. ..
a e .5 2. ..

.‘I aim . 2.11

sfihamn .131... .

 

.. \hdumwmn.
9.1.3...
as. ,

u h.
1:45.94»: in

.595: .1
ti L...

 

 

 

3 LIBRARY
70.x? MIL J '. i dtate
v-W University

 

 

This is to certify that the
dissertation entitled

SCALING STELLAR FEEDBACK: A STUDY OF THE
PHYSICAL PROCESSES INVOLVED IN STAR-FORMING
REGIONS OF VASTLY DIFFERENT SIZES

presented by
Eric W. Pellegrini

has been accepted towards fulfillment
of the requirements for the 3

PhD. degree in Physics and Astronomy {

LCx C4 Erukk —

7‘ Major Professor’s Signature

3d(;/ 74 2009

I
Date '

 

 

Doctoral Dissertation
BOOKPLATE INSTRUCTIONS

MSU is an Affirmative Action/Equal Opportunity Employer

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5108 K:IProj/Acc&Pres/CIRCIDaIeDue.indd

 

 

SCALING STELLAR FEEDBACK: A STUDY OF THE PHYSICAL PROCESSES
INVOLVED IN STAR-FORMING REGIONS OF VASTLY DIFFERENT SIZES

By

Eric W. Pellegrini

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY
Physics and Astronomy

2009

ABSTRACT

SCALING STELLAR FEEDBACK: A STUDY OF THE PHYSICAL PROCESSES
INVOLVED IN STAR-FORMING REGIONS OF VASTLY DIFFERENT SIZE

By
Eric W. Pellegrini

Regions of recent or ongoing star formation often contain massive stars capable of
ionizing the surfaces of nearby molecular clouds. These layers of ionized gas, called H II
regions, produce emission lines that serve as beacons of star formation as we look out
into distant parts of our Galaxy and the universe. The complex physical processes of star
formation are responsible for the chemical and structural evolution of galaxies throughout
the history of the universe on many size scales. Light and winds from massive stars heat
and compress nearby clouds, acting to simultaneously inhibit and enhance further star
formation. To disentangle the importance of competing processes such as
photoionization, supernovae, stellar winds, magnetic fields, radiation pressure, I have
studied the dominant physical processes in nearby H 11 regions to determine the relative
contribution of each feedback mechanism as a function of star formation intensity. The
Orion Nebula is an H 11 region that is visible to the naked eye. Due to its proximity to
the Sun and brightness, it has been studied extensively in all wavelengths. It is
dominated by a single 0 star and offers the least complex environment to compare with
models of H II regions. The most complex site of star formation in the local universe is
30 Doradus in the Large Magellanic Cloud. Hundreds of 0 stars dominated a region
thousands of times larger than the Orion Nebula. Together these two examples provide

the constraints necessary to quantify stellar feedback on different scales.

Dedication
This dissertation is dedicated to all those family and friends who helped me overcome
various obstacles as I struggled toward the finish line. With out the friendship and family
support Joe, Joni and Dick Guerne provided the outcome of the effort would have far
more uncertain. And to those who have suffered the most, my son Connor and wife

Jennifer I dedicate this work.

iii

Acknowledgments
I would like to thank the Center for the Study of Cosmic Evolution for their support via a
fellowship in my last semester, the Graduate School for providing a Dissertation
Completion Fellowship. I wish to also thank Dr. Horace Smith for maintaining my
position as a TA at the campus observatory for many semesters.
I owe Gary Ferland a special thanks, not only for the financial support, but also your
time. You were always available, even during Christmas. And while completing this
dissertation I found it helpful to remind myself that I was building a model of a plane,
and not an actual plane, or I would never have finished.
I owe my largest debt of gratitude to Jack Baldwin, most of all for h unending patience.
When I first started you always prioritized my class work during the school year and I
thank you for that. You also taught me the importance of knowing now only how to
operate an instrument, but to understand how it works. Thank you for allowmg me to
“drive” during observing runs and teaching me the finer detail of observing. The
knowledge you have passed on will serve me well until the end of my career.
You have always demonstrated a well balanced attitude toward work and home. You
never asked my to put my research before my family, and were conscience of when I did

it anyway. For that, Jennifer, Connor and I are grateful.

iv

TABLE OF CONTENTS

 

 

LIST OF TABLES ............................................................................................................. vii

LIST OF FIGURES ............................................................................................................ ix

SYMBOLS AND ABRERVIATIONS .............................................................................. xv
CHAPTER 1

Orion’s Bar: Physical Conditions across the Definitive H+ / HO / H2 Interface .................. 1

Abstract .................................................................................................................... 1

1. Introduction .......................................................................................................... 3

2. The Observational Data Set ................................................................................. 7

3. A ray through the H+ / HO / H2 Layers of the Bar .......................................................... 9

3.1 Numerical simulations of the Bar ..................................................................... 9

3.2 The Gas Pressure Model ................................................................................. 13

3.3 The Magnetic Pressure Model ........................................................................ 18

3.4 The Enhanced Cosmic Ray Model .................................................................. 19

4. Discussion ........................................ -- ................................................. 21

4.1 The parameters needed to fit the observations ................................................ 21

4.2 Predicted column densities of additional molecules ....................................... 22

4.3 Sensitivity of final model to input parameters ................................................ 27

4.4 Magnetostatic equilibrium .......................................................... 29

4.5 Heating mechanisms ....................................................................................... 30

4.6 Comparison to previous models of the Bar ..................................................... 31

4.7 Is enhanced cosmic ray heating a realistic prospect ........................................ 35

4.8 Comparison to the magnetic field in other PDRs ............................................ 39

4.9 The effect of radiation pressure on gas density ............................................... 40

5. Conclusions .................................................................................................................... 41

Appendix A. The gas equation of state .............................................................................. 44
CHAPTER 2

Deconstructing the Structure and Physical Processes of 30 Doradus ................................ 63

Abstract .................................................................................................................. 63

1. Introduction ........................................................................................................ 64

2. Observations ....................................................................................................... 68

2.1 Existing optical passband data sets ...................................................... 68

2.2 New Narrow-Band Images ................................................................... 70

2.3 Spectrophotometry ............................................................................... 72

3. The spectroscopic 'data cube' ............................................................................. 76

3.1 Emission Line Flux Measurements ...................................................... 76

3.2 Noise Estimates .................................................................................... 78
3.3 Detector Saturation ............................................................................... 79
3.4 Reddening Correction .......................................................................... 79
3.5 Electron Density and Temperature ....................................................... 80-
3.6 A Publicly Available Data Set ............................................................. 81
4. Observational Results ........................................................................................ 83
4.1 Overview .............................................................................................. 83
4.2 The [0 111] Gas Temperature ............................................................... 84
4.3 Ionization Mechanism .......................................................................... 87
4.4 Temperature Fluctuations .................................................................... 88
4.5 Structural Details .................................................................................. 89
5. Photoionization Simulations .............................................................................. 94
5.1 Rationale and Purpose .......................................................................... 94
5.2 Basic Simulation Parameters ................................................................ 95
5.3 Initial Chemical Abundance Set ........................................................... 99
5.4 The Ionizing Continuum Shape .......................................................... 101
5.5 Revised Chemical Abundances .......................................................... 103
5.6 The Physical Parameters at Each Point in the Nebula ....................... 106
6. Discussion ........................................................................................................ 107
6.1 Photoionization by R136 vs. other Sources of Excitation .................. 107
6.2 The Current Geometry of 30 Dor ....................................................... 109
6.3 What Determined the Structure of 30 Dor ......................................... 115
6.4 The Global Abundances ..................................................................... 121
6.5 Abundance Comparison With Other Methods ................................... 123
7. Conclusions ...................................................................................................... 127
CHAPTER 3
Spectroscopic Observations of NGC 3603 ...................................................................... 183
1. NGC3603 — A Large Star-Forming Region in the Milky Way ........................ 183
2. SOAR Imaging ................................................................................................. 184
3. Spectroscopic Observations with the Blanco Telescope .................................. 185

vi

LIST OF TABLES

Table 1.1 ............................................................................................................................ 47
Observed and predicted quantities for the Orion Bar.

Table 1.2 ............................................................................................................................ 48
Comparison of face-on predicted optical lines to BFM spectrum.

Table 2.1 .......................................................................................................................... 131
Summary of imaging observations used in 30 Doradus optical mosaics.

Table 2.2 .......................................................................................................................... 132
Summary of Blanco and SOAR spectroscopic observations of 30 Doradus. Included in
columns 1-5 are position number corresponding to Figures 2.1 and 2.2, RA and
Declination of the slit center, PA and exposure time.

Table 2.3 .......................................................................................................................... 134
Line IDs and wavelengths. Measured strengths for each of these lines are listed in Table
2.4 and/or 2.5 at every extracted point in the nebula. Rest frame wavelengths with an
asterisk indicate the lines used below to fit the Blanco data to models at each point in the
nebula as described below.

Table 2.4 .......................................................................................................................... 135
A portion of the reddening-corrected Blanco spectroscopic data cube in table form as

described in the text. The units of electron density and temperature are cm'3 and K,
respectively. The dereddened HB surface brightness are reported in units of erg s'1 cm’2

arcsec'z. The other emission lines are reported as 100XS(line)/S(H7t). The entries for
position number 38 are average values for the whole data set, as described in the text.

Table 2.5 ......................................................................................................................... 136
A portion of the reddening corrected SOAR data set, in the same format as Table 2.4 but
with additional columns because more emission lines are measured. These include [8 111]
39069 which was used with [S III] A6312 to measure the gas temperature using [S III] in
a manner identical to that described using [0 111].

Table 2.6 .......................................................................................................................... 137
Possible regions with a locally enhanced ionization parameter due to nearby massive
stars. Columns 1-6 correspond to a unique ID, RA and Dec, RA and Dec offsets from
R136 of the ionizing star and radius of obvious influence visible in the [S II]/Hoc image.

Table 2.7 .......................................................................................................................... 138
Prominent IFS suitable for follow up study with multi-wavelength data. From left to right

vii

the columns are ID, RA, Dec, RA and DEC offsets from R136, IF length and PA. IDs
with an asterisks identify IFs with PAH emission closer to R136 than the [8 II] emission.

Table 2.8 .......................................................................................................................... 140
Catalog of bright dense pillars and protruding IFs. Each pillar has an ID, RA and Dec
(J2000), length and position angle.

Table 2.9 .......................................................................................................................... 142
Cataloged massive stars in 30 Dor with spectral types.

Table 2.10....... ................................................................................................................. 143
Abundances of Selected Elements.

Table 2.11 ........................................................................................................................ 144
X-ray pressures of selected regions in 30 Doradus. The region number corresponds to the
regions identified in Townsley et al. 2006.

Table 2.12 ........................................................................................................................ 145
Abundance Ratios from Different Studies.

Table 3.1 .......................................................................................................................... 187
The direct imaging observing journal for N GC 3603. These observations span 2 years and
the exact number and duration of exposures in each 4 fields vary slightly by field.

Table 3.2 .......................................................................................................................... 188
Journal of NGC 3603 Spectroscopic Observations. The coordinates of the NGC 3603 slit
positions with position angle and total exposure time, excluding short exposures used to
fix saturation.

Table 3.3 .......................................................................................................................... 189
N GC 3603 Line IDs. Emission lines identified in the NGC 3603 survey.

Table 3.4 .......................................................................................................................... 190
A portion of the reddening-corrected NGC3603 Blanco spectroscopic survey in table
form as described in the text. The first several rows of slit position 1 are shown Together
there are 28 unique slit positions with line-strength measurements at 2950 extracted

locations with a 6x6 arcmin area. The units of electron density are cm'3. The dereddened

HB surface brightnesses are reported in units of erg s'1 cm'2 arcsec'z. The other emission
lines are reported as 100 x S(line)/S(HB)

viii

LIST OF FIGURES

Figure 1.1 ........................................................................................................................... 49
Positions of data across the Orion Bar used in this analysis, shown superimposed on a
dereddened Ha image provided by GR. O’Dell. The lines included for each cut are: Wen

8t O’Dell (1995), H0; Tauber et al. (1994), H2 and 12CO. The image is rotated so that the
ionizing radiation strikes the Bar from the left, the same as in Figs. 1.2, 1.3, 1.4, 1.5, 1.7
and 1.8.

Figure 1.2 ........................................................................................................................... 50
Observations of the Orion Bar from Tielens et al. 1993 (”CO ), Wen and O’Dell 1995
(H G), Young Owl et al. 2000 (H2) and this paper (811 [A6716+ 63731]), all relative to the

IF defined by the peak in the [5 II] emission. 61 Ori c is to the left at —111 arcsec. There
is clear stratification indicating an ionized region viewed nearly edge-on.

Figure 1.3 ........................................................................................................................... 51
The geometry derived from the [S II] emission at the ionization front. The Bar is well
represented by a slab 0.115 pc long inclined at 7 deg to the viewing angle.

Figure 1.4 ........................................................................................................................... 52
Pressure and density results from the three basic models developed here. The left-hand
column shows the various pressure components (gas, magnetic, absorbed radiation from
stars, turbulence) as a function of depth into the cloud from its illuminated front surface.
The gas, magnetic, integrated starlight and turbulent pressures are shown as a solid, long
dash, dotted and short dashed lines. The right-hand column shows the number density of

H atoms in the H+, H0 and H2 zones, as a function of depth, so that the pressures shown
on the left can easily be related to specific zones in the model. The densities are shown
using solid, dashed and dotted lines, respectively.

Figure 1.5 ........................................................................................................................... 53

The surface brightness distributions in key emission lines, in units of erg s'1 cm'2 arcsec ‘2
or antenna temperature, as computed for the three basic models (solid lines), compared to
the observed distributions (dotted lines).

Figure 1.6 ........................................................................................................................... 54
Predicted 12CO brighmess temperature as a function of cosmic ray density normalized by
the Galactic background cosmic ray density mm.

Figure 1.7 ........................................................................................................................... 55
Diatomic molecular column densities (in cm'z) for (a) CO+, (b) CN, (c) 80+, ((1) SO, (e)
CS, and (f) 810. Modeled and observed values in cm'2 as a function of angular projection

from the ionization front. Shown with a short dash, long dash and solid lines are the gas
pressure, magnetic pressure and enhanced cosmic ray models. Dots show the various
observations of each molecular species.

Figure 1.8 ........................................................................................................................... 56
Heating mechanisms in the three models. The line styles indicating each mechanism are
the same in each panel. Photoelectric, H2, C I, dust and O I (63 pm) heating are as
defined by Tielens 8t Hollenbach (1985). We also show heating by H I and He II
photoionization in the H+ region, and heating of the molecular gas by direct cosmic ray
heating and also by cosmic ray excitation of permitted FUV lines.

Figure 1.9 ........................................................................................................................... 57

Magnetic field strength vs. H2 gas density, adopted from Crutcher (1999). The star
indicates our new result for the Orion Bar. The filled circles are other systems for which

B105 measurements are available, and the triangles are other systems for which upper
limits on B105 are available.

Figure 1.10 ......................................................................................................................... 58
Predicted [S 11] ratio and density vs. the incident ionizing photon flux @(H).

Figure 2.1 ......................................................................................................................... 146
Figure 2.1. - New narrow band Ha image of 30 Dor from the SOAR telescope rotated 13 degrees. The
center of R136 is marked as a white cross. a) the outline of the region covered by our maps made from the
Blanco spectra; b): The individual slit positions of our Blaco spectroscopic data set.

Figure 2.2 ......................................................................................................................... 147
Ratio of SOAR Hor / [S II] images where darker indicates a lower ratio. The orientation
is the same as Figure 2.1. For reference the SOAR spectroscopic slit positions are plotted
on top of the image.

Figure 2.3 ......................................................................................................................... 148
A sample SOAR spectrum near [0 I] 716300 and [O 11] ”0320,7330. The solid and
dashed lines respectively show the spectrum after and before sky subtraction.

Figure 2.4 ......................................................................................................................... 149
A demonstration of the source of the line profile shapes with a wide slit width. Panel a is
a 2D image of the sky in Hon emission. The box represents the region on the sky the
spectrum in panel b was extracted. Panel b is the 2D spectrum at the same scale as panel
a. Panel c is the line profile for the region extracted. The solid line shows the data of the
He I M678 at the LMC redshift. The top dashed line is the best Gaussian fit and the
bottom is the residual of that fit.

Figure 2.5 ......................................................................................................................... 150
Two examples of extraction windows used to measure line flux. He I 2.6678 shows an
isolated emission line. The [S 11] doublet is slightly blended. Between the two peaks of
the [S 11] indicated by a horizontal bar, a search is performed for the minimum to define
the wings of each emission line as described in the text.

Figure 2.6 ......................................................................................................................... 151
Repeatability of results at overlapping points along different Blanco slit positions. The
curves are histograms of the distributions of ratios of intensity ratios; for example the [O
III]/HB measured from one slit position is divided by the [O III]/HB measure at the same
position on the sky but from a different slit position.

Figure 2.7a ....................................................................................................................... 152

2.7a Interpolated dereddened Ha surface brightness in erg s'1 cm'2 arcsec'z. The region
shown is the same as that outlined in Figure 2.1 and figures 7b — 7g.

Figure 2.7b ....................................................................................................................... 152
Interpolated Av.

Figure 2.7c ....................................................................................................................... 153
Log([o III]/Hor).

Figure 2.7d ....................................................................................................................... 153
Log([N II]/Hor)

Figure 2.7e ....................................................................................................................... 154

Log ([S 111] 7.6312 / ([s 11] 16716+Ms731))

Figure 2.7f ....................................................................................................................... 154
Log ( ([S 11] k6716+16731) / Hor ).

Figure 2.7g ....................................................................................................................... 155
Log me in cm'3 measured from the [8 II] A. 6716/}. 6731 ratio.

Figure 2.7b ....................................................................................................................... 155
Temperatures measured from R defined in equation 5. The scale is shown in increments
of 1,000K.

Figure 2.83 ....................................................................................................................... 156
The temperature and density profiles for our Blanco slit positions. From top to bottom
Figure 2.8a shows positions 1-9. These figures are further described in the text.

Figure 2.8b ....................................................................................................................... 157

The temperature and density profiles for our Blanco slit positions 10-17 and position 20.

Figure 2.8c ....................................................................................................................... 158
The temperature and density profiles for our Blanco slit positions 21-29.

Figure 2.8d ....................................................................................................................... 159
The temperature and density profiles for our Blanco slit positions 30-38.

Figure 2.9 ......................................................................................................................... 160
Prominent ionization fronts listed in Table 2.7, drawn on the [S Ill/Ha ratio image. Top-
left: A blow up of the central region around R136 including IFs 1 and 2.

Figure 2.10 ....................................................................................................................... 161
Left SOAR [S II]/Hoc; Right SPITZER 8pm PAH. A selection of bright pillars are shown
with arrows indicating their location and direction. These Dense IF are detected in both
optical and IR passbands and show a connection with the background molecular cloud.

Figure 2.11 ....................................................................................................................... 162
The profile of IF1. Top: Ionized gas is traced by Her, [0111] and [S III] emission. Middle:
The IF is traced by [8 II], [N 11] and [0 II]. Bottom: The electron density profile
measured from [S 11] as described in the text.

Figure 2.12 ....................................................................................................................... 163
Plots of diagnostic line ratios with observations from SOAR and Blanco spectra. Lines
represent photoionization models with ne 200 cm’3. Arrows indicate the effect increasing
the modeled SED temperature has on the line ratios. From top left to bottom right: (a)
[SIl]/[ 3111] vs. [0 II]/[O 111]; (b) [SID/Hon vs [OIII]/HB; (c) [SII]/[S III] vs. [OII]/[OI];
(d) [O II]/Hor vs [0 III]/HB.

Figure 2.13 ....................................................................................................................... 164
Predicted and observed [0 III] M363/7LSOO7 ratios for models with TP05 abundances.

Figure 2.14 ....................................................................................................................... 165
Similar to Figure 2.13 but with modeled with our adopted abundances.

Figure 2.15 ....................................................................................................................... 166
Various predicted and Observed line ratios described as described in Figure 2.12 using
our adopted abundances.

Figure 2.16 ....................................................................................................................... 167
Predicted and Observed intensity ratio of He I 26678 / Hon.

Figure 2.17 ....................................................................................................................... 168
Predicted and Observed intensity ratio of the commonly used [N 11] 7.6584/H0t
diagnostic.

Figure 2.18 ....................................................................................................................... 169
An interpolated map of the dimensionless quantity U, derived from fitting models to our
Blanco spectra. Region mapped is the same describing Figs. 2.7a — 2.7h.

Figure 2.19 ....................................................................................................................... 170
Top: Map of of [2] , defined in Eq. 12. Bottom: Profile of z for position 8. The offset in
the declination offset from R136. By definition the height of R136 is z = 0.

Figure 2.20 ....................................................................................................................... 171
Profile of |z| in slit positions 1 and 2. Top: |z| from the best fitting model; bottom : |z|
calculated from equation 13 assuming a smooth density distribution.

Figure 2.21 ....................................................................................................................... 172
A cartoon of possible geometries in 30 Doradus consistent with the changes in U across
the nebula. The region labeled “a” is a continuos ionization front of finit height, seen
edge on, facing the ionizing cluster. Region “b” represents possible geometries that
would be seen as discontinuities in modeled R or |z|.

Figure 2.22 ....................................................................................................................... 173
Observed thermal gas pressure interpolated from the Blanco spectra.

Figure 2.23 ....................................................................................................................... 174
The ratio of pressure from integrated star light (Eq. 17) to gas pressure.

Figure 2.24 ....................................................................................................................... 175
Observed gas density vs modeled ionization parameter U.

Figure 2.25 ....................................................................................................................... 176
PM.my from the regions of diffuse x-ray emission described in Townsley et al. (2006). The

pressure was calculated using the reported T may, surface brightness and area.

Figure 2.26 ....................................................................................................................... 177
The region discussed in section 6.3.4 demonstrating an outflow. The contours in both left
and right represent the soft, diffuse x-ray emission. Left: SOAR [S 11]; Right:0.5 -— 0.7
keV emission. The x-ray data was made available by Liesa Townsley.

Figure 2.27 ....................................................................................................................... 178
The ratio of Px—ray to P905.

xiii

Figure 2.28 ....................................................................................................................... 179
The spatial distribution of the [N 11] k6584/ [S 11] (16716 + A6731) ratio.

Figure 3.1 ......................................................................................................................... 191
Narrow-band Hor mosaic image of NGC 3603, from observations using the SOAR
telescope. The NGC3603 Blanco slit positions are shown with labels. Slits with position
angles PA = 147 deg are labeled 1-15. Observations with PA = 57 deg are labeled 20-32.
A 1 arcmin scale bar is shown along with the projected physical scale assuming a
distance of 6.5 kpc.

Figure 3.2 ......................................................................................................................... 192
The same Hor mosaic image of NGC 3603 as in Figure 3.1, without the observed slit
positions super imposed.

Figure 3.3 ......................................................................................................................... 193
An [S 11] mosaic covering the same field as Figure 3.1, also created from observations
using the SOAR telescope. Two predominate pillars, P1 and P2 are indicated. These
were observed extensively in our spectroscopic survey, with multiple position angles.

Figure 3.4 ........................................................................................................................ 194
The emission line diagnostic ([8 II] 6716+6731)/Hor vs [OIII]/HB from the NGC 3603
Blanco data. A higher degree of ionization corresponds to lower [S II]/Hoc and high [0
IIIVHB.

Figure 3.5 ........................................................................................................................ 194
The emission line diagnostic plot [N 11] 6584/Hor vs. [0 IlI]/H[3 from the NGC 3603
Blanco data. A higher degree of ionization corresponds to lower [N Ill/H01 and high

[0 III]/HB. This is analogous to Figure 3.4.

xiv

SYMBOLS AND ABBREVIATIONS
c9 - Solar

3 O Dor - 30 Doradus

A LMA — Atacama Large Millimeter Array
arcmin — arcminute

G EHIIR — Giant Extragalactic HII Region
I—I ST — Hubble Space Telescope

IR - InfraRed

I S N — InterStellar Medium

JW ST - James Webb Space Telescope

n — volume density

P — Pressure

PA — Position Angle

PAH — Poly-Aromatic Hydrocarbons
SED -— Spectral Energy Distrobution
T ~ Temperature

U ~ Ionization parameter

Z - Metallicity, usually defined by the O/H ratio.

XV

Chapter 1

Pellegrini, Eric W., Baldwin, Jack A., Ferland, Gary J ., Shaw, Gargi, Heathcote, S.
( 2 009), Orion's Bar: Physical Conditions Across the Definitive H7 H°/ H2 Interface.

Will;

Orion’s Bar: Physical Conditions across the Definitive H+ / H” / H2
Interface

E.W. Pellegrini & J.A. Baldwin
Physics and Astronomy Department, Michigan State University, 3270 Biomedical
Physical Sciences Building, East Lansing, MI 48824

G.J. F erland
Department of Physics and Astronomy, University of Kentucky, 1 77 Chemistry/Physics
Building, Lexington, KY 40506

Gargi Shaw
Department of Astronomy and Astrophysics, Tata Institute of Fundamental Research,
Mumbai-400-005, India

8. Heathcote
SOAR Telesc0pe, Casilla 603, La Serena, Chile

pelleg10@pa.msu.edu

Abstract
Previous work has shown the Orion Bar to be an interface between ionized and molecular gas,
Viewed roughly edge on, which is excited by the light from the Trapezium cluster. Much of the
emission from any star-forming region will originate from such interfaces, so the Bar serves as a
fOllndation test of any emission model. Here we combine X-ray, optical, IR and radio data sets to
defive emission spectra along the transition from H“ to H° to H2 regions. We then reproduce the
SIDeCtra of these layers with a simulation that simultaneously accounts for the detailed
microphysics of the gas, the grains, and molecules, especially H2 and CO. The magnetic field,
0t)served to be the dominant pressure in another region of the Orion Nebula, is treated as a free
Parameter, along with the density of cosmic rays. Our model successfully accounts for the
0Dtical, IR and radio observations across the Bar by including a significant magnetic pressure and
a180 heating by an excess density of cosmic rays, which we suggest is due to cosmic rays being

trapped in the compressed magnetic field. In the Orion Bar, as we had previously found in M17,

momentum carried by radiation and winds from the newly formed stars pushes back and
compresses the surrounding gas. There is a rough balance between outward momentum in
starlight and the total pressure in atomic and molecular gas surrounding the H” region. If the gas
starts out with a weak magnetic field, the starlight from a newly formed cluster will push back the
gas and compress the gas, magnetic field, and cosmic rays until magnetic pressure becomes an
important factor.

1. Introduction

The interactions between light and winds from a newly-formed cluster and the molecular cloud in
which the stars were born sculpts the geometry of the regions, produces the observed spectrum,
and is a feedback mechanism that throttles the rate of star formation. To explore these processes
in detail we are revisiting a series of well-studied nearby star forming regions. In particular we
are examining objects with geometries viewed nearly edge-on, allowing us to measure the effect
magnetic fields have at different depths as the starlight penetrates into the cloud. We use the
observed stellar parameters, gas densities, and multi-wavelength emission-line spectrum to
strongly constrain a numerical simulation of the physical conditions and emission along a ray
from the central stars through the H+, H0, and H2 regions. We include the effects of dust and of
detailed molecule destruction and formation processes, and treat the detailed micro-physics of the
H+, Ho, and H2 regions self-consistently. The only free parameters in simulations of a well
observed cloud will be the cosmic ray density and magnetic field strength. In some cases the
field can be directly measured. It is never possible to measure the cosmic rays directly so this
approach is one of the few ways to infer their properties, although they are known to be produced
in star-forming regions and can energize emission-line regions.

The modeling approach described in the preceding paragraph was recently applied to the Galactic

H 11 region M17 (Pellegrini et al 2007; hereafter Paper I). That object contains a heavily obscured

and nearly edge-on interface that is excited by about a dozen 0 stars. This interface is of
particular interest because it is a rare case where the magnetic field can be measured in the
adjacent H° region (or photodissociation region, the PDR). This is possible because radio
continuum emission from the H+ zone provides a background light source against which Zeeman
polarization can be measured with the H I 21cm line (Brogan et al. 1999; Brogan 8t Troland
2001). The magnetic field is strong and magnetic pressure is important (see § 3 below).
Combining the Zeeman measurements together with existing radio, infrared, and X-ray maps and
new optical spectroscopy, we found that the structure of M17 is well described by a model in
which the outward momentum carried by the stellar radiation field, together with pressure from a
stellar wind-blown bubble, has compressed the gas and its associated magnetic field until the
magnetic pressure built up sufficiently to be able to halt the process. The overall geometry is set
by hydrostatic equilibrium. In addition, the density of cosmic rays is enhanced as a result of
partial trapping of the charged cosmic ray particles by the compressed magnetic field, so that
cosmic ray heating is important in atomic regions. We consider this to be a very natural cause-
and-effect explanation of why the M17 gas cloud has taken on its present form.

Here we investigate whether this is also a good description of another edge-on interface - the
well-known Bar in the Orion Nebula. Because it is very close to us (here we adopt a distance of
437 pc; Hirota et al. 2007), Orion is perhaps the best studied of all H 11 regions, with data across
the entire electromagnetic spectrum. A schematic of the geomeu'y is shown as Figure 8.4 of
Osterbrock & Ferland (2006, hereafter AGN 3). The ionizing radiation field is dominated by the
hot 0 star 9 1 Ori C. Light from this star is steadily dissociating the background molecular cloud,
resulting in a blister type geometry in which the H+ region is a hot skin on the surface of the
molecular cloud. A large cavity has been carved out of the molecular gas, breaking out of the

cloud on the side nearest the Earth so that we can see through the bubble to the H‘ region on the

back wall (Zuckerrnan 1973; Balick et al. 1974; Baldwin et al. 1991, hereafter BFM91; Wen 8t
O’Dell 1995; Ferland 2001; O’Dell 2001). The hot gas filling this cavity has recently been
detected by Gi‘idel et a1 (2008).

Given the important role that magnetic pressure plays in M17, it is natural to ask whether it might
also be important in Orion. While there are no direct observations of the field strength in the
atomic gas associated with the Orion Bar, the Orion complex shows a well-structured polarization
pattern that drops to a low level of polarization in the Bar, suggesting that the magnetic field in
the Bar is directed more or less along our line of sight (Schleuning 1998). This is the expected
orientation for an initially tangled magnetic field frozen into gas that has been compressed by
radiation pressure or stellar winds from 91 Ori C. There are Zeeman measurements for the Orion
Veil (a foreground structure that is associated with the Orion Molecular Cloud) showing that Bios
~ 50 [LG (Brogan et al. 2005). This is already an order of magnitude greater than the Galactic
background of 5-10pG (Tielens 8r Hollenbach 1985), even though Bro, only represents the line of
sight component. Detailed analysis suggests that the ratio of magnetic to gas pressure in the Veil
is large (Abel et al. 2004, 2006). A further indication of the presence of a strong B field in the
Veil is an indirect study using infrared line ratios (Abel and Ferland 2006), which indicated that
along the line of sight to 6 1 on C the ratio of magnetic/gas pressure is Pmag/ Pgas > 1. Aside
from the suggestive results of the Veil, turbulent velocities in the atomic region of the background
cloud are supersonic and it has been suggested that magnetic fields are responsible (Kristensen et
al. 2007; Roshi 2007). These results lend plausibility to the possible presence of a strong
magnetic field in the Bar region.

The Bar appears as a bright ripple on the background ionization front, lying at a projected
distance from 910ri C of 111 arcsec (0.23 pc). In an important series of papers Tielens et a1

(1993), Tauber et al. (1994), and Young Owl et al. (2000) showed that the H0 and H2 regions in

the Bar are easily resolved on the sky as separate structures displaced from each other in a way
that clearly demonstrates that the Bar is indeed a roughly edge-on interface. They showed that
most of the observed emission in PAHs, [O I], [C II], H2 and the 12CO J=1—0 lines can be
understood as coming from a homogeneous region with density n" ~ 5x 104 cm'3. However, they
argued that large (9 arcsec = 0.02 pc) clumps with about 20 times higher density must be
irnbedded in this homogeneous medium to produce the observed high-level CO lines (J=14-13,
7-6) and also the HCO+ and HCN emission. Indeed, their interferometer images directly show
clumpy structures in these lines but not in the many lines attributed to the homogeneous medium.
Those investigations considered only the molecular and neutral atomic regions of the Bar. The
density chosen by Tauber et al. (1994) for their model was inferred from the observed offsets of
the H2 and 12CO emission in the Bar relative to the ionization front, combined with the column
density required to allow the observed CO emission to be produced. In doing so, the authors
assumed constant gas density in the H0 region, arguing that turbulence dominated the pressure.
Van der Werf et al. (1996) combined new H2 observations with the existing data set of the PDR
and postulated that H2 emission from the interclump medium required a filling factor less than
unity for the interclump gas. Contradicting this result, Allers et al. (2005) also modeled the
interclump region with constant pressure, not constant density. They concluded that a filling
factor of unity described the region well. However they also found that an extra heating agent
must be present in the Bar, but were unable to establish what it might be.

Here we investigate the physical conditions across the Bar that are implied by the variations of its
spectrum across its full width. Like the previous work, we use the observed gas density, stellar
parameters, and emission peak offsets. However, we consider the emission from the H+ and PDR
regions together in a self-consistent picture. By more fully understanding what determines the

structure of the Bar we can better understand the general nature of such interfaces. Here we focus

on a line of sight through the interclump medium, not on the clumps described by Young Owl et
al. (2000) and others. Like many (but not all) previous papers we conclude that the interclump
medium dominates and determines the overall structure of the Bar. We find that the observed
densities and the radial extent of the various stratified emission regions through the Bar are in fact
the natural consequences of a cloud in roughly hydrostatic equilibrium, in which magnetic
pressure and cosmic ray heating play major roles in the H0 region. This study, combined with the
companion work by Shaw et al. (2008; hereafter Paper 111), which focuses on the detailed H2
emission spectrum, lead to a better picture of the nature of these interfaces.

2. The Observational Data Set

The Orion Nebula has been observed extensively at all wavelengths. For this study of the Bar we
draw on the following published data sets. From Tauber et al. (1994) we use maps of the 12CO
and 13CO emission lines, and their summary of previous mid-IR observations of the CO 7-6, CO
14—13, 01163 pm, 0 121143 um, C 113.158 um, Si 112135 um, C I 71609 um and C ”1370 um
emission lines. The H2 1-0 8(1) data are originally from van der Werf et al. (1996), who find an
average surface brightness of 5.9x 10’15erg s'1 cm'2 arcsec'z. However, the H2 surface
brightness varies significantly along the Bar. A05 and Young Owl et al. (2000) used the van der
Werf et al. H2 data set but extracted surface brightness profiles along different narrow lines
perpendicular to the Bar. A05 did this for a particularly bright region and found a peak brightness
of about 9x 10'15 erg s'1 cm'2 arcsec'z. Young Owl et a1. (2000) did the same for a fainter region
which is along the same cut for which most of the other published molecular data used here were
measured, and found the peak S(H2) 1-0 8(1) = 3.3x 10'15 erg s.1 cm'2 arcsec'z. For consistency
we adopt the Young Owl et al. H2 surface brightness profile. The peak surface brightnesses of
these emission lines are summarized in Table 1.1 with observed values in column 2 and results of

our three models described below in columns 3 through 5.

Besides H2 and CO, many other molecules are detected in the Bar. Here we will also compare our
predictions with observations for CO+ (Storzer et al. 1995), SO+ (Fuente et al. 2003), CN (Simon
et al. 1997), CS (Simon et al. 1997, Hogerheijde et a1. 1995), SiO (Schilke et a1 . 2001), and SO
(Leurini et al. 2006).

Pogge et al. (1992) made optical passband Imaging Fabry-Perot maps of Her, H3, [0 III] A5007,
[N 11] AA6548, 6583, [8 II] AA6716, 6731 and He I A6678, and in their Figure 5 show a density
profile across the Bar, measured from the [S 11] A6716/A6731 intensity ratio. Wen 8t O’Dell
(1995) used the Pogge et al. data to make a 3-dirnensional map of the H+ region in Orion and
their Figure 4 shows intensity profiles across the Bar for several lines including Hor. Garcia-Diaz
& Henney (2007) mapped a number of optical emission lines by taking an extensive grid of
echelle long-slit spectra, and in their Figure 9 show a density profile across the Bar (from the

[S 11] intensity ratio) that is very similar to the Pogge et al. (1992) result. From these papers, we
have especially depended on several figures showing line intensities along the cuts across the Bar
shown here in Figure 1.1.

To this existing data set we added an intensity profile across the Bar in [S 11] A6713+A6731,
measured from a continuum-subtracted, narrow-band [8 11] image taken on 06 April, 2007 with
the Southern Astrophysical Research (SOAR) Telescope 1. The [S 11] filter is 45A wide and is
centered at A6723, while the continuum filter is centered at 6850A with a width of 95A.2 The
SOAR Optical Imager ($01) was used; it provides a 5x 5 arcmin field of view with a 2x 2 binned
pixel scale of 0.15 arcsec/pixel and yielded images with 0.6 arcsec FWHM in both filters. The

images were processed in IRAF in the usual way, then the continuum image was scaled to

 

1 The Southern Astrophysical Research Telescope is a joint project of Michigan State University,
Ministério da Ciéncia e Tecnologia-Brazil, the University of North Carolina at Chapel Hill, and the
National Optical Astronomy Observatory. Further information about SOAR and its instruments may be
found at www.soartelescope.org.

2 We are grateful to Dr. Frank Winkler for loaning us the filters.

unsaturated stars in the nebula and subtracted from the emission line image. Next, the continuum-
subtracted [S 11] image was flux calibrated using the emission-line surface brightness
measurements from the long-slit spectrophotometry of BFM91, following the example of O’Dell
and Doi (1999). Finally, the [S 11] intensity profile was measured along the same cut as is shown
in Figure 1.1 for the Wen & O’Dell Hor profile.

Figure 1.2 combines together the four key intensity profiles that we will use to constrain our
models of the Bar. The ionizing radiation comes in from the left. Using the new SOAR [S II]
images, the different profiles have all had their zero points set to match the distance from the peak
in the [S 11] emission along each cut. There are several key features to note. The ionization front
(IF) is assumed to be at the position of the peak [S 11] emission. Hor emission comes from a broad
plateau before (to the left of) the IF. The H2 brightness profile peaks about 12 arcsec after the IF
(i.e. deeper into the cloud). In the 12CO I = 1—0 line the peak intensity occurs about 20 arcsec
after the IF and is very broad.

As can be seen in Figure 1.1, these cuts across the Bar are not at identical locations. The Wen 8t
O’Dell Hor profile and our [S 11] profile are averaged over 20 arcsec wide swathes extending in
PA 322 deg from 9 1 Ori C, while the 12CO (from Tauber et al. 1994) and H2 profiles (Young
Owl et al. 2000) are taken in PA 315 deg crossing the Bar about 60 arcsec SE of the H01 and [S
11] cuts. However, close examination of maps of the [S 11] ratio in the vicinity of the Bar (e.g. Fig.
11 of Henney et al. 2005a), and the similarities of the same lines published for a different position
along the Bar (Young Owl et al. 2000) indicates that the relative offsets and shapes of the
different line profiles are typical of the interclump component of the Bar as a whole, and
therefore are valid criteria for fitting our models.

3. A ray through the H+/ Ho / H2 Layers of the Bar

3.1 Numerical simulations of the Bar

The bright ridge of emission seen as the Bar in Figure 1.1 is a region where the gas goes from H+,
closest to the central stars, through H0 when ionizing radiation has been attenuated, eventually
becoming H2 when ultraviolet light is sufficiently extinguished (see also Figures 8.4 and 8.6 of
AGN 3). We derive physical conditions across the Bar by comparing predicted and observed
spectra at various distances away from the star cluster (as represented by 8 1 Ori C). The
simulations are done with the spectral synthesis code Cloudy. Last described by Ferland et al.
(1998),3 Cloudy has since been updated to include a large model of the hydrogen molecule
(Shaw et al. 2005), more complete grain physics (van Hoof et al. 2004) and the chemistry of a
PDR (Abel et al. 2005). We used the publicly available version of Cloudy, but with updated H2
collision rates as described in Paper III.

We begin the calculation at the ionized or illuminated face of the H+ region and follow a beam of
starlight away from the central cluster into the molecular cloud. The H+, H0 and H2 regions are
actually a flow from cold molecular gas through the atomic region into hot and ionized gas with
the ionizing radiation gradually eating into the molecular cloud. However, we approximate the
situation here with a hydrostatic model. We will present predicted and observed quantities along
this line from the stars into the molecular cloud, with the ionizing radiation coming in from the
left in all relevant figures.

The most important assumption in all of this is that the conditions within the various regions are
related to one another by continuous variations in the gas, radiation, and magnetic pressures.

The equation of state, the relationship between these pressures and the density, is described in
Paper I and further in Appendix A. Because the structure of the H+, H0 and H2 regions are all the

result of a single calculation, with a single set of initial conditions, we minimize the number of

 

3 Many additional physical processes described in Cloudy are documented and referenced in Hazy, the
approximately 1500 page Cloudy manual. It can be downloaded from
ftp://gradj.pa.uky.edu/gary/cloudy_gold/docs

10

adjustable parameters compared to the number of observational constraints. As will be shown in
section 4.1 our best model will have only 6 arbitrary input parameters, but will satisfactorily
match 17 observational constraints. For example, the UV radiation that penetrates into the PDR
is the result of a detailed treatment of radiation transport through the H+ region. By treating the
H+ region and PDR together in a single calculation, the radiation affecting the PDR is guaranteed
to be consistent with the observed properties of 8 1 on C and the gas density of the H+ region. The
emission from ionized and neutral atoms, grains and PAH chains, and several molecular species
including H2 and C0, are all predicted by self-consistently treating the microphysics of the gas.
Starlight is attenuated by each layer and passed on to more distant regions. All models were also
required to reproduce the observed 111 arcsec projected offset of the [S 11] emission from 81 on
C, which clearly defines the Bar’s ionization front.

The location of the illuminated face of the H+ region, and the gas density there, are set by the H01
emission profile. There is a steep drop in the Ho: surface brightness where the H‘ zone ends and
the H0 zone begins. There is no similar sharp rise to mark the illuminated face of the Bar because
the light emitted by the edge-on Bar structure is diluted by additional emission coming from parts
of the H+ region which are on the background cavity surface but which lie along the same line of
sight. Therefore, the initial density and radius of our models can only be constrained by their
effects beyond the illuminated face. These initial parameters are chosen so that the observed H01
and [S 11] surface brightness and [S II] doublet ratio are matched. Since we assume hydrostatic
equilibrium the gas pressure increases as starlight with its associated momentum is absorbed.
This causes an increase in density and emission measure, n2 (W, producing the observed

brightness profile.

11

The parameters assumed in our simulations are largely observationally based and are summarized

next. The main free parameters will be the magnetic field intensity and cosmic ray density. We

assume flux freezing and a scrambled magnetic field to relate the gas density and field.

Three models are presented here, designated (1) Gas Pressure Model, (2) Magnetic Field Model

and (3) Enhanced Cosmic Ray Model. They all share the following assumptions:

The grain optical properties, described by BFM91, are based on the observed extinction
in Orion. Many previous investigations, going as far back as Tielens 8: Hollenbach
(1985), assume standard ISM extinction. Orion grains have a more nearly grey
dependence of extinction on wavelength compared with ISM grains, changing the
structure of the layers. The Orion size distribution is deficient in small particles so
produces less heating of the gas by grain electron photo-ejection.

The PAH abundance is npAH/nHo = 3x 10'7 (Draine et al. 2007) with a power-law
distribution of PAH sizes with 10 size bins, according to Bakes & Tielens (1994).

We assume overall hydrostatic equilibrium as described in Appendix A, with magnetic
fields (when included) described by Eq. A1 with y = 2.

The gas-phase abundances are the standard Cloudy values for H 11 regions, and are
largely based on observations of the Orion Nebula described by BFM91, Osterbrock et a1
(1992), and Rubin et a1 (1991, 1993). These abundances are listed in Table 2 of Paper 111.
We have taken the distance to the Orion Nebula to be 4371: 19 pc (Hirota et al. 2007).
This is a VLBI parallax measurement, which should be more accurate than previous
results.

All models are constrained to reproduce the observed [8 II] ratio I(A6716)/I(A6731) =
0.63 :l: 0.05, converted from the reported electron density of n. ~ 3200 cm'3 at the peak

of the Bar (Pogee et al. 1992). This is consistent with the [S 11] ratios reported by Garcia-

12

Diaz & Henney (2007) for the Bar. The calculations predict the full spectrum including
this ratio. We match this ratio instead of using a deduced density since the conversion
from ratio to density depends on the kinetic temperature and ratio of electrons to atoms.
The observed stellar X-ray emission caused by wind activity immediately around the
Trapezium stars is modeled with a bremsstrahlung distribution with a temperature

32'6 erg s'1 over the 0.5—8 keV passband

T = 106K and an integrated luminosity Lx = 10
(Feigelson et al. 2005). This represents the X-ray emission from just 81 Ori C. There are
additional X-ray point sources in this region besides 9 l Ori C, but their positions along
the line of sight are poorly known. Omitting them is not likely to be a large source of
error because 6 1 Ori C accounts for 68% of the total observed X-ray flux from the Orion
region. The diffuse X-ray emission detected by Giidel et a1 (2008) is much weaker than
the stellar contribution.

A constant turbulent velocity was assumed. The observed 13CO line width is 1.8 km s'1
FWHM (Tauber et al. 1994), while the H2 line widths are in the range 2—4 km s'1 after
allowance for a poorly known instrumental profile (A05). Here we adopt a constant

turbulent velocity of 2 km s'1 FWHM throughout the nebula and include it as a source of

pressure in our models.

The three models are described below. Each builds on the results from the preceding one by

adding additional physical processes.

3.2 The Gas Pressure Model

This first model represents the BFM91 hydrostatic model of the H+ region. It does not include a

magnetic field but does include radiation and turbulent pressure. We shall refer to this as the gas

pressure model even though turbulent and radiation pressures make significant contributions.

Cosmic ray heating and ionization are included in the calculation using a cosmic ray density set to

13

the Galactic average of 2.6x 10'9 cm'3 (Williams et al. 1998). The ionization rate per particle
corresponding to this density is 2.5x 10'17 s'1 for H0 and 5x 10'17 s'1 for molecular H2. For this
model the heating due to cosmic rays is mostly insignificant, with the cosmic rays peaking at 15%
of the heating in the coldest and most neutral regions, where the starlight is heavily extinguished.
Collisional heating of grains (Drain 1978) is the most important heating mechanism in the
molecular region, providing 70% of the total heating. In this model we find the heating in the
deep molecular gas is not dominated by cosmic rays but by grain heating processes, consistent
with the conclusions of A05.

Figure 1.3 shows our assumed geometry. The Bar is a thin slab tilted slightly to our line of sight.
Because the hydrogen ionization front is thin, the observed [S 11] profile on the sky is very
sensitive to the angle of inclination. The thin slab extends 0.115 pc along the line of sight and is
inclined 7 deg relative to the line of sight. This is very similar to many previous models of the
Bar’s geometry (eg. Tielens et al. 1993; Hogerheijde et al. 1995; Wen & O’Dell 1995; Walmsley
et al. 2000; A05).

1 cm’z, the shape and intensity of the [S 11]

For a given incident ionizing flux ¢(H) photons s'
profile is set by the initial hydrogen density no(H). This also defines the position of the hydrogen
ionization front. The problem is that we do not initially know either <D(H) or no(H).

The number of ionizing photons per second Q(H) emitted by 61 on C and hence the ionizing flux
¢(H) incident upon the slab is uncertain. Although the star’s spectral type is clearly 06.5, there is
a significant range in the Q(H) and effective temperature values that are thought to be appropriate

for even a “normal” 06.5 star. At the high end is Q(H) = 1049'23 s'1 with Teff = 42,300 K (Vacca

et al. 1996), while Hanson et a1. (1997) adopted the much lower value Q(H) = 1048'89

s'1 and T3)?“
= 41,200 K for this spectral type. Intermediate values are suggested by Smith et al. (2002) and

Stemberg et al. (2003). In addition to this uncertainty, 8 1 OriC has an unusually strong magnetic

14

field, which is thought to channel the flow of stellar winds from the star in ways that modulate the
observed spectrum (Wade et a1. 2006). This could produce a significantly anisotropic radiation
field.

Given the above uncertainty, we explored a range of values of Teff and Q(H). Our models have
only a very slight dependence on the exact value of Tent, but the value used for Q(H) clearly does
matter. Changing Q(H) while matching all the observations of the H+ region, including the
density ne which is a measured quantity, does not significantly affect the predictions for the H0
and H2 regions, but it does change the deduced distance between 6 1 Ori C and the ionization
front. Since the projection of this distance on the sky is fixed, the position of the slab along our
line of sight must change relative to 81 on C to match the observed geometry. For the largest
Q(H) value cited above the slab would have to lie about 0.125 pc farther away from us than in the
adopted model, while maintaining the same inclination angle relative to our line of sight. We
eventually adopted the Kurucz (1979) stellar model as the continuum shape of 0 1 Ori C with
Q(H) = 1049'00 s'1 and T9)? = 39,700 K. This Q(H) is the same as was used in the 3D model for
which Wen and O’Dell (1995) show figures, and also was used in the hydrodynamic wind models
computed by Henney et al. (2005a), simplifying comparison to those papers.

However, this still left many combinations of the Bar’s thickness 1 along the ray from 6 1 on C
and in its density no(H) at the illuminated face, and consequently of the radial gas density
distribution nH( r), which were consistent with the constraints used so far. We used the additional
constraints of the Hor and [S 11] brightness profiles together with the [S 11] A6716/A6731 intensity
ratio to determine the remaining properties of the entire H+ zone. The electron density ne ~ m; is
directly measured in the region where the [S 11] lines are formed, so the [S II] surface brightness

was used to determine the distance h through the Bar along the line of sight using

ms 11]) cc ns+ neh. (1)

15

Then the Ho: surface brightness at other points on the Bar was used to determine the relationship
between density and position on the sky for the known h. Besides increasing the peak surface
brightness, a higher n" also decreases the deduced thickness 1 of the H+ region.

The result of this was a model, with predicted emission line strengths and magnetic field
summarized in column 3 of Table 1.1, which reproduced the structure and emission of the H+
region. In this model, h = 0.115 pc, the illuminated face of the cloud lies 0.114 pc from 81 on C,
I = 0.141 pc, and ¢(H) = 6.45x 1012 s'1 cm'z.
The bottom row of Figure 1.4 shows the pressure sources and H+, HO and H2 density distributions
calculated for the gas pressure model. This figure also includes the same plots for the other two
models which are described in the following two sections.

Figure 1.5 shows, again for all three models, how well the computed surface brightness
distributions of [8 II], Hor, H2 and CO lines match the observations. The results for the gas
pressure mode] are shown in the left-hand column. We computed the conditions and locally
emitted spectrum for each point along a ray from the central star through the layers shown in
Figure 1.3. For the case of optically thin lines ([8 II], Ho: and H2), the comparisons of the models

to the observed surface-brightness distributions were made by integrating the volume emissivity

along the line of sight into the cloud according to

-dAl.(r)/2.5
si=jei(r)/4nx10 dh (2)

The integration is along the line of sight into the modeled region of thickness h, £5 is the volume
emissivity for the ith line, F is the radial vector from 8 1 Ori C, and M: is the amount of internal
reddening, in magnitudes, providing a correction to the observed line intensities for internal
extinction by dust. The computed surface brightness of the visible-passband emission lines shown

in Figure 1.5 and listed below in Table 1.1 have been increased by a factor 1.5 to account for an

16

additional component reflected from dust in the molecular cloud (Wen and O’Dell 1995; O’Dell
et al. 1992). At longer wavelengths the Orion dust does not scatter efficiently (BFM91) so no
similar correction is needed for the infrared and mm-wavelength lines.

The reddening correction is specific for each line according to the R = 5.5 reddening curve used
for Orion. The mid-IR and longer wavelength lines are unaffected by extinction due to their long
wavelength and the H+ region observations have been dereddened (BFM91, Wen & O’Dell
1995). Only the analysis of the H2 2.121jrm line is affected by internal extinction. This amounts
to a flux decrement of roughly 40% for the models presented below. Increasing the depth of the
slab along the line of sight (i.e. parallel to the ionization front) increases the intensities of all lines
except that of the H2 2.121 um line which stays approximately constant due to the effects of
extinction by dust.

The 12CO J = 1—0 line is also different because of its large optical depth. For optically thick
therrnalized lines the emission is directly related to the kinetic temperature of the gas via the
antenna temperature T antenna. Deep in the cloud the density is very high and the optical depth
increases rapidly. The antenna temperature quickly approaches the kinetic temperature according
to the equation

Tame"... = Trina-e") (3).

where r is the CO line optical depth along our line of sight into the cloud and Tran is the
computed spin temperate of the 12CO levels. Our calculations solve for 1 toward the illuminated
face, which we then scale to account for the enhanced path-length caused by our viewing angle
(see Figure 1.3). The computed CO surface brightness curves in Fig. 1.5 show Tamenna starting at
the edge of the slab farthest from us, which from our viewing angle is projected to lie closest to

91011 c.

17

The top-left panel in Figure 1.5 shows the excellent fit of the gas pressure model’s [S 11] surface
brightness distribution to the data. The gas pressure model’s match to the observed Ha surface
brightness profile (Figure 1.5, second panel down in left-hand column), as well as [0 III] 5007A
and [N 11], with no further adjustments to the model, validates our assumptions.

The gas pressure model accurately describes the H+ region, but does not correctly predict the
positions of the H2 and 12CO emission peaks on the sky, as can be seen in the left-hand panels of
Figure 1.5. This model was tuned to reproduce the observed H+ emission region using only gas,
radiation, and turbulent pressures. In this situation, the gas density in the H0 region that is
required to maintain hydrostatic equilibrium is 1.4x 105 cm'3 (Fig. 1.4). Since the depth of the H0
region is set by the path length required to reach an Av ~ 1 where H2 forms, this high-density H0
region is quite narrow. The result of the high density is that the computed H2 emission peak
occurs two times closer to the ionization front than is observed (5 arcsec predicted separation
rather than the observed 12 arcsec), as is seen in Fig. 1.5. This result is consistent with the theory
of H2 emission as presented by Black and van Dishoeck (1987) and Draine and Bertoldi (1996).
For further discussion of H2 we refer the reader to Paper 111. For the same reason, the CO
emission also peaks at a point too close to 910fi C. Table 1.1 shows that the predicted peak H2
emission is equal to 1.51 times the peak value, while the predicted CO emission is several times
fainter than is observed.

3.3 The Magnetic Pressure Model

This second model uses the same constraints as the gas pressure model described above, but
includes a magnetic field similar in strength to that seen in the Veil (Abel et al. 2005). The cosmic
ray density was maintained at the Galactic background level. We ran a series of models with
increasing values of the strength of the magnetic field at the illuminated cloud face, in the same

way that we had done in Paper I for M17. The initial magnetic field sets the field strength

18

throughout the model since we assume flux freezing (Eq. A1 in the Appendix). The resulting
magnetic pressure contributes to the total pressure according to Eq. A2. The center two panels in
Figure 1.4 show the pressure contributions and densities as a function of depth in the final version
of this model.

In the H+ region where the temperatures are of the order 104 K, gas pressure still dominates, so the
gas pressure and magnetic pressure models are very similar in this region. The observed [S 11]
density and [S 11] and H01 surface brightness profiles are matched by either model using the same
initial conditions.

In the H0 region the temperature drops and the gas density increases. According to equation Al,
the magnetic field is amplified as well, so magnetic pressure support becomes important and
further compression of the gas is halted. With a lower average density in the H0 region a longer
path length is required to absorb the UV photons that prevent H; from forming, so the H0 region
becomes more extended. Stronger fields at the illuminated face produce more magnetic pressure
in the H0 region, resulting in a lower density and larger thickness. We changed the initial
magnetic field at the face of the cloud so that the H2 emission peak occurred at its observed offset
of about 10 arcsec (0.021 pc) from the ionization front. We found that an initial field of 8 uG best
fits the observed brightness distribution of the H2 emission line. In the H0 region this field is

<B> = 438 uG and the density in the H0 region is 8x 104 cm’3, about 2 times lower than the
density in the gas pressure model.

With the magnetic pressure model, the brightness and position of the H2 peak now match the
observations (Fig. 1.5). However, the model still does not reproduce the position and antenna
temperature of the 12CO J = 1—0 emission peak or the high surface brightness in the higher-level
12CO lines (Table 1.1).

3.4 The Enhanced Cosmic Ray Model

19

The magnetic field model fails because it does not provide sufficient heating in the deeper parts
of the H0 region. Based on the results from M17 in Paper I, we next explored the effect of
assuming that cosmic ray particles are trapped by the compressed magnetic field, so that the
cosmic ray density is also increased. The increased density of cosmic rays acts as an additional
heat source, which becomes important in the region emitting the H2 and 12CO lines. The cosmic
rays also increase the ionization level in molecular regions, increasing the speed of ion—molecule
interactions and enhancing the CO formation rate. This moved the 12CO J = 1—0 peak inwards
toward the central stars. A second effect is that the kinetic temperature is increased and the H2
emission is enhanced deep into the cloud. The model with enhanced cosmic rays produces
extended H2 emission beyond the emission peak. We ran models with the cosmic ray density
increased by different values over the Galactic background density, up to the point where the
cosmic ray energy density is in equipartition with the magnetic energy density. Figure 1.6 shows
the dependence of the computed 12CO J =1-0 brightness temperature on the cosmic ray density.
The observed CO brightness temperature, peak H2 intensity, and shapes and positions of the H2
and CO profiles all are matched best by a cosmic ray density in equipartition with the magnetic
field. That is the cosmic ray density we adopted in our final best model, which we call the
enhanced cosmic ray model.

The average magnetic field, weighted by the 21 cm opacity T5pin/"(H0), for the enhanced cosmic
ray model is <B> = 516uG, almost 100uG higher than in the magnetic field model, where the
cosmic ray density was set to the Galactic background. The difference is due to the weighting of
B by T5pin/"(H0) that is used to calculate <B>. The ratio is the 21 cm opacity used by Zeeman
measurements to derive B. In the enhanced cosmic ray model, deep regions of the cloud which
would normally be fully molecular have a significant amount of H0 produced by cosmic ray

dissociation. The effects of the cosmic rays on the chemistry are shown in Figure 1.4. Neutral

20

hydrogen persists deep into the molecular core, where B peaks at about 530uG. Here the cosmic
ray density is enhanced by a factor of 103‘6 over the Galactic background. This corresponds to an
ionization rate of 1x 10'13 s'1 for H0 and 2x 10'13 s.1 for H2. In the case of the magnetic model,
the neutral hydrogen does not coexist with Hz. This extended distribution of neutral hydrogen is a
result of the enhanced cosmic rays and is not dominated by FUV photons. It would be incorrect to
consider it part of the classically defined “PDR”, but it will still affect the weighted <B> an
observer will measure.

4. Discussion
4.1 The parameters needed to fit the observations

Our final model (the enhanced cosmic ray model) provides an integrated description of the
ionized, neutral and molecular regions of the Bar. It is based on the idea that the pressure of
photons from 0 1 Ori C has compressed the surface of the molecular cloud, and along with it a
magnetic field that was already present, until the combination of gas, magnetic and turbulent
pressure became high enough to halt the compression. This model indicates that the Bar is in
(quasi)hydrostatic equilibrium. This picture of the Orion Bar, combined with straightforward
assumptions about the geometry, provides a good fit to the observed parameters. It reproduces the
observed surface brightness profiles in the sense of both the position and the peak brightness of
the H01, [S 11], H2 and 12CO emission lines. As is shown in Table 1.1, it satisfactorily reproduces
the [S 11] A6716/A6731 ratio, the integrated strengths of many other lines from the H+ region, and
the peak surface brightness of important atomic lines from the Ho region including [0 I] A63um,
[Si 11] A34.1p.m and [C II] A158um. We do not have a good optical spectrum. with the slit set
across the Bar, but we checked the predicted optical emission lines in Table 1.2 and verified that
the computed optical spectrum is similar to that found by BFM91 at their position 5 on the west

side of the Orion Nebula and that no unusual lines are predicted to be strong. Position 5

21

represents a region with a density comparable to the Bar. Our final model provides a quite good
fit to the data. We matched 17 observed properties with 6 free parameters: the location of the
illuminated face and the gas density at the face, the tilt of the IF and its depth along the line of
sight, the magnetic field strength, and the cosmic ray density.

We arrived at this model in three steps.gThe properties of the H+ region were determined by
fitting a hydrostatic model that included only gas, radiation, and turbulent pressure — the gas
pressure mode]. The location of the illuminated face and the gas density at the face, the tilt of the
IF and the depth of the IF along the line of sight needed to be adjusted to reproduce the properties
of the H+ region. However, the resulting gas pressure model produced an H0 region that is too
narrow. To accurately fit the distance from the IF to the H2 emission peak, it was necessary to add
magnetic pressure. For the case of hydrostatic equilibrium this results in a decrease in the gas
density, causing the H0 region to become more extended and pushing the peak emission of H2
and 12CO to the observed value. However, this magnetic pressure model still failed to reproduce
the observed H2 and CO emission from deep in the cloud because the kinetic temperature was too
low. We propose that the extra heating comes from cosmic rays trapped by the compressed
magnetic field, the same thing that appears to be happening in M17. This fully reproduces the
emission and geometry.

4.2 Predicted column densities of additional molecules

Cloudy includes 94 molecules in its calculations, using data mainly from the UMIST data base1
(Abel et al. 2005). Chemical fractionation is not included in UMIST, so we cannot now deal with
molecular isotopes. The UMIST data base does not include information about the internal
structure of molecules. Therefore, we can only compute the molecular space density and derive

column densities for this full set of molecules. Here we compare results for the subset of

 

‘ www.udfa.net

22

molecules which we consider most reliable and for which observations are available. Figure 1.7
shows this comparison as a function of projected distance from the IF for CO+, CN, 80+, SO, CS
and SiO. Our models directly predict the volume density of each molecule. We have then
computed a predicted column density N,-( r) at an angular offset r arcsec from the IF using
N}.(r)=thj(r)cm—2 (4)
where h is the depth into the cloud along our line of sight. From §3.2 h is found to be 0.115 pc.
The number density from the model is nj for the fh molecule.
In most cases observations of diatomic molecules are presented as column densities. In cases
where the observed surface brightness is the quantity reported, we have converted it into a
column density using the assumption that the line is optically thin and the energy levels of the
molecule are in LTE. Under these assumptions the relation between column density NT and

surface brightness <I> averaged over frequency v is given by Miao et al. (1995)

 

E
S Q exp —i
1
_2.04 I > B (Trot) f rot Trot 20 _2
NT’eo B(T )—1 2 3 X10 cm (5)
a b rot back glngu v

where Qrot=2kTror/hv, and B(Tro¢) is the Plank function at a temperature Trot with Iback equal to
the background continuum. In the analysis done by Young Owl et al. (2000) the conversion from
I to N assumed that Trot = Tkin = 100K, a value similar to the CR enhanced model. This is
approximately the observed temperature of the 12CO gas. When it is not possible to derive Trot,
Tia-n is often used. This underestimates the true column density if the line is subthennally
populated as occurs when the gas density is below the critical density of the line.

Figure 1.7a shows the CO+ comparison. Storzer et al. (1995) and Fuente et al. (2003) both
measured N(CO+) as a function of depth into the Bar along the same line of sight as the other

molecules measured by Tielens et al. (1993) and Tauber et al. (1994). The CO+ observations are

23

characterized by a rise in column density with distance from the IF reaching a peak column
density of 3x 1012 cm'2 at r =17i7”. After the peak a gradual decrease is observed out to a
distance of 40” where N(CO+) = 4.8x 1011 cm'z. The column densities found by Storzer et al. are
uncertain and may actually be higher based on uncertainties in the excitation mechanism assumed
in the analysis, as explained in their paper. The CO+ column density computed using the
magnetic model with the canonical cosmic ray ionization rate peaks too early and is a factor 100
too low. With increasing cosmic ray ionization rate the peak column density increases and moves
farther from the IF. For the enhanced cosmic ray model the peak column density is

4.85x 1012 cm'2 and occurs at r = 12”. The subsequent decrease in the modeled N(CO+) follows
the observations very closely. CO+ is shown to be very sensitive to the presence of cosmic rays.
Decreasing the ionization rate by a factor of 10 decreases the column density by a factor of 100 at
r 2 20”. We conclude that the enhanced cosmic ray model is in good agreement with the observed
CO+ column densities.

The CN column density (Fig. 1.7b) of our enhanced cosmic ray model is equal to the observed
value of 1x 1014 cm'2 at 20” (Simon et al. 1997), and is 5 times higher than the observed value of
2x 1014 at r = 30”. These are decreased to 0.5 and 3.75, respectively, if we adopt a gas phase C
abundance of 30 percent of the solar value (Jansen et al. 1995). In contrast our magnetic model
underpredicts CN by more than a factor of 10 at 20” and by a factor of 3 at r = 30”. We consider
our enhanced model to match the CN observations to an acceptable level, given that the
uncertainties in the molecular data result in uncertainties of up to an order of magnitude in the
absolute chemical abundance of species like CN (Simon et al. 1997).

The SO+ column density (Fig. 1.7c) is observed to steadily increase with distance from the IF
reaching 8.4x 1012 cm'2 (Fuente et al. 2003) at r = 28”. The computed column densities for the

cases including the magnetic field without and with enhanced cosmic rays rise steeply at around

24

r = 20” to 3.4x 109 cm'2 and 1.1x1013 cm'2 respectively, followed by a nearly constant plateau.
Our cosmic ray enhanced model reproduces the observed peak column density, although

additional SO+ emission is observed near the IF and may be due to the background molecular gas.

The peak measured column densities of 50 (Fig. 1.7d) are 6x 1014 cm'2 at 28” (Jansen et al
1995). Taking beam dilution into account, this corresponds to a factor of 6 greater than the
predicted value, which is reasonable agreement.

The observed column density of CS (Fig. 1.7e) is 5x 1013 cm'2 at r = 20” and increases to

5x 1014 cm'2 at 30” (Simon et al. 1997). Hogerheijde et al. (1995) find N(CS) = 1.5x 1015 an2 at
28” for a similar position. The profiles of our magnetic model and enhanced cosmic ray model
have nearly the same value as the observations at r = 20”, however these profiles are not
convolved with the resolution of the observations which is very important in this case due to the
rapid rise in the enhanced cosmic ray model at that point. If the resolution were taken into
account, the enhanced cosmic ray model would have a column density ~2x 1015 cm'2 or 40 times
the observed value, while the magnetic model would match the observed column density to
within a factor of 2. At 30” the situation is the same with the enhanced cosmic ray model
overpredicting CS by a factor of 40. These conclusions are tenuous given the large range in
assumed S abundance relative to H present in the literature. For example Jansen et al. (1995) and
Simon et al. (1997) found the gas phase S abundance relative to H to be 2x 107, while Young
Owl et al. (2000) assumed 7.9x 106. If S/H in the PDR is changed to 2x 107, the abundance
assumed by Simon et al. (1997), our modeled N(CS) would drop to 1.25x 1015 at r = 28”, which
is 0.83 times the observed value. Thus the CS column density taken by itself could easily be
adjusted to fit the observations. However if the lower abundance of S that reproduces CS is used,
the predicted SO and SO+ become many orders of magnitude too faint. Therefore, the predicted

ratio of CS to 30 and 80+ does not match the observations.

25

SiO observations (Fig. 1.70 are available for three areas across the Bar (Schilke et al. 2001). The
observations along their cut labeled “Bar-CO” lie in the same region of the Bar that we are
studying. However, the exact location of the observations is unimportant because the
measurements are statistically consistent with a constant value near 2x 1012 cm"2 (Schilke et al.
2001). All of our models are in disagreement with this result. Our enhanced cosmic ray model
predicts a column density equal to 1x 1013 cm'z, a factor of 5 larger than observed at depths
between 10” 8: 20”. Then N(SiO) sharply increases to 4x 1014 cm'z. Any decrease in the cosmic
ray density by a factor greater than 10 results in an underprediction of SiO for depth shallower
than 20”, but even with the canonical Galactic value used in the magnetic model, there is an
overprediction by at least an order of magnitude at depth greater than 25”. There are two
explanations that may account for the discrepancy. First the gas phase Si abundance is likely to
be depleted by at least an order of magnitude in the PDR compared with the ionized H’ region
(Schilke et a1 2001). However the rate of depletion would have to be matched in such a way as to
maintain a constant gas phase Si density with depth. The second possibility is that the emission is
from an outflow in the foreground, although the velocity profiles suggest this is not the case
(Schilke et al. 2001)

We conclude that for 4 of the 6 diatomic molecules shown, the addition of cosmic rays brings our
model into general agreement (within a factor of 6) with the observations. For CS, our “magnetic
model” agrees well with the observations and the further inclusion of enhanced cosmic ray
heating hurts the agreement, but this depends on the fraction of S depleted onto grains in the
molecular region. For SiO, none of our models agree with the observations. We stress that we did
not use the molecular column densities discussed in this subsection as constraints when we fit our
models to the observations. Rather, we are using them as ex post facto tests of how well our

simulations of the inter-clump medium reproduce a wider body of data. We should also again

26

mention that we did not do a formal calculation of the predicted emission from these molecules.
Some transitions, of CO“ for example, may be significantly sub—thermally populated and would
therefore produce little emission and would have to be attributed to unmodeled clumps. Still
Simon et al. (1997) found that the CN and CS emission could be characterized by a diffuse gas
with a density of 1—4x105 cm"3 despite the critical densities of CN N =3-)2 and CS J =7-96 being

9x106cm'3 and 3x107 cm'3, respectively. This is comparable to our peak density of 9x104 cm'a.

While some moderately complex molecules are included in the UMIST database and therefore in
our computations, the predictions for anything more complex than diatomic molecules are highly
uncertain. There are two sources of uncertainty. The first is in the rate coefficients for the many
processes that lead to the formation or destruction of a species. The systematic errors in the rates
are not possible to quantify. The second systematic uncertainty is in the assumptions that go into
creating an equilibrium model. The comparisons presented in Rollig et al. (2007) between
models computed with different codes show nearly an order of magnitude scatter in the computed
number densities of these complex molecules even in cases where all models used the same fixed
gas kinetic temperature and an agreed upon subset of the UMIST data set. The scatter between
actual thermal equilibrium models was far worse. Because of the resulting large uncertainty in the

computed column densities of these complex molecules, we will not consider them further.

4.3 Sensitivity of final model to input parameters

49.00

Q(H). During the course of this investigation, we explored the parameter space 10 S Q(H) S

1049.23 s-l

, and distance to the nebula 437 S d S 500 pc. We could always find a placement and
tilt of the Bar that would provide a good fit to the observations. It was always the case that the H+
region could be described by a version of the gas pressure model, but that magnetic pressure

support and cosmic ray heating were needed to also match the H0 region properties. Specifically

we found no combination of parameters that produced a gas pressure model in which the H2

27

emission was displaced from the ionization front by the observed distance, nor in which the
observed peak H2 intensity was reproduced.

Distance. If the distance to the nebula were in fact 500 pc (the distance adopted by Wen 8t O’Dell
1995 and many other authors), <B> would be affected in two ways. The projected distance of the
H2 emission from the IF increases from 0.021 pc to 0.024 pc. To match the larger offset a higher
value of Pmag/Pgas is required, with a nonlinear relation between the offset distance and <B>.
Countering this effect, the inferred radiation pressure responsible for compressing the magnetic
field also would drop, since 610fi C would have to be farther away from the illuminated face.
The average magnetic field in the H0 region from Paper 1, eq. 7 has <B> cc 1/R, where R is the
distance to the IF. For a distance of 500 pc, <B> is expected to decrease by 12 per cent. The
combination of these two effects was calculated for the magnetic model without enhanced cosmic
rays. The magnetic field dropped from 448uG to 435 uG.

Inclination. If the Bar is inclined by more than the 7 deg angle assumed here, the observed offsets
would require a greater radial distance between the observed emission peaks and 8 1 on C. Take
for example a model with a calculated radial separation Ax between two emission peaks. For a tilt
angle 4), the projected separation on the sky is Ax’ = Ax cos¢. For increased values of ti, the
modeled radial separation must increase. This increased radial separation would in turn imply a
lower density H0 region, so that a higher magnetic field would be required to maintain hydrostatic
equilibrium. The effect at 7 deg is less than 1 percent. Therefore our models with magnetic fields
represent the lower limit for <B> in the Bar when considering the geometry.

Density and scattered Ha light. The [S 11] ratio found in our model is 0.05 lower than the
observed value, within the 15% uncertainty in the collision rate (AGN3). Wen & O’Dell (1995)
estimate that 1/3 of the surface brightness of Hor and other optical emission lines is due to light

reflected from dust in the molecular cloud, which we have included in our current computed

28

results. If this is wrong and scattered light is not such a large effect, we could compensate by
making the slab somewhat larger along the line of sight.

Turbulence. We have considered turbulence in two ways. For all of the models presented here
the turbulence was fixed to be 2 km 5'1 FW HM, consistent with observations of H2 and 12CO
lines. This turbulence was counted as part of the total pressure. Another approach possible in any
model with a magnetic field is to assume that Pmag = Pturb, motivated by equipartition arguments.
In that case the turbulent velocity would vary as a function of depth and would be about 3 km s"1
in the molecular gas. The associated increase in Pmrb then would lead to a lower magnetic field.
In the enhanced cosmic ray model, the final derived cosmic ray density is in equipartition with
the magnetic field, so the density of cosmic rays would also decrease. We rule out this type of
model for two reasons: (1) the 12CO emission would peak 10 arcsec farther from the ionization
front than is observed; and (2) the molecular gas temperature is predicted to be only 70 K , at
least 20 K colder than the observed temperature of the Bar. The offset is geometry dependent,
while the temperature is not.

4.4 Magnetostatic equilibrium

Each of the three models assumes hydrostatic equilibrium. The left-hand column of Fig. 1.4
shows how the various pressure components from Eq. A2 adjust themselves to maintain this
condition. As was discussed in detail in Paper I, the integrated radiation pressure from absorbed
starlight steadily builds up with depth until all of the incoming photons have been used up at the
bottom of the H0 region. The sum of the other pressure terms must steadily rise to balance this.

It is clear from Fig. 1.4 that in each model there is a large residual gas pressure at the illuminated
face of the cloud (depth = 0). This has to happen in these models because they describe the cloud
as suddenly beginning with some gas density and temperature. The H+, H0 and H2 regions are

actually a flow from cold molecular gas through the atomic region into hot and ionized gas with

29

the ionizing radiation gradually eating into the molecular cloud. For the case of M17 (Paper I) we
found that the residual pressure at the cloud face actually is in equilibrium with the hot bubble of
X-ray emitting gas that is observed to surround the ionizing stars. A similar bubble of diffuse X-
ray emitting gas has recently been found to the SW of the Bar region, although the close
proximity to 6 1 on C and foreground absorption by the Veil blocks a direct view of the Bar in
diffuse X-ray emission (Giidel et al. 2008). The estimated pressure from the X-Ray emitting gas
was found to be roughly equal to the gas pressure of the H+ region. The observed champagne
flow and pressure equilibrium suggest the bubble is a leaky cavity (Giidel et al. 2008).

Our final enhanced cosmic ray model should be a realistic simulation of the Bar for a snapshot in
time. The key feature is that at the same time that the radiation field from 0 1 Ori C is dissociating
and then ionizing the original molecular gas, the momentum canied by the photons has pushed
the gas back into the molecular cloud, compressing both the gas and any magnetic field that is
frozen into it. The natural and straight-forward result is that this compression is halted when the
combination of gas, turbulent, and magnetic pressure has risen enough to offset the radiation
pressure, so the system is in “magnetostatic equilibrium”. The current quasi-equilibrium situation
in the Orion Bar might represent an extrapolation of the situation described in recent MHD
models (Krumholz et al. 2007) of the effects of magnetic fields on the early stages of expansion
of an H 11 region.

4.5 Heating mechanisms

Figure 1.8 shows the relative contribution of each important heating mechanism for each of the
three models. Cosmic ray heating is a minor effect in the gas pressure and magnetic field models,
but in the enhanced cosmic ray model it completely dominates the heating beyond a depth of
about 0.2 pc, which is well into the molecular (H2) region. Cosmic ray heating accounts for 80

percent of the total heating in this region, while the remaining 20 percent is due to cosmic ray

3O

excitation of permitted FUV lines. We note that cosmic ray heating is generally thought to be
responsible for heating molecular clouds (Lequeux 2005).

4. 6 Comparison to previous models of the Bar

Geometry. Almost all previous models of the Orion Bar PDR agree that the observed stratified
H+, PAH and H2 emission must come from a diffuse gas with density 1-5x104 cm'3, rather than
from the superposition of many small optically thick clumps ( i.e. Tielens et al (1993); Tauber et
al. (1994); Hogerheijde et a1. (1995); Young Owl et a1 (2000)). The pressure of this gas is of the
same order as the gas pressure at the IF inferred from the ratio of [8 II] A6716/AG731. The
geometry of the homogenous region can be estimated from the use of optically thin emission
lines. With estimates of the emissivity per unit volume we have used [S 11] A6716+A6731 to find
the depth of the bar to be 0.115pc. Estimates using [0 I] A6300 and FIR continuum measurements
from 20pm to 100pm lead to a similar conclusion. Since these are observations of a region close
to the IF, they only directly trace the geometry at the narrow H+lH0 transition.

A more complete, 3-dimensional model of the IF over the entire nebula was constructed by Wen
& O’Dell (1995), working backwards from the observed Hor surface brightness and projected
positions. Their work indicated that the Bar is an upward corrugation of the main IF, which is the
basic geometry that we have adopted here. We initially attempted to rather closely follow their
result by describing the Bar as a surface steadily curving upward towards the observer, with 6 1
Ori C at the center of curvature. We also experimented with a slab inclined at 20 deg to our line
of sight, which is the tilt of the layer in the Wen & O’Dell models. In neither case were we able to
reproduce the observed [8 II] brightness profile and intensity simultaneously. It should be noted
that Wen 8r O’Dell warned in their paper that their model was not expected to be very accurate in
the region of the Bar. These problems led us to switch to the more nearly edge-on slab geometry

described above. However, our model is still generally consistent with the basic Wen 8t O’Dell

31

picture, especially if the Bar is connected to the background cloud in the way sketched in Fig. 1.3.
We note that the computed H2 profile shown in Figure 1.5 is narrower than the observed one,
which suggests that the H0 region in the Bar does have a more complicated geometry (curvature
or corrugations) than we have assumed here, as Wen & O’Dell suggested. Note that the
separation between 61 on C and the ionization front immediately behind it is fixed by Q(H) and
the Ha surface brightness to be 0.183 pc (this is different than the value found by Wen 8t O’Dell

because we have adopted a different distance to the Orion Nebula, which changes Q(H)).

Hogerheijde et al. (1995) showed that the molecular gas also has a geometry that changes from
face-on to edge-on (the Bar) and then back to face—on. Figure 13 of their paper illustrates the PDR
geometry derived from C180. They assumed a constant abundance ratio Hz/CIBO = 5x 106 and
n(H2) = 5x 105 cm'3 to estimate the volume density of C180. Using the measured column density
of C180 = 1.3x 1016 cm'z, the line of sight path length would be 0.6 pc. Applying our more
detailed calculations and assuming CO/C‘BO ~ 500 (Wilson & Rood 1994) our enhanced cosmic
ray model predicts N(CIBO) = 1.25x 1016 cm'z, in good agreement with Hogerheijde et al. (1995).
However the path length came out to be 0.115 pc, which is smaller than the one found by.
Hogerheijde et al. but is in good agreement with the path length derived above from the [S 11]
lines. In contrast our magnetic model has a predicted N(C 18O) = 2.2x 1016 cm'2 which would
require the geometry to decrease in size by almost a factor of 2 between regions in the PDR. The
similarities of the geometry between the H+ region, IF and PDR therefore support our enhanced
cosmic ray model.

Interclump Region. Our final model largely reproduces the interclump densities used by Tielens
et al (1993), Tauber et al. (1994) and Young Owl et a1 (2000) to model the H0 region. This is a
factor of 5 greater than the interclump medium density proposed by van der Werf et al. (1996).

The main difference in our work is in the starting point of the calculation. Previously, the

32

observed offsets between the emission line peaks were used to establish the H0 region density.
Having assumed a H0 region temperature of 1000 K and a constant density, Tielens et al. (1993),
Tauber et al. (1994) and Young Owl et a1 (2000) argued that the H0 region gas density was
consistent with the ionized gas density for a system with gas pressure equilibrium across the IF. It
is this assumption that our calculations treat in detail by including not only gas and turbulent
pressure but also a pressure gradient caused by absorbed starlight. Starting from the assumption
of hydrostatic equilibrium we constrained our models to match the observed properties of the H+
region of the Bar. Our model using only gas pressure did not correctly describe the density and
geometry of the H0 region. We then added a magnetic field whose pressure is also in hydrostatic
equilibrium, and the H0 region density deduced by Tielens et al. then came out as a natural result.
Our models are most similar to those presented by A05 who considered a constant pressure model
with an equation of state that determines the density as a function of depth, as we have, as
opposed to using a constant density. Unlike van der Werf et al. (1996) for example, A05 used a
single-component medium with a filling factor of unity for the H2 emitting region. This was
motivated by a lack of unambiguous evidence that clumping is important in the H2 emitting
region. A pressure of P/k = 8 x 107 cm.3 K was derived from the electron density and
temperature at the IF. This is the same total pressure found in our models significantly beyond the
IF. However, the spot measured by A05 has a higher surface brightness than the interclump
region we have modeled here. Paper 111 shows that their H2 measurements are matched by the
enhanced cosmic ray model as described here but with a factor two increase in peak H2 density.
The detailed study by A05 determined that a source of extra heating was required to reproduce
the observed level populations of H2 as well as the 100-120 K temperatures where other
molecules form. They were unable to determine a realistic heating mechanism to account for this,

concluding that “future modeling must address the high temperatures in the CO/HCOVN H3 zone

33

of the Orion Bar.” In this paper we propose enhanced cosmic ray heating as a major source of that
extra heating and present the effects of such a high cosmic ray ionization rate on the chemistry of
the Bar.

Clumping. Tielens et al. (1993), Tauber et al. (1994) , Young-Owl et. a1, (2000), van der Werf et
al. (1996) and many others have also considered the effects of small (< 1 arcsec) and large scale
(5—10 arcsec) regions of higher density (clumps). Clumping is directly observed in maps made in
molecular lines from optically thing high levels, such as 12CO J(14- 13) as well as H13CN (Lis &
Schilke 2003). We have not attempted to address this sub-structure in our model.

The Tielens et al. (1993) model for the inter-clump gas component which dominated most of the
observed emission lines under-predicted the strengths of the CO (7—6) and (14—13) lines by a
large factor, so these lines were attributed to emission from clumps. Our final model does in fact
reproduce the observed 12CO (7—6) line strength (as well as 12CO (1—0)) to within a factor of two
(Table 1.1). However, we under-predict the J = 14-13 line by a factor of 4, and clumps clearly are
visible on direct images taken in this line (e.g. Young Owl et al. 2000), so they likely are
significantly affecting the strengths of the higher-level molecular lines. We conclude in general
agreement with all the previously mentioned studies except van der Werf et al. (1996) that the
UV penetration responsible for the observed stratification in the Orion Bar is dominated by a
roughly homogeneous inter-clump medium, with filling factor close to unity. In our view the
density concentrations comprising the clumps are just details on top of this.

Van der Werf et al. (1996) arrived at a model that had a rather lower density interclump medium
(n(H2) = 1 X 104 cm'3) in combination with clumps characterized by two different densities. A
high density component with n(H2) ~ 106 cm'3 was deduced from photochemical models in
which high densities are necessary to effectively produce hot HCO+ and CO+. The 13CO I = 3-2

brightness temperatures of 40 to 50K are typically only reached in PDRs with densities of at least

34

106 cm'3. However, the observed CS line ratios indicate a lower-density clump component with
n(H2) = 2.5x 105 cm'a. Our model overpredicts the CS column density (Fig. 1.7e) that is derived
from the observed surface brightness assuming LTE and a constant gas phase S abundance from
the H+ region through the PDR. However, if the space density of H2 is lower than the critical
density of a few times 105 cm'3, the levels will be subtherrnally populated and the column
densities deduced from the observed CS surface brightness assuming LTE will be too small. Our
derived n(H2) = 4.6x 104 cm'3 is significantly below the critical density of CS so we conclude
that the reported column densities are a lower limit. Likewise if the gas phase S abundance were

lowered in the PDR our predicted CS column density would be lowered.

Neglecting clumps in our treatment is unlikely to affect our conclusions regarding the presence of
a magnetic field for two reasons. First, every paper about clumps (except van der Werf et al.
1996) find the clumps to have a low filling factor (i.e. 0.3% clump filling factor according to
Hogerheijde et al. 1995) so that they do not strongly affect the transport of the UV light
responsible for the overall structure of the H2 emission. Second, the studies that include clumps
such as that of Young Owl et al. (2000) do not require a clumpy ridge to match the HCN and
HCO+ until depths greater than 20” are reached, significantly beyond the peak H2 emission. Even
a clumped medium does not offer a perfect explanation of the emission at this depth. As Young
Owl et al. (2000) noted, significant differences in the HCN and HCO+ surface brightness
distributions indicates that localized variations in the production mechanisms for HCN and
destruction mechanisms for HCO+ are required. These may be explained by density

enhancements caused by local variations in the magnetic field.
4.7 Is enhanced cosmic ray heating a realistic prospect?

Our best-fitting model is for the case where the cosmic ray energy density is in equipartition with

the magnetic field’s energy density. Equipartition occurs in the local ISM (W ebber 1998) where

35

B ~ 8uG and the energy density of cosmic rays is ~1.8eV cm'3, although we do not know of any
first-principle physical reason why this must occur. Still, the idea of equipartition has been used
to argue for the existence of a high cosmic ray energy density in the Arches cluster near the
galactic center (Y usef-Zadeh et al. 2007), with an energy density of 6x 104 eV cm'3. This is an
order of magnitude larger than what we are suggesting here. In the Sagittarius B region the
cosmic rays are thought to be enhanced by a factor of 10 over the Galactic background (van der
Tak et al. 2006).

Gamma ray observations provide a limit to the cosmic ray ionization rate (Ramaty 1996). A
preliminary analysis of COMPTEL satellite observations (Bloemen et al. 1994) suggested a very
high gamma ray flux. Although Bloemen er al. (1999) revised this to a 20 upper limit three times
lower than the originally-claimed detection, Giammanco & Beckman (2005) used the preliminary
value to find a cosmic ray ionization rate of 2—7x 10'13 5'1, depending on the cloud mass. We
rescaled their result to the revised measurement to find a 30 upper limit on the cosmic ray
ionization rate of 3x 10'13 s'1 for H2, as compared to our predicted value of 2x 10'13 5'1. Thus the
predicted gamma-ray flux from our enhanced cosmic ray model is consistent with the upper limit
detected by COMPTEL.

These high energy particles in the presence of magnetic fields also produce synchrotron radiation,
so we next examine whether they are ruled out by radio continuum observations. The total power

emitted per unit volume for a power law distribution of particles is, from chapter 6 of Rybicki and

Lightrnan (2004),

(p-I)
21rmcv T

_\/3q3CBsinor
3quinor

— 2 X1"
mc(p+U

 

eTotaIM 4 12 4 12

 

£+£)X1‘(E_l_

 

 

 

(Q

36

where B is the strength of the magnetic field, in the electron mass, c the speed of light, and a is
the angle between the electron velocities and the magnetic field. C and p parameterize the
distribution of the volume density of relativistic electrons nm as a function of energy E

n CR = C X E '"p (7)
where p ~ 2.4 as measured from the spectra of radio-loud active galaxies (Kellerman 1966) and C
is a normalization constant. We have assumed an initially tangled magnetic field. If the field were
still tangled the synchrotron emission would be isotropic. However the field would become less
tangled as it is swept up with the gas. This will result in beaming that could increase or decrease
the observed flux depending on the exact orientation of the magnetic field relative to our line of
sight. If to first order, we assume the radiation will still be fairly isotropic, the predicted surface
brightness due to 20 cm synchrotron emission from the PDR for our enhanced cosmic ray model
is

S syn (3216? X L = 1.08 X 10.27 erg s'1 cm'2 arcsec.2 Hz'l . (8)
Continuum measurements have been made of the Orion Nebula using both single dish and
interferometric radio telescopes at 20 and 2cm (Felli et al 1993). The single dish measurements
provide a total flux but do not have adequate angular resolution to resolve the Bar. Interferometric
observations using the VLA C and D configurations with a beam size of 28” show that the
observed surface brightness at the same position in the Bar is 1.35x 10'26 erg s‘1 cm'2 arcsec'2
Hz'1 at 20cm. This is a lower limit to the true flux because the total interferometric observation
contains only 60 percent of the single dish flux. Assuming that this lower limit is the true value,
the predicted synchrotron radiation is 8% of the observed value and therefore is consistent with

the observations. The observed flux is dominated by thermal emission.

37

A further question is how long an enhanced rate of cosmic ray heating could go on for, even if the
cosmic ray density does become large at some point in time. With no inflow of new cosmic rays,
the rate at which cosmic rays heat the gas is the same as the rate at which the overall cosmic ray
energy density is decreasing, and this can be used to determine a time scale. An analytical form of
the cosmic ray heating rate can be found in Tielens & Hollenbach (1985). Assuming that the
cosmic rays are trapped in the cloud by a tangled magnetic field, we can then define the lifetime
of a high-energy cosmic ray by

TCR = UCR/ 1‘ CR (9)
' where UCR is the energy density of cosmic rays and I‘CR is the heating rate from Tielens and
Hollenbach. This lifetime depends on the H2 and H0 density, but is independent of the cosmic ray
density. For the gas density m; = 4.6x 105 cm'3 which exists in the appropriate zone of our best
model, the cosmic ray lifetime is only 2000 years. This short lifetime applies to all PDRs with
similar densities. The transport of cosmic rays through a magnetized partially ionized medium is
a rich and complex problem with no easy solutions (see, for example, Lazarian & Beresnyak
2006; Snodin et al. 2006). The details of their motion depend on the field geometry, which is
unknown in this region of Orion. However, the timescale problem we have pointed out is shared
by most models of the neutral and molecular regions. The short timescale is a consequence of the
hydrogen density and not the cosmic ray density.
We have shown that an enhanced cosmic ray density can account for the observed properties of
the Bar. Enhanced cosmic rays will result from compression of magnetic field lines. If the field
is well ordered and connected to the diffuse ISM the cosmic rays will leak out at nearly the speed
of light and the CR density quickly will go back to the background value. If field lines are
tangled, as assumed in our work, the CRs will be trapped and lose their energy through collisional

processes. This is clearly a crucial issue. Recent studies (Indriolo et al. 2007) have shown that the

38

cosmic ray ionization rate varies widely along various sight lines through the diffuse ISM with a
maximum value over 1 dex larger than the accepted background value. Clearly the picture of a
single Galactic cosmic ray background with ordered fields sustaining this rate is an
oversimplification.

4.8 Comparison to the magnetic field in other PDRs

The other star forming region where this type of analysis has been performed is M17 (Paper I).
The edge-on ionization front in M17 has a considerably lower gas density than does the Orion
Bar (the [S 11] ratio indicates ne ~560 cm"3 in M17, as compared to ne ~3200 cm'3 in the Orion
Bar). At the same time, the magnetic fields are measured to be roughly equal in the two Ho
regions. This means that the ratio Pmag/Pgas is much larger in M17, so that the magnetic field has
a much larger (and hence more noticeable) effect. Paper I showed that the magnetic field
produces a very large increase in the extent of the H0 region in M17, as is observed. In the case of
the Orion Bar, Pmag/Pgas ~ 1 in the H0 region (Fig. 1.4) so the Ho region size is increased by only
about a factor of two. In addition, while the magnetic pressure also affects the region where the
[S 11] doublet is formed in M17, this does not really happen in Orion. N one-the-less, the magnetic
pressure does change the overall structure of the Orion Bar by a significant amount, and so should
be included in models of it.

At a depth greater than 0.15pc into the cloud our best model predicts most hydrogen is molecular
and has a density n(H2)=104'66 cm'3, and a magnetic field of 543uG. Crutcher (1999) has
combined most of the available direct measurements of IBIosl for a number of systems, including
molecular cores and star forming regions such as M17, along with estimates of n(H2). Figure 1.9
has been adapted from Figure 1 of Crutcher (1999) to include our point for the Orion Bar. The
filled circles are measured points for other systems, while the triangles represent systems for

which only an upper limit on Bios is available. The Orion point is consistent with the general

39

trend, showing that the magnetic field strength that we have deduced for the Bar is in line with
the values found in the cases where direct measurements are possible.

4.9 The effect of radiation pressure on gas density

In the simplest case, the Bar is an edge-on ionization front where the observed increase in surface
brightness can be attributed to the depth along the line of sight. However, this cannot be the
whole story. The maps of the measured [S II] A6716/A6731 ratio shown by both Pogge et al.
(1992) and Garcia-Diaz & Henney (2007) show that the Bar is also a density enhancement over
the regions to either side of it. We suggest that this increase in density is due to the Bar being
more face-on to the light from 8 1 Ori C than is the surrounding ionized cloud face, so that
radiation pressure due to starlight is greater in the Bar. To illustrate this, we perturbed the
momentum (as set by ¢(H) cc Q(H)) at the illuminated face upwards and downwards by a factor
of 2. Figure 1.10 shows that the density at the ionization front responds roughly according to my
oc ¢(H), indicating that in the H+ zone the momentum p oc (D (H) is balanced by gas pressure.
This effect, the changing flux of momentum in the ionizing radiation field, may be what drives
the density variations across many H II regions. It appears to occur over the Orion Nebula as a
whole. For example, in a strip reaching radially to the W of 9 l on C that is covered by the long-
slit spectrum of BFM91, the electron density falls off steadily as a function of the distance r from
910ri C approximately as ne oc r'l'7 (see their Fig. 7), consistent with the dilution of the incident
momentum entering the cloud per unit area along this strip. Wen & O’Dell (1995) find it oc r'l'63
from a more comprehensive plot of ne vs. r covering the two-dimensional face of the Orion
Nebula (their Fig. 6). While this is not quite as rapid a falloff as the expected r'2 cos(6)
dependence (where 6 is the angle at which the ionizing radiation strikes the IF), it certainly
suggests that the decrease in momentum flux is an important factor. Radially decreasing density

gradients also are found in a number of additional H 11 regions, both Galactic (Copetti et al. 2000)

40

and extragalactic (Castafieda et al. 1992). These may be other cases in which the ionization front
forms a background sheet behind the ionizing star(s), similar to the situation in Orion.

5. Conclusions

We simultaneously modeled the Orion Bar’s H+ and PDR regions in order to find out which
physical processes are important in determining the overall structure. We assumed that the Bar is
in hydrostatic equilibrium throughout; while not likely to be strictly true, this should be a fair
approximation. We compared models that include a magnetic field that is well-coupled to the gas
to models without magnetic fields and developed a final best-fitting model that includes both
magnetic fields and an enhanced cosmic ray density.

Our final model reproduces the observed surface brightness profiles, in the sense of both their
position and the actual brightness, of the H01, [S 11], H2 and 12CO emission lines and also the

[S 11] A6716/A6731 ratio, and the strengths of [0 III], [N 11], [O I], [Sill], [C II] and additional
CO emission lines. This is achieved by varying the location of the illuminated face and the gas
density at the face, the tilt of the IF and its depth along the line of sight, the magnetic field
strength, and the cosmic ray density. These emission lines come from what are really three very
different regions (ionized, atomic and molecular) associated with star formation. Self consistently
modeling all the zones together provides us with additional constraints that are not available when

treating either the H+ or the combined H0 and molecular zones as independent entities.

We find that in the case of the Bar, magnetic pressure plays an important role in the Ho region,
providing about half of the total pressure support. The evidence for this is the offset of the H2
emission peak from the position of the ionization front as marked by the [S 11] emission peak,
which is mostly a measurement of the gas density in the HO region. Since in hydrodynamic
equilibrium the magnetic pressure offsets some of the need for gas pressure, the result is a lower

gas density. For this model we find <B> = 438uG ill the H0 region.

41

The above model still did not provide a very good fit to the location and brightness of the 12CO
J=1—O emission peak, nor did it reproduce the extended H2 emission reaching into the deeper
regions of the cloud. We find that these features can be explained if there is an associated
enhancement in the density of cosmic rays by the factor expected if the cosmic ray and magnetic
field energy densities are in equipartition, with <B> = 516er in the H0 region and the cosmic ray
density 103'6 times higher than the Galactic background value. Although we don’t know exactly
how this equipartition would occur, and there may be problems about the timescales over which
these cosmic rays lose their energy, this is the same situation that we found in M17. This suggests
that an increase in the cosmic ray density may be a natural consequence of the compression of
magnetic fields frozen into these gas clouds. We find that an enhanced abundance of cosmic rays
in a diffuse inter-clump medium can explain the observed CO+ surface brightness profile as well
as the surface brightness profiles of a number of other molecules, although some molecular lines

clearly do come from clumps.

We note that the initial magnetic field (before compression) needed to create such a scenario is
very close to the average field for the Galaxy, suggesting that magnetic fields may be important
in many similar regions of star formation. In both this current model of the Orion Bar and in our
earlier model of M17, the basic concept is that the momentum carried by radiation and winds
from the newly formed stars compresses the surrounding gas until enough pressure builds up
inside the gas cloud to resist (i.e. until an approximate hydrostatic equilibrium is set up). If the
gas starts out with a weak magnetic field, that field can be compressed along with the gas until
magnetic pressure becomes an important factor and magnetostatic equilibrium is established. The
enhancement of the cosmic ray density would then just be an additional side effect. We believe
that this is a very straight-forward, cause-and-effect description of some of the key processes that

shape the gas clouds that are found in star-fomling regions of all sizes.

42

EWP and JAB gratefully acknowledge support from NSF grant AST-0305833. GJF thanks NSF
for support through AST-0607028, NASA for support through NNGOSGD81G and STScI for

support through HST-AR-10653.

43

Appendix A. The gas equation of state

The equation of state is the relationship between the gas density and pressure. The terms we

include in the equation of state are described here.

There are three broad classes of simulation codes that could be applied to a region such as the
Orion molecular cloud. Hydrodynamics codes such as Zeus (Hayes et. al 2006) or Gadget
(Springel 2005) will follow the gas dynamics in detail, often in three dimensions and sometimes
with a treatment of the magnetic field. This class of codes generally does not do the atomic
physics in detail but rather treats thermal and ionization processes with generalized fits to
universal functions and neglect the radiative transfer. Radiative transfer codes such as ATLAS
(Kurucz 2005) or Phoenix (Hauschildt et al. 1997) will do the radiative transfer with great care
but at the expense of the atomic physics and dynamics. Finally, plasma simulation codes such as
Cloudy, which we use in this paper, treat the atomic, molecular, and emission physics with great
care but do the dynamics and radiative transfer with more approximate methods.

All three classes of codes are trying to do the same thing, a true simulation of what occurs in
nature, but are limited by available computers and coding complexity. All are being improved in
the areas in which they are weak but we are still many years away from being able to perform a
true simulation of the spectral emission of a magnetized molecular cloud with an advancing
ionization front.

In this paper we model the H+ / H0 / H2 layers as a hydrostatic atmosphere. This is clearly a
simplification — the layers are actually a flow from cold-molecular into hot-ionized regions. For a
D-critical ionization front the ram pressure, the pressure term due to the motions of the gas, will
equal the gas pressure once the gas has attained its full motion (Henney eta12005b; Henney
2007). This term is not present in a static geometry, so the pressure we use may be off by as

much as a term equal to the gas pressure. In most of the cloud the gas pressure is only a fraction

44

of the total pressure, which includes terms from turbulence, radiation pressure, and the magnetic
field. The uncertainty introduced by the hydrostatic approximation is likely to be of the same
order as uncertainties in the chemistry network or the grain properties.

We treat magnetic pressure as a scalar which is added to the gas and radiation pressures. This is
formally correct if the field is highly disordered. For an ordered magnetic field the forces acting
on a charged particle produce a directed motion rather than a scalar pressure term. We know that
the magnetic field in the Veil, the atomic layer of gas in front on the Orion Nebula (Abel et al.
2005), is ordered since a disordered field would produce no net Zeeman polarization. The field
strength and geometry across the Bar are unknown. The approach we take is simple but

reasonable given the uncertainties and complexities.

The magnetic field is computed assuming flux freezing, that is, that the field and gas are well
coupled. The field at any place in the cloud is related to the field at the illuminated face of the
cloud by the ratio of densities. The gas density at the ionization front is in turn constrained by the
observed [8 II] I(A6716)/I(A6731) intensity ratio. As is explained in detail in Paper I, the

magnetic field is assumed to scale with the gas density according to the relation

n y/2

B=B X
0 n0

 

 

(A1)

Here Bo and no are the magnetic field and the gas density at the illuminated face of the cloud and
7 depends on the geometry of the system. For spherical collapse 7 is 4/3 while 7 = 2 describes the
2D compression of a shell. We choose 7 = 2 on the assumption that the Bar is an edge-on section
of the general shell of material which has been swept up and compressed by the radiation pressure
from 01 Ori C. This is similar to the situation in M17, for which this choice was justified in some

detail in Paper I.

45

The total pressure Pm; is then given by

2

_ B
Ptot(r)"nkT+fi+Pmrb+Plines Pstars(r) (A2)

This is the equation of state assumed in BFM91, who ignored the magnetic and turbulent pressure
terms. They called this a constant pressure model, and Cloudy continues to do so today, although
hydrostatic is a better term. The first term in Eq. A2 is thermal gas pressure, the second is the
magnetic pressure, Pmrb is the pressure from non-thermal turbulent motions, Panes is the radiation
pressure due to trapped emission lines (mainly Lyor), and Psrars is the net pressure resulting from

the absorption of starlight. This last term is given by

r

Pstars(r)=I aradpdr

rl (A3)

This is equal to zero at the illuminated face, where r = r1, and then grows with depth, reaching its
full value near the ionization front, where r = RH. The total outward force is approximately given

by the total momentum in ionizing radiation,
_ o (H,) <h v>
stars 4 "RE! C . (A4)

 

The local pressure is not actually constant, although there is no net force acting on the gas, since
the Pstars term increases with increasing depth. Paper I gives a more detailed description of these

effects.

46

 

Table 1.1
Observed and predicted quantities for the Orion Bar.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

- 1 Gas Pressure Ma netic Enhanced Cosmic

u Ob d r2 8

Quan ty serve (Re ) Rays
[5 II] ratio 0.62 0.59 0.59 0.59

j 5([5 II]A6716+A6731) 5.7e—13 (1) 5.6e-13 5.5e-13 5.6e-13

'1' 5(Ha) 6.6e-12 (2) 6.5e-12 6.5e—12 6.5e-12

l 5(0111) 6.8e-12 (2) 8.2 e-12 8.2e-12 8.2e-12

j S(N II) 2.2e-12 (2) 2.8e—12 2.7e-12 2.7e-12

l S(H2 2.121l1m) 3.3e-15(3) 5.0e—15 3.3e-15 3.1e-15
5(12co J(1-0)) 9.4e-18 (4) 1.7e-17 8.0e-18 7.5e-18
5(12CO J(7-6)) 4.7e-15 (4) 3.0e—15 4.3e-17 7.2e—15
3(12c0(14-13)) 7.1e-15 (4) 2.0e-17 2.6e-21 2.6e-15
5(01145um) 4.7e-14(4) s 2.9e—13 s 1.3e-13 s 1.2e-13
5(0163um) 9.4e-13(4) s 4.4e-12 s 1.8 e-12 s 2.2e-12

l S(Si1134um) 2.1e-13(4) 8.8e-13 3.7e—13 3.9e—13

l. S(C11158 um) 1.2e-13 (4) 2.4e—13 1.2e-13 1.2e-13
<B> ”Gs Unknown our; 438uo 516ttc

 

 

 

 

 

' 1 Peak surface brightness S is in erg s'1 cm.2 arcsec _2.
2 References for peak surface brightness: (1) new SOAR observations; (2) Wen and O’Dell 1995; (3)

| Young Owl et al. 2000; and (4) Tauber et al. 1994.

L

3 <B> is weighted by Tspin/ n(Ho).

47

 

 

Table 1.2
Comparison of face-on predicted optical lines to BFM gpectrum

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Emission Line Obs ervedl Enhanced Cosmic Model/Observed
Ray Model2
[0 II] 3727 1.246 1.43 1.14
[Ne IIIB869 0.165 0.366 0.22
Hy 0.460 0.467 1.02
He I A4471 0.044 0.045 1.04
[0 lg] A4959 1.052 0.955 0.91
[O 1le A5007 3.144 2.875 0.91
[0 1] A55773 0.003 0.0002 0.06
[N II] A5755 0.007 0.008 1.10
He I A5876 0.133 0.136 1.02
[O I] 1163003 0.025 0.010 0.41
[0 I] 7163633 0.007 0.003 0.50
Ha 2.960 2.893 0.98
[N 11] A6584 0.548 0.588 1.07
He I A6678 0.034 0.035 1.04
[S 11] A6725 0.070 0.113 1.61
[S II] A6717 0.026 0.042 1.61
[S 11] A6731 0.051 0.071 1.39
[S II] A6731/AG717 1.642 1.663 1.02
He I A7065 0.059 0.085 1.44
[Ar III] A7751 0.036 0.049 1.36
[S III] A9069+A9532 1.705 1.600 0.94
[S III] 9532 1.452 1.140 0.79

 

 

 

1Line strengths from BFM91, relative to H8.

2 . . . .
Model predictions are for a face-on observation, relative to H8.

3Observed line is blended with night sky emission.

 

48

 

H II (N
I III

III

“(II All“

WI“ I

“\“ZIII N‘jjjII‘ “\le vaII‘\
I

(Milli “n1 1 l
‘“ .“IIIII‘“fl

will“ “WW“ ‘1» (WI “II":“IIII NW (‘1 W11
III M

 

II . ‘ ‘ ‘ I
Figure 1 osrtions of data across the Orion Bar used in this analysis, shown

superimposed on a dereddened Ha image provided by C. R. 0’ Dell. The lines
included for each cm are: Wen 8i O’Dell (1995), Ha; Tauber et al. (1994), H2 and

12CO. The image is rotated so that the ionizing radiation strikes the Bar from the left,
the same as in Figs. 1.2, 1.3, 1.4, 1.5, 1.7 and 1.8.

49

The Orion Bar

 

12.....-

Ha [3 ll] H2 12co

0.8
0.6
0.4

0.2

Intensity / Peak Bar Intensity

 

 

 

 

o. o I n n n n n A 1 A A n
40 -20 0 20 40

Figure 1.2. Observations of the Orion Bar from Tielens et al. 1993 (”CO ), Wen and O’Dell 1995 (Ha),
Yo ung Owl et al. 2000 (H2) and this paper (811 [A6716+AG731]), all relative to the IF defined by the peak

in the [S [I] emission. 61 Ori C is to the left at —111 arcsec. There is clear stratification indicating an
ionized region viewed nearly edge-on.

50

Scale: 472 arcseclpc

Line of sight

Q

Orion Bar IF, inclined
01 on C 7° to our line of sight

0.182pc

 

’
-' " Illuminated 0-1159c

Face

 

 

 

J
l

——

0.235 pc (111 arcsec)

Figure 1.3. The geometry derived from the [S 11] emission at the ionization front. The Bar is well
represented by a slab 0.115 pc long inclined at 7 deg to the viewing angle.

51

H'essure (P/k )/ 107

 

 

'7

 

 

   
 

 

 

    

 

 

 

wicks/incl _,____..m_._; .-

60?" 0-5

20 """""""" 0.2 I

1 .

6.0;-

20%-

12.

6.0:

o‘hgahthll 1.1; :...r
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.00 0.05 0.10

Depth (PC)

 

1.0 E

 

1.0 l
0.6 :-
0.2

 

5.0
3.0 ,-
1.0 :'

 

(8)113) S01 / htsuau

I’l

J—JIK J

 

 

0.15 0.20 0‘25 030

Depth (PC)

Fi gure 1.4. Pressure and density results from the three basic models developed here. The left-hand column
shows the various pressure components (gas, magnetic, absorbed radiation from stars, turbulence) as a
fig nction of depth into the cloud from its illuminated front surface. The gas, magnetic, integrated starlight
and turbulent pressures are shown as a solid, long dash, dotted and short dashed lines. The right-hand

column shows the number density of H atoms in the HI, H0 and H2 zones, as a function of depth, so that
the pressures shown on the left can easily be related to specific zones in the model. The densities are shown
us in g solid, dashed and dotted lines, respectively.

52

Offset from ionization front (pc)

8 -0.10 .0.05 0.00 0.05 0.10 -0.10 -0.05 0.00 0.05 0.10 -0.10 -0.05 0.00 0.05 0.10

I I Y V V V Y ' 1 I I

 

vvr'vvw—vtfirva—Vf‘v—VT fifV‘V’V'fTV’fi Y

Gas Pressure Model Magnetic Pressure Model 3% Enhanced Cosmic Ray Model?

< I-

     
   
  

I
r

   
  
 

sas II])x1013
A
N .

I-

I- I

t .........000 I.
. > o

rifts II

p
o
.0
r .00....
I

'0

‘ S"

o
vtwvivvvv
.

AAAAAAAAAAAAAAAAAAAAA

 

iiurmrnii‘
'7 '7 Y‘

O
144;;1111
Y VIVY

11LJLL4ALLLL‘LL,ALLAJLLJALL
vvvvvvvvvvv

 

 

 

 

 

 

 

 

 

 

v—d
C
v—l
K
A
8
(.1)
d F P
., r- I-
. i- u
. n .. _ +
‘1 P ‘
'4 4t I- <
K r > C .
A d
A «:- .r <
" h
é‘ .; u .
b O‘ L I I .C‘
m 1 :' ..:: I: cl
. i 1
11250-0) - 1 +123 ~st(1-0) 4
b h 1 > '4
, X140_:H::¢44¢H44.+ AAAAAAAA I-I'E HHfi, .. ........... .
o D : I E
s 120 L' .00. 1E. .00. _ — 1
‘5 : : .'_ E: .' o d E
‘- 100 r .‘ - . 1 : .' '- .. ‘ ' .J
F I O . . _, p . . O <
3 : 0 0.. 0 : >- . O . . :1
E 80 : 5 .115 f 'o: r :1
3 ' ~2: '.: 4
.2 so a ,: a; . , _ .5
' o .. r _ 0‘
: e l‘: . :
2 40 = - i: .- l: 1
c : '0‘. 1: no. : :
:3, :12 d” 1:12 : :12 <
q o '4 #141 1‘LL4IA L 14 l A l A 1 14+; 1 L 1" " ,_L A 1 1 5 L4 1" L4 A L A . A l l 1 A 1 L A 11111111111111

 

so-4o-2o o 2040-0040-20 0 20404304040 0 204a
Offset from ionization front (arcsec)

F i gure 1.5.The surface brightness distributions in key emission lines, in units of erg s'1 cm.2 arcsec _2 or
ElIoitexama temperature, as computed for the three basic models (solid lines), compared to the observed
(118 tributions (dotted lines).

53

 

 

 

 

T I I l I I I l l I l l l I ‘l l I l l I l I l l
: Equipartiton_
120 ~— ———————————————————————————— N"—
- Observed Range in Bar J
(Tauber et al. 1994) _
I ........................... 1.--
$2 80 e // _
I—— I / '
l- ‘/ _
t _

/

4O — / ,z _
_ ”’. _.
.. r” ..

0 I l l m I l I I I l L I I I I l I 1 l L I I I I
0 1 2 3 4

Log [ nCR/ nCRO]

Figure 1.6. Predicted 12CO brightness temperature as a function of cosmic ray density normalized by the
Galactic background cosmic ray density nCRo.

54

13

12

11»

10-

15

14

\oglColumn Density)

11

15

14

13-

12-

11

(a) CO 3
o o O . .
r . Storzer eta. 1995
II o Fuente et al. 2003
I
, I
I
l fl . ,J .....................
E 0 Simon etal. 1997
f (b) CN
t
E o Fuente etal. 2(D3 1
r (c) 80” 3
' i
4
0 fi‘ 3 /j
o "'
I / 1
........... :.......r.—~-I.l.
-1o 0 10 20 30 4o

 

  
 

vvvvvvv

 

 

13-

12>

 

 

 

 

 

 

 

18E-
17g
16%
15;
14:-
13f

17;.

16

153

13’

16

14

12

o Jansen et al. 1995 ;
E((1)50 3
I’ 3
I E
I .
I
‘I
' I
.LQJLA.
.Simonetal. 1997
oI-bgerheijde etal. I1995
(e)CS
r
III I!
l I
I i’
I I”
I
v vvvvvv lvruvvvvvva‘vaerIC.
2 o Schilkeetal. 2001
[(0 SiO
L
E o
-10 o

 

 

 

 

14E

 

  

 

 

 

 

 

IF offset (arcsec)

Figure 1.7. Diatomic molecular column densities (in cm'z) for (a) CO+, (b) CN, (c) 80+, ((1) 50, (e) CS,

and (f) SiO. Modeled and observed values in cm_2 as a function of angular projection from the ionization
front. Shown with a short dash, long dash and solid lines are the gas pressure, magnetic pressure and
enhanced cosmic ray models. Dots show the various observations of each molecular species.

55

 

1.4 '
1.2
1.0
0.8
0'6 . ' photo-
0’4 IOl‘tlZ. electric >V cosmic ray excitation
0.2 ________ ’//>.. ”Mt. ..... ...

0.0 A '
1.2
1.0
0.8

I I I I I I I I I I f

Enhanced Cosmic Ray
cosmic ray heating
r '— _____
\ //

   
 

  
  
  
 

l l
I I

u-lt—
+ '
unr-
.b

F

l
I I I

H ionization

 
 
   
 

 

  

0.6
photo-
3'4 yctric
.2 _____
0.0 \l I I — T 1“ ~I

 

I I I T I I

Fraction at iotat Heating

1.2
1.0
0.8
0.6 ~
0.4

H ionization

  
 
  

'fil/ dust
I _

‘t‘o I (63 um)

 
  

photo-
electric

 
  
 

I5.—

 

 

o.2 ; _______ - ________
0.0 a .
0.0 0.1 0.2 0.3 4
Depth (pc)

Figure 1.8. Heating mechanisms in the three models. The line styles indicating each mechanism are the
same in each panel. Photoelectric, H2, C I, dust and O I (63 um) heating are as defined by Tielens 8r

H 01 lenbach (1985). We also show heating by H I and He II photoionization in the H+ region, and heating of
the molecular gas by direct cosmic ray heating and also by cosmic ray excitation of permitted FUV lines.

56

 

 

 

 

2 4 6
log (H2)

Figure 1.9. Magnetic field strength vs. H2 gas density, adopted from
Crutcher (1999). The star indicates our new result for the Orion Bar. The

filled circles are other systems for which 3105 measurements are
available, and the triangles are other systems for which upper limits on

Bios are available.

57

[6 ll] 6716/6731

0.66

 

0.64
0.62
0.60
0.58

0.56

 

IIIIIIIIIIIIIIIIIIIIIIITIIIIIIIjIIIII

 

O) \l to
log [Ne(104/Te)1’2]

9”
U1

llllllllllllllllllllllllllLlllllll
w
m

 

 

12.8
log (D

Figure 1.10. Predicted [S II] ratio and density vs. the incident ionizing photon flux (DO-l).

58

References
Abel, N.P., Brogan, C.L., Ferland, G.J., O'Dell, C.R., Shaw, G. 8: Troland, TH. 2004, ApJ, 609,

247
Abel, N.P., Ferland, G.J., Shaw, G. 81 van Hoof, P.A.M. 2005, ApJS, 161, 65

Abel, N.P., Ferland, G.J., O'Dell, C.R., Shaw, G. & Troland, TH. 2006, ApJ, 644, 344

Abel, N .P. 8: Ferland, G.J. 2006, ApJ, 647, 367
Allers, K. N., Jaffe, D. T., Lacy, J. H., Draine, B. T. & Richter, M. J. 2005, ApJ, 630, 368 (A05)

Bakes, E. L. 0. 8t Tielens, A. G. G. M. 1994, ApJ, 427, 822

Baldwin, J. A., Ferland, G. J., Martin, P. G., Corbin, M. R., Cota, S. A., Peterson, B. M. &
Slettebak, A. 1991, ApJ, 374, 580 (BFM91)

Balick, B., Gammon, R. H. 8: Hjellming, R. M. 1974, PASP, 86, 616
Black, J. H. 8: van Dishoeck, E. F., 1987, ApJ, 322 412
Bloemen, H. et al 1994, A&A, 281, L5

Bloemen, H. et al., 1999, ApJ, 521, L137

Brogan, C.L. & Troland, TH. 2001, ApJ, 560, 821

Brogan, C. L., Troland, T. H., Abel, N. P., Goss, W. M. & Crutcher, R. M. 2005, ASP
Conference Series, 343, 183

Brogan, C. L., Troland, T. H., Roberts, D. A., & Crutcher, R. M. 1999, ApJ, 515, 304

C astafieda, H.O, Vilchez, J .M. & Copetti, M.V.F 1992, A&A, 260, 370

C rutcher, R. M., 1999, ApJ, 520, 706
C Opetti, M.V.F., Mallmann, J .A.H., Schmidt, A.A. & Castaneda, H0. 2000, A&A, 357, 621

Draine, B. T. & Bertoldi, F. 1996, ApJ, 468, 269

Draine, B. T. 1978, ApJS, 36, 595
Draine, B. T. et al. 2007, ApJ, 663, 866

Fei gelson, E. et al. 2005, ApJS, 160, 379
Fella, M., Churchwell, E., Wilson, T. L. & Taylor, G. B., 1993, A&AS, 98, 137

Ferland, G.J. 2001, PASP, 113, 41

Fel‘land, G. J. Korista, K.T. Vemer, D.A. Ferguson, J. W. Kingdon, J.B. & Vemer, EM. 1998,
PASP, 110, 761

I”Illente, A., Rodnguez-Franco, A., GarCIa-Burillo, S., Martm-Pintado, J. 8: Black, J. H 2003,
A&A, 406, 899

59

Garcia-Diaz, Ma. T. 8: Henney, W. J. 2007, AJ, 133, 952
Giammanco, C. & Beckman, J. E., 2005, A&A, 437, L11

Giidel, M., Briggs, K.R., Montmerle, T., Audard, M., Rebull, L. & Skinner, S.L. 2008, Science,
319, 309

Hanson, M. M., Howarth, I. D. & Conti, RS. 1997, ApJ 489, 698
Hauschildt, P. H., Baron, E. & Allard, F. 1997, ApJ, 483, 390

Hayes, J. C., Norman, M. L., Fiedler, R. A., Borden, J. 0., Li, P. S., Clark, S. E., ud-Doula, A. &
Mac Low, M-M. 2006, ApJS, 165, 188

Henney, W. J. 2007, Diffuse Matter from Star Forming Regions to Active Galaxies, Hartquist,
T.W., Pittard, J .M., Falle, S.A.E.G (Dordrecht : Springer), 103

Henney, W. J., Arthur, S. J. & Garcia-Dial, Ma. T. 20053, ApJ, 627, 813

Henney, W. J., Arthur, S. J., Williams, R. J. R., Ferland, G. J. 2005b, ApJ, 621, 328
Hirota, T. et al. 2007, PASJ, 59, 897

Hogerheijde, M.R., Jansen, DJ. 8: van Dishoek, E. F. 1995, A&A, 294 792

Indriolo, N ., Geballe, T.R., Oka, T. & McCall, B.J. 2007, ApJ, 671, 1736

Jansen, D. J., Spaans, M., Hogerheijde, M. R. & van Dishoeck, E. F. 1995, A&A, 303, 541

Kellerman, K]. 1966 ApJ, 146,621

Kristensen, L.E., Ravkilde, T.L., Field, D., Lemaire, J. L. & Pineau Des Foréts, G. 2007, A&A,
469, 561

Krumholz, M.R., Stone, J.M. & Gardiner, TA. 2007, ApJ, 671, 518
Kurucz, R. L., 2005, MSAIS, 8, 189

Kurucz, R. L., 1979, ApJS, 40, 1

Lazarian, A. & Beresnyak, A. 2006, MNRAS, 373, 1195

Lequeux, J. 2005, The interstellar medium, Translation from the French language edition of: Le
Milieu Interstellaire by James Lequeux, EDP Sciences, 2003 Edited by J. Lequeux.

Astronomy and Astrophysics Library, Berlin: Springer, 2005.

L eurini, s. et al. 2006, A&A, 454, L47

Miao, Y., Mehringer, D.M., Kuan, Y-J & Snyder, LE 1995, ApJ, 445, L59
0 ’ Dell, CR. 2001, ARA&A, 39, 99

0'Dell, C. R & Doi, T. 1999, PASP, 111, 1316

O'Dell, C.R., Walter, D.K. 8t Dufour, R.J. 1992, ApJL, 399, L67.

60

Osterbrock, D.E. & Ferland, G.J. 2006, Astrophysics of Gaseous Nebulae and Active Galactic
Nuclei, 2nd Edition, University Science Books, Sausalito (AGN3)

Osterbrock, D.E., Tran, H.D. & Veilleux, S. 1992, ApJ, 389, 196
Pellegrini, E. W. et al. 2007, ApJ, 658, 1119 (Paper I)

Pogge, R.W., Owen, J. M. & Atwood, B. 1992, ApJ, 399, 147
Ramaty, R. 1996, A&AS, 120, 373

Roshi, D.A. 2007, ApJ(Letters), 658, L41

Rollig, M., et al., 2007, A&A, 467, 187

Rybicki, G. B. & Lightman, A. P. 2004, Radiative Processes in Astrophysics, lst edition, Wiley-
Interscience,Weinheim

Rubin, R. H., Simpson, J. R., Haas, M. R., & Erickson, E. F. 1991, ApJ, 374, 564

Rubin, R.H., Dufour, R.J. & Walter, DR. 1993, ApJ, 413, 242

Schilke, P., Pineau des F oréts, G., Walmsley, C. M. & Martin-Pintado, J. 2001, A&A, 372, 291
Schleuning, D.A., 1998, ApJ, 493, 811

Shaw, G., Ferland, G. J ., Abel, N. P., Stancil, P. C., 8: van Hoof, P. A. M. 2005, ApJ, 624, 794

Shaw, G., Ferland, G. J., Henney, W. J., Stancil, P. C., Abel, N. P., Pellegrini, E.W., Baldwin,
J .A. & van Hoof, P. A. M. 2008, ApJ, (submitted) (Paper 111)

Simon, R., Stutzki, J., Stemberg, A., Winnewisser, G., 1997, A&A, 327, 12

Snodin, A.P., Brandenburg, A., Mee, A.J. & Shukurov, A. 2006, MNRAS, 373, 643
Smith, L.J., Norris, R.P.F. & Crowther, PA. 2002, MNRAS, 337, 1309

Springel, V., 2005, MNRAS, 364, 1105

Stemberg, A., Hoffmann, TL. 8: Pauldrach, A.W.A. 2003, ApJ, 599, 1333.

Sttirzer, H., Stutzki, J. 8:; Stemberg, A. 1995, A&A, 296, L9

Tauber J. A., Tielens, A. G. G. M., Meixner, M. & Goldsmith, P. 1994, ApJ, 422, 136
Tielens A. G. G.M. & Hollenbach, D. 1985, ApJ, 291, 722

Tielens, A. G. G. M., Meixner, M. M., van der Werf, P. P., Bregman, J., Tauber, J. A., Stutzki, J .,
& Rank, D., 1993, Science, 262, 86

Vacca, W. D., Garmany, C. D. & Shull, J. M. 1996, ApJ, 460, 914
van der Tak, F.F.S. et al. 2006, A&A, 454L, 99
van der Werf, P. P., Stutzki, J., Stemberg, A., Krabbe, A., 1996, A&A, 313, 633

61

Wade, G.A., Fullerton, A.W., Donati, J.-F., Landstreet, J .D., Petit, P. 8: Strasser, S. 2006, A&A,
451, 195

Walmsley, C. M., Natta, A., Olivia, E., & Testi, L. 2000, A&A, 364, 301

Webber, W.R. 1998, ApJ, 506, 329

Wen, Z. 8: O’Dell, C. R. 1995, ApJ, 438, 784

Williams, J. P., Bergin, E.A., Caselli, P., Myers, P. C. 8: Plume. R. 1998, ApJ, 503, 689
Wilson, TL. 8: Rood, R. 1994, ARA8:A, 32, 191

Young Owl, R.C., Meixner, M.M., Wolfire, M., Tielens, A. G. G. M. & Tauber, J. 2000, ApJ,
540, 886

Yusef-Zadeh, F., Wardle, M. 8: Roy, S., 2007 ApJL, 665, 123
Zuckerman, B. 1973, ApJ, 183, 863

62

Chapter 2

Deconstructing the Structure and Physical Processes of 30 Doradus
Abstract
We have completed a study of a 140 x 80 pc region of 30 Doradus centered on R136
based on a new optical imaging and spectrophotometric survey covering key diagnostic
emission lines, together with photoionization modeling with CLOUDY. We focus on the
physical conditions, geometry and importance of radiation pressure on a point by point
basis to replace with constraints some of the assumptions that go into modeling and
interpreting distant sites of star formation. The observed survey is a compromise between
spatial density of the grid and areal coverage. The final science quality data products are
available to the public via the NV 0. While stellar winds and supernovae undoubtedly
produce shocks, we find no spectral signature to indicate shocks are currently important
for driving the large scale structure. We conclude that the considerable region covered by
our survey is well described by photoionization from the central cluster where the
ionizing continuum is dominated by the most massive 0 stars. The completeness of our
survey allows us to create a composite spectrum of 30 Doradus, simulating the
observable spectrum of a spatially unresolved, distant HII region. Using this spatially
integrated spectrum we confront various strong line abundance techniques, which are
typically applied to an entire HII region or galaxy, against abundance estimates using
spatially resolved observations of 30 Doradus. We find that despite averaging over more
than an order of magnitude in ionization parameter and density, the abundances estimated
using the diagnostic lines of the composite spectrum are in agreement with spatially

resolved abundance studies.

63

1 Introduction

Star formation is a fundamental step in the continuous chemical and structural evolution
of the universe. Intense star formation is an ongoing process of cloud collapse, stellar
birth and the inevitable enrichment and destruction of natal clouds. Light and winds from
a newly-formed star cluster interact with the parent molecular cloud. This sculpts the
geometry of the regions, produces the observed emission-line spectrum, and most
importantly throttles the rate of star formation. This ongoing process is tracked in other
galaxies throughout the universe via strong emission lines that are characteristic of H II
regions. Because of the distances between galaxies, resolving all but the largest spatial
structures of most extragalactic H II regions is not possible. Therefore our understanding
of star formation in distant parts of the universe is largely based on models of the [SM
and extrapolation of nearby systems to explain the observed emission lines.
To determine how massive clusters throttle star formation, we are making detailed studies
of a succession of larger and larger local star-forming regions, so far of the Orion Nebula
and M17 and now 30 Doradus (30 Dor). These nearby objects offer high spatial
resolution of the many structural details that in more distant objects would be blended
together and thus confuse the physical interpretation. The Orion Nebula is dominated by
a single 0 star and is the smallest scale we will consider. The 0 star forms a blister H II
region on the surface of a background molecular cloud. Multiple studies have explored
the correlation between photon flux and gas density. In Pellegrini et al. (2009; hereafter
P09), we showed that the detailed physical conditions of a bright IF called the Orion Bar
were set by the absorbed energy and momentum of stellar radiation. A similar situation

was realized in the larger H II region of M17 (Pellegrini et al. 2008; P08) where the

physical conditions were in part determined by the integrated momentum in starlight. In
both regions we found a magnetic field of plausible strength was likely supporting the
molecular gas against collapse from the integrated starlight's pressure.

To move these calibrations up to a scale more like that of distant giant star forming
regions, we turn here to 30 Doradus. At a distance of 48.5 kpc (Marci 2006) it is the only
giant H 11 region in the local group we are capable of studying on a sub-parsec scale
using ground based observations.

30 Doris the largest star-forming region in Local Group, with almost 500 times more
ionizing photons than the Orion Nebula. Other nearby large H 11 regions include
NGC3603, the largest optically visible star forming region in the Milky Way. This very
young system is found in the plane of the galaxy and suffered significant extinction. The

cluster in NGC3603 is very compact, containing more than 50 O-stars within a 2 parsec
diameter, generating at least 1051 ionizing photons 5’1. NGC 604 the second largest star

forming region in the Local Group, is approximately 1/4 the size of 30 Doradus, but is 20
times farther from the Sun.

30 Doradus stands out in any wide-scale optical image of LMC. and is nearly visible to
the naked eye as a smudge in dark skies. The central 13 pc contain most of the
approximately 2400 massive O and B stars (Parker 1993) ranging in age from 1-10 Myr.
This includes the dense star cluster known as R136 which includes more 04 and 03 stars

with masses in excess of 100 M0, among the highest luminosity stars that are known.

The total cluster mass is of the order 104 — 105 Me with 2-3 times more ionizing stars

than any cluster in the Milky Way.

65

The violent history of the region has led to vast cavities and shell like structures around

R136. These structures are visible in optical and IR ionization-front tracers like [S II] and

Burn PAH emission. Many of these structures have Ha velocity profiles which show

them to be expanding shells. Filling these cavities is a 106 K gas visible in X-ray
emission. The correlation between this emission and the expanding shells lead to the

conclusion that the SNe were likely the cause of the current dynamics (Chu 8: Kennicutt
1994). These shells have typical kinetic energies equal to 1051 erg. The total kinetic

energy in the vicinity of R136 is 1052 ergs. Within reasonable estimates about the
timescale for injecting energy from stellar winds into the ISM it is equally likely that
stellar winds are the energy source. These types of kinematic arguments suggest SNe or
stellar winds are a likely source of energy.

The remaining molecular gas cloud has been observed in optically thick CO emission
(Poglitsch 1995). This molecular cloud runs N E-SW through R136. The brightest arcs
appear to trace the surface of this molecular cloud, and are bright because of their high
density. 30 Dor has been named the “Tarantula Nebula” because images showing
emission lines are dominated by the interwoven pattern of these bright arcs. However the
majority of the line emission originates in the extended regions of low surface brightness,
which tend to be associated with X-ray emission.

Through the use of modeling, we test the assumption that 30 Doradus can be explained as
a scaled up version of smaller H 11 regions. Specifically we test whether the optical
emission-line spectrum can be explained by a nebula with a constant metal and dust

abundance, as determined by deep spectroscopic studies, being ionized by a single young

66

stellar central cluster. If not, the nebula may be too chaotic to be adequately described

by just a few global parameters, as is commonly done for distant GEHRs. We will
conclude by considering what we could tell about 30 Doradus if it were at a distance such
that it could only be observed via an integrated spectrum.

The modeling process of such distant H 11 regions has been refined to eliminate as many
free parameters as possible (Dopita et al. 2006). The energy injected into a system can by
modeled with the stellar synthesis code Starburst99 (Leitherer et al. 1999). This provides
the time dependent evolution of a stellar population of a given metallicity, including the
spectral energy distribution (SED), the total number of ionizing photons, and mechanical
energy released by stellar winds and supernova. Other necessary input parameters
include an initial mass function (IMF), and either total cluster mass for an instantaneous
burst of star-formation, or a stellar formation rate if continuous. With the assumption of
pressure balance between the H II region and surrounding material, the radius and
density of the nebula at any given time are described by a shell, swept up by the
mechanical energy of the supernova and winds (Dopita et al. 2006b). These assumptions
uniquely determine the instantaneous ionization parameter U of the idealized H 11 region,

where U is defined as is the flux of photons per hydrogen atom defined as

Q
U =—°2—- (1)

41rcr nH

where Q; is the number of ionizing photons from the ionizing source, r is the distance
from the ionizing source to the cloud and 11;; is the density of hydrogen. In a typical H 11

region the fraction of neutral hydrogen is very low and m; = ne.

Finally, the emission line spectrum as a function of metallicity can be modeled with

67

plasma simulation codes such as CLOUDY or MAPPINGS to determine the gas
abundances. This paper will test how well this procedure works in the case of a single,
resolvable Giant Extragalactic H 11 Region (GEHR).
2 Observations

2.1 Existing optical passband data sets

Most of the strong optical emission lines from star-forming regions trace ionized gas at

electron temperatures Te ~104 K. A few key lines from O I, O II, N 11 and 5 11 form in the

interface region at the edge of ionization-bounded clouds. The elements 5, O, N, Ar have

transitions visible from the ground that are sensitive to both Te and the electron density
n,, and also to gas phase abundances. Even though 30 Doris a key example of a GEHR,
the available measurements of it in the emission lines of these elements, using either
direct imaging or especially spectroscopy, are surprisingly limited.

The best existing publicly available narrow-band optical image data sets are the

Magellanic Cloud Emission Line Survey (MCELS), and archival HST images. The

MCELS survey (Smith 2005) covers the central 8X8 degrees2 of the LMC including 30

Dor, in Ho: + [N II], [0 III] and [S II] (using many of the same emission line and

continuum filters listed below in Table 2.1). However, the data were taken with the 0.9m

Schmidt telescope at CTIO and have a spatial scale of 2.3 arcsec pixel”1 with a resolution
closer to 5 arcsec FWHM. This means that structures smaller than 1.2 pc are unresolved,
blurring fine details in the ionization structure that otherwise could be used to remove
ambiguities in the interpretation of emission line spectra.

The HST archival data covers the central 4.5 by 3 arcmin of 30 Dor with at least 0.1

68

arcsec resolution (Scowen 1998, Walbom et al. 2001) in Ha, [0 III] and [S 11], revealing
vastly more detail. However the HST data are limited to the brighter central region
around R136. Since the brighter nebula accounts for only 25 percent of the total nebular
emission in Ha the limited spatial coverage of HST misses the bulk of the emission that
would be detected if 30 Dor were viewed from a much greater distance.

Turning to the available spectroscopy, Krabbe 8: Copetti (2002) obtained a set of long-

slit observations which covered H8, [0 III] M363 and [0 III] 76007 at 135 points along
three slit positions. The sensitivity of these spectra is comparable to our Blanco survey
described below, and the data were used to measure temperature fluctuations (see Section
4.4, below).

Mathis, Chu & Peterson (1985) had previously obtained spectra covering a similar
wavelength range at four other slit positions. They presented line strengths measured for
22 extracted regions covering typically 30 to 60 arcsec each along the slit.

Chu 8: Kennicutt (1994) used the CTIO echelle spectrograph in a single order to obtain a

sparse grid of long-slit spectra that covered the Ha and [N 11] It)» 6548, 6584 emission

lines with high (20 km sec'l) velocity resolution. They found that about half of the kinetic

energy in 30 DOT is contained in shells in the central regions which are expanding with

characteristic velocities v ~ 20-200 km sec'l, and that the kinetic energy contained in this
expansion greatly exceeds the gravitational binding energy.

There has also been a limited amount of deep echelle spectroscopy covering a much
wider wavelength range (Tsamis et al. 2003; Peimbert 2003). These spectra measured

hundreds of emission lines that can be used for detailed chemical abundance analysis,

69

and also cleanly resolve the density sensitive [0 II] 78737263729 doublet and many other
lines which are blended at lower spectral resolution. However, they require relatively
long exposures and very short slit lengths, so have only very small areal coverage on 30
DOT.
Each of these existing measurements will be discussed below as needed.

2.2 New Narrow-Band Images

To the existing data sets we added a new set of narrow-band images taken with the

SOAR Optical Imager (801) on the 4.1m SOAR Telescopes. We used the Ha 6563 x75,

[S II] 6738X50, [0 III] 5019><15 emission-line filters and the 6850 x100 and 5130><155

continuum filters from the CTIO 3X3 and 4X4 in2 filter sets, where the filter names refer
to the approximate central wavelengths and FW HM bandpasses. A summary of the 801

observations is given in Table 2.1.

In each passband we took grids of SXS arcmin2 801 images, overlapping them to give a

12X 13 arcmin2 field of view with a scale of 0.15 arcsec pixel‘l. The individual images
were then combined to create final mosaic images in each passband. Due to the
significant spatial overlap of the individual images, the total integration time in a given
filter varies across the mosaic. Table 2.1 includes for each filter the resulting minimum
and maximum integration times at any point in the mosaic. The data are seeing-limited,
with full width at half maximum intensity FWHM = 0.5—0.9 arcsec, much better than the

5 arcsec FWHM in the MCELS survey, and are three times more sensitive to diffuse

 

5The Southern Astrophysical Research (SOAR) Telescope is a joint project of Michigan State
University,Ministério da Ciéncia e Tecnologia-Brazil, the University of North Carolina at Chapel Hill, and
the National Optical Astronomy Observatory. Further information about SOAR and its instruments may be
found at www.soartelescope.org.

70

nebular emission than the HST data set.

As an example of the results, Figure 2.1a shows the mosaic image made with the Ha +
[N II] filter, prior to continuum subtraction. The central cluster R136 is marked with a
white cross. The 1 arcmin scale bar is equivalent to 14.1 pc for an LMC distance of 48.5
kpc (Marci 2006). Figure 2.1b shows, on the same image, the slit positions used in the
Blanco spectroscopic survey described below.

The two continuum filters were used to measure the emission from point sources (stars),
together with the nebular continuum and starlight scattered off dust. The continuum

image count rates were scaled and subtracted from the emission line image to give a
nebular count rate Rline in each filter, according to

_ _ narrow
line — Rnarrow Rcont X W ’ (2)
cont

R

where Rnarrow and Ram are equal to the atmospheric-extinction corrected count rate for

the narrow band and continuum filters, respectively, and the effective filter width W,- in

the ith filter is

wi=I deA (3)

where the transmission curve T}, was measured by the SOAR staff before the beginning

of the program.

Changes in stellar scattered light, seeing or sky brightness between the time the emission
lines were measured and the continuum was measured were found to be a limiting factor
in the quality of the continuum-subtracted images, so the observations with the

continuum and emission filters were made sequentially one after the other before

71

offsetting the telescope to each new point in the mosaic grid.

2.3 Spectrophotometry

2.3.1 The Blanco Telescope Spectral Grid

In Feb 2008 we obtained a grid of long-slit spectra of 30 Dor using the RC spectrograph
on the 4m Blanco telescope at CTIO. The spectrograph slit was 5 arcmin in length, and
was positioned at a total of 37 different locations spanning the nebula (Figure 2.1b). Two
sets of spectra were taken at orthogonal angles. The slit position angles PA = 13 deg and
PA = 103 deg were chosen to maximize the number of key ionization fronts that could be
covered with the slit either crossing the ionization fronts at an approximately
perpendicular angle or running directly along them. The Blanco telescope uses an
equatorial mount, which results in a constant position angle of the slit once the
instrument has been rotated to the correct PA. We minimized the uncertainty in the PA
by rotating the slit only once each night, after all observations at the initial PA were
completed. As a result we have a high degree of confidence in our stated position angles,
and therefore of the mapping onto the sky of individual points along the slit.
Table 2.2 lists for each slit position the identifying position number, the RA and Dec of
the slit center in J 2000 coordinates, the PA, and the total exposure time. For simplicity,
slit positions with PA = 13 deg are numbered beginning at 1 at the easternmost position
and incrementing up through 17 increasing toward the west, including the two most
extreme north and south slit positions (position numbers 16 and 17, respectively).
Numbering of slit positions with PA =103 deg then begins at 20 and increases towards
the south, with even numbered slit positions centered to the east of R136 and odd

numbered ones to the west. The only exception is position 31 which has a PA equal to 98

72

degrees. Positions 40-44 were taken with the SOAR Telescope and are described in the
following sub-section.
The Blanco Telescope observations were taken under photometric conditions with a slit
width of 3.5 arcsec. Accurate offsetting was achieved by moving the autoguider probe by
known amounts, and fiducial stars of known coordinates were placed in the slit at most of
the slit positions in order to internally calibrate the position on the sky. An autoguider
maintained the position to better than 1 arcsec.
The standard IRAF 'LONGSLIT' package was used to generate bias and flat-field
corrected, flux calibrated long-slit spectra, rectified to remove optical distortions and to
convert to a linear wavelength scale. The wavelength solution had an RMS uncertainty of
0.06 A. After performing the distortion correction using the wavelength calibration, any
additional wavelength shifts due to flexure within the spectrograph were corrected by
shifting each spectrum so that the [O I] 765577.33 A night sky line fell at the correct
wavelength. The spectrophotometric standard stars LTT 2415, Hiltner 600 and CD32
from Hamuy et al. (1994) were observed on each night, through a 7 arcsec slit, to obtain
an absolute flux calibration.

2.3.2 Additional SOAR Telescope Spectroscopy
Six additional locations in 30 DOT were observed with the Goodman High-Throughput
Spectrograph on the SOAR telescope, on 5 Feb 2009. The instrument was used with a

single slit 3.9 arcmin long and 0.46 arcsec wide, with its 3001ine/mm grating. The

wavelength range was 3950 A to 9335 A with 1.32 A pixel'1 sampling and a spectral

resolution of 4.9 A FWHM at Her . This setup easily resolved the [3 II] doublet near

73

6720 A and the [0 II] lines at 7617320, 7330 A.

Calibration of the data was performed with the spectroscopic packages in IRAF in a
fashion similar to the method described above for the Blanco data, with a few notable
exceptions. The spectrophotometric standard stars LTT 2415 and L'I'I‘ 4816 were
observed, without a slit, to obtain an absolute flux calibration. There is second order
contamination in the spectra beyond 7700A, with the sensitivity in second order below
3850 A being about 36 per cent of that in first order at the same wavelength. This is not a
major problem for measurements of the the nebular emission lines in the red (observed in
first order), because at the positions of these lines the contaminating second-order light
is mostly continuum emission. But the second-order contamination does strongly affect
the flux calibration. This was accounted for by observing the standard stars through two
different second-order blocking filters, GG385 and GG495, which cut off light blueward
of 3850 and 4950 A, respectively. The GG385 filter was used for the nebular
observations and for the flux calibration for wavelengths lower than 7000 A, while the
GG495 filter was used only for the flux calibration beyond 7000 A. To stitch together a
flux calibration combined over the full wavelength range, a sensitivity response curve
was generated separately for each filter, using the IRAF 2D spectroscopic reduction
package. A final response curve was made using the GG 385 curve below 7000 A and
the GG495 curve at longer wavelengths.

Fringing with the current CCD is severe, reaching 32 percent at 9000 A. To calibrate this,
an internal quartz lamp was observed immediately after the series of nebular exposures at
each slit position, before moving the telescope. In spite of this precaution, flexure within

the instrument still caused significant, obvious offsets between the fringes in the nebular

74

data and in the quartz calibration frame. We used the fringe pattern seen in the spectra of
fiducial stars that fell in the slit as a guide for shifting and scaling in amplitude the quartz
fringe frames before dividing out the fringe pattern. This procedure decreased the
amplitude of the residual fringing to 10 percent.

The SOAR spectra sample an area of very low [S 111/Ha intensity ratio, lying to the east

of R136. Figure 2.2 shows the SOAR slit positions superimposed on a map of the [S II]/

Ha intensity ratio made from our SOAR images. The orientation and spatial coverage of
Figure 2.2 is the same as that of Figure 2.1. At the time of the observations the SOAR
Telescope lacked an atmospheric dispersion corrector, which is important with our broad
wavelength coverage and narrow slit. As a result our observations were restricted to
nearly the parallactic angle, resulting in the pattern shown in Figure 2.2. The positions of
the Goodman spectra are numbered beginning at 40, to minimize confusion between the
two spectroscopic data sets.

To remove night sky contamination we obtained a spectrum of the sky 2 deg north and
2.687 deg east of R136. The sky spectrum was smoothed with a median filter spanning
15 arcsec along the slit, to decrease noise and remove stars along the slit. The intensity in
the night sky lines was then scaled to match the object frames and subtracted. The

redshift of the LMC helps to separate night sky forbidden lines from the same lines from

30 Bar. For example, in the case of the [O I] 7&6300 line, the nebular emission line is
redshifted to 6306 A. Figure 2.3 shows a typical spectrum before and after night sky

subtraction, in the regions around 6300 A and 7330 A.
The instrument setup used includes the low ionization species [0 I] 16300, [0 II]
M7320, 7330 as well as the highly ionized [S 111] 19068 line and offers a clean

75

unblended measurement of [S 111] 16312 redshifted to 6318A. All of these lines except

for [S 111] 19069 are also within the wavelength range of the Blanco spectra but in that
case are heavily blended with night sky lines and/or nebular emission and so were not
measured. The region east of R136 was re-observed with SOAR because of the
additional constraints which these lines provide when determining the ionization
parameter U, the energy distribution of the ionizing radiation, and the O/H and S/H
abundances.
3 The spectroscopic 'data cube'

3.1 Emission Line Flux Measurements
Table 2.3 lists the emission lines measured from the Blanco and SOAR spectra. Columns
1 through 3 indicate the observed and rest wavelengths and ionic species. The fourth
column lists the extinction coefficient used to deredden the observations, using a standard
Galactic extinction curve with a ratio of total to selective extinction R=3.1 (Cardelli et al.
1989). The final column indicates which data sets include the emission line.
We wrote a FORTRAN program which extracted 1D spectra binned over successive 2.5
arcsec increments along the slit in the 2D spectra. The 2.5 arcsec width of the extraction
window along the slit was chosen so that the in the fainter regions the line flux equaled
the noise in the continuum. The program then automatically measured, in each extracted
1D spectrum, the emission line fluxes and their uncertainty. This produced a 'data cube'
of emission-line intensity measurements at 4238 points on the 30 Dor nebula.
We needed to use a summation approach to measure the flux in each emission line profile
because of the often complicated shape of the profiles. For the Blanco spectra, a slit

width of 3.5 arcsec was chosen to maximize the signal of the faintest lines while still

76

separating the [S 11] doublet. However, with such a wide slit, spatial variability of the
surface brightness of the nebula is imprinted onto the line profile in the spectrum, as is

seen in Figure 2.4. Figure 2.4a shows an enlarged portion of the SOAR [0 III] image at

the position of a ring-like structure, and Figure 2.4b shows the [0 III] 14959 line in the
2D image of the Blanco spectrum crossing this same location. The black dashed lines on
Figure 2.48 indicate the region over which the spectrum in Figure 2.4b was extracted.
The resulting profiles in Figure 2.4b clearly reflect the spatial structure rather than the
velocity structure of the gas. This resulted in often-ragged line profiles which were better
measured by summing the total flux above a fitted continuum rather than by fitting a
Gaussian. Figure 2.4c shows the extracted line profile at the location marked in Figure
2.4, with a best-fitting Gaussian shape superimposed to demonstrate that line shape is not
well represented by a simple Gaussian.

The wavelength windows used in the automated line-measuring procedure were
manually set in advance, using the brightest regions of the nebula. Two examples of

wavelength windows are presented in Figure 2.5. In the case of an isolated line like He I

16678, two 'wing' boundaries are chosen to include all of the line flux shown by the
vertical dashed lines. The continuum regions (indicated by horizontal dashed lines) are
separately defined on either side of the emission line, avoiding other emission lines. In
the case of mildly blended lines like the [S II] 116716, 6731 doublet, the wings are
defined to include the total flux from both lines. A search for the minimum flux over the
width of the solid horizontal bar shown in Figure 2.5 is is used to define the point of

separation between the extraction windows of the two lines. This does not perfectly

deblend the two lines, but since the ratio of [S 11] 16716/ 16731 is always of order unity

77

the error is negligible. This procedure for measuring the line strengths works well even in
the cases where stars are present in the extraction region so long as the star is not
extremely bright or the nebula extremely faint. Possible problem cases were identified
using the continuum bins, and were checked manually and for a few cases all emission

lines at that position were left out of analysis. In the case of bright stars with strong

absorption features the surface brightness of HB will be underestimated. Star clusters
were intentionally avoided and the number of regions affected by stellar absorption are
small.

3.2 Noise Estimates
We estimated the uncertainties in the measured line strengths by assuming Poisson
statistics in the original photon count rate and then carrying parallel noise images all the
way through the same reduction process as the data images. At the beginning of the data
reduction the measured counts were converted to photons using the known detector gain.

An initial noise image was then created according to

+gain‘XReadNoise2 )“2 (4)

0' = ( N electrons

photons

All multiplicative factors to the data such as flat fields, illumination corrections and the

flux calibration were applied to the the noise image by multiplying the values in the
noise image. In the step where the spectra were linearized and distortion was corrected,
there was a re-binning of the data. In this case the variance of the noise was formed and
rebinned in the same way so that the noise in each pixel was summed in quadrature. We
manually inspected a random set of extracted spectra and verified that these estimated

errors do in fact reasonably measure the observed pixel-to-pixel scatter in the continuum

78

points. An exception to this is [O III] 14363, where uncertainties in the continuum level

and blending with the wings of H I 14340 cause special problems, as is discussed below.
3.3 Detector Saturation
The large range in the surface brightness of the nebula meant that the 5 arcmin long slit

included both bright and faint regions. In order to get a sufficiently high signal/noise ratio

in the faint regions, long exposures were used and the [0 III] 15007 and/or Ha lines were
often saturated at various places along the slit. Where necessary additional short
exposures were made to correct for saturation. All repeated observations of a single

position, long and short, were gray-shifted to the highest value according to the total flux

of the [0 III] 14959 line summed over the full 300 arcsec length of the slit. The 14959
line was used because it is a moderately strong line but is never saturated. Summing
along the slit minimizes any effects of variable seeing or telescope drift. As a final check
to the data, line ratios of extracted regions were compared wherever two slits intersected.
While the extracted regions at these points overlap, they do not sample completely

identical parts of the nebula. The normalized histogram shown in Figure 2.6
demonstrates the repeatability of [0 III] 15007/H6 and ([S 11] 16716+16731)/H0t
measured for these overlapping observations. The histogram shows that the data do
repeat to within the typical range of differences due to the mismatch between the areas
sampled on th sky by the two slits , with especially good repeatability in the [O III]/HB
ratio.

3.4 Reddening Correction

The reddening was determined separately for each extracted 1D spectrum, using the

79

observed Her/H8 intensity ratio and assuming an intrinsic ratio of 2.87 appropriate for

Case B and a gas temperature 104K (Osterbrock and Ferland 2006) and R=3.1. This is
adequate even though 30 Dor is in the LMC because (1) although the LMC extinction
curve departs considerably from the Galactic curve in the ultraviolet, the two are very
similar in the optical passband, and (2) some considerable part of the reddening is in any
case due to foreground material within our own Galaxy. We then applied the reddening

to each measured line from the same extracted 1D spectrum, using the wavelength-

dependent extinction coefficients f7, from Table 2.3. A check on the validity of this

procedure is that the dereddened Hy/HB ratios agree with the predicted Case B values to

within an average of 3 percent, with a 10' scatter of 4 percent which is consistent with the
observational errors.
3.5 Electron Density and Temperature

At each position along each spectrograph slit we determined the electron density n. from
the dereddened [S II] 16716/16731 line ratio. The IRAF/STSDAS task nebular was used

to derive the electron densities using a 5 level sulpher atom assuming a gas temperature

of 104K. The errors in the density measurements were calculated using the 1 sigma

uncertainty in the line ratio. The difference between the densities at these extremes and at

the nominal value are the reported density uncertainties. If either the nominal or 10 [S 11]

line ratio was greater than 1.41 the density was assumed to be ne 5 10 cm'3 (the low-
density limit).

The electron gas temperature Te was also calculated at each position, from the

80

dereddened [O 111] (15007+14959) / 14363 ratio again using the IRAF/STSDAS nebular
task. The density used in the calculation is from the [S II] lines. Error bars were
computed in the same way as for the density values.

3.6 A Publicly Available Data Set

The Blanco instrument setup was chosen to cover the strong optical nebular emission

lines from 4100 to 7400 A including HB, H01 and the [S II] doublet at 116716, 6731. The
Blanco observations were made in the form of a grid with fairly regular spacing which
allow the data to be turned into a 'data cube'.

Our Blanco slit positions were chosen to produce a relatively unbiased sample of the
diffuse emission as well as to include discrete structures of interest. Post-observing
verification of the location of each slit position on the sky was done using narrow band

emission line and continuum images taken at the SOAR telescope by (1) comparing the

extracted surface brightness profile of H01 along the slit to the Hat images, (2) using
positions, measured from our narrow band continuum images, of stars purposely placed
in the slit as references and (3) where necessary using 2D nebular features visible in both
the spectra and images as seen in Figure 2.4.

Table 2.4 is a sample of the final data product from the Blanco spectra. It lists the
dereddened line strengths and several additional parameters for each extraction window
along each slit position, and their uncertainties. The whole table, which is available
electronically, includes measurements at 4238 positions with one row per position.
Column 1 lists the slit position number (as used in Figure 2.1 and Table 2.2). Columns 2-

4 contain the RA and Dec offsets in arcsec from R136 and the central pixel row of the

extraction window along the slit. Then columns 5-10 list the electron density and its 10

81

uncertainty limits, followed by the electron temperature with its 10' uncertainties.

Column 11 is the reddening Av deduced from the Hon/H6 ratio. These are followed in

columns 12 and 13 by the dereddened H6 surface brightness and its 1 uncertainty. The

remaining 44 columns list, in pairs, the dereddened surface brightness of each measured

emission line relative to H6 and the corresponding uncertainty.
The last three rows of Table 2.4, collectively labeled position 38, list the average
properties of 30 Doradus derived from the entire data set. Position 38 row 1 is the

average of the dereddened fluxes, and the physical properties derived from those. Row 2
is the average of the observed values. Row 2 was then dereddened using the average HB

and Her fluxes to produce row 3. Rows 2 and 3 represent the global spectroscopic
properties that would be measured for 30 Dor if it were spatially unresolved.

Table 2.5 is the similar data product from the SOAR spectra, computed in exactly the
same way as for Table 2.4. The SOAR spectra are presented in a separate table because

there are a different number of columns due to the additional emission lines which were
measured, including [S III] 1 9069. The ratio [S III] 19069 / 16312 is sensitive to the gas

temperature like ([0 III] 1 5007 + 14959 ) / 14363. The temperatures measured from the
[S III] lines are included in Table 2.5.
In addition to the tabulated data the fully reduced Blanco 2D spectra, noise and

calibration frames are publicly available. They currently are posted on a permanent web

site at the Michigan State University Department of Physics 8: Astronomys. The

complete version of Tables 2.4 and 2.5 will be available in electronic form at that site,

 

6http/www.pa.msu.edu/astro/thesis/pellegrini/3Odor/

82

and on the publisher's web site when this current paper is published.

4 Observational Results
4.1 Overview

Figures 2.7a-f are maps of selected emission lines, nebular diagnostics, and physical
conditions derived from those emission lines, interpolated in the spatial plane of the
Blanco data cube. The two extreme slit positions, 16 and 17, are excluded from these
maps. These maps are rotated 13 deg with respect to the N -S and E-W directions, and
are labeled with offsets in that rotated coordinate system. An outline of the area covered
by these maps is shown on Figure 2.1.

Figure 2.7a shows the dereddened Ha surface brightness and demonstrates the effective

resolution of the grid interpolated from the 'data cube'. It can be compared with the
SOAR Ho: image in Figure 2.1. Most features, including the bright arcs centered around
R136, are visible. To the east the cavity-like region of low surface brightness as well as
the bright rim on its eastern edge are clearly seen.

Figure 2.7b maps the reddening Av as measured using the observed Her/H8 intensity
ratio. As noted above, we used a simple Galactic reddening law to describe what is a

really a more complicated situation in which the line of sight to 30 Dor passes through

both the LMC and the Milky Way. With the exception of the region 80 arcsec south of

R136, the overall smoothness in AV over the nebula suggests there are no large scale

obstructions along the line of sight.

The ionization level of the gas is traced by intensity ratios which include ([0 III]
15007)/HB (Figure 2.7c) , ([N 11] 16584)/Ha (Figure 2.7d), ([8 III]16312)/([S II]16716

+ 16731) (Figure 2.7e), and [S II]/H0t (Figure 2.7f). Figure 2.7f, shows the ratio derived

83

from the Blanco spectra, as opposed to the ratio made from the SOAR images that is
shown in Figure 2.2. These line ratios all show a rough circular symmetry around a point
about 40 arcsec E of R136. The same distribution was identified by Indebetouw et al.
(2009; 109) using a Spitzer Space Telescope survey of 30 Dor using higher ionization
lines. This will be discussed further below.

Figure 2.7g plots the electron density measured from the [S II] 16717/16731 doublet
ratio. In this plot various structures seen in emission lines are also detected. The
brightness enhancement to the east of R136 ARA = +200 arcsec is visible 8 density
enhancement. The region around the central cluster has a density almost an order of

magnitude higher than the bulk of the nebula. Using the published contour plots of log
ne derived from [S III] in 109 we compared the electron densities measured from [S 11]
and [S 11]. We find they agree to 0.1 to 0.2 in log ne, noting that the increments in the

density contours of 109 are 0.1 in log ne.
4.2 The [0 III] Gas Temperature

The electron gas temperature is mapped in Figure 2.7b. It is computed from the [O 111]

(15007+14959) / 14363 ratio . We find the nebula to be fairly isothermal within a few

thousand degrees, in the range 9,000 S Te 5 12,000 K for the cases with error bars
smaller than 10 percent. The regions of hotter emission tend to be of lower density, as is
expected since the the cooling rate of a plasma is proportional to square of the density.
The accuracy of the overall scale for the gas temperature will be an important point in
our analysis. Here we compare our mean [0 III] temperature to that found by Krabbe &

Copetti (2002), who used three long-slit spectra which criss-crossed the central region of

84

30 Dor with much of their length falling on the bright arcs.

There are a number of valid ways to calculate an average temperature <Te> for 30 Dor.
For the most direct comparison to distant, unresolved GEHRs, we should spatially

integrate, over the whole nebula, all of the light in each emission line, then apply a single

reddening correction based on the ratio of the spatially integrated Hat and HB fluxes.
From those reddening-corrected line strengths we can then find <Te> using the observed

[0 III] (M959 + 15007) / M363 ratio R0111 given by

0 O
f H4959 A+5007A)d.(2 . (5)

 

Rom: O
jF(43634)da

This method uses the total line fluxes listed for Position 18, row 3 in Table 2.4, and

yields an equivalent temperature of <Te> = 10,680 K +/- 5.0 (Rom = 172).

An alternative is to use [0 111] line fluxes that have first been individually dereddened at
each point on the nebula and then spatially integrated. That method leads to <Te> =

10,760K +/- 8.0 ( Rom = 167.5). This corresponds to the values listed for Position 18,
row 1 in Table 2.4. While the 1% difference between the results of these two methods is

statistically significant, it is much smaller than typical observational errors for more
distant extragalactic H 11 regions with measured [0111] 14363.

When individual measurements of T., have been made at multiple locations in a nebula,
as is the case here, another common approach is to compute

jT(r )(XF (HB)d.Q

6
jF()H8dr2 U

 

<T,>=

85

where the Te is weighted by the HB flux at each position. Using this method we find <Te>

= 10,680K +/- 8.0, equivalent to Rom = 172.

For comparison, using this last method, Krabbe 8: Copetti (2002) found <Te> = 10,2 70K
+/- 46.7 (Rom = 194.5) using measurements along all three of their slit positions. An

even lower <Te> = 9990K was found by Tsamis (2003) who summed over the length of
a 160 arcsec slit.

To explore the source of the discrepancy between these values from other others and our

measurement of <Te>, we coadded all of our Blanco spectra to produce a single 1D
composite spectrum, excluding only the regions containing the brightest stars. The entire

spectrum was then dereddened as in the first of the averaging methods described above.

We then measured Rom and computed the associated <Te> from this high signal/noise

ratio spectrum in several ways. Using our automated line measuring software on this
spectrum produced <Te> = 10,565 K (Rom = 179), in reasonable agreement with the
value <Te> = 10,680 K obtained from the first of the averaging methods described
above. The automated software fit the continuum in windows covering the wavelength
ranges 114280-4325A (blueward of H7) and 114410-4150 (redward of [0 111] 14363). A
manual measurement of the [0 111] line strengths using the IRAF splot routine and
similar continuum levels produced essentially the same Rom and <Te>, verifying that
our automated routine works correctly. However, inspection of the coadded spectrum
showed that the region between H7 and [0 III] 14363 is partly filled in by a faint plateau

due to line wings or weak emission, and that the continuum also has complications in the

86

region around [0 111] 14959 and 15007. We tried various choices of how to draw in the

continuum and were able to measure line strengths corresponding to a temperature as low
as <Te> = 10,280K (Rom: 194). We surmise that the value <Te> = 10,270K found by

Krabbe 8: Copetti (2001) agrees with our corresponding value <Te> = 10,673K to within
these measurement uncertainties. Our automated process is an objective and repeatable
technique, and we conclude that it returns temperature measurements that are as accurate
as the data allow.

4.3 Ionization Mechanism
The observed gas temperature provides an important constraint to the energy sources
that ionize the gas. The violent history of 30 Doradus is evident in the diffuse X-ray

emission that is present throughout the nebula (Townsley et al. 2006), as well as in the

high velocity expansion features with speeds up to 200 km s'1 seen in Ha emission (Chu

& Kennicutt 1994). These show that supernovae and strong winds from massive O and

WR stars have combined to heat gas to the observed 3.5 — 7><106 K temperatures derived

from Chandra X-ray spectra. Despite these indications of high-velocity flows, the

moderate 104 K temperatures in the ionized gas do not indicate shock heating as the
current source of ionization. Additionally there is no detection of 0 IV or Ar V emission
lines (Lebouttler et al. 2008), nor are there any significant detections of nebular He II
emission which would indicate strong shocks being responsible for the ionization
structure in 30 Doradus.

Instead, photoionization is implied as the energy input mechanism. This is assumed in

our further analysis in this paper.

87

4.4 Temperature Fluctuations
Temperature fluctuations in nebulae are usually invoked to explain the observed
differences in chemical abundances determined empirically from collisionally excited
lines, and those measured using emission lines formed by recombination (Peimbert,
2003; P03) and also those found using the strong-line techniques discussed below. 30
Doradus is a nearby, resolved example in which temperature variations can be directly

measured, as a point of comparison for what may happen in distant unresolved GEHRs.

The 1D spatial profiles of Te and ne along each of our slit positions are shown in Figure

2.8. For clarity only every fourth error bar is shown. The electron temperature is

presented on a linear scale and the density measurements on a loglo scale. For slit

positions 1—17 positive displacement is toward the North, and for 18—37 it is to the East.

An artificial lower limit of log10(ne) = 1 was used as a lower limit to the density

measurements. It can be seen that in many places where the error bars are small there are

obvious 1000+ K fluctuations in Te over distances of tens of arcsec along the slit.

Krabbe & Copetti (2002) used their long-slit spectra to measure the [0 III] temperature

fluctuations t2, formally defined as

2 I(Te—(Te>)2ninedV
r = 2 v. (7)
(Te) IninedV

 

They reported t2 = 0.0025, using the approximation

1 (T-<T,>)2F(HB)dQ
(reflpmmdo

2<

 

(8)

with <Te> as defined in Eq. 7 and after correcting the observed value for their

88

observational uncertainties. Working backwards from the information in their paper, we

find (2 = 0.0035 before their correction for observational errors.

The directly measured value from the Blanco data set, using Eq. 8, is t2 S 0.024. After

accounting for our error bars, this becomes t2 5 0.014. This is considerably larger than
the Krabbe & Copetti result. The Krabbe 8: Copetti slit positions sampled a much
smaller area of 30 Dor than our data set and covered regions dominated by the bright
arcs, which we find to be fairly isothermal. The fluctuations in temperature become much
more significant when the distant lower density regions are included, as in our data set.

However, we also note that in these fainter regions we have likely underestimated the
error in our measurements because of the difficulty in measuring the [0 III] 14363 line.

The large t2 value that we measure here is very close to the value needed to bring
abundances measured from collisionally excited lines into agreement with those
measured from recombination lines. Unfortunately, we cannot be confident of this result
due to the possibility of systematic errors in the continuum placement that might change
with position along the slit.

4.5 Structural Details
The geometry and ionization structure of 30 Doris quite complex. The SOAR direct
images provide a powerful tool for distinguishing edge-on ionization fronts, regions
ionized by localized sources of radiation besides R136, edge-on IFs, optically thick

pillars like those found in M16 (Hester 1996), and line-emitting foreground structures. In

particular, the nearly reddening-free [S II]/H0t image (Figure 2.2), formed from the ratio

of the SOAR images in those two passbands, offers considerable insight into the

89

ionization structure on many scales.

4.5.1 Embedded ionizing stars
A key question is the degree to which R136 dominates the ionizing radiation field at
different positions in the nebula. This will depend on the extent to which additional,
isolated O stars provide local contributions to the radiation field. Such stars can be

recognized by localized, circular variations in the ionization level as indicated in the [S

II]/Hor maps, in combination with brightness enhancements in the Hat emission. We
carefully searched the SOAR images for such features, and located the 49 isolated
ionizing stars listed in Table 2.6. Some percentage of these are likely to have been
selected by chance, so this estimate should represent an upper limit to the number of
locally ionizing stars. The area of the nebula in which the ionization level is significantly
affected by the radiation field of these stars, as projected on the sky, covers only 2% of
the full nebula, so is unimportant in our analysis below of the overall properties of the
nebula.

4.5.2 Edge-on ionization fronts

Edge-on ionization fronts stand out in the [S II]/Hor image as narrow lines of high [S II]/

Hor ratio butted up against regions of low [S II]/Hor For an edge-on IF to be detected in
this way, the ionization parameter U must be high enough that the thickness of the

ionized gas is spatially resolvable, and the density must be high enough that the optically

thin Hor emission is distinguishable from background emission. Where this is true, the
Hon emission is separated from the [8 II] emission and the edge-on IF is detected as an

inversion in the [S II]/H0t ratio, with lower values found on the side toward the ionized

90

gas and higher values toward the IF.

Table 2.7 lists prominent, isolated, edge—on ionization fronts found in both our [S II]/Hoc

image and the Spitzer Bum PAH mosaic. These unobstructed IFs are perfectly suited for
detailed, high angular resolution (e.g. with HST and/or ALMA) studies of the neutral and
molecular gas beyond the IF, commonly called the Photon Dominated Region (PDR). A
finding chart with the all IFs listed in Table 2.7 is shown in Figure 2.9. Two examples of

such features, IF1 and IF2, are shown in an inset in the upper comer of Figure 2.9, which

is a blow-up of a portion of the [S II]/Hor image shown in Figure 2.2. Visible in both IF1

and IF 2 is the spatially resolved layer of H+.
IF 1 and IF2 are likely to be similar in geometry to the Orion Bar (see P09), and enough
information can be measured about them to provide a test of whether or not they are

likely to be photoionized by R136. The thickness dr of the ionized gas depends on the

density and distance r0 from the source of ionizing radiation according to

drocQO/(41'rornenp r3) . (9)

3 sis the H+ recombination rate.

where a: 2.59><10'13 cm'
There is a line of stars coincident with IF1 that could be an alternative to R136 as the
source of ionization, or which could be stars formed in a region of gas that was
compressed by radiation and wind pressure from R136, or which could simply be
foreground stars. Most of these stars fell in one or another of our slit positions. We have
examined their spectra and found them to lack any absorption features that would

indicate they are massive stars. With the exception of one star, they do not appear to be

particularly reddened, suggesting they are not embedded in or around the IF.

91

Additionally, we note that the overall structure of IF 1 is oriented almost exactly
perpendicular to the direction to R136. There are also many prominent pillars in this
region (see below), all pointing back up to R136, suggesting the nearby stars are
unimportant to the structure.

Our SOAR spectrographic slit position number 40 cuts directly across the large
ionization front IF 1, avoiding the pillars noted by Scowen (1998) and targeting the larger
structure. The emission line strengths in the region of IF 1 were measured as described
above, but were extracted in smaller (0.75 arcsec) windows along the slit to better resolve

the detailed structure. Figure 2.10 shows the profile of IF 1. In the top panel are the

relative intensities of Her, [0 111] 15007 and [S 111] 19069 which trace highly ionized gas.

The middle panel shows the same for [8 II], [0 II] and [N 11], all tracers of the IF. The

bottom panel shows the linear density profile of the structure in cm'3. At the illuminated

3

face the structure has an electron density ne S 100 cm" , increasing to 800 cm"3 at the

peak of the [S 11] emission, and then decreasing to a minimum of 200 cm'3.

There is a bright knot in the [5 II] images at the location of the peak density. Excluding

this point, the density at the IF is likely to be 560 cm'3. Using this density, and assuming
that R136 is the ionizing source, that the 13.6 pc (58 arcsec) projected distance from

R136 is the true three-dimensional distance and that the 1.18 pc projected thickness of

the H+ zone is the true thickness, Eq. 9 requires Q; = 5.9 X 1051 5'1. In addition to the
uncertainties in projected vs. true distances, this value does not include the effects of

absorption by He and neglects the effect the observed gradient will have in lowering the

measured average he, so the derived Q0 could be either too low or too high. However, it is

92

still interesting to compare this estimate based on our measurements of IF2, to Q0
estimates for other edge-on ionization fronts at varying distances from R136, as a test of
the hypothesis that R136 is the predominant source of ionizing radiation.

IF2 is the inner edge of one of the most prominent bright arcs, located at a projected

distance of 17 pc (70 arcsec) NE of R136 (see Figure 2.1) . The thickness of the H+

layer perpendicular to the direction to R136 is approximately 0.94 pc. Within the
observational error the density at the IF along a ray from R136 into IF1 is ne = 280 cm'3.

Applying Eq. 9 with the same assumption used with IF1, we find Q; = 2.0 X1051 5'1.

IF4 is a bright rim lying much further out in the nebula, 220 arcsec (53 pc) to the east of
R136 on the far side of the large faint low-density region that is easily seen in figures
2.7a-e. This large wall is remarkably homogeneous in density and has a thickness

between 3.5 and 4 arcsec all along its length of 140 arcsec (33 pc). The limb shows up as
a density enhancement in Figure 2.7c, with ne = 125 cm’3. It has a nearly constant

projected distance from R136. Using the previous technique we find 00 = 4.58 X 10515'1.

We will return to this particular IF in the Discussion section below.

The above estimates of Q0 for R136 are all in reasonable agreement with each other and

also with Q0 = 4.2x 10 51 s'1 estimated by Crowther and Dessart (1998) from adding up
the contributions expected from the individual stars in the cluster. These results support
the conclusion that R136 dominates the ionizing radiation field throughout most of the
volume of 30 Dor that is included in this present study.

However, there is at least one counter example, to the SW of R136 where the observed

structure indicates a different ionization source. IFs 1*-5* show PAH emission closer to

93

R136 than the [S 11] emission. Forming an elongated shell structure with a diameter of 15
pc around another star cluster, thes IFs and likely to be ionized by it.

4.5.3 Dense pillars
There are numerous bright fingers and protruding IFs scattered over the face of the
nebula, but particularly concentrated in the region SW of R136. Many of these are
probably elephant trunks similar to the famous “Pillars of Creation” in M16. A small
number of these have been commented on by Scowen et a1. (1998). 106 of the brightest
such features are cataloged here in Table 2.8 with RA and Dec positions, projected length
and PA, but there are hundreds more that blend together as one looks on finer and finer
scales, until they are indistinguishable from small examples of the edge-on ionization
fronts discussed above. Almost all of these point back to R136, again showing that the
central cluster is the dorrrinant source of ionizing radiation.

Many of these features are clearly connected to larger bodies of molecular gas, as can be

seen by comparing our [S II]/Hor ratio image to the Bum image from the Spitzer Space
Telescope archives, which shows PAH errrission (Figure 2.11). However, some of them
do appear to be isolated tubes or blobs of gas rather than protrusions from background

walls of molecular material. Scowen (1998) have commented on one bright structure of

this latter type. Some of these might correspond to the expanding shells seen in Ha
emission in the high-resolution echelle spectra of Chu & Kennicutt (1994).
5 Photoionization Simulations
5.1 Rationale and Purpose
As discussed above, the observed emission line intensity ratios and morphology indicate

photoionization from a central source as the dominate process energy source. The general

94

radial symmetry of these ratios around a point on the sky near R136 (Figs 2.7a-g)
strongly suggest that central cluster is the major source of this photoionization. The
center of these circular patterns is offset to the east of R136, but that can be immediately
understood as being the result of a blowout into the lower density gas toward the east (fig
2.7g), very similar to what is seen on a smaller scale in M17 (P08) The identification of
R136 as the source of ionization is supported by the lack of strong competition to the
stars within 15 pc of R136 from local ionizing stars (Sect. 4.5.1), by the thickness of
edge-on ionization fronts as a function of distance from R136 (Section 4.5.2), and by the
large numbers of elephant trunks pointing back towards R136 from points all over 30
DOT.

In order to estimate U at different locations within 30 DOT, we computed a grid of
photoionization simulations as a tool for using the observed emission line strengths to
determine U and other physical conditions in the gas on a point-by-point basis across the
nebula. The purposes of this grid of photoionization simulations are ( 1) to determine the
shape of the ionizing continuum radiation field assuming that most of the gas is
photoionized by the same source (R136); (2) to fix the global chemical abundances, as

averaged over the full nebula; and (3) to provide a look up table that can be used to

determine U and ne at each point in the extended nebula.

5.2 Basic Simulation Parameters
We used the plasma simulation code Cloudy (Ferland et al. 1998). All of the models

described below begin at the illuminated face of a plane parallel cloud, externally

illuminated by a source of ionizing radiation with an incident flux Qo/(41t r02).

For simplicity, we assumed a constant-density equation of state (EOS). The EOS may be

95

approximated by equation A2 in P09. The constant density assumption used here affects

the relative densities of the H”, H0 and H2 regions. When only considering the H+ region
the gas temperature is nearly constant and the differences between equations of state
become minimal. As a result, the relative strength of the emission lines formed in the H II
region are largely unaltered by the EOS.

All of our models assume the physical thickness of the gas cloud to be 10 pc (which

projects to 43 arcsec at the distance of 30 Dor). The physical thickness of the H+ zone
will change as described by Eq. 9 in response to changes in density and ionizing flux. In
cases where the ionized layer extends to a depth greater than 10 pc, the cloud was treated
as matter bounded (the 'optically thin' case). In the model fitting described below, this
happened only in a few regions close to R136. Elsewhere, the ionized layer is radiation
bounded ('optically thick'), in which case the models were stopped when their
temperature reached 500 K. At this temperature the ionization front has fully formed, but
the models have not gone into the PDR.

For the optically thick models, the computed emission line strengths correspond to what

would be seen if the ionized cloud were viewed face-on, so that the emission coming

from all depths into the H+ zone was added together. This would simulate either gas seen
on the ionized back surfaces of a large bubble (similar to our view of the main part of the

Orion Nebula, for instance), or what would be seen for a more edge-on IF with a

sufficiently small ionized thickness so that the full depth of the H+ zone fit into one of
our 1D spectral extraction windows. For the Blanco spectra the extraction window was

2.5><3.5 arcsec, corresponding to 0.6X0.8 pc. The ionized layer in gas with a typical

96

density 11;; = ne = 200 cm’3 and lying more than 50 pc (200 arcsec) from R136 would fit
within one of these extraction windows.
For the optically thin case the models do not reflect a particular geometry. Instead they

represent a fully ionized zone which is matter bounded. These are indicated by regions

with very weak emission from IF tracers like [N 11] 16584 and [S 11] (16716+16731)

relative to H recombination lines. For matter bounded nebulae the strengths of these IF

tracers are typically 1/ 10 Ha. Line ratios characterizing a fully ionized plasma with no IF
are an order of magnitude lower than this. Our models of these regions follow the
radiation transfer through a 10 pc thick region. However, these models assume
illumination by an unfiltered ionizing spectrum, which would not be realistic if the
observed gas is part of an ionized region larger than the area taken in by the slit. To deal

with this additional complication we would need prior knowledge about the geometry to

reconstruct the entire H+zone, which is what we are trying to work out. Despite the lack
of total realism, the optically thin models are still useful since the line ratios are most
sensitive to ionization parameter and not thickness.

We estimated the numbers by spectral type of massive O and WR stars in the central
cluster by using SIMBAD to find all stars within 45 arcsec of R136 with spectral types

(Table 2.9). Using the conversions from spectral type to ionizing flux and effective

temperature Tefl' given by Vacca et al. (1996) for the O-stars and Crowther (2007) for the

WR stars, the corresponding luminosity of ionizing photons generated by the central

cluster is Q0 = 7X10 51 5'1 . This is slightly above the value 4.2)( 10 51 s'1 estimated by

Crowther and Dessart ( 1998) for all of R136. Using the stellar atmosphere parameters

97

given by Heap, Lanz & Hubeny (2006) would reduce our computed Q0 by about a factor

of two. Fixing 00 means that in our models the incident ionizing flux was determined by
r, the distance from R136.

The global X-ray emission in 30 Doris well studied (T ownsley et al. 2006) and found to
show considerable surface brightness structure and to have plasma temperatures between
3 — 9 million deg. However, the X-rays are unlikely to significantly affect the ionized

gas, since the X-ray luminosity is three orders of magnitude lower than the UV ionizing
luminosity Q0 from the stars. We have compared models with and without likely X—ray

fluxes and conclude that for models of the H+ region the X-rays can be ignored.

The models include silicate and graphite dust using the average LMC size distribution
from Weingartner and Draine (2001). We have assumed a quite small dust/gas ratio. To
first order the gas to dust ratio is expected to scale with metallicity, where the gas to dust

ratio in the LMC is equal to the gas to dust ratio near the Sun times the ratio of
metallicity of the LMC to the Sun, Z/Zo. We have adopted a lower gas to dust ratio is

equal to 0.1 Z/Za. This choice is motivated by the lack of correlation between FIR
emission and Av (Chu et al 1985). The hypothesis that the extinction is caused by

foreground dust is further supported by the similarity of Av shown in Figure 2.8b to that
measured along the line of sight to the stars. Recently a study of the dust to gas ratio in
the LMC was carried out in the IR, using heavily extinguished lines of sight though

molecular clouds (Dobashi et al. 2008). IR colors of stars behind molecular clouds were

used to measure Av. Av was compared to the hydrogen column density N(I-I) of the

98

molecular clouds, where N(H) is the sum of neutral H traced by H I 21-cm maps (Kim et
al. 2003) and H2 traced by 12CO maps using a common conversion factor between N(

12GO ) and N( H2). Using the ratio of Av /N(H), the relative amount of dust in the LMC
was found to vary between 8.5 and 2 times smaller than the solar value. There is a trend
toward 30 Dor for the ratio to increase, but this may be the result of an undetected gas
component in that direction. As a result the gas to dust ratio we have assumed for our HII
region is about half the lower value found in the molecular clouds in the LMC. This is

not to say that there is no significant amount of dust outside the highly ionized region. In
the Spitzer images the correlation between the more easily destroyed PAHs and the Hat

emission is very strong, but separated spatially as in more quiescent H II regions. Since

PAHs are more fragile than dust, the dust must also be present in significant quantities

outside the H II region. We only suggest a lower dust abundance in the H“ zone itself due
to the extreme nature of the cluster.
We have assumed a constant turbulent velocity of 5km s'1 .This does not affect the
pressure and is included only for a realistic treatment of the radiation transfer.

5.3 Initial Chemical Abundance Set
The chemical abundances in 30 Dor have been studied extensively. Historically this has
been done using optical spectroscopy, which has access to a great number of elements
and is easily obtained from the ground. Empirical total gas phase abundances are

determined by using either a recombination or collisionally excited emission line relative

to a hydrogen recombination line, often using the (14959+ 15007)) / ([0 111] 14363) ratio

to estimate the gas temperature. The relative intensities of emission from a given ion

99

provide an accurate measure of the relative ionic abundances. The total abundance is
measured by observing emission from all the dominant ionization states or by making an
ionization completeness correction to account for ionization states not observed. Some
key studies of this type are by Vermeij eta12002 (V02), Mathis, Chu 8: Peimbert 1985;
Rosa & Mathis 1987; Tsarrris et a1. 2003; and P03).

A more complicated, model-dependent alternative is to use photoionization codes to
account for emission from all ionization states simultaneously. There is significant
difficulty in breaking the degeneracy of models with different U, abundances and SED.
Studies of this type have been carried out by Tsamis & Pequignot (TP05) and
Lebouteiller et al. 2008.

As a starting point for our simulations, we drew on the results of TP05. They assembled a
set of observed emission line strengths by combining published results from UV (V02),
optical (P03) and IR Spitzer (V02) spectroscopy, and then used exceptionally detailed
photoionization simulations to determine the total gas-phase abundances of 30 Dor. They
concluded that at the spot in the nebula they were studying, a multiple component model
is required, in which two gas components with different chemical abundances, densities
and temperatures are in pressure equilibrium. One component represented low-
temperature high-density filaments that are H and He poor, consistent with wind-blown
material from pre-SN winds. We adopt here the abundances from their other component,
representing the homogeneous surrounding material, because the filaments would
contribute less than 10 percent of the total emission in most of the emission lines that we
are studying. The abundances for the homogeneous component of TPOS, their model D2,

are listed in Table 2.10, along with other empirical and model-based abundances. The

100

abundances quoted from P03 are those derived assuming a mean square temperature

fluctuation t2 =0.003. We take the scatter between these methods to represent the range of
uncertainty in the abundances in 30 Dor. The last line of Table 2.10 lists the abundances
which we determine here, as described below.
Our initial set of simulations used CLOUDY with the same abundances and the same
COSTAR (Schaerer et al. 1996) stellar atmosphere (an atmosphere with half-solar metal
abundances and an effective temperature of 38,000 K) that was used by TP05 in their
model D2, to compute a grid of models in which we varied the incident ionizing flux and
the gas density. We then iterated to a final choice of ionizing continuum shape and
cherrrical abundances as described below.

5.4 The Ionizing Continuum Shape
We first explored the degree to which the ionizing continuum shape is constrained by the
observed emission line intensity ratios and by their variation across the nebula.

Figures 2.12a-d show a set of diagnostic diagrams which compare observed emission line

intensity ratios to the predicted ratios using COSTAR models of different Tefi. The
assumption that the gas throughout 30 Doris to first order all illuminated by the same
continuum source gives these diagrams great leverage in identifying the shape of the

ionizing continuum. In these figures, predictions and observations are shown only for

cases with 100 s n. s 200 cm'3. In Figures 2.12a, 2.12c and 2.12d, only SOAR

observations could be used because not all of the necessary emission lines could be

measured in the Blanco data set. The solid lines show results for CLOUDY grids with

the density was fixed at 102 cm'3, and with an ionizing flux corresponding to distances

101

from the central cluster between 13 s r S 140 pc. Multiple SEDs are a shown using the

COSTAR continuum shapes ranging from Tefi = 37,000 K to 41,000 K in 1,000 K steps,
increasing in temperature in the direction of the arrow shown on each panel.

Figure 2.12a (top left) shows ([S 11] 16716+16731)/([S III] 1.9069) vs ([0 II]

17320+17330)/([O III] 15007), which is very similar to the radiation softness
parameter originally defined by Vilchez & Pagel (1998). It is also nearly equivalent to
the ratio of $23/R23 where 823 and R23 are two widely used ratios to indicate
abundances of O and S. R23 was originally purposed by Pagel et al. (1997) and is often
used in strong line abundance measurements because it includes the strongest lines from
the most abundance phases of O. This diagnostic is strongly dependent on the shape of
the spectral energy distribution (SED) and on U, and is only weakly dependent on
abundance and density. As the effective temperature of the modeled SED increases, the
emission from the low ionization species relative to those with higher ionization potential
decreases. It was demonstrated by Oey et al. (2000) that these line ratios are sensitive to
effective stellar temperatures less than 40,000 K. As a result, the observational scatter in
Figure 2.98 may be the result of observational error such as imperfect night sky
subtraction of the [O 11] lines or improperly corrected fringing in the [S 111] line, or it
may indicate real variability in the shape of the ionizing radiation reaching some points.
Figure 2.12b shows [S II]/Hor vs [0 III]/HB and is often used to study large samples of
different H 11 regions or emission-line galaxies. It was first proposed by Veilleux 8:
Osterbrock ( 1987) as an additional way of separating H 11 regions from active galactic
nuclei using emission lines, to supplement BPT diagrams (Baldwin et al. 1981). This

diagnostic depends on the O/S abundance ratio as well as on the gas temperature and U.

102

Both the SOAR and Blanco data sets are used on this diagram.

Figure 2.12c is a diagnostic that includes the [O I] 16300 line that comes from deep
within the H*-H° boundary zone and hence is a stringent test on how well our constant
density models are fitting at the point of transition into the Partial Dissociation Region
(PDR). Figure 2.12d is similar to Figure 2.12b except that it uses only O/H line ratios,
thus removing the uncertainties concerning the O/S abundance ratio.

From the four diagnostic diagrams in Figure 2.12, we found that the best-fitting models
using COSTAR atmospheres lie between the Te” = 38,000 and 39,000 K curves, so we

ran a final grid using Te” = 38,500 K. The results for that grid, which is our best-fitting
case using the TP05 abundances, are shown on Figure 2.12 as a heavy dashed line. The
fit to the ridge line of the observations is good except that the [O II]/[O 1] ratio is under-
predicted, meaning that [O I] is too strong in the models since other line ratios involving

[O 11] give good fits. It is important to note that the [O 11] line strength used on these

diagrams is 17320+17330, not the much more commonly-used 13726+13729.

We also tried fitting a grid of WMBASIC models (Pauldrach 2001) of different
temperatures. These stellar atmosphere treat the radiation transfer including line
absorption by metals, and as a result produce a significantly softer continuum shape.
They gave poor fits on to the [S III]/ [ 511] data from the SOAR observations and were
dropped from further consideration. As a further possible continuum shape, we
constructed a composite continuum made by adding together WMBASIC models at
different temperatures weighted by an the number of stars of each spectral type as listed

in Table 2.9.

103

5.5 Revised Chemical Abundances

The above exercise led us to adopt a COSTAR Teff = 38,500 K model as a reasonable
approximation to the ionizing continuum shape produced by R136. This model gave an
acceptable fit to the line ratios shown on Figure 2.12.

However, the resulting CLOUDY models significantly under-predict the observed

electron temperatures Te. This can be seen by comparing the predicted [0 III] 14363/

15007 ratio to the observations, as is shown in Figure 2.13. TP05 computed the average
model temperature weighted by the H“ density to be <T(n(H+))> = 9895 K. Using the
same initial conditions our typical model produces <T(n(H+))> = 9310 K with the same

weighting. Both of these temperatures are significantly lower than our measured <Te> =

10,270-10,760 K weighted by F(HB) as in Eq. 6.

Since the temperature of an H 11 region is largely regulated by cooling through forbidden
metal lines, and a harder SED is already ruled out, we conclude that the overall metal
abundances found by TP05 are too high. Oxygen lines account for about 25% of the
cooling in the various models described here, so the gas temperature is essentially set by
the O/H abundance ratio. To arrive at a final set of models we first decreased the O/H
abundance until the model temperatures increased to the observed value, leaving the
other abundances unchanged (i.e. with the values used in TP05 model D02). The
resulting O abundance is log(N(O)/N(H)) = -3.75, as shown in the bottom row of Table
2.10.

Since N, S and Ar do not dominate the cooling, their emission varies almost linearly with

abundance. The abundances of these elements were adjusted so the models again

104

matched the emission line diagnostics used in Fig 2.12. The resulting adopted
abundances for these elements are also listed in Table 2.10.
This model-based method of determining the global abundances in 30 Dor should be

reasonably accurate for elements for which the atomic parameters are well known (i.e.

elements from the 2nd row of the periodic table, including 0 and N from Table 2.10) and
also for elements for which lines from more than one important ionization state are

observed (0 and S from Table 2.10). However, our abundance estimate for Ar is based

on only one pair of emission lines from a single ionization state of this 3rd row element,
and therefore is quite uncertain.

Finally, we also adjusted the He abundance to match the observed He I emission line
strengths. The He I lines are due to recombination of He+ which has well-known atomic
data. The observations provide tight constraints on N(He+)/ N(H+). This should be nearly
identical to N(He)/N(H). Our new observations do not detect any significant He"+ 14686
nebular emission. This rules out a significant fraction of He”. There exist many
arguments (Pagel 1992, AGN3) that there should also be no He0 coexisting with H". This
3

is confirmed in all of our models, where a typical place in the nebula with n" = 102 cm"

at a distance of 50 pc has an Heo/He+ fraction of 0.009.
A comparison of the observations to the predicted intensity ratios from this final set of

models, with the revised abundances, is shown in Figures 2.14-2.17. Figure 2.14 shows

the much improved predictions of [0 III] 14363/15007 relative to observed values.

Figure 2.1Sa-d is a repeat of Figure 2.1Za-d, showing that the models fit just as well with

105

the revised abundances as with the TP05 abundances. Figure 2.16 shows the fit to He I

16678/H0t, and Figure 2.17 shows the commonly used [N II]/Hor vs. [0 III]/HB diagram.
Agreement between the predictions and the observations is now good in all cases.
The final adopted abundances logro(N(X)/N(H)) are (HzHezCzN :O:Ne:Si:S:Cl:Ar:F e) =
(Oz-1.08:-4.3:-4.91:-3.75:-4.36:-5.51:-5.32:-7.16:-5.99:-5.95). Here the abundances of all
elements for which we did not measure line strengths were left unchanged from the
values used in TP05 model D02. An alternative strategy might have been to alter the
abundances of the unobserved elements in lockstep with the O/H abundance ratio. We
verified that making that change did not alter any of the predicted intensity ratios shown
in Figures 214- 2.17, or the values of U found in the following section, by more than
1%.

5.6 The Physical Parameters at Each Point in the Nebula

With the SED and chemical abundances now fixed, we fit the observations to a full grid

of 987 CLOUDY models with varying distance r0 from R136 and density n”. The [S 11]

16716/16731 ratio, SZr, is primarily dependent on the gas density and offers a strong
observational constraint in this procedure. We began with a grid spacing in density such
that SZr varied by 1%. For each point in the nebula SZr was used to eliminate any model
where

szrobs- SZr > 3 0cm - (10)

model
This sub-grid of n, r0 is like the original, but with fewer possible It”. For each m; in the

SZr selected sub-sample, we compared the models with different r0 by a convergence

criteria X2 defined as

106

2
R Rimodel ’ (11)

l 0'0b5min { Riobs’ Rimodel}

iobs

 

X2(r0,nH)=Z

where R.- obs and R.- model is the observed and modeled dereddened ratio of emission lines

relative to H6. The emission lines with an asterisk in Table 2.3, as well as $2r, were used
in the fitting. They were chosen because of their brightness, dependence on ionization

parameter and the absence of contamination by night sky lines.

The Ro(n) with the lowest X was then used as the starting point for a search between

neighboring models. Using linear interpolation along r we found the n, which minimized

f for each n. Finally, the n, ro(n) pair with the lowest 12 was selected as the best model
for that particular data point. Over almost all of the nebula good fits were achieved w/
optically thick models. However, we do not claim that we have accurately described the
relatively small region fitted by our optically-thin models — there is obviously some
radiation transfer left out of that part of the fitting and there is a bright, obviously
embedded star in this region which is very prominent at infrared wavelengths, which may
be locally ionizing this relatively small part of the larger nebula.
The final result of this procedure was a grid of values of the fitted ionization parameter U
and electron density n, measured at every point along our slit positions. The resulting
map of U over the face of 30 Doris shown in Figure 2.18.
6 Discussion

6.1 Photoionization by R136 vs. Other Sources of Excitation
We have fitted photoionization models to the measured emission-line intensity ratios at a

large number of points across the surface of 30 Dor. This produces a point-by-point

107

estimate of the ionization parameter all across the nebula. This is very similar in outline
to the recent study of a somewhat smaller region of 30 Dor by Indebetouw et al. (2009;
109), who used the strengths of emission lines measured with the Spitzer Space
Telescope. A goal of I09 was to use variations in the shape of the SED photoionizing
each part of the nebula to disentangle the relative contribution from local ionizing
sources compared to the central cluster. The SOAR spectroscopic data used here allows
us to estimate the effective stellar temperature using [0 II]/[O III] and [S II]/[S III] ratios
for a selected portions of the nebula, similar to way in which 109 used [Ne IV] /[Ne III]
and [Ar III]/[Ar II].

109 analyzed their infrared line intensity ratios using “one zone” Cloudy models, which
we take to mean that they stopped the Cloudy integration through the ionized layer after
the first zone, so did not include the effects of radiative transfer inside the gas cloud and
instead fit all line ratios as if the gas were fully ionized. Such models cannot be expected
to correctly reproduce the strengths of the low-ionization lines that are included in our
observations. They allowed the SED in their one zone models to vary on a point-by-point
basis, which presumably represents some combination of true variations in the intrinsic
shape of the ionizing continuum, correction for the missing radiation transfer calculation
within each gas cloud, and noise.

Here we have used more complete Cloudy models which properly predict lines from

lower ionization species like [N 11], [S 11] and [0 II] along with H+, [0 III] and [Ar 111]
so that the entire H II region is used as a constraint. We have also assumed that all gas
clouds in 30 Dor see the same ionizing continuum source, and have neglected any

filtering of that radiation by any intervening gas within the 30 Dor nebula. We deduce

108

from the derived parameters that this is a reasonable approximation in most parts of the
nebula. A quantitative comparison of our results to the ionization parameters found by
109 is not possible because their published results are only descriptive. Qualitatively, the
the ionization parameter maps determined here from our optical data are in reasonable
agreement with the results 109 derived using their IR data.
The dominance of the X—ray emitting gas in the equation of state is a validation of our
constant density/models, which show the integrated radiation pressure to be negligible on
global scales. This is in contrast to the results for the smaller H II regions M17 and the
Orion bar which we have previously studied in detail (P08, P09). While the radiation
pressure from the stars is negligible over much of 30 Dor, we find excellent agreement
between photoionization models and our observations for the entire region. Within the
constraints of our data we find the emission line spectrum is set by photoionization from
the central cluster. We have identified regions that are suitable for follow up that may be
locally ionized, but they are quite limited in number. The additional constraints provided
by further optical [S III] and [0 II] observations should in the future allow the nature of
those regions to be firmly established. By considering the entire system, we conclude that
the approximation of a single ionizing source is a sufficient description for the
complicated 30 DOT region, and hence of similar regions that may be unresolved in
distance galaxies.

6.2 The Current Geometry of 30 Dor
The projected structure of 30 DOT as seen on an Hor image such as Figure 2.1 is
dominated first of all by the very bright arcs near R136, and after that by the extensive

regions of low surface brightness which have the appearance of large cavities and are

109

often referred to by terms such as “merged supernova remnants”.
Our Cloudy model fitting shows that all of the regions of higher Ha surface brightness

are optically thick to ionizing radiation; this is determined most clearly by the observed

[S II]/Hor intensity ratio but all of the other measured lines bolster the result. Spitzer
images (Meixner 2006) show that there are extensive regions of PAH emission associated
with most of these same regions, verifying that a transition zone into molecular gas is
associated with them.

CO maps (Poglitsch 1995) show that the main bulk of the molecular gas falls in a broad
N-S swathe that includes the region of the bright arcs. Direct near-infrared narrowband

images (Rubio & Probst, unpublished) and also long-slit infrared K-band spectra that we

have obtained with the SOAR Telescope show emission in the Hz 2.12 |.1m line coming
from extensive regions on the opposite sides of the bright arcs from R136. This all
combines together to very clearly show that the bright arcs are edge-on walls of
molecular clouds whose faces are being photoionized by R136.

In Section 4.5.2 we have cataloged a number of other examples of similar walls, but at
greater distances, which also are photoionized primarily by R136.

The remainder of the surface area on 30 Dor that our model-fitting identified as optically
thick gas is a mix of smaller, often blended edge-on ionization fronts (Figure 2.2), along
with broad surfaces of optically-thick gas seen more nearly face-on. This latter gas must
lie further away from us than R136 if it is indeed photoionized by direct radiation from
R136.

Turning to the “cavities”, Townsley et al. (2006) showed from Chandra Space Telescope

maps that there is a correlation between X-ray emission and these regions of low Ha

110

surface brightness, and argued that the correlation between these structures implies that
the cavities are supported by the pressure of the hot X-ray gas. We explored the

possibility that the optical emission lines in these directions were coming from optically

thin inclusions of warm (104 K) gas somehow surviving within these cavities. However,

we find that the optical emission from only a fairly small portion of only one of these

regions is actually fitted by optically thin models. Over most of the regions of low Ha
surface brightness the optical emission lines are from optically thick gas, which we
interpret as coming from an optically thick back wall.

This all shows that, as has long been realized, the optical emission from 30 Bar traces a
very layered, three-dimensional structure. We know the projection on the sky (the x, y
coordinates) of the observed features. We now use the results from our photoionization
model fits to quantify as much as possible the shape and positions of these structures in
the line-of-sight (z) direction.

The model-fitting returns a three-dimensional distance Rmoder between the ionizing source

R136 and the gas cloud in question. We can compare this to the projected distance

rprojeaed to estimate 2, the difference between the line-of-sight distance to the gas cloud

and the line-of—sight distance to R136. We use

IZI=(R2 r2 )1’2 . (12)

model — projected
Our Slit Position 8 passes very near to R136 and through the bright arcs associated with
[PS 2 and 3, which are to the NE and SW of R136 (see Figure 2.9). Figure 2.19 a map of
the fitted z in units of pc at the top. The bottom of Figure 2.19 shows 2 along Position 8,

plotted vs. the declination offset in pc from R136. Equation 12 does not constrain z to be

111

positive or negative, but in the vicinity of R136 we can safely assume a face-on geometry
where z is positive (i.e. the gas is behind the cluster). This places the background gas 40
pc behind the central cluster. The sharp increase in 2 at zero declination offset is an
artifact due to contamination from the light of R136 at that point. The bright IFs 2 and 3,

situated to either side of R136 at a projected distance of approximately 9 pc (40 arcsec),
have z ~ 0. If the assumed Q0 were 10 percent smaller they would have 2 exactly equal to

0. Thus, within the uncertainty of Q0, the IF 5 are at the same line-of—sight depth as R136.
A region N of R136 was identified by 109 to be of especially high excitation, and is also
shown by our spectra to have a very high ionization parameter. This region is part of the
bright arc to the NE of R136. 109 concluded that the high excitation was the result of
local photoionization by a group of three WR stars which lie close to the IF at least in
projection. If this were the case then the measurement of 2 would be decreased due to
enhanced flux from the WR stars. Our Slit Position 6 samples the IF part of the region of
interest, but at a different location further from the possible influence of the WR stars
mentioned by 109. At this position there are no obvious stars that might be local
ionization sources. The 2 values found along this slit position are very close to zero,
similar to those found along Position 8. The simplest explanation is that despite the
presence of the WR stars, R136 is still the dominant ionization source along the IF,
including at the position identified by 108.

The two most eastern slit positions, 1 and 2, sample the cavity to the east of R136. In
the top panel of Figure 2.20, as in Figure 2.19, we plot 2 as a function of the offset along
the slit in pc, with the location closest to R136 being the zero point in offset. There is

considerable scatter in the 2 values. To explore the possibility that this scatter is caused

112

by uncertainties in the electron density, we recomputed 2 using

1/2

2 - +( M R2 _ r2 ) (13)
corr — — n model projected
avg

where nayg is an assumed average density in the local part of 30 Dor and nmodel is the
density from the best fitting model. For the low-density or very faint regions of the
nebula, where the density is poorly constrained (as is the case for this position), it would
have been desirable to add an additional term to the goodness of fit parameter which

included a comparison of the modeled density to the local density averaged over some

area. To show the effect this would have we use equation 13 with atypical ”avg = 101'75
for the background wall in this region. The bottom panel of Figure 2.20 shows that

assurrring a constant typical density decreases the noise in z significantly. This would be

valid so long as there are no significant density gradients effecting "avg and the modeled
U is correct.

The inclination of the ionized face to the line of sight back to R136 is also a major source
of uncertainty. It is assumed that the gas forms a plane parallel slab at a distance r from

the ionizing source, but the inclination relative to the ionizing source modifies the
derived distance Rmodei from R136 such that the true distance to the illuminated face Rm.e
would be

R =R (c056)1/2 , (14)

true model
where 9 is the inclination angle of the illuminated face relative to the direction to the
ionizing source. There are at least three special cases of the modeled Rmodei which should

be considered in connection with possible inclination effects. The first is where Rmodei

113

equals the projected distance r on the sky. If 00 is correct, this indicates that the z
coordinate of the ionized gas along the line of sight is zero. This also implies 9 is equal to

zero and the ionization front is being viewed nearly edge on. Another special case is

when R0 < r, which means that the ionizing source(s) must be closer than the model
indicates. This requires a different or additional ionization source than R136.

The third case of special interest is an abrupt, discontinuous change in in the derived

Rmodei. This could be due to a discontinuity in either or Ruue. Sudden changes in the

inclination angle (where = 0 is equivalent to a plane-parallel slab oriented
perpendicular to the direction back to the ionizing source) occur at the lip of a tilted
ionization front, as in the example labeled “a” in Figure 2.21. The horizontal face at the
far left in the figure would receive a much lower ionizing flux than the tilted, face-on
portion. A situation like this occurs with the Orion Bar (see P09).

There is significant evidence throughout 30 Dor that such changes in 0 exist. Take for
example the eastern limb that includes IF 4, or the limb near the molecular gas
corresponding to IF 3. Both of these IFs border the central region of high ionization

parameter discussed in Section 4. In both cases there is a strong, large-scale increase in

the [S II]/Hor ratio just beyond the peak in the [S 11] emission, typical of all the other
edge-on IFs, which indicates a decrease in U. This occurs despite a lowered gas density
and similar projected distance from the central cluster. The simplest explanation for this
is a geometry sirrrilar to that shown in example “a” of Figure 2.21.

An alternative explanation for a discontinuity in 0 is a true geometric discontinuity in

Rm, such as a free-floating filament or an optically thick overlapping shell which blocks

114

our line of sight to the background gas, as sketched in example “b” in Figure 2.21. If
such an effect were important, the reddening free 3 cm and 6 cm radio continuum images
from Lazendic et al. (2003) that trace the ionized gas would show a different structure
than the optical data. On a large scale such a geometry is ruled out by comparing the Av
measured with optical observations to those that are made with optical-radio data. We
conclude that with the exception of the regions 190 arcsec south and approximately 190
arc north of R136 covered by slit positions 16 and 17, the observed H II region all across
the face of 30 Dor is largely a continuous, unobstructed structure.

6.3 What Determined the Structure of 30 Dor?
The gas component of 30 Dor has a very complex distribution in space which is seen in
all wavelengths from the IR to the X-ray. The gas got where it is today in response to
pressure forces that have moved it there. To try to understand better why it has taken on
its current distribution, we take an inventory of the various kinds of pressure currently at

work in the nebula, and of their relative strengths at different points in the gas.
6.3.1 Thermal gas pressure in the H” zone

As a first step, we consider the thermal gas pressure observed at the IF at each point in

the nebula. This is mapped in Figure 2.22, where the [S II] density and [0 III]
temperature measurements have been used to compute Pgas = 2 nekTe. There is very little

change in T., from the illuminated face of a cloud to the ionization front (the IF).

Therefore the reported pressure is equal to the gas pressure at the IF. We want to see if

P903 is in equilibrium with the the sum of all the external pressures acting on the ionized

layer.

115

The region of highest pressure shown in Figure 2.22 forms circular ring around the
central cluster, reflecting the circular regions of high excitation and high density shown
in Figs. 2.7c-g. Immediately outside this region (RA +/- 100 arcsec ) the gas pressure

declines by a factor of 3, with the exception of the pressure in [PS 4 and 6.
6.3.2 Magnetic fields

In Orion (P08) and M17 (P09) we have explored the effect of pressure due to the
compression of tangled magnetic fields, in relation to other pressure sources included in
the overall equation of state. With no magnetic field measurements available for 30 DOT
like those available for M17, nor highly resolved ionization fronts like in Orion, we
cannot constrain the magnetic pressure. As is the case in Orion and M17, magnetic
pressure may be important inside the PDRs and molecular regions of the clouds in 30
Dor, as part of the mechanism by which the cooler regions of the clouds push back
against the gas pressure exerted by the H“ zones. The only observational evidence for a
structured magnetic field in 30 Doradus come from dust polarization (Nakajirna et al.
2007) which should follow a local magnetic field (Davis 8: Greenstein 1951). These
polarization maps do not provide constraints to B but show shell like structures 2.5' north
of R136 between the bright arcs and the region 24 identified by Townsley et al. 2006. In
the expanding shell to the east of R136 including IF 4 the field also seems to form a shell
with the B vector perpendicular to a ray toward R136. However based on arguments
below, the expected stellar radiation pressure is low, and the expected gradient in the H+
region density profile will be small. This will prevent the amplification of any magnetic
fields present in the H11 region constant until the PDR. Therefore the presence of a

magnetic field should not affect our conclusions derived from our spectroscopic data for

116

Whit
Ullfe

gird

lit" C
and
tilt
lie

all

\‘a‘.

rim

the Hi", but the existence of a magnetic field and it's effects should be considered in future

studies of the PDR.
6.3.3 Radiation pressure from the ionizing stars

While there is significant uncertainty in the true geometry of 30 Dor, there is much less
uncertainty in the ionization parameter U. Accounting for the inclination angle, U is
given by

90
U: 2 £059 (15)
4 1r Rman

 

or in terms of the model parameter R0

00
4TrR0an

recovering Equation 1. While the degeneracy between the distance to the illuminated face
and the inclination angle remain unresolved, the solution for the ionization parameter is
unique.

The pressure associated with the momentum absorbed from the incident photons can be
approximated in terms of U by

P =UXnH(hv) (17)

stars
where <hv> ~ 20eV is the average energy per photon of the SED. This approximation is
valid if the optical depth to ionizing radiation is greater than unity. Using this

approximation we compare the observed thermal pressure to the derived radiation

pressure, as in Figure 2.23 which shows a map of loglo(Psta,~,/Pgas). There is a very strong

correlation between the gas pressure and the radiation pressure within the highly ionized

117

legit
press

10112

fiat
(195
30 p

in 51

air?

the

the

region, indicating that the radiation pressure is having a strong effect. Here, the radiation
pressure is approximately 1/3 of the total gas pressure gas pressure. Outside of the highly
ionized region around R136, this ratio drops to below 1/ 10. For example, in the outlying
IF 4 and IF 6, the radiation pressure is equal to 0.05 of the gas pressure.
From this it is clear that while in the high-pressure ring the effect of direct radiation
pressure is likely to be quite important, it is negligible outside of that region ( i.e. for r >
30 pc). This dichotomy can be seen in the plot of electron density vs. U in Figure 2.24.
In such a plot two populations are evident. The first shows a correlation between 11., and
U in the region near the cluster. This is a scaled up and more complicated form of the
scaling between density and U found in the Orion Nebula by Baldwin et al. (1991) and
by Wen & O'Dell (1995). The second population has no correlation between density and
radiation pressure suggestive of another important physical process.

6.3.4 Pressure from the hot X-ray-emitting gas

In the cases of M17 and Orion, the O-stars contribute to the dynamics via stellar winds

which thermalize with a preexisting 106 - 107 K plasma. This hot plasma will either
escape into the ISM or be confined by the surrounding molecular cloud. In the later case,
the pressure associated with this gas is thought to form the bubbles and cavities which
characterize the central parts of many H 11 regions.

In 30 DOT, this situation is made much more complicated because of the many SN e that
presumably also have contributed to the hot gas. The definitive study of the diffuse X—ray
emitting structures in 30 Dor is by Townsley et al. (2006), who identified 17 unique X-
ray emitting regions that lie within the area covered by our spectroscopic maps. For each

region, Townsley et al. gave the temperature, the absorption-corrected luminosity, and

118

the area on the sky. If we make the working assumption that the volume containing each

region is spherical, we can compute a volume density and then a gas pressure Pym for the
X-ray emitting gas in each region. Pram, will be the sum of the mechanical energy input

from stellar winds and SNe. PXq'gy is shown in Figure 2.25 in cgs units and listed in table
2.11.

The Px-my in the different regions are all very similar to each other. This suggests that the
X-ray emitting volumes are interlinked, an idea that has previously been put forth on
other grounds by Townsley et al. (2006). Could the pressures have equalized within the

lifetime of the existing star cluster? Using the densities to derive a sound speed, we find a

sound crossing time of 105 yr for the largest bubbles, much shorter than the lifetimes of

the current generation of O stars.

The energy stored in the X-ray emitting gas is very large. The cooling time can be

approximated by two; = A ne / kT , where A = 5 x 10'23 erg cm'3 is the estimated

emissivity of an X-ray gas with kT = 0.7 keV ( Landi & Landini 1999 ). For the

parameters for 30 Dor, it is of the order two; ~ 15 Myr, so a confined plasma will remain
at X-ray emitting temperatures for the typical lifetime of a few million years for an H II
region, at which time the molecular cloud will be dispersed. Therefore the heat energy
currently stored in the X-ray emitting gas could have been accumulated over the livetime
of the star cluster.

We also checked to make sure that the X-ray emitting gas does not prevent ionizing

radiation from reaching the outer parts of the 30 Dor nebula. We used Cloudy to compute

a coronal model of the X—ray plasma, and found a neutral H fraction log[n(H°)/n(H)] =

119

-7.41. Assuming a typical hydrogen density 11“ = 0.14 cm'3 the optical depth to ionizing

radiation over the distance r = 50 pc to IF4 (on the far eastern wall of the large cavity) is
T = n(Ho)r o = 5.1X10'6, where o is the photoionization cross section of hydrogen equal

to 6><10'8 cm'z. Although the X-ray emitting gas has a big effect on the dynamics of the
H 11 region, it does not impede the stars from providing the ionization.

There is one region not identified by Townsley et al. (2006) that is especially interesting.
North of their region 16 and east of their Region 24 is an area of very low emission at all
wavelengths. Figure 2.26 shows this area in the SOAR [S II] (left) and Chandra 0.5-
0.7keV (right) images of the region. The contours trace the X-ray surface brightness and
outline the Townsely et al. 2006 Regions 16 and 24. Both Regions 16 and 24 are very
bright in X—ray errrission and are bounded by well-defined IFs seen in [8 II]. The
intervening region of low X-ray surface brightness is the only region within 30 Doradus
where the optical morphology clearly suggests a large-scale outflow of gas. This
manifests itself as long filaments all oriented parallel away from R136. It is
understandable that this was not noticed before. This region was not covered by the HST
observations and the filaments are completely unresolved at the resolution of the MCELS
survey. This seems to be an example of a region where the the X-ray gas is not contained.
The expected result is an outflow and low density. This region was not covered by the
kinematic study by Chu and Kennicutt (1994). The lack of this type of structure
anywhere else in 30 Dor supports our conclusions that the X-ray gas is for the most part

confined by the molecular gas, resulting in a pressure equilibrium between it and the

104 K gas in the IFs.

120

IF 4 seems to be a clear example of a place where the X-ray pressure is dominant. The
gas pressure along the wall is 5.8 ><10'10 dyne cm'z. The expected contribution from stars

in the form of radiation pressure at that distance is 2.8 ><10'11 dyne cm’z. The other
observed pressure source in the region is the diffuse X-ray emitting gas (T ownsely et al.
2006). X—ray Region 12 Townsley et al. (2006) seems to fill the cavity that is bounded by

IF4. Using the observed surface brightness and assuming a spherical geometry we
estimate the electron density of the hot gas to be equal to 0.14 cm‘3. Using the

temperature reported for Region 12 the thermal X—ray pressure is equal to 5 ><10'11 dyne

2

cm' , in close agreement with the measured pressure in the wall. This also means the gas

pressure at the illuminated face, where the H 11 region is in contact with the X-ray gas, is
equal to the gas pressure at the IF, in close agreement with our constant density models.

In Figure 2.27 we show the ratio of the X-ray to gas pressure. Aside from the [PS
southwest and northeast of R136 which outline the molecular cloud, the ratio of Px-my/
P905 is between 1 and 10 for the entire region. Given the uncertainty in the assumptions
regarding the geometry of the X-ray-emitting gas and how it affects Px-my, the typical
Px-my/ Pgas ratio is higher than 1. Except in the inner regions, the pressure in the H+ zone

clearly is set by Prom}. which dominates the equation of state by an order of magnitude,
and is likely driving the current outflows, expansion and compression of the remaining

molecular material.
6.4 The Global Abundances

The most reliable abundance measurements from this study are for He, 0, N, and S

121

relative to H. Our He abundance is consistent with all values in the literature, which is
reassuring since there is very little scatter in the reported values of the He/H ratio. As is

shown in Table 2.10, there is much less agreement for O, N and S between different

studies. The reported ranges in the (O/H) abundance ratio are -3.75 S log(O/H) S -3.5.

We find log(O/I-I) = -3.75, in agreement with the lower limit. Like O, our S abundance

log(S/H) = -0.32 is within the range -5.32 S log(S/H) S -5.01 found in other studies. The
(S/O) ratio found here, log(S/O) = -1.57, is equal to the average of the (S/O) ratios found
in other studies.

The place where we find significant discrepancies with previous work is in the resulting
(N/O) and (MS) ratios (Table 2.12). Although our measurement of log(N/H) = 4.91 falls
within the range of previous measurements, our measured log(N/O) = -1.16 and log(N/S)
= 0.41 are both 0.1 in the logarithm higher than in any other study. This could
conceivably be an artifact of the of the SED adopted here, but it is fair to say that our
abundances represent a much broader average over the full 30 Dor nebula than do the
previous results, and the difference may be real.

It is well known that N can be enhanced due to secondary processing in the CNO cycle in
stars more massive than the Sun (Pettini et al. 2002). N 11 and S11 have very similar

ionization potentials and trace the same physical conditions in the nebula. The ratio of
the [N 11] 16584 to [S II] 16716+16731 surface brightness, which is mapped in Figure
2.28, indicates changes in relative abundances of MS. Using this ratio as an indicator we
find two regions with a possible enhancement of the (N/S) abundance, one around R136

and the bright rim to its east, and the other around Hodge 301. Hodge 301 is a 25 Myr old

cluster which lies 190 arcsec NW of R136, and its vicinity has likely been polluted by

122

SNe.

These results seem to correlate with the surface brightness of the nebula. To check for
any systematic errors in the line measurements due to changes in surface brightness, we
measured line ratios by hand (using the IRAF SPLOT routine) at five locations with
low surface brightness and five locations with high surface brightness. We found that
uncertainties in placing the continuum could lead to a 5 percent increase over the
uncertainties returned by our automated measuring routine, but that there is no evidence
of systematic errors. We conclude that the line measurements are accurate and that real

variations in elemental abundances and/or unaccounted-for physical processes are

causing the ([N II] 16584 )/ [S 11] (16716+16731) intensity ratio to change across the
nebula.

6.5 Abundance Comparison With Other Methods
The technique for determining abundances used here has not, to our knowledge, been
used previously for a single H 11 region. It is similar to the strong line methods developed
by Pagel et al. 1992; Tremonti et al. 2004; Kewley & Doptia 2002; Pettini & Page12004
and others, which are used to study the spatially integrated spectra of distant H [1 regions

and galaxies. Our technique improves on the strong-line approach because it includes

direct empirical measurements of Te and ne. It also removes the need to describe an
entire, complex nebula by a single ionization parameter and density, and instead

examines many small regions with many different densities and ionizing fluxes. Using
the observed correlations between ([0 III] 15007)/H[3, ([N 11] 16584)/Hor, ([S III]

16312)/([S II] 16716 + 16731) and [S II]/Hor in conjunction with Cloudy models and

123

temperature constraints, we can measure the ionization parameter and total gas phase
abundances using relatively few emission lines.

Our approach has certainly glossed over many of the finer details often considered when
doing a an empirical measurement at one place in the nebula, but averages over a much
greater fraction of the full nebula and returns results within the range of reported
empirical measurements except for the unusually high (N/O) and (N/S) abundance ratios.
The next question is: What would the result be if 30 Dor were instead observed from a
much larger distance so that only the strong-line methods could be used? It is important
to make this test because the strong-line technique is now routinely being applied to very
large samples of objects seen out to large look back times, and many big-picture results
are being deduced from the resulting abundance measurements based on the integrated
spectrum of entire HII regions and galaxies. These include a survey of over 500 galaxies
from z = 0 — 1 ( Hu et al. 2009), as well as single galaxies at z = 1.7 (Yuan & Kewley
2009).

There are a number of warning flags that there may be systematic errors in the strong-line
abundances. The different strong-line methods do not agree among themselves, although
calibrations can transform them all onto any one arbitrarily chosen scale (Kewley &

Ellison 2008 ). The strong-line abundances also are known to disagree with purely

empirical abundance measurements based on direct measurements of Te, ne and the
ionization fractions for different elements (Kennicutt et al. 2003; Tremonti et al. 2004;
Kewley 8: Doptia 2002; Pettini 8: Pagel 2004). Bresolin et al. (2009) have recently
determined the abundance gradient in the nearby spiral galaxy NGC 300 using empirical

measurements of individual H 11 regions, and find that its slope and absolute value are in

124

good agreement with abundances determined from stars in the same galaxy, but
systematically different than the strong-line results.

All of the comparisons of strong-line to empirical techniques made to date are for entire
H 11 regions or entire galaxies each treated as a single idealized point. For example, the
Bresolin et al. (2009) study uses data for 28 H 11 regions spread across the face of NGC
300. Each of those H II regions is a complex thing similar in size and nature to 30 Dor.
Yin et al. (2007) also sought to calibrate strong line methods using the integrated spectra
695 galaxies with [0 III] 4363 detections using mostly data from the SDSS-DR4.

For this reason, our study which takes into account the detailed structure within 30 Dor is
a valuable complement to those earlier results for the H11 regions. We first compare our
30 Dor abundance results to those that would be determined using the strong line method
of Dopita et al. (2006). These authors developed a grid of models incorporating an
assumed history of stellar evolution in the H 11 region, which in turn determines the
current SED and mechanical energy input via SNe and stellar winds. The models are
parametrized by three variables which determine the physical conditions and emission
line spectrum. They are the age of the cluster t, the metallicity Z, and a (gravitating
mass)/pressure ratio R defined as

R=1og10[(MCLUSER/Mo)/(Plk)] (18)

in cgs units.
To compare our results to strong line methods we will use line ratios, the [S 11] density

and the O 111 temperature measured from our globally integrated spectrum. For the

observed cluster mass of the order 104-105 Me, m; = 102, T=10,500K, R is observed to

be between -1 and -2. Z is primarily characterized by the oxygen abundance and given in

125

units relative to solar abundances.

In the Dopita et al. models, la is defined as log(O/H) = -3.34 . On this scale our derived

oxygen abundances becomes Z = 0.41 20. The available models cover the span -6 S R S
2 in increments of 2. The modeled ages range from 0.1 to 4.5 Myr in increments of 0.5

Myr beginning at 0.5 Myr. Finally the available abundances are 0.05, 0.2, 0.4, 1 and 2 in

units of Z/Zo,
We compared the line strengths predicted by each Dopita et al. (2006) model to the lines

measured from our 30 Bar composite spectrum, and ranked them using Equation 11. We
did this using the intensities of He I 16678, [0 111] 15007, [O I] 16300, ([0 II]
17320+17330), [N 11] 16584, and ([S 11] 16716 + 16731) relative to H8 and the [S 11]

16716/ 16731 ratio. Identifying the initial best fit model, we interpolated the models
between different values of R. We find the best agreement with the observations come
from models with Z = -0.4, -2 < R < -1.5 and ages 0.5 < t < 2.0 Myr. The ages are

consistent with the youngest observed population of 0 stars in 30 Dor. For younger ages

the difference between the observed and modeled ([0 111] 15007)/ HB ratio is less than 10
percent, but the lower ionization lines are under predicted by about a factor of two. The
situation is reversed for older ages, when the models seem to be to be too weakly ionized,
with the best fit at 2.0 Myr. We find that despite the complexities of the region, the
assumptions and simplifications of this strong line method are sufficient to reproduce the
known physical properties of 30 Dor derived from our more detailed treatment. In
particular, the abundances Z are in almost exact agreement.

A simpler abundance measurement can be made using the calibration involving only [N

126

II]/H0t (Pettini 8: Pagel 2004). This is an empirical measurement based on a sample of
137 extragalactic H 11 regions with well-constrained (O/H) and [N II]/Hor ratios.

Inserting the value of [N II]/Hor from our composite spectrum into Equation 1 from
Pettini & Pagel (2004) gives a derived oxygen abundance [O/H] = -3.87 +/- 0.38. This is
-0.12 dex lower than the value from our study or from the strong line method of Kewley
and Dopita (2002) using the models from Dopita et al. (2006), but agree with those
results to within the quoted uncertainty of the method.

In the comparisons between the strong-line methods and the purely empirical methods,
the empirical results are often considered suspicious because temperature fluctuations
can weight hotter vs. cooler regions of the gas in different ways in different emission

lines. We have already described (Sect. 4.4) our measurement of the temperature

fluctuation parameter (2 in 30 Bar. We find a larger value than had been reported
previously, one which would have a significant effect on the abundances determined
from the empirical method, but we are not confident of that result because of the
possibility of systematic errors.

In summary, the abundances derived from our point-by-point analysis of 30 DOT are in
good agreement with those that would have been found with the strong-lined methods.
This result in favor of the strong—lined abundance determinations is contrary to the
conclusions reached by Bresolin et al. (2009) from their very different test of the strong-

lined method.
7 Conclusions

We have obtained and are making publicly available a dense grid of long-slit optical

127

spectra which measure key emission lines in the 114100-7400A wavelength region over

8 140x80 pc2 (10x6 arcminz) region of 30 Bar. These spectra were taken at 37 slit
positions and then extracted and measured as 4238 individual 1D spectra. The resulting
'data cube' of emission line intensity measurements is also publicly available.
Supplementary spectra at a few additional slit positions extended the coverage to a wider
wavelength range.

We also obtained a set of subarcsec-resolution direct images in multiple narrow-band

filters, covering a 170 X 180 pc2 (12X 13 arcminz) field of view, which we have used to
identify and catalog a large number of structures of special interest. These include
regions likely to be locally ionized by embedded stars, as well as edge-on ionization
fronts seen in both the visible and infrared and 'elephant trunk' pillars. With JW ST and
ALMA on the horizon and the newly upgraded HST, these structures will provide
important sites for further constraining the physical conditions in the PDRs of distant
low-metallicity extragalactic H II regions through calibrating measurements in this
closest example of such an object.

We find that the cluster of O stars centered around R136 is the dominate source of
radiation for almost all of the nebula. We find no compelling evidence of large scale
variations in U resulting from multiple, embedded ionization sources.

We have combined photoionization models with empirically determined nebular
temperatures and densities to measure point-by-point variations in the ionization
parameter in order to reconstruct the ionization structure of 30 Doradus. This technique
provides us with a comprehensive measurement of the global abundances. We find an

oxygen abundance 0.15 dex lower than that found in recent studies of the bright arcs, but

128

within the observed range of reported abundances. Attempts to use the SED and
abundances used in the photoionization modeling of 30 Bar by TP05 failed. We found
differences in the gas temperatures depending on the photoionization code used. This led
us to adopt a new, lower oxygen abundance consistent with V02. We find systematically
higher N/O and N/ S ratios than previous studies, which may suggest we are more

sensitive to secondary enhancement of N.

Despite a clear correlation between ne and U in 30 Bar, the geometric dilution of
radiation pressure over many tens of parsecs is significant. As has been suggested by
other authors in previous papers, we have shown that the dynamics and large scale
structure are set by a confined system of X—ray bubbles in rough pressure equilibrium
with each other and with the confining molecular gas. The long cooling time of the X-ray
emitting gas means that it will dominate the dynamics until it is no longer confined. This
is unlike the the situations in the much smaller Orion and M17 H II regions which we
have also studied in detail.

The often-studied bright arcs do not represent the average physical conditions in the
nebula, nor do they account for the majority of the line emission. They are of higher
density and ionization parameter than would characterize the whole nebula if viewed as a
single point-like source at a great distance. Studies focusing on the fainter emission are
likely to examine the physical conditions characterizing those observed in distant
nebulae.

We use the results from our spatially resolved survey to test the accuracy of two strong-
line abundance measurement methods. Despite the general complexity and large scale

changes in U across 30 Dor, there is excellent agreement between our measured

129

abundances and those determined using the strong-line method outlined by Kewley 8:
Dopita. Their models representing entire H 11 regions and H II galaxies as being
characterized by a single, time dependent ionization parameter accurately reproduce the
global spectrum of 30 Dor. This includes correctly estimating the ratio of cluster mass to

gas pressure, the cluster age and the oxygen abundance.

Acknowledgements
EWP gratefully acknowledges financial support from the National Science Foundation
(grant AST-0305833), NASA (07-ATFPO7-0124, STSCI GOO9736.02-A and STSCI AR-

10932) and Michigan State University's Center for the Study of Cosmic Evolution.

130

Table 2.1: Summary of imaging observations used in 30 Doradus optical mosaics.

 

 

 

 

 

 

 

 

 

 

 

 

Table 2.1
DETAILS OF DECEMBER 2007 NARROWBAND OBSERVATIONS OF 30 DORADUS
Filter No. of Exp x Duration No. Positions FWHM Min/Max
at each position (arcsec) Exposure(s)
6563 3 x 1505 6 0.9 300/1480
6738 3 x 4505 6 1.04 9005700
6850 3 x 4505 6 0.8-1.0 900/5700
5019 3 x 2255 5 1.01 450/2500
5130 3 x 2255 5 1041.1 450/2500

 

131

 

Table 2.2: Summary of Blanco and SOAR spectroscopic observations of 30 Doradus. Included in columns
1-5 are position number corresponding to Figures 2.1 and 2.2, RA and Declination of the slit center, PA

and exposure time.

 

Table 2.2

SURVEY OF 30 DORADUS

OBSERVATIONAL DETAILS FOR THE SPECTROSCOPIC

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Position RA. (2000) Dec. (2000) PA Exposure
No. (slit center) (slit center) (deg E of N) Time (s)
Blanca Telescope
1 05:39:16.53 -69:06:37.04 13 405
2 05:39:11.25 -69:06:31.18 13 425
3 05:39:06.29 -69:06:29.68 13 405
4 05:39:00.44 -69:06:22.80 13 250
5 05:38:56..29 -69:06:11.60 13 500
6 05:38:52.21 -69:06:11.81 13 260
7 05:38:45.00 -69:06:03.65 13 260
8 05:38:41.38 -69:05:55.66 13 320
9 05:38:37.98 -69:05:53.44 13 300
10 05:38:35.17 -69:05:48.71 13 120
11 05:38:30.08 -69:05:38.24 13 400
12 05:38:24.82 -69:05:29.57 13 450
13 05:38:17.42 -69:05:28.52 13 500
14 05:38:08.61 -69:O5:18.33 13 300
15 05:37:59.70 -69:04:56.19 13 250
16 05:38:49.47 -69:01:28.18 13 600
17 05:38:37.48 -69:08:52.47 13 600
20 05:39:16.50 -69:04:51.76 103 400
21 05:38:28.34 -69:03:50.99 103 400
22 05:39:15.65 -69:05:28.77 103 405
23 05:38:27.16 -69:04:29.19 103 308
24 05:39:12.49 -69:05:54.62 103 500
25 05:38:25.24 -69:04:55.67 103 355
26 05:39:12.16 -69:06:22.34 103 500
27 05:38:25.90 -69:05:23.89 103 460
28 05:39:06.04 -69:06:36.65 103 600
29 05:38:22.74 -69:05:43.05 103 520
30 05:39:07.60 -69:07:05.02 103 700
31 05:38:24.16 -69:06:13.86 98 450

 

132

 

 

Table 2.2 (cont'd).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

32 05:39:06.61 -69:07:26.38 103 900
33 05:38:21.35 -69:06:31.58 103 500
34 05:39:06.01 -69:07:58.72 103 400
35 05:38:20.45 -69:07:00.61 103 300
36 05:39:01.11 -69:08:34.64 103 600
37 05:38:15.70 -69:07:38.32 103 450
SOAR Telescope
40 05:38:46.430 -69:05:35.06 10 1000
41 05:39:03.142 -69:06:40.45 33.9 3000
42 05:39:02.763 -69:06:45.82 61 2000
43 05:39:21.240 -69:06:55.44 61 300
44 05:39:08.649 -69:07:02.80 80 3000

 

 

 

 

 

133

 

Table 2.3. Line IDs and wavelengths. Measured strengths for each of these lines are listed in Table 2.4
and/or 2.5 at every extracted paint in the nebula. Rest frame wavelengths with an asterisk indicate the lines
used below to fit the Blanca data to models at each point in the nebula as described below.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 2.3. Line IDs
AObs 10 ion [1(R = 3.1) Data Set
4106 4101 H I 1.43 SOAR
4344 4340 H I 1.35 Blanco+SOAR
4366 4363 [O 111] L 1.34 Blanco+SOAR
4476 4471 He 1 1.30 Blanco+SOAR
4800 4800 Continuum 1.19 Blanco+SOAR
4866 4861* H I 1.16 Blanco+SOAR
4926 4922 He 1 1.14 SOAR
4963 4959 [0 111] 1.13 Blanco+SOAR
5012 5007* [0 111] 1.12 Blanco+SOAR
5625 5625 Continuum 0.97 Blanco+SOAR
5880 5875* He 1 0.93 Blanco+SOAR
6308 6300 [O l] 0.86 SOAR
6318 6312 [S III] 0.86 Blanco+SOAR
6364 6364 NS [0 I] - Blanca
6552 6548 [N 11] 0.82 Blanco+SOAR
6570 6563 H I 0.82 Blanco+SOAR
6590 6584* [N 11] 0.81 Blanco+SOAR
6684 6678 He 1 0.80 Blanco+SOAR
6721 6716* [S 11] 0.79 Blanco+SOAR
6738 6731* [S 11] 0.79 Blanco+SOAR
7072 7065* He 1 0.74 Blanco+SOAR
7142 7135* [Ar 11]] 0.73 Blanco+SOAR
7288 7281 He I 0.71 SOAR
7325 7320 [0 II] 0.7 SOAR
7337 7330 [0 II] 0.7 SOAR
7758 7751 [Ar 111] 0.63 SOAR
9075 9069 [S 111] 0.48 SOAR

 

 

 

 

 

134

Table 2.4. A portion of the reddening-corrected Blanca spectroscopic data cube in table farm as described

in the text. The units of electron density and temperature are cm.3 and K, respectively. The dereddened HB

surface brightness are reported in units of erg s'1 cm'

2

arcsec'z, The other emission lines are reported as

100><S(line)/S(H1). The entries for position number 38 are average values for the whole data set, as

described in the text.

 

TABLE 2.4
EXAMPLE OF PUBLICLY AVAILABLE SCIENCE PRODUCT FROM Blanca
SPECTROPHOTOMETRIC SURVEY OF 30 DORADUS

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Pas ARA [lDec raw log 11., -err +err T(03) -err +err Av F(HB) err

1 2 3 5 6 7 8 9 10 11 12 13

1 148.1 -181.6 1.41 0.41 0.27 13600 1010 1500 1.12 6.56E-15 5.65E-17
1 148.6 -179.1 1.70 0.24 0.16 10900 1280 3210 1.13 63813-15 63813-15
1 149.2 -176.7 13 1.01 0.01 0.51 9690 1360 6450 1.21 6.80E-15 6.80E-15
1 149.8 -174.2 18 1.73 0.22 0.15 12900 1270 2320 1.23 7.07E—15 7.07E-15
37 1.4 -128.0 598 2.01 0.09 0.08 19900 2040 3270 2.50 2.28E-14 3.86E-16
37 3.9 -128.5 603 1.88 0.11 0.09 13900 1320 2280 2.86 3.03E-14 6.13E-16
38 0.0 0.0 1 2.11 0.001 0.001 10800 8 8 0.00 6.09E-14 4.09E-18
38 0.0 0.0 2 2.08 0.001 0.001 0 4 4 1.28 1.43E-l4 7.96E-019
38 0.0 0.0 3 2.08 0.001 0.001 10700 5 5 0.00 5.75E-14 3.20E-18

 

 

 

135

Table 2.5. A portion of the reddening corrected SOAR data set, in the same format as Table 2.4 but with
additional columns because more emission lines are measured. These include [5 III] 19069 which was used

with [S 111] 16312 to measure the gas temperature using [S III] in a manner identical to that described
using [0 III].

 

Table 2.5. A portion of the dereddened SOAR spectroscopy results

Pas RA De raw log(ne) T(03) K -err +err [S3]T -err +err Av F(HB)

1 2 3 4 5 6 7 8 9 10 1 1 12 13

40 4.65 -72.58 260 2.34 13112 579 724 11960 43 43 1.49 1.07E-13
40 5.03 -70.42 275 2.43 14065 1116 1706 9875 66 68 1.89 7.86E-14
40 5.41 -68.26 290 2.67 11542 1048 1884 11792 68 70 1.88 8.76E-14
40 5.79 -66.09 305 2.55 13248 968 1443 11673 57 58 1.93 9.77E-14
40 6.18 -63.93 320 2.7 11483 578 766 11972 34 34 1.67 1.40E-13

 

 

 

 

 

 

 

 

40 6.56 -61.77 335 2.8 11915 448 547 11648 26 26 1.48 1.67E-13

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

136

Table 2.6: Possible regions with a locally enhanced ionization parameter due to nearby massive stars.
Columns 1-6 correspond to a unique ID, RA and Dec, RA and Dec offsets from R136 of theianizing star

and radius of obvious influence visible in the [S II]/Ha image.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 2.6 Possible [Legions of Enhanced Flux Due to Nearby Stars
radius
Object ID RA Dec RA Dec (arcsecL
1 05:39:03.44 -69:06:35.72 113 -32 3.97
2 05:39:10.17 —69:06:22.63 149 -19 3.97
3 05:38:57.10 -69:06:06.62 79 -3 1.43
4 05:39:05.43 -69:04:16.27 124 108 2.85
5 05:39:05.51 -69:04:31.61 124 92 2.77
6 05:38:24.46 -69:07:57.28 -97 -113 2.78
7 05:38:38.86 -69:08:16.05 -19 -132 4.76
8 05:38:01.61 -69:04:50.49 -219 73 1.69
9 05:38:01.31 -69:04:50.25 -221 74 1.35
10 05:38:41.19 -69:02:57.75 -7 186 3.07
11 05:38:45.39 -69:02:50.84 16 193 3.83
12 05:38:13.98 -69:07:47.63 -153 -104 10.11
13 05:39:20.55 -69:06:S4.42 205 -51 4.06
14 05:39:03.36 -69:09:33.58 113 -210 3.02
15 05:38:45.08 -69:08:08.06 14 -124 6.26
16 05:38:52.82 -69:06: 12.01 56 -8 3.81
17 05:38:55.79 -69:05:24.90 72 39 2.41
18 05:38:04.84 -69:07:34.84 -202 -91 3.54
19 05:38:31.75 -69:02:14.28 -57 230 3.59
20 05:38:41.76 -69:01:58.79 -4 245 4.27
21 05:38:53.54 -69:02:00.56 60 243 2.43
22 05:39:11.63 -69:02:01.02 157 243 3.07
23 05:38: 15.64 -69:04:37.64 -144 86 9.94
24 05:38:54.94 -69:08:45.76 67 -162 3.77
25 05:38:56.83 -69:08:42.47 77 -159 4.25
26 05:38:54.70 -69:07:45.81 66 -102 5.29
27 05:38:36.43 -69:06:58.7O -32 -55 2.98
28 05:38:36.06 -69:06:47.52 -34 -44 2.04
29 05:38:51.32 -69:06:41.99 48 -38 2.84
30 05:38:49.79 -69:06:44.12 40 -40 1.7
31 05:38:57.40 -69:07: 10.75 81 -67 2.73
32 05:38:46.60 -69:04:28.07 22 96 2.3
33 05:38:36.91 -69:05:08. 19 -30 56 5.97
34 05:39:01.05 -69:06:30.16 100 -26 2.44
35 05:38:58.80 -69:05:24.55 88 39 2.46
36 05:38: 17.62 -69:05:42.83 -133 21 13.67
37 05:39:22.86 -69:07:46.88 217 -103 3.39
38 05:39:12.35 -69:06:02.74 161 1 4.78
39 05:38:14.88 -69:04:31.80 -148 92 2.31
40 05:38:08.53 -69:05:44.36 ~182 19 15.11
41 05:38:10.64 -69:06:17.52 -171 -14 3.69
42 05:38:09.46 -69:06:22.26 -177 -18 3.69
43 05:37:49.22 -69:06: 14.27 -286 -10 4.41
44 05:38:24.70 -69:07:43.55 -95 —100 3.06
45 05:38:30.51 -69:06:46.21 -64 ~42 3.07
46 05:38:29.73 -69:06:57.41 -68 —54 3.07
47 05:38:31.02 -69:06:37.47 -61 -34 2.63

 

137

 

Table 2.7: Prominent IFs suitable for follow up study with multi-wavelength data. From
left to right the columns are ID, RA, Dec, RA and DEC offsets from R136, IF length and
PA. [US with an asterisks identify IFs with PAH emission closer to R136 than the [S 11]

emission.

 

Table 2.7: Prominent Ionization Fronts Detected in IR and Optical Wavelengths

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

IF id RA (2000) Dec (2000) DRA DDec Length PA
1 05:38:44.15 -69:06:59.71 9.3 -55.9 31.0 257.7
2 05:38:55.02 -69:05:42.02 67.7 21.8 39.8 317.9
3 05:38:36.07 -69:06:27.99 -34.1 -24.2 92.2 349.1
4 05:39:22.75 -69:07:17.56 216.8 -73.8 87.3 34.7
5 05:39:11.80 -69:08:14.33 157.9 -130.5 49.0 68.9
6 05:38:59.15 -69:05: 14.79 89.9 49.0 38.2 120.0
7 05:38:51.70 -69:05:06.06 49.9 57.7 38.2 70.0
8 05:37:55.71 -69:05:43.70 -251.1 20.1 75.2 321.9
9 05:38:58.56 -69:08:43.34 86.8 -159.5 2.4 324.7
9 05:37:54.88 -69:05:11.42 -255.6 52.4 59.5 306.1
10 05:37:45.34 -69:05:14.08 -306.8 49.7 28.5 0.0
11 05:38:06.22 -69:07:43.53 -194.6 -99.7 37.3 344.1
12 05:38:11.38 -69:08:06.38 -166.9 -122.6 37.3 283.8
13 05:39:37.31 -69:07:17.48 295.1 -73.7 28.0 26.0
14 05:39:37.57 -69:08:25.22 296.5 -141.4 96.0 63.0
15 05:39:23.05 -69:08:33.57 218.4 -149.8 31.0 310.0
16 05:39:19.76 -69:08:21.01 200.7 -137.2 14.4 29.2
17 05:38:58.60 -69:09:36.36 87.0 -212.6 63.8 16.9
18 05:38:32.82 -69:09:17.55 -51.6 -193.8 63.8 340.7

20 05:39:03.35 -69:08:04.44 112.5 -120.6 13.5 67.9
21 05:39:09.79 —69:04:36.61 147.1 87.2 18.3 142.7
22 05:38:52.60 -69:04:40.27 54.7 83.5 20.3 128.7
23 05:38:44.74 -69:04:19.25 12.5 104.5 20.3 120.0
24 05:38:37.67 -69:04:27.59 -25.5 96.2 13.6 45.0
25 05:38:39.17 -69:03:31.66 -17.5 152.1 25.9 90.0
26 05:38:28.95 -69:02:54.60 -72.4 189.2 29.8 25.7
27 05:38:23.40 -69:02:24.23 -102.2 219.6 62.5 111.4
28 05:38:12.58 -69:02:16.68 -160.4 227.1 50.4 66.2
29 05:38:10.17 -69:02:44.58 ~173.4 199.2 11.4 61.0
30 05:38:07.60 -69:03:16.01 -187.2 167.8 15.7 39.1
31 05:38:14.24 -69:03:29.95 -151.5 153.8 38.8 53.0
32 05:38:38.16 -69:00:34.55 -22.9 329.2 28.8 63.8

 

138

 

 

Table 2.7 (cont'd).

 

 

 

 

 

 

 

 

 

 

 

32 05:38:04.92 -69:04:38.91 -201.6 84.9 9.2 26.3
33 05:38:34.69 -69:01:05.06 -41.6 298.7 20.8 32.5
1* 05:37:57.50 -69:07:44.65 -241.5 -100.9 107.0 322.2
2* 05:38:09.50 -69:08:58.47 -177.0 -174.7 83.8 298.5
3* 053811948 -69:09:55.73 -123.3 -231.9 54.8 343.7
4* 05:38:06.23 -69:10:47.75 -194.5 -284.0 54.8 60.8
5* 05:37:58.58 -69:08:32.56 -235.7 -l48.8 18.4 317.2

 

 

 

 

139

 

Table 2.8 ~ Catalog of bright dense pillars and protruding IFs. Each

pillar has an ID, RA and Dec (J2000), length and position angle.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 2.8 - Catalog of Bright Dense Pillars and protruding IF

Piller ID RA ( 2000) Dec (2000) A RA 6 Dec Leflth PA
1 05:37:28.35 ~69:01:51.86 ~400 252 37 216
2 05:37:30.68 ~69:04:12.44 -384 111 17 120
3 05:37:32.85 ~69:04:16.79 ~373 107 19 115
4 05:37:42.14 -69:02:30.70 ~325 213 26 209
5 05:37:42.85 ~69:04:42.79 ~319 81 10 176
6 05:37:42.95 ~69:04:34.28 ~319 90 10 183
7 05:37:50.28 ~69:06:02.92 ~282 1 19 155
8 05:37:54.67 ~69:05:09.56 ~255 54 9 167
9 05:37:58.57 ~69:03:44.17 ~233 140 7 182
10 05:37:59.29 ~69:05:49.22 ~233 15 8 163
11 05:38:03.81 -69:05:22.05 ~207 42 10 164
12 05:38:05.74 ~69:06:36.40 ~196 ~33 6 138
13 05:38:08.17 ~69:06:51.16 ~185 ~47 33 167
14 05:38:09.49 ~69:07:05.29 ~180 ~61 19 159
15 05:38:11.95 ~69:07:08.39 ~164 -65 26 142
16 05:38:13.19 ~69:04:25.76 ~158 98 13 236
17 05:38:14.22 ~69:07:00.28 ~153 ~56 10 131
18 05:38:15.89 ~69:05:25.19 ~142 39 6 260
19 05:38:17.03 -69:05:57.00 ~137 7 3 112
20 05:38:17.56 ~69:05:02.66 ~131 61 14 200
21 05:38:24.12 ~69:04:56.88 ~99 67 24 206
22 05:38:28.52 -69:08:26.14 ~72 ~142 16 166
23 05:38:28.74 ~69:03:37.29 ~72 147 6 238
24 05:38:30.07 ~69:08:38.73 ~67 ~155 24 138
25 05:38:33.14 ~69:06:02.88 ~51 1 5 153
26 05:38:34.01 ~69:03:33.01 ~45 151 12 268
27 05:38:34.26 -69:06:26.96 ~45 ~23 12 153
28 05:38:36.05 ~69:07:43.52 ~35 -100 12 148
29 05:38:38.27 ~69:08:13.16 ~24 ~129 9 81
30 05:38:40.51 ~69:09:58.77 ~8 ~235 9 111
31 05:38:40.97 ~69:07:27.27 ~8 ~83 12 84
32 05:38:43.16 ~69:07:04.03 3 ~60 7 96
33 05:38:43.20 ~69:01:19.02 3 285 15 173
34 05:38:43.23 ~69:10:21.48 3 ~258 11 132
35 05:38:44.16 ~69:06:58.89 8 ~55 5 90
36 05:38:44.25 ~69:11:10.68 8 -307 9 127
37 05:38:45.18 ~69:05:05.74 14 58 8 304
38 05:38:45.41 ~69:04:19.41 14 104 8 268
39 05:38:45.65 ~69:10:25.07 19 ~261 18 108
40 05:38:45.85 ~69:09:51.28 19 ~227 15 62
41 05:38:46.15 ~69:07:04.11 19 ~60 8 86
42 05:38:46.86 -69:09:33.58 25 -210 7 73
43 05:38:47.05 ~69:07:56.66 25 ~113 13 61
44 05:38:47.85 ~69:07:15.15 30 ~71 14 65
45 05:38:50.55 ~69:10:17.26 46 -253 16 77
46 05:38:53.45 ~69:08:00.23 57 ~116 8 70
47 05:38:54.71 ~69:05:36.11 68 28 3 343
48 05:38:55.01 ~69:05:39.94 68 24 3 309
49 05:38:55.29 ~69:05:40.63 68 23 3 267

 

140

 

 

Table 2.8 (cont'd).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

50 05:38:56.40 ~69:07:21.25 73 ~77 13 30
51 05:38:56.81 ~69:05:30.58 78 33 10 297
52 05:38:57.29 ~69:01:56. 18 78 248 5 32
53 05:38:57.51 ~69:06:27.65 84 ~24 16 16
54 05:38:57.81 ~69:07:56.53 84 ~113 19 68
55 05:38:58.07 ~69:01:51.67 84 252 7 23
56 05:38:58.58 ~69:02: 16.27 89 228 9 233
57 05:38:58.84 ~69:07:28.60 89 ~85 15 43
58 05:38:59.48 ~69:08:04.34 89 ~121 37 67
59 05:39:00.01 ~69:11:45.42 95 -342 22 83
60 05:39:00.14 ~69:06:08.15 95 ~4 6 340
61 05:39:00.29 ~69:08:56.01 95 ~172 21 52
62 05:39:02.13 ~69:08:22.12 105 ~138 17 46
63 05:39:02.47 ~69:06:40.10 105 ~36 17 29
64 05:39:03.49 ~69:07:31.02 111 ~87 19 33
65 05:39:04. 13 ~69:08:14.74 116 ~131 23 57
66 05:39:05.07 ~69:05:54.83 121 9 7 279
67 05:39:07.00 ~69: 10:49.34 132 ~286 7 39
68 05:39:07.51 ~69:10:26.01 138 ~262 14 12
69 05:39:08.05 ~69:08:47.38 138 ~164 18 53
70 0513921 1.46 ~69:08:15.27 154 ~131 9 49
71 05:39: 11.76 ~69:10:53.42 159 ~290 14 55
72 05:39:12.36 ~69:07:44.75 159 ~101 13 26
73 05:39:13.25 ~69:08:56.22 164 ~172 9 91
74 05:39:14.28 ~69:01:45. 16 170 259 23 308
75 05:39:14.63 ~69:10:54.29 175 ~290 24 54
76 05:39:14.72 ~69:07:39.20 175 ~95 13 37
77 05:39: 16.70 ~69:05:28.48 186 35 8 350
78 05:39: 17.98 ~69:10:36.02 191 ~272 11 32
79 05:39:18.95 ~69:05:56.61 197 7 11 355
80 05:39: 18.97 ~69:07:45.44 197 ~102 8 63
81 05:39: 19.05 ~69:07:45.59 197 ~102 8 55
82 05:39:21.91 ~69:04:31.65 213 92 15 335
83 05:39:22.10 ~69:05:27.13 213 37 18 341
84 05:39:22.91 ~69:06:08.13 218 ~4 7 359
85 05:39:23.79 ~69:07:00.26 224 ~56 24 14
86 05:39:24.53 ~69:04:45.88 229 78 9 339
87 05:39:26.37 ~69:05:59.02 234 5 15 27
88 05:39:26.59 ~69:04:42.32 240 81 4 344
89 05:39:26.62 ~69:07:32.13 240 ~88 22 28
90 05:39:27.52 ~69:06: 11.28 245 ~7 9 349
91 05:39:27.79 ~69:04:04.10 245 120 28 320
92 05:39:30.83 -69:05:55.85 261 8 20 25
93 05:39:30.85 ~69:10:11.75 261 ~248 19 44
94 05:39:31.05 ~69:05:29.34 261 34 8 341
95 05:39:31.14 ~69:05:24.43 261 39 6 330
96 05:39:31.23 ~69:05:20.09 261 44 16 357
97 05:39:34.55 ~69:05:25.43 283 38 6 324
98 05:39:37.13 ~69:07:28.57 293 ~85 13 12
99 05:39:37.64 ~69:07: 19.20 299 ~75 10 13
100 05:39:37.86 ~69:07:09.53 299 ~66 8 8
101 05:39:39.19 ~69:11:25.06 304 ~321 7 240
102 05:39:39.53 ~69:08:34.63 310 ~151 5 29

 

 

 

 

 

 

 

141

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 2.8 Eont'd).
103 05:39:39.73 -69:08:31.87 310 ~148 4 58
104 05:39:39.77 ~69: 11:25.05 310 ~321 12 258
105 05:39:44.91 ~69:06:49.21 336 ~45 17 2
106 05:39:47.81 ~69:08:49.79 353 ~166 39 19

 

Table 2.9 ~ Cataloged massive stars in 30 Dor with spectral types.

 

Table 2.9: Ionizing Stars in 30 Dor

 

 

 

 

 

 

 

 

 

 

Spec Type Number Spec Type Number
0313 3 07V 16
03111 12 08.5V 26

03V 22 08V

04V 28 09V 21

05V 11

06V 12 WR 19

 

 

 

142

 

Table 2.10. Abundaces of Selected Elements

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 2.10

Abundances of Selected Elements
He 0 N 5 Ar ref
-~-~ -3.6 ~5.1 ~5.3 ~5.8 Garnett (1999) (for full LMC)
~1.05 -3.75 -5.42 ~5.16 ~5.86 Vermeij et al. 2002
~1.07 -3.5 ~4.79 ~5.01 ~5.74 Peimbert (P03)
~-~- ~~~~ ~~~~ ~5.23 -5.68 Lebouteiller et al. 2008
~~~~~ ~3.69 ~5.21 ~5.32 ~5.84 Mathis, Chu 8: Peterson
~1.1 -3.6 ~4.87 ~5.19 ~5.89 Tsamis & Pequngnot (TP05)
~1.08 ~3.75 ~4.91 ~5.32 ~5.99 Adopted Here

 

 

Units of Loglo [N(X) / N(H)].

 

143

 

Table. 2.11 X-ray pressures of selected regions in 30 Doradus. The
region number corresponds to the regions identified in Townsley et
al. 2006.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 2.11
X-ray Pressures of Selected Regions in 30 Doradus

Region Number Pressure1 x 1010
Townsley et al. 2006
1 1.7
2 1.3
3 3.0
4 10.2
5 7.9
6 10.0
7 13.4
8 7.9
9 7.2
10 5.8
12 5.0
13 6.0
14 6.6
15 5.6
16 2.8
17 2.4
20 3.0
22 1.5
24 1.4

 

 

 

1Units of dyne cm-2

 

 

144

Table 2.12 Abundance Ratios from Different Studies.

 

Table 2.12
Abundance Ratios from Different Studies

 

N/O N/S Reference

 

~l.50 0.20 Garnett (1999) (for full LMC)

 

~ 1 .67 -0.26 Vermeij et al. 2002

 

-l.29 0.22 Peimbert (P03)

 

~l .52 0.11 Mathis, Chu & Peterson

 

~l .27 0.32 Tsamis & Pequngnot (TP05)

 

 

 

 

~1.16 0.41 Adopted Here

 

 

145

 

 

“W
,, , x

 

1‘“ m

"w ‘miifljllii'mi‘fiw . 1 1ll“ ‘ i l

' ‘ 4
, 23,-. «ll, M 't

_‘ “M.
‘llhi’hlll. b l

 
   

(b)

 

 

V iiilir ‘

 

Figure 2.1. - New narrow band Ha image of 30 Dor from the SOAR telescope rotated 13 degrees. The
center of R136 is marked as a white cross. a) the outline of the region covered by our maps made from the
Blanca spectra; b): The individual slit positions of our Blaco spectroscopic data set.

146

”fin", :

    

1 arcmin

Figure 2.2. ~ Ratio of SOAR I-Ia/ [S 11] images where darker indicates a lower ratio. The orientation is the
same as Figure 2.1. For reference the SOAR spectroscopic slit positions are plotted on top of the image.

147

 

    

Sky su'btracted '—
1.6 - “ Nebula + Sky ------- "
1.4 - -- -
[01] 6300A
[o H] 7330A ~
[0 II] 7320A -

F). x 1016 (erg s_1 cm_2 arcsec'2 A‘l)

 

 

 

 

6260 6280 6300 6320 6340 7300 7320 7340 7360 7380
Wavelength (A)

Figure 2.3. - A sample SOAR spectrum near [0 1] 71.6300 and [0 11] 1173207330. The solid and dashed
lines respectively show the spectrum after and before sky subtraction.

148

 

 

 

02>
00w0060£1giisfifif-

 

 

—02 - . (c)
6660 6680 6700

 

 

Figure 2.4. - A demonstration of the source of the line profile shapes with a wide slit width. Panel a is a 2D
image of the sky in Ha emission. The box represents the region on the sky the spectrum in panel b was
extracted. Panel b is the 2D spectrum at the same scale as panel a. Panel c is the line profile for the region
extracted. The solid line shows the data of the He 1 2.6678 at the LMC redshift. The top dashed line is the
best Gaussian fit and the bottom is the residual of that fit.

149

 

1.8 I I I I I I

1.6 - [s 11] 6717A -

1.4 _ [311'] 6731A

p—
N
I
1

pa
I
l

.O
00
I
1

He 1 6678A -

.9
ON
I

F). x 1014 (erg s'1 cm‘2 A4)

.o
A

I
A;

g:
N

 

 

 

 

 

r ---------- -l
L l l

0
6640 6660 6680 6700 6720 6740 6760 6780
Wavelength (A)

 

 

 

 

Figure 2.5. ~ TWO examples of extraction windows used to measure line flux. He 1 16678 shows an isolated
emission line. The [S 11] doublet is slightly blended. Between the two peaks of the [S 11] indicated by a
horizontal bar, a search is performed for the minimum to define the wings of each emission line as
described in the text.

150

 

[o I'm/Hp '———
[s II]/Ha ---------

0.8 -

0.6 -

N/N tot

0.4 L

 

 

 

 

Figure 2.6. - Repeatability of results at overlapping points along different Blanco slit positions. The curves
are histograms of the distributions of ratios of intensity ratios; for example the [O III]/HB measured from
one slit position is divided by the [O III]/HB measure at the same position on the sky but from a different
slit position.

151

30 Doradus Log(S(Hoc))

 

 

 

 

 

~10.5
15° 7 -11.0
§ 100 _ 115
o ' .
5 50 -
45,-; 0 — ‘1: L'r' ”12.0
35 u:
o -50 _ -12.5
8 100
' -13.o
~150 -
-13.5

300 200 100 0 ~100 ~200 ~300
RA offset (arcsec)

Figure 2.7a - Interpolated dereddened Ha surface brightness in erg s'1 cm-
the same as that outlined in Figure 2.1 and figures 7b — 7g.

2 arcsec'z. The region shown is

 

30 Doradus
I l l 1

Dec offset (arcsec)

 

 

 

 

 

300 200 100 0 ~100 ~200 -300
RA offset (arcsec)

Figure 2.7b ~ Interpolated Av,

152

30 Doradus Log([O “ll/(Hm)
0.75

 

0.70
0.65

i i 0.60

y offset (arcsec)

0.55

 

 

 

0.50

 

300 200 100 0 -100 ~200 ~300
x offset (arcsec)

Figure 2.7c - L08([0 [In/HIS).

 

 

 

 

 

 

30 Doradus Log([N ||]/Ha)
| I I I I I
150 _ -O.60
g 100 _ -O.80
g 50 — ~1.00
3
.82 0 ‘ -1.20
”4: ~50 —
g -1.40
~100 —
~1.60
~150
I I I I I I “1.80

300 200 100 0 -100 ~200 ~300
x offset (arcsec)

Figure 2.7d _ Log([N III/Ha)

153

30 Doradus [S ||l]/[S II]

 

 

 

 

 

 

-0.60
33‘ ~0.80
(D
2
g; 1 .1 -1.00
am) I
5
> -1.20
-1.40
300 200 100 O ~100 ~200 ~300
x offset (arcsec)
Figure 2.7e — Log ([5 111] 16312 / ([s 11] 16716+16731»
30 Doradus Logf [S ||]/H0t)
150 _ -O.60
g 100 — ~0.80
g 50 _ ,\“j\<‘: '1.00
5.“,
.5 0 — -1.20
s -50 _
: ~1.4O
~100 —
~1.60
~150 —
~1.80

 

 

 

 

300 200 100 O ~1OO ~200 ~300
x offset (arcsec)

Figure 2.7f — Log ( ([s 11] 16716+16731)/ Ha).

154

3O Doradus

Log(ne)
3.0

 

_L

_L

O 01
O O
I I

01
O
I

Dec offset (arcsec)
O
I

—'~ .
c> tn
c> c:
I

-150 -

 

  

1

2.8
2.6
2.4
2.2

 

2.0
1.8

 

1.6

 

300 200 100 0 -100

RA offset (arcsec)
Figure 2.7g - Log ne in cm-3 measured from the [S 11] 3671606731 ratio.

30 Doradus

~200

    
  
 

 

_L_l.
01001
000

Dec offset (arcsec)
0

—'~ .
o 01
o o

-150

 

 

 

 

300 200 100 0 ~1OO
RA offset (arcsec)

~200

-300

Figure 2.7h - Temperatures measured from R defined in equation 5. The scale is shown in increments of

1,000K.

155

 

 

 

 

 

Te( 1000K)

 

 

 

 

 

 

 

 

 

. . I . I . I . I . I
~150 ~100 ~50 0 50 100 150 ~150-100 ~50 O 50 100 150

offset slit center (arcsec) offset slit center (arcsec)

Figs. 2.8a — 2.8d — The temperature and density profiles for our Blanco slit positions. From top to bottom
Figure 2.8a shows positions 1-9. These figures are further described in the text.

156

 

~3

Logmnecm

 

 

 

 

 

Te( 1000K)

 

3.0

 

 

 

 

 

3.0

 

 

~150 ~100 ~50 0 50 100 150 ~150 ~100 ~50 0 50 100 150
offset slit center (arcsec) offset slit center (arcsec)

Figure 2.8b - Temperature and density along Blanco slit positions 10 — 17 and position 20.

157

 

5”
c

2.0

4.0

2.0

' 4.0

3.0

2.0

4.0
3.0
2.0

4.0

2.0

Log", ne cm‘3

Tc(1000K)

   

 

I I I
I I I

   

lll'll'l

 

j'j'jf

  

llllll
I I

 

I'I‘I‘

  

 

I I I
I I l

 

 

r

'_Irl

 
  

 

 

 

 

 

l I I

 

 

 

 

~150 ~100

~50 0 50
offset slit center (arcsec)

100

150 ~150 ~100

~50 0 50 100 150
offset slit center (arcsec)

Figure 2.8c — Temperature and density along Blanco slit positions 21-29.

158

4.0
3.0
2.0

4.0
3.0
2.0

4.0
3.0
2.0

 

-3

Logm ne cm

 

I'I'I'

 

N
I'I‘I'

 

 

5
I'I'l

 

 

   

 

   
     
   

 

 

 

 

 

 

 

A 16
g 14 r j— - 3.0
_ 12 _' T- - 2.0
[:6 10 - -
16 L ~. ' I . I I I I I I I I I ‘ 4.0
14 r 1~ - 3.0
12 '- —' ‘
_ I. q— - 2.0
10 - -‘
16 + i i 'r I ' I i i I . 4-0
14 1— - 3.0
12 - 1 -' ‘
- I.' I. 4- ~ 2.0
10 .. fl . | ..
16 _i ' i i I ifi ' I i I :11 i i film I I T i _ 4‘0
14 _— I1- — 3.0
12 - -' ‘
. -- - 2.0
10 - —
16 I I I I I I I I I I I . I I ' I I I I I I I 4.0
14 '_ P0538 I—I—I 1 Pos 38 |-—+—-l '
. r - 3.0
12 — 4' ‘
- — + - 2.0
10 - + I
J . I I I I l I I . J LJ I I l I I I . I
~150 ~100 ~50 0 50 100 150 ~150 ~100 ~50 0 50 100 150

offset slit center (arcsec)

offset slit center (arcsec)

Figure 2.8d — Temperature and density along Blanco slit positions 30-38.

159

~3

Logm ne cm

q Min 2:; -.

‘ «Wk L j."
1 i I I II ' ‘1 1' i 1‘:

‘ 4

. ‘, ’__ .‘ “I
- muffin" ‘
I" I. 1!.

It}. 's‘ . - .
i '9 I ( I I'
7“ shifts“? 'K
’ I

 

Figure 2. 9- Prominent ionization fronts listed in Table 2. 7, drawn on the [S III/Ha ratio image. Top—left: A
blow up of the central region around R136 including IFs 1 and 2.

160

 

Figure 2.10 - Left SOAR [S II]/I-Io.; Right SPITZER 8pm PAH. A selection of bright pillars are shown
with arrows indicating their location and direction. These Dense IF are detected in both optical and IR
passbands and show a connection with the background molecular cloud.

161

 

 

 

I T I I fl
1 .4 ‘- Ha ‘
[0111] 5007 -------

12 " [$111] 9069 -------- ‘

 

 

 

 

 

 

 

 

 

 

 

 

 

80
T I I
-I
-I
80
50 55 60 65 7O 75 80

Projected distance from R136 (arcsec)

Figure 2.11 ~ The profile of IF1. Top: Ionized gas is traced by Hat, [0111] and [S III] emission. Middle: The
IF is traced by [S 11], [N 11] and [0 11]. Bottom: The electron density profile measured from [S 11] as
described in the text.

162

 

 

   

 

 

 

 

 

 

0.0 I I I 10 l l I I
TP05 Costar — ,,
38.5kK ------------
0.8 - I
~ “‘3‘
E. a 0.6 -
.8 g
E E 0.4
0 b0
- 3
0.2 - _
l I l l 0.0 L 4 4 J 1
~08 -0-4 0-0 0-4 —2.4 .20 ~16 -1.2 08
[SII]/[SIII] Log10([SII]/HOI)
4_() . . . . 1.0

 

 

Log10([OII]/ [01])
Log10([OIII]/HB)

   

 

 

 

 

~2.0 ‘ v ‘

A

 

 

-O.8 -04 0.0 0.4 0-0
Log10([SII]/[S 111])

-2.8 -2.4 .2.0 ~1.6 -1.2
Log10([OII]/H0t)

Figure 2.12 - Plots of diagnostic line ratios with observations from SOAR and Blanco spectra. Lines

represent photoionization models with ne 200 cm'3. Arrows indicate the effect increasing the modeled
SED temperature has on the line ratios. From top left to bottom right: (a) [SII]/[ 8111] vs. [0 II]/[O 111]; (b)
[SII]/Ha vs [0111]/118; (c) [SII]/[S 111] vs. [OII]/[OI]; (d) [O Ill/Ha vs [0 III]/HB.

163

 

 

38.5kK -------
Blanco

TP05 CoStar —

 

 

 

m
0.9

1.0

0.6 0.7 0.8
LogloflOIIH/HB)

0.5

0.4

 

 

-l.U

-l.5 -

_
0.
2

-2.5 IL

§3ch £5333

-3.0
0.3

Figure 2.13 - Predicted and observed [0 III] 14363/2.5007 ratios for models with TP05 abundances.

164

 

'1.0 T I I I I
New CoStar
Blanco

 

 

 

 

 

—1.5 — g . —
g 0
O
O
‘Q
m
\D
m
v
E 1
Q.
‘6
63
3
_3.0 I I . I . "I. I I
0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Log10([OIII]/HB)

Figure 2.14 - Similar to Figure 2.13 but with modeled with our adopted abundances.

165

 

 

   

 

 

 

 

 

 

 

 

   

 

 

 

 

 

 

"1-0 I l’ l 1.0 7 l I
New CoStar —— .
0.8
E
E E 0.6
9. o
E“ “2 0.4
g ' go 02
4.0 I " ' - '
- . - . 0.0 . - -
-03 -()_4 0.0 0.4 -2.5 -2.0 -1.5 -1.0 -0.5
[SII]/[SIII] LogioflSHl/Ha)
40 . r . . 10 1 .
A A 0.8 -
Z a.
o .
‘5“ E 0.6 ' -
<3. 9.
V2 ‘§ 0.4 —
DO 60
° 3
4 0.2 -
4.0 . . M 0.0 - - A
-0.8 -04 0.0 0.4 -3.0 -2.5 -2.0 -1.5 -1.0
Log10([SH]/[S 111]) LogmUOIIIIHa)

Figure 2.15 — Various predicted and Observed line ratios described as described in Figure 2.12 using our
adopted abundances.

166

 

 

 

 

 

 

I I I New toStar
glance + SOAR
0.8 _ V ..
V v v n
W. .'
a r,
0.6 - 97 -
E v
§
E
E” 0.4 - _
0.2 '- V _
0.0 l l I I l l
-4.5 —4.0 -3.5 -3.0 -2.5 -2.0 -1.5
Log10(He I/HOL)

Figure 2.16 - Predicted and Observed intensity ratio of He I A6678 / Hut.

167

 

New CdStar —
Blanco + SOAR

V

Log 10([OH11/Hfi)

 

 

 

0.0 1 I 1
-2.5 -2.0 -l.5 —1.0 -0.5

L0g10([NII]/Ha)

 

Figure 2.17 - Predicted and Observed intensity ratio of the commonly used [N II] 16584/Ha diagnostic.

168

 

150 -
100 -

01
O
I

y offset(arcsec)
61
O O
1 I

-100 -
-150 -

 

 

 

 

l l l I | l
300 200 100 O -100 -200 -300
x offset(arcsec)

 

Figure 2.18 - An interpolated map of the dimensionless quantity U, derived from fitting models to our
Blanco spectra. Region mapped is the same describing Figs. 2.7a — 2.7b.

169

50

Dec offset(arcsec)
O

IZI (pc)
300.0

 

 

 

250.0
200.0

‘ 150.0

 

100.0

50.0

 

 

300 200

2 (PC)

80

100

0.0
0 -1 00 -200 -300

RA offset(arcsec)

 

 

 

 

 

—10 -5 0

51015203303540
offset(pc)

Figure 2.19 — Top: Map of of |z| , defined in Eq. 12. Bottom: Profile of z for position 8. The offset in the
declination offset from R136. By definition the height of R136 is z = 0.

170

 

 

.l J‘J u a . _ _
151-III-
. in":
. IIIIII
r " lihnhuhllhll HHHH
. IDS-{Illa .
u rrrrrrrr
u uuuuuuuuuuuuu
III
ulllllul
ll 2 inunnuuuun llll
S S Pull-u nuuur
0 O
P P - ...
Illlluflfl‘lll
I II||
.Ial ......
uuuuuuuuuuuuuuuu .-
Llll
I. II
I Italic-Illa III
III” "l
r II'I II-
AI
Ilnluiatr
f 4 "

 

 

-20

100-
120-
140-

30

10 20

0

-10

-20

-30

-50 40

offset (pc)

 

l

Pos 1

 

Pos 2

 

 

-20

 

10 20 30

-40 -30 -20 -10 0

-50

offset (pc)

Figure 2.20 -Profile of |z| in slit positions 1 and 2. Top: |z| from the best fitting model; bottom : [2]

calculated from equation 13 assuming a smooth density distribution.

171

 

Line of sight Illuminated Face ' ' - " -

A ‘F

 

Geometric Discontinuities

    

Edge on IF

 

 

 

Figure 2.21 - A cartoon of possible geometries in 30 Doradus consistent with the changes in U across the
nebula. The region labeled “a” is a continuos ionization front of finit height, seen edge on, facing the

ionizing cluster. Region “b” represents possible geometries that would be seen as discontinuities in
modeled R or Izl.

172

150

.5
O
o

50

y offset(arcsec)
o

L0910(Pgas)

 

 

 

 

 

300

200 100 O -100 -200 -300
x offset(arcsec)

Figure 2.22 - Observed thermal gas pressure interpolated from the Blanco spectra.

173

 

-8.8
-9.0
-9.2
-9.4
-9.6

*- . -9.8
-10.0

-10.2
-10.4
-10.6
-10.8

L091O(Psta1/Pgas)

 

0.0
150 -0.2
100 -0.4

’8‘ -0.6

g 50 - -0.8

to

:5: 0 _ -,‘ 1| -1.0

‘0 " , -1 2

:t: .
-50 _

g -1.4
-100 - -1.6
-150 _ -1.8

-2.0

 

 

 

 

300 200 100 0 -100 -200 -300
x offset(arcsec)

Figure 2.23 — The ratio of pressure from integrated star light (Eq. 17) to gas pressure.

174

 

10g (U)

 

 

 

-3 —
—3.5 —
_4 I I I l I l I l l
l 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8
log necm'3

Figure 2.24 - Observed gas density vs modeled ionization parameter U.

175

 

150
100

01
O

y offset(arcsec)
<31
0 O

-100
-150

LO910(F,X-ray)

 

 

 

 

 

 

I l I I l I '9-0
WW III III _ -9.1
MI I“. I' "m I‘ I'
IIII‘ “IIIII III “III, ..
..II III“"'II\III“‘: III“ III“? II II d 9'2
III?“ III“ I I ‘ -9.3
“11‘“th “I III;‘ III :LIIU‘ II‘ _ 9 4
._ -9.5
" -9 6
WWII “III“:‘I'III -— I
“III“ lIII‘U'
II“ I - .7
I MW MW“ ‘1“ ;“I\ Wk“ III‘“::‘\I\I\“‘\\ ”I“ m. 9
ll“ “w“:tlh‘"““I‘I‘I‘II“W“:I‘I‘I‘lwm“m I“ “:II “mm“ -9.8
Illll‘mtlI‘I‘ellllY“\‘l‘lllllllll‘.“t|t\lllll‘\ul " l‘I‘I‘IIIIII‘l‘.\“ IIIIII‘I'IIIIIIII ..IIIII .. _ -9.9
I I I I I I _1 0.0
300 200 1 OO 0 -1 00 -200 -300

x offset(arcsec)

Figure 2.25 - Px-ray from the regions of diffuse x-ray emission described in Townsley et al. (2006). The

pressure was calculated using the reported Tx-ray, surface brightness and area.

176

 

Figure 2.26 — The region discussed in section 6.3.4 demonstrating an outflow. The contours in both left and
right represent the soft, diffuse x-ray emission. Left: SOAR [S 11]; Right:0.5 — 0.7 keV emission. The x-ray
data was made available by Liesa Townsley.

177

Log PX-ray/Pgas

 

           

 

I I I I I I 1-5
_ IIIIIIII'III 1 ‘ _
150 IIIIIIWI 1.0
o 1
a: 1 ‘ 11
g 50 — 1 1 1.; - 0.5
5! IIIIIIIIIIIIIIIIIIII N
° - ~ « III “I“ ~ ‘ 00
.1 I 1 . 1 III III" _
”5 -50 — 1 1. ~ _
>
-100 - — _0_5
-15o —
I I I I I 4.0

 

 

300 200 100 0 -100 -200 -300
x offset(arcsec)

Figure 2.27 — The ratio of Px.my to Pgas.

178

so Doradus Log([N l|]/[S ”D
j l I l l l

 

 

0.30
150 -
0.20
g 100 "‘
g 50 _ 0.10
as
v .q 0.00
E ° ~
g _50 — “'1 '0.10
>.
-100 — -0.20
-150 - -0.30

 

 

 

L I I L I I
300 200 100 O -100 -200 -300
x offset (arcsec)

 

Figure 2.28 - The spatial distribution of the [N 11] 16584/ [8 11] (2.6716 + M731) ratio.

179

‘
‘.

References

Adelman—McCarthy, J. K., et al., 2006, ApJS, 162, 38
Baldwin, J. A., Phillips, M. M., & Terlevich, R., 1981, PASP, 93, 5 (BPT)

Bresolin, F., Gieren, W., Kudritzki, R., Pietrzynski, G., Urbaneja, M. A., Carroro, G.,
2009, ApJ, 700, 309

Cardelli, J. A., Clayton, G. C., Mathis, J. S., 1989, ApJ, 345, 245
Chu, Y., and Kennicutt, R. C. 1994, ApJ, 425, 720

Crowther, P. A., 2007, ARA&A 45, 177

Davis, L., Jr., & Greenstein, J. L., 1951, ApJ, 114, 206

Dopita, M. A., Fischera, J ., Sutherland, R. S. 2006, ApJ, 167, 177

Dobashi, K., Bernard, J. P., Paradis, D., Reach, W. T., Kawamura, A., 2008, A&A,
484,205

Garnett, D. R. 1999, IAUS, 190, 266

Heap, S.R., Lanz, T. & Hubeny, I. 2006, ApJ, 638, 409

Hu, E. M., Cowie, L. L., Kakazu, Y. Barger, A., 2009, ApJ, 698, 2014
Indebetouw, R., et al. 2009, ApJ, 694, 84 ( 109)

Kewley, L. J., Geller, M. J., Jansen, R. A., & Dopita, M. A. 2002, AJ, 124, 3135
Kewley, L.J. & Ellison, S.L. 2008, ApJ, 681, 1183

Krabbe, A. C., and Copetti, M. V. F, 2002, A&A, 387, 295

Landi, B., & Landini, M., 1999, A&A, 347, 401

Lazendic, J. S., Dickel, J.R., & Jones, P. A., 2003, ApJ, 569, 287

Lebouteiller, V., Bemard-Salas, J ., Brandl, B., Whelan, et al., 2008, ApJ, 680, 398

180

Marci, M. L., Stanek, K. 2., Bersier, D., Greenhill, L. J ., Reid, M. J., 2006, ApJ,
652,1133

Massey, P., et al., 2009, ApJ, 692, 618
Mathis, J. S., Chu, Y.H. and Peterson, D. E., 1985, ApJ, 292, 155
Meixner, M., et al., 2006, AJ, 132, 2268

Nakajima, Y., Kandori, R., Tamura, M. Kusakabe, N. Hashimoto, J ., Nagayma, T.,
Nagata, T., Hatano, H., Kata, D., Hough, H., 2007, PASJ, 59, 519

Osterbrock, D.E. & Ferland, G. J ., 2006, Astophysics of Gaseous Nebula and Active

Galactic Nuclei 2nd Edition, Sausalito, California: University Science Books
(AGN3)

Oey, M. S., Dopita, M. A., Sheilds, J. C., Smith, R. C., 2000, ApJS, 128, 511

Pagel, B. E. J., Edmunds, M. G., Blackwell, D. E., Chun, M. S., & Smith, G., 1997,
MNRAS, 189, 95

Pagel, B. E. J., Simonson, E. A., Terlevich, R. J. & Edmonds, M., 1992, MNRAS, 255,
325

Peimbert, A., 2003, ApJ, 584, 735 (P03)

Pellegrini, E. W., Baldwin, J. A., Ferland, G., to submitted MNRAS, 2009
Pettini, M., Ellison, S.L., Bergeron, J & Petitjean, P. 2002, A&A, 391, 21.
Pettini, M., Pagel, B., 2004, MNRAS, 348, L59

Poglitsch, A, etal., 1995, ApJ, 454, 293

Rosa, Michael, & Mathis, J. S. 1987, ApJ, 317, 163

Schaerer, D., 1996, A&A, 310, 837

Scowen, P. A. et al., 1998, AJ, 116, 163

Smith, R. C., Points, S. D., Chu, Y.H., Winkler, P.F., Augilera, C., Leiton, R., MCELS
Team, 2005, AAS, 207, 2507

181

 

....

1.

Townsley, L. K., et al. 2006, AJ, 2006, 131, 2140

Tsamis, Y. G., Barlow, M. J ., Liu, X. W., Danziger, I. J ., Storey, P. J., 2003, MNRAS,
338, 687

Tsamis, Y. C., Pequignot, D., 2005, MNRAS, 364, 687 (T P03)

Vacca et al. 1996

Veilleux, S., & Osterbrock, D. E., 1987, ApJS, 63, 295

Vermeij et a1. 2002

Vilchez, J. M., & Page], B. B., 1988, MNRAS, 231, 257

Walbom, N. R., Mais-Apellaniz, J ., Barba, R. H., 2002, AJ, 124, 1601

Weingartner, J .C., & Draine, B.T., 2001, AJ, 548, 296

Wen, Z. & O'Dell, 1995, ApJ, 438, 784

Yan, T. T., Kewley, L. J ., 2009, ApJ, 699, 161

Yin, S. Y., Liang, Y. C., Hammer, F ., Brinchmann, J ., Zhang, B., Deng, L. C., & Flores,

H., 2007, A&A, 462, 535

182

O
,f!

Chapter 3
Spectroscopic Observations of NGC 3603

l N GC3603 — A Large Star-Forming Region in the Milky Way

NGC 3603 is a large, yet compact Galactic HII region with dozens of OS to O4 ( M > 35

Mo) stars within a 1 pc diameter. It clearly is one of the most luminous star-forming
regions in our own Galaxy. The cluster mass has been estimated to be as high as 10!5 Mo,

half that of R136 with a luminosity equal to 6.1 x 105 Lo (Eisenhauer et al. 1998). A
recent spectroscopic study of 26 cluster members by Melena et al. (2008) place the
nebula at 7.6 kpc, more than 6 times closer than 30 Doradus. In addition to its proximity,
the age of the massive stars studied by Melena et al. (2008) suggest the cluster age is
between 1-2 Myr. Because of it's youth, the nebula has not yet been disturbed by
supernovae and is likely to present a less complex example of star formation than 30
Doradus. The region is further simplified because there are no additional sources of
ionization competing with the central cluster, unlike the case of 30 Doradus.

Empirical abundance from optical studies show the log (O/H ) abundance to be -3.29
(Garcia-Rojas et al. 2008), almost 3 times greater than our value for 30 Doradus. In
addition to 30 Doradus, Lebouteiller et al. (2008) also obtained abundances for S, Ar, Ne
and Fe in N GC 3603 at 7 locations. On average, the abundance of these elements are 0.2
dex greater in NGC 3603 than in 30 Doradus with a scatter of about 0.08. These factors
make NGC3603 a logical intermediate step in our study of H 11 regions as a function of
size and abundance as we work our way toward GEHIIRs.

However an optical study of NGC3603 must face a number of challenges. The region

183

lies in the disk of the Milky Way and has an average extinction Av: 5.23 (see Table 3.3
below). Additional H II regions along the line of sight contaminate any study which, like

ours, does not have sufficient velocity resolution to separate the various components.

The extinction makes detecting a faint auroral line like [0 III] M363 in our survey
impossible in all but the brightest parts of the nebula. Without a measure of the global gas
temperature we cannot directly apply the technique for determining the metal abundances

that we used to study 30 Doradus.

In keeping with our original NOAO observing proposal, the spectroscopic grid of N GC
3603 will also be made available to the public along with the reduced data and calibration
frames. Due to the additional complications regarding NGC 3603, these data will be
presented in a short journal article describing the observations and a few basic results that

can be obtained without extensive Cloudy modeling.
2 SOAR Imaging

In the same way as was done for 30 Doradus, we took a series of narrow band images
with the SOAR Telescope in order to map the spatial structure of N GC 3603 at sub-
arcsec resolution. For present purposes, we present mosaic images (Figures 3.1, 3.2 and
3.3) constructed from these observations in two of the band-passes to serve as high-
resolution finding charts to show the slit positions of the spectroscopic observations
described below. The mosaics presented represent a combination of 4 fields spanning an
area 5.3 across. A journal of the full set of SOAR images is presented in Table 3.1. The
filters used were a combination of those available through CTIO ( 6850 red continuum,

6563x75, 5130 green continuum) and those generously provided by Frank Winkler

184

( 6734, 6572 and 5007). As the filters became available over a 2 year time period we
observations during three runs: April 4 — 8 2007, Jan 6 8r 7 2008 and April 16 & 19. A
continuum subtraction of each filter to account for both stellar and scattered light was
done in accordance with the description of the 30 Doradus images. The surface brightness
profile of each of slit position was checked against these direct images using a
combination of emission line and stellar band-passes. The RAs and Declinations reported

in Table 3.2 were computed using such a comparison and are accurate to 1 arcsec.
3 Spectroscopic Observations with the Blanca Telescope

A full grid of spectra of NGC 3603 were obtained during two observing runs, one on 22
March 2007 and one 8-9 Feb 2008. The NGC 3603 spectra were part of the same
program as the 30 Doradus survey, and used the same instrument setup described in
section 2.3.1 of chapter 2 of this thesis. The data were reduced using the calibration steps
and extraction processes for emission lines that have already been described for 30

Doradus.

It should be noted that there are many fewer detected and extracted lines in the case of
NGC 3603 than for 30 Dor, due to the lower apparent surface brightness of NGC 3603.
The lines that are strong enough to be extracted automatically are shown in Table 3.3.
These are presented at their rest wavelengths because NGC 3603 is essentially at zero

velocity with respect to the Sun. Following the format of Tables 2.4 and 2.5 of this thesis,
all reddened line fluxes are presented relative to the dereddened HB strength. The lines

listed here in Table 3.2 include HB, Ha and the [S II] M 6716,6731 doublet. From these

185

 

 

we measured Av and the electron density. Both of these quantities are included here in

Table 3.4. The average extinction in the nebula is Av = 5.23 $1.11, without correction for

outliers effected by the considerable number of field stars. Melena et a1 (2009) report an

E(B-V) of 1.394 :0.012 for the stars in the cluster, corresponding to an AV: 4.31.
Position 8 passes near to the cluster but is far enough that underlying stellar absorption

features do not effect the measured emission line strengths. At it's closest approach of 19

arcsec we find Av = 4.27. Tables 3.2-3.4 and the reduced 2D spectra are all available

electronically to the public on the same permanent web site where we have posted the 30

Dor data 1.

There are two common emission line diagnostic diagrams available in in Table 3.4 used

through out the study of 30 Doradus. These are shown below in Figures 3.4 and 3.5
without a detailed analysis. Figure 3.4 is the diagnostic ([8 II] M6716+6731)/HOI vs. [0

III]/HB. Figure 3.5 is the [N lI]/Hoc vs [0 III]/H[3. Just as in the case of 30 Doradus a
large range in the ionization state of the gas is visible in the observations. In an upcoming
paper we will apply the results from 30 Doradus and combine the existing empirical
abundances in the literature with photo ionization models to measure the ionization
parameter. This will require data to constrain the gas temperature. Some observations
may be found in the literature for a small number of positions, but given the importance
of the [0 III] abundance to the gas temperatures in 30 Dor, additional observations are

likely required to proceed.

 

1 http/www.pa.Insu.edu/astro/thesis/pellegrini/ngc3603/

186

 

Table 3.1: The direct imaging observing journal for NGC 3603. These
observations span 2 years and the exact number and duration of exposures in
each 4 fields vary slightly by field.

 

 

 

 

 

 

 

 

Table 3.1
DETAILS OF NARROWBAND OBSERVATIONS OF NGC3603

Filter No. of Exp x Duration No. Positions FWHM Min/Max

at each position (arcsec) Exposure(s)
6563 3 x 3005 4 0.7 BOO/4000
6572 3 x 3005 4 0.5-0.8 590/3200
6734 7 x 6005 5 0.85 4000/14000
6850 (Number of exposures 5 0.5 -1 900/5700

matched to each read frame)

5007 3 x 2005 4 0.7-0.8 400/1800
5130 3 x 3005 4 0.7-0.8 600/2400

 

 

 

 

 

 

 

187

Table 3.2: The coordinates of the NGC 3603 slit positions with position

angle and total exposure time, excluding short exposures used to fix
saturation.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 3.2 Journal of NGC 3603 Spectroscopic Observations
Pos RA. (2000) Dec. (2000) PA (deg E of N) Exposure
Time (s)
1 11:15:29.45 -61:18:17.46 57 750
2 11:15:26.36 -61:17:32.62 57 600
3 11:15:18.90 -61:17:28.03 57 900
4 11:15:14.99 -61:17:13.20 57 600
5 11:15:13.39 -61:16:41.09 57 900
6 11:15:10.79 -61:16:28.09 S7 450
7 11:15:07.96 -61:16:25.27 S7 900
8 11:15:03.38 -61:16:20.31 57 600
9 11:15:02.79 -61:16:01.81 57 800
10 11:15:02.79 -61:16:01.81 57 900
11 11:15:03.44 -61:15:34.46 57 900
12 11:15:01.07 -61:15:03.31 57 900
13 11:14:58.46 -61:14:41.34 57 900
14 11:15:27.72 -61:13:27.59 57 600
15 11:15:18.38 -61:11:53.69 57 630
20 11:15:28.48 -61:16:23.42 147 600
21 11:15:20.86 -61:15:32.58 147 900
22 11:15:16.51 -61:15:43.90 147 900
23 11:15:21.61 -61:17:04.51 147 550
24 11:15:13.52 -61:15:55.11 147 900
25 11:15:14.81 -61:16:38.09 147 500
26 11:15:07.71 -61:15:43.98 147 900
27 11:15:07.01 -61:16:02.37 147 500
28 11:15:07.36 ~61:16:24.85 147 450
29 11:14:53.93 -61:16:46.67 147 900

 

 

188

1‘.

 

 

 

Table 3.2 (cont'd)

 

 

 

30 11:14:51.16 -61:17:OO.13 147 1350
31 11:15:18.62 -61:16:33.Sl 147 300
32 11:15:13.00 -61:14:05.72 147 630

 

 

 

 

 

 

 

Table 3.3: Emission lines identified in the NGC 3603 survey.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 3.3 NGC 3603 Line IDs
M ion {HR = 3.1)
4341 H I 1.35
4800 Continuum 1.19
4861 H I 1.16
4959 [0 HI] 1.13
5007 [0 HI] 1.12
5650 Continuum 0.97
5875 He I 0.93
6300 [O I] 0.86
6312 [S III] 0.86
6548 [N 11] 0.82
6563 H I 0.82
6584 [N II] 0.81
6678 He I 0.80
6716 [S 11] 0.79
6731 [8 II] 0.79
7065 He I 0.74
7135 [Ar III] 0.73

 

 

 

 

 

189

Table 3.4: A portion of the reddening-corrected NGC 3603 Blanco spectroscopic survey in table form as
described in the text. The first several rows of slit position 1 are shown. Together there are 28 unique slit
positions with line-strength measurements at a total of 2950 extracted locations within a 6x6 arc min area.

The units of electron density are cm'3. The dereddened HB surface brightnesses are reported in units of erg

s.1 cm'2 arcsec-2. The other emission lines are reported as 100><S(line)/S(HB).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

EXAMPLE OF PUBLICLY AVAEEABEIEESCI‘ENCE PRODUCT FROM BLANCO
SPECTROPHOTOMETRIC SURVEY OF NGC 3603

SlIt AR

# A ADec row log(ne) -err +err Av F(Hb) I(Hb) sig 4341 sig 4959 sig
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
1 31.6 -240.9 3 <1.00 1.00 1.91 5.49 1.33E-15 5.21E-13 4.47E-14 64.9 32.9 36.00 6.38
1 33.8 -239.6 8 <1.00 1.00 0.00 4.28 2.12E-15 2.23E-13 9.65E-15 0.54 0.13 37.79 3.32
1 35.9 -238.2 13 1.76 0.47 0.24 3.9 2.39E-15 1.65E-13 6.34E-15 0.57 0.11 31.13 2.97
1 38.0 -236.8 18 1.83 0.33 0.20 4.05 2.38E-15 1.94E-13 7.44E-15 0.49 0.11 32.97 2.94
1 40.1 -235.5 23 1.37 0.37 0.41 4.47 1.99E-15 2.55E-13 1.17E-14 0.40 0.15 37.59 3.46
1 42.2 -234.1 28 1.49 0.49 0.34 4.37 2.21E-15 2.55E-13 LOSE-14 0.43 0.13 38.48 3.09
1 44.3 -232.7 33 1.49 0.49 0.32 4.48 2.23E-15 2.89E-13 1.17E-14 0.58 0.13 38.79 3.05
1 46.5 -231.3 38 1.86 0.23 0.16 4.36 2.46E-15 2.81E-13 1.04E-14 0.47 0.12 39.64 2.77
1 48.6 -230.0 43 1.91 0.20 0.14 4.42 2.55E-15 3.12E-13 1.11E-14 0.40 0.11 47.03 2.67
1 50.7 -228.6 48 2.16 0.11 0.09 4.52 229151-15 311131-13 1.22E-14 0.50 0.13 58.25 2.97
1 52.8 -227.2 53 2.21 0.12 0.10 4.43 2.12E-15 2.63E-13 1.11E-14 0.37 0.14 36.88 3.18

 

190

 

SOAR Hcr + [N II]

1 arcmin = 189 pc

 

Figure 3.1: Narrow-band Ha mosaic image of NGC 3603, from observations using the SOAR telescope.
The NGC3603 Blanco slit positions are shown with labels. Slits with position angles PA = 147 deg are
labeled 1-15. Observations with PA = 57 deg are labeled 20—32. A 1 arcmin scale bar is shown along with
the projected physical scale assuming a distance of 6.5 kpc.

SOAR Hu +[N II]

IaI'chn = 1.89 pc

 

Figure 3.2: The same HOI mosaic image of NGC 3603 as in Figure 3.1, without the observed slit positions
super imposed.

192

‘1‘] II ' ' “'_ l
t , -. “l" ‘11," I“VI

_ 1112'
' ., . . . , 7 IIIIIIW““"‘ ‘
with “ ' ._" , . ‘ ‘ , 1w I-IIIIIIIII'.

1“” I

Figure 3. 3: An [S II] mosaic coverIng the same field as Figure 3.1, also created from
observations using the SOAR telescope. Two predominate pillars, P1 and P2 are indicated.
These were observed extensively in our spectroscopic survey, with multiple position angles.

 

193

 

[o III]/Hfi

 

 

 

 

' -2.o -l.5 4.0 -0.5 0.0
[s II]/Ha

Figure 3.4: The emission line diagnostic ([S 11]
6716+6731)/Hot vs [0111]/FIB from the NGC 3603
Blanco data. A higher degree of ionization
corresponds to lower [5 II]/Ha and high [0 III]/HB.

 

 

 

 

 

1.5 I I I
+ + I +
1.0 - 'I
Q
E
E 0.5 -
9 .
0.0 ~ * -
‘0-5 1 I l
-2.0 -l.5 -l.0 -0.5 0.0

[N II]/HOL
Figure 3.5: The emission line diagnostic plot [N [I]
6584/Ha vs. [0 HIM-{B from the NGC 3603 Blanco
data. A higher degree of ionization corresponds to
lower [N II]/Hu and high [0 IIIJ/HB. This is
analogous to Figure 3.4.

194

References
Eisenhauer, F., Quirrenbach, A., Zinnecker, H., Genzel, R., 1998, ApJ, 498, 278

Garcia-Rojas, J., Esteban, C., Peimbert, M., Costado, M. T., Rodriguez, M., Peimbert, A.,
Ruiz, M. T., 2006, MNRAS, 368, 253

Lebouteiller, V., Bernard—Salas, J ., Brandl, B., Whelan, D. G., Wu, Yanling,
Charmandaris, V., Devost, D., Houck, J. R., 2008, ApJ, 680, 398L

Melena, N. W., Massey, P., Morrell, N. I., Zangari, A. M., 2008, AJ, 135, 878

195

          

VE

mlllllflfiifllfllflllfilflljlfllflllflflflfllillm