Astron. Astrophys. 318, 60-72 (1997)
7. Discussion
7.1. The H Emission Line
A significant result of these observations is that no obvious
excess absorption feature has been detected for ER Vulpeculae which
would indicate the presence of extended material surrounding the
system. This confirms the results of Barden (1984), Newmark (1990),
Lazaro & Arevalo (1994) and Eker, Hall & Anderson (1995).
Mention should be made of the small excess absorption feature
approximately 200 km s-1 blue-ward of the secondary star.
This feature appears to be real but is so weak that analysis was
impossible.
The observations have revealed a significant amount of excess
emission in the core of the H line for ER Vul.
Analysis of the velocity of this component strongly suggests that the
emission is arising predominantly on the secondary (G5 V) star
although the variation in the FWHM indicates that there is a small
amount of emission from the G0 V primary. The modulation of the excess
emission with phase would indicate the appearance and disappearance of
discrete active regions as different hemispheres of the stars are
presented to the observer. No obvious variation of the emission has
been detected for ER Vul over the phase of the observations. Although
the phase coverage of the observations is small (0.14) and therefore
inadequate to produce a significant variation, the velocity of the
emission is highly suggestive that it is not associated with a
transient phenomenon such as a flare or with localized individual
active regions but arises due to a global phenomenon. Although it has
been assumed that the velocity of the emission component is consistent
with it arising on the secondary star there is clearly a consistent
velocity shift of the emission by about 10 km s-1 blue-ward
of the secondary position. It is interesting to note that in the solar
case, observations in H reveal systematic
blue-shifts in the emission as a result of spicules (Beckers 1968;
Beckers 1972). Spicules are rapidly changing thin filamentary features
permeating the solar chromosphere at chromospheric temperatures (T
104 K) which often extend upward into
the hotter corona (T 106 K). However
these features cover only about 1-2% of the solar surface and
consequently the disk-integrated H solar
spectrum does not display the systematic velocities associated with
spicules. As will be demonstrated the plage coverage or filling factor
for the ER Vul secondary appears to be very high (see below) so this
observation may be of disk-integrated blue-shifts associated with
active regions such as spicules.
In solar-type stars the radiation temperatures are sufficiently
large or the electron temperatures are sufficiently low so that H
is dominated by photoionization; the Balmer
lines are consequently coupled to the photospheric radiation field
rather than the local line-formation region. For later stars or those
with enhanced chromospheres the H source
function can become dominated by collisional processes as the electron
density becomes higher and the photospheric radiation temperature
becomes lower. In this case the H core becomes
filled-in and indicates the presence of a highly active chromosphere.
Very high temperature chromospheres found in M-dwarfs actually drive
the H line into emission due to the dominance of
collisional excitation in the line-forming region. The actual details
of H formation in late-type stars are extremely
complex since many mechanisms and structures contribute to the line
profile. However the observation of substantial excess emission in ER
Vul is consistent with the general picture of an extremely active
chromosphere (see for example Houdebine, Doyle & Koscielecki
1995).
For single stars the filling-in of the H core
has been well documented (Zarro & Rodgers 1983; Herbig 1985;
Thatcher & Robinson 1993) and has also been observed in
chromospherically active binaries (Strassmeier et al. 1990;
Fernandez-Figueroa et al. 1994; Frasca & Catalano 1994; Eker, Hall
& Anderson 1995; Montes et al. 1995). For such binaries the
behaviour of the H line is often found to be
inconsistent with the photometric evolution and has been variously
attributed to emission arising from star-spot regions or plages
(Ramsey & Nations 1984; Newmark et al. 1990), from chromospheric
network-like structures (Huenemoerder, Ramsey & Buzasi 1990) or
from extended prominence-like material (Hall & Ramsey 1992b; Hall
& Ramsey 1994). Unfortunately the present study does not encompass
the variation of excess emission with phase since the primary goal was
to detect the presence of extended material during the eclipse of the
ER Vul system. However the analysis of the data has revealed evidence
that the filling-in of the H profile is due to
plage regions. This is also confirmed by the appearance of the He
I D3 absorption line in the spectrum (see
below).
It is extremely difficult to quantify excess emission for a single
star and most studies to date have involved surveys of large samples
of active stars with a view to deciphering correlations between
different chromospheric diagnostics and between these and other
stellar parameters such as spectral type or rotational period (Young
et al. 1989; Herbig 1985). Using subtracted spectra to derive physical
parameters is dangerous since the technique is best suited to
observing line variability in binaries across wide ranges in phase.
However some simple calculations concerning the size of the emitting
regions are possible as follows.
The continuum flux density in the H region
for ER Vul was calculated using the stellar atmosphere models of
Kurucz (1979) for a G5 V star with solar abundances. The actual
wavelength of the calculation was 6575Å and the effective
temperature was taken as 5770 K. The resulting flux density is 2.14
106 erg s-1 cm-2 Å-1
which differs from the black-body flux at H by
only 3%. The corresponding luminosity at H for
ER Vul is 1.5 1029 erg s-1 Å-1
assuming a radius of 1.07 . The mean equivalent
width of the excess H emission (0.596Å)
was then converted to flux units using the relation;
![[EQUATION]](img24.gif)
where is the H
continuum flux, is the equivalent width of the
excess emission and was taken as 2Å, the
width of the region used to define . The
resulting excess emission luminosity is 3.9 1029 erg
s-1.
Fraquelli (1984) gives a relationship between the volume emissivity
j, the electron density and temperature
of formation of the H line T based on the
assumption that the dominant emission mechanism is recombination. This
mechanism was considered in detail by Burgess (1958). In case B
of that analysis the assumption is made that the plasma is very opaque
to Lyman line radiation and that the absorption from level 1 to level
2 in hydrogen is exactly balanced by the inverse spontaneous
transition. The H line radiation then results
from cascades after electron capture and following absorption of Lyman
line radiation. The approximation of Burgess (1958) is better than 1%
for electron temperatures less than about 2.5 104 K. By
comparing the emissivity variation in a non-LTE scaled VAL C model of
solar plage regions for H (Vernazza, Avrett
& Loeser 1981), it has been shown (Andretta 1995, private
communication) that within an order of magnitude the results of
Fraquelli (1984) are valid. Hence the flux in the H
emission region is given by;
![[EQUATION]](img29.gif)
where V is the volume of the H
emitting region. The volume emissivity of H is
given by;
![[EQUATION]](img30.gif)
where is the ionization energy of hydrogen.
This equation is also based on the assumption that the proton and
electron densities are equal ( =
) in this region of the chromosphere which is a
reasonable assumption for a star of this spectral type. The same VAL C
models show that the region of formation of the H
line core is at least a factor of three in
electron density (1.0 1011
3.5 1010) and
at least 50% in temperature (6000 T
10000). Taking the ionization energy of hydrogen
as 2.18 10-11 erg then the corresponding range in the
volume of the emission region is 1.6 1032 - 2.0
1033 cm3. Assuming the emission region is a
homogeneous hemispherical shell then the chromospheric thickness for H
lies in the range = 0.045
- 0.42 . The chromospheric thickness reported for
the Sun is of the order 0.004 (Athay 1971) while
Fraquelli (1984) shows by a similar method that the two components of
HR 1099 (consisting of a G5 IV primary and a K1 IV secondary) have
chromospheric thicknesses of 0.2 and 0.06
respectively. This may mean that the higher
value is appropriate for ER Vul so that the main emission excess is
formed at higher temperatures and lower electron densities than
assumed by Fraquelli (1984). More detailed chromospheric modelling of
the H excess emission in active stars is
required before any further definitive statements can be made.
7.2. The Ca II IRT emission lines
The observations of ER Vulpeculae have also revealed the presence
of significant excess emission in the Ca II IRT line at
8498. Unfortunately the remaining two lines in
this triplet were just on the edge of the spectral coverage for these
observations and were impossible to analyse. Analysis of the velocity
of the excess emission has suggested that it arises almost equally on
both components of the system (probably the secondary is slightly more
active in these lines). The FWHM variation for ER Vul is also
consistent with a dual emission peak. This is similar to observations
of the BY Draconis-type variable DH Leonis studied by Newmark et al.
(1990) in which the secondary was found to be more active in the
Balmer lines while both stars showed almost equal Ca II
IRT emission.
Enhanced emission cores in the Ca II H and K lines
at 3933.66 and 3968.47 are
the primary optical indicators of chromospheric activity. Their source
functions are collisionally controlled and so these lines are
sensitive probes of the electron density and temperature in the
chromosphere. The H and K lines are favoured for chromospheric
modelling since they are extremely important cooling mechanisms and
their interpretation is relatively straight-forward. However less work
has been done on the Ca II IRT although they are formed
deeper in the atmosphere and are thus sensitive probes of the
temperature minimum region and the temperature rise to the so-called
Lyman plateau. Foing et al. (1989) observed the
8498 and 8542 IRT lines in a sample of stars
from F9 to K4 and found that these lines correlate well with the H and
K emission peaks. Linsky et al. (1979) also showed that filling-in of
the IRT lines was a good indicator of stellar activity. The IRT lines
are formed in the lower chromosphere of the Sun (Vernazza, Avrett
& Loeser 1981) and in the temperature plateau in active
chromosphere K2 dwarfs (Thatcher, Robinson & Rees 1991). In the
quiet solar atmosphere these lines are simple absorption lines but as
one goes to plages of brighter Ca II H and K emission
the IRT line cores brighten and eventually develops self-reversed
emission cores (Shine & Linsky 1972). On this basis it might then
be expected that stars with Ca II H and K emissions
comparable to those in solar plages will also exhibit IRT emission
cores rather than simple filling-in. However this has not been found
to be the case. Anderson (1974) surveyed 28 stars from F8 to M2 in the
8498 line and Linsky et al. (1979) studied 49
stars from F9 to K3 at 8542. Both studies
revealed no distinct emission features even in the most active stars,
although they did however display filling-in of the line cores. This
behaviour is also displayed by the active components of ER Vul; on the
evidence for plage regions on these stars we might expect to see
emission cores in the Ca II IRT but instead see only
filling-in. It has been suggested that such an emission core may be
smeared out by rotation or large velocity fields and thus appear as
filling-in (Linsky et al. 1979). However Thatcher & Robinson
(1993) pointed out that rotational broadening could not account for
the lack of emission cores in all cases. For ER Vul the rotation rate
is probably not sufficient to smear the emission core beyond detection
and so this star remains part of the Ca II IRT enigma.
It can therefore be assumed that some additional broadening mechanism
is at work in the plage emission from ER Vul. This is therefore
another area where detailed chromospheric modelling is required.
Although the observation of excess emission in the IRT suggests the
presence of non-radiative heating in the lower chromosphere for both
components of ER Vul, and is consistent with plage emission, a
reliable quantitative interpretation would be difficult without more
detailed modelling.
7.3. The He I D3 absorption line
The observations have revealed obvious excess absorption in the He
I D3 ( 5876) line for
ER Vulpeculae. The use of He I D3 as an
activity indicator has been largely ignored because it is extremely
weak in normal stars and is generally blended with difficult water
vapour lines when observed from low-to-medium altitude observatories.
However observations of the stronger 10830 line
have been presented for large numbers of stars of different classes by
Vaughan & Zirin (1968), Zirin (1976) and Zirin (1982) although
observations of the D3 line are less common.
The triplet lines of He I at
5876 and 10830 appear in absorption in the solar
spectrum; the weaker D3 absorption feature appears to be
cospatial with plage regions and absent elsewhere. At
10830 the absorption is strongest above active
regions and very weak (with a tendency to be in emission) in coronal
hole regions (McCabe & Mickey 1981). Landman (1981) studied high
resolution spectra of He I D3 taken for
solar plage regions. In stellar work Wolf & Heasley (1984)
observed He I in 18 late-type stars and showed that the
line depths, widths and the ratio 10830/
5876 in dwarf stars was similar to that measured
in solar plages. However they found that the line ratio in giants was
much larger than in either solar plages or active dwarfs and suggested
that the conditions under which He I forms may be very
different in highly luminous stars. The factors which control the
formation of the He I triplet lines, in particular
D3 and 10830, are however not well
understood. The basic problem is that the transition region models
based on UV and EUV resonance lines cannot account for the observed
intensity levels in the quiet Sun He I resonance lines
(Jordan 1975). Zirin (1975) suggested that the He I
triplet levels are populated by radiative recombinations following
photoionization of He atoms by coronal far-UV and X-ray line and
continuum radiation. Giampapa et al. (1978) argued that He
I D3 line they found in AD Leonis was not
excited by recombination following photoionization since such an
assumption would also require unrealistically high X-ray luminosity.
Instead they suggest that He I
5876 is excited by collisions from the ground state in the hotter
(T 8000 K) region of the stellar
chromosphere. Recently Andretta, Giampapa & Jones (1995) suggest
on the basis of their non-LTE calculations that at least some of the
He I spectral features in the Sun and late-type stars
could be linked to regions of enhanced UV and X-ray emission but that
He I formed by photospheric ionization and decay and
collisional processes seemed to be present even for the case of an
inactive corona.
Despite the confusion as to the details of He I
formation in the Sun and stars it is now almost certain that active
regions on the stellar surface are the dominant area contributing to
the observed flux profiles. The equivalent width of D3 is
controlled by the temperature-density profile in the middle
chromosphere and the fraction of disk covered by plages. The
observation of He I D3 absorption in ER Vul
is therefore very suggestive of large areas of plage-like plasma in
the chromospheres of one or both components in this binary. The
equivalent widths of the absorption remain fairly constant with a mean
value of 78 mÅ. The velocity measurements of the excess
absorption are however confused for all measurements and it can only
be assumed that this is due to the weakness of the line and noise
effecting the gaussian fitting. It is a reasonable assumption that the
majority of the absorption is occurring on the secondary component of
the binary since this appears to be the more active star with a higher
probability of plage regions. The FWHM variation however clearly
implies that absorption is occurring for both stars.
Andretta & Giampapa (1995) presented a computational approach
to the He I triplet lines in dwarf F and G stars. They
show that the lines can be utilised to infer the fractional area
coverage or filling factor of active regions on stellar surfaces if
the intrinsic absorption strengths are known. They compute He
I profiles using a grid of model chromospheres
superposed on late-type dwarf photospheric models. They estimate that
in both F and G type stars a conservative value for the maximum
absorption equivalent width ( ) in He
I D3 is 100-150
mÅ. This calculation ignores the quiet component and is
therefore appropriate for measurements on the subtracted spectra. A
lower limit to the filling factor is then given by;
![[EQUATION]](img36.gif)
where is the observed He I
equivalent width. Using this relationship the filling factor for the
secondary component of ER Vul is 0.53. This is extremely high even if
we apportion this value between the two blended components.
Unfortunately there are no measurements of the He I
10830 line for this system which could be used
to verify this result. However Andretta & Giampapa (1995) report a
filling factor of between 0.5 and 0.6 for the active G8 V star
Boo A while O'Neal, Neff & Saar (1994)
deduced a filling factor of between 0.41 and 0.52 for II Peg. So
although ER Vul has a high filling factor this is not inconsistent
with other observations of highly active stars (Wolf & Heasley
1984).
Andretta & Giampapa (1995) also computed the theoretical
dependency of the 5876 equivalent width with the
excess emission equivalent width in the core of the H
line from plage regions, and these predictions
showed good agreement with numerous observations of solar plage
regions. The results for ER Vul show that the He I
D3 and H measurements are again
entirely consistent with both observational and theoretical
considerations of active plage regions in the solar analogy. In
conclusion the He I D3 observations for ER
Vul have demonstrated the existence of very large filling factors
presumably associated with the secondary component. ER Vul appears to
be highly active with regards to plage regions and the existence of
such slab-like structures is entirely consistent with the lack of
detection of extended material around the system.
7.4. The Mg I b and Na I D1 & D2 lines
Strong neutral metal absorption lines are of special interest for
stellar activity studies since they are formed in the lower
chromosphere and the region of temperature minimum for solar type
stars. Basri, Wilcots & Stout (1989) concluded in their study of
29 main-sequence stars that the Mg I b lines were good
diagnostics of photospheric activity. The Na I lines
are collision dominated due to the small photoionization cross-section
and relatively large collisional cross-section for Na I
(Johnson 1964). This means that the source functions of the Na
I lines are functions of electron temperature and
pressure and so are good indicators of changes in the lower
chromosphere.
The observations have revealed a significant amount of excess
emission in the Mg I b lines from ER Vulpeculae.
Measurements have been made on the lines at
5172.68 and 5183.61 while the line at
5167.33, although present, was too weak to give
reliable measurements. The velocities of the two measured lines show
that the emission is again predominantly from the secondary component
although the FWHM variation is consistent with a small amount of
emission from the primary. In contrast, no excess emission has been
detected in the Na I D1
( 5889.95) and D2
( 5895.92) lines in the observations of ER Vul.
This is significant since both these neutral metals form in the lower
chromosphere. Furthermore the ionization energies of Mg I and Na I are
7.6 eV and 5.1 eV respectively indicating a disparity between the
observed emission and that expected from simple energetics. We regard
it unlikely that crowded atmospheric absorption could account for a
difference in EW between the standard and target stars. An explanation
may lie in the existence of a complex chromospheric structure and/or
opacity effects.
Further theoretical work is required before any definitive
statements about the lack of excess Na I emission in ER
Vul can be made. Current work underway by Andretta, Doyle & Byrne
(1996) using a more complete set of opacities will provide a better
insight into the use of these lines in studies of stellar
activity.
© European Southern Observatory (ESO) 1997
Online publication: July 8, 1998
helpdesk.link@springer.de  |