Skip to content

Optics: media and wavelength

Media

A detector's medium is the material of its geometry config, read from config/materials/<material>.json. Three ship:

material file contents
water config/materials/water.json Rayleigh and Mie scattering and absorption from the SK calibration (arXiv:1307.0162); above 464 nm the absorption follows Pope & Fry (1997), blended smoothly across the join. Refractive index 1.33
WbLS config/materials/wbls.json 10% water-based liquid scintillator: water's bulk optics plus a scintillation block (below)
ice config/materials/ice.json a placeholder: water's functional forms with ice coefficients and index 1.309. Fine for geometry and workflow studies, not for ice-optics precision

All three use the water SIREN emitters ("siren": "water"). The refractive index is one constant per medium: the Cherenkov angle and the light speed c/n used for timing have no dispersion.

Scalar and wavelength mode

setup_event_simulator(..., wavelength_mode=...) picks one of two regimes:

  • wavelength_mode=False (scalar): every photon sees the same optics, the scalar fields of DetectorParams (scatter_length, mie_scatter_length, absorption_length, qe, ...). Faster, and enough when the wavelength dependence is not the question.
  • wavelength_mode=True (the default): each photon carries a wavelength, and scattering, absorption and QE are evaluated for it from the medium's curves and the PMT's QE curve. The medium covers 300 to 700 nm.

A physics config that omits a scalar optical length, g or qe gets it from the medium and QE curves at 400 nm, so the two regimes agree there. The optical lengths given under scalar_mode in a physics config are used only in scalar mode.

Reference and deviation

In wavelength mode, each optical property is a fixed reference curve (from the medium or QE file) times a fittable deviation curve in DetectorParams: rayleigh_dev, mie_dev, abs_dev and qe_dev. Each has one value per control wavelength (337, 375, 398, 405 and 445 nm, the SK laser lines), interpolated to the photon's wavelength and held flat outside that range. The deviation multiplies the interaction strength:

qe(λ) = qe_ref(λ) × dev(λ)
L(λ)  = L_ref(λ) / dev(λ)      # Rayleigh, Mie and absorption lengths

All-ones deviations reproduce the reference, so calibration fits departures from it.

QE

qe in DetectorParams is the QE at 400 nm, for light arriving at the PMT with its reflection included, as PMT QE is quoted. In wavelength mode each photon is weighted by qe × qe_fn(λ) / qe_fn(400 nm), where qe_fn is the QE curve named by the physics config's qe_curve (config/pmt/; the SK curve spans 294 to 648 nm). The photon step reflects part of the light off the PMT and converts the rest at QE / (1 − R0), with R0 the sensor's normal-incidence reflectance, so the net detection is the quoted QE. Setup rejects parameters with a peak QE above 1 − R0.

How wavelengths are sampled

Track sources sample the Cherenkov 1/λ² spectrum over the emitter's emission_band_nm, the band its photon yield N_photons(E) was counted over (274.91 to 673.83 nm for the shipped emitters). Photons outside the QE curve or the medium range count nothing, as they should. cherenkov_emission_band overrides the band.

When LUCiD samples wavelengths (a track, or a calibration source with wavelength=None), wavelength_sampling decides how:

  • 'cherenkov' (default): sample 1/λ² and weight each photon by qe_fn(λ).
  • 'cherenkov_qe': sample QE(λ)/λ² directly, so every photon has the same weight. Lower variance for an expected forward, but the QE is then inside the sampling distribution and a gradient with respect to the QE curve has nothing to act on. Do not use it when fitting QE or qe_dev. It is rejected with wavelength_mode=False, without a QE curve, and for source='photons'.

WbLS scintillation

Scintillation needs wavelength_mode=True; setup raises for a scintillating medium in scalar mode. The physics config inherits S, kB, C, tau_rise and tau_fall from the material file unless it sets them.

Light yield, Chou form (Birks with C = 0):

dL/dx = S · (dE/dx) / [1 + kB · (dE/dx) + C · (dE/dx)²]

The bundled file is 10% WbLS: S = 1387.3 photons/MeV, kB = 1.65e-5 mm/keV, C = 1.33e-9 mm²/keV².

  • S is linear in the WbLS mass fraction c over 1 to 10%: S(c) = (127.9 ± 17.0)·c + (108.3 ± 51.0) photons/MeV (Caravaca 2020, arXiv:2006.00173; CHESS, cosmic muons and ⁹⁰Sr, scintillation separated from Cherenkov).
  • kB and C were measured for 5% WbLS on 2 to 20 MeV proton recoils (Callaghan 2023, arXiv:2210.03876, Table 6; oxygenated sample):
form kB C
Chou (1.65 ± 0.81) × 10⁻⁵ mm/keV (1.33 ± 0.27) × 10⁻⁹ mm²/keV²
Birks only (5.95 ± 0.43) × 10⁻⁵ mm/keV —

The S-kB correlation in the Chou fit is 87.4% (full covariance in Callaghan 2023, Table 7).

Emission time. The measured profile has two decay components with a shared rise time:

p(t) = R₁ · g(t; τ_r, τ₁) + (1 − R₁) · g(t; τ_r, τ₂)
g(t; τ_r, τ_d) = (e^(−t/τ_d) − e^(−t/τ_r)) / (τ_d − τ_r),   t ≥ 0
parameter 1% WbLS 5% WbLS 10% WbLS
τ_r [ns] 0.00 ± 0.06 0.06 ± 0.11 0.13 ± 0.12
τ₁ [ns] 2.25 ± 0.15 2.35 ± 0.13 2.70 ± 0.16
τ₂ [ns] 15.10 ± 7.47 23.21 ± 3.28 27.05 ± 4.20
R₁ 0.96 ± 0.01 0.94 ± 0.01 0.94 ± 0.01

Uncertainties are statistical plus systematic (Caravaca 2020, Table 2; ⁹⁰Sr betas). LUCiD simulates the fast component only: one rise and one fall time (tau_rise = 0.13, tau_fall = 2.7 ns for 10%). The slow component, 6% of the light, is not simulated.

The emission spectrum is a Moyal distribution, given in the material file.