6 The H$\alpha$ line behaviour

Figure 8: Spectral image for the H$\alpha$ order. All 32 spectra have been aligned on the velocity frame of the K2 star and normalized to the continuum. Orbital phase increases from 0 at the bottom to 1 in the middle and data are then repeated to have a better view of the variations with phase. The white rows correspond to phase intervals larger than 0.08 in which we have no spectra. Telluric H2O lines display as sinusoidal features. The broad H$\alpha$ extra-absorption feature around phase 0.3 and a blue-shifted emission component near phase 0.5 are apparent.

The H$\alpha$ line appears significantly variable in width, intensity and shape.

In Fig. 8 all the spectra of HR 7428 we have obtained in the H$\alpha$ region are shown, in orbital-phase order, as a trailed spectral image. In particular, from inspection of Fig. 8, variable broad absorption and emission components (red-shifted and blue-shifted, respectively) are apparent around the H$\alpha$ absorption line of the K2 II-III star.

6.1 The difference profile behaviour

In order to analyze the behaviour of the H$\alpha$ feature in more detail, we have used the technique of spectral subtraction. The standard low-active star $\alpha$ Ari, whose spectral type (K2 III) is very similar to that of the cool component of HR 7428, has been used to provide a "normal" H$\alpha$ template.

In the synthesis of the "inactive template'' we have neglected the contribution of the hot component because, according to the spectral model fit developed in Sect. 3.2, its weight to the total flux in the H$\alpha$ region is only 4%.

The difference H$\alpha$ profile, obtained with the spectral subtraction, therefore represents the contribution from the chromosphere of the K2 II-III star and from circumstellar material (see e.g. Richards 1992; Frasca et al. 2000).

Figure 9, left panel, displays a sample of observed spectra with superimposed the spectrum of $\alpha$ Ari artificially broadened to $v\sin i=17.2~{\rm km}\,{\rm s}^{-1}$ (De Medeiros & Mayor 1995). The difference of the observed profile with that of the reference star $\alpha$ Ari, right panel of Fig. 9, displays a complicated structured profile. The H$\alpha$ line core appears always filled-in with emission, and, in many cases, emission and/or absorption excess is present in the line wings. We can distinguish three typical situations:

a)

normal wings and filled-in core, with the difference showing only a narrow central emission;

b)

emission wings and filled-in core which lead to a difference profile with emission broad wings and a narrow central emission peak separated by absorption features on both sides;

c)

broad absorption wings and filled-in core, which produce an emission core inside broad absorption wings in the difference profile.

In particular, nearly all the spectra between $\phi=0\hbox{$.\!\!^{\scriptscriptstyle\rm p}$ }50$ and $\phi=0\hbox{$.\!\!^{\scriptscriptstyle\rm p}$ }70$ show emission wings, with the blue side generally brighter than the red one. The spectrum at phase 3 $\hbox{$.\!\!^{\scriptscriptstyle\rm p}$ }$50 shows excess emission on the red wing with a profile very similar to that observed by Xue-fu & Hui-song (1984) in 1983 at the same phase. Since our spectra taken in the three years at similar phases also show similar profiles, it clearly appears that this wing excess emission is related to the orbital phase. Also, the excess absorption in both wings appears to be related to the orbital phase. Starting at phase 0 $\hbox{$.\!\!^{\scriptscriptstyle\rm p}$ }$2, a broad absorption wing progressively develops and reaches maximum strength near phase 0 $\hbox{$.\!\!^{\scriptscriptstyle\rm p}$ }$30-0 $\hbox{$.\!\!^{\scriptscriptstyle\rm p}$ }$32. This effect almost completely disappears at phase 0 $\hbox{$.\!\!^{\scriptscriptstyle\rm p}$ }$42, when the blue wing has a normal shape and the red wing still displays some absorption.

Figure 9: A sample of spectra of HR 7428 (thick lines) compared to the rotationally broadened spectra of $\alpha$ Ari (thin lines).

We have taken a great care in the reduction process to avoid bias in the normalization procedure. The blazing function in the spectra has been accurately cancelled by the flat-field division, and the normalization has been made by selecting the continuum peaks outside the H$\alpha$ wings (10-15 Å far away from the line center) and by interpolating the continuum-fitting polynomial function in the H$\alpha$ line region. Therefore we conclude that the observed excess emission and absorption are intrinsic to the system.

6.2 The integrated equivalent width

Such excess emission and/or absorption is not easily interpreted in terms of chromospheric activity. Broad H$\alpha$ emission wings are sometimes observed in RS CVn binaries. The phenomenon is always associated with transient events like strong flares (Catalano & Frasca 1994; Foing et al. 1994) or intense active regions, but also the line core fills in with emission and the overall profile appears to be composed of a narrow bright emission component and a broad fainter one (Montes et al. 1997, 2000; Lanzafame et al. 2000).

Although this H$\alpha$ behaviour is difficult to ascribe to chromospheric inhomogeneities and may arise from circumstellar matter, as we will propose below, a chromospheric emission component should be present in the H$\alpha$ profile, as suggested by the Ca  II emission shown in Fig. 7 (see also Fernández-Figueroa et al. 1994), and by the Mg  II h and k emission. We have re-determined, from archive IUE spectra, a Mg  II flux $F_{\rm Mg~II}=2.38\times10^6\,{\rm erg\,s^{-1}\,cm^{-2}}$ at the stellar surface, and then, according to the average correlation between H$\alpha$ and Mg  II emission established by Frasca & Catalano (1994), we estimate an expected net H$\alpha$ equivalent width of 0.2 Å.

Figure 10: Equivalent width of net H$\alpha$.

We have determined, from the difference spectra, the excess H$\alpha$ equivalent width (EW $_{\rm H\alpha}$) by integrating the difference profile including the extended wings. Values are displayed in Fig. 10 including error bars determined as the product of the integration range and the (S/N)^-1 ratio, evaluated in two line-free windows selected on the residual spectrum at the two sides of the H$\alpha$ line.

These measurements show a large spread, significantly greater than the estimated errors, with some intrinsic phase-dependent trend. In particular, the steady decrease from positive to negative values between phase 0 $\hbox{$.\!\!^{\scriptscriptstyle\rm p}$ }$2 and 0 $\hbox{$.\!\!^{\scriptscriptstyle\rm p}$ }$4 clearly reflects the development of the extended absorption wings. Fluctuation of the excess emission (positive values) reflects changes with phase and from cycle to cycle.

The average net H$\alpha$ ${\it EW} \simeq 0.3~$Å measured between phase 0 $\hbox{$.\!\!^{\scriptscriptstyle\rm p}$ }$7 and 0 $\hbox{$.\!\!^{\scriptscriptstyle\rm p}$ }$2, when no excess emission or absorption is present, fairly well agrees with the expected chromospheric emission estimated from the Mg  II emission flux. Emission EW values significantly larger than $\simeq$0.3 Å are found between phase 0 $\hbox{$.\!\!^{\scriptscriptstyle\rm p}$ }$5 and 0 $\hbox{$.\!\!^{\scriptscriptstyle\rm p}$ }$8 when the extended emission wings are present, thus we conclude that the real chromospheric H$\alpha$ emission is at level of about $\simeq$0.3 Å. No emission modulation, ascribable to the rotation effect of inhomogeneous distribution of emission, can be distinguished.

6.3 The wing emission and absorption radial velocity

The emission wing profile, even if of lower intensity, is quite similar to that observed in shell stars (Marlborough & Cowley 1974) or in mass-losing red giants (Cacciari & Freeman 1983) as well as in the spectrum of the supergiant component of the binary system $\epsilon$ Aurigae, which is supposed to have a ring-like structure of moving gaseous clouds. On the other hand, the extended absorption wing profile seen between phase 0 $\hbox{$.\!\!^{\scriptscriptstyle\rm p}$ }$2 and 0 $\hbox{$.\!\!^{\scriptscriptstyle\rm p}$ }$3 closely resembles the extra absorption present in the H$\beta$ line of the T Tau star SU Aur (Petrov et al. 1996), explained as the effect of simultaneous outflows and inflows of matter.

These similarities do suggest that some flow of matter can be responsible of the excess emission and absorption in HR 7428. Here we propose that a single cloud localized in the region between phase 0 $\hbox{$.\!\!^{\scriptscriptstyle\rm p}$ }$0 and 0 $\hbox{$.\!\!^{\scriptscriptstyle\rm p}$ }$5 can produce both the extra emission and extra absorption in the H$\alpha$ wings. This cloud is mainly seen projected against the K star disk at phase 0 $\hbox{$.\!\!^{\scriptscriptstyle\rm p}$ }$2-0 $\hbox{$.\!\!^{\scriptscriptstyle\rm p}$ }$3 and produces the extra absorption in the H$\alpha$ wing spectrum of that star. Between phase 0 $\hbox{$.\!\!^{\scriptscriptstyle\rm p}$ }$5 and 0 $\hbox{$.\!\!^{\scriptscriptstyle\rm p}$ }$75 we see the same cloud illuminated and excited by the A2 star and projected against the sky, so its contribution is an extra-emission, due to radiative recombinations and de-excitations.

 

 

Table 5: Velocity shift of the broad extra-absorption and wing-emission centroid with respect the K star velocity, and the full width half maximum of the feature.

HJD	Phase	Feature	Vel. Shift	Vel. disp.
- 2440000			kms-1	kms-1
1385.4111	0.2154	Abs.	11.2	49.2
1386.3447	0.2240	Abs.	12.6	54.5
1387.3975	0.2337	Abs.	4.6	50.8
1387.4258	0.2339	Abs.	4.3	55.2
1388.3750	0.2427	Abs.	1.0	59.1
631.5586	0.2724	Abs.	20.8	89.6
632.5635	0.2817	Abs.	28.1	125.3
633.5440	0.2907	Abs.	31.9	115.9
634.5732	0.3002	Abs.	43.6	136.6
636.5156	0.3181	Abs.	12.4	99.4
1407.6045	0.4198	Abs.	18.7	51.3
982.5674	0.5052	Em.	15.0	108.1
1417.5703	0.5115	Em.	-41.7	133.2
1423.4961	0.5661	Em.	-23.5	174.0
1426.4814	0.5936	Em.	-31.5	146.1
1427.4102	0.6022	Em.	-44.6	133.8
1000.5713	0.6710	Em.	-16.0	87.9
1002.5664	0.6894	Em.	-16.8	89.8
678.4854	0.7046	Em.	-11.3	144.5
679.4756	0.7137	Em.	-5.4	101.6
680.4775	0.7229	Em.	3.1	106.7
681.5195	0.7325	Em.	-2.0	104.7

Both the broad extra-emission and absorption components exhibit asymmetric profiles compared to the chromospheric excess emission line of the K2 II-III star, indicating a relative global motion of the cloud with respect to the star. To quantify this motion we have estimated the central wavelength positions of the extra emission/absorption by fitting Gaussian profiles to the broad components and to the sharp central emission in the difference spectrum (see Fig. 11 for the description). Measured velocity shifts of the broad component are reported in Table 5 together with the velocity dispersion derived from the FWHM of the feature. From this table one sees the red-shifted behaviour of the extra-absorption feature, that reaches its maximum shift (44 km$\,$s^-1), intensity and width at phase 0 $\hbox{$.\!\!^{\scriptscriptstyle\rm p}$ }$30. The broad emission component is instead mainly blue-shifted with respect to the K2 II-III star with a maximum of about 45 km$\,$s^-1 at phase $\simeq$0 $.\!\!^{\scriptscriptstyle\rm p}$6.

Figure 11: Upper panel: simulation of the difference spectrum at phase $\phi = 0.30$ with the sum of two Gaussians, one for the broad extra-absorption ( dashed line) and the other for the sharp core emission ( dotted line). The thin line is the sum of the two fitting functions. Short vertical lines show the central positions for the two fitting functions. The red-shift of the extra-absorption component is evident. Although the red absorption wing is rather well fitted, we note that this simple function is not able to simulate the blue absorption wing. This indicates an asymmetric shape of the extra-absorption feature. Lower panel: example of fitting of Gaussians to the broad and narrow component for a spectrum showing excess wing emission.

The average FWHM of the broad components ranges from 50 to 140  ${\rm km}\,{\rm s}^{-1}$for the extra-absorption, and from about 90 to 150  ${\rm km}\,{\rm s}^{-1}$ for the wing emission. This strengthens our hypothesis of a unique structure as being responsible for both the extra absorption and the wing-emission, with turbulent velocity of the order of 100-150  ${\rm km}\,{\rm s}^{-1}$.