AI generated image

The enduring mystery of the DAMA/LIBRA experiment’s annual modulation signal has puzzled physicists for over two decades. In our latest paper, 2510.05216, we perform a detailed statistical analysis to assess whether this signal is compatible with the latest results from other sodium iodide (NaI) experiments, ANAIS-112 and COSINE-100.

The Long-Standing Puzzle

For years, the DAMA/LIBRA collaboration has reported a highly significant modulation in their detector’s event rate, peaking in the summer and troughing in the winter, consistent with the expected signal from a halo of dark matter particles (astro-ph/0307403). This interpretation, however, stands in stark contrast to results from numerous other direct detection experiments. Experiments using different target materials, such as liquid xenon detectors like LUX-ZEPLIN (10.1103/PhysRevLett.131.041002), have set limits on dark matter interactions that are orders of magnitude stronger than what would be needed to explain the DAMA signal. The persistence of the DAMA claim has often been defended by the possibility that dark matter interacts in a peculiar way, specific to the NaI crystals used by the experiment.

A New Generation of NaI Detectors

This defense is now being put to the direct test by a new generation of experiments using the same NaI target material. The ANAIS-112 and COSINE-100 experiments are designed to replicate the DAMA search and either confirm or refute the modulation signal. Their latest results, such as the full dataset from COSINE-100 (2409.13226), have found no evidence of a signal compatible with DAMA’s, creating significant tension.

Our Analysis: A Deep Dive into the Data

Our new work, led by Giorgio Busoni and including our group’s Will Handley, moves beyond simple comparisons to conduct a comprehensive statistical assessment. The core of our analysis is to use the full spectral information—the event rates binned by energy—from all three experiments. This is critical because detector-specific properties like energy resolution and quenching factors (how efficiently a nuclear recoil’s energy is converted to a measurable light signal) can dramatically alter the shape of an observed spectrum, even if the underlying physical signal is the same.

We approached this challenge from two distinct angles:

  • A Physically-Motivated Test: We first assume the signal arises from dark matter scattering within a standard astrophysical halo model. To remain general, we employ an Effective Field Theory (EFT) framework, allowing us to test a wide range of possible interactions beyond simple spin-independent or spin-dependent scattering. Using the powerful statistical machinery of GAMBIT, DarkBit, and DDCalc, we determine the best-fit dark matter model for DAMA’s data and then calculate the expected signal in ANAIS-112 and COSINE-100.
  • A Model-Agnostic Test: To see how generic the tension is, we also adopt a more flexible parameterization of the nuclear recoil spectrum. Instead of assuming a particle physics model, we describe the recoil energy spectrum with general mathematical functions (like a polynomial multiplied by an exponential, or as a series of independent bins). This allows us to find the “best-case” scenario that could possibly reconcile the datasets, free from the constraints of known physics.

Decisive Tension and Contrived Cancellations

The results are striking. In our physically-motivated EFT analysis, the tension between the DAMA/LIBRA dataset and the combined ANAIS-112/COSINE-100 dataset exceeds 5σ. This represents a decisive statistical incompatibility, strongly disfavoring a common dark matter origin for the signals (or lack thereof) in these experiments.

Even more revealing are the results from the model-agnostic approach. To reduce the tension to statistically acceptable levels, the models must be finely tuned in physically bizarre ways. For instance, a common feature of the “best-fit” solutions is a requirement for the modulation signal from sodium recoils to have the opposite sign to that from iodine recoils. This would imply that dark matter particles scatter more frequently off sodium in the summer, while simultaneously scattering more frequently off iodine in the winter. Such a “contrived cancellation” is not a feature of any known dark matter model and highlights the extreme measures needed to reconcile the experimental results.

Conclusion

Our analysis demonstrates that the long-standing DAMA/LIBRA anomaly is not just in tension with experiments using different target materials, but is now in severe tension with experiments using the very same NaI technology. Under the standard assumptions of dark matter physics and astrophysics, the datasets are irreconcilable. While the mystery of the DAMA signal’s origin remains, our work shows that a standard dark matter interpretation is exceptionally unlikely, requiring either unknown, complex detector effects or contrived, physically unmotivated cancellations. The era of testing DAMA with like-for-like experiments is pointing towards a decisive conclusion.

Will Handley

Content generated by gemini-2.5-pro using this prompt.

Image generated by imagen-4.0-generate-001 using this prompt.