INFORMATION TO USERS

This manuscript has been reproduced from the microfilm master. UMI films the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter face, while others may be from any type of computer printer.

The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrough, substandard margins, and improper alignment can adversely affect reproduction.

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 copyright material had to be removed, a note will indicate the deletion.

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

Photographs included in the original manuscript have been reproduced xerographically in this copy. Higher quality 6" x 9" black and white photographic prints are available for any photographs or illustrations appearing in this copy for an additional charge. Contact UMI directly to order.

University Microfilms International A Bell & Howell Information Company 300 North Zeeb Road. Ann Arbor. Ml 48106-1346 USA 313/761-4700 800/521-0600

Order Number 9307846

Multiwavelength studies of classical nova shells

Saizar, Pedro, Ph.D.

The Ohio State University, 1992

UMI 300 N. Zeeb Rd. Ann Arbor, MI 48106

M ultiwavelength S t u d i e s o f C l a s s ic a l N o v a S h e l l s

dissertation

Presented in Partial Fulfillment of the Requirements for

the Degree Doctor of Philosophy in the Graduate

School of The Ohio State University

By

Pedro Saizar,

The Ohio State University

1992

Dissertation Committee: Approved by

Dr. Gary J. Ferland

Dr. Kristen Sellgren Adviser Di. Darren L. DePoy Department of Astronomy To my parents,

Alicia E. Fernandez

and

Pedro Saizar (h) A cknowledgements

I would like to express my deepest gratitude to my adviser, Dr. Gary J. Ferland,

for guiding me through the fascinating world of nebular astrophysics. His well-known

experience and insight were often crucial to point out the right direction in a complex

subject. But, at the same time, I had from him a high degree of independence and a

sense of being a colleague, rather than a student, all of which I feel also contributed

to my scientific formation. Finally, I need to thank his financial support for my

participation in the 1991 Workshop on Cataclysmic Variables, in Vina del Mar, Chile,

and the General Assembly of the International Astronomical Union, in my hometown

Buenos Aires, Argentina.

I am also greatly indebted to Dr. Sumner Starrfield, from Arizona State Univer­ sity, who also taught me many wonders in the field of Classical Novae. His constant support, and critical review of this work are greatly appreciated.

Next, I need to acknowledge Drs. Starrfield, Mark Wagner, and Steve Shore, for their warm hospitality during my visits to ASU, Lowell Observatory and NASA

Goddard Space Flight Center, respectively. Many thanks also to Drs. R. Williams,

J. Truran, W. Sparks, S. Kenyon, L. Stryker, M. Livio and R. Pogge, for their various contributions in the form of data, reviews and/or comments; to the IUE Observatory, for the ultraviolet data; and to the Lowell Observatory Computer Center for providing the facilities where the optical data were reduced and measured.

After 5 years at Ohio State, I could not possibly mention every person who helped me go through these difficult times. I would like to mention here a few names, but I certainly treasure many great feelings from as many friends. First, my appreciation to Dr. Capriotti, former Chairman of the Department, and through him, to all the

Faculty, for accepting me into the program and for their constant support. Among them, I must mention Drs. R. Wing and J. Villumsen. My appreciation also to my fellow students, and particularly to my good friends Bill Welsh, with whom I shared many sleepless nights and coffees in our quest for survival, and Nancy Jo Lame. To my friends in the USA, especially Fernando Fischer, Alejandra Zanetta, and Martin

Olivera, and to those back in Argentina, to whom I have never forgotten. Among my former teachers at La Plata Observatory, I need to acknowledge Dr. Hugo Levato for guiding me through my first steps in scientific research. And my gratitude to the scientists, journalists, and amateur astronomers, who helped me develop over the years a love for science writing and teaching.

Finally, the most sincere word of gratitude and love to my family. My trip to the United States was only possible thanks to the financial support of my father and

Roberto and Ana Marfa Campi. My mother has been a permanent driving force behind me, and I certainly owe this degree to her. And as I ponder about the future,

I can only express a last word of love to my future wife, Marina Folatelli, with whom

I hope to share all the delights and hardships of being an astronomer. V it a

July 23, 1959 ...... Born in Buenos Aires, Argentina.

1984 ...... Licenciado en Astronomia Universidad Nacional de La Plata, La Plata, Argentina.

1985 ...... Teaching Assistant, Universidad de Buenos Aires, Buenos Aires, Argentina.

1988-1992 ...... Graduate Research Assistant, Department of Astronomy, The Ohio State University, Columbus, Ohio.

P ublications

“PW Vulpeculae: A Nova with Nearly Solar Abundances”; Saizar, P., Starrfield, S., Ferland, G., Wagner, R. M., Truran, J. W., Kenyon, S. J., Sparks, W. M., Williams, R. E., and Stryker, L. L., 1991, Ap. J. 367, 310.

“PW Vul: A Classical Nova with Nearly Solar Abundances”; Saizar, P., Starrfield, S., Austin, S., Ferland, G. J., Wagner, R. M., Truran, J. W., Sonneborn, G., Kenyon, S. J., Sparks, W. M., Williams, R. E., Wade, R., and Gehrz, R. D., 1990, In Evolution in Astrophysics, ed. E. Rolfe (ESA SP-310), p. 435.

“Late Stages in the Evolution of Classical Novae”; Starrfield, S., Krautter, J., Son­ neborn, G., Shore, S. N., Wagner, R. M., Austin, S., Saizar, P., Ferland, G. J., Wade, R., Gehrz, R. D., Truran, J. W., Sparks, W. M., Shaviv, G., and Williams, R. E., 1990. In Evolution in Astrophysics, ed. E. Rolfe (ESA SP-310), p. 451.

“Ultraviolet Light Curves of Galactic and Extragalactic Classical CNO Novae: PW Vul, OS And, LMC 1988 # 1 and # 2” ; Austin, S., Starrfield, S., Saizar, P., Shore, S. N., and Sonneborn, G., 1990, In Evolution in Astrophysics, ed. E. Rolfe (ESA SP-310), p. 367.

F ie l d s o f S t u d y

Major Field: Astronomy

Studies in: Novae, Gaseous Nebulae - Dr. Gary J. Ferland

Infrared Photometry - Dr. Robert F. Wing T a b l e o f C o n t e n t s

ACKNOWLEDGEMENTS ...... iii

VITA ...... v

LIST OF FIGURES ...... ix

LIST OF TABLES ...... xii

CHAPTER PAGE

I Introduction ...... 1

1.1 Overview of the Classical Nova O utburst ...... 2 1.1.1 Thermonuclear R u n aw ay s ...... 2 1.1.2 T he Ejected S h e ll ...... 5 1.2 Presentation of the P r o j e c t ...... 6

II Nova PW Vulpeculae 1984 ...... 9

2.1 The Outburst ...... 9 2.2 Observational Material ...... 10 2.2.1 O ptical S p ectra...... 10 2.2.2 Ultraviolet S p e c tra ...... 16 2.3 Reddening and Distance ...... 22 2.4 The N ebular Gas R eg io n ...... 24 2.4.1 Evolution of Emission L in es ...... 24 2.4.2 Temperatures and D ensities ...... 31 2.4.3 Chemical Composition ...... 34 2.5 The Coronal Line Region ...... 40 2.6 A Model of the Ejecta ...... 45 2.6.1 M odel P a ra m e te rs ...... 45 2.6.2 Comparison with the Observations ...... 46 2.7 Discussion and Sum m ary ...... 48

III Nova QU Vulpeculae 1984: The Cold Gas Phase ...... 50

3.1 Overview of the O u tb u r s t ...... 50 3.2 Observational Material ...... 54 3.2.1 Optical Spectra ...... 54 3.2.2 Ultraviolet Spectra...... 60 3.2.3 Interpolation of Spectra ...... 69 3.3 Reddening and Distance ...... 70 3.4 The Nebular Region ...... 72 3.4.1 Electron Densities and Temperatures ...... 73 3.4.2 Abundances ...... 79

IV Nova QU Vul 1984: The Hot Gas P h a se ...... 87

4.1 Expansion V elo cities ...... 88 4.2 Temperature of the Hot Gas ...... 89 4.3 The Density of the Hot G as ...... 94

V Nova QU Vul 1984: The Overall Shell Structure ...... 100

5.1 T he Mass of the E j e c t a ...... 101 5.2 Energetics...... 104 5.2.1 Evidence for Pressure Equilibrium ...... 104 5.2.2 Evidence for a “Clumpy” Structure ...... 105 5.2.3 Where does the Emission Spectrum O riginate? ...... 106 5.2.4 Thermal Stability ...... 109 5.2.5 Energy Budget ...... 109 5.3 S u m m a r y ...... 113

VI Concluding Remarks ...... 115

6.1 Implications for Nova Physics ...... 115 6.2 Implications for Galactic Nucleosynthesis ...... 118 6.3 C onclusions...... 120

BIBLIOGRAPHY ...... cxxii

viii L is t o f F ig u r e s

FIGURE PAGE

1 The standard model of a Cataclysmic Variable Star. Courtesy of W.

Welsh and K. H orne ...... 3

2 Montage of optical spectra of PW Vul during outburst, (a) September

25, 1985 (day 413); (b) April 20, 1986 (631); (c) June 8, 1986 (680);

(d) May 4, 1987 (1011) ...... 12

3 Enlargement of the spectrum of PW Vul obtained on September 15,

1985 showing the principal emission features ...... 14

4 High dispersion spectrum obtained on March 27, 1987 near H /3. (Inset)

Enlargement of the region between 4400 and 4900 A ...... 17

5 Montage of ultraviolet spectra of PW Vul during outburst, (a) April

7, 1985 (day 253); (b) June 24, 1985 (331); (c) July 16, 1985 (353); (d)

July 25, 1985 (362); (e) October 29, 1985 (458); (f) M arch 31, 1986

(611) 18

6 Reddening in the field of PW V u l ...... 26

7 Light curve of H/? including a least square fit of the data ...... 30

8 Evolution of [0 III] lines ...... 32 9 Electron temperature vs. density from [0 III] lines ...... 35

10 The continuum spectrum of PW Vul relative to H/3 ...... 41

11 The first 2 months: a light curve of QU Vul, from visual estimates

reported in IAU Circulars ...... 52

12 The first 3 years. Visual observations are averages of amateur esti­

mates published by the AAVSO. Least-squares fits to the data and

the respective slopes are included ...... 53

13 Sample spectra of QU Vul. (a) Day 145-155; (b) Day 1064. Note that

the scale follows the flux of the strongest line ...... 55

1 4 Evolution of the He I lines ...... 61

15 Ultraviolet spectrum of QU Vul at day 126 (April 30, 1985): (a) ob­

served spectrum, showing the depression centered at 2200 A, due to in­

terstellar absorption; (b) same spectrum corrected for reddening. Note

that the feature has virtually disappeared ...... 63

16 A gallery of ultraviolet spectra: (a) day 158, (b) d ay 191, (c) day 204,

(d) day 340, (e) day 462, (f) day 627. Note that the scale is the same

in all panels to show the evolution of the emission lines. The spectra

are not corrected for reddening ...... 64

17 Observed He II 1640/He II 4686, useful for reddening correction. . . . 71

18 Evolution of th e [0 III] l i n e s ...... 76

19 Velocity fields from various emission lines, as a function of time: open

circles, N V 1240; closed circles, [0 III] 1663; squares: H/?...... 90 20 The continuous spectrum of QU Vul normalized to the continuum flux

at H/3...... 96

21 Density diagnostics: model predictions of the UV to optical continuum

ratio as a function of density, and for various stellar temperatures . . 99

22 The continuum spectrum of QU Vul ...... 102

23 Thermal stability in a very hot gas: solid line: cooling rate; dotted line:

heating rate. Units in erg cm-3 sec-1 ...... I ll

24 Thermal stability of the cold phase. Axes and units as in the previous

figure...... 112

xi L ist o f T a b l e s

TABLE PAGE

1 Photometry and distances in the field of PW Vul ...... 25

2 Light Curve of H ( i ...... 27

3 Optical Emission Lines ...... 28

4 Ultraviolet Emission Lines ...... 29

5 Electron Temperatures and Densities ...... 36

6 Abundance of Helium ...... 37

7 Abundances of Oxygen, Nitrogen and Neon ...... 38

8 Abundances in PW Vul ...... 39

9 Continuum Fluxes at day 330 ...... 42

10 A M odel for Day 4 1 3 ...... 47

11 Optical Lines ...... 56

12 Weaker Optical Lines ...... 59

13 Journal of IUE Observations ...... 62

14 Ultraviolet Lines ...... 68

15 Electron density and temperature ...... 75

16 Intensity and Luminosity of H /? ...... 78

xii 17 Abundances in QU Vul ...... 86

18 The Continuous Spectrum of QU Vul ...... 91

19 Emission from the Hot P h a s e ...... 92

20 Observed Fluxes for Continuum Diagnostics ...... 95

21 Emission from the Hot Gas Phase ...... 97

22 Model Parameters and Predictions ...... 101

23 Emission Line Fluxes in the Hot G a s ...... 107

24 Model Parameters ...... 108

25 A Model for Day 554 ...... 110

26 The Ejected Mass of QU Vul ...... 117

xiii CHAPTER I

Introduction

Well known to vigilant amateur astronomers, novae are the “new stars” that suddenly become visible to the telescope or even to the naked eye, reaching a peak brightness in a matter of hours, to then fade away out of sight in a period of a few months to a few years.

The rapid photometric and spectral changes in a nova event generated an enor­ mous amount of scientific interest, especially since the first major work of classification and interpretation by Cecilia Payne-Gaposchkin (1957). Since then, it has been re­ alized that the term “nova” included many objects of very different properties, and therefore narrower definitions became available. Based on their observed properties,

Classical Novae are defined as a class of Cataclysmic Variable stars which have only one recorded outburst in their history. They have other common properties which complete their defining properties (Gallagher and Starrfield 1978), namely, an optical brightness increase of at least 9 mag. in a few days or less; significant photometric and spectroscopic changes in times scales of less than 3 years; and expansion velocities between 100 and 5000 km sec-1, as deduced from their spectral lines.

This dissertation is concerned with two classical novae, PW Vulpeculae (1984

#1), and QU Vulpeculae (1984 #2). But before describing the nature of this project and its importance to Astronomy, let us briefly review our current understanding of

classical nova outbursts. A basic frame of reference is presented here as more details

are presented throughout this work as needed.

1.1 Overview of the Classical Nova Outburst

The wealth of data and curiosity for the nova phenomenon rapidly contributed to

establish the basic frame for their understanding. Payne-Gaposchkin (1957) and

McLaughlin (1960) made the first systematic efforts to understand the complexities

of the spectral development. But it was only in the seventies, through the work of

Starrfield and collaborators (Starrfield, Truran and Sparks 1972, Starrfield, Sparks,

and Truran 1974), that a successful framework was found to explain many of the

observed properties of nova outbursts.

1.1.1 Thermonuclear Runaways

A classical nova outburst originates from a thermonuclear runaway (TNR) in the

surface layers of the white dwarf component of a close binary star. The secondary

star (usually, but not always, a red dwarf star), has filled its Roche equipotential lobe

and transfers matter to the white dwarf, which accretes it through a disk (Figure 1).

The deposition of hydrogen-rich material onto the surface of the metal-rich white

dwarf will eventually lead to the thermonuclear runaway. The transfer and accretion mechanisms are discussed by Starrfield (1988) and Shara (1989).

Once the temperature at the base of the accreted envelope reaches the point where hydrogen thermonuclear burning starts, the system abruptly increases in brightness as Figure 1: The standard model of a Cataclysmic Variable star. The bigger and cooler companion transfers matter to the white dwarf which accretes it through a disk. The accretion process can eventually lead to a nova outburst. A complete orbit is shown here, where the dot indicates the system's center of mass. Picture courtesy of W. Welsh and K. Home. the outer layers expand with velocities of several hundreds to thousands of km sec-1.

The details of the outburst depend on parameters such as the mass and composition of

both the envelope and the white dwarf, as well as the accretion rate (Starrfield 1989).

The same general features, however, occur in the early evolution of most novae.

Once the peak brightness is reached, corresponding to the point of maximum ex­

pansion of the pseudo-photosphere, the ejection of a gaseous shell follows. This can

be seen in the spectra as the change from a pure absorption spectrum to a spectrum

showing P-Cygni profiles, with broad emission lines, and absorption components usu­

ally showing high blue-shifted terminal velocities. These absorption components,

however, disappear after a few weeks or months, and a spectrum consisting of pure

emission lines indicates that the nova has entered in a “nebular” stage, a few weeks

or months after the peak of the outburst.

Very early in nova studies it was suspected that the chemical composition of the

shell must be anomalous, which is now interpreted in the context of TNR theory.

Starrfield’s hydrodynamic simulations show that the energetics of the outburst is also

sensitive to the abundance of carbon, nitrogen, oxygen and/or neon. Some of these

and other metals are mixed into the envelope from the white dwarf’s core material,

but some are also synthesized during the runaway, as the temperature in the envelope

reaches the critical temperature for the CNO cycle. As a result, the ejecta will be

rich in intermediate mass elements, depending whether the outburst took place on

a carbon-oxygen or an oxygen-neon-magnesium white dwarf. Being related to the energetics of the outburst, the enhanced abundances are also related to the “speed class” of the nova (Truran 1982). The “fastest” novae usually show the strongest enhancements.

1.1.2 The Ejected Shell

Rushing its way from the binary system into the interstellar medium, the ejected gaseous shell goes through periods where the physical conditions change at a rate rarely seen in other astrophysical objects. The gas density decreases from almost photospheric values (~ 1012 cm-3), to near interstellar ones (~ 103 cm-3) in just a few years. The gas temperature as well as the radiation field also change appreciably as the source of radiation continues evolving.

A considerable effort has been invested over the years in determining the chemical composition of classical nova shells in order to compare them to the predictions of outburst theories. However, the observed abundances cannot be interpreted in terms of the thermonuclear reactions alone. It is well known that many novae form dust a few months after the outburst (see, e.g., a review by Gehrz 1989). Ablation from the secondary star can also change the observed composition of the gas. It is not possible, therefore, to make a direct comparison between the theoretical predictions and the observed abundances. A qualitative comparison shows, nevertheless, a good agreement in many cases, although the number of reliable determinations is relatively small.

The structure of the nebular shell has offered many challenges to researchers as well. It is clear that the expanding shells are neither spherical nor homogeneous. In only a handful of cases, resolved images are available, such as those of Nova DQ Her 1934, showing bright condensations across the shell. Indirect evidence from other studies also indicate their inhomogeneous nature (Williams et al. 1991). The lack of sphericity has been discussed by Livio et al. (1991), as due to a “common envelope phase” in which both components of the binary system share material during the initial moments of the outburst. This effect would induce a very high mass loss rate on the white dwarf and give the shell a non-spherical geometry.

In addition, there is some evidence that the shell might have more than one gas phase. Grasdalen and Joyce (1976) were the first to suggest that a hot “coronal line” region might exist in addition to the cooler “nebular” phase. Despite the few optical signatures, this coronal line region seems to play an important role in the decline.

Ferland, Lambert and Woodman (1986) have observed this phase in Nova Cygni 1975, but little else is known about its properties and physical conditions.

1.2 Presentation of the Project

We present here a detailed study of two galactic novae, PW Vul 1984 and QU Vul

1984. We will be mostly concerned with the physical conditions in their shells during the later stages of their evolution, starting about 4 months after the optical maxi­ mum. In each case, we have combined ground-based optical and satellite ultraviolet spectrophotometric data to measure the temperature, density and chemical compo­ sition of their shells, as well as to investigate the nature of the source of continuous emission.

In addition, we also performed radiative-equilibrium photoionization model cal­ culations and their predictions are compared to the observations to further constrain the physical conditions in the shell.

There are several motivations for a study of this nature.

• The launch in 1978 of the International Ultraviolet Explorer opened the ultra­

violet window for astronomical research and has kept it that way for the last 14

years. This orbiting telescope allowed observers to follow bright novae for sev­

eral years, and through the “Target of Opportunity” program, they have been

able to obtain spectra within hours of their optical discovery. Since then, IUE

has helped to greatly advance our understanding of classical novae, but still

only two dozen or so novae could be followed in both optical and ultraviolet

wavelengths as they declined from peak magnitude (Friedjung 1989). At this

point, then, there are very few novae which have been studied in detail using

multiwavelength analysis. In this sense, we have also tried to include results and

data from other published studies in order to develop a picture of the outburst

and its relation to the parent system.

• As mentioned above, the coronal line region is a poorly known part of the gas

shell due to its slight effect on the optical emission spectrum. Based on careful

analyses of the ultraviolet spectra we have been able to further investigate its

nature. We will see below that the coronal line region dominates both the mass

and the energetics of the outburst in each nova studied here.

• From the discussion in the previous section, it is clear that the shell can give

us information about the underlying binary system. The calculation of abun­

dances in the ejecta is the only observational test of the theory of thermonuclear runaway, and, although an exact comparison is not yet possible, an order of

magnitude agreement would provide strong evidence in their favor.

• Our final motivation is relevant to studies of chemical enrichment of the Galaxy.

It is well known that mass loss from evolved stars can provide the interstellar

medium with significant amounts of material which has been processed by var­

ious thermonuclear reactions and, therefore, is substantially richer in heavy

elements than most of the gas present in the interstellar medium (see, e.g.,

Audouze and Tinsley 1976).

A single nova does not return much material to interstellar space compared to

supernova explosions, novae being typically between 10-6 to 10-3 M®. Their

higher frequency of occurrence, however, more than compensates for this smaller

mass and make them suitable candidates to contribute to the chemical enrich­

ment of the Galaxy. Weiss and Truran (1990), and Nofar, Shaviv, and Starrfield

(1991) show that novae can provide the ISM with significant amounts of nuclei

such as 13C, 15N, 170, 22Na, 26A1. An important result of this work is that it

is indeed possible that a few novae of the type of QU Vul can provide much of

the observed mass of galactic 26A1.

We begin this project with our study of PW Vul (Chapter II). In Chapters III and IV, we discuss the properties of the cold and hot gas phases, respectively. Then we use that information to examine the overall structure of the shell in Chapter V.

In the final chapter, we consider the implications of this study for the field of nova physics, and close with a summary. CHAPTER II

Nova PW Vulpeculae 1984

2.1 T he Outburst

PW Vul reached its maximum visual brightness of 6.3 mag on August 4, 1984 (J.D.

2,445,918). A light curve covering observations made on a period of over 800 days has been published by Gehrz et al. (1988). We adopt July 29, 1984, as day 1 of the outburst.

The early optical development was discussed by Kenyon and Wade (1986), whereas the infrared evolution was discussed by Gehrz et al. (1988). Based on their data, we can make a preliminary description of the early outburst.

Before the optical maximum, an optically thick pseudo-photosphere expanded during the first 8 days of the outburst. An infrared spectrum at the time of maximum light presented by Gehrz et al. (1988) indicates a blackbody temperature of about

6,700 K for the m aterial.

The optical decline was somewhat irregular. From the light curve by Gehrz et al. we infer that the time to fall 3 mag. is t3 ~ 140 days. This time scale, normally taken as a measure of the “speed class” of the nova (Payne-Gaposchkin 1957), implies that

PW Vul was a very slow nova. After the time of maximum light, the ejecta became an

9 10 optically thin shell both in visible and infrared light. The width of H/3, measured from the data to be discussed in the next section, indicates an average expansion velocity of about 600 km/sec, which agrees well with Kenyon and Wade’s (1986) measurement of the FWHM of Ha. Ionized gas in the “principal” ejecta was moving at a somewhat slower rate. A spectrum taken in March 27, 1987 with the Perkins 1.8-m telescope is shown in Fig. 4 on page 17. The double peak observed in the [0 III] 5007 line implies an expansion velocity of about 400 km/sec. Infrared observations (cf. Gehrz et al. 1988) also suggest ejection velocities of 300-400 km/sec during this phase of the decline.

Gehrz et al. also presented flux measurements of the infrared continuum, which we shall supplement with ultraviolet data. Up to day ~100, the entire continuum had an energy distribution characteristic of optically thin free-free emission (i.e., f„ ~const.). At about day 280, a flux excess developed in the infrared, which th e y interpret as thermal emission from optically thin dust. The possibility that this free- free continuum is due to very hot gas in a coronal line region will be discussed further in Section 2.5.

2.2 Observational Material

2.2.1 Optical Spectra

Optical spectrophotometric observations of PW Vul were obtained by Dr. R. M.

Wagner (Ohio State University), beginning on September 15, 1985 (day 413) and continued through May 4, 1987 (day 1011). These observations were obtained using the Ohio State University Image Dissector Scanner (Byard et al. 1981) attached to the Perkins 1.8-m telescope of the Ohio Wesleyan and Ohio State Universities at the

Lowell Observatory. These data supplement those already reported by Kenyon and

Wade (1986), for the period October 1984-April 1986.

Dual 7” diameter entrance apertures and a 600 line-per-mm grating blazed at

A5500 A were employed which covered about 2600 A of the spectrum a t 10 A resolu­

tion. Two different grating tilts were required to cover the region between 3700 and

8600 A. Higher dispersion spectra in the H/3 region were obtained on March 27, 1987

using the same telescope and instrument. For these observations, a 1800-line-per-

m m grating blazed at A4650 was used, which, when combined with 5” dual entrance

apertures, yielded a spectral resolution of about 2.5 A and covered the region 4400-

5100 A. The spectrum of a quartz-halogen lamp was observed to remove pixel-to-pixel

variations in response, and the spectrum of an FeNe-He source provided wavelength

calibration. The observations of one or more standard stars permitted the removal

of the instrumental response and provided absolute photometric calibration. The

reduction procedures are described in Wagner (1986).

In Figure 2(a)-(d), we show a montage of our late-time optical spectra of PW Vul scaled to enhance the weaker features. The principal emission features are identified in Figure 3 from the spectrum obtained on day 413.

As shown in Figure 2, the optical spectra consist of a mixture of both permitted and forbidden emission lines of both high and low ionization. The strongest emission lines throughout the late time development are those due to [O III] 5007 and Ha.

Weaker permitted emissions are due to other members of the Balmer series of hy- 12

a) 413 in

in

o

b) 631 in

in

o

4000 5000 6000 7000 8000 Wavelength (A)

Figure 2: Montage of optical spectra of PW Vul during outburst, (a) September 25, 1985 (day 413); (b) April 20, 1986 (631); (c) June 8, 1986 (680); (d) May 4, 1987 (1011). 13

Figure 2 (continued)

c) 600 in

in I V) ftI © BO U60 0) 1011 n I to

o l i .... i .... i . . 4000 5000 6000 . 7000 8000 Wavelength (A) iue3 Elreeto h pcrmo WVlotie nSpebr1, 1985 15, September on Vul obtained spectrumPW the of of Enlargement 3: Figure showing the principal emission features. principalemission the showing Fx (10“13 erg cm -* s'* A.-1) W 00 00 00 00 8000 7000 6000 5000 4000 [0 III][0 Wavelength Wavelength . h II (A) a 14

15

drogen, and also, He I AA 4471,5876,6678,7065; He II AA 4686,5411; N III 4640, and

N V 4600. The strongest forbidden lines arise from [0 III] 4959,5007, [O III] 4363;

[0 II] 7320,7330; [N II] 5755; [Ne III] 3869,3968; [A III] 7136; [A IV] 7237,7263,7170

(blended with [A III] 7136); and weaker emission due to [Fe II]. In the later spectra,

the pair [N II] 6584,6548 begins to appear as structure in the H a profile. Many coro­

nal lines are present in the late-time spectra, including those of [Fe VII] 6085, and

[Fe VI] 5677. [Fe X] 6374 is also present as suggested by the anomalous doublet ratio

of [O I] 6300,6363.

The luminosity of the emission lines decreases quickly with time. The relative line

strengths also change with time indicating changes in the physical conditions of the

ejecta. For example the [0 III] 5007 to Hct ratio changes from about 1.6 on day 413 to

about 3.7 on day 1011 even though the total flux of [0 III] has decreased by a factor

of 16. Note also the changes in the relative strength of the coronal lines, especially

[Fe X] 6374. On day 631 the strength of [0 I] 6300 is comparable to [0 I] 6363 and

[Fe X] 6374, but by day 680, [Fe X] has decreased significantly to the point where the [O I] lines have a relative intensity as predicted by the ratio of their transition probabilities. Other obvious changes in relative line strength produced by decreasing density in the emission region include the rapid decrease in intensity of [N II] 5755 and [O III] 4363.

In Figure 4, we show the high dispersion spectrum of PW Vul in the H (3 region obtained on March 27, 1987. Note the well known castellated structure of the emission lines, especially of the doublet [O III] 4959,5007. At this resolution (150 km/sec), the profiles break up into at least three major components. The expansion velocity

as inferred from the line profiles is about 400 km/sec. However, a comparison of the

[0 III] or H/3 line profiles with those of He II 4686 and N III 4640 suggests that the

latter two species may arise or originate in a different manner or location than [0 III].

The observed fluxes relative to H/3 and corrected for reddening as discussed in

Section 2.3, are presented in Table 3 (see p. 28).

2.2.2 Ultraviolet Spectra

Ultraviolet observations of PW Vul with the IUE satellite began on August 2, 1984,

two days after discovery, and continued until June 23, 1988. Here we will concentrate

on the late time spectra that are suitable for abundance determinations from nebular

emission line analyses. In Figure 5(a)-(f) we display combined short wavelength

prime (SWP) plus long wavelength prime (LWP) spectra obtained during the same

shift. In the following paragraphs we shall discuss each of these spectra in turn. Note

that while we display the same wavelength scale for most of the spectra, we allow the

peak flux to follow the emission lines and it slowly decreases with time.

As shown in Figure 5, the spectrum at late times contained a mixture of both high

and low ionization lines with both resonance and intercombination lines present. In

particular, the presence of N V 1240 indicates a strong ionizing continuum which is

borne out by the EXOSAT detection of PW VUL in June 1985 (Ogelman, Krautter,

and Beuermann 1987).

Figure 5(a) shows the spectrum obtained on April 7, 1985, about 9 months after discovery. This is the first spectrum taken after PW Vul had reappeared from behind CM I 'I I I I I I II I < I I I I ' I I I I I I 4400 4500 4600 4700 4800 4900

_A Vj L I i i I I I I I i t i i I i i i i ttii MIL ill' u 4400 4500 4600 4700 4000 4900 5000 5100 Wavelength (A)

Figure 4: High dispersion spectrum obtained on March 27, 1987 in the H/3 region. Note the castellated structure of the emission lines implying an expansion velocity of about 400 km/sec. (Inset) Enlargement of the region between 4400 and 4900 A showing the line profiles ofNIII A4640, Hell4686, and H/?. iue : otg o lrvoe setao W u drn otus, a Arl 7, April (a) outburst, during Vul PW of spectraultraviolet of Montage 5: Figure 32; e Otbr2, 95 48; f March (611). 1986 1985 (f) 31, (458); 29, October (e) (362); 95 dy23; b Jn 4 18 (3) () uy 6 18 (5) () uy2, 1985 25, July (d) (353); 1985 16, July (c) (331); 1985 June 24, (b) 253); (day 1985

Ffc (I0 'w erg cm in in in in 50 00 50 3000 2500 2000 1500 Wavelength Wavelength ) 331 b) ) 253 a) (A) A i i ___ 18

iue5 (continued) 5 Figure

Fk (I0',a erg 1500 00 2500 2000 Wavelength Wavelength (A) 3000 19 iue5 (continued) 5 Figure F* (10”,a erg cm ”* s”1 A.”1) 1500 00 2500 2000 Wavelength Wavelength (A) 3000 20 the sun. The previous spectrum, obtained on November 13, 1984, still shows a very strong continuum. The spectrum in Fig. 5(a) is dominated by lines, showing that the expanding shell has become optically thin, the continuum has decreased in intensity, and emission lines characteristic of a low density gas have become prominent. The strongest lines are N IV] 1486, C IV 1550, N III] 1750, C III] 1909, C II] 2326, Mg

II 2800, and O III 3133. Other lines are listed in Table 4 (see p. 29). The region between about 2000 A and 2400 A shows the noise level at the short wavelength edge of the LWP camera.

Figure 5(b) was obtained on June 24, 1985 and while the emission lines are steadily decreasing in intensity, they appear to be retaining virtually the same intensity ratio.

However, Mg II 2800 has dropped m uch faster than C IV 1550. A t first glance, the strength of C II], C III], and C IV would suggest that carbon was enhanced in the ejecta. However, they are strong only because of the characteristics of the shell and not, as we show below, because carbon is enhanced in abundance which suggests that one must be careful in interpreting line strengths as face value indicators of enhanced abundances. Figure 5(c) was obtained on July 16, 1985 and shows the continuing decline of the emission lines. It was followed soon thereafter by the spectrum shown in Figure 5(d) obtained on July 25, 1985. In this spectrum we extend the red edge to 3400 A to show that the neon lines are starting to become strong. Below, we show that this is because of the physical conditions in the expanding shell, and not because of any abundance enhancement. The sharp feature just to the red of 0 III 3133 is a cosmic ray “hit”. By October 29, 1985 (Figure 5(e)), the peak flux in C IV has 22 declined by more than a factor of three and some weaker features are beginning to appear.

The last usable combined spectrum, obtained in March 31, 1986, is shown in

Figure 5(f). It was somewhat underexposed accounting for the increased noise level.

Mg II has clearly disappeared although C IV and N IV] are still present. We believe that all of the “features” between 2000 A and 2400 A are noise. A slight rise in the flux is observed towards the red end of the LWP camera; however, we do not know whether this feature is real.

2.3 Reddening and Distance

An estimate of the reddening can be obtained by using the ratio of intensities of He II

1640 relative to He II 4686. Seaton (1978) calculated the intensities of several Case-B

He II recombination lines, including several radiative and collisional processes. He predicted that j(He II 1640)/j(He II 4686)=6.81, for an electron density of Ne = 106 cm-3, and a temperature of 10,000 K, which is within the range of observed values as we shall show below. The observed ratios for several dates can be computed from the data presented in Tables 3 and 4 (pp. 28 and 29). This ratio is a weak function of temperature and density, so we assume that it was constant over our observations.

Thus, we can derive an E(B-V) for each day, following the method described by

Seaton (1979). For days 253, 331, 340, and 681, the B-V excesses are 0.55, 0.71, 0.62, and 0.52, respectively. Therefore, the average is E(B-V)=0.60±0.06.

The color excess just derived is somewhat higher than the one adopted by Duer- beck et al. (1984), based on the study of the galactic extinction by Neckel and Klare (1980), who assumed E(B-V)~0.45, in the direction to the nova and, thus, a visual absorption of 1.4 mag.

Duerbeck et al. (1984) obtained a distance to the nova of 1.2 kpc, based on the equivalent width of the interstellar Ca II K line. On the other hand, Gehrz et al.

(1988), derived a distance of 6.35±0.35 kpc from the angular expansion rate of the pseudo-photosphere, and from several estimates of the luminosity from the decline of the light curve. They assumed that the absorption was negligible. Since, as discussed at below, the reddening is not zero, their estimate can be considered as an upper limit to the distance.

A first estimate of the nova’s distance can be obtained from the relationship between speed class, t 2, and absolute magnitude at maximum, Mv (Warner 1989):

Mv = 1.76(±0.31) x logt2 - 10.42(±0.38).

Using t2=53 days, from Evans et al. (1990) light curve, we obtain Mv= -7.4+0.9.

Then, with mv(max) = +6.3 mag., and E(B-V) = 0.60+0.06, the distance is kpc.

We can obtain an estimate of the distance by measuring the amount of reddening in field stars within 3 degrees of the nova, following the method discussed by Schild

(1976). The advantage of this method is that it is independent of the somewhat uncertain properties of the nova. For this purpose, we used the latest version of the uvby/3 photometric catalogue by Hauck and Mermilliod (1985), kindly provided by the Astronomical Data Center. 24

The indices ci and mi can be corrected for absorption using the formulas given by

Stromgren (1966). From the plot [mi] vs. [ci] (see Stromgren’s review), we can esti­

mate the spectral type and luminosity class, and thus, the absolute magnitude of the

stars in the catalog. We have used Mermilliod’s (1987) UBV photometric catalogue,

also provided by ADC, to obtain the color excesses and corrected visual magnitudes.

In Table 1 we present, for brevity, only the star HD numbers, the excesses, and the

distances. The individual data can be found in the catalogs mentioned. The resulting

distances are plotted against individual E(B-V) excesses in Fig. 6.

Unfortunately, an all-sky survey does not provide enough stars for a good estimate.

A second problem is that uvby/3 photometry is not good for the red giant stars which,

in general, are the more distant objects. However, from the plot we infer that the nova

cannot be closer than about 1.5 kpc, since the stellar color excesses are significantly

smaller, and probably it is farther than about 3 kpc. In addition, a map of the galactic

gas by Lucke (1978) shows a large cloud complex at a distance of about 3 kpc. Since

the color excess of this cloud seems somewhat large, we can infer that the star might

be in or beyond this cloud. So, we shall adopt d~3 kpc as a plausible distance for

PW Vul. Note that this distance is also consistent with our previous determination from the speed class-absolute magnitude relationship.

2.4 The Nebular Gas Region

2.4.1 Evolution of Emission Lines

The spectra show emission lines typical of the nebular phase of a nova. The intensities and luminosities for a recombination line like Hf3 are collected in Table 2 as a function 25

Table 1: Photometry and distances in the field of PW Vul

HD# E{B - V) D(pc) 180 502 0.06 47.9 180 553 0.15 316.2 180 583V 0.07 1202.0 180 584 0.36: 2399.0 180 615 0.08 57.5 180 917 0.11 91.2 181 602 0.08 69.2 182 008 0.09 83.2 182 255 0.01 158.5 182 568 0.14 263.0 182 618 0.07 302.0 182 807 0.03 1000.0 183 032D 0.11 57.5 183 033 0.23 10470.0 183 560 0.12 100.0 183 614 0.05 631.0 183 914 0.00 114.8 184 058 0.05 120.2 184 151 0.12 60.3 184 384 0.14 83.2 184 720 0.36 8710.0 184 979 0.06 63.1 184 998 0.00 138.0 185 242 0.15 229.1 185 269 0.08 33.1 185 270 0.04 87.1 26

hi

0 1 2 3 4 Log d (pc)

Figure 6: Reddening in tlie field of PW Vul. The dashed line is based on. Neckel and Klare’s (1980) work. The full line is just a linear growth of absorption with distance typical for our Galaxy. The “stars” represent field stars. “Sun" symbols are the available estimates for PW Vul, from left to right: Duerbeck et al. (1984), this paper, Gehrz et al. (1988). 27

Table 2: Light Curve of Hf3

t I(Hf3) W ) 0 L(Hf3) 235 1.68(-11) 9.23(-ll) 9.94(34) 309 5.90(-12) 3.24(-ll) 3.49(34) 413 2.60(-12) 1.43(-11) 1.54(34) 631 2.84(-13) 1.57(-12) 1.69(33) 680 2.62(-13) 1.44(42) 1.55(33) 1011 7.71 (-14) 4.2 (-13) 4.52(32)

of time. In Table 3, we present the ratios with respect to H(3 for several optical lines.

The ultraviolet emission lines are listed in Table 4. All reddening corrections have been calculated using the fits to the interstellar extinction curve obtained by Seaton

(1979). We shall use these data to derive temperatures and densities and to compare them to model calculations.

Fig. 7 shows that H/? declines monotonically from day 235 to day 1011, as can be expected for a matter-bounded photoionized expanding shell. A least squares fit of the data allows us to express the flux in H/? as a power law given by

f{H0) = const x r 3-84 . (2.1)

If hydrogen is assumed to be fully ionized, then the intensity of H/3 is given by

I(H/3) ~ hvH(3 NeNp a '" ■ V , (2.2) where a eJp is the effective recombination coefficient for H/3 and V is the volume of the emitting region. If we consider the mass of the ejecta M=/x Ne x V to be constant, Table 3: Optical Emission Lines

Element Days since outburst® 235 309 413 631 680 1011 HP 1.00 1.00 1.00 1.00 1.00 1.00 //e/5 8 7 6 0.15 0.15 0.14 0.20 0.15 0.12 0.10 0.10 0.10 0.14 0.10 0.08 H ell4686 0.29 0.36 0.40 0.47 0.49 0.57 0.31 0.39 0.43 0.51 0.53 0.62 [Nil]5755 0.82 1.46 1.98 2.16 1.55 0.68 0.60 1.06 1.44 1.57 1.13 0.49 [07]6300 — 0.15 0.30 0.43 0.45 0.30 — 0.09 0.19 0.27 0.28 0.19 [0/7)7330 — 0.66 3.23 1.15 0.76 0.35 — 0.33 1.62 0.58 0.38 0.18 [0777)4363 1.08 1.80 1.53 0.99 0.88 0.46 1.36 2.28 1.93 1.25 1.11 0.58 [0777)4959 + 5007 3.80 9.71 23.19 43.31 45.42 49.94 3.59 9.18 21.93 40.96 42.95 47.23 [Neill] 3869 0.27 0.51 0.75 0.38: 0.56 — 0.42 0.79 1.16 0.59: 0.86 — [7eV7/]6087 — 0.19 0.29 0.44 0.27 0.36 — 0.13 0.19 0.29 0.18 0.24 “Each entry: line 1, observed ratio; line 2, corrected Table 4: Ultraviolet Emission Lines

Element Days since outburst a 253 331 340 681 H e i n m 3.30(-12) 1.08(-12) 1.22(-12) — 0.26 0.25 0.30 — 2.58 2.48 2.98 — C7V1549 — 1.30(-11) 1.12(-11) 5.42(-13) — 2.82 2.62 2.06 — 31.38 28.59 22.48 JVV1240 — 5.13(-12) 4.75(-12) 3.40(-12): — 1.10 1.11 1.29 — 30.31 30.58 35.54 A/7 V] 1486 — 7.08(-12) 7.29(-12) 5.56(-13) — 1.51 1.71 2.11 — 18.11 20.50 25.30 A /7//] 1750 — 6.23(-12) 4.89(-12) 1.47(-13) — 1.33 1.15 0.56 — 12.74 11.01 5.36 OIV] 1402 3.36(-12) 1.15(-12) 9.93(-13) — 0.22 0.25 0.23 — 3.08 3.50 3.22 — [ o n i] m o ,6 4.31(-12) 8.43(-13) 1.10(-12) 1.32(-13) 0.29 0.18 0.26 0.50 2.80 1.76 2.52 4.89 M g i n m — 2.00(-12) 5.24(-13)i> 7.76(-14) — 0.44 0.15 0.30: — 1.63 0.56 1.11: “Line 1: flux in erg cm-2 sec-1; 2: ratio wrt H/?; 3: corrected ratio Observation made on day 353 Figure 7: Light curve of curve Lightof 7: Figure Log L(H/I) (erg/sac) n n n m 2 K0 2.4 o t dy sns outburst) sines (days t Log including a least square fit of the data. the of including square fit least a 2.6 2.8 3 30 then the intensity of H/3 has a linear dependence on Ne, and both have the same

time dependence. The deduced slope of -3.84, is a somewhat faster decline than that

corresponding to a free expansion of the shell, t-3.

2.4.2 Temperatures and Densities

We use two procedures to calculate densities and temperatures for the nebular phase.

In the first case, we use [0 III] 4959,5007, [0 III] 4363, and H/?. In the second case, we use both optical [0 III] lines and the ultraviolet doublet [O III] 1660,1666. This will give us a check on the derived values. This check is necessary since we do not have simultaneous observations in the optical and ultraviolet, and the scarcity of the data may introduce large errors during the interpolation. The techniques used to calculate line intensities as a function of the level populations, the density, and the temperature of the gas are described by Osterbrock (1989).

Fig. 8 shows the observed [0 III]/H/3 intensity ratio as a function of time. The figure suggests that at about day 450-500, the XD level em itting the AA4959,5007 lines reaches its critical density, where collisional excitation and radiative deexcitation are equal. Thus, we can expect that by t2=680 days, the density of 0 ++ was well below its critical value of N ^t ~ 7 x 105 cm-3, so collisional de-excitation is negligible. On the other hand, at ti =235 days, we can assume that the levels are thermalized, since

N > Ncr.t and LTE can be assumed for the level populations.

A detailed balance equation lets us write the ratio of the densities in the XD2 and

3P i i2 levels giving rise to AA4959,5007. Then, we can easily o b ta in the intensity ratios for this blend with respect to H/? in the high (at day tj), and in the low (at day t2), 4363)/I(H£). iue : vlto, f 0 I] ie. h upr uv crepns o h inten­ III] the to R=I([0 corresponds curve to corresponds III]4959,5007)/I(H/3);curve upper lowerthe The R=I([0 lines. III]ratio [0 sity Evolution, of 8: Figure

Log R in 0 o in 1 2 Z •L. 2.4 2.8 2.3 2.3 2.8 2.4 o t(as ie outburst) sines (days Log t [OUI]A4363/H/f 3.2 3 3 [OII1JX5007/H/? 3.4 32

33

density limits. If we assume that the electron temperature is constant throughout the

interval (which is not a bad assumption considering the observational uncertainties),

we can take the ratio between both expressions to obtain a relation between density

and temperature at time ti. This is given by

Ne{ti)/\JrJ^ j = 7128.7 X j4959,5007(^2)/i49S9,5007(^l) • (2-3)

A second relation between Ne(ti) and Te(ti), can be obtained by calculating the populations of the 1So and *D2 levels of 0 ++, which form the line at A4363, and then forming the intensity ratio j(4363)/j(4959,5007). We have used the solution to the five- level atom, using the atomic parameters from Mendoza (1982). The complete solution is quite involved, but it is only function of the electron density and temperature,

J 4363 / ,74959,5007 = /[JV,(t,),r,(i,)]. (2.4)

Both equations should allow us to determine the temperature and the density, at time ti, and the density at any other time can be obtained from the constraint imposed by H (3, nam ely,

N e(t2) / N e ( h ) = (t2/* i)-3-84 . (2.5)

The results are summarized in Table 5, for both values of the color excesses.

The critical density of [0 III] 4959,5007 occurs at about day 470, as suggested by inspection of Fig. 8. 34

Another way to estimate densities and temperatures is by using the set of [0 III] lines at AA1660,1666, A4363 and AA4959,5007. The first two lines originate in the

5S2~3P i ,2 levels of 0 ++. When we form the ratios j( 1660,1666)/j(4959,5007) and j (4363)/j (4959,5007), we obtain expressions dependent on the electron density and temperature, and both parameters can in principle be found. In this case, however, we do not have simultaneous UV and optical observations, and an approximate solution can be only obtained by interpolating in the data.

Another important source of error, especially for the UV lines, are the uncertainties in E(B-V). To see this, we plotted curves of equal intensity ratio in the plane log Ne- log T e (Figure 9). The curves correspond to values of j(1660,6)/j(4959,5007) obtained by applying color excesses of 0.60+0.06 (dot-dashed line), and 0.60-0.06 (dashed line).

We include a single curve for j(4363)/j(5007) since this ratio changes very little with reddening. From the plot, we estimate that the uncertainty in the electron density is about 0.5xl06 cm-3 and, in the temperature, about 2,000 K, due to reddening. The results obtained from this method are also included in Table 5.

2.4.3 Chemical Composition

For the most part, the methods and atomic parameters needed to derive the abun­ dances have been taken from Osterbrock (1989). Electron densities and temperatures are also used as calculated in the previous subsection. The errors in the abundances are estimated assuming that the uncertainty in the temperature is ~ 2,000 K. 33

[0IH]M363/[0nife5a07 day 233 [oai]xi 663/[om]xsao7

\

day 340 o»

6 6.4 6.6 7 Log Ne

Figure 9: Electron temperature vs. density from [OIII] lines. Full lines corre­ spond to A4363/AA4959,5007 and an E(B-V)=0.60. Dashed lines correspond to AA1660,1666/AA4959,5007 and E(B-V)=0.60-0.06. Dot-dashed lines are for the same ratio, but E(B-V)=0.60+0.06. The intersection of the full line with the other two gives the errors in the measurement of density and temperature due to a 10% uncertainty in the reddening. 36

Table 5: Electron Temperatures and Densities

day Ne(cm~3) r .( K ) Ne(cm 3) r.(K ) E(B-V) = 0.45 E(B - V) = 0.60 a) lines 4363, 4861, 4959/5007 253 7.7(6) 13,200 7.6(6) 13,200 11 11 331 2.8(6) 2.7(6) 11 11 340 2.5(6) 2.4(6) 11 11 681 1.7(5) 1.7(5) b) lines 1660/1666, 4363,4959/5007 253 1.6(6) 23,300 3.0(6) 21,900 331 3.3(6) 11,900 4.0(6) 12,700 340 2.7(6) 12,400 2.2(6) 15,800 681 3.2(4) 16,000 1.0(4): 18,300

Helium The abundances of single and double ionized helium can be obtained from the intensities of He 15876 and He II4686 relative to H/?, respectively. Assuming pure radiative recombination, the density of helium with respect to hydrogen can be easily obtained by assuming that He/H = He+/H ++ He++/H +, and the ion abundances are given by

j{HeIhm) N(He+) aHeI(T) 4861 j(H/3) ~ N(H+) am {T) 5876

j(//e/74686) N(He++) aHeII{T) 4861 j(H(3) N(H+) aHp(T) 4686 1 ' where c * a ( T ) is the recombination coefficient for the specified line. The ratio of re­ combination coefficients is not very sensitive to temperature (Peimbert 1975).

However, the assumption stated above would make us overestimate the abundance of helium. Collisional excitation from the metastable 2s 3S level of He I can be impor- 37

Table 6: Abundance of Helium

day He+/H+ He++/H+ He/H no coll. coll. 235 .074 .050 .027 .077 309 .074 .050 .033 .083 413 .074 .050 .037 .087 631 .104 .071 .043 .114 680 .074 .051 .045 .096 1011 .059 .041 .052 .093

tant and it will contribute to the observed intensity of recombination lines like A5876.

Clegg (1987) provides simple formulae to account for this effect. The correct abun­ dance of He+/H+ is obtained by multiplying the abundance derived from equation

(2.6) by the factor (1+C /R )-1, where C /R is the ratio of excitations by collisions to recombinations. The abundances of ionized helium, both uncorrected and corrected by collisions, and the total abundances are listed in Table 6.

The resulting average abundance by number is He/H=0.092±0.009, which is al­ most coincident with cosmic abundances.

Oxygen The line ratio j(4959,5007)/j(H/?) can be used to obtain the number density of double ionized oxygen relative to ionized hydrogen, 0 ++/H +. Since He and 0 both have about the same ionization potentials, we assume that 0 ++/He+ ~ 0/He, and, by using the He abundances just obtained, we obtain the 0/H abundance ratios listed in Table 7. An average of these results gives O/H=(1.29±0.75)xl0-3. The probable error is mainly influenced by the uncertainties in the temperature. 38

Table 7: Abundances of Oxygen, Nitrogen and Neon

day 0++/H+ O/H N+/H+ 0+/H+ N/H Ne/O 235 6.68(-4) 1.03(-3) ——— 0.038 309 6.89(-4) 1.14(-3) 4.38(-5) 0.62(-5) 4.94(-3) 0.044 413 7.55(-4) 1.31(-3) 5.56(-5) 2.70(-5) 1.61(-3) 0.046 631 7.55(-4) 1.21(-3) 5.93(-5) 1.02(-5) 4.43(-3) 0.020: 680 7.49(-4) 1.41(-3) 4.25(-5) 0.65(-5) 4.93(-3) 0.030 1011 7-21 )(-4) 1.63(-3) 3.77(-5) 0.28(-5) 9.68(-3) —

N itrogen The nitrogen abundance can be expressed in terms of the ionic abun­

dances as

(28) H H+ 0+ ’ '

where the oxygen abundance can be expressed as

0 /H = (0 + + 0 ++)/H +. (2.9)

The ion abundances are obtained from the emissivity ratios j([N II] 5755)/j(H 0), for N+/H+, j([0 III] 5007)/j(H/?), for 0++/H+, and j([0 II] 7330)/j(H/?), for 0+/H+.

Note that in doing this, we are assuming that 0 + has the same temperature as

0 ++. The observational errors here might be somewhat larger since [O II] 7330 is a

relatively weak line. The average is N/H=(5.1±4.2)x 10-3, which is over 50 times the

solar abundance (Table 7).

Neon The most useful line is [Ne III] 3869, and N e/0~ Ne++/0 ++ can be assumed,

again using the [0 III] 5007 line. Results are collected in the final column of Table 7. 39

Table 8: Abundances in PW Vul

Element PW Vul Sun PW /Sun H e/H 0.09 0.10 0.9 C /H 4.0(-4) 4.7(-4) 0.8 N /H 5.1(-3) 9.8(-5) 52.0 O /H 1.3(-3) 8.3(-4) 1.6 N e/H 4.8(-5) 1.0(-4) 0.5

In forming the average, we gave half weight to the uncertain observation at day 631.

The average is Ne/0~0.037, or Ne/H=(4.5±2.6)xl0-5, i.e., about half the solar abundance.

Carbon The ratio of ultraviolet lines j([C III] 1907,1909)/j([0 III] 1660,1666) give the ionic abundance C++/0 ++ ~ C/O. Table 4 shows the line ratios. For a tem­ perature of 13,200 K, C/O=0.45 at day 330, and C/O=0.17 at day 681. Using the previously derived oxygen abundance, we find an almost solar abundance of

C/H=(4.0±2.3) x 10-4.

Sum m ary of Abundances In Table 8 we collect all average abundances and com­ pare them to accepted solar values. It can be seen that the ejecta has approximately solar composition, with the exception of nitrogen, which is enhanced even including the observational errors. The oxygen abundance might be a little enhanced.

This enhancement can be understood in terms of the mechanism proposed by

Starrfield and collaborators, in which carbon from the white dwarf’s core is mixed up by convection onto the accreted envelope, and converted into nitrogen by runaway 40 proton capture (see, e.g., Starrfield, Sparks, and Truran 1986). The modest over­ abundance in nitrogen, and possibly oxygen, appears to be consistent with a very slow nova.

2.5 The Coronal Line Region

The optical and IUE spectra suggest that the continuum spectrum is predominantly free-free emission. The absence of a strong Balmer jump and the strength of the continuum relative to H/3 are clear indications that the gas producing the continuum is very hot. We computed the observed continuum fluxes in the form f„ (erg cm - 2 sec-1 Hz-1), from the equivalent widths and intensities of several ultraviolet lines

(see Table 9), for day 330. Also, we added infrared fluxes for day 319 published by

Gehrz et al. (1988). Fig. 10 shows the observed continuum intensities relative to the intensity of H/? (in units of Hz-1), for day 330.

To account for the observed continuum, we first try a fit with the theoretical predictions from nebular theory. To do this, we assume that the observed continuum flux f„ (erg sec-1 cm - 3 Hz-1) can be expressed as the sum of the continuum fluxes of the cooler “nebular” gas and the hotter “coronal” gas. Both fluxes, relative to the intensity of H/?, can be expressed in terms of their respective continuous emission coefficients, electron densities and volumes of the emitting region. Assuming that H j3 is formed in the nebular region, they are given by,

fv ,C O T corl'cor'y^.cor

I Hi.J ( N e N + )neb Vneb Airj h[) 41

x" t I

to I

(Q I 13.5 14 1 5 Log v

Figure lOt The continuum spectrum of PW Vul relative to H/3. Dots represent observed continuum intensities. The dotted line includes free-free and bound-free emission by H and He at 10,000 K (nebular component). The upper curve includes hot coronal gas (10® K) as described in the text. A blackbody continuum at 690 K was also added to fit the infrared excess to the left. 42

Table 9: Continuum Fluxes at day 330

A“ /„ 1240 8.53 -15) 1639 9.18 -15) 1663 4.89 -15) 1749 1.96 -14) 1909 1.39 -14) 2798 1.94 -14) 3133 1.57 -14) 5550 1.42 -14) 2.3 1.59 -14) 3.6 2.56 -14) 8.7 1.01 -13) 1 0 .0 8.73 -14) 11.4 7.87 -14) “Last five entries in microns

fv,neb _ 7fi/,neb /n i i \ i m ~ ‘ Ferland (1980) calculated the continuous emission coefficients of H and He as a function of electron temperature. The lower curve of Fig. 10 shows the resulting neb­ ular phase continuous emission relative to the H/3 emission coefficient, from a mixture of H and 10% He (as calculated in section 2.4), at Te = 10 4 K. This theoretical contin­ uum includes both free-free and bound-free transitions. For the optical continuum we used the continuous coefficients tabulated by Osterbrock (1989), since they include more accurate Gaunt factors. We can see from the figure, that the continuum emis­ sion from the “nebular” phase is too weak to explain the observed free-free spectrum, and it also predicts a strong Balmer jump, which is not observed. This suggests that 43 the nebular gas represents some fraction of the total mass, and the rest is at a sub­ stantially higher temperature. The hot gas radiation is mainly a free-free continuum and X-ray lines, with very little recombination or collisionally excited line emission in the optical.

We now estimate the contribution to the continuum from very hot gas. We adopted a combination of H and He continua at T=10 6 K, as tabulated by Fer- land (1980). Let us assume that the observed continuous flux can be expressed as the sum of the fluxes of the coronal and the nebular line regions. Then, dividing the observed relative density flux by Equation 2.11, we obtain,

f v ,o b a//?) _ J ^ c o r ^ u ,c o r ^ y t y

~1u,neb I ■£'ne6'7t/,ne6 where the emission measure E~N* V. The left-hand side of this equation can be obtained from Fig. 10, where we choose to fit the optical region (A5550). On the right side, we obtain the ratio of continuous coefficients, 7 , from Ferland’s (1980) tabulations, and, then, we can solve for the ratio of emission measures. This gives the amount of hot gas to add to the nebular gas to fit the observed spectrum, as the equation above indicates. We obtained a factor of about 5.5, and the resulting spectrum is shown as a solid line in Fig. 10. The fit, specially the Balmer jump, is greatly improved, except for an infrared excess, which can be attributed to thermal emission from dust in the shell (Gehrz et ol. 1988), which, as mentioned in section 2 . 1 , started to develop by day 280. In Fig. 10, we fit the infrared continuum with a blackbody at 690 K. Since this excess is still small, the temperature is only a crude 44 estim ate.

We can now use the ratio of emission measures to estimate the mass of coronal gas. Recalling that the mass is given by M=/ieNeV, we obtain,

/V2 V N M a r T l F ' = ArC°V°r = 5-50» (2.13) neb Vneb N nebMneb w here fie is the electron molecular weight.

If the nebular and coronal line regions are in pressure equilibrium, the ratio of electron densities equals the inverse ratio of the temperatures, and therefore, Mcor ~

550 x Mne6. If, rather, they have the same density, then Mcor ~ 5.5 x Mne 6 - In either case, we see that a surprisingly large mass is in the coronal line region. The mass of nebular phase gas can also be calculated from the luminosity emitted in H/3 and the nebular electron density, which we obtained in the previous subsection. Then, at day

330, Lup ~2.71xl034 erg sec-1, and Nne6=(3.4±0.5)xl0 6 cm-3, the nebular mass is

Mneb ~ L m = (1.1 ± 0.3) x 1029 g ; (2.14) WneblHP that is, the mass of nebular gas is ~ 5.3 xlO - 5 M®, and, therefore, the total mass of the ejecta lies in the range 3xlO- 4-3 x lO -2 M®.

Finally, we can use these estimates to evaluate the energetics of the ejecta. As­ suming an expansion velocity of 600 km/sec, the kinetic energy is of the order of

1 .1 x 1045 to 1.1 x 1047 ergs, by day 330. 45

2.6 A Model of the Ejecta

2.6.1 Model Parameters

We used a photoionization model of the ejecta to reproduce the observed spectrum as a check on our abundances. We chose to match the optical data for day 413, for which we felt the abundances and observations were best estimated, and made a rough interpolation for some of the available ultraviolet lines. The model was calculated using the photoionization code CLOUDY (Version 72: Ferland 1989).

The model consists of an expanding spherical shell, which has a hydrogen density of 6.3x10s cm"3, at day 413. Based on an expansion velocity of 600 km/sec, as discussed above, we adopted a radius of 2xl0 15 cm for day 413. For the continuum, we chose a lum inosity of 6.3 x 10 36 erg/sec and shape given by optically thin hydrogen bremsstrahlung at a temperature of 2xl0 6 K. These numbers were chosen to give an acceptable match of the observed helium ionization and the luminosity of H/3.

This luminosity is also consistent with the observed brightness of the optical free-free continuum, assuming full covering of the nebula.

The metal abundances had to be increased by about 20% from those given in

Table 8 to have a reasonable match of the ultraviolet spectrum. The only exception was carbon, which had to be adjusted by a substantially larger factor. We should remember that our derivation of carbon abundance is based on data from only two days and, therefore, highly uncertain. The set of abundances we used is as follows, as a fraction of solar abundances: He, 0.9; C, 2.4; N, 60; O, 1. 8 ; all other elements,

1 .2 times solar. 46

2.6.2 Comparison with the Observations

In Table 10, we present a comparison between the predictions of this model and the observed intensities. It can be seen that, in most cases, they agree within a factor of 2 or less, which we find reasonable within the uncertainties discussed in previous sections.

Silicon depletion? We can use these results to further investigate the problem of dust production already mentioned. If dust has been formed in the ejecta, we expect the observed intensities of the silicon lines to be lower than those predicted by our model, since these atoms are incorporated in the dust grains. On the other hand, the intensity of a line such as Mg II 2798 will not show signs of depletion, since this element does not play a significant role in forming dust.

Our data fail to show Si III] 1892 at all dates considered here. For dates close to our model calculation, the detection limit of this line, is of the order of 2 x l 0 -14 erg cm- 2 sec- 1 (2cr), which implies that the intensity of Si III] 1892 relative to H/3 is < 0.12, corrected for reddening. Our model, predicts a relative intensity of 0.24

(Table 10), for a nearly solar abundance. The observed relative intensity of Mg II

2798, interpolated for day 413, is about 0.6, while our model predicts a ratio of 0.4.

Both figures are comparable within the observational errors. From these results, we infer that silicon might be depleted by at least a factor of 2. This depletion factor might be consistent with the small production rate reported by Gehrz et al. (1988). Table 10: A M odel for Day 413

Line I(A)/I(H/?) Observed Model H/3a 34.2 34.2 H a — 2.78 [NII\ 6584 — 8.69 Ha + [Nil] 9.62 11.47 He 11640 3.: 2 .2 He 75876 0 .1 0 0.09 He ii me 0.42 0.41 C III] 1909 7: 9.5 CIV 1549 27.5: 17.4 [Nil] 5755 1.4 0.9 N III] 1750 1 0 .: 13.9 NIV] 1486 21.5: 14.6 NV 1240 32.: 32.3 [01] 6300 0 .2 0.4 [0II] 7325 1.6 0 .2 [ 0 1 11] 1660 + 1666 3.: 1 .2 [0 III]4363 1.8 1.1 [ 0 1 11] 4959 + 5007 2 2 .0 2 2 .0 0 IV] 1402 3. : 0 .8 [Ne III] 3869 1.1 1 .2 M g i n m 0 .6 0.4 [Fe VII] 6087 0 .2 0 .1 [FeX] 6374 0.24 0 .1

< Te [0 III] > K 13,200 12,700 < Ne > cm~3 1.14(6) 7.1(5) “This entry is Log Lhp in erg/sec High Excitation Lines In previous work on other novae (e.g., Nova Cygni 1975;

Ferland, Lambert, and Woodman 1977), it has been shown that the presence of high excitation lines is a clear indication of the existence of a hot coronal line region. A useful line for this purpose is [Fe X] 6374, which is blended with [0 I] 6363, but it is possible to separate both contributions by comparing the intensity of this blend with the neighboring line [0 I] 6300. The ratio of transition probabilities for both

[0 I] lines is A(6363)/A(6300)=0.32 (Osterbrock, 1989). At days 73, 309, and 413, the reddening corrected ratios were 1.20, 0.54, and 0.57, which implies that [Fe X]

6374/[0 I] 6300~0.88, 0.22, and 0.24, respectively.

As indicated in Table 10, the model predicts that, at day 413, this ratio should be about 0.1 in the warm component of the ejecta. This is somewhat lower than the observed value. We conclude that these ions have become too weak in the coronal line region at this point of the outburst, and the observations reflect their presence in both the nebular and the coronal gas.

2.7 Discussion and Summary

In this final section, we summarize the topics presented in this chapter and discuss some of their implications.

The distance to the nova is of the order of 3 kpc, which is within the range of previously determined measures.

Ultraviolet lines of [O III] can be used in conjunction with the optical lines to obtain electron temperatures and densities. The results are uncertain due to errors in the reddening and in the temporal interpolation of the data, since the optical and 49 ultraviolet observations are not simultaneous.

We obtained a fairly reliable set of abundances for the ejecta of PW Vul. The only element with a significantly higher-than-solar abundance is nitrogen. Some re­ searchers have reported anomalous abundances for some specific elements in other novae (e.g., neon; Gehrz, Grasdalen, and Hackwell 1985), a phenomenon that has been interpreted as a consequence of the evolutionary state of the progenitor star at the time of the outburst (Starrfield, Sparks, and Truran 1986). In this case, the high nitrogen abundance could be a by-product of hydrogen burning through the CNO cycle. The nearly solar abundances of C, 0, and Ne indicates that this star did not go as far in the burning cycle at the time of the outburst. This is consistent with a very slow nova.

The ejecta has a hot (T~ 106 K) component where lines of highly ionized species and the free-free continuum originate, and a warm (T~ 10 4 K) component which give rise to the recombination lines. Free-free emission from the hot component is the main contribution to the optical and ultraviolet continuum, and may be the source of heating of the ejecta during the early decline.

A third component to the observed continuum is characterized by thermal emis­ sion, presumably from dust. Silicon might be depleted for at least a factor of two over solar abundances, providing further evidence for the presence of this material. CHAPTER III

Nova QU Vulpeculae 1984: The Cold Gas Phase

In the previous chapter, we found that the method of measuring abundances from line ratios can lead to fairly reliable estimates, as shown by our model of the nebular line region. In this study of Nova QU Vul 1984, we will use m ore extensively th e method of the “ionization correction factors”, introduced by Torres-Peimbert and Peimbert

(1977), which proved to be a powerful tool in planetary nebula research.

In the next section we describe the history of the 1984 outburst as well as some of its observational properties. The spectroscopic material is discussed in Section 3.2.

The reddening and distance to QU Vul are evaluated in Section 3.3, and the physical conditions in the shell are discussed in Sections 3.4.

3.1 Overview of the Outburst

QU Vul was discovered by P.Collins (IAU Circular 4023), at visual magnitude 6 .8 , on

1984 December 22.13 UT. A precise position given by Klemola (IAU Circular 4024), implies that the nova is at b11 ~ —6 °. The maximum apparent magnitude of +5.6m occurred approximately on 1984 December 25.0 UT, which we take as day 0 of the outburst. Note, however, that a single visual observation reports a value as high as

5.1 on December 25.4 (T.Yusa, IAU Circular 4024).

50 Figure 11 shows a light curve for the first 60 days of the outburst, from observa­ tions reported in several IAU Circulars. The time to decline 2 mag. from maximum, usually taken as a measure of the “speed class” of the nova, is about 25 days, which corresponds to an intermediate type of classical nova (Payne-Gaposchkin 1957). An­ other visual light curve covering the first three years of the outburst is presented in

Figure 12. This curve includes the average of amateur observations reported in the monthly AAVSO Circulars, and it is not corrected for interstellar reddening.

An ultraviolet light curve, compiled by Austin et al. (1990) from IUE spectra, is also displayed in Fig. 12, where each point represents the uncorrected integrated flux between 1200 and 3300 A.

The light curves do not show any signature of dust formation, such as DQ Her’s deep minimum (see, e.g., Martin 1989). However, grain formation has been detected in the infrared from a photometric study by Gehrz et al. (1986). Their observations of 1 0 - and 2 0 -/xm features have been interpreted as originating from silicon grains, with a mass Ms,- > 10- 8 M®. A search in the millimeter wave band by Shore and

Braine (1991), failed to detect any molecular material surrounding the nova.

Taylor et al. (1987) presented a radio map of the nova at 14.9 GHz taken on day

497 of the outburst. The image shows a non-spherical profile of about 0.23x0.13”.

They also report that an early radio outburst occurred between 100 and 300 days, preceding the normal radio emission from the shell. To explain their observations, they suggest that a shock front developed by the interaction between a high velocity wind from the nova and the principal ejecta. 52

O -o a • • *5o» o 2 o a •f • 0 5

10

t (days from maximum)

Figure 11: The first 2 months: a light curve of QU Vul, from visual estimates reported in IAU Circulars. 53

® ® n □ □ O- - i* ° 0 ~®> \ -3.6 .-10 u u 01 N -0.83 uE wOJ ,-lt >

-3.9 Visual observations UV integrated flux Hpflux

100 1000 t (days)

Figure 12: The first 3 years. Visual observations are averages of amateur estimates published by the AAVSO. Least-squares fits to the data and the respective slopes are included. 54

3.2 Observational Material

3.2.1 Optical Spectra

The analysis is based on new and archival spectra taken also by Dr. Wagner with the

same instrument described in chapter II.

The first optical spectrum was obtained on day 145, and the star has been routinely

observed since that time. The I RAF spectroscopic reduction program has been used

to measure the lines and continua. Whenever possible, the lines were deblended by

fitting either simple or multiple gaussians. The weaker lines were measured by direct

integration under the line.

A sample of selected spectra is shown in Figure 13. The spectra show a rich vari­

ety of forbidden and recombination lines. The typical nebular lines, [0 III] 4959,5007,

[Ne III] 3869,3968, [Ne IV] 4720 are the most prominent features. The Balmer spec­

tru m of H, and several He I and II lines are also present in emission.

The recombination line, N II 5680, is also observed and we will use it in order to

derive the electron temperature in the shell. The strongest identified lines and their

fluxes relative to H/3 are listed in Table 1 1 . T he fluxes are corrected for a reddening

of E b - v = 0.61, as calculated in the next section. The observed intensities of H/3 are

listed in Table 16 (see p. 78).

In Table 12, we list the average of the observed fluxes for th e weakest lines, and the number of spectra where the line is observed. Note that some highly ionized species, such as [Fe XI] 7892, [Ar XIV] 4413, and others might be present. The presence of

[Fe X] 6374 can be inferred from the anomalous ratio of [0 I] 6363/[0 I] 6300. Often, scale follows the flux of the strongest line. strongest the of flux the follows scale Figure 13: Sample spectra of QU Vul. (a) Day 145-155; (b) Day 1064. Note that the that Note 1064. Day (b)145-155; Day (a) QU Vul. spectraof Sample 13: Figure Fjk (10-,a erg cm"* s"' A”') 50 00 50 00 50 00 50 00 50 8000 7500 7000 6500 6000 5500 5000 4500 4000 3500 H - H -f-h I-1 -f-h H 1 - 1 1-1 H I-lJ-f Day 1064 1064 Day a 145-155 Day 1-llftU Wavelength (A) | 1 1 1 | 1 I | •, X f++4 £

Table 11: O ptical Lines®’6

Day H I H I H I He I He I He I He I He II He II 6563 4340 4102 4471 5876 6678 7065 4686 5411 (1 ) (2 ) (3) 145 - 0 .6 6 0.36 0.16 --- 0.30 -

155 5.30 --- 0.40 0.115 0.26 - - 258 3.20 0.54 0.55 0.16 0.34 0.088 0 .2 0 0.26 0.024

311 - 0.61 0.57 0.15 0.32 -- 0.34 0.053 481 4.61 0.50 0.52 0.16 0.40 0.118 0.25 0.46 0.046 530 2 .8 6 0.50 0.54 0.14 0.25 0.057 0.14 0.49 0.055 560 3.26 0.52 0.45 0.17 0.24 0.076 0.14 0.51 0.059 561 2.74 0.48 0.51 0.16 0 .2 2 0.061 0 .1 1 0.50 0.045 582 2.50 0.49 0.55 0.14 0 .2 2 0.050 0 .1 0 0.53 0.054 614 1.01 0.50 0.53 0.18 0.24 0.064 0 .1 2 0.56 0.053

638 - 0.45 0.47 0 .2 2 0 .2 0 -- 0 .6 6 0.053 648 2.80 0.45 0.58 0.18 0.23 0.057 0 .1 1 0.64 0.051

652 3.36 0.46 0.53 0.16 0.23 - 0.14 0.59 0.056

655 2.78 --- 0 .2 2 0.058 0 .1 2 -- 820 3.43 0.44 0.48 0.23 0.23 0.068 0 .1 1 0 .6 6 0.079 821 2.58 0.46 0.52 0.23 0 .2 1 0.051 0.08 0.73 0.065 860 3.38 0.42 0.49 0.16 0 .2 2 0.070 0 .1 0 0 .6 6 0.074 880 2.85 0.50 0.48 0 .2 1 0 .2 0 0.056 0.08 0.75 0.065 916 3.47 0.46 0.52 0.25 0 .2 0 0.073 0 .1 0 0.70 0.074

917 - 0.63 - 0.24 --- 0.79 - 978 3.73 0.47 0.47 0.19 0 .2 2 0.069 0.09 0.67 0.066

1033 - 0.57 0.40 0.23 0.25 -- 0.65 - 1063 5.60 0.26 0.28 0 .2 1 0.29 0.125 0.16 0.58 0.052 1064 3.74 0.40 0.44 0 .2 1 0 .2 2 0.073 0.08 0 .6 6 0.054 “ Notes: numbers below the wavelength indicate possible blends. (1) [N II] 6548,6584; (2) He II 6683; (3) [Fe XV] 7060. 6 Entries are observed intensities relative to H/3, corrected for reddening. 57

Table 11 (continued): Optical Lines®

Day N II [N II] N III 0 I [0 I] [0 I] [0 II] [0 III] [0 III] 5680 5755 4640 8446 6300 6363 7325 4363 5007 (4) (5) (6 ) 145 0.064 0.109 0.30 - --- 0.29 1.50

155 0.079 0.135 -- 0.150 0.086 0.34 -- 258 0.052 0.083 0.46 0.082 0.105 0.087 0 .1 0 0.76 2.54

311 0.060 0.115 0.46 ---- 1.06 3.96

481 0.046 0.208 0.55 - 0.119 0.113 0.36 1.45 7.34 530 0.057 0.141 0.54 0.017 0.079 0.061 0.24 1.60 8.75 560 0.069 0.150 0.57 0.016 0.082 0.071 0.24 1.54 9.66 561 0.059 0.138 0.54 0.016 0.078 0.062 0 .2 0 1.58 9.56 582 0.059 0.148 0.52 0 .0 1 0 0.072 0.055 0 .2 0 1.43 9.97

614 0.059 0.143 0.55 - 0.078 0.053 0 .2 2 1.34 11.16

638 0.127 0.132 ----- 1.17 13.73

648 0.060 0.134 0.59 - 0.081 0.068 0 .2 2 1.41 13.38

652 0.063 0.138 0.53 - 0.085 0.074 0.31 1.27 13.81

655 ------0.25 --

820 0.064 0 .1 2 1 0.51 - 0.080 0.067 0.28 0.90 19.76

821 0.071 0.130 0.58 - 0.060 0.040 0 .2 0 0.98 19.73

860 0.058 0.113 0.49 - 0.072 0.050 0.24 0.82 2 1 .8 6

880 0.055 0.105 0.61 - 0.052 0.025 0 .2 0 0.74 2 2 .1 2

916 0.041 0.087 0.55 - 0.061 0.040 0 .2 1 0.63 21.35

917 - - 0.57 ---- 0.63 23.09

978 0.056 0.075 0.52 - 0.048 0.039 0.19 0.42 20.93

1033 -- 0.52 —--- 0.37 20.24 1063 0.044 0.054 0.40 0.070 0.068 0.077 0.34 0.28 20.61

1064 0.044 0.051 0.50 - 0.033 0.046 0.19 0.38 17.56 ® Notes: (4) C III 4649; (5) [Fe X] 6374; ( 6 ) [0 III] 4959. T able 11 (continued): O ptical Lines'*

Day [Ne III] Ne IV] Mg I [S I] [Ar III] [Fe VII] [Fe VII] 3869 4720 4571 7727 7135 5720 6087 (7) (8 ) (9)

145 2.47 0.08 0 .2 0 2 - - -

155 ------

258 9.67 0.49 0.136 ----

311 12.79 0 .8 8 0 .1 1 1 - - - - 481 19.62 1.38 0.061 0.246 0.054 0.082 0.118 530 23.37 1 .6 6 0.043 0.205 0.029 0.055 0.109 560 22.16 1.56 0.067 0.225 - 0.057 0.098 561 24.17 1.49 0.042 0.197 0.028 0.052 0.089 582 23.68 1.52 0.038 0.147 0.028 0.063 0 .1 0 : 614 20.60 1.52 0.049 0.169 0.044 0.067 0.130

638 2 1 .8 6 1.48 --- 0.096 - 648 24.32 1.58 0.046 0.203 0.039 0.076 0.135

652 22.29 1.53 - 0.181 0.059 0.073 0.149

655 ---- 0.044 - 0.124 820 17.79 1 .2 2 0.059 0.197 0.050 0.096 0.156 821 20.85 1.29 0.024 0.136 0.038 0.090 0.134

860 19.73 1 .0 2 - 0.180 0.050 0.084 -

880 18.28 1 .0 0 - 0.090 0.043 0.072 0.113

916 16.98 0.74 0.031 0.196 - 0.050 -

917 - 0.85 0.030 - ---

978 13.26 0.56 - 0.139 0.051 0.039 0.073

1033 10.09 0.53 ---- 0.040

1063 3.94 0.41 - 0.088 0.094 0.034 -

1064 8.89 0.47 - 0.078 0.041 0 .0 2 1 0.046 <* Notes: (7) [Ne III] 3968; ( 8 ) 0 III 4569; [Ca V] 6087. 59

Table 12: Weaker Optical Lines

Identification < fu/fw3 > N He I 4026 3 8 x l0 -3 6 0 II 4415-17, [Ar XIV] 4413 28 5 He 14517, N III 4511-34 50 12 N V 4604, [Fe III] 4607 51 5 ? (A5470) 13 9 [Ca VII] 5615, [Fe VI] 5631 26 11 [Fe III] 6133 27 7 He II 6407 ? 18 5 C III 6744 8 7 He II 6891 12 9 [Ar V] 7007 14 8 He II 7595 21 14 [Fe XI] 7892 17 2 ? (A7926 - 35) 45 15 He II 8237, N II 8242 52 12

the presence of these lines are taken as a signature of hot coronal gas in nova shells

(cf. C hapter IV).

Typical observational errors in the line fluxes have been estimated at about 5 to

10% for the strongest lines, and as large as 40% for the weakest lines listed in Table 12

(Wagner 1986). Various tests of error propagation show that the observational errors

make a contribution no larger than a factor of two to the total abundances (see

Section 3.4).

The Helium Lines: He I 4471 is particularly strong in our spectra, and in several spectra as strong as He I 5876. From Case B calculations, the former line is expected to be much weaker. In order to check for possible blends, we compare the He I 60 spectrum to theoretical predictions presented by Osterbrock (1989).

Since the helium line ratios are not strongly temperature sensitive, we adopt values of T e ~ 104 K, and Ne ~ 10 6 cm"3. Then, the predicted ratios are j„(He 14471)/]^, (He

1 5876)= 0.37, and j„(He I 6678)/jt,(He I 5876)= 0.29. Both ratios and the observed values as a function of time are included in Fig. 14.

The A6678/A5876 ratio closely follows the predicted value, and A4471/A5876 pro­ gressively departs from it. A check of the wavelengths as a function of time, also shows a shift of the A4471 line towards shorter wavelengths at later times, while

A5876 remains relatively steady at all times. The measured blueshifts are 12 and

2 A, respectively. We conclude that there is clear indication for another emission line blended with He I 4471. We have found no suitable candidate.

3.2.2 Ultraviolet Spectra

The UV spectra have been obtained from the IUE archives and reduced by using the

IUE Regional Data Analysis Facility (RDAF). Several spectra taken within the first

20 days of the outburst show typical P-Cygni profiles indicating that the expanding atmosphere is still optically thick. These profiles are still present, but very weak, in the spectra of days 126 and 158. Since we are concerned with the late development of the shell, we defer analysis of the earliest spectra to another paper.

A journal of the IUE observations is presented in Table 13 to facilitate archival analysis. A gallery of selected spectra are presented in Figures 15 and 16.

The first spectrum in our sample, a long exposure taken on day 126 (Fig. 15(a)), shows many weak lines, whereas the strongest features are strongly saturated, such as lines the predicted ratios at constanttemperature. at ratios predicted the lines iue 4 Eouino h eIlns Tedt ersn bevdvle,ad the and values, dotsrepresentobserved The lines.IHe the of Evolution 14: Figure

H « I R a d o 1 H« I <1 H* I p. . n n e s M71A337S). OtMVad MB,gha l M7B), am h g O O r c 1000 61

62

Table 13: Journal of IUE Observations t-t0 Date Tim e Image # Time days yr/day UT min:sec Long Wavelength Prime 126 85/119 03:17 LWP 5851 1 0 :0 0 158 85/151 22:34 LWP 6102 0 0 :2 0 182 85/175 15:42 LWP 6265 00:30 191 85/183 23:45 LWP 6328 1 0 :0 0 204 85/197 18:15 LWP 6409 0 1 :0 0 309 85/302 11:47 LWP 7015 0 2 :0 0 340 85/334 05:15 LWP 7200 0 2 :1 0 462 86/090 15:48 LWP 7926 03:00 554 86/182 21:31 LWP 8525 0 1 :0 0 627 86/255 00:56 LWP 9058 06:00 Short Wavelength Prime 126 85/119 02:56 SWP 25805 15:00 158 85/151 21:52 SWP 26047 07:30 182 85/175 15:30 SWP 26245 08:00 191 85/184 00:16 SWP 26341 30:00 204 85/197 18:27 SWP 26428 15:00 309 85/302 11:55 SWP 26999 11:40 340 85/334 04:54 SWP 27191 14:00 462 86/090 15:12 SWP 28069 2 2 :0 0 554 86/182 20:49 SWP 28587 26:00 627 86/255 0 0 :1 0 SWP 29196 38:00 63

t— i— i— r “1 i i— i— i— |— i— i— i— i— [— i— i— i— i— |— i— p

Zxlff" a) Day 126, Observed

1 0 "

o «

uE - i | I i i- 1 1 I I I—I—| -h i- I—i- 1 I i i i ; VE>

1 2x10" b) Corrected, E(B-V)=0.61

1000 1500 2000 2500 3000 3500

Figure 15: Ultraviolet spectrum of QU Vul at day 126 (April 30, 1985): (a) observed spectrum, stowing the depression centered at 2200 A, due to interstellar absorption; (b) same spectrum corrected for reddening. Note that the feature has virtually diss- apeared. 64

a) Day 158

o

b) Day 191

o

1000 1500 2000 2500 3000 3500 X

Figure 16: A gallery of ultraviolet spectra: (a) day 158, (b) day 191, (c) day 204, (d) day 340, (e) day 462, (f) day 627. Note that the scale is the same in all panels to show the evolution of the emission lines. Figure 16 (continued) Figure 16(continued) Figure Flux (erg cm'* sac ') 2xlff" 2xt0” Iff" nr" nr" 0 1000 o

i H 1 - I 1 l M | I - | I h | -I + I I | M I 'l '1 -I I- I 1 -i—H 1 1500 » « » I

____ I ____ I « I I » 2000

0 Day 627 0 e) Day 462e) ____ X I ____ « « « I L 2500

1 « 1 1 « 3000

_____ 1 ____ I ____ L. 3500 66 Mg II 2798. The exposure time, in this and the other spectra were chosen to obtain accurate fluxes for the weaker lines and, therefore, the Mg II line is overexposed at all times. The P Cygni profiles are rapidly disappearing at this time, indicating that the nova shell is becoming optically thin. The continuum is still strong, but rapidly declining, as can be seen by comparing it to another long exposure taken on day 191

(Fig. 16.b).

All spectra show a rich variety of permitted, intercombination, and forbidden transitions, of both high and low ionization stages (see the other panels in Figure 16).

Some of these transitions are blended in our low resolution spectra, which complicates the analysis of the shell’s evolution. From Table 14, which lists the strongest UV lines, we note that the highly ionized species tend to appear later and become stronger with time. This increase of the ionization with time is typical of classical nova shells

(Gallagher and Starrfield 1978, Hauschildt et al. 1992).

Several strong neon lines in various ionization stages are also present, which imme­ diately suggested that the outburst occurred on an ONeMg white dwarf (Starrfield,

1986). T he blend at A3346, is particularly strong after day 400, and it is due to

[Ne III] 3343 and [Ne V] 3346 lines. Unfortunately, the other component of the [Ne

V] doublet, at X 3426, was out of the range of both optical and ultraviolet spectra.

On the other hand, although C IV 1549 is present, it is very weak throughout this period compared, for instance, to Mg II 2798. This situation is reversed in PW

Vul, a slow CO nova, which suggests that carbon might not be enhanced, and gives further support to the idea that the progenitor was an ONeMg white dwarf rather 68

Table 14: Ultraviolet Lines0

Day He II C II C III] C IV N III] N IV] [NV] 0 I 1640 1335 1909 1549 1750 1486 1240 1305

126 1.63 -- 1.15 2.25 - 4.07 6.09

158 1.25 0.90 0.63 1.77 2.29 - 3.95 2.83

182 0.92 - 0.86 0.84 2.50 0.60 3.66 1.94 191 1.87 0.44 0.84 1.07 2.29 2.66 2.17 1.61 204 1.82 0.32 1.38 1.02 2.92 2.44 2.02 1.40

309 2.48 - 1.16 1.17 2.93 3.87 2.79 1.00

340 2.51 - 0.96 1.54 3.05 4.82 3.82 -

462 -- 1.19 2.18 3.19 6.56 9.16 -

554 3.57 - 0.89 1.86 2.27 5.39 10.30 - 627 3.98 - 0.70 1.76 1.74 4.89 10.55 - ° Entries are observed intensities relative to H (3, corrected for reddening

Day 0 III [0 III] 0 IV [Ne III] [Ne IV] [Ne V] [Ne V] 3132 1663 1402 1815 1602 1575 3346 126 - 0.92 3.86 0.88 --- 158 - 1.22 1.83 0.88 - - - 182 - 1.62 1.08 1.11 - - - 191 0.85 1.93 2.46 0.84 1.67 - 0.90 204 0.95 1.95 1.99 1.36 1.69 - 1.56 309 1.21 2.03 3.44 1.43 3.39 - 2.68 340 1.46 1.96 3.90 1.01 3.98 0.56 3.37 462 1.77 1.96 5.01 1.95 7.55 1.74 11.27 554 1.67 1.36 4.33 2.08 7.02 1.16 11.89 627 1.63 1.39 3.80 1.64 6.82 1.49 16.79 69

Table 14 (continued): Ultraviolet Lines

Day Mg II (Mg V] Si III] A1 II] A1 III 2798 2928 1892 2669 1860 126 - - 1.95 0.54 2.42 158 17.74 - 2.39 1.62 2.13 182 16.96 - 2.38 1.48 2.05 191 -- 1.77 0.94 1.72 204 11.59 - 2.22 1.58 2.21 309 11.53 - 2.68 1.51 1.94 340 11.49 0.50 1.67 1.57 1.42 462 11.34 0.86 1.41 1.17 1.91 554 9.78 - 1.04 0.39 - 627 7.87 0.82 0.70 0.76 1.22

than a CO star.

3.2.3 Interpolation of Spectra

Unfortunately, simultaneous observations in different wavelength ranges are rarely available. It is thus necessary to interpolate fluxes in order to construct composite spectra.

In Fig. 12, least-square fits to portions of the light curves have been included together with the respective slopes and 1-sigma standard deviations. The slopes for the UV integrated fluxes are: -1.2±0.1 and ~3.6±0.1, for early and late dates, respectively. For the H/3 fluxes, they are: -0.8±0.2, and -3.9±0.2. 3.3 Reddening and Distance

The flux ratio He II A 1640/He II A 4686 has been tabulated by Seaton (1978), includ­ ing radiative and collisional excitation processes for conditions of radiative recombi­ nation. Since this ratio does not have a strong dependence on the physical conditions in the nebula (i.e., temperature and density), we can assume a predicted ratio for typical conditions in the shell and compare it to the observed ratio. This comparison will then yield the interstellar reddening towards the nova.

We assume typical values for a relatively young nova shell of 104 K and 106 cm-3, respectively, for which a flux ratio j(He II 1640)/j(He II 4686)~6.8 is predicted. The observed values can be obtained from Tables 11 and 14, and the ratio as a function of time is shown in Fig. 17, where we have interpolated the optical data for the dates of the UV observations. As expected, the ratio is fairly constant except for the first two data points, which is probably due to internal reddening. Dust has indeed been observed to form a few months after the explosion (Gehrz et al. 1986).

Following Seaton (1979), we derived the color excess by applying his interstellar extinction curves, and obtained an average E(B-V)=0.61±0.10, between days 204 and 627. The error includes the uncertainty due to the a priori unknown physical conditions in the shell.

A check on this value can be obtained from the 2200A graphite feature in the ul­ traviolet extinction curve (Seaton 1979). A fortunate combination of low noise level in a relatively long exposure and a still strong ultraviolet continuum on a spectrum taken on day 126, clearly shows the effect of this feature (see Fig. 15.a). The depres- Ha II 1640/He II 4686 Figure 17: Observed He II 1640/He H 4686, useful for reddening correction. reddening for useful 4686, H 1640/He II He Observed 17: Figure .6 .8 .4 0 1 100 200 300 I I (days) 400 0 600 500 700 71 sion is completely removed when we apply the above color excess, as shown in the

lower panel of Fig. 15. As mentioned before, the Mg II 2798 line is highly saturated

on account of the long exposure, as are many of the strongest lines in this spectrum.

The distance to the nova can be estimated from the luminosity- “speed class”

relationship. From the visual light curve discussed in Section 3.1, and the relationship

given by Warner (1989), we obtain My = —8.0 ±0.9, and a distance of 3.5 ± 1.5 kpc,

assuming a visual absorption, A y ~ 1.9m. Gehrz, Grasdalen and Hackwell (1985),

using somewhat different numbers but the same method, obtain a distance of 3 kpc,

which we will adopt.

Another method for distance determinations, namely, by measuring the reddening

in field stars around the nova, could not be used in this case, since all-sky photometric

catalogs provide insufficient stars in this area for a reliable determination (an applica­

tion of this method is presented by Schild 1976). Careful uvby photometry in a small field around novae would provide with more accurate, object-independent estimates of the distance.

3.4 The Nebular Region

In previous studies of classical novae (e.g., Grasdalen and Joyce 1980, Ferland, Lam­

bert and Woodman 1977, Greenhouse et al. 1988), evidence has been found for the existence of at least two regions in nova ejecta, the “coronal-line” and the “nebular- line” regions.

The coronal-line region is a hot (T~ 106K), gas that produces an almost flat optical-ultraviolet continuum, showing a very weak Balmer jump (Ferland, Lambert 73 and Woodman, 1986). In contrast, the nebular-line region is produced by a cooler

(T~ 104K), higher density gas, which radiates most of the collisionally excited and recombination lines. As we will show in the next section, its continuum is predicted to have a strong Balmer jump.

In this section we discuss the physical characteristics of the nebular-line region, using the standard techniques of nebular physics.

3.4.1 Electron Densities and Temperatures

To calculate the chemical composition of the shell, we first need to determine its electron density and temperature as a function of time. As mentioned before, the optical data have been interpolated to match the UV data, so our study will cover the period from 126 to 627 days.

A good temperature indicator can be obtained by forming the ratio between a recombination line and an ultraviolet collisionally excited line of the same ion in the next lower ionization state. UV lines have, in general, very high transition probabili­ ties, and at this late point in the outburst the electron density is assumed to be low enough for further collisional deexcitation to be negligible.

The strongest recombination line we consistently observe in the data is N II 5680, which we will use in conjunction with N III] 1750, a collisionally excited intercombi­ nation line. The respective emission coefficients are given by

47rjrec( AT/75680) = NeN(N ++) a'Jf {Te) x hvrec (3.1)

4irjcoit{NIII]m0) = NeN(N ++) (uii/ui) Q 21 e~h,/lkTe x hvcou (3.2) 74 where aeJf (Te) is the total recombination coefficient for the multiplet, Q2l = 8.63 x

10~6 (Cl/u>2) T~os, is the collisional deexcitation rate, ui2 and u>i are the statistical weights of the upper and lower levels, respectively, and fl is the collisional strength.

The other symbols have their usual meaning. Values for the collisional strengths have been obtained from Mendoza (1983).

Wilkes et al. (1981) provide an expression for the radiative part of the effective recombination coefficient of N II 5680 as a function of temperature, based on an earlier study by Burgess and Seaton (1960). They show that

<*56 so = 5-2 x 10-13te-°'84 cmV1, (3.3) where te = Te/10,000 K. Often, dielectronic recombination plays an important role in the recombination coefficient. However, a check of the values calculated by Nuss- baumer and Storey (1984) shows that the contribution is negligible in this case. There­ fore, the ratio of emission coefficients yields,

;'5680/ii750 = 5.49 x 10-6 *e-°-34 e8-23/‘« (3.4)

The results are listed in Table 15. The electron temperature maintains a fairly constant level (te ~ 1.00 ± 0.08) throughout the period of the observations. The uncertainty in the temperature has been estimated from the error in the reddening given in the previous section.

In order to estimate the electron density, we use the [0 III] 4363/[0 III] 4959,5007 ratio (a function of both Ne and Te). Since 0 ++ and N++ have similar ionization potentials, we can assume that they are located in nearly coincident zones within 75

Table 15: Electron density and temperature

Time Te{K) Ne (c m '3) 158 9,400 1.2(7) 182 9,600 1.1(7) 191 9,600 1.1(7) 204 9,900 1.1(7) 309 10,000 1.2(7) 340 10,100 1.1(7) 462 10,400 8.2(6) 554 9,600 8.3(6) 627 9,500 5.0(6)

the shell, and therefore, we can use the temperature derived above from the nitrogen lines. The results are also presented in Table 15.

We can check these results from the temporal behavior of some other emission lines. The light curve of [0 III] 4363, as can be seen in Fig. 18, rises almost linearly and reaches a maximum by day 530. When this plateau occurs, the electron density has fallen below the critical density of the XS level, and the gas is efficiently radiating through this transition. From the equations of statistical equilibrium (cf. Osterbrock

1989), the critical density of the 2S level occurs at an intensity equal to 1/2 the intensity when the gas is collisionless. By examining Fig. 18 we conclude that the critical density of [0 III] 4363, Nc ~ 3 x 107 cm-3, was reached by day 230.

A similar analysis for the AA4959,5007 doublet shows that the electron density was probably about 7 x 105 cm-3 on day 600. In this case, however, it is not certain that the emission has reached a maximum due to its proximity to the end of the available observations, and therefore, this is only a lower limit to the electron density. ratio I([0 III] 4959,5007)/I(H/3); the lower curve corresponds to I([0 HE] 4363)/I(H/3). I([0 intensity to the corresponds to curve lower the corresponds curve upper The 4959,5007)/I(H/3); III] lines. ID] [0 I([0 the of ratio Evolution 18: Figure

Ralatlva flux • - - . ------0 MjM363/ty £0 0 [o tnF£oo7/ty • o • —, J ! ------0 •

1 , 0 ---- •

t (dayi) t _L._ .1. 1 1 I 1 J 1 1 I 1 1 1 o r • ••• m --- • 1--- 1 -- 0 L L °

77

The results shown in Table 15 suggest that the electron density maintains a some­ what constant level for the period 150-350 days, although the uncertainty in these calculations is expected to be large, especially at the earlier times. It is possible, however, that high densities are maintained by continuous ejection of matter up to day 350. Indeed, a radio study of QU Vul by Taylor et al. (1987) suggests that a high velocity wind existed in addition to the principal ejecta which might account for this effect on the density. This wind would have declined in intensity after day 300, when the density began to fall. As discussed below (see discussion on helium abundance), the wind may have its origin in steady hydrogen burning in the remnant’s surface layer past the peak of the outburst.

Additional evidence for this hypothesis comes from the light curve of H/?. The observed and reddening corrected fluxes of H/3 are listed in Table 16. Figure 12 also shows that H0 declines monotonically with time, as expected for a matter-bounded

(i.e., fully ionized) photoionized shell. The dependence of the electron density with time can be deduced from the light curve of H/3 assuming both that the gas is fully ionized and that the mass of the shell is constant. If these conditions apply, then

In (t) « M s x Ne(t) a e" ( T e) oc r \ (3.5) where M5 is the mass of the ejecta, assumed constant, a^p is the recombination coefficient of H/?, corrected for reddening, and the last proportionality is the time dependence of H/3, which we can evaluate by means of a least squares fit to the observed intensities. The resulting slope, s=-3.9, for t—to > 350 days, is shown as a fit to H/3 in Fig. 12. A slope of 3 in the decline of the electron density is to be Table 16: Intensity and Luminosity of H/?

Day I H0a L H0T 145.25 6.93E-11 4.74(34) 258.25 5.29E-11 3.62(34) 311.40 3.32E-11 2.27(34) 481.44 1.39E-11 9.50(33) 530.37 1.40E-11 9.57(33) 560.44 1.04E-11 7.11(33) 561.42 1.23E-11 8.41(33) 582.40 1.13E-11 7.73(33) 614.40 1.02E-11 6.97(33) 638.19 6.04E-12 4.13(33) 648.21 5.91E-12 4.04(33) 652.21 6.81E-12 4.66(33) 820.50 3.84E-12 2.63(33) 821.49 2.78E-12 1.90(33) 860.44 2.08E-12 1.42(33) 880.36 1.88E-12 1.29(33) 916.43 1.56E-12 1.07(33) 917.40 1.85E-12 1.26(33) 978.20 1.57E-12 1.07(33) 1033.24 1.13E-12 7.72(32) 1063.20 4.11E-13 2.81(32) 1064.12 9.44E-13 6.45(32) ° Observed Intensity (erg cm-2 sec-1 ) 6 Luminosity (erg sec-1) corrected for reddening 79 expected for a freely expanding shell, which seems to be the case during most of the period. From the figure, we can see, however, that the decline seems to be slower at earlier times, suggesting that early in the outburst the conditions were not exactly those of a freely expanding shell with constant mass. Therefore, it seems reasonable to interpret these observations as evidence for a wind.

In summary, the electron temperature in the nebular region shows an approxi­ mately constant value of 104 K. The electron density is subject to a larger uncer­ tainty, but diiferent methods seem to indicate that it shows a definite decline only after day 350. From our first observation and until day 350, the density was approx­ imately constant at slightly over 107 cm-3. Next, we will use these results to derive the abundances in the nebular region.

3.4.2 Abundances

The determination of abundances in classical novae can have significant uncertainties due to the inhomogeneous nature of their shells (Williams et al. 1991). We have carefully chosen pairs of lines that are expected to be produced in regions within the shell with nearly the same physical conditions, so that their ratio will be independent of the particular local conditions.

For each line, the ion number densities are calculated by applying the equations of statistical equilibrium to a five-level model of the atom, with collision strengths obtained from Mendoza (1983). Total abundances are then estimated by, first, cal­ culating ion abundances relative to H+ from line fluxes relative to H/3, and, then, by applying the appropriate “ionization correction factor” (hereafter, ICF) to account for the unobserved stages of ionization. An earlier discussion of this concept can be found in Torres-Peimbert and Peimbert (1977).

The ICF is formed by assuming that the ratio of the abundances of two ions with similar ionization potentials is equal to the ratio of their total abundances. Then, the total elemental abundance relative to hydrogen is proportional to the ionic abundance relative to H+, where the factor of proportionality is the ICF. This approach minimizes the uncertainties due to fluctuations in the density and the ionization structure.

In general, we expect our abundances to have relatively large errors. At the elec­ tron densities considered here (~ 107cm-3), most forbidden optical line ratios are sensitive to the electron density, which is not well determined. On the other hand, many permitted UV lines have transition probabilities that are high enough for col­ lisions to be negligible, and, therefore, they are linearly dependent on the density.

Then, the ratio of two of these lines is independent of density. Unfortunately, ultra­ violet lines are more sensitive to uncertainties in the reddening and the temperature.

An additional source of error comes from the assumptions involved in the ICF.

Since the distribution of the gas within the shell is not known a priori, we assume that ions with similar ionization potentials occupy the same physical region within the shell. By taking ratios we expect that the uncertainties due to inhomogeneities in the shell will be minimized.

Next, we estimate the abundances in the ejecta of QU Vul. All abundances are given by number, relative to hydrogen, and where appropriate, relative to the solar composition. A summary of our results is presented in Table 17 (see p. 86). 81

H elium The recombination lines He I 5876 and He II 4686 can be used to estimate

the relative abundances of He+ and He++. The former, however, should be first

corrected by collisional excitations from the metastable 2s 3 S level of He I. Expressions for these corrections are given by Clegg (1987) in the form of a ratio of collisions to

recombinations, C/R. The corrected He I abundance is obtained by multiplying the

uncorrected value by the factor (1+C/R)-1.

Neglecting the unobserved fraction of neutral helium, which is probably small because of the high ionization of the shell, we can obtain the total helium abundance by adding up its two higher stages of ionization,

(3 .6 ) H H+ i.e., a factor of 5 above the accepted helium cosmic abundance. This result is the average abundance obtained from the line fluxes measured on the optical spectra between days 258 and 652. Since the ratio of two recombination lines is not very sensitive to the physical conditions in the shell, we expect any systematic errors to be small.

The high abundance of helium seems to imply a mechanism for continuous pro­ duction past the peak of the outburst, such as hydrogen burning in the remnant’s envelope. Other possibility could be the enrichment of the envelope by mixing in a helium-rich white dwarf.

Next, we discuss the abundances of heavy elements. The first term in each equa­ tion below is the number density of the respective element relative to hydrogen, while the last factor of the second term is the ICF as described above. In each equation, 82 we also include the average abundance obtained from our data at various dates.

N itro g e n

N _ jV+ (5755) + jV*"*-(5680) H t , , v1n-a H H* tfe+(5876) ~ * * W+(5755) 4- AT++(1750) He H+ He+(5876) ' ’

Equation 3.7 can be somewhat uncertain if the abundance of neutral nitrogen were large. However, given the presence of highly ionized species, we expect the abundance of neutral elements to be small. The second equation is included only as a check on our calculations since N III] 1750 and N II 5680 have been used to calculate the temperature.

The average abundances are 47.5 and 39.5 times solar, respectively, with probable errors of about a factor of 2. The ICF is of the order of 1 for all measured abundances, and thus the errors introduced by this correction are expected to be small.

O x y g en

_ 0^(1402) _ JV _ H + JV+3(1486) 1 1

In Equation 3.9 we also assume that the neutral abundances of oxygen and helium are small, so the error introduced by their different lower-stage ionization potentials should be small. The higher ionization potentials of oxygen and helium are almost identical, and we expect this equation to be a good indicator of the total oxygen

abundance.

The assumption that N /0 ~ N+/0 +, commonly used in planetary nebulae, might

not work as well in the nova environment. First, the intensities of [N II] 5755 and especially [0 II] 7325 are small, and therefore the observational errors are larger.

Second, the difference between the higher ionization potentials of N and 0 is about 6 eV, which is a significant fraction of the range between their lower and higher stages.

These ions, then, coexist in a very thin region within the shell, so they can poorly reflect the total ratio of abundances in the shell.

The next two equations above involve the 0 IV 1402 line which is most certainly blended in our low resolution spectra with Si IV 1397. They are, however, useful in further constraining the abundance obtained from Equations 3.9.

From Equation 3.9 we obtain a total oxygen abundance of 4 times solar, with errors better than a factor of 2. Equations 3.10 and 3.11 yield upper limits to the abundances of less than 22 and 29 times solar, respectively.

C a rb o n

C _ C++(1909) w N onc .. ,n_4 x t v , , 7 , - 7 ;; ~ 3.05 x 10"4 (3.12) H H+ JV++(l750)

In this case, each equation represents the ratio of abundances in consecutive zones within the shell. The ionization ranges of the triply ionized atoms are remarkably similar, and we expect Equation 3.12 to be a good abundance indicator. In the second equation, the difference amounts to 13 eV, but at this point there is probably little carbon left in the next ionization stage, so the errors involved in this assumption must not be large.

The results show an excellent agreement: 0.84 and 0.76 times solar, respectively.

No major blends are present in these strong lines, and the ICF are also near unity, so the total error, including errors in the observed fluxes, reddening and temperature, is estimated to be of a factor of 1.5. The good agreement gives us further confidence in the ICF method of abundance calculations.

Neon

Ne _ Ne++(3869) ^ 0 H H+ x 0++(5007) ~ 3-8 x 10 (3-14) _ iVe++(1815) „ C _ _ 1ft_2 x < 1.6 x 10-2 (3.15) H+ C7++(1549) N e++(1815)+iVe+3(1602) N ^ 1T in_2 M < 1.7x10_2(3.16) H+ iV+3( 1486)+iV+4 (1240)

Equation 3.14 involves another standard assumption in nebular physics. We should note, however, that the ionization potentials of 0 ++ and Ne++ differ by about

10 eV, and, since the shell is highly ionized, we expect that a considerable fraction of Ne++ might exist in the 0 +3 zone. This equation gives an abundance of about 31 times solar, with errors of about a factor of 2 or 3.

The next two equations involve [Ne III] 1815, which is probably blended with

Mg VI 1806 (radiative-equilibrium model calculations to be presented in the next chapter show that this line should be strong). We can, then, give only upper limits to the neon abundance. The results show that the neon abundance is less than about 85

130 times solar.

Magnesium, Silicon and Argon

M g = M ^(29_28)'S '! _ NNe e 94vin_4 2.4 x 1(T4 (3.17) H H+ Ae+4( 1575) Si S'i++(1892) N 3 H H+ x N+(5755) ~ 4J x 10 (3-18) Ar Ar++(7135) N (3.19) H H+ AT++(5680) - ‘ = Ar^TOOT) x He 10_6 H+ i/e++(4686) v ’

Equation 3.17 involves higher ionization species, and therefore we are less certain about the temperature in that zone. Equation 3.18 has similar problems as the

N+/0 + ratio, namely, a very narrow ionization range, and the error is expected to be large. The last two equations involve weak argon lines, and also are uncertain.

These equations yield a magnesium abundance of 6 times solar, within a factor of

2; a silicon abundance of 133 times solar, and an argon abundance of 0.7 times solar, with errors of factors of 3 or 4.

Sum m ary The results are presented in Table 17, where a “quality grade” has been introduced to indicate the degree of uncertainty in the measurement. Roughly, a letter “a” indicates abundances which are expected to be better than a factor of 2.

A letter “b” corresponds to measurements with errors ranging from a factor of 2 to

4. Finally, all upper limits have been graded with a letter “c”. All abundances are given by number, relative to hydrogen, and where appropriate, relative to the solar composition. 8 6

Table 17: Abundances in QU Vul

z Optical lines U V lines Equation i m \ ' Qb Equation [Z/HY Q b He 6 5.1 a --

C -- 13 0.84 a

C -- 14 0.76 a N 7 47.6 a 15 39.5 a 0 8 4.0 a 16 < 22.4 c

0 -- 17 < 29.5 c Ne 10 31 c 18 < 131 c

Ne -- 19 < 140 c

Mg -- 20 6.2 b

Ar 11 > 0.3 c -- Ar 12 0.7 b --

Si - - 21 133 c “ Abundance by number relative to the Sun b Quality of the measurement: a, good; b, poor; c, very poor

To summarize our results, the ejected shell of QU Vul shows an enhancement of

N, O, Ne, and Mg by factors of 44, 10, 60, and 6 with respect to the Sun. Carbon and argon, on the other hand, are nearly solar. These averages have been calculated by arbitrarily giving a weight of 0.2 to the abundances showing only an upper limit.

The observed abundances in QU Vul are qualitatively in line with the picture of mixing of core material of an oxygen-neon-magnesium white dwarf into the accreted envelope (Starrfield, Sparks, and Truran 1986). The low abundance of argon could be an indicator of the mass of the white dwarf (Starrfield et al. 1992). Finally, note the relatively high abundance of helium, which could have been produced by steady hydrogen burning in the envelope or also from mixing from the core.

In the next chapter we will proceed to investigate the nature of the hot gas phase. C H A P T E R IV

Nova QU Vul 1984: The Hot Gas Phase

The first suggestion that a hot coronal line region is present in classical novae was put forward by Grasdalen and Joyce (1976). They interpreted the appearance of bright coronal lines in the infrared spectrum of Nova Cygni 1975 (V1500 Cyg), as produced in a region at ~ 106 K. Ferland, Lambert and Woodman (1977, 1986) discuss the presence of coronal line emission in the optical, concluding that the coronal line region is an important source of ionizing radiation. QU Vul also showed infrared coronal lines (Greenhouse et al. 1988), which have been interpreted as produced by photoionization in a low density hot gas. Greenhouse and collaborators also estimated the mass of this coronal line region, and we will discuss their results in connection with our study.

Little else is known about the properties and physical conditions in the coronal line region, in part due to its very nature. The emission lines in novae are broadened by the expansion of the shell, and, therefore, line blends make it difficult to study the weaker optical coronal lines. This has been less of a problem in the infrared where the coronal emission is much brighter and sometimes dominant. The continuum emission from the coronal phase, on the other hand, is expected to dominate in the ultraviolet region and beyond, and a goal of this chapter is to use the information contained

87 88 in the UV to sort out the contribution to the spectrum from the various sources of radiation present.

In the next section, we will use the expansion velocities to obtain the size of the shell. Then, we will use direct calculations and photoionization modeling to constrain the physical conditions in the hot gas phase, namely, the electron temperature, the density, and the mass of the ejecta.

4.1 Expansion Velocities

Expansion velocities from line widths have been reported by various authors since early in the outburst. Andrillat (1985, IAUC 4026), and Lyons and Bolton (1985,

IAUC 4027), reported two velocity systems from absorption features, V/ ~ —650 km/sec, and V// ~ —1380 km/sec, between days 5 and 15.

Higher velocities were also observed a few months later. Ogelman and Krautter

(1985, IAUC 4065), reported that P Cygni profiles on a strong continuum were seen in the ultraviolet, with terminal velocities of ~5000 km/sec. Gehrz, Grasdalen and

Hackwell (1985, IAUC 4065), observed a very bright Ne II emission at 12.8 fim, whose width implies an expansion velocity of ~2500 km/sec. Greenhouse et al. (1988), using a radio study by Taylor et al. (1987), derived an expansion velocity of ~ 752 km/sec in the coronal line region.

In Fig. 19, we show the velocities from selected emission lines in our spectra, as a function of time. We can see that the widths of the [0 III] 1663 and N V 1240 followed similar declines, from about 1600 km/sec, around day 200 to about 1000 km/sec, after day 350. H/3, on the contrary, remained nearly constant at about 674±37 km/sec. 89

Also, the average for He II 1640 is 836±52 km/sec, and for He II 4686, is 737±43 km/sec. [O III] 5007 did not follow the behavior of the UV line, staying at 752±48 km /sec.

Therefore, it seems clear that at least three kinematically distinct systems exist in the shell, moving at about 700, 1200, and 2500 km/sec. For day 554, the implied radii are: R/ ~ 3.4 x 1015 cm, R// ~ 5.7 x 1015 cm, and R/// ~ 1.2 x 1016 cm.

4.2 Temperature of the Hot Gas

The absence of a strong Balmer jump in the continuous spectrum is a clear signature that the gas must be at a high temperature. In fact, the depth of the Balmer jump is a function of temperature and we can use this information to measure this parameter.

The observed spectrum contains, however, a contribution from the cold gas phase.

Since we have already measured its temperature from emission-line diagnostics, we can subtract its continuous emission from the observed spectrum, and the remaining emission may thus be attributed to the hot gas phase. The emission coefficient for a gas at Te = 104 K has been computed by Ferland (1980). Therefore, the observed continuous emission luminosity relative to the line luminosity of H/9, which is assumed to form in the cold phase, can be expressed as:

vL„ { Ehot v-)v(hot) ^

Lhp obs coid Ecold4irjH0(cold) ’ where i/ju and 4?rj//0 are the emission coefficients of the continuum and H/?, respec­ tively, in erg cm3 sec-1, and E=NeN+V, is the emission measure for each gas phase, where V is the volume. Figure 19: Velocity fields from various emission lines, as a function of time: time: of function a as lines, emission various from fields Velocity 19: Figure circles

Velocity (km/sec) 1400 1000 1600 1200 600 800 N V 1240; V N , • o • 200 closed circles 400 [ 3 1663; D3] [0 , t (days) 600 squares: ■ H/3. 800 1000 open 90

91

Table 18: The Continuous Spectrum of QU Vul

A vLvj Lhp Obs. Cold Hot 1500 30.0 0.3 29.7 1800 19.1 1.5 17.6 2500 21.4 3.3 18.1 3300 19.1 12.6 6.5 3500 18.8 12.6 6.2 4100 10.9 5.5 5.4 4700 7.6 2.9 4.7 5400 4.6 4.3 ~ 0 6800 4.9 6.1 ~ 0

In order to subtract the cold phase from the observed spectrum, we have averaged

portions of the data at day 554 around the wavelengths listed by Ferland (1980). In

Table 18, we list these average fluxes relative to H/9, in dimensionless units. In the

third column, we include the predicted relative continuum emission coefficient for a

mixture of hydrogen and 50% helium (as obtained in the previous chapter), at T=104

K. The subtracted spectrum, shown in the last column, indicates that the Balmer jump is virtually absent. The depth of the Balmer jump, which we measure as the

ratio i//1,(3500)/t'/„(4100), is listed in Table 19, as a function of temperature. From

Table 18 and 19, we obtain a crude estimate of the temperature of 1.7xl05 K. This is only an order of magnitude estimate, since we are trying to measure a small effect in a noisy spectrum.

A lower limit to the temperature of the hot gas can be obtained from the observed emission from H/?, which we assume to be formed by the sum of the intensities emitted 92

Table 19: Emission from the Hot P h a se

T (x l0 4 K 7(3646+)/7(3646-) ^ Ot(1500)/I^ 1 18.10 1.00 2 4.57 5.30 5 1.79 23.2 10 1.30 52.3 20 1.11 144. 50 1.04 288. 100 1.02 589.

by both gas phases. The emissivity of H/3 is very sensitive to temperature (Ferland

1980), and therefore, we expect that most of the line radiation is produced in the cold phase. The minimum temperature of the hot gas corresponds to the case in which half of the emission is originated in each phase. In this case we are also underestimating the cold phase emission.

The continuum spectrum at short wavelengths (e.g., A1500 A), is also sensitive to temperature, but not as strongly as H/?. Taking the ratio of the continuum flux at

A1500, i/f^, relative to the line intensity of H/3, we obtain

_ vfujhot) + vfy(cold) Ihp obs hp{hot) + iHp(cold) f IHp(cold)\ \ iHp(hot)) Ihp h o t where we have neglected the contribution to the continuum by the cold gas. The minimum temperature of the hot gas corresponds to I/^hot) = IW/g(cold) = 1/2

I///?(obs). From Table 18, we see that the observed continuum to line ratio is about 30, and therefore, the ratio produced by the hot gas is equal to 60. In the second

column of Table reftb:bjp, we list the continuum to line ratio predicted by Ferland

(1980) for a mixture of hydrogen and helium at various temperatures. A ratio of 60

implies that the minimum temperature of the hot phase is near 10s K, which is in good agreement with our previous estimate.

An upper limit to the temperature can be obtained from the theory of nova out­

burst. Bath and Shaviv (1976), and Gallagher and Starrfield (1976), show that the white dwarf decreases in radius at constant luminosity, roughly equal to the Ed-

dington luminosity, for several months to years after the outburst. The maximum

temperature that the star reaches will be given by its preoutburst radius, as

T _ Le u!Lq (4.2) Tq (Rw d/R q)2. where the Eddington luminosity, the white dwarf radius and its temperature are given in units of solar values. For a 1 Mg star, the temperature is of the order of 9xl05

K, and increases by about a factor of 2 for a 1.4 Mg white dwarf. The hottest the gas can be is the Compton temperature of the radiation field, which is roughly the photospheric temperature of the white dwarf. Therefore, the above estimate is an upper limit to the temperature of the gas.

To summarize our results, we find clear evidence that there is material surrounding

QU Vul which is at a substantially higher temperature than the material emitting most of the emission-line spectrum, as discussed in the previous chapter. Although the data do not allow for an accurate estimate, they suggest that the temperature lies in the range ~ 105 — 106 K. In the next section we will attempt to set limits to 94 the electron density in the hot gas phase.

4.3 The Density of the Hot Gas

In the previous section we showed that the continuous spectrum is mostly free-free emission from the hot gas, and that the ultraviolet emission from the cold phase is negligible. If the hot gas has a very low density, we expect it to be optically thin to the ionizing continuum, and the observed continuum would approach the stellar continuum. X-ray observations suggest that the stellar source is at a temperature of about 3x10s K (Ogelman, Krautter and Beuermann 1987). At these temperatures, a blackbody distribution peaks in the extreme UV to X-ray bands, and therefore, we would see the Rayleigh-Jeans tail of the distribution in the optical. As the density increases, the material becomes optically thick and the incident radiation would be reprocessed into bremsstrahlung, producing a fainter UV continuum.

In order to quantify these ideas, we have constructed a series of models using a radiative-equilibrium photoionization code, CLOUDY, which we have already de­ scribed in Chapter II. In this case, we have chosen a luminosity, L e

In Fig. 20 we show the predicted emergent spectrum for various hydrogen den­ sities, and for a temperature T bb — 5 x 10s K. This value of T bb is intermediate between the temperatures obtained above, and consistent with that suggested by X- ray observations (Ogelman, Krautter and Beuermann, 1987). Note that the scale is 95

Table 20: Observed Fluxes for Continuum Diagnostics

t — to n m / a (1250) fA(4861) vM125O)/IH0 ^ ( 4 8 6 1 ) / / ^ 158 6.75(-ll) 7.56(-14) 9.36(-14) 39.13 6.76 182 6.40(-ll) 1.11(-13) 8.86(-14) 60.63 6.71 191 6.27(-ll) 6.72(-14) 8.67(-14) 37.50 6.71 204 6.08(-ll) 5.18(-14) 8.40(-14) 29.75 6.71 309 3.41 (-11) 1.04(-14) 5.39(-14) 1 0 .6 6 7.68 340 2.99(-ll) 2.09(-14) 4.69(-14) 24.50 7.63 462 1.61(-11) 1.12(-14) 2.08(-14) 24.38 6.27 554 1 .1 2 (-1 1 ) 6.34(-15) 1.51(-14) 19.75 6.56 627 7.98(-12) 3.00(-15) 1.34(-14) 13.13 8.17

normalized to the continuum at H/3. We can see that the density must lie in the range

2xl04-10 6 cm - 3 to fit the continuum. At the higher densities, the gas becomes too cold, showing a weaker UV continuum and the stronger Balmer jump.

Another indication of the point at which the cold gas becomes too cold due to a high electron density, is given by the luminosity of H/3 predicted by our models. The observed luminosity for day 554 is hup ~ 8 x 1033 erg sec-1. In the example discussed above, with Tab ~ 5 x 10s K, models with electron densities larger than ~ 4 x 10s cm - 3 predict H/? luminosities larger than observed.

The procedure described above suggests a method of examining in more detail the relationship between the continuum “color”, i.e., the flux ratio f>( 1250)/fA(4861), the electron density and the temperature. In Table 20, we list the observed line intensity of H/3, and the continuum fluxes at 1250 and 4861 A, respectively. In the last two columns, we list the continuum fluxes relative to H/3, in dimensionless units and corrected for reddening. 96

10

1

2.0(4) = N, 1.0(5) 1.0(6 )

1 10” v (Hz)

Figure 21: The continuous spectrum of QU Vul normalized to the continuum flux at H (3. Closed circles , observed fluxes in regions assumed free of emission lines; open circles, fits to the continuum around emission lines. The errors are assumed to be similar for all points. Lines are calculated emergent continua, with densities as indicated. The model parameters are described in the text. 97

Table 21: Emission from the Hot Gas Phase

t-to vf„{ 1250)/IW (4861 )//#/? i//„ (1250)/i//„ (4861) 158 39.0 4.96 7.9 182 60.4 4.91 12.3 191 37.3 4.91 7.6 204 29.6 4.91 6 .0 309 10.5 5.88 1 .8 340 24.3 5.83 4.2 462 24.1 4.47 5.4 554 19.6 4.76 4.1 627 13.0 6.37 2 .0

Next, we need to subtract the cold gas phase. From Ferland’s (1980) tables, we obtain: 1/7 ,, (1250)/47t ja p = 0.30, and ^7l,(4861)/4ffj//Jg = 1.80, for gas at T = 1 0 4 K.

The resulting fluxes are listed in Table 21. The last column shows the UV to optical flux ratio, which we will use to constrain the density.

The photoionization model CLOUDY has been used in order to find the relation­ ship between the flux ratio and the electron density. The results are plotted in Fig. 21.

The dotted lines are the model predictions for various temperatures as indicated. The vertical solid lines delimit the area for which the line luminosity of H/3 is greater than the value shown. The dark gray area, therefore, indicates those densities which are excluded by the observations. In order to evaluate the errors due to the poorly deter­ mined distance, an additional boundary line has been added which corresponds to an order of magnitude increase in the luminosity of H/3. Finally, the middle horizontal solid line indicates the observed ratio for day 554, and the remaining solid lines, the upper and lower limits given by the observational errors. They have been obtained by setting limits to the noise in an area assumed to be free from line contamination.

From this graph, we obtain a probable density for the hot gas of 8.4 x 10 4 cm - 3 w ith

errors of the order of a factor of 4. This is substantially lower than the density of the

cold gas phase, and even if the distance was underestimated by as much as a factor of 3, the hot gas would still be one order of magnitude less dense than the cold gas.

To summarize, the shape of the continuum spectrum suggests that the gas has a

temperature Thot ~ 10s — 106 K, and a density Ne,/iot ~ 2 x 104 — 3 x 1 0 s cm-3. 99

10

’v star V

T- CO TfGO ? o in CVJ

2x10'

5x10 1x10'

10* 10* N, (cm'3)

Figure 21: Density diagnostics. Dotted lines: model predictions for various stellar temperatures. Vertical solid lines: H/? isoluminosities, labeled as Log hup (erg sec-1). Thus, the region at the right of the “34” contour curve corresponds to conditions in which models predict that L///j > 1 0 34 erg sec-1, which is larger than observed. Horizontal solid lines: observed continuum flux ratio (middle line), and its upper and lower limits. The light gray area delimits the region of densities and temperatures that are allowed by the observations. CHAPTER V

Nova QU Vul 1984: The Overall Shell Structure

In the previous chapters, we have examined each gas phase separately. Now, we will

combine the results obtained so far, in order to derive additional properties of the

shell, such as the mass, geometry, and energetics. Unfortunately, the hot gas has

been poorly constrained, so we have to derive the additional parameters for a grid of

possible hot gas phases, and then see which models offer more reasonable solutions.

In Table 22, we list the parameters adopted for our grid of photoionization models.

Each entry, except the last one, corresponds to one point within the light gray area in

Figure 21, so these are our acceptable models for the hot gas. The last entry shows the

values for the cold gas phase. The first two columns show the white dwarf temperature

and the gas density adopted as input data for the photoionization models. The next

column lists the predicted average electron temperature in the shell. The mass ratio of

both gas phases and the filling factors, listed in the remaining columns, are discussed in the following sections.

The model parameters used for these calculations are as follows: inner shell radius, R, = 10 15 cm; outer radius, R 0 = 7 x 10 15 cm; white dwarf luminosity,

Lb b = 2 x 1038 erg sec-1; filling factor, 1 for the hot phase. The abundances are the same as obtained in Chapter III. The remaining input parameters, the white dwarf

100 1 0 1

Table 22: Model Parameters and Predictions

Star Ejecta T b b N h Te N* T M///M c tcold Hot gas phase 1.0(6) 2.0(4) 3.8(5) 8(9) 920 2 .2 (-6 ) 3.0(4) 3.0(5) 9(9) 548 5.5(-6) 5.0(4) 2.5(5) 1 ( 1 0 ) 300 1.7(-5) 5.0(5) 6.0(4) 7.0(4) 4(9) 132 4.5(-5) 8.5(4) 6.3(4) 5(9) 89 9.6(-5) 1.4(5) 5.4(4) 8(9) 50 2.8(-4) 2(5) 2.0(5) 2.2(4) 4(9) 2 2 9.0(-4) 3.0(5) 3.1(4) 9(9) 18 1.7(-3) Cold gas phase ~ 1-0(7) 1.0(4) 1( 1 1 ) 1 —

temperature and the shell’s hydrogen density, are to be varied in our grid, and are listed in the first two columns of Table 22.

5.1 The Mass of the Ejecta

In order to find the mass of the ejecta, we will follow a procedure similar to that of

Section 2.5. Fig. 22 shows the observed continuum at day 554 (filled dots), where the reddening corrected flux (erg cm - 2 sec- 1 Hz-1) relative to H/? is plotted against frequency.

In Section 3.4 we measured the electron temperature in the nebular shell to be nearly 10 4 K. The combined free-free and free-bound emission from hydrogen and he­ lium at this temperature, predicted by scaling from the H/3 flux, is shown in Figure 22 as the dotted line. As discussed in the previous two sections, this cold continuum poorly reflects the observed levels, so we added the hot gas component, using the 102

v (Hz)

Figure 22: T he continuum spectrum of QU Vul relative to H/3. Dots represent ob­ served continuum fluxes. The dotted line includes free-free and free-bound emission by H and He at 104K (nebular component). The upper curve includes also the hot coronal gas at 106K as described in the text. 103

emission coefficients computed by Ferland (1980). The total observed flux relative to

H/? is given by

fi/,o b s 7 i/ Ekot''fv,hot(j'hot') /_ ... Ih/3 4xjnpeold E cotd4x jHp(cold)

where the emission coefficient j^hot is a function of t/lot=T/lot/10 6 K, and E hot is the

emission measure of the hot gas as previously defined. We have assumed here that

the recombination line H/3 originates in the cold component, thus canceling out the

ratio of emission measures in the first term of the sum above.

The ratio of emission measures provides us with an estimate of the amount of hot

gas needed to fit the level of the observed continuum. A good fit of the observations

is obtained with

= 3 C - (5-2) ■E'nefc

The result is shown in Figure 22, where the dotted line represents the cold phase on

the right-hand side of Equation 5.1. The addition of the hot phase with the factor given by Equation 5.2, assuming t/K,t=1.0, is shown as the solid line in Figure 22.

There may also be a possible additional contribution to the cold phase from the

2 -photon decay of the 2 2S level of H I (Osterbrock 1989). However, at the densities

calculated above, the 2 -photon emission represents a 1 0 % effect, at best, which is well

within the observational errors, and, thus, we neglect its contribution.

The ratio of coronal avid nebular masses can be obtained assuming that E~N 2 V

~ Ne M, where V and M are the volume and the mass of the emitting region, we obtain

Mcor = 3 x ^ t°Jrc o r* . (5.3) M n efc Ncor t 104

This implies that the coronal line region has a mass between 18 and 900 times the nebular mass, as shown in Table 22 (cf. p. 101).

The individual masses can be obtained from the luminosity of H 3, since

L h 0 ~ ( M neb /m p ) NneblHP- (5.4)

At day 554, Lnp = 7.7 x 1 0 33 erg/sec, Ne = 1 0 7 cm-3, and, thus, Mnei> ~ 5 x 1 0 - 6

Mq. From the arguments above, the total mass of the ejecta lies in the range 1.0 x 10- 4 to 4.5 x 10- 3 Mq. Based on an analysis of high ionization infrared lines, Greenhouse et al. (1988), obtained a lower limit to the coronal mass of 9 x 10 ~ 4 M 0 .

5.2 Energetics

5.2.1 Evidence for Pressure Equilibrium

The fact that the densities of the hot and cold phases are different suggests that they might be in pressure equilibrium, i.e.,

(N ekTe)hot = (N ekTe)cold. (5.5)

In Section 4.3, we showed that the hot gas density has a range of values between

2 x 104 and 3 x 105 cm-3. These values were obtained from Figure 21 for a variety of stellar temperatures and for an specific geometry and chemical composition in the gas.

The photoionization models used to predict the relationship between the continuum color and density also give us the average temperature in the shell, which we need to evaluate the pressure.

In Table 22 we show the electron density and temperature for several models that lie within the light gray area in Fig. 21. The pressure of the hot gas is roughly one 105 order of magnitude lower than the pressure of the cold gas, shown in the last line of

Table 22. Given the large uncertainties, this may not be a conclusive indication that the gas is not in pressure equilibrium. It is possible that pressure equilibrium occurs at some point within the shell, since the electron temperature varies with radius and we have also made the simplifying assumption of constant density for both phases.

Naturally, there is no compelling reason to demand pressure equilibrium as both phases may be only loosely associated.

5.2.2 Evidence for a “Clumpy” Structure

The assumption of spherical geometry for the hot gas is in part inspired from the appearance of supernova shells. The Crab Nebula, for instance, shows bright filaments of gas embedded in a diffuse shell. If the situation has any resemblance, we can adopt a filling factor of 1 for the hot phase, and express the ratio of emission measures as:

E{hot) Ng(hot)V(hot) 1 N 2(hot) E(cold) N*(cold)V (cold) ~ e N?(cold) 1 j

= (5.7) where e is the filling factor of the cold gas, and the second equation is the same as

Equation 4.5. The resulting values of e from the grid of models range from 2xl0 - 6 to 2xl0-3, and they are listed in the 6 th column of Table 22. Despite the large uncertainties, the results suggest that the cold gas phase has a “clumpy” structure, with m any cool, dense cloudets. 106

5.2.3 Where does the Emission Spectrum Originate?

Now that most of the relevant physical parameters have been constrained by the

observed properties of the continuum, we need to check that they are also consistent

with the emission-line spectrum. Unfortunately, this task is very difficult, as both gas phases might emit emission line radiation. In particular, Williams (1989) showed

that gas at temperatures typical of the cold gas phase can emit high-ionization lines, which may also be found in a very hot gas. This, as a matter of fact, has been the main argument to question the presence of a “coronal line region” in novae in favor of

a “coronal line stage”. According to Williams, as the remnant’s photosphere shrinks

after the outburst, its temperature increases up to ~ 10 6 K, and the hardening of

the radiation field would be capable of photoionizing the ejected gas to produce the observed coronal lines, while still having a low (~ 10 4 K) temperature. This can

certainly be the case, as we showed in Chapter II, for PW Vul, where [Fe X] 6374 was predicted to appear relatively strong in the cold gas.

To examine the relative contributions of each phase, we use our grid of models, to look at the predicted strengths of various emission lines. We have chosen some lines which can be expected to appear in the hot gas. The infrared fluxes have been interpolated from the data given by Greenhouse et al. (1988). Using the parameters listed at the beginning of this chapter, we obtain the line intensities listed in Table 23.

All lines have been normalized with respect to the observed intensity of H/? for day

554.

The results suggest that the coldest models in our grid predict fluxes which are 107

Table 23: Emission Line Fluxes in the Hot Gas

______I a / I H0 (day 554) Observ. Model Predictions T star(x l0 6 K) 1 .0 1 .0 1 .0 0.5 0.5 0.5 0 .2 0.3 N //(x l0 5cm -3) 0 .2 0.3 0.5 0 .6 0 .8 1.4 2 .0 3.0

He II 4686 0.5 - - - 0 .1 0 .2 0.7 3.0 5.1

N V 1240 10.4 --- 1.4 3.4 14.6 5.2 70 Ne VI 7. 6 /i 2 .8 0 .1 1 2 0 . 150. Mg VII 2629 1.1 2.9 1.6 Mg VIII 3n 1.4 0 .1 0.4 9.4 Si VI 2/i 0 .2 0.9 0.3 Si VII 2.5/i 0.4 0.4 1.5 Si IX 3.9/i 0 .8 - - - - 0 .1 0.3 - 1.0

much stronger than observed, and thus, we can discard them as unlikely to be repre­ sentative of the conditions in the hot gas phase. The other models, show weaker lines as the temperature increases. However, as the cold gas might emit coronal emission, the fact that these lines are weak in some models does not rule them out.

In order to verify that our results are consistent with the emission line spectrum, we need to compare the predictions from the cold gas phase with the observations.

We have used the photoionization code CLOUDY with initial input parameters as calculated in the previous sections. The code was then used as a subroutine of an optimization program that makes a non-parametric search for a best solution.

In one search, we used a 300,000 K blackbody to photoionize a shell whose char­ acteristics are listed as “Model 1” in Table 24. The abundances have been chosen as calculated in Chapter III. For the second search, we assumed that the gas is pho­ toionized by bremssthralung from the hot phase (i.e., the star ionizes the hot phase, 108

Table 24: Model Parameters

Model 1° Model 2 6 T.( K) 3.0 x 10s 3.0 x 105 L*(erg/sec) 2 .0 x 1038 2 .0 x 1038 Rsheiii cm) 1 .0 x 1 0 16 4.5 x 1015 N h ( cm-3) 6 .2 x 106 8 .0 x 106 Filling factor 2 .2 x 1 0 " 4 8.3 x 10~ 4 Adopted Abundances 0 H e/H 5.0 C /H 0.7 N /H 44.0 O/H 1 0 .0 N e/H 60.0 M g/H 6 .0 A l/H 90.0 Si/H 2 .0 Others 1 .0 ° Photoionization by a blackbody b Photoionization by free-free emission from hot gas c By number, relative to Solar

and radiation from the hot gas ionizes the cold phase). The best parameters are listed in the second column of Table 24. We should stress the fact that these are not necessarily the best possible models for QU Vul, as we have not explored the whole parameter space. In addition, other effects, such as the presence of dust, have also been neglected.

The predicted and observed intensities relative to H/3 are listed in Table 25. It is clear that, from the point of view of the emission lines, there is not a significant difference between different sources of ionization, at least in this case, as each model 109 represents reasonable well the observations. Some lines, such as [A1 III] 1860, show very large deviations from the observations, and this may be due to errors in the chemical composition assumed for each phase. The last line includes the predicted temperatures in the 0 ++ zone, which are comparable with the temperatures directly derived from the observations.

5.2.4 Thermal Stability

The models for the hot gas phase predict temperatures in the gas which are near

10s K (cf. Table 22). At these temperatures, the cooling reaches a maximum, and therefore it is important to check for thermal stability in the gas. Figure 23 shows the heating and cooling curves as a function of temperature for the hottest model of our grid (first entry in Table 22), where the solid and the dotted lines represent the cooling and heating, respectively. Since the cooling exceeds the heating for larger temperatures than the equilibrium point, and viceversa, we conclude that the model is stable. We also show the heating-cooling map for the bremss photoionized cold model (Fig. 24), which is also thermally stable.

5.2.5 Energy Budget

In the previous sections we have assumed that QU Vul maintained a luminosity level equal to the Eddington level for a 1 M® star. Is this a reasonable assumption? From the visual light curve discussed in section 3.1, we found that the time to decline 2 magnitudes was 25 days. From the luminosity-“speed class” relationship given by

Warner (1989), we obtain an absolute magnitude at maximum, My = —8.0 ± 0.9, Table 25: A Model for Day 554“

Line Model l 6 Model 2C Observed H (3d 33.41 33.73 33.89 He 1 5876 0 .0 2 0.03 0.24 H e 11 m O 1 2 . 2 2 . 3.60 He 7/4686 1.3 2.4 0.51 C III] 1909 1 .2 0.91 0.90 CIV 1549 5.3 5.2 1.90 [Nil] 5755 0 .0 2 < 0 .0 1 0 .2 0 N III] 1750 3.9 4.1 1.80 NIV] 1486 1 1 . 1 2 . 5.40 N V 1240 1 2 . 5.1 10.30 [07] 6300 < 0 .0 1 < 0 .0 1 0 .1 0 [0 77] 7325 0 .0 2 0 .0 1 0 .2 0 [0777] 1660+ 1666 1.9 2 .2 1.40 [0 777] 4363 1.5 2 .0 1.50 [0 777] 4 9 5 9 + 5007 2.9 4.2 6.50 OIV] 1402 4.2 4.7 blend Si IV] 1398 0.9 1.3 blend Total 1400 5.1 6 .0 4.33 [Ne III] 3869 + 3968 4.5 18. 14.8 [Ne IV] 1602 9.0 14. 7.02 [Ne IV] 4720 2 .2 3.3 1.60 [Ne V] 1575 1 .0 0.36 1 .2 0 [Ae V] 3346 13. 5.7 blend Ne III 3343 0 .1 0 .2 blend Total 3346 13. 5.9 11.89 Mg 772798 0.09 0.38 9.78 AlIII 1860 1.9 5.1 1.60 Si III] 1892 0.70 0.62 1 .0 0 [Ft VII] 6087 0.15 0.09 0 .1 0 ° Entries give line fluxes with respect to observed b Photoionization by a blackbody c Photoionization by free-free emission from a hot gas. d Entry is Log L(H/3) etn rt. nt i eg cm erg in Units rate. heating Figure 23: Therm al stability in a very hot gas: gas: hot very a in stability al Therm 23: Figure Heating-Cooling ,-ta ,•15 '13 1000 10000 3 - sec-1. TOO solid line: solid cooling rate; rate; cooling otd line: dotted 1 1 1 figure. Figure 24: Therm al stability of th e cold phase. Axes and units as in the previous previous the in as units and Axes phase. cold e th of stability al Therm 24: Figure Heating-Cooling 0 1 10* 1000 ® 1111 10000 10s TOO 10* 2 1 1 10s 113 which corresponds to a peak luminosity of 5.3 x 10 38 erg sec-1. This super-Eddington level has been observed in other novae (see, e.g., Starrfield 1989), and it is expected to be short-lived, quickly reaching levels at or near the Eddington limit. The slow decline of the ultraviolet and optical light curves until day ~ 500 suggests that the integrated luminosity may have maintained a constant level during that period.

Let us assume that the white dwarf maintained a luminosity to the Eddington level equal limit for its mass, LEdd ~ 2 x 1038 erg sec-1. At day 554, the total radiated energy was 9.6 xlO 45 erg. From the arguments above, this is clearly a lower limit to the emitted energy.

On the other hand, the kinetic energy of the shell (hot plus cold phases), is 1/2

Ms/te// v2 ~ 5.6 x 10 44 and 2.5 x 10 46 ergs, for the lowest and highest mass estimates derived at the beginning of this chapter, respectively.

For a 1 M0 white dwarf, and the ejected masses listed above, the liftoff energies required lie in the range 1.0 x 10 46 to 5 x 1047 ergs. These numbers suggest that a significant fraction of the energy of the outburst is being carried by the hot gas phase.

5.3 Summary

In this chapter we have derived additional properties of the shell, based on our previ­ ous results. The mass of the ejecta lies in the range l.OxlO - 4 to 4 .5 x l0 - 3 M0, which is consistent with other independent estimates.

The cold gas seems to be formed of many “cloudets”, as shown by the very low filling factors derived. We have no definitive evidence that these clouds are in pressure equilibrium with their surroundings. Radiative equilibrium model calculations indicate that it may not be possible to show from the emission-line spectrum alone what the source of ionization is. In addition, the parameters derived from the observed continuum reproduce reasonably well the emission-line fluxes, although a few large deviations are observed. Stellar temperatures between 300,000 and 500,000 K seem to reproduce most of the observed features, and are consistent with the X-ray observations. c h a p t e r VI

Concluding Remarks

6.1 Implications for Nova Physics

In order to put our findings in perspective, we will first summarize the general prop­ erties of Nova QU Vul and its expanding shell, and then proceed to discuss the implications of our results for nova theory.

QU Vul reached a peak brightness of 5 -6 m on December 25, and faded three magnitudes in about 40 days. P-Cygni profiles were observed within the first few days of the outburst (Andrillat, IAU Circular 4026), showing absorption components with velocities of -650 km /s and -1380 km /s- Most of the absorption disappeared within 4 months of the outburst, indicating that the nova had reached the nebular stage. An additional high-velocity component of the ejecta may have been present during the early times, with velocities of several thousand km /s, as evidenced by a very broad infrared Ne II em ission (Gehrz, G rasdalen and Hackwell, 1985).

Early in the outburst, the emission was dominated by low ionization permitted transitions of hydrogen, Fe II, Mg II and other ions (Andrillat, IAUC 4026). As these lines weakened, the spectra became dominated by medium to high ionization intercombination and forbidden lines of Ne, O, N, and other elements. Coronal-line

115 116

emission was also observed in the infrared (Greenhouse et al. 1988) approxim ately

500 days after the outburst. They observed lines of Al V and IX, Si VI, VII and IX,

Mg VIII, and others, and estimated that the excitation temperature of the gas was

T cor ~ 6 x 105 K, assuming photoionization by the white dwarf.

EXOSAT X-ray observations by Ogelman, Krautter and Beuermann (1987) show

that the X-ray emission was increasing between days 115 and 307. They combined

these data with those of other novae and constrained the temperature of the white

dwarf to the range 2.2-4.4xlOsK. A prediction of the X-ray photon count by Mac­

Donald and Vennes (1991) also estimated that Il w d ~ 3 x 105K. The error of their

estimate, however, might be large on account of the poor statistics.

Our estimate of the temperature from forbidden-line diagnostic methods yields a

relatively constant value of T ne4 ~ 104K and a density of about 10 7 cm - 3 between

days 120 and 500, with the density decreasing after that time. This cold gas phase has a very low filling factor, as derived in the previous chapter. This suggests that the clumps of cold gas may actually have had a roughly constant density during this period. This can result from magnetic confinement or other mechanisms, and this point should be further studied.

We have also estimated the abundances in the shell by selecting ratios of elements with similar ionization potentials in order to minimize the influence of the particular local conditions, such as density inhomogeneities. The abundances by number rel­ ative to hydrogen are, with respect to the solar abundance: He=5, C=0.8, N=44,

0=10, Ne=60, Mg=6 , and Ar=0.7. Most abundances are good within a factor of 2. 117

Table 26: The Ejected Mass of QU Vul

Reference Mej (M©) Notes This work (lower limit) 9.5 x 10" 5 2 -phase model Taylor et al. (1987) 8 x 1 0 - 4 Radio emission Greenhouse et al. (1988) > 9 x 10“ 4 Infrared emission This work (higher limit) 4.5 x 10" 3 2 -phase model

Greenhouse et al. (1988), obtained the following ratios in the coronal zone, by num­ ber: Al/Si=5.8, and Mg/Si=5.1. Using our Mg abundance, in combination with these results, we obtain Si/H=1.3 solar, and Al/H=92 solar. These abundances strongly support the idea that the outburst took place in an ONeMg white dwarf, and they can be qualitatively explained with a model in which part of the enhancement is due to mixing of core material from the white dwarf into the accreted envelope with some nitrogen and oxygen possibly synthesized during the outburst (Starrfield, Sparks and

Truran 1986).

The continuum spectrum could not be reproduced with a simple model of bound- free and free-free emission from a gas at T~ 10 4K. The flatness of the continuum, particularly in the far ultraviolet, and the lack of a strong Balmer jump suggests that the emission is dominated by free-free emission from a hot gas. This featureless continuum is observed in all the UV spectra, between days 200 and 627. A fit of the continuum at day 554 shows that the coronal-line region should dominate the mass of the outburst, being at least 18 times more massive than the nebular-line region.

Several estimates of the total mass of the ejecta are now available, and we have listed them in Table 26. 118

The large values derived for the ejected masses of PW Vul and QU Vul has im­ portant implications for nova studies. First, it is believed that ONeMg white dwarfs tend to be massive stars. Ritter et al. (1991) present a detailed theoretical study of the frequency distribution of nova outbursts, allowing for selection effects introduced by certain physical characteristics of the system. They show that novae occurring on massive ONeMg white dwarfs are expected to be the most frequent types, despite the fact that these systems are extremely rare among white dwarfs. For instance, they expect that one quarter to one half of the observed outbursts should occur on primaries with masses over 1.3 M®.

6.2 Implications for Galactic Nucleosynthesis

How does the white dwarf mass relate to the ejected material? It has been shown by

MacDonald (1983) that a thermonuclear runaway will occur in the accreted envelope when the pressure at its base reaches about 1020 dynes cm-2. Starrfield (1989) showed that the critical envelope mass is an inverse function of the white dwarf mass, and, for a 1.3 M® star, it is 3.1 x 10-5 M®. This value is lower than any estimated value of the shell’s mass (Table 26). The higher masses obtained by this and other studies are consistent with a white dwarf mass of 1 M® or less. These considerations suggest that QU Vul is an unusually slow nova for its type, whose primary star appears to be less massive than expected. This presents a theoretical problem that needs to be examined in detail.

Another important implication of this study concerns the chemical abundances.

ONeMg novae are considered to be a significant source of 26A1 in our Galaxy (Weiss 119 and Truran, 1990; Nofar, Shaviv, and Starrfield, 1991). In their study of nucleosyn­ thesis in novae, they estimate the mass of 26Al in the Galaxy due to ONeMg nova outbursts, as

where Rnot;a is the galactic nova rate, f is the fraction of novae that produce 26A1, i.e., ONeMg novae, Mej- is the total mass of the ejecta, and X 26 is the fractional mass of ejected 26A1. Their choice of an average Mej ~ 2 x 1 0 '5 M© was motivated by the idea that ONeMg white dwarfs are massive? as explained above. The total mass of 26A1 in the interstellar medium has been estimated by Mahoney et al. (1984) from gamma-ray observations. Their value of 3 M© is almost an order of magnitude higher than the predictions from Equation 6.1. Nofar, Shaviv and Starrfield (1991), using hydrodynamic simulations to predict the galactic mass of 26 Al, reached similar conclusions in the sense that an additional class of novae should provide the missing m aterial.

From Equation 6.1 we have estimated the fraction of novae of the QU Vul type needed to explain the observed mass of aluminum in the Galaxy- Using the results of this paper for the mass of the ejecta and the fractional mass of aluminum, we find that the rate of occurrence of novae of the QU Vul type is °f the order of 1-2 %. Therefore, although extremely rare, the occurrence of a few novae like QU Vul in the Galaxy would be sufficient to account for most of the observed abundance of aluminum. 1 2 0

6.3 Conclusions

In this work we have used infrared, optical and ultraviolet spectrophotometry to study the physical properties of the shells of Novae PW Vul 1984 and QU Vul 1984. Based on these analyses, we can draw the following conclusions:

• The chemical composition of both shells shows a qualitative agreement with

the hydrodynamic calculations of nova outbursts. PW Vul’s nearly solar abun­

dances, with the exception of nitrogen, is consistent with an outburst taking

place on a very slow CO white dwarf. In the case of QU Vul, we observe an

enhancement of intermediate mass species, ranging from carbon to magnesium,

and possibly silicon. A very high helium abundance is also observed which

might indicate the presence of steady hydrogen burning past the peak of the

outburst.

• The shells are dominated, both in mass and energetics, by a tenuous, hot gas

phase that emits copious amounts of continuum radiation. Most emission-line

radiation is produced in a denser, cooler nebular region. Coronal-line radiation,

however, may be produced in both regions, depending on the characteristics

of the ionizing radiation field, and the local conditions in the gas. In this

sense, it is necessary to look at both the emission lines and the continua to

determine whether more than one gas phase is present, and sort out their relative

contributions. • Nova QU Vul seems to be a rare type of slow ONeMg nova, possibly involving a

low mass white dwarf, and certainly capable of ejecting a large amount of mass

into the interstellar medium. The high abundance of aluminum in the ejecta

makes this nova an extraordinary contributor of this element to our Galaxy.

Although the frequency of occurrence of such an event is very small, their very

existence is difficult to understand with our current theories of nova outbursts. B ibliography

Audouze, J., Tinsley, B. M. 1976. Ann. Rev. Astron. Ap., 14,43.

Austin, S. J., Starrfield, S. G., Saizar, P., Shore, S. N., and Sonneborn, G. 1990. In Evolution in Astrophysics, ESA SP-310, p. 367.

Burguess, A. and Seaton, M. 1960. M.N.R.A.S., 121, 76.

Bvard, P. L., Foltz, C. B., Jenkner, H., and Peterson, B. M. 1981. P. A. S. P., 93, 147.

Clegg, R. E. S. 1987. M.N.R.A.S., 229, 31p.

Duerbeck, H. W., Geffert, M., Nellse, B., Dummler, R., and Nolte, M. 1984. Inf. Bull. Var. Stars No. 2641.

Ferland, G. J. 1978. Ap. J., 219, 589.

Ferland, G. J. 1980. P. A. S. P., 92, 596. Ferland, G. J. 1989. Astronomy Dept. Internal Report 89-001 (The Ohio State University).

Ferland, G. J., and Shields, G. A. 1978. Ap. J., 226, 172.

Ferland, G. J., Lambert, D. L., and W oodman, J. H. 1977. Ap. J., 213, 132.

Ferland, G. J., Lambert, D. L., and Woodman, J. H. 1986. Ap. J. Suppl., 60, 375.

Friedjung, M. 1989. In Classical Novae, ed. M. F. Bode and A. E. Evans (New York: Wiley), p. 187.

Gallagher, J. S., and Starrfield, S. 1978. Ann. Rev. Astron. Ap., 16, 171. Gehrz, R. D. 1988. Ann. Rev. Astron. Ap., 26, 377.

Gehrz, R. D., Grasdalen, G. L., and Hackwell, J. A. 1985. Ap. J. (Letters), 298, L47.

Gehrz, R. D., Grasdalen, G. L., Greenhouse, M., Hackwell, J. A., Hayward, T., and Bentley, A. F. 1986. Ap. J. (Letters), 308, L63.

1 2 2 123

Gehrz. R. D., Harrison, T. E., Ney, E. P., Matthews, K., Neugebauer, G., Elias, J., G rasdalen, G. L., and Hackwell, J. A. 1988. Ap. J., 329,894. Grasdalen, G., and Joyce, R. R. 1976. Nature, 259, 187.

Greenhouse, M. A., Grasdalen, G. L., Hayward, T. L., Gehrz, R. D., and Jones, T. J. 1988. Astron. J., 95, 172.

Hacke. G. 1987. Inf. Bull. Var. Stars No. 2979.

Hauck, B., and Mermilliod, M. 1985. Astr. Ap. Suppl., 60, 61. Hauschildt et al. 1992. Preprint.

Kenyon, S. J., and Wade, R. A. 1986. P. A. S. P., 98, 935.

Livio, M. 1992. Submitted to the Ap. J.

Livio, M., Shankar, A., Burkert, A., and Truran, J. W. 1990. Ap. J., 356, 250. Lucke, P. B. 1978. Astr. Ap., 64,367.

M acDonald, J. 1983. Ap. J., 267, 732.

M acDonald, J., and Vennes, S. Ap. J. (Letters), 373, L51.

M cLaughlin, D. 1960. In Stellar Atmospheres, ed. J. L. Greenstein (Chicago: Univ. Chicago Press), p. 585.

Mahoney, W. A., Ling, J. C., Jacobson, A. S., and Liggenfelter, R. E. 1984. Ap. J., 286, 578.

M artin, P. G. 1989. In Classical Novae, ed. A. Evans and M. Bode (New York: Wiley), p. 93.

M endoza, C. 1983. In IAU Coll. 103 Planetary Nebulae, ed. D. R. Flower (Dordrecht: D. Reidel), p. 143.

Mermilliod, J. C. 1987. Astr. Ap. Suppl., 71, 119.

Neckel, T h., and Klare, G. 1980. Astr. Ap. Suppl., 42, 251.

Nofar, I., Shaviv, G., and Starrfield, S. 1991.Ap. J., in press.

Nussbaumer, H., and Storey, P. J. 1984. Astr. Ap. Suppl., 56, 293.

Ogelman, H., Krautter, J., and Beuermann, K. 1987. A str. Ap., 177, 110.

O sterbrock, D.E. 1989. Astrophysics of Gaseous Nebulae and Active Galactic Nuclei (Mill Valley, CA: University Science Books).

Ritter, H., Politano, M., Livio, M., and Webbink, R. F. 1991. Ap. J.. 376, 177. 124

Payne-Gaposchkin, C. 1957. The Galactic Novae (New York: Dover).

Peimbert, M. 1975. Ann. Rev. Astron. Ap., 13, 113. Schild, R.E. 1976. Ap. J. (Letters), 209, L35.

Seaton, M. J . 1978. M.N.R.A.S., 185, 5p.

Seaton, M. J . 1979. M.N.R.A.S., 187, 73p.

Shara, M'. M. 1989. P. A. S. P., 101,5.

Shore, S. N., and Braine, J. 1991. Preprint.

Sowell, J. R., Laird, J. B., and Thorstensen, J. R. 1985. Inf. Bull. Var. Stars No. 2700.

Starrfield, S. 1986. In Radiation Hydrodynamics, ed. D. Mihalas and K.-H. Winkler, (Dordrecht: Reidel), p. 225.

Starrfield, S. 1988. In Multiwavelength Astrophysics, ed. F. A. Cordova (Cam­ bridge University Press), p. 159.

Starrfield, S. 1989. In Classical Novae, ed. A. Evans and M. Bode (New york: W iley), p. 123.

Starrfield, S. 1991. In Binary Stars, ed. J. Sahade and Y. Kondo (Kluwer Press), to be published.

Starrfield, S., Sparks, W. M., and Truran, J. W. 1974. Ap. J., 192,647.

Starrfield, S., Sparks, W. M., and Truran, J. W. 1986. Ap. J. (Letters), 303, L5.

Starrfield, S., Truran, J. W., and Sparks, W. M., and Kutter, G. S. 1972. Ap. J., 176,169.

Starrfield, S., Truran, J. W., Sparks, W. M., Krautter, J., and MacDonald, J. 1990. In Physics o f Classical Novae, ed. A. Cassatella and R. Viotti (Heidelberg: Springer-Verlag), p. 306.

Starrfield, S., Shore, S. N., Sparks, W. M., Sonneborn, G-, Truran, J. W.. and Politano, M. 1992. Ap. J. (Letters), „ in press. Taylor, A. R., Seaquist, E. R., Hollis, J. M., and Pottasch, S. R. 1987. Astr. Ap., 183, 38. Stromgren, B- 1966. Ann. Rev. Astron. Ap., 4, 433. Torres-Peim bert, S., and Peim bert, M. 1977. Rev. Mex. Astron. Ap., 2, 181.

Truran, J. W- 1982. In Essays in Nuclear Physics, eds. C. A. Barnes, D. D. Clayton, and D. N. Schramm (Cambridge: Cambridge U. Press), p. 467. 125

Wagner, R. M . 1986. Ap. «/., 308, 152.

Warner, B. 1989. In Classical Novae , ed. N. Evans and M. Bode (New York: Wiley), p. 1

Weiss, A., and Truran, J. W. 1990. Astr. Ap., 238, 178.

Wilkes, B. J., Ferland, G. J., Hanes, D., and Truran, J. W. 1981. M.N.R.A.S.. 197, 1.

Williams, R. E-, Hamuy, M., Phillips, M. M., Heathcotte, S. R., Wells, L., and Navarrete, M. 1991. Ap. J., 376, 721.