1 THE GLOBAL MDF OF THE BULGE

The stars of the bulge of the Milky Way have a broad metallicity distribution, with [Fe/H] between − 3.0 and + 1 dex. Spectroscopic surveys of the bulge, both high and medium resolution, have reported a metallicity gradient in latitude. This was first measured to be about − 0.35 dex kpc⁻¹ along the minor axis of the bulge by Minniti et al. (Reference Minniti1995), and then − 0.6 dex kpc⁻¹ by Zoccali et al. (Reference Zoccali, Hill, Lecureur, Barbuy, Renzini, Minniti, Gómez and Ortolani2008) from observations of about 900 K giants across b = −4° to b = −12° on the minor axis. Ness et al. (Reference Ness2013a) later found a similar minor axis gradient of − 0.45 dex kpc⁻¹ from spectroscopy of about 2 000 red clump giants with R _GC < 3.5 kpc, across b = −5° to b = −10°. Numerous studies (e.g. Hill et al. Reference Hill2011; Babusiaux et al. Reference Babusiaux2010; Gonzalez et al. Reference Gonzalez2011; Uttenthaler et al. Reference Uttenthaler, Schultheis, Nataf, Robin, Lebzelter and Chen2012; Rojas-Arriagada et al. Reference Rojas-Arriagada2014; Johnson et al. Reference Johnson, Rich, Kobayashi and Fulbright2012; Ness et al. Reference Ness2013a; Bensby et al. Reference Bensby2013) have reported similar metallicity distribution functions (MDFs) for the bulge. Figure 1, from Bensby et al. (Reference Bensby2013) compares the MDFs from a number of different studies in various bulge fields. In addition to the vertical metallicity gradient, there is also a weaker gradient in longitude, at |b| > 4°. This is seen in the photometric metallicity distribution by Gonzalez et al. (Reference Gonzalez, Rejkuba, Zoccali, Valent, Minniti and Tobar2013). The vertical abundance gradient was initially interpreted as inconsistent with the formation of the bulge from the disk via dynamical instabilities, which results in a boxy/peanut morphology similar to that already observed in the Milky Way (Dwek et al. Reference Dwek1995). This large abundance gradient was instead interpreted as evidence for a bulge formed via mergers or dissipational collapse (Zoccali et al. Reference Zoccali, Hill, Lecureur, Barbuy, Renzini, Minniti, Gómez and Ortolani2008).

Figure 1. Bulge MDFs from several sources assembled by Bensby et al. (Reference Bensby2013) showing: (a) The MDF for the microlensed dwarf sample (the white-dashed line shows the 26 microlensed dwarf stars from Bensby et al. [Reference Bensby2011]); (b) 166 red giant stars in Baade′s window from Hill et al. (Reference Hill2011) (c) 220 red clump stars in Baade′s window from Hill et al. (Reference Hill2011) (d) 1 813 red giant stars from the ARGOS survey fields at (l, b) = (0°, –5°), (5°, –5°), (–5°,–5°) from Ness et al. (Reference Ness2013); (e) 363 red giants at (l, b) = (0, –10°) from Uttenthaler et al. (Reference Uttenthaler, Schultheis, Nataf, Robin, Lebzelter and Chen2012); The curves are generalised histograms and the dotted lines mark the peaks of the MDF components proposed by Ness et al. (Reference Ness2013a).

Subsequently, however, bulge metallicity gradients in (l, b) like those measured for the Milky Way have been shown to be a natural consequence of bulge formation via instability of the disk, due to the preferential redistribution of stars according to their initial phase space energies in the disk Martinez-Valpuesta and Gerhard (Reference Martinez-Valpuesta and Gerhard2013); Di Matteo et al. (Reference Di Matteo2015). Furthermore, the overall large negative gradient in the mean of the MDF is now believed simply to reflect the changing contribution of populations which have different spatial distributions in the innermost regions. The more metal-rich population of stars is concentrated to the plane, while the more metal-poor populations are present in greater number at larger heights from the plane. Although the various observers find very similar MDFs (see Figure 1), there is no consensus yet in the interpretation of this MDF (see Section 2).

The vertical and (lesser) radial metallicity gradients reported for the MDF in the bulge by numerous studies have been found to flatten for latitudes |b| <5°. This was first reported from studies using small samples of stars near the minor axis (Ramírez et al. Reference Ramírez, Stephens, Frogel and DePoy2000; Rich, Origlia, & Valenti Reference Rich, Origlia and Valenti2007, Reference Rich, Origlia and Valenti2012). From the APOGEE survey, it has been revealed that the flattening of the gradient near the plane is seen right across the bulge in longitude, out to l < 10° and not just on the minor axis, again for latitudes |b| < 5°. This is shown in Figure 2 which includes 13 000 stars from the APOGEE survey and 8 000 stars from the ARGOS survey (those at positive longitudes where APOGEE has observed). These stars are at distances from 4 to 12 kpc from the Sun, thus showing stars in the bulge region (and stars in the disk away from the Sun to illustrate the smooth transition between the bulge and the disk). The mean [Fe/H] of these stars is shown in Figure 2(a), and Figure 2(b) shows the mean [α/Fe]. The ARGOS fields, which each have about 600 stars, are shown in the larger boxes with black outlines for Figure 2(a) only. There are typically ⩾ 30 stars in each of the bins which show the mean measurement for the APOGEE sample.

Figure 2. (a) [Fe/H] and (b) [α/Fe] maps for the 13 000 bulge and disk stars from APOGEE and 8 000 ARGOS bulge stars (in the larger outlined boxes for the [Fe/H] map only) spanning heliocentric distances of 4–12 kpc. The dashed line indicates approximate outline of the 180 pc thin bar identified by Wegg, Gerhard, and Portail (Reference Wegg, Gerhard and Portail2015) and the larger box represents the approximate outline of the boxy bulge in the COBE image (Dwek et al. Reference Dwek1995).

From Figure 2(a), it is clear that in the inner region with (l,|b|) < (10°, 5°) the metallicity is nearly constant at constant latitude with the exception of stars near the plane (|b| < 2°) at l = 5–10°. The most metal-rich and correspondingly alpha-poor stars are concentrated to the layer |b| < 2°, and these metal-rich stars show a smooth transition out into the disk. In the plane, the stars reach the highest metallicity at the end of the thin long bar reported by Wegg et al. (Reference Wegg, Gerhard and Portail2015), the profile of which is shown by the thin long box in Figure 2, extending out to l = 30°. The outline of the COBE boxy/bulge profile is shown in this Figure as the larger box, going out to longitudes of l ≈ 10°. The metallicity measurements for b > 4° from APOGEE and ARGOS are in good agreement with the photometric metallicities of Gonzalez et al. (Reference Gonzalez, Rejkuba, Zoccali, Valent, Minniti and Tobar2013).

2 MULTIPLE STELLAR POPULATIONS IN THE BULGE

Several studies (e.g. Babusiaux et al. Reference Babusiaux2010; Hill et al. Reference Hill2011; Gonzalez et al. Reference Gonzalez2011; Freeman Reference Freeman, Aoki, Ishigaki, Suda, Tsujimoto and Arimoto2012; Rojas-Arriagada et al. Reference Rojas-Arriagada2014) have interpreted the MDF of the bulge across a broad region in (l, b) as being made up of multiple populations. Ness et al. (Reference Ness2013a) interpret the MDF as having five populations which are shown in Figure 3: 3 dominant populations (A–C) with [Fe/H] > −1 and two minor metal-poor populations (D & E). From the ARGOS bulge survey described in Ness et al. (Reference Ness2013a), the three populations A–C with [Fe/H] > −1 have peaks at metallicities of about +0.15, − 0.25, − 0.7 dex, respectively and provide about 95% of stars in the bulge. They associate these populations with the stars of the boxy/peanut bulge (A and B), the thick disk (C), the metal weak thick disk (D) and stellar halo (E). They find that these populations are present in different proportions throughout the inner region of the Milky Way.

Figure 3. The MDF for stars within R _GC < 3.5 kpc from the ARGOS survey: (a) for stars at b = −5°, (b) stars at b = −7.5°, and (c) stars at b = −10°, for all stars across longitudes |l| < 15° showing the changing contribution of metallicity fractions with latitude. The Gaussian components A–E are indicated.

Using red clump stars, Ness et al. (Reference Ness2012) and Uttenthaler et al. (Reference Uttenthaler, Schultheis, Nataf, Robin, Lebzelter and Chen2012) showed that only the more metal rich stars in populations A and B (i.e. stars with [Fe/H] > −0.5) are part of the split density distribution of stars in the innermost region that reflects the X-shaped morphology of the bulge (e.g. Ness and Lang Reference Ness and Lang2016). This implies that stars with [Fe/H] < −0.5 that are present in the inner region are not part of the boxy/peanut bulge morphology and are not on X-profile-supporting x1 orbit families. The most metal-rich stars show the largest minima between peaks in the K-magnitude distribution of stars and are the most strongly involved in the X-shape (Ness et al. Reference Ness2012). Population A was found by Ness et al. (Reference Ness2013a) to be concentrated toward the plane and to be the thinner part of the boxy/peanut bulge. Population B corresponds to the vertically thicker stars in the bulge, with a similar contribution fraction across b = 5 to 10°. Population C is not significantly involved in the split density distribution or boxy/peanut structure but transitions smoothly from the bulge to disk with longitude; it is identified with the inner thick disk. The two populations of stars (A and B) associated with the boxy/peanut structure have similar peaked velocity dispersion profiles with longitude. The velocity dispersion profiles for the more metal-poor stars have a different shape (e.g. Shen et al. Reference Shen, Rich, Kormendy, Howard, De Propris and Kunder2010; Ness et al. Reference Ness2013b; Portail et al. Reference Portail, Wegg, Gerhard and Martinez-Valpuesta2015) (see below). The 5% of stars in the bulge with metallicities [Fe/H] < −1.0 are chemically similar to stars of the metal-weak thick disk and halo near the Solar neighbourhood (e.g. Alves-Brito et al. Reference Alves-Brito, Meléndez, Asplund, Ramírez and Yong2010; Bensby et al. Reference Bensby2013) and were associated by Ness et al. (Reference Ness2013a) with these populations.

The rotation and dispersion profiles reported by Ness et al. (Reference Ness2013b) and shown in Figure 4 as a function of [Fe/H] support the differentiation of the populations. The kinematics of stars in populations A & B are related and are different from the other components. The stars in populations A and B, which are part of the split clump and the boxy/peanut bulge, show the same characteristic peaked pattern of velocity dispersion in the two left hand panels of Figure 4, with population A being a colder replica of population B. Population C, associated with the thick disk in the inner Galaxy, is rotating as rapidly as the more metal-rich populations (all show the latitude-independent rotation profiles that are commonly seen in boxy bulges) but its velocity dispersion profile has a different shape. The most metal-poor stars with [Fe/H] < −1 have the slow rotation profile and high dispersion that might be expected of a population that had no ancestral association with a disk, such as a spheroidal population in the bulge or the stars of the inner halo or an underlying merger component.

Figure 4. The rotation curves (top) and dispersion profiles (bottom) for the 17 500 bulge stars from the ARGOS survey within distances 5–11 kpc. The populations from most metal rich to most metal poor indicated in Figure 3 correspond to populations A,B,C, and D/E. (Ness et al. Reference Ness2013b).

Other studies (Babusiaux et al. Reference Babusiaux2010; Hill et al. Reference Hill2011; Gonzalez et al. Reference Gonzalez2011; Rojas-Arriagada et al. Reference Rojas-Arriagada2014) have proposed that the MDF of the bulge comprises two populations, a metal-rich population that is part of the boxy/peanut bulge and a metal-poor population that is an old spheroid (i.e. with a possible formation history that is distinct from the disk and halo components of the Milky Way). Although note that Gonzalez et al. (Reference Gonzalez2015) concludes that whether their two MDF components have different formation histories is yet to be determined. The decomposition of stars into two roughly equal components from Gonzalez et al. (Reference Gonzalez2015) is shown in Figure 5, from the GIBS survey (Zoccali et al. Reference Zoccali2014), where these two populations peak at [Fe/H] of about +0.25 dex and − 0.3 dex.

Figure 5. From Gonzalez et al. (Reference Gonzalez2015): The MDF obtained from the combination of four GIBS (high-resolution) fields and the red clump stars from Hill et al. (Reference Hill2011). The probability density distribution is shown in the upper panel as a dashed line, with the variance on the probability densities in blue. The lower panel shows the best two Gaussian fit to the upper panel distribution.

The two-component decomposition places fewer stars within the boxy/peanut or X-shaped structure, attributing approximately 50% to an old spheroid (Hill et al. Reference Hill2011). The five-population model of Ness et al. (Reference Ness2013a) attributes 95% of stars to be disk stars, with the population C (at [Fe/H] = -0.7) being disk stars but simply not part of the X-shaped morphology (possibly because they were originally part of a hotter thick disk which was dynamically less responsive to the instability). These two interpretations therefore have different implications for the stars in the inner region in terms of their origin from the disk and the contribution of any additional population that is distinct from any other Milky Way population and specifically from the bulge, such as an old spheroid formed via mergers at high redshift. Whereas a large fraction, up to 50% of stars in the two-component decomposition, are associated with an old-spheroid component, only ~ 5% of the population in the five-population decomposition is associated with any old spheroid component, and this component may simply be metal-poor halo stars in the inner region and not a unique population of the bulge.

3 THE ALPHA-ENHANCEMENT OF THE BULGE

The alpha-enhancement of stars in the bulge shows similar trends to that of the thin and thick disk in the solar neighbourhood, as reported by Alves Brito et al. (Reference Alves-Brito, Meléndez, Asplund, Ramírez and Yong2010), Bensby et al. (Reference Bensby2013), and Gonzalez et al. (Reference Gonzalez2015).

The APOGEE survey includes observations across the Milky Way's disk and bulge and allows the [α/Fe] trends to be homogeneously mapped from the outermost to the innermost region of the Galaxy (see e.g. Hayden et al. Reference Hayden2015; Nidever et al. Reference Nidever2014). Comparing the [α/Fe] trends and the individual abundance patterns ([X/Fe]) for bulge stars with stars at other galactic radii is important for understanding the detailed formation of the bulge and its relationship with the disk. A large fraction of the bulge stars is likely to have come from the disk, and therefore the broad expectation would be that disk and bulge abundance trends will be similar. However, there are a number of perturbing processes that may show abundance signatures. These include the star formation rates in the inner and outer disk and the growth of the disk over time, gas inflow driving star formation, and radial migration. The alpha and abundance trends of the rare metal-poor stars in the inner Milky Way are particularly interesting due to the prediction that the oldest stars of the Galaxy will now be concentrated to the bulge (e.g. Doménech-Moral et al. Reference Doménech-Moral, Martínez-Serrano, Domínguez-Tenreiro and Serna2012; Diemand et al. Reference Diemand, Kuhlen and Madau2008; Tumlinson Reference Tumlinson2010).

Figure 6 shows the [Fe/H]–[α/Fe] maps for the APOGEE stars for different Galactic zones in R and z. The stellar parameters have been obtained with The Cannon (Ness et al. Reference Ness, Hogg, Rix, Ho and Zasowski2015) and demonstrate that the inner Galaxy has the narrowest alpha sequence, which is offset to higher metallicity than in the outer galaxy. The dotted line in Figure 6 marks the approximate trend of the high-alpha sequence that is seen in the innermost region. The density distribution shows that, in the zones further from the Galactic plane at a given radius, more stars are found in the low metallicity/high alpha sequence than in the high metallicity/low alpha sequence.

Figure 6. The [Fe/H]–[α/Fe] for the Milky Way from the Galactic center to R _GC = 13 kpc. Three intervals of height z are shown, at increasing radii, for 0 < z < 0.5 kpc, 0.5 < z < 1 kpc and 1 < z < 2 kpc, for the bottom, middle, and top panels respectively. The number of stars is indicated in the left hand corner of each sub-panel. The dashed line shows the sequence in the inner bulge and is included to help follow the trends in the [α/Fe] sequence with increasing radius. Note that the alpha sequence narrows at small galactic radii in the bulge, the low and high alpha sequences are both present near the sun, and only the low-alpha sequence is present at the largest radii in the disk.

Figure 6 shows that there is a smooth transition from the outer to inner region at a given height from the plane, with an increasing mean metallicity. The low-alpha sequence dominates in the outer region and a narrower sequence extending to lower [Fe/H] and higher [α/Fe] dominates in the inner region. At intermediate radii, near the Sun, both low and high alpha sequences are present. The narrower and high-alpha sequence seen in the inner Galaxy implies that the star formation and chemical evolution rate was high at the early epoch in the disk at which the bulge stars were forming. The star formation continued over a relatively brief period, and this single sequence is a snapshot in time of the rapid chemical evolution of the early disk from which the bulge was formed. Athanassoula (Reference Athanassoula2005) showed that bulges form early on after the disk, and over a short period of 1–2 Gyr. This rapid formation at early times is consistent with the single continuous alpha sequence seen in the inner 3 kpc of the Milky Way that is observed in APOGEE data. The simulations of Di Matteo et al. (Reference Di Matteo2015) show that stars are mapped into the bulge from the disk according to their initial phase space location. Stars now in the bulge that reside nearest to the plane originate from the innermost region of the disk. Similarly, the stars in the bulge whose orbits take them to higher z came from larger radii in the disk and preferentially at larger heights above the plane. Simulations that include star formation (e.g. Obreja et al. Reference Obreja, Domínguez-Tenreiro, Brook, Martínez-Serrano, Doménech-Moral, Serna, Mollá and Stinson2013; Calura et al. Reference Calura2012) combined with new observational data e.g. from APOGEE which covers from the bulge and into the disk may offer insight into the observed chemodynamical distribution of stars across (l, b).

4 THE METAL-POOR [Fe/H] < –1.0 POPULATION

About 5% of the stars in the bulge have metallicities < −1.0. The ARGOS survey showed that the fraction of these stars, which they associated with the Galactic halo population, increases with height above the plane. This population also rotates more slowly, at about 50% of the speed of more metal-rich stars (Ness et al. Reference Ness2013b). It also has a high and relatively latitude-independent velocity dispersion as seen in Figure 4. Kunder et al. (Reference Kunder2015) have also suggested that the halo population in the bulge is not insignificant, even within 1 kpc of the Galactic center. Bensby et al. (Reference Bensby2013) measured the individual abundances of microlensed dwarf bulge stars and found that the stars with [Fe/H] ≈ –1.0 have very similar chemical abundance trends to the nearby thick disk stars.

The bulge RR Lyrae population, whose MDF peaks around [Fe/H] = −1.0 (Pietrukowicz et al. Reference Pietrukowicz2015), has been used to trace the distribution of the old metal-poor population of stars in the bulge. The RR Lyrae stars are ideal tracers of structure because their distances can be accurately estimated. Although Dékány et al. (Reference Dékány, Minniti, Catelan, Zoccali, Saito, Hempel and Gonzalez2013) reported the spatial distribution of the RR Lyrae to be spheroidal with only a slight elongation, the survey of Pietrukowicz et al. (Reference Pietrukowicz2015) showed that the metal-poor RR Lyrae very closely trace the barred structure of intermediate-age red clump giants. Pietrukowicz et al. (Reference Pietrukowicz2015) show that the spatial distribution of the RR Lyrae population is barred but that these stars are not part of the X-shape structure. They also report two sequences of RR Lyrae with marginally different metallicities, proposing evidence for the role of mergers in the initial formation of the bulge. Although they may follow the elongated distribution of the bar, the RR Lyrae stars may still be part of the halo but which has been pulled into a bar shape by the potential of the other stars (e.g. Saha and Gerhard Reference Saha and Gerhard2013).

The high-resolution detailed chemical abundance analysis of a handful of the most metal-poor bulge stars with [Fe/H] < −2.0 by Howes et al. (Reference Howes2014) and García Pérez et al. (Reference García Pérez2013) reported similar individual element enhancement and alpha-enhancement trends to halo stars of the same metallicity (although Howes et al. (Reference Howes2014) noted an abundance scatter that may be larger). From the analysis of the dynamics and detailed chemical abundances of three bulge stars with [Fe/H] < −2.7, Casey and Schlaufman (Reference Casey and Schlaufman2015) report that these stars follow the abundance trends identified previously for metal-poor halo stars, except for scandium. Simulations (e.g. Brook et al. Reference Brook, Kawata, Scannapieco, Martel and Gibson2007; Diemand et al. Reference Diemand, Kuhlen and Madau2008; Tumlinson Reference Tumlinson2010) predict that the oldest stars in the Galaxy will be concentrated to the bulge. These very metal-poor stars found by Casey and Schlaufman (Reference Casey and Schlaufman2015) may well be members of the oldest populations of stars in the Milky Way, formed at redshift z > 10. The abundance patterns of this rare stellar population which is concentrated to the bulge could provide critical insight into the composition of the universe at the highest redshift and the conditions at the beginning of the Milky Way's formation.