UVES and X-Shooter spectroscopy of the emission line AM CVn systems GP Com and V396 Hya

1.4 MB·

by  T. Kupfer

Mon.Not.R.Astron.Soc.000,1–14(2002) Printed13January2016 (MNLATEXstylefilev2.2)
UVES and X-Shooter spectroscopy of the emission line
AM CVn systems GP Com and V396 Hya
6
1 T. Kupfer1,2⋆†, D. Steeghs3, P. J. Groot2, T. R. Marsh3, G. Nelemans2,4 and
0
G. H. A. Roelofs2
2
1Divisionof Physics, Mathematics, and Astronomy, California Institute of Technology, Pasadena, CA 91125, USA
n
2Department of Astrophysics/IMAPP, Radboud University Nijmegen, P.O. Box 9010, 6500 GL Nijmegen, The Netherlands
a
J 3Department of Physics, University of Warwick, Coventry CV4 7AL, UK
4Institute for Astronomy, KU Leuven, Celestijnenlaan 200D, 3001 Leuven, Belgium
2
1

] Accepted —Received—;inoriginalform—
R
S
. ABSTRACT
h
We present time-resolved spectroscopy of the AMCVn-type binaries GPCom and
p
V396HyaobtainedwithVLT/X-ShooterandVLT/UVES.Wefullyresolvethenarrow
-
o central components of the dominant helium lines and determine radial velocity semi-
r amplitudesofK =11.7±0.3kms−1 forGPComandK =5.8±0.3kms−1 for
t spike spike
s V396Hya.Themeanvelocitiesofthenarrowcentralcomponentsshowvariationsfrom
a line toline.Comparedtocalculatedline profilesthatinclude Starkbroadeningweare
[
able to explain the displacements, and the appearance of forbidden helium lines, by

1 additional Stark broadening of emission in a helium plasma with an electron density
v ne ≃ 5×1015 cm−3. More than 30 nitrogen and more than 10 neon lines emission
1 linesweredetectedinbothsystems.Additionally,20nitrogenabsorptionlinesareonly
4 seen in GP Com. The radial velocity variations of these lines show the same phase
8 and velocity amplitude as the central helium emission components.
2
The small semi-amplitude of the central helium emission component, the consistency
0
of phase and amplitude with the absorption components in GPCom as well as the
.
1 measured Stark effect shows that the central helium emission component, the so-
0 called central-spike, is consistent with an origin on the accreting white dwarf.
6 We use the dynamics of the bright spot and the central spike to constrain the binary
1 parameters for both systems and find a donor mass of 9.6 - 42.8M for GPCom
Jupiter
:
v and 6.1 - 30.5MJupiter for V396Hya.
i We find an upper limit for the rotational velocity of the accretor of v < 46kms−1
X forGPComandv <59kms−1 forV396Hyawhichexcludesafastrortoattingaccretor
rot
r in both systems.
a
Key words: accretion, accretion discs – binaries: close – stars: individual: – stars:
individual: GP Com, V396 Hya
1 INTRODUCTION evolved semi-degenerate helium star companion (e.g. Nele-
mans et al. 2001). Observationally they are characterized
Accretion ontowhite dwarfs in cataclysmic variables(CVs)
by a high deficiency of hydrogen (< 1%) and short orbital
commonly proceeds via Roche-lobe overflow of hydrogen-
periods (<1hour), indicating an advanced stage of binary
rich material from a main-sequence type donor star. A
evolution.
smallnumberofsystemshavebeenidentifiedwithhydrogen-
deficient,degenerate,donorstars;theAMCVnsystems(see
Since the identification of the prototype, AMCVn,
Solheim2010forarecentreview).Thesesystemsconsistofa
as a semi-detached pair of degenerate dwarfs (Smak
whitedwarf(WD)primaryandeitheraWDorsignificantly
1967; Paczyn´ski 1967; Faulkneret al. 1972), over 40
additional systems and candidates have been identi-
fied. Confirmed orbital periods range from 5-65 min-
⋆ E-mail:[email protected]
† BasedonobservationsmadewithESOtelescopesattheParanal utes (e.g Natheret al. 1981; O’Donoghueet al. 1987, 1994;
Observatory under programme ID 69.D-0562(A) and 084.D- Ruizet al. 2001; Woudt& Warner 2003; Roelofs et al.
0814(A). 2005, 2006, 2007; Anderson et al. 2005, 2008; Roelofs et al.
2 T. Kupfer et al.
Table 1.SummaryoftheobservationsofV396HyaandGPCom
System Tele./Inst. Nexp Exp.time(s) Total time(h) Coverage(˚A) Resolution
V396Hya
2002-04-07 VLT/UVES(blue) 28 360 3.33 3826-5053 ∼40000
2002-04-07 VLT/UVES(red) 28 360 3.33 5763-9462 ∼40000
2010-02-18 VLT/X-Shooter(UVB) 3 900 0.75 3050-5550 4300
2010-02-18 VLT/X-Shooter(VIS) 3 900 0.75 5340-10200 7400
2010-02-18 VLT/X-Shooter(NIR 3 900 0.75 9940-24800 5400
GPCom
2002-04-07 VLT/UVES(blue) 110 120 5.25 3826-5053 ∼40000
2002-04-07 VLT/UVES(red) 110 120 5.25 5763-9462 ∼40000
2010-02-18 VLT/X-Shooter(UVB) 45 60 1.00 3050-5550 4300
2010-02-18 VLT/X-Shooter(VIS) 45 60 1.00 5340-10200 7400
2010-02-18 VLT/X-Shooter(NIR) 45 60 1.00 9940-24800 5400
2010; Levitan et al. 2011; Kupferet al. 2013; Levitan et al. dwarf accretor (Marsh 1999; Morales-Rueda et al. 2003).
2013; Carter et al. 2014). AMCVn binaries are impor- Suchso-called‘central-spikes’havesofaronlybeenobserved
tant as strong, low-frequency, Galactic gravitational wave in AMCVn systems and He-rich dwarf novae but never
sources (e.g. Nelemans et al. 2004; Roelofs et al. 2007; in hydrogen dominated cataclysmic variables (Breedt et al.
Nissanke et al.2012),thesourcepopulationoftheproposed 2012).Iftheaccretororiginofthecentral-spikescanbecon-
.Ia supernovae (Bildsten et al. 2007), and as probes of the firmed,the central-spikes would be a powerful tool to trace
final stages of low-mass binary evolution. themotion of theaccreting white dwarf.
The accreting white dwarf accretes helium with a spe-
Several formation channels have been proposed
cific angular momentum from the inner edge of the ac-
for these systems; a double white dwarf channel
cretion disc. This will spin up the accretor and is ex-
(Tutukov& Yungelson 1979), a channel in which the
pected to result in a minimum equatorial velocity of ∼
donors are low-mass helium stars (Savonijeet al. 1986;
1250kms−1 (Bildsten et al. 2006). The rapid surface rota-
Tutukov& Fedorova 1989; Yungelson 2008), and one
tion will broaden any spectral features originating on the
withevolvedpost-main-sequencedonors(Thorstensen et al.
whitedwarfsignificantly.Sofarnodetailedstudyofthero-
2002; Podsiadlowski et al. 2003). A way to distinguish be-
tational velocity has been performed. However, the narrow
tween thesescenarios is toobtain thechemical composition
central-spike features already suggest a much lower rota-
of the donor, and in particular the C/O, N/O and N/C ra-
tionalvelocity,iftheoriginofthecentral-spikeonthewhite
tios, due to different levels of CNO and He burning in the
dwarf can be confirmed. This would imply an effective loss
progenitor of the donor (Nelemans et al. 2010). High N/O
of angular momentum from the rapidly spinning accretor.
andN/Cratiosareexpectedforaheliumwhitedwarfdonor
In an attempt to understand the origin of the pecu-
and significantly lower N/O and N/C ratios expected for a
liar spike, we secured echelle spectroscopy of GPCom and
semi-degenerate donor. Although the donor has never been
V396Hya. The UVES and X-Shooter spectra allow us to
observed directly, the accreted material in the disc and the
fully resolve the kinematics of the line profiles in general,
photosphereoftheaccretingwhitedwarfisexpectedtorep-
and their central-spikesin particular.
resent thecomposition of thedonor.
In this paper we present high-resolution optical spec-
troscopy of the AMCVn systems GP Com (Natheret al.
1981) and V396 Hya (Ruizet al. 2001). These two systems 2 OBSERVATIONS AND DATA REDUCTION
haverelativelylongorbitalperiods(46and65minutes)and
represent, evolutionary speaking, the bulk of the AMCVn 2.1 VLT/UVES observations
systems, which should be long period, low mass-transfer-
We employed the UV-Visual Echelle Spectrograph (UVES)
rateobjects(∼10−12M⊙yr−1).Inthesesystems,themass- mounted on the Unit 2 telescope of the ESO-VLTat Cerro
transfer-rate was thought to be below the threshold for ac-
Paranal,Chile.Thedatawereobtainedinvisitormodedur-
cretiondiscinstabilitiestooccur(Hirose & Osaki1990),and
ingthenightofApril7/8,2002.Theinstrumentwasconfig-
indeednooutbursthassofarbeenreported forGPCom or
uredindichroicmode(dichroic#2)permittingsimultaneous
V396 Hya.
dataacquisitionfromboththeredandthebluecamera.No
GP Com has been observed extensively, and shows er- pre-slit optics were put in the beam, instead atmospheric
ratic flaring in optical, UV (Marsh et al. 1995) and X-ray dispersionwasminimizedbymaintainingthe1′′ wideslitat
(van Teeseling & Verbunt 1994) wavebands, which is at- theparallacticanglethroughoutthenight.Onthebluecam-
tributed to accretion. One of the more intriguing spectral era,a2048x4096pixelEEVdetectorcovered3826–5053˚Ain
features is the presence of sharp, low-velocity components 29 echelle orders. The chip was read out using 1 port and
in the optical helium emission lines. The fact that these 2x3 on-chip binning to reduce read-out time to 22s. In this
‘central-spikes’ contribute to the flare spectrum and fol- mode,abinnedpixelcorrespondsto4.5km/salongthedis-
low a low-amplitude radial velocity curve as a function of persionaxisandaslit-limitedresolutionprofileof1.7pixels.
the orbital phase suggests an origin on, or near, the white The red camera detector system consists of two 2048x4096
UVES andX-Shooterspectroscopyofthe emissionlineAMCVnsystems GP ComandV396 Hya 3
f (mJy)ν23 HeI 3187 GP Com HeI 3705HeI 3819 HeI 3888HeI 3964 HeI 4026 HeI 4120 HeI 4388 HeI 4471 HeII H4eI6 845713 HeI 4921 HeI 5015
1 V396 Hya
3200 3400 3600 3800 4000 4200 4400 4600 4800 5000
λ (Å)
4 5
f (mJy)ν23 GP Com HeI 587 HeI 6678 atmos. HeI 7065 HeI 7281 aattmmooss.. sky residuals N I sky residuals N I sky residuals
1
V396 Hya
6000 6500 7000 7500 8000 8500 9000
λ (Å)
0
1
HeI 4471
HeI 10830 Å
2
x
u
zed fl1.5 GP Com mJy)5
mali f (ν
or
n1
V396 Hya GP Com
5
0. 0
V396 Hya
−400 −200 0 200 400 −2000 −1000 0 1000 2000
velocity (km/s) velocity (km/s)
Figure 1. Gaussian smoothed average spectrum of GPCom and V396Hya obtained with VLT/X-Shooter. Helium emission lines are
indicated. The lower left panel shows the central-spike feature with its blue-shifted forbiddencomponent in Hei4471˚A observed with
UVES.ThelowerrightpanelshowsthestrongestheliumlinesintheNIRarmobservedwithX-Shooter.Thenarrowspikesatwavelength
> 7500˚A areresidualsfromthenightsskylineremoval.
pixelCCDdetectors,anEEVdevicecovering5763–7502˚Ain mined for optimal extraction (Horne 1986). Target expo-
24 orders and a red optimized MIT-LL chip covering 7760– sures were then optimally extracted after subtracting the
9462˚A in 15 orders. Both chips were readout in 32s using 1 sky background. In addition, the same profile weights were
port each with 2x3on-chipbinning.Thespectra pixelscale usedtoextractasuitableThArexposureaswellastheme-
corresponds to 3.6 km/s after binning. dianflat-fieldframe.Thewavelengthscalesweredetermined
Frames were first de-biased using a median of 9 bias byfitting4-6orderpolynomialstoreferencelinesintheex-
exposures. In addition, any residual bias contribution was tractedThArspectrum,deliveringzero-pointRMSresiduals
subtractedusingtheoverscanareasofthethreeCCDdetec- of60.6km/s.Theobject orderswereblaze corrected bydi-
tors. The orders were extracted using the echomop echelle visionofasmoothedversionoftheextractedflatfieldspec-
package (Mills et al. 2014). Flat field correction was per- trum.Orderswerethenmergedintoasinglespectrumusing
formed using the median of a series of well-exposed tung- inverse variance weights based on photon statistics to com-
sten flat field exposures obtained during the day. The in- bineoverlappingordersegments.Finally,themergedspectra
dividual orders were traced and profile weights were deter-
4 T. Kupfer et al.
1.21.21.2 NI NI HeI NINI NI 1.11.11.1 GP Com NI NI NI NI 111 NeI 1.51.51.5 GP Com NININI NI NI NI NI
normalized fluxnormalized fluxnormalized flux0.810.810.81 GP Com normalized fluxnormalized fluxnormalized flux0.910.910.91 V396 Hya normalized fluxnormalized fluxnormalized flux0.80.80.8 GP Com normalized fluxnormalized fluxnormalized flux111 V396 Hya
V396 Hya
V396 Hya
0.60.60.6 444000888000 444111000000 444111222000 444111444000 444111666000 0.80.80.8 444222222000 444222444000 0.60.60.6 666444000000 666444000222 666444000444 0.50.50.5 888666888000 888777000000 888777222000
λλλ (((ÅÅÅ))) λλλ (((ÅÅÅ))) λλλ (((ÅÅÅ))) λλλ (((ÅÅÅ)))
Figure 2.ZoomedregionoftheaverageX-ShooterspectraaroundlocationsofvariousmetallinesinGPComandV396Hya.
Table 2.EmissionlinepropertiesforGPCom Table 3.EmissionlinepropertiesforV396Hya
UVES X-Shooter UVES X-Shooter
Heline EWrange EWmean EWrange EWmean Heline EWrange EWmean EWrange EWmean
(˚A) (˚A) (˚A) (˚A) (˚A) (˚A) (˚A) (˚A)
3187.745 - - 0.5-1.6 1.0 3187.745 - - 3.4-4.0 3.8
3705.005 - - 2.6-3.8 3.2 3705.005 - - 2.0-2.5 2.2
3819.607 - - 1.7-4.0 3.0 3819.607 - - 5.2-5.7 5.5
3871.791 3871.791
6.2-11.1a 8.6a 10.9-14.9a 12.6a 14.0-27.9a 20.8a 15.9a 15.9a
3888.643 3888.643
3964.730 0.7-2.1 1.4 1.2-2.3 1.8 3964.730 2.8-5.2 3.7 1.9-2.6 2.2
4026.191 4.6-8.7 6.6 5.2-9.5 7.3 4026.191 16.4-25.4 20.9 10.3-11.4 11.0
4387.929 1.1-2.9 1.9 2.9-4.9 3.9 4387.929 4.6-7.7 6.1 3.0-3.5 3.3
4471.502 8.3-14.9 11.0 11.7-20.3 14.8 4471.502 24.4-39.0 31.1 17.5-19.6 18.2
4685.710 4685.710
3.5-6.5b 4.9b 5.9-9.0b 7.2b 8.1-14.4b 11.1b 5.4-6.5b 5.8b
4713.170 4713.170
4921.930 5.0-8.7 6.8 8.1-12.4 10.4 4921.930 11.6-20.1 15.3 8.6-9.8 9.1
5015.678 5015.678
9.9-14.8c 12.4c 17.4-22.2c 20.1c 17.5-29.9c 22.7c 13.2-13.9c 13.6c
5047.738 5047.738
5875.661 30.4-62.7 44.1 46.7-58.8 52.4 5875.661 58.3-101.8 76.4 50.4-53.7 51.5
6678.152 24.8-38.1 30.7 36.9-46.1 42.5 6678.152 45.0-71.7 57.2 32.8-35.9 34.6
7065.251 35.3-51.0 41.5 47.8-56.2 51.5 7065.251 54.2-89.1 70.9 48.6-51.7 50.6
7281.351 11.0-36.5 18.2 16.1-19.4 18.2 7281.351 19.3-38.0 24.5 12.6-13.7 13.1
10830.34 - - 581.0-736.6 666.9 10830.34 - - 922.3-982.6 969.5
12784.79 - - d d 12784.79 - - d d
17002.47 - - d d 17002.47 - - d d
20586.92 - - d d 20586.92 - - d d
a Combinedequivalent widthofHei3871andHei3888 a CombinedequivalentwidthofHei3871andHei3888
b CombinedequivalentwidthofHeii4685andHei4713 b CombinedequivalentwidthofHeii4685andHei4713
c CombinedequivalentwidthofHei5015andHei5047 c Combinedequivalent widthofHei5015andHei5047
d Linepresentbutcontaminated withatmosphere. d Linepresentbutcontaminated withatmosphere.
2.2 VLT/X-Shooter observations
GPCom and V396Hya were also observed using the
medium resolution spectrograph X-Shooter (Vernet et al.
were flux calibrated using wide slit exposures of the B-star 2011) mounted at the Cassegrain focus on the Unit 2 tele-
fluxstandard HD60753. scopeoftheVLTarrayinParanal,Chile,duringthenightof
Fortheredarmdata,wealsoperformedacorrectionfor 2010-02-18 as part of the Dutch GTO program. X-Shooter
thetelluricabsorptionfeatures.Spectrawerefirstalignedby consistsof3independentarmsthatgivesimultaneousspec-
cross-correlating with the telluric standard. This corrected tra longward of theatmospheric cutoff (0.3 microns) in the
smallwavelengthshiftsandcorrectionswerefoundtobeless UV(the‘UVB’arm), optical (the‘VIS’arm) and upto2.5
than 3 km/s. The telluric features in each spectrum were microns in the near-infrared (the ‘NIR’ arm). We used slit
then removed as far as possible by adjusting the depth of widthsof1.0′′,0.9′′ and0.9′′ inX-Shooter’sthreearmsand
thetelluric template. binnedby2x2intheUVBandVISarmsresultinginvelocity
UVES andX-Shooterspectroscopyofthe emissionlineAMCVnsystems GP ComandV396 Hya 5
Heiλ3888 Heiλ5015 Neiλ6402 Neiλ6506 Niλ7423 Niλ7468
2
1.8
1.6
1.4
1.2
e
s
a 1
h
P
0.8
0.6
0.4
0.2
0
-100 -50 0 50 100-100 -50 0 50 100-100 -50 0 50 100-100 -50 0 50 100-100 -50 0 50 100-100 -50 0 50 100
Velocity(km/s) Velocity(km/s) Velocity(km/s) Velocity(km/s) Velocity(km/s) Velocity(km/s)
Figure3.Tracingofthecentral-spikeofselectedheliumlinesandsomeneonandnitrogenemissionlinesofGPComfromUVESdata.
Table 4.Velocityofthecentral-spike ber of prominent lines in all spectra. In Table 2 and 3 we
list the ranges of EWs observed together with their mean
value. Large intrinsic EW modulations are detected in all
GPCom V396Hya linesthroughoutourobservingrun.InthecaseofGPCom,
HeIline γ K1 γ K1 the measured EWs from UVES and X-Shooter are consis-
(˚A) (kms−1) (kms−1) (kms−1) (kms−1) tently smaller than previously reported values. In compar-
ison with Marsh et al. (1991), we find that our mearured
3888.643 –4.7±0.2 12.7±0.3 –9.9±0.3 6.1±0.4 EWs are smaller by a factor between 1.4 for Hei7281˚A
3964.730 –4.1±1.1 13.0±1.6 –2.5±1.2 5.8±1.6
and 3.3 for Hei4387˚A respectively. Thus, on top of short
4387.929 14.5±2.3 11.5±3.2 20.1±3.4 8.6±4.2
4471.502 42.4±0.5 11.1±0.7 39.4±0.6 5.4±0.9 timescale flaring activity, GP Com also displays significant
4685.710 17.4±0.3 11.7±0.5 16.1±0.5 5.2±0.6 variability on longer timescales. V396Hya shows very simi-
4713.170 32.6±0.3 11.1±0.5 27.3±0.5 4.3±0.6 larvariations,withEWvariationsclosetoafactoroftwofor
4921.930 52.6±1.0 12.3±1.6 47.8±0.8 5.8±1.1 UVES.Wecalculate flarespectraaccording tothemethods
5015.678 6.4±0.3 12.3±0.4 7.5±0.4 6.0±0.5 described in Marsh et al. (1995) and find that in both GP
5875.661 0.4±0.3 11.3±0.3 2.5±0.3 5.1±0.5 Com and V396 Hya the continuum as well as the disc and
6678.152 18.2±0.2 11.6±0.2 11.2±0.3 6.8±0.3 spike components of the lines contribute to the flaring. As
7065.251 23.2±0.2 13.3±0.3 16.0±0.3 6.2±0.4
discussedinMarsh(1999)andMorales-Rueda et al.(2003),
7281.351 17.2±0.2 8.0±0.4 16.3±0.4 4.1±0.5
this strongly suggests that the spike originates at or near
Mean 11.7±0.3 5.8±0.5 the accreting dwarf and not in an extended nebula in both
of these objects.
Despite the detailed similarities between the profile
resolutions of14kms−1,7kms−1 and11kms−1 perbinned shapes in GP Com and V396 Hya, a few differences can
pixel for the UVB, VIS and NIR arm respectively. The re- be identified in the average spectra. GP Com displays a
duction of the raw frames was conducted using the stan- number of sharp absorption features, most notably be-
dard pipeline release of Reflex (version 2.3) for X-Shooter tween 4050˚Aand 4250˚A. We identify these lines as Ni.
data(Freudlinget al.2013).Thestandardrecipeswereused No such features are present in V396 Hya (Fig.2). A
to optimally extract and wavelength calibrate each spec- large number of Nii absorption lines have been observed
trum.Theinstrumentalresponsewasremovedbyobserving in the high state system SDSSJ1908 (Kupferet al. 2015).
thespectrophotometricstandardstarEG274(Hamuy et al. Morales-Rueda et al.(2003)alreadyreportedanabsorption
1992, 1994) and dividing it by a flux table of the same star feature near Heii 4686˚Ain GPCom. Indeed this feature is
to producethe response function. also present in our X-Shooter and UVES spectra but its
Table1 gives an overview of all observations and the origin remains unidentified.
instrumental set-ups. In the blue spectral region the strongest Nii lines are
detected in absorption whereas in the red spectral region,
complexNiemissiondominatesthespectrum.Alargenum-
3 ANALYSIS berofindividualtransitionsofNicanbeidentifiedthanksto
our high spectral resolution (Fig.2). The equivalent widths
3.1 Average spectra oftheseblendsofNiarelargerinV396HyathaninGPCom.
Wealso identify anumberof weak Nei emission featuresin
We present the time averaged spectra of GP Com and
the spectra of both systems (see Fig.2 and 3). See Tab.A1
V396 Hya in Fig.1, illustrating the striking similarity be-
andA2foran overviewofthedetectedlineswith measured
tween thespectra ofthesetwo systems.Thefamiliar triple-
equivalent widths.
peaked helium emission lines dominate the spectra. The
high-resolution spectra obtained with UVES nicely resolve The X-Shooter spectra allow us to search for spec-
the central-spike feature in both systems (lower left panel troscopic signatures in the near-infrared part of the spec-
Fig.1).Wemeasuredtheequivalentwidths(EW)foranum- trum.WefindastrongHei10830˚A linewith anequivalent
6 T. Kupfer et al.
30 50
NeI 640(cid:23)A2
25
45
20
15
40
s) 10 s)
km/ 5 km/ 35
( (
V 0 V
R R
30
-5
-10 25
-15
HeI3888(cid:23)A
-20 20
0 0.5 1 1.5 2 0 0.5 1 1.5 2
Phase Phase
Figure 4. Radial velocity curves of the central-spike of Hei Figure 6. Radial velocity curve with measured velocities of the
3888˚A and the emission line Nei 6402˚A in GPCom. Two or- nitrogen absorption line Ni 4143˚Ain GPCom. Two orbits are
bitsareplottedforbetter visualization. plottedforbettervisualization.
3.2 The orbital ephemeris
20
Marsh(1999)andMorales-Rueda et al.(2003)havedemon-
15 NI 746(cid:23)A8 strated that the central-spike in GP Com shows significant
10 radialvelocityshiftsasafunctionoftheorbitalphase,with
anamplitudeof∼10−15kms−1,compatiblewithitslikely
s) 5 origin close to the accretor.
/
m We perform multi-Gaussian fits to the individual line
k 0
( profiles, modelling the line as a combination of two broad
V
R -5 Gaussiansrepresentingthedouble-peakeddiscemissionplus
a narrow Gaussian for the spike. As a first step, we fit the
-10
velocitiesofthecentral-spiketotheindividualspectra.The
derivedradialvelocitycurvesarethenusedtodeterminethe
-15
HeI3888(cid:23)A orbital phasing of thespike.
-20 Although the UVES data and X-Shooter data were
0 0.5 1 1.5 2
taken eight years apart the orbital period could not be re-
Phase
finedbecausetheX-Shooterdataonlycoveraboutoneorbit
Figure 5. Radial velocity curves of the central-spike of Hei of GPCom. Therefore, for GPCom, we fix the orbital pe-
3888˚A and the emission line Ni 7468˚A in V396Hya. Two or- riod to 46.57 minutes (Marsh 1999). If the radial velocity
bitsareplottedforbetter visualization.
curve of the spike traces the white dwarf accretor, we can
define orbital phase zero as the phase of superior conjunc-
tion of the accretor (or the red to blue crossing point of
widthof666.9˚A and969.5˚A,whichismorethantentimes the accretor’s radial velocity curve). Individual lines give
stronger than the strongest optical helium lines. Addition- identicalzero-pointsforthephasing,wethusfittothethree
allywefindtheHeilines11969,12784,17002and20586˚A. strongestheliumlinessimultaneously,andderivethefollow-
None of the helium lines in the near-infrared spectrum of ing ephemeris for GPCom taking either the UVES or the
GPCom and V396Hya show a sharp central-spike feature X-Shooterdata:
(lower right panel Fig.1), a somewhat surprising difference
HJDGPCom;UVES=2452372.5994(2)+0.0323386E
to the optical regime. A possible explanation could be that
thedouble-peakedheliumdisclinesinthenear-infraredform HJDGPCom;X-Shooter =2455245.851(2)+0.0323386E
atalowertemperaturecomparedtothecentral-spike.Inthis
In the case of V396Hya, the orbital period is not that
casethenear-infraredheliumdisclinesbecomestrongerrel-
wellconstrained,buttheshort durationofourobservations
ativetothecentral-spikefeaturesandpossiblyoutshinethe
prevents an improvement on the value of 65.1 minutes as
central-spikes in the near-infrared.
derivedbyRuizet al.(2001).TheX-Shooterdatacouldnot
The peak-to-peak velocity of the double-peaked disc
beusedbecausetheexposuretimeoftheindividualspectra
emission line Hei10830˚A was found to be 1237±7kms−1
was 15min which covers a significant fraction of the orbit.
forGPComand1069±6kms−1forV493Hyawhichissignif-
We again use the radial velocity curve of the spike in the
icantly lower than the 1379-1414kms−1 range for GPCom
three strongest lines to derive the following ephemeris for
andthe1111-1124kms−1rangeforV493Hyausingthelines
V396 Hya
in the optical regime, showing an emission profile that is
more centered in theouter disc. HJDV396Hya;UVES=2452372.5263(3)+0.0452083E
UVES andX-Shooterspectroscopyofthe emissionlineAMCVnsystems GP ComandV396 Hya 7
❍✕✐✖✸✗✗✗ ❍✕✐✖✹✹✜✢ ❍✕✐✖✹✣✤✢ ❍✕✐✖✥✦✢✥
✷
✶(cid:0)✁
✶(cid:0)✂
✶(cid:0)✄
✛ ✶(cid:0)☎
✚
✘✙ ✆
P ✵(cid:0)✁
✵(cid:0)✂
✵(cid:0)✄
✵(cid:0)☎
✝
✏✑✑✑
❇
❆
❅
❄ ✺✠✠
❃
❂
✿❀❁ ✝
✾
✼✽
✻ ✲☛☞☞
✳✴
❨
✒✓✔✔✔
✌✍✎✎✎✲☛☞☞ ✡ ✺✠✠✞✟✟✟ ✌✍✎✎✎✲☛☞☞ ✡ ✺✠✠✞✟✟✟ ✌✍✎✎✎✲☛☞☞ ✡ ✺✠✠✞✟✟✟ ✌✍✎✎✎✲☛☞☞ ✡ ✺✠✠✞✟✟✟
❳✧★❡❧✩❝✪✫②✭✬✮✯✰✱ ❳✧★❡❧✩❝✪✫②✭✬✮✯✰✱ ❳✧★❡❧✩❝✪✫②✭✬✮✯✰✱ ❳✧★❡❧✩❝✪✫②✭✬✮✯✰✱
❍✗✐✘✙✚✛✙ ❍✗✐✘✻✻✛✚ ❍✗✐✘✛✼✻✙ ✥✦✧★✩✪✫✬✪
✷
✶(cid:0)✁
✶(cid:0)✂
✶(cid:0)✄
✤ ✶(cid:0)☎
✣
✜✢ ✆
P ✵(cid:0)✁
✵(cid:0)✂
✵(cid:0)✄
✵(cid:0)☎
✝
✏✑✑✑
■
●
❋
❊ ✺✠✠
❉
❈
❅❆❇ ✝
❄
❂❃
❁ ✲☛☞☞
✿❀
❨
✒✓✔✔✔
✌✍✎✎✎✲☛☞☞ ✡ ✺✠✠✞✟✟✟ ✒✓✔✔✔✲☛☞☞ ✡ ✕✖✖✞✟✟✟ ✒✓✔✔✔✲☛☞☞ ✡ ✕✖✖✞✟✟✟ ✒✓✔✔✔✲☛☞☞ ✡ ✕✖✖✞✟✟✟
❳✭✮❡❧✯❝✰✱②✳✴✸✹✽✾ ❳✭✮❡❧✯❝✰✱②✳✴✸✹✽✾ ❳✭✮❡❧✯❝✰✱②✳✴✸✹✽✾ ❳✭✮❡❧✯❝✰✱②✳✴✸✹✽✾
Figure 7. Trailedspectra and maximum-entropy Doppler tomograms of selected Hei lines of GPCom obtained from X-Shooter data.
The disc,the central-spikeas well as two brightspots arevisible. Note that inthe plotting of somelines the displayed intensity of the
central-spikewassaturatedtoemphasizebothbrightspots.
Armedwiththeorbitalephemerides,wefoldallspectra malerrors.Table4liststhepropertiesofthespikeinallthe
inordertoobtainhighersignal-to-noise spectra,permitting heliumlines forwhich areliable fitcouldbemade,for both
ustoderivemoreaccuratevaluesfortheradialvelocitycurve ourtargets.
of the various emission line components. All orbital phases We find that all 12 usable helium lines covered by the
reported in thispaper are based on theabove ephemerides. UVES spectra are consistent with the same radial veloc-
ity amplitude for the central-spike component and thus we
calculate a weighted mean from the 12 fitted amplitudes
to derive K = 11.7 ± 0.3kms−1 for GP Com and
3.3 The radial velocity of the central-spike and spike
K = 5.8±0.3kms−1 for V396 Hya. The fact that all
the metal lines spike
lines share the same phasing and velocity amplitudes pro-
The final radial velocity curves of the central-spike com- vides strong support for placing the origin of the spike at
ponents are measured by fitting a single Gaussian to the orneartheprimary white dwarf, thusprovidinguswith an
central-spikes in the phase-folded spectra. A pure sinusoid accuratedeterminationofitsradial velocityamplitude,K1.
wasfittedtotheradialvelocitiestodeterminethemeanand However,themeanvelocitiesofthecentral-spikeshowvari-
amplitude of the orbital motion of the spike and their for- ations from line to line, from −5 to +53kms−1. The seem-
8 T. Kupfer et al.
❍✕✐✖✸✗✗✗ ❍✕✐✖✜✢✣✜ ❍✕✐✖✻✻✤✗ ❍✕✐✖✤✢✻✜
✷
✶(cid:0)✁
✶(cid:0)✂
✶(cid:0)✄
✛ ✶(cid:0)☎
✚
✘✙ ✆
P ✵(cid:0)✁
✵(cid:0)✂
✵(cid:0)✄
✵(cid:0)☎
✝
✏✑✑✑
❅
❄
❃
❂ ✺✠✠
❁
❀
✽✾✿ ✝
✼
✴✹
✳ ✲☛☞☞
✰✱
❨
✒✓✔✔✔
✌✍✎✎✎✲☛☞☞ ✡ ✺✠✠✞✟✟✟ ✌✍✎✎✎✲☛☞☞ ✡ ✺✠✠✞✟✟✟ ✌✍✎✎✎✲☛☞☞ ✡ ✺✠✠✞✟✟✟ ✌✍✎✎✎✲☛☞☞ ✡ ✺✠✠✞✟✟✟
❳✥✦❡❧✧❝★✩②✭✪✫✬✮✯ ❳✥✦❡❧✧❝★✩②✭✪✫✬✮✯ ❳✥✦❡❧✧❝★✩②✭✪✫✬✮✯ ❳✥✦❡❧✧❝★✩②✭✪✫✬✮✯
Figure 8.Trailedspectraandmaximum-entropyDopplertomogramsofselectedHeilinesofV396HyaobtainedfromtheUVESdata.
Thedisc,thecentral-spikeaswellasbothbrightspotsarevisible.Notethatinsomelinesthecentral-spikewassaturatedtoemphasize
bothbrightspots.
ingly random systemic velocity shifts reported by Marsh trajectories of the accretion stream to the observed bright
(1999) and Morales-Rueda et al. (2003) are thus confirmed spot velocities in a Doppler tomogram, as has been done
inourhigh-resolutiondataandextendedtoalargernumber for several of the AMCVn stars that show a central-spike
oflines.Theshiftsstronglycorrelatebetweenthesamelines (Roelofs et al. 2005, 2006).
in both stars. Fig.7 and 8 show maximum-entropy Doppler tomo-
Notonlythecentral-spikefeatureshowsradialvelocity grams of GP Com and V396 Hya. The central-spikes are
variations. We find that the narrow emission lines of neon fixedtothenegativeY-velocityaxis,theconventionalphase
and nitrogen, as well as the absorption lines in GPCom, of the accreting white dwarf. A strong bright spot shows
tracethemotionofthecentral-spike.Figure3showsatrailed up near the expected accretion stream/disc impact region,
spectrogramofthecenteroftwostrongheliumemissionlines with a faint secondary bright spot at ∼120-degree phase-
aswellasthetwostrongestnitrogenandneonemissionlines offset, which has also been observed in other AMCVn-type
in GP Com. A sinusoidal fit was made to the radial veloc- systems(Roelofs et al.2005;Kupfer et al.2013).Weassume
ities of the metal lines. The radial velocity amplitudes and that the strong bright spot corresponds to the first impact
phases are consistent with those of the central-spike fea- pointoftheaccretionstreamandthedisc,whiletheweaker
ture in the helium lines. For the co-added neon and nitro- bright spots may represent accretion stream overflow and
gen emission lines, we derivea mean amplitude Kemission = re-impact further downstream. The latter effect has been
12.3±0.5kms−1 andameanvelocityγ =11.9±0.9kms−1 seen in numerical simulations of accretion discs (M. Wood,
for GPCom (Fig.4) and Kemission = 5.6±0.8kms−1 and in preparation; privatecommunication).
γ =10.2±1.6kms−1 for V396Hya(Fig.5). Figure9 shows the allowed mass ratios and accretion
The same approach was used for the Ni 4143˚A ab- disc radii for GP Com and V396 Hya that we obtain from
sorption line in GPCom. We find a radial velocity ampli- thephasesandamplitudesofthecentral-spikesandthepri-
tude of K4143 = 8.5±2.1kms−1 with a mean velocity of mary bright spot. We solve the equation of motion for a
γ =35.5±2.5kms−1.Theradial velocityamplitudeiscon- free-falling stream of matter through the inner Lagrangian
sistent with the amplitude obtained for the central-spikes point, based on theresults of Lubow & Shu(1975),and we
(Fig.6).Therefore,weconcludethatthemetalemissionand seewhethertheresultingaccretion streamand/oraccretion
absorption lines are linked to the accreting white dwarf as disc velocities and phases at the stream-disc impact point
well. match with the measured values. The bright spot in inter-
actingbinaries isnot always observed torepresentthepure
ballistic stream velocities at the stream-disc impact point;
3.4 System parameters via Doppler tomography
thegasvelocitiescouldinprinciplelieanywherebetweenthe
Doppler tomography (Marsh & Horne 1988) of GPCom ballisticstreamvelocitiesandtheaccretiondiscvelocitiesin
has been used in the past to study the properties of thedisc-streamimpactregion.Wethereforeconsiderthetwo
its accretion disc and bright spot emission (Marsh 1999; limiting cases of pure ballistic stream and pure Keplerian
Morales-Rueda et al. 2003). Armed with an accurate esti- disc velocities in the bright spot. The mass ratio and ac-
mate for the primary radial velocity K1, we can constrain cretion disc radii (i.e., where thebright spot occurs) ranges
the mass ratios by comparing gas velocities along ballistic we obtain in this way for GPCom are 0.015 < q < 0.022
UVES andX-Shooterspectroscopyofthe emissionlineAMCVnsystems GP ComandV396 Hya 9
Figure9.AllowedmassratiosqandeffectiveaccretiondiscradiiRforGPCom(top)andV396Hya(bottom).Theleftpanelsassume
ballisticaccretionstreamvelocitiesinthebrightspot,therightpanelsassumeKepleriandiscvelocities.Sincethebrightspotcouldbea
mixofthese,weshowinthecenterpanelstheresultsforamixof80%ballisticstreamvelocitiesand20%Kepleriandiscvelocities.The
upper andlower dotted linesindicate theedge of the primaryRoche lobeand the circularizationradius,respectively, whilethe dashed
lineshowsthe3:1resonanceradius.Thegray-scaleindicatestheexclusionlevelinstandarddeviations.
and 0.60 < R/RL1 < 0.80 assuming pure ballistic stream Table5.Systemparametereitherassumingpureballisticstream
velocity and 0.021 < q < 0.028 and 0.45 < R/RL1 < 0.58 velocitiespureKepleriandiscvelocities.
assumingpureKepleriandiscvelocity.ForV396Hyawefind
0.010 < q < 0.016 and 0.54 < R/RL1 < 0.84 assuming pureballisticstream pureKepleriandisc
pure ballistic stream velocity and 0.013 < q < 0.018 and
0.50 < R/RL1 < 0.62 assuming pure Keplerian disc veloc- GPCom
ity. RL1 corresponds to the distance from the center of the q 0.020-0.022 0.024-0.028
accreting white dwarf to theinner Lagrangian point. M1 (M⊙) >0.54 >0.33
While these mass ratios and disc radii lie in the range M2 (MJupiter) 12.5-33.8 9.6-42.8
where superhumps might be expected due to the 3 : 1 res- iR(/◦R)L1 405.68-7-40.80 303.45-7-80.58
onance, numerical simulations indicate that the mass ra-
V396Hya
tios are in fact so low that the (eccentric) accretion disc
q 0.010-0.016 0.013-0.018
remains stationary in the binary frame (Simpson & Wood
M1 (M⊙) >0.37 >0.32
1998). This matches the absence of any reports of ‘super- M2 (MJupiter) 6.1-26.8 6.1-30.5
humps’in GP Com and V396 Hya. R/RL1 0.54-0.84 0.50-0.62
With the derived mass ratios and the primary veloc- i(◦) 29-75 25-79
ity amplitudeK1 as well as themeasured disc velocities we
can limit theinclination angles, the component masses and
the disc sizes. Above a certain inclination angle the donor
will start to eclipse the outer edges of the disc and, at high For a given mass ratio and donor mass we calculate
enough inclination, also the accretor. No eclipses are ob- the expected primary velocity amplitudes. The ratio of the
served in both systemseither in thelines norin thecontin- calculated velocity amplitudes and the measured velocity
uum. amplitudesdependsonlyontheinclinationangle.Addition-
10 T. Kupfer et al.
80
80
75
70 70
) )
◦ ◦
( 65 (
n n 60
o o
ti 60 ti
a a
n n
cli 55 cli 50
n n
i i
50
40
45
40 30
10 20 30 10 20 30 40 50
M (M ) M (M )
2 Jupiter 2 Jupiter
80 80
70
70
) )
◦ ◦
( ( 60
n 60 n
o o
ti ti
a a 50
n 50 n
cli cli
n n 40
i i
40
30
30
20
5 10 15 20 25 30 10 20 30
M (M ) M (M )
2 Jupiter 2 Jupiter
Figure 10.Calculated inclinationangles fordifferent donor masses forGPCom (upper panels) and V396Hya (lower panels). Theleft
vertical dashed linemarksthe zero temperature massfor thedonor star andthe curved dashed linemarks the limitwhenthe accretor
reaches the Chandrasekhar limit. The grey shaded area marks the region where the system would show eclipses assuming the largest
possible disc radius. The dotted line corresponds to the largest disc radius and the dashed-dotted line to the largest mass ratio. The
hatched area shows the allowed parameter range. Upper left: Assuming pure ballistic stream velocities for GP Com. Upper right:
AssumingpureKepleriandiscvelocitiesforGPCom.Lowerleft:AssumingpureballisticstreamvelocitiesforV396Hya.Lowerright:
AssumingpureKepleriandiscvelocitiesforV396Hya.
ally,themeasuredpeak-to-peak(Keplerian)velocitiesinthe massratioswouldshiftthedasheddottedlinestotheright.
double-peaked lines correlate to the velocity in the outer Hence, the largest mass ratio is a lower limit. Combining
disc,andalsodependontheinclinationinasimilarfashion. these constraints leaves a small region of parameter space
(Horne& Marsh 1986). that is allowed.
Fig.10showsthecalculatedinclinationanglesfordiffer- Theupperpanelsin Fig.10 show theallowed ranges in
entdonormassesforGPCom(upperpanels)andV396Hya inclinationforagivendonormassinGPComandthelower
(lowerpanels) assumingeitherpureballistic streamorpure panels in Fig.10 for V396Hya.Table5 gives an overview of
Keplerianvelocityofthebrightspot.Asalowerlimittothe thederivedsystemparameterseitherassumingpureballistic
mass of the donor, the mass for a Roche lobe filling zero- stream velocities or pureKeplerian disc velocities.
temperatureheliumobjectiscalculated(Deloye et al.2007).
The upper limit for the donor masses are set by the Chan-
drasekhar mass of the primary. The dotted line in Fig.10 4 DISCUSSION
corresponds to the largest allowed disc radius (0.80R/RL1 4.1 Stark broadening and the behaviour of the
for pure ballistic stream velocities and 0.58R/RL1 assum- central emission spikes
ing pure Keplerian disc velocity in GPCom). The largest
allowed disc radius is an upper limit. The dashed-dotted The forbidden components of neutral helium that are ob-
lineinFig.10correspondstothelargest possiblemassratio servedinsomeofthecentral-spikes(Fig.11)haveledtothe
(q=0.022 for pureballistic stream velocities and q=0.028 suggestionthatthecentral-spikeprofilesmaybeaffectedby
assumingpureKepleriandiscvelocitiesinGPCom).Smaller Stark broadening (Morales-Rueda et al. 2003), as modelled