3. Study of reddening and extinction of NGC 6913
3.1. Comparison among the different observational data
Many E(B-V) values derived from our spectral classification are different from those based on photometry (Ep(B-V)). Some show a very large difference. 32 stars have absolute values of (E(B-V)-Ep(B-V)) greater than 0.3 mag., the largest one of as much as 1.48 mag. (#219)! We find that the stars with large absolute values of (E(B-V)-Ep(B-V)) are mainly some early B type stars and most of the F type stars. The values for Ep(B-V) of the B type stars are far lower than for E(B-V), and the values for Ep(B-V) of the F type stars are higher than for E(B-V). These results are mainly due to the effect of reddening and the uncertainty of the photometric classification; moreover, they have different R values. The difference shows the importance of the spectroscopic observations in studying young star clusters which include many pre-main sequence stars.
In addition, For the same cluster, Crawford et al. (1977) selected a small sample to do four-color and H photometry. They found the disagreement between the distance modulus derived from the photometry alone and from that using spectral types. What is particularly striking is that they never found it in their previous cluster work. This indicates once more that to study the clusters with large variable extinction, using the photometric method only is insufficient.
3.2. The distance modulus and H-R diagram
In order to construct the physical H-R diagram for NGC 6913, the distance modulus must be known accurately. The distance of NGC 6913 estimated by various authors ranges from 0.8 kpc to 2.8 kpc. Up to now a relatively dependable value of the distance is () kpc, corresponding to a distance modulus mag (Joshi et al. 1983). This is obtained by fitting the ZAMS given by Allen onto the unevolved part of the cluster main sequence present in the (V0, (B-V)0) and (V0, (U-B)0) diagrams respectively.
We obtained the cluster distance by means of the ZAMS fitting method. Three main-sequence stars (#124, 126 and 130) satisfied the strict conditions, i.e. membership probability around 80% (or more), OB type stars, location within the central region, 0.70 E(U-B)/E(B-V) 0.80 (normal reddening law) and a value of E(B-V) not too high compared with other OB stars in this cluster. After fitting to the ZAMS locus, a mean distance modulus of mag. is derived, which corresponds to a distance of 1.08 kpc.
Fig. 3 is the H-R diagram constructed by using the dereddened (B-V)0 on the basis of our spectroscopic classification and the V magnitudes, where the ZAMS line is evaluated based on the distance modulus of 10.17. The ((U-B), (B-V)) colour-colour diagrams of NGC 6913 are plotted in Fig. 4, in which the zero-age main sequence line, III and Ib type luminosity line are drawn. In this diagram, the reddening line is also plotted with a slope of 0.72.
On the basis of the above two diagrams, we found that most of the stars in the cluster are still in the pre-main-sequence stage. They have an abnormal reddening slope. Only 10% of the stars in this open cluster approximately satisfy the normal interstellar reddening law. This problem is usual in star forming regions and very young star clusters.
3.3. Membership of NGC 6913
Crawford et al. (1977) did four-colour and H photometry for 25 stars in NGC 6913 and found that of stars called members by Sanders, 14 stars were real members and 6 probable non-members. Since some of the stars with membership probability greater than 50% in the cluster field are non-member stars, we can try to identify those possible non-members of the cluster in the full field.
In order to rule out most of the possible foreground stars, a reddening method is used. Considering the distance of NGC 6913 to be 1.08 kpc, to avoid removing the real cluster members, a distance of 1.0 kpc is used to do the preliminary work. Neckel & Klare (1980) (hereafter NK) gave the galactic distribution of the interstellar extinction in the galactic belt and presented reliable Av(r)-relations of 325 fields. As NGC 6913 is at galactic coordinate l=76.92, b=+0.61, from the NK relation between distance and optical extinction Av in different regions of sky a value of Av of about 2.0 is derived (field 286(76/+1) in NK), corresponding to a value of E(B-V)=0.65 if we assume R=3.1. Therefore, those stars with E(B-V) less than 0.65 could be foreground stars. These comprise two fifths of the stars Sanders lists as members. Clearly, most of the OB stars remain in the list and most later-type stars are removed.
Even so, the value of E(B-V) 1.03 in the cluster is still very large. Extremely non-uniform reddening across NGC 6913 can still be inferred.
However, there is something very important to which we should pay special attention. In NK, the extinction in this region at a distance below 0.8 kpc is only about 0.1 mag., and sharply increases to over 2 mag. from about 0.8 kpc to 1.1 kpc. This could be caused by the high reddening star clusters. Therefore, the stars with E(B-V) smaller than 0.65 discussed above are not definitely non-members of the cluster, Some of which could be members lying on the edge of the cluster. We can use a mean value of the extinction of 1.0 mag. (with an E(B-V) value of 0.32) to judge whether a star is a real non-member of NGC 6913. Those stars with E(B-V) values between 0.32 and 0.65 are possibly either members or non-members. Based on thi principle, the percentage of non-members in Sanders' list of members amounts to . Supposing that half the stars with reddening values between 0.32 and 0.65 are non-members of the cluster, a total amount of of the stars with membership probability greater than 50% can be assumed to be non-members, coincident with the proportion of non-members evaluated by Crawford.
3.4. Reddening and extinction across the NGC 6913 field based on the Sanders' cluster members ( membership probability)
The number distribution of E(B-V) is presented in Fig. 5. We note that there are 9, 20, 21, 28 and 16 stars respectively corresponding to , between 0.9 and 1.2, between 0.6 and 0.9, between 0.3 and 0.6, as well as 0.3 mag. The average value of E(B-V) is 0.71 with E(B-V) = - = 1.82 mag. (the two negative reddening values are excluded here). It has been pointed out by Burki that E(B-V) greater than 0.11 is an indication of the presence of nonuniform extinction across the cluster (Burki 1975). Thus the extremely variable reddening across NGC 6913 can be inferred.
Fig. 6 shows the spatial distribution of extinction across the NGC 6913 field, in which the diameter of the circle is 12´, corresponding to twice the possible smallest cluster diameter given by Lyngå (1987). From this figure, we find that most of the nine stars with E(B-V) greater than 1.2 lie on the edge of the field. Most of the stars with E(B-V) between 0.9 and 1.2 are concentrated in the central area. All of the stars except one with E(B-V) between 0.6 and 0.9 lie in the western part (right-hand side of the figure) of the cluster. Those with relatively small values of E(B-V) are distributed randomly over the entire cluster region. It is clear that the region with the largest mean value of extinction is in the central part. The western region comes second.
On the same scale as Fig. 6, Fig. 7 shows the spatial distribution of those stars with high mass and relatively low mass respectively. Surprisingly, it was found that the distribution of OB stars coincides very well with the distribution of large colour excess across the field. All nine stars with E(B-V) greater than 1.2 are B-type stars. From Figs. 6 and 7, clearly the surroundings of the massive stars seem to be more obscure than those of the low-mass stars.
Fig. 8 shows (B-V)0 versus E(B-V). It clearly shows a dependence of E(B-V) on spectral type. For those stars earlier than F8 (i.e. (B-V)0 less than 0.6), the colour excess increases together with the effective temperature of the stars. But the stars with spectral type later than F8 seem to show reversed characteristics. Because the number is too small, we cannot confirm the latter point. A similar result for the whole field was obtained by Sagar (1987). In order to understand if the tendency is caused by a systematic error in Joshi's observations, we have compared the data for the open cluster NGC 6823 (both NGC 6823 and 6913 show a tendency like Fig. 8 in Sagar's paper), observed by Sagar & Joshi (1981) using the same instruments, with the data observed by other people (e.g. Guetter 1992; Stone 1988). We find that all of them have the same trend. Therefore, the trend should be reflection of reality.
3.5. The interstellar diffuse bands in the stellar spectra
The correlation between interstellar reddening and the interstellar diffuse bands has been proven by Greenberg & Chlewicki (1983). A cluster is a very good sample to check this relationship. There are many different types of stars in a cluster. By comparing the intensity of the IDBs in the spectra of stars of different types, the relation between IDBs and the stellar parameters (e.g. temperature, luminosity etc.) could be derived.
The IDBs appear distinctly in the spectra shown in Fig. 1 (e.g. at 6284Å etc.). Although in a low resolution spectrum it is difficult to measure the equivalent width of the relatively weak IDBs, the spectra of early type stars in NGC 6913 clearly show relatively strong absorption of the interstellar diffuse bands, and the low-mass late-type stars show relatively weak absorption, or even no absorption. This means that the strength of the IDBs is well correlated with effective temperature of stars.
Krelowski et al. (1998) stated that IDBs are not all of the same origin and usually quite well correlated to E(B-V). In our present work a correlation between spectral type and E(B-V) of NGC 6913 has been shown, thus the IDBs also correlate well with the reddening, in agreement with Krelowski's conclusion.
© European Southern Observatory (ESO) 2000
Online publication: March 28, 2000