A SPECTROSCOPIC CENSUS IN YOUNG STELLAR REGIONS: THE σ ORIONIS CLUSTER

2.1.1. OSMOS Photometry

2.1.2. Additional Photometry

The initial sample in this study includes all Two Micron All Sky Survey (2MASS) sources (4659 sources; Cutri et al. 2003) in a region of 48' × 48' centered at R.A. = 847 and decl. = −26. This region covers the field studied in H07b using the four channels of the InfraRed Array Camera (IRAC; Fazio et al. 2004). The 2MASS catalog is complete down to J < 15.8, which includes stars beyond the substellar limit expected for the σ Orionis cluster (e.g., J ∼ 14.6; H07b). We compared sources from the 2MASS catalog and from the United Kingdom Infrared Telescope (UKIRT) Infrared Deep Sky Survey (UKIDSS; Lawrence et al. 2007, 2013). When 2MASS sources have poor photometric quality flag ("U," "F," or "E") for the J magnitude, we use photometric measurements reported in the UKIDSS catalog.

Out of 31 UKIDSS sources with J < 15.8 and without 2MASS counterparts, 10 sources are galaxies based on the profile classification of UKIDSS and 16 sources are close visual binaries not resolved in 2MASS images (2MASS reports photometry only for the brightest companion). From the remaining five sources, four stars are background candidates and only one star is a photometric member based on its location in the Z versus Z−J color–magnitude diagram (Lodieu et al. 2009). This star has an optically thick disk (SO 566; H07b) and the non-detection in 2MASS images suggests large variability (variability amplitude in J band ≳ 2.7). The star SO 566 was added to the initial sample of 4659 2MASS sources. Finally, there are four additional sources studied in H07b without 2MASS counterpart (SO 406, SO 336, SO 361, and SO 950). Since all these sources are fainter (J > 17.9 in UKIDSS) than the 2MASS completeness limit, they were not included in this study.

2.2.1. 2MASS Sources without Optical Photometry

We compiled lists of spectroscopically confirmed members of the σ Orionis cluster on the basis that they exhibit Li i λ6708 in absorption or they have radial velocities (RVs) expected for the kinematic properties of the cluster (hereafter "spectroscopic known members").

Using RV measurements, Jeffries et al. (2006) showed that young stars located in the general region of the cluster consist of two stellar groups kinematically separated by 7 km s⁻¹ in RV and with different mean ages and distances. One group has RVs from 27 km s⁻¹ to 35 km s⁻¹, which is consistent with the RV of the central star σ Ori AB (29.5 km s⁻¹; Kharchenko et al. 2007). The other group has RVs from 20 km s⁻¹ to 27 km s⁻¹ and could be in front of the first group and could have a median age and distance similar to older stellar groups associated with the sparser Orion OB1a subassociation (age ∼ 10 Myr; distance ∼326; Briceño et al. 2007). This older and kinematically distinct stellar population (hereafter "the sparser stellar population") is located to the northwest of the central system and more difficult to detect in RV distributions of stars located near the center of the cluster (Sacco et al. 2008). Using the limit defined by Jeffries et al. (2006), we included in our study 162 kinematic members of the σ Orionis cluster, with RV between 27 and 35 km s⁻¹, from Zapatero Osorio et al. (2002), Kenyon et al. (2005), Caballero (2006), Burningham et al. (2005), Maxted et al. (2008), Sacco et al. (2008), and González Hernández et al. (2008).

We also included as spectroscopic known members, 181 stars with Li i λ6708 in absorption from Wolk (1996), Alcalá et al. (1996), Zapatero Osorio et al. (2002), Muzerolle et al. (2003), Barrado y Navascués et al. (2003), Andrews et al. (2004), Kenyon et al. (2005), Caballero (2006), Caballero et al. (2006, 2012), González Hernández et al. (2008), and Sacco et al. (2008). Since the presence of lithium is an indicator of youth in late-type stars, our selection includes stars with reported spectral types K or M. Our list includes all late-type stars with Li i λ6708 in absorption compiled in the Mayrit Catalog (Caballero 2008a).

Additionally, we have compiled 168 X-ray sources reported by Caballero (2008a), Skinner et al. (2008), López-Santiago & Caballero (2008), Caballero et al. (2009, 2010), or X-ray sources in the XMM-Newton Serendipitous Source Catalog 2XMMi-DR3 (XMM-SSC, 2010) with source detection likelihoods (srcML) larger than eight (sources with scrML < 8 may be spurious). Since young stars are strong X-ray emitters, likely members of the σ Orionis clusters are X-ray stellar sources located above the zero-age main sequence (ZAMS).

Finally, another indicator of youth is the infrared excess present when stars are surrounded by circumstellar disks. Using Spitzer Space Telescope observations, we reported 114 photometric candidates with infrared excesses (H07b).

Youth indicators based on the presence of Li i λ6708 in absorption, X-ray observations and infrared excesses do not discriminate interlopers from the sparser stellar population. Regardless of this expected contamination, we define the sample of known likely members (hereafter "known members") compiling kinematic members based on RVs, stars with Li i λ6708 in absorption, X-ray sources above the ZAMS and stars with infrared excesses. Table 1 shows information about the known member sample.

Using the photometric data compiled in Sections 2.1.1 and 2.1.2, we selected photometric candidates of the σ Orionis cluster. Although we have V, R_C, I_C, and J magnitudes for most stars in the initial sample, we use the V−J color because is the most complete for our sample and it has the greatest color leverage for constraining PMS populations.

Figure 1 shows the color–magnitude diagram V versus V−J for sources in the initial sample with optical counterparts. We have estimated an empirical isochrone as the location of known members compiled in Section 2.3; stars with infrared excesses, young stars confirmed using the presence of Li i λ6708 in absorption, kinematic members confirmed using RVs, and X-ray sources above the ZAMS (Siess et al. 2000) located at 385 pc (Caballero 2008c). The empirical isochrone in Figure 1 was estimated using the median V−J color of the known members for 1 mag bins in the V band. Standard deviations (σ) were calculated using the differences between the observed colors and the expected colors from the empirical isochrone. We fitted two straight lines to the points represented by the empirical isochrone color + 3σ and by the empirical isochrone color −3σ. Photometric candidates are stars that fall between these lines (Hernández et al. 2008, 2010).

Zoom In Zoom Out Reset image size

Our photometric sample includes the photometric candidates reported in H07b using a more limited optical photometric data set. The present sample is larger because we have included an updated version of the CVSO catalog (Mateu et al. 2012), photometry from the cluster collaboration, and the new OSMOS photometry described in Section 2.1.1.

Out of 255 spectroscopic known members and disk bearing candidates, 249 stars are located in the photometric candidates region. Based on this, we lose about 2.4% of known members of the σ Orionis cluster with our photometric selection. Since targets for spectroscopic follow up generally are selected using photometric cuts, the exclusion percentage of know members could be a lower limit of the percentage of actual members excluded from our photometric selection. On the other hand, Burningham et al. (2005) found that photometric selection cuts do not miss significant numbers of bona-fide members of the σ Orionis cluster and thus the percentage of actual member excluded by our photometric selection can not be much higher than ∼2.4%. Since our photometric criteria to select candidates is conservative, contamination by background stars is present in the entire color range. At a color V−J ∼ 1.5, where a branch of old field stars crosses the PMS of the σ Orionis cluster, the expected contamination level by non-members of the cluster is quite high (e.g., Hernández et al. 2008). Table 2 includes optical magnitudes for the photometric candidates and for the known members not located in the photometric candidates region.

Optical low-resolution spectra is a fundamental tool to identify and characterize young stars by using indicators as the presence of lithium in absorption, strong Hα in emission or weak gravity-sensitive features. We have obtained low-resolution spectra for 340 stars located in the σ Orionis cluster. This spectroscopic data comes from several spectrographs with similar spectral coverage and resolution (see Table 3). Targets for spectroscopic followup were selected mainly from the sample of photometric members with infrared excesses or the sample of photometric members with X-ray counterparts. Except for observations obtained using the Hectospec multifibers spectrograph, target selection includes stars with V magnitude brighter than 16.5. Using the 3 Myr isochrone from Siess et al. (2000) and assuming a distance of 385 pc (Caballero 2008c) without reddening, this limit corresponds to stars of spectral type M3 or earlier (M_* > 0.35 M_☉).

2.5.1. Hectospec

2.5.2. FAST

2.5.3. OSU-CCDS

2.5.4. Boller and Chivens, SPM

2.5.5. Boller and Chivens, Cananea

We obtained high-resolution spectra of a subset of candidates of the σ Orionis cluster using the Hectochelle fiber-fed multiobject echelle spectrograph mounted on the 6.5 m Telescope of the MMTO. Hectochelle can record up to 240 high-resolution spectra simultaneously in a 1° circular field. Each fiber subtends 15 on the sky (Szentgyorgyi et al. 2011). We use the order-sorting filter OB26 that provides a resolution of R ∼ 34,000 with 180 Å of spectral coverage centered at 6625 Å. In spite of the fact that the order-sorting filter OB26 is not optimal for RV measurements (Fűrész et al. 2008), the spectral coverage of this filter allows us to analyze the Hα profile and the Li i λ6708 line to identify young stars, accretors, and non-accretors.

One Hectochelle field was obtained on the night of 2007 February 27 that provides high-resolution spectra for 134 photometric candidates and 8 additional stars in the region studied in this work. The data were reduced using an automated IRAF pipeline developed by G. Furesz, which utilizes the standard spectral reduction procedures using IRAF and the tasks available under the packages mscred and specred. A more detailed description of Hectochelle data reduction can be found in Sicilia-Aguilar et al. (2006).

Spectral types were derived applying the SPTCLASS code on the sample of stars with low-resolution spectra (Section 2.5). SPTCLASS is a semi-automatic spectral analyzing program that uses empirical relations between spectral type and equivalent widths (EWs) to classify and characterize stars based on selected features.¹³ For spectral typing, it has three schemes optimized for different mass ranges (K5 or later, from late F to early K, and F5 or earlier), which use different sets of spectroscopic features (Hernández et al. 2004; Sicilia-Aguilar et al. 2005; Downes et al. 2008). The user has to manually choose the best scheme for each star based on the prominent features in the spectrum and the consistency of several spectral indices. The EW for each spectral feature is obtained by measuring the decrease in flux due to photospheric absorption line from the continuum that is expected when interpolating between two adjacent bands. The spectral features measured by this procedure are largely insensitive to reddening and the signal-to-noise ratio (S/N) of the spectra (as long as we have enough S/N to detect the spectral feature). Moreover, spectral types obtained by SPTCLASS are largely independent of luminosity because most of the indices selected are not sensitive to the surface gravity of the star (Hernández et al. 2004). However, SPTCLASS does not take into account the effect on the lines of the hot continuum emission produced by the accretion shocks. This continuum emission makes the photospheric absorption lines appear weaker and for K-type and M-type highly veiled stars the SPTCLASS outputs should be considered as the earliest spectral-type limits (Hsu et al. 2012). SPTCLASS is unable to give spectral types for highly veiled stars earlier than K-type (e.g., continuum stars in Hernández et al. 2004). In Table 4, we report spectral types for the sample observed in Section 2.5.

In Figure 2, we compare spectral types derived in this work with those published previously from selected works (Wolk 1996; Houk & Swift 1999; Zapatero Osorio et al. 2002; Caballero 2006; Caballero et al. 2012; Rigliaco et al. 2012). Additionally, our spectral types were compared with spectral types compiled in the SIMBAD¹⁴ database from different authors (Cohen & Kuhi 1979; Nesterov et al. 1995; Houk & Swift 1999; Zapatero Osorio et al. 2002; Barrado y Navascués et al. 2003; Muzerolle et al. 2003; Hernández et al. 2005; Briceño et al. 2005; Oliveira et al. 2006; Caballero et al. 2006, 2010, 2012; Gatti et al. 2008; Sherry et al. 2008; Skinner et al. 2008; Renson & Manfroid 2009; Nazé 2009; Cody & Hillenbrand 2010; Townsend et al. 2010; Rigliaco et al. 2011, 2012). Most spectroscopic studies of the σ Orionis cluster have derived spectral types for low-mass stars and very low-mass stars (spectral types K5 or later). Thus, in Figure 2 we have few points of comparison in the solar-type range (from F to early K). In general, and with the exception of Wolk (1996), spectral types reported previously by other authors agree within the uncertainties with spectral types calculated using SPTCLASS. Wolk (1996) used absorption line ratios of Ca i lines (λ6122, λ6162, and λ6494) and Fe i lines (λ6103, λ6200, and λ6574) as indicators of spectral types. The spectral type dependencies of these indicators are weak for stars with spectral types early K or earlier. Additionally, the spectroscopic data set of Wolk (1996) have poor S/N, which severely affects the measurements (Zapatero Osorio et al. 2002). A subset of the stars studied by Wolk (1996) were observed by Zapatero Osorio et al. (2002). Their spectral types, based on the measurements of molecular bands like TiO and CaH, match our determinations. On the other hand, Caballero (2006) used indices based on the EW of some lines (Fe i λ6400, Ca i λ6439, Ca i λ6450, Ca i λ6462, and Ca i λ6717) to estimate spectral types. The spectral type dependencies of these indices are very weak in the spectral type range from G0 to K0, where we observe larger discrepancies between our spectral types and the spectral types reported by Caballero (2006).

Zoom In Zoom Out Reset image size

EWs of Li i λ6708 and Hα lines were measured using the interactive mode of SPTCLASS in which two adjacent continuum points must be manually selected on the spectra to fit a Gaussian line to the feature.¹⁵ We find that the typical uncertainties of EWs measured in this way are ∼10%. In Table 4, we assign a flag to each star based on the presence of Li i λ6708 in absorption: "2" means that the star has Li i λ6708 in absorption; "1" means that there are doubts about the measurement of Li i, generally, due to low S/N of the spectra; and "0" means that Li i λ6708 is not present in the spectra.

Based on the EW of Hα and the criterion defined by Barrado y Navascués et al. (2003) to distinguish between accretors and non-accretors for stars with spectral type G5 or later, in Table 4, we classify stars into different groups: accretors (Acr), non-accretors (nAcr), and stars in which uncertainties for EWs of Hα and spectral types fall on the limit between accretors and non-accretors (Acr?). We also have identified stars with spectral types earlier than G5 with Hα in emission (ET_em) and with Hα in absorption (E_abs). Figure 3 shows the relation between the EW of Hα and the spectral type. In order to plot our data in a logarithmic scale, EWs of Hα have been shifted by 10 units (the dotted line is the limit between absorption and emission of Hα). Solid line delimits the area for accretors and non-accretor stars based on the EWs of Hα emission line (Barrado y Navascués et al. 2003). Using different symbols we plot results obtained using different spectrographs (see Table 3). We also plot stars bearing protoplanetary disks (H07b). In general, stars above the Barrado's accretion cutoff were classified as stars bearing full disks (II) or transitional disks (TD) in H07b. There are some stars with optically thick disks below the Hα EW criterion (SO 435, SO 467, SO 566, SO 598, SO 682, SO 823, and SO 967). These stars would be similar to CVSO 224, which was classified as a weak-line T Tauri star based on the EW of Hα. However, using high-resolution spectra, Espaillat et al. (2008) show that is a slowly accreting T Tauri star with accretion rate of 7 × 10⁻¹¹ M_☉ yr⁻¹ estimated using the magnetospheric accretion models (Muzerolle et al. 2001). On the other hand, there are diskless stars that were classified as accretors (SO 229, SO 1123, and SO 1368). These diskless objects with Hα in emission could have substantial contribution produced by strong chromospheric activity which is related to fast rotation. In Section 4.2, we study in more detail slowly accreting stars and diskless stars that mimic the Hα width expected in stars with accretion disks.

Zoom In Zoom Out Reset image size

For very low-mass stars (VLMS) observed with HECTOSPEC, we can use the line Na i λ8195 as an additional indicator of youth (Downes et al. 2008). Since VLMS in the σ Orionis cluster are still contracting, their surface gravities are lower than that expected for a main-sequence star with similar spectral type. The line Na i λ8195 is the strongest one of the sodium doublet (Na i λλ8183, 8195), which is sensitive to surface gravity and varies significantly between M-type field dwarf and PMS objects (Schlieder et al. 2012). The EW of the line Na i λ8195 was estimated fitting a Gaussian function to the feature on the normalized spectrum. We use the continuum bands suggested by Schlieder et al. (2012) to normalize each spectrum. The uncertainties in EWs were calculated as in Kenyon et al. (2005) using the scale of HECTOSPEC (∼1.2 Å pixel⁻¹) and the S/N of the continuum bands. Figure 4 shows normalized spectra of three stars with spectral type M5 illustrating the procedure to measure the line Na i λ8195. The sources 05400029+0142097 and 05373723+0140149 are CVSO's stars spectroscopically confirmed as a dwarf field star and as a ∼10 Myr old star, respectively (C. Briceno et al., in preparation). The star SO 460 has Li i λ6708 in absorption and has been confirmed previously as member of the σ Orionis cluster using RV (Zapatero Osorio et al. 2002; Sacco et al. 2008; Maxted et al. 2008). It is apparent that the M-type field star exhibits stronger absorption in the sodium doublet in comparison to the PMS stars.

Zoom In Zoom Out Reset image size

Figure 5 shows the EW of Na i λ8195 versus spectral type for stars in the σ Orionis sample, illustrating the procedure of using the sodium line as an additional criteria supporting membership in the VLMS of σ Ori. We display the median and the second and third quartiles of Na i λ8195 measured for a sample of PMS stars (lower solid line) and for a sample of M-type field dwarfs (upper solid line) in the Orion OB1a and OB1b associations (C. Briceno et al., in preparation). These PMS stars, with ages ranging from ∼5 to ∼10 Myr, were confirmed by the presence of Li i λ6708 in absorption. In general, stars confirmed as members by the presence of Li i λ6708 in the σ Orionis cluster (open squares) are below the median population of PMS stars in the OB1a and OB1b associations. On the other hand, stars selected as non-members of the σ Orionis cluster (open circles) follow the median line of M-type field dwarfs in the OB1a and OB1b associations. Figure 5 shows that the separation between M-type field stars and PMS stars is more clear for later spectral types, so that for stars with spectral type M3 or later we use the EW of Na i λ8195 as an additional criteria of youth; those stars located below the median line of the PMS stars in the OB1a and OB1b associations were identified as young stars (labeled with "Y" in Column 10 of Table 4). Most of the σ Ori VLMS with uncertain membership based on Li i λ6708 (crosses) have smaller values of EW of Na i λ8195 and thus are likely members of the cluster. The stars SO 576 and SO 795 exhibit strong Na i λ8195 line and thus are classified as M-type field stars (labeled with "N" in Table 4)

Zoom In Zoom Out Reset image size

In order to estimate reddening for our sample, we calculate the rms of the differences between the observed colors and the standard colors

$$\begin{equation} {{\rm rms}(A_V)=\sqrt{\displaystyle\frac{\sum {([V-M_\lambda ]_{{\rm obs}}- \left[1-\displaystyle\frac{A_\lambda }{A_V} \right]*A_V - [V-M_\lambda ]_{{\rm std}})^2}}{n}}, } \end{equation} \tag{ 1 }$$

where [V − M_λ]_obs are the observed colors V-Rc, V-Ic, and V−J (when they are available), n is the number of colors used to calculate rms(A_V), and [V − M_λ]_std are the corresponding standard colors for a given spectral type. The values of A_λ/A_V were obtained from Cardelli et al. (1989) assuming the extinction law normally used for interstellar dust (A_λ/A_V = 0.832, 0.616 and 0.288 for Rc, Ic, and J, respectively). We vary the visual extinction (A_V) from 0 to 10 mag in steps of 0.01 mag and estimate visual extinction as the A_V value with the lowest rms(A_V). We use the intrinsic colors for main-sequence stars from Kenyon & Hartmann (1995) and the intrinsic colors of 5–30 Myr old PMS stars from Pecaut & Mamajek (2013). Since these PMS intrinsic colors are given for stars F0 or later, we complete the standard colors for stars earlier than F0 using the intrinsic color of 09–M9 dwarf stars compiled by Pecaut & Mamajek (2013). Intrinsic colors of PMS stars earlier than G5 appear to be consistent with the dwarf sequence (Pecaut & Mamajek 2013).

Since PMS stars are in a different evolutionary stage than dwarf field stars, there may be systematic errors in the calculation of interstellar reddening when using main-sequence calibrators. Figure 6 shows a comparison between the reddening calculated using the intrinsic colors for main-sequence stars from Kenyon & Hartmann (1995) and the intrinsic colors for PMS stars from Pecaut & Mamajek (2013). In general, for stars in the spectral type range from G2 to M3, the visual extinctions estimated from Kenyon & Hartmann (1995), are larger than those estimated using PMS colors (Pecaut & Mamajek 2013). On the other hand, for stars later than M3 (open squares), visual extinctions estimated from Kenyon & Hartmann (1995), are slightly smaller than those estimated using PMS colors (Pecaut & Mamajek 2013).

Zoom In Zoom Out Reset image size

Sorted by spectral types, Table 4 summarizes the results obtained from the low-resolution data set. Column 3 indicates the spectrograph used for each star (see Table 3). Column 4 shows spectral types. Columns 5, 7, and 9 show the EWs of Li i λ6708, H_α, and Na i λ8195, respectively. Flags based on these lines are shown in Columns 6, 8, and 10. Columns 11 and 12 show visual extinctions calculated using standard colors for main-sequence stars (Kenyon & Hartmann 1995) and using standard colors for PMS stars (Pecaut & Mamajek 2013).

We measured heliocentric RVs in the high-resolution spectra of the sample observed with Hectochelle. RVs were derived using the IRAF package "rvsao" that cross-correlate each observed spectrum with a set of templates of known RVs (Tonry & Davis 1979; Mink & Kurtz 1998). We used the synthetic stellar templates from Coelho et al. (2005) with solar metallicity, surface gravity of log(g) = 3.5 and effective temperature ranging from 3500 to 7000 K in steps of 250 K. The use of synthetic templates enables us to explore a wider range of stellar parameters than a few observed templates (Tobin et al. 2009). We estimated the RV and the template as those that gave the highest value of the parameter R, defined as the S/N of the cross correlation between the observed spectrum and the template. The RV error is calculated from the parameter R as (Tobin et al. 2009):

$$\begin{equation} {\rm RV}_{{\rm err}}=\frac{14 ({\rm km\, s}^{-1})}{1+R}. \end{equation} \tag{ 2 }$$

In Table 5, we report RVs for 95 stars in which R is larger than 4 (Rverr ≲ 2.8 km s⁻¹). This sample includes 36 stars with known RVs compiled in Section 2.3. In general, the differences between our measurements and known RVs were less than 2 km s⁻¹ (rms = 1.9 km s⁻¹). There are five stars (SO 616, SO 662, SO 791, SO 929, and SO 1094) with differences larger than 2 km s⁻¹. Since during its orbit around the center of mass RVs of the components of binaries could change, these five stars are labeled as binary candidates by radial velocity variability (RVvar). The cross-correlation function also can be used to identify double-lined spectroscopic binaries (SB2). We have identified four SB2 stars that show double peaks in the cross-correlation function.

EWs of the Li i λ6708 line reported in Table 5 were calculated fitting a Gaussian function to the line in the high-resolution spectra. The errors of EWs were calculated as in Kenyon et al. (2005) using the scale of HECTOCHELLE (00786 pixel⁻¹) and the S/N of the continuum used to obtain the EW. Figure 7 shows a diagram of EWs of Li i λ6708 versus RVs. The distribution of RVs can be described by a Gaussian function centered at 30.8 km s⁻¹ with a standard deviation (σ_RV) of 1.7 km s⁻¹. Applying a 3σ_RV criteria, we can define a RV range for kinematic members of the cluster from 25.7 km s⁻¹ to 35.9 km s⁻¹. These values are similar to those reported by Jeffries et al. (2006) for the stellar group associated with the star σ OriAB (27–35 km s⁻¹). In Table 5, we label as kinematic members of the σ Orionis cluster stars with RV in the membership range (flag "2") and as kinematic candidates of the sparser stellar population (see Section 2.3) stars with RVs from 20 km s⁻¹ to 25.7 km s⁻¹ (flag "1"). Stars with RV smaller than 20 km s⁻¹ and larger than 35.9 km s⁻¹ were flagged as kinematic non-member (flag "0"). Three stars with Li i λ6708 in absorption fall in the last group. The stars SO 362 and SO 1251 have RV > 35.9 km s⁻¹ and the star SO 243 has RV < 20 km s⁻¹. The star SO 362 has a protoplanetary disk and its discrepant RV value could be caused by binarity. The Li i λ6708 measured in the star SO 1251 is lower than those expected for the σ Orionis stellar population and thus its membership is uncertain. Since the spectral type of the star SO 243 is relatively early (G5; see Section 3.1), the youth of this object is also uncertain. The shallow depth of the convective zone in stars earlier than K0 can allow them to reach the main sequence with a non-negligible amount of their primordial lithium content and thus Li i λ6708 in absorption is not a reliable indicator of the PMS nature of these stars. Therefore, the presence of lithium in absorption in F-type and G-type stars is only evidence that these objects are not old disk, post-main-sequence stars (Briceno et al. 1997). Finally, we were unable to calculate RVs for 17 stars with Li i λ6708 in absorption. These stars were included as spectroscopic members of the cluster.

Zoom In Zoom Out Reset image size

The full width of H_α at 10% of the line peak (WH_α_10%) is an indicator that allows to distinguish between accreting and non-accreting young stellar objects. White & Basri (2003) suggested that WH_α_10% > 270 km s⁻¹ indicates accretion independent of the spectral type. Jayawardhana et al. (2003) adopted a less conservative accretion cutoff of 200 km s⁻¹ studying young very low-mass stars and brown dwarfs. A newly born star is surrounded by a primordial optically thick disk that evolves due to viscous processes by which the disk is accreting material onto the star while expanding to conserve angular momentum (Hartmann et al. 1998). As a consequence of this and other evolutionary processes, the frequency of accretors and the accretion rate in the inner disks steadily decreases from 1 to 10 Myr (e.g., Calvet et al. 2005; Williams & Cieza 2011). Thus signatures of accretion can be used as indicators of membership in a young stellar region. In the case in which Hα has a symmetric or quasi-symmetric profile in emission, we measured automatically the peak of the profile fitting a Gaussian function to the Hα line. When the fitting of a single Gaussian function does not work (e.g., non-symmetric profiles), the peak of the Hα line was selected manually. The WH_α_10% was measured as the width of the Hα velocity profiles at 10% of the emission peak level (Column 6 in Table 5). Following the criterion from White & Basri (2003), stars with WH_α_10% > 270 km s⁻¹ were identified as accretors. Stars without measurements of WH_α_10% have comments in Table 5 about the Hα profile.

Since the criteria used in Section 3.1 as indicators of youth are useful mainly for stars with spectral types K and M, we need to use additional criteria to confirm or reject candidates as member of the σ Orionis cluster. In general, the memberships described in this section are less reliable than those obtained from low-resolution spectroscopic analysis (Section 3.1) and those obtained from high-resolution spectroscopic analysis (Section 3.2).

3.3.1. Photometric Membership Probabilities and Variability

PMS stars are characterized by having a high degree of photometric variability of diverse nature (Herbst et al. 1994; Cody & Hillenbrand 2010, 2011). The variability is due to chromospheric and magnetic activity in the stellar surface (e.g., cool spots, flares, and coronal mass ejections) and to protoplanetary disks around the stars (e.g., hot spots produced by accretion shocks, variable accretion rates, and variable extinction produced by inhomogeneities in the dusty disk). Since the expected photometric variability in PMS stars are so diverse, it is difficult to apply variability criteria such as range of periods or light curve types to refine the selection of possible members of a young stellar group. However, we can use the location of variable stars in color–magnitude diagrams to select them as PMS candidates. Briceño et al. (2001, 2005) indicate that selecting variable stars above the ZAMS located at the typical distance of a young stellar group clearly picks a significant fraction of members of that stellar group. Moreover, using the differences between the observed colors and the expected colors defined by empirical or theoretical isochrones, we can calculate photometric membership probabilities for the sample.

Figure 8 shows the procedure to estimate photometric membership probabilities. Based on [V−J] colors and V magnitudes, we tailored new indices (C_rot and M_rot) using the following equations to rotate the color–magnitude diagram of Figure 1:

$$\begin{equation} C_{{\rm rot}}=[V-J]*\cos (\theta)-(V-V_0)*\sin (\theta) \end{equation} \tag{ 3 }$$

$$\begin{equation} M_{{\rm rot}}=V_0+[V-J]*\sin (\theta)+(V-V_0)*\cos (\theta), \end{equation} \tag{ 4 }$$

where V₀ is the V value where the empirical isochrone has [V−J] = 0 and θ is the angle where the rms of C_rot for the known member sample is minimal (∼255). The distribution of C_rot for known members (open circles) describes a Gaussian function with standard deviation of σ = 0.32. Assuming a standard normal distribution we can transform C_rot values to probabilities for our sample. The 3σ criterion applied in Section 2.5.5 to select photometric candidates means that stars that fall outside of this criteria have membership probabilities lower than ∼0.3% (dotted lines in Figure 8). Notice that the photometric probability obtained here does not take into account the distribution of non-members in Figure 8. The contamination by non-members is much higher at negative values of C_rot than at positive values of C_rot. Also, the expected contamination is higher at M_rot ∼13 where a branch of old field stars crosses the stellar population of the σ Orionis cluster.

Zoom In Zoom Out Reset image size

We plot in Figure 8 variable stars reported in the CVSO catalog (Briceño et al. 2005; Mateu et al. 2012), stars flagged as variable in the cluster collaboration's photometric catalogs (Kenyon et al. 2005; Mayne et al. 2007), variable low-mass stars and brown dwarfs of the σ Orionis cluster (Cody & Hillenbrand 2010), and stars listed as known or suspected variables in The AAVSO International Variable Star Index (Watson 2006). In spite of the fact that this compilation of variable stars in the σ Orionis cluster is not complete, 60% of the stars within the 1σ criteria (dotted lines; photometric probability higher than ∼32%) are reported as variable objects.

3.3.2. Proper Motion and Spatial Distribution

In general, the motion of young stellar groups of the Orion OB1 association are mostly directed radially away from the Sun. Thus, the expected intrinsic proper motions in right ascension (cos (δ)*μ_α) and declination (μ_δ) are small and comparable to measurement errors (Brown et al. 1998; Hernández et al. 2005). Particularly, Brown et al. (1998) reported an average proper motion for this stellar association of 0.44 mas yr⁻¹ and −0.65 mas yr⁻¹ for cos (δ)*μ_α and μ_δ, respectively. Although we cannot use proper motion to separate potential cluster members from field star non-members, we can use criteria based on proper motions to identify and reject high proper motion sources as potential members of the σ Orionis cluster (e.g., Lodieu et al. 2009; Caballero 2010).

We follow a similar method used in Hernández et al. (2009) to identify potential non-members of the cluster. Figure 9 shows the vector point diagram for the photometric candidates with proper motions reported in the fourth US Naval Observatory CCD Astrograph Catalog (UCAC4; Zacharias et al. 2013). The distributions of proper motions for the known members (green X's) can be represented by Gaussians centered at cos (δ)*μ_α ∼ 2.2 mas yr⁻¹ and μ_δ ∼0.7 mas yr⁻¹, with standard deviations of 5.3 mas yr⁻¹ and 4.1 mas yr⁻¹, respectively. These values agree within the errors with previous estimations for the entire OB association (e.g., Brown et al. 1998) and for the cluster (e.g., Kharchenko et al. 2005; Caballero 2007, 2010). Most known members (∼86%) are located in a well defined region represented for the 3σ criteria (dotted ellipse) in the vector point plot. We also use a less conservative criteria of 5σ (solid line ellipse) similar to the criteria used in previous works (Caballero 2007, 2010; Lodieu et al. 2009). With the exception of the stars SO 592, SO 936, and SO 1368, the 5σ criteria includes all the known member sample. The star SO 592 is a RV member with Li i λ6708 in absorption (see Section 2.6). The star SO 1368, reported as diskless accretor in Section 3.1, has large error in cos (δ)*μ_α ∼ and thus the proper motion criteria for this object is uncertain. The star SO936 lies outside the vector point diagram. Its infrared excess at 8 μm (H07b) is the only youth feature reported for this photometric candidate. Astrometric followup could help to find out if those stars are members of binary systems, they were ejected from the cluster early on during the formation process or they belong to a moving group associated with Orion (Lodieu et al. 2009).

Zoom In Zoom Out Reset image size

Based on the distribution of stars in the vector point diagram, we have classified our photometric candidates into three groups, stars inside of the 3σ limit (proper motion flag "2"), stars between the 3σ and the 5σ limit (proper motion flag "1") and stars with proper motions larger than the 5σ criteria (proper motion flag "0"). In general, the proper motion flags agree with previous proper motion studies to identify high proper motion interlopers (Lodieu et al. 2009; Caballero 2010). Out of 29 stars studied by Caballero (2010) and located in the region studied in this work, 14 stars have proper motion in UCAC4. All the 13 photometric candidates identified by Caballero (2010) as high proper motion interlopers have proper motion flag "0." The other star is a background star (Section 2.4) rejected using optical colors (star #38 in Caballero 2010). Only one star (05401975-0229558) reported by Lodieu et al. (2009) as proper motion non-member have proper motion flag "1." Additional studies are necessary to obtain youth features of this object.

Finally, Caballero (2008b) suggests that the σ Orionis cluster has two components: a dense core that extends from the center to a radius of 20' and a rarefied halo at larger separations. Members of the σ Orionis cluster have higher probability to be located in the dense core than in the rarefied halo. Thus, we also include a flag for the distance from the center of the cluster. Stars located closer than 20' have flag "1," otherwise they have flag "0."

Depending on the spectral type range, obtaining memberships of stars that belong to a young stellar group can be a difficult process and some times we need to combine different membership indicators. Based on the spectroscopic and photometric analysis developed in Sections 2.4, 4.1, 3.2, 3.1, and 3.3, and the information compiled in Section 2.3 about several membership criteria, we compile in Table 6 membership indicators for stars with spectral types obtained in Section 3.1. Columns 1 and 2 show source designations from H07b and Cutri et al. (2003), respectively. Column 3 shows the spectral type. Column 4 indicates the membership flag based on the presence of Li i λ6708 in absorption from low-resolution spectra (Section 3.1) and from high-resolution spectra (Section 3.2). Column 5 shows the references of known members based on Li i λ6708. Columns 6 and 7 show the RV information from our analysis (Section 3.2) and from previous works, respectively. Column 8 shows the classification based on the Hα line and the accretion criteria from Barrado y Navascués et al. (2003, Section 3.1) and from White & Basri (2003, Section 3.2). Column 9 indicates the membership flag based on the line Na i λ8195 (Section 3.1). Additional membership information based on the presence of protoplanetary disks, X-ray emission, proper motion, distance from the center of the cluster, and variability are in Columns 10, 11, 12, 13, and 14, respectively. Column 15 gives the photometric membership probability calculated in Section 3.3 and Column 16 indicates the visual extinction estimated using the PMS standard colors of Pecaut & Mamajek (2013, Section 3.1). Finally, our membership flags and comments are given in the last column. Similar to the membership study by Kenyon et al. (2005), some of our membership flags are arguable, but the reader can reach their own conclusions based on the information in Table 6.

In general, stars with spectral types B or A in Table 6 were included as young stars by Caballero (2007) and Sherry et al. (2008) using proper motion, RV, X-ray, and infrared observations and main sequence fitting analysis. The star SO 602 was not included in those studies. This star has discrepant RV and very high visual extinction to be considered member of the σ Orionis cluster. The star SO 147 was not included in Sherry et al. (2008) and was rejected by Caballero (2007) based on the proper motions reported by Perryman et al. (1997). SO 147 is a X-ray source and our RV measurements indicate that could be a young star member of the sparser population (Maxted et al. 2008) or a binary of the σ Orionis cluster. However, the visual extinction of SO 147 is higher than that expected for the cluster, thus its membership is uncertain. Sherry et al. (2008) suggested that SO 956 is too bright to be a member of the cluster and SO 521 is too faint to be a member of the cluster. The star SO 956 has youth features like infrared excess and X-ray emission and the star SO 521 is located above the ZAMS but with relatively low photometric membership probability. Thus is not clear the membership status of these two objects. Finally, the star SO 139 is relatively bright to be a member of the cluster and Sherry et al. (2008) suggested that it could be a member of the cluster if this star is an equal mass binary. In summary, the memberships of the stars SO 139, SO 956, SO 521, and SO 147 reported in Table 6 are uncertain.

It is more difficult to estimate membership for solar-type stars (F and G). In this spectral type range, old stars cross the sequence defined by the σ Orionis cluster and thus the contamination level by non-members of the cluster is quite high. For some stars, Li i λ6708 appears in absorption. However, for solar-type stars, this is not a robust criteria of youth and it is only evidence that these objects are not old disk, post-main-sequence stars (Briceno et al. 1997; Hernández et al. 2004). Thus, we need to evaluate other membership indicators like presence of disks, X-ray emission, variability, RV proper motions, reddening, and photometric membership probabilities to confirm or reject stars as members of the cluster. In Table 6, solar-type stars with Li i λ6708 in absorption have additional youth features that support their membership of the cluster (mainly X-ray source, infrared excesses, and variability). Four stars with Li i λ6708 in absorption (SO 1123, SO 243, SO 92, and SO 379) have discrepant RV, which causes us to suspect that they are not members of the cluster. Another possibility is that these stars are binaries with variable RVs. Since, the frequency of binaries increases with the stellar mass (e.g., Duchêne & Kraus 2013), the probability to find a binary member with discrepant RV in solar-type stars is higher in comparison with low-mass stars. We labeled these stars as probable members. We rejected as members of the cluster stars with very low (<0.3%) photometric membership probability and stars with discrepant RV that do not have additional features of youth. Finally, some stars do not have enough information to define their membership and were labeled as uncertain members.

For low-mass stars (from K0 to M2.5) and for very low-mass stars (M3 or later), the presence of Li i λ6708 in absorption is a reliable indicator of youth. For very low-mass stars when the presence of Li i λ6708 in absorption is uncertain (flag "1"), we used the measurements of Na i λ8195 to support the membership status. Moreover, stars with uncertain flag of Li i λ6708 that exhibit other indicators to be members of the cluster (including previous Li i λ6708 reported for the star) are also classified as members of the σ Orionis cluster. We also considered member of the cluster stars with protoplanetary disks classified as accretors using the Hα line. Finally, RV can be used as a strong membership criteria for stars in the entire spectral type range and could be use to separate the σ Ori stellar population and the sparser stellar population (see Section 2.3).

Optical spectra and spectral energy distributions (SEDs) for the entire sample with spectral types obtained in Section 3.1, plus additional information compiled from Tables 2, 4, and 6 for each star, is available online.¹⁶ This online tool is a work in progress and we expect to add new objects and to extend the information about the population of the σ Orionis cluster.

Finally, Table 7 compiles membership information for a set of stars studied in high resolution (Section 3.2) but without low-resolution analysis or spectral types (Section 3.1). Our memberships are based on RV criteria, presence of Li i λ6708 in absorption, and the accretion criteria from White & Basri (2003).

In H07b, we studied the disk population in the region covered by the four channels of IRAC. Since mosaics of channels 4.5 μm and 8.0 μm have a 65 displacement to the north from the mosaics of channels 3.6 μm and 5.8 μm, the region with the complete IRAC data set is smaller than the field of view of individual IRAC mosaics. MIPS observations in the σ Orionis cluster also cover more area in the north–south direction in comparison to the region covered by the complete IRAC data set. A more detailed description of MIPS and IRAC data can be found in H07b.

We searched for infrared excesses at 24 μm in the newly identified candidates within the photometric sample (Table 2) reanalyzing the MIPS observation of the σ Orionis cluster. We followed the procedure described in Hernández et al. (2007a, 2007b, 2008, 2009, 2010) to identify stars with excess at 24 μm. Figure 10 shows the color–color diagram used to select new disk bearing candidates of the cluster. Photospheric limits (dotted lines) were defined by H07b using the typical K-[24] color of diskless stars detected in the MIPS observation. An arbitrary limit between optically thick disks (II) and debris disks (dashed line) was defined using the K-[24] color of a sample of known debris disks located in several young stellar groups (Hernandez et al. 2011). Open squares represent stars not included in the previous MIPS analysis (H07b). Out of 14 new disk bearing candidates, 12 have 24 μm infrared excess expected in stars with optically thick disks. The remaining two stars show 24 μm expected in stars with debris or evolved disks (K-24 ≲ 3.0; Hernandez et al. 2011). Two disk bearing candidates are potential galaxies based on the profile classification of UKIDSS (red X's). Spitzer photometry and disk type for each source are provided in Table 8.

Zoom In Zoom Out Reset image size

Independently of the spectral type, WH_α_10% is an indicator of accretion (White & Basri 2003; Jayawardhana et al. 2003). Studying very low-mass young objects, Natta et al. (2004) found that WH_α_10% can be used to roughly estimate accretion rates ($\dot{M}_{{\rm acc}}$). The relation of $\dot{M}_{{\rm acc}}$ as a function of WH_α_10% from Natta et al. (2004) indicates that stars accreting below the detectable level have log ($\dot{M}_{{\rm acc}}$) ≲ −10.3 M_☉ yr⁻¹ and log ($\dot{M}_{{\rm acc}}$) ≲ −11 M_☉ yr⁻¹ for the accretion cutoff of 270 km s⁻¹ (White & Basri 2003) and 200 km s⁻¹ (Jayawardhana et al. 2003), respectively, although a large chromospheric contamination is expected if the mass accretion rates have such low levels (Ingleby et al. 2011).

Figure 11 shows the relation between the WH_α_10% versus the IRAC SED slope determined from the [3.6]–[8.0] color. The horizontal solid line indicates the limit between optically thick disks and evolved disk objects based on their infrared excess at 8 μm (Lada et al. 2006; H07b). We plotted stars with optically thick disks (open circles), evolved disks (open squares), and TDs (open triangles) from H07b. In general, diskless stars have WH_α_10% < 270 km s⁻¹, which is the accretion cutoff proposed by White & Basri (2003). We found eight objects with infrared excess at 8 μm consistent with optically thick disks but that are not accreting or are accreting below the detectable level (hereafter very slow accretor). These stars are SO 467, SO 738, SO 435, SO 674, SO 451, SO 247, SO 697, and SO 662. The star SO 823 was also included as a very slow accretor candidate. This star has a strong absorption component at Hα, we were unable to measure WH_α_10% and thus it was not included in Figure 11. On the other hand, low-resolution spectra of SO 823 shows a single emission Hα profile, which suggests variability in the profile of Hα. Additionally, photometric measurements from CVSO indicate that SO 823 is a variable star with amplitude of ΔV ∼ 2.5 mag. Objects such as UX Ori-type and AA Tau-type show photometric and spectroscopic variability in which Hα could change from single emission line profile to a profile with a strong central absorption component.

Zoom In Zoom Out Reset image size

Figure 12 shows the distribution on the IRAC color–color diagram ([5.8]–[8.0] versus [3.6]–[4.5]) for stars studied in Figure 11. The very slow accretor candidates (including the star SO 823) were plotted with solid circles. Stars with optically thick disks classified as accretor and very slow accretor candidates fall in similar region in this plot. Thus, very slow accretors candidates have similar infrared excesses in the IRAC bands in comparison with stars with optically thick disks classified as accretors using the accretion cutoff proposed by White & Basri (2003).

Zoom In Zoom Out Reset image size

Low-resolution spectroscopic analysis (Section 3.1) supports the low accretion level for stars SO 823, SO 467, SO 435, and SO 451. Moreover, in Figure 3, we can identify four additional very low accretor candidates (SO 566, SO 598, SO 682, and SO 967) bearing optically thick disks that were classified as non-accretor following the criteria from Barrado y Navascués et al. (2003). High-resolution studies are necessary to determine whether these stars are accreting or not (e.g., Espaillat et al. 2008; Ingleby et al. 2011).

Figure 13 shows normalized SEDs and H_α profiles of the very slow accretor candidates detected in our high-resolution spectra. For comparison, we show SEDs and profiles of a diskless star (SO 669) and two stars classified as accretor in Figure 11 (SO 397 and SO 462). Particularly, SO 462 has WH_α_10% on the accretion cutoff defined by White & Basri (2003). The star SO 397 exhibits infrared excesses below the median SED for class II stars in the σ Orionis cluster. High inclination of the disk (edge-on) or dust settling could be responsible for the relatively small infrared excesses observed in this star. The star SO 823 has a jump in the SED between the 2MASS and the IRAC measurements, which could be caused by variability or an unresolved binary. Some stars like SO 467, SO 435, SO 451, and SO 462 exhibit asymmetries in the Hα profile that are characteristic of accreting disks. The emission of Hα could have a contribution from the stellar chromosphere or the stars could have variable accretion rate, so the WH_α_10% could vary and would not be a robust quantitative indicator of accretion (e.g., Nguyen et al. 2009a, 2009b). Thus, we need additional studies to understand whether these stars have stopped accreting or if they are in a passive phase in which accretion is temporally stopped or if accretion is below the measurable levels in WH_α_10%. Previous studies have found very low accretors stars in other young stellar populations. Studying the accretion properties in Chameleon I and ρ Oph, Natta et al. (2004) found a population of very low-mass objects with evidence of disks but no detectable accretion activity estimated using several indicators of accretion. Nguyen et al. (2009a) also found a group of stars in Chameleon I (∼2 Myr old) with excess at 8 μm and accretion rates below the measurable levels in WH_α_10%. They also found that non-accreting objects with disks do not seem to exist in the Taurus-Auriga star forming region.

Zoom In Zoom Out Reset image size

On the other hand, Figure 11 shows two diskless stars (SO 616 and SO 229) exhibiting WH_α_10% larger than the accretors limits. One of these stars, SO 229, was identified as a double lined spectroscopic binary from the cross-correlation function while the star SO 616 was recognized as possible binary based on RV variability. These two stars have Li i λ6708 in absorption and spectral types of G2.5 and K7, respectively (see Section 3.1). In Figure 14, we compare the Hectochelle spectra of SO 229 and SO 616 to stars with similar spectral types (G3 for SO 229 and K7 for SO 616). It is apparent that SO 299 and SO 616 have wider photospheric features. It is possible that wider spectral features are combined features from components in binary stars with similar spectral types. Another possibility is that the broadening of spectral lines in these objects can be caused by fast rotation. Non-accreting objects with projected rotational velocities larger than ∼50 km s⁻¹ can generate WH_α_10% > 270 km s⁻¹ (see Figure 5 of Jayawardhana et al. 2006). A multi-epoch spectroscopic analysis will help to reveal the nature of these objects.

Zoom In Zoom Out Reset image size

Low resolution spectroscopic analysis (Section 3.1) shows three diskless stars (SO 229, SO 1123, and SO 1368) that mimic the Hα width expected in stars with accretion disks (see Figure 3). The star SO 299 shows broad photospheric features that could be produced by binarity or fast rotation (Figure 14). The star SO1123 has been classified as spectroscopic binary (Caballero 2007) and fast rotator with a very high projected rotational velocity (Alcalá et al. 2000). The star SO 1368 is a variable star (Mayne et al. 2007) with spectral type K6.5 and still does not have information about multiplicity or rotation. These diskless objects with Hα in emission could have substantial contribution produced by strong chromospheric activity related to fast rotation.