EDP Sciences
Free Access
Issue
A&A
Volume 510, February 2010
Article Number A22
Number of page(s) 21
Section Stellar atmospheres
DOI https://doi.org/10.1051/0004-6361/200913120
Published online 02 February 2010
A&A 510, A22 (2010)

The chemical composition of the Orion star forming region

I. Homogeneity of O and Si abundances in B-type stars[*],[*]

S. Simón-Díaz1,2

1 - Instituto de Astrofísica de Canarias, 38200 La Laguna, Tenerife, Spain
2 - Departamento de Astrofísica, Universidad de La Laguna, 38205 La Laguna, Tenerife, Spain

Received 14 August 2009 / Accepted 27 October 2009

Abstract
Context. Recent accurate abundance analyses of B-type main sequence stars in the solar vicinity has shown that abundances derived from these stellar objects are more homogeneous and metal-rich than previously thought.
Aims. We investigate whether the inhomogeneity of abundances previously found in B-type stars in the Ori OB1 association is real (hence a signature of enrichment of the newly formed stars in an induced star formation scenario) or a consequence of intrinsic errors induced by the use of photometric indices to establish the stellar parameters prior to the abundance analysis.
Methods. We obtained a new (improved) spectroscopic data set comprising 13 B-type stars in the various Ori OB1 associations, and performed a detailed, self-consistent spectroscopic abundance analysis by means of the modern stellar atmosphere code FASTWIND.
Results. We detect systematic errors in the stellar parameters determined previously which affect the derived abundances. Once these errors are accounted for, we find a high degree of homogeneity in the O and Si abundances for stars in the four Ori OB1 subgroups. The derived abundances are in very good agreement with recent determinations in other B-type stars in the solar vicinity. We also compare our results with those obtained for the Sun during the epoch of the ``solar crisis'', and the Orion nebula.

Key words: stars: early-type - stars: atmospheres - stars: fundamental parameters - stars: abundances

1 Introduction

For many years, our knowledge about the chemical composition of early-B main sequence stars in the solar vicinity has been characterized by two main results: (i) the derived abundances seemed to be highly inhomogeneous (with a dispersion of up to 0.5 dex); and (ii) the mean values indicated lower abundances than the standard (Grevesse & Sauval 1998) set of solar abundances (see reviews by Morel 2009; Herrero & Lennon 2004). These results were not very encouraging, because the inhomogeneity of stellar abundances contradicted with the homogeneity in oxygen abundance found from studying the local diffuse interstellar medium (e.g. Meyer et al. 1998; Cartledge et al. 2006). On the other hand, chemical evolution models of the Galaxy (e.g. Carigi et al. 2005; Chiappini et al. 2003) predict a small enrichment of the ISM in metals during the lifetime of the Sun (i.e. because they are younger than the Sun, nearby OB-type stars are expected to be slightly metal-rich).

Some recent results have began to change this situation. The solar oxygen abundance traditionally considered as a cosmic abundance reference (Grevesse & Sauval 1998) was reviewed by Asplund et al. (2004), who obtained log (O/H) = 8.66 dex, 0.17 dex lower than the standard value. This was the beginning of what has been called the epoch of the``solar crisis'': between 2004 and 2008, several studies by different authors (Ayres 2008; Caffau et al. 2008; Centeno & Socas-Navarro 2008; Socas-Navarro & Norton 2007; Meléndez & Asplund 2008; Ayres et al. 2006; Allende Prieto 2008) presented solar oxygen abundances derived by means of different approaches. The calculated values range from 8.63 dex (Socas-Navarro & Norton 2007) to 8.86 dex (Centeno & Socas-Navarro 2008). The debate about its actual value is still open.

Przybilla et al. (2008) have recently analyzed a representative sample of six unevolved early B-type stars in nearby OB associations and the field, and found a very narrow distribution of abundances, with mean values that are more metal-rich compared to previous works (e.g. log (O/H) = 8.76 dex, a value that is within the range of solar abundances calculated during the ``solar crisis''). These authors indicate the importance of properly determining the atmospheric parameters and using robust model atoms to avoid systematic errors in the abundance determination. (See also Nieva & Przybilla 2009, for a summary of the main sources of systematic errors affecting the abundance analyses of B-type stars.) The study by Przybilla et al. show that the chemical inhomogeneity previously found for B-type stars in the solar vicinity may be spurious and an artificial effect of those systematic errors. It also reinforces the importance of self-consistent spectroscopic abundance analyses (i.e., the stellar parameters and the metal abundances are determined exclusively from spectroscopic diagnostics by using the same set of stellar atmosphere models). Less accurate photometric $T_{\rm eff}$ estimates must be handle with care (or better avoided whenever possible!) in the abundance analysis of B-type stars (see also Nieva & Przybilla 2008).

The Orion complex, containing the Orion molecular cloud and the Orion OB1 (Ori OB1) association, is one of the most massive active star-forming regions in the 1 kpc centered on the Sun. Blaauw (1964) divided Ori OB1 into four subgroups of stars - namely Ia, Ib, Ic, and Id - having different locations in the sky and ages. Brown et al. (1994) derived mean ages of $11.4 \pm 1.9$, $1.7 \pm 1.1$[*], $4.6 \pm 2$, and <1 Myr for subgroups Ia to Id, respectively. The youngest subgroup Ori OB1 Id is associated with the Orion nebula (M 42), the most studied H  II region and the closest ionized nebula to the Sun in which a high accuracy abundance analysis can be performed.

The correlation between the ages of the stellar subgroups, their location, and the large scale structures in the interstellar medium around Orion OB1 have been interpreted as features of sequential star formation and type-II supernovae (Cowie et al. 1979; Brown et al. 1994; Reynolds & Ogden 1979). Cunha & Lambert (1994,1992) obtained C, N, O, Si, and Fe abundances of 18 B-type main sequence stars from the four subgroups comprising the Ori OB1 association. They found a range in oxygen abundances[*] of $\sim$0.4 dex, with the highest values corresponding to the stars in the youngest (Id and some Ic) subgroups. In this case, the inhomogeneity in stellar abundances (mainly oxygen and silicon) seemed to be real and coherent with a scenario of induced star formation in which the new generation of stars are formed from interstellar material contaminated by type-II supernovae ejecta.

The study by Cunha & Lambert was based on a photometric estimation of $T_{\rm eff}$and the fitting of the H$\gamma$ line computed from Kurucz (1979) LTE model atmospheres to the observed one to derive log g. In a more recent work, Simón-Díaz et al. (2006) used one of the new generation of NLTE, line blanketed, model atmosphere codes ( FASTWIND, Puls et al. 2005; Santolaya-Rey et al. 1997) and a self-consistent spectroscopic approach to derive the stellar parameters and oxygen abundances for the three B0.5 V stars in Ori OB1d (the youngest and, supposedly, more metal-rich subgroup). The resulting stellar parameters were somewhat different and, more important, the derived abundances were systematically lower than the previous values by Cunha & Lambert (1994) by $\sim$0.2-0.3 dex.

This result motivated us to review the chemical composition of the other B-type stars in Ori OB1 to investigate whether the inhomogeneity of abundances previously found is real or a consequence of intrinsic errors induced by the use of photometric indices to establish the stellar parameters prior to the abundance analysis. To this aim, we obtained a totally new (improved) observational data set and performed a self-consistent abundance analysis of 13 of the stars considered by Cunha & Lambert (1994,1992).

We used FASTWIND to derive the stellar parameters, oxygen, and silicon abundances. The observational data set is described in Sect. 2. The whole spectroscopic analysis is presented in Sect. 3. Then, we compare our stellar parameters and abundance with results from previous works (Sect. 4). The homogeneity of stellar abundances in Ori OB1 and its comparison with other B-type stars in the solar vicinity determinations, the Sun, and the Orion nebula is discussed in Sect. 5. The main conclusions of this work are summarized in Sect. 6.

2 Observational data set

Table 1:   List of B-type stars from Ori OB1 considered in this study, ordered by spectral type.

\begin{figure}
\par\includegraphics[angle=90,scale=0.62]{13120f1.eps}
\end{figure} Figure 1:

Example of FIES@ NOT spectra for three of the observed stars. The complete atlas is available in the electronic version of the paper (Fig. 8). The Si II-IV and O II lines used for the abundance analysis are indicated as solid and dashed vertical lines, respectively.

Open with DEXTER

The observations used here were carried out with the FIES cross-dispersed, high-resolution echelle spectrograph attached to the NOT2.5 m telescope at El Roque de los Muchachos observatory on La Palma (Islas Canarias, Spain) on 5-8 November 2008. The medium-resolution mode ( R = 46 000, $\delta\lambda = 0.03$ Å/pix) was selected, and the entire spectral range 3700-7300 Å was covered without gaps in a single fixed setting. A sample of 14 stellar candidates was observed, after being selected from the list of stars in the Ori OB1 association analyzed by Cunha & Lambert (1994,1992). These are early-B type, non-evolved, stellar objects (B0 V-B2 V) with low projected rotational velocities (v sin i $\leq$ 60 km s-1). The signal-to-noise ratio achieved for all the spectra was always above 250. The list of observed stars is presented in Table 1. Apart from the new set of stars, we re-observed HD 37020 and HD 37042, two of the three B-type stars[*] analyzed in Simón-Díaz et al. (2006).

The spectra were reduced with the FIEStool[*] software in advanced mode. The FIEStool pipeline provided wavelength calibrated, blaze-corrected, order-merged spectra of high quality. These spectra were then normalized with our own developed IDL programs. An example of the FIES@ NOT spectra for three of the observed stars is presented in Fig. 1, where the Si II-IV and O II lines used for the abundance analysis are also indicated.

3 Spectroscopic analysis

The analyses were performed following a self-consistent spectroscopic approach with the spherically extended, NLTE, line-blanketed stellar atmosphere code FASTWIND (Puls et al. 2005; Santolaya-Rey et al. 1997). Basically, the stellar parameters were derived by comparing the observed H Balmer line profiles and the ratio of Si III-IV and/or Si II-III line equivalent widths with the output from a grid of FASTWIND models. Whenever possible, the He I-II ionizing equilibrium was also considered. Then, the same grid of models was used to derive the stellar abundances by means of the curve-of-growth method.

3.1 Projected rotational velocities

The projected rotational velocities (v sin i) were obtained by applying the Fourier method (Gray, 1976; see also Simón-Díaz & Herrero, 2007, for its application to OB-type stars) to the Si III $\lambda$4552 line. The 0.03 Å/pix resolution of the FIES spectra implies that the lowest v sin i that could be detected is 2-3 km s-1; however, for those cases with v sin i < 10-15 km s-1, identifying of the first zero of the Fourier transform was difficult due to the effect of the noise (an maybe the microturbulence).

In many cases, an additional extra Gaussian-type broadening ( $\Theta_{\rm g}$) was needed to properly fit the line profile for the derived value of v sin i. This extra-broadening account for the microturbulence (and maybe the macroturbulence), also affecting the shape of the line. The corresponding derived values are summarized in Table 2.

3.2 Measurement of equivalent widths and identification of problematic lines

Table 2:   Projected rotational velocities derived for the studied stars.

Table 3:   Final results of the HSi analysis: stellar parameters and Si abundances.

The strategy we followed in our analyses is based on the equivalent widths (EW) of metal lines. Therefore, proper identification of the lines of interest, along with an accurate measurement of their EWs (also including the associated uncertainties), is a very important step. We have developed IDL routines automatically identifying metal lines in the spectra, measuring the EWs and their uncertainties, and detecting the possibility that other lines affect these measurements. To this aim we compiled a list of C, N, O, Si, Mg, S, Ne, and Ar lines, extracted from the atomic[*] line list v2.05. For a given line, the program performs a multi-Gaussian fit of the observed line profile accounting for all the lines expected to be present in a certain spectral range ($\lambda_0$ $\pm 2~\max$[v sin i$\lambda_0$/c, 0.5$\lambda_0$/R]) around the wavelength indicated in the line list. The uncertainty in the EW measurement is obtained by assuming the location of the local continuum at $\pm$1/SNR and, in those cases in which the line is isolated, comparing the value obtained by means of the Gaussian fitting with the value derived by integrating the line.

The high quality of our spectra allowed us to identify and measure the EWs of up to 27 Si II-IV lines and 47 O II lines. Some of the lines were labeled as problematic because of the presence of one or more lines from other elements (e.g. the O II $\lambda$4673.73 line may be affected by the C III $\lambda$4673.95 in same cases; similarly occurs for the O II $\lambda$4641.81 line, coincident with the N III $\lambda$4641.85 line). These lines were treated with special care in the abundance analysis, since they may be giving wrong values for the abundance.

Generally, the Gaussian fit provides reliable results for the EWs; however, in those cases in which the v sin i of the star is above 30-35 km s-1, the use of a Gaussian to fit the profile may result in an under or overestimation of the EW (in a few percent), depending on the line strength. For those cases, the EW resulting from the integration of the observed line profile was preferred.

3.3 A grid of HHeOSi FASTWIND models

For this study, we constructed a grid of FASTWIND models with $T_{\rm eff}$ and log g ranging from 17 to 36 kK (1 kK steps) and 3.7 to 4.3 dex (0.1 dex steps). As the studied stars are not expected to be evolved, the He abundance was fixed to 0.09 dex. In addition, since FASTWIND is a spherically extended code, the radius and other wind parameters need to be indicated, and are grouped in the Q-parameter. We fixed this parameter to $\log Q= -15$ as a representative value for which the wind effect over the optical spectrum is practically negligible. (H$\alpha$ and He II$\lambda$4686 show no sign of wind contamination.) The metallicity was assumed to be solar (following the set of abundances by Grevesse & Sauval 1998).

For each pair of stellar parameters, a sub grid of models varying the microturbulence ( $\xi_{\rm t}$ = 1, 3, 5, 7, 9 km s-1), the Si abundance ( $\epsilon_{\rm Si}$ = -5.10, -4.80, -4.50, -4.20 dex), and the O abundance ( $\epsilon_{\rm O} = -4.00$, -3.65, -3.30, -2.95 dex) was calculated. The O and Si atomic models used for the grid came mainly from Becker & Butler (1988,1990). However, two updates were considered: (a) a extended Si II model atom (see Trundle et al. 2004), and (b) the most recent log gf values indicated in the atomic line list v2.05 for the formal solution calculations.

The final grid consists of $20 \times 7 \times 4$(=560) models and a total of 2800 formal solutions (5 microturbulence values per model). It includes line profiles and EWs for H , He I-II, Si II-IV and O II lines, along with the spectral energy distribution for each set of stellar parameters.

3.4 Determining stellar parameters

The use of the Si III-IV and/or Si II-III ionization equilibrium, along with the H Balmer lines, for determining the stellar parameters of early B-type stars is a longstanding method described elsewhere (see e.g. Kilian et al. 1991; Urbaneja et al. 2005; Crowther et al. 2006; Markova & Puls 2008). The ratios EW(Si IV $\lambda$4116)/EW(Si III $\lambda$4552) and/or EW(Si II $\lambda$4128)/EW(Si III $\lambda$4552), depending on the temperature of the star have been traditionally used as $T_{\rm eff}$ indicators. This decision has been probably motivated by the fact that these are the strongest, unblended lines in the spectral range commonly observed for the stellar abundance analysis (i.e. $\sim$4000-5000 Å).

We derived the stellar parameters, along with the Si abundance, for all stars in our sample using this methodology. We initially considered that these Si line ratios determine the effective temperatures (Cols. 3 and 4 in Table 3). However, motivated by a couple of problems found when deriving the stellar parameters in the cooler objects (see discussion below), we decide to include the EW(Si II $\lambda$6347)/EW(Si III $\lambda$4552) ratio as a temperature indicator (Col. 5). The measured values are indicated in brackets in the corresponding columns. As expected, the Si IV/Si III ratio decreases when we move to later spectral types, and the Si II/Si III behaves in the opposite way. For three of the stars (HD 36959, HD 35299, and HD 37744), lines from the three ions are clearly and simultaneously present in the spectra.

Four parameters need to be determined at the same time in an iterative way: $T_{\rm eff}$, log g, $\xi_{\rm t}$(Si), and $\epsilon_{\rm Si}$. First, we use the Si line ratios indicated above and the wings of the H Balmer lines (fixing $\xi_{\rm t}$(Si), and $\epsilon_{\rm Si}$) to obtain an initial guess for $T_{\rm eff}$and log g. Then we apply the curve of growth method to a proper set of Si II-IV lines to iteratively obtain final values for the four parameters (a detailed description of the used lines and the results of the Si abundance analysis is presented below). Normally, the final values of $T_{\rm eff}$and log g are quite close to the initial values, since the mentioned line ratios are only slightly dependent on $\epsilon_{\rm Si}$and $\xi_{\rm t}$.

3.4.1 Hotter objects ( $ T_{\mathsf {eff}}$$\ge $27 000 K)

In the hotter objects, Si II lines are not present in the spectra. The EW(Si IV $\lambda$4116)/EW(Si III $\lambda$4552) ratio is thus used to obtain the initial guess values of $T_{\rm eff}$for each of the log g values considered in the grid. Then, the full set of reliable Si III-IV lines, along with the H Balmer line profiles, is used for fine determination of the four parameters indicated above (see an example in the upper panel of Fig. 2, and in Fig. 3). The uncertainties in the EW measurements of the Si lines have been taken into account for establishing the uncertainties associated with the derived stellar parameters. Generally, the $T_{\rm eff}$and log g can be determined with an accuracy better than 500-600 K and 0.1 dex, respectively. The final results of the analysis are presented in Table 3.

\begin{figure}
\par\includegraphics[angle=0,width=9cm,clip]{13120f2.eps}
\vspace*{2mm}
\end{figure} Figure 2:

Silicon abundance vs. EW diagnostic diagrams for three of the analyzed stars (with representative effective temperatures).

Open with DEXTER

\begin{figure}
\par\includegraphics[angle=0,width=9cm,clip]{13120f3.eps}
\vspace*{2mm}
\end{figure} Figure 3:

Example of fitting the H Balmer lines to determine log g. Two FASTWIND models are compared with the observed spectrum (red line) of HD 36960: in black, model with the stellar parameters resulting from the FASTWIND HSi analysis (28 900, 3.9); in blue, model with the stellar parameters derived by Cunha & Lambert (1992), for comparison (28 900, 4.3).

Open with DEXTER

For main sequence stars with $T_{\rm eff}$ $\gtrsim$ 28 000 K the He I-II ionization equilibrium can also be used (see e.g. Herrero et al. 1992; Simón-Díaz et al. 2006). (Below this temperature, He II lines are too faint or not present in the spectrum.) We could determine the stellar parameters in this way for four of the stars in our sample. An example of this type of analysis has already been presented in Simón-Díaz et al. (2006), so we only present here the corresponding results (see Col. 6 in Table 3). In general, there is good agreement between the $T_{\rm eff}$determined through the Si III-IV and the He I-II ionization balance with differences in $T_{\rm eff}$ not larger than $\sim$500 K.

3.4.2 Cooler objects

For the cooler objects, the Si II/Si III must be used. We initially considered the EW(Si II $\lambda$4130)/EW(Si III $\lambda$4552) ratio for obtaining a first estimation of the stellar parameters; however, we found two facts that warned us of a possible problem with the Si II $\lambda$4130 (or the Si II $\lambda$4128) line. First, it was not possible to properly fit the H Balmer lines for any of the ( $T_{\rm eff}$, log g)-pairs indicated by the Si II $\lambda$4130/Si III $\lambda$4552 line ratio, because the core of the H  synthetic lines were somewhat narrower than the observed ones (even if high values of log g are considered). For these temperatures, the cores of the H Balmer lines begin to be sensitive to changes in $T_{\rm eff}$. For all the cases studied, these lines require somewhat lower effective temperatures to properly fit their cores. Second, the $\epsilon_{\rm Si}$-EW diagnostic diagrams show two sets of Si II lines giving different abundances by $\sim$0.2 dex (see lower panel in Fig. 2). Curiously, each subset of lines corresponds to transitions with very different energy levels (Si II $\lambda$$\lambda$4128, 4130, 5041, 5056, in one hand, coming from higher energy levels; Si II $\lambda$$\lambda$6347, 6371, 3856, 3862, on the other, comming from lower energy levels; we suggest the reader have a Grotrian diagram in hand). This different behavior of the various lines warned us of possible problems with the Si II atomic model and of using Si II $\lambda$4128 line alone to establish the stellar temperature.

Columns 4 and 5 in Table 3 show the different effective temperatures obtained depending on the Si II line that is used (Si II $\lambda$4128, Si II $\lambda$6371). Temperatures given by the EW(Si II $\lambda$6371)/EW(Si III $\lambda$4552) ratio are systematically lower by up to $\sim$2000-3000 K. In addition, a proper fit of the cores of the H Balmer lines can be achieved with these new values of the temperature.

3.4.3 HD 36595, HD 37744, and HD 35299

These three stars show Si II-III-IV lines in their spectra, so can contribute with new decisive clues to the problem mentioned above. In this case we count with three $T_{\rm eff}$ indicators. As can be noticed from inspection of Table 3, the $T_{\rm eff}$ indicated by the EW(Si II $\lambda$6371)/EW(Si III $\lambda$4552) ratio results in better agreement than the other Si II/Si III ratio for two of the three cases. For the hotter object (HD 36959), the Si II $\lambda$6371 is too faint to be measured. Although the $T_{\rm eff}$indicated by the EW(Si II $\lambda$4130)/EW(Si III $\lambda$4552) ratio agrees with the Si IV/Si III diagnostic in this object, this is not the case for the other objects.

In Fig. 2 (middle panel) we show the $\epsilon_{\rm Si}$-EW diagnostic diagram for HD 37744. The stellar parameters considered in this plot are those indicated by the Si IV/Si III ratio. The Si II $\lambda$$\lambda$6371, 3856 lines fit the other Si III-IV lines perfectly, but not the Si II $\lambda$$\lambda$4128, 4130 lines.

3.4.4 Concluding

We found some indications of a problem in the Si II model atom that led to bad modeling of the Si II $\lambda$$\lambda$4128, 4130, 5041, 5056 lines. Fortunately, our observations also include other Si II lines that seem to behave properly. A review of the Si II model atom is needed (and could be tested with the type of detailed analysis presented here), but in the meantime, several arguments allow us to trust[*] the Si II $\lambda$$\lambda$6347, 6371, 3856, 3862 set of lines for the stellar parameter determination and for abundance analysis of the cooler objects in our sample: (a) the coherence of results between Si IV-III and Si III-II in terms of $T_{\rm eff}$; (b) the good fit of the H Balmer lines for the cooler stars when the lower $T_{\rm eff}$ is considered; (c) the coherence of results in Si and O abundances that is obtained for all the stars in our sample when this solution is adopted (see below).

3.4.5 Uncertainties

Columns 3 and 5 indicate the uncertainties in $T_{\rm eff}$obtained from considering errors in the EWs measurement of Si II-IV lines. These uncertainties, obtained by assuming a fixed gravity are $\sim$100-600 K. Gravity can be normally established with an accuracy better 0.1 dex. Uncertainties from both quantities are correlated, e.g. a positive variation of log g of 0.1 dex needs to be compensated by an increase of $T_{\rm eff}$$\sim$ 100-300 K to guarantee again the Si ionization balance. Although in many cases the formal errors in temperature obtained from the propagation of EWs errors are smaller than 500 K, our experience warns us to be conservative (since other parameters can also slightly affect the derived temperatures, such as the considered microturbulence). Therefore, we adopt 500 K and 0.1 dex as characteristic uncertainties in $T_{\rm eff}$and log g, respectively, from our analyses.

3.5 Silicon and oxygen stellar abundances

We applied the curve-of-growth method to derive Si and O abundances. This method considers a grid of models for a given set of stellar parameters ( $T_{\rm eff}$and log g) in which the microturbulence and the abundance of the element to be studied are varied. An abundance is obtained for the various values of microturbulence for each of the considered lines (given the measured EW of the lines in the observed spectrum). Then, the final abundance is given by that microturbulence that results in all lines giving the same abundance. More details on this method can be found in e.g. Kilian (1992) or Simón-Díaz et al. (2006).

Our preference for this methodology in performing the abundance analysis is that we find the curve-of-growth method very powerful for identifying problematic lines (as shown e.g. in the previous section), which can affect the final abundance determination, and provide precise estimations of the uncertainties associated with the dispersion of line-to-line abundances, microturbulence, and the stellar parameters.

3.5.1 Silicon abundances

As mentioned, silicon abundance is obtained at the same time as to the stellar parameters. Once a first estimation of the stellar parameters was obtained through the EW(Si IV $\lambda$4116)/EW(Si III $\lambda$4552) and/or EW(Si II $\lambda$6371)/ EW(Si III $\lambda$4552) line ratios, the curve-of-growth method was applied to a proper set of Si lines to derive the Si abundance, together with the final values of the stellar parameters and a microturbulence.

Initially, we included all the measured lines in the analysis, but identified some problematic lines. By problematic lines we mean lines giving abundances that are too high or too low compared with the mean value provide by a set of lines initially considered as reliable. These problematic lines were removed in the final analysis. To illustrate the procedure, we consider the case of HD 36960. The $\epsilon_{\rm Si}$-EWdiagram with whole set of measured Si lines (Si III and Si IV lines, in this case) is shown in Fig. 4. As the initial set of reliable lines we consider the Si III $\lambda$$\lambda$4552, 4567, 4574 triplet and the Si IV $\lambda$4116 line. The other Si IV lines provide similar abundances to the first set of reliable lines, except Si IV $\lambda$6701. The situation is less encouraging for the other Si III lines, since all of them lie below the mean abundance value from reliable lines. We thus label these lines as problematic. The same procedure was followed for the selection of reliable Si II lines (see Sect. 3.4).

\begin{figure}
\par\includegraphics[angle=0,width=8.5cm,clip]{13120f4.eps}
\vspace*{2mm}
\end{figure} Figure 4:

Silicon abundance vs. EW diagnostic diagrams for HD 36960. All the observed Si III-IV lines are included in the plot. Uncertainties in the individual line abundances (propagated from the errors in the measured EWs) are indicated as vertical lines.

Open with DEXTER

A similar behavior is found for the problematic lines in all the analyzed stars. Since the EW of these lines are accurately measured and no lines from other elements are expected to be affecting them[*], we argue that the discrepancy may be related to the definition of the model atom. For example, the Si III $\lambda$$\lambda$4813, 4819, 4829 triplet is known to give different results than the Si III $\lambda$$\lambda$4552, 4567, 4574 triplet because of the boundary problems of the Si III model atom (Becker & Butler 1990). Normally, there is an explanation for the bad behavior of the problematic lines; therefore, the exclusion of lines from the analysis is not arbitrary (e.g. lines from the same multiplets behave in the same way and are normally excluded all together).

It is important to notice that the analysis with the whole set of lines (including the problematic ones) could lead to an incorrect determination of the microturbulence, hence of the final value of Si abundance.

Final results of the Si abundance analysis[*] are indicated in Table 3 (Cols. 9 and 10). Three sources of errors must be considered for estimating of the final uncertainty: (1) the dispersion in the line-by-line abundances; (2) the uncertainty associated with the microturbulence, (indicated in Table 3, and Tables 6-18, only available in the electronic version); and (3) the contribution of uncertainties in the stellar parameters. These can be added quadratically to obtain the total uncertainty.

Table 4:   Final results of the O abundance analysis. Detailed results (line-by-line) of the O abundance analysis are presented in Tables 6 to 18.

For illustrative purposes in the last column of Table 3, we indicate the effect of a change of $\pm$500 K in $T_{\rm eff}$ on the Si abundance if this was derived exclusively by using Si III lines[*]. The minimum effect occurs for $T_{\rm eff}$ $\sim$ 27 000 K and increases towards higher and lower temperatures (note the change in the sign of the uncertainties), reaching values up to 0.12 dex. This behavior is a consequence of the dependence of the EW of the Si III lines with temperature. The maximum EW is achieved around 27 000 K (so the abundance is quite insensitive to small changes in $T_{\rm eff}$), and decreases towards lower and higher temperatures (also increasing its sensitivity to $T_{\rm eff}$ variations). When lines from two different ions are used, the final Si abundance can be constrained with better accuracy, since the EW of lines from different ions behaves in a opposite way for a given $T_{\rm eff}$. As a consequence in this case, the uncertainty associated with the stellar parameters is always negligible compared with the dispersion in the line-by-line abundances.

3.5.2 Oxygen abundances

Up to 47 O II lines were identified in the observed spectral range. However, not all the lines were finally used for the O abundance determination. The selection of the final set of lines used for the analysis was based on a detailed analysis by multiplets (following similar criteria to the case of Si, but in this case we studied the behavior of lines resulting from different multiplets). We found a similar behavior of lines to the one described in Simón-Díaz et al. (2006).

The derived O abundances are indicated in Table 4, and the results of the line-by-line analyses in Tables 6-18 (in the electronic version). For this element, only lines from one ionization state were available, so we could not test whether the corresponding ionization equilibrium (O III/II or O II/I) is achieved for the considered stellar parameters. We adopt for the oxygen abundance analysis the same stellar parameters as in the Si analysis.

We found (as in many previous works) that the microturbulence derived from the O II lines ( $\xi_{\rm t}$(O)) differs from the Si analysis ( $\xi_{\rm t}$(Si)), and a somewhat larger microturbulence is derived. Some authors assume the $\xi_{\rm t}$(Si) value (or a mean value of the microturbulences obtained for the various elements analyzed) to perform the oxygen abundance analysis. In our opinion, this can lead to significant systematic errors in the analysis. Since this is an ad-hoc parameter that is still not well understood (see, however, Cantiello et al. 2009), we adopt the microturbulence derived from the oxygen analysis itself, for consistency.

In fact, determination of the microturbulence value that will be adopted in the final steps of the abundance determination is an important task. Unidentified problems in the O II line modeling or bad measurements of the corresponding EWs (due to blends, noise, or a bad placement of the continuum) can enormously affect the $\xi_{\rm t}$value that produces a zero slope in the $\epsilon_{\rm O}$-EW diagrams. A detailed analysis by multiplets (see Simón-Díaz 2005) can help identify problematic lines and to better decide on the final microturbulence to be adopted.

Table 4 also indicates the uncertainties associated with errors in the line-to-line abundance dispersion, the microturbulence, and the stellar parameters. In addition, last column shows the derived abundances if the effective temperature is varied $\pm$500 K. (Although the exact values are not shown here, the contribution of the log g uncertainty to the oxygen abundance can be considered negligible in comparison with the $T_{\rm eff}$ contribution.) The derived oxygen abundance is very sensitive to changes in $T_{\rm eff}$ for the cooler and hotter objects, and there is a change in the behavior of the oxygen abundance with $T_{\rm eff}$ around 27 000 K. This behavior is similar to the one illustrated in Table 3.

4 Comparison with previous works

Two of the stars included in this analysis have already been analyzed in Simón-Díaz et al. (2006). These are HD 37020 and HD 37042. We observed the two stars again to have spectra with the same characteristics as the other stars in the sample. The stellar parameters presented in Table 2 correspond to the analysis of the new spectra. The new analysis resulted in slightly higher effective temperatures (but within the errors) and gravities $\sim$0.1-0.2 dex larger. Several factors produced this difference in the derived stellar parameters. First, we based our $T_{\rm eff}$ - log g determination on the HSi analysis, instead of using the H and He I-II lines. The best $T_{\rm eff}$ - log g pair reproducing the Si III-IV ionization equilibrium and the wings of the H lines simultaneously is the one indicated in Table 3. A lower log g requested an effective temperature somewhat lower (which was not fitting the He II lines). In addition, the better quality of the new spectra allowed us to better constrain the gravity of the stars. The wings of the H lines are not very sensitive to changes of $\sim$0.1-0.2 dex in log g in this range of stellar parameters. In Simón-Díaz et al., we based our decision about the best solution on the faint He II $\lambda$4541 line; however, it is better to rely on the HSi criterion, which is more sensitive in this range of stellar parameters. In fact, we find that the new $T_{\rm eff}$ - log g pair also fits nicely the He I-II lines. We thus prefer this last solution.

\begin{figure}
\par\includegraphics[angle=90,width=17cm,clip]{13120f5.eps}
\end{figure} Figure 5:

Comparison of our derived stellar parameters, Si and O abundances, with those obtained by Cunha & Lambert (1994,1992). Results from both studies for each star are connected by a dashed line. Numbers follow the order of stars presented in Tables 3 and 4. The size of the uncertainties of the various quantities are shown by the crosses.

Open with DEXTER

We also obtain $\sim$0.05-0.10 dex higher oxygen abundances. Part of this difference is caused by the change in stellar parameters; in addition, we measured sligthly larger EWs for the O II lines in both stars. Note, however, that the old and new abundances agree when taking the error bars into account. The discrepancy for these two objects between Simón-Díaz et al. (2006) and these new studies serves as an example of the effect of intrinsic uncertainties in determination of stellar abundances. Following a similar methodology the same author can find slightly different results (but within the errors) when analyzing spectra from two different observing campaigns[*].

Recently, Przybilla et al. (2008) analyzed a representative sample of unevolved early B-type stars in nearby OB associations and the field using a similar technique but a different stellar atmosphere code (line blanketed ATLAS9 LTE model atmospheres, Kurucz 1993) and NLTE line-formation calculations (with updated versions of DETAIL and SURFACE, Giddings 1981; Butler & Giddings 1985). They have one star in common with our work, HD 36591 (HR 1861). We obtain very similar results for the stellar parameters, as well as the Si and O abundances. They used several $T_{\rm eff}$ spectroscopic indicators (apart from Si II-IV), finding very good overall agreement.

Gummersbach et al. (1998) included a sample of 5 stars from Ori OB1 in their study of the abundance gradient of the Galaxy. We have three stars in common. Gummersbach et al. used a self-consistent spectroscopic approach. We find similar effective temperatures for the two hotter objects (when Si IV lines are available), but $\sim$0.2 dex lower gravities. For these stars (HD 36959 and HD 36960), the derived Si and O abundances agree (within the errors) with our values. The difference in gravity could explain their slightly lower O abundances.

Interestingly, for the cooler object in common (HD 35039), we obtained lower $T_{\rm eff}$and log g. It is remarkable that they obtained very low O and Si abundances (8.20 and 7.08 dex, respectively) for this star. Probably, this result is related to the Si II problem we described in Sect. 3.4 (see also lower panel in Fig. 2); they used line Si II 4130 to establish $T_{\rm eff}$, so obtained too high a value (23 500 K vs. 19 900 K, the value we obtained). For this range of stellar parameters, higher $T_{\rm eff}$ supposes lower O abundances, and a change of $\sim$3000 K in $T_{\rm eff}$can perfectly explain a 0.5 dex variation in the O abundance.

Figure 5 presents a comparison our results with those obtained by Cunha & Lambert (1994,1992) concerning the stellar parameters, Si and O abundances. There is a clear discrepancy for the majority of the objects, in terms not only of effective temperatures, but also of gravities. Cunha & Lambert (1994,1992) made use of the calibrations of Strömgen photometry coupled with the fits to the pressure-broadened line wings of H$\gamma$ (from Kurucz 1979, LTE stellar atmopsheres) to derive $T_{\rm eff}$ and log g. Following Gies & Lambert (1992), they adjusted the $T_{\rm eff}$calibrations by Lester et al. (1986) and Balona (1984) upward by 4.2% and 5.2% , respectively. The discrepancy in effective temperatures obtained through commonly used photometric calibrations and spectroscopic line diagnostic has already been pointed out by several authors (e.g. Nieva & Przybilla 2008; Kilian et al. 1991). Our result once more brings out these discrepancies.

Note also the large discrepancy found for the gravity. The Cunha & Lambert values are systematically higher ($\sim$0.3 dex). As shown by Nieva & Przybilla (2007), the use of LTE profiles for the gravity determination lead to overestimated log g values, in particular for hotter stars. In Fig. 3 we show a comparison of synthetic hydrogen lines from two FASTWIND models with the observed profiles for HD 36960 (labeled with #3 in Fig. 5. One of the models considers the stellar parameters derived from our analysis, the other one corresponds to the $T_{\rm eff}$-log g pair provided by Cunha & Lambert (1992). The difference in the wings of the lines is clear.

We also find big discrepancies in the derived Si and O abundance for most of the stars, with no systematic trend (i.e. differences are found in both positive and negative directions). There are several factors to take into account to explain this disagreement. One of them is the imprecise determination of the stellar parameters from photometric indices. But maybe the most important one is the following. Once the stellar parameters were calculated as described above, Cunha & Lambert use those values to perform the abundance analysis, computing line-blanketed LTE model atmospheres with ATLAS6 (Kurucz 1979) for the LTE case, or using a grid of EWs based on Gold's models (Gold 1984) for the NLTE abundance calculations. The risk in this strategy is that the stellar abundance analysis is decoupled of the stellar parameter determination, and this could produce inconsistencies in the analysis process. Briefly, the calculation of EWs of those lines used for the abundance determination is based on the stellar atmosphere structure defined by a stellar atmosphere model computed for a given set of stellar parameters; on the other hand, the stellar parameters derived for a given star will depend on the characteristics of the stellar atmosphere model we have used[*]. Therefore, it may be dangerous to do an abundance analysis with a given stellar atmosphere code using the stellar parameters obtained from a different code or a photometric calibration. This argument is crucial in the case of stars with $T_{\rm eff}$ $\ge $ 30 000 K, where photometric methods become completely unreliable discriminators of temperatures and gravities, because of the insensitivity of the Rayleigh-Jeans tail of the spectral energy distribution on temperature.

A self-consistent spectroscopic approach allows minimizing this problem. In this case, the stellar parameters are determined by fitting certain spectroscopic diagnostics, and then the same models are used for the abundance analysis. This way we are certain that the stellar atmosphere structure used for the computation of the abundance diagnostics is coherent with the derived stellar parameters for the studied star. To a first order, when the whole analysis is performed with stellar atmosphere codes with different characteristics, the derived abundances should be quite similar (although the derived stellar parameters could be somewhat different). This approach can be strengthened if multiple independent spectroscopic indicators are considered (i.e. Si III-IV, He I-II, C II-IV; see e.g. this study or Nieva & Przybilla 2008).

5 Chemical composition of B-type stars in the Ori OB1 association

Figures 6 and 7 again show the derived oxygen and silicon abundances in our sample of B-type stars in the Ori OB1 association, along with the results by Cunha & Lambert (1994). This time, the stars are ordered following the subgroups suggested by Blaauw (1964). As discussed in Cunha & Lambert (1994), they found that the stars in the youngest group (Id) and some of the stars in subgroup Ic were enriched in oxygen by about 40% relative to the stars belonging to the older subgroups. They interpreted this result as possible proof of enrichment of the new generation of stars in the association with the products from supernovae ejecta from the older subgroups. In addition, they found features of this enrichment in silicon (also expected from the triggered star formation scenario). In contrast, we did not find any systematic difference between the O and Si abundances in stars from the various associations. In fact, our results indicate that the B-type stars in the Ori OB1 association are chemically homogeneous (at least in terms of oxygen and silicon), having a dispersion in abundances (0.04 and 0.03 dex, respectively) smaller than the intrinsic uncertainties of the derived abundances (0.10 and 0.08 dex, respectively). The mean abundances are $\epsilon$(O) = 8.73 dex and $\epsilon$(Si) = 7.51 dex. These values agree with those obtained by Przybilla et al. (2008) for their sample of six stars in the solar neighborhood.

\begin{figure}
\par\includegraphics[angle=90,scale=.53]{13120f6.eps}
\end{figure} Figure 6:

Oxygen abundances derived for our sample of early B-type stars in the Orion association, and comparison with the previous results from Cunha & Lambert (1994). Vertical lines separates stars from the various subgroups. Solid and dashed horizontal lines represent the mean value and the dispersion (1$\sigma $) of our results; dot-dashed horizontal lines indicate the characteristic intrinsic uncertainty of the derived abundances (accounting for uncertainties in stellar parameters, microturbulence, and the line-by-line abundance dispersion). The oxygen abundance derived by Esteban et al. (2004) for M 42, along with the values determined for the Solar abundance in the past 4 years (the epoch of the ``solar crisis''), are also presented for comparison.

Open with DEXTER

\begin{figure}
\par\includegraphics[angle=90,scale=.53]{13120f7.eps}
\end{figure} Figure 7:

As Fig. 2 but for silicon. This time, the reference for the nebular abundance is Rubin et al. (1993) and Garnett et al. (1995) (first and second column, see text). Solar values from Asplund et al. (2005) and Grevesse & Sauval (1998).

Open with DEXTER

5.1 Comparison with the Sun

We also include in Fig. 6 the resulting solar oxygen abundances appearing in the literature in the past 4 years (during the so-called solar oxygen crisis[*] epoch, not yet finished). Various authors have published values for the solar oxygen abundance based on improved model atmospheres (either 1D and 3D), line formation codes, atomic and molecular data, and detailed treatment of blends. In the plot, we present results by Centeno & Socas-Navarro (2008), Meléndez & Asplund (2008), Ayres (2008), Caffau et al. (2008), Allende Prieto (2008), Socas-Navarro & Norton (2007), Ayres et al. (2006), Asplund et al. (2004), and Grevesse & Sauval (1998). This last value (marked with an internal black dot) was considered as the standard solar O abundance until a few years ago. The derived solar values range from 8.63 dex (Socas-Navarro & Norton 2007) to 8.86 dex (Centeno & Socas-Navarro 2008). The O abundances in our sample of stars in Ori OB1 lie in the middle of all these values. In view of the present-day results, the only thing we can say is that oxygen abundances in the Sun and B-type stars in the solar vicinity are the same within the uncertainties. However, we consider it too premature to draw any firm conclusion or hypothesis about the chemical evolution of the local interstellar medium during the lifetime of the Sun.

Figure 7 also shows a comparison of our derived Si abundances with the Solar value. We only present here the old abundance (Grevesse & Sauval 1998) and the new value proposed by Asplund et al. (2005). In contrast to previous results, the mean Si abundance in B-type stars in Ori OB1 is very close to the Solar value (similar to other B-type stars in the solar vicinity Przybilla et al. 2008).

5.2 Comparison with the Orion nebula

The most recent and detailed analysis of the optical spectrum of the Orion nebula was done by Esteban et al. (2004). The high-quality UVES@VLT spectrum they used allowed them to derive the oxygen gas phase abundance of the nebula by using collisionally excited lines (CELs) and recombination lines (RLs). The final value they propose is $8.65 \pm 0.03$ dex. This value was calculated assuming the ionic abundances given by O 2+ RLs and O + CEL (plus a t2 = 0.022)[*]. Note, however, the abundance given by the O 2+ and O + CELs (and t2=0) is $8.51 \pm 0.03$ dex. These values have been included in Fig. 6 for comparison with the early B-type stellar abundances presented in this study. The mean value of the derived stellar oxygen abundances is 0.25 and 0.11 dex higher than the nebular abundances given by the CELs and RLs, respectively.

Although the mean stellar abundance seems to agree better with the one given by the faint recombination lines, we have to consider that the analysis of the nebular emission line spectrum can only provide abundances for the ionized gas phase of the ISM. This means a lower limit to the actual ISM abundance, since part of the oxygen can be depleted, forming part of the dust grains.

Silicon is one of the elements expected to be more depleted onto dust grains, along with Mg and Fe (Draine 2003). We can therefore compare our derived Si abundances with those obtained from the study of the emission line spectrum of the Orion nebula to try to find some clues about the amount of oxygen depleted. One should note that the determination of Si nebular abundances is not straightforward and must be based on results from photoionization models and certain ionization correction factors (ICFs). The only determinations of Si abundance in the Orion nebula we found in the literature are those by Rubin et al. (1993) and Garnett et al. (1995). Both studies used the same observations of the FUV Si III $\lambda\lambda$1883, 1892, and C III$\lambda\lambda$1907, 1909 lines to estimate the Si abundance, and obtained 6.65 and 6.58 dex, respectively. They computed photoionization models to obtain Si/H (Rubin et al. 1993) and the ICF needed to transform Si 2+/C 2+ into Si/C, hence the C/O and O/H ratios to derive Si/H (Garnett et al. 1995). In addition, Garnett et al. (1995) discuss how the effect of assuming a t2 = 0.04 (Peimbert et al. 1993) in the calculation of the C/O and O/H ratios could affect the derived Si abundance. They obtained log (Si/H)+12 = 7.14 dex for this case. These three values of the Si abundance are included in Fig. 7 for comparison with the stellar abundances. The difference between stellar and gas phase Si abundances are $\sim$1 dex, or $\sim$0.3 dex, depending on the nebular abundances we trust more.

Since a detailed comparison of stellar and nebular abundances in Ori OB1 and the Orion nebula within a dust-depletion scenario requires more extended study, which is beyond the scope of this paper, we decide to present a more detailed discussion in a separate paper (in preparation).

6 Summary and conclusion

In this work, we applied a self-consistent spectroscopic approach to determining of the stellar parameters, the Si and O abundances of a sample of early-B type stars from the various subgroups of the Ori OB1 association. We made use of a high-quality spectroscopic data set, obtained with FIES@ NOT, and the modern NLTE, line-blanketed, spherically extended stellar atmosphere code FASTWIND.

We developed several IDL programs to automatically identify and measure the EW of metal lines in high-resolution spectra. We also constructed a grid of FASTWIND HHeSiO models optimized for the analysis of early B-type, main sequence stars.

The availability of a large number of Si II-III-IV lines in the FIES@ NOT spectra allowed us to obtain the stellar parameters with high accuracy and detect some problems related to some Si lines commonly used for the stellar parameter and Si abundance determination. Once these problems were accounted for, a high degree of homogeneity was found for Si abundances in the analyzed sample of stars.

The oxygen abundance analysis also result in a small dispersion of abundances, in contrast to previous determinations. The mean oxygen and silicon abundances agree with those resulting from a similar analysis of a representative sample of unevolved early B-type stars in the solar vicinity (Przybilla et al. 2008). Both results indicate that abundances derived from these stellar objects are more homogeneous and metal-rich than previously though.

We also compared the O and Si stellar abundance in Ori OB1 with those obtained for the Sun during the epoch of the ``solar crisis''. The O abundances in our sample of stars in Ori OB1 lie in the middle of all these values. In view of the present-day results, the only thing we can say is that oxygen abundances in the Sun and B-type stars in the solar vicinity are the same within the uncertainties. However, we consider it too premature to draw any firm conclusion or hypothesis about the chemical evolution of the local interstellar medium during the lifetime of the Sun. Silicon abundances are also very similar (contrary to what was previously found from the study of B-type stars).

Finally, we compared the stellar abundances with those derived from the study of the emission line spectrum of the Orion nebula. In a forthcoming paper we will present a more detailed discussion accounting for the possible depletion of O, Si, Mg, and Fe into dust.

This work points out one more time the importance of self-consistent spectroscopic abundance analyses for determining of the chemical composition of the photospheres of OB-type stars. Photometric $T_{\rm eff}$diagnostics must be treated with caution in the context of abundance analyses of these type of objects. It is also dangerous to combine the stellar parameters determined with any given code and the abundance analysis performed with a different code. Systematic errors inherent in those techniques, or possible biases between the various codes, can lead to incorrect abundances.

Acknowledgements
Financial support by the Spanish Ministerio de Ciencia e Innovación under the project AYA2008-06166-C03-01. This work has also been partially funded by the Spanish MICINN under the Consolider-Ingenio 2010 Program grant CSD2006-00070: First Science with the GTC (http://www.iac.es/consolider-ingenio-gtc). I am very grateful to A. Herrero (Spain), G. Stasinska (France), and D. Schaerer (Switzerland) for their support and hospitality during the development of this work. I also acknowledge A. Herrero, M. Urbaneja, D. Lennon, F. Najarro, C. Trundle, and N. Castro for fruitful discussions. J. Puls, N. Przybilla, F. Nieva, and M. Urbaneja for the careful reading of the first version of this paper and all their comments. Finally, I thank J. Puls for allowing me to use the stellar atmosphere code FASTWIND.

References

Online Material

Table 5:   List of log gf values for the Si II-IV lines considered in this study (from Atomic Line List v2.05). For the O II lines we refer to Simón-Díaz et al. (2006).

Table 6:   Results from the abundance analysis of HD 36512 (B0 V).

Table 7:   Results from the abundance analysis of HD 37020 (B0.5 V).

Table 8:   Results from the abundance analysis of HD 36960 (B0.5 V).

Table 9:   Results from the abundance analysis of HD 37042 (B0.7 V).

Table 10:   Results from the abundance analysis of HD 36591 (B1 V).

Table 11:   Results from the abundance analysis of HD 36959 (B1 V).

Table 12:   Results from the abundance analysis of HD 37744 (B1.5 V).

Table 13:   Results from the abundance analysis of HD 35299 (B1.5 V).

Table 14:   Results from the abundance analysis of HD 36285 (B2 V).

Table 15:   Results from the abundance analysis of HD 35039 (B2 V).

Table 16:   Results from the abundance analysis of HD 36629 (B2 V).

Table 17:   Results from the abundance analysis of HD 36430 (B2 V).

Table 18:   Results from the abundance analysis of HD 35912 (B2 V).

\begin{figure}
\par\includegraphics[angle=180,scale=0.85]{13120f8a.eps}
\end{figure} Figure 8:

The complete atlas of FIES@ NOT spectra (part 1 of 3). The Si II-IV and O II lines used for the abundance analysis are indicated as solid and dashed vertical lines, respectively.

Open with DEXTER

\begin{figure}
\par\includegraphics[angle=180,scale=0.85]{13120f8b.eps}
\end{figure} Figure 8:

continued.

Open with DEXTER

\begin{figure}
\par\includegraphics[angle=180,scale=0.85]{13120f8c.eps}
\end{figure} Figure 8:

continued.

Open with DEXTER

Footnotes

... stars[*]
Based on observations made with the Nordic Optical Telescope, operated jointly on the island of La Palma by Denmark, Finland, Iceland, Norway, and Sweden, in the Spanish Observatorio del Roque de los Muchachos of the Instituto de Astrofisica de Canarias.
...[*]
Tables 5 to 18, and Fig. 8 are available in electronic form at http://www.aanda.org
...$1.7 \pm 1.1$[*]
Briceño et al. (2005) determined an age for Ori OB1b $\sim$4-6 Myr by studying its low mass, young stellar population; this value is more consistent with what would be expected from the presence of the evolved blue supergiant $\epsilon$ Ori in this subgroup.
... abundances[*]
Other authors (Gies & Lambert 1992; Gummersbach et al. 1998; Kilian 1992) obtained a similar range in oxygen abundances from the analysis of smaller samples of B-type stars in Orion OB1.
... stars[*]
HD 37023 was also re-observed, but it was not included in the analysis because we found in the new spectrum that the star is is actually a binary system (SB2), hence not optimal for an abundance analysis.
... FIEStool[*]
http://www.not.iac.es/instruments/fies/fiestool/FIEStool.html
... atomic[*]
http://www.pa.uky.edu/ peter/newpage/
... trust[*]
This hypothesis is also supported by a comparison of $T_{\rm eff}$ obtained for the cooler objects in our sample from other spectroscopic diagnostic based on O II/I and C III/II ratios of lines (F. Nieva, private communication).
... them[*]
This is not the case for Si IV $\lambda$$\lambda$4089, 4631, and Si II $\lambda$$\lambda$6347, 3856; these lines may be blended with O II $\lambda$4089, N II $\lambda$4631, Mg II $\lambda$6347, and O II$\lambda$3856, respectively, in some cases. The high resolution of the FIES@ NOT spectra normally allows separating both line contributions; however, results with these lines are always treated with care.
... analysis[*]
Results of the line-by-line analysis are presented in Tables 6-18.
... lines[*]
A similar test assuming a change of 0.1 dex in log g indicates negligible effects when compared with the $T_{\rm eff}$ contribution.
... campaigns[*]
The different results may also be caused by actual changes in the stellar spectra (e.g. due to binarity).
... used[*]
As an example, we want to mention the consequences that including the line blanketing and wind blanketing effects in the stellar atmosphere models of O and early B-type stars had on the SpT -  $T_{\rm eff}$ calibrations of these stars (see e.g. Repolust et al. 2004; Martins et al. 2005).
... crisis[*]
Ayres et al. (2006)
...[*]
The t2 parameter was introduced by Peimbert (1967) to account for temperature fluctuations in ionized nebulae. The temperature fluctuation scenario has been proposed to explain the CELs vs. RL abundance discrepancy (see García-Rojas & Esteban 2007, and references therein).
Copyright ESO 2010

All Tables

Table 1:   List of B-type stars from Ori OB1 considered in this study, ordered by spectral type.

Table 2:   Projected rotational velocities derived for the studied stars.

Table 3:   Final results of the HSi analysis: stellar parameters and Si abundances.

Table 4:   Final results of the O abundance analysis. Detailed results (line-by-line) of the O abundance analysis are presented in Tables 6 to 18.

Table 5:   List of log gf values for the Si II-IV lines considered in this study (from Atomic Line List v2.05). For the O II lines we refer to Simón-Díaz et al. (2006).

Table 6:   Results from the abundance analysis of HD 36512 (B0 V).

Table 7:   Results from the abundance analysis of HD 37020 (B0.5 V).

Table 8:   Results from the abundance analysis of HD 36960 (B0.5 V).

Table 9:   Results from the abundance analysis of HD 37042 (B0.7 V).

Table 10:   Results from the abundance analysis of HD 36591 (B1 V).

Table 11:   Results from the abundance analysis of HD 36959 (B1 V).

Table 12:   Results from the abundance analysis of HD 37744 (B1.5 V).

Table 13:   Results from the abundance analysis of HD 35299 (B1.5 V).

Table 14:   Results from the abundance analysis of HD 36285 (B2 V).

Table 15:   Results from the abundance analysis of HD 35039 (B2 V).

Table 16:   Results from the abundance analysis of HD 36629 (B2 V).

Table 17:   Results from the abundance analysis of HD 36430 (B2 V).

Table 18:   Results from the abundance analysis of HD 35912 (B2 V).

All Figures

  \begin{figure}
\par\includegraphics[angle=90,scale=0.62]{13120f1.eps}
\end{figure} Figure 1:

Example of FIES@ NOT spectra for three of the observed stars. The complete atlas is available in the electronic version of the paper (Fig. 8). The Si II-IV and O II lines used for the abundance analysis are indicated as solid and dashed vertical lines, respectively.

Open with DEXTER
In the text

  \begin{figure}
\par\includegraphics[angle=0,width=9cm,clip]{13120f2.eps}
\vspace*{2mm}
\end{figure} Figure 2:

Silicon abundance vs. EW diagnostic diagrams for three of the analyzed stars (with representative effective temperatures).

Open with DEXTER
In the text

  \begin{figure}
\par\includegraphics[angle=0,width=9cm,clip]{13120f3.eps}
\vspace*{2mm}
\end{figure} Figure 3:

Example of fitting the H Balmer lines to determine log g. Two FASTWIND models are compared with the observed spectrum (red line) of HD 36960: in black, model with the stellar parameters resulting from the FASTWIND HSi analysis (28 900, 3.9); in blue, model with the stellar parameters derived by Cunha & Lambert (1992), for comparison (28 900, 4.3).

Open with DEXTER
In the text

  \begin{figure}
\par\includegraphics[angle=0,width=8.5cm,clip]{13120f4.eps}
\vspace*{2mm}
\end{figure} Figure 4:

Silicon abundance vs. EW diagnostic diagrams for HD 36960. All the observed Si III-IV lines are included in the plot. Uncertainties in the individual line abundances (propagated from the errors in the measured EWs) are indicated as vertical lines.

Open with DEXTER
In the text

  \begin{figure}
\par\includegraphics[angle=90,width=17cm,clip]{13120f5.eps}
\end{figure} Figure 5:

Comparison of our derived stellar parameters, Si and O abundances, with those obtained by Cunha & Lambert (1994,1992). Results from both studies for each star are connected by a dashed line. Numbers follow the order of stars presented in Tables 3 and 4. The size of the uncertainties of the various quantities are shown by the crosses.

Open with DEXTER
In the text

  \begin{figure}
\par\includegraphics[angle=90,scale=.53]{13120f6.eps}
\end{figure} Figure 6:

Oxygen abundances derived for our sample of early B-type stars in the Orion association, and comparison with the previous results from Cunha & Lambert (1994). Vertical lines separates stars from the various subgroups. Solid and dashed horizontal lines represent the mean value and the dispersion (1$\sigma $) of our results; dot-dashed horizontal lines indicate the characteristic intrinsic uncertainty of the derived abundances (accounting for uncertainties in stellar parameters, microturbulence, and the line-by-line abundance dispersion). The oxygen abundance derived by Esteban et al. (2004) for M 42, along with the values determined for the Solar abundance in the past 4 years (the epoch of the ``solar crisis''), are also presented for comparison.

Open with DEXTER
In the text

  \begin{figure}
\par\includegraphics[angle=90,scale=.53]{13120f7.eps}
\end{figure} Figure 7:

As Fig. 2 but for silicon. This time, the reference for the nebular abundance is Rubin et al. (1993) and Garnett et al. (1995) (first and second column, see text). Solar values from Asplund et al. (2005) and Grevesse & Sauval (1998).

Open with DEXTER
In the text

  \begin{figure}
\par\includegraphics[angle=180,scale=0.85]{13120f8a.eps}
\end{figure} Figure 8:

The complete atlas of FIES@ NOT spectra (part 1 of 3). The Si II-IV and O II lines used for the abundance analysis are indicated as solid and dashed vertical lines, respectively.

Open with DEXTER
In the text

  \begin{figure}
\par\includegraphics[angle=180,scale=0.85]{13120f8b.eps}
\end{figure} Figure 8:

continued.

Open with DEXTER
In the text

  \begin{figure}
\par\includegraphics[angle=180,scale=0.85]{13120f8c.eps}
\end{figure} Figure 8:

continued.

Open with DEXTER
In the text


Copyright ESO 2010

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.