Heavy-Element Paleodetectors in Ancient Minerals

• Heavy-element paleodetectors are ancient mineral-based solid-state track detectors that use heavy nuclei to enhance dark matter scattering sensitivity.
• They employ nanoscale imaging techniques to resolve latent nuclear recoil tracks accumulated over millions to billions of years, leveraging coherent scattering properties.
• Optimization challenges include balancing electronic stopping power, radiopurity, and microphysical track formation to accurately extract recoil energy and inelastic scattering signals.

Heavy-element paleodetectors are paleo-detectors implemented in ancient minerals whose target nuclei are sufficiently heavy to enhance coherent scattering and, in inelastic scenarios, to overcome kinematic thresholds for scattering. In the broad paleodetector program, insulators or poor semiconductors preserve latent damage tracks from nuclear recoils over geological timescales; heavy-element variants emphasize minerals containing high-mass nuclei such as lead, while related work on ultra-basic rocks such as olivine establishes the microphysics of track formation and readout in candidate solids (Graham et al., 3 Jun 2026, Drukier et al., 2018). The resulting detectors are time-integrating solid-state track detectors rather than real-time instruments, with sensitivity controlled by recoil kinematics, stopping powers, mineral radiopurity, annealing stability, and nanoscale three-dimensional microscopy (Baum et al., 2021).

1. Concept and detection principle

Heavy-element paleodetectors inherit the general paleo-detector concept: a recoiling nucleus produced by dark matter or neutrino scattering deposits energy in a solid and creates a permanent damage track of broken bonds, provided the local energy deposition exceeds a material-dependent threshold and the mineral’s annealing time far exceeds its age (Drukier et al., 2018). In this framework, the detector is the mineral itself, and the measurement is performed by post hoc imaging of the accumulated damage record rather than by instrumenting a large target mass in real time.

The basic attraction of the method is the combination of long integration times and nanometric readout. Paleo-detectors use small samples of naturally occurring rocks that have been deep underground, typically for $\mathcal{O}(1)$ Gyr, and modern microscopy techniques promise nanometer-resolution readout in macroscopic samples (Baum et al., 2021). In heavy-element proposals, integration over geological times $T\sim10^6$–$10^9$ yr builds up $O(10^3$–$10^6)$ t·yr exposure in even $\sim100$ g samples, while conventional forecasts for the broader paleo-detector program reach keV recoil thresholds and 100 kilotonne-yr exposures (Graham et al., 3 Jun 2026, Baum et al., 2021).

The preference for heavy nuclei follows directly from the scattering kinematics and rates. For elastic spin-independent scattering, the sensitivity scales roughly as $A_T^2$, so large mass number nuclei are preferred (Drukier et al., 2018). For inelastic dark matter, the dependence is stronger still: the maximum accessible mass splitting is

$\delta_{\max,A}=\tfrac12\,\mu_A\,v_{\max}^2,$

so heavy nuclei increase the kinematic reach because $\mu_A$ grows with target mass (Graham et al., 3 Jun 2026).

2. Stopping powers, latent-track formation, and morphology

Track formation in insulating crystals is governed by the competition between electronic stopping and nuclear stopping. An energetic ion loses energy through ionization and excitation of target electrons,

$S_e=\frac{dE}{dx}_e,$

and through elastic collisions with target nuclei,

$T\sim10^6$0

A useful diagnostic is the fractional nuclear stopping

$T\sim10^6$1

which parameterizes the transition from electronically dominated energy loss to collision-dominated cascade damage (Calabrese-Day et al., 9 Apr 2026).

In the electronic, thermal-spike regime, track radius is commonly parameterized by the Szenes scaling law,

$T\sim10^6$2

where $T\sim10^6$3 is the electronic stopping threshold for continuous track formation (Calabrese-Day et al., 9 Apr 2026). In the nuclear-dominated regime, by contrast, damage is described as discontinuous “islands” of amorphous or vacancy-rich zones rather than a smooth cylinder, and empirically tracks become spotty when $T\sim10^6$4 (Calabrese-Day et al., 9 Apr 2026).

The clearest experimental benchmark for this morphology in a candidate paleo-detector mineral is the transmission electron microscopy study of Au-irradiated olivine. In that work, natural olivine Mg$T\sim10^6$5Fe$T\sim10^6$6SiO$T\sim10^6$7 was irradiated with 15 MeV Au$T\sim10^6$8 and examined by focused ion-beam sectioning and scanning transmission electron microscopy at target depths from 423 to 2920 nm, without etching (Calabrese-Day et al., 9 Apr 2026). The measured widths changed only modestly across the energy range studied: approximate weighted means for linear tracks were 6.0, 6.1, 6.3, 6.8, 7.1, 7.4, 7.9, 7.7, and 7.0 nm from 423 to 2920 nm, while circular-track estimates ranged from 6.5 to 9.2 nm over the same depths (Calabrese-Day et al., 9 Apr 2026). The paper reports a significant change in track continuity across 0.4–12.9 MeV, with smooth, continuous tracks at shallow depths and spotty, discontinuous tracks at larger depths, consistent with the transition from electronic to nuclear stopping dominance predicted by SRIM (Calabrese-Day et al., 9 Apr 2026).

This result is significant for heavy-element paleodetectors for two reasons. First, it demonstrates that latent tracks in a paleo-detector candidate can be resolved directly at the few-nanometer scale without etching. Second, it anchors the broader claim that track morphology is not determined by recoil energy alone but also by how that energy partitions between $T\sim10^6$9 and $10^9$0. A plausible implication is that heavy-element target optimization must account for morphology changes, not only total scattering rate.

3. Minerals, radiopurity, and geological provenance

Material choice is constrained simultaneously by scattering physics, track retention, and backgrounds. The principal criteria stated in the paleo-detector literature are large mass number for coherence enhancement, electrical resistivity $10^9$1 to allow track recording, high melting or annealing temperature for long track retention, ultra-low concentrations of radioactive contaminants such as $10^9$2U and $10^9$3Th, and in some cases the presence of hydrogen for neutron moderation (Drukier et al., 2018).

Geological setting or class	Examples	Noted properties
Marine evaporites	Halite, gypsum, epsomite	Typically $10^9$4 in weight; some contain H
Ultra-basic rocks	Olivine, phlogopite, nickelbischofite	Typically $10^9$5 in weight
Brine precipitates from deep geothermal aquifers	Laurionite, cottunite, paralaurionite, galena	Deep $10^9$6 sources; Pb-bearing; extremely low U is possible

The explicit heavy-element proposal centers on ancient, radiopure minerals containing lead. Brine precipitates from deep geothermal aquifers are identified as a possible source because hot $10^9$7, reducing conditions render UO$10^9$8 and USiO$10^9$9 insoluble, leading to measured uranium concentrations $O(10^3$0 g U/g in Gulf-Coast aquifers and a theoretical minimum $O(10^3$1 g/g (Graham et al., 3 Jun 2026). These aquifers lie at 4–5 km depth, where cosmogenic neutrons are negligible relative to radiogenic ones, and their ages can range from Oligocene values of 23–34 Myr to older horizons up to $O(10^3$2 Gyr (Graham et al., 3 Jun 2026).

Laurionite is the benchmark Pb-bearing mineral in this literature. Its chemical formula is PbClOH, its Pb mass fraction is $O(10^3$3, its electrical resistivity is $O(10^3$4–$O(10^3$5, and its hydrogen content helps moderate fast-neutron backgrounds (Graham et al., 3 Jun 2026). Other Pb-bearing minerals explicitly noted are cottunite PbCl$O(10^3$6, paralaurionite PbOHCl, and galena PbS (Graham et al., 3 Jun 2026). By contrast, olivine appears as an ultra-basic rock target with Fe and Mg, important both as a general paleo-detector material and as the mineral in which track-formation microphysics has been directly characterized (Drukier et al., 2018, Calabrese-Day et al., 9 Apr 2026).

4. Readout modalities and the observable track spectrum

The observable in paleodetection is usually the damage-track length distribution rather than the recoil-energy spectrum itself. For a recoiling nucleus of energy $O(10^3$7, the track length is approximated by its range,

$O(10^3$8

with stopping powers obtained in practice from SRIM or TRIM (Baum et al., 2021). The differential track-length spectrum is then written as

$O(10^3$9

or, more generally, as a convolution over the conditional distribution of track lengths at fixed recoil energy when stochastic cascade effects are included (Fung et al., 11 Apr 2025).

Two readout architectures dominate the literature. High-resolution Helium-Ion-Beam Microscopy is associated with $10^6)$0 nm and $10^6)$1 mg, corresponding to an effective threshold $10^6)$2 nm, approximately 1 keV in heavy-element minerals (Baum et al., 2021). Small-Angle X-ray Scattering tomography is associated with $10^6)$3 nm and $10^6)$4 g, corresponding to $10^6)$5 nm, approximately a few keV (Baum et al., 2021). In the heavy-element Higgsino study, the corresponding processable volumes are given as $10^6)$6 mm$10^6)$7 for HIBM and $10^6)$8 cm$10^6)$9 for SAXS (Graham et al., 3 Jun 2026).

Microscopy-based microphysical validation has also been demonstrated directly in olivine. The 2026 STEM study used a Thermo-Fisher Spectra 300 at 300 keV, bright-field imaging for track measurements, pixel sizes of 0.25 nm or 0.36 nm, and focused ion-beam staircase lift-outs at 10 depths (Calabrese-Day et al., 9 Apr 2026). Track extraction employed ridge detection plus transverse Gaussian fits for oblique tracks and ParticleAnalyzer plus circular-Gaussian fits for near-normal tracks (Calabrese-Day et al., 9 Apr 2026). This establishes that few-nanometer tracks can be measured without etching, but the scalable readout challenge remains the transition from $\sim100$0m$\sim100$1-scale microscopy to macroscopic sample volumes.

A central revision introduced by TRIM-based sensitivity studies is that recoil energy and track length are not in one-to-one correspondence. The full distribution is

$\sim100$2

where $\sim100$3 is the conditional track-length distribution and $\sim100$4 is the probability that a recoil produces any visible track at all (Fung et al., 11 Apr 2025). This refinement becomes critical near threshold.

5. Backgrounds, statistical inference, and revised sensitivities

Background control in paleodetectors is a mineralogical and geological problem as much as a detector-physics problem. Cosmogenic backgrounds are negligible if samples spend $\sim100$5 km-water-equivalent underground, or at depths $\sim100$6 km rock, where the cosmogenic neutron flux is negligible relative to the relevant radiogenic and neutrino backgrounds (Baum et al., 2021, Drukier et al., 2018). Radiogenic backgrounds are dominated by the $\sim100$7U chain, spontaneous fission, and $\sim100$8 neutrons; one particularly important feature is the isolated U-234 recoil from the first $\sim100$9-decay of U-238, which produces a monoenergetic 72 keV heavy recoil and a characteristic track-length signature near 200 nm in analyses that do not observe low-$A_T^2$0 tracks (Drukier et al., 2018).

Hydrogen-bearing minerals are repeatedly emphasized because they moderate radiogenic neutrons efficiently. This is why gypsum and sinjarite are favorable in the general dark-matter forecasts, and why laurionite is attractive among Pb-bearing candidates (Baum et al., 2021, Graham et al., 3 Jun 2026). Fiducial uranium concentrations are $A_T^2$1 g/g for marine evaporites and $A_T^2$2 g/g for ultra-basic rocks, while the heavy-element Higgsino proposal targets $A_T^2$3 g/g for ultimate reach (Baum et al., 2021, Graham et al., 3 Jun 2026).

The modern sensitivity forecasts rely on binned spectral analyses with nuisance parameters for sample age, neutrino flux normalizations, and radioactive contamination. One formulation writes the expected counts in track-length bin $A_T^2$4 as

$A_T^2$5

with a Gaussian window function of width $A_T^2$6, and uses a Poisson likelihood with Gaussian priors on nuisance parameters (Baum et al., 2021). In the 2021 projections, the 90% CL exclusion criterion follows the profile-likelihood ratio with Asimov value $A_T^2$7 (Baum et al., 2021). Earlier work also formulated sensitivity through a signal-to-noise requirement

$A_T^2$8

with $A_T^2$9 for neutrinos and $\delta_{\max,A}=\tfrac12\,\mu_A\,v_{\max}^2,$0 for radioactive backgrounds (Drukier et al., 2018).

For elastic spin-independent dark matter, the updated 2021 projections give the following representative numbers. In a high-resolution scenario with 10 mg, $\delta_{\max,A}=\tfrac12\,\mu_A\,v_{\max}^2,$1 nm, and $\delta_{\max,A}=\tfrac12\,\mu_A\,v_{\max}^2,$2 Gyr, paleo-detectors can probe down to the conventional neutrino floor in a Xe-based direct detection experiment for masses $\delta_{\max,A}=\tfrac12\,\mu_A\,v_{\max}^2,$3 GeV/$\delta_{\max,A}=\tfrac12\,\mu_A\,v_{\max}^2,$4; for $\delta_{\max,A}=\tfrac12\,\mu_A\,v_{\max}^2,$5 GeV/$\delta_{\max,A}=\tfrac12\,\mu_A\,v_{\max}^2,$6 the sensitivity reaches $\delta_{\max,A}=\tfrac12\,\mu_A\,v_{\max}^2,$7 cm$\delta_{\max,A}=\tfrac12\,\mu_A\,v_{\max}^2,$8, and for $\delta_{\max,A}=\tfrac12\,\mu_A\,v_{\max}^2,$9 GeV/$\mu_A$0 it reaches $\mu_A$1 cm$\mu_A$2 (Baum et al., 2021). In a high-exposure scenario with 100 g and $\mu_A$3 nm, the projected sensitivity is $\mu_A$4 cm$\mu_A$5 at $\mu_A$6 GeV/$\mu_A$7 and $\mu_A$8 cm$\mu_A$9 at 100 GeV/$S_e=\frac{dE}{dx}_e,$0 (Baum et al., 2021).

A major correction to earlier optimism comes from realistic TRIM-based track modeling. In olivine, below $S_e=\frac{dE}{dx}_e,$1 keV the track-formation probability falls below 50%; at $S_e=\frac{dE}{dx}_e,$2 keV only $S_e=\frac{dE}{dx}_e,$3–30% of recoils yield any visible track, while by $S_e=\frac{dE}{dx}_e,$4 keV the yield is approximately 100% (Fung et al., 11 Apr 2025). At the opposite end, once electronic stopping dominates above $S_e=\frac{dE}{dx}_e,$5–100 keV, the average track length saturates at $S_e=\frac{dE}{dx}_e,$6 nm, producing a “track length barrier” (Fung et al., 11 Apr 2025). The paper concludes that previous studies overestimated the number of tracks caused by weakly interacting particles, and that the realistic limits degrade by 0.5–1 decade in $S_e=\frac{dE}{dx}_e,$7 at low $S_e=\frac{dE}{dx}_e,$8 and by $S_e=\frac{dE}{dx}_e,$9 in $T\sim10^6$00 relative to one-to-one energy–length treatments (Fung et al., 11 Apr 2025). For heavy-element minerals, the same work notes that because electronic stopping scales roughly as $T\sim10^6$01, heavy-$T\sim10^6$02 targets exhibit this plateau at even shorter lengths and lower recoil energies (Fung et al., 11 Apr 2025).

6. Heavy nuclei for inelastic dark matter and halo-history sensitivity

The most distinctive role of heavy-element paleodetectors emerges in inelastic dark matter models such as the Higgsino, where existing paleo-detector targets lack sufficiently heavy nuclei to overcome the kinematic threshold for scattering (Graham et al., 3 Jun 2026). In this scenario, dark matter $T\sim10^6$03 up-scatters to a heavier state $T\sim10^6$04 with mass splitting $T\sim10^6$05, and the minimum velocity for recoil energy $T\sim10^6$06 is

$T\sim10^6$07

No scattering is kinematically allowed if $T\sim10^6$08, so target mass becomes the decisive figure of merit (Graham et al., 3 Jun 2026).

This is why Pb-bearing minerals are central. Since $T\sim10^6$09 for $T\sim10^6$10, the ordering $T\sim10^6$11 follows directly, and Pb with $T\sim10^6$12 materially extends the accessible inelastic frontier (Graham et al., 3 Jun 2026). The differential recoil rate used in the Higgsino benchmark is

$T\sim10^6$13

with benchmark parameters $T\sim10^6$14 GeV/cm$T\sim10^6$15, $T\sim10^6$16 TeV, and $T\sim10^6$17 cm$T\sim10^6$18 (Graham et al., 3 Jun 2026).

The benchmark heavy-element forecast uses laurionite with $T\sim10^6$19 cm$T\sim10^6$20, age $T\sim10^6$21 Gyr, SAXS readout at $T\sim10^6$22 nm, $T\sim10^6$23 g/g, and depth $T\sim10^6$24 km (Graham et al., 3 Jun 2026). In the Standard Halo Model this yields sensitivity up to $T\sim10^6$25 keV at $T\sim10^6$26 cm$T\sim10^6$27 (Graham et al., 3 Jun 2026). When a Large Magellanic Cloud-induced high-velocity tail is included, the reach extends to $T\sim10^6$28 keV for the present-day velocity distribution and to $T\sim10^6$29 keV at pericenter; with the theoretical minimum uranium concentration, the ultimate reach is stated as $T\sim10^6$30 keV (Graham et al., 3 Jun 2026).

The same study argues that heavy-element paleodetectors are uniquely sensitive to the history of the dark matter high-velocity tail. Simulations of Auriga halo 13 indicate that the Large Magellanic Cloud pericenter about 50 Myr ago injected a fast, unbound dark matter population, increasing the maximum Earth-frame speed from the Standard Halo Model value to present-day $T\sim10^6$31 km/s and pericenter $T\sim10^6$32 km/s (Graham et al., 3 Jun 2026). For $T\sim10^6$33 keV, upscattering is kinematically forbidden until the LMC boost, so a Gyr-old sample would accumulate approximately 950 Myr of background before any signal; a younger $T\sim10^6$34 Myr sample therefore optimizes signal-to-noise for large $T\sim10^6$35 (Graham et al., 3 Jun 2026). This temporal selectivity is a distinctive feature of time-integrating detectors and has no direct analogue in conventional real-time experiments.

Relaxed scenarios are also quantified. Even radio-impure samples with $T\sim10^6$36 up to $T\sim10^6$37 g/g from depths of only 2 km still probe $T\sim10^6$38 keV, and a $T\sim10^6$39 smaller volume, $T\sim10^6$40 mm$T\sim10^6$41, remains competitive for $T\sim10^6$42 keV (Graham et al., 3 Jun 2026). This does not eliminate the challenges of radiopurity and depth, but it indicates that the unusually large Higgsino-nucleon cross section partially relaxes requirements that are otherwise stringent in the paleodetector program.

Taken together, the literature defines heavy-element paleodetectors as a specialized branch of paleo-detection in which mineralogy is tuned to high-mass nuclei, especially lead, while track-formation microphysics, background rejection, and nanoscale readout remain inherited from the broader ancient-mineral program. The central design tension is now clear: heavy nuclei increase coherent rate and inelastic reach, but stronger electronic stopping compresses the observable track spectrum and pushes the track-length barrier to shorter scales (Fung et al., 11 Apr 2025). This suggests that the future development of heavy-element paleodetectors will depend on jointly optimizing target $T\sim10^6$43, hydrogen content, radiopurity, age, depth, and readout resolution rather than maximizing any single parameter alone.