Content uploaded by Michael A. Dopita
Author content
All content in this area was uploaded by Michael A. Dopita
Content may be subject to copyright.
THE ASTROPHYSICAL JOURNAL, 474:188È204, 1997 January 1
1997. The American Astronomical Society. All rights reserved. Printed in U.S.A.(
HUBBL E SPACE T EL ESCOPE OBSERVATIONS OF PLANETARY NEBULAE IN THE
MAGELLANIC CLOUDS. V. MASS DEPENDENCE OF DREDGE-UP AND THE
CHEMICAL HISTORY OF THE LARGE MAGELLANIC CLOUD
M. A. DOPITA,1,2 E. VASSILIADIS,3,4 P. R. WOOD,1S. J. MEATHERINGHAM,1J. P. HARRINGTON,5
R. C. BOHLIN,3H. C. FORD,3T. P. STECHER,6AND S. P. MARAN6
Received 1996 March 11; accepted 1996 July 17
ABSTRACT
A photoionization analysis of Hubble Space Telescope UV and ground-based optical spectrophoto-
metry is given for eight more planetary nebulae (PNs) in the Large Magellanic Cloud (LMC). This
allows the central stars to be placed accurately on the H-R diagram and permits the determination of
the He, C, N, O, Ne, S, and Ar abundances. In some cases, the gas-phase abundances of Mg and Si may
also be determined. We have combined these results with the analysis of two other objects published by
us in the Ðrst two papers of this series. The observed abundance patterns are qualitatively consistent
with the (mass-dependent) operation of the various chemical dredge-up processes as predicted by theory.
Dredge-up of C during the thermal pulsing stage appears to be most important, and ““ hot bottom
burning ÏÏ transforms much of this C to N in the more massive stars. There is no sign of dredge-up of
22Ne. We show that the spread in the a-process element abundances can be understood as being due to
di†erences in core mass of the planetary nebula nucleus, which is related directly to initial mass of the
precursor star. This is, therefore, a tracer of the age-metallicity relationship for stars in general, and we
derive, for the Ðrst time, the chemical history of the LMC based on PNs. We Ðnd that the base metal-
licity of the LMC almost doubled D2 Gyr ago. This is consistent with studies of Ðeld stars and of clus-
ters that show that there was a major burst of star formation at that time.
Subject headings: galaxies: abundances È galaxies : Magellanic Clouds È planetary nebulae : general È
stars: evolution
1.INTRODUCTION
The planetary nebula (PN) phase of evolution represents
the Ðnal phase of late stellar evolution of low- to
intermediate-mass stars as they pass on their way to becom-
ing white dwarfs. During the asymptotic giant branch
(AGB) phase that precedes the PN phase, there occurs a
complex series of thermal-pulsing events, chemical pro-
cessing, and dredge-up into the envelope, accompanied by
rapid and episodic mass loss. This is terminated by a Ðnal
envelope ejection event that may, or may not, be correlated
with the phase of the He-Ñash cycle. All these processes
a†ect the subsequent evolution in the PN phase and are still
relatively poorly understood, although rapid theoretical
progress into these questions has been made in recent years
et al.(Scho nberner 1983; Weidemann 1987; Alongi 1993;
& Wood et al. &Vassiliadis 1993; Bressan 1993; Chiosi
Marigo 1996).
In our major Hubble Space Telescope (HST ) program,
results of which were reported in the Ðrst four papers of this
series (Dopita et al. hereafter Papers1993, 1994, 1996, I, II
and respectively; et al. hereafterIV, Vassiliadis 1996, Paper
we have circumvented the distance scale problem thatIII,
plagues studies of Galactic PNs by investigating a sample of
PNs in the Magellanic Clouds (MCs) drawn from the
1Mount Stromlo and Siding Spring Observatories, Institute of
Advanced Studies, Australian National University, Private Bag, Weston
Creek P.O., ACT 2611, Australia.
2Michael.Dopita=anu.edu.au.
3Space Telescope Science Institute, 3700 San Martin Drive, Baltimore,
MD 21218.
4Currently at Instituto de AstroÐsica de Canarias, Via Lactea, E-38200,
La Laguna, Tenerife, Canary Islands, Spain.
5Department of Astronomy, University of Maryland, College Park,
MD 20742-2421.
6NASA Goddard Space Flight Center, Greenbelt, MD 20771.
catalog of MacConnell, & Davis PhilipSanduleak, (1978).
At optical wavelengths, this population has already been
the subject of a systematic and detailed observational study
both by us and by the University College group. Data on
the line Ñuxes, the nebular densities, the expansion veloci-
ties, and the kinematics have all been accumulated in the
last few years (see the review by also etBarlow 1989; Dopita
al. Ford, & Webster et1985, Dopita, 1985; Meatheringham
al. Dopita, & Morgan1988a, Meatheringham, 1988b,
Bessell, & Dopita et al. andWood, 1986, Wood 1987,
Ciardullo, & Walker The distance to theJacoby, 1990).
MCs has been established accurately by a variety of other
techniques (among which determinations based upon
Cepheid variables, and upon SN 1987A, are the most
accurate).
At the distance of the LMC, the PNs typically subtend
less than 1Aon the sky. Nonetheless, they remain well
resolved by the HST , and Ðne details of their internal struc-
ture may often be distinguished et al. In(Dopita 1996).
addition, they typically display low systemic reddening.
This is particularly important for their spectroscopic study
in the UV, where we can not only detect directly the central
star through its continuum and absorption line system, but
we can also use the nebular spectrum to determine directly
the chemical abundances of many elements that are not
accessible to analysis either at optical or IR wavelengths. In
particular, we can observe all the elements thought to be
involved in dredge-up chemical processing; He, C, N, and
O. In many respects, therefore, the LMC and SMC PNs
represent a key to our understanding of the details of evolu-
tion beyond the AGB.
The results of the FOS spectroscopy of these objects has
already been given by et al. Here weVassiliadis (1996).
present detailed photoionization models for individual
objects, which allows us to infer observationally details of
188
HST MAGELLANIC CLOUD PN OBSERVATIONS. V. 189
the operation of the chemical dredge-up processes, and
which gives us Ðrst results on the chemical evolution of the
Large Magellanic Cloud as a whole.
The distance to the LMC is assumed to be 50 kpc in the
analysis that follows. This now appears to be well estab-
lished through a number of techniques. From Cepheid
period-luminosity (PL) and period-luminosity-color (PLC)
relations, distance modulii of 18.4È18.55 are obtained
& Laney An analysis of the ionization of(Caldwell 1991).
the ring in SN 1987A gave a modulus of 18.5 et al.(Panagia
Finally, an extremely accurate distance modulus has1991).
been determined by Ðtting theoretical light curves to a
bump Cepheid in the LMC. This gives 18.53 ^0.04 (Wood,
Arnold, & Sebo Any systematic error in the assumed1996).
distance will propagate to all luminosities quoted, and this
will in turn a†ect systematically estimates of core mass,
initial mass, and age of the central stars.
2.PHOTOIONIZATION MODELING
The UV spectrophotometry et al. was(Vassiliadis 1996)
combined with ground-based data (Meatheringham &
Dopita to generate a self-consistent set of1991a, 1991b)
de-reddened line intensities extending from 1216 toÓ
beyond 7300 in the red. These spectra are interpreted inÓ
conjunction with the HST imaging data using the gener-
alized modeling code MAPPINGS II Dopita, &(Binette,
Tuohy & Dopita The goals of this1985; Sutherland 1993).
self-consistent nebular modeling are as follows:
1. As far as possible, match the modeled size, density, and
internal structure of the nebula with observations.
2. Determine a luminosity of the central star that agrees
with both the absolute luminosity of the nebula and the
observed continuum Ñux, particularly in the UV.
3. Find the temperature of the central star and the
strength of the local UV Ðeld (the ionization parameter)
that matches the observed degree of ionization and excita-
tion.
4. Determine the chemical abundances for each element
in the nebula (the ““ metallicity ÏÏ) that matches the observed
line spectrum as closely as possible.
5. Determine whether, and to what extent, the nebula
leaks UV radiation (optically thin or thick).
Extremely valuable constraints can be obtained from the
HST imaging data. For all the PNs for which we have
spectra, we had earlier obtained et al. Planet-(Dopita 1996)
ary Camera (PC) images in the [O III]j5007 line, chosen
because it is generally the brightest emission line in the
optical. These images provide both size and structural infor-
mation that can be used as input to detailed photoioniza-
tion modeling analysis.
From the analysis of a sample of 15 planetary nebulae in
the Magellanic Clouds, et al. found clear evi-Dopita (1996)
dence for size evolution across the H-R diagram. However,
the younger, low-excitation, compact PNs tend to be sys-
tematically smaller than photoionization models based on
ground-based data would predict, suggesting these have a
central reservoir of dense atomic and molecular gas. This
gas lies close to the central star and is currently undergoing
ionization and being accelerated into outÑow. PNs classi-
Ðed previously as nitrogen-rich objects with massive central
stars (Peimbert type I) show mostly the bipolar ““ butterÑy ÏÏ
symmetry that is also a characteristic of their Galactic
counterparts.
The derived kinematic ages range from less than 1000 yr
up to almost 5000 yr but show little sign of systematic
increase along the evolutionary tracks. Overall, the lack of
any clear correlation of dynamical age and position on the
H-R diagram seems to result from the dynamical evolution
of the PN nebular shells themselves, the details of which
also depend upon whether the central star leaves the AGB
as an H or He burner. A correlation between the excitation
class and expansion velocity was found by et al.Dopita
In its turn, excitation class is correlated closely with(1988).
the logarithm of the e†ective temperature & Mea-(Dopita
theringham Subsequently found that1991a). Dopita (1993)
an even better correlation existed between a combination of
log log and expansion velocity :(L/L_), (Teff),
(Vexp/km s~1)\[128 ^4]38 ^2
][log (Teff)[0.25 ^0.05 log (L/L_)] .
The true ages of the larger objects are therefore systemati-
cally underestimated because of acceleration of the nebular
shell during its lifetime. Apparently the PN shells are accel-
erated initially by the increase of thermal energy content
during their nuclear burning phase of evolution and coast
at almost constant expansion velocity during the fading
track. In this model, the stellar wind provides the pressure
that conÐnes the ionized material to a (somewhat) thin shell.
The terminal velocity reached by the nebular shell is corre-
lated with the mass of the PNs, with more massive PNnÏs
driving faster expansion. The increase in thermal energy
content is due in part to the temporal increase in ionized
mass, but is mostly due to thermalization of the fast stellar
wind, which develops as the PNn tracks to high tem-
perature on the H-R diagram.
Using the relation, along withVexp:log (Teff) : log (L/L_)
the & Wood theoretical evolutionaryVassiliadis (1993)
tracks of the central star, et al. derived twoDopita (1996)
semiempirical estimates for the evolutionary timescales
based upon the nebular size and the measured dynamical
age. They then demanded that these evolutionary time-
scales should be consistent with the evolutionary age
derived from theory. With this assumption, they concluded
that He-burning PNnÏs outnumber the H burners in the
approximate ratio 2:1 in the LMC.
2.1. T he Nebular Density Model
The essential features of the imaging data summarized
above provide the basis for the nebular density model that
we have adopted. This model must be reasonably physical,
but it must also be fairly simple so that the nebular param-
eters can be well constrained by the available observational
material. Clearly, the features of the dynamical and size
evolution support the ““ interacting winds ÏÏ model of Kwok,
Purton, & Fitzgerald in which a fast (V[100 km(1978)
s~1) wind interacts with much slower ejecta (VD10 km
s~1) ejected during the AGB. The fast wind both ionizes and
sweeps out the slow-moving ejecta. Provided either that the
slow-moving ejecta were ejected with nonspherical sym-
metry, or that instabilities in the interaction with the fast
wind has allowed the slow-moving ejecta to be broken up
into clumps at a characteristic radius, the photoionization
model can be simpliÐed to a two-zone model, one dealing
with the material close to the clumps, and the other describ-
ing the (lower density) ionized Ñow region. Both these com-
ponents are necessary in order to produce a satisfactory
photoionization model, as we found in the Ðrst two papers
of this series (Dopita et al. 1993, 1994).
190 DOPITA ET AL. Vol. 474
In our Ðtting procedure, therefore, we have adopted
mostly a two-zone model. Zone 1 is an optically thick, iso-
baric region having a certain covering factor as seen at the
central source. This region is ram-pressure conÐned at its
inner boundary by the stando† shock in the fast wind of the
PNn, and it will produce the bulk of the emission in lines of
low-ionization potential, which are produced close to ion-
ization fronts. The density in this region is therefore well
constrained by electron densities inferred from the [S II]or
[O II] density-sensitive line ratios.
Zone 2 is a region with density law of the form nH(R)\
A]BR~2, which covers the remainder of the solid angle as
seen at the central source. This region may be either opti-
cally thick or thin. The rationale behind this choice is that
since there is an ionization front eating into the store of
neutral gas ahead of zone 1, there is a continuous supply of
gas available to be photoionized and accelerated into the
surrounding space either through its own pressure gradient
or with the assistance of the ram pressure of the fast stellar
wind. This sets up a density distribution that is decreasing
steeply in the radial direction. The HST nebular photo-
metry et al. indicates this to be the case for(Dopita 1996)
those objects that are well resolved. However, in the outer-
most regions, the density will tend to become approx-
imately constant with the pressure of the ionized gas being
determined by the ram pressure of the stellar wind at the
mean radius of the stellar wind shock, which generally will
lie well outside the inner edge of zone 1. Since the fast wind
that is accelerating the nebular shell originates at the central
star, we assume also that the radii of the inner boundaries of
both zones are the same. To allow the Ñow to escape from
the region of the ionization front, the inner density of zone 2
must be always lower than that of zone 1. The radius of the
inner boundary of the nebular shell is inferred from inspec-
tion of the nebular morphology, in the cases in which this is
possible.
2.2. Derivation of Stellar Parameters
The Ðrst object of nebular modeling is to be able to place
the central star of the PN upon the Hertzsprung-Russell
(H-R) diagram. For the hot central stars[log L: log (Teff)]
of PNs, by far the greatest fraction of the energy is put out
below the Lyman limit. In an optically thick nebula, in
which the ionizing radiation from the central star is totally
absorbed, the luminosity is directly proportional to the sum
of all the nebular emissions (line, continuum, and IR). This
is the principle of the ““ energy balance ÏÏ concept. This tech-
nique, discussed Ðrst by and improved byStoy (1933) Kaler
and & Pottasch can also be(1976) Priete-Martinez (1983),
used to derive the e†ective temperature of the central star,
since the mean energy per photoionization becomes larger
for hotter stars, so there is a direct correlation between the
nebular and the stellar temperature. This method, extremely
accurate in principle, is often difficult to apply, since it
depends on having a complete set of measurements of the
nebula. It has most notably been applied by inRatag (1991)
the context of the Galactic bulge population of bright PNs.
For PNs in the Magellanic Clouds, prior to HST we
usually had available only optical spectrophotometric mea-
surements. Photoionization modeling of the PNs (Dopita &
Meatheringham had therefore to rely mainly1991a, 1991b)
on the HbÑux (as a measure of the recombinations, and
therefore of the ionizing Ñux) to derive the luminosity of the
central star, and upon the degree of excitation of the nebula
(which is related to the ionization temperature of the central
star) to estimate the temperature of the central star. The
main problem with this is that many PNs absorb the UV
radiation completely in some directions, but are very opti-
cally thin in other directions. From ground-based data
alone, the optically thin portions may contribute little to the
total Ñux, and they often contribute signiÐcantly only to
lines of high ionization potential, which are not the main
coolants in the optical. Nebular modeling based on optical
spectra alone will in general lead to an underestimate of the
luminosity, and, to a lesser extent, it will provide an over-
estimate of the temperature of the central star.
The HST spectral data provides vital complementary
data for our understanding of the distribution of PNs on
the H-R diagram, and for their dynamical and physical
evolution. From the spectral distribution in the continuum,
in many cases we can detect directly the central star at
far-UV wavelengths. In one or two cases, we can even
measure the line proÐles produced in the atmosphere of the
central star to estimate stellar wind parameters. For a
blackbody stellar source in the LMC, the apparent bright-
ness (ergs cm~2 s~1 at the edge of the EarthÏs atmo-Ó~1)
sphere at wavelength is given byj(Ó)
B(j)\4.956 ]10~19(R/cm)2(j/Ó)~5
Mexp [1.439 ]108(j/Ó)~1(T/K)~1][1N, (2.1)
where Ris the radius of the star, and Tis its e†ective
temperature. This formula can be used to determine the
luminosity of the star directly.
In an optically thick PN, the ratio of stellar to nebula
continuum intensities gives a direct estimate of the e†ective
temperature of the central star, and in the UV we can detect
many resonance and intercombination lines of highly
ionized species that are sensitive to the optically thin
regions of the nebula, allowing these estimates to be cor-
rected for the escape of UV photons from the nebula.
2.3. Self-consistent Photoionization Models
With the density model presented above, we proceed with
photoionization modeling in the manner already described
in Dopita et al. Our MAPPINGS II photoion-(1993, 1994).
ization models must match the observed linear size, mor-
phology, densities determined from [O II], [S II], or the UV
lines, and the reddening-corrected HbÑux. The central star
is assumed to have a blackbody distribution in frequency.
Stellar and nebular parameters and chemical abundances of
each element are changed by hand at each iteration until
the dispersion of the observed versus theoretical line inten-
sities is minimized for all ionization stages of each element.
In this procedure, the overall degree of excitation is deter-
mined largely by the assumed e†ective temperature of the
central star. Typically four or Ðve iterations are required
until a satisfactory model is obtained. For most objects,
unless otherwise noted, the random errors in the
abundances of individual ions may be as large as ^25%,
but random errors in the abundances of individual elements
are mostly only ^15%, since these are determined by the
error in the mean abundance of the dominant ionization
stages, which are smaller than the errors associated with
minor ionic constituents.
Some lines prove to be quite sensitive to the degree of
optical thinness. In particular, strong He II,[NV], [Ne V],
and [Ne IV] lines frequently show the need for an optically
thin component, as do weak C II], [N I], [N II], [O I], or
[O II] lines. These lines provide very useful constraints on
the parameters of both zones. In we show separa-Figure 1
No. 1, 1997 HST MAGELLANIC CLOUD PN OBSERVATIONS. V. 191
tely the contributions of the two components for one object
(LMC-SMP 8) that displays both optically thick and opti-
cally thin components. The requirement for the contribu-
tions of both components to ensure the goodness of Ðt is
clearly indicated. The discussion on the Ðts for individual
objects in expands this point.°2.4
For cooler objects, the central star is directly visible
through its UV continuum. For our Ðnal models of each
object, we have constructed the theoretical nebular contin-
uum and added to this the predicted stellar continuum pre-
dicted by A comparison with the observedequation (2.1).
continuum always provides a ““ sanity check ÏÏ of the model,
and in some cases it also allows us to obtain an independent
estimate of the luminosity of the central star, provided we
assume that the temperature of the central star has been
determined accurately from the nebular model. An attempt
to determine directly the temperature from a variant of the
classical Zanstra method would often be invalidated by the
optical thinness of zone 2 of the model.
The parameters estimated for each of the models are
given in The derived elemental abundances forTable 1.
these models, and for those of Papers and are given inIII,
The detailed spectral model Ðts are given inTable 2. Table
and the goodness of Ðt of the spectral line model is shown3,
in Figures Model continuum Ðts for objects in which2aÈ2b.
the central star is detectable are given in Figures 3aÈ3c.
Finally, gives the logarithmic abundances withTable 4
respect to the solar values from & Grevesse inAnders (1989)
accordance with the usual practice of stellar spectroscopists.
We have used the mean of the heavier a-process element
abundances with respect to the solar values
[a/H] \([Ne/H] ][S/H] ][Ar/H])/3 (2.2)
as our ““ metallicityÏÏ indicator. These elements are less likely
to be a†ected by dredge-up processes (see the discussion in
below). Such an average reduces the uncertainty in the°3.4,
abundance with respect to the solar values. In Tables and2
we have also added the results for LMC-SMP 83 and4
FIG. 1.ÈThe theoretical line Ðt for LMC-SMP 8 showing the contribu-
tions made by each of the zones of the photoionization model, showing
clearly the need for both an optically-thick and an optically thin com-
ponent.
LMC-SMP 85 from Papers and of this series (Dopita etIII
al. and1993 1994).
It is useful to conclude this section with a few remarks
about temperature-sensitive lines such as [O III]j1363. In
general, observers spend a great deal of e†ort measuring
such lines in order to determine the nebular temperatures.
However, temperature-sensitive lines such as this one are
enormously a†ected by Ñuctuations in the temperature, as
has been pointed out many times over the years by the
Peimberts and their coworkers (see the recent review by
and references therein). Any temperaturePeimbert (1995),
enhancements that are due to shock waves and turbulence,
or any temperature enhancements near ionization fronts,
will enhance the intensity of such lines, leading to an over-
estimate of the mean temperature, and in consequence, an
underestimate of the elemental abundances. To quote from
that paper ““...the best determinations of abundance ratios
are those that do not depend, or depend weakly, on the
electron temperature.ÏÏ In this sense, the self-consistent
global modeling that we present here should approach this
ideal since we seek simply to minimize the scatter between
the observed and modeled intensities of all lines on a
logarithmic plot, without any regard to temperature or cor-
rection for unseen stages of ionization. However, it should
be remarked that this procedure is only made possible by
the fact that we have both the UV and visible spectra, which
permit a good determination of the abundances of all of the
important coolants in the nebula.
2.4. Notes on Individual Objects
2.4.1. LMC-SMP 2
This PN is a rather faint, low-excitation object (excitation
class EC \1.1) with low expansion velocity (9.9 km s~1).
The HST image shows it to be a fairly compact object with
a steep radial density distribution. The central star is
strongly detected in the UV with deep C IV absorption and
other absorption lines visible. It is also detectable at optical
wavelengths against the nebular continuum.
The weak C II], [S II], and [N II] lines all indicate that
this PN is optically thin. The [O II] density has not been
measured directly, but the relative strengths of the [O II]
jj7320,7330 and jj3727,3729 lines indicate that the density
in the outer parts must be quite low.
An adequate Ðt is obtained with zone 2 only. The stellar
luminosity is quite low. The nebular model gives
L\1345 ^200 while a Ðt to the nebular continuumL_,
would yield L\1500 ^300 This implies a low-mass,L_.
old central star.
2.4.2. LMC-SMP 8
This PN is a low-excitation object (EC \2.5) with mod-
erate expansion velocity (25 km s~1). The absence of [S II]
and the weakness of the [N II] lines in the red suggest
optical thinness and/or high density. The ratio of the [O II]
jj7320,7330 and jj3727,3729 lines would require a very
high density, but the ratio of the near-UV lines of [O II]
jj3727/3729 indicates a density of only 5500 cm~3. The
only way to reconcile these is to have two components. One
of these must be very dense and compact, and the other
must be much less dense in the outer regions, but very much
more extended to provide the required emissivity. This is
consistent with the appearance of the HST image, which
strongly suggests a core ]halo structure.
The Ðt to the continuum yields an extraordinary result.
TABLE 1
SUMMARY OF TWO-ZONE NEBULAR MODELS FOR THE LMC PNS
NEBULAR PARAMETERS (ZONE 1) NEBULAR PARAMETERS (ZONE 2)
STELLAR PARAMETERS
n(H) Rin Rout Coverage BA R
in Rout
OBJECT L/L_LUV/L_Teff/104K(cm~3) (cm) (cm) Factor (cm~1)(cm~3) (cm) (cm) qout
LMC-SMP 02 ...... 1345 ^150 1500 ^300 3.9 ^0.15 . . . .. . . . . . . . 25000 1600 2.00E]16 1.72E]17 3.05
LMC-SMP 08 ...... 3930 ^300 (See text) 5.2 ^0.40 120000 2.00E]16 3.04E]16 0.29 20000 2000 2.00E]16 1.85E]17 0.79
LMC-SMP 20 ...... 3300 ^300 . . . 15.1 ^0.8 1000 1.00E]17 4.01E]17 0.73 2000 2300 8.00E]16 2.93E]17 ...
LMC-SMP 35 ...... 2405 ^800 . . . 10.8 ^0.8 2900 2.50E]17 3.58E]17 0.16 1000 600 2.50E]16 5.52E]17 7.8
LMC-SMP 40 ...... 1330 ^250 . . . 17.6 ^0.7 1600 1.70E]17 3.60E]17 0.50 850 . . . 1.80E]17 3.83E]17 11.2
LMC-SMP 47 ...... 4750 ^200 . . . 14.1 ^0.5 16000 4.00E]16 1.27E]17 0.50 6500 . . . 8.00E]16 2.29E]17 ...
LMC-SMP 76 ...... 5432 ^300 5500 ^500 5.8 ^0.30 100000 2.78E]16 3.97E]16 0.20 40000 10000 2.50E]16 1.18E]17 ...
LMC-SMP 87 ...... 4900 ^400 . . . 16.8 ^0.5 3500 3.60E]17 4.09E]17 0.45 1000 . . . 1.50E]17 3.86E]17 2.5
HST MAGELLANIC CLOUD PN OBSERVATIONS. V. 193
TABLE 2
SUMMARY OF ABUNDANCE DETERMINATIONS FOR LMC PNSa
Object He C N O Ne Mg Si S Ar
LMC-SMP 02 ...... 11.01 8.57 7.36 8.26 7.45 . . . . . . 6.60 6.00
LMC-SMP 08 ...... 11.28 7.93 7.48 8.26 7.32 . . . . . . 6.69 6.04
LMC-SMP 20 ...... 11.15 7.52 8.58 8.49 7.74 . . . 7.52 7.26 6.34
LMC-SMP 35 ...... 10.95 8.53 7.78 8.38 7.48 . . . 6.78 6.70 5.93
LMC-SMP 40 ...... 11.01 8.81 7.79 8.45 7.65 6.11 . . . 6.79 5.83
LMC-SMP 47 ...... 11.24 8.18 8.15 8.36 7.60 6.20 6.52 6.85 6.28
LMC-SMP 76 ...... 10.88 9.04 7.40 8.30 7.34 . . . . . . 6.60 5.90
LMC-SMP 83 ...... 11.11 7.41 8.06 8.62 7.71 . . . . . . 7.38 6.45
LMC-SMP 85 ...... 11.02 8.74 7.63 8.40 7.48 . . . . . . 6.30 6.24
LMC-SMP 87 ...... 11.25 7.90 8.95 8.56 7.90 . . . 6.90 7.11 6.18
SLMCTb............ 10.95 8.04 7.14 8.35 7.61 7.47 7.72 6.81 6.29
aGiven in terms of 12.0 ]log (x/H).
bTaken from Russell & Dopita 1992.
The stellar continuum that is observed is completely incon-
sistent with the luminosity of 3930 ^300 inferred fromL_
the nebular model. At 1300 the observed stellar contin-Ó,
uum is about 20 times too weak. Furthermore, there is a
pronounced 2200 absorption feature in the continuum.Ó
Therefore, we must conclude that the central star is heavily
reddened compared with the nebula. This could occur if the
dense reservoir of dusty neutral gas was obscuring the
central star, but not the ionized gas. The strength of the
2200 feature is inconsistent with an LMC reddening lawÓ
for this dust, and even a Galactic(Fitzpatrick 1985)
reddening law with a logarithmic reddening(Seaton 1979)
constant, c\0.48, can hardly reproduce the depth of the
2200 feature (see the resulting continuum Ðt inÓFig. 3).
This result proves that the dust ejected by PNs can be a
carrier of the 2200 feature. The most likely explanation ofÓ
the feature is that it results from a surface ““ charge slop ÏÏ
resonance on very small carbon spheres or shells, possibly
even on Buckminsterfullerene (see the discussion inC60
& Wickramasinghe Such C-rich dust is likelyHoyle 1991).
to be ejected during the carbon star phase. However, the
abundance analysis for this particular PN shows it not to be
particularly C-rich. This may simply reÑect the fact that
photodestruction of C-bearing grains has not yet had a
chance to occur, given the extremely high density and the
youth of this nebula.
2.4.3. LMC-SMP 20
This PN is a fairly large (DD0.2 pc) bipolar, high-
excitation (EC \8.1) type I object. The accuracy of the
modeling was a†ected somewhat by the loss of lock event,
which caused the G130H exposure to fail. Nonetheless, a
quite satisfactory Ðt could be obtained using only the
portion of the spectrum that remained. The nitrogen abun-
dance is likely to be somewhat less accurate for this object,
since the N IV] and the N Vlines were not observed, and
these are dominant ionization states.
2.4.4. LMC-SMP 35
This PN is a large (DD0.3 pc) shell-like intermediate-
excitation object (EC \5.4) with a high expansion velocity
(41 km s~1). At diameter, it is appreciably larger than1A.2
the FOS aperture, so that only about 40% of the total1A.0
Ñux is collected in the UV et al. As a(Vassiliadis 1996).
consequence, it is very difficult to model, since the UV spec-
trum samples a di†erent region of nebula than does the
optical, which accepts all the nebular light. These cali-
bration problems are apparent in the region of the Balmer
jump, which is too deep in the merged spectrum. The UV
lines appear to be overestimated with respect to the optical
by a factor of order 1.6.
From the scatter in the model versus observed line inten-
sities, we estimate the errors in the abundances to be D30%
for this object.
2.4.5. LMC-SMP 40
This PN is a fairly large (DD0.2 pc) high excitation
(EC \7.6) object with a very clear distorted ring morphol-
ogy and a high expansion velocity (54 km s~1). A simple,
optically thick isobaric model provides a fairly good
description of the observed spectrum. However, this model
overestimates the strengths of [O I] and [N I], underesti-
mates the C IV/C III] ratio, and predicts the [Ne V] lines far
too weak. It is clear, therefore, that a substantial contribu-
tion of optically thin gas is required, and that the object is
probably a bipolar PN seen almost pole-on. This would
provide a natural explanation for the rather high expansion
velocity. Given the large inner radius, we adopted an
isochoric density law for zone 2.
2.4.6. LMC-SMP 47
This PN is a compact bipolar type I object with moderate
excitation (EC \6.8) and a very high expansion velocity (79
km s~1). It proved to be very easily modeled, and an iso-
baric optically thick component by itself gave a good
description of the model. However, we adopted a model
with two optically thick isobaric components of di†erent
densities and inner radii to improve the Ðt to the observed
image morphology.
2.4.7. LMC-SMP 76
This PN is very compact with low excitation (EC \2.6)
and moderate expansion velocity (29 km s~1). The central
star is strongly detected, along with a number of stellar
absorption lines. The C IV line displays a pronounced P
Cygni proÐle. The nebula is characterized by a very high C
abundance, as was inferred by et al. TheVassiliadis (1996).
luminosity inferred from the photoionization model and
from the stellar UV continuum agree well with each other.
2.4.8. LMC-SMP 87
This is a very large (DD0.25 pc) optically thin type I PN
of very high excitation (EC \9.0). It proved very difficult to
model because the PN is larger than the FOS and therefore,
as for LMC-SMP 35, there is a mismatch between the UV
and optical spectra. The extreme intensity of the N V,NIV],
TABLE 3
OBSERVED AND MODEL FLUXES FOR LMC PLANETARY NEBULAE
LMC-SMP 02 LMC-SMP 08 LMC-SMP 20 LMC-SMP 35
jLINE Observed Model Observed Model Observed Model Observed Model
(Ó)IDENTIFICATION log L(Hb)\34.33 log L(Hb)\34.4 log L(Hb)\34.97 log L(Hb)\35.00 log L(Hb)\34.72 log L(Hb)\34.70 log L(Hb)\34.72 log L(Hb)\34.70
1239, 1243... N V... ... ... ... ... 428.2 ... ...
1336 ......... C II . . . . . . 7.2 6.5 . . . . . . 9.7 20.7
1483 ......... N IV] ... ... ... ... ... ... 8 5.0
1548 ......... C IV . . . . . . 7.9 3.4 . . . . . . 282 123.2
1551 ......... C IV . . . . . . 5.4 1.7 . . . . . . 196 62.5
1640 ......... He II . . . .. . .. . . . . 494 519.1 . . . . . .
1661 ......... O III] . . . . . . . . . . . . 13.3 10.4 18.2 7.3
1666 ......... O III] . . . . . . . . . . . . 43.9 25.5 26.4 17.7
1740, 1742... N III] . . . . . . . . . . . . 214 176.0 11 20.7
1884 ......... Si III] . . . . . . . . . . . . 30.3 14.9 . . . . . .
1892 ......... Si III] . . . . . . . . . . . . 14.2 17.0 8.4 9.2
19079 ........ C III] 128 116.6 63.4 81.9 111 91.8 558 636.0
2139 ......... N II] . . . . . . . . . . . . 28.3 15.0 . . . . . .
2321, 2325... [OIII],CII] 19.1 32.8 16.4 15.7 . . . . . . . . . . . .
2424 ......... [Ne IV] . . . . . . . . . . . . 160.7 122.9 54.6 . . .
2470 ......... O II] 8.5 6.2 . . . . . . 10.2 6.7 . . . 9.3
3343 ......... [Ne V] . . . . . . . . . . . . 27 16.8 8.3 . . .
3426 ......... [Ne V] . . . . . . . . . . . . 76 44.2 14.4 . . .
3727, 3729... [OII] 180.4 140.4 10.1 15.3 141.2 132.1 84.2 120.8
3869 ......... [Ne III] . . . . . . 32.2 36.2 109.6 123.7 79.2 80.5
3969 ......... [Ne III], Hv15.7 29.8 23.7 27.0 50.7 50.3 43.8 40.0
4069 ......... [S II] ... ... ... ... 8.2 5.0 ... ...
4076 ......... [S II] ... ... ... ... 1.5 1.7 ... ...
4102 ......... Hd25.7 25.9 26 26.1 28.6 26.0 26.3 26.1
4340 ......... Hc47.4 46.9 49.1 47.1 47 47.0 46.5 47.1
4363 ......... [O III] 2.9 1.9 4.8 8.1 20.5 15.8 18.4 13.8
4472 ......... He I... ... ... ... 3.6 3.8 ... ...
4686 ......... He II . . . . . . . . . . . . 67.6 74.3 18.3 22.7
4725 ......... [Ne IV] ... ... ... ... 2.3 1.2 ... ...
4740 ......... [Ar IV] . . . . . . . . . . . . 6 13.7 4.1 2.2
4861 ......... Hb100 100.0 100 100.0 100 100.0 100 100.0
4959 ......... [O III] 88.4 98.5 182.4 204.0 226.2 324.2 428.9 369.2
5007 ......... [O III] 249.2 283.9 553 587.5 651.5 934.0 1235.3 1062.5
5199 ......... [N I] . . . . . . . . . . . . 9.4 10.1 2.2 0.9
5577 ......... [O I] ... ... ... ... ... ... ... ...
5755 ......... [N II] . . . . . . . . . . . . 14.4 9.0 . . . . . .
5876 ......... He I14.2 13.6 23 23.4 10.6 10.2 9.8 9.2
6300 ......... [O I] . . . . . . . . . . . . 20.2 14.8 4.9 5.1
6312 ......... [S III] . . . . . . 2.4 2.3 5.8 6.6 1.8 2.5
6364 ......... [O I] ... ... ... ... 9 4.9 3.5 1.7
6548 ......... [N II] 13 11.9 6.9 6.0 159.4 134.4 15.7 17.2
6563 ......... Ha284.9 285.0 289.4 281.2 278.4 284.7 279 282.4
6584 ......... [N II] 37.6 35.1 20.1 17.8 468.1 395.8 46.2 50.8
6678 ......... He I3.7 3.9 7 6.6 . . . . . . 3.4 2.6
6717 ......... [S II] . . . . . . . . . . . . 24.8 18.2 2.8 3.6
6731 ......... [S II] . . . . . . . . . . . . 35.4 25.8 3.6 4.5
7006 ......... [Ar V] ... ... ... ... 4.3 3.4 ... ...
7136 ......... [Ar III] 10.1 11.2 13.8 13.1 25.9 24.0 10.5 12.5
7320 ......... [O II] . . . 4.6 13.4 9.6 . . . . . . 2.5 3.4
7330 ......... [O II] . . . 3.6 9 7.7 . . . . . . 1.5 2.7
7751 ......... [Ar III] ... ... ... ... ... ... 3.1 3.0
194
TABLE 3ÈContinued
LMC-SMP 40 LMC-SMP 47 LMC-SMP 76 LMC-SMP 87
jLINE Observed Model Observed Model Observed Model Observed Model
(Ó)IDENTIFICATION log L(Hb)\34.33 log L(Hb)\34.4 log L(Hb)\34.97 log L(Hb)\35.00 log L(Hb)\34.72 log L(Hb)\34.70 log L(Hb)\34.72 log L(Hb)\34.70
1239 ......... N V. . . . . . 54.0 65.0 . . . . . . 872.0 533.7
1243 ......... N V. . . . . . 30.2 33.1 . . . . . . 520.0 272.4
1336 ......... C II 37.6 91.8 . . . . . . 24.0 21.2 . . . . . .
1394 ......... Si IV 7.3 . . . . . . . . . . . . . . . 17.7 . . .
1400, 1401... O IV 9.9 11.0 22.9 20.5 . . . . . . 81.3 71.1
1483, 1487... N IV] 21.6 11.9 . . . 88.8 . . . . . . 756.0 672.4
1548 ......... C IV 418.0 553.0 392.0 367.4 24.1 23.4 257.0 263.2
1551 ......... C IV 311.0 279.0 251.0 185.0 11.8 11.8 142.0 132.6
1602 ......... [Ne IV] ... ... 7.7 3.5 ... ... ... ...
1640 ......... He II 454.0 559.9 314.0 274.9 . . . . . . 648.0 891.0
1661 ......... O III] 8.6 7.1 8.0 12.7 5.6 2.7 28.4 10.4
1666 ......... O III] 18.3 17.4 23.5 31.3 6.8 3.3 66.0 25.6
1740, 1742... N III] . . . . . . 70.9 62.5 . . . . . . 607.0 418.8
1884 ......... Si III] ... ... 3.9 2.8 ... ... 8.8 6.7
1892 ......... Si III] ... ... 5.2 6.1 ... ... 8.8 7.5
1907, 1909... C III] 1610.0 1585.0 893.0 472.5 841.0 623.4 294.0 232.4
2139, 2143... N II] . . . . . . 12.9 11.5 . . . . . . 54.9 28.5
2321, 2325... [OIII],CII] 243.0 143.1 125.0 29.6 155.0 59.3 36.9 11.3
2424, 2436... NeIV] 67.5 34.7 70.3 43.0 . . . . . . 179.0 163.7
2470 ......... [O II] 8.9 11.0 15.0 14.2 16.2 5.4 13.9 13.2
2800 ......... Mg II 4.5 4.8 18.9 18.8 . . . . . . . . . . . .
3343 ......... Ne V] 11.7 1.9 .. . . . . . . . . . . 26.2 25.1
3426 ......... [Ne V] 29.0 5.3 42.6 16.4 . . . . . . 104.0 66.1
3727, 3729... [OII] 233.5 289.9 47.8 105.2 20.6 19.7 211.0 108.4
3869 ......... [Ne III] 105.5 117.8 125.7 129.9 8.3 8.0 122.6 157.7
3889 ......... H, He I.. . . . . 22.8 26.8 32.9 31.6 26.1 18.5
3969 ......... [Ne III], Hv49.7 52.3 51.6 56.0 27.6 25.7 56.5 65.0
4069 ......... [S II] 10.7 4.1 7.0 4.7 . . . . . . . . . . . .
4076 ......... [S II] ... ... 4.2 1.6 ... ... ... ...
4102 ......... Hd26.7 25.9 26.4 26.0 25.7 26.0 26.5 26.0
4340 ......... Hc46.7 46.9 46.1 47.0 47.2 47.0 46.0 47.0
4363 ......... [O III] 15.4 12.0 19.1 20.1 6.6 5.2 43.9 14.9
4472 ......... He I. . . . . . 5.2 6.6 6.1 3.7 8.0 3.4
4686 ......... He II 59.2 74.6 45.2 40.4 . . . . . . 84.5 127.5
4725 ......... [Ne IV] ... ... 3.0 0.5 ... ... ... ...
4740 ......... [Ar IV] . . . . . . 5.9 8.1 . . . . . . 10.0 8.6
4861 ......... Hb100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0
4959 ......... [O III] 295.5 269.8 365.8 390.0 198.1 197.3 322.0 273.4
5007 ......... [O III] 855.8 777.1 1053.0 1123.5 565.5 568.4 913.6 787.4
5199 ......... [N I] 5.8 4.5 2.7 2.2 . . . . . . 43.5 29.6
5755 ......... [N II] ... ... 9.5 6.4 ... ... ... ...
5876 ......... He I6.6 5.7 20.2 17.7 9.6 10.1 13.0 9.3
6300 ......... [O I] 16.8 13.4 15.0 20.3 . . . . . . 24.6 19.1
6312 ......... [S III] 4.6 3.8 3.3 4.0 . . . . . . 4.6 4.4
6364 ......... [O I] 4.5 4.4 5.5 6.8 . . . . . . 7.9 6.4
6435 ......... [Ar V] ... ... 1.6 0.7 ... ... ... ...
6548 ......... [N II] 45.8 45.5 80.2 68.8 4.2 4.5 307.3 308.9
6563 ......... Ha279.1 286.9 276.1 284.5 281.9 283.2 269.3 284.7
6584 ......... [N II] 137.9 133.8 238.9 202.6 12.6 13.4 921.0 909.6
195
TABLE 3ÈContinued
LMC-SMP 40 LMC-SMP 47 LMC-SMP 76 LMC-SMP 87
jLINE Observed Model Observed Model Observed Model Observed Model
(Ó)IDENTIFICATION log L(Hb)\34.33 log L(Hb)\34.4 log L(Hb)\34.97 log L(Hb)\35.00 log L(Hb)\34.72 log L(Hb)\34.70 log L(Hb)\34.72 log L(Hb)\34.70
6678... He I. . . . . . 4.9 5.0 2.7 2.9 6.5 2.7
6717... [S II] 15.9 16.8 6.8 5.2 . . . . . . 22.5 11.3
6731... [S II] 18.7 19.7 11.4 9.8 . . . . . . 28.9 18.1
7006... [Ar V] ... ... 2.9 1.6 ... ... 3.1 1.9
7136... [Ar III] 11.3 11.2 22.3 27.4 5.3 8.3 14.1 16.4
7237... [Ar IV] ... ... 1.0 0.2 ... ... ... ...
7320... [O II] 7.2 8.2 11.4 10.5 4.8 4.1 5.0 4.5
7330... [O II] 3.9 6.5 8.1 8.4 3.9 3.2 3.2 3.6
7751... [Ar III] ... ... ... ... ... ... 3.6 4.9
HST MAGELLANIC CLOUD PN OBSERVATIONS. V. 197
FIG.2a
FIG.2b
FIG. 2.È(a) The Ðt of the photoionization model to the observed emission-line intensities for the LMC PNs modeled in this paper. In many cases the
goodness of Ðt is comparable with the measurement errors of the line intensities (see Meatheringham & Dopita, (b) Same as (a). The e†ect of1991a, 1991b).
the Ðnite FOS aperture and its associated calibration errors is most marked in the case of LMC-SMP 87.
CIV, and [Ne V] lines all show the need for an optically thin
component. However, the very strong [N I] and [O I] lines
require an optically thick region with low ionization param-
eter illuminated by a hard photon source. The di†erence
between the logarithmic reddening constants andcopt cHe II
found by et al. shows that there is likely toVassiliadis (1996)
be a calibration problem for this nebula. We believe that the
measured [O II] and [Ne V] relative Ñuxes are overesti-
mates. It is clear that this object has an enormous N abun-
dance, as has been inferred by et al. InVassiliadis (1996).
FIG. 3.È(a) A comparison of the observed continuum spectrum and the theoretical continuum model predicted by the photoionization model for those
cases in which the central star is clearly detected. Note the very deep (100%) stellar C IV absorption feature in LMC-SMP 02. (b) As explained in the text, in
the case of LMC-SMP 08, the central star appears to be heavily absorbed compared with the nebular, and the reddening law of the circumstellar dust
displays a very deep 2200 feature, presumably due to C-rich grains ejected by the central star in the late AGB phase. (c) Same as (a). Note the P CygniÓ
proÐle of the stellar C IV line in LMC-SMP 76.
HST MAGELLANIC CLOUD PN OBSERVATIONS. V. 199
TABLE 4
SUMMARY OF NEBULAR ABUNDANCES GIVEN WITH RESPECT TO SOLAR
Object He [C/H] [N/H] [O/H] [Ne/H] [Mg/H] [Si/H] [S/H] [Ar/H] [a/H]
SMP 02...... 0.102 0.01 [0.69 [0.67 [0.64 . . . . . . [0.61 [0.56 [0.60
SMP 08...... 0.192 [0.63 [0.57 [0.67 [0.77 . . . . . . [0.52 [0.52 [0.59
SMP 20...... 0.142 [1.04 0.53 [0.44 [0.35 . . . [0.03 0.05 [0.22 [0.14
SMP 35...... 0.090 [0.03 [0.27 [0.55 [0.61 . . . [0.77 [0.51 [0.63 [0.58
SMP 40...... 0.102 0.25 [0.26 [0.48 [0.44 [1.47 . . . [0.42 [0.73 [0.51
SMP 47...... 0.172 [0.38 0.10 [0.57 [0.49 [1.38 [1.03 [0.36 [0.28 [0.37
SMP 76...... 0.076 0.48 [0.65 [0.63 [0.75 . . . . . . [0.61 [0.66 [0.67
SMP 83...... 0.130 [1.15 0.01 [0.31 [0.38 . . . . . . 0.17 [0.11 [0.05
SMP 85...... 0.105 0.18 [0.42 [0.53 [0.61 . . . . . . [0.91 [0.32 [0.55
SMP 87...... 0.176 [0.66 0.90 [0.37 [0.19 . . . [0.65 [0.10 [0.38 [0.21
our modeling, we have adopted an isochoric density model
for zone 2, but the density of this region is not well deÐned
by the observations.
3.DISCUSSION
3.1. Masses of Central Stars
The photoionization modeling presented here and in
Papers and has allowed us to place accurately the PNIII
central stars onto the theoretical [log (L/L_) :log (Teff)]
Hertzsprung-Russell (H-R) diagram. We can, therefore, use
the & Wood evolutionary models toVassiliadis (1993)
determine both the current core mass and the original mass
of the central star. This analysis depends upon whether the
central star leaves the AGB as a H- or an He- burning
object, since this a†ects radically the subsequent evolution,
both in the rate of evolution across the H-R diagram, and in
the position of the central star on the H-R diagram. In
we had attempted to sort the central stars accord-Paper IV,
ing to their dynamical evolution, demanding that the
dynamical ages be consistent with the theoretical evolution-
ary age. With the improved determination of the stellar log
and log given by these models, we can now(L/L_)(T
eff)
eliminate the ambiguities, so that all except LMC-SMP 85
can now be assigned deÐnite classiÐcation as either H or He
burners. The types that we have assigned are given in Table
In we show the theoretical H-R diagrams for5. Figure 4
both H and He burners from & WoodVassiliadis (1993),
with the inferred locations of the LMC PNs.
The initial massÈcore mass relationship is critical to the
interpretation of these results. This is determined essentially
by the mass loss on the giant and AGB phases, which chan-
nels a wide range of initial stellar masses into a narrow
range of core mass. The mass-loss formulation adopted by
& Wood and used by Bressan, &Vassiliadis (1993) Marigo,
Chiosi is to be preferred, since this ensures that the(1996)
models match the period-luminosity relationship of the
long-period variables, the maximum luminosity of the AGB
stars, and the bolometric luminosity distribution of the
carbon stars observed in the LMC. Combining the results
of these two papers, we Ðnd that the relationship between
initial mass, M, and Ðnal core mass, can be expressedMcore,
as
(Mcore/M_)\0.5241 ]0.0438(M/M_)
]0.00949(M/M_)2. (3.1)
This expression Ðts the models to better than 0.01 M_
throughout the range. We also Ðt the & WoodVassiliadis
models to smooth functions of the H-burning and(1993)
He-burning lifetimes, to give the age of the star in terms of
its mass by
(q/Gyr) \11[M/M_]~3.1 ] 0.46[M/M_]~4.6 . (3.2)
We have used and equations and toFigure 4 (3.1) (3.2)
derive the initial mass, core mass, and age of each PN.
These are also given in Because of the extremeTable 5.
nonlinearity of parameters, and the dependence on location
of the PNs on the H-R diagram, full measurement errors are
given.
3.2. Chemical Evolution History of the L MC
& Jacoby and & KalerKaler (1990, 1991) Jacoby (1993)
had already shown that there is a relationship between core
mass and He and N abundance. However, it is difficult to
separate changes in abundance of elements a†ected by
dredge-up processes from the changes in abundance pro-
duced by the chemical evolution over time. andTables 4 5
show that, with these new data, we can now distinguish a
core mass:metallicity or, equivalently, an age : metallicity
relationship for the LMC using [a/H] as a metallicity indi-
cator on the grounds that this is little a†ected by dredge-up.
TABLE 5
DERIVED PARAMETERS OF THE PN CENTRAL STARSa
log (Teff) Core Mass Initial Mass Age
Object log (L/L_) (K) Type (M_)(M
_
) (Gyr)
SMP 02...... 3.128 4.591 He 0.571(]0.006 [0.002) 0.90(]0.08 [0.03) 12.2È17.8
SMP 08...... 3.594 4.716 He 0.600(]0.006 [0.005) 1.35(]0.08 [0.08) 3.7È5.4
SMP 20...... 3.519 5.179 H 0.604(]0.007 [0.007) 1.40(]0.10 [0.10) 3.2È5.0
SMP 35...... 3.381 5.033 He 0.590(]0.021 [0.013) 1.20(]0.30 [0.15) 3.2È9.8
SMP 40...... 3.124 5.246 H 0.634(]0.016 [0.008) 1.80(]0.20 [0.10) 1.30È2.16
SMP 47...... 3.677 5.149 H 0.618(]0.016 [0.011) 1.60(]0.20 [0.15) 1.81È3.56
SMP 76...... 3.735 4.763 H 0.590(]0.010 [0.010) 1.20(]0.15 [0.15) 4.5È9.8
SMP 83...... 4.431 5.230 He 0.980(]0.15 [0.13) 5.00(]1.0 [1.0) 0.04È0.15
SMP 85...... 3.857 4.653 He 0.650(]0.05 [0.02) 2.00(]0.60 [0.20) 0.57È1.8
H 0.614(]0.020 [0.014) 1.55(]0.25 [0.15) 1.9È4.0
SMP 87...... 3.690 5.225 He 0.693(]0.014 [0.018) 2.50(]0.15 [0.20) 0.54È0.84
aThe type of SMP 85 remains ambiguous.
200 DOPITA ET AL. Vol. 474
FIG.4aFIG.4b
FIG. 4.È(a) The location on the H-R diagram of PN central stars for those objects identiÐed as having left the AGB as H burners. The theoretical tracks
are from & Wood The 1.0 track is for SMC rather than LMC abundances. (b) Same as (a), but for the He burners.Vassiladis (1993). M_
These relationships are shown in andFigures 5a5b.
Because of the highly nonlinear relationship between core
mass and age, the resolution of the age becomes progres-
sively poorer as we go back in time, so the age axis is
plotted on a logarithmic scale. Despite the relatively few
number of points, reveals that in the LMC thereFigure 5
was a long period of quiescence between D15 Gyr and D4
Gyr ago, during which the metallicity remained close to
SMC values. There is no evidence in this sample of any
““ halo ÏÏ abundance objects. About 1È3 Gyr ago, there
appears to have been a strong burst of star formation that
more than doubled the metallicity. In recent times, star for-
mation and the rate of chemical evolution slowed again.
This is in remarkable agreement with the data on both
Ðeld stars and clusters. From color-magnitude diagrams of
the Ðeld stars et al. & Blanco(Hardy 1984; Frogel 1983 ;
Stryker show that in the regionButcher 1977; 1983, 1984),
of the bar of the LMC, most of the star formation occurred
in a major burst D3 Gyr ago and continued at a lower rate
until D0.1 Gyr ago. The latest cluster data et al.(Girardi
show a long period of quiescence, followed by a large1995)
rate of cluster formation 1È2 Gyr ago with a secondary
peak at D0.1 Gyr. The ““ age gap ÏÏ 3È12 Gyr ago, and a
rapid increase in metallicity 2È3 Gyr ago, is also seen in the
color-magnitude diagrams of individual clusters (Costa
Finally, recent HST data on the distribution of stars1991).
in the color-magnitude diagram for a Ðeld in the outer disk
of the LMC et al. shows that there was a(Gallagher 1996)
strong burst in star formation D2.3 Gyr ago. During this
burst, the star formation rate was enhanced by at least a
factor of 3, and possibly by as much as a factor 50.
Clearly, the results presented here show the great poten-
tial o†ered by the PNs in the investigation of the chemical
history of the LMC. should be much improvedFigure 5
when the results of cycles 4È6 are available.
3.3. T he Refractory Elements
From Tables and we see that abundances of Mg are24,
determined in two objects, and Si abundances are estimated
for four objects altogether. The LMC current interstellar
abundances derived by Russell & Dopita are(1990, 1992)
also given in Table 2. In that work, the Mg abundances
were determined from supergiant atmospheres, and the Si
abundance was determined from both supergiants and from
energetic supernova remnants, in which refractory grains
are expected to have been destroyed by sputtering (Draine
1995).
By inspection, it is clear that Mg is depleted in the ionized
gas by D1.3 dex, and Si by D1.1 dex, apart from
LMC-SMP 20. Since both Mg and Si are elements produc-
ed by the a-process, and they are essentially una†ected by
dredge-up processes occurring during the giant and the
AGB phases of evolution, we would expect that the abun-
dances of these elements in the material ejected by the
central star would reÑect those of the star at the time of its
formation. This result demonstrates, therefore, that both
Mg and Si are depleted onto refractory grain materials,
presumably silicates, which are not destroyed by the UV
Ðeld of the central star over the evolutionary lifetime of the
PN. In this regard, it is signiÐcant that the four objects in
which Si III] lines are seen are all hot type PNs. We would
expect that conditions in these objects would lead to signiÐ-
cant photodestruction of grains.
In the local interstellar medium, the observed depletions
of refractory elements are tightly correlated with the
average value of the neutral hydrogen density. In cold dense
clouds, both Mg and Si are depleted by 1.0 dex or more
Savage, & Spitzer However, in(Jenkins, 1986; Joseph 1988).
the di†use interstellar medium (ISM), Mg is almost unde-
pleted. This e†ect seems to be correlated with grain destruc-
tion by shocks in the di†use ISM. hasGondhalekar (1984)
shown that signiÐcant shock processing occurs in high-
velocity gas, and even at velocities as low as 50 km s~1,at
least 10% of the Si is returned to the ISM. It is interesting to
note that this velocity is lower than the observed expansion
velocities of both LMC-SMP 40 and LMC-SMP 47, both
No. 1, 1997 HST MAGELLANIC CLOUD PN OBSERVATIONS. V. 201
FIG.5aFIG.5b
FIG. 5.È(a) Core mass: age and (b) the metallicity: age relationship inferred from the LMC PNs. Errors in the determination of the core mass, and hence of
the initial mass, lead to fairly substantial errors in the determination of ages and loss or resolution in time at early epochs. Nonetheless, the main features of
the chemical evolution history can be clearly distinguished; a long period of quiescence, followed by a short period of activity within the past 3 Gyr, which
more than doubled the base metallicity of the LMC. Evidence of a burst of star formation and rapid enrichment at this time is also found in the study of
color:magnitude arrays of clusters and Ðeld stars (see text).
of which display large depletions in the refractory elements.
This seems to require that the gas currently in the ionized
gas has passed through an ionization front and has been
subsequently accelerated up to the observed expansion
velocity, rather than being shocked directly up to this veloc-
ity. This picture is consistent with our dynamical model,
presented above.
3.4. Thea-Process Elements
The chemical enrichment of the a-process elements
should be essentially una†ected by the dredge-up processes
discussed in the following section. This is conÐrmed by
in which we have plotted the abundances ofFigure 6,
the individual a-process elements against the mean;
([Ne/H] ][S/H] ][Ar/H])/3. The line shown is the two-
dimensional mean square best Ðt. Correlation coefficients
vary between 0.66 (Ar) and 0.83 (S). The slope of the (O:a)Ðt
is appreciably less than 1 (0.56), and that of the (S:a)Ðtis
appreciably greater than unity (1.51).
The result for O may be marginal evidence for O-N pro-
cessing, but if so, this is occurring at a much lower rate than
suggested by & Meatheringham As for S,Dopita (1991b).
there is no known dredge-up process that could cause this
e†ect. However, it is signiÐcant that the derived S abun-
dance relies heavily on the optical [S II] lines. In our spec-
trophotometry, the only other S line seen frequently is the
(temperature-sensitive) [S III]j6312 line. Thus, sulfur is not
seen in its dominant ionization stage, and systematic errors
in the abundance determination may occur. Many of the
objects showing high apparent S abundance are high-
excitation objects. In these objects, much of the [S II] line
intensity arises in a partially ionized zone ionized by the
hardest photons. & Bennett made aPetuchowski (1995)
convincing case that a signiÐcant fraction of [S II] emission
in such regions can arise by direct photoionization to
excited states by the di†use stellar continuum below 912 Ó.
This process of line emission has not so far been included in
MAPPINGS II (despite the fact that photoionization to
excited states has long been included in the photoionization
rate calculations), and it could lead to a systematic overesti-
mate of the S abundance in high-excitation objects.
3.5. Mass Dependence of Chemical Dredge-up in the L MC
The chemical evolution of the material that is ejected as a
PN shell is determined by three major dredge-up episodes
and by ““ hot-bottom ÏÏ burning & Renzini(Iben 1983 ;
& Voli summarized as follows:Renzini 1981)
1. The Ðrst dredge-up, operating as the star becomes a
red giant for the Ðrst time, is produced by the penetration of
the convective envelope into regions that are partially CNO
burned. The dredged-up material is mixed throughout the
envelope, with enhancement of the 13C and 14N abun-
dances and a decrease in 12C abundance.
2. The second dredge-up appears in the early AGB evol-
ution of stars more massive than 3È5 when theM_,
hydrogen-burning shell extinguishes, and once again the
base of the convective envelope dips into burned material.
Envelope enhancements of 4He, 14N, and 13C are produced.
3. The third dredge-up occurs in the thermally pulsing
AGB phase in which, after each He-burning pulse, the con-
vective envelope dips down, dredging up nuclear processed
material rich in 4He, 12C, and the s-process elements.
4. Hot-bottom burning occurs in the more massive AGB
stars (M[3 when convection in the stellar envelopeM_)
cycles matter through the hydrogen-burning shell during
the interpulse phase, with resultant partial CNO cycling of
the whole envelope. SigniÐcant 14N, and possibly 4He,
production may occur.
These processes are expected to be dependent upon both
mass and initial metallicity of the star.
In we show the abundances of elements likely toFigure 7,
be a†ected by dredge-up and hot-bottom burning (He, C,
and N) as a function of the abundance of the a-process
elements, as deÐned in the previous sections. Thanks to the
metallicity-age relation, the axis is a measure of both initial
metallicity and of mass of the central star.
We Ðnd a striking and systematic trend with very little
scatter in the abundances of He, C, and N with a-process
abundance and/or mass. The e†ect of the second dredge-up
is apparent at intermediate masses through the enhance-
ment of He. The enhancement of C at the low mass/
202 DOPITA ET AL. Vol. 474
FIG. 6.ÈMetallicity-metallicity plots for the a-process elements. Relative abundances are normalized to solar values, following the convention used in
stellar abundance analyses. There is no evidence for a contribution to Ne through dredge-up of 22Ne, but the slope of the correlation with O is appreciably
less than the average, while the slope of the S metallicity-metallicity relationship is steeper than the average.
abundance end indicates that the third dredge-up of C is
important for such stars in the LMC, as is well known from
studies of carbon stars. An important new result is that the
sum of the C ]N abundances shows little systematic trend,
in agreement with & Jacoby This indi-Kaler (1990, 1991).
cates that the third dredge-up is signiÐcant at all masses. In
stars of higher mass and/or abundance, hot-bottom burning
appears to be operating efficiently to produce the very high
N abundances observed, with much of the dredged-up C
converted to N.
In we compare the measured He/H abundanceFigure 8,
as a function of stellar mass (see with the results of°3.1)
et al. for LMC abundances. It should beMarigo (1996)
remarked at the outset that such a direct comparison is not
exactly fair, since we have shown in that the use of a°3.2
constant initial abundance is invalid because of the metal-
licity:age relationship in the LMC. The e†ect of the lower
initial abundance in the older, low-mass stars is to make the
third dredge-up operate much more efficiently, increasing
the C/O ratio The observed amount of He(Wood 1981).
increase is about 2.5 times that predicted for masses in the
range 2È3 The models need to produce more heliumM_.
either by more second dredge-up or more hot-bottom
burning. If the latter case prevails, a very high N enhance-
ment would be expected to accompany the He enhance-
ment.
The top panel of plots N/O against He/H for theFigure 9
observed LMC PNs and the models of et al.Marigo (1996).
From this plot, and Figure 13 of et al. it isMarigo (1996),
clear that high He/H is generally associated with high N/O,
indicating that hot-bottom burning plays a more important
role in intermediate mass stars than current models suggest.
also compares the C/O abundance ratio with theFigure 9
theoretical predictions. The agreement is again poor. For
masses greater than D2 the C/O (and N/O) versusM_,
He/H plot would be brought into much better agreement
with observation if hot-bottom burning were operating in
the models, increasing envelope He while at the same time
decreasing envelope C/O (and increasing N/O). As a Ðnal
point, we note that the PNs associated with the lowest mass
progenitors have C/O ratios much larger than the models,
showing that the third dredge-up remains more important
to lower masses than current theory predicts.
The enhancement of Ne predicted by the theoretical
models of et al. is not seen at allMarigo (1996) (Figure 9).
This enhancement is in the form of 22Ne and is produced
from 14N via the chain 14N(a,c)18F(b`,l)18O(a,c)22Ne.
et al. assume, following & Sack-Marigo (1996) Boothroyd
mann that essentially all 14N left over from the CNO(1988),
cycle is converted through to 22Ne in a helium shell Ñash.
They assume also, following et al. that theGallino (1988),
conversion of 22Ne to 25Mg via the neutron source reaction
No. 1, 1997 HST MAGELLANIC CLOUD PN OBSERVATIONS. V. 203
FIG. 7.ÈThe variation of He, C, N, and O abundances with the abun-
dance of the heavier a-process elements: [a/H] \([Ne/H] ][S/H] ]
[Ar/H])/3. Error bars have been omitted for O to increase clarity, but they
are similar to those of C and N. As discussed in the text, the x-axis can be
interpreted as a mass sequence. This Ðgure shows the systematic depen-
dence of dredge-up processes on mass and metallicity.
22Ne(a,n)25Mg is very low at low core masses. The lack of
22Ne we observe would require, on the contrary, almost
complete conversion of 22Ne to 25Mg. However, there are
two problems with this suggestion. First, the depletion
factor for Mg would have to be much larger than what we
estimated in above and, second, the observed isotopic°3.3
ratios of 24Mg, 25Mg, and 26Mg in thermally pulsing AGB
stars do not show any evidence for the dredge-up of 25Mg
& Lambert The observational evidence there-(Smith 1986).
fore suggests that the reaction sequence 14N(a,c)18F(b`,l)
18O(a,c)22Ne(a,n)25Mg must stop before signiÐcant
amounts of Ne or Mg are produced. But the sequence
FIG. 8.ÈVariation of the helium abundance with initial mass of the
central star. The curve is from et al. Clearly, much moreMarigo (1996).
dredge-up is occurring than is predicted by the theoretical models.
FIG. 9.ÈThe variation of the relative abundances of He, C, N, and O
compared with the helium abundance for the LMC PNs. The curves are
from et al. The large di†erences between theory and obser-Marigo (1996).
vation are discussed in the text.
cannot stop at 18O either, since thermally pulsing AGB
stars do not show evidence for enhancements in 18O (Smith
& Lambert On the other hand, current reaction rate1990).
estimates have the reactions 14N(a,c)18F(b`,l)18O
occurring early in each helium shell Ñash. The Ðnal pro-
ducts of the 14N]areaction are clearly not identiÐed at
the present time.
204 DOPITA ET AL.
4.CONCLUSIONS
The results presented in this paper have demonstrated
the utility of Hubble Space Telescope UV and ground-based
optical spectrophotometry in the analysis of the evolution
of planetary nebulae in the Magellanic Clouds, in examin-
ing the mass dependence of dredge-up processes, and in
inferring details of the star formation history of the Magel-
lanic Clouds. We have been able to place the central stars
on the H-R diagram and so determine core masses and (via
theoretical tracks) infer the initial masses and ages of 10
objects.
The observed abundance patterns are qualitatively con-
sistent with the (mass-dependent) operation of the various
chemical dredge-up processes as predicted by theory.
Dredge-up of C during the thermal pulsing stage appears to
be most important, and ““ hot-bottom burning ÏÏ transforms
much of this C to N in the more massive stars. There is no
sign of dredge-up of 22Ne. It is clear that, although useful
for qualitative descriptions of the dredge-up processes,
current theoretical models are very inadequate for the task
of describing these processes in detail. The observations
presented here provide an important constraint on future
theoretical models. By the time our HST program is com-
plete, we will have data of similar quality on a further 20 or
so LMC PNs, and for at least 10 SMC PNs. Therefore, we
expect to be able to determine the e†ect of lower metallicity
on these dredge-up processes, reÐne our understanding of
dredge-up processes in LMC PNs, and infer details of the
metallicity history of the SMC at about the same resolution
the results for the LMC presented in this paper.
The observed spread in the a-process element abun-
dances can be understood as being due to di†erences in core
mass of the PNn, which is directly related to initial mass of
the precursor star. On this basis, we have derived the chemi-
cal history of the LMC. This is the Ðrst time that data
derived from PNs have been used in this way. We Ðnd that
the base metallicity of the LMC almost doubled D2 Gyr
ago. This is consistent with studies of Ðeld stars and of
clusters, which show that there was a major burst of star
formation at that time. It is also consistent with the mean
kinematic age of the PN population derived by
et al.Meatheringham (1988a).
The results presented here are based on observations
with the NASA/ESA Hubble Space Telescope, obtained at
the Space Telescope Science Institute, which is operated by
the Association of Universities for Research in Astronomy,
Inc., under NASA contract NAS 5-26555. Support for this
work was provided by NASA through grant number
GO-2266 from the Space Telescope Science Institute. Aus-
tralian collaborators wish to acknowledge travel and pub-
lication support under a major grant from the International
Science and Technology Division of the (Australian)
Department of Industry, Science, and Technology. The
authors would particularly like to thank the referee, George
Jacoby, for his careful reading of the manuscript and his
suggestions which both improved the end product for the
reader and prevented the propagation of careless errors in
to print.
REFERENCES
M., Bertelli, G., Bressan, A., Chiosi, C., Fagotto, F., Greggio, L., &Alongi,
Nasi, E. 1993, A&AS, 97, 851
E., & Grevesse, N. 1989, Geochim. Cosmochim. Acta, 53,Anders, 197
M. 1989, in IAU Symp. 131, Planetary Nebulae, ed. S. Torres-Barlow,
Peimbert (Dordrecht: Kluwer), 319
L., Dopita, M. A., & Tuohy, I. R. 1985, ApJ, 297,Binette, 476
A. I., & Sackman, I.-J. 1988, ApJ, 328,Boothroyd, 653
A., Fagotto, F., Bertelli, G., & Chiosi, C. 1993, A&AS, 100,Bressan, 647
H. R. 1977, ApJ, 216,Butcher, 372
J., & Laney, C. D. 1991, in IAU Symp. 148, The MagellanicCaldwell,
Clouds, ed. R. Haynes & D. Milne (Dordrecht: Reidel), 249
C., & Marigo, P. 1996, A&A, inChiosi, press
Costa, G. S. 1991, in IAU Symp. 148, The Magellanic Clouds, ed.Da
R. Haynes & D. Milne (Dordrecht: Kluwer), 183
M. A. 1993, in IAU Symp. 155 Planetary Nebulae, ed. R. Weinber-Dopita,
ger & A. Acker (Dordrecht: Kluwer), 433
M. A., Ford, H. C., Bohlin, R. C., Evans, I. N., & Meatheringham,Dopita,
S. J. 1993, ApJ, 418, 804 (Paper I)
M. A., Ford, H. C., Lawrence, C. J., & Webster, B. L. 1985, ApJ,Dopita,
296, 390
M. A., Ford, H. C., & Webster, B. L. 1985, ApJ, 297,Dopita, 593
M. A., & Meatheringham, S. J. 1991a, ApJ, 367,Dopita, 115
1991b, ApJ, 377,ÈÈÈ. 480
M. A., Meatheringham, S. J., Ford, H. C., & Webster, B. L. 1988,Dopita,
ApJ, 327, 651
M. A., et al. 1996, ApJ, 460, 320 (PaperDopita, IV)
1994, ApJ, 426, 150 (PaperÈÈÈ. II)
B. T. 1995, Ap&SS, 233,Draine, 111
E. L. 1985, ApJ, 299,Fitzpatrick, 219
J. A., & Blanco, V. C. 1983, ApJ, 274,Frogel, L57
J. S., et al. 1996, ApJ, 466,Gallagher, 732
R., Busso, M., Piccio, G., Raiteri, C. M., & Renzini, A. 1988,Gallino,
A&AS, 104, 365
L., Chiosi, C., Bertelli, G., & Bressan, A. 1995, A&A, 298,Girardi, 87
P. M. 1984, in RAL Workshop, Gas in the InterstellarGondhalekar,
Medium (SERC:RAL-84-101), 66
E., Buonanno, R., Corsi, C., Janes, K., & Schommer, R. 1984, ApJ,Hardy,
278, 592
F., & Wickramasinghe, N. C. 1991, The Theory of Cosmic GrainsHoyle,
(Dordrecht: Kluwer)
I., Jr., & Renzini, A. 1983, A&A, 21,Iben, 271
G. H., Ciardullo, R., & Walker, A. R. 1990, ApJ, 365,Jacoby, 471
G. H., & Kaler, J. B. 1993, ApJ, 417,Jacoby, 204
E. B., Savage, B. D., & Spitzer, L. 1986, ApJ, 301,Jenkins, 355
C. L. 1988, ApJ, 335,Joseph, 157
J. B. 1976, ApJ, 210,Kaler, 843
J. B., & Jacoby, G. H. 1990, ApJ, 362,Kaler, 471
1991, ApJ, 382,ÈÈÈ. 134
S., Purton, C. R., & Fitzgerald, P. M. 1978, ApJ, 219,Kwok, L125
P., Bressan, A., & Chiosi, C. 1996, A&A, inMarigo, press
S. J., & Dopita, M. A. 1991a, ApJS, 75,Meatheringham, 407
1991b, ApJS, 76,ÈÈÈ. 1085
S. J., Dopita, M. A., Ford, H. C., & Webster, B. L. 1988a,Meatheringham,
ApJ, 327, 651 S. J., Dopita, M. A., & Morgan, D. H. 1988b, ApJ, 329,Meatheringham,
166 N., Gilmozzi, R., Macchetto, F., Adorf, H.-M., & Kirshner, R. P.Panagia,
1991, ApJ, 380, L23
M. 1995, in STScI Symp. Ser. 8, The Analysis of Emission Lines,Peimbert,
ed. R. E. Williams & M. Livio (Cambridge: Cambridge Univ. Press), 165
S. J., & Bennett, C. L. 1995, ApJ, 438,Petuchowski, 735
A., & Pottasch, S. R. 1983, A&A, 126,Priete-Martinez, 31
M. A. 1991, Ph.D. thesis, Univ.Ratag, Gro ningen
A., & Voli, M. 1981, A&A, 94,Renzini, 175
S. C., & Dopita, M. A. 1990, ApJS, 74,Russell, 93
1992, ApJ, 384,ÈÈÈ. 508
N., MacConnell, D. J., & Davis Philip, A. G. 1978, PASP, 90,Sanduleak,
621 D. 1983, ApJ, 272,Scho nberner, 708
M. J. 1979, MNRAS, 187,Seaton, 73
V. V., & Lambert, D. L. 1986, ApJ, 311,Smith, 843
1990, ApJS, 72,ÈÈÈ. 387
R. H. 1933, MNRAS, 93,Stoy, 588
L. L. 1983, ApJ, 266,Stryker, 82
1984, ApJS, 55,ÈÈÈ. 127
R. S., & Dopita, M. A. 1993, ApJS, 88,Sutherland, 253
E., et al. 1996, ApJS, 105, 375 (PaperVassiliadis, III)
E., & Wood, P. R. 1993, ApJ, 413,Vassiliadis, 641
V. 1987, A&A, 188,Wiedemann, 74
P. R. 1981, ApJ, 281,Wood, 311
P. R., Arnold, A., & Sebo, K. 1996, inWood, preparation
P. R., Bessell, M. S., & Dopita, M. A. 1986, ApJ, 311,Wood, 632
P. R., Meatheringham, S. J., Dopita, M. A., & Morgan, D. H. 1987,Wood,
ApJ, 320, 178
