Calculation of crystal defects induced in CaWO4 by 100 eV displacement cascades using a linear Machine Learning interatomic potential

Aug 5, 2026

Upload by Miles

Energy stored in crystal defects created by O(10−100) eV nuclear recoils in low-threshold CaWO4 cryogenic detectors is determined using molecular-dynamics simulations. A machine-learning interatomic potential is developed to enable accurate atomic-scale modeling at ab initio–level fidelity with reduced computational cost. Crystal defects modify the recoil energy spectra relevant to Dark Matter and neutrino coherent scattering, and reference predictions are provided. The neutron-capture recoil case is discussed for its potential to uniquely validate the stored-energy calculations.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages6

Reading Time15 min

AccessFree to download

Introduction

Low-threshold cryogenic detectors for DM and CEνNS

Role of crystal defects in phonon signals

Method

Building the DFT database for CaWO4

Training the linear machine-learning interatomic potential

Calculation of crystal defects induced in CaWO4 by 100 eV displacement cascades using a linear Machine Learning interatomic potential arXiv :2407 .00133v1 [physics .ins-det] 28 Jun 2024 G. Soum-Sidikov, 1, 2 J. P. Crocombette,2, ∗ M. C. Marinica,2 C. Doutre, 1 D. Lhuillier, 1 and L. Thulliez 1 1 IRFU, CEA, Universite´ Paris-Saclay, 91191 Gif-sur-Yvette, France 2 Service de recherche en Corrosion et Comportement des Mat e´riaux, SRMP, Universite´ Paris-Saclay, CEA, 91191 Gif Sur Yvette, France (Dated: July 2, 2024) We determine the energy stored in the crystal defects induced by O(10 − 100) eV nuclear recoils in lowthreshold CaWO4 cryogenic detectors. A Machine Learning interatomic potential is developed to perform molecular dynamics simulations. We show that the energy spectra expected from Dark Matter and neutrino coherent scattering are affected by the crystal defects and we provide reference predictions. We discuss the special case of the spectrum of nuclear recoils induced by neutron capture, which could offer a unique sensitivity to the calculated stored energies. Cryogenic low-threshold detectors are at the crossroads of two vast experimental programs: the search for light Dark Matter (DM) and the study of Coherent Elastic NeutrinoNucleus Scattering (CEνNS) . Despite abundant evidence of a predominantly “dark” component to the mass of the universe [1], no direct interaction with ordinary matter has yet been observed. The strongest constraints have been placed on the elastic scattering of DM with mass ≥1 GeV [2] . Lighter masses are now being considered, implying lower nuclear recoils in the detectors at the O(10 − 100) eV scale. Access to low detection thresholds also paves the way for the study of CEνNS, experimentally demonstrated for the first time in 2017 [3] . The precise study of this new neutrino-matter interaction offers opportunities for original tests of the Standard Model at low energies [4] . Detection thresholds at the 10 eV scale have recently been demonstrated with Ge and Si cryogenic detectors with masses of the order of several grams [5–8] . Here, we focus on cryogenic detectors made of CaWO4 crystals equipped with a Transition Edge Sensor to detect the phonon (heat) signal induced by the nuclear recoils. This technology was initially developed by the CRESST experiment [9, 10] for DM searches and is now applied to the detection of CEνNS by the NUCLEUS experiment [11, 12] . As the elastic scattering of low speed DM particles and low energy neutrinos are coherent processes, their cross-sections are respectively proportional to the square of the number of nucleons and neutrons in the target nuclei, motivating the use of crystals containing heavy atoms like tungsten. The trade-off is a very low energy nuclear recoil of hundreds of eV or lower. Such a low energy range means that we need to study in detail the atomic-scale material effects that can affect the phonon signal from cryogenic detectors. In particular, part of the energy deposited in the target material can be stored in crystal defects created along the displacement cascade, affecting the energy scale of the detector. Molecular dynamics (MD) simulations of this phenomenon have already been performed with empirical potentials for some commonly used detector materials [13, 14] not including CaWO4 . In this paper, we present MD calculations based on a new Machine Learning (ML) interatomic potential developed for this study. Bypassing complexity walls, ML empowers simulations to achieve atomic-level fidelity atthe ab initio level, dramatically reducing computational costs. The predicted DM and CEνNS recoil energy spectra are then expressed as a function of the detected energy, taking into account the energy stored in crystal defects. We discuss as well the calibrated nuclear recoils induced by neutron capture on tungsten [15–17] as a potential experimental validation of these calculations. To design the CaWO4 ML potential we first

FAQs

What physical quantity does the study calculate in CaWO4 detectors?

The study calculates the energy stored in crystal defects induced by nuclear recoils of O(10−100) eV in CaWO4 cryogenic detectors.

How are the molecular-dynamics simulations performed?

Molecular dynamics simulations are carried out using a newly developed machine-learning interatomic potential trained from a DFT database.

How do crystal defects affect the expected signals for Dark Matter and neutrino scattering?

Crystal defects change the recoil energy spectra expected from Dark Matter and neutrino coherent scattering, and the work provides reference predictions including the stored defect energy.

Aug 5, 2026

Upload by Miles