Thesis FIN.Pdf (3.768Mb)

Study of the Earliest Stages of Galactic Star Formation: BLAST Survey of the Vela Molecular Ridge

Daniel Angl´esAlc´azar

A thesis submitted in partial fulﬁllment of the requirements for the degree of

Master of Science in Physics

University of Puerto Rico, Rio Piedras Campus

2009

Program Authorized to Oﬀer Degree: Department of Physics University of Puerto Rico, Rio Piedras Campus Graduate School

This is to certify that I have examined this copy of a master’s thesis by

Daniel Angl´esAlc´azar

and have found that it is complete and satisfactory in all respects, and that any and all revisions required by the ﬁnal examining committee have been made.

Chair of the Supervisory Committee:

Professor Luca Olmi

Reading Committee:

Professor Daniel Altschuler

Professor Carmen Pantoja

Date: In presenting this thesis in partial fulﬁllment of the requirements for a master’s degree at the University of Puerto Rico, Rio Piedras Campus, I agree that the Library shall make its copies freely available for inspection. I further agree that extensive copying of this thesis is allowable only for scholarly purposes, consistent with “fair use” as prescribed in the U.S. Copyright Law. Any other reproduction for any purpose or by any means shall not be allowed without my written permission.

Signature

Date University of Puerto Rico, Rio Piedras Campus

Abstract

Study of the Earliest Stages of Galactic Star Formation: BLAST Survey of the Vela Molecular Ridge

Daniel Angl´esAlc´azar

Chair of the Supervisory Committee: Professor Luca Olmi Department of Physics

Stars form from the gravitational collapse of dense clouds of gas and dust in the interstellar molecular medium. However, little is known about the origin and evolution of these early stages of star formation. In this work, we present a detailed multi-wavelength characterization of a sample of these dense cores detected by the

Balloon-borne Large-Aperture Submillimeter Telescope (BLAST) in the Vela-D molecular cloud. Combining the BLAST photometry at 250, 350, and 500 µm with additional existing observations from millimeter to mid-infrared wavelengths, we have constrained the spectral energy distribution of 141 cores. Their physical parameters have been estimated assuming an isothermal modiﬁed blackbody model. In addition, associations of BLAST cores with mid-infrared sources allow us to separate starless from proto-stellar cores. We ﬁnd that proto-stellar cores are characterized by higher luminosities and temperatures than starless cores, indicating an evolutionary sequence. TABLE OF CONTENTS

Page

List of Figures ...... iii

Glossary ...... v

Introduction ...... x

Chapter 1: Star Formation ...... 1 1.1 Introduction ...... 1 1.2 The Interstellar Medium ...... 4 1.3 From Molecular Clouds to Pre-Stellar Cores ...... 6 1.4 The Observation of Dense Cores ...... 10

Chapter 2: Submillimeter Observations and the BLAST Telescope ...... 14 2.1 Methods in Submillimeter Astronomy ...... 14 2.2 The BLAST Telescope ...... 23

Chapter 3: Observations ...... 34 3.1 The Vela Molecular Ridge ...... 34 3.2 BLAST Data ...... 40 3.3 Archive Data ...... 45

Chapter 4: Analysis ...... 54 4.1 Overview of the Analysis ...... 54 4.2 1200–8 µm Source Identiﬁcation ...... 59 4.3 Source Photometry ...... 64 4.4 Spectral Energy Distribution of the Dense Cores ...... 67 4.5 Bonnor-Ebert Masses ...... 73

i Chapter 5: Results ...... 76 5.1 SEDs and Morphology of the Cores ...... 76 5.2 Separating Starless and Proto-Stellar Cores ...... 78 5.3 Distribution of Physical Parameters ...... 83 5.4 Do Starless and Proto-Stellar Cores Have Diﬀerent Properties? . . . . . 91 5.5 Mass Spectrum ...... 94 5.6 Dynamical State of the Cores ...... 97 5.7 Summary and Conclusions ...... 100

Bibliography ...... 102

ii LIST OF FIGURES

Figure Number Page 1.1 Evolutionary sequence of low-mass star formation ...... 3 2.1 Atmospheric transmission ...... 16 2.2 Diﬀraction from a circular aperture ...... 19 2.3 Schematic of a bolometer ...... 20 2.4 BLAST broadband ﬁlters ...... 25 2.5 Model of the BLAST06 gondola ...... 26 2.6 Schematic of the BLAST gondola ...... 28 2.7 Spectral energy distribution of the BLAST06 calibrator ...... 31 2.8 BLAST06 point spread functions ...... 31 3.1 12CO map of the VMR ...... 35 3.2 The VMR and the Vela SNR ...... 36 3.3 BLAST image of the VMR ...... 41 3.4 BLAST 250 µm image of Vela-D ...... 43 3.5 Size distribution of the BLAST cores ...... 45 3.6 SIMBA map of Vela-D ...... 47 3.7 Color-composite image of Vela-D at 250, 70, and 24 µm ...... 51 3.8 IRAC map of Vela-D ...... 53 5.1 SED and thumbnails of a warm core ...... 77 5.2 SED and thumbnails of a cold core ...... 77 5.3 Thumbnails of starless/pre-stellar cores ...... 82 5.4 Distributions of physical parameters ...... 85 5.5 Color–color plot ...... 87 5.6 Luminosity–mass plot ...... 89 5.7 Spatial distribution of starless and proto-stellar cores ...... 90

iii 5.8 Mass spectrum ...... 95 5.9 Bonnor-Ebert masses ...... 98

iv GLOSSARY

BE: Bonnor-Ebert

BLAST: Ballooon-borne Large Aperture Submillimeter Telescope.

CMF: Core Mass Function.

DC: Direct Current.

EM: Electromagnetic.

FIR: Far-Infrared.

FWHM: Full Width at Half Maximum.

GMC: Giant Molecular Cloud.

GPS: Global Positioning System.

IGA: IRAS Galaxy Atlas.

IMF: Initial Mass Function.

v IR: Infrared.

IRAC: Infrared Array Camera.

IRAM: Institut de Radio Astronomie Millimetrique.

IRAS: Infrared Astronomical Satellite.

ISM: Interstellar Medium.

ISO: Infrared Space Observatory.

ISSA: IRAS Sky Survey Atlas.

LDB: Long Duration Balloon.

MAMBO: Max-Planck Millimeter Bolometer.

2MASS: Two Micron All Sky Survey.

MIPS: Multiband Imaging Photometer for Spitzer.

MIR: Mid-infrared.

MSX: Midcourse Space Experiment.

NIR: Near-Infrared.

vi PAH: Polycylic Aromatic Hydrocarbon.

PSC: Point Source Catalogue.

PSF: Point Spread Function.

SANEPIC: Signal and Noise Estimation Procedure Correlations.

SCUBA: Submillimetre Common-User Bolometer Array.

SED: Spectral Energy Distribution.

SEST: Swedish ESO Submillimeter Telescope.

SHARC-II: Submillimeter High Angular Resolution Camera.

SIMBA: SEST Imaging Bolometer Array.

SNR: Supernova Remnant.

SWIRE: Spitzer Wide-area Infrared Extragalactic Survey.

UV: Ultraviolet Radiation.

VMR: Vela Molecular Ridge.

YSO: Young Stellar Object.

vii ACKNOWLEDGMENTS

First of all, I would like to acknowledge my thesis advisor, Luca Olmi, for his continuous guidance, support, and patient during these two years. I also would like to thank my advisor and my reading committee members, Daniel Altschuler and Carmen

Pantoja, for their helpful suggestions and corrections for the completion of this thesis.

I am very grateful to Gerardo Morell and the NASA Puerto Rico Space Grant Con- sortium for their continuous support. I also would like to acknowledge the graduate students, faculty, and staﬀ of the Department of Physics for their thoughtful assis- tance. In particular, thanks to Jorge Morales for his companionship during hundreds of hours in front of a computer.

I want to thank my friends from Torre Norte: Baldomero, Victorius, Josete, Ruben,

Pascal, Fumero, Noel, and Julissa for an unforgettable ﬁrst year in Puerto Rico (thanks to them I could spend the rest of the time researching). I also would like to thank my family for all their personal support, and Jos´e,Fanny, “los gordillos”, and Colibri, for being my family here.

Finally, the most special thanks go to Julissa, my wife, for her unconditional support and because she makes me happy every day.

viii A mi princesita...

ix INTRODUCTION

Star formation is one of the most important research areas in modern Astrophysics.

Understanding the formation of stars, particularly in its earliest phases, is still a fundamental and partially unsolved problem, which also has profound consequences on both planet formation and the physics of galaxies. A considerable theoretical and observational eﬀort is currently being done in order to understand the initial conditions and mechanisms leading to the formation of stars from the gas and dust in the interstellar medium. Star form out of huge volumes of gas known as molecular clouds, composed predominantly by molecular hydrogen, H2. Along with their gas, molecular clouds also contain an admixture of small solid particles, known as interstellar dust grains. These particles eﬃciently absorb visible and infrared light and reradiate this energy at longer wavelengths, particularly in the submillimeter regime (∼ 200–

1000 µm). Molecules are imperfect tracers because they undergo complex physical and chemical changes. In contrast, dust is stable in most phases of the interstellar medium and its emission at submillimeter wavelengths gives a direct measure of the mass content along the line of sight. Ground-based observations in the submillimeter regime are partially or completely limited by the atmospheric absorption and its

x rapid changes in time and along diﬀerent line of sights. Cold dust emission is observed much more eﬃciently from telescopes on (sub)orbital platforms, above all or most of the atmosphere. Recently, the Balloon-borne Large-Aperture Submillimeter Telescope

(BLAST) has obtained the most sensitive large-scale maps of the sky at 250, 350, and

500 µm to date.

The aim of this thesis is to exploit the submillimeter survey performed by BLAST towards the Vela constellation in order to investigate the physical properties of the earliest stages of star formation. In this work we make use of additional existing observations from mid-infrared to millimeter wavelengths (including SEST-SIMBA,

IRAS, AKARI, Spitzer MIPS, IRAC, and MSX data) to properly characterize the dense cores detected by BLAST. This multi-wavelength approach is needed given the wide range of scales and physical conditions in which star formation occurs.

In Chapter 1 we introduce the current accepted model for the formation of low- and intermediate-mass stars. Then, we discuss in more depth some fundamental aspects regarding the study of the earliest stages of star formation, from giant molecular clouds to pre-stellar cores. Some observational aspects of submillimeter Astronomy are presented in Chapter 2, where we also present a general description of the BLAST telescope. In Chapter 3 we introduce the region analyzed in this work, the Vela-D cloud. Then, the speciﬁc observations performed towards this cloud are described in details, with special emphasis on the maps and source catalogs used in this work. An

xi overview of the data analysis is given in Chapter 4. Here, we describe the speciﬁc methods used to correlate the information at diﬀerent wavelengths, constraining the spectral energy distribution of the dense cores, and estimating their physical parameters. Finally, we discuss the results of this work and summarize our conclusions in

Chapter 5.

This thesis is based on material not yet published. Part of the work presented here will appear in Netterﬁeld et al. (2009) and Olmi et al. (2009).

xii 1

Chapter 1 STAR FORMATION

Over the past two decades, a reasonably robust evolutionary sequence has been es- tablished and widely accepted for the formation and evolution of low- and intermediate- mass stars (M ≤ 8 M¯) within molecular clouds (e.g., Shu et al. 1987, 2004; McKee and Ostriker 2007, and references therein). However, several fundamental questions related to the earliest stages of star formation and the formation of high-mass stars remain unanswered. The formation of high-mass stars is not a simple scaled-up version of low-mass star formation, it is subject to further theoretical and observational diﬃculties (Zinnecker and Yorke 2007), and will not be speciﬁcally discussed here.

1.1 Introduction

Stars form from the gravitational collapse of dense cores in molecular clouds. In the current accepted scenario of low-mass star formation, the fragmentation of the parent cloud into cold, dense condensations leads to the formation of gravitationally-bound cores, which are initially supported against gravity by a combination of thermal, turbulent, and magnetic pressures. Subsequently, these pre-stellar cores/condensations become gravitationally unstable and evolve toward the formation of an accreting pro- 2

tostar: a central compact object in hydrostatic equilibrium surrounded by an accretion disk and embedded within an infalling envelope of dust and gas. While the central compact object continues accreting material, the gradual dissipation of the disk and envelop characterize the evolutionary phase of the system.

Fig. 1.1 shows schematically the theoretical evolutionary sequence of low-mass star formation, from pre-stellar to proto-stellar and ﬁnally pre-main sequence phases.

Different stages are observationally characterized by different shapes of the spectral energy distribution (SED, see section 2.1.4), and the bolometric temperature (the temperature of a blackbody whose spectrum has the same mean frequency as the observed SED). Dense cores are observable at millimeter and submillimeter wavelengths, but are opaque in the near-infrared (NIR) and mid-infrared (MIR) bands (e.g. Alves et al. 2001). Proto-stellar and pre-main sequence objects are classified according to the relative mass distribution among the envelope, the circumstellar disk, and the central compact object. Observationally, we distinguish among Class 0 to Class III protostars, from less to more evolved stages (see Lada and Wilking 1984; Andre et al.

1993, 2000). Previous IR studies using IRAS, MSX, and ISO satellites in addition to ground-based NIR surveys have provided a complete census of Class I–III sources (e.g.,

Bontemps et al. 2001). However, only a few bona-ﬁde Class 0 protostars are known to date. Observations at longer wavelengths are essential to detect and characterize even earlier evolutionary stages. 3

Figure 1.1: Theoretical evolutionary sequence of low-mass star formation and the observational classiﬁcation scheme (De Luca 2008). 4

1.2 The Interstellar Medium

The interstellar medium (ISM) is a mixture of ∼ 99 % gas and ∼ 1 % dust by mass which permeates the space between stars. Interstellar hydrogen amounts ∼ 89 % of the gas content in the ISM1 and it is found in a variety of chemical forms, temperatures, and densities which characterize diﬀerent phases coexisting in the ISM (see, e.g.,

Stahler and Palla 2004).

The largest portion of Galactic mass (∼ 34 % in volume) is in the form of (neutral) atomic hydrogen gas, observable in the 21 cm radio spectral line. This neutral gas is believed to be composed of two different phases in pressure equilibrium, diffuse cold clouds (called HI regions; T ∼ 80 K) confined by a surrounding, hotter (∼ 8 ×

103 K), and rareﬁed intercloud medium. In addition, it is thought that ∼ 65 % of the

Galaxy volume is ﬁlled with a hot and extremely hot (∼ 8 × 103 K and ∼ 5 × 105 K respectively) low density ionized gas primarily originated from supernova explosions.

Thus, the physical conditions of neutral and ionized phases of the ISM are far from ideal regarding the formation of stars which therefore must be restricted to particular regions of the ISM.

The appropriate conditions for the gravitational collapse of interstellar gas into stars are found in the molecular phase of the ISM, characterized by higher densities

(> 300 particles/cm3) and very low temperatures (∼ 10 K) compared to the neutral and ionized phases (molecules facilitate the gravitational collapse emitting excess grav-

1Percentage by number of nuclei; ∼ 9 % is helium and ∼ 2 % heavier elements. 5

itational energy as electromagnetic radiation in a process called “cooling radiation”).

Molecular clouds comprise a signiﬁcant fraction of the mass content in the Galaxy in ∼ 1 % of its volume and, unlike other phases of the ISM, they are not necessarily in pressure equilibrium with the environment since self-gravity governs the cloud’s mechanical equilibrium.

Molecular hydrogen (H2), being the principal component of molecular clouds, is an homonuclear diatomic molecule and therefore its rotational transitions are forbid- den. Thus, molecular clouds are usually observed in radio rotational lines of tracers such as CO, OH, and other molecules, providing information about the gas spatial distribution from line integrated intensity maps, and its dynamical state from velocity gradients determined from doppler shifts. Furthermore, molecular clouds are also indirectly visible in optical and NIR images as obscured regions in rich stellar

ﬁelds. Dust grains, though much less abundant than gas, are responsible for most of the interstellar extinction and reddening of the optical/NIR light from background stars. Remarkably, this eﬀect is commonly used to obtain extended maps of molecular clouds. In addition, dust grains, which in the regions of interest to us can be considered in thermal equilibrium, re-radiate the absorbed optical/NIR emission at longer wavelengths. Therefore, with a characteristic Black-Body continuous spectrum, the

−1 peak of the dust emission occurs at λmax [cm] = 0.29 × T [K], which corresponds to the submillimeter band at the typical temperatures in molecular clouds. 6

Cloud Type M n T

−3 [M¯] [cm ] [K]

GMC ∼ 104 − 105 ∼ 102 ∼ 15

Clump ∼ 102 − 103 ∼ 103 − 104 ∼ 10 − 20

Core ∼ 1 − 10 ∼ 104 − 105 ∼ 10 − 30

Table 1.1: Physical properties of molecular clouds (adapted from Stahler and Palla 2004).

1.3 From Molecular Clouds to Pre-Stellar Cores

Molecular clouds, often characterized by a complex morphology, are found in a wide range of sizes and masses. For practical reasons, they can be separated into diﬀerent types which are summarized in Table 1.1.

Giant molecular clouds (GMC) are the largest coherent structures in the Galaxy,

5 with sizes ∼ 50 pc, masses ∼ 10 M¯, and containing ∼ 80 % of the molecular hydrogen. These large reservoirs of gas and dust, preferentially located within Galactic arms, are closely related to the formation of massive stars. Actually, every observed

Galactic OB association (group of stars with members of M > 8 M¯) seems to be physically associated with a GMC. Direct evidence is found in the habitual presence of HII regions in GMCs2. Furthermore, massive stars play a crucial role in the evolu-

2Regions of ionized hydrogen (HII) within GMCs are produced by the ultraviolet (UV) radiation emitted from young massive stars. They are observed in radio continuum emission (from free-free electron transitions) and hydrogen line emission (from electron-ion recombination, including the optical Balmer series of hydrogen). 7

tion of GMCs, and it is thought that the intense winds powered by embedded O and

B stars constrain their lifetimes to ∼ 3 × 107 years (Stahler and Palla 2004).

The origin and evolution of GMCs is still not well understood. Various molecular tracers probe regions with diﬀerent densities, revealing a complex internal structure and dynamics. They appear to be composed of large fragments with masses

2 3 ∼ 10 −10 M¯ and sizes ∼ 0.5 pc, called clumps, connected by ﬁlaments, and smaller, denser sub-units called cores, with typical masses ∼ 1−10 M¯ and sizes ∼ 0.1 pc. The distinction between cores and clumps is often related to diﬀerent modes of star formation: individual cores lead to the formation of one or few stars (isolated star formation) while clumps are related to the formation of clusters of stars (cluster star formation).

However, we note that there is no net distinction between them; in fact, observations with enough resolution may separate clumps into individual cores. Ultimately, the detection of infrared (IR) point sources driving jets and outﬂows inside dense cores suggests that these are the actual sites of ongoing star formation. The formation and evolution of dense cores in molecular clouds is, thus, critical for understanding the earliest stages of star formation.

The virial theorem may be used to estimate the initial conditions needed for the gravitational collapse of clouds. The minimum mass necessary for a spherical homogeneous cloud with temperature, T , and density, ρ, to initiate a spontaneous collapse is known as the Jeans mass: µ ¶ µ ¶ 5kT 3/2 3 1/2 Mcloud > MJeans = , (1.1) GµmH 4πρ0 8

where, G and k are the gravitational and Boltzmann constants respectively, µ is the mean molecular weight, and mH the mass of the hydrogen atom. Assuming an isothermal contraction (i.e., the gravitational energy released during the collapse is radiated away eﬃciently), the cloud would collapse in a characteristic free-fall time independent of the initial size of the cloud: µ ¶ 3π 1 1/2 t = . (1.2) ff 32 Gρ

Given the mean density and temperature in a typical GMC, equation 1.1 implies that the entire cloud should be collapsing. The internal structure observed in GMCs, however, does not support the idea of large scale collapsing of clouds, suggesting that initial inhomogeneities in density could originate a gravitational fragmentation into clumps and cores, which would individually satisfy the Jeans mass limit. In addition, molecular line observations show that the clump velocities within GMCs appear to be random and not directed toward a collapsing center. Molecular clouds, thus, seem to be in approximate force balance over their lifetimes, supported against gravity by additional mechanisms besides thermal pressure, namely, turbulent and magnetic pressures.

The clump mean velocities are comparable to the virial velocity, the typical speed of masses under the influence of gravitational field, and thus, the associated kinetic energy accounts for a significant fraction of the gravitational energy and contribute to the support of the cloud against collapse. This internal velocity field, along with characteristic filamentary structures observed in GMCs, can be explained as a con- 9

sequence of large scale supersonic turbulence. In addition, turbulent motion could also generate large density ﬂuctuations on small scales, some of which could become gravitationally unstable and collapse to form stars. Thus, turbulence provides an al- ternative model to the purely gravitational fragmentation and collapse of clouds into clumps and cores. Furthermore, it is known that large scale magnetic ﬁelds pervade

GMCs and could play a crucial role in the formation and evolution of sub-structures.

Magnetic field can slow or even prevent the gravitational collapse of clouds by a process known as ambipolar diffusion: the ionized fraction of the cloud remains coupled to the magnetic field and the cloud contraction may be opposed by collisions between ions and neutrals.

The relative importance of self-gravity, turbulence, and magnetic ﬁelds in shaping the clouds structure is currently a matter of debate (Andr´eet al. 2008). It is thought that some combination of these factors govern not only the formation of dense cores but also their evolution, which could be considered, in some sense, a process analogous to the evolution of molecular clouds. Once a dense core has been formed, its evolution depends on the relative strength of the thermal, turbulent, and magnetic pressures against the cohesive gravitational force.

There are currently two major open issues that still need theoretical and observational work: (i) whether core formation is initiated by gravitational fragmentation or turbulent fragmentation; and (ii) whether magnetic ﬁelds control core evolution or not. Several theoretical models have been proposed for the formation and evolution 10

of dense cores in molecular clouds. In summary, we can consider two extreme models to be compared with observations:

• A slow, quasi-static evolution in which cores form by gravitational fragmenta-

tion in magnetically-supported clouds (Shu et al. 1987; Mouschovias and Ciolek

1999). The gradual core collapse is mediated by magnetic ﬁelds and dissipation

of low-level turbulence.

• A dynamic process (known as fast star formation), in which supersonic turbu-

lence creates large density inhomogeneities which may become gravitationally

unstable and collapse to form stars (Ward-Thompson 2002). Magnetic ﬁelds

may further contribute to the dissipation of excess turbulent energy.

As we will see later, the observation and characterization of statistically-signiﬁcant samples of dense cores is fundamental to constrain the diﬀerent star formation models.

1.4 The Observation of Dense Cores

The observational identiﬁcation of dense cores has been carried out primarily by molecular and millimeter/submillimeter dust continuum surveys. Molecular transitions that

18 trace relatively high density gas, such as C O(1–0) and NH3, provide information about the dynamics of the system. However, the interpretation of line integrated intensity maps can be complicated by variations in molecular abundance and chemical evolution within clouds and cores. Moreover, the sensitivity of line observations 11

critically depends on the gas density: a given molecular line is most sensitive above the critical excitation density and below a saturation limit imposed by the increasing optical depth. In contrast, millimeter/submillimeter continuum emission maps can trace the column density (dust content along the line of sight, commonly measured in units of mass per area) of dense cores without the confusing eﬀects of line opacity and chemical evolution. Therefore, though the estimation of core physical parameters is still subject to certain assumptions (see section 4.4), dust continuum surveys are vitally important for probing the nature of these objects. In particular, they represent the only way of deriving statistically-signiﬁcant samples of pre-stellar cores.

Dense cores can be separated by observational criteria into three broad categories

(e.g., Di Francesco et al. 2007; Ward-Thompson et al. 2007): starless, pre-stellar, and proto-stellar cores. Starless cores are deﬁned as concentrations of molecular gas without any evidence of star formation, such as embedded young stellar objects (YSO), molecular outﬂows, or compact radio emission. They are possibly transient objects supported by external pressure. Pre-stellar cores are the subset of starless cores which are gravitationally bound and are expected to form stars. Finally, dense cores which are actually forming stars and are therefore associated with signs of star formation, are commonly known as proto-stellar cores. Such cores with an embedded YSO can be viewed actually as the surrounding envelope of an accreting protostar.

In practice, the observational distinction of cores among diﬀerent categories is not generally straightforward, and we note that there is no sharp transition between them, 12

but rather a continuous evolution from pre-stellar to proto-stellar cores. The distinction between starless and proto-stellar cores is ultimately dependent on the availability and sensitivity of ancillary data. All-sky IR surveys, traditionally far-infrared (FIR) data from IRAS (Beichman et al. 1986) and more recently MIR and NIR data from

MSX and 2MASS respectively, have been used to find evidence of embedded YSOs associated with previously detected cores. However, deep NIR observations towards specific cores and the high sensitivity of the Spitzer Space Telescope have shown the inability of previous all-sky surveys to detect the youngest and most embedded protostars. Thus, the ratio of the number of proto-stellar to starless cores usually increases with the sensitivity of the additional IR data used in the classification. In addition, the observation of molecular outflows and jets may be used to identify proto-stellar cores. On the other hand, molecular line profiles can be used to detect inward mo- tions in cores and therefore distinguish between pre-stellar and starless cores, since they would provide direct evidence of gravitational binding. An analysis of the core stability can also be performed using the Bonnor-Ebert criterion (Bonnor 1956), which will be described later in section 4.5.

The dense, dust cores observed with submillimeter telescopes represent a critical link between the evolution of GMCs and the formation of stars. The identiﬁcation and characterization of these cores provide valuable information about the physical conditions prior to star formation. Furthermore, comparing the physical parameters

(temperature, mass, luminosity) of starless, pre-stellar, and proto-stellar cores may 13

help to characterize an evolutionary sequence which, in addition, may help to distinguish among diﬀerent star formation models. In particular, there are some key parameters which can be directly used to discriminate between the two extreme models mentioned above: the characteristic timescale of core evolution, the mass distribution of pre-stellar cores, and the star formation rate and eﬃciency. 14

Chapter 2 SUBMILLIMETER OBSERVATIONS AND THE BLAST TELESCOPE

The submillimeter band is the region of the electromagnetic (EM) spectrum lying between the infrared and microwave bands, and may be deﬁned as the wavelength range between 200 µm and 1000 µm. Submillimeter observations of spectral lines and dust continuum emission contribute signiﬁcantly to many areas of astronomy, from planetary science to cosmology. In particular, during the past ten years, the study of star formation has been revolutionized with the advent of telescopes and cameras able to map the cold dust component of molecular clouds, providing direct evidence of dense cores and the youngest protostars. In this chapter, we introduce some basic concepts in submillimeter astronomy and present a general description of the BLAST telescope.

2.1 Methods in Submillimeter Astronomy

Despite its clear importance in a wide range of astrophysical contexts, submillimeter astronomy has been the least explored portion of the EM spectrum. Even today, the submillimeter sky has not been yet systematically surveyed, because of the high absorption in the atmosphere at these wavelengths and the technical complexity of the 15

required instrumentation. Between the IR and radio regimes, submillimeter astronomy makes use of techniques from both IR and radio astronomy.

2.1.1 The Submillimeter Regime

As we have seen in the previous chapter, the very ﬁrst stages of star formation are invisible at optical and NIR wavelengths. At the typical densities of pre-stellar cores, all of the NIR-optical-UV light emitted by the forming star is completely absorbed by the surrounding dust, which is heated to temperatures ∼ 10–30 K. At these temperatures, the bulk of the emission is radiated in the submillimeter band which, in contrast to optical and IR wavelengths, remains optically thin even in the densest regions of molecular clouds. Therefore, submillimeter dust continuum observations can probe even the center of dense cores, providing a direct measurement of the dust content along the line of sight (i.e., the column density).

The atmospheric absorption of submillimeter radiation is mainly due to water vapor and therefore ground-based observatories are built at the highest and driest possible sites (the Atacama desert in Chile and Antarctica are currently the best places on Earth). Fig. 2.1 shows a model of atmospheric transmission under good observing conditions at an altitude of ∼ 4000 m. Ground-based observations at submillimeter wavelengths are possible only through atmospheric windows of partial transparency centered on 350, 450, and 850 µm. However, in the wavelength range 200–600 µm the atmospheric transmission is ≤ 40 % even under the best observing conditions. 16

Figure 2.1: Zenith atmospheric transmission at the Caltech Submillimeter Observa- tory (Mauna Kea, Hawaii) for 1 mm precipitable H2O (Weisstein 1996).

Dust continuum surveys at ≥ 1.2 mm suﬀer less from atmospheric absorption but they probe the Rayleigh-Jeans tail of the SED of the dense cores, far from the peak of the emission. Therefore, millimeter/submillimeters surveys have been considerably limited in sensitivity and/or spatial coverage to date.

The full submillimeter band can only be observed from space, avoiding the absorption of the atmosphere. However, the design, construction, and operation of space telescopes imply huge budgets and long timescales. A cheaper and shorter-timescale solution can be found in balloon-borne telescopes. Helium-filled balloons of large volume can lift telescopes above most of the atmosphere, near space-like conditions, during flights of ≤ 2 weeks duration, called Long Duration Balloon (LDB) flights. 17

2.1.2 Telescope Performance

The telescope performance depends primarily on the diameter of the main reﬂector

D divided by the wavelength of the observed emission, λ , also known as the electric diameter. A large electric diameter improves the sensitivity of the telescope (proportional to D2) and its angular resolution (see below). However, a larger electric diameter demands a better pointing accuracy and a more accurate reflector surface, which are made difficult by the effects of gravity, wind forces, and temperature gradients (see Olmi 2003, for a review).

Submillimeter telescopes have the appearance of radio telescopes though some of them are sheltered by an enclosure similar to optical telescopes (in order to reduce wind and temperature gradients). Due to the relatively short wavelengths, the primary mirror must be built to very high accuracy. The design of most submillimeter telescopes is based on a Cassegrain configuration, in which a hyperbolic secondary mirror direct the light collected by a parabolic primary mirror through a hole in the later. In the Ritchey-Chrétienversion, a hyperbolic primary mirror substitutes the parabolic mirror in order to correct for spherical aberration at a flat focal plane, suitable for wide field observations.

In submillimeter telescopes it is very important to block stray radiation, i.e., radiation propagated and/or generated by a non-ideal optical system which does not come from the object of interest (the field-of-view of the telescope). Stray radiation may be diffracted or scattered radiation originated from sources outside the field-of-view and, 18

even more important, thermal emission of objects inside or outside the optical system.

At submillimeter wavelengths, the thermal emission radiated by any component of the telescope at ambient temperature may contribute to the detectors response far more than the total radiation coming from an astronomical source. Thus, every component in the detectors ﬁeld of view (and the detectors themselves) must be cooled to very low temperatures (∼ 0.3 K) for correct operation in the submillimeter regime. The cryogenic system is therefore an essential element in submillimeter instruments.

The angular resolution (the ability to separate nearby sources) of submillimeter telescopes is limited by diffraction (assuming that the image of a point source is affected only by diffraction and not by geometrical aberrations, or equivalently that the optical system is “diffraction-limited”). The beam of the telescope is defined as the portion of the sky that a single detector in the focal plane can “see”. Therefore, two adjacent sources falling within the beam of the telescope would be unresolved. Due to the limited diameter of the telescope, the image of a point source (an unresolved source whose angular size is smaller than the beam) resembles a typical Airy diffraction pattern (see Fig. 2.2). The response of a telescope to a point source, in particular the x ,y intensity distribution in the image, is called the point spread function (PSF).

A good knowledge of the telescope’s PSF is essential in order to interpret correctly astronomical images. 19

Figure 2.2: Left panel: Diffraction pattern produced for light passing through a circular aperture (“Airy pattern”). The first minimum determines the angular resolution λ of the instrument: θmin = 1.22 D , where θmin is the angular displacement from the centre (expressed in radians), λ is the observed wavelength, and D is the diameter of the telescope. Right panel: A point source will appear as an extended image with the total emission distributed according to the diffraction pattern and the size of the recon- structed image pixels. Taken from http://laser.physics.sunysb.edu/marissa/report.

2.1.3 Detectors Used in Submillimeter Astronomy

In general, there are two different types of detectors of EM radiation, which are related to the wave-particle duality of the EM field. Coherent detectors, commonly used in radio astronomy, are sensitive to the wave nature of the EM field and detect

ﬂuctuations in the amplitude and phase of the radiation. In addition, they record the spectral information of the astronomical signal. Therefore, coherent detectors are very useful to generate interferometric signals1 and obtain the spectrum of astronomical sources. In contrast, incoherent detectors are sensitive to the particle nature of the

EM ﬁeld, and can measure only the intensity of the EM radiation which is a measure of

1Interferometry combines the phase information from several receivers on diﬀerent telescopes in order to substantially increase the angular resolution. 20

Figure 2.3: Schematic of a bolometer.

the ﬂux of incoming photons. They are useful for broad-band continuum observations.

Both coherent and incoherent detectors are used at submillimeter wavelengths, and are known as heterodyne receivers and bolometers, respectively. In the present work, we are mostly interested in submillimeter dust continuum observations which are obtained by means of bolometers (Fig. 2.3). These are very sensitive semiconductor devices which detect EM radiation over a wide spectral range, from millimeter to X- ray wavelengths. Bolometers consist of an absorber and a thermistor and are made of diﬀerent materials to be optimized for each spectral band. Incoming photons increase the temperature of the thermistor which results in a change in its electric resistance.

This change in resistance is measured as a change in voltage across the thermistor which provides, once calibrated, a measurement of the power coming from the detector

field of view at a given instant. The wavelength range of the absorbed photons, the effective bandwidth, is determined by a series of filters placed in front of the detector. 21

Over the past decade, the advent of cameras with many (increasing with time from a few tens to thousands) individual bolometers operating at the same time, called bolometer arrays, has signiﬁcantly increased the mapping capabilities of millimeter and submillimeter telescopes.

2.1.4 Observables in Submillimeter Astronomy

The basic quantity measured in observational astronomy is the speciﬁc intensity Iν, which is the energy per unit time, area, solid angle, and frequency interval emitted or received in/from a given direction. Thus, the speciﬁc intensity, also called brightness or simply intensity, has units W m−2 Hz−1 srad−1.

When we observe an extended source (i.e., with angular size larger than the beam), we resolve its intensity distribution. However, observing a point source we can only measure the ﬂux density, i.e. the total energy received per unit time, area, and frequency, and we have no information about the spatial distribution of the source.

The ﬂux density, is deﬁned as:

Fν = Iν cos θ dΩ , (2.1)

where θ is the angle between the propagation direction and the surface normal, and dΩ is the diﬀerential solid angle. Flux density is, therefore, measured in units of W m−2 Hz−1 (submillimeter ﬂux densities are usually expressed in Jansky units:

1 Jy = 10−26 W m−2 Hz−1). The ﬂux density received from an extended source may 22

be calculated using equation 2.1, and cos θ = 1 with a very good approximation.

The intensity received from an astronomical source does not depend on the distance between detectors and source, provided that submillimeter emission is optically thin, i.e., the extinction by interstellar dust can be neglected. However, the observed ﬂux density decreases with the square of the source distance. Flux density measurements over a wide range of frequencies allow the characterization of the energy emitted by a given source as a function of frequency, the SED, whose shape does not depend on the distance to the source. If the distance d is known, then, the total energy per unit time emitted by the source, called luminosity, can be estimated by integrating the R 2 SED over the whole spectral range: L = 4πd Fν dν.

In relatively high density regions such as molecular clouds, the absorption, scat- tering, and thermal emission of dust grains must be considered to appropriately characterize the radiation ﬁeld. The equation of radiative transfer describes the change in intensity with distance, s, along the line of sight:

dI ν = −ρκ I + j , (2.2) ds ν ν ν

−3 2 −1 where ρ [g cm ] is the total density, κν [cm g ] is the opacity at frequency ν, and jν is the emissivity of the medium (measured in units of intensity divided by length). If we consider a cloud in equilibrium at temperature T , the internal radiation ﬁeld must be described by the Plank function Bν(T ) everywhere and therefore, from equation 2.2 jν = ρκνBν(T ). Then, in the approximation of ρ, κν, and T constant across the 23

cloud, and if we consider the dust thermal emission optically thin (appropriate in the submillimeter regime), the intensity leaving the cloud is Iν = Bν(T ) ∆τν, where

∆τν = ρκν∆s is the optical depth through the cloud. Finally, if the cloud subtends a solid angle ∆Ω, the received ﬂux would be:

Fν = Bν(T ) ∆Ω ∆τν . (2.3)

We note that the wavelength dependence of ∆τν is poorly known and thus it is usually written as a power law, ∼ νβ. In section 4.4.1 we will use equation 2.3 in order to model the observed SEDs of the dust cores and obtain their physical parameters.

2.2 The BLAST Telescope

BLAST (Pascale et al. 2008) is a 2-meter telescope designed to perform three-band photometry at 250, 350, and 500 µm with an in-flight angular resolution of 3600, 4200, and 6000 respectively. Operated during LDB (Long Duration Balloon) flights above most of the atmosphere (at an altitude of about 40 km), BLAST achieves the highest sensitivity in these wavebands obtained to date. This allows a mapping speed approximately an order-of-magnitude faster than any other existing submillimeter facility in terms of detecting compact sources and even a greater improvement in terms of mea- suring diffuse structures in the interstellar medium. The design of BLAST is driven by its scientific primary goals: (i) to identify and characterize cold dust pre-stellar cores representing the earliest stage of star formation; (ii) to measure the structure of 24

the interstellar medium and molecular clouds where star formation takes place; (iii) to constrain the angular and redshift distribution of optically obscured star-forming galaxies; and (iv) to study the evolution and the spatial clustering of this extragalactic population.

Until the regular operation of the Herschel Space Observatory, BLAST is unique in its ability to derive large statistical samples of pre- and proto-stellar cores, providing the critical spectral coverage needed to constrain column densities, masses, luminosities and temperatures. In addition, BLAST large-scale surveys enable the study of the environmental eﬀects on the early evolution of star formation, from the scale of

GMCs to individual cold cores. The Galactic and extragalactic maps obtained by

BLAST and the techniques used for map reconstruction and analysis will serve as a legacy to be followed by Herschel. In this chapter we present a brief description of the telescope and summarize the observations performed during the 2005 (BLAST05) and 2006 (BLAST06) LDB ﬂights.

2.2.1 Telescope Design

The BLAST05 telescope had a 2 m CFRP (carbon ﬁber reinforced plastic) spherical primary mirror and a highly aspherical secondary mirror to compensate the spherical aberration of the primary (Olmi 2001, 2002). However, the optical design of the

BLAST06 telescope was based on a Ritchey-Chr´etienconﬁguration. The radiation collected by the warm optics of the telescope is redirected to a cryostat where cold 25

-5

-10

-15

-20

0 Relative Response (db) -5

-10

-15

-20

600 800 1000 1200 1400 1600 Frequency (Hz) Figure 2.4: Relative spectral response of the three BLAST broadband ﬁlters in the 2005 (top panel) and 2006 (bottom panel) conﬁgurations (Pascale et al. 2008).

reimaging optics (three additional mirrors) corrects the sky image for aberrations. A set of ﬁlters placed inside the cryostat block IR emission, split radiation into three beams (one for each wavelength; see Fig. 2.4), and deﬁne the wide band seen by each detector (∼ 30 % bandwidth). BLAST makes use of almost the same bolometer arrays as the SPIRE instrument on Herschel, consisting of 149, 88, and, 43 elements at 250,

350, and 500 µm respectively, organized in a hexagonal pattern.

A rigid and relatively light aluminum structure, the gondola frame, supports all the elements of the telescope. The gondola consists of two main components, an outer frame which provides the attachment point to the balloon ﬂight train, and an inner frame attached to the outer frame along a horizontal axis and free to move in elevation from 25◦ to 60◦ with respect to the outer frame. Fig. 2.5 shows a model of 26

Figure 2.5: Model of the BLAST gondola in the 2006 conﬁguration (Pascale et al. 2008).

the telescope and gondola in the 2006 conﬁguration. Sun shields are used to protect the telescope from solar radiation during the day and a set of solar panels provides the power required by the electronics, estimated to be ∼ 650 W.

BLAST can be controlled from the ground station only while the gondola is within line-of-sight and thus, it must be fully autonomous. Two redundant ﬂight computers receive and process information from all systems in the telescope, control the gondola motion, and write the data to disk. BLAST is primary driven by three torque motors which provide the active pointing of the telescope (see Fig. 2.6). Placed in the outer frame, a high-moment of inertia reaction wheel controls the azimuth pointing by rotating the gondola against it (due to angular momentum conservation). A pivot 27

motor at the attachment point to the ﬂight train prevents the reaction wheel from high speeds, counteracting torques from the balloon and diﬀerential wind speeds. Pointing control in elevation is provided by a torque motor mounted at the attachment point of the inner frame to the outer frame.

In order to produce fully sampled images and modulate the bolometer signals the telescope must scan each field several times. Cross-linking (scanning in different directions) is also necessary to reduce large scale noise in the map. We note that a balloon-borne telescope experiences a very complex motion during its flight, travelling free in latitude and longitude, undergoing rotations and oscillations due to variable torques. Thus, accurate pointing sensors and responsive motors are required, as well as an automated flight scheduler which provides a list of observations or actions in an optimal way according to the desire list of targets, scan parameters, and observational constrains at each moment. The primary pointing sensors are a set of gyroscopes and two optical star cameras which are complemented by secondary sensors as a magnetometer, Sun sensor, and a differential GPS. Gyroscopes measure the telescope’s angular velocities about three orthogonal axes, while star cameras are necessary to calculate its absolute orientation (by comparing the relative position of three stars, at least, to the position of stars in a reference catalog). The telescope’s orientation on the sky as a function of time, the pointing solution, is calculated in-flight with an accuracy of ∼ 3000. 28

Figure 2.6: Front and side schematic drawings of the BLAST gondola (Pascale et al. 2008).

2.2.2 Data Reduction: Calibration, Pointing Solution, and Map Making

BLAST data reduction is a complex process involving many steps before making a

final map of the sky specific intensity. In general terms, time-varying bolometer data must be calibrated, combined with a post-flight pointing solution, and projected into the associated map pixels (for a complete description of the process, see Pascale et al.

2008; Patanchon et al. 2008; Truch et al. 2008, 2009).

First of all, bolometer time streams, in voltage units and sampled at 100 Hz, must be cleaned of spikes from cosmic-ray hits and deconvolved from the filters applied by the electronics. Bolometer responsivities (change in voltage due to a change in incident power) fluctuate across the flight and therefore a correction must be considered. An internal calibration lamp provides a stable signal used to track these fluctuations and generate a constant responsivity time stream for each bolometer, which can be 29

combined with the pointing solution to make individual maps.

Post-ﬂight pointing reconstruction provides an absolute pointing of the telescope with an uncertainty ≤ 500. Gyroscope rates are integrated from the absolute reference provided by the star camera (given every two seconds) and sampled in phase with the bolometer time streams. A rotation must be applied to change from the star camera reference frame to the bolometer array coordinate frame and a correction accounts for the relative alignment between the star camera and the telescope. Thus, for each bolometer, the voltage “coming” from some speciﬁc sky coordinates is known as a function of time.

In order to make a map of the sky, the region of interest is conveniently divided in pixels so that each data sample can be assigned to the map pixel to which the detector points at each time interval. Pixel size must be smaller than the telescope’s

PSF in order to avoid loss of information. Initial pixel values, still in voltage units, are calculated as the average of all data samples corresponding to each pixel. It is important to note that the ﬁnal map at each wavelength will be the result of the convolution of the “true” sky emission with the telescope’s PSF and then, integrated over the normalized band-pass ﬁlter.

Before averaging data samples from different detectors, bolometer global responsivities must be compared to a reference bolometer for each wavelength. The relative calibration of different detectors, known as flat-fielding in optical astronomy, is performed by comparing the total integrated flux density from a point source, indi- 30

vidually calculated from maps made for each bolometer. For that purpose, a bright and point-like calibration source is mapped with the slow scan velocity needed to be fully-sampled from single bolometer data.

Now, constant and calibrated responsivity time streams can be used to make high signal-to-noise ratio multibolometer maps. However, the large number of detectors used by BLAST in addition to the on-axis design results in a significant effect of the correlation between time streams. Correlated noise, unlike uncorrelated noise, does not integrate down while increasing the number of detectors; therefore, different time streams cannot be considered as independent data in the mapmaking process.

To address this problem, which is common to all BLAST data, a new mapmaking method was developed: Signal and Noise Estimation Procedure Including Correlations

(SANEPIC), which is fully explained in Patanchon et al. (2008).

Finally, multibolometer maps, in voltage units, must be converted into ﬂux-unit

(or intensity) calibrated maps which are then used in all the subsequent analysis.

Astronomical flux calibration is performed by comparing the band-averaged flux of an isolated, point-like source as obtained from the BLAST maps to the value predicted from its SED (Fig 2.7). Thus, the primary calibration source must have a well-constrained SED from millimeter to FIR wavelengths and must be accessible throughout the flight. Available photometry is used to construct the SED and then

fitted to an isothermal modified blackbody model (see section 4.4 for a description of the model and the fitting procedure). Then, predicted band-averaged fluxes are 31

Figure 2.7: SED of the absolute flux calibrator for BLAST06 (the red hypergiant star VY CMa), as determined by published data from IRAS, SCUBA, SHARC-II, and Bolocam. The black line represents the best-fit model and the grey lines show the 68 % confidence intervals. Black diamonds indicate the model predictions at 250, 350, and 500 µmBLAST˙ broad-band filters (normalized to an arbitrary amplitude) are shown for reference (Truch et al. 2009).

Figure 2.8: BLAST06 PSFs at 250, 350, and 500 µm obtained from observations of VY CMa. The open circles represent the expected diﬀraction limited full width at half maximum (FWHM) for each waveband (Truch et al. 2009). 32

calculated by integrating the SED model over the normalized band-pass ﬁlters (see equation 4.7). The calibration coeﬃcient is calculated as the ratio of the BLAST

fluxes to those predicted and it is finally applied to the maps. We note that BLAST is insensitive to absolute brightness and the final maps have an arbitrary mean (the DC level is unconstrained). Therefore, background subtraction is necessary to perform any flux density measurement.

2.2.3 BLAST 2005 and 2006 Science Flights

BLAST has made one test flight and two LDB science flights to date. A first 1-day

flight was devoted to test the instrument performance in order to ensure a proper operation during the following flights. In 2005, a 5-day science flight was conducted from Sweden to northern Canada. Extragalactic observations could not be performed due to a failure in the optical system (with a significant degradation of the beam shape) and thus, the full observing time was spent in Galactic surveys for which the data are still very relevant. The second science flight took place in 2006 from

Antarctica. During the 11-day ﬂight the telescope performed successfully, achieving the main Galactic and extragalactic goals.

BLAST surveyed a total of ∼ 20 deg2 of the Galactic plane visible from the northern hemisphere during its 2005 LDB ﬂight, including extensive maps toward Vulpecula,

Cygnus, Aquila, and Sagitta, well-known sites of high-mass star formation. A detailed study of the 4 deg2 Vulpecula map has been published recently (Chapin et al. 2008), 33

and the other regions are currently under analysis. During the 2006 LDB ﬂight,

BLAST surveyed ∼ 200 deg2 of the southern Galactic plane at nearly diﬀraction- limited resolution. The surveyed area includes a ∼ 50 deg2 deep map of the Vela

Molecular Ridge nested inside a wider and shallower map, ∼ 3 deg2 in Eta Carina,

∼ 1 deg2 map of the Gum nebula, and a ∼ 3 deg2 map towards the supernova remnant

Pup A. In addition, during this ﬂight BLAST conducted unprecedented extragalactic surveys towards the GOODS South ﬁeld and the South Ecliptic Pole. First Galactic and extragalactic results have been published recently and may be accessed from the

BLAST web page2.

2http://blastexperiment.info/results.shtml 34

Chapter 3 OBSERVATIONS

In this chapter we introduce the Vela Molecular Ridge (VMR), a region exten- sively mapped by BLAST in the submillimeter dust continuum emission. We brieﬂy describe the current understanding of star formation, based on previous observations by diﬀerent authors, and introduce the region analyzed in the present work, the Vela-

D cloud. Then, the speciﬁc observations performed towards this cloud at diﬀerent wavelengths are described in detail, with special attention to the maps and catalogs used in this work.

3.1 The Vela Molecular Ridge

The VMR is one of the nearest GMC complexes. Located in the Galactic plane (b

= ± 3◦) and extended over 17◦ in longitude (l ∼ 257◦–274◦), it was ﬁrstly observed by May et al. (1988) in the 12CO(1–0) transition. Murphy and May (1991) subdivided the region into four main GMCs, named A, B, C, and D, and estimated their 12CO

5 masses to be ∼ 10 M¯ (see Fig. 3.1). They found that most of the molecular gas content is located behind the Vela supernova remnant (SNR; estimated to be at a distance of ∼ 500 pc) and noted that the CO emission is well correlated with optical

HII regions but uncorrelated with OB stars observed in the same ﬁeld. 35

Figure 3.1: 12CO integrated intensity map of the VMR (Murphy and May 1991).

Liseau et al. (1992) went over distance evaluations and estimated that GMCs A,

C, and D are located at a distance of 700 ± 200 pc while GMC-B is ∼ 2 kpc away. A more detailed analysis of the dense molecular gas in the VMR was carried out by Yam- aguchi et al. (1999) in the 12CO, 13CO, and C18O (1–0) emission lines. They resolved

GMCs C and D into smaller clouds and clumps and showed that GMC-C is more massive and probably in an earlier evolutionary state than GMC-D. Higher resolution

12CO observations of Moriguchi et al. (2001) showed a general correlation between the CO emission and the X-ray distribution from the Vela SNR, suggesting a possible interaction between molecular clouds and the supernova remnant (see Fig. 3.2).

Through a number of continuum and spectral line observations, the VMR appears as an actively low- and intermediate-mass star forming region. It has been searched for YSOs by diﬀerent authors, usually through the IRAS and/or MSX Point Source

Catalogues (PSC). NIR observations of selected IRAS sources performed by Liseau 36

Figure 3.2: 12CO integrated intensity map of the VMR (contours) overlayed to the X-ray image of the Vela SNR (grey scale) (Moriguchi et al. 2001).

et al. (1992) and Lorenzetti et al. (1993) led to the ﬁrst catalogue of Class I objects in the VMR. They found that low- and intermediate-mass star formation has been occurring for at least 106 years, as suggested by the presence of Class II sources. Based on bolometric luminosity estimations, they concluded that there is no current massive star formation in the region, at least in GMCs A, C, and D.

Subsequent NIR observations towards GMCs C and D by Massi et al. (2000, 2003)

3 showed that IRAS sources with bolometric luminosities Lbol > 10 L¯ are in fact associated with young star clusters (< 100 members) which present a stellar mass distribution similar to the canonical initial mass function (IMF). They found that Class 37

I-consistent IRAS sources are the most massive and less evolved components in each cluster. In addition, the presence of IRAS sources with low bolometric luminosities, associated with small groups of YSOs or isolated, suggests that both cluster and diffuse star formation modes are present in the GMC-D. Furthermore, protostellar jets found by Lorenzetti et al. (2002) and De Luca et al. (2007) using NIR imaging and spectroscopy evidenced a high activity in the region, supporting also the picture of diﬀerent modes of star formation. The interaction between some of these protostellar jets and the surrounding medium has been studied in great detail by Giannini et al.

(2005).

New 12CO(1–0) and 13CO(2–1) arcminute-resolution observations performed by Elia et al. (2007) highlighted the complex dynamics dominating the GMC-D. They found different structures such as clumps and filaments associated with FIR sources and suggested the existence of expanding shells which could be the cause of the observed cluster star formation. Expanding shells appear as arc-like and elliptical emitting regions using integrated intensity maps and velocity field analysis. Their origin was discussed in terms of possible interactions between the molecular cloud and near expanding HII regions, stellar winds from a previous generation of young stars, and even the Vela SNR. However, none of these descriptions seemed to be convincing and the hypothesis of filamentary structures being a natural consequence of turbulent gas evolution cannot be rejected. In addition, Elia et al. (2007) identified a number of molecular outflows originated in the vicinity of the IRAS sources. 38

FIR sources were also found to be associated with dense dust cores as shown by Massi et al. (2007) from 1.2 mm dust continuum observations, confirming that they are in an early evolutionary state. De Luca et al. (2007) compiled a list of ∼ 60 dust cores and identified their IR counterparts using the IRAS, MSX, and 2MASS catalogs in addition to further NIR data. They found that the majority of the core counterparts have SEDs consistent with being Class I protostars and about one third of the cores present jets driven by sources located into their interior. The preceding description of the GMC-D young stellar population was significantly improved by means of a recent Spitzer MIPS imaging survey carry out by Giannini et al. (2007).

They considerably increased the number of known YSOs, identiﬁed some of the sources driven protostellar jets, and proposed some MIPS sources as potential candidates for

Class 0 objects.

The millimeter dust continuum survey performed by Massi et al. (2007) was not sensitive enough to obtain a complete census of pre- and proto-stellar cores in the

GMC-D region, essential for understanding the very early stages of star formation. In a very recent submillimeter continuum survey presented in Netterﬁeld et al. (2009),

BLAST has characterized the cold dust emission in the VMR in a wide scale range, from GMCs to individual cores (see Fig. 3.3). They have identiﬁed over a thousand compact sources simultaneously in the three bands. Through the BLAST three-band photometry alone, they have estimated the temperature, mass, and luminosity of the dense cores, characterizing the youngest population in the GMC complex. In addition, 39

using the source temperature to discriminate between starless and proto-stellar cores

(they considered a separation limit of 14 K based on correlations with the IRAS and

MSX PSCs) they inferred a mass dependent cold core lifetime signiﬁcantly longer than free fall and turbulent decay models, suggesting the existence of an additional non-thermal support.

However, though BLAST bands sample the peak of the SED of the coldest objects, temperature and therefore, mass and luminosity estimations of warmer cores are subject to signiﬁcant uncertainties. Therefore, additional FIR-MIR data are needed to better constrain the dense core physical parameters. Furthermore, MIR/NIR data are essential to directly discriminate between starless and proto-stellar cores, without the need of assumptions about the temperature of the cores. In this work, we analyze in detail the portion of the GMC-D encompassing the majority of the millimeter to NIR data mentioned above, in order to carry out a detailed multi-wavelength analysis of the BLAST cores. Thus, for practical reasons we limit our analysis to the area con- tained within 262◦.80 < l < 264◦.60 and −1◦.15 < b < 1◦.10, which we refer hereafter as Vela-D. 40

3.2 BLAST Data

3.2.1 BLAST Observations

BLAST mapped ∼ 50 deg2 of the VMR in a 21-hour survey performed during the 2006 science ﬂight. Fig. 3.3 shows a composite RGB image1 made from the BLAST 250,

350, and 500 µm maps, where color scale is an indicative of temperature: blue and red colors indicate warmer and cooler regions respectively. Spanning ∼ 10◦ in Galactic longitude and ∼ 5◦ in Galactic latitude, BLAST maps encompas the main four GMCs previously detected in molecular line observations (Murphy and May 1991). Dust emission follows a similar distribution to the gas traced by the 12CO(1–0) transition

(see Fig. 3.1 and Fig. 3.2). A large number of diﬀerent structures such as clumps,

filaments, and individual dense cores are observed across the field. Netterfield et al.

(2009) have identiﬁed and measured the ﬂux densities of 1282 sources simultaneously at 250, 350, and 500 µm. The sample of 141 BLAST sources located within Vela-D

(see Fig. 3.3 and Fig. 3.4) will be used as a reference catalog for all the subsequent analysis.

1False color image in which the color of individual pixels is specified by three values, indicating a particular combination of red, green, and blue colors. 41 m channels µ Figure 3.3: Compositerespectively RGB image (Netterfield of et thesection al. VMR 3.1. 2009). using blue, green, The and region red enclosed colors by for white 250, 350, solid and lines 500 is the Vela-D cloud, as defined in 42

3.2.2 Source Identiﬁcation and Flux Extraction in the BLAST Maps

The source extraction technique applied to the Vela map is similar to the method described by Chapin et al. (2008) for the analysis of the BLAST-2005 Vulpecula maps. After a MHW (mexican hat wavelet; i.e., the negative normalized second derivative of a Gaussian function) type convolution is applied to the 250 and 350 µm maps, intensity peaks above a certain threshold are identiﬁed as candidate sources (the

500 µm map is not used for source identification due to its intrinsic lower resolution which results in higher source-source and source-background confusion). Then, 2- dimensional Gaussians are fitted to each candidate source, determining their FWHMs and centroid positions, and the 250 and 350 µm candidate lists are merged to create a final source list containing positions and sizes. Finally, flux densities are calculated as integrals of Gaussians of fixed size and position and variable amplitude fitted to the 250, 350, and 500 µm maps. The bias and completeness of the source and flux extraction method is evaluated using Monte Carlo simulations (see Fig. 3.5). Fake sources (Gaussians of known amplitude, size, and position) are randomly added to the maps and the source extraction pipeline is applied again. The comparison of the input to the output values allows quantifying the completeness and determining a flux correction factor as the mean ratio of the input simulated flux to the recovered flux.

Once local emission peaks have been identiﬁed, ﬂux extraction could be performed using aperture photometry or PSF photometry instead of Gaussian photometry. We have applied and compared these methods in the Vela-D map. In summary we found 43

Figure 3.4: Colour-scale image of the BLAST 250 µm Vela-D map, with the locations of compact sources indicated by circles of diameter proportional to the observed source sizes. Dust cores are usually found in the densest regions of the cloud, often associated to shell-like structures and ﬁlaments. 44

that: (i) aperture photometry is seriously affected by contamination in crowded fields and therefore cannot be generally applied to the BLAST maps; (ii) PSF photometry is very efficient in crowded fields but provides accurate fluxes only for point sources; and (iii) multi-Gaussian photometry can deal with both crowded fields and resolved sources.

In addition, Gaussian ﬁtting provides us information about the source size. The high signal-to-noise ratio of the BLAST maps allows us to infer the intrinsic size of the BLAST sources using the standard Gaussian deconvolution formula:

2 2 1/2 θdec = (θfwhm − θbeam) , (3.1)

where θfwhm is the observed FWHM from the best Gaussian ﬁt and θbeam is the

BLAST beam size. The FWHM median value in the VMR is 6200, which corresponds to an intrinsic deconvolved size of 0.15 pc at a distance of 700 pc (the estimated distance of the GMCs A, C, and D).

BLAST has a beam width of 3600 at 250 µm (i.e eﬀective spatial resolution of

∼ 0.1 pc) and therefore it seems that BLAST sources are individual resolved objects.

This spatial resolution is comparable to the average size of the dense cores found by

Motte et al. (2007) in Cygnus-X (with a resolution of ∼ 1100 using MAMBO at IRAM ) but considerably larger than values reported by Ward-Thompson et al. (1999) and

Enoch et al. (2008) in nearby star forming regions. Therefore, we cannot exclude that some of the BLAST cores may be actually composed of multiple objects. Furthermore, 45

0.8

0.6

0.4

Completeness 0.2

60 50 40 30 20 Number in Bin 10 0 40 50 60 70 80 90 100 Source FWHM ["] Figure 3.5: Top panel. Completeness of the source extraction method as a function of size. Bottom panel. Size distribution of the BLAST cores in the VMR. The blue bars correct for the size completeness shown in the top panel. (Netterﬁeld et al. 2009).

it has been shown that the ratio θdec/θbeam is independent of the spatial resolution in star forming regions, and it seems to be a consequence of fitting Gaussians to power- law structures convolved with the beam. Dust cores appear as extended sources of power-law radial profiles immersed on a medium populated with structures of a wide range of scales. Therefore, they cannot be described as isolated objects with well defined boundaries. The size of the dust cores is not a well-defined concept and should be regarded as a typical scale rather than an absolute physical parameter.

3.3 Archive Data

3.3.1 SIMBA

A ∼ 1 deg2 area of Vela-D was mapped in the 1.2 millimeter continuum using the bolometer array SIMBA at the SEST (Massi et al. 2007). The 1.2 mm map (see 46

Fig. 3.6) has a resolution of ∼ 2400 (the SIMBA beam at this wavelength), and the r.m.s is ∼ 20 mJy/beam with signiﬁcant variations over the map. The dust emission was analyzed using the CLUMPFIND algorithm which allows recognizing compact condensations with arbitrary shapes. From the CLUMPFIND output, Massi et al.

(2007) reported a robust list of 29 dust cores with sizes greater than the SIMBA beam

(“MMS” cores following their deﬁnition) and a sample of 26 candidate cores with sizes smaller than the SIMBA beam (“umms” cores) and therefore initially discarded. The deconvolved sizes of the MMS cores are in the range ∼ 0.03–0.25 pc (at a distance of

700 pc) and have masses in the 0.4–88 M¯ range (estimated by assuming a temperature

2 −1 of 30 K, a dust-to-mass ratio of 100, and a dust opacity κ1.2 = 0.5 cm g at 1.2 mm).

The umms sources were studied in detail by De Luca et al. (2007) and they found that many of them show signposts of star formation (associations with IR sources and/or protostellar jets), suggesting that these objects are probably real millimeter sources and not just instrumental artifacts. Therefore, we consider the complete SIMBA catalog, MMS and umms cores, to look for possible associations with the BLAST cores.

An immediate comparison between the SIMBA and BLAST 250 µm maps (see

Fig. 3.4 and Fig. 3.6) reveals a good correspondence between them. However, though both wavelengths are supposed to trace basically the same neutral material (cold dust grains), we can clearly see that most of the extended emission is not detected by

SIMBA, due to its lower sensitivity compared to BLAST. The bulk of the SIMBA 47

Figure 3.6: SIMBA 1.2 mm emission overlaid with the BLAST 250 µm contours at 100, 300, and 1000 MJy/srad. The position of the BLAST sources are indicated using white crosses. Also shown are the IRS sources from Liseau et al. (1992). 48

1.2 mm emission essentially originates from known sites of recent star formation dis- covered by IRAS. In fact, a signiﬁcant fraction of the SIMBA cores, the most massive and luminous, are associated with IRAS sources, denoted as IRS by Liseau et al.

(1992) when associated with NIR objects. The brightest IRS sources are associated with young stellar clusters (Massi et al. 2007).

3.3.2 IRAS, AKARI, and MSX

During a period of ten months in 1983, the Infrared Astronomical Satellite (IRAS) performed an all sky survey at 12, 25, 60, and 100 µm with an angular resolution of approximately 0.5, 0.5, 1.0, and 2.0 arcmin, respectively. In this work we use the

IRAS PSC version 2.0 (Beichman et al. 1988) to obtain ﬂux measurements of the

BLAST cores at 60 and 100 µm. This catalog provides positions, ﬂux densities, and uncertainties of ∼ 250 000 point sources with completeness limits of about 0.4, 0.5,

0.6, and 1.0 Jy at 12, 25, 60, and 100 µm. We ﬁnd 157 IRAS PSC sources within

Vela-D. In addition, we use the IRAS Galaxy Atlas2 (IGA), a reprocessed version of the maps at 60 and 100 µm, to complement the PSC photometry. These maps are produced using the resolution-enhancing algorithm HIRES (Aumann et al. 1990) and the PSFs vary across the ﬁeld, showing strong elongation along the scan direction.

The Far-Infrared Surveyor (FIS) on the AKARI space telescope (Yamamura 2008) is an instrument designed to perform an all-sky survey at 60, 90, 140, and 160 µm with

2http://irsa.ipac.caltech.edu/data/IGA. 49

much better sensitivity and spatial resolution than IRAS. The AKARI data should be very useful to characterize the FIR emission of the dense cores observed by BLAST. In this work we use a preliminary version of the AKARI source catalog to ﬁnd the FIR counterparts of the BLAST cores. In addition, we make use of the MIR all-sky survey performed by the Midcourse Space Experiment 3 (MSX ) at wavelengths 8.28, 12.13,

14.65, and 21.3 µm. Though this survey is much less sensitive than the observations performed by the Spitzer MIPS and IRAC instruments (see below), the MSX PSC is valuable to ﬁnd MIR counterparts for the BLAST cores located outside the area covered by MIPS and IRAC. We ﬁnd a total of 362 AKARI sources and 194 MSX sources within Vela-D.

3.3.3 MIPS

Our multi-wavelength analysis takes advantage of the Spitzer MIPS imaging survey presented in Giannini et al. (2007). They mapped an area of ∼ 1.5 deg2 at 24, 70, and

160 µm, although the coverage of the latter was incomplete due to the very small ﬁeld of view of the 160 µm array (the mapping parameters were optimized for the 24 and

70 µm bands). Therefore, only the 24 and 70 µm maps were exploited for scientiﬁc analysis (see Fig. 3.7). Final calibrated and mosaiced images were obtained using the standard MOPEX package4. Though the main instrumental artefacts were removed from the mosaiced images, some residual stripes oriented in the scan direction are

3http://irsa.ipac.caltech.edu/Missions/msx.html. 4Provided by the Spitzer Science Center. 50

quite evident in the 70 µm map, aﬀecting signiﬁcantly the point source extraction and photometry at this wavelength.

Giannini et al. (2007) compiled a catalog of 849 and 61 point sources detected at

24 and 70 µm respectively. Spitzer is diffraction limited in the MIPS bands with a spatial resolution of ∼ 600 at 24 µm and ∼ 1800 at 70 µm and the survey was complete down to 5 mJy and 250 mJy respectively. The point source extraction and photometry were performed using the standard astronomical package IRAF. Peaks were identified in a differential image produced by subtracting a “sky” image from the final mosaic, and flux densities were measured using aperture photometry. After applying a spatial

filter to have a more uniform sensitivity level across the field, they performed PSF photometry by fitting an empirical point response function derived from a sample of

“bona ﬁde” point sources in the same image. This approach allowed them to obtain an enlarged catalog of 1347 and 63 point sources at 24 and 70 µm respectively with better photometric accuracy.

3.3.4 IRAC

An area of ∼ 1 deg2 has been recently mapped by means of Spitzer-IRAC observations.

Though this survey has not being published yet (Strafella et al. 2009), we use the IRAC data to characterize the BLAST sources up to the shortest possible wavelengths. The

IRAC sensitivity allow us to detect objects with ﬂuxes down to 50 µJy, reaching a completeness limit of ∼ 400 µJy. As for the MIPS data, the IRAC maps have been 51

Figure 3.7: False colour image of the Vela-D map using the MIPS 24 µm channel for blue, the MIPS 70 µm channel for green, and the BLAST 250 µm channel for red. The regions observed by the diﬀerent instruments are shown for reference: green and blue lines show the MIPS 70 and 24 µm coverages respectively, the purple line encloses the region covered by IRAC, and the contours represent the SIMBA map. 52

analyzed using the DAOPHOT task of the IRAF package and PSF photometry was performed instead of aperture photometry. The ﬁnal point source catalog contains the position and ﬂuxes of more than 170 000 sources detected above 5σ at least in one of the IRAC bands, 3.6, 4.5, 5.8, and 8.0 µm.

The high sensitivity of the IRAC image (see Fig. 3.8) allows us to see a general correspondence between the emission at 8 µm and 250 µm as partially expected because both trace neutral material. However, the 8 µm emission is known to be heavily inﬂuenced by PAH (polycylic aromatic hydrocarbon) emission and therefore is not proportional to the column density. In addition, we found that some of the BLAST sources are coincident with local minima or darker regions in the IRAC image, suggesting the existence of the so-called Infrared Dark Clouds (see, e.g., Egan et al.

1998). 53

Figure 3.8: Overlay of the IRAC 8 µm image (color scale) with the BLAST 250 µm contours. The position of the BLAST sources are indicated using white crosses. 54

Chapter 4 ANALYSIS

The primary goal of this work is to characterize the SED of the dust cores found by BLAST from millimeter to MIR wavelengths and then analyze their physical properties. For that purpose, we need: (i) to construct the SEDs of the BLAST sources and thus to associate the flux densities measured by BLAST at 250, 350, and 500 µm with the emission observed at other wavelengths; and (ii) to fit these SEDs assuming a theoretical model and relate the fitting parameters to the core physical properties.

4.1 Overview of the Analysis

In order to construct the SED of the BLAST cores we have essentially two possibilities:

• To look for candidate counterparts from catalogs at diﬀerent wavelengths, adopt-

ing the ﬂux densities of the sources associated with the BLAST cores.

• To estimate ﬂux measurements at diﬀerent wavelengths directly from the original

maps, performing our own source photometry.

In practical terms, we often need to use both methods depending on the available data, catalogs and/or images, at diﬀerent wavelengths. However, care must be taken in determining the correct set of ﬂux densities which appropriately describes the SED 55

of the BLAST cores. Generating a source catalog is a complex process which usually involves the use of source and flux extraction methods specifically dedicated and developed according to the particular characteristics of the data. Therefore, as a first attempt to characterize the SED of the BLAST cores, we initially used the available catalogs at different wavelengths. However, source and flux extraction techniques are also dependent on the specific goals of the analysis. We must, thus, define the object under analysis and decide about the correct flux extraction method to be used at different wavelengths.

4.1.1 What are the Correct Measurements?

We are interested in the study of the earliest stages of star formation, which are closely related to the evolution of dense dust cores. Therefore, our analysis is focused on the submillimeter dust cores found by BLAST and how they emit radiation at other wavelengths. As we have seen, BLAST sources are defined by Gaussian fits which delimit the specific areas of the sky from where flux densities are extracted. Thus, in determining the SED of a BLAST core, we should ideally integrate the specific intensity emitted at other wavelengths over the same area of the sky, even though this approach still does not ensure that the integrated emission in each waveband comes from the same volume of material (see below).

Following this idea, a rigorous flux extraction method would be to convolve the maps (at different wavelengths) to the BLAST resolution and then to fit the Gaussian 56

profiles extracted from the BLAST maps, with fixed sizes and positions. Then, the integral of these Gaussians would provide the flux densities associated to the BLAST cores. Alternatively, we could use the original maps at native resolution, fitting directly the BLAST Gaussian profiles convolved to account for the different beam sizes at different wavelengths. This is essentially the method applied to the BLAST maps where the final fits are performed using the same Gaussian profiles at the three wavebands. However, at shorter wavelengths the brightness distribution is quite different compared to the BLAST emission (see Fig. 5.1 and Fig. 5.2) and therefore this method cannot be generally applied.

Aperture photometry allows to integrate the emission over any desired aperture.

Selecting a dynamical aperture of radius Rdyn = 1.3 × FWHM, centered over the

BLAST coordinates, the resulting flux density would be equivalent to 95 % of the integral of a Gaussian of a given FWHM. However, the typical large apertures needed for BLAST cores often result in a significant contamination from nearby sources and diffuse emission due to the characteristic crowding in GMCs. The question whether or not this “contamination” at different wavelengths is actually associated with the

BLAST core emission must be addressed in a case by case basis. We also note that integrating the emission in a dynamic aperture can often lead to the estimate of only an upper limit. Thus, in order to constrain the SED of the BLAST cores more effectively, we need to analyze in detail each case individually, often through the visual inspection of the same fields at different wavelengths. 57

Submillimeter dust cores present density and temperature gradients, particularly proto-stellar cores, implying that the bulk of the emission at different wavelengths may come from different volumes of material. In addition, the optical depth depends on the observed wavelength (increasing towards shorter wavelengths) and the instrumental response is also different. It is thus not possible to define a “unique core” at different wavelengths or even state that “the same core” is seen at different wavelengths. Theory and observations suggest that proto-stellar cores (with a protostar or proto-stellar cluster) are likely composed of a warmer central part embedded in a colder and less dense envelope1. At wavelengths shortward of ∼ 100 µm the bulk of the emission comes from the inner part and the contribution from the envelope can be neglected. Therefore, integrating over a reasonable small aperture we can approximately measure the flux density emitted by the system at a given wavelength, avoiding major contamination problems. In the MIR regime we can even consider that the bulk of the emission comes from point sources inside the BLAST cores and thus perform PSF photometry, which is the most accurate flux extraction method in crowded fields. However, PSF photometry may underestimate flux densities in the case of compact or extended sources.

1Many Starless cores are thought to be cooler inside and warmer outside due to the absence of an inner YSO and the presence of an external heating source, the interstellar radiation ﬁeld. 58

4.1.2 The Approach

In summary, we have seen that submillimeter dust cores are very complex systems and there is no universally valid ﬂux extraction method. Therefore, in order to construct the SEDs, we perform the following steps:

• execute the cross-correlation between the BLAST and ancillary catalogs.

• obtain additional photometry for comparison with catalog ﬂuxes and estimation

of upper limits.

• create multi-wavelength images, or “thumbnails”, showing the position of cat-

alog sources, the size of BLAST cores, and the apertures used for aperture

photometry.

• ﬁt preliminary SEDs and visually analyze each source, using the thumbnails as

a qualitative reference.

• fit final SEDs using the appropriate fluxes.

For those cases where catalog ﬂuxes need to be revised, we perform speciﬁc aperture photometry centered over the coordinates of the catalog sources. In addition, for those BLAST sources without counterparts at a given wavelength, we estimate upper limits using aperture photometry centered over the BLAST coordinates (except at wavelengths ≤ 24 µm, see section 4.4). The thumbnails allow us to visually assess 59

the presence of multiple sources, as well as other morphological characteristics (e.g., a “core + halo” structure), in the diﬀerent wavebands. An additional comparison can be made with the preliminary SEDs obtained using only BLAST ﬂux densities.

4.2 1200–8 µm Source Identiﬁcation

In order to ﬁnd candidate counterparts of the BLAST cores we initially adopt a spatial criterion and perform cross-correlations with all the existing catalogs. We consider a source being associated with a BLAST core when their distance is less than the following dynamical search radius:

2 2 1/2 Rsearch = [(²blast) + (²archive) ] (4.1)

where ²archive is the pointing error associated with the position of the archive source, and we conservatively deﬁne a positional uncertainty of the BLAST source:

2 2 2 1/2 ²blast = [(FWHMdec/2) + (²ptg) + ²extr] (4.2)

Where FWHMdec is the deconvolved FWHM of the BLAST source, ²ptg is the

BLAST pointing error and ²extr is the uncertainty in the source position due to the source ﬁnding technique (both estimated to be ' 500). This search criterion takes into consideration that BLAST cores are extended sources which may be associated with a number of point-like sources. In fact, we have found that some BLAST cores appear to be associated with NIR clusters. 60

4.2.1 SIMBA

The higher signal-to-noise ratio of the BLAST maps compared to the SIMBA map is evident in terms of the number and distribution of dust cores detected in the same region. As we have seen in Fig. 3.6, most of the SIMBA emission seems to be associated with a BLAST core. In order to perform a cross-correlation between the BLAST and

00 SIMBA catalogs we assume ²simba = 5 , the estimated SIMBA pointing error (Massi et al. 2007). After applying our search criteria, we ﬁnd that 31 BLAST cores have one or more SIMBA counterparts, corresponding to ∼ 36 % of the BLAST sources located within the area observed by SIMBA. We note that some BLAST sources are actually composed of more than one core, as revealed by the higher angular resolution of the

SIMBA map. In addition, the fact that some of the SIMBA cores are associated with the brightest IRS sources (De Luca et al. 2007) implies that individual BLAST cores can be linked to the formation of young stellar clusters rather than individual stars.

We have analyzed in detail the positional offset between the SIMBA and BLAST sources to check the consistency of the association criteria. We find that these offsets are randomly distributed with an average value approximately within the BLAST and SIMBA pointing errors. Therefore, we consider that our associated sources are in fact the same objects observed by BLAST and SIMBA and the relative offsets can be explained by instrumental pointing and source extraction algorithm. 61

4.2.2 IRAS, AKARI, and MSX

The IRAS bands nicely complement the BLAST maps, especially the 100 µm waveband which samples the peak of the SED at temperatures ∼ 30 K. However, the coarse angular resolution of the IRAS map in most cases prevents assigning reliable ﬂuxes to the BLAST cores emission. Nevertheless, IRAS data can provide upper limits to constrain the SEDs, which is useful for BLAST sources located outside the MIPS spatial coverage. For the cross-correlation between the IRAS PSC and the BLAST catalog we conservatively set the positional error ²iras equal to the semi-major axis of the IRAS error ellipse (∼ 1500 to 8000 for PSC sources in Vela-D). We ﬁnd associations for only 26 of the 141 BLAST sources in Vela-D.

In order to characterize the emission of the BLAST cores at 60, 90, 140, and

160 µm, we perform a cross-correlation between the BLAST and AKARI catalogs,

00 for which we consider a positional error ²akari = 8 (the pixel size of the images used to create the source catalog). We ﬁnd 44 BLAST cores associated with at least one

AKARI source (4 BLAST cores seem to be resolved into two components). However, the AKARI flux densities have not been used in estimating the best-fit SED (see section 4.4.2), as we have noted that they are not often consistent with the SED as determined using all other wavebands. Since we do not have access to the AKARI maps we have not been able to compare the emission at different wavelengths, and thus we have conservatively decided to use the AKARI flux densities only as a reference.

As for the correlation between the BLAST catalog and the MSX PSC, we have set 62

00 the positional error ²msx = 3 . We ﬁnd MSX counterparts for 21 of the 141 BLAST cores in Vela-D. For comparison with the more sensitive Spitzer data, we ﬁnd only 15

MSX counterparts in the area cover by MIPS at 24 µm, while we ﬁnd 55 BLAST cores associated with 24 µm sources (see below). This conﬁrms the importance of having

MIPS data to identify the MIR sources associated with dense cores.

4.2.3 MIPS

In order to search for possible 24 and 70 µm counterparts to the BLAST sources, we

00 assumed a MIPS pointing error of ²mips = 2 and initially considered all the sources falling within the search radius being associated with a given BLAST core. We found that 55 and 32 BLAST sources have at least one counterpart at 24 and 70 µm respectively. However, our sample could be aﬀected by extragalactic contamination which is expected to be signiﬁcant at the 24 and 70 µm completeness levels (as determined from a comparison of the counts per deg2 between the present survey and the Spitzer

Wide-area Infrared Extragalactic Survey—SWIRE—legacy program). Enoch et al.

(2008), in a similar work, considered a threshold of 3 mJy at 24 µm to eliminate most of the extragalactic interlopers. We ﬁnd that only 2 BLAST cores have a 24 µm counterpart below this threshold and thus we assume that our analysis is not being aﬀected by extragalactic contamination. In addition, given the large number of MIPS

24 µm sources in the ﬁeld, we also explore the possibility of ﬁnding a chance association within the search radius. By calculating the mean number of MIPS sources in 63

an area equivalent to that deﬁned by the search criteria, we estimate the probability of chance associations to be < 20 %. However, if we consider only sources with ﬂux densities > 3 mJy the probability drops to < 10 %, and therefore we can conclude that the 24 µm counterparts of the BLAST cores are unlikely to be the result of chance associations.

4.2.4 IRAC

We perform a cross-correlation between the IRAC and BLAST catalogs also assuming

00 a pointing error ²irac = 2 . Now, the probability of ﬁnding chance association between the two catalogs is signiﬁcantly higher than the estimated value for the 24 µm catalog.

We found that, with a probability of chance associations > 100 %, all the BLAST cores within the area covered by IRAC seem to have at least one IRAC counterpart, and many of them would be associated with NIR clusters. While we have seen that

Vela-D presents a high eﬃciency clustered star formation mode, the 24 µm data suggest that the majority of those cannot be real associations. A study of the potential association of NIR sources with dust cores is beyond the scopes of this work. Though

NIR photometry has little eﬀect on our SED ﬁtting model, it can provide valuable information about the core evolutionary state. Therefore, we have carried out a preliminary search of IRAC counterparts using the following criteria: (i) spectral index

α = d log(λFλ)/d log(λ), estimated using 2MASS, IRAC, and MIPS 24 µm data, with values α > 0.3; (ii) the source must be detected in the IRAC 8 µm band; and (iii) 64

when no MIPS 24 µm flux is available, the flux extrapolated from 2MASS and IRAC data must be consistent with an upper limit in the MIPS 24 µm band. With these criteria, we find that 48 BLAST sources have at least an IRAC counterpart.

4.3 Source Photometry

In this section, we describe the specific flux extraction methods finally used to construct the SED of the BLAST cores. The supplementary photometry performed at

1200, 100, 70, and 60 µm using the SIMBA, IRAS, and MIPS maps is discussed in detail.

4.3.1 SIMBA

In order to create the SED of the BLAST cores, we adopt the flux densities of the associated SIMBA cores listed in Massi et al. (2007). While CLUMPFIND is expected to provide consistent flux densities for the resolved MMS cores, flux densities and sizes of the umms cores are unreliable. In fact, we realized that only the MMS fluxes can be consistent with the SEDs as defined by the BLAST fluxes. Therefore, we preferred to estimate the flux densities of the umms cores using aperture photometry. We chose

00 00 a circular aperture of radius Rapr = 1.3 × 35 (35 being the BLAST FWHM at 250 micron), centered over the SIMBA coordinates. Baseline pixel values were estimated as the median in an annulus between the outer-edge of the main aperture and a second √ circle with a radius Rsky = 2 Rapr in order for the annulus to have the same area as the aperture. If more than a SIMBA MMS source is associated with a single BLAST 65

object, their ﬂuxes are co-added and considered as an upper limit in the SED ﬁt.

However, when a multiple source is composed of at least one umms core, we integrate the map in an area equivalent to the size of the BLAST source. For that purpose, we perform aperture photometry using a dynamical aperture of radius Rdyn = 1.3×Dsource centered over the BLAST coordinates and an annulus region of the same area. Here

Dsource is the BLAST deconvolved FWHM, convolved with the SIMBA beam (thus, an estimate of the source size as seen by SIMBA). For those BLAST sources falling inside the area mapped by SIMBA but without a SIMBA counterpart, we estimate an upper limit to the 1.2 mm ﬂux density performing aperture photometry with the same aperture and annulus. However, when the resulting upper limit is less than 0.4 Jy we retain this value, since it corresponds to the average noise of ∼ 20 mJy/beam integrated over an aperture of size equal to the BLAST beam.

4.3.2 IRAS

IRAS PSC ﬂuxes are used to complement the BLAST photometry at 100 and 60 µm.

For those BLAST sources without an IRAS counterpart we estimate upper limits performing direct aperture photometry in the IRAS maps, centered over the BLAST coordinates. We integrate the 100 µm emission over a circular aperture of radius 14400 and subtract a background estimated as the median in an annulus region of the same area. For the 60 µm band we take into consideration the strong elongation of the

PSF along the scan direction. Therefore, we use an elliptical aperture aligned with 66

the scan direction, with semi-axes of 10800 × 7200 and an elliptical annulus region of the same area. The size of the apertures was chosen so that aperture measurements approximately reproduce the flux densities from the IRAS PSC. We made this comparison using the sample of BLAST sources with an IRAS counterpart and found no significant bias between both flux extraction methods, with scatters of 23 % at 60 µm and 26 % at 100 µm.

4.3.3 MIPS

After experimenting different photometric techniques (see discussion in section 4.1.1), we found that catalog fluxes (i.e., PSF fluxes in this case) are more appropriate in most cases. Therefore, we take the flux densities of the MIPS sources associated with

BLAST cores, as determined from the cross-correlation between the BLAST and MIPS catalogs. If more than a MIPS source is associated with a single BLAST object, their

fluxes are co-added and the total flux is considered as an upper limit in the SED fit.

However, for various 70 µm sources with a “core + halo” structure the ﬂux densities are clearly underestimated. In these cases, we performed aperture photometry and ignored the ﬂux densities from PSF photometry. We used and aperture of radius

00 Rapr = 8.75 pixel (with 1 pixel = 4 in the MIPS 70 µm image) and an annulus region delimited by two circles of radii Rsky1 = 9.75 pixel and Rsky2 = 16.25 pixel. Following the MIPS manual2 we also applied a correction factor c = 1.308. For those BLAST

2http://ssc.spitzer.caltech.edu/mips/apercorr/ 67

sources falling inside the area mapped by MIPS but without a 70 µm counterpart, we estimate upper limits using dynamical aperture photometry. As for the SIMBA map, we used a dynamical aperture of radius Rdyn = 1.3 × Dsource centered over the

BLAST coordinates and an annulus region of the same area, where now Dsource is the

BLAST deconvolved FWHM, convolved with the MIPS 70 µm beam. Flux densities at wavelengths ≤ 24 µm do not affect significantly the fit of the SED (see section 4.4.2) and therefore we do not need to estimate additional upper limits.

4.4 Spectral Energy Distribution of the Dense Cores

Dense cores present a complex structure which may depend on the environmental conditions and their evolutionary stage. In order to model the observed emission, the equation of radiative transfer (equation 2.2) should be solved for a given set of dust properties and temperature and density distributions. A number of radiative transport codes have been developed by diﬀerent authors to model starless cores (e.g.

Evans et al. 2001; Shirley et al. 2005) and accreting protostars (e.g. Whitney et al.

2004) in the case of isolated star formation. In this section we describe how we model the observed SEDs in order to estimate the physical parameters of the dense cores.

4.4.1 The SED Model

Here, we follow the approach adopted by Chapin et al. (2008) and Netterﬁeld et al.

(2009) who use a simple, single-temperature SED model to ﬁt the millimeter–MIR photometry of the BLAST cores (described in section 4.3), which will allow us to infer 68

their main physical parameters. The temperature and density variations are much less critical in starless than proto-stellar cores, and therefore an isothermal model should provide an appropriate ﬁt of the SED. Proto-stellar cores, in contrast, consist of a central IR source embedded within a cooler envelope and thus cannot be described with a single-temperature model. In such cases, ﬁtting the submillimeter/FIR part of the SED with a single-temperature model (and leaving MIR data as upper limits) one can estimate the envelope properties.

In section 2.1.4 we have calculated the observed ﬂux density of a cloud of constant temperature, density, and dust properties, subtending a solid angle ∆Ω. In the case of a dense core at a given distance, we can rewrite equation 2.3 in terms of the dust mass, Md, the dust opacity, κν, and the distance, d:

M κ F = d ν B (T ) . (4.3) ν d2 ν

The dependence of the dust opacity with radiation wavelength is conventionally written as ∼ λ−β in the range 30 µm ≤ λ ≤ 1 mm, where β is the dust emissivity index.

Then, we can express the model SED of the dense cores as a function of three ﬁtting parameters: a constant factor, A, the dust temperature, T , and the emissivity index,

β. Following the notation adopted in Chapin et al. (2008):

µ ¶ ν β Sν = A Bν(T ) , (4.4) ν0

where the constant factor is written in terms of the total core mass, Mc, the dust 69

opacity evaluated at ν0, κ0, and the distance to the object:

Mcκ0 A = 2 . (4.5) Rgdd

Here, we adopt a dust-to-gas mass ratio Rgd = 100, required in the denominator for Mc to be the total mass of the core rather than just the dust mass. The value of κ0 has been estimated in diﬀerent environments, from cold dense regions to the diﬀuse ISM, (e.g. Hildebrand 1983; Draine and Li 2007) though it is still uncertain

2 −1 by an order of magnitude. Netterﬁeld et al. (2009) have obtained κ0 = 16 cm g , evaluated at ν0 = c/250 µm by comparing the BLAST dust emission to the estimated gas mass from C18O data (Yamaguchi et al. 1999). Here, we use the same value for internal consistency.

Equation 4.4 contains three free parameters, A, β, and T , and therefore, at least three flux densitites in different wavebands are needed to perform the fit. However, there is a strong degeneracy between β and T , and there may be a large spread in the values of these parameters which can provide an equally good fit of the SED, particularly for the coldest objects non detected at IR wavelengths. Simple dust emission models predict a value β ' 2 at millimeter wavelengths, though it is thought to vary in the range 1 to 2, from warmer and denser regions to more diffuse environments.

Dupac et al. (2002) attempted to ﬁt both parameters and suggested that there is an inverse correlation between T and β. However, it has been shown recently (Shetty et al. 2009a) that this apparent correlation naturally arises from least square ﬁts due 70

to the typical uncertainties in ﬂux density measurements. Therefore, the observed T –

β correlation may not be a physical property of dust in the ISM. We thus choose to ﬁx a value β = 2 to reduce the inferred errors in temperature, though possible variations in β must be considered for a correct interpretation of the estimated temperatures.

4.4.2 The Fitting Procedure

An isothermal modified blackbody model (equation 4.4) is fitted to all of the available photometry, from millimeter to MIR wavelengths, using χ2 optimization. Then, core temperatures are obtained directly from the fit, and core masses are calculated from the fitted constant factor and the assumed dust properties (equation 4.5). The FIR luminosity of the cores is estimated by integrating the SED model (see section 2.1.4) over a wide spectral range (from 1 µm to 5 mm).

The SED ﬁtting procedure has been explained in detail in Chapin et al. (2008) and Truch et al. (2008). The χ2 function can be written in the simplest form as:

X (˜s − s )2 χ2 = i i , (4.6) σ2 i i

where si are the flux density measurements at wavelength λi, σi their associated errors, ands ˜i are the model predictions. BLAST measurements are band averaged flux densities and therefore, they must be compared with the band averaged flux densities of the SED model: 71

s˜i = Ti(ν) Sν dν , (4.7)

where, the SED model, Sν, is deﬁned in equation 4.4, and Ti(ν) is the normalized

ﬁlter transmission proﬁle for each BLAST band (see Fig. 2.4). Flux density errors,

σi, must include statistical as well as calibration uncertainties. BLAST calibration uncertainties are highly correlated, since the three bands were calibrated using just one object (Truch et al. 2009). Therefore, equation 4.6 should be written in terms of the data covariance matrix, C, in order to account for correlated errors. In matrix notation:

χ2 = (˜s − s) C−1 (˜s − s)T , (4.8)

where now s and ˜s are the vector of measured ﬂuxes and model predictions respectively, and “T ” denotes the transposed matrix. The diagonal elements of C cor-

2 respond to the variances, σi , of the ﬂux density measurements. For the BLAST data,

2 σi , are calculated as the quadrature sum of the statistical (photometric) and calibration uncertainties (10 %, 12 %, and 13 % at 250, 350, and 500 µm respectively). The off-diagonal terms are estimated from the Pearson correlation coefficients, assuming completely correlated uncertainties (see Truch et al. 2009, table 1). Errors associated with flux measurements from other instruments are considered uncorrelated.

In section 4.3 we have discussed how we estimate upper-limits for BLAST cores without any counterpart at a given wavelength, to help constraining their SEDs. In 72

addition, we noted that BLAST and MIR flux densities cannot be fitted simultaneously with a single-temperature model, as they originate from volumes of material at different temperatures. Therefore, in most cases flux density measurements at MIR wavelengths (≤ 24 µm) must be considered as upper-limits. In order to include them in the calculation of χ2 we use the “survival analysis” (see Chapin et al. 2008). Given un upper-limit, the likelihood of the model flux density is calculated and the negative log-likelihood function is added to the right side in equation 4.8, giving the final expression that we minimize to perform the fit.

As a general rule, we have used all photometry at wavelengths ≤ 24 µm as upper- limits. In addition, when more than one catalog source is associated with a BLAST core, the total ﬂux is also considered an upper-limit (see section 4.3). IRAS PSC

flux densities are estimated over an effective area which may encompass the emission of nearby sources and/or any plateau structure (a circular area of ∼ 50 diameter at 100 µm), as revealed by maps at other wavelengths. Therefore, IRAS photometry measurements are always used as upper-limits. A critical point is to determine whether 70 µm flux densities are consistent with the isothermal SED model or should be considered also as upper-limits. MIPS 70 µm data lie on the Wien’s side of the distribution, between the MIR emission from the warmer, inner parts, and the submillimeter emission radiated from a cooler envelop, and therefore, they can strongly constrain the SED. A careful analysis of preliminary SEDs and maps is needed to address this point. 73

Uncertainties for the fitting parameters, A and T , and other derived quantities such as core masses and FIR integrated fluxes, are estimated using Monte Carlo simulations. Simulated data sets are generated from realizations of Gaussian noise, including correlated and uncorrelated errors, and the χ2 minimization process is repeated for each of them. Then, each parameter is characterized by a distribution of values originated from the simulations, which is used to calculate the 68 % Bayesian confidence interval.

Finally, once the SED has been fitted, BLAST band-averaged flux measurements are corrected to quote effective flux densities at precisely 250, 350, and 500 µm, allow- ing a direct comparison with the SEDs. Color-corrected flux densities are calculated using the expression:

S S = B S˜ , (4.9) ν ˜ ν SB ˜ where Sν is the SED model ﬂux density evaluated at 250, 350, and 500 µm, SB

˜ is the band-averaged BLAST measurement, and SB is the SED model band-averaged

ﬂux density.

4.5 Bonnor-Ebert Masses

Once the physical parameters of the BLAST cores have been estimated, it is interesting to investigate their dynamical state. In this section we introduce a speciﬁc criterion for core stability, derived from a careful modelling of starless cores. 74

Observations of intensity proﬁles of nearby starless cores (see Di Francesco et al.

2007, for a review) reveal that they are far from homogeneous. Detailed modelling of these proﬁles have shown that the density structure of starless cores may be correctly described by Bonnor-Ebert (BE) spheres, i.e., non-singular solutions of the equations of hydrostatic equilibrium which can be critically stable in the presence of external pressure.

BE spheres are characterized by radial density proﬁles with a central “plateau” and a power-law decrease at large radii, being gravitationally unstable when the mean density is ∼ 14 times the density at the BE boundary radius. This stability criterion can be written in terms of a critical mass, known as the Bonnor-Ebert mass. Following, for example, Stahler and Palla (2004): µ ¶ µ ¶ T 3/2 hni −1/2 M = 1.0 M , (4.10) BE ¯ 10 K 104 cm−3

Where T is the core temperature and hni is the particle density at the edge of the core

(where the internal pressure is equal to the external pressure). However, given the

BLAST resolution and the Vela-D cloud distance, we cannot deﬁne the boundaries of the cores. We, thus, approximate hni as the mean particle density which can be estimated from the total core mass, Mc, and the linear deconvolved radius, Rdec:

Mc hni = 4 3 . (4.11) 3 π Rdec µ mH

Here we deﬁne Rdec = Ddec/2, where the linear deconvolved diameter is calculated in terms of the Vela-D cloud distance, d, and the intrinsic angular size, θdec (deﬁned in 75

equation 3.1): Ddec = d θdec. As we deﬁned in equation 1.1, mH is the mass of the hydrogen atom, and here we use a mean molecular weight per particle µ = 2.33.

The size of the cores estimated from Gaussian ﬁtting (see section 3.2.2) and the core temperatures and masses (see section 4.4) are used in this work to evaluate equation 4.10 and infer the dynamical state of the cores. 76

Chapter 5 RESULTS

5.1 SEDs and Morphology of the Cores

The SED of the 141 BLAST cores found in Vela-D have been characterized and ﬁtted across the available spectral range, from millimeter to MIR wavelengths. In addition, we have constructed multi-wavelength thumbnails for each source, providing an appropriate visual reference in terms of the millimeter-MIR catalogs and images used in this work.

Fig. 5.1 and Fig. 5.2 show two representative examples of SEDs and thumbnails of the dust cores. The ﬁrst one corresponds to one of the brightest sources in Vela-D,

BLAST J084848-433225. The black line represents the best-fit modified blackbody model, where SIMBA, BLAST, and MIPS 70 µm flux densities are fitted while IRAS and MIPS 24 µm data are left as upper limits (see section 4.4.2). From the best

fit, we obtain a temperature of 20.7 K. The grey lines indicate the 68 % confidence interval of fitting models, from which we estimate the uncertainties in the fitting parameters (constant factor and temperature; see section 4.4.1). In this case, we obtain a total mass of 16 M¯ (Mtot = Rgd × Mdust, where we adopt a dust-to-gas mass ratio Rgd = 100) and a FIR luminosity of 142 L¯. From the thumbnails, we identify 77

Figure 5.1: Left. SED of one of the brightest sources in Vela-D (see text). Red and blue colors are used to show the BLAST color-corrected flux densities and the ancillary data, respectively. The yellow asterisks represent the AKARI flux densities, though they are not used in the fit (see section 4.2.2). Right. Multi-wavelength thumbnails show the source emission at 1200, 500, 350, 250, 100, 70, 60, 24, and 8 µm from SIMBA, BLAST, IRAS, MIPS, and IRAC instruments. Each map covers an area of 20000 × 20000 centered on the BLAST source, which is represented by a white circle of diameter equal to its FWHM. The position of catalog sources at other wavelengths is shown using crosses or open squares, depending on whether they fall within the search radius or not, respectively (see section 4.2). IRAC catalog sources are shown using white dots for clarity.

Figure 5.2: Same as Fig. 5.1 for one of the coldest sources in Vela-D (see text). 78

an isolated core well detected at all wavelengths. We note that the combination of source intrinsic structure and instrumental responses results in a set of clearly diﬀerent images from 1.2 mm to 8 µm.

The second example (see Fig. 5.2) shows one of the coldest objects in the sample,

BLAST J084542-432721, with an estimated temperature of 11.7 K, and mass and FIR luminosity of 5.7 M¯ and 1.5 L¯ respectively. Here, we clearly see that the BLAST measurements reveal a turnover in the SED, contrary to the previous example of a warmer source where the BLAST bands sample the Rayleigh-Jeans tail of the SED.

As seen in the thumbnails, the BLAST core is not detected at shorter wavelengths, though it is located near a warmer source revealed by the MIPS 24 and 70 µm bands.

The IRAC images in this and the previous figure illustrate the large number of sources detected at wavelengths ≤ 8 µm. At these wavelengths even pre-main sequence field stars can be identified, making it very difficult the search for counterparts of the

BLAST cores (see section 4.2.4).

5.2 Separating Starless and Proto-Stellar Cores

One of the goals of this work is to try to identify an evolutionary sequence from starless to proto-stellar cores in terms of the estimated physical parameters. In section 1.4 we have mentioned how we can separate dense cores between these two categories by observational criteria. Here we deﬁne as proto-stellar cores those BLAST sources associated with a MIPS 24 µm counterpart(s) (see section 4.2.3), which provides evidence 79

of the presence of an embedded protostar. In addition, we use the association with

MSX point sources for those BLAST cores falling outside the MIPS 24 µm coverage.

Following this criterion we ﬁnd a fraction of starless to proto-stellar cores Ns/Np = 1.3 in Vela-D. However, we note that if we restrict the starless/proto-stellar comparison to the area covered by MIPS at 24 µm, we obtain a signiﬁcantly lower ratio Ns/Np = 0.6.

This may be a consequence of lacking sensitive Spitzer observations in part of the

Vela-D region considered here (see section 4.2.2 for a comparison between the number of MIR counterparts found by MIPS and MSX ), but also because the area mapped at 24 µm is in a later evolutionary state.

In order to check the robustness of this method, we have used our preliminary search for IRAC counterparts (see section 4.2.4) as well as other signpost of star formation such as bipolar jets and molecular outﬂows available from the literature

(Giannini et al. 2005, 2007; De Luca et al. 2007). We find that ∼ 69 % of the BLAST sources classified as starless cores within the IRAC coverage do not have an IRAC candidate counterpart. As for the cores classified as proto-stellar within the IRAC coverage, ∼ 73 % of them do have an IRAC candidate counterpart(s). Therefore, our quick search for NIR associations show that MIPS 24 µm data provide an appropriate statistical characterization of these two stages of evolution, though a more careful analysis should be done to extend the SED of the dense cores to NIR wavelengths (see

De Luca et al. 2007, for example). In addition, we have found that 13 BLAST cores, classiﬁed as proto-stellar (except BLAST J084822-433152), all are associated with H2 80

jets. The core BLAST J084822-433152 represents an example of the need of NIR observations to ensure a correct classification. Finally, we have identified the BLAST cores associated with molecular outflows in Vela-D (Elia et al. 2007), all of which have been correctly classified as proto-stellar cores except BLAST J084805-435415 (also having an IRAC counterpart).

In Fig. 5.3 we show a few examples of BLAST cores in Vela-D. Panel (a) shows that the core BLAST J084805-435415 is clearly detected by IRAS at 60 and 100 µm, and presents a strong emission at 24 and 70 µm, though it is not identiﬁed in the

MIPS catalogs. Panel (b) shows another example of a core detected by IRAS in which the presence of MIPS compact emission but no MIPS counterparts makes somewhat uncertain the starless/proto-stellar classification. Furthermore, we find 26 BLAST cores detected in the MIPS 24 µm band but not detected in the 70 µm band (see panel(c) in Fig. 5.3). In ≥ 30 % of these cases, the MIPS 70 µm map shows a compact, weak emission associated with the BLAST core, as seen in Fig. 5.3, panel (d). It seems that a combination of different instrumental sensitivity in the MIPS bands and geometrical projection effects (relevant for sources with an embedded protostar; see, e.g., Whitney et al. 2004; Robitaille et al. 2007) may be masking the 70 µm counterpart. However, we have seen (section 4.2.3) that the probability of chance associations between the BLAST and MIPS 24 µm catalogs is ∼ 10 %, and therefore we cannot discard a possible false association, especially in some of the cases with no significant 70 µm emission. In addition, we found two BLAST cores associated 81

with 24 µm sources with ﬂuxes below the threshold selected to eliminate most of the extragalactic objects (see section 4.2.3). 82 m µ Panels (c) but without MIPS m (from left to right). µ IRAS m counterparts but no 70 µ shows a core (BLAST J084928-440426) 1 K. . Panel (e) show two cores detected by (b) and m emission. µ Panels (a) m with an estimated temperature of only 13 µ m counterparts (BLAST J084805-435415 and BLAST J084902-433802 respectively; see text). µ show two cores (BLAST J084842-431735 and BLAST J085010-431704) with 24 (d) Figure 5.3: Multi-wavelengthSymbols thumbnails have of the same 5 meaning as representative in BLAST Fig. cores, 5.1. from 1200 tocounterparts, 8 the latter with signiﬁcant compact 70 70 or 24 and detected at 24 and 70 83

5.3 Distribution of Physical Parameters

Once we have separated starless and proto-stellar cores, we are interested in the analysis of their physical parameters. As the dense cores evolve towards higher central condensations, the accretion process results in an increase in luminosity, followed by a gradual decrease of the core mass (material from the envelope is transferred to a central compact object and part is also ejected into the ISM). In addition, we have seen that most of the luminosity is absorbed and re-processed by dust, leading to an increase of the dust temperature. Therefore, we could expect that the luminosity-to-mass ratio, L/M, and the temperature of the cores will increase with time, characterizing and evolutionary sequence. Fig. 5.4 shows the distribution of temperatures, masses, luminosities, and luminosity-to-mass ratios of the BLAST cores in Vela-D.

5.3.1 Temperature, Mass, and Luminosity of the BLAST Cores

Before analyzing the distribution of physical parameters, we should understand the typical uncertainties on the values estimated from the SED ﬁts. Statistical errors depend on the ﬂux density uncertainties and the number of data points used in the

ﬁt, and are calculated using Monte Carlo simulations (see section 4.4.2). We have obtained the following mean errors for the physical parameters of the dense cores:

∼ 10 % for temperature, ∼ 40 % for mass, ∼ 30 % for luminosity, and ∼ 50 % for the luminosity-to-mass ratio. In addition, there might be signiﬁcant uncertainties related to the values of distance, opacity, emissivity index, and dust-to-gas ratio used in the 84

SED ﬁtting model (see section 4.4.1). For example, if we use β = 1.5 (instead of β = 2) the estimated temperatures would increase by ≈ 10 %. On the other hand, using

2 −1 2 −1 κ0 = 10 cm g (from Hildebrand 1983, instead of κ0 = 16 cm g ) the estimated masses would result ∼ 60 % higher. In addition, a similar mass increase would result if we consider a distance d = 900 pc (with 700 ± 200 pc being the estimated distance to Vela-D). However, if we assume that these parameters are nearly constant for the population of dense cores in the Vela-D cloud, we still can compare the physical properties of starless and proto-stellar cores, taking into consideration only statistical errors.

The temperature distribution (Fig. 5.4, top-left panel) is characterized by a wide peak at T ∼ 15 K, with sharp cutoffs at the low (11–12 K) and high (18–20 K) end of the distribution. The median temperature for the whole sample is 15.4 K, higher than, for example, in the Pipe cores (Rathborne et al. 2008) and in the Vela-C cores, suggesting that cores in Vela-D are in a later evolutionary stage. However, we find little difference between starless and proto-stellar cores, with median temperatures of

14.6 K and 15.7 K respectively.

The overall mass distribution is presented in the top-right panel of Fig. 5.4, showing a clear decrease from lower to higher masses. With a median value for the whole sample of 4.7 M¯, very few cores are found with masses > 10 M¯. The range of observed values is consistent with the formation of low and intermediate-mass stars. There is no signiﬁcant diﬀerence between the median masses of starless and proto-stellar cores, 85

Figure 5.4: Temperature (top-left panel), mass (top-right panel), (FIR) luminosity (bottom-left panel), and luminosity-to-mass ratio (bottom-right panel) distributions of the BLAST cores in Vela-D (grey histograms). The distributions of starless and proto-stellar cores are shown using red dashed lines and yellow solid lines respectively. 86

with values of 4.6 M¯ and 4.8 M¯ respectively.

As for the FIR luminosities (Fig. 5.4, bottom-left panel), the distribution shows a strong peak at ∼ 4 L¯, with a median luminosity of 4.6 L¯ (11 cores with luminositites

> 40 L¯ are not shown in the histogram). Starless and proto-stellar cores seem to have diﬀerent median luminosities, 4.4 L¯ and 7.5 L¯ respectively.

The luminosity-to-mass ratio of the cores is essentially equivalent to the temperature and depends, therefore, only on the shape of the SED (being a distance- independent quantity). In Fig. 5.4, bottom-right panel, we see that starless and proto-stellar cores show a somewhat diﬀerent L/M distribution, having median val-

−1 −1 ues of 1.0 L¯M¯ and 1.6 L¯M¯ respectively.

5.3.2 Color–Color, Luminosity–Mass, and Spatial Distribution Plots

In order to further characterize the core population in Vela-D, we have investigated the distribution and potential correlation between diﬀerent parameters by means of

2-dimensional plots. In this section we present the color–color, luminosity–mass, and spatial distribution plots for the BLAST cores in Vela-D (see Fig. 5.5, 5.6, and 5.7).

Fig. 5.5 shows a color–color plot where we represent the ﬂux density ratio F250/F350 as a function of F250/F500, which are quantities depending on the core temperature.

We see a large concentration of BLAST sources located near the line of constant emissivity index β = 2, being consistent with the adopted value in our SED-ﬁtting model (∼ 44 % and 56 % of the cores are located above and below the line β = 2). 87

Figure 5.5: Color–color plot for the BLAST cores in Vela-D (see text). The ratios F250/F350 and F250/F500 have been calculated from the color-corrected ﬂux densities (see section 4.4.2). Modiﬁed blackbody models with β = 0, 1, 2, 3, and temperatures ranging from 3 to 100 K are overplotted as dashed lines. Starless and pre-stellar cores are represented by open circles and crosses respectively. The error bars account for statistical as well as calibration uncertainties and are shaded for clarity. 88

Starless and proto-stellar cores show a similar distribution in the color–color plot, though the latter tend to be located at lower β values: the ratio of the number of sources consistent with β < 2 to β > 2 is ∼ 1.5 for proto-stellar cores, compared to a ratio of ∼ 1.1 for starless cores.

A plot of the luminosity of the cores as a function of mass, the luminosity–mass plot or simply L−M plot, is shown in Fig. 5.6. The plot shows a high concentration of

−1 sources at temperatures . 15 K, or equivalently . 0.65 L¯ M¯ , consistent with the temperature and luminosity-to-mass ratio distributions shown in Fig. 5.4 (bottom- right panel). If the BLAST cores have actually different temperatures they would move along their constant flux density locus (evaluated at 250 µm in the figure), down and right if they are colder than the estimated temperatures. We note that a small group of nine bright objects (F250 > 100 Jy) appear clearly separated from the main concentration of sources, which have typical flux densities . 30 Jy at 250 µm. They are mostly proto-stellar cores, detected by AKARI and MIPS (at 24 and 70 µm), and

ﬁve of them correspond to known IRS sources (Liseau et al. 1992) associated with small NIR clusters.

In Fig. 5.7 we present the spatial distribution of dense cores in Vela-D (also shown in Fig. 3.4) with additional information of core temperatures and masses, as well as the starless/proto-stellar core classiﬁcation. This ﬁgure shows some interesting features.

For example, we see that the BLAST cores associated with IRS sources (shown in

Fig. 3.6) and thus, classiﬁed as proto-stellar cores, are among the warmer and more 89

Figure 5.6: FIR luminosity versus mass for the BLAST cores in Vela-D. Error bars show the uncertainties in luminosity and mass estimated from the range of SED models consistent with the available photometry (see section 4.4.2). Errors involving the adopted values of distance, β, κ0, and gas-to-dust mass ratio are not included. The dashed lines are loci at constant T = 10 to 30 K assuming a modiﬁed blackbody model with β = 2. Also shown are the loci at constant 250 µm ﬂux density, evaluated from 3 to 1000 Jy, using the same model. The “+”, “¦”, and “¤” signs mark the BLAST sources without AKARI, MIPS 24 µm, or MIPS 70 µm counterparts respectively. 90

Figure 5.7: Spatial distribution of cores in Vela-D. Starless and proto-stellar cores are represented by open circles and crosses respectively, overlaid with a grey-scale image of the BLAST 250 µm map (using the same saturation levels as Fig. 3.4). Increasing core temperatures (arranged in four intervals delimited by T = 13, 15, and 17 K) are shown using red, yellow, green, and blue colors. Core masses are indicated by the size of the signs, circles and crosses, separated in three mass intervals: M < 2 M¯, 2 < M < 10 M¯, and M > 10 M¯. 91

Starless Proto-stellar

All M > 4.2 M¯ M > 11 M¯ All M > 4.2 M¯ M > 11 M¯

T [K] 14.6 13.1 12.9 15.7 15.5 17.2

Mcore [M¯] 4.6 7.0 15.3 4.8 8.7 16.5

Lﬁr [L¯] 4.4 4.5 7.0 7.5 10.9 84.7

−1 Lﬁr/Mc [L¯M¯ ] 1.0 0.5 0.5 1.6 1.5 2.7

Table 5.1: Temperature, mass, luminosity, and luminosity-to-mass ratio median values of starless and proto-stellar cores in Vela-D, evaluated for the whole sample (All), and the fraction of cores above the completeness limits M > 4.2 M¯ and M > 11 M¯ for sources with T > 12 K and T > 10 K respectively.

massive cores in Vela-D (these are some of the brightest sources shown in Fig. 5.6).

We see that the more massive cores are rarely found in isolation, often being part of small groups of sources and associated with strong dust continuum emission. In contrast, BLAST sources found in isolation tend to be low-mass cores. We also note an interesting group of sources found along the ﬁlament at coordinates l ∼ 262.88◦, b ∼ 0.27◦, that is composed of mainly cold, massive starless cores.

5.4 Do Starless and Proto-Stellar Cores Have Diﬀerent Properties?

As we have seen, it is not completely clear that starless and proto-stellar cores can be separated according to their observed physical parameters. They are placed in similar regions in the L − M and color–color plots, and the histograms do not show a sharp separation between the two types of core, though the proto-stellar cores appear to be somewhat warmer than the starless cores. 92

However, if we restrict our analysis to mass ranges for which our source-extraction technique is more complete (Netterfield et al. 2009), we find that the temperature difference between starless and proto-stellar cores becomes more significant. For example, sources with masses > 11 M¯ (the completeness level for T > 10 K) have median temperatures of 12.9 and 17.2 K for starless and proto-stellar cores respectively (see table 5.1). In addition, the differences in median luminosities and L/M ratios between starless and proto-stellar cores, which were mentioned earlier, become larger if we consider only those cores above this completeness level. Thus, starless

−1 cores with masses > 11 M¯ have medians of 7.0 L¯ and 0.5 L¯M¯ respectively, sig- niﬁcantly lower than the median luminosity and median L/M ratio of 84.7 L¯ and

−1 2.7 L¯M¯ found for proto-stellar cores. Remarkably, the diﬀerence between median masses of starless and pro-stellar cores in Vela-D is not signiﬁcant even for the sample of cores above the completeness level.

These results suggest that starless and proto-stellar cores are characterized by different values of temperature and luminosity-to-mass ratio. However, median values may be affected by the low-number statistics at the high mass end of the distribution, as we find only 29 BLAST cores with masses > 11 M¯. Therefore, we should be careful interpreting the observed differences between starless and proto-starless cores.

While we do observe an increase in median temperature and L/M ratio from starless to proto-stellar cores and thus, consistent with an evolutionary sequence, we also note that there should be a smooth transition between these phases. This is in fact 93

suggested by the partial overlap between the distribution of physical parameters of starless and proto-stellar cores.

Besides the uncertainties introduced by completeness and low-number statistics, the different instrumental resolution and sensitivities may affect the distinction between starless and proto-stellar cores. An example of proto-stellar core with an unex- pected low temperature is shown in Fig. 5.3, panel (e). It is clearly detected by MIPS at 24 and 70 µm, but its temperature from the best-fit SED is only 13.14 K. Such a cold source is not expected to emit a significant flux density at wavelengths ≤ 70 µm and therefore we can considerer two possible scenarios: (i) either the 70 µm flux density is emitted by a warmer central core and therefore the emission detected by BLAST comes from a colder envelope; or (ii) there are two separate nearby sources, a cold, starless core detected only by BLAST, and a warmer proto-stellar source detected by

MIPS. With the current available data we are not able to discriminate between these two possible scenarios and, while recognizing some ambiguity in this and similar cases, we opt to be consistent with our association criteria and classify as proto-stellar those cores with 24 µm counterparts.

We also ﬁnd starless cores signiﬁcantly warmer than the typical temperatures we could expect (> 15 K; see, e.g., Di Francesco et al. 2007). This might be a consequence of lacking enough sensitivity at MIR wavelengths to detect the most embedded protostars. On the other hand, it has been shown (e.g., Shirley et al. 2005) that starless cores may be colder on the inside and warmer on the outside, because they are heated 94

externally by the interstellar radiation ﬁeld and they lack an internal source of heating. In fact, we have seen that the temperature of the most massive starless cores

(M > 11 M¯) seem to be lower than that of the low-mass starless cores (see table 5.1 and Fig. 5.7), being consistent with the dust temperature depending on the degree of shielding from an external radiation ﬁeld (as shown by radiative transfer calculations).

Therefore, the higher temperature of some starless cores (especially the less-shielded low-mass and/or isolated cores) could be the consequence of a warmer surrounding medium1 heated by the interstellar radiation ﬁeld.

The eﬀect of temperature variations along the line of sight has been analyzed elsewhere (Shetty et al. 2009b) and could have implications in our results obtained from single-temperature models. Furthermore, the use of a constant emissivity index

(β = 2) in our SED ﬁtting model may contribute to mask possible temperature variations between starless and proto-stellar cores. We note that proto-stellar cores, since they are warmer and denser than starless cores , could be better characterized by an emissivity index β < 2, thus resulting in higher model ﬁtting temperatures.

5.5 Mass Spectrum

We have mentioned in section 1.4 that the mass distribution of pre-stellar cores, i.e., the core mass function (CMF), is one of the key concepts which can be used to discriminate between diﬀerent star formation models. Furthermore, recent studies

1Schlegel et al. (1998) estimated an equilibrium temperature of ∼ 18 K for low density interstellar dust in the Galactic plane. 95

Figure 5.8: Combined starless and proto-stellar CMF of the BLAST cores in Vela-D. Error bars show the Poisson uncertainty for each bin. The best-ﬁt to the CMF (a power law with slope α = −2.2) is represented by the solid line. The vertical dot- dashed and dotted lines show the completeness limits for sources warmer than 10 and 12 K respectively. The turnover seen at the low mass end is likely to be a consequence of incompleteness.

suggest that the distribution of stellar masses at birth, i.e., the IMF, is determined at the pre-stellar stage. Thus, a considerable eﬀort is currently being done to constrain the IMF both theoretically and observationally.

We have investigated the combined mass spectrum in Vela-D, including starless2 and proto-stellar cores (see Fig. 5.8). Individual core masses are placed in logarith- mically spaced bins so that we can fit a standard power law (dN/dM ∝ M α) to the resulting mass function. The error bars are estimated from the Poisson uncertainty for each bin. Then, fitting the CMF for masses M > 4 M¯ we find a slope α = −2.2 with a correlation coefficient r = 0.99. However, we note that the best-fit power law

2We are thus using starless cores that may not be pre-stellar (see section 5.6). 96

is somewhat dependent on the histogram binning and therefore there is no unique slope α for a given mass distribution. In fact, by varying the bin width from ∼ 1.7 to

4.2 M¯, we ﬁnd values from α ' −2.1 to α ' −2.5. Therefore, we empirically assign the slope α = −2.3 ± 0.2 to the CMF for masses M > 4 M¯ in Vela-D.

This value is very similar to the slopes found in other star forming regions such as

Orion (α = −2.35 in the mass range M > 2.4 M¯; Nutter and Ward-Thompson 2007) and Perseus, Serpens and Ophiucus (α = −2.3 ± 0.4 in the mass range M > 0.8 M¯;

Enoch et al. 2008). In contrast, we have found a CMF steeper than the previous mass functions obtained in Vela-D, with slopes of α ∼ −1.4 to −1.9 (Massi et al. 2007) and

α ∼ −1.3 to −2.0 (Elia et al. 2007), which are probably aﬀected by the smaller source samples used to compute the CMF and the inability to detect many of the coldest, low-mass sources. We note that evaluating the CMF only for the BLAST cores with

SIMBA counterparts we ﬁnd a slope consistent with the range of values reported by

Massi et al. (2007). On the other hand, Netterfield et al. (2009) have found a steeper slope of α = −2.77 ± 0.16 in the mass range M > 14 M¯ (the completeness limit for sources warmer than ∼ 8.5 K) for the BLAST cores in the VMR. Thus, it seems that the slope of the CMF depends on the mass range used to evaluate the mass function as well as the statistical significance of the sample, besides other effects such as variations of β among different cores or temperature variations inside them. In addition, we note that the VMR overall slope may also be affected by the great variation in physical conditions and by distance effects. Netterfield et al. (2009) have shown that cold 97

cores follow a steeper mass function than warmer cores. In that sense, their ﬁnding of

α = −2.55 ± 0.20 for the BLAST cores in Vela-C is consistent with our result, being the Vela-D cloud in a later evolutionary phase.

The CMFs computed in Vela-D and other star-forming regions have similar shapes to the Salpeter IMF (Salpeter 1955) for main-sequence stars in the solar neighbour- hood, characterized by the slope α = −2.35. All of these observations suggest that the shape of the IMF is a direct consequence of the CMF. However, compared to previous measurements of the CMF, our result is based on a robust determination of the core temperature.

5.6 Dynamical State of the Cores

Ideally, the CMF should be computed for pre-stellar cores only, rather than the combined population of starless and proto-stellar cores. However, the relatively small number of sources in the two sub-samples prevents us from evaluating separately the

CMF. In addition, based on our current data, we cannot distinguish the subset of starless cores which are gravitationally bound and are therefore truly pre-stellar, which would require the analysis of the velocity dispersion within the cores using molecular line observations. However, in section 1.4 we have seen that an indirect separation between starless and pre-stellar cores could be attempted in terms of the comparison between the observed core masses and the theoretical masses for cores in critical equilibrium. We have used the BE mass criterion introduced in section 4.5 to analyze the 98

Figure 5.9: Ratio of the total core mass to the Bonnor-Ebert mass plotted as a function of the total mass. BE masses are calculated using the core temperature and the average density, estimated from the deconvolved size and mass of the BLAST cores (see section 4.5). Starless and proto-stellar cores are represented by red and black symbols respectively.

dynamical state of the cores.

Evaluating equation 4.11 for those cores with deconvolved diameter larger than the beam (thus, eliminating those cores with unrealistically small sizes), we have obtained a median particle density of 1.6×104 cm−3, with values ranging from 3.7×103 cm−3 to

1.5×105 cm−3. Then, the BE masses of these cores are estimated using equation 4.10.

Fig. 5.9 shows a plot of the ratio of the total core mass to the BE mass, Mc/MBE, as a function of core masses. If we consider that the dense cores are approximately spherical in shape and assuming that they are supported against gravity only by 99

thermal pressure, then the horizontal line drawn at Mc/MBE = 1 would separate those cores in equilibrium from those in non-equilibrium configurations. Thus, dense cores with Mc/MBE > 1 could be potentially collapsing to form stars. However, we note that elongated cores can be more stable than spherical BE-like cores. In addition, we have seen that non-thermal support due to magnetic and/or turbulent pressures may have significant effects on the evolution of cores. Therefore, the Mc > MBE criterion should be regarded as a necessary but not sufficient condition for core collapse. In addition, we note that the ratio Mc/MBE critically depends on how Mc/Rdec is observationally defined, which is clearly instrumental dependent.

We see in Fig. 5.9 that the Mc/MBE ratio seems to be proportional to Mc, and we note that this is a consequence of its operational deﬁnition, giving that Mc/MBE ∝

3/2 (Mc/T Rdec) . This ﬁgure suggests that the low-mass cores observed by BLAST are more likely to be transient objects rather than gravitationally bound cores, being the transition from Mc/MBE < 1 to Mc/MBE > 1 in the mass range Mc ' 3 to 7 M¯.

We find a median value MBE = 4.6 M¯ or Mc/MBE = 1.1 for all cores in Vela-D, which is near the critical value. If we restrict our analysis to the more massive cores, with masses Mc > 14 M¯, we find a median value significantly higher, Mc/MBE =

8.6, indicating that high-mass cores are more likely to be gravitationally unstable.

Furthermore, we ﬁnd very little diﬀerence between starless and proto-stellar cores, with ∼ 50 % and ∼ 60 % of them being consistent with Mc/MBE > 1, respectively.

The ratio Mc/MBE < 1 for ∼ 40 % of the proto-stellar cores could be a consequence of 100

the presence of a central YSO having accreted a signiﬁcant fraction of the core mass and increased the temperature of the envelope. On the other hand, the fraction of starless cores with Mc/MBE > 1 could be classiﬁed as pre-stellar. However, given the uncertainties in the estimated physical parameters and, even more importantly, the various assumptions required by the BE criterion, the distinction between starless and pre-stellar is still quite uncertain.

5.7 Summary and Conclusions

In this work, we have presented a detailed analysis of the earliest stages of star formation in the Vela-D region. Using the BLAST maps at 250, 350, and 500 µm and additional data from previous MIR, FIR, and millimeter continuum observations, we have investigated the physical nature of the dense core population in the Vela-D molecular cloud. We summarize our results as follows:

• We have found that ∼ 30 % of the BLAST cores falling within the area covered

by SIMBA (at 1200 µm) and MIPS (at 24 and 70 µm) are not detected at any of

these wavelengths, thus demonstrating the importance of observing the dense,

cold cores near the peak of the SED.

• Combining the available photometry from millimeter to MIR wavelengths we

have constrained and ﬁtted the SED of the cores found by BLAST, thus obtain-

ing more accurate estimates of their physical parameters, particularly for the

warmer sources. 101

• Based on the associations of BLAST sources with YSO we have separated the

population of cores between starless and proto-stellar cores, ﬁnding a ratio

Ns/Np = 1.3 in Vela-D. However, we note that the sensitivity of the available

MIR data critically aﬀects the intrinsic starless to proto-stellar core ratio for a

given star-forming region.

• We ﬁnd a smooth transition from starless to proto-stellar cores in terms of

their physical parameters. Proto-stellar cores have a median temperature and

a median L/M ratio higher than starless cores, suggesting that they are in a

later evolutionary stage. The similarity between the median masses of starless

and proto-stellar cores suggests that the mass is not signiﬁcantly aﬀected by the

transition between the two phases.

• The CMF for all the BLAST cores in Vela-D is characterized by a power law

with slope α = −2.3 ± 0.2, consistent with other millimeter surveys and the

Salpeter IMF.

• The median ratio of the observed masses to the BE masses is found to be near

the critical value, Mc/MBE = 1.1. The Mc/MBE versus Mc distribution suggests

that the observed low-mass cores tend to be gravitationally unbound. 102

BIBLIOGRAPHY

J. F. Alves, C. J. Lada, and E. A. Lada. Internal structure of a cold dark molecular cloud inferred from the extinction of background starlight. Nature, 409:159–161, Jan. 2001.

P. Andre, D. Ward-Thompson, and M. Barsony. Submillimeter continuum observations of Rho Ophiuchi A - The candidate protostar VLA 1623 and prestellar clumps. ApJ, 406:122–141, Mar. 1993.

P. Andre, D. Ward-Thompson, and M. Barsony. From Prestellar Cores to Proto- stars: the Initial Conditions of Star Formation. Protostars and Planets IV, pages 59–+, May 2000.

P. Andr´e,S. Basu, and S.-i. Inutsuka. The Formation and Evolution of Prestellar Cores. ArXiv e-prints, Jan. 2008.

H. H. Aumann, J. W. Fowler, and M. Melnyk. A maximum correlation method for image construction of IRAS survey data. Astron. J., 99:1674–1681, May 1990.

C. A. Beichman, P. C. Myers, J. P. Emerson, S. Harris, R. Mathieu, P. J. Benson, and R. E. Jennings. Candidate solar-type protostars in nearby molecular cloud cores. ApJ, 307:337–349, Aug. 1986.

C. A. Beichman, G. Neugebauer, H. J. Habing, P. E. Clegg, and T. J. Chester, editors. Infrared astronomical satellite (IRAS) catalogs and atlases. Volume 1: Explanatory supplement, volume 1, 1988.

W. B. Bonnor. Boyle’s Law and gravitational instability. MNRAS, 116:351–+, 1956.

S. Bontemps, P. Andr´e, A. A. Kaas, L. Nordh, G. Olofsson, M. Huldtgren, A. Abergel, J. Blommaert, F. Boulanger, M. Burgdorf, C. J. Cesarsky, D. Ce- sarsky, E. Copet, J. Davies, E. Falgarone, G. Lagache, T. Montmerle, M. P´erault, P. Persi, T. Prusti, J. L. Puget, and F. Sibille. ISOCAM observations of the rho Ophiuchi cloud: Luminosity and mass functions of the pre-main sequence embedded cluster. A&A, 372:173–194, June 2001. 103

E. L. Chapin, P. A. R. Ade, J. J. Bock, C. Brunt, M. J. Devlin, S. Dicker, M. Griﬃn, J. O. Gundersen, M. Halpern, P. C. Hargrave, D. H. Hughes, J. Klein, G. Marsden, P. G. Martin, P. Mauskopf, C. B. Netterﬁeld, L. Olmi, E. Pas- cale, G. Patanchon, M. Rex, D. Scott, C. Semisch, M. D. P. Truch, C. Tucker, G. S. Tucker, M. P. Viero, and D. V. Wiebe. The Balloon-borne Large Aper- ture Submillimeter Telescope (BLAST) 2005: A 4 deg2 Galactic Plane Survey in Vulpecula (l = 59deg). ApJ, 681:428–452, July 2008.

M. De Luca. The Vela Molecular Ridge Young Population: from Pre-stellar Condensations to Young Clusters. PhD thesis, Universit`adegli Studi di Roma “Tor Vergata”, 2008.

M. De Luca, T. Giannini, D. Lorenzetti, F. Massi, D. Elia, and B. Nisini. Near- and far-infrared counterparts of millimeter dust cores in the Vela molecular ridge cloud D. A&A, 474:863–872, Nov. 2007.

J. Di Francesco, N. J. Evans, II, P. Caselli, P. C. Myers, Y. Shirley, Y. Aikawa, and M. Tafalla. An Observational Perspective of Low-Mass Dense Cores I: Internal Physical and Chemical Properties. In B. Reipurth, D. Jewitt, and K. Keil, editors, Protostars and Planets V, pages 17–32, 2007.

B. T. Draine and A. Li. Infrared Emission from Interstellar Dust. IV. The Silicate- Graphite-PAH Model in the Post-Spitzer Era. ApJ, 657:810–837, Mar. 2007.

X. Dupac, M. Giard, J.-P. Bernard, N. Boudet, J.-M. Lamarre, C. M´eny, F. Pajot, E.´ Pointecouteau, I. Ristorcelli, G. Serra, B. Stepnik, and J.-P. Torre. Submil- limeter dust emission of the M 17 complex measured with PRONAOS. A&A, 392:691–698, Sept. 2002.

M. P. Egan, R. F. Shipman, S. D. Price, S. J. Carey, F. O. Clark, and M. Cohen. A Population of Cold Cores in the Galactic Plane. ApJ, 494:L199+, Feb. 1998.

D. Elia, F. Massi, F. Strafella, M. De Luca, T. Giannini, D. Lorenzetti, B. Nisini, L. Campeggio, and B. M. T. Maiolo. Mapping Molecular Emission in Vela Molec- ular Ridge Cloud D. ApJ, 655:316–331, Jan. 2007.

M. L. Enoch, N. J. Evans, II, A. I. Sargent, J. Glenn, E. Rosolowsky, and P. My- ers. The Mass Distribution and Lifetime of Prestellar Cores in Perseus, Serpens, and Ophiuchus. ApJ, 684:1240–1259, Sept. 2008.

N. J. Evans, II, J. M. C. Rawlings, Y. L. Shirley, and L. G. Mundy. Tracing the Mass during Low-Mass Star Formation. II. Modeling the Submillimeter Emission from Preprotostellar Cores. ApJ, 557:193–208, Aug. 2001. 104

T. Giannini, F. Massi, L. Podio, D. Lorenzetti, B. Nisini, A. Caratti o Garatti, R. Liseau, G. Lo Curto, and F. Vitali. Star formation in the Vela Molecular Clouds: A new protostar powering a bipolar jet. A&A, 433:941–954, Apr. 2005.

T. Giannini, D. Lorenzetti, M. De Luca, B. Nisini, M. Marengo, L. Allen, H. A. Smith, G. Fazio, F. Massi, D. Elia, and F. Strafella. Spitzer MIPS Survey of the Young Stellar Content in the Vela Molecular Ridge-D. ApJ, 671:470–482, Dec. 2007.

R. H. Hildebrand. The Determination of Cloud Masses and Dust Characteristics from Submillimetre Thermal Emission. QJRAS, 24:267–+, Sept. 1983.

C. J. Lada and B. A. Wilking. The nature of the embedded population in the Rho Ophiuchi dark cloud - Mid-infrared observations. ApJ, 287:610–621, Dec. 1984.

R. Liseau, D. Lorenzetti, B. Nisini, L. Spinoglio, and A. Moneti. Star formation in the VELA molecular clouds. I - The IRAS-bright Class I sources. A&A, 265: 577–596, Nov. 1992.

D. Lorenzetti, L. Spinoglio, and R. Liseau. Star Formation in the VELA Molec- ular Clouds - Part Two - the Luminosity Function of the Class-I Sources. A&A, 275:489–+, Aug. 1993.

D. Lorenzetti, T. Giannini, F. Vitali, F. Massi, and B. Nisini. Near-infrared h2 and [fe ii] imaging and spectroscopy of new jets in the vela molecular clouds. ApJ, 564:839–852, Jan. 2002. doi: 10.1086/324305.

F. Massi, D. Lorenzetti, T. Giannini, and F. Vitali. Star formation in the Vela molecular clouds. IV. Young embedded star clusters towards D-cloud class I sources. A&A, 353:598–616, Jan. 2000.

F. Massi, D. Lorenzetti, and T. Giannini. Star formation in the Vela molecular clouds. V. Young stellar objects and star clusters towards the C-cloud. A&A, 399:147–167, Feb. 2003.

F. Massi, M. de Luca, D. Elia, T. Giannini, D. Lorenzetti, and B. Nisini. Star formation in the Vela molecular ridge. Large scale mapping of cloud D in the mm continuum. A&A, 466:1013–1023, May 2007.

J. May, D. C. Murphy, and P. Thaddeus. A wide latitude CO survey of the third galactic quadrant. A&AS, 73:51–83, Apr. 1988. 105

C. F. McKee and E. C. Ostriker. Theory of Star Formation. ARAA, 45:565–687, Sept. 2007.

Y. Moriguchi, N. Yamaguchi, T. Onishi, A. Mizuno, and Y. Fukui. A 12CO (J = 1 – 0) Survey of Molecular Clouds toward the Vela Supernova Remnant with NANTEN. PASJ, 53:1025–1036, Dec. 2001.

F. Motte, S. Bontemps, P. Schilke, N. Schneider, K. M. Menten, and D. Brogui`ere. The earliest phases of high-mass star formation: a 3 square degree millimeter continuum mapping of Cygnus X. A&A, 476:1243–1260, Dec. 2007.

T. C. Mouschovias and G. E. Ciolek. Magnetic Fields and Star Formation: A Theory Reaching Adulthood. In C. J. Lada and N. D. Kylaﬁs, editors, NATO ASIC Proc. 540: The Origin of Stars and Planetary Systems, pages 305–+, 1999.

D. C. Murphy and J. May. Molecular clouds in VELA. A&A, 247:202–214, July 1991.

C. B. Netterﬁeld, P. A. R. Ade, J. J. Bock, E. L. Chapin, M. J. Devlin, M. Griﬃn, J. O. Gundersen, M. Halpern, P. C. Hargrave, D. H. Hughes, J. Klein, G. Mars- den, P. G. Martin, P. Mauskopf, L. Olmi, E. Pascale, G. Patanchon, M. Rex, A. Roy, D. Scott, C. Semisch, N. Thomas, M. D. P. Truch, C. Tucker, G. S. Tucker, M. P. Viero, and D. V. Wiebe. BLAST: The Mass Function, Lifetimes, and Properties of Intermediate Mass Cores from a 50 Square Degree Submillime- ter Galactic Survey in Vela (l = 265). ArXiv e-prints, Apr. 2009.

D. Nutter and D. Ward-Thompson. A SCUBA survey of Orion - the low-mass end of the core mass function. MNRAS, 374:1413–1420, Feb. 2007.

L. Olmi. Int. J. IR Millimeter Waves, 22:791, 2001.

L. Olmi. Optical designs for submillimeter-wave spherical-primary (sub)orbital telescopes and novel optimization techniques. In H. A. MacEwen, editor, Soci- ety of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, volume 4849 of Presented at the Society of Photo-Optical Instrumentation Engineers (SPIE) Conference, pages 245–256, Dec. 2002.

L. Olmi. The next generation of large millimeter- and submillimeter-wave radio telescopes. Recent Res. Devel. A&A, 1:433–456, 2003.

L. Olmi et al. In preparation, 2009. 106

E. Pascale, P. A. R. Ade, J. J. Bock, E. L. Chapin, J. Chung, M. J. Devlin, S. Dicker, M. Griﬃn, J. O. Gundersen, M. Halpern, P. C. Hargrave, D. H. Hughes, J. Klein, C. J. MacTavish, G. Marsden, P. G. Martin, T. G. Mar- tin, P. Mauskopf, C. B. Netterﬁeld, L. Olmi, G. Patanchon, M. Rex, D. Scott, C. Semisch, N. Thomas, M. D. P. Truch, C. Tucker, G. S. Tucker, M. P. Viero, and D. V. Wiebe. The Balloon-borne Large Aperture Submillimeter Telescope: BLAST. ApJ, 681:400–414, July 2008.

G. Patanchon, P. A. R. Ade, J. J. Bock, E. L. Chapin, M. J. Devlin, S. Dicker, M. Griﬃn, J. O. Gundersen, M. Halpern, P. C. Hargrave, D. H. Hughes, J. Klein, G. Marsden, P. G. Martin, P. Mauskopf, C. B. Netterﬁeld, L. Olmi, E. Pascale, M. Rex, D. Scott, C. Semisch, M. D. P. Truch, C. Tucker, G. S. Tucker, M. P. Viero, and D. V. Wiebe. SANEPIC: A Mapmaking Method for Time Stream Data from Large Arrays. ApJ, 681:708–725, July 2008.

J. M. Rathborne, C. J. Lada, A. A. Muench, J. F. Alves, and M. Lombardi. The nature of the dense core population in the pipe nebula: A survey of nh3, ccs, and hc5n molecular line emission. ApJS, 174:396–425, Feb. 2008.

T. P. Robitaille, B. A. Whitney, R. Indebetouw, and K. Wood. Interpreting Spectral Energy Distributions from Young Stellar Objects. II. Fitting Observed SEDs Using a Large Grid of Precomputed Models. ApJS, 169:328–352, Apr. 2007.

E. E. Salpeter. The Luminosity Function and Stellar Evolution. ApJ, 121:161–+, Jan. 1955.

D. J. Schlegel, D. P. Finkbeiner, and M. Davis. Maps of Dust Infrared Emission for Use in Estimation of Reddening and Cosmic Microwave Background Radiation Foregrounds. ApJ, 500:525–+, June 1998.

R. Shetty, J. Kauﬀmann, S. Schnee, and A. A. Goodman. The Eﬀect of Noise on the Dust Temperature-Spectral Index Correlation. ApJ, 696:676–680, May 2009a.

R. Shetty, J. Kauﬀmann, S. Schnee, A. A. Goodman, and B. Ercolano. The Eﬀect of Line-of-Sight Temperature Variation and Noise on Dust Continuum Observations. ApJ, 696:2234–2251, May 2009b.

Y. L. Shirley, M. K. Nordhaus, J. M. Grcevich, N. J. Evans, II, J. M. C. Rawlings, and K. Tatematsu. Modeling the Physical Structure of the Low-Density Pre- Protostellar Core Lynds 1498. ApJ, 632:982–1000, Oct. 2005. 107

F. H. Shu, F. C. Adams, and S. Lizano. Star formation in molecular clouds - Observation and theory. ARAA, 25:23–81, 1987.

F. H. Shu, Z.-Y. Li, and A. Allen. Does Magnetic Levitation or Suspension Deﬁne the Masses of Forming Stars? ApJ, 601:930–951, Feb. 2004.

S. W. Stahler and F. Palla. The Formation of Stars. Willey-VCH, 2004.

F. Strafella et al. In preparation, 2009.

M. D. P. Truch, P. A. R. Ade, J. J. Bock, E. L. Chapin, M. J. Devlin, S. Dicker, M. Griﬃn, J. O. Gundersen, M. Halpern, P. C. Hargrave, D. H. Hughes, J. Klein, G. Marsden, P. G. Martin, P. Mauskopf, C. B. Netterﬁeld, L. Olmi, E. Pascale, G. Patanchon, M. Rex, D. Scott, C. Semisch, C. Tucker, G. S. Tucker, M. P. Viero, and D. V. Wiebe. The Balloon-borne Large Aperture Submillimeter Telescope (BLAST) 2005: Calibration and Targeted Sources. ApJ, 681:415–427, July 2008.

M. D. P. Truch, P. A. R. Ade, J. J. Bock, E. L. Chapin, M. J. Devlin, S. Dicker, M. Griﬃn, J. O. Gundersen, M. Halpern, P. C. Hargrave, D. H. Hughes, J. Klein, G. Marsden, P. G. Martin, P. Mauskopf, L. Moncelsi, C. B. Netterﬁeld, L. Olmi, E. Pascale, G. Patanchon, M. Rex, D. Scott, C. Semisch, C. Tucker, G. S. Tucker, M. P. Viero, and D. V. Wiebe. The Balloon-borne Large Aperture Submillimeter Telescope (BLAST) 2006: Calibration and Flight Performance. Submitted to ApJ, 2009.

D. Ward-Thompson. Isolated Star Formation: From Cloud Formation to Core Collapse. Science, 295:76–81, Jan. 2002.

D. Ward-Thompson, F. Motte, and P. Andre. The initial conditions of isolated star formation - III. Millimetre continuum mapping of pre-stellar cores. MNRAS, 305:143–150, May 1999.

D. Ward-Thompson, P. Andr´e,R. Crutcher, D. Johnstone, T. Onishi, and C. Wil- son. An Observational Perspective of Low-Mass Dense Cores II: Evolution To- ward the Initial Mass Function. In B. Reipurth, D. Jewitt, and K. Keil, editors, Protostars and Planets V, pages 33–46, 2007.

E. W. Weisstein. Millimeter/Submillimeter Fourier Transform Spectroscopy of Jovian Planet Atmospheres. PhD thesis, California Institute of Technology, 1996.

B. A. Whitney, R. Indebetouw, J. E. Bjorkman, and K. Wood. Two-Dimensional Radiative Transfer in Protostellar Envelopes. III. Eﬀects of Stellar Temperature. ApJ, 617:1177–1190, Dec. 2004. 108

N. Yamaguchi, N. Mizuno, H. Saito, K. Matsunaga, A. Mizuno, H. Ogawa, and Y. Fukui. A Study of Dense Molecular Gas and Star Formation toward the Vela Molecular Ridge with NANTEN. PASJ, 51:775–790, Dec. 1999.

I. Yamamura. AKARI Far-Infrared All-Sky Survey point source catalgoue. In 37th COSPAR Scientiﬁc Assembly, volume 37 of COSPAR, Plenary Meeting, pages 3529–+, 2008.

H. Zinnecker and H. W. Yorke. Toward Understanding Massive Star Formation. ARAA, 45:481–563, Sept. 2007.