A submillimeter continuum study of massive star formation in NGC 7538 and M17.

摘要

I have used sensitive submillimeter continuum images to probe the earliest stages of massive star formation in NGC 7538 and M17, two of the nearest massive star-forming regions. I have located almost 200 cold, dense, potentially star-forming clumps: 77 in NGC 7538 and 121 in M17. Using images at two wavelengths, 850 and 450 mum, I have estimated the mean dust temperature of each clump. At least 4 of the NGC 7538 clumps and 38 of the M17 clumps appear to be cold and starless; these clumps include many excellent candidates for some of the youngest massive star-forming objects yet known. I have studied the mass distribution of the clumps in both regions. Fitting both regions with double power laws, I find best-fit power-law exponents for the high-mass ends of each mass function of -2.0+/-0.3 in NGC 7538 and -2.4+/-0.3 in M17. On the same scale, the Salpeter stellar initial mass function has exponent -2.35, which suggests a correlation between the clump and stellar mass functions. I have used a consistent fitting technique to fit 11 submillimeter clump mass functions drawn from studies of 7 different star-forming regions whose clumps collectively span five orders of magnitude in mass. I find that the clumps in low-mass star-forming regions appear to obey a lognormal mass distribution, while those in massive star-forming regions obey a double power law mass distribution. Intermediate-mass star-forming regions have clump mass functions which are equally well described by either lognormal or double power law distributions. I suggest that it is the stochastic evolution of the clump masses, governed by processes such as turbulent fragmentation and gravitational N-body interactions, which gives rise to this apparent evolution of the clump mass function from power-law at high masses to lognormal at low masses. This interpretation is consistent with theories which invoke the Central Limit Theorem to explain the origin of the clump and stellar mass functions. I have also used non-parametric test to show that, when their different intrinsic mass scales are normalized away, most of these 11 observational mass functions are consistent with being random samplings from the same parent distribution. This suggests a previously-unrealized fault in the measurement of such mass functions, an intriguing commonality among their shapes, or possibly both. The latter interpretation hints at new evidence for the origin of the stellar mass function in the mass function of pre-stellar clumps on larger mass and spatial scales than previously considered.