Annu. Rev. Astron. Astrophys. 2005. 43: 861-918 Copyright © 2005 by Annual Reviews. All rights reserved

6. KINEMATICS

As described in the previous sections, the introduction of echelle spectrometers on 8- to 10-m-class telescopes has led to precise column-density measurements of weak transitions in the damped Ly systems. It has also led to another significant advance in damped Ly research, namely the resolution of kinematic characteristics from the velocity profiles of the metal lines. In Section 3.1 we showed that damped Ly profiles exhibit multiple components that are qualitatively similar to the component structure of similar transitions observed in the Galaxy ISM (e.g., Wolfe et al. 1994). At high signal-to-noise ratios the line profiles generally decompose into 5 to 30 velocity components, i.e., "clouds," with column densities spanning roughly an order of magnitude. The velocity components comprising the low-ion profiles (e.g., Si II 1808.0) are typically broadened by turbulent motions and have velocity dispersions of _v 4-7 km s^-1. These characteristics are also observed for the high-ion profiles (e.g., C IV 1548.1; Wolfe & Prochaska 2000a), although the velocity dispersions are generally larger.

Prochaska & Wolfe (1997) presented the first modest sample of measurements on the low-ion kinematic characteristics of the damped Ly systems. Their results and subsequent surveys have shown that the damped Ly systems exhibit velocity widths v ranging from 15 km s^-1 to several hundred km s^-1 with a median of 90 km s^-1 (Figure 10). Prochaska & Wolfe (1997, 1998) demonstrated that the observed v distribution matched that predicted for rotating disks with typical velocity speed v_c ~ 200 km s^-1 under the important assumption that the gas disk is thick (h 0.1R_d), where h is the disk scale height and R_d is the radial exponential scale length. The authors also stressed that the damped Ly line-profiles tend to show the "edge-leading asymmetry" expected for rotating disks. The observations, therefore, suggested a population of disk galaxies with rotation speeds corresponding to present-day galaxies.

Figure 10. Histogram of the low-ion velocity widths for the current sample of damped Ly systems with HIRES, UVES, or ESI observations. The median v value is 90 km s-1 and the distribution shows a significant tail to beyond 200 km s-1.

The difficulty with this scenario, however, is that hierarchical cosmology implies that galaxies in the young Universe have smaller masses and lower rotation speeds (on average) than the current population. Indeed, Prochaska & Wolfe (1997) emphasized that the circular velocity distribution of damped Ly systems predicted within the CDM cosmogony (Kauffmann 1996) is incompatible with the observations. This includes models (e.g., Mo, Mao & White 1998) that allow for a cross-section weighted distribution of spin parameters (Wolfe & Prochaska 2000a). As emphasized by Jedamzik & Prochaska (1998) and Prochaska & Wolfe (2001), there is a tension within CDM theory between including enough low mass galactic halos to match the observed incidence of damped Ly systems without severely underpredicting the v distribution. Granted the many successes of CDM theory, perhaps it will remain a true coincidence that the damped Ly kinematic characteristics are best described by a population of galaxies similar to (albeit thicker than) the present-day disk population.

If we are to adopt the CDM power spectrum at z ~ 3, then the velocity fields of a significant fraction of damped Ly systems must have contributions from nonrotational dynamics in order to explain the kinematic data. Numerical simulations within the CDM context describe the damped Ly systems as multiple "proto-galactic clumps" bound to a virialized dark matter halo (Haehnelt, Steinmetz & Rauch 1998). The kinematics within this scenario are due to the combination of infall, random motions, and rotational dynamics. In order for this model to be consistent with the observations, however, one requires that the damped Ly cross-section A is proportional to v_c with 2 (Haehnelt, Steinmetz & Rauch 1998; Maller et al. 2001). As discussed in Section 2.5.2, early numerical work indicated A v_c^1.1, which implied a crisis between theory and observation (Prochaska & Wolfe 2001). More recent results, however, support A v_c^2.5 (Nagamine, Springel & Hernquist 2004a) and the protogalactic clump scenario remains a viable option. ⁵ We stress, however, that no cosmological simulation to date has self-consistently matched the low-ion damped Ly kinematic observations.

Lu et al. (1996) first remarked that the high-ion profiles (e.g., C IV 1548.1) of the damped Ly systems have significantly different kinematic characteristics from the low-ion gas. Wolfe & Prochaska (2000a) examined a large sample of high-ion damped Ly profiles and stressed that while the component structure is generally disjoint from the low-ion profiles, the high ions are roughly centered on the low-ion transitions. Furthermore, there is a connection between the velocity fields of the two ionization states in that the high-ion gas nearly always shows comparable or larger velocity width than the low-ion gas (Figure 11). These trends place important constraints on the nature of the damped Ly systems. Wolfe & Prochaska (2000b) considered a simple model where the low-ion gas is confined to a disk enshrouded in a halo of high-ion gas with kinematics described by the infall model of Mo & Miralda-Escudé (1996). Wolfe & Prochaska (2000b) demonstrated that this model could not simultaneously match the low- and high-ion kinematics, in particular the co-alignment of the profile centers. In contrast, Maller et al. (2003) demonstrated that a damped Ly model based on multiple satellites bound to a single dark matter halo can satisfy the low-ion and high-ion kinematics provided each satellite has a halo of C IV producing gas. Their model only marginally reproduced the correlation between low-ion and high-ion gas and the authors suggested that a kinematic correlation exists beyond their simple model. Perhaps numerical simulations of the hot and cold phases will show that the hot gas co-rotates with the cold gas.

Figure 11. Comparison of the C IV velocity width with the low-ion velocity width for the current sample of damped Ly systems with HIRES or ESI observations. With only one or two exceptions, vCIV ~ vLow.

Because damped Ly systems are selected solely on the basis of the H I column density, it is possible that other mechanisms contribute to their velocity fields. This could include winds induced by mergers or star formation (Nulsen, Barcons & Fabian 1998; Schaye 2001) or even Hubble flow motions in the ambient IGM. At present such scenarios have not been quantitatively developed or tested against observation.

One can gain further insight into the nature of damped Ly systems by synthesizing the results on kinematic characteristics with their other properties. Wolfe & Prochaska (1998) first noted relationships between the velocity width and the H I column density and metallicity of the damped Ly systems. They found that the damped Ly systems with larger velocity width tend to have higher metallicity. This trend matches one's physical intuition: If the velocity width is correlated with the galactic mass, a correlation between v and [M/H] is natural provided more massive galaxies have higher metallicity. In contrast to the metallicity- v distribution, the damped Ly systems show smaller N(H I) at large v values. The observations run contrary to expectation for disk galaxies where the gradient of rotation projected along the line of sight peaks toward the center. The protogalactic clump scenario also would predict a mild, positive correlation (e.g., Maller et al. 2001) or no correlation depending on whether H I surface density correlates with halo mass. Therefore, the damped Ly observations are unexpected and difficult to interpet in terms of single or multiple rotating disks. Additional studies, both observational and theoretical, on correlations between the kinematics and other damped Ly properties would place further constraints on the processes of galaxy formation in the young Universe.

It is important to consider the implications of the damped Ly kinematic characteristics with respect to the observed relative abundance patterns (e.g., Prochaska 2003). As noted in Section 3.2.2, several authors have recently argued that dust-corrected / Fe ratios of damped Ly systems are roughly solar and indicate star-formation histories representative of dwarf or irregular galaxies instead of massive systems (e.g., Calura, Matteucci & Vladilo 2003; Tolstoy et al. 2003). We note, however, that the velocity widths of the majority of damped Ly systems exceed 60 km s^-1 and cannot be attributed to the gravitational velocity field of a single dwarf galaxy. Furthermore, there is no apparent correlation between the gas-phase Si/Fe ratio and velocity width. Therefore, a contradiction exists between the observed velocity widths and the interpretation of the damped Ly abundance patterns in terms of absorption by a single dwarf galaxy along the line of sight. Regarding the CDM clump scenarios, in which multiple dwarf galaxies are found along the line of sight, one notes that the majority of damped Ly systems are predicted to be embedded in dark matter halos that exceed the masses of present-day dwarf galaxies (e.g., Maller et al. 2001; Nagamine, Springel & Hernquist 2004a) and instead correspond to the progenitors of galaxies like the Milky Way (Steinmetz 2003). It would be particularly valuable to perform a simulation that traced the star-formation history and resolved the galaxy kinematics within a cosmological context to examine abundance pattern correlations with gas kinematics.

⁵ We note that in the clump model the `edge-leading asymmetry' can be reproduced if 3 or fewer clumps are intercepted by the majority of damped Ly sightlines. Back.