2008A&A...480..335G

The detection of the rotational lines of CO in proto-galaxies in the early Universe provides one of the most promising ways of probing the fundamental physical properties of a galaxy, such as its size, dynamical mass, gas density, and temperature. While such observations are currently limited to the most luminous galaxies, the advent of ALMA will change the situation dramatically, resulting in the detection of numerous normal galaxies at high redshifts. We investigate the morphology and strength of the CO rotational line emission emerging from z∼3 progenitor systems of normal field galaxies seen in the present-day Universe, examine how well ALMA will be able to detect such systems and how accurately CO line widths, gas morphologies and ultimately dynamical masses can be inferred. Maps and spectra of rotational CO line emission were calculated from a cosmological N-body/hydrodynamical TreeSPH simulation of a z∼3 ``Lyman break galaxy'' of UV luminosity about one order of magnitude below L^*. To simulate a typical observation of our system with ALMA, we imposed characteristic noise, angular, and spectral resolution constraints. The CO line properties predicted by our simulation are in good agreement with the two Lyman break systems detected in CO to date. We find that while supernovae explosions from the ongoing star formation carve out large cavities in the molecular ISM, they do not generate large enough gas outflows to make a substantial imprint on the CO line profile. This implies that for most proto-galaxies - except possibly the most extreme cases - stellar feedback effects do not affect CO as a dynamical mass tracer. Detecting CO in sub-L^* galaxies at z≃3 will push ALMA to the limits of its cababilities, and whether a source is detected or not may depend critically on its inclination angle. Both these effects (sensitivity and inclination) will severely impair the ability of ALMA to infer the gas kinematics and dynamical masses using line observations.