Forum Springer Astron. Astrophys.
Forum Whats New Search Orders

Astron. Astrophys. 358, 869-885 (2000)

Previous Section Next Section Title Page Table of Contents

6. The features of the Milky Way SFH

We now can return to the discussion of the meaning of each feature found in the SFH derived in Sect. 3.

6.1. Burst A

The most recent star formation burst is also the most likely burst to have occurred, since it has occurred in the very recent past, and so is less affected by the age errors. A recent enhancement in the SFH is also present in nearly all previous investigations of the SFH (Scalo 1987; Barry 1988; Gómez et al. 1990; Noh & Scalo 1990; Soderblom et al. 1991; Micela et al. 1993; Rocha-Pinto & Maciel 1997; Chereul et al. 1998), and is consistent with the distribution of spectral types in class V stars (Vereshchagin & Chupina 1993). It is not present in the isochrone age distributions (Twarog 1980; Meusinger 1991a) most probably due to the difficulty to measure ages for stars near the zero-age main sequence, where we expect to find the components of this burst in a HR diagram.

We can conclude with confidence that it is a real feature of the SFH. However, being the youngest, it is also the most local feature, because the younger stars have had no time to diffuse to larger distances from their birthsites. Thus, we cannot be sure (from out data only) whether this feature applies to the Milky Way as a whole.

On the other hand, it is known that the Large Magellanic Cloud appears to have experienced also a recent burst of star formation (Westerlund 1990; Alcock et al. 1999) which is very well represented by its young population of open clusters, cepheids, OB associations and red supergiants. At the time of this burst, both galaxies have been closer than ever in their history (Lin et al. 1995). This suggests that burst A could be caused by tidal interactions between our Galaxy and the LMC.

6.2. AB gap

A substantial depression in the star formation rate 1-2 Gyr ago was found by many studies, beginning with Barry (1988; see also the SFH derived from the massive white dwarf luminosity function derived by Isern et al. 1999). This gap appears, although not directly, in the chromospheric age distribution (the so-called Vaughan-Preston gap) and in the spectral type distribution, between A and F dwarfs (Vereshchagin & Chupina 1993). A quiescence between 1 and 2 Gyr is also visible in Chereul et al. (1998), in their study of the kinematical properties of A and F stars in the solar neighbourhood.

This feature has been present in all steps of our work, from the initial age distribution in Fig. 1 to the SFH. Note that the volume corrections have deepened this lull, but it has not changed its duration.

The AB gap is likely to have lasted for a billion years. Previous studies have given a more extended duration for it, but we believe that it was caused by the use of a highly incomplete sample, together with a chromospheric age calibration that does not account for the different chemical composition of the stars. Since it is a relatively recent feature, it only samples birthsites over a radial length scale of 1-2 kpc.

6.3. Burst B

The small lull between the peaks B1 and B2 is not present in the initial age distribution (Fig. 1), appearing only after the volume corrections. It is very narrow, which could be most probably caused by hazardous small weights of the stars in these age bins, during the volume correction. This is why we have presently no means to distinguish burst B from a single burst or an unresolved double burst. At its age of occurrence, considerable broadening of the original features is expected. Either way, our simulations give strong support to this feature.

Previous studies have found star formation enhancements around 4 Gyr ago (Scalo 1987; Barry 1988; Marsakov et al. 1990; Noh & Scalo 1990; Soderblom et al. 1991; Twarog 1980; Meusinger 1991a). Note that a strong concentration of stars around this age can also be found in the age distribution of Edv93's stars, that we show in Fig. 19.

[FIGURE] Fig. 19. Age distribution of Edv93's sample. The concentration of stars around 4 Gyr shows evidence of burst B. The lack of objects with ages lower than 2 Gyr is not an evidence of the AB gap, since Edv93 avoided the inclusion of very young stars due to the difficulty to measure an isochrone age for them.

A significant exception is the SFH found by some of us (Rocha-Pinto & Maciel 1997). This paper suggests that burst B would be much smaller than the preceding burst C. To find the SFH, Rocha-Pinto & Maciel used a method to extract information from the G dwarf metallicity distribution (Rocha-Pinto & Maciel 1996) aided by the AMR (see also Prantzos & Silk 1998). The authors have used several AMRs from the literature and different SFHs were found for each AMR. The SFHs recovered with the AMR from Twarog (1980) and Meusinger et al. (1991) were preferred compared to that found with Edvardsson et al. 1993 (hereafter Edv93) AMR. To be consistent with our present result, we need to compare the present SFH with that coming from Rocha-Pinto & Maciel's method for an AMR similar to that found from our sample (paper I). Our AMR now looks very similar to the mean points of Edv93's AMR. Rocha-Pinto & Maciel (1997) have found, using Edv93's AMR, that Burst B could have around the same intensity as burst C, and also a narrow AB gap lasting 1 Gyr at most. Fig. 20 shows a comparison between their SFH (for Edv93's AMR) and the present history binned by 1 Gyr intervals.

[FIGURE] Fig. 20. Comparison between Rocha-Pinto & Maciel (1997)'s SFH (for Edv93's AMR) and the present chromospherically-based SFR binned by 1 Gyr intervals.

6.4. BC gap and Burst C

The existence of the BC gap is directly linked with how much credit we are going to give to Burst C. From Fig. 15, one could say that no burst could be found around 8-9 Gyr, and all supposed features are artificial patterns created by statistical fluctuations. To reinforce this theoretical expectation, we have done a simulation to show how the features above could be formed by a bursty SFH. We have considered a SFH composed by three bursts, one occurring at 0.3 Gyr, lasting 0.2 Gyr, and the other at 4 Gyr, also lasting 0.2 Gyr, and the last ocurring at 9 Gyr, lasting 0.5 Gyr. The first burst and the last burst are composed by 300 stars, while the second burst is three times more intense. The star formation at other times is assumed to be highly inefficient, forming only 60 more stars at the whole lifetime of the galaxy. The recovered SFR is shown in Fig. 21. Although the two more recent bursts can be well recovered, there is no sign of burst C at 9 Gyr. We have tried other combinations between the amplitude and time of occurrence of them, but in all cases the stars of burst C were much scattered from its original age.

[FIGURE] Fig. 21. Results from an example simulation to show how the observed features: Burst A and B and AB gap could be caused by two bursts separated by a lull of star formation. See text for description of the original SFH used in this simulation.

If on theoretical grounds there is no convincing arguments to accept the existence of burst C, the same does not occur on observational grounds. This puzzling situation comes from the fact that burst C has appeared in a number of studies that have used not only different samples, but also different methods (Barry 1988; Noh & Scalo 1990; Soderblom et al. 1991; Twarog 1980; Meusinger 1991a; Rocha-Pinto & Maciel 1997). And it appears double-peaked in some of them, as we saw in Sect. 3.

The magnitude of the age errors prevents us from assigning a good statistical confidence to this particular feature.

However, it is not implausible that we have overestimated the age errors. A decrease of 0.05 dex in the age errors could alleviate the situation and allow the identification of peaks (although highly broadened) younger than 10 Gyr, which would suggest that burst C is a real feature. A better estimate of the age errors would not create new bursts, or flatten the recovered SFH in these age bins, but would give confidence limits for the ages where the features found are likely to be real and not just artifacts.

6.5. Burst D

The so-called burst D was proposed by Majewski (1993), as a star formation event that would be responsible for the first stars of the disk, before the formation of the thin disk.

A superficial look at Fig. 8 could give us the impression that the peaks beyond 11 Gyr were remnants of this predicted burst. However, as we have shown above, it is presently impossible to recover the SFH correctly at this age range, even if our age errors are overestimated by as much as 0.05 dex. The SFH at older ages are dominated by fluctuations, superimposed on the original strongly broadened structures, in such a way that it is imposible to disentangle statistical fluctuations from real star formation events.

Theoretically, patterns as old as 13 Gyr could be found in the SFH, provided that they occurred not very close to younger ones, if the age errors were decreased by 0.10 dex, but that is hardly possible to be attained at the present moment since it would need to be of the order of magnitude of the error in the [FORMULA] index.

For these reasons, we give no credit to the peaks beyond 11 Gyr in Fig. 8. If burst D has ever occurred, probably the present chromospheric age distribution is not an efficient tool to find its traces.

Previous Section Next Section Title Page Table of Contents

© European Southern Observatory (ESO) 2000

Online publication: June 20, 2000