![]() ![]() |
![]() |
UFDC Home | myUFDC Home | Help |
|
|
||||||||||||||||||||||||||||||||||||||||||||
Full Text | |||||||||||||||||||||||||||||||||||||||||||||
NEAR INFRARED STUDY OF THE STAR-FORMING PROPERTIES OF THE ROSETTE COMPLEX By CARLOS G. ROMAN-ZUNIGA 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 2006 Copyright 2006 by Carlos G. Romin-Zifiiga This work is dedicated to the memory of: Richard J. Elston (1961-2004) and Leonel Hernandez (1972-2001) ACKNOWLEDGMENTS This is the second time in my life I have to write an acknowledgment section for a thesis. Just as then, I will make my best effort to to avoid omitting important names. However, if I do, let me say -as some sort of disclaimer- that it was not on purpose. My memory is very selective and it tends to retain too much movie trivia, old jingles and bad jokes, while randomly erasing names, telephone numbers and birthdays. But seriously: I want to acknowledge in a general way the Department of Astronomy at the Uni- versity of Florida accepting me as a student and for giving me a Teaching Assistantship during my first two and a half years in graduate school. Also, for giving me an office to study, access to a computer, free photocopying services and a marvelous work environ- ment. I want to acknowledge CONACYT-Mexico for a fellowship that sponsored a major fraction my doctoral studies at the University of Florida. During 4 years CONACYT gave me a living stipend, paid my tuition in full and paid for a significant part of my health insurance costs. My most sincere gratitude to this excellent program. It is also important to mention that FLAMINGOS, the instrument we used to collect most of our data was designed and constructed by the IR instrumentation group (PI: R. Elston) at the University of Florida, Department of Astronomy, with support from NSF grant AST97-31180 and Kitt Peak National Observatory. Next, I want to thank first my supervisor, Dr. Elizabeth Lada. Her efforts to direct my graduate career have been enormous to say the least, and I feel that it is only fair to say that without her, none of this would have been possible. When I asked her for a summer job at the end of my first year at UF, she told me about this 'nice little project on the Rosette Molecular Cloud, and how she would be happy to have me in charge of it. Well, I took the summer project, the project grew to a thesis, and I fell in love with it completely. Six years later, I am still working on it, pondering about its many consequences, and hoping to keep working on it for a while. You see, I already traveled to the Rosette in my dreams, and I got to know the place well. Yes, there were many challenges along the way, both technical and personal, but I cannot but admire Elizabeth for always being so enthusiast -and for being patient with my own flashes of over- enthusiasm-, for never allowing me to give up no matter how discouraging were the problems, and mostly for being there, sometimes as a supervisor, sometimes as some sort of basketball couch, sometimes as a psychologist, but most of the time as a friend. On top of that, Elizabeth generously used part of her grant money to take care of my salary and tuition requirements during my last year and my first two summers. Dr. Richard Elston constructed the instrument FLAMINGOS it with the skills and patience of a clockmaker, and then he made sure we cared about it the same way he did. I remember those first two years of the survey, where the goals were still confusing and foggy, our pipelines too buggy and the piles of data overwhelming: Richard always had a simple and clear way to solve any problem. Then he got ill and had to leave us, but I am glad I had enough time to learn from him a very important lesson, one about keeping up the courage intact even in the most difficult of circumstances. On another comer of the FLAMINGOS project is the effort and amazing stamina of Dr. Nick Raines, who not only knows every single cable and bolt of the instrument, but also every inch of the telescope facilities and every fine dinning corner of Tucson. I want to thank Dr. Jonathan Williams, from the Institute for Astronomy at the University of Hawaii, and former professor of Astronomy at UF. He took the necessary time to direct one entire chapter of this thesis, showing me the world of radioastronomy all the way to the big leagues at FCRAO and IRAM. And because he is an expert on the Rosette, there were many things I learned straight from his papers. I want to thank the faculty of the Astronomy Department for their patience, their help, their classes, their advice and their restless effort to integrate what is on its way to be one of the best Astronomy programs in the world. I want to mention Dr. Stanley Dermott, chairman of the department, who gave me trust and support even in difficult academic moments, Dr. Francisco Reyes for being such a great supervisor as coordinator of the Astronomy Laboratory, and a great example to follow; Dr. Rafael Guzman for being always a supportive friend and informal advisor, and Dr. Ata Sarajedini for always being attentive to my career as Graduate Coordinator. To the secretary staff of the Astronomy Department I wish I could dedicate a chapter of my thesis. There are no words to describe their help, specially with all the extra paperwork that my condition of foreign student implies. I want to specially thank Catherine Cassidy for dedicating so much time to remind me of academic deadlines, always with that extra wit and animosity. Debra and Deborah for all that help with grants, assistanships and travel, Glenda for helping me with my employee documentation, and also Tracey and Ann, who are in different departments now, but always were beautiful and patient and showed me how to fight the bureocratic monsters. Joanna Levine was my officemate, my project colleague, and a wonderful friend. Joanna is the kind of person that you can talk to about almost anything, and believe me, in 5 years of sharing an office with someone you get to cover a lot of conversation material. Now I really hope time will keep us being friends and collaborators, even if she becomes the first one in the world to combine astronomy and dance into a single, indivisible discipline and become insanely famous. Thank you to Dr. August Muench, also ex-officemate, for his brilliant lessons on computing, scientific passion and insomnia. Many ideas in this thesis became clear after commenting them with him. Other "ideas", for the same reason, went safely to the garbage pail before being incorrectly stated. Now we will be officemates again, which is a good opportunity to learn to follow very high standards like he always does. I also want to thank Dr. Charles Lada, from the Harvard Smithsonian Center for Astrophysics for many useful comments and encouraging words about this project, some of them crucial for its completion. And nowadays for giving me an opportunity to work with him. Astronomy does not get much better than this. I want to thank Eric McKenzie for his patience coping with the many versions of the pipeline and to enjoy observing and reducing star formation data just as he enjoys observing and reducing those long integration plates of galaxy clusters. And thanks for his anecdotes about walking across the United States, reading a thousand books and the nutritional virtues of peanut butter. I want to thank Dr. Anthony Gonzalez (nowadays a professor at UF), Dr. Matthew Horrobin, Dr. Andrea Stolte and Dr. Aaron Steinhauer, all former postdoctoral fellows, for taking our graduate student mess and helping to convert it into a refined scientific effort. At the mountain, they mastered the many twiggles of FLAMINGOS and the KPNO telescopes. Back at the office, they were always available for questions and comments, and I cannot remember one single time when the answers were not given to us with a big side of smiling: Anthony always was inquisitive enough to discover obscure bugs in the pipelines, but also patient enough to wait for us to correct them. With Matthew I learned tons too, and shared with him the experience of a data quality assessment trip to Boston, complete with bar hopping and all. From Aaron I will always admire his neat style and enthusiasm, everything to him was an opportunity to learn, which is fantastic. Oh, and his Simpsons collection kept us awake many times at the telescope! Finally, Andrea was to put it in simple words, the person who saved my project. She was patient enough to understand the problems we were dealing with and then even more patient to fix them. She even accepted my crazy idea of making a catalog software application in Supermongo and then refined the effort!. That is something to admire. Really. I want to thank Noah Rashkind and Chris Foltz for being the guinea pig users of LongLegs, with all that amazing vitality and enthusiasm. They were also great trip pals in the Boston trip, which was a lot of fun. Many thanks to Bruno Ferreira and Jorge Gallego. Bruno for instance, made possible a crucial gear of my analysis, digesting successfully a cryptic little paper I reccomend him to read. But that is nothing compared with the many many great moments we shared as part of the Star Formation Crib, and many friendly gatherings. Jorge, the other creature lurking in the cave 319 always kept me afloat with his cheerful style, along with many discussions about movies, spanish rock bands and the meaning of life. Nowadays these guys are so important in the Gainesville community (specially Maestro Yogi Bruno) that I wonder how the city will cope with after they leave. Thanks also for pizza, for the Indian Air sessions, for FILMINGOS and for many other adventures. I want to thank the many UF graduate students that I met along the years. Names like Lauren, Elisha, Christos, Sue, Barbara, Jim, James, Scott, Rob and TJ might sound ancient to some people now, but they were the same as the rest of us not a long ago. In a second layer are Kelly, David, Debbie, Doug, Pimol, Paty, Veera, Derrick, Bill, Catherine and Manuel among those who left. Craig, Gator extraordinaire is a separate case. And then all of those who keep the joy going on: Eric, Ana, Ileana, Naibi, Cynthia (thanks for the hospitality!), Suvrath, Margaret (the coffee hour angel), Michelle, David, Aaron, Mike, Ashley, Valerie, Paola, Miguel, Sung, Justin, Lauren, Curtis, Scott, Audra, Alison, Justin, Leah, Julian and Andrew. To the Mexicans in Gainesville Student Association, for being my family all these years. Without their support, adapting to a whole new country would have been near to impossible. I have a special mention for Julio Castro for being my lunch pal, joke sidekick, movie critic partner and friend all these years. I also want to thank Eugenio y Milena (and her parents Sofia and Fernando), Velia y Luis, Diego y Erica, Jorge, Rocio, Juan, Hussein, Alicia, Maria Jose y Leo, Horacio y Maru, Arturo y Rosa Isela, Antonio y Roxanna, Sebastian y Paula, Nicasio y Miriam. I owe an apology for not putting the names of their children -I am already several pages above expected-. Too many moments we shared: meetings, barbecues, carnivals, September 15th parties, birthdays, you name it, but the best thing are the memories. To my family, for ALWAYS believing in me and being close to me despite distance and the time. My parents, Hector and Rosario have always been my greatest motivation, and everything I am now I owe to them and their efforts and many sacrifices. To my brother and sister, Esteban and Daniela, I want to say that we are all part of the same love, and I am proud of you every moment. I love you all so much, and I need to be with you again so much... I only pray for that moment to come soon. To beloved Sheikha Amina Teslima and all the members of the Al-Jerrahi commu- nity for keeping the very essential light of my heart always lit on. And for making time and distance invisible. Alhamdulilah. And Finally, the person who I decided to share the road of life with. Fabiola, can you believe that I am writing this for the second time around? And how many words do I need now to explain what I feel, if there are infinite reasons for being thankful and happy? No, I cannot start nor finish because you are my beggining and my end. I love you with every part of me. TABLE OF CONTENTS page ACKNOWLEDGMENTS ............................ iv LIST OF TABLES ...... ................. ........ xiii LIST OF FIGURES ........... ...... ........ ........ xiv KEY TO ABBREVIATIONS ............................. xviii KEY TO SYMBOLS .. ............................ xix ABSTRACT. ..................... ................. xx CHAPTER 1 INTRODUCTION ................... ......... 1 1.1 A Global Picture of Star Formation .......... ....... ... 1 1.2 Motivations for the Study of the Rosette Complex ............ 4 2 THE ROSETTE COMPLEX IN MONOCEROS ........ ......... 7 2.1 Historical Perspective ........................... 7 2.2 The Rosette Nebula and the Young Cluster NGC 2244 ......... 9 2.2.1 The Rosette Nebula ......... ............... 9 2.2.2 NGC 2244 ............................. 10 2.2.3 Spectroscopic Studies .......... ....... ..... 12 2.2.4 Near Infrared Studies ......... ............. 13 2.2.5 X-ray Studies ................... ...... 14 2.3 The Rosette Molecular Cloud: Structure ....... ......... 14 2.3.1 CO studies ............................. 14 2.3.2 Interaction with the Rosette nebula ......... ....... 18 2.4 The Rosette Molecular Cloud: Embedded Populations . ... 23 2.4.1 Dominance of Cluster Formation in the Rosette Complex . 23 2.4.2 The Hypothesis of Sequential Star Formation . ... 25 3 A NEAR-IR SURVEY OF THE ROSETTE COMPLEX: OBSERVATIONS 27 3.1 The FLAMINGOS GMC Survey ....... . .... 27 3.2 Data Reduction ........... . . . ......29 3.2.1 The Data Reduction Pipeline: LongLegs . . ... 29 3.2.2 The Photometry and Astrometry Pipeline: PinkPack ....... .31 3.3 Completeness of Sample .................. ..... 32 3.4 Positional Correction of Photometry . . ...... 35 3.5 Quality and Uniformity of the Survey .... . . . 40 3.6 Construction of Final Catalog ...... . . . .... 44 3.6.1 Intrinsic quality: 2MASS Addendum . . ... 44 3.6.2 Survey Area Merging ........ . . .... 44 3.7 Intrinsic Detection Constraints .............. . 45 4 NEAR-IR SURVEY: ANALYSIS AND RESULTS . . . 48 4.1 Introduction .............. . . ..... 48 4.2 Analysis .......................... .... ........ 50 4.2.1 The Nearest Neighbor Method . . 50 4.2.2 Detection of Embedded Populations . . . 52 4.2.3 Infrared Excess Stars .................. ..... .. 53 4.2.4 Magnitude Depth Restriction for IRX stars . . .... 55 4.2.5 Nearest Neighbor Analysis for Infrared Excess Stars ....... 58 4.2.6 Identification of Clusters .............. ... .. 60 4.2.7 Properties of Clusters ................ .... .. 63 4.3 The Fraction of Stars in Clusters . . . .... 69 4.3.1 Distribution of Sources with Respect to the Rosette Nebula 75 4.3.2 A Case for a Distributed Population? . . . 79 4.4 Discussion and Future Work ................ .... .. 82 5 OBSERVATIONS OF CLUSTER DENSE GAS ENVELOPES ......... 87 5.1 Introduction ............... . . . 87 5.2 Observations and Data Reduction ..... . .... 89 5.3 Analysis and Results.. . . . ..103 5.3.1 Presentation of the Data ....... . . ... 103 5.3.2 Local Extinction ......... . . . ... 104 5.3.3 Calculation of Physical Parameters . . . .. 105 5.3.4 Tracer Abundances . . . ..106 5.3.5 Gas Mass ....... ..................... 108 5.3.6 Clump Sizes .................. ........ 109 5.3.7 Line W idths ................... . .. 110 5.3.8 Clump Masses .......... . . ... 110 5.3.9 Gas Dynamics .......... . . ... 114 5.4 The Embedded Stellar Population .... . . .... 116 5.4.1 Star Forming Efficiencies . . . 116 5.4.2 The Gas Stars Connection ..... . . .. 117 5.5 Chemical Differentiation in Cluster Envelopes . . ... 119 5.6 Summary and Discussion .................. ...... ..121 6 GLOBAL ASPECTS .......... ....................... 125 6.1 A Near Infrared Extinction Map for the RMC . . ... 125 6.1.1 Motivations ..... . . . . . 125 6.1.2 Dust Extinction from Near-Infrared Colors: NICE and NICER 126 6.1.3 The Extinction Map . ....... . . 130 6.2 Individual Extinction Cores .............. ..... 134 6.2.1 Identification and Estimation of Properties . . .... 134 6.2.2 Core Sizes ............ . . .... 143 6.2.3 Core M asses ................... . .. 145 6.2.4 The Embedded Cluster Mass Function . . ... 148 6.2.5 Star Formation Efficencies .... . . .. 153 6.3 Summary and Discussion .................. ...... ..157 7 CONCLUSIONS AND FUTURE WORK ..... . . .. 161 7.1 Distribution of Young Stellar Populations in the Rosette Complex .. 161 7.2 The Local Environments of Young Clusters . . .. 162 7.3 Extinction in the Rosette Complex and Global Results . ... 163 7.4 Future Work .................. ............. 164 APPENDIX A NEAR-IR SURVEY. DETAIL OF OBSERVATIONS . . ... 167 B MILLIMETER SURVEY. DETAIL OF OBSERVATIONS . . ... 173 C MILLIMETER SURVEY. DETAIL MULTIPANEL MAPS ..... ... .. 174 REFERENCES ................... . . .... 177 BIOGRAPHICAL SKETCH .......................183 LIST OF TABLES Table 2-1 4-1 5-1 5-2 5-3 5-4 6-1 6-2 A-1 A-1 A-1 A-1 A-2 B-1 page . . 64 . . 64 . . 102 . . 106 . . 116 . . 139 . . 157 . . 168 . . 169 . . 170 . . 171 . . 172 . . 173 Distance Estimates to the Rosette (NGC 2244) . ... Young Clusters Rosette Complex . .......... Relevant Properties of Rosette Clusters ............ Clump properties for Rosette Clusters . . Molecular Line Parameters . ............. Star Formation Efficiencies (13CO(2-1)) . ...... Extinction Cores in the Rosette Complex ........... Star Formation Efficiencies (Av cores) . ....... Summary of near-IR observations. FLAMINGOS KPNO-2.1m Summary of near-IR observations. FLAMINGOS KPNO-2.1m Summary of near-IR observations. FLAMINGOS KPNO-2.1m Summary of near-IR observations. FLAMINGOS KPNO-2.1m Mean Photometric Scatter by Field . ......... IRAM Observations: Area Coverage by Tracer . .. LIST OF FIGURES Figure page 2-1 A photograph of the Rosette Nebula .................. .... 7 2-2 Location of the Rosette Cloud in the Monoceros Complex . . 8 2-3 A Ha vs. V-I diagram for NGC 2244 .................. ..13 2-4 A CO map of the Rosette Molecular Cloud ... . . ..... 16 2-5 Molecular Clumps in the Rosette Cloud .................. ..17 2-6 IRAS 12pm and 1400 Mhz map overlay ..... . . 20 2-7 A 0.5-2 keV Chandra image of the Rosette Complex . . ... 22 2-8 Location of the Phelps & Lada clusters ................. ..24 3-1 UF/NOAO Rosette Complex Survey Map ................. ..30 3-2 Completeness Limits by Filter .................. ....... 33 3-3 Completeness by type of field ..... ........ ...... 34 3-4 Photometric Quality Areas .................. .. ...... 37 3-5 Photometric Correction: Radial (J,H) .................. ..38 3-6 Photometric Correction: Radial (K) ................... . 39 3-7 Photometric Correction: Colors .................. ....... 40 3-8 Photometric Correction: Color and Magnitude Diagrams . ... 41 3-9 Photometric Correction: Photometric Scatter ..... . . . 42 3-10 Distribution of Photometric Uncertainties by Filer . . .... 43 3-11 Detectability of an Embedded Populations ................. ..46 4-1 Areas of the Color-Color Diagram .................. 54 4-2 Contour level J- H vs. H- K Diagram for All stars in the Survey . 56 4-3 Contour level J- H vs. H- K Diagrams Divided by Brightness ...... 57 4-4 Nearest Neighbor Distributions for Bright IRX Stars . . .. 59 4-5 Location of IRX Stars with Brightness K < 15.75 mag ..... 4-6 Identification of Clusters in the Rosette Complex .............. 4-7 Distribution of Cluster Core and Total Radii .. .............. 4-8 Analysis Plots for Cluster PL01 ........................ Analysis Plots Analysis Plots Analysis Plots Analysis Plots Analysis Plots Analysis Plots Analysis Plots Analysis Plots Analysis Plots Analysis Plots for Cluster PL02 ........................ for Cluster PL03 ........................ for Cluster PL04 ........................ for Cluster PL05 ...................... for Cluster PL06 ........................ for Cluster PL07 ........................ for Cluster RLE08 .. ................... for Cluster RLE09 .. ................... for Cluster RLE10 ....................... for Cluster NGC 2237 ..................... 4-19 Analysis Plots for Cluster NGC 2244 ..................... 4-20 Distribution of IRX stars as a function of distance to the Rosette Nebula 4-21 Cumulative Counts of IRX sources in Field 09 ............... 4-22 Images of Distributed Formation in Field 09 of the Survey .......... 5-1 Molecular Emission Maps: Cluster PL01 ................... 5-2 Molecular Emission Maps: Cluster PL02 ................... 5-3 Molecular Emission Maps: Cluster PL03 ................... 5-4 Molecular Emission Maps: Cluster PL04 .. .............. 5-5 Molecular Emission Maps: Cluster PL05 ................... 5-6 Molecular Emission Maps: Cluster PL06 ................... 5-7 Molecular Emission Maps: Cluster PL07 ................... 5-8 Molecular Emission Maps: Cluster RLE08A ................. 5-9 Extinction in 13CO(2-1) Map Areas (1) .................... 5-10 Extinction in 13CO(2-1)Map Areas (2) . 5-11 Abundance Ratios by Tracer . . 5-12 Distribution of Clump Sizes by Tracer 5-13 Distribution of Line Widths by Tracer 5-14 Distribution of Virial Mass by Tracer . 5-15 Distribution of LTE Mass by Tracer ... 5-16 Distribution of Virial to LTE Mass ratios 5-17 Comparison to Curves of Binding Pressur 5-18 Distribution of Velocity Gradients by Trac 5-19 Distribution of Star Formation Efficiencie 5-20 Correlation Between Cluster Sizes and Er 5-21 Overlap of HCO+(1-0) and CS(2-1) Emis 6-1 Near-Infrared Extinction Map of the Rose 6-2 13CO emission map of the Rosette Compl 6-3 Contour Extinction Map with Cluster Pos 6-4 Correlation between 13CO and Av . 6-5 Distribution of individual Av values betw 6-6 Extinction maps for Individual Cores (1) 6-7 Extinction maps for Individual Cores (2) 6-8 Extinction maps for Individual Cores (3) 6-9 Extinction maps for Individual Cores (4) 6-10 Extinction Core Profiles (1) . . 6-11 Extinction Core Profiles (2) . . 6-12 Extinction Core Profiles (3) . . 6-13 Extinction Core Profiles (4) . . 6-14 Distribution of Core Radii . . 6-15 Comparison of Cluster and Core Radii . . . . 100 .. . 107 . . 109 ....... . ... 110 ..... . . .. 111 ........ . . .. 112 ...... . . . 113 e ..... . . . 114 cer . . . 115 s ..... . . . 117 mission Offsets . . 118 sion for Cluster PL03 . 119 :tte Complex . . 128 ex . . . 129 itions .. . . 131 . . .. . . 132 een0 and30mag . ... 133 ..... . . . 136 ..... . . . 137 ..... . . . 138 ..... . . . 139 . .. . 140 . . . . . 14 1 . . . . . 142 . . . . . 143 . . . . . 144 ....... . . 145 -16 Cluster Radii and IRXF vs. Extinction . . -17 Mean Extinction vs. Core Radii . .... -18 Distribution of Extinction Core Masses . . -19 Distribution of Core Mass Compared to Clusters . 20 The Embedded Cluster Mass Distribution Function 21 Rosette Clusters in the ECMDF . .... 22 Extinction Cores SFE vs. Cluster Radii . . 23 SFE as a Function of Distance to NGC 2244 . 24 Schematic Map of the Rosette Complex . . . . . 146 . . . 147 . . . 148 . . . 149 . . . . 150 . . . 15 1 . . . 154 . . . 155 . . . .. 156 xvii KEY TO ABBREVIATIONS 2MASS ECMDF FCRAO FITS FLAMINGOS trometer FOV GMCs HBL IMF IRAF IRX ISM NNM NOAO OBAs PMS PSF ROSAT SFE SQIID SSF ZPT Two Micron All Sky Survey Embedded Cluster Mass Distribution Function Five College Radio Astronomy Observatory Flexible Image Transport System Florida Multi-object Imaging Near-IR Grism Observational Spec- Field of View Giant Molecular Clouds Hydrogen Burning Limit Initial Mass Function Image Reduction and Analysis Facility Infrared Excess Interstellar Medium Nearest Neighbor Method National Optical Astronomy Observatory OB Associations Pre-main sequence star Point Spread Function fitting method Rosetta X-Ray Satellite Star Formation Efficiency Simultaneous Quad Infrared Imaging Device Sequential Star Formation Photometric calibration zero point xviii KEY TO SYMBOLS avir Ratio of Virial to LTE clump mass Av Visual Extinction E(B-V) Visual Color Excess Ha Ha emission HII Ionized Hydrogen J, H, K Near Infrared Bands at 1.2, 1.6 and 2.2 pm Mcc Mass of star forming extinction core Mclus Cluster Mass Memb Mass of Embedded Stars MLTE Clump LTE mass Msc Mass of starless extinction core M. Units of Solar Mass Mvir Clump virial mass Nemb Number of Embedded Stars Rcc Radius of Extinction Core Rcore Cluster Core Radius Req Cluster Equivalent Radius Rv Visual Extinction to Excess Ratio Abstract of Dissertation Presented to the Graduate School of the University of Flr 'iida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy NEAR INFRARED STUDY OF THE STAR-FORMING PROPERTIES OF THE ROSETTE COMPLEX By Carlos G. Rom6n-Ztfiiga May 2006 Chair: Elizabeth A. Lada Major Department: Astronomy The Rosette Complex is one of the most important astrophysical laboratories for the study of star formation. In this region we can study the interaction of an expanding HII region -impulsed by the stellar winds from the large OB association NGC 2244- with a large remnant molecular cloud, which is known to host seven embedded clusters. As part of a large observational program to study the nature of young stellar populations in giant molecular clouds, we made a complete near-infrared imaging survey of the Rosette Complex using the detector FLAMINGOS. This survey is deep enough to detect stars near the brown dwarf limit, improving considerably over available databases. However, given the location of the Rosette Complex at a large distance from the Sun and at a latitude close to the galactic disk, the contamination of the survey data by field populations is high. In order to facilitate the detection of young populations, we combined a selection of cloud members by means of their infrared excess emission with a technique to detect star clusters using distances to nearest neighbors. This way we were able to confirm the seven clusters previously identified, and to discover four new clusters. For every stellar cluster we determined for the first time their approximate extensions and number of members. We found that the fraction of stars in clusters in the Rosette Complex is close to 87%, which is similar to other clouds like Orion. However, the formation of clusters in the Rosette seems to be heavily influenced by the interaction with the expanding nebula, as evidenced by the fact that the core of the molecular cloud, coincident with the shock front of the expanding nebula contains 50% of the total cluster population. The clusters in the core are also more extended and more populated. Our study was complemented with a high resolution millimeter wavelength radio survey of the dense gas emission around the 8 most prominent clusters in the sample. We confirmed that all of the clusters observed are still embedded in what appear to be very compact parental clump remnants, but in many cases these gaseous envelopes are possibly becoming gravitationally unbound, due to the partial emergence of the young cluster stars. The dense gas maps show features characteristic of the interaction of clusters their local environment, particularly significant offsets of tracer emission peaks, possibly due to chemical differentiation effects. Our near-infrared observations also allowed us to construct an extinction map for the fields observed. The map shows an good agreement with 13CO emission radio maps, and allowed us to identify the main molecular cores in the complex. Using the mass of stars in the clusters and the mass of the emission cores we calculated star formation efficiencies, which resulted to be significantly larger at the central core of the cloud. Also, extinction appears to be inversely proportional to the size of the clusters, but directly proportional to the fraction of IRX sources, which is suggestive of evolutive effects and a rapid dispersion of the gas after clusters are formed. The cluster emergence time scales could be similar and even shorter than the T Tauri phase of the stars. CHAPTER 1 INTRODUCTION 1.1 A Global Picture of Star Formation Star Formation is one of the main puzzles in present day Astrophysics. Along the years, it has been possible to construct a relatively detailed picture of the physics involved in the formation of individual stars (Shu et al., 1987), but the problem of how to extrapolate that picture to explain the formation of large groups of stars is more complicated. For example, a complete model of Star Formation should formulate correctly the necessary rates and efficencies of formation required to populate a galaxy like ours, but also those of more or less active galaxies. It would also need to be a sort of general scheme that could explain the formation of stellar populations with similar characteristics (for example their mass distributions) in completely independent environments. It would also need to unify the physics relevant to the prime material (interstellar clouds) and the final product (the stars). Progress has been made, but nowadays Star Formation, as a global theory, still has many untied knots. Stars form in molecular clouds, composed mainly of molecular hydrogen, which are the densest (n> 103 cm3) and coldest (T~ 10 K) components of the Interstellar Medium. A significant fraction of this molecular material exists in the form of large complexes called Giant Molecular Clouds (GMCs), with masses of 104_106 M. and typical sizes of 10-100 pc.) GMCs are usually surrounded by extended envelopes of atomic Hydrogen with typical masses of 106 Mo. Practically all known GMCs with distances of less than 3 kpc have been forming stars during the last 10 million years and we have direct evidence for this assumption: * First, many HII regions located at the edges of molecular clouds are being expanded by the winds of young, massive stars. By young and massive we understand O and B spectral types, with fast nuclear burning rates that result in lifetimes much shorter than the age of the galaxy. Also, these objects are usually located in groups, called OB Associations (OBAs). These associations usually have spatial densities below the threshold for Galactic tidal disruption (Ambartsumian, 1947). This fact provides further evidence -in this case dynamical- that star formation is recent. * Second, with the aid of infrared and millimeter-wave detectors developed in the last three decades, we are able today to see through the optically thick clouds where stars form -an impossible task for optical telescopes. This way we have been able to observe stars and even proto-stars while they are still embedded in their parental clouds. These embedded stellar populations are even younger than OBAs, with typical ages of 1 Myr or less. Also, from observations of embedded clusters in nearby GMCs (d < 2 kpc), there is observational evidence that the majority of the stars in GMCs are formed in clusters (Lada et al., 1991b; Carpenter, 2000). Moreover, rich clusters (100 members or more) clearly dominate over small groups, as they contain more than 80% of the observed embedded stellar population (Porras et al., 2003; Lada & Lada, 2003). Unfortunately, the dominance of large embedded clusters in available catalogs might be slightly biased by an incompleteness at the small cluster regimes. Among the reasons for this are: a) systematic surveys of molecular clouds aiming for the detection of an embedded population are rather scarce; b) searches for embedded clusters, if any, are usually limited to those zones with signposts of formation (e.g. the presence of very luminous infrared sources); c) surveys are mostly based on monotonic wavelength counts, with poor corrections for background contamination. This way available surveys have led to the spotting of only the richest clusters. It is only in a few cases when there is a search -either additional or separate- for low density groups and distributed embedded populations1 The main reason is that small groups are logically much more difficult to detect, especially if there are mostly composed of low mass stars (which are fainter), with large spatial distributions and projected against a high background of reddened sources. Embedded low mass stars are clearly harder to observe because they are intrinsically faint. Even so, spectroscopic studies reveal that OB associations have a much larger number of low mass than massive stars, in proportions that are coincident with the distribution or Initial Mass Function (IMF) of stars in the field. However, the spatial density of the low mass component is rarely above that of faint field stars and thus the exact fraction of low mass stars in young populations is difficult to calculate from stellar density counts alone (Lada & Kylafis, 1999). Fortunately, if stellar associations are young enough (3 Myr or less) then low mass stars have circumstellar material that causes them to have an excess of infrared emission, and this makes them distinguishable from field stars. These kinds of objects, known as Line-Emission or T Tauri stars, are thus a good tracer of the low mass component of young stellar populations, but their observation is subject to uncertainties related to the eventual weakness of line emission, the quality of the photometry required to observe the excess, and the eventual contamination from field stars (see Chapter 4). The process of formation of low mass stars, which leads to their coexistence with massive stars is also poorly understood. Some existent models of cluster formation are able to account for the observed spatial distributions of stars in clusters, but fail to match the observed physical conditions of dense cores where clusters form (Bonnell et al., 1997), or do not fit the number distribution of the observed IMF (Mouschovias, 1 The term distributed refers to stars for which their formation process cannot be di- rectly associated to a group or a cluster. For example, it could refer to stars formed in isolation or stars originally formed in a group but dispersed to the point that the group is no longer distinguishable (e.g. Li et al., 2002; G6mez et al., 1993; Carpenter, 2000,.) 1991). Also, dense molecular cores are expected to experience significant fragmenta- tion prior to condensation of proto-stars, a process that is not completely understood either. The current hypothesis is that marginally stable cores experience cooling via dissi- pation of magnetohydrodynamic turbulence in highly extinguished cores (Myers, 1998), which leads to the fragmentation of the core into a matrix of molecular kernels. The kernels will end up forming stars of different masses via competitive accretion, with the most massive stars either forming closer to the center of the core where accretion rates are higher or from initially larger kernels (Bonnell et al., 2001). In this scheme low mass stars will form preferentially in the outer parts of cores, resulting in a primordial mass segregation. The puzzle of the global properties of star formation in GMCs, with a complete understanding of the mechanisms that lead to the dominion of large groups and the for- mation and role of low mass stars, can only be solved by studying GMCs in a systematic approach. This means that entire GMCS should be observed one at a time, with instru- ments powerful enough to detect low mass stars. Also we need to cover as much area of the cloud as possible, independently of the presence of rich clusters signposts, so that low density populations if any, can also be taken into account. In such surveys, we could obtain unbiased statistics of the embedded stellar populations, and it would be easier to form a global picture of the stellar birth phenomenon. 1.2 Motivations for the Study of the Rosette Complex The Rosette Complex is a giant star forming region where a very large OBA, NGC 2244 has formed. This OBA is evacuating the center of its original cloud by means of a powerful ionization front created by the winds of its members. At the southeast edge of this region, there is a large molecular cloud, where several embedded clusters have been detected. These characteristics make the Rosette an excellent laboratory to investigate the properties of very young stellar populations. The region has been stud- ied extensively in terms of its main features, the Rosette Nebula, its OB association NGC 2244 and the physical properties of the molecular cloud. However, the charac- teristics of the embedded populations have been studied only to a very superficial level and it is unknown if there are other clusters, if they share the cloud with a low density population and more important, what is their relation to the prominent NGC 2244. The molecular cloud and the nebula appear to be in clear interaction, and a basic question is how the formation of the new clusters is related to this interaction. One approach to this problem, for example, would be to study the properties of clusters as a function of distance to the nebula and see if any significant differences arise, which would be proof of the influence of the OB association in the new episode of formation occurring in the cloud. One of the main goals of this thesis is to determine, to the best possible level, the total number of young stars in the Rosette Complex, as well as their distribution, and relative properties. The core of the thesis is a new near infrared survey of the region made with the instrument FLAMINGOS, developed at the University of Florida, which can detect stars in the Rosette down to the low mass regimes -a task that has not been accomplished yet. After separating from the catalog the best candidates for young stars, we apply a technique based on the calculation of local surface densities in order to determine the location and extent of the known clusters. The selection of stars by their infrared excess -determined with the use of near-infrared colors, improves the cluster detection techniques used in single band photometry studies for other clouds. In the first chapter of this thesis we make a review of the previous studies of the Rosette Complex region. The review follows a roughly historical line, and ends with the few embedded population studies done previous to this work, motivating the necessity for our new observations. The second chapter of this thesis is dedicated to the description of our Rosette Complex near-infrared survey, detailing our observations, data reduction methods, and data quality assessments. The third chapter describes the analysis applied to the photometric catalogs resultant from the survey. From this analysis we attempt to improve the discussion about the distribution of star formation in the RMC. The fourth chapter describes a complementary millimetric radio wave study of eight RMC clusters, which has the goal of discussing the interaction between embedded star clusters and the remnants of their parental cores. The fifth chapter describes the use of near-infrared colors of stars to create a detailed extinction map of the Rosette Cloud, which allows us to compare some properties of the clusters with those of their forming cores. We also include a first approach to the calculation of the cluster masses, which allows us to study star forming efficiencies in the complex. Finally, we present a summary of the results of the thesis and a discussion on future work. CHAPTER 2 THE ROSETTE COMPLEX IN MONOCEROS 2.1 Historical Perspective The Rosette Complex (1=207.0, b=-2.1) is located at the anticenter of the galactic disk in the constellation of Monoceros. The region is very popular, partly because of the lIigg.rinfg beauty of its main feature: a very extended emission nebula which hosts a large central HII region, evacuated by the winds of a central OB association (see Figure 2.1). Figure 2-1: The Rosette Nebula. Credit: Canada-France-Hawaii Telescope / 2003 The complex is part of a much larger structure known as the Northern Monoceros Region. This region comprises the Mon OB 1 Cloud (host of NGC 2264 and the Cone 8 Nebula), the Monoceros Loop, and the Mon OB2 Cloud in which the Rosette is one of the most prominent features (see Figure 2-2). 13 L 11 L 6h 48rh 40m 32m 24m a (1950) 6h16m Figure 2-2: The Location of the Rosette Molecular Cloud in the Context of the Mono- ceros Complex region, from Perez (1991). The catalog name for the Rosette can be somewhat confusing because it is not unique: The nebula itself is usually cataloged as NGC 2237 or NGC 2246, (especially by amateur observers) although NGC 2237 originally referred to the brightest patch at its west side and NGC 2246 originally pointed to a bright zone at the eastern side. In addition, while the central cluster is usually known as NGC 2244, it has also been I I I NORTHERN MONOCEROS i i i lql I IT I I cataloged as NGC 2239. However, this designation historically referred to the brightest star in the region, 12 Monocerotis. The cluster was first noticed by Flamsteed in the late 17th century and later reported by William Herschel -who did not notice the nebulosity- and John Herschel, who discovered several of its most conspicuous features and reported them in his general catalog (Herschel, 1864, NGC 2239 = GC 1420). Other parts of the nebula (NGC 2237 and NGC 2246) were reported by Swift (1886) who cataloged the object as being "pretty bright [pB], very, very large [vvL] and diffuse [diff]." Afterwards, the region was formally known as the "Swift Nebula," until the name "Rosette" became more popular. The total extent of the Rosette was not determined until the first photographic plates were obtained by Barnard (1894). Two of the first applications of Rosette Nebula photographic data were made by Hubble (1922) in his study of diffuse nebulae associated with massive stars, and Minkowski (1949), who published a photographic plate study along with a first discussion on the expansion of the HII region by the central cluster O stars and the possible existence of Bok globules. He estimated the mass of the nebula to be 104 M.and suggested that it could be "surrounded and probably embedded in obscuring material," thus proposing the existence of the companion molecular cloud. 2.2 The Rosette Nebula and the Young Cluster NGC 2244 2.2.1 The Rosette Nebula A tabulation of the different methods used to determine the age of the Rosette Nebula was done by Ogura & Ishida (1981). These varied from studies of the properties of the central cavity (Kahn & Menon, 1961; Lasker, 1966) to evolutionary models of the HII region based on the luminosity of the stars (Hjellming, 1968). Other methods involve time scales of radiation pressure (Mathews, 1966, 1967), estimates of the formation time for dark globules in the nebula (Herbig, 1974), and the separation of [OIII] emission lines (Smith, 1973). The mean value of all these age estimates is approximately 3 1 x 106 yr. A series of studies by Celnik 1983; 1985; 1986 discussed the global physical characteristics of the Rosette Complex. The first two of these are dedicated to the nebula, while the third one is a model of the interaction with the molecular cloud. In the first of the articles, he presented a map of the Ha emission in the nebula region, and calculated a total integrated flux density of 5 x 10 11W-m2 within 60' from the center of the HII cavity. He suggested that the emission is contained in a more or less symmetric ring with a peak at 16' from the center. In the second paper, Celnik reported radio continuum observations (1410 and 4750 MHz) from which he was able to determine that the nebula is bound by ionization, forming a spherical shell with radius of 40 pc (about 85') and a total ionized matter mass of 2.3 x 104 M Using the H112a and Hel 12a recombination lines (4619 and 4621 Mhz) Celnik calculated a He+ abundance of 0.120.03 and a non-LTE electron temperature for the nebula of Te = 5800 700 K almost 1100 K above the LTE with no observable gradient with respect to the radial distance from the center. However, the investigation by Shipman & Clark (1994) revealed a good fit to a T oc r-,, a = 0.4 model for the temperature gradient in the nebula cavity which, interestingly, could not be adjusted to the observed IRAS emission. Instead, they found that this temperature gradient was better adjusted to a = 0.05 for r < 47' and a = 0.2 for 47' < r < 65'. 2.2.2 NGC 2244 The prominent OB association that is presumed responsible for the evacuation of the central part of the nebula has been the subject of many interesting studies over the years. The distance to this young cluster (and therefore to the entire complex) has been estimated many times with slightly different results. Table 2-1 is a compilation of these values, from which the most commonly used is 1600 to 1700 pc. Some of the first visual photometric studies on NGC 2244 were made by Johnson (1962), who estimated the mean color excess in the cluster to be E(B-V)= 0.46 for Table 2-1: Distance Estimates to the Rosette (NGC 2244) Author Value (pc) Method Johnson (1962) 1660 Photoelectric Photometry Ogura & Ishida (1981) 1420 Visual Photometry Perez et.al (1987) 1670 Visual Photometry Park & Sung (2002) 1660 Visual Photometry Hensberge et.al (2002) 1390 Spectroscopy Rv = Av/E(B V) = 3.0. This was confirmed by Turner (1976) and later Ogura & Ishida (1981), who suggested a value of R = 3.2 0.15. Ogura & Ishida (1981) also pro- posed an age of 41 Myr and a star formation efficiency of 22% for the cluster. Later, Marschall et al. (1982) completed a proper motion study of 287 stars in the NGC 2244 area. They confirmed membership for 113 objects, 52 of them from the list of Ogura & Ishida (1981). A study that combined photometry as well as spectroscopy was completed by Perez et al. (1987). They found that some members of NGC 2244 presented anomalous values of R, possibly suggesting the coexistence of main sequence stars with very young objects -likely T Tauri stars. This was confirmed with a uvbyp photometry study by Perez et al. (1989), in which 4 members presented evidence of being true pre-main sequence (PMS) objects. They also confirmed the age of NGC 2244 to be below 4 Myr but spread towards younger values, thus confirming a model of continuous formation. A study of great importance was performed by Park & Sung (2002). They obtained UBVI and Ho photometry for the cluster. They were able to determine membership for a total of 30 cluster sources and to extend the list of known PMS candidates to 21. They subsequently identified members coincident with ROSAT point sources catalogs and spectral types from Verschueren (1991) (see section 2.2.3). Six of the PMS candidates were confirmed as X-ray sources. In Figure 2-3, we show the Park & Sung (2002) relation between alpha emission and V-I color for NGC 2244. (the relationship actually would hold for any optical or infrared color). In this figure, PMS stars are clearly located above the main sequence. Later, using evolutionary models Park & Sung (2002) showed that most of the PMS stars and PMS candidates in their sample appear to have masses close to 1M. and an approximate mean age of 0.4 to 0.9 Myr. Because they also estimated the main sequence turn-off age of the cluster to be 1.9 Myr, they showed that the cluster has not stopped forming stars yet. Another important calculation in this article is the Initial Mass Function (IMF) of NGC 2244. They found it has a flat (F=-0.7) IMF slope in the range 0.5 < logm < 2.0. By comparing directly to the IMF model of Scalo and to the observed mass function of NGC 2264, they demonstrated that NGC 2244 is highly dominated by massive stars, thus confirming its status as a giant OB association. 2.2.3 Spectroscopic Studies The most complete spectroscopic study of NGC 2244 was done by Verschueren (1991), (see section 2.2.2) and it has been widely used in the literature. In particular, Park & Sung (2002) used data from this study to identify the spectral types of candidate T Tauri stars in NGC 2244. A low resolution, single slit investigation by Hensberge et al. (1998) of 2 members and 3 field stars in the region of NGC 2244 yielded evidence that these were chemically peculiar, possibly magnetic stars. Later, Hensberge et al. (2000) performed spectroscopic analysis of the binary member V578 Mon, which resulted in an estimated distance slightly lower than other photometric estimates (see Table 2-1). They also calculated the age of the system to be 2.30.2 Myr. Finally, Li et al. (2002) presented low resolution spectra for a sample of X-ray counterparts from the ROSAT PSPC survey (see also Gregorio-Hetem et al. (1998)). They were able to confirm that five sources had strong Hc emission. Two of the stars were -2 A Ha- A 0A -- &.._ 0 1 2 3 V-I Figure 2-3: A Ha vs. V-I diagram for NGC 2244 from Park & Sung (2002). The solid line represents a ZAMS relation while the dashed line is a selection limit. Filled triangles are PMS stars while open triangles are PMS candi- dates. Bright members are marked with dark filled circles. X symbols are X-ray sources and dots are non-members. found to be Herbig Ae/Be and two others had WTTS profiles. These data indicate that X-rays are an efficient tracer of young populations. 2.2.4 Near Infrared Studies Recent surveys in the near-infrared permit investigation of the extension and structure of the cluster. In the study by Li (2005) they analyzed data from the 2MASS all sky survey, and suggested that NGC 2244 had a second component located approximately 6.6 pc west of the core center. Data from the FLAMINGOS survey reported in this thesis, appear to confirm the existence of this second association (see 4), which is coincident with the area originally labeled as NGC 2237. This area is particularly interesting because it contains large dust structures known as "elephant trunks", as well as other types of very young condensations of material which suggest very recent formation. 2.2.5 X-ray Studies NGC 2244 is an important target for X-ray studies due to the interest in investigating the nature of massive stars as sources of high energy photons. The ROSAT Consortium observations yielded 34 X-ray sources in NGC 2244, with typical energies of 1030 1032 ergs s 1. Six of these X-ray sources are PMS candidates as reported by Park & Sung (2002). Also, Berghifer & Christian (2002) studied NGC 2244 ROSAT sources and found that objects with the faintest X-ray emission have very high X-ray to optical luminosity ratios. They noted that the number of X-ray emitters associated with Ha emission in NGC 2244 is remarkable. Taken together, these data give strength to the hypothesis that many X-ray emitters are young late type stars. 2.3 The Rosette Molecular Cloud: Structure 2.3.1 CO studies A substantial part of the Interstellar Medium (ISM) exists in molecular form. Molec- ular hydrogen (H2) is stable and abundant, but unfortunately is not easily detectable because H2has no permanent dipole moment and therefore its transition probabilities are very small. CO is considered instead the best tracer of molecular gas because of its high and constant abundance in molecular hydrogen clouds. First attempts to detect CO emission associated with the Rosette nebula were unsuccessful as they pointed at the nebula region, which is mostly composed of neutral and ionized gas. The observations reported by Blitz & Thaddeus (1980), which targeted the southeast adjacent region of the nebula, were the first successful detections of molecular gas in the Rosette. Their NRAO survey mapped over 80% of the 12CO emission in the area of the cloud with a 1' beam size, and yielded information about its large scale distribution. They estimated the angular extent of the cloud to be 3.5deg (98 pc at a distance of 1600 pc) and labeled the most prominent sub-structures. In the adapted map of Figure 2-4 we show an optical picture of the Rosette nebula from the DSS overlayed with contours of 12CO from the Bell Labs maps of Blitz & Stark (1986). We included the nomenclature of Blitz & Thaddeus (1980). Particularly important regions are the Monoceros Ridge (region A1-2) which is literally a region of gas compression at the cloud-nebula interface; the Cloud Central Core (Al-1), which hosts the most massive clumps in the cloud and is the strongest region of star formation; the cores D and G, which are separated from the main body of the cloud but have ongoing star formation; the IRS core, which hosts the massive proto- binary AFGL-961 (see 4 and 5.1); the Back Core B, which is more loose in structure than the regions near the nebula; and the Arm or E core, which despite its brightness contains no significant star formation (no IRAS sources, or near-infrared clusters have been found in this core so far). In a subsequent study Blitz & Stark (1986) mapped the 12CO and 13CO emission with improved sensitivity at the AT&T Bell Labs, uncovering the high degree of clumpi- ness of the cloud. The study of Williams et al. (1994) also made use of the data from Blitz & Stark (1986) and listed a total of 95 clumps. The clumps with evidence of star formation had larger peak temperatures, larger densities and also were more gravitation- ally bound compared to clumps from the Maddalena Complex, a cloud with very low star formation. Later, Williams et al. (1995) showed that about half of the clumps in the RMC were gravitationally bound and the rest were supported by pressure from the interclump medium, which was shown to be mostly atomic and about 40 times less dense. In figure 2-5 we show the locations and relative sizes of the clumps from Williams et al. (1995). From the clump central velocities Williams et al. (1995) found that the cloud has a well defined velocity gradient of about 0.08 km.s- 1pc1. Also, the negative correlation between clump mass and clump to clump velocity dispersion suggested that the system is still far from equipartition even though it is dynamically evolved. Williams et al. (1995) also found the star forming activity to be more intense in the ridge and the central core areas, near the interface cloud-nebula: clumps located near the nebula presented larger ROSETTE COMPLEX 6.07297 IRAS 25 micron. Digital Sky Survey 5.47608 - 0 P 4.87919 - U . 04.28231 - 3.68542 - AT&T Bell 12CO Levels 3.08853 I I II 6.65547 6.61515 6.57483 6.53451 6.49419 6.45387 Right Acension (J2000) Figure 2-4: A map of the Rosette Complex area. The background image is a DSS plate of the IRS survey at 25 microns. The contours represent 12CO integrated inten- sity levels from the survey of Blitz & Stark (1986). Indicated with labels are the main regions of the molecular complex identified by Blitz & Thaddeus (1980) excitation temperatures, average densities and star forming efficiencies and could be translated as brought clues of evolution. Other properties of the clumps (mass, sizes or line widths) did not show any significant variations along the cloud. Another CO study was done by Schneider et al. (1998a). The observations focused on the central part of the cloud, detailing the structure of the midplane star forming cores. They paid special attention to the IRS core, where the source AFGL-961 is located 2I '' d:_ 1; ,. <* -_ -2 208.5 208 207.5 207 206.5 206 205.5 Galactic Longitude () Figure 2-5: Locations and relative sizes of clumps in the Rosette Molecular Cloud from Williams et al. (1995). The size of the symbol is proportional to the mass of the clump. (see also section 5.1) and pointed out the ample blue wing emission due to the powerful outflow from this object. In a complementary study (Schneider et al., 1998b), examined the CII emission (158pm) at the ridge, the central core and the IRS core. They found weak but significant C+ emission deep into the molecular cloud cores and suggested that the distribution agrees well with a clumpy molecular cloud exposed to a low level UV radiation field. The penetration of UV photons in the cloud is apparently facilitated by a high density contrast clump-interclump medium. The clump mass spectrum in the RMC has the form dN/dM oc M where x =1.6, with small variations in the exponent depending on the range, bin size and beam reso- lution used (for example Williams et al. suggest that x is closer to 1.3). The exponent in this power law is similar to other clouds (Blitz, 1993), but what is more important, it shows that the clump mass spectrum is much shallower than the observed stellar IMF (x = 2.35). This shows that although small clumps have a larger number proportion, most of the mass is contained in only a few big clumps, while for stars both numbers and total mass are dominated by the lowest mass bins. Interestingly enough, the power law index is in fact similar to that corresponding to the mass distribution function for embedded clusters (Lada & Lada, 2003), which is suggestive of a uniform star formation efficiency for most star forming cores. The more recent survey of Heyer et al. (2005), obtained with the wide field array SEQUOIA at the FCRAO 14m telescope have a resolution of 45" at 115 GHz and 47" at 110 GHz. The maps reveal "textural variations" in the 12CO emission across the complex, with a brighter emission component within the nebula projected radius (approx. 40 pc from the center as defined by Celnik (1986)) and weaker, more extended emission outside this ionization edge. HBW05 suggest that the weaker emission is probably due to subthermally excited material with lower densities. They also calculated the total molecular mass of the cloud to be 1.6 x 105M. from 12CO, and found a LTE mass of 1.16 x 105M. from 13CO. Moreover, they were able to apply a Principal Component Analysis (Heyer & Schloerb, 1997) to determine the turbulent flows and the turbulence scale in the RMC. This analysis reveals more significant variations in the velocity structure of the cloud at the regions located within the ionization than in the more diffuse, external component. This fact reveals the interaction of the cloud and the HII region. They suggested, however, that these interactions are still very localized, and have not affected the global dynamics of the cloud yet. 2.3.2 Interaction with the Rosette nebula In his third study of the Rosette, Celnik (1986) focused on comparing his Ha map and radio continuum observations of the nebula (see section 2.1) with the CO map of the molecular cloud from Blitz & Thaddeus (1980). Celnik constructed a complex model of the distribution of the main CO cores (see Fig 2-4) in the context of the HII region and estimated the rotation center of the cloud at (u, 8) = (98.1615,4.3287,J2000.0). Finally, he re-calculated of the mass of the entire complex by adding the total mass of ionized atoms, stars, dust and molecular gas, resulting in 3.3 x 105 M.. Cox et al. (1990), used the available IRAS data (12, 25, 60 and 100pm), and determined in great detail the distribution of dust and compared this to the distributions of ionized and molecular gas. Additionally, they were able to estimate a total infrared luminosity of roughly 1.1 x 106 L. or about 50% of the available luminosity from the cluster NGC 2244. Warm dust (usually present near an OB association) typically emits strongly at the four IRAS bands. However, Cox et al. also showed that in the Rosette, while the 60 and 100pm emission were quite strong at regions of ionized and neutral gas (nebula), the 12pm emission was preferently located beyond the limits of the ionization front (molecular cloud), suggesting a heavy rate of destruction of dust grains from UV radiation from the cluster. Surprisingly, the 25pm emission was found to be significant in some parts of the ionized nebula, possibly due to the existence of a second type of dust particle that is more resistant to UV photons. This was also suggested by Shipman & Clark (1994), who found that the maximum temperature in the shallow temperature gradient found in the nebula (see section 2.1) seems too low to sublimate ice mantles in grains and too low for grains to emit significantly in 12 or 25pm -a problem possibly solved with a second type of grain in the region. Later, Shipman & Carey (1996) suggested that line emission could be contributing strongly to the IR emission of the nebula, and suggested, once more that the presence of a "hot dust" component is necessary to model this emission, especially for the 25pm. Figure 2-6 shows the superposition of the 12pm emission from the IRAS survey and a 1400 MHz radio continuum emission map from Holdaway, Braun & Liszt (un- published). The infrared contours indicate that the warm dust emission defines a shell that encloses the ionization front, showing the effect of heavy dust destruction by the nebula. The overposition of these maps defines very clearly the region where the HII region impacts the molecular cloud. Kuchar & Bania (1993) made a complete map of HI emission at 21cm using the Arrecibo telescope. They found that atomic gas in the Rosette Complex is distributed ROSETTE COMPLEX image: L-band 1400 Mhz - IRAS 12/. 5.5 o O 0 oL 5.0 4.5 4.0 98.0 97.5 97.0 96.5 RA (J1950.0) Figure 2-6: IRAS 12pm emission map superimposed on a 1400 MHz radio continuum emission map by Holdaway, Braun and Liszt, NRAO. in three main regions which form a rough, extended shell around the optical nebula and beyond the molecular cloud, with a center of expansion at (a, 8) =(97.95,4.97, J2000). This shell (according to their calculations) would have a mass close to 2x 104 M which implies a budget of kinetic energy for the shell expansion of approx. 4x 1048 ergs, or 2% of the total energy available from the stars in NGC 2244. 'I I r F'I i; High Energy studies The possibility of interaction between the HII region and the star forming cloud, as well as the location of the Rosette Nebula near the edge of the Monoceros Loop -also known as the Monoceros Supernova Remnant- (Davies, 1963), has motivated a number of studies aimed at investigating the high energy photon emission in the interaction regions. Deep HO+[NII] photographic plates by Davies et al. (1978) suggested a correlation between a filamentary structure observable in Ha emission and a Rosette nebula feature observable in 240 MHz radio waves. This feature was proposed as evidence of loop- nebular interaction and confirmed by decameter (Odegard, 1986) and diffuse X-ray emission (Leahy et al., 1986) observations. Later, high energy (100 MeV) y-ray images from EGRET (Jaffe et al., 1997), revealed a feature partly coincident with the filaments and apparently significant (7o) over expected diffuse emission. If real, these y energy photons would be a product of the interaction of charged particles with the dense ambient medium at the shock region. Recently, the HEGRA system of atmospheric Cerenkov telescopes at IAC was used to calculate the cosmic ray emission from the loop-nebula interaction region, but no significant TeV energies were found (Aharonian et al., 2004). When the Rosette Molecular Cloud was confirmed as a region of star formation, Gregorio-Hetem et al. (1998) used ROSAT data again, this time to map the MonR2 cluster and the Rosette Molecular Cloud areas in order to confirm a correlation between star forming cores and clusters of X-rays sources. They found strong X-ray emission in NGC 2244, the ridge of the cloud (A1-2 in fig 2-4), and at the cloud core area (Al-1), but the resolution was poor and individual sources could not be resolved. They suggested that molecular cores known to have active star formation but failing to show significant X-ray emission, could be predominantly forming low-mass stars. They also suggested that detectable X-ray counterparts are in most cases Herbig/AeBe or T Tauri stars, as found in NGC 2244 (Li et al., 2002). The more recent observations of the Rosette Complex done with Chandra (Townsley et al., 2003) have resolutions of only a few arcseconds, thus allowing for the detection of X-ray counterparts for 75% of the OB members of NGC 2244. One of the most interesting results of this study was the confirmation of a second, soft diffuse emission which probably originates from the O star winds and is later brought to thermal- ization by wind-wind interactions or by the shock with the surroundings, in this case the molecular cloud (see figure 2-7). This X-ray plasma surrounds the OB association and fills the nebula cavity completely. 6:3f00 W. .. 34:p0 3p 33:p0 3 32:p0 31:p0 . Right Ascension (J2000) Figure 2-7: A 0.5-2 keV Chandra image of the Rosette Complex. The emission has been smoothed to highlight the soft diffuse emission that originates in the neb- ula and propagates into the molecular cloud. Credit: Townsley et al. and NASA/Chandra X-Ray Observatory (2003). 2.4 The Rosette Molecular Cloud: Embedded Populations The coincidence of massive clumps and luminous IRAS sources pointed out by the study of Williams et al. (1995) strongly suggested that star formation had already taken place across the molecular cloud. However, the poor spatial resolution of the IRAS point source survey did not allow the resolution of individual members of an embedded population. Early near-infrared studies (e.g. Perez et al., 1987) did not cover the molecular cloud areas, and optical photometric studies were incapable of detecting obscured populations. An exploratory near-infrared survey (JHK) by Phelps & Lada (1997) that made use of the imager SQIID finally confirmed the existence of embedded clusters in some of the most massive clumps from the list of Williams et al. (1995) that were associated with an IRAS source. They were able to distinguish seven deeply embedded clusters with bright nebulosities, and suggested that clumps not forming a cluster, might not be physically bound. The location of the seven Phelps & Lada (1997) clusters is shown in Figure 2-8 Complete area coverage of the Rosette Complex in the near-infrared was first accomplished with the release of the All-Sky 2MASS survey catalogs. The 2MASS survey was a major gain in data uniformity but unfortunately not in sensitivity. Due to the distance to the Rosette (1.6 kpc), the 2MASS completeness limit (K=14.3 mag) is not deep enough to study the low mass end of the IMF. So, what is the next logical step in the study of the Rosette Complex? The existence of embedded clusters in the Molecular Cloud means that the cloud is actively forming stars and that at least a fraction of the new stars were formed in clusters from the collapse of some of the most massive clumps of molecular gas. This leads to two problems of importance: 2.4.1 Dominance of Cluster Formation in the Rosette Complex The first problem is to determine if star formation in the RMC leads to a dominance of rich clusters. Are there any low density groups as well? Is there any evidence for a Galctic Longitude (1) Figure 2-8: The location of the clusters identified in the study of Phelps & Lada 2.0 -2.2 . -5I onurvey Bell Lobsntours indicate CO emission from the maps of Blitz & Stark -2.4 +Embedded Clusters P 208.0 207.5 207.0 206.5 206.0 Golactic Longitude (M) Figure 2-8: The location of the clusters identified in the study of Phelps & Lada (1997). The background image is an optical plate from the Digital Sky Survey. The contours indicate 12C0 emission from the maps of Blitz & Stark (1986) distributed population? For example,Carpenter (2000) showed that the MonR2 region, might be harboring a low density population, counting for up to 9% of the total number of young stars. The nature of such low spatial density members is not clear, as it could be formed independently of the cluster population, but also could be the result of the dispersal of an older high spatial density population. Another study that attempts to account for the contribution of distributed populations in star forming clouds was done by Li et al. (1997), who found that for the L1630 cloud, where Lada et al. (1991b) found unequivocal dominance of cluster formation. The fraction of infrared excess stars in the inter-cluster areas of the cloud was found to be very small, suggesting that the contribution of low-density formation was almost negligible. We need to consider that the RMC is located 4 times further away than the well studied Orion or Perseus molecular clouds (d = 300 500 pc), where reasonably deep observations can easily detect low mass stars. Equivalent detections in the RMC would need observations at least 3 magnitudes deeper. Furthermore, the RMC is located in the direction of the galactic anticenter (1=207deg) and closer to the galactic disk (b=-2) than Orion (b=-16.3) or Perseus (b=-20.6). As a result, the density of the field population towards the RC is very high, and any faint, low mass stars are probably well mixed with foreground and background sources. These problems would make very difficult to detect other clusters or a low density population by means of single band stellar density counts, as in other cloud surveys (Lada et al., 1991b; Carpenter, 2000).Also, depth limited databases like 2MASS are not sensitive enough to study the Rosette Complex. Deep multi-wavelength photometry, capable of rendering infrared colors, for even low mass populations is necessary to separate members from field stars by means of accurate extinction statistics and counts of infrared excess stars. 2.4.2 The Hypothesis of Sequential Star Formation The second problem to be understood, relates to the physical processes that led to the formation of stellar clusters in the Rosette: Are those processes similar to those occurring in other star forming clouds?. The current hypothesis is that the formation of star clusters in the RMC was possibly stimulated by the interaction of the HII region and the cloud. The expansion of the Nebula via the ionization front generated by the strong stellar winds of the massive association NGC 2244, results in a shock front which interacts with the gas of the molecular cloud, as shown in some of the studies mentioned above. The hypothesis is that the shock front directly stimulated the collapse of clumps which then formed the clusters. This model is known as sequential star formation (SSF), and was developed theoretically by (Elmegreen & Lada, 1977). In the study of Williams et al. (1995), it was shown that the cloud had larger values of excitation temperature, clump density and possibly, star formation efficiency near the HII region. However, there are not significant differences among characteristics of cluster forming clumps, namely mass, size or line width across the cloud. Could this mean that other massive clumps, either those not included in the areas of the Phelps & Lada (1997) survey, or those not associated with a luminous IRAS point source could also have formed stars recently, even if their location is not favorable with respect to the shock front? In other words, how feasible is the hypothesis of SSF? The detection of additional embedded populations would allow us to determine for once if star formation is preferentially located near the shock front of the nebula expansion. We might also be able to find a relation between the characteristics of the embedded clusters and their distance to the HII region that could support or discard the hypothesis of SSF. CHAPTER 3 A NEAR-IR SURVEY OF THE ROSETTE COMPLEX: OBSERVATIONS 3.1 The FLAMINGOS GMC Survey As we discussed in Chapter 1, a systematic and thorough investigation of the young star population of GMCs is the key to understanding the global aspects of the problem of Star Formation. Historically, the large distances and large angular sizes of GMCs, made it difficult and costly to perform surveys of embedded populations which could render both photometric depth and area coverage. Technological limitations were also a factor, with infrared detectors having very small areas: until the early 1990s, near-infrared arrays were no larger than 256 x 256 pixels which resulted in rather poor resolution and a small field of view (FOV). For example, the survey of the region L1630 in the Orion Molecular Cloud by Lada et al. (1991b) used a 58 x 62 pixel device which rendered a FOV of only 1' x 1', and thus required of 2800 images to cover an area of approx. 0.7 square degrees in the K band. Large devices were developed then, with the instrument SQIID (Ellis et al., 1993) being the first versatile instrument to allow a high resolution and a large FOV (1024 x 1024 InSb device with simultaneous quadrant detection in J,H and K). This camera was used for the first time to survey large areas of star formation regions in multi-band mode, like rho Ophiuchi (Barsony et al., 1997) and the Rosette Molecular Cloud (PL97). Near the end of the decade, the first 2048 x 2048 HgCdTe devices for the use in astronomical instrumentation were developed (Kozlowski et al., 1998), opening even better possibilities. The instrument FLAMINGOS (Elston, 1998), developed at the University of Flhtida., takes advantage of the 4 million pixel detectors by being designed as a combination wide field near-IR imager and multi-object spectrometer. The camera has a Lyot stop wheel with a number of stops customized to receive different input beams slower than f/7 and therefore can provide a wide range of pixel scales for imaging. For example, on the Kitt Peak 2.1m telescope it renders 0.606" pixels and a 20'x 20' arcminute FOV. This particular setup makes FLAMINGOS an excellent survey imager as entire square degree areas can be surveyed with a few observed fields. The instrument has a suite of four filters: J, H, K and K, which cover the whole near-IR wavelength range from 1.6 to 2.2pm. Intended as one of the first large scale applications of the instrument, the NOAO survey program Toward a Complete Near-Infrared Spectroscopic and Imaging Survey of Giant Molecular Clouds (PI E. A. Lada) is dedicated to the global study of several giant molecular clouds using FLAMINGOS. One of the two main goals of the survey is to do a complete imaging coverage in J, H and K of comprehensive areas of the clouds with a photometric depth that assures coverage down to the Hydrogen Burning Limit (HBL). Four important GMCs were selected for this survey: Orion B, Perseus, Monoceros, Cepheus, Serpens and the Rosette. Observations for the survey program have been carried out over the course of 6 winter observing seasons from 2000 to 2005. The survey was done at the 2.1 and 4.0m telescopes of the Kitt Peak National Observatory, where FLAMINGOS is a commissioned instrument. Although FLAMINGOS is suitable for multi-object spectroscopy (MOS) and imaging mode in both telescopes, we preferentially performed imaging at the 2.1m telescope, where FLAMINGOS has a larger FOV, while the 4.0m telescope has been used essentially for MOS. The imaging observations for the GMCs targets were carried out iteratively, with Orion B and the Rosette being the first clouds to be completed. After reduction of the first batch of observations, a first quality assessment was performed by members of the team and collaborators at the Center for Astrophysics in Cambridge, Massachussetts during the fall of 2003. Those fields that yielded poor results were assigned for re-observation during the winters of 2003 and 2004. In the particular case of the Rosette Complex, a total of 22 FLAMINGOS fields were observed during the winters of 2001 to 2004 in the 3 available filters J, H and K, resulting in a very complete coverage of the Rosette Nebula the Rosette Molecular Cloud areas. We selected the area of the Complex to be covered from the 12CO and 13CO emission maps of Blitz & Stark (1986) and the 25pm emission map from the IRAS survey. Twenty of our 22 fields are adjacent, while 3 of them (areas 4, G1 and G2 ) were added to enhance the quality of the observations in some particularly interesting regions. In order to account for the field contamination, two control fields were observed at close distance from the survey areas but away from the main molecular cloud emission. The control fields were observed with an equivalent method to the main survey fields, and have the same depth as any of our on-source fields. A map showing the positions of the observed fields in the context of the molecular gas emission (12CO) and the 25 micron IRAS flux in the area can be seen in Fig 3-1. For each field we aimed for a total of 1000 sec. on source integration in each filter (for some fields, weather conditions and defective frames kept us a tad below this goal), which was done by obtaining a number of short, dithered exposures. For the J and H filters, we used dither exposures of 60 sec. each each dither, and for the K band, with a higher sensitivity, we used 20 or 30 sec. dithers, depending on the weather conditions. Details of the observations, including dates, total integration times, average seeing, and airmass for each field can be consulted in Appendix A. 3.2 Data Reduction 3.2.1 The Data Reduction Pipeline: LongLegs Each FLAMINGOS individual image is stored as a FITS file with a size of 16 Megabytes. Each field is observed in three filters and requires combining groups of dithered pointing. The resultant amount of data for the survey is therefore very large, and required the development of automated processing pipelines for reduction and photometry. ROSETTE COMPLEX 5.67218 IRAS 251 Digital Sky Survey T-- 14 15 :i 4.96886- 10 2 21 3 . 4.26554 1 6 SI20 3.56222- 2 CF2 2.85891 - AT&T Bell '2CO Levels + Rosette Embedded Clusters 2.15561 I I I I 99.5010 98.8144 98.1278 97.4412 96.7547 96.0681 Right Acension (J2000) FLAMINGOS Giant Molecular Cloud Survey Figure 3-1: Scheme of the University of Florida/NOAO Rosette Molecular Cloud Survey. The boxes delimit individual FLAMINGOS fields (20 x 20' after trimming) over an image of the IRAS 25pm emission in the region. The labels at the left side of each box refer hereafter to the fields detailed in Ap- pendix A and the text. Light solid contours represent the extension of the Rosette Molecular Cloud in CO emission from the survey of Blitz & Stark (1986). Crosses mark the centers of known embedded clusters from the previous study of Phelps & Lada (1997). Our image reduction pipeline, nicknamed LongLegs and programmed by the author, is a standard IRAF routine script divided into three main phases: During the first phase, the pipeline rejects defective images and removes bad pixels. Then, it applies a 3rd degree polynomial linearization correction for every image on a pixel by pixel basis (IRAF routine IRLINCOR). Dark and flat field data groups are combined into master flat and dark fields, which are then used to create bad pixel masks. The second phase of LongLegs is a two-step preparation of pre-combined data. The algorithm combines groups of 8 adjacent images to create a local sky. Then, after each data frame has been sky-subtracted and divided by the normalized flat field, the program reconstructs the individual images dither pattern based on the positions of the 200 brightest sources in each frame. The program does a first combination of data, extracts the positions of sources with fluxes larger than a pre-selected sigma level, and masks them out from individual images to create a new set of "starless" local sky frames. These are used for a second pass of sky subtraction, flat field division and shift-and-add combination. The final result of this phase is a set of precombined frames and a first combined image with analysis quality that only lacks a geometric distortion correction. On the third phase the pipeline program corrects for geometric distortion, using a sixth order Chebyshev polynomial solution map constructed from the positional distortions of a 20x20 pinhole grid mask that is pre-imaged each time the instrument is corrected internally (corrections indicated slight variations in the geometric distortion from season to season). The pre-combined data is also re-sampled to half-size pixels, the dithers are centroid corrected, and re-combined into a final image that is 4096 x 4096 pixels in size and is ready for the photometry pipeline. 3.2.2 The Photometry and Astrometry Pipeline: PinkPack Our photometry and astrometry pipeline, nicknamed PinkPack and programmed by Joanna Levine (Levine, 2006), performs stellar profile fitting (also known as Point Spread Function or PSF fitting) photometry on a LongLegs final product. The script also uses standard IRAF-DAOPHOT tasks (Stetson, 1987), except for the detection, which is performed using the S-extractor algorithm (Bertin & Arnouts, 1996). A full description of Pinkpack can be found in Levine's PhD thesis. We will only mention that the pipeline gives out a full photometric calibration and an astrometric solution with respect to the 2MASS All Source Catalog Release data. Calibration of data is done in the range K = 11 to 14.5 mag. Once a photometric catalog is obtained and an astrometric solution is calculated, this pipeline combines the data from different filters into a final merged catalog that contains, for each object, an ID, final RA-DEC coordinates, pixel positions (in the K band image), and photometry for all the bands, including profile fitting uncertainties. The photometry pipeline also has the option of creating and subtracting a median value image for a specific field in order to enhance the detection of sources in regions with bright nebulosities. We used this option in all frames that contained bright nebulosi- ties, although doing the same in non-nebulous regions had no effect whatsoever on the number of detections obtained. 3.3 Completeness of Sample In order to estimate the completeness of our sample, we performed intensive artificial star experiments in 3 selected fields of the survey, each one considered to be characteristic of a type of region: crowded (with low extinction), sparse (high extinction, no nebulosity) and with bright nebulosity. Our main goal was to determine mean values of completeness to apply to the entire survey for our statistical purposes. The completeness limits for the region containing bright nebulosity emission were not affected in a greater way than in zones of higher stellar density, although in both cases the experiments performed slightly better in the sparse fields (see Figures 3-2 and 3-3). For all the fields, the artificial stars were added partially, in consecutive annuli of 250 pixels from the center of the frame. For each annuli, 100 artificial images with 50 stars in an uniform distribution were created based on the resultant PSF profiles from Pinkpack for that specific field. Their magnitudes were adjusted according to the mean zero point value calculated from the 2MASS calibration. The resultant images were then reduced Rosette Mo ecular Cloud Artificial Star Experiments (FILTER) 9-20 S16 0 500 1000 1500 2000 20 19 I Crowded 1 8 I0----- -. ,__ Nebulous -cc S17 Sparse 1 16 Sparse 2 15 I 0 500 1000 1500 2000 20 19 1 9 L 1 8 - S17 16 0 500 1000 1500 2000 Figure 3-2: Results of the artificial star experiments described in section 2.4. a) the left side panels show the turnoff magnitudes (limits of 90% object recovery) by filter. The average values in this graphs were used as our general complete- ness limits for the survey. with the same set of parameters as the original frame, and the positions and magnitudes of the artificial stars were recovered using the XYXYMATCH routine from IRAF. The stars in the recovery catalogs were divided by brightness in bins of 0.25 mag, and the completeness limit was calculated as the bin at which the recovery fraction descended below 90 percent. Rosette Molecular Cloud Artificial Star Experiments (AREA) s-"- S20 I 19 .. .................... .... I I .: ....... ... . 16 S0 500 1000 1500 2000 ......................... ..... .. J b n 9 1 ...... J bond 18 ------ -cc-H bond The resultant average completeness limits are K=17.25, H=18. and J=18.50 mag S 17 I I I SI I I .... K band S0 500 1000 1500 2000 within the limits of acceptable focus quality of the images (see section 3.4); these results < 19 .. . .I ... a_ -- ----- --- *--- E 17 I. o 1 6 E I I I IL 0 500 1000 1500 2000 Figure 3-3: Results of the artificial star experiments separated by type of field. We de- tected a subtle variation of the recovery limits in the case of too crowded or too nebulous fields. The resultant average completeness limits are K=17.25, H=18.00 and J=18.50 mag within the limits of acceptable focus quality of the images (see section 3.4); these results rapidly degrade in the areas of high optical distortion. However, these stand for now as some of the deepest observations of the region, going about 3 magnitudes fainter than 2MASS, and thus assuring the detection of stars around and below the HBL. 3.4 Positional Correction of Photometry During the assessment of data quality it was noticed that our pipelines had difficulties adjusting correctly the PSF in certain areas of the chip. The problem was worse towards the corners of the images, where the stellar profiles were in some cases clearly aberrated and even presented prominent comas. It is known that the parabolic shame of the primary mirror has an effect on large detectors, which can be usually corrected with a second degree surface variation of the PSF profile, but apparently the distortions we observed had a different origin, because the distortions are not symmetrical, i.e., the four comers of the images are affected differently. The distortions can also worsen with poor focusing and bad weather (i.e. mediocre seeing values). One hypothesis based on optical path simulations (Eikenberry, S. Univ. of Florida, personal communication) is that the alignment between the primary and secondary mirrors of the KPNO-2.lm telescope has lost accuracy along the years, affecting the symmetry of the focus and shifting the center of optical alignment. This defect is unfortunately enhanced by the large FOV of FLAMINGOS. The size of the affected area varied slightly from season to season. The area with minimal distortion is nearly circular, with a center that falls systematically on the pixel position (3000,2400) for observations made before the fall 2004, and on the position (3200,2170) for more recent observations. The radius of this area within which the PSF values stay uniform is variable, with an average of 3200600 pixels depending on the observing conditions, mainly seeing value, which is affected respectively by the airmass and the weather conditions at the observatory. For most of our fields, about 75-95% of the area of the detector contained minimal distortion, with reasonably smooth PSF profiles and small (< 0.025 mag) photometric differences with respect to the 2MASS catalogs. Outside this area, the optical distortion increases quickly, and therefore the shape of the stars and the PSF profiles degraded to the point that stars presented noticeable aberration comas and larger PSF FWHM values,especially at the two eastward comers of the detector, which resulted in poor fittings to the average PSF profile generated from good quality stars in the acceptable area, and generated a net flux loss with respect to 2MASS that raises sharply to 0.5 mag in the bad psf fitting areas, independently of the filter. In terms of the net output to our photometric catalogs, this effect resulted in a variation of the photometric calibration zero point (ZPT) value across the images, and this affected the uniformity of the survey from field to field. In order to correct for this effect, we applied a 6th order Legendre polynomial correction of the ZPT values as a function of the radial position with respect to the pixel centers of the optical distortion circles. The solution was developed and constructed as interactive software by Andrea Stolte. The correction is applied done by fitting a polynomial to a fiducial line made by the median values of the 2MASS vs FLAMINGOS differences in radial bins of 300 pixels from the optical distortion center. The correction was calculated within the ranges 10.0 to 14.0, 10.0 to 15.0 and 10.0 to 16.0 mag in K, H and J only due to the limitations of sensitivity of the 2MASS catalogs, but was applied to every star detected by our pipelines. This method allowed us to reduce the scatter in the ZPT values, and to determine (by field and by filter) which was the cutoff radius from the minimal distortion center, at which ZPT differences with respect to 2MASS rose above a maximum tolerance of 0.3 mag. Inside the area marked by this cutoff radius, the polynomial correction reduced the ZPT differences significantly, and this also results in a decrease of the noise in the color terms. The areas located beyond the cutoff circles, towards the east (left) corners of the detector, have too large optical distortions, and so objects detected in those areas were removed from our final catalogs Figure 3-4 shows schematically the positions and extensions of every field observed, as well as the cutoff radii of the ZPT correction for each filter. The K band circle, being the most conservative, always defines the area of the field that was kept for analysis. This of course, has an exception for those areas that have an overlap with the good quality regions of another coincident field, in which case our catalog joining program selected systematically the star from the good frame into the final catalog. Rosette Complex 5.567111 1 5.12495 4.68279 4.24063 3.79847 3.35631 99.3568 98.9089 98.4611 98.0132 Right Acension (J2000) FLAMINGOS GMC Survey 97.5654 97.1175 Figure 3-4: The extension of the areas of acceptable optical distortion are marked for each field as circles with radii equal to the center of the maximum bin at which the ZPT polynomial correction to the zero points (see text and fig- ure 5) can be applied within the detector. This effect varies by field (size of the acceptable area) and filter: the solid, dotted and dashed linestyle circles represent the tolerance radii for J, H and K respectively. In the various panels of Figures 3-5 to 3-9, we show, as an example, the effects of the polynomial ZPT correction in the area 01 of our survey (which coincides approxi- mately with the center of the Rosette Nebula).. The first group of plots (Figures 3-5) shows the polynomial ZPT correction applied to the photometric differences FLAMINGOS vs 2MASS as a function of radial position from the minimal distortion center (3000,2400). As can be noticed, the scatter in the zero point per magnitude is clearly corrected and the differences at large radii now converge closer to zero and stay within a 0.1 mag range. Figure 3-5: Example of the results of the ZPT polynomial correction of the zero point as a function of pixel radial distance from the center of low optical distortion (3000,2400) in the FLAMINGOS detector for region 1 of our Rosette survey in filters J and H. The dots represent matches of the FLAMINGOS data with 2MASS sources in the ranges 11.0 to 15.0 and 11.0 to 16.0 H and J respec- tively. The solid line represents a 6th order Legendre polynomial fit to the median values of the scatter in bins of 300 pixels. The dashed lines indicate levels of 0.1 mag of scatter. -- -- -- -- -- - .. -0.5 1 .0 . . . . 0 1000 2000 3000 R (xc=3000, yc=2400) 2002 Jan 13 ROSETE 01 K see 1.00 1.0 -- 0.000+-0.092 0.5 0.0 -0.5 1 .0 . __.. . .___ . ..___. . 0 1000 2000 3000 R (xc=3000, yc=2400) Figure 3-6: Same as Figure 3-5 but for the K filter. In the second group of plots (Figure 3-7) we can see how the values of the pho- tometric differences with respect to the observed colors are also reduced intrinsically, decreasing the overall uncertainty of our photometry. The third set of plots (Figure 3-8) shows the net effect of the ZPT correction on the color-magnitude and color-color diagrams. As it can be noticed, the color-magnitude sequences for background and for members in the field get more confined and better separated, which in the color-color space results in a reduction of the scatter around the zero age and giant sequences. This, consequently, reduces the number of spurious detections in the infrared excess region of the color-color diagram,especially in the area located closer to the intersection of the T-tauri reddening band and the main sequence. In the fourth set of plots of Figure 3-9 we show the scatter of these photometric differences for field 01 before and after the correction is applied. In the left panels of the 0.0 0.0 0.5 0.5 1.0 1.0 0.0 0.5 1.0 1.5 2.0 2.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 H < 15.00 & R < 3387 H < 15.00 & R < 3387 1.0 1.0 0.5 0.5 M* V 0.0 o0.0 . 0.5 -0.5 1 .0 . 1 .0 ._.._._.._._.._. 00 0.5 1.0 1.5 2.0 2.5 0.0 0 5 1.0 1.5 2 0 2.5 3.0 3.5 H K (2MASS) J H (2MASS) Figure 3-7: Example of the results of the ZPT correction in the J -H and H K color differences of FLAMINGOS vs 2MASS as a function of magnitude in area 1 of our Rosette survey. Top and bottom panels indicate color differences before and after the polynomial correction. The color differences are also reduced. The solid line indicates the zero level. figure we plot the FLAMINGOS 2MASS differences as a function of magnitude, which has larger scatter values towards fainter magnitudes. The mean value of the differences is closer to zero after the correction. The right panels show how the correction results also in a reduction of the net photometric scatter measured by the standard deviation of the differences within consecutive magnitude bins. The median value of these deviations is indicated with a dashed line. 3.5 Quality and Uniformity of the Survey The overall uniformity in the quality of the photometry of our survey can be simply assessed by comparing the mean values of the scatter in the FLAMINGOS vs.2MASS differences after applying the positional correction, as described above. The median 18 1i or ol 0.5 - 20 251 excess sources 0 .0 1 1 1 , 0 1 2 3 4 -0.5 0.0 0.5 1.0 1.5 2.0 J K H K counterfeit sigma PSF < 0.1 & R < 2705 counterfeit sigma PSF < 0.1 & R < 2705 3.0 - 2.5 - 14 / /r St 2.0 / S16 8 1.5 .. S18 -1 .0 " 0.5 20 R < 3105 240 excess sources . . . i . i . .0 0 1. 1 i . i . 0 1 2 3 4 0.5 0.0 0.5 1.0 1.5 2.0 J K corrected H K corrected Figure 3-8: Example of the results of the zero point correction in the color-magnitude and color-color spaces for the sources in area 1. The top and bottom panels represent data and after before the polynomial correction is applied. scatter values were compiled for every field and filter, and we use these numbers the indicators of the net photometric quality of an observation. The medin scatter values for each field are shown in Table A-2 of Appendix A. The correction allowed us to reduce our internal photometric scatter in individual bands by an average of 0.02 mag within the low optical distortion areas. The average of these scatter values over the whole survey are 0.058 + 0.012, 0.064 + 0.018 and 0.056 0.014 in J, H and K respectively. In addition to our completeness limits, which are statistical, there is another set of limits which represent the sensitivity of the survey, i.e. the faintest magnitude at which an object in our catalog can be consider to have good photometric quality within our errors These sensitivity limits are also different for each filter, and we estimated them -^ - - -0.5 0.05 - 0.00 10 12 14 16 18 10 12 14 16 18 JFLMN JFLMN 2002 Jan 13 ROSETTE 01 J see 1.00 1.0 0.25 - -0 002+-0.071 fit lmnt llo 16o = 0.061 0.20 0.5 hs S a 0.15 ^ -- 0.0 0 sp1ec4tandvKe1ly.'7 ma.o -0.5d a 0.05t0 estmatd c eten lits.I 0.00 10 12 14 16 18 10 12 14 16 18 JFLMN JFLMN Figure 3-9: Example of the results of the ZPT correction in the values and scatter of the FLAMINGOS vs 2MASS differences as a function of magnitude (J band). Top and bottom panels show data before and after the correction re- spectively. The numbers on the top of left panels are median and standard deviations within the fit range. For the right panels, the dashed line and the numbers on the top represents the median of the standard deviation per mag- nitude bin. Solid and dotted lines mark the 0.0, 0.05 and 0.1 scatter levels. simply as the points where the fiducial curve represented by the average values of the FLAMINGOS photometric errors crosses the standard 107 limit defined as 0.109 mag level in J, H and K. The result is shown in fig 3-10; our 107 crossing values are J=19.4, H=18.4 and K=17.7 mag. These values are also a good indication of the limits at which our photometric values are consistent in quality, and are slightly higher than our estimated completeness limits. In fact, it is possible that within certain areas of the fields,especially near the centers of the low optical distortion areas, the completeness of the data is in fact higher than the average values calculated from the artificial star experiments, but as stated in the section 3.7, Figure 3-10: The solid fiduciall) line joins the points which mark the median value of the PSF fitting photometric uncertainty in a given magnitude bin. Error bars in- dicate the standard deviation in each bin. The horizontal dashed lines marks the zero level and the sensitivity level, estimated at the standard 10-o limit of 0.109 mag. Vertical dotted lines indicate the sensitivity limits, marked as the bin at which the fiducial curve generated by the median values crosses the 10-o line. The dash-dotted line represents the 3-0 level from the fiducial line; any star in our catalog with errors higher than this levels were rejected from the analysis. this depends greatly on the variations of the extinction across the Molecular Cloud, and logically, even the completeness limits can be compromised accordingly in certain spots. 3.6 Construction of Final Catalog 3.6.1 Intrinsic quality: 2MASS Addendum Our photometric uncertainties also degrade relatively steeply at the bright end, approximately below 11.0 magnitudes in J, H and K. This is because this coincides with the level at which the counts per pixel in the detector reach values above 3 x 104 counts, right at the limit of linearity and saturation. In order to account for this effect, we rejected, for our analysis, every source in our catalog with a magnitude brighter than H=1 1.0, as this filter seem to be affected slightly worse by the saturation effect. To complete this end of the magnitude spectrum, we added 2MASS sources to complete our catalogs within the range 5.0 to 11.0 mag in all filters. The total addendum of 2MASS objects to our survey is 798 objects. 3.6.2 Survey Area Merging A final catalog that includes the identifications, astrometry, photometry and basic information about variability in overlap areas for all the sources accepted from our survey, was created using a catalog merging code that put together catalogs for individual fields and calculated the weighted average values for matches (duplicates) in overlap areas. In the case where a match source was located in an area of high optical distortion, null values were ignored. Most of the regions in the survey are adjacent, so that the overlapping regions were usually small; in fact, the outermost 1 arcmin ribbon of the final combined images was in most cases trimmed out of the final images, because Pinkack only runs on the regions of the image that contained a combination of at least 50 percent of the individual dithers. However, regions 4, G1 and G2 have large overlapping regions within the RMC area and in those cases the selective matching came in handy. After merging all of the individual catalogs, including the 798 2MASS sources for the bright end, our final survey catalog contained a total of 153,266 objects. For our analysis however, we restricted our study to those stars with an uncertainty values below 3.0o above the median (see Figure 3-10). The total number of sources in this selection is 146,868. Completeness limits at this level were not affected as 97.3% of these 3-sigma rejections comprised magnitude ranges above J=18.50, H=18.25 and K=17.50 mag respectively. 3.7 Intrinsic Detection Constraints We consider important to mention that surveys with an emphasis on embedded young populations, even with the aid of infrared detectors, cannot assure full detection of all embedded cluster members. Most of young clusters are embedded in the remnants of their original molecular gas cores, which are expected to carry massive amounts of dust and dense gas, to the point that some objects (specially low mass protostars) will be so highly extinguished that they will not be detected even in the near infrared, and thus a fraction of the true number of cluster stars will be left off the counting. In addition, cluster populations will always be mixed with a significant number of foreground and background objects, whose contribution has to be estimated from the control fields. Contributions of field stars are corrected by extinction effects, counted per magnitude bin and subtracted from equivalent counts on-field. After these subtractions are applied, final counting for the number of members in a cluster region are only statistical estimates and cannot determine the membership of individual sources. Another related factor that has to be taken into consideration is that low mass pre- main sequence stars become fainter at older ages, with the consequence that at a certain sensitivity limit they will just not be detectable (Carpenter, 2000). This sensitivity limit depends on the age and mass distribution, as well as extinction, and the net effect is only known accurately for a few regions. In our figure 3-11, which is similar to figure 18 from (Carpenter, 2000), we try to show the intrinsic limitations on mass and age detection that our RMC survey has. For our completeness limit of K=17.25 we expect to detect stars well beyond the HBL if all the stars were younger than 2.5 Myr and no extinction was present in the line of sight; if stars are older, the range of observable stellar mass is reduced. In a typical molecular DM=1 1.02, DM97 models 108 107 106 105 0.05 0.1 1.0 2.0 Log(Mass) [Mo] Figure 3-11: Contours of equal K magnitude value as a function of stellar mass and age at the estimated distance of the Rosette Molecular Cloud (1600 pc, distance modulus=1 1.02). For the construction of these plots we used the pre-main sequence evolution models of D'Antona & Mazzitelli (1997). The solid lines represent iso-magnitude levels with no extinction, while dotted lines represent the same levels with a Av=5.0 mag extinction. The contour at K=17.25 coincides with the completeness limit of our FLAMINGOS survey, and the dotted vertical line marks the 0.08 M. HBL limit. cloud, typical extinction values range from a few to 50 magnitudes in visual wavelengths, which would cause the completeness limits to compromise up to 5 magnitudes in K, causing the range of detectable ages and masses to be even shorter in some areas. 47 Most of the known clusters in the Rosette are deeply or partially embedded in their parental cores (see 5), which means that their ages could be possibly no older than 1-2 Myr (the most recent spectroscopically estimated age of NGC 2244 is about 2 Myr), and thus, more than 80% of the embedded stars could be detected, depending on the extinction level, but this effect is not uniform even at sub-cluster scales. CHAPTER 4 NEAR-IR SURVEY: ANALYSIS AND RESULTS 4.1 Introduction As discussed in Chapter 1, the study of embedded clusters in Giant Molecular Clouds (GMCs) is of capital importance to understand the problem of star formation: at the embedded stage, star clusters have not evolved significantly, and therefore their densities and mass distributions are still close to the original fragmentation of their forming cores. From observations of embedded clusters in nearby GMCs (d < 2 kpc), there is good evidence that a major fraction of the stars are formed in a cluster environment, with rich clusters (100 members or more) clearly dominating over small groups, as they contain more than 80% of the embedded stellar population in GMCs. Unfortunately, available catalogs of embedded clusters are still incomplete at the small cluster sizes. Systematic surveys of molecular clouds that are focused on the detection of young clusters are usually limited to those regions with signposts of star formation (like the presence of luminous IRAS sources), which lead to the discovery of the richest clusters. Only in a very few cases has there been an specific search for low density groups or distributed embedded populations, given that they are logically more difficult to detect, especially if they are mostly composed of low mass stars, with large spatial distributions and projected against a high background of reddened sources. The RMC is a particularly active star formation region, with a large OB association, NGC 2244, whose winds have generated an expanding HII region. One unsolved problem is to determine if the shock front generated by this photodissociation bubble was the principal trigger of the formation of the observed embedded clusters found by (Phelps & Lada, 1997) in the adjacent, highly structured Rosette Molecular Cloud (RMC) (Williams et al., 1995). All of the PL97 clusters are associated with a luminous IRAS source and a massive molecular clump, and until now, no other study has been able to determine the existence of additional clusters. One of the main problems is the large distance to the Rosette: the cloud is located at d = 1.6 kpc, roughly 4 times further away than nearby clouds like Orion or Perseus and for this reason previous studies with limited photometric sensitivity (depth) were unsuccessful at giving any new information on the distribution of young populations in the Rosette. For example, the 2MASS survey is complete only to K = 14.3 mag, which is not enough to detect low mass stars or deeply embedded high mass stars. This makes difficult to detect highly embedded clusters, specially if their local surface densities are low. Another difficulty is that the Rosette is located at a very low galactic latitude (b = -1.8 2.0), which implies a very high density of field objects, and thus, if a search for clusters was performed using monochromatic wavelength counts like it was done for other clouds, then corrections for background contamination would be large and difficult to apply. Our survey is designed to address these problems by a) doing deep observations of the region, capable of detecting stars close or below the HBL and thus improving the detectability of previous studies, b) replacing the use of monotonic wavelength counts by emulating the technique of Li et al., in which the detection of young populations is done by photometric color selection, and c) using the method of Nearest Neighbors (Casertano & Hut, 1985) to distinguish areas with surface densities intrinsically larger than the overpopulated field. Among the main goals of our survey are to study the characteristics of the known clusters, to determine if there are more, and to study their distribution across the complex, and in the context of NGC 2244. Because our observations are deep enough to detect low mass stars we should be able to trace well the structure and extension of embedded populations in the Rosette Complex, adding valuable information about the nature of stellar nurseries in GMCs. 4.2 Analysis 4.2.1 The Nearest Neighbor Method Single band monotonic detection of embedded clusters is done by subtracting normalized control field counts (corrected by extinction) from counts towards the cloud line of sight. This is expected to give the total number of expected members of a certain region, and clusters are defined as regions with surface densities significantly higher than the field. Unfortunately the method is biased because it will only be able to detect very large or very dense clusters. Another problem is that the subtraction of field counts is more difficult for more distant clouds, because members are fainter and are mixed with a larger number of foreground and background stars. The Rosette Complex is located 4 times further away than other star forming complexes where systematic searches for embedded clusters have been performed, like Orion or Perseus. Furthermore, the Rosette is located at a very low galactic latitude (b = -2) and towards the anticenter of the Galaxy (1 = 210), which results in an intrinsically high density of field sources located in the foreground and background of the cloud. For example, one typical off-source field in the Rosette observed with FLAMINGOS can have an average of 6x 103 sources, almost 5 times higher than a field in Orion. To detect embedded populations in the Rosette we applied a density selection technique, the Nearest Neighbors Method (hereafter NNM), to distinguish populations with surface densities above the uniform field levels. This method has already been applied succesfuly to nearby clouds by Ferreira et al. (2005) and gives reasonable results for a large distance region like the Rosette. However, we improved the method with the use of a color selection to separate the youngest members in a field as near infrared excess (IRX) objects, assuring the detection of embedded clusters by increasing their probabilities of membership. We expect this combined approach to give a non-biased detection of young stellar groups and to give insight into their nature at the same time. Before describing our color selection, we will review briefly the terminology and concepts from the NNM relevant to this paper. A more detailed description of the general use of the method for embedded populations can be found at Ferreira et al. (2005). The calculation of nearest neighbor densities to detect clusters in a crowded field was proposed by Casertano & Hut (1985). They proposed to estimate the local surface density of objects in a certain field from the individual, relative surface density of each object. The generalized form of the individual density estimator is: j-1 j =- (4.1) where Dj is the distance of the star to its jth member. This estimator has only one degree of freedom, the number j of neighbors used to calculate the local density. Accord- ing to Casertano & Hut, the larger the value of j, the smaller the fluctuations in the local density estimations due to local irregularities, which is very useful to determine extension and structure of large systems. However, j also defines the minimum number of particles in the smallest substructure to be considered, and so j should be small if structures looser or smaller than typical clusters are to be detected. They showed that j = 6 was the mini- mum number at which fluctuations could be acceptable for populations of the order of 30 to 1000 particles. The NNM also allows the definition of a "density center" and a "density weighted radius" which define cluster centers and cluster cores, respectively. The density center is defined as the density weighted average of the star positions in a field: Xd = X( (4.2) -iPj(i) And the density or core radius, Roreis defined as the density weighted average of the distance of each star to X ,j: Rcore = -i djlPj(i) (4.3) Ci pj(i) Using the NNM, clusters can be detected as regions where average individual pj values are larger than those of a uniform control field, and ifj is small enough, groups of the order of N = 101 stars should be detected without bias. If we follow the definition of N 35 stars as a minimum number to represent a cluster (Lada & Lada, 2003; Adams & Myers, 2001), then any groups with less members could be considered loose enough as to account for a non-cluster (distributed) population. For example, Ferreira et al. (2005) applied a j = 20 estimator to 2MASS catalogs of nearby (d < 1 kpc) molecular clouds and confirmed locations and sizes of clusters with radii as small as 0.3 pc and total number of members as low as 205 stars. 4.2.2 Detection of Embedded Populations We calculated the 20th nearest neighbor densities for stars down to the completeness limit in our final RMC catalog and the control fields, expecting to be able to distinguish at least NGC 2244 and the seven clusters from PL971 as regions with densities unequiv- ocally higher than the field. The result was that NGC 2244 is indeed, very well traced as a high density region, as were the zone of clusters PL04 and PL05 in the core of the cloud. Clusters PL01, PL03, and PL07 were also distinguishable but their apparent ex- tensions were not more noteworthy than some groups of stars that rose above the 30 level only because they coincided with "patches" of low extinction around the main molecular gas emission. Finally, clusters PL02 and PL06, the smallest in the cloud, presented den- sities below the 30 level because their small number of members (around 30) resulted in lower than average 20th NN densities, and thus are not distinguishable among the noise levels of the distribution. 1 We use from now on the nomenclature "PL01" to "PL07" to refer to these clusters. Particularly, cluster PL06, which is associated with the B-type proto binary AFGL961, contains large quantities of obscuring material near its center, which makes difficult the detection of embedded faint members even in carefully constructed near- infrared maps (e.g Aspin, 1998); any clusters like PL06 will be difficult to detect in a large j scheme because their number of members will be intrinsically small. Given these difficulties, we repeated the NN analysis using only infrared excess (IRX) stars, to assure the detection of only the youngest component of the cloud embed- ded population, to minimize the contribution of the field and to promote the detection of deeply embedded, low surface density clusters. Infrared excess stars are not expected to exist in the control fields as they located away from the star forming clouds, and so the density of these objects should be always higher for embedded populations. Also, as the fraction of IRX stars in an embedded cluster is larger if the cluster is younger and therefore more embedded, so that high extinction regions could actually have higher IRX overdensities, improving detection. 4.2.3 Infrared Excess Stars In the J H vs. H K color-color diagram, IRX stars fall to the right of the reddening band defined by the projection of the Classic T Tauri star (CTTS) locus (Meyer et al., 1997), along the direction of the extinction vector (Cohen et al., 1981). The fraction of stars with infrared excess emission in a cluster is known to decrease with time as early stellar evolution leads to the destruction of disks by photospheric UV radiation, but for a deeply embedded population with ages of 1 to 2 Myr like the one expected in the RMC (the OB association NGC 2244 is estimated to have an age of 1.9 Myr and the embedded clusters cannot be older), the circumstellar disk fraction will be significant enough and IRX stars counted from JHK excess will trace well the presence of the most recent episode of formation (see e.g Lada et al., 1996; Carpenter et al., 1997). For our study, we define an IRX star as one with colors that place it 0.1 mag (5 times the standard deviation of the H K uncertainty) to the right of the ZAMS and above J H = 0.47(H K) + 0.46 which defines the lower limit of the Classic T Tauri Star (CTTS) locus of Meyer et al. (see Figure 4-1). 2.5gion 1 2.0 I / S,-K 0.0 0.5 1.0 1.5 2.0 2.5 1.0 Figure 4-1: The near-infrared color-color space. The thick dark solid lines represent the loci of the zero age main sequence and the giant branch (Bessell & Brett, 1988). The thick colored line is the Classic T-Tauri locus (Meyer et al., 1997), which in this diagram is extended to the right to large H K values and above and below by its observational error. The other dashed lines rep- resent extinction along the direction of the reddening vector indicated by the arrow on the left. The shadow region indicates where, under our definitions, stars with possible infrared excess emission fall. Stars falling in the regions labeled as 1 and 2 colors are usually affected by spurious detections and high photometric color scatter. The first constraint avoids contamination from non-IRX stars with large H K uncertainties located close to right edge of the Zero Age Main Sequence Reddening Band (ZAMSRB). The second constraint helps us to avoid including objects that locate in the region below the CTTS line. Sources fall in this region mainly due to high color scatter (see section 4.2.4 below), however unresolved galaxies with large color dispersions (Labb6 et al., 2003) or distant background galaxies with highly inclined reddening vectors (Heraudeau et al., 1996) may have colors that fall in this region of the diagram. For stellar sources, there are cases in which nebulosity can add a blue J H component to background stars with a low extinction vector and push stars to the region, as shown by Montecarlo simulations of Muench et al. (2001). 4.2.4 Magnitude Depth Restriction for IRX stars The combined effects of variable seeing quality over the seasons, and variability of the focus quality across the wide detector field of FLAMINGOS, resulted in a high intrinsic dispersion of color values for our sample, which cannot be eliminated with the zero point corrections or the uncertainty restrictions. In figure 4-2 we illustrate this effect in a contour level color-color diagram all of the stars in our working catalog. In the diagram, made with a nyquist box size of 0.1 mag, the lowest level shown represents the mean color-color space surface density and each subsequent level represents a step of 1 standard deviation. There is a noticeable bloating in the dispersion of colors at both sides of the ZAMSRB at the core of the diagram, near the regions of lowest extinction. The color dispersion is larger for faint stars. The scatter in H K has a major incre- ment at approximately K=15.75, where the field object density increases significantly. In Figure 4-3 we show equivalent contour level color-color diagrams made with separate samples for stars with K < 15.75 mag and for stars with 15.75 < K < 17.25 mag respec- tively. The diagrams show that the scatter for the bright end bins is smaller than for the faint end bins. Our calculations indicate that the standard deviation of the H K color uncertainties is twice as large for the faint end bins (0.11 vs. 0.21 mag). This effect is 11.0 -0.51 , -0.5 0.0 0.5 1.0 1.5 2.0 H-K Figure 4-2: Contour level color-color diagram for all stars in the FLAMINGOS RMC survey within the restrictions described in section 2.1. The diagram was con- structed using a sampling box of size 0.1 mag. The lowest level represents the mean value of the object counts, at 825 dex 2. Subsequent levels represent steps of 1 sigma (3550 dex2). slightly worse for some fields which were observed under less favorable weather con- ditions or at a higher than average airmass. There are also regions of the survey where the scatter is smaller and the quality of the colors is kept to fainter limits (see section 4.3.2). The statistical cuts we present are mostly conservative and assure the uniformity of our statistics across the entire survey area. I I I I I I I I I I I I I ''''' 1 1 1 1 1 1 / // ) I a) 11 v di i.n tw a o- o r n /B 2 0 2 0 - 1.0 1.0 0 5 0 5 - 0 0- 0 0 . -05 , I 0 5 -0.5 00 0.5 1.0 1 5 20 2.5 -0.5 00 0.5 1.0 1 5 20 2.5 H-K H-K Figure 4-3: Contour level color-color diagrams for stars in the FLAMINGOS RMC survey divided in two ample groups of brightness. Both diagrams were constructed with a Nyquist box size of 0.1 mag. The diagram a) shows the distribution of colors for 'bright' stars within 5.0 < K < 15.75 mag, and the diagram b) is for 'faint' stars within 15.75 < K < 17.25 mag. The contour levels start at the mean level (360 and 466 dex 2 for a) and b) respectively) with subsequent steps of 1 sigma (1770 and 1910 dex 2 for a) and b) respec- tively). The high scatter in near infrared colors affects directly the calculation of the number of IRX stars in the survey, which locate to the right of the ZAMSRB. We performed Montecarlo experiments in which we simulated the colors of stars drawn from a model population with an age of 1 Myr (D'Antona & Mazzitelli, 1997) located at the distance of the Rosette and we added to our simulated stars, color errors and extinction similar to those observed in the survey areas. We found that the resultant number of stars with colors similar to those of IRX stars was 5 times larger for stars in the faint end. Because we are basing our analysis in the detection of infrared excess sources, we had to limit the primary aspect of our analysis, the identification of embedded clusters, to those stars in the bright end of the sample to assure an IRX sample with a minimum of contamination. However, these bright IRX sources are only helping us to trace the location and rough extension of clusters, and although our color uncertainties are high, individual K band magnitudes are still good within 0.1 mag, which allows us later to include stars down to the HBL in the areas traced by the IRX sources and calculate correct luminosity functions with a generous bin resolution of 0.25 mag. Also, a photometric depth limit of K=15.75 means a sample still almost 1.5 mag- nitudes deeper than 2MASS and is equivalent (for dwarf type stars) to a stellar mass range of 0.09 to 0.18 M. for a population of 1 Myr embedded in a cloud with a typical extinction of 0 to 10 visual magnitudes (D'Antona & Mazzitelli, 1997). Thus, we should be able to count IRX populations slightly above the HBL for a typical young cluster. 4.2.5 Nearest Neighbor Analysis for Infrared Excess Stars A preliminary exploration using false color image combos of the less populated clusters, PL01, PL02 and PL06, indicated that the typical number of stars in a modest size cluster that can be detected "by eye" within areas of bright nebulosity, usually coincident with embedded cluster cores, could be of the order of 30 members. This is close to the minimum number that defines an association of stars to be a cluster, and below that, any groups could be considered a distributed population. The expected JHK infrared excess fraction in an embedded cluster less than 3 Myr old is 20 to 60 percent, so the number of IRX sources is much smaller than the number of sources in the full catalogs, and cluster have to be identified with less neighbors. Because of this, instead of j = 20 as in FL06, we selected a value of j = 10, which assures the detection of clusters with less than 20 IRX members (60 to 100 total members if the fraction of stars with circumstellar emission is 20 to 60%) but gives local surface densities with 15% more accuracy than the minimum j = 6 described by Casertano & Hut (1985). Also we are able to determine the existence of populations distributed in groups almost three times less dense than the minimum expected for a cluster. We detected a total of 116834 sources IRX sources under our definition in the bright end sample. In Figure 4-4 we show their 10th Nearest Neighbor distributions of RX 10th Ne Figure 4-4: Nearest Neighbor distributions for bright IRX stars. The top panel shows the distribution of 10th neighbor distances. The bottom panel is the distribution of 10th neighbor densities. In the top panel line A indicate the limit of dis- tances shorter than 1.0 pc, while the dashed line B indicates the midpoint value at 1.83 pc. At the bottom panel the equivalent limits in density space are also indicated. distances, Do1 and local surface densities p o. The mean value found for Do1 was 1.83 pc which corresponds to a pio = 0.2 (')2. This limit is indicated in Figure 4-4. For the control fields we found 3 sources that had IRX colors down to a maximum brightness of K = 15.75. However, we added another 16 sources which fall to the right of the reddening band below the CTTS line but would have IRX J H colors if reddened by an average value ofAv = 5.0 mag, typical of the cloud regions. The mean 10th neighbor density among these 19 sources was 0.18 (') 2 which compares relatively well with the mean value 0.2 (') 2 of the distribution of distances in the survey areas, so that we considered a round value of 0.2 (') 2 as a background limit, below which we cannot assure that an IRX source has a density high enough to be distinguished from the field. The minimum value of the Do1 distribution in the survey to be 0.145 pc, which represents a density of 29.5 (') 2 and is a good estimation of the typical local surface density in the central regions of RMC clusters. The midpoint between this minimum distance and the mean is 0.987 (roughly 1 pc) which corresponds to a p o value of approx. 0.6 (')2. This can be considered as a good estimate of the average embedded cluster size in the RMC. 4.2.6 Identification of Clusters In Figure 4-5 we show the location of IRX stars with the levels of density described above. All of the known clusters seem to be traced well in this selection, and we confirm that they are the main regions of star formation in the Rosette Complex. In Figure 4-6 we present a contour level plot of the local surface densities calculated with the NN method (j = 10). The contours were constructed using a Nyquist sampling box of 90". We define a cluster as a region for which a closed contour at 0.2 (') 2 con- tains at least 10 IRX sources. Using this definition we found, in addition to NGC 2244 and the seven PL97 clusters, 4 additional areas that arise as significant but have not been studied before: The first is a region at the "Core" of the cloud, to the east of cluster PL05 and south of cluster PL04. This region, which we designate as RLE08, contains a large number of highly reddened sources, differing from the clusters PL04 and PL05 which have a number of sources already visible in DSS plates and can be considered partially emerged. RLE08 appears to be a more recent episode of formation in this ample zone of formation at the 8 4.68( O U o 4.32( 3.96( 3.60( Figure 4-5: 99.0000 98.6800 98.3600 98.0400 97.7200 97.4000 Right Acension (J2000) FLAMINGOS GMC Survey. NOAO/UF The location of IRX stars in the Rosette survey with brightness K 15.75 mag. The "plus" symbols are IRX stars with 10th neighbor den- sities higher than 0.2 (') 2, while black dots are stars with densities below 0.2 (')2. We also indicate the expected position of the cen- ter of NGC 2244. Contours indicate levels of CO emission in steps of 20K km s-1. The dotted line indicates the limits of the survey coverage. center of the cloud, in which clusters PL04 and PL05 are the largest and most brilliant clusters. The second is a substantially large, highly reddened cluster located in the southeast- ern edge of the cloud, designated as RLE09. Along with RLE08, these clusters are clear examples of clusters which are located in regions with large extinction values and thus 5.40000 R LE1 5.04000- C C 2244 GC2 0V 0 8 4.68000 - N PL02 .2 0I0 SI PLO 3.60000 1 I I I 99.0000 98.6800 98.3600 98.0400 97.7200 97.4000 Right Acens-on (J2000) FLAMINGOS GMC Survey. NOAO/UF Figure 4-6: Identification of clusters in the Rosette Complex. The contours indicate 10th Nearest Neighbor densities and were constructed with a nyquist box size of 1.5 arcmin. Labels for individual clusters are explained in text. The dotted thin lines indicate the 15.0 KO km s level of 12CO emission, which we use to define the extension of the main molecular cloud regions. have average surface densities comparable or lower than the field. However, they contain a large number of red sources that are easily distinguishable in JHK false color composite images and such a large number of IRX sources that they stand out clearly as embedded clusters. A third new cluster is located to the east of NGC 2244, in the region of the cloud identified as NGC 2237, which is well known for its content of gas pillar structures (Carlqvist et al., 1998), and thus we assign it to that name. The existence of this cluster was also suggested by Li (2005) in their study of 2MASS data. NGC 2237 is distinguish- able as a zone of high surface density in all star Nearest Neighbor counts. Other patches in the fields that coincided with the Nebula areas also presented high densities but when we applied the IRX color selection, NGC 2237 was the only one -besides NGC 2244- that was confirmed to coincide with a cluster. A fourth group designated as RLE10 is located North of NGC 2244, and although it has a very low surface density compared to the rest of the clusters, it contains 13 IRX sources, from which at least six have colors suggestive of very large infrared excess, while the region has a small extinction value. 4.2.7 Properties of Clusters We analyzed each cluster individually, isolating appropriate sub-regions that varied roughly from 25 to 120 arcmin2 depending on the apparent extension of the clusters in the maps. This way we were able to determine cluster structures with total areas of 6 to 60 -2 arcmin2. For embedded clusters in the Molecular Cloud regions we calculated the extension of a cluster as the area Ap inside the polygon defined by the 0.2 (') 2 contour in each analysis box, and consider as potential members all of the sources (IRX and non IRX) down to K=17.25 inside it. Equivalent radii, Reqcan be defined as /Ap/7u and can be considered as standard estimates of the total extensions of clusters. In Table 4-1 we present for each cluster, its center coordinates (as defined from equation 2), core radii (equation 3) and equivalent radii. We also show the number of IRX sources to K < 15.75 and the corresponding fraction it represents. The IRX percentages in the nebula clusters NGC 2244 and NGC 2237 are intrin- sically smaller than those in the embedded clusters, roughly 10 vs 18-76 percent. This Table 4-1: Young Clusters Rosette Complex Cluster RA DEC Rcore Requiv NIRX VNIX a IRXF b ID center, J2000 [pc] K < 15.75 PL01 97.96 4.32 0.37 1.16 29-5 0.28 PL02 98.31 4.59 0.94 1.46 32-6 0.33 PL03 98.38 4.00 0.32 1.69 80-9 0.44 PL04 98.53 4.42 1.10 1.85 89-9 0.24 PL05 98.63 4.32 0.86 1.31 57-8 0.18 PL06 98.66 4.21 0.73 0.75 13-4 0.52 PL07 98.88 3.98 0.38 0.88 22-5 0.61 RLE08 98.56 4.32 0.99 1.30 49-7 0.33 RLE09 98.78 3.69 0.74 1.49 65-8 0.76 RLE10 97.78 5.27 1.19 1.15 15-4 0.32 NGC 2237 97.59 4.93 1.94 1.91 36-6 0.15 NGC 2244 97.95 4.94 1.56 2.30 62-8 0.12 aNumber of IRX stars with 10th Nearest Neighbor densities above 0.2 (')2. bIRX fraction with respect to total number of stars with K < 15.75 inside 0.2 (')2 contour. is due to an expected lower rate of disk survival in the presence of UV radiation from numerous OB stars (Dolan & Mathieu, 1999), as well as disk evolution in older stars which results in reduced circumstellar excess emission. For the clusters embedded in the Molecular Cloud areas (PL01-PL07, RLE08 and RLE09) the large IRX fractions are suggestive of ages of 1 to 1.5 Myr or younger (Haisch et al., 2001; Hillenbrand, 2006). The core radii, Reore of the Rosette clusters (see equation 4.3) have a range of 0.3 to 2.0, with an average of 0.930.48 pc. The equivalent radii, Req range from 0.75 to 2.30, with an average of 1.440.44 pc. The distributions of these size estimates are shown in Figure 4-7. We also show in the figure the distribution of the Rcore /Req ratios, which peak at 0.650.27 and have in two cases (clusters PL01 and PL03) values below 0.5. The clusters NGC 2244 and NGC 2237 are extended and their core radii are too close in value to their equivalent radii, so that we considered them equal. The distribution of core radii and core to total ratios is consistent with the study of Ferreira et al. (2005), and suggests that clusters, in most cases, have a tight center but with well extended edges. 65 4 _j 3 1-) 0 0 1 2 3 CORE 4 0 2 3 REQ 3 z2 0 0.5 1 1.5 2 RCORE/REQ Figure 4-7: From top to bottom: distribution of core radii, equivalent radii and core to equivalent radii ratios for the Rosette clusters We constructed color-magnitude and color-color diagrams, which we show in Fig- ures 4-8 to 4-19. In the K vs. H K color-magnitude diagrams we show the photometry for all of the stars inside the corresponding 0.2 (') 2 contour, and mark separately those with infrared excess. We include the ZAMS locus and a PMS evolution isochrone of 1 Myr, as well as extinction vectors corresponding to 3 times the mean value (Av) in the cluster analysis box. Stars falling to the right of the isochrone are affected by extinction towards the line of sight of the cluster, revealing their embedded nature. In the J H 66 vs. H K color-color diagrams, the same stars are located above the dwarf and giant sequences along the reddening bands, with the IRX sources located to the right of the MS reddening strip. Those objects located at or near the zero age sequences, which in the color-magnitude diagram locate preferentially to the left of the isochrone, are most probably foreground stars or evolved cloud members that coincide with the line of sight of the clusters. Ara (orcmin) 3 2 1 0 -1 -2 - I '.' I '' ' ''" ; ,' .'. . C .4 : o [ "' . 98.010 97.985 97.960 97.935 Right Acension () t" .t 97.910 A- Figure 4-8: a) K band image, b) control magnitude diagram, c) color color diagram and d) Radial Density Profile for the area corresponding to cluster PL01. See text for explanation. IRX - j Ara (arrcmin) 2 0 -2 *.. .'- . ** ^ W, "* " 98.37 98.34 98 31 98.28 98 25 Right Acens. on S* ., tl* 9 98.37 9834 9831 98.28 9825 Right Arension () Pt Figure 4-9: Same as Figure 4-8, for cluster PL02. The fourth plot in each panel are radial density distributions calculated with a method of equivalent areas (see e.g. Muench et al. (2003)) for all stars down to K = 17.25 inside the analysis boxes and calculated from the cluster centers. In this plots we indicate the core and equivalent radii calculated from the IRX Nearest Neighbor distri- butions. With the exception of PL02, PL06 and RLE10, which are the clusters with the lowest surface densities, the rest present well defined radial profiles, which unfortunately, due to poor statistics cannot be fit successfully to standard King or Plummer cluster models, but show well extended tails that in some cases (e.g clusters PL01, PL04, PL07, %o k 71,;r : Ara (orcmin) 4 2 0 2 4 -4 ." C I * *-., : .. .. ;, .. ; dI ., , .''. .' i . : 98,476 98 431 98.386 98.341 98.296 Right Acension (") .'f f . o '[a a ^^~~X s o I Figure 4-10: Same as Figure 4-8, for cluster PL03. RLE09) present well defined secondary bumps suggestive of structure. In the case of the nebula clusters NGC 2244 and NGC 2237 the radial distribution profiles show a slow decline that implies a negligible core peak, and might be suggestive of an extended structure. However, the counts in each of the equivalent areas used to construct these profiles are not corrected by background, and as the extinction is lower in the nebula, these profiles might be showing the effect of field contamination. - - . -, ,,- Ara (orcmin) 4 2 0 -2 4 4.489 I 4 4 1 4 ' "- I- i ^ -1 437 - U..'. ^". -4-4 4339 .., -, .. I 98.621 98 584 98.546 98.509 98.471 R ght Acension (a ) Figure 4-11: Same as Figure 4-8, for cluster PL04. 4.3 The Fraction of Stars in Clusters Under the assumption that the IRX are tracing the correct distribution of embedded populations in the Rosette complex, we can use them to estimate the fraction of stars that belong to clusters. We made our calculations inside the molecular cloud areas first, to account for deeply embedded clusters only, and then for the whole survey area which includes the emerged clusters located in the Nebula area. The total number of IRX stars detected in the survey is 116934, out of which 63025 stars have NN densities larger than the mean, 0.2 (')2. A total of 43621 stars 't.4.1. *i 'C-."6 * r^ Ara (orcmin) 4 2 0 2 4 4.388 .,- ..r .. ... .. .. .. A, :'V . 4353-- .... - S. t i2 4- -- 98,702 98 667 98.632 98 597 98.562 Right Acension (a ) -------- Figure 4-12: Same as Figure 4-8, for cluster PL05. are contained within the estimated areas of the 9 embedded clusters PL01 to RLE09, which occupy a total of 242 sq. arcmin. The remaining 53923 stars have local surface densities lower than the mean, and thus cannot be distinguished from the background field. The area of the molecular cloud covered by our survey was calculated as the one contained inside integrated intensity contour levels higher than 15 K km s 1. This area is equal to 2747 sq. arcmin, which means the clusters occupy roughly 9% of the cloud. Inside the molecular cloud areas we counted 12411 stars with densities lower ,- ....... , 3 2 98719 98 98.719 98 ara (orcmin) 0 1 2 3 -2 - '1 4 .. ,; 694 98.669 98.644 98.619 Right Acension (') Figure 4-13: Same as Figure 4-8, for cluster PL06. than the mean, and 437 are stars with densities larger than the mean but not associated with the cluster areas. For background correction purposes, we use a scale factor equal to the ratio of the non-cluster areas of the molecular cloud to the area of the control fields. Using this factor, we expect to see a total of 9410 field IRX stars, which leaves a total of 738 IRX sources in the cloud areas that are not associated with a cluster. From this, we estimate that the fraction of stars in clusters in the Rosette Molecular Cloud is 865%. ~ .. . r ; ~ i' r-, ? ~t~~ X Ara (orcmin) 3 2 1 0 1 -2 -3 4.037 .' '. ; I ** D .* - i 3937 .... I- - 98.932 98907 98.882 98.857 98.832 Right Acension (a) Figure 4-14: Same as Figure 4-8, for cluster PL07. If we repeat these estimates for the whole area of the FLAMINGOS RMC survey, 7308 sq. arcmin, we find that there is a total of 54923 sources associated with clusters after including NGC 2244, NGC 2237 and RLE10 in the counts. The clusters occupy a total area of 390 sq. arcmin, or 5.3% of the total survey areas. The number of IRX stars with densities lower than the mean is 53923, and there are 819 IRX stars with high densities but no association with a cluster. The scaled number of expected IRX field stars from the control fields is 26117 in this case, which results in a total of 35919 it~a, :P .. "8 ~ did 73 a) b) Ara (aricmin) 2 0 -2 438 .' *L .... I .o. ". ,.38 `L... - .2 *2 9 98 5 9 5 0 4: *. ." .'.. 1 . Right Acenslon () H-K c) d) F r 1 S a F e 8 o Radius from center (pc)8 0.0 0.5 2 . 10 A i n I t ihce dd clter popu s o c. T I a f ,* 4 : Q 0- 0 1 2 3 4 0 1 2 3 H-K Radius from center (arcmin) Figure 4-15: Same as Figure 4-8, for cluster RLE08. stars non-associated with clusters and a total fraction of 60+5% of stars associated with clusters in the whole survey. Another interesting result is that in the case of the embedded cluster population, 208+15 sources are contained in clusters PL04, PL05, PL06 and RLE08, at the "Central Core" of the cloud, which corresponds to 48+3% of the total number of embedded sources. This means that approximately half of the recent births in the RMC occurred at the most dense region of the cloud, which coincides with the main zone of interaction with the Nebula (Heyer et al., 2005). If the whole survey is considered, then the Central with the Nebula (Heyer et al., 2005). If the whole survey is considered, then the Central a) Arc (orcmin) 4 2 0 -2 -4 S- ., ,- .- ''.,: '. o , 98.871 98 826 98.781 98.736 98.691 Right Acension (") c* 98,871 98826 98.781 98.756 98.691 Right Acension (a) c) B",P XX. : : ~~~ , 1 Figure 4-16: Same as fig 4-8, for group RLE09. Core clusters plus the clusters in the Rosette Nebula, NGC 2244, NGC 2237 and RLE10, account for 563% of the recent stellar formation in the Rosette Complex, which suggests that the formation occurred in two main episodes which resulted in the generation of the biggest clusters, and then, a number of secondary episodes resulted in the smaller, remaining clusters which are distributed in the remaining areas of the Complex. x .. ... .. 0 1 2 3 4 H-K d) Radius from center (pc) .0 0.5 1.0 1.5 2,0 2.t \ l i ;. *:/ \, ( i 2 Arc (a-cmin) 2 0 2 P -- --- *... * .. '% .. .. _, t .." ;, :'. ."" "" *** *. .m ~.- *. .,' :: 9,.^ *'*.; U 4.., _-q .'* @',- 4. ' . . -I- I *. I 1 I I I *. .. I . . 97.81 97.79 97.77 97 75 97.73 97,71 9 Right Acension () x IRx ir. -1 .. ~ , i, 7.69 Figure 4-17: Same as Figure 4-8, for cluster RLE10. 4.3.1 Distribution of Sources with Respect to the Rosette Nebula It is important to mention that with the possible exception of RLE10, NGC2237 and NGC2244, all of the clusters are associated with a massive molecular clump, which confirms their deeply embedded stage. From this, it is clear that the Nebula and the Molecular Cloud areas expose different episodes of formation. Also, high density IRX stars in the RMC area are mostly confined to the limits of the cloud, while in the Nebula area, the stars that trace cluster populations are already exposed out from the molecular I I a) b) Ara (orcmin) 6 4 2 0 2 4 6 -. - C. E .. 14 - 2- S. 16 .X -ux IRX ... 97.745 97 708 97,672 97.635 97 598 97,562 97.525 0 1 2 3 4 Right Acension () H K c) d) Radius from center (pc) 0.0 0.5 1.0 1.5 2.0 2,5 5,0 4A j4- S........ I ......... I ......... I ......... II........ I ......... I 0 1 2 3 4 1 2 3 4 5 6 H-K Radius from center (arcmin) Figure 4-18: Same as Figure 4-8, for cluster NGC 2237. gas, possibly showing a more evolved population which evacuated most of the molecular material in the northern half of the complex. We calculated the distribution of IRX sources with densities higher than the mean as a function of the distance to the center of NGC 2244. To do this, we counted the number of IRX sources inside the central parsec of NGC 2244 (11 sources) and then we counted those outside this area in concentric annular wedge sectors with a constant width of 1.0 pc, but assuring that these sectors were always contained within the survey map areas. We scaled and normalized the star counts in each wedge area with respect to the area of the 77 a) b) Arac (aorcn)in) 6 4 2 0 2 4 6 | ., ,-, .- .-.. .- io:: o -.'. S ,ooo 14 .. : -,.. .... i --- 'I16 .. .. 4 98.060 98005 97950 97895 97.840 0 1 2 3 Right Acension () H-K c) d) Radius from center (pc) 00 0.5 1.0 1,5 A2. 2.5 3,0 t 100- Radus from center (pc) r in OO 0.5 1,0 1,5 1 5 -,0 Figure 4-19: Same as Figure 4-8, for cluster NGC 2244. first parsec circle. The result is shown in Figure 4-20, and we marked in the figure the approximate locations of clusters and main cores of the molecular cloud. *I I I 1I . 0 1 2 3 4 0 1 2 3 4 5 6 In the top panel of the figure, we see how the prominence of the Rosette Nebula clusters indicate they are the largest stellar groups in the complex. The Molecular Cloud "Rigde", where clusters PL01 and PL02 are located, and which is the part of the molecular cloud that is in direct contact with the ionization front from the Nebula, appears to be moderate in its star forming efficiency. The "Core" or central part of the cloud, which contains most of its mass and which has been suggested as the main region CLOUD CENTRAL CORE PL03 PL06 Figure 4-20: Top panel: Distribution of IRX stars with NN densities higher than 0.2 (1)2 as a function of distance from the center of the Rosette Nebula (NGC 2244). The counts are made in sectors of 1.0 pc in length and counts in each sector have been scaled and normalized to the area and counts in the central 1.0 pc circle in NGC 2244. Labels indicate the approximate locations of clusters described in this paper, as well as the main 'regions' of the complex. Bottom panel: equivalent distribution only for sources non associated with cluster areas. of interaction between the molecular and atomic hydrogen clouds (see Celnik, 1985; Cox et al., 1990), and where the clusters PL04, PL05, RLE08 and PL06 are located, seems to be carrying most of the cluster mode production, enhanced by the presence of cluster PL03, which is however, located in a separated sub-cloud but at the same radial distance. Clusters PL03, PL04, PL05, PL06 and RLE08 account fot 58% of the total cluster population. At the "Back Core" of the cloud, there are two clusters, PL07 and RLE09 which, al- though smaller than those in the central core, still have significant extensions. Particularly RLE09 has an extraordinary number of young sources despite its location well beyond the interaction front of the Nebula. For these clusters, it is possible that a mechanism different than triggering by interaction with the expanding HII region need to be proposed. In the bottom panel of Figure 4-20 we repeated the counting but only considering stars in the wedges that are located outside of the clusters. The sources not associated with clusters were defined as those located at least two cluster radii away from each cluster center. The scaled and normalized counts for these stars are of course much smaller but it can be seen that there are two major zones of along the wedge where non- cluster young sources accumulate: the first one is the area between the NGC 2244 and the "Cloud Ridge", and the second one is the region in between the cloud "Central" and "Back" cores. 4.3.2 A Case for a Distributed Population? Noticing how there is a significant number of sources not associated with clusters in the region between the central and back cores of the cloud, we used Field 09 of the survey for a separate analysis. This field lies precisely to the south of the cloud "Central Core" and north of the cloud "Back Core". The seeing and observing conditions for this field were particularly good, and 93% of the original area of the field was kept after the polynomial correction. The average scatter of colors down to K = 17.25 remains below an acceptable 0.109 mag across the whole field, probably because the southeastern quadrant, which for other fields presents high stellar profile distortions, overlaps in this case with the good quality northwestern quadrant of the Gap 1 field. The weighted averaging |