M.-A. Dupret 1 - J. De Ridder 2 - P. De Cat
2,
- C. Aerts 2 - R. Scuflaire 1 -
A. Noels 1 - A. Thoul 1,
1 - Institut d'Astrophysique et de Géophysique de
l'Université de Liège, allée du 6 Août 17,
4000 Liège, Belgium
2 -
Instituut voor Sterrenkunde, Katholieke Universiteit
Leuven, Celestijnenlaan 200, 3001 Leuven, Belgium
Received 17 September 2002 / Accepted 7 November 2002
Abstract
We present an improved version of the method of photometric mode
identification of Heynderickx et al. (1994).
Our new version is based on the inclusion of precise
non-adiabatic eigenfunctions determined in the outer stellar atmosphere
according to the formalism recently proposed by Dupret et al. (2002a).
Our improved photometric mode identification technique is
therefore no longer dependent on ad hoc
parameters for the non-adiabatic effects. It contains the complete physical
conditions of the outer atmosphere of the star, provided that rotation does
not play a key role. We apply our method to the two slowly
pulsating B stars HD 74560 and HD 138764 and to the Cephei star EN (16) Lac. Besides identifying the degree
of the pulsating
stars, our method is also a tool for improving the knowledge of
stellar interiors and atmospheres, by imposing constraints on parameters
such as the metallicity and the mixing-length parameter
(a procedure we label non-adiabatic asteroseismology).
Key words: stars: oscillations - stars: atmospheres - stars: variables: general
A crucial problem in asteroseismology is mode identification. Firstly
because, from a theoretical point of view, despite the linear non-adiabatic
predictions, the mode selection mechanisms are not well understood for many
kinds of pulsating stars ( Scuti,
Cephei, slowly pulsating B stars,
Doradus, roAp stars, ...). Secondly because, from an
observational point of view, we do not resolve the disks of stars
other than the Sun so that we can only observe disk-integrated
quantities. Thirdly, the rotational splittings
and the "avoided crossing'' effect produce such a complicated
power spectrum that a mode identification based on the frequencies
alone is generally impossible.
Currently, both spectroscopic and photometric mode identification techniques are being used. The latter methods are based on multi-colour photometry, and are the subject of this paper. The principle of these methods is to observe the photometric variations due to stellar oscillations in different colours and compare them to the theoretical predictions at the appropriate wavelengths (Dziembowski 1977b; Stamford & Watson 1981; Watson 1988; Garrido et al. 1990; Heynderickx et al. 1994; Garrido 2000). However, all these methods have an important drawback: their theoretical predictions are very sensitive to the non-adiabatic temperature variations at the photosphere (Cugier et al. 1994; Balona & Evers 1999).
Dupret et al. (2002a) developed a non-adiabatic code including a detailed
treatment of the pulsation in the outer atmosphere. In this paper, we show how
this non-adiabatic treatment opens the way to a significant improvement of the
discriminant power of mode identification methods based on multi-colour
photometry. Indeed, by using Dupret et al.'s calculations, we are able to
eliminate the weakest point of the mode identification method, namely the
ad hoc parameters to express the non-adiabaticity. We present the results
obtained for two slowly pulsating B stars (SPBs) observed by De Cat & Aerts
(2002) and for the Cephei star EN (16) Lac.
Theoretical expressions for the monochromatic magnitude variations of a
non-radially pulsating star have been derived by different authors. Dziembowski
(1977b) was the first to derive an expression for the bolometric
magnitude variation of a non-radially pulsating star. He suggested also
that a Wesselink technique could be formulated from these expressions.
Balona & Stobie (1979) recast the suggestion of Dziembowski in an
observationally feasible way.
Stamford & Watson (1981) derived an expression for the monochromatic
magnitude variations of a non-radially pulsating star. They proposed to compute
the local emergent monochromatic flux variation on the base of equilibrium
atmosphere models (see Eq. (2) below) and they simplified the way to
compute the influence of the stellar surface distortion. Watson (1988)
improved the method by taking the variation of the limb darkening into account
(see Eq. (3) below), and discussed the importance of the different
terms in the equation giving the
monochromatic magnitude variations of a non-radially pulsating star. Garrido et al. (1990) and Garrido (2000) derived a method of mode identification
using Strömgren photometry, based on the formalism of Watson (1988), and
applied it to Scuti and
Doradus stars. Heynderickx et al.
(1994) derived an expression for the surface
distortion of a non-radially pulsating star in a Lagrangian formalism. He
developed a method of mode identification based on photometric amplitude ratios
and applied it to
Cephei stars.
In all the previously cited papers, the non-adiabatic character of the pulsation
was neglected or treated with an ad hoc parameter. Cugier et al. (1994) were the first to use non-adiabatic computations for
photometric mode identification, using the non-adiabatic pulsation
code of Dziembowski (1977a). The same code was also used by
Balona & Evers (1999) for mode identifications of Scuti stars.
Townsend (2002) used a non-adiabatic code for the photometric modelling
of SPBs.
It is useful to detail the assumptions made, sometimes implicitly, by the previous authors and in our method, in order to derive an expression for the monochromatic magnitude variations of a non-radially pulsating star. In Sects. 3 and 4, we will put forward the improvement and specificities of our method, compared to the one of the previous authors.
We note that assumption (a7) concerns only the flux. We do not make this hypothesis for the temperature which depends strongly on the optical depth in stellar atmospheres (see Dupret et al. 2002a, Fig. 1).
Assumptions (a5) and (a9) have the same physical justification as explained in
Dupret et al. (2002a). Because of the very small thermal relaxation time of
the atmosphere, we assume that, at each time of the pulsation cycle, the local
atmosphere remains in radiative equilibrium. Using this assumption,
Dupret et al. (2002a) assumed that, during the pulsation cycle, the
temperature distribution ( law) in the local atmosphere was, for each
given time, the same as the temperature distribution of an equilibrium
atmosphere model. For the same physical reasons, we make now the same
assumption for the local monochromatic outwards flux
and
for the limb darkening law
.
From assumptions (a5), (a7) and (a8), the monochromatic outward flux
in the local atmosphere is given by:
We denote by
the amplitude of relative radial displacement
in the photosphere:
The quantities
and
can be computed by the non-adiabatic code of Dupret et al. (2002a). Because the Eulerian variation of the gravitational potential at
the surface is always very small (
), it appears that
in very good approximation,
is in opposite phase
with the radial displacement. These two quantities can be expressed in term of the
associated Legendre functions:
![]() |
![]() |
![]() |
(8) |
![]() |
![]() |
![]() |
(9) |
![]() |
![]() |
![]() |
(10) |
In Eq. (7), the term proportional to
corresponds
to the influence of the stellar surface distortion, the term proportional to
corresponds to the influence of the local effective temperature variation
and the term proportional to
corresponds to the influence of the local effective
gravity variation. In our applications, we computed the coefficients
and
(derivatives of the monochromatic
flux) from the models of Kurucz (1993). An analytical law for the limb
darkening is needed for the computation of
and its
derivatives. For the present paper, we used a quadratic law (Wade & Rucinski
1985). We note that an improved non-linear limb darkening law has been
proposed by Claret (2000), but his computations were only done for
Strömgren filters, while our applications concern Geneva and Johnson
filters.
In multi-colour photometry, one observes the integral
of the monochromatic magnitude variation over the response of the filter:
The linear theory does not permit us to predict the amplitudes of the
eigenfunctions. Therefore, it is appropriate to use amplitude ratios and phase
differences between different filters when comparing the theoretical predictions
to the observations. On one hand, for Scuti stars, the observations and
the theoretical predictions give significant phase-lags between the different
filters. Mode identification methods using these phase-lags have been proposed
by Garrido et al. (1990) and Balona & Evers (1999). On the other
hand, for
Cephei stars, SPBs and
Doradus stars, no phase-lags
are observed (in agreement with the small phase-lags predicted by the theory).
For the latter stars, mode identification methods based on amplitude ratios are
thus appropriate (Heynderickx et al. 1994). We adopt here such a
method. The theoretical procedure of our mode identification method is the
following:
![]() |
(13) |
The difference between our method and the one proposed by other authors is in the way of estimating the influence of the effective temperature variation and the effective gravity variation in Eqs. (7) and (12).
As a preliminary for the estimation of the relative effective temperature
variation
and the relative effective gravity
variation
,
the first step of the procedure followed by
Watson (1988), Heynderickx et al. (1994),
Cugier et al. (1994) and Balona &
Evers (1999) was to compute
at the photosphere. As
initially proposed by Buta & Smith (1979), they used the following formula:
![]() |
(15) |
Watson (1988), Garrido et al. (1990) and Heynderickx et al.
(1994) related the Lagrangian temperature variation to the Lagrangian
pressure variation at the photosphere by introducing a free
ad-hoc parameter R describing the departure from adiabatic conditions.
Moreover, they assumed that the Lagrangian temperature variation is equal
to the local effective temperature variation at the photosphere:
Our approach does not make any of these assumptions. We rigorously compute both the amplitude and the phase of the local effective temperature variation by non-adiabatic computations throughout the entire star and, in particular, throughout the entire non-grey atmosphere.
Cugier et al. (1994) and Balona & Evers (1999)
performed non-adiabatic computations in order to determine
and
in a more rigorous way. In their method, they assume also that
the Lagrangian temperature variation is equal
to the local effective temperature variation at the photosphere
(Eq. (18)).
We do not make this assumption in our method, since we have shown in Dupret et al. (2002a) that the Lagrangian variation of the temperature at the
photosphere is different from the variation of the local effective temperature,
because of the significant Lagrangian variation of the optical depth.
Stamford & Watson (1981), Watson (1988), Garrido et al.
(1990) and Heynderickx et al. (1994) related the coefficient
corresponding to the effective gravity
variation (i.e. the gravity variation corrected for the pulsational
acceleration) to the Lagrangian pressure variation by the following equation:
Dupret et al. (2002a, Eq. (24)) proposed a more accurate way to
determine
.
In our method,
is given by:
In this section, we present the application of our mode identification
method to two SPBs and one Cephei star. The theoretical stellar
models we used have been computed by the Code Liégeois d'Évolution Stellaire
(CLÉS).
The two SPBs for which we performed non-adiabatic computations and a photometric mode identification are HD 74560 and HD 138764. Data obtained with Geneva photometry were taken from De Cat & Aerts (2002). We give in Table 1 the effective temperature and the gravity of these two stars, as derived from the most recent calibration of Künzli et al. (1997). We selected then theoretical models closest to these observations and their global characteristics are given in Tables 2 and 3. Numerous additional applications to other SPBs will be presented in De Cat et al. (in preparation).
HD 74560 |
![]() |
![]() |
HD 138764 |
![]() |
![]() |
![]() |
![]() |
![]() |
![]() |
![]() |
age (My) = 24.5 |
X = 0.7 | Z = 0.02 | no overshooting |
![]() |
![]() |
![]() |
![]() |
![]() |
age (My) = 38 |
X = 0.7 | Z = 0.02 | no overshooting |
![]() |
gn | ![]() |
![]() |
1 |
![]() |
10.08 | -23.0 |
2 | g29 | 22.38 | -8.2 |
3 | g41 | 36.01 | 1.3 |
![]() |
gn | ![]() |
![]() |
1 | g16 | 5.59 | -23.2 |
2 | g28 | 17.23 | -9.3 |
3 | g40 | 31.57 | 1.4 |
![]() |
Figure 1:
![]() ![]() ![]() ![]() ![]() |
Open with DEXTER |
![]() |
Figure 2:
![]() ![]() ![]() ![]() ![]() |
Open with DEXTER |
![]() |
Figure 3:
![]() ![]() ![]() ![]() ![]() |
Open with DEXTER |
![]() |
Figure 4:
![]() ![]() ![]() ![]() ![]() |
Open with DEXTER |
![]() |
Figure 5:
Amplitude ratios obtained with Geneva photometry for the dominant mode of
the SPB star HD 74560. The bullets with error bars correspond to the
observations. The lines correspond to the theoretical predictions for
different degrees ![]() ![]() ![]() ![]() |
Open with DEXTER |
![]() |
Figure 6: Amplitude ratios obtained with Geneva photometry for the dominant mode of the SPB star HD 138764, same caption as in Fig. 5. |
Open with DEXTER |
In Figs. 1 and 3, we give the values of (local effective temperature variation for a normalized radial displacement at
the photosphere) and in Figs. 2 and 4 those of
(phase difference between the local effective temperature variation and
the radial displacement at the photosphere, in degrees) for different modes of the
two SPBs, as a function of the
pulsation frequency in c d-1, as computed by our non-adiabatic code. In
these figures, the vertical line corresponds to the observed frequency of the
dominant mode.
We see that, for a given frequency, the amplitude and phase-lag
depend strongly on the degree
.
The physical explanation is the
following. The term corresponding to the transversal compression in the equation
of mass conservation is proportional to
.
For high-order g-modes,
this term dominates, which implies a strong dependence on
of the
eigenfunctions. The phase-lags (
)
are relatively close to zero, which is
in agreement with the observations.
In Tables 4 and 5, we give for the two stars, the
degree ,
the radial order,
and
for the modes with
theoretical frequency closest to the observed frequency. The identified mode is
given in bold.
In Figs. 5 and 6, we give the amplitude
ratios obtained from Geneva photometry for the dominant modes of the two stars.
The bullets with error bars correspond to the observations. The full lines
correspond to the
theoretical predictions for different degrees :
solid line for
,
dashed line for
and dot-dashed line for
.
For both stars, a
solution inside the error bars is found, and the identified degree is
.
These photometric mode identifications are in very good agreement with
the spectroscopic mode identifications performed by De Cat et al. (in
preparation) using the moment method.
We present now the application of our method to the Cephei star
EN Lac. This star has been studied by many authors. We refer to Chapellier et al.
(1995) and Lehmann et al. (2001) for a summary of the observational
studies and to Dziembowski & Jerzykiewicz (1996) for the first
seismic study of this
Cephei star. In this section, we illustrate the
process we term non-adiabatic asteroseismology, by deriving constraints on the
metallicity of the star. The three observed frequencies used in our study were
taken from Lehmann et al. (2001) and the photometric amplitudes obtained
with Johnson filters were derived by Jerzykiewicz (1993).
Model 1a | ||
![]() |
![]() |
![]() |
![]() |
age (My) = 16.2 | Z = 0.015 |
Model 1b | ||
![]() |
![]() |
![]() |
![]() |
age (My) = 15.15 | Z = 0.015 |
Model 2a | ||
![]() |
![]() |
![]() |
![]() |
age (My) = 15.9 | Z = 0.02 |
Model 2b | ||
![]() |
![]() |
![]() |
![]() |
age (My) = 13.95 | Z = 0.02 |
Model 3a | ||
![]() |
![]() |
![]() |
![]() |
age (My) = 14.85 | Z = 0.025 |
Model 3b | ||
![]() |
![]() |
![]() |
![]() |
age (My) = 12.9 | Z = 0.025 |
Aerts et al. (2002) showed
convincingly that the spectroscopic mode identification is fully compatible with
the photometric one for this star, and points towards a radial mode for the
first frequency (
f1=5.9112 c d-1), an
mode for the second
frequency (
f2=5.8551 c d-1) and an
mode for the third
frequency (
f3=5.5033 c d-1).
We therefore adopt this result here. We present the results of the amplitude
ratios obtained for models with different metallicities. The choice
of the models has been made with the following procedure.
We computed models with 3 different metallicities: Z = 0.015, Z = 0.02 and
Z = 0.025. Because of the uncertainties in the calibration of the effective
temperature of EN Lac (see Jerzykiewicz & Sterken 1980;
Shobbrook 1985 and Dziembowski & Jerzykiewicz 1996),
we computed for each metallicity two evolutionary tracks with two different
masses.
For each evolutionary track, we subsequently selected the model giving
the best agreement between the theoretical and observed frequencies, relying on
the unambiguous mode identification.
In all these models, X = 0.7 and there is no overshooting.
Their global characteristics are given in Table 6.
In Fig. 7, we present the values of as a function of the pulsation frequency in c d-1, for different modes and
for the six models of EN Lac given in Table 6. The three vertical
lines correspond to the three observed frequencies.
We see that, the higher
the metallicity, the lower the amplitude of the local effective temperature
variation for a normalized radial displacement (
). The physical origin of this
phenomenon is explained in Fig. 8, where we show the amplitudes of
the luminosity variation
as a function of the logarithm of
temperature, from the center to the surface of the star, for the radial
fundamental mode and for the models 1a, 2a and 3a of Table 6. The
higher the metallicity, the more efficient the
mechanism, which implies
a more important decrease of the luminosity variation in the driving region.
Therefore, the amplitude of the luminosity variation and of the local effective
temperature variation at the photosphere are smaller for a normalized
displacement. The phase differences between the local effective temperature
variation and the radial displacement at the photosphere we obtained for the
different models are very close to
(in agreement with the observations) and we do not give them here.
By comparing in Fig. 7 the results obtained for the cold models
(models 1a, 2a and 3a) and the hot models (models 1b, 2b and 3b),
we see that changing the values of
within its observational error bars (keeping the
metallicity constant) has only a very small effect on the non-adiabatic results.
![]() |
Figure 7:
![]() ![]() ![]() ![]() ![]() ![]() |
Open with DEXTER |
We present in Fig. 9 the theoretical amplitude ratios (Johnson photometry) obtained for three models of EN Lac with different metallicities (Table 6, models 1a, 2a and 3a), and for the fundamental radial mode. We see that the model with Z=0.015 gives the best agreement between the theoretical and observed amplitude ratios. We have checked explicitly that all the modes in the observed range of frequencies remain unstable for this low metallicity; lower values are not compatible with mode excitation.
The confrontation between the theoretical and observed amplitude ratios
can thus be used as a constraint on the metallicity of stars driven by the metal
opacity bump ( Cephei and Slowly Pulsating B stars), once we know
the identification of the mode.
We have seen in Fig. 7 that the non-adiabatic predictions and
thus the theoretical amplitude ratios are little affected by the
uncertainties on
for a given metallicity,
so that the constraints we derived on the metallicity are reliable.
This way of determining the metallicity may even turn out to be more
precise than the classical method based on the analysis of the spectrum. We plan
to validate our method to derive the metallicity by this feed-back process by
confronting our predictions to those of
Cephei stars for which the
metallicity is known with high accuracy.
We note that, for rapidly rotating Cephei stars, the interaction between
pulsation and rotation can affect significantly the photometric amplitudes and
phase-lags as shown recently by Daszynska et al. (2002). This
interaction was not yet taken into account in our current non-adiabatic
treatment.
![]() |
Figure 8:
Amplitude of luminosity variation
![]() |
Open with DEXTER |
![]() |
Figure 9: Observed and theoretical amplitude ratios (Johnson photometry) for the radial fundamental mode obtained for three models of EN Lac with different metallicities (Table 6, models 1a, 2a and 3a). The solid line corresponds to the model with Z = 0.015, the dashed line to the model with Z = 0.02 and the dotted line to the model with Z = 0.025. The error bars correspond to the observations. |
Open with DEXTER |
We have presented an improvement of the often-used photometric mode
identification method. Our version of this method is based on precise
non-adiabatic computations in which special attention is paid to the treatment
of the pulsation in the stellar atmosphere (Dupret et al. 2002a). We have
applied our new version of the method to identify the main mode of two SPBs
(HD 74560 and HD 138764) and one Cephei star (EN Lac, see also Aerts et al. 2002). In both cases, our photometric mode identifications were in very
good agreement with the spectroscopic mode identifications, which are far less
sensitive to temperature variations (De Ridder et al. 2002). We have
shown also that the confrontation between the non-adiabatic theoretical
predictions and the observations can give interesting constraints on the models.
We have used the term non-adiabatic asteroseismology for this feed-back
process. More precisely, for
Cephei stars and SPBs, the non-adiabatic
predictions are very sensitive to the metallicity, so that this parameter can be
constrained for these stars once definite mode identification is achieved. For
Scuti and
Doradus stars, the non-adiabatic predictions are
very sensitive to the characteristics of the thin superficial convection layer
(Balona & Evers 1999; Moya et al. 2002; Dupret et al. 2002b).
Therefore, it is to be
expected that a feed-back process similar to the one we presented will lead to a
significant improvement of our understanding of this convection layer.
Acknowledgements
Part of this work was supported by the F.R.I.A. (Fonds pour la formation à la Recherche dans l'Industrie et dans l'Agriculture). The authors are members of the Belgian Asteroseismology Group (BAG, http://www.ster.kuleuven.ac.be/~conny/bag.html).