A&A 435, 1013-1030 (2005)
DOI: 10.1051/0004-6361:20042368
J. Petrovic1,2 - N. Langer1 - K. A. van der Hucht3,4
1 - Sterrenkundig Instituut,
Universiteit Utrecht,
Princetonplein 5, 3584 CC Utrecht,
The Netherlands
2 -
Astronomical Institute,
Radboud Universiteit Nijmegen,
Toernooiveld 1, 6525 ED,
Nijmegen, The Netherlands
3 -
SRON, Nationaal Instituut voor Ruimte Onderzoek,
Sorbonnelaan 2, 3584 CA Utrecht, The Netherlands
4 -
Sterrenkundig Instituut Anton Pannekoek, Universiteit van Amsterdam,
Kruislaan 403, 1098 SJ Amsterdam, The Netherlands
Received 15 November 2004 / Accepted 18 February 2005
Abstract
Since close WR+O binaries are the result of a strong interaction
of both stars in
massive close binary systems, they can be used to constrain
the highly uncertain mass and angular momentum budget
during the major mass transfer phase.
We explore the progenitor evolution of the three
best suited WR+O binaries HD 90657, HD 186943 and HD 211853,
which are characterized by
a WR/O mass ratio of 0.5 and periods of 6...10 days.
We are doing so at three different levels of approximation:
predicting the massive binary evolution through
simple mass loss and angular momentum loss estimates, through
full binary evolution models with parametrized mass transfer
efficiency, and through binary evolution models including rotation
of both components and a physical model which allows to
compute mass and angular momentum loss from the binary system
as function of time during the mass transfer process.
All three methods give consistently the same answers.
Our results show that, if these systems formed through stable mass transfer,
their initial periods were smaller than their current ones,
which implies that mass transfer has started during the
core hydrogen burning phase of the initially more massive star.
Furthermore, the mass transfer in all three cases must have
been highly non-conservative, with on average only
10%
of the transferred mass being retained by the mass receiving star.
This result gives support to our system mass and angular momentum
loss model, which predicts that, in the considered systems,
about 90% of the overflowing matter is expelled by the
rapid rotation of the mass receiver close to the
-limit,
which is reached through the accretion of the remaining 10%.
Key words: stars: binaries: close - stars: evolution - stars: fundamental parameters - stars: rotation - stars: Wolf-Rayet
The rotational properties of binary components may play a key role in this respect. The evolution of massive single stars can be strongly influenced by rotation (Heger & Langer 2000; Meynet & Maeder 2000), and evolutionary models of rotating stars are now available for many masses and metallicities. While the treatment of the rotational processes in these models is not yet in a final stage (magnetic dynamo processes are just being included Maeder & Meynet 2003; Heger et al. 2004), they provide first ideas of what rotation can really do to a star. Effects of rotation, as important they are in single stars, can be much stronger in the components of close binary systems: Estimates of the angular momentum gain of the accreting star in mass transferring binaries show that critical rotation may be reached quickly (Langer et al. 2000; Yoon & Langer 2004b; Packet 1981). In order to investigate this, we need binary evolution models which include a detailed treatment of rotation in the stellar interior, as in recent single star models. However, in binaries, tidal processes as well as angular momentum and accretion need to be considered at the same time. Some first such models are now available and are discussed below.
Angular momentum accretion and the subsequent rapid rotation of the mass gainer may be essential for some of the most exciting cosmic phenomena, which may occur exclusively in binaries: type Ia supernovae, the main producers of iron and cosmic yardsticks to measure the accelerated expansion of the universe (Yoon & Langer 2004b,a), and gamma-ray bursts from collapsars, which the most recent stellar models with rotation and magnetic fields preclude to occur in single stars (Heger et al. 2004; Woosley 2004; Petrovic et al. 2004). For both, the type Ia supernova progenitors and the gamma-ray burst progenitors, it is essential to understand how efficient the mass transfer process is and on which physical properties it depends. Further exciting astrophysical objects whose understanding is affected by our understanding of mass transfer comprise X-ray binaries (Chevalier & Ilovaisky 1998) and type Ib and Ic supernovae (Podsiadlowski et al. 1992).
How much matter can stars accrete from a binary companion? As mentioned above, non-magnetic accretion, i.e. accretion via a viscous disk or via ballistic impact, transports angular momentum and can lead to a strong spin-up of the mass gaining star. For disk accretion, it appears plausible that the specific angular momentum of the accreted matter corresponds to Kepler-rotation at the stellar equator; this leads to a spin-up of the whole star to critical rotation when its initial mass is increased by about 20% (Packet 1981). It appears possible that mass accretion continues in this situation, as viscous processes may transport angular momentum outward through the star, the boundary layer, and the accretion disk (Paczynski 1991). However, as the star is rotating very rapidly, its wind mass loss may dramatically increase (Langer 1998,1997), which may render the mass transfer process inefficient.
Observations of massive post-mass transfer binary systems constrain this
effect. Langer et al. (2003) and Langer et al. (2004) points out that there is
evidence for both extremes occurring in massive close binaries, i.e. for
quasi-conservative evolution as well as for highly non-conservative evolution.
In the present study, we are interested in those binaries that contain a
Wolf-Rayet and a main sequence O star.
We have chosen to focus on three WN+O systems (HD 186943, HD 90657 and HD 211853)
which have similar mass ratios (0.5) and orbital periods (6...10 days).
As clearly the two stars in these systems must have undergone
a strong interaction in the past, an understanding of their
progenitor evolution may be the key to constrain the mass transfer
efficiency in massive binaries: which fraction of the mass leaving
the primary star is accumulated by the secondary star during a mass
transfer event?
Evolutionary calculations of massive close binaries were performed by various authors.
General ideas about the formation of WR+O binary systems were given by
Paczynski (1967), Kippenhahn et al. (1967), van den Heuvel & Heise (1972).
Vanbeveren et al. (1979) modelled the evolution of massive Case B binaries
with different assumptions for mass and angular momentum loss from the binary system.
Vanbeveren (1982) computed evolutionary models of massive close Case B binaries with primary
masses between
and
.
He concluded that most of the
WR primaries are remnants of stars initially larger than
and
that the accretion efficiency in these systems should be very below 0.3
in order to fit the observations.
de Loore & de Greve (1992) computed detailed models of massive Case B binary
systems
for initial mass ratios of 0.6 and 0.9, assuming
an accretion efficiency of 0.5.
Wellstein & Langer (1999) and Wellstein et al. (2001) modelled
massive binary systems mass range 12...60
assuming conservative evolution,
and Wellstein (2001) presented the first rotating binary evolution
models for initial masses of
and initial mass ratios
.
While it was realized through these models that different mass accretion may be needed to explain different observations, these efforts did not have the potential to explore the physical reasons for non-conservative evolution. I.e., there is no reason to expect that the mass transfer efficiency remains constant during the mass transfer process in a given binary system, nor that its time-averaged value is constant for whole binary populations.
It is not yet known which physical processes can expel matter from a binary system.
Vanbeveren (1991) proposed that if a binary component is more
massive than 40-50
it will go through an LBV phase of enhanced
mass loss, which will prevent the occurrence of RLOF.
Dessart et al. (2003) investigated the possibility that radiation pressure from the
secondary prevents the accretion. They found that even for moderate mass transfer rates
(
)
the wind and photon momenta
which emerge from the accretion star can not alter the
dynamics of the accretion stream. Here, we follow the suggestion
that the effective mass accretion rate can be significantly decreased
due to the spin-up of the mass receiving star (Petrovic & Langer 2004; Wellstein 2001; Langer et al. 2003,2004).
The remainder of this paper is organized as follows. In Sect. 2 we briefly discuss the observational data available for WR+O binary systems. In Sect. 3 we derive estimates for the masses of both stars in WR+O systems for given initial masses and accretion efficiencies. In Sect. 4 we present the physics used to compute our detailed evolutionary models. Non-rotating binary evolution models with an adopted constant mass accretion efficiency are presented in Sect. 5. Our rotating models in which the mass accretion efficiency is obtained selfconsistently are discussed in Sect. 6. We briefly compare our models with observations in Sect. 7. Conclusions are given in Sect. 8.
WN+O systems that also have short orbital periods are V444 Cyg, CX Cep,
CQ Cep, HD 94546, HD 320102 and HD 311884.
V444 Cyg has period of 4.2 days
and can be result of stable mass transfer evolution, but since mass ratio
of this
system is 0.3 we did not include it in this paper.
Orbital periods of CX Cep and CQ Cep
are very short (
2 days) and these systems
are probably the result of a contact evolution.
HD 94546 and HD 320102 are systems with very low masses of WR and O components
(4
and 2.3
respectively)
and HD 311884 is extremely massive WR+O binary system
(51
). Recently, an even more massive WR+O system has been observed 83
(Bonanos et al. 2004; Rauw et al. 2004).
Table 1: Basic parameters of selected WN+O SB2 binaries.
The mass ratio of a binary system is determined from its radial velocity
solution, with an error of 5-10%. However, to determine the exact value
of the masses of the binary components, the value of the inclination of the
system has to be known. Without knowledge of the inclination, only minimum
masses of the components can be determined, i.e., M sin3i.
Massey (1981) determined the minimum mass for the WR star in
HD 186943 to be 9-11
.
Niemela & Moffat (1982)
determined the masses of the components of HD 90657 in the range
11-14
for the WN4 component and 21-28
for the O-type component. The masses of the WR components in HD 186943
and HD 90657 given in Table 1 have been determined by
Lamontagne et al. (1996) on the basis of improved values for the
inclination of these systems. Demers et al. (2002) determined
minimum masses of the components of the system GP Cep. Previously,
Lamontagne et al. (1996) suggested values of
and
for this system.
There is no obvious hydrogen contribution in the WR spectrum in any of
these systems (Niemela & Moffat 1982; Massey 1981).
Massey (1981) showed that hydrogen absorption lines are
fairly broad in the spectrum of HD 186943, equivalent to
,
thus the O-type star is rotating much faster
than synchronously.
Beside the fact that the binary system GP Cep has a similar mass ratio and
period as the other two systems, it has some very different properties as
well. The spectral type of the WR component in GP Cep is a combination of
WN and WC (WN6/WCE Demers et al. 2002). Also,
Massey (1981) showed that, next to the main period of
6.69 days of the binary system GP Cep, radial velocities of
absorption lines vary also with a period of 3.4698 days. He proposed that
GP Cep is a quadruple system, consisting of two pairs of stars, WR+O and O+O.
Panov & Seggewiss (1990) suggested that in both pairs one component is a
WR star. However, Demers et al. (2002) showed that there is only
one WR star in this quadruple system.
If the initial binary system is very close (an initial period is of the order of few days), RLOF occurs while the primary is still in the core hydrogen burning phase and Case A mass transfer takes place (fast and slow phase). When the primary expands due to shell hydrogen burning, it fills its Roche lobe and Case AB mass transfer starts. During this mass transfer the primary star loses the major part of its hydrogen envelope. After Case AB mass transfer, the primary is a helium core burning Wolf-Rayet star. During all this time, the secondary is still a main sequence star, but with an increased mass due to mass transfer. When the initial binary period is of the order of one to few weeks, the primary fills its Roche lobe for the first time during shell hydrogen burning and Case B mass transfer takes place. The primary loses most of its hydrogen envelope, becomes a WR star and the secondary is an O star with an increased mass. Case C mass transfer occurs when initial period is of the order of years. The primary fills its Roche lobe during helium shell burning and mass transfer takes place on the dynamical time scale. This scenario is not likely for chosen systems, since some of the secondary stars in WR+O systems have been observed to rotate faster than synchronously. This means that they have accreted some matter which increased their spin angular momentum.
We constructed a simple method to quickly estimate the
post-mass transfer parameters for a large number of binary systems
for a given accretion efficiency .
This allows us to narrow the space of possible initial
parameters (primary mass, secondary mass and orbital period)
that allows the evolution into a specific observed WR+O systems.
We considered binary systems with
initial primary masses
...100
and secondaries masses
/1.7...100
with an initial period of 3 days. We assumed that the primary is transferring matter
to the secondary until it reaches the mass of its initial helium core (Eq. (1)).
Matter that is not accreted on the secondary leaves the system with the specific angular momentum which corresponds to the secondary's orbital angular momentum (King et al. 2001), which is consistent with our approach for mass loss from the binary system (cf. Sect. 4). Stellar wind mass loss is neglected.
More massive initial primaries produce more massive WR stars (helium cores) in general, but if the star is in a binary system that goes through mass transfer during hydrogen core burning of the primary (Case A), this depends also on other parameters:
We estimated the minimum initial helium core masses, which are obtained
by the earliest Case A systems, for systems
with a mass ratio of
and an initial period of
3 days (
)
from the detailed
evolutionary models shown later in this paper
(Sect. 5):
For Case B binaries, the initial WR mass does not depend on the initial period
and the initial mass ratio of the system, since during core hydrogen burning, the primary evolves as a single
star, without any interaction with the secondary.
We estimated the relation between initial main sequence mass and initial WR mass
as a linear fit from the Case B binary systems with initial primaries
(Wellstein & Langer 1999):
We calculate binary systems for early Case A (
days) and for late Case A
(
).
![]() |
Figure 1:
Masses of both components of post-Case A mass transfer WR+O binary
systems resulting from our simple approach, for
initial primary masses in the range 25...100
![]() ![]() ![]() ![]() ![]() |
Open with DEXTER |
![]() |
Figure 2:
Masses of both components of post-Case A mass transfer WR+O binary
systems resulting from our simple approach, for
initial primary masses in the range 25...100
![]() ![]() ![]() ![]() |
Open with DEXTER |
The results are shown in Fig. 1 for four different
accretion efficiencies (,
0.1, 0.5, 1.0 respectively) for early Case A evolution
(p=3 days). Figure 2 shows the results for early Case A systems (p=3 days) for
all values of
...1 and for Case B/late Case A, also for all
.
We notice from Fig. 1, that when the assumed
is larger,
the resulting WR+O systems lie further from the line defined by q=0.5. The reason is clear:
if the accretion efficiency is higher, the secondary will become
more massive while the initial mass of the WR star stays the same.
Conservative evolution (Fig. 1d) produces WR+O systems that have
small mass ratios,
q=1/5...1/6.
We conclude that if the considered three observed WR+O binary systems evolved through a stable mass transfer, a large amount of matter must have left the system. On the other hand, since some of the secondary stars in WR+O binaries have been observed to rotate faster than synchronously (Underhill et al. 1988; Massey 1981), a certain amount of accretion may be required.
Figure 2 shows the resulting WR+O masses in Case A and Case B (latest Case A) for accretion
efficiency ...1. If the primary star does not lose
mass in a mass transfer during core hydrogen burning (
),
it will form a more massive WR star, as we already explained.
There will be less mass to transfer from the primary to the secondary, and for fixed
the corresponding O star will become less massive.
However, since the observed periods of HD 186943, HD 90657 and HD 211853 are shorter than
10 days and Case A+Case AB widens the binary orbit, the initial orbital period should be shorter
than observed, so we can conclude roughly that
is between 3 and 10 days.
The orbital period of WR+O systems depends on their initial orbital period,
their initial mass ratio and on the parameter .
If the initial period increases and there is no contact during the evolution,
the orbital period in the WR+O stage will also increase.
However, the orbital period of WR+O systems will be shorter if the initial mass ratio is larger.
If the initial masses are very similar, the primary will become less massive than the
secondary very early during the mass transfer, and afterward matter is
transfered from the less to the more massive star, which results in a widening of the orbit.
Conversely, the final period is shorter for a larger difference in initial masses in the
binary system.
We can draw the following conclusions:
We showed in Sect. 3 that we can roughly estimate the parameters of the progenitor systems of observed WR+O binaries HD 186943, HD 90657 and HD 211853. However, detailed numerical models are required in order to verify that the assumption of contact-free evolution can in fact be justified. And finally, we want to check whether the required mass and angular momentum loss can be reproduced by our detailed selfconsistent approach.
We are using a binary evolutionary code which was originally
developed by Braun (1998) on the
basis of an implicit hydrodynamic stellar evolution code for single stars
(Langer 1998,1991).
It calculates simultaneous evolution of the two stellar
components of a binary system in a circular orbit
and the mass transfer within the Roche approximation
(Kopal 1978).
Mass loss from the Roche lobe filling component through the first Lagrangian point
is given by Ritter (1988) as:
![]() |
(3) |
Stellar wind mass loss for O stars on the main sequence is calculated according to
Kudritzki et al. (1989).
For hydrogen-poor stars (
)
we assume mass loss based on the empirical
mass loss rates for Wolf-Rayet stars derived by Hamann et al. (1995):
The treatment of a convection and a semiconvection which is applied here is described in Langer (1991) and Braun & Langer (1995). Changes in chemical composition are computed using a nuclear network including pp chains, the CNO-cycle, and the major helium, carbon, neon and oxygen burning reactions. More details are given in Wellstein & Langer (1999) and Wellstein et al. (2001). We use the OPAL Rosseland-mean opacities of (Iglesias & Rogers 1996). For all models, a metallicity of Z=0.02 is adopted. The abundance ratios of the isotopes for a given element are chosen to have the solar meteoritic abundance ratios according to Grevesse & Noels (1993). The change of the orbital period (orbital angular momentum loss) due to the mass transfer and stellar wind mass loss is computed according to Podsiadlowski et al. (1992), with the specific angular momentum of the stellar wind material calculated by Brookshaw & Tavani (1993).
The influence of the centrifugal force in the rotating models
is implemented according to Kippenhahn & Thomas (1970).
The stellar spin vectors are assumed to be perpendicular to the
orbital plane. Synchronization due to tidal
spin-orbit coupling is included with a time scale given by Zahn (1977).
Rotationally enhanced mass loss is included as follows:
![]() |
(5) |
When the star approaches ,
the mass loss rate is increased according
to the previous equation. However, mass loss also causes a spin-down of the star
and equilibrium mass loss rate
results (Langer 1998).
If
,
the corresponding angular momentum loss is so large
that the star evolves away from the
-limit.
The transport of angular momentum through the stellar interiour is formulated as a diffusive process:
![]() |
(6) |
The specific angular momentum of the accreted matter is determined by integrating the equation of motion of a test particle in the Roche potential in case the accretion stream impacts directly on the secondary star, and is assumed Keplerian otherwise Wellstein (2001). Rotationally induced mixing processes and angular momentum transport through stellar interior are described by Heger et al. (2000). Magnetic fields generated due to differential rotation in the stellar interior (Spruit 2002) are not included here (however, see Petrovic et al. 2004).
We calculated the evolution of the binary systems in detail until Case AB
mass transfer starts. Then we estimated the outcome of this mass transfer
by assuming that it ends when WR star has 5% of the hydrogen left at the surface.
For this purpose we calculate the Kelvin-Helmholtz time scale of the primary:
![]() |
(7) |
We concluded in Sect. 3 that massive O+O binaries can result in
WR+O systems similar to observed the (HD 186943,
HD 90657 and HD 211853) if accretion efficiency
is low.
Since some O stars in WR+O binaries have been observed to rotate faster than synchronously,
we concluded that
and assumed a constant value of
in our detailed evolutionary models.
We already mentioned that the orbital periods of the observed systems are between 6 and 10 days. Since the net effect of Case A+Case AB mass transfer is a widening
of the orbit, the initial periods should be shorter than the observed ones, so
we modelled binary systems with initial orbital periods of 3 and 6 days.
Table 2:
Non-rotating WR+O progenitor models for .
N is the number of the model,
and
are initial masses of the primary and
the secondary,
is the initial orbital period and
is the initial mass ratio of
the binary system.
is time when Case A mass transfer starts,
is the duration of the fast phase of Case A mass transfer,
is the maximum mass
transfer rate,
and
are mass loss of the primary and mass gain
of the secondary (respectively) during fast Case A,
is the duration of slow Case A mass
transfer,
and
are mass loss of the primary and mass gain
of the secondary (respectively) during the slow Case A,
is the orbital period at the onset of Case
AB,
is the mass loss of the primary during Case AB (mass gain of the secondary is
1/10 of this, see Sect. 4),
is the WR mass when the hydrogen surface abundance
is
,
the WR mass at
is given in brackets,
is the mass of the corresponding O star, q is the mass ratio
,
and p is the orbital period of the WR+O system.
The models are computed with a stellar wind mass loss of Hamann/6, except
Hamann/3,
Hamann/2.
c indicates a contact phase that occurs
for low masses due to a mass ratio too far from unity, for high masses
due to the secondary expansion during slow phase of Case A.
We chose initial primary masses to be in the range 41...75
.
The masses of the secondaries are chosen so that the initial mass
ratio (
)
is
-2.0.
An initial mass ratio of
1.55 is estimated to be the limiting value
for the occurrence of contact between the components
in Case A systems by Wellstein et al. (2001) for conservative mass transfer.
Contact occurs when the accretion time scale of the secondary (
)
is much longer than the thermal (Kelvin-Helmholtz) time scale of the primary
(
), so the secondary expands
and fills its Roche lobe. In our models, only 10% of matter lost by the primary
is accreted on the secondary star, so it reaches hydrostatic equilibrium faster
and expands
less than in the case of larger
.
This is the reason why we adopted a weaker condition for contact formation and
calculate models with mass ratios
...2.0.
All modelled systems (except the ones that enter contact) go
through Case A and Case AB mass transfer.
Details of the evolution of all calculated binary systems are given in Table 2.
We discuss the details of the binary evolution taking the system number 11 as an example.
Figure 3 shows the evolutionary tracks of the primary and the secondary in the HR
diagram until the onset of Case AB mass transfer.
This system begins its evolution with the initial parameters
,
,
days.
Both stars are core hydrogen burning stars (dashed line, Fig. 3), but since the primary is more
massive, it evolves faster and fills its Roche lobe, so
the system enters Case A mass transfer (solid line, Fig. 3)
years after the beginning of core hydrogen burning.
The first phase of Case A is fast process and takes place on the Kelvin-Helmholtz (thermal) time
scale (
years). The primary loses matter quickly and
continuously with a high mass transfer rate
(
).
In order to retain hydrostatic equilibrium, the envelope
expands, which requires energy
and causes a decrease in luminosity (Fig. 3). At the same time the secondary is
accreting matter and is expanding.
Due to this, its luminosity increases and the effective temperature decreases (Fig. 3).
During fast phase of Case A mass transfer the primary loses
19
and the secondary
accretes 1/10 of that matter.
After the fast process of mass transfer,
the primary is still burning hydrogen in
its core and is still expanding, so
slow phase of Case A mass transfer takes place on a nuclear time scale (
years)
with a mass transfer rate of
.
After this, the primary is the less massive star,
with decreased hydrogen surface abundance.
Stellar wind mass loss of the primary increases when its surface becomes hydrogen poor
(
).
At the end of core hydrogen burning the primary contracts (effective temperature increases)
and thus RLOF stops (Fig. 3 dotted line).
When the primary starts
shell hydrogen burning it expands (dash-dotted line, Fig. 3),
fills its Roche lobe and Case AB mass transfer starts.
Figures 4 and 5 show the evolution of the interior of the primary and the secondary until Case AB mass transfer. The primary loses huge amounts of matter during fast Case A mass transfer and its convective core becomes less than a half of its original mass. At the same time, the secondary accretes matter from the primary and the heavier elements are being relocated by thermohaline mixing. In Figs. 6 and 7 we see the mass transfer rate and the surface abundances of hydrogen, carbon, nitrogen and oxygen.
During Case AB mass transfer the primary
star loses the major part of its hydrogen envelope. After Case AB mass transfer,
the primary is a helium core burning star (WR) and the secondary is still
a core hydrogen burning O star.
The masses of the modelled WR stars are in the range from 8...18.5
The orbital periods of the modelled WR+O systems vary
from
9.5 to
20 days, and the mass ratios are between 0.33 and 0.53.
![]() |
Figure 3:
HR diagram of the initial system
![]() ![]() ![]() |
Open with DEXTER |
![]() |
Figure 4:
The evolution of the internal structure of the 56
![]() |
Open with DEXTER |
![]() |
Figure 5:
The evolution of the internal structure of the 33
![]() |
Open with DEXTER |
![]() |
Figure 6:
Upper plot: mass transfer rate during Case A mass transfer in the binary system
with
![]() ![]() ![]() ![]() |
Open with DEXTER |
![]() |
Figure 7:
Surface abundance of carbon (solid line), nitrogen (dotted line) and
oxygen (dashed line) of the primary ( upper plot)
and the secondary ( lower plot) in the system with
![]() ![]() ![]() |
Open with DEXTER |
In our models the primary starts losing mass by stellar wind as WR star when its hydrogen surface
abundance goes below
.
However, the observed WR stars in HD 186943, HD 90657 and HD 211853 do not have
obvious hydrogen on the surface, so we
assume that these WR stars are the result of Case AB mass transfer, with
a hydrogen surface abundance of
.
We also calculated the
corresponding WR masses with
.
We plotted in Fig. 8 the initial WR masses (
and
)
versus the initial primary (progenitor)
masses. With "star'' symbols we indicated WR stars that originate from binary
systems with an initial mass ratio of
and an initial period
p=3 days (Table 2: N 4, 8, 9, 13, 14). Large "star'' symbols represent
WR stars with 5% of hydrogen at the
surface and small symbols indicate WR stars that
have a hydrogen surface abundance of less than 1%.
We derive a relation between the initial primary mass and the
initial WR mass (derived as a linear fit)
for p=3 days and
,
(
):
![]() |
(9) |
![]() |
(10) |
![]() |
Figure 8:
Initial WR mass as a function of initial (progenitor) mass.
Large and small symbols indicate WR stars with hydrogen surface abundance of
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Open with DEXTER |
Note that the WR masses that are the result of early Case A progenitor evolution
are significantly lower than ones that are the result
of Case B evolution (Wellstein & Langer 1999), because of the mass transfer
from the primary during the core hydrogen burning phase.
![]() |
Figure 9:
Mass transfer rate ( upper plot) and orbital period ( lower plot) during Case A
mass transfer as a function of the change of the primary mass for
systems with the initial primary
![]() ![]() |
Open with DEXTER |
During the mass transfer phase, the mass transfer rate increases roughly until the
masses of both components are equal.
The maximum mass transfer rate during Case A increases with the increase of the initial mass ratio
(
)
and the resulting WR star is less massive.
To analyse the influence of the initial mass ratio on the evolution of the binary system, we compared
systems with an initial primary mass of
,
an initial
orbital period of
days for five different initial mass ratios: 2.05, 2.00, 1.71, 1.52 and 1.37.
(Table 2 N 1, 3, 4, 6, 7, Fig. 9).
The system with
enters contact during fast Case A mass transfer. The
mass transfer rate
in this case is very high (
),
the secondary expands, fills its Roche lobe and the system enters a contact phase.
The system with an
initial mass ratio of
loses
during the fast phase of Case A. The maximum mass transfer rate of this system is
.
The helium surface abundance of the primary
after this mass transfer is 65%, so the
primary shrinks, loses mass through a WR stellar wind
and there is no slow phase of Case A mass transfer (
).
For the other three models q=1.71, 1.52, 1.37, the primaries lose less mass
(
15, 14, 13
respectively) during the fast phase of Case
A mass transfer. The helium surface abundances in these systems after fast Case A mass transfer
are
30-35%. The primaries
expand (
-15
)
on a nuclear time scale and transfer
mass to the secondaries (slow phase of Case A).
We can conclude the following:
first, if the initial mass ratio is larger, the mass transfer rate from the primary during fast phase
Case A mass transfer is higher.
Second, if the mass transfer rate is higher, the helium surface abundance of the star
increases faster and if it reaches 58%, the primary starts losing mass with
a higher (WR) mass loss rate and slow Case A mass transfer can be avoided.
We also show in Fig. 9 (lower plot) how the period changes during Case A mass transfer for binary systems N 3, 4, 6, 7. Roughly, when the mass is transfered from the more to the less massive star, the binary orbit shrinks, and when the mass is transfered from the less to the more massive star, the orbit widens. If the initial period is close to unity, the absolute difference between stellar masses is small, and more mass is transfered from the less to the more massive star during the evolution of the system. This results in a longer final period after Case A mass transfer. Systems with initial mass ratios of 2.00, 1.71, 1.52 and 1.37 enter Case AB mass transfer with orbital periods of 2.9, 3.9, 4.4 and 5.2 days respectively. However, the final period is also (more significantly) influenced by the stellar wind mass loss rate and the amount of matter lost from the primary during Case AB mass transfer (see Sect. 5.4).
![]() |
Figure 10:
Mass transfer rate during Case A mass transfer for systems
41
![]() ![]() ![]() ![]() ![]() ![]() |
Open with DEXTER |
If the initial orbital period increases for 3 days, a 41
star will
enter Case A mass transfer
later and a 56
star
later.
So, there are two things to point out:
first, the more massive star (56
)
evolves faster, and second,
a 3 days longer
initial period postpones Case A mass transfer, for this star, by about 10
more than for a 41
star.
The net effect is a more significant increase of the convective
core (i.e. initial helium core, i.e. initial WR mass),
for more massive star, due to the initial orbit widening.
![]() |
Figure 11:
Primary mass ( first plot), mass transfer rate ( second plot) and stellar wind
mass loss rate from the primary (third plot) until the onset of Case AB mass transfer for the
system
![]() ![]() ![]() ![]() |
Open with DEXTER |
We show in Fig. 11 the influence of the stellar wind mass loss (Plot c) on the primary mass
(Plot a) and mass transfer rate (Plot b) of
systems with
,
,
and three different stellar wind mass loss rates (from
): 1/6 (solid line),
1/3 (dotted line) and 1/2 (dashed line) of the mass loss proposed by Hamann et al. (1995).
We notice that for higher mass loss rates, the slow phase of Case A
stops earlier, due to the decrease of the stellar radius.
The orbit is widening due to the stellar wind mass loss and
the final period increases with the increasing mass loss rate.
However, the orbit is more significantly widening during Case AB mass transfer.
The more mass there is to
transfer from the primary to the secondary during Case AB mass transfer,
the larger the final orbital period.
So, if the stellar wind removes most of the hydrogen envelope of the primary,
there will be less mass to transfer during Case AB and
the net effect of a higher mass loss rate is a shorter orbital period of the WR+O system.
When mass transfer in a binary system starts, the primary loses matter through the first Lagrangian
point (). This matter carries a certain angular momentum that will be transfered to the
secondary. If there is an accretion disk, the angular momentum of the transfered matter is assumed to
be Keplerian. If there is a direct impact accretion, like in our models, we calculate the angular
momentum following a test particle moving through
.
This angular momentum spins up the top layers of the secondary star, and
angular momentum is transfered further into the star due to rotationally induced mixing processes.
Every time the secondary spins up to close to critical rotation it starts losing
more mass due to the influence of centrifugal force (Eq. (5)). High mass loss decreases the net
accretion efficiency and
also removes angular momentum from the secondary star. The secondary star is also spun down by tidal
forces that tend to synchronize it with the orbital motion. Wellstein (2001)
investigated these processes in binary systems with
initial mass ratios close to unity and concluded that
the accretion efficiency does not decrease significantly for Case A mass transfer, but in
the Case B the parameter
can be significantly decreased by rotation.
We present Case A rotating models with larger mass ratio
...2 and
find
that accretion can be significantly decreased during Case A mass transfer. The reason is the following:
if the initial mass ratio increases, so does the maximum mass transfer rate (
increases roughly until the masses in binary system are equal). If there is more mass
transfered from the primary to the secondary, the rotational velocity of the secondary is higher
as well as its mass loss, which leads to a smaller accretion efficiency.
We compare the evolution of non-rotating and
rotating binary systems on the example
,
an initial orbital period of p=6 days,
and Hamann/3 WR mass loss stellar wind rate (Table 3: N 6).
The rotating binary system is synchronized as it starts core hydrogen burning and it stays that way until
mass transfer starts. The radius of the primary increases during the main sequence phase
(from
10 to
25
,
Fig. 19b),
but the rotation of the primary stays synchronized with the orbital
period. This is why the rotational velocity of the primary also increases from
100 to
200
(Fig. 19d). The radius of the rotating primary increases faster
than the radius of the non-rotating primary due to the influence of the centrifugal force.
The result is that Case A mass transfer starts earlier for the rotating binary system
(
%) then for the corresponding nonrotating one(
%,
Fig. 15).
When the fast phase of Case A starts, the secondary spins up (Fig. 20d)
and stellar wind mass loss rapidly increases
(
,
Fig. 20c).
The accretion efficiency during this phase in the rotating system is
(Table 3).
We see in Fig. 15 that the orbital period after Case A mass transfer of the
rotating binary system is
shorter than for the non-rotating system (4.5 compared with 6.6 days).
The orbital angular momentum of the binary is changing due to mass transfer,
mass loss from the system and spin-orbit coupling. The rotating binary system loses more angular
momentum and the final orbital period is shorter than in the corresponding non-rotating system.
Angular momentum loss in our systems is calculated according to Podsiadlowski et al. (1992)
as already mentioned
in Sect. 4, and parameter
that determines the efficiency of angular momentum loss is
calculated according to Brookshaw & Tavani (1993). It increases with the mass ratio
and
the ratio between the secondary radius and its Roche radius
.
In rotating system the secondary
accretes slightly more matter (
)
compared to
in non-rotating systems,
so the mass ratio
is larger in the rotating system. Second, the secondary is spinning fast and its
radius is larger than in the non-rotating case, and so is the ratio
.
The result is that the angular
momentum is more efficiently removed from the system in the rotating binary system.
After the fast phase of Case A mass transfer, the two primaries, non-rotating and rotating, have almost the same
mass
34
and helium surfa- ce abundance
%. However, since the orbital periods are
Table 3:
Rotating WR+O progenitor models.
N is the number of the model,
and
are initial masses of the primary and
the secondary,
is the initial orbital period and
is the initial mass ratio of
the binary system.
is the time when Case A mass transfer starts,
is the duration of the the fast phase of Case A mass transfer,
is the maximum mass
transfer rate,
and
are mass loss of the primary and mass gain
of the secondary (respectively) during the fast Case A,
is the duration of slow Case A mass
transfer,
and
are mass loss of the primary and mass gain
of the secondary (respectively) during the slow Case A,
is the orbital period at the onset of Case
AB,
is the mass loss of the primary during Case AB (mass gain of the secondary is
1/10 of this, see Sect. 4),
is the WR mass when the hydrogen surface abundance is
,
is the mass of the corresponding O star, q is the
mass ratio
,
p is the orbital period of the WR+O system and
is WR mass with
.
The models are computed with a stellar wind mass loss of Hamann/6 except :
Hamann/3,
Hamann/2.
c indicates a contact phase.
Table 4:
Comparison of resulting WR masses and orbital periods from
non-rotating and rotating binary systems with the same initial parameters.
are
initial primary and secondary mass,
is the initial orbital period,
,
are WR masses at
and
respectively
and p is the orbital period in the initial WR+O system where the hydrogen surface abundance of WR star is
.
Systems are modelled with WR stellar wind mass loss H/6 except
which are done with H/3,
indicates rotating models.
When the fast phase of Case A is finished, the non-rotating primary has still 20% of
hydrogen to burn (
), and the rotating primary has
10% more than
that (
). When the surface hydrogen abundance is less than 40%,
the primaries start losing mass as WR stars, i.e., their stellar wind mass loss rate increases.
Since the rotating primary has more time to spend on the main sequence, it also has more time to lose
mass by WR stellar wind mass loss (
compared with
for non-rotating system).The result is that the non-rotating primary enters Case AB
mass transfer as a
26
star with
,
while the rotating one is a
17
star
with
.
Clearly, the rotating primary has less hydrogen in its envelope, i.e. less mass to
transfer to the secondary during Case AB mass transfer, and the orbit widens less than in the non-rotating
system.
We can draw the conclusion that if rotation is included in our calculations,
the initial WR mass is smaller and the orbital period of the WR+O system is
shorter than in the corresponding non-rotating system (Table 4).
We present in Fig. 12 the evolutionary tracks of the rotating primary and secondary in the
HR diagram.
Both stars are core hydrogen burning stars (dashed line, Fig. 12),
but since the primary is more
massive, it evolves faster and fills its Roche lobe, so
the system enters Case A mass transfer (solid line, Fig. 12).
The primary loses matter quickly with a high mass transfer rate
(
)
and its luminosity decreases (Fig. 12).
At the same time
the secondary accretes matter and its luminosity increases, but due to
change in rotational velocity (Fig. 20d) its radius and effective temperature are changing
as well (Figs. 12d, 20a,b).
During fast Case A mass transfer the primary lost
19
and the secondary
accreted 15% of that matter.
After the fast mass transfer,
the primary is still burning hydrogen in
its core and is still expanding, so
slow Case A mass transfer takes place. After the primary starts losing mass with a WR
stellar wind mass loss rate (
)
its radius will decrease and the slow phase of
Case A stops (Fig. 19c).
However, the primary continues expanding on the nuclear time scale (Fig. 19b)
and it fills its Roche lobe once again (Fig. 15, upper plot).
At the end of core hydrogen burning the primary contracts (effective temperature increases)
and thus RLOF stops. This phase is presented in Fig. 12 with a dotted line.
When hydrogen starts burning in
a shell, the primary star expands (dash-dotted line, Fig. 12),
fills its Roche lobe and Case AB mass transfer starts.
![]() |
Figure 12:
HR diagram of the initial system
![]() ![]() ![]() |
Open with DEXTER |
![]() |
Figure 13:
The evolution of the internal structure of the rotating 56
![]() |
Open with DEXTER |
![]() |
Figure 14:
The evolution of the internal structure of the rotating 33
![]() |
Open with DEXTER |
The initial helium core masses are 18.6
for the non-rotating and 14.8
for
the rotating primary. When Case AB mass transfer starts, the orbital periods are 7.9 d and 6.6 d
for the non-rotating and the rotating system respectively (Fig. 15, lower plot).
The non-rotating primary loses
7
and the rotating one
2
during Case AB. When there is more mass to be transfered from the less to the more massive star in a binary
system, the orbit widens more and the final orbital period is longer.
Figures 13 and 14 show the structure of the primary and the secondary
before Case AB mass transfer.
The primary loses large amounts of matter during the fast phase of Case A mass transfer (20
),
and its convective core becomes less than half of its original mass. At the same time, the
secondary accretes matter from the primary and the heavier elements are being relocated by thermohaline
mixing.
Figures 16 and 17
show surface abundances of the primary and the secondary.
The secondary is accreting material from the primary and its surface abundances
change due to this, but also due to thermohaline and rotational mixing.
Figure 18 shows the orbital angular momentum of the system and the spin periods of both components. The orbital angular momentum of the system decreases rapidly due to mass loss from the system during fast Case A mass transfer, and then further due to stellar wind mass loss. The primary slows down rapidly during fast Case A and further due to stellar wind mass loss. The secondary spins up due to the accretion from the primary during fast Case A mass transfer and then slows down due to stellar wind mass loss. It spins up again during slow Case A mass transfer.
The masses of modelled WR stars are in the range from 11
to
15.7
.
Period of modelled WR+O systems vary
from
7.6 to
12.7 days and mass ratios are between 0.35 and 0.46 (Table 3).
![]() |
Figure 15:
Upper plot: the mass transfer rate during
Case A mass transfer in the binary systems with
![]() ![]() ![]() |
Open with DEXTER |
![]() |
Figure 16:
The hydrogen surface abundance (solid line) in the primary in system
with
![]() ![]() ![]() |
Open with DEXTER |
![]() |
Figure 17:
Surface abundance of carbon (solid line), nitrogen (dotted line) and
oxygen (dashed line) in the primary ( upper plot)
and the secondary ( lower plot), in the system with
![]() ![]() ![]() |
Open with DEXTER |
![]() |
Figure 18:
The orbital angular momentum ( upper plot) of
the non-rotating (dotted line) and the rotating (solid line) binary systems with
![]() ![]() ![]() |
Open with DEXTER |
![]() |
Figure 19:
Effective temperature (plot a)), stellar radius (plot b)), stellar wind mass
loss rate (plot c)) and rotational velocity (plot d)) of the primary star
in the non-rotating (dotted line) and
rotating (solid line) binary system with
![]() ![]() ![]() |
Open with DEXTER |
![]() |
Figure 20:
Effective temperature (plot a)), stellar radius (plot b)), stellar wind mass
loss rate (plot c)) and rotational velocity (plot d)) of the secondary star
in the non-rotating (dotted line) and
rotating (solid line) binary system with
![]() ![]() ![]() |
Open with DEXTER |
We show in Table 5 average accretion efficiencies of rotating binary systems during different
mass transfer phases, and total average values with and without stellar wind mass loss from the primary
included.
During fast Case A mass transfer, the primary stars are losing matter with very high mass transfer rates
(3...
). The angular momentum of surface layers in the secondary
increases fast, they spin up to close to the critical rotation and start losing mass
with high mass loss rate
(
10
). The average accretion efficiency during fast Case A in our
models is 15-20%. Since this phase takes place on the thermal time scale, stellar wind mass loss from the
primary is negligible during this phase.
Slow Case A mass transfer takes place on the nuclear time scale. The primary stars start losing
their mass due to a WR stellar wind when their surfaces become hydrogen deficient (
).
The WR stellar wind mass
loss rates are of the order of:
,
and
we have to take into account stellar wind mass loss of the primary during slow Case A.
We calculate the mass loss of the primary only due to mass transfer and total mass loss including stellar
wind mass loss, and the two corresponding average accretion efficiencies.
If we calculate
only for mass transfer, we notice that the slow Case A is
almost
a conservative process. The average mass transfer rates are
10
and the secondary stars are able to accrete almost everything without spinning up to critical
rotation.
![]() |
Figure 21:
Upper plot: mass transfer (solid line) and accretion rate (dotted line) of the rotating
initial system
56
![]() |
Open with DEXTER |
Figure 21 shows how mass transfer rate, accretion rate and
change in the rotating model with 56
,
days
(WR mass loss Hamann/3) depending on the amount of matter lost by the primary.
We also see in this figure the mass transfer rate from the primary in the non-rotating case.
We can notice in the upper plot, what we previously discussed,
that during most of the fast Case A mass transfer,
the mass accretion rate of the secondary is about
one order of magnitude lower than the mass loss rate of the primary.
The primary loses
19.3
during the fast phase and the secondary gains
2.9
,
which means that on
average
15% of the mass has been accreted.
However, the mass loss of the primary due to mass transfer during the slow phase is
4.5
,
and the secondary accretes
3.9
which means that
.
If we take into account stellar wind mass loss of the primary stars,
the average accretion efficiencies are lower.
For example, the total mass loss of the
primary during slow Case A mass transfer, including the stellar wind, in the previous example is
10.9
,
which means that
.
We neglected accretion during Case AB mass transfer since Wellstein (2001) showed that
it is inefficient, and since the primary stars in the modelled systems
have relatively low mass hydrogen envelopes, and masses of secondary stars will not
significantly change due to this mass transfer.
(However, let us not forget that even the accretion of very small amounts of matter
can be important for spinning up the secondary's surface layers and making it rotate
faster than synchronously in a WR + O binary system.)
Also, since this mass transfer
takes place on the thermal time scale, stellar wind mass loss can be neglected.
Finally, we can estimate the total mass loss from the binary systems including stellar wind,
or only due to mass
transfer, and calculate corresponding values of .
In the binary systems we modelled, the primary stars lose between 30
and 45
due to mass
transfer and stellar wind, until they
ignite helium in their core. The amount of lost mass increases with initial mass. At the same time
the secondaries accrete 3...10
.
This means that in most cases 80...90% of the mass lost by the primary leaves the binary system.
On the other hand, the primary stars lose
20...30
only due to mass transfer, so
the average accretion of secondary stars in our models is between 15 and 30%.
Table 5:
Mass loss from binary systems. N is number of the model corresponding to
Table 3.
is the average accretion efficiency of the secondary during the
fast phase of
Case A mass transfer.
is the accretion efficiency of the secondary during the
slow phase of
Case A mass transfer taking into account matter lost by the primary only due to the mass transfer.
is the average accretion efficiency of the secondary during the slow phase of
Case A mass transfer taking into account matter lost by the primary due to the mass transfer and stellar
wind.
is the average accretion efficiency of the secondary during the
progenitor evolution of WR+O binary
system taking into account matter lost by the primary only due to the mass transfer and
taking also into account stellar wind mass loss of the primary.
Our rotating models give generally similar results as our non-rotating models
for .
The rotating binary systems R6 (56
,
p=6 days,
Hamann/3 WR mass loss) and R1 and R2
(41
,
p=6 days) agree quite well with the observed
systems HD 186943 and HD 90657,
as well as the non-rotating systems N11 and N12
(56
,
p=6 days; WR mass loss rate Hamann/2 and
Hamann/3).
The system R6
evolves into a WR+O configuration with
15
+39
and p=8.5 days.
I.e., its masses and period are close to those found in HD 186943 and HD 90657,
even though its mass ratio of 0.38 is somewhat smaller than what is observed.
Systems R1 and R2
evolve into a
11
WR+O system with a 9.8 day orbital period.
I.e., period and mass ratio (0.46) agree well with the observed systems,
but the stellar masses are somewhat smaller than observed (cf. Sect. 2).
Systems N11 and N12 evolve into a WR+O system of
19
with an orbital period of 12...14 days.
In this case, both masses and the mass ratio (0.53) agree well with the observed
ones, but the orbital period is slightly too large.
I.e., although none of our models is a perfect match of HD 186943 or HD 90657
- which to find would require many more models, however, might not teach us
very much - it is clear form these results that both systems can in fact be well
explained through highly inefficient Case A mass transfer.
The situation is more difficult with HD 211853 (GP Cep):
neither the models with nor those without rotation reproduce it satisfactory.
HD 211853 has the shortest period (6.7 d) and largest mass ratio (0.54) of the three
chosen Galactic WR+O binaries. While we can not exclude that
a Case A model of the kind presented here can reproduce this systems,
especially the small period makes it appear more likely that this
system has gone through a contact phase: contact would reduce the orbital angular momentum,
and increase the mass loss from the system, i.e. result in a larger WR/O mass ratio
(Wellstein et al. 2001).
This reasoning is strengthened by the consideration that, in contrast to HD 186943 or HD 90657,
the WR star in HD 211853 is of spectral type WN6/WNC. I.e., as this spectroscopic
signature is not interpreted in terms of a binary nature of the WR component,
but rather by assuming that the WR star is in the transition phase from the WN to the
WC stage (Langer 1991; Massey & Grove 1989). This implies that the WR star
in HD 211853 must have already lost several solar masses of helium-rich matter,
which causes the orbit to widen.
For example, system R6, which evolved into a
14.8
+39.0
WR+O system with p=8.53 days,
evolves into a WC+O system
after losing
5
more from the Wolf-Rayet star,
which increases its orbital period by
3 days.
I.e., HD 211853 might have entered the WR+O stage with an orbital period of about
4 days, which would put it together with the shortest period WR binaries like
CX Cep or CQ Cep whose periods are 2.1 and 1.64 days respectively.
During the evolution of WR+O binary system, the primary loses
mass due to WR stellar wind mass loss.
WR stellar wind mass loss of the primary
decreases mass ratio of the system and increases the orbital period, which means
that, for example, WC+O binary system HD 63099 (
,
and p=14 days) could have evolved into present state
through a WN+O binary system with q=0.5.
In an effort to constrain the progenitor evolution of the three WN+O
binaries HD 186943, HD 90657, and HD 211853,
we calculated the evolution of non-conservative Case A binary systems with
primaries
and initial mass ratios
between 1.7 and 2 until the WN+O stage.
We performed binary evolution calculations neglecting rotational processes in the
two stellar components, and assuming a constant mass accretion efficiency of 10% for all three phases of the mass transfer, fast Case A, slow Case A, and Case AB.
Those models could match two of the three systems reasonably well,
while HD 211853, which has the shortest orbital period, the largest mass ratio, and
a WN/WC Wolf-Rayet component, was found to be not well explained by contact-free
evolutionary models: While models with shorter initial orbital periods result in short periods
during the WR+O stage, the initial WR mass is decreasing at the same time, which
leads to smaller initial WR/O mass ratios.
We then computed binary evolution models including the physics of rotation in
both stellar components as well as the spin-up process of the mass gainer due
to angular momentum accretion. In these models, the surface of the accreting star
is continuously spun-up by accretion, while at the same time angular momentum
is transported from the outer layers into the stellar interior by rotationally
induced mixing processes. By employing a simple model for the mass loss of
rapidly rotating luminous stars - the so called -limit, which
was actually worked out to describe the mass loss processes in Luminous Blue Variables
(Langer 1997) - accretion is drastically reduced once the star reaches
critical rotation at its surface. The mass accretion rate is then controlled
by the time scale of internal angular momentum transport.
Some first such model for Case A and early Case B have been computed
by Wellstein (Langer et al. 2003,2004) for a primary mass of 15
and
a mass ratio close to one. The result was that rather high mass accretion efficiencies
(
)
could be obtained for initial periods shorter than about 8 days.
Here we find that, with the same physical assumptions although at higher system mass,
the accretion efficiency drops to about 10% at an initial mass ratio of 1.7.
As Wellstein (2001) computed one early Case A model for a 26
system which gave
,
it is like the high initial mass ratio
in our models which is responsible for the low accretion efficiency:
larger initial mass ratios lead to larger mass transfer rates and, as the time scale
of internal angular momentum transport in the accreting star is rather unaffected,
to smaller accretion efficiencies.
Our rotating models - in which the accretion efficiency is no free parameter any more but is computed selfconsistent and time-dependent - reproduce the observed WR+O binaries quite well, i.e. as good as our models without rotation physics, where the accretion efficiency is a free parameter. Our simplified considerations in Sect. 3 have shown that this is unlikely attributable to the freedom in the choice of the initial parameter of the binary system, i.e. initial masses and period - at least under the assumption that contact was avoided. In case of contact, various new parameters enter the model, similar to the case of common envelope evolution. And indeed, also our rotating models can not reproduce HD 211853 very well, mostly because it currently has a too short orbital period, which was likely even significantly shorter at the beginning of its WR+O stage. However, this of course only confirms the result of the simpler approaches that a contact-free approach does not work well for this system.
In summary we can say that the system mass and angular momentum loss model used here - which is the first detailed approach to tackle the long-standing angular momentum problem in mass transferring binaries - has passed the test of WR+O binaries. However, it still needs to be explored over which part of the space spanned by the initial binary parameters this model works well, and to what extent its results are sensitive to future improvements in the stellar interior physics. The inclusion of magnetic fields generated by differential rotation (Spruit 2002) will be the next step in this direction (Petrovic et al. 2004).