The evolution of emission lines in HII galaxies

Abstract

We constructed diagnostic diagrams using emission line ratios and equivalent widths observed in several samples of HII galaxies. The diagrams are compared to predictions from new photoionization models for evolving starbursts. We find that HII galaxies from objective-prism surveys are not reproduced by models of instantaneous starbursts surrounded by constant density, ionization bounded HII regions. The observed relations between emission line ratios and Hb equivalent width (W(Hb)) can be understood if older stellar populations are not negligible in HII galaxies. Also, different dust obscuration for stars and gas and leakage of Lyman continuum photons from the observed HII regions can be important. As a result, HII galaxies selected from objective-prism surveys are unlikely to contain many objects in which the most recent starburst is older than about 5~Myr. The observed increase of [OI]/Hb with decreasing W(Hb) can result from the dynamical effects of winds and supernovae. This interpretation provides also a natural explanation of the small range of ionization parameters in giant HII regions. The [OIII]/Hb vs [OII]/Hb diagnostic diagram cannot be fully understood in terms of pure photoionization models and indicate the need for additional heating sources. The [NII]/[OII] ratio is shown to increase as W(Hb) decreases. A possible explanation is an N/O increase due to gradual enrichment by winds from Wolf-Rayet stars on a time scale of 5 Myr. (abridged abstract) Comment: Accepted for publication in A&A. LaTeX, 26 pages including figures. The photoionisation models are available at http://webast.ast.obs-mip.fr/people/schaerer/

Full text

Page 1

arXiv:astro-ph/0102228v1 13 Feb 2001
A&A manuscript no.
(will be inserted by hand later)
Your thesaurus codes are:
08(11.01.1;11.05.2;11.09.4;11.19.3;11.19.5;09.08.01; 08.23.2)
ASTRONOMY
AND
ASTROPHYSICS
The evolution of emission lines in H ii galaxies
Gra˙ zyna Stasi´ nska1, Daniel Schaerer2, and Claus Leitherer3
1DAEC, Observatoire de Paris-Meudon, 92195 Meudon Cedex, France (grazyna.stasinska@obspm.fr)
2Laboratoire d’Astrophysique, Observatoire Midi-Pyrenees, 14, Av. E. Belin, F-31400 Toulouse, France (schaerer@ast.obs-
mip.fr)
3Space Telescope Science Institute⋆, 3700 San Martin Drive, Baltimore, MD 21218 (leitherer@stsci.edu)
Received 27 october 2000 / Accepted 13 february 2001
Abstract. We constructed diagnostic diagrams using
emission line ratios and equivalent widths observed in
several independent samples of H ii galaxies. Significant
trends are seen, both in the line ratio diagrams, and in di-
agrams relating line ratios to the equivalent width of Hβ.
The diagrams are compared to predictions from photoion-
ization models for evolving starbursts. This study extends
the work of Stasi´ nska & Leitherer (1996) by including ob-
jects with no direct determination of the metallicities, and
by using updated synthesis models with more recent stel-
lar tracks and atmospheres.
We find that H ii galaxies from objective-prism sur-
veys are not satisfactorily reproduced by simple models
of instantaneous starbursts surrounded by constant den-
sity, ionization bounded H ii regions. The observed re-
lations between emission line ratios and Hβ equivalent
width can be understood if older stellar populations gen-
erally contribute to the observed optical continuum in H ii
galaxies. In addition, different dust obscuration for stars
and gas and leakage of Lyman continuum photons from
the observed H ii regions can be important. As a result,
H ii galaxies selected from objective-prism surveys are not
likely to contain significant numbers of objects in which
the most recent starburst is older than about 5 Myr. This
explains the success of the strong line method to derive
oxygen abundances, at least in metal poor H ii galaxies.
The observed increase of [O i]/Hβ with decreasing Hβ
equivalent width can result from the dynamical effects of
winds and supernovae. This interpretation provides at the
same time a natural explanation of the small range of
ionization parameters in giant H ii regions. The classi-
cal diagnostic diagram [O iii]/Hβ vs [O ii]/Hβ cannot be
fully understood in terms of pure photoionization models.
The largest observed [O ii]/Hβ ratios require additional
heating.
The [N ii]/[O ii] ratio is shown to increase as the Hβ
equivalent width decreases. A possible explanation is an
N/O increase due to gradual enrichment by winds from
Wolf-Rayet stars on a time scale of ∼ 5 Myr. Alternatively,
Send offprint requests to: grazyna.stasinska@obspm.fr
⋆Operated by AURA for NASA under contract NAS 5-26555
the relation between N/O and O/H could be steeper than
N/O ∝ O/H0.5, with a previous stellar generation more
important at higher metallicities.
Key words: Galaxies: abundances – Galaxies: evolution
– Galaxies: ISM – Galaxies: starburst – Galaxies: stellar
content – ISM: H ii regions – Stars: Wolf-Rayet
1. Introduction
Isolated extragalactic H ii regions, which we will simply
refer to as H ii galaxies, are powered by clusters of hot,
massive stars that ionize their environment. Recent analy-
ses of such objects, comparing the observed nebular emis-
sion lines, colors, and stellar features with evolutionary
models (Mas-Hesse & Kunth 1991, 1999, Schaerer et al.
1999, Stasi´ nska & Schaerer 1999) found that most have
experienced a recent, quasi instantaneous burst of star for-
mation. In addition, there is growing evidence for an older
stellar population (Telles & Terlevich 1997, Papaderos et
al. 1998, Raimann et al. 2000a). Some clues regarding the
evolution of giant H ii regions, their properties and stel-
lar content may be obtained by considering H ii galaxies
in different evolutionary stages with respect to the most
recent star formation event.
The present paper follows this approach: we compare
the emission line properties of a large sample of H ii
galaxies with those of photoionization models in which
the ionizing continuum is provided by synthetic models of
evolving starbursts. In a previous paper (Stasi´ nska & Lei-
therer 1996, hereinafter SL96), we considered a sample of
metal poor galaxies in which the oxygen abundance could
be derived directly from observations using the electron
temperature sensitive [O iii] λ4363 line. Thus the derived
abundances were unambiguous and model independent.
Unfortunately, we were automatically restricted to young
starbursts by this requirement. As the most massive stars
disappear, the radiation field gradually softens inducing a
decrease of the [O iii] emission, and the weak [O iii] λ4363
becomes undetectable after about 5 Myr. Alternatively,
the oxygen abundances can be estimated via strong line

Page 2

2 Gra˙ zyna Stasi´ nska et al.: The evolution of emission lines in H ii galaxies
methods (Pagel et al. 1979, McGaugh 1991, 1994). The
relevance and accuracy of these methods is however still
under debate. We address this question in Sect. 4.2.
Lifting the restriction to the youngest starbursts im-
mediately leads to inclusion of objects with a priori un-
known metallicities. However, the distribution in metallic-
ities should be the same for old and young objects, per-
mitting an interpretation of diagnostic diagrams based on
emission lines. This advantage of a large sample of bona
fide young and old H ii galaxies is explored in the present
work. With respect to SL96, we also use updated evolu-
tionary synthesis models with more recent stellar tracks
and stellar atmospheres as input for our photoionization
models.
The study in the present paper is complementary to
work on giant H ii regions in spiral galaxies, such as pub-
lished by Garc´ ıa-Vargas & D´ ıaz (1994), Garc´ ıa-Vargas et
al. (1995), and more recently Bresolin et al. (1999) and
Dopita et al. (2000). First, our observational samples are
based on extragalactic H ii regions, for which the equiva-
lent width of Hβ can be used as a direct first-order indica-
tor of the age of the latest starburst (Dottori 1981). Sec-
ond, our samples are biased towards dwarf galaxies with
correspondingly low average metallicities.
In Section 2 we present the observational samples, dis-
cuss possible selection effects, and display a series of ob-
servational diagrams. In particular, we show that emission
line trends are seen not only in classical emission line ratio
diagrams, but also in diagrams relating emission line ra-
tios with the Hβ equivalent width. In Section 3 we describe
the models, present our model grid, and show theoretical
diagrams for evolving starbursts. In Section 4 observations
and models are compared. The main conclusions are sum-
marized in Section 5.
2. Sample selection
2.1. The database
The prerequisites for our observational data base are the
following. We require a large sample of objects with uni-
form observation parameters. The sample must be large
since there are several factors determining the line spec-
tra, the most important being metallicity and starburst
age. We wish to use information from the He i λ5876 line,
which is directly related to the mean effective tempera-
ture of the ionizing stars (cf. Stasi´ nska 1996). Deep spec-
troscopy is required for this generally weak line. In ad-
dition, we want to make use of a direct estimator of the
starburst age, such as provided by the equivalent width of
Hβ (hereafter EW(Hβ)). While this estimator has its well-
known drawbacks (effects of dust, incomplete absorption
of the stellar ionizing photons by the observed nebula), it
is relatively independent of the assumed nebular proper-
ties, as compared with emission line ratios such as [S ii]
λλ6717+30/Hβ or [S ii] λλ6717+30/[Siii] λ9069 proposed
by Garc´ ıa-Vargas et al. (1995). However, EW(Hβ) is diffi-
cult to use for giant H ii regions with a strong underlying
old stellar component, such as in nuclear starbursts or gi-
ant H ii regions in evolved galaxies. Our sample is thus
restricted to giant H ii regions in irregular or blue com-
pact galaxies, where the old stellar component is expected
to be weaker.
The largest, homogeneous sample meeting these re-
quirements is the spectrophotometric catalogue of H ii
galaxies by Terlevich et al. (1991). It contains spectra of
425 emission line galaxies, among which about 80% are
classified as genuine H ii galaxies by the authors (as op-
posed to active galactic nuclei and nuclear starbursts).
Further imaging of these objects identified some of these
H ii galaxies as giant H ii regions in spiral galaxies (Telles,
private communication). Those objects were removed to
produce the final sample, hereafter referred to as the Ter-
levich sample, which consists of 305 objects. Most of the
observations for the Terlevich catalogue were made with
2-m class telescopes, and the typical signal-to-noise in the
continuum is 5. Note, however, that second-order contam-
ination could affect some of the line ratios for λ > 6000˚ A
(Terlevich et al. 1991).
As a control sample, we consider a much more re-
stricted collection of isolated extragalactic H ii regions
with very high quality observational data. These are the
blue compact galaxies observed by Izotov and coworkers
(Izotov et al. 1994, 1997, Thuan et al. 1995, Izotov &
Thuan 1998, Guseva et al. 2000) on 2-m and 4-m class
telescopes. Their signal-to-noise in the continuum is typi-
cally 20 – 40. The latter sample, referred to as the Izotov
sample, comprises 69 objects.
Recently, another survey of emission line galaxies
meeting our requirements has been published. This is the
catalogue of Popescu & Hopp (2000) which contains 90 ob-
jects, out of which 70 are classified as H ii galaxies. The
quality of the data is not as good as for the Izotov sample,
but the sample is more homogeneous, as explained below.
2.2. Selection effects
Objective-prism surveys of emission line objects tend to
underestimate the true proportion of objects with weak
emission line equivalent widths. On the other hand, gen-
uine, isolated giant extragalactic H ii regions will always
be discovered in such surveys if they are bright enough.
The Terlevich catalogue comprises 75% (25%) of the non-
quasar emission line objects identified in the Michigan
(Tololo) objective-prism survey, and additional biases are
not expected to be important. Therefore, one expects a
roughly uniform distribution of ages of the most recent
starbursts, and the same distribution of metallicities at
each age bin above a certain threshold in Hβ equivalent
width in the Terlevich sample.
In the case of the Izotov sample, the objects are mainly
blue compact galaxies from the First and Second Byu-

Page 3

Gra˙ zyna Stasi´ nska et al.: The evolution of emission lines in H ii galaxies3
rakan objective-prism surveys. The sample is obtained
by merging two sub-samples. One is composed of ob-
jects which had been selected for their low metallicities
(Z < Z⊙/10) as estimated from low signal-to-noise 6-m
telescope spectra. The other subsample (the 39 objects
studied by Guseva et al. 2000) is composed of objects se-
lected on the basis of broad Wolf-Rayet features seen in
the 6-m telescope spectra. The metallicities in the latter
subsample range between Z⊙/40 and Z⊙1. The Hβ equiva-
lent widths in the Guseva subsample are, on average, lower
than those in the other Izotov et al. subsample, and con-
tamination from old stellar populations is likely present at
least for some objects known as nuclear starbursts. Com-
pared to the Terlevich sample, the Izotov sample is af-
fected by strong biases which must be kept in mind when
interpreting the diagnostic diagrams shown below.
The Popescu & Hopp sample is a complete subsam-
ple of emission line galaxies found on the objective-prism
plates for the Hamburg Quasar Survey. The single selec-
tion criterion was the location of the galaxies with respect
to voids. Therefore, as for the Terlevich sample, we ex-
pect a roughly uniform distribution of ages, and the same
metallicity distribution at each age bin. The sample is
however much smaller than the Terlevich sample; there-
fore statistical fluctuations may not be negligible.
2.3. Reddening corrections
Izotov and coworkers used an iterative procedure to dered-
den their sample where the reddening constant and stellar
absorption at Hβ are determined simultaneously. Such a
procedure was feasible due to the high signal-to-noise of
the data. They assumed the Whitford (1958) extinction
law and took for the intrinsic Hα/Hβ ratio the recom-
bination value corresponding to the electron temperature
derived from [O iii] 4363/5007, when available. In most
cases, the contribution of the stellar absorption in Hβ
turns out to be only a few percent, except in some ob-
jects from the Guseva sample which includes a few nuclear
starbursts (see Guseva et al. 2000, Schaerer et al. 2000).
The line intensities in the Popescu & Hopp (2000) sam-
ple are published corrected for reddening. For the galax-
ies with a strong underlying continuum relative to the Hβ
emission (EW(Hβ) < 20˚ A), they assumed a constant
underlying absorption at Hβ of 2˚ A. The reddening was
computed from Hα/Hβ assuming an intrinsic ratio of 2.87
and using the Howarth (1983) extinction law.
In the case of the Terlevich sample, the data are pub-
lished without reddening correction. We have dereddened
them adopting the Seaton (1979) reddening law and an
intrinsic Hα/Hβ ratio of 2.87. We assumed a constant un-
derlying stellar absorption of 3˚ A for all the spectra in the
1For 9 objects the electron temperature was too low for tem-
perature sensitive ratios to be measured, and the metallicity
was estimated with the strong line method (Pagel et al. 1979).
Hα and Hβ lines, which is predicted by models for syn-
thetic absorption line spectra in starburst galaxies (Olofs-
son 1995a, Gonz´ alez Delgado et al. 2000). Unfortunately,
spectra in the Hα region have second order contamination
in many objects (cf. Terlevich et al. 1991), affecting the
reddening correction. Alternatively, one could use Hγ/Hβ
instead of Hα/Hβ to determine the reddening correction,
but this procedure is plagued by the fact that Hγ is a much
weaker line than Hα, introducing substantial errors due to
the reduced signal-to-noise, and by the lack of published
Hγ equivalent widths. In any event, for objects without
published Hα data, we used Hγ/Hβ to derive the redden-
ing, assuming that the stellar (negative) contribution to
the observed Hγ line is three times as large as for the Hβ
line.
The reddening correction introduces some uncertain-
ties into each sample. Even in the case of the Izotov sam-
ple, where dereddening was performed in a completely
self-consistent way, the reddening correction may not be
perfect. First, departures from the standard reddening law
exist (e.g. Mathis & Cardelli 1988 or Stasi´ nska et al. 1992).
Second, the intrinsic Hα/Hβ ratio in a nebula is not neces-
sarily the recombination value. In objects with high elec-
tron temperatures, collisional excitation increases Hα/Hβ
with respect to the recombination value, in a proportion
that depends on the quantity of neutral hydrogen in the
emitting regions (Davidson & Kinman 1985, Stasi´ nska &
Schaerer 1999). Adopting the same underlying absorption
of 3 or 2˚ A at Hβ for all the objects of the Terlevich and
Popescu & Hopp samples may not be justified if stellar
populations from previous star-forming events contribute
significantly to the continuum (see below). Among the line
ratios discussed in this paper, the one most affected by
reddening is [O ii]/Hβ. It is prudent to assume that the
[O ii]/Hβ ratios may well be uncertain by up to about
30%, and even more for some objects in the Terlevich sam-
ple.
2.4. Observed emission line trends
The nine panels of Figs. 1, 2, and 3 are various obser-
vational diagrams for the Terlevich, Izotov, and Popescu
& Hopp samples, respectively. Panels a through f in these
figures show the behavior of various emission line ratios as
a function of EW(Hβ). In spite of some (expected) disper-
sion, striking line trends are seen. These trends are very
similar in the three samples, and are shown for the first
time.
Some differences among the samples are likely due to
the quality of the data. The smaller dispersion in He i
λ5876/Hβ in the Izotov sample, for example, is probably
due to the better signal-to-noise. We also note that the
Terlevich and Popescu & Hopp samples contain some ob-
jects with rather large [O ii]/Hβ: ∼ 20 % of the Popescu
& Hopp sample and 3% of the Terlevich sample have
[O ii]/Hβ ≥5, while no object of the Izotov sample has

Page 4

4Gra˙ zyna Stasi´ nska et al.: The evolution of emission lines in H ii galaxies
such large values for this ratio. However, if we retain only
the 50 objects from the Terlevich sample with measure-
ments in the four lines [O ii] λ3727, [O iii] λ5007, He i
λ5876 and [N ii] λ6584, the diagrams become very similar
to those corresponding to the Izotov sample. Therefore, we
will ignore the objects with [O ii]/Hβ > 5 since such high
ratios may be attributed to poor signal-to-noise and/or to
inadequate reddening corrections.
We generally note a gradual increase in the disper-
sion and a decrease of the average value of [O iii]/Hβ as
EW(Hβ) decreases, while the He i λ5876/Hβ ratio re-
mains remarkably constant. The [O i]/Hβ ratio, on the
other hand, steadily increases with decreasing EW(Hβ),
extending the trend already noted by SL96 to lower
EW(Hβ). The [O ii]/Hβ increases and tends to level off
at EW(Hβ) about 30˚ A. The ([O ii]+ [O iii])/Hβ ratio
is rather constant at the largest EW(Hβ), and becomes
more dispersed with a tendency to decrease as EW(Hβ)
decreases. The [N ii]/[O ii] ratio shows a clear tendency
to increase as EW(Hβ) decreases, although the dispersion
is larger than in the remaining diagrams. Overall, all the
trends of emission line ratios with Hβ equivalent width
are significant and require an explanation.
Finally, panels g, h, and i in Figs. 1 – 3 show the
standard emission line ratio diagrams ([O iii]/Hβ vs
[O ii]/Hβ, [O iii]/Hβ vs [S ii]/Hβ and [N ii]/[O ii] vs
([O ii]+[O iii])/Hβ), similar to those that have been
widely used in the literature to distinguish H ii regions
from active galaxies (Baldwin et al. 1981, Veilleux & Os-
terbrock 1987, van Zee et al. 1998, Martin & Friedli 1999)
and for studies of abundance ratios in H ii regions (Mc-
Gaugh 1994, Ryder 1995, Kennicutt & Garnett 1996, van
Zee et al. 1998, Bresolin et al. 1999). For comparison, we
have plotted in Fig. 4 the same three emission line ra-
tio diagrams for the sample of giant H ii regions in spiral
galaxies from McCall et al. (1985).
The H ii galaxies of our samples clearly appear like an
extension of the giant H ii region sequence of McCall et
al. toward the high [O iii]/Hβ end and towards the low
[N ii]/[O ii] end. Some of the H ii galaxies (mainly in the
Terlevich sample) reach values of [N ii]/[O ii] which are as
high as those observed in the giant H ii regions of the cen-
tral parts of spiral galaxies. The true scatter in [O ii]/Hβ
is probably smaller than it appears in our figures: as dis-
cussed in Section 2.3, [O ii]/Hβ is strongly affected by
uncertainties in the dereddening procedure2. Note the
pronounced bending of the [O iii]/Hβ vs [O ii]/Hβ and
[O iii]/Hβ vs [S ii]/Hβ sequences toward the left at high
[O iii]/Hβ in the three H ii galaxies samples. The relation
between [N ii]/[O ii] and ([O ii]+[O iii])/Hβ is extremely
tight in all the samples we consider. This was also seen
in diagrams originally studied by the authors mentioned
2
Plotting [O iii]/Hβ vs [O ii]/Hβ from the Terlevich sample
using line ratios not corrected for reddening results in a smaller
scatter.
above. Considering that line ratios in nebulae are deter-
mined by several factors (abundance ratios, intensity and
spectral distribution of the ionizing radiation field, density
distribution of the nebular gas), the question arises as to
why the observed sequences are so narrow.
We will try to understand the observational data in
the light of the photoionization models described in the
next section.
3. The models
3.1. Technique and assumptions
The evolutionary synthesis models of Schaerer & Vacca
(1998) are used to predict the theoretical spectral energy
distributions3. They are based on the non-rotating Geneva
stellar evolution models, with the high mass-loss tracks of
Meynet et al. (1994). The spectral energy distributions for
massive main-sequence stars are those given by the CoStar
models (Schaerer & de Koter 1997) taking into account
the effects of stellar winds, non-LTE, and line blanketing.
The pure He models of Schmutz et al. (1992) are used for
Wolf-Rayet stars. The spectral energy distributions from
the plane-parallel LTE models of Kurucz (1991) are used
for the remaining stars which contribute to the continuum.
To illustrate the differences between the ionizing spec-
tra adopted here with the recent Starburst99 models (Lei-
therer et al. 1999) and the earlier Leitherer & Heckman
(1995) models adopted in SL96, we plot the ratio of the
He0and He+ionizing photons (QHe, QHe+ respectively)
to QH, the number of H ionizing photons issued from var-
ious models in Figure 5. Differences with the Leitherer
& Heckman (1995) models (left panel) are mostly due
to the updated stellar tracks, which in particular, lead
to a shorter Wolf-Rayet (WR) phase in the solar metal-
licity model shown, and thus to a more rapid decline of
the hardness of the ionizing spectrum. From this figure
it is also apparent that at solar (and higher) metallici-
ties the changes brought about by recent stellar tracks
have a rather important impact on the ionizing fluxes,
which affects photoionization models based on the older
Leitherer & Heckman (1995) spectra (e.g. SL96, Bresolin
et al. 1999). As both the Schaerer & Vacca (1998) and
Starburst99 models use the same stellar tracks, the differ-
ences (right panel) in the predicted spectra are only due
to different treatments of stellar atmospheres. Outside the
phases with WR stars (i.e. at ages<∼3 Myr or>∼6 Myr),
the Schaerer & Vacca spectrum in the He0ionizing con-
tinuum is somewhat harder due to the use of improved O
star model atmospheres (cf. also Schaerer & Vacca 1998,
Fig. 5). Differences in the He+ionizing flux (non-negligible
during part of the WR phase; cf. Schaerer & Vacca) are
due to slightly different assumptions on the link between
3The predictions from these synthesis models are available
on the Web at http://webast.ast.obs-mip.fr/people/schaerer/.

Page 5

Gra˙ zyna Stasi´ nska et al.: The evolution of emission lines in H ii galaxies5
the stellar evolution and atmosphere models and other
numerical differences.
As also apparent from Fig. 5, WR stars are predicted
to provide a non-negligible fraction of the ionizing flux, es-
pecially at high energies during the WR-rich phase, whose
duration increases with metallicity (cf. Schaerer & Vacca
1998). The predicted ionizing fluxes are likely more uncer-
tain during these phases, due to 1) the neglect of metals
in the WR atmosphere models of Schmutz et al. (1992),
and 2) uncertainties in relating the stellar interior and at-
mosphere models (cf. review by Schaerer 2000). However,
while the predicted spectra are possibly too hard at high
metallicities (see also Crowther et al. 1999, Bresolin et al.
1999), there are good indications that at low metallicities
– of main concern here – the non blanketed WR atmo-
spheres combined with the latest stellar tracks provide
a reasonable description of the ionizing spectra (e.g. de
Mello et al. 1998, Luridiana et al. 1999, Gonz´ alez Delgado
& P´ erez 2000, Schaerer 2000 ).
We used the code PHOTO to build the photoionization
models, with exactly the same atomic data as in SL96.
3.2. The model grid
We built a wide model grid in order to cover the full pa-
rameter space found in our observational samples4.
Most models are computed for instantaneous bursts
with a Salpeter initial mass function and an upper stel-
lar limit Mup = 120 M⊙. The evolution of the burst is
followed in time steps of 106yr, starting from an age of
104yr. Such a time step may be too long for a detailed
description of the Wolf-Rayet stage in the evolution of
the burst at metallicities lower than Z⊙/5 (see Schaerer &
Vacca 1998), but is a good compromise to cover the evolu-
tion until 10 Myr. A second set of models was computed
for an upper stellar limit of 30 M⊙ instead of 120 M⊙.
Another set of models was built with Mup= 120 M⊙and
a Salpeter IMF but for a star formation extended over a
period of 10 Myr.
We considered five metallicities: 2, 1, 0.4, 0.2 and 0.02
times the solar metallicity. The metallicities of the stars
and the nebular gas are assumed to be the same. As in
SL96, the metallicity is defined as the oxygen abundance
(oxygen is the main gas coolant). O/H is taken equal to
8.51 × 10−4for the solar abundance set. The abundances
of the other elements relative to oxygen are given by the
prescription of McGaugh (1991). In particular, He/H =
0.0772 + 15 (O/H) and log N/O = 0.5 log (O/H) + 0.4.
The ionized gas is assumed to be dust free.
The third main parameter that determines the emis-
sion line ratios and equivalent widths of a model is the
gas distribution. The photoionization models in our grid
4Tables giving infrared, optical and ultraviolet line intensi-
ties for all the models discussed in this paper are available on
the Web at http://webast.ast.obs-mip.fr/people/schaerer/
are spherically symmetric with constant density, and are
ionization bounded. Most models presented in this paper
are spheres uniformly filled with gas at a density of n =
10 cm−3. We also constructed some models represented
by hollow spheres with f = 3 and n = 200 cm−3, where
f is the inner radius in units of Str¨ omgren radius for a
full sphere (see Stasi´ nska & Schaerer 1997). These two ex-
treme cases are intended to highlight the sensitivity of the
results on the adopted density distribution. It can easily
be shown that such hollow sphere models have the same
ionization parameter (averaged over the nebular volume)
as the filled sphere models with n = 10 cm−3and a total
ionizing luminosity 1000 times lower.
Four values of the ionization parameter are consider-
ered: those that would be produced by a star cluster of
M⋆= 1, 103, 106, and 109M⊙5, providing a range in ion-
ization parameters of a factor 1000 at a given epoch. One
should realize that our choice of density distribution is
quite arbitrary. So far, little is known theoretically or ob-
servationally about the systematics of the evolution of the
nebular geometry with time in giant H ii regions. Our
model sequences are not necessarily meant to describe the
true evolution of a giant H ii region with time. For ex-
ample, it may well be that, due to the combined effects
of expansion or stellar winds, the ionization parameter of
giant H ii regions decreases with time more than by the
sole effect of the decrease of the number of ionizing pho-
tons. This was suggested by SL96 from the comparison of
a sample of young H ii galaxies with their grid of models.
Furthermore, the time variation of the ionization param-
eter depends on the adopted evolution of the geometry.
For example, the ionization parameter U in our models
varies as Q1/3
H
(QH being the number of stellar photons
able to ionize hydrogen per unit time) while in the models
of Garc´ ıa-Vargas & D´ ıaz (1994) and Garc´ ıa-Vargas et al.
(1995), who constructed shells with a fixed outer nebular
radius, U varies as QH.
Figures 6,7 and 8 correspond to our full sphere model
sequences for instantaneous bursts with Mup = 120 M⊙
and M⋆ = 109, 106, and 103M⊙, respectively. Figure 9
corresponds to a M⋆= 106M⊙burst uniformly extended
over a period of 10 Myr. (In this case the time step is 2
Myr instead of 1 Myr.) Figure 10 is for a M⋆= 106burst
with Mup= 30 M⊙. Figure 11 corresponds to our hollow
sphere model sequences with Mup= 120 M⊙ and M⋆=
109M⊙. Each model is represented by a symbol: circle for
Z/Z⊙= 2, square for Z/Z⊙= 1, triangle for Z/Z⊙= 0.4,
diamond for Z/Z⊙= 0.2 and plus sign for Z/Z⊙= 0.05,
and the successive models in a sequence are linked by a
thin line.
It is important to realize that, while the model predic-
tions for optical forbidden lines like [O iii] λ5007, [O ii]
λ3727, [N ii] λ6584 are relatively robust and accurate for
metallicities below solar, the situation worsens as one goes
5A lower IMF mass cut-off of Mlow = 0.8 M⊙ was adopted.