[Paper Review] Reply to the Comments on the XENON100 First Dark Matter Results

The XENON Collaboration|arXiv (Cornell University)|May 14, 2010

Dark Matter and Cosmic Phenomena4 citations

TL;DR

This paper defends the XENON100 experiment's first dark matter results against criticisms regarding the extrapolation of scintillation efficiency ($\mathcal{L}_{\text{eff}}$) to low nuclear recoil energies. It demonstrates that, even with conservative assumptions on $\mathcal{L}_{\text{eff}}$, the experiment excludes the light WIMP interpretation of DAMA and CoGeNT data at 90% CL, while retaining a small allowed region for CoGeNT-compatible WIMPs at 7–9 GeV/$c^2$ under the most conservative assumptions.

ABSTRACT

The recently submitted preprint on the first results from the XENON100 dark matter experiment (arxiv:1005.0380) was followed by a criticism by J.I. Collar and D.N. McKinsey (arxiv:1005.0838), focused on our extrapolation of the scintillation efficiency L_eff to the lowest nuclear recoil energies, where no data and no theoretical model exist. Here we add clarifications on our analysis and comment on their criticism.

Motivation & Objective

• To address criticisms from Collar and McKinsey regarding the extrapolation of scintillation efficiency ($\mathcal{L}_{\text{eff}}$) to low nuclear recoil energies in the XENON100 experiment.
• To justify the use of a global fit to all direct $\mathcal{L}_{\text{eff}}$ measurements rather than individual or unverified datasets.
• To demonstrate that the XENON100 result remains robust under conservative assumptions, including a 90% lower confidence limit extrapolation of $\mathcal{L}_{\text{eff}}$ and a lowered 3 PE threshold.
• To clarify that the experiment's sensitivity is not compromised by threshold effects, even with finite energy resolution and low photoelectron counts at the threshold.

Proposed method

• The analysis uses a global fit to all published direct measurements of $\mathcal{L}_{\text{eff}}$ across 5–100 keVr, with systematic uncertainties quantified via ±90% confidence level contours.
• The authors exclude certain neutron calibration data (e.g., XENON10 and ZEPLIN-III) from the global fit due to higher systematic uncertainties from simulation-based comparisons.
• A conservative logarithmic extrapolation of the 90% lower contour of $\mathcal{L}_{\text{eff}}$ is applied below 5 keVr to assess the most pessimistic sensitivity scenario.
• The energy resolution effects at the threshold are modeled by convolving the true WIMP spectrum with a Poisson distribution to account for statistical fluctuations in photoelectron counts.
• The experiment's detection efficiency is evaluated down to 3 PE, showing it remains high enough to maintain sensitivity despite low signal levels.
• Astrophysical parameters such as WIMP density (0.3 GeV/cm³), solar velocity (220 km/s), and galactic escape velocity (544 km/s) are used to compute the expected WIMP event rate.

Experimental results

Research questions

• RQ1 Does the XENON100 experiment’s exclusion of light WIMPs hold under conservative assumptions about $\mathcal{L}_{\text{eff}}$ extrapolation to low energies?
• RQ2 How does the choice of $\mathcal{L}_{\text{eff}}$—particularly its 90% lower confidence limit—affect the sensitivity to low-mass WIMPs?
• RQ3 To what extent do threshold effects, due to low photoelectron counts and finite energy resolution, impact the detection of sub-4 PE events?
• RQ4 Can the XENON100 result exclude the light WIMP interpretation of the DAMA annual modulation signal under the most conservative assumptions?
• RQ5 How do different $\mathcal{L}_{\text{eff}}$ datasets, including unpublished or simulation-based results, compare in reliability and systematic uncertainty?

Key findings

• The XENON100 experiment excludes the light WIMP interpretation of the DAMA annual modulation signal at 90% confidence level, even when using the 90% lower contour of $\mathcal{L}_{\text{eff}}$ and a conservative extrapolation to 1 keVr.
• For a 7 GeV/$c^2$ WIMP, the expected number of events above a 3 PE threshold is 0.73–1.5 under the conservative $\mathcal{L}_{\text{eff}}$ assumption, consistent with the absence of observed events.
• The global fit to all direct $\mathcal{L}_{\text{eff}}$ measurements yields a monotonically falling $\mathcal{L}_{\text{eff}}$ curve with ±90% confidence level contours, which is used as the basis for sensitivity limits.
• Even with a lowered 3 PE threshold (8.2 keVr for conservative $\mathcal{L}_{\text{eff}}$), the experiment expects 4.4 events for a 10 GeV/$c^2$ WIMP with a cross section of $1 \times 10^{-41}$ cm², but observes zero events, leading to a 90% CL exclusion.
• The XENON100 result remains robust under conservative assumptions: the CoGeNT-favored region is excluded at 90% CL when $\mathcal{L}_{\text{eff}}$ is assumed constant, but a small fraction remains uncovered under the most conservative extrapolation.
• The authors reject the use of a single unpublished $\mathcal{L}_{\text{eff}}$ measurement (ref:manzur) due to low trigger efficiency and unverified Monte Carlo corrections, favoring instead the more systematically controlled data from Aprile et al. (ref:aprileLeff).