2016MNRAS.461.3001T

The existence of long (>100 kpc) H I streams and small (<20 kpc) free-floating H I clouds is well known. While the formation of the streams has been investigated extensively, and the isolated clouds are often purported to be interaction debris, little research has been done on the formation of optically dark H i clouds that are not part of a larger stream. One possibility is that such features result from the fragmentation of more extended streams, while another idea is that they are primordial, optically dark galaxies. We test the validity of the fragmentation scenario (via harassment) using numerical simulations. In order to compare our numerical models with observations, we present catalogues of both the known long H i streams (42 objects) and free-floating H i clouds suggested as dark galaxy candidates (51 objects). In particular, we investigate whether it is possible to form compact features with high velocity widths (>100 km/s), similar to observed clouds which are otherwise intriguing dark galaxy candidates. We find that producing such features is possible but extremely unlikely, occurring no more than 0.2% of the time in our simulations. In contrast, we find that genuine dark galaxies could be extremely stable to harassment and remain detectable even after 5 Gyr in the cluster environment (with the important caveat that our simulations only explore harassment and do not yet include the intracluster medium, heating and cooling, or star formation). We also discuss the possibility that such objects could be the progenitors of recently discovered ultra diffuse galaxies.