close

Вход

Забыли?

вход по аккаунту

?

306647

код для вставкиСкачать
THE ASTROPHYSICAL JOURNAL, 511 : 374�8, 1999 January 20
( 1999. The American Astronomical Society. All rights reserved. Printed in U.S.A.
QUANTITATIVE NEAR-INFRARED SPECTROSCOPY OF Of AND WNL STARS
BRUCE BOHANNAN1
Kitt Peak National Observatory, P.O. Box 26732, Tucson, AZ 85726-6732
AND
PAUL A. CROWTHER1
Department of Physics and Astronomy, University College London, Gower Street, London, WC1E 6BT, UK
Received 1998 February 16 ; accepted 1998 August 25
ABSTRACT
From new high signal-to-noise ratio (S/N) 1�2 km spectroscopy of nine extreme early-type stars�
including O Iaf, O Iafpe and WN9 types葁e determine stellar parameters from detailed atmospheric
analysis and evaluate results from near-IR analogues of well-known spectral diagnostics in the optical.
We conclude that accurate stellar parameters can be measured from near-IR spectroscopy alone, an
analysis technique important to studies of luminous stars in the Galactic center and other galaxies.
Derived stellar parameters萴ass-loss rates, luminosities, surface abundances, temperatures萻how good
agreement between optical and near-IR analyses, provided that IR data are of sufficient spectral
resolution (R [ 2000) and S/N (S/N [ 30). Wind velocities derived from He I 1.0830 km are consistent
with those from ultraviolet P Cygni pro衛es. Temperatures 200�00 K systematically lower are determined from the near-IR diagnostics, a di?erence not signi衏ant in determining the stellar properties of
these objects ; which set of spectral lines provides the more accurate physical parameters萶ptical or
IR萩annot at present be ascertained. The strength of He I 2.0581 km is very sensitive to the extreme
ultraviolet energy distribution where line blanketing by heavy elements plays an important role ; this line
should not on its own be considered a reliable temperature diagnostic. The three peculiar, extreme
emission-line stars萾he O Iafpe stars HD 152386, HD 152408, and HDE 313846萢re more similar in
both morphological and physical characteristics to WNL-type Wolf-Rayet stars than to normal O Iaf
supergiants and should be classi衑d as W-R. Their classi衏ation should be WN9ha, in which they
remain a unique subgroup.
Subject headings : infrared : stars � stars : early-type � stars : mass loss � stars : Wolf-Rayet
1.
INTRODUCTION
in a galaxy must be known in order to constrain models of
stellar structure through the late stages of massive-star evolution. Morphological classi衏ation alone poorly discriminates physical characteristics in the Of-WNL transition
region, and direct measurement of physical parameters
must be made. This situation is well illustrated by the three
Galactic O Iafpe stars, HD 152386, HD 152408, and HDE
313846, which Walborn (1982) concluded should be considered as extreme late-O supergiants because their characteristics are more similar to those of the Of type than the
WN. HDE 313846, originally classi衑d as O7 : Iafpe by
Hutchings (1979), has since become accepted as a WN9type Wolf-Rayet star (van der Hucht et al. 1981), while HD
152408 has remained as O8 : Iafpe (Walborn 1972) and HD
152386 as O6 : Iafpe (Walborn 1973 ; Leep 1978). Walborn
(1982) argued that, because of essentially identical spectroscopic appearance, all three stars must be given the same
spectroscopic category. We will apply the atmosphere
analysis evaluated here along with spectroscopic criteria to
arrive at an appropriate spectral classi衏ation for these
three peculiar, extreme emission-line stars.
Crowther & Bohannan (1997 ; hereafter Paper I) concluded from model atmosphere萣ased analysis of ultraviolet and optical spectroscopy that HD 152408 had
stellar-wind properties more like HDE 313846 and other
WNL stars than would be found in the normal late O Iaf�
type supergiant HD 151804 (O8 Iaf) and proposed that HD
152408 should be considered, like HDE 313846, a W-R star
with a similar WN9ha subtype. We showed that the distinguishing characteristic of the O Iaf and WNL spectral
classi衏ations is the surface mass 製x (M0 /4nR2 ). HD
*
While only a tiny fraction of the energy of hot stars is
emitted in the near-infrared (1�km), understanding the
physics of spectral lines in the near-IR and evaluation of
their e?ectiveness as diagnostics of hot star atmospheres are
critical for investigation of luminous hot stars that are
obscured by interstellar dust at ultraviolet and optical
wavelengths but readily accessible in the near-IR (e.g., the
Galactic Center He I emission-line sources of Krabbe et al.
1991). In addition, near-IR spectroscopy combined with
low-order adaptive optics will soon a?ord improved spatial
resolution relative to optical wavelengths for spectroscopic
observations of crowded 衑lds in external galaxies. Our
goal here is to evaluate near-IR spectroscopic diagnostics of
stars that have physical parameters bridging the Of and late
WN (WNL) morphologies萢 problematic distinction in
stellar evolution萬or accurate measurement of the key
stellar parameters of temperature, mass-loss rate, and wind
velocity, parameters that are the distinguishing features of
these two stellar types.
The path or paths between hydrogen-burning Of supergiants and helium-burning WN stars2 is crucial to understanding massive-star evolution, particularly in de衝ing the
role that metallicity plays. The true population of each type
1 Visiting Astronomer, Cerro Tololo Inter-American Observatory.
CTIO is operated by AURA, Inc. under contract to the National Science
Foundation.
2 Although some WNL stars have been proposed to be in a phase of
core H burning (see, e.g., de Koter, Heap, & Hubeny 1997).
374
NIR SPECTROSCOPY OF Of AND WNL STARS
152386 was not included in our original analysis (Paper I)
because it could not be observed in the UV at high
resolution with the International Ultraviolet Explorer (IUE)
for measurement of terminal wind velocity, which makes it
a perfect candidate for IR observations and an excellent
prototype for our test of IR diagnostics of stellar properties.
Critical hot-star diagnostics in the near-IR include the
metastable He I transition at 1.0830 km (2p 3Po� 3S),
which provides a measure of the wind terminal velocity
(Howarth & Schmutz 1992), and the He I 2.0581 km
(2p 1Po� 1S) line, which reveals information about the
extreme-ultraviolet energy distribution (Najarro et al. 1994 ;
Crowther, Hillier, & Smith 1995a, 1995b). The numerous
recombination lines in the near-IR allow more accurate
abundance studies of Wolf-Rayet stars in the IR than in the
optical (see, e.g., Eenens & Williams 1992 for WC stars ;
Crowther & Smith 1996 for WN stars) ; abundance determination from spectral features in the optical is less accurate
because the Balmer and Pickering lines are often blended
with other elements that cannot be readily modeled (e.g., N
IIIjj4859�97 with Hb) and carbon transitions in the
ultraviolet su?er large optical depth e?ects. To date,
near-IR spectroscopy of O-type stars has mostly been of a
qualitative nature signi衏ant to the de衝ition of spectral
morphology in this wavelength region (see, e.g., Hanson &
Conti 1994 ; Conti et al. 1995).
Throughout this paper we shall refer to HD 152386, HD
152408, and HDE 313846 as the three O Iafpe stars,
although HDE 313846 has, as noted above, come to be
accepted as a WN-type W-R star. In � 2 we present new
near-infrared and optical observations of representative Of
and WNL stars, and in � 3 near-IR diagnostics are used to
determine the atmospheric parameters of these stars. In � 4
we discuss the appropriate spectral classi衏ation of the
O Iafpe stars, peculiar extreme Of or WN9ha-type W-R.
In � 5 we consider the reliability of quantitative near-IR
spectroscopy of Of and WNL stars and compare the stellar
properties of the WN9ha stars to representative stars of
Of and WNL types.
2.
OBSERVATIONS
New spectroscopic observations were obtained of the
three O Iafpe stars菻D 152386, HD 152408, HDE
313846萾wo WN9h stars, and four O Iaf supergiants.
Table 1 lists the program stars with spectral types, photometry taken from the recent literature, and various quantities necessary for the atmospheric analysis. In this section
we summarize the new observations, provide details on the
spectroscopic appearance of these stars, comment on the
determination of terminal wind velocities from various
spectral features, and compare interstellar reddening results
obtained exclusively from the near-IR.
2.1. New Spectroscopic Observations
Our analysis is based on new near-infrared spectroscopy
from Cerro Tololo Inter-American Observatory (CTIO)
and the United Kingdom Infrared Telescope (UKIRT)
combined with optical spectroscopy from the 74 inch telescope at Mt. Stromlo Observatory (MSO), the AngloAustralian Telescope (AAT), and the Isaac Newton
Telescope (INT).
At CTIO we used the Infrared Spectrometer (IRS) on the
4.0 m Blanco telescope on 1996 March 2� The IRS was
375
con術ured with the 210 line mm~1 grating in 2nd (K@-band)
and 4th (J-band) orders. Four individual settings were used
to cover He II 1.012 km (0.995�025 km), He I 1.083 km
(1.070�099 km), He I 2.058 km (2.016�075 km), and Brc
(2.145�203 km). A slit width of 1A yielded 3 pixel spectral
resolution of approximately 100 km s~1. Data reduction
was performed within IRAF using CTIO pipeline tools,
with special care taken to remove telluric features through
division by suitable standard stars (HR 1936 [G5 IV] in the
J-band and HR 6197 [F5 Iab] in the K@-band) ; stellar
features having 衦st been removed as necessary. A signalto-noise ratio (S/N) of 70�0 was achieved for Galactic
targets, while that for the LMC star BE 381 was D25.
We also acquired new K@-band observations of HDE
313846 and NS 4 at UKIRT on 1996 May 4 using the
cooled grating spectrograph CGS4 with the 150 mm
camera and a 150 line mm~1 grating to cover 2.045�205
km. A slit width of 1A yielded a 2 pixel spectral resolution of
100 km s~1, with a signal-to-noise ratio of between 100�0
achieved. Observations were bias corrected, 補t 衑lded,
extracted, and sky subtracted using CGS4DR (Daly &
Beard 1992). Subsequent reductions and analysis were
carried out using FIGARO (Meyerdierks 1993) and DIPSO
(Howarth et al. 1995). In order to remove atmospheric
features, the observations were divided by the standard star
BS 7034 (F7 V), with spectral features arti衏ially removed.
We undertook complementary optical spectroscopy of
the stars in this program in 1995 April at the AAT, in 1995
June at MSO, and in 1996 July at the INT. The University
College London (UCL) echelle spectrograph (UCLES) was
used at the AAT to observe HD 152386 with a 31.6 line
mm~1 grating and a 10242 pixel Tektronix CCD. Complete
spectral coverage of jj3820�30 required three echelle settings ; a slit width of 1A. 5 yielded a 2 pixel spectral resolution
of D8 km s~1.
The Coude� spectrograph was used on the MSO 74A (1.9
m) telescope with a 20482 pixel Tektronix CCD and a 600
line mm~1 grating in 衦st order. Three overlapping settings
resulted in near complete spectral coverage between
jj3860�40 at a 2 pixel resolution of 70 ^ 10 km s~1. Flux
calibrated MSO-coude� spectroscopy of HD 152386 allowed
a valuable check on the reliability of the AAT echelle data
reduction. Further details relating to the instrumental setup
and reduction technique for our MSO and AAT observations are described in Paper I.
NS 4 was observed at the INT with the Intermediate
Dispersion Spectrograph with the 235 mm camera, a 10242
pixel Tektronix CCD (24 km pixels), the 1200Y grating in
four overlapping settings, and a 1A. 5 slit to yield complete
spectral coverage between jj3650�60 at a 2 pixel
resolution of 70�0 km s~1. Spectrophotometry of NS 4
was also obtained with a nearly identical setup萿sing
instead the 300V grating萾o cover the same spectral region
in a single integration with a 5A slit (D1000 km s~1
resolution) in seeing conditions of 1A. 3. The INT spectrophotometry of NS 4 was convolved with appropriate 衛ter
pro衛es to measure broadband B and V magnitudes (Table
1). Our spectrophotometry translates to narrowband ubvr
Smith (1968) photometry of v \ 13.15, b [ v \ 1.91, and
v [ r \ 1.57. Di?erences in magnitude and color we determine from those of Massey (1984 ; v \ 12.92, b [ v \ 1.83
and v [ r \ 1.39) appear to be real since equivalent v-band
measures for AS 306 (WR 116, WN8h), obtained under
identical conditions, di?er by only 0.04 mag.
He 3�49
HR 6245
He 3�78
HR 6272
He 3�20
MR 78
WR 108
NS 4
BE 381
150958
151804
152386
152408
153919
163758
313846
O6.5 Ia(n)f`
O8 Iaf
O6 : Iafpe
O8 : Iafpe
O6.5 Iaf`
O6.5 Iaf
WN9ha
WN9h
WN9h
Spectral
Type
7.30
5.22
8.13
5.77
6.53
7.34
10.16
12.38
13.26
V
(mag)
0.36
0.07
0.51
0.16
0.25
0.01
0.68
2.09
[0.05
B[V
(mag)
6.39
5.06
6.92
5.21
5.75
7.27
7.75
7.07
J
(mag)
6.24
4.92
6.74
5.02
5.67
7.25
7.40
6.33
H
(mag)
6.13
4.86
6.61
4.92
5.58
7.24
7.13
5.78
K
(mag)
1, 2
1, 2
3, 4
1, 2
3, 4
3, 4
5, 6
5, 6, 7
8
References
1.6
0.7
4.7
3.1
2.3
0.7
5.8
6.2
6.4
He II j4686
(A� )
0.7
0.2
1.6
3.3
0.8
0.4
4.0
18.3
13.1
He I j5876
(A� )
8.2
11.2
18.9
25.1
12.6
4.8
21.9
49.4
52.2
Ha
(A� )
0.8
[1.1
6.2
2.1
2.6
[0.5
7.8
6.6
4.0
He II j10124
(A� )
10.8
9.0
21.1
40.4
11.5
5.0
49.7
183.0
120.0
He I j10830
(A� )
8.7
7.6
14.7
21.1
12.1
4.6
22.0
54.9
50.6
Brc
(A� )
NOTE.萅ear-IR photometry is from the references quoted, and selected emission-line equivalent widths are from this investigation.
REFERENCES.�(1) Walborn 1972 ; (2) The�, Wesselius, & Janssen 1986 ; (3) Walborn 1973 ; (4) Williams, van der Hucht, & The� 1987 ; (5) Smith et al. 1996 ; (6) Crowther et al. 1995a ; (7) see � 2.1 ; (8)
Leitherer & Wolf 1984.
Other Name
HD(E)
TABLE 1
SPECTROSCOPIC AND PHOTOMETRIC OBSERVATIONS FOR THE STARS ANALYZED
NIR SPECTROSCOPY OF Of AND WNL STARS
2.2. Comments on Near-IR Spectroscopy and Near-IR
Analogues to Optical Spectral Diagnostics
Previous near-IR observations of the program stars and
those of similar spectral type have generally been made at
somewhat lower spectral resolution, a resolving power not
sufficient to discern critical spectral features and to enable
accurate determination of stellar parameters. Hanson &
Conti (1994) and Hanson, Conti, & Rieke (1996) presented
lower resolution K@-band observations of our program Of
stars, except HD 163758. Crowther & Smith (1996) included
I-, H-, and K-band spectra of HDE 313846 and NS 4 in
their morphological comparison of Galactic WN stars.
McGregor, Hyland, & McGinn (1989) previously presented
K-band spectroscopy of BE 381 ; Howarth & Schmutz
(1992) and Eenens & Williams (1994) have made observations of HDE 313846 (I-band) and NS 4 (K@-band)
respectively.
A montage of near-IR spectroscopy (Fig. 1) illustrates the
appearance of photospheric absorption and wind emission
features of our set of stars. Absorption features switch to
emission (e.g., He II 1.012 km, Pc) in the most ?? extreme
stars, 舷 the O Iaf`, O Iafpe and WN9 types. As we previously concluded from optical spectroscopy (Paper I), little
in the near-IR distinguishes the O Iafpe supergiants from
377
each other. The He II features of NS 4 and BE 381 are
somewhat similar to HDE 313846 ; both show much stronger H I and He I, in keeping with their optical spectral
morphology (� 2.3).
We demonstrate in Figure 2 that given sufficient spectral
resolution, corresponding near-IR lines exist for the optical
classi衏ation features. For example, He II 2.1885 km (10�
shows an identical morphology to the important spectral
class discriminator He II j5412 (7�; Crowther et al. 1995a),
suggesting that for late Of stars these lines are formed in
similar regions in the stellar atmosphere. Note that these
features have clear P Cygni pro衛es for HD 152386 (O6 :
Iafpe), unlike other late Of stars (� 4). In contrast, Conti et
al. (1995) found that these lines behave di?erently in some
early Of types, with He II 2.1885 km in emission and He II
j5412 remaining in absorption, suggesting di?erent line formation regions for these transitions in early Of stars.
He II 1.0124 km and Brc show comparable morphologies,
albeit weaker, to the optical diagnostics He II j4686 and Ha,
respectively.
Figure 3 shows the strong correlation in strength between
optical and near-IR hydrogen and helium classi衏ation
diagnostics. He I 2.0581 km emission does not follow these
correlations, however ; compare BE 381 and NS 4 in Figure
FIG. 1.萐pectral comparison of I and K@-band spectra of our program stars obtained with the 4 m CTIO-IRS. The smooth morphological sequence from
Of to WN9 is apparent with the exception of the unexpected strength of He I 2.0581 km emission in NS 4 (see � 3.2). Successive stars are shifted vertically by
0.5 continuum units. The peak intensity of He I 1.0830 km in NS 4 reaches 8.5 times the local continuum ; that for BE 381 is 10 times.
378
BOHANNAN & CROWTHER
Vol. 511
FIG. 2.菴omparison of selected IR pro衛es with their optical counterparts. The peak intensities of He I jj5876, 10830 and Ha in NS 4 are, respectively,
2.8, 8.4, and 4.3 times the local continuum.
1. As we will discuss in � 3 the line strength of this transition
is particularly sensitive to the extreme UV energy distribution.
2.3. Optical Spectroscopy of HD 152386 and NS 4
Beyond basic morphological descriptions, HD 152386
and NS 4 have received little quantitative consideration.
Walborn (1973) classi衑d HD 152386 as O6 : Iafpe and
grouped it with HD 152408 and HDE 313846 as one of
three O Iafpe stars. Leep (1978) proposed a marginally
earlier spectral type of O5.5 If but noted that the P Cygni
nature of He I j4472 complicated any attempt at a precise
O-type classi衏ation. She remarked upon the spectroscopic
similarity in He II j4686 emission between it and the Carina
WN stars. NS 4萶ne of the weakest lined WNL stars
known in the Galaxy葁as the 衦st WN9h-subtype identi衑d in the Galaxy (Smith, Crowther, & Willis 1995) and is
now a classi衏ation standard for that subtype (Smith,
Shara, & Mo?at 1996).
The spectral morphology of HD 152386 closely resembles
HDE 313846 in the key spectral regions around Hd and He
II j4686 with critical di?erences from Of stars (see Fig. 8 in
Paper I). For example, in HD 152386 N III jj4097�03 are
in emission, in contrast with normal Of stars. HD 152386
shows slight spectroscopic di?erences from HD 152408 and
HDE 313846, di?erences consistent with a range of temperatures and wind characteristics in stars whose spectra
are dominated by strong wind features (see � 3) ; He II j4686
has broader emission, He I j5876 has a shallower P Cygni
absorption, and overall Balmer emission is weaker. As
mentioned earlier, He II j5412 is observed as a P Cygni
pro衛e, in common with HDE 313846, while a blueshifted
absorption pro衛e is observed for HD 152408. The radial
velocity of HD 152386 from narrow, metal emission
featuresformed close to the stellar photosphere was
measured to be ]44 km s~1.
NS 4 closely resembles the LMC WN9h stars BE 381 and
Sk [69�9c (a classi衏ation standard for WN9h). Crowther & Smith (1997) have recently discussed the difficulty in
accurately classifying some WNL stars in the Galaxy due to
the N IV j4058 classi衏ation line being difficult to observe in
heavily reddened objects. This problem has been partially
FIG. 3.菴orrelation between observed optical and near-IR hydrogen and helium emission-line strengths (W in A� ) for O Iaf (circles), O Iafpe (triangles),
j
and WNL (squares) stars.
No. 1, 1999
NIR SPECTROSCOPY OF Of AND WNL STARS
379
FIG. 4.菴omparison of optical spectra of NS 4 (WR 105, WN9h) with two stars of similar morphology. NS 4 most closely resembles the LMC star BE
381 (WN9h) and is clearly of lower excitation than Galactic WN8 stars such as HD 86161 since N IV j4058 emission is absent. All data are plotted to the same
scale, successively shifted vertically by one continuum unit.
avoided in the WN classi衏ation scheme of Smith et al.
(1996) where He II j5412 and He I j5876 are the primary
criteria. NS 4 was originally classi衑d as WN8 (Massey &
Conti 1983), a classi衏ation revised to WN9h by Smith et al.
(1996) using an unpublished spectrum P. A. C. obtained in
1994 June with the WHT-ISIS. In Figure 4 we present new
blue INT-IDS spectra of NS 4 and, for comparison, BE 381
(Crowther et al. 1995a) and the Galactic WN8h star HD
86161 (Crowther, Hillier, & Smith 1995b, 1995c). A WN9h
classi衏ation is certainly appropriate for NS 4 since N IV
j4058 is absent, with N III jj4634�41 and He II j4686
emission present, though relatively weak, and strong
Balmer emission.
2.4. T erminal W ind V elocities
The terminal wind velocity of early-type stars is most
accurately measured from the blueward edge of the absorption trough in the UV P Cygni resonance pro衛es of C IV or
Si IV (Prinja, Barlow, & Howarth 1990) ; no accurate wind
velocity diagnostic is available for hot stars in the optical.
Most measures of terminal wind velocity come from IUE ;
measurement must be made in the high-resolution mode to
be of sufficient accuracy for atmospheric analysis. For earlytype stars that were too faint for high-resolution IUE
observations萫ither because of their distance or because of
interstellar extinction萢nd that have only low-resolution
IUE spectra available, Prinja (1994) provides an empirical
relation (with an accuracy of D400 km s~1) to transform
low-resolution IUE measurement of the absorption
minimum to the scale of the violet limit of ?? black 舷 absorption troughs in high-resolution IUE spectroscopy.
For very distant or extremely heavily reddened O stars
the ultraviolet is either very difficult to observe or not available at all and the near-IR is the obvious spectral region to
use for measurement of terminal wind velocities. Outside of
the UV, the most accurate transition for wind velocity
determinations is He I 1.0830 km ; the lower level of this
transition is metastable and the line is therefore formed in
the outermost regions in the wind. The superiority of this
transition to He I j5876萢s previously illustrated for W-R
stars by Howarth & Schmutz (1992) and Eenens &
Williams (1994)萯s well illustrated by HD 151804. In this
star, P Cygni absorption for j5876 reaches 95% of the local
continuum at D400 km s~1 in contrast to D1300 km s~1
for 1.0830 km, a value that is in good agreement with that
derived from ultraviolet resonance line measurements (1445
km s~1 ; Prinja et al. 1990).
In Figure 5 we compare velocities from our He I 1.0830
km observations (the terminal velocity is de衝ed to be the
velocity at which 95% of the local continuum is achieved in
the absorption component) of our set of stars with previous
determinations from UV resonance line measurements
(Prinja et al. 1990 ; Pasquali et al. 1997) or values based on
lower resolution near-IR data (Eenens & Williams 1994). In
the cases of HD 152386 and HD 150958, for which only
low-resolution IUE observations are available, we include
estimates (and sizeable error bars) following Prinja (1994).
Overall, terminal velocities from He I 1.0830 km line pro衛es
of the stars analyzed here show excellent agreement with
previous determinations (Fig. 5). For HD 163758 (O6.5 Iaf),
the near-IR terminal velocity is lower than that obtained
FIG. 5.菴omparison of terminal wind velocities derived here from He I
1.0830 km observations with previous UV measurements (衛led circles)
from Prinja et al. (1990) and Pasquali et al. (1997) or from lower resolution
near-IR measurements (open triangles) by Eenens & Williams (1994). We
have used the relation from Prinja (1994) for cases in which only lowresolution IUE observations are available (HD 152386 and HD 150958).
The terminal wind velocity is consistently de衝ed to be the velocity at
which 95% of the local continuum is achieved in the P Cygni absorption
component.
380
BOHANNAN & CROWTHER
Vol. 511
TABLE 2
INTERSTELLAR REDDENING AND ABSOLUTE MAGNITUDE FOR THE PROGRAM STARS
HD(E)
V
150958 . . . . . .
152386 . . . . . .
152408 . . . . . .
313846 . . . . . .
NS 4 . . . . . . . .
BE 381 . . . . . .
7.30
8.13
5.77
10.16
12.38
13.26
E
B~V
0.60
0.80
0.42
1.15
2.40
0.30
Reference for
E
B~V
1
1
3
4
1
4
m[M
Reference for
m[M
12.0
12.7
11.4
13.5
11.0
18.5
2
3
2
5
2
5
M
V
[6.6
[7.1
[7.0
[7.0
[6.3
[6.2
NOTES.萊eddening and magnitude determined from 衪ting of theoretical energy distributions with observed spectrophotometry or broadband photometry. All values are in mag. The
distance modulus corrected for extinction is denoted by m [ M. We have estimated the
distance modulus to HD 150958 from membership in NGC 6204 and to NS 4 from Sgr OB1,
both from the catalog of Humphreys 1978. Absolute magnitudes were calculated with R \
A /E
\ 3.2.
V B~V
REFERENCES.�(1) Lundstro巑 & Stenholm1984 ; (2) Humphreys 1978 ; (3) this work ; (4)
Paper I ; (5) Crowther et al. 1995a.
from UV P Cygni pro衛es because of its weaker He I signature compared with that of other more extreme Of stars.
2.5. Interstellar Reddening and Distance
We determine interstellar extinction using the standard
method of 衪ting theoretical continua to dereddened UV-IR
energy distributions (Paper I). Results from this technique,
including those taken from Paper I, are in Table 2 ; representative 衪s for NS 4 and HD 152386 are shown in Figure
6. We describe below an alternative estimate of interstellar
extinction based on IR photometry alone.
Since ultraviolet and optical observations may not be
feasible for near-IR sources because of heavy reddening, we
test a second method of measuring interstellar reddening
based on IR photometry alone. We obtain the extinction at
the K-band, A , using the theoretical intrinsic colors from
K
our IR-only spectroscopic
analysis (model K as de衝ed in �
3.2), together with the observed (J [ K) colors from Table 1
via A \ 0.61 ] [(J [ K) [ (J [ K) ] (Howarth 1983).
K
0
Interstellar
reddenings determined exclusively
from near-IR
measurements, EK , are compared with those based addiB~Voptical observations, E
tionally on UV and
, in Table 3.
B~V from excluWe obtain slightly lower interstellar reddenings
sively near-IR observations, typically D0.1 mag (equivalent
to D0.04 dex lower luminosity).
HD 152386 lies along a line of sight close to Sco OB1
(Humphreys 1978), which includes HD 151804 and HD
152408, but is not a member of this association. We are able
to estimate its distance from the velocity structure of the
FIG. 6.萒heoretical continuum 衪s to observations of NS 4 and HD
152386 for model V (solid line) and model K (dotted line). A small variation
in temperature dramatically changes the EUV 製x with an associated
change in He I 2.0581 km emission. Energy is in ergs cm~1 s~1.
TABLE 3
INTERSTELLAR REDDENING FOR THE PROGRAM STARS FROM IR PHOTOMETRY
ONLY
HD(E)
(J [ K)
(J [ K)
150958 . . . . . .
152386 . . . . . .
152408 . . . . . .
313846 . . . . . .
NS 4 . . . . . . . .
0.26
0.31
0.29
0.62
1.29
[0.07
]0.01
]0.03
]0.01
]0.16
0
A
K
0.20
0.18
0.16
0.37
0.69
EK
B~V
0.62
0.57
0.50
1.15
2.14
E
B~V
0.60
0.80
0.42
1.15
2.40
*E
B~V
]0.02
[0.23
[0.08
]0.00
[0.26
NOTES.華ll values are in mag. EK ( \ 3.1 ] A ; Howarth 1983), and conKE
tinuum 衪s are to the full UV, optical,B~V
and IR data set,
. Di?erences between
B~V
the two methods are indicated by *E
. An estimated uncertainty
of calculating
B~V
reddening from continuum 衪s is B0.05�1 mag.
No. 1, 1999
NIR SPECTROSCOPY OF Of AND WNL STARS
interstellar Na I lines present in our AAT-UCLES spectra
(following Crowther et al. 1995a). We 衝d that the Na I D1
and D2 lines have two broad components at LSR velocities
centered at [10 and [30 km s~1, with the latter extending
to [45 ^ 5 km s~1. From the standard Galactic rotation
curve of Brand & Blitz (1993), these velocity measurements
can be interpreted as the low-velocity component arising
from local gas while the high-velocity component is due to
more distant material, with [45 km s~1 corresponding to a
distance of D3.5 ^ 0.5 kpc (distance modulus 12.7 ^ 0.3),
this interpretation locates HD 152386 beyond Sco OB1 (1.9
kpc ; Humphreys 1978). The absolute magnitude corresponding (M \ [7.1 ^ 0.3) to this assumed distance is
V the absolute magnitudes found for other late
consistent with
Of supergiants in this analysis (Table 2).
3.
SPECTROSCOPIC ANALYSIS
Our spectroscopic analysis of HD 152386, HD 152408,
HD 313846, and a selection of O Iaf and WNL stars uses
Hillier (1987, 1990) theoretical model atmosphere as a diagnostic tool. His model atmosphere employs an iterative
technique to solve the transfer equation in the comoving
frame subject to statistical and radiative equilibrium in an
expanding, spherically symmetric, homogeneous, and
steady state atmosphere. We shall perform two sets of
analyses, one based on the usual optical diagnostics
(?? model V 舷), the other based on near-IR lines alone
(?? model K 舷). Our overall analysis technique is that of
Paper I.
3.1. T echnique
For the near-IR萴odel K萻tellar parameters are determined from 衪s of theoretical pro衛es to near-IR observations of H I Brc, He I (1.0830 km) and He II (1.0124 km) ;
the K-band magnitude determines the absolute 製x. (For
BE 381, its V -band continuum magnitude was used as
K-band photometry is not available for this star.) Where
possible, we take the terminal wind velocity from the UV
resonance C IV doublet (Prinja et al. 1990) ; otherwise, we
adopt a near-IR measurement from our data.
Because near-IR Brackett and Paschen H I series are
contaminated by He I萯n contrast with optical Balmer
transitions葁e include a detailed model atom of He I (49
levels in total), with l states merged into a single z state for
l � 3 and n � 7 (following Najarro et al. 1994) and grouped
by singlets and triplets for 8 � n � 10. Although carbon (C
III菼V) and nitrogen (N II萔) are included in these model
atmospheres to allow for their e?ect on wind cooling
(Hillier 1988), metal abundance determinations are not in
the scope of the present work. We have previously esti-
381
mated nitrogen and carbon abundances for HDE 313846
(Crowther et al. 1995a) ; from spectral morphology similarities we shall assume identical nitrogen enrichments for
the O Iafpe and WN9 stars (N/He \ 0.005 and C/N \ 0.1
by number) ; the exact choice is not critical for our analysis.
With the exception of NS 4 and BE 381, line pro衛es of
the program stars could not be simultaneously reproduced
using a standard b \ 1 velocity law (see, e.g., Crowther et al.
1995a). As detailed in Paper I, improved 衪s resulted from
a combined b \ 1 and b \ 2 law, weighted toward the
former.
To compare the results of our near-IR analyses with
optical studies, we include previous model V results for HD
152408 and HDE 313846 from Paper I and BE 381 from
Crowther et al. (1995a). We have repeated this analysis for
HD 152386 and NS 4, based on their optical He I j5876, He
II j4686 and Ha diagnostics, together with the absolute
V -band 製x. We use He I 1.0830 km for HD 152386 since its
optical He I j5876 pro衛e is extremely weak and shallow.
Earlier optical studies of NS 4 by Schmutz et al. (1989) and
Hamann, Koesterke, & Wessolowski (1995a) using pure
helium and hydrogen-helium analyses, respectively, are in
reasonable agreement with our model V results (Table 4).
3.2. Near-IR Analysis
Best 衪s of synthetic line pro衛es to the CTIO-IRS spectroscopy are plotted in Figure 7. The relatively poor 衪 of
the He II 1.0124 km and He I 1.0830 km pro衛es are a result
of the compromise velocity law. As we discuss below, the
failure of the models to 衪 He I 2.0581 km has signi衏ant
consequences on the reliability of stellar parameters derived
from studies that use this line alone. Stellar parameters for
our optical and IR analysis are in Table 5.
Best 衪s to the IR diagnostics based on He II 1.012 km
(model K) are some 200�00 K lower than the optical
results from He II j4686 (model V), a di?erence that does
not signi衏antly a?ect the stellar parameters. Currently, we
cannot determine which transition is the more accurate
diagnostic. Model K hydrogen atmospheric abundances are
generally in good accord with optical results, similar to
those found for WNE stars by Crowther & Smith (1996) ;
the analysis of HD 152386 produces slightly lower helium
abundances from model K (X \ 45% by mass) than from
He
model V (X \ 58%).
He
The one exception to the general good 衪 of theoretical
pro衛es with observations is He I 2.0581 km. For example,
in HD 152408, a strong P Cygni emission feature is predicted, yet a weak absorption feature is observed. The interpretation of this discrepancy lies in the physics of the formation
of this line and in the extreme ultraviolet (EUV) energy
TABLE 4
COMPARISON OF STELLAR PARAMETERS OBTAINED HERE FOR NS 4 (MODEL V) WITH EARLIER
OPTICAL DETERMINATIONS
T
*
(kK)
31.0 . . . . . .
30.4 . . . . . .
30.2 . . . . . .
R
2@3
(R )
*
1.12
1.20
1.13
log L
*
(L )
_
5.7
5.8
5.6
log M0
(M yr~1)
_
[4.1
[4.1
[4.3
v
=
(km s~1)
H/He
M
V
(mag)
Reference
900
700
700
�
0.85
1.00
[6.7
[6.7
[6.3
1
2
3
REFERENCES.�(1) Schmutz, Hamann, & Wessolowski 1989 ; (2) Hamann et al. 1995a ;
(3) model V.
382
BOHANNAN & CROWTHER
FIG. 7.菳est 衪 synthetic 1.0�5 km spectra (dotted lines ; model K) of our program stars with CTIO-IRS observations (solid lines). We include
predictions from model V for He II 1.0124 km and He I 2.0581 km (dashed lines).
distribution of early-type stars (Najarro et al. 1994 ;
Crowther et al. 1995a). A slight decrease in the stellar
temperature萶r equivalently, greater EUV blanketing�
leads to an increase in the emission strength of this line
because the alternative, preferred decay route from the
upper 2p 1Po level at 584 A� becomes optically thick. In
general, we 衝d that the temperature obtained from our
optical model V analysis reproduces the observed 2.0581
km emission-line strength ; e.g., for HD 152408 and BE 381
the lower excitation resulting from the IR analysis (model
K) produces signi衏antly more emission than is observed.
The e?ect of EUV line blocking on He I 2.0581 km is well
illustrated by NS 4 and BE 381, two stars with otherwise
similar spectroscopic characters (both WN9h) and very
similar measured stellar parameters. NS 4, in the Galaxy,
shows a moderately strong P Cygni emission pro衛e for this
line, while BE 381, in the LMC, shows a very weak P Cygni
pro衛e (Fig. 1). This di?erence appears to be a re裡ction of
their di?erent heavy-element abundances (Galactic vs.
LMC, respectively) upon their EUV 製x distributions.
Initial calculations (Crowther, Bohannan, & Pasquali 1998)
demonstrate that line-blanketed model atmospheres lead to
a signi衏antly modi衑d EUV energy distribution for WNL
types and to major improvements in 衪s to He I 2.0581 km.
Indeed, without line blanketing, a poor 衪 is obtained for
NS 4, unless a temperature D1,000 K lower than our model
K analysis were to be used (Crowther et al. 1995a found
similar results for the WN8 star HD 86161).
In Paper I we found that spectroscopic mass-loss rates
for HD 151804 and HD 152408 were in good agreement
O8 : Iafpe
WN9ha
WN9h
WN9h
152408 . . . . . .
313846 . . . . . .
NS 4 . . . . . . . .
BE 381 . . . . . .
K
V
K
V
K
V
K
V
K
V
Model
28.4
28.6
27.2
28.5
28.2
28.6
29.5
30.2
29.0
30.6
T
*
(kK)
R
*
(R )
_
33.7
33.2
35.4
32.4
34.3
33.1
22.8
22.6
22.8
20.8
26.8
27.2
26.2
27.6
27.3
27.7
27.8
28.4
26.8
27.5
T
eff
(kK)
R
2@3
(R )
*
1.12
1.11
1.07
1.07
1.06
1.07
1.13
1.13
1.17
1.25
log L
*
(L )
_
5.82
5.83
5.79
5.80
5.83
5.82
5.55
5.58
5.52
5.54
log M0
(M yr~1)
_
[4.58
[4.55
[4.60
[4.62
[4.51
[4.53
[4.33
[4.35
[4.68
[4.65
1650
1650
915
955
1250
1170
700
700
375
375
v
=
(km s~1)
3.5
3.4
1.8
1.8
2.8
2.6
4.5
4.0
1.2
1.2
M0 v
=
(L /c)
4.8
2.7
1.6
1.5
1.8
1.2
0.8
1.0
2.3
2.3
H/He
49.29
49.35
49.25
49.32
49.33
49.37
49.03
49.10
48.98
49.09
log Q
0
(s~1)
(J [ K)
0
(mag)
]0.01
[0.03
]0.03
]0.01
]0.01
]0.00
]0.16
]0.17
]0.19
]0.21
(B [ V )
0
(mag)
[0.31
[0.31
[0.32
[0.33
[0.33
[0.33
[0.31
[0.32
[0.31
[0.30
[7.1
[7.1
[7.0
[7.0
[7.0
[7.0
[6.3
[6.3
[6.2
[6.2
M
V
(mag)
M0 v /(L /c) is the wind performance number, and Q is the number of Lyman continuum ionizing photons. The stellar radius (R )萪e衝ed as the inner boundary of the model atmosphere萯s
*
located= at a* Rosseland optical depth of 20. The stellar0 temperature (T ) is related to R by the Stefan-Boltzmann relation. Similarly,
the e?ective temperature (T ) is determined at the radius
*
*
eff
(R ) at which the Rosseland optical depth equals 2/3.
2@3
O6 : Iafpe
Spectral Type
152386 . . . . . .
HD(E)
TABLE 5
DERIVED STELLAR PARAMETERS FOR OF, OFPE AND WNL STARS FROM NEAR-IR ANALYSES (MODEL K) WITH RESULTS FROM OPTICAL STUDIES (MODEL V)
384
BOHANNAN & CROWTHER
with rates obtained from radio observations. Here,
however, we 衝d that mass-loss rates for the O Iafpe萕N9
stars from the optical and near-IR may be systematically
greater by B0.1�2 dex from those derived from the radio.
Inaccurate values of k, Z, and c used in Paper I led to
erroneous 6 cm radio mass-loss rates. Corrected radio rates
are log M0 /M yr~1 of [5.01 for HD 151804 and [4.74 for
_
HD 152408 ; in comparison, we determine here log M0 /M
_
yr~1 [4.9 and [4.6, respectively. An additional observation of a optical-IR versus radio discrepancy is that of
Leitherer, Chapman, & Koribalski (1995), who were not
able to detect HD 152386 at either 8.64 GHz (3 cm) or 4.80
GHz (6 cm) at an upper limit of 0.3 mJy. Their upper limit is
equivalent to a radio mass-loss rate of log M0 /M yr~1 [
_
[4.70 using our model V and ([[4.78) from model K,
rates that are B0.2 dex below our derived mass-loss rate
from optical and IR observations (Table 5). (Leitherer et al.
calculated log M0 /M yr~1 [ [4.47 based on di?erent
_
estimates for the distance and atmospheric mean molecular
weight, etc.)
4.
SPECTRAL CLASSIFICATION OF HD 152386 AND
HD 152408
HD 152386, HD 152408 and HDE 313846 were classi衑d
as peculiar emission stars with an ?? O Iafpe 舷 designation ;
they did not 衪 into either the Of or WN sequences as
de衝ed at the time of their earliest morphological consideration. Our spectroscopic analysis has demonstrated the
strong similarity of the stellar atmospheres of these three
stars萢n argument that similar spectral classi衏ation
should be assigned萢nd that the three have atmospheric
parameters signi衏antly more similar to low-excitation
WNL stars than to stars with Of types. Spectral classi衏ation, however, like all true morphological classi衏ation,
must remain independent of theory or interpretation based
upon external evidence. In this section we review the history
of the spectral classi衏ation of these three stars and present
quantitative spectroscopic criteria to distinguish the Of and
WNL classes. Our working position is that these three stars
should all be classi衑d either as Of or WN, not as members
of a peculiar, intermediate class. Our criteria re裡ct this
assumption by seeking quantitative measures of line
strengths and line shifts that clearly distinguish the Of and
low-excitation WN classes.
HD 152386 was initially classi衑d as O6 : Iafpe (Walborn
1973 ; Leep 1978), HD 152408 as O8 : Iafpe (Walborn 1972),
and HDE 313846 as O7 : Iafpe (Hutchings 1979). All three
have emission-line strengths that are much stronger than
those in normal Of supergiants (e.g., Fig. 8 from Paper I),
hence the suffix indicating ?? peculiar. 舷 In all three, He II
j4542 is in absorption but blueshifted from the systemic
velocity de衝ed by the emission features, and He I j4472
has a distinctly P Cygni pro衛e with the absorption component clearly not photospheric in origin ; the spectral
subtype is therefore uncertain. HDE 313846 has the strongest emission features and is somewhat intermediate in ionization between the other two ; HD 152386 has the highest
ionization. While labeled ?? the most extreme Of star 舷 by
Hutchings, HDE 313846 was classi衑d as a WN9-type W-R
by van der Hucht et al. (1981) because emission from the
extended atmosphere dominates over photospheric absorption. While one could consider that a physical argument
and not strictly relevant in morphological classi衏ation, one
could also point to the signi衏antly greater line widths of
Vol. 511
Ha, He I j5876, and He II j4686 in these three stars, line
widths that are much more consistent with a WN classi衏ation than with O Iaf stars like HD 151804 (Paper I, and
references therein).
Walborn (1982) preferred to retain the type ?? O Iafpe 舷 for
all three stars, noting, however, that ?? if HDE 313846 is
classi衑d as W N9, then HD 152408 and 152386 must also be ;
these three spectra are too similar to be in qualitatively
di?erent categories. 舷 At that time the WN9 classi衏ation
was poorly de衝ed and the smooth sequence of characteristics to lower ionization classes from WN8 required by a
meaningful classi衏ation scheme had not yet been demonstrated. In the decade following Walborn蟬 important classi衏ation work on early-type stars, additional stars that
extended the W-R sequence to lower ionization character
than WN8 came to be identi衑d. HDE 313846 has since
been classi衑d as WN9ha by Crowther & Smith (1997) following the nomenclature in the latest WN classi衏ation
scheme of Smith et al. (1996) for WNL types. WN9 is
de衝ed by the absence of N IV j4058 emission (at a
resolution of D2�A� ), and the presence of N III j4634�41
is much greater than N II j3995. The ?? h 舷 suffix indicates
the presence of signi衏ant hydrogen content from the
Pickering-Balmer decrement. The ?? a 舷 suffix is added when
the upper Balmer lines show intrinsic absorption components.
Conti (1973) distinguished W-R and Of types through the
nature of their respective absorption features : ?? the only
absorption lines seen (in W-R types) are violet shifted (P
Cygni type). Although in some cases emission lines appear
that are similar to those found in some Of stars, the latter
types always have some unshifted absorption lines present. 舷
While He II j4542 is in absorption in all three stars, it
appears asymmetric with blueshifts signi衏antly larger
(D110 km s~1 for HDE 313846 and HD 152386, D80 km
s~1 for HD 152408) than that measured in our prototypical
O Iaf star, HD 151804 (20 km s~1 ; Paper I). Based on
Conti蟬 criterion alone, all three of the O Iafpe stars should
have been classi衑d as W-R.
In Figure 8 we compare the emission strength of He II
j4686 versus the ionization indicator, He IIj4686/He I
j5876, of our program stars with those of representative
WN5� and Of stars. The He II j4686 emission strengths
of HD 152386, HD 152408, and HDE 313846 are comparable with those of WN9h stars. With the exception of the
LMC WN9h star AB 18 (Brey 44a ; Crowther & Smith
1997), the three original O Iafpe stars show apparently
higher ionization than WN9 stars because of weaker He I
line strengths. This di?erence is not simply one of ionization ; rather, it arises principally from lower surface mass
製xes in the O Iafpe stars with resulting lower He I line
strengths. The surface mass 製xes of the O Iafpe stars, while
intermediate between the Of and WNL types, are signi衏antly closer to WNL stars (cf. Fig. 9 of Paper I). By
analogy, similar conclusions were reached for the Car OB1
WN6�a stars by Crowther et al. (1995c). As illustrated in
Figure 8, HD 152386 deserves a genuine Wolf-Rayet classi衏ation (WN9ha) as much as does HD 93162 (WN6ha) ;
both stars are at the division between Of and WNL types
because of their relatively weak winds.
At the dividing line, the distinction between extreme Of
and low-excitation WN types is largely one of degree : how
large should the blueshift of an absorption feature be or
how broad should emission lines be before a star is classi-
No. 1, 1999
NIR SPECTROSCOPY OF Of AND WNL STARS
5.
FIG. 8.菴omparison of emission equivalent widths (in A� ) of He II
j4686 versus the ratio He II j4686/He I j5876 for WN5� and Of stars
demonstrating that WNLha stars ( 衛led symbols) show emission-line
strengths between that for O Iaf and WNL types but more consistent with
low-excitation WNLs. Data are from Crowther & Smith (1997) for LMC
WNL stars, or Crowther et al. (1995c) and Hamann, Koesterke, & Wessolowski (1995b) for Galactic stars.
衑d as W-R instead of O Iaf ? Expanding upon our arguments in Paper I (p. 543), we propose that, if any of the
following quantitative criteria are met, a star should be
classi衑d as WNL, rather than as O Iaf. (1) The photospheric temperature classi衏ation line He II j4542 has a P
Cygni pro衛e, or, if in absorption only, has a central wavelength obviously blueshifted from the systemic velocity by
Z50 km s~1 (i.e., 2.5 times that seen in HD 151804). (2) W
(He I j5876) Z 3.0 A� , with He II j4686 in emissionj
(Crowther & Smith 1997). (3) W (He II j4686) Z 12 A�
j
(Crowther & Dessart 1998). The application
of the 衦st two
criteria would result in all three stars being classi衑d as
W-R. The third criterion is used to distinguish WN5�types from early O supergiants ; He II j4686 becomes the
sole discriminator in this temperature range.
Our conclusion is that, from both spectroscopic characteristics and atmospheric analysis, all three stars should be
classi衑d as WN9ha.3 We note that the three stars still
represent a distinct group as the WN9ha subtype contains
only these three. All three are in the Galaxy, while with the
exception of NS 4 others of the WN9 class are located in the
Magellanic Clouds. A question remains : How homogeneous is the WN9萣oth h and ha萾ype ? At 衦st glance, it
is no more so than other WN types ; however, we leave this
question to another paper.
3 Following the nomenclature of the W-R catalog by van der Hucht et
al. (1981) we suggest WR 79a and WR 79b for HD 152408 and HD 152386,
respectively.
385
DISCUSSION AND CONCLUSIONS
The essentially identical results from our near-IR atmospheric study of low-excitation, hot emission-line stars to
previous optical and ultraviolet analyses demonstrate that
accurate stellar parameters can be determined from near-IR
spectroscopy alone, provided that appropriate transitions
and observed line pro衛es of sufficient spectral resolution
and signal-to-noise ratio are used. Understanding of the
properties of these stars is important because their high
luminosity and strong emission signature lead them to be
among the 衦st to be studied in newly explored regions.
Because the calibration of stellar parameters with morphological classi衏ation is neither well de衝ed nor unique for
strong emission-line stars, atmospheric analysis is required
to determine their stellar properties with sufficient accuracy
for broad studies of massive-star evolution and to de衝e the
quantitative character of their parent galaxies. Our detailed
atmospheric analysis of the three O Iafpe stars indicates
that their physical properties are very similar to each other,
quite distinct from normal Of stars and much like that of
low-excitation WNL types. The most appropriate classi衏ation of these three is WN9ha.
5.1. T he E?ectiveness of Spectral Analysis Based on the IR
The new capability provided in the near future by the
combination of near-IR moderate-resolution spectroscopy
with low-order adaptive optics will allow astrophysical
studies in previously inaccessible regions such as the Galactic center, where optical observations are not possible
because of extreme interstellar obscuration, and in external
galaxies, where luminous stars are often in such crowded
衑lds that spectroscopy of what appears to be a single star
may indeed be of multiple stars. The spectroscopic diagnostics evaluated here demonstrate that near-IR quantitative
spectral analysis is of comparable precision to optical and
ultraviolet studies if one employs appropriate transitions
for analysis and if one uses sufficiently high-resolution
(R [ 2000) and high-S/N (S/N [ 30) spectra. The I- and
J-band spectral lines used here, which are usually not available in extremely obscured environments, also allow us to
designate suitable diagnostics at longer wavelengths.
A single-temperature model could not be 衪 to both the
optical and near-IR spectroscopic features ; temperatures
systematically 200�00 K lower resulted from IR 衪ting
alone, an insigni衏ant di?erence in the determination of
stellar parameters. It is not clear which temperature is the
more accurate. Schmutz (1991) 衦st noted the discrepancy
for WN types between observed He II optical (j4686, 4�
and near-IR (1.012 km, 5� line strengths with theory. The
difficulty in interpreting this discrepancy lies in understanding how the theoretical population of He II n \ 5 could be
in error while other levels are correct. Helium atomic data
cannot be invoked as excellent agreement is seen in stronglined WNE stars (see Crowther & Smith 1996). The di?erences in He II line strengths for the extreme O and WNL
stars may be reconciled by realistic consideration of the
underlying photosphere, particularly the velocity structure
at depth.
Di?erences between the optical and IR results provide
important clues to the nature of the extreme UV 製x. For
example, the optical spectral morphologies of the two
WN9h stars NS 4 and BE 381 are nearly identical (Fig. 4),
386
BOHANNAN & CROWTHER
yet their He I 2.0581 km (Fig. 1) character is quite di?erent
owing to the sensitivity of this line to the EUV 製x (� 3.2).
He I 2.0581 km should not be used as a temperature diagnostic. Instead, He I 1.0830 km or He I 1.7002 km should be
modeled. He I pro衛es must not be used in isolation ; rather,
they should be combined with an appropriate He II transition (e.g., 1.012 km, 3.09 km) so that the stellar temperature is determined from the ionization of the
atmosphere.
In environments of increasingly large interstellar absorption the I- and J-bands become difficult to observe ; the
temperature, abundance, and wind velocity diagnostics we
evaluate here would not be readily available. When the Iand J-bands are too faint, He I 1.7002 km and He I 2.0581
km (if a strong P Cygni pro衛e is present) are available for
terminal velocity measurement and Bra, Pa, and Pb can be
used for H abundance determinations. Synthetic spectra of
these transitions and He II temperature diagnostics are presented in Figure 9 for stellar parameters equivalent to HD
152408 (model K). Pa and He II 1.864 km (6� (while
resolved in HD 152408, they will be signi衏antly blended
for broader line stars) are in a spectral region poorly accessible from ground-based telescopes. He II 3.09 km (7�,
which is more accessible from the ground despite its location in the thermal IR, does not su?er signi衏ant blending
and provides crucial diagnostic power for stars of later
spectral type than WN8 within very heavily reddened
environments.
The principal gain of the higher spectral resolution observations used here is the separation of blended spectral features. For example, the 2.11 km emission feature observed
in Of and WNL stars has been assumed to be formed principally by He I 4s� with N III 8�providing a minor
contribution (Hillier 1985), a prediction con衦med in our
high-S/N UKIRT observations of HDE 313846 (Fig. 10). A
double-peaked pro衛e is seen at 2.11 km, with components
at 2.113 km from He I (4s�, 2.113 km) and at 2.115 km
produced by N III (8j� 2.1149 km and 8i� 2.1146 km).
The presence of N III is con衦med by the presence of weak
emission at 2.1031 km, which can be attributed to N III
8h� at 2.10315 km.
Carbon and nitrogen lines in the near-IR provide critical
metal diagnostics for objects without optical or ultraviolet
data. In HDE 313846 weak C IV 3p 2Po� 2D emission
lines are seen at 2.069 km (J \ 1/2�2) and 2.078 km
(J \ 3/2�2) (Fig. 10) ; these features are not present in NS
Vol. 511
FIG. 10.菻igh-resolution UKIRT K@-band spectroscopy of NS 4 and
HDE 313846. Weak emission features are revealed given sufficient spectral
resolution. In particular, 2.11 km shows a double peaked pro衛e for HDE
313846 due to He I and N III components.
4, an observation consistent with the behavior of the optical
C IV counterparts at jj5801�12 in these two stars.
5.2. Quantitative Comparison of W N9ha Stars with Of and
W NL Stars
Our perspective on the spectroscopic characterization of
the three O Iafpe stars is that they should be classi衑d in the
same subgroup and that they should be categorized with
objects having similar properties. As we demonstrate in this
paper, the spectra and physical properties of HD 152386,
HD 152408, and HDE 313846 indicate that these stars are
essentially equivalent and that their stellar properties are
intermediate between those of normal Of stars and WNLtype W-R stars but that their properties are quite distinct
from normal O Iaf-type stars and are more similar to weaklined WNL stars. We conclude that all three should be
considered as WN9ha. In this section we will brie褃
compare the physical properties of these stars to other luminous hot stars and make a few comments on their evolutionary signi衏ance.
FIG. 9.萐ynthetic near-IR spectra for HD 152408 from model K. He II 3.09 km is an important temperature diagnostic for WNL stars in heavily
reddened environments. The Pb synthetic pro衛e is in excellent agreement with ESO-IRSPEC observations of HD 152408 by F. Najarro (1998, private
communication).
No. 1, 1999
NIR SPECTROSCOPY OF Of AND WNL STARS
In Paper I we identi衑d the surface mass 製x (M0 /4nR2 )
*
as the principal diagnostic distinguishing O and WNL
stars. The surface mass 製x of HD 152386 [log(M0 /4nR2 ) \
*
[8.69 M yr~1 R~2) is identical to that found for HDE
_
_
313846, consistent with their very similar spectroscopic
appearance. The surface mass 製x of HD 152408 is D0.1
dex lower than for the other two, a di?erence seen spectroscopically in He II j5412 and He II 2.1885 km, which show
the smaller stellar wind e?ects in this star. The slightly
higher terminal velocity for HD 152386 compared with HD
152408 and HDE 313846 can be interpreted as a higher
present mass. The somewhat lower helium content for HD
152386 relative to HD 152408 and HDE 313846 may re裡ct
a slightly earlier evolutionary state.
Figure 11a illustrates that the mass-loss rate of HD
152386 is comparable with those of other WNL stars with
hydrogen detected and absorption lines present (WNL-ha).
The wind performance numbers [M0 v /(L /c)] and surface
= *
hydrogen content of HD 152386 are more representative of
other WNL stars than of normal Of stars (Fig. 11b).
The classi衏ation of HD 152386, HD 152408 and HDE
313846 in the sequence of WNL types instead of unique,
peculiar O-type emission-line supergiants serves to 衪 them
smoothly into scenarios for massive stellar evolution. From
the continuity of spectroscopic, physical, and chemical
properties, we proposed in Paper I that the WN9ha stars
directly evolve from Of stars, without an intermediate luminous blue variable (LBV) phase. We extend this argument
as follows. (1) In general, LBVs have lower surface helium
enrichments than the WN9ha stars, a result that implies
that processes other than LBV eruptions are responsible for
the surface abundance enhancements of the WN9ha stars.
(2) The nebulae associated with HD 152408 and HD 152386
(RCW 113�6 and RCW 111, respectively) are not of the
?? ejecta 舷 type often associated with LBVs and other WNL
stars (Lozinskaya 1982). (3) WNL stars that have probably
passed through an LBV stage萯.e., those seen with LBVtype nebulae萫xhibit sizeable photometric variabilities,
have WN8 spectral types, and are generally not associated
with clusters or associations.
If the subsequent evolution of the WN9ha stars follows
that proposed for other W-R stars (see, e.g., Conti 1976),
they should advance to higher temperature, higher wind
density, and lower surface hydrogen contents. WN7 stars�
e.g., HD 151932 (Crowther et al. 1995c), located in Sco OB1
with HD 151804 and HD 152408萴ost readily 衪 the morphological, physical, and environmental properties for the
next stage. (A close evolutionary connection between HD
152408 and HD 151932 was 衦st hinted at by Walborn 1975
from their strikingly similar He I j5876 pro衛es.) However,
we cannot exclude the possibility that WN9ha stars subsequently undergo LBV eruptions at a later stage. Recent
theoretical models by Pasquali et al. (1997) incorporating
pulsational instabilities lead to an LBV outburst at a phase
at which the surface content is highly chemically evolved ;
evidence for such highly enriched LBVs has yet to be provided observationally (Crowther 1997).
We wish to thank John Hillier for providing his atmosphere code. We appreciate the support of the sta? at the
Cerro Tololo, Mount Stromlo, and Anglo-Australian
Observatories, especially Richard Elston (CTIO), Sean
Ryan (AAT), and Mike Bessell (MSO). We are grateful to
Linda Smith and Dave King for obtaining K-band obser-
387
FIG. 11.�(a) Mass-loss rates vs. luminosity for various O-type giants
and supergiants (circles), WNL-types with absorption lines (triangles), and
WNL stars without absorption lines (squares), with results from the optical
analysis (model V) of the program Ofpe and WN9ha stars shown as 衛led
symbols. Data not determined here are from Puls, Kudritzki, & Herrero
(1996), Crowther et al. (1995b), Hamann et al. (1995a). The Howarth &
Prinja (1989) mass-loss萳uminosity relation for O-type stars is shown with
a dotted line. (b) Wind performance numbers vs. surface hydrogen mass
fraction, with the single-scattering limit indicated (dotted line). (c) Stellar
temperature (log K) vs. surface mass 製x (M yr~1R~2). The discrimi_ I is shown
_
nation between Of and WNL stars de衝ed in Paper
(dotted line).
vations of HDE 313846 and NS 4 at UKIRT and to Orsola
De Marco for coobserving NS 4 at the INT. We appreciate
the careful reading of the manuscript and constructive comments by the referee, N. R. Walborn. Calculations have
been performed at the Cray J90 of the RAL Atlas center and
at the UCL node of the U.K. STARLINK facility. P. A. C.
gratefully acknowledges 衝ancial support from PPARC.
This research has made use of the SIMBAD database, operated at the CDS, Strasbourg, France. IRAF was written and
supported by the National Optical Astronomy Observatories.
388
BOHANNAN & CROWTHER
REFERENCES
Brand, J., & Blitz, L. 1993, A&A, 275, 67
Howarth, I. D., & Prinja, R. K. 1989, ApJS, 69, 527
Conti, P. S. 1973, in IAU Symp. 49, Wolf-Rayet and High-Temperature
Howarth, I. D., & Schmutz, W. 1992, A&A, 261, 503
Stars, ed. M. K. V. Bappu & J. Sahade (Dordrecht : Reidel), 96
Humphreys, R. M. 1978, ApJS, 38, 309
热�. 1976, in Me峬oires Socie峵e� Royale des Sciences de Lie?ge, 6e se峳ie,
Hutchings, J. B. 1979, PASP, 91, 361
tome IX, 193
Krabbe, A., Genzel, R., Drapatz, S., & Rotaciue, V. 1991, ApJ, 382, L19
Conti, P. S., Hanson, M. M., Morris, P. W., Willis, A. J., & Fossey, S. J.
Leep, E. M. 1978, ApJ, 225, 165
1995, ApJ, 445, L35
Leitherer, C., Chapman, J. M., & Koribalski, B. 1995, ApJ, 450, 289
Crowther, P. A. 1997, in ASP Conf. Proc. 120, Luminous Blue Variables :
Leitherer, C., & Wolf, B. 1984, A&A, 132, 151
Massive Stars in Transition, ed. A. Nota & H. Lamers (San Francisco :
Lozinskaya, T. A. 1982, Ap&SS, 87, 313
ASP), 51
Lundstro巑, I., & Stenholm, B. 1984, A&AS, 58, 163
Crowther, P. A., & Bohannan, B. 1997, A&A, 317, 532 (Paper I)
Massey, P. 1984, ApJ, 281, 789
Crowther, P. A., Bohannan, B., & Pasquali, A. 1998, in ASP Conf. Proc.
Massey, P., & Conti, P. S. 1983, PASP, 95, 440
131, Boulder-Munich II : Properties of Hot, Luminous Stars, ed. I. D.
McGregor, P. J., Hyland, A. R., & McGinn, M. T. 1989, A&A, 223, 237
Howarth (San Francisco : ASP), 38
Meyerdierks, H. 1993, Starlink User Note 86.9 (Chilton, UK : Rutherford
Crowther, P. A., & Dessart, L. 1998, MNRAS, 296, 622
Appleton Laboratory)
Crowther, P. A., Hillier, D. J., & Smith, L. J. 1995a, A&A, 293, 172
Najarro, F., et al. 1994, A&A, 285, 573
热�. 1995b, A&A, 293, 403
Pasquali, A., et al. 1997, ApJ, 470, 340
热�. 1995c, A&A, 293, 427
Prinja, R. K. 1994, A&A, 289, 221
Crowther, P. A., & Smith, L. J. 1996, A&A, 305, 541
Prinja, R. K., Barlow, M. J., & Howarth, I. D. 1990, ApJ, 361, 607
热�. 1997, A&A, 320, 500
Puls, J., Kudritzki, R. P., & Herrero, A. 1996, A&A, 305, 171
Daly, P. N., & Beard, S. M. 1992, Starlink User Note 27 (Chilton, UK :
Schmutz, W. 1991, in Proc. IAU Symp. 143, Wolf-Rayet Stars and InterRutherford Appleton Laboratory)
relations with Other Massive Stars in Galaxies, ed. K. A. van der Hucht
de Koter, A., Heap, S. R., & Hubeny, I. 1997, ApJ, 477, 792
& B. Hidayat (Dordrecht : Kluwer), 39
Eenens, P. R. J., & Williams, P. M. 1992, MNRAS, 255, 227
Schmutz, W., Hamann, W.-R., & Wessolowski, U. 1989, A&A, 210, 236
热�. 1994, MNRAS, 269, 1082
Smith, L. F. 1968, MNRAS, 140, 409
Hamann, W.-R., Koesterke, L., & Wessolowski, U. 1995a, A&A, 299, 151
Smith, L. F., Shara, M. M., & Mo?at, A. F. J. 1996, MNRAS, 281, 163
热�. 1995b, A&AS, 113, 459
Smith, L. J., Crowther, P. A., & Willis, A. J. 1995, A&A, 302, 830
Hanson, M. M., & Conti, P. S. 1994, ApJ, 423, L139
The�, P. S., Wesselius, P. R., & Janssen, I. M. H. H. 1986, A&AS, 66, 63
Hanson, M. M., Conti, P. S., & Rieke, M. J. 1996, ApJS, 107, 281
van der Hucht, K. A., Conti, P. S., Lundstro巑, I., & Stenholm, B. 1981,
Hillier, D. J. 1985, AJ, 90, 1514
Space Sci. Rev., 28, 227
热�. 1987, ApJS, 63, 947
Walborn, N. R. 1972, AJ, 77, 312
热�. 1988, ApJ, 327, 822
热�. 1973, AJ, 78, 1067
热�. 1990, A&A, 231, 111
热�. 1975, PASP, 87, 735
Howarth, I. D. 1983, MNRAS, 203, 301
热�. 1982, ApJ, 256, 452
Howarth, I. D., Murray, J., Mills, D., & Berry, D. S. 1995, Starlink User
Williams, P. M., van der Hucht, K. A., & The�, P. S. 1987, A&A, 182, 91
Note 50.16 (Chilton, UK : Rutherford Appleton Laboratory)
Документ
Категория
Без категории
Просмотров
0
Размер файла
318 Кб
Теги
306647
1/--страниц
Пожаловаться на содержимое документа