Theorists aim to present a model that explains three basic observational facts:
Young stars are difficult to satisfy these facts and thus thought unlikely to be the main driver of the UV upturn. The focus has been on how an old population can develop hot stars. Post-AGB stars (central stars of planetary nebulae) are too short-lived and more fatally too hot most of their lifetime, hence violating item 3. There is a good consensus that hot (low-mass) horizontal-branch (HB) stars are the more natural candidates. Here I introduce two classical solutions based on the HB hypothesis.
3.1. Metal-poor HB hypothesis
It is widely known that metal-poor HB stars can be hot and make good UV sources when they are old (e.g., Lee et al. 1994). Thus, the first scenario was naturally that an order of 20% of the stellar mass of bright elliptical galaxies are extremely old and metal-poor populations (Park & Lee 1997). The strength of this scenario is that the oldest stars in a galaxy are likely the most metal-poor and to be in the core, where the UV upturn is found to be strong. In this scenario, the UV vs Mg2 relation does not present any causality connection but simply a result of tracing different populations in terms of metallicity. Mg2 is exhibited by the majority metal-rich stars while the UV flux is dominated by the old metal-poor stars. The narrow range of temperature is easily explained as well. On the other hand, the mass fraction of order ~ 20% is too high by the standard galactic chemical evolution theory. Canonical models suggest the metal-poor fraction of 10%. If metal-poor stars are present at such a high level, there must also be a large number of intermediate-metallicity (20-50% solar) stars, which will make galaxy's integrated metallicity too low and integrated colours too blue, compared to the observed values. Moreover, the age of the oldest stars, i.e. the main UV sources, is required in this scenario to be 20-30% older than the average Milky Way globular clusters (Yi et al. 1999). This would pose a big challenge but there may be a rescue (see Section 4).
3.2. Metal-rich HB hypothesis
Through a gedanken experiment Greggio & Renzini (1990) noted a possibility that extremely low-mass HB stars may completely skip the AGB phase and dubbed it "AGB Manqúe stage". Through this stage metal-rich populations could become UV bright. This is particularly effective for a high value of helium abundance (Dorman et al. 1995). If galactic helium is enriched with respect to heavy elements at a rate of Y / Z 2.5 this means that the stage would be very effective in galaxy scales as well (Horch et al. 1992). It could be similarly effective if the mass loss rate in metal-rich stars is 30-40% higher than that of metal-poor stars (Yi et al. 1997a). Either of the two conditions would be sufficient while they can also complement each other. Both of these conditions are difficult to validate empirically but plausible (Yi et al. 1998). In this scenario, metal-rich stars may become UV bright in two steps: (1) they lose more mass on the red giant phase due to the opacity effect and become low-mass HB stars, and (2) extremely low-mass HB stars stay in the hot phase for a long time and directly become white dwarfs, effectively skipping the red, asymptotic giant phase (Yi et al. 1997a, 1997b). This scenario reproduces most of the features of the UV upturn (Bressan et al. 1994; Yi et al. 1998). The UV vs Mg2 relation is naturally explained as a UV vs metallicity relation. However, its validity heavily hinges upon the purely-theoretical (and hence vulnerable to criticisms) late-stage stellar evolution models of metal-rich stars.
3.3. Metal-poor or metal-rich HB?
Both of these scenarios are equally appealing but their implications on the age of bright elliptical galaxies are substantially different. The metal-poor hypothesis suggest UV-upturn galaxies are 30% older than Milky Way and requires the universe to be older than currently believed, suggesting a large cosmological constant. The metal-rich hypothesis on the other hand suggests that elliptical galaxies are not necessarily older than the Milky Way halo.
Figure 2. The two classic models (Model A: metal-poor HB, Model C: metal-rich HB) predict different evolution history. While precise calibrations are difficult, the UV developing pace is in general predicted to be faster for more metal-rich populations. Excerpted from Yi et al. (1999).