CLUMPY GALAXIES IN CANDELS. I. THE DEFINITION OF UV CLUMPS AND THE FRACTION OF CLUMPY GALAXIES AT 0.5 < z < 3

Although giant clumps of stars are thought to be crucial to galaxy formation and evolution, the most basic demographics of clumps are still uncertain, mainly because the definition of clumps has not been thoroughly discussed. In this paper, we carry out a study of the basic demographics of clumps in star-forming galaxies at 0.5 < z < 3, using our proposed physical definition that UV-bright clumps are discrete star-forming regions that individually contribute more than 8% of the rest-frame UV light of their galaxies. Clumps defined this way are significantly brighter than the H II regions of nearby large spiral galaxies, either individually or blended, when physical spatial resolution and cosmological dimming are considered. Under this definition, we measure the fraction of star-forming galaxies that have at least one off-center clump (f(clumpy)) and the contributions of clumps to the rest-frame UV light and star formation rate (SFR) of star-forming galaxies in the CANDELS/GOODS-S and UDS fields, where our mass-complete sample consists of 3239 galaxies with axial ratio q > 0.5. The redshift evolution of f(clumpy) changes with the stellar mass (M-*) of the galaxies. Low-mass (log(M-*/M-circle dot) < 9.8) galaxies keep an almost constant f(clumpy) of similar to 60% from z similar to 3 to z similar to 0.5. Intermediate-mass and massive galaxies drop their f(clumpy) from 55% at z similar to 3 to 40% and 15%, respectively, at z similar to 0.5. We find that (1) the trend of disk stabilization predicted by violent disk instability matches the f(clumpy) trend of massive galaxies; (2) minor mergers are a viable explanation of the f(clumpy) trend of intermediate-mass galaxies at z < 1.5, given a realistic observability timescale; and (3) major mergers are unlikely responsible for the f(clumpy) trend in all masses at z < 1.5. The clump contribution to the rest-frame UV light of star-forming galaxies shows a broad peak around galaxies with log(M-*/M-circle dot) similar to 10.5 at all redshifts. The clump contribution in the intermediate-mass and massive galaxies is possibly linked to the molecular gas fraction of the galaxies. The clump contribution to the SFR of star-forming galaxies, generally around 4%-10%, also shows dependence on the galaxy M-*, but for a given galaxy M-*, its dependence on the redshift is mild.