A&A 404, 1099-1106 (2003)
M. S. Dimitrijevic1,2 - T. Ryabchikova3,4 - L. C. Popovic1,2 - D. Shulyak5 - V. Tsymbal5
1 - Astronomical Observatory, Volgina 7, 11160 Belgrade 74, Serbia
2 - Isaac Newton Institute of Chile, Yugoslavia Branch
3 - Institute of Astronomy, Russian Academy of Science, Pyatnitskaya 48, 119017 Moscow, Russia
4 - Institute for Astronomy, University of Vienna, Türkenschanzstrasse 17, 1180 Vienna, Austria
5 - Tavrian National University, Yaltinskaya 4, 330000 Simferopol, Crimea, Ukraine
Received 12 November 2002 / Accepted 27 March 2003
We study the influence of Stark broadening and stratification effects on Sii lines in the rapidly oscillating (roAp) star 10 Aql, where the Sii 6142.48 Å and 6155.13 Å lines are asymmetrical and shifted. First we have calculated Stark broadening parameters using the semiclassical perturbation method for three Sii lines: 5950.2 Å, 6142.48 Å and 6155.13 Å. We revised the synthetic spectrum calculation code taking into account both Stark width and shift for these lines. From the comparison of our calculations with the observations we found that Stark broadening + the stratification effect can explain asymmetry of the Sii 6142.48 Å and 6155.13 Å lines in the atmospere of roAp star 10 Aql.
Key words: stars: chemically peculiar - line: profiles - atomic processes
The Stark broadening is the most significant pressure broadening mechanism for A and B stars and one has to take into account this effect in investigation, analysis and modeling of their atmospheres. In one of our previous works (Popovic et al. 2001) we have shown that the Stark effect may change the equivalent width of spectral lines by 10-45%, hence neglecting this mechanism, we may introduce a significant error in abundance determinations. On the other hand, high resolution spectra allow us to study different broadening effects using line profiles. In the course of the abundance analyses of peculiar (Ap) stars we noticed that some of the Sii lines are shifted relative to the laboratory wavelength. Moreover, few strong lines mainly from the multiplets and have asymmetrical line profiles, in particular the Sii 6155.13 Å line. We found that this line is slightly shifted and asymmetrical even in the solar spectrum, while in the hotter, e.g. Ap stars, the shift and asymmetry are more pronounced.
The aim of this paper is to explain the asymmetry of Sii lines within the framework of the Stark broadening effect. First we calculated the Stark broadening parameters for these lines, after that we calculated synthetic spectra and compared them with the observations, and finally, we discuss the effects causing shift and asymmetry in the line profile of the Si 6155.13 and 6142.48 Å lines.
For our analysis we used observations of one normal star HD 32115, two Ap stars HD 122970 and 10 Aql and Solar Flux Atlas by Kurucz et al. (1984). High resolution CCD spectra of 10 Aql and HD 122970 are described by Ryabchikova et al. (2000). High resolution CCD spectra ( ) of HD 32115 in the wavelength region 4000-9500 Å were obtained with the coude-echell spectrometer mounted on the 2 m "Zeiss'' telescope at Peak Terskol Observatory, Russia (see Bikmaev et al. 2002 for more details).
More Ap stars show peculiar line profiles of Sii lines but most stars have rather strong magnetic fields which distort line profiles through Zeeman splitting. Rather weak magnetic fields in Ap stars HD 122970 and 10 Aql allow us to ignore magnetic effects on line shape.
Calculations have been performed within the semiclassical perturbation formalism, developed and discussed in detail in Sahal-Bréchot (1969a,b). This formalism, as well as the corresponding computer code, have been optimized and updated several times (see e.g. Sahal-Bréchot 1974; Dimitrijevic & Sahal-Bréchot 1984a; Dimitrijevic 1996).
Within this formalism, the
full width of a neutral emitter isolated spectral line broadened by
electron impacts can be expressed in
terms of cross sections for elastic and inelastic processes as
The formulae for the ion-impact broadening parameters are analogous to the formulae for electron-impact broadening. We note that the fact that the colliding ions would impact in the far wings should be checked, even for stellar densities.
Model atmosphere calculations as well as calculations of the absorption coefficients were made with the local thermodynamical equilibrium (LTE) approximation. Model calculations were performed with the ATLAS9 code written by Kurucz (1993). The next step is the calculation of the outward flux at corresponding wavelengths points using the given model. For this we used the STARSP program written by Tsymbal (1996). In its current state this code includes the possibility of calculating a synthetic spectrum for an atmosphere with a vertical stratification of chemical elements.
The computational scheme is as follows. For each line we find the
central opacity as
The Doppler width is
Next, we compute the total damping parameter
In order to include Stark broadening effects we added the approximate
formulas (see Eqs. (13) and (14) in Sect. 5.1) in
The Stark width and shift are
Thus, at each point of a given spectral region (with resolution 0.01 Å for both lines)
we computed line absorption coefficient
The flux is given by the expression
Table 1: Stark broadening parameters for Sii spectral lines. This table shows electron-, proton-, and ionized helium-impact broadening parameters for Sii for a perturber density of 1014 cm-3 and temperatures from 2500 up to 50 000 K. The quantity C (given in Å cm-3), when divided by the corresponding full width at half maximum, gives an estimate for the maximum perturber density for which tabulated data may be used. The asterisk identifies cases for which the collision volume multiplied by the perturber density (the condition for validity of the impact approximation) lies beetwen 0.1 and 0.5. For higher densities, the isolated line approximation used in the calculations breaks down. FWHM(Å) denotes full line width at half maximum in Å, while SHIFT(Å) denotes line shift in Å.
Table 2: The parameters A, B and C of the approximate formulae for Stark widths and shifts.
The atomic energy levels needed for Stark broadening calculations were taken from Martin & Zalubas (1983) and Moore (1971), but LS determination of , , and 7s1P terms have been adopted according to Moore (1971) and therfore, in order to obtain a consistent set of data, energy levels from Moore (1971) have been used as the principal source. Oscillator strengths have been calculated by using the method of Bates & Damgaard (1949) and tables of Oertel & Shomo (1968). For higher levels, the method described in van Regemorter et al. (1979) has been applied.
The spectrum of neutral silicon is complex and not known well enough for a good calculation of the considered lines. First of all upper and lower energy levels for 6142.48 Å and 6155.13 Å lines are not known reliably. According to Striganov & Sventickij (1966) they belong to the and multiplets respectively, while Moore et al. (1966) stated that the lower level of the corresponding transitions is . It is also worth noting that in NIST (2002) the lower and upper levels of the corresponding transitions are not specified. Consequently, we adopted the identification of Striganov & Sventitskij (1966) as the only one enabling the corresponding calculations. Moreover, the level is in fact a mixture of 39% of 3pn and 56% of states, and 5% is unknown. Also the 3pn state is a mixture of states with . In calculations, we assumed that we have 30% of state, and our checks show that the difference in final results is negligible if we assume that this state is involved up to 20% only. An additional complication was that g levels, which might be very important for the perturbation of the considered 5f levels, are unknown. In accordance with the decrease of distance between and levels, we estimated that the distance from 5 or 5 term to the 5g levels should lie between 500 and 1500 cm-1. We checked results without 5g levels and with a fictive 5g level at 500, 1000 and 1500 cm-1 from the corresponding 5f terms. In all cases, line widths differed by less than 1%, while the shift varies within the limits of 3.5% for 1000 or 1500 cm-1. If the distance is 500 cm-1, the shift value e.g. at 10 000 K for 6142.48 Å line differs by about 20%. If we include a fictive 6g level 500 cm-1 distant from 5g levels, the difference is negligible. Hence, in order to obtain the needed set of atomic energy levels, we adopted a fictive 5g level 1000 cm-1 distant from the 5f levels and a fictive 6g level 500 cm-1 distant from the 5g level. Since the average estimated error of the semiclassical method is 30%, due to additional approximations and uncertainties, we estimate the error bars of our results to be 50%.
Our results for electron-, proton-, and ionized helium-impact line widths and shifts for the three considered Sii spectral lines for a perturber density of 1014 cm-3and temperatures T = 2500-50 000 K, are shown in Table 1. For perturber densities lower than those tabulated here, Stark broadening parameters vary linearly with perturber density. The nonlinear behaviour of Stark broadening parameters at higher densities is the consequence of the influence of Debye shielding and has been analyzed in detail in Dimitrijevic & Sahal-Bréchot (1984b).
In Table 1 the Stark widths and shifts for electron-, proton-
He II ion-impact broadening are presented as a function of temperature for a
of 1014 cm-3. After testing the density dependence
of Stark parameters, we have found that the width and shift are linear functions
of density for perturber densities smaller than 10
and can be scaled by the simple formula:
|Figure 1: The analytic fit of HeII-impact broadening data for Sii 6142 Å line: a) Stark width, and b) Stark shift. The shift and width are given for one perturber per cm3.|
|Open with DEXTER|
In order to simplify the use of Stark broadening data in the codes for stellar
spectral synthesis, we have found an analytical expression for Stark
Table 3: The atmospheric parameters and rotational velocities of the investigated stars.
The asymmetry and shift of Sii 6142.48 Å and (in particular) 6155.13 Å lines were observed, and it was obvious that the asymmetry is higher for the hotter stars. The most asymmetrical are 6142.48 Å and 6155.13 Å, the strongest lines of and multiplets, respectively. This line is slightly shifted and shows an asymmetry in the red wing even in a rather cool star like the Sun. It is not surprising if the Stark effect is responsible for the observed shifts and asymmetries. From Table 1 one can see that neither widths nor shifts have a strong temperature dependence, but they depend linearly on the perturber density. If a line is strong enough so that line core and line wings are formed at different atmospheric layers where electron density, differs significantly, then we should observe line asymmetry due to different shifts at different atmospheric layers even with a homogeneous vertical distribution of the absorbing element in stellar atmosphere (the case of the observed asymmetries in normal stars). Vertical abundance stratification makes this effect more pronounced, and it is observed in cool Ap stars, where an element stratification takes place (Savanov et al. 2001; Ryabchikova et al. 2002; Wade et al. 2001).
Table 3 summarizes the model atmosphere parameters and rotational velocities for the stars of our sample. These data are taken from Ryabchikova et al. (2000) and Bikmaev et al. (2002). For the Sun corresponding parameters were taken from Valenti & Piskunov (1996) except . Si abundances used in the present calculations are given in the last column.
In the Sun the most important broadening is due to collisions with neutral hydrogen and helium. In the hotter stars the importance of the Stark broadening increases. Barklem (priv. communic.) has provided us with the value of the broadening parameter due to collisions with atomic hydrogen for the Sii lines of our interest. This value per perturbing particle for a temperature 10 000 K, , is by 0.4 dex higher than the corresponding value obtained from the Unsöld (1955) approximation. According to Barklem his value may be overestimated by 0.1 dex which is proved by a comparison between the calculated and the observed blue wing of the solar Sii lines. The final value used in our calculations is . As for the shift due to collisions with the atomic hydrogen, a corresponding theory is not well developped, and the value of the shift is rather uncertain (Barklem, priv. communic.). The core and the wings of the strongest Sii 6155 line are formed in the layers with the temperatures 5000 and 5800 K respectively. In these layers the corresponding values for electron-, proton- and atomic hydrogen densities are cm-3, cm-3, cm-3 - line core, and cm-3, cm-3, cm-3 - line wings. The Stark width is 20 times smaller than at the depth of line core formation and 10 times smaller than at the depth of line wing formation. In the solar atmosphere, the Stark shift is 3 mÅ at the line core and about 20 mÅ for the line wings, which is not negligible. In the atmosphere of the hottest star of our sample, 10 Aql, the Stark width is of the order of at the depth of line core formation and exceeds by 5 times at the depth of line wing formation. Stark shifts are 10 and 100 mÅ, respectively. In 10 Aql the line core is formed at the layers with T=6500 K, cm-3, cm-3, cm-3 while line wings are formed at the layers with T=7500 K, cm-3, cm-3, cm-3.
|Figure 2: A comparison between the observed Sii 6155 Å line profile in the solar spectrum and synthetic spectra calculated with Stark widths and shifts from Table 1 (solid line) and with Stark widths calculated by the approximate formulae and without Stark shift taken into account (dotted line). X- and Y-coordinates are wavelenghths and surface fluxes (normalized to unity).|
|Open with DEXTER|
We first calculated Sii lines in the solar spectrum to check the Stark parameters. It was mentioned by Ryabchikova et al. (2002) that available theoretical oscillator strengths for the multiplets and are overestimated by about 0.5 dex compared to the multiplet ( 5948.55 Å). The authors fitted the lines of and multiplets using the solar spectrum. Corrected oscillator strengths - ( 6142.48 Å) and ( 6155.13 Å) were published in Bikmaev et al. (2002). The latter value was reduced to to get a better fit for the solar Sii 6155 line. We also reduced the oscillator strength ( ) of very weak Si line of the same multiplet, 6155.69 Å. We need to decrease both Stark widths and shifts for Sii 6142 and 6155 Å lines by 40% to get a reasonable fit of the synthetic line profiles to the solar ones. Due to relative weakness of the Sii 6142 Å line Stark effect is negligible but it is noticeable for the stronger line Sii 6155 Å. Figure 2 shows a comparison between the observed profile of the Sii 6155 Å line in the solar spectrum and synthetic profiles calculated with (solid line) and without (dashed line) taking into account the Stark width and in particular, shift. The strongest Sii 5948.55 Å line in our study is fully symmetric as expected taking into account an order of magnitude smaller values for both the Stark width and shift (see Table 1).
With corrected Stark parameters we calculated Sii line profiles in spectra of other stars.
Spectra of this normal late A-type star have the smallest spectral resolution of our sample. As it may be seen from Fig. 3 the fit of the synthetic line profile of 6155.13 Å line to the observed one is much better when it was calculated with both Stark width and shift. As in the solar case, the Stark effect is negligible for the weaker Sii 6142.48 Å line. To get a better fit of the line profiles we decrease the Si abundance by 0.13 dex compared to the results published by Bikmaev et al. (2002) which were based mainly on the equivalent width measurements.
This is one of the coolest Ap stars and is expected to have a chemically stratified
atmosphere. Results of the abundance determinations show that except for the rare-earth
elements (REE) abundance stratification is marginal if it exists at all (Ryabchikova et al. 2000).
Spectral synthesis of Sii lines seems to support this conclusion. We
could reproduce the asymmetric and shifted line profile of Sii 6155.13 Å
reasonably well, using the uniform distribution of Si and Stark
broadening parameters presented here. Figure 4 shows a comparison
between the observed and computed line profiles for two cases similar to
Fig. 3. The best fit of Sii line profiles was obtained with the
Si abundance as was derived by Ryabchikova et al. (2000).
is the hottest star in our sample. It has the most asymmetrical Sii 6155.13 Å line profile, which could not be reproduced by any combination of Stark parameters in a homogeneous atmosphere (Fig. 5). The even weaker line, Sii 6142.48 Å shows a noticeable line shift (Fig. 6). Ryabchikova et al. (2000) mentioned a possibility of Fe and the REE stratification in 10 Aql. Therefore we tried to find a simple distribution of Si in the atmosphere of 10 Aql by trial-and-error which would fit both Sii 6142.48 Å and 6155.13 Å lines. The simplest distribution shown in Fig. 7 gives a reasonable fit to the observed profiles of both Sii lines (Figs. 5 and 6). Moreover, the same Si distribution seems to fit much better the profiles of the strong Si II 6347, 6371 Å spectral lines compared to the calculations with the homogeneous Si abundance (-4.19) obtained by Ryabchikova et al. (2000). It is beyond the scope of the present paper to derive an exact Si abundance profile in the atmosphere of 10 Aql. We would like to stress here that with the present Stark parameters a sensitivity of 6155.13 Å line asymmetry to Si abundance changes in the stellar atmosphere can be successfully used in empirical studies of abundance stratification in the atmospheres of cool Ap stars.
|Figure 5: A comparison between the observed Sii 6155 Å line profile in the spectrum of Ap star 10 Aql (thick line) and synthetic spectra calculated with Stark widths and shifts from Table 1 and Si abundance stratification (thin line), with the same Stark parameters but for homogeneous Si distribution (dashed line), and with Stark width calculated by approximate formulae for the same stratification (dotted line). X- and Y-coordinates are wavelenghths and surface fluxes (normalized to unity).|
|Open with DEXTER|
|Figure 7: Si abundance distribution in the atmosphere of Ap star 10 Aql.|
|Open with DEXTER|
In order to discuss the contribution of the Stark broadening effect to the asymmetry and the shift of Sii 6142.48 Å and 6155.13 Å lines we have calculated Stark broadening parameters for these lines by using the semiclasical perturbation method. The obtained results are the first calculated data for the considered lines. In order to include Stark broadening data we have changed the STARSP code and we have synthesized the considered lines. From our analyzis we can conclude that:
|Figure 8: A comparison between the observed Si II 6371 Å line profile in the spectrum of Ap star 10 Aql (double line) and the synthetic spectrum calculated with Si abundance distribution shown in Fig. 7 (solid line) and Si homogeneous abundance (dotted line). X- and Y-coordinates are wavelenghths and surface fluxes (normalized to unity), respectively.|
|Open with DEXTER|
We are very grateful to Paul Barklem who provided us with the unpublished calculations of damping constants for Sii lines due to collisions with the atomic hydrogen. This work is a part of the projects "Influence of collisional processes on astrophysical plasma lineshapes'' and "Astrophysical Spectroscopy of Extragalactic Objects'' supported by the Ministry of Science, Technologies and Development of Serbia. The research was supported also by the Fonds zur Förderung der wissenschaftlichen Forschung P14984 and Osterreichische Nationalbank (Jubilaäumsfonds Nr. 7650). TR also thanks Russian Federal program "Astronomy'' and the RFBR (grant 00-15-96722) for partial funding.