next previous
Up: The peculiar B[e]

6. Discussion of the variations

6.1. The photometric variations

6.1.1 Long time scale variations/obscuration

Let us consider the decrease in brightness of HD 45677 from V = 7tex2html_wrap3407 2 to V = 8tex2html_wrap3409 8 as being due to long-term effects (the brighter data-points of the minimum V = 8tex2html_wrap3411 8 may be due to shorter time scale variations, which will be discussed in the next two subsections). If we think of this long-term variation as being caused by some obscuration effect, one would need to cover tex2html_wrap_inline337977% of the stellar surface. As was discussed for the pre-main sequence star UX Ori (Bibo & Thé 1991), these amounts are difficult to realize by intrinsic stellar activities such as star spots, but a slow recovery would explain the relatively slow increase in brightness.

The large changes in V at the long time scale, decennial, and the form of the light curve are also not easily understood by means of obscuration effects like eclipsing by companion(s). The large tex2html_wrap_inline3383 would indicate a nearly total eclipse. Because the complete minimum itself already takes several decennia, the period would be extremely long, indicating a very wide binary or one with a very slow moving eclipsing object. The orbital period of the companion can be much shorter if one explains the long duration of the minimum as one at apastron in a very eccentric orbit, but this can not explain the asymmetric minimum. Another argument against an eclipsing companion is the absence of any periodicity in the radial velocity variations of HD 45677 (Swings & Allen 1971). However, these amounts of obscuration could be realized by the presence of large (> 1 tex2html_wrap_inline3387m), presumably circumstellar, dust particles in our line of sight, which will not redden the light of the star, but simply obscure the stellar radiation from the UV to the near infrared.

The colour changes on the long time scale are quite remarkable. The U-B and B-V are getting redder throughout the available data, while the opposite colour changes are seen for V-R and V-I during the period of increasing brightness. The complete photometric behaviour can thus not be explained as a simple obscuration effect, as hypothetically discussed above: In that case the getting bluer would be followed by a getting redder again when the star recovers in brightness. In case of pure extra extinction one expects all the colours to get redder. However, if we have a distribution consisting of two distinct grain sizes, as proposed before by SchulteLadbeck et al. (1993), of small grains (tex2html_wrap_inline3397 0.5 tex2html_wrap_inline3399m) located in the polar lobs and of large grains located in the disk, an increase of both grain populations can produce the observed colour effect. In the polar lobs scattered light will increase and being more reddened by the extra extinction, whereas in the optically thick disk extinction due to larger grains is dominant more efficiently to the longer wavelengths. Such a mechanism effects the visual radiation the less and can explain the colour behaviour. The production of grey grains shortly after a blow-out or evaporation of a cometary body could account for the large obscuration as measured in all pass-bands and the possible getting redder of the V-R and mainly the V-I colours due to extra extinction before the minimum if this was the case.

The difficulty with this scenario is the steady continuation of the UBV colour effects. To account for the required continuous production of small grains out of the larger ones, we need a constant production mechanism to do so, such as a stellar wind, stellar radiation field or mass accretion or a combination of these processes. (a) If accretion is the process, the destruction of the large grains will increase the brightness, small grains will be produced in the polar lobs. Accretion could produce a more compact disk, also explaining the decrease of total infrared flux (Sitko et al. 1994), and therefore getting optically thicker accounting for the extra extinction in the red. In such a model the production of grey grains is then followed by accretion, contributing to the destruction of the large grains mostly close by the central star. (b) Destruction of the grey grains by the stellar wind and the stellar radiation field will create the small grains in outer regions. Scattering at these grains will be added by the scattering in and near a bipolar flow. Radiation and stellar wind pressure could produce a more compressed ``ring'' of the blown out material to produce the extra extinction. However, when getting thinner reduced extinction due to its expansion must be compensated in this case. The brightening of HD 45677 in this case would be due to evaporation of dust by the stellar radiation and a stellar wind.

Additionally to (a) and (b), if the grey grains were produced by a blow-out around 1950, this material will continue to expand, becoming less efficient in the obscuration. We must note here that the obscuration due to grey grains supports the hypothesis by Sitko et al. (1994). Support for our suggestion that the small grains can also be (continuously) generated by accretion, which is then also seen by an energetic bipolar flow, is given by Grady et al. (1993).

Another model to produce changes in the grain size distribution is to introduce a magnetic field. After the blow-out, the magnetic field recovers in strength, causing a more efficient alignment of the dust particles in time. For this effect one spheroid grain population is needed. Then the colour effects will continue until the original magnetic field is recovered. The brightness variations can be explained in the same way as above. However, this model does not need a bipolar flow of which the existence is still hypothetical.

6.1.2 Short time scale variations/flickering

In the paper of Pérez et al. (1993) a strong near-UV excess was reported. If this relatively strong near-UV excess is explained by scattering, as mentioned before in the two grain population model, it will be stable throughout the detected short time scale variations. However, variable extra extinction, due to the production of small grains by accretion, on the scattered and the blue stellar light will effect the near-UV colour more than the blue. If the grains are small enough they will redden the blue colours most efficient, which will result in the colour behaviour as detected. A support for the production of small grains is that, if we explain the photometric short time scale variations as due to variable extinction, their reddening slope indicates tex2html_wrap_inline3413 values tex2html_wrap_inline3415 3.1.

The shortest time scale variations are then seen as brightening effects in the visual by: re-radiation from the extinction in the blue and near-UV and by the extra accretion luminosity. The resulting ``flickering'' acts on a time scale of days, as resolved by our new Strömgren 1993 dataset (E) as shown in Fig. 3 (click here).

In plots E1-3 of Fig. 3 (click here) only a part of one ``pulse-like'' variation, with a daily trend of diminishing brightness, during the three following nights is seen. Superimposed on this, small brightenings are seen on a time scale of several minutes. As both kinds of ``flickering'' show the same trends in the colour-magnitude diagrams, see plot E6 of Fig. 3 (click here), they are probably of the same origin and are not due to scatter in the data as their errors are of about 0tex2html_wrap3423 008 (de Winter et al. 1996) and even less when taken during one night, as is the case. In this case the short time scale ``flickering'' of minutes are instabilities in the ``pulse-like'' variations of days. However, here all the colour variations are orientated towards the red at increasing light with a large scatter in the u-v colour, which becomes visible as the points are chronological connected to each other in these plots by lines. We explain these shortest time scale variations as being brightenings due to accretion which will be reflected by grains, being more effective in the less bluer passbands of the observed colours. In this period, 1993, either no extra extinction due to produced small grains seems to be present, which explained the reddening of some colours of the other short time scale variations, as before and during the minimum at 1981 or the monitoring period was to short to detect this effect.

The amplitude of the short time scale variations within the different datasets does not seem to vary much, this is best seen in Fig. 1 (click here)b. The flickering seems therefore present before and after the maximum obscuration around 1981.

We must note that due to inhomogeneities in the circumstellar material there may be moments during which we can look to regions closer to the star which can produce similar effects as mentioned above. In this case variations within a few minutes of 0tex2html_wrap3425 015, in the 1993 Strömgren data, is equivalent with about 1 to 2% less obscuration of the surface of the central regions. But also in this case these variations are brightening effects.

6.1.3 Intermediate time scale variations/semi periodic

Although colour changes of the intermediate time scale variations could only be studied in B-V, we can fit them well with the other variations seen when we explain them as obscuration effects by concentrations of dust. The higher extinction then causes the extra reddening with decreasing brightness. If the average grain-size is large this effect will not be significant. A support for this hypothesis is the large tex2html_wrap_inline3429-value of the reddening slope in the V versus B-V diagram of about 7.0-7.5.

Rotation of such dust clouds around the central star could cause the obscurations, extra reddening and recovering in brightness as seen in the lightcurve of HD 45677, Fig. 1 (click here)b. However, the decrease of brightness in 1989 and 1992 seems to occur slowly compared to the getting brighter in 1991. This can be explained as due to the superpositions of the brightening in the long time scale variations. Another explanation for the intermediate time scale variations is the production of large grains at certain moments. The getting brighter is then due to the subsequent destruction of such grains or the rotating away of these regions from our line of sight.

As we have seen that the short time scale variations are brightening effects, the same could be true for the intermediate time scale variations. In that case we have maxima during 1988 and 1992. Just before these maxima we see in Fig. 1 (click here)b that the ``flickering'' amplitude is relatively high, followed by a decrease during the maxima and after that getting larger again. In this picture there seems to be a connection between the short time scale accretion and the intermediate time scale variations. The latter can be explained by infall of circumstellar material which lasts 1-2 years and which produces minima. Then the dense obscuring material will be destroyed by accretion as seen by the short time scale variations. The infall explanation will not violate the B-V colour effects for the intermediate time scale variations, it is still due to extra CS extinction. Here the brightening and fading are simply due to the accretion efficiency and do not need a complex explanation.

Continued monitoring, both in magnitude and in colours, of HD 45677 is necessary to select the correct mechanism.

6.1.4. Variations in tex2html_wrap_inline3439 values

From Table 4 (click here) we see that the obtained tex2html_wrap_inline3441 values range from 4.4 to tex2html_wrap_inline3443 5.8, and vary roughly with V, resulting in a nearly constant intrinsic stellar brightness. Also note that the maximum value of tex2html_wrap_inline3447 in the grid of extinction curves by Steenman & Thé (1991) is 5.8 and our SED fit resulted in that value for several fits near the state of minimum brightness of HD 45677 around 1980. Therefore, it seems likely that the true value of tex2html_wrap_inline3449 near minimum brightness will be somewhat higher than 5.8.

These variations of the tex2html_wrap_inline3451 values with stellar brightness had previously also been found by Brown et al. (1995) from UV extinction curves, in which they found the value of tex2html_wrap_inline3453 for the total extinction towards HD 45677 to vary between 4 and 5. Brown et al. (1995) did also note that the value of tex2html_wrap_inline3455 for individual dust clouds around HD 45677 must be much larger (tex2html_wrap_inline3457 > 7), see also Sect. 6.3.

From the variations of tex2html_wrap_inline3461 with brightness we conclude also that the long time scale variations are not caused by intrinsic variations of the central star, but to an external effect. A change in the average particle size in the circumstellar dust would indeed change the circumstellar extinction law. Note that the derived tex2html_wrap_inline3463 values describe a theoretical extinction law of Steenman & Thé (1991) which is taken to be the observed average extinction law from the UV to the near-IR, but wavelength-dependent deviations of this law due to other grain populations can be expected.

6.2. The spectroscopic behaviour

6.2.1 Long time scale variations

Concerning the long time scale variations, the major point is that the observations and conclusions of Swings et al. (1980) and all the observations made since 1928 (Swings & Allen 1971) are confirmed. The lines as detected in the new spectra are in majority very constant in relative strength, radial velocity and profiles, when detectable, compared to those seen before.

An exception is the FeII spectrum. As reported the FeII lines are double peaked and more dispersed that the [FeII] lines. The FeII lines are thus formed in another region, closer to the star with gases moving at higher velocities and therefore with wider emission profiles. Since there is no central re-absorption, the outer material is optically thin for this radiation. This leaves the double peaked FeII to be explained by a rotating disk or ring with 16 km stex2html_wrap_inline3465. In all observations made FeII shows an emission spectrum including the forbidden transitions. Note the remarkably stable average radial velocity of tex2html_wrap_inline3467 km stex2html_wrap_inline3469 (Swings & Allen 1971). The so-called stellar lines are stable at a somewhat higher displacement, tex2html_wrap_inline3471 km stex2html_wrap_inline3473. This is explained by outflowing material as the bright lines show absorption cores in their violet wing and low V/R values for most hydrogen lines (Merrill 1928).

In the 1992 intermediate resolution spectra the double peaked profiles show separations of about 40 - 60 km stex2html_wrap_inline3479, which is somewhat higher than in the spectra taken before the photometric minimum. This can be explained by assuming that the FeII lines are formed in the disk region which contains gas, moving faster than before 1982, and dense enough for some self-absorption, or with a higher rotation velocity. The stable identified nebular lines must be produced in a region further out, but in a region which is not much flattened and of which the velocity dispersion is low as they are not as broad as the FeII lines.

In combination with the dramatic photometric changes (thus in the continuum), we conclude that these do not seem to affect the spectral lines much. The mechanism causing the photometric changes is thus either separated from the line forming regions or not influencing the amount and other properties of the gaseous material. Furthermore, indications for the existence of a disk like region by spectral lines and their profiles are based on data taken and models proposed far before the 1950 event.

6.2.2 Short time scale variations

Although the long time scale variations are only seen for the photometric behaviour, short time scale changes in line profiles were reported above as well as before in many papers (Merrill 1928; Swings et al. 1980). This indicates that HD 45677 was already active in the beginning of this century. To have a better view of such behaviour today we monitored HD 45677 for several days around the NaID lines, including HeI, and Htex2html_wrap_inline3483.

First of all we detected the NaID lines to be in emission. If we take a one atom thick shell of neutral sodium at a distance of roughly 1 AU from a B2 V central star, all of this material will be evaporated and ionized within 1 second. The presence of tex2html_wrap_inline3487 and tex2html_wrap_inline3489 emission profiles, therefore, indicates the existence of cool material far away from the hot central star. On the one hand the slight P Cygni nature of these lines proves the existence of outflowing radiating cool gas from the central star. On the other hand, the strong absorption cores at 20 tex2html_wrap_inline3491 2 km stex2html_wrap_inline3493 indicate that a significant part of this material is at zero velocity, adopting a systemic velocity of 20 km stex2html_wrap_inline3495. The equal strength of the tex2html_wrap_inline3497 and tex2html_wrap_inline3499 absorption components, which are stable throughout all spectra, indicates the saturation of these components and therefore the presence of a high column density of the circumstellar material in the disk, of which the existence was indicated before.

Dense outflowing material, released in the disk, could be present as indicated by the blueshifted absorption lines being dynamical on a daily time scale. The outer sides of the disk can then be responsible for the wide sodium emission. It is not clear what the central velocity of this broad emission line is and therefore it remains unclear if the strong absorption at systemic velocity could be due to self-absorption in the outer part of the disk. Furthermore, collisions of the outflowing material with the accreting or study disk material at rest, causing inhomogeneous dense structures, could also be the origin of the blueshifted narrow absorption components; see also Israelian et al. (1996). This could explain the wide variety of the velocities, strength and duration of visibility of such components. However, the enrichment of the disk with new material could also act as the environment for the formation of the necessary large grains although the destruction of the material could be more significant, as discussed before.

It must also be noted that the variable brightness of HD 45677 could be the cause of the slight differences in the NaID emission strength, when produced in the outer parts of the circumstellar disk.

The variation in the strength of the HeI absorption line can be partly caused by obscurations of the central star as the photometry indicates such time scale variations. However, the variations are also seen in the line profiles and are too strong to be accounted for by the relatively weak photometric changes on the daily variations. As compared to 09/10/1993, Fig. 9 (click here)b, we see especially on 13 and 14/10/1993 extra ``emission''. Note that this component is asymmetric, the red side is extended to higher velocities, especially on 14/10/1993, when this relative emission is at its strongest. The extra emission or less absorption at the red side, exactly the opposite as what we see for the NaID lines in the same spectra, must originate from infalling material very close to the stellar surface. This could also produce collisionally ionized gas when it coincides with the inner disk material. Because of the redshifted origin, this component of a hot wind, can not be accounted for by a bipolar outflow. The existence of this emission component and the short time scale indicates a strong dynamical process close to the central star. This infall process is necessary for the existence of a bipolar flow anyway.

Because of the co-variations of the emission components that are red- and blueshifted for the HeI and NaID lines, respectively, there must be some common origin. Although the processes are of different velocity sign and the line formation region must be different we assume that on moments that dust-concentrations obscure the central regions and the star, we will not see much of the HeI emission region but only a contribution of the stellar spectrum. However, also parts of the disk itself will be obscured and these high dust densities can also cause the narrow absorption lines in NaID. It would be interesting to test this hypothesis by simultaneous high resolution spectroscopy and photometry.

In addition to these effects Israelian et al. (1996) reported emission at the violet side of the HeI profile as well at times when its central absorption is at its strongest.

Figure 12: E(B-V) versus photometric distance for stars from the Hipparcos Input Catalogue within 1tex2html_wrap3515 of HD 45677

next previous
Up: The peculiar B[e]

Copyright by the European Southern Observatory (ESO)