6. The role of the Galactic Bar
There is by now substantial evidence that the Milky Way hosts a small Bar in its inner 3 kpc or so. The idea was originally suggested to explain the kinematics of the atomic and molecular gas near the Galactic Centre (de Vaucouleurs 1964, Peters 1975, Liszt & Burton 1980, Mulder & Liem 1986). In recent years further evidence for a Galactic Bar has piled up from several tracers: gas dynamics (Binney et al. 1991; Weiner & Sellwood 1996, 1999; Yuan 1993; Wada et al. 1994; Englmaier & Gerhard 1999; Fux 1999), IR photometry and star counts (Blitz & Spergel 1991; Weinberg 1992; Nikolaev & Weinberg 1997; Dwek et al. 1995; Binney et al. 1997; Unavane & Gilmore 1998), stellar kinematics (Zhao et al. 1994; Weinberg 1994; Fux 1997; Sevenster 1997, 1999; Sevenster et al. 1999; Raboud et al. 1998), OGLE data (Stanek et al. 1994, 1997; Paczynski et al. 1994; Evans 1994; Zhao et al. 1995, 1996; Zhao & De Zeeuw 1998, Ng et al. 1996). For a review on the Galactic Bar see e.g. Gerhard (1996, 1999). The determination of the characteristic parameters (size, axis ratio, rotation speed, orientation and so forth) is even more difficult for the Bar of our own Milky Way than for external galaxies. Broadly speaking, the various studies mentioned above indicate a Bar with a major axis of 2-4 kpc viewed at an angle of 15-45o in the first longitude quadrant, an axis ratio around 3:1 and a pattern speed km sec-1 kpc-1.
PC99 underlined that the dynamical influence of the Galactic Bar is likely to account for the peak at 4 kpc displayed by the gas profile in the Disc, which static chemical models are unable to reproduce if other constraints, like the observed metallicity gradient, are to be matched as well. In fact, gravitational torques in a barred, or non-axisymmetric, potential are thought to induce gas accumulation and formation of rings at the corresponding Lindblad resonances (e.g. Combes & Gerin 1985; Schwarz 1981, 1984). In brief, bar-induced flows sweep gas away from the co-rotation (CR) radius, where the bar roughly ends, toward the inner and outer Lindblad resonances (ILR, OLR). In fact, we developed our new chemical model with radial flows also with the aim to mimic such effects of the Bar upon the gas distribution, by simulating suitable flow velocity profiles.
As mentioned above, though the existence and gross features of the Galactic Bar are by now established, there is no general agreement on details like its size and pattern speed, and on the corresponding radii for its CR and ILR, OLR. In this paper, with the aim to reproduce the molecular ring at 4-6 kpc, we will consider two Bar models covering the range of scenarios suggested in literature:
In any case, we will assume here that the Bar influences only the inner 5-6 kpc of the Galactic Disc, where its OLR is supposed to lie at the outermost according to current understanding, while leaving regions outside the OLR unaffected (see also Gerhard 1999). Actually, the possible influence of the Bar over a larger Disc region than its formal extent is still an open problem, as we will comment upon in the final conclusions (Sect. 7).
In the framework of our chemical model, the inclusion of the Bar translates into imposing a suitable velocity profile for the radial flows in the inner regions of the disc. Namely, we run the "successful models" with radial flows presented in Sect. 5, but at a suitable age the Bar is assumed to form and the radial velocity profile is altered correspondingly, by modifying the coefficients , , describing the radial flow pattern (Sect. 3). In case A, we will impose fast radial inflow velocities within CR at 3-4 kpc to mimic the rapid drift of the gas toward the ILR (Sect. 6.1). In case B, we will impose outflows from CR to the OLR around 4.5 kpc (Sect. 6.2).
As to the age of the Bar, an upper limit is set by the typical age of its stellar population, 8-9 Gyr (Ng et al. 1996); an age of 5 to 8 Gyr has been suggested by Sevenster (1997, 1999). The results from our simulations turned out to be quite insensitive to a change in the Bar's age from 9 to 5 Gyr; therefore, we will present simulations with a 5 Gyr old Bar () as representative of the generic case of age Gyr. We will also consider, for the sake of completeness, the case of a much younger Bar of 1 Gyr of age ().
6.1. Modelling the effects of the Bar: case A
To mimic case A, where the Bar induces fast inflows from CR to its ILR, the inflow velocity is typically increased for kpc (the Bar's CR radius) with respect to the drift pattern adopted before the onset of the Bar. Models corresponding to case A give good results when combined with a Schmidt SF law with radial inflows. Starting from the corresponding "successful" model S15RFe of Sect. 5 and changing the velocity law as in Fig. 9 at the onset of the Bar, we obtained the models shown in Fig. 10, compared to the original model S15RFe with no Bar's effects included. Obviously, if the Bar is younger ( Gyr) faster induced inflows need to be assumed in the inner regions so that the observed sharp dip in the gas profile at kpc is obtained in a shorter time (Fig. 9). Anyways, no extreme speeds need to be induced by the Bar ( km sec) at these radii yet, to resemble the observed gas profile, so the models remain plausible.
This type of solution actually corresponds to the one originally suggested by Lacey & Fall (1985), who in fact assumed a Schmidt SF law, radial inflows of the order of -1 km sec-1 to reproduce the metallicity gradient, and for kpc a raise in the inflow speed up to -10 km sec-1 to reproduce the gas profile. Without such a spike in the inflow velocity, the gas distribution keeps rising inward, with no depression (cfr. model S15RFe). This picture also corresponds to the situation for viscous models as suggested by Thon & Meusinger (1998): the detailed gas profile of the Disc can be reproduced only by artificially increasing the viscosity in the inner regions, so as to mimic the influence of the Galactic Bar.
Case A can also well combine with model O10RFe , namely with an Oort-type SF law with plus the corresponding "successful" inflow pattern. In fact, also in model O10RFe (which is actually the only model able to predict a peak in the gas profile reminding of the molecular ring even without including the Bar, see Sect. 5) the gas profile rises inward, and by increasing the inflow velocities inside 4 kpc one can fit the observed gas depression. As an example, we show models O10RF with inflow patterns as in Fig. 11 ; the corresponding results are plotted in Fig. 12 together with the base model O10RFe .
Notice how in all these "case A" models the metallicity gradient is only negligibly affected by the switch of the inner inflow pattern at the time of onset of the Bar, ( Gyr is the present Galactic age). The effects are at most limited to kpc, where they are hard to check from observations, since in that region the gas distribution is depressed and the metallicity tracers are missing as well, since they are young objects strongly correlated to present-day SF activity and therefore to the presence of gas. We remark that abundance data at in the plots for the abundance gradient are to be disregarded as a constraint for the model, since they refer to the Galactic Centre population, not to the Disc (see PC99).
6.2. Modelling the effects of the Bar: case B
According to what we labelled above as "case B", the Galactic Bar ends in correspondence to its CR around 2.5 kpc, and it has an OLR around 4.5 kpc. Therefore, the gas is expected to drift from CR outward and accumulate at the OLR, while gas inflowing from outer regions slows down and accumulates as well at the level of the resonance at 4.5 kpc.
Models for case B will adopt, starting from , a radial flow pattern with positive velocities (outflow) from kpc to kpc, and negative velocities (inflow) from outside dropping to zero at kpc. Case B can be combined with model DRRFe , namely with the SF law by Dopita & Ryder (1994) and the corresponding "successful" overall inflow pattern (Sect. 5). The relevant models with their detailed velocity patterns, for the usual two values of , are shown in Figs. 13 and 14. Notice once again that rather low drift velocities ( km sec) suffice to reproduce the gas peak, which reinforces the plausibility of the models.
In case B models, the metallicity gradient is clearly affected in the inner regions by the switch in the gas flow pattern, at least if this lasted for some time (Fig. 14, case ). This feature, though, is again hard to check from observations since there are no tracers of Disc metallicity for kpc (see Sect. 6.1).
Case B can also reproduce the observed gas profile combined with an Oort-type SF law with (model O10RFe), especially provided the Bar formed recently, as displayed in Fig. 16 with . If the Bar-induced flows activate much earlier, the predicted peak is very sharp and narrow (Fig. 16, case Gyr), due to the milder sensitivity of SF to the gas density in this case (Schmidt-like exponent ): with respect to other SF laws, this SF is relatively less efficient where the gas density is high, namely where the gas accumulates around 4.5 kpc, while it is relatively more efficient where the density drops, below 4 kpc. Such a sharp peak seems in contrast with the observed, quite broad distribution (the observational uncertainty on the gas density profile in the inner Galactic region is less than a factor of 2, Dame 1993). Models O10RF with an "old" Bar are therefore less appealing in case B.
6.3. Concluding remarks
We referred to the current understanding of the structure and features of the Galactic Bar to simulate its dynamical influence on the surrounding gas flows and distribution. Chemical models accounting for the effect of the Bar make it finally possible to reproduce the gas profile properly, which could not be accomplished by static models (PC99).
Broadly speaking, two main scenarios are presented.
Within these simplified models it is unfeasible to discuss any further on the scenarios for Bar structure and age, and related gas flows. Only detailed dynamical simulations for Bar formation, evolution and potential can tell how the molecular ring consequently formed, which is beyond the goals of this paper (see also Sect. 7). Here we were just interested in showing how even simple qualitative models for Bar-induced gas flows in the inner disc solve in fact the puzzle encountered in PC99. Namely, they can reproduce at the same time both the metallicity gradient and the gas distribution, in particular the peak corresponding to the molecular ring around 4 kpc. This can be accomplished already with quite slow, and largely plausible, flow velocities. The present modelling therefore provides a simple tool for qualitative understanding of possible behaviours.
Regardless of details, however, one condition is necessary for the scheme to work: there must be enough gas in the inner regions of the Disc, which the Bar can then "shape" to resemble the detailed observed distribution. This favours chemical models with radial inflows where the metallicity gradient can coexist with high gas fractions in the inner shells. Our simulations showed that no Bar-induced gas flow superimposed on otherwise static models (as those by PC99) can produce the observed gas peak: whatever the assumed age of the Bar or gas velocity profile, there is not enough gas left in the inner Galactic regions if we are to reproduce the metallicity gradient as well. Gas must be continuously replenished by inflows from outer regions; that's why we presented here "barred" models based only on the "successful" models with radial inflows from Sect. 5 and never on the static models of PC99.
© European Southern Observatory (ESO) 2000
Online publication: March 21, 2000