4. Discussion and conclusions
In this paper I have considered the constraints on the contribution to the halo dark matter (DM) of a population of zero-metallicity, hydrogen-burning stars with mass below (VLM stars).
Though there are already strong constraints on the VLM star scenario, this present work is motivated by 2 considerations: (1) previous studies have adopted colour cuts for their star samples which effectively eliminate any chance they have of discovering zero-metallicity VLM stars; (2) previous studies only consider limits on a smoothly distributed population of VLM stars, where as some theories of baryonic DM formation predict that the stars should be grouped into globular-cluster configurations.
Using data from 20 of the 22 HST fields obtained by Gould et al. (1996 ), together with the photometric predictions of Saumon et al. (1994 ) for zero-metallicity VLM stars, I find that the contribution from a smoothly distributed population of such stars to the halo can be no more than 1.4% at the 95% confidence level on the basis of 75 candidate VLM stars with appearing in the fields when at least 6 310 are predicted. In reality, the true fraction is likely to be less than this value since many of the candidates may belong to the spheroid or disc components. This limit is comparable to previous analyses for stars of non-zero metallicity and therefore the inescapable conclusion is that any smoothly distributed population of VLM stars, regardless of its metallicity, makes at best only a tiny contribution to the halo DM.
Clustering allows the possibility of much larger fluctuations in the expected number of stars appearing in the fields. Additionally, highly compact clusters may not be completely resolvable, thereby decreasing the number of available point sources. These two effects can permit a halo fraction in clusters which is compatible with the halo fraction inferred by the MACHO gravitational microlensing experiment provided that the cluster mass M and radius R satisfy the inequality pc.
One also requires that the cluster scenario satisfies the dynamical constraints which exist on the allowed mass and radius of clusters. Comparison of the dynamically allowed region with the region which satisfies both HST and MACHO observations reveals that only a very small portion of parameter space, characterised by and pc, satisfies all three requirements. However, the implicated cluster mass is intriguingly close to the mass predicted by some DM formation theories (Ashman 1990 ), and to the baryon Jeans mass at the cosmological epoch of recombination.
For a cluster mass , which is required by dynamical arguments, the effect on microlensing statistics is indistinguishable from the unclustered case, and in particular the inferred halo fraction should be within a few percent of the value inferred by assuming the DM distribution to be unclustered. However, MACHO and HST observations place a strong limit on the efficiency with which VLM stars need to be clustered in order to remain compatible with both surveys. This limit corresponds to a 92% present-day clustering efficiency and therefore provides a very stern test of the scenario.
Whilst recent microlensing results seemingly provide strong evidence for a substantial baryonic contribution to Galactic halo DM, when taken together with other observational and theoretical constraints it has become increasingly difficult to provide a unique baryonic candidate which can simultaneously explain the high microlensing fraction and the event timescales. One way out is to invoke a substantial modification in the shape of the halo though, for this to work, the microlensing results require a component resembling something closer to a maximal disc, leaving the status of any halo (and the role of non-baryonic DM on Galactic scales) much reduced. Another option is to attribute the lensing events to some non-baryonic candidate, such as primordial black holes, though in this case one requires an additional non-baryonic candidate to explain the rest of the halo DM. Here I have shown that it is still possible, if only barely, to construct a baryonic scenario which is compatible with all known constraints and which does not require a major modification in the halo dynamics. It is also conceivable that the scenario could provide an explanation for the recent detections of faint extended emission around the edge-on spiral galaxy NGC 5907 (Sackett et al. 1994 ; Lequeux at al. 1996 ) which, under the assumption of constant mass-to-light ratio, appears to trace the distribution of a halo.
© European Southern Observatory (ESO) 1997
Online publication: June 5, 1998