The changing nebula around the hot R Coronae Borealis star DY Centauri

0.34 MB·English

by  N. Kameswara Rao

Mon.Not.R.Astron.Soc.000,1–??() Printed25January2013 (MNLATEXstylefilev2.2)
The changing nebula around the hot R Coronae Borealis
star DY Centauri
N. Kameswara Rao1,2, David L. Lambert2, D. A. Garc´ıa-Herna´ndez3,4,
3
1 & Arturo Manchado3,4,5
0 1 543, 17th Main, IV Sector, HSR Layout, Bangalore 560102, India
2 2The W.J. McDonald Observatory, Universityof Texas, Austin, TX 78712-1083, USA
n 3Instituto De Astrof´ısica De Canarias, V´ıa L´actea s/n, E-38200 La Laguna, Tenerife,Spain
a 4Departmento de Astrof´ısica, Universidad de La Laguna(ULL), E-38206 La Laguna, Tenerife,Spain
J 5Consejo Superior de Investigaciones Cient´ıficas(CSIC), Spain

4
2
Accepted Received;inoriginalform

]
R
S ABSTRACT
.
h
p AmongthedistinguishingcharacteristicsoftheremarkablehotRCoronaeBorealis
- star DY Cen, which was recently found to be a spectroscopic binary, is the presence
o
of nebular forbidden lines in its optical spectrum. A compilation of photometry from
r
t 1970tothepresentsuggeststhatthestarhasevolvedtohighereffectivetemperatures.
s Comparison of spectra from 2010 with earlier spectra show that between 2003 and
a
[ 2010, the 6717 and 6730 ˚A emission lines of [Sii] underwent a dramatic change in
their fluxes suggesting an increase in the nebula’s electron density of 290 cm−3 to

1 3140cm−3 from1989to2010whilethestellartemperatureincreasedfrom19500Kto
v
25000K.The nebularradiusis about 0.02pc, 60000times biggerthanthe semimajor
3
axis of DY Cen binary system. Rapid changes of stellar temperature and its response
7
by the nebula demonstrate stellar evolution in action.
7
5 Key words: Star:individual:DY Cen:variables:RCBtype:nebula,stellarevolution
.
1
0
3
1
1 INTRODUCTION linedspectroscopicbinary(Raoetal.2012),theonlyknown
:
v binary among RCBs.
i
X The origins of the hydrogen-deficient RCB and EHe
DY Centauri is known as a hot R Coronae Borealis (RCB)
r stars are not presently fully understood. Two proposals are
a star. RCB stars are a rare class of peculiar variable stars
commonly advocated. In one, a final helium shell flash or a
with two principal defining characteristics: (i) they exhibit
verylatethermalpulseoccursonacoolingwhitedwarfstar
a propensity to fade at unpredictable times by up to about
(Iben et al. 1983; Renzini 1990; Herwig 2000) which swells
eightmagnitudesasaresultofobscurationbycloudsofsoot,
the envelope of the star to supergiant dimensions for a few
and(ii)theyhaveasupergiant-likeatmospherethatisvery
thousandyears and theremaining hydrogenis convectedin
H-deficient, He-rich and C-rich. The subject of this paper,
andconsumedwhileheliumandcarbonareconvectedoutto
the remarkable star DY Cen, one of the hottest RCBs, is
thesurface.LivingexamplesarebelievedtobeSakurai’sob-
even a peculiar RCB member on several accounts. It is one
ject(V4334Sgr)andFGSge(Asplundetal.1997;Jeffery&
of the most hydrogen-rich RCBs. Its circumstellar environ-
Sch¨onberner2006). The other proposal involves themerger
mentmaybehometothefullereneC60 ormorelikelyproto-
of two white dwarfs: a carbon-oxygen white dwarf accretes
fullerenes (Garc´ıa-Hern´andez et al. 2011a; 2012). In terms
a helium white dwarf as a close binary orbit shrinks under
of its chemical composition, DY Cen may have a composi-
theinfluenceofenergylossbygravitationalradiation(Web-
tion that sets it apart from most RCB and extreme helium
bink 1984; Iben & Tutukov 1986). The merger leads to a
(EHe)stars–seeJeffery &Heber’s(1993) abundanceanal-
swollen envelope around the C-O white dwarf which lasts
ysis: DY Cen is not only uncharacteristically H-rich but is
afewthousandyears.Evidencefrom thechemicalcomposi-
veryFe-poorwithaveryhighS/Feratio,thushasbeenclas-
tionsofRCBandEHestarssuggeststhatmostareproducts
sifiedasaminorityRCBstar(Lambert&Rao1994); i.e.,a
ofa merger(Garc´ıa-Hern´andez et al. 2009, 2010; Pandey &
RCBwithextraordinarilyhighSi/FeandS/Feratios.How-
Lambert 2011; Jeffery et al. 2011).
ever,Jefferyetal.(2011)suggestthattheFeabundancewas
greatlyunderestimatedin1993.Finally,DYCenisasingle- DY Cen with an orbital period of about 39 days is ex-
(cid:13)c RAS
2 N. Kameswara Rao et al.
periencingmasslossandpresumablymasstransfertoalow In addition to these spectra from 2010, we draw on a
masscompanion.1 Moreover,itislikelytohaveexperienced spectrum obtained in 2003 from the 3.9m AAT with the
mass loss and transfer previously in arriving at its H-poor University College London Echelle Spectrograph (UCLES).
condition. Rao et al. (2012) suggest DY Cen may evolve to The UCLES spectrograph was used with the 1.4” slit (slit
aHe-richsdBstar.ManysuchsdBstarsarebinaries.Inthis lengthof6”)2 andthreeindividualexposuresof1200seach
sense, DY Cen is a representative of a third way to form a wereobtained.Wavelengthcoverageisfrom4780to8800˚A.
RCB star. The fact that its supergiant phase is short-lived The final S/N in the summed spectrum is 38 to 40 in the
relative to life as a sdB accounts for the rarity of such hot continuum at 6300 ˚A to 6750 ˚A spectral region. Resolving
RCBs relative to thesdB population. power is about 40000 as estimated from telluric [Oi] lines.
The presence of forbidden emission lines in DY Cen’s Also, considered here is the 1989 spectrum used by Rao et
optical spectrum, was noticed in 1989 (Rao et al. 1993). al. (1993). This was acquired at CTIO with the Blanco 4-
These lines were attributedto anebulaaround DYCen. In metertelescope andaCassegrain echellespectrograph. The
thispaper, weanalyse thenebularlines and theirevolution slit width was 2”3, which gives a resolving power of about
overtwodecades.Nebularlineswereespeciallyprominentin 18000withthewavelengthcoveragefrom 5480to6830˚Ain
a2010high-resolutionspectrumwithbroadwavelengthcov- 1989. A 1992 spectrum with Blanco telescope covered the
erage. Inspection of earlier spectra show that physical con- interval 5480 to 7080 ˚A at a resolving power of 35000 (see
ditionsintheregionemittingthenebularlineshavechanged Giridhar, Rao & Lambert 1996).
since 1989. In particular, theelectron density inferred from DY Cen was identified as a RCB variable by Hoffleit
theratioof the[Sii]6717 and 6731 ˚A lines increased about (1930) from well-determined minima in 1897, 1901, 1924,
eight-fold from 1989 to 2010. and 1929. No RCB type minima have been recorded since
1958 (Bateson 1978, A. A. Henden 2010, private communi-
cation). Thus, DY Cen is not an active RCB star and all
2 SPECTROSCOPIC OBSERVATIONS thespectra discussed here were taken when the star was at
maximum light.
ThegeneralpropertiesofDYCen’sspectrumhavebeende-
Quantitativeinterpretationofnebularemissionlinesre-
scribed by Pollacco & Hill (1991), Jeffery & Heber (1993),
quires knowledge of the line fluxes and, hence, of the con-
Rao, Giridhar & Lambert (1993), Giridhar, Rao & Lam-
tinuum fluxes across a spectrum. To obtain such fluxes, we
bert (1996) and De Marco et al. (2002). Optical spectra
use UBVRI photometry of the star reported for the period
arepresentlyacombinationofphotosphericabsorptionlines
1989 - 2010. DY Cen has been slowly fading for the last 40
(e.g. Oii, Nii, Siiii lines), emission lines due to a stellar
to 50 years (Bateson 1978; Rao et al. 1993). This fading is
wind (mainly Cii, Hei often superposed on underlying ab-
independent of RCB-like events. We have collected UBVRI
sorption lines), and nebular emission lines of [Sii], [Nii],
photometry reported in the literature. Figure 1 shows the
[Oi], [Feii], etc. In addition, circumstellar and interstellar
variation of B, V,and R from 1970 to 2010.
lines from Caii, Nai, Ki and other species as well as the
The earliest published UBV magnitudes are from mid-
enigmatic diffuse interstellar bands (DIBs) are present in
1970 by Marino & Walker (1971): V, B-V and U-B are
absorption (Garc´ıa-Hern´andez et al. 2012).
12.23, 0.39 and -0.62, respectively. U band measurements
This paper considers new spectra from 2010 acquired
arefewandwereobtainedonlyin1970(seeabove)anddur-
as a result of an ESO Director’s Discretionary Proposal.
ing the 1982-83 period at SAAO by Kilkenny et al. (1985).
ThespectrawereobtainedonfournightsinFebruary-March
The 1972 measurements are from V.E. Sherwood (quoted
2010withthecross-dispersedechellespectrographUVESat
by Rao, Giridhar & Lambert 1993). Pollacco & Hill (1991)
the VLT at ESO’s Paranal Observatory. The UVES spec-
obtained BV mesurements during1987 May andJune. The
trograph was used with the 1.2” slit (slit length of 8” and
2007 BVRI measurements are from the AAVSO. It is clear
P.A.=0degrees) andthestandard settingDIC2(390+760).
fromthefigurethatthereisagradualincreaseinmagnitude
Theseeingwas always betterthan1.5” andweobtained 11
over 40 years. The total variation is about 0.90 magnitude
individual exposures of 1800 s each (total exposure time of
inV and0.83 magnitudein Bin37 years:the(B-V)colour
5.5hours).Sincethenebularlinesarenotexpectedtochange
variations are minimal. Even (U-B) seems to be same in
in a short time, the individual exposures were combined to
1970 and 1982-83.
increasethesignal-to-noise(S/N)ratio.TheS/Ninthecon-
Because of these very slow changes, we adopted the
tinuum at 4000 ˚A is ∼200 and at wavelengths longer than
colours observed in 1987 for our spectroscopic observations
6000 it is higher than 250. The slit width is just about the
in1989 andthemid-2007 photometryfor our2010 spectro-
size of the nebular diameter, as estimated later. The spec-
scopic observations to estimate continuum fluxes. The con-
tral resolving power is about 37000, as estimated from O2 tinuum fluxes are estimated from the UBVRI colours after
telluriclinesat6970˚A.Wavelengthcoverageisfrom3300to
applying a correction for interstellar reddening for an E(B-
4500˚A,5700to7525˚Aand7660to9460˚A.AlongwithDY
V) of 0.47 (Garc´ıa-Hern´andez et al. 2011b). Standard re-
Cen, HD 115842 a nearby B supergiant was observed with lations: AV≃ 3.1 E(B-V) and E(B-V)/E(U-B) =0.72, AR/
thesamesetup.HD115842wasobservedsoonaftertheDY
Cenexposuresonallnightstocorrectfortelluriclinesinthe
DY Cen spectrum. The S/N of the HD 115842 spectrum is
300 even for a single night’s observation. 2 Thestarwasconsideredasapointsourceandstandardsettings
wereusedfortheslitposition.
3 Nonebularemissionsperpendicular tothedispersionwerede-
1 Amergerof thetwostars through lossofgravitational energy tected (Rao et al. 1993). Unfortunately, the information about
willnotoccurforabout1015 years. thepositionangleoftheslitisnotavailable.
(cid:13)c RAS,MNRAS000,1–??
DYCentauri’s nebula 3
Gradual fading of DY Cen seems unlikely to be due to
11 an increased presence of dust along the line of sight to the
star; a colour variation would likely result from the dust.
ForamagnitudeextinctiontheE(B-V)changeof0.32isex-
DY Cen pected, if the reddening law of the diffuse ISM is adopted.
Even otherwise RCBs in recovery phases typically show
Av/E(B-V) < 4 - so one would expect a change in E(B-V)
12 of 0.25. Such colour changes are large and noticeable. Any
colour change >0.1 in E(B-V) is noticeable. We speculate
that the fading may be due to the star evolving to higher
temperatureatconstantbolometricluminosity,asexpected
for such stars. Jeffery & Heber (1993) from a spectrum ob-
tained on 1987 April 17 derived an effective temperature
13 of 19500 K. Our 2010 spectra suggest a spectroscopic ion-
ization/excitation temperatureof25000 K.Acrossthetem-
perature range 19000 K to 25000 K, (B-V) is insensitive
to temperature: the change is only 0.07. If the bolometric
magnitude is constant, the V change over this temperature
range is about 0.9 magnitude with V increasing as the ef-
14
fective temperature increases. This is close to the observed
changein V (and B). Therate of evolution may besurpris-
ingly rapid. Jeffery et al. (2001a) have monitored 12 EHe
stars over a period of 10 years looking for the changes in
their T and radii. The surface of four stars was found
eff
to show heating (contraction) at rates between 20 and 120
15
1970 1980 1990 2000 2010 degrees per year in agreement with theoretical predictions
Year basedonmodelsbySaio(1988).DYCen,whichmaybecon-
sidered an EHe star, shows a heating rate of 239 K degrees
Figure1.Longtermvariationsoflightatmaximum.Thegradual peryear somewhat more rapid than EHestars.
fadingoflightfrom1970to2010ofDYCeninBVandRbands.
The arrows indicate the dates of our spectroscopic observations
mentionedinthetext.
3 PHYSICAL CONDITIONS IN THE NEBULA
Thesuiteofnebularlinesprovidesmeasuresofthenebula’s
AV≃0.75 and AI/AV≃0.479 have been used (Cardelli et al electron density, temperature, and radius. Comparison of
1989). Fluxcalibrations given byDrilling & Landolt (2000) the spectra from 1989 to 2010 shows evolution of physical
are adopted. conditions overthesetwo decades. Comparison of thelines’
Reddening-correctedlinefluxesaregivenintheTable1 radial velocities with the binary’s systemic velocity suggest
forthe1989,1992,2003,and2010observations.TheS/Nof thatthenebulamaybesymmetricallydistributedaboutthe
the 2010 spectrum is over 200 to 250. The uncertainities in binary.
equivalent widths (EQWs) are 3% in weaker lines (and less
instronglineslikeHα).Thephotometriccoloursaremoreor
3.1 Radial velocity
lessunchangedsoresultingerrorsinfluxarealsoexpectedto
bearound3%orbetter.Thus,therelativefluxratiosareex- Inobtainingthenebula’sradialvelocity,weusealine’sRitz
pectedtobebetterthanapercent.The2003spectrumS/N wavelengthasprovidedbytheNISTwebsite.4 Radialveloc-
isabout40andtheuncertaintyinfluxesarearound5%level. ities from the individual lines in the 2010 spectra are given
The 1989 and 1992 entries are from Rao et al. (1993) and inTable1.Thereisnoobviouschangeofvelocitywithwave-
Giridhar et al. (1996), respectively. For some lines, correc- lengthorionicspeciesand,therefore,weaveragethesemea-
tionsforblendingwithabsorptionorotheremissionlinesare surements to obtain the mean velocity of 20.9±0.3 km s−1
applied.Inparticular,the[Sii]6730.8˚Alineisblendedwith from 10 lines measured on each of the four nights. From
theCii6731.1˚Aabsorptionlineontheredside,particularly the 1989 and 2003 spectra, we obtain a radial velocity of
on the 1989 July spectrum. A Gaussian fit to the blueside 22±2 and 22.3±1.2 km s−1, respectively. Thus, there is no
of the forbidden line was used in estimating the absorption indication of avelocity change from 1989 to 2010, although
line correction. Since the FWHM (after a minor correction physical conditions were evolving, i.e., the electron density
instrumentalbroadening)agreeswiththeFWHMfrom2010 increased substantially (see below) overthe same period.
spectrum,wesuggest thatproperallowance ismadefor the The sequence of 2010 VLT/UVES spectra shows very
blending. In the 2010 spectra several Cii lines are in emis- clearlytheindependenceofthenebularradialvelocityfrom
sion. The emission is seen only in the core of absorption thatofthestellarphotosphere.The2010andothervelocities
linesofthemultipletofwhich6731.1˚Aisamember(Figure ledRaoetal.(2012)toproposethatDYCenisasingle-lined
2).Thecontributionofthislinetothestrongemission [Sii]
6730.8 ˚A is fortunately minor. Thus, we trust theline ratio
of [Sii]6716 ˚A to 6731 ˚A is not undulyaffected. 4 http://physics.nist.gov/PhysRefData/asd.cfm
(cid:13)c RAS,MNRAS000,1–??
4 N. Kameswara Rao et al.
Table 1. Dereddened Nebular line fluxes in DY Cen
2010 2003(1992)a 1989
Ion λc Eq.w Fluxb (FWHM)0 Rad.Vel Eq.w Fluxb Eq.w Fluxb (FWHM)0 Rad.Vel
˚A (m˚A) x10−14 kms−1 kms−1 (m˚A) x10−14 (m˚A) x10−14 kms−1 kms−1
[Oii] 3726.0 433 10.2 34 20.9
[Oii] 3728.8 224 5.3 34 18.8
[Nii] 5754.6 38 0.28 41 19.8
[Oi] 6300.3 131 0.76 28.5 20.2 74 0.46 87 0.68 26 23.3
[Oi] 6363.8 52 0.29 26.6 21.1 33(67) 0.20(0.47)
[Nii] 6548.0 1425 7.25 41 21.3 478(46) 2.59(0.29) 47 0.32 25.8
Hα 6562.8 9620 48.6 44.6 3798(2000) 20.4(12.4) 1473 9.95
[Sii] 6716.5 350 1.61 34 22.6 157(67) 0.77(0.38) 92 0.56 34 22.2
[Sii] 6730.8 588 2.69 34 22.0 169 0.82 83 0.50 34 22.4
[Oii] 7319.1 93 0.31 43.7
[Oii] 7320.1 273 0.92 46.6
[Oii] 7329.7 154 0.52 45
[Oii] 7330.8 157 0.53 45
[Feii] 4287.4 25 0.36 10.4 21.1
[Feii] 4359.3 17 0.26 8.8 21.5
[Feii] 4413.8 13 0.20 8.6
[Feii] 4243.9 7 0.10
a 1992datafromGiridharetal.(1996) aregivenbetweenbracketts
b inergscm−2 s−1
i
Table 2. Flux ratios of H Paschen emission (nebular) lines in DY Cen in 2010
Te 9425K
ne 103 cm−3
Line λc Eq.w Fluxa (FWHM)0 Rad.Vel. j/jHα(obs) j/jHα(CaseB)
˚A (m˚A) x10−14 kms−1 kms−1
p11 8862.8 162 0.34 34.7 20.4 0.0069 0.0059
p12 8750.5 125 0.27 34.8 19.7 0.0056 0.00489
p14 8598.4 72 0.17 36.7 20.2 0.0034 0.0029
p17 8467.3 55 0.13 34.4 20.1 0.00275 0.0015
p20 8392.4 20 0.05 41 0.0010 0.00084
a inergscm−2 s−1
spectroscopic binary. Over four nights of the 2010 observa- 1989 and 2010 spectra. The instrument-corrected Gaussian
tions,thephotosphericvelocityvariesoverarangeof22km FWHMs (FWHM)0 of various lines are given in Table 1.
s−1 butthenebularvelocityisconstanttowithin1kms−1. There seems to be a mild trend for (FWHM)0 of forbidden
The overall mean velocity from the nebular lines is in good linesofN,O,andStoincreasewithionizationpotential.All
agreement with theγ-velocity of 21.3±0.5 km s−1 provided widths are greater than the thermal widths. The suggested
by the binary solution from the photospheric absorption expansion velocities are in the range 17 to 22 km s−1.
lines (Rao et al. 2012). This close agreement in radial ve-
locity suggests thatthenebulaissymmetrically distributed Clearly, the [Feii] lines have much the smallest width
about the binary. With the estimate of the nebula’s radius (comparable to the instrumental width) but share the ra-
(seebelow) farexceedingtheorbital size,‘nebula’seemsan dialvelocityofotherforbiddenlines.Several[Feii]linesare
appropriatedescriptionfortheregionemittingtheseforbid- present in the blue part of the 2010 spectra (Table 1). The
den lines. spectrum of DY Cen obtained by Jeffery & Heber (1993)
on1987 April16ataspectral resolvingpowerof27000 and
displayed in Leuenhagen, Heber & Jeffery (1994) shows no
3.1.1 Nebular line widths & [Feii] lines traceof[Feii]emissionlines.DeMarcoetal.(2002)obtained
coupleofbluespectraataresolutionof35000on2001April
The nebular lines have full-widths at half maximum inten- andSeptember.Theregionaround4267˚Adisplayedintheir
sity (FWHMs) larger than the instrumental width for the figure12showsweak[Feii]emissionat4287˚Aispresentbut
(cid:13)c RAS,MNRAS000,1–??
DYCentauri’s nebula 5
2 2
[S II]
DY Cen
DY Cen
1.5
1.5
C II(21) C II(21)
2010 Feb 27
1 2003 July 18
1
1992 May
1989 July
0.5
0.5 O II
6700 6720 6740 6760 3720 3725 3730

Figure 2. Dramatic changes of the intensity ratio of [Sii] 6716 Figure3.The[Oii]3727and3729˚Alinesinthe2010spectra.An
˚A to 6731 ˚A between 1989 and 2010. From 1989 to 2003, the absorptionlineofOiiat3727.3˚Aaffects thelinesratioslightly;
6716 ˚A line is stronger than or comparable in strength to the the estimated effect of this line is to reduce the 3726/3729 line
6731˚Alinebutinthe2010spectrumthe6731˚Aisobviouslythe ratiobyonlyapercent.
stronger line. The apparent doubling of the 6731 ˚A line in the
1989spectrummaybeduetocontamination bya[Feii]line.
Fromthe2010spectra,the[Oii]ratiogivesNe =3105±
not as strongly as we see in our 2010 spectra: the peak rel- 200cm−3 andthe[Sii]givesNe =3170±200cm−3 andwe
ative flux in 2001 was ∼ 1.07 whereas it is ∼ 1.13 in 2010. adopttheaverageNe =3140cm−3.Inderivingthesevalues
Theradial velocities agree with thoseof othernebularlines weadopttheelectrontemperatureTe=9350K(seebelow).
Figure 3 shows the [Oii] lines in the UVES spectrum. The
(Table1).Thelinewidthsobtainedin 1989 spectrumagree
presence of Oii absorption line at 3727.3 ˚A between the
with theones determined from 2010 spectra.
Usually[Feii]linesarisefromlowexcitationhigherelec- two [Oii] lines affects the estimation of the [Oii] line ratio
tron density regions (Ne ∼ 104- 106cm−3; Perinotto et al. slightly. Figure 2 shows the [Sii] lines in the 1989, 1992,
2003, and 2010 spectra and the reversal from 1992 to 2010
1999, Viotti 1976). Where from around the region around
DY Cen could they arise? The [Feii] lines share the radial in theflux ratio of the lines. The [Sii] ratio gives Ne =620
velocity of otherforbidden lineswhich suggests that there- cm−3 in 2003, Ne ≃450 cm−3 in 1992, andNe ≃290 cm−3
in1989.Figure4displaysthewavelengthregion aroundthe
gion of formation is linked to that of the general nebula.
Then,might theybefrom theinterfacebetweentheHiire- [Nii] at 5755 ˚A in the 1989, 1992, 2003, and 2010 spectra,
showing theappearanceof [Nii]5755˚A in 2010. In addition,
gionandthesurroundingneutralregionatwhichthehigher
velocity expanding hot nebular gas pushes into the neutral the6540-6590˚Aspectralregion,whichcoversHα,[Nii]6548
and6583˚Anebularlines,andCiistellarwindlines,in2003
region?
and 2010 is compared in Figure 5.
For the 2010 spectra, the electron temperature is ob-
3.2 Electron density and temperature tainable from the flux ratio of the [Nii] lines (6583 +
6548)/5755. Because of blending of the 6583 ˚A line with
For the 2010 spectra, the electron density Ne is obtainable emission from a Cii stellar wind line, we calculate its
from two line ratios: [Oii] 3726˚A/3729˚A, and [Sii] 6716 strengthfromthe6548˚Alineandtheelectrondensity.From
˚Alim/6it7e3d1˚Asp.eFctorralthceov1e9ra8g9e,anthde2[0S0i3i]srpaeticotrias twhiethsotlheeeilrecmtroorne Nthoete[Nthiai]troautriom, ewaesuorbedtauinpptehrelitmemitpteorathtuer[eNTiie]5=75953˚A50liKne.
density indicator.5
5 Physicalparameters–electrondensityandtemperature–were 2012), usingtheatomicparameterslistedinGarc´ıa-Rojas,etal.
derived using the ‘PyNeb’ software package (Luridiana et al. (2012).
(cid:13)c RAS,MNRAS000,1–??
6 N. Kameswara Rao et al.
2
DY Cen DY Cen
N II Al III Si III C III
C II
1989 Jul
10
1.5
1992 May
[N II]+C II
2003 Jul
1
5
2010 Feb-Mar
[N II] C II
2010 Feb-Mar ISM
0.5 2003 Jul

5700 5750 5800 6540 6550 6560 6570 6580 6590

Figure 4. The wavelength region around 5750 ˚A showing the Figure 5. The Hi 6562 ˚A, [Nii] and Cii lines in the 2003 and
appearancein2010ofthenebularlineof[Nii]at5755˚A. 2010spectra.
inthe2003spectrum(seeFigure4)providesanupperlimit Hestars:theratiooftheintegralsisabout42infavourofthe
of 11000 K for Te. 25000 Kmodelrepresentativeofthestarin2010relativeto
Jeffery & Heber (1993) provide the composition of the the19000Kmodelfor1989.ThisfactorwiththeR∗termas-
starasHe:Hof7.5:1(or11.52to10.65inlogscale).Ourre- sumingevolutionatconstantluminosityleadstoapredicted
centunpublishedanalysisofthe2010spectrumisconsistent increase in electron density by a factor of (42/2.25)0.5 or 4
with this number (11.52 to 10.76±0.20). After helium next from 1989 to 2010. This is less than the measured increase
dominant gas in the atmosphere of DY Cen is hydrogen. If from the[Sii]lines of afactor of 11between 1989 and 2010
the nebula has a similar composition to that of DY Cen’s but is comparable to the factor of 5 increase between 2003
photosphere, the leading electron donor is expected to be and 2010. This might suggest non-uniform increase of Teff
hydrogen despite the fact that, as a RCB star, DY Cen is ofthestarduringthisperiodsinceweexpectthebolometric
H-poor.Inthenebula,theelectrondensityissetbythebal- luminosity to remain constant. The rate of change of elec-
ance between photoionization and radiative recombination. trondensitybetween1989and2003isonlyafactorof2and
Photoionization bystellar radiation is proportional to from 2003 to 2010 is a factor of 5.
NHR∗2Z πFνσνdν/hν=NHR∗2P(ν) (1) 3.3 Ionic abundances
where NH is the density of neutral hydrogen, Fν the KnowledgeofTe andNe enablestheionicabundancestobe
stellar flux, R∗ the stellar radius, σν the photoionization calculated (see Table 3). With the estimated Te and Ne of
cross-section and the integral is evaluated from the Lyman 9350Kand3140cm−3,the2010linefluxes(Table1)provide
limit to higher frequencies. Radiative recombination occurs theO0/H+,O+/H+,N+/H+,andS+/H+ ionicratioslisted
atarateNe2α(T)whereα(T)isthetemperature-dependent inTable3.TheN+/O+ratiois0.46andtheS+/O+is0.019.
total radiative recombination coefficient. AssumingO0/O+ =N0/N+andalsoS0/S+,andestimating
If the nebula was largely undisturbed with a low de- the total O (= O0 + O+), N (=N0 + N+), and S (=S0 +
gree of ionization (i.e., Ne << NH from 1989 to 2010, the S+) reducesthe aboveratios by about 15 %.
electron density scaled as On the assumption that N/O ≃ N+/O+, this nebular
(Ne2α(T))2010 ∝ R∗(2010)2 P(ν)2010 (2) rvaatluioeiswahbicohutisaNfa/cOto=r o0f.t1h4referolmargJeerfftehrayn&thHeepbheort’sos(p1h9e9r3ic)
(Ne2α(T))1989 R∗(1989)2 P(ν)1989 LTEanalysisand0.07fromourunpublishednon-LTEanal-
TheintegralsP(ν)maybeestimatedfrommodelatmo- ysis (in preparation). Interestingly, the photospheric S/O
spheres computed by Jeffery et al. (2001b) for carbon-rich ratiois0.018(Jeffery&Heber1993),avalueingoodagree-
(cid:13)c RAS,MNRAS000,1–??
DYCentauri’s nebula 7
Table 3. Derived ionic abundances
2.5
(Te=9350K,Ne=3140cm−3) DY Cen
sys
Ion Lines Abundance
N+/H+ λ6548, λ5755 3.1x10−5 2
O0/H+ λ6300, λ6363 1.2x10−5
O+/H+ λ3726, λ3729 6.6x10−5
S+/H+ λ6716, λ6731 1.3x10−6
1.5
mentwiththenebula’sionicratio. Thus,comparison ofthe
H I P17(x 6)
nebulaandphotosphericN/O/Sratiossuggeststhatthefor-
merisN-richrelativetothelatter.Thiscouldbeaccounted
for if the nebularepresents the envelopeof DY Cen ejected 1
afteritsfirstdredge-upwhentheenvelopeisN-richfollowing
the dredge-up of CN-cycled material. To a rough approxi-
mation, much of the envelope at that point is N-rich owing
cs
to the conversion of large amounts of the C to N, say N/O
≃ 0.5. The He shell and deeper layers of the star surviving 0.5
the episode of envelope ejection may well be far less N-rich
because theN has been processed by α-captures to 22Ne.
0
-100 0 100
3.4 Hydrogen emission lines
The2010spectrumshowsBalmerandPaschenlinesinemis-
sion. The Paschen lines, particularly the lower members of Figure6.ProfilesoftheOi8446˚A(red)alongwithHiPaschen
the series, occur as photospheric absorption lines on which 17 (six times magnified- blue) shown with Caii 3933 ˚A (black).
nebular emission is superposed. We estimate emission line Oiprofileisverydifferentshowingasuperposedabsorption(dou-
fluxes from higher members of the series where absorption blepeak)unlikeothernebularlines,HαorPaschenlines.Oi8446
is minimal. Fluxes are given in Table 2. ˚AlineisproducedbyresonancefluorescenceofHiLβ withOiλ
1025. Note the coincidence of the circumstellar absorption com-
Comparison between observed and predicted flux ra-
tiosisaccomplishedusingHirecombinationlineratioscom- ponent in Caii 3933 ˚A with superposed absorption in Oi 8446
˚A.Thisabsorptioncomponentisassociatedwiththenebularand
puted by Storey & Hummer (1995) for Case B - see Oster-
systemicradialvelocity.
brock & Ferland (2006 - Table 4.4). We interpolate to con-
ditions Te =9350 K and Ne =1000 cm−3 for the fluxratio
of a Paschen line to Hα; the predictions are mildly sensi- sorptionorasymmetry.Itiscenteredonthenebularvelocity
tiveto the adopted Te and insensitive to Ne. Observed and and has a range -42.5 to 88.5 km s−1 in radial velocity.
predicted ratios for five clean Paschen lines with respect to The superposed absorption suggests the presence of
HαaregiveninTable2.Goodagreementisseensuggesting neutral gas surrounding the nebula. An estimate of the
that the lines including Hαare optically thin. amount of thisgas is made from theabsorption component
TheOi8446˚Alineispresentstronglyinemission(Fig- ofCaii3933 ˚Aline(Figure6).Theequivalentwidthofthis
ure6)butpresentsadifferentprofilethaneitherHαorother componentisabout55m˚A.Theexpansionvelocityfromthe
nebularlines.Figure6showsacomparisonwithHiPaschen FWHMisestimatedtobeabout8.5kms−1.Assumingitis
17line.SincetheFWHMoftheOilineisaboutthesameas opticallythin,thecolumndensityofCa+is6.4x1011cm−2.
othernebularlines, it is plausible toattributetheapparent The gf values from Morton (2003). Assuming that Ca+/Hi
doublingtosuperposedabsorptionratherthantoablendof ∼ Ca/H and further assuming solar abundance ratio, the
twoemission lines.Oi8446˚Aisproducedbyresonanceflu- Hi column density estimated is 2.7 x 1017 cm−2. For the
orescence of Hi Lβ with λ1025.76 of Oi. The excitation of nebular radius estimated (in the next section) the mass of
Oibyλ1025.76isfollowedbyemissionat11287˚Aandthen neutralgasis estimated as 1.2 x10−5 M⊙.This estimate is
byemission at 8446 ˚A.TheOiexistsineithertheH+ zone dependentonhowthelineofsighttothestarinterceptsthe
thankstochargeexchangebetweenoxygenandhydrogenor nebula.
the neutral zone immediately surrounding the Hii region.
The superposed absorption component in Oi 8446 ˚A line
3.5 The size of the nebula
implies that even Hi Lβ might be experiencing the super-
posed absorption. It is a curious fact that a circumstellar DeMarcoetal.(2002)suggestthatthedistancetoDYCen
absorption component at the same radial velocity exists in ismuchmorethan4.8kpc,aswassuggestedearlierbyGirid-
Caii λ3933 and is associated with stellar velocity (not the har,Rao&Lambert(1996).AssumingthatDYCenhasthe
systemic velocity!). Although Hα profile shows much wider same Mv (-3.0 mag) as its spectroscopic twin the hot RCB
radial velocity range it does not show any superposed ab- star HV 2671 in the Large Magellanic Cloud (De Marco et
(cid:13)c RAS,MNRAS000,1–??
8 N. Kameswara Rao et al.
al.2002), theestimated distanceofDYCenisabout 7kpc. Interestingly, DY Cen shows a heating rate (contrac-
In addition, the very nearby B-type star HD 115842 (with tion)morerapidthanEHestars.Ineitherofthetwopresent
E(B-V)=0.5ofISMidenticaltoDYCen)mayprovideacon- evolutionaryscenarios(whitedwarfmergersoralasthelium
straint that the distance should be greater than 5 kpc. HD flash)for theorigin of EHestarsand RCBs, thestar hasto
115842 seemstobeahypergiantverysimilar toHD169454 become a cooler supergiant (for a second time) and starts
(Mv=-9.2mag;Federman&Lambert1992) asindicatedby contractionatconstantluminosity.Whethertherateofcon-
the presence of Fe III emission lines in its spectrum. For tractionisuniformintimeornotisnotclear.IncaseofDY
Mv=-9.2 mag, HD 115842 would be at a comparable dis- Cen this path of evolution might not be the one the star
tanceof5.4kpc.Thus,assumingadistanceof7kpcforthe passed through. The binarity of the star might introduce a
star (De Marco et al. 2002) and the Hα flux given in Ta- different course (see Rao et al. 2012 for some discussion).
ble 1 for 2010, the Hα effective recombination coefficient of Furtherobservationsofthestararemostcrucialandimpor-
8.6 x 10−14 cm−3 s−1 for the nebula’s density and temper- tant in understanding the evolution of hydrogen deficient
atureleadstoanangularradiusofthenebulaofabout0.64 stars.
arcsec. Thus, the slit width of 1.2 arcsec used for the 2010 One is hopeful that DY Cen’s future behaviour will be
observations should havecovered the nebularemission. followed intensively spectroscopically and photometrically.
The geometrical radius corresponding to the angular In particular, it will be important to separate behaviour
radius is 6.5 x 1016 cm. The semi-major axis of the binary linked to the 39 day orbital period from the supposedly
systemisestimatedtobe9.3/siniR⊙(Raoetal.2012)where slower evolutionary trends as DY Cen crossed the top of
i is the angle of inclination of the binary orbit to the line theH-Rdiagram.
of sight. For sin i ≃ 0.6, the semi-major axis is 1.1 x 1012
cm. Thus, the nebular radius is ∼0.02 pc or 60000 times
larger than the binary orbit. The earlier estimated nebular
expansion velocity (17 to 20 km s−1), suggests an age of 5 ACKNOWLEDGEMENTS
about 1100 years for the nebula. The ionized mass (mainly
Weacknowledgetheanonymousrefereeforusefulcomments
Hwhichisprovidingelectrons)withinthenebulaisabout3
x10−3 M⊙.6 Notethatthemainelectrondonorishydrogen that help to improve the paper. We also acknowledge with
thanks the variable star observations from the AAVSO in-
because: i) no nebular emission lines of He I are observed
ternational database contributed by world wide observers
(i.e.,nodominantnebularHeIrecombinations);andii)the
used in this research. This research has made use of the
photoionizations of H I dominate over He I by a factor of
∼35. Thus most of theelectrons are contributed byH I. SIMBAD database, operated at CDS, Strasbourg, France.
Our sincere thanks are due to Simon Jeffery and Vincent
WoolfforsupplyingtheAATarchivalspectrumofDYCen.
N.K.R.wouldliketothankDavidandMelodyLambertfor
4 CONCLUDING REMARKS their hospitality at UT, Austin where part of this work is
done.D.A.G.H.andA.M.acknowledgesupportforthiswork
The nebula centred on DY Cen, a hot RCB star, is surely providedbytheSpanishMinistryofEconomyandCompet-
intimately related to the star because its radial velocity is itiveness undergrant AYA-2011-27754.
coincidentwithinaboutonekms−1withthesystemicveloc-
ityofbinarysystemtowhichDYCenbelongs.Thenebula’s
expansion velocity as estimated from the width of nebular
forbiddenlinessuggeststhatthenebulamayhavebeencre- REFERENCES
ated about 1000 yearsago.
Spectrafrom1989to2010indicatethatthestarandthe Asplund, M., Gustafsson, B., Lambert, D. L., Kameswara
nebulaareevolving.Inapparentconcertwithanincreasein Rao, N.1997, A&A321,17
theeffectivetemperatureofthestar,theelectrondensityin Bateson, F. M. 1978, Publs. Var. Star, Sect., R. Astron.
the nebula has increased several-fold over the two decades Soc. New Zealand, 6, 39
sampled by available high-resolution spectra. Cardelli, J.A.,Clayton,G.C., Mathis,J.S.1989, ApJ,345,
Itisquitepossiblethattherecentformationoftheneb- 245
ula,thechangesinnebularconditionsoverthepreceding20 De Marco, O., et al , 2002, A.J.,123, 3387
years, and the increase in the stellar effective temperature Drilling,J.,Landolt,A.U.,2000, in‘Allen’sAstrophysical
since1987andthesteadyfadingofthestaratopticalwave- Quantities’ ed.Cox
lengthsareall manifestations ofDYCen’spast andpresent Flower, P.J., 1996, ApJ., 469, 355
evolution. The nebula may be the product of an episode of Federman, S.R., & Lambert, D.L. 1992, AJ, 104, 691
common envelopeevolution that resulted ina primary(DY Garc´ıa-Hern´andez, D. A., Hinkle, K.H., Lambert, D.L.,
Cen) which became a H-deficient supergiant. Such super- Eriksson, K. 2009, ApJ, 696, 1733
giantsareexpectedtoevolverapidlytoaveryhotluminous Garc´ıa-Hern´andez,D.A.,Lambert,D.L.,KameswaraRao,
star at thetip of thewhite dwarf cooling track. N.,Hinkle, K. H.,Eriksson, K., 2010, ApJ,714, 144
Garc´ıa-Hern´andez, D. A., Rao, N. K., Lambert, D. L.,
2011a, ApJ, 729, 126
6 TheionizedHmassisestimatedfromtheexpression(4π/3R3) Garc´ıa-Hern´andez, D. A., Rao, N. K., Lambert, D. L.,
x (np(=ne)/mp) where R is the estimated radius of the nebula 2011b, ApJ,739, 37
(=6.5x1016 cm),np isnumberofprotons(=no.electrons/cm3) Garc´ıa-Hern´andez,D.A.,Rao,N.K.,Lambert,D.L.,2012,
andmp istheprotonmass. ApJ,759, L21
(cid:13)c RAS,MNRAS000,1–??
DYCentauri’s nebula 9
Garc´ıa-Rojas, J., Pen˜a, M., Morisset, C., Mesa-Delgado,
A., & Ruiz,M. T. 2012, A&A,538, A54
Giridhar,S.,Rao,N.K.,Lambert,D.L.,1996,JAp&A,17,
23
Herwig, F., 2000, A&A,360, 952
Hoffleit, D. 1930, Harvard College Observatory Bulletin,
874, 13
Iben, I., & Tutukov,A. V.1986, ApJ., 311, 753
Iben,I.,Kaler,J. B., Truran,J. W.,et al. 1983, ApJ,264,
605
Jeffery, C.S., Heber,U., 1993, A&A,270, 167
Jeffery, C.S., Starling, R.L. C., Hill, P.W., & Pollacco, D.
2001a, MNRAS,321, 111
Jeffery, C.S., Woolf, V.M., & Pollacco, D.L. 2001b A&A
376, 497
Jeffery, C. S., Sch¨onberner,D. 2006, A&A,459, 885
Jeffery, C.S., Karakas, A.I., Saio, H., 2011, MNRAS, 414,
3599
Kilkenny,D.,et al , 1987, SAAO.cir., 9, 87
Lambert, D. L., Rao, N. K., 1994, JAp&A,15, 47
Leuenhagen,U.,Heber,U.,Jeffery,C.S.,1994,A&AS,103,
445
Luridiana,V.,Morisset,C.,&Shaw,R.A.2012,IAUSym-
posium, 283, 422
Marino, B.F., Walker, W.S.G., 1971, Cir. Roy. Astr. Soc.
New Zealand. Var. StarSec., 184, 1
Morton, D.C., 2003, ApJ, suppl., 149,
Pandey.G, Lambert,D.L, 2011, ApJ, 727,122
Perinotto,M.,Bencini,C.G.,Pasquali,A.,Manchando,A.,
Rodriguez Espinosa, J.M., Stanga, R., 1999, A&A, 347,
967
Pollacco, D.L., Hill, P.W., 1991, MNRAS,248, 572
Osterbrock, D.E., Ferland, G.J., 2006, ‘Astrophysics of
Gaseous Nebulae and ActiveGalactic Nuclei’, University
Science Books, Sausalito, California
Rao, N. K., Giridhar, S., Lambert, D.L., 1993, A&A, 280,
201
Rao N. K., Lambert, D. L., Garc´ıa-Hern´andez. D. A., Jef-
fery, C. S., Woolf, V. M. McArthur,B., 2012, ApJ, 760,
L3
Renzini,A.,1990, in‘Confrontation betweenstellar pulsa-
tion and evolution’, eds. Cacciari,C., Clementini, G., As-
tronomical Society of the Pacific Conf. Ser. 11, 549, San
Francisco, CA.
Saio, H., 1988, MNRAS,260, 465
Storey, P.J., Hummer, D.G., 1995, MNRAS,272, 41
Viotti, R.,1976, ApJ, 204, 293
Webbink,R.F., 1984, ApJ, 277, 355
(cid:13)c RAS,MNRAS000,1–??