On the universality of interstellar filaments: theory meets simulations and observations

Filaments are ubiquitous in the universe. Recent observations have revealed that stars and star clusters form preferentially along dense filaments. Understanding the formation and properties of filaments is therefore a crucial step in understanding star formation. Here we perform three-dimensional high-resolution magnetohydrodynamical simulations that follow the evolution of molecular clouds and the formation of filaments and stars. We apply a filament detection algorithm and compare simulations with different combinations of physical ingredients: gravity, turbulence, magnetic fields and jet/outflow feedback. We find that gravity-only simulations produce significantly narrower filament profiles than observed, while simulations that include turbulence produce realistic filament properties. For these turbulence simulations, we find a remarkably universal filament width of 0.10 +/- 0.02 pc, which is independent of the star formation history of the clouds. We derive a theoretical model that provides a physical explanation for this characteristic filament width, based on the sonic scale (lambda_sonic) of molecular cloud turbulence. Our derivation provides lambda_sonic as a function of the cloud diameter L, the velocity dispersion sigma_v, the gas sound speed c_s, and the ratio of thermal to magnetic pressure, plasma beta. For typical cloud conditions in the Milky Way spiral arms, we find lambda_sonic = 0.04-0.16 pc, in excellent agreement with the filament width of 0.05-0.15 pc from observations. Consistent with the theoretical model assumptions, we find that the velocity dispersion inside the filaments is subsonic and supersonic outside. We further explain the observed p=2 scaling of the filament density profile, rho ~ r^(-p) with the collision of two planar shocks forming a filament at their intersection.