Baltic Astronomy, vol. xx, yy 2011 The long-term spectroscopic misadventures of AG Dra with a nod toward V407 Cyg: Degenerates behaving badly 1 1 S. N. Shore,1 K. Genovali,1,2 G. M. Wahlgren3,4 0 2 1 Dipartimento di Fisica ”Enrico Fermi”, Universit`a di Pisa p largo B. Pontecorvo 3, Pisa 56127, Italia; [email protected] e 2 Dipartimento di Fisica, Universit`a di Roma Tor Vergata S ViadellaRicercaScientifica,1,I-00133Roma,Italia;[email protected] 5 3 Department of Physics, Catholic University of America 2 620MichiganAve.,N.E.,Washington,DC20064USA;[email protected] 4 NASA Goddard Space Flight Center, Code 667, Greenbelt, MD 20771 USA ] R S Received: . h p - Abstract. We present some results of an ongoing study of the long-term o spectroscopic variations of AG Dra, a prototypical eruptive symbiotic system. r Wediscusstheeffectsoftheenvironmentandorbitalmodulationinthissystem t s and some of the physical processes revealed by a comparison with the nova a outburst of the symbiotic-likerecurrentnova V407Cyg 2010. [ Key words: Symbiotic stars; stars-individual (AG Dra, V407 Cyg); stars: 1 novae;physical processes v 7 1. INTRODUCTION 9 3 The symbiotic binary AG Dra certainly needs no introduction to readers of 5 these proceedings and we refer to Leedja¨rv’s report in these proceedings for the . 9 relevant background and the excellent introductions in Leedj¨arv et al. (2004), 0 Viotti et al. (2007), G´alis et al. (2007), and Gonz´alez-Riestraet al. (2008). Pho- 1 tometric monitoring of this system extends now over nearly a century with the 1 highest precision spanning more than 40 years from both amateurs and profes- v: sionalastronomers(see Munari, theseproceedings,and Skopalet al. (2009)). We i concentrate, instead, on the line profiles and flux measurements. We have used X new and archival optical grating, grism, and echelle spectra from Asiago (Pennar r and Ekar), Loiano, the Telescopio Nazionale Galileo (TNG), Ondˇrejov, and the a Nordic Optical Telescope (NOT). Part ofthese data havealreadybeen published (Shore et al. 2010) and we refer the reader to that paper for details coveringthe 2006-2009 part of the observations. The full details of the data set are available inGenovali(2010). HereweoutlinesomeoftheresultsfromourstudyofAGDra thatbeganwiththemajoroutburstof2006-2008andhassinceexpandedtocover the last 30 years(Genovali et al. in preparation). 2 S. N. Shore et al. Fig.1. TheBalmerlinevariationsduringthe2006-2008outburstofAGDra,corrected for variations in the continuum flux as described in Shore et al. (2010). Square: Hα; cross, Hβ; asterisk: Hγ; diamond, Hδ. The outburst covered the orbital interval 5.p88- 7.p15withthepeakfortuitouslyoccurring nearsuperiorconjunctionofthegiant. 2. EMISSION LINE FLUX VARIATIONS Webeginwithtwosummaryplots. Figure1isthevariationoftheHαthrough Hδ lineequivalentwidths, correctedforthe continuumvariations,fromthemajor outburstof2006ofAGDra. Figure2showstherelativeHeI,HeII4686˚A,andO VI6825˚Avariabilityduringthe2006outburst. Wedrawthereader’sattentionto the bifurcation in the last panel that maydistinguish the hot and cool outbursts; thosefromthe startof the2006eruptionarethe upperbranchforHe I.As noted by Gonz´alez-Riestra et al. (2008) and Skopal et al. (2009), there is a saturation in the He II emission level during the major eruptions. This could be due to the increasedopticaldepthofthewindofthehotcomponent. Viottietal. (2007)have already conjectured that a burst may have two stages depending on its intensity: a hot event may, if it continues long enough, become a cool outburst. Passing to the long haul, we show in Fig. 3 the variations in the equivalent widths of the main emission features and the UBV photometry for about two decades.1 It is an impressively rich phenomenology (see also G´alis et al. 2007). Informed by the dense coverageof the 2006outburst, goingbackwardsin time to the major outburst of 1995, it appears that the pattern of Raman 6825˚A feature variationwasthesame. Asteeprisefromwhatwaslikelythedisappearanceofthe line. Arelativelyminoroutburst,aboutthreeyearslater,showsthesamepattern. Leedja¨rv reported at the meeting that the same thing appears to be happening again with the most recent (2010) minor outburst. Since the variations of this line are directly linked to the down-conversionof O VI 1032˚A photons, there are two possible, likely linked inferences: the column density in the wind around the white dwarf(WD) couldhavechangedand/ortheintensityof theresonancelines could have changed. The first implies a variability of the mass loss rate from the red giant (RG) or a change in the ambient density around the WD. The second, instead,issomethingintrinsictotheWD massloss,achangeintheopticaldepth oftheOVIlinesimplyingavariationinthewindfromthedegeneratecomponent. Two close peaks around day 1000 (Fig. 3) show the same Raman line variation 1 Thedisplayeddata arefromLeedj¨arvetal. (2004)(star),Tomova &Tomov(1999)(pen- tagon)andspectrafromLoiano(plus),Asiago/Ekar(asterisk),Catania(diamond),TNG(cross), Asiago/Pennar(filledcircle). Inthelowerpanel: U(filledcircle),B(plus),andV(circle)from SkopalandcollaboratorsandtheAAVSO. AG Dra spectroscopic history 3 100.0 40 W 10.0 30 He I/He II E 1.0 He II 4686 20 0.1 10 3800 4000 4200 4400 4600 4800 5000 5200 2.0 2.5 3.0 3.5 4.0 4.5 5.0 JD (2450000+) He I 6678 40 5 4 30 He II 4686 20 He I 6678 23 1 10 0 0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14 O VI Raman 6825 O VI Raman 6825 Fig.2. Variations of the He I, He II, and O VI Raman 6825˚A line during the 2006- 2008eruptionfromtheLoianoandOndˇrejovspectra. Topleft: HeI6678˚A(square). He I 7065˚A (cross), and He II 4686˚A (diamond) variations in time. Top right, He I 6678˚A vs.He II 4686˚A. Bottom left, He II 4686˚A vs. O VI Raman 6825˚A; bottom right: He I 6678˚A vs. OVIRaman6825˚A. Note,inthelastplot,thebifurcation. eventuallyreachingthesamemaximum. Ifthesewereso-calledhot outbursts,the OVIlinewouldhaveturnedlessopticallythickthaninthecooloutbursts,andthe featurewouldnothavedisappeared. If,instead,theWDmasslosswashigher,and the covering factor complete, then the increase in the WD wind could have been so large that the O VI 1032, 1036˚A was smuthered with a consequent vanishing of the optical features (cf. Schmid (1995) and Harries & Howarth (1997) for a discussion). The 2006 outburst confirmed what had, implicitly, already been reported in the literature, that the Raman lines can disappear (Munari et al. 2009). Having been seen on only one occasion, however, and lacking dense coverage it was not obvious that this is a generic feature of a certain magnitude of event. Instead, with 20-20 hindsight, it seems that the 6825˚A feature flux curve observed during the last major outburst is actually typical, the slow recovery of the line relative to its sharp initial drop and its subsequent return to the quiescent state being recognizable in earlier events. The interval of disappearance was brief, about a week, and due to weather, season, and scheduling, other similar events might have been (barely) missed. The rise from minimum line strength is similar for all such large U outbursts. An important feature of the long term variations is the appearance of a sort of saturation level for the emission line strengths, in particular He II 4686˚A as a function of the U magnitude. From Leedj¨arev et al. 4 S. N. Shore et al. Fig.3. ThelongtermspectroscopichistoryofAGDra. Equivalentwidthvariationsof (toptobottom)Hα,Hβ,OVIRaman6825˚A,andHeII4686˚Aemissionlines,corrected forcontinuumvariations, togetherwiththeUBVphotometry. AG Dra spectroscopic history 5 (2004), it was already known that the emission varies in phase with the near UV but the relation was a simple power law up to U≈ 9m. At brighter levels, the strict increase breaks down with a large dispersion, greater for He I 6678˚A than for He II 4686˚A. The O VI Raman 6825˚A feature actually peaks at U ≈10 and then declines by about a factor of two for increasing brightness. The maximum equivalentwidthreachedbytheRaman6825˚Alineis-15˚Athatis,however,rarely reachedin the historical data. The optical He II lines, formed by recombination in the circum-WD region, are useful, but not unique, proxy measures of the state of the hot source. The recombination time is density dependent and can be of the same order as the interval between activity phases. It should be recalled that the environment is a mess given the many dynamical and radiative processes acting simultaneously. These lines convolve the ionization source variations with the response function oftheenvironment,anunknowngiventheuncertaintiesregardingthemechanism producing the ultravioletand X-rayvariations. The Raman features are,instead, instantaneousmeasuresoftheFUVsincetheyareopticallythinandproducedonly byscattering. Usingthis,itisinterestingthatonerecoversthesameclassification ofhotandcooloutburstsobtainedfromthedirectmeasurementoftheX-rayswith ROSAT and XMM analyzed by Gonz´alez-Riestra et al. (2008) and Skopal et al. (2009). Those events during which the Raman line fell below the quiescent floor and the U magnitude increased by the largest amount were judged cool, while the the two events that saw the line increase from the floor level and return to it after a short time were hot. The other intriguing feature is that during the hot outbursts the Raman line reachesa systematically greaterflux. The H+ region of the WD is always ionization bounded, we are not dealing herewithaninterstellarcloud,butitisdenseandnotwithoutintrinsiclineopac- ity that can affect the spectrum formation at other wavelengths. For example, changes in the Lyman series optical depths, that can be caused by variations in the FUV luminosity of the hot source or a change in the absorbing column den- sity,canonlybeextractedbydetailedmodelingofthelineformation. Decreasing the density increases the ionized volume and decreases the Lyman series optical depth, changing the optical Balmer line intensities and profiles. Changing the mass loss rate from the hot star displaces the location of the stagnation point of thewind-windshockandalsomaychangethespectrumofhardphotonsavailable to further ionize the environment. This requires self-consistent, time dependent photoionization and hydrodynamic models of the sort described by Folini et al. (1998,2003). Like the Raman events, there seems to be a characteristic time development of the U outbursts but these also indicate different natures. A change in the covering factor and/or the column density without an accompanying change in the luminosity of the hot source will produce only flux redistribution and the bolometric luminosity should remain constant. This is well known from the post- explosion stage of classical novae and is not far from correct for LBVs. In such cases,anincreaseinthemasslossratefromthehotstarcouldhavethesameeffect. Instead,assuggestedbySokoloskietal. (2006),iftheoutburstsareactuallymini- TNR events (see Starrfield, these proceedings), then the bolometric luminosity should vary. At least the major event of 2006appears to be the latter case. This is a further wayto separatethe sourcesof the variations. 6 S. N. Shore et al. Fig.4. The equivalentwidth variations of the O VI Raman 6825˚A, and He I 6678˚A emission lines in AG Dra, corrected for continuum variations, phased on the orbital solution byFekeletal. (2000). Seetextfordetails. 3. ORBITAL MODULATION OF STRONG EMISSION LINES Line profile variationsare direct probes of the environmentaleffects alongthe lineofsight. ToillustratethecomplexityoftheAGDraphenomenology,weshow somesamplephasedvariationsoftheHαandheliumlines. Figure4showstheHe I 6678˚Aand O VI 6825˚Ascaled flux variationsplotted on the Fekel et al. orbital solution. No distinction is made in this plot between phases of activity from JD 49030 to JD 54639 (Ondˇrejov and Loiano spectra only). For the Raman feature, the minimum line strength (< 0.3˚A) from the 2006 outburst happens to have occurred at orbital phase zero. This plot also suggests the presence of multiple periodicities in the line variations. While the Raman line shows little orbital modulation, there is a clear indication of such in the He I variations. Further data cuts, based on the activity state, show that the line fluxes from pre-selected quiescentstagesarethosewiththestrongestorbitalmodulation,ascanbegleaned from Fig. 3. The Raman features are immune to self-absorption, that alters the permitted Balmer and helium profiles, and Rayleigh scattering that affects the UV lines. Electron scattering should not be important since it is local to the source, not to the conversion region. Instead, they should be sensitive to changes in the line of sight column densities of neutral hydrogen.2 The modulation of the ratio of the two optical O VI Raman features may, consequently, be indicative of structural changes in the wind resulting from the hydrodynamic interaction of the winds from the two components (see Fig. 4 in Shore et al. 2010). The minimum in the lineratiohappenedintheintervalbetweenthetwoUmaximaoftheoutburstand is, perhaps coincidentally, approximately at inferior conjunction of the WD. We urgeobserverstoinclude,wheneverpossible,bothfeaturesinfuturespectroscopic studies. In Figs. 5a,b and 6a,b we show sample high resolution line profiles at a few phases to illustrate the environmental effects. The He I line profiles show a nebular structure, the He II 4686, 5412˚A lines show, instead, a comparatively single peaked profile that remains at almost con- stant radialvelocity,about -200km s−1, but whose symmetrychanges. The vari- ations of the profile of the 1640˚A line, studied by Gonza´lez-Riestra et al. (2008), 2 Asimilareffect wasdiscussed forthe OI]1641˚A lineinsymbiotics byShore &Wahlgren (2010)asaprobeoftheneutralwindoftheRG.IntheabsenceofavailableUVmeasurements, the[OI]6300˚Aline,formedfromthesametransitioncascade,maybeanappropriatesubstitute. AG Dra spectroscopic history 7 Fig.5a. Sample profiles for Hα in AG Dra from the TNG spectra at several orbital phases. Thedates areJD 51826, 53151, 53589, and54731. The phasesare according to theephemerisofFekeletal., (2000) withφ=0p isRGinferior conjunction. Fig.5b. The Balmer Hγ andHδ line profiles from the NOTspectrum on JD53100, atorbitalphase0.p23. are more extreme in displaying extended wings with maximum velocities of 2000 km s−1 that do not appear to show orbital modulation. 8 S. N. Shore et al. Fig.6a. ExamplesoftheHeI5875˚A line profilesinAGDraas afunction oforbital phase from Catania, TNG, and Asiago spectra. The dispersions were 0.02˚A px−1 and 0.15˚A px−1. ThetimeperiodisfromJD51826 toJD55076 coveringeightorbits. Fig.6b. ExamplesofthestructureontheHeI6678˚A singlet line profiles; as inFig. 6a, fromTNGandAsiago. Figure 7 shows a simple model of the line profile variations for Hα as a func- tion of orbital phase (see Dumm et al. (1998) for an application to BX Mon). AG Dra spectroscopic history 9 Fig.7. Modellineprofilevariationswithorbitalphase. AsyntheticHαline,assumed tobeagaussian,wasabsorbedinthewindoftheRGassumingaterminalvelocityof30 kms−1 andasimpleβ-law(seetext). Themass ratio andorbitalperiod approximately matchBXMon. Differentialdisplacement(center-of-massframe)producesthevariability oftheabsorptionfeature. Theemission lineintensitywas keptconstant. The emission line from the ionized region around the hot star, assumed to be a gaussian for simplicity, is observed with orbital radial velocity shifts through the extended neutral wind of the RG that has a β-type velocity law. In this case, a velocityamplitudeoforder10kms−1fortheRGandamassratioof3:1(RG:WD) were assumed to simulate BX Mon (see Dumm et al. (1998)). No quantitative information is intended here, just a warning that even in the visual part of the spectrum, environmental line absorption effects that are well known to affect ul- travioletlineprofilestudiescanproducesignificantchangesintheprofiles. Itthen suffices to recall that in the realsystems, the interactionsof the winds of the two components, complex structures within the orbital plane, non-spherical outflows, andtimedependentvariations,combinetocomplicatethequantitativeanalysesof such variations. In AG Dra, for instance, the He II 1640˚A emission line intensity variations and profiles are clearly affected by both intrinsic changes and orbital modulation bythe UV Fe-curtainand detailed modelsareneeded toseparatethe variouscontributions. 4. LESSONS LEARNED FROM THE 2010NOVA ERUPTION OF V407 CYG Oneofthemanythingslearnedfromthe2010eruptionofV407Cyg,asymbi- oticthatshowedallthesymptomsinthiseventofasymbiotic-likerecurrentofthe RS Oph variety (Munari et al. 2011, Shore et al. 2011a,b), is that the presence of broad wings on the emission lines, especially intrinsically asymmetric profiles, can also have hydrodynamic origins aside from steady outflows. The interactions inthe environmentsofthesesystemsarecomplexandverytime dependent, espe- ciallyinthelowerdensitypartsofthecircumstellarmedium,asshownbytwoand three dimensional simulations. We show some examples of the late-time Balmer lineprofilesthatallowadeterminationoftheindividualregionsandtheirreaction to the expanding shock and photoionization changes caused by the post-eruption emission from the WD. The lines are asymmetric, a characteristic feature of an off-centershockpropagatinginthedensitygradientoftheRGwind. Suchprofiles have, intriguingly, been observed on He II 1640˚A in AG Dra and other symbi- 10 S. N. Shore et al. Fig.8. The Balmer line profile in the symbiotic-like recurrent nova V407 Cyg on 2010 Jul. 15 (top) and 2011 Aug. 21 (bottom) (date of outburst 2010 Mar. 10) from theNOT(0.02˚Apx−1). TheprofilesareHβ (dot-dash),Hγ (dash),andHδ (solid). The maximum shock velocity was >3000 km s−1. The stellar velocity is -54 km s−1, note thewindabsorptionatanexpansionvelocityofabout10kms−1 thatbecomesoptically thinastheoutburstprogresses. Thebroadwingsareduetoashockfromtheejectaand subsequentoutflowintheMira (seediscussion andShoreetal. 2011a,b). otics in outburst. The subsequent passage of the shock and the onset of wind recombinationrenderstheprofilesmoresymmetricwithtime,muchlinetheNOT observationof Hγ and Hδ in AG Dra (Fig. 5b). ACKNOWLEDGMENTS. We thank T. Iijima, C. Rossi, and R. Viotti for providing some of the spectra used in this study. SNS thanks the organizers for their kind invitation and patience, and D. Folini, J. Jos´e, P. Koubsky, L. Leedja¨rv,J.Mikolajewska,K.Mukai,C.Rossi,A.Siviero,A.Skopal,J.Sokoloski, S. Starrfield, R. Viotti, and R. Walder for stimulating discussions. Extensive use hasbeenmadeduringthisprojectoftheSimbad,ADS,andMASTdatabases. We also thank the AAVSO for providing archival photometry. GMW acknowledges support from NASA grant NNG06GJ29G. REFERENCES Dumm, T., Muerset, U., Nussbaumer, H. et al. 1998,A&A, 336,637 Fekel, F. C., Hinkle, K. H., Joyce, R. R., & Skrutskie, M. F. 2000,AJ, 120,3255 Folini, D. & Walder, R. 1998,AGAb, 14, 108 Folini, D., Walder, R., Psarros,M. & Desboeufs, A. 2003,ASPC, 288,433 G´alis, R., Hric, L., Leedj¨arv, L., & Sˇuhada, R 2007,IAUS, 240,121 Genovali, K. 2010,MSc Thesis - Physics, Univ. of Pisa (URL: http://etd.adm.unipi.it/theses/available/etd-05052010-002805/) Gonzlez-Riestra,R., Viotti, R., Iijima, T. & Greiner, J. 1999,A&A, 347,478 Gonz´alez-Riestra,R.; Viotti, R. F.; Iijima, T. . et al. 2008,A&A, 481,725 Harries,T. J. & Howarth,I. D. 1997,A&AS, 121, 14