A new study of diffuse dwarf galaxies is challenging the prevailing galaxy formation model within the standard Cold Dark Matter (CDM) framework, leading to a proposed new model of dark matter.

Under the direction of Prof. WANG Huiyuan from the University of Science and Technology of China (USTC) of the Chinese Academy of Sciences, the research team identified for the first time an exceptionally strong clustering pattern in diffuse dwarf galaxies.

The study was published in Nature.

Dwarf galaxies, like all galaxies, sit within a halo of dark matter. These halos formed early in the universe and shaped where galaxies could form.

Nevertheless, not all dark matter halos are the same. Some are more likely to be found in denser regions of the universe than others. This is called “halo bias” and comes in two types—“mass bias,” which holds that massive halos cluster more strongly, and “assembly bias,” which holds that among halos of the same mass, those with different halo properties exhibit different clustering. For example, the halos formed earlier (old halos) cluster more strongly than those formed later (young halos).

Historically, massive galaxies were the primary focus for detecting halo assembly bias, due to their higher luminosity and more efficient observability by surveys such as the Sloan Digital Sky Survey (SDSS). In contrast, dwarf galaxies have often been underrepresented in such studies because of their low luminosity and the challenges associated with sparse sampling.

However, the USTC researchers have revealed that dark matter halos hosting dwarf galaxies also exhibit halo bias, which is largely unaffected by uncertainties in halo mass estimations. This finding suggests that halo assembly bias may be more effectively traced through dwarf galaxies compared to their more massive counterparts.

In this study, Prof. WANG’s team analyzed a sample of isolated dwarf galaxies from the SDSS, revealing that diffuse dwarf galaxies—whose stars are farther apart—display unexpectedly strong large-scale clustering compared to compact dwarf galaxies—whose stars are closer together. This unexpected finding fundamentally contradicts the established understanding of galaxy clustering derived from studies of massive galaxies.

Through their proprietary Exploring the Local Universe with reConstructed Initial Density field (ELUCID) cosmological simulation, the researchers found that this “inverted” phenomenon was intrinsically linked to the formation time of halos. Specifically, the spatial distribution of diffuse dwarf galaxies closely aligned with old halos, while compact dwarf galaxies followed patterns similar to young halos. This represents the first high-confidence observational evidence for halo assembly bias based on real-world data, bridging the gap between cosmological simulations and empirical validation.

However, existing galaxy formation models under the standard CDM paradigm fail to explain the formation of diffuse dwarf galaxies in old halos, implying potential contradictions between current galaxy formation models and dark matter models, on the one hand, and the actual Universe, on the other. To overcome this contradiction, the researchers introduced the Self-Interacting Dark Matter (SIDM) model. 

This model posits that dark matter particles interact not only via gravity, but also via weak non-gravitational interactions. These interactions cause structural expansion and weaken the central gravitational strength in old halos, thereby promoting the formation of diffuse dwarf galaxies. Conversely, young halos exhibit weaker such effects, favoring the formation of compact dwarf galaxies. This theory well explains the observed correlation between halo age and galaxy density, suggesting that the nature of dark matter may be more complex than previously thought.

Reviewers from Nature highly commended this work: “This is an original and very surprising (and thus significant) result. Testing predictions of dark matter self-interactions through galaxy clustering is a novel approach and could have a lasting impact.”

This work represents the first observational confirmation of significant halo assembly bias—a breakthrough that defines critical parameters for modeling the nature of dark matter, the evolution of cosmic large-scale structures, and the mechanisms governing galaxy formation and evolution. It reveals a unique correlation between the structures of baryonic components and the ages of their host halos in dwarf galaxies, fundamentally challenging the standard CDM paradigm and necessitating potential modifications.