Providing observational tests for theories of mass accretion and angular momentum transfer in low mass star formation is the unifying theme of my research efforts. My thesis is devoted to the analysis of photospheric and dynamical properties of young embedded low mass stars in nearby star formation regions, as well as the characterization of the mass function of low mass stars and brown dwarfs in the solar neighborhood. To investigate these scientific issues, I have carried out large scale surveys of low mass stars using spectroscopy and photometry at optical and near-infrared wavelengths.

Near Infrared Spectroscopy of Embedded Class I Sources

     Measuring the properties of forming stars provides direct insight into the physics of star formation; creating an age sequence of such measurements places observational constraints on how young stars accrete mass and dissipate angular momentum, as well as the dependence of each mechanism on environment. NIR spectroscopy represents a prime opportunity to observe directly the photospheric properties of embedded young stars whose outer envelopes strongly absorb emission at shorter wavelengths. In collaboration with Thomas Greene (NASA Ames), Greg Doppman (NASA Ames; Gemini South) and Charles Lada (SAO), I have analyzed high resolution near infrared (NIR) spectra of 52 embedded young stars (hereafter referred to as Class I sources using the SED classification scheme of Lada 1987) in nearby star formation regions. From this sample we have measured photospheric effective temperatures, luminosities (separating stellar and accretion components), emission line strengths, surface gravities, as well as radial and rotational velocities for 41 sources with detected absorption lines (see Fig. 1). Results of the survey have been reported in a series of papers in the Astronomical Journal (Doppmann et al. 2005, Covey et al. 2005, Covey et al. 2006a).

Figure 1 - NIR spectra of two embedded protostars (bottom objects; from Covey et al. 2006c, in prep) and three standard stars for comparison (top objects; from Doppmann et al 2005). These spectral standards demonstrate the temperature sensitivity of the Mg/Al (2.106 & 2.11 microns) and Na (2.206 & 2.209 microns) line pairs, while the CO bandhead (2.293 microns) is sensitive to surface gravity. An unidentified feature (`Line X': 2.105 microns) appears at very cool temperatures. The spectra of 04154+2823 (shown with feature depths enhanced by a factor of 4) indicates a heavily veiled K7/M0 star, while 04181+2654 displays features similar to an M4 star.

     My analysis of the sample reveals significant differences in the observed rotation of Class I and presumably older Classical T Tauri stars (CTTS), suggesting that photospheric angular momentum must be extracted between the two phases (see Fig. 2). This result, interpreted within the context of magnetospheric accretion models, implies a factor of five difference in the mass accretion rate that characterizes each stage. An elevated Class I mass accretion rate is also consistent with our observations of emission lines tracing accretion (HI Br Gamma), though the observations do not appear to support accretion rates differing by two orders of magnitude as some models suggest (Covey et al. 2005). As well, a comparison of the radial velocities of Class I sources and the local CO gas constrains the dynamical state of Class I objects. I have measured a 3 km/sec upper limit for the 1-dimensional velocity dispersion for Class I sources that is fully consistent with the dispersion observed for CTTSs, limiting the amount of dynamical evolution which can occur between the two phases (Covey et al. 2006a). Lastly, current estimates of the stellar luminosities for Class I sources and CTTSs prevent a clear segregation of the two phases by age in the HR diagram, possibly due to uncertainties in de-reddening Class I sources given the complex geometrical structure of their circumstellar material.

Figure 2 - Observed projected rotation velocity as a function of mid-infrared SED slope (alpha). Larger values of alpha indicate redder, more deeply embedded stars. Symbols show objects in the Taurus (squares), rho Ophiuchi (triangles) and Serpens (diamonds) star forming regions; filled symbols indicate a rotation velocity upper limit. The dashed vertical line shows the canonical dividing line between embedded objects, on the right, and optically revealed sources, on the left. While a variety of rotation rates are seen at every evolutionary stage, the mean and maximum observed rotation velocities decline as sources become more optically visible, indicating angular momentum must be extracted from these sources as they emerge from their protostellar envelopes. (Data from White et al. 2004, Covey et al. 2005 and references therein.

     I have recently begun a new observational program to study the properties of very low luminosity Class I sources using moderate resolution NIR spectroscopy. Protostars with very low luminosities may be the precursors to fully formed substellar objects; identifying and analyzing Class I brown dwarfs will inform our understanding of the mass accretion process and its dependence on protostellar mass. Followup observations will be useful for confirming Class I brown dwarfs, however, as the near/mid-infrared colors and magnitudes expected for Class I brown dwarfs are shared by some extragalactic sources, while the effects of extinction and scattering from a highly structured circumstellar environment can lead one to mistake a Class I star seen through its disk for a Class I brown dwarf. In support of this program I authored a successful observing proposal (with a second proposal submitted) to obtain time with the NIRSPEC spectrograph on the Keck II telescope to follow up promising proto-brown dwarf candidates. An initial description of this research program, its motivation and observational technique will be published in a special issue of Astronomical Notes (Covey et al. 2006b), with publication of final results to follow shortly after our prospective July 2006 observing run.

The Mass Function of Nearby Low Mass Stars

     Though long separated from their initial formation events, main sequence stars carry encoded in their masses information about the star formation process. The distribution of stellar masses which result from star formation, described mathematically as the initial mass function (IMF; Salpeter 1955), helps us understand the physics of star formation, as well as providing a foundation for numerous studies concerning the evolution of stellar populations. Working with Suzanne Hawley (Univ. of Washington), I have carried out an observational program to construct a mass function of low mass stars from matched Sloan Digital Sky Survey (SDSS) and 2 Micron All Sky Survey (2MASS) point sources stretching over 30 square degrees on the sky. I have derived a preliminary mass function for this sample of 30,000 stars; I am currently using models of galactic structure to correct for the effects of sampling stars of different luminosities from different median heights within the Galactic disk.

     To calibrate this photometric sample, I have defined a complete subset of 500+ low mass stars for spectroscopic followup. Using this data, I can confirm the accuracy of SDSS photometric parallax relations, as well as quantify the effects of photometric errors on my derived mass and luminosity functions. The spectroscopic dataset, the product of 15+ nights of observing time over three years on the Apache Point Observatory 3.5m telescope, has verified that the photometric sample is free of contamination from non-stellar sources and provides traction for discriminating between dwarf and giant stars using luminosity sensitive spectral features. Measuring the strength of spectral features tracing magnetic activity and metallicity for each of these low mass stars allows them to be statistically assigned to the thin and thick disks of the Galaxy. Lastly, the spectroscopic sample has also identified a danger for matching between two surveys with quite different sensitivity limits; visual binaries in SDSS with a bright and faint component can be mismatched when the faint SDSS component is matched to the bright component's 2MASS counterpart. This results in an extremely red optical/NIR color as expected for very late type stars and brown dwarfs. As a result of spectroscopically targeting the reddest objects in the photometric sample, I have modified my selection routine to eliminate the vast majority of such spurious matches.

     Analysis of the complete photometric and spectroscopic dataset is ongoing, with submission of results expected by April 2006. Once this effort is complete, a fully calibrated, spectroscopically verified method for creating a stellar mass and luminosity function from matched SDSS/2MASS data can be applied to the entire SDSS photometric footprint, covering nearly 10,000 square degrees. This huge footprint will allow the creation of a luminosity and mass function from individual detections of millions of stars, surpassing previous efforts by multiple orders of magnitude and providing immense statistical traction for investigating slight deviations of the slope as a function of mass. The vast number of independent sight lines in the SDSS footprint will also allow the investigation of IMF variations as a function of position within the Galaxy. I am currently collaborating with Suzanne Hawley and John Bochanski (Univ. of Washington) to create the computational infrastructure needed for such an effort; once the last holes in the SDSS imaging footprint have been filled (expected spring 2006), a full analysis will begin.

Skills and Tools

     Over the course of these projects I have developed a set of skills and computational tools for analyzing multi-wavelength survey scale photometric and spectroscopic datasets. Creating a mass function from the intersection of the SDSS and 2MASS surveys required the management of many gigabytes of photometric data and the development of a catalog matching algorithm that robustly resolves matches for sources with multiple potential counterparts in a deeper survey. My work on the NIR spectroscopic survey of Class I sources, as well as the spectroscopic component of the SDSS/2MASS mass function, has led me to develop software routines to significantly automate the reduction and analysis of these large spectroscopic datasets. I have also created analysis pipelines to consistently measure the strength and shape of spectral features in these survey spectra; the set of indices I defined to analyze my SDSS spectroscopic sample are being investigated for inclusion as part of the automated SDSS II: SEGUE spectroscopic pipeline.