Deep near-infrared imaging of W3 Main: constraints on stellar cluster formation

Context. Embedded clusters like W3 Main are complex and dynamically evolving systems that represent an important phase in the star formation process. Aims. We aim to characterize of the entire stellar content of W3 Main in a statistical sense, which will then identify possible differences in the evolutionary phase of the stellar populations and find clues about the formation mechanism of this massive embedded cluster. Methods. Deep JHK(s) imaging is used to derive the disk fraction, K-s-band luminosity functions, and mass functions for several subregions in W3 Main. A two-dimensional completeness analysis using artificial star experiments is applied as a crucial ingredient for assessing realistic completeness limits for our photometry. Results. We find an overall disk fraction of 7.7 +/- 2.3%, radially varying from 9.4 +/- 3.0% in the central 1 pc to 5.6 +/- 2.2% in the outer parts of W3 Main. The mass functions derived for three subregions are consistent with a Kroupa and Chabrier mass function. The mass function of IRSN3 is complete down to 0.14 M-circle dot and shows a break at M similar to 0.5 M-circle dot. Conclusions. We interpret the higher disk fraction in the center as evidence that the cluster center is younger. We find that the evolutionary sequence observed in the low-mass stellar population is consistent with the observed age spread among the massive stars. An analysis of the mass function variations does not show evidence of mass segregation. W3 Main is currently still actively forming stars, showing that the ionizing feedback of OB stars is confined to small areas (similar to 0.5 pc). The FUV feedback might be influencing large regions of the cluster as suggested by the low overall disk fraction.