Using Photometry to Probe the Circumstellar Environment of delta Scorpii
C. E. Jones, Paul Wiegert, C. Tycner, G. W. Henry, R. P. Cyr, R. J. Halonen, M. W. Muterspaugh
aa r X i v : . [ a s t r o - ph . S R ] J un Draft version September 14, 2018
Preprint typeset using L A TEX style emulateapj v. 5/2/11
USING PHOTOMETRY TO PROBE THE CIRCUMSTELLAR ENVIRONMENT OF δ SCORPII
C. E. Jones , Paul Wiegert , C. Tycner , G. W. Henry , R. P. Cyr , R. J. Halonen , M. W. Muterspaugh Draft version September 14, 2018
ABSTRACTWe acquired Johnson BV photometry of the binary Be disk system δ Scorpii during its 2009, 2010,2011, and 2012 observing seasons and used it to probe the innermost regions of the disk. We foundthat several disk building events have occurred during this time, resulting in an overall brighteningin the V -band and reddening of the system. In addition to these long-term trends, we found cyclicalvariability in each observing season on timescales between 60 and 100 days. We were able to reproducethe changes in the magnitude and colour of δ Sco using our theoretical models and found that variablemass-loss rates in the range 2 . − . × − M ⊙ / yr over ∼
35 days can reproduce the observed increasein brightness.
Subject headings: binaries—circumstellar matter—stars: Be, emission-line, mass-loss, individual δ Sco INTRODUCTION
The classical Be (B-emission) stars are well known fortheir characteristic spectral emission lines in the Balmerseries, the defining property of the group, and infraredexcess due to radiative processes occurring within thedisk-like distribution of circumstellar material. Be starsare also characterized by rapid rotation, of typically sev-eral hundred km s − , and this property certainly playsa role in the release of material from the stellar surfaceto form a disk. Their rotation rates are difficult to de-termine because of the effects of gravity darkening and,as a result, the precise values remain a contentious is-sue (Townsend et al. 2004; Cranmer 2005). Despite theseunknowns, this group of stars offers a valuable test bedto study disk physics and the effects of rapid rotation onstellar evolution.In addition to rapid rotation, emission lines, and in-frared excess, there are other commonly observed fea-tures among this group of stars, such as polarized contin-uous light due to Thomson scattering of stellar radiationand variability on a range of time-scales from minutesto decades (Porter & Rivinius 2003). The longer timescales are associated with significant disk building anddisk dissipation events. Short term variations on typicalscales of 0.5 to 2 days are often associated with stellarpulsation (Rivinius et al. 2003).There have been a variety of models proposed to ex-plain the formation of these disks and associated char-acteristics beginning with Struve (1931) who suggestedthat Be star disks are formed exclusively by the rapidrotation of the central star with the variety of spectralline shapes due to viewing angle. However, this viewmay be too simplistic, since it is generally believed thatBe stars rotate below critical so there must be another Department of Physics and Astronomy, Western University,London, ON Canada N6A 3K7 Department of Physics, Central Michigan University, Mt.Pleasant, MI 48859 USA Center of Excellence in Information Systems, TennesseeState University, 3500 John A. Merritt Blvd., Box No 9501,Nashville, TN 37209, USA Department of Mathematical Sciences, College of Engineer-ing, Tennessee State University, Boswell Science Hall, Nashville,TN 37209, USA mechanism(s) to lift material off the stellar surface.Though there have been many different models pro-posed since Struve’s time to explain these systems, theviscous disk model has been most promising (Carciofi2011). Basically, this model operates similarly to thestandard α -disk theory for accretion disks of inward flow-ing material except, in this case, the gas is outward flow-ing. The viscous model was first applied to Be stars byLee et al. (1991), later investigated by Okazaki (2001),and more recently is being used to explain Be disk sys-tems in exceptional detail (Carciofi et al. 2006, 2012).This model naturally predicts near-Keplerian disk rota-tion, a property that is consistently implied through ob-servations (Meilland et al. 2007; Oudmaijer et al. 2011;Wheelwright et al. 2012). Viscous disks may also be-come unstable to density perturbations allowing the for-mation of under- and over-dense volumes of gas thatcould give rise to the characteristic variability of doublypeaked emission lines that are often observed. Viscousdisk models have also successfully predicted the observedIR excess (see Carciofi et al. (2009)).The actual mechanism for launching material from therapidly rotating star remains unknown, and often thevalue of the mass loss rate is a free parameter in themodeling process (see e.g., Carciofi et al. (2006)). V -band photometry has been shown to probe the inner-most regions of Be disks (Carciofi 2011) and so will showthe quickest reaction to photospheric activity (Carciofi2012). Therefore, V -band photometry combined withmodeling may be used to place bounds on the mass-lossrates and contribute to a better understanding of thelaunching mechanism(s).Interestingly, δ Sco (HD143275, HR5953) was orig-inally designated as an uvbyβ standard star ofspectral type B0.3IV (Perry et al. 1987) but laterCote & van Kerkwijk (1993) observed emission in thewings of H α providing evidence that δ Sco had become aBe star. In that same year, Bedding (1993) confirmedspectroscopic binarity and reported an elliptical orbitwith a 10.5 year orbit. A significant brightening oc-curred in 2000 (Otero et al. 2001) and since that time,the highly variable early-type Be star δ Sco has beena focus of much attention. It is interesting to speculate Jones et al.that the periastron passage in 2000 contributed to the ob-served brightening. The secondary companion has a veryeccentric orbit with a period of about 11 years (for recentcalculations see Tango et al. (2009); Tycner et al. (2011);Meilland et al. (2011)) and the periastron passage of itscompanion in 2011 offered an opportunity to study thissystem. We collected V - and B -band photometry for the2009, 2010, 2011 and 2012 observing seasons with the T30.4m automatic photoelectric telescope (APT) operatedby Tennessee State University (TSU) and located at Fair-born Observatory in southern Arizona.In this work we use our V and B photometry to probethe innermost regions of the disk of δ Sco in order to de-termine the mass loss rate and investigate the variabil-ity of these outbursts for this system. Our observationsand results are presented in Section 2 and Section 3, re-spectively, and a discussion and summary is provided inSection 4. OBSERVATIONS
The T3 0.4 m automatic photoelectric telescope (APT)at Fairborn Observatory in southern Arizona acquiredseveral hundred observations of δ Sco during the 2009,2010, 2011, and 2012 observing seasons. T3 is oneof eight automatic telescopes operated by TennesseeState University at Fairborn for automated photom-etry, spectroscopy, and imaging (Henry 1995, 1999;Eaton, Henry, & Fekel 2003; Eaton & Williamson 2007).T3 is equipped with a precision photometer that em-ploys an EMI 9924B photomultiplier tube (PMT) for thesuccessive measurements of photon count rates throughJohnson B and V filters. To maximize the stability ofthe photometer and the precision of the data, the PMT,voltage divider, pre-amplifier electronics, and photomet-ric filters are all mounted within the temperature- andhumidity-controlled body of the photometer. The preci-sion of a single observation on a good night is usually inthe range ∼ . − .
005 mag (e.g., Johnson et al. 2011,Table 20), depending primarily on the brightness of thetarget and the airmass of the observation. Year-to-yearbrightness means are stable to 0.0001 - 0.0002 mag overdecadal time scales (Henry 1999).We programmed the APT to make one or two observa-tions of δ Sco each clear night in the following sequence,termed a group observation: K,S,C,V,C,V,C,V,C,S,K,where K is the check star (HD 144217, V = 2 . B − V = − .
06, B0.5 V), C is the comparison star(HD 144470, V = 3 . B − V = − .
05, B1 V), V isthe program star δ Sco ( V = 2 . B − V = − . V − C and two unbracketed K − C differ-ential magnitudes, which are averaged together to creategroup means for both B and V bands. Group-mean dif-ferential magnitudes with internal standard deviationsgreater than 0.01 mag were discarded to eliminate ob-servations taken under non-photometric conditions. Thesurviving group means were corrected for differential ex-tinction with nightly extinction coefficients, transformedto the Johnson system with yearly-mean transformation Table 1
Photometric Observations of δ Sco .Reduced Julian Date Var B Var V Chk B Chk V (mag) (mag) (mag) (mag)54,890.0405 − − − − − − − − − − − − − − − − − − − − − − − − − − − − Note . — Table 1 is presented in its entirety in the electronicedition of the Astrophysical Journal. A portion is shown here forguidance regarding the form and content. coefficients, and treated as single observations thereafter.T3 acquired a total of 511 group observations over thefour observing seasons of our campaign. The observa-tions that survived the cloud-filtering process are plottedin the four panels of Figure 1 and listed in Table 1. Allfour panels of Figure 1 are plotted using the same ver-tical and horizontal scales to allow direct comparison ofthe brightness variations in the V and B filters, in the B − V color index, and any brightness changes in the K − C differential magnitudes during the four seasonsof our observational campaign. The K − C observationsscatter around their grand mean with a standard devi-ation of 0.0089 mag. This is somewhat larger than thetypical precision of ∼ .
004 mag referenced above, pri-marily because δ Sco lies at a declination of − ◦ andso is observed through a larger than usual air mass. The K − C observations in the bottom panel of Figure 1 arequite flat compared to the V − C observations plottedin the top three panels, indicating that the variability inthe V − C observations are intrinsic to δ Sco. We dosuspect slight variability in the comparison star at thevery end of our time series (see Section 3 below). Thearrow in the top panel marks the time of the latest peri-astron passage on UT 2011 July 6 ± V magnitudes and ∆( B − V ) color indices to facilitatetheir comparison. The vertical axis scale of the individualpanels in those two figures are no longer the same asthey were in Figure 1 but are chosen individually to bestdisplay the individual light curves. RESULTS
As previously mentioned, δ Sco is highly variable andthis behaviour can clearly be seen in Figure 1. The arrowin the top panel of this Figure marks the periastron pas-sage of the secondary and reveals no significant change inthe photometry surrounding this time. Cyclical behaviorwith timescales of approximately 60–100 days is visible inall four seasons with obvious differences in amplitude andmean brightness of the system. This variability of simi-lar frequency was observed much earlier by Gandet et al.(2002) so these variations have persisted for at least adecade, even at times when the disk density was muchdifferent. As mentioned previously, the V -band photom-etry is sensitive to the emitting volume within the firstfew stellar radii (see figure 1 of Carciofi (2011)). There-fore, this variability could be due to changes in the den- Scorpii’s Circumstellar Disk 3
Table 2
YEARLY PHOTOMETRIC PERIODS FOR δ SCOObserving Photometric N obs Period Full AmplitudeSeason Passband (days) (mag)(1) (2) (3) (4) (5)2009 B
91 64 . ± . . ± . V
94 64 . ± . . ± . B
64 72 . ± . a . ± . V
61 73 . ± . a . ± . B
127 94 . ± . a . ± . V
126 94 . ± . a . ± . B
152 83 . ± . . ± . V
145 101 . ± . b . ± . a Due to the slope in these light curves, we first removeda linear trend before applying the period analysis. b This period determination is probably much more un-certain than the formal error indicates. See the text fordetails.sity of the disk within the first few stellar radii or as aresult of shielding of the inner disk by warping or flaringof the outer disk.The top two panels of Figure 2 show that the bright-ness and color index in 2009 were anti-correlated, i.e.,the star became redder as it got brighter. These trendsare relatively straight forward to explain. As the systembrightens in the V -band with the injection of additionalgas into the disk, the system becomes correspondinglyredder due to an increase in radiative processes, espe-cially free-free emission that increases with wavelength.The bottom two panels demonstrate the same relation in2010. However, the top two panels of Figure 3 show a re-versal in the sense of the correlation; the star gets bluer asit gets brighter. Finally, the brightness and color changesfor 2012, plotted in the bottom two panels of Figure 3,appear to be significantly out of phase.We determined individual photometric periods for allfour observing seasons in both the B and V passbands.The complete results are given in Table 2. Figure 4 showsone example of our period determination, for the B dataof 2010. The frequency spectrum is computed by fittingleast-squares sinusoids to the data over a range of trialfrequencies and finding the frequency that best reducesthe total variance of the data set. Because of the slopein the observations for 2010 and 2011, we first removeda linear trend from the data before computing the fre-quency spectrum and the corresponding phase curves.Table 2 shows that both the photometric period andamplitude of δ Sco varied dramatically from year to year,from 64 to 94 days and 0.03 to 0.18 mag, respectively.One result, namely the 2012 V period determination,may be suspect. The 2012 V light curve has a much lowerS/N ratio than most of the 8 light curves. In addition, thecomp star exhibits possible low-amplitude variability forthe first time near the end of the 2012 observing season.This combination renders the period determination forthe 2012 V light curve unreliable.Figure 5 shows the ∆ V magnitude versus ∆( B − V ) forall four observing seasons with each season 2009, 2010,2011, 2012 indicated by the four shades of gray from lightto dark, respectively. The general trend of the data fromthe bottom left to upper right is due to the increase inV magnitude and a reddening of colour during the four years of observations as the disk builds. Note that the∆( B − V ) colour index becomes redder (more positive)as expected with the building of the disk. Overall the V magnitude begins to saturate and then most variationsare changes mainly in colour. This is particularly evidentin the observations for 2011 and 2012 since the changes in V magnitude are not as significant compared with 2009and 2010. Also note that the 2012 season is not quite asbright as the 2011 season. It will be interesting to see ifthis trend continues in future years.It is also interesting to replot Figure 5 as a function oftime for each observing season. We start with Figures 6and 7 showing ∆ V versus ∆( B − V ) for the 2009 seasonas a function of time. Although δ Sco is not as bright inthe V -band in 2009, this season shows two well-definedincreases in the V - and B -band, both of which returnto near their original values at the start of the season(see Figure 1). Figure 6 covers the period starting fromthe Reduced Julian date 54861 for 66 days, and Figure 7covers the remaining season of 76 days beginning at Re-duced Julian date 54928. The division of the 2009 observ-ing season into these two parts was rather arbitrary; wetried to divide the data where the loop seemed to repeat.Each figure shows a distinct loop during the specifiedtime-frame due to an increase in brightness and redden-ing that is followed by a subsequent decline in brightnessand by a slow return to bluer color-index. Notice in Fig-ures 6 and 7 that the maximum and minimum boundsof ∆ V and ∆( B − V ) are remarkably similar. This isnot surprising since Figure 1 shows that the peak bright-ness for both ∆ V and ∆ B are almost identical in 2009.Figure 8 shows that, in 2010, we also have two loops.However, the system remains bright in the V -band afterthe first loop, so the second loop starts at brighter valuesof the V -band and continues to increase during the for-mation of the loop. This can also be seen in Figure 1 forthe 2010 season. At the end of the 2010 observing season,both the V - and B -band magnitudes remain well abovetheir values at the beginning of this season. de Wit et al.(2006) studied the light and colour variability of Be starsin the Small Magellanic Cloud and found similar loopsto those displayed in Figure 6 to Figure 8 for 40% ofthe stars having a photometric variability of greater than0.2 m . Among their group of stars that exhibited this be-haviour, 90% of these loops were traced out in a clockwisedirection analogous to the behaviour we find for δ Sco.de Wit et al. (2006) attributed clockwise loops to slowlyoutward flowing material as a disk builds followed by anemptying of the disk from the inside-out as mass lossfrom the stellar surface is terminated. They suggest thatanticlockwise loops are indicative of accretion in youngsystems such as the Herbig Be stars.In 2011 and 2012, the loops are not as distinctive sincethe changes in ∆ V and ∆ B are smaller (see Figure 1 andFigure 5) so we have not included plots of these two ob-serving seasons. Nevertheless, as mentioned previously,the cyclical variability is still noticeable (see Figure 1). Amovie showing the ∆ V versus ∆( B − V ) as a function oftime for all four seasons is available in the online versionof the paper.We also produced models of the δ Sco system forthe recent periastron passage following the approach byHolman & Wiegert (1999). These models rely on grav-ity to follow the particles orbiting in a Keplerian fash- Jones et al.ion within the disk; gas dynamics are not included. Weadopt the star/disk system and orbital parameters fromCarciofi et al. (2006) and Tycner et al. (2011), respec-tively. The total mass of the disk for our simulationof 3 × − M ⊙ was calculated from the base density, ρ o = 4 . × − g cm − (Carciofi et al. 2006) with anaverage power-law density fall off of 3.5 for a disk trun-cated at 7 R ∗ . While the mass of δ Sco’s disk certainlychanges over time, and perhaps substantially in the innerdisk during mass ejection, this is a reasonable estimateof the total disk mass for our preliminary modeling. Fig-ure 9 shows the system 3.4 years after periastron passage.For this particular simulation, we assume that the disk,the equatorial plane of the star and the secondary alllie in the same plane. Although the secondary does notdirectly impact the disk, material is pulled away fromthe disk, forming a tidal tail that extends out to 5 AUat the time of the snapshot. Assuming a distance of135 pc (Tycner et al. 2011), the tail size is equivalentto ∼
40 milliarcsesconds. Over time, the tail will con-tinue to expand. Despite the fact that the secondarydoes not impact the disk, it does change the geometryof the disk system and these changes could potentiallyblock parts of the primary star and inner disk from thefield of view where the largest contribution to the V -band is produced. However, since the orbital time scalefor the secondary passage is of order of a decade, theremust be some other process that is responsible for thecyclical variability. We note, however, that if the disksize is much smaller than 7 R ∗ , then a tidal tail will notform since disks smaller than 5 R ∗ do not produce tailsin our simulations. We also note that tail production iscurtailed if the secondary orbit is not in the same planeas that of the disk. This prediction is consistent withChe et al. (2012), who find that the secondary passagedid not cause any mass outflow during periastron. DISCUSSION AND SUMMARY
Nonradial pulsations are generally believed to be thesource of short term variability, especially for the early-type Be stars where they are observed to be present in80 - 90% of the disk systems (Porter & Rivinius 2003).Typical periods vary from less than a day to several days(Rivinius et al. 2003). Time scales corresponding to theviscous disk model (discussed above) are much longer,from 10 to 10 days (Jones et al. 2008). The cyclicalvariability observed in δ Sco is intermediate in length andit is interesting to speculate on the possible mechanismrequired to produce variability on such a scale.A possible explanation for the variation in V -bandmagnitude is that δ Sco is subjected to both gravity dark-ening caused by rapid rotation and precession due to anunseen third companion. If so, as the star precesses, dif-ferent latitudes would come into view. Because the staris gravity darkened the variation in the effective temper-ature from pole to equator causes the amount of lightradiated to change with latitude. The star’s apparentbrightness would therefore change as the star is observedat different times during its precession.To test the effect of precession, a gravity darkened starwas simulated and the difference in apparent flux in thevisual band was estimated for different inclinations of thestar. To simplify the calculation, the star was assumed tobe emitting as a perfect black body. The star was divided into several latitudinal sections, each of which are con-sidered to be at constant effective temperature. Then,for the considered inclination angle, the projected areaof each latitudinal section was calculated. For simplic-ity, the star was assumed to be a perfect sphere insteadof an oblate spheroid. To estimate the relative appar-ent flux at a specific inclination angle, the projected areaand effective temperature of each latitudinal section wascalculated. The effective temperature was deduced bycalculating the average surface gravity for each latitudi-nal section and using the von Zeipel theorem (that statesthat the effective temperature is proportional to fourthroot of the local surface gravity (von Zeipel 1924)). Thetotal relative apparent flux in the V -band for a partic-ular inclination angle was then obtained by adding upthe relative apparent flux of each section and integratingover the V -band.Two cases were considered. The first case was for aprecessing star without any obstruction by a disk, whilein the second case, a static disk was added. It was as-sumed that the structure of the disk was unchanging andthat it completely blocks the V -band emission of theobstructed part of the star. The results of these testsshowed that even for the most extreme case by allowingthe inclination to vary pole-on to edge-on, the changesin V -band magnitude were orders of magnitude lowerthan the observed variation. Therefore it is unlikely thatthe variation in magnitude can be attributed to a pre-cession of the star. We note that the critical velocityof δ Sco is 620 km/s and with an inclination of 38 ± o (Carciofi et al. 2006) combined with observed vsini of 148km/s (Brown & Verschueren 1997) means that δ Sco isrotating well below critical by approximately 40%. Formore rapidly rotating Be stars, the variation in temper-ature from pole to equator would be more significant sothat it would be possible for precession to have a largereffect in other stars.Carciofi et al. (2006) considered whether or not achange in disk geometry could help explain the opti-cal fadings observed for δ Sco. They concluded, basedon δ Sco’s low inclination, that a significant warp-ing of the disk could block enough of the stellar sur-face to account for the fadings if enough material wasmoved to higher latitudes. In fact, tidal warping isfrequently invoked to explain Be star transitions fromsingly or doubly peaked emission lines to shell lines orvice versa (Martin et al. 2011). The size of the H α emitting region should be roughly equivalent to the sizeof the disk (Miroshnichenko et al. 2003) with the op-tical fading anti-correlated with H α line strength (seeMiroshnichenko (2011); Carciofi et al. (2006)). Recall,as discussed above, the V -band contribution will origi-nate in a volume of gas very close to the star and notover the full extent of the H α emitting region. Since δ Sco’s H α line transitioned from doubly to singly peakedin 2003 (Miroshnichenko et al. 2003), there is evidence tosupport this suggestion. The fact that the fadings havebeen observed to be anti-correlated with line emission(Miroshnichenko et al. 2003) seems to also require an in-crease in disk mass. Interestingly, Smith et al. (2012)noted for the binary system γ Cas that the quasi-secularbrightening occurred in the optical during 2010, andearly 2011 was correlated with a higher column densitywhich was manifested by an attenuation of the soft X-ray
Scorpii’s Circumstellar Disk 5flux. Figure 9 shows how material could be stripped fromthe disk at periastron, which could potentially shield por-tions of the inner disk where the disk V -band is produced.However, the ∼
10 year binary period is much too longto explain the cyclical variability.It is also interesting to consider whether or not an un-seen companion could cause the periodic variations. If ayet unseen companion orbits the primary with a 70 dayperiod, for example, we can determine the semi-majoraxis, a, of its orbit from Kepler’s Third Law. Assuminga stellar mass of 12.4M ⊙ for the primary (Tycner et al.2011), we find a = 0 .
77 AU. Here we assume that themass of the unseen companion is much less than the massof the primary. Tycner et al. (2011) give a radius for theprimary of 0.45 milliarcsecs. If δ Sco is at a distance of135 pc from Earth, the primary radius is then 0.061 AU,and a is 0.77/0.061 = 13R ∗ . Thus, if an unseen com-panion orbiting the primary is the source of the cyclicalvariability (70 days in this example), it must orbit theprimary near this distance. The minimum separation atperiastron of the secondary is 14 R ∗ (Tycner et al. 2011).This is very near the size of the orbit of the putative un-seen companion and raises the question of whether or notthe companion could remain in such an orbit in the long-term or would be destabilized by repeated passages of thesecondary. Holman & Wiegert (1999) examined the sta-bility of planets in binary star systems, and we can usetheir result to examine the stability of the hypotheticalunseen companion.Taking the secondary mass to be 8 M ⊙ (Tycner et al.2011), we can use their expression (1) to calculatethat the smallest stable orbit expected around the pri-mary is less than the primary star’s radius. We note,however, that δ Sco has a more extreme eccentric-ity, e = 0 . ∗ ; thus, our hypothetical companion on a 70 day orbitis still outside the regime of stability under this relaxedassumption. This does not mean that such a companioncould not survive a single periastron passage at 13 R ∗ ,only that it cannot do so over many such passages. Thus,it seems unlikely that an unseen companion with a ∼ V versus ∆( B − V ) colour-magnitude diagram with so-phisticated computational codes. Our initial models sim-ulate the addition and depletion of material in the innerregion of an axisymmetric circumstellar disk constructedusing a conventional power-law radial distribution. Gasis added to the inner edge of the disk during predeter-mined periods of constant mass loss. Similarly, gas is re-moved from the inner edge of the disk once mass transferfrom the star to the disk ends. The thermal structure ofthe disk is recomputed after each adjustment to the dis-tribution of gas in the disk using the radiative transfercode of Sigut & Jones (2007). The theoretical observ-ables are calculated using the Monte Carlo simulation ofHalonen & Jones (2012). Using this procedure, we cananalyze the evolving physical conditions of the circum-stellar gas through trends in predicted observables such as the colour-magnitude diagrams.Our results indicate that we can reproduce the ob-served changes in the magnitude and colour of the starusing models with different initial parameters. By chang-ing the base density of the disk and the size of the in-ner region that is cleared and refilled, we have deter-mined that the short-term mass-loss rates that occurover roughly 35 days range from 2 . − . × − M ⊙ / yr.These models reproduce the photometric trends plottedin Figure 2 that show the system becoming redder as itgets brighter. We expect that assiduous comparison ofthe shapes of the observed and predicted loops will re-duce much of the degeneracy between the models andfurther constrain our predicted mass-loss rates. Whilewe have initially restricted ourselves to a disk truncatedat 7 R ∗ to be consistent with Carciofi et al. (2006), weacknowledge that the truncation radius of the disk is anadditional parameter that should be constrained by otherobservations. Rivinius et al. (2012) suggest that, afterthe periastron in 2011, a disturbance propagated inwardthroughout the disk. It is interesting to speculate thatperhaps this caused the disk to become smaller in 2011and 2012. We note that the photometric cycles in the B and V passbands vary in length and amplitude (see Ta-ble 2) from year to year. This interesting complicationwill also require further study. It is clear that detailedmodeling, including changes in disk geometry, episodicand/or asymmetrical mass-loss rates must be invoked toexplain all of the observed peculiarities of this system.We are currently constructing models with various den-sities, disk sizes, evacuated regions and other importantdisk properties to see if the observations can be usedto constrain the nature of the changing condition in thedisk. The presentation and analysis of these models isthe focus of a follow-up paper.C.E.J. and R.J. H. acknowledges research supportedby NSERC, the Natural Sciences and Engineering Re-search Council of Canada. Astronomy at Tennessee StateUniversity is supported by NASA, NSF, Tennessee StateUniversity, and the state of Tennessee through its Cen-ters of Excellence programs. REFERENCESBedding, T. R. 1993, AJ, 106, 768Brown, A. G. A., & Verschueren, W. 1997, A&A, 319, 811Cranmer, S. R. 2005, ApJ, 639, 1081Carciofi, A. C., & Bjorkman, J. E. 2006, ApJ, 639, 1081Carciofi, A. C., Miroshnichenko, A. S., Kusakin, A. V., Bjorkman,J. E., Bjorkman, K. S., & 7 coauthors 2006, ApJ, 652, 1617Carciofi, A. C., Okazaki, A. T., Le Bouquin, J.-B., ˘Stefl, S.,Rivinius, Th., Baade, D., Bjorkman, J. E., & Hummel, C. A.2009, A&A, 504, 915Carciofi, A. C. 2011 in Proceedings of the InternationalAstronomical Union, Symposium S272, Active OB stars:structure, evolution, mass loss, and critical limits, ed. C.Neiner, G. Wade, G. Meynet, & G. Peters (Cambridge, IAU)272, 384Carciofi, A. C., Bjorkman, J. E., Otero, S. A., Okazaki, A. T.,˘Stefl, S., Rivinius, Th., Baade, D., & Hauvois, X. 2012, ApJ,744, L15Carciofi, A. C. 2012 in AIP Conference Proceedings, StellarPolarimetry: From Birth to Death (Cambridge, AmericanInstitute of Physics) 1429, 121Che, X., Monnier, J. D., Tycner, C., and 10 co-authors 2012,ApJ, 757, 29
Jones et al.
Cot´e, J., & van Kerkwijk, M. H. 1993, A&A, 274, 870de Wit, W. J., Lamers, H. J. G. L. M., Marquette, J. B., &Beaulieu, J. P. 2006, A&A, 456, 1027Eaton, J. A., Henry, G. W., & Fekel, F. C. 2003, in The Future ofSmall Telescopes in the New Millennium, Vol. II, TheTelescopes We Use, ed. T. D. Oswalt (Dordrecht: Kluwer), 189Eaton, J. A., & Williamson, M. H. 2007, PASP, 119, 886Gandet, T. L., Otero, S., Fraser, B., West, J. D. 2002,Information Bulletin on Variable Stars, 5352, 1Halonen, R. J., & Jones, C. E. 2012, ApJ, submittedHenry, G. W. 1995, in ASP Conf. Ser. 79, Robotic Telescopes:Current Capabilities, Present Developments, and FutureProspects for Automated Astronomy, ed. G. W. Henry & J. A.Eaton (San Francisco: ASP), 44Henry, G. W. 1999, PASP, 111, 845Holman, M. J., & Wiegert, P. A. 1999, AJ, 117, 621Johnson, J. A., Clanton, C., Howard, A. W., Bowler, B. P.,Henry, G. W., Marcy, G. W., Crepp, J. R., Endl, M., Cochran,W. D., MacQueen, P. J.,Wright, J. T., & Isaacson, H. 2011,ApJS, 197, 26Jones, C. E., Sigut, T. A. A., & Porter, J. M. 2008, MNRAS, 386,1922Lee, U., Saio, H., & Osaki, Y. 1991, MNRAS, 250, 432McDavid, D., Bjorkman, K. S., Bjorkman, J. E., & Okazaki, A.T. 2000 in Proceedings of the International AstronomicalUnion, Colloquium 175, The Be Phenomenon in Early-TypeStars, ed. M. A. Smith, H. F. Henrichs, & J. Fabregat(Orem,ASP), 214, 460Martin, R. G., Pringle, J. E., Tout, C. A., & Lubow, S. H. 2011,MNRAS, 416, 2827McGill, M. A., Sigut, T. A. A., & Jones, C. E. 2007, ApJ, 743, 111Meilland, A., Stee, P., Vannier, M., Millour, F., Domiciano deSouza, A., Malbet, F., Martayan, C., Paresce, F., Petrov, R.G., Richichi, A., & Spang, A. 2007, A&A, 464, 59Meilland, A., Delaa, O., Stee, Ph., Kanaan, S., Millour, F., &coauthors 2011, A&A, 532, 80Miroshnichenko, A. S., Bjorkman, K. S., Morrison, N. D.,Wisniewski, J. P., Manset, N., & 5 coauthors, 2003, A&A, 408,305 Miroshnichenko, A. S. 2011, in Proceedings of the InternationalAstronomical Union, Symposium S272, Active OB stars:structure, evolution, mass loss, and critical limits, ed. C.Neiner, G. Wade, G. Meynet, & G. Peters (Cambridge, IAU),272, 304Okazaki, A. T. 2001, PASP, 53, 119Otero, S., Fraser, B., & Lloyd, C. 2001, Information Bulletin onVariable Stars, 5026, 1Oudmaijer, R. D., Wheelwright, H. E., Carciofi, A. C., Bjorkman,J. E., & Bjorkman, K. S. 2011, in Proceedings of theInternational Astronomical Union, Symposium S272, ActiveOB stars: structure, evolution, mass loss, and critical limits, ed.C. Neiner, G. Wade, G. Meynet, & G. Peters (Cambridge,IAU), 272, 418Perry, C. L., Olsen, E. H., & Crawford, D. L. 1987, PASP, 99,1184Porter, J. M., Rivinius, T. 2003, PASP, 115, 1153Rivinius, Th., Baade, D., & ˘Stefl, S. 2003, A&A, 411, 229Rivinius, Th., ˘Stefl, S., Baade, D., Carciofi, A. C., Otero, S.,Miroshnichenko, A. S., & Manset, N. 2012, in ASP ConferenceSeries 464, in pressSigut, T. A. A., & Jones, C. E. 2007,ApJ, 668, 481Smith, M. A., Lopes de Oliveira, R., Motch, C., Henry, G. W.,Richardson, N. D. & 15 coauthors 2012, A&A, 540, 53Struve, O. 1931, ApJ, 73, 94Tango, W. J., Davis, J., Jacob, A. P., Mendez, A., North, J. R., &3 coauthors 2009, MNRAS, 396, 842Townsend, R. H. D., Owocki, S.P., & Howarth, I.D. 2004,MNRAS, 350, 189Tycner, C., Ames, A., Zavala, R. T., Hummel, C. A., Benson, J.A., & Hutter, D. J. 2011, ApJ, 729, L5von Zeipel, H. 1924, MNRAS, 84, 665Wheelwright, H. E., Bjorkman, J. E., Oudmaijer, R. D., Carciofi,A. C., Bjorkman, K. S., & Porter, J. M. 2012, MNRAS, 423,L11
Scorpii’s Circumstellar Disk 7
Figure 1.
Photometric data of δ Sco acquired over four years with the T3 0.4 m APT at Fairborn Observatory. The vertical and horizontalscales for all four panels are identical to facilitate direct comparison of the brightness variations in the V and B filters, in the B − V colorindex, and brightness changes in the K − C differential magnitudes. The arrow in the top panel marks the time of the latest periastronpassage on UT 2011 July 6 ± σ , of the K − C is provided in the lower left of the figure. Jones et al.
Figure 2.
Top two panels: Individual ∆ V magnitudes and ∆( B − V ) color indices for 2009. Brightness and color are anti-correlated.Bottom two panels: Same as the top two panels but for 2010. Scorpii’s Circumstellar Disk 9
Figure 3.
Top two panels: ∆ V magnitudes and ∆( B − V ) color indices for 2011. Brightness and color are now directly correlated.Bottom two panels: Same as the top two panels but for 2012. V magnitudes and ( B − V ) color indices appear to be significantly out ofphase. Figure 4.
T op : Frequency spectrum of the 2010 Johnson B photometry after removing a linear trend from the data. Best frequency is0.01387 c d − . Bottom : B data from 2010 phased with the corresponding best period of 72.1 days. The peak-to-peak amplitude is 0.156mag. See Table 2 for the complete results of our period determinations. Scorpii’s Circumstellar Disk 11
Figure 5.
This figure shows the ∆ V magnitude versus ∆( B − V ) for all four observing seasons. The 2009, 2010, 2011, and 2012 seasonsare indicated by the four shades of gray dots from light to dark, respectively. The error estimate for each value is represented by one errorbar in the lower right of the figure. Figure 6.
This figure shows ∆ V versus ∆( B − V ) as a function of time for the first 65 days of the 2009 observing season from the ReducedJulian dates 54861-54922. The color from light gray to dark gray indicates the progression with time. The observational error for eachdata value is given in the lower right. The lines connect the data in time with the arrows indicating the sense of direction. Scorpii’s Circumstellar Disk 13
Figure 7.
Same as Figure 6 for the second half of the observing season corresponding to Reduced Julian dates 54923-55006.
Figure 8.
Same as Figure 6 for the entire 2010 from Reduced Julian dates 55222-55384.
Scorpii’s Circumstellar Disk 15 −10 −5 0 5 10 − − AU A U Figure 9.
Disk particles (open circles) after periastron passage of the secondary (shown by the grey dot in the upper right). In this case,the disk is 7 R ∗∗