The pressure exerted by massive stars' radiation fields is an important mechanism regulating their formation. Detailed simulation of massive star formation therefore requires an accurate treatment of radiation. However, all published simulations have either used a diffusion approximation of limited validity; have only been able to simulate a single star fixed in space, thereby suppressing potentially-important instabilities; or did not provide adequate resolution at locations where instabilities may develop. To remedy this we have developed a new, highly accurate radiation algorithm that properly treats the absorption of the direct radiation field from stars and the re-emission and processing by interstellar dust. We use our new tool to perform three-dimensional radiation-hydrodynamic simulations of the collapse of massive pre-stellar cores with laminar and turbulent initial conditions and properly resolve regions where we expect instabilities to grow. We find that mass is channeled to the stellar system via gravitational and Rayleigh-Taylor (RT) instabilities, in agreement with previous results using stars capable of moving, but in disagreement with methods where the star is held fixed or with simulations that do not adequately resolve the development of RT instabilities. For laminar initial conditions, proper treatment of the direct radiation field produces later onset of instability, but does not suppress it entirely provided the edges of radiation-dominated bubbles are adequately resolved. Instabilities arise immediately for turbulent pre-stellar cores because the initial turbulence seeds the instabilities. Our results suggest that RT features are significant and should be present around accreting massive stars throughout their formation.

Using numerical hydrodynamics calculations and a novel method for densely sampling parameter space, we measure the accretion and torque on a binary system from a circumbinary disk. In agreement with some earlier studies, we find that the net torque on the binary is positive for mass ratios close to unity, and that accretion always drives the binary toward equal mass. Accretion variability depends sensitively on the numerical sink prescription, but the torque and relative accretion onto each component do not depend on the sink timescale. Positive torque and highly variable accretion occurs only for mass ratios greater than around 0.05. This means that for mass ratios below 0.05, the binary would migrate inward until the secondary accreted sufficient mass, after which it would execute a U-turn and migrate outward. We explore a range of viscosities, from α = 0.03 to α = 0.15, and find that this outward torque is proportional to the viscous torque, so that torque per unit accreted mass is independent of α. Dependence of accretion and torque on mass ratio is explored in detail, densely sampling mass ratios between 0.01 and unity. For mass ratio q > 0.2, accretion variability is found to exhibit a distinct sawtooth pattern, typically with a five-orbit cycle that provides a smoking gun prediction for variable quasars observed over long periods, as a potential means to confirm the presence of a binary.

Star clusters larger than ∼ 10 3 M contain multiple hot stars that launch fast stellar winds. The integrated kinetic energy carried by these winds is comparable to that delivered by supernova explosions, suggesting that at early times winds could be an important form of feedback on the surrounding cold material from which the star cluster formed. However, the interaction of these winds with the surrounding clumpy, turbulent, cold gas is complex and poorly understood. Here we investigate this problem via an accounting exercise: we use empirically determined properties of four well-studied massive star clusters to determine where the energy injected by stellar winds ultimately ends up. We consider a range of kinetic energy loss channels, including radiative cooling, mechanical work on the cold interstellar medium, thermal conduction, heating of dust via collisions by the hot gas, and bulk advection of thermal energy by the hot gas. We show that, for at least some of the clusters, none of these channels can account for more than a small fraction of the injected energy. We suggest that turbulent mixing at the hot-cold interface or physical leakage of the hot gas from the H ii region can efficiently remove the kinetic energy injected by the massive stars in young star clusters. Even for the clusters where we are able to account for all the injected kinetic energy, we show that our accounting sets strong constraints on the importance of stellar winds as a mechanism for feedback on the cold interstellar medium.

Stellar feedback is needed to produce realistic giant molecular clouds and galaxies in simulations, but due to limited numerical resolution, feedback must be implemented using sub-grid models. Observational work is an important means to test and anchor these models, but limited studies have assessed the relative dynamical role of multiple feedback modes, particularly at the earliest stages of expansion when H II regions are still deeply embedded. In this paper, we use multiwavelength (radio, infrared, and X-ray) data to measure the pressures associated with direct radiation (P dir ), dust-processed radiation (P IR ), photoionization heating (P H II ), and shockheating from stellar winds (P X ) in a sample of 106 young, resolved H II regions with radii 0.5 pc to determine how stellar feedback drives their expansion. We find that the P IR dominates in 84% of the regions and that the median P dir and P H II are smaller than the median P IR by factors of ≈6 and ≈9, respectively. Based on the radial dependences of the pressure terms, we show that H II regions transition from P IR -dominated to P H II -dominated at radii of ∼3 pc. We find a median trapping factor of f trap ∼ 8 without any radial dependence for the sample, suggesting this value can be adopted in sub-grid feedback models. Moreover, we show that the total pressure is greater than the gravitational pressure in the majority of our sample, indicating that the feedback is sufficient to expel gas from the regions.

Stellar feedback in the form of radiation pressure and magnetically driven collimated outflows may limit the maximum mass that a star can achieve and affect the star formation efficiency of massive prestellar cores. Here we present a series of 3D adaptive mesh refinement radiation-magnetohydrodynamic simulations of the collapse of initially turbulent, massive prestellar cores. Our simulations include radiative feedback from both the direct stellar and dust-reprocessed radiation fields, and collimated outflow feedback from the accreting stars. We find that protostellar outflows punch holes in the dusty circumstellar gas along the star's polar directions, thereby increasing the size of optically thin regions through which radiation can escape. Precession of the outflows as the star's spin axis changes due to the turbulent accretion flow further broadens the outflow, and causes more material to be entrained. Additionally, the presence of magnetic fields in the entrained material leads to broader entrained outflows that escape the core. We compare the injected and entrained outflow properties and find that the entrained outflow mass is a factor of ∼3 larger than the injected mass and the momentum and energy contained in the entrained material are ∼25% and ∼5% of the injected momentum and energy, respectively. As a result, we find that, when one includes both outflows and radiation pressure, the former are a much more effective and important feedback mechanism, even for massive stars with significant radiative outputs. Unified AstronomyThesaurus concepts: Interstellar magnetic fields (846); Star formation (1569); Radiative magnetohydrodynamics (2009); Massive stars (732); Stellar jets (1607); Stellar feedback (1602)