We now discuss the results of the SED fits for the programme stars for which we altered the originally derived astrophysical parameters or which show peculiarities.
Table 7 (click here) lists all astrophysical parameters of the programme stars. Parameters obtained from the SED fits were used and superseded previous determinations.
For most of the SED fits special remarks are given in Sect. 5.1.
of Koulis (1993). The SEDs for which we could not obtain satisfactory fits,
encountered some problems, or for which several fits were
possible, are discussed now:
W213: we consider two options, denoted by W213(1) and W213(2), below. In both cases the photometry used is identical except for the UBV set. The subscript 2 in indicates the reference of the data-set used in the last column of Tables 2 and 3, and in Sect. 2.1.
W213(1): we take the data and the spectral type A7. Because of the IR-excess, we have corrected with = 3.4. Except for the Balmer discontinuity, the obtained fit is reasonable (except that the J datapoint lies below the model).
W213(2): we use the averaged data and fit a spectral type F9. The obtained fit becomes acceptable after we have applied an extinction law with = 3.4. However, the and J datapoints still lie above and below the Kurucz model respectively. We keep both options for further consideration.
W232: we use spectral type F8. The fit shows an UBV as well as a near-IR excess, while the observed datapoint lies below the fit. We eliminate the IR-excess by correction with = 4.0. However, the fit remains unsatisfactory for the UBV datapoints.
W273: this proves to be an interesting case. We have started using a spectral type B9, but have to modify this to A0, with E(B-V) = 0 64. The obtained fit is not good, because most of the photometric blue and visual datapoints are just covered by the fit, whereas the datapoint is too high. Furthermore, the star shows an IR-excess for which we have corrected by considering two possibilities:
(a) Fitting an anomalous extinction law with = 4.8, we found that the blue, visual and red part of the fit are improved. However, the IR-excess remains, and the J and K datapoints are asymmetrically above and below the fit.
(b) W273sh: we have fitted the IR-excess by a Planck curve (see Sect. 5.3) with 200 K. The combination of such a near-IR explanation and a fit with = 3.1 gives reasonably good results at all pass-bands. The addition sh in W273sh indicates that the IR-excess is assumed to be due to a dust shell.
W276: like W213 here we also consider two options, in which the JHK sets are different:
W276(1): with the set the fit becomes satisfactory if an anomalous extinction with = 3.4 is applied, although the U datapoint does not fit in the Balmer discontinuity and the K datapoint is located slightly below the fit.
W276(2): we use the same procedure for , but with = 4.0 instead. Although the K datapoint fits well, the J stays below the fit. Furthermore, the blue part of the spectrum appears to fit less well compared to the SED fit of W276(1). Option W276(1) provides a relatively better fit, which we adopt.
W299: here again we have two options:
W299(1): we use the averaged data.
W299(2): we adopt the data.
For both options we have used a spectral type B6 and the same sets of Walraven and near-IR data. If the Walraven data set proved not to be in mutual agreement, we discarded them. Both fits show approximately the same characteristics. Because of the presence of an IR-excess we have fitted with = 3.8. The J and K datapoints fall asymmetrically below and above the fit. This motivates us to consider fitting a Planck curve, see Sect. 5.3.
W336: we have W336(1), where we have fitted the data, and W336(2) the data. W336(1) shows a JHK depletion and W336(2) an IR-excess. We chose to consider W336(2) and rejected W336(1). We then derived = 3.5 for a B3 spectral type with E(B-V) = 0 69. This is a satisfactory fit.
W339: we considered 3 different options:
W339(1): based on data. The spectral type is adopted to be B3 and E(B-V) = 1 08 is taken. This SED fit shows no IR-excess.
W339(2): has been used and we obtained an extinction law with = 3.9, the fit was very satisfactory.
W339(2)sh: we have fitted the IR-excess with a Planck curve, see Sect. 5.3.
We notice that the SED fit of W339(2) is the best one, that of W339(2)sh being of somewhat poorer quality. We therefore discarded option W339(1). Note that the CCD spectrum of W339 seems to have characteristics of a symbiotic system.
W374: with spectral type F2 we have obtained a satisfactory SED fit after correcting the presence of a large IR-excess with = 4.4. Since the IR-excess is strong, we decided to make a second fit with a CS dust shell (see Sect. 5.3).
W396: two possible spectral types were considered, because it lies low in the TCDs.
W396(a): spectral type ranging from A8 to F5 (photometry) and F9 from the CCD spectrum. Several weak emission lines are visible.
W396(b): spectral type ranging from G2 to G6 (photometry), in which G2 is deduced from an IDS spectrum.
The SED fits for both W396(a) (spectral type F9, E(B-V) = 0 64) and W396(b) (spectral type G2, E(B-V) = 0 56) are both very satisfactory. However, we have corrected the SED with = 3.5 for W396(a) and with = 3.7 for W396(b). Probably W396(a) and W396(b) are two different neighbouring stars, lying almost along the same line of sight. This would explain the two different spectra as well as the relative discrepancy in their value; they are not resolved in the photometric observations.
W440(1): Originally we have also considered W440(2). From and a very poor CCD spectrum (not shown here) the spectral classifications (type B9) seems to be of another star, as the SED fit is also poor. This option is therefore discarded.
W455: the SED fit shows a near-IR depletion. Changing the spectral type of the star from B8 to B7 and fitting for different luminosity classes gave no satisfactory result. Finally we have decided to keep the original fit.
W489: we have two somewhat different sets of JHK data. For both an SED fit is made with the same Walraven and data. Although both fits are of poor quality in the near-IR (the J and K data lying asymmetrically below and above the fit) we have obtained the same results: spectral type B7 (note that initially we have taken B9) and an anomalous extinction law with = 3.3 are found.
W536: initially four different SED fits, using 2 different sets of Walraven data and two different sets of JHK data, are made. We have discarded the two SED fits with the data, because the data yield better fits. In addition to this, we have averaged the 2 Walraven data sets as they are in agreement with each other. The spectral type (B1) remains the same, but we have to correct for an anomalous extinction with = 3.5.
W611: we have made numerous SED fits by varying the temperature and luminosity class, and staying within the spectral range determined from our photometric and spectroscopic results. The data fitted well and the extinction appeared to be normal. Because of a remaining depletion at the JHK data, we finally have classified W611 as G8 V. Initially we used K0 V.
W617: we have two very different UBV data sets. The data do not match with the other photometric data. Furthermore, we have to discard as these data lie far below the fit. Although the blue data still lie somewhat lower than the fit, the resulting fit with spectral type K5 V is satisfactory. No indication for an anomalous extinction law is found.
After we have selected self-consistent data and removed erroneous data points at the SED-fits, several objects still show a remarkable depletion in the near-IR: W103, W349, W402, W406, W411, W504, W525 and W534.
The IR-depletion was explained in Thé et al. (1990) by values lower than 3.1. However, the depletion remained after trying to vary the extinction law and the Kurucz model. If the photometry and spectroscopy are reliable and the stellar astrophysical parameters derived from them are correct, the only adjustable parameter left (besides the presence of strong absorption bands in the near-IR) is the luminosity class. Indeed, by fitting at the red, optical and blue data we are able to change the log g in such a way that reasonable fits in the near-IR for several late type objects can be made.
We have obtained satisfactory fits for W349: G8 I, W406: K0 III, W525: K5 III and W534: K2 III.
For the other two objects, W402 and W411, the depletion at the JHK(LM) pass-bands can not easily be compensated using this procedure. The best fits obtained are spectral type G9 III for W402 and K5 III for W411.
As we have mentioned before, a PMS characteristic is brightness variability. Normally this will not alter the stellar spectral type. However, for several PMS objects the colours can change rapidly (see Bibo & Thé 1991). This happens especially in the optical and the blue pass-bands. We must, therefore, keep in mind that the inability to obtain a satisfactory SED fit in cases where there is a near-IR depletion, is probably because the photometric data were taken at different brightness stages. Furthermore, strong molecular bands can influence the near-IR photometry significantly. This could be the origin of the near-IR depletion in the SEDs of W103: B9 and W504: B5.
In Sect. 5.1. we encountered some objects for which a fit of a CS dust shell to their IR-properties was proposed. Additionally, we have some remaining cases for which the IR-excess could not be fitted by anomalous extinction: W245, W262, W266, W494 and W605. Sometimes it is possible to obtain a reasonable fit by using an anomalous extinction law with extreme values, far above 4.0, but then the obtained fit shows a poor agreement in the blue.
A strong IR-excess, other than one due to anomalous extinction or free-free emission, is a well known PMS characteristic. Such excesses are normally explained as due to thermal re-radiation of circumstellar dust grains. Also an M-type ``companion'' (in a star forming region could be a T Tauri- or a proto-star) to explain the IR-excess, show approximately, a Planck spectrum. This latter explanation is presented as the explanation for the variable IR-excess of NGC 6611-W409 (Thé et al. 1985). To study these two options we calculate the amount of near-IR excess in each passband and try to fit the flux differences to a black body. We thus obtain a temperature of the black body by using Wien's law ( cm K, where is the wavelength corresponding to the maximum flux of the used Planck curve). This temperature is used to approximate the temperature of the radiating dust.
The stars showing
one of these two properties are: W245sh, W273sh, W299sh, W339(2)sh,
W374sh, W494sh and W605sh, where the abbreviation sh stands for
shell (we will not discuss here the correct morphology of the circumstellar
material distribution: disk or spherical shell).
W245sh: its SED fits a spectral type B6 well, with a strong IR excess. An IR-excess for this star was also found by Hillenbrand et al. (1993), but a photometric spectral type of B9.5 was used. We have tried to fit 3.4, but it was soon clear that the SED at the UBV passbands became tilted off the Kurucz model significantly, while the JHKL points could not be fitted appropriately. However, we obtained a satisfactory fit for a Planck curve of 1880150 K. We assume therefore that this radiation is caused by a CS dust shell.
W262sh: the IR-excess has been fitted with a Planck curve of 2550200 K. The SED fit is very satisfactory. However the temperature is much too high for CS dust. We therefore suggest the presence of an M-type companion.
W266sh: has an SED with a black body of 1970150 K. This temperature is somewhat high for a CS shell. However, considering the errors in the temperature estimate, the presence of a CS dust shell is possible.
W273sh: has been considered in sect. 4.1. We obtained a temperature of 2440200 K for the Planck curve, which indicates the presence of an M-type companion. Judging the two SED fits, the Planck curve fit is relatively better.
W299sh: an SED fit with = 3.8 gives a reasonable result. However, we also tried a 2070200 K black body. The best fit is the one for the anomalous extinction law.
W339: the different possibilities of the SEDs are already discussed. For W339sh we have also tried a Planck curve of 2550200 K to fit the near-IR-excess. However, this fit is of poorer quality compared to the one of W339(2), especially in the near-IR. We are thus compelled to assume that this object suffers from anomalous extinction with = 3.9. This option must compete with the possibility that the IR-excess is due to an M-type companion, as is noticed from the spectroscopic observations, Fig. 1 (click here).
W374sh: is another star for which we considered different options. We compared the SED fit for = 4.4 with the one for a Planck curve of 2550200 K. The quality of the SED fit of the former is the best.
W494sh: a Planck curve of 1760100 K fits very well to the IR-excess in the SED. The radiating body can thus be a CS dust shell. For this object an IR-excess was also found by Hillenbrand et al. (1993).
W605sh: a CS dust shell with 136080 K is found to be the best fit for the near-IR-excess.
In conclusion, we find that W245sh, W266sh, W494sh and W605sh clearly present dust properties as known in HAeBe candidates. On the other hand W262sh, W273sh and W339(2) appear to have an M-type companion, whereas W299sh and W374sh show a strong near-IR-excess due to highly anomalous extinction ( = 3.8-4.4).
In the cases above, free-free emission might also be considered as young objects have a significant amount of gaseous circumstellar material.
Note that for those objects showing a strong near-IR-excess we have adopted a normal extinction law ( = 3.1). Although we suspect that the cluster is mostly influenced by anomalous extinction, we are not able to make appropriate conclusions about the extinction of the previously discussed shell-type stars. If their extinctions are indeed anomalous, they could change the amount of near-IR-excess, thereby lowering the estimated temperatures. Consequently, the temperatures derived above should be considered as upper limits. However, the overall conclusions will not change significantly.
Figure 1: Low resolution spectra of the PMS programme stars of NGC 6611
Figure 2: The U-B, B-V diagram for our stars in NGC 6611. The intrinsic luminosity class V relation (solid line) and the one for class III (dashed line) of Schmidt-Kaler (1982) are drawn
Figure 3: SEDs of our sample stars. In some cases the different options are also given, see text and Table 7. Note that a correction with a high value causes a small UV-depletion when the IR-excess is probably not due to anomalous extinction