A surprisingly high pair fraction for extremely massive galaxies at z ∼ 3 in the GOODS NICMOS survey

We calculate the major pair fraction and derive the major merger fraction and rate for 82 massive (M-* > 10(11) M-circle dot) galaxies at 1.7 < z < 3.0 utilizing deep Hubble Space Telescope NICMOS data taken in the GOODS north and south fields. For the first time, our NICMOS data provide imaging with sufficient angular resolution and depth to collate a sufficiently large sample of massive galaxies at z > 1.5 to reliably measure their pair fraction history. We find strong evidence that the pair fraction of massive galaxies evolves with redshift. We calculate a pair fraction of f(m) = 0.29 +/- 0.06 for our whole sample at 1.7 < z < 3.0. Specifically, we fit a power-law function of the form f(m) = f(0)(1 + z)(m) to a combined sample of low-redshift data from Conselice et al. (2007) and recently acquired high-redshift data from the GOODS NICMOS survey. We find a best fit to the free parameters of f(0) = 0.008 +/- 0.003 and m = 3.0 +/- 0.4. We go on to fit a theoretically motivated Press-Schechter curve to this data. This Press-Schechter fit, and the data, shows no sign of levelling off or turning over, implying that the merger fraction of massive galaxies continues to rise with redshift out to z similar to 3. Since previous work has established that themerger fraction for lowermass galaxies turns over at z similar to 1.5-2.0, this is evidence that higher mass galaxies experience more mergers earlier than their lower mass counterparts, i.e. a galaxy assembly downsizing. Finally, we calculate a merger rate at z = 2.6 of R < 5 x 10(5) Gpc(-3) Gyr(-1), which experiences no significant change to R < 1.2 x 10(5) Gpc(-3) Gyr(-1) at z = 0.5. This corresponds to an average M-* > 10(11) M-circle dot galaxy experiencing 1.7 +/- 0.5 mergers between z = 3 and 0.