We study the evolution of quiescent galaxies at 0.5 < z < 2.5 as a function of stellar mass , using very deep NIR imaging data taken with the Multi-Object Infrared Camera and Spectrograph on the Subaru Telescope in the GOODS-North region .
The deep NIR data allow us to construct a stellar mass-limited sample of quiescent galaxies down to \sim 10 ^ { 10 } M _ { \odot } even at z \sim 2 for the first time .
We selected quiescent galaxies with age/ \tau > 6 by performing SED fitting of the multi broad-band photometry from the U to Spitzer 5.8 \mu m bands with the population synthesis model of \citet bru03 where exponentially decaying star formation histories are assumed .
The number density of quiescent galaxies increases by a factor of \sim 3 from 1.0 < z < 1.5 to 0.5 < z < 1.0 , and by a factor of \sim 10 from 1.5 < z < 2.5 to 0.5 < z < 1.0 , while that of star-forming galaxies with age/ \tau < 4 increases only by factors of \sim 2 and \sim 3 in the same redshift ranges .
At 0.5 < z < 2.5 , the low-mass slope of the stellar mass function of quiescent galaxies is \alpha \sim 0 – 0.6 , which is significantly flatter than those of star-forming galaxies ( \alpha \sim -1.3 – -1.5 ) .
As a result , the fraction of quiescent galaxies in the overall galaxy population increases with stellar mass in the redshift range .
The fraction of quiescent galaxies at 10 ^ { 11 } –10 ^ { 11.5 } M _ { \odot } increases from \sim 20–30 % at z \sim 2 to \sim 40–60 % at z \sim 0.75 , while that at 10 ^ { 10 } –10 ^ { 10.5 } M _ { \odot } increases from \lesssim 5 % to \sim 15 % in the same redshift range .
These results could suggest that the quenching of star formation had been more effective in more massive galaxies at 1 \lesssim z \lesssim 2 .
Such a mass-dependent quenching could explain the rapid increase of the number density of \sim M ^ { * } galaxies relative to lower-mass galaxies at z \gtrsim 1–1.5 .