The disk of the Milky Way contains both a thin layer of young stars and a thicker layer of old stars. For a long time they were described as separate components, with intermediate age stars being included as part of the "thin" disk (Gilmore & Reid 1983; Liu & Chaboyer 2000; Munn et al. 2004; Juric et al. 2008; Ivezic et al. 2008). However, Bovy et al. (2012a) suggested that there is no clear distinction between the two populations but rather a continuous variation in thickness, metallicity and radial scale-length, with the oldest, most metal-poor and hottest component having the shortest radial scale length. Others have challenged this conclusion (e.g. Bensby 2013), arguing that the thick disk is a distinct component. Whichever way this discussion is settled, the thick and thin terminology remains useful to distinguish the two ends of the thickness range.
Burstein 1979, Mould 2005, and others found evidence for a thicker layer of older stars in other galaxies, which Yoachim & Dalcanton (2006) suggested may be more massive, relative to the thin disk, in lower mass galaxies. Furthermore, Comerón et al. (2011) suggested that the thick components may be more massive than previously believed.
The thin and thick disks of the Milky Way can be distinguished not only by their scale heights and velocity dispersions, but the thick disk lags in its net rotational velocity (Chiba & Beers 2000), contains older stars with lower metallicities (Majewski 1993), its stars have enhanced [α/Fe] ratios (Bensby et al. 2005; Reddy et al. 2006; Fuhrmann 2008; Ruchti et al. 2011; Schlesinger et al. 2012; Liu & van de Ven 2012). As noted Bensby et al. (2011) and Cheng et al. (2012) suggest a shorter radial scale length for the α-enhanced thick disk, although their estimates are still quite uncertain. These distinctions are not clear cut, and the assignment to a population may depend somewhat on whether a spatial, kinematic, or chemical abundance criterion is applied (Fuhrmann 2008; Schönrich & Binney 2009b; Loebman et al. 2011).
4.1. Formation of thickened disks
Whether the thick disk is or is not a separate component has important implications for its formation. A distinct component suggests some event, such as a minor merger (see Section 4.2) in the past, stirred up the old disk and what is now described as the thin disk began to form subsequently through gas accretion and star formation, creating two chemically and dynamically distinct populations. However, evidence of such an event could well be obscured by one or more of a number of other mechanisms that may also contribute to the currently observed properties.
In addition to the minor merger hypothesis, Abadi et al. (2003) proposed that the debris accreted from disrupted satellite galaxies could form part of the thick disk, but chemical analysis of thick-disk stars (Ruchti et al. 2010, 2011) argued against this suggestion. Brook et al. (2005) and Bournaud et al. (2009) have suggested that stars formed in a thicker gas layer during galaxy assembly could have given rise to a thick disk. A fourth suggestion is that stars migrating outward from the inner Galaxy would have a thick distribution. Both the simulations of Loebman et al. (2011) and the semianalytic model for Galactic chemical evolution that includes radial migration by Schönrich & Binney (2009a), showed that outward radial migration of old stars from the inner disk can create a thick population of old, metal-poor, stars with enhanced [α/Fe] ratios. Schönrich & Binney ( 2009a, b) and Scannapieco et al. (2011) pointed out that it naturally gives rise to both a thin and a thick disk, under the assumption that thick-disk stars experience a similar radial churning. This assumption was validated by Solway et al. (2012).
Sales et al. (2009) proposed a test, based on orbit eccentricity, to distinguish these formation mechanisms, that has not proven decisive (Dierickx et al. 2010; Casetti-Dinescu et al. 2011; Wilson et al. 2011), although it does disfavor the accretion scenario. In this context, it should be noted that the peculiar velocity components, even of stars having highly eccentric orbits, could be redirected by GMC scattering (Section 3.2.4), weakening the power of such tests. Furthermore, it is likely that more than one of these mechanisms has been at play as the disk of the Milky Way has built up.
Sridhar & Touma (1996) suggested that the thick disk was formed by "levitation", in which radially eccentric in-plane orbits were converted to near-circular inclined orbits through resonant trapping as the potential of the Galaxy became more flattened during disk growth. However, the observed orbital eccentricities in the thin and thick disks today are the other way around.
4.2. Survival of thin disks
The hierarchical model of galaxy assembly (Section 2.1) is challenged by the thinness of disk galaxies (Tóth & Ostriker 1992), which are stirred and thickened by the infall of satellite galaxies (Quinn et al. 1993; Walker et al. 1996; Velazquez & White 1999; Berentzen et al. 2003; Read et al. 2008; Villalobos & Helmi 2008; Kazantzidis et al. 2009). The severity of the challenge to the current CDM paradigm involves many questions that are not easily answered. Wyse (2009) summarizes the evidence that the thick disk of the Milky Way, and perhaps that of other galaxies (Mould 2005), contains essentially no stars younger than ~ 1010 yr. If this critical piece of evidence holds up, it implies that no gravitational disturbance to the disk could have scattered stars into the thicker layer throughout that time.
The survival of the so-called superthin galaxies (see e.g. Matthews 2000) presents a similar challenge. They are believed to be low-surface-brightness galaxies viewed edge-on that are probably embedded in a massive halo. If the low-luminosity density represents a low disk mass density, then their disks are less coherently held together by their self-gravity than are normal disks, making them all the more fragile to gravitational perturbations. Thus, not only are these disks remarkably thin, but they would be more easily thickened by perturbations than would heavier disks.
The expected rate of infall of subhalos as a function of their mass can be estimated from simulations of the growth of dark matter halos in the appropriate cosmology (e.g. Purcell et al. 2009). However the infalling pieces of substructure can be tidally disrupted, and may merge into the smooth inner halo (Gao et al. 2011). The Sagittarius stream (e.g. Belokurov et al. 2006) provides a clear example of the tidal stripping of a satellite as it falls into the Milky Way halo.
If the core of a dwarf galaxy is dense enough to survive until it interacts strongly with the disk, it may deposit some of its orbital energy into the disk, the remainder being absorbed by the halo through dynamical friction (Section 6). A proper calculation of this process needs to take into account the damping of the vertical oscillation by dynamical friction (Quinn & Goodman 1986), the reorientation of the disk plane in response to the absorption of misaligned angular momentum (Huang & Carlberg 1997), and the excitation of bending waves that can travel some distance across the disk before depositing their energy into vertical random motion (Sellwood et al. 1998). The coherence of the disk needed to support these last two mechanisms depends both on its self-gravity and on the degree of random motion (Debattista & Sellwood 1999).
The simulations by Kazantzidis et al. (2009) reveal that the disk is significantly distorted and thickened by the infall of a sequence of massive subclumps. The larger clumps which arrived first caused the most disruption, while the smaller fragments did less damage. All three velocity components of the disk particles rose substantially, while the disk also developed a pronounced flare.
Many (e.g. Moster 2010; Villalobos et al. 2010; Puech et al. 2012) have pointed out that gas infall subsequent to a minor merger can form a new thin disk, and that the attraction of the additional mass in the disk squeezes the thickened layer of older stars. However, stars formed prior to the merger remain in a thickened layer (see e.g. Brook et al. 2004; Robertson et al. 2006; Governato et al. 2009). The Milky Way may have a continuum of disk populations of increasing thickness and age (Bovy et al. 2012a); but if the conclusion of Wyse (2009) that the thickest subcomponent contains no stars with ages 1010 yr holds, then the disk could not have been gravitationally stirred for all that time. Such a constraint would present a significant challenge to current cosmological models.
4.3. Challenge to radial migration models
The old age of the thick disk also raises a challenge for radial migration models (Schönrich & Binney 2009b). If some thick-disk stars have migrated from the inner Milky Way through the action of spirals, why are they all so old? One might expect at least a tail of young stars that have migrated rapidly from the center, although there were very few in the simulation by Loebman et al. (2011). Solway and Sellwood (in preparation) suggest that the formation of the bar in the Milky Way prevented any stars born in the inner Galaxy from being caught up by the corotation resonance of spirals and carried to the outer disk. If this suggestion is correct, then we may be able to date the formation of the bar in the Milky Way from the oldest stars that have inner disk metallicities. Other processes may have added stars to the thick disk, but there is at least hope that the abundance ratios of some elements might be unique signatures that the star originated in the inner Milky Way.