A MAGELLAN MIKE AND SPITZER MIPS STUDY OF 1.5–1.0 M_☉ STARS IN SCORPIUS-CENTAURUS

Planets are believed to form in circumstellar disks of gas and dust. Recent high contrast imaging has revealed the presence of a 9 ± 3 M_Jup planet in the dusty, debris disk around β Pic at 8–13 AU from the central star, suggesting that giant planets may form at ages as young as ∼12 Myr (Lagrange et al. 2010). Although planets have not been widely detected in young debris disks (with ages 5–100 Myr), coronagraphic imaging has revealed the presence of structures, such as brightness peaks, asymmetries, and warps, that indicate that planets may have formed or may be forming (Wyatt 2008). Since debris disks are expected to be common around stars with ages ∼10–20 Myr, many Spitzer Multiband Imaging Photometer for Spitzer (MIPS) photometric surveys (e.g., Rebull et al. 2008; Low et al. 2005) have been carried out to search for infrared excess around stars in nearby (<75 pc from the Sun) moving groups with ages <100 Myr (e.g., the ∼10 Myr old TW Hya association, ∼12 Myr old β Pic moving group, etc.). Observations of stars in nearby, star-forming regions are needed to improve statistics.

Numerical coagulation N-body models have been developed to describe the formation of oligarchs (1000 km radius planetary embryos) in outer planetary systems and the destruction of planetesimals by collisions in circumstellar disks (e.g., Kenyon & Bromley 2005, 2008). These models assume that disks initially possess (1) planetesimals (with radii between 1 m and 1 km) at 30–150 AU from the central star and (2) gas that maintains circular orbits and low relative velocities between small bodies, leading to constructive collisions that build successively larger planetesimals. They predict that large bodies continue to grow until they reach a radius of ∼1000 km (at an age of ∼10⁴ years) and are massive enough to gravitationally perturb smaller objects (with radii 0.1–10 km) into crossing orbits, generating debris dust that can be detected via thermal emission. The models are able to reproduce the general amplitude and evolution of 24 μm excess around A-type stars at ages >20 Myr (Su et al. 2006); however, they may or may not correctly describe the production of debris dust at younger ages. In particular, the Kenyon & Bromley (2005, 2008) models predict a peak in 24 μm excess at an age of ∼10–20 Myr.

The Scorpius-Centaurus OB association (ScoCen), with typical stellar distances of ∼100–200 pc, is the closest OB association to the Sun and contains three subgroups: Upper Scorpius (US), Upper Centaurus Lupus (UCL), and Lower Centaurus Crux (LCC), with estimated ages of ∼10 Myr (Pecaut et al. 2011), ∼15 Myr, and ∼17 Myr (Mamajek et al. 2002), respectively. The close proximity of ScoCen and the age of its constituent stars make this association an excellent laboratory for studying the formation and evolution of planetary systems. Several hundred candidate members have been identified to date; although, the association probably contains thousands of low-mass members. Member stars with spectral type F and earlier have been identified using moving group analysis of Hipparcos positions, parallaxes, and proper motions (de Zeeuw et al. 1999), while later type members have been identified using youth indicators (i.e., high coronal X-ray activity and large lithium abundance; Preibisch & Mamajek 2008; Slesnick et al. 2006). Demographic studies of infrared excess in ScoCen, combined with demographic studies of other young clusters, is expected to provide constraints on debris production in young debris disk systems and input into self-stirred disk models.

We report the results of a Spitzer MIPS 24 μm and 70 μm survey of 182 F- and G-type Hipparcos common proper motion members of ScoCen, building on our initial results (Chen et al. 2005). Our previous study included preliminary MIPS 24 and 70 μm photometry of 41 candidate ScoCen single and binary systems. The current study has been expanded to include (1) Magellan/Magellan Inamori Kyocera Echelle (MIKE) stellar radial and rotational velocity, lithium equivalent width, and Ca ii activity measurements for 181 candidate ScoCen members, (2) membership analysis for the F- and G-type candidate ScoCen members in the de Zeeuw et al. (1999) sample, and (3) new MIPS 24 μm and 70 μm observations of 142 candidate ScoCen single and binary systems. We re-examine membership using Li abundance to determine stellar youth and convergent point analysis to determine whether stellar luminosity, radial velocity, and proper motion are consistent with ScoCen membership. Refinements in the calibration of the MIPS data have allowed us to improve the accuracy of the measured fluxes; therefore, we rereduce and reanalyze the complete sample. We list the targets for the full sample, along with their spectral types, distances, and subgroup memberships in Table 1.

Finally, our ScoCen sample is sufficiently large that it can be used to search for signatures of mass-dependent disk evolution. The physical properties of central stars are expected to shape the evolution of circumstellar disks. For example, Spitzer photometric measurements of 204 stars in Upper Sco indicate that disk properties depend on spectral type. At ∼10 Myr, M-type stars possess optically thick, accreting disks; F- and G-type stars do not apparently possess disks; and B- and A-type stars posses optically thin, debris disks, suggesting that disk evolution is most advanced around intermediate-mass stars (Carpenter et al. 2006). During the debris disk phase, (1) infrared excess is expected to be dependent on the luminosity of the central star with more luminous stars warming larger surface areas within their circumstellar disks, producing larger infrared excesses and (2) the timescale for disk evolution is expected to be dependent on the orbital timescale with disks evolving more rapidly around higher mass stars. Therefore, we also search for disk trends as a function of stellar properties to determine how stellar properties impact the production and removal of debris dust around main-sequence stars.

The Hipparcos satellite enabled high-precision measurements of stellar position, parallax, and proper motion for stars with V-band magnitudes, m_V < 9, providing the ability to identify candidate members of nearby OB associations with spectral types later than B for the first time. De Zeeuw et al. (1999, hereafter dZ99) analyzed the Hipparcos measurements of stars in a dozen nearby OB associations using de Bruijne's refurbished convergent point method and Hoogerwerf & Aguilar's "Spaghetti method" to determine the average position and space motions of each association, including the US, UCL, and LCC subgroups of ScoCen. By cross correlating the Hipparcos measurements of individual stars with mean subgroup properties, dZ99 carried out a detailed census of high- and intermediate-mass stars in ScoCen. In particular, they identified almost 200 probable new solar-like members of US (22 F-type, 9 G-type, 4 K-type, and 2 M-type stars), UCL (55 F-type, 25 G-type, 6 K-type, and 1 M-type stars), and LCC (61 F-type, 15 G-type, 6 K-type, and 1 M-type stars). However, they cautioned that up to ∼30% of their candidate members may be interlopers because the stellar radial velocities were not measured.

We sought to obtain MIPS 24 μm and 70 μm photometry of all of the solar-like ScoCen members identified by dZ99. In fact, we obtained MIPS observations of 13 F-type and 4 G-type stars in US. Carpenter et al. (2009b) observed eight F-type and two G-type dZ99 US candidates as part of a detailed study focusing on US. We obtained MIPS observations of 54 F-type, 25 G-type, 6 K-type, and 1 M-type star in UCL. The remaining one F-type star (HIP 73777) was observed by F. Low as part of a GTO program search for MIPS excess around nearby, young stars (Spitzer PID 72). Smith et al. (2006) published the excess sources discovered in this survey; HIP 73777 did not apparently possess a MIPS excess. We obtained MIPS observations of 61 F-type, 13 G-type, 6 K-type, and 1 M-type star in LCC. The remaining two G-type stars (HIP 62445 and HIP 66001) were observed by the formation and evolution of planetary systems (FEPS) legacy team led by M. Meyer. Carpenter et al. (2009a) published the MIPS results from this study; neither HIP 62445 and HIP 66001 apparently possess MIPS excess.

Similarly, we sought to obtain Magellan/MIKE spectra of all of the solar-like ScoCen members identified by dZ99. The higher angular resolution of the 6.5 m Magellan Clay Telescope at visual wavelengths additionally afforded us the opportunity to obtain spectra of primary and secondary stars in ScoCen binary systems that could not be resolved by Spitzer at mid-infrared wavelengths. Therefore, we collected MIKE spectra of individual components whenever possible. We obtained Magellan/MIKE observations of the majority of candidate solar-like ScoCen members that we observed with MIPS including the primary and secondary components of nine binary systems (HIP 63975, HIP 66285, HIP 70376, HIP 70833, HIP 72033, HIP 75924, HIP 77135, HIP 77520, and HIP 79610). We were not able to obtain Magellan/MIKE observations for three F-type (HIP 80663, HIP 80921, and HIP 81136), four G-type (HIP 78133, HIP 78432, HIP 81447, and HIP 81775), and two K-type (HIP 77157 and HIP 82135) stars in UCL; and two F-type stars in LCC (HIP 59781 and HIP 68534).

2.1. MIPS Observations

2.2. MIKE Observations

Since the dZ99 analysis was published, we have obtained stellar radial velocities for the majority of the candidate ScoCen members and van Leeuwen (2007) have rereduced the Hipparcos data and re-derived the parallaxes for all of the stars in the Hipparcos catalog. Therefore, we re-examine ScoCen membership for the targets in our MIPS sample using (1) new and previously published lithium abundance measurements, (2) previously published stellar spectral types and new stellar luminosity estimates (based on the updated van Leeuwen 2007 distances), (3) our new radial velocity measurements, and (4) previously published Hipparcos proper motions (see Table 1). We list our best estimates for stellar spectral type (drawn from the literature) and updated distances (drawn from van Leeuwen 2007) for the single and binary target systems in our MIPS sample in Table 1. No published spectral types existed for the secondary stars in the binary systems we observed; therefore, we determined their spectral types by convolving our MIKE spectra to a lower resolution (R ∼ 1000) and comparing them with SMARTS 1.5 m spectral standards. We calculated the reddening implied by our spectral types and found that it was consistent with that estimated for other stars in our sample (A_V ∼ 0.1). We estimated that the uncertainty for these spectral types is two subtypes.

The presence of strong lithium absorption in the spectra of late-type stars has long been used as a diagnostic of stellar youth (e.g., Herbig 1965). Mamajek et al. (2002) measured Li 6707 Å equivalent widths toward 30 G- through K-type dZ99 candidate UCL and LCC members to determine whether the lithium observed toward these stars was consistent with youth and therefore membership in ScoCen. By plotting the lithium equivalent width for these stars as a function of spectral type and comparing their observations with those of other young clusters (e.g., IC 2602 at 30 Myr, the Pleiades at 70–125 Myr, and M34 at 250 Myr), Mamajek et al. (2002) concluded that the lithium abundances toward HIP 63797, HIP 68726, HIP 74501, HIP 77015, HIP 79610, and HIP 81775 were too small to be consistent with ScoCen membership. More recent spectroscopic observations of HIP 56420, HIP 65891, HIP 66285, HIP 66782, HIP 70833, HIP 75916, and HIP 76197 suggest that the lithium abundances for these sources or their cooler companions (HIP 70833B and HIP 66285B) are also too small to be consistent with ScoCen membership (M. Pecaut 2010, private communication). We measured lithium equivalent widths from our Magellan/MIKE spectra and plotted their values as a function of stellar effective temperature (see Figure 1). Stars with T_eff < 6000 K and EQ(Li) <60 mÅ are inconsistent with ScoCen membership and possess lithium membership flags (Li) set to "N" in Table 1. Stars with 6000 K < T_eff < 6600 K and EQ(Li) <40 mÅ or 6600 K < T_eff < 7200 K and EQ(Li) <20 mÅ are not lithium-rich and possess lithium membership flags (Li) set to "Y?" in Table 1.

Zoom In Zoom Out Reset image size

We examined the measured spectral types, distances, and stellar luminosities to determine whether measured stellar properties are consistent with ScoCen membership. Stars with spectral types and luminosities consistent with ScoCen membership possess stellar luminosity membership flags (L_*) set to "Y" in Table 1. Giant stars with luminosity class III are expected to be significantly older than the mean ScoCen subgroup ages and therefore to be interlopers. We plot the Hertzsprung–Russell diagram (HRD) positions for all of the giant stars in our sample (see Figure 2). We estimate stellar effective temperatures from stellar spectral types using the Kenyon & Hartmann (1995) conversion. We calculate stellar luminosities from extinction-corrected V-band magnitudes, Hipparcos distances, and Flower (1996) bolometric corrections. The HRD positions for the majority of giant stars are consistent with an older age; however, three F-type giants have HRD positions that are inconsistent with their luminosity class: HIP 56227, HIP 62428, and HIP 72164. HIP 56227 is a F0III star (Houk & Cowley 1975); however, its HRD position is clearly near the zero-age main-sequence (ZAMS) therefore we have set its luminosity membership flag to "Y" in Table 1. HIP 62428 is a F0III star (Houk & Cowley 1975); its HRD position is consistent with the DM97 1 Myr isochrone, significantly above the other ScoCen F-stars. Since this star does not appear to have an age of 1 Myr—it is not embedded and does not possess an infrared excess, we set its L_* flag to "N." HIP 72164 is a F2III/IV star (Houk 1982); however, its HRD position (with large errors) is consistent with the ZAMS therefore we have set its L_* flag to "Y."

Zoom In Zoom Out Reset image size

We calculated updated UVW velocities for the subgroup members of de Zeeuw et al. (1999), using their updated Hipparcos positions, parallaxes, and proper motions (van Leeuwen 2007), previously published radial velocities (Gontcharov 2006), and the new radial velocity measurements described here. With these revised UVW velocities, we calculated new mean subgroup velocities using an iterative clip to remove obvious outliers. The new subgroup velocities are listed in Table 3, and represent the most precise modern values ever calculated, taking into account the best available astrometry and radial velocities. We also compared the projected and radial motions of the stars to that for an "ideal" member following the convergent point techniques discussed in Mamajek (2005). In our calculations, we assumed that the intrinsic one-dimensional velocity dispersion of each subgroup is 1.3 km s⁻¹, based on the results of de Bruijne (1999) and Madsen et al. (2002). We flagged any star with a radial velocity more than 3σ away from its predicted radial velocity, based on the star's position and subgroup membership, with a "Y?" in the RV Membership column of Table 1. We could not exclude stars with inconsistent radial velocities as non-members because we obtained only one epoch of MIKE data for each star; stars with inconsistent measured radial velocities may be members of binary or multiple systems. A second epoch of measured radial velocities is needed for stars with discrepant radial velocities to determine whether they are members of multiple systems or interlopers. Similarly, we flagged any star with a proper motion more than 2σ away from its predicted proper motion with a "Y?" in the PM Membership column of Table 1. In our analysis, we defined proper motion as the velocity in the plane of the sky that is perpendicular to the velocity toward the convergent point. One system (HIP 62677) possesses a 3σ discrepant proper motion; however, we do not reject this system because it is a member of a wide binary system with measured orbital motion during the past century.

Based on lithium, stellar luminosity, radial velocity, and proper motion membership flags, we made a summary assessment of the likelihood of ScoCen membership for all of the stars in our sample. Stars with any membership flag set to "N" possess a final membership flag of "N." We find that 0/17 Upper Sco, 18/86 UCL, and 13/81 LCC candidate single and binary systems in our sample are excluded as members. Combining all of the ScoCen MIPS observations together, our final sample is composed of 27 members of Upper Sco (including 21 F-type and 6 G-type members), 69 members of UCL (including 56 F-type, 12 G-type, and 2 K-type members), and 71 members of LCC (including 59 F- and 11 G-type members).

We estimate the stellar photospheric fluxes for our sample based on Two Micron All Sky Survey (2MASS; Cutri et al. 2003) K_s-band magnitudes and intrinsic main-sequence colors calculated by E. Mamajek.⁶ First, we assembled the Hipparcos B- and V-band photometry, Cousins I-band photometry (where available), and 2MASS J-, H-, and K_s-band photometry and constructed measured B−V, V − I_c, V−J, V−H, and V − K_s colors for each star. Second, we calculated the extinction in each color assuming that the stars possess intrinsic main-sequence colors. Third, we calculated an average visual extinction, A_V, and its uncertainty based on the standard deviation of the extinction measurements. Fourth, we extrapolated the average A_V to the K_s band assuming that A_K = 0.116 A_V, consistent with R_V = 3.1 and a Cardelli et al. (1989) extinction law. Fifth, we corrected the measured 2MASS K_s-band magnitudes for extinction. Finally, we used Kurucz models to estimate the predicted 24 μm and 70 μm fluxes based on the extinction-corrected K_s-band magnitudes and stellar spectral types, assuming the Kenyon & Hartmann (1995) conversion between spectral type and effective temperature. For example, a star with T_eff = 7200 K, is expected to possess F_ν(K_s)/F_ν(24 μm) = 93.9 and F_ν(K_s)/F_ν(70 μm) = 857.7.

For comparison with our measured (but not color-corrected) fluxes, we list the predicted photospheric 24 and 70 μm fluxes integrated over the MIPS bandpasses, F_ν, in Table 2. We calculate the excess significance of each detected source, χ = (measured flux−predicted flux)/uncertainty, where the uncertainty includes absolute calibration uncertainty (2%), repeatability uncertainty (0.4%), the photospheric model uncertainty (∼3%), and statistical uncertainty of the 24 μm photometry summed in quadrature. We list excess objects with χ ⩾ 7 at 24 and/or 70 μm in Table 4. We verify our stellar atmosphere model fits by examining the K_s−[24] colors of our sources. All of our excess sources possess K_s−[24] > 0.3 mag. We plot the K_s−[24] color as a function of J − H color (as a proxy for spectral type) for all of the sources in our study in Figure 3. We show the distribution of significance of the 24 μm excesses (χ₂₄ = (measured F_ν(24 μm)−predicted F_ν(24 μm))/measured σ_F24) in Figure 4. Our K_s−[24] selection criteria robustly identified sources with large excesses but may not identify sources with weaker excesses.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size