A detector for very light dark matter needs a material that absorbs tiny amounts of energy. That means an electronic energy gap far below the 1.2 eV of silicon, ideally at or near zero. Some crystals have exactly such gapless electronic states, and they are guaranteed by symmetry. In these materials, a "nonsymmorphic" symmetry, meaning a reflection or rotation combined with a half-step shift of the lattice, forces electron bands to touch along whole lines or planes in momentum space. That symmetry guarantee has a catch. It forbids a gap from ordinary single-electron terms, but it does not stop electron–electron interactions from breaking the symmetry on their own and opening a gap. When that happens, the gap is set by "dimensional transmutation": an energy scale that appears on its own, growing out of how an interaction strength changes logarithmically with energy. BaVS₃ is the standing warning. Its symmetry-enforced node surface disappears at a structural transition near 70 K (Node-surface and node-line fermions from nonsymmorphic lattice symmetries).
The central claim of this note is that one simple property decides whether a symmetry-protected nodal material is a usable zero-threshold target. That property is the dimension of the set of momenta where bands are forced to touch. A two-dimensional node surface, found in BaVS₃ without spin–orbit coupling, has a finite density of states at zero energy. Interactions in the symmetry-breaking channel are then marginal, like a Peierls or BCS instability, so a dynamically generated gap is guaranteed even at weak coupling. A one-dimensional node line, found in BaTaS₃ with spin–orbit coupling and in SrIrO₃ protected by an n-glide, has a density of states that vanishes linearly at zero energy. Short-range interactions are then irrelevant at low energy, and the material stays gapless until the coupling passes a finite critical value. We propose to turn this distinction into a quantitative target-selection rule. For the node-line materials that pass the test, we will compute the first dielectric response functions and dark-matter scattering rates for any nonsymmorphic target. We will also exploit a feature no fixed-gap detector has: strain can reversibly break the glide and open a controlled gap of a few meV (Strain-induced nonsymmorphic symmetry breaking and removal of Dirac semimetallic nodal line in an orthoperovskite iridate).
Core Claim
Let $\nu(E)$ be the electronic density of states near the nodes and $g$ the dimensionless interaction strength in the channel that breaks the protecting symmetry. For a node surface, $\nu(E)\to\nu_0>0$. The one-loop flow of $g$ is then logarithmic, and the self-generated gap is
$$\Delta_{\rm dyn}\simeq \Lambda\, e^{-1/(\nu_0 g)},$$
where $\Lambda$ is the bandwidth of the nodal bands. This is the standard marginal-coupling result, and it is nonzero for any attractive $g>0$.
For a node line, $\nu(E)\propto E$. The coupling then has a finite critical value $g_c$, and
$$\Delta_{\rm dyn}=0 \ \ (g<g_c), \qquad \Delta_{\rm dyn}\propto \Lambda\,(g-g_c)^{\nu} \ \ (g>g_c),$$
where $\nu$ is the correlation-length exponent of the resulting quantum critical point. The node-line case is the three-dimensional analogue of the (2+1)-dimensional Dirac problem in Dynamical generation of topological masses in Dirac fermions. There, quantum Monte Carlo on a symmorphic checkerboard lattice found interaction-driven mass generation only beyond a critical coupling, with $\nu=0.85(2)$ in the N=8 chiral Ising class.
The detector-relevant consequence follows directly. Spin–orbit coupling shrinks the BaVS₃ node surface to BaTaS₃ node lines (Node-surface and node-line fermions from nonsymmorphic lattice symmetries). In doing so, it converts a material that is guaranteed to gap into one that is protected twice: by symmetry, and by the irrelevance of weak interactions. Long-range Coulomb repulsion in a node line is marginal rather than irrelevant, and it must be treated separately. It renormalizes the band velocity logarithmically, but in the charge channel it does not break the glide. We will check this explicitly.
Relation to Prior Work
This note draws on three kinds of prior work.
The dimensional-transmutation literature supplies the method. Spontaneous symmetry breaking at two loop in 3-d massless scalar electrodynamics shows that one loop can leave symmetry breaking ambiguous while two loops settle it. For us, this means one-loop estimates of $g_c$ need a non-perturbative cross-check, which we get from sign-problem-free quantum Monte Carlo in the style of Dynamical generation of topological masses in Dirac fermions. Neither of these works includes a nonsymmorphic lattice. Neither asks how the projective, momentum-dependent phases of glides and screws restrict which order-parameter channels can gap the nodes. That restriction is our addition.
The nonsymmorphic literature motivates the proposal. Node-surface and node-line fermions from nonsymmorphic lattice symmetries establishes the node surface in BaVS₃, the node lines with Berry phase π in BaTaS₃, and the 70 K instability of BaVS₃. We read that instability as evidence for our node-surface prediction rather than as a nuisance. Strain-induced nonsymmorphic symmetry breaking and removal of Dirac semimetallic nodal line in an orthoperovskite iridate shows that a 0.4° monoclinic shear in SrIrO₃ breaks only the n-glide and opens a gap of about 5 meV. This gives an experimentally demonstrated gap control that we reuse as a switch. Dirac semimetal and topological phase transitions in A3Bi provides point-node Dirac semimetals protected by rotation, which we use as a comparison class.
The light-dark-matter literature supplies the detector machinery and constrains what we can claim. DarkELF computes scattering rates from the energy loss function (ELF), $\mathrm{Im}[-1/\epsilon(\mathbf q,\omega)]$, the quantity that measures how strongly a material absorbs energy $\omega$ at momentum $\mathbf q$. It does so only for Si, Ge, Al₂O₃, GaAs, GaN, ZnS, SiO₂ and Al. Migdal Effect in Semiconductors ties the Migdal rate to the ELF. The Migdal effect is ionization caused when a struck nucleus jolts its own electron cloud. Knapen et al. also warn that dropping the off-diagonal "local-field" terms of the dielectric matrix causes O(1) errors. Describing the Migdal effect with a bremsstrahlung-like process and many-body effects supplies the random-phase-approximation workflow with Quantum ESPRESSO and YAMBO, and a validation fingerprint, a 13 eV feature in diamond. Phonon-mediated Migdal effect in semiconductor detectors explicitly suggests Dirac materials as targets but computes nothing for them. These Migdal papers also constrain us. The free-recoil picture fails below roughly 50–70 MeV, and Migdal effect of Phonon-mediated neutrino nucleus scattering in semiconductor detectors stresses that the Migdal effect has never been directly observed. For both reasons, our headline channel is direct dark matter–electron scattering. Migdal rates appear only above 70 MeV and are labelled as conditional on the effect existing.
Research Program
The work has three stages, each gating the next.
The first stage is a symmetry and renormalization-group analysis for the space groups P6₃/mmc (BaMX₃) and Pbnm (SrIrO₃). We will enumerate the order-parameter channels whose condensation can gap the nonsymmorphic nodes, using the representation theory of Constructions and applications of irreducible representations of spin-space groups where magnetic channels arise. For each channel, we will derive the one-loop flow of the coupling from Wannier-projected DFT bands, following the VASP workflow of the BaMX₃ paper. The outputs are $\nu_0 g$ for BaVS₃ and $g/g_c$ for the node-line materials. We will then run quantum Monte Carlo on a minimal glide-symmetric node-line lattice model to test whether the one-loop $g_c$ survives, since the two-loop lesson says it might not.
The second stage computes the ELF of BaTaS₃ and of SrIrO₃ in both its glide-preserving Pbnm and glide-broken P112₁/m structures. We will use DFT plus the random phase approximation in YAMBO, with the full dielectric matrix $\epsilon_{\mathbf G\mathbf G'}$, focusing on energy transfers from 1 meV to 1 eV. These ELFs are then fed into DarkELF's rate routines for dark matter–electron scattering with heavy and light mediators, and for Migdal above 70 MeV.
The third stage computes the directional daily modulation. The node lines run along specific high-symmetry paths, such as A–L in BaTaS₃, so the ELF should be strongly anisotropic. As the Earth rotates, the dark-matter wind sweeps across the crystal axes, producing a daily modulation whose phase is fixed by the crystal orientation.
Baselines
The naive reference point is a free-electron gas with the same carrier density, computed with DarkELF's Lindhard and Mermin modules. It shares the zero threshold but has no nodal structure. If our nodal-line ELF gives the same projected reach as this model, the nonsymmorphic physics is irrelevant to detection.
The strongest existing sub-MeV competitor is QROCODILE (First Sub-MeV Dark Matter Search with the QROCODILE Experiment Using Superconducting Nanowire Single-Photon Detectors). It reports a 0.11 eV threshold, sensitivity down to $m_\chi\approx 30$ keV, and 15 non-periodic pulses in a 415.15-hour run. The file gives no numerical cross-section limit for it, so we report "no reported result found on Lacuna for the QROCODILE cross-section limit." The comparison will therefore use its threshold and observed rate as inputs to a projection at matched exposure and matched background.
For masses near 200 MeV, the conventional-detector reference is the PandaX-4T dark matter–electron limit of $1.5\times10^{-41}$ cm² for a heavy mediator (Search for Light Dark Matter with 259 Days of Data in PandaX-4T). The semiconductor reference is DarkELF's Si and Ge reach at identical exposure. All projections use the same halo model, the same 1 kg·yr exposure, and the same assumed flat background. Our targets do not tune any parameter the baselines lack.
Beating the baselines means two things. First, the projected reach in the node-line target must exceed the Lindhard model by more than the spread between DFT functionals. Second, the node-line target must extend reach below 30 keV, which QROCODILE's threshold cannot reach.
Two ablations isolate the key ingredient. The first compares glide-intact Pbnm SrIrO₃ with glide-broken P112₁/m SrIrO₃ at identical composition, which isolates nonsymmorphic protection. The second compares ELFs computed with and without local fields, which tests whether the O(1) caveat of Knapen et al. is larger in anisotropic nodal crystals. A symmorphic-protection comparison with Na₃Bi separates line-node phase space from point-node phase space.
Why not just…?
Why not just compute the ELF of SrIrO₃ and BaTaS₃ and skip the renormalization-group analysis? That calculation is stage two of this program, and it is the obvious filling of Gap 1. But a DFT ELF for a nodal material describes a ground state the real material may not have. A DFT calculation of BaVS₃ would report a gapless target, yet the material gaps near 70 K. The instability analysis decides which ELFs are physically meaningful at detector temperatures, so it cannot be skipped.
Why not just use a superconducting sensor like QROCODILE, which already works at 0.11 eV? This is a real threat. If the node-line reach at 30 keV–1 MeV does not beat the matched QROCODILE projection, the target class is not worth pursuing, and we will say so. What a superconductor lacks is a reversible, symmetry-selective gap. Opening the SrIrO₃ gap by strain should remove events below about 5 meV in a way the ELF predicts, while leaving detector noise that is unrelated to the electronic spectrum unchanged. That gives an in-situ background subtraction. Our rate calculations will quantify how much signal this subtraction retains.
Why not just use a symmorphic Dirac semimetal such as Na₃Bi? Its protection is rotational, and its nodes are points. Point nodes have fewer low-energy states than lines, and any strain that lowers the rotation symmetry gaps them without selectivity. The Na₃Bi baseline tests directly whether line-node phase space and glide-selective gapping change the projected reach. If they do not, the nonsymmorphic ingredient adds nothing for detection.
Decisive Checks
First, the pipeline must reproduce DarkELF's Si ELF and the 13 eV diamond feature before any new material is trusted.
Second, the node-surface formula must give a BaVS₃ transition scale of the right order. A temperature of 70 K corresponds to roughly 6 meV. If the estimate misses this by orders of magnitude, the one-loop couplings are not trustworthy for BaTaS₃ either.
Third, BaTaS₃ and Pbnm SrIrO₃ must come out at $g<g_c$, confirmed by quantum Monte Carlo. If they are above critical, the double-protection claim fails and the materials drop out.
Fourth, the glide ablation must show that the change in rate between Pbnm and P112₁/m is confined below the 5 meV gap, with a predicted spectral shape. That shape is the signature a future strain-switched detector would test.
References
- Node-surface and node-line fermions from nonsymmorphic lattice symmetries
- Strain-induced nonsymmorphic symmetry breaking and removal of Dirac semimetallic nodal line in an orthoperovskite iridate
- Dynamical generation of topological masses in Dirac fermions
- Spontaneous symmetry breaking at two loop in 3-d massless scalar electrodynamics
- Dirac semimetal and topological phase transitions in<mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline"><mml:msub><mml:mi>A</mml:mi><mml:mn>3</mml:mn></mml:msub></mml:math>Bi (<mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline"><mml:mrow><mml:mi>A</mml:mi><mml:mo>=</mml:mo><mml:mtext>Na</mml:mtext></mml:mrow></mml:math>, K, Rb)
- python package for dark matter scattering in dielectric targets
- Migdal Effect in Semiconductors
- Describing the Migdal effect with a bremsstrahlung-like process and many-body effects
- Phonon-mediated Migdal effect in semiconductor detectors
- Migdal effect of phonon-mediated neutrino nucleus scattering in semiconductor detectors
- Constructions and applications of irreducible representations of spin-space groups
- First Sub-MeV Dark Matter Search with the QROCODILE Experiment Using Superconducting Nanowire Single-Photon Detectors
- Search for Light Dark Matter with 259 Days of Data in PandaX-4T