A&A 459, 215-227 (2006)
DOI: 10.1051/0004-6361:20065055
M. A. Smith
Science Programs, Computer Sciences Corporation, Space Telescope Science Institute, 3700 San Martin Dr., Baltimore, MD 21218, USA
Received 20 February 2006 / Accepted 1 August 2006
Abstract
Using the International Ultraviolet Explorer data archive,
we have examined the SWP echellograms of 74 B0-B2.5e stars for
statistically significant fluctuations in the He II ("H'')
1640 line profile. In this sample we found
that the He II line is occasionally variable in 10 stars over short
to long timescales. The He II-variable stars discovered are
Eri,
Ori,
Cen, 6 Cep, HD 67536,
Ori,
Cen,
Aqr, 2 Vul, and
19 Mon. The most frequent two types of variability are an extended
blue wing absorption and a weakening of the line along the profile.
Other types of variability are a weak emission in the red wing
and occasionally a narrow emission feature. In the overwhelming number of
cases, the C IV resonance doublet exhibits a similar response; rarely, it
can exhibit a variation in the opposite sense.
Similar responses are also often seen in the Si IV doublet,
and occasionally even the Si III
1206 line. We interpret the
weakenings of He II and of high-velocity absorptions of C IV to
localized decreases in the photospheric temperature, although this may not
be a unique interpretation. We discuss the variable blue wing absorptions
and red wing emissions in terms of changes in the velocity law and mass
flux carried by the wind. In the latter case, recent experimental models by
Venero, Cidale, & Ringuelet require that during such events the wind
must be heated by 35 000 K at some distance from the star.
Key words: stars: general - stars: emission-line, Be - stars: winds, outflows - ultraviolet: stars
Together with the hydrogen lines, the lines of helium are among the
most important in the spectra of hot stars. In atmospheres of B stars and
most O stars, the dominant ion stage of helium is He+, and the strongest
of the He II features is the 1640 ("H
'') line.
In O stars this line is formed substantially in the wind - so much
so in supergiants that the line generally develops a strong P Cygni
emission structure. For spectral types later than O8-09, the line
decreases in strength, but it remains visible as a photospheric diagnostic
for spectra as late as B2.5 (Peters 1990; Rountree & Sonneborn 1991).
The He II 1640 line is actually a complex of seven permitted
transitions arising from lower levels at 40.8 eV. Its effective centroid
wavelength is 1040.42 Å. In the outer atmospheres of hot stars
this line is formed by the
photoionization of He2+ by extreme UV (<
228) photons,
followed by recombination. The density and temperature sensitivities
insures that the line's formation occurs substantially in
the base of the wind or within the photosphere
for O and B stars near the main sequence, respectively.
The 1640 line is mildly sensitive to departures from LTE in the
He1+ atom. As a result, the strengths computed from non-LTE models tend
to be slightly stronger than those from LTE models.
Since these effects are relatively small,
there appears to be no major difficulties fitting this line approximately
with conventional blanketed non-LTE model atmospheres. Auer & Mihalas (1972)
suggested that the near coincidence of central wavelengths of the He II
and some hydrogen lines could enhance emission of He II
4686
and
1640 through optical pumping (Bowen fluorescence). However,
using more recent atomic parameters, Herrero (1987) demonstrated that
these effects are negligible. Recently, Venero et al. (2000;
"VCR'') have considered the behavior of the
1640 line
for model atmospheres with
= 25 000 K and strong, isotropic,
and heated winds. These authors find that even for model
atmospheres of early-type B stars a fast and/or heated wind can alter the
underlying photospheric profile. For example, in these models
1640
undergoes a near maximum absorption strength in winds having a temperature determined by
radiative equilibrium, i.e., with
/
= 0.6-0.8. However,
if the wind is heated to 10 000 K above the
,
then emissions
will be produced in one or both of the line wings. Thus, isotropic, heated
(
)
winds produce a P Cygni-type profile, that is,
with a distinctly blueshifted absorption and slightly redshifted emission.
For standard wind models for early-type Be stars
(unheated winds, with
1
10
yr-1),
emission should be absent or undetectable.
Even in the spectrum of the O9 V star 10 Lac, with its
mass loss rate of 1.7
10
yr-1, the
1640 profile is unshifted and has no redshifted emission (see Fig. 18a of Brandt et al. 1998). It can be added that unpublished thesis work by
Cidale (1993) suggests that these same trends continue with emission in C IV. As a postulated hot temperature rise in these models is moved
outward, a mild red emission component in C IV is enhanced while
the absorption component remains almost the same. Some of our examples
of observed variations below will illustrate this behavior.
The 1640 line has been interpreted by some authors to be stable in
strength and thus to be a good measure of an O or early B star's effective
temperature. As detailed below, variability in this line has only seldom
been reported in hot, chemically homogeneous, single stars.
According to the VCR models, any variability of this line
would require substantial changes in the star's effective temperature,
mass loss rate, or restructuring of the wind stratification. In cool
stars
1640 variations arise from
variable EUV irradiation in their chromospheres (Linsky et al. 1998).
In this paper we will test several claims in the literature for
1640 variability in Be star spectra and extend the search
for this variability to a larger sample of B stars. Positive
results of this search will be placed in the general context of the
predictions of the VCR's ad hoc and nonstandard models of winds in
hot stars.
VCR's models predict that 1640 variations are caused
by dramatic changes in an O or B star's wind structure. This is
consistent with the report by Peters (1990) that a weak correlation exists
between the strength of the
1640 line and the wind component of CIV
1550i of
Eri (B2e). Peters noted that these variations
depend in part on where the star is in its "wind oscillation cycle.''
A second report of correlations between variations of He II and another
line in a B star came from a campaign with a ground-based telescope
and the International Ultraviolet Explorer (IUE) during three 8-h "shifts'' on Eri. According to Smith
et al. (1996), decreases in the
1640 and C IV resonance doublet
absorptions coincided with the creation of a "dimple'' on five occasions on
1990 October 21 and October 22 (see also Smith & Polidan 1993; Smith et al.
1996) in the line profile of
6678 and other optical He I lines
.
The timescale for these changes was as short as
h.
These line profile transients hint at the presence of rapid, possibly
magnetic, activity close to the surface of the star.
A connection with
1640 changes would
further tie this activity to the photosphere. In view of this short
history, we began our search for
1640 activity in spectra
of early-type Be stars for which claims have been made preferentially for
the presence of magnetic fields. Because the theoretical predictions are
that changes in
1640 strength should be found in the dense,
rapidly accelerating regions of winds of Be stars. It is logical to search
the spectra of stars which have known histories of variable wind components
of UV resonance lines.
The ultraviolet data for these programs are extant high-dispersion
IUE echellograms obtained through the large aperture of the Short
Wavelength Prime (SWP) camera. These data were obtained from the MAST
archive. An IDL program was written
to read all spectra obtained for a star in the orders containing the
echelle orders of
1640 and the resonance lines of C IV, N V,
Si III, and Si IV. The spectra were cross correlated against one another
to place them on a common wavelength scale an co-added. Small rescalings
were then applied to the individual spectra to force their continua to the
same level. Generally, these orders contain imprints of the
instrumental calibration "reseaux'' etched
onto the faceplates of the camera. The flux dips they cause are omitted in
our plots to avoid possible confusion.
![]() |
Figure 1:
High dispersion IUE spectra of the ![]() ![]() |
Open with DEXTER |
In order to study the He II 1640 line's behavior with
respect to physical variables, we utilized the SYNSPEC line
synthesis code (Hubeny et al. 1994)
using LTE and non-LTE atmospheres for the effective
temperature interval 21 000-29 000 K at log g = 4 and for
= 5 km s-1.
LTE models were taken from Kurucz (1993), while non-LTE models were taken
from an extension of the OSTAR2002 grid for B stars (Lanz & Hubeny 2006).
Although the
1640 line is an important transition complex in
the He1+ atom, its proximity in wavelength
to several iron-group lines of comparable strength is one reason why it has been so little studied in B stars. The SYNSPEC program ameliorates this problem by permitting the
user to identify the primary lines by eliminating candidates from the input
line library and seeing if a feature of interest has disappeared in the
recomputed spectrum. The program provides the ability to convolve the
spectrum to mimic the effects of instrumental and rotational broadening.
We utilized these functions
to determine the contribution of the He II line to the
aggregate "
1640 feature'' as a function of stellar
in spectra broadened by rotation.
To show why a spectral line synthesis approach is important to the study
of 1640, we exhibit in Fig. 1 the wavelength region
surrounding this line taken from
IUE spectra of two B1 stars,
Sco and HR 1886. The spectra
of both stars are sharp-lined. As expected, we see that the He II line is stronger in
Sco (
= 30 200 K, log g = 4.2;
Hunter et al. 2005) than in HR 1886 (
= 23 300 K, log g = 4.1; Lyubimkov et al. 2005). The figure also shows that the
absorptions of nearby lines must be included in the measurement of the
total strength of "
1640'' in broad lined spectra. Our spectral
line syntheses have enabled us to identify these blends as follows.
Started from the blue edge of the aggregate, that the strong feature at
1639.4 Å is itself a blend of a Zn III (
1639.42) and an Fe line.
The latter may be either Fe IV
1640.40 or Fe II
1640.40
for spectra of type B0 or B2, respectively. Dominating the blue half of
the He II line is a feature at 1640.0 Å. For late O to B0.2-type stars,
this feature is a blend of excited Fe IV lines at
1640.04
and
1640.15. For B1-B2 stars the composition of this blend
has shifted to Ni III
1639.99 and secondarily to Fe II
1640.15. Just to the red of the He II line, a strong blend of Fe IV
1640.78 line at type B0 gives way to a weak
Fe II
1640.86 line.
![]() |
Figure 2:
The equivalent width contributions to the He II ![]() ![]() ![]() |
Open with DEXTER |
The fractional contribution of the He II line within the 1640 feature, along with the dependence of the strength of the total aggregate,
is plotted in Fig. 2. This figure depicts the equivalent
width-temperature relations for three pairs of spectra computed with
SYNSPEC for LTE and non-LTE atmospheres.
In constructing the first ("True'') pair of
models, we have removed all lines from metal ions from the input line
library and computed the equivalent width between "continuum'' points
at 1638.7 Å and 1641.5 Å. This is almost identical to the wavelength
interval selected by Peters (1990) for her measurements. Second, and beneath
the "True'' relations in Fig. 2, we show the relation for the
same lines "spun up'' to a rotational velocity of 300 km s-1 and
measured according to the now lower "continuum'' points in the same
wavelength interval as in the first case.
A third pair of equivalent relations is measured with
the metal lines reinserted in the synthesized spectrum. The equivalent
width for a fully broadened feature of the
1640 aggregate is shown
for reference in the middle of the diagram for
= 25 000 K.
Altogether, Fig. 2 brings out several characteristics of the
1640 feature. First, all relations run roughly parallel to one
another. Therefore, whichever relation one follows will undergo the same
fractional change as one moves to a new effective temperature. For example,
it is a remarkable coincidence that the blends to the blue and red of He II
maintain their strengths relative to He II through the early B spectral types.
Second, the loss of equivalent width from an undersetting of
the continuum in a rotationally broadened spectrum is substantial, about 30%. This affirms the necessity of measuring the strengths of the
1640 aggregate in the same way for the range of stellar rotational
velocities. Third, the contribution of the iron lines is almost half the
strength of the total aggregate (for example, 0.30 Å of the total
of 0.62 Å at
= 25 000 K).
Fourth, the enhancement of the line strength due to non-LTE
effects is very small.
In order to undertake an quantitative analysis of the 1640 variations, we first "conditioned'' the data, that is we standardized
the continuum levels and slopes (which changed over the IUE lifetime
as the detector degraded) of the constituent spectra. We performed these
steps by coadding all the spectra and choosing two generally line-free
regions across the order containing the He II line and resonance doublets
of interest. For this purpose we chose two velocity regions (relative to
line center) at -1200--700 km s-1 and +350-900 km s-1).
The blue window of each spectrum was scaled relative to the mean.
The spectrum were then individually detrended relative to the mean again
by an interactive computer routine
(generally by
5-7% from one end of our echelle order to the other).
We then binned the spectra in wavelength by
a ratio of 2 to 1 pixels, thus making the value of each binned pixel
substantially independent of the values of its neighbors. Although in
a few cases we have smoothed spectra for our plotting presentations, our
statistical tests described below were performed on the unsmoothed data.
To obtain an estimate of the rms noise level of our spectral comparisons
shown in Figs. 3-21, we used the median of the
absolute value of the differences of spectra in the quasi-continuum windows.
The characteristic signal-to-noise ratios derived in this fashion were
20-27 per binned pixel for pairs of spectra,
35 for comparisons
of one spectrum against a seasonal average, and
70 for averages
of two seasons represented by large numbers of spectra.
![]() |
Figure 3:
IUE spectra of ![]() ![]() ![]() |
Open with DEXTER |
Table 1: Relevant parameters for program Be stars (ordered by right ascension).
To determine the statistical significances of our trial 1640 variations, we made the assumption that the data noise is gaussian. Veteran users of IUE data will
recognize that this is formally a risky assumption for flux excursions of
perhaps 2 rms or more. However, since most flux differences within
the profile do not exceed 1-1
rms, the errors caused
by this assumption are unlikely to be serious. We quantified the
statistical significances, "
'', of the line strength variations
with a computer a program we wrote that uses simple Monte Carlo approach.
Our procedure
was to define the wings of the entire profile and to use the rms level
determined above to determine the statistical likelihood of random
variations across the profile causing a difference in the absorption
anywhere in the profile by at least the observed amount. Notice that this
procedure estimates the significance level irrespective of "sympathetic''
responses in the C IV or Si IV lines.
As an initial check on our technique, we compared the fluctuations
of He II line profiles of the rapidly rotating B3 V star
UMa. The IUE satellite observed this star a total of 64 times with the SWP camera as a calibration standard
during the interval 1978-1994. Using our Monte Carlo program,
we searched for statistically significant line profile variations from the
mean profile. We found fluctuations neither over the whole profile
or the red or blue halves greater than 1.6-1.7
.
Moreover, in the instances of greatest fluctuations from the mean there
were no sympathetic responses in the C IV or Si IV lines.
Because the "eye'' is a good judge of sustained flux variations
of several pixels, it is not surprising that we found
the overwhelming number of candidate variations we initially selected
turned out to be significant to at least the 3
(0.13%) level.
We relaxed this criterion only in Fig. 15, in which evidence from a simultaneous strong C IV variation in the same velocity range is overwhelming.
Note also that our algorithm tests only changes in overall line strength.
Thus, it is not an effective tool to measure the significance of high
frequency variations of opposing signs across the profile.
For this reason we withdrew two examples of possible "emission
spikes'' because they were not found to be
statistically significant when tested against simulations over the whole
line profile (typically
300 km s-1). In several cases the
line strength contribution in one region of the profile overwhelmed a variation of opposite sign in another and nevertheless was
significant over the whole profile. For example, for
Fig. 20 our annotated significances
refer to the
net difference of the opposing contributions. In two
of our examples (Figs. 9 and 10) opposite contributions of two segments of the line profile are about equal. In these cases we will give the significances for the
corresponding halves of the profiles.
Overall, it is likely that our procedures have excluded
several true variations of marginal significances.
The impetus for this program was the example of 1640 variations in the B2e star
Eri and
Cen. As discussed below,
1640 and various optical He I lines in the spectra of these two stars undergo
rapid activity. This fact has led several authors to suggest
that magnetic fields play a role in this activity. Recently, Neiner
et al. (2003) have reported the detection of a rotationally modulated
magnetic signature in
Ori. Thus, we will start our
survey of He II line variability by discussing these three stars.
Various authors (e.g., ten Hulve 2004) have suggested that magnetic fields play a role in aperiodic variability of the photospheric components
of the C IV and other resonance lines. We will therefore treat these
stars as well.
To extend the search for He II line variability further, we
surveyed all B0-B2.5 Be stars that the IUE observed at high dispersion through the large aperture of the SWP camera at least 10 times. This search netted a sample of 74 stars.
To this number we also added several early-Bn stars,
such as UMa, that were observed many times, but none of them
exhibited He II line variations.
Likewise, we note that some Bp stars exhibit variations in
1640
because of their heterogeneous He surface abundances (Bp stars such as
Ori E), and these are not included in our program.
Our search is admitted not exhaustive, and it may
contain selection biases. However, note that rotational velocity was not
a search criterion, except implicitly through our choice of Be stars.
Table 1 lists 10 program stars for which we have found
variable He II lines, representing a data sample
of 558 IUE SWP-camera echellograms.
The table also gives spectral types,
and
values
according to the cited references.
We have given preference to spectral types determined at high
resolution and for velocities and temperatures to determinations by
recent authors. Rough estimates of
for HD 67536, 2 Vul, and 6 Cep are based on the observed He II line
strength, but these were not used in this work.
Eri is a rapidly rotating B2 IVe star that
has been studied extensively in the optical, UV, and X-ray wavelength
ranges. Searches for velocity variations have produced no evidence that
the star is in a binary (Bolton 1982; Smith 1989).
Since the discovery of H
emission in this star by Irvine (1975),
this line has been observed to cycle between emission and
absorption states. Bolton (1982) first reported that the star
to be a periodic velocity variable.
These variations are now generally recognized to be due
to nonradial pulsations (NRP; e.g., Rivinius et al. 2003, but cf. Balona & James 2002). The star's broad spectral lines have prevented
the direct detection of a magnetic field. However, magnetic activity
might explain several types of activity. The first example of this,
"dimples'', has been noted already. Second, Smith (2000) has noted
the occasional presence of "high velocity absorptions'' in the He I
6678 profiles.
These events last several hours and have been interpreted as ejections
of blobs, many of which return to the star (Smith et al. 1991;
Smith 2000; see also
Cen discussion). A third possible case for
magnetic activity is the observation of "flows'' across the
He I
6678 line profile over several hours (Smith 1989). This
suggests that the matter is channeled along prominence-like structures
over the star's surface. A fourth example is the observation by the
Rosat satellite of a strong soft X-ray flare, lasting several hours
(Smith et al. 1993). Groote & Schmitt
(2004) have pointed to the similarity of this event and a flare observed
in the magnetic Bp star,
Ori, E.
Evidence also exists for a periodicity or cyclicity of 475 days
for the star's "Be outbursts'' (Mennickent et al. 1998; Balona
& James 2002). Several observers (e.g., Peters
1990) have also reported increases in H
emission strength, and the
appearance of Discrete Absorption Components (DACs) of the C IV and
Si IV resonance doublets at about -900 km s-1.
We found variations over both short and long timescales in our
examination of the He II 1640 line of
Eri in
the IUE archives. As a start, we note
that the He II line equivalent widths
decreased substantially, and with almost no overlap between their
respective ranges between the epochs 1993.79-1994.06
and 1995.7. A comparison of the mean profiles during these times
is depicted in Fig. 3.
This plot shows that the He II line is either partially filled in on the
red side (0 to +300 km s-1), or Doppler shifted to to the blue.
This interpretational ambiguity is decisively settled by the filling in
of the red wing of the C IV doublet at 0-240 km s-1 in this figure.
A similar difference is found in the Si IV doublet (not shown).
![]() |
Figure 4:
Comparison of IUE observations SWP 29272 and 29301 (2.0 days later) on ![]() ![]() |
Open with DEXTER |
![]() |
Figure 5:
Comparison of IUE observations SWP 39904 and 39911 (19 h later) for He II, the C IV doublet and the Si IV ![]() ![]() |
Open with DEXTER |
The MAST/IUE archive contains 145 SWP high-dispersion observations of
Eri distributed over a large range of timescales, including
monitoring campaigns in 1982 and 1996. We have found rapid variations
of the He, II, Si IV and C IV lines during these times. Figures 4
and 5 document changes over the central
region of the photospheric profile. The intervals between
these two pairs of observations are 48 and 19 h, respectively. Figure 5 is especially interesting because it compares the
profiles during a pair of dimple-active and -inactive states (Smith et al. 1996; see Figs. 5 and 6). As opposed to the He I lines, the
He II respond to dimples by small weakenings.
In Fig. 6 we show variations over 4-h in the red
wings of the He II, C IV and Si IV
1394 lines. Because the
increased flux may not exceed the continuum level, we do not know if this
is due to emission or to a weakening of absorption.
![]() |
Figure 6:
Comparison of IUE observations SWP 32228 and 32232 (3.9 h later) for the He II, the C IV, and Si IV ![]() ![]() |
Open with DEXTER |
Ori is a typical B2e star that seems to oscillate between B-normal and Be H
emission states. So far, these oscillations seem
to be cyclical rather than strictly periodic. The high-velocity (wind)
components of the resonance lines often exhibit large changes. The profiles
are relatively narrow for a classical Be star. This fact has permitted the
discovery of a weak dipolar surface magnetic field that modulates on a rotational period of 1.29 days (Neiner et al. 2003). Given the expected
radius of a B2 main sequence star, this period implies that
we view this star from an intermediate aspect.
Neiner et al. (2003) also reported an enhancement of nitrogen from an optical line. C IV variations inform us that the wind of
this star can be variable on timescales as short as 1
h.
Such variations suggest that localized and hence anisotropic changes in the
wind in the rapidly accelerating zone occur close to the star (Sonneborn
et al. 1988).
The IUE archive includes 189 SWP high-resolution observations of
Ori. Of these, 110 are included during intensive campaigns in 1982
and 1996 and intermittent monitoring in 1983. According to ground-based
polarization studies,
the star underwent a strong, rapidly evolving outburst in 1983
(Sonneborn et al. 1983). This event was accompanied by the emergence of even
stronger DACs in the C IV lines, though these were shifted to lower velocities.
The star's H
line was in strong emission in December, 1996, and
by 1999 its equal V, R emissions were still unchanged (Peters 2005).
Thus, it is likely that when the 1996 campaign was conducted,
Ori was likewise in a Be active state.
Although the star's H
line showed strong emission
during the 1982-3 episode, there are apparently no published accounts of
its status during early 1996. However, Neiner et al. (2003) documented that
the emission was moderately strong in 1998 and declining through 1999.
For completeness, we note that the 1996 spectra of
Ori
exhibit stronger N V doublet absorptions than during 1982. This is among
the few cases in our study for which we could find a correlation
between these He II and N V temperature indicators.
The difficulty of seeing the correlation for other stars
is mainly due to the weakness of N V lines in their spectra.
![]() |
Figure 7:
IUE observations of ![]() |
Open with DEXTER |
We found several examples of rapid and long-term variability in
the He II line variability for this star.
Starting again with the long-variations, we noticed systematic differences
in the mean photospheric profile strengths for 1982 and 1996 observations.
The 1982 profiles exhibit a filling in of the red wing
and a tapered blue absorption wing.
The variations in the He II line of Ori are not always replicated
in the C IV and Si IV doublet lines, even though the doublets
show substantial inter alia variations at high velocities.
Sonneborn et al. (1988) discussed the wind activity in this star during 1982-3. Figure 7 exhibits one observation, SWP 21018, discussed in their paper and another obtained 16 days earlier. During this time the He II profile showed substantial activity over sometimes broad, and at other times narrow, wavelength ranges. The Si IV and C IV doublets exhibited apparent incipient emission fluctuations in their red wings during these times. In the blue/central parts of the line, the Si IV doublet changed very little and the C IV lines not at all. Some 13 days later, the He II line shows the same filling in on the red side as for SWP 21018, but the blue side of the line again shows full-strength absorption. An important interpretation from these comparisons of He II and resonance line variations is that changes observed at the base of the wind do not necessarily correlate well with those further out in the flow.
![]() |
Figure 8:
IUE observations of ![]() |
Open with DEXTER |
These lines also exhibit variations on a rapid timescale of 12 h; see
Fig. 8. In this case the central core of 1640 (solid
line) has deepened while the red wing has filled in. The
C IV doublet shows an overall weakening over the whole photospheric
components. The Si IV lines exhibit no change. We speculate that
because
Ori has a magnetic field,
these rapid variations might arise from dissipative magnetic processes
in the outer atmosphere.
This B2e star has been extensively observed and shows a rich activity
in its light curve, H,
and other spectral lines. The light
curve undergoes sporadic brightenings of up a few tenths
of a magnitude (Baade et al. 2001) for reasons unknown.
The H
emission component exhibits activity episodes over
a variety of amplitudes and timescales (e.g., Hanuschik et al. 1993).
The star's spectral lines are sharp, suggesting that it is viewed at a low
inclination.
Cen has been extensively monitored spectroscopically.
Rivinius et al. (1998b) reported that its line profile variability can
be decomposed into six nonradial pulsation periods. These modes cluster at 0.51 and 0.28 days. Rivinius et al. (1998a) have attempted to interpret
the Be outburst in this star as an outcome of nonlinearities associated
with the beating of these modes. In addition to activity associated with
intermodal beating, aperiodic activity also appears to be present. Peters
(1984b) noted that an optical He I line exhibited a rapidly appearing absorptions at velocities outside the photospheric
profile over several minutes. Such transients, since dubbed
"high velocity absorptions'' seem to represent
discrete ejections of blobs (Rivinius et al. 1998).
Peters (1998) has also documented evidence of rapid, large-amplitude emission variations
along the
6678 profile over several hours, suggesting that matter
is channeled in arc-line prominences.
![]() |
Figure 9:
A comparison of ![]() ![]() |
Open with DEXTER |
![]() |
Figure 10:
A comparison of the He II and the C IV lines for the interval
April 1986-February 1989 for ![]() |
Open with DEXTER |
In our survey of the 34 available SWP echellograms for
Cen, we found several variations of the
1640 line.
This included a narrowing of the He II, Si III, Si IV, and Al III lines
from September 1980 to February 1981 reported by Peters (1984a). However,
whereas Peters interpreted these differences in terms of line broadening
during an H
emission episode, we believe these differences should
be interpreted as a weakening of absorption and emergence of P Cygni emission
component during the later epoch, making the profiles appear narrow.
These differences can be seen in Fig. 9. Any possible doubt that
the faint bump in the red wing of the C IV doublet are emission in this
figure is dispelled by its behavior in Fig. 10.
This plot compares observations of the He II and C IV lines made in 1989
and 1996. The 1989 profiles are weak and narrow.
Although there is much activity in the red wings, the photospheric
components of C IV show no variations.
Just as for
Eri and
Ori, we were surprised to find
that relatively large variations in the He II line of
Cen can occur
on short as well as long timescales. Although we have found convincing
evidence for at least four cases in the He II line, residual red wing emission
is actually more common for the C IV and
Si IV doublets (Fig. 9). Moreover, short-term variations can occur
in resonance lines of less excited ions like Si III and Al III
(Peters 1984a). This suggests that the region of the wind affected extends
further downstream than is the typical for other stars in our sample.
Like Cen, 6 Cep is a B2.5e star with unusually narrow spectral
lines for a Be star. Pavlovski et al. (1997) monitored the star's optical
flux but were unable to find optical continuum variations.
In contrast, both its optical and UV spectrum are variable.
The C IV resonance lines exhibit strong variations
during oscillations of its wind state (e.g., Barker & Marlborough
1985; Grady et al. 1987; "GBS''). Abraham et al. (1993) have
speculated that the star's wind is responsible for the
creation of a "stellar wind bubble'', which they were able to image with
the IRAS satellite. Koubsky et al. (2005) have found a period of 1.621 days in the optical line profiles of 6 Cep.
These authors believed that they could
be attributed either to nonradial pulsations or a corotating disturbance or
cloud over the star. However, we favor the pulsation alternative because
in our present study we can see variations in lines arising from moderately
excited exitation states, as would be expected in the photosphere.
![]() |
Figure 11: A comparison 6 Cep's He II and C IV lines for epochs 1989.0 and 1990.5. Note the inactivity of the red wing emission in the C IV line. The wind was in a different state during 1982-86 (C IV, dotted line). |
Open with DEXTER |
The IUE archive contains 34 SWP echellograms of 6 Cep. From our inspection, we suspect that a few profiles undergo small amplitude variations. However, our statistical tests of these disclosed that only one was significant. A comparison of He II and the C IV doublet is shown in Fig. 11 for a pair of observations taken in 1989 and 1990. In this example, the He II line shows a general weakening of the photospheric profile and a distinctly raised red wing. The corresponding profile of the C IV doublet is consistent with true emission: its red wing is raised above the continuum level. Although the red wings of the C IV exhibit no changes during this time, the profiles develop a narrow absorption at about -150 km s-1. This is just one of a few examples shown in the present paper of different types of simultaneous variations of He II and C IV. The example in Fig. 11 also adheres to our more general finding that a filling in of red wing emission in the He II line tends to accompany strong absorption at low negative velocity in the C IV doublet. To give some perspective as to how these C IV profiles differ from the "norm'', we exhibit in Fig. 11 the profiles of both pairs of observations from 1989 and 1990 and the mean profile for 1982-6 (dotted line).
HD 67536 is a somewhat understudied variable B2.5-B3n star. On some
occasions its Balmer lines show emission (Hanuschik 1996).
The structure and significant variability of its C IV lines is similar
to that found in 6 Cep (GBS). Using this star as a prototype, ten Hulve
(2004) and Henrichs et al. (2005) have identified a class of "magnetic''
candidates based on the presence of variability between the rotational
velocity limits
in the line profiles of this doublet.
This working definition is reminiscent of the
characteristic periodic absorption/emission behavior of these lines noted
by Shore and colleagues (e.g., Shore & Brown 1990). However, the datasets
for most or all of the 24 Be stars so characterized are too sparse
to determine whether these variations are periodic.
![]() |
Figure 12:
A montage of four variations of the C IV and He II lines in
1983, 1986, 1994.7, and 1994.9 for HD 67536. The dashed line represents
the mean of the available 22 spectra. The photospheric components of the
C IV doublet of observation SWP 27503 (bold line) are nearly
completely filled in. The average spectra (dashed plot) are binned to 4
instead of 2 pixels. Note the slightly sharp, red displaced features (in
the first three cases in emission) in these lines. "5.1![]() |
Open with DEXTER |
In examining the 22 available IUE SWP echellograms for
HD 67536, we found that variations at velocities above -600 km s-1are readily apparent in the C IV and Si IV lines.
Figure 12 exhibits four spectra obtained at epochs 1983, 1986,
1994.7, and 1994.9 along with the mean of all observations.
Interestingly, the He II profiles in the first three exposures have a P Cygni-like character, which includes a narrow emission spike at about
+100 km s-1. (Because the profile variations are so complicated we
have not attempted to determine their statistical significances.)
The C IV lines also show arguably weak emission at
this velocity. Indeed in the SWP 27503 observation the doublet components
are almost completely filled in out to the DAC at -300 km s-1.
Incidentally, this peculiar phenomenon cannot be explained by
small changes in the
over the whole stellar surface because
the iron-group lines seem unaffected.
Judging from previous exposures, the C IV line spectrum was in this
filled-in state for at least 4 days when SWP 27503 was recorded.
In the fourth example, depicted in Fig. 12, an observation taken
in 1995.1, a narrow absorption appears, again at +100 km s-1, in
C IV and He II lines. We not know if it is significant that each of
these events occurs at nearly the same velocity.
In contrast to the previous Be stars, spectra in the second half of our sample exhibit variations that are usually correlated with blue wing variations in C IV and/or Si IV, and occasionally even Si III resonance lines.
1 Ori is in many ways a typical early-type Be star.
The spectral temperature diagnostics,
including
1640 and resonance line wind features, are
consistent with this classification. Like many other Be stars, it
exhibits cyclic H
emission episodes. This emission was strong
during the late 1970's and early 1980's (Barker 1983) when
Lamers & Waters (1987) estimated the mass loss of
1 Ori
to lie in the range 10-8-4
10
yr-1 from Copernicus, IUE, and IRAS data. This
is among the highest mass loss rate range noted in these authors' sample.
The IUE archive contain 26 SWP camera observations of this star. These are well enough distributed to give a sense of both rapid and long-term variations. From these data we found no evidence of rapid variability. Even over the long term, the fluctuations of this line are small compared to the moderate amplitudes of C IV activity. These variations extend over all possible velocities. Curiously, a strong "DAC-like'' feature is present in the C IV complex at about -200 km s-1 in all the observations.
![]() |
Figure 13:
The comparison of the stronger than unusual
He II and C IV spectra for ![]() |
Open with DEXTER |
The He II variations in 1 Ori seem to bridge the red-central
profile activity seen in the examples discussed above and for the
remaining stars in our sample. Up until the spectra of this star,
we have not encountered He II variations in the blue wing. In fact,
our first example, depicted in Fig. 13, exhibits
an increased absorption over the range +50 to -300 km s-1. The C IV lines show a similar variation in their blue
wings, but they also exhibits a weakening at high velocities
of -1000 to -1500 km s-1. Figure 14 exhibits a second example of He II variability. On this occasion absorptions are present, a narrow feature
at +100 km s-1 and a broad one in the blue wing. The narrow but
not the broad feature is present in the corresponding C IV observation.
![]() |
Figure 14:
A comparison of the He II and C IV spectra
for ![]() |
Open with DEXTER |
A rapidly rotating B2e star, Cen has been the subject of much recent
study. Its optical and UV lines and UV continuum
undergo regular short-term variations with
a dominant period near 0.64 days (Leister et al. 1995; Stefl et al. 1995;
Peters & Gies 2000). In addition, the star's H
line exhibits
strong oscillations over long timescales (Dachs et al. 1986). Rivinius (2005)
has suggested that the star ejects blobs that develop into ring-like structures.
Inspection of the C IV lines discloses considerable "slow''
variability in the range -1500 to +200 km s-1.
The line cores show a peculiar double-lobed
structure, and the core centroid positions vary between -50 and
-250 km s-1. Although the red wings of the C IV doublet are typically
filled in by emission, the changes in their profiles are small. To a lesser extent, these statements also apply to the Si IV doublet and even the Si III
1206 line.
The IUE obtained 28 SWP echellograms of this star
from 1983 through 1991. These were roughly evenly distributed between
two observing campaigns during each epoch.
The He II line shows correlated variations with C IV in several instances. First, we noticed long-term differences in
the sense that the wings of both lines were more depressed in 1991.
Rapid variability is also evident, and we display two examples. Figure 15 exhibits variations in these lines over the same velocity
range and over an interval of 7.4 h. During this time the He II line and the C IV doublet developed a low-velocity absorption at
about -200 km s-1. Figure 16 shows a weakening of emission over an interval of 7 h. Taken together, these two examples are the
only cases we have of activity in the blue and red wing of 1640 line over a time interval that is not much longer than the star's rotational period (1.5 days).
![]() |
Figure 15:
A comparison of the He II and C IV spectra during the
1991.2 campaign on ![]() |
Open with DEXTER |
![]() |
Figure 16:
A comparison of the He II and C IV lines during the
1991.2 campaign on ![]() |
Open with DEXTER |
Aqr is an active B1e star and is also a double-lined,
84-day spectroscopic binary. The component mass ratio is 0.16, which implies
that the secondary is an early-type A star (Bjorkman et al. 2002).
Detailed fits of the H
emission profiles suggest that we view this
star at a high inclination, i.e., i
70
(Hanuschik et al.
1996). The X-ray flux of this star is high for a Be star, indeed about
one half that of the X-ray anomalous Be star
Cas. Also, unusually
for a Be star, high energy emission has been detected by the EUVE satellite (Christian et al. 1999). Moreover, its IR flux is quite variable, even for a Be star.
Bjorkman et al. have suggested that the star's variable H
emission
can be used to measure a time-dependent mass transfer from the secondary star
on to the Be star's disk. (This assumes that the disk is formed by binary
accretion and not by decretion, as is ordinary for Be stars.)
The abnormal absorption strength of the He I
5876 is consistent
with the formation of part of this line close to the Be star's surface,
perhaps in the inner region of the disk.
The IUE observed Aqr 23 times during 1978-9
and 1985-91 with the SWP camera. Unusually for a Be star,
the N V doublet is not only present but strikingly variable.
Ringuelet et al. (1981) have reported emission in
these lines, and we attribute this to mass transfer from the companion
to the Be star. The
resonance lines of C IV, Si IV, and Si III exhibit a characteristic
range and type of variability for an active Be star. Moreover,
because its measured mass loss rate of
2.5
10
yr-1 (Freitas Pacheco 1982; Snow 1981) is typical for
a Be star, we believe that the variable blue wings of these lines
are due to fluctuations in the Be star's wind.
It remains to be added that the Be star's optical lines reveal the
presence of traveling bumps due to a 1.88 h oscillation (Peters &
Gies 2005; the grayscale in this paper offers an unusually clear
depiction of the increased acceleration of NRP bumps at the edges
of the line profile).
This star's He II line exhibits remarkable blue-wing
strengthenings which track strengthenings of the
strong blue wings of the C IV and Si IV doublets. Figure 17 exhibits the variation of these lines during epochs 1979.5 and 1979.8. Although not plotted, the Si III 1206 line shows the same enhanced absorption out to a common edge of
-1300 km s-1.
In Fig. 17 the blue wing of the He II line is enhanced out to -500 km s-1.Likewise, Fig. 18 exhibits both a blue wing strengthening
red wing weakening in 1993.9 relative to 1979.5.
The Si III and Si IV resonance lines show similar variations as C IV
over the intervals covered in our two figures. Because the statistical tool
we described in Sect. 3.2.2 evaluates changes for a chosen wavelength
interval, we exhibit the significance of our test for just the
low negative velocities of He II defined by the comb symbol.
![]() |
Figure 17:
A comparison of the He II and C IV
spectra
between epochs 1979.5 and 1979.8 for ![]() |
Open with DEXTER |
![]() |
Figure 18:
A comparison of the He II and C IV spectra between
epochs 1988.4 and 1993.9 for ![]() |
Open with DEXTER |
Although 2 Vul has not been classified as a Be star, we
included it in our sample because Zaal et al. (1997) had detected emission
in its near-infrared Brackett hydrogen lines. This
emission implies the presence of a thin disk. Although Percy et al.
(1988) have reported that this star has a photometric period near 0.61 days,
Balona (1995) found a period of 1.27 days. Hanula & Gies (1994)
discovered regular line profile variations, suggesting that these
variations are due to nonradial pulsation. Prinja (1989) has determined
a mass loss rate of 1
10
yr-1 from
resonance lines of several ions.
The IUE archive includes 39 SWP observations, of which
19 were recorded in a campaign at 1992.7. GBS and
ten Hulve (2004) found that the C IV lines are variable. Ten Hulve
considered it a "magnetic star''. The mean C IV profiles are extremely
strong, indeed so much so that the two components merge into a single broad trough with a minimum at -400 km s-1. Occasionally, the small-scale variations take the form of
narrow emissions centered at a few high velocities along the
profile. In this sense the resonance lines are more characteristic of an active O star than a Be star.
Rapid He II variations are not discernible in the IUE spectra of
this star. We have selected two examples of
long-term variability. Figure 19 shows that a difference
in the profiles of He II, the C IV doublet, and Si IV 1394
between an observation in 1989 and the epochal average for 1992. It
is clear that the wind was far stronger during the latter epoch and
produced enhanced absorption out to -1200 km s-1.
The behavior of the Si III
1206 line, not shown, is similar
to Si IV. This fact suggests that the variations are due to
large increases in mass rather than a shift in wind ionization.
In contrast, the DAC in the Si III profile is centered only at -800 km s-1, indicating that the position of this feature is governed by
wind ionization. The blue wing of the He II line is uniformly stronger in
1989 than 1992 and extends to -400 km s-1.
![]() |
Figure 19:
A comparison of the He II, C IV, and Si IV ![]() ![]() |
Open with DEXTER |
![]() |
Figure 20:
A comparison of the He II, C IV, and Si IV ![]() |
Open with DEXTER |
Our second example of 1640 variability
is shown in Fig. 20. This figure compares the behavior of He II, C IV, and Si IV lines between 1983 and 1992. This contrast is
smaller than the previous one with respect to 1989.
The C IV, Si IV, and Si III (not shown)
lines show variations in an opposite sense from He II -
for example, in the range -700 to -1000 km s-1. This suggests
that ionization shifts are at play in this case. The strengthening
of the blue wing of the He II line is small but consistent out to
at least -500 km s-1. Although this slow change occurs along
the photospheric profile too, this is another case in which the
blue-shifted absorption is formed in the wind.
19 Mon, the final star for which we found 1640 variability,
is also a rapidly rotating B1e star near the main sequence. It
exhibits at least two large-amplitude prograde nonradial pulsations with
periods near 5 h (Balona et al. 2002). Although the star's
Be type is based on emission twice detected at low dispersion, Balona et al.
have disputed whether the H
line really has ever displayed emission.
However, GBS and ten Hulve
(2004) have noted the variations of its C IV lines.
Ten Hulve considered this another example of a "magnetic star''.
For this reason, we are inclined to believe the original spectroscopic
emission reports and have included 19 Mon in our Be star sample.
Figure 21 exhibits the only possible example of a likely deviation of the He II line from its mean profile from the 15 available IUE spectra. The diminished absorption in this observation (SWP 50412) begins at line center and extends out to -700 km s-1. Remarkably, this variation is anticorrelated with the C IV and Si IV strengthenings. Yet, it appears once again that the He II formation region extends into the wind.
![]() |
Figure 21:
A comparison of the He II, C IV, and Si IV ![]() ![]() |
Open with DEXTER |
The He II 1640 variations we have found fall into one
of the following patterns:
Most of the morphologies just summarized involve similar variations in
the C IV and/or Si IV doublets. Therefore, they probably reflect changes in
the structures of the stars' winds. For the spectra of Eri at least,
we also note that some of these events correlate
with the appearance of He I optical line dimples - that is,
a sympathetic response in the C IV lines can also be found (Smith et al. 1996). We have stipulated that there are some He II events for which
correlations with extant optical or UV data do not occur. We cannot
conclude much from these particular cases.
A common event class, second in overall frequency to those displaying
an extended blue wing type, is the filling of the central and red-wing
profile regions, and sometimes the whole line. Attempts to interpret such
events encounter the ambiguity of weakened absorption versus true emission,
and thus cannot be straightforwardly attributed to changes in the wind.
For example, weakened absorption might be caused by a decreased effective
temperature over the visible disk, or by a decrease of the temperature
gradient within the photosphere. Were either a lowering of the
or the temperature gradient to occur for whatever reason, it would have the
effect of decreasing the equivalent widths of all excited photospheric lines,
as well as the wind components of the lines we have studied.
This prompts the question of whether the general
weakenings of the He II line also correspond to less blue wing absorption
in the C IV lines. In the examples we have shown with general line
weakenings, the answer seems to be "yes'' in about
of the cases.
This includes the events shown for
Cen, 2 Vul, 19 Mon,
and the Fig. 17 event for
Aqr. Thus, it
appears that an argument can be made that the thermal conditions of the
photosphere have a effect on the velocity and density
relations of the wind.
Smith et al. (1997) have constructed non-LTE models of static atmospheres
in order to examine the requirements needed to reproduce observed emissions
(or absorption weakenings) in 1640 and the red He I lines
of
Eri.
They found that emission can be produced within a moderately dense, heated
slab above the atmosphere by "Lyman pumped recombination''. In this
process the slab's helium atoms feel the effects of the
slab's own Lyman continuum radiation. He I line emission will result if the
slab is heated to about 50 000 K or illuminated by EUV flux having an equivalent radiation temperature. The process is efficient
for a slab density of
1011-12 cm-3 (which is incidentally the
typical density where the line cores of the He II line are formed in B dwarf atmospheres). Moreover, the slab should be thick enough for the
helium lines to be optically thick, while at same time allowing the
Lyman continuum to remain thin. Such emitting structures might take
the form of mild density blobs suspended above the photosphere.
Equivalent conditions might be produced
by a flattening of the density lapse rate in its upper regions.
What causes the filling in of the red wing of the He II line?
According to VCR, a heated, isotropic
wind, even with the mass loss rate expected for an early Be star, is capable
of producing visible P Cygni signatures in the He II, Si IV, and
C IV lines. However, except in Fig. 12 (HD 67536) the incipient
redshift emissions we have found are not accompanied by blueshiftings
of the absorption components: when the red wing is raised, the blue half
of the profile is usually unchanged. This observation runs contrary to
the predictions of P Cygni profiles for 1640 and
4686
from VCR models; the models predicting these features included
strong, fast winds (
=
10
yr-1) and
an equivalent
1) heated to about 10 000 K above
;
see also Hamann & Schmutz (1987).
Under these conditions line emission is produced efficiently because
recombinations to He1+ are sensitive to high density and
temperature. If the region coincides with
the base of the wind, the acceleration of the flow reduces flux shielding
in the wind, permitting more atoms to be exposed to the deep photospheric
radiation flow. In Sect. 4.1 we emphasized that the red emissions
of
1640 should not be described as true P Cygni profiles.
The VCR models of
1640 indicate that P Cyg profiles are most
easily produced if a heated chromosphere exists very close to a star.
In two of the examples we discussed, Figs. 10 and 11, the
response of the C IV doublet to weak red wing emissions in 1640 is
accompanied by increased blue wing absorption.
This fits with VCR's modeling results that a heated region can be placed
at a position too far from the star to influence the He II line but yet
where it would be still responsible for C IV absorption in the wind.
It is also important to point out that because VCR's "distant, heated
wind slab'' models produce red wing
1640 emission, one does not
have to resort to ad hoc "returning blobs'' to explain this
emission in the observed profiles.
Before undertaking their work on the He II line, Venero et al. (2000) had produced anomalous wind models that led to absorption and emission signatures in the C IV doublet similar to those they found later for He II. Although similar models for C IV have not been explored yet by any authors, one can surmise that the responses of these lines would be similar to those just outlined. One might guess that the effects on C IV would be amplified for those models in which the wind is heated far from the star. For example, weak red wing emissions are most visible in the C IV observations - see Figs. 9, 11, and arguably 12. The mildness of this emission may be used in the future models to constrain the distance of heated regions above the star.
In addition to the wind heating requirements, the work of Venero et al. demonstrates the intuitive result that variable He II characteristics, whether in absorption or emission, increase with the mass loss rate. In cases where we have found correlated blue wing variations in both Si III and Si IV lines (e.g., Figs. 17 and 19) we estimate from tests of moving slab models using the CIRCUS program (Hubeny & Heap 1996) that the mass loss rate must be enhanced by a factor of at least a factor of 10. This enhancement is too large to be an effect of refocusing of wind in a magnetic dipolar field.
We have argued that the high velocity absorptions of the He II line can be best understood by a change in the mass loss rate and probably the velocity acceleration law. In addition, the most likely explanation for the faint red wing emissions is that an unknown instability, possibly magnetic, heats an accelerating region of the wind. Finally, we have suggested that conditions within the photosphere are responsible for the relatively common line weakenings across the photospheric profile.
This paper provides a mini-atlas of He II 1640
variability for a group of 10 Be stars selected from
a much larger sample of early-type Be, Bn, and B normal stars.
We have identified several basic types of variability. Weak red emissions
and line weakenings occur over timescales of a few hours or less.
In terms of the variability timescales, we have noted that
the pattern of strengthening blue wings occurs over long timescales.
In our view this is most likely explained by changes in the wind
velocity law (cause unknown). Second, line weakenings
likewise occur preferentially over long timescales. Long-term
weakenings occur in half of our ten
1640-variable stars.
Third, weakenings over the whole line or only the red wing can occur
even within a few hours. We also point out that rapid variability was
found preferentially in the stars
Cen and
Eri.
We believe these events speak to intrinsic properties of these stars rather
than to observational sampling.
We have suggested that the properties which
change the surface and wind properties of these stars are mediated
by magnetic instabilities. This is among the few ways of
interpreting aperiodic variations of a single star on a timescale sometimes
much less than its rotation period. We note that our examples of
variable 1640 do not include known Bp stars. Magnetic
fields in these stars are thought to be dipolar and, most importantly,
stable over at least several years. Strong stable fields resist the influence
of hydrodynamical instabilites that might alter a wind's structure
or its geometrical flow. Therefore, we conjecture that at least for
some of these stars magnetic fields must be localized on the
surface. Velocity perturbations due to nonradial pulsations and
differential surface rotation would then offer plausible ways to
trigger magnetic instabilities in these multipolar configurations.
Acknowledgements
We thank Dr. Roberto Venero for clarifications on his work on the He II1640 line, Dr. Geraldine Peters for making available H
observations of
Ori, and an anonymous referee for improving the quality of this paper.