3 Spectral classification

3.1 Methods

To properly classify stellar spectra in the MKK system, the spectra of program stars have to be carefully compared with those of MKK standards. MKK standards should be observed with the same instrument and under the same conditions as the program stars. The reobservation of MKK standards is necessary to link the local instrumental system to the MKK system which is defined by its standards (Morgan & Keenan 1973; Morgan 1984). The spectra of a complete grid of MKK standards observed by us served as the basic frame for our spectral comparison and classification. Published atlases of grating spectra with similar resolution were also referenced. The main atlases used are those from Jacoby et al. (1984) and Turnshek et al. (1985).

Criteria of MKK classification in the classical region (3900 Å - 4900 Å) are described and illustrated in many sources (Keenan & McNeil 1976; Morgan et al 1978; Jaschek & Jaschek 1987). These criteria were used in our spectral classification. As our spectra also covered the yellow-red region (5000 Å - 8000 Å), all the spectral features present in the whole wavelength range from blue to red could be examined and compared. Many authors have been engaged in seeking suitable criteria for MKK classification in non-classical regions (Turnshek et al. 1985; Silva & Cornell 1992; Torres-Dodgen & Weaver 1993; Danks & Dennefeld 1994; Allen & Strom 1995; Shi & Hu 1999). As stated by these authors and illustrated by our spectra of observed MKK standards (Shi & Hu 1999), a lot of spectral features (lines, blends or bands) show temperature or luminosity effect in the yellow-red region. Among these, the intensity ratio of blend 6495/H$\alpha$ is a very good luminosity indicator for late-type stars, with the intensity of the blend significantly enhanced in supergiants. These new criteria did help in our stellar classification, especially in cases when good S/N could only be gained in the yellow-red wavelength range.

When classifying a stellar spectrum, we first figured out an approximate temperature type (e.g., early-B, mid-B or late-B). The second step was to determine its luminosity class. After that a more accurate temperature type was obtained. The last two steps were often performed several times before a final accurate MKK type could be fixed. When a program spectrum and a standard spectrum were compared, the whole spectrum as well as individual features corresponding to classification criteria were carefully examined. The program spectra were also compared with each other, to check that nothing went wrong during the process of spectral classification.

  

Table 5: Statistics of Be and B-type stars in the observed clusters

The errors in the MKK types classified by us have been estimated by: 1) comparing classification results of the two authors of this paper; 2) comparing results derived from spectra of different gratings (when existing); 3) comparing our results with those published in the literature for some brighter cluster stars; and 4) reclassifying the observed MKK standards; The errors of the resulting MKK types generally are smaller than 1 subclass in temperature types and about 1 subclass in luminosity classes. It should be pointed out that the luminosity class could be more easily determined using the spectra obtained with the grating of 600 lines/mm than the grating of 300 lines/mm. But even in the spectra of the lower dispersion luminosity classes I, III and V distinguished themselves from each other clearly.

3.2 Results

3.2.1 IC 1805

IC 1805 is the central cluster of the association Cas OB6. A proper motion survey was made by Vasilevskis et al. (1965) for 354 stars located in an area about two-thirds of a square degree and centered on BD +60$^\circ$502 (HD 15558) which is the brightest star of IC 1805. Spectra of 140 stars from this survey list were obtained by us. Of these observed stars, 126 have proper motion membership probabilities $p\geq 50$% according to Zhao et al. (1985). Reduced spectra of all the program stars in IC 1805 are shown in Fig. 3 and Fig. 4, for 600 lines/mm and 300 lines/mm grating respectively.

<


   Figure 3: Spectra of program stars in IC 1805 taken with 600 lines/mm grating

   Figure 4: Spectra of program stars in IC 1805 taken with 300 lines/mm grating

  Figure 4: continued

  Figure 4: continued

The results of the spectral classification on stars in IC 1805 are shown in Table 2. The star numbers from Vasilevskis et al. (1965) are listed in Col. 1. Proper motion membership probabilities from Zhao et al. (1985) are given in Col. 2. Spectral types published previously and their sources are presented in Cols. 3 and 4. MK types classified in this work are given in Col. 5. The grating(s) used in obtaining the spectra are listed in Col. 6, with "1" for the 300 lines/mm grating, "2" for 600 lines/mm and "3" for 1200 lines/mm.

IC 1805 contains a lot of massive stars, about forty from O4.5 through B2. These are most probable cluster members, judging from both their proper motion membership probabilities and early spectral types. The presence of very early O type stars indicates a very young cluster age and makes this cluster an desirable object for the study of the star formation process as well as for the determination of IMF at higher masses.

3.2.2 NGC 654

NGC 654 is a young cluster also located in the Cassiopeia region as IC 1805. Stone (1977) conducted a proper motion study for 186 stars in the cluster area. Spectra of 46 stars from this study were taken by us. This sample of observed stars corresponding to all the stars with membership probabilities $p\geq 50$% (Zhao et al. 1985) and brighter than $V=14\hbox{$.\!\!^{\rm m}$}0$. Only one exception is star No. 101 with p=36%. Reduced spectra of all the program stars in NGC 654 are shown in Fig. 5 and Fig. 6.

  

Figure 5: Spectra of program stars in NGC 654 taken with 600 lines/mm grating

  

Figure 6: Spectra of program stars in NGC 654 taken with 300 lines/mm grating

 

Figure 6: continued

The results of the spectral classification on stars in NGC 654 are shown in Table 3. The star numbers from Stone (1977) are listed in Col. 1. Proper motion membership probabilities from Zhao et al. (1985) are given in Col. 2. The other columns in Table 3 are similar to those in Table 2.

The earliest spectral type of around B0 means that NGC 654 is a more evolved young cluster than IC 1805. The presence of a few luminous stars, i.e., No. 111 (HD 10494, F5Ia), No. 68 (BD+61$^\circ$315, A0Ib-II), No.  2 (B1.5Ie) and No. 109 (B2II) makes NGC 654 an interesting cluster for the study of the evolution of massive stars.

3.2.3 NGC 6823

NGC 6823 is the central cluster of the association Vul OB1. Erickson (1971) made a proper motion survey of a region about 30${'}\times 30{'}$ centred on the central Trapezium system of NGC 6823. 146 stars were listed in that survey, but membership probabilities were computed only for the 92 stars brighter than $V=13\hbox{$.\!\!^{\rm m}$}0$ whose proper motions were better determined (Erickson 1971; Zhao et al. 1985). Spectra of 77 Erickson's stars were obtained by us. The observed stars went into three catagories: 1) 40 proper motion members, including all the stars with probability $p\geq 50$% except for star No. 73 which is too close to a bright star; 2) 19 proper motion non-members; and 3) 18 stars fainter than $V=13\hbox{$.\!\!^{\rm m}$}0$ to which no proper motion membership probabilities were assigned. Reduced spectra of all the program stars in NGC 6823 are shown in Fig. 7 and Fig. 8.

   Figure 7: Spectra of program stars in NGC 6823 taken with 600 lines/mm grating

Figure 7: continued

   Figure 8: Spectra of program stars in NGC 6823 taken with 300 lines/mm grating
The results of the spectral classification on stars in NGC 6823 are shown in Table 4. The star numbers from Erickson (1971) are listed in Col. 1. Proper motion membership probabilities from Zhao et al. (1985) are given in Col. 2. The other columns are similar to those in Table 2.

The earliest spectral type is O7, which suggests that the age of NGC 6823 may come between those of IC 1805 and NGC 654. The MKK types of two massive stars, No. 8 and No. 72, show some discrepancies among ours and other authors'. No. 8 is an O8V star here whereas it was classified as O9.5III by Turner (1979). No. 72 was assigned O7V((f)), O9Ia and O9.5III respectively by Turner (1979), Massey et al. (1995) and here. The spreads among spectral types for the above two stars(and maybe some others of similar cases in Table 2 through 4) as classified from time to time may indicate possible changes of spectral features due to some reason(s), rather than misclassification in the spectral types.