Discovery of the First Eclipsing Binary Barium Star

'~uro~eanSouthern Observatory; 21nstitut dHstrophysique, Unlversit4 de Li&ge, Belgium; vrije Universiteit hssel, Betglum

gramme operating at €SO since 1982 I.Barium Stars ly have cireul~orbii (e-g., Webbink, 1986), which is not the case for all (Stmian. 1983). Barium stars are a family of peculiar barium stars. Boffin and Jorlssen (1988) The Str6mgren uvby magnitudes of a red giant stars whose envelopes exhibh suggested instead that ttse accretion by sample of 19 barium stars have been Werabundances of carbon as well as of the mum star of tfie wind from the monitored since July 1984 in a differ- dements heavier than iron, flrst iden- former AGB primary may be efficient ential way: fw each programme star, tifled by Biddman and Keenan (1951), enwgh to account far the obsewed two comparison stars were selected they represent about 1% of all red dwmlcal peculiarities. among nearby G or K giants. The obser- ghnts of types G and K. Their kinematic vation frequency is about meek during behaviwr ap- to be similar to that typical one-rnonth observing runs. That 2 Interest of a Photomettic frequency was increased to one mea- of A or early F main-sequeom siars; of acmdlngly, barium stars shwtd have Modbring Barium Stars surementlnlght during three observing an average mass of about 1.5 to 2 MQ Slnce barium stars are single-llned campaigns around the predicted time of (Hakkila 1989a). smscopic blnarie (SBI), mly par- eclipse of the companion of HD 48407 McCture (1983) found that a large tial Information cm be obtained abut by the md gM(Section 4). fraction of barlum stars belong to spec- their orbit parametem The detection Obmm&ns were performed an the tmoopic binaries. Jorissen and Mayor of possible photometric variations re- ESO 50-cm telescope or on the Danish (1988) showed that virtually all known lated to the binary nature of these stars 50-cm telescope (four-channel simut- southem barium stars with large barium could provide furthw insight in the taneous photometer). The reduction to overabundance are spectroscopic characteristics of these systems. How- the standard uvby system is performed binash, and McClure and Woodswwth, ever, the relatiJely Large orbital separa- with the "multi-night" algorithm d+ (1990) obtainad orbital periods In the tion (estimated as sevef8l AU) and the sctibed by Manfroid (1985), taking into range 80 d to more than 10 y. Whiie low lumlnos'ty of the companion (dic- account the total set of measurments dwarf (WD) companions are inferred tated by the SB1 nature of the system) carried out at the ESO 50-cm w Danish fmthe mass function distribution, al- Imply that photomettlc variations, if 50-cm telescopes since the beginning though very few of them can be de- present, should b of rather small am- of the monitoring. bbehnreen ob- tected directly from their UV radiation plitude and have a long period. serving runs are avoided in that way, (Btihrn-Vltense, 1980, Dominy and Lam- LandoR (1983) was the first to look for since thsre is lust one colour matrix for bed, 1983, BBhrn-Vitense. Nmnee and photometric vdability of Mumstars. the whole set of measurements in a pm,1984). He obtained UBV photoetdc obsw- given system. Checks were of course The fact that all barium stars belong vatlons of several barium stars at KPNO made to control the stablllty of the sys- to binary systems clearly rneplns that and CTlO at Irregular internal3 over a tem over the whob period cons id^. binarity must mehow be responsible time span of more than a decde. His The accuracy of the differential mag- for their chemical peculiarities. It was photometry was baslcalfy non-differ- nitudes can be estimated from the stan- sometimes suggested that the presence ential, but included the measurements dard deviations of the dWmces be- of a companion could perhaps modify of a numbw of standard stars of similar tween comparison stars. For both the the outcome of the hellurn flash occur- brightness as the barium stars of his Danish 50 and ESO 50 systems (referred ring in tow-mass stars, possibly trigger- programme. Assuming the mnmr to In the following as D50 and W-6, ing the extra-mixing leading to the of obsmdons in V for standard stars to respecttvely, according to the notation synthesis of heavy elements (e.g., be about 0.007 magnitude, he wn- of Jobsen, Manfroid and Sterken, McClure* 1984). However, soma barium cluded that 6 stars of the sample (which 1991; see also Manfroid et al., 1991), systems are quite wide (P > 10 y), and It containd 17 Mum showed v&- 50% of the consldsred comparison is difficult to imagine that such a distant ations at or above the 3 o level. All stars pairs have a standard deviation In the y companion could have an effect on the were however obsened less than a channel smatler than 0.003 mag for ob- internal structure of the Mumstar. dozen times under very different ob servations spannlng several years. The Mass transfer from the former primary mIng conditions and no lightcum is mean value of these r.m.s. deviations Is towards the barium star, when the pti- provided. Our more accurate photomet- somewhat smaller for the 1350 system mary was a bvy-etement rlch S or ric mwtitorlng, described in Section 3, (0.007 mag) than for the E50-6 system carbon star on the asymptotic giant does not confirm these variations In all (0.009 mag). Colour Indices in the E5M branch (AGB), seems more likely. The but one case. system are however much less accurate observation of a 10 pm (N band) excess than In the D50 system, mainly because in many barium stars (Hakkila, 1989b), The the latter are obtained with a fwr- which is not correlated with the other 3. Long-Term Photometry of channel photometer. In what follows, we atmospheric peculiarities, may polnt to- Variables Programme at ESO fill therefore malnly make use of the wards the presence tn the system of High accuracy and homogeneity over more accurate D50 measurements. dust left over from a former mass-loss periods of several years are required for More details about the obsenrhg policy, episode. Again, many barium systems monitoring barium stam. Both require the redudon method and the resulting appear to be too wide for mass transfer ments, together w'h the potential of accuracy can be found in Manfroid et al. through Roche lobe overflow (RLOF) to extenslve observing time (4 to 6 months/ (1991). occur (Tout and Eggleton, 1988)b year), are offered by the Long-Term A deteriled discussion of the results of Moreover, post-RLOF systems general- Photombtry of Variables (LTW) pro- the photometric monitoring for the sam- P-(A+B)/2 A-B McCIum and Woodsworth, 1990). HS -0.01^_1I1'LI'1~I11LI1L1I'11IC-11111I1L1I11111111111I~spectroscopic ephemeris was used in e order to predict the times of a posslbb 4 eclipse of the companion by the barium star. During the ?-year span of the - -- 4:. .', - monitoring, only thrw ecll p~eswere 0 \ .I*. pmdicted to occur at a time when the I =' 4 star was wily ohwabk HD 46407 was obsmed once a night fw 20 to 40 - 4 - -- - nights duhg these periods (FebmarY 0.01 1985, Noamher 1488 and February 1990). A clear dlp is seen in the lightcunre In -0.0 1 November 1968 while the comparison pair remains stable (Fig. 1). Since this dip is exactly centred on the predicted

I time for the total acllpse, there ts littler * a -1 m. A * t doubt that an eclipse has %ctudlybeen , * AY 0'. : 0 - 1. 8 I: +* . detected. Very accurate measurements - *v7* were carried out ddng the FebnmY I a. v #mom 8 1990 campaign at the Danish 50-cm 8 telescope, and meal that, I any, the 0.01 - A lo* -- - eclipse was much shallower. The mea- surements of February 1985 are puz- zllg, since a Mstidly significant M trend is ind& obsed, but If It were 1t1rrI1iii1111 11 1111--I 1111 I I -0.01 TTT--Ctt to twrsspond to the eclipse Ingress, there wwld be a > 0.03 phase lag a m wml respect to the spzhamplc I ephemeris. - 4 *t -- Figure 2 presents the phase diagram 0 - I = a, .'s rn 0 7 for all D50 measurements, adopting P = 458.1 d and T = JD 2 445 296.0 (time of *, maximum velocity), whbh cormpond - t~ the lower bounds of McClure and 0.01 - -- WoodsWs elements. The measure mmi of February 1985 are lagging in that phase diagram as well, so that the current uncwtalnty on ttw circular orbi- 0.7 0.72 0.74 0.76 11.78 0.8 0.7 0.72 0.74 0,760.18 0.8 tal elements (f0.5 d on P and k1 d on T) annot resolve the discremcy. The phase superposition of the February 1985 eclipse on the Novembw 1988 Figml: R~ofm~~ngwlmparsn~mndthetlmedthepredWedl~ingress d&t would require a 451.7 d period, not (Q - UX): hbmy 1% (ypper pmI; cycle I), lyov~rnbr1m (middle pamid; cycle 4, FmyTssO low^; c* 5). Phases am omputlad adqothrg ma c&bI &emen& of vwy dMmnt from McClure'e spwtros- McGIure and Wmdwmh (I-). Tha martainty on Me prt3dicted tlme of iota! adlp is coplc period, although well outside its dispiayedbythehQMmtaImbaramndthe valwpl-0.75 5.~ft~~tthe formal mr bar. ditY~8I-+Bp mgdtu& in the SlrSmgren y channel, whereas the right paddlspfay The phase diagram .of Flgure 2, h61~~1lA-Bymag~,mwhere,PstanderWMeMumemandAandBfarMnuo adopting McClure and Woodsworth's ~slars~p~tivdyHRP373andHR236~~Themprnhtofthe~tuds~ (1990) s~lcperiod, reveals C the mean Wwe of the mmapondlng dfmmntrsl mgnftude In a glven system (P * MO that a wty broad secondary eclipse may A S~SWI:r = Em~pt~m; - Em8 stem; 7 - ES050-8 m; - E5U-x system). be present as well. A trend was in dwaxiaiswlenWinsucnawaymta~h~~stoadpinW clear Em. fact &sewed for 0.35 c cp c 0.5 in cycle 2, and the drop in brightness around phase 0.35 was conf~rmedby pie of 19 barium stars is presehd in between the cormpanding comparison the observations of cycle 6. Again, the Jarissen, ManfroId and Wrken (1991), stars) we specifically related to the bi- large scatter in that phase range indl- The main result of our rnonltoring Is that nary nature of tbetar, or whether HD cat= that the spcmmopb period is the two Mum stars with the shortsst 121447, which is also the coolest known probabjy not the best ~Mcefor con- orbital perlods in the sample, HD badurn star (K7 Ill, LU et al., 1983), is structing a photometric phase diwram. f 21447 (- 185 dl JoPissen and Mayor, in simply a mIcrwariable as any other very A careful pert& analysis of our data pepmation) and HD 46407 (458.6 d; red star. remains to be done. McClure and Woodsworth, 1890), are In summary, the Itghrcunre of HD the only ones to present small, albelt 46407 derived from the currently avail- 4. HD 48407: The Rrst Eclipsing in significant, light variations, In the casa of Binary Barium Star able photomettfc observations the HD 121447, it Is not yet clear, however, 050 system displays a sharp "primaryb whether these variations at a Iwel o, HD 46407 (HR 2392, y - 6.27, b - y a eclipse (companion &hind the barlum 0.020 mag (to be compared with ay - 0.60, KO111 8a3) has the second shortest star) and a pdbb shallow "secon- 0.009 mag for the magnitude dMemnce otbital period in our sample (P = 458.6 d, dmyn eclipse. In November 1988, the -0.02 ,n.,n.#,v.n,.-.,.m, ~~~ll~~l~.~l~"l"'tMvefy: the depth of the eclipse will yield an &mate of the mount of dust qulred and the wavelength depen- P-(A+B)/2 A-B dence of the swtterlng cr~ss-sectIrn should be oompatlble with the relattve I --8 - variations of b - y ewld v - b. The follow- -0.01 -* a, ing questions also remain to be en- is such an explanation cornpat- -..4". #Is- with the absence of IR excess for a I * r HD 46407 (Hakklla and McNmara 1987, HWla 1989b)? And how can dust grains remain trapped around the companion for long periods of the (Poynblw-R- -I? Ofl 8 'b t The pfiatwnetric monitoring of HD 2 a a 4W7is still going on; potarlmetric mea- 0.01 - -- - surements are also In progress. Finally, 0 a similar photometric mbnitotim is be-

I ing &ed out fw fhb two barium stars, + with the &omst known ohW persods (HD 77247 and HD 12144?), and it will be Interesting to know whether a be- 0.2 0.4 0.0 0.8 0 0.2 O,4 0.0 0.a haviwr similar to that of HD 46407 will phase pbaae be observed. b2:#wedlagram forthe wtmk setofy m~infheD51lsWm NRpensl: In conclusb, the behaviour of HE) ~~~&tRneto~~c~tpanel:mparlsQn~,aWW7~sPee~- 48407 Is a clear example of a phenome- ~~liovnIWeC~andWooc~swwth(f~~hevarlws5~refartofhenon that can wly be studied as part of a d~~ydeS(slcd&fff#n~~~r145~,fl:~*cflle1,r =cyele2,e=~b4,*~Gyda s,D-~8.Wffie~oiawwide~ed~Wd*~iflcarefully planned long-term project rely- ing obsenrfng: If) ftpnt of the barium slarb Note that, bemuse P-,WEWinvolves the ~WPof the Wo on mdce continurns smwme88MWl#h P-A tvld P-8, CR~W+&&k I/Z tlmeS 8dler than photometric monitoring should be theontronA-B, carried out In order to get the general light cum; (3) eaeh eclipse naeds mom Primary eclipse had a depth of - 0.02 since dust ought not to be distributed In than one month of dally obswing; (iii) mag In y and a full wldth of ~rp- 0.05 ;a spherically symrnet* way mund the they only occur every 7.3 years and one 6. e. &out 20 dl; the primly eclipse was companion. third of them am missed because of the much shmllwer in February 19%). The ~lthough this mdseems to proximily to the Sun. In dditron, we shdtow secondary ec~ipseoaurs when account qcratitativdy for bhe must also emphasiz~hat only a small the companion is in frmt of the barium beviour, It remains to be tested Wan- tdwmpe is needed. The LTPV project *. b depth ts at least 0.01 mag in the Y band and It mdsover about 60% of the orbital period. It mlgM well be that the brig- of HD 4W07 b slowly WFng over the whole OWperiod. lha b - y index is mqinally variable, at variance with the v - b index which exhibfts large vlvlatlons. Quite interest- ingly, the variations 05 the v - b index and of the y magnitude appear to be roughly comM& when ths star is falnter, R is alm redder, as shown by Figure 3. This mhaviour L tvpicaI d light-- p-, SO that we suggest that the ecllpmd light is actuaily tho light from the barium star ttself which is backscamred by dust trapped fin a disk?) andthe companion. Rust must be present in a rather ex- tended region around tlw companion, since the November 1988 eclipse was about thm$ times longer than ft would have been expectid for the eclipse of a point-like swroe by a red giant of mdlus 15 Rg (and a mimajoraxis of 1.5 AU; total duratlon of 7 d or Arp = 0.015). Mwewer, then does not smto be a flat battom in the diplightcum, in- dicating that the eclw is never total. The phase tag of February 1- can dm be acoounfed for by mls Wanatlon, at ESO clearly represents the only avail- M., de Groot, M., DuW,H.. Duemmler, McC~U~,R.D., WQO&WOT~~,AW. 1980, able opportunity to address that kind of R., Figer, A, Hageman, T., Henaberg, H., Ap.J*, 932, m. problem. Jorissen, A., Madejsky, R., MandeI, H., Ott, Starken, C. 1983, The Mesengw, 33,IO. H., ReItermann, A, Schulte-Ladbeck, R., Tout, C.A.. Eggleton, P.P. 1988, MNM Staht, O., Steenman, H., vander Unden D.. 291,823. irckgral, FA. 1991, Astm, Astrophyw Webblnk, R.F. 1886, In: WalObwvatlons 8up@., 87,48 1 , vsPhysicalM&forCl~~131~Sys- Bldelrnan, W.P., Keenan, P.C. 1951, &J., McClure, R.D. 1983, ApJ., 268,264. tsms, eds. KC. Leuw and D.S. Zhal, Mew t14,473. McChrre, R.O. 1984, PASP,#, 11 7. Yo&, Gordon and Breach. bffin, H.M.J., Jodwm,A. 1988,m.A~- mRYs,, m 155. mm-w,E 19m, Ap.J., 2aa, L79. BBnm-Vltense, E., Nemc, J., Proffltt, Ch. 1984, ApJ., 270,726. Domlny, J.F.,LamM,D.L.lW,Ap.J.,270, The Rise of SN 198749 180. Hakkila, J. 198Oa, A.J., 98,699. It is now almost five years since SN intensky and the angular tarte have Hakkifa, J. 1989b, mn.Astmphysys, 219 1987A explded in the LMC. In the been rescaled to allow direct corn@- 204. meantime, the visual brightness has de son. A rwnt, careful remeasurement d Hakkita, J., MeNamara, 6.J. f 987, Astron. ma& to about one millionth of what it the magnitudes of SM 1987A on these -, 10B, 255. was at the time of maximum. If is still plates has shown that earlier publlskd Jotissen, A., Mayor, M. 1988, Asm. Ash- being observed with large telescopes, estimates are too faint by 0.2-0.8 mag- phm, 1!W, 187. also af ta Silla, but the elusive pulsar nitude (McNaugM and West, to appear JoriSWI, A, ManWd, J., Sterken, C. IWl, has not yet bean directly detected. in Astronomy and Astrophysics). The &irona Asiqhys., In p-. Larwiok, AU. 1983, PAW, H, 644. The four plctures were taken during new values are in better agreement with U,P.K.. Darn,D.W., Upgren, AR., Wels, ambur patrols in Australia at the time the theoretical Ilghtcuwes, but they do E,W. 1983, &J., 62, la. of the explaslon by Robert H. McNaugM not by themselves permit to decide Manhold, J. 1985, habiT~t, theds, Univ. of (Plates 2.3 and 7)and frank 6. Zoltows- which of the two neutrino events that we. ky (Plate 5). They show the early rise in were observed In the morning of Febm- hamfroid, J., Swken, C., Bruch, A.. Bwger, brightness of this famous object; the ary 23, 1987, w~ the actual time-zero.