FE II EMISSION FROM HI1 REGIONS AND

ACTIVE GALACTIC .NUCLEI

Ekaterina Vemer

A thesis submitted in confonnity with the requirements

for the degree of Doctor of Philosophy

Graduate Department of Astronomy

University of Toronto

S Copyright by Ekaterina Verner uXM

u i s i i s and Acquisitions et Bib iographic Services services bibbg~phipues "f

The author has granted a non- exclusive licence allowing the National Library of Canada to reproduce, loan, distriiute or seii copies of dus thesis in microform, paper or electronic formats.

L'auteur a accordé une licence non exclusive permettant a la Biblioth&que nationale du Canada de reproduire, prêter, distri'buer ou vendre des copies & cette thèse sous la fonne de microfiche/film, de reproduction sur papier ou sur format électronique.

The author retains ownership of the L'auteur conseme la propriété du copyright in this thesis. Neither the droit d'auteur qui protège cette thèse. thesis nor substantial extracts fiom it Ni la thèse ni des extraits substantiels may be printed or otherwise de celle-ci ne doivent être imprimés reproduced without the author's ou autrement reproduits sans son permission. autorisation.

FE II EMIÇSION

FROM HI1 REGIONS AND ACTIVE GALACTIC NüCLEI

Ekaterina Verner

Doctor of Philosophy, 200

Department of Ashonomy, University of Toronto

ABSTRACT

This thesis focuses on the elaboration of numerical models of Fe II emission

spectra and applications of these models to the interpretation of new observational

data. A mode1 of the Fe II atom, wh.ich indudes the lowest 371 energy levels (up

to 11.6 eV) and predicts intensities of 68635 lines, is first developed and then

incorporated into the radiative-collisional code Cloudy. The atomic data and

numerical methods used to determine level populations are descriid. The basic

equa tions of ioniza tion and thermal balance, level populations, and radia tive

tramfer are solved self<onsistently. Test cases show that the atom goes to local

therrnodynamical equüiirium in the limits of high particle and radiation densities.

The quantitative models have been applied to sunulate emission spectra of

H II regions. The general behavior of the Fe II emission lines and their semitivity

to d d t y and radiathe pumping conditions are investigated. New spectroscopie

observations of the Orion Nebula obtained at the Cerro Tololo I n t e r - A e a

Observatory are presented. The spectral range is 3498 - 7468 A. The specha are

high resolution and sensitivity and remlt in the most extensive line atlas of the

Orion Nebuia yet made. More than 400 emission iines have been identified,

inciuding 40 forbidden F e lines. Our theoretid mode1 of Fe II emission is in

gwd agreement with the observational data. The velocity field in the Orion

Nebula, which shows a marked dependence on the ionization structure, has been

analyzed.

An o v e ~ e w is presented of general features of the Fe II spectra of typical

quasar Broad Emission Line Regions and the dependence of these features on the

basic parameters (density, flux of radiation, microturbulent velocity, the Fe

abundance, and Lya pumping). Calcuiations of Fe II emission for grids spaMing

several orders of magnitude in density and ionizing flux are performed in the

framework of "locally optimally-emitting douds." It is shown that strong

seledion are at work.

Due to the richness of the Fe II spectra and their sensitivity to the excitation

parameters, the Fe II emission lines provide a powerful tool for diagnostics of

physical conditions and Fe abundance in a variety of astronomicai objects.

ACKNOWLEDGEMENTS

First of al l 1 greatly appreciate the guidance of my thesis supervison Gary

Ferland and Peter Martin.

1 would like to thank Gary Ferland forhis help in modifying his photo-

ionization code Cloudy. Without all of this support that 1 had al1 the four years,

this thesis would not have been possible. 1 thank Peter Martin for careful reading

and most useful recommendations and advices. 1 am graiefui to Jack Baldwin for

new observations of the Onon Nebula, which we used for the applications of our

Fe II emission models. 1 acknowledge collaboration and fiuitfui discussion with

Kirk Korista, Fred Hamann, Jason Ferguson, Peter van Hoof and Dima Vemer. 1

extend my thanks to the cornputer group for all of their support in Lexington,

Kentucky. 1 would like to thank John Connolly for his consistent great interest in

advancing forefront research in astrophysics. I thank Chu& Fisher and the late

Shashi Shitae who have helped me on many parallel - coding problems. 1 am

grateful to Lillian Lanca for her kind help. 1 thank the Hubble Space Telescope,

the Center of Computer Science at the University of Kentucky and the

Department of Astronomy at the University of Toronto for their support through

l3-h-

TABLE OF CONTENTS

1 . htrduction ........................................................................................... 1 1.1. Background ..................................................................................... 1

1.1.1. Fe II emission lrom H II regions ..................... .. .................... 3

1.1.2. Fe II emission Gom Active Galactic Xud& ............................... A

1.1.3. Fe II emission lrom steiiar sources ............................................... 7

1.2. Thesis outline ................. .. ..... ........... ................................. 8

.................................................. 2 . The Fe II Mode1 and Atomic Data 11

2.1. The mode1 atom and its atomic data.. ...................................... 11 2 2 . Balance equations .................................................................... 15

2.3. Line transfer .................................. .......................................... 16 2.4. Heating-cooling balance ............................................................ 18

2.5. Interaction with the environment: tests of approach to local

thermodynamical equilibrium ..................... .... .... d 2.5.1 . Cloud structure: collision-dorninated case ................................. 20

97 ........................................................... 2.5.2. Radiation-dorninated case.. - 2.5.3. Large op tical dep ths .......................... .. ....

.............................. .................... 2.5.4. T h 4 equilibrium ...... 24

3 . Models of Fe II Emission in the Orion Nebula ...m..........mL............ 28

3.1. The predicted F e III spectrum ........................ ..... ...... 28

...... 3.1 .1 . Introduction ..... t., ............................ ..r. 28

3.12. Level popdation sensitivity to the pumping conditions .......... 29

3.1.3. The sensitivity of Fe II miission to eiectron densiy ............... 38

3.2. Prehinary assessment of the observed spectrum

of the Orion Nebula .................................................................... 38

3 2 1 . The [O Il specmun ......-................... ............ ......................... 43

................................ 3.2.2. The [Fe II] specaum ...................... .. .... .. 49

32.3. Summarg ....................................................................................... 3 3

4 . Orion Nebula . New Observations and Emission Line

4.1. Observations and reductions ................................................ 5 4

4.2. Anaiysis of errors ............................................................... 7 4

4.3. Line identifications ....................................................................... 77

.............................................. 4.4. F e n] h e s in the Orion Nebula 79

4.4.1. Line list ............................................................................................ 79

4.4.2. Cornplrison with mode1 predictions ........................................... 83 4.5. VeloQty field analysis in the Onon Nebula ................... .. ..... 94

4.5.1. Introduction ........................... 4.5.2. Forbidden lines ............................................................................... 97

............................................................................. 4.5.3. Permitted h e s 101

5 . Fe II Emission from Bmad Emission Line Regions

of Active Galactic Nuclei ....................... ..L, ...... ............ 108

5.1. General picture of AGN and observational facts ................ .. 108 5.2. BEZR cloud stnicture .................... ...... .................................... 113

5.71. The ionizaaon hncaons of Ho. H'. He" . He'. and ~ e ' ~

vs . depth ........................... ...... .......................................... 113

5.22. The ionization hctions of Fe 1-Fe V vs . depth within

...................................................................................... the cloud 115

5.2.3. Totai and Fe II cooiing vs . depth within the doud ............... 115

5.24. The resulang spectnun... ........................................................... 115

................. 5.25. Elcctron temperature behavior .. ....................... 118

............................ 5.3. Fe II &sion modeling for BELR of AGN 121

................. 5.3.1. D&y and flux dependence .. ............................ 121

5.35 Dependence on microturbulent veiocity .............................. 128

5.33. L p pumping ..,................ ........................................................ 132

5.3.4. Fe abundance and cosmological scde ................................ .,.. 135

5.3.5. ' 2ocdy optimdy emitting doud" concept and Fe II emission

6. Summary ~ o ~ ~ e e e o ~ ~ o e e e e e o e e e e e o o o e o e o e e m o . e e o e e o e e e o e e o o o o e ~ e ~ e o o ~ o e e e e e e e e o e e e e e o e a e o e e e e e 147

6.1. Fe II emission line physics ...................... ..o.o....o...-................. 147

62. [Fe IlJ emission from the Orion Nebula .............................. ... 148

6.3. Analysis of other emission lines in H 11 regions ................... 149

6.4. Fe II Iine emission frorn Active Gdactic Nuclei ......... ..... . ..... 151

References o.~o.oeo*o*oeoooeoooaeeoeeeeeeeeteeeeooeeee*eeeoee.eeaeeeo.oe*eoeeo**e*eeeeoeeoe*****.e**ooeooe.* 155

LIST OF TABLES

3.1 Observations of the Orion Nebda ................................................... 46

3 2 Dereddened and Predicted Spectrum ...................... .............. 48

3.3 Dereddeneâ and Predicted [Fe n] Luie Intensities ....................... M

4.1 Orion Emission Line Measwments from 3498 A to 7468 A ...... 60

4.2 Ratios of I(Line)/I(6678) .............................................................. 76

4.3 Observed [Fe II] Emission Lines ....................... .. ........................ 80

4.4 Obsewed and Predicted F e II] Line Intensities ......................... ... û4

5.1 Parameters of the BELR Fe U Emission Models .. ........................ 122

LIST OF FIGURES

The energy distriiution of the 371 lowest levels of Fe II ion ................... .. ...... 13

Log-log distribution of density of Fe II lines vs . wavelength .......................... 14

.................... DepartuTe coefficients of selected Fe II leveis vs . dectron density 21

Departure coefficients of selected Fe II levels vs . energy density of radiation

field ............................................................................................................................ 23

Departure coeffiaents of selected Fe II levels vs . the 912 A continuum optical

depth of the doud ................................................................................................... 25

The calcdated electron temperature vs . the energy densi ty temperature ....... 26

..................... Log of the ionization fractions of H 1 and H II vs . depth in doud 30

Log of the ionkation fiactions of He 1 and He II vs . depth in doud ................. 31

Log of the ionization fractions of O 1 O iII vs . depth in doud ........................... 32

Log of the ionization fractions of Fe 1 O IV vs . depth in doud ................. ..... 33

Log of the electron temperature (K) vs . depth in doud ...................................... 34

3.6a Log of Fe II level population ratio (with and without pumping) similar to the

...................... Won Nebula with blackbody 35. 000 K continuum ................... ... 35

3.6b Log of Fe II level population ratio under conditions (with and without

pumping) with Trapezium stars continuum ................................................................. 36

....... 3.7a Fe II emWion line fiuxes for one-zone caldations and density 102 a+ 39

....... 3% Fe II emisgion line fluxes for onezone calcuIations and density l(yl a+ 40

....... 3.7~ Fe II emission Iine fluxes for one-zone calcuIations and density 106 cm3 41

........ 3.7d Fe II emission line fluxes for one-zone caldations and densitv 108 cm.3 42 d

Parts of the spectra of the Onon Nebula observed with HST FOS and

....................................................................................................... CTIO 4 m echeile 45

......................... Sample of the fully calibrated spectrum at Balmer limit region 57

Sample of the fuily calibrated spectrum showing OII near 4650 A ................... 58

Speainun with strong teiiuric emission and nebular [O I] 5577 A emission

........................................................................................................ .................... line ,., 78

The lowest 43 levels of Fe II ion and the observed multiplets in the Orion

................................................................ ................ Nebda .............................. 8 2

The ratio of the predicted intensities of F e II] lines to the obsewed values .... 86

................ . The predicted intensities vs the observed intensities of F e IIl lines 87

The ratio of the predicted intensities to the obsewed values for the Fe II

................... a4F - a4H multiplet .. ......................................................................... 88

The predicted intensities vs . the observed intensities for the Fe II aJF - a+I

...................................................................................................................... multiplet 89

The ratio of the predicted intensities to the observed values for the Fe II

................................................................. ................... a4F- azG multiplet .... ........ 90

4.10 The ratio of the predicted intensities to the obsenred values for the

............................................................................................. . Fe II a6D b4F multiplet 91

4.11 The ratio of the predicted intensities to the observecl values for the

. Fe II a6D a% multiplet ............................................................................................. 92

4.12 The ratio of the predicted intensities of F e III lines to the observecl values vs . the energy of upper leveis of the transitions .,., ............ , .................................... 93

4.13 H 1 velocities in the Onon Nebula .......................................................................... 95

4.14 He 1 velocities in the Orion Nebula ........................ ......................... ........... 96

4.15 Fe n] velocities in the Onon Nebula ............... ........................................... 98

4.16 Fe III] velocities in the Orion Nebula ........................................................... 99

4.17 F e m] velocities vs . excitation potential in the Orion Nebda ..................... ... 100

........ 4.18 Average velocities of forbidden Lines vs . ionization pot& al ..... ............ 102

4.19 N II velocities in the Orion Nebula .. ........................... ..................................... 103

4.20 O 11 veloaties in the Orion Nebula ..................................................................... 104

4.21 Veioaties of al1 emission lines obsemed in the Orion Nebula .................... .... 105

.................................. 5.1 A p e r d basis for the photoionization mode1 of AGN 110

5.2 Composite quasar spectrum (fiom Bright Quasar Sarnple) ............................ ... 111

............ . 5.3 Log of the ionization fractions of H 1, H II, He 1-iü vs depth in cloud 114

5 A Log of the ionkation fractions of Fe0 - F&+ vs . depth in doud .......................... 116

5.5 Log total and Fe II cooling vs . depth in doud .................... .......................... . 117

5.6 The diMw ernission from a doud with gas density log n(H) [an-31 = 12.

ionization parameter log UN) = -3 (log@ r 19.5) and column density log N(H)

......... [ m 2 ] = 23 and illurninated by an ionizing spectnim typical of quasars 119

5.7 Log of the electron temperature (K) vs . depth in doud ...........................cc....... 120 5.8a Theoretical UV and optical Fe II speaia caldated for the parameters

expected in quasar broad emission Luie regions ................................................. 124

5.8b The same as Figure 5.8a, but with density increased at constant ionization

parameter ........................................................... ........................................ 126

5 .8~ The same as Figure 5.8a, but with density increaseâ at constant flux, loweting

the ionization parameter .......... .................... . . . . . . . . . . . . . ......... 1 2 7

5.9a Same as Figure 5Ba, with same ionization parameter, intermediate density

and miaoturbulent veioaty v,= O km s-1 ................... .. ...... .............................. 129 5.9b Same as Figure 5.8a, but with P, = 10 km s-1. .................................................... 130

5.9~ Same as Figure 5.8a, but with V, = 100 km si. ...................... ............................. 131

5.10 The predicted Fe II spectrum when the effeds of Lya pumping are tumed

off........... ............ ....... ... ............................................... *...* ...... *......* ........ . ............ 1 3 4

5.1 1 Age-redshift dependence on cosmological parameter q for different values of

the Hubble constant (from Hamann & Ferland 1993) ..................... ..... ......... .... 136

5.12 Like Figure 5.8a showing dependence of spectrum on Fe ............................ ... 138

5.13 The flux of the total Fe II ernission in the spectral band 2430, from 2200 to

2660 A, shown as a function of the hydrogen density and flux of hydrogen-

. 0 . i o m g photons ........................... .-....... ....... .. .... ....... . .................... . 141 5.14 The same as Figure 5.13 for the Fe II 5270 emission band,

horn 5080 to 5460 A .................... ................................ ..... ................. . . 142

5.15 The same as Figure 5.13 for the Fe II 2780 emission band,

from 2660 to 2900 A ..................... .. ..... . ...... . ......... . ........ . . . . . 1 4 3

5.16 The same as Figure 5.13 but for the broad Mg II doublet at 2800 A .............. 144

5.17 Relative contriiution of the Fe II 2780 emission compared to the a broad

Mg II doublet Une for the same grid of caldations ................... ....................... 145

Chapter 1

Introduction

1.1 Background

A large number of the allowed and forbidden Fe II lines are present in a

broad range of emission spectra in a variety of astronomicai sources. The Fe11

emission spectrum &ses from thin "envelopes" around a variety of objects mch

as stars, H II regions, novae, supernova remnants, active galaxies and quasars.

The high abundance of iron in the Universe and the very rich spectrum of Fe II

explain why this is so. The iron abundance is 3.105 that of hydrogen, ranks only

after H, He, C, N, O, and Ne in a solar composition, and is comparable in

abundance to S, Mg, and Si.

Fe II has a very line-rich spectnun. The richness in lines of Fe iI originates

in its about half-fiIIed 3d-shell. Its ground state configuration is 6 2 6 ls%?&~ 3s 3p 3d4s 6 ~ g n . The pioneeruig work on the term anaiysis of sin&

ionized iron (Russell 1926) reported 61 energy Ieveis. Now more than 1ûûû

energy levels of Fe II are known (Johansson 1978,1999). The Fe II spectrum has a

rich and complicated mixture of forbidden and permitted iines. The lowest

energy permitted transition occurs from the 64" levd to the first Ievel, and this

transition produces a strong W line at 4.77 eV (A = 2599.40À). Netzer & WUs

(1983) and Vemer et al. (1999) discussed the critical density for this transition,

and it is about 1010 cm-3. in a case when ody Iower Ieveis are populated, oniy

forbidden line transitions take place. With inaeasing density and temperature,

higher levels are populated by collisions, and the pemWed lines are observed. In

addition, fluorescence (and even recombination) can populate levels.

Many tens of te-, corresponding to hundreds of leveis, contxibute.

Laboratory spectroscopy is not cornplete and only in the past few years have

many of the radiative and collision rates been determined. First optical and then

W and IR spectral windows have become available to meamre the fÙil spectnun

in astrophysical sources.

The excitation and ionization conditions for populating Fe II levels are

very favorable in a great number of astronomicd objects. Because the ionization

potential is 7.9 eV for Fe 1 and 16.2 eV for Fe II, the energy of hard ultraviolet

photons (X 1 15704 is sufficient to produce singly ionized iron. Electron

temperatures between about 5000 to 20000 K are typical for producing strong

Fe II lines. The atom is far from equilibriurn in most emission objects, so the

spectrum is sensitive to the detailed local conditions. A simulation of the

physical processes affecting the Fe II spechvm wodd make it possible to deduce

the density, temperature, and iron abundance of the emitting regions. One fourth

of dI energy emitted by Broad Emission Line clouds (discussed in Chapter 5)

cornes fiom Fe+ ernission. Fe+ is the main coolant at high densities. Some stars

also show extreme Fe+ emission (Carpenter & Wing, 19û5).

Our goal is to completely simulate the emission of the Fe Ii atom in a wide

range of physical conditions. This goal is diffidt because the physics goveming

Fe II emission is complex. A variety of processes affect the spectrum, including

coilisionai excitation, pumping by the incident continuum, fluorescence by line

overlap, and line transport. It is this sensitivity to details that &es Fe II a

potentiaiiy powerful probe of physid conditions in the ernitting gas. Even the

resuIts are not simple however; tens of thousands of Iines often fom a pseudo-

continuum that can only be compaxed with observations afkr a complete

spectral synthesis. Fe II emission presents a ChalIenge at the very forefront of

computational nebular astrophysics.

The ubiquity and variety of Fe II iines in the Universe d e s their analysis

of such extraordinary interest. Moreover, the complicated Fe CI spectrum analysis

demands collaboration between atomic physicists and astrophysiasts to provide

the large amount of atomic data such as energy levels, collisional strengths and

transition probabilities.

An important long-te= goal is to determine the abundance of iron in

many contexts. For example, most of the evolutionary modeis of galaxies predict

an increase in Fe/O abundances after -1-2 Gyr to factors of -2 to 10 relative to

solar ratios (Hamann & Ferland, 1993). This overabundance might explain the

strong Fe II emission observed in many Q W s and AGN. The expected -1 Gyr

delay from the onset of star formation could therefore be used as a dock to

constrain QSO ages if accurate Fe abundances are measured. Age constraints

could in turn constrain the cosrnology when applied to hi@-redshift sources.

1 .f .1 Fe Il emission from H II regions

A very active field of research is the diagnostics based upon F e II] Iines of

Iow-density plasmas around a variety of different objects with the aim to obtain

consistent models of the vdoaty field? density, excitation conditions, etc. H II

regions are good Iaboratories to check how Fe II emission works.

Recent studies of the F e II] lines observed in the Orion Nebula have

s h o w that some intensity ratios cannot be expIained by only colIisional

excitation at the densities ne - lW cm-3 deduced £rom the analysis of other

forbidden iines. In order to explain these disaepanaes Bautista et al. (1994)

postdated the existence of high-density partiany ionized regions with n, - 10'

cm+ Lucy (1995) has shown that the h e ratios of [Ni 11] can be explained by the

fluorescent excitation of Ni+ IeveIs by W neMar radiation. The fluorescent

excitation may be important for Fe II iine formation process (Baldwin et al. 19%)

and some Fe II emission can be explained by the presence of continuum

pumping. Recentiy, Rodriguez (1999) showed that the intensities of F e 4 lines

in different H II regions are strongly correiated with the intensity of the diEuse

continuum,

A study of the Onon Nebula forms a major part of this thesis (Chapters 3

and 4).

1.1.2 Fe II emission from Active Galaetic Nuclei

A forest of thousands of Fe II lines competes with Lya as the strongest

contributor to the broad emission line spectrum of @Os, the nudei of radio

galaxies and Seyfert 1 galaxies. The most prominent feature, always present in

broad line spectm, is between 2000 and 3500 A. The Fe II Iines in the opticd

region, strongest near 4600 A, vary widely in strength kom one object to another.

Fe II emission was seen in extragalactic objects fist by Greenstein &

Schmidt (1964) for quasars 3C4û and 3C273. Features near 4450450 and 5100-

5500 A were resbserved by Wampler & Oke (1967) who suggested their

identification with blended Fe II multiplets. Later, Sargent (1968) made the sarne

identification in the Seyfert 1 galaxy I Zw 2. From the absence of [Fe II], Warnpler

& Oke conduded that the emission region has ne > 106 an-î or higher. More

Seyfert galaxies were observed (e.g., Fairail 1968, DeVeny 6r L p d s 1969, Adams

& Weedrnan 1972, Baldwin 1975, Boksenberg & Netzer 1976, Oke & Shields 1976)

and other Fe II emission features were recognized. Grandi (1981) and Netzer &

Wis (1983) have shown that Fe II Lines are prominent in the spectra of quasars.

The f h t systematic spectroscopie survey of Seyfert galaxies by Osterbrack

(1977) showed that Fe II was present in 90% of the Seyf& 1 galaxies, and in

parüdar that the Fe II emission was associated onIy with the broad emission

iine objects. He &O noted the similarity of the Fe II spedra fkom one object to

another.

Phillips (1977, 1978a,b) discussed the properties of the Fe II ernission

region in Seyfert galaxies in more detail. He derived a lower Mt on ne -107 cm;?

also £rom the absence of [Fe iines. At the same t h e he supported the BLR

origin of the Fe II. He also searched for Fe II emission in a number of low redshift

@Os, finding it to be generaliy very weak.

The opticai Fe U emission has been observed not only in Seyfert galaxies,

but in other types of AGNs in low-redshift quasars (Boroson & Green 1992) and

some high-redshift quasars (Hiil, Thompson, & Elston 1993). In some AGNs, the

intensity of the opticai Fe II multiplets near 4570 À is 4 times the HP intensity

(Bergeron & Knuth 1980; Lipari, Terlevich, & Macchetto 1993) while it is

undetected at a level of a few percent of HP in some other AGNs (Boroson &

Green 1992; Osterbrock 1977).

The ionization structure in AGN is very different from that of H II regions.

The douds can be thick enough to be highly ionized (N*+, 05+) at their

illuminated face and almost completely neuttal at the bak. The large flux of X-

ray photons maintains a low degree of ionization (-10%) over a large part of the

doud, more than 90% of its thickness in some cases. Such extended low

ionization regions are thought to be the orîgin of the strong Fe II and Mg iI Iines.

The low excitation Iines of Fe II are produced in the partly neutral region

of the BLR clouds. They are characterized by Te - 104 K. Understanding the Fe II excitation and radiative processes in quasars wiIl

Iead to a better understanding of the energy balance in the broad Iine region and

the abundance of metals. It seems likely that, in the UV, Fe II feahws that are

dearly seen are only local peaks where strong multiplets are seen lyuig over a

generai forest of weaker h.

I Zw 1 is the prototype narrow Iine Syl gdwy. Its narrow line profdes

rninimize blending effects. The prominent Fe II multiplet, UV 191 according to

the Moore (1972) classification, is seen, together with weak features at 1294 À

and 187ï A. The presence of such Fe II emission indicates that Fe iI is not excited

simply by collisions.

Because of the extremely complicated energy level configuration of the

Fe II atom, Iine intensities are not simple to calculate. There are up to thousands

of transitions of Fe II to be considered, many with large optical depth. An

additional complication is the large number of wavelength coincidences of

different Fe II lines. Their number at high densities (>IO10 d) can reach 1000.

h e overlap and self-pumping is a very important population process for the

levels that must be taken into account.

There have been a number of pioneering theoretical investigations into

Fe II emission spedra in Active Galactic Nudei (AGN). For example, Philiips

(1978a,b) discussed continuum pumping as one of the excitation mechanisms

that is responsible for the Fe II emission. Brown, Jordan, & Wilson (1979) and

Penston et al. (1983) suggested that Lya pumping might excite high excitation

transitions in Iate-type and symbiotic stars (see also Johansson & Jordan 1984).

Penston (1987) showed that this process can be very important in the overalI Fe II

flux from the symbiotics and might, therefore, contribute substantially to the

strong Fe U emission kom AGNs. Netzer & Wis (1983) and WiIls, Netzer &

WiUs (1985) deveioped the most detailed models of the Fe II emission kom the

Broad Emission Line Region (BELR) of quasars. These modds took into account

the processes of continuum and line pumping, induding Lya pumping and Fe II

self-fluorescence. The models were based on atornic data available in the 1980s,

and assumed LTE population for some FeII Levels. These models were able to

explain the apparent strmgths of the UV and optical Fe II features fairly welI, but

there were large disarepancies berneen the predicted and obsenred shapes of the

spectca. The observed W Fe II (2200-2650À) profile is reiatively smooth

compared with the dips and bumps in their mode1 spectra. Another very

interesting observed fa& is that the Fe [I features between 2200 and 2650 A are

always present, but the strength of Fe II 450-4680 A is quite different in different

obi-.

1.1.3 Fe II emission Rom stellar sources

In contrast to the situation in AGN research, there is a large body of more

recent theoretical investigations of Fe11 emission in steiiar sources. Applications

include supernova envelopes (Hauschildt, Baron, Starrfield, & Nard, 1996; Li &

McGay 1996; Koma & Fransson 1998a,b; Hoeflich, Wheeler, & Thielemann

1998), and more normal stars (Lanz, Hubeny, & Heap, 1997; Young, Landi, &

Thomas 1998). This work is largeiy based on stellar atmosphere tediniques (see,

for example, Mihalas 1978), in particular using fairly exact radiative transfer

methods.

The approach we take here cornes fiom a plasma and nebular physics

background (see, for example, Netzer 1990) and tries very hard to do the best

possible simulation of the miaophysics, within the restrictions of an

approximate radiative transfer treatment, based on escape probabiIity methods.

This approach has advantages in both speed and portability (Kallrofen 1984). It

is applicable to a wide range of astrophysical enviroments fiom H 11 regions

and circumstellar enveiopes to the various emission and absorption line regions

of AGNs. We also note that for many of these applications we have Little or no

howledge of the locaI geometry and veldty field. Therefore, the escape

probabilities and gaussian microturbulent velocities we will use are appropriate

(maybe even preferable) for exploring the gros sensitivities of the Ihe spectra to

various parameters.

1.2 Thesis outline

The main goal of this thesis is to aeate an elaborate numericd model of

the Fe II atom coupled to selfconsistent photoionization caiculations and to

make a contribution to the understanding of the origin of Fe II emission. From

the previous review it has become dear how complicated the problem is, even

with new atornic data available. As currently implemented, the atom model

indudes the lowiist 371 leveh (up to 11.6 eV) and predias intensities of 68635

lines. An initial emphasis for appücation of o u model has been the Orion H II

region due to its relative simpliaty.

In Chapter 2 we descnbe our data sources, which indude the most recent

experimental energy levels, transition probabilities and collision strengths.

Although we use detailed fits to temperature-dependent collision strengths

where possible, in many cases the uncertain gaPproximation is the only source

for collision data. The atom is designed to be readily expandable to indude more

levels and to incorporate more accurate sets of collision and radiative data as

cornputers grow faster and the atomic databases expand. We introduce the basic

equations of level populations, Iine transfer and heathgcoohg balance, and

describe the numericd methods of their solution.

We describe the methods and approximations we have used to

incorporate a large Fe+ ion modei into Cioudy (a spectral synthesis code

designed to simulate conditions within a plasma and model the resulting

speanim (Ferland et al. 1998)) to predict the emission of Fe II lines. S e v d test

cases showing that the atom goes to LTE in the limits of high partide and

radiation densities are presented.

ui Chapter 3, the photoionization calculations are applied to Fe II emission

from H II regions. W e investigate the sensitivity of the populations of di 371

levels in our model to the pumping conditions, and the sensitivity of Fe II

emission to electron density for conditions similar to the Onon Nebula. Some

levels are very sensitive to pumping conditions and can't be used for the density

diagnostics. We also investigate the general behavior Fe II emission flues over

the whole wavelength range. After that, we model the observed spectra of the

Orion Nebula. We reexamine the spectroscopie underpinnings of recent

suggestions that [O I] and [Fe 11] lines from the Orion H II region are produced in

gas where the iron-carrying grains have been destroyed and the electron density

is surprisingly high. Our observations show that the previous detection of [O I]

5577 was dominated by telluric emission. Our limits are consistent with a

moderate density (=lW &) photoionized gas. We show that a previously

proposed model of the Orion H II region ieproduces the observed [O q and F e II] spectnun: these lines are fuliy consistent with formation in a dusty region

of moderate density.

Chapter 4 presents a new set of observations of the Orion Nebuia. New

detailed observations with the cassegrain d e l l e spectrograph on the Blanco 4x11

telescope at CTI0 gave a possibiüty to idenhfy 411 emission lines in the 3498-

7468À range for two sets of observations. We have identified 40 [Fe II] hes ,

double the previous number of F e II] iines from the Orion Nebula. We present

detailed atlases of emission lines from the Orion Nebula dong with error

analysis. We separately present the detaüed table of Fe iI Ihes hom the Orion

Nebula and compare our theoretical modeis with the new set of observations.

Finaily, we present a detailed analysis of the velocity field in the Orion

Nebula, and show that the forbidden iines spiit into two distinct dusters. The

first duster has ions with an ionization potential Iess than 20 eV, narndy O 1, N 1,

Ni 11 and Fe II. The lines of these ions have velocities (relative to the local

standard of rest) hmn +10 to +15 km SI. There is a sharp diange io the second

duster, which indudes the ions with the ionization potentid imger than 20 eV,

namely S II, O II, N II, and Fe m. These lines have vdocities around 3 km d,

with a slight trend of decreasing velocity with the inaeasing ionkation potentid.

This is consistent with a dynamic model in which h e s of ions with different

ionization potentials originate at different distances from the ionizing star.

in Chapter 5 we apply the photoionization calculations to modeling of

Fe II emission from AGN. We discuss how Fe II effects Uoudy code calculations

and present the calculations for a single doud for illustration.

We present a general photoionization modei of AGN emission. We find

that in order to measure the strength of the Fe II emission with any reliability we

need a full model. This is because in broad Une spectra the actuai continuum may

lie far below the apparent continuum due to blending of thousands of broad

Fe II lines. We apply our 371 level Fe II model for typical AGN parameters with

moderate abundances. The main grid of caldations has two main parameters,

density and flux of ionizing radiation. Then we changed other parameters of the

model, abundances and microturbulent veIocities. Special attention haç been

paid to Lya pumping of Fe II lines, and to cosmologicaI aspects of Von

abundance that might be deduced from the shape of Fe II spectra. General

features (lines, multiplets, and bands) of the Fe II spectra and thir dependence

on the basic parameters (density, microturbulent veiocity, Lya pumping,

influence of Fe abundance) of the modds are discussed. Findy, we describe the

concept of "locdy opamally emitüng douds" as an approach to understanding

quasars spectra, and present the grids of calcuiations of Fe U emission over

several orders of magnitude in variation of density and ionking flux. We show

that powerful selection effects are at work. The pemiitted W multiplets, the

forbidden colIisionalIy excited iines, and the pumped Fe II lines are fonned most

effiaently at quite different cioud parameters.

In the Appendix, we show how to use the Fe II model in Cloudy giving

the commands added to Uoudy and an example of an input file that indudes the

Fe II atorn

Chapter 2

The Fe II Model and Atomic Data

2.1 The model atom and its atomic data

Fe II is a complicated atom with the following structure of electronic shells

in the ground state: 1 s ' 2 s ~ ~ ~ 3 s ~ 3 ~ ~ 3 d 4 s 6 D g E . TO date, energies of more than 1000

excited levels of F d have b e n determined experimentally (Johansson 1978,

1999). Astronomical spectra show emission bumps formed by hundreds of the

Fe 11 lines. To model these bumps and individual lines theoreticaîîy, we have to

include in our calcdations as many Fe II levels as possible. However, there are

some bits posed by the existing atomic data. In order to compute intensities of

the emission lines, we need the following data for the Fe II atom:

1. Experimental energy levels.

2. Einstein coefficients (oscillator strengths) of spontaneous transitions

between levels.

3. Collision strengths (excitation rates) for transitions between levels.

Our m e n t model of the Fe II atom includes d371 levels with energies

below 93487.650 cm-1, or 11.59 eV. AU level energies are experimental (Johansson

1978, 1999) and should be quite accurate. Wavdengths calculated from

experimental ewgy levels are more teliable than meamwments of individuai

lines, because the experimental energy levels are based on the analysis of the

whole spectnun. In the case of Fe LI, errors in the caldated wavelenm are

generally less than 0.001 A. Figure 2.1 shows the energy distribution of the 371

lowest levels of Fe+. The horizontal avis shows terms ranging fiom octets to

doublets. The vertical sale indicates the energy of levels in eV. For each

multipliaty and orbital momentum, even ternis are on the left, and odd terms on

the right.

Figure 2.2 shows how the number of iines within equal logarithmic

wavelength intervals predicted by our mode1 depends on the wavelength. Ail

prediaed lines are induded for the upper c w e , and the lower curve shows the

pennitted lines. The density of lines is different for shorter and longer

wavelengths, which shows that a different approach must be taken for UV and

IR iines. In the IR, individual lines cm be identified. For visible wavelengths

many lines are blended. For the UV a complete spectral synthesis must be used

because the number of lines is large enough for the h e s to fonn a blended

pseudo-continuum. The number of Iines can reach several hundred per A in the

W. This synthesis approach will be used in Chapter 5.

Transition probabilities are taken from theoretical calculations by Nahar

(1995, allowed transitions) and Quinet et al. (1996, forbidden transitions), and

supplemented by data from compilations by Fuhr, Martin, & Wiese (1988) and

Giridhar & Arellano Ferro (1995). Transition probabilities for all

intercombination lines not covered by these compilations are taken from the

Kurucz (1995) database. Uncertainties are generally smder than 20% fot strong

permitted lines, but can be larger than 50% for weak permitted and

intercombination lines. For the strongest forbidden transitions (transition

probabilities Ah > 0.01 81) uncertainties are Iess han 15%. The mors in

probabilities of the forbidden transitions with 0.01 c Aa < 0.01 pl are not higher

than 30°h (Quinet et al. 19%). The forbidden lines with Ak < 0.001 s-1 are hardly

observed and not used for our analysis of observational data.

Figure 2.1: The energy distribution of the 371 lowest levels of Fe II ion that are

included in our calculations. The horizontal axis shows te- ranged hrom octets

to doublets. For ewample, 6D stands for 6D (left) and 6Do (right). The vertical

scale indicates energy of levels in eV. AU levels below 4.77 eV are of even parity.

The Cirst level of odd parity z~DQ/~ is shown in box. The first allowed transition,

a Q / z -z6D%/2, corresponds to the h e h=2599.40 A. The energy of the 371st level

is 11.59 eV, the ionization iimit is 16.19 eV.

Figure 2.2: Log-log distribution of density of Fe II lines vs. wavelength (solid line,

including al i lines; dotted he, pennitted lines).

Collision strengths for al l quartet and sextet transitions among the 142

levels (10,011 lines) are taken tr~rn.~rnatr ix calculations made by Zhang dr

Pradhan (1995). For other Iines, we use interpolations and the gapproximation

(Seaton 1962, Van Regemorter 1962, Mewe 1972). It is hard to judge the

uncertainties in the theoretical calculations, but they might be within 30% for

strong Lines, and 0.5 dex for weak and lines.

2.2 Balance equatlons

The first step in modeling the Fe II emission spectrum is to calculate the

atom's level populations. A realistic mode1 requires a suffiaently large number

of levels and must include al1 excitation and de-excitation mechanisms. We

assume that the atom is in steaciy state and solve for al1 level populations by

balanhg processes that populate and depopulate the levels.

Let ni be the population of Fe+ ions in level i. For steady-state conditions,

ni is constant with the , and the number of transitions into level i is equal to the

corresponding number of transitions out per second. For the ia level population,

the complete form of the rate (master) equation is

Let us illustrate the tenninology. A',i is the effeaive spontaneous

transition probability, A ii = Aui(L;i + dii). Here Aw is the Einstein coefficient, C. is

the escape probability and & is the destruction probability (Ferland et al. 1992).

The rate of absorption upward from i is B :unit where B & is given by B,K, Bk is the induced transition probability and J , is the mean intensity averaged

over the line profile. zincludes the attenuated inadent con t i nu~~~ l and net

radiation field produced by other lines. The flux r i s calculateci self

consistently, again using the escape probability formalism (see Elitzur & Netzer

1985). Similarly, B is the rate of the induced transitions downward from u to i.

In Equation 1 qiu is the collisional exatation rate coefficient (cm3 +) from

level i to u, and ne the electron density. We use our fits to the temperature

dependence of the electron collision strengths to obtain q. Colhional de-

excitation rates qui are obtained by detailed balance, qb = q, exp(-hv/kT) &,,, where the g's are the statisticai weights. Compare to electron collisions, proton

and hydrogen atom collisions are not important if carbon is at least once ionized,

as should be true in any region where Fe is ionized to Fe+.

The system of equations given by Equation 1 is overdetermined. We

replace the balance equation for the ground level with the conservation equation

Dti = n(Fe+), where n(Fe+) is the total number density of Fe+ ions. The level

populations can then be obtained numerically by solving the set of algebraic

equations by matrix inversion. However, if the self-pumping of Fe II lines is

hduded in the caiculation, the matrix coefficients, in tum, indirectly depend on

the Fe+ level populations. Thus, the final solution of the problem is obtained by

iteration in each Cloudy zone (see Chapter 25), with the feedback effects of the

atom on the radiation environment explicitly taken into account.

2.3 Line transfer

In the current version of our mode1 a l l lines are transported using the

escape probabiüty formaiism (Elitzur 1984, 1992). However, the form of the

coding was done to d o w eventual incorporation of exact transfer methods when

cornputers become fast enough.

Both continuum and line pumping are considered in the radiative rates.

This uidudes pwnping by alI lines of other atoms and ions induded in the

simulation as well as self-pumping by overlapping Fe II lines. Wills, Netzer, &

Wills (1985) stressed the importance of the last &ect. In the escape probability

fomalism radiation produced by a line is not counted as an exatation process

for that particular line, but is counted for aIl other lines. Line purnping of a

partidar transition is efficient if the wavelength of the FeII line coincides with

another optically thidc he. In the master set of equations, the 106s of a photon in

a particda transition due to absorption in another transition counts in the

destruction probability of the fîrst, and fluorescence of the second.

Selective excitation of high-1ying leveis by Lya pumping rnight be very

important in stars and AGNs (Pemton et al. 1983). The Lya line width is given by

Adams (1972):

where iîiaop is the Doppler width, a the darnping parameter and ?the line-center

optical depth. According to Adams (1972, FigA) we approximate the profile of

Lya Iule as a trapezoid with the upper base 2 times less than lower base. in UUs

case the width of the line (FWHM) is the average of lower and upper bases. The

number of Fe II lines that can be pumped by Lya is very sensitive to the Lya

optical depth and resulting line width, and to the electron temperature since this

determines the lower level populations of Fe+. The number of the pumped lines

can range fiom only a few lines to tens of hes .

For ail other optically thick lines we use a simplified rectangular profile to

calculate the line widths. Again the line widths are a function of the optical

depths, whîch are onIy known &er s e v d iterations. We use the h e profile

suggested by Elitzur & Ferland (1986):

2.4 Heating-cooling balance

The balance of al l cooling and heating processes detennines the electron

tempera- Te. We define heatuig and cooling relative to the continuum

(Osterbrock 1989) so only processes that change the electron kinetic energy

count. The total collisional cooling (erg cm3 s-1) produced by Fe+ is

where h vu[ is the energy of the transition. If a level inversion occurs, so that the

population of a level relative to its LTE value increcises with excitation potential,

the cooling will be negative and the atom will act to heat the gas. This will occur

in some of the following models. It happens since some pumping mechanisms

selectively populate very highly excited leveis, which are then collisionally de-

exci ted effectively converthg radia tive into kinetic energy .

Fe II o h dominates the thermal equiübrium of a cloud, and cm account

for well over 90% of the cooling in some conditions. In such cases the

temperature predictor-corredor must have a very good analytic estimate of the

derivative of the net Fe II cooling to apply the next temperature correction. We

have tried various prescriptions for this derivative for large muitilevel atoms.

We found that taking the excitation temperature of lines as the energy between

the upper level and the ground level can best represent the derivative of the

cooling between two levels. Then the cooling derivative at constant ionization

will be

In practice the ionization often changes with temperature and it is

necessary to iterate further on the solution. The caldations are repeated until a

stable solution is obtained. Typically ten iterations were required. Specific v

examples of the effects of Fe II on the thermal structure of a doud are given

2.5 Interaction with the environment: tests of approach to local thermodynamic equilibrium

The Fe Ii mode1 has been incorporated into the development version of

the radiativeîollisional code Cloudy (Ferland 1996, Ferland et al. 1998). These

references give an ovewiew of the methods and approximations used by the

code. We briefly summarize parts relevant to Fe II here.

An optically thick slab is divided into a large number of mesh points,

referred to as zones, chosen to be thin enough for the ionization and temperature

to be nearly constant across each bin. We then solve the set of statistical

equilibrium equations (see Equation 1) in each zone. On the first iteration, the Fe

II code gets the radiation field calculated by Cloudy with no contribution from

the Fe II emission, and then calculates the flues of all the Fe II lines. On the

second iteration, Cloudy includes the Fe II contribution as it is given by the first

iteration of the Fe II code. We recalculate ali values and iterate to convergence.

Thus, the electron temperature Te, the local field z, the electron density ne, and

the ionization balance of ail elements are al l calculated self-consistently. These

result in a net heating or cooling of the gas, with a l l Fe II interactions induded.

The electron temperature is determined by making the ciifference between the

heating and cooling rates esaller than a pre-set tolerance. The local radiation

field is determined from the level populations of d spdes and the local opacity.

The final spectrum is the integration over the computed structure.

There are fundamental uncertainties in the physics of Fe II emission due

to mors in the basic atomic data. These are both systematic and random mors,

and are hard to evaluate. However, our caldation has energy balance as its

foundation, and this limits the largest possible excursion due to m o r s in the

basic rates. The purposes of the large numerical simulations are basicaily to

detennine how much energy is absorbed from the incident continuum and then

find into which channels the energy will be reradiated.

Before attempting to simulate the Fe II spectnim in a non-equilibrium (but

steady state) envuonment we verified that the model atom gws to local

thermodynamic equilibrium (LE) in four limiting cases. We define the

departure coeffiaents as a ratio bi = (n i /n r ) / (n i ln l )~~~ where n/hr is the actuai ratio

of population of the P level relative to the ground level, and (ni/nl)L~E is specified

by the Boltzmann equation. Each of these limiting cases has (different)

parameters chosen sa that Fe would have a significant fraction that is singly

ionized.

2.5.1 Cloud structure: collision-ûominated case

Figure 2.3 shows a test case where collisional processes are dominant. The

iron abundance has b e n set to a very large number, (Fe/H)=100, to insure that

the Fe+ dominates the calculation. AU of the radiative processes are adually

hduded but the extemal radiation flux was set to a very low intensity. The

model is of a very thin ceii of gas that is opticalIy thin in the lines and conünuum,

is. the column density is kept small enough for optical depth effects to be

unimportant. The level populations are determined by solving the full set of

equations of staüstical equiliirium for a preset electron temperature Tt = 5,000 K.

ûeparture coefficients for the ground state (l* level) and for the 10ü, 25% 64th

and 295th levels are shown as a fundion of the electron density. Note that the first

63 levels have even parity and therefore radiative transitions between these

levels and ground have very low probability. As a result, these levels corne into

LTE at much lower densities than the upper levels starhg from the 64*, whîch

can decay by permitted transitions.

I

4 6 8 10 12 14 16

log Qectron Density

Figure 2.3: Departue coefficients of selected Fe II levels vs. electron density.

Populations of all levels reach LTE at densities above 10%1+.

For low densities the ground state (level 1) is overpopulated relative to its

LTE value. This is because upward collisional processes are much slower than

downward radiative processes. When the densities are high these rates become

equal and the depamire coefficients approach unity. Similarly, highly d t e d

levels are underpopulated relative to their LTE values at low densities, since

radiative decays to lower levels are faster than collisional processes. For

densities > loi6 cm-3 collisional processes completely dominate the rates, and

the entire atom reaches LTE.

2S.2 Radiation-dorninated case

Figure 2.4 shows a series of tests where the extemal radiation field

becomes increasingly dominant. This culminates in the atom going to

equiliirium in the case of a mie of bladcbody radiation field. Again, the fidi set

of equations coupling the levels is solved. The density is low enough (nu = 109

cm-3) and the radiation field becomes intense enough that spontaneous and

induced radiative processes domhate the atom's level populations. The Fe

abundance is set to the same high value as in the collisional model described

above. The electron temperature is a constant 8000 K. Again the model is of a

very thin cell of gas, so that all lines and continuum are opticaliy thin. The gas is

exposed to a bladc body continuum with a color temperature of Tc = 8000 K but

with various values of the energy density of the radiation field. The equivalent

energy density temperature is Tu =(ula)"',where u is the total energy density in

al l radiation (erg anJ) and a is the Stefan radiation density constantt

For thÏs electron density the lowest 63 even levels are close to LTE for any

external radiation field, but the higher levels, from which p d t t e d transitions

exist, are far nom LTE. For low energy densities the Ievel populations of excited

States are below their LTE value while the ground Ievel is slightly overpopdated.

Increasing the radiation field has the effkct of increasing populations of high

Figure 2.4: Departue coefficients of selected Fe II levels vs. energy density of

radiation field. Populations of ai l levels reach LTE with increasing energy

density. The unusual behavior (overpopulated at low energy density) of the level

249 can be explained by low probabilities of transitions d o m horn this doublet

state.

levels, as pumping of these levels from ground grows more important. A

radiation field givm by Planck's law (i.e., Tu = Tc) forces the ionization and levei

populations of the atom to LTE in much the same way that high eledron

densities do. In the limit where Tu = Tc, the departure coeffiaents reach unity and

the Fe Ii goes to the strict themodynamic equilibrium (SE) limit.

2.5.3 Large optical depths

The third test checks that the Fe 11 atom goes to LTE with increasing

continuum optical depth at 912 A (Figure 2.5). The model assumes a constant

temperature of 15,000 K, a very high iron abundance (Fe/H)=lW, a gas density

nn = l o g a n . ' , and a very low ionization radiation density. Figure 2.5 shows the

departwe coeffiaents for the ground state (1st level) and for the 124*, 206% and

347th levels as a function of increasing optical depth in the Lyman continuum. At

large optical depth photons scattered many times and become thermalized. AU

the levels reach their LTE populations at sufficiently large optical depth and the

atom becomes effectively collisionaliy dominated.

2.5.4 Thermal equilibrium

The ultimate test is that a fully uon-dominated environment equilibrates

at the correct ternperature and population when the radiation field approadies

strict thermodynamic equilibrium (ir. Tu = Tc). Partides and light must corne

into equilibrium in this Mt, and this sirnuitaneously tests the ionization,

excitation, and thermal effects of the model atom. These tests are similar to the

applications to quasars below, in that the electron temperature is determined

sekonsistently from the balance between heating and cooling. The conditions

are similar to those in subsection 2-53 with the exception that the electron

ternperature Te is determined self-consistently. We adopt Tc = 8000K. Figure 2.6

shows how the calculated Te changes with Tu and reaches the same value

-4 -3 -2 -1 O

log Optical Deplh

Figure 2.5: Departure coeffiaents (note linear scale) of selected Fe II levels vs.

logarithm of the 912 A continuum optical depth of the doud. The continuum

optical depth is shown at 912 A for iiiustrative purposes but is not related to

hydrogen column density. Actually the continuum opticai depth is determined

by neutral iron given the abundance of iron. The calcdations started in

conditions where aü lower levels are already in equüiarium (a - 10'3 ana). This

is why low levels do not appear on the graph. The lowest 100 Ievels of Fe T[

approach LTE closeiy at s-1, levels up to 200 reach LTE at t-10, and populations

of a i l levels are in LTE at opticd depth - 10".

Figure 2.6: The calculated electron temperature vs. the energy density

temperature. They reach the same value when the energy density temperature

reaches the 8000 K blackbody temperature.

Tu=Tc - 8000 K. In this case we have recovered the iimit where the photons and

partides have corne into complete statistical equiliirium, verifying that the

caldation is fidy sekonsistent.

Chapter 3

Models of Fe II Emission in the Orion Nebula

3.1 The predicted [Fe II] spectrum

3.1 .1 Introduction

One of the sirnplest applications of the Fe II emission model is to the

interpretation of the infrared and optical [Fe II] lines from H II regions. The

Orion Nebula is the defining blister H II region (Zuckerman 1973; Balidc,

Gammon, & Hjeiiming 1974). There are four so-cailed "Trapezium" stars located

at the center of the core. This cluster ionizes the skin of the molecular doud,

causing an expansion away from the molecular cloud toward us. The H II region

is in photoionization equilirium, with a density in concert with that of the

background photodissociation region (PDR; Tielens & Hollenbach 1985). Both

regions are dusty with high depletions of the refractory elements (Rubin, hifour,

& Walter 1993).

In this Chapter, we presertt H II region model calculations, spectroscopic

observations, and their interpretation. We used Cloudy to analyze the physical

conditions: ionkation fractionsf the electron temperature and density of the gas.

W e cddated a modd under conditions similar to that in the Orion Nebula:

electron density ttc = IOl d, a depleted iron abundance Fe/H = 3406, nebdar

abundances for aI l other elements (Baldwin et al. 1991, Osterbrock et al. 1992),

with the continuum pumping by a blackbody radiation with T=35,000K (like 01

ûri C). Figures 3.1-3.4 show ionization fractions of hydrogen, helium, oxygen,

and iron as a function of doud depth. Figure 3.5 presents the electron

temperature against the depth of the doud. M e r the hyàrogen ionization front,

the temperature rapidly goes down and the gas becomes neutral.

3.1.2 Levei population renaitivity to the pumping conditions

The forbidden oxygen and nitrogen he s , e.g. [O II] ~727A, are

collisionally excited in H II regions and used for density diagnostics (Osterbrodc

1989). Observed Fe n] optical lines have similar excitation energies. However,

the structure of the Fe II atom is much more complicated. Other excitation

mechanisms, induding pumping of high Fe II levels by UV stellar continuum

and subsequent cascades down, can contribute significantly to the obsewed

intensities of the optical [Fe 11] lines. In order to determine what [Fe II] lines can

be used as density diagnostics it is necessary to investigate the behavior of the

level populations using the atom model. Simple one-zone calculations were

made assuming conditions similar to the Orion Nebula: nt = 103 cm-3, a depleted

iron abundance, and nebuiar abundances for ail other elements (see Section

3.2.2). We intentiondy chose a low density to be safe in determinhg which lines

are excited by collisions only. The first caldation indudes the process of

continuum pumping by blackbody radiation with temperature 35,00OK, and the

second one did not. Then the ratio of level populations for these two cases was

calculated for each level to indicate the sensitivity to continuum pumping. In

Figure 3.6a the x-axis shows the number of the level in the atom and the y-&

shows the logarithm of the level population ratios. The levels which have s m d

sensitivity, l e s than about 20% (0.08 dex) can be considered as upper levels for

hes that are not very sensitive to pumping. Our calculations show that only the

very Iowest levels (16 in 4 multiplets) are so highly collisionally dominateci that

O 1-17 2-97 3611 7 4e+17

Depth, cm

Figure 3.1: Log of H 1 and H II ionization fiactions vs. depth of the cloud.

O 1e+17 2e+17 3-17 40+17

Depth, cm

Figure 3.2: Log of He I and He II ionization fractions vs. depth of the doud.

-10 1 I t 0 t O 1 e+17 2e+17 3e+17 48+17

Depth, cm

Figure 3.3: Log of O 1-III ionkation fractions vs. depth of the doud.

Depth, cm

Figure 3.4: Log of Fe 1-IV ionization fractions vs. depth of the doud.

O 1e+17 2e+17 3e+17 4w17

Depth, cm

Figure 3.5: Log of the electron temperature (K) vs. depth of the cloud. The

Cloudy mode1 was stopped ai a zone beyond which there was no M e r

emission (the temperature in the plot plunges artificially to zero there).

Number of the level

Number of th8 level

Figure 3.6a: Log of Fe II level population ratio (with/without pumping) under

conditions s d a r to the Orion Nebula with blackbody 35,ûûOK continuum.

Nurnber of the level

Nurnber of the level

Figure 3.6b: Log of Fe Ii levd population ratio (with/without pumping) under

conditions sixnilar to the Orion Nebuia with continuum fiom al l four Trapezium

stars*

they can be diable basis for density diagnostics (these levels can be considered

as populated due to collisions). Transitions "safe" for density diagnostics are

marked with numbers 12 (~8616.96& ~8891.88~k). 13 (1112567A) and 14 (h16435A)

(Figure 4.4, Chapter 4; see also Table 3.3). As seen from Chapter 4, we do not

have any observations, ourselves, of these lines.

Calculations were made with electron temperature Te - 10,000K. As one

goes up in energy level (Figure 2.1) it becomes harder and harder to excite by

collisions, while the pumping is not so dependent on the Te value chosen. This

accounts for pumping being relatively more dominant for the upper Ievels (trend

upwards to right in Figure 3.6a). Nevertheless, even though pumped levels are

106 times overpopulated, relatively, they still can have very low populations

(because of the small Boltvnann factors) and so do not produce strong lines.

Figure 3.6b shows the logarithrnic of Fe II level population ratio as Figure

3.6a, but with continuum pumping by al1 four Trapezium stars (an estimated

70% of the total flux at ZOOA is due to 8' On C). Cornparison between Figure

3.6a and Figure 3.6b shows the enhanced importance of continuum pumping. in

this case even the lowest Ievels populations are infîuenced by pumping at nu=lB

cm-3.

The Uustrative one-zone caldations somewhat overesümate the

importance of pumping relative to collisions in the actual (more realistic) Orion

Nebula mode1 described below. One zone rnodels tend to overestimate the

radiative pumping (the radiation field is unattenuated) and the collisions are

underestimated (nw1B cm-3 compared to n ~ = l @ cm-3 in the Onon Nebula

modei). As we will see below, in the Onon Nebula, neither ~ 1 7 A (often used as

a nonnalizing he; from Ievei 14) or ~ 8 9 a (from level 15) are affkted by

pumping uables 32,3.3).

3.1.3 The sensitlvity of Fe II emisslon to electron density

To assess the behavior of Fe II emission with changing density, we

repeated the general one zone caldations for the density range 18 - 108 cm-3

and conditions othemvise similar to the Orion Nebula with the included process

of continuum pumping by blackbody radiation with temperature 35,ûOOK. To

make graphical presentations, we first integrated the total Fe II flux over ail

wavelen* (as a nomalizing value), then divided the whole range into about

18 cells with resolution 3-1û-3. Figures 3.7a-d show the emission line flues in

each cell aaoss the whole (logarithmic) wavelength range, with an upper panel

on a logarithmic scale to show all ünes, and a lower panel on a linear scale.

At densities - 1 W l W c d the spectrum is very rich throughout the entire

wavelength range. The near-infrared and far-Wared lines are the strongest, and

the optical lines, which are the focus of Chapter 4, are relatively weak. The

situation dramatically changes near density 106 an-3: the idkared lines become

very weak, while the UV lines become more important. At higher density, the

trend continues with weaker optical lines as well.

The Fe II atom structure has no permitted transitions within the lowest 63

levels. No permitted lines are seen when levels higher than 63 are not suffiaendy

populated and so only forbidden lines can be obsewed. In the case of higher

density (and/or temperature) only permitted lines are strong. For some

intermediate conditions a mixture of them is quite possible.

3.2 Prelimlnary assessrnent of the obsewed spectrum of the Orion Nebula

BauWa? Pradhan, dr Osterbrodc (1994; hereafter BPO), Bautista h

Pradhan (1995; hereafter BP95), and Bautista, Peng, O Pradhan (1996; hereafter

BPP) have suggested that [O and F e n] iines are produced by a w m (Te =

Log Wavelength (A)

3 .O 3.5 4.0 4.5 5.0 5.5 6.0

Log Wavelength (A)

Figure 3.7a: Fe II ernission line &es (normalized and binned), with an upper

p d on a logarithmic scale to show ai l lines, and a lower panel on a h e m scale,

for one-zone caldations with conditions similar to the Orion Nebula and

density 102 a+.

5 .O 5.5 6.0

Log Wavelength (A)

3.0 3.5 4 .O 4.5 5.0 5.5 6 .O Log W avelength (A)

Figure 3.7b: Same as Figure 3.7a for density lW an".

Log Wavelength (A)

3.0 3.5 4.0 4.5 5.0 5.5 6.0

Log Wavelength (A)

Figure 3.7~: Same as Figure 3.7a for d d t y 106 cm-?

Log Wavelength (A) I 1 I , 1

L 1 l 1 I

3.0 3.5 4.0 4.5 5.0 5.5 6.0 Log Wavelength (A)

Figure 3-7d: Same as Figure 3.7a for density 108 cm%

1W K) region with very high electron density (ne = 2-106 cm-3) and a solar iron

abundance. This would be surprishg for mch a dusty environment, but would

have important implications for grain destruction medianisms. Here we use

existing spedroscopic observations and more sophisticated mode1 caldations to

reexamine these daims.

3.2.1 The [O Il spctrum

[Fe II] lines are formed in the region where the ionization conditions are

favorable for existence of the Fe II ion. The ionization potential of neutral Fe is 7.9

eV, and the ionization potential of singly ionized Fe is 16.2 eV. The ionization

potential of neutral O is 13.6 eV. Therefore, [Fe II] and [O I] lines ought to be

formed in mauily the same region. Detailed photoionization models confirm this

(see Figures 3.3 and 3.4).

BP95's daims were based in part on their analysis of the [O Il intensities

reported by Osterbrock, Tran, & Veilleux (1992; hereafter OTV) and Baldwin et

al. (1991; hereafter BFM). However, in these low-resolution spectra the Orion

[O Il iines were completely blended with the tellunc emission that varies with

time and position on the sky. In the case of [O 1] 5577, the telluric emission is

much stronger than the nebular emission, and even carehil subtraction could not

yield an accurate measurement of this line. Both BFM and OTV assigned large

uncertainties to theU rneasurement of [O Il 6300, and did not daim any

significant measurement of [O I] 5577.

We have two sets of observations which dramatically improve on these

measurements. First, we used the Faint Object Spectrograph on the Hubble Space

Telescope (HST) with the 0".86 circuiar aperhw and the G570H grating on 1995

October 23/24 (UT). We made 225 s integrations at positions 1W (ai;mao, b) =

(05h35m14.7î, -05023'41''.5) and x2 (OYiW1&92, 45023'57"5), roughly 3û"

W W and S, respectively, frorn 01 On C. The summed spectrum is presented in

Figure 3.8. Line strengths were measured with Gaussian fits to the profdes and

are presented in Table 3.1 where they are given relative to He 1 6678 and are

uncorrected for hterstellat extinction. We note that the HST spectra are not

affècted by telluric emission. The diode giitch between [Cl III] 5538 and [O I] 5577

in the Bat fields does not impact our analysis.

Also, a spectnun was taken with the Cassegrain echeile spectrograph on

the 4m Blanco Telescope at CTIO, on 12 January 1996. This provided an 11- km s-l

resolution spectrum at a time when the Earth's motion split the nebular lines

from the telluric emission. The spectra were extracteci over a slit width of 1" and

length of 12.5", centered on the BFM position 2 at 37" W of 01 Ori C. The line

intensities are in Table 3.1 and parts of the spectra are shown in Figure 3.8.

Onon has a spectrum with a strong point to point variation (e.g., Peirnbert

& Torres-Peimbert 1977) causing differences in the two measurements. The CTIO

spectnim at the BFM position has a higher level of ionization. The HST spectnim

is simüar to that reported by OTV for another position in the nebda. Table 32

gives reddening corrected intemities relative to HP, using BFMs extinction at

position 2 for the C n O data; for our HST data, we adopt C(Hp) = 0.51 from the

composite Ha /HP ratio. The CTIO data in Table 3 2 were renormalized using

the dereddened He 1 6678/HP intensity ratio reported by BFM for position 2.

The CTIO [S n] 6716 /673l ratio (0.53) indicates a Te-insensitive density of

- 6600 cm4 (Osterbrodc 1989). The [N II] 5755/6583-ratio (0.018) indicates Te of

10,100 K, fairly independent of n, in this range. The HST derived properties are

similar.

Limits to the [O I] ratio R = 1(6300+6364)/1(5577) are needed to set ne for

an assumed Te. In neither spectnun was [O I] 5577 detected, but we have

1 O-" i l . , . ' . , , s , - . ,

Waveiength (A)

Figure 3.8: Spectrurn of the Onon Nebula obsewed with the HST FOS and the

CTIO 4 m echeile. The lower half shows the HST spectnun. The two inserts show

portions of the echelle spectrum. In ail spectra the [O Il lines from the Orion

Nebula are indicated by tickmarks. In the inserts the positions of the telluric [O q h e s in the çn0 data are also indicated, just to the left (blue) of the lines

produced in the nebula. In the HST spectnim [O I] 6300 is blended with [S III]

6312.

Table 3.1

* Cil0 echelle data near BFM position 2. Not dcreddened. HST observations. Not dereddened- Poorly calibrated because of broad Ha absorption in siondard star.

obtained a much stricter upper Mt. The obsewed limits are R > û4 (mu) and

R > 33 (Hm and the reddening correded ratios are R > 71 and R > 29. The

superior spectral resolution is the primary reason the CTIO spectrwn has the

more stringent limit- This iimit is very different from the ratio of 22 adopted by

BP95.

Assuming 104 K, the CTIO [O Il ratio limit sets the limit nc < 1.6 x 105 an-3,

consistent with the ndS II] and much lower than the density deduced by BP95.

Assuming n,[S II], we find that the [O I] ratio Iimit sets the limit Te < 11,600 K,

consistent with the [N II] spectrum. These are all conventional Orion [H II]

region numbers; we condude that the [O I] spectnun is consistent with its

formation in gas of moderate density (-10J -3). The photoionization

calculations presented here (Table 3.2) predict the [O I] ratio to be R = 95 and that

warm dense gas dws not contriiute to the [O I] spectrum. New obsenraüons of

the Orion Nebula ailowed us to resolve and detect the [O Il 5577 line (see

Chapter 4). The new observations give R = 77. The difference of 20% compared to

theory can be assigned to the mors of observations ([O I 55771 is very weak) and

dereddenîng corrections. Independently, this line was observed by Esteban et al.

(1999). Diagnostic diagrams there based on ratios of the [O I] ~ 5 7 7 A , 6300 A and [N I] M 3467 A, 5198 A, 5200A emission h e s indicate that the bdk of the

nebular emission in the partially ionized zone of the Onon Nebula is produced in

regions with electron densities 2-103 - 4-101 cm-3. The density derived from these

observations contradicts the high density hypothesis and confirrns the lunits

obtained here.

Table 3.2 ûereddened and Prodicteci Spectrum '

Line intcnsities arc scakd to 0.0 lI(HB), position 2 Data references: 1 - new C ï I 0 . 2 - table 7 of BFM. 3 - O n . ne - 10' cm') (A) Cdculotions for the parameters given by BFM. purnping by 0' Ori C. (B) Purnping by al1 four Tmpezium stars, gns pressure constant. (C) Like B. with totai pressure constant.

Intensities of other [Fe II] lines scûled to ï(8617). n line in a multiplet (ais0 containing 18892) little aected by pumping in this model, are given in Table 3.3.

3.2.2 The [Fe II] spectrurn

Table 3.3 shows the obsemed (extinction corrected) F e n] line ratios fiom

OTV a . Lowe et al. (1979) and some predictiow.

The daim by BPO that the [Fe II] lines corne from a high-density region in

ûrion is based on a collisional model of the [Fe II] atom, which reproduces the

obsenred optical to near-infrared h e ratios at density ne -106 a n - 3 . Table 3.3

shows the collisional model reproduced kom Table 2 of BPP (ne = 2x106 cm-3 and

Te = 101 K, using the new [Fe II] collision strengths described by Bautista &

Pradhan 1996, hereafter BP96).

A simple limit of Our Fe II model (one zone, constant temperature,

constant density, no continuum radiation flux, no line pumping, no interactions

with other species) corresponds to the BPO and BPP collisional model, and we

can reproduce in this limit the Fe II h e ratios presented in Figure 2 of BPO,

using their atornic data. Note that Figure 2b of BPO erroneously compared the

predicted 1(8892)/1(5334) ratio with the observed inverted 1(5334)/1(8892) ratio,

although their caption for Figure 2b stated the ratio to be 1(5262)/1(5159).

Our model I (Table 3.3) recaldates the line ratios for the same collisional model,

using the publicly available collisionai data from Zhang & Pradhan (1995;

hereafter ZP) also used in o u photoionization models below. This shows that the

lines AA4û15, 5159, 5262, and 5334 are stronger by a factor of two than in the

BF9dbased model in BPP because of different collision strengths. Because the

predictions of the cobional model are sensitive to the adopted collision

strengths, finely hined matching of the data is unwarranted.

Also shown in Table 3.3 are line ratios we calculated for a density more typical of

standard Onon models, n, =1P on3 (model II). At this density, there is dearly a

Table 3.3 Dereddened anâ Predicted [Fe ïIJ Wne Intensities

' Line intcnsities are scaled to ((86 17). Lowe et ai. 1979 for 12567, OTV for al1 other Iines. Collisiond model from BPP; T~ l@ K, n p 2 x 10'cm'~. ~ollisiond model h m the prerent work 7,=104 K: model 1: n,=2 xl~~crn*' . Mode1 ïï: ne = lo4 cm4. ' A, B and C models are the same as in Table 3.2. ' Lines in blends rhat would be unresoived at 4244 and 5 1 59A are not included in the BPP rnodel.

requirenient for some mechanism other than simply collisions to populate the

upper levels that produce the optical lines.

The alternative to collisional excitation, pumping of [Fe II] lines by the

incident continuum, was considered by BPP (see also Lucy 1995). The equation

used by BPP is correct in the optically thin Mt. However, the ultraviolet lines

dominating the pumping are usually optically thick and so line self-shielding

must be included if energy is to be conserved (Ferland 1992). The correct

formulation of the constant temperature problem outlllied in BPP should indude

the column density as an additional free parameter.

Tables 3.2 and 3.3 present results of our pho toioniza tion calculations.

Model A is calculated with the parameters taken from BFM with improved gas-

phase elemental abundances based on resdts hom OTV and Rubin et al. (1991,

1992): H:He:C:N:O:Ne:Na:Mg:Al:Si:Cl:Ar:Ca:Fe=1:0.095:3~10~:

7.10.5: 4 4û-k 6. 10-5: 3 * lW: 3- 10": 2 * 10-7: 4 404: 10-5: IO-': 3 -10-6: 2 408:

3 -106, grains, ionkation by the central star 0' On C, and a hydrostatic blister.

Resdts of model A differ fiom the original BFM model, since the underlying

atomic database has been improved considerably. Model B has the same

parameters as model A, but the inadent continua from the other three

Trapezium stars are included (see also the one zone models in Section 3.1.2). In

models A and B, the gas pressure was kept constant. Model C is the same as

model B, except that the total (gas and radiative) pressure is kept constant; this

reveals some sensitivity of the predided spectrum to the assumed pressure law.

Table 3.3 highlights the [Fe II] Line ratios obtained fiom our

photoionization models. A cornparbon with the observed line ratios and with the

line ratios calculated by the coUisiona1 model at the same density ne = 104 cm-3

(model II) indicates that the lower levels that produce the 8617,8892, and 12567

lines (levels 14, 15, and 10, respecüvely) are mody populated by collisions, as

antiapated kom the simple models in 3.12 Likewise, the upper levels of the

Fe II atom, those that produce the optical lines, c m be effeaively populated by

continuum pumping. Note therefore the sensitivity of the optical to near-infrared

[Fe II] line ratios to the increased pumping from the other Trapeziurn stars

(modd B versus model A, and both of these versus model II).

In proposing high electron densities in the F e II] emitting region in Otion

BPO did not consider the near infrared observations by Lowe et al. (1979). They

find that the 12567A iine is the only [Fe II] line dearly deteded in that part of the

infrared spectrum of Orion-it is aduaily the strongest F e n] iine in the infrared

and op ticd (Table 3.3). Our more detailed models at n, - 104 a n - 3 all predict that

the 12567A line is -3 times stronger than the 8617A line, in good agreement with

the observations. Lf the 12567A line were 2-3 times weaker than the 8617A line,

as predicted at ne -106 a n - 3 (model 1), it would not have been detected at all. This

is further evidence against high n,

Both photoionization and collisional models predid that F e II] and [O I]

emission Iine regions should be nearly coincident. BP95, using [Fe II] /[O I] line

ratios, inferred that the Fe /O abundance ratio in the p roposed high-density

region is dose to solar. Based on the given upper limit on the [O Il 5577 strength,

which is 3.2 times Lower than one used by BP95, the Fe/O abundance ratio

would be higher than 3 times solar, if we adopt their high dewity analysis. If

iron was depleted ont0 dust grains, these abundance ratios would be even larger.

Table 3.2 indudes the relative strength of F e I[18617 compared to lines of

other species, information cnidal to abundance analysis. Uur photoionization

models match the observations in Tables 3.2 and 3.3 quite well. The strongly

depleted Fe/H in the gas phase, 3 x1W cm4 (about 1/10 solar), is close to the

value - 27 x 106 found from KA0 obse~ations of the ground state Fe III 229p

line (EnciCson et al. 1989) and the even lower Fe/H abundance ratio determined

hom the flux of [Fe IV] 2836.56 UV h e (Rubin et al. 1997). Even with solar

abundances, iron is not a major codant in Orion, and so the intetlsities of F e III

lines scaie linearly with abundance. Therefore, if solar Fe/H (BPO, BP95) were

used instead, the predided F e II] lines would be 11 ümes stronger and very

disaepant with obsenrations.

1. Previous measutements of [O I] 5577 A were affected by telluric

emission. Our limit to the lhe sets limits on the density and temperature

consistent with the physical conditions inferred from the other emission lines.

2. The high-density collisional mode1 for the formation of the [Fe II] lines

is inconsistent with the strongest obsewed [Fe II] line at 12567 A.

3. Our photoionization models of the [Fe II] spectnim with parameters

dose to those found by BFM predict a spectrum in good overail agreement with

observations. The infiared F e II] lines are produced by collisional excitation, and

the optical F e II] lines by photon pumping by the ultraviolet continuum in a

region of moderate density, ne - lW &.

4. in combination with the upper Limit on the [O I] 5577 line, the inferred

Fe/O ratio in any putative high-density region would be at least 3 times solar,

not solar as found by BP95. Our models show thaï Fe/H must be depleted,

-3 xlW an-3, in agreement with independent deteminations from KA0

observations and UV HST measurements for different Fe ions.

5. While dense condensations undoubtedly exist within the Onon Nebula,

these regions do not contribute significantly to the observed F e and [O I]

spectra under consideration.

Chapter 4

Orion Nebula - New Observations and Emission Line Analysis

This chapter is related to two new observations of the Orion Nebula

aimed at detecting very faint lines. We wili describe the observations dong with

the signal to noise analysis, and then present an atlas of ail the observed emission

lines. F e IiJ emission lines are listed separately and compared with our

theoretical models. We also discuss the velocity field in the Orion Nebuia.

4.1 Observations and reductions

We obtained two deep, 10 km s-1 resolution echelle spectra (red and blue)

on two different nights approximately one year apart. We used the Cassegrain

echeile spectrograph on the Blanco 4rn telescope at CTIO. Data were taken at the

same position studied by 896 with earlier echeiie spectroscopy (see Section 3.2).

Our red spectrum was obtained on UT 27 Feb 1997 using a 31.6 g~ooves

mm-i de l l e grating and a 316 grooves mm-i cross disperser (grating G181) with

a GG495 order separaior fiiter. This gave full spectral coverage over the

wavelength range 5100-7485 A. The slit width was 1.0" and the decker length

137, but the area actually extracted dong the slit was 1" x 10". The sIit was

oriented in the parallactic angle, PA 1420. A series of srposures of lengths los,

lûûs, and 3 x lûûûs were taken so that measurements of strong emission iines

from the shorter exposures could be combined with those of weak emission iines

kom the sum of the three 1000s exposures without having saturation problems.

In practice, only the Hu and the [N II] lines had to be measured from the 10s

exposure, while al l other lines were unsaturated on the 1000s exposures.

The blue spectrum was taken on UT 17 Jan 1998 using a 79 grooves mm1

echelle grating and 316 grooves mm-1 cross disperser, with no order separator

filter. This spechum completely covered the wavelength range 3510-5940 A. The slit center was placed at the same position as for the red spectrum, and the slit

again rotated to the parallactic angle, which in this case was PA 2030. Again we

used a l'*.O slit width and extracted data from 10" along the süt. Eight exposutes

were taken of the Orion Nebula, with lengths 100s, 6 x 1000s, and 100s. Orion

aossed the meridian halfway through this sequence.

On both nights we also took two 500s exposures at a sky position 4m in

HA East of the Onon position. The sky spectnun was not subtracted from the

Orion spectnim, but rather was used as a reference in the identification of sky

lines in the Onon data. On eadi night we also observed two additional positions

in the Orion Nebula, which have not yet been analyzed.

The spectra were reduced using the IRAF echelle package. A wavelength

alignment was then made between the red and blue spectra by comparing the

wavelengths of 28 lines measurable in both spectra in the M. 5100-5940 A region

where they overlapped. We expected a ciifference of 10.9 km s-1 due to the Earth's

orbital motion; the actual adjustment needed was hred - h u e = 10.3 km si. The

wavelength scale was then shifted to the Orion rest frame correspondhg to a

veloaty of -59 k 0.3 km s-1 with respect to the LSR (Local Standard of Rest, Solar

system). In this rest fiame, the mean average measwd velocity of the first six

Balmer lines is 0.0 km s-1 (see Figure 4.13 below).

The flux calibration for each night was perfomed usïng specta of the

standard stars Eta Hya, Theta Crt and 108 Vir taken through a 10"-wide dit

oriented in the parallactic angle, comparing them to the low-tesolution

calibrating spedra taken over contiguous 16 A intervals by Hamuy et al. (1994).

The caliirated one-dimensionai spectra for the 1000s exposures were then CO-

added using median filtering to reject cosmic rays.

Samples of the fully calibrated spectnun are presented in Figures 4.1-42

These are purposeiy chosen to coinade in wavelength with similar plots shown

by Esteban et al. (1998), so that readers can easily compare the resolution and

signal to noise ratio (S : N) in those two sets of echelle data for Orion.

We used Gaussian fits to rneasure the peak line wavelength and full

width at half-maximum intensity (FWHM). However, it cm be seen for the

brighter lines that the line profiles are dearly nonGaussian, and so the

integrated intensities for all h e s were measured by adding the flux above the

continuum level within the wavelength interval where the lines were dearly

above the continuum noise level. The observational data are listed in Table 4.1.

The columns are the following:

obsenred wavelength in the Orion rest frame, in A laboratory wavelength, in A identification (see Section 4.3 for details)

"?" for lines with no identification, or with smail signal-to-noise ratio

(Section 4.21, or otherwise uncertain

velocity relative to Orion rest frame; velocities are not determined in the

case of bleds and for severai L e s with inaccurate laboratory wavelengths

FWHM

ratio of the observeci intensity to the intensity of He 1 A6678 A h e

reddening-corrected line ratios relative to the intensity He 1 A6678 line

signal to noise ratio

10, notes

I I I I

3640 3660 3680 3700 Wavelength (A)

Figure 4.1: Sarnple of the fully calibrated spectnim at Balmer limit region,

illustratirtg the high resolution and signal to noise. Hydrogen lines up to n=28

are detectable.

I l t 1 L

4620 4640 4660 4680

Wavelength (A)

Figure 4 2 Sarnple of the fully calibrated spectrum in wavelength range

coinadent with Esteban et al. (1998) and showing the individual emission lines

of multiplet 1 of O II as weU as much weaker N III, N II lines. A few sharp cosmic

ray hits are present too.

The line k t from one set of observations is partly overlapped with the

other one. The overlapped part is between the bold iines in the Table 4.1. Lines

within the overlapped part are marked by B (blue spectnim) or R (red spectnun).

We initiaily measured everything that appeared to be an emission line in

each order of the one-dimensional CO-added red and blue spectra. We then went

badc and inspected the position of each line in the original two-dimensional

images, which allowed us to exclude from the line list any night sky iines,

remnants of cosrnic ray hits, and artifacts due to scattered light, ghosts or interna1

refleaions in the spectrograph. The superimposed sky lines are obviously

narrower than the Orion lines and have a more uniform intensity dong the dit,

and we could check against our offset sky spectnun as needed. On the two-

dimensional image, cosmic ray hits are easiiy recognized as round spots, while

most ghosts and reflectiow are either fuuy or do not fail exactly on the echelle

orders,

A few real emission lines were blended with these ghosts, and their

intensities were measured after subtracting off the ghosts. Those lines can be

identified in Table 4.1 by the notation "after subtracting ghost".

A few of the strongest emission lines were saturated on the 1000s exposures, and

so were measured from shorter exposures (ZOOS in the blue, 10s in the red). Their

measured wavelengths and fluxes were tied into the measurements from the

lOûûs exposures by using h e s of intermediate strength measured in both the

long and short exposure data. The h e s affected are [O III] h4959, 5007 in the

blue, and Ha, II] h W , 6584 in the red.

The signal to noise ratio of each detected emission Line was estimated by

first measuring the rms pixel-to-pixei scatter in the continuum ( ~ n t i n u m , in units

of flux perpixd), and Uien comparing the Iine intensity to the expected noise

L W L IFSES-O m l c o [w L m 1 lorwo cwr-O lot

S'69 ~ O L * O 96WO LZ 2'0- I H L'09 1~180.0 ~ ~ 0 . 0 a 6.0 I H l'a 11080.0 29~1.0 a 6.0 I H * ~BBBOD LLWO n ri I H l'O9 11280'0 O S W O SZ 8'0 I H 0'6E (0190.0 W O VZ '21 I H

0.a BOSO'O 98Wl'O L'O I H L' LE csw-O sszo"o n TL I H iT*Z 6 W O 16LO'O OZ VO I H 9'81 OLZO'O 2SLO'O LZ 6'0 I H Z' t l ZOZO'O tL LO'O 61 l'O l H SEL f9L0'0 H00'0 L l '29- 1 I H 9'8 BOLO'O 1900'0 91 1%- I H

1

1 1

WON

OZ

CZ

(-1 WHMâ

I eH

01

P' LZ

L'L

IMI

t'l- L1

IW1SZlSC W86X

v ~ u m & m i

ai

SLWO

LLW'O

Jo, sLOsll

(hr 2 96VZlçC

û99'88W

v ~ u q a L i

muu

ç920'0 SM'O

.sr Wsll Lai

R t i t 10 Wmvdmgth W r h g t h ID ID? mr FWHM j A) (A ) (m 6') (km S.']

i

0.0069 0.0122 1 2 4 'ld&tiffed by Grandi (1 976)

? 60 0.0134 0.0236 8.7

O II -6.1 21 0.0168 0.0295 2.3 avg; two orâem in paor agreement ,

H 1 -0.5 26 0.6863 1.2047 486.2 avg ? 18 0.0072 O.Ot26 6.5

Hel 1 1

0.7653 1.3400 593.2 avg frorn the same

4.2 multiplet as 3604.21 2

4.9 large FWHM 0.01 05

0.0022 0.0038

--

3 13 0.0014 0.0024 5.6 He l 0.7 19 0.1863 0.3238 290.6 avg S Ill 32 0.0088 0.01531 12.6 large FWHM C II 1

3838.31 SI II 2 4 24 0 . m 0.0062 7.7

SI II 1.8 19 0.0266 0.0460 blend with pumped N Il . llne 3856.06

3833.544 3835.381

3837.729

3838,307

S 111 3 -0.2 14 0,0020 0.0035 6.6 SI II 1.t t8 0.0148 0.0255 40.9 He 1 1.1 18 0,0151 0.0260 41.3 gh [Ne III) -0.5 13 26885 4.6349 9401.0 He l 4.5 16 0.0126 0.0217 38.0 Hel 3 4.1 15 0,0019 0.0033 6.3

0.0024 0.0041 stmngest line from "Ine muiüNet

3835.384 3837.726

3838.374

H I S III

N II

?

7 -0.2 0.2

-5.2

18 26 17

22

0.01 11 1.5007 0.0043

0.0095

0.0192 2,601 1

23.6 1424.9

gh

avg gh pumped line ; possible blend with S lil

0.0075

0.0165

9.1

16.1

W-gth W8vohiigth (A) !(A)

s III He l

He l [Ne III] H I OII S 111

'

7

?

[Fe III] He l He l

0.5

-0.1

-0.3

4026.201 1 4026.1 9

6V (km

-0.3 4.2

-0.3

O

-0.2

-0.9 0.0

S III [NiIl] N 111

r

31 .O

168.2

16.3

16 17

15

He I

[S II]

O II

O II

427.1 33

4oSe.663

0.7 15.5

1

?

3

4068.600

fwnu en d)

14

18

17

17

13

25 11 14

,S 11 13 O+ooeO 0+0330

0.0045

?

11

10

13

0.0101

0.0554

0.0075

4069.801

4072148

YlCn )ob.

0 . m

0.0023

0.8

4.6

-6.0

-0.4

-0.1

4060.882

4072,153OII

4075.8601 4075.162

10.31 18.6 7.1

lm78 c#

0.0051 0.0038

0.0022 0.- 0.0013

18

36

18

34

No-

13nl

7.6

0.0037

0.0056 0.0022

6.1 32431avg

1702.01 25925'

5.5 11.6

O.ûû24

0.1820 1.3926

0.-

0,0042

0.2617

0.01 85

0.0041

0.3252

23618

141 0.0115

0.6803

0.0070

0.4338

0.0307

4.5235'

O.ûû13

OJû31

0.0191

7.6475

0.0022 0.0052

50.0 131 0.01131 0.0187

855.8

7.2

867.8

38.2

56.9

avg broad bump on He I wing, after subtracting ghost

01' 406930 lfne

Wauahgth Wnwkngîh ( A ) ( A )

Fe11 3 -126 15 0.0017 0.0020 8.4

011 3 4.8 12 0.0014 0.0023 7.7

Si 11 ? 13 0.0012 0.0020 6.2

si 11 3

NIH ? 14.7 38 0.0040 0.0065 8.7

O II -1.8 14 0.0049 0.0000 23.0 Hel 1 1 -0.41 171 0.05521 0.09001 214.4

!

He l 3.3 26 0.0092 0.0149 26.4

[Fe LI] ? 15.0 16 0.0016 0.0026 5.6

[Fe Il] 3 5 0.0006 0.0010

17 0.0033 0.0053

14 0.0039 0 . m 120

O II

same multiplet as 4128.054

large: FWHM

possible blend with C 111 41 56.504 btend

anomalously low FWHM avg avg

421 1.289

4219.747

4236.980

4241 395

4241 .762

4244,145

4249.039

4253.539

4267.168

4275554

4276.987

4284.931

S II Al III Al 111

[Ni Il] 4201.376

4205.581

?

3

4189.681 4189.70

41 89-76

4201.172

421 1.099

4219.74

4236.91

4237.05

4241.246

4241.78

4243.969

4248.799

4253,499

14.6 13

10

[Fe Il]

Ne 11 N 11

N 11 CI li N 11 (Fe II] (Nt II]

S 111

9.0

4.6

m e multiplet as 4326.237 O.OOt7

?

4253.381

0.0027

? Cl lt

0.00071 O.ûû11

135

426fM)l I~l l

426?.183I~ II

multiplet as 4241.246 avg, Mend

L I

4.4

26

? 1 0.5

426ï.261

4275.551

same multiplet as 4253,498, 4332653,4361 A68

anomaiousiy ~ow W H M

? ?

? ?

3

?

C Il

O 11

7.0

6.6

14.7

6.8 37.6

14

24

17

17

17

10.6

-1.3 12.4

16.9

0.0449

0.2 0.0025 0.0058

13

20 4276.749 4276.829

4284.904

0.0010( O.ûû16

0.0712

O II [Fe III

S II1

possible b l d with O 11 4253.74.4253.98 from the same

13

13

O.Oû1S

0,0028

0.0013

0.0074

85.1

0.0040

0,0092

1.9

0.0024

0.0045 0.0021

0.0118

10.6

18,t

0.0036

0.0014

0.0029

10.0 11 0 . m

0.0022

0.0046

7.1

11-7

[Fe H)

SI1 7 -4.4 10 0.0008 0.0013 5.5

O 11 7 -0.3 13 0.0012 0.0019 6.9

O II 16 0.0031 0.0049 15.2

O II

Fe11 ? -8.9 29 0.0031 O.OW9 9.5 lame FWHM

I

4326.W 4326.237 LN[ II] '7 15.2 13 0.0058 0.0091 m e multfplet as 4201.1 72

4332.746 4332653 S 111 6.4 27 0.0055 0.0086 128

4319.650

4325.749

4319.63û

4325.761

4352.951 4359.528 4361.545

O II

0 11

4352.n8

4359.333 4361.468

1.4

-0.9

[Fe 113 [Fe II] S III

17

15

?

0.0035

0.0020

11.9

13.4

5.3

0.0055

0.003t

19 13

8.9

0.0033

0.0103

blend with [Fe II] 4319.62

231 0.0020

0.0051

0.0160

10.6

t 8.6 0.0031 6.3

O II 1.0 16 0.0018 0.0027 1 2 1

[Fe il] ? t3.2 11 0.0009 0.0014 7.0

![NI tll) 13.6 18 0.0019 0.0028 10.5 I N II -0.4 15 0.0025 0.0037 15.9 I

10 Il -3.2 20 0.0015 0.0022 7.5

[Fe Ill] 14 0.0094 0.0138 57. t

1

4613.848

4621 355

N III -4.6 13 0.0013 0.0019 (mm muitiplet Ath strongest Hne 4640.64 ,

O II 4.3 14 0.0166 0.0241 109.0

4628.294

4630.536

4634.072

N II 0.0 151 0.0037 0.(1054 22.5 this Is the strwigest

O 11 -0.1 14 0.0266 0.0386 171.5 line of the multiplet, all lines are oôserved

O II -0.7 13 0.0081 0.0117 54.9

4613.868

4621.41 8

4621 -696

4621,722

[Fe Ill] 6.4 15 0.1246 OMO4 844.6

O Il -0.7 14 0.0091 0.0132 67.4

4621 -393

4628,046

4630.539 4630.61

4634.13

t

[Fe Ill] 24 16 0.0050 0.0072 35.91

O II -0.8 13 0.0014 0 . m x

11.5 O II -0.7 16 0.0060 0.0067 43.5 coérmic my hit oci wC-

N II

SI II

SI II

Si l t

N II

[Ni II)

N II N 111

N III

1

?

?

9

21

? -1.3

16.1

0.0015

0.0030

9

14

avg; MaMiousiy low FWHM

avg obsewed line might include SI II triplet

L

0.0008

0.0076

-3.8

0.0022

0.0008 13

, ,

0.0012

0.0111

0.0012

0.0044

from multiplet wim süongest line 4640.64

4.7

37.2

7.3

main contributor anomalously low FWHM

Cor - lm weaicest lin8 from I multi let 4649.1 35 I

[Fe III] 1 multiplet wiih 4888.6171

[Fe 1 II]

from multiplet with strongest line at 481 4.534

L

[Fe III] 4.2 15

the same multiplet as 4774,4779,4781, 4708,4793,4810 1 avg; from multiplet with strongest line at 4814,534

0 N III

wu- [Fe III]

[Fell] 3 133 12

He I 0.4 17

[Fe III] 18

O II [Fe IIfl 1 1

W8Wngth WiWangth ID J A) ( A ) 1 4947.5571 4997.3f31tFe II]

-6.0 16 0.0137 0.01 85 31 ,9 same multiplet as 4994.37

3 12 0.0024 0.00Û2 4.2 ?

5220.275 5220.297

5261.854

5261 .ES6

5269228

52ïOS3û

5270.W

5220.059

5220.059

526t .62t

5261 .a1

[Fe 11) [Fe II]

(Fe l l]

[Fe l l]

3

5268.874

52ïO-SO

52'70.4

3 [le Ill

[Fe IIQ

[Fe 1 Il]

124 13.7

13.3

13.4

20.2

7.4

8.1

19 20

15

13

20

15

15

0.001 1

0.0012

0.01 07

0.0088

0.0007

0.0638

0.0648

0.00t4

0.0015

0.01 36

0.0112

0.0009

5.5 9.9

38.2

, ,

R avg B B possiMe contribution from Cal 5261 .?O4 R possiôie conhibutlm fram Cal 526lJO4

4.1 R

0.0812 335.1 R avg; posdbie contribution hwn Cal 5270.270

0.0826 252.8 6 avg; possible ContriMion from Cai 5270.2iO

b 8 t W " W - ~ ( A )

6 can't see on 2D image B

-.Sn

5273.598 5275.306

R B same multiplet as 5159,5262 R avg: funy on 20 B

-

R sarne multiplet as 5453.855

ID wm*n* ( A )

5275.1ôô O I R

5273.346 5273.346 5275.123

R avg ; sarne multiplet

5432.797

b

BI 9h 1 R severai weak lines of

ID

[Fe II] (Fe II] O 1

this mult. pcmibly observed

10

F e l ? 10 0.0005 0.0006 5.3

S I ? 1

FWHM

13.2

14.3

average of sMet wiîh

8,

approximate, averaged for severai lines in ttie

I M i ô abs Note8

16

16

iM6ï8

1 19

0.0057

0.0053

0.0018

0.0073

0.0067 0.0023

35.3 26.4

R avg B

8.5 R avg

mnu mm m m ID " ? a.') (bm s*) Ob. car Nom

[Cl 1111 14 0.0832 0.101 1 258.4 ' blend* line IV) 8

N I ? 14 0.0050 0.0061 26.9 R

N I ? 14 Fi avg, M with stfong l in9

N I ? 14 0.0047 0.0057 B avg, bl 4th strong ilne

N I ? B avg

1 21 0.0027 O.ûO33 16J R not s m in blue spectnim

[Ci Ill] (Cl Ili] 1 this multjPlet are

R avq I R several weak tlnes off r~yp*,tm

ibl observed

I

171 0.00501 0.00601 24.81~ avg I

15.6 R B strongest Iine in multi let

15 0.0045 0.0053 15 0.0063 0.0075 55.3 R 25 0.0038 0.0045 9.5 B

N Il -1.8 18 0.0020 0.0023 120 B N Il -0.7 13 0.0018 0.0021 14.3 R Si Ill -20 13 0.W7 0.0055 44.9 R si Ill 7

-1.6 14 0.0049 0.0057 423 6 (Fe lu ? 11.7 17 0.0007 0.0008 5.5 B [N III 3.8 18 0.1727 0.2014 974.0 R avg

3.91 19 0.1- 0.1974 116226 ? 11 0.001 1 0.00-13 10.6 Flavg

AIIl ? 32 0.0017 0.0019 6 probabiy nat mal; l m FWHM

AiII ? 6 AiIl 3 B He l -0.6 22 3.0441 3.4745. 17583.2 B He I 4.5 19, 3.1916 3 . W i 18100.6 Ravg

B includes nlght sky

Fk8t wavalwigttt J A )

0.00181 0.002~~ ~;l./; nlght sky? 1 0.0016 0.0018 16.9 R 0.0017 0.0019 0.0031 0.0035 31.3 R blended wilh sky

5958.584 O I 5958.W O I

5979.047 5978.93 SI 11 5.9 24 0.0242 0.0271 125.9 avg, blend

6000.222 6W02 fNi Ill1 1.1 14 0.0027 0.0030 29.3

10 WavaImW { A )

0.0038 avg; same multiplet as L

'4 6946.4 3.6260 3.70861 96üs.7 avg

ID

6402273

6440.522 6461.836

avg; on vdng of [N Il]; 0.0849 0.0868 109.8

73.7560 75.221 1 6091 -4 avg 0.0502 0.0592 273.2 avg

11.3392 116179 22973avg

7808.0 avg

ID

r

6402.246

644ô.400

FWHM en

Ne 1

[Fe Il]

lm76

3

?

w a ow

1.3

5.7

s/N

13

No-

0.0018 0.0019

19 14

0.0005

0.0062 0.0005 0.0060

17'7 identifid by Esteban 1998

4.6

39.2 avg

Nel ? -0.2 10 0.0005 0.0005 6.4

He 1 2 0 17 0.0050 0.0047 25.2 He 1 t8 21055 1.9813 10405.8

He 1

mnu lm78 vran

[Ar lllj -0.7 131 4.6787 43507 38291.5

[Fe 111 8.6 191 0.0162 0.0150 76.4

Nom

1 , l li -i:l 161 0.~073~ 0.0068~ 4::;

[Fe Il] ? 13 O.ûO38 0,0035

-1.8 17 0.0162 0.0148 103.5

C lt O P 14 0.0508 0.0483 359.4

-

avg flux uncertain; cut by atm a b cutbyatmabs cutbyabn abs lambda uncertain, Mend with sûur~g line; m e multiplet as 7236.42: M31.33 avg , Memi or cut by atm

,wl Fut by ami abs

avg; rtmngert h multiplet same multiplet as 6700.59 same multiplet as 6179.94, but strongest line (6783.91) is missing; amalously low FW HM

? 15.4 11 0.0015 0.0015 15.4 avg

? 18 0.0017 0.0017 avg; measured after 'OS tubtracthg s b ? -7.0 16 0.0016 0.0016 avg; measured after O.' subtractfng sky

? 13 0.0007 0.0007 4.4 avg; after subtmctlng sky: doublet of SI Il

? 14 0.0015 0.0015 13.3avg ? 10 0.0006 0.0006 7.1 atter subtracting sky ? 16 0.0019[ 0.0018 14.7 avg

-1.4 14 0.0013 0.0013 12.0 avg

? -1.2) 15 0.0008 0.0008 7.0 avg

? 1 12 0.0009 0.0009 8.3 avg 15 0.0027 0.0026 14.3

l l ? 12 0,0008 0.0008 'same multiplet as 4.0 4586.84.6533.8

l 2.6 20 0.0032 0,0030 17.7 avg

24 0.0194 0.0184 avg. blend or cut by atm abs *

ga

'Og7

6.7

? 12 0.0010 0.0010

? -0.4 14 0.0010 0.0010

? -0.3 9 0.0005 0.0005

[O 111 ( 0 111 ( 0 ID [NI II) [Fe II] . ![Ni 19 N 1

C) 7.6

ID? r - ~

Wmkngth ( A )

7319.108

- 7452782

7468.689

FWCIM (bn C)

18

8.0 2 4 3.8

120

8.8 15.5 14.6

10 Wawîongth (A)

7318.92 17

16

18

20

20 13

11

745254

746ô.312

ID

10 111

[Fe II] N 1

9.7

15.1

19

13

level taking into account both the measurement error based on the number of

pixels across the iine and the error in determinhg the continuum level. For the

signal to noise in the emission line flux we used the relation S : N = Iiim/aünr =

Ibe/[aontjnuum -4 n ü , ( 1+ nune / nmntinum)]. Here nün is the number of pixels

induded in the line profile (estimated as 2eFWHM in units of pixels), and

nconanuU = 10 is typical of the number of pixels averaged together to determine

the continuum level. Our adopted expression for S : N should be approximately

corne& for weak h e s where the continuum noise dominates. Visual inspection

of plots of the actual spectra showed that the resulting S : N values are about

what is suggested by the appearance of the data, with lines having S : N < 3.5 at

the noise k t , while lines having S : N > 7 are dear detections. We therefore

rejected d l lines with S : N < 3.5, and kept lines with 3.5 < S : N < 7 in our line list

but with a "?" notation in column 4 of Table 4.1.

The calculated signal-to-noise is given in column 8. Note that for strong

lines the signal-to-noise is grossly overestimated, but this does not matter

because we are interested ody in having an objective criterion for including or

excluding lines at the faint limit. In cases where an emission line was detected in

two orders, the entry in Table 4.1 was computed by taking the average value of

each measured quantity (including signal-to-noise) if either the two individual

signal-to-noise measurements were within a factor of 2 of each other, or both had

signal-to-noise > 24. Sudi cases are marked "avg" in the Notes column of Table

4.1. If these aiteria were not met, we just used the measurements having higher

S : N. The reason for taking an average rather than weighting by S : N is that the

diffaences between measurements are more likely to be dominated by

systematic enors (partidarly in the flux calibration) than by noise.

4.2 Analysis of errors

Here we assess the uncertainties in the present measurernents first by

studying the intemal consistency of the data set, and then by comparison to

previous remlts.

Particularly in the case of the blue spechum, the d e l l e flux calibration is

liable to have been affected by scattered üght (grating ghosts plus internal

refiections in the spectrograph optics), and by crowding of the orders in the UV.

In the wavelength range where the blue and red spectra overlapped, the mean

blue/red flux ratio is 1.11, but this is strongly influenced by noise in the

measurements of the many weak lines included in this average. Using only the

five strongest lines, the mean ratio is 1.00, while for He 1 AS876 (by far the

strongest line in the overlap region) the ratio is 0.95. We concluded that the flux

calibrations of the biue and red spectra are in good agreement.

In the red spebnun, there are 37 cases where the same nebular emission

line is recorded in more than one order. The lovelocity scatter between these

duplicate measurements is 21.5 km SI, and the mean ratio of stwigths is 1.0,

with a lascatter of M.2. This method depends on iines appearing in two orders

and so tends to compare rneasurements taken far off the echeile blaze, where the

flux calibration is worst. It therefore overestimates the typical error. By contrast,

intensities of intermediate strength lines measured fiom exposures of different

lengths agree to about 3%, but of course in this comparison flux calîration mors

cancel out.

The blue spectnim has approximately the same internal accuracy as the

red spectnun. Comparing Unes measured in two orders in the blue spectmm (35

cases), we fhd that the lavdoaty scatter is 11.5 km si and the mean fh ratio is

0.97 with a 1 ascatter of I0.X (again an overestimate of the typical error).

We can also make extemal checks. Table 4.2 compares out new

measurements to previous remlts for some of these lines measured at this same

location using both the same echelle spectrograph (B96) and at nearby positions

using the low-resolution RC spectrogaph @FM) on the Blanco Telescope. The

agreement with the previous echelle data is in fact very good; comparing the h e

strengths normalized to He 1 h6678, the mean ratio between the new and

previous results is 1.01, with a luscatter of H.05. This shows the measurements

a i the same location using the same instrumentation and calibration method are

reproducible at 5% level of accuracy.

However, comparing the low-resolution spectra taken at nearly the same

position (BFM position 2), the mean is 1.06 with a 10 scatter of M.14 (leaving out

of this tabulated comparison [01] s577,6300 and 6363, which are blended with

night sky lines in the low-resolution data, and Cm A4652 which in the low-

resolution data is a blend of many weak lines). Table 4.2 also includes a

comparison to the low resolution results at BFM positions 1 and 3, which lie 8" E

and W of position 2, respectively, in order to give an idea of the differences

which can be expected from small positionhg changes. As weii as different

centering, we also have different orientation of the slit dong which the spectra

were extracted: EW in case of BFM and at different parallactic angles here. In

addition to measurernent errors, real dïlferences in the spectrum may be

important. These Merences can be in the intrinsic spectrum (e.g., higher

ionization) or c m result from different foreground reddening (BFM).

Considering ail of these interna1 and external checks, we condude

consewatively that due to systematic errors in the del le flux calibration, the

relative strengths of individual intermediate and strong lines generally may be

accurate only to f1520%.

- - [Ne 111) 3869 1 [ 1.14 1 1.12 1 1.19

- - i Mend 5538 1.09 / (N Il] 5755 (blue) 1 1 1.26 1.27 1 -74 [N 111 5755 ( r d ) ( 0.96 1 1.23 1.24 1 .71

k 1 [N Il] 6584 1.08 1 0.97 1 1.19 / 1 . ] L *

He 1 5876 (blue) 1 / 1.W 1 1.07 1 1.16 1

! blend 671 7+6731 1 / 0.92 1 0.96 ! 1.38 i

He 1 5876 (red)

10 11 - [S 111) 6312 (SI II 6347

1.00 1 1.01 1 .O3

1 -03 0.99

' 1.05 1 1.13

I 1 1 I I

1 1

0.97 mean 1 -06 1.30 1 1 .O1

std d~ 0.05 0.15 , 0.14 0.27 1

The flux caübration potentidy has an additional problem near the

positions of the Balmer lines, because the standard stars are all white dwarves

with very broad hydrogen absorption lines whose wings cover aknost half of an

echelle order. This is why we use He 1 6678 A as a normalizing iine instead of

HP. Table 4.2 shows that the echelle measurements of the lower Balmer lines

(Ha, HP, Hy) have the same sort of scatter compared to the RC spectrograph

values as do the other hes , and so this problem does not appear to be espeaally

senous for well-sepaated Balmer lines. However, near the Balmer jump the lines

in the standard star spedra overlap. Comparing the intewity ratios of the higher

Balmer lines as given in Table 4.2 to the predictions of Hummer and Storey

(1987), it is quite clear that the echelle order which indudes iines higher than Hl5

(3711A) and the Balmer jump does have significant flux calibration problems.

4.3 Line identifications

Table 4.1 lists 444 emission lines measured in the 3498-7468A range for

cornbined sets of observations. Of these we were able to iden* 411. From them

37 lines are seen in both spectra, blue and red. Therefore, the total number of the

identified distinct lines is 374. A number of lines are blends or have several

possible identifications.

To identify these lines, we use atornic data from compiiations by Moore

(1972), Wiese et al. (1966,1969,1996), Fuhr et al. (1987), Vemer et al. (1996), on-

line databases (van Hoof 1999, Vemer 1999, and NET 1999), and earlier

identifications of lines in the Orion nebda made by Kaler et al. (1965), OTV, and

Esteban et al. (1998). We assumed the standard chernical abundance in the Orion

Nebula, and searched for lines fiom the most abundant elements fmt. We also

asnimed that the caiculated velocities for lines fiom the same multiplet are

within the mors of our measurements and within 15 km si of one another. To

NigM sky

Orion

Wavelength (A)

Figure 4.3: Spectnun with strong teiluric emission and nebuiar A5577 [O I]

emission line (Orion test frame).

make line identifications we paid special attention to branching ratios in

multiplets and checked if other hes, espeaally the strongest ones, were

obsewed. The search for other possible theoretical Unes fiom the same multiplet

allowed us to avoid some misidentifications.

The red observations were made purposely at a t h e when the Earth's

orbital velouty caused a maximum separation between the night sky h e s and

the same features from Orion. The [O Il A5577 line is dearly separated from the

night çky feature (by +49 km si) and is detected with reasonable signal to noise

(Table 4.1 and Figure 4.3). The blue observations are very similas.

4.4 [Fe II] lines in the Orion Nebula

4.4.1 Line list

Using our mode1 calculations, we compiled a Iist of [Fe III lines that c m be

expected kom the observations of the Orion Nebula. Based on this iist, we

identified 40 distinct forbidden lines of Fe II. From hem, 7 Iines are obsewed in

both spectra, blue and red. This is the largest number of [Fe II] lines ever

obsewed in the Orion Nebula. Note that the previous record, 21 identified [Fe II]

lines, was set by recent echelle observations by Esteban et al. (1998). Table 4.3

contains F e II] data in 8 columns:

obsenred wavelength in the Orion rest hame, A laboratory wavelength, A "?" for lines with low signal-to-noise ratio (3.5 c S : N < 7)

calculated veloaty, km si, relative to the Orion test frame

FWHM,kms'

reddening-corrected line ratios relative to the intensity He I -78 h e

signal-to-noise ratio

notes-

I Tibk 4.3 O b m e â [Fe II] Emirriori L W (Rddening Comctd)

possible overiap with 4452.378 011

-. . - - - - -

aftersubtracting ghost from multiplet with I I lines, strongest at 441 6.27

R u t Wmnkngth

( A )

L

possible llne from multiplet with stronqest line at 4889.63

10 7

ID Wlinkngth

( A )

FWHY (km 8.')

t

20

13.5 12

4276.987

4287.587

8V (km a*')

4114.566

4177.405

4179.218 4211.289

1 4244.145

SIN -6

4276.029

4287.394

? ?

? ?

4114.470

4177.196

4178.958 4211.099

4243.969

mt.8

I

0.0092

0.0232

J

4889.833

4905.555

4947.557 I

15.6

19.0

possible blend wiîh O 11 4276.748 18COmbinatIan posslble blend with O 11 4287.727 -mMnsUon

4889.61 7

4905.339

4947.373

5.7

4.6 3.2 3.2

31.1

0.0025 0.0026

0.0010

0.WOB

0.0118

7.0

15.0

18.7

strorigest lin0 in multiplet 20

16

5

?

4951 .O38 4950.744 1 17.8

13.51 8

t2.41 17

34

20 ---- 12

21

21

19

20

13.2

11.2

16.0

13.3

10.1

4973.653 4973.388 ?

0,0035

5158.951

12

16

5.4

5158.777

5158.956

5220.275 5220.297

0.0023

O.Oû4Q

0.0036 3.1

5158.t77

5220.060 5220.059

0.0038

0.0182 0.0147

0.0014

0.0015

229

8.8

J

avg; fm multiplet with Dtmngest line at 481 4.55 but posstbly [Fell] 4889.70 from a dlfferent multi piet

21.1

49.6

49.7R

4.7R

8.6

? 10.4 124

13.7

6

B A

W m i r n g t h Winkngai ? (knr"] ( A ) ( A )

0.01 12 56.3 R

0.0073 28.9 R avg 6 possible contribution from Ca I

'*Oûû7 21.4 5261 -704 3DûP 3.8 R not seen in blue spectnim

5269- 6 can't tell if reai; from same multiplet as 5273,5269 6

,tlw ': uncertain; cut by aimi abs

Figure 4.4: The lowest 43 levels of Fe II and the obsewed multiplets in the Onon

Nebula. We did not show each observed line in diagram for simplicity. The

numbers in diagram correspond to the Moore classification of multiplets shown

here in parenthesis, whidi contain multiple lines (in A). 1 (7F) - 4287,4359,4414,

4452,4475; 2 (4F) - 4728,4889; 3 (18F) - 5269,5273,5433; 4 (6F) - 4416,4458,4493,

4515; 5 (20F) - 4775,4815,4905,4947,4951,4973; 6 (21F) - 417ï, 4-44, 4277,4353; 7

(34F) - 5747; 8 (19F) - 5112,5159,5220,5262,5297,5334,5376; 9 (none) - 6440; 10

(23F) - 4114,4179,4211; 11 (14F) - 7155,7172,7388,7453; 12 (13F), 13 (none), 14

(none)- show observed near-infrared multiplets (not from current observations).

A U [Fe lines are within the velocity range 13fB km SI (see Figure 4.15

below). AU the obsemed lines are produced by transitions within the f h t 43

levels (see Table 4.4 and Figure 4.4). In some multiplets we see up to 6 hes .

Within al l the multiplets, the strongest lines predicted by the model have been

obsemed.

4.4.2 Cornpariaon with model predictionr

The new observations give another opportunity to check our theoretical

models (Chapter 3,896). Table 4.4 shows the atomic data for the observed [Fe II]

lines in the blue and red spectra, observed line ratios relative to the intensity He 1

A6678 line from both spectra, and predidions from our models. The Table

contains data in 14 columns:

1. Fe II laboratory air wavelength, A 2. Moore multiplet nomenclature

3. multiplet identification

4. order number of lower level

5. order number of upper level

6. energy of lower level, cm-'

7. energy of upper level, an-'

8. statistical weight of lowet level

9. statistical weight of upper levei

10. transition probability, s-1 (Quinet et al. 1996)

11. reddening-corrected line ratios relative to the intensity He 1 h6678 line

from the blue spectnun

12. reddening-conected line ratios frorn the red spectnim

13. predictions from pumpmg modd (mudel C, Chapter 3)

14. predictions from collisional model (model II, Chapter 3).

Tabk 4.4 Obmweâ and Pmdictod [Fe II] Une Intondtb

Figure 4.5 shows a comparison of the intensities of the obsenred F e II]

lines with the theoretical intensities calculated in the frame of model C (Chapter

3). h e s at 4277 and 4287 A are not exduded from comparison, ahough their

intensities might be enhanced by the contribution of the O II tecombination lines.

Generally, the predicted intensities are within 60% of the observed ones.

Agreement in the red part of the spectnim is better than in the blue part. Overall

agreement is consistent with the uncertainties of observations and the

dereddening corrections combined with the quality of the involved atomic data.

There are no strong theoretical lines in model C which have not been obsewed.

The theoretical intensities of [Fe II] lines in the pure collisional model (model II,

Chapter 3) are also shown. Clearly, the pumping mechanisms are important to

get the mean of the predicted intensities closer to the observed values.

Figure 4.6 plots all the predicted intensities vs. the observed ones, and

there is no spetial trend dependhg on intensity of the line.

Figures 4.7-4.11 demonstrate the comparison for the strong multiplets, in

the format of Figures 4.5 and 4.6. In the strongest multiplet, a4F-a4H, 7 lines have

been identified, and 5 from hem have been obsewed in both red and blue

spectra (as shown by the pairs of points). The a4F-a2G, a6Db4F, and a6D-a6S

multiplets have 4 observed lines each. Note that alî the theoretically strongest

lines in multiplets have been observed. AU the theoretical intensities of the lines

from asPa% multiplet are 50% weaker than the obsenred ones. Obviously, there

is a missed exatation mechanism for the a% level (see Figure 4.4), or collisional

data are wrong for this level.

Figure 4.12 presents the ratio of the predicted intensities to the observed

ones against the energies of upper levels of the transitions. There is only a slight

trend that the higher levels are underpopulated. This dependence is much

stronger in the pure collisional model. Definiteiy, density is not the only factor to

determine the population of upper levels.

4500 Sûûû 5500 6000 6500 7000 7500

Wavelength (A)

Figure 4.5: The ratio of the predicted intensities of [Fe II] lines to the o b s e ~ e d

values for mode1 C (with purnping, filied &des) and mode1 II (no pumping,

open citdes). The data for h e s obsewed in both blue and red spectra are joined

with a bar.

0.000 0.005 0.010 0.015 0.020 0.025 Obsenred intensity, l(line)/1(6678)

Figure 4.6: The predicted intensities vs. the observed intensities of F e II] lines

mode1 C.

Figure 4.7: The ratio of the predicted intensities (mode1 C) to the obsewed values

for the [Fe II] a4F-aJH multiplet. The data for h e s obsewed in both blue and red

spectra are joined with a bar.

2-0 .

1.5

VI O O

c I

p 1.0

4 - 0.5

0.0

t b I 8 t

- T 1 a

1 a

I

& a

- 1

- . , 9

5100 5150 5200 5250 5300 5350 5400

Wavelength (A)

0.000 0.005 0.010 0.015 0.020 0.025

Observed intensity, l (line)/1(6678)

Figure 4.8: The predicted intensities (mode1 C) vs. the observed intensities for the

[Fe a aJF-a*H multiplet.

Figure 4.9: The same as Figure 4.7 for the aAF-a2G multiplet.

2.0

1.5

V) P O

I 1.0

r -1 -

0.5

0.0

I 1 1 1 I 1 L

-

a

O

-

I I 1 ! 1 1 1

7100 7150 7200 7250 7300 7350 7400 7450 7500

Wavelength (A)

4400 4420 4440 4460 4480 4500 4520 4540

Wavelength (A)

Figure 4.10: The same as Figure 4.7 for the a6Dbf multiplet.

4250 4300 4350 4400 4450 4500

Wavelength (A)

Figure 4.11: The same as Figure 4.7 for the a6DabS multiplet (single upper level).

t4000 16000 18000 20000 22000 24000 26000 28000

Energy of the upper leval. cm"

Figure 4.12: The ratio of the predided intensities of F e II] iines (mode1 C) to the

obsemed values vs. the energy of upper levek of the transitions. Linear

regression is shown by a straight he.

4.5 Velocity field analysis

4.5.1 Introduction

Emission lines from ions with different ionization potentials originate at

ditferent positions dong the line of sight to the PDR. As an example, Figures 3.3

- 3.4 (Chapter 3) show the ionization structure expeded for 01-III and Fe II-III in

an ionized cloud versus the depth for a typical mode1 of the Orion Nebula

(Baldwin et al. 1996) calculated using the pho toionization code Cloud y (Ferland

et al. 1998). The goal of this section is to use the velouty structure of the nebula to

identify experimentally which lines corne from which ionization zones.

Previous work by Balick et al. (1974) shows, from the dependence of

velocities on ionization potential for a few speaes, that the gas is flowing off the

blister and accelerating toward us. High ionization lines (S3+, @+, and NeZ+) have

negative velocities from about -2 to -1 km s-1 (blueshifted relative to the LRS);

intermediate veloaties are observed for intermediate ionization lines (N+ and

O+), and positive veloaties of about 9 km s-1 are found for low ionization lines

(Fe+ and C+).

Our atlas adds many new lines and additional ions to this pibure. There

are a few ions with a large number of emission lines. We have information for

about 47 [FeII] emwion lines from both spectra, 41 lines of H 1' 39 lines of He 1,

32 lines of N II, 56 Iines of O IL, and 17 hes of [Fe m].

Veloaties of H 1 lines (Figure 4.13) have values of O& km s-1, with a few

exceptions.

AU He 1 lines (Figure 4.14) are withui 0 6 km si.

Forbidden collisionaIly exated Iines, permitted recombination lines and

pumped lines originate from different regions in a doud. For example, [[O a

5000 6000 Wavelength (A)

Figure 4.13: H I veloaties in the Onon Nebula.

lines arise in the O II zone, O II lines formed by recombination arise in the O III

zone, and O II Iines which are pumped arise in the O II zone. The recombination

contribution to [O II] h e s arises in the O III zone, but this contribution is very

small.

For the neutral lines of interest, collisional excitation will depend on

temperature (as for ail iines), but the temperature falls off past the H I front

(Figure 3.5), and ço much of the extended neutral O 1 region in Figure 3.3 does

not produce strong lines.

In the following subsections we will study separately forbidden and

permitted hes.

4.5.2 Forbidden lines

Most forbidden iines are collisionally excited in the Orion Nebula

(Osterbrock et al. 1992). The radiative-collisional mode1 predicts that the

emission kom different ionization stages cornes from different parts of the Orion

Nebula and depends on physical conditions (temperature and distance fiom

central radiative sources). The forbidden lines of nine different ions from our

observations were used to investigate the dependence of velocity on ionization

potential.

F e n] lines (Figure 4.15) have a range of veloaties from 6 km s-1 to 22 km

s-1 with most of them between 10 and 15 km sl.

The obsewed F e III] lines (Figure 4.16) Lie between 4000 to 560d and

show a systematic increase in their vdoaties with their wavdengths. No other

ion shows this, and so it is not an instrumental effeb. Figure 4.17 plots the F e III]

velocity agauist the excitation potential of the upper level of the correspondhg

line. We see a definite trend: the higher the exatation potential the lower the

velocity- The most probable reason for this trend is the foilowing.

Wavelength (A)

Figure 4.15: F e II] velouties in the Orion Nebuia.

Wavelength (A)

Figure 4.16: Fe III] vdocities in the Orion Nebula.

I I 1 L 1 8 , , . l , , , , ~ , I

2.4 2.6 2.8 3.0 Excitation potential (eV)

Figure 4.17: F e III] velocities vs. excitation potentid of upper Ievel.

The fiaction of Fe III is up to 4 orders of magnitude higher than the fraction of Fe

11 in the ionked zone (see Fig. 3.4). Therefore, we can observe the [Fe lines

from a much wider range of distances from the ionizing star than we can for the

II] Lines. Different exatation conditions are dominant at different distances

from the star. The doser to the star the easier it is to exate the higher levels. We

cannot see this effect on the II] lines, since they corne fkom a much narrower

region than the [Fe III] lines do.

Figure 4.18 plots the average velocities of various forbidden lines against

the ionization potentials of the corresponding speaes. Two distinct clusters are

obvious. The first duster has ions with ionization potentials les than 20 eV,

namely O 1, N 1, Fe II, and Ni II. The lines of these ions have velocities from +10

to +15 km SI. This is different from the second duster, which includes ions with

ionization potential larger than 20 eV (S II, N II, Fe III, 0 II, and O III). These iines

have veloaties around 3 km s-1, with a süght trend of deaeasing velocity with

increasing ionizaüon potential.

The similar ionization potentials suggest that collisionaîIy exated Lines of

Ni II, Fe II, N 1, and O 1 should al l form in the same region, and this is confirmed

by the fact that lines of these ions have similar velocities. Therefore, the average

conditions where the [Fe II] Lines originate shouid be close to the density and

temperature determhed from intensity ratios of the (N 11 Lines and the [O Il

(Section 3.2).

Velocîties of all iines preliminarily identified as N II and O II Iines are

plotted on Figures 4.19 and 4.20, respectively. Most of the permitted N II and O II

lines are obsezved at similar veloaties, about O to - 5 km s-1. In a few cases, the

measured velocities are quite different fiom this average. There are several

possible explanations. First, the four lines with !VI> 10 km sl are probably mors.

20 30 40 50 lonization potential (eV)

Figure 4.18: Veloaties of forbidden lines vs. the ionization potential.

Errors are standard deviations of the distribution of veiocities from

severai lines for a given ion.

4000 5000 6000 Wavelength (A)

Figure 4.19: N II velocities in the Orion Nebula.

W avelength (A)

Figure 4.20: O II velocities in the Orion Nebula.

Figure 4.21: Veloaties of al l emission iines observed in the Orion Nebula. The

arrow shows CO veloaty taken from Balick et al. (1974). The middle line is a

median, and boxes and error bars are 10% 25% 75th and 90" percentiles.

Velocities out of the 10th and 90th percentiles are shown by dots.

Second, for the milder deviations the lines may have a Merent ongin

(recombination versus radiatively purnped lines). These contriiutions to lines

may corne from the different parts of the Orion Nebula. This should be taken into

account if these lines are used for temperature and density diagnostics.

Cloudy calculations show that the origin of permitted O II lines is mainly

due to recombination from O III. Moreover, it is seen (Figure 4.21) that the range

of O II emission line velocities overlaps more closely with that of [O III] lines

than of [O II]. We have poor information about the N 1, [N 11, [N II], N III ion

velocities. Available N 1 and [N Il veloaties are close to each other. Probably N 1

emission is due to purnping rather than recombination. Cloudy calculations of

N II lines indicate that the recornbination contribution should be much higher

than that due to pumping.

Figure 421 summarizes in pidorail form al1 the veloaty information. Al1

elements are piaced in order of increasing atomic number. The middle line for

each symbol is a median, and boxes and error bars are IO*, 25% 75th and 90th

percentiles. Velocities outside the 10" and 90th percentiles are shown by dots. We

exduded from this plot lines with very discrepant velocities. They are either due

to misidentifications or due to uncertain position of line center in unresolved

blends and low signal-to-noise ratio lines. We also show by anow the velocity of

CO molecule taken from Balick et al. 1974. The CO emission cornes from the

background Giant Moledar Cloud.

Figure 4.21 summarizes the preceding discussion, showing that:

1. Most of the obsewable ions have veloaties in the range 0 6 km s" (lines of

H 1, He 1, N IL [O IZI, [O ml, 0 n, [Ne ml, [Sul, [Si nl, [Si ml, [Ar III1 and

[Ar IV], F e a). 2. Emission from another set of ions which includes lines of [NI], N 1, [O Il,

Fe ITJ (and probably mi II]) mainly arises in the velocity range 13I3 km SI.

The CO velocity of the background cloud (Balick et al. 1974) is close to this

value.

3. For the same chernical element the higher ionization stages have the

smaller veloaty: [N Il, N 1, [N III, N II, N III; 0 II, [O a, [O II], and [O ID]; Si II, Si III, and Si IV; and S II, S III, and [S Ilj, [S m], and [Fe II], [Fe III]. The

[Ar Iül lines and [Ar IV] show the opposite behavior, but this conclusion is

based on very poor statistics.

4. This is consistent with a dynamic mode1 in which lines of ions with

different ionization potentials originate in gas flovving away from the

ionization front toward the ionivng stars (and the observer too in this

case). The ünes from the species that are most ionized originate in the

most apidly moving (relative to the ionization front) gas dosest to the

stars.

Chapter 5

Fe II Emission from Broad Emission Line Regions of Active Galactic Nuclei

5.1 General picture of AGN and obsewational tacts

Active galaxies are distinguished from other galaxies in that they show

indications of having energy output not related to ordinary stellar processes. The

activity is centered in a small nuclear region and associated with strong emission

lines. The nuclei of such galaxies are narned Active Galactic Nudei (AGN). This

group of objects indudes bright quasars, with luminosity exceeding 1047 erg s*.

The emission iine spectra of AGN divide broadly in two dasses. One of

them is characteristic of the "Broad Ernission Line Region*' (BELR). The electron

density in the BELR is at least 108 &, as judged from the absence of strong

broad emission forbidden lines, and the typical gas velocity is 3000-10000 km si.

The second region is the 'Warrow Emission Line Region" (NELR) with narrow

permitted h e s of H, He, Fe II, Mg Il, C IV. Its typical density is 105-106 cm-3, and

the gas velocity 300-1000 km si. Photoionization is the most iikely source of

excitation for the emission hes gas in BELRs. The intensef nonstellar continuum

is dwctly obsewed in many objects. The correlated continuum and (delayed)

line variations in many Seyfert 1 galaxies, where the üne luminosity changes are

ciearly a response to the continuum luminosity changes, is direct evidence for

photoionization (Peterson 1993) and the scde of the system. Moreover, the

observed C III A977/CIII] A1909 line ratio is a good BELR temperahire indicator

for densities (below 10'0 cd). The observed line ratio in several quasars shows

that the electron tempera- is below W X H l K in the C2+ zone. This temperahire

is less than that required to produce C2+ by collisions.

One of the most discussed but highly idealized and simplified pictures of

a photoionization model of AGN is the following (Netzer 1990). The s m d

continuum source with luminosity about 1014 erg s-l< LBH < 1W erg s-1 is around

a supermassive blackhole with mass range 106Mo < MBH < logMo. The

continuum source is surrounded by a much larger emission line region; typical

distances from central object to the douds are about 0.1 pc. The electron density

of the douds range hoom 109 cm-3 < n~ < 1012 cm-? Figure 5.1 shows this simplified

model.

The ionization structure of the BELR is very different from that of H II

regions and planetary nebulae. The douds can be thick enough to be highly

ionized (W, P) at their illuminated face and almost cornpletely neutral at the

badc. The large flux of X-ray photons maintains a low degree of ionization

(-10%) over the large part of the doud, more than 90% of its thickness in some

cases. Such extended low ionization regions are thought to be the ongin of the

strong Fe II and Mg II Unes. This is illustrated in more detail in section 5.2 below.

A composite quasar spectrum is shown on Figure 5.2. Prominent

emission lines are indicated. There are a lot of sttong features of Fe II on both

sides of the Mg II line and in parts of the optical spectnirn. The conglomerations

of the strong Fe II lines, the Baimer continuum, and other spectral features,

aeates a noticeable energy excess between the wavelengths of ~OOOA and 4000A.

This kature is sometimes refkrred to as the "smd bump".

The Fe II spebnim is one of the unsoIved problems of AGN study.

Because of the extremely compiicated energy level configuration of the Fe II

P h o t o i o n i z a t i o n m o d e l

S u p e r m a s r i v e b l a c k h o l e

Figure 5.1: A general basis for the photoionization model of AGN.

atom, üne intensities are not simple to caldate. There are several hundreds of

transitions of Fe II to be considered, many with large optical depth. An

additionai complication is the large number of wavelength coincidences of

düferent Fe II lines. Their number at high densities (>lOl*cmJ) can reach more

than 1000. A very large number of Fe II lines which cannot be resolved

individuaily may produce a quasi-continuum in some wavelength intervals. As

discussed by Netzer (1990), the large number of Fe II lines fom severai distinct

emission bands at 2200-2600 A, 3000-3400 A, 4500-4600 A and 5250-5350 A.

Line overlap and self-pumping is a very important population process for

the Ievels that must be taken into account. Figure 5.8a demonstrates the

complicated nature of the Fe II spectrum. It shows a calculated Fe II spectrum,

for an AGN doud, with 65638 lines. The total observed strength of these lines

could reach that of Lya, while the calculated strength is only about 1/3 or 1/2 of

that. The ratio of the optical Fe iI lines to the hydrogen Balmer lines presents a

similar problem and there is also a difficulty in explaining the observed ratio of

optical Fe II lines to ultraviolet Fe II lines. Suggested explanations, within the

general framework of photoionization, indude very high densities, large iron

abundance and emission from the outer regions of central accretion disks

(Netzer, 1990).

This section gives an overview of the effects of Fe II emission upon the

thermal structure of a typicd Broad Ernission Line Region (BELR) doud of

Active Galactic Nudei (AGN). The Fe II h e s are treated within a

photoionkation model whose parameters are those that have often been

adopted. The central part of AGN contains a massive object, possibly a black

hole, with a very small region around it where the optical - UV and X-ray

continuum originate. The douds were illuminated by a spectral energy

distribution thought typical of AGN (Mathews & Ferland 1987).

The conditions were chosen so tfiat low-ionization h e s like Fe II will

dominate the emission. Solar abundances for the base elements and constant

density were used. The hydrogen density is n~ = 10'2 cm-3, the column density is

10n cm-2, the flux 9 of hydrogen ionizing photons at the üluminated face is

3x1019 ar2 s-l, and these correspond to an ionization parameter U(H) = / n~ c,

of 1û-3. Figure 5.3 through 5.7 show the results of model calculations. The plotted

quantities appear discontinuous at the hydrogen ionization front since

conditions change rapidly there - these regions were actually well resolved in the

calculations. Most of the foilowing applies to emission of nebulae, planetary

nebulae, the Broad Line Region, novae at later stages of evolution, and sorne

symbiotic stars.

5.2.1 The ioniution fractions ot Ho, H+, ~ e ' , He', and He2+ vr. depai

The ionizing radiation from the central source aeates an H+, He+ zone,

and a partly neutral zone where HO is dominant. Figure 5.3 shows how the

ionization fractions of HO, H+, Heo, He+, HeZ' change with depth into doud. H+

and He+ are alrnost coînadent.

log Depth, cm

Figure 5.3: Log of the ionization fractions of HO, H+, HeO, He+, and HG+ vs. depth

within the cloud. The ionization fionts appear as discontinuities, but are weli

resolved in the numerical caidations.

The reason for the sharp dip in H+, He+, and H$+ cwes is the

photoionization of the n=2 level of hydrogen by the Balmer continuum and the

corresponding texnperattue fluctuation (see Section 5.2.5).

5.2.2 The ionization fractions of Fe 1-Fe V vs. depth within the cloud

Figure 5.4 shows how the ionization fiactions of Fe0 - Fe& change with

depth into doud. The ionking radiation fiom the central source is responsible for

predominant F e and F e fractions in H+ zone and predominant Fe+ fraction in a

partly neutrd zone.

5.2.3 Total and Fe II cooling vs. depth within the cloud

The total and Fe II cooling are shown in Figure 5.5. As can be suspected

from the figure, Fe II is the most important coolant behind the ionization front,

where Fe+ is the main fraction of iron. In fa&, very near the hydrogen ionization

ftont, Fe II carries essentially 100% of the total cooling, and totally dominates

conditions in the doud. This occurs when cooiing shifts away from the Balmer

hes, which are becorning optically thick, so that the gas grows hotter (see

Figure. 5.5) and is able to exate Fe II more effectively. Behind the ionkation front

Fe II carries about 1/3 of the total cooling, and remains the most important

coolant speûes.

,r quasars

5.2.4 The resulting spectrum

The Fe II lines are strong in man] and we expect that in some

conditions Fe II can dominate the heating or cooling of the cioud. Fe II is an

efficient coolant because of its large number of lines, offering many charnels for

emission despite the strong trapping due to large optical depths. In situations

where collisional deexcitation of radiatively pumped exated levels is significant

log Depth,cm

Figure 5.4: Log of the ionization fractions of Fe0 -F& vs. depth within the doud.

Ionization structures of the heavy elements are often not sharp, as descri'bed by

Ferland & Persson (1989).

1 total cooling

log Depth.cm

O

C -2 .- - O

O F 4 - -

-6

-8

Figure 5.5: Log total and Fe II cooling vs. depth within the cloud.

Fe II cooling 1

: 1 /

11 ,f

\ - \

* . 6 8 10

heating of the gas can occur as well (see sedion 5.3.3). These effects are

fundamental since the caldations desdbed in this thesis have energy balance

as their foundation: the energy in the incident continuum is only being

repartitioned among the various lines and continua that are emitted. If Fe II

heating or coohg alters the thermal balance of the cloud, the rest of the

spectrum will also change to maintain energy conservation. Thus the details of

Fe II emiçsion will feed back h t u other lines emitted from the same region such

as the Balmer lines or Mg II 22798.

The resulting spectrum of this cloud (continuum and emission lines,

induding all h e s of Fe II) is shown in Figure 5.6.

5.2.5 Electron temperature behavior

As is typical of a X-ray illuminated atmosphere, a large column density of

partially ionized gas with a warm temperature exists beyond the hydrogen

ionization front. This is the combined result of penetrating X-rays and

photoionization of the n=2 level of hydrogen by the Balmer continuum. The

latter depends exponentially on the electron temperature since this affects the

population of the n=2 levei. There is also strong feedbadc between Fe II emission

and the hydrogen n=2 population because there is an extra Loss mechanism

through Lya pumping of iron. Figure 5.7 shows how the electron temperature

changes with depth into cloud.

-2 -1 O 1 log photon energy, Ry

Figure 5.6: The diffuse emission hom a doud with gas density log n(H) [an-31 =

12, ionization parameter log U(H) = -3 (log@ n 19.5) and column density log N(H)

[an-21 = 23 and illuminated by an ionizing spectrum typical of quasars. The inset

illustrates the cornparison of the diffuse emission in the near-W region with

(soüd) and without the Fe II mode1 induded in the calculation. Note the

significant bump in the spectral region corresponding to 2200-2700 A (-0.38 to

-0.47 in log hv, Ry), and the weaker Balmer continuum resuiting from its

destruction in the Fe Iï emitting zone.

log Oepth,cm

Figure 5.7: The log of the electron temperature (K) vs. depth within the doud.

The temperature changes rapidly on the shielded side of hydrogen ionization

front due to Lya trapping becoming important, with the associated increase in

photoionization from n=2 by the Balmer continuum.

5.3 Fe II emission modeling for the BELR of AGN

Here we will be concemed primarily with Fe II emission from the BELR of

AGN. Observationally, the Fe II spectrum is rich across a broad wavelength

rangeI from the W to IR, and various astronomical objeds show quite different

types of Fe II emission spectra. The goal is to detennine the physical conditions

in Fe II line-forming regions by modeling the Fe II emission. In this section we

discuss general features of the Fe II spectra as they depend on the basic

parameters of o u models (density, flux, microturbulent velocity, the Fe

abundance, and Lya pumping). Findy, we discuss the Fe II emission bands that

we use to make the problem more tractable and the "locally optimally emitting

doud" concept.

Here we show a few Fe II ernission models from more than 120

photoionization calculations that span 4 ordea of magnitude in hydrogen

ionking flux, 5 orders of magnitude in particle density, from O to 100 km si in

microturbulent velocity, and about 1 order of magnitude in Fe abundance. A

summary of model parameters, induding the model studied in section 5.2, is

given in Table 5.1.

5.3.1 Density and fîux dependence

Electron collisions are one of the main excitation mechanisms populating

the Fe II atom. The redting spectnim cames with it many dues to a doud's

electron density. For instance, some levels have no permitted decays to ground

and build up relatively high populations. Forbidden or intercombination lines

kom such levels will become strong ai relatively low densities. Transitions with

metastable states as lower levels will become opticaily thick and thermalize at

Table 5.1

Parameters of the BELR Fe11 emiasion models

vt, kms -l Abundance Cont. *

. relatively low densities too. ûther lines corne from levels that are depopdated by

strong transitions and remah strong to high densities.

First we investigate the density dependence of the spectntm. We consider

the UV and optical Fe II spectra produced at the high densities expected in

quasar broad emission line regions. For the purposes of plotting, the lines are

assumed to be broadened by a Gaussian maaoturbulence with FWHM = 103 km

si, chosen to represent buk motion of separate douds within the broad line

region. No microturbulence is assumed to exist within a doud; the lines were

only broadened by the Doppler width appropriate br the local electron

temperature. Figure 5.8a (model 5.8a) shows the UV and optical Fe II spectrum

calcuiated at n~ = 109 an-3, a flux of hydrogen-ionizing photons of log (OH) [ar2

si] = 17.5 and total hydrogen column density of 1024 cm-*. We w d the baseline

AGN continuum described in Korista et al. (1997), consisting of a UV bump

peaking near 44 eV, a f, a v" X-ray power law, and the UV to optical spectral

index ax = -1.4. The plotted specba are expressed as the "flux" f, where units of

&are in erg cm%'.

Two peaks around 2600 and XOOA are the strongest features in all the

models. They are mainly formed by UV1 and UV2, the first resonance multiplets

of Fe II, although more than 3000 Fe II h e s have wavelengths between these two

peaks. Hundreds of lines give significant contributions to the total Fe II

emission, filling the gap between the two strongest peaks. Spectral features

within this gap are produced by a pseudo-continuum from many lines of various

multiplets. Depending on model parameters, tens or even hundreds of weak

hes , both pennitted and forbidden ones, may produce a higher total intensity at

a given wavelength than the intensity of one strong permitted line. This makes it

meannigless to assoaate any blend with a partidar line in most cases, and

makes complete spectral synthesis mandatory.

Figure 5.8a: Theoretical UV and optical Fe II spectra calculated for the parameters expected in quasar

broad emission line regions. Clouds illuminated by a spectral energy distribution typical in AGN, total

hydrogen column density of 1024 cm. solar abundances, hydrogen density n~ = 109 cm-3, flux of

hydrogen-ionizing ph0 tons 10g(@~)[crn-W 147.5.

Figure 5.8b presents the Fe U spednim at a thousand times higher particle

density, nH=1012 -3, and similarly inaeased flwc log(mH) [cm-2 SI]= 20.5 (model

5.8b). This means that the ionization parameter is the same as in model 5.8a. The

spectra of models 5.8a and 5.8b are quite different in the optical window, 3600A

< L < 5600A. A detailed study of the contnbuting transitions shows that the

forbidden Lines at these wavelengths are the strongest ones in model 5.8a.

whereas the permitted optical lines are mu& stronger in the higher density

model 5.8b. An intermediate density case with the same ionization parameter

shown in Figure 5.9a shows the same trend. This agrees with the conclusions of

Netzer (1988), who found that the forbidden lines become much weaker than the

permitted opücal Lines at densities above log(nH) = 9.5. This provides an obvious

density probe.

We find that typical blends of Fe II emission have effective critical

densities (densities where collisional rates equal radiative rates) varying between

Iog(nH) of 9.5 to 10.5, depending on other parameters of calculation (mainly the

ionking flw and microturbulent velocity, since these affect the iine optical

depth). In the ultraviolet two broad blends of Fe II lines at 1000-1800A and a

larger one at ~OOO-WIOA demonstrate this density dependence. The shortward

feature becomes weaker relative to the Iongward one as density increases. The

ionization parameter also affects the spectnun. Mode1 5 .8~ (Figure 5 .8~) was

computed with the same flux as in model 5.8a, log(<PH) [& cl]= 17.5, and the

same density as in model 5.8b, n~ = 10'2 cm-3. The ionization parameter in mode1

5 . 8 ~ is therefore three orders of magnitude lower than in models 5.8a and 5.8b,

and the temperature of the gas cloud decreases rapidly with depth. The emission

cornes from oniy a relatively thin layer with Iowa optical depth, and therefore,

the strongest resonance lines are relativdy less saturated and rem& efficient

ernitters. For instance the strength of multiplets UV1 (2600 A) and UV2 (2400 A) is much larger than the emission that lies between these emission clusters. The

Figure 5,8b: The same as Figure 5.8a, but n~ =IO12 cm-3, log(QH) [cm-2 SI]= 20.5 The density has increased at

constant ionization parameter.

weaker transitions in this waveiength interval are only strong relative to UV1

and W2 when there is a signüicant contriiution from the continuum pumping.

5.33 Dependence on microturbulent velocity

Netzer & Wills (1983) suggested that the miaoturbulent veloaty could

influence the intensity of Fe II emission. This is because broader line widths

increase the efficiency of continuum pumping and decrease line optical depths.

Effects of microturbulent velocity on emission spectra of various objects have

been discussed recently by Alexander & Netzer (1997) and Murray & Chiang

(1997). The main effect is to add an extra term to the line width, v' = kT / rn + v,' , and this has the effect of desaturating the iine and ailowing it to absorb radiation

more effiaently.

We computed several BELR douds with a density of nH=lOIO cm-3, a

hydrogen-ionuing photon flux log(@H) [cm-2 sl]=18.5 (same ionization parameter

as models 5.9a and 5.9b but intemediate density) with various values of the

microturbulent velocity and the same column density as models 5.8. Mode1 5.9a

(Figure 59a) has thermal widths ody, but the two others have added

microturbulent velocities of v, =10 km s-1 (model 5.9b), and v, =IO0 km s-1 (model

5.9~). Note that the flux scale in al1 the three Figures is the sarne. As the

microturbulent velocity increases the local line width does too, and pumping by

both lines and the continuum becornes more effective. This seiectively increases

the intensities of hes arising from levels that can be pumped.

The relative intensities of UV and optical bumps of Fe II lines are also very

sensitive to the rniaohirbdent velocity v,. Not only the form of the net Fe II

e s i o n is changed, but also the full flux and profile of the Fe Ii emission. The

qualitative changes with increasing turbulent velocity are simiiar to those with

Encreashg density (ai constant ionization parameter). However, a high turbulent

velocity enhances the W spectmn of Fe II (1000-1800A and 2400-

2600A) far more than a high density does. This sensitivity to the asnimed line

profile is also a complication - the spectral synthesis r e d t s are dearly sensitive

to the exact treatment of the line broadening function and radiative transfer.

5.3.3 Lya pumping

There has been a long and active debate conceming how important Lya is

for exciting Fe II (Brown et al. 1979, Penston et al. 1983, Johansson & Jordan 1984,

Netzer 1988). Netzer made calcdations assuming a high density and optical

depth: a y a ) is about 108 for gas where Lya fluorescence is included. These

caldations showed that Lya exates a few Fe II h e s but that the energy in all of

them is very srnail.

The importance of Lya pumping is actually a function of the intrinsic

width of Lya, the excitation temperature or source function of Lya, and the

number of exated Fe+ levels energeticdy accessible. We find that this process

cari be very efficient for some conditions. The most important condition for

efficient Lya pumping is for the optical depth and resulting Lya width to be

large enough to cause overlap with many optically thidc Fe II lines (the greater

the number of levels that can contribute, the more effectively Lya pumping is).

ûtherwise, Fe II atoms cannot absorb the Lya photons effectively.

The full NLTE level populations of hydrogen are determined self-

consistently for the local conditions, as described in Ferguson & Ferland (1997)

and the papers they cite. The mode1 hydrogen atom cornputes level populations

for 2s and & independently, including now the effects of Lya loss ont0 Fe+. We

assume that the Lya source function is given by the Planck function at the 1s-@

exatation temperature. The optical der& detennines the line width as outlined

by Elitzur and Ferland

optically thick Lya lines

&

(1986). Using an approximation of the profiles of

(see section 23), we d e c t all the o p t i d y thidc Fe II

lines which have the wavelengths within the Lya profile, and detennine the

intensities of Lya emission at the corresponding energies. These intensities are

then used to calculate the coefficients of the induced transitions due to Lya

radiation. Lya pumping is significant in populating the corresponding upper

levels of the Fe II atom if these coefficients are significant compared to the

collisional exatation and continuum pumping coefficients. The subsequent

cascades of the spontaneous transitions change the populations of the lower

levels, and therefore the whole spectnim of the Fe II emission.

We computed a BELR with the Mathews & Ferland (1987) continuum,

d a r abundances, a column density of 1023 &, U = 10-2, a hydrogen density n~ =

1012 &, (lower column density version of mode1 5.8b with slightly different

continuum, see Table 5.1) and rniaoturbulent velocity of 100 km s-1, and did two

calculations, with Lya pumping switched on and off, to determine its effects. We

found that the spectmm between 1000 and VOOA, shown on Figure 5.10, is the

most sensitive to Lya pumping. It c m be seen that Lya pumping makes many

emission lines stronger, thus increasing the flux across the UV region. Lya

pumping does not affect some h e s at all.

In general Lya pumping is most effective for transitions whose lower level

is adD and upper level is u4P. heasing the microturbulent velouty allows

more optically thidc Fe II transitions to be pumped by the wider Lya In

partida., more upper states, inciuding u~P, x6F0, vsF0 (which are energeticdy

close to u4P) c m become accessible. Another consequence is that the gas tends

to be hotter when Lya pumping is induded. This is the renilt of the process

depositing electrons into very highly exated states that can then be collisionally

de-exated.

Figure 5.10: The solid line represents the Fe II spebnim predicted when the effects of Lya pumping are

turneci off. The dashed line shows the predicted spectrum with the inclusion of these effects. A

microturbulent velocity of 100 km s-1 was assumed in both cases.

5.3.4 Fe abundance and coamological scale

Interpretation of the Fe II emission from quasars has a major cosmological

motivation. In any mode1 of galactic nudeosynthesis, iron is predorninantly

formed as a "delayed primary" element (Hamann & Ferland 1993) as the result

of Type 1 supernova explosions. These occur when acaetion ont0 white dwarfs

in dose binary systems causes them to exceed the Chandrasekhar limit. As a

result of the explosion, radioactive nickel is fonned which then decays to Fe.

White dwah only form after a duster age of roughly 109 year, the üfeüme of the

shortest lived progenitor star. Hamann & Ferland (1993) showed that in a typical

QSO model the iron abundance should increase by more than an order of

magnitude relative to C when the age of the stellar cluster passes through 1 Gy.

The long-term hope is determining the abundance of iron. If accurate Fe

abundances are measured the expected -1 Gyr delay could therefore be used as a

dock to constrain QSO ages. Age constraints could in tum constrain the

cosmology when applied when the Universe is 0.5 to 3 Gyr old, hence to high-

redshift (redshift - 5) sources (see Figure 5.11). The age and line spectnim place

strong linnits on the chemical evolution of the host galaxy and the onset of galaxy

formation.

To investigate the influence of Fe abundance we did a series of

calculations for a quasar BELR with a column density NH =IO24 cm-2, a density

nwlO*l cm-3 and a hydrogen-ionizing photon flw< log(Qrr) [an-2 d]=185. This

corresponds to an ionization parameter of 10-3. Only the iron abundance was

changed, by scale factors from 1/3 (model 5.12a) to 5 times solar (model5.12b,

see Table 5.1).

Figure 5.11: Age-redshift dependence on cosmological parameter q. for

different values of the Hubble constant, Ho. From Hamann & Ferland (1993).

Figure 5.12 shows spectra where the dashed üne is for the smaller

abundance and solid line for the higher one (15 ümes larger). These show that

the lhe strength does not scale 1:l with Fe/H is and different dusters of lines

read differently to changes in the iron abundance. This is due to a competition

between two effeds as the abundance increases; the density of Fe iI atoms

increases, so îhat they have a greater effed but the lines tend to become more

sahirated and so less efficient emitters. The spednim does not depend linearly

on abundance and these dependenaes show that a complete synthesis of the

spednun will be necessary to deduce reliable uon abundances.

5.3.5 "Locally optimally emitting cloud" concept and Fe II etnisrion bands

In early attempts to reproduce the observed quasar BELR spectra by

photoionization calcuiations Davidson (1977), Davidson & Netzer (1979), Kwan

& Krolik (1981) and others used a mode1 with a single "doud" and thus a single

column density (log(NH)[ cm-21-23), gas density (log(n.)[ ana]-lO), and ionkation

parameter (U -0.03). Later spectroscopie observations (Gaskell 1982, Wilkes 19û4,

Espey et al. 1989, Corbin 1990, Clavel et ai. 1991, Peterson 1993) îndicated the

presence of a wide distribution of clouds with various parameters. These

observations prompted the more recent multi-cloud photoionization caldations

by Rees, Netzer, & Ferland (1989), Krolik et al. (1991), Goad, O'Brien, &

Gondhalekar (1993), and O'Brien, Goad, & Gondhalekar (1994,1995).

Recently, Baldwin et al. (1995) found that integrating the emission from

clouds with a very wide distriiution in properties r ed t s in a spectnim which is

consistent with the observed composite quasar spectra. They introduced a

concept of locaUy optimally emitting douds" (LOCs) whidi states that for each

emission line there is only a narrow range of density and distance continuum

source resulting in a maximum reprocessing effïciency. The scenario based on

the LOC concept offers effective solutions to the observational problems posed

by B U

As we discussed previously, a very large number of Fe II iines that

originate at high densities cannot be resolved individually. Rather, observational

data are to be compared with theoretical caldations of the Fe II emission

integrated over spectral bands (blends of iines). Study of the Fe II emission in

different spectral intervals can be especiaily valuable for theoretical

interpretation of quasar emission spectra.

Theoretical calculatiow for the LCX concept require the analysis of a large

number of "single-doud" calculations simultaneously. A "single-doud" model

with Fe II induded into the photoionization model took about 5 hours to

calculate with about 98% CPU on one processor of an HP Exemplar

supercornputer. Here we present the resdts of 99 photoionization calculations

that Vary in hydrogen ionizing flux between log (QH) of 17 to 21 and in partide

density between log(ns) [an.'] of 9 to 14. We assurned solar chemical abundance,

hydrogen column dewity 10U cm-2, and constant hydrogen density throughout

each doud. The incident continuum dope was chosen according to Korista et al.

(1997). Our calculations confirmed that for the high-density objects the Fe II

emission cornes mainly fkom bands, not from individuai lines.

The total flues of two prominent Fe II emission bands relative to the

inadent conhuum fiux at 1215 A are shown in Figures 5.13 and 5.14 as contours

in the n~ - aH plane. The first Fe II band, marked "2430N, accumulates emission

Lines from 22ûû to 2660 A and uidudes the resonant multiplets UV1, UV2, UV3,

UV4 and UV5. In the left part of the plane, this relative flux provides a measure

of the ionization parameter, LI. The relative flux in this band increases as U

decreases (see discussion in Section 5.3.1).

The second Fe II band, "5270f', ranges from 5080 to 5460 A and covers as

permitted optical multiplets (48,49) and strong forbidden multiplets (18F, 19F).

The maximum intensity of the 2430 bump occurs at high density (log ( n ~ ) [a+]

> 14.0) and flux log(@^) [cm-2 SI] - 19). W e is no simple relationship to the

ionization parameter. The 5270 bump intensity distribution has two maxima.

One of them is due to contribution of the optical permitted lines. These lines are

produced by transitions from the levels which are dose to the upper levels of the

resonant ünes responsible for the 2430 burnp, and the position of its maximum at

log ( n ~ ) [cm*3]- 13.5 and log(@^) [ a n - 2 SI] - 19 is close to the position of the 2430

maximum intensities. However, the 5270 bump has one more maximum at log

( n ~ ) [cm41 c 9 and log(@^) [cm-2 s-1) < 17 which is partially due to the

contribution of the forbidden multiplets. In fa&, the position of this maximum

occurs where conditions are favorable for both the permitted and forbidden

transitions. It is obvious that powerfd selection effects work here. The permitted

W multiplets and the forbidden collisionally excited h e s form most effiaently

at quite different parameters of clouds.

Figure 5.15 shows the Fe II emission in another spectral band "2780", from

2660 to 2900 A. This band indudes several strong Fe II multiplets (UV62, UV63).

It behaves like the "2430" band. It also covers the region of the broad Mg II

doublet at 2800 A whidi is prominent in quasars. Figure 5.16 presents the sarne

grid of calculations for the Mg II emission, and Figure 5.17 shows the relative

contniution of the Fe II emission compared to the Mg II line. In the main part of

the n~ - aH plane, the Fe II is responsble for less than 30% of the total emission

line i n t d t y in this spectral range. However, at high flues of ionking photons,

the Fe II emission is stronger than the Mg II emission, and at high density still

significant .

9 10 11 12 13 14

log hydrogen density

Figure 5.13: The flux of the total Fe II emission in the spectral band "2430", from

2200 to 2660 A, shown as a function of the hydrogen density nH and flux of

hydrogen-ionking photons a. The chexnical abundances are solar, the doud

column density is 1023 cm-2, and the continuum spectral energy distribution has

an UV bump that peaks near 44 eV. The flwes are given relative to the incident

continuum 18, at 1215 A for full source coverage.

9 10 11 12 13 14

log hydtogen density

Figure 5.14: The same as Figure 5.13 for the Fe II "5270" emission band, from

5080 to 5460 A.

10 I l 12 13 14

log hydrogen density

Figure 5.15: The same as Figure 5.13 for the Fe II "2780" emission band from 2660

to 2900 A. Note the similar behavior compared to the "2430" band.

10 11 12 13 14

log hydrogen density

Figure 5.16: The same as Figure 5.13 but the broad Mg II doublet at 2800 A (see

Figure 5.6).

r i t

log hydrogen density

Figure 5.17: Relative contribution of the Fe II "2780" emission compared to the

Mg II doublet for the same grid of cddations. In the upper right the ratio dimbs

significantly where Mg II emission becomes weak while Fe II remains strong

(bo th are weak in the upper Mt).

Our general condusions are the folIowing:

Study of the Fe II emission in optical and W spectral bands is a

powerful method for diagnostics of the physical conditions in BELR of

quasars.

Different bumps of the Fe II emission lines are the most intensive

quite different physical conditions. Powerful selection effects are

work in quasar spectra.

The Fe II emission might contribute significantly to strong broad

emission lines in quasar spectra, like the Mg II 2800 doublet. This

contribution must be taken into account in the general models of the

strong quasar emission hes.

Chapter 6

Summary

Our primary research interests lie in the theoretical modeling of ernission

spectra and interpretation of spectroscopic observations. In the thesis we placed

speaal emphasis on the Fe II emission. We have developed a numerical Fe II line

emission code, which has been incorporated into the radiative-collisional code

Cloudy. The Fe II code has been applied to the theoretical interpretation of

spectroscopic observations based on data obtained from HST and the Cerro

Tololo Inter-American Observatory (in coilaboration with Jack Baldwin). Future

research plans build upon this approach.

6.1 Fe II emission line physics

We developed a large Fe+ modei ion that indudes the lowest 371 energy

levels (up to 11.6 eV) and predicts the emission of 68635 Fe II lines. The mode1

atom estirnates the heating, cooüng and emission characteristics of Fe II in a

computationally expedieni way, and has been fully incorporated into Cloudy.

A large number of dowed and forbidden Fe II lines are present in a broad

range of emission spectra in a variety of asfronomical sources. The Fe II emission

spectrum arises from thin "envelopes" around a variety of objects such as stars,

H II regions, novae, supernova remnantsI active gaiaxies and cpasars. The high

abundance of iron in the Universe, and the very rich spectrum of Fe IïI explain

why this is so. The imn abundance is 3-1W that of hydrogeh ranking only after

H, He, C, N, O, and Ne in a solar composition, and is comparable in abundance

to S, Mg, and Si. Fe II has a very line-rich spectnim because the valence shell is a

half-nlled 3d-shell. Its gound state configuration is ls22~22p63sz3p63d64s 6D912.

More than lûûû energy levels of Fe II are currently known. We plan to extend the

number of Fe II levels and to incorporate more accurate sets of collision and

radiative data into our code as computers grow faster and atornic databases

expand.

6.2 [Fe II] emission from the Orion Nebula

[Fe II] lines provide powemil diagnostics for physical conditions in low-

density plasmas. These are applied to a variety of different objects with the aim

to measure the Fe abundance, veloaty field, density, and excitation conditions.

H II regions are good laboratones to check how Fe II emission works. We made

an initiai application of the Fe II emission model to the Orion H II region due to

its relative sirnplicity. The detailed cornparison of model predictions with new

observations with the Cassegrain echelle spectrograph on the Blanco 4m gave a

unique possibility to identiEy more than 40 [Fe II] lines, almost double the

previously known number of [Fe II] lines from the Orion Nebula. The sensitivity

of the populations of ali 371 levels to radiative pumping, and changes in the Fe II

emission lines as the result of density and radiative pumping conditions, were

investigated.

The theoretical models (pure collisional mode1 and pumping model) were

compared with the observational data. These cornparisons dearly show that

radiative pumping is required. A pure collisional model, which had been

proposed previously and would have recpired densities 100 times higher Uian

indepependently know to exist in Orion, is not viable. The models show that Fe

is significantly depleted in the gas phase in the Orion H II region (probably in the

form of dust).

Generally, the prediaed intensities are w i t h SOYO of the observed ones.

The overall agreement is consistent with the combined uncertainties of the

observations, the dereddening corrections, and the quality of the underlying

atomic data. Furthemore, here are no strong iines Fe II iines predicted in our

best model (model C, Chapter 3) that have not been deteded in our observations.

Recently, Rodnguez (1999) showed that the intensities of [Fe Dl] lines in

different H II regions are strongly correlated with the intensity of the diffuse

continuum. Using an updated version of Cloudy, we are intending to investigate

the pumping effects of the diffuse continuum quantitatively.

6.3 Analysis of other emission lines in H II regions

We are also participatirtg in studies of emission spectra of other ions in H

II regions. A primary motivation is to investigate physical conditions, velocity

fields and abundances of elements and to develop a theoretical model consistent

with the whole suite of observations.

New detailed spectroscopie obsewations of the Onon Nebula (in

coiIaboration with Jack Baldwin) have made it possible for us to identify 411

emission Iines in the 3498-7468A spectral range. This is the most extensive line

atlas of the Orion Nebula ever made in this spectral range.

We present a detailed analysis of the velouty field in the Orion Nebula,

and show that the forbidden lhes split to two distinct dusters, depending on

ionization potential. This is consistent with a dynamïc model in which lines of

ions with different ionization potentiais originate in gas flowing away from the

ionization front toward the ionipng stars (and the observer too m this case). The

Iines from the Speoes that are most ionized originate in the most rapidly moving

(relative to the ionization fiont) gas dosest to the stars.

The [Fe m] spedrum is unique in displaying a correlation between

veloaty and exatation potentid for individual lines. The obsewed IlIJ lines

lie between 4000 to 5600 A and show a systematic deaease in their velocities

with decreasing wavelengths; this means the higher the excitation potential the

lower the velocity (lower veloaty means gas moving more rapidly relative to the

ionization front). Our interpretation is that the ionization structure of Fe is such

that these iines Corm aaoss the region where the gas receives the greatest

acceleration as it moves out of the PDR and ionization front into the H II region.

This is the first direct measurement of a pressure-induced acceleration w i h h the

ionized gas.

The observation of an excitation po tential - velocity correla tion requires

that there be a correlation between the temperature and veloaty gradients,

something not predicted in stationary photoionization models üke Cloudy. The

possïble importance of advection in the heating-cooling balance will have to be

investigated. It will also be interesting to calculate self-consistently the

acceleration which is needed to produce the obsewed diange in velocity in this

particular ionization zone. This wiil be an important test of hydrodynamical

models of realistic H II regions.

Our future efforts will be directed to work on the permitted C, N, and O

emission lines from the Orion Nebula. These are important lines that have in the

past been used to suggest the presence of temperature fluctuations in the gas. It

appears from our preliminary assessrnent that in addition to fomtation of these

lines by simple recombination cascades, the eff- of radiative pumping (and

photoionization to excited states in some cases) might also be important, which

wouid dramatically change the physical interpretation of these lines.

6.4 Fe II line emission from Active Galactic Nuclei

Fe II emission is very cornplex, with a vast number of physical processes

at work. We plan to work with archival HST observations of the old nova RR

Tel, q Carinae, and the Seyfert galaxy I Zw 1 to verify the mode1 calculations.

These are sharp h e d objects with strong Fe II emission, and the steiiar sources

have the advantage that the geometry, ionization source, and the abundances are

largely understood. These offer an astrophysical laboratory to test and validate Y

the completeness of our numerical simulations.

in the thesis we gave an o v e ~ e w of general features of the Fe II spectra of

typical quasar Broad Emission Line Regions and discussed the dependence of

these features on the basic parameters (density, flux of radiation, microturbulent

velocity, the Fe abundance, and Lya pumping). We applied the concept of

"locally optimally ernitting douds" (LOC) to a few prominent Fe II emission

bands (unresolved blends of rnany overlapping lines) as an approach to

investigating Fe II emission in quasars.

The eventual goal of our work is to use Fe II emission of high redshift

quasars to determine physical parameters of the emission line regions and their

iron abundance- We have identified severai broad blends that are sensitive to

density. We will search for others which are affected by temperature, turbulence

md Lya pumping. By combining HST W data with ground-based optical data

we will try to identify the physics governing the overall appearance of the strong

Fe II emission. The synthesis approach to be taken is descnid in this thesis. W e

produce very large grids of \caldations fonning a data cube in doud density - doud column density - flux of ionizing photons space, and then integraie over

this ensemble using distribution functions (LOC being an extreme). Object to

object spedral ciifferences in the Fe II lines are due to changes in the distribution

functions, doud composition, and shape of the ioniPng radiation field. Since Fe

CI is an important coolant, when Fe II changes in response to changes in one of

these parameters, other lines change too, red t ing in quantitative predidions

that we can check. It will be interesting to look for changes in the Fe II spectrum

in AGNs that have intrimic vasiability.

The HST d v e s are needed to obtain specha of the UV Fe II multiplets

for objects where the optical multiplets also have been or can be measured. It is

necessary to cover a wide range in the W to optical Fe II intensity ratio in order

to test predictions of the effects of dianges in turbulent line broadening and other

doud parameters. We plan to start by analyzing data for low-redshift quasars

and then move to high-redshift quasars.

The aim is to determine the abundance of iron at high redshifts. The

evoiutionary models of the central regiow of galaxies, in which the quasars form,

predict an inaease in Fe/O abundances after -1-2 Gyr by factors of -2 to 10

relative to solar (Hamann & Ferland, 1993). This overabundance might help

explain the strong Fe II emission observed in many Q 5 0 s and AGN. The

expected -1 Gyr delay from the onset of star formation could therefore be used

as a dock to constrain QÇO ages if accurate Fe abundances are measured. Age

conshaints could in turn constrain the cosmology when applied to high-redshift

sources.

Appendix

How to use the Fe II model in Cloudy For an overview of how to use Cloudy, see the "Hazy" documentation (Ferland

1996).

Commands added to Cloudy input are:

atom-ii vemer comand initiates the large 371 level Fe II atom. This optional

command determines which model of the Fe II atom is used. The simple

option is atom eii netzer which adivates a different smaii atom with only the

16 lowest levels model. The netzer option is very fast. The verner option is far

more accurate, but also much larger and slower.

atom feN verner levels N sets number of levels. N must be at least 16, but not

greater than 371.

iterate N This command specifies the number of iterations to be performed

(N>l). The command is required in the input set because the Cloudy code

has no interaction with the Fe Il atom within the first iteration.

punch vemer Ilastl "file.trtN lfaint l f enerml enermî l This command produces

the output of Fe II line intensities after each iteration. Optional parameter

[lastl produces the output after the last iteration only. Parameter "fi1e.m" is

the name of the output file. It is mandatory and must be surrounded by

double quotes. The remaining parameters are optional and can be omitted.

faint is the faintest line to be punched, relative to the nomaikation line,

expressed logarithmically if negative. Wamhg. if this parameter is not set, all

h e s will be punched. . The size of total output He is 2 Mb. enernvl and

e n e r e show the range of energies, in Rydberg, over which lines are to be

included in the output. If this parameter is not set, all h e s will be punched

The output file includes Fe II level numbers, Fe II line intensities and optical

depths of Fe II hes . The output includes the flux integrated up hom enerm 1

It is possible to work with Fe II bands to minimize output when surveying

oves a large grid in parameter space. In this case a main program with

subroutine, cailed "cùline", needs to be written to drive Cloudy. The label

and wavelength of the line should be specified, and the subroutine retums

the relative intensity.

punch feii de~artzwe coefficients "de~arture.txt" lalll punches the departue

coefficients for selected levels, optionai keyword & does all the levels.

An example of a Cloudy input file that indudes the Fe II atom:

titlm Fa II ta#+ for tha Orion conditions Mas 3,OO black body 35000. iodxation par -3 abuaâ orion conmtaat tampmraturm 4 itaratm 2 rtop +oiu 1 atoam f m i i varno+ atom f a i i vaznmr lavals 3 7 1 punch vamer mfila.trrtm 0.01 0 , 0 0 0.05 +********************************************************

References .

Adams, T. F., 1972, ApJ, 174,439

Adams, T. F., & Weedman, D. W., 1972, ApJ, 173, LI09

Alexander, T., & Netzer, H., 1997, MNRAS, 284,967

Baldwin, J. A., 1975, ApJ, 196, L91

Baldwin, J. A., Crotts, A., Dufour, R. J., Ferland, G. J., Heathcote, S., Hester, J. J.,

Korista, K. T., Martin, P. G., O'DeU, C. R., Rubin, R. H., Tielens, A. G. G. M.,

Verner, D. A., Vemer, E. M., Walter, D. K. & Wen, Z., 1996, ApJ, 468, LI15

(B96)

Baldwin, J. A., Ferland, G. J., Martin, P. G., Corbin, M. R., Cota, S. A., Peterson,

B. M., & Slettebak, A., 1991, ApJ, 374,580 @FM)

Baldwin, J., Ferland, G., Korista, K., & Vemer, D., 1995, ApJ, 455, LI19

Balidc, B., Gammon, R. H., 6r Hjellming, R. M., 1974, PASP, 86,616

Bautista, M. A., Peng, J., & Pradhan, A. K., 1996, ApJ, 460,372 (BPI?)

Bautista, M. A., & Pradhan, A. K., 1995, ApJ, 442, L65 (BP95)

Bautista, M. A., & Pradhan, A. K., 1996, A&AS, 115,551 (BP96)

Bautista, M. A., Pradhan, A. K., & Osterbrock, D. E., 1994, ApJ, 432, LI35 (BPO)

Bergeron, J., & Kunth, D., 1980, A U , 85, LI1

Boksenberg, A., & Netzer, H., 1977, ApJ, 212,37

Boroson, T. A., & Green, F. F., 1992, ApJS, 80,109

Brown, A., Jordan, C., dr Wilson, R., 1979, in The First Year of IUE, ed. A. Wills,

University College London, p. 232

Carpenter, K. G., Wimg, R F., & Stencel, R. E., 1985, ApJÇ, 57,405

ciavel, J., et al., 1991, ApJ, 366.64

Corbin, M. R, 1990, ApJ, 357,346

Davidson, K., 19R, ApJ, 218,20

Davidson, K., & Netzer, H., 1979, Rep. Prog. Phys., 51,715

DeVeny, J. B., & Lynds, C. R 1969, PASP, 81,535

Elitzur, M., 1984, Ap J, 280,653

Elitzur, M., 2992, Astronomical Masers, Kluwer, Dordrecht

Elitzur, M., & Ferland, G. J., 1986,ApJ, 305,35

Elitzur, M., L Netzer, H., 1985,ApJ, 291,464

Enckson, E. F., Rubin, R. H., Haas, M. R, Simpson, J. P., & Colgan, S. W. J., 1989,

BAAS, 21,1156

Espey, B. R., Carswell, R. F., Bailey, J. A., Smith, M. G., & Ward, M. J., 1989, ApJ,

342,666

Esteban C., Peimbert, M., Torres-Peimbert, S., & Escalante,V., 1998, MNRAS,

295,401

Esteban C., Peimbert, M., 6r Torres-Peimbert, S., 1999, A U , 342, L37

Fairall, A. P., 1968, PASP, 80, W

Ferguson, J. W., & Ferland, G. J., 1997, ApJ, 479,363

Ferland, G. J., 1992, ApJ, 389, L63

Ferland, G. J-, 1996, Hazy, a Brief Introduction to Cloudy, University of

Kentucky Physics Department Interna1 Report (available on the Intemet at

http://www.pa.uky.edu/-gqlcloudy )

Ferland, G. J., Korista, K. T., Vemer, D. A., Ferguson, J. W., Kingdon, J. B., dE

Vemer, E. M., 1998, PASP, 110,761

Ferland, G.J. 6r Persson, S. E., 1989, ApJ, 347,656

Ferland, G. J., Peterson, B. M., Home, K., Welsh, W. F., & Nahar, S. N., 1992, ApJ,

387,95

Fuhr, J. R, Martin, G. A., 6r Wiese, W. L., 1988, J. Phys. Chem. Ref. Data, 17,

SuppL 4

Gaskell, C. M., 1982, ApJ, 263,79

Gindhar, S., & Arellano Ferro, A., 1995, Revista Mex. Astrob., 31,23

Goad, M. R., OBrien, P. T., & Gondhalekar, P. M., 1993, MNRAS, 263,149

Grandi, S. A., 1976, ApJ, 206,658

Grandi, S. A., 1981, ApJ, 251,451

Greenstein, J. L., & Schmidt, M., 1964, ApJ. 140,l

Hamann, F., & Ferland, G., 1993, ApJ, 41% 11

Hamuy, M., Suntzeff, N. B., Heathcote, S. R., Walker, A. R, Gigow, P. &

Phillips, M. M., 1994, PASP, 106,566

Hauschildt, P. H., Baron, E., Starrfield, S., & Allard, F., 1996, ApJ, 462,386

Hill, G. J., Thompson, K. L., & Elston, R., 1993, ApJ, 414, L1

Hoeflich, P., Wheeler, J. C., & Thielemann, F. K., 1998, Ap J, 495,617

Hummer, D. G., & Storey, P., 1987, MNRAS, 224,801

Johansson, S., 1978, Phys. Scripta, 18,217

Johansson, S., 1999, private communication

Johansson, S., &Jordan, C., 19û4, MNRAS, 210,239

Kaler J. B., Mer LX, & Bowen 1. S., 1965, ApJ, 141,912

Korista, L, Baldwin, J., Ferland, G., 6r Verner, D., 1997, ApJS, 108,401

Kozma, C., & Fransson, C., 1998a, ApJ, 496,946

Kozma, C., & Fransson, C., 1998b, ApJ, 497,431

Kurucz, R. L., 1995, CD ROM 23

Kwan, J., h Krolik, J. H., 1981, ApJ, 250,478

Lanz, T., Hubeny, I., & Heap, S. R., 1997, ApJ, 485,843

Li, H., & McCay R., 1996, ApJ 456,370

Lipari, S., Terlevich, R., & Macchetto, F., 1993, ApJ, 357,338

Lowe, R. P., Moorehead, J. M., & Wehlau, W. H., 1979, ApJ, 228,191

Lucy, L., 1995, A&A, 294,555

Kalkofen, W., 19û4, Methods in Radiative Transfer, (Cambridge: CUP)

Krolik, J. H., Home, K., Kahan, T. R., Malkan, M. A., Edelson, R. A., & Kriss, G.

A., 1991, ApJ, 371,541

Mathews, W.C., & Ferland, G.J., 1987, ApJ, 323,456

Mewe, R., 1972, A U S , 20,215

Mihalas, D., 1978, Stellar Atmospheres, 2nd Edition, W. H. Freeman: San

Francisco

Moore C.E., 1972, A Multiplet Table of Astrophysical Interest, NSRDSNBS,

Washington: US Department of Commerce, Rev. edition

Murray, N., & CXang, J., 1997, ApJ, 474,91

Nahar, S. N., 1995, A U , 293,967

Netzer, H., 1988, in Physics of Formation of Fe11 Lines Outside LTE, eds. R.

Viotti, A. Vittone, & M. Frîedjung, p. 247

Netzer, H , 1990, in Active Galactic Nudei, Saas-Fee Advanced Course 20, eds.

Counrorsier, T. J.-L, k Mayor, M., Springer-Verlag; Berlui

Neber, H., & Wik, 8. J., 1983, ApJ, 275,445

NIST Atomic Spectra Database, 1999, NIST Standard Reference Database #78,

http:/ /physics.nist.gov /cgi-bin/AtData/maui-asd

O'Brien, P. T., Goad, M. R., & Gondhalekar, P. M., 1994, MNRAS, 268, û45

Oke, J. B., & Shields, G. A., 1976, ApJ, 207,713

Osterbrodc, D. E., 1977, ApJ, 215,733

Osterbrodc, D. E., 1989, Astrophysics of Gaseous Nebulae and Active Galactic

Nuclei, University Science Books: Mü1 Vailey, California

Osterbrodc, D. E., Tran, H. D., & Veillew, S., 1992, ApJ. 389,305 @TV)

Peimbert, M., & Torres-Peimbert, S., 1977, MNRAS, 179,217

Penston, M. V., et al., 1983, MNRAS, 202,833

Penston, M. V., 1989, MNRAS, 229,lP

Peterson, B. M., 1993, PASP, 105,247

Peterson, B. M., 1997, An Introduction to Active Galactic Nudei, Cambridge

University Press.

Phiilips, M. M .,1977, ApJ, 215,746

Phillips, M. M., 1978a, ApJS, 38,187

Phillips, M. M., 1978b, Ap J, 226,736

Quinet, P., Le Doumeuf, M. & Zeippen, C. J., 1996, A M , 120,361

Rees, M. J., Netzer, H., & Ferland, G. J , 1989, ApJ, 347,640

Rodnguez, M., 1999, A U , 348,222

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

Rubin, R. H., hifour, R J., Ferhd, G. J., bfartin, P.G., OI>ell, C. R, Bddwin,

J,A, Hester, J.J., Walter, D. K, W e . Z., 1997, ApJ, 474L, 131R

Rubin, R. H., Erihon, E. F., Haas, M. R, Colgan, S. W. J., Simpson, J. P., &

Dufour, R. J., 1992, in The Astrochemistry of Cosmic Phenornena-IAU

Symposium No. 150, ed. P. D. Singh, Kluwer, p. 281

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

Russell, H. N., 1929, ApJ, 70,11

Sargent, W. L. W., 1968, ApJ, 152, L31

Seaton, M. J., 1962, in Atomic and Molecular processes, ed. D. R. Bates, Academic

Press hc., New York, p. 414

Tielens, A. G. G. M., & Hollenbach, D., 1985, ApJ, 291,722

van Hoof, P., 1999, Atomic Line List v2.02,

http:/ /www.pa.uky.edu/-peter/atomic/

Van Regemorter, H., 1962, Ap J, 136,906

Vemer, D. A., 1999, Atomic Data for Astrophysics,

http://www.pa.uky.edu/-vemer/atomM

Vemer, D. A., Vemer, EM., & Ferland G. J., 1996, Atomic Data and Nudear Data

Tables, 64,l

Vemer, E. M., Verner, D. A., Korista, K. T., Ferguson, J. W., Hamann, F., &

Ferland, G. J., 1999, ApJS, 120,101

Wampier, E. J., dr Oke, J. B., 1967, ApJ, 148,695

Wiese, W. L., Fuhr, J. R., & Deters, T. M., 1996, J. Phys. Chem. Ref. Data,

Monogaph No.7

Wiese, W. L., Smith, M. W., dE Glennon, B. M., 1966, Atomic Transition

Probabilities, Vol. 1, National Bureau of Standards

Wiese, W. L., Smith, M. W., & Mes, B. M., 1969, Atomic Transition Probabilities,

Vol. 2, National Bureau of Standards

Wilkes, B. J., 1984, MNRAS, 207,73

WiUs, B. J., Netzer, H., & Wih, D., 1985, ApJ, 288,94

Young, P. R., Landi, E., & Thomas, R. J., 1998, A U , 329,291

Zhang, H. L., & Pradhan, A. K., 1995, A U , 293,953

Zuckeman, B., 1973, ApJ, 183,863