LSND oscillation. If the LSNDoscillation is confirmed by futureexperiments, then Nature must con-tain at least four different kinds ofneutrinos. As Nature contains onlythree kinds of neutrinos that inter-act with matter via the weak nuclearforce, any additional kinds wouldhave to be of the almost totally non-interacting sterile variety. Thus, it isvery interesting, to say the least, tofind out whether the oscillation re-ported by LSND is genuine.

NOW THAT physicistshave ascertained that neu-trinos almost certainly

have nonzero masses, there are somemajor questions we would like to seeanswered:

• How many different neutrinoflavors are there in all? Just νe, νµ andντ? These three plus one sterile fla-vor? These three plus many sterile

flavors? Equivalently, how many dif-ferent neutrinos of definite mass—that is, mass eigenstates—are there?

• What are the masses of theseeigenstates? From neutrino oscilla-tion experiments, we already knowsomething about the squared-masssplittings between the differenteigenstates, but what are their indi-vidual masses?

• Is each neutrino mass eigen-state identical to its antiparticle, asis a photon, or distinct from it likean electron?

Alone among all the spin-1/2 con-stituents of matter—the quarks andleptons—neutrinos may be their ownantiparticles. No quark or chargedlepton can be identical to its anti-particle because it is electricallycharged, so its antiparticle must beoppositely charged. However, a neutrallepton—that is to say, a neutrino—is electrically neutral and as far as weknow does not carry any othercharge-like attribute either. Thus, aneutrino could well be identical toits antiparticle.

An important question is whetherthe properties of neutrinos are relatedin a deep way to the fact that weexist. That existence depends on thefact that the Universe containsnucleons and electrons—the parti-cles of which we are made—but veryfew of the corresponding antiparti-cles. That is, the Universe currentlycontains far more matter than anti-matter. Since the Big Bang presum-ably produced equal amounts of mat-ter and antimatter, the presentasymmetry must have arisen afterthe Big Bang. But for this to haveoccurred, matter particles and theircorresponding antiparticles mustbehave differently, at least under

δM2atm(Mass)2

δM2solar

ν3

ν2ν1

A three-neutrino (Mass)2 spectrum thatcan accommodate both the solar andatmospheric neutrino oscillations. Anequally possible spectrum places theclosely-spaced pair of neutrinos at thetop rather than at the bottom.

40 FALL 2001

experiment known as KARMEN,carried out at the Rutherford Labo-ratory in England, does not see anysuch oscillation. However, the twoexperiments have different sensitiv-ities and are not necessarily in con-flict. To find out whether or not theLSND oscillation signal is genuine,an experiment known as BooNE,presently under construction atFermilab, will be carried out (seearticle by Paul Nienaber, this issue).

The splittings δM2 among the dif-ferent neutrino eigenstates will bedetermined through a broad programof neutrino-oscillation studies.Together, these splittings will tell usthe M2 spectrum, but not its base-line. Determining this baseline—thatis, the absolute scale of neutrinomass—will not be easy. One ap-proach will be to study very care-fully the high-energy end of the elec-tron energy spectrum in tritium betadecays: 3H → 3He + e− + νm. Here, 3His the nucleus of a tritium atom (oneproton plus two neutrons), 3He is thenucleus of a helium-3 atom (two pro-tons plus one neutron), and νm(m = 1,2,3. . .), is any one of the neu-trino mass eigenstates. Through thekinematics of these decays, themasses of the neutrinos νm will bereflected in the energy spectrum ofthe electron. However, this effectwill be experimentally invisibleunless at least one νm has a mass bigenough to be within range of themass sensitivity of the experiment,and this particular νm is emitted in asubstantial fraction of all the tritiumdecays.

A tritium experiment sensitive toneutrino masses as small as 0.3 eV isnow in the planning stages. Knownas KATRIN, it will be performed in

The most popular explanation ofthe fact that the neutrinos are somuch lighter than the charged lep-tons and quarks is called the “see-saw mechanism.” This mechanismcan apply only to the neutrinos—itcannot be the physics behind themasses of quarks or charged leptons.The see-saw mechanism requiresthat the neutrinos are their own anti-particles. It predicts that a typicalneutrino will have a mass of orderM2

quark/MBig. Here Mquark is a typicalquark (or charged lepton) mass, andMBig is a very large mass far beyondthe reach of present-day accelerators.The new physics responsible for theneutrino masses and other phenom-ena supposedly resides at the high,presently-inaccessible mass scale rep-resented by MBig. Obviously, the big-ger MBig is, the smaller the neutrinomass M2

quark/MBig will be. Hence, thename “see-saw mechanism.”

HOW CAN WE answer allthese neutrino questions?The evidence that there are

more than three neutrinos is theneutrino-oscillation signal observedby the LSND experiment. A similar

certain conditions. Otherwise, amatter-antimatter symmetrical Uni-verse would have remained that way.One possibility is that there was atime in the early Universe whenelectrons were produced morecopiously than positrons by thedecays of some very heavy parentparticles. Subsequently, the electronsparticipated in processes that creatednucleons but not antinucleons.

If indeed there was an early epochin which electrons were producedmore abundantly than positrons,then neutrinos may exhibit relatedbehavior. For example, it may be thatthe probability for an antineutrino tooscillate is different from that for thecorresponding neutrino. If so, thiswould be a clear difference betweenthe behavior of antimatter and thatof matter. Any such difference wouldbe extremely interesting to observe.

Essentially this same differencecan still occur even if a neutrino andits antineutrino are the same particle.If so, an interacting neutrino whosespin is antiparallel to its momentumwill make a charged lepton (e−, µ−, orτ −), while one whose spin is parallelto its momentum will make anantilepton (e+, µ+, or τ+). The oscillationpatterns of the lepton- and antilepton-producing neutrinos can differ. If theydo, we will once again have a distinc-tion between the behavior of matterand that of antimatter.

The biggest question of all aboutthe neutrinos is: What is the under-lying physics responsible for the neu-trino masses, mixings, and otherproperties? And is this physics alsoresponsible for the masses of thecharged leptons and quarks, or can itonly be glimpsed by studying theneutrinos?

Thanks to the

dramatic discoveries

of the past few years,

we now know that

there is a rich world

of massive neutrinos

waiting to be explored,

but we have only begun

to explore it.

BEAM LINE 41

Karlsruhe, Germany. If the LSNDneutrino oscillation is genuine, thenthe LSND data tell us that there aretwo definite-mass neutrinos, νm andνn, separated from one another byδM2

mn = M2m – M2

n > 0.2 eV2. It followsthat Mm, the mass of νm, is nosmaller than 0.4 eV, which is withinrange of the experiment.

It is amusing that the search forthe absolute scale of neutrino massbrings us back to take a very detailedlook at the energy spectra in nuclearbeta decay. It was the observationthat these spectra are continuous,rather than discrete, that originallyled Pauli to propose the existenceof neutrinos!

To test whether the electron neu-trino is identical to its antiparticle,we can search for neutrinoless dou-ble beta-decay—in which one nu-cleus decays into another with theemission of two electrons, and noth-ing else. Since there are no electronsbefore the decay, and two electronsafter it, such a process clearly doesnot conserve the hypothetical,charge-like “lepton number” thatwould distinguish charged and neu-tral leptons like e− and νe on the onehand from charged and neutral an-tileptons like e+ and ν−e on the otherhand. Indeed, the occurrence of thisdecay would indicate that ν−e = νe,so that there is no conserved leptonnumber to distinguish leptons fromtheir antiparticles. Several groupsaround the world are hoping to carryout sensitive searches for this po-tentially very informative decay. Inprinciple, these searches could yielda bonus: The rate for neutrinolessdouble beta-decay depends, albeit ina nontrivial way, on the absolutescale of neutrino mass. Thus the

observation of this process wouldyield information about this scale.The proposed searches should be sen-sitive to a mass scale an order ofmagnitude smaller than the plannedtritium experiment can reach.

To probe the possible connectionbetween neutrinos and our existence,we will need very intense neutrinobeams produced by accelerators.These beams will probably have tobe “bright” enough to remain rela-tively intense even after travelingseveral thousand kilometers to a dis-tant detector. Then it will be possi-ble to use them to see whether neu-trino oscillations depend on whetherthe neutrinos in the beam are of thekind that produce particles of mat-ter or of the kind that produce par-ticles of antimatter.

AS FOR the big question—what’s behind it all—whocan say how we will answer

that? The answer may include in-gredients presently being considered,and it may well include physics thatnobody has yet dreamed of.

Thanks to the dramatic discover-ies of the past few years, we now knowthat there is a rich world of massiveneutrinos waiting to be explored, butwe have only begun to explore it.Major, fundamental questions aboutthe neutrinos need to be answered. Toattempt to answer them, a diverseexperimental program will occur inthe coming years. This program willinclude experiments on accelerator-generated neutrinos that travel to bothnear and far detectors, and others onneutrinos produced by tritium decay,nuclear reactions in power plants, cos-mic rays, the Sun, and perhaps super-nova explosions. It will also entail

searches for neutrinoless double betadecay and astrophysical studies of theneutrinos that, eons ago, helped shapethe Universe that we see today. If therecent past is any guide, this future ex-ploration of the enigmatic world ofneutrinos will be very exciting indeed.

42 FALL 2001

THREE YEARS AGO, it was already evident thatour standard picture of neutrinos needed work.Unexpected shortfalls in the fluxes of solar and

atmospheric neutrinos were not about to go away; thesedeficits had primed many physicists to accept the clear evi-dence for neutrino mass that the Super-Kamiokande Collabo-ration presented in 1998. But the Super-Kamiokande resultsindicated only that at least one kind of neutrino indeed hasmass. Many other questions remain unanswered. What is thecomplete pattern of neutrino masses? How are the threeknown kinds of neutrinos connected to the mass eigen-states? Do we really need a fourth neutrino, which wouldhave to be “sterile”? And how does the new information fitwithin the framework of the Standard Model, which hasworked so well for us for so long?

The next steps are to confirm and extend the Super-Kamiokande results and then try to map out, at least in out-line, the pattern of neutrino masses. Groups in the UnitedStates, Europe and Japan are now working to develop theneutrino beams and detectors needed for this next generationof neutrino experiments, in which beams of well-knowncomposition are directed over a distance of hundreds of kilo-meters from an accelerator to a remote neutrino detector.One group, however, has a big head start on the others. Since1999, the K2K (KEK to Kamioka) experiment, the world’s firstlong-baseline neutrino experiment to begin operations, hasslowly but surely been logging data.

The K2K Neutrino Eby KOICHIRO NISHIKAWA & JEFFREY WILKES

The world’s first

long-baseline

neutrino experiment

is beginning

to produce results.

BEAM LINE 43

Experiment

PREHISTORY OF THE EXPERIMENT

In the early 1980s, underground water Cerenkov detectors were built in Japanand the United States by Masatoshi Koshiba, Frederick Reines, and others tosearch for nucleon decay. A few thousand tons of water contained the requi-site nucleons, and the water itself also served as the track-sensitive medium,generating Cerenkov light from relativistic particles. To reduce cosmic raybackgrounds to manageable levels, the detectors had to be located deepwithin the earth in underground mines. Kamiokande (Kamioka NucleonDecay Experiment), built in the Kamioka-Mozumi mine in the JapaneseAlps, was essentially a 10 m diameter by 10 m tall cylindrical tank of ultra-pure water, viewed by hundreds of large photomultiplier tubes (PMTs). TheIMB (Irvine-Michigan-Brookhaven) experiment, built in the Morton Saltmine under Lake Erie, near Cleveland, had a similar arrangement, with acubic tank of water about 10 m on a side. Neither experiment found convinc-ing evidence for nucleon decay, but they were fortunately sensitive toneutrinos from extraterrestrial sources.

Thus nucleon-decay seekers became astrophysical neutrino observers, andfound plenty to write home about. Previous solar neutrino observations, con-ducted by Ray Davis and collaborators since 1967 at the Homestake goldmine in South Dakota, had reported a significant deficit compared totheoretical predictions. The Kamiokande and IMB experiments confirmedthis result and added a new puzzle to the picture. While the Homestakeexperiment was insensitive to high-energy neutrinos, the two waterCerenkov detectors could easily observe neutrinos created in the upperatmosphere by cosmic rays. But the experimental observations of these“atmospheric” neutrinos fell well short of theoretical expectations, addinganother deficit to the growing list of neutrino puzzles.

44 FALL 2001

in April 1996, two separate analysisgroups began to reduce the data andprovide independent checks uponone another. One group took as itsstarting point the analysis methodsand software from IMB, and the otherfrom Kamiokande. Such careful pro-cedures helped ensure the promptacceptance of Super-Kamiokanderesults by the physics community.

By June 1998, collaboration mem-bers had enough data and confidencein their analyses to present a trulyhistoric result at the Neutrino-98Conference, held in nearby Taka-yama, Japan: the first convincingevidence for neutrino oscillations,which emerged from their atmos-pheric neutrino data (see JohnLearned’s article in the Winter 1999issue of Beam Line, Vol. 29, No. 3).Distributions of neutrino flux versusincidence angle showed a clear deficit(relative to expected fluxes) for up-ward-going muon neutrinos comingfrom the other side of the Earth,while downward-going muon neu-trinos (and all electron neutrino dis-tributions) agreed with theoreticalpredictions. The difference could notbe explained by the fact that upward-going neutrinos had to pass througha lot more matter, since the Earthis essentially transparent to theseneutrinos. But the deficit could beexplained by neutrino oscillations,since the upward-going neutrinosmust have trekked 10,000 km or sofrom their point of birth, while down-ward going neutrinos traveled only15 or 20 km. In fact, the hypothesisthat particles produced as muon neu-trinos oscillate into tau neutrinos fitsthe data very well, since the Super-Kamiokande detector cannot effi-ciently identify tau neutrinos—

It became clear that a new, muchlarger detector could be of great valuein solving both the solar andatmospheric neutrino puzzles. ThusYoji Totsuka and collaborators beganplanning the next-generation detec-tor called Super-Kamiokande, even asthe original experiments continuedcollecting data. Super-Kamiokandewas designed on a truly gargantuanscale, with a 40 m diameter, 40 m tallcylindrical tank holding 50,000 tonsof water that is monitored by over11,000 giant phototubes, each 50 cmin diameter (see photo on page 36 anddrawing below). To construct andoperate such a huge detector requiredpooling resources, so members ofboth the Kamiokande and IMB groupsjoined forces.

The two groups each brought theirown independent ideas and ap-proaches to the experiment. WhenSuper-Kamiokande began taking data

Artist’s conception of the Super-Kamiokande detector.

BEAM LINE 45

leading to an apparent deficit ofmuon neutrinos. Such oscillationscan occur only if at least one of theseneutrinos has mass.

HISTORY OF K2K

The idea of directing a neutrino beamtoward a distant detector had longbeen discussed by Koshiba, Reines,Al Mann, and others. Until recently,however, the high cost of construct-ing a far detector as well as a new,downward-directed beam line at anysuitable accelerator was prohibitive.But Super-Kamiokande could serveas a perfect detector for a neutrinobeam from KEK, the Japanese na-tional particle-physics laboratory atTsukuba, 150 miles from Tokyo. The12 GeV proton synchrotron at KEKcould produce a neutrino beam withthe energy range required to test theoscillation parameters suggested bythe early Kamiokande results; a rel-atively modest 1 degree downwardangle was required for this beam.And many components of past ex-periments could be recycled to builda near detector on the KEK site, tosample the outgoing neutrino beam.In particular, a 1000-ton water tankwas already available, as well as amuon detector system (iron platesand proportional tubes) and a lead-glass detector. The only major equip-ment needed was a scintillating-fibertracking detector plus new photo-multiplier tubes for the water tank.The KEK synchrotron was steadilyupgraded, and a significant increasein its luminosity was achieved. Bend-ing magnets on loan from SLAC wereused to bend the primary protonbeam about 90 degrees into the orien-tation needed to direct the neutrino

beam toward Super-Kamiokande (seephoto above).

With remarkable rapidity, plansfor the K2K experiment were ap-proved in 1995. It was organized asa scientific collaboration distinctfrom Super-Kamiokande, althoughthere is substantial overlap in mem-bership. Civil construction of theneutrino beam line and experimentalhall began in 1996, with an extremelytight project timeline. This wasespecially true when KEK was aprime contender to be the future siteof the Japan Hadron Facility (nowofficially named the High IntensityProton Accelerator Facility), whichwould have required upgrading theproton synchrotron from 12 to 50GeV and left K2K only two years ofrunning time. The decision to sitethe facility at Tokai relieved muchof this pressure—but only after theconstruction and commissioningphases were completed.

ProtonBeam

TargetHall

DecayPipe

NeutrinoHall

Aerial view of the KEK neutrino beamline. The extracted 12-GeV proton beamenters at left, goes through a 90-degreebend and then enters the target hall.Neutrinos from the 200 m decay pipepass through the neutrino hall at rightbefore heading toward Super-Kamiokande.

46 FALL 2001

designed. Thus K2K may take longerto reach its experimental goals. Theshakedown process was completedby April 1999, and data-taking runsbegan in earnest. Since then, K2K hasbeen operating for about six monthsa year, as the accelerator is shut downduring the summer months andother experiments take beam in theautumn.

THE K2K BEAM LINE

K2K uses a fairly conventional wide-band neutrino beam design. Theprimary beam is composed of 12 GeV(kinetic energy) protons from the KEKproton synchrotron. Every 2.2 sec-onds, approximately 6 trillion pro-tons are extracted and focused ontoa 66 cm long aluminum target. Adouble-horn magnet system focusespositive pions produced by p-Alcollisions, while sweeping negativesecondary particles out of the beamline. Between the horn system andthe 200 m long decay volume (wherethe π+ decay to νµ and µ+) the pionmonitor—a gas-Cerenkov detector ofnovel design—is periodically insert-ed in the beam to measure the pionkinematics before they decay. Withthese kinematic distributionsknown, it is possible to predict theneutrino spectrum at any distancefrom the source.

Following the decay volume, aniron-and-concrete beam dump filtersout essentially all charged particlesexcept muons with energy greaterthan 5.5 GeV. Downstream of thedump there is a muon monitor sys-tem, which measures the residualmuons to check beam centering andmuon yield. Steering and monitoringof the neutrino beam are carried out

Civil construction was rapidlycompleted, and experimental areasbecame available to scientists by thesummer of 1998. All beam line andtarget area components were quicklyinstalled and tested. Meanwhile, theexperimenters took a few months toinstall the detector elements in thenew Neutrino Hall, a cylindrical pit24 m in diameter located about 300 mdownstream of the production tar-get. Filling the tank and water puri-fication began in December 1998,with commissioning of other detec-tor elements completed during thenext few months.

Tuning of the neutrino beambegan in early 1999, followed by aseries of engineering runs to test thebeam and near detectors together.The 20 millimeter diameter alu-minum target used to produce neu-trinos failed mechanically under thesevere thermal stresses and wasreplaced by a 30 millimeter target,which meant that the neutrino in-tensity is a bit lower than originally

MuonChamber

LeadGlass Array

SciFi Tracker

1 kt Water CerenkovDetector

Neutrino BeamTo Super-Kamiokande

p

µ

Fine-GrainedDetector

Schematic view of the K2K neardetectors, showing (right to left) the1 kiloton water Cerenkov detector, theSciFi tracker, lead-glass array, and themuon detector.

BEAM LINE 47

based on data from this system. Theνµ and µ+ in the decay volume orig-inate from the same pion decays, sothe measured central position of themuon profile is well correlated withthe neutrino beam direction. Thebeam line was aligned by an engi-neering survey using the GlobalPositioning System (GPS) to deter-mine the line from the target to theSuper Kamiokande far detector tobetter than 0.01 mrad; the construc-tion precision for the beam align-ment is better than 0.1 mrad, whichis much better than required.

THE NEAR DETECTOR

The near detector array (see illust-tration on opposite page and photo-graph at right) is located in an under-ground experimental hall about 300m from the pion production target,after an earth berm about 70 m thickthat eliminates virtually all promptbeam products other than neutrinos.The one kiloton (1 kt) waterCerenkov detector uses the sametechniques and analysis algorithmsas the far detector, with 680 20-inchdiameter hemispherical PMTs lininga cylindrical volume 8.6 m in diam-eter and 8.6 m high. Because thisdetector has the same target mater-ial, PMTs, and basic geometry asSuper-Kamiokande and uses thesame event identification and analy-sis methods, many potential biasesand systematic errors cancel betweenthe two detectors. Thus the 1 ktdetector is the primary source of dataon neutrino event rates at the nearsite. A beam simulation based onthese data is then used to estimatethe expected νµ signal at the fardetector. Other components of the

near detector provide independentchecks on the event rates, as wellas supplementary data on inter-actions and rates.

Following the 1 kt detector, thescintillating-fiber detector (SciFi)

View from ground level into the NeutrinoHall at KEK, showing near detectorcomponents: 1 kiloton detector (at topin this view); SciFi tracker; lead-glassarray; and muon detector (bottom).

48 FALL 2001

allowing for the 250 km flight path.Duplicate GPS clock systems areused at the near and far sites to gen-erate time data with 20 nanosecondprecision. Beam-induced neutrinointeractions at Super-Kamiokandeare selected by comparing these twoindependent time measurements,TKEK for the starting time of the KEKbeam spill, and TSK for the Super-Kamiokande event-trigger times. Thetime difference ∆T = TSK − TKEK −TOF, where TOF is the time of flightat the speed of light between the nearand far detectors, should be distrib-uted within an interval of 1.1 µsec,matching the duration of the pro-ton beam spill at KEK. Using a con-servative 200 nanosecond estimateof the synchronization accuracybetween the two sites, we establisha 1.5 µsec long time interval insidewhich candidate neutrino interac-tions must occur. All neutrino eventsoccurring within the time windoware carefully examined. Cuts andanalysis procedures similar to thoseused in atmospheric neutrino analy-ses in Super-Kamiokande are appliedto select events that are fully con-tained within the same 22.5 kilotonfiducial volume. The fiducial-volumeand timing cuts have a powerfuleffect in separating beam-inducedevents from cosmic-ray and otherbackground events (see graph on thispage).

Up until April 2001, 44 fully-contained beam-induced neutrinoevents were observed in Super-Kamiokande. This number should becompared with the number of eventsexpected in the absence of any neu-trino oscillations, which is 60–68events based on the number of neu-trino interactions observed in the

provides detailed tracking capability,and allows identification of differenttypes of interactions such as quasi-elastic or inelastic scattering. It con-sists of 19 planes of water targettanks, interleaved with scintillatingfiber modules. Downstream of theSciFi detector is the lead-glass array,for tagging electromagnetic showers.The muon detector then measuresthe energy, angle, and productionpoint of muons from charged-currentneutrino interactions. The charac-teristics and stability of the νµ beamare directly monitored at the near siteby this detector, whose large trans-verse area and mass make it possibleto measure the beam direction, sta-bility, and intensity profile over a rea-sonable angular range. The neutrinobeam direction has been stablewithin ±1 mrad throughout all data-taking periods.

PRESENT RESULTS

The basic strategy in a long base-line experiment is to sample thebeam at a near detector, thus deter-mining its characteristics before anyoscillation effects can occur. Thenone uses these observations to deter-mine the expected beam characteris-tics at the far detector in the absenceof oscillations. We use a beam sim-ulation to make this extrapolation;it is validated by comparing its pre-dictions with the pion monitor mea-surements in the beam line, beforethe decay volume. The agreement be-tween the simulation and the pionmonitor data has been excellent.

Candidate events observed inSuper-Kamiokande must be syn-chronized with the arrival of protonsat the production target at KEK,

103

101

–500 –250 0

00

4

8

∆T (µs)2 4–2–4

250 500

Eve

nts

Top plot: ∆T distribution over a ±500 µsecwindow for events remaining a) afterrequiring that the total collected chargein a 300 nsec time window is equivalentto more than 20 MeV deposited energy(hatched area); b) after all cuts excepttiming (white bars). Bottom plot: ∆Tdistribution over a ±5 µsec window forevents that originate in the fiducialvolume of Super-Kamiokande.

BEAM LINE 49

1 kt near detector. The expectednumbers of Super-Kamiokandeevents based on analyses of interac-tions in the other near detectors areconsistent with this value. Theseresults so far are consistent with theoscillations of muon neutrinos intosome other kind of neutrino, asreported by the Super-Kamiokandeexperiment, but more events areneeded before the K2K experimentcan claim to have reached definitiveconclusions to this question.

FUTURE PLANS

It will take another two years toachieve the original K2K goal, whichwas to provide a statistically signif-icant test of neutrino oscillationsin a long-baseline experiment. Itappears likely that the experimentwill have time to accumulate at mosta few hundred events in Super-Kamiokande before the KEK syn-chrotron is shut down in 2005. Thisnumber is sufficient to provideimproved confidence limits on oscil-lation parameters, which will givevaluable guidance for future long-baseline experiments. Meanwhile,KEK and the Japan Atomic EnergyResearch Institute will be jointlybuilding a very high-intensity protonaccelerator in Tokai, scheduled forcompletion in 2006. After K2K ends,our physics goals will move fromexploratory to precision measure-ments of the two sets of ∆m2 andmixing angles (representing thesquared-mass difference and thedegree of mixing between each pair-wise combination of the three under-lying neutrino-mass states). A next-generation experiment using theTokai accelerator will again use

Super-Kamiokande as the far detec-tor, this time about 300 km distantfrom the production target. Narrow-band neutrino beams will be used,with mean energy below 1 GeV—an energy range well matched tothe detection efficiency of Super-Kamiokande.

With such a neutrino beam, weshould be able to achieve the fol-lowing goals:

• Observations of muon neutrinodisappearance will allow the para-meter ∆m2

µx for νµ → νx oscillationsto be determined with an accuracyof 10−4 eV2 in one year; a 1 percentmeasurement of the correspondingmixing parameter can be obtained infive years.

• Although recent results fromSuper-Kamiokande and SNO makethe existence of sterile neutrinosseem unlikely, observations ofneutral-current neutrino interactionswill allow improved limits to be seton possible sterile neutrino contri-butions.

• The possibility of neutrino decaycan be definitely tested by searchingfor a clear dip in the νµ disappearancerate as a function of energy.

Measurement of CP violation inthe lepton sector may even be fea-sible, but it will require a much largerfar detector, with mass of about amegaton. There are no fundamen-tal engineering showstoppers, andthe design of a megaton detector(dubbed “Hyper-Kamiokande”) iscurrently under study. The discov-ery of CP violation in the lepton sec-tor, and a great improvement of lim-its on—or the possible observationof—proton decay may finally revealthe secret of the baryon-antibaryonasymmetry in the Universe.

For Further Information

To learn more about the K2K andSuper-Kamiokande experiments,go to:

Official K2K home page:http://neutrino.kek.jp/

K2K FAQ (for the layman):http://www.phys.washington.edu/

~superk/k2k/k2k_faq.html

Super-Kamiokande official homepage:

http://www-sk.icrr.u-tokyo.ac.jp/and SK-USA home page: http://www.phys.washington.edu/

~superk/

Maury Goodman’s Neutrino Oscil-lation Industry page: http://www.hep. anl.gov/ndk/hyper-

text/nuindustry.html

Next Generation Long-baselineNeutrino Experiment (JHFnu):

http://neutrino.kek.jp/jhfnu/

50 FALL 2001

HE LIGHTEST OF THE FUNDAMENTALmatter particles, neutrinos may play importantroles in determining the structure of the Universe.They helped to control the expansion rate of theUniverse during its first few minutes, when deu-

terium and most of its helium were formed. Neutrinos andphotons each accounted for nearly half the energy in theUniverse during its first few thousand years. The fact thatneutrinos are so ubiquitous—with hundreds of them occupy-ing every thimbleful of space today—means that they canhave an impact upon how matter is distributed in the Uni-verse. Over the past two decades, the likelihood of this possi-bility has risen and fallen as more and more data has becomeavailable from laboratory experiments and the latest tele-scopes. We now know from the Super-Kamiokande and SNOexperiments that neutrinos indeed have mass. Current esti-mates suggest that the total neutrino mass in the Universemay even be comparable to that of the visible stars.

By 1979 cosmologists had become convinced that most ofthe matter in the Universe is completely invisible. This grav-itationally dominant component of the Universe was named“Dunkelmaterie,” or “dark matter,” by the astronomer FritzZwicky, who in 1933 first described evidence for dark matterin the Coma cluster (see article on Fritz Zwicky by StephenMaurer, Winter 2001 Beam Line, Vol. 31, No. 1). Its galaxieswere moving much too fast to be held together by the gravityof its visible matter. This phenomenon was subsequentlyseen in other galaxy clusters, but since the nature of this

by JOEL R. PRIMACK

Neutrinos

may have had

a significant

impact on

the structure

of the Universe.

T

Whatever Happened to H

BEAM LINE 51

dark matter was completelyunknown, it was often ignored.During the 1970s, it became clearthat the motion of stars and gas ingalaxies—and of satellite galaxiesaround them—required that thisdark matter must greatly outweighthe visible matter in galaxies. Thedata gathered since then providesvery strong evidence that most ofthe matter in the Universe isinvisible.

DARK MATTER made of light neutrinos, withmasses of a few electron volts (eV) or less, is called“hot dark matter” (HDM) by cosmologists, because

those neutrinos would have been moving at nearly the speedof light in the early Universe. (See the tables on the nextpage for a summary of dark-matter types and associated cos-mological models.) For a few years in the late 1970s and early1980s, hot dark matter looked like the best dark-matter can-didate. But HDM models of cosmological structure forma-tion led to a “top-down” formation scenario, in whichsuperclusters of galaxies are the first objects to form afterthe Big Bang, with galaxies and clusters forming through asubsequent process of fragmentation. Such models wereabandoned by the mid-1980s after cosmologists realized that

ot Dark Matter?

The Coma cluster of galaxies, which isabout 50 million light years distant fromthe Milky Way.

52 FALL 2001

on nearby galaxies and clusters onlyif the average density of matter in theUniverse were at or close to the crit-ical density (Ωm = 1). But like all suchcritical-density models, CHDMrequired that galaxies and clustersmust have formed fairly recently,which disagrees with observations.The evidence now increasingly favorsΛCDM models, in which cold darkmatter makes up about a third of thecritical density, with a cosmologicalconstant Λ or some other form of“dark energy” contributing theremainder. This model also helpedresolve the crisis regarding the age ofthe Universe (see box on next page).The question has now become oneof how much room is left for neu-trinos in these cosmological models.

To describe the possible role ofneutrinos as dark matter, I must nowexplain how structures such as galax-ies formed as the Universe expanded.The expansion itself is described byour modern theory of gravity andspacetime, Einstein’s theory of gen-eral relativity. In order for structureto form, there must have been somesmall fluctuations in the initial den-sity of matter, or else some mecha-nism had to generate such fluctua-tions afterward. The only suchmechanisms that have been investi-gated are “cosmic defects” such ascosmic strings, and the pattern offluctuations produced by suchdefects is inconsistent with the tem-perature fluctuations observed in thecosmic microwave background(CMB) radiation. On the other hand,“adiabatic” fluctuations—in whichall components of matter and energyfluctuate together—occur naturallyin the simplest cosmic inflation mod-els and are in excellent agreement

if galaxies had formed early enoughto agree with observations, their dis-tribution would be much moreinhomogeneous than is the case.

Since 1984, the most successfulstructure-formation models havebeen those in which most of themass in the Universe comes in theform of cold dark matter (CDM)—par-ticles that were moving sluggishly inthe early Universe. But the HDMstock rose again a few years later, andfor a while in the mid-1990s it ap-peared that a mixture of mostly CDMwith 20–30 percent HDM gave a bet-ter fit to the observations than eitherone or the other. This “cold plus hotdark matter” (CHDM) theory fit data

Types of Dark Matter

Ωi represents the fraction of the critical density ρc = 10.54 h2 keV/cm3 needed to close the Universe, where h is the Hubble constant H0 divided by 100 km/s/Mpc.

Dark Matter Fraction ofType Critical Density Comment

Baryonic Ωb ~ 0.04 about 10 times the visible matter

Hot Ων ~ 0.001–0.1 light neutrinos

Cold Ωc ~ 0.3 most of the dark matter in galaxy halos

Dark Matter and AssociatedCosmological Models

Ωm represents the fraction of the critical density in all types of matter.ΩΛ is the fraction contributed by some form of “dark energy.”

Acronym Cosmological Model Flourished

HDM hot dark matter with Ωm = 1 1978–1984

SCDM standard cold dark matter with Ωm = 1 1982–1992

CHDM cold + hot dark matter with Ωc ~ 0.7 and Ων = 0.2–0.3 1994–1998

ΛCDM cold dark matter Ωc ~ 1/3 and ΩΛ ~ 2/3 1996–today

BEAM LINE 53

with the latest CMB results an-nounced at the American PhysicalSociety meeting in April 2001.

The evolution of adiabatic fluc-tuations into galaxies and clusters iseasy to understand if you just thinkof gravity as the ultimate capitalistprinciple: the rich always get richerand the poor get poorer. A “rich”region of the Universe is one that hasmore matter than average. Althoughthe average density of the Universesteadly decreases due to its expan-sion, those regions that start out witha little higher density than averageexpand a little slower than averageand become relatively more dense,while those with lower densityexpand a little faster and become rel-atively less dense. When any regionhas achieved a density about twicethe average, it stops expanding andbegins to collapse—typically firstin one direction, forming a pancake-shaped structure, and then in theother two directions.

I can now explain why the firsthot-dark-matter boom occurredabout two decades ago. Improvingupper limits on CMB anisotropieswere ruling out the previouslyfavored cosmological model, whichincluded only ordinary matter. Therewas also evidence from a Moscowexperiment suggesting that the elec-tron neutrino mass was about20–30 eV, which would have corre-sponded to a nearly critical-densityUniverse in which neutrinos wouldhave dominated the total mass (seebox on page 56). In such a cosmology,the primordial fluctuations on galaxyscales are erased by “free streaming”of the relativistic neutrinos in theearly Universe. One year after the BigBang, a region about one light year

across contained the amount of mat-ter (both ordinary and dark) in a largegalaxy like our own Milky Way. Butthe temperature was then about 100million degrees, so each particle hada thermal energy far higher than therest energy of light neutrinos. As theywould therefore have been movingat nearly the speed of light, theseneutrinos would have rapidly spreadout, and any fluctuations in densityon the scale of galaxies would soonhave been smoothed back to the aver-age density.

The first scales to collapse in sucha HDM scenario would therefore cor-respond to the mass inside the cos-mic horizon when the temperaturedropped to a few eV and the neutri-nos inside it became nonrelativistic.This mass turns out to be about10,000 times the mass of our galaxy,including its dark halo. Evidence wasjust then becoming available fromthe first large-scale galaxy surveysthat the largest cosmic structures—superclusters—have masses of ap-proximately this size, which at firstglance appeared to be a big successfor the HDM scenario.

Superclusters of roughly pancakeshape were observed to surroundroughly spherical voids (regionswhere few galaxies are found), inagreement with the first cosmologi-cal computer simulations, whichwere run for the HDM model. In thispicture superclusters should haveformed first, since any smaller-scalefluctuations in the dominant hotdark matter would have been erasedby free streaming. Galaxies then hadto form by fragmentation of thesuperclusters. But it was alreadybecoming clear from observationsthat galaxies are much older than

The Ageof the Universe

IN THE MID-1990S there was acrisis in cosmology, because the age ofthe old globular-cluster stars in the MilkyWay, then estimated to be tGC = 16±3billion years (Gyr), was higher than theexpansion age of the Universe, which forΩm = 1 is texp = 9±2 Gyr. Here I haveassumed that the Hubble parameter hasthe value h = H0/(100 km/s/Mpc)= 0.72±0.07, the final result from theHubble Space Telescope projectmeasuring H0.

But when the data from theHipparcos astrometric satellite becameavailable in 1997, it showed that the dis-tances to the globular clusters had beenunderestimated, which implied in turnthat their ages had been overestimated;now tGC = 13±3 Gyr. The age of theUniverse is consequently tU ~ tGC + 1= 14±3 Gyr.

Several lines of evidence now showthat the Universe does not have Ωm = 1but rather Ωm + ΩΛ = 1.0±0.1 and Ωm= 0.3±0.1. Taken together, these yieldtexp = 13±2 Gyr, in excellent agreementwith the revised globular cluster age. Thehigh-redshift supernova data alone givetSN = 14.2±1.7 Gyr.

Moreover, a new type of agemeasurement based on radioactivedecay of Thorium-232 (half-life 14.1 Gyr)measured in a number of very old starsgives a completely independent agetdecay = 14±3 Gyr. A similar measure-ment, based on the first detection in astar of Uranium-238 (half-life 4.47 Gyr),reported this past February, gives tdecay= 12.5±3 Gyr. Work in progress shouldsoon improve the precision of thesemeasurements.

All the recent measurements of theage of the Universe are therefore inexcellent agreement. It is reassuring thatthree completely different cosmologicalclocks—stellar evolution, expansion ofthe Universe, and radioactive decay—agree so well.

54 FALL 2001

predicted (to within a normalizationuncertainty factor of about 2) themagnitude of the CMB temperaturefluctuations, which were discoveredin 1992 using the COBE satellite. Butthe simplest CDM model, standardCDM (or SCDM) with the matter den-sity equal to the critical value (Ωm

= 1), had already begun to run intotrouble.

Cosmological theories predict sta-tistical properties of the Universe—for example, the size of density fluc-tuations on various scales, describedmathematically by a power spec-trum. Sound or other fluctuationphenomena can be described in thesame way—for example, low fre-quencies might be loud, correspond-ing to relatively high power at longwavelengths. Put simply, SCDM haddifficulty fitting the full spectrum ofdensity fluctuations at both short andlong wavelengths—or at small andlarge scales. With a given amount

superclusters, contrary to what theHDM scenario implies. And theapparent detection of electron neu-trino mass by the Moscow experi-ment was soon contradicted byresults from other laboratories. Hotmatter fell into decline.

IHELPED TO DEVELOP thecold dark matter model in1982–1984 just as the problems

with the hot dark matter model werebecoming clear. Proto-galaxies formfirst in a CDM cosmology, and galax-ies and larger-scale objects form byaggregation of these smaller lumps—although the cross-talk betweensmaller and larger scales in the CDMtheory naturally leads to galaxiesforming earlier in clusters than inlower-density regions. In this andother respects, CDM models appearedto fit observations much better thanHDM. The first great triumph of colddark matter was that it successfully

The Andromeda Galaxy, which togetherwith the Milky Way and several dwarfgalaxies, forms the Local Group ofgalaxies.

BEAM LINE 55

of fluctuation power on the largescales probed by COBE (billions oflight years), that is, SCDM has a lit-tle too much power on small scalesrelevant to galaxies and clusters (mil-lions of light years and less). It pro-duces too many of them. But the factthat the SCDM theory could workfairly well across such a wide rangeof size scales suggested that it hada kernel of truth. Cosmologists be-gan to examine whether some vari-ant of SCDM might work better.

Although these COBE results didnot come in until 1992, I became wor-ried that SCDM was in trouble whenthe large-scale flows of galaxies werefirst observed by my UC Santa Cruzcolleague Sandra Faber and her “Sev-en Samurai” group of collaborators.Earlier studies had established thatthe local group of galaxies (includingthe Milky Way and the “nearby”Andromeda galaxy) is speeding at avelocity of about 600 kilometers persecond with respect to the CMB ref-erence frame. But the Seven Samu-rai and others found that bulkmotions of other galaxies often hadsimilar high velocities across regionsseveral tens of millions of light yearswide. It soon became clear that thisresult was inconsistent with theexpectations of the “biased” SCDMmodel that seemed to fit the prop-erties of galaxies on smaller scales.In the late 1980s Jon Holtzman, thenFaber’s graduate student, did a the-oretical dissertation with me inwhich he calculated detailed expec-tations for 96 variants of cold darkmatter. When we compared thesepredictions with the data availablein early 1992, it was clear that thebest bets were CHDM and ΛCDM,each of which could fit the data on

small scales better than SCDM. Bothof these variants had been proposedin 1984, when cold dark matter wasstill a new idea, but their detailedconsequences were not worked outuntil the problems with SCDM beganto surface.

EVEN IF MOST of the darkmatter is cold, a little hot darkmatter can still have dramat-

ic effects on the small scales relevantto the formation and distribution ofgalaxies. In the early Universe, thefree streaming of fast-moving neu-trinos would have washed out anyspatial inhomogeneities on the scalesthat later became galaxies, just asin the HDM scenario. Consequently,CDM fluctuations did not grow asfast on these scales, and at the rela-tively late times when galaxiesformed there was less flutuationpower on small scales in CHDM

models. The main problem with Ωm

= 1 cosmologies containing only colddark matter plus a small admixtureof ordinary baryonic matter is thatthe galaxy-scale inhomogeneities aretoo big compared to those on largerscales. Adding a little hot dark mat-ter appeared to be just what wasneeded to solve this problem.

And there was even a hint froman accelerator experiment that neu-trino mass might lie in the relevantrange. The experiment was theLiquid Scintillator Neutrino Detector(LSND) experiment at Los AlamosNational Laboratory (see photographon page 38), which recorded a num-ber of events that appear to be νµ → νe

oscillations. Comparison of the LSNDdata with results from other neutrinoexperiments allows two discretevalues of ∆mµe

2, (see box, next page)around 10.5 and 5.5 eV2, or a rangeof values between 0.2 and 2 eV2. Iftrue, this means that at least one neu-trino has a mass greater than 0.5 eV,which would imply that the contri-bution of hot dark matter to the cos-mological density is much greaterthan that of all the visible stars. Suchan important conclusion requires in-dependent confirmation. The KArl-sruhe Rutherford Medium EnergyNeutrino (KARMEN) experimentresults exclude a significant portion(but not all) of the LSND parameterspace, and the numbers quoted abovetake into account the current KAR-MEN limits. The Booster NeutrinoExperiment (BooNE) at Fermilab (seearticle by Paul Nienaber, this issue)should attain greater sensitivity andhelp to resolve this issue.

By 1995 simulation techniquesand supercomputer technology hadadvanced to the point where it

The detailed studies

of cosmological

structures now

going on or about

to begin may

eventually reveal

something about

neutrino mass itself.

56 FALL 2001

was possible to do reasonably high-resolution cosmological-scale simu-lations including the random veloc-ities of a hot dark matter component.The results at first appeared veryfavorable to CHDM. Indeed, as late as1998 a CHDM model (with Hubbleparameter h = 0.5, mass density Ωm

= 1, and neutrino density Ων = 0.2) wasfound to be the best fit of any cos-mological mode to the galaxy dis-tribution in the nearby Universe. Butcosmological data was steadily im-proving, and even by 1998 it had be-come clear that h = 0.5 and Ωm = 1were increasingly inconsistent withseveral observations, and that h ~ 0.7and Ωm ~ 1/3 worked much better.For example, CHDM predicts thatgalaxies and clusters formed rela-tively recently, but around 1998increasing numbers of galaxies werediscovered to have formed in the firstfew billion years after the Big Bang.And the fraction of baryons found inclusters, together with the reason-able assumption that this fraction isrepresentative of the Universe as awhole, again gives Ωm ~ 1/3. Thatthere is a large cosmological constant(or some other form of dark energy)yielding ΩΛ ~ 2/3 then follows fromany two of the following threeresults: (1) Ωm ~ 0.3, (2) CMB aniso-tropy data implying that Ωm + ΩΛ =1, and (3) high-redshift supernovadata implying that ΩΛ – Ωm ~ 0.4.

The abundance of galaxies andclusters in the early Universe agreeswell with the predictions of theΛCDM model. However, the highest-resolution simulations of this modelthat were possible in the mid-1990sgave a dark matter spectrum that hadmore power on scales of a fewmillion light years than did the

Neutrino Mass and Cosmological Density

THE ATMOSPHERIC-NEUTRINO DATA from the Super-Kamiokande underground neutrino detector in Japan provide strongevidence of muon to tau neutrino oscillations, and therefore that theseneutrinos have nonzero mass (see the article by John Learned in theWinter 1999 Beam Line, Vol. 29, No. 3). This result is now being confirmedby results from the K2K experiment, in which a muon neutrino beam fromthe KEK accelerator is directed toward Super-Kamiokande and the numberof muon neutrinos detected is about as expected from the atmospheric-neutrino data (see article by Jeffrey Wilkes and Koichiro Nishikawa, thisissue).

But oscillation experiments cannot measure neutrino masses directly,only the squared mass difference ∆m2

ij = |mi2− mj

2 | between the oscillatingspecies. The Super-Kamiokande atmospheric neutrino data imply that1.7×10−4 < ∆m2

τµ < 4×10−3 eV2 (90 percent confidence), with a central value∆m2

τµ = 2.5×10−3 eV2. If the neutrinos have a hierarchical mass pattern mνe

<< mνµ<< mντ

like the quarks and charged leptons, then this impliesthat ∆m2

τµ ≅ m2ντ

so mντ~ 0.05 eV.

These data then imply a lower limit on the HDM (or light neutrino)contribution to the cosmological matter density of Ων > 0.001—almost asmuch as that of all the stars in the disks of galaxies. There is a connectionbetween neutrino mass and the corresponding contribution to the cosmo-logical density, because the thermodynamics of the early Universe speci-fies the abundance of neutrinos to be about 112 per cubic centimeter foreach of the three species (including both neutrinos and antineutrinos). Itfollows that the density Ων contributed by neutrinos is Ων = m(ν)/(93 h2 eV),where m(ν) is the sum of the masses of all three neutrinos. Since h2 ~ 0.5,mντ

~ 0.05 eV corresponds to Ων ~ 10-3.This is however a lower limit, since in the alternative case where the

oscillating neutrino species have nearly equal masses, the values of theindividual masses could be much larger. The only other laboratoryapproaches to measuring neutrino masses are attempts to detect neutrino-less double beta decay, which are sensitive to a possible Majorana compo-nent of the electron neutrino mass, and measurements of the endpoint ofthe tritium beta-decay spectrum. The latter gives an upper limit on theelectron neutrino mass, currently taken to be 3 eV. Because of the smallvalues of both squared-mass differences, this tritium limit becomes anupper limit on all three neutrino masses, corresponding to m(ν) < 9 eV. Abit surprisingly, cosmology already provides a stronger constraint on neu-trino mass than laboratory measurements, based on the effects of neutri-nos on large-scale structure formation.

BEAM LINE 57

observed galaxy power spectrum,although the simulations and dataagreed on larger scales. This resultwas inconsistent with the expecta-tions that galaxies would be (if any-thing) more clustered than the darkmatter on small scales, not less.When it became possible to do evenhigher-resolution simulations thatallowed the identification of the darkmatter halos of individual galaxies,however, their power spectrumturned out to be in excellent agree-ment with observations. The galax-ies were less clustered than dark mat-ter because galaxies had merged orwere destroyed in very dense regionsdue to interactions with each otherand with the cluster center. Thisexplanation turned a troubling dis-crepancy into a triumph for ΛCDM.

Thus ΛCDM is the favorite theorytoday. But we know from the Super-Kamiokande evidence for atmos-pheric neutrino oscillations thatthere is enough neutrino mass to cor-respond to some hot dark matter, atleast Ων ~ 10−3 (see box on oppositepage) about one-fourth as much asthe visible stars. So the remainingquestion regarding neutrinos in cos-mology is how much room remainsfor a little hot dark matter in ΛCDM.cosmologies. There could be perhaps10 times the lower limit just quoted,but probably not 100 times. The rea-son there is any upper limit at allfrom cosmology is because the freestreaming of neutrinos in the earlyUniverse must have slowed thegrowth of the remaining CDM fluc-tuations on small scales. Thus tohave the galaxy structures we see to-day, there must be much more coldthan hot dark matter. For the obser-vationally favored range 0.2 ≤ Ωm ≤

0.5, the limit on the sum of the neu-trino masses is m(ν) < 2.4 (Ωm/0.17 – 1)eV. Thus for Ωm < 0.5, the sum of allthree neutrino masses is less than5 eV. This limit on the sum of the neu-trino masses is much stronger thanthe best current laboratory limit.

Astronomical observations thatmay soon lead to stronger upperlimits on m(ν)—or perhaps even a de-tection of neutrino mass—includedata on the distribution of low-density clouds of hydrogen (the so-called “Lyman-alpha forest”) in theearly Universe, large-scale weak grav-itational lensing data, and improvedmeasurements of the fluctuations inthe CMB radiation temperature atsmall angles. These data are sensi-tive to the presence of free-streamingneutrinos in the early Universe,which can lead to fewer galaxies onsmall scales, depending on the val-ues of the neutrino masses.

The hot dark matter saga thusillustrates once again the fruitfulmarriage between particle physicsand cosmology. While neutrino-oscillation experiments can tell usonly about differences between thesquared masses of the neutrinos, thestructure of the Universe can give usinformation about the neutrinomasses themselves. In an earlierexample of this connection, cosmo-logical arguments based on Big Bangnucleosynthesis of light elements puta strict limit on the possible numberof light neutrino species; this limitwas eventually borne out by high-energy physics experiments on Zbosons at CERN and SLAC. Thedetailed studies of cosmologicalstructures now going on or about tobegin may eventually reveal some-thing about neutrino mass itself.

For Further Reading

For a more technical article on thepresent subject with extensivereferences, see “Hot Dark Mat-ter in Cosmology,” by Joel R.Primack and Michael A. K.Gross, in Current Aspects ofNeutrino Physics, D. O. Cald-well, Ed. (Berlin: Springer,2001); also available on theInternet as astro-ph/0007165and interactively as nedwww.ipac.caltech.edu/level5/Primack4/frames.html

The latest published upper limiton the sum of neutrino massesfrom cosmological structureformation is Rupert A. C. Croft,Wayne Hu, and Romeel Davé,Phys. Rev. Lett 83, 1092(1999).

X. Wang, M. Tegmark, and M. Zaldarriaga, astro-ph/0105091, gives a tighterupper limit of 4.4 eV on m(ν).

58 FALL 2001

PAUL NIENABER has the greatgood fortune to be part of twoextraordinary neutrino groups atFermilab. He joined the NuTeV col-laboration in 1992 and is now work-ing on data analysis from thatexperiment; in addition, he becamea member of the BooNE collaborationin 2000 and is participating in con-struction and testing of its detector.

The photograph above shows Paulinside the BooNE tank, surroundedby the photomultiplier tubes hehelped install. He is a Jesuit priest, aGuest Scientist at Fermilab, and amember of the physics faculty at theCollege of the Holy Cross in Worces-ter, Massachusetts, where he thor-oughly enjoys infecting college stu-dents’ minds with the manifoldpleasures of doing physics. This is hisfirst article for the Beam Line.

JOSHUA KLEIN received hisPh.D. from Princeton University in1994. He then moved to the Univer-sity of Pennsylvania where he hasworked on building and makingmeasurements with the SudburyNeutrino Observatory. He is cur-rently Research Assistant Professorof Physics at Pennsylvania.

SNO is in many ways the quin-tessential particle astrophysicsexperiment featuring the Sun as aneutrino laboratory and neutrinos assolar probes. Klein’s interest in SNOencompasses both of these features,including an interest in neutrinosthemselves as the Standard Model’smost enigmatic particle.

MICHAEL RIORDAN is Assis-tant to the Director at the StanfordLinear Accelerator Center, Lecturerin the History and Philosophy ofScience Program at Stanford, andAdjunct Professor of Physics at theUniversity of California, Santa Cruz.A Contributing Editor of the BeamLine, he is author of The Huntingof the Quark, and coauthor of TheShadows of Creation: Dark Matterand the Structure of the Universeand Crystal Fire: The Birth of theInformation Age. He leads a group ofhistorians and physicists researchingand writing the history of the Super-conducting Super Collider. In 1999 hereceived a Guggenheim Fellowshipto pursue research on this subjectat the Smithsonian Institution inWashington, DC. While there, he alsogot married again.

CONTRIBUTORS

BEAM LINE 59

BORIS KAYSER is a theoreticalphysicist who has been particularlyinterested in the physics of massiveneutrinos and the asymmetry be-tween matter and antimatter. Healso enjoys brachiating around inquantum mechanics. An author ofwell over 100 scientific papers, heis also co-author of a popular slenderbook on neutrino physics and a fre-quent, enthusiastic speaker on par-ticle physics topics.

For nearly three decades, Kayserserved as Program Director forTheoretical Physics at the NationalScience Foundation, in which capac-ity he was instrumental in estab-lishing the Institute for Theoreti-cal Physics at the University ofCalifornia, Santa Barbara. He hasnow joined Fermilab as a Distin-guished Scientist and looks forwardto spending full time on his firstlove—physics research.

Kayser enjoys nature, and he isone of the primates in the photoabove.

KOICHIRO NISHIKAWA isProfessor of Physics at Kyoto Uni-versity. He received his Ph.D. in 1980from Northwestern University,where his thesis research involveddi-muon production using the highenergy neutrino beam at Fermilab.As a postdoc at the University ofChicago, he was subsequently in-volved in an experiment on CPviolation in the neutral kaon system.

Nishikawa returned to Japan—tothe Institute for Nuclear Study of theUniversity of Tokyo—and joined theSuper-Kamiokande experiment. Hemoved to Kyoto University in 1999.He serves as the spokesman for theK2K long-baseline neutrino experi-ment.

JEFFREY WILKES is Kenneth K.Young Memorial Professor of Physicsat the University of Washington inSeattle. His thesis research at theUniversity of Wisconsin involved thelast of many cosmic-ray experimentsconducted at Echo Lake, Colorado.As a postdoc at the University ofWashington, he learned the nuclearemulsion trade as Jere Lord's ap-prentice and has been there eversince.

After an interlude with thequixotic and habit-forming enterprisecalled DUMAND, he joined Super-Kamiokande, where he is convenerof an analysis group. He is US co-spokesman for the K2K long-baselineneutrino experiment, to which hismain contribution seems to be thecare and feeding of a wildly fluctu-ating shift schedule. Most recently,he has been developing WALTA, aschool-network cosmic-ray detectorproject in Seattle.

60 FALL 2001

JOEL PRIMACK is Professor ofPhysics at the University of Cali-fornia, Santa Cruz. He has beenworking at the interface between par-ticle physics and cosmology since theearly 1980s, when he and several col-leagues conceived the idea of colddark matter, which has since becomethe “standard model” of cosmologi-cal structure formation. He is cur-rently doing research in Germanyunder an award from the HumboldtFoundation.

Primack has also worked exten-sively on science and public policyissues. He helped initiate the Con-gressional Fellowship program spon-sored by the American Associationfor the Advancement of Science, aswell as the American PhysicalSociety’s Forum on Physics andSociety.

A Brief Neutrino Bibliography

John N. Bahcall, Neutrino Astrophysics (Cambridge UniversityPress, New York, 1989)

John N. Bahcall, Raymond Davis, et al., Eds., Solar Neutrinos: TheFirst Thirty Years (Addison-Wesley, Reading, 1994).

David O. Caldwell, Ed., Current Aspects of Neutrino Physics(Springer Verlag, Berlin, 2001)

Boris Kayser, The Physics of Massive Neutrinos (World Scientific,Singapore, 1989)

Nickolas Solomey, The Elusive Neutrino (Scientific American Library,New York, 1997)

Christine Sutton, Spaceship Neutrino (Cambridge University Press,Cambridge, 1992)

Jan 7 - Mar 15 Joint Universities Accelerator School: Courses in the Fundamentals of the Physics,Technologies, and Applications of Particle Accelerators (JUAS 2002), Archamps,France, (J. Delteil, ESI/JUAS, Centre Universitaire de Formation et de Recherche, F-74166 Archamps, France or juas@esi.cur-archamps.fr or http://www.esi.cur-archamps.fr)

Jan 14 - 25 US Particle Accelerator School, Long Beach, CA (uspas@fnal.gov or http://www.indiana.edu/~uspas/programs/ucla.html)

Feb 4 - 8 9th International Workshop on Linear Colliders, Menlo Park, CA (Robbin Nixon,SLAC, MS 66, Box 4349, Stanford, CA 94309 or lc02@slac.stanford.edu)

Mar 3 - 8 International Spring School on the Digital Library and E-Publishing for Science andTechnology, Geneva, Switzerland (http://cwis.kub.nl/~ticer/spring02/index.htm)

Mar 7 - 9 IUPAP International Conference on Women in Physics, Paris, France (Dr. Judy Franz,American Physical Society, One Physics Ellipse, College Park, MD 20740 orhttp://www.if.ufrgs.br/~barbosa/conference.html)

Mar 18 - 22 Annual Meeting of the APS, Indianapolis, IN (American Physical Society, One PhysicsEllipse, College Park, MD 20740 or http://www.aps.org/meet/MAR02/)

Apr 20 - 23 Meeting of the APS and the High-Energy Astrophysics Division (HEAD) of theAmerican Astronomical Society, Albuquerque, NM (American Physical Society,One Physics Ellipse, College Park, MD 20740 (http://www.aps.org/meet/APR02/)

May 8 - 17 CERN Accelerator School: Superconductivity and Cryogenics for Accelerators andDetectors, Erice, Sicily (CERN Accelerator School, AC Division, 1211 Geneva 23,Switzerland or suzanne.von.warburg@cern.ch)

May 24 - 28 DPF 2002: Meeting of the Division of Particles and Fields of the American PhysicalSociety, Williamsburg, VA (Marc Sher, Physics Department, College of William andMary, Williamsburg, VA 23187 or sher@physics.wm.edu or http://www.dpf2002.org)

May 25 - 30 20th International Conference on Neutrino Physics and Astrophysics (Neutrino 2002),Munich, Germany (NEUTRINO 2002 Secretariat, Technische Universitat Munchen,Physik Department E15, D-85747 Garching, Germany or neutrino2002@ph.tum.de orhttp://neutrino2002.ph.tum.de/)

Jun 20 - 22 Physics in Collisions, Stanford, CA (Maura Chatwell, MS 81, 2575 Sand Hill Road,Menlo Park, CA 94025 or maura@slac.stanford.edu)

DATES TO REMEMBER