Dark matter experiments are approaching the "neutrino floor." This is the point at which neutrinos from the Sun, which no shielding can stop, produce detector signals indistinguishable from dark matter. Past that point, collecting more data barely helps. The standard way through the floor is direction: the dark matter "wind" arrives from a fixed direction in the galaxy, while solar neutrinos come from the Sun. As Earth rotates, a detector whose response depends on direction sees the two sources modulate with different daily patterns. Today's low-threshold detectors have little or no directional response. The best example, a superconducting microwire that reached a 0.11 eV threshold in the QROCODILE search, owes its directional dependence only to the flat geometry of the film.
We propose a different source of direction. In a d-wave superconductor such as the cuprate Bi2212, the energy gap varies with the direction of electron momentum: it vanishes along four "nodal" lines and changes sign between neighbouring lobes. The energy gap is the minimum energy needed to break a bound electron pair, called a Cooper pair. When dark matter scatters off electrons and transfers momentum $\mathbf{q}$, whether pairs break efficiently depends on whether $\mathbf{q}$ connects regions of momentum space where the gap has the same sign or the opposite sign. The material's response therefore depends on the direction of $\mathbf{q}$ relative to the crystal axes, and that dependence is fixed at the microscopic scale.
Angle-resolved photoemission spectroscopy (ARPES) measures the energy and momentum of electrons knocked out of a crystal by light. It can now measure both the size and the sign of the gap across momentum space (Gao et al.). The core claim is that ARPES data are enough to build the directional response function of such a target. From that function one can quantify how much the gap's sign structure, beyond ordinary crystal anisotropy, improves separation of dark matter from neutrino backgrounds.
Core Claim
The quantity that decides directional power is the target's energy-loss function $W(\mathbf{q},\omega)$, the rate at which the material absorbs energy $\omega$ at momentum transfer $\mathbf{q}$. In a superconductor, the electron-density part of this response carries a weighting factor called a coherence factor:
$$ W(\mathbf{q},\omega) \propto \sum_{\mathbf{k}} \tfrac{1}{2}\left(1 - \frac{\xi_{\mathbf{k}}\xi_{\mathbf{k+q}} + \Delta_{\mathbf{k}}\Delta_{\mathbf{k+q}}}{E_{\mathbf{k}}E_{\mathbf{k+q}}}\right)\,\delta(\omega - E_{\mathbf{k}} - E_{\mathbf{k+q}}). $$
In this expression, $\xi_{\mathbf{k}}$ is the normal-state band energy and $\Delta_{\mathbf{k}}$ is the momentum-dependent gap. The quasiparticle energy is $E_{\mathbf{k}}=\sqrt{\xi_{\mathbf{k}}^2+\Delta_{\mathbf{k}}^2}$. Everything is then screened by the material's dielectric response, the same ingredient QROCODILE used.
The product $\Delta_{\mathbf{k}}\Delta_{\mathbf{k+q}}$ is what matters. When $\mathbf{q}$ joins opposite-sign lobes, the coherence factor is enhanced near threshold. When it joins same-sign lobes, the factor is suppressed. For sub-MeV dark matter moving at about $10^{-3}c$, momentum transfers reach roughly a keV, which is comparable to the Brillouin zone of a cuprate, the basic repeating cell of crystal momentum. Energy transfers reach up to about half an eV. So the node-to-node and lobe-to-lobe scattering vectors fall inside the kinematic window.
Our claim is testable: the gap sign adds a daily-modulation signature that a normal metal with the same Fermi surface lacks, and we can measure how large it is. If it turns out small, the conclusion is that ordinary anisotropic metals or geometry suffice, and that is a useful result too.
Research Program
The first stage builds the model of the target from measurements rather than from a toy model. We fit a tight-binding band structure, a simple model of electrons hopping between lattice sites, to ARPES maps of overdoped Bi2212. We then fit $\Delta_{\mathbf{k}}$ including its sign, using the band-hybridization constraints of Gao et al.. Those constraints are the shift in Fermi momentum from $0.023\pi/a$ to $0.018\pi/a$ and the band breaking at 14 meV binding energy. Earlier sign determinations, such as the Josephson-junction evidence summarized in Tsuei and Kirtley, give the overall d-wave symmetry but not a momentum-resolved gap. ARPES gives both, which is why the response can be built directly from data.
As an independent check of the model's anisotropy in the long-wavelength limit, we compare its predicted in-plane optical response against direction-dependent conductivity data of the kind reported for detwinned YBCO in the a-axis optical conductivity study. This is a consistency check, not a fit. A mismatch would mean that interactions not captured by the single-particle model control the response, and the program would stop there.
The second stage computes the signals. The dark matter–electron rate follows from $W(\mathbf{q},\omega)$ folded with the standard galactic halo and Earth's rotation, giving a rate as a function of time of day for a given crystal orientation. The neutrino side is harder. Coherent neutrino–nucleus scattering deposits energy in the electronic channel at these energies through the phonon-mediated Migdal effect, in which a recoiling nucleus jolts the lattice and the lattice in turn excites electrons. This effect has been formulated by Li and Xia for neutrinos and for dark matter, but only for an isotropic, single-element crystal resembling silicon. We extend that formalism to an anisotropic lattice whose electronic final states are the Bogoliubov quasiparticles of the ARPES-fitted model.
This extension also probes a tension the file leaves open. In isotropic silicon, the simpler free-recoil calculation agreed closely with the full phonon calculation even where it was expected to fail. We will check whether that agreement survives in an anisotropic, correlated lattice. Solar neutrinos then give a background whose daily pattern tracks the Sun, and reactor neutrinos give a fixed-direction calibration source. For the reactor case we reuse the Taishan-1 flux setup from Li and Xia, with 4.6 GW of thermal power at a 30 m baseline.
The third stage is the statistics. We compute the exposure needed to separate dark matter from the neutrino-induced background at 3σ using the full time-and-energy likelihood. We also compute the corresponding discovery limit on the scattering cross section, scanned over threshold. We report the threshold at which gap structure starts to matter, because it may sit below 0.11 eV. Gap-scale features in Bi2212 lie at tens of meV.
Baselines
The naive reference point is an isotropic target with the same density of states. Its daily modulation is zero by construction, so it gains nothing beyond exposure scaling. That is the regime CDEX-50 faces: Geng et al. project 5.1 × 10⁻⁴⁵ cm² at 5 GeV with a 160 eVee threshold and note that solar neutrinos become irreducible there. The low-mass reference from noble liquids is the PandaX-4T ionization-only analysis: the 259-day PandaX-4T search reports a dark matter–electron limit of 1.5 × 10⁻⁴¹ cm² at 200 MeV with a heavy mediator, with no directional capability. These figures set the scale of what isotropic scaling achieves. They are not direct competitors on the directional question.
The strongest existing method is geometric directionality in a conventional superconducting film, as in QROCODILE. That search reached a 0.11 eV threshold and observed 15 unexplained pulses in 415.15 hours. We found no reported result on Lacuna for QROCODILE's daily-modulation amplitude or for its power to discriminate dark matter from neutrinos. We will therefore compute that baseline ourselves, using the same thin-film dielectric-tensor treatment, the same halo model, the same threshold scan, and the same likelihood and exposure, so that only the source of anisotropy differs.
Two ablations isolate the key ingredient. The first keeps the ARPES-fitted Bi2212 band structure but replaces $\Delta_{\mathbf{k}}$ with an isotropic gap of equal average size. This removes the nodes and the sign change while keeping Fermi-surface anisotropy. The second uses the normal state, with zero gap. For this case we also evaluate a strongly anisotropic, non-superconducting metal: the altermagnet KV₂Se₂O, whose spin-split, anisotropic Fermi surfaces were mapped by ARPES in Jiang et al.. It serves as the "ordinary crystal anisotropy" competitor.
The proposal beats these baselines only if the full sign-changing model reaches 3σ separation with materially less exposure than both the isotropic-gap ablation and the best normal-anisotropic case. We set that bar in advance at a factor of two in exposure. A smaller gain would not justify the added difficulty of operating a cuprate target.
Why not just…?
Why not just keep scaling germanium and xenon? Exposure scaling stops paying once the neutrino background dominates, and CDEX-50 states this explicitly for its mass range. Ionization-only xenon analyses also suffer from micro-discharge and cathode backgrounds (PandaX-4T) and have no directional handle. Our isotropic reference point measures this ceiling directly.
Why not just use film geometry, as QROCODILE does? Geometry might be enough, and that is a real threat. A flat film's dielectric anisotropy is smooth in $\mathbf{q}$, whereas gap nodes and sign changes produce sharp, crystal-locked features near threshold. Which one discriminates better is an empirical question, and the like-for-like comparison in our baselines settles it. If geometry matches the cuprate within a factor of two, we will recommend geometry.
Why not just use any anisotropic crystal, with no superconductivity at all? This is the most serious alternative. Anisotropic Fermi surfaces and phonons alone give daily modulation, and they avoid the thermal quasiparticles that a gapless nodal superconductor produces. The isotropic-gap and altermagnet ablations exist to measure exactly this. If the gap sign contributes little beyond Fermi-surface shape, the proposal becomes a recommendation for anisotropic normal metals. The ARPES-based response pipeline remains useful in that case.
Why not just use the atomic Migdal effect in liquid xenon? At the momentum and energy transfers relevant here, the free-atom picture breaks down and lattice effects dominate, as argued in Li and Xia. Liquids also have no crystal axes that could carry a directional signature.
Decisive Checks
The program has three go/no-go points. First, the ARPES-built model must reproduce the measured in-plane optical anisotropy without refitting. Second, in the reactor-neutrino calibration geometry, the anisotropic Migdal calculation must predict an orientation dependence that a fixed-direction source could detect. If it does not, the neutrino background is effectively isotropic and only the dark matter side carries direction. Third, and most important, is the exposure ratio between the full sign-changing model and the isotropic-gap ablation at the lowest threshold we consider achievable. We will also state plainly the threshold that is required, since it is likely to lie below what current superconducting sensors reach. If the ratio is near one, the gap sign is irrelevant to directional detection, and we will report that negative result.
References
- First Sub-MeV Dark Matter Search with the QROCODILE Experiment Using Superconducting Nanowire Single-Photon Detectors
- ARPES Detection of Superconducting Gap Sign in Unconventional Superconductors
- Robust<mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline"><mml:msub><mml:mi>d</mml:mi><mml:mrow><mml:msup><mml:mi>x</mml:mi><mml:mn>2</mml:mn></mml:msup><mml:mo>−</mml:mo><mml:msup><mml:mi>y</mml:mi><mml:mn>2</mml:mn></mml:msup></mml:mrow></mml:msub></mml:math>Pairing Symmetry in Hole-Doped Cuprate Superconductors
- <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline"><mml:mi>a</mml:mi></mml:math>-axis optical conductivity of detwinned ortho-II<mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline"><mml:mrow><mml:mi mathvariant="normal">Y</mml:mi><mml:msub><mml:mi mathvariant="normal">Ba</mml:mi><mml:mn>2</mml:mn></mml:msub><mml:msub><mml:mi mathvariant="normal">Cu</mml:mi><mml:mn>3</mml:mn></mml:msub><mml:msub><mml:mi mathvariant="normal">O</mml:mi><mml:mn>6.50</mml:mn></mml:msub></mml:mrow></mml:math>
- Migdal effect of phonon-mediated neutrino nucleus scattering in semiconductor detectors
- Phonon-mediated Migdal effect in semiconductor detectors
- Projected WIMP sensitivity of the CDEX-50 dark matter experiment
- Search for Light Dark Matter with 259 Days of Data in PandaX-4T
- A metallic room-temperature d-wave altermagnet