Three-Dimensional Doppler Tomography of the RS Vulpeculae Interacting Binary
DDraft version November 2, 2018
Preprint typeset using L A TEX style emulateapj v. 08/22/09
THREE-DIMENSIONAL DOPPLER TOMOGRAPHY OF THE RS VULPECULAE INTERACTING BINARY
Mercedes T. Richards
Department of Astronomy & Astrophysics, Pennsylvania State University, 525 Davey Laboratory, University Park, PA, 16802, USA
Olga I. Sharova
Radiophysical Research Institute (NIRFI), 25/12a, Bolshaya Pecherskaya St., Nizhny Novgorod, 603950, Russia andMichail I. Agafonov
Radiophysical Research Institute (NIRFI), 25/12a, Bolshaya Pecherskaya St., Nizhny Novgorod, 603950, Russia
Draft version November 2, 2018
ABSTRACTThree-dimensional Doppler tomography has been used to study the H α emission sources in the RSVulpeculae interacting binary. The 2D tomogram of this binary suggested that most of the emissionarose from the cool mass losing star with additional evidence of gas flowing close to the predictedtrajectory. However, the 3D tomogram revealed surprising evidence of a more pronounced gas streamflow at high V z velocities from -240 to -360 km s − . This behavior is most likely caused by magneticactivity on the cool star since the central velocity plane, defined by V z = 0 km s − , should be coincidentwith the orbital plane of the binary if the flow is dominated by gravitational forces only. RS Vul hasbeen detected as both an X-ray and a radio source, and it is possible that the RS Vul gas stream mayhave been deflected by magnetic field lines. This flow is distinctly different from that found in thestreamlike state of U CrB, in which the gas flow was confined mostly to the central velocity plane. Subject headings: techniques: image processing – accretion, accretion disks – stars: binaries: close –(stars:) binaries: eclipsing – (stars:) circumstellar matter – stars: imaging – stars:individual (RS Vulpeculae,U Coronae Borealis) INTRODUCTION
The extension from 2D to 3D tomography is now areality. Two-dimensional Doppler images can be cre-ated with the standard Filtered Back Projection (FBP)Method ( e.g.,
Kaitchuck et al. 1994) by projecting thesources of emission onto a single central ( V x , V y ) velocityplane. This plane is expected to coincide with the orbitalplane of the binary since this is the plane in which thegravitational forces dominate the gas flows. However,these sources may have non-zero velocities beyond thecentral 2D plane since they are not geometrically thinand are not confined to the orbital plane. Instead, 3Dimages can be derived over a range of V z values transverseto the central plane. The resulting 3D maps are consis-tent with the 2D Doppler tomograms obtained with theFBP method and reveal substantial flow structures be-yond the central velocity plane. The 3D reconstructionmethod applied in this work is the RadioastronomicalApproach (RA) developed by Agafonov (2004a,b); Aga-fonov & Sharova (2005a,b). A detailed discussion of thedifferences between this technique and the FBP methodcan be found in Agafonov, Richards & Sharova (2006)and Agafonov, Sharova & Richards (2009). In particu-lar, the RA method is especially useful when the numberof projections is relatively small because it can achievethe same resolution as FBP while using fewer profiles(see Figure 1 of Agafonov, Richards & Sharova 2006).The effectiveness of the RA method has been examined Electronic address: [email protected] address: shol@nirfi.sci-nnov.ruElectronic address: agfn@nirfi.sci-nnov.ru by using initial 3D test models, calculating 1D profilesfrom these models, and then performing the reconstruc-tion. (Sharova 2006) compared the computed modelswith the initial models and showed that the computedemission maxima coincided with the initial model, andall details were reproduced. So, there was very goodagreement between the initial test model and the recon-structed 3D tomogram.The two-dimensional Doppler tomograms of the Algol-type binaries (Richards, Albright & Bowles 1995; Al-bright & Richards 1996; Richards 2004) display a di-verse range of circumstellar structures. These include agas stream, accretion annulus, transient accretion disk,shock regions and sometimes a chromospheric emissionsource in the short-period (P < ≥ α spectra of this binary col-lected in 1993 and 1994 displayed two distinct patterns:a transient accretion disk representing the disklike state(in epoch 1993) and a bright gas stream flowing along thepredicted ballistic trajectory representing the streamlikestate (in epoch 1994). The 3D tomograms of these emis-sion sources were even more distinct than the 2D ver-sions, and both states showed evidence of a prominent a r X i v : . [ a s t r o - ph . S R ] S e p Richards, Sharova, & Agafonovemission source associated with the mass gainer, result-ing from the impact of the gas stream onto the photo-sphere of that star (Agafonov, Richards & Sharova 2006;Agafonov, Sharova & Richards 2009). Chromosphericemission from the mass loser was also visible especiallywhen the binary was in the disklike state. The 3D im-age also showed distinct evidence of gas flows (jets) withhigh V z velocities. These jets were associated with threeemission sources: (1) the impact/splash region where thegas stream strikes the mass gainer in the streamlike state( V z ∼ -300 to +100 km s − ); (2) the interaction betweenthe gas stream and accretion disk in the disklike state( V z ∼
300 to 400 km s − ); and (3) the Localized Re-gion where the disk material interacts with the incominggas stream after traveling around the mass gainer ( V z ∼ +200 to +500 km s − ). These high values are ex-pected since the star-stream impact occurs at speeds ofup ∼ − , compared to the slower rotationof the mass gainer (Richards 1992; Blondin, Richards &Malinkowski 1995; Richards & Ratliff 1998). Moreover,the pattern of V z velocities associated with the disk emis-sion implies that the disk may be inclined to the orbitalplane; which subsequently suggests that the rotation axisof the mass gainer could be precessing as a consequenceof the star-stream impact. This result would not havebeen possible from 2D tomograms alone.In this paper, we have continued our exploration of3D tomography by studying the RS Vul system. Specifi-cally, we have 1) identified the features in the 3D imagesand related them to those seen in the 2D tomograms; 2)compared the 3D tomogram of RS Vul with those of UCrB; and 3) provided a physical interpretation of thesefeatures. The system parameters and 2D results are de-scribed in Section 2, the 3D tomograms are described inSection 3, a model for RS Vul is given in Section 4, andthe Conclusions are outlined in Section 5.
2D DOPPLER TOMOGRAPHY OF RS VUL
The Algol-type binary RS Vulpeculae is an interact-ing system in which the cooler G1 III secondary staris transferring mass to its B5 main-sequence companion(the primary) through Roche lobe overflow. RS Vul isa short-period Algol ( P orb = 4.4776635 days; Holmgren1989) with an orbital inclination of 78.7 ◦ (Hutchings &Hill 1971). The properties of the system are M p = 6.59 M (cid:12) , M s = 1.76 M (cid:12) , R p = 4.71 R (cid:12) , R s = 5.84 R (cid:12) , sys-temic velocity V o = -20.1 km s − , velocity semi-amplitude K p = 54.0 km s − , with a mass ratio, q =0.27 (Holmgren1989). Eighty-one H α (6562.8 ˚A) spectra with a reso-lution of 0.166 ˚A/pixel or 7.6 km s − were collected in1993 with the KPNO 0.9m Coud´e Feed Telescope (seeRichards & Albright 1999). The spectra were collectedat closely spaced positions around the entire orbit of thebinary. Since the observed spectra are dominated bythe spectrum of the B5 mass gaining star, model atmo-spheres calculations were used to represent the photo-spheric contributions of the stars, and these stellar con-tributions were subtracted from the observed spectra tocreate difference spectra (see Richards & Albright 1999for details). The difference spectra were used to calculatethe tomograms.The standard 2D Doppler tomogram of RS Vul 1993 isshown in Figure 1 along with the 2D tomogram of U CrB in the streamlike state (from Richards (2001)). These 2Dimages were compared with the 3D Doppler tomogramsbased on the same data. In the tomograms, the solidtrajectory is the gravitational free-fall path of the gasstream; and the circles along this trajectory are markedat intervals of a tenth of the distance from the L point tothe distance of closest approach to the mass gainer. Thelargest solid circle and the smaller dashed circle mark theinner and outer edge of a Keplerian disk, respectively; theasterisk is the predicted location where the gas streamshould strike the photosphere of the mass gainer; and theplus sign marks the center of mass of the binary. TheRS Vul tomogram shows the presence of the gas stream,a circumprimary emission source, chromospheric emis-sion, and a region dominated by absorption called theabsorption zone (Richards 2001). The strongest emis-sion source corresponds to the chromospheric H α emis-sion (Richards & Albright 1996).
3D DOPPLER TOMOGRAPHY
Reconstruction of the 3D Tomogram
For the reconstruction of the 3D Doppler tomograms,we analyzed the same set of H α line profiles that wereused by Richards (2001) to construct the standard 2DDoppler tomogram of RS Vul. The 3D Doppler tomo-gram was calculated as a set of numerical values in thecells of a cube, with dimensions ( V x , V y , V z ) having val-ues ranging from -700 to +700 km s − . The maximumof this function was normalized to the unit to permitthe visualization of the image based on any slice takenfrom any direction. The velocity resolution of the imagedepends on the number of projections (i.e., number ofspectra) and their distribution in orbital phase; and theorbital inclination influences the ratio of the resolutionsin the V x , V y , and V z directions. The wavelength resolu-tion of the spectra is 7.6 km s − , so it plays a smaller rolein setting the resolution of the image since it is alreadysufficient to produce the best possible velocity resolutionfor the tomogram. In particular, the velocity resolutionsin the ( V x , V y ), ( V z , V x ), or ( V z , V y ) planes are equal onlyfor an inclination of 45 ◦ ; and the resolutions in the V x and V y directions are better than that in the V z directionfor high values of inclination. Since the orbital inclina-tion of RS Vul is high ( i = 78.7 ◦ ), the 3D tomogramwas restored with a resolution of 30 km s − in the V x and V y directions and 105 km s − in V z direction. Sincethe resolution in the V z direction is approximately 3.5times lower than in the other directions, features on thetomogram may look elongated in this direction. The ap-pearance of the images has been simplified by displayingthe ( V x , V y ) slices in the horizontal plane to be consis-tent with the direction of the orbital plane of the binary,and the slices containing the V z axis are in the verticaldirection.A direct comparison was made between the observedspectra and those computed from the reconstructed 3DDoppler tomograms. Figure 2 shows this match for the3D Doppler tomograms of RS Vul. The radial velocity, V r is plotted versus orbital phase, with a resolution of 30km s − . The observed spectra are displayed in the leftframes and the spectra computed from the 3D Dopplertomograms are displayed in the middle frames. Two S-wave patterns are noticeable in the observed and com-hree-dimensional Doppler Images of RS Vul 3puted spectra corresponding to the motions of the twostars. The overall agreement between the observed andcomputed spectra is very good. The quality of the fit isshown in the right frame of Figure 2, which displays therelative chi-square statistic, χ /χ c , versus orbital phase.Here, χ is normalized to the critical value of χ c , whichcorresponds to the largest acceptable value of χ at the99% confidence level (e.g., Skilling & Bryan 1984). Fig-ure 2 shows that χ was less than 20% of the critical valueat most orbital phases and it was only as high as 65% ofthe critical value over a narrow phase range from 0.47 to0.49, perhaps because the 3D code assumes that the gasis optically thin when it might be optically thick at thesephases. So, in all cases, χ was much lower than the crit-ical value which confirms that the 3D reconstruction hasproduced a reliable computed model for RS Vul. Description and Interpretation of EmissionFeatures
RS Vul and U CrB have similar geometries, so we ex-pect to identify similar features in their Doppler images.Figure 3 displays 20 slices in the ( V x , V y ) plane for val-ues of V z ranging from −
540 km s − to +540 km s − ,in 60 km s − steps. The superposition of the color orgray-scale images with the contour images allows us toemphasize the main features in each V z slice of the 3Dimage. Figure 4 displays the images of seven representa-tive V z slices over the range from V z = −
480 km s − to+480 km s − . A similar figure for U CrB in the stream-like state is shown in Figure 5 for comparison. Figure 6displays several cross-sections of the flow in the ( V y , V z )plane. In Figures 3 – 6, the light (yellow) areas with bluecontours represent absorption while the dark (deep pink)areas with red contours represent emission.Agafonov, Sharova & Richards (2009) identified sev-eral features in the 3D tomogram of U CrB that wereconsistent with those found in the 2D image studied byRichards (2001): (1) circumprimary emission (centeredon the mass gainer) and an accretion annulus (a circularregion of low velocity); (2) chromospheric emission fromthe cool donor star (the secondary); (3) the gas stream flowing along its predicted ballistic trajectory; (4) the star-stream impact region , where the gas stream strikesthe stellar surface; (5) emission within the predicted locusof the accretion disk ; (6) a localized region between thestars where the gas stream strikes material that has cir-cled the mass gainer; (7) a second localized region wherethe gas stream makes impact with the outer edge of theaccretion disk; (8) a jet or flare associated with the chro-mosphere of the secondary star; and (9) a high-velocitystream or jet extending in the Vz-direction beyond thestream-star impact region.The 3D image of RS Vul displays all of the emissionsources found in U CrB (see Figs. 3 and 4), except thatthere was almost no emission from an accretion disk andlittle evidence of the star-stream impact region. Figure 3shows that the strongest emission sources are associatedwith the cool mass loser as well as the mass gainer. Theemission sources are described in Table 1 along with theirtypical velocity ranges and a brief description of eachfeature is provided below.1. The circumprimary emission centered on the velocityof the mass gainer in RS Vul is labeled ( ) in Figure 4. It is the strongest source of emission in the 3D image(see Figure 3, slice at V z = −
160 km s − ). This featurecan be seen in the slices with V z velocities from 0 to -320km s − in Figures 3 and 6 (see also Table 1). Moreover,it is distinctly different from the same feature seen in UCrB 1993, which displayed a symmetric V z distributionfrom V z = −
180 to +180 km s − (see Table 1 in Aga-fonov, Sharova & Richards 2009). The circumprimaryemission has higher velocities than expected from syn-chronous rotation, so the star may have been spun up bythe impact of the gas from the donor star.2. Chromospheric emission from the cool mass losingstar is feature ( ) and it is comparable in strength tothe circumprimary emission. It is associated with a widerange of V z velocities from −
530 to +350 km s − (seeFigures 3, 4, 6, and Table 1). This range is illustratedin Figure 6 and resembles a jet-like feature in the V z direction. The feature has the same velocity as the masslosing star.3. The gas stream flowing along its predicted ballistictrajectory is feature ( ). It flows from the L1 point to-wards the mass gaining star with speeds up to 450 km s − in the ( V x , V y ) plane, with a V z velocity of −
300 km s − .RS Vul is unusual because the flow along the gas streamis not strongest in the V z = 0 km s − frame but in the V z = -240 to -360 km s − range (see Figs. 3 and 4). Thisis distinctly different from U CrB in its streamlike state(Fig. 5), in which the gas stream flow was strongest inthe central velocity plane ( V z = 0 km s − ) and locateddirectly on the predicted trajectory of the gas stream inthe ( V x , V y ) plane. This is not the case for RS Vul.4. The predicted location of the star-stream impact re-gion , where the gas stream strikes the stellar surface, isfound at ( ) near the location where the gas stream tra-jectory intersects the inner part of the disk. Even in itsmost extended state (when it is at V z = -300 km s − ),the gas stream (Feature ) never gets to this predictedvelocity, but is truncated at the velocity corresponding tothat of Keplerian flow at the surface of the mass gainer.5. There is very little evidence of any emission within the predicted locus of the accretion disk (Feature ) in the RSVul images, although some clumping is seen within thisregion. This feature was much stronger in U CrB, andthere is slightly more space around the mass gainer inU CrB to form a disk. However, gas may have traveledaround the mass gainer and created the localized regions(described below) when the interaction with the streamslowed the gas flow.6. A Localized region (LR) between the stars labeled ( )was found near the center of the tomogram where theflow around the mass gainer is expected to strike the in-ner edge of incoming gas stream (see Cartesian models of β Per (Richards 1992, 1993), RW Tau (Vesper & Hon-eycutt 1993), and other direct-impact systems). In RSVul, this region is detected at high positive V z velocitiesranging from +50 to +300 km s − and seems to developinto the structure seen at the L1 point from V z = +360to 540 km s − (Fig. 3). The classical localized regionbetween the stars is seen along the V x =0 km s − line for V z velocities from 50 to 540 km s − in Figure 3. How-ever, it is difficult to say whether the part with L1 pointvelocities in the ( V x , V y ) plane for V z = +180 to +300km s − (Fig. 4) should be considered part of the cool Richards, Sharova, & Agafonovstar or part of the LR. This feature is also seen in Figure6 slices for V x = −
20 to +20 km s − . This kind of brakingwas first suggested by Richards (1992) in the study of theAlgol-type binary β Persei and confirmed using hydrody-namic simulations of Blondin, Richards & Malinkowski(1995) and Richards & Ratliff (1998). Similar resultswere found later from gas dynamical modeling of cata-clysmic variables by Kuznetsov et al. (2001), Bisikalo etal. (2000a), and Bisikalo et al. (2000b).7. A second compact region labeled ( ) also has high V z velocities from 0 to 350 km s − and is located on the V y = 0 km s − line (see Figures 3 and 4). In Figure 6,it can be seen in the V x = 100 to 120 km s − frames,with an average V z velocity of about 250 km s − . Thisfeature may later merge with the other localized regionthat runs along the line of centers between the stars.8. The feature labeled ( ) has a velocity similar to the V x and V y velocities of the donor star, but with highpositive V z velocities in the range from V z =+180 to +540km s − (see Figures 3 and 4). It has V x and V y velocitiesof 100 to 300 km s − (see Figs. 3, 4, and 6). Thisfeature could be the result of chromospheric emission onthe cool mass loser, and the high velocities suggest thatit could correspond to a jet or flare. However, anotherexplanation is that it is related to the interaction betweenthe inner edge of the gas stream and the material thathas circled the mass gainer. A similar emission sourcewas found in U CrB.9. The absorption feature in RS Vul seen in the lowerleft and right quadrants of Figure 3 from V z = -120 to+540 km s − was called the absorption zone by Richards(2001, 2004), and is associated with the locus wherethe gas temperature becomes too high to emit at H α .A structure at a similar location was identified in theultraviolet tomogram of U Sge by Kempner & Richards(1999) and in the H α tomogram of U CrB by Agafonov,Sharova & Richards (2009).The 3D tomograms also reveal an interesting symmetryin the emission sources found in the V z slices. A strongsource near the mass losing star (in velocity space) is de-tected with intensities of 0.852 and 0.955 in symmetricallocations at V z = −
300 and +300 km s − , respectively(see Figure 4). A similar symmetric intensity pattern(I = 0.343 and 0.333) is seen at V z = −
480 and +480km s − . In the case of the mass gaining star, anothersymmetric pattern is revealed. The most intense emis-sion feature in the V z = 180 km s − frame in Figure 4,with I=0.612, is located along the V x = 0 km s − lineat V y = +80 km s − , in the opposite V y direction fromthe circumprimary emission. These symmetries may bepurely coincidental or perhaps we can see the same fea-ture with motions on both sides relative to the central ve-locity plane, with a − V z component in − z direction andwith a + V z component in + z direction. Nevertheless, itis expected that the gas flows should be symmetric aboutthe central velocity plane if gravitational forces dominatethe gas flows. So, the displacements of emission features (1) , (2) , and (3) from the central velocity plane shouldnot be unusual. A MODEL FOR RS VUL
Figure 7 displays the 3D tomograms of RS Vul and UCrB in a 3D velocity cube together with the 2D inte- gral images for the main planes: ( V x , V y ), ( V x , V z ), and( V y , V z ). These 2D integral images represent the inte-grals along the V z , V y , and V x directions, respectively;and the image in the ( V x , V y ) plane is equivalent to the2D Doppler tomogram. The U CrB images in this figureare derived from the combined disk and stream states ofthat system. The 3D velocity cube of RS Vul illustratesthat RS Vul has no gas stream emission along the pre-dicted trajectory in its central velocity plane, althoughthe 2D integral image in the ( V x , V y ) plane shows thatthe gas stream emission is indeed prominent, but at V z velocities much higher than expected. In addition, bothbinaries display a gas velocity distribution that is inclinedto the vertical direction ( V x = 0 km s − ) in the 2D inte-gral ( V x , V z ) plane, but vertical in the 2D integral ( V y , V z )plane.A pattern has emerged from the study of the 3D to-mogram of RS Vul. Both the circumprimary emissionsource labeled earlier as (1) and the gas stream (3) wereoffset toward negative V z velocities, while the chromo-spheric emission (2) had V z velocities over the range from −
530 to +350 km s − (see also Figure 6). So, the cir-cumprimary emission was strong over the same range of V z velocities corresponding to the strongest gas streamemission. Moreover, the gas stream emission extendedto the Keplerian velocity at the photosphere of the massgainer, while the mass gainer was spinning faster thanthe synchronous value. Therefore, the results are consis-tent with the assumption that the circumprimary emis-sion (from a photospheric bulge) is produced when thegas stream strikes the surface of the star. In addition,the mostly negative V z velocities for the gas stream andthe circumprimary emission suggest that the gas does notmove in the central velocity plane, as expected if grav-itational forces dominate the gas flows, and so the flowaround the mass gainer is concentrated on one side of thecentral velocity plane.While the image reconstruction process can now pro-duce 3D velocity images, we are still unable to create thedesired 2D or 3D Cartesian images because the variousemission features have different velocity fields. However,an estimate of the gas stream flow can be made frominformation provided in Figure 1. Based on the pointwhere the gas stream is truncated, we can produce aCartesian model to show the dominant emission sourcesin RS Vul. The distance from the L point to the centerof the primary star has been divided into 10 equal seg-ments in the center frame of Figure 1, and the similarsmall circles on the Doppler tomograms are marked atvelocity intervals corresponding to those distance inter-vals. Comparing the center and right frames of Figure 1,we see that the gas stream flow is truncated after trav-eling 7 small circles on the gas stream trajectory, whichis equivalent to 70% of the distance from the L pointto the center of the primary star. The star-stream in-teraction occurs along the path of the gas stream rightwhere the gas stream ends (at 70% of the distance fromthe L point). Both Figures 3 and 4 show that the gasflow reaches the stellar surface corresponding to the solidcircle in the tomogram, which corresponds to the Keple-rian velocity at the surface of the mass gainer. Beyondthis point, there is no evidence of any significant disk inthis binary at this epoch.hree-dimensional Doppler Images of RS Vul 5Evidence of circumstellar gas distributed well aboveand below the orbital plane of the binary has been de-rived from studies of ultraviolet resonance lines and theH α line of several Algol systems ( e.g., Peters and Pol-idan 1984, Richards 1993). This gas can reach heightscomparable to the radius of the mass gaining star. How-ever, semi-analytical ballistic studies ( e.g.,
Lubow andShu 1975, 1976) or hydrodynamic simulations of the masstransfer process ( e.g.,
Richards & Ratliff 1998) assumethat the gas should be found very close to the orbitalplane, if magnetic fields are not involved. However, Sternet al. (1992) suggested that coronal mass ejections couldaccount for more than 10% of the mass transfer expectedsolely from Roche lobe overflow by gravitational pro-cesses alone. The coronal behavior of the cool stars in Al-gol binaries is enhanced because these stars rotate morerapidly than the slowly rotating Sun. If we can scalethe solar coronal behavior to the enhanced levels seenin many Algol binaries, then radio and X–ray flares de-tected from these systems can be interpreted as the peri-odic reconnection of the coronal field (Stern et al. 1992).RS Vul was detected as an X-ray source by White & Mar-shall (1983) with an X-ray luminosity, L x = 2 . × erg s − , which is about half of the x-ray luminosity of β Per. Radio emission from RS Vul was also detectedby Umana et al (1998) with a flux density of 0.26 mJyand radio luminosity, L radio = 3 . × ergs cm − s − Hz − , compared to 0.30 mJy and 9 . × ergs cm − s − Hz − for U CrB. So, the magnetic fields of the activestars in RS Vul and U CrB should be comparable.The long-term radio flare survey of the prototype Algolsystem β Per by Richards et al. (2003) demonstratedthat major flares (up to 1Jy) occur regularly and pre-dictably every 48.9 ± − and could be as high as100 yr − (for weaker flares every orbit). Mass ejectionsfrom X-ray flares observed with EXOSAT and GINGAare estimated to be (cid:39) × − M (cid:12) (Stern et al.1992), and a total mass ejection rate of (cid:39) × − M (cid:12) yr − would correspond to the detected flar- ing rate of 7.5 yr − for β Per. These arguments suggestthat coronal mass ejections may influence the transfer ofgas between stars and even the angle at which the flow isdirected. The differences between the gas distributionsin RS Vul and U CrB cannot be readily explained bydifferences in magnetic activity based on their x-ray andradio luminosities. However, it is evident that magneticfields may play a role in the distribution of circumstellargas in both systems. CONCLUSIONS
Three-dimensional Doppler tomography has been usedto study the H α emission sources in the RS Vulpeculaeinteracting binary and has led once again to the discoveryof gas flows with significant V z velocities. The 2D tomo-gram of this binary suggested that most of the emissionarose from the cool mass losing star with additional ev-idence of gas flowing close to the predicted trajectory.However, the 3D tomogram revealed surprising evidencethat a more pronounced flow of gas along the stream wasnot found as expected in the central velocity plane ( V z = 0 km s − ) but at V z velocities from -240 to -360 kms − . Magnetic activity on the cool mass losing star mayexplain the deflection of the gas stream flow away fromthe orbital plane of the binary where the gravitationalforces dominate the interactions between the stars. Thisflow behavior is distinctly different from that found in thestreamlike state of U CrB, in which the gas flow was con-fined mostly to the central velocity plane. Exploration ofthe 3D velocity structures of other binary systems shouldimprove our understanding of these out-of-plane gas mo-tions.We thank the referee for comments on the manuscript.This research was partially supported by the RussianFoundation for Basic Research (RFBR) grants 06-02-16234 and 09-02-00993, and National Science Foundationgrant AST-0908440. REFERENCESAgafonov, M. I. 2004a, Astron. Nachr., 325, 259Agafonov, M. I. 2004b, Astron. Nachr., 325, 263Agafonov, M. I., Richards, M. T. & Sharova, O. I. 2006, ApJ,652, 1547Agafonov, M. I. & Sharova, O. I. 2005a, Astron. Nachr., 326, 143Agafonov, M. I. & Sharova, O. I. 2005b, Radiophysics andQuantum Electronics, 48, No.5, 329Agafonov, M. I., Sharova, O. I. & Richards, M. T. 2009, ApJ,690, 1730Albright, G. E. & Richards, M. T. 1996, ApJ, 459, L99Bisikalo, D. V., Boyarchuk, A. A., Kuznetsov, O. A., &Chechetkin, V. M. 2000, Astronomy Reports, 44, 26Bisikalo, D. V., Harmanec, P., Boyarchuk, A. A., Kuznetsov, O.A., & Hadrava P, P. 2000, A&A, 353, 1009Blondin, J. M., Richards, M. T., & Malinkowski, M. 1995, ApJ,445, 939Holmgren, D. 1989, Space Sci Rev. 50, 347Hutchings, J. B. & Hill, G. 1971, ApJ, 166, 373Kaitchuck, R. H., Schlegel, E. M., Honeycutt, R. K., Horne, K.,Marsh, T. R., White, J. C., & Mansperger, C. S. 1994, ApJS,93, 519Kempner, J. C., & Richards, M. T. 1999, ApJ, 512, 345Kuznetsov, O. A., Bisikalo, D. V., Boyarchuk, A. A., Khruzina, T.S., & Cherepashchuk, A. M. 2001, Astronomy Reports, 45, 872Lubov, S. H. & Shu, F. H. 1976, ApJ, 207, L53Richards, M. T. 1992, ApJ, 387, 329 Richards, M. T. 1993, ApJS, 86, 255Richards, M. T. 2001, Astrotomography, Indirect ImagingMethods in Observational Astronomy, Edited by H. M. J.Boffin, D. Steeghs and J. Cuypers, Lecture Notes in Physics,573, 276Richards, M. T. 2004, Astron. Nachr., 325, 229Richards, M. T., & Albright, G. E. 1996, in Stellar SurfaceStructure, ed. K. Strassmeier and J. Linsky (Dordrecht:Kluwer), 493Richards, M. T. & Albright, G. E. 1999, ApJS, 123, 537Richards, M. T., Albright, G. E. & Bowles, L. M. 1995, ApJ, 438,L103Richards, M. T., & Ratliff, M. A. 1998, ApJ, 493, 326Richards, M. T., Waltman, E. B., Ghigo, F., & Richards, D. St.P. 2003, ApJS, 147, 337Sharova, O. I. 2006, Proc. of the IAU,
Planetary Nebulae in ourGalaxy and Beyond , IAU Symp. 234, 507Skilling, J. & Bryan, R. K. 1984, MNRAS, 211, 111Stern, R. A., Uchida, Y., Tsuneta, S., & Nagase, F. 1992, ApJ,400, 321Umana, G., Trigilio, C., & Catalano, S. 1998, A&A, 329, 1010Vesper, D. N., & Honeycutt, R. K. 1993, PASP, 105, 731White, N. E., & Marshall, F. E. 1983, ApJ, 268, L117
Richards, Sharova, & Agafonov
Fig. 1.—
Two-dimensional Doppler tomogram of RS Vul - 1993 (left) taken from Richards (2001), a Cartesian model of RS Vul (center),and the 2D tomogram of U CrB in the streamlike state (right). In the tomograms, the solid trajectory is the gravitational free-fall path ofthe gas stream; and the circles along this trajectory are marked at intervals of a tenth of the distance from the L point to the distance ofclosest approach to the mass gainer. The largest solid circle and the smaller dashed circle mark the inner and outer edge of a Kepleriandisk, respectively; the asterisk is the predicted location where the gas stream should strike the photosphere of the mass gainer; and theplus sign marks the center of mass of the binary. Fig. 2.—
Comparison between the original data (left frame) and the spectra computed from the reconstructed 3D Doppler map of RS Vul(middle frame) in terms of the radial velocity, V r versus orbital phase. The right frame displays the orbital phase variation of the relativechi-square statistic, χ /χ c , where χ c is the critical value corresponding to the 99% confidence level. The agreement between the observedand computed spectra is very good. hree-dimensional Doppler Images of RS Vul 7 Fig. 3.— ( V x , V y ) slices of the 3D Doppler tomogram of RS Vul, for velocities beyond the central velocity plane from V z = -540 km s − to V z = +540 km s − , in steps of 60 km s − . The images are shown for intensity levels: +0.01, 0.05, 0.1, 0.3, 0.5, 0.7, 0.9 (red) and -0.01,-0.05, -0.1, -0.3 (blue). The central source, corresponding to the mass gainer, is strongest at V z = - 160 km s − and very weak at Vz= 0km s − , while the gas stream emission is strongest from V z = -360 km s − to V z = -240 km s − . Richards, Sharova, & Agafonov
Fig. 4.—
Visualization of the seven most interesting ( V x , V y ) slices in the 3D Doppler Tomogram of RS Vul (1993) displayed symmetrically.The main features are: (1) circumprimary emission, (2) chromospheric emission from the donor star, (3) gas stream, (4) star-stream impactsite (the asterisk ∗ ), (5) the predicted locus of accretion disk, (6) localized region between stars, (7) other localized region, and (8) highintensity region with velocity close to that of the donor star. hree-dimensional Doppler Images of RS Vul 9 Fig. 5.—
Visualization of the seven most interesting ( V x , V y ) slices in the 3D Doppler Tomogram of U CrB (1994) in the streamlike statedisplayed symmetrically. The main features are: (1) circumprimary emission, (2) chromospheric emission from the donor star, (3) gasstream, (10) a high-velocity jet in the Vz direction (see Agafonov, Richards & Sharova (2006); Agafonov, Sharova & Richards (2009). Fig. 6.—
Cross sections of the RS Vul 3D Doppler tomogram in the ( V y , V z ) plane for V x = -380, -200, -120, -80, -20, 0, 20, 100, 120, 200km s − with intensity levels of 0.01; 0.05; 0.10; 0.30; 0.50; 0.70; 0.90. The numbered features correspond to those in Figure 3. The mapsfor V x values from +200 to +380 km s − contain no information. hree-dimensional Doppler Images of RS Vul 11 Fig. 7.—
The 3D tomograms of RS Vul and U CrB displayed in a 3D velocity cube (top row) together with the 2D integral images forthe ( V x , V y ) plane (second row), ( V x , V z ) plane (third row), and ( V y , V z ) plane (last row). The 2D integral image in the ( V x , V y ) plane isequivalent to the 2D Doppler tomogram. TABLE 1Characteristics and locations of prominent emission features in the 3D tomogram
Location: central velocity or velocity range (km s − )No. Emission Feature V x V y V z a
100 to 300 a -300 (-500 to -60)4 Star-stream impact region no source found5 Locus of the accretion disk very weak6 Localized Region (LR) – Part 1 (between stars) 0 0 to 100 180 (50 to 300), 450 (360 to 540)7 Localized Region – Part 2 100 -80 to 30 240 (150 to 350)8 Jet or flare near donor star 200 (100 to 300) ∼
200 360 (180 to 540) aa