Spontaneous damage annealing reactions as a possible source of low energy excess in semiconductor detectors

TL;DR

The paper investigates the persistent low-energy excess observed in semiconductor dark matter and neutrino detectors, proposing that spontaneous annealing of radiation-induced defect pockets can yield energy releases with an exponential spectrum similar to experiment. Using a five-stage atomistic workflow (cascade damage, long-time annealing, 0 K quenching, energy-release extraction, and time-scale analysis) in silicon, and incorporating quantum zero-point effects via a quantum thermal bath, the authors show energy releases follow $f(E)=A\exp(-\alpha E)$ with temperature-insensitive slopes, and that annealing can proceed via avalanche-like cascades driven by small barriers around $\sim$0.1 eV. They locate most energy releases to amorphous pockets and interfaces, and demonstrate that such processes can occur even at cryogenic temperatures, potentially contributing to, but not fully explaining, the observed excess signal. The work suggests detector-background modeling should include defect-pocket dynamics and advocates further studies across materials and time-resolved experiments to validate and leverage these insights for dark matter detection.

Abstract

In semiconductor detectors designed for capturing dark matter particles or neutrinos, when the detection threshold is constantly improved to increasingly low energies, an "excess" signal of apparent energy release events below a few hundred eV is observed in several different kinds of detectors. This becomes a big obstacle to the observation of actual dark matter signals, hindering the detectors' sensitivity for rare events in this energy range. Using atomistic simulations with a classical thermostat and a quantum thermal bath, we show that this kind of signal is consistent with energy release from long-term annealing events of complex defects that can be formed by any kind of nuclear recoil radiation events. Such energy releases are shown to have a very similar exponential dependence on energy release magnitudes as that observed in experiments. By detailed analysis of the annealing events, we show that crossing very low energy barriers can trigger larger energy releases in an avalanche-like effect. This explains why large energy release events can occur even down to cryogenic temperatures, where the significant migration of point defects in silicon is hardly ever possible.

Figures (18)

• Figure 1: Cross-sectional view of the time evolution of damage formation in a 5 keV cascade in Si, as modeled with the Stillinger-Weber potential and a classical thermostat. In these simulations, one atom in the cell was given a recoil energy of 5 keV towards the center of the cell, leading to a collision cascade producing damage. For clarity, only atoms in a 2 unit cell thick cross section in the center of the cell are shown. Atoms are of course colliding and damage is produced also in other atom layers than those shown. The top part shows the time evolution of the cascade at the specific times 0, 0.1, 0.4 and 1 ps, and the larger frame at the bottom the final damage state at 10 ps. This cascade was ran at 0 K ambient temperature to ensure there is no thermal annealing. The precise size of the plotted simulation cell region is 81.2 Å in the $x$ (horizontal), 71.3 Å in the $y$ (vertical) and 10.86 Å in the $z$ (out of plane) direction. The colors indicate the potential energy $E_{\rm pot}$ of the atoms relative to the ground state energy $E_{\rm pot,0}$, as indicated in the legend on the top right.
• Figure 2: Heat capacity of silicon crystal, calculated with the quantum thermal bath (QTB) and the classical Berendsen thermostat. Since the Berendsen thermostat and molecular dynamics describes only classical atom motion, its heat capacity prediction corresponds to the value given by the equipartition theorem, which does not have a temperature dependence Mandl. The results are compared against experimental values (red plus signs).
• Figure 3: a) Annealing of damage produced in a single 5 keV cascade in Si at 600 K. The data shows that both the average potential energy and the number of Wigner-Seitz defect pairs decreases with time. b) Same as subfigure (a) but analyzed after each time step was quenched to 0 K. The inset shows the atomic positions in one disordered pocket before and after the annealing event occuring at $t=1.236$ ns. c) Potential energy release per 2 ps intervals from the same cascade. The inset shows that annealing at well-separated times keep occurring also at longer times and even at the cryogenic temperature of 100 K.
• Figure 4: Annealing of damage produced in a single 5 keV cascade in Si at different temperatures. The data shows that both the average potential energy and number of Wigner-Seitz defect pairs decreases with time at all the considered temperatures.
• Figure 5: Statistics of the magnitude of the energy release events during annealing runs at different temperatures. To enable easy comparison of the slopes, the curves at different temperatures have been scaled to having a value of 1 at 1 eV. a) Classical molecular dynamics, b) Molecular dynamics with the quantum thermal bath