Oscillation Results From IceCube

Oscillation Results From IceCube Summer Blot Neutrino Telescopes Conference Padua, Italy 30 September 2025 Photo credit: NSF / Ilya Bodo 2 Introduction |να⟩=∑U*αk |νk⟩ Flavour Mass U = Ue1 Ue2 Ue3 Uμ1 Uμ2 Uμ3 Uτ1 Uτ2 Uτ3 •Neutrino flavors states are superposition of nonequal mass states •Mixing is governed by a unitary matrix U parameterized by: •Three mixing angles: θ12, θ13, θ23 •and potential CP-violating phase: δCP •Neutrinos accumulate phase differences as they propagate due to mass differences: Δm221, Δm232 •Leads to neutrino flavor oscillation as a function of L / E Unitarity implies, e.g., |Ue3|2 + |Uμ3|2 + |Uτ3|2 = 1 •Cosmic ray air showers produce abundant flux of νμ/anti-νμ and νe/anti-νe •These atmospheric neutrinos oscillate as they travel through the Earth •Can measure νμ disappearance (νμ→νμ) and ντ appearance (νμ→ντ) •Latter explicitly tests Uτ3 and unitarity 3 Atmospheric Neutrino Oscillations 12,714 km L∝ cosθzenith Δm232 θ23 P(νμ→ντ) ≈ sin2(2θ23)sin2(1.27 Δm232 L / E) 4 The IceCube Neutrino Observatory 1450 m 2450 m IceTop: cosmic ray science IceCube DeepCore Upgrade Astrophysics & neutrino properties } 1 km3 ice Cherenkov detector ~1 km 5 The IceCube Neutrino Observatory 1450 m 2450 m IceTop: cosmic ray science IceCube DeepCore Upgrade Astrophysics & neutrino properties Part 1: recent results Part 2: new detector status & prospects 1 km3 ice Cherenkov detector ~1 km } Tomorrow: Talk by Angela Zegarelli 6 IceCube DeepCore •5680 Digital Optical Modules (DOMs) embedded 1450-2450 m deep into the ice sheet •Detect Cherenkov photons from particle interactions •Spacing, quantum efficiency and ice properties drive detector performance Spacing [m] Approximate energy range Horiz. VerWcal IceCube 125 17 100 GeV – PeV DeepCore ~50 7 5 – 150 GeV 10” PMT HV control Digital electronics Calibration LEDs Glass pressure resistant sphere DeepCore IceCube Astropart.Phys. 35 (2012) 615-624 JINST 12 (2017) 03, P03012 DOM Tracks: Decent energy resolution Excellent direction/ pointing 7 NC (Z0) CC (W±) νe νe + N → νe + X νe + N → e- + X νμ νμ + N → νμ + X νμ + N → μ- + X ντ ντ + N → ντ + X ντ + N → τ- + X •Most interactions are in Deep Inelastic Scattering regime •Sensitive to all flavors, neutrino and anti-neutrinos, but cannot easily distinguish (no B-field) High-Energy neutrino signatures in IceCube Cascades: Good energy resolution Challenging direction/ pointing Tracks: Decent energy resolution Excellent direction/ pointing 8 High-Energy neutrino signatures in IceCube NC (Z0) CC (W±) νe νe + N → νe + X νe + N → e- + X νμ νμ + N → νμ + X νμ + N → μ- + X ντ ντ + N → ντ + X ντ + N → τ- + X •Most interactions are in Deep Inelastic Scattering regime •Sensitive to all flavors, neutrino and anti-neutrinos, but cannot easily distinguish (no B-field) Cascades: Good energy resolution Challenging direction/ pointing Tracks: Decent energy resolution Excellent direction/ pointing 9 High-Energy neutrino signatures in IceCube NC (Z0) CC (W±) νe νe + N → νe + X νe + N → e- + X νμ νμ + N → νμ + X νμ + N → μ- + X ντ ντ + N → ντ + X ντ + N → τ- + X •Most interactions are in Deep Inelastic Scattering regime •Sensitive to all flavors, neutrino and anti-neutrinos, but cannot easily distinguish (no B-field) Cascades: Good energy resolution Challenging direction/ pointing Double Bangs: Distinct cosmic ντ signature 16 Analysis signals νμ disappearance (νμ→νμ) ντ appearance (νμ→ ντ) Vary θ23 and Δm232 Maintain PMNS unitarity Vary θ23 and Δm232 + vary ντ normalization Deficit of tracks Excess of cascades 17 •Perform simultaneous fit of θ23 and Δm232 •Best fit θ23 is in upper octant, consistent with maximal mixing •Normal mass ordering is preferred •Dedicated NMO analysis ongoing •Excellent agreement between simulation and data, with p-value = 37.1 % •Result is independent of δCP, with minimal influence of cross-section systematics νμ disappearance (νμ→νμ) sin2θ23 = 0.516 Δm232 = 2.44 x 10-3 Best fit point: New! 18 νμ disappearance (νμ→νμ) •Compare measurement to expectation from Monte Carlo pseudo experiments •Mass splitting aligns well with expectation •Mixing angle is within 95% of trials, but more constraining than median expectation •Consistent with previous results •Continue to investigate potential causes •Final uncertainties will be calculated with Feldman-Cousins corrections, caution extracting too much from these plots! 19 ντ appearance (νμ→ ντ) •Additionally fit ντ normalization to test PMNS unitarity and cross-section •Best-fit: 1.11 +0.11/-0.10 consistent with standard expectation at 1σ • Goodness of fit p-value = 39.4% Standard expectation New! 20 Systematics uncertainties •Many sources of systematic uncertainty were studied •Detector calibration, atmospheric flux, muon rate, cross section, and other oscillation parameters •Include only impactful systematics as nuisance parameters in the fit with priors where applicable •No significant pulls with respect to priors were observed 21 Nuisance Parameter Pulls Detector & ice calibration contribute largest source of uncertainty, no priors! 22 The IceCube Upgrade detector 350 m 3 m 100 m Coming online in 2026 23 ~400 mDOMs ~300 DEggs 24 x 3” PMTs 2 x 8” PMTs + Increased photocathode area + Improved angular acceptance 350 m 3 m 100 m Coming online in 2026 New Optical Modules IceCube Gen2 TDR Next Generation Optical Sensor Performance 24 25 Improved ν detection •Increased sensor density in 2 Mton core •Lowers energy threshold down to ~1 GeV •Increased efficiency and improved resolution at GeV-scale 4 GeV up-going νμ with Upgrade DeepCore-only 32 Graph Neural Network - Energy resolution 33 Graph Neural Network - Zenith resolution 34 New atmospheric neutrino mixing analysis •Over 100k neutrinos in final sample •>5k charged current ντ! •Oscillation valley visible by eye in track-like sample from detector data Oscillation valley 35 •New result is consistent with previous IceCube DeepCore analyses •Somewhat expected due to overlap between samples and similar systemic treatments •Still, very different approaches to reconstruction and background rejection lead to similar results νμ disappearance (νμ→νμ) 36 Breakdown of event rate with ICUpgrade •Increased effective area at lower energies compared to DeepCore •Deep inelastic scattering events remain the dominant contribution •Larger relative contributions from resonant, quasi elastic and coherent scattering •Investigating potential for new systematic uncertainties using GENIE 37 Reconstruction performance with IC93 •Improved reconstruction, particularly at low energies 38 ICUpgrade: atmospheric neutrino mixing 39 NMO Sensitivity with IceCube Upgrade