arXiv:0802.2022v1 [astro-ph] 14 Feb 2008
A Far Ultraviolet Archival Study of Cataclysmic Variables: I. FUSE and
HST/STIS Spectra of the Exposed White Dwarf in Dwarf Nova Systems.
Patrick Godon2, Edward M. Sion
Department of Astronomy and Astrophysics Villanova University, Villanova, PA 19085,
Paul E. Barrett
United States Naval Observatory, Washington, DC 20392 email@example.com
Department of Astronomy, University of Arizona, AZ, firstname.lastname@example.org
Albert P. Linnell, Paula Szkody
Department of Astronomy, University of Washington, Seattle, WA 98195,
We present a synthetic spectral analysis of Far Ultraviolet Spectroscopic Ex-
plorer (FUSE) and Hubble Space Telescope/Space Telescope Imaging Spectrograph
(HST/STIS) spectra of 5 dwarf novae above and below the period gap during qui-
escence. We use our synthetic spectral code, including options for the treatment of
the hydrogen quasi-molecular satellite lines (for low temperature stellar atmospheres),
NLTE approximation (for high temperature stellar atmospheres), and for one system
(RU Peg) we model the interstellar medium (ISM) molecular and atomic hydrogen lines.
In all the systems presented here the FUV flux continuum is due to the WD. These
spectra also exhibit some broad emission lines. In this work we confirm some of the
previous FUV analysis results but we also present new results. For 4 systems we com-
bine the FUSE and STIS spectra to cover a larger wavelength range and to improve
the spectral fit. This work is part of our broader HST archival research program, in
which we aim to provide accurate system parameters for cataclysmic variables above
and below the period gap by combining FUSE and HST FUV spectra.
Subject headings: Stars: white Dwarfs, Stars: dwarf novae (SS Aur, EY Cyg, VW Hyi,
RU Peg, EK TrA).
1Based on observations made with the NASA-CNES-CSA Far Ultraviolet Spectroscopic Explorer. FUSE is oper-
ated for NASA by the Johns Hopkins University under NASA contract NAS5-32985
2Visiting at the Space Telescope Science Institute, Baltimore, MD 21218, email@example.com
– 2 –
1.1.Dwarf Novae in Cataclysmic Variables
Cataclysmic variables (CVs) are short-period, semi-detached compact binaries in which the
primary star, a white dwarf (WD), accretes matter and angular momentum from the secondary, a
main-sequence star filling its Roche lobe (the mass donor) (Warner 1995). The matter is transferred
by means of either an accretion disk around the WD, or an accretion column or curtain - when the
WD has a strong magnetic field. Ongoing accretion at a low rate (quiescence) can be interrupted
every few weeks to months by intense accretion (outburst) of days to weeks (a dwarf nova accretion
event), and every few thousand years by a thermonuclear runaway explosion (TNR - the classical
nova event, due to the ignition of the accreted layer of hydrogen-rich material). CVs are divided in
sub-classes according to the duration, occurrence and amplitude of their outbursts. The two main
types of CVs are Dwarf Nova systems (DNs; weakly- or non-magnetic disk systems found mostly
in their quiescent state which lasts much longer than their outburst), and nova-like (NL) systems
which form a less homogeneous class. NLs includes disk systems found mostly in their high state,
intermediate polars (IPs; with a magnetically truncated inner disk), polars (devoid of disk due to
the strong magnetic field of the WD), and other systems e.g. that never go into outburst or which
cannot be classified as DNs (Warner 1995) (this includes the helium systems - AM CVn - with the
shortest period). Dwarf nova systems includes the U Gem systems, the SU UMa systems, and the
Z Cam systems. The U Gem systems are the typical DNs, i.e. those systems exhibiting normal DN
outbursts; the SU UMas exhibit both normal DN outbursts, and superoutbursts, which are both
longer in duration and higher in luminosity than normal DN outbursts; and the Z Cam systems
have standstills where they remain in a state of intermediate optical brightness for a long time
(here we adopt the classification according to Ritter & Kolb (2003)).
Interestingly enough, while the binary orbital period in CV systems ranges from a fraction of an
hour (AM CVn systems) to about a day (e.g. GK Per), there is a gap in the orbital period between
2 and 3 hours where almost no systems are found (hereafter the “period gap”). For example U
Gem and Z Cam DN systems are found above the period gap, while the SU UMa DN systems are
found below the period gap. It is not known whether the systems are evolving from a longer period
to a shorter period (across the gap) or whether the systems above the gap are altogether different
from the systems below the gap.
The now widely accepted interpretation of the quiescence/outburst cycle is the disk instability
model (DIM, Cannizzo (1998)). It is assumed that during the quiescent phase the matter in the
disk is cold and neutral and the disk is optically thin because of its low density, while during
outburst the matter in the disk is ionized and becomes optically thick as the mass accretion rate
within the disk increases. The basic principle of the DIM theory depends heavily on the unknown
viscosity parameter α (Shakura & Sunyaev 1973) and on the mass transfer rate during the different
phases. DNs, unlike other CVs, offer a fairly reliable estimate of their distances via the absolute
magnitude at maximum versus orbital period relation for DNs found by Warner (1995). This
– 3 –
relationship is consistent with theory (Cannizzo 1998). The mass accretion rate within the disk
has been taken from Patterson (1984), which is, however, only a first order estimate. In the last
decade, the disk mass accretion rate of many systems has been deduced more accurately at given
epochs of outburst or quiescence using spectral fitting techniques. The accretion rate is usually
a function of time (especially during the outburst itself; and it is also a function of radius r due
to wind-outflow from the disk) and, consequently, it is difficult to assess its time-averaged value
Recent advances in theory (Townsley & Bildsten 2004) have shown that the average mass ac-
cretion rate onto the WD in DNs can be deduced if one knows the mass of the accreting WD and its
effective surface temperature during quiescence, therefore providing an additional and independent
way to assess ˙M (or more precisely ˙M(r = R∗) - the mass accretion rate at one stellar radius R∗
onto the stellar surface, which might be different than the mass accretion rate in the disk
at a radius r if there is an outflow from the disk: ∂˙ M/∂r < 0. ). Consequently, in order to put
more constraints on the theories we need to know the properties (mainly the temperature and mass
of the WD) of these systems above as well as below the period gap. There is, however, a critical
shortage in knowledge of the WD properties (effective temperature Twd, gravity Log(g), projected
rotational velocity Vrotsin(i), chemical abundances, accretion belts?) in DNs above the period gap.
Thus, detailed comparisons of accreting WDs above and below the gap cannot be made.
For systems below the gap, with orbital periods near the period minimum (Warner 1995),
the distribution of temperatures are centered at ∼15,000K with only a narrow range identified at
present. This distribution appears to manifest the effect of long term compressional heating at a
time averaged accretion rate of 2 × 10−11M⊙yr−1(Townsley and Bildsten 2002; Sion et al. 2003).
It appears that WD Teff’s for systems above the gap are higher than WD temperatures in systems
below the gap, due to the systems above the gap having larger disks (with higher mass transfer
rates) and more massive (somewhat earlier-type) secondaries. Some disks may remain optically
thick even during quiescence so that the WDs are heated to a greater extent than systems below
the gap. It is not yet known whether the WDs in systems above the gap are rotating more slowly
than WDs in systems below the gap where presumably the CVs are older and with a longer history
of angular momentum transfer via disk accretion. Thus far the only DNs above the gap whose
WDs and disks/boundary layers have been analyzed with FUSE, IUE and HST have been Z Cam
(Hartley et al. 2005), RX And (Sion et al. 2001; Sepinsky et al. 2002), U Gem (Sion et al. 1998;
Long & Gilliland 1999; Froning et al. 2001), SS Aur (Sion et al. 2004a), EY Cyg (Sion et al.
2004b), RU Peg (Sion & Urban 2002; Sion et al. 2004a) and WW Cet (Godon et al. 2006b); a
total of 2 Z Cam systems, 4 U Gem systems and the peculiar DN WW Ceti (although classified as
a U Gem, others suggest it is a Z Cam).
Under our broader HST archival research we have started to secure accurate system parameters
(˙M, i, Mwd, Teff, Vrotsin(i), chemical abundances,etc..)
below the period gap by fitting synthetic spectral models (WDs and accretion disks) to combined
FUSE+HST spectra from the MAST archive. We have identified 25 CV systems for which the
for CVs (DNs and NLs) above and
– 4 –
FUSE and HST (STIS, FOS or GHRS) spectra match and can be combined. As a part of this HST
archival research, we present in this paper the analysis of HST/STIS and (FUSE) archival spectra
of 5 DNs during quiescence using synthetic stellar spectra together with the FUSE spectrum of RU
1.2.Five Dwarf Nova Systems
The five DNs are listed in Table 1 with their system parameters as follows: column (1) Name,
(2) CV subtype, (3) reddening value E(B-V), (4) distance, (5) orbital period in days, (6) orbital
inclination in degrees, (7) spectral type of the secondary, (8) mass of the primary in solar masses,
(9) mass of the secondary in solar masses, (10) apparent magnitude in outburst, and (11) apparent
magnitude in quiescence. The references are listed below the table. Three systems are above the
period gap: EY Cyg, SS Aur and RU Peg; and two systems are below the period gap: VW Hyi
and EK TrA.
Among these 5 objects, 4 DNs in quiescence are directly from our HST archival program with
matching FUSE and HST/STIS spectra: EY CYG, SS Aur, VW Hyi, and EK TrA. We supple-
ment this study here with the analysis of RU Peg’s FUSE spectrum. For SS Aur, VW Hyi and
EK TrA we assess the WD parameters by modeling for the first time for these 3 objects the com-
bined FUSE+STIS spectra. For RU Peg’s FUSE spectrum, and EY Cyg’s combined FUSE+STIS
spectrum we improve the fittings that were carried out previously.
For all the spectra we present here, we use the latest versions of the stellar model atmospheres
and synthetic spectra codes (see section 3); this includes options for the treatment of the hydrogen
molecular satellite lines (for modeling cooler WDs), and NLTE atmosphere models (for modeling
hotter WDs). We also adopt the extinction values E(B-V) given in Bruch & Engel (1994) for
the 5 objects and we deredden the spectra accordingly. For some systems we identify the ISM
molecular absorption lines in the FUSE spectra; this greatly helps improve the fitting of the chemical
abundances and WD’s temperature. RU Peg has stronger ISM absorption lines and, for this system
only, we model the ISM atomic and molecular hydrogen opacities. For these 5 systems we obtain
the temperature, projected rotational velocity and chemical abundances (of C, N, S and Si) of the
exposed WD atmosphere. We also give upper and lower limits for the mass of the WD and the
distance to the system.
2. The Archival Spectra
The observations log is presented in Table 2. From the AAVSO (American Association of
Variable Stars Observers) data, we found that all the systems were in optical quiescence at the
time of the observation (see the last column of Table 2). The STIS spectra of EY Cyg and SS Aur
are snapshots (lasting 600-700s) with a lower resolution and S/N than the STIS spectra of VW Hyi
– 5 –
and EK TrA. With fluxes ≤ 1 × 10−14ergs s−1cm−2˚ A−1, both EY Cyg and EK TrA are actually
weak FUSE sources, and RU Peg (though with a FUSE flux ten times higher) is under-exposed
with an exposure time of only 2800s.
2.1.The FUSE Archival Spectra
2.1.1.Processing the FUSE Data
FUSE’s optical system consists of four optical telescopes (mirrors), each connected to a Row-
land spectrograph. The four diffraction gratings of the four spectrographs produce four independent
spectra on two detectors. Two mirrors and two gratings are coated with SiC to provide wavelength
coverage below 1020˚ A, and the other two mirrors and gratings are coated with Al and LiF. The
Al+LiF coating provides about twice the reflectivity of SiC at wavelengths >1050˚ A, and very little
reflectivity below 1020˚ A. These are known as the SiC1, SiC2, LiF1 and LiF2 channels.
All the FUSE spectra presented here were obtained through the 30”x30” LWRS Large Square
Aperture in TIME TAG mode. The data were processed with CalFUSE version 3.0.7 (Dixon et al.
2007), which automatically handles event bursts. Event bursts are short periods during an exposure
when high count rates are registered on one or more detectors. The bursts exhibit a complex
pattern on the detector, their cause, however, is yet unknown (it has been confirmed that they are
not detector effects). The main change from previous versions of CalFUSE is that now the data
are maintained as a photon list (the intermediate data file - IDF) throughout the pipeline. Bad
photons are flagged but not discarded, so the user can examine, filter, and combine data without
re-running the pipeline. A number of design changes enable the new pipeline to run faster and use
less disk space than before. Processing time with CalFUSE has decreased by a factor of up to 10.
To process FUSE data, we follow the same procedure used previously for the analysis of other
systems (such as WW Ceti, Godon et al. (2006b)); consequently we give only a short account of
this procedure. The spectral regions covered by the spectral channels overlap, and these overlap
regions are then used to renormalize the spectra in the SiC1, LiF2, and SiC2 channels to the flux in
the LiF1 channel. We then produced a final spectrum that covers the full FUSE wavelength range
905 − 1187˚ A. The low sensitivity portions of each channel were discarded. In most channels there
exists a narrow dark stripe of decreased flux in the spectra running in the dispersion direction.
This stripe has been known as the “worm” and it can attenuate as much as 50% of the incident
light in the affected portions of the spectrum; - this is due to shadows thrown by the wires on
the grid above the detector. Because of the temporal changes in the strength and position of
the “worm”, CALFUSE cannot correct target fluxes for its presence. Therefore, we carried out a
visual inspection of the FUSE channels to locate the worm and we manually discarded the portion
of the spectrum affected by the worm. We combined the individual exposures and channels to
create a time-averaged spectrum weighting the flux in each output datum by the exposure time
and sensitivity of the input exposure and channel of origin.
– 6 –
2.1.2. The FUSE Lines
The FUSE spectra of DNs in quiescence exhibit mainly absorption lines from the WD itself,
as the exposed WD is the main FUV component of the system. A second FUV component is
sometimes present as a flat and featureless continuum contributing in the very short wavelengths of
FUSE (λ < 970˚ A ). This second component could be a hot region of the accretion disk (the inner
disk) or the boundary/spread layer. Sharp absorption lines from circumstellar (or circumbinary)
material and/or the ISM are also often seen and, if the system is a weak FUSE source, sharp
emission lines from air glow (geo- and helio-coronal in origin) are present
Since the WD is the main component of the FUV spectra of DNs in quiescence, the main
characteristic of the FUSE spectra is the broad Lyβ absorption feature due to the gravity (Log(g) ≈
8) and temperature (15,000K< T < 25,000K) of the exposed WD (in the disk this feature is usually
smoothed out due to velocity broadening, unless the disk is almost face-on).
common absorption features observed in the spectra of DNs WDs are due to Ciii (1175˚ A), Cii
(1066˚ A), Siiii (≈1108-1114˚ A and ≈1140-1144˚ A), and Nii (1085˚ A when not contaminated by air
glow). At higher temperatures (T> 25,000K), as the continuum rises in the shorter wavelengths,
the higher orders of the Lyman series also become visible; however, they become narrower. At
these temperatures the Siv (1073˚ A) absorption line starts to appear, and, as there is more flux
in the shorter wavelengths, the Cii (1010˚ A) absorption line also becomes visible. At still higher
temperature (T> 50,000K), the Cii and Siiii lines disappear, and the spectrum becomes dominated
by high order ionization lines such as Niv (≈923˚ A), Svi (933.5 & 944.5˚ A), Siv (1063 & 1073˚ A),
Siiv (1066.6˚ A), and Oiv (1067.8˚ A).
The other most
On top of the spectrum of the WD, broad emission lines are found in quiescent DN systems,
usually the Ovi doublet and Ciii (977˚ A and 1175˚ A). These, together with broad emission lines
from Niv (≈923˚ A) and Svi (933.5 & 944.5˚ A) are often observed in nova-like systems (e.g. such
as AE Aqr, V347 Pup, DW UMa, see the MAST FUSE archives) and are not usually present in
low-inclination DN systems in quiescence. This implies that the gas could possibly be heated by a
shock (e.g. in the boundary layer) and the broadening and variation of the lines suggests they are
originating in the disk. This scenario has recently been supported by Kromer et al. (2007) who for
the first time modeled the broad H and He emission lines ab initio as irradiation of the inner disk
by the hot boundary layer and/or white dwarf. One fully expects that the metallic emission lines
form in the same way. The DN system here that exhibits the strongest emission lines is EK TrA
(i=58o). RU Peg (i=33o) does have some strong emission lines too, but lacks a STIS spectrum for
complete comparison; however, its IUE spectrum clearly show strong emission lines such as Civ
(1550˚ A) and Siiv (1400˚ A).
The FUSE spectra of the DN systems (and CVs in general) often show some ISM molecular
hydrogen absorption, which appears as sharp lines at almost equal intervals (12˚ A) starting at
wavelengths around 1110˚ A and continuing towards shorter wavelengths all the way down to the
– 7 –
hydrogen cut-off around 915˚ A . In the affected FUSE spectra, we identified the most prominent
molecular hydrogen absorption lines by their band (Werner or Lyman), upper vibrational level
(1-16), and rotational transition (R, P, or Q) with lower rotational state (J=1,2,3).
2.2.The HST/STIS Archival Spectra
2.2.1. Processing the HST/STIS Archival Data
All the STIS spectra were processed with CALSTIS version 2.19. Except for the spectrum
of EK TrA (obtained in TIME TAG operation mode), all the spectra were obtained in ACCUM
operation mode. All the STIS spectra used the FUV MAMA detector and all were centered on the
wavelength 1425˚ A . The STIS snapshots (with an exposure time of about 600-700s, for EY Cyg
and SS Aur) were obtained through the 52x0.2 aperture using the G140L optical element. These
snapshots consist of one spectrum. The STIS spectra of VW Hyi and EK TrA (with an exposure
time of 4,000s and above) were obtained through the 0.2x0.2 aperture using the E140M optical
elements, and with 42 echelle spectra each, are of a much higher resolution than the snapshots.
Towards the longer wavelengths, the echelle spectra do not overlap and five gaps are apparent
around λ ≈1634˚ A, 1653˚ A, 1672˚ A, 1691˚ A, and 1710˚ A.
2.2.2.The HST/STIS Lines
The STIS spectra of quiescent DNs are also dominated by the WD and their main characteristic
is the broad Lyα absorption feature at ≈1216˚ A. The other very common absorption lines are the
carbon lines Ci (1266˚ A, 1561˚ A, 1657˚ A), Cii (1335˚ A), and Ciii (1175˚ A); and the silicon lines
Siii (1260˚ A, 1300˚ A, 1530˚ A), Siiii (1300˚ A), and Siiv (1400˚ A).
The HST/STIS spectrum also exhibits some broad emission lines. The most prominent ones
are Ciii (1175˚ A), Civ (1550˚ A), Siiii (1206˚ A), Nv (1240˚ A), Heii (1640˚ A Balmer α), and Siiv
(1400˚ A). With their broad emission lines, both the FUSE and HST/STIS spectra show evidence of
2.3. Preparation of the Spectra
For four objects (EY Cyg, SS Aur, VW Hyi and EK TrA) we wish to combine the FUSE
spectrum with the HST/STIS spectrum in order to have a larger wavelength coverage and improve
the synthetic spectral modeling.
However, for each object, before combining the FUSE spectrum with the STIS spectrum we
must make sure that (i) the system was observed in the same (high or low) state; (ii) the shape of
– 8 –
the spectra (continuum and lines) is similar (if the spectra are not too noisy); (iii) the flux level in
the FUSE and HST/STIS spectra do not differ more than about 50% (FUSE and HST/STIS are
very different instruments, we do not expect their flux level to match exactly); (iv) for long period
systems for which only short exposure time spectra exist, the FUSE and STIS spectra have most
probably been obtained at a different binary orbital phase and it is not clear one can combine such
spectra. We address these points here in detail as follows.
(i) The observations listed here were not coordinated and consequently some were carried out
during early quiescence while others were carried out in late quiescence. For EY Cyg, the outburst
recurrence time is believed to be about 2000 days and the FUSE and STIS spectra were obtained
in deep quiescence. EK TrA has also a long outburst recurrence time and it was observed in deep
quiescence (45 days and 155 days after outburst). For SS Aur, both spectra were obtained about
a month into quiescence. For VW Hyi the FUSE spectrum was obtained 11 days into quiescence
while the main STIS spectrum was obtained 15 days into quiescence (we discuss VW Hyi in some
more detail in the results section).
(ii) For all the systems, a visual analysis shows that the shape of the continuum and lines is
similar in the FUSE and STIS spectra.
(iii) 0nly 2 systems (EY Cyg and SS Aur) have FUSE and STIS flux levels that do not match
exactly. For EY Cyg the STIS spectrum had to be multiplied by 1.5 to match the flux level of the
continuum in the FUSE spectrum, while for SS Aur it was multiplied by 0.93. At temperatures of
about 25,000K a change of 2,000K can produce a change of 50 percent in the flux; the accuracy of
the temperature we obtained for EY Cyg is therefore of this order (i.e. ±2,000K).
(iv) Since the systems were observed in quiescence, we expect mainly to see the exposed WD
with basically no contribution or occultation from the accretion disk (none of these systems is
eclipsing). The spectra, whether observed at a given phase or averaged over an entire orbit, should
not differ significantly. Only for spectra with a high S/N ratio and a high inclination do we expect
the absorption and emission lines to exhibit some red- or blue- shift if they were obtained at a given
orbital phase, while the orbit-averaged spectra will have broader absorption and emission features.
Except for RU Peg, all the FUSE spectra exposure times cover a significant fraction of the
orbital period (and in some cases several periods); therefore, the FUSE spectra are averaged over a
significantly large fraction of the orbit. The STIS spectra of VW Hyi and EK TrA are also averaged
over the orbit, and for these two systems one can therefore combine the FUSE and STIS spectra.
The STIS snapshots of EY Cyg and SS Aur are very short (less than 1ksec) and were taken
at a particular orbital phase. One might therefore argue that the STIS and FUSE spectra of EY
Cyg and SS Aur cannot be combined because of that. However, for both EY Cyg and SS Aur, they
fit except that the absorption lines in the STIS range are not as deep as in the FUSE range. The
effect of observing the systems at a particular orbital phase is not very pronounced, and this might
be due to the lower resolution of the STIS snapshots (in comparison to VW Hyi and EK TrA’s
STIS spectra) and the moderate inclination of EY Cyg and SS Aur. Therefore, for EY Cyg and
– 9 –
SS Aur, we also combine the FUSE and STIS spectra without concern about the orbital phase.
The only object for which we find a significant shift in the lines is the long period system RU
Peg, for which we only have a FUSE spectrum of 2,800s.
Finally, we deredden the spectra according to the E(B-V) values given in Verbunt (1987);
La Dous (1991); Bruch & Engel (1994).
3. Spectral Modeling
3.1. The Synthetic Stellar Spectral Codes
We create model spectra for high-gravity stellar atmospheres using codes TLUSTY and SYN-
SPEC1(Hubeny 1988; Hubeny & Lanz 1995). Atmospheric structure was computed (using TLUSTY)
assuming a H-He LTE atmosphere; the other species were added in the spectrum synthesis stage
using SYNSPEC. For hot models (say T > 50,000K) we switched the approximate NLTE treatment
option in SYNSPEC (this allows to consider and approximate NLTE treatment even for LTE mod-
els generated by TLUSTY). We generate photospheric models with effective temperatures ranging
from 12,000K to 75,000K in increments of about 10 percent (e.g. 1,000K for T≈15,000K and
5,000K for T≈70,000K). We chose values of Log(g) ranging between 7.5 and 9.5 for consistency
with the observed mass. We also varied the stellar rotational velocity Vrotsin(i) from 100km s−1
to 1000km s−1in steps of 100km s−1(or smaller if needed). In order to try and fit the absorption
features of the spectrum, we also vary the chemical abundances of C, N, S and Si. For any WD mass
there is a corresponding radius, or equivalently one single value of Log(g) (e.g. see the mass radius
relation from Hamada & Salpeter (1961) or see Wood (1990); Panei et al. (2000) for different
composition and non-zero temperature WDs).
Our suite of stellar spectra generator codes has been implemented and includes also the treat-
ment of the quasi-molecular satellite Lyman lines. The satellite hydrogen lines appear as strong
absorption features near 1400 and 1600˚ A (Lyα, in the IUE and STIS range; Koester et al. (1985);
Nelan & Wegner (1985)) and are somewhat weaker near 1060 and 1078˚ A (Lyβ, in the FUSE range;
e.g. Dupuis et al. (2006)). Following Dupuis et al. (2003, 2006), we used opacity tables computed
by Allard et al. (1998, 1994, 2004a) to take into account the quasi-molecular satellite line opac-
ities of H2 and H+
2for Lyα, Lyβ, and Lyγ. The quasi-molecular lines are expected to appear
when the temperature of the WD is ≈20,000K or lower, though the gravity, pressure and magnetic
field of the WD can also play a role in the formation of these lines (larger electronic density will
favor recombination, see e.g. Dupuis et al. (2003); G¨ ansicke et al. (2006) for a more complete
1http://nova.astro.umd.edu; TLUSTY version 200, SYNSPEC version 48
– 10 –
3.2. Modeling the ISM Hydrogen Absorption Lines
For all the systems showing ISM atomic and molecular hydrogen absorption lines, we identify
these lines in the figures to avoid confusing them with the WD lines. For RU Peg, however, some
ISM lines are deep and broad and we decided to model them, especially since some of the WD lines
(such as Siv λλ1062.6 & 1073) are located at almost the same wavelengths.
For RU Peg only, we model the ISM hydrogen absorption lines to assess the atomic and
molecular column densities. This enables us to differentiate between the WD lines and the ISM
lines, and helps improve the WD spectral fit. The ISM spectra models are generated using a program
developed by P.E. Barrett. This program uses a custom spectral fitting package to estimate the
temperature and density of the interstellar absorption lines of atomic and molecular hydrogen.
The ISM model assumes that the temperature, bulk velocity, and turbulent velocity of the medium
are the same for all atomic and molecular species, whereas the densities of atomic and molecular
hydrogen, and the ratios of deuterium to hydrogen and metals (including helium) to hydrogen can
be adjusted independently. The model uses atomic data of Morton (2000, 2003) and molecular
data of Abgrall et al. (2000). The optical depth calculations of molecular hydrogen have been
checked against those of McCandliss (2003).
The ratios of metals to hydrogen and deuterium to hydrogen are fixed at 0 and 2 × 10−5,
respectively, because of the low signal-to-noise ratio data.
used to estimate the density of atomic hydrogen and the depth of the unsaturated molecular lines
for molecular hydrogen. The temperature and turbulent velocity of the medium are primarily
determined from the lines of molecular hydrogen when the ISM temperatures are < 250K.
The wings of the atomic lines are
The ISM absorption features are best modeled and displayed when the theoretical ISM model
(transmission values) is combined with a synthetic spectrum for the object (namely a WD synthetic
3.3. Synthetic Spectral Model Fitting
Before carrying out a synthetic spectral fit of the spectra, we masked portions of the spectra
with strong emission lines, strong ISM molecular absorption lines, detector noise and air glow.
These regions of the spectra are somewhat different for each object and are not included in the
fitting. The regions excluded from the fit are in blue in Figures 1, 3, 5, 7, and 10. The excluded
ISM quasi-molecular absorption lines are marked with vertical labels in figures 2, 4, and 9. For RU
Peg, we model the ISM absorption features and the WD separately; namely, we also mask the ISM
absorption features when we model the WD.
After having generated grids of models for each target, we use FIT (Press et al. 1992), a χ2
minimization routine, to compute the reduced χ2
factor values for each model fit. While we use a χ2minimization technique, we do not blindly
ν(χ2per number of degrees of freedom) and scale
– 11 –
select the least χ2models, but we examine the models that best fit some of the features such as
absorption lines (see the fit to the FUSE spectrum alone) and, when possible, the slope of the wings
of the broad Lyman absorption features. We also select the models that are in agreement with the
known distance of the system.
However, in the model fitting, for a given WD mass and radius, the resulting temperature
depends mainly on the shape of the spectrum in the FUSE range. The flux level at 1000˚ A (between
Lyδ and Lyγ) is close to zero for temperatures below 18,000K, at 30,000K it is about 50% of the
continuum level at 1100˚ A and it reaches 100% for T>45,000K. At higher temperature (T>50,000K)
the spectrum becomes pretty flat and there is not much difference in the shape of the spectrum
between (say) a 50,000K and a 80,000K model. When fitting the shape of the spectrum in such a
manner, an accuracy of about 500-1,000K is obtained, due to the S/N. In theory, a fine tuning of
the temperature (say to an accuracy of about ±50K) can be carried out by fitting the flux levels
such that the distance to the system (if known) is obtained accurately. However, the fitting to the
distance depends strongly on the radius (and therefore the mass) of the WD. In all the systems
presented here the error on the measured mass of the WD is so large (see Table 1) that it produces
an error in the temperature much larger than 1,000K. This is because the Lyα and Lyβ profiles
depend on both temperature and gravity. For example G¨ ansicke et al. (2001) derived a best fit
temperature for EK Tra Teff ≈ 2360 × Log(g) − 95, where g is the (unknown) surface gravity
of the WD (7.0 < Log(g) < 9.0). Additional errors are further introduced as the distance and
reddening are rarely known accurately, therefore increasing significantly the inaccuracy of assessing
the temperature by scaling the synthetic flux to the observed flux. For these, we decided to fit the
temperature of each model, based on the shape of the FUSE spectrum in the shorter wavelengths,
for a temperature accuracy of only about 5%, namely accurate to within 500K for T< 20,000K;
1,000K for 20,000K<T<35,000K; and 2,000K for T>40,000K (see the Results section). While the
error in temperature is only 500K for a 18,000K model and 2,000K for a 50,000K model, the relative
error ∆T/T is the same for all the models.
For all models we first tried fitting solar abundance models and then tried changing the abun-
dances of some species to improve the spectral fit. In particular, for each system we fit obvious
absorption features which can determine the abundance of some specific element (C, Si, S, N). For
example, the Cii (≈1065˚ A ), the Siiii (≈1110˚ A ) and the Si-C blend (≈1138-1146˚ A ) are the
main absorption features in the FUSE range of a T=20,000K WD that help assess the Si and the
C abundances (these are discussed in detail for each system in the next section).
The WD rotation (Vrotsin(i)) rate is determined by fitting the WD model to the spectrum
while paying careful attention to the line profiles in the FUSE portion of the combined spectrum.
We did not carry out separate fits to individual lines but rather tried to fit the lines and continuum
in the same fit while paying careful attention to the absorption lines.
It is important to note that, in the modeling, the depth of the absorption features depends
not only on the abundances but also on the rotational velocity. Increasing the rotational velocity
– 12 –
decreases the depth of the absorption features, but broadens the wings - the net effect being that
higher metal abundances result with faster rotational velocity. With sufficiently high S/N data, it
becomes possible to assess the rotational velocity and abundances independently.
4.Results and Discussion
All the results are presented in Table 5. For some systems for which the mass of the WD is
relatively unknown, we list more than one optimized fit result, assuming different values of Log(g)
(i.e. for each value of Log(g) we find an optimized fit). In the first column we list the name of the
system; in the second column the assumed surface gravity on a logarithmic scale Log(g) (where
it is understood that Log is the logarithm to the base 10 - as opposed to log which denotes the
logarithm to the base e) in cgs units; in column (3) we list the effective surface temperature we
obtained for the WD; in column (4) we list the projected rotational velocity which was obtained
when matching the stellar absorption lines; in columns (5), (6), (7) and (8) we list the abundances
of Carbon, Silicon, Nitrogen and Sulfur (respectively) in solar units; in column (9) we list the
distance we obtained from the fitting; in column (10) we list the parameter by which the STIS flux
had to be multiplied to match the FUSE flux; in column (11) we give the χ2
(12) we indicate the number of the figure displaying the model fit.
νvalue; and in column
4.1. EY Cygni
EY Cygni is a peculiar long period U Geminorum-type of DN, with a very massive accreting
WD (M1= 1.26M⊙or possibly larger), an outburst amplitude of 4.1 magnitude with a recurrence
time of 2000 days, and an orbital period of 11.0238 hrs (Smith et al. 1997; Costero et al. 1998;
Tovmassian et al. 2002; Costero et al. 2004). Hα and Sii ground-based imaging has revealed it
is associated with a non-homogeneous shell with a size of about 25arcsec (Tovmassian et al. 2002;
Sion et al. 2004b). It is possible that this shell is the result of a recent nova explosion. The
spectral type of the secondary is not certain, it is believed to be a K5V to M0V star, from which
the lower limit of the distance to EY Cyg has been inferred to be at least 250pc. Since EY Cyg
is not embedded in the background Cygnus superbubble (Bochkarev & Sitnik 1985), the upper
limit for its distance is 700pc. An additional characteristic which makes EY Cyg a peculiar DN
is its anomalous Nv/Civ ratio. Winter & Sion (2001) first reported a very strong Nv emission
and a very weak (or absent) Civ, which is atypical of most DNs in which Civ is usually the most
prominent and common emission line in their FUV spectra. More recently G¨ ansicke et al. (2003)
reported an analysis of HST/STIS snapshots of 4 CVs (including EY Cyg) with anomalously large
Nv/Civ line flux ratios, similar to those observed in AE Aqr. So far 10 systems have been identified
with such an anomalous Nv/Civ ratio.
EY Cyg is an extremely weak FUSE source and the spectrum is therefore very noisy, which
– 13 –
is the reason we binned its FUSE spectrum at 0.5˚ A for the fitting (Figure 1). The STIS snapshot
spectrum was binned at the default value of 0.58˚ A. Consequently the relative weight of the STIS
spectrum is about twice as large as that of the FUSE spectrum. In preparation for the model fitting
we had to multiply the STIS flux by 1.5 to match the FUSE flux level; this change of 50% in flux
corresponds to a (0.5)
our spectral modeling of the combined FUSE+STIS spectrum of EY Cyg is of the order of ∼ 0.1
ab initio. Since the distance to the system is relatively unknown, there is no restriction on the flux
level to minimize the number of best fits to one in the Log(g) − Teffparameter space. Therefore,
we restricted the modeling assuming only two different values for the mass of the WD. Because
of the extremely large WD mass of the system (which within the margin of error well exceeds the
Chandrasekhar mass limit for a WD) we ran models assuming Log(g) = 9.0 (corresponding to
Mwd= 1.21M⊙) and Log(g) = 9.5 (corresponding to Mwd= 1.35M⊙).
4 ≈ 10% in the value of T. Therefore, the relative error in temperature in
Within the range of the temperature error (10%) the best fit models spread between about
T=27,000K and T=33,000K. However, the high temperature (T > 32,000K) models do not produce
enough flux in the longer wavelengths of STIS, while the lower temperature (T < 28,000K) models
do not have enough flux in the shorter wavelengths of the FUSE portion of the spectrum. Because
of that we rejected the T = 27,000K and T = 33,000K models. In addition, in the observed
STIS spectrum the bottom of the Lyα does not go to zero and this corresponds to models with
T ∼ 30,000K and higher. Because of the difference in fluxes between STIS and FUSE we decided
to check how the results would change if we fit the spectra separately. We found that the STIS
spectrum gave the same temperature, however with a distance about 20 percent larger as the main
driving elements in choosing the best fit model were the same, namely: the Lyα profile and the
longer wavelengths of STIS. Fitting the FUSE spectrum alone gave a temperature about 2,000K
higher than for the combined spectrum. We therefore inferred for EY Cyg T = 30,000K±2,000K
assuming Log(g) = 9.0. The 2,000K error in T produces an error of ±60pc in the distance as listed
in Table 5. This 30,000K model is shown in Figure 1. In Figure 2 the 30,000K WD model assuming
Log(g) = 9.0 is shown with the FUSE spectrum binned here at 0.1˚ A to show the sharp emission
and absorption lines.
We also checked how the inclusion of the quasi-molecular hydrogen affects the result and found
that it does not improve the fits.
Next, we assume Log(g) = 9.5 to check the effect of larger WD mass. Using the same consid-
erations as before we found that the best fit models range between 30,000K and 34,000K, or about
2,000K higher than the Log(g) = 9.0 models.
The projected rotational velocity we obtained for all the models is relatively low: 100km s−1.
We confirm the anomalously low Carbon abundance, namely we found that the best Carbon line fits
are obtained for abundances ranging between 0.01 and 0.05 solar. The low carbon abundance was
set to match the FUSE absorption features of Ciii 1175˚ A, Cii 1066˚ A, and Ci-Ciii at 1128 and
1140˚ A. There is also a complete absence of Ciii emission at 977˚ A and 1175˚ A, and Civ emission at
– 14 –
1550˚ A . There is a strong and broad Nv 1240˚ A emission line. The Nitrogen abundance (2× solar)
was obtained by fitting the Ni (1135˚ A), Nii (1185˚ A), and Niii (1189˚ A) lines. However, these lines
are not very pronounced and are affected by air glow emission. The Sulfur abundance (6× solar)
was obtained by fitting the Siv (1073˚ A) and Si (1097˚ A) lines. The Silicon abundance (0.6× solar)
was obtained by fitting the Siiv (1125˚ A), and Siiii (1140-1145˚ A) lines.
Sion et al. (2004b) modeled the combined HST/STIS and FUSE spectra of EY Cygni. Despite
the noisy, weak FUSE spectrum, they managed to be obtain two possible solutions which in a
ν) sense differ little from each other. If the combined FUV spectra was due to a single
temperature white dwarf alone, then for their adopted mass of 1.2M⊙(assuming Log(g) = 9) with
no interstellar reddening (see below), then the WD has Teff= 24,000K. However, the combined
FUSE plus STIS spectra also yielded a similar statistical fit if a two-temperature white dwarf
model was applied to the data. This fit corresponded to a cooler, slowly rotating WD photosphere
(Teff = 22,000K) with the second temperature (Teff = 36,000K) representing a putative hot,
rapidly rotating, equatorial accretion zone (an accretion belt). We note here that their noisy FUSE
spectrum showed a sharp increase in flux extending even beyond the Lyman limit. This extra flux
is an artifact of the CaLFUSE background subtraction and, therefore, in the present woork we
process the FUSE data with latest (and final) version of CalFUSE (v3.2) which lowered the flux
at wavelengths ¡ 950˚ A, removing the shortward flux excess. The new combined HST/STIS plus
reprocessed FUSE data is fit in the present paper with the FUSE spectral region shortward of
950˚ Amasked in the fitting.
Our single WD component best fit, assuming Log(g) = 9.0, has Teff = 30,000K, with a
distance d ∼500pc; and assuming Log(g) = 9.5 we found a WD temperature of 32,000K with a
distance of ∼300pc. EY Cyg does not show any sign of quasi-molecular satellite lines, and this
is consistent with our higher temperature model (see the Discussion section), and no need to add
a second component since this component was driven by the flux excess which the re-processing
SS Aur is a U Gem type DN with an orbital period Porb=4.3872 hrs (Shafter & Harkness
1986), a WD mass Mwd= 1.08 ± 0.40 M⊙, a secondary mass M2= 0.39 ± 0.02 M⊙, and a system
inclination i = 38o± 16o(Shafter 1983). SS Aur has an HST FGS parallax measurement of 497
mas (Harrison et al. 1999) giving a distance of 201pc.
Lake & Sion (2001) analyzed the IUE archival spectra of SS Aur; their best-fit model photo-
sphere has Teff = 30,000K, Log(g) = 8.0, and solar composition abundances; while the best-fit
accretion disk model has Mwd= 1.0 M⊙, i = 41o, and ˙M = 10−10M⊙yr−1.
Using a newly available Hubble FGS parallax, Lake & Sion (2001) showed for the first time
– 15 – Download full-text
that the FUV flux of SS Aur during quiescence was dominated by, if not provided entirely by, a
hot WD, not an accretion disk. This was something of a milestone because the prevaling view was
that the underlying degenerate was detected in the FUV only in those dwarf novae which clearly
revealed the broad Lyα profile of a WD, namely U Gem, VW Hyi and WZ Sge.
More recently Sion et al. (2004a) analyzed the FUSE spectrum of the system. Their best
fit model is a 27,000K WD contributing 73% of the flux with a hot belt (48,000K) contributing
the remaining 27% of the flux, with a corresponding distance of 267pc. However, a single WD
component also gives a good fit with a temperature Twd=33,000K and a distance of 303pc. They
also suggested that the absence of the Ciii (1175˚ A) absorption line might be a sign of composition
deficit in carbon. It is important to note here that in these previous studies the spectra of SS Aur
were not dereddened.
The STIS spectrum of SS Aur was obtained as part of a snapshot program (G¨ ansicke et al.
2003) and the single STIS spectrum is being analyzed also in Sion et al. (2008). Here, we ana-
lyzed the combined FUSE+STIS spectrum of SS Aur which we dereddened assuming E(B-V)=0.08
(Bruch & Engel 1994). The FUSE spectrum of SS Aur was obtained 28 days after outburst, while
the STIS spectrum was obtained 34 days after outburst. The fit was carried out with a binning of
0.2˚ A in the FUSE spectrum and 0.58˚ A in the STIS spectrum, consequently the relative weight of
the FUSE spectrum is about 50% larger than that of the STIS spectrum.
The best fit model for a ≈1.1M⊙(Log(g) = 8.71) mass WD and a distance of 200pc gives a
temperature of 31,000K, solar abundances, a rotational velocity of 400km/s and χ2
fit is the best when fitting the flux level while keeping the distance and radius of the star fixed. If
we relax the restriction on the distance, the best fit model with Log(g) = 8.71 gives a temperature
of 33,000K, a distance of about 240pc and a slightly lower χ2
of the χ2
νbetween the 31,000K and the 33,000K models is not significant. However, it is this
discrepancy between the apparent temperature of the WD and its flux that has led to the adoption
of second model components in other CVs. However, the WD mass of SS Aur is given with a
large error (±0.4M⊙). We, therefore, computed more models with Log(g) = 8.31 and 8.93. The
Log(g) = 8.31 model, for a distance of 200pc gave a temperature of 27,000K, while the lowest χ2
best fit model for that value of Log(g) gave a temperature of 30,000K and a distance of 254pc. For
the Log(g) = 8.93 model, the distance of 200pc is achieved with a temperature of 34,000K. This
model, however, is also the lowest χ2
ν(1.472). The difference in the values
νbest fit model for that value of Log(g) with the least χ2
A model fit to the STIS spectrum alone of SS Aur (which was carried out in (Sion et al.
2008)) gave the same results, namely a 34,000K WD assuming Log(g) = 8.8. The only difference
with our modeling is that the STIS spectrum in Sion et al. (2008) was NOT normalized to the
FUSE spectrum, which explain the slight difference in the value of Log(g). We then ran a model to
fit the FUSE spectrum alone and found total agreement with the FUSE+STIS spectral fits, namely
that the best fit model is again a T=34,000K WD, spinning at 400km/sec with solar abundances,