Main-Sequence Fitting Distance to the σ Ori Cluster W. H. Sherry National Solar Observatory, Tucson, AZ 85719 8 [email protected] 0 0 2 F. M. Walter n a J SUNY Stony Brook, Stony Brook, NY 11790 6 1 [email protected] ] h p - S. J. Wolk3 o r t s [email protected] a [ 1 and v 5 8 5 N. R. Adams3 2 . 1 0 [email protected] 8 0 Center for Astrophysics, 60 Garden Street, Cambridge, Cambridge, MA 02138 : v i X Received ; accepted r a – 2 – ABSTRACT The σ Ori cluster is an unbound aggregate of a few hundred young, low–mass stars centered on the multiple system σ Ori. This cluster is of great interest becauseitisatanagewhenroughlyhalfofthestarshavelosttheirprotoplanetary disks, and the cluster has a very large population of brown dwarfs. One of the largest sources of uncertainty in the properties of the cluster is that the distance is not well known. The directly measured Hipparcos distance to σ Ori AB is 350+120 pc. On the other hand, the distance to the Orion OB1b subgroup (of −90 which σ Ori is thought to be a member), 473±40 pc, is far better determined, but it is an indirect estimate of the cluster’s distance. Also, Orion OB1b may have a depth of 40 pc along our line of sight. We use main sequence fitting to ± 9 main sequence cluster members to estimate a best fit distance of 420 30 pc, − ± assuming a metallicity of 0.16 0.11 or 444 pc assuming solar metallicity. A distance as close as 350 pc is inconsistent with the observed brightnesses of the cluster members. At the best fit distance, the age of the cluster is 2–3 Myrs. Subject headings: open clusters: general — young stars: individual(σ Ori) – 3 – 1. Introduction The bright O9.5V star σ Ori is a trapezium–like system with six known components. The brightest component, σ Ori AB (V=3.m80), is a 0.25′′ binary (Horch et al. 2001) with an O9V primary and a B0.5V secondary (Edwards 1976) and an orbital period of ∼170 years or ∼158 years (Heintz 1974, 1997). The O9V primary, σ Ori A, is itself a double lined spectroscopic binary1 (Peterson 2006; Bolton 1974; Miczaika 1950). The spectral type of σ Ori C is A2V. The D and E components are B2V stars with V≃6.8 and ≃ 6.6 respectively (see Table 1). The σ Ori cluster was first recognized as a group of high–mass stars by Garrison (1967), and as cluster of low–mass pre-main-sequence stars by Walter, Wolk, & Sherry (1998). Continuing work on the cluster has revealed a young cluster of several hundred low–mass stars (Sherry et al. 2004; Burningham et al. 2005; Kenyon et al. 2005) and a rich population of brown dwarfs (B´ejar et al. 1999, 2001; B´ejar, Zapatero Osorio, & Rebolo 2004; Zapatero Osorio et al. 2002). About one third to one half of the stars retain their accretion disks (Oliveira, Jeffries & van Loon 2004; Hern´andez, et al. 2007). The cluster is considered part of the Orion OB1b association. Age estimates for Orion OB1b range from less than 2 Myrs (Brown, de Geus, & de Zeeuw 1994) up to 7 Myrs (Blaauw 1991). This is an exceptionally interesting age because it is the age when protoplanetary disks are making the transition from optically thick to optically thin and may be the age when giant planets form. The more accurately the cluster age can be measured, the tighter the constraint on disk lifetimes and the time available for giant planet formation will be. See Walter, et al. (2008) for a recent review of observational data on the low–mass population of the σ Ori cluster. 1We wish to thank Deane Peterson for sharing unpublished results of his observations of σ Ori AB with us. – 4 – The most significant source of uncertainty for the age of this cluster is the uncertain distance to the cluster. Many authors adopt the Hipparcos distance of 350+120 pc for σ Ori −90 (Perryman et al. 1997) as the distance to the center of the cluster. This has the virtue of being a direct measurement of the distance to the cluster, but it has an uncertainty of 30% (see Schroeder et al. (2004) for a discussion of biases on Hipparcos parallaxes of O stars). Others use the Hipparcos distance to the Orion OB1b subgroup (of which σ Ori is ± a member), 473 40 pc (deZeeuw et al. 1999), as the distance to the cluster. This value is more precise because it is the average distance to 42 members of the association, yet it is only an indirect measurement of the distance to σ Ori. Similarly, Hern´andez, et al. (2005) ± find a distance of 443 16 pc from the Hipparcos parallaxes of the combined Orion OB1b ∼ and 1c subgroups. The Orion OBIb association has a size of 30–40 pc across the sky, so it is likely to have a similar depth along our line of sight. The cluster could easily lie >20 pc in front of or behind the center of Orion OBIb. It would be preferable to have a direct measure of the cluster’s distance that is more precise than the Hipparcos distance to σ Ori. In a brief abstract, Garrison (1967) said that main sequence fitting to 15 B stars near σ Ori yielded a narrow main sequence at a distance modulus of 8.2 (440 pc). Garrison did not correct for the small values of reddening that some of the likely cluster members have. Garrison does not appear to have ever published a more detailed description of this result. In this paper we re–examine the main sequence fitting distance for the σ Ori cluster using published spectroscopy and photometry for the stars that lie within 30′ of σ Ori AB and have spectral types earlier than F0. – 5 – 2. Analysis and Data We searched the literature2 for photometry and spectral types for all of the stars within 30′ of σ Ori AB that have spectral types earlier than F0. Several of the stars have B and V photoelectric photometry from multiple observations taken prior to 1980. Table 1 collects our adopted colors and magnitudes for the 19 stars we selected. We have also obtained new spectra of those stars whose spectral types or colors appeared discrepant. These spectra, obtained with the SMARTS/CTIO 1.5m RC spectrograph, have 1.6˚A resolution between about 3880 and 4500˚A. Spectral types have been determined through visual comparison with a grid of spectral standards obtained with the same same instrument. 2.1. Magnitudes and B−V Colors for Individual Stars Several of the stars listed in Table 1 can not be directly compared to the main sequence because they are binaries or have known problems with their photometric data. Their photometry must be corrected before being included in any estimate of the cluster distance. 2.1.1. σ Ori Aa, σ Ori Ab, and σ Ori B σ Ori A and σ Ori B are the two brightest cluster members. This visual pair, with a separation of 0.25′′, lies roughly at the center of the cluster. Horch et al. (2001) used speckle observations to derive a V band magnitude difference of 1.m25. Similar results were found by Ten Brummelaar et al. (2000) who report ∆V=1.m24. Given a combined magnitude of V=3.m8 and an E(B−V) of ∼0.m06 (see section 2.2), a magnitude difference 2Our initial search relied heavily upon the SIMBAD data base. – 6 – of 1.m25 indicates that σ Ori A has V ≃3.m91 while σ Ori B has V ≃5.m16. 0 0 σ Ori A is itself a double-lined spectroscopic binary. Bolton (1974) estimated a ∆V of ∼0.m5 between σ Ori Aa and σ Ori Ab. This would require σ Ori Aa and Ab to have V ∼4.m4 and ∼4.m9, respectively. There is no measured spectral type for σ Ori Ab yet, so 0 its UBV colors are unknown. Assuming that it is on the ZAMS, the spectral type a star that is 0.m5 fainter than an O9V star should B0V. Edwards (1976)quotesspectraltypesofO9VandB0.5Vforσ OriAandB,respectively. Assuming an uncertainty of ±0.5 subtypes, we adopt values of (B−V) ≃ −0.m31±0.01 for 0 σ Ori Aa and (B−V) ≃ −0.m28±0.02 for σ Ori B (Kenyon & Hartmann 1995). 0 The observed B−V for σ Ori AB is −0.m24. Assuming an intrinsic color (B−V) = 0 −0.m30 for σ Ori AB , E(B−V) must be ∼0.m06. This is consistent with the observed N(H) column density of 3.3×1020 cm−2 (Fruscione et al. 1994; Bohlin et al. 1983). The uncertainty on the column density is 20%. This value is also consistent with published estimates of the line of sight reddening to σ Ori AB (e. g. Lee (1968)) and with the N(H) column density of 3.6×1020 cm−2 measured by Shull & Van Steenberg (1985). 2.1.2. σ Ori C Greenstein & Wallerstein (1958) measured the B and V magnitudes of σ Ori C. They noted that the observed B−V color of σ OriC, −0.m02, is too blue for its spectral type, A2V, which should have (B−V) =0.m06 (Kenyon & Hartmann 1995). Greenstein & Wallerstein 0 (1958) account for the exceptionally blue color of σ Ori C as the result of scattered light from σ Ori AB (11′′ away) in the aperture of the photometer. The published V magnitude is 8.m79 (Greenstein & Wallerstein 1958), but that measurement was also contaminated by scattered V band light from σ Ori AB. Greenstein & Wallerstein (1958) estimate that, – 7 – after correcting for scattered light, the true V magnitude of σ Ori C is ∼9.m2. Sherry et al. (2008) report recent differential V and I photometry for stars within C 6′ of σ Ori AB. While σ Ori AB is saturated, C, D, and E are not saturated. They find that σ Ori C is 2.m63±0.01 magnitudes fainter than σ Ori D, and 2.m74±0.01 magnitudes fainter than σ Ori E in the V band. Using the V magnitudes from Table 1 for σ Ori D and σ Ori E yields V=9.m44 and V=9.m40 respectively for σ Ori C. These measurements were done using small apertures which are not significantly contaminated by scattered light. We will adopt V=9.m42±0.02 for the magnitude of σ Ori C (uncorrected for reddening). 2.1.3. σ Ori D The three papers that report UBV photometry for σ Ori D quote significantly different V magnitudes, and to a lesser extent, colors. Greenstein & Wallerstein (1958) report a V magnitude of 6.m62. Eighteen years later, Vogt (1976) reported a V magnitude of 6.m73. Guetter (1979) reported a V magnitude of 6.m84. These discrepant V magnitudes may indicate variability, or that stray light from σ Ori AB affected the measurements of σ Ori D. Mermilliod & Mermilliod (1994) list a weighted averaged of the UBV photometry for σ Ori D which we have used in Table 1. 2.1.4. σ Ori E The B2Vp star σ Ori E has unusually strong He lines (Greenstein & Wallerstein 1958) which make it spectroscopically peculiar. It has variable line widths and photometric variations (∆mag∼0.m03–0.m15) with a period of 1.19 days (Hesser, Walborn, & Ugarte 1976; Townsend, Owocki, & Groote 2005). There are conflicting opinions as to whether σ Ori E is physically associated with σ Ori AB. Much of the uncertainty surrounding – 8 – the membership of σ Ori E with σ Ori AB follows from uncertainty on the mass and evolutionary status of σ Ori E. Greenstein & Wallerstein (1958) estimated the absolute magnitude of σ Ori E using three different methods, thereby placing the star on or near the main sequence (which would put σ Ori D, and σ Ori E at the same distance). They found that the equivalent widths of two components of the interstellar K line are similar for both σ Ori AB and σ Ori E, as are the radial velocities. Attempts to model the UV flux from the V band flux and spectroscopic features lead to models of σ Ori E that have ∼ ∼ masses that are far too small ( 3 M ) for an early B main sequence star (M 9 M ). ⊙ ⊙ The main reason for the low masses in these models is that the profiles of the Balmer and helium lines indicate a low gravity (Hunger, Heber, & Groote 1989). More recent models postulate emission from plasma clouds magnetically confined above the photosphere (Townsend, Owocki, & Groote 2005) which may explain the discrepancy between the gravity estimated from line profiles and data that indicate that σ Ori E is a main sequence star. Given the significant uncertainties on the models, we take the observations indicating that σ Ori E is a main sequence star with a normal mass and radius for its spectral type at face value. A B2V star has (B−V) of −0.m24 (Kenyon & Hartmann 1995). The measured 0 B−V for σ Ori E is −0.m18 (Guetter 1979), which makes E(B−V) 0.m06. This is consistent with the observed N(H) column density of 4.5×1020 cm−2 (Fruscione et al. 1994; Shull & Van Steenberg 1985). The uncertainty on the column density is 20%. 2.1.5. BD −02 1323C and HD 294272 − BD 02 1323C was not found by our initial SIMBAD search for early type stars within 30′ of σ Ori. This star came to our attention because SIMBAD notes that HD 294272 is a member of a triple system, ADS 4240 (Aitken 1932). ADS 4240A is HD 294271. – 9 – ADS 4240B is HD 294272 which is separated from HD 294271 by ∼68′′. ADS 4240C (BD −02 1323C) is separated from HD 294272 by 8.5′′. SIMBAD listed a V magnitude of 10.m3 for BD −02 1323C, and no other photometric measurements. This value is not correct. − − Guetter (1979) reported photometry for BD 02 1323A and BD 02 1323B. Guetter (1981) used the same names when he reported the spectral types. SIMBAD, which does − − not recognize the names BD 02 1323A and BD 02 1323B, assigned the Guetter (1979) − − photometry for BD 02 1323B (ADS 4240C) to BD 02 1323 which is ADS 4240B or HD 294272. Consequently, HD 294272 (ADS 4240B) was listed in SIMBAD as having the − photometry and spectral type of BD 02 1323C (ADS 4240C). The measurements for the two stars from Guetter (1979, 1981) and Mermilliod & Mermilliod (1994) have been correctly assigned in Table 1. 2.1.6. HD 294273 & HD 294279 We obtained new spectra for HD 294273 & HD 294279 since the only published spectral types that we could find are A2 and A3, respectively, in the HD catalog. The Ca II K lines are far too strong for early A spectral types. The revised spectral types are early F (F0-F3) for HD 294279 and A7 for HD 294273. We do not assign a luminosity class, but it is likely that they are both class V. 2.2. Reddening We estimate the reddening of most of the stars in our sample by comparing the − − observed B V color to the (B V) expected for each star’s spectral type, and computing 0 A assuming R=3.1. Column 8 of Table 1 lists these A values for probable cluster V V – 10 – members. The mean E(B−V) for probable cluster members is 0.m06±0.005 (σ Ori A and B were treated as a single measurement). The median E(B−V) is also 0.m06. All of the probable cluster members have values of E(B−V) between 0.m04 and 0.m09, which is consistent with the 0.m015 uncertainty on E(B−V) for individual stars. The mean E(B−V) of 0.m06±0.005 makes the mean A of the cluster 0.m19±0.02, in agreement with the values V quoted by Lee (1968); Shull & Van Steenberg (1985) for σ Ori AB. Assuming that E(U−B)=0.72E(B−V), we expect a mean E(U−B)∼0.m04 mag. This − − is comparable to or smaller than the uncertainties on (U B) due to the change in (U B) 0 0 − from one spectral type to the next along the ZAMS. We found a median E(U B) of 0.m02±0.03 with most of the stars in Table 1 having values ranging from −0.m06 to 0.m07. HD 37633 (V1147 Ori), a known variable, does have an exceptionally large, negative E(U−B)=−0.m16. This may be due to its variability or a U band excess. Our spectrum − shows a spectral type of B9.5. HD 37699 also has a very blue U B color excess with E(U−B)=−0.m11. Since these two have values of E(U−B) that are significantly less than − zero, we excluded these stars from the calculation of the median E(U B). Our median − − E(U B) is consistent with our mean E(B V) and a normal reddening law. The small reddening has a disproportionate impact on main-sequence fitting on the V, B-V plane because the ZAMS has a slope ∆V ∼18 for stars near B5V. If we were to ∆(B−V) ∼ ignore the cluster’s reddening, we would find a distance that is 100 pc smaller than we find when correcting for the observed color excess. 2.3. The Main Sequence Turn–On Since the σ Ori cluster is roughly 3 Myrs old, most of the cluster members have not reached the ZAMS. Figure 1 compares the ZAMS of Turner (1976, 1979) to theoretical