Multiphase Interstellar Medium Research Articles

C iv and He ii line emission of Lyman α blobs: powered by shock-heated gas

Utilizing {\it ab initio} ultra-high resolution hydrodynamical simulations, we investigate the properties of the interstellar and circum-galactic medium of Ly$\alpha$ Blobs (LABs) at $z=3$, focusing on three important emission lines: Ly$\alpha$ 1216\AA, \heii 1640\AA\ and \civ 1449\AA. Their relative strengths provide a powerful probe of the thermodynamic properties of the gas when confronted with observations. By adjusting the dust attenuation effect using one parameter and matching the observed size-luminosity relation of LABs using another parameter, we show that our simulations can reproduce the observed \civ/\lya\ and \heii/\lya\ ratios adequately. This analysis provides the first successful physical model to account for simultaneously the LAB luminosity function, luminosity-size relation, and the \civ/Ly$\alpha$ and \heii/Ly$\alpha$ ratios, with only two parameters. The physical underpinning for this model is that, in addition to the stellar component for the \lya\ emission, the \lya\ and \civ\ emission lines due to shock heated gas are primarily collisional excitation driven and the \heii\ emission line collisional ionization driven. We find that the density, temperature and metallicity of the gas responsible for each emission line is significantly distinct, in a multi-phase interstellar and circumgalactic medium that is shock-heated primarily by supernovae and secondarily by gravitational accretion of gas.

The anatomy of a star-forming galaxy: pressure-driven regulation of star formation in simulated galaxies

We explore the regulation of star formation in star-forming galaxies through a suite of high-resolution isolated galaxy simulations. We use the SPH code GASOLINE, including photoelectric heating and metal cooling, which produces a multi-phase interstellar medium. We show that representative star formation and feedback sub-grid models naturally lead to a weak, sub-linear dependence between the amount of star formation and changes to star formation parameters. We incorporate these sub-grid models into an equilibrium pressure-driven regulation framework. We show that the sub-linear scaling arises as a consequence of the non-linear relationship between scale height and the effective pressure generated by stellar feedback. Thus, simulated star-formation regulation is sensitive to how well vertical structure in the ISM is resolved. Full galaxy disks experience density waves which drive locally time-dependent star formation. We develop a simple time-dependent, pressure-driven model that reproduces the response extremely well.

TRACING THE EVOLUTION OF HIGH-REDSHIFT GALAXIES USING STELLAR ABUNDANCES

ABSTRACT This paper presents the first results from a model for chemical evolution that can be applied to N-body cosmological simulations and quantitatively compared to measured stellar abundances from large astronomical surveys. This model convolves the chemical yield sets from a range of stellar nucleosynthesis calculations (including asymptotic giant branch stars, Type Ia and II supernovae, and stellar wind models) with a user-specified stellar initial mass function (IMF) and metallicity to calculate the time-dependent chemical evolution model for a “simple stellar population” (SSP) of uniform metallicity and formation time. These SSP models are combined with a semianalytic model for galaxy formation and evolution that uses merger trees from N-body cosmological simulations to track several α- and iron-peak elements for the stellar and multiphase interstellar medium components of several thousand galaxies in the early (z ≥ 6) universe. The simulated galaxy population is then quantitatively compared to two complementary data sets of abundances in the Milky Way stellar halo and is capable of reproducing many of the observed abundance trends. The observed abundance ratio distributions are best reproduced with a Chabrier IMF, a chemically enriched star formation efficiency of 0.2, and a redshift of reionization of 7. Many abundances are qualitatively well matched by our model, but our model consistently overpredicts the carbon-enhanced fraction of stars at low metallicities, likely owing to incomplete coverage of Population III stellar yields and supernova models and the lack of dust as a component of our model.

Stellar and quasar feedback in concert: effects on AGN accretion, obscuration, and outflows

We study the interaction of feedback from active galactic nuclei (AGN) and a multi-phase interstellar medium (ISM), in simulations including explicit stellar feedback, multi-phase cooling, accretion-disk winds, and Compton heating. We examine radii ~0.1-100 pc around a black hole (BH), where the accretion rate onto the BH is determined and where AGN-powered winds and radiation couple to the ISM. We conclude: (1) The BH accretion rate is determined by exchange of angular momentum between gas and stars in gravitational instabilities. This produces accretion rates ~0.03-1 Msun/yr, sufficient to power luminous AGN. (2) The gas disk in the galactic nucleus undergoes an initial burst of star formation followed by several Myrs where stellar feedback suppresses the star formation rate (SFR). (3) AGN winds injected at small radii with momentum fluxes ~L/c couple efficiently to the ISM and have dramatic effects on ISM properties within ~100 pc. AGN winds suppress the nuclear SFR by factors ~10-30 and BH accretion rate by factors ~3-30. They increase the outflow rate from the nucleus by factors ~10, consistent with observational evidence for galaxy-scale AGN-driven outflows. (4) With AGN feedback, the predicted column density distribution to the BH is consistent with observations. Absent AGN feedback, the BH is isotropically obscured and there are not enough optically-thin sightlines to explain Type-I AGN. A 'torus-like' geometry arises self-consistently as AGN feedback evacuates gas in polar regions.

The Relation between Star-Formation Rate and Stellar Mass of Galaxies atz~ 1–4

AbstractThe relation between the star-formation Rate and stellar mass (M⋆) of galaxies represents a fundamental constraint on galaxy formation, and has been studied extensively both in observations and cosmological hydrodynamic simulations. However, the observed amplitude of the star-formation rate—stellar mass relation has not been successfully reproduced in simulations, indicating either that the halo accretion history and baryonic physics are poorly understood/modelled or that observations contain biases. In this paper, we examine the evolution of the SFR −M⋆relation ofz~ 1–4 galaxies and display the inconsistency between observed relations that are obtained using different techniques. We employ cosmological hydrodynamic simulations from various groups which are tuned to reproduce a range of observables and compare these with a range of observed SFR −M⋆relations. We find that numerical results are consistent with observations that use Spectral Energy Distribution techniques to estimate star-formation rates, dust corrections, and stellar masses. On the contrary, simulations are not able to reproduce results that were obtained by combining only UV and IR luminosities (UV+IR). These imply star-formation rates at a fixed stellar mass that are larger almost by a factor of 5 than those of Spectral Energy Distribution measurements forz~ 1.5–4. Forz< 1.5, the results from simulations, Spectral Energy Distribution fitting techniques and IR+UV conversion agree well. We find that surveys that preferably select star-forming galaxies (e.g. by adopting Lyman-break or blue selection) typically predict a larger median/average star-formation rate at a fixed stellar mass especially for high mass objects, with respect to mass selected samples and hydrodynamic simulations. Furthermore, we find remarkable agreement between the numerical results from various authors who have employed different cosmological codes and run simulations with different resolutions. This is interesting for two reasons. (A) simulations can produce realistic populations of galaxies within representative cosmological volumes even at relatively modest resolutions. (B) It is likely that current numerical codes that rely on similar subgrid multiphase interstellar medium models and are tuned to reproduce statistical properties of galaxies, produce similar results for the SFR −M⋆relation by construction, regardless of resolution, box size and, to some extent, the adopted feedback prescriptions.

On the nature of star-forming filaments – II. Subfilaments and velocities

We show that hydrodynamic turbulent cloud simulations naturally produce large filaments made up of a network of smaller and coherent sub-filaments. Such simulations resemble observations of filaments and fibres in nearby molecular clouds. The sub-filaments are dynamical features formed at the stagnation points of the turbulent velocity field where shocks dissipate the turbulent energy. They are a ubiquitous feature of the simulated clouds, which appear from the beginning of the simulation and are not formed by gradual fragmentation of larger filaments. Most of the sub-filaments are gravitationally sub-critical and do not fragment into cores, however, there is also a significant fraction of supercritical sub-filaments which break up into star-forming cores. The sub-filaments are coherent along their length, and the residual velocities along their spine show that they are subsonically contracting without any ordered rotation on scales of ~0.1 pc. Accretion flows along the sub-filaments can feed material into star forming cores embedded within the network. The overall mass in sub-filaments and the number of sub-filaments increases as the cloud evolves. We propose that the formation of filaments and sub-filaments is a natural consequence of the turbulent cascade in the complex multi-phase interstellar medium. Sub-filaments are formed by the high wavenumber, small scale modes in the turbulent velocity field. These are then stretched by local shear motions and gathered together by a combination of low wavenumber modes and gravitational contraction on larger scales, and by doing so build up the extended filaments observed in column density maps.

IMPACT OF SUPERNOVA AND COSMIC-RAY DRIVING ON THE SURFACE BRIGHTNESS OF THE GALACTIC HALO IN SOFT X-RAYS

The halo of the Milky Way contains a hot plasma with a surface brightness in soft X-rays of the order $10^{-12}$erg cm$^{-2}$ s$^{-1}$ deg$^{-2}$. The origin of this gas is unclear, but so far numerical models of galactic star formation have failed to reproduce such a large surface brightness by several orders of magnitude. In this paper, we analyze simulations of the turbulent, magnetized, multi-phase interstellar medium including thermal feedback by supernova explosions as well as cosmic-ray feedback. We include a time-dependent chemical network, self-shielding by gas and dust, and self-gravity. Pure thermal feedback alone is sufficient to produce the observed surface brightness, although it is very sensitive to the supernova rate. Cosmic rays suppress this sensitivity and reduce the surface brightness because they drive cooler outflows. Self-gravity has by far the largest effect because it accumulates the diffuse gas in the disk in dense clumps and filaments, so that supernovae exploding in voids can eject a large amount of hot gas into the halo. This can boost the surface brightness by several orders of magnitude. Although our simulations do not reach a steady state, all simulations produce surface brightness values of the same order of magnitude as the observations, with the exact value depending sensitively on the simulation parameters. We conclude that star formation feedback alone is sufficient to explain the origin of the hot halo gas, but measurements of the surface brightness alone do not provide useful diagnostics for the study of galactic star formation.

The SILCC (SImulating the LifeCycle of molecular Clouds) project – I. Chemical evolution of the supernova-driven ISM

The SILCC project (SImulating the Life-Cycle of molecular Clouds) aims at a more self-consistent understanding of the interstellar medium (ISM) on small scales and its link to galaxy evolution. We simulate the evolution of the multi-phase ISM in a 500 pc x 500 pc x 10 kpc region of a galactic disc, with a gas surface density of $\Sigma_{_{\rm GAS}} = 10 \;{\rm M}_\odot/{\rm pc}^2$. The Flash 4.1 simulations include an external potential, self-gravity, magnetic fields, heating and radiative cooling, time-dependent chemistry of H$_2$ and CO considering (self-) shielding, and supernova (SN) feedback. We explore SN explosions at different (fixed) rates in high-density regions (peak), in random locations (random), in a combination of both (mixed), or clustered in space and time (clustered). Only random or clustered models with self-gravity (which evolve similarly) are in agreement with observations. Molecular hydrogen forms in dense filaments and clumps and contributes 20% - 40% to the total mass, whereas most of the mass (55% - 75%) is in atomic hydrogen. The ionised gas contributes <10%. For high SN rates (0.5 dex above Kennicutt-Schmidt) as well as for peak and mixed driving the formation of H$_2$ is strongly suppressed. Also without self-gravity the H$_2$ fraction is significantly lower ($\sim$ 5%). Most of the volume is filled with hot gas ($\sim$90% within $\pm$2 kpc). Only for random or clustered driving, a vertically expanding warm component of atomic hydrogen indicates a fountain flow. Magnetic fields have little impact on the final disc structure. However, they affect dense gas ($n\gtrsim 10\;{\rm cm}^{-3}$) and delay H$_2$ formation. We highlight that individual chemical species, in particular atomic hydrogen, populate different ISM phases and cannot be accurately accounted for by simple temperature-/density-based phase cut-offs.

Star formation rates, laws, thresholds

AbstractOn scales of ≳ several hundred parsec, there are relatively well defined observed relations between the star formation rate surface density of a galaxy and its gas content. I present a short summary of the so-called ‘star formation laws’ and discuss the most established models to explain their origin. Nevertheless, none of the models can explain the offset of these relations, which requires a low star formation efficiency within the total and/or molecular gas. Stellar feedback is proposed to limit the star formation efficiency locally. I discuss the available feedback mechanisms and show recent simulation results on the evolution of the multi-phase interstellar medium under the influence of gravity, cooling, feedback-driven outflows and heating.

Modelling the supernova-driven ISM in different environments

We use hydrodynamical simulations in a $(256\;{\rm pc})^3$ periodic box to model the impact of supernova (SN) explosions on the multi-phase interstellar medium (ISM) for initial densities $n = 0.5-30$ cm$^{-3}$ and SN rates $1-720$ Myr$^{-1}$. We include radiative cooling, diffuse heating, and the formation of molecular gas using a chemical network. The SNe explode either at random positions, at density peaks, or both. We further present a model combining thermal energy for resolved and momentum input for unresolved SNe. Random driving at high SN rates results in hot gas ($T\gtrsim 10^6$ K) filling $> 90$% of the volume. This gas reaches high pressures ($10^4 < P/k_\mathrm{B} < 10^7$ K cm$^{-3}$) due to the combination of SN explosions in the hot, low density medium and confinement in the periodic box. These pressures move the gas from a two-phase equilibrium to the single-phase, cold branch of the cooling curve. The molecular hydrogen dominates the mass ($>50$%), residing in small, dense clumps. Such a model might resemble the dense ISM in high-redshift galaxies. Peak driving results in huge radiative losses, producing a filamentary ISM with virtually no hot gas, and a small molecular hydrogen mass fraction ($\ll 1$%). Varying the ratio of peak to random SNe yields ISM properties in between the two extremes, with a sharp transition for equal contributions. The velocity dispersion in HI remains $\lesssim 10$ km s$^{-1}$ in all cases. For peak driving the velocity dispersion in H$_\alpha$ can be as high as $70$ km s$^{-1}$ due to the contribution from young, embedded SN remnants.

THERMAL AND RADIATIVE ACTIVE GALACTIC NUCLEUS FEEDBACK HAVE A LIMITED IMPACT ON STAR FORMATION IN HIGH-REDSHIFT GALAXIES

The effects of Active Galactic Nuclei (AGNs) on their host-galaxies depend on the coupling between the injected energy and the interstellar medium (ISM). Here, we model and quantify the impact of long-range AGN ionizing radiation -- in addition to the often considered small-scale energy deposition -- on the physical state of the multi-phase ISM of the host-galaxy, and on its total Star Formation Rate (SFR). We formulate an AGN Spectral Energy Distribution matched with observations, which we use with the radiative transfer (RT) code Cloudy to compute AGN ionization in a simulated high-redshift disk galaxy. We use a high-resolution ($\sim6$ pc) simulation including standard thermal AGN feedback and calculate RT in post-processing. Surprisingly, while these models produce significant AGN-driven outflows, we find that AGN ionizing radiation and heating reduce the SFR by a few percent at most for a quasar luminosity ($L_{bol}=10^{46.5}$ erg s$^{-1}$). Although the circum-galactic gaseous halo can be kept almost entirely ionized by the AGN, most star-forming clouds ($n\gtrsim10^{2-3}$ cm$^{-3}$) and even the reservoirs of cool atomic gas ($n\sim0.3-10$ cm$^{-3}$) -- which are the sites of future star formation (100 - 200 Myrs), are generally too dense to be significantly affected. Our analysis ignores any absorption from a putative torus, making our results upper limits on the effects of ionizing radiation. Therefore, while the AGN-driven outflows can remove substantial amounts of gas in the long term, the impact of AGN feedback on the star formation efficiency in the interstellar gas in high-redshift galaxies is marginal, even when long-range radiative effects are accounted for.

Stellar feedback in molecular clouds and galactic discs

Feedback from massive stars in the form of stellar winds, ionizing radiation, and Supernova explosions shapes the structure and dynamics of the multi-phase interstellar medium. Here, I will briefly summarise some of our recent studies and findings on these subjects. I will discuss the effects of ionizing radiation and stellar winds in individual molecular clouds. Furthermore, I will shortly introduce the SILCC project, in which we model the evolution of the Supernova- and wind-driven, multi-phase ISM in stratified galactic discs.

Inclination dependence of Lyman-αproperties in a turbulent disk galaxy

We present simulations of Lyman-Alpha radiation transfer in an isolated disk galaxy with a turbulence sub-grid model, multi-phase interstellar medium and detailed star formation modelling. We investigate the influence of inclination on the observed Lyman-Alpha properties for different snapshots. The Lyman-alpha spectrum, equivalent width distribution and escape fractions vary significantly with the detailed morphology of the disk, leading to variations from one snapshot to another. In particular, we find that supernova-driven cavities near star-forming regions in the simulation can dominate the transmitted Lyman-alpha fraction, suggesting a variability of LAEs on the timescales of the star formation activity.

INVESTIGATING NEARBY STAR-FORMING GALAXIES IN THE ULTRAVIOLET WITHHST/COS SPECTROSCOPY. I. SPECTRAL ANALYSIS AND INTERSTELLAR ABUNDANCE DETERMINATIONS

This is the first in a series of three papers describing a project with the Cosmic Origins Spectrograph on the Hubble Space Telescope to measure abundances of the neutral interstellar medium (ISM) in a sample of 9 nearby star-forming galaxies. The goal is to assess the (in)homogeneities of the multiphase ISM in galaxies where the bulk of metals can be hidden in the neutral phase, yet the metallicity is inferred from the ionized gas in the HII regions. The sample, spanning a wide range in physical properties, is to date the best suited to investigate the metallicity behavior of the neutral gas at redshift z=0. ISM absorption lines were detected against the far-ultraviolet spectra of the brightest star-forming region(s) within each galaxy. Here we report on the observations, data reduction, and analysis of these spectra. Column densities were measured by a multi-component line-profile fitting technique, and neutral-gas abundances were obtained for a wide range of elements. Several caveats were considered including line saturation, ionization corrections, and dust depletion. Ionization effects were quantified with `ad-hoc' CLOUDY models reproducing the complex photoionization structure of the ionized and neutral gas surrounding the UV-bright sources. An `average spectrum of a redshift z=0 star-forming galaxy' was obtained from the average column densities of unsaturated profiles of neutral-gas species. This template can be used as a powerful tool for studies of the neutral ISM at both low and high redshift.

Large-eddy simulations of isolated disc galaxies with thermal and turbulent feedback

We present a subgrid-scale model for the Multi-phase Interstellar medium, Star formation, and Turbulence (MIST) and explore its behaviour in high-resolution large-eddy simulations of isolated disc galaxies. MIST follows the evolution of a clumpy cold and a diffuse warm component of the gas within a volume element which exchange mass and energy via various cooling, heating and mixing processes. The star formation rate is dynamically computed from the state of the gas in the cold phase. An important feature of MIST is the treatment of unresolved turbulence in the two phases and its interaction with star formation and feedback by supernovae. This makes MIST a particularly suitable model for the interstellar medium in galaxy simulations. We carried out a suite of simulations varying fundamental parameters of our feedback implementation. Several observational properties of galactic star formation are reproduced in our simulations, such as an average star formation efficiency ~1%, a typical velocity dispersion around ~10 km/s in star-forming regions, and an almost linear relationship between the column densities of star formation and dense molecular gas.

Black hole feedback in a multiphase interstellar medium

Ultrafast outflows (UFOs) from supermassive black holes (SMBHs) are thought to regulate the growth of SMBHs and host galaxies, resulting in a number of observational correlations. We present high-resolution numerical simulations of the impact of a thermalized UFO on the ambient gas in the inner part of the host galaxy. Our results depend strongly on whether the gas is homogeneous or clumpy. In the former case all of the ambient gas is driven outward rapidly as expected based on commonly used energy budget arguments, while in the latter the flows of mass and energy decouple. Carrying most of the energy, the shocked UFO escapes from the bulge via paths of least resistance, taking with it only the low-density phase of the host. Most of the mass is however in the high-density phase, and is affected by the UFO much less strongly, and may even continue to flow inwards. We suggest that the UFO energy leakage through the pores in the multiphase interstellar medium (ISM) may explain why observed SMBHs are so massive despite their overwhelmingly large energy production rates. The multiphase ISM effects reported here are probably under-resolved in cosmological simulations but may be included in prescriptions for active galactic nuclei feedback in future simulations and in semi-analytical models.

Non-equilibrium chemistry and cooling in the diffuse interstellar medium - I. Optically thin regime

An accurate treatment of the multiphase interstellar medium (ISM) in hydrodynamic galaxy simulations requires that we follow not only the thermal evolution of the gas, but also the evolution of its chemical state, including its molecular chemistry, without assuming chemical (including ionisation) equilibrium. We present a reaction network that can be used to solve for this thermo-chemical evolution. Our model follows the evolution of all ionisation states of the 11 elements that dominate the cooling rate, along with important molecules such as H2 and CO, and the intermediate molecular species that are involved in their formation (20 molecules in total). We include chemical reactions on dust grains, thermal processes involving dust, cosmic ray ionisation and heating and photochemical reactions. We focus on conditions typical for the diffuse ISM, with densities of 10^-2 cm^-3 < nH < 10^4 cm^-3 and temperatures of 10^2 K < T < 10^4 K, and we consider a range of radiation fields, including no UV radiation. In this paper we consider only gas that is optically thin, while paper II considers gas that becomes shielded from the radiation field. We verify the accuracy of our model by comparing chemical abundances and cooling functions in chemical equilibrium with the photoionisation code Cloudy. We identify the major coolants in diffuse interstellar gas to be CII, SiII and FeII, along with OI and H2 at densities nH > 10^2 cm^-3. Finally, we investigate the impact of non-equilibrium chemistry on the cooling functions of isochorically or isobarically cooling gas. We find that, at T < 10^4 K, recombination lags increase the electron abundance above its equilibrium value at a given temperature, which can enhance the cooling rate by up to two orders of magnitude. The cooling gas also shows lower H2 abundances than in equilibrium, by up to an order of magnitude.

THE INTERSTELLAR MEDIUM AND STAR FORMATION IN LOCAL GALAXIES: VARIATIONS OF THE STAR FORMATION LAW IN SIMULATIONS

We use the Adaptive Mesh Refinement code Enzo to model the interstellar medium in isolated local disk galaxies. The simulation includes a treatment for star formation and stellar feedback. We get a highly supersonic turbulent disk, which is fragmented at multiple scales and characterized by a multi-phase interstellar medium. We show that a Kennicutt-Schmidt (KS) relation only holds when averaging over large scales. However, values of star formation rates and gas surface densities lie close in the plot for any averaging size. This suggests an intrinsic relation between stars and gas at cell-size scales, which dominates over the global dynamical evolution. To investigate this effect, we develop a method to simulate the creation of stars based on the density field from the snapshots, without running the code again. We also investigate how the star formation law is affected by the characteristic star formation timescale, the density threshold and the efficiency considered in the recipe. We find that the slope of the law might vary from ~1.4 for a free-fall timescale, to ~1.0 for a constant depletion timescale. We further demonstrate that a power-law is recovered just by assuming that the mass of the new stars is a fraction of the mass of the cell $m_\star=\epsilon\rho_{\rm gas}\Delta x^3$, with no other physical criteria required. We show that both efficiency and density threshold do not affect the slope, but the right combination of them can adjust the normalization of the relation, which in turn could explain a possible bi-modality in the law.

COSMIC RAY SAMPLING OF A CLUMPY INTERSTELLAR MEDIUM

How cosmic rays sample the multi-phase interstellar medium (ISM) in starburst galaxies has important implications for many science goals, including evaluating the cosmic ray calorimeter model for these systems, predicting their neutrino fluxes, and modeling their winds. Here, we use Monte Carlo simulations to study cosmic ray sampling of a simple, two-phase ISM under conditions similar to those of the prototypical starburst galaxy M82. The assumption that cosmic rays sample the mean density of the ISM in the starburst region is assessed over a multi-dimensional parameter space where we vary the number of molecular clouds, the galactic wind speed, the extent to which the magnetic field is tangled, and the cosmic ray injection mechanism. We evaluate the ratio of the emissivity from pion production in molecular clouds to the emissivity that would be observed if the cosmic rays sampled the mean density, and seek areas of parameter space where this ratio differs significantly from unity. The assumption that cosmic rays sample the mean density holds over much of parameter space; however, this assumption begins to break down for high cloud density, injection close to the clouds, and a very tangled magnetic field. We conclude by evaluating the extent to which our simulated starburst region behaves as a proton calorimeter and constructing the time-dependent spectrum of a burst of cosmic rays.

A dynamical model of supernova feedback: gas outflows from the interstellar medium

We present a dynamical model of supernova feedback which follows the evolution of pressurised bubbles driven by supernovae in a multi-phase interstellar medium (ISM). The bubbles are followed until the point of break-out into the halo, starting from an initial adiabatic phase to a radiative phase. We show that a key property which sets the fate of bubbles in the ISM is the gas surface density, through the work done by the expansion of bubbles and its role in setting the gas scaleheight. The multi-phase description of the ISM is essential, and neglecting it leads to order of magnitude differences in the predicted outflow rates. We compare our predicted mass loading and outflow velocities to observations of local and high-redshift galaxies and find good agreement over a wide range of stellar masses and velocities. With the aim of analysing the dependence of the mass loading of the outflow, beta (i.e. the ratio between the outflow and star formation rates), on galaxy properties, we embed our model in the galaxy formation simulation, GALFORM, set in the LCDM framework. We find that a dependence of beta solely on the circular velocity, as is widely assumed in the literature, is actually a poor description of the outflow rate, as large variations with redshift and galaxy properties are obtained. Moreover, we find that below a circular velocity of 80km/s the mass loading saturates. A more fundamental relation is that between beta and the gas scaleheight of the disk, hg, and the gas fraction, fgas, as beta hg^(1.1) fgas^(0.4), or the gas surface density, \Sigma_g, and the gas fraction, as beta \Sigma_g^(-0.6) fgas^(0.8). We find that using the new mass loading model leads to a shallower faint-end slope in the predicted optical and near-IR galaxy luminosity functions.