Neutrinos as Lorentz Invariance Probes: Tighter Than the Planck Scale

Special relativity sits at the foundation of every relativistic quantum field theory we use. Lorentz invariance, the assumption that the laws of physics are the same for all uniformly moving observers, is so deeply embedded that it is often taken as self-evident. Yet there are theoretical reasons — coming from candidate theories of quantum gravity, from cosmologies with a preferred frame, and from string-theoretic constructions with spontaneously broken Lorentz symmetry — to suspect that Lorentz invariance is an emergent low-energy property rather than a fundamental one. If so, the breakdown would happen at some very high scale, plausibly the Planck mass $M_{\mathrm{Pl}}\approx 1.2\times 10^{19}$ GeV, and would leave only tiny residual effects at accessible energies.

The challenge for any experimental probe is to leverage some kind of amplification that magnifies a Planck-suppressed effect into something measurable. Neutrino oscillations turn out to be one of the most powerful amplifiers known. Oscillations are interferometric: they accumulate the phase difference between mass eigenstates over the propagation distance, so even a tiny modification to the dispersion relation of a particular flavor shows up as a measurable energy-dependent distortion of the oscillation pattern. Atmospheric neutrinos pumping through the Earth and astrophysical neutrinos travelling cosmological distances are natural Lorentz-invariance probes, and the limits IceCube has set on Lorentz-violating coefficients now reach below the Planck scale — a remarkable achievement for a detector that was built to do neutrino astronomy, not quantum gravity.

This post is about how Lorentz violation enters the neutrino sector, what the dominant phenomenological framework looks like, and where the current experimental bounds sit.

A simple form of Lorentz violation

The simplest way to imagine Lorentz violation in a neutrino is a modification of its dispersion relation. In Lorentz-invariant physics, a particle of mass $m$ and momentum $p$ has energy $E^2=p^2+m^2$. Adding a small correction yields

$E^2=p^2+m^2+\xi \frac{p^{n+2}}{M_{\mathrm{Pl}}^n}$

where $\xi$ is a dimensionless coefficient and $n$ controls how rapidly the correction grows with energy. For $n=1$ the correction is linear in $p/M_{\mathrm{Pl}}$; for $n=2$ it is quadratic. A different flavor of neutrino could in principle have a different coefficient $\xi$, in which case the propagation velocities of the mass eigenstates would differ by an energy-dependent amount on top of the ordinary mass-squared difference.

What matters for oscillations is the difference between coefficients of different mass eigenstates. The vacuum oscillation phase becomes

${\varphi }_{ij}=\frac{\Delta m_{ij}^2}{2E}L+{\xi }_{ij}^{(n)}{E}^{n}L/M_{\mathrm{Pl}}^n$

The first term decreases with energy — the standard oscillation. The second term increases with energy. At low energies the mass term dominates and one sees ordinary oscillation; at high energies the Lorentz-violating term takes over and the pattern shifts. Looking for an energy-dependent distortion of the oscillation probability is therefore the principal experimental handle.

The Standard Model Extension

A more systematic framework is the Standard Model Extension, or SME, which catalogues all possible Lorentz-violating operators that can be added to the Standard Model Lagrangian without spoiling gauge invariance. For neutrinos the relevant operators introduce two kinds of coefficients:

$\delta {\mathcal{L}}_{\nu }\supset -\frac{1}{2}{\bar{\psi }}_L\left[(a_L{)}_{\alpha \beta }^{\mu }{\gamma }_{\mu }+(c_L{)}_{\alpha \beta }^{\mu \nu }{\gamma }_{\mu }{\partial }_{\nu }\right]{\psi }_L+\text{h.c.}$

The $a_L$ coefficients have dimensions of energy and produce energy-independent (relative to the standard vacuum term) Lorentz-violating contributions to the Hamiltonian. The $c_L$ coefficients are dimensionless and produce contributions that grow linearly with neutrino energy. Both come with flavor and Lorentz indices, so they catalogue a large collection of distinct physical effects.

The advantage of the SME is that it provides a common language for all Lorentz-violation experiments. Atomic-clock measurements bound certain $a$ and $c$ coefficients in the electron and proton sectors; cosmic-ray observatories bound coefficients in the photon and quark sectors; neutrino oscillations bound coefficients in the lepton sector. Each experiment is sensitive to specific combinations, and the SME framework makes it possible to combine them into global constraints.

What atmospheric and astrophysical neutrinos see

The atmospheric neutrino flux at IceCube spans energies from roughly 100 GeV to several hundred TeV and arrives from all directions, with the deepest crossings traversing the full Earth diameter — about 12,700 km. The combination of high energy and long baseline gives an unmatched lever arm for detecting energy-dependent oscillation distortions.

In standard three-flavor oscillation theory, atmospheric muon-neutrino disappearance shows a clean characteristic energy dependence: oscillation is strongest at GeV energies and weakens as the energy rises and the oscillation phase falls below one. A Lorentz-violating term would produce a different energy dependence — typically, instead of disappearance fading at high energies, it could re-emerge with a modified pattern. The absence of any such re-emergence is what bounds the Lorentz-violating coefficients.

IceCube’s 2018 analysis of atmospheric muon-neutrino disappearance, using one year of high-energy starting events plus additional through-going tracks, placed limits on SME coefficients at the level of

• $c_{\mu \tau }^{TT}\lesssim 1.0\times 10^{-27}$ (dimensionless)
• $a_{\mu \tau }^T\lesssim 2.9\times 10^{-23}$ GeV

These are, very roughly, ten thousand times stronger than the natural Planck-suppressed expectations. For comparison, the natural expectation for a Planck-scale dimensionless coefficient is something like $M_W^2/M_{\mathrm{Pl}}^2\sim 10^{-34}$ — IceCube is not quite there yet, but it has decisively pushed below the most naive Planck-scale estimates that scale as $E/M_{\mathrm{Pl}}$.

Astrophysical neutrinos in IceCube’s sample, with energies up to several PeV travelling from kiloparsec to gigaparsec distances, provide an even longer lever arm. Their flavor ratios, currently consistent with a 1:1:1 mix from oscillation-averaged pion-decay sources, would be modified by Lorentz violation through changes in the effective propagation Hamiltonian. Limits from this sector are still developing but already constrain a complementary set of SME coefficients.

Astrophysical neutrinos as a long-baseline test

A separate line of attack uses the highest-energy neutrinos and the longest baselines available. Astrophysical neutrinos detected by IceCube at PeV energies travel from distant accelerators across megaparsecs to gigaparsecs. The product $E\times L$ that controls the sensitivity to a Lorentz-violating term is enormous — for a 1 PeV neutrino from a source 100 Mpc away, $E\times L\approx 10^{37}$ eV·s, dwarfing anything achievable on Earth.

The observable in this case is the arrival flavor ratio. As discussed in the previous post on IceCube’s tau neutrino detection, the standard expectation for a pion-decay source after oscillation-averaging is approximately ${\nu }_e:{\nu }_{\mu }:{\nu }_{\tau }\approx 1:1:1$ at Earth. A Lorentz-violating term that distinguishes the propagation of one flavor from another would push the ratio out of the standard allowed region — into a corner of the so-called flavor triangle that no astrophysical production model can reach.

The current IceCube tau-neutrino measurement is consistent with the standard ratio within uncertainties, and the constraint on Lorentz-violating coefficients from this measurement is competitive with the atmospheric one. Higher-statistics samples expected from IceCube-Gen2 and KM3NeT will sharpen the bound further.

Timing tests from supernova neutrinos

A complementary, independent test came from SN1987A. The neutrinos and the visible-light photons from that supernova travelled the 168,000 light-years to Earth within a few hours of each other, after a journey lasting essentially the same age as the supernova itself. Any energy-dependent difference in propagation velocity between photons and neutrinos would have spread the arrival times over time scales proportional to the propagation time, and the observed simultaneity placed a strong bound on the difference. The corresponding limit on a Lorentz-violating coefficient affecting neutrino propagation is around $10^{-9}$ at neutrino energies of tens of MeV — competitive with laboratory bounds at the time and obtained from a single astrophysical burst.

A future galactic supernova would extend this test by orders of magnitude. The next supernova in our galaxy, when it occurs, will likely be observed in multiple wavelengths and across multiple neutrino energy ranges, providing a clean baseline-energy combination to bound Lorentz violation freshly.

What it means

The current state of the field is: no Lorentz violation in the neutrino sector has been observed. The atmospheric, astrophysical and supernova-timing limits all sit at or below the naive Planck-scale expectation, which already tells us that if Lorentz violation does happen at the Planck scale, it must be more subtle than the simplest “linear in $E/M_{\mathrm{Pl}}$” estimates would suggest. This is itself useful information — it constrains the way Lorentz symmetry can be broken in quantum-gravity models.

The bounds will continue to tighten as IceCube accumulates more data, as the next generation of detectors comes online, and as new astrophysical neutrino sources are detected at higher energies. The eventual reach is set by how cleanly the relevant systematic uncertainties can be controlled — atmospheric neutrino flux modelling and cross-section uncertainties for the atmospheric sample; flavor identification efficiencies for the astrophysical sample. Steady progress on both fronts continues.

A Lorentz-violating signal, if it ever appears, would be among the most consequential results in modern physics. The current strong bounds notwithstanding, neutrinos remain among the cleanest places to look — and the analysis machinery is in place to make any future hint visible.

Summary

Lorentz invariance, the foundational symmetry of special relativity, may be broken at energies near the Planck scale by quantum-gravity physics. Neutrino oscillations are powerful probes because they accumulate phase interferometrically over enormous baselines, amplifying tiny modifications of the dispersion relation. The Standard Model Extension provides a systematic catalogue of Lorentz-violating operators; IceCube’s atmospheric-neutrino analysis bounds the leading dimensionless coefficients at the level of $10^{-27}$ and the leading dimensionful coefficients at the level of $10^{-23}$ GeV, decisively below naive Planck-scale expectations. SN1987A timing provides an independent test at lower energies. The combined constraints leave room for Lorentz violation that is more subtle than the simplest models predict, and they remain among the strongest empirical bounds on the physics of the Planck-scale frontier — derived not from gravitational experiments but from neutrinos travelling through the Earth.