Hα Emission From Active Equal-Mass, Wide M Dwarf Binaries 1

Transcription

1 PUBLICATIONS OF THE ASTRONOMICAL SOCIETY OF THE PACIFIC, 1: , 01 Decemer 01. The Astronomical Society of the Pacific. All rights reserved. Printed in U.S.A. Hα Emission From Active Equal-Mass, Wide M Dwarf Binaries 1 HEATHER C. GUNNING,,3 SARAH J. SCHMIDT,, JAMES R. A. DAVENPORT, SAURAV DHITAL, 5, SUZANNE L. HAWLEY, AND ANDREW A. WEST Received 01 August 08; accepted 01 Octoer 0; pulished 01 Novemer 19 ABSTRACT. We identify a sample of near-equal mass wide inary M dwarf systems from the SLoWPoKES catalog of common proper-motion inaries and otain follow-up oservations of their chromospheric activity as measured y the Hα emission line. We present optical spectra for oth components of 8 candidate M dwarf inaries, confirming their mid-m spectral types. Of those 8 coeval pairs, we find eight with Hα emission from oth components, three with weak emission in one component and no emission in the other, and 37 with two inactive components. We find that of the 11 pairs with at least one active component, only three follow the net trend of decreasing activity strength (L Hα =L ol ) with later spectral type. The difference in quiescent activity strength etween the A and B components is larger than what would e expected ased on the small differences in color (mass). For five inaries with two active components, we present 7 hr of time-resolved spectroscopy, oserved on the ARC 3.5-m over 1 different nights. For four of the five pairs, the slightly redder (B) component exhiits a higher level of Hα emission during the majority of the oservations and the redder ojects are the only components to flare. The full range of Hα emission oserved on these variale mid-m dwarfs is comparale to the scatter in Hα emission found in single-epoch surveys of mid-m dwarfs, indicating that variaility could e a major factor in the spread of oserved activity strengths. We also find that variaility is independent of oth activity strength and spectral type. Online material: color figures 1. INTRODUCTION M dwarfs are well known for haroring surface magnetic fields as strong as several kg (e.g., Johns-Krull & Valenti 199; Valenti & Johns-Krull 001), persisting on oth sides of the fully convective oundary ( M; Charier & Baraffe 1997). The strength of the magnetic fields in M dwarfs may e intrinsically tied to stellar rotation (on oth sides of the convective oundary; Doler et al. 00; Browning et al. 00), which has in turn een shown to depend on oth mass and age (Irwin et al. 011). These magnetic fields result in chromospheric heating, often oserved through the presence and strength of Hα emission (e.g., Hawley et al. 199). 1 This pulication is partially ased on oservations otained with the Apache Point Oservatory 3.5-meter telescope, which is owned and operated y the Astrophysical Research Consortium. Astronomy Department, Box , University of Washington, Seattle, WA, Space Telescope Science Institute, Box , 3700 San Martin Drive, Baltimore, MD, 118. Department of Astronomy, Ohio State University, 10 West 18th Avenue, Columus, OH, 310, schmidt@astronomy.ohio state.edu. 5 Department of Physical Sciences, Emry-Riddle Aeronautical University, 00 South Clyde Morris Blvd., Daytona Beach, FL, 311. Department of Astronomy, Boston University, 75 Commonwealth Avenue, Boston, MA, 015. The presence of Hα emission has een oserved to depend on oth mass (or its proxy, spectral type) and age: early-m dwarfs are active for a much shorter time ( 1 Gyr) than the mid- ( Gyr) and late-m dwarfs (>7 Gyr; West et al. 008). Chromospheric activity strength (usually quantified y the ratio of the luminosity of Hα to the olometric luminosity, L Hα =L ol ) is also related oth to spectral type and age. L Hα =L ol is relatively constant for M0 M dwarfs, then declines with later spectral type through M9 (West et al. 00, 011). The activity trends exhiit significant scatter compared to their dynamic range and measurement errors. Since the West et al. studies were ased on an ensemle of single-epoch measurements, the oserved spread in activity could originate from the intrinsic variaility of the Hα emission line, or could reflect some intrinsic range in magnetic activity strength. Repeat oservations of Hα emission in M dwarfs often show variaility (e.g., Bopp & Schmitz 1978), ut it has only recently een investigated over oth long and short timescales. On timescales of hours to months, 80% of active M dwarfs show significant variaility (Lee et al. 010; Kruse et al. 010), usually varying y factors of 1.5 inhα EW. In data averaged over 150 day windows, Gomes da Silva et al. (011) found variaility on 5 10 year timescales in 10 out of 30 M0 M5 dwarfs. These variaility timescales could e linked to stellar rotation, the formation and dissipation of active regions, or long-term variations of the magnetic field. The relationships etween 1081

2 108 GUNNING ET AL. Hα variaility, stellar mass, and age have not een thoroughly investigated. Bell et al. (01), for example, found no net trend in Hα variaility as a function of spectral type, and an increase in variaility with declining activity strength. It is unknown how the road trends in activity strength with respect to mass and age translate to individual systems, and whether the scatter in the oserved trends is due to Hα variaility or is intrinsic to the M dwarf magnetic field strengths. To proe M dwarf activity strength and variaility while controlling for mass and age, we have chosen to monitor a sample of near-equal mass, wide M dwarf inary systems selected from the Sloan Low-mass Wide Pairs of Kinematically Equivalent Stars catalog (SLoWPoKES; Dhital et al. 010). With presumaly the same age and metallicity and near-equal masses, these are nearly identical twins. In addition, these are wide ( AU) inary systems in which oth stellar components have evolved independently, without affecting each other s activity (as oserved in tighter systems, e.g., Meiom et al. 007; Morgan et al. 01). In this paper, we seek to characterize the differences in activity and variaility of Hα emission of M dwarfs y examining Hα activity over timescales of hours to days for these coeval twins. These inary M dwarfs are a unique test case for the relationships etween mass, age, and activity. In, we descrie the oservations to select and characterize our active M dwarf sample in addition to our time-series spectroscopic oservations of the active SLoWPoKES inaries. We descrie the individual Hα light curves and variaility trends in 3. In, we discuss our Hα oservations compared to the mean trends for large samples of M dwarfs.. OBSERVATIONS AND DATA REDUCTION The SLoWPoKES catalog, constructed from the Sloan Digital Sky Survey (SDSS; York et al. 000), comprises 13 wide ( 500 AU) common proper motion pairs with at least one lowmass (K5 or later) component (Dhital et al. 010). While these systems are classified as inaries ased on common proper motion and distance alone, their proaility of chance alignment is low. To e part of the SLoWPoKES catalog, each pair was required to have a false positive proaility (calculated for each pair using a Galactic model including oth stellar density and kinematics) of less than 5%. Dhital et al. (01) otained follow-up oservations of 111 of these pairs, finding that 87% showed good agreement in their radial velocities. The remaining 13% of those pairs were likely to e contaminated with low signal-to-noise ratio (S/N) oservations of spectroscopic inaries (providing a challenge for the accurate measurement of radial velocities). As these common proper motion pairs have a high proaility of eing physically associated, we assume that they are wide inaries. With a large numer of inaries that extend down to M spectral types, the SLoWPoKES catalog is an ideal source of coeval laoratories to conduct follow-up studies of M dwarfs. Because the SLoWPoKES inaries were identified from photometry and proper motions without spectroscopic information, the catalog does not include the magnetic activity as measured y Hα emission for the component stars. Therefore, we selected an initial sample of SLoWPoKES inaries for spectroscopic oservations. The components of the inaries were restricted to M0 and later dwarfs (r z 0:9) and to have similar r z colors (Δðr zþ 0:5), meaning the components are roughly within one-half spectral type. We also selected for well-resolved components that fit along the length of the slit (separations of 7 10 ) and a minimum rightness of r 18. This resulted in 17 candidate pairs, 3 of which had een determined to e inactive y Dhital et al. (01), giving an initial sample of 10 near-equal mass, M dwarf inaries. While this sample was chosen to e near-equal mass, there exists a small difference in their r z colors. Hereafter, we refer to the component with the luer r z color as the primary (A) and the one with the redder r z color as the secondary (B). Out of the sample of 10 candidate SLoWPoKES inaries, we otained optical (λ Å) spectra for 8 pairs (9 total stars) in our sample using the Dual Imaging Spectrograph (DIS) on the Astrophysical Research Consortium (ARC) 3.5-m telescope at Apache Point Oservatory (APO). No selection criteria were initially applied, ut righter and redder systems were preferentially oserved. The complete sample of oserved M dwarf inaries is listed in Tale 1. DIS uses a dichroic to split light into a red and a lue channel; for this study, we only used the red channel. These data were taken with a 10 long, 1.5 wide slit, dispersed with the R300 (:31 Å=pix) grating, resulting in an average resolution of R Spectra for the components of a inary were otained simultaneously y placing oth targets on the spectroscopic slit. Figure 1 shows spectra of SLW during time-resolved oservations; those spectra are representative of typical spectra otained during the oservations. For each inary, we initially otained three exposures of 1 10 min each to achieve a minimum total S/N of 0. Flats, iases, and HeNeAr comparison arcs were also taken at the eginning or end of every night. The data were ias sutracted, flat fielded, extracted, and wavelength calirated (onto an air scale) using standard IRAF 7 routines. To identify a sample of Hα-active M dwarf inaries for timeresolved oservations, we first determined the spectral types y comparing each spectrum y eye with templates using the Hammer software (West et al. 00; Covey et al. 007). In five cases, the spectral type of the A component was one sutype later than that of the B component, in contrast with their r z colors. We did not reassign types for these systems ecause of the uncertainties associated with spectral types; since our types are good to 0:5 sutype, these types are still consistent with a slightly 7 IRAF is distriuted y the National Optical Astronomy Oservatories, which are operated y the Association of Universities for Research in Astronomy, Inc., under cooperative agreement with the National Science Foundation.

3 HΑ EMISSION FROM TWIN M DWARF BINARIES 1083 TABLE 1 LIST OF ALL OBSERVED BINARIES Designation Comp R.A. Decl. Dist a (pc) P false (%) SpT EW Hα (Å) r z i z L Hα =L ol ( 10 ) Active? SLW A M5.17 ± ± ± y B M.5 ± 1.3. ± ± y SLW A M 0.1 ± ± ± 0.03 n B M 0.10 ± ± ± 0.03 n SLW A M ± ± ± 0.03 n B M ± ± ± 0.03 n SLW A M 0. ± ± ± 0.01 n B M 0.07 ± ± ± 0.01 n SLW 01+0 A M3 0.0 ± ± ± 0.01 n B M3 0.0 ± ± ± 0.01 n SLW A M3 0.1 ± ± ± 0.0 n B M 0.0 ± ± ± 0.03 n SLW A M 1.95 ± ± ± y B M 5.5 ± ± ± y SLW A M 0.3 ± ± ± 0.0 n B M 0.19 ± ± ± 0.0 n SLW A M 0.1 ± ± ± 0.01 n B M ± ± ± 0.01 n SLW A M 0. ± ± ± 0.03 n B M 0.08 ± ± ± 0.03 n SLW A M ± ± ± 0.03 n B M3 0.1 ± ± ± 0.03 n SLW A M ± ± ± 0.01 n B M ± ± ± 0.01 n SLW A M3 0.0 ± ± ± 0.03 n B M3 0.1 ± ± ± 0.03 n SLW A M. ± ± ± y B M.90 ±.07.0 ± ± y SLW A M5 0.3 ± ± ± 0.0 n B M ± 0.7. ± ± 0.0 n SLW A M 0.1 ± ± ± 0.0 n B M 0.39 ± ± ± 0.0 n SLW A M 0.38 ± ± ± 0.0 n B M 0.19 ± ± ± 0.0 n SLW A M5.8 ± ± ± y B M ± ± ± y SLW A M3.37 ± ± ± y B M3 5.9 ± ± ± y SLW A M3 0.5 ± ± ± w B M ± ± ± 0.03 n SLW A M3 0. ± ± ± 0.03 n B M 0.1 ± ± ± 0.03 n SLW A M3 0.5 ± ± ± 0.03 n B M3 0.0 ± ± ± 0.03 n SLW A M ± ± ± 0.03 n B M3 0.3 ± ± ± 0.03 n SLW A M 1.5 ± ± ± y B M 5.9 ± ± ± y SLW A M 0. ± ± ± 0.01 n B M 0.3 ± ± ± 0.01 n SLW A M5 0.7 ± ± ± 0.0 n

4 108 GUNNING ET AL. TABLE 1 (Continued) Designation Comp R.A. Decl. Dist a (pc) P false (%) SpT EW Hα (Å) r z i z L Hα =L ol ( 10 ) Active? B M ± ± ± 0.0 n SLW A M 1.3 ± 0..1 ± ± y B M 3.1 ± 0..8 ± ± y SLW 11+5 A M0 0.3 ± ± ± 0.0 n B M 0.15 ± ± ± 0.0 n SLW A M 0. ± ± ± 0.0 n B M ± ± ± 0.0 n SLW A M3 0.8 ± ± ± 0.03 n B M3 0. ± ± ± 0.03 n SLW A M 0.0 ± ± ± 0.03 n B M 0.33 ± ± ± 0.03 n SLW 1+53 A M 0.1 ± ± ± 0.03 n B M 0.87 ± ± ± w SLW A M 0.01 ± ± ± 0.0 n B M 0.1 ± ± ± 0.0 n SLW A M 0.10 ± ± ± 0.0 n B M5 0.0 ± ± ± 0.0 n SLW A M 0.03 ± ± ± 0.03 n B M3 0.1 ± ± ± 0.03 n SLW A M 0.0 ± ± ± 0.0 n B M3 0.3 ± ± ± 0.0 n SLW A M 0.1 ± ± ± 0.03 n B M 0.18 ± ± ± 0.03 n SLW A M ± ± ± 0.01 n B M ± ± ± 0.01 n SLW A M 0.18 ± ± ± 0.03 n B M 0.0 ± ± ± 0.03 n SLW A M 0.3 ± ± ± 0.01 n B M 0.9 ± ± ± 0.01 n SLW +318 A M 0. ± ± ± 0.0 n B M 0.7 ± ± ± 0.0 n SLW 53+7 A M ± ± ± W B M ± ± ± 0.03 n SLW A M 0.5 ± ± ± 0.0 n B M5 0.0 ± ± ± 0.0 n SLW A M5 0.0 ± ± ± 0.03 n B M 0.0 ± ± ± 0.03 n SLW A M 0.07 ± ± ± 0.0 n B M 0.08 ± ± ± 0.0 n SLW A M ± ± ± 0.03 n B M ± ± ± 0.03 n SLW A M5 0.3 ± ± ± 0.03 n B M ± ± ± 0.03 n SLW A M5 3.9 ± ± ± y B M ± ± ± y NOTE. Units of right ascension are hours, minutes, and seconds, and units of declination are degrees, arcminutes, and arcseconds. a Distances were calculated using the Bochanski et al. (010) M r as a function of r z. For inaries with time-resolved oservations, the mean and standard deviation of oservations, with flares included, are listed for Hα EW.

5 HΑ EMISSION FROM TWIN M DWARF BINARIES 1085 Normalized Flux + Constant 3 1 Normalized Flux Wavelength (Å) SLW B SLW A Wavelength (Å) FIG.1. Red optical spectra (normalized flux as a function of wavelength) for the A and B components of SLW during the time-resolved oservations on UT 011 Decemer 9. Both the A component (ottom six spectra) and the B component (top six spectra) are shown at six different times 1 hr apart, with time increasing from the ottom (purple) spectrum to the top (red) spectrum. The inset figure shows the Hα emission from the B component; the green spectrum was taken during a flare. See the electronic edition of the PASP for a color version of this figure. higher mass A component. We also used the Hammer to measure the Hα equivalent width (EW; hereafter EW Hα) for each component. The Hα line region is defined as Å, and the continuum regions are Å and Å. The Hammer applies a multipart criteria to determine the activity of each oject (descried in detail y West et al. [00]). When considering spectra with S=N > 3 in the continuum region (a condition met y all our spectra), those with EW Hα > 0:75 Å are considered active (y in Tale 1), those with EW Hα < 0:75 Å and an EW greater than 3 times its uncertainty are considered weakly active (w in Tale 1), and those with EW Hα elow 3 times its uncertainty are considered inactive (n in Tale 1). While the resolution of these spectra is somewhat lower than those taken y SDSS (R 1000 compared to R 000), our Hα EW measurements should e comparale to those performed on SDSS data. Nineteen of the 9 stars exhiited Hα in emission. Of the 8 inaries oserved, eight exhiited Hα emission in oth components, and three exhiited Hα emission in one component. In addition to measuring EW Hα, we also calculated the spectral type-independent measure of activity strength (L Hα =L ol ) for the active dwarfs. The conversion from EW Hα to L Hα =L ol relies on the χ factor, calculated y West & Hawley (008) as a function of i z color or spectral type. We adopted χ as a function of spectral type to calculate L Hα =L ol for each oject. Spectral types, EW Hα, and L Hα =L ol are given in Tale 1. We selected the five rightest of the eight inaries with two active components for time-resolved spectroscopy. These five systems were each oserved for 1 5 hr per night for a total of 7 hr over 11 nights. We used the same setup for the time-resolved oservations as for the single epoch oservations, with typical exposure times of 1 5 minutes for each system. The UT dates and hours oserved are given in Tale 1. Four of the systems were oserved on three or four different nights while SLW was only oserved on one night. We measured EW Hα using the Hammer for each spectrum; the Hα variations are discussed for each pair in the next section. 3. CHARACTERIZING Hα LIGHTCURVES Our initial goal was to determine if oserved differences in EW Hα etween the two stars in each inary were due to variaility, e.g., two ojects with the same mean EW Hα would appear different in single epoch oservations if their instantaneous measured values differed from the mean. The light curves from the time-resolved oservations are shown in Figure. Our initial examination of the light curves revealed the presence of flares, so we identified and removed the flares efore examining the quiescent light curves in more detail Removing Flares Both flares and quiescent activity are caused y the interaction of strong surface magnetic fields with the stellar atmosphere, ut the details of that interaction are likely to e different. Flares are thought to e triggered y magnetic reconnection events resulting in dramatic heating over short timescales (e.g., Cram & Mullan 1979), while quiescent activity is characterized y emission over longer timescales from lower temperature material (e.g., Roinson et al. 1990). Our primary focus is the quiescent variaility, ut Hα emission traces oth quiescent emission and flares. We calculated the mean ( EW Hα ) and standard deviation (σ EW ) of the Hα emission oth from the entire light curves and from the light curves with flares removed. While flares have een traditionally identified y eye (e.g., Pettersen et al. 198), to clearly separate flares from smaller-scale variations we chose to remove flares using the quantitative method developed y Hilton (011), which we summarize here. We first determined a quiescent mean and standard deviation EW Hα for each inary component on each night of data. For stars that appeared to exhiit a flaring event, we used the quiescent period efore and/or after the flare to otain our EW Hα and σ EW. We then defined a flare as an event with at least one point aove 3σ and at least five points aove σ of the quiescent EW Hα. Figure 3 shows a suset of the EW Hα measurements of SLW 0858 þ 093AB (from UT 011 Decemer 9) with the quiescent mean as well as the 1σ and σ deviations given as dashed lines. The A component shows variations, ut none are large enough to meet the flare criteria. The B component also varies throughout the night, with one of the variations eing large enough to e considered a flare.

6 108 GUNNING ET AL Aug Oct 15 SLW 019+A SLW 019+B 011 Oct Fe Oct SLW A SLW B 011 Nov 8 01 Mar EWHα EWHα Mar 09 SLW A SLW B 011 Dec 9 01 Mar May Fe 1 01 Fe 1 SLW 110+0A SLW 110+0B 01 Mar 1 EWHα 10 8 EWHα EWHα SLW A SLW B 011 Aug FIG.. EW Hα as a function of time for each of the inaries. Upper left: SLW Upper right: SLW Middle left: SLW Middle right:slw Bottom: SLW Each panel is laeled with the UT dates when the inary was oserved. The B component data are shown as open circles and the A component data are filled circles. The red data indicate the portion of the curve that has een classified as a flare event, as descried in 3.1. Formal uncertainties are shown on each point; many are approximately the size of the point. These stars have separations far greater than those of interacting inaries, so any variations that are similar etween the two stars are either coincidental or due to oservational effects (e.g., changing airmass). See the electronic edition of the PASP for a color version of this figure. Using these criteria, we initially identified six flares. We then confirmed each flare y eye, accepting five flares and rejecting one spurious flare (on SLW B; see discussion elow), resulting in a total of five flares out of 7 hr of monitoring of the five inary systems (comparale to 9 hr on single M dwarfs). All five of the flares occurred on the B components of the inaries, including two on SLW B. The numer of flares on each star on each night is noted in Tale and the flares are indicated in the lightcurves shown in Figure.

7 EW Hα (Å) SLW A SLW B 3σ σ Mean HΑ EMISSION FROM TWIN M DWARF BINARIES 1087 TABLE 3 BINARIES WITH TIME-DOMAIN SPECTRA Including flares Designation SLW+ Comp EW Hα (Å) σ EW Hα EW Hα Excluding flares EW Hα (Å) σ EW Hα EW Hα A.17 ± ± B.5 ± ± A 1.95 ± ± B 5.5 ± ± A. ± ± B.90 ± ± A 1.5 ± ± B 5.9 ± ± A 3.9 ± ± B.01 ± ± FIG. 3. EW Hα as a function of time for SLW (top) and SLW B (ottom) on UT 011 Decemer 9. The data (lack filled circles with error ars) are shown along with the mean and multiples of the standard deviation (dashed lines). According to the flare criteria descried in 3.1, the A component shows only quiescent variations while the B component flares at t ¼ 10:3 hr. 3.. EW Hα Light Curves The EW Hα light curves shown in Figure are descried in detail elow. The EW Hα and σ EW were calculated oth with and without flares and are given in Tale 3. Here we discuss the properties with flares removed SLW A total of 7.3 hr of data were otained for this system over the course of three nights. The A component maintained a relatively constant EW Hα of EW Hα ¼ :17 Å with TABLE LOG FOR TIME-DOMAIN OBSERVATIONS # Flares UT Designation Total t Os (hrs) A B 011 Feruary 18 SLW AB March 9 SLW AB May SLW 110+0AB August 7 SLW AB SLW AB Octoer 15 SLW AB Octoer SLW AB SLW AB Novemer 8 SLW AB Decemer 9 SLW AB Feruary 1 SLW 110+0AB Feruary 1 SLW 110+0AB March 1 SLW 110+0AB March SLW AB SLW AB σ EW ¼ 0:30 Å. Hα emission from the B component was more variale, showing a Å variation etween the two longer nights of oservations with EW Hα ¼ :5 Å and σ EW ¼ 1:3 Å. It is unclear whether the ase value surrounding the flare on UT 011 Octoer 15 is true quiescence or part of a longer trend of elevated activity, ut it is clear that the B component has stronger and more variale Hα emission than the A component SLW We oserved this system for 7.7 total hr on four different nights. The emission from A component was relatively constant with an EW Hα ¼ 1:95 Å and σ EW ¼ 0:57 Å. The emission from the B component was oth stronger ( EW Hα ¼ 5:5 Å) and more variale (σ EW ¼ 0:80 Å) than the A component. The emission from the B component included a flare, while the emission from the A component did not. At t ¼ 9:8 hr, the A component showed an increase that was initially marked as a flare, ut it was rejected when reviewed y eye ecause it lacked a decay phase after the initial rise SLW This system was oserved for 13. hr on three separate nights. The Hα emission from the A component was relatively strong ( EW Hα ¼ : Å) ut not variale (σ EW ¼ 0:39 Å). The Hα emission from the B component was very strong and variale ( EW ¼ :90 Å; σ EW ¼ :07 Å) even with the two flares excluded. On UT 011 March 9, the Hα emission showed a large flare that peaked at EW Hα ¼ 15 Å and on UT 011 Decemer 9, a small flare peaked at an EW of Hα ¼ Å SLW This inary was oserved for a total of 17.8 hr during four nights. The Hα emission from the A component was the weakest of all the oserved M dwarfs ( EW Hα ¼ 1:5 Å) ut more variale than other A components (σ EW ¼ 0:7 Å). The mean

8 1088 GUNNING ET AL. Hα emission from the B component was stronger than that of other B components ( EW Hα ¼ 5: Å) with relatively low variaility (σ EW ¼ 0:5 Å; similar to SLW B). On the last night of oservations for this pair, we oserved a small flare on the B component which peaked at EW Hα 8 Å SLW This inary pair was oth the latest-type dm in our sample (M5/M5) and has the least oservations (one hour on a single night). This inary had the smallest difference in strength and variaility etween the two components, with values of EW Hα Å and σ EW 0:3 Å for oth A and B. During the short duration of out oservations, SLW was less variale than the other oserved systems Variaility The production of Hα emission y an active chromosphere is a dynamic process on timescales of minutes to decades. Several groups have used different statistics to characterize that variaility; the ratio of maximum to minimum EW Hα (Kruse et al. 010; Lee et al. 010), the standard deviation of EW Hα (Gizis et al. 00), and normalized standard deviation of EW Hα (Bell et al. 01). We choose characterize variaility with the standard deviation of EW Hα (σ EW Hα ) and normalized variaility with σ EW Hα divided y EW Hα. Tale 3 includes these quantities for each of the twins with time-resolved data, and they are shown with respect to mean EW Hα and r z color in Figure. The σ EW Hα and σ EW Hα = EW Hα are slightly higher than the mean values reported y Bell et al. (01) for M dwarfs with similar spectral types and emission levels, which could e due to the differences in cadence and timescale of the oservations. The unnormalized variaility is on average higher for the B component than the A component. This is, perhaps, simply due to the B components having stronger activity than their twins. More Hα emission allows a larger dynamic range for variation. Increasing variaility with larger EW Hα is not visile when the normalized variaility is instead examined. The four pairs with strong variaility are evenly divided etween the A and.5.0 A Component B Component.5.0 Flare No Flare σ EW Hα (Å) σ EW Hα (Å) <EW Hα> (Å) r z σ EW Hα/<EW Hα> (Å) σ EW Hα/<EW Hα> (Å) <EW Hα> (Å) r z FIG.. Variaility of the twins with time-resolved oservations. The top two panels show the variaility of EW Hα (σ EW Hα ) as a function of the mean EW Hα ( EW Hα ; top left panel) and r z color (top right panel). The ottom two panels show the normalized variaility of EW Hα (σ EW Hα = EW Hα ) as a function of the mean EW Hα ( EW Hα ; ottom left panel) and r z color (ottom right panel). Each A (circles) and B (squares) component is connected, and values computed without flares (solid lack symols) are distinguished from values computed with flares (open red symols). See the electronic edition of the PASP for a color version of this figure.

9 HΑ EMISSION FROM TWIN M DWARF BINARIES 1089 B components showing stronger variaility. If variaility was a strong function of age, we would expect the coeval companions to have similar variaility, ut our sample reveals no relationship etween age, activity level, and variaility. If there are correlations etween these quantities, larger samples of M dwarfs or longer timescale Hα monitoring may e needed to detect them.. M DWARFS ACTIVITY TRENDS WITH MASS AND AGE Comparing the Hα emission of the A and B components to overall trends can provide unique constraints on the relationship etween M dwarf mass, age, and activity. Figure 5 shows the activity strength (L Hα =L ol ) of the active M dwarfs from West et al. (011) as a function of r z color (often a proxy for spectral type; Bochanski et al. 011). As expected, the activity strength declines slowly with decreasing mass (redder r z color) for these mid-m dwarfs (Hawley et al. 199; Gizis et al. 00; West et al. 008, 011), though with relatively large scatter: etween 1:5 <r z<3:0, there is at least an order of magnitude range of activity strengths (L Hα =L ol 0:0001 to 0.001). We also show the activity strength of the 11 inaries with at least one active component (including upper limits for the inactive component in those pairs). Of the 11 inaries, four have a more active A component, while the rest have more active B components. The inaries pairs do not, on average, follow the ensemle trend of decreasing activity strength with redder r z color. To understand the significance of the differences etween the A and B components compared to the scatter in emission strength for the ensemle of measurements, in Figure we show the percent difference etween the A and B components L Hα =L ol as a function of the A component s L Hα =L ol compared to the normalized standard deviation of L Hα =L ol values from Bell et al. (01). If magnetic activity monotonically decayed for oth A and B twin stars together, we would naively expect all active pairs to have a difference close to zero on the vertical axis. However, this zero difference only falls within the uncertainties of one system. This is true even of the five pairs with dedicated monitoring; the variaility over timescales of hours to months does not ring the A and B emission strengths closer together. While no clear trend is seen as a function of the A component s L Hα =L ol, the difference in activity spans from 10 to 80%, with the majority of systems having stronger Hα emission in the slightly redder B component. Additionally, four of the pairs in our sample show differences in their activity significantly greater than the typical scatter from Bell et al. (01), all in which the B component has stronger activity. These are particularly interesting systems that would enefit from additional oservations. It is possile that despite their coevality, these twin M dwarfs have different rotation rates and magnetic field strengths. Another possiility is the existence of Solar-like activity cycles that could e revealed though additional Hα monitoring. 100 M M3 M M5 L Hα /L ol A component B component Percent Difference r-z FIG. 5. The ratio of the Hα luminosity to the olometric luminosity as a function of r z color for the active inaries in our sample. The distriution of active SDSS M dwarfs is also shown for comparison (contours; Bell et al. 01). A (triangle) and B (square) components of each pair are connected (solid and dashed lines). For stars with single epoch data (lue), the measured L Hα =L ol is shown while for pairs with time-resolved data (red), oth the median and total spread in L Hα =L ol are shown. For the three pairs consisting of one active and one inactive component, the upper limit of Hα emission (the measured EW from the highest individual spectrum, which was not high enough to meet our criteria for active) is shown as a down arrow. See the electronic edition of the PASP for a color version of this figure L Hα /L ol (A Component) FIG.. The percent difference in activity strength etween A and B components for each pair as a function of the A component activity strength. Here, systems with a positive percent difference have stronger activity in the A component. Binaries with time-resolved oservations (lack circles) are distinguished from those with only a single set of oservations (lue triangles). Systems with only one active component are represented with arrows denoting upper or lower limits; for SLW 1+53, the limit is on oth the A component Hα and the difference etween the A and B component. For comparison, the normalized standard deviation of the ensemle activity level for the M0-M5 dwarfs from Bell et al. (01) is shown (gray shaded region). Systems falling outside the gray shaded region have significantly different levels of activity compared to the typical range of the ensemle. See the electronic edition of the PASP for a color version of this figure.

10 1090 GUNNING ET AL. While there is not evidence for a strong trend etween activity and age, our data are consistent with the activity lifetime model of M dwarf Hα emission. Our full sample included 37 inactive pairs, three pairs with one active component, and eight pairs with two active components. The 37 systems with two inactive components may e older than the active lifetimes for their spectral types. Based on the West et al. (008) age-activity relations, this places lower limits on the stellar ages of 1 Gyr for M0 M dwarfs, and 7 Gyr for M3 M5 dwarfs, which are typical of disk stars. In the three inary systems from our sample for which only one stellar component showed Hα activity, the active components (two A components and one B component) had relatively weak Hα emission (EW < 0:5 Å). Given the moderate S/N and low spectral resolution of our oservations, very weak emission from the inactive components may have een elow our measurement sensitivity. These systems may e in fact e undergoing the transition from active to inactive states. This would in turn imply that the inary s age is approximately at the expected activity lifetime (e.g., Gyr for M). The relative scarcity of these systems in our dataset indicates that this transition must occur relatively quickly. Variaility may instead e a cause of the scatter in activity strength with respect to mass and age. If variaility is the main cause for the scatter, we would expect the activity of all mid-m dwarfs to e aout equal when oserved over long timescales. For the inaries with time-resolved data, we plot the median value and the full range of oserved activity strengths in Figure. These ranges show the possile range of values that would e oserved for each M dwarf in a single spectrum. This variaility ranges from a factor of to an order of magnitude and is comparale to the full range of scatter in the singe-epoch data, indicating that variaility could e the main cause for the scatter. Variaility does not, however, explain the differences etween the A and B components ecause they persist on timescales of weeks to years. To understand the different Hα strengths of coeval inaries, additional oservations are needed. 5. SUMMARY We presented optical spectra for 9 early- to mid-m dwarfs found in 8 coeval twin inary systems. Of those inaries, eight had Hα emission from oth components and three showed Hα emission from one component. The active wide inary components did not show the similar levels of Hα emission that would initially e expected from M dwarfs of the same mass and age. Instead, the A and B components exhiited different levels of emission, with four of the 11 having differences larger than the standard deviation of the scatter in L Hα =L ol for the West et al. (011) M dwarfs. Despite the mean trend of decreasing activity strength with decreasing mass, we found that seven of the 11 inaries had stronger activity on the less massive (B) component. We also presented 7 hr spectroscopic monitoring for five of the wide inaries. We examined their Hα light curves in detail, identifying five flares that all occurred on the B components of the inaries. The majority of these M dwarfs exhiited significant variaility; the range of activity strength was comparale to the scatter in L Hα =L ol as a function of r z color. Over the 1 year timescale of the time-resolved oservations, the oserved inaries continued exhiit stronger activity from their B components, ut variations on longer timescales (e.g., decades) may further modify their relative activity strengths. The variaility from our time-resolved data showed no trends with either r z color or Hα EW; we find that variaility is independent of oth stellar mass and activity strength. While M dwarf ensemle trends indicate strong relationships etween mass, age, and the presence and strength of activity, our oservations reveal a complicated relationship etween mass, age, and activity for individual stars. The differences in activity in coeval twins could indicate long timescale Solar-type variations in the magnetic field strength or could e due to intrinsic differences in rotation rate and magnetic field generation. Timeresolved oservations over longer timescales, in addition to more detailed studies of coeval twins, will e instrumental in a complete understanding of M dwarf activity. J. R. A. D. acknowledges funding from NASA ADP grant NNX09AC77G. S. D. acknowledges funding from NSF grant AST A. A. W. acknowledges funding from NSF grants AST and AST and also the support of the Research Corporation for Science Advancement s Cottrell Scholarship. Our work relies on data from the Sloan Digital Sky Survey. Funding for the SDSS and SDSS-II has een provided y the Alfred P. Sloan Foundation, the Participating Institutions, the National Science Foundation, the U.S. Department of Energy, the National Aeronautics and Space Administration, the Japanese Monukagakusho, the Max Planck Society, and the Higher Education Funding Council for England. The SDSS We Site is The SDSS is managed y the Astrophysical Research Consortium for the Participating Institutions. The Participating Institutions are the American Museum of Natural History, Astrophysical Institute Potsdam, University of Basel, University of Camridge, Case Western Reserve University, University of Chicago, Drexel University, Fermila, the Institute for Advanced Study, the Japan Participation Group, Johns Hopkins University, the Joint Institute for Nuclear Astrophysics, the Kavli Institute for Particle Astrophysics and Cosmology, the Korean Scientist Group, the Chinese Academy of Sciences (LAMOST), Los Alamos National Laoratory, the Max-Planck- Institute for Astronomy (MPIA), the Max-Planck-Institute for Astrophysics (MPA), New Mexico State University, Ohio State University, University of Pittsurgh, University of Portsmouth, Princeton University, the United States Naval Oservatory, and the University of Washington.

11 REFERENCES HΑ EMISSION FROM TWIN M DWARF BINARIES 1091 Bell, K. J., et al. 01, PASP, 1, 1 Bochanski, J. J., et al. 010, AJ, 139, 79 Bochanski, J. J., Hawley, S. L., & West, A. A. 011, AJ, 11, 98 Bopp, B. W., & Schmitz, M. 1978, PASP, 90, 531 Browning, M. K., Miesch, M. S., Brun, A. S., & Toomre, J. 00, ApJ, 8, L 157 Charier, G., & Baraffe, I. 1997, A&A, 37, 1039 Covey, K. R., et al. 007, AJ, 13, 398 Cram, L. E., & Mullan, D. J. 1979, ApJ, 3, 579 Dhital, S., West, A. A., Stassun, K. G., & Bochanski, J. J. 010, AJ, 139, 5 Dhital, S., West, A. A., Stassun, K. G., Bochanski, J. J., Massey, A. P., & Bastien, F. A. 01, AJ, 13, 7 Doler, W., Stix, M., & Brandenurg, A. 00, ApJ, 38, 33 Gizis, J. E., Reid, I. N., & Hawley, S. L. 00, AJ, 13, 335 Gomes da Silva, J., et al. 011, A&A, 53, A 30 Hawley, S. L., Gizis, J. E., & Reid, I. N. 199, AJ, 11, 799 Hilton, E. J. 011, University of Washington Ph.D. Thesis Irwin, J., et al. 011, ApJ, 77, 5 Johns-Krull, C. M., & Valenti, J. A. 199, ApJ, 59, L 95 Kruse, E. A., et al. 010, ApJ, 7, 135 Lee, K.-G., Berger, E., & Knapp, G. R. 010, ApJ, 708, 18 Meiom, S., Mathieu, R. D., & Stassun, K. G. 007, ApJ, 5, L 155 Morgan, D. P., et al. 01, AJ, 1, 93 Pettersen, B. R., Coleman, L. A., & Evans, D. S. 198, ApJS, 5, 375 Roinson, R. D., Cram, L. E., & Giampapa, M. S. 1990, ApJS, 7, 891 Valenti, J. A., & Johns-Krull, C. 001, in ASP Conf. Ser. 8, Magnetic Fields Across the Hertzsprung-Russell Diagram, ed. G. Mathys, S. K. Solanki, & D. T. Wickramasinghe (San Francisco: ASP), 179 West, A. A., et al. 008, AJ, 135, , AJ, 18,. 011, AJ, 11, 97 West, A. A., & Hawley, S. L. 008, PASP, 10, 111 York, D.G., et al. 000, AJ, 10, 1579