DarkSide-50 Links Delayed Electrons to Impurities

Dual-phase argon time-projection chambers can detect very small ionization signals, a useful capability for low-mass dark-matter searches. The same sensitivity exposes a difficult background: delayed, spurious electrons that can resemble genuine small S2 signals. DarkSide-50 presents a detector-lifetime study of these events in argon, using their timing, position, parent-event dependence, pulse shapes, and response to purification conditions to divide the population into correlated and apparently uncorrelated components.

The paper's central result is empirical rather than a single confirmed microscopic diagnosis. For events with fewer than four extracted electrons, the authors estimate that roughly 30–70% are associated with delayed release of electrons captured by impurities. The fraction changes over the experiment's lifetime, alongside the purification system's operating state. That makes the result operationally useful: this background is not simply an irreducible feature of argon electroluminescence.

Core Contribution

Earlier discussions of single-electron backgrounds in noble-liquid detectors had substantial xenon evidence but no comparably comprehensive argon characterization. DarkSide-50 contributes a phenomenological decomposition of the argon population. Temporally correlated spurious-electron events have an identifiable preceding S2 parent and show spatial and temporal correlations with it; uncorrelated events have no parent in the roughly one-second searchable window.

The distinction matters because it separates mechanisms that may respond to different interventions. The correlated component scales linearly with the parent S2 size and drift time, consistent with a probability of electron capture during drift followed by delayed release. The uncorrelated component instead tracks the overall ionization rate, which leaves open missed parents, long-lived charge reservoirs, or other production channels.

Technical Approach

The analysis starts from isolated low-electron S2 pulses and compares their rate with prior events. Fits to the parent-to-spurious-electron delay distribution require at least two exponential components: a fast component near $\tau_1\sim5\,\mathrm{ms}$ and a slower component varying from about $\tau_2\sim40$ to $80\,\mathrm{ms}$. A component near $\tau_3\sim16\,\mathrm{ms}$ appears when the gas getter is bypassed. The changing $\tau_2$ value and its correlation with radon-trap temperature are used as evidence for more than one impurity species or impurity state.

The paper also tests whether pulse observables are stable across parent conditions. The mean extracted-electron signal, $\mu_{\rm SE}$, shows no obvious trend with parent drift time or parent S2 size. This supports an interpretation in which the extracted multiplicity is governed by a common release process rather than by a changing single-electron response. For two- and three-electron signals, however, the lower early-light fraction $f_{5000}$ indicates that the electrons reach the gas pocket at different times but are reconstructed as one pulse.

A separate argument addresses multi-electron events. The authors estimate that 128-nm argon electroluminescence can photo-ionize the stainless-steel extraction grid with a 2–4% probability per one-electron event, using the grid transparency, measured steel photoelectric yield, and photon incidence estimates. The observed pulse-shape behavior is consistent with this mechanism, though it is not a direct material-level measurement.

Results and Analysis

The strongest quantitative support is the correlated population's dependence on both parent S2 and drift length. The paper reports a trapping probability of $1.74\times10^{-8}\,\mathrm{mm}^{-1}$ per drift length from these correlations. Its fitted temporal structure spans milliseconds, not the much longer timescale of the uncorrelated component, whose lifetime is inferred to exceed one second.

The authors' interpretation is appropriately conditional. Impurities explain a substantial measured fraction of the low-electron population and offer a concrete route to reduction through purification and impurity studies. The grid hypothesis offers a second route: suppressing photo-ionization could allow multi-electron S2s to enter future low-threshold background models. These are meaningful detector-design implications, but neither identifies the responsible impurity nor closes the accounting of uncorrelated events.

Caveats in Practice

The evidence is based on one argon detector and on correlation fits rather than controlled impurity injections. The paper explicitly calls for dedicated studies with spiked impurity concentrations to establish species-level causation and test mitigation. It also notes that the long-lived component is statistically difficult to study within the available live period, so an apparently uncorrelated event may still have an unrecognized parent. The work therefore supports targeted background reduction strategies, not a complete microscopic model of every spurious electron.