B fields in OB stars (BOB): The magnetic triple stellar system HD 164492C in the Trifid nebula

4.3 MB·

by  J. F. Gonzalez

MNRAS000,1–15(2016) Preprint16January2017 CompiledusingMNRASLATEXstylefilev3.0
B fields in OB stars (BOB): The magnetic triple stellar system
HD164492C in the Trifid nebula.⋆
J. F. González1†, S. Hubrig2, N. Przybilla3, T. Carroll2, M.-F. Nieva3, I. Ilyin2,
S. Järvinen2, T. Morel4, M. Schöller5, N. Castro6,7, R. Barbá8, A. de Koter9,10,
7 F. R. N. Schneider11, A. Kholtygin12, K. Butler13, M. E. Veramendi14, N. Langer7,
1
0 and the BOB collaboration
2
1InstitutodeCienciasAstronómicas,delaTierraydelEspacio,CasilladeCorreo49,5400SanJuan,Argentina

n 2Leibniz-InstitutfürAstrophysik,AnderSternwarte16,14482Potsdam,Germany
a 3InstitutfürAstro-undTeilchenphysik,UniversitätInnsbruck,Technikerstr.25/8,6020Innsbruck,Austria
J 4Spacesciences,TechnologiesandAstrophysicsResearch(STAR)Institute,UniversitédeLiège,QuartierAgora,Alléedu6Août19c,Bât.B5C,B4000-Liège,Belgium
2 5EuropeanSouthernObservatory,Karl-Schwarzschild-Str.2,85748Garching,Germany
1 6DepartmentofAstronomy,UniversityofMichigan,1085S.UniversityAvenue,AnnArbor,MI48109-1107,USA
7Argelander-InstitutfürAstronomie,UniversitätBonn,AufdemHügel71,53121Bonn,Germany
] 8DepartamentodeFísicayAstronomía,UniversidaddeLaSerena,Av.Cisternas1200Norte,LaSerena,Chile
R 9AntonPannekoekInstituteforAstronomy,UniversityofAmsterdam,SciencePark904,POBox94249,1090GE,Amsterdam,TheNetherlands
S 10InstituutvoorSterrenkunde,KULeuven,Celestijnenlaan200D,3001,Leuven,Belgium
. 11DepartmentofPhysics,DenysWilkinsonBuilding,KebleRoad,Oxford,OX13RH,UnitedKingdom
h 12ChairofAstronomy,AstronomicalInstitute,St.PetersburgStateUniversity,Universitetskipr.28,198504,St.Petersburg,Russia
p 13Universitäts-SternwarteMünchen,Scheinerstr.1,81679München,Germany
- 14ComplejoAstronómicoElLeoncito,Casilladecorreo467,5400SanJuan,Argentina
o
r
t
s
a AcceptedXXX.ReceivedYYY;inoriginalformZZZ
[

1
ABSTRACT
v
HD164492Cisaspectroscopictriplestellarsystemthathasbeenrecentlydetectedtopossess
4
1 astrongmagneticfield.Wehaveobtainedhigh-resolutionspectroscopicandspectropolarimet-
5 ric observationsovera timespanoftwo yearsandderivedphysical,chemical,andmagnetic
3 properties for this object. The system is formed by an eccentric close spectroscopic binary
0 (Ca1-Ca2)withaperiodof12.5days,andamassivetertiaryCb.Wecalculatedtheorbitalpa-
. rametersoftheclosepair,reconstructedthespectraofthethreecomponents,anddetermined
1
atmosphericparametersandchemicalabundancesbyspectralsynthesis.Fromspectropolari-
0
7 metricobservations,multi-epochmeasurementsofthelongitudinalmagneticfieldswereob-
1 tained.Themagneticfieldisstronglyvariableontimescalesofafewdays,withamostprob-
: ableperiodintherangeof1.4–1.6days.StarCbwithT ∼25000Kistheapparentlyfastest
v eff
rotator and the most massive star of this triple system and has anomalous chemical abun-
i
X danceswithamarkedoverabundanceofhelium,0.35±0.04bynumber.Weidentifiedthisstar
r asbeingresponsiblefortheobservedmagneticfield,althoughthepresenceofmagneticfields
a in the componentsofthe Ca pair cannotbe excluded.Star Ca1 with a temperatureof about
24000Kpresentsanormalchemicalpattern,whiletheleastmassivestarCa2isamid-Btype
star(T ∼15000K)withanoverabundanceofsilicon.Theobtainedstellarparametersofthe
eff
systemcomponentssuggestadistanceof1.5kpcandanageof10–15Myr.
Keywords: stars:early-type–binaries:spectroscopic–stars:magneticfields–stars:abun-
dances–stars:fundamentalparameters–stars:individual:HD164492C
1 INTRODUCTION
The multiple star HD164492 (=ADS 10991) is a trapezium-like
⋆ Based on observations obtained in the framework of the ESO
stellarsystembelongingtotheTrifidNebula,anactivestarforming
Prgs. 191.D-0255(B,F,H,I,J), 093.D-0267(B), 091.C-0713(A), 088.D-
0064(A),and070.D-0191(A). region.ThenebulaisionisedbytheO7.5Vz(Sotaetal.2014)star
† E-mail:[email protected] HD164492A,whichisthecentralobjectofthismultiple.Thecen-
(cid:13)c 2016TheAuthors
2 J. F.Gonzálezet al.
tralpartofthesystemisformedbyfivebright(V =7.4−12.2mag) nebularemissionlines.Emissionfeaturesappearwithdifferentin-
visual components in a non-hierarchical configuration (A to E; tensity in spectra taken with the fibre-fed spectrograph HARPS,
Kohouteketal.1999).Therelevanceofstudyingthismultiplehas while they are absent in observations taken with the slit spectro-
increased following the recent discovery of a magnetic field of graph UVES. The fact that these emissions are removed through
about500Ginoneofitscomponents,thesubsystemHD164492C backgroundsubtractionduringthereductionprocessofUVESdata,
(Hubrigetal.2014),by theBOB(“BfieldsinOBstars”) collab- indicatesthattheoriginoftheseemissionsisnotcircumstellarbut
oration(Moreletal.2014,2015).Inaddition,Hubrigetal.(2014) nebular,sinceanycircumstellarstructurewouldhavebeenincluded
reportedthatHD164492Cisanearly-Btypedouble-linedspectro- intheobjectextractionaperture.Weobtainedanemissionlinemas-
scopicbinary,andpossiblyatriple.Noquantitativeanalysesofthe terspectrumbyaveragingbackground-aperture spectraof several
componentsofHD164492Cisavailableintheliteraturesofar. UVES observations. This emission line spectrum was scaled to
Themaingoalofthepresentworkistodescribethephysical mimictheintensityofemissionlinesineachHARPSspectrumand
andchemicalpropertiesofthisinterestingsystemascompletelyas thensubtracted.
possible, in particular its magnetic field, in order tocontribute to During this process we identified and measured all relevant
the understanding of the conditions under which magnetic fields nebular lines. The line identification was done using the lists by
aredeveloped inmassivestars. Onlyasmall fractionofstars(5– Stasin´ska (2009). The average radial velocity of nebular lines is
7%)withradiativeenvelopespossessstronglarge-scaleorganised −4.8±1.8kms−1.Mostemissionlinesdonotoverlapcrucialstel-
magneticfields(Grunhut&Wade2013;Moreletal.2015),which larlines,butthedetectionandremovalofsmallHeiemissionswas
can be generated during the star formation process (fossil fields, particularlyimportanttoavoidthedistortionofstellarHeilinepro-
e.g.Borraetal.1982;Moss2001),ordynamoactiontakingplace files.
intherotatingstellarcores,ortheycouldbeproductsofamerger
process.Whilethefirsttwoscenariosareunabletoexplainanum-
berofobservationalphenomena(e.g.Ferrarioetal.2015),themag-
3 SEPARATIONOFSPECTRALCOMPONENTS
neticfieldscouldformwhentwoproto-stellarobjectsmergelateon
theirapproachtothemainsequenceandwhenatleastoneofthem LineprofilesofHD164492Cshowclearvariationsontime-scales
hasalreadyacquiredaradiativeenvelope(Ferrarioetal.2009).The of a few days. The majority of the line profiles presents a com-
originofmagneticfieldsinmassivestarsisstillamajorunresolved plex structure suggesting the existence of more than two spec-
probleminastrophysics. troscopic components. We propose as the most simple interpre-
Sections 2 to4 describe theobservations, the reconstruction tation, the presence of a fixed broad-line spectrum superimposed
ofthespectraofthestellarcomponents,andtheorbitalanalysisof withadouble-lined spectroscopic binary.Theresultsofthespec-
theclosebinarysubsystem.Thephysicalparametersandchemical tral disentangling performed under this hypothesis and the mod-
abundancesarederivedinSect.5and6,whilethemagneticfieldis ellingofcompositespectrawithspectralsynthesisconfirmedthat
discussedinSect.8.ThemainresultsaresummarisedinSect.9. thissystemisindeedahierarchicaltriple.Thepresenceofathird
source is also supported by the fact that HD164492C is a visual
binary with a separation of 0.08′′(González&Veramendi 2016;
Yusef-Zadehetal.2005).Thisvisualpaircanbeconfidentlycon-
2 SPECTROSCOPICANDSPECTROPOLARIMETRIC
sideredasphysicallybound,sincetheprobabilityofchancealign-
OBSERVATIONS
mentfortwostarsofmagnitudeV=9.0–9.5mag(J=8.6–9.2mag)
Wehaveobtainedhigh-resolutionspectroscopicandspectropolari- isverylow.Fromthetwo-dimensional stardensity,derivedusing
metric observations over a timespan of two years with the ESO the 2MASS Catalogue, we estimated that the probability of hav-
spectrographs UVES, FEROS, and HARPS. The FEROS echelle ing a companion brighter than J=10 mag at an angular distance
spectrograph on the 2.2-m telescope at La Silla produces spec- smallerthan0.08′′isabout1:1.3×106.Followingtheconventional
tra with a resolving power of 48000 and a wavelength coverage nomenclature of multiplestars, wenamed thecompanions of the
from3500to9200Å.OurUVESobservationscovertwospectral closebinaryasHD164492Ca1andHD164492Ca2,andthemore
regions:thebluespectrumbetween3300and4530Åwitharesolv- distantstarwiththebroad-linespectrumasHD164492Cb.
ingpowerofR = 80000andtheredspectrumcoveringtherange We separated the three spectral components applying the
from 5700 to 9450Å with R = 110000. The polarimetric spec- methodofGonzález&Levato(2006),andassumingthatthethird
traof HARPShave aresolvingpower of R = 115000 and cover star (Cb) has constant radial velocity. Considering the different
thespectralrangefrom3780to6910Å,withagapbetween5259 wavelength coverage of the HARPS and UVES spectra, we per-
and5337Å.ThefourFEROSspectraweretakeninAugust2013, formedthecalculationsseparatelyinthreespectralregions:3700–
whilethe15UVESspectraweretakenbetweenAprilandAugust 4530Å (UVES+HARPS), 3800–5250Å (HARPS), and 5680–
2014.Thetypicalsignal-to-noiseratioatλ=4200Åis280forthe 6800Å (UVES+HARPS). From the composite spectra we made
UVESspectraand110fortheFEROSspectra.WithHARPSweac- afirstestimationofthespectraltypes(B1forcomponentsCa1and
quired11spectropolarimetricobservations in5observing runsin Cb,andB4-B6forCa2)andaccordingtothisweselectedspectraof
June2013,April2014,March2015,June2015,andOctober2015. referencestarsasastartingpointfortheiterativedisentanglingpro-
The signal-to-noise ratio in the Stokes I spectra is about 300 per cess. We used high-resolution archival spectra of slowly-rotating
pixelabove4600Å,decreasinginthebluetoabout150at4000Å. starsHR2222(B1V)forCa1andCbandHR1288(B4V)forCa2.
Thebasicdatareductionandcalibrationwasperformedusing ThesespectrahavebeentakenwiththeUVESandFEROSspectro-
specific ESO reduction pipelines. UVES spectra were re-reduced graphsaspartofotherobservingprogrammes.Thereferencespec-
usingstandardtasksoftheNOAO/IRAFpackage. trawerescaledandconvolvedwiththeoreticalrotationalprofilesto
Intheredspectralregion,thespectrarequiredsomeprocess- mimicthelinesofthecomponentsofHD164492C.
ing toremove featuresnot belonging tothe stars: telluricand in- Torecover theintrinsicintensitiesof thespectra oftheindi-
terstellarabsorptionlines,and,inthecaseoftheHARPSspectra, vidual components, it is necessary to scale the separated spectra
MNRAS000,1–15(2016)
ThemagnetictriplestellarsystemHD164492C. 3
Figure1.SelectedregionsintheUVESspectrashowingtheseparationofthespectralcomponentsatthreedifferentphases.Thetwoupperpanelscorrespond
toUVESspectrawithorbitalphases0.14,0.35,and0.81(frombottomtotop)andthelowerpaneltoHARPSspectrawithphases0.02,0.33,and0.56.For
eachphaseweshowintheupperpartthereconstructedspectraofthecomponentsCb(green),Ca1(red),andCa2(blue),andbelowtheobservedspectrum
(smallblackdots)overplottedwiththesumofthethreereconstructedspectra(redline).Inthelowerpanelmodelledtelluriclinesareshowninlightblue.The
nebular[Oiii]emissionlineisvisibleat5007Å.
accordingtothefluxcontributionofeachstar.Tothisaim,wede- Thenoiseleveloftheseparatedspectraisabout1/1300ofthe
rivedfromthespectraltypesafirstestimateofT andlogg.Then original continuum at λ 4200Å in the first spectral region, about
eff
we estimated the flux contribution of each component as a func- 1/650 at5000Å,and1/800 intheredregion. Thefinalsignal-to-
tionofwavelengthusingsyntheticspectrafromthegridcalculated noiseofthereconstructedspectradependsalsoontherelativeflux
byLanz&Hubeny(2007)andthestellarradiiestimatedfromthe contribution of each star, which are approximately 0.55–0.60 for
Genevastellarmodels(Ekströmetal.2012)assumingthatallthree Cb,0.35–0.40forCa1,and0.05–0.07forCa2(seeSect.5).Thus,
starshavethesameageandrequiringthemassesofstarsCa1and after scaling, theseparated spectrahave asignal-to-noise ratioof
Ca2 tobe consistent withthe spectroscopic mass ratio. In a later theorderof500forthemostmassivestarsandofonly50forthe
stage, the relative flux of the components was recalculated using starCa2.
syntheticspectrawiththeappropriateatmosphericparametersand As illustration, Figure 1 shows for three different orbital
chemicalabundances(seeSect.5). phaseshowthereconstructedspectracombinetoreproducetheob-
MNRAS000,1–15(2016)
4 J. F.Gonzálezet al.
Table1.RadialvelocitiesofthespectroscopicbinaryHD164492Ca.Column1:HeliocentricJuliandate;Col-
umn2:orbitalphasecalculatedwithrespecttotheperiastronpassageTπ(seeTable2);Column3and5:radial
velocitiesforstarsCa1andCa2;Columns4and6:residualsobserved-minus-calculated; Column7:Spectro-
graph.
HJD φ RVCa1 (O-C)1 RVCa2 (O-C)2 Instrument
kms−1 kms−1 kms−1 kms−1
2456445.7946 0.5559 −28.4±1.0 1.7 44.1±0.6 0.0 HARPS
2456446.7934 0.6355 −36.6±0.9 2.0 62.3±0.7 −0.9 HARPS
2456522.6844 0.6898 −42.7±1.4 1.2 75.0±1.9 0.0 FEROS
2456523.7039 0.7711 −49.9±1.2 0.4 88.9±1.9 −0.7 FEROS
2456524.6224 0.8444 −52.9±1.6 −0.5 98.8±2.6 4.5 FEROS
2456525.6624 0.9273 −32.3±1.4 2.6 59.1±2.0 4.3 FEROS
2456770.7700 0.4808 −22.5±1.0 −1.2 24.0±0.9 −0.2 HARPS
2456771.7850 0.5618 −28.3±0.9 2.5 46.3±0.9 0.7 HARPS
2456772.7707 0.6404 −39.4±0.8 −0.2 63.1±0.8 −1.2 UVES
2456777.8085 0.0423 73.9±0.9 −1.2 −192.7±0.7 0.2 UVES
2456779.8951 0.2088 25.4±1.0 −0.3 −83.5±0.9 −2.0 UVES
2456786.8817 0.7661 −51.2±0.8 −1.2 88.2±0.5 −0.6 UVES
2456837.6044 0.8125 −53.4±0.8 −1.2 94.3±0.5 0.3 UVES
2456869.5077 0.3576 −4.2±0.9 −0.1 −12.7±0.5 1.9 UVES
2456879.6971 0.1705 36.1±0.9 −0.2 −105.2±0.7 0.4 UVES
2456883.4874 0.4729 −22.5±0.9 −2.2 20.7±0.4 −1.2 UVES
2456887.4936 0.7925 −52.4±0.9 −0.9 93.3±0.5 1.1 UVES
2456894.6261 0.3614 −3.8±0.9 0.9 −12.3±0.6 0.9 UVES
2456898.5681 0.6759 −43.8±0.9 −1.2 71.8±0.7 −0.3 UVES
2456919.5607 0.3506 −3.1±1.0 −0.2 −17.0±0.5 0.1 UVES
2456923.5111 0.6658 −41.7±0.9 −0.1 67.9±1.0 −2.0 UVES
2456926.5385 0.9073 −42.8±0.8 0.7 72.2±0.7 −2.0 UVES
2456929.4996 0.1435 43.8±0.8 −1.2 −126.4±0.9 −1.4 UVES
2457091.8426 0.0944 59.1±1.3 −3.6 −163.8±1.0 1.3 HARPS
2457092.8282 0.1730 35.9±0.9 0.4 −103.4±0.6 0.5 HARPS
2457093.8478 0.2544 17.2±0.7 2.3 −57.5±0.6 −0.1 HARPS
2457094.8153 0.3316 3.2±0.7 3.0 −24.0±0.7 0.3 HARPS
2457178.6948 0.0231 69.5±1.4 0.6 −180.9±1.1 −2.0 HARPS
2457179.7426 0.1067 60.2±1.4 2.0 −156.5±0.9 −1.8 HARPS
2457317.5185 0.0977 64.6±2.0 3.1 −163.9±2.0 −1.7 HARPS
servedprofiles.ForthestarCa2,linesofSiii(λλ4128,4130,5041, tra, which were not considered in the reconstruction of compo-
5056),Feii(λλ4233,5018),Cii(λ4267),andHei(λλ4121,4144, nent spectra because of their lower resolution, were included in
5015,5047)canbedistinguished.Forthemostmassivestars,be- the radial velocity measurements. We used as templates the ob-
sidesthementionedlinesofCiiandHei,themostnoticeablelines servedspectraofreferencestarsmentionedintheprevioussection:
in the plot correspond to Aliii (λ 4150), Feiii (λ 4165), and Oii HR2222 for Ca1 and HR1288 for Ca2. Before carrying out the
(λλ 4133, 4153, 4169, 4185, 4190). A complete spectral atlas of cross-correlations, radial velocities of template spectra were cor-
thecomponentspectrawiththelineidentificationisincludedinthe rectedbymeasuringseveraltensofmetallicspectrallines.Table1
Appendix(availableonline).Weconcludethatthethreementioned liststhemeasuredradialvelocities.Orbitalphasezerocorresponds
spectral components represent satisfactorilytheobserved spectral totheperiastronpassage.
contentandvariations. Duringthespectraldisentanglingonlyonefixedvalueforthe
ThereconstructedspectrumofCa1,theprimarycompanionof radialvelocityofstarCbwasfitted.Afterwards,tolookforpossi-
theclosebinary,showsamorphologytypicalofaB1Vstar,with ble variations in the spectrum of this star, we removed the spec-
lines of Hei, Cii, Nii, Oii, Aliii, Siiii-iv, Sii-iii, and Feiii. The tra of Ca1 and Ca2 from the observed spectra using the recon-
spectrumofCbseemstobeverysimilartoCa1,butwithbroader structedspectra of these components, and measured the resulting
metallinesbecauseofahigherrotationalvelocity.Remarkableare spectra,whichwouldbeindividualspectraofstarCb.Thevelocity
theunusuallystrongHeilines,indicatingaHe-richnatureforthis of thisstar isconstant withinuncertainties, withamean value of
star.ThefaintstarCa2isamid-Btypestar:onlyafewstronglines −6.3kms−1andastandarddeviationof2.8kms−1,whichislower
ofHei,Cii,Mgii,Siii,SiiandFeiicanbeidentifiedinitsspectrum than thetypical measurement errors. Themean radial velocity of
(seeAppendix). CbagreeswiththebarycentricvelocityofthebinaryCaandthatof
theTrifidnebula(−4.8±1.8kms−1,seeSect.2).
AKeplerianorbitwasfittedtotheobservedradialvelocities
byleastsquares.Figure2showstheradialvelocitycurvesandTa-
4 ORBITALELEMENTS
ble2liststhefittedorbitalparameters:periodP,timeofperiastron
Radial velocitiesoftheclose binaryCaweremeasured bycross- passage T ,radial velocityamplitudes K and K ,centre-of-mass
π 1 2
correlations during the spectral separation process. FEROS spec- velocity γ, argument of periastron ω, and eccentricity e. We list
MNRAS000,1–15(2016)
ThemagnetictriplestellarsystemHD164492C. 5
Table 3. Model atoms for non-LTE calculations. Updated models as
describedbyNieva&Przybilla(2012)aremarkedwith*.
Ion Modelatom
Hi Przybilla&Butler(2004)
Hei/ii Przybilla(2005)
Cii/iii Nieva&Przybilla(2006,2008)
Nii Przybilla&Butler(2001)*
Oi/ii Przybillaetal.(2000),Becker&Butler(1988)*
Nei Morel&Butler(2008)*
Mgii Przybillaetal.(2001)
Aliii Przybilla(inprep.)
Siii-iv Przybilla&Butler(inprep.)
Sii/iii Vranckenetal.(1996)*
Feii/iii Becker(1998),Moreletal.(2006)*
requires the physical model of the system to match the observed
compositeanddisentangledspectraandtoprovideconsistentages,
Figure2.RadialvelocitycurvesofHD164492Ca.Thebottompanelshows spectroscopic distances, themassratiofor Ca2/Ca1(fromtheor-
residuals(observed-minus-calculated).Filled(open)symbolscorrespondto
bitalsolution)andfluxscalingfactorsforthecomponentssimulta-
the primary (secondary) star, circles indicate measurements with UVES,
neously.Observationally,thisiscomplicatedbythefactthatlow-
squareswithHARPS,andtriangleswithFEROS.
frequency modulations are not recovered by disentangling tech-
niquesbasedonrelativeDopplershifts.Forthatreasonbroadfea-
tures like hydrogen lines and broad He lines, whose profiles are
Table2.OrbitalparametersofthebinaryHD164492Ca.
significantly wider than the shifts due to the orbital motion, can-
not be fitted individually for each component. Moreover, relative
Parameter Units Value
fluxes are in principle unknown. Unless they are adopted from a
P d 12.5349±0.0008 differentinformationsource–ase.g.indetachedeclipsingbinaries
TI d 2456692.94±0.03 –,theyareadditionalparameters,howevermutuallydependenton
Tπ d 2456689.54±0.03 stellareffectivetemperaturesandradii.Thesefactsmakethespec-
K1 kms−1 63.8±0.8 tralanalysisofthevariousstellarcomponentstobeinterdependent.
K2 kms−1 143.8±0.7 We used a hybrid non-LTE approach for the quantitative
ω rad 5.29±0.01
e 0.532±0.004 analysis of the composite and separated spectra, in analogy to
q 0.444±0.006 the methods described by Nieva&Przybilla (2007, 2012) and
a1sini R⊙ 13.4±0.2 Przybillaetal. (2016) for the study of single (He-strong) B-type
a2sini R⊙ 30.2±0.2 stars, extending them here where necessary for the multiple star
asini R⊙ 43.5±0.2 case.Inbrief,thisapproachisbasedonplane-parallel,hydrostatic,
M1sin3i M⊙ 4.89±0.06 chemically homogeneous and line-blanketed model atmospheres
M2sin3i M⊙ 2.17±0.04 that were computed with the code Atlas9 (Kurucz 1993) under
γ kms−1 −7.3±0.3
theassumption ofradiativeandlocalthermodynamic equilibrium
n 30
(LTE).Fortheparameter rangestudiedhere, theseareequivalent
σ1 kms−1 1.65 tohydrodynamicorhydrostaticline-blanketednon-LTEmodelat-
σ2 kms−1 1.61
mosphereswithintheline-formationregions(Nieva&Simón-Díaz
2011; Nieva&Przybilla 2007; Przybillaetal. 2011). Non-LTE
alsothefollowingderivedparameters:timeofprimaryconjunction level populations and synthetic spectra were calculated with re-
T,projectedorbitalsemiaxisasini,andminimummassesMsin3i. cent versions of the codes Detail and Surface (Giddings 1981;
I
Thelastthreerowsarethenumberofobservationsandtheglobal Butler&Giddings1985,bothupdatedbyoneofus(KB)),employ-
RMSoftheresidualsfortheprimary(σ )andthesecondary(σ ). ingcomprehensivemodelatomsassummarisedinTable3.
1 2
Infact,standarddeviations differ forthethreeinstruments,being Grid-based procedures for the analysis of composite spectra
about1.1kms−1forUVES,1.7kms−1forHARPS,and2.5kms−1 (e.g.Irrgangetal.2014)couldnotbeappliedwithoutprohibitively
forFEROS.Thesevaluesaresomewhatlargerthantheformalmea- large extensions of the precomputed grids because of the chem-
surement errors, and are probably a better estimate of the actual ically peculiar nature of the star Cb (see below). Therefore, the
velocityuncertainties. line-profilefittingwasperformedonthebasisofdedicatedmicro-
gridcomputationscentredaroundasetofatmosphericparameters
for each star, which were refined iteratively. The aim was to de-
rive those parameter combinations for effective temperature T ,
5 SPECTRALANALYSIS eff
(logarithmic)surfacegravitylogg,microturbulenceξ,projectedro-
Thedeterminationofphysicalandchemicalpropertiesofstellarat- tational velocity vsini, (radial-tangential) macroturbulence ζ, he-
mospheresthroughspectralanalysisandmodellingismorecompli- lium abundance y (by number), and individual metal abundances
catedinthecaseofspectroscopicmultiplesystems,becausemore thatfacilitateanoverallgoodfittobeachievedfori)thedisentan-
parameters need to be accounted for. A comprehensive solution gledspectra(weakmetallinesonly)andii)theobservedcomposite
MNRAS000,1–15(2016)
6 J. F.Gonzálezet al.
Table4.Atmosphericandfundamentalstellarparametersandchemicalabundances(logX/H+12)forthethreecomponentsofHD164492C.Uncertainties
are∼0.2dexforabundances.Standardvalues(CAS)weretakenfromNieva&Przybilla(2012),exceptthosemarkedwithasterisks,whicharepreliminary
valuesfromPrzybillaetal.(2013).UncertaintiesfortheCASvaluesarestandarddeviations.
Parameter HD164492Cb HD164492Ca1 HD164492Ca2 CAS
SpectralType B1V/IVHe-strong B1V B4-6V
Teff(K) 25000±500 24000±500 15000±1000
logg(cgs) 3.90±0.15 4.00±0.20 4.30a
ξ(kms−1) 2±1 2±1 2±1
vsini(kms−1) 135±5 48±5 19±2
ζ(kms−1) ... 25±5 12±2
y(numberfraction) 0.35±0.04 0.089±0.010 0.089±0.010 0.089±0.002
Cii/iii 8.13 8.50 ... 8.33±0.04
Nii 7.78 7.93 ... 7.79±0.04
Oi/ii 9.16 8.91 ... 8.76±0.05
Nei 7.59 8.15 ... 8.09±0.05
Mgii 7.36 7.62 ... 7.56±0.05
Aliii 6.23 6.49 ... *6.30±0.07
Siii-iv 7.81 7.59 ... 7.50±0.05
Sii 7.15 7.21 ... *7.14±0.06
Feiii 7.52 7.61 ... 7.52±0.03
M/M⊙ 11.5±1.1 10.0±1.0 4.2±0.5
R/R⊙ 6.5±1.5 5.1±1.4 2.2a
logL/L⊙ 4.1±0.2 3.9±0.2 2.4a
MV(mag) −3.2±0.5 −2.6±0.6 +0.1a
τ(Myr) 13.5+1.6 14.0+2.2 consistent
−2.1 −8.0
τ/τMS 0.7±0.1 0.6±0.1 <0.1
scalefactor 0.59±0.05 0.37±0.03 0.04a
aAdoptedfromstellarmodelsaccordingtotheageofthemassivecompanions.
spectra(broadHandHelines,metallines).Thederivedparameters imposedbythetheoreticalisochrones.Fromthispreliminaryanal-
shouldalsofacilitatethestarstoshareacommonisochroneanda ysis,weconfirmedthatCbisaHe-richstar.
common spectroscopic distance, and lead to consistent flux scal- A final check within an iteration step was performed on the
ingfactors.Manyoftheparametersareinterrelatedintheanalysis observedcompositespectrausingtherefinedatmosphericparame-
methodology,hencetheneedforaniterativeapproach. ters,chemicalabundances,andscalingfactors.Syntheticcompos-
itespectraforparticularorbitalphaseswerebuiltbycombiningthe
Inaninitialstep,weestimatedstartingvaluesforprojectedro-
threesyntheticspectraofthecomponentstarsaftershiftingbyra-
tationalvelocitiesbyapplyingtotheseparatedspectraofthethree
dialvelocity,broadeningwithappropriaterotationalandmacrotur-
componentsthetechniqueintroducedbyDíazetal.(2011).Initial
bulentprofiles,andscalingaccordingtotheirrelativefluxes.Atthis
valuesforatmosphericparameterswerebasedonspectraltypesand
stagewewereabletoincludeHiandbroadHeilinesintheanaly-
consideredtwoadditional externalconstraints:i)theevolutionary
sis.Thisyieldedfurtherestimatesforcorrectionsofparameters,for
age must be the samefor allthree components, and ii) themass-
whichdedicated syntheticspectrawereagaincomputed. Wethen
ratio M(Ca2)/M(Ca1)mustbeconsistentwiththeqvaluederived
returnedtotheinvestigationoftheseparatedspectraasdiscussed
from the spectroscopic orbit. When all three stars are considered
inthepreviousparagraph,etc.,requiringseveraliterationstepsun-
tobelongtothesameisochrone,atmosphericparametersandrela-
tilnofurther need forchanges intheparameterswasfound from
tivefluxesarenolongerindependent. Forfixedtemperatures,one
thespectroscopicanalysis,meetingalsotheadditionalconstraints
singleparameter(e.g.age)determinesloggandrelativefluxesfor
onconsistentages,spectroscopicdistances,theCa2/Ca1massratio
allthreestars.Inthesecalculationsweusedthestellarmodelgrids
andthefluxscalingfactors.Theprocedure turnedouttobechal-
byEkströmetal.(2012).Thisstrategyallowedustoadoptstarting
lengingandtimeconsumingduetothelargenumberofvariables,
gravityvalueswithouthavingindividualhydrogenlineprofiles.
furthercomplicatedbecauseofthechemicalpeculiaritiesofCb.
Inasecond step, bothT and loggwererefinedon theba- Thefinallyadopted atmospheric and fundamental stellar pa-
eff
sisofionisationequilibria,i.e.allionisationstagesofanobserved rameters,andtheabundancesforthecomponentsofHD164492C
element have to indicate the same abundance (within the uncer- aresummarisedinTable4.Forcomparison,chemicalabundances
tainties),employing narrow Hei/ii lines,Cii/iii, Oi/ii, Siii-iv, and asderivedfromasampleofchemicallynormalandsingleearlyB-
Sii/iii. Individual metal lines were fitted using spectral synthesis. type starsin thesolar neighbourhood (Nieva&Przybilla2012) –
Analysingthedisentangledspectrarequiredadouble-iterativepro- the‘cosmicabundancestandard’(CAS)–,arealsogiven.
cedure, one on the atmospheric parameters and chemical abun- Figure3showsaselectionofspectrallinesusedintheabun-
dancesandanotherinthescalingfactorsofthedisentangledspec- dancedetermination.Anoverallgoodtoexcellentmatchbetween
tra.Newscalingfactorswerederivedfromstellarparameterscal- the model and the disentangled spectra is obtained. Even so, for
culatedduringtheanalysisoftheseparatedspectraviametalion- some lines the residuals are significantly above the noise level,
isation equilibria. In this instance we relaxed the strict constraint which is low in the reconstructed spectra (typically 1/400–1/500
MNRAS000,1–15(2016)
ThemagnetictriplestellarsystemHD164492C. 7
Figure3.Synthetic(red)vs.separatedspectra(black)ofcomponentsCa1(upper)andCb(lowersetofcurves).
of the continuum level). These discrepancies might be a sign of Ca1hasachemicalcompositioncompatiblewiththeCAS,though
surface chemical spots, although it is also possible that they are bytrendaslightlymoremetal-richcompositionisindicated.
remains of the spectral separation process. Withthevelocity am-
plitudeofstarCa1beingwithinthewidthofthelineprofileofstar
Cb,theresultinglineintensitiesinthereconstructedspectraofCb
andCa1arestronglycorrelated.
Despitethehighsignal-to-noiseratiooftheseparatedspectra
ofthecomponentsCa1andCb,theuncertaintiesinchemicalabun-
Thefaintcomponent Ca2wasassessedonlythroughthedis-
dancesareestimatedtobeabout0.2dex.Themainerrorsourceis
entangledspectrumsinceitscontributiontobroadHandHelines
the scale factor introduced by the relative fluxes and in the case
iscomparativelysmallmakingthefittingofcompositespectrain-
ofCbtheuncertaintyinthecontinuumdefinitionintheregionof
sensitivetotheparametersofthisstar.However,itsnarrowmetal
broad lines or blends. The fitting of Balmer lines was consistent
lines are well reconstructed in the spectral disentangling, though
withatmosphericparametersderivedfrommetallinesthroughout
theoverallsignal-to-noiseratioislowandtheuncertaintyinitsrel-
theorbitalperiod,seeFig.4.
ativefluxislarge,whichaffectsystematicallythelineintensities.
The star Cb is a He-rich star (y=0.35±0.04, i.e. about Moreover, without good neutral or doubly ionised metal lines, it
+0.6dexabovetheCASvalue)withsomemetalpeculiarities,par- wasnotpossibletoderivethesurfacegravityforthisstar.Instead,
ticularly moderate overabundance of O and Si, and underabun- we adopted a value consistent with the age of the more massive
dances of C, Ne and Mg. Abundance anomalies are frequently stars.Asaconsequence,werefrainfromprovidingabundanceval-
found in He-strong stars as a consequence of the different cou- uesforCa2inTable4.Ontheotherhand,foranyreasonablescale
plingofindividualionstothestellarwindandduetotheformation factor the spectrum appears peculiar, the star being apparently a
ofchemicalinhomogeneities(spots)becauseofthemagneticfield Si-richBptypewithitseffectivetemperatureranging fromabout
(seee.g.Przybillaetal.2016),andCbisnotanexception.Thestar 14000to16000K.
MNRAS000,1–15(2016)
8 J. F.Gonzálezet al.
Figure4.ProfilesofseveralHilinesatdifferentorbitalphases.Comparisonofcompositesyntheticspectra(redline)withUVESspectra(smallblackdots).
The spectroscopic logg value of stars Cb and Ca1 suggests that
theyaresomewhatevolvedwithinthemain-sequence,withanage
of∼12-15MyraccordingtoGenevamodelsforrotatingstarswith
metallicityZ = 0.014.Usingthesemassesandtheradialvelocity
curve,weestimatetheorbitalinclinationoftheCa1-Ca2systemto
bei=52±3◦.
For the adopted isochrone, assuming star Ca2 is a normal
main-sequence star of the same age, the relative flux of star Ca2
would be about 0.040±0.013. The observed intensity of spectral
lines of this star in the composite spectrum, however, suggests a
scalingfactor 0.061±0.012.Inotherwords, theintensityofCa2
lines in the spectrum suggests that stars Ca1 and Cb are some-
whatlessluminousthanfound inthespectral analysis. Iftherel-
ativecontributionoffluxforstarCa1isoftheorderof0.37,then
the intensity of the Ca2 lines suggests that the flux ratio Ca2-to-
Ca1 is 0.16±0.03. The strip between blue lines in Fig. 5 marks
the positions of star Ca1 that are consistent with a temperature
15000±1000 K for Ca2 and a flux ratio 0.16±0.03. In short, the
spectrumof Ca2, under theassumption of being anormal star of
Figure5.logTeff−loggdiagram.IsochronescorrespondtoGenevastellar thesameage,suggeststhesystemtobesomewhatyounger,atmost
modelsforrotatingstarswithZ=0.014.Dottedlinesare(fromlefttoright)
10Myr.
isochronesforlogage6.9,7.0,7.1,7.2,and7.3.Continuousblacklinesare
(from right toleft) stellar tracks for4,5,7,9, 12,and 15M⊙.Thestrip At periastron the components of the Ca pair are at a sepa-
markedwiththe twobluedashedlines corresponds tothestellar models ration of about 26R , only 3-4 times larger that the sum of their
⊙
ofstarCa1forwhichtheflux-ratioCa2-to-Ca1is0.16±0.03,asssuminga radii.Thisisnotcloseenough,however,fortheorbitandtherota-
temperature15000±1000KforCa2. tionvelocityofthecompanionstohavesufferedsignificantchanges
by tidal effects. Calculations with the binary evolution code by
Hurleyetal. (2002) indicate that no significant evolution of the
6 PHYSICALPARAMETERS
orbit and rotational velocities would occur until the primary ap-
We fitted the observed spectra assuming that they are formed by proaches the end of the main-sequence at about 25Myr. With an
threestarsofthesameage.Figure5showsthepositionofthethree eccentricityofe=0.532thepseudo-synchronousrotationalperiod
companionsinthelogT −loggdiagram. wouldbe3.994days.Itisinterestingthattheprojectedrotationcor-
eff
TheadoptedparametersarelistedinthethirdblockofTable4. respondingtothepseudo-synchronous regime,ascalculatedfrom
MNRAS000,1–15(2016)
ThemagnetictriplestellarsystemHD164492C. 9
theadoptedstellarparameters,51±9kms−1,isinagreementwith
theobservedvsinivalue.Inotherwords,theprimaryCa1seemsto
berotatingclosetothepseudo-synchronous regime,althoughthis
wouldnotbecausedbytidalforces.
Thestellarparametersdeterminedforthecomponentsofthis
multiplesystemoffertheopportunitytocalculatethespectroscopic
distancemodulus.Eventhoughvariousworkshavedeterminedthe
distancetotheTrifidNebulaclusterusingotherstarsorproperties
of the nebula, this issue is still controversial. The high and vari-
ableextinction,andmainlytheanomalousextinctionlawhascon-
tributedtothis.Lyndsetal.(1985)reportedanextinctionlawwith
R = 5.1 and estimated a distance of 1.67kpc. Cambrésyetal.
V
(2011) confirmed this high R value, but obtain a significantly
V
largerdistance.Fromthecolourdistributionofstarsalongtheline-
of-sighttostronglyabsorbedregionswithintheTrifidnebula,they
derivedtheextinctiondistributionalongtheline-of-sightandcal-
culated adistance of 2.7±0.5kpc to thenebula. A similarvalue
hadbeenobtainedfromUBVphotometryofstarsHD164492A,B,
C,andE,byKohouteketal.(1999).
Unfortunately none of the stellar components of the mul-
tiple system HD164492 is included in the Gaia first data re- Figure6.Profilevariations ofHiandHeilines.Upperpanels:compari-
sonofUVESspectraatHJD2456869.5077(red)andHJD2456919.5607
lease (GaiaCollaboration,etal. 2016). Parallaxesare availablein
(blue).Lowerpanels:comparisonofHARPSspectraatHJD2457091.8426
this catalogue for a few other probable members of the cluster
(red)andHJD2457317.5185(blue).Inthelowerpartofeachpanel,the
NGC6514buttheuncertaintiesaretoolarge(∼50%)forareliable
differencebetweenthetwoplottedspectraisshown.
determinationofthedistance.
The adopted stellar parameters for the three stars studied in
thepresent paper correspond toatotalabsolute visual magnitude
of−3.72±0.53mag.TheapparentV magnitudeofthistriplesys-
tem is not well known. The photometric measurements given by
Kohouteketal. (1999) involve also the component HD164492D. tral variations. One possible explanation is a temperature varia-
Using the light ratio estimated by the same authors (3.9 at λ = tionover thesurface of thisstar. Inthetemperature range of star
514nm), we obtain: V = 8.91mag for the integrated magnitude Cb,adropintemperaturewouldstrengthenbothHeandHlines.
of the system HD164492C. Assuming a reddening E(B−V) = In chemical peculiar stars, the regions with a great concentra-
0.33±0.03mag(Kohouteketal.1999,forstarsA,B,andC),and tionofsomeparticularelements(chemicalspots)presentahigher
R=5.1,wederiveadistanceof1.55+0.45kpc,supportingtheresults opacity,whichincreasesthetemperatureduetobackwarming(see
−0.35
ofLyndsetal.(1985). Krticˇkaetal.2007foraclearexampleinaHe-strongstar).Onthe
other hand, inmassivemagnetic stars,temperaturevariations can
berelatedtobrightspotsthatariseduetothelowerdensity(lower
optical thickness) of the gas in surface regions with higher mag-
7 SPECTRALVARIABILITY
neticpressure(Cantiello&Braithwaite2011).
Even though the combination of the reconstructed spectra of the BesidesthementionedvariationsinthecoreofBalmerlines,
three components reproduces satisfactorily the observed spectra, the line Hα shows variable wings which in a few spectra appear
particularlythemetalliclines, someobservations show small dif- in emission. To analyse these variations we subtracted from the
ferencesinthecoreofBalmerlines(seeFig.4and6).Inthespectra UVES spectra synthetic spectra of the three components as cal-
withstrongHlinessomestrongHelinesappearalsoenhanced,al- culatedinSect.5.Thesedifferencespectrawereusedtocalculate
thoughthesevariationsarealwaysbelow1%ofthecontinuumand residualequivalent widthsintegratinginawavelengthwindow of
shouldbeconsideredasmarginal.Theseline-profilevariationsare 80ÅaroundHα.WefindthatallUVESspectraexhibitanHαex-
notcorrelatedwiththeorbitalphaseofthesystemCa1-Ca2.This cess in the range 0.5–2.3 Å over the synthetic model, except for
is shown in Fig. 6 where we compare two UVES spectra (HJD thespectrumtakenatHJD2456779.9,whichhasamuchstonger
2456869.5077 and2456919.5607) takenatsimilarorbitalphase emission (3.6 Å), probably due to the contamination by the Her-
(nearconjunction)butpresentingdifferentintensityinHlinesand big Bestar (HD164492D) inthe vicinity of HD164492C. A pe-
someHelines.TwoHARPSspectraexhibitingdifferencesinHand riod search on the sample of fourteen UVES difference spectra
Helinesarealsoshown. Inthiscasetheobservations weretaken showsthatthesevariationsmightpresentaperiodicbehaviour,al-
onHJD2457091.8426and2457317.5185,whichcorrespondsto thoughtheperiodisnotunambiguouslydeterminedwiththeavail-
phase∼0.09(nearquadraturewithstarCa1shiftedtothered).In ableUVESobservations. Themost probable periodsare0.557d,
some cases these variations are noticeable on consecutive nights, 1.370 d, and 3.68 d. Figure 7 shows the equivalent width curves
suggestingashortperiodthatmightberelatedwithanon-uniform built with these three periods. The estimated uncertainties are of
surfaceoftherotatingstarCb.Inmostofthespectra,howeverthe theorderof0.05Å.SinceFEROSandHARPSarefibre-fedinstru-
spectralvariationsarecomparabletothespectrumnoiseorthecon- ments,itisdifficulttotakeintoaccountthebackgroundvariations.
tinuumuncertainties,makingdifficultitsuseforderivingtherota- As a results, we do not detect in the spectra obtained with these
tionperiod. instruments clear emission variations similar to those detected in
Morethanonemechanismcouldberesponsibleforthesespec- UVESspectra.
MNRAS000,1–15(2016)
10 J. F.Gonzálezetal.
tainedbyusingtheSVDmethodwith109metalliclinesintheline
mask and with the line mask including He lines (125 lines) are
presented in Figs. 8 and 9. Null polarisation spectra were calcu-
latedbycombiningthesubexposures insuchawaythatpolarisa-
tioncancelsout. Thelinemaskwasconstructed using theVALD
database(e.g. Kupkaetal.2000).Themeanlongitudinalmagnetic
field was estimated from the SVD reconstructed Stokes V and I
usingthecentre-of-gravitymethod(seee.g.Carroll&Strassmeier
2014).Thevelocitywindowusedinthesemeasurementwasabout
400kms−1wide,whichincludesallthreecomponentsinallphases.
We note that due to the fact that HD164492C is a hierarchical
triplesystem and all three components remain blended in all ob-
served spectra, the estimation of the magnetic field can only be
doneassumingthatHD164492Cisasinglestar.Adefinitedetec-
tionofameanlongitudinalmagneticfield,hB i,withafalsealarm
z
probability (FAP) smaller than 10−6, was achieved on all epochs
apart fromthe observation on HJD 2457091.8426 (orbital phase
0.09) where a marginal detection was achieved for the line mask
includingalllines,andnodetectionforthelinemaskincludingex-
Figure7.FluxexcessinHαphasedwiththreeprobableperiods. clusively metal lines.The analysisof thediagnostic Null profiles
showednon-detectionsatallphasesapartfromthesameepochat
theorbitalphase0.09whereweachieveamarginaldetectionwith
Table 5. Logbook of the HARPS spectropolarimetric observations of FAPofabout2×10−5forthesamelinemaskwithmetallines.
HD164492C.Column1:HeliocentricJulianDate;Column2:Orbitalphase
IntheframeworkofourESOLargeProgramme191.D-0255,
of the subsystem Ca1-Ca2; Column 3: Rotational phase of star Cb us-
thefirstspectropolarimetricobservationsofHD164492Cwerecar-
ingP=1.59689dandT0=2457001.263;Column4:Longitudinalmagnetic
ried out in 2013 with the low-resolution FORS2 spectrograph
fieldmeasuredusingmetallicandHelines;Column5:Longitudinalmag-
neticfieldmeasuredusingonlymetalliclines. and were complemented in the same year by two observations
using HARPS in spectropolarimetric mode (Hubrigetal. 2014).
HJD Orbital Rotational hBziall hBzinoHe While FORS2 observations indicated a strong longitudinal mag-
Phase Phase [G] [G] neticfieldofabout500–600G,theHARPSobservationsrevealed
that HD164492C could not be considered asingle star as it pos-
2456445.7946 0.5559 0.1548 615±22 309±64
sessed one or two companions. Given the complex configuration
2456446.7934 0.6355 0.7803 351±24 203±65
and shape of the Stokes profiles, it was also impossible to con-
2456770.7700 0.4808 0.6606 −52±16 −78±41
clude exactly which components possessed a magnetic field. The
2456771.7850 0.5618 0.2963 11±15 18±45
2457091.8426 0.0944 0.7225 −72±27 −27±83 detectionofasignificantStokesV signatureintheSVDandLSD
2457092.8282 0.1730 0.3397 48±26 −89±77 profilesatabout−100kms−1and+150kms−1fromthelinecoreof
2457093.8478 0.2544 0.9782 404±22 419±56 theprimarysuggestedthatmorethanonecomponentmightholda
2457094.8153 0.3316 0.5840 325±32 42±67 magneticfield.Ontheotherhand,assumingthatHD164492Cisa
2457178.6948 0.0231 0.1110 784±20 467±37 singlestar,weobtainedresultsverysimilartothoseobtainedwith
2457179.7426 0.1067 0.7671 −46±36 228±76 FORS2: between 500 and 700G for the first HARPS observing
2457317.5185 0.0977 0.0450 905±37 406±62 night,andbetween400and600GforthesecondHARPSnight.
Since2013 we wereabletoacquire nine additional HARPS
spectropolarimetric observations. Admittedly, even having at our
8 MAGNETICFIELDS
disposal observations at eleven epochs, the interpretation of the
The reduction and magnetic fieldmeasurements were carried out temporal behaviour of theStokes V profilesisachallenging task
usingdedicatedsoftwarepackagesdevelopedforthetreatmentof because the line profiles of the three stars overlap in all orbital
high-resolution spectropolarimetric data. The spectrum reduction phases. In fact, the radial velocity amplitude of the star Ca1, the
and calibration was performed using the HARPS data reduction primaryof thespectroscopic binary, issmaller thantheprojected
pipeline available at the ESO 3.6-m telescope in Chile. The nor- rotationalvelocityofthestarCb,whichhasasimilarspectraltype.
malisationof thespectratothecontinuum levelconsistedof sev- Theradialvelocityamplitudeofthecomponent Ca2islarger,but
eral steps described in detail by Hubrigetal. (2013). The soft- stilloverlapsthewings of theSVDlineprofileof therapidlyro-
ware package used to study the magnetic field in HD164492C, tatingcomponentCb.Wenote,however,thatatseveralepochs(for
the so-called multi-line Singular Value Decomposition (SVD) exampleon02-06-2013and24-04-2014,correspondingbothtothe
method for Stokes profile reconstruction was recently introduced orbitalphase0.56)weobservesignificantfeaturesintheStokesV
byCarrolletal.(2012).MoredetailsonthebasicideaofSVDcan profilesthatarelocatedinthevelocityspacefarfromthepositions
befoundinCarrolletal.(2009).Theresultsofourmagneticfield correspondingtostarsCa1andCa2.Thesefeaturescanonlybeat-
measurements,thoseforthelinelistincludingalllinesandtheline tributedtothestarCb.Itispossibletherefore,toidentifythespec-
list excluding He lines, are presented in Table 5, where we also troscopiccomponentCbasastarpossessingamagneticfield.This
collectinformationabouttheheliocentricJuliandateforthemid- isinlinewithitbeingaHe-richstar.ThegroupofHe-peculiarstars
dleoftheexposure,andtherotationalphaseassumingaperiodof areasarulestronglymagnetic(Borra&Landstreet 1979).Given
1.5969d(seebelow). theratherfastchangesintheshapeofZeemanfeaturesoverconsec-
TheresultingmeanStokesI,StokesV,andNullprofilesob- utivenights,weexpectthattherotationperiodofthiscomponentis
MNRAS000,1–15(2016)