Tuesday 5 June 2018

Can MiniBooNE be right?

The experimental situation in neutrino physics is confusing. One one hand, a host of neutrino experiments has established a consistent picture where the neutrino mass eigenstates are mixtures of the 3 Standard Model neutrino flavors νe, νμ, ντ. The measured mass differences between the eigenstates are Δm12^2 ≈ 7.5*10^-5 eV^2 and Δm13^2 ≈ 2.5*10^-3 eV^2, suggesting that all Standard Model neutrinos have masses below 0.1 eV. That is well in line with cosmological observations which find that the radiation budget of the early universe is consistent with the existence of exactly 3 neutrinos with the sum of the masses less than 0.2 eV. On the other hand, several rogue experiments refuse to conform to the standard 3-flavor picture. The most severe anomaly is the appearance of electron neutrinos in a muon neutrino beam observed by the LSND and MiniBooNE experiments.


This story begins in the previous century with the LSND experiment in Los Alamos, which claimed to observe νμνe antineutrino oscillations with 3.8σ significance.  This result was considered controversial from the very beginning due to limitations of the experimental set-up. Moreover, it was inconsistent with the standard 3-flavor picture which, given the masses and mixing angles measured by other experiments, predicted that νμνe oscillation should be unobservable in short-baseline (L ≼ km) experiments. The MiniBooNE experiment in Fermilab was conceived to conclusively prove or disprove the LSND anomaly. To this end, a beam of mostly muon neutrinos or antineutrinos with energies E~1 GeV is sent to a detector at the distance L~500 meters away. In general, neutrinos can change their flavor with the probability oscillating as P ~ sin^2(Δm^2 L/4E). If the LSND excess is really due to neutrino oscillations, one expects to observe electron neutrino appearance in the MiniBooNE detector given that L/E is similar in the two experiments. Originally, MiniBooNE was hoping to see a smoking gun in the form of an electron neutrino excess oscillating as a function of L/E, that is peaking at intermediate energies and then decreasing towards lower energies (possibly with several wiggles). That didn't happen. Instead, MiniBooNE finds an excess increasing towards low energies with a similar shape as the backgrounds. Thus the confusion lingers on: the LSND anomaly has neither been killed nor robustly confirmed.     

In spite of these doubts, the LSND and MiniBooNE anomalies continue to arouse interest. This is understandable: as the results do not fit the 3-flavor framework, if confirmed they would prove the existence of new physics beyond the Standard Model. The simplest fix would be to introduce a sterile neutrino νs with the mass in the eV ballpark, in which case MiniBooNE would be observing the νμνsνe oscillation chain. With the recent MiniBooNE update the evidence for the electron neutrino appearance increased to 4.8σ, which has stirred some commotion on Twitter and in the blogosphere. However, I find the excitement a bit misplaced. The anomaly is not really new: similar results showing a 3.8σ excess of νe-like events were already published in 2012.  The increase of the significance is hardly relevant: at this point we know anyway that the excess is not a statistical fluke, while a systematic effect due to underestimated backgrounds would also lead to a growing anomaly. If anything, there are now less reasons than in 2012 to believe in the sterile neutrino origin the MiniBooNE anomaly, as I will argue in the following.

What has changed since 2012? First, there are new constraints on νe appearance from the OPERA experiment (yes, this OPERA) who did not see any excess νe in the CERN-to-Gran-Sasso νμ beam. This excludes a large chunk of the relevant parameter space corresponding to large mixing angles between the active and sterile neutrinos. From this point of view, the MiniBooNE update actually adds more stress on the sterile neutrino interpretation by slightly shifting the preferred region towards larger mixing angles...  Nevertheless, a not-too-horrible fit to all appearance experiments can still be achieved in the region with Δm^2~0.5 eV^2 and the mixing angle sin^2(2θ) of order 0.01.     

Next, the cosmological constraints have become more stringent. The CMB observations by the Planck satellite do not leave room for an additional neutrino species in the early universe. But for the parameters preferred by LSND and MiniBooNE, the sterile neutrino would be abundantly produced in the hot primordial plasma, thus violating the Planck constraints. To avoid it, theorists need to deploy a battery of  tricks (for example, large sterile-neutrino self-interactions), which makes realistic models rather baroque.

But the killer punch is delivered by disappearance analyses. Benjamin Franklin famously said that only two things in this world were certain: death and probability conservation. Thus whenever an electron neutrino appears in a νμ beam, a muon neutrino must disappear. However, the latter process is severely constrained by long-baseline neutrino experiments, and recently the limits have been further strengthened thanks to the MINOS and IceCube collaborations. A recent combination of the existing disappearance results is available in this paper.  In the 3+1 flavor scheme, the probability of a muon neutrino transforming into an electron  one in a short-baseline experiment is
where U is the 4x4 neutrino mixing matrix.  The Uμ4 matrix elements controls also the νμ survival probability
The νμ disappearance data from MINOS and IceCube imply |Uμ4|≼0.1, while |Ue4|≼0.25 from solar neutrino observations. All in all, the disappearance results imply that the effective mixing angle sin^2(2θ) controlling the νμνsνe oscillation must be much smaller than 0.01 required to fit the MiniBooNE anomaly. The disagreement between the appearance and disappearance data had already existed before, but was actually made worse by the MiniBooNE update.
So the hypothesis of a 4th sterile neutrino does not stand scrutiny as an explanation of the MiniBooNE anomaly. It does not mean that there is no other possible explanation (more sterile neutrinos? non-standard interactions? neutrino decays?). However, any realistic model will have to delve deep into the crazy side in order to satisfy the constraints from other neutrino experiments, flavor physics, and cosmology. Fortunately, the current confusing situation should not last forever. The MiniBooNE photon background from π0 decays may be clarified by the ongoing MicroBooNE experiment. On the timescale of a few years the controversy should be closed by the SBN program in Fermilab, which will add one near and one far detector to the MicroBooNE beamline. Until then... years of painful experience have taught us to assign a high prior to the Standard Model hypothesis. Currently, by far the most plausible explanation of the existing data is an experimental error on the part of the MiniBooNE collaboration.

14 comments:

Luboš Motl said...

Your explanation must be an error by both LSND and MiniBooNE, right? How does it happen that they made errors with the compatible consequences - in two rather different experiments? What kind of an error is it? I also think it should be an error - these are my real questions, not rhetorical ones.

Unknown said...

Yes, both are probably wrong. The similarity between the two experiments is the lack of a near detector that would allow one to get a robust measurement of the background. Moreover, the two collaborations have a significant overlap (which may or may not be relevant for the error propagation). If I knew what the error is, our trouble would be over; certainly, the experimentalists involved are more competent than me in this respect. Sometime pi0 decays to photons are pointed out as the most difficult and suspicious background.

StevieB said...

I'm American, and I never knew Benjamin Franklin said that; I always learn something on this blog. Thanks, Jester!

andrew said...

The Daya Bay and Reno collaborations have both suggested strongly that the composition of the reactor fuel (which changes over time) is a likely source of the reactor anomalies observed at LSDN and MiniBooNE, something that the recent preprint from MiniBoonNE does not analyze rigorously or figure into its systemic error budget in a meaningful way.

A preprint from Daya Bay in 2017 (https://arxiv.org/abs/1704.01082) notes in its abstract that:

"The Daya Bay experiment has observed correlations between reactor core fuel evolution and changes in the reactor antineutrino flux and energy spectrum. Four antineutrino detectors in two experimental halls were used to identify 2.2 million inverse beta decays (IBDs) over 1230 days spanning multiple fuel cycles for each of six 2.9 GWth reactor cores at the Daya Bay and Ling Ao nuclear power plants. Using detector data spanning effective 239Pu fission fractions, F239, from 0.25 to 0.35, Daya Bay measures an average IBD yield, σ¯f, of (5.90±0.13)×10−43 cm2/fission and a fuel-dependent variation in the IBD yield, dσf/dF239, of (−1.86±0.18)×10−43 cm2/fission. This observation rejects the hypothesis of a constant antineutrino flux as a function of the 239Pu fission fraction at 10 standard deviations. The variation in IBD yield was found to be energy-dependent, rejecting the hypothesis of a constant antineutrino energy spectrum at 5.1 standard deviations. While measurements of the evolution in the IBD spectrum show general agreement with predictions from recent reactor models, the measured evolution in total IBD yield disagrees with recent predictions at 3.1σ. This discrepancy indicates that an overall deficit in measured flux with respect to predictions does not result from equal fractional deficits from the primary fission isotopes 235U, 239Pu, 238U, and 241Pu. Based on measured IBD yield variations, yields of (6.17±0.17) and (4.27±0.26)×10−43 cm2/fission have been determined for the two dominant fission parent isotopes 235U and 239Pu. A 7.8% discrepancy between the observed and predicted 235U yield suggests that this isotope may be the primary contributor to the reactor antineutrino anomaly."

The Reno Collaboration also points to fuel discrepancies as the source of the reactor anomaly. It's abstract to a June 2, 2018 preprint (https://arxiv.org/abs/1806.00574) states:

"We report a fuel-dependent reactor antineutrino yield using six 2.8\,GWth reactors in the Hanbit nuclear power plant complex, Yonggwang, Korea. This analysis uses an event sample acquired through inverse beta decay (IBD) interactions in identically designed near and far detectors for 1807.9 live days from August 2011 to February 2018. Based on multiple fuel cycles, we observe a fuel dependent variation in the IBD yield of (6.03±0.21)×10−43~cm2/fission for 235U and (4.17±0.29)×10−43~cm2/fission for 239Pu while a total average IBD yield per fission (y⎯⎯⎯f) is (5.79±0.11)×10−43~cm2/fission. The hypothesis of no fuel dependent IBD yield or identical spectra of fuel isotopes is ruled out at more than 6\,σ. The measured IBD yield per 235U fission shows the largest deficit relative to a reactor model prediction. Reevaluation of the 235U IBD yield per fission may mostly solve reactor antineutrino anomaly. We also report a hint of correlation between the 5\,MeV excess in observed IBD spectrum and the reactor fuel isotope fraction of 235U."

Unknown said...

You're confusing the MiniBooNE and reactor anomalies. These are different things.

Avelino said...

Great post, Jester! With so many news about the discovery of sterile neutrinos a more realistic opinion was very much needed.

Nicophil said...

The other lines of evidence have been weakening. For example, theorists have for years noted that nuclear reactors appear to produce about 6% fewer electron antineutrinos than standard theory predicts (suggesting that some of them were perhaps oscillating into sterile neutrinos). However, in April 2017, physicists with the Daya Bay Reactor Neutrino Experiment near Shenzhen, China, reported that the entire deficit could be explained if theorists had simply overestimated the number of antineutrinos produced by one component of the complex fuel, uranium-235. Ha ha !

Chris said...

StevieB, I like Jester's paraphrasing, but the actual quote is "In this world nothing can be said to be certain, except death and taxes."

Pedro said...

Just to add on the discussion, an underestimated π0 background may be responsible for MiniBooNE low energy excess, but LSND is completely different. Their signal is inverse beta decay, that is anti-nue plus proton to e+ plus neutron. The experimental signature consists of prompt 511 keV photons from e+e- annihilation, and delayed 2.2 MeV photon from neutron capture. So π0's cannot explain LSND...

milkshake said...

I wouldn't be so quick as dismissing LSND /MiniBooNE anomaly as error just because there is a poor fit with results from other experiments: the energy range is substantially different from solar/cosmic ray-generated neutrinos. And unlike the astronomy sources and reactor-produced antineutrinos, they have a precise control over the production rates of their muon neutrinos and the direction of the beam.

If it does not fit the Planck-measured microwave background, well it is too bad but don't forget that the data from microwave background is not something taken straight from the detector display - there is a long chain of analysis dependent on the chosen cosmology model, which does contain adjustable parameters "to make things come out right" and now this piece of data ruins everyone's favorite setting of those knobs...

Maybe there is a source of systematic error that affects LSND /MiniBooNE - or maybe all other experiments too, just to a different degree, and it wasn't noticed. But that in itself can be interesting. With regards to "most likely scenario is just plain vanilla SM" - neutrino oscillations are already an extension of SM, initially neutrinos were thought to me massless; maybe the current model is simplistic: what if there is more than one right-handed massive neutrinos...

John Duffield said...

Good post, Mad Hatter. IMHO there's an awful lot of inference when it comes to neutrinos, and not a lot of evidence. For example a lot of people take neutrino mass for granted, but I'm still waiting for somebody to show me a neutrino moving slower than light.

Unknown said...

That small window from LSND at around 1eV would remain if one just ignored the MiniBooNE result.

Unknown said...

I remain astonished that muon neutrino fluxes are so well-determined at high energy that a 1% loss of flux can be ruled out at the several sigma level as that should be the criterion for contradicting LSND and MiniBooNE. I am also astonished that particle physicists continue to take astrophysical constraints so seriously, as they depend on being certain in detail about 15 billion years of physics history which has undergone a number of significant revisions since the last millennium. (That's a reaction to the blog's historical reference style.) Consistency is great but can anyone rule out intermediate epochs of dark energy style acceleration of the expansion rate? The conclusions (oft revised despite Mike Turner's assurances) depend on knowing that we do know the all of the possible variations from the current best fit model. One must always be skeptical but I recall Felix Boehm's long-held certainty that neutrinos were massless and could not oscillate. Besides, if any dark matter is spinorial in nature, we are assured that sterile neutrinos do exist -- and likely couple somehow.

Ara Ioannisian said...

arXiv:1909.08571
A Standard Model explanation for the excess of electron-like events in MiniBooNE

ABSTRACT:
We study the dependence of neutral current (NC) neutrino-induced π0/photon production (νμ+A→νμ+1π0/γ+X) on the atomic number of the target nucleus, A, at 4-momentum transfers relevant to the MiniBooNE experiment: Δ resonance mass region. Our conclusion is based on experimental data for photon-nucleus interactions from the A2 collaboration at the Mainz MAMI accelerator. We work in the approximation that decays of Δ resonance unaffected by its production channel, via photon or Z boson. 1π0+X production scales as A2/3, the surface area of the nucleus. Meanwhile the photons created in Δ decays will leave the nucleus, and that cross section will be proportional to the atomic number of the nucleus. Thus the ratio of photon production to π0 production is proportional to A1/3. For carbon 12C this factor is ≈2.3. MiniBooNE normalises the rate of photon production to the measured π0 production rate. The reduced neutral pion production rate would yield at least twice as many photons as previously expected, thus significantly lowering the number of unexplained electron-like events.