Searching for Neutrino Oscillations by MAURY GOODMAN

Experiments of the past forty years have revealed three families of the ghostly particles called neutrinos. Continuing studies hint that a neutrino of one family might sometimes change into a neutrino of a different family, by a mechanism known as neutrino oscillation. The author describes why understanding this phenomenon might be critical to the question of whether neutrinos have mass.

HYSICISTS FROM AROUND the world are engaged in a wide variety of experiments to determine whether neutrinos have mass. This possibility has intrigued physicists and cosmologistsP for two decades, ever since neutrinos emerged as a leading candidate for the dark matter thought to inhabit the Universe. A comprehensive new experiment is being built in Illinois and Minnesota to study neutrinos from an intense new Fermilab beam impinging on a detector 500 miles away. It is one of the most ambitious of a new round of experiments being planned and proposed to search for neutrino oscillations, a process in which neutrinos can transform from one kind into another—if they have mass. A positive result could have implications for the density of the Universe, as well as for the generation of energy by the Sun. Often, given a well-deﬁned physics problem, one or two well-designed experiments can answer the question one way or the other. But this is not the case for neutrino mass, because there are three different kinds of neutrinos and a wide range of possible mass scales. This situation has led physicists to at- tempt a large number of experiments that are quite different from each other. In this article, I relate why so many physicists are excited by neutrino- oscillation experiments. First, I describe the properties of neutrinos them- selves. Then I cover some of the experimental hints supporting neutrino oscillations. Finally, I close with a description of the Fermilab-to-Soudan, Minnesota, long-baseline neutrino project, an ambitious program to search for changes in the properties of a neutrino beam as it speeds silently beneath the farms and prairies of the American Midwest.

BEAM LINE 9 Melvin Schwartz in front of the Brookhaven detector that showed experimenters in 1962 that the muon had its own neutrino, different from the electron neutrino. Schwartz, Leon Lederman, and Jack Steinberger won the Nobel Prize in 1987 for this discovery. (Courtesy Brookhaven National Laboratory)

HERE ARE THREE source of copious neutrinos, but I am “flavors” of neutrinos, the not aware of any experiment that has electron neutrino ν , the detected them.) Neutrinos are either T e muon neutrino νµ, and the tau neu- massless or far lighter than the trino ντ. Each is closely related to the quarks and other leptons. This dif- corresponding lepton: the electron, ference might be related to the fact muon, and tau lepton. These six lep- that neutrinos have no electric tons together with six quarks con- charge, while the quarks and other stitute the fundamental “matter” leptons do have charge. The question particles of the Standard Model of of mass remains one of the big mys- high energy physics. teries remaining in particle physics. The three neutrinos interact very There is no clear prediction relat- weakly with ordinary matter. Physi- ing the masses of the nine charged cists originally thought that the great fermions, and none for whether the weakness of the interaction would neutrino masses are zero or just very make them impossible to detect, but small. neutrinos have been seen coming However, most physicists expect from accelerators, from nuclear re- that if neutrinos do have mass, even actors, from cosmic-ray interactions a tiny amount, the phenomenon of in the atmosphere, from the sun, and neutrino oscillations should exist from Supernova 1987A. (Nuclear (see the box on the opposite page). weapon explosions are also the These transformations are closely related to the quantum-mechanical phenomenon of mixing. If neutrinos oscillate, they can be produced in one flavor, such as νµ, and be detected as another flavor, such as ντ, some distance away. When Pauli predicted the existence of the neutrino in 1930, he did not suppose there would be more than one kind. He was only trying to explain the wide distribution of electron energies observed in nuclear beta decay. The idea that neutrinos come in different ﬂavors became ac- cepted in 1962, when an experiment at Brookhaven National Laboratory directed neutrinos from pion decay at a target and found that almost all of the events had a muon, and not an I II III of Matter electron, emerging from the point of Three Generations the neutrino interaction. This result led to the idea that each lepton flavor (e, µ, τ) has a conserved quantity— something that doesn’t change in an

10 SPRING 1998 Probability of Neutrino Oscillations

IN ORDER TO MEASURE neutrino oscillations, the experimenter wants the probability that one neutrino transforms into another to be as large as possible. This probability is given by

2 2 2 Pν → ν = sin (2θ )sin (1.27∆m L/Eν), interaction—associated with it. The two factors affecting neutrino 1 2 12 12 When the pion decays, it almost al- oscillations (see adjacent box) that 2 θ ways becomes a muon and a neutri- are under the control of the experi- where sin (2 12) is the mixing angle, ∆m2 = m2 − m2 is the difference in no and hardly ever an electron and menter are the neutrino energy Eν 12 1 2 a neutrino. The Brookhaven National and the distance L between their mass squares of the two neutrinos, L Laboratory result could be explained source and the detector. These ap- is the distance (in km) from the + + neutrino production point to the ex- if the π decays into a µ and a νµ ; pear in the ratio L/Eν, so an experi- when they interact with the target ment designer needs a large distance periment, and Eν is the neutrino energy in GeV. If either sin2(2θ) = 0 or nuclei, the νµ’s generate muons, not and low energies in order to measure 2 electrons. small values of the mass difference ∆m = 0, the phenomenon of neutrino When the third lepton, the tau, between two neutrino types. This re- oscillations does not exist. If all three 2 was discovered at the Stanford Lin- quirement must be balanced against neutrinos are massless, ∆m = 0. ear Accelerator Center in 1975, it was the fact that large distance and low As a result of the above equation, natural to conjure up a third neutri- energy both make it more difficult the neutrino “oscillates” with a strength sin2(2θ) and an “oscillation no, the ντ, to account for missing en- to detect a large number of neutrino ergy in tau decays. In the 1980s physi- events. length” ______πEν cists discovered and began producing Let’s go back to the Brookhaven Losc = 1.27∆m2 copious numbers of Z bosons; this experiment that discovered the particle served as a neutrino counter muon neutrino. If the mixing The oscillation probability varies as 2 π because its decay rate is proportion- strength and mass difference had sin ( L/Losc). It is the sinusoidal na- al to the number of fundamental par- both been large enough, that exper- ture which gives the name to ticles with less than half its mass. iment would not have been able to “neutrino oscillations.” Measurements of this rate at CERN discover the νµ. It would have seen and SLAC confirmed that there are both electrons and muons coming only three neutrino-like particles in from the point of the neutrino in- the elementary particle zoo—a result teractions! We can use the success of that had been predicted by cosmolo- that experiment to place limits on gists. So far, there has not been any the combination of the two parame- convincing evidence that the ντ in- ters. We usually do this by making a teracts with nuclei to make taus in graph in the parameter space called a manner equivalent to the other two the “∆m2 − sin2 (2θ) plane,” these be- neutrinos. But a current Fermilab ex- ing two parameters that specify the periment is expected to ﬁnd these ντ mixing strength and mass difference interactions. (see the box on the next page). An

BEAM LINE 11 Relevant Neutrino Parameter Space

experiment that is consistent with Simply put, five solar-neutrino 3 10 small or no neutrino oscillations cor- experiments have measured a responds to a curve in that plane that signiﬁcantly smaller number of neu- Missing Matter? excludes the values of mixing trino interactions than expected, strength and mass difference above based on the measured heat output and to the right of the curve. of the Sun and nuclear physics mod- 100 LSND Since the early 1960s, neutrino ex- els of both the Sun and the detectors. periments at Brookhaven, Fermilab, Each experiment observed fewer neu- ) 2 CERN, and the Institute for High En- trinos than expected, but the actual (eV Atmospheric

2 ergy Physics at Serpukhov, Russia, deﬁcit each measures depends on the Ratio m –3 ∆ 10 have grown from tens to thousands detecting medium and energy thresh- to millions of neutrino events. None old. While it is not possible to explain

Up/Down of these experiments has witnessed the data with alternate models of the

ν → ν ν → ν evidence for µ e or µ τ os- Sun, one can account for all the data MSW cillations. And at the same time, ex- within the framework of neutrino 10–6 periments at nuclear reactors have oscillations. Mass Difference, ν found no evidence for e oscillations As discussed in the box on in detectors situated up to a kilo- page 11, the length scale of an ex- meter from the reactor. The pub- periment provides a possible oscil- lished limits have steadily crept to lation length. There are two possible –9 10 lower values of the neutrino mix- scales for solar neutrinos: the dis-

Vacuum ing strength and mass difference. tance from the Sun to the Earth and

the radius of the Sun. Each length 10–5 10–4 10–3 10–2 10–1 1 UT THE STORY by no scale leads to a separate neutrino- Mixing Strength, sin2(2θ) means ends there. While ex- oscillation solution for the solar neu- Bperiments at reactors and high trino deficit. One (labeled “vacuum” energy accelerators have found no in the illustration on the left) arises This graph shows the regions of neutrino mass evidence for them, four hints have from a straightforward solution of ∆ 2 2 θ ( m ) and mixing strength [sin (2 )] which are emerged suggesting the real possi- the relationship between neutrino suggested and ruled out by present data. The bility of neutrino oscillations and mass and oscillations (see the box on shaded regions are ruled out above and to the hence mass. These are the solar neu- the left). The other solutions (labeled ν → ν ν → ν right of the curves labeled µ e and µ τ. trino deficit, the atmospheric neutri- “MSW” after Stanislav Mikheyev, The hatched areas are suggested regions of pa- no deficit, the Liquid Scintillator Neu- Alexei Smirnov, and Lincoln Wolfen- rameter space from the LSND, atmospheric, and trino Detector (LSND) experiment at stein, who formulated the relevant solar neutrino experiments. The band labeled Los Alamos National Laboratory, and theory) obey more complicated equa- “Missing Matter” is where one might expect to the missing matter problem. These tions that take into account the huge find neutrino oscillations if neutrinos contribute hints suggest the existence of neutri- density and density gradients of mat- significantly to the Dark Matter problem. New no oscillations in regions of the ter in the Sun, and how they can af- long-baseline experiments will explore the re- parameter space that have not been fect neutrinos emerging from its core. gion of parameter space suggested by the at- completely ruled out by accelerator Both of these solutions involve os- mospheric ratio and up/down results. experiments. cillations of electron neutrinos into The solar-neutrino deficit has other kinds. been around for thirty years. (See The atmospheric neutrino deficit “What Have We Learned About Solar takes us underground to experiments Neutrinos” by John Bahcall in the that were originally built for another Fall 1994 issue of the Beam Line.) purpose—to search for proton decay.

12 SPRING 1998 These massive detectors, which weigh from one to fifty thousand tons, haven’t discovered proton decay, but they do observe about a hundred interactions of atmospheric neutrinos per year for every thousand tons of detector mass. These neutrinos are created near the top of the atmosphere when cosmic-ray pro- tons initiate a particle cascade, making one or more charged pions, each of which decays into a muon and a νµ. The muon subsequently decays ν ν into an electron, a µ, and a e. Thus, ν ν the ratio of µ flux to the e flux observed in an underground detector should be about two. This is quite a strong prediction, regardless of cosmic-ray rates and the subtleties of calculating the number of parti- would be between 25 km and The LSND detector is designed to search cles produced in the cosmic-ray cas- 12,000 km. There is strong evidence for the presence of electron anti- cades. Underground detectors seem from the SuperKamiokande experi- neutrinos with great sensitivity. Over to be measuring the expected num- ment this is the case. This “up/down 1200 photomultiplier tubes line the inner surface of the oil tank, shown above with ber of electron neutrinos, but only asymmetry” observed seems to favor LSND physicist Richard Bolton of Los ∆ 2 −4 −2 sixty percent of the expected muon a value of m between 10 and 10 Alamos. (Courtesy Los Alamos National ν 2 neutrinos. This µ deficit could be ex- eV (region labeled “Up/Down” in Laboratory) plained by νµ → ντ oscillations, with the illustration). the ντ too low in energy to produce The LSND experiment at Los a tau lepton by interacting with a nu- Alamos, unlike the solar and atmos- cleus. This deficit seems to indicate pheric neutrino experiments, was ex- − a value of ∆m2 betyween 10 3 and plicitly built to look for neutrino os- 1 eV2 (see region labeled “Atmos- cillations. Operating near the target pheric Ratio” in the illustration on of the LAMPF accelerator, it uses a page 12). very intense π+ beam. The pions stop + The distance that atmospheric in the target, decay into a µ and a νµ, + + neutrinos travel before hitting a de- and the µ decays into an e , a νµ, and ν tector varies from 25 kilometers for a e. Except for a small and calcula- those coming from overhead to ble background from negative pion ν– 12,000 kilometers for those coming decays, there are no e’s in the beam. from the other side of the Earth. This So if excess numbers of these parti- provides an opportunity to determine cles are detected, they probably arose ν– → ν– whether there is any difference in the from µ e’s oscillations. The ex- signal between the up-going and the periment has a 170 ton tank of min- down-going neutrinos. If so, the os- eral oil that can detect the reaction ν– → + cillation length for typical atmos- ep ne by measuring a 15–30 MeV pheric neutrino energies (500 MeV) positron in coincidence with a signal

BEAM LINE 13 from neutron capture, which yields density should just equal the critical a 2.2 MeV gamma ray. The experi- value, though there are recent ob- ment has reported a signal that could servational data which suggest only be explained by neutrino oscillations twenty percent of that value. Ordi- nary baryons, the stuff that makes with a strength P(νµ→νe) = 0.003. Unlike the atmospheric and solar up stars and stuffed pizza, is only neutrino deficits, the LSND signal has about five percent of the critical val- been observed in only one experi- ue, based on both observational data ment. In fact, other experiments that and the rates of light element pro- are sensitive to these oscillations duction during the Big Bang. Some over similar regions of parameter of this missing matter may well be space have obtained negative results. neutrinos; there are hundreds of The region favored by LSND but not them in every cubic centimeter of ruled out by other experiments the Universe. If they had a mass of (labeled “LSND” in the figure on just 5 eV, neutrinos would outweigh page 12) suggests a value of ∆m2 all the stars and pizza in the around 1 eV2. Universe. The final hint, the missing mat- Experimenters want to confront ter problem, is really suggestive of these hints of neutrino oscillations neutrino mass rather than oscilla- with more definitive measurements. tions. There is a critical density of New solar neutrino experiments are matter in the Universe (see article by determining the size, time depen- Alan Guth in the Fall 1997 issue of dence, and energy dependence of the the Beam Line, Vol. 27, No. 3), about solar neutrino deficit. New short- one hydrogen atom per cubic me- baseline oscillation experiments are ter, above studying mass differences in the re- Select Present/Future Neutrino Experimentsa which the gion of the missing matter problem. Universe is The reported LSND effect will be Neutrino closed and sought by another collaboration at Experiments Energy Location Status will some- the Rutherford Laboratory in Britain, Solar and there is a proposal for a future SuperKamiokande 7 MeV Kamioka, Japan Current day collapse Sudbury (SNO) 4 MeV Ontario, Canada Beginning back into a follow-up detector at Fermilab (see Atmospheric single point. table on the left for a small selection Soudan 2 600 MeV Minnesota Current If the densi- of these experiments). MACRO 5 GeV Gran Sasso, Italy Current ty is at or Reactor below this HILE UNDERGROUND Chooz 5 MeV France Current critical den- experiments will continue Palo Verde 5 MeV Arizona Future Short-baseline sity, the Uni- Wto study the atmospheric NOMAD 50 GeV Geneva, Switzerland Current verse is open neutrino deficit, there is another plan LSND 40 MeV Los Alamos Current and will ex- to study possible neutrino oscilla- Long-baseline pand forev- tions in the same region of parame- MINOS 20 GeV Fermilab to Minnesota Future er. There ter space. These are the long-baseline ICARUS 30 GeV Switzerland to Italy Future are strong experiments. While short-baseline K2K 1 GeV Tsukuba to Kamioka, Japan Future theoretical experiments are typically one kilo- a A more complete list of neutrino experiments can be found at arguments meter from the point where the neu- http://www.hep.anl.gov/ndk/hypertext/nu_industry.html that the trinos are produced, long-baseline

14 SPRING 1998 Lake Soudan Superior

Duluth

MN Lake WI Michigan

Madison experiments in the United States and the pion decay results in an average MI Europe will be located 730 km away angle between a neutrino and the IA from the source. And another exper- original beam of about 1/20th of a Fermilab iment in Japan will have 250 km be- degree. tween neutrino production and de- One obvious concern in aiming IL IN tector. All three of these choices are a beam at a target so far away is the matters of convenience—the dis- precision required to hit it, but this MO tances between existing accelerators turns out to be only a minor prob- and existing underground facilities. lem. Hitting the target is a similar to As luck would have it, however, all aiming a flashlight at the moon. three projects will substantively ad- Most people could hold the ﬂashlight dress the possibility that the and point accurately enough. The Fermilab10 km Soudan atmospheric neutrino deﬁcit is due problem comes in seeing the flash- 730 km to neutrino oscillations. light while standing on the moon. As an example of one of the most This could only be accomplished 12 km ambitious new neutrino oscillation with a powerful enough light. The Map showing long-baseline neutrino projects, I will now focus on the long-baseline neutrino problem is experiment planned from Fermilab in Fermilab-to-Soudan long-baseline similar. The neutrino beam spreads Illinois to Soudan in Minnesota. project (see map on the right). There out as it recedes from Fermilab, los- are three elements to the project: the ing its intensity. And, neutrinos are neutrino beam, a near detector at Fer- very weakly interacting, so one needs milab, and a far detector at the a very massive target in order to de- Soudan underground physics labora- tect just a few of them. In order to tory in northern Minnesota. study such long oscillation lengths, A high-intensity neutrino beam the detectors must be far away from from Fermilab will be made possible the source and can only intercept a by a new high-intensity 120 GeV pro- small fraction of the beam. Thus it The home of the MINOS detector—a ton source called the Main Injector. is necessary to make the beam very cavern in Soudan, Minnesota—being Scheduled for completion in 1999, intense at its origin. installed in the 1980s. this facility is being built to replace The far detector will be located in (Courtesy Fermilab) the present Main Ring as one stage an old iron mine beneath the Soudan of acceleration. The Main Injector State Park in Minnesota. A half mile will also allow a very high-intensity beneath the surface—at the deepest neutrino program, known as NuMI level of a historic iron mine—is the for “Neutrinos at the Main Injector,” existing Soudan 2 fine-grained de- to be run simultaneously with other tector. The mine, which operated for experiments . The intense proton almost one hundred years, is currently beam makes a neutrino beam by hit- being maintained for tourists by the ting a target to make the maximum State of Minnesota Department of number of pions and kaons, which Natural Resources. Scientists plan to are focused forward to give a beam bring ten thousand tons of iron to with as little divergence as possible. build the MINOS detector, which will Then they travel through a one kilo- join the one thousand tons already in meter pipe where many of them de- Soudan 2, to study neutrinos from cay into neutrinos, which continue Fermilab. It’s a bit like taking coal to moving forward. The kinematics of Newcastle.

BEAM LINE 15 The headframe atop a former iron mine at Soudan in northern Minnesota. (Courtesy Fermilab)

The new MINOS (for “Main In- new MINOS detector in 2002. Given jector Neutrino Oscillation Search”) all the other neutrino experiments detector will measure about twelve around the world, I can promise that thousand neutrino interactions per there will be substantial progress in year out of the five trillion that pass understanding neutrinos and the pos- through. It will consist of six hundred sibility of neutrino oscillations dur- layers each of scintillation counters ing the next decade. But I suspect and magnetized iron. If the atmos- that progress will come slowly and pheric neutrino deficit is due to νµ → gradually. The large and growing ντ oscillations, MINOS will observe effort being devoted to neutrino ex- different rates of events, different periments is indicative not only of fractions of events with muons, and the interest in these ghostly particles different energy distributions from but also the difficulty of doing pre- those seen in the near detector. cision work in this field. Even if we A small version of the MINOS de- definitively show that neutrino os- tector at Fermilab is a necessary part cillations exist, there will be a large of the experiment. This detector will set of neutrino mass and mixing be used to understand the beam and parameters to determine. And if neu- calibrate its intensity, by measur- trino oscillations do not turn up, we ing the interactions of neutrinos will need alternative explanations for before they have had any chance to the present observational hints. In oscillate into other species. one form or another, the experimen- Physicists hope to begin taking tal study of neutrino oscillations will data with the existing Soudan de- probably continue for the next tector and the first sections of the twenty years!

16 SPRING 1998