3 Analysis

3.1 Spectral variations

In the spectra of Hen 1761 we confirm the presence of emission lines of HI, HeI, HeII, CIII, NIII, [OIII] and FeII and detect OI, MgII, TiII, [FeII], [SII], [NeIII], [FeIV], [FeVI], [FeVII] and [CaV]. Some members of the Paschen Series are also present as very weak emission lines. A Balmer jump in emission is observed in the images of October 1994.

  

Figure 1: Low resolution optical spectra for Hen 1761. Spectral regions a) between $\lambda\lambda$ 3750 - 5200 Å and b) $\lambda\lambda$ 6200 - 7200 Å. Numerous TiO molecular bands are present in the red region. The Raman scattered OVI emission lines at $\lambda$ 6825 Å and $\lambda$ 7082 Å are absent. Fluxes are indicated in unit of 10-13 erg cm-2 s-1 Å-1

The Figs. 1a and b show two spectral regions between $\lambda\lambda$ 3750 - 5170 Å and $\lambda\lambda$ 6200 - 7200 Å with the fluxes indicated in units of 10^-13 erg cm^-2 s^-1 Å^-1. As seen in Fig. 1a, numerous TiO molecular bands are present in the red region and the Raman scattered OVI emission lines at $\lambda$ 6825 Å and $\lambda$ 7082 Å (Schmid 1989) are not detected in Hen 1761. The flux ratios of the most important emission lines for the different observing runs, scaled with respect to H$\beta$ = 100, are given in Table 2. The estimated error in the determination of the integrated emission line fluxes is in the order of $\pm10$% for strong emission lines and $\pm25$% for weaker lines. The fluxes are not corrected for reddening.

  

Table 2: Line fluxes in the spectrum of Hen 1761 scaled to H$\beta$ = 100

Notes:
- : the line is either outside of the observed spectral range or the image does not include H$\beta$.
nd: undetected line
w: very weak line.
bl: blended line.
(*): The mean absolute fluxes in units of 10^-12 erg cm^-2 s^-1.

  

Figure 2: a) Intensity ratio variations of HeI, HeII, NIII and [OIII] observed in Hen 1761. b) "color index" (m4895 - m6975). c) depth of TiO absorption bands at $\lambda$ 4954 Å and $\lambda$ 6159 Å. d) and e) Radial velocities obtained from different ions. Radial velocity values published by van Winckel et al. (1993), were included in the plots between HJD 2447360 and 2447790

  

Figure 3: The spectrograms show intensity variations in the lines of HeI, FeII and TiII. a) In August 1995, FeII is very weak and the spectrum develops forbidden lines of high ionization such as [FeVI]. b) In August 1990, there is a greater number of absorptions in the blue region, particularly TiII absorption lines, which disappear in April 1991 or are visible as weak emissions

  

Figure 4: Variations in the shape of the H$\beta$profile along the time. The double-peaked profile observed in November 1990, becomes a single asymmetric emission. Later, in 1995, the H$\beta$ profile shows an incipient central absorption

We can see in Table 2 that during 1990-95 Hen 1761 shows variations, in the line intensities of HeI, HeII, NIII and [OIII] in the same sense, reaching in general their minimal values in April 1991, and then increasing until August 1995. HeII $\lambda$4686 Å and the forbidden line [OIII] $\lambda$5007 Å display the most significant intensity variations, but the HeI and NIII intensities change with a much lower amplitude (see Fig. 2a). In contrast, the FeII line emissions show a systematic decrease of intensities until 1995. The Fig. 3a shows a spectral region where the variations in the HeI, [OIII] and FeII lines are displayed.

Van Winckel et al. (1993) already found dramatic changes in the [OIII] $\lambda$5007 Å and FeII $\lambda$5018 Å lines in time scales of several months. In our spectra H$\alpha$ is slightly asymmetric on the blue side and a weak central absorption is more noticeable in August 1995. Otherwise, the shape of the H$\beta$ profile undergo important changes along the time (see Fig. 4). In August 1990 the profile has a symmetrical appearance but a double-peaked profile can be observed in the image of November 1990, being the radial velocity of the central absorption of 49 km s^-1. H$\beta$ becomes a single emission with an asymmetry on the blue wing, few months later, in April 1991. An incipient central absorption is observed in August 1995.

In the blue region of the spectra the TiII lines are in absorption in August 1990, then they practically disappear in November 1990, and in April 1991, are visible as weak emissions. Later, up to the last observations, these lines are present as absorptions. Figure 3b shows the dramatic variations when the absorptions are filled by emissions.

In 1994 double peak profiles of [FeVI] $\lambda$4967 Å $\lambda$4972 Å and $\lambda$5176 Å as well as [CaV] $\lambda$5309 Å appear in the spectrum. Then, in 1995 these lines become stronger and other forbidden lines of higher ionization, like [FeVII] $\lambda$5721 Å and $\lambda$6087 Å can be detected.

In order to establish the behavior of the continuum as function of a time, we have chosen to measure fluxes in 50 Å bands centered at 4895 and 6975 Å since these spectral ranges are free of important emission lines. The respective fluxes have been converted into a color index CI = m₄₈₉₅ - m₆₉₇₅ = - 2.5 log F₄₈₉₅/ F₆₉₇₅ listed in Table 4. In Fig. 2b we can see that the system is much bluer during August-November 1990 and later, in July 1992 the red continuum becomes stronger.

From 1990 to 1992 the spectra are dominated by many narrow absorption lines which were identified more frequently as FeI and TiI. Moreover, other absorption lines of neutral and once ionized metals, e.g. CaI $\lambda$4227 Å NiI, CrI, NaI, OI, VI, CaII and NiII are also present.

3.2 Radial velocities

A complete summary of the heliocentric radial velocities of Hen 1761 are given in Table 3. The individual emission and absorption lines have been measured by fitting a Gauss function to the profiles. The average values of the radial velocities, the corresponding mean errors and the measured line number within brackets are indicated for differents epochs.

  

Table 3: Heliocentric radial velocities of the absorption and emission lines in Hen 1761 in $\rm{km~s^{-1}}$

Notes:
(S)= singlet; (T)= triplet; (D)= doublet; E= emission line; A= absorption line
(n)= number of measured lines
: dubious value.

The radial velocities have also a variable behavior with time. The Fig. 2d shows the radial velocity curves corresponding to the high ionization potential ions, HeII and [OIII], while the Fig. 2e shows those of the low ionization potential ions H, HeI and FeII separetely. The largest amplitudes of the radial velocity curves and a systematic decrease from August 1990 to April 1991 is observed in the first figure.

Radial velocities of several Balmer lines, H$\alpha$ to H9, can be measured in two epochs, 1992 and 1994, except for H8 since it is affected by blending with the HeI $\lambda$3888 Å. A regression of the Balmer series radial velocities increasing towards the first members of the series may be an evidence that the lines are emitted in an expanding region with a finite optical thickness in these lines.

The singlet and triplet series of the HeI lines are shown separately in Table 3. Their radial velocities show differences which are more noticeable in the Echelle spectra, 17 km s^-1 in October 1994 and 14 km s^-1 in August 1995, the triplet emission lines being redshifted with respect to the singlet ones.

In the case of [FeVI], three double peaked lines $\lambda$4967 Å, $\lambda$4972 Å and $\lambda$5176 Å are measured in August 1995. The mean radial velocities of each peak are $68\pm8~$km s^-1 and $109\pm4$ km s^-1 respectively.

3.3 Classification of the cool component

The spectral classification of the cool component was based on the TiO optical molecular-band strengths at $\lambda\lambda$ 6180 and 7100 Å as proposed by Kenyon & Fernández-Castro (1987). We adopted the calibration of the [TiO] indices with the spectral types of M giants given by these authors and both indices, [TiO]₁ and [TiO]₂, were calculated with the observations of August 1992 and October 1994. The spectral type obtained from the first index [ST]₁, is earlier than that given by the second one [ST]₂, indicating that the first index is more affected by the contribution of the hot component. In the case of the 1992-1995 observations, the fluxes were measured separately in each spectra to obtain, for each epoch, the mean indices with their respective internal mean errors.

In Table 4 we can see that the spectral type of the cool star gets increasingly later during the period of our observations. This behavior could be interpreted either as a heating suffered by the giant during the first observations or as the result of variations in the blue continuum. Figure 2c shows variations of the depth of TiO absorption bands in connection with the local continuum level. Two strong absorption bands at $\lambda$4954 and $\lambda$6159 have been measured (see Table 4), and we notice that the relative depth of each band increases at the time the spectral type of the giant becomes later, but the amplitude of these variations seems to be more important toward the shorter wavelengths. Besides, if we consider the behavior of the color index CI = (m₄₈₉₅ - m₆₉₇₅) (Fig. 2b) we may conclude that the observed changes in the spectral type of the giant are the result of a dilution effect of an additional blue continuum, coming from the hot source and veiling the M spectrum. The influence of the blue continuum was decreasing in time, as it is shown in the Figs. 2b and c. Consequently we might consider that the spectral type of the cool component in Hen 1761 is as later as M5 III and according to the published near-infrared data from 1972 (Glass & Webster 1973) and 1990 (Munari et al. 1992), no change is registered in the IR colors, specially in the K and L bands.

  

Table 4: Spectral type of the cool component, color index CI = (m₄₈₉₅ - m₆₉₇₅) and TiO relative depths

3.4 Reddening and distance

Before deriving the physical parameters of Hen 1761, it is important to determine the interstellar extinction and distance of the system. We present several ways to determine the reddening, adopting in all the cases, the interstellar extinction curve given by Savage & Mathis (1997).

The Balmer decrement provides a good reddening diagnostic considering that the optical HI recombination lines are formed under case B conditions. But this is not the case in the majority of the S-type symbiotic stars, which show significant departures from case B in the lower Balmer series members, due to self-absorption effects. Netzer (1975) has computed the self-absorption effects on hydrogen line intensities in dense nebulae, as functions of Ly-$\alpha$ and H$\alpha$ optical depths ($\tau_{\alpha}$, $\tau_{\rm H\alpha}$). By using his results and comparing them with the mean values of our intensity ratios observed in August 1992, October 1994 and August 1995, $I({\rm H}\alpha)/I({\rm H}\beta)=5.4\pm1.0$; $I({\rm H}\gamma)/I({\rm H}\beta)=0.37\pm0.05$; $I({\rm H}\delta)/I({\rm H}\beta)=0.22\pm0.07$; we can derive $E(B-V)=0.26\pm0.04$, corresponding to $\tau_{\rm H\alpha}\sim20$ and $\tau_{\alpha}$ = 10⁴, and densities of $n_{\rm e}$= 10⁹ cm^-3. The upper Balmer series members have nearly constant intensity ratios in August 1992 and October 1994, with mean values $I({\rm H}\epsilon)/I({\rm H}\beta)=0.13$; $I({\rm H}9)/I({\rm H}\beta)=0.07\pm0.03$ and $I({\rm H}10)/I({\rm H}\beta)=0.05\pm0.02$.The lines showing severe blending were not considered, such as H$\epsilon$ on the lower resolution spectrum of August 1992, which was blended with [NeIII] and H8 with HeI $\lambda$3888 Å. In this case a value of $E(B-V)=0.21\pm0.12$ is obtained by comparison with theoretical values of case B-recombination taken from Brocklehurst (1971) for $T_{\rm e}=10\,000$ K and $n_{\rm e}$= 10⁶ cm^-3.

Since the near infrared colors of the M star have remained virtually constant in Hen 1761, reddening migh also be obtained from the JK photometry and the spectral type of the cool component. By combining the observed (J-K) colour index (Munari et al. 1992), with intrinsic colors given by Bessell & Brett (1988), and considering that the cool component is a normal M5 giant, we obtain E(B-V) = 0.20. Then a distance of $d\sim 2.2$ kpc is obtained by means of the absolute magnitude M_K= - 6.26 calculated from the $M_{V}\sim - 0.3$ (Schmidt-Kaler 1982) and A_K = 0.08.

Whitelock & Munari (1992) found that many of S-type symbiotics have IR characteristics very similar to those of nonsymbiotics M-giants of the Galactic bulge. Tacking into acount this considerations, a color excess E(B-V) = 0.73 was obtained by comparing the observed (J-K) colour index with the intrinsic colors given by Frogel & Whitford (1987) for a bulge M5-giant. Considering the distance modulus to the galactic center, 14.43 mag (Tiede et al. 1995) and the K magnitude given by Frogel & Whitford (1987), we estimate M_K = -5.2 and A_K = 0.28. A distance of $d\sim 1.3$ kpc is obtained in this case. The reddening resulting when we consider the giant of the Hen 1761 as belonging to the galactic bulge, E(B-V) = 0.73, conduces to a high reddening, A_V = 2.3 mag. Consequently the intrinsic colors (H-K) would be correlated with earlier spectral type $\sim$ M0 - M2 (see Figs. 7a and 7b Whitelock & Munari 1992), and this is inconsistent with our considerations about the spectral type of the cool star. While the reddening estimated for the HI line emissions is very similar than that derived by comparing the J-K colors observed in Hen 1761 with those for normal giants. This means that the cool component is as reddened as the nebular region of the system. Circumstellar dust around the giant is not important, as usual in S-type symbiotic systems. In this sense, the IRAS data for S-type symbiotic stars (Kenyon et al. 1988), indicated that Hen 1761 has flux ratios, [25/12] $\sim 0.2$ and [60/25] $\sim 1.0$, similar to those of normal M giant stars.

In the next sections we will correct the line fluxes for interstellar reddening adopting the values: E(B-V)=0.20 and A_V=0.6.

3.5 Hot component

The temperature of the hot component has been estimated from the intensity ratios of the HI, HeI and HeII recombination lines under certain assumptions (Kenyon (1986) and references therein). We give in Table 5, the temperatures of the hot component $T_{\rm h}$ for each epoch of observation, by using Iijima`s method and the de-reddened flux ratios of (HeII$\lambda$4686)/H$\beta$ and (HeI$\lambda$4471)/H$\beta$.

  

Table 5: Physical parameters of the hot component

The $T_{\rm h}$ in turn leads to a crude estimation of the blackbody luminosity and radius of the hot star from HI, HeI and HeII recombination line fluxes assuming theoretical emissivities of case B (Harman & Seaton 1966), adopting $T_{\rm e}=10\,000$ K and our estimation of distance $d\sim 2.2$ kpc.

3.6 Nebulae

We are constrained to estimate physical conditions in the nebulae of Hen 1761 with the emission lines observed in the optical region. The most popular line ratio that has been used for the determination of $T_{\rm e}$ and $n_{\rm e}$ is [OIII] [I(4959)+I(5007)]/I(4363). When it was possible to measure the three [OIII] emission lines in our spectra, the dereddened intensity ratio increased slightly from 2.2 in August 1992 to 2.5 in October 1994 and to 2.9 in August 1995, which corresponds to $T_{\rm e}=10\,000$ K, using the Kafatos & Lynch (1980) diagnostic diagram and the corresponding densities between $n_{\rm e} = 6\,10^{7} - 3\,10^{7}\,{\rm cm}^{-3}$ have been calculated on the basis of Seaton's (1975) relation.

In a large sample of galactic symbiotic stars, Proga et al. (1994) found that, besides distinguishing between S and D type symbiotics, the HeI $\lambda6678$/$\lambda5876$ ratio provides an useful tool for approximating the physical conditions within the nebulae. The $I(\lambda6678)/I(\lambda5876)$ ratio in Hen 1761 is 0.5 which corresponds to a S-type symbiotic. Some comments on the position of our object in the $I(\lambda6678)/I(\lambda5876)$ vs. $I(\lambda7065)/I(\lambda5876)$ diagram may be worthwhile. In Fig. 5 we can see that the positions of the points fall near Proga et al.'s models for $n_{\rm e}\sim 10^{10}\,{\rm cm}^{-3}$ and temperature $10\,000$ K. The figure shows also that the line ratios appear to be variable, increasing with decreasing $n_{\rm e}$ since 1990 to 1995. Proga et al. found that the $I(\lambda6678)/I(\lambda5876)$ ratio variations are, in most systems, correlated with V, in the sense that lower intensity ratios seem to be associated with optical maxima, either during an eruption when the system changes from quiescence to an outburst stage (i.e. RS Oph) or as a function of the photometric phases along an orbital motion (i.e. V443 Her, AG Peg). It is not clear at the moment which is the case of Hen 1761, and it would be necessary to ascertain the origin of the observed variations first.

  

Figure 5: The HeI intensity ratios measured on logarithmic scale in Hen 1761. A comparison with theoretical models taken from Proga et al. is shown for $n_{\rm e} = 10^{12}\,{\rm cm}^{-3}$ (dot-dashed lines), $n_{\rm e} = 10^{10}\,{\rm cm}^{-3}$ (dashed lines), $n_{\rm e} = 10^{8}\,{\rm cm}^{-3}$ (solid lines), $T_{\rm e}$ of 10000 K (light lines) and 20000 K (heavy lines). The observed intensity ratios in Hen 1761 are indicated with symbols for different epochs: 1990 (solid circle), 1992 (open triangle), 1994 (open circle) and 1995 (solid triangle)

The [FeVII] lines at $\lambda\lambda$ 4942, 4989, 5159, 5721 and 6087 Å, are the higher ionization forbidden lines found in Hen 1761 in the spectra of 1995. Nussbaumer & Storey (1982) give relative intensities of [FeVII] lines as a function of the electron density and electron temperature in the region where these lines are formed. Our dereddened values of the optical line ratios 5159/6087 and 4942/6087 are 0.63 and 0.20 respectively. The first one is larger than the maximum theoretical values presented by Nussbaumer & Storey but the second line ratio suggests that the [FeVII] lines are emitted in a region where the electron temperature is very high, K and the density is in the order of $n_{\rm e}=10^{7}\,{\rm cm}^{-3}$.

Some considerations about the observed variations in the physical conditions in the nebulae of Hen 1761 will be made in the next sections.