Different dust and gas radial extents in protoplanetary disks: consistent models of grain growth and CO emission

Context. ALMA observations of protoplanetary disks confirm earlier indications that there is a clear difference between the dust and gas radial extents. The origin of this difference is still debated, with both radial drift of the dust and optical depth effects suggested in the literature. Aims. In thermo-chemical models, the dust properties are usually prescribed by simple parametrisations. In this work, the feedback of more realistic dust particle distributions onto the gas chemistry and molecular emissivity is investigated, with a particular focus on CO isotopologues. Methods. The radial dust grain size distribution is determined using dust evolution models that include growth, fragmentation, and radial drift for a given static gas density structure. The vertical settling of dust particles is computed in steady-state. A new version of the code DALI is used to take into account how dust surface area and density influence the disk thermal structure, molecular abundances, and excitation. Synthetic images of both continuum thermal emission and low J CO isotopologues lines are produced. Results. The difference of dust and gas radial sizes is largely due to differences in the optical depth of CO lines and millimeter continuum, without the need to invoke radial drift. The effect of radial drift is primarily visible in the sharp outer edge of the continuum intensity profile. The gas outer radius probed by (CO)-C-12 emission can easily differ by a factor of similar to two between the models for a turbulent alpha ranging between 10(-4) and 10(-2), with the ratio of the CO and mm radius R-CO(out)/R-mm(out) increasing with turbulence. Grain growth and settling concur in thermally decoupling the gas and dust components, due to the low collision rate with large grains. As a result, the gas can be much colder than the dust at intermediate heights, reducing the CO excitation and emission, especially for low turbulence values. Also, due to disk mid-plane shadowing, a second CO thermal desorption (rather than photodesorption) front can occur in the warmer outer mid-plane disk. The models are compared to ALMA observations of HD 163296 as a test case. In order to reproduce the observed CO snowline of the system, a binding energy for CO typical of ice mixtures, with E-b >= 1100 K, needs to be used rather than the lower pure CO value. Conclusions. The difference between observed gas and dust extent is largely due to optical depth effects, but radial drift and grain size evolution also affect the gas and dust emission in subtle ways. In order to properly infer fundamental quantities of the gaseous component of disks, such as disk outer radii and gas surface density profiles, simultaneous modelling of both dust and gas observations including dust evolution is needed.