Dust Evolution in Galaxy Cluster Simulations

Gaseous astrophysical media are splattered with solid agglomerates of molecules we call cosmic dust. Observations, in particular, are quite sensitive to dust properties such as composition and grain size. A need emerges to include dust within theoretical models of galaxy and galaxy cluster evolution. The INAF Astronomical Observatory of Trieste has developed a custom state-of-the-art cosmological N-body simulation of galaxy clusters based on the GADGET-3 code. Adding dust to this framework has not been feasible until now. Tracing the continuous grain size (or its discrete approximation) would burden with additional dimensions the already heavy particle structure and calculations, slowing down the runs to impractical rates. Dust was instead treated in post processing Granato+15, and its properties were assumed a priori. Then Hirashita15 proposed an approximation. Instead of computing the grain size continuum, he postulated that dust grains are divided between large (nominally 0.1 micrometers) and small (nominally 0.01 micrometers). He therefore adapted a comprehensive one-zone dust evolution model Asano+13 to this approximation. The binary grain size was selected for both observational and modeling reasons. When dust grains are produced in the envelopes of evolved stars or in supernova remnants, the dominant size is around 0.1 micrometers. In the ISM however, smaller dust is often just as prevalent or at times even dominant. This suggests that ISM evolution alters the grain size distribution. The phenomenon is captured in the modeling. Some processes are most efficient on one grain size over the other, or at times they have opposite effects on each size domain. We successfully adapted the Hirashita15 model to our custom GADGET-3 cosmological zoom-in simulation code, specifically we embedded the model so that each simulated gas particle will trace, on top of the usual gas elements obtained from stellar and supernovae yields, also small and large dust grains. We tested our method on four massive ( two M200 > 3 x 10e14 solar masses and two M200 > 10e15 solar masses) galaxy clusters. We also improved on previous dust production routines, which assumed a fixed dust condensation efficiency for each element. Instead, we form the two most representative dust species observed in nature: carbonaceous dust and astrophysical silicates, based on the element abundance produced by stellar or supernovae yields. At the peak of star formation activity at z > 3 when proto-clusters start to assemble, we find that the gas particles in our simulations are rich in dust, as expected. In order to test the impact of dust processes on dust growth other than stellar production, we ran simulations with dust production and destruction alone, without any grain-gas or grain-grain interactions. Dust is enhanced by a factor of two to three due to the processes occurring in the ISM. We investigated variations of the model through different runs to understand the interdependence of all the processes. We were able to reproduce the dust abundance to metallicity relations observed in local galaxies, however we under-produced the dust content of galaxy clusters around z < 0.5 observed by IRAS, Planck, and Herschel observations. This discrepancy can be mended only by assuming a lower sputtering efficiency, which erodes dust grains in the hot Intracluster Medium (ICM). The abundance of the two dust species, silicates and carbonaceous dust, is also slightly different from the Milky Way average, and from the common values adopted in calculations of dust reprocessing. These differences may have a strong impact on the predicted SED. This method lays the groundwork for further developments, such as cosmological simulations of single galaxies, or the refinement of radiative cooling routines and H2 catalysis on grain surfaces.