INFORMATION TO USERS

Tbis manusa^t has been reproduced from the microfilm master. UMl films the text direct^ from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter face, while others may be from aiy type of conqmter printer.

The quality of this reproduction Is dqiendent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrou^ substandard margins, and inqtroper alignment can adverse^ affect reproductioiL

In the unlikely event that the author did not send UMI a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyri^t material had to be ranoved, a note wiH indicate the deletioiL

Oversize materials (e.g., maps, drawings, charts) are reproduced by sectioning the original, beginning at the upper left-hand comer and continuing from left to right in equal sections with small overlaps. Each original is also pbotogr^hed in one e:q)osure and is included in reduced form at the back of the book.

Photographs included in the original m anuscr^ have been rq>roduced xerographically in this copy. Higher quali^ 6” x 9" black and white photographic prints are available for any phot(%raphs or illustrations appearing in this copy for an additional charge. Contact UMl direct^ to order.

UMI A Bell & Howell Information Company 300 NorthZeeb Road. Ann Arbor. Ml 48106-1346 USA 313/761-4700 800.'521-0600

I m a g in g S pectrophotometry o f P l a n e t a r y N e b u l a e

DISSERTATION

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University

By

Nancy Joanne Lame, M.S.

* * * *

The Ohio State University

1995

Dissertation Committee; Approved by

Prof. Richard W. Pogge | ^

Prof. Bradley M. Peterson Advisor Prof. Donald Terndrup Department of Astronomy UMI Number: 9534014

Copyright 1995 by Lame, Nancy Joanne All rights reserved.

OHI Microform 9534014 Copyright 1995, by UMI Company. All rights reserved.

This microform edition is protected against unauthorized copying under Title 17, United States Code.

UMI 300 North Zeeb Road Ann Arbor, MI 48103 © Copyright by

Nancy Joanne Lame

1995 To my father, who inspired me to start, and to John, who gave me the strength to finish

11 A cknowledgements

I am indebted to ray advisor, Rick Pogge for many things, but especially for his excellent teaching and for believing in rae even when I didn’t. Significant support has also come from ray first advisor. Brad Peterson, who could always be counted on for good advice. I’d like to thank those at Lowell Observatory who helped with this dissertation: Ray Bertram for all his help in observing and for being a such a nice guy, and Mark Wagner for always being there in an emergency. I am grateful to Bruce

Balick and Julie Lutz for arranging the loan of a Hell filter from Manastash Ridge

Observatory. I’d also like to thank Bruce Balick for useful discussions and advice, and for permission to reproduce Figure 15 of Balick (1989).

The support and encouragement of my family and friends sustained me throughout this dissertation. I owe thanks to Anita Krishnamurthi and Mark Houdashelt for being my “bondmates”, and to all the other graduate students for their friendship.

I appreciate the help from all who read drafts (especially Anita and Alice Quillen), listened to practice talks, and discussed data reduction/analysis with me. Others who made a difference are: Don Terndrup, from whom I enjoyed great talks and lots of smiles, and Karen Kepler, who supplied lots of chocolate. A special tribute belongs to the Columbus Folk Dancers who have been like family to me. Finally, the path that led to the completion of this dissertation wouldn’t have been started without

iii the encouragement of my parents, who told me I could do anything. And, of course, a huge thanks is owed my husband and partner, John.

IV V ita

May 9, 1966 ...... Born - Buffalo, NY

1988 ...... B.S. Physics, Washington State Uni­ versity, Pullman, Washington 1988 ...... Ohio State University Fellowship

1991 ...... M.S. Astronomy, Ohio State Univer­ sity, Columbus, Ohio 1991-1993 ...... Graduate Teaching Associate, Ohio State University 1993-present ...... Graduate Research Associate, Ohio State University.

Publications

Research Publications

Papers Published in Refereed Journals “NGC 6302: Ionized by a Very Hot Star or by a Wind?,” Lame, N. J. and Ferland, G. J. 1991, ApJ, 367, 208.

“Steps Toward Determination of The Size and Structure of the Broad-Line Region in Active Galactic Nuclei. II. An Intensive Study of NGC 5548 at Optical Wavelengths,” B. M. Peterson, et al. (64 authors, including N. J. Lame), 1991, ApJ, 368, 119.

“Steps Toward Determination of The Size and Structure of the Broad-Line Region in Active Galactic Nuclei. III. Further Observations of NGC 5548 at Optical Wave­ lengths,” B. M. Peterson, et al. (35 authors, including N. J. Lame), 1991, ApJ, 392, 470. “The Structure of the Broad-Line Region in the Seyfert Galaxy Mrk 590,” Peterson, B.M., Ali, B., Home, K., Bertram, R., Lame, N.J., Pogge, R.W., and Wagner, R.M. 1993, ApJ, 402, 469.

“Imaging Spectrophotometry of the Planetary Nebula NGC 6720 (the Ring Nebula),” Lame, N. J. and Pogge, R. W. 1994, AJ, 108, 1860.

Contributions to Published Conference Proceedings “CCD Images of Three Planetary Nebulae with Binary Nuclei,” Lutz, J. H. and Lame, N. J., in Planetary Nebulae, lAU Symposium 131, S. Torres-Piembert, ed. (Kluwer, Dordrecht), p. 462, 1989, (contributed poster).

“CCD Images of Southern Hemisphere Planetary Nebulae,” Lutz, J. H., Lame, N. J. and Balick, B., in Planetary Nebulae, lAU Symposium 131, S. Torres-Piembert, ed. (Kluwer, Dordrecht), p. 173, 1989, (contributed poster).

“Two-dimensional Spectrophotometry of the Ring Nebula,” Lame, N. J. and Pogge, R. W., in Planetary Nebulae, lAU Symposium 155, Weinberger, R. and Acker, A., eds. (Kluwer, Dordrecht), 1993, p. 194 (contributed poster).

“Hybrid 3-D Spectrophotometry of Emission-line Nebulae”, Pogge, R.W., & Lame, N.J. 1994, in Tridimensional Optical Spectroscopic Methods in Astrophysics, lAU Colloquium 149, ed. G. Comte, ASP Conf Ser, in press.

Fields of Study-

Major Field: Astronomy

Studies in: Planetary Nebulae Prof. R. Pogge, Prof. G. Ferland Active Galactic Nuclei Prof. B. Peterson

VI T a b l e o f C o n t e n t s

DEDICATION ...... ii

ACKNOWLEDGEMENTS ...... iii

VITA ...... V

LIST OF TABLES...... x

LIST OF FIGURES ...... xi

CHAPTER PAGE

I Introduction ...... 1

1.1 Planetary Nebulae ...... 1 1.1.1 Evolution of P N s ...... 2 1.1.2 Fine Structure in P N s ...... 4 1.2 Diagnostic Emission L ines ...... 6 1.3 Two-dimensional Spectrophotometry ...... 7 1.4 Focus of this project ...... 8

II Observations and Data Reduction ...... 13

2.1 Sample Selection ...... 14 2.2 Imaging Fabry-Perot Spectroscopy ...... 15 2.2.1 O bservations ...... 15 2.2.2 Basic Reductions ...... 17 2.2.3 Analysis ...... 18 2.3 Direct Imaging ...... 21 2.3.1 O bservations ...... 21 2.3.2 Reduction ...... 22 2.4 Longslit Spectroscopy ...... 23

vii 2.4.1 O bservations ...... 23 2.4.2 Reduction ...... 24 2.4.3 Analysis ...... 24 2.5 Putting it all together ...... 25 2.5.1 Flux Calibration of the Emission-line Images ...... 25 2.5.2 Ratio M ap s ...... 25

III NGC 6720 - The Ring Nebula ...... 37

3.1 Introduction ...... 38 3.2 Observations and Data Reduction ...... 42 3.2.1 O bservations ...... 42 3.2.2 Data Cube A nalysis ...... 44 3.2.3 Emission-line Image A n aly sis ...... 46 3.2.4 Flux C alibration ...... 48 3.3 Results ...... 49 3.3.1 Line Flux M aps ...... 50 3.3.2 Diagnostic Emission-Line Ratio Maps ...... 52 3.3.3 Integrated flu x e s ...... 59 3.4 Discussion ...... 60 3.4.1 Ionization Edge ...... 60 3.4.2 Filam ents ...... 61 3.4.3 Nebular G eom etry ...... 63 3.5 Conclusions ...... 66

IV NGC 7662 and NGC 7009 ...... 81

4.1 Introduction ...... 81 4.1.1 Previous Studies of NGC 7662 and NGC 7009 ...... 83 4.2 Observations, Data Reduction and Analysis ...... 84 4.3 Results ...... 86 4.3.1 NGC 7662 ...... 86 4.3.2 NGC 7009 ...... 90 4.4 Discussion ...... 93 4.5 S u m m ary...... 98

V Conclusion ...... 116

5.1 Summary...... 116

vin 5.2 Future W o r k ...... 118 5.3 Final R e m a rk s ...... 119

BIBLIOGRAPHY ...... 121

IX L i s t o f T a b l e s

TABLE PAGE

1 Sample of Planetary Nebulae ...... 28

2 Journal of Fabry-Perot Observations ...... 28

3 Interference Filter Characteristics ...... 29

4 Direct Images...... 30

5 Longslit Spectra Observations ...... 31

6 NGC 6720: Integrated Line Strengths ...... 68

7 NGC 6720: Integrated Spectroscopic Properties ...... 68

8 NGC 7662: Integrated Line Strengths ...... 99

9 NGC 7009: Integrated Line Strengths ...... 99 L i s t o f F i g u r e s

FIGURE PAGE

1 PN Spectra ...... 10

2 Balick (1987) Morphological Sequence ...... 11

3 NGC 7662: Identification of Morphology ...... 12

4 NGC 6720 Ha+[N II] Spectrum ...... 32

5 NGC 7662 [SII] S p e c tru m ...... 33

6 NGC 7662 Ha data cube Fitting Regions ...... 34

7 NGC 7662 Ha Spectra ...... 35

8 NGC 6720 Flux calibration ...... 36

9 NGC 6720: Emission-line flux maps from the data c u b e s ...... 69

10 NGC 6720: Emission-line flux maps from direct im a g in g ...... 70

11 NGC 6720: [0 I]A6300 em ission ...... 71

12 NGC 6720: Baimer decrement ...... 72

13 NGC 6720: Electron density in the S'*" region ...... 73

14 NGC 6720: Electron temperature in the N"*" region ...... 74

XI 15 NGC 6720: Ionization structure ratio maps - HeIIA4686/H/3 ...... 75

16 NGC 6720: Ionization structure ratio maps - [0 I]A6300/[0 III]A5007 76

17 NGC 6720: Ionization structure ratio maps - [0 III] A5007/H/3 .... 77

18 NGC 6720: Ionization structure ratio maps - [0 I]A6300/H/3 ...... 78

19 NGC 6720: Ionization structure ratio maps - [N II]A6583/Ha ...... 79

20 NGC 6720: Ionization structure ratio maps - [S II]A6717+6731/H^ . 80

21 NGC 7662: Direct Im ages ...... 100

22 NGC 7662: Fabry-Perot Im ages ...... 101

23 NGC 7662: Balmer Decrement ...... 102

24 NGC 7662: Electron density ...... 103

25 NGC 7662: [0 III]A5007/H/3 ...... 104

26 NGC 7662: HeIIA4686/H/? ...... 105

27 NGC 7662: [0 III]A5007/HeIIA4686 ...... 106

28 NGC 7662: [0 I]A6300/H^ ...... 107

29 NGC 7662: [N II]A 6584/H a ...... 108

30 NGC 7009: Direct Im ages ...... 109

31 NGC 7009: Fabry-Perot Images ...... 110

32 NGC 7009: Balmer Decrement ...... I l l

33 NGC 7009: Electron Density ...... 112

Xll 34 NGC 7009: HeIIA4686/H/3 ...... 113

35 NGC 7009: [O III]A5007/H;3 ...... 114

36 NGC 7009: [N II]A 6584/H a ...... 115

Xlll CH APTER I

Introduction

1.1 Planetary Nebulae

Planetary nebulae (PNs) occur near the end of the life of an intermediate-mass star.

During the transition from a red giant star into a white dwarf, the outer atmosphere of the red giant star is ejected and forms an expanding gaseous shell. Meanwhile, the remnant stellar core continues to contract and heat up as it evolves into a degener­ ate white dwarf. Ultraviolet radiation from the hot remnant core photoionizes the surrounding gas shell which cools by emitting line radiation, and thus an emission- line nebula, the PN, is seen. In time, the gas shell vanishes into the surrounding interstellar medium and the remnant core cools, eventually radiating all of its energy away.

The PN phase is relatively short-hved, lasting only about 10,000 years. There are only about 1500 known PN in our galaxy, and of these only a few are sufficiently close

and bright enough to allow us to study their structure in detail.

Unfike stars or galaxies, the spectra of PNs are dominated by emission lines su­

perimposed on a weak continuous spectrum. PNs are identified by their characteristic

emission-line spectra, especially the strong [0 III]A5007Â emission line. Spectra from

1 two PNs are shown in Figure 1. PNs can be distinguished from other galactic emission line objects (such as H II regions) by their morphology and spectral-line intensities

(see Acker et al. [1992] for selection criteria).

The most important connection between current issues in astronomy and PNs is to stellar evolution. In the structure of PNs we hope to read the history of the asymptotic giant branch (AGB) phases of stellar mass-loss. A clear understanding of the stellar mass-loss rates and chemical abundances in the ejected envelopes are necessary to establish their contribution to the chemical enrichment and evolution of our galaxy. Additionally, PNs are a unique laboratory for atomic physics. PNs are, by terrestrial standards, extremely low density (10^ — 10^ cm“^) and spectral lines

(in particular, the collisionally excited forbidden lines) are observed in PNs which cannot be generated in laboratory conditions. Observations of spectral lines in PNs have provided excellent tests of atomic physics calculations.

1.1.1 Evolution of PNs

The exact details of the transition from an AGB star to a white dwarf are not well understood, but some recent progress has been made. One especially difi&cult problem was the extreme range of PN morphologies. The morphologies of most PNs can perhaps be unified into one evolutionary scheme proposed by Balick (1987), who considered the Interacting Stellar Winds model first applied to PNs by Kwok et al.

(1978). He proposed that the morphologies of PNs are determined by the interaction of winds from the AGB/PN central star. Generally, mass loss from the AGB star is characterized by three distinct stages, each with different mass-loss rates and wind velocities. The first stage is a slow AGB wind {v ~10 km s“^, about the escape velocity from the AGB star) with a mass loss rate of about M = 10“® — 10“^ M© yr“^. In the second stage the mass loss increases to M ~ 10““* M® yr“^ and there is a slight increase in the wind velocity.

It is during this stage that most of the mass loss occurs. Occasionally the simplest models of evolution will combine these two stages, although some observations suggest that mass loss may occur in several episodes or “pulses” (Frank 1993). Finally, as the star begins its evolution into a white dwarf, a very fast but tenuous wind

{v ~ 1000 km s“^,M = 10“^ — 10“® M© yr“^) driven by radiation pressure from the increasingly hot remnant core is produced. This completes ejection of the star’s envelope.

The morphologies of most PNs can be explained within this framework. The in­ teraction of the fast wind with previous winds alters the original density structure in the nebula (established by mass loss during the earlier AGB star phase) via the combined effects of hydrodynamic shocks and ionization fronts. The fast wind creates a hot central bubble, or core, that expands and plows the material in the AGB wind into a thin shell. Balick (1987) proposed that the morphologies of many PNs can be explained by a density gradient (from equator to pole, generally with with a higher equatorial density) in the ABG wind. Because the hot core expands faster in the lower density polar direction, three basic morphological types are created depend­ ing upon the degree of density contrast: (1) a uniform density produces a spherical nebula, (2) a low density contrast produces an elliptical nebula, and (3) a high den­ sity contrast produces a butterfly-shaped nebula. Figure 2, reproduced from Balick

(1987), shows this morphological sequence. Some of the morphological features that appear in NGC 7662 have been identified in Figure 3. Hydrodynamic models using an axisymmetric slow wind have been successful in reproducing the basic run of mor­ phologies observed in most PNs (Balick 1987; Icke et al. 1992; Mellema 1993; Frank

& Mellema 1994).

The differing velocities and densities set up a complex hydrodynamic situation, requiring detailed hydrodynamic modeling of the evolution in order to investigate the original mass-loss process. In the last five years, hydrodynamic models of the evolution of PNs have become quite sophisticated, calculating the evolution of the density, temperature, and velocity of the gas in great detail. The most complex models to date are those of astronomers V. Icke, A. Frank, and G. Mellema (see, for example,

Mellema &: Frank [1994] and references therein). Their numerical models are based on three interacting winds with a variety of mass-loss scenarios, including aspherical slow winds and time-dependent fast winds. A limited treatment of radiation processes

(including the evolution of the central star) has been incorporated. These models have successfully reproduced the overall morphologies, the spectral-line profiles of several lines, and the kinematics of a few PNs.

1.1.2 Pine Structure in PNs

When a PN is imaged in the light of low ionization potential emission lines, such

as [N II] and [0 I], it often takes on a very filamentary appearance. In contrast. the same PN will appear smooth when imaged in emission lines of high ionization potential, such as Hell or [0 III]. This can be seen in my images of the Ring Nebula

(Chapter 3), and was first noticed in other PNs by Capriotti et. al. (1971). Low ionization small-scale structures (knots and FLIERS, defined below) are identified in

NGC 7662 (Figure 3).

Filamentary-looking nebulae are expected when [0 1] line emission is strong (Kaler

1980) because strong [0 I] emission requires the presence of cool, optically thick, dense condensations (Williams 1973). [0 I] emission can be produced in the observed amounts by the ionized “skins” of the condensations and the “shadowed” regions behind them. These shadowed regions, which appear as filaments, are shielded from the radiation from the central star and are photoionized by diffuse nebular radiation.

A successful model for the Helix Nebula (Capriotti 1973) was composed of an optically thin gas with filaments (which were the shadowed regions behind optically thick condensations) that were 1.7 times denser than the adjacent non-shadowed regions and had a temperature of 6000 K. Boeshaar’s (1974) study of nine PNs also observed the higher densities predicted by the models. However, Hawley & Miller

(1977) and Kupferman (1983) have found no density enhancements in the bright filaments of the Ring Nebula, and Balick et al. (1993) have found fine structures, which they call FLIERS (Fast Low Ionization Emission Regions), that also do not fit the Capriotti’s model, and in fact so far cannot be adequately modeled by either photoionization or shock heating. 1.2 Diagnostic Emission Lines

The physical processes occuring in a PN axe measurable from its emission lines. The basic principles behind emission-line analysis is discussed, for example, by Osterbrock

(1988), Aller (1984), and Pottasch (1984). With exact atomic calculations, accurate measurements of the physical conditions in the nebula can be made (e.g. temperature, density, ionization state, and total and ionic abundances).

Typically, a PN will exhibit strong radial stratification of ionization, with high ionization species predominating at smaller radii close to the ionizing star, and with low ionization species occurring at large radii. The wide range of ionization states observed in PNs is diagnostic of the heating and cooling processes occuring in the nebula. Emission lines from very high ionization species can be observed in PN, such as Hell A4686Â with an ionization potential 54.4 eV, up to [Si VII]A2.48/fm with an ionization potential of 205.0eV (Ashley & Hyland 1988). These species are photoion­ ized by radiation from hot staxs (T* up to 200,000 K) or ionized by shock heating.

Emission lines from low-ionization species are also observed: [N II] (ionization poten­ tial 14.5 eV), and [0 IJA6300Â at ionization fronts or in condensations.

Especially important in the study of nebulae are emission-line ratios that are sen­ sitive to the local density and temperature of the gas. The ratio of [SII]A6717/6730Â or [0 H]A3726/3729Â (both ions have the same electronic structure) is an electron density diagnostic in the range = 10^ - 10^ cm“®, a typical observed range of densities in PN (Osterbrock 1988). Electron temperature (Te) can be determined with the lines [N H]AA6584,5755Â or [0 IH]AA5007,4363Â (Osterbrock 1988; Kaler 1986). The Balmer decrements H q:A6563/H/3A4861/H7A4340 can be used to accu­ rately measure interstellar or internal nebular reddening, which is diagnostic of the presence of dust in the nebula or intervening ISM. Very accurate measurements of the reddening can be made when the density and temperature of the emitting gas are

known because the atomic transitions probabilities for hydrogen are well established

(even reasonable estimates of Te and Ng will suffice as the Balmer decrements are not

particularly sensitive to these parameters).

1.3 Two-dimensional Spectrophotometry

A look at the photographs in any PN catalog (such as Balick 1987 or Perek & Kohutek

1967) shows that PNs are richly structured. Within a given PN the density, tempera­

ture and ionization state can vary significantly not just as a function of distance from

the central star, but also in each direction from the central star. A necessary step in

trying to understand how the general and detailed structure of a PN is produced is

to make spatially resolved spectroscopic observations of representative PNs.

Besides yielding a large increase in the number of points in the nebula in which

we know the physical conditions, two-dimensional maps also make it easier to discern

patterns that might otherwise go unnoticed. For example, it is possible to look at

specific morphological components such as the entire outer edge of a nebula or all of

the filamentary regions.

Most previous spectrophotometric studies of PNs have been limited to longslit

spectra that only measures the physical conditions in a small portion of the nebula.

Two-dimensional spectrophotometry using narrowband filter imaging, which samples 8 the whole nebula, has been done for only a few nebulae because of the inherent diffi­ culties of the method (see Jacoby, Quigley & Africano 1987) especially the problems eliminating or assessing the contamination of line fluxes by nearby lines falling into the filter bandpass. For example, the [S II]A6717/6730 density diagnostic depends on small differences in intensity between the two closely spaced lines, making conven­ tional narrowband filter imaging impractical. There is also a particularly troublesome problem in some low-ionization PN (e.g. the Ring Nebula) where [N II]A6583Â is strong relative to Ha, making cross-talk between adjacent emission-line bands signif­ icant.

When closely-spaced lines need to be observed, more accurate two-dimensional spectroscopy can be accomplished using Fabry-Perot interferometry. The utility of this technique has been demonstrated in a recent study of the Orion Nebula by Pogge,

Owen & Atwood (1992). In this project I have used a combination of Fabry-Perot interferometry, narrowband filter imaging, and longslit spectrophotometry to achieve good spatial and spectral resolution of emission lines.

1.4 Focus of this project

There are two main points to explore with the data set for each PN. First, the large scale structure of PNs can be examined. This is particularly important since current hydrodynamic evolution models can now predict not only the shapes but also the

detailed physical characteristics of PNs, such as density and temperature. Ionization fronts are apparent in our data, the critical parameter of density can be examined,

and possibly temperature changes due to shocks can be looked for. Second, small-scale structure in these PNs can be examined. Although appearing smooth in Ha emission, some PNs look very clumpy and filamentary in low ioniza­ tion potential (LIP) emission lines such as [N II] and [0 I] (Capriotti 1971). This fine structure appears in several forms: ansae (NGC 7009), knots (NGC 7662), conden­ sations (the Helix Nebula), and filaments (the Ring Nebula). These microstructures are very tiny (1-2"). The density and temperature diagnostics used in this project are for low-ionization ions. The good spatial coverage and spatial resolution of my data will enable a thorough look at these small scale structures, especially in the Ring

Nebula.

In this project I make maps of emission-line ratios measuring the Baimer decre­ ment, electron density in the S’*" zone, electron temperature in the N"** zone, and ionization state for the PNs NGC 6720 (the Ring Nebula), NGC 7662, and NGC 7009.

In Chapter 2 ,1 describe the observations and data reduction methods. This is a hybrid approach using three different types of observations, and so each of the meth­ ods of reduction as well as the process of combining the observations are discussed.

My initial target, NGC 6720, served to work out these methods and the results of that work are presented in Chapter 3. Results for two additional nebulae, NGC 7662 and NGC 7009, (which, in contrast to the Ring Nebula, are high-excitation PNs) are presented in Chapter 4. A summary of the results on the nebulae and future work is in Chapter 5. 10 600 n—1—r "I " I—r o NGC 6720 400 I i l s (O 0 ? CO 200 3 I I 1 z œ s- m I X X ^ I 1 n I I I Li 20 : 15 0 I g « i| uT' I X 10 iI 1 3 I 0 (D CO 1 ^ 1 X < 1 5 I T I

-uùsi I ■ I . I I r - I i Jtl 3000 a> NGC 7662 (O 2000 s s 1000

o 100 s CO

IL 60 Soo 3 ^ 40

4500 5000 5500 6000 6500 7000 wavelength (A)

Figure 1: Sample spectra of two PNs are shown. Bottom panel in each is an enlarge­ ment showing the faint emission lines. The differences between low-excitation PNs (NGC 6720) and high-excitation PNs (NGC 7662) are seen. 11

MORPHOLOGICAL TYPES, DESIGNATIONS, AND EXAMPLES

E5

! O 1

R8 E8 ‘- 0 - - -BatiNP. JELLroAL -BOIERIIY CLQSED INNER HALO CLOSED INNER HALO PRE LOBE NGC 1535, 1C 3568, NGC 2022,3242,6826,7662, 1C 5217 BD-»303639, H91-4 1C418, M^51 OPEN LOBE OPEN INNER HALO OPEN INNER HALO NGC 2346, 6781*, NGC 6894,7139, NGC 40,2392*, 2610,6905,7009, 101747*, M2-9 1C 1454 7027,7048,7354, 1C 289, A 82 BIPOLAR OUTER HALO ROUND OUTER HALO BIPOLAR OUTER HALO NGC 650-1,2440, NGC 6543,6826, A 30 NGC 2438*, 2371-2,6720*, 6853,6905,7026 6309,6445, A 79

Figure 2: Balick (1987) Morphological Sequence for Planetary Nebulae (reproduced with the permission of the author). 1 2

NGC 7662 [O l]16300 + Hp

FLIERS

Figure 3: Morphological features in NGC 7662 axe identified on this composite image of [0 I] and H/3. The bright rim is a high-ionization structure, and the knots and FLIERS axe low-ionization structures. CH APTER II

Observations and Data Reduction

In this chapter, the observations and the data reduction and analysis techniques used in this study are described. I have taken a hybrid approach in which several different methods of observations are combined to exploit the advantages and mitigate the problems of each. The full data set for a PN is comprised of three parts: (1) direct images of single emission lines made through narrowband interference filters,

(2) Fabry-Perot spectral data cubes covering more than one emission line and (3) calibrated longslit spectra.

The advantage of observing a spectroscopic data cube of the lines of interest over conventional narrowband filter imaging techniques is twofold. First, because closely spaced lines such as Ha, [N II]A6548,6584Â and [S II]AA6717,6731Â may be deblended using model line profiles, uncontaminated line fluxes can be measured. Second, the line modeling procedure also provides estimates of the continuum fluxes at the line wavelengths, providing a more accurate estimate of the continuum than would be obtained by extrapolating from images taken though filters centered on adjacent continuum bands. It is important to emphasize that I am not simply using the

Fabry-Perot as a tunable filter (i.e. Jacquinot spot imaging, see e.g., Bland-Hawthorn

1994), but as a true three-dimensional spectrophotometer with the expressed goal of

13 14 calibrated imaging spectrophotometry.

As stated before, using Fabry-Perot spectroscopy for measuring emission-line flux is an unconventional technique, and new methods of hybrid spectroscopy were re­ quired. I demonstrate that the results are comparable to longslit spectroscopy in accuracy, but with a factor of >100 greater spatial coverage.

2.1 Sample Selection

The criteria for the ideal object for this project was that the PN be bright, have a large angular size, and show interesting structure. The objects chosen needed to be not only bright overall, but also bright in [S II]A6717,6731Â emission so that the data cube for these lines could be taken within one night. PNs larger than about 30" (but less than 4') were considered.

Other considerations were a constraint in position of the target (a limit of -1-52°N declination and about —10°S declination), a desire that the final set of objects be close in position (i.e. to observe more than one object per observing run), and the poor weather during the summer monsoon season in Arizona.

Balick’s (1987) PN catalog was a good source from which to choose targets. The final set of targets (NGC 6720, NGC 7662 and NGC 7009), while conforming to the ideal criteria stated (see Table 1), was mostly the result of when the weather cooperated. 15

2.2 Imaging Fabry-Perot Spectroscopy

2.2.1 Observations

Observations were obtained at the 1.8m Perkins Telescope in FlagstaiF, AZ (at Ander­ son Mesa) using the Ohio State University Imaging Fabry-Perot Spectrograph (IFPS;

Pogge et al. 1995). Fabry-Perot observations for each nebula consists of two data cubes: one centered at A6563Â (covering the Ha and [N II]AA6584,6548Â lines) and the other centered at A6724Â (covering the [S II]AA6717,6731Â doublet). A list of the six data cubes taken is in Table 2.

The IFPS is a focal reducing camera using a Queensgate ET50 Fabry-Perot étalon with a T I800 x 800 CCD detector at a pixel scale of 0.5"/pixel. A low-resolution étalon

(200 km/sec) was used, producing a spectral resolution of about 4Â FWHM. This resolution is insufficient to resolve the expansion velocity of the nebula (see Table 1), but well suited for clearly resolving the Ha-f [N II] and [S II] line blends. All line profiles are determined by the instrumental profile, which is well approximated by a Lorentzian function. Adjacent orders of interference were rejected with narrow­ band interference filters. The filters used had central wavelengths/bandwidths at

6630/150Â and 6780/150Â, and were tilt-tuned to the central wavelength of the data cube (see Section 2.3.1).

Reflections between the CCD detector and the optical surfaces of the étalon plates will produce faint (2-3%) “ghost” images of the object. The ghosts will appear on the CCD image at a position reflected about the optical axis (see Bland-Hawthorn

[1994] for a discussion of ghost families in Fabry-Perots). Because the PNs observed 16 occupy a significant portion of the CCD frame it is not possible to position the object so that the ghost reflections do not overlap with the primary image. The étalon was therefore tilted by 3° relative to the optical axis in order to deflect ghost reflections away from the CCD detector. This does not affect the position of the object in each image, but does affect the surface of constant wavelength in the data cube in that the equations describing this surface become more complex. It is then necessary to follow the wavelength calibration procedure described in Section 2.2.2.

The wavelength sampling of the data cubes was chosen so as to minimize time- dependent effects such as changes in seeing, temperature, or air transparency through­ out the night. Interferograms covering the cores of the emission lines were obtained first, in the event that conditions deteriorated, followed by the continuum interfero­ grams which are less sensitive to seeing variations.

Since the IFPS has no active temperature control for the etcdons, the étalon gap was monitored for drifts caused by changes in the ambient temperature and continu­ ally adjusted during observations. Temperature changes affected the observations in two ways: (1) the zero-point of the wavelength calibration “drifts” and (2) the effec­ tive finesse of the étalon degrades. The zero-point drift was monitored with regular, interspersed observations of a calibration lamp during the data cube observations.

The success of the étalon drift corrections is seen in the uniform, symmetric line- profiles in the data cubes. The techniques for compensation for étalon drift were worked out and implemented after the the initial Ring Nebula Ha data cube obser­ vations were obtained. There is, however, no evidence of significant étalon gap drifts 17 in this data cube as judged by the quality of the spectra.

2.2.2 Basic Reductions

The raw data are in the form of a series of interferograms taken at different étalon gap spacings. The process of converting the raw interferograms into a wavelength calibrated cube is what has given Fabry-Perots their reputation for being difficult to

use. Explicit procedures for data reduction and analysis with the IFPS are given in

Pogge (1991).

Basic reduction of the interferograms is similar to standard CCD imaging meth­

ods (see Pogge 1992). For the IFPS system, the necessary steps were overscan bias

subtraction, flat-fielding, and sky subtraction (sky emission was taken to be con­

stant across each frame). The frames were registered with respect to a fiducial using

field stars on the frames. Two additional corrections were made: (1) atmospheric

extinction and (2) white-light corrections.

The white-light correction is used to correct small differences in filter transmission

(as a function of wavelength) in the order-separation interference filters used in the

IFPS. Filter transmission can also vary as a function of position, but that was not

significant in these data sets. To calculate the correction needed at each wavelength, a

white-light cube (a coarsely sampled series of flat field images covering the data cube

spectral range) was obtained, and the relative transmission at the wavelength of each

interferogram (for the position of the nebula) was measured. Each interferogram

was then divided by a normalized relative transmission factor. This method only

works for the case of small corrections or angularly small objects because wavelength 18 is not constant across an interferogram. However, even in the large Ring Nebula, the phase surface (wavelength as a function of position on the frame) was shallow enough that a constant correction for each frame sufficed. When a white-light cube was not available, the longslit spectra were used to determine the overall effect upon emission-line strengths due to filter transmission.

Wavelength calibration of the data cubes is typically the most difficult part of the Fabry-Perot data reduction. Wavelength is a linear function of étalon spacing.

However, a surface of constant wavelength in the data cube is parabolic rather than flat (i.e. the pixels in each frame are not at the same wavelength). Thus, the data cube is a collection of spectra in which both the dispersion (A/pixel) and starting wavelength are functions of position. A map of the relative wavelength zero-point of each spectrum in the data cube (the “phase map”) was constructed from observations of a Neon or Hydrogen calibration lamp. The phase map is used in the automated line- fitting routines (described in section 2.3) to put each spectra onto the same absolute wavelength scale before the line-fitting.

2.2.3 Analysis

The data cubes were analyzed using the Fabry-Perot reduction package OASIS (Pogge

1991). The end result of the analysis is a measurement of the spectral line parameters

(line center, total flux, and FWHM), continuum, and RMS uncertainty of the line

profile fit at each location in the object, which are assembled into maps of flux,

continuum, etc., across the nebula. Fits in regions of good S/N can be made to ±2%

RMS of residuals of the data and fit, and fits worse than ±10% were not accepted. 19

In low S/N spectra, the most difficult problem was fitting the continuum, which contributed the major portion of the uncertainty to the line measurements. Sample spectra from the data cubes are shown in Figures 4 and 5.

The line-profile fits were constrained using physically reasonable assumptions to reduce the number of free parameters. For example, all lines in a blend were con­ strained to have the same FWHM. The lines are expected to have the same FWHM since velocity structure of the lines was not resolved. Another fit constraint was that the flux of the weak [N II]A6548Â emission line was constrained to be one-third the flux of the [N II]A6584Â line (simply the ratio of the transition probabilities as both of these lines arise from the same excited state) and also constrained by relative wavelength when necessary.

However, the line-fitting routines used to analyze the data cubes were not able to converge in the case of a strong Ha line blended with very weak [N H]AA6584,6548Â lines, even in a strongly constrained case ([N II]A6548=0.333x[N II]A6584, FWHM of all lines equal in every spectrum). In regions where [N H]A6584Â was >5% of Ha, a full three-line deblend was possible. In regions where [N H]A6584Â was 1-5% of

Ha, a two-line deblend was performed (only Ha and [N H]A6584Â). Lastly, where

[N II]A6584Â was <1% of Ha, the line-fitting routines were not able to converge and only the Ha line could be fit. At this point [N II]A6584Â is not visible in the spectrum, or only marginally detected after subtraction of the best fit model Ha line profile.

We can thus determine the [N II] flux accurately to 1% of Ha, a level which would be very difficult to achieve with narrow-band filters (see, for example, the NGC 7662 2 0

[N II] image in Balick 1987). The fitting regions for NGC 7662 are shown in Figure 6 and sample spectra from each region are shown in Figure 7. There is good agreement between fits on the boundaries of the regions.

Emission from HeIA6678Â from an adjacent order of interference appeared in some

[S II] data cubes (only in high-excitation PNs where this line is very strong). This contaminating line appeared in the red continuum portion of the [S II] spectra (see

Figure 5). The contamination was treated by using only the blue continuum points for continuum fitting (the contaminated red points were easily rejected statistically as they generally deviated by more than 2a from the mean continuum level). With fewer points to determine the continuum, the continuum measurement and thus the

[S II] fluxes have a higher uncertainty. The [S II] A6731Â line flux measurement was checked in several ways to ensure that the Hel line did not systematically affect it.

The [S II] line fluxes have an accuracy of ±2-7% even in the rim region (where there is weak [S II] emission and strong Hel emission).

To investigate the halos of the PNs, where the emission is too faint for line fitting in the data cubes, line fluxes were measured by direct integration over wavelength bands by simulating “on-band” and “off-band” top-hat filters. Contamination from the wings of neighboring emission lines was corrected for by using the measured instrumental profile from the brighter regions to estimate contribution from line wings.

Although the results of this “band averaging” technique are slightly less accurate than those obtained from profile fitting in bright regions, this technique proved useful for inherently low flux spectra where the nonlinear least-squares fitting procedure gave 2 1 fits with laxge variations in parameters or nnacceptably large RMS residuals.

2.3 Direct Imaging

2.3.1 Observations

Isolated spectral lines were observed with the IFPS in direct-imaging mode (with the

étalon removed from the beam). The emission lines observed in this manner were:

HeIIA4686Â, Hjg, [0 III]A5007Â, [N II]A5755Â, and [0 I]A6300Â. A list of the filter wavelengths and bandwidths is given in Table 3 and observations for each PN are listed in Table 4.

Continuum images were made by “tilt-tuning” the filters by 15° relative to the optical axis to observe the blue continuum. The central wavelength of interference filters can be shifted to the blue by tilting the filter. The new central wavelength is given by the equation: sin^'g 1/2 Ag = Ac 1 - N2 (2.1)

where 6 is the tilt in degrees, and N is the index of refraction of the filter (N = 1.7 was

used for the IFPS narrow-band filters). The A4861/30Â, A5755/30Â and A6300/30Â filters were tilted to provide continuum images. A description of the properties of

interference filters is in Pogge (1992).

It should be noted that the optics on the IFPS were upgraded in September 1992.

Images from the two different optical configurations were kept separate in subsequent

reduction and analysis to avoid introducing an additional source of error from the

geometric coordinate transformation necessary to put them on the same effective 2 2 pixel scale.

2.3.2 Reduction

The images taken with the IFPS in direct-imaging mode were reduced with IRAF or

VISTA using standard techniques (see Pogge 1991). The basic procedures for each image were: overscan bias subtraction, flat-fielding, sky subtraction (as for the data cube interferograms, sky was assumed constant across the image), and registration.

Multiple exposures were summed together.

Nebular continuum was subtracted from all images except those of [0 III]A5007Â.

The continuum contribution to the [0 IIIJA5007Â image is small, and the lack of subtraction of nebular continuum from the [0 III]A5007Â image did not affect line- ratio maps. For weak emission lines such as [N II]A5755Â, the nebular continuum

W e i s a significant contribution to the on-band image. The H/?, [N II]A5755Â and

[0 I]A6300Â images were all corrected for nebular continuum emission by using off- band images using the same tilted filter. Field star profiles in the continuum images were not distorted, evidence that tilting the interference filters does not introduce undesirable effects in the images. The Hell image was corrected for nebular continuum using the same continuum image as used for the H(3 images.

There was some contamination by adjacent emission lines in the narrowband filter images, but in all cases it was small enough that the results were not significantly affected. The [0 I]A6300Â images were contaminated by [S III]A6312Â emission that falls inside the wing of the filter bandpass (filter transmission T\ 63i2 = 0.4Ta63oo)-

Longslit spectroscopy shows that the [S III] emission is limited in extent to the inner. 23 high-ionization core of nebulae where it is strongly correlated with [0 III]A5007Â emission. Thus, in the outer [0 I]-bright regions, little contamination of the [0 I] flux by [S III]A6312Â is expected. The Hell images were also slightly contaminated by the weak adjacent lines [Ar V]-fHeI A4711Â that fall inside the filter bandpass. Again,

longslit spectroscopy was used to determine the extent of the contamination and to

determine what, if any, corrections should be applied. Finally, in the H/3 bandpass

no correction was made for the weak HeIIA4859Â emission line as its contribution is

expected to be < 1.5% (Brocklehurst 1971; Richer, McCall & Martin 1991). Overall,

contamination in any direct image was not large enough to cause problems.

2.4 Longslit Spectroscopy

2.4.1 Observations

The longslit spectra were obtained on the 1.8m Perkins Telescope using the Boiler

& Chivens CCD spectrograph with a 350 I mm“^ grating (FWHM=5Â). A list of

observations is in Table 5. Usually, the slit was placed on the central star, and

oriented along position angle 90°. The slit width was 2" and had a length of 3'.5

(unvignetted portion). It was not possible to observe the entire spectral range of

interest in one spectrum, so two spectra covering the desired range were taken. The

wavelength ranges were either A4725 — 5952Â and A5613 — 6848Â or A4250 — 5475Â

and A5630 — 6870Â. Non-overlapping spectra were placed onto the same absolute

flux calibration scale using lower-dispersion spectra taken with a 150 I mm“^ grating

in the same position (these spectra were not used for calibration of individual lines

because the spectral resolution was inadequate to deblend the emission lines observed 24 with the Fabry-Perot).

2.4.2 Reduction

The longslit spectra were reduced with IRAF using the IMRED and TWODSPEC packages. Flux calibration used spectra of standard stars observed throughout the night.

Sky subtraction was not routine for these spectra because many PNs have faint extended halos whose spectral lines contaminated the sky regions. These emission lines were removed from the sky spectrum by interpolation from adjacent spectral regions. The only nebular emission line for which this could not be done was the

[0 IJA6300Â line which is blended with the strong [0 I] night-sky line. Thus, night sky emission may have been slightly oversubtracted in the longslit spectra on that particular line. It is unlikely to be the case in NGC 7662 and NGC 7009 as they have very faint outer halos (Balick et al. 1992) but for the Ring Nebula it is more of a concern. The [0 I] image shows [0 I] emission is very weak in the outer halo of the Ring Nebula except at the extreme edges. It is estimated that the nebular [0 I] contribution to the [0 I] night-sky line is no more than 12%.

2.4.3 Analysis

The longslit spectra were used primarily to provide a relative flux-calibration of the

emission-line images. Spectra were extracted from 3"x2" apertures along the slit, and

all emission lines were fit using a Gaussian profile where possible (the fluxes for weak

emission lines were measured using band integration.) The relative flux calibration 25 is very accurate, to ±5%, but the absolute flux calibration is not as well determined

(±30%) since usually the longslit spectra were not taken in photometric conditions.

2.5 Putting it all together...

2.5.1 Flux Calibration of the Emission-line Images

Relative flux calibration between the data cubes and individual direct images was accomplished using flux-calibrated longslit spectra of known positions in the neb­ ula. Intensity (counts) in the narrow-band fllter image or Fabry-Perot flux map was measured in synthetic apertures the same size and position as those extracted from the longslit spectra. A least-squares fit is made to the points (CCDS flux against

IFPS counts); the slope of the line is the flux calibration coefficient. Calibration of the emission-line images (both Fabry-Perot and direct imaging) was done individ­ ually for each line, and yields spectrophotometric accuracy comparable to longslit spectrographic observations (Figure 8).

2.5.2 Ratio Maps

The final calibrated emission-line flux maps from the data cubes and direct imaging were used to produce the diagnostic line-ratio maps that indicate the ionization state, reddening, electron temperature, and electron density. Two additional steps need to be taken to generate the line-ratio maps: spatial registration and matching the seeing of the images.

Registration between image frames was done using stars on the frames. Flux images from data-cube analysis were registered using an unmasked “mash” (the sum 26 of all the interferograms in the cube) of the data cube which showed field stars.

Registration in this manner should be adequate, but it was sometimes found to be off by as much as few tenths of a pixel. This is due to a systematic charge-transfer- efficiency problem with the CCD which smeared low-level point sources vertically.

Because planetary nebulae have such a sharp dropoff in intensity at the edge of the nebula (or in certain features), misregistration by even this small amount was usually very obvious and easily corrected.

Most direct images and data cubes were taken in about l'/5-2'/0 seeing, the best seeing available at Anderson Mesa. Worse seeing conditions were generally not consid­ ered acceptable for imaging. Dividing images taken during different seeing conditions will produce artifacts in the line-ratio maps reminiscent of the effects of “unsharp- mask” processing (a form of high-pass spatial filtering used to enhance the contrast of fine structures in complex images). The Ha cube for the Ring Nebula was taken during 3" seeing and so it was necessary to match the stellar point-spread functions between images whenever constructing ratio maps with Ha or [N II]A6584Â (the two diagnostic maps affected are the [N II] temperature map and the Baimer decrement map). This is particularly important in the Ring Nebula because of the rapid changes in intensity found in the filaments and the outer edge of the nebula. The direct image was first registered with respect to the Ha data cube, then smoothed to Z" effective

seeing using a Gaussian kernel. The smoothing kernel width was determined by com­

paring the stellar point-spread functions in the Ha data cube “mash” and the direct

image (before continuum subtraction). 27

Maps of the electron density and temperature were constructed from flux-calibrated, extinction-corrected line-ratio maps of [S II]A6717/673l and [N II]A6584/5755. The line-ratio maps were converted to density and temperature following Osterbrock

(1989).

Images used in ratio maps have been corrected for extinction using the interstellar extinction curve of Cardelli, Clayton & Mathis (1989). The reddening parameter c{HP) was calculated from the ratio of the integrated Ha and fluxes, assuming an intrinsic Case B recombination ratio of Ha/Hyd=2.86 (Hummer & Storey 1987).

Because the reddening in all of the PNs in my sample is small, using a different intrinsic ratio will not affect the data significantly. Also, the variations in the observed

Baimer decrement across the PNs was found to be small, and so a constant correction in each case is acceptable. 2 8

Table 1: Sample of Planetary Nebulae®

Angular Object Diameter log HjS Flux^ NGC 6720 76" -10.08 26.5 km s ^ NGC 7662 17" —9.99 27.5 km s~^ NGC 7009 28.5" -9.78 20.6 km s~^

“data from Acker et al. (1992) ^total flux (erg cm“^) “expansion velocity measured in 0 III]A5007Â emission line

Table 2: Journal of Fabry-Perot Observations

Lines AA Int.“ N L f N“spec AA"' Seeing® UT Date N G C 6720 Ha, [N II] 6523 - 6596 180s 33 6328 3.7Â 2.8" 1991 May [SII] 6690 - 6758 300s 36 4514 3.8Â 2.0" 1992 June N G C 7662 Ha, [N II] 6526- 6600 60s 51 1280 4.2Â 2" 1991 Nov [S II] 6692- 6755 600s 36 769 3.3Â 1.5" 1992 Sept N G C 7009 Ha, [N II] 6515- 6609 60s 48 1374 3.9Â 1.0" 1992 Sept [S II] 6692- 6750 360s 33 489 3.7Â 1.5" 1992 Sept

“integration time in seconds of each interferogram ^number of interferograms in the data cube “number of spectra fit in the data cube “^FWHM of the observed emission line profiles “mean FWHM of field stars in a sum over interferograms 29

Table 3: Interference Filter Characteristics

Filter Parameters rnc ID A: AA^ •^peak Emission Lines

468.6/6 4692Â 56Â 0.73 HeIIA4686, [Ar IV]A4711^ 486nb3 4861Â 30Â 0.64 H/3A4861 501nb3 5007Â 30Â 0.70 [OIII]A5007 576nb3 5755Â 30Â 0.77 [N II]A5755 630nb3 6300Â 30Â 0.77 [0 I]A6300, [S III]A6312“

“central wavelength ^Full Width at Half Maximum “peak filter transmission “^filter transmission T=0.68 at A4711Â “filter transmission T=0.30 at A6312Â 30 Table 4: Direct Images

Line Filter® Tilt^ A: Int.'^ Seeing® UT Date N G C 6720 Hell 468.6/6 0° 4692Â 2400s 175 1992 Sept 29 H/3 486nb3 0° 4861Â 1200s 2" 1992 June 1 1200s 175 1992 Sept 29 H/3 cont. 486nb3 15° 4804Â 1200s 175 1992 Sept 29 [0 HI] 501nb3 0° 5007Â 540s 2" 1992 June 1 [Nil] 576nb3 0° 5755Â 1200s 2" 1992 June 1 [N II] cont. 576nb3 15° 5688Â 1200s 2" 1992 June 1 [01] 630nb3 0° 6300Â 900s 2" 1992 June 1 630nb3 0° 6300Â 2700s 175 1993 Sept 9 [0 I] cont. 630nb3 15° 6227Â 900s 2" 1992 June 1 630nb3 15° 6227Â 2700s 1.5" 1993 Sept 9 N G C 7662 Hell 468.6/6 0° 4692Â 900s 177 1992 Sept 28 H/3 486nb3 0° 4861Â 720s 177 1992 Sept 28 H/3 cont. 486nb3 15° 4804Â 900s 177 1992 Sept 28 [0 HI] 501nb3 0° 5007Â 70s 177 1992 Sept 28 [Nil] 576nb3 0° 5755Â 3600s 177 1992 Sept 28 [N II] cont. 576nb3 15° 5688Â 3600s 177 1992 Sept 28 [01] 630nb3 0° 63G0Â 1200s 177 1992 Sept 28 [0 I] cont. 630nb3 15° 6227Â 1200s 177 1992 Sept 28 N G C 7009 Hell 468.6/6 0° 4692Â 600s 1725 1992 Nov 23 H/3 486nb3 0° 4861Â 240s 175 1992 Sept 29 240s 1725 1992 Nov 23 240s 175 1993 Sept 9 H/3 cont. 486nb3 15° 4804Â 1800s 1725 1992 Nov 23 600s 176 1993 Sept 9 [0 HI] 501nb3 0° 5007Â 75s 175 1992 Sept 29 [Nil] 576nb3 0° 5755Â 3600s 2" 1993 Sept 8 [N II] cont. 576nb3 15° 5688Â 3600s 2" 1993 Sept 8 [01] 630nb3 0° 6300Â 3600s 275 1993 Sept 8 [0 I] cont. 630nb3 15° 6227Â 3600s 3" 1993 Sept 8

“Filter ID (see Table 3) ^Filter tilt “effective central wavelength of filter “^total exposure time in seconds “FWHM of field stars 31

Table 5: Longslit Spectra Observations

Object A Range Grating PA“ Int.^ UT Date NGC 6720 4725-5952À 350 i mm“^ 90° 900s 1992 June 6 5613-6848Â 350 i mm~^ 90° 2100s 1992 June 5 4250-5475Â 350 I mm~^ 129° 1200s 1993 June 26 4280-7100Â 150 i mm“^ 90° 1410s 1992 Sept 3 NGC 7662 4250-5475À 350 I mm“^ 90° 750s 1993 June 28 5630-6870Â 350 i mm"^ 90° 630s 1992 Sept 1 4280-7100Â 150 € mm“^ 90° 395s 1992 Sept 3 NGC 7009 4250-5475À 350 £ mm“^ 90° 840s 1993 June 28 5630-6870Â 350 I mm“^ 90° 410s 1992 Sept 1 4280-7100Â 150 I mm“^ 90° 180s 1992 Sept 3

“position angle of 2"-widtli slit (90° is East-West) ^exposure time in seconds 32

NGC 6720 Ha+[N II] spectrum

1000 Ha

800

B 600

400 [N ll]X6548

200

10 15 20 25 30 35 40 étalon gap (z)

Figure 4: Spectrum from one location in the NGC 6720 Ha data cube. The data are plotted as a histogram and the line fit as a solid line. The line fit is a Lorentzian profile, which matches the instrumental profile. This particular data cube is slightly undersampled (FWHM of the emission lines is 1.7 pixels). 33

NGC 7662 [S II] spectrum 300

250

200

3 150

100

étalon gap (z)

Figure 5; Spectrum from one location in the NGC 7662 [SII] data cube (in the SE knot). The HeIA6678Â emission line comes from an adjacent order, and appears to the right of the [S II]A6731 emission line (the Hel emission line is partially cut-off). 34

NGC 7662 Fit Regions

I I I I I I I I I I I I I I I I I I I I I r I I I

20

10

-10

-20

-■3 line fit , ■ 2 line fit 1 line fit - - II I I I I I I I I I I I I I I I I I I I I I I 4 -20 -10 0 10 20

Figure 6: Map of line-fitting regions for the NGC 7662 Ha data cube. In the black regions a three-line deblend was possible ([N II]A6584 > O.OSHa). In the dark gray regions, a two-line deblend was used. In the light gray regions, only Ha could be fit ([N II]A6584 < O.OlHa). 35

NGC 7662 sample spectra

3 line fit

[N ll]A.6584 = 17% Ha

[N li]X6584

Ha

2 line fit

[N ll]^6584 = 1% Ha

[N ll]X6548 [N ll]X6584 o c Ha

1 line fit

[N ll]^6584 < 1% Ha

[N ll]À6584

Ha

20 30 40 50 60 étalon gap (z)

Figure 7: Sample spectra from each of the regions in Figure 6. 36

40 7 a) Hp

30 c 3 8 CO 20 CL IL 10

0 0 1 2 3 4 Longslit Flux (xlO"^® erg sec"^1 _cm~“) 3\

a) Ha

IL 5000

Longslit Flux (xlO"^^ erg sec"’ cm"®)

Figure 8: Presented are the flux calibration for a) and b) Ho: emission lines in NGC 6720. Each point represents flux in a 2"x2" aperture along a cut through the central star at position angle PA=90°. Flux from the longslit spectra apertures is plot­ ted against counts from the IFPS apertures. The solid line represents a least-squares fit to the points; the slope of the line is the flux calibration coefficient. CHAPTER III

NGC 6720 - The Ring Nebula

Published Work

The following chapter has been previously published as Lame, N. J. & Pogge, R.

W., 1994, “Imaging Spectrophotometry of the Planetary Nebula NGC 6720 (the Ring

Nebula)”, Astronomical Journal, 108, 1860.

Reformatting for the dissertation included renumbering of figures and tables to ensure continuity, and moving bibliographic material to the general list of references.

37 38

3.1 Introduction

Within a planetary nebula (PN) the density, temperature and ionization state can vary significantly not just as a function of distance from the central star, but also in each direction from the central star. Emission-line spectroscopy can be used to study these physical parameters across the nebula. Most previous spectrophotomet- ric studies of the Ring Nebula, the most notable of which are Dufour (1993), Hawley

& Miller (1977), and Barker (1987), have been limited to only partial spatial cover­ age, which only measures the physical conditions in a small portion of the nebula.

Two-dimensional spectrophotometry using narrowband filter imaging, which covers the whole nebula, has been done by Reay & Worswick (1977), and Kupferman (1983).

This method of narrowband filter imaging is difficult, however, (see Jacoby, Quigley

& Africano 1987) and can be problematic due to contamination of the filter bandpass from nearby lines. In the Ring Nebula this is a particularly troublesome problem due to strong [N II]A6583 relative to Ha, making contamination between bands signifi­ cant. The [S II]6717/6730 density diagnostic depends on small differences in intensity between the two closely spaced lines, making conventional narrowband filter imaging impractical.

When closely-spaced lines need to be observed, more accurate two-dimensional spectroscopy can be accomplished using Fabry-Perot interferometry. The utility of this technique has been demonstrated in a recent study of the Orion Nebula by Pogge,

Owen & Atwood (1992). In this paper we present a study of the Ring Nebula that uses a combination of Fabry-Perot interferometry, narrowband filter imaging, and longslit 39 spectrophotometry to achieve good spatial and spectral resolution of emission lines.

We make maps of emission-line ratios measuring the Balmer decrement, electron density in the S'*’ zone, electron temperature in the N"^ zone, and ionization state.

Besides yielding a large increase in the number of points in the nebula in which we know the physical conditions, two-dimensional maps also make it easier to discern patterns that might otherwise go unnoticed. For example, it is possible to look at the entire outer edge of the nebula, all the filamentary regions, and specific morphological components. The large spatial coverage of our data combined with its more accurate spectral resolution allows us to investigate the ionization edge, the filaments, and the geometry of the Ring Nebula.

As discussed by Balick et al. (1992), the Ring Nebula at first glance appears to fit the classical model of an ionization bounded nebula. Ionization is observed to be radially stratified, with high ionization lines such as Hell produced closer to the central star than low ionization lines such as [N II] (Reay & Worswick 1977).

Molecular hydrogen is observed along the outer edge of the nebula (coincident with the [0 I] emission) (Greenhouse et al. 1988, Zuckerman & Gatley 1988, Kastner et al. 1994). The Ring Nebula, however, has two partially ionized halos that are outside, or at least are projected outside of the ionization front (Balick et al. 1992).

Kinematical observations of the Ring Nebula show that the inner halo is probably bipolar and projected outside the ionization front (Bryce et éd. 1994). There is no such evidence for the outer halo. The ionization front can be studied in detail with our data because the entire outer edge of the nebula is continuously covered. 40

The intensity of low ionization potential lines, especially [0 I], with respect to high ionization lines, indicates conditions at the edge of the nebula.

When the nebula is imaged in the [N II] and [0 I] emission lines, it has a very filamentary appearance. Emission from the low ionization potential (LIP) species 0° and N"*" is quite strong in the Ring Nebula compared to other PNs (Campbell 1968;

Hawley & Miller 1977). Filamentary-looking nebulae are expected when [0 I] line emission is strong (Kaler 1980) because strong [0 I] emission requires the presence of cool, optically thick, dense condensations (Williams 1973). [0 I] emission can be produced in the observed amounts by the ionized “skins” of the condensations and the

“shadowed” regions behind them. These shadowed regions, which appear as filaments, are shielded from the radiation from the central star and are photoionized by diffuse nebular radiation. While Boeshaar (1974) found the expected higher densities in the filaments of the Ring Nebula, other studies (Hawley & Miller 1977, Kupferman 1983) have claimed that no density enhancements existed in the filaments. With our two- dimensional spectrophotometry, we study the physical conditions across the filaments in order to compare to these models and to observations of filaments in other PN.

The geometry of the Ring Nebula has never been firmly established. Fabry-Perot kinematics of the high ionization [OIII]A5007 emission show a spherical expansion

(Atherton et al. 1978), and Kupferman (1983) found that Hell emission is consistent with a closed sphere geometry. These lines, however, arise primarily from the hot inner regions immediately surrounding the central star. It has long been known that emission from low ionization lines, such as [N II] and even the hydrogen Balmer lines, 41 has a distribution that is not consistent with a spherical geometry and indicates a deficit of material of these ionization species along the line of sight to the central star of the Ring Nebula (Curtis 1918; Kupferman 1983). Minkowski & Osterbrock (1960) suggested that the Ring Nebula is a bipolar nebula similar to NGC 650-1 that is being viewed along the polar axis. Recently Volk & Leahy (1993) also suggested this based on their deprojected emission images. Current models of planetary nebula structure and evolution are based on the interacting stellar winds theory (Kwok et al. 1978). In the simplest case of these models, a fast (1000 km/sec) tenuous wind from the central star interacts with a slow dense AGB mass-loss wind. The fast wind creates a hot bubble core which expands and snowplows material in the AGB wind into a thin shell.

The morphologies of many PNs can be explained by a density gradient (from equator to pole, generally with with a higher equatorial density) in the ABG wind (Balick

1987). Because the hot bubble expands faster in the lower density polar direction, three basic morphological types are created depending upon the density contrast: a uniform density produces spherical nebulae, a low density contrast produces elliptical nebulae, and a high density contrast produces butterfly-shaped nebulae. We use our electron density map and emission-line flux images to place constraints on the possible geometry of the Ring Nebula, vis-à-vis these basic forms.

In Section 3.2, we detail the observations and data analysis. In Section 3.3, we describe the results, namely, the line flux maps and diagnostic emission-line ratio maps. A discussion and summary follows in Section 3.4. 42

3.2 Observations and Data Reduction

In this section we describe the observations and data reductions. Using Fabry-Perot spectroscopy for measuring emission-line flux-maps is an unconventional technique.

We demonstrate that our results are comparable to longslit spectroscopy in accuracy, but with extended spatial coverage.

3.2.1 Observations

All observations of the Ring Nebula were obtained with the 1.8m Perkins Telescope in Flagstaff AZ. The full data set is comprised of single emission-line images, data cubes covering more than one emission line, and calibration longslit spectra. The advantage of observing a spectroscopic data cube of the lines of interest over conven­ tional narrowband filter imaging techniques is twofold. First, because closely spaced lines such as Ho:,[N II]A6548,6583 and [S II]AA6717,6731 are deblended using model

(Lorentzian) line profiles, we can obtain uncontaminated line fluxes. Second, the line fitting procedure also calculates continuum fluxes at the line wavelengths, providing a more accurate estimate of the continuum than would be obtained by extrapolating off-band images. We emphasize that we are not simply using the Fabry-Perot as a

crude tunable filter, but as a true three-dimensional spectroscope with the expressed

goal of calibrated imaging spectrophotometry.

The Imaging Fabry-Perot Spectrograph (IFPS) (more fully described in Pogge,

Owen & Atwood 1992) was used for acquisition of the data cubes. The IFPS is

a focal reducing camera using a Queensgate ET50 Fabry-Perot étalon with a TI 43

800x800 CCD detector at a pixel scale of 0.5"/pixel. A low-resolution étalon (200 km/sec) was used, producing a spectral resolution in the data cubes of about 4Â

FWHM. This resolution is insufficient to resolve the expansion velocity of the nebula

(26 km/sec, Weinberger 1989), but well suited for clearly resolving the H q +[N II] and [S II] line blends. All line profiles are determined by the instrumental profile, which is well approximated by a Lorentzian function. Our observations were taken in the form of two data cubes, centered at A6563 (Ha and [N II] lines) and A6724

([S II]AA6717,6731 doublet) in May 1991 and June 1992. In Table 2 (Chapter 2) we list the wavelength coverage (Col. 2), exposure time of each frame (Col. 3), number of interferograms in each data cube (Col. 4), the number of spectra that were fit in each data cube (Col. 5), FWHM of the emission lines (Col. 6) and the seeing during the data cube observations (Col. 7). Since the IFPS has no temperature control, the étalon was monitored for reactions to changes in ambient dome temperature and continually adjusted during observations for the [S II] data cube. We saw no evidence of wavelength variation due to étalon temperature changes in the Ha data cube line profiles.

Since using the IFPS in spectroscopic mode (with the étalon) requires several hours of photometric conditions, it is more efficient to observe isolated lines in direct imaging mode using narrowband filter techniques (Pogge 1992). In direct imaging mode, the étalon is removed from the IFPS, making it a conventional focal-reducing system. We observed the HeIIA4686, H^, [0 III]A5007, [N II]A5755, and [0 I]A6300 emission lines in this mode during June and September 1992 (see Table 4, Chapter 2). 44

Continuum images were made by “tilt-tuning” several of the filters 15° to observe the blue side continuum. Field star profiles in the continuum images were not distorted, evidence that tilting filters does not introduce undesirable effects in the images.

Relative flux calibration between the data cubes and individual direct images was

accomplished using flux-calibrated longslit spectra of known positions in the nebula.

The longslit spectra were obtained on the 1.8m Perkins Telescope in June 1992 and

June 1993 using the Boiler & Chivens CCD spectrograph with a 350 In/mm grating

(FWHM=5Â). The slit was placed on the central star, and oriented along position

angle 90° (1992) and 129° (1993). The slit width was 2" and had a length of 3'.5

(unvignetted portion). It was not possible to obtain the entire spectral range on one

spectrum, so two spectra covering the range were taken. The wavelength ranges were

A4725 — 5952Â and A5613 — 6848Â for the 1992 longslit spectra and A4250 — 5475Â

and A5630 — 6870Â for the 1993 longslit spectra. The 1992 spectra were used for all

flux calibrations except the HeIIA4686 image.

3.2.2 Data Cube Analysis

The data cubes were analyzed using the data cube reduction package OASIS (Pogge

1991). The spectral line parameters (line center, total flux, and FWHM), continuum,

and RMS uncertainty of the line profile fit were measured at each location in the

object and assembled into maps of flux, continuum, etc., across the nebula. Fits can

be made to ±2% RMS of residuals of the data and fit, and fits worse than ±10%

were not accepted. In low S/N spectra, the most difficult problem was in fitting the

continuum, which contributed the major portion of the uncertainty to the fit. A mask 45 over the field of view was applied before analysis of the data cubes to remove spectra with very faint emission. The edge of the mask is slightly outside the sharp edge of emission, and none of the structures observed are artifacts caused by masking.

Although both data cubes were analyzed in a similar manner, analysis of each data cube had its own idiosyncrasies. For the Ha data cube, the FWHMs for all 3 lines was constrained to be a specific number (i.e. the instrumental width). This was necessary because the spectra are slightly undersampled in wavelength. The regions selected for continuum determination were weighted to the blue side of the spectra to

avoid contamination of the continuum with [NII] emission, due to the inherently large

Lorentzian wings in the line. The Ha continuum measured in the data cube does not

show a more filamentary structure than the direct image continuum at A6300, and so

we do not see significant contamination from [N II]. For the [S II] cube, the FWHM

was allowed to vary as a function of position, however we constrained the FWHM of

the two lines to be the same. Because the [S II] emission in the center of the nebula

is weak relative to the continuum, we were not able to make good fits to the lines in

that region.

To investigate the halo, where emission is too faint for line fitting in the data

cubes, integration over wavelength bands was used in the data cube. Integration is

over simulated “on-band” and “off-band” top-hat filters, and contamination from the

wings of neighboring emission lines can be corrected for by assuming the measured

instrumental profile. Although results from band averaging are slightly less accurate

than those from line fitting in bright regions, we used this technique only for inherently 46 low flux spectra where the nonlinear least-squares fitting procedure gave fits with large variations in parameters or unacceptably large RMS residuals.

3.2.3 Emission-line Image Analysis

The images taken with the IFPS in direct-imaging mode were reduced with IRAF using standard techniques. The optics on the IFPS were upgraded between June and

September, and so the two sets of images are kept separate to avoid introducing an additional source of error from the necessary geometric coordinate transformation.

The [N II]A5755 and [0 I]A6300 images used in the final analysis were corrected for nebular continuum emission with off-band images using the same filter. In these weak lines, the nebular continuum was a significant contribution to the on-band image (22% of [N II]A5755). The Hell image was corrected for nebular continuum observed with the tilted (off-band) H^ filter. A continuum image near was not obtained in June

1992. However, the nebular continuum emission is weak at A4861, about 2% of H/3 flux. This was determined by comparing the total flux in the Sept. 1992 H/5 image to the total flux in the Sept. 1992 continuum subtracted H/3 image, as well as from the longslit observations. Because the continuum contribution in H/3 is small, subtraction of nebular continuum from the June 1992 H/5 image was not necessary as it did not affect line-ratio maps.

There was some contamination by adjacent lines in our narrowband filter images,

but in all cases it was small enough that our results were not significantly affected. The

[O I]A6300 image is contaminated by [S III]A6312 emission that falls inside the wing

of the filter bandpass (filter transmission Ta6312=0.4 Ta 63oo)- Longslit spectroscopy 47 shows that the [S III] emission is weak ([S III]A6312/[0 I]A6300=0.01-0.05 in strong

[0 I] emission regions, depending on radius), and limited in extent to the inner, high-ionization core of the nebula where it is strongly correlated with [0 III]A5007 emission. Thus, in the outer [0 I]-bright regions, we expect little contamination of the

[0 I]. The Hell image is also slightly contaminated by the weak adjacent lines [Ar V] and HeIA4711 that fall inside the filter bandpass. Longslit spectra show this blended line has a maximum strength of ~ 7.5% of HeIIA4686 strength. In the H/3 bandpass, no correction was made for the HeIIA4859 line as the contribution should be less than

1.5% (Brocklehurst 1971; Richer, McCall k Martin 1991). Overall, contamination in any direct image was not large enough to cause problems.

Although the direct images and the [S II] data cube all have comparable seeing, the H q data cube was taken during 3" seeing and so it is necessary to match the point spread functions between images whenever constructing ratio maps with Ha or

[N II]A6583 (the two diagnostic maps affected are the [N II] temperature map and the

Balmer decrement map). This is particularly important in the Ring Nebula due to the spatially sharp changes in intensity formed by the filaments and the outer edge of the nebula. The direct image was first registered with respect to the Ha data cube, then smoothed to 3" effective seeing using a Gaussian kernel. The smoothing kernel width was determined by comparing the stellar point spread functions in Ha data cube mash (the sum of all the interferograms in the cube) and the direct image (not continuum subtracted). 48

3.2.4 Flux Calibration

To calibrate our data cubes and direct image fluxes we compared them to flux- calibrated longslit spectra. Longslit spectra at PA=90° were used. The regions used for sky subtraction in the longslit spectra were at a radius of about 100", which is still within the faint halo of the nebula (Balick et al. 1992 give an outer halo diam­ eter of 270"). Faint nebular emission lines from the halo could be seen in the sky spectrum (the strongest, [N II]A6583, had a peak about 3.5 times sky level). These emission lines were removed from the sky spectrum by interpolation from spectrally adjacent regions. The only nebular emission line for which this was not done was the

[0 I]A6300 line, which is blended with the strong [0 I] night sky line. Our [0 I] image shows [0 I] emission is very weak in the outer halo except at the extreme edges. We estimate that nebular [0 I] contribution is no more than 12% of the [0 I] night sky line. Thus, sky may be slightly oversubtracted in the longslit spectra on that par­ ticular line. Apertures (3"x2") along the slit were extracted, and all emission lines were fit using a Gaussian profile where possible. (The flux for weak emission lines was measured using band integration.)

Calibration of the emission-line images (both Fabry-Perot and filter) was done individually for each line, and as can be seen in Figure 8 (Chapter 2), we have achieved photometric accuracy comparable to our longslit spectrographic observa­ tions. Fabry-Perot derived line ratios for [N II]A6583/Ho: are in excellent agreement with the longslit ratio. The flux calibrations show the need for minor corrections to ac­ count for structure in the order-separation filter transmission curve. This white-light 49 correction is seen as a 25% difference between the Ha and [N II]A6583 flux calibra­ tions. Longslit calibrations for the AA6717,6731 lines showed that data cube also required a small white light correction (3%). We stress that the relative flux calibra­ tion is very accurate, to 5%, but the absolute flux calibration is not as well determined

(30%) since the calibration longslit spectra were not taken in photometric conditions.

Images used in ratio maps have been corrected for extinction using the interstellar extinction curve of Caxdelli, Clayton & Mathis (1989) and using Av=0.69 (c=0.322) derived from the ratio of the integrated Ha and H/? fluxes (discussed in section 3.3), assuming an intrinsic Case B recombination ratio of 2.86. Because the reddening is small, using a different intrinsic ratio will not affect the data significantly. The variation in observed Balmer decrement across the nebula is also small, and so a constant correction is acceptable.

3.3 Results

We present the final line-emission and line-ratio maps in this section. A definition of our morphological terminology is followed by a description of the line flux maps in

§3.1. In §3.2 we present maps of physical conditions in the nebula (electron density, etc.) with a description of how we estimate these physical quantities and an esti­ mation of errors. We compare these results to previous spectrophotometric studies, particularly Barker (1987), Hawley & Miller (1977) and Kupferman (1983). We also present results from our calibration longslit spectra when relevant. 50

3.3.1 Line Flux Maps

We identify two important morphological features; the shell and the core. The shell is the bright “ring” of the nebula with a round inner radius of 20" and an outer elliptical semi-major axis of 50". The core is the inner 20" of the ring, and appears as the “hole” in the center of the shell. We shall adopt this terminology throughout the rest of the paper. Note that Balick et al. 1992 refer to the “ring” as the core of the nebula. However, their morphological assessment does not allow us to distinguish between the “ring” and the “hole”. In general, the Ring Nebula does not fit Balick’s

(1993) bright rim 4- attached shell morphological picture of a PN, hence the confusion.

Emission-line flux images from data cube reductions and direct images are shown in

Fig. 9. Below we discuss the emission features in each of the three morphological components: the shell, the core, and the halo.

Flux from most emission lines is strongest in the shell. As has been noted before

by Capriotti et al. (1971), the images show varying amounts of filamentary structure

depending upon the emission line imaged. In general, higher ionization images are

smoother (e.g. [0 III]A5007, Fig. 10) and lower ionization images are more filamen­

tary, with the [0 I] image (Fig. 10) showing extreme filamentary structure. In the

[0 I] images, radial scalloped-like features can be seen in the filaments along the inner

edge of the shell.

The central core is strong primarily in high ionization potential (HIP) lines.

He IIA4686 emission (Fig. 10) fills the center inside the bright shell. The Hell emission

is rounder than the emission (e.g. Ha) in the surrounding shell. Reay & Worswick 51

(1977) showed that there is relationship between ionization potential and apparent el- lipticity of the nebula in a particular emission line, with the nebula appearing rounder in images of higher ionization potential lines. Structure can be seen in the Hell images in the form of bands across the nebula parallel to the major axis. These stand out quite clearly in the HeIIA4686 images (about 15% brighter than surrounding emis­ sion) and can also be seen traversing the center of the nebula in the Ha, H^ and

[0 III] emission-line images (Fig. 10). Although low ionization emission lines such as [0 I] are extremely weak in the core, anomalous structures can be seen in these lines (the most obvious is the two blobs of emission inside the hole seen in the [0 I] image).

In general, our observations of the halo confirm the findings of Chu, Jacoby &

Arendt (1987); Moreno &: Lopez (1987); Balick et al. (1992); and Bryce, Balick &

Meaburn (1994). The nebula is surrounded by two halos, shown in a deep [0 I]A6300

(Fig. 11) image. The brighter inner halo is more structured than the larger outer halo. Strong edge brightening in the inner halo is clear. A cut across the one of the southeast edges shows that the edge is about 3" wide and is three times brighter than the inner halo. There is extreme contrast because the inner halo is almost as faint as the outer halo in this emission line, unlike in [N II] and [0 III]. However, faint radial spokes of [0 I] emission are seen from the outer edge of the shell, extending into the inner halo. Edge brightening of the outer halo can also be seen in the [0 I]A6300 image. It is extremely circular with a radius of 109". The edge brightening is well- resolved (7" on the southeast arc) and rather clumpy. Both the inner and outer 52 halos emit diffuse [0 III]A5007, but edge brightening in this emission line is just barely seen in the inner halo only (emission is ~ 10% brighter in the edge). A deep

Ha+[N II]A6583 emission hne image formed by summing all the frames in the data

cube also shows the loops, filaments, and arcs seen in the [0 I] image. Emission

from the two lines can be separated using band integration. Overall [N II] is ~ 2.0

times stronger than Ha in the halo. Our observations confirm the finding of Moreno

& Lopez (1987) that most of the fine structure in the halo seen in the the light of

Ha and [N II] is due to [N II] emission. The [N II] edge brightening is ~ 7% - 25%

brighter than [NII] inner halo emission. In contrast, the Ha halo emission is relatively

smooth. An azimuthal average of Ha emission in the outer halo, fit with a power

law, shows emission decreasing in brightness as r~“, where a = 1.85±0.10, consistent

with confinement of the outer halo (Frank et al. 1990). Halo emission is not seen in

the summed interferograms from the [S II] data cube, but this is an artifact of our

relatively brighter detection limits in this band.

3.3.2 Diagnostic Emission-Line Ratio Maps

Balmer Decrement

The Ha/H/3 map is shown in Fig. 12. Recall that the H/? image was smoothed to

match seeing with the Ha flux map. Careful attention was paid to the registration

process, with registration accurate to ±071 along rows and ±0725 along columns. Fur­

ther refinement of the registration was accomplished by keying on the sharp increase

in [N II]A6583/H;S ratio at the edge of the nebula (see section 2.5.2). Duplicating the 53 structure of the [N II]A6583 /H q: ratio provides registration accurate to ~ ±0.1". As the variation in Ha/H/3 is less than 20%, this method produces good results. The flux calibration between the two frames (which would only affect the mean ratio) depends strongly on our ability to connect the separate red and green longslit spectra accurately, and has an error bar of ~ ±6% in the ratio.

The mean H q /H/? ratio in the map is 3.74±0.22. There is a smooth steepening of the Balmer decrement, to a ratio of ~3.93, around the 3/4 of the shell of the nebula.

This steepening is a smooth increase in Balmer decrement from the inner edge of the shell to the outer edge. Hawley & Miller (1977) noted a steepening to a ratio of

4.7 in one position of their aperture spectra of the Ring Nebula and attributed it to increased reddening as it appeared on only one of their apertures. The mean ratio in the inner 15" of the nebula is 3.67, significantly lower than in the shell of the nebula

(excepting the northeast side). The northeast side has a mean ratio of 3.39, and this does not have the same radial dependence as the ratio in the rest of the shell.

The Balmer decrement is slightly sensitive to temperature and we see a difference in the Balmer decrement between core and shell which is consistent with the expec­ tation that the core is at a higher temperature than the shell. We observe a ratio of

Ha/H|8=3.68 in the core compared to 3.74 (average) in the shell. If we correct for interstellar extinction by using our mean observed Ha/H/3 ratio of 3.74 and assuming the intrinsic ratio is 2.86 (the value for Tg=10,000, Ng = 10^ cm~^, assuming Case

B recombination as computed by Hummer & Storey 1987), then the corrected ratio in the center of the nebula is 2.81. This ratio is expected of a gas at Tg=13,500 K. 54

Barker (1987) estimates aa electron temperature of Te=13,000 K in the O^"*" region.

Since it is expected that optical depth in the nebula varies as a function of position

and that the gas is not perfectly Case B at all points, the lower Balmer decrement in the core could alternatively be due to stronger deviations from the assumption of

Case B recombination. An examination of the H 7 /H /3 ratio in our calibration longslit

spectra shows similar behavior as H/3/Ha.

[S II] Density

The electron density (Ne) was estimated using the [S II]A6716/6731 line ratio using

computed line strengths from Cai &: Pradhan (1993), for an assumed constant electron

temperature of 10^ K. A map of Ng across the nebula is shown in Fig. 13. Since the

density is derived from the line ratio of two lines originating in the same data cube,

errors in the line ratio are relatively small. The major sources of error are uncertainty

in fitting the continuum and in the line deblending procedure. The rms errors in the

flux were ~ ±2-3% for most spectra. The precision of the derived Ng depends on

the relation of Ng to the [S II] ratio. For the range of Ng in this nebula, a ±3% error

in the [S II] doublet ratio corresponds to ±70 cm“^ in Ng. The small systematic

white-light correction applied to the [S II] data cube only affects the average density,

not the point-to-point density variations seen.

The highest density regions (Ng ~ 700 cm“^) appear as clumps in a broad band

about the minor axis in the shell of the Ring. These density enhancements are located

~ 8" inward from the outer edge of the nebula (as defined as the sharp dropoff of Ha 55 emission). The density is lower (Ng ~ 500cm“^) in the rest of the nebula. It appears that the density enhancements extend in filaments toward and across the major axis on the inner edge of the shell. However, the [S II] flux drops off rapidly at the inner edge of the shell, and so the density measurement is less certain. Band integration in the central region of the nebula does not show high density structure along the inner edge. Our density map is not entirely dissimilar to Kupferman’s (1983) density map, but the higher spatial resolution of our map shows that the density enhancements in our map are resolved and appear quite sharp.

The density enhancements coincide with the general locations of some of the fila­ ments identified in the [0 I]A6300 emission image, but there is no detailed correlation.

Projection effects would make it possible to see interior, sharp density enhancements if more filaments were crossed along the line of sight on the minor axis than on the

major axis. In this case, the [S II] density map and the [S II] emission maps should

look similar, but they do not (also noted by Kupferman 1983). An attempt to sepa­

rate the filaments and the surrounding nebula into separate components showed that

the filaments are not significantly more or less dense than in the general density map.

Comparing the direct flux maps at [0 I]A6300 and [S II] that best show the filaments,

we find no density contrast between the high surface-brightness filaments and the

nebula in which they are embedded. We discuss this surprising result in Section 3.4.2

below. 56

[N II] Temperature

The electron temperature (Te) was calculated using the [N II]A6548+6583/5755 line ratio following the five-level atom calculation of Osterbrock (1989), assuming a con­ stant density of 500 cm“^ (consistent with our mean Ng estimated from the [S II] ratio map). The ratio was corrected for extinction as described in Section 3.2.4. Although

the density map clearly shows that the density is not constant across the nebula, this

simplifying assumption is acceptable as the electron temperature calculation is not

very sensitive to electron density at our measured Te and Ng.

Uncertainties in the [N II] ratio map are dominated by the signal-to-noise in the

direct image of the weak [N II]A5755 line. An estimate of the uncertainty in the

[N II]A5755 image was made in the following way. Although the seeing was 1.5", the

initial pixel size before binning was 0.5". The standard deviation of a running mean

of a 3x3 box (within the seeing disk) was taken to be an estimate of a. This is an

overestimate in areas where the emission is changing rapidly, such as at the edge of

the shell. In the brightest regions, errors in fiux are approximately ±10%. Because

the ratio map was derived from images not taken at the same time and in the same

manner ([N II]A6548,6584 are from a data cube, [N II]A5755 is from a direct image)

a determination of the errors must also include the additional data reduction steps

of registration and flux calibration. Emission from the nebula drops off sharply, and

so errors in registration are very obvious in line-ratio maps. We are confident of our

image registration, with errors estimated to be ±0.15". Errors in the fiux calibration

factor (which would effectively shift the mean temperature up or down) we estimate 57 to be about ±3%, dominated by the uncertainty in the calibration of the [N II]A5755 image (due to the weakness of the line in our longslit spectra). A ±10% error in the

[N II] ratio corresponds to ~ ± 375 K at a temperature of 9900 K, but this is an absolute error, and we are much more sensitive to relative temperature variations.

The resulting [N II] temperature map (Fig. 14) shows a smooth, fairly constant temperature distribution across the nebula. The mean temperature is 9643 K with a standard deviation of 230 K (disregarding the noisy center and edge pixels). Barker’s

(1987) aperture spectra give a mean [0 III] temperature of 10,000 K and a mean [N II] temperature of 9400 K. However, our mean [N II]A6584/5755 ratio (not corrected for extinction) is 75, which is comparable to Barker’s (1987) median ratio of 85. The difference in temperature is partly due to using different atomic constants (Barker uses

Osterbrock 1974). Kupferman’s (1983) mean [0 III] electron temperature is 150 K lower than our average [N II] electron temperature. Our calibration longslit spectra include the [0 III]A4363 emission line, and confirm that the [0 III] temperature is higher than the [N II] temperature in the Ring Nebula. Thus, Kupferman’s (1983)

[0 III] electron temperature is probably too low. Kaler (1986) predicts a nebula with HeIIA4686=0.24H/3 should have an [0 III] temperature of 11092 K and a [N II] temperature of 10300 K.

Azimuthal averaging of the temperature map shows a relatively flat curve in the main body of the nebula and there does not appear to be a systematic tempera­ ture trend with radius. The radial decrease in [0 III] electron temperature seen by

Kupferman (1983) is not present in the [N II] temperature (this behavior in both line 58 ratios is also confirmed by our calibration longslit spectra). This difference between the two temperature diagnostics in radial behavior is understandable since the zone is expected to be confined to a smaller radial extent than the 0 ^^ zone. Could we be missing temperature variations due to the poorer seeing in the [N II] images?

Possibly, but we note that filamentary structure is still apparent in a [0 1] flux im­ age smoothed to 3" so it is unlikely that strong temperature differences between the filaments and the surrounding gas are being missed by our temperature map. More significantly, our longslit spectra, taken in 1?5 seeing, confirm a lack of temperature contrast between the filaments and their surroundings.

Ionization Structure

Maps of the diagnostic emission-line ratios: HeIIA4686/H;3,[0 I]A6300/[0 III]A5007,

[0 I1I]A5007/H^, [0 I]A6300/HyS, [N II]A6583/Ha, and [S II]A6717+6731/H;5 are shown in Figures 15-20. Line ratios formed from the Fabry-Perot spectral maps have been masked, showing zeros where no fits can be made because of low S/N.

Ratios formed entirely from direct imaging show mostly noise outside the visible shell of the nebula.

The most striking feature of these line ratio maps is the very sharp drop in ion­

ization coincident with the sharp drop in emission at the edge of the nebula. This

sharp ionization edge is resolved, with a width of in the [0 I]/[0 III] map. The

ionization edge also appears clumpy, particularly in this image. This ionization edge

is elliptical in form, following the outer edge of the shell, but it is irregular at the 59 extreme major axis ends and more clumpy and broken out than along the minor axis sides of the nebula (in the sense of extending to larger projected radius than expected for a simple ellipse).

The center of the nebula also has ionization structure as shown in both the

HeII4686/H/3 and the [0 III]5007/H/3 map. The inner ionization edge (between

[0 III]5007/H/?) is nearly circular in shape like the Hell emission, not elliptical. It is also not continuous, but appears as several overlapping arcs.

3.3.3 Integrated fluxes

The flux in each hne was measured in a 100" diameter circular aperture centered on the central star (this included part of the halo). These fluxes, with respect to H/3, are given in Table 6. For the data cubes, a correction was made for flux in the masked regions by using band integration fluxes. These corrections were less than 2%. H/3 and [0 III]A5007 images were not corrected for nebular continuum, and line ratios have been adjusted using a 2.5% continuum contribution to H/3, determined from

Sept. 1992 images.

Most of the low ionization potential (LIP) line emission in the Ring Nebula is

rather strong relative to Ha, even for a PN (Hawley & Miller 1977). A look at the

PN samples of Campbell (1968), Kaler (1980), and Kondratjeva (1993) shows that

only about ~10% of PNs have stronger [0 I] and [NII] emission (with respect to H/3).

Estimations of physical quantities based on the above integrated line strengths

are given in Table 7. The mean Balmer decrement is 3.72. The line ratios for the

entire nebula give [N II] temperature Te=9540 K and [S II] density Ng=550 cm“^. 60

3.4 Discussion

The Balmer decrement, [S II] density, [N II] temperature, and ionization state maps show physical conditions inside the nebula. Although it would be best to know these parameters at each point in the nebula, we are limited to the integration along the line of sight, yielding two-dimensional images. These two-dimensional maps reveal patterns not seen in one-dimensional spectra: a sharp drop in ionization at the outer edge of the nebula, an increase in the Balmer decrement in part of the nebula, and significant density variations along the minor axis. The maps are used to study the ionization edge, the filaments, and the possible geometry of the nebula.

3.4.1 Ionization Edge

The ionization structure maps show a sharp drop in ionization state covering the entire outer edge of the shell. An ionization front in the Ring Nebula is expected since molecular hydrogen is detected in the halo (Zuckerman & Gatley 1988, Kastner et al.

1994). In the [0 I]/[0 III] ratio map, this ionization edge appears very clumpy in that there are multiple breaks in the high [0 I]/H/3 and [0 I]/[0 III] rims around the edge of the nebula. The density map does not show actual dense clumps of material; only ionization variations are seen. The dumpiness at the ionization front may explain the partially ionized halos observed around the Ring Nebula. Mechanisms for ionization of the inner and outer halos have been discussed in several papers (Zuckerman &

Gatley 1988, Balick et al. 1992, Bryce et al. 1994, and Kastner et al. 1994). The outer, spherical halo is unlikely to be the projection of a bipolar halo because it is 61 too perfectly round (that would imply an inclination angle of exactly 0°). Instead, the outer halo may be partially ionized from radiation leaking through the ionization front. The radial spokes of [0 I] seen outside shell could be shadowed regions behind dense clumps, with radiation leaking between the spokes. Such a situation is seen in the Helix Nebula on its inner edge (Meaburn et al. 1992).

The radial increase in Balmer decrement towards the outer edge of the shell is most likely due to differential reddening in the nebula. Regions of large H a/H ^ in the shell of the nebula are spatially correlated with regions of strong molecular hydrogen emission observed by Zuckerman & Gatley (1988). Their Hg u = 1 ^ 0

8(1) flux map shows emission from molecular hydrogen at the outer edge of the shell.

The Hg emission is weakest on the northeast side where our map shows the lowest reddening, and strongest near the region of largest Balmer decrement. A recent Hg flux map by Kastner et al. (1994) shows Hg emission is spatially correlated with

[0 I] emission, again increasing toward the outer edge of the shell. The variations in Balmer decrement in the shell are thus consistent with reddening due to dust, probably in a neutral atomic and molecular sheath surrounding the ionized gas shell.

3.4.2 Filaments

We find that the filaments are not significantly denser or cooler than the surrounding nebula. Our results disagree with the measured densities of Boeshaar (1974) in the

Ring Nebula filaments of 4000-11000 cm“^ using the [S II] doublet, but confirm

Kupferman’s (1983) low-resolution electron density map which did not show higher densities in the filaments. Although the density enhancements coincide with some of 6 2 the filaments, a density structure that follows the filament structure (most obviously traced in the [0 I] flux map) is not seen. The temperature map is smooth and shows no filamentary structure even though the filaments are prominent in the [N II] flux maps. Our data show no density or temperature differences between the filaments and surrounding gas, only variations in ionization state (i.e. enhanced [0 I], [N II] and [S II] emission in spatially sharp regions).

The lack of temperature and density differences between the filaments and the rest of the nebula is unexpected since only models composed of cool, dense filaments have been able to explain the high levels of [0 I] and [N II] in certain PNs. A successful model for the Helix Nebula (Capriotti 1973) was composed of an optically thin gas with filaments (which were the shadowed regions behind optically thick condensations) that were 1.7 times denser than the adjacent non-shadowed regions and had a temperature of 6000 K. The densities and temperatures we measure for the filaments in the Ring Nebula do not show similar conditions as in this model. The

Helix Nebula is similar to the Ring Nebula in its morphology, filamentary appearance, and strong [NII] and [0 I] emission. However, its filaments are predicted and observed to be radial, unlike the filaments in the Ring Nebula, which also indicates that the

Ring Nebula filaments are not the same structures as observed (and explained) in the

Helix Nebula.

There must be a reason that the filaments in the Ring Nebula are in a different ionization state than the rest of the nebula. There are two possibilities. One is that there really are differences in temperature and density in the filaments that we have 63 not detected, either because the variations are low contrast or are strongest on small spatial scales and very high spatial resolution would be needed to see them. The other is that the filaments simply represent local variations in ionization at the nebula edge and not dense condensations embedded within. The clumpy nature of the edge (seen in the [0 I]/[0 III] map, Fig. 16) might be reflecting an overall filamentary edge to the nebula. Possibly the ionization “edge” is not a smooth surface but a network of filaments.

3.4.3 Nebular Geometry

Many different models have been proposed for the geometry of the Ring Nebula. The most recent model, based on the interacting stellar winds theory of PN evolution, is by Bryce et al. (1994). From observations of kinematics in the halo, they surmise that the nebula consists of a dense equatorial band (what we call the shell) with a bipolar inner halo. The outer halo is the relatively undisturbed red giant envelope

(RGE) wind. An unresolved question with their model is the larger than expected observed ellipticity of the shell, as compared to the ellipticity of the inner halo and their estimate of the nebula’s inclination angle. We propose that the shell of the

Ring Nebula is an intrinsically oval torus, a geometry which can be explained by the observed density structure.

The shell of the Ring Nebula is thought to be a toroid because strong emission lines from the shell are non-spherical (Kupferman 1983). Besides showing a lack of low ionization emission in the center of the nebula, our observations further support a toroid geometry. The thickness of the nebula along the line of sight at each point 64 in the nebula can be estimated using our electron density map and our H/3 flux map, since F(H/3) goes as / iVgiVpdV. The average thickness, given filling factor e, is about

2xl0^^e~^ cm (assuming a distance of 600pc). The brightest emission areas (along the minor axis) are not substantially thicker than the rest of the nebula (assuming a uniform and constant filling factor throughout the shell). The increased emission from the minor axis is therefore due entirely to the higher density observed along the minor axis. Our observations are consistent with a toroidal geometry for the shell.

Projection effects from tilting the toroidal shell cannot account for the large el­ lipticity of the outer edge of the shell (axis ratio of 0.7 indicating a tilt of 45°).

Suggestions that the Ring Nebula is bipolar and viewed at a small angle to the polar axis (Minkowski & Osterbrock 1960; Volk & Leahy 1993) have been confirmed by kinematic observations of the halo (Bryce et ai. 1994). We also estimate the tilt of the shell by looking at the [0 I] morphology. The filamentary [0 I]A6300 emission- line flux image (Fig. 10) is suggestive of a tilted cylinder geometry (see, for example, the model images in Frank et al. 1993), where the filaments trace out the ends of a cylindrical inner wall to the shell. Assuming this geometry, from the positions of the [0 I] filaments we estimate that the tilt of the nebula is ~ 30°, and the height of the cylindrical inner wall is 13" (1.2x10^^ cm at a distance of 600pc). This height is in reasonable agreement with the thickness estimated above, but the tilt angle is much lower than would be estimated from the ellipticity of the outer edge of the shell

(45°) (the tilt is consistent, however, with the ellipticity of the inner halo; Bryce et

al. 1994). Since the outer edge of the shell is more elliptical than can be accounted 65 for by tilt projection, the shell of the nebula must be intrinsically oval in shape.

We look to the observed density structure to explain the oval shape of the shell.

The current theory of PN evolution accepts density structure as the cause of asym­

metrical evolution, specifically, evolution of a PN into a bipolar nebula is caused by

a density gradient from equator to pole. The shaping of the nebula occurs as a fast

tenuous wind from the central star interacts with a previous slow dense wind. When

there is a density contrast in the slow wind, the fast wind encounters lower density

along the polar axis and will accelerate faster in that direction, producing a bipolar

nebula. Analogously, the density contrast observed in the Ring Nebula between the

major and minor axes may be responsible for the fragmentation of the nebular shell

along the major axis first.

The gas in an evolved nebula like the Ring Nebula has already undergone so much

hydrodynamic shaping that it will be difficult (or impossible) to determine the origin

of the high density structure from our observations. Either the high density region

as we see it now was inherent in the mass-loss wind, or is the result of a snowplow

effect. The morphology of the observed density enhancements does not match either

case. The observed density enhancements are clumpy and not a smooth gradient as is

usually assumed in theoretical models. Also, hydrodynamic models show the initial

smooth density gradient to be snowplowed by shocks into density enhancements at

the inner and outer edge of the shell (Frank 1994), not in the interior as seen in our

density map. 66

3.5 Conclusions

This combination of observing techniques has produced line-emission and line-ratio maps of high accuracy and good spatial coverage. These properties are essential for investigating the physical conditions in large-scale structures in the nebula.

We see no density or temperature differences between the filaments and surround­ ing nebular gas. This leaves no explanation for the strong [0 I] and [N II] emission in the filaments. Further observations which can put strong limits on the temperature and density difference between the filaments and surrounding nebula are needed. Re­ cently it has been shown that the properties of some fine structure features in other

PNs are not entirely compatible with either photoionization or shock explanations

(i.e. FLIERS, Balick et al. 1993). Thus, it is important to determine if the proper­ ties of the filaments in the Ring Nebula can be explained by the established theories of their origin and structure or if they have something in common with unexplained fine structure features in other PNs.

Our data is consistent with a intrinsically elliptical toroid geometry for the shell, which can be explained by the observed density structure using the principles of the interacting stellar winds theory. The density structure in the shell and its part in the evolution of the nebula should be more rigorously considered, since the density structure is a critical parameter in hydrodynamic modeling. Current models (see

Frank 1994; Mellema 1993) assume axisymmetry, which is starting to break down in the Ring Nebula. The Ring Nebula might make a good example case for triaxial 67 models since its structure is fairly uncomplicated (unlike highly structured PN such as NGC 6543). 68

Table 6: NGC 6720: Integrated” Line Strengths

Identification A F/F(H^) V i m ” 4861 1.00 1.00“ H e ll 4686 0.24 0.24 [0 III] 5007 10.61 10.27 [Nil] 5755 0.09 0.07 [01] 6300 0.36 0.29 Ha 6563 3.72 2^6 [Nil] 6583 6.41 4.92 [S II] 6717+6731 0.64 0.48

“100" diameter circular aperture ^corrected for extinction using c=0.322 “F(H|5)=2.45xlO~^° erg s“^ cm“^

Table 7: NGC 6720: Integrated Spectroscopic Properties

H a/H ^ 3.72 ± 0.22 c 0.32 ± 0.04 Ay 0.69 ± 0.09 [N II]A6584+6548/5755 93.7 + 1.0 Te 9540 K [S II]A6717/6731 1.01 + 0.01 Ne 550 cm"^ 69

Ha [N ll]X6583

10 " 10 "

[S ll]À6717 [S ll]X6731

10 " 10"

Figure 9: NGC 6720: Emission-line flux maps from the data cubes. Ha, [N II]A6583, [S II]A6717, [S II]A6731. Images are 120" across, with north at the top and east to the left. Flux is zero where the data cube was masked and no spectra were fit. 70

HellU686 [O lll]X5007

10" 10"

HP [OIJX6300

10" 10 "

Figure 10: NGC 6720: Emission-line flux maps from direct imaging. HeIIA4686, [0 III]A5007, H/?, [0 I]A6300, Images are 120" across, oriented with north at the top and east to the left. The Hell and [0 I] images have been corrected for nebular continuum and starlight. 71

1 0 0

in ts' < ±a

50

g (D g

< '55'^ -5 0 ■'i.'

-100

100 50 0 -50 -100 Aa (arcsecs)

Figure 11: NGC 6720: [0 I]A6300 émission. This image was obtained September 1993. Contrast has. been adjusted to clearly show the inner and outer halo and edge brightening. 72

1 I I I I I I I r 1 I I I I I I I r _ Ho/HB Balmer Decrement

40

20

8 0 g & < -2 0

- 4 0

L I I I I -I I I I I I - 1 I I I j — I.. I I 40 20 0 -20 -4 0 Aa (arcsecs)

3 3.2 3.4 3.6 3.8 4 Ha/HP Figure 12: NGC 6720: Balmer decrement 73

1 I I I I I I I I I I I I I I I I I I I I I r Ü Electron Density in Region

40

20

Q) 2 a < -20

-40

L I___L J L I I I I I I I I I I I I I I I 40 20 0 -20 -40 Aa (arcsecs)

200 300 400 500 600 700 K (cm" ) Figure 13: NGC 6720: Electron density in the S"*" region. Density was calculated using the [S II]A6716/6731 line ratio for an assumed constant temperature of 10^ K. 74

|~n I I I I I I I I I I I I I I I I I I I I I r _ Electron Temperature in Region

40

20

(D g

< -20

-4 0

L I I I I I I I I I I I ■ I I I... I.. I 40 20 0 -20 -40 Aa (arcsecs)

8500 9000 9500 10000 10500 11000 T,(K) Figure 14: NGC 6720: Electron temperature in the N""" region. Temperature was calculated using the [N II]A6548+6583/5755 line ratio following Osterbrock (1989), assuming a constant electron density of 500 cm“^. 75

_ HellÀ4686/Hp

*r% I

-20

20 0 -2 0 -4 0 Aa (arcsecs)

Hell 4686/Hp Figure 15: NGC 6720: Ionization structure ratio maps - HeIIA4686/H/3. Images are 120" across. Line ratios formed from the data cube fits have been masked, showing zeros where no fits were made. Ratios formed entirely from direct imaging show mostly noise outside the visible shell of the nebula. 76

_ [O l]À6300/[0 lll]À5007

8 0 g a <

-4 0

20 0 -20 -40 Aa (arcsecs)

.2 .3 .4 [O l ] / [ 0 III] Figure 16: NGC 6720: Ionization structure ratio maps - [0 I]A6300/[0 III]A5007 77

_ [O lll]À5007/Hp

8 0) i 5 . < -20

-4 0

40 20 0 -20 -40 Aa (arcsecs)

0 5 10 [O 1II]/H P Figure 17: NGC 6720: Ionization structure ratio maps - [0 III]A5007/H;S 78

I I I I I . [O l]À6300/HB

8 m g a <

20 0 -20 Aa (arcsecs)

0 .5 1 [O 1]/H p Figure 18: NGC 6720: Ionization structure ratio maps - [0 I]A6300/H/3 79

1 I I I I I I I I I I I I I I I I I I I I I r . [N ll]X6584/Ha

40

20

8 0 g s < -2 0

-4 0

r . I L - i I I I I I I I I I I I I I I I I I I I I I I 40 20 0 -20 -40 Aa (arcsecs)

0 1 2 3 4 [N ll]/H a Figure 19: NGC 6720: Ionization structure ratio maps - [N II]A6583/Ha 80

T I I I I I I I I I I I I I I I I I I I I I r ■ [S ll]^6717+6731/Hp

40

20

8 CD g

< -20

-4 0

l~ I I I I I I I I I I I I I I I I I I I I I I I 40 20 0 -20 -40 Aa (arcsecs)

[S ll]/Hp Figure 20: NGC 6720: Ionization structure ratio maps - [S II]A6717+6731/H;5. Note that the mask for the [S II] data cube is smaller than the mask for the Ha data cube. The ionization edge is at the same position in both the [N II]/Ha and [S II]/Hj5 ratio maps. CH APTER IV

NGC 7662 and NGC 7009

4.1 Introduction

Our understanding of the evolution and structure of PNs has advanced quickly in the past 10 years. A unified explanation for the observed wide range in morphologies was proposed by Balick (1987), based on the interacting stellar winds theory first applied to PNs by Kwok et al. (1978). Before Bahck’s paper, there was no encompassing explanation of how PNs were shaped, and no theory that could predict the wide range in observed morphologies. ISM environment, chemical abundances, progenitor

mass, and many other factors were used to classify PNs to explain their appearances,

but these schemes tended to leave out large classes of PNs. Balick asserted that

hydrodynamics is the principal agent that shapes PNs, and that the critical parameter

is the equator/pole density contrast.

Based on this foundation, researchers V. Icke, G. Mellema, and A. Frank (see

Icke et. al. 1992; Mellema 1993; Frank & Mellema 1994 and references therein) have

made sophisticated radiative-hydrodynamical models of PN evolution. The models

predict the run of temperature, density, and kinematics throughout the nebula as a

function of time, and the appearance of such physical characteristics as ionization

81 8 2 and shock fronts. The result is a large grid of models that encompass a wide range of equator/pole density contrasts, central star masses, and mass-loss rates and histo­ ries. The set of model morphologies produced has a close correspondence to the set of observed PN morphologies, and the predicted emission-line flux profiles and the kinematics match some observations of PNs. This is an exciting result, and shows that no important physical processes have been neglected in the modeling and that the input AGB mass-loss parameters are not inconsistent.

Not only are these models exciting from the standpoint of understanding PNs, but they may also provide the best tool for understanding the AGB mass-loss history.

It has been shown that the mass-loss history cannot be derived from observations of PNs because of the overwhelming hydrodynamic effects of the interacting stellar winds (Frank et al. 1990). A detailed comparison (not just using morphology) of these models with observations of PNs will hopefully result in a refinement of the

AGB mass-loss history input into the models.

Both NGC 7662 and NGC 7009 are considered to be good examples of how well current hydrodynamic evolution models can predict the morphology and emission-line structure of PNs. NGC 7662 and NGC 7009 are bright, well-studied planetary nebulae

and share many similarities. They are high-excitation PNs, with strong emission in high ionization lines and overall little emission from low ionization species. They both

have a bright-rim/attached-shell morphology (using the nomenclature of Frank et al.

1990, see Chapter 1, Figure 3 for a picture) and are classified as “middle elliptical” in

Balick’s (1987) morphology-evolution classification (Balick’s morphological sequence 83 is reproduced in Chapter 1, Figure 2). Observations place both of their central stars in a similar position on the Harmon-Seaton sequence (Gathier & Pottasch 1989).

My observations of NGC 7662 and NGC 7009 provide an opportunity to rigor­ ously test the current hydrodynamic evolution models because density, temperature and ionization state can be compared in detail. The models that are most useful for comparison with observations are those of Frank and Mellema (Frank 1994, Mellema

1993, Frank & Mellema 1994b). Although neither NGC 7662 nor NGC 7009 is a spherical nebula, the spherical-geometry models of Frank (1994) are nonetheless very useful in that they give exact numbers for density and temperature. When compar­ ing my observations to the models, small-scale structure (such as condensations and

FLIERS) will be ignored, and only the large-scale structure will be considered. In particular, I will first try to determine the density and temperature as a function of radius in the context of the spherical geometry case. Asymmetrical structure will be examined later and compared to the predictions of Mellema (1993).

4.1.1 Previous Studies of NGC 7662 and NGC 7009

NGC 7662, because it is bright and of reasonable angular diameter (about 30"), is a well studied PN. The most comprehensive spectral-line study in the UV and optical is by Barker (1986) using aperture spectroscopy. Two studies exploring morphology are Frank et al. (1990), which observes shell emission profiles, and Balick, Preston,

&: Icke (1987), which observes kinematics using a stepped high-resolution longslit spectrometer. Harrington et al. (1982) did the definitive photoionization modeling work on this PN. 84

NGC 7009 is also a well studied nebula. Again, there is a spectroscopic study by

Barker (1983), and Cyzak & Aller (1979) compared the spectroscopic properties of the ansae and the rim. Recently, Hyung & Aller (1995) presented extensive spectroscopy of NGC 7009 from regions along the minor axis (the high-excitation rim) and the major axis (the low excitation caps). They also attempted photoionization modeling based on their spectra. A narrowband filter imaging study of NGC 7009 similar to this study has been presented by Bohigas et al. (1995).

In addition to their similar large-scale structure, both NGC 7662 and NGC 7009 show significant small-scale structure (e.g. condensations and ansae), and both have

FLIERS (Fast Low Ionization Emission Regions, Balick 1993). Although the FLIERS appear similar to the condensations, the condensations are cool and dense, while the

FLIERS have the same temperature and density as the rest of the nebula. FLIERS also have anomalously large velocities relative to the expansion of the nebula. Balick et al. (1993, 1994) has studied the FLIERS in NGC 7662 and NGC 7009 using longslit spectroscopy.

4.2 Observations, Data Reduction and Analysis

The observations of NGC 7662 and NGC 7009 are listed in Tables 2 and 4 (Chapter 2), and the methods of data reduction and analysis are also described in Chapter 2. There were a few additional steps taken in the method used to analyze the data cubes of

NGC 7662 and NGC 7009 that were not needed for the Ring Nebula data cubes.

These differences are because, unlike the Ring Nebula, NGC 7662 and NGC 7009 are high excitation PNs. The [N II] and [S II] emission lines are relatively weak 85 in NGC 7662 and NGC 7009, leading to three complications: (1) difficulty fitting the relatively weak [N II]A6584Â line in the presence of strong Ha in the Ha data cube, (2) order overlap in the [S II] data cube in which the HeIA6678Â emission line from the adjacent order of interference contaminates the red continuum channels of the [S II] spectra, and (3) problems obtaining an accurate flux calibration for faint low-ionization emission lines. The first two items are discussed in Section 2.2.3.

The strong emission lines of Ha, H/3, [0 III]A5007Â, and HeIIA4686Â are all ac­ curately flux calibrated using the method described in Chapter 2, but there are larger uncertainties in the flux calibration of the [N II]AA6584,5755Â, [S II]AA6717,6731Â and [0 I]A6300Â lines (and in some case no reasonably accurate calibration was pos­ sible) because these lines are so faint (e.g. [0 I]A6300/H^ ~ 0.004). All lines in the data cube have an accurate internal flux calibration. Thus the Ha flux calibration

(which has small uncertainties) can be used for the [N H]A6584Â emission line, and the [S II]AA6717,6731Â emission lines yield an accurate [S II] ratios for the purposes of deriving the electron density maps.

[N II] emission is weak in both PNs, and so it is not possible to measure the temperature in all parts of the nebula using the [NII] lines. Unfortunately, the regions of strong [N II] emission are in the dense condensations in these two nebulae. At the observed densities (Ne > 6000 cm“^, see Section 4.3), [N II]A6584Â is collisionally de­ excited, so that even an accurate measurement of the [N II]6584/5755 line ratio results in an uncertain temperature. The temperature maps are therefore not presented, as they contain little useful information. 86

4.3 Results

4.3.1 NGC 7662

Flux Maps

Emission-line flux maps of H/?, [0 III]A5007Â, HeIIA4686Â, and [ 0 IJA6300Â obtained from direct imaging axe presented in Figure 21. Emission-line flux maps of Ha,

[N II]A6584Â, and [S II]AA6717,6731Â derived from the data cubes are presented in Figure 22. These latter images have been masked (see Section 2.2.3), hence the

“blank” sky regions. The mask for the Ha data cube extends partway into the halo while the mask for the [S II] data cube extends only just beyond the edge of the shell.

All images except [0 III]A5007Â have been continuum subtracted.

The change in the nebular morphology that occurs when progressing from high ionization potential emission lines (e.g. Hell) to low ionization lines (e.g. [0 I]) is seen in Figure 21. Hell radiates almost entirely in the rim and core. In Hell, the rim is an unbroken ring with two bright spots corresponding to the broken parts of the rim seen in H/3 and [0 III]. In H^ and [0 III] the shell emission is much more prominent relative to the rim. In [0 I], the NE/SW parts of the rim are only barely distinguishable, and microstructures — knots and FLIERS — dominate the emission-line images.

The knots surround the outer edge of the shell except for a noticeable void in the NW quadrant. The north and south low-ionization structures have been shown to be FLIERS (Balick et al. 1993). The SE knot has significant H^d and [0 III] emission compared to the other knots. 87

These images point out a surprising asymmetry: the core is offset from the central star and the shell by about 1.5" W. Note that the uncertainty on the central star position (~ ±0.5") is much less that this difference.

Integrated line intensities with respect to HjS are presented in Table 8. The intensities have been corrected for extinction using c(ffj3) = 0.16 (Ay = 0.32) as

derived from the integrated Ha/H/3 ratio. The errors in the relative total intensities

are almost entirely due to uncertainties in the flux calibration. Only the lines with a good flux calibration (better than ±10%) are given in Table 8.

Balmer Decrement Ha/Hy5

The Ha/Hy5 line ratio map (uncorrected for extinction) is shown in Fig. 23. The

mean ratio is Ha/H/3=3.2±0.2 in the shell plus the rim of the nebula, indicating

very little interstellar extinction. Overall, our measured ratio is somewhat lower than

previous measurements by Barker (1986), 3.6 compared with 3.2, but consistent with

the measurements of Harrington et al. (1982). The value of Ha/H/3=3.2 leads to a

reddening parameter c{H^) = 0.16 ± 0.04 (Ay = 0.32) assuming an intrinsic value

for the ratio Ho:/H^=2.86 (Case B, 3^=10,000 K, iVe=10^ cm“®. Hummer & Storey

1987).

There is little variation in Ha/H^d in the shell and the rim. The high density

rim and condensations appear to have a slightly lower Balmer decrement (~7%), as

would be expected from collisional de-excitation of Ha at high densities (Hummer &

Storey 1987). 88

Although the nebulosity is considerably fainter in the halo, we can average over a large number of halo pixels and arrive at a reasonable estimate of Ha/H^S in the halo.

The mean in the halo is 4.4±0.2, significantly higher than in the shell. This applies to the inner part of the halo (our observations, constrained by the Ha observations, extend to a radius of 40", but the halo extends to ~72" Chu et ai. 1987). The most likely explanation for this is dust mixed into the halo gas.

Electron Density

The electron density derived from the [S II]6717/6731 line ratio is shown in Figure 24.

Uncertainties on the [S II] fluxes are about ±5%, the largest contribution coming from the continuum measurement because the HeIA6678Â line reduces the number of continuum pixels in the spectra. The problem is the worst in the bright rim region, where HeIA6678Â is strong and the [S II] lines have relatively low equivalent widths.

However, even in the worst case, the RMS uncertainties in the line flux fits are no more than ±10%. All the high density regions are at densities near the high density limit of the [S II] diagnostic. In the rim the A6717/6731Â line ratios were as low as 6717/6731=0.36 which is below the high density limit of 6717/6731=0.44. These points have been masked in Figure 24.

The high-density regions are distinct from the rest of the nebula. The shell

(without the knots) is uniformly at Ne ~ 4000 cm“^. The high-density regions are all

at Ne > 6000 cm“^. The high-density regions fit theoretical predictions. The [0 I]-

bright knots around the outer edge of the shell (excepting the FLIERS) are dense 89 condensations, which is predicted by photoionization models (Capriotti 1973). A dense rim is predicted by radiative-hydrodynamic evolution models (Frank 1994).

Ionization State

NGC 7662 is a high excitation PN. Emission from the core, rim, and shell are pri­ marily, of the lines observed, in Hell, [0 III] and Ha. Low ionization line emission is confined to the knots/condensations at the outer edge of the shell. In general, the smaller the radius, the higher the excitation state of the gas. The ratio maps are shown in Figures 25-29.

The innermost ionization front is the transition from 0 ’*’^ to at a radius of 8" in the [0 III]/H/3 image (Figure 25). The innermost 8" is probably entirely 0+^

(or higher); 0 IV]AA1403,1409Â emission is observed from the core by Barker (1986).

The faint [0 III] emission seen interior to 8" is likely projected [0 III] emission from the shell. No to O"*" ionization front is seen in these ratio maps. [0 III]/H/3 is constant in the shell (19.3±0.1) and drops only slightly in the halo. [0 I]A6300Â is observed only in the condensations and FLIERS, not as an ionization front in contrast to what is seen in NGC 6720 (Chapter 3). [N II]A6584Â, another low ionization line, shows the same behavior as [0 I].

The Helium ionization front is circular with a radius of 9". Photoionization models of Harrington et al. (1982) predict the He++ and 0"^^ ionization fronts to be at about the same radius. There does not appear to be a hydrogen ionization front, as no [0 I]A6300 (a good indicator of partially ionized hydrogen) or molecular hydrogen 90

(Geballe et al. 1991) is observed around the edge of the nebula.

4.3.2 NGC 7009

Flux Maps

The emission-line flux maps of HeIIA4686Â, [0 III]A5007Â, [ 0 1]A6300Â-h[S III]A6312Â and H/S obtained from the direct imaging are shown in Figure 30. The emission-line flux maps of Ha, [N IIJA6584Â, and [S II]A6717,6731Â derived from the data cubes are shown in Figure 31. There are a number of important morphological features seen in these images. The rim is elliptical and bright in Hell and [0 HI] emission (best seen in the [0 III] image). Interior to the rim is the core, and exterior to the rim is the shell, out to the elliptical outer edge of the nebula. The low-ionization structures are point-symmetric (best seen in the [0 I] image). The caps are the inner pair and the ansae are the outer pair (outside the shell).

The point-symmetric structures seen in NGC 7009 are not aligned. The sym­ metry axes of the caps and the ansae are rotated 13.5° with respect to each other (the rotation is counterclockwise in the North up. East left orientation of Figure 30). This angle is measured from the points of peak [0 I] emission. Reay & Atherton (1985) proposed a geometrical model to explain this. A line from the central star to the soli­

tary northern [0 I]-bright knot is almost exactly perpendicular to the line connecting

the outer ansae. The points of peak Hell emission are not on the minor axis of the

rim. The major axis of the rim is aligned with the outer ansae. In the Hell image,

the line connecting the points of peak Hell emission is rotated 60° counterclockwise 91 from the line connecting the outer ansae. Hell emission forms a bar across the central star along the line connecting the peaks.

Integrated line intensities with respect to H/? are presented in Table 9. The intensities have been corrected for extinction using c{H^) = 0.24 as derived from the integrated Hq/H^ ratio of 3.5.

Balmer Decrement Ha/H/3

The Ha/H^ image is presented in Figure 32. Rather than using the continuum- subtracted image from 1992 November, the H/3 image from the same run as the

Ha data cube (1992 September) was used instead because the lack of field stars in the images makes registration, particularly removing rotation, difficult. The effect of the nebular continuum was considered less of a problem than rotation between two frames as the 1992 November H/? and continuum images show that the nebular continuum at H^ is only 2% of H/3. The final registration process to determine linear shift first used the positions of field stars in the image, then used the sharp nebular structure to refine registration. This method was successful for the Ring Nebula, and is particularly effective when the field stars used for registration are smeared due to poor guiding or faintness. The registration was checked carefully as the rim structure is similar to a registration-error artifact, but the structure appears to be real.

Only slight structure is seen in the Ha/H^d ratio map. The Balmer decrement is

3.5±0.2 in the shell. As in the NGC 7662 Ha/Hyd ratio map, the small variations seen are likely due to temperature and density variations rather than dust or interstellar 92 reddening. The interstellar extinction to this nebula is very low: c{H^) = 0.24 ±

0.04 [Ay = 0.51), using the previous intrinsic Ha/Hy5 ratio. Distance estimates for

NGC 7009 range from 0.6 to 2.4 kpc (Terzian 1993).

Electron Density

The electron density is shown in Figure 33. As in NGC 7662, the electron density in the high-density regions of NGC 7009 is close to the high-density limit of the [S II] diagnostic.

Electron density in the central region is >8000 cm“^. This high-density region is elliptical in shape with a major axis rotated clockwise 50° from the major axis of the rim. The elliptical [0 Illj-bright rim is not clearly a high-density structure, as was the case in NGC 7662. The shell, not including the caps, is at significantly lower density (Ng ~4000 cm“^).

The ansae and caps are not high density structures. The electron density is only slightly higher in the caps than in the rest of the shell, and slightly lower in the ansae than in the shell. This is typical for FLIERs (Balick et al. 1994).

Ionization State

The ratio maps are shown in Figures 34-36. In the [0 III]A5007/H^ ratio map

(Figure 35), the edge of the rim is marked by an increase in the ratio. The [0 III]/H/? ratio does not significantly drop interior to the rim. In the HeIIA4686/H/3 ratio map

(Figure 34), the ratio drops sharply at a radius of 7", suggesting a Hell ionization front. The ratio is enhanced along an S-shaped curve tying the core and caps together. 93

The low-ionization emission-Iine/Balmer-line ratios (e.g. [N II]/Ha, [S Ilj/H^S,

[0 I]/H/3) all look similar because the ansae and caps are very strong in low-ionization line emission (Figure 36). The [N II]A6584/Ha ratio map shows several features. In each of the ansae, there is an outward gradient in [N II]/Ha in which it increases to

[N II]/Ha ~ 2 over ~4", and drops sharply back to zero over ^2". Similar behavior in this ratio was reported by Bohigas (1994). In the east cap, there are three points where the ratio [N II]/Ha >0.3.

4.4 Discussion

I have compared my observations and those from other studies of these two nebulae to the radiative-hydrodynamic evolution models of A. Frank and G. Mellema (Frank

1992; Mellema 1993; Frank & Mellema 1994a; Frank 1994; Mellema 1994; Frank

& Mellema 1994b; Mellema & Frank 1995; Mellema 1995). The intention is to go beyond a comparison that only considers morphology, to a more detailed comparison including density and ionization structure (especially ionization fronts). For each

PN, the process of finding a model that best matches the observations has several stages. First, I will compare the observed morphology of the PN to the morphologies in the models to narrow down to the class of models reflecting approximate age and density contrast. A similar, but coarser, classification was done by Balick (1987) for a large number of PN (see Figure 2). Frank (1992, 1994) divides his spherical models into four phases of evolution: (1) Pre-Ionization, (2) Hydrogen Ionization, (3)

Helium Ionization, and (4) Shock Break-out, and lists defining characteristics of each phase. Next, I will compare the observations of key parameters with the subset of 94 models to find a best matching model (or two). The available parameters are two- dimensional density, two-dimensional ionization fronts, temperature at some points, some kinematics, and the approximate age of the central star. The final goal is to use the discrepancies between the best matching model(s) and the observations to refine to the model’s input density distribution, which reflects the AGB mass-loss history.

NGC 7662 and NGC 7009 are classified as early/middle ellipticals by Balick

(1987), and have a distinctive bright-rim/attached-shell morphology (seen in [0 III] or Ha emission lines). This morphology appears in the “young” models of Frank and

Mellema (Frank 1994; Mellema 1993). The rate at which a PN central star travels along its evolutionary track depends sensitively on its mass. Central star evolution models predict that a 0.64 M q central star will take 4200 years to reach its maximum temperature of 200,000 K (Schonberner 1981), while a 0.546 M q star will take 300,000 years to reach its maximum temperature of 100,000 K (Schonberner 1983). Thus, it is more useful to specify age by a central star’s position on the Harmon-Seaton evolution tracks, rather than in years. In the “young” models of Frank and Mellema referred to above, the central star is still on the horizontal part of its evolution track, i.e. it is still increasing in temperature.

The locations of the Helium and Hydrogen ionization fronts in both PNs also indicate an early period of evolution. The position of the He++ to He"*" ionization front is at a radius of 9" for NGC 7662, just inside the rim (0.52 pc assuming a distance of 1.2 kpc, Terzian 1993). This ionization front is at a radius of 7" for NGC 7009. In

Frank (1993), the He"*"*" ionization front is at the same distance as the rim only for 95 young stages (Hydrogen Ionization stage). No Hydrogen ionization front is seen at the outer edge of the shell in either NGC 7662 or NGC 7009.

One kinematic study of NGC 7662 is relevant here. Chu et al. (1987) observe

the bright rim expanding into the surrounding envelope (i.e. Vrim > ^envelope)- This

kinematic situation only occurs in models of young nebulae in Frank (1992; 1994)

and Frank & Mellema (1994).

The central star of NGC 7662 has temperature T*=91,200 K and luminosity

Zr* = 10 ^‘^^Zq (Gathier & Pottasch 1989). Using the Schonberner central star evo­

lution tracks presented in Gathier & Pottasch (1989), the central star falls on the

M* = 0.546Mq evolutionary track at an age of ~250,000 years (on the knee of the

evolutionary track). The central star of NGC 7009 has a temperature T*=70,800 K

and luminosity T* = 10 ^’®^X q (Gathier & Pottasch 1989), which places it on the

M*. = 0.546Mq evolutionary track at an age of ~350,000 years (past the knee in the

evolutionary track). The central stars of NGC 7662 and NGC 7009 thus appear to

be low-mass stars that have reached or passed their maximum temperature epochs.

However, the other observations (morphology, kinematics, and ionization fronts)

indicate that the best matching models are those in which the central stars have not

yet reached their maximum temperature. Although the uncertainties in the central

star temperatures and luminosities are large (20-25%, Gathier k Pottasch 1989), they

are nonetheless not large enough that the central stars of these two PNs could be as

young as the central stars in the matching models. Furthermore, uncertainties in the

assumed central star evolutionary tracks are irrelevant in this case, as the central 96 star observations are placed on the same evolution tracks that were input to the hydrodynamic models (i.e. the models of Schonberner 1981). Since the masses of the central stars of NGC 7662 and NGC 7009 are not in the set of masses of the models run by Frank and Mellema, these observations should not be compared to those models.

Why, then, do the models appear to match the observations of these two PNs

so well? There are two significant effects a lower-mass central star could have on the

surrounding PN and its evolution.

In Frank’s (1994) spherical models, the position of the He"*"*" ionization front is a

good indicator of its evolutionary state, because it tracks the increase in temperature

of the central star. In the model with a 0.644 M q central star, a sharp Helium

ionization front moves outward as the central star evolves until the ionization front

reaches the outer edge of the nebula. However, a low-mass central star will never get

hot enough to entirely ionize Helium throughout the PN. The Helium ionization front

will go no further than the Stromgren radius and not propagate through the nebula.

This would then mimic a younger stage in the evolution in the models and possibly

contribute to the misidentification.

The other important effect is that a low-mass central star evolves more slowly.

A comparison of radiative and non-radiative hydrodynamic models shows how impor­

tant the central star evolution and the related ionization fronts are to the structure

of the PN. The ionization fronts will propagate through the surrounding PN later

than the interacting stellar wind shocks (relative to a higher-mass central star PN), 97 and thus the nebula will be more dispersed when these ionization fronts go through.

Although Frank (1992) discusses the effect of time scales in the modeling, it would be difficult to estimate the actual values of the density, temperature, and velocity by using simple scaling arguments. As the final goal is to use the differences in these parameters between the models and the observations to refine the initial mass distri­ bution, scaled models will not be sufficient, and so radiative-hydrodynamic models need to be run for the low-mass central star case.

Since the morphologies, ionization fronts, and density structures of NGC 7662 and NGC 7009 match those of Frank’s and Mellema’s models so well, one has to wonder if there is a uniqueness problem. The successful production of the dense- rim/attached-shell morphology was a significant result of the model of both Frank

(1994) and Mellema (1994). Is it a coincidence that a young PN with a high-mass central star has the same morphology as an older PN with a low-mass central star?

Or is this morphology generic, occuring in many situations or for a long time, and thus less useful in determining the evolution state of a PN than one might have been led to expect? Again, models using a low-mass central star should be able to address this point.

Although there are still many PNs with central stars in the right mass range that matched the models of Frank and Mellema, it is perhaps premature to say that their models are matching real PNs. Strong testing of the models against observations needs to include aU relevant nebular parameters: morphology, density, temperature, velocity, and the properties of the the central star, namely its age and mass. 98

4.5 Summary

A range of diagnostic optical emission lines in the planetary nebulae NGC 7662 and

NGC 7009 were observed to determine the electron density and ionization state across the nebulae. These observations, combined with previous studies, are compared to the current radiative hydrodynamic evolution models of PNs.

The [S II] density maps for NGC 7662 and NGC 7009 are the most illuminat­ ing, indicating that the bright-rim structure is dense in both PNs. The low-ionization knots (except for the previously-identified FLIERS) seen in both PNs are dense con­ densations, as predicted by Capriotti (1973).

The structure of NGC 7662 and NGC 7009 (dense-rim/attached shell morphol­ ogy, locations of He''"’' and H'*' ionization fronts) matches the predictions of some recent radiative-hydrodynamic evolution models. However, the models that best match the observations are younger than the ages one would infer for NGC 7662 and NGC 7009 based on their central stars. The likely cause of this discrepancy is that the central stars of NGC 7662 and NGC 7009 have a lower mass than the central stars used in the models. A lower-mass central star significantly affects the evolution of the PN because the star evolves much slower, and has a different spectral evolution.

Simply scaling the models will not sufficiently deal with this, so there is a need for

models with lower central star masses. 99

Table 8: NGC 7662: Integrated® Line Strengths

Identification A F/F(H^) 4861 1.00 1.00 He II 4686 0.45 0.46 [0 III] 5007 12.71 12.51 Ha 6563 3.253 2.86 [Nil] 6583 0 .055 0.049

“ 40" diameter circular aperture ^corrected for extinction using c=0.16 T(H)8)=7.32xlO-" erg s'^ cm“2

Table 9: NGC 7009: Integrated® Line Strengths

Identification A F/F(H^) i / i m ^ H^= 4861 1.00 1.00 He II 4686 0.15 0.15 [0 III] 5007 11.97 11.68 Ha 6563 3.48 2.86 [N II] 6583 0.13 0.11

“ 40" diameter circular aperture ^corrected for extinction using c=0.24 ‘^j0)=1.10xl0"^^ erg s“^ cm~^ 100

Heia4686 [Ol]^6300

10 " 10 " H(3 [0 lll]15007

10 " 10 "

Figure 21: NGC 7662: [0 I]A6300Â, HeIIA4686Â, and [0 III]A5007Â emis­ sion-line flux maps. Images are 60" across, oriented with north at the top and east to the left. 101

[S ll]X6717 [S ll]X6731

10" 10"

Ha [N ll]^6583

10" 10 "

Figure 22: NGC 7662: [S II]A6717Â, [S II]A6731Â, Ha, and [N II]A6584Â. Images are 60" across, with north at the top and east to the left. Flux is zero where the data cube was masked and no spectra were fit. 102

NGC 7662 Ho/Hp

h i I ! I I I I t I I I I I 1 I I I I I I I I I I I I I I I

20

10

5 (D i 6 <

-10

-20

' l I I I I I I I I I I I I I 1 I I I I I I I I I I I I I I 1 20 10 0 -10 -20 -30 Aa (arcsecs)

* 4 H o / H p

Figure 23: NGC 7662: Balmer Decrement 103

NGC 7662: Electron Density in S Region

-I I T I I I I FT I I I I I I I I I r~n r I i r i i i i 1

20

10

I 0 & <

-10

-20

~i I I I I > I I I 1 I I I I 1 1 I I I 1 I I I I I I I I I 1 20 10 0 -10 -20 -30 Aa (arcsecs)

2000 4000 6000 8000 10000 NJcm-")

Figure 24; NGC 7662: Electron density was calculated using the [S II]A6716/6731 line ratio for an assumed constant temperature of 10“* K. North is up, East is left. The masked points in the southeast rim had a [S II]A6717/6731 ratio lower than the high-density limit. 104

NGC 7662 [O lll]À5007/Hp

40

20 ' m uI#/';'- .r - .» • .

8 (D • S-2 <

■ l v S 4 -20 i& **'K ^

-40 m40 20 Aa (arcsecs)

10 15 20 [O lll]/Hp

Figure 25: NGC 7662: [0 III]A5007/H^ 105

NGC 7662 Helll4686/Hp

40 20 0 -20 -40 Aa (arcsecs)

Hell/HB

Figure 26: NGC 7662: HeIIA4686/H^ 106

NGC 7662 [O III]À5007/Helll4686

, 'Vij „.s«» ..*. jnl" ■.. S 0} g - : k #:L . <

/.:v j y f : . V . ÿ :u. zT "*%i

-20 -40 Aa (arcsecs)

100 200 300 [0 Ill]/Hell

Figure 27: NGC 7662: [0 III]A5007/HeIIA4686 o

m §

m 8 %L O § cû. i Si CO. X ' i o 0 ü 1 % I % i R ü O

(soasojB) gv 108

NGC 7662 [N ll]^6584/Ha

- I I I r i I I I I I I I I "I I I I I I I I I I I I I I I I

20

10 s 0 2 s . <

-10

-20 k m .

I I I I I I I I I I I I I I I - I I I I I I I I I I I I I I 1 20 10 0 -10 —20 —30 Aa (arcsecs)

.1 .2 [ N l l ] / H a

Figure 29: NGC 7662: [N II]A6584/Ha 109

HellM 686 [O lll]X5007

10" 10 "

HP [O l]>.6300

10" 10 "

Figure 30: NGC 7009: H^, [0 I]A6300Â, HeIIA4686Â, and [0 III]A5007Â emis­ sion-line flux maps. Images are 75" across, oriented with north at the top and east to the left. 110

[S II]16717 [S II]16731

[N ll]À6583

Figure 31: NGC 7009: [S II]A6717Â, [S II]A6731Â, Ha, and [N II]A6584Â. Images are 75" across, with north at the top and east to the left. Flux is zero where the data cube was masked and no spectra were fit. I ll

NGC 7009 Hot/Hp

r-i .t . ,ifci |.;i r-T-ç|ï-i i i i | * i i- i&U 1.1, ihm

8 0) g

<

l - ' l I I» I I 1 * 1 II I I I I I T I I . I M ' I 1 . 1 I I 111 I T ' . i t i ' A k 30 20 10 0 -10 -20 -30 Aa (arcsecs)

Figure 32: NGC 7009: Baimer Decrement 112

NGC 7009 Electron Density in Region

8 0) I i & <

-10 — % w

30 20 10 0 -10 -20 -30 Aa (arcsecs)

4000 6000 10000 Ne (cm-®)

Figure 33: NGC 7009: Electron density was calculated using the [S II]A6716/6731 line ratio for an assumed constant temperature of 11000 K. North is up, East is left. The masked points in the central region are where the [S II] spectrum could not be fit. The contour overlay is of [S II]A6717 emission-line flux (evenly spaced levels). o eo T- I «O

w î 1 î ô> cû. 4 o g ü O 0 % 1 Ü 0) X o O) CM 1 g 0

(SO0SOJB) gV 114

NGC 7009 [O lll]15007/Hp

S 0} g s <

30 20 10 0 -10 —20 —30 Aa (arcsecs) mmÆ

[O lll]/H p

Figure 35: NGC 7009: [0 III]A5007/H^ 115

NGC 7009 [N 11]À6584/Ha

8 (D g

< #

Aa (arcsecs)

[N ll]/H a

Figure 36: NGC 7009: [N II]A6584/Ha CH APTER V

Conclusion

5.1 Summary

I have presented two-dimensional spectrophotometry of three planetary nebulae:

NGC 6720, NGC 7662, and NGC 7009. Imaging Fabry-Perot interferometry, narrow­ band filter imaging, and longslit spectrophotometry are combined in order to take advantage of the strengths of each technique. The final data set has good spatial cov­ erage (5'x5') and angular resolution (2"), good spectral resolution on closely spaced emission lines (4Â), and accurate spectrophotometry (±5%). There was a practical limitation to the total number of lines observed, so good diagnostics were chosen: the

[S II]6717/6731 density diagnostic, the [N II]6584/5755 temperature diagnostic, the

Ha/H/? Baimer decrement, and a range of both high ionization potential emission lines (e.g. HeIIA4686Â, [0 III]A5007Â) and low ionization potential emission lines

(e.g. [0 I]A6300Â) . The scientific objective was to study the structure and evolution of PNs.

The initial target was NGC 6720 (the Ring Nebula). There are three principal results. (1) The Ring Nebula is ionization-bounded and there is a clumpy ionization front around the entire outer edge of the nebula. (2) The prominent, [0 I]-bright

116 117 filaments were expected to be cool and dense, but they are not — no differences

were seen between the temperature and density in the filaments and the surrounding

nebula. (3) The “ring” of the nebula is intrinsically oval-shaped, rather than being

a tilted circular ring. Molecular hydrogen observations led to the suspicion that

the Ring Nebula was ionization bounded, but the ionization front has never been

demonstrated so clearly as in this work. The results for the filaments were very

surprising since theoretical modeling of [0 I]-bright microstructures had led us to

expect that the filaments would be cool and dense structures as seen in other PNs. The

conspicuous filaments are not due to changes in density, but rather are changes in the

ionization state of the gas. The three-dimensional geometry of the Ring Nebula has

always been a puzzle despite the many observations of the Ring Nebula. My density

map, which shows density enhancements along the minor axis, can be interpreted

along the lines of the Interacting Stellar Winds view of PN evolution to finally pin

down the geometry of the PN, in this case a tilted oval seen in projection.

The other two PNs observed, NGC 7662 and NGC 7009, are quite different from

the Ring Nebula in that they are high-excitation PNs. NGC 7662 and NGC 7009

are similar in a number of ways, and thus are considered together. One gratifying

result was that in these two PNs, the low-ionization microstructures do follow the pre­

dictions: the [0 I]-bright knots are dense structures (except for the FLIERS, which

was expected). The density maps also showed that the observed bright-rim structure

is dense, an observation that is in line with predictions of radiative-hydrodynamical

models. In fact, the morphology, density structure, and ionization fronts in NGC 7662 118 and NGC 7009 all fit the predictions of the models. However, the central star observations of NGC 7662 and NGC 7009 indicate that these PNs are significantly older than the models they match most closely. It turns out that the central stars of these two PNs have lower masses than the central stars assumed by the models. This contradiction is not easily resolved, and it may be premature to say that the models reproduce real PNs.

5.2 Future Work

A further step in this project would be to extend the list of observed emission lines.

With the addition of the UV optics and the blue-sensitive detector to the IFPS, some useful emission lines will be observable. The most important is the [0 III]A4363Â emission line, which is used in the [0 III]A5007-|-4959/4363 temperature diagnostic.

This temperature diagnostic is more appropriate for high-excitation PNs, such as

NGC 7662 and NGC 7009, for which the [N II] temperature diagnostic produces no useful information. The [0 II]A3727Â and H 7 emission lines would complement the set of lines already observed by completing the oxygen ionization set and the Baimer set. Lastly, the [Ne III]A3869Â and [Ne V]A3426Â emission lines would provide a look at the high-ionization inner regions of PNs. Beyond the optical capabilities of the IFPS, emission lines in the near-infrared would be useful in exploring the neutral

and molecular regions surrounding PNs.

There are two immediate followup projects suggested by this dissertation. First,

the unexpected result for the Ring Nebula filaments demands further investigation.

One approach would be to look for very small condensations by obtaining observations 119 with increased spatial resolution (such as with the Hubble Space Telescope). It may also be worthwhile to re-examine other filamentary PNs to fully explore the distinction between ionization features and density features. Second, FLIERS could be clearly identified in my data by comparing the [S II] emission-line flux maps and the [S II] density maps. A single Fabry-Perot data cube covering the [S II] doublet would yield both of these maps, and so a good application of the techniques developed in the course of this dissertation would be a search for FLIERs in a large sample of PNs.

5.3 Final Remarks

The set of data cubes, narrowband images, and longslit spectra for each PN was a large amount of data. However, this was not a case of overkill. The results described above justify the extensive observations. The clumpy, continuous ionization front in the Ring

Nebula would never have been noticed with aperture or longslit spectroscopy. Most

PNs exhibit rich structure which can only be properly explored with two-dimensional imaging. The Fabry-Perot interferometer produced large data cubes covering only a few lines, but was essential to get the desired accuracy in the emission-line fluxes.

This was particularly important for the density maps. Lastly, the problem with the central star mass/age mismatch for NGC 7662 and NGC 7009 shows that this integrated approach is exactly what is needed to properly compare evolution models

and observations. A few emission-line flux images, say of Ha and [0 III], are just not

enough.

It is worth noting that all of the PNs in my sample were already among the

most studied PNs in the sky, and even then this level of effort was required to take the 120 next steps in understanding them. However, this type of project would not be suited to a general study of structure and evolution of PNs as the amount of data would be overwhelming. A more focussed project, though, has the potential to produce good results. For example, a set of [S II] density maps for a large group of PNs would be useful for comparisons with the predictions of the radiative-hydrodynamic models. B ibliography

Aller, L. H. 1984, Physics of Thermal Gaseous Nebulae, (Dordrecht iReidel) Ashley, M., and Hyland, A. R. 1988, ApJ, 331, 532. Atherton, P. D., Hicks, T. R., Reay, N. K., Worswick, S. P., 6 Smith, W. H. 1978, A&A, 66, 297 Balick, B. 1987, AJ, 94, 671 Balick, B. 1993, Proceedings of I AU Symposium 155, ed Weinberger, R. & Acker, A. (Kluwer, Dodrecht), p. 131 Balick, B., Gonzalez, G., Frank, A., & Jacoby, G. 1992, ApJ, 392, 582 Balick, B., Perinotto, M., Maccioni, A., Terzian, Y., & Hajian, A.1994. ApJpreprint. Balick, B., Preston, H., and Icke, V. 1987 AJ, 94, 1641. Balick, B., Rugers, M., Terzian, ¥., & Chengalur, J. 1993. ApJ, 411, 778 Barker, T. 1983, ApJ, 267, 630 Barker, T. 1986, ApJ, 308, 314 Barker, T. 1987 ApJ, 322, 922 Bland-Hawthom, J. 1994. in Tridimensional Optical Spectroscopic Methods in Astro­ physics, ed. G. Conte & M. Marcelin, ASP Conf. Ser., 71, 72. Boeshaar, G. 1974, ApJ, 187, 283 Bohigas, J., Lopez, J., and Aguilar, L. 1995, A&A, preprint Brocklehurst, M. 1971, MNRAS, 153, 471 Bryce, M., Balick, B., & Meaburn, J. 1994, MNRAS, 266, 721 Cai, W. & Pradhan, A.K. 1993, ApJS, 88, 329 Campbell, W. A. 1968 PASP, 80, 689

121 122

Capriotti, E. R., Cromwell, R. H., and Williams, R. E. 1971. Astrophysical Letters, 7,241 Capriotti, E. R. 1973, ApJ, 179, 495 Cardelli, J. A., Clayton, G. C., & Mathis, J. S., 1989 ApJ, 345, 245 Chu, Y.-H., & Jacoby, G. 1989, Proceedings of lAU Symposium 131, ed. Torres- Peimbert, S. (Kluwer, Dodrecht), p. 198. Chu, Y.-H., Jacoby, G., & Arendt, R. 1987, ApJS, 64, 529 Curtis, H. 1918, Publ. Lick Obs. 13, 57 Cyzak, S. J. and Aller, L. H. 1979, MNRAS, 188, 229 Dufour, R. J. & Quigley, R. 1993, in Proceedings of lAU Symposium 155, ed Wein­ berger, R. & Acker, A. (Kluwer, Dodrecht), p. 198 Frank, A. 1992, Thesis University of Washington Frank, A. 1994, AJ, 107, 261 Frank, A., Balick, B., Icke, V., & Mellema, G. 1993 ApJ, 404, L25 Frank, A., Balick, B., & Riley, J. 1990 AJ, 100, 1903 Frank, A., Balick, B., van der Veen, W. 1993, in Proceedings of lAU Symposium 155, ed Weinberger, R. & Acker, A. (Kluwer, Dodrecht), p. 365 Frank, A., & Mellema, G., 1994a, A&A, 289, 937. Frank, A., & Mellema, G., 1994b, ApJ, 430, 800. Geballe, T.R., Burton, M.G. and Isaacman, R., 1991, MNRAS, 253, 75 Gathier, R., & Pottasch. S. R. 1989, A&A, 209, 369 Greenhouse, M. A., Hayward, T. L., & Thronson, H. A. 1988, ApJ, 325, 604 Harrington, J.P., Seaton, M.J., Adams, S., & Lutz, J.H., 1982, MNRAS, 199, 517 Hawley, S. A., & Miller, J. S. 1977 ApJ, 212 94 Hummer, D. G., & Storey, P. J. 1987, MNRAS, 224, 801 Hyung, S. and Aller, L. H. 1995, preprint Icke, V., Balick, B., Frank, A. 1992 A&A, 253, 224 Jacoby, G. H., Quigley, R. J., & Africano, J. L. 1987, PASP, 99, 672 123

Kaler, J. B. 1980, ApJ, 239, 592 Kaler, J. B. 1986, ApJ, 308, 322 Kastner, J.H., Gatley, I., Merrill, K.M., Probst, R., & Wein.traub, D. A. 1994, ApJ, 421, 600 Kondratjeva, L. N. 1993, Proceedings of I AU Symposium 155, ed Weinberger, R. & Acker, A. (Kluwer, Dodrecht), p. 195 Kupferman, R N. 1983, ApJ, 266, 689 Kwok, S., Purton, C.R., & Fitzgerald, P.M., 1978, ApJ, 219, L125 Meaburn, J., Walsh, J.R., Clegg, R.E.S., Walton, N.A., Taylor, D., Berry, D.S., 1992, MNRAS, 255, 177 Mellema, G. 1993. Ph.D. Thesis, Leiden University Mellema, G. 1994, A&A, 290, 915 Mellema, G., and Frank, A. 1994 in Asymmetrical Planetary Nebulae, Annals of the Israel Physical Society, Vol. 11, eds. A. Harpaz and N. Soker Mellema, G. and Frank, A. 1995, preprint Middlemans, D., Clegg, R.E.S., Walsh, J.R., & Harrington, J.P., 1991, MNRAS, 251, 284 Minkowski, R. & Osterbrock, D. E. 1960, ApJ, 131, 537 Moreno, M. A., & Lopez, J. A. 1987, A&A, 178, 319 Osterbrock, D. E. 1989, Astrophysics of Gaseous Nebulae and Active Galactic Nuclei, (Mill Valley: University Science Books) Perek, L., and Kohoutek, L. 1967, Catalogue of Galactic Planetary Nebulae (Acad. Pub. House of the Czech. Acad. SCi., Prague) Pogge, R. W. 1991, OASIS; An Imaging Fabry-Perot Reduction Package, OSU As­ tronomy Internal Report 91-02 Pogge, R. W. 1992, in Astronomical CCD Observing and Reduction Techniques, ed. S. Howell, ASP Conf. Ser., 23, 195 Pogge, R., Atwood, B., Byard, P., O’Brien, T., Peterson, B., and Lame, N.J. 1995, in prep. Pogge, R. W., Owen, J. M., & Atwood, B. 1992, ApJ, 399, 147 124

Pottasch, S. 1984, Planetary Nebulae, (Dordrecht:Reidel) Reay, N. K., & Atherton, P. D. 1985, MNRAS, 215, 233 Reay, N. K., & Worswick, S. P. 1977, MNRAS, 179, 317 Reay, N. K., & Worswick, S. P. 1982, MNRAS, 199, 581 Richer, M.G., McCall, M.L. & Martin, P.O. 1991, ApJ, 377 210 Storey, P. J., and Hummer, D. G. 1995, MNRAS, 272, 41 Terzian, Y. 1993, Proceedings of lAU Symposium 155, ed Weinberger, R. & Acker, A. (Kluwer, Dodrecht), p. I l l Volk, K. & Leahy, D. 1993, AJ, 106, 1954 Weinberger, R. 1989 A&AS, 78, 301 Williams, R. E. 1973, MNRAS, 164, 111 Zuckerman, B., k Gatley, 1 .1988 ApJ, 324, 501