
5
are relatively less affected compared to νeand νµ. In our
forecasts, we account in detail for the in-Earth propaga-
tion of UHE neutrinos from the surface of the Earth to
the radio array of IceCube-Gen2. Below, we elaborate.
At neutrino energies above a few GeV, the leading neu-
trino interaction channel is neutrino-nucleon (νN ) deep
inelastic scattering (DIS) [76–78]. In it, a neutrino scat-
ters off of one of the partons, i.e., a quark or a gluon,
inside a nucleon, N,i.e., a proton or a neutron. The
final-state parton promptly hadronizes into final-state
hadrons, X. A neutral-current (NC) DIS interaction,
mediated by a Zboson, produces in addition a final-
state neutrino, i.e.,να+N→να+X(α=e, µ, τ ). A
charged-current (CC) DIS interaction, mediated by a W
boson, produces in addition instead a final-state charged
lepton, i.e.,να+N→lα+X. The νN DIS cross sec-
tion has been measured at sub-TeV neutrino energies by
a variety of accelerator neutrino experiments [79–96]; in
the few-TeV range, by FASER [97] (and the upcoming
FASERν[98]), and in the TeV–PeV range, using Ice-
Cube data [99–101]. At higher energies, the cross section
is predicted [102–111] and may be measured in upcoming
UHE neutrino telescopes [51,112–114].
Computing the UHE νN DIS cross section requires
knowing the parton distribution functions in protons and
neutrons, which are measured in lepton-hadron collisions,
and extrapolating them beyond the regime where they
have been measured. (Concretely, it requires extrapolat-
ing them to values of Bjorken-x—the fraction of nucleon
momentum carried by the interacting parton—beyond
the measured ones.) At ultra-high energies, the NC and
CC νN cross sections, σNC
νN and σCC
νN , respectively, grow
roughly ∝E0.36
ν[103], are essentially equal for all flavors
of ναand ¯να, and σNC
νN ≈σCC
νN /3. Below, to produce
our forecasts, we adopt the state-of-the-art BGR18 cal-
culation of the νN DIS cross sections [111] in the prop-
agation and detection of neutrinos. The BGR18 is built
using recent experimental results and sophisticated next-
to-leading-order calculations, including the major correc-
tions described in Appendix B4 of Ref. [111]; for details,
see Ref. [111] andRef. [115], for a summary, see Ref. [51].
In a DIS interaction, the final-state hadrons receive
a fraction y—the inelasticity—of the neutrino energy,
and the final-state lepton receives the remaining frac-
tion (1 −y). In each interaction, the value of yis ran-
domly sampled from a probability density that is propor-
tional to the differential DIS cross sections, dσNC
νN /dy and
dσCC
νN /dy. At the energies relevant for our work, the aver-
age value of yis about 0.25 [102]. However, because the
distribution of values of yhas a large spread (see Fig. 4
in Ref. [51]), when propagating neutrinos through the
Earth below (and also when computing the event rates
that they induce, in Section IV D), we do it by using the
distributions of y, separately for NC and CC DIS, rather
than by using its average value.
Inside the Earth, NC interactions shift the UHE neu-
trino flux to lower energies, by regenerating lower-energy
neutrinos, while CC interactions dampen the flux alto-
gether, by replacing neutrinos with charged leptons. The
one exception is the CC interaction of ντ: in them, the
final-state tauon may propagate for some distance inside
the Earth before decaying and generating a new, high-
energy ντ. As a result of this “ντregeneration,” the flux
of ντis less attenuated than that of νeand νµ.
The severity of the effects of in-Earth propagation on
the neutrino flux varies with neutrino energy, Eν, and
direction, expressed via the zenith angle, θz, measured
from the South Pole, where IceCube-Gen2 will be lo-
cated. Higher energies and directions corresponding to
longer path lengths inside the Earth yield more severe
effects. To illustrate this, we use a simplified calculation
of the number of neutrino-induced events in the detector,
Nsimp
ν, similar to the one in Ref. [100], i.e.,
Nsimp
ν(Eν, θz)∝Φν(Eν)σνN (Eν)e−L(θz)/LνN (Eν,θz),
(1)
where Φνis the neutrino flux at the surface of the Earth,
σνN is the νN cross section (for this simplified calcula-
tion, it is the sum of NC and CC cross sections), L(θz) =
qR2
⊕−2R⊕dcos2θz+ 2R⊕d−(R⊕−d) cos θzis the
distance traveled inside the Earth by a neutrino with in-
coming direction θz, where R⊕= 6371 km is the radius
of Earth, dis the detector depth, approximately 200 m
for the radio array of IceCube-Gen2, LνN ≡(σνN nN)−1
is the neutrino mean free path inside the Earth along this
direction, and nNis the average number density of nu-
cleons along this direction, based on knowledge of the in-
ternal matter density of Earth (more on this later). (We
use Eq. (1) only for illustration; later we describe the
detailed calculation with which we produce our results.)
Equation (1) accounts for flux attenuation during in-
Earth propagation, via the exponential dampening term,
but ignores the regeneration of lower-energy neutrinos.
Even so, it embodies essential features of the propaga-
tion and detection of high-energy and ultra-high-energy
neutrinos. Upgoing neutrinos (cos θz<0), i.e., neutrinos
that reach the detector from below after traveling under-
ground a distance of up to the diameter of the Earth, are
more strongly attenuated than downgoing (cos θz>0)
and horizontal neutrinos (cos θz≈0). For UHE neutri-
nos, the attenuation is so strong that virtually no upgoing
neutrinos reach the detector (see Fig. A2 in Ref. [100]),
unless the neutrino flux at the surface is extraordinarily
large; e.g., benchmark flux model 4 in Figs. 2,3, and 4.
This means that our forecasts below, which factor in the
contribution of neutrinos from all directions, are driven
primarily by downgoing and horizontal neutrinos.
Further, Eq. (1) shows that while flux attenuation is
∝e−σνN , the rate of neutrino interactions in the detec-
tor is ∝σνN . The interplay between these competing
effects is accentuated at high energies, where the cross
section is larger: a larger cross section makes the already
tiny flux of upgoing neutrinos vanish, which has little
marginal effect, but it appreciably increases the number
of downgoing and horizontal neutrinos detected.
Finally, Eq. (1) reveals important nuance in the rate