Observations of both star-forming regions and young, gas-free stellar associations indicate that most nearby molecular clouds form stars only over a short time span before dispersal; large-scale flows in the diffuse interstellar medium have the potential for forming clouds sufficiently rapidly, and for producing stellar populations with ages much less than the lateral crossing times of their host molecular clouds. We identify four important factors for understanding rapid star formation and short cloud lifetimes. First, much of the accumulation and dispersal of clouds near the solar circle might occur in the atomic phase; only the high-density portion of a cloud's lifecycle is spent in the molecular phase, thus helping to limit molecular cloud ``lifetimes''. Second, once a cloud achieves a high enough column density to form $\h2$ and CO, gravitational forces become larger than typical interstellar pressure forces; thus star formation can follow rapidly upon molecular gas formation and turbulent dissipation in limited areas of each cloud complex. Third, typical magnetic fields are not strong enough to prevent rapid cloud formation and gravitational collapse. Fourth, rapid dispersal of gas by newly-formed stars, and reduction of shielding by a small expansion of the cloud after the first events of star formation, might limit the length of the star formation epoch and the lifetime of a cloud in its molecular state. This picture emphasizes the importance of large-scale boundary conditions for understanding molecular cloud formation, and implies that star formation is a highly dynamic, rather than quasi-static, process.Comment: 41 pages, 10 figures, accepted by Ap

We study the formation of giant dense cloud complexes and of stars within them using SPH numerical simulations of the collision of gas streams (''inflows'') in the WNM at moderately supersonic velocities. The collisions cause compression, cooling, and turbulence generation in the gas, forming a cloud that then becomes self-gravitating and begins to collapse globally. Simultaneously, the turbulent, nonlinear density fluctuations induce fast, local collapse events. The simulations show that (1) The clouds are not in a state of equilibrium. Instead, they undergo secular evolution. During its early stages, the cloud's mass and gravitational energy jE g j increase steadily, while the turbulent energy E k reaches a plateau. (2) When jE g j becomes comparable to E k , global collapse begins, causing a simultaneous increase in jE g j and E k that maintains a near-equipartition condition jE g j $ 2E k . (3) Longer inflow durations delay the onset of global and local collapse by maintaining a higher turbulent velocity dispersion in the cloud over longer times. (4) The star formation rate is large from the beginning, without any period of slow and accelerating star formation. (5) The column densities of the local star-forming clumps closely resemble reported values of the column density required for molecule formation, suggesting that locally molecular gas and star formation occur nearly simultaneously. The MC formation mechanism discussed here naturally explains the apparent ''virialized'' state of MCs and the ubiquity of H i halos around them. Also, within their assumptions, our simulations support the scenario of rapid star formation after MCs are formed, although long (k15 Myr) accumulation periods do occur during which the clouds build up their gravitational energy, and which are expected to be spent in the atomic phase.

We suggest that molecular clouds can be formed on short timescales by compressions from large scale streams in the interstellar medium (ISM). In particular, we argue that in the Taurus-Auriga complex, with Ðlaments of 10È20 ] 2È5 pc, most have been formed by H I Ñows in Myr, explaining the [3 absence of postÈT Tauri stars in the region with ages Myr. Observations in the 21 cm line of the H I Z3 "" halos ÏÏ around the Taurus molecular gas show many features (broad asymmetric proÐles, velocity shifts of H I relative to 12CO) predicted by our MHD numerical simulations, in which large-scale H I streams collide to produce dense Ðlamentary structures. This rapid evolution is possible because the H I Ñows producing and disrupting the cloud have much higher velocities (5È10 km s~1) than are present in the molecular gas resulting from the colliding Ñows. The simulations suggest that such Ñows can occur from the global ISM turbulence without requiring a single triggering event such as a supernova explosion.

Motivated by our previous paper, in which we argued for the formation of molecular clouds from large-scale flows in the diffuse galactic interstellar medium, we examine the formation of molecular gas behind shocks in atomic gas using a one-dimensional chemical/dynamical model. In our analysis we place particular emphasis on constraints placed on the dynamical evolution by the chemistry. The most important result of this study is to stress the importance of shielding the molecular gas from the destructive effects of UV radiation. For shock ram pressures comparable to or exceeding typical local interstellar medium pressures, self-shielding controls the formation time of molecular hydrogen but CO formation requires shielding of the interstellar radiation field by dust grains. We find that for typical parameters the molecular hydrogen fractional abundance can become significant well before CO forms. The timescale for (CO) molecular cloud formation is not set by the H 2 formation rate on grains, but rather by the timescale for accumulating a sufficient column density or extinction, A V 0.7.The local ratio of atomic to molecular gas (4:1), coupled with short estimates for the lifetimes of molecular clouds (3-5 Myr), suggests that the timescales for accumulating molecular clouds from atomic material typically must be no longer than about 12-20 Myr. Based on the shielding requirement, this implies that the typical product of pre-shock density and velocity must be nv 20 cm −3 kms −1 . In turn, depending upon the shock velocity, this implies shock ram pressures which are a few times the typical estimated local turbulent gas pressure, and comparable to the total pressures (gas plus magnetic plus cosmic rays). Coupled with the rapid formation of CO once shielding is sufficient, flow-driven formation of molecular clouds in the local interstellar medium can occur sufficiently rapidly to account for observations.We also provide detailed predictions of atomic and molecular emission and absorption that track the formation of a molecular cloud from a purely atomic medium, with a view toward helping to verify cloud formation by shock waves. However, our predictions suggest that the detection of the pre-CO stages will be challenging. Finally, we provide an analytic solution for time-dependent H 2 formation which may be of use in numerical hydrodynamic calculations.

We present a unified description of the scenario of global hierarchical collapse (GHC). GHC constitutes a flow regime of (non-homologous) collapses within collapses, in which all scales accrete from their parent structures, and small, dense regions begin to contract at later times, but on shorter time-scales than large, diffuse ones. The different time-scales allow for most of the clouds’ mass to be dispersed by the feedback from the first massive stars, maintaining the cloud-scale star formation rate low. Molecular clouds (MCs), clumps, and cores are not in equilibrium, but rather are either undergoing contraction or dispersal. The main features of GHC are as follows: (1) The gravitational contraction is initially very slow, and begins when the cloud still consists of mostly atomic gas. (2) Star-forming MCs are in an essentially pressureless regime, causing filamentary accretion flows from the cloud to the core scale to arise spontaneously. (3) Accreting objects have longer lifetimes than their own free-fall time, due to the continuous replenishment of material. (4) The clouds’ total mass and its molecular and dense mass fractions increase over time. (5) The clouds’ masses stop growing when feedback becomes important. (6) The first stars appear several megayears after global contraction began, and are of low mass; massive stars appear a few megayears later, in massive hubs. (7) The minimum fragment mass may well extend into the brown-dwarf regime. (8) Bondi–Hoyle–Lyttleton-like accretion occurs at both the protostellar and the core scales, accounting for an IMF with slope dN/dM ∝ M−2. (9) The extreme anisotropy of the filamentary network explains the difficulty in detecting large-scale infall signatures. (10) The balance between inertial and gravitationally driven motions in clumps evolves during the contraction, explaining the approach to apparent virial equilibrium, from supervirial states in low-column density clumps and from subvirial states in dense cores. (11) Prestellar cores adopt Bonnor–Ebert-like profiles, but are contracting ever since when they may appear to be unbound. (12) Stellar clusters develop radial age and mass segregation gradients. We also discuss the incompatibility between supersonic turbulence and the observed scalings in the molecular hierarchy. Since gravitationally formed filaments do not develop shocks at their axes, we suggest that a diagnostic for the GHC scenario should be the absence of strong shocks in them. Finally, we critically discuss some recent objections to the GHC mechanism.

We discuss the nature of the velocity dispersion versus size relation for molecular clouds. In particular, we add to previous observational results showing that the velocity dispersions in molecular clouds and cores are not purely functions of the spatial scale but involve surface gas densities as well. We emphasize that hydrodynamic turbulence is required to produce the first condensations in the progenitor medium. However, as the cloud is forming, it also becomes bound, and gravitational accelerations dominate the motions. Energy conservation in this case implies |E g | ∼ E k , in agreement with observational data, and providing an interpretation for two recent observational results: the scatter in the δv-R plane, and the dependence of the velocity dispersion on the surface density δv 2 /R ∝ . We argue that the observational data are consistent with molecular clouds in a state of hierarchical and chaotic gravitational collapse, i.e. developing local centres of collapse throughout the whole cloud while the cloud itself is collapsing, and making equilibrium unnecessary at all stages prior to the formation of actual stars. Finally, we discuss how this mechanism need not be in conflict with the observed star formation rate.

We examine the idea that diffuse and giant molecular clouds and their substructure form as density fluctuations induced by large scale interstellar turbulence. We do this by closely investigating the topology of the velocity, density and magnetic fields within and at the boundaries of the clouds emerging in high-resolution two-dimensional simulations of the ISM including self-gravity, magnetic fields, parameterized heating and cooling and a simple model for star formation. We find that the velocity field is continuous across cloud boundaries for a hierarchy of clouds of progressively smaller sizes. Cloud boundaries defined by a density-threshold criterion are found to be quite arbitrary, with no correspondence to any actual physical boundary, such as a density discontinuity. Abrupt velocity jumps are coincident with the density maxima, indicating that the clouds are formed by colliding gas streams. This conclusion is also supported by the fact that the volume and surface kinetic terms in the Eulerian Virial Theorem for a cloud ensemble are comparable in general. The topology of the magnetic field is also suggestive of the same process, exhibiting bends and reversals where the gas streams collide. However, no unique trend of alignment between density and magnetic features is observed. Both sub-and super-Alfvénic motions are observed within the clouds in the simulations.In the light of these results, we argue that thermal pressure equilibrium is irrelevant for cloud confinement in a turbulent medium, since inertial motions can still distort or disrupt a cloud, unless it is strongly gravitationally bound. Turbulent pressure confinement appears self-defeating, because turbulence contains large-scale motions which necessarily distort Lagrangian cloud boundaries, or equivalently cause flux through Eulerian boundaries.We then discuss the compatibility of the present scenario with observational data. We find that density-weighted velocity histograms are consistent with observational line profiles of comparable spatial and velocity resolution, exhibiting similar FWHMs and similar multi-component structure. An analysis of the regions contributing to each velocity interval indicates that the histogram "features" do not come from isolated "clumps", but rather from extended regions throughout a cloud, which often have very different total velocity vectors.

We discuss the lifetimes and evolution of clumps and cores formed as turbulent density fluctuations in nearly isothermal molecular clouds. In the non-magnetic case, clumps are unlikely to reach a hydrostatic state, and instead are expected to either proceed directly to collapse, or else ``rebound'' towards the mean pressure and density of the parent cloud. Rebounding clumps are delayed in their re-expansion by their self-gravity. From a simple virial calculation, we find re-expansion times of a few free-fall times. In the magnetic case, we present a series of driven-turbulence, ideal-MHD isothermal numerical simulations in which we follow the evolution of clumps and cores in relation to the magnetic criticality of their ``parent clouds'' (the numerical boxes). In subcritical boxes, magnetostatic clumps do not form. A few moderately-gravitationally bound clumps form which however are dispersed by the turbulence in < 1.3 Myr. An estimate of the ambipolar diffusion (AD) time scale t_AD in these cores gives t_AD > 1.3 Myr, only slightly longer than the dynamical times. In supercritical boxes, some cores become locally supercritical and collapse in typical times ~ 1 Myr. We also observe longer-lived supercritical cores that however do not collapse because they are smaller than the local Jeans length. Fewer clumps and cores form in these simulations than in their non-magnetic counterpart. Our results suggest that a) A fraction of the cores may not form stars, and may correspond to some of the observed starless cores. b) Cores may be out-of-equilibrium structures, rather than quasi-magnetostatic ones. c) The magnetic field may help reduce the star formation efficiency by reducing the probability of core formation, rather than by significantly delaying the collapse of individual cores.Comment: Accepted in ApJ. Originally submitted as astro-ph/0208245. Completely rewritten, now including numerical simulations. Animations available at http://www.astrosmo.unam.mx/~e.vazquez/turbulence_HP/movies/VKSB04.htm