1997A&A...322..709K

Hubble Space Telescope (HST) limits on the amount of halo dark matter (DM) in the form of very low-mass (VLM) stars close to the hydrogen-burning limit have been interpreted as excluding such stars as viable candidates. However, these limits assume that the stars are smoothly distributed and have at least the metallicity of Population II stars, whilst some baryonic DM formation theories predict that they may instead be clumped into globular-cluster configurations and have close to zero metallicity. I re-analyse the HST data employing the zero-metallicity VLM star models of Saumon et al. (1994), which predict V-I colours below the cuts of previous analyses for stars below 0.2M_☉. From the models I derive new limits on the allowed halo fraction comprising VLM stars for both the unclustered and clustered cases. In the unclustered regime I find a 95% confidence upper limit on the allowed halo fraction of 1.4% inferred from 20 HST fields, comparable to limits derived by previous studies for non-zero metallicity populations. In the cluster scenario I show that clusters of mass M and radius R can satisfy both HST and the recent MACHO gravitational microlensing results, which indicate a lens halo fraction of 40% for a standard halo model, provided R≲1.2(M/10⁴M_☉)^0.74pc. However, existing dynamical limits restrict the allowed range to a tiny region characterised by M∼4x10⁴M_☉ and R∼3pc. Furthermore, consistency between MACHO and HST demands a present-day clustering efficiency of 92% or better. Intriguingly however, the cluster mass implied by these limits is theoretically well motivated and the VLM star scenario may also help to provide an explanation for the faint red `halo' light recently reported around another galaxy.