8. Concluding remarks

From an intensive study of the photometric and spectroscopic properties of HD 45677, including published data as well as own observations, we have detected many different variations and properties. We have tried to explain these variations, from short to long time scales. Different mechanisms must produce the variations, each with their own characteristics. Furthermore, we have discussed different options for the evolutionary status of HD 45677. Although the properties of HD 45677 are disputable we can make the following concluding remarks:

• Photometric variations and spectroscopic properties prove the hypothesis of a significant disruption of material around HD 45677. By extrapolation to the maximum brightness as measured by Swings & Swings (1972), before any large photometric changes occured, we determined this event took place around 1950. The minimum brightness at 8 8 was reached around 1981. Linear extrapolation of the current increase of brightness indicates that the maximum brightness of 7 2 will be reached again about the year 2020.
• During this long time scale obscuring effect we observed that the colours do not change significantly at both the obscuration and brightening phases. To explain this, formation of large grey grains is proposed which caused the strong obscuration of HD 45677 just after the 1950 event. The existence of such grains was also detected by extinction studies of Sitko et al. (1994) and already proposed by Low et al. (1970). We suggest that the origin of this material around 1950 is either due to the evaporation of a large cometary-like body, equal to less dramatic but similar as events for UX Ori (Grinin et al. 1994), or caused by a blow-out of material. Probably this blow-out was not extremely violent because of a fall back of material causing accretion effects.
• The relatively small and continuous changes of the colours during the period 1971-1993 are remarkable because of the opposite sign of the blue colours versus the red colours. If part of the erupted material falls (back) in the direction of the central star, a more compact inner disk region will cause higher extinction for the red colours, while the blue colours can be compensated with the production of small grains by accretion probably being visible in the bipolar flow.
• Variations on a time scale of days down to several minutes are detected. These short time scale variations are explained by brightenings or ``flickerings'' in the brightness as well as in the colour due to accretion. To understand the colour variations a production of small grains by accretion is necessary. The existence of small grain particles, producing scattering, is detected by an UV-excess (Pérez et al. 1993) which could be due to a bipolar flow (Schulte-Ladbeck et al. 1993), maybe produced but certainly enriched by the accretion itself. This accretion supports the explanations for the origin of the long time scale variations in which the presence of accretion is suggested.
• Photometric variations on an intermediate time scale are also detected and well known. These variations need more observations to be well explained. They seem to exist before and after the minimum in 1981 but their changes of 0 4 within a few years were probably absent before 1930. The intermediate time scale variations could be produced by revolving ``dust clouds'' or by shock fronts. In the latter case the blow-out hypothesis suggests certain eruptions. The occurrence of larger variations and the slightly getting brighter just before the strong fading around 1950 would support this. Then the intermediate time scale variations following the 1981 minimum are more probably due to infall. Support for this is that the brightening seems to occur in certain ``waves'' and dust destruction by accretion is seen by the higher amplitudes of the short time scale variations. The latter can vary then up to about 0.1 within a few days. Obscurations by ``dust clouds'' will also cause completely different colour effects as well discussed for the HAe UXOR type objects (e.g. Grinin et al. 1994 and references therein).
• The fact that changes from 4.8 in 1970 to 5.8 in 1981 and back to 4.4 in 1992 means that the average particle size in the circumstellar matter responsible for the extinction increased from 1970 to 1982 and decreased from 1982 until 1992.

  Probably the large grains were directly formed after 1950, due to a blow out or by the evaporation of a cometary-like body. Then the combination of the ongoing wind/outflow, coinciding with the previously erupted material and the pressure at the edges of the shockfront can produce larger grains by collision. The increase of the average grain size seems then to be most effective in obscuring the central star.

  At a certain point, 1982, the large grains were destroyed again. Infall of material seems to occur in waves, the intermediate time scale variations, which stimulates the accretion as seen as the higher amplitudes of the short time scale variations. As a result, the average grain size will decrease, producing the observed lower values and giving rise to brightening of the object again, apart from the effect of an expanding ring of material. A dramatic colour change on the longer time scale due to the production of these smaller grains is not necessary when the material is either effectively evaporated or destroyed and blown away in a bipolar flow. It supports the conclusion of Sitko et al. (1994) of a decrease of the total mass of the star's circumstellar dust envelope.

• Spectroscopically we have noticed that forbidden lines are narrow compared to other observed emission lines and are well centered on 20 km s. This velocity is also valid for the narrow absorption components of NaID and CaIIK lines and the central velocity of the FeII spectral lines. We adopted therefore a systemic velocity of 20 km s. Relative to this velocity in- and out-flow of gaseous hot and cool material are seen.
• Remarkable blueshifted absorption lines of NaID are detected at -10 to -20 km s. They are shifted with about 30 to 40 km s compared to the systemic velocity. They are probably caused by material swept up by the radiation pressure and stellar wind. This cool grain material will collide with the outer disk causing extra cogglumeration and obscurations on short time scales. Although the intermediate time scale variations can be explained in a similar way, by material moving away from the central star, a more plausible explanation is accretion activity. During the observations of the high resolution time series, this activity was decreasing, see Fig. 10 (click here)b.
• H profiles are detected to contain absorption components on the large emission profile at certain velocity regimes of which the red components are the most prominent. The H profiles are therefore interesting as they show velocity regimes of both infalling and outflowing material. Although the infalling gaseous material seems to be of higher densities, the strongest narrow absorption core is blueshifted with 9 km s with respect to the systemic velocity adopted as 20 km s. We might expect that especially this very strong H component originates from a very dense gaseous ring of which the velocity-dispersion will be as low as is observed. It could be questioned if this ``wave'' might exist for about 40 years.
• Accretion close to the stellar surface is indicated by the HeI and H short time scale variations, probably followed by a reaction of outflowing gas. This dynamical process could replenish the bipolar flow. It could be questioned if the spectroscopic short time scale variations show co-variations to the ``flickering'' as being of a similar, daily, time scale. Furthermore, the line variations are very different from those seen for many of the UXOR type stars (Grinin et al. 1994 and 1995) for which redshifted absorption components of the NaID are detected as evidence of the evaporation of cometary-like bodies.

  Observations as for the HeI lines are also reported by Israelian et al. (1996), whose conclusions do support our detection of accreting gas.

• Outflow of cool material is seen by emission lines that are stronger towards the blue as in MgI and in NaID. The strong absorption cores of the NaID and CaIIK indicate the presence of an optically thick disk at zero-velocity. This suggests that this disk was already present and is now enriched by the 1950 event.
• The presence of a disk before 1950 is also indicated by other emission lines which are stable throughout the large and long-term photometric variations and even far before 1950. In this case the 1950 event is due to an instability which is probably not the first one. Evidence of ``ancient'' activities, probably close to the stellar surface, were seen from 1926 on (Swings 1973) until now by activity in several of the Balmer lines.
• The evolutionary status of HD 45677 is discussed extensively. However, no strong conclusions can be drawn about its age. HD 45677 seems to fit best in the picture of Zickgraf & Schulte-Ladbeck (1989) of non-luminous B[e] stars in contradiction to the B[e] supergiants of the Magellanic clouds (Zickgraf et al. 1986). Probably in the case of HD 45677 the dust does not occur by a dense, slow expanding equatorial wind, but by eruptions. HD 45677 could therefore act as a key object to understand the reason of dust formation rather than to evolve to an early-type emission line star without dust.

Acknowledgements

The authors wish to thank Dr. B. Gilbert for providing the photometric data of HD 45677 in the Geneva system and R. van Duuren for his assistance to obtain the CAT/CES observations. We are thankful to Dr. J.P. Swings for his help, inspiration and patience in tracing some unpublished observations and to Dr. G. Israelian, Dr. C.A. Grady, Dr. M.R. Pérez and the referee for their many useful comments. This research has made use of the Simbad data base, operated at CDS, Strasbourg, France.