2009ApJ...699..150S

We present and categorize Spitzer infrared spectrometer spectra of 294 objects in the Large Magellanic Cloud (LMC) to create the largest and most complete catalog of massive young stellar object (YSO) spectra in the LMC. Target sources were identified from infrared photometry and multiwavelength images indicative of young, massive stars highly enshrouded in their natal gas and dust clouds. Several objects have been spectroscopically identified as non-YSOs and have features similar to more-evolved stars such as red supergiants, asymptotic giant branch (AGB), and post-AGB stars. Our sample primarily consists of 277 objects we identify as having spectral features indicative of embedded YSOs. The remaining sources are comprised of seven C-rich evolved sources, eight sources dominated by broad silicate emission, and one source with multiple broad emission features. Those with YSO-like spectra show a range of spectral features including polycyclic aromatic hydrocarbon emission, deep silicate absorption, fine-structure lines, and ice absorption features. Based upon the relative strengths of these features, we have classified the YSO candidates into several distinct categories using the widely used statistical procedure known as principal component analysis. We propose that these categories represent a spectrum of evolutionary stages during massive YSO formation. Using our catalog we put statistical constraints on the relative evolutionary timescale of processes involved in massive star formation. We conclude that massive pre-main-sequence stars spend a majority (possibly as high as 90%) of their massive, embedded lives emitting in the UV. Half of the sources in our study have features typical of compact H II regions, suggesting that massive YSOs can create a detectable compact H II region half-way through the formation time present in our sample. This study also provides a check on commonly used source-selection procedures including the use of photometry to identify YSOs. We determine that a high success rate (>95%) of identifying objects with YSO-like spectra can be achieved through careful use of infrared color-magnitude diagrams, spectral energy distributions, and image inspections.