TRACING STAR FORMATION IN THE ROSETTE MOLECULAR CLOUD

Transcription

1 TRACING STAR FORMATION IN THE ROSETTE MOLECULAR CLOUD By JASON ERIC YBARRA A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2013

2 2013 Jason Eric Ybarra 2

3 I dedicate this dissertation to Jennifer Mary Camille Cain and to the memory of my grandfather Jesse Ybarra ( ) 3

4 ACKNOWLEDGMENTS There are so many people who have supported and helped me to get to this point. This is my attempt to acknowledge them. I would like to acknowledge my advisor, Elizabeth Lada. She continues to think I am a better writer than I think I am; Throughout these last few years, this expectation has driven three published papers and this dissertation. I feel I was given relative freedom, but I also had to explain what I was doing and justify the directions I wanted to take my research. My ideas were both nurtured and challenged. Without her guidance this dissertation would not be possible. I would like to thank my NASA GSRP advisor, Malcolm Niedner, for encouragement and memorable conversations, many which extended beyond science to the areas of film, culture, and food. I acknowledge my previous advisor and collaborator, Mary Barsony, from whom I learned the importance of being thorough and the importance of being a good scientific citizen. I also acknowledge my undergraduate advisor, Randy Phelps, who first introduced me to the field of star formation. Science is not done in a vacuum, and therefore, I thank Carlos Roman-Zúñiga, Jonathan Tan, John Bally, Charles Lada, Phil Myers, Naibi Marinas, Scott Fleming, Krista Romita, Zoltan Balog, Junfeng Wang, and Eric Feigelson for all the useful discussions, questions, and comments. I also thank my dissertation committee of Ata Sarajedini, Fred Hamann, and Steve Detweiler. Without friends such a journey would not be as interesting or fun. I thank Andro Rios, Bret van den Akker, Brandi Gartland, Raman Narayan, Ameer Thompson, Mr. Dave Sexton, Alejandro Robin Banks Rojas, Jessica Giles, Ralph Mayer, Nicky Cunningham, Kathryn Harris, Heather Furlong, Josh Spurgin, Emma Bickerstaff, Kathryn Cain, and Judi Cain. I also acknowledge the support given by River Saenz, Harriet Taniguchi, William DeGraffenried, John Flaherty, and the CSUS McNair Program. I thank Gary Busey for inspiration. 4

5 I want to thank my brother, David Shary, who has continued to be a source of encouragement and inspiration. I also acknowledge my parents, Winfried and Arleen Bauer, for encouraging my curiosity and appreciation of the natural world as a child. Finally, I would like to thank my fiancée, Jennifer Mary Camille Cain, for her love, support, and infinite patience. This work is based in part on archival data obtained with the Spitzer Space Telescope, which is operated by the Jet Propulsion Laboratory, California Institute of Technology under a contract with NASA. Financial support for my studies was provided by an award issued by JPL/Caltech, a Florida Space Grant fellowship, and a NASA Graduate Student Researcher Program (GSRP) fellowship through Goddard Space Flight Center. 5

6 TABLE OF CONTENTS page ACKNOWLEDGMENTS LIST OF TABLES LIST OF FIGURES ABSTRACT CHAPTER 1 INTRODUCTION Approach to the Problems Rosette Molecular Cloud Description of Chapters PROGRESSION OF STAR FORMATION IN THE ROSETTE MOLECULAR CLOUD Background Observations Spitzer IRAC observations and data reduction Chandra observations and data reduction Near-infrared photometry data Analysis Dust extinction distribution in the RMC Identification of YSOs Spatial distribution of the different YSO classes Age distribution through YSO ratios Discussion Star formation and column density Cluster properties Age gradients in the RMC main core Star formation as a function of distance from NGC Conclusion SPITZER IRAC DETECTION AND ANALYSIS OF SHOCKED MOLECU- LAR HYDROGEN EMISSION Background Calculations Molecular Hydrogen Emission in IRAC Color Space Example Summary

7 4 SPITZER AND NEAR-INFRARED OBSERVATIONS OF A NEW BI-PLOAR PROTOSTELLAR OUTFLOW IN THE ROSETTE MOLECULAR CLOUD Background Observations and Data Reduction Spitzer IRAC and MIPS data reduction Near-infrared molecular hydrogen observations and data reduction Results and Analysis IRAC color space of shocked gas IRAC color analysis of MHO Outflow source Discussion Structure of the outflow Deflection of the outflow Conclusions MOLECULAR HYDROGEN EMISSION SURVEY OF THE ROSETTE MOLEC- ULAR CLOUD Background Observations and Data Reduction Distinguishing Between Shocked and UV Excited Molecular Hydrogen Results Molecular hydrogen emission features Association of outflow activity and embedded clusters Driving sources Summary CONCLUSION Future Directions REFERENCES BIOGRAPHICAL SKETCH

8 Table LIST OF TABLES page 2-1 Disk fraction and age Cluster Properties Associated CO clumps Fractional contribution of the strongest H 2 lines to the IRAC bands Temperature estimates of HH Positions and flux estimates for the NIR H 2 knots of MHO IRAC color analysis of H 2 knots Outflow driving sources List of H 2 emission features

9 Figure LIST OF FIGURES page 1-1. Graphical representation of Young Stellar Object (YSO) evolutionary classes Optical R-band Digitized Sky Survey image of the Rosette Nebula and Rosette Molecular Cloud Spitzer 3-color image of the Rosette Molecular Cloud NICEST extinction map of the Rosette Molecular Cloud Nearest-Neighbor (NNM) density maps of the RMC for Class I/0, Class II, and Class III sources Nearest-Neighbor (NNM) density maps of the RMC main core region for Class I/0 and Class III sources Ratio maps Plot of extinction bin versus mean Class II to Class III ratio Plot of extinction bin versus mean Class I/0 to Class II ratio Plot of estimated age versus mean extinction IRAC [3.6] [4.5] vs. [4.5] [5.8] color-color plot indicating the region occupied by shocked H IRAC color-color plot for identified knots in HH IRAC 4.5 µm image and temperature map of HH Spitzer IRAC images of the outflow Near-infrared images of the outflow IRAC color-color plot indicating the region occupied by shocked gas composed of H 2 and CO for T max = K IRAC color-color plot indicating the region occupied by dissociatively shocked gas composed of H 2 and CO The outflow in IRAC color space Thermal map of the outflow based on color analysis of the IRAC data Column density map for H 2 of the outflow based on color analysis of the IRAC data MIPS 24 µm image of the outflow source

10 5-1 2MASS K-band image of the RMC with blue boxes indicating the regions surveyed in the H µm line NIR images and UV excited H 2 map of the PL01 cluster region NIR images and UV excited H 2 map of the PL02 cluster region NIR images and UV excited H 2 map of the PL03 embedded cluster NIR images and UV excited H 2 map of the PL06 embedded cluster NIR images and UV excited H 2 map of the PL07 embedded cluster NIR images and UV excited H 2 map of the REFL08 embedded cluster NIR images and UV excited H 2 map of the PL04 embedded cluster NIR images and UV excited H 2 map of the PL05 embedded cluster NIR images and UV excited H 2 map of the REFL09 embedded cluster Summed H µm line emission vs. number of protostars within 1 pc (2.14 ) of the cluster centers Summed H µm line emission vs. Class I/0 to Class II ratio within 1 pc (2.14 ) of the cluster centers Summed H µm line emission vs. total luminosity of protostars within 1 pc (2.14 ) of the cluster centers NIR H 2 and Spitzer images of an outflow in PL NIR H 2 and Spitzer images of outflow HH 871 ( ) in PL NIR H 2 and Spitzer images of outflows in REFL NIR H 2 and Spitzer images of an outflow in PL NIR H 2 and Spitzer images of the center of cluster PL

11 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy TRACING STAR FORMATION IN THE ROSETTE MOLECULAR CLOUD Chair: Elizabeth A. Lada Major: Astronomy By Jason Eric Ybarra August 2013 Most stars form in embedded clusters distributed throughout a molecular cloud. To fully understand how clusters and molecular clouds evolve over time, I used observations of the Rosette Molecular Cloud (RMC) to study the progression of star formation, the effects of environment, and the role outflows have. In order to investigate the progression of star formation, I conducted an analysis of the distributions of young stellar objects (YSOs) and the gas from which they form. Relationships between ratios of YSO evolutionary states and extinction reveal that stars form preferentially in high extinction regions and that the column density of gas rapidly decreases as the region evolves on timescales comparable to the protostar stage. The rapid removal or relocation of gas may account for the low star formation efficiencies observed in molecular clouds. The small age spread between the embedded clusters in the cloud suggests the H II region has negligible on the global star formation history of the cloud. To investigate the role outflows from forming stars have on cluster evolution, I developed a technique using mid-infrared imaging data from the Spitzer Space Telescope to detect and study the thermal structure of shocked molecular hydrogen (H 2 ) gas. I investigated the thermal structure of a prominent outflow in the RMC and studied interactions with its immediate surroundings. I used near-infrared H 2 observations to investigate the distribution of shocked H 2 emission throughout the cloud and find the emission appears to be more prominent in 11

12 younger regions. I find strong correlations between the total measured H 2 line emission, number of protostars, and ratio of protostars to Class II objects suggesting younger clusters have more outflow activity, and that outflow activity in the RMC decreases with age. This suggests outflows play a significant role in the gas removal within the clusters and subsequently affecting cluster and molecular cloud evolution. 12

13 CHAPTER 1 INTRODUCTION Most stars form in embedded clusters within giant molecular clouds (Lada & Lada, 2003). Molecular clouds are large structures of mostly molecular gas with sizes that range tens to hundreds of parsecs and with masses M. Embedded clusters are structures deeply embedded within a cloud, with typical spatial sizes less than 1 pc, and masses M (Gutermuth et al., 2009; Lada & Lada, 2003). Only a small fraction of gas within the clouds is dense, forms stars, and is associated with the clusters (Lada, 1992). Observations of molecular clouds suggest lifetimes comparable to cluster lifetimes, on order of a few Myr (Elmegreen, 2000). However, the connection between embedded cluster evolution and the evolution of the molecular cloud is not well understood. There are some important questions that still need to be answered: How does environment and location within a cloud affect star formation properties? How does star formation progress through a molecular cloud? What role do outflows play in the evolution of embedded clusters and clouds? Answering these questions is important in understanding the processes by which clusters and molecular clouds evolve over time. 1.1 Approach to the Problems Regions where stars form are heavily obscured by dust. Visual extinction prohibits observations of these dense regions where clusters are evolving. However, light at infrared wavelengths can penetrate further into these dusty environments. For every 10 magnitudes of visual extinction, wavelengths greater than 2 µm suffer less than 1 magnitude of extinction. Thus, in order to probe these regions of star formation, this study primarily uses data observed within the infrared regime (1 to 350 µm). This study makes extensive usage of data from the Spitzer Space Telescope which during its cryogenic phase ( ) had wavelength coverage of 3 to 180 µm. Near-infrared observations are used to map the extinction and also to study shocked emission from outflows. X-rays can also penetrate 13

14 the dusty cluster environments and young stars happen to be very X-ray active. Therefore, I also make use Chandra X-ray Observatory data to identify young stellar objects (YSOs), especially the older objects that do not display infrared excess. In order to understand how star formation progresses though a cloud, I compare the distribution of extinction to distribution of YSOs. I use the relative proportion of YSOs in various stages (Figure 1-1) as a probe of age, and subsequently use age gradients as a probe of progression. The progression of star formation should map the evolution of the dense gas and reveal clues about the evolution of the cloud and the timescale for gas removal. Additionally, the distributions of YSOs and gas can reveal the effect of environment and external influences such as nearby H II regions. In order to understand the role outflows play I measure the outflow activity within embedded clusters that are in various evolutionary stages. Molecular gas shocked by outflows emits strongly in the ro-vibrational lines of molecular hydrogen (H 2 ). The H 2 S(1 0) 2.12 µm line is particularly useful as it is one of the brightest lines and also resides within the K-band atmospheric window. I use near-infrared narrow-band imaging in the 2.12 µm line to compile a census of H 2 emission features and study the outflow activity in the clusters. Figure 1-1. Graphical representation of Young Stellar Object (YSO) evolutionary classes. 14

15 1.2 Rosette Molecular Cloud The target of this study is the Rosette Molecular Cloud (RMC; Figure 1-2). The Rosette is located at a distance of 1.6±0.2 kpc, with a mass of M, and is a known star forming region (Román-Zúñiga & Lada, 2008). This cloud has a ideal layout for studying star formation and the effects of environment. It is adjacent to an expanding H II region driven by the optically revealed cluster NGC Young embedded star clusters are distributed across the cloud into to four main regions corresponding to different environments (Phelps & Lada, 1997; Román-Zúñiga et al., 2008). The first region is the Rosette Nebula, which contains the least embedded and oldest cluster. The remaining cloud can then be separated into the ionization front of the H II region, the central core where the ionization front and the cloud are interacting, and a back core of the cloud far from any interaction with the nebula. In a near-infrared imaging survey of the RMC, Román-Zúñiga et al. (2008) found evidence of a possible temporal gradient across the cloud; Near-infrared excess measurements indicate clusters may be younger with increasing distance from nebula. This study builds upon this work by taking a closer look at the possible age gradients by using mid-infrared observations to distinguish between the protostars (Class I/0) and more evolved disk sources (Class III). Additionally, new Chandra observations increase the number of known older Class III sources and thus allow a comparison of the different YSO classes. 1.3 Description of Chapters In Chapter 2, I analyze the distributions of young stellar objects (YSOs) and compare that with the distribution of gas traced by extinction. I use observations from the Spitzer Space Telescope to identify young YSOs (Class I/0 and Class II) through their infrared excess, and I use observations from the Chandra X-ray Observatory to identify more evolved YSOs (Class III) through their X-ray activity. An extinction map of the RMC is made by measuring the reddening to background stars using JHK photometry. I develop a 15

16 Figure 1-2. Optical R-band Digitized Sky Survey image of the Rosette Nebula and Rosette Molecular Cloud. Red diamonds show the locations of embedded clusters. Blue contours reveal location of integrated 13 CO emission. Digitized Sky Survey images are produced at the Space Telescope Science Institute using photographic data obtained using the Oschin Schmidt Telescope on Palomar Mountain and the UK Schmidt Telescope. technique using YSO ratios to study age gradients across the cloud to better understand how star formation has progressed. In Chapter 3, I describe a technique for identification and analysis of shocked molecular hydrogen (H 2 ) using photometry from Spitzer InfraRed Array Camera (IRAC) observations. In Chapter 4, I use Spitzer IRAC color analysis and near-infrared narrow-band H 2 observations to study a particularly bright bi-polar outflow traced by bright H 2 emission. 16

17 The strong emission in the IRAC bands allow for for the mapping of the thermal structure of the flow. I study the interaction of the outflow with its immediate surroundings. In Chapter 5, I survey the embedded cluster regions for shocked H 2 using nearinfrared narrow-band observations. I use Spitzer IRAC colors to distinguish between shocked and UV excited emission. I describe outflow activity as it correlates to cluster evolution. In Chapter 6, I summarize the results of this study and discuss future directions. 17

18 CHAPTER 2 PROGRESSION OF STAR FORMATION IN THE ROSETTE MOLECULAR CLOUD 2.1 Background Determining the effect environment has on the star forming process is imperative to gaining a full understanding of star formation. Most stars form in embedded clusters distributed throughout a molecular cloud (Lada & Lada, 2003). These embedded clusters are often subject to different environmental conditions based on their physical location in the cloud. The Rosette Molecular Cloud (RMC) has a ideal layout for studying star formation and the effects of environment. At its northeastern end lies the Rosette Nebula which contains the OB association NGC The RMC contains many embedded clusters throughout the length of the cloud. The near-infrared study of Phelps & Lada (1997) identified 7 embedded clusters (PL01 07) within the cloud. Subsequent mid-infrared, near-infrared, and X-ray studies have refined and added to this list. Román-Zúñiga et al. (2008) confirmed the NIR clusters and also discovered two more (REFL08-REFL09) using the density distributions of near-infrared excess (NIRX) sources. Within the PL04 region, Román-Zúñiga et al. (2008) found the peak of NIRX sources (PL04a) to be spatially coincident with the NIR nebulosity. Poulton et al. (2008) found a concentration of Spitzer identified YSOs (cluster D = PL03b) offset by 4.3 (2.0 pc) from the center of the NIRX distribution PL03a. Additionally, a distribution of YSOs (cluster C = PL02b) was found just south of the NIR cluster PL02. Wang et al. (2009) found a cluster of X-ray sources in the northern end of PL04 indicating the presence of a population of Class III sources. The center of this distribution (cluster XC = PL04b) is offset from the peak of the NIRX Work in this chapter appears in Ybarra et al. 2013, The Astrophysical Journal, Volume 769, Issue 2, p 140. Reprinted with permission from the American Astronomical Society American Astronomical Society 18

19 distribution by 3.3 (1.5 pc). Additionally, a small distribution of X-ray sources (XB2) was found spatially coincident with MIR cluster PL02b. These embedded clusters are found in different environments. Clusters near the OB association are within the ionization front of the H II region. Further away from the ionization front, is a region that is associated with a high concentration of gas and star formation, where it is estimated almost half of the star formation in the cloud is taking place (Román-Zúñiga et al., 2008). This region may have experienced a shock front passing through it. Finally at the back of the cloud is a region where two embedded clusters have formed beyond the influence of the nebula. Román-Zúñiga et al. (2008) studied the RMC with a deep near-infrared JHK survey. In their analysis of NIR excess sources they found that the excess fraction within the clusters increased with distance away from the the center of the Rosette Nebula. This suggested an underlying age sequence in the cloud where cluster age decreases with increasing distance from the nebula. This sequence extends beyond the influence of the H II region and it was suggested that its origin is from the formation and evolution of the cloud. In this study we take a closer look at the distributions of these young sources and focus on tracking the progression of star formation within the cloud. 2.2 Observations This study makes use of data obtained from the Spitzer Space Telescope, the Chandra X-ray Observatory, and the FLAMINGOS instrument on the KPNO 2.1 m telescope. Figure 2-1 is a Spitzer IRAC 3-color image of the RMC survey region. The blue dashed lines show the extent of the FLAMINGOS JHK survey and the red lines show the boundaries of the Chandra X-ray observations. The locations and names of the embedded clusters are shown. 19

20 2.2.1 Spitzer IRAC observations and data reduction For this study we use IRAC µm and MIPS 24 µm data from program 3394 (PI: Bonnel) available in the Spitzer archive. The IRAC mapping covers a total projected area of degrees 2 for all four IRAC bands. Each IRAC pointing has a field of view of with 3 dither positions and an exposure time of 12 seconds per frame. Additionally, we observed the NIR cluster REFL09 with IRAC from a separate program (PI: Rieke). The REFL09 IRAC observation (α = 06:35:09.11 δ = +03:41:13.7) is a single pointing with a field of view of , 3 dither positions, and an exposure time of 12 seconds per frame. The IRAC frames were processed using the Spitzer Science Center (SSC) IRAC Pipeline v14.0, and mosaics were created from the basic calibrated data (BCD) frames using a custom IDL program (see Gutermuth et al. (2008) for details). The MIPS frames were processed using the MIPS Data Analysis Tool (Gordon et al., 2005). Source detection and aperture photometry was perform using the IDL software package PhotVis which is based in part on DAOPHOT (Gutermuth et al., 2004, 2008). The MIPS 24 µm frames were processed using the MIPS Data Analysis Tool. PSF photometry was performed on the MIPS data using the DAOPHOT IRAF package Chandra observations and data reduction The embedded Rosette clusters studied in this paper were observed with the Imaging Array of the Chandra Advanced CCD Imaging Spectrometer (ACIS-I; Garmire et al. (2003)) onboard of the Chandra X-ray Observatory, which has a field of view in a single pointing. These observations were taken on 2010 December 3 (ObsID 12388, covering the PL03 cluster), 2010 December 10 (ObsID 12142, PL06), 2011 January 14 (ObsID 12387, PL01), and 2011 January 18 (ObsID 12386, REFL09), with net exposure time of 19.6 ks, 39.3 ks, 24.5 ks, and 34.6 ks, respectively. They significantly supplement the previous Chandra campaign of the Rosette complex reported in Townsley et al. (2003) and Wang et al. (2010, 2009, 2008) which consisted of four 20 ks ACIS-I snapshots 20

21 Figure 2-1. Spitzer 3-color image of the Rosette Molecular Cloud [4.5 µm (blue), 5.8 µm (green), 24 µm (red)]. Blue dashed lines show the boundaries of the FLAMINGOS JHK survey (Román-Zúñiga et al., 2008). Red lines show the boundaries of the Chandra X-ray observations (Wang et al., 2009, this study). 21

22 in January 2001, a deep 75 ks ACIS-I image in January 2004 centered on the O5 star HD in NGC 2244, and one 20 ks ACIS-I pointing at the NGC 2237 sub-cluster in All images were taken in Timed Event, Very Faint mode (5 pixel 5 pixel event islands). We follow the same customized data reduction described in Wang et al. (2008, 2007) using the Chandra Interactive Analysis of Observations (CIAO, Fruscione et al. (2006); version 4.3) package provided by the Chandra X-ray Center. A detailed description of the X-ray data analysis will be presented in a separate paper (Wang et al. 2013, in preparation). For each of the ACIS fields, we extracted X-ray images in the kev band, and applied the source detection algorithm wavdetect 1 (Freeman et al., 2002) with a range of wavelet scales (from 1 to 16 pixels in steps of 2) and a source significance threshold of to produce a list of candidate sources. X-ray event extraction was made with our customized IDL script ACIS Extract 2 (AE; Broos et al., 2010). Using the AE-calculated probability P B that the extracted events are solely due to Poisson fluctuations in the local background, we rejected sources with P B > 0.01, i.e. those with a 1% or higher likelihood of being a background fluctuation. The trimmed source list includes 431 valid X-ray sources Near-infrared photometry data This study makes use of photometry data from the FLAMINGOS JHK imaging survey of the RMC (Román-Zúñiga et al., 2008) which is part of the NOAO survey program "Toward a Complete Near-Infrared Spectroscopic and Imaging Survey of Giant Molecular Clouds" (PI: E. A. Lada). The completeness limits for the survey are J = 17.25, H = 18.00, and K =

23 2.3 Analysis Dust extinction distribution in the RMC A dust extinction map was made with the NICEST algorithm (Lombardi, 2009) on the FLAMINGOS photometry catalog. We removed Class I/0 and Class II sources from the catalog as those sources have large intrinsic red colors and tend to bias the map near the centers of clusters. The extinction is estimated toward background sources by comparing the source NIR colors with the intrinsic color distribution measured from a nearby extinction-free control region. The control field is located at a similar galactic latitude as the RMC and was selected from an IRAS 25 µm map as being devoid of dust thermal emission (see Román-Zúñiga et al., 2008, Figure 2). Spatial smoothing is then applied to the extinction values to create the extinction map. The smoothing creates a map of the weighted mean as a function of position on the sky. The weighting function is composed of a Gaussian which weighs the individual extinction measurements based on angular distance from the center of the map point plus a correction term which compensates for the bias due to the low numbers of background sources at high extinctions. In the case of the RMC, the optimal Gaussian width was found to be 90. This is capable of resolving column density structures of the molecular cloud with projected sizes of about pc. Figure 2-2 shows the extinction map of the RMC. The contour levels indicate A V = 8, 10, 12, 14, 16, 20 mag. The locations of the embedded clusters are indicated. Comparison of the location of the embedded clusters with the extinction distribution suggest a correspondence between star formation with the highest extinction regions of the cloud Identification of YSOs Spitzer and Chandra ACIS data were used to identify Young Stellar Objects (YSOs) in the Rosette Molecular Cloud. In order to select the Class I/0 and Class II sources, we employ the color cuts of Gutermuth et al. (2008) and Kryukova et al. (2012) to our Spitzer catalog. For objects which have IRAC (3.6 µm, 4.5 µm, 5.8 µm, 8.0 µm) and MIPS 24 µm 23

24 A v PL04b PL05 PL04a a PL02 b PL01 PL06 REFL08 Dec. (J 2000) PL07 PL03b PL03a 3.8 REFL R.A. (J2000) Figure 2-2. NICEST extinction map of the Rosette Molecular Cloud. The contours represent extinction levels A V = 8,10,12,14,16,20 mag. Location of embedded clusters are indicated. detections, we identify Class I/0 sources using the following criteria: [4.5] [5.8] > 1 or [4.5] [5.8] > 0.7 and [3.6] [4.5] >

25 and [4.5] [24] > 4.7 From the remaining catalog we select Class II sources that have all of the following criteria: [4.5] [8.0] > 0.5 [3.6] [5.8] > 0.35 [3.6] [5.8] 3.5([4.5] [8.0]) 1.75 The Gutermuth [4.5] [5.8] color cut that distinguishes between Class I/0 and Class II sources is particularly useful as the extinction curve is relatively flat in that wavelength range and thus the color cut is insensitive to extinction. For sources without an IRAC 4.5 µm detection, we select Class I/0 sources by [5.8] [24] > 4.5 and [24] < 6 and for Class II sources [3.6] [5.8] < 0.35 and 2.0 < [5.8] [24] 4.5 For objects that do not have a MIPS 24 µm detection, we use the previous IRAC color cuts for objects with a [4.5] [5.8] color and the additional requirement of [5.8] [8.0] < 1 in order to filter out AGN and PAH galaxies. Class III sources do not display significant infrared excess and thus need to be identified another way. Fortunately, YSOs are known to emit X-rays at levels that can range many orders of magnitude above main sequence stars (Feigelson et al., 2007; Preibisch et al., 2005). Thus, X-ray observations can efficiently identify YSOs in molecular clouds. Class III candidate sources were selected from our Chandra ACIS observations and 25

26 the previously published X-ray catalog of Wang et al. (2009). Sources with colors of Class I/0 or Class II objects were then removed to create a catalog of Class III sources. In order to deal with extragalactic contaminants in our sources list, we also removed sources that did not have a NIR counterpart in the FLAMINGOS catalog. In our analysis of the X-ray properties, we found that the sources without a NIR counterpart had an average hardness ratio consistent with extragalactic sources (Wang et al., in preparation) Spatial distribution of the different YSO classes By analyzing the distributions of YSO classes separately one can probe the evolution of star forming regions. The different YSO classes represent different evolutionary stages, with Class 0 and I sources representing the youngest sources still embedded in their envelopes, Class II sources are a later stage of sources still accreting material from their disks, and Class III sources are diskless pre-main sequence stars. We employed the k-nearest neighbor (k-nn) density estimation algorithm, often referred to as the nearest neighbor method (NNM), to analyze the structures and distributions of the YSO classes. This method allows us to map the density distributions and subsequently identify regions of clustering (Casertano & Hut, 1985). The algorithm measures the distance, D k, between a source point and its k th nearest neighbor and estimates the local stellar number density as µ k = (k 1)/πDk. 2 For the Class I/0 objects we used a k=5, and for the Class II and Class III sources we used k=10. Using k values larger than one has the advantage of lessening the influence possible non-members of the set, e.g. background galaxies, have on the local density estimate (Ferreira, 2009). This is important when using Spitzer data as many background galaxies have colors that are similar to those of YSOs in both the near- and mid-infrared (Foster et al., 2008). Figure 2-3 presents the NNM density maps of the Class I/0, Class II, and Class III sources in the RMC. The dashed lines show the spatial boundaries of the survey data used to identify the sources. The embedded clusters are spatially coincident with the high stellar density regions in the maps. Using the NNM maps of the Class II and Class III 26

27 sources, we have identified a new cluster (PL06b). The maps reveal that the different YSO classes can have very different spatial distributions. For example the Class I/0 and Class III concentrations appear to be separate while the Class II sources appear more wide spread. This is especially evident in the central region of the cloud. Figure 2-4 shows the NNM density maps of the central region for the Class I/0 and Class III sources. A high density of Class I/0 sources is found in the center of this region and is spatially coincident with the highest extinction. In contrast the older Class III sources are found primarily around the perimeter on north and east sides of the central region. Class II sources are found throughout the central region. By numbers, the Class II sources dominate the cloud. The different density distribution of the YSO classes in the central region suggest a progression of star formation from the outer north and east sections towards the central Age distribution through YSO ratios In order to study the progression of star formation in the RMC, we investigate how the ratios between different YSO classes change throughout the cloud. As time progresses, Class I/0 sources will lose their envelopes and evolve into Class II sources. Because the lifetime of the Class I/0 phase is less than the Class II phase, the ratio of Class I/0 to Class II sources should decrease with the age of the region (cf. Myers, 2012). Similarly, as time progresses, Class II sources will lose their disks and evolve into Class III sources. Disk fractions in clusters have an empirical relationship with age (Haisch et al., 2001). From different Spitzer IRAC studies of young clusters, the average disk fractions at different cluster ages are 75% at 1 Myr, 50% at 2 3 Myr, 20% at 5 Myr, and 5% at 10 Myr (Williams & Cieza, 2011). Thus, the Class II to Class III ratio within a region will also decrease with age (Table 2-1). This relationship between the Class II to Class III ratio and age allows us to probe age progressions within the cloud. We constructed ratio maps by estimating the ratio of the number of one class of sources, N 1, to that of another, N 2, (e.g. Class I/0 to Class II) within a projected region of the sky across a grid. We chose the size of the region by trying to minimize both 27

28 Class I/0 PL02a PL04b PL02b PL04a PL01 PL05 REFL08 PL06 PL06b REFL09 PL07 PL03b PL03a Class II PL02a PL04b PL02b PL04a PL01 PL05 REFL08 PL06 PL06b REFL09 PL07 PL03b PL03a Class III PL02a PL04b PL02b 4.4 PL04a PL01 Dec. 4.2 PL05 PL06 PL06b REFL08 REFL PL07 PL03b PL03a R.A Figure 2-3. Nearest-Neighbor (NNM) density maps of the RMC for Class I/0, Class II, and Class III sources. The density contours are µ = 2.6, 4.6, 7.7, 12.8 stars pc 2. The background contours are A V = 8, 10, 12, 14, 16, 20 mag. Dotted lines in the first two panels show the coverage of the Spitzer observations and the dotted lines in the last panel show the coverage of the Chandra observations. 28

29 Class III Class I/0 4.5 PL04b PL04b 4.4 PL04a PL04a PL05 PL REFL08 REFL Dec. PL06 PL06b PL06 PL06b R.A Figure 2-4. Nearest-Neighbor (NNM) density maps of the RMC main core region for Class I/0 and Class III sources. The density contours are µ = 2.6, 4.6, 7.7, 12.8 stars pc 2. The background contours are A V = 8, 10, 12, 14, 16, 20 mag. resolution and uncertainty. The probability density function of the ratio, R, is p(r N 1, N 2 ) = RN 1 (N 1 + N 2 + 1)! (R + 1) N 1+N 2 +2 N 1!N 2! assuming a uniform prior for the ratio (Jin et al., 2006). We use the expectation value of the ratio as our estimator, ˆR = N N 2, 29

30 with variance, σ 2 R = (N 1 + 1)(N 1 + N 2 + 1) N 2 2 (N 2 1) In each point on the grid, for a non-zero ratio value, we require σ R / ˆR For the Class I/0 to Class II ratio we chose a circular region with radius 1.1 pc and for the Class II to Class III ratio a circular region with radius 1.4 pc. The radius for the Class II to Class III ratio is larger because the magnitude cuts (see next paragraph) reduce the number of sources. The regions are sampled across the grid at intervals of one-third the region radius. The created ratio map provides a visual representation of age gradients and can then map the progression of star formation. In order to compare Class II and Class III sources it is necessary to make sure the samples are uniform and unbiased (Gutermuth et al., 2004). We first restrict the samples to sources which are detected in the FLAMINGOS JHK survey. We estimate the Class II sample is complete to H = 15 and the Class III sample to be complete to H = 14. We use the H-band luminosity as the emission from disks at this wavelength is negligible. We de-redden our samples to a 2 Myr isochrone (Baraffe et al., 1998) and limit the samples to H 14, making the samples complete to the same depth across the cloud. Figure 2-5 shows the two ratio maps, the Class II to Class III on the left and the Class I/0 to Class II on the right. These maps are over-plotted on top of the extinction contours. For the Class II to Class III ratio, the contour levels are R II:III = 0.5, 1.0, 2.0, 3.0. For the Class I/0 to Class II ratio, the contour levels are R I:II = 0.1, 0.3, 0.4, 0.5. Visual inspection of the maps suggest a correlation between the ratios and extinction, where higher ratio values are spatially correlated with higher extinction. To further investigate the correlation between these ratios and extinction we also determine the average extinction in each region. We binned the regions by extinction into 1 mag bins and calculated the weighted mean ratio in each bin. Figure 2-6 presents a plot of extinction bin versus mean ratio of Class II to Class III sources. The figure shows that the ratio increases monotonically with extinction. The plot can be fitted with a shallow 30

31 A B Figure 2-5. Ratio maps. A) Class II to Class III ratio map. The contour levels are R II:III = 0.5, 1.0, 2.0, 3.0. B) Class I/0 to Class II ratio map. The contour levels are R I:II = 0.1, 0.3, 0.4, 0.5, 0.7. The background contours are A V = 8,10,12,14,16. Dotted lines show the coverage of the Spitzer observations and dash dotted lines show the coverage of the Chandra observations. linear fit for A V < 13 mag, and a steeper linear fit for A V > 13 mag. Thus, decreasing age is related to increasing extinction. Figure 2-7 shows the extinction bin versus mean Class I/0 to Class II ratio. At high extinctions, A V > 17 mag, the Class I/0 to Class II ratio also steeply increases with extinction. However, this ratio appears flat for the extinction range A V = 8 17 mag. To investigate this further we look at the contributions to the graph from the main core and cluster PL07 separately (Figure 2-7 B & C). We find that for the main core of the cloud, where most of the star formation is taking place, the ratio has a positive monotonic relation with extinction. Cluster PL07 region is different; It appears to have a relatively flat relationship between ratio and extinction, with a possible increase at A V = 7 8 mag. There is a small offset between the extinction peak and Class I/0 density peak; which suggests a recent expulsion of gas from the center of PL07. For the cloud as a whole, the mean ratios between the Class I/0 and Class II sources have a small 31

32 Class II : Class III ratio A v Figure 2-6. Plot of extinction bin versus mean Class II to Class III ratio. range ( ) which may correspond to a narrow range of ages or perhaps evidence for continuous star formation. The relationships between the two ratios and extinction reveal a trend of decreasing age with increasing extinction. This implies that stars may be preferentially forming in the highest extinction parts of the molecular cloud. 2.4 Discussion Star formation and column density The Class II to Class III ratio map traces regions with estimated ages of Myr, and these regions are spatially coincident with extinction values of A V = 4 18 mag. Using 32

33 A! B C Figure 2-7. Plot of extinction bin versus mean Class I/0 to Class II ratio. the empirical relationship between disk fraction and cluster age, we can fit a power law relation to the Class II to Class III ratio and age. Then we can directly study the relation between age and extinction. Figure 2-8 shows a plot of estimated age versus extinction. The plot suggests that the column density of gas decrease exponentially with time above A V = 5 mag, with a half-life, t 1/2 = 0.4 Myr. We estimate the average rate at which the column density of gas decreases as Σ 10 4 M yr 1 pc 2 which is over an order of magnitude larger than the star formation rate measured in nearby molecular clouds (Evans et al., 2009; Lada & Lada, 2003). Thus most of this gas is not removed through formation of stars and is possibly being relocated to other regions 33

34 A v Estimated Age (Myr) Figure 2-8. Plot of estimated age versus mean extinction. The age is estimated from the Class II to Class III ratio. The red dotted line shows the exponential fit with half-life, t 1/2 = 0.4 My, above A V = 5 mag. of the cloud. This is consistent with the study of nearby molecular clouds by Lada et al. (2010) which demonstrated that star formation on the scale of few Myr has a negligible effect on the total mass of the cloud. In the RMC main core, we find a similar relationship between the Class I/0 to Class II ratio and extinction. This suggests that star formation occurs preferentially in high extinction regions. Additionally, we find that over half of the clustered (µ 1.3 stars pc 2 ) Class I/0 sources are found at A V > 15 mag and all of them at A V > 7.5 mag. This too is consistent with the study of nearby molecular clouds by Lada et al. 34

35 (2010) that found an extinction threshold of A V 7 mag, above which the star formation rate was proportional to the mass of the cloud measured above that threshold Cluster properties The relationship between star formation and column density may have consequences for the formation and evolution of clusters. Table 2-2 shows the properties and YSO content of the clusters. The stellar content is measured within a 1 parsec (2.14 ) radius of the listed cluster center. Most of the embedded clusters in the RMC have a Class II to Class III ratio, R II:III, between 1 and 2. The weighted mean ratio of all the embedded clusters is R II:III = 1.2 ± 0.2, suggesting that most of the clusters started forming around the same time. There is a group of clusters with ratios suggesting a more recent episode of star formation, having R II:III > 3.0. This group includes clusters PL2a, PL02b, PL03b, and REFL09. These clusters however do not appear to have a significant Class I/0 population. This suggests that the star formation in these clusters is more coeval. The cluster REFL08 appears to have the most recent episode of star formation. This cluster has Class I/0 sources concentrated in dense filamentary structures seen in extinction. It has the highest ratio of Class II to Class III sources which is consistent with an age less than 1 Myr. Our X-ray observations do not cover embedded cluster PL07. This cluster has the second highest concentration of young Class I/0 objects. However, without knowledge of its Class III content, an age estimate for this region is not possible. Although there are some differences between the clusters, the age spread of these clusters is nonetheless small. The ages inferred from the YSO ratios for most of the clusters are between 1 to 3 Myrs Age gradients in the RMC main core The RMC main core is composed of clusters PL04(a & b), PL05, and REFL08. Each of the clusters is associated with a CO clump identified in Williams et al. (1995) CO survey of the RMC (Table 2-3). These three clusters are characterized by having more 35

36 YSOs than the other clusters in the cloud, which is consistent with the study by Román- Zúñiga et al. (2008) where it is estimated that half of the star formation in the whole cloud happens in this region. The REFL08 sub-region has the highest density of protostars and is spatially coincident with the gas surface density peak. Its ratio of Class I/0 to Class II sources and its dearth of Class III sources suggest the age of this region to be less than 1 Myr. Hennemann et al. (2010) using Herschel observations found 27 protostars in this region, 7 of which were classified as very young Class 0 candidates. This region appears filamentary in the MIR Spitzer images with the main filament running north to south. The southern end of the main filament is coincident with the center of NIR cluster REFL08 and is at the intersection of two smaller filaments. This region appears to be the youngest region of the cloud. Cluster PL04 is located north of REFL08, while PL05 is located to the east. Both clusters, PL04 and PL05, appear to be older with Class II to Class III ratios consistent with ages 1 2 Myrs. The age gradient across this region as a whole suggests that star formation began in clusters PL04 and PL05 first, followed by star formation in REFL08 as a more recent event. Based on the YSO content, we estimate the age difference across the region to be about 1 Myr. It is possible that star formation feedback from clusters PL04a and PL05 pushed the gas into its present state and thus triggered the formation of REFL08. Alternatively, the formation of cluster REFL08 may not have been triggered by PL04 and PL05. The age progression may be a consequence of the formation of the cloud. This scenario is consistent with the numerical simulations of dynamic molecular cloud formation by Hartmann et al. (2012). In these simulations, stars can form from the initial density fluctuations resulting from turbulence during the formation of a molecular cloud. These stars form before the global gravitational collapse of the cloud leads to a main phase of star formation. 36

37 2.4.4 Star formation as a function of distance from NGC 2244 We find that the spatial distribution of young Class I/0 sources throughout the cloud has little or no correlation to location in the cloud or distance form NGC The age of NGC 2244 is estimated to be 2 3 Myrs which is consistent with the ages of the other clusters inferred from the YSO ratios. Age progressions appear to be limited to small regions, where surface density enhancements lead to more recent star formation episodes. Clusters PL02a and PL02b, spatially coincident with the visible edge of the H II region, have Class II to Class III ratios consistent with a recent ( 1 Myr) episode of star formation. The formation of these clusters may have been triggered by NGC 2244, consistent with the X-ray analysis of Wang et al. (2009). Our analysis finds a local age progression surrounding REFL08 which is consistent with the studies of Román-Zúñiga et al. (2008). This progression does not extend through the cloud. It appears that any effect the H II region has on the star formation history is secondary to that of the primordial collapse of the cloud. Our analysis suggests that star formation started throughout the complex due to a primordial collapse and has progressed through a series of localized episodes of formation closely following each other. Table 2-1. Disk fraction and age. Age (Myr) D.F. R II:III 1 75% % % 0.3 Columns 1 and 2 are cluster ages and average disk fractions from the literature (Williams & Cieza, 2011). Column 3 is the associated Class II to Class III ratio. 2.5 Conclusion In our analysis of the different YSO classes in the RMC we find that they often have different density distributions. We use the YSO ratios to study age gradients across the cloud to better understand how star formation has progressed. The relationships 37

38 between the YSO ratios and extinction suggest that star formation in the cloud occurs preferentially in high extinction regions and that the column density of gas rapidly decreases as the region evolves. This suggests rapid removal of gas may account for the low star formation efficiencies observed in molecular clouds. We find that progressions of star formation appear to be localized with a small overall age spread across the cloud, consistent with star formation starting across the cloud at roughly the same time. The distribution of YSOs in the RMC show little or no effect of NGC 2244 on the relative ages of the clusters except for the possible triggering of clusters PL02a and PL02b. Table 2-2. Cluster Properties. Cluster R.A. Dec. I/0 II III II III R II:III A V H 14 H 14 PL ± ±1.4 PL ± ±2.2 PL02b ± ±1.0 PL03a ± ±2.5 PL03b ± ±1.9 PL04a ± ±2.4 PL04b ± ±0.8 REFL ±2.4 PL ± ±2.2 PL ± ±1.6 PL06b ± ±1.2 REFL ± ±2.7 PL ND 10.6±1.4 The number of sources are counted within 1 pc (2.14 ) of the cluster centers Table 2-3. Associated CO clumps. Cluster Clump ν ν PL PL REFL CO clump data from Williams et al. (1995) 38

39 CHAPTER 3 SPITZER IRAC DETECTION AND ANALYSIS OF SHOCKED MOLECULAR HYDROGEN EMISSION 3.1 Background Protostellar outflows have strong molecular hydrogen emission in the wavelength range covered by the Spitzer InfraRed Array Camera (IRAC). Spitzer studies of known outflows reveal that shocked H 2 emission appears particularly strong in the 4.5 µm IRAC band (Noriega-Crespo et al., 2004; Teixeira et al., 2008). Many groups are beginning to visually inspect IRAC data to search for outflows and objects with extended H 2 emission based on the strong emission in the 4.5 µm band. Smith & Rosen (2005) created synthetic Spitzer images from their models of precessing protostellar jets. These models calculated the population of the first three vibrational levels by solving for statistical equilibrium and assuming local thermal equilibrium (LTE) for the rotational levels. Their simulations showed that the emission in the 4.5 µm band to be the strongest, which was consistent with observations. In taking a census of the young stellar objects (YSOs) in NGC 1333, Gutermuth et al. (2008) empirically determined an IRAC color cut based on observations of known shocked emission within NGC This was used to remove any possible shocked emission in their source list of YSOs. Due to the multitude of lines in the IRAC bands it was thought that information on the physical parameters of the gas could not be ascertained from the Spitzer IRAC data. In analyzing the IRAC images of IC 443, Neufeld & Yuan (2008) calculated the IRAC band ratios for shocked H 2 using the 13 strongest lines covered by IRAC but only included collisional excitation by H 2 and He in their calculations. They found the measured flux in the 3.6 µm band to be stronger than predicted in their calculations, which may have been due to neglecting collisional Work in this chapter appears in Ybarra & Lada, 2010, The Astrophysical Journal Letters, Volume 695, Issue 1, pp. L120-L123. Reprinted with permission from the American Astronomical Society American Astronomical Society 39

40 excitation with atomic hydrogen. Until now, the color space of shocked H 2 emission due collisional excitation with atomic hydrogen, He, and H 2 has not been calculated. In order to use Spitzer IRAC data to find outflows and study their properties we have calculated the IRAC color space of shocked H 2 emission. 3.2 Calculations We have calculated the intensities of shocked molecular hydrogen and determined the location of the shocked emission within IRAC color space. At typical shock temperatures and densities, the excitation of molecular hydrogen is through collisions with H atoms and He atoms, and with ground state ortho- and para-h 2 molecules. The typical densities in outflows are less than critical so we do not assume LTE. Instead the populations of the first 47 ro-vibrational excited states were calculated from solving the equations of statistical equilibrium where we set n(he)/n H = 0.10, n H = n(h) + 2n(H 2 ), and n(h)/n(h 2 ) = The atomic hydrogen fraction was set to the median value consistent with shock models and the temperature range we chose (Le Bourlot et al., 2002; Timmermann, 1998). To solve the equations we employed a non-lte code based on the method by Li et al. (1993). We used the latest H H 2 non-reactive collisional rate coefficients calculated by Wrathmall et al. (2007). The reactive collisional rate coefficients were derived from the relations of Le Bourlot et al. (1999). The rate coefficients for He H 2 and H 2 H 2 collisions are from Le Bourlot et al. (1999). The degeneracy of the states is given by g J = (2J + 1) for even J, and g J = 3(2J + 1) for odd J. The quadrupole transition probabilities used are from Wolniewicz et al. (1998). We calculated the populations of the states over a wide range of atomic hydrogen densities, n(h) = cm 3, and gas temperatures, T = K. The relative intensities of 49 lines that fall within the IRAC bands (3.08 µm < λ < 10.5 µm) were determined and used to calculate the IRAC band fluxes using the latest published IRAC spectral response (Hora et al., 2008) and calibration data (Reach et al., 2005). Table 3-1 lists the fractional contribution of the 40

41 strongest H 2 lines to each IRAC band for n(h) = 10 4 cm 3 and temperatures 2000 K and 4000 K. Table 3-1. Fractional contribution of the strongest H 2 lines to the IRAC bands. Transition λ (µm) IRAC T=2000 K T=4000 K ν = 1 0 O(5) ν = 2 1 O(5) ν = 1 0 O(6) ν = 2 1 O(6) ν = 0 0 S(14) ν = 1 0 O(7) ν = 0 0 S(13) ν = 0 0 S(12) ν = 0 0 S(11) ν = 0 0 S(10) ν = 1 1 S(11) ν = 0 0 S(9) ν = 1 1 S(9) ν = 0 0 S(8) ν = 0 0 S(7) ν = 1 1 S(7) ν = 0 0 S(6) ν = 0 0 S(5) ν = 1 1 S(5) ν = 0 0 S(4) Fractional contribution to the total emission from the H 2 lines in the bands convolved with the IRAC spectral response for n(h) = 10 4 cm Molecular Hydrogen Emission in IRAC Color Space Figure 3-1 shows location of shocked H 2 gas in IRAC [3.6] [4.5] vs. [4.5] [5.8] color space. We find that the observed emission in the bands is a function of kinetic gas temperature and atomic hydrogen density. We restrict our plot to gas temperatures below 4000 K where H 2 emission is expected to be the dominant. Shocks that would produce gas temperatures in excess of 4000 K are likely to be dissociative (J-type) decreasing the abundance of H 2 molecules. These shocks are also likely to produce vibrationally excited CO emission and fine structure [Fe II] emission in addition to the H 2 emission. The CO molecule, which has a higher dissociation energy than H 2, is able to survive 41

42 Figure 3-1. IRAC [3.6] [4.5] vs. [4.5] [5.8] color-color plot indicating the region occupied by shocked H 2. Constant density (dotted) and temperature (solid) lines are indicated. The region to the left and above the dashed-dotted line was empirically determined by Gutermuth et al. (2008) to contain outflows. at higher shock velocities and temperatures. In these high energy shocks, CO becomes vibrationally excited and emits in ν = 1 0 (4.45 µm λ 4.9 µm) lines and can contribute significantly to the total emission from the shocked gas. (Draine & Roberge, 1984; González-Alfonso et al., 2002). Furthermore, the majority of [Fe II] lines in the range covered by IRAC fall within the 4.5 µm band. Consequently, 4.5 µm emission in excess to the color space defined by H 2 at 4000 K is likely to trace gas with T > 4000 K placing shocked emission into the upper left portion of the color-color diagram. We chose not to use the 8 µm band as there may be dust and PAH contamination in this band. PAH emission is particularly strong in the 8 µm band and it is still unclear what contribution PAHs may have in the emission of shocked gas from outflows. Dust 42

43 continuum emission also becomes a possibility as dust may survive the shock. Furthermore our calculations are restricted to a single temperature along the line of sight. In the case of multiple temperature components, the cooler gas will significantly contribute to the lower excitation pure rotational lines, 0 0 S(4) and 0 0 S(5), that dominate the emission in the 8 µm band. The 8 µm band would then trace the cooler temperature component while the other bands would be more sensitive to the hotter gas. Thus our analysis can include gas containing multiple temperature components along the line of sight but will still be restricted to analyzing only the hotter gas. As seen in Figure 3-1, the shocked H 2 lies in a well defined location in the colorcolor diagram. For comparison, the location of YSOs is also shown in the figure based on the criteria of Gutermuth et al. (2008) and Megeath et al. (2004). Shocked H 2 gas with sufficiently high temperature and atomic hydrogen density is found in a unique location on this diagram and can be distinguished from YSOs. This is consistent with the empirically determined color cut of Gutermuth et al. (2008). Therefore these colors offer an unambiguous method for searching for shocked gas from outflows/jets. However, there is overlap in IRAC color space between low temperature shocked H 2 gas and protostellar sources. Consequently surveys searching for outflows using IRAC color analysis will be restricted to finding flows containing higher excitation gas. Additional data at different wavelengths (2MASS, MIPS, etc.) and/or morphology may be able to break this degeneracy. Our results are consistent with empirical observations of strong 4.5 µm emission in outflows and with the hydrodynamic simulations of Smith & Rosen (2005). Using this color analysis, two methods can be applied to identify outflows in the images: 1) Visual inspection of 3-color images constructed out of the 3.6 µm, 4.5 µm, and 5.8 µm band data with appropriate scaling to enhance the shocked emission, and 2) analysis of the photometry of the field where features consistent with colors of shocked H 2 emission are selected. This can be accomplished by evaluating the color pixel by pixel across the field. 43

44 The location of the shocked H 2 in color space depends on its temperature. Therefore color analysis provides a new way to probe the temperature structure of the gas. This can be accomplished by evaluating the colors pixel by pixel across the field. The colors can then be compared to the colors of shocked H 2 and the temperature of the gas can be estimated. Resulting temperature maps are restricted to temperatures between K and high atomic hydrogen densities. The color space for H 2 at temperature less than 2000 K moves further into the color space of Class I/0 objects and it becomes more likely to misidentify scattered light from YSOs as H 2 emission. Note that, at low atomic hydrogen densities the constant temperature lines start to converge and thus temperature estimates from this region of color space may have large uncertainties. 3.4 Example We applied our analysis to Spitzer archival data of the known outflow HH 54. We evaluated the IRAC colors at each pixel in the field of HH 54. The median value of the image was used to estimate the background and was subtracted from the images. Figure 3-2 shows the color-color diagram for the knots (A,B,C,E,K,M) of shocked H 2 previously studied by Giannini et al. (2006). We plot all the pixels on the diagram that are encompassed by the knots. The majority of points fall within our calculated color space for shocked H 2 emission. One exception is knot A which contains several pixels with colors that fall more than 3σ outside the 4000 K boundary. This region of color space with excess 4.5 µm emission is consistent with higher gas temperatures and possible additional emission from CO ν = 1 0 and [Fe II]. Giannini et al. (2006) obtained spectroscopic data of various knots in HH 54 in the near-infrared and used H 2 emission lines to estimate the rotational and vibrational excitational temperatures of the gas. We created a temperature map of HH 54 (Figure 3-3) by selecting the pixels whose colors fall within the range we identified as belonging to shocked H 2 (T = K) and [4.5] [5.8] 1.5, and then estimated the temperature of the gas based on the pixels location in color space. The estimated temperatures are 44

45 compared to the vibrational (ν 1) excitation temperatures obtained by Giannini et al. (2006) in Table 3-2. Because typical atomic hydrogen densities in shocks are less than critical, the gas cannot be assumed to be in LTE. In these environments the rotational temperatures are often far below the kinetic gas temperature, while the ro-vibrational ν 1 excitation temperatures are close to the kinetic gas temperature. For most of the knots the temperatures we estimate from color analysis and the spectroscopically determined temperatures are consistent. There is an additional knot, labeled I (α J2000 =12:55:54.8, δ J2000 =-76:56:06), 5 arcseconds to the east of knot C seen in the IRAC images that is not found in the NIR images of HH 54 which we identify as the mid-infrared counterpart to the optical knot HH 54I (Sandell et al., 1987). The temperature of the gas in this knot peaks at 3500 K as seen in Figure 3-2 and 3-3. This knot is also found on the edge of the [Ne II] 12.8 µm map of HH 54 by Neufeld et al. (2006). The presence of the fine-structure [Ne II] line is consistent with the high temperature structure of this knot. Knot B (T 2600 K)is also found to be spatially coincident with strong [Ne II] 12.8 µm emission. The distribution of the [Fe II] 26 µm line (Neufeld et al., 2006) covers a region containing knots A, B, M. We find that spatial distribution of fine-structure emission lines is consistent with regions within our temperature map where T 2600 K. Knots A, B, and I are spatially aligned with each other as indicated by the green line in Figure 3. These high temperature knots may trace the jet component of the outflow. This line points in the direction of IRAS which is believed to be the source of the outflow. 3.5 Summary We have quantitatively shown that analysis of Spitzer data can be used to discover and characterize emission from protostellar outflows. Shocked H 2 with sufficiently high temperature and neutral atomic hydrogen density can be distinguished unambiguously from stellar objects in IRAC color space. IRAC color analysis is useful for studying intermediate-excitation shocked gas within the temperature range T = K. 45

46 Figure 3-2. IRAC color-color plot for identified knots in HH 54. Each point represents the colors at an individual pixel within the knots. Different knots are represented with different symbols. Table 3-2. Temperature estimates of HH 54. Knot α(j2000) δ(j2000) T gas T ν 1 A 12:55: :56: ±500 K 3000±150 K B 12:55: :56: ±300 K 3000±140 K C 12:55: :56: ±100 K 2960±150 K E 12:55: :56: ±100 K 2500±90 K K 12:55: :56: ±100 K 2500±90 K M 12:55: :56: ±400 K 3000±140 K Higher temperature gas may contain significant contribution from ionic lines like [Fe II] and also from the CO ν =1 0 emission band. Even with this limitation, Spitzer IRAC data provides a useful tool in the study of outflows. In particular, IRAC color analysis can be used to probe the thermal structure of the gas without the need of using spectroscopic data. 46

47 Figure 3-3. IRAC 4.5 µm image and temperature map of HH 54. A) IRAC 4.5µm image. Image center is at α(j2000) = 12 h 55 m 49 ṣ 5 δ(j2000) = Individual knots are identified and labeled according to Giannini et al. (2006). Knot I, not seen on the NIR, is identified as the optical counterpart to HH 54I. B) Temperature map of HH 54 based on IRAC color-color analysis for 2000 K T 4000 K. The black contour levels are 2000 K, 2500 K, 3000 K, and 3500 K. The green line connecting the higher temperature knots A, B, and I points toward the proposed source IRAS

48 CHAPTER 4 SPITZER AND NEAR-INFRARED OBSERVATIONS OF A NEW BI-PLOAR PROTOSTELLAR OUTFLOW IN THE ROSETTE MOLECULAR CLOUD 4.1 Background Outflows and jets from young stellar objects (YSOs) accompany the early stages of star formation. Outflows can manifest themselves as jets and knots of shocked material visible at optical and near-infrared wavelengths and also molecular emission observable at longer wavelengths. The outflowing material plays a role in removing the excess angular momentum from the YSOs allowing them to evolve into stars. Outflows are able to trace the history of mass loss and accretion of their driving sources. Studying the structure and properties of these flows may provide clues to understanding the connection between jets and the associated wide angle molecular flows (Reipurth & Bally, 2001). Additionally, this outflowing material interacts with its surroundings and may affect its environment, possibly regulating further star formation and cluster evolution. The energy and momentum inputted by outflows may disrupt the surrounding ambient gas, contribute to the turbulence in the cloud, and affect chemical processes (Bally, 2007). Ybarra & Lada (2009) developed a technique to study the thermal structure of shocked H 2 gas using color analysis of observations from the Spitzer InfraRed Array Camera (IRAC). Given the vast amount of Spitzer data available, this technique can be used to survey large regions and simultaneously find and analyze shocked emission. The IRAC color analysis enables the construction of temperature maps of the shocked gas which may in turn be used to probe the interaction of outflow with its surroundings. These maps may also be used to compare the properties of outflow with those of simulations allowing a better understanding of the Work in this chapter appears in Ybarra et al., 2010, The Astrophysical Journal, Volume 714, Issue 1, pp Reprinted with permission from the American Astronomical Society American Astronomical Society 48

49 Figure 4-1. Spitzer IRAC images of the outflow. The origin is set at (α,δ)(j2000) = (06 h 35 m 25 ṣ 0, ). physics involved and estimating the energy and momentum inputted by outflows into their environment. The Rosette Molecular Cloud (RMC) is a star forming region located at a distance of 1.6 kpc. Near-infrared imaging studies have revealed nine embedded clusters across the cloud (Phelps & Lada, 1997; Román-Zúñiga et al., 2008). Outflow activity in the cloud has been revealed through the [S II] narrowband imaging survey of Ybarra & Phelps (2004) and the 12 CO survey of Dent et al. (2009). In an analysis of the Spitzer IRAC images of the Rosette Molecular Cloud, we have discovered a structure with the morphology of a bipolar outflow that is visible in the images from all four IRAC bands (Figure 4-1). This structure can be seen in the images published by Poulton et al. (2008) although it is not discussed in their paper. In this study we analyze the outflow using near-infrared (NIR) narrow-band imaging of the flow to confirm the presence of shocked gas inferred from analysis of the the IRAC images. We improve the IRAC color analysis of Ybarra & Lada (2009) and use it to create temperature and column density maps of the outflow. Using both the NIR and IRAC data, we probe the physical conditions and structure of the outflow. 4.2 Observations and Data Reduction Spitzer IRAC and MIPS data reduction We used MIPS 24µm and IRAC µm data from program 3391 (PI: Bonnel) available in the Spitzer archive. The IRAC frames were processed using the Spitzer Science 49

50 Center (SSC) IRAC Pipeline v14.0, and mosaics were created from the basic calibrated data (BCD) frames using a custom IDL program (see Gutermuth et al. (2008) for details). The MIPS frames were processed using the MIPS Data Analysis Tool (Gordon et al., 2005) Near-infrared molecular hydrogen observations and data reduction Near-infrared, narrow-band observations of the outflow were obtained with the Infrared Side Port-Imager (ISPI) on the Blanco 4 meter telescope at the Cerro Tololo Inter-American Observatory (CTIO). ISPI employs a HgCdTe Hawaii-2 array with a arcmin field of view and a plate scale of pixel 1. The outflow was imaged using the H S(2) µm filter (λ c = µm, λ/λ =.007), H S(1) µm filter (λ c = µm, λ/λ =.01), and K cont filter centered at µm. The telescope was dithered with a 20 point dither pattern with a integration time of 60s at each dither position. The images were taken on the nights of 2008 December 17 and 2008 December 19 with total integration time of 40 minutes in each filter. The raw images were flat fielded, corrected for bad pixels, and linearized using the task osiris from the CTIO Infrared Reduction package. For each image, a sky frame was created by median combining the dithered images closest in time to the image. The IRAF tasks msctpeak and mscimage, which are part of the IRAF Mosaic Data Reduction Package, mscred, were used to correct the images for geometric distortions. The high order polynomial distortion terms were calculated with msctpeak using the 2MASS Point Source Catalog as the reference catalog and the distortion correction was applied with mscimage. The corrected images were aligned and then combined to form the final science images. 4.3 Results and Analysis Figure 4-1 shows the outflow in all four IRAC bands. The outflow appears as patchy regions of diffuse emission with an overall structure that is elongated and collimated in the E-W direction. Figure 4-2 shows the narrowband near-infrared emission images of the 50

51 outflow. The NIR H 2 knots coincide with the diffuse emission seen in the IRAC images. The NIR H 2 images confirm the presence of shocked gas and the interpretation that this structure is an outflow. The eastern end appears to truncate at a bow shock. Slightly west of the bow shock, the H 2 images reveal a small scale chaotic structure followed by a more linear chain of knots.the western end of the outflow appears slightly deflected northward followed by a bright knot (g) and then a complex structure of smaller knots. The NIR images were flux calibrated to the 2MASS Ks band by determining the magnitude difference between the 2MASS Ks band catalog values and the ISPI image magnitudes for stars in common. The zero point flux in each NIR image after calibration is then the filter bandwidth multiplied by the 2MASS Ks-band zero point flux density. This results in a relation between the counts sec 1 in each filter image and the flux in W cm 2. The narrow band continuum K cont images were scaled and subsequently subtracted from the H 2 line images. Figures 4-2c and 4-2d show the continuum subtracted H S(1) µm and H S(2) µm line images. The quality of the subtraction is good although there are some subtraction residuals from the brightest stars present in the subtracted images due to differences in wavelength and PSF combined with changing atmospheric conditions. Based on the observation, this outflow has been given the designation MHO 1321 in the Catalogue of Molecular Hydrogen Emission-Line Objects (MHOs) in Outflows from Young Stars 1 (Davis et al., 2010). The fluxes of the individual H 2 knots comprising this outflow were determined using a circular aperture on the continuum subtracted images. Table 4-1 lists the NIR fluxes of the H 2 emission knots. The flux uncertainty is composed of the rms background, poisson noise, and uncertainty from the flux calibration. 1 MHO catalogue is hosted by Liverpool John Moores University. ljmu.ac.uk/mhcat/ 51

52 Figure 4-2. Near-infrared images of the outflow. A) H S(1) µm line. B) K cont µm line. C) Continuum subtracted H S(1) µm. D) Continuum subtracted H S(2) µm. The horizontal scale is in arcminutes and the vertical scale is in arcseconds. The origin is set at (α,δ)(j2000) = (06 h 35 m 25 ṣ 0, ) IRAC color space of shocked gas In order to study the structure of the shocked gas, we applied the IRAC color analysis method developed by Ybarra & Lada (2009). We improved the color analysis method by including the effects of CO ν=1 0 band emission in the total emission of the shocked gas. The distribution of the population of pure rotational levels of CO due to collisional excitation with H 2, H, and He was calculated using the rate coefficients of Draine & Roberge (1984). We employed the method of González-Alfonso et al. (2002) to calculate the relative rotational population for the CO ν=1 vibrational level. The Einstein A values for the CO ν=1 0 rovibrational transitions were obtained using the oscillator 52

53 strengths of Hure & Roueff (1993). In our calculations, we set n H = n(h) + 2n(H 2 ), n(he)/n H = 0.1 and n(co)/n H = The fraction of atomic to molecular hydrogen was estimated by considering the rate of collisional dissociation by H atoms, R d = exp( /T ) cm 3 s 1 (Le Bourlot et al., 2002) and the rate of formation on grains, R f = T 1.5 cm 3 s 1, derived from Hollenbach & McKee (1979) with the cooling rates for H 2 and H 2 O (Le Bourlot et al., 1999, 2002). Figure 4-3 shows the location of shocked gas in IRAC [3.6] [4.5] versus [4.5] [5.8] color space for maximum shock temperature of T max = K for gas temperatures T = K and densities n H = cm 3. The square brackets refer to IRAC magnitudes. The post-shock fraction of atomic hydrogen was found to be n(h)/n H which is consistent with simulations of non-dissociative C-shocks (Wilgenbus et al., 2000). In the simulations by Wilgenbus et al. (2000) it was found that the atomic fraction is relatively constant over a wide range of maximum temperatures. Therefore we will assume our color space to be representative of non-dissociative C-shocks in general. The location of the shocked emission in IRAC color space depends on the gas density, fraction of atomic hydrogen, and the kinetic temperature of the gas. The [4.5] [5.8] color is strongly dependent on temperature, while the [3.6] [4.5] color has a strong dependence on the atomic hydrogen density. At high densities, the dependence on density decreases as the H 2 gas moves toward local thermal equilibrium (LTE). The slope of the reddening vector is similar to the approximate slope of the constant temperature lines. Thus temperature maps of high extinction regions remain accurate even if the extinction cannot be accounted for. However, unless extinction can be corrected for, accurate density information may not be attainable. 53

54 We fit an analytic form to the relationship between color and temperature for the non-dissociative case, T 3 = ([3.6] [4.5]) 2.11([4.5] [5.8]) +0.59([3.6] [4.5]) ([4.5] [5.8]) 2 where T 3 = T/10 3 in the color space defined by 2.0 > [3.6] [4.5] > 0.21([4.5] [5.8]) + 1.5, and 0.2 < [4.5] [5.8] < 2.0. The difference between the analytic fit and the calculated temperature-color relation over the defined range is less than 10%. Additionally, one can use the flux in the 3.6 µm image to estimate the column density of the shocked H 2. Using our calculations we fit the following analytic form to the relationship between column density, IRAC 3.6 µm flux density, and temperature, log(n H2 /F 3.6 ) = T T T n n 6 T 3 where N H2 is the column density of shocked H 2 in cm 2, F 3.6 is the IRAC 3.6 µm band flux density in units of MJy sr 1, and n 6 = n H /10 6 for 1.5 < T 3 < 5.0 and 2 < n 6 < 10. In order to investigate the color space of dissociative J-type shocks we simulated the gas properties of a dissociative shock by setting the maximum temperature of the gas to T max = K. Our calculations show significant dissociation of H 2 with increasing density. Figure 4-4 shows the IRAC color space for dissociatively shocked gas. As the molecular hydrogen gets dissociated, emission from the CO ν=1 0 band begins to dominate the 4.5 µm IRAC channel. The relationships between temperature, density and color are different for the case of the non-dissociative shock and the case of the dissociative shock. There is some degeneracy at the low density and low dissociation region of color space for the dissociative shock and at the high density non-dissociative shock region. These two models meet in a region of color space defined by H 2 gas in LTE. Although there is degeneracy, we expect 54

55 Figure 4-3. IRAC color-color plot indicating the region occupied by shocked gas composed of H 2 and CO for T max = K. the distribution in color space for the dissociative shock to primarily lie at [4.5] [5.8] < 0. Thus we define the domain of the dissociative shocked gas in color space to be [4.5] [5.8] < 0 and [3.6] [4.5] > 1.5, whereas we define the color domain of non-dissociate gas to lie primarily at [4.5] [5.8] > 0. By analyzing the color distribution of an outflow, it may be possible to distinguish between the cases. A pixel density distribution, produced from binning the colors of each pixel in the the outflow, can reveal the nature of the outflow by showing where most of the pixels lie in color space. It should be noted that the IRAC color analysis assumes dust and PAH emission is negligible. In order to prevent dust emission from affecting the color analysis, the 8 µm IRAC color is not used as there is evidence in some shocks of continuum dust emission within that wavelength range covered by the 8 µm channel (eg. Smith et al., 2006). 55

56 K 1500 K 2000 K [3.6] [4.5] CO ν=1 0 dominated region n H = 5x10 6 n(h)/n H = 0.7 n H = n(h)/n H = [4.5] [5.8] Figure 4-4. IRAC color-color plot indicating the region occupied by dissociatively shocked gas composed of H 2 and CO. Additionally, PAHs are very likely destroyed in shocks. In a recent study, Micelotta et al. (2010) show that strong shocks can destroy PAHs or severely denature them IRAC color analysis of MHO 1321 The IRAC 8 µm image of the outflow region reveals patches of absorption against the diffuse background (Figure 1). Of particular interest is a dark patch seen in absorption that bisects the outflow. We created an extinction map from the 8 µm data using a small scale median filter assuming a uniform background. We applied this extinction map to the images of the outflow region using the mid-infrared reddening law (KP, v5.0) of Chapman et al. (2009). However, this is not able to account for the total extinction in the line of sight toward the outflow. Nonetheless, the temperature-color relation is insensitive to extinction for non-dissociative shocks. We estimate the background using a ring median filter and subsequently remove this background from the IRAC 3.6 µm, 4.5 µm, and

57 µm images. A ring median filter is a median filter from which only the pixels within an annulus are used in calculating the median (Secker, 1995). The scale of this filter needs to be larger than the scale of the shocked emission otherwise the background will be overestimated, yet small enough to account for the large scale background fluctuations. The images were shifted and registered with each other and then IRAC colors at each pixel location were determined. Figure 4-5 shows the pixel density in IRAC color space for the knots of the outflow. Due to the lack of pixels whose colors are in or near the CO dominated region and our criteria above for non-dissociative shocks, we conclude that this shock is mostly non-dissociative and we can therefore estimate the thermal structure based on color analysis. We compared the colors to those of non-dissociative shocked gas with the cutoff [4.5] [5.8] 1.5. A thermal map was created by estimating the gas temperature based on the location of the pixels in color space. Figure 6 shows the thermal map of the outflow. We find that most of the NIR H 2 knots are spatially coincident with the high temperature regions of the flow. However, knots a and v do not have a corresponding IRAC derived temperature. The NIR images reveal stars in the line of sight for these knots which add to the emission and prevents IRAC color analysis from deriving temperatures. Seven of the knots have estimated temperatures greater or equal to K. By combining the NIR infrared and temperature data it is possible to estimate the extinction towards the brightest knots. For this we used the extinction cross-sections of Weingartner & Draine (2001). The median extinction to the knots is A V = 27. The knots j and k in the vicinity of the dark clump have higher extinction compared to the rest of the knots. We use the median extinction value to de-redden the 3.6 µm flux and use it create a column density map with our column density temperature relation. Figure 4-7 show the column density map of shocked H 2 in the flow. We find that there is also a correspondence between the NIR H 2 knots and regions of higher column density. Using the established 57

58 2.5 n H = [3.6] [4.5] K 4000 K 3000 K 2000 K n H = 2 x n H = 5x [4.5] [5.8] Figure 4-5. The outflow in IRAC color space. Contours indicate the pixel density of the outflow knots. The distribution of pixels and the lack of pixels in the CO dominated region indicate non-dissociative shocks. Figure 4-6. Thermal map of the outflow based on color analysis of the IRAC data. The contour levels indicate T = 1500 K, 2500 K, 3000 K, 4000 K. The origin is set at (α,δ)(j2000) = (06 h 35 m 25 ṣ 0, ). distance to the RMC of 1.6 kpc and the column density map we calculate the total mass of the shocked H 2 (T > 2000 K) in the outflow to be g ( M ). 58

59 Figure 4-7. Column density map for H 2 of the outflow based on color analysis of the IRAC data. The contour levels indicate N H2 = cm 2, cm 2, cm 2, cm 2. The origin is set at (α,δ)(j2000) = (06 h 35 m 25 ṣ 0, ) Outflow source The source of the outflow is not seen in the NIR nor in the IRAC images. However, inspection of the MIPS 24 µm image reveals a source ((α,δ)(j2000) = (06 h 35 m 25 ṣ 0, )) bisecting the outflow (Figure 4-8). Moreover, this source is spatially coincident with a dark patch seen in the 8 µm image. This patch is elongated nearly perpendicular to the outflow and the northwest part of it has the morphology of an outflow cavity. The dark patch is seen in the contours that indicate mass surface densities obtained through extinction mapping of the 8 µm imaging data by the method of Butler & Tan (2009). The contour levels correspond to mass surface densities of Σ = (2.5, 4.0, 5.0, 6.0) 10 3 g cm 2. This small cloud may be a remnant of the core from which the protostar formed. The morphology of the northern end of the cloud appears as a bi-polar outflow cavity with an opening angle θ H S(1) surface brightness contours are overlaid on the MIPS 24 µm image that bisect the MIPS source and a coincident with the cavity of the dark cloud. This source is also detected in the MIPS 70 µm and MIPS 160 µm imaging data. However, we are unable to estimate the flux in the MIPS 70 µm band image due to incomplete coverage and possible contamination from the adjacent knot (j) which may have emission from dust and the 63 µm [O I] line (Reipurth & Bally, 2001). Similarly, the MIPS 160 µm band image may include emission from the knot in addition to the source. This source is not detected at shorter wavelengths and thus can be classified as a Class 0 protostar. We propose this newly discovered protostar to be the source of the 59

60 outflow. Additionally, using the column density map, we find that the mass of the shocked H 2 to the east ( g) of this source is almost equal to the mass west of the source ( g). Faint extended 24 µm emission is also detected at the locations of the brightest H 2 knots (g & j). This may arise from fine-structure [Fe II] lines within the MIPS 24 µm band (Velusamy et al., 2007). This is consistent with our IRAC color analysis of these knots that reveal them to be high temperature regions (T K). This consistency between the IRAC color analysis and the MIPS 24 µm emission validates our usage of the color space for non-dissociative shocks. 4.4 Discussion Structure of the outflow The long axis of the flow extends to 3.3. With a distance of 1.6 kpc to the RMC, the flow would have a projected total length of 1.5 pc. The east lobe extends 2.3 from the MIPS source to the bow shock, while the west lobe extends only 1. Assuming a projected outflow velocity of 100 km s 1, using the east lobe we estimate the age of the outflow to be 10 4 years. This age is consistent with the typical age of a Class 0 source and thus this outflow may provide an accretion record of the protostar (Reipurth & Bally, 2001). We can estimate the mass flux of the outflow as Ṁ M(H 2 )v t l t where v t is the projected outflow velcocity and l t is the projected outflow length. Assuming the typical value of v t = 100 km s 1 and using the values for the H 2 mass and length of the east lobe, we estimate a mass flux of Ṁ 10 7 M yr 1. This value is consistent with those obtained from other outflows using spectroscopic data (Podio et al., 2006). The NIR H 2 data reveals the higher temperature regions of the outflow seen in the IRAC images. The IRAC images also show the cooler regions of the flow as the IRAC bands contain pure rotational H 2 lines in addition to ν = 1 0 and ν = 2 1 ro-vibrational lines. 60

61 Figure 4-8. MIPS 24 µm image of the outflow source. The green contours show the dark cloud that bisect the outflow and indicate the Σ values (2.5, 4.0, 5.0, and 6.0) 10 3 g cm 2 obtained through extinction mapping using the IRAC 8 µm imaging data (Butler & Tan, 2009). The blue contours are H S(1) surface brightness contours with levels indicating (0.5, 1.0, 2.0, 5.0) W m 2 sr 1. The H 2 contours reveal the location of the outflow. The scale of the image axes is in arcseconds. The origin is set at (α,δ)(j2000) = (06 h 35 m 25 ṣ 0, ). 61

62 The flow is spatially coincident with two lobes of high velocity CO gas observed by Dent et al. (2009). Similar to many HH flows which extend further than their CO counterpart, we find the eastern half to extend beyond the the eastern CO lobe. The east end of the H 2 flow ends in a large bow shock, while the west end reveals a complex structure resembling either a broken up bow shock or multiple smaller bow shocks. Although the outflow is linear on large scales, the IRAC and H 2 data reveal a region in the eastern lobe before the bow shock with a more chaotic structure. This deviation from a linear progression of knots may be due to a possible interaction with another outflow. The CO observations of Dent et al. (2009) reveal another flow in the NE-SW direction originating from the embedded cluster PL07 (Phelps & Lada, 1997) that points toward this region. A collision between the two flows may explain the morphology and high temperature of knot r. The distribution of knots may also be due to variations in in jet direction over time and thus an indication of jet precession Deflection of the outflow The western end of the outflow appears slightly bent northward. The outflow is deflected by an angle θ d = 20 where it appears to graze the densest region within the dark patch. The deflection angle remains small and appears to decrease slightly beyond the interaction region. This is consistent with the simulations by Baek et al. (2009) of outflows colliding with dense cloud cores where the impact parameter is large. The deflection of the outflow may explain why the western end of the flow is shorter than the eastern end as the outflow velocity is expected to decrease after the collision (Raga & Canto, 1995). This is consistent with the IRAC color analysis that reveals high temperature shocked gas (knot j) to the east of the collision. If the outflow is composed of episodic ejection of material, there may be collision between clumps of material moving through the flow due to the velocity change (Raga & Cantó, 2003). As these successive clumps collide they may give rise to the high temperature and high column density region (g) found slightly west of the deflection. This 62

63 interaction may also explain why the western lobe lacks the bow shock structure seen in the eastern end. 4.5 Conclusions We present the discovery of a new bi-polar outflow in the Rosette Molecular Cloud and use NIR narrowband and Spitzer imaging data to study the flow. We show that IRAC color analysis can be used to interpret the interaction of an outflow with its surrounding environment. Using our calculations of the IRAC space of non-dissociative shocked gas we fit analytic forms to the color-temperature and column density-temperature relationships. We verify that IRAC color analysis can reveal regions of shocked gas and find that the NIR H 2 knots correspond to regions of high temperature and or column density determined through color analysis. We find diffuse MIPS 24 µm emission, most likely from [Fe II] lines, to be coincident with regions of high temperature thus confirming the validity of using the non-dissociative shock IRAC color space The NIR line ratios combined with the temperature estimates allow for the determination of extinction along the line of sight which is used to create a column density map of the shocked H 2 gas. We deduce that the asymmetry in the outflow is due to interactions with the dense material to the west of the outflow source causing deflection and possibly deceleration of the outflowing material. Table 4-1. Positions and flux estimates for the NIR H 2 knots of MHO Knot R.A. (J2000) Dec. (J2000) H S(2) H S(1) a 06:35: :56: ± ±2.2 b 06:35: :56: ± ±2.2 c 06:35: :56: ± ±2.4 d 06:35: :56: ± ±2.8 e 06:35: :56: ± ±2.4 f 06:35: :56: ± ±2.1 g 06:35: :56: ± ±5.8 h 06:35: :56: ± ±1.8 i 06:35: :56: ± ±1.9 63

64 Table 4-1. Continued. Knot R.A. (J2000) Dec. (J2000) H S(2) H S(1) j 06:35: :56: ± ±3.4 k 06:35: :56: ± ±2.4 l 06:35: :56: ± ±2.1 m 06:35: :56: ± ±1.8 n 06:35: :56: ± ±2.5 o 06:35: :56: ± ±1.6 p 06:35: :56: ± ±2.6 q 06:35: :56: ± ±2.6 r 06:35: :56: ± ±5.0 s 06:35: :56: ± ±2.0 t 06:35: :56: ± ±1.8 u 06:35: :56: ± ±2.6 v 06:35: :56: ± ±2.1 w 06:35: :56: ± ±2.0 Flux is in units of W m 2. Aperture radius used is 7 pixels. Flux uncertainty includes calibration uncertainty. Table 4-2. IRAC color analysis of H 2 knots. Knot T (10 3 K) N H2 (10 18 cm 2 ) b 2.5± ±0.3 c 2.7± ±0.2 d 2.5± ±0.2 e 2.5± ±0.2 f 2.5± ±0.2 g 3.2± ±0.1 h 3.0± ±0.1 i 3.0± ±0.1 j 3.3± ±0.1 k 2.8± ±0.2 l 3.3± ±0.1 m 2.5± ±0.2 n 2.3± ±0.4 o 2.3± ±0.3 p 2.8± ±0.2 q 2.6± ±0.2 r 4.0± ±0.1 64

65 Table 4-2. Continued. Knot T (10 3 K) N H2 (10 18 cm 2 ) s 2.8± ±0.2 t 2.6± ±0.2 u 3.3± ±0.1 w 2.7± ±0.2 The estimated temperature is the column density averaged temperature determined from pixel colors within each aperture. The temperature uncertainty is the column density weighted standard deviation of the temperatures corresponding to the individual pixels. 65

66 CHAPTER 5 MOLECULAR HYDROGEN EMISSION SURVEY OF THE ROSETTE MOLECULAR CLOUD 5.1 Background Mass loss in the form of outflows is ubiquitous during the early stages of star formation. These outflows can provide information on the accretion and mass loss history of individual forming stars (Reipurth & Bally, 2001). On global scales outflows can provide feedback into the cloud and assist in regulating star formation. The outflows can be traced by emission that arises from shock heating when the outflow interacts with surrounding gas. The Rosette Molecular Cloud (RMC) provides an ideal layout in which to study various aspects of star formation. Previous studies of the cloud have revealed the presence of outflow activity. An optical narrow-band [S II] imaging survey of the cloud found evidence of shocked gas (Ybarra & Phelps, 2004) and a 12 CO survey revealed several molecular flows (Dent et al., 2009). Optical searches are limited by extinction that is present in molecular clouds and radio surveys often lack the resolution needed to disentangle flows and study the structures in detail. Shock excited gas can also be traced in the near-infrared (NIR), in particular, through the molecular hydrogen emission (H 2 ) lines. NIR narrow-band H 2 imaging of the PL06 region by Aspin (1998) revealed the existence of multiple bow shocks surrounding the central AFGL 961 binary. A follow up study of one of the [S II] emission features by Phelps & Ybarra (2005) reveal a series of shocked H 2 knots originating from an embedded YSO near cluster PL01. An unbiased search for H 2 emission within all the known embedded clusters has not previously been done. In this paper we present the results of our narrow-band H 2 imaging survey of the RMC embedded clusters. We use available Spitzer Infrared Array Camera (IRAC) photometry to distinguish between shocked and UV excited emission. We discuss the 66

67 correlation between the shocked emission and the clusters and also discuss the properties of the driving sources. 5.2 Observations and Data Reduction Near-Infrared, narrow-band observations of the RMC embedded clusters were obtained with the Infrared Side Port-Imager (ISPI) on the Blanco 4 meter telescope at the Cerro Tololo Inter-American Observatory (CTIO). ISPI employs a HgCdTe Hawaii-2 array with a arcmin field of view and a plate scale of pixel 1. The fields were imaged using a H S(1) µm filter (λ c = µm, λ/λ =.01) and a narrow-band continuum, K cont, filter centered at µm. The observations were taken during the nights of 2008 December 17 through December 19. Figure 5-1 shows the regions surveyed (blue boxes) over plotted on a gray-scale 2MASS 1 K-band image of the cloud. The Flamingos II Data Pipeline (FATBOY) was used to reduce the data (Warner et al., 2012). First a linearity correction was applied to the data with the known ISPI linearization coefficients 2. Flat fielding and sky subtraction were then applied. Astrometry of the individual frames was performed using SExtractor (Bertin & Arnouts, 1996) and SCAMP (Bertin, 2006). Distortion correction, alignment, and stacking were performed with SWarp (Bertin et al., 2002). The NIR images were flux calibrated to the 2MASS Ks band by determining the magnitude difference between the 2MASS Ks band catalog values and the ISPI image magnitudes for stars in common. 1 This publication makes use of data products from the Two Micron All Sky Survey, which is a joint project of the University of Massachusetts and the Infrared Processing and Analysis Center, funded by the National Aeronautics and Space Administration and the National Science Foundation

68 Figure MASS K-band image of the RMC with blue boxes indicating the regions surveyed in the H µm line. 5.3 Distinguishing Between Shocked and UV Excited Molecular Hydrogen Molecular hydrogen emission is a good tracer of shocked gas from outflows. However, H 2 emission can also arise from UV fluorescence in cold gas (Black & Dalgarno, 1976; Black & van Dishoeck, 1987). In this process external UV photons are absorbed by H 2 into its excited electronic Lyman (B 1 Σ + u ) and Werner (C 1 Π u ) bands. These excited molecules will either decay into the ground electronic state continuum and disassociate or decay into discrete ro-vibrational levels of the ground electronic state and then cascade through ro-vibrational transitions giving rise to infrared photons. The resulting line ratios from the H 2 fluorescent emission can often mimic those of shock heated gas. In order to 68

69 compile a census of outflow activity in the cloud it is important to be able to distinguish between the two excitation mechanisms. Ybarra & Lada (2009) developed a method of analyzing shocked H 2 using Spitzer IRAC photometry. The location of the shocked gas in IRAC color space was determined to be a function of temperature, density, and extinction. In this chapter we expand this analysis to investigate the IRAC colors of UV excited H 2 with the hope of be able to distinguish between UV and shock excited emission. We use the population of rovibrational levels from Sternberg & Neufeld (1999). We also included the pure rotational (ν=0) levels j = 8 10 through extrapolation using T gas = 500 K. Similar to Ybarra & Lada (2009) we used the Einstein-A transition coefficients (Wolniewicz et al., 1998), IRAC spectral response (Hora et al., 2008), and calibration data (Reach et al., 2005) to calculate the IRAC colors. We calculate the IRAC colors for the UV excited H 2 emission from the model of Sternberg & Neufeld (1999) to be [3.6] [4.5] = 1.2 and [4.5] [5.8] = 1.1. Taking into consideration reddening, uncertainties in the ortho-to-para ratio, and possible PAH contribution to the 5.8 µm band we define the following criteria as the color space of probable UV excited H 2 : 1.2 < [3.6] [4.5] < < [4.5] [5.8] Potential H 2 features that fall within this color space will be removed from the catalog and will not used for the calculation of outflow luminosities. 5.4 Results Molecular hydrogen emission features In Table 5-2 we present our census of the H 2 features in the RMC. The listed coordinates are that of the center of the emission. The flux is measured in pixels for which S/N > 2σ. MHO designations refer to the objects which have entries in the Catalogue of 69

70 Molecular Hydrogen Emission-Line Objects (MHOs) in Outflows 3 (Davis et al., 2010). The features listed in our census do not display Spitzer IRAC colors consistent with UV excited emission and thus we can assume the emission is due to shocked gas from outflows. In total we identify 181 individual H 2 knots in the RMC. Many of these individual knots comprise large scale outflows Association of outflow activity and embedded clusters In this section we investigate the association of outflow activity with the embedded clusters. The following plots are composed of four panels: an image of the region using the H µ filter, an image of the same region with the K cont 2.15 µm filter, an image showing the continuum subtracted pure H 2 emission, and a map showing regions with IRAC colors consistent with UV excited fluorescent H 2 emission. The clusters within the nebular region (PL01 and PL02) display much diffuse H 2 emission with morphology consistent with the optical edge of the nebula (Figures 5-2 and 5-3. IRAC color analysis reveals the probable excitation mechanism for the emission is UV fluorescent excitation. There is some H 2 that can be attributed to shock excitation. In cluster PL01 there are H 2 knots ( ) which appear to originate from an embedded Class I/0 object in the cluster. To the east of the cluster center there is also the HH 871 outflow (Phelps & Ybarra, 2005) also originating from a Class I/0 object. Cluster PL02 displays very little shocked emission; Only one H 2 emission feature (013) is found in the cluster center. The PL03 region contains a chain of knots ( ) NW of the cluster. The cluster itself is surrounded by diffuse H 2 emission (Figure 5-4). We find the majority of this emission to have IRAC colors consistent with UV excited fluorescent H 2. The geometry of 3 MHO catalogue is hosted by Liverpool John Moores University. ljmu.ac.uk/mhcat/ 70

71 Figure 5-2. NIR images and UV excited H 2 map of the PL01 cluster region. A) H µm image. B) K cont 2.15 µm image. C) Continuum-subtracted pure H 2 emission image. D) Map of probable UV excited H 2 from IRAC color analysis. the diffuse UV excited emission suggests that the source of UV photons is from within the cluster. Our image of PL06 (Figure 5-5 shows the H 2 knots first that were discovered by Aspin (1998). The bright center of the cluster (AFGL 961) is oversaturated in the Spitzer IRAC band images and therefore IRAC color analysis is unable to determine the nature of the H 2 emission in the center. The young clusters PL07 (Figure 5-6) and REFL08 (Figure 5-7) display more outflow activity than most of the other clusters. These two clusters have a high fraction of protostars which indicates youth (Ybarra et al., 2013). The H 2 knots of REFL08 are 71

72 Figure 5-3. NIR images and UV excited H 2 map of the PL02 cluster region. A) H µm image. B) K cont 2.15 µm image. C) Continuum-subtracted pure H 2 emission image. D) Map of probable UV excited H 2 from IRAC color analysis. distributed throughout the region and the distribution is consistent with multiple outflows, with possible overlap of flows along the line of sight. PL07 contains multiple H 2 emission features predominantly in the center of the cluster. Many of the features appear to trace one or more outflows. Clusters PL04, PL05, and REFL09 shows a dearth of H 2 emission (Figures 5-8,5-9,5-10). (Wang et al., 2009) found a concentration of Class III sources in PL04, and (Ybarra et al., 2013) found the YSO ratios of PL04 and PL05 to be consistent with ages of 2-3 Myr, indicating an older population of stars. The lack of H 2 emission in REFL09 may be 72

73 Figure 5-4. NIR images and UV excited H 2 map of the PL03 embedded cluster. A) H µm image. B) K cont 2.15 µm image. C) Continuum-subtracted pure H 2 emission image. D) Map of probable UV excited H 2 from IRAC color analysis. due to high extinction associated with the cluster. However, this may also be correlated to the low number of protostars detected in this cluster. In order to quantitatively investigate the amount of amount activity within the embedded clusters I summed up the H 2 emission from the knots with 1 pc of the cluster centers for each embedded cluster. Figure 5-11 shows a plot of summed H µm 73

74 Figure 5-5. NIR images and UV excited H 2 map of the PL06 embedded cluster. A) H µm image. B) K cont 2.15 µm image. C) Continuum-subtracted pure H 2 emission image. D) Map of regions with IRAC colors consistent with UV excited H 2. emission from knots versus number of protostars. The H 2 emission and the number of protostars are measured within 1 pc (2.14 ) of the cluster centers. Cluster PL06 is not plotted as it displays anomalously strong H 2 and due to oversaturation in Spitzer imaging its protostar count is at best a lower limit. There appears to be a relationship where clusters with higher numbers of protostars have brighter H 2 emission, which suggests more outflow energy within the cluster. The Pearson s correlation coefficient between these variables is We also investigate the relationship between summed H 2 emission and the ratio of protostars to Class II objects. Figure 5-12 shows a plot of H µm emission versus the ratio of Class I/0 to Class II sources within 1 pc (2.14 ) of the cluster centers. 74

75 Figure 5-6. NIR images and UV excited H 2 map of the PL07 embedded cluster. A) H µm image. B) K cont 2.15 µm image. C) Continuum-subtracted pure H 2 emission image. The location of bright knots are indicated. D) Map of regions with IRAC colors consistent with UV excited H 2. The ratio and ratio uncertainty were calculated from the formulas in Ybarra et al. (2013). The Pearson s correlation coefficient between these variables is We find that clusters with higher Class I/0 to Class II ratios tend to have more H 2 emission, which suggests outflow activity is stronger for younger regions. Interestingly, there does not appear to be a correlation between H 2 emission and total luminosity of the protostars in the clusters. We estimate the luminosity of the protostars with the method developed by Kryukova et al. (2012). using Spitzer IRAC and MIPS 24 75

76 Figure 5-7. NIR images and UV excited H 2 map of the REFL08 embedded cluster. A) H µm image. B) K cont 2.15 µm image. C) Continuum-subtracted pure H 2 emission image. The location of bright knots are indicated. D) Map of regions with IRAC colors consistent with UV excited H 2. µm photometry. Kryukova et al. (2012) found the bolometric luminosity of protostars to be a function of their MIR luminosity and slope of their Spectral Energy Distribution (SED). Figure 5-13 shows a plot of H µm emission versus total luminosity of protostars within 1 pc (2.14 ) of the cluster centers. The Pearson s correlation coefficient is , suggesting a very weak to no correlation between the variables. Even though a single object can dominate the total luminosity of a region, it is the number of sources that appears to be more important to the total outflow activity. 76

77 Figure 5-8. NIR images and UV excited H 2 map of the PL04 embedded cluster. A) H µm image. B) K cont 2.15 µm image. C) Continuum-subtracted pure H 2 emission image. D) Map of regions with IRAC colors consistent with UV excited H 2. Ybarra et al. (2013) find evidence that regions in the RMC remove gas very quickly as they age. The relationship between column density of gas and age was found to have a half-life of 0.4 Myr, which is similar to the lifetime of protostars. We compare this to the relationship between outflow activity and age and suggest that outflows play a significant role in gas removal early on in the evolution of a cluster Driving sources Many of the H 2 emission features are arranged in linear structures which allow for the identification of the driving sources. In the center of cluster PL01 we find H 2 emission knots ( ) that appear to trace out a outflow originating from an embedded 77

78 Figure 5-9. NIR images and UV excited H 2 map of the PL05 embedded cluster. A) H µm image. B) K cont 2.15 µm image. C) Continuum-subtracted pure H 2 emission image. D) Map of regions with IRAC colors consistent with UV excited H 2. protostar (Figure 5-14). In the NIR the source appears to have the morphology of a reflection nebula, possibly revealing a cavity formed from the outflow. The outflow source has IRAC colors of a Class I/0 object and is one of the brightest MIPS 24 µm sources in the cluster. The other outflow in the PL01 region is HH 871 (Figure 5-15) discovered by Phelps & Ybarra (2005). This flow is composed of many H 2 knots ( ) and appears to originate from a Class I/0 object. Within the young cluster REFL08 there is a chain of knots ( ,041,042) which appear to trace out an outflow of point symmetry about an embedded MIPS 24 µm source (s4; Figure 5-16). The western end (033) of the outflow appears equidistant to the eastern end (042) from the proposed driving source with an 78

79 Figure NIR images and UV excited H 2 map of the REFL09 embedded cluster. A) H µm image. B) K cont 2.15 µm image. C) Continuum-subtracted pure H 2 emission image. D) Map of regions with IRAC colors consistent with UV excited H 2. angular distance of Cluster PL03 contains a group of H 2 knots ( ) that has the morphology of an outflow (Figure 5-17). This flow appears to originate from an embedded protostar northwest of the center of the cluster. In the center of cluster PL07 (Figure 5-18) there are two linear chains of H 2 knots that appear to originate from an embedded Class I/0 protostar. These may represent two separate flows that can occur in the case of a binary, or these knots may be tracing the edges of an outflow cavity. 79

80 Total H 2 flux (10-4 Jy) # of protostars Figure Summed H µm line emission vs. number of protostars within 1 pc (2.14 ) of the cluster centers. Table 5-1 lists the proposed driving sources of the outflows. The driving sources appear to be predominantly embedded protostars. While YSOs through Class II are known to power outflows, studies suggest outflows are strongest during the protostar stage. This is consistent with our identified outflow sources and also the correlation between the H 2 flux and the number of protostars in a region. Table 5-1. Outflow driving sources. ID R.A. Dec. Class Knots s I/ s I/ s I/0 037,038 s ,041,042 s

81 H 2 flux (10-4 Jy) R (I/0:II) Figure Summed H µm line emission vs. Class I/0 to Class II ratio within 1 pc (2.14 ) of the cluster centers. 5.5 Summary We catalog and investigate the distribution of shocked H 2 emission throughout the cloud. The distribution of shocked H 2 emission is not uniform and that the emission appears to be more prominent in younger regions. We identify the driving sources for many of the outflows traced by H 2 emission and find the driving sources are predominantly young Class I/0 protostars. We find strong correlations between the total measured H 2 line emission, number of protostars, and ratio of protostars to Class II objects. The relationship between the total emission and the protostar to Class II ratio suggests younger clusters have more outflow activity, and that outflow activity in the RMC decreases with age. We compare this to the relationship between age and extinction from Ybarra et al. (2013) and suggest outflow play a significant role in the gas removal within the clusters and subsequently affecting cluster and molecular cloud evolution. 81

82 H 2 flux (10-4 Jy) Total Luminosity of Protostars (L sun ) Figure Summed H µm line emission vs. total luminosity of protostars within 1 pc (2.14 ) of the cluster centers. Figure NIR H 2 and Spitzer images of an outflow in PL01. A) H µm image. The red cross indicates the proposed driving source. B) IRAC 4.5 µm image. C) MIPS 24 µm image. 82

83 Figure NIR H 2 and Spitzer images of outflow HH 871 ( ) in PL01. A) H µm image. The red cross indicates the proposed driving source. B) IRAC 4.5 µm image. C) MIPS 24 µm image. Figure NIR H 2 and Spitzer images of outflows in REFL08. A) H µm image. The red crosses indicate proposed driving sources. B) IRAC 4.5 µm image. C) MIPS 24 µm image. 83

84 Figure NIR H 2 and Spitzer images of an outflow in PL03. A) H µm image. B) IRAC 4.5 µm image. C) MIPS 24 µm image. Figure NIR H 2 and Spitzer images of the center of cluster PL07. A) H 2 (2.12 µm) image. Red contours reveal the pure H 2 emission. The red cross indicates the location of the proposed driving source, a deeply embedded protostar. B) IRAC 4.5 µm image. C) MIPS 24 µm image. 84