The DarkSide-50 detector is a dual-phase time projection chamber filled with 46.4 kilograms of argon harvested from deep underground wells. This underground argon is precious because its concentration of the radioactive isotope argon-39 is at least 1400 times lower than that of ordinary atmospheric argon, dramatically suppressing one of the most stubborn backgrounds in rare-event searches. When a particle interacts inside the cylindrical target, it produces a prompt flash of scintillation light, called S1, and liberates ionization electrons that drift upward through the liquid under a 200 volts-per-centimeter electric field. At the liquid surface, a stainless steel grid extracts these electrons into a thin layer of argon gas, where they generate a second, amplified flash of electroluminescence known as S2.

Each extracted electron produces roughly 23 detected photoelectrons, making the S2 signal extraordinarily sensitive. This sensitivity underpins the so-called S2-only analysis strategy, in which researchers hunt for dark matter collisions so feeble that they generate ionization electrons but too little scintillation light to register an S1. Because the S2 channel can be read with near-perfect efficiency, it opens a window onto dark matter particles with masses below about 10 gigavolts per square of the speed of light, a regime where conventional analyses lose their grip. The same technique is proposed for detecting coherent neutrino scattering from supernovae and reactors, making its reliability a matter of broad interest across astroparticle physics.

But every experiment that has run in this mode, whether filled with xenon or argon, has been plagued by an excess of tiny signals corresponding to just a few extracted electrons. These spurious electrons, or SEs, appear at rates that no model of radioactive background can explain. Their presence forced DarkSide-50’s earlier light dark matter searches to discard their lowest-energy data, capping their sensitivity at WIMP masses above roughly 1.2 gigavolts per square of the speed of light. Until now, no one had systematically characterized these events in liquid argon, leaving theorists to extrapolate from xenon experiments such as XENON1T and LUX.

The new analysis draws on nearly 1000 days of data collected between April 2015 and February 2018. The collaboration classified every triggered event by its pulse structure and isolated those containing a single S2-like pulse with fewer than four extracted electrons. The resulting time series revealed something striking: while ordinary background categories remained rock-steady over three years of operation, the spurious-electron rate decayed rapidly during the first 200 days after the detector was filled, with a characteristic timescale of about 65 days, before settling into a much slower decline. Even more telling, when the hot getter that purifies the recirculating argon gas was bypassed for five days of maintenance, the spurious-electron rate jumped sharply and then fell back with a 36-hour time constant after the getter was restored, matching the expected purification speed of the circulation loop.

That spike pointed the finger at chemical impurities. The team therefore measured two independent purity metrics: the free electron lifetime, which quantifies how long electrons survive drifting through the liquid before being captured by electronegative contaminants such as oxygen, and the lifetime of the long-lived triplet component of the argon scintillation, which is sensitive to nitrogen contamination. Surprisingly, neither metric degraded during the getter-off period, and neither correlated with the spurious-electron rate over the full dataset. Whatever impurities are capturing and later releasing electrons, they are evidently not the same species that dominate electron attachment or scintillation quenching, hinting at a subtle population of contaminants that standard purity monitors cannot see.

To pin down the mechanism, the collaboration searched for temporal coincidences between spurious electrons and ordinary particle interactions, treating the latter as potential parent events. The distribution of time delays between parents and SEs resolved cleanly into two exponential components with decay constants of about 5 milliseconds and 40 to 80 milliseconds, plus a third component with a roughly 16-millisecond constant that appeared only while the getter was off. At the start of the dataset, these correlated components accounted for about 70 percent of all spurious electrons, falling to about 30 percent after 200 days as the responsible impurities were gradually removed. A longer-lived component extending beyond one second was also visible, and the team found that the apparently uncorrelated SE rate scales linearly with the total ionization activity in the detector, suggesting that even these events may trace back to parents whose delayed signatures stretch over seconds.

The spatial and energetic fingerprints reinforced the impurity picture. Spurious electrons tended to appear within about 5 centimeters of their parent’s reconstructed horizontal position, and the probability of an SE following a parent grew linearly with both the parent’s S2 charge and its electron drift time. Normalizing for the parent’s charge, the collaboration derived a trapping probability of about 1.74 times ten to the minus eight per electron per millimeter of drift, a tiny but consequential efficiency. Notably, the number of electrons lost to genuine capture by electronegative impurities is far larger than the number that resurface as spurious signals, meaning only a small fraction of captured electrons ever escape their traps.

The multi-electron events yielded perhaps the most unexpected clue. The distribution of electron multiplicity in SE events follows a Poisson law with a mean of just 0.062 extra electrons per event, independent of impurity species, parent energy, and drift time. The team proposes that the 128-nanometer ultraviolet photons of the electroluminescence process themselves photo-ionize the stainless steel extraction grid, whose measured photoelectric yield at these wavelengths is around one percent. A simple calculation predicts a 2 to 4 percent chance that a single-electron event liberates a secondary electron from the grid, beautifully consistent with the observed mean. Pulse-shape analysis supports this: two- and three-electron events show light arriving at slightly different times, exactly as expected if secondary electrons reach the gas pocket about a microsecond after the photons from the first electron strike the metal.

The implications reach well beyond DarkSide-50, which has since completed its mission. Its successor, DarkSide-20k, and other next-generation argon and xenon detectors plan to push S2-only analyses to even lower thresholds in pursuit of dark matter particles with sub-GeV masses. If the impurity-driven components can be eliminated through tighter purification and material control, and if the grid photo-ionization can be suppressed by coating or redesigning the extraction electrode, the spurious-electron floor that currently limits these searches could drop dramatically. The DarkSide-50 results provide the first quantitative roadmap for that effort, turning a mysterious background into a characterized, and potentially conquerable, phenomenon.

Subject of Research: Spurious-electron background signals in the dual-phase liquid argon time projection chamber of the DarkSide-50 dark matter experiment

Article Title: Characterization of spurious-electron signals in the double-phase argon TPC of the DarkSide-50 experiment

Article References: DarkSide-50 Collaboration, Agnes, P., Albuquerque, I. F., Alexander, T., Alton, A. K., Ave, M., Back, H. O., Batignani, G., Berzin, E., Biery, K., Bocci, V., Bonivento, W. M., Bottino, B., Bussino, S., Cadeddu, M., Cadoni, M., Calaprice, F., Caminata, A., Campos, M. D., … Zuzel, G. (2026). Characterization of spurious-electron signals in the double-phase argon TPC of the DarkSide-50 experiment. The European Physical Journal C, 86(9), Article 1099. https://doi.org/10.1140/epjc/s10052-026-16193-y

Image Credits: AI Generated

DOI: 10.1140/epjc/s10052-026-16193-y

Keywords: DarkSide-50, dark matter, liquid argon, time projection chamber, spurious electrons, WIMP, low-mass dark matter, ionization signal, Gran Sasso, electronegative impurities, photo-ionization, S2-only analysis